[PATCH 2/5] lib: zstd: sync with Linux

Ahmad Fatoum a.fatoum at pengutronix.de
Wed Jul 13 03:09:19 PDT 2022


Import the Linux v5.18.3 state of the zstd decompression code. The
compressor was omitted and error strings were disabled for PBL.

Signed-off-by: Ahmad Fatoum <a.fatoum at pengutronix.de>
---
 fs/squashfs/zstd_wrapper.c                    |    4 +-
 fs/ubifs/ubifs.c                              |    4 +-
 include/linux/decompress/unzstd.h             |   11 +
 include/linux/limits.h                        |    1 +
 include/linux/zstd.h                          | 1252 ++------
 include/linux/zstd_errors.h                   |   77 +
 include/linux/zstd_lib.h                      | 2432 ++++++++++++++++
 lib/zstd/Makefile                             |   11 +-
 lib/zstd/bitstream.h                          |  374 ---
 lib/zstd/common/bitstream.h                   |  437 +++
 lib/zstd/common/compiler.h                    |  177 ++
 lib/zstd/common/cpu.h                         |  194 ++
 lib/zstd/common/debug.c                       |   24 +
 lib/zstd/common/debug.h                       |  101 +
 lib/zstd/common/entropy_common.c              |  544 ++--
 lib/zstd/common/error_private.c               |   56 +
 lib/zstd/common/error_private.h               |   69 +
 lib/zstd/common/fse.h                         |  710 +++++
 lib/zstd/common/fse_decompress.c              |  594 ++--
 lib/zstd/common/huf.h                         |  356 +++
 lib/zstd/common/mem.h                         |  259 ++
 lib/zstd/common/zstd_common.c                 |  118 +-
 lib/zstd/common/zstd_deps.h                   |  125 +
 lib/zstd/common/zstd_internal.h               |  450 +++
 lib/zstd/decompress.c                         | 2528 -----------------
 lib/zstd/decompress/huf_decompress.c          | 2076 ++++++++------
 lib/zstd/decompress/zstd_ddict.c              |  241 ++
 lib/zstd/decompress/zstd_ddict.h              |   44 +
 lib/zstd/decompress/zstd_decompress.c         | 2085 ++++++++++++++
 lib/zstd/decompress/zstd_decompress_block.c   | 1540 ++++++++++
 lib/zstd/decompress/zstd_decompress_block.h   |   62 +
 .../decompress/zstd_decompress_internal.h     |  202 ++
 lib/zstd/decompress_sources.h                 |   28 +
 lib/zstd/error_private.h                      |   53 -
 lib/zstd/fse.h                                |  575 ----
 lib/zstd/huf.h                                |  212 --
 lib/zstd/mem.h                                |  151 -
 lib/zstd/zstd_decompress_module.c             |  105 +
 lib/zstd/zstd_internal.h                      |  263 --
 lib/zstd/zstd_opt.h                           | 1014 -------
 40 files changed, 11947 insertions(+), 7612 deletions(-)
 create mode 100644 include/linux/decompress/unzstd.h
 create mode 100644 include/linux/zstd_errors.h
 create mode 100644 include/linux/zstd_lib.h
 delete mode 100644 lib/zstd/bitstream.h
 create mode 100644 lib/zstd/common/bitstream.h
 create mode 100644 lib/zstd/common/compiler.h
 create mode 100644 lib/zstd/common/cpu.h
 create mode 100644 lib/zstd/common/debug.c
 create mode 100644 lib/zstd/common/debug.h
 create mode 100644 lib/zstd/common/error_private.c
 create mode 100644 lib/zstd/common/error_private.h
 create mode 100644 lib/zstd/common/fse.h
 create mode 100644 lib/zstd/common/huf.h
 create mode 100644 lib/zstd/common/mem.h
 create mode 100644 lib/zstd/common/zstd_deps.h
 create mode 100644 lib/zstd/common/zstd_internal.h
 delete mode 100644 lib/zstd/decompress.c
 create mode 100644 lib/zstd/decompress/zstd_ddict.c
 create mode 100644 lib/zstd/decompress/zstd_ddict.h
 create mode 100644 lib/zstd/decompress/zstd_decompress.c
 create mode 100644 lib/zstd/decompress/zstd_decompress_block.c
 create mode 100644 lib/zstd/decompress/zstd_decompress_block.h
 create mode 100644 lib/zstd/decompress/zstd_decompress_internal.h
 create mode 100644 lib/zstd/decompress_sources.h
 delete mode 100644 lib/zstd/error_private.h
 delete mode 100644 lib/zstd/fse.h
 delete mode 100644 lib/zstd/huf.h
 delete mode 100644 lib/zstd/mem.h
 create mode 100644 lib/zstd/zstd_decompress_module.c
 delete mode 100644 lib/zstd/zstd_internal.h
 delete mode 100644 lib/zstd/zstd_opt.h

diff --git a/fs/squashfs/zstd_wrapper.c b/fs/squashfs/zstd_wrapper.c
index 86740edd4855..aae981f6cee0 100644
--- a/fs/squashfs/zstd_wrapper.c
+++ b/fs/squashfs/zstd_wrapper.c
@@ -39,7 +39,7 @@ static void *zstd_init(struct squashfs_sb_info *msblk, void *buff)
 		goto failed;
 	wksp->window_size = max_t(size_t,
 			msblk->block_size, SQUASHFS_METADATA_SIZE);
-	wksp->mem_size = ZSTD_DStreamWorkspaceBound(wksp->window_size);
+	wksp->mem_size = zstd_dstream_workspace_bound(wksp->window_size);
 	wksp->mem = vmalloc(wksp->mem_size);
 	if (wksp->mem == NULL)
 		goto failed;
@@ -75,7 +75,7 @@ static int zstd_uncompress(struct squashfs_sb_info *msblk, void *strm,
 	ZSTD_inBuffer in_buf = { NULL, 0, 0 };
 	ZSTD_outBuffer out_buf = { NULL, 0, 0 };
 
-	stream = ZSTD_initDStream(wksp->window_size, wksp->mem, wksp->mem_size);
+	stream = zstd_init_dstream(wksp->window_size, wksp->mem, wksp->mem_size);
 
 	if (!stream) {
 		ERROR("Failed to initialize zstd decompressor\n");
diff --git a/fs/ubifs/ubifs.c b/fs/ubifs/ubifs.c
index 88a4340a3856..e7b434c1960e 100644
--- a/fs/ubifs/ubifs.c
+++ b/fs/ubifs/ubifs.c
@@ -525,13 +525,13 @@ static int zlib_decomp_init(void)
 
 static int zstd_decomp_init(void)
 {
-	const size_t wksp_size = ZSTD_DCtxWorkspaceBound();
+	const size_t wksp_size = zstd_dctx_workspace_bound();
 	void *wksp = malloc(wksp_size);
 
 	if (!wksp)
 		return -ENOMEM;
 
-	ubifs_zstd_cctx = ZSTD_initDCtx(wksp, wksp_size);
+	ubifs_zstd_cctx = zstd_init_dctx(wksp, wksp_size);
 	if (!ubifs_zstd_cctx)
 		return -EINVAL;
 
diff --git a/include/linux/decompress/unzstd.h b/include/linux/decompress/unzstd.h
new file mode 100644
index 000000000000..ac078a3ba151
--- /dev/null
+++ b/include/linux/decompress/unzstd.h
@@ -0,0 +1,11 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef LINUX_DECOMPRESS_UNZSTD_H
+#define LINUX_DECOMPRESS_UNZSTD_H
+
+int unzstd(unsigned char *inbuf, int len,
+	   int (*fill)(void*, unsigned int),
+	   int (*flush)(void*, unsigned int),
+	   unsigned char *output,
+	   int *pos,
+	   void (*error_fn)(char *x));
+#endif
diff --git a/include/linux/limits.h b/include/linux/limits.h
index bda9c94bb5ae..e1aa2de84519 100644
--- a/include/linux/limits.h
+++ b/include/linux/limits.h
@@ -18,6 +18,7 @@
 #define ULLONG_MAX	(~0ULL)
 #define SIZE_MAX	(~(size_t)0)
 #define PHYS_ADDR_MAX	(~(phys_addr_t)0)
+#define UINTPTR_MAX	ULONG_MAX
 
 #define U8_MAX		((u8)~0U)
 #define S8_MAX		((s8)(U8_MAX >> 1))
diff --git a/include/linux/zstd.h b/include/linux/zstd.h
index 249575e2485f..113408eef6ec 100644
--- a/include/linux/zstd.h
+++ b/include/linux/zstd.h
@@ -1,138 +1,97 @@
+/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of https://github.com/facebook/zstd) and
+ * the GPLv2 (found in the COPYING file in the root directory of
+ * https://github.com/facebook/zstd). You may select, at your option, one of the
+ * above-listed licenses.
  */
 
-#ifndef ZSTD_H
-#define ZSTD_H
+#ifndef LINUX_ZSTD_H
+#define LINUX_ZSTD_H
 
-/* ======   Dependency   ======*/
-#include <linux/types.h>   /* size_t */
+/**
+ * This is a kernel-style API that wraps the upstream zstd API, which cannot be
+ * used directly because the symbols aren't exported. It exposes the minimal
+ * functionality which is currently required by users of zstd in the kernel.
+ * Expose extra functions from lib/zstd/zstd.h as needed.
+ */
 
+/* ======   Dependency   ====== */
+#include <linux/types.h>
+#include <linux/zstd_errors.h>
+#include <linux/zstd_lib.h>
 
-/*-*****************************************************************************
- * Introduction
+/* ======   Helper Functions   ====== */
+/**
+ * zstd_compress_bound() - maximum compressed size in worst case scenario
+ * @src_size: The size of the data to compress.
  *
- * zstd, short for Zstandard, is a fast lossless compression algorithm,
- * targeting real-time compression scenarios at zlib-level and better
- * compression ratios. The zstd compression library provides in-memory
- * compression and decompression functions. The library supports compression
- * levels from 1 up to ZSTD_maxCLevel() which is 22. Levels >= 20, labeled
- * ultra, should be used with caution, as they require more memory.
- * Compression can be done in:
- *  - a single step, reusing a context (described as Explicit memory management)
- *  - unbounded multiple steps (described as Streaming compression)
- * The compression ratio achievable on small data can be highly improved using
- * compression with a dictionary in:
- *  - a single step (described as Simple dictionary API)
- *  - a single step, reusing a dictionary (described as Fast dictionary API)
- ******************************************************************************/
-
-/*======  Helper functions  ======*/
+ * Return:    The maximum compressed size in the worst case scenario.
+ */
+size_t zstd_compress_bound(size_t src_size);
 
 /**
- * enum ZSTD_ErrorCode - zstd error codes
+ * zstd_is_error() - tells if a size_t function result is an error code
+ * @code:  The function result to check for error.
  *
- * Functions that return size_t can be checked for errors using ZSTD_isError()
- * and the ZSTD_ErrorCode can be extracted using ZSTD_getErrorCode().
+ * Return: Non-zero iff the code is an error.
+ */
+unsigned int zstd_is_error(size_t code);
+
+/**
+ * enum zstd_error_code - zstd error codes
  */
-typedef enum {
-	ZSTD_error_no_error,
-	ZSTD_error_GENERIC,
-	ZSTD_error_prefix_unknown,
-	ZSTD_error_version_unsupported,
-	ZSTD_error_parameter_unknown,
-	ZSTD_error_frameParameter_unsupported,
-	ZSTD_error_frameParameter_unsupportedBy32bits,
-	ZSTD_error_frameParameter_windowTooLarge,
-	ZSTD_error_compressionParameter_unsupported,
-	ZSTD_error_init_missing,
-	ZSTD_error_memory_allocation,
-	ZSTD_error_stage_wrong,
-	ZSTD_error_dstSize_tooSmall,
-	ZSTD_error_srcSize_wrong,
-	ZSTD_error_corruption_detected,
-	ZSTD_error_checksum_wrong,
-	ZSTD_error_tableLog_tooLarge,
-	ZSTD_error_maxSymbolValue_tooLarge,
-	ZSTD_error_maxSymbolValue_tooSmall,
-	ZSTD_error_dictionary_corrupted,
-	ZSTD_error_dictionary_wrong,
-	ZSTD_error_dictionaryCreation_failed,
-	ZSTD_error_maxCode
-} ZSTD_ErrorCode;
+typedef ZSTD_ErrorCode zstd_error_code;
 
 /**
- * ZSTD_maxCLevel() - maximum compression level available
+ * zstd_get_error_code() - translates an error function result to an error code
+ * @code:  The function result for which zstd_is_error(code) is true.
  *
- * Return: Maximum compression level available.
+ * Return: A unique error code for this error.
  */
-int ZSTD_maxCLevel(void);
+zstd_error_code zstd_get_error_code(size_t code);
+
 /**
- * ZSTD_compressBound() - maximum compressed size in worst case scenario
- * @srcSize: The size of the data to compress.
+ * zstd_get_error_name() - translates an error function result to a string
+ * @code:  The function result for which zstd_is_error(code) is true.
  *
- * Return:   The maximum compressed size in the worst case scenario.
+ * Return: An error string corresponding to the error code.
  */
-size_t ZSTD_compressBound(size_t srcSize);
+const char *zstd_get_error_name(size_t code);
+
 /**
- * ZSTD_isError() - tells if a size_t function result is an error code
- * @code:  The function result to check for error.
+ * zstd_min_clevel() - minimum allowed compression level
  *
- * Return: Non-zero iff the code is an error.
+ * Return: The minimum allowed compression level.
  */
-static __attribute__((unused)) unsigned int ZSTD_isError(size_t code)
-{
-	return code > (size_t)-ZSTD_error_maxCode;
-}
+int zstd_min_clevel(void);
+
 /**
- * ZSTD_getErrorCode() - translates an error function result to a ZSTD_ErrorCode
- * @functionResult: The result of a function for which ZSTD_isError() is true.
+ * zstd_max_clevel() - maximum allowed compression level
  *
- * Return:          The ZSTD_ErrorCode corresponding to the functionResult or 0
- *                  if the functionResult isn't an error.
+ * Return: The maximum allowed compression level.
  */
-static __attribute__((unused)) ZSTD_ErrorCode ZSTD_getErrorCode(
-	size_t functionResult)
-{
-	if (!ZSTD_isError(functionResult))
-		return (ZSTD_ErrorCode)0;
-	return (ZSTD_ErrorCode)(0 - functionResult);
-}
+int zstd_max_clevel(void);
+
+/* ======   Parameter Selection   ====== */
 
 /**
- * enum ZSTD_strategy - zstd compression search strategy
+ * enum zstd_strategy - zstd compression search strategy
  *
- * From faster to stronger.
+ * From faster to stronger. See zstd_lib.h.
  */
-typedef enum {
-	ZSTD_fast,
-	ZSTD_dfast,
-	ZSTD_greedy,
-	ZSTD_lazy,
-	ZSTD_lazy2,
-	ZSTD_btlazy2,
-	ZSTD_btopt,
-	ZSTD_btopt2
-} ZSTD_strategy;
+typedef ZSTD_strategy zstd_strategy;
 
 /**
- * struct ZSTD_compressionParameters - zstd compression parameters
+ * struct zstd_compression_parameters - zstd compression parameters
  * @windowLog:    Log of the largest match distance. Larger means more
  *                compression, and more memory needed during decompression.
- * @chainLog:     Fully searched segment. Larger means more compression, slower,
- *                and more memory (useless for fast).
+ * @chainLog:     Fully searched segment. Larger means more compression,
+ *                slower, and more memory (useless for fast).
  * @hashLog:      Dispatch table. Larger means more compression,
  *                slower, and more memory.
  * @searchLog:    Number of searches. Larger means more compression and slower.
@@ -141,1017 +100,348 @@ typedef enum {
  * @targetLength: Acceptable match size for optimal parser (only). Larger means
  *                more compression, and slower.
  * @strategy:     The zstd compression strategy.
+ *
+ * See zstd_lib.h.
  */
-typedef struct {
-	unsigned int windowLog;
-	unsigned int chainLog;
-	unsigned int hashLog;
-	unsigned int searchLog;
-	unsigned int searchLength;
-	unsigned int targetLength;
-	ZSTD_strategy strategy;
-} ZSTD_compressionParameters;
+typedef ZSTD_compressionParameters zstd_compression_parameters;
 
 /**
- * struct ZSTD_frameParameters - zstd frame parameters
- * @contentSizeFlag: Controls whether content size will be present in the frame
- *                   header (when known).
- * @checksumFlag:    Controls whether a 32-bit checksum is generated at the end
- *                   of the frame for error detection.
- * @noDictIDFlag:    Controls whether dictID will be saved into the frame header
- *                   when using dictionary compression.
+ * struct zstd_frame_parameters - zstd frame parameters
+ * @contentSizeFlag: Controls whether content size will be present in the
+ *                   frame header (when known).
+ * @checksumFlag:    Controls whether a 32-bit checksum is generated at the
+ *                   end of the frame for error detection.
+ * @noDictIDFlag:    Controls whether dictID will be saved into the frame
+ *                   header when using dictionary compression.
  *
- * The default value is all fields set to 0.
+ * The default value is all fields set to 0. See zstd_lib.h.
  */
-typedef struct {
-	unsigned int contentSizeFlag;
-	unsigned int checksumFlag;
-	unsigned int noDictIDFlag;
-} ZSTD_frameParameters;
+typedef ZSTD_frameParameters zstd_frame_parameters;
 
 /**
- * struct ZSTD_parameters - zstd parameters
+ * struct zstd_parameters - zstd parameters
  * @cParams: The compression parameters.
  * @fParams: The frame parameters.
  */
-typedef struct {
-	ZSTD_compressionParameters cParams;
-	ZSTD_frameParameters fParams;
-} ZSTD_parameters;
+typedef ZSTD_parameters zstd_parameters;
 
 /**
- * ZSTD_getCParams() - returns ZSTD_compressionParameters for selected level
- * @compressionLevel: The compression level from 1 to ZSTD_maxCLevel().
- * @estimatedSrcSize: The estimated source size to compress or 0 if unknown.
- * @dictSize:         The dictionary size or 0 if a dictionary isn't being used.
+ * zstd_get_params() - returns zstd_parameters for selected level
+ * @level:              The compression level
+ * @estimated_src_size: The estimated source size to compress or 0
+ *                      if unknown.
  *
- * Return:            The selected ZSTD_compressionParameters.
+ * Return:              The selected zstd_parameters.
  */
-ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel,
-	unsigned long long estimatedSrcSize, size_t dictSize);
+zstd_parameters zstd_get_params(int level,
+	unsigned long long estimated_src_size);
 
-/**
- * ZSTD_getParams() - returns ZSTD_parameters for selected level
- * @compressionLevel: The compression level from 1 to ZSTD_maxCLevel().
- * @estimatedSrcSize: The estimated source size to compress or 0 if unknown.
- * @dictSize:         The dictionary size or 0 if a dictionary isn't being used.
- *
- * The same as ZSTD_getCParams() except also selects the default frame
- * parameters (all zero).
- *
- * Return:            The selected ZSTD_parameters.
- */
-ZSTD_parameters ZSTD_getParams(int compressionLevel,
-	unsigned long long estimatedSrcSize, size_t dictSize);
+/* ======   Single-pass Compression   ====== */
 
-/*-*************************************
- * Explicit memory management
- **************************************/
+typedef ZSTD_CCtx zstd_cctx;
 
 /**
- * ZSTD_CCtxWorkspaceBound() - amount of memory needed to initialize a ZSTD_CCtx
- * @cParams: The compression parameters to be used for compression.
+ * zstd_cctx_workspace_bound() - max memory needed to initialize a zstd_cctx
+ * @parameters: The compression parameters to be used.
  *
  * If multiple compression parameters might be used, the caller must call
- * ZSTD_CCtxWorkspaceBound() for each set of parameters and use the maximum
+ * zstd_cctx_workspace_bound() for each set of parameters and use the maximum
  * size.
  *
- * Return:   A lower bound on the size of the workspace that is passed to
- *           ZSTD_initCCtx().
+ * Return:      A lower bound on the size of the workspace that is passed to
+ *              zstd_init_cctx().
  */
-size_t ZSTD_CCtxWorkspaceBound(ZSTD_compressionParameters cParams);
+size_t zstd_cctx_workspace_bound(const zstd_compression_parameters *parameters);
 
 /**
- * struct ZSTD_CCtx - the zstd compression context
- *
- * When compressing many times it is recommended to allocate a context just once
- * and reuse it for each successive compression operation.
- */
-typedef struct ZSTD_CCtx_s ZSTD_CCtx;
-/**
- * ZSTD_initCCtx() - initialize a zstd compression context
- * @workspace:     The workspace to emplace the context into. It must outlive
- *                 the returned context.
- * @workspaceSize: The size of workspace. Use ZSTD_CCtxWorkspaceBound() to
- *                 determine how large the workspace must be.
- *
- * Return:         A compression context emplaced into workspace.
- */
-ZSTD_CCtx *ZSTD_initCCtx(void *workspace, size_t workspaceSize);
-
-/**
- * ZSTD_compressCCtx() - compress src into dst
- * @ctx:         The context. Must have been initialized with a workspace at
- *               least as large as ZSTD_CCtxWorkspaceBound(params.cParams).
- * @dst:         The buffer to compress src into.
- * @dstCapacity: The size of the destination buffer. May be any size, but
- *               ZSTD_compressBound(srcSize) is guaranteed to be large enough.
- * @src:         The data to compress.
- * @srcSize:     The size of the data to compress.
- * @params:      The parameters to use for compression. See ZSTD_getParams().
- *
- * Return:       The compressed size or an error, which can be checked using
- *               ZSTD_isError().
- */
-size_t ZSTD_compressCCtx(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize, ZSTD_parameters params);
-
-/**
- * ZSTD_DCtxWorkspaceBound() - amount of memory needed to initialize a ZSTD_DCtx
- *
- * Return: A lower bound on the size of the workspace that is passed to
- *         ZSTD_initDCtx().
- */
-size_t ZSTD_DCtxWorkspaceBound(void);
-
-/**
- * struct ZSTD_DCtx - the zstd decompression context
- *
- * When decompressing many times it is recommended to allocate a context just
- * once and reuse it for each successive decompression operation.
- */
-typedef struct ZSTD_DCtx_s ZSTD_DCtx;
-/**
- * ZSTD_initDCtx() - initialize a zstd decompression context
- * @workspace:     The workspace to emplace the context into. It must outlive
- *                 the returned context.
- * @workspaceSize: The size of workspace. Use ZSTD_DCtxWorkspaceBound() to
- *                 determine how large the workspace must be.
- *
- * Return:         A decompression context emplaced into workspace.
- */
-ZSTD_DCtx *ZSTD_initDCtx(void *workspace, size_t workspaceSize);
-
-/**
- * ZSTD_decompressDCtx() - decompress zstd compressed src into dst
- * @ctx:         The decompression context.
- * @dst:         The buffer to decompress src into.
- * @dstCapacity: The size of the destination buffer. Must be at least as large
- *               as the decompressed size. If the caller cannot upper bound the
- *               decompressed size, then it's better to use the streaming API.
- * @src:         The zstd compressed data to decompress. Multiple concatenated
- *               frames and skippable frames are allowed.
- * @srcSize:     The exact size of the data to decompress.
- *
- * Return:       The decompressed size or an error, which can be checked using
- *               ZSTD_isError().
- */
-size_t ZSTD_decompressDCtx(ZSTD_DCtx *ctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-
-/*-************************
- * Simple dictionary API
- **************************/
-
-/**
- * ZSTD_compress_usingDict() - compress src into dst using a dictionary
- * @ctx:         The context. Must have been initialized with a workspace at
- *               least as large as ZSTD_CCtxWorkspaceBound(params.cParams).
- * @dst:         The buffer to compress src into.
- * @dstCapacity: The size of the destination buffer. May be any size, but
- *               ZSTD_compressBound(srcSize) is guaranteed to be large enough.
- * @src:         The data to compress.
- * @srcSize:     The size of the data to compress.
- * @dict:        The dictionary to use for compression.
- * @dictSize:    The size of the dictionary.
- * @params:      The parameters to use for compression. See ZSTD_getParams().
- *
- * Compression using a predefined dictionary. The same dictionary must be used
- * during decompression.
- *
- * Return:       The compressed size or an error, which can be checked using
- *               ZSTD_isError().
- */
-size_t ZSTD_compress_usingDict(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize, const void *dict, size_t dictSize,
-	ZSTD_parameters params);
-
-/**
- * ZSTD_decompress_usingDict() - decompress src into dst using a dictionary
- * @ctx:         The decompression context.
- * @dst:         The buffer to decompress src into.
- * @dstCapacity: The size of the destination buffer. Must be at least as large
- *               as the decompressed size. If the caller cannot upper bound the
- *               decompressed size, then it's better to use the streaming API.
- * @src:         The zstd compressed data to decompress. Multiple concatenated
- *               frames and skippable frames are allowed.
- * @srcSize:     The exact size of the data to decompress.
- * @dict:        The dictionary to use for decompression. The same dictionary
- *               must've been used to compress the data.
- * @dictSize:    The size of the dictionary.
- *
- * Return:       The decompressed size or an error, which can be checked using
- *               ZSTD_isError().
- */
-size_t ZSTD_decompress_usingDict(ZSTD_DCtx *ctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize, const void *dict, size_t dictSize);
-
-/*-**************************
- * Fast dictionary API
- ***************************/
-
-/**
- * ZSTD_CDictWorkspaceBound() - memory needed to initialize a ZSTD_CDict
- * @cParams: The compression parameters to be used for compression.
+ * zstd_init_cctx() - initialize a zstd compression context
+ * @workspace:      The workspace to emplace the context into. It must outlive
+ *                  the returned context.
+ * @workspace_size: The size of workspace. Use zstd_cctx_workspace_bound() to
+ *                  determine how large the workspace must be.
  *
- * Return:   A lower bound on the size of the workspace that is passed to
- *           ZSTD_initCDict().
- */
-size_t ZSTD_CDictWorkspaceBound(ZSTD_compressionParameters cParams);
-
-/**
- * struct ZSTD_CDict - a digested dictionary to be used for compression
+ * Return:          A zstd compression context or NULL on error.
  */
-typedef struct ZSTD_CDict_s ZSTD_CDict;
+zstd_cctx *zstd_init_cctx(void *workspace, size_t workspace_size);
 
 /**
- * ZSTD_initCDict() - initialize a digested dictionary for compression
- * @dictBuffer:    The dictionary to digest. The buffer is referenced by the
- *                 ZSTD_CDict so it must outlive the returned ZSTD_CDict.
- * @dictSize:      The size of the dictionary.
- * @params:        The parameters to use for compression. See ZSTD_getParams().
- * @workspace:     The workspace. It must outlive the returned ZSTD_CDict.
- * @workspaceSize: The workspace size. Must be at least
- *                 ZSTD_CDictWorkspaceBound(params.cParams).
+ * zstd_compress_cctx() - compress src into dst with the initialized parameters
+ * @cctx:         The context. Must have been initialized with zstd_init_cctx().
+ * @dst:          The buffer to compress src into.
+ * @dst_capacity: The size of the destination buffer. May be any size, but
+ *                ZSTD_compressBound(srcSize) is guaranteed to be large enough.
+ * @src:          The data to compress.
+ * @src_size:     The size of the data to compress.
+ * @parameters:   The compression parameters to be used.
  *
- * When compressing multiple messages / blocks with the same dictionary it is
- * recommended to load it just once. The ZSTD_CDict merely references the
- * dictBuffer, so it must outlive the returned ZSTD_CDict.
- *
- * Return:         The digested dictionary emplaced into workspace.
+ * Return:        The compressed size or an error, which can be checked using
+ *                zstd_is_error().
  */
-ZSTD_CDict *ZSTD_initCDict(const void *dictBuffer, size_t dictSize,
-	ZSTD_parameters params, void *workspace, size_t workspaceSize);
+size_t zstd_compress_cctx(zstd_cctx *cctx, void *dst, size_t dst_capacity,
+	const void *src, size_t src_size, const zstd_parameters *parameters);
 
-/**
- * ZSTD_compress_usingCDict() - compress src into dst using a ZSTD_CDict
- * @ctx:         The context. Must have been initialized with a workspace at
- *               least as large as ZSTD_CCtxWorkspaceBound(cParams) where
- *               cParams are the compression parameters used to initialize the
- *               cdict.
- * @dst:         The buffer to compress src into.
- * @dstCapacity: The size of the destination buffer. May be any size, but
- *               ZSTD_compressBound(srcSize) is guaranteed to be large enough.
- * @src:         The data to compress.
- * @srcSize:     The size of the data to compress.
- * @cdict:       The digested dictionary to use for compression.
- * @params:      The parameters to use for compression. See ZSTD_getParams().
- *
- * Compression using a digested dictionary. The same dictionary must be used
- * during decompression.
- *
- * Return:       The compressed size or an error, which can be checked using
- *               ZSTD_isError().
- */
-size_t ZSTD_compress_usingCDict(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize, const ZSTD_CDict *cdict);
+/* ======   Single-pass Decompression   ====== */
 
+typedef ZSTD_DCtx zstd_dctx;
 
 /**
- * ZSTD_DDictWorkspaceBound() - memory needed to initialize a ZSTD_DDict
+ * zstd_dctx_workspace_bound() - max memory needed to initialize a zstd_dctx
  *
- * Return:  A lower bound on the size of the workspace that is passed to
- *          ZSTD_initDDict().
- */
-size_t ZSTD_DDictWorkspaceBound(void);
-
-/**
- * struct ZSTD_DDict - a digested dictionary to be used for decompression
+ * Return: A lower bound on the size of the workspace that is passed to
+ *         zstd_init_dctx().
  */
-typedef struct ZSTD_DDict_s ZSTD_DDict;
+size_t zstd_dctx_workspace_bound(void);
 
 /**
- * ZSTD_initDDict() - initialize a digested dictionary for decompression
- * @dictBuffer:    The dictionary to digest. The buffer is referenced by the
- *                 ZSTD_DDict so it must outlive the returned ZSTD_DDict.
- * @dictSize:      The size of the dictionary.
- * @workspace:     The workspace. It must outlive the returned ZSTD_DDict.
- * @workspaceSize: The workspace size. Must be at least
- *                 ZSTD_DDictWorkspaceBound().
- *
- * When decompressing multiple messages / blocks with the same dictionary it is
- * recommended to load it just once. The ZSTD_DDict merely references the
- * dictBuffer, so it must outlive the returned ZSTD_DDict.
+ * zstd_init_dctx() - initialize a zstd decompression context
+ * @workspace:      The workspace to emplace the context into. It must outlive
+ *                  the returned context.
+ * @workspace_size: The size of workspace. Use zstd_dctx_workspace_bound() to
+ *                  determine how large the workspace must be.
  *
- * Return:         The digested dictionary emplaced into workspace.
+ * Return:          A zstd decompression context or NULL on error.
  */
-ZSTD_DDict *ZSTD_initDDict(const void *dictBuffer, size_t dictSize,
-	void *workspace, size_t workspaceSize);
+zstd_dctx *zstd_init_dctx(void *workspace, size_t workspace_size);
 
 /**
- * ZSTD_decompress_usingDDict() - decompress src into dst using a ZSTD_DDict
- * @ctx:         The decompression context.
- * @dst:         The buffer to decompress src into.
- * @dstCapacity: The size of the destination buffer. Must be at least as large
- *               as the decompressed size. If the caller cannot upper bound the
- *               decompressed size, then it's better to use the streaming API.
- * @src:         The zstd compressed data to decompress. Multiple concatenated
- *               frames and skippable frames are allowed.
- * @srcSize:     The exact size of the data to decompress.
- * @ddict:       The digested dictionary to use for decompression. The same
- *               dictionary must've been used to compress the data.
+ * zstd_decompress_dctx() - decompress zstd compressed src into dst
+ * @dctx:         The decompression context.
+ * @dst:          The buffer to decompress src into.
+ * @dst_capacity: The size of the destination buffer. Must be at least as large
+ *                as the decompressed size. If the caller cannot upper bound the
+ *                decompressed size, then it's better to use the streaming API.
+ * @src:          The zstd compressed data to decompress. Multiple concatenated
+ *                frames and skippable frames are allowed.
+ * @src_size:     The exact size of the data to decompress.
  *
- * Return:       The decompressed size or an error, which can be checked using
- *               ZSTD_isError().
+ * Return:        The decompressed size or an error, which can be checked using
+ *                zstd_is_error().
  */
-size_t ZSTD_decompress_usingDDict(ZSTD_DCtx *dctx, void *dst,
-	size_t dstCapacity, const void *src, size_t srcSize,
-	const ZSTD_DDict *ddict);
+size_t zstd_decompress_dctx(zstd_dctx *dctx, void *dst, size_t dst_capacity,
+	const void *src, size_t src_size);
 
-
-/*-**************************
- * Streaming
- ***************************/
+/* ======   Streaming Buffers   ====== */
 
 /**
- * struct ZSTD_inBuffer - input buffer for streaming
+ * struct zstd_in_buffer - input buffer for streaming
  * @src:  Start of the input buffer.
  * @size: Size of the input buffer.
  * @pos:  Position where reading stopped. Will be updated.
  *        Necessarily 0 <= pos <= size.
+ *
+ * See zstd_lib.h.
  */
-typedef struct ZSTD_inBuffer_s {
-	const void *src;
-	size_t size;
-	size_t pos;
-} ZSTD_inBuffer;
+typedef ZSTD_inBuffer zstd_in_buffer;
 
 /**
- * struct ZSTD_outBuffer - output buffer for streaming
+ * struct zstd_out_buffer - output buffer for streaming
  * @dst:  Start of the output buffer.
  * @size: Size of the output buffer.
  * @pos:  Position where writing stopped. Will be updated.
  *        Necessarily 0 <= pos <= size.
+ *
+ * See zstd_lib.h.
  */
-typedef struct ZSTD_outBuffer_s {
-	void *dst;
-	size_t size;
-	size_t pos;
-} ZSTD_outBuffer;
+typedef ZSTD_outBuffer zstd_out_buffer;
 
+/* ======   Streaming Compression   ====== */
 
-
-/*-*****************************************************************************
- * Streaming compression - HowTo
- *
- * A ZSTD_CStream object is required to track streaming operation.
- * Use ZSTD_initCStream() to initialize a ZSTD_CStream object.
- * ZSTD_CStream objects can be reused multiple times on consecutive compression
- * operations. It is recommended to re-use ZSTD_CStream in situations where many
- * streaming operations will be achieved consecutively. Use one separate
- * ZSTD_CStream per thread for parallel execution.
- *
- * Use ZSTD_compressStream() repetitively to consume input stream.
- * The function will automatically update both `pos` fields.
- * Note that it may not consume the entire input, in which case `pos < size`,
- * and it's up to the caller to present again remaining data.
- * It returns a hint for the preferred number of bytes to use as an input for
- * the next function call.
- *
- * At any moment, it's possible to flush whatever data remains within internal
- * buffer, using ZSTD_flushStream(). `output->pos` will be updated. There might
- * still be some content left within the internal buffer if `output->size` is
- * too small. It returns the number of bytes left in the internal buffer and
- * must be called until it returns 0.
- *
- * ZSTD_endStream() instructs to finish a frame. It will perform a flush and
- * write frame epilogue. The epilogue is required for decoders to consider a
- * frame completed. Similar to ZSTD_flushStream(), it may not be able to flush
- * the full content if `output->size` is too small. In which case, call again
- * ZSTD_endStream() to complete the flush. It returns the number of bytes left
- * in the internal buffer and must be called until it returns 0.
- ******************************************************************************/
+typedef ZSTD_CStream zstd_cstream;
 
 /**
- * ZSTD_CStreamWorkspaceBound() - memory needed to initialize a ZSTD_CStream
- * @cParams: The compression parameters to be used for compression.
+ * zstd_cstream_workspace_bound() - memory needed to initialize a zstd_cstream
+ * @cparams: The compression parameters to be used for compression.
  *
  * Return:   A lower bound on the size of the workspace that is passed to
- *           ZSTD_initCStream() and ZSTD_initCStream_usingCDict().
- */
-size_t ZSTD_CStreamWorkspaceBound(ZSTD_compressionParameters cParams);
-
-/**
- * struct ZSTD_CStream - the zstd streaming compression context
- */
-typedef struct ZSTD_CStream_s ZSTD_CStream;
-
-/*===== ZSTD_CStream management functions =====*/
-/**
- * ZSTD_initCStream() - initialize a zstd streaming compression context
- * @params:         The zstd compression parameters.
- * @pledgedSrcSize: If params.fParams.contentSizeFlag == 1 then the caller must
- *                  pass the source size (zero means empty source). Otherwise,
- *                  the caller may optionally pass the source size, or zero if
- *                  unknown.
- * @workspace:      The workspace to emplace the context into. It must outlive
- *                  the returned context.
- * @workspaceSize:  The size of workspace.
- *                  Use ZSTD_CStreamWorkspaceBound(params.cParams) to determine
- *                  how large the workspace must be.
- *
- * Return:          The zstd streaming compression context.
+ *           zstd_init_cstream().
  */
-ZSTD_CStream *ZSTD_initCStream(ZSTD_parameters params,
-	unsigned long long pledgedSrcSize, void *workspace,
-	size_t workspaceSize);
+size_t zstd_cstream_workspace_bound(const zstd_compression_parameters *cparams);
 
 /**
- * ZSTD_initCStream_usingCDict() - initialize a streaming compression context
- * @cdict:          The digested dictionary to use for compression.
- * @pledgedSrcSize: Optionally the source size, or zero if unknown.
- * @workspace:      The workspace to emplace the context into. It must outlive
- *                  the returned context.
- * @workspaceSize:  The size of workspace. Call ZSTD_CStreamWorkspaceBound()
- *                  with the cParams used to initialize the cdict to determine
- *                  how large the workspace must be.
+ * zstd_init_cstream() - initialize a zstd streaming compression context
+ * @parameters        The zstd parameters to use for compression.
+ * @pledged_src_size: If params.fParams.contentSizeFlag == 1 then the caller
+ *                    must pass the source size (zero means empty source).
+ *                    Otherwise, the caller may optionally pass the source
+ *                    size, or zero if unknown.
+ * @workspace:        The workspace to emplace the context into. It must outlive
+ *                    the returned context.
+ * @workspace_size:   The size of workspace.
+ *                    Use zstd_cstream_workspace_bound(params->cparams) to
+ *                    determine how large the workspace must be.
  *
- * Return:          The zstd streaming compression context.
+ * Return:            The zstd streaming compression context or NULL on error.
  */
-ZSTD_CStream *ZSTD_initCStream_usingCDict(const ZSTD_CDict *cdict,
-	unsigned long long pledgedSrcSize, void *workspace,
-	size_t workspaceSize);
+zstd_cstream *zstd_init_cstream(const zstd_parameters *parameters,
+	unsigned long long pledged_src_size, void *workspace, size_t workspace_size);
 
-/*===== Streaming compression functions =====*/
 /**
- * ZSTD_resetCStream() - reset the context using parameters from creation
- * @zcs:            The zstd streaming compression context to reset.
- * @pledgedSrcSize: Optionally the source size, or zero if unknown.
+ * zstd_reset_cstream() - reset the context using parameters from creation
+ * @cstream:          The zstd streaming compression context to reset.
+ * @pledged_src_size: Optionally the source size, or zero if unknown.
  *
  * Resets the context using the parameters from creation. Skips dictionary
- * loading, since it can be reused. If `pledgedSrcSize` is non-zero the frame
+ * loading, since it can be reused. If `pledged_src_size` is non-zero the frame
  * content size is always written into the frame header.
  *
- * Return:          Zero or an error, which can be checked using ZSTD_isError().
+ * Return:            Zero or an error, which can be checked using
+ *                    zstd_is_error().
  */
-size_t ZSTD_resetCStream(ZSTD_CStream *zcs, unsigned long long pledgedSrcSize);
+size_t zstd_reset_cstream(zstd_cstream *cstream,
+	unsigned long long pledged_src_size);
+
 /**
- * ZSTD_compressStream() - streaming compress some of input into output
- * @zcs:    The zstd streaming compression context.
- * @output: Destination buffer. `output->pos` is updated to indicate how much
- *          compressed data was written.
- * @input:  Source buffer. `input->pos` is updated to indicate how much data was
- *          read. Note that it may not consume the entire input, in which case
- *          `input->pos < input->size`, and it's up to the caller to present
- *          remaining data again.
+ * zstd_compress_stream() - streaming compress some of input into output
+ * @cstream: The zstd streaming compression context.
+ * @output:  Destination buffer. `output->pos` is updated to indicate how much
+ *           compressed data was written.
+ * @input:   Source buffer. `input->pos` is updated to indicate how much data
+ *           was read. Note that it may not consume the entire input, in which
+ *           case `input->pos < input->size`, and it's up to the caller to
+ *           present remaining data again.
  *
  * The `input` and `output` buffers may be any size. Guaranteed to make some
  * forward progress if `input` and `output` are not empty.
  *
- * Return:  A hint for the number of bytes to use as the input for the next
- *          function call or an error, which can be checked using
- *          ZSTD_isError().
+ * Return:   A hint for the number of bytes to use as the input for the next
+ *           function call or an error, which can be checked using
+ *           zstd_is_error().
  */
-size_t ZSTD_compressStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output,
-	ZSTD_inBuffer *input);
+size_t zstd_compress_stream(zstd_cstream *cstream, zstd_out_buffer *output,
+	zstd_in_buffer *input);
+
 /**
- * ZSTD_flushStream() - flush internal buffers into output
- * @zcs:    The zstd streaming compression context.
- * @output: Destination buffer. `output->pos` is updated to indicate how much
- *          compressed data was written.
+ * zstd_flush_stream() - flush internal buffers into output
+ * @cstream: The zstd streaming compression context.
+ * @output:  Destination buffer. `output->pos` is updated to indicate how much
+ *           compressed data was written.
  *
- * ZSTD_flushStream() must be called until it returns 0, meaning all the data
- * has been flushed. Since ZSTD_flushStream() causes a block to be ended,
+ * zstd_flush_stream() must be called until it returns 0, meaning all the data
+ * has been flushed. Since zstd_flush_stream() causes a block to be ended,
  * calling it too often will degrade the compression ratio.
  *
- * Return:  The number of bytes still present within internal buffers or an
- *          error, which can be checked using ZSTD_isError().
+ * Return:   The number of bytes still present within internal buffers or an
+ *           error, which can be checked using zstd_is_error().
  */
-size_t ZSTD_flushStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output);
-/**
- * ZSTD_endStream() - flush internal buffers into output and end the frame
- * @zcs:    The zstd streaming compression context.
- * @output: Destination buffer. `output->pos` is updated to indicate how much
- *          compressed data was written.
- *
- * ZSTD_endStream() must be called until it returns 0, meaning all the data has
- * been flushed and the frame epilogue has been written.
- *
- * Return:  The number of bytes still present within internal buffers or an
- *          error, which can be checked using ZSTD_isError().
- */
-size_t ZSTD_endStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output);
+size_t zstd_flush_stream(zstd_cstream *cstream, zstd_out_buffer *output);
 
 /**
- * ZSTD_CStreamInSize() - recommended size for the input buffer
- *
- * Return: The recommended size for the input buffer.
- */
-size_t ZSTD_CStreamInSize(void);
-/**
- * ZSTD_CStreamOutSize() - recommended size for the output buffer
+ * zstd_end_stream() - flush internal buffers into output and end the frame
+ * @cstream: The zstd streaming compression context.
+ * @output:  Destination buffer. `output->pos` is updated to indicate how much
+ *           compressed data was written.
  *
- * When the output buffer is at least this large, it is guaranteed to be large
- * enough to flush at least one complete compressed block.
+ * zstd_end_stream() must be called until it returns 0, meaning all the data has
+ * been flushed and the frame epilogue has been written.
  *
- * Return: The recommended size for the output buffer.
+ * Return:   The number of bytes still present within internal buffers or an
+ *           error, which can be checked using zstd_is_error().
  */
-size_t ZSTD_CStreamOutSize(void);
+size_t zstd_end_stream(zstd_cstream *cstream, zstd_out_buffer *output);
 
+/* ======   Streaming Decompression   ====== */
 
-
-/*-*****************************************************************************
- * Streaming decompression - HowTo
- *
- * A ZSTD_DStream object is required to track streaming operations.
- * Use ZSTD_initDStream() to initialize a ZSTD_DStream object.
- * ZSTD_DStream objects can be re-used multiple times.
- *
- * Use ZSTD_decompressStream() repetitively to consume your input.
- * The function will update both `pos` fields.
- * If `input->pos < input->size`, some input has not been consumed.
- * It's up to the caller to present again remaining data.
- * If `output->pos < output->size`, decoder has flushed everything it could.
- * Returns 0 iff a frame is completely decoded and fully flushed.
- * Otherwise it returns a suggested next input size that will never load more
- * than the current frame.
- ******************************************************************************/
+typedef ZSTD_DStream zstd_dstream;
 
 /**
- * ZSTD_DStreamWorkspaceBound() - memory needed to initialize a ZSTD_DStream
- * @maxWindowSize: The maximum window size allowed for compressed frames.
+ * zstd_dstream_workspace_bound() - memory needed to initialize a zstd_dstream
+ * @max_window_size: The maximum window size allowed for compressed frames.
  *
- * Return:         A lower bound on the size of the workspace that is passed to
- *                 ZSTD_initDStream() and ZSTD_initDStream_usingDDict().
+ * Return:           A lower bound on the size of the workspace that is passed
+ *                   to zstd_init_dstream().
  */
-size_t ZSTD_DStreamWorkspaceBound(size_t maxWindowSize);
+size_t zstd_dstream_workspace_bound(size_t max_window_size);
 
 /**
- * struct ZSTD_DStream - the zstd streaming decompression context
- */
-typedef struct ZSTD_DStream_s ZSTD_DStream;
-/*===== ZSTD_DStream management functions =====*/
-/**
- * ZSTD_initDStream() - initialize a zstd streaming decompression context
- * @maxWindowSize: The maximum window size allowed for compressed frames.
- * @workspace:     The workspace to emplace the context into. It must outlive
- *                 the returned context.
- * @workspaceSize: The size of workspace.
- *                 Use ZSTD_DStreamWorkspaceBound(maxWindowSize) to determine
- *                 how large the workspace must be.
- *
- * Return:         The zstd streaming decompression context.
- */
-ZSTD_DStream *ZSTD_initDStream(size_t maxWindowSize, void *workspace,
-	size_t workspaceSize);
-/**
- * ZSTD_initDStream_usingDDict() - initialize streaming decompression context
- * @maxWindowSize: The maximum window size allowed for compressed frames.
- * @ddict:         The digested dictionary to use for decompression.
- * @workspace:     The workspace to emplace the context into. It must outlive
- *                 the returned context.
- * @workspaceSize: The size of workspace.
- *                 Use ZSTD_DStreamWorkspaceBound(maxWindowSize) to determine
- *                 how large the workspace must be.
+ * zstd_init_dstream() - initialize a zstd streaming decompression context
+ * @max_window_size: The maximum window size allowed for compressed frames.
+ * @workspace:       The workspace to emplace the context into. It must outlive
+ *                   the returned context.
+ * @workspaceSize:   The size of workspace.
+ *                   Use zstd_dstream_workspace_bound(max_window_size) to
+ *                   determine how large the workspace must be.
  *
- * Return:         The zstd streaming decompression context.
+ * Return:           The zstd streaming decompression context.
  */
-ZSTD_DStream *ZSTD_initDStream_usingDDict(size_t maxWindowSize,
-	const ZSTD_DDict *ddict, void *workspace, size_t workspaceSize);
+zstd_dstream *zstd_init_dstream(size_t max_window_size, void *workspace,
+	size_t workspace_size);
 
-/*===== Streaming decompression functions =====*/
 /**
- * ZSTD_resetDStream() - reset the context using parameters from creation
- * @zds:   The zstd streaming decompression context to reset.
+ * zstd_reset_dstream() - reset the context using parameters from creation
+ * @dstream: The zstd streaming decompression context to reset.
  *
  * Resets the context using the parameters from creation. Skips dictionary
  * loading, since it can be reused.
  *
- * Return: Zero or an error, which can be checked using ZSTD_isError().
+ * Return:   Zero or an error, which can be checked using zstd_is_error().
  */
-size_t ZSTD_resetDStream(ZSTD_DStream *zds);
+size_t zstd_reset_dstream(zstd_dstream *dstream);
+
 /**
- * ZSTD_decompressStream() - streaming decompress some of input into output
- * @zds:    The zstd streaming decompression context.
- * @output: Destination buffer. `output.pos` is updated to indicate how much
- *          decompressed data was written.
- * @input:  Source buffer. `input.pos` is updated to indicate how much data was
- *          read. Note that it may not consume the entire input, in which case
- *          `input.pos < input.size`, and it's up to the caller to present
- *          remaining data again.
+ * zstd_decompress_stream() - streaming decompress some of input into output
+ * @dstream: The zstd streaming decompression context.
+ * @output:  Destination buffer. `output.pos` is updated to indicate how much
+ *           decompressed data was written.
+ * @input:   Source buffer. `input.pos` is updated to indicate how much data was
+ *           read. Note that it may not consume the entire input, in which case
+ *           `input.pos < input.size`, and it's up to the caller to present
+ *           remaining data again.
  *
  * The `input` and `output` buffers may be any size. Guaranteed to make some
  * forward progress if `input` and `output` are not empty.
- * ZSTD_decompressStream() will not consume the last byte of the frame until
+ * zstd_decompress_stream() will not consume the last byte of the frame until
  * the entire frame is flushed.
  *
- * Return:  Returns 0 iff a frame is completely decoded and fully flushed.
- *          Otherwise returns a hint for the number of bytes to use as the input
- *          for the next function call or an error, which can be checked using
- *          ZSTD_isError(). The size hint will never load more than the frame.
+ * Return:   Returns 0 iff a frame is completely decoded and fully flushed.
+ *           Otherwise returns a hint for the number of bytes to use as the
+ *           input for the next function call or an error, which can be checked
+ *           using zstd_is_error(). The size hint will never load more than the
+ *           frame.
  */
-size_t ZSTD_decompressStream(ZSTD_DStream *zds, ZSTD_outBuffer *output,
-	ZSTD_inBuffer *input);
+size_t zstd_decompress_stream(zstd_dstream *dstream, zstd_out_buffer *output,
+	zstd_in_buffer *input);
 
-/**
- * ZSTD_DStreamInSize() - recommended size for the input buffer
- *
- * Return: The recommended size for the input buffer.
- */
-size_t ZSTD_DStreamInSize(void);
-/**
- * ZSTD_DStreamOutSize() - recommended size for the output buffer
- *
- * When the output buffer is at least this large, it is guaranteed to be large
- * enough to flush at least one complete decompressed block.
- *
- * Return: The recommended size for the output buffer.
- */
-size_t ZSTD_DStreamOutSize(void);
-
-
-/* --- Constants ---*/
-#define ZSTD_MAGICNUMBER            0xFD2FB528   /* >= v0.8.0 */
-#define ZSTD_MAGIC_SKIPPABLE_START  0x184D2A50U
-
-#define ZSTD_CONTENTSIZE_UNKNOWN (0ULL - 1)
-#define ZSTD_CONTENTSIZE_ERROR   (0ULL - 2)
-
-#define ZSTD_WINDOWLOG_MAX_32  27
-#define ZSTD_WINDOWLOG_MAX_64  27
-#define ZSTD_WINDOWLOG_MAX \
-	((unsigned int)(sizeof(size_t) == 4 \
-		? ZSTD_WINDOWLOG_MAX_32 \
-		: ZSTD_WINDOWLOG_MAX_64))
-#define ZSTD_WINDOWLOG_MIN 10
-#define ZSTD_HASHLOG_MAX ZSTD_WINDOWLOG_MAX
-#define ZSTD_HASHLOG_MIN        6
-#define ZSTD_CHAINLOG_MAX     (ZSTD_WINDOWLOG_MAX+1)
-#define ZSTD_CHAINLOG_MIN      ZSTD_HASHLOG_MIN
-#define ZSTD_HASHLOG3_MAX      17
-#define ZSTD_SEARCHLOG_MAX    (ZSTD_WINDOWLOG_MAX-1)
-#define ZSTD_SEARCHLOG_MIN      1
-/* only for ZSTD_fast, other strategies are limited to 6 */
-#define ZSTD_SEARCHLENGTH_MAX   7
-/* only for ZSTD_btopt, other strategies are limited to 4 */
-#define ZSTD_SEARCHLENGTH_MIN   3
-#define ZSTD_TARGETLENGTH_MIN   4
-#define ZSTD_TARGETLENGTH_MAX 999
-
-/* for static allocation */
-#define ZSTD_FRAMEHEADERSIZE_MAX 18
-#define ZSTD_FRAMEHEADERSIZE_MIN  6
-static const size_t ZSTD_frameHeaderSize_prefix = 5;
-static const size_t ZSTD_frameHeaderSize_min = ZSTD_FRAMEHEADERSIZE_MIN;
-static const size_t ZSTD_frameHeaderSize_max = ZSTD_FRAMEHEADERSIZE_MAX;
-/* magic number + skippable frame length */
-static const size_t ZSTD_skippableHeaderSize = 8;
-
-
-/*-*************************************
- * Compressed size functions
- **************************************/
-
-/**
- * ZSTD_findFrameCompressedSize() - returns the size of a compressed frame
- * @src:     Source buffer. It should point to the start of a zstd encoded frame
- *           or a skippable frame.
- * @srcSize: The size of the source buffer. It must be at least as large as the
- *           size of the frame.
- *
- * Return:   The compressed size of the frame pointed to by `src` or an error,
- *           which can be check with ZSTD_isError().
- *           Suitable to pass to ZSTD_decompress() or similar functions.
- */
-size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize);
-
-/*-*************************************
- * Decompressed size functions
- **************************************/
-/**
- * ZSTD_getFrameContentSize() - returns the content size in a zstd frame header
- * @src:     It should point to the start of a zstd encoded frame.
- * @srcSize: The size of the source buffer. It must be at least as large as the
- *           frame header. `ZSTD_frameHeaderSize_max` is always large enough.
- *
- * Return:   The frame content size stored in the frame header if known.
- *           `ZSTD_CONTENTSIZE_UNKNOWN` if the content size isn't stored in the
- *           frame header. `ZSTD_CONTENTSIZE_ERROR` on invalid input.
- */
-unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize);
+/* ======   Frame Inspection Functions ====== */
 
 /**
- * ZSTD_findDecompressedSize() - returns decompressed size of a series of frames
- * @src:     It should point to the start of a series of zstd encoded and/or
- *           skippable frames.
- * @srcSize: The exact size of the series of frames.
+ * zstd_find_frame_compressed_size() - returns the size of a compressed frame
+ * @src:      Source buffer. It should point to the start of a zstd encoded
+ *            frame or a skippable frame.
+ * @src_size: The size of the source buffer. It must be at least as large as the
+ *            size of the frame.
  *
- * If any zstd encoded frame in the series doesn't have the frame content size
- * set, `ZSTD_CONTENTSIZE_UNKNOWN` is returned. But frame content size is always
- * set when using ZSTD_compress(). The decompressed size can be very large.
- * If the source is untrusted, the decompressed size could be wrong or
- * intentionally modified. Always ensure the result fits within the
- * application's authorized limits. ZSTD_findDecompressedSize() handles multiple
- * frames, and so it must traverse the input to read each frame header. This is
- * efficient as most of the data is skipped, however it does mean that all frame
- * data must be present and valid.
- *
- * Return:   Decompressed size of all the data contained in the frames if known.
- *           `ZSTD_CONTENTSIZE_UNKNOWN` if the decompressed size is unknown.
- *           `ZSTD_CONTENTSIZE_ERROR` if an error occurred.
- */
-unsigned long long ZSTD_findDecompressedSize(const void *src, size_t srcSize);
-
-/*-*************************************
- * Advanced compression functions
- **************************************/
-/**
- * ZSTD_checkCParams() - ensure parameter values remain within authorized range
- * @cParams: The zstd compression parameters.
- *
- * Return:   Zero or an error, which can be checked using ZSTD_isError().
+ * Return:    The compressed size of the frame pointed to by `src` or an error,
+ *            which can be check with zstd_is_error().
+ *            Suitable to pass to ZSTD_decompress() or similar functions.
  */
-size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams);
+size_t zstd_find_frame_compressed_size(const void *src, size_t src_size);
 
 /**
- * ZSTD_adjustCParams() - optimize parameters for a given srcSize and dictSize
- * @srcSize:  Optionally the estimated source size, or zero if unknown.
- * @dictSize: Optionally the estimated dictionary size, or zero if unknown.
- *
- * Return:    The optimized parameters.
- */
-ZSTD_compressionParameters ZSTD_adjustCParams(
-	ZSTD_compressionParameters cParams, unsigned long long srcSize,
-	size_t dictSize);
-
-/*--- Advanced decompression functions ---*/
-
-/**
- * ZSTD_isFrame() - returns true iff the buffer starts with a valid frame
- * @buffer: The source buffer to check.
- * @size:   The size of the source buffer, must be at least 4 bytes.
- *
- * Return: True iff the buffer starts with a zstd or skippable frame identifier.
- */
-unsigned int ZSTD_isFrame(const void *buffer, size_t size);
-
-/**
- * ZSTD_getDictID_fromDict() - returns the dictionary id stored in a dictionary
- * @dict:     The dictionary buffer.
- * @dictSize: The size of the dictionary buffer.
- *
- * Return:    The dictionary id stored within the dictionary or 0 if the
- *            dictionary is not a zstd dictionary. If it returns 0 the
- *            dictionary can still be loaded as a content-only dictionary.
- */
-unsigned int ZSTD_getDictID_fromDict(const void *dict, size_t dictSize);
-
-/**
- * ZSTD_getDictID_fromDDict() - returns the dictionary id stored in a ZSTD_DDict
- * @ddict: The ddict to find the id of.
- *
- * Return: The dictionary id stored within `ddict` or 0 if the dictionary is not
- *         a zstd dictionary. If it returns 0 `ddict` will be loaded as a
- *         content-only dictionary.
- */
-unsigned int ZSTD_getDictID_fromDDict(const ZSTD_DDict *ddict);
-
-/**
- * ZSTD_getDictID_fromFrame() - returns the dictionary id stored in a zstd frame
- * @src:     Source buffer. It must be a zstd encoded frame.
- * @srcSize: The size of the source buffer. It must be at least as large as the
- *           frame header. `ZSTD_frameHeaderSize_max` is always large enough.
- *
- * Return:   The dictionary id required to decompress the frame stored within
- *           `src` or 0 if the dictionary id could not be decoded. It can return
- *           0 if the frame does not require a dictionary, the dictionary id
- *           wasn't stored in the frame, `src` is not a zstd frame, or `srcSize`
- *           is too small.
- */
-unsigned int ZSTD_getDictID_fromFrame(const void *src, size_t srcSize);
-
-/**
- * struct ZSTD_frameParams - zstd frame parameters stored in the frame header
- * @frameContentSize: The frame content size, or 0 if not present.
+ * struct zstd_frame_params - zstd frame parameters stored in the frame header
+ * @frameContentSize: The frame content size, or ZSTD_CONTENTSIZE_UNKNOWN if not
+ *                    present.
  * @windowSize:       The window size, or 0 if the frame is a skippable frame.
+ * @blockSizeMax:     The maximum block size.
+ * @frameType:        The frame type (zstd or skippable)
+ * @headerSize:       The size of the frame header.
  * @dictID:           The dictionary id, or 0 if not present.
  * @checksumFlag:     Whether a checksum was used.
+ *
+ * See zstd_lib.h.
  */
-typedef struct {
-	unsigned long long frameContentSize;
-	unsigned int windowSize;
-	unsigned int dictID;
-	unsigned int checksumFlag;
-} ZSTD_frameParams;
+typedef ZSTD_frameHeader zstd_frame_header;
 
 /**
- * ZSTD_getFrameParams() - extracts parameters from a zstd or skippable frame
- * @fparamsPtr: On success the frame parameters are written here.
- * @src:        The source buffer. It must point to a zstd or skippable frame.
- * @srcSize:    The size of the source buffer. `ZSTD_frameHeaderSize_max` is
- *              always large enough to succeed.
+ * zstd_get_frame_header() - extracts parameters from a zstd or skippable frame
+ * @params:   On success the frame parameters are written here.
+ * @src:      The source buffer. It must point to a zstd or skippable frame.
+ * @src_size: The size of the source buffer.
  *
- * Return:      0 on success. If more data is required it returns how many bytes
- *              must be provided to make forward progress. Otherwise it returns
- *              an error, which can be checked using ZSTD_isError().
+ * Return:    0 on success. If more data is required it returns how many bytes
+ *            must be provided to make forward progress. Otherwise it returns
+ *            an error, which can be checked using zstd_is_error().
  */
-size_t ZSTD_getFrameParams(ZSTD_frameParams *fparamsPtr, const void *src,
-	size_t srcSize);
-
-/*-*****************************************************************************
- * Buffer-less and synchronous inner streaming functions
- *
- * This is an advanced API, giving full control over buffer management, for
- * users which need direct control over memory.
- * But it's also a complex one, with many restrictions (documented below).
- * Prefer using normal streaming API for an easier experience
- ******************************************************************************/
-
-/*-*****************************************************************************
- * Buffer-less streaming compression (synchronous mode)
- *
- * A ZSTD_CCtx object is required to track streaming operations.
- * Use ZSTD_initCCtx() to initialize a context.
- * ZSTD_CCtx object can be re-used multiple times within successive compression
- * operations.
- *
- * Start by initializing a context.
- * Use ZSTD_compressBegin(), or ZSTD_compressBegin_usingDict() for dictionary
- * compression,
- * or ZSTD_compressBegin_advanced(), for finer parameter control.
- * It's also possible to duplicate a reference context which has already been
- * initialized, using ZSTD_copyCCtx()
- *
- * Then, consume your input using ZSTD_compressContinue().
- * There are some important considerations to keep in mind when using this
- * advanced function :
- * - ZSTD_compressContinue() has no internal buffer. It uses externally provided
- *   buffer only.
- * - Interface is synchronous : input is consumed entirely and produce 1+
- *   (or more) compressed blocks.
- * - Caller must ensure there is enough space in `dst` to store compressed data
- *   under worst case scenario. Worst case evaluation is provided by
- *   ZSTD_compressBound().
- *   ZSTD_compressContinue() doesn't guarantee recover after a failed
- *   compression.
- * - ZSTD_compressContinue() presumes prior input ***is still accessible and
- *   unmodified*** (up to maximum distance size, see WindowLog).
- *   It remembers all previous contiguous blocks, plus one separated memory
- *   segment (which can itself consists of multiple contiguous blocks)
- * - ZSTD_compressContinue() detects that prior input has been overwritten when
- *   `src` buffer overlaps. In which case, it will "discard" the relevant memory
- *   section from its history.
- *
- * Finish a frame with ZSTD_compressEnd(), which will write the last block(s)
- * and optional checksum. It's possible to use srcSize==0, in which case, it
- * will write a final empty block to end the frame. Without last block mark,
- * frames will be considered unfinished (corrupted) by decoders.
- *
- * `ZSTD_CCtx` object can be re-used (ZSTD_compressBegin()) to compress some new
- * frame.
- ******************************************************************************/
-
-/*=====   Buffer-less streaming compression functions  =====*/
-size_t ZSTD_compressBegin(ZSTD_CCtx *cctx, int compressionLevel);
-size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx *cctx, const void *dict,
-	size_t dictSize, int compressionLevel);
-size_t ZSTD_compressBegin_advanced(ZSTD_CCtx *cctx, const void *dict,
-	size_t dictSize, ZSTD_parameters params,
-	unsigned long long pledgedSrcSize);
-size_t ZSTD_copyCCtx(ZSTD_CCtx *cctx, const ZSTD_CCtx *preparedCCtx,
-	unsigned long long pledgedSrcSize);
-size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx *cctx, const ZSTD_CDict *cdict,
-	unsigned long long pledgedSrcSize);
-size_t ZSTD_compressContinue(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-size_t ZSTD_compressEnd(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-
-
-
-/*-*****************************************************************************
- * Buffer-less streaming decompression (synchronous mode)
- *
- * A ZSTD_DCtx object is required to track streaming operations.
- * Use ZSTD_initDCtx() to initialize a context.
- * A ZSTD_DCtx object can be re-used multiple times.
- *
- * First typical operation is to retrieve frame parameters, using
- * ZSTD_getFrameParams(). It fills a ZSTD_frameParams structure which provide
- * important information to correctly decode the frame, such as the minimum
- * rolling buffer size to allocate to decompress data (`windowSize`), and the
- * dictionary ID used.
- * Note: content size is optional, it may not be present. 0 means unknown.
- * Note that these values could be wrong, either because of data malformation,
- * or because an attacker is spoofing deliberate false information. As a
- * consequence, check that values remain within valid application range,
- * especially `windowSize`, before allocation. Each application can set its own
- * limit, depending on local restrictions. For extended interoperability, it is
- * recommended to support at least 8 MB.
- * Frame parameters are extracted from the beginning of the compressed frame.
- * Data fragment must be large enough to ensure successful decoding, typically
- * `ZSTD_frameHeaderSize_max` bytes.
- * Result: 0: successful decoding, the `ZSTD_frameParams` structure is filled.
- *        >0: `srcSize` is too small, provide at least this many bytes.
- *        errorCode, which can be tested using ZSTD_isError().
- *
- * Start decompression, with ZSTD_decompressBegin() or
- * ZSTD_decompressBegin_usingDict(). Alternatively, you can copy a prepared
- * context, using ZSTD_copyDCtx().
- *
- * Then use ZSTD_nextSrcSizeToDecompress() and ZSTD_decompressContinue()
- * alternatively.
- * ZSTD_nextSrcSizeToDecompress() tells how many bytes to provide as 'srcSize'
- * to ZSTD_decompressContinue().
- * ZSTD_decompressContinue() requires this _exact_ amount of bytes, or it will
- * fail.
- *
- * The result of ZSTD_decompressContinue() is the number of bytes regenerated
- * within 'dst' (necessarily <= dstCapacity). It can be zero, which is not an
- * error; it just means ZSTD_decompressContinue() has decoded some metadata
- * item. It can also be an error code, which can be tested with ZSTD_isError().
- *
- * ZSTD_decompressContinue() needs previous data blocks during decompression, up
- * to `windowSize`. They should preferably be located contiguously, prior to
- * current block. Alternatively, a round buffer of sufficient size is also
- * possible. Sufficient size is determined by frame parameters.
- * ZSTD_decompressContinue() is very sensitive to contiguity, if 2 blocks don't
- * follow each other, make sure that either the compressor breaks contiguity at
- * the same place, or that previous contiguous segment is large enough to
- * properly handle maximum back-reference.
- *
- * A frame is fully decoded when ZSTD_nextSrcSizeToDecompress() returns zero.
- * Context can then be reset to start a new decompression.
- *
- * Note: it's possible to know if next input to present is a header or a block,
- * using ZSTD_nextInputType(). This information is not required to properly
- * decode a frame.
- *
- * == Special case: skippable frames ==
- *
- * Skippable frames allow integration of user-defined data into a flow of
- * concatenated frames. Skippable frames will be ignored (skipped) by a
- * decompressor. The format of skippable frames is as follows:
- * a) Skippable frame ID - 4 Bytes, Little endian format, any value from
- *    0x184D2A50 to 0x184D2A5F
- * b) Frame Size - 4 Bytes, Little endian format, unsigned 32-bits
- * c) Frame Content - any content (User Data) of length equal to Frame Size
- * For skippable frames ZSTD_decompressContinue() always returns 0.
- * For skippable frames ZSTD_getFrameParams() returns fparamsPtr->windowLog==0
- * what means that a frame is skippable.
- * Note: If fparamsPtr->frameContentSize==0, it is ambiguous: the frame might
- *       actually be a zstd encoded frame with no content. For purposes of
- *       decompression, it is valid in both cases to skip the frame using
- *       ZSTD_findFrameCompressedSize() to find its size in bytes.
- * It also returns frame size as fparamsPtr->frameContentSize.
- ******************************************************************************/
-
-/*=====   Buffer-less streaming decompression functions  =====*/
-size_t ZSTD_decompressBegin(ZSTD_DCtx *dctx);
-size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx *dctx, const void *dict,
-	size_t dictSize);
-void   ZSTD_copyDCtx(ZSTD_DCtx *dctx, const ZSTD_DCtx *preparedDCtx);
-size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx *dctx);
-size_t ZSTD_decompressContinue(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-typedef enum {
-	ZSTDnit_frameHeader,
-	ZSTDnit_blockHeader,
-	ZSTDnit_block,
-	ZSTDnit_lastBlock,
-	ZSTDnit_checksum,
-	ZSTDnit_skippableFrame
-} ZSTD_nextInputType_e;
-ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx *dctx);
-
-/*-*****************************************************************************
- * Block functions
- *
- * Block functions produce and decode raw zstd blocks, without frame metadata.
- * Frame metadata cost is typically ~18 bytes, which can be non-negligible for
- * very small blocks (< 100 bytes). User will have to take in charge required
- * information to regenerate data, such as compressed and content sizes.
- *
- * A few rules to respect:
- * - Compressing and decompressing require a context structure
- *   + Use ZSTD_initCCtx() and ZSTD_initDCtx()
- * - It is necessary to init context before starting
- *   + compression : ZSTD_compressBegin()
- *   + decompression : ZSTD_decompressBegin()
- *   + variants _usingDict() are also allowed
- *   + copyCCtx() and copyDCtx() work too
- * - Block size is limited, it must be <= ZSTD_getBlockSizeMax()
- *   + If you need to compress more, cut data into multiple blocks
- *   + Consider using the regular ZSTD_compress() instead, as frame metadata
- *     costs become negligible when source size is large.
- * - When a block is considered not compressible enough, ZSTD_compressBlock()
- *   result will be zero. In which case, nothing is produced into `dst`.
- *   + User must test for such outcome and deal directly with uncompressed data
- *   + ZSTD_decompressBlock() doesn't accept uncompressed data as input!!!
- *   + In case of multiple successive blocks, decoder must be informed of
- *     uncompressed block existence to follow proper history. Use
- *     ZSTD_insertBlock() in such a case.
- ******************************************************************************/
-
-/* Define for static allocation */
-#define ZSTD_BLOCKSIZE_ABSOLUTEMAX (128 * 1024)
-/*=====   Raw zstd block functions  =====*/
-size_t ZSTD_getBlockSizeMax(ZSTD_CCtx *cctx);
-size_t ZSTD_compressBlock(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-size_t ZSTD_decompressBlock(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-size_t ZSTD_insertBlock(ZSTD_DCtx *dctx, const void *blockStart,
-	size_t blockSize);
+size_t zstd_get_frame_header(zstd_frame_header *params, const void *src,
+	size_t src_size);
 
-#endif  /* ZSTD_H */
+#endif  /* LINUX_ZSTD_H */
diff --git a/include/linux/zstd_errors.h b/include/linux/zstd_errors.h
new file mode 100644
index 000000000000..58b6dd45a969
--- /dev/null
+++ b/include/linux/zstd_errors.h
@@ -0,0 +1,77 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_ERRORS_H_398273423
+#define ZSTD_ERRORS_H_398273423
+
+
+/*===== dependency =====*/
+#include <linux/types.h>   /* size_t */
+
+
+/* =====   ZSTDERRORLIB_API : control library symbols visibility   ===== */
+#define ZSTDERRORLIB_VISIBILITY 
+#define ZSTDERRORLIB_API ZSTDERRORLIB_VISIBILITY
+
+/*-*********************************************
+ *  Error codes list
+ *-*********************************************
+ *  Error codes _values_ are pinned down since v1.3.1 only.
+ *  Therefore, don't rely on values if you may link to any version < v1.3.1.
+ *
+ *  Only values < 100 are considered stable.
+ *
+ *  note 1 : this API shall be used with static linking only.
+ *           dynamic linking is not yet officially supported.
+ *  note 2 : Prefer relying on the enum than on its value whenever possible
+ *           This is the only supported way to use the error list < v1.3.1
+ *  note 3 : ZSTD_isError() is always correct, whatever the library version.
+ **********************************************/
+typedef enum {
+  ZSTD_error_no_error = 0,
+  ZSTD_error_GENERIC  = 1,
+  ZSTD_error_prefix_unknown                = 10,
+  ZSTD_error_version_unsupported           = 12,
+  ZSTD_error_frameParameter_unsupported    = 14,
+  ZSTD_error_frameParameter_windowTooLarge = 16,
+  ZSTD_error_corruption_detected = 20,
+  ZSTD_error_checksum_wrong      = 22,
+  ZSTD_error_dictionary_corrupted      = 30,
+  ZSTD_error_dictionary_wrong          = 32,
+  ZSTD_error_dictionaryCreation_failed = 34,
+  ZSTD_error_parameter_unsupported   = 40,
+  ZSTD_error_parameter_outOfBound    = 42,
+  ZSTD_error_tableLog_tooLarge       = 44,
+  ZSTD_error_maxSymbolValue_tooLarge = 46,
+  ZSTD_error_maxSymbolValue_tooSmall = 48,
+  ZSTD_error_stage_wrong       = 60,
+  ZSTD_error_init_missing      = 62,
+  ZSTD_error_memory_allocation = 64,
+  ZSTD_error_workSpace_tooSmall= 66,
+  ZSTD_error_dstSize_tooSmall = 70,
+  ZSTD_error_srcSize_wrong    = 72,
+  ZSTD_error_dstBuffer_null   = 74,
+  /* following error codes are __NOT STABLE__, they can be removed or changed in future versions */
+  ZSTD_error_frameIndex_tooLarge = 100,
+  ZSTD_error_seekableIO          = 102,
+  ZSTD_error_dstBuffer_wrong     = 104,
+  ZSTD_error_srcBuffer_wrong     = 105,
+  ZSTD_error_maxCode = 120  /* never EVER use this value directly, it can change in future versions! Use ZSTD_isError() instead */
+} ZSTD_ErrorCode;
+
+/*! ZSTD_getErrorCode() :
+    convert a `size_t` function result into a `ZSTD_ErrorCode` enum type,
+    which can be used to compare with enum list published above */
+ZSTDERRORLIB_API ZSTD_ErrorCode ZSTD_getErrorCode(size_t functionResult);
+ZSTDERRORLIB_API const char* ZSTD_getErrorString(ZSTD_ErrorCode code);   /*< Same as ZSTD_getErrorName, but using a `ZSTD_ErrorCode` enum argument */
+
+
+
+#endif /* ZSTD_ERRORS_H_398273423 */
diff --git a/include/linux/zstd_lib.h b/include/linux/zstd_lib.h
new file mode 100644
index 000000000000..b8c7dbf98390
--- /dev/null
+++ b/include/linux/zstd_lib.h
@@ -0,0 +1,2432 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_H_235446
+#define ZSTD_H_235446
+
+/* ======   Dependency   ======*/
+#include <linux/limits.h>   /* INT_MAX */
+#include <linux/types.h>   /* size_t */
+
+
+/* =====   ZSTDLIB_API : control library symbols visibility   ===== */
+#define ZSTDLIB_VISIBILITY 
+#define ZSTDLIB_API ZSTDLIB_VISIBILITY
+
+
+/* *****************************************************************************
+  Introduction
+
+  zstd, short for Zstandard, is a fast lossless compression algorithm, targeting
+  real-time compression scenarios at zlib-level and better compression ratios.
+  The zstd compression library provides in-memory compression and decompression
+  functions.
+
+  The library supports regular compression levels from 1 up to ZSTD_maxCLevel(),
+  which is currently 22. Levels >= 20, labeled `--ultra`, should be used with
+  caution, as they require more memory. The library also offers negative
+  compression levels, which extend the range of speed vs. ratio preferences.
+  The lower the level, the faster the speed (at the cost of compression).
+
+  Compression can be done in:
+    - a single step (described as Simple API)
+    - a single step, reusing a context (described as Explicit context)
+    - unbounded multiple steps (described as Streaming compression)
+
+  The compression ratio achievable on small data can be highly improved using
+  a dictionary. Dictionary compression can be performed in:
+    - a single step (described as Simple dictionary API)
+    - a single step, reusing a dictionary (described as Bulk-processing
+      dictionary API)
+
+  Advanced experimental functions can be accessed using
+  `#define ZSTD_STATIC_LINKING_ONLY` before including zstd.h.
+
+  Advanced experimental APIs should never be used with a dynamically-linked
+  library. They are not "stable"; their definitions or signatures may change in
+  the future. Only static linking is allowed.
+*******************************************************************************/
+
+/*------   Version   ------*/
+#define ZSTD_VERSION_MAJOR    1
+#define ZSTD_VERSION_MINOR    4
+#define ZSTD_VERSION_RELEASE  10
+#define ZSTD_VERSION_NUMBER  (ZSTD_VERSION_MAJOR *100*100 + ZSTD_VERSION_MINOR *100 + ZSTD_VERSION_RELEASE)
+
+/*! ZSTD_versionNumber() :
+ *  Return runtime library version, the value is (MAJOR*100*100 + MINOR*100 + RELEASE). */
+ZSTDLIB_API unsigned ZSTD_versionNumber(void);
+
+#define ZSTD_LIB_VERSION ZSTD_VERSION_MAJOR.ZSTD_VERSION_MINOR.ZSTD_VERSION_RELEASE
+#define ZSTD_QUOTE(str) #str
+#define ZSTD_EXPAND_AND_QUOTE(str) ZSTD_QUOTE(str)
+#define ZSTD_VERSION_STRING ZSTD_EXPAND_AND_QUOTE(ZSTD_LIB_VERSION)
+
+/*! ZSTD_versionString() :
+ *  Return runtime library version, like "1.4.5". Requires v1.3.0+. */
+ZSTDLIB_API const char* ZSTD_versionString(void);
+
+/* *************************************
+ *  Default constant
+ ***************************************/
+#ifndef ZSTD_CLEVEL_DEFAULT
+#  define ZSTD_CLEVEL_DEFAULT 3
+#endif
+
+/* *************************************
+ *  Constants
+ ***************************************/
+
+/* All magic numbers are supposed read/written to/from files/memory using little-endian convention */
+#define ZSTD_MAGICNUMBER            0xFD2FB528    /* valid since v0.8.0 */
+#define ZSTD_MAGIC_DICTIONARY       0xEC30A437    /* valid since v0.7.0 */
+#define ZSTD_MAGIC_SKIPPABLE_START  0x184D2A50    /* all 16 values, from 0x184D2A50 to 0x184D2A5F, signal the beginning of a skippable frame */
+#define ZSTD_MAGIC_SKIPPABLE_MASK   0xFFFFFFF0
+
+#define ZSTD_BLOCKSIZELOG_MAX  17
+#define ZSTD_BLOCKSIZE_MAX     (1<<ZSTD_BLOCKSIZELOG_MAX)
+
+
+
+/* *************************************
+*  Simple API
+***************************************/
+/*! ZSTD_compress() :
+ *  Compresses `src` content as a single zstd compressed frame into already allocated `dst`.
+ *  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+ *  @return : compressed size written into `dst` (<= `dstCapacity),
+ *            or an error code if it fails (which can be tested using ZSTD_isError()). */
+ZSTDLIB_API size_t ZSTD_compress( void* dst, size_t dstCapacity,
+                            const void* src, size_t srcSize,
+                                  int compressionLevel);
+
+/*! ZSTD_decompress() :
+ *  `compressedSize` : must be the _exact_ size of some number of compressed and/or skippable frames.
+ *  `dstCapacity` is an upper bound of originalSize to regenerate.
+ *  If user cannot imply a maximum upper bound, it's better to use streaming mode to decompress data.
+ *  @return : the number of bytes decompressed into `dst` (<= `dstCapacity`),
+ *            or an errorCode if it fails (which can be tested using ZSTD_isError()). */
+ZSTDLIB_API size_t ZSTD_decompress( void* dst, size_t dstCapacity,
+                              const void* src, size_t compressedSize);
+
+/*! ZSTD_getFrameContentSize() : requires v1.3.0+
+ *  `src` should point to the start of a ZSTD encoded frame.
+ *  `srcSize` must be at least as large as the frame header.
+ *            hint : any size >= `ZSTD_frameHeaderSize_max` is large enough.
+ *  @return : - decompressed size of `src` frame content, if known
+ *            - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
+ *            - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small)
+ *   note 1 : a 0 return value means the frame is valid but "empty".
+ *   note 2 : decompressed size is an optional field, it may not be present, typically in streaming mode.
+ *            When `return==ZSTD_CONTENTSIZE_UNKNOWN`, data to decompress could be any size.
+ *            In which case, it's necessary to use streaming mode to decompress data.
+ *            Optionally, application can rely on some implicit limit,
+ *            as ZSTD_decompress() only needs an upper bound of decompressed size.
+ *            (For example, data could be necessarily cut into blocks <= 16 KB).
+ *   note 3 : decompressed size is always present when compression is completed using single-pass functions,
+ *            such as ZSTD_compress(), ZSTD_compressCCtx() ZSTD_compress_usingDict() or ZSTD_compress_usingCDict().
+ *   note 4 : decompressed size can be very large (64-bits value),
+ *            potentially larger than what local system can handle as a single memory segment.
+ *            In which case, it's necessary to use streaming mode to decompress data.
+ *   note 5 : If source is untrusted, decompressed size could be wrong or intentionally modified.
+ *            Always ensure return value fits within application's authorized limits.
+ *            Each application can set its own limits.
+ *   note 6 : This function replaces ZSTD_getDecompressedSize() */
+#define ZSTD_CONTENTSIZE_UNKNOWN (0ULL - 1)
+#define ZSTD_CONTENTSIZE_ERROR   (0ULL - 2)
+ZSTDLIB_API unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize);
+
+/*! ZSTD_getDecompressedSize() :
+ *  NOTE: This function is now obsolete, in favor of ZSTD_getFrameContentSize().
+ *  Both functions work the same way, but ZSTD_getDecompressedSize() blends
+ *  "empty", "unknown" and "error" results to the same return value (0),
+ *  while ZSTD_getFrameContentSize() gives them separate return values.
+ * @return : decompressed size of `src` frame content _if known and not empty_, 0 otherwise. */
+ZSTDLIB_API unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize);
+
+/*! ZSTD_findFrameCompressedSize() :
+ * `src` should point to the start of a ZSTD frame or skippable frame.
+ * `srcSize` must be >= first frame size
+ * @return : the compressed size of the first frame starting at `src`,
+ *           suitable to pass as `srcSize` to `ZSTD_decompress` or similar,
+ *        or an error code if input is invalid */
+ZSTDLIB_API size_t ZSTD_findFrameCompressedSize(const void* src, size_t srcSize);
+
+
+/*======  Helper functions  ======*/
+#define ZSTD_COMPRESSBOUND(srcSize)   ((srcSize) + ((srcSize)>>8) + (((srcSize) < (128<<10)) ? (((128<<10) - (srcSize)) >> 11) /* margin, from 64 to 0 */ : 0))  /* this formula ensures that bound(A) + bound(B) <= bound(A+B) as long as A and B >= 128 KB */
+ZSTDLIB_API size_t      ZSTD_compressBound(size_t srcSize); /*!< maximum compressed size in worst case single-pass scenario */
+ZSTDLIB_API unsigned    ZSTD_isError(size_t code);          /*!< tells if a `size_t` function result is an error code */
+ZSTDLIB_API const char* ZSTD_getErrorName(size_t code);     /*!< provides readable string from an error code */
+ZSTDLIB_API int         ZSTD_minCLevel(void);               /*!< minimum negative compression level allowed */
+ZSTDLIB_API int         ZSTD_maxCLevel(void);               /*!< maximum compression level available */
+
+
+/* *************************************
+*  Explicit context
+***************************************/
+/*= Compression context
+ *  When compressing many times,
+ *  it is recommended to allocate a context just once,
+ *  and re-use it for each successive compression operation.
+ *  This will make workload friendlier for system's memory.
+ *  Note : re-using context is just a speed / resource optimization.
+ *         It doesn't change the compression ratio, which remains identical.
+ *  Note 2 : In multi-threaded environments,
+ *         use one different context per thread for parallel execution.
+ */
+typedef struct ZSTD_CCtx_s ZSTD_CCtx;
+ZSTDLIB_API ZSTD_CCtx* ZSTD_createCCtx(void);
+ZSTDLIB_API size_t     ZSTD_freeCCtx(ZSTD_CCtx* cctx);  /* accept NULL pointer */
+
+/*! ZSTD_compressCCtx() :
+ *  Same as ZSTD_compress(), using an explicit ZSTD_CCtx.
+ *  Important : in order to behave similarly to `ZSTD_compress()`,
+ *  this function compresses at requested compression level,
+ *  __ignoring any other parameter__ .
+ *  If any advanced parameter was set using the advanced API,
+ *  they will all be reset. Only `compressionLevel` remains.
+ */
+ZSTDLIB_API size_t ZSTD_compressCCtx(ZSTD_CCtx* cctx,
+                                     void* dst, size_t dstCapacity,
+                               const void* src, size_t srcSize,
+                                     int compressionLevel);
+
+/*= Decompression context
+ *  When decompressing many times,
+ *  it is recommended to allocate a context only once,
+ *  and re-use it for each successive compression operation.
+ *  This will make workload friendlier for system's memory.
+ *  Use one context per thread for parallel execution. */
+typedef struct ZSTD_DCtx_s ZSTD_DCtx;
+ZSTDLIB_API ZSTD_DCtx* ZSTD_createDCtx(void);
+ZSTDLIB_API size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  /* accept NULL pointer */
+
+/*! ZSTD_decompressDCtx() :
+ *  Same as ZSTD_decompress(),
+ *  requires an allocated ZSTD_DCtx.
+ *  Compatible with sticky parameters.
+ */
+ZSTDLIB_API size_t ZSTD_decompressDCtx(ZSTD_DCtx* dctx,
+                                       void* dst, size_t dstCapacity,
+                                 const void* src, size_t srcSize);
+
+
+/* *************************************
+*  Advanced compression API
+***************************************/
+
+/* API design :
+ *   Parameters are pushed one by one into an existing context,
+ *   using ZSTD_CCtx_set*() functions.
+ *   Pushed parameters are sticky : they are valid for next compressed frame, and any subsequent frame.
+ *   "sticky" parameters are applicable to `ZSTD_compress2()` and `ZSTD_compressStream*()` !
+ *   __They do not apply to "simple" one-shot variants such as ZSTD_compressCCtx()__ .
+ *
+ *   It's possible to reset all parameters to "default" using ZSTD_CCtx_reset().
+ *
+ *   This API supercedes all other "advanced" API entry points in the experimental section.
+ *   In the future, we expect to remove from experimental API entry points which are redundant with this API.
+ */
+
+
+/* Compression strategies, listed from fastest to strongest */
+typedef enum { ZSTD_fast=1,
+               ZSTD_dfast=2,
+               ZSTD_greedy=3,
+               ZSTD_lazy=4,
+               ZSTD_lazy2=5,
+               ZSTD_btlazy2=6,
+               ZSTD_btopt=7,
+               ZSTD_btultra=8,
+               ZSTD_btultra2=9
+               /* note : new strategies _might_ be added in the future.
+                         Only the order (from fast to strong) is guaranteed */
+} ZSTD_strategy;
+
+
+typedef enum {
+
+    /* compression parameters
+     * Note: When compressing with a ZSTD_CDict these parameters are superseded
+     * by the parameters used to construct the ZSTD_CDict.
+     * See ZSTD_CCtx_refCDict() for more info (superseded-by-cdict). */
+    ZSTD_c_compressionLevel=100, /* Set compression parameters according to pre-defined cLevel table.
+                              * Note that exact compression parameters are dynamically determined,
+                              * depending on both compression level and srcSize (when known).
+                              * Default level is ZSTD_CLEVEL_DEFAULT==3.
+                              * Special: value 0 means default, which is controlled by ZSTD_CLEVEL_DEFAULT.
+                              * Note 1 : it's possible to pass a negative compression level.
+                              * Note 2 : setting a level does not automatically set all other compression parameters
+                              *   to default. Setting this will however eventually dynamically impact the compression
+                              *   parameters which have not been manually set. The manually set
+                              *   ones will 'stick'. */
+    /* Advanced compression parameters :
+     * It's possible to pin down compression parameters to some specific values.
+     * In which case, these values are no longer dynamically selected by the compressor */
+    ZSTD_c_windowLog=101,    /* Maximum allowed back-reference distance, expressed as power of 2.
+                              * This will set a memory budget for streaming decompression,
+                              * with larger values requiring more memory
+                              * and typically compressing more.
+                              * Must be clamped between ZSTD_WINDOWLOG_MIN and ZSTD_WINDOWLOG_MAX.
+                              * Special: value 0 means "use default windowLog".
+                              * Note: Using a windowLog greater than ZSTD_WINDOWLOG_LIMIT_DEFAULT
+                              *       requires explicitly allowing such size at streaming decompression stage. */
+    ZSTD_c_hashLog=102,      /* Size of the initial probe table, as a power of 2.
+                              * Resulting memory usage is (1 << (hashLog+2)).
+                              * Must be clamped between ZSTD_HASHLOG_MIN and ZSTD_HASHLOG_MAX.
+                              * Larger tables improve compression ratio of strategies <= dFast,
+                              * and improve speed of strategies > dFast.
+                              * Special: value 0 means "use default hashLog". */
+    ZSTD_c_chainLog=103,     /* Size of the multi-probe search table, as a power of 2.
+                              * Resulting memory usage is (1 << (chainLog+2)).
+                              * Must be clamped between ZSTD_CHAINLOG_MIN and ZSTD_CHAINLOG_MAX.
+                              * Larger tables result in better and slower compression.
+                              * This parameter is useless for "fast" strategy.
+                              * It's still useful when using "dfast" strategy,
+                              * in which case it defines a secondary probe table.
+                              * Special: value 0 means "use default chainLog". */
+    ZSTD_c_searchLog=104,    /* Number of search attempts, as a power of 2.
+                              * More attempts result in better and slower compression.
+                              * This parameter is useless for "fast" and "dFast" strategies.
+                              * Special: value 0 means "use default searchLog". */
+    ZSTD_c_minMatch=105,     /* Minimum size of searched matches.
+                              * Note that Zstandard can still find matches of smaller size,
+                              * it just tweaks its search algorithm to look for this size and larger.
+                              * Larger values increase compression and decompression speed, but decrease ratio.
+                              * Must be clamped between ZSTD_MINMATCH_MIN and ZSTD_MINMATCH_MAX.
+                              * Note that currently, for all strategies < btopt, effective minimum is 4.
+                              *                    , for all strategies > fast, effective maximum is 6.
+                              * Special: value 0 means "use default minMatchLength". */
+    ZSTD_c_targetLength=106, /* Impact of this field depends on strategy.
+                              * For strategies btopt, btultra & btultra2:
+                              *     Length of Match considered "good enough" to stop search.
+                              *     Larger values make compression stronger, and slower.
+                              * For strategy fast:
+                              *     Distance between match sampling.
+                              *     Larger values make compression faster, and weaker.
+                              * Special: value 0 means "use default targetLength". */
+    ZSTD_c_strategy=107,     /* See ZSTD_strategy enum definition.
+                              * The higher the value of selected strategy, the more complex it is,
+                              * resulting in stronger and slower compression.
+                              * Special: value 0 means "use default strategy". */
+
+    /* LDM mode parameters */
+    ZSTD_c_enableLongDistanceMatching=160, /* Enable long distance matching.
+                                     * This parameter is designed to improve compression ratio
+                                     * for large inputs, by finding large matches at long distance.
+                                     * It increases memory usage and window size.
+                                     * Note: enabling this parameter increases default ZSTD_c_windowLog to 128 MB
+                                     * except when expressly set to a different value.
+                                     * Note: will be enabled by default if ZSTD_c_windowLog >= 128 MB and
+                                     * compression strategy >= ZSTD_btopt (== compression level 16+) */
+    ZSTD_c_ldmHashLog=161,   /* Size of the table for long distance matching, as a power of 2.
+                              * Larger values increase memory usage and compression ratio,
+                              * but decrease compression speed.
+                              * Must be clamped between ZSTD_HASHLOG_MIN and ZSTD_HASHLOG_MAX
+                              * default: windowlog - 7.
+                              * Special: value 0 means "automatically determine hashlog". */
+    ZSTD_c_ldmMinMatch=162,  /* Minimum match size for long distance matcher.
+                              * Larger/too small values usually decrease compression ratio.
+                              * Must be clamped between ZSTD_LDM_MINMATCH_MIN and ZSTD_LDM_MINMATCH_MAX.
+                              * Special: value 0 means "use default value" (default: 64). */
+    ZSTD_c_ldmBucketSizeLog=163, /* Log size of each bucket in the LDM hash table for collision resolution.
+                              * Larger values improve collision resolution but decrease compression speed.
+                              * The maximum value is ZSTD_LDM_BUCKETSIZELOG_MAX.
+                              * Special: value 0 means "use default value" (default: 3). */
+    ZSTD_c_ldmHashRateLog=164, /* Frequency of inserting/looking up entries into the LDM hash table.
+                              * Must be clamped between 0 and (ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN).
+                              * Default is MAX(0, (windowLog - ldmHashLog)), optimizing hash table usage.
+                              * Larger values improve compression speed.
+                              * Deviating far from default value will likely result in a compression ratio decrease.
+                              * Special: value 0 means "automatically determine hashRateLog". */
+
+    /* frame parameters */
+    ZSTD_c_contentSizeFlag=200, /* Content size will be written into frame header _whenever known_ (default:1)
+                              * Content size must be known at the beginning of compression.
+                              * This is automatically the case when using ZSTD_compress2(),
+                              * For streaming scenarios, content size must be provided with ZSTD_CCtx_setPledgedSrcSize() */
+    ZSTD_c_checksumFlag=201, /* A 32-bits checksum of content is written at end of frame (default:0) */
+    ZSTD_c_dictIDFlag=202,   /* When applicable, dictionary's ID is written into frame header (default:1) */
+
+    /* multi-threading parameters */
+    /* These parameters are only active if multi-threading is enabled (compiled with build macro ZSTD_MULTITHREAD).
+     * Otherwise, trying to set any other value than default (0) will be a no-op and return an error.
+     * In a situation where it's unknown if the linked library supports multi-threading or not,
+     * setting ZSTD_c_nbWorkers to any value >= 1 and consulting the return value provides a quick way to check this property.
+     */
+    ZSTD_c_nbWorkers=400,    /* Select how many threads will be spawned to compress in parallel.
+                              * When nbWorkers >= 1, triggers asynchronous mode when invoking ZSTD_compressStream*() :
+                              * ZSTD_compressStream*() consumes input and flush output if possible, but immediately gives back control to caller,
+                              * while compression is performed in parallel, within worker thread(s).
+                              * (note : a strong exception to this rule is when first invocation of ZSTD_compressStream2() sets ZSTD_e_end :
+                              *  in which case, ZSTD_compressStream2() delegates to ZSTD_compress2(), which is always a blocking call).
+                              * More workers improve speed, but also increase memory usage.
+                              * Default value is `0`, aka "single-threaded mode" : no worker is spawned,
+                              * compression is performed inside Caller's thread, and all invocations are blocking */
+    ZSTD_c_jobSize=401,      /* Size of a compression job. This value is enforced only when nbWorkers >= 1.
+                              * Each compression job is completed in parallel, so this value can indirectly impact the nb of active threads.
+                              * 0 means default, which is dynamically determined based on compression parameters.
+                              * Job size must be a minimum of overlap size, or 1 MB, whichever is largest.
+                              * The minimum size is automatically and transparently enforced. */
+    ZSTD_c_overlapLog=402,   /* Control the overlap size, as a fraction of window size.
+                              * The overlap size is an amount of data reloaded from previous job at the beginning of a new job.
+                              * It helps preserve compression ratio, while each job is compressed in parallel.
+                              * This value is enforced only when nbWorkers >= 1.
+                              * Larger values increase compression ratio, but decrease speed.
+                              * Possible values range from 0 to 9 :
+                              * - 0 means "default" : value will be determined by the library, depending on strategy
+                              * - 1 means "no overlap"
+                              * - 9 means "full overlap", using a full window size.
+                              * Each intermediate rank increases/decreases load size by a factor 2 :
+                              * 9: full window;  8: w/2;  7: w/4;  6: w/8;  5:w/16;  4: w/32;  3:w/64;  2:w/128;  1:no overlap;  0:default
+                              * default value varies between 6 and 9, depending on strategy */
+
+    /* note : additional experimental parameters are also available
+     * within the experimental section of the API.
+     * At the time of this writing, they include :
+     * ZSTD_c_rsyncable
+     * ZSTD_c_format
+     * ZSTD_c_forceMaxWindow
+     * ZSTD_c_forceAttachDict
+     * ZSTD_c_literalCompressionMode
+     * ZSTD_c_targetCBlockSize
+     * ZSTD_c_srcSizeHint
+     * ZSTD_c_enableDedicatedDictSearch
+     * ZSTD_c_stableInBuffer
+     * ZSTD_c_stableOutBuffer
+     * ZSTD_c_blockDelimiters
+     * ZSTD_c_validateSequences
+     * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
+     * note : never ever use experimentalParam? names directly;
+     *        also, the enums values themselves are unstable and can still change.
+     */
+     ZSTD_c_experimentalParam1=500,
+     ZSTD_c_experimentalParam2=10,
+     ZSTD_c_experimentalParam3=1000,
+     ZSTD_c_experimentalParam4=1001,
+     ZSTD_c_experimentalParam5=1002,
+     ZSTD_c_experimentalParam6=1003,
+     ZSTD_c_experimentalParam7=1004,
+     ZSTD_c_experimentalParam8=1005,
+     ZSTD_c_experimentalParam9=1006,
+     ZSTD_c_experimentalParam10=1007,
+     ZSTD_c_experimentalParam11=1008,
+     ZSTD_c_experimentalParam12=1009
+} ZSTD_cParameter;
+
+typedef struct {
+    size_t error;
+    int lowerBound;
+    int upperBound;
+} ZSTD_bounds;
+
+/*! ZSTD_cParam_getBounds() :
+ *  All parameters must belong to an interval with lower and upper bounds,
+ *  otherwise they will either trigger an error or be automatically clamped.
+ * @return : a structure, ZSTD_bounds, which contains
+ *         - an error status field, which must be tested using ZSTD_isError()
+ *         - lower and upper bounds, both inclusive
+ */
+ZSTDLIB_API ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter cParam);
+
+/*! ZSTD_CCtx_setParameter() :
+ *  Set one compression parameter, selected by enum ZSTD_cParameter.
+ *  All parameters have valid bounds. Bounds can be queried using ZSTD_cParam_getBounds().
+ *  Providing a value beyond bound will either clamp it, or trigger an error (depending on parameter).
+ *  Setting a parameter is generally only possible during frame initialization (before starting compression).
+ *  Exception : when using multi-threading mode (nbWorkers >= 1),
+ *              the following parameters can be updated _during_ compression (within same frame):
+ *              => compressionLevel, hashLog, chainLog, searchLog, minMatch, targetLength and strategy.
+ *              new parameters will be active for next job only (after a flush()).
+ * @return : an error code (which can be tested using ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value);
+
+/*! ZSTD_CCtx_setPledgedSrcSize() :
+ *  Total input data size to be compressed as a single frame.
+ *  Value will be written in frame header, unless if explicitly forbidden using ZSTD_c_contentSizeFlag.
+ *  This value will also be controlled at end of frame, and trigger an error if not respected.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Note 1 : pledgedSrcSize==0 actually means zero, aka an empty frame.
+ *           In order to mean "unknown content size", pass constant ZSTD_CONTENTSIZE_UNKNOWN.
+ *           ZSTD_CONTENTSIZE_UNKNOWN is default value for any new frame.
+ *  Note 2 : pledgedSrcSize is only valid once, for the next frame.
+ *           It's discarded at the end of the frame, and replaced by ZSTD_CONTENTSIZE_UNKNOWN.
+ *  Note 3 : Whenever all input data is provided and consumed in a single round,
+ *           for example with ZSTD_compress2(),
+ *           or invoking immediately ZSTD_compressStream2(,,,ZSTD_e_end),
+ *           this value is automatically overridden by srcSize instead.
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize);
+
+typedef enum {
+    ZSTD_reset_session_only = 1,
+    ZSTD_reset_parameters = 2,
+    ZSTD_reset_session_and_parameters = 3
+} ZSTD_ResetDirective;
+
+/*! ZSTD_CCtx_reset() :
+ *  There are 2 different things that can be reset, independently or jointly :
+ *  - The session : will stop compressing current frame, and make CCtx ready to start a new one.
+ *                  Useful after an error, or to interrupt any ongoing compression.
+ *                  Any internal data not yet flushed is cancelled.
+ *                  Compression parameters and dictionary remain unchanged.
+ *                  They will be used to compress next frame.
+ *                  Resetting session never fails.
+ *  - The parameters : changes all parameters back to "default".
+ *                  This removes any reference to any dictionary too.
+ *                  Parameters can only be changed between 2 sessions (i.e. no compression is currently ongoing)
+ *                  otherwise the reset fails, and function returns an error value (which can be tested using ZSTD_isError())
+ *  - Both : similar to resetting the session, followed by resetting parameters.
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset);
+
+/*! ZSTD_compress2() :
+ *  Behave the same as ZSTD_compressCCtx(), but compression parameters are set using the advanced API.
+ *  ZSTD_compress2() always starts a new frame.
+ *  Should cctx hold data from a previously unfinished frame, everything about it is forgotten.
+ *  - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
+ *  - The function is always blocking, returns when compression is completed.
+ *  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+ * @return : compressed size written into `dst` (<= `dstCapacity),
+ *           or an error code if it fails (which can be tested using ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_compress2( ZSTD_CCtx* cctx,
+                                   void* dst, size_t dstCapacity,
+                             const void* src, size_t srcSize);
+
+
+/* *************************************
+*  Advanced decompression API
+***************************************/
+
+/* The advanced API pushes parameters one by one into an existing DCtx context.
+ * Parameters are sticky, and remain valid for all following frames
+ * using the same DCtx context.
+ * It's possible to reset parameters to default values using ZSTD_DCtx_reset().
+ * Note : This API is compatible with existing ZSTD_decompressDCtx() and ZSTD_decompressStream().
+ *        Therefore, no new decompression function is necessary.
+ */
+
+typedef enum {
+
+    ZSTD_d_windowLogMax=100, /* Select a size limit (in power of 2) beyond which
+                              * the streaming API will refuse to allocate memory buffer
+                              * in order to protect the host from unreasonable memory requirements.
+                              * This parameter is only useful in streaming mode, since no internal buffer is allocated in single-pass mode.
+                              * By default, a decompression context accepts window sizes <= (1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT).
+                              * Special: value 0 means "use default maximum windowLog". */
+
+    /* note : additional experimental parameters are also available
+     * within the experimental section of the API.
+     * At the time of this writing, they include :
+     * ZSTD_d_format
+     * ZSTD_d_stableOutBuffer
+     * ZSTD_d_forceIgnoreChecksum
+     * ZSTD_d_refMultipleDDicts
+     * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
+     * note : never ever use experimentalParam? names directly
+     */
+     ZSTD_d_experimentalParam1=1000,
+     ZSTD_d_experimentalParam2=1001,
+     ZSTD_d_experimentalParam3=1002,
+     ZSTD_d_experimentalParam4=1003
+
+} ZSTD_dParameter;
+
+/*! ZSTD_dParam_getBounds() :
+ *  All parameters must belong to an interval with lower and upper bounds,
+ *  otherwise they will either trigger an error or be automatically clamped.
+ * @return : a structure, ZSTD_bounds, which contains
+ *         - an error status field, which must be tested using ZSTD_isError()
+ *         - both lower and upper bounds, inclusive
+ */
+ZSTDLIB_API ZSTD_bounds ZSTD_dParam_getBounds(ZSTD_dParameter dParam);
+
+/*! ZSTD_DCtx_setParameter() :
+ *  Set one compression parameter, selected by enum ZSTD_dParameter.
+ *  All parameters have valid bounds. Bounds can be queried using ZSTD_dParam_getBounds().
+ *  Providing a value beyond bound will either clamp it, or trigger an error (depending on parameter).
+ *  Setting a parameter is only possible during frame initialization (before starting decompression).
+ * @return : 0, or an error code (which can be tested using ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_setParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int value);
+
+/*! ZSTD_DCtx_reset() :
+ *  Return a DCtx to clean state.
+ *  Session and parameters can be reset jointly or separately.
+ *  Parameters can only be reset when no active frame is being decompressed.
+ * @return : 0, or an error code, which can be tested with ZSTD_isError()
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_reset(ZSTD_DCtx* dctx, ZSTD_ResetDirective reset);
+
+
+/* **************************
+*  Streaming
+****************************/
+
+typedef struct ZSTD_inBuffer_s {
+  const void* src;    /*< start of input buffer */
+  size_t size;        /*< size of input buffer */
+  size_t pos;         /*< position where reading stopped. Will be updated. Necessarily 0 <= pos <= size */
+} ZSTD_inBuffer;
+
+typedef struct ZSTD_outBuffer_s {
+  void*  dst;         /*< start of output buffer */
+  size_t size;        /*< size of output buffer */
+  size_t pos;         /*< position where writing stopped. Will be updated. Necessarily 0 <= pos <= size */
+} ZSTD_outBuffer;
+
+
+
+/*-***********************************************************************
+*  Streaming compression - HowTo
+*
+*  A ZSTD_CStream object is required to track streaming operation.
+*  Use ZSTD_createCStream() and ZSTD_freeCStream() to create/release resources.
+*  ZSTD_CStream objects can be reused multiple times on consecutive compression operations.
+*  It is recommended to re-use ZSTD_CStream since it will play nicer with system's memory, by re-using already allocated memory.
+*
+*  For parallel execution, use one separate ZSTD_CStream per thread.
+*
+*  note : since v1.3.0, ZSTD_CStream and ZSTD_CCtx are the same thing.
+*
+*  Parameters are sticky : when starting a new compression on the same context,
+*  it will re-use the same sticky parameters as previous compression session.
+*  When in doubt, it's recommended to fully initialize the context before usage.
+*  Use ZSTD_CCtx_reset() to reset the context and ZSTD_CCtx_setParameter(),
+*  ZSTD_CCtx_setPledgedSrcSize(), or ZSTD_CCtx_loadDictionary() and friends to
+*  set more specific parameters, the pledged source size, or load a dictionary.
+*
+*  Use ZSTD_compressStream2() with ZSTD_e_continue as many times as necessary to
+*  consume input stream. The function will automatically update both `pos`
+*  fields within `input` and `output`.
+*  Note that the function may not consume the entire input, for example, because
+*  the output buffer is already full, in which case `input.pos < input.size`.
+*  The caller must check if input has been entirely consumed.
+*  If not, the caller must make some room to receive more compressed data,
+*  and then present again remaining input data.
+*  note: ZSTD_e_continue is guaranteed to make some forward progress when called,
+*        but doesn't guarantee maximal forward progress. This is especially relevant
+*        when compressing with multiple threads. The call won't block if it can
+*        consume some input, but if it can't it will wait for some, but not all,
+*        output to be flushed.
+* @return : provides a minimum amount of data remaining to be flushed from internal buffers
+*           or an error code, which can be tested using ZSTD_isError().
+*
+*  At any moment, it's possible to flush whatever data might remain stuck within internal buffer,
+*  using ZSTD_compressStream2() with ZSTD_e_flush. `output->pos` will be updated.
+*  Note that, if `output->size` is too small, a single invocation with ZSTD_e_flush might not be enough (return code > 0).
+*  In which case, make some room to receive more compressed data, and call again ZSTD_compressStream2() with ZSTD_e_flush.
+*  You must continue calling ZSTD_compressStream2() with ZSTD_e_flush until it returns 0, at which point you can change the
+*  operation.
+*  note: ZSTD_e_flush will flush as much output as possible, meaning when compressing with multiple threads, it will
+*        block until the flush is complete or the output buffer is full.
+*  @return : 0 if internal buffers are entirely flushed,
+*            >0 if some data still present within internal buffer (the value is minimal estimation of remaining size),
+*            or an error code, which can be tested using ZSTD_isError().
+*
+*  Calling ZSTD_compressStream2() with ZSTD_e_end instructs to finish a frame.
+*  It will perform a flush and write frame epilogue.
+*  The epilogue is required for decoders to consider a frame completed.
+*  flush operation is the same, and follows same rules as calling ZSTD_compressStream2() with ZSTD_e_flush.
+*  You must continue calling ZSTD_compressStream2() with ZSTD_e_end until it returns 0, at which point you are free to
+*  start a new frame.
+*  note: ZSTD_e_end will flush as much output as possible, meaning when compressing with multiple threads, it will
+*        block until the flush is complete or the output buffer is full.
+*  @return : 0 if frame fully completed and fully flushed,
+*            >0 if some data still present within internal buffer (the value is minimal estimation of remaining size),
+*            or an error code, which can be tested using ZSTD_isError().
+*
+* *******************************************************************/
+
+typedef ZSTD_CCtx ZSTD_CStream;  /*< CCtx and CStream are now effectively same object (>= v1.3.0) */
+                                 /* Continue to distinguish them for compatibility with older versions <= v1.2.0 */
+/*===== ZSTD_CStream management functions =====*/
+ZSTDLIB_API ZSTD_CStream* ZSTD_createCStream(void);
+ZSTDLIB_API size_t ZSTD_freeCStream(ZSTD_CStream* zcs);  /* accept NULL pointer */
+
+/*===== Streaming compression functions =====*/
+typedef enum {
+    ZSTD_e_continue=0, /* collect more data, encoder decides when to output compressed result, for optimal compression ratio */
+    ZSTD_e_flush=1,    /* flush any data provided so far,
+                        * it creates (at least) one new block, that can be decoded immediately on reception;
+                        * frame will continue: any future data can still reference previously compressed data, improving compression.
+                        * note : multithreaded compression will block to flush as much output as possible. */
+    ZSTD_e_end=2       /* flush any remaining data _and_ close current frame.
+                        * note that frame is only closed after compressed data is fully flushed (return value == 0).
+                        * After that point, any additional data starts a new frame.
+                        * note : each frame is independent (does not reference any content from previous frame).
+                        : note : multithreaded compression will block to flush as much output as possible. */
+} ZSTD_EndDirective;
+
+/*! ZSTD_compressStream2() :
+ *  Behaves about the same as ZSTD_compressStream, with additional control on end directive.
+ *  - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
+ *  - Compression parameters cannot be changed once compression is started (save a list of exceptions in multi-threading mode)
+ *  - output->pos must be <= dstCapacity, input->pos must be <= srcSize
+ *  - output->pos and input->pos will be updated. They are guaranteed to remain below their respective limit.
+ *  - endOp must be a valid directive
+ *  - When nbWorkers==0 (default), function is blocking : it completes its job before returning to caller.
+ *  - When nbWorkers>=1, function is non-blocking : it copies a portion of input, distributes jobs to internal worker threads, flush to output whatever is available,
+ *                                                  and then immediately returns, just indicating that there is some data remaining to be flushed.
+ *                                                  The function nonetheless guarantees forward progress : it will return only after it reads or write at least 1+ byte.
+ *  - Exception : if the first call requests a ZSTD_e_end directive and provides enough dstCapacity, the function delegates to ZSTD_compress2() which is always blocking.
+ *  - @return provides a minimum amount of data remaining to be flushed from internal buffers
+ *            or an error code, which can be tested using ZSTD_isError().
+ *            if @return != 0, flush is not fully completed, there is still some data left within internal buffers.
+ *            This is useful for ZSTD_e_flush, since in this case more flushes are necessary to empty all buffers.
+ *            For ZSTD_e_end, @return == 0 when internal buffers are fully flushed and frame is completed.
+ *  - after a ZSTD_e_end directive, if internal buffer is not fully flushed (@return != 0),
+ *            only ZSTD_e_end or ZSTD_e_flush operations are allowed.
+ *            Before starting a new compression job, or changing compression parameters,
+ *            it is required to fully flush internal buffers.
+ */
+ZSTDLIB_API size_t ZSTD_compressStream2( ZSTD_CCtx* cctx,
+                                         ZSTD_outBuffer* output,
+                                         ZSTD_inBuffer* input,
+                                         ZSTD_EndDirective endOp);
+
+
+/* These buffer sizes are softly recommended.
+ * They are not required : ZSTD_compressStream*() happily accepts any buffer size, for both input and output.
+ * Respecting the recommended size just makes it a bit easier for ZSTD_compressStream*(),
+ * reducing the amount of memory shuffling and buffering, resulting in minor performance savings.
+ *
+ * However, note that these recommendations are from the perspective of a C caller program.
+ * If the streaming interface is invoked from some other language,
+ * especially managed ones such as Java or Go, through a foreign function interface such as jni or cgo,
+ * a major performance rule is to reduce crossing such interface to an absolute minimum.
+ * It's not rare that performance ends being spent more into the interface, rather than compression itself.
+ * In which cases, prefer using large buffers, as large as practical,
+ * for both input and output, to reduce the nb of roundtrips.
+ */
+ZSTDLIB_API size_t ZSTD_CStreamInSize(void);    /*< recommended size for input buffer */
+ZSTDLIB_API size_t ZSTD_CStreamOutSize(void);   /*< recommended size for output buffer. Guarantee to successfully flush at least one complete compressed block. */
+
+
+/* *****************************************************************************
+ * This following is a legacy streaming API.
+ * It can be replaced by ZSTD_CCtx_reset() and ZSTD_compressStream2().
+ * It is redundant, but remains fully supported.
+ * Advanced parameters and dictionary compression can only be used through the
+ * new API.
+ ******************************************************************************/
+
+/*!
+ * Equivalent to:
+ *
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_refCDict(zcs, NULL); // clear the dictionary (if any)
+ *     ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
+ */
+ZSTDLIB_API size_t ZSTD_initCStream(ZSTD_CStream* zcs, int compressionLevel);
+/*!
+ * Alternative for ZSTD_compressStream2(zcs, output, input, ZSTD_e_continue).
+ * NOTE: The return value is different. ZSTD_compressStream() returns a hint for
+ * the next read size (if non-zero and not an error). ZSTD_compressStream2()
+ * returns the minimum nb of bytes left to flush (if non-zero and not an error).
+ */
+ZSTDLIB_API size_t ZSTD_compressStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
+/*! Equivalent to ZSTD_compressStream2(zcs, output, &emptyInput, ZSTD_e_flush). */
+ZSTDLIB_API size_t ZSTD_flushStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
+/*! Equivalent to ZSTD_compressStream2(zcs, output, &emptyInput, ZSTD_e_end). */
+ZSTDLIB_API size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
+
+
+/*-***************************************************************************
+*  Streaming decompression - HowTo
+*
+*  A ZSTD_DStream object is required to track streaming operations.
+*  Use ZSTD_createDStream() and ZSTD_freeDStream() to create/release resources.
+*  ZSTD_DStream objects can be re-used multiple times.
+*
+*  Use ZSTD_initDStream() to start a new decompression operation.
+* @return : recommended first input size
+*  Alternatively, use advanced API to set specific properties.
+*
+*  Use ZSTD_decompressStream() repetitively to consume your input.
+*  The function will update both `pos` fields.
+*  If `input.pos < input.size`, some input has not been consumed.
+*  It's up to the caller to present again remaining data.
+*  The function tries to flush all data decoded immediately, respecting output buffer size.
+*  If `output.pos < output.size`, decoder has flushed everything it could.
+*  But if `output.pos == output.size`, there might be some data left within internal buffers.,
+*  In which case, call ZSTD_decompressStream() again to flush whatever remains in the buffer.
+*  Note : with no additional input provided, amount of data flushed is necessarily <= ZSTD_BLOCKSIZE_MAX.
+* @return : 0 when a frame is completely decoded and fully flushed,
+*        or an error code, which can be tested using ZSTD_isError(),
+*        or any other value > 0, which means there is still some decoding or flushing to do to complete current frame :
+*                                the return value is a suggested next input size (just a hint for better latency)
+*                                that will never request more than the remaining frame size.
+* *******************************************************************************/
+
+typedef ZSTD_DCtx ZSTD_DStream;  /*< DCtx and DStream are now effectively same object (>= v1.3.0) */
+                                 /* For compatibility with versions <= v1.2.0, prefer differentiating them. */
+/*===== ZSTD_DStream management functions =====*/
+ZSTDLIB_API ZSTD_DStream* ZSTD_createDStream(void);
+ZSTDLIB_API size_t ZSTD_freeDStream(ZSTD_DStream* zds);  /* accept NULL pointer */
+
+/*===== Streaming decompression functions =====*/
+
+/* This function is redundant with the advanced API and equivalent to:
+ *
+ *     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
+ *     ZSTD_DCtx_refDDict(zds, NULL);
+ */
+ZSTDLIB_API size_t ZSTD_initDStream(ZSTD_DStream* zds);
+
+ZSTDLIB_API size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
+
+ZSTDLIB_API size_t ZSTD_DStreamInSize(void);    /*!< recommended size for input buffer */
+ZSTDLIB_API size_t ZSTD_DStreamOutSize(void);   /*!< recommended size for output buffer. Guarantee to successfully flush at least one complete block in all circumstances. */
+
+
+/* ************************
+*  Simple dictionary API
+***************************/
+/*! ZSTD_compress_usingDict() :
+ *  Compression at an explicit compression level using a Dictionary.
+ *  A dictionary can be any arbitrary data segment (also called a prefix),
+ *  or a buffer with specified information (see dictBuilder/zdict.h).
+ *  Note : This function loads the dictionary, resulting in significant startup delay.
+ *         It's intended for a dictionary used only once.
+ *  Note 2 : When `dict == NULL || dictSize < 8` no dictionary is used. */
+ZSTDLIB_API size_t ZSTD_compress_usingDict(ZSTD_CCtx* ctx,
+                                           void* dst, size_t dstCapacity,
+                                     const void* src, size_t srcSize,
+                                     const void* dict,size_t dictSize,
+                                           int compressionLevel);
+
+/*! ZSTD_decompress_usingDict() :
+ *  Decompression using a known Dictionary.
+ *  Dictionary must be identical to the one used during compression.
+ *  Note : This function loads the dictionary, resulting in significant startup delay.
+ *         It's intended for a dictionary used only once.
+ *  Note : When `dict == NULL || dictSize < 8` no dictionary is used. */
+ZSTDLIB_API size_t ZSTD_decompress_usingDict(ZSTD_DCtx* dctx,
+                                             void* dst, size_t dstCapacity,
+                                       const void* src, size_t srcSize,
+                                       const void* dict,size_t dictSize);
+
+
+/* *********************************
+ *  Bulk processing dictionary API
+ **********************************/
+typedef struct ZSTD_CDict_s ZSTD_CDict;
+
+/*! ZSTD_createCDict() :
+ *  When compressing multiple messages or blocks using the same dictionary,
+ *  it's recommended to digest the dictionary only once, since it's a costly operation.
+ *  ZSTD_createCDict() will create a state from digesting a dictionary.
+ *  The resulting state can be used for future compression operations with very limited startup cost.
+ *  ZSTD_CDict can be created once and shared by multiple threads concurrently, since its usage is read-only.
+ * @dictBuffer can be released after ZSTD_CDict creation, because its content is copied within CDict.
+ *  Note 1 : Consider experimental function `ZSTD_createCDict_byReference()` if you prefer to not duplicate @dictBuffer content.
+ *  Note 2 : A ZSTD_CDict can be created from an empty @dictBuffer,
+ *      in which case the only thing that it transports is the @compressionLevel.
+ *      This can be useful in a pipeline featuring ZSTD_compress_usingCDict() exclusively,
+ *      expecting a ZSTD_CDict parameter with any data, including those without a known dictionary. */
+ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict(const void* dictBuffer, size_t dictSize,
+                                         int compressionLevel);
+
+/*! ZSTD_freeCDict() :
+ *  Function frees memory allocated by ZSTD_createCDict().
+ *  If a NULL pointer is passed, no operation is performed. */
+ZSTDLIB_API size_t      ZSTD_freeCDict(ZSTD_CDict* CDict);
+
+/*! ZSTD_compress_usingCDict() :
+ *  Compression using a digested Dictionary.
+ *  Recommended when same dictionary is used multiple times.
+ *  Note : compression level is _decided at dictionary creation time_,
+ *     and frame parameters are hardcoded (dictID=yes, contentSize=yes, checksum=no) */
+ZSTDLIB_API size_t ZSTD_compress_usingCDict(ZSTD_CCtx* cctx,
+                                            void* dst, size_t dstCapacity,
+                                      const void* src, size_t srcSize,
+                                      const ZSTD_CDict* cdict);
+
+
+typedef struct ZSTD_DDict_s ZSTD_DDict;
+
+/*! ZSTD_createDDict() :
+ *  Create a digested dictionary, ready to start decompression operation without startup delay.
+ *  dictBuffer can be released after DDict creation, as its content is copied inside DDict. */
+ZSTDLIB_API ZSTD_DDict* ZSTD_createDDict(const void* dictBuffer, size_t dictSize);
+
+/*! ZSTD_freeDDict() :
+ *  Function frees memory allocated with ZSTD_createDDict()
+ *  If a NULL pointer is passed, no operation is performed. */
+ZSTDLIB_API size_t      ZSTD_freeDDict(ZSTD_DDict* ddict);
+
+/*! ZSTD_decompress_usingDDict() :
+ *  Decompression using a digested Dictionary.
+ *  Recommended when same dictionary is used multiple times. */
+ZSTDLIB_API size_t ZSTD_decompress_usingDDict(ZSTD_DCtx* dctx,
+                                              void* dst, size_t dstCapacity,
+                                        const void* src, size_t srcSize,
+                                        const ZSTD_DDict* ddict);
+
+
+/* ******************************
+ *  Dictionary helper functions
+ *******************************/
+
+/*! ZSTD_getDictID_fromDict() :
+ *  Provides the dictID stored within dictionary.
+ *  if @return == 0, the dictionary is not conformant with Zstandard specification.
+ *  It can still be loaded, but as a content-only dictionary. */
+ZSTDLIB_API unsigned ZSTD_getDictID_fromDict(const void* dict, size_t dictSize);
+
+/*! ZSTD_getDictID_fromDDict() :
+ *  Provides the dictID of the dictionary loaded into `ddict`.
+ *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
+ *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
+ZSTDLIB_API unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict* ddict);
+
+/*! ZSTD_getDictID_fromFrame() :
+ *  Provides the dictID required to decompressed the frame stored within `src`.
+ *  If @return == 0, the dictID could not be decoded.
+ *  This could for one of the following reasons :
+ *  - The frame does not require a dictionary to be decoded (most common case).
+ *  - The frame was built with dictID intentionally removed. Whatever dictionary is necessary is a hidden information.
+ *    Note : this use case also happens when using a non-conformant dictionary.
+ *  - `srcSize` is too small, and as a result, the frame header could not be decoded (only possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`).
+ *  - This is not a Zstandard frame.
+ *  When identifying the exact failure cause, it's possible to use ZSTD_getFrameHeader(), which will provide a more precise error code. */
+ZSTDLIB_API unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize);
+
+
+/* *****************************************************************************
+ * Advanced dictionary and prefix API
+ *
+ * This API allows dictionaries to be used with ZSTD_compress2(),
+ * ZSTD_compressStream2(), and ZSTD_decompress(). Dictionaries are sticky, and
+ * only reset with the context is reset with ZSTD_reset_parameters or
+ * ZSTD_reset_session_and_parameters. Prefixes are single-use.
+ ******************************************************************************/
+
+
+/*! ZSTD_CCtx_loadDictionary() :
+ *  Create an internal CDict from `dict` buffer.
+ *  Decompression will have to use same dictionary.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Special: Loading a NULL (or 0-size) dictionary invalidates previous dictionary,
+ *           meaning "return to no-dictionary mode".
+ *  Note 1 : Dictionary is sticky, it will be used for all future compressed frames.
+ *           To return to "no-dictionary" situation, load a NULL dictionary (or reset parameters).
+ *  Note 2 : Loading a dictionary involves building tables.
+ *           It's also a CPU consuming operation, with non-negligible impact on latency.
+ *           Tables are dependent on compression parameters, and for this reason,
+ *           compression parameters can no longer be changed after loading a dictionary.
+ *  Note 3 :`dict` content will be copied internally.
+ *           Use experimental ZSTD_CCtx_loadDictionary_byReference() to reference content instead.
+ *           In such a case, dictionary buffer must outlive its users.
+ *  Note 4 : Use ZSTD_CCtx_loadDictionary_advanced()
+ *           to precisely select how dictionary content must be interpreted. */
+ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize);
+
+/*! ZSTD_CCtx_refCDict() :
+ *  Reference a prepared dictionary, to be used for all next compressed frames.
+ *  Note that compression parameters are enforced from within CDict,
+ *  and supersede any compression parameter previously set within CCtx.
+ *  The parameters ignored are labelled as "superseded-by-cdict" in the ZSTD_cParameter enum docs.
+ *  The ignored parameters will be used again if the CCtx is returned to no-dictionary mode.
+ *  The dictionary will remain valid for future compressed frames using same CCtx.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Special : Referencing a NULL CDict means "return to no-dictionary mode".
+ *  Note 1 : Currently, only one dictionary can be managed.
+ *           Referencing a new dictionary effectively "discards" any previous one.
+ *  Note 2 : CDict is just referenced, its lifetime must outlive its usage within CCtx. */
+ZSTDLIB_API size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict);
+
+/*! ZSTD_CCtx_refPrefix() :
+ *  Reference a prefix (single-usage dictionary) for next compressed frame.
+ *  A prefix is **only used once**. Tables are discarded at end of frame (ZSTD_e_end).
+ *  Decompression will need same prefix to properly regenerate data.
+ *  Compressing with a prefix is similar in outcome as performing a diff and compressing it,
+ *  but performs much faster, especially during decompression (compression speed is tunable with compression level).
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Special: Adding any prefix (including NULL) invalidates any previous prefix or dictionary
+ *  Note 1 : Prefix buffer is referenced. It **must** outlive compression.
+ *           Its content must remain unmodified during compression.
+ *  Note 2 : If the intention is to diff some large src data blob with some prior version of itself,
+ *           ensure that the window size is large enough to contain the entire source.
+ *           See ZSTD_c_windowLog.
+ *  Note 3 : Referencing a prefix involves building tables, which are dependent on compression parameters.
+ *           It's a CPU consuming operation, with non-negligible impact on latency.
+ *           If there is a need to use the same prefix multiple times, consider loadDictionary instead.
+ *  Note 4 : By default, the prefix is interpreted as raw content (ZSTD_dct_rawContent).
+ *           Use experimental ZSTD_CCtx_refPrefix_advanced() to alter dictionary interpretation. */
+ZSTDLIB_API size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx,
+                                 const void* prefix, size_t prefixSize);
+
+/*! ZSTD_DCtx_loadDictionary() :
+ *  Create an internal DDict from dict buffer,
+ *  to be used to decompress next frames.
+ *  The dictionary remains valid for all future frames, until explicitly invalidated.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Special : Adding a NULL (or 0-size) dictionary invalidates any previous dictionary,
+ *            meaning "return to no-dictionary mode".
+ *  Note 1 : Loading a dictionary involves building tables,
+ *           which has a non-negligible impact on CPU usage and latency.
+ *           It's recommended to "load once, use many times", to amortize the cost
+ *  Note 2 :`dict` content will be copied internally, so `dict` can be released after loading.
+ *           Use ZSTD_DCtx_loadDictionary_byReference() to reference dictionary content instead.
+ *  Note 3 : Use ZSTD_DCtx_loadDictionary_advanced() to take control of
+ *           how dictionary content is loaded and interpreted.
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_loadDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);
+
+/*! ZSTD_DCtx_refDDict() :
+ *  Reference a prepared dictionary, to be used to decompress next frames.
+ *  The dictionary remains active for decompression of future frames using same DCtx.
+ *
+ *  If called with ZSTD_d_refMultipleDDicts enabled, repeated calls of this function
+ *  will store the DDict references in a table, and the DDict used for decompression
+ *  will be determined at decompression time, as per the dict ID in the frame.
+ *  The memory for the table is allocated on the first call to refDDict, and can be
+ *  freed with ZSTD_freeDCtx().
+ *
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Note 1 : Currently, only one dictionary can be managed.
+ *           Referencing a new dictionary effectively "discards" any previous one.
+ *  Special: referencing a NULL DDict means "return to no-dictionary mode".
+ *  Note 2 : DDict is just referenced, its lifetime must outlive its usage from DCtx.
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_refDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);
+
+/*! ZSTD_DCtx_refPrefix() :
+ *  Reference a prefix (single-usage dictionary) to decompress next frame.
+ *  This is the reverse operation of ZSTD_CCtx_refPrefix(),
+ *  and must use the same prefix as the one used during compression.
+ *  Prefix is **only used once**. Reference is discarded at end of frame.
+ *  End of frame is reached when ZSTD_decompressStream() returns 0.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Note 1 : Adding any prefix (including NULL) invalidates any previously set prefix or dictionary
+ *  Note 2 : Prefix buffer is referenced. It **must** outlive decompression.
+ *           Prefix buffer must remain unmodified up to the end of frame,
+ *           reached when ZSTD_decompressStream() returns 0.
+ *  Note 3 : By default, the prefix is treated as raw content (ZSTD_dct_rawContent).
+ *           Use ZSTD_CCtx_refPrefix_advanced() to alter dictMode (Experimental section)
+ *  Note 4 : Referencing a raw content prefix has almost no cpu nor memory cost.
+ *           A full dictionary is more costly, as it requires building tables.
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_refPrefix(ZSTD_DCtx* dctx,
+                                 const void* prefix, size_t prefixSize);
+
+/* ===   Memory management   === */
+
+/*! ZSTD_sizeof_*() :
+ *  These functions give the _current_ memory usage of selected object.
+ *  Note that object memory usage can evolve (increase or decrease) over time. */
+ZSTDLIB_API size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx);
+ZSTDLIB_API size_t ZSTD_sizeof_DCtx(const ZSTD_DCtx* dctx);
+ZSTDLIB_API size_t ZSTD_sizeof_CStream(const ZSTD_CStream* zcs);
+ZSTDLIB_API size_t ZSTD_sizeof_DStream(const ZSTD_DStream* zds);
+ZSTDLIB_API size_t ZSTD_sizeof_CDict(const ZSTD_CDict* cdict);
+ZSTDLIB_API size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);
+
+#endif  /* ZSTD_H_235446 */
+
+
+/* **************************************************************************************
+ *   ADVANCED AND EXPERIMENTAL FUNCTIONS
+ ****************************************************************************************
+ * The definitions in the following section are considered experimental.
+ * They are provided for advanced scenarios.
+ * They should never be used with a dynamic library, as prototypes may change in the future.
+ * Use them only in association with static linking.
+ * ***************************************************************************************/
+
+#if !defined(ZSTD_H_ZSTD_STATIC_LINKING_ONLY)
+#define ZSTD_H_ZSTD_STATIC_LINKING_ONLY
+
+/* **************************************************************************************
+ *   experimental API (static linking only)
+ ****************************************************************************************
+ * The following symbols and constants
+ * are not planned to join "stable API" status in the near future.
+ * They can still change in future versions.
+ * Some of them are planned to remain in the static_only section indefinitely.
+ * Some of them might be removed in the future (especially when redundant with existing stable functions)
+ * ***************************************************************************************/
+
+#define ZSTD_FRAMEHEADERSIZE_PREFIX(format) ((format) == ZSTD_f_zstd1 ? 5 : 1)   /* minimum input size required to query frame header size */
+#define ZSTD_FRAMEHEADERSIZE_MIN(format)    ((format) == ZSTD_f_zstd1 ? 6 : 2)
+#define ZSTD_FRAMEHEADERSIZE_MAX   18   /* can be useful for static allocation */
+#define ZSTD_SKIPPABLEHEADERSIZE    8
+
+/* compression parameter bounds */
+#define ZSTD_WINDOWLOG_MAX_32    30
+#define ZSTD_WINDOWLOG_MAX_64    31
+#define ZSTD_WINDOWLOG_MAX     ((int)(sizeof(size_t) == 4 ? ZSTD_WINDOWLOG_MAX_32 : ZSTD_WINDOWLOG_MAX_64))
+#define ZSTD_WINDOWLOG_MIN       10
+#define ZSTD_HASHLOG_MAX       ((ZSTD_WINDOWLOG_MAX < 30) ? ZSTD_WINDOWLOG_MAX : 30)
+#define ZSTD_HASHLOG_MIN          6
+#define ZSTD_CHAINLOG_MAX_32     29
+#define ZSTD_CHAINLOG_MAX_64     30
+#define ZSTD_CHAINLOG_MAX      ((int)(sizeof(size_t) == 4 ? ZSTD_CHAINLOG_MAX_32 : ZSTD_CHAINLOG_MAX_64))
+#define ZSTD_CHAINLOG_MIN        ZSTD_HASHLOG_MIN
+#define ZSTD_SEARCHLOG_MAX      (ZSTD_WINDOWLOG_MAX-1)
+#define ZSTD_SEARCHLOG_MIN        1
+#define ZSTD_MINMATCH_MAX         7   /* only for ZSTD_fast, other strategies are limited to 6 */
+#define ZSTD_MINMATCH_MIN         3   /* only for ZSTD_btopt+, faster strategies are limited to 4 */
+#define ZSTD_TARGETLENGTH_MAX    ZSTD_BLOCKSIZE_MAX
+#define ZSTD_TARGETLENGTH_MIN     0   /* note : comparing this constant to an unsigned results in a tautological test */
+#define ZSTD_STRATEGY_MIN        ZSTD_fast
+#define ZSTD_STRATEGY_MAX        ZSTD_btultra2
+
+
+#define ZSTD_OVERLAPLOG_MIN       0
+#define ZSTD_OVERLAPLOG_MAX       9
+
+#define ZSTD_WINDOWLOG_LIMIT_DEFAULT 27   /* by default, the streaming decoder will refuse any frame
+                                           * requiring larger than (1<<ZSTD_WINDOWLOG_LIMIT_DEFAULT) window size,
+                                           * to preserve host's memory from unreasonable requirements.
+                                           * This limit can be overridden using ZSTD_DCtx_setParameter(,ZSTD_d_windowLogMax,).
+                                           * The limit does not apply for one-pass decoders (such as ZSTD_decompress()), since no additional memory is allocated */
+
+
+/* LDM parameter bounds */
+#define ZSTD_LDM_HASHLOG_MIN      ZSTD_HASHLOG_MIN
+#define ZSTD_LDM_HASHLOG_MAX      ZSTD_HASHLOG_MAX
+#define ZSTD_LDM_MINMATCH_MIN        4
+#define ZSTD_LDM_MINMATCH_MAX     4096
+#define ZSTD_LDM_BUCKETSIZELOG_MIN   1
+#define ZSTD_LDM_BUCKETSIZELOG_MAX   8
+#define ZSTD_LDM_HASHRATELOG_MIN     0
+#define ZSTD_LDM_HASHRATELOG_MAX (ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN)
+
+/* Advanced parameter bounds */
+#define ZSTD_TARGETCBLOCKSIZE_MIN   64
+#define ZSTD_TARGETCBLOCKSIZE_MAX   ZSTD_BLOCKSIZE_MAX
+#define ZSTD_SRCSIZEHINT_MIN        0
+#define ZSTD_SRCSIZEHINT_MAX        INT_MAX
+
+/* internal */
+#define ZSTD_HASHLOG3_MAX           17
+
+
+/* ---  Advanced types  --- */
+
+typedef struct ZSTD_CCtx_params_s ZSTD_CCtx_params;
+
+typedef struct {
+    unsigned int offset;      /* The offset of the match. (NOT the same as the offset code)
+                               * If offset == 0 and matchLength == 0, this sequence represents the last
+                               * literals in the block of litLength size.
+                               */
+
+    unsigned int litLength;   /* Literal length of the sequence. */
+    unsigned int matchLength; /* Match length of the sequence. */
+
+                              /* Note: Users of this API may provide a sequence with matchLength == litLength == offset == 0.
+                               * In this case, we will treat the sequence as a marker for a block boundary.
+                               */
+
+    unsigned int rep;         /* Represents which repeat offset is represented by the field 'offset'.
+                               * Ranges from [0, 3].
+                               *
+                               * Repeat offsets are essentially previous offsets from previous sequences sorted in
+                               * recency order. For more detail, see doc/zstd_compression_format.md
+                               *
+                               * If rep == 0, then 'offset' does not contain a repeat offset.
+                               * If rep > 0:
+                               *  If litLength != 0:
+                               *      rep == 1 --> offset == repeat_offset_1
+                               *      rep == 2 --> offset == repeat_offset_2
+                               *      rep == 3 --> offset == repeat_offset_3
+                               *  If litLength == 0:
+                               *      rep == 1 --> offset == repeat_offset_2
+                               *      rep == 2 --> offset == repeat_offset_3
+                               *      rep == 3 --> offset == repeat_offset_1 - 1
+                               *
+                               * Note: This field is optional. ZSTD_generateSequences() will calculate the value of
+                               * 'rep', but repeat offsets do not necessarily need to be calculated from an external
+                               * sequence provider's perspective. For example, ZSTD_compressSequences() does not
+                               * use this 'rep' field at all (as of now).
+                               */
+} ZSTD_Sequence;
+
+typedef struct {
+    unsigned windowLog;       /*< largest match distance : larger == more compression, more memory needed during decompression */
+    unsigned chainLog;        /*< fully searched segment : larger == more compression, slower, more memory (useless for fast) */
+    unsigned hashLog;         /*< dispatch table : larger == faster, more memory */
+    unsigned searchLog;       /*< nb of searches : larger == more compression, slower */
+    unsigned minMatch;        /*< match length searched : larger == faster decompression, sometimes less compression */
+    unsigned targetLength;    /*< acceptable match size for optimal parser (only) : larger == more compression, slower */
+    ZSTD_strategy strategy;   /*< see ZSTD_strategy definition above */
+} ZSTD_compressionParameters;
+
+typedef struct {
+    int contentSizeFlag; /*< 1: content size will be in frame header (when known) */
+    int checksumFlag;    /*< 1: generate a 32-bits checksum using XXH64 algorithm at end of frame, for error detection */
+    int noDictIDFlag;    /*< 1: no dictID will be saved into frame header (dictID is only useful for dictionary compression) */
+} ZSTD_frameParameters;
+
+typedef struct {
+    ZSTD_compressionParameters cParams;
+    ZSTD_frameParameters fParams;
+} ZSTD_parameters;
+
+typedef enum {
+    ZSTD_dct_auto = 0,       /* dictionary is "full" when starting with ZSTD_MAGIC_DICTIONARY, otherwise it is "rawContent" */
+    ZSTD_dct_rawContent = 1, /* ensures dictionary is always loaded as rawContent, even if it starts with ZSTD_MAGIC_DICTIONARY */
+    ZSTD_dct_fullDict = 2    /* refuses to load a dictionary if it does not respect Zstandard's specification, starting with ZSTD_MAGIC_DICTIONARY */
+} ZSTD_dictContentType_e;
+
+typedef enum {
+    ZSTD_dlm_byCopy = 0,  /*< Copy dictionary content internally */
+    ZSTD_dlm_byRef = 1    /*< Reference dictionary content -- the dictionary buffer must outlive its users. */
+} ZSTD_dictLoadMethod_e;
+
+typedef enum {
+    ZSTD_f_zstd1 = 0,           /* zstd frame format, specified in zstd_compression_format.md (default) */
+    ZSTD_f_zstd1_magicless = 1  /* Variant of zstd frame format, without initial 4-bytes magic number.
+                                 * Useful to save 4 bytes per generated frame.
+                                 * Decoder cannot recognise automatically this format, requiring this instruction. */
+} ZSTD_format_e;
+
+typedef enum {
+    /* Note: this enum controls ZSTD_d_forceIgnoreChecksum */
+    ZSTD_d_validateChecksum = 0,
+    ZSTD_d_ignoreChecksum = 1
+} ZSTD_forceIgnoreChecksum_e;
+
+typedef enum {
+    /* Note: this enum controls ZSTD_d_refMultipleDDicts */
+    ZSTD_rmd_refSingleDDict = 0,
+    ZSTD_rmd_refMultipleDDicts = 1
+} ZSTD_refMultipleDDicts_e;
+
+typedef enum {
+    /* Note: this enum and the behavior it controls are effectively internal
+     * implementation details of the compressor. They are expected to continue
+     * to evolve and should be considered only in the context of extremely
+     * advanced performance tuning.
+     *
+     * Zstd currently supports the use of a CDict in three ways:
+     *
+     * - The contents of the CDict can be copied into the working context. This
+     *   means that the compression can search both the dictionary and input
+     *   while operating on a single set of internal tables. This makes
+     *   the compression faster per-byte of input. However, the initial copy of
+     *   the CDict's tables incurs a fixed cost at the beginning of the
+     *   compression. For small compressions (< 8 KB), that copy can dominate
+     *   the cost of the compression.
+     *
+     * - The CDict's tables can be used in-place. In this model, compression is
+     *   slower per input byte, because the compressor has to search two sets of
+     *   tables. However, this model incurs no start-up cost (as long as the
+     *   working context's tables can be reused). For small inputs, this can be
+     *   faster than copying the CDict's tables.
+     *
+     * - The CDict's tables are not used at all, and instead we use the working
+     *   context alone to reload the dictionary and use params based on the source
+     *   size. See ZSTD_compress_insertDictionary() and ZSTD_compress_usingDict().
+     *   This method is effective when the dictionary sizes are very small relative
+     *   to the input size, and the input size is fairly large to begin with.
+     *
+     * Zstd has a simple internal heuristic that selects which strategy to use
+     * at the beginning of a compression. However, if experimentation shows that
+     * Zstd is making poor choices, it is possible to override that choice with
+     * this enum.
+     */
+    ZSTD_dictDefaultAttach = 0, /* Use the default heuristic. */
+    ZSTD_dictForceAttach   = 1, /* Never copy the dictionary. */
+    ZSTD_dictForceCopy     = 2, /* Always copy the dictionary. */
+    ZSTD_dictForceLoad     = 3  /* Always reload the dictionary */
+} ZSTD_dictAttachPref_e;
+
+typedef enum {
+  ZSTD_lcm_auto = 0,          /*< Automatically determine the compression mode based on the compression level.
+                               *   Negative compression levels will be uncompressed, and positive compression
+                               *   levels will be compressed. */
+  ZSTD_lcm_huffman = 1,       /*< Always attempt Huffman compression. Uncompressed literals will still be
+                               *   emitted if Huffman compression is not profitable. */
+  ZSTD_lcm_uncompressed = 2   /*< Always emit uncompressed literals. */
+} ZSTD_literalCompressionMode_e;
+
+
+/* *************************************
+*  Frame size functions
+***************************************/
+
+/*! ZSTD_findDecompressedSize() :
+ *  `src` should point to the start of a series of ZSTD encoded and/or skippable frames
+ *  `srcSize` must be the _exact_ size of this series
+ *       (i.e. there should be a frame boundary at `src + srcSize`)
+ *  @return : - decompressed size of all data in all successive frames
+ *            - if the decompressed size cannot be determined: ZSTD_CONTENTSIZE_UNKNOWN
+ *            - if an error occurred: ZSTD_CONTENTSIZE_ERROR
+ *
+ *   note 1 : decompressed size is an optional field, that may not be present, especially in streaming mode.
+ *            When `return==ZSTD_CONTENTSIZE_UNKNOWN`, data to decompress could be any size.
+ *            In which case, it's necessary to use streaming mode to decompress data.
+ *   note 2 : decompressed size is always present when compression is done with ZSTD_compress()
+ *   note 3 : decompressed size can be very large (64-bits value),
+ *            potentially larger than what local system can handle as a single memory segment.
+ *            In which case, it's necessary to use streaming mode to decompress data.
+ *   note 4 : If source is untrusted, decompressed size could be wrong or intentionally modified.
+ *            Always ensure result fits within application's authorized limits.
+ *            Each application can set its own limits.
+ *   note 5 : ZSTD_findDecompressedSize handles multiple frames, and so it must traverse the input to
+ *            read each contained frame header.  This is fast as most of the data is skipped,
+ *            however it does mean that all frame data must be present and valid. */
+ZSTDLIB_API unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize);
+
+/*! ZSTD_decompressBound() :
+ *  `src` should point to the start of a series of ZSTD encoded and/or skippable frames
+ *  `srcSize` must be the _exact_ size of this series
+ *       (i.e. there should be a frame boundary at `src + srcSize`)
+ *  @return : - upper-bound for the decompressed size of all data in all successive frames
+ *            - if an error occurred: ZSTD_CONTENTSIZE_ERROR
+ *
+ *  note 1  : an error can occur if `src` contains an invalid or incorrectly formatted frame.
+ *  note 2  : the upper-bound is exact when the decompressed size field is available in every ZSTD encoded frame of `src`.
+ *            in this case, `ZSTD_findDecompressedSize` and `ZSTD_decompressBound` return the same value.
+ *  note 3  : when the decompressed size field isn't available, the upper-bound for that frame is calculated by:
+ *              upper-bound = # blocks * min(128 KB, Window_Size)
+ */
+ZSTDLIB_API unsigned long long ZSTD_decompressBound(const void* src, size_t srcSize);
+
+/*! ZSTD_frameHeaderSize() :
+ *  srcSize must be >= ZSTD_FRAMEHEADERSIZE_PREFIX.
+ * @return : size of the Frame Header,
+ *           or an error code (if srcSize is too small) */
+ZSTDLIB_API size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize);
+
+typedef enum {
+  ZSTD_sf_noBlockDelimiters = 0,         /* Representation of ZSTD_Sequence has no block delimiters, sequences only */
+  ZSTD_sf_explicitBlockDelimiters = 1    /* Representation of ZSTD_Sequence contains explicit block delimiters */
+} ZSTD_sequenceFormat_e;
+
+/*! ZSTD_generateSequences() :
+ * Generate sequences using ZSTD_compress2, given a source buffer.
+ *
+ * Each block will end with a dummy sequence
+ * with offset == 0, matchLength == 0, and litLength == length of last literals.
+ * litLength may be == 0, and if so, then the sequence of (of: 0 ml: 0 ll: 0)
+ * simply acts as a block delimiter.
+ *
+ * zc can be used to insert custom compression params.
+ * This function invokes ZSTD_compress2
+ *
+ * The output of this function can be fed into ZSTD_compressSequences() with CCtx
+ * setting of ZSTD_c_blockDelimiters as ZSTD_sf_explicitBlockDelimiters
+ * @return : number of sequences generated
+ */
+
+ZSTDLIB_API size_t ZSTD_generateSequences(ZSTD_CCtx* zc, ZSTD_Sequence* outSeqs,
+                                          size_t outSeqsSize, const void* src, size_t srcSize);
+
+/*! ZSTD_mergeBlockDelimiters() :
+ * Given an array of ZSTD_Sequence, remove all sequences that represent block delimiters/last literals
+ * by merging them into into the literals of the next sequence.
+ *
+ * As such, the final generated result has no explicit representation of block boundaries,
+ * and the final last literals segment is not represented in the sequences.
+ *
+ * The output of this function can be fed into ZSTD_compressSequences() with CCtx
+ * setting of ZSTD_c_blockDelimiters as ZSTD_sf_noBlockDelimiters
+ * @return : number of sequences left after merging
+ */
+ZSTDLIB_API size_t ZSTD_mergeBlockDelimiters(ZSTD_Sequence* sequences, size_t seqsSize);
+
+/*! ZSTD_compressSequences() :
+ * Compress an array of ZSTD_Sequence, generated from the original source buffer, into dst.
+ * If a dictionary is included, then the cctx should reference the dict. (see: ZSTD_CCtx_refCDict(), ZSTD_CCtx_loadDictionary(), etc.)
+ * The entire source is compressed into a single frame.
+ *
+ * The compression behavior changes based on cctx params. In particular:
+ *    If ZSTD_c_blockDelimiters == ZSTD_sf_noBlockDelimiters, the array of ZSTD_Sequence is expected to contain
+ *    no block delimiters (defined in ZSTD_Sequence). Block boundaries are roughly determined based on
+ *    the block size derived from the cctx, and sequences may be split. This is the default setting.
+ *
+ *    If ZSTD_c_blockDelimiters == ZSTD_sf_explicitBlockDelimiters, the array of ZSTD_Sequence is expected to contain
+ *    block delimiters (defined in ZSTD_Sequence). Behavior is undefined if no block delimiters are provided.
+ *
+ *    If ZSTD_c_validateSequences == 0, this function will blindly accept the sequences provided. Invalid sequences cause undefined
+ *    behavior. If ZSTD_c_validateSequences == 1, then if sequence is invalid (see doc/zstd_compression_format.md for
+ *    specifics regarding offset/matchlength requirements) then the function will bail out and return an error.
+ *
+ *    In addition to the two adjustable experimental params, there are other important cctx params.
+ *    - ZSTD_c_minMatch MUST be set as less than or equal to the smallest match generated by the match finder. It has a minimum value of ZSTD_MINMATCH_MIN.
+ *    - ZSTD_c_compressionLevel accordingly adjusts the strength of the entropy coder, as it would in typical compression.
+ *    - ZSTD_c_windowLog affects offset validation: this function will return an error at higher debug levels if a provided offset
+ *      is larger than what the spec allows for a given window log and dictionary (if present). See: doc/zstd_compression_format.md
+ *
+ * Note: Repcodes are, as of now, always re-calculated within this function, so ZSTD_Sequence::rep is unused.
+ * Note 2: Once we integrate ability to ingest repcodes, the explicit block delims mode must respect those repcodes exactly,
+ *         and cannot emit an RLE block that disagrees with the repcode history
+ * @return : final compressed size or a ZSTD error.
+ */
+ZSTDLIB_API size_t ZSTD_compressSequences(ZSTD_CCtx* const cctx, void* dst, size_t dstSize,
+                                  const ZSTD_Sequence* inSeqs, size_t inSeqsSize,
+                                  const void* src, size_t srcSize);
+
+
+/*! ZSTD_writeSkippableFrame() :
+ * Generates a zstd skippable frame containing data given by src, and writes it to dst buffer.
+ *
+ * Skippable frames begin with a a 4-byte magic number. There are 16 possible choices of magic number,
+ * ranging from ZSTD_MAGIC_SKIPPABLE_START to ZSTD_MAGIC_SKIPPABLE_START+15.
+ * As such, the parameter magicVariant controls the exact skippable frame magic number variant used, so
+ * the magic number used will be ZSTD_MAGIC_SKIPPABLE_START + magicVariant.
+ *
+ * Returns an error if destination buffer is not large enough, if the source size is not representable
+ * with a 4-byte unsigned int, or if the parameter magicVariant is greater than 15 (and therefore invalid).
+ *
+ * @return : number of bytes written or a ZSTD error.
+ */
+ZSTDLIB_API size_t ZSTD_writeSkippableFrame(void* dst, size_t dstCapacity,
+                                            const void* src, size_t srcSize, unsigned magicVariant);
+
+
+/* *************************************
+*  Memory management
+***************************************/
+
+/*! ZSTD_estimate*() :
+ *  These functions make it possible to estimate memory usage
+ *  of a future {D,C}Ctx, before its creation.
+ *
+ *  ZSTD_estimateCCtxSize() will provide a memory budget large enough
+ *  for any compression level up to selected one.
+ *  Note : Unlike ZSTD_estimateCStreamSize*(), this estimate
+ *         does not include space for a window buffer.
+ *         Therefore, the estimation is only guaranteed for single-shot compressions, not streaming.
+ *  The estimate will assume the input may be arbitrarily large,
+ *  which is the worst case.
+ *
+ *  When srcSize can be bound by a known and rather "small" value,
+ *  this fact can be used to provide a tighter estimation
+ *  because the CCtx compression context will need less memory.
+ *  This tighter estimation can be provided by more advanced functions
+ *  ZSTD_estimateCCtxSize_usingCParams(), which can be used in tandem with ZSTD_getCParams(),
+ *  and ZSTD_estimateCCtxSize_usingCCtxParams(), which can be used in tandem with ZSTD_CCtxParams_setParameter().
+ *  Both can be used to estimate memory using custom compression parameters and arbitrary srcSize limits.
+ *
+ *  Note 2 : only single-threaded compression is supported.
+ *  ZSTD_estimateCCtxSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
+ */
+ZSTDLIB_API size_t ZSTD_estimateCCtxSize(int compressionLevel);
+ZSTDLIB_API size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams);
+ZSTDLIB_API size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params);
+ZSTDLIB_API size_t ZSTD_estimateDCtxSize(void);
+
+/*! ZSTD_estimateCStreamSize() :
+ *  ZSTD_estimateCStreamSize() will provide a budget large enough for any compression level up to selected one.
+ *  It will also consider src size to be arbitrarily "large", which is worst case.
+ *  If srcSize is known to always be small, ZSTD_estimateCStreamSize_usingCParams() can provide a tighter estimation.
+ *  ZSTD_estimateCStreamSize_usingCParams() can be used in tandem with ZSTD_getCParams() to create cParams from compressionLevel.
+ *  ZSTD_estimateCStreamSize_usingCCtxParams() can be used in tandem with ZSTD_CCtxParams_setParameter(). Only single-threaded compression is supported. This function will return an error code if ZSTD_c_nbWorkers is >= 1.
+ *  Note : CStream size estimation is only correct for single-threaded compression.
+ *  ZSTD_DStream memory budget depends on window Size.
+ *  This information can be passed manually, using ZSTD_estimateDStreamSize,
+ *  or deducted from a valid frame Header, using ZSTD_estimateDStreamSize_fromFrame();
+ *  Note : if streaming is init with function ZSTD_init?Stream_usingDict(),
+ *         an internal ?Dict will be created, which additional size is not estimated here.
+ *         In this case, get total size by adding ZSTD_estimate?DictSize */
+ZSTDLIB_API size_t ZSTD_estimateCStreamSize(int compressionLevel);
+ZSTDLIB_API size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams);
+ZSTDLIB_API size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params);
+ZSTDLIB_API size_t ZSTD_estimateDStreamSize(size_t windowSize);
+ZSTDLIB_API size_t ZSTD_estimateDStreamSize_fromFrame(const void* src, size_t srcSize);
+
+/*! ZSTD_estimate?DictSize() :
+ *  ZSTD_estimateCDictSize() will bet that src size is relatively "small", and content is copied, like ZSTD_createCDict().
+ *  ZSTD_estimateCDictSize_advanced() makes it possible to control compression parameters precisely, like ZSTD_createCDict_advanced().
+ *  Note : dictionaries created by reference (`ZSTD_dlm_byRef`) are logically smaller.
+ */
+ZSTDLIB_API size_t ZSTD_estimateCDictSize(size_t dictSize, int compressionLevel);
+ZSTDLIB_API size_t ZSTD_estimateCDictSize_advanced(size_t dictSize, ZSTD_compressionParameters cParams, ZSTD_dictLoadMethod_e dictLoadMethod);
+ZSTDLIB_API size_t ZSTD_estimateDDictSize(size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod);
+
+/*! ZSTD_initStatic*() :
+ *  Initialize an object using a pre-allocated fixed-size buffer.
+ *  workspace: The memory area to emplace the object into.
+ *             Provided pointer *must be 8-bytes aligned*.
+ *             Buffer must outlive object.
+ *  workspaceSize: Use ZSTD_estimate*Size() to determine
+ *                 how large workspace must be to support target scenario.
+ * @return : pointer to object (same address as workspace, just different type),
+ *           or NULL if error (size too small, incorrect alignment, etc.)
+ *  Note : zstd will never resize nor malloc() when using a static buffer.
+ *         If the object requires more memory than available,
+ *         zstd will just error out (typically ZSTD_error_memory_allocation).
+ *  Note 2 : there is no corresponding "free" function.
+ *           Since workspace is allocated externally, it must be freed externally too.
+ *  Note 3 : cParams : use ZSTD_getCParams() to convert a compression level
+ *           into its associated cParams.
+ *  Limitation 1 : currently not compatible with internal dictionary creation, triggered by
+ *                 ZSTD_CCtx_loadDictionary(), ZSTD_initCStream_usingDict() or ZSTD_initDStream_usingDict().
+ *  Limitation 2 : static cctx currently not compatible with multi-threading.
+ *  Limitation 3 : static dctx is incompatible with legacy support.
+ */
+ZSTDLIB_API ZSTD_CCtx*    ZSTD_initStaticCCtx(void* workspace, size_t workspaceSize);
+ZSTDLIB_API ZSTD_CStream* ZSTD_initStaticCStream(void* workspace, size_t workspaceSize);    /*< same as ZSTD_initStaticCCtx() */
+
+ZSTDLIB_API ZSTD_DCtx*    ZSTD_initStaticDCtx(void* workspace, size_t workspaceSize);
+ZSTDLIB_API ZSTD_DStream* ZSTD_initStaticDStream(void* workspace, size_t workspaceSize);    /*< same as ZSTD_initStaticDCtx() */
+
+ZSTDLIB_API const ZSTD_CDict* ZSTD_initStaticCDict(
+                                        void* workspace, size_t workspaceSize,
+                                        const void* dict, size_t dictSize,
+                                        ZSTD_dictLoadMethod_e dictLoadMethod,
+                                        ZSTD_dictContentType_e dictContentType,
+                                        ZSTD_compressionParameters cParams);
+
+ZSTDLIB_API const ZSTD_DDict* ZSTD_initStaticDDict(
+                                        void* workspace, size_t workspaceSize,
+                                        const void* dict, size_t dictSize,
+                                        ZSTD_dictLoadMethod_e dictLoadMethod,
+                                        ZSTD_dictContentType_e dictContentType);
+
+
+/*! Custom memory allocation :
+ *  These prototypes make it possible to pass your own allocation/free functions.
+ *  ZSTD_customMem is provided at creation time, using ZSTD_create*_advanced() variants listed below.
+ *  All allocation/free operations will be completed using these custom variants instead of regular <stdlib.h> ones.
+ */
+typedef void* (*ZSTD_allocFunction) (void* opaque, size_t size);
+typedef void  (*ZSTD_freeFunction) (void* opaque, void* address);
+typedef struct { ZSTD_allocFunction customAlloc; ZSTD_freeFunction customFree; void* opaque; } ZSTD_customMem;
+static
+__attribute__((__unused__))
+ZSTD_customMem const ZSTD_defaultCMem = { NULL, NULL, NULL };  /*< this constant defers to stdlib's functions */
+
+ZSTDLIB_API ZSTD_CCtx*    ZSTD_createCCtx_advanced(ZSTD_customMem customMem);
+ZSTDLIB_API ZSTD_CStream* ZSTD_createCStream_advanced(ZSTD_customMem customMem);
+ZSTDLIB_API ZSTD_DCtx*    ZSTD_createDCtx_advanced(ZSTD_customMem customMem);
+ZSTDLIB_API ZSTD_DStream* ZSTD_createDStream_advanced(ZSTD_customMem customMem);
+
+ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict_advanced(const void* dict, size_t dictSize,
+                                                  ZSTD_dictLoadMethod_e dictLoadMethod,
+                                                  ZSTD_dictContentType_e dictContentType,
+                                                  ZSTD_compressionParameters cParams,
+                                                  ZSTD_customMem customMem);
+
+/* ! Thread pool :
+ * These prototypes make it possible to share a thread pool among multiple compression contexts.
+ * This can limit resources for applications with multiple threads where each one uses
+ * a threaded compression mode (via ZSTD_c_nbWorkers parameter).
+ * ZSTD_createThreadPool creates a new thread pool with a given number of threads.
+ * Note that the lifetime of such pool must exist while being used.
+ * ZSTD_CCtx_refThreadPool assigns a thread pool to a context (use NULL argument value
+ * to use an internal thread pool).
+ * ZSTD_freeThreadPool frees a thread pool, accepts NULL pointer.
+ */
+typedef struct POOL_ctx_s ZSTD_threadPool;
+ZSTDLIB_API ZSTD_threadPool* ZSTD_createThreadPool(size_t numThreads);
+ZSTDLIB_API void ZSTD_freeThreadPool (ZSTD_threadPool* pool);  /* accept NULL pointer */
+ZSTDLIB_API size_t ZSTD_CCtx_refThreadPool(ZSTD_CCtx* cctx, ZSTD_threadPool* pool);
+
+
+/*
+ * This API is temporary and is expected to change or disappear in the future!
+ */
+ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict_advanced2(
+    const void* dict, size_t dictSize,
+    ZSTD_dictLoadMethod_e dictLoadMethod,
+    ZSTD_dictContentType_e dictContentType,
+    const ZSTD_CCtx_params* cctxParams,
+    ZSTD_customMem customMem);
+
+ZSTDLIB_API ZSTD_DDict* ZSTD_createDDict_advanced(
+    const void* dict, size_t dictSize,
+    ZSTD_dictLoadMethod_e dictLoadMethod,
+    ZSTD_dictContentType_e dictContentType,
+    ZSTD_customMem customMem);
+
+
+/* *************************************
+*  Advanced compression functions
+***************************************/
+
+/*! ZSTD_createCDict_byReference() :
+ *  Create a digested dictionary for compression
+ *  Dictionary content is just referenced, not duplicated.
+ *  As a consequence, `dictBuffer` **must** outlive CDict,
+ *  and its content must remain unmodified throughout the lifetime of CDict.
+ *  note: equivalent to ZSTD_createCDict_advanced(), with dictLoadMethod==ZSTD_dlm_byRef */
+ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict_byReference(const void* dictBuffer, size_t dictSize, int compressionLevel);
+
+/*! ZSTD_getDictID_fromCDict() :
+ *  Provides the dictID of the dictionary loaded into `cdict`.
+ *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
+ *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
+ZSTDLIB_API unsigned ZSTD_getDictID_fromCDict(const ZSTD_CDict* cdict);
+
+/*! ZSTD_getCParams() :
+ * @return ZSTD_compressionParameters structure for a selected compression level and estimated srcSize.
+ * `estimatedSrcSize` value is optional, select 0 if not known */
+ZSTDLIB_API ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long estimatedSrcSize, size_t dictSize);
+
+/*! ZSTD_getParams() :
+ *  same as ZSTD_getCParams(), but @return a full `ZSTD_parameters` object instead of sub-component `ZSTD_compressionParameters`.
+ *  All fields of `ZSTD_frameParameters` are set to default : contentSize=1, checksum=0, noDictID=0 */
+ZSTDLIB_API ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long estimatedSrcSize, size_t dictSize);
+
+/*! ZSTD_checkCParams() :
+ *  Ensure param values remain within authorized range.
+ * @return 0 on success, or an error code (can be checked with ZSTD_isError()) */
+ZSTDLIB_API size_t ZSTD_checkCParams(ZSTD_compressionParameters params);
+
+/*! ZSTD_adjustCParams() :
+ *  optimize params for a given `srcSize` and `dictSize`.
+ * `srcSize` can be unknown, in which case use ZSTD_CONTENTSIZE_UNKNOWN.
+ * `dictSize` must be `0` when there is no dictionary.
+ *  cPar can be invalid : all parameters will be clamped within valid range in the @return struct.
+ *  This function never fails (wide contract) */
+ZSTDLIB_API ZSTD_compressionParameters ZSTD_adjustCParams(ZSTD_compressionParameters cPar, unsigned long long srcSize, size_t dictSize);
+
+/*! ZSTD_compress_advanced() :
+ *  Note : this function is now DEPRECATED.
+ *         It can be replaced by ZSTD_compress2(), in combination with ZSTD_CCtx_setParameter() and other parameter setters.
+ *  This prototype will be marked as deprecated and generate compilation warning on reaching v1.5.x */
+ZSTDLIB_API size_t ZSTD_compress_advanced(ZSTD_CCtx* cctx,
+                                          void* dst, size_t dstCapacity,
+                                    const void* src, size_t srcSize,
+                                    const void* dict,size_t dictSize,
+                                          ZSTD_parameters params);
+
+/*! ZSTD_compress_usingCDict_advanced() :
+ *  Note : this function is now REDUNDANT.
+ *         It can be replaced by ZSTD_compress2(), in combination with ZSTD_CCtx_loadDictionary() and other parameter setters.
+ *  This prototype will be marked as deprecated and generate compilation warning in some future version */
+ZSTDLIB_API size_t ZSTD_compress_usingCDict_advanced(ZSTD_CCtx* cctx,
+                                              void* dst, size_t dstCapacity,
+                                        const void* src, size_t srcSize,
+                                        const ZSTD_CDict* cdict,
+                                              ZSTD_frameParameters fParams);
+
+
+/*! ZSTD_CCtx_loadDictionary_byReference() :
+ *  Same as ZSTD_CCtx_loadDictionary(), but dictionary content is referenced, instead of being copied into CCtx.
+ *  It saves some memory, but also requires that `dict` outlives its usage within `cctx` */
+ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_byReference(ZSTD_CCtx* cctx, const void* dict, size_t dictSize);
+
+/*! ZSTD_CCtx_loadDictionary_advanced() :
+ *  Same as ZSTD_CCtx_loadDictionary(), but gives finer control over
+ *  how to load the dictionary (by copy ? by reference ?)
+ *  and how to interpret it (automatic ? force raw mode ? full mode only ?) */
+ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_advanced(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType);
+
+/*! ZSTD_CCtx_refPrefix_advanced() :
+ *  Same as ZSTD_CCtx_refPrefix(), but gives finer control over
+ *  how to interpret prefix content (automatic ? force raw mode (default) ? full mode only ?) */
+ZSTDLIB_API size_t ZSTD_CCtx_refPrefix_advanced(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType);
+
+/* ===   experimental parameters   === */
+/* these parameters can be used with ZSTD_setParameter()
+ * they are not guaranteed to remain supported in the future */
+
+ /* Enables rsyncable mode,
+  * which makes compressed files more rsync friendly
+  * by adding periodic synchronization points to the compressed data.
+  * The target average block size is ZSTD_c_jobSize / 2.
+  * It's possible to modify the job size to increase or decrease
+  * the granularity of the synchronization point.
+  * Once the jobSize is smaller than the window size,
+  * it will result in compression ratio degradation.
+  * NOTE 1: rsyncable mode only works when multithreading is enabled.
+  * NOTE 2: rsyncable performs poorly in combination with long range mode,
+  * since it will decrease the effectiveness of synchronization points,
+  * though mileage may vary.
+  * NOTE 3: Rsyncable mode limits maximum compression speed to ~400 MB/s.
+  * If the selected compression level is already running significantly slower,
+  * the overall speed won't be significantly impacted.
+  */
+ #define ZSTD_c_rsyncable ZSTD_c_experimentalParam1
+
+/* Select a compression format.
+ * The value must be of type ZSTD_format_e.
+ * See ZSTD_format_e enum definition for details */
+#define ZSTD_c_format ZSTD_c_experimentalParam2
+
+/* Force back-reference distances to remain < windowSize,
+ * even when referencing into Dictionary content (default:0) */
+#define ZSTD_c_forceMaxWindow ZSTD_c_experimentalParam3
+
+/* Controls whether the contents of a CDict
+ * are used in place, or copied into the working context.
+ * Accepts values from the ZSTD_dictAttachPref_e enum.
+ * See the comments on that enum for an explanation of the feature. */
+#define ZSTD_c_forceAttachDict ZSTD_c_experimentalParam4
+
+/* Controls how the literals are compressed (default is auto).
+ * The value must be of type ZSTD_literalCompressionMode_e.
+ * See ZSTD_literalCompressionMode_t enum definition for details.
+ */
+#define ZSTD_c_literalCompressionMode ZSTD_c_experimentalParam5
+
+/* Tries to fit compressed block size to be around targetCBlockSize.
+ * No target when targetCBlockSize == 0.
+ * There is no guarantee on compressed block size (default:0) */
+#define ZSTD_c_targetCBlockSize ZSTD_c_experimentalParam6
+
+/* User's best guess of source size.
+ * Hint is not valid when srcSizeHint == 0.
+ * There is no guarantee that hint is close to actual source size,
+ * but compression ratio may regress significantly if guess considerably underestimates */
+#define ZSTD_c_srcSizeHint ZSTD_c_experimentalParam7
+
+/* Controls whether the new and experimental "dedicated dictionary search
+ * structure" can be used. This feature is still rough around the edges, be
+ * prepared for surprising behavior!
+ *
+ * How to use it:
+ *
+ * When using a CDict, whether to use this feature or not is controlled at
+ * CDict creation, and it must be set in a CCtxParams set passed into that
+ * construction (via ZSTD_createCDict_advanced2()). A compression will then
+ * use the feature or not based on how the CDict was constructed; the value of
+ * this param, set in the CCtx, will have no effect.
+ *
+ * However, when a dictionary buffer is passed into a CCtx, such as via
+ * ZSTD_CCtx_loadDictionary(), this param can be set on the CCtx to control
+ * whether the CDict that is created internally can use the feature or not.
+ *
+ * What it does:
+ *
+ * Normally, the internal data structures of the CDict are analogous to what
+ * would be stored in a CCtx after compressing the contents of a dictionary.
+ * To an approximation, a compression using a dictionary can then use those
+ * data structures to simply continue what is effectively a streaming
+ * compression where the simulated compression of the dictionary left off.
+ * Which is to say, the search structures in the CDict are normally the same
+ * format as in the CCtx.
+ *
+ * It is possible to do better, since the CDict is not like a CCtx: the search
+ * structures are written once during CDict creation, and then are only read
+ * after that, while the search structures in the CCtx are both read and
+ * written as the compression goes along. This means we can choose a search
+ * structure for the dictionary that is read-optimized.
+ *
+ * This feature enables the use of that different structure.
+ *
+ * Note that some of the members of the ZSTD_compressionParameters struct have
+ * different semantics and constraints in the dedicated search structure. It is
+ * highly recommended that you simply set a compression level in the CCtxParams
+ * you pass into the CDict creation call, and avoid messing with the cParams
+ * directly.
+ *
+ * Effects:
+ *
+ * This will only have any effect when the selected ZSTD_strategy
+ * implementation supports this feature. Currently, that's limited to
+ * ZSTD_greedy, ZSTD_lazy, and ZSTD_lazy2.
+ *
+ * Note that this means that the CDict tables can no longer be copied into the
+ * CCtx, so the dict attachment mode ZSTD_dictForceCopy will no longer be
+ * useable. The dictionary can only be attached or reloaded.
+ *
+ * In general, you should expect compression to be faster--sometimes very much
+ * so--and CDict creation to be slightly slower. Eventually, we will probably
+ * make this mode the default.
+ */
+#define ZSTD_c_enableDedicatedDictSearch ZSTD_c_experimentalParam8
+
+/* ZSTD_c_stableInBuffer
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable.
+ *
+ * Tells the compressor that the ZSTD_inBuffer will ALWAYS be the same
+ * between calls, except for the modifications that zstd makes to pos (the
+ * caller must not modify pos). This is checked by the compressor, and
+ * compression will fail if it ever changes. This means the only flush
+ * mode that makes sense is ZSTD_e_end, so zstd will error if ZSTD_e_end
+ * is not used. The data in the ZSTD_inBuffer in the range [src, src + pos)
+ * MUST not be modified during compression or you will get data corruption.
+ *
+ * When this flag is enabled zstd won't allocate an input window buffer,
+ * because the user guarantees it can reference the ZSTD_inBuffer until
+ * the frame is complete. But, it will still allocate an output buffer
+ * large enough to fit a block (see ZSTD_c_stableOutBuffer). This will also
+ * avoid the memcpy() from the input buffer to the input window buffer.
+ *
+ * NOTE: ZSTD_compressStream2() will error if ZSTD_e_end is not used.
+ * That means this flag cannot be used with ZSTD_compressStream().
+ *
+ * NOTE: So long as the ZSTD_inBuffer always points to valid memory, using
+ * this flag is ALWAYS memory safe, and will never access out-of-bounds
+ * memory. However, compression WILL fail if you violate the preconditions.
+ *
+ * WARNING: The data in the ZSTD_inBuffer in the range [dst, dst + pos) MUST
+ * not be modified during compression or you will get data corruption. This
+ * is because zstd needs to reference data in the ZSTD_inBuffer to find
+ * matches. Normally zstd maintains its own window buffer for this purpose,
+ * but passing this flag tells zstd to use the user provided buffer.
+ */
+#define ZSTD_c_stableInBuffer ZSTD_c_experimentalParam9
+
+/* ZSTD_c_stableOutBuffer
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable.
+ *
+ * Tells he compressor that the ZSTD_outBuffer will not be resized between
+ * calls. Specifically: (out.size - out.pos) will never grow. This gives the
+ * compressor the freedom to say: If the compressed data doesn't fit in the
+ * output buffer then return ZSTD_error_dstSizeTooSmall. This allows us to
+ * always decompress directly into the output buffer, instead of decompressing
+ * into an internal buffer and copying to the output buffer.
+ *
+ * When this flag is enabled zstd won't allocate an output buffer, because
+ * it can write directly to the ZSTD_outBuffer. It will still allocate the
+ * input window buffer (see ZSTD_c_stableInBuffer).
+ *
+ * Zstd will check that (out.size - out.pos) never grows and return an error
+ * if it does. While not strictly necessary, this should prevent surprises.
+ */
+#define ZSTD_c_stableOutBuffer ZSTD_c_experimentalParam10
+
+/* ZSTD_c_blockDelimiters
+ * Default is 0 == ZSTD_sf_noBlockDelimiters.
+ *
+ * For use with sequence compression API: ZSTD_compressSequences().
+ *
+ * Designates whether or not the given array of ZSTD_Sequence contains block delimiters
+ * and last literals, which are defined as sequences with offset == 0 and matchLength == 0.
+ * See the definition of ZSTD_Sequence for more specifics.
+ */
+#define ZSTD_c_blockDelimiters ZSTD_c_experimentalParam11
+
+/* ZSTD_c_validateSequences
+ * Default is 0 == disabled. Set to 1 to enable sequence validation.
+ *
+ * For use with sequence compression API: ZSTD_compressSequences().
+ * Designates whether or not we validate sequences provided to ZSTD_compressSequences()
+ * during function execution.
+ *
+ * Without validation, providing a sequence that does not conform to the zstd spec will cause
+ * undefined behavior, and may produce a corrupted block.
+ *
+ * With validation enabled, a if sequence is invalid (see doc/zstd_compression_format.md for
+ * specifics regarding offset/matchlength requirements) then the function will bail out and
+ * return an error.
+ *
+ */
+#define ZSTD_c_validateSequences ZSTD_c_experimentalParam12
+
+/*! ZSTD_CCtx_getParameter() :
+ *  Get the requested compression parameter value, selected by enum ZSTD_cParameter,
+ *  and store it into int* value.
+ * @return : 0, or an error code (which can be tested with ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_getParameter(const ZSTD_CCtx* cctx, ZSTD_cParameter param, int* value);
+
+
+/*! ZSTD_CCtx_params :
+ *  Quick howto :
+ *  - ZSTD_createCCtxParams() : Create a ZSTD_CCtx_params structure
+ *  - ZSTD_CCtxParams_setParameter() : Push parameters one by one into
+ *                                     an existing ZSTD_CCtx_params structure.
+ *                                     This is similar to
+ *                                     ZSTD_CCtx_setParameter().
+ *  - ZSTD_CCtx_setParametersUsingCCtxParams() : Apply parameters to
+ *                                    an existing CCtx.
+ *                                    These parameters will be applied to
+ *                                    all subsequent frames.
+ *  - ZSTD_compressStream2() : Do compression using the CCtx.
+ *  - ZSTD_freeCCtxParams() : Free the memory, accept NULL pointer.
+ *
+ *  This can be used with ZSTD_estimateCCtxSize_advanced_usingCCtxParams()
+ *  for static allocation of CCtx for single-threaded compression.
+ */
+ZSTDLIB_API ZSTD_CCtx_params* ZSTD_createCCtxParams(void);
+ZSTDLIB_API size_t ZSTD_freeCCtxParams(ZSTD_CCtx_params* params);  /* accept NULL pointer */
+
+/*! ZSTD_CCtxParams_reset() :
+ *  Reset params to default values.
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_reset(ZSTD_CCtx_params* params);
+
+/*! ZSTD_CCtxParams_init() :
+ *  Initializes the compression parameters of cctxParams according to
+ *  compression level. All other parameters are reset to their default values.
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel);
+
+/*! ZSTD_CCtxParams_init_advanced() :
+ *  Initializes the compression and frame parameters of cctxParams according to
+ *  params. All other parameters are reset to their default values.
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_init_advanced(ZSTD_CCtx_params* cctxParams, ZSTD_parameters params);
+
+/*! ZSTD_CCtxParams_setParameter() :
+ *  Similar to ZSTD_CCtx_setParameter.
+ *  Set one compression parameter, selected by enum ZSTD_cParameter.
+ *  Parameters must be applied to a ZSTD_CCtx using
+ *  ZSTD_CCtx_setParametersUsingCCtxParams().
+ * @result : a code representing success or failure (which can be tested with
+ *           ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* params, ZSTD_cParameter param, int value);
+
+/*! ZSTD_CCtxParams_getParameter() :
+ * Similar to ZSTD_CCtx_getParameter.
+ * Get the requested value of one compression parameter, selected by enum ZSTD_cParameter.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_getParameter(const ZSTD_CCtx_params* params, ZSTD_cParameter param, int* value);
+
+/*! ZSTD_CCtx_setParametersUsingCCtxParams() :
+ *  Apply a set of ZSTD_CCtx_params to the compression context.
+ *  This can be done even after compression is started,
+ *    if nbWorkers==0, this will have no impact until a new compression is started.
+ *    if nbWorkers>=1, new parameters will be picked up at next job,
+ *       with a few restrictions (windowLog, pledgedSrcSize, nbWorkers, jobSize, and overlapLog are not updated).
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_setParametersUsingCCtxParams(
+        ZSTD_CCtx* cctx, const ZSTD_CCtx_params* params);
+
+/*! ZSTD_compressStream2_simpleArgs() :
+ *  Same as ZSTD_compressStream2(),
+ *  but using only integral types as arguments.
+ *  This variant might be helpful for binders from dynamic languages
+ *  which have troubles handling structures containing memory pointers.
+ */
+ZSTDLIB_API size_t ZSTD_compressStream2_simpleArgs (
+                            ZSTD_CCtx* cctx,
+                            void* dst, size_t dstCapacity, size_t* dstPos,
+                      const void* src, size_t srcSize, size_t* srcPos,
+                            ZSTD_EndDirective endOp);
+
+
+/* *************************************
+*  Advanced decompression functions
+***************************************/
+
+/*! ZSTD_isFrame() :
+ *  Tells if the content of `buffer` starts with a valid Frame Identifier.
+ *  Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
+ *  Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
+ *  Note 3 : Skippable Frame Identifiers are considered valid. */
+ZSTDLIB_API unsigned ZSTD_isFrame(const void* buffer, size_t size);
+
+/*! ZSTD_createDDict_byReference() :
+ *  Create a digested dictionary, ready to start decompression operation without startup delay.
+ *  Dictionary content is referenced, and therefore stays in dictBuffer.
+ *  It is important that dictBuffer outlives DDict,
+ *  it must remain read accessible throughout the lifetime of DDict */
+ZSTDLIB_API ZSTD_DDict* ZSTD_createDDict_byReference(const void* dictBuffer, size_t dictSize);
+
+/*! ZSTD_DCtx_loadDictionary_byReference() :
+ *  Same as ZSTD_DCtx_loadDictionary(),
+ *  but references `dict` content instead of copying it into `dctx`.
+ *  This saves memory if `dict` remains around.,
+ *  However, it's imperative that `dict` remains accessible (and unmodified) while being used, so it must outlive decompression. */
+ZSTDLIB_API size_t ZSTD_DCtx_loadDictionary_byReference(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);
+
+/*! ZSTD_DCtx_loadDictionary_advanced() :
+ *  Same as ZSTD_DCtx_loadDictionary(),
+ *  but gives direct control over
+ *  how to load the dictionary (by copy ? by reference ?)
+ *  and how to interpret it (automatic ? force raw mode ? full mode only ?). */
+ZSTDLIB_API size_t ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx* dctx, const void* dict, size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType);
+
+/*! ZSTD_DCtx_refPrefix_advanced() :
+ *  Same as ZSTD_DCtx_refPrefix(), but gives finer control over
+ *  how to interpret prefix content (automatic ? force raw mode (default) ? full mode only ?) */
+ZSTDLIB_API size_t ZSTD_DCtx_refPrefix_advanced(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType);
+
+/*! ZSTD_DCtx_setMaxWindowSize() :
+ *  Refuses allocating internal buffers for frames requiring a window size larger than provided limit.
+ *  This protects a decoder context from reserving too much memory for itself (potential attack scenario).
+ *  This parameter is only useful in streaming mode, since no internal buffer is allocated in single-pass mode.
+ *  By default, a decompression context accepts all window sizes <= (1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT)
+ * @return : 0, or an error code (which can be tested using ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_setMaxWindowSize(ZSTD_DCtx* dctx, size_t maxWindowSize);
+
+/*! ZSTD_DCtx_getParameter() :
+ *  Get the requested decompression parameter value, selected by enum ZSTD_dParameter,
+ *  and store it into int* value.
+ * @return : 0, or an error code (which can be tested with ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_getParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int* value);
+
+/* ZSTD_d_format
+ * experimental parameter,
+ * allowing selection between ZSTD_format_e input compression formats
+ */
+#define ZSTD_d_format ZSTD_d_experimentalParam1
+/* ZSTD_d_stableOutBuffer
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable.
+ *
+ * Tells the decompressor that the ZSTD_outBuffer will ALWAYS be the same
+ * between calls, except for the modifications that zstd makes to pos (the
+ * caller must not modify pos). This is checked by the decompressor, and
+ * decompression will fail if it ever changes. Therefore the ZSTD_outBuffer
+ * MUST be large enough to fit the entire decompressed frame. This will be
+ * checked when the frame content size is known. The data in the ZSTD_outBuffer
+ * in the range [dst, dst + pos) MUST not be modified during decompression
+ * or you will get data corruption.
+ *
+ * When this flags is enabled zstd won't allocate an output buffer, because
+ * it can write directly to the ZSTD_outBuffer, but it will still allocate
+ * an input buffer large enough to fit any compressed block. This will also
+ * avoid the memcpy() from the internal output buffer to the ZSTD_outBuffer.
+ * If you need to avoid the input buffer allocation use the buffer-less
+ * streaming API.
+ *
+ * NOTE: So long as the ZSTD_outBuffer always points to valid memory, using
+ * this flag is ALWAYS memory safe, and will never access out-of-bounds
+ * memory. However, decompression WILL fail if you violate the preconditions.
+ *
+ * WARNING: The data in the ZSTD_outBuffer in the range [dst, dst + pos) MUST
+ * not be modified during decompression or you will get data corruption. This
+ * is because zstd needs to reference data in the ZSTD_outBuffer to regenerate
+ * matches. Normally zstd maintains its own buffer for this purpose, but passing
+ * this flag tells zstd to use the user provided buffer.
+ */
+#define ZSTD_d_stableOutBuffer ZSTD_d_experimentalParam2
+
+/* ZSTD_d_forceIgnoreChecksum
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable
+ *
+ * Tells the decompressor to skip checksum validation during decompression, regardless
+ * of whether checksumming was specified during compression. This offers some
+ * slight performance benefits, and may be useful for debugging.
+ * Param has values of type ZSTD_forceIgnoreChecksum_e
+ */
+#define ZSTD_d_forceIgnoreChecksum ZSTD_d_experimentalParam3
+
+/* ZSTD_d_refMultipleDDicts
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable
+ *
+ * If enabled and dctx is allocated on the heap, then additional memory will be allocated
+ * to store references to multiple ZSTD_DDict. That is, multiple calls of ZSTD_refDDict()
+ * using a given ZSTD_DCtx, rather than overwriting the previous DDict reference, will instead
+ * store all references. At decompression time, the appropriate dictID is selected
+ * from the set of DDicts based on the dictID in the frame.
+ *
+ * Usage is simply calling ZSTD_refDDict() on multiple dict buffers.
+ *
+ * Param has values of byte ZSTD_refMultipleDDicts_e
+ *
+ * WARNING: Enabling this parameter and calling ZSTD_DCtx_refDDict(), will trigger memory
+ * allocation for the hash table. ZSTD_freeDCtx() also frees this memory.
+ * Memory is allocated as per ZSTD_DCtx::customMem.
+ *
+ * Although this function allocates memory for the table, the user is still responsible for
+ * memory management of the underlying ZSTD_DDict* themselves.
+ */
+#define ZSTD_d_refMultipleDDicts ZSTD_d_experimentalParam4
+
+
+/*! ZSTD_DCtx_setFormat() :
+ *  Instruct the decoder context about what kind of data to decode next.
+ *  This instruction is mandatory to decode data without a fully-formed header,
+ *  such ZSTD_f_zstd1_magicless for example.
+ * @return : 0, or an error code (which can be tested using ZSTD_isError()). */
+ZSTDLIB_API size_t ZSTD_DCtx_setFormat(ZSTD_DCtx* dctx, ZSTD_format_e format);
+
+/*! ZSTD_decompressStream_simpleArgs() :
+ *  Same as ZSTD_decompressStream(),
+ *  but using only integral types as arguments.
+ *  This can be helpful for binders from dynamic languages
+ *  which have troubles handling structures containing memory pointers.
+ */
+ZSTDLIB_API size_t ZSTD_decompressStream_simpleArgs (
+                            ZSTD_DCtx* dctx,
+                            void* dst, size_t dstCapacity, size_t* dstPos,
+                      const void* src, size_t srcSize, size_t* srcPos);
+
+
+/* ******************************************************************
+*  Advanced streaming functions
+*  Warning : most of these functions are now redundant with the Advanced API.
+*  Once Advanced API reaches "stable" status,
+*  redundant functions will be deprecated, and then at some point removed.
+********************************************************************/
+
+/*=====   Advanced Streaming compression functions  =====*/
+
+/*! ZSTD_initCStream_srcSize() :
+ * This function is deprecated, and equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_refCDict(zcs, NULL); // clear the dictionary (if any)
+ *     ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
+ *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
+ *
+ * pledgedSrcSize must be correct. If it is not known at init time, use
+ * ZSTD_CONTENTSIZE_UNKNOWN. Note that, for compatibility with older programs,
+ * "0" also disables frame content size field. It may be enabled in the future.
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t
+ZSTD_initCStream_srcSize(ZSTD_CStream* zcs,
+                         int compressionLevel,
+                         unsigned long long pledgedSrcSize);
+
+/*! ZSTD_initCStream_usingDict() :
+ * This function is deprecated, and is equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
+ *     ZSTD_CCtx_loadDictionary(zcs, dict, dictSize);
+ *
+ * Creates of an internal CDict (incompatible with static CCtx), except if
+ * dict == NULL or dictSize < 8, in which case no dict is used.
+ * Note: dict is loaded with ZSTD_dct_auto (treated as a full zstd dictionary if
+ * it begins with ZSTD_MAGIC_DICTIONARY, else as raw content) and ZSTD_dlm_byCopy.
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t
+ZSTD_initCStream_usingDict(ZSTD_CStream* zcs,
+                     const void* dict, size_t dictSize,
+                           int compressionLevel);
+
+/*! ZSTD_initCStream_advanced() :
+ * This function is deprecated, and is approximately equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     // Pseudocode: Set each zstd parameter and leave the rest as-is.
+ *     for ((param, value) : params) {
+ *         ZSTD_CCtx_setParameter(zcs, param, value);
+ *     }
+ *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
+ *     ZSTD_CCtx_loadDictionary(zcs, dict, dictSize);
+ *
+ * dict is loaded with ZSTD_dct_auto and ZSTD_dlm_byCopy.
+ * pledgedSrcSize must be correct.
+ * If srcSize is not known at init time, use value ZSTD_CONTENTSIZE_UNKNOWN.
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t
+ZSTD_initCStream_advanced(ZSTD_CStream* zcs,
+                    const void* dict, size_t dictSize,
+                          ZSTD_parameters params,
+                          unsigned long long pledgedSrcSize);
+
+/*! ZSTD_initCStream_usingCDict() :
+ * This function is deprecated, and equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_refCDict(zcs, cdict);
+ *
+ * note : cdict will just be referenced, and must outlive compression session
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict);
+
+/*! ZSTD_initCStream_usingCDict_advanced() :
+ *   This function is DEPRECATED, and is approximately equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     // Pseudocode: Set each zstd frame parameter and leave the rest as-is.
+ *     for ((fParam, value) : fParams) {
+ *         ZSTD_CCtx_setParameter(zcs, fParam, value);
+ *     }
+ *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
+ *     ZSTD_CCtx_refCDict(zcs, cdict);
+ *
+ * same as ZSTD_initCStream_usingCDict(), with control over frame parameters.
+ * pledgedSrcSize must be correct. If srcSize is not known at init time, use
+ * value ZSTD_CONTENTSIZE_UNKNOWN.
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t
+ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs,
+                               const ZSTD_CDict* cdict,
+                                     ZSTD_frameParameters fParams,
+                                     unsigned long long pledgedSrcSize);
+
+/*! ZSTD_resetCStream() :
+ * This function is deprecated, and is equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
+ *
+ *  start a new frame, using same parameters from previous frame.
+ *  This is typically useful to skip dictionary loading stage, since it will re-use it in-place.
+ *  Note that zcs must be init at least once before using ZSTD_resetCStream().
+ *  If pledgedSrcSize is not known at reset time, use macro ZSTD_CONTENTSIZE_UNKNOWN.
+ *  If pledgedSrcSize > 0, its value must be correct, as it will be written in header, and controlled at the end.
+ *  For the time being, pledgedSrcSize==0 is interpreted as "srcSize unknown" for compatibility with older programs,
+ *  but it will change to mean "empty" in future version, so use macro ZSTD_CONTENTSIZE_UNKNOWN instead.
+ * @return : 0, or an error code (which can be tested using ZSTD_isError())
+ *  Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize);
+
+
+typedef struct {
+    unsigned long long ingested;   /* nb input bytes read and buffered */
+    unsigned long long consumed;   /* nb input bytes actually compressed */
+    unsigned long long produced;   /* nb of compressed bytes generated and buffered */
+    unsigned long long flushed;    /* nb of compressed bytes flushed : not provided; can be tracked from caller side */
+    unsigned currentJobID;         /* MT only : latest started job nb */
+    unsigned nbActiveWorkers;      /* MT only : nb of workers actively compressing at probe time */
+} ZSTD_frameProgression;
+
+/* ZSTD_getFrameProgression() :
+ * tells how much data has been ingested (read from input)
+ * consumed (input actually compressed) and produced (output) for current frame.
+ * Note : (ingested - consumed) is amount of input data buffered internally, not yet compressed.
+ * Aggregates progression inside active worker threads.
+ */
+ZSTDLIB_API ZSTD_frameProgression ZSTD_getFrameProgression(const ZSTD_CCtx* cctx);
+
+/*! ZSTD_toFlushNow() :
+ *  Tell how many bytes are ready to be flushed immediately.
+ *  Useful for multithreading scenarios (nbWorkers >= 1).
+ *  Probe the oldest active job, defined as oldest job not yet entirely flushed,
+ *  and check its output buffer.
+ * @return : amount of data stored in oldest job and ready to be flushed immediately.
+ *  if @return == 0, it means either :
+ *  + there is no active job (could be checked with ZSTD_frameProgression()), or
+ *  + oldest job is still actively compressing data,
+ *    but everything it has produced has also been flushed so far,
+ *    therefore flush speed is limited by production speed of oldest job
+ *    irrespective of the speed of concurrent (and newer) jobs.
+ */
+ZSTDLIB_API size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx);
+
+
+/*=====   Advanced Streaming decompression functions  =====*/
+
+/*!
+ * This function is deprecated, and is equivalent to:
+ *
+ *     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
+ *     ZSTD_DCtx_loadDictionary(zds, dict, dictSize);
+ *
+ * note: no dictionary will be used if dict == NULL or dictSize < 8
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize);
+
+/*!
+ * This function is deprecated, and is equivalent to:
+ *
+ *     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
+ *     ZSTD_DCtx_refDDict(zds, ddict);
+ *
+ * note : ddict is referenced, it must outlive decompression session
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* zds, const ZSTD_DDict* ddict);
+
+/*!
+ * This function is deprecated, and is equivalent to:
+ *
+ *     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
+ *
+ * re-use decompression parameters from previous init; saves dictionary loading
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_resetDStream(ZSTD_DStream* zds);
+
+
+/* *******************************************************************
+*  Buffer-less and synchronous inner streaming functions
+*
+*  This is an advanced API, giving full control over buffer management, for users which need direct control over memory.
+*  But it's also a complex one, with several restrictions, documented below.
+*  Prefer normal streaming API for an easier experience.
+********************************************************************* */
+
+/*
+  Buffer-less streaming compression (synchronous mode)
+
+  A ZSTD_CCtx object is required to track streaming operations.
+  Use ZSTD_createCCtx() / ZSTD_freeCCtx() to manage resource.
+  ZSTD_CCtx object can be re-used multiple times within successive compression operations.
+
+  Start by initializing a context.
+  Use ZSTD_compressBegin(), or ZSTD_compressBegin_usingDict() for dictionary compression,
+  or ZSTD_compressBegin_advanced(), for finer parameter control.
+  It's also possible to duplicate a reference context which has already been initialized, using ZSTD_copyCCtx()
+
+  Then, consume your input using ZSTD_compressContinue().
+  There are some important considerations to keep in mind when using this advanced function :
+  - ZSTD_compressContinue() has no internal buffer. It uses externally provided buffers only.
+  - Interface is synchronous : input is consumed entirely and produces 1+ compressed blocks.
+  - Caller must ensure there is enough space in `dst` to store compressed data under worst case scenario.
+    Worst case evaluation is provided by ZSTD_compressBound().
+    ZSTD_compressContinue() doesn't guarantee recover after a failed compression.
+  - ZSTD_compressContinue() presumes prior input ***is still accessible and unmodified*** (up to maximum distance size, see WindowLog).
+    It remembers all previous contiguous blocks, plus one separated memory segment (which can itself consists of multiple contiguous blocks)
+  - ZSTD_compressContinue() detects that prior input has been overwritten when `src` buffer overlaps.
+    In which case, it will "discard" the relevant memory section from its history.
+
+  Finish a frame with ZSTD_compressEnd(), which will write the last block(s) and optional checksum.
+  It's possible to use srcSize==0, in which case, it will write a final empty block to end the frame.
+  Without last block mark, frames are considered unfinished (hence corrupted) by compliant decoders.
+
+  `ZSTD_CCtx` object can be re-used (ZSTD_compressBegin()) to compress again.
+*/
+
+/*=====   Buffer-less streaming compression functions  =====*/
+ZSTDLIB_API size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel);
+ZSTDLIB_API size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel);
+ZSTDLIB_API size_t ZSTD_compressBegin_advanced(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, ZSTD_parameters params, unsigned long long pledgedSrcSize); /*< pledgedSrcSize : If srcSize is not known at init time, use ZSTD_CONTENTSIZE_UNKNOWN */
+ZSTDLIB_API size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict); /*< note: fails if cdict==NULL */
+ZSTDLIB_API size_t ZSTD_compressBegin_usingCDict_advanced(ZSTD_CCtx* const cctx, const ZSTD_CDict* const cdict, ZSTD_frameParameters const fParams, unsigned long long const pledgedSrcSize);   /* compression parameters are already set within cdict. pledgedSrcSize must be correct. If srcSize is not known, use macro ZSTD_CONTENTSIZE_UNKNOWN */
+ZSTDLIB_API size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* preparedCCtx, unsigned long long pledgedSrcSize); /*<  note: if pledgedSrcSize is not known, use ZSTD_CONTENTSIZE_UNKNOWN */
+
+ZSTDLIB_API size_t ZSTD_compressContinue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTDLIB_API size_t ZSTD_compressEnd(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+
+
+/*
+  Buffer-less streaming decompression (synchronous mode)
+
+  A ZSTD_DCtx object is required to track streaming operations.
+  Use ZSTD_createDCtx() / ZSTD_freeDCtx() to manage it.
+  A ZSTD_DCtx object can be re-used multiple times.
+
+  First typical operation is to retrieve frame parameters, using ZSTD_getFrameHeader().
+  Frame header is extracted from the beginning of compressed frame, so providing only the frame's beginning is enough.
+  Data fragment must be large enough to ensure successful decoding.
+ `ZSTD_frameHeaderSize_max` bytes is guaranteed to always be large enough.
+  @result : 0 : successful decoding, the `ZSTD_frameHeader` structure is correctly filled.
+           >0 : `srcSize` is too small, please provide at least @result bytes on next attempt.
+           errorCode, which can be tested using ZSTD_isError().
+
+  It fills a ZSTD_frameHeader structure with important information to correctly decode the frame,
+  such as the dictionary ID, content size, or maximum back-reference distance (`windowSize`).
+  Note that these values could be wrong, either because of data corruption, or because a 3rd party deliberately spoofs false information.
+  As a consequence, check that values remain within valid application range.
+  For example, do not allocate memory blindly, check that `windowSize` is within expectation.
+  Each application can set its own limits, depending on local restrictions.
+  For extended interoperability, it is recommended to support `windowSize` of at least 8 MB.
+
+  ZSTD_decompressContinue() needs previous data blocks during decompression, up to `windowSize` bytes.
+  ZSTD_decompressContinue() is very sensitive to contiguity,
+  if 2 blocks don't follow each other, make sure that either the compressor breaks contiguity at the same place,
+  or that previous contiguous segment is large enough to properly handle maximum back-reference distance.
+  There are multiple ways to guarantee this condition.
+
+  The most memory efficient way is to use a round buffer of sufficient size.
+  Sufficient size is determined by invoking ZSTD_decodingBufferSize_min(),
+  which can @return an error code if required value is too large for current system (in 32-bits mode).
+  In a round buffer methodology, ZSTD_decompressContinue() decompresses each block next to previous one,
+  up to the moment there is not enough room left in the buffer to guarantee decoding another full block,
+  which maximum size is provided in `ZSTD_frameHeader` structure, field `blockSizeMax`.
+  At which point, decoding can resume from the beginning of the buffer.
+  Note that already decoded data stored in the buffer should be flushed before being overwritten.
+
+  There are alternatives possible, for example using two or more buffers of size `windowSize` each, though they consume more memory.
+
+  Finally, if you control the compression process, you can also ignore all buffer size rules,
+  as long as the encoder and decoder progress in "lock-step",
+  aka use exactly the same buffer sizes, break contiguity at the same place, etc.
+
+  Once buffers are setup, start decompression, with ZSTD_decompressBegin().
+  If decompression requires a dictionary, use ZSTD_decompressBegin_usingDict() or ZSTD_decompressBegin_usingDDict().
+
+  Then use ZSTD_nextSrcSizeToDecompress() and ZSTD_decompressContinue() alternatively.
+  ZSTD_nextSrcSizeToDecompress() tells how many bytes to provide as 'srcSize' to ZSTD_decompressContinue().
+  ZSTD_decompressContinue() requires this _exact_ amount of bytes, or it will fail.
+
+ @result of ZSTD_decompressContinue() is the number of bytes regenerated within 'dst' (necessarily <= dstCapacity).
+  It can be zero : it just means ZSTD_decompressContinue() has decoded some metadata item.
+  It can also be an error code, which can be tested with ZSTD_isError().
+
+  A frame is fully decoded when ZSTD_nextSrcSizeToDecompress() returns zero.
+  Context can then be reset to start a new decompression.
+
+  Note : it's possible to know if next input to present is a header or a block, using ZSTD_nextInputType().
+  This information is not required to properly decode a frame.
+
+  == Special case : skippable frames ==
+
+  Skippable frames allow integration of user-defined data into a flow of concatenated frames.
+  Skippable frames will be ignored (skipped) by decompressor.
+  The format of skippable frames is as follows :
+  a) Skippable frame ID - 4 Bytes, Little endian format, any value from 0x184D2A50 to 0x184D2A5F
+  b) Frame Size - 4 Bytes, Little endian format, unsigned 32-bits
+  c) Frame Content - any content (User Data) of length equal to Frame Size
+  For skippable frames ZSTD_getFrameHeader() returns zfhPtr->frameType==ZSTD_skippableFrame.
+  For skippable frames ZSTD_decompressContinue() always returns 0 : it only skips the content.
+*/
+
+/*=====   Buffer-less streaming decompression functions  =====*/
+typedef enum { ZSTD_frame, ZSTD_skippableFrame } ZSTD_frameType_e;
+typedef struct {
+    unsigned long long frameContentSize; /* if == ZSTD_CONTENTSIZE_UNKNOWN, it means this field is not available. 0 means "empty" */
+    unsigned long long windowSize;       /* can be very large, up to <= frameContentSize */
+    unsigned blockSizeMax;
+    ZSTD_frameType_e frameType;          /* if == ZSTD_skippableFrame, frameContentSize is the size of skippable content */
+    unsigned headerSize;
+    unsigned dictID;
+    unsigned checksumFlag;
+} ZSTD_frameHeader;
+
+/*! ZSTD_getFrameHeader() :
+ *  decode Frame Header, or requires larger `srcSize`.
+ * @return : 0, `zfhPtr` is correctly filled,
+ *          >0, `srcSize` is too small, value is wanted `srcSize` amount,
+ *           or an error code, which can be tested using ZSTD_isError() */
+ZSTDLIB_API size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize);   /*< doesn't consume input */
+/*! ZSTD_getFrameHeader_advanced() :
+ *  same as ZSTD_getFrameHeader(),
+ *  with added capability to select a format (like ZSTD_f_zstd1_magicless) */
+ZSTDLIB_API size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format);
+ZSTDLIB_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize);  /*< when frame content size is not known, pass in frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN */
+
+ZSTDLIB_API size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx);
+ZSTDLIB_API size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);
+ZSTDLIB_API size_t ZSTD_decompressBegin_usingDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);
+
+ZSTDLIB_API size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx);
+ZSTDLIB_API size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+
+/* misc */
+ZSTDLIB_API void   ZSTD_copyDCtx(ZSTD_DCtx* dctx, const ZSTD_DCtx* preparedDCtx);
+typedef enum { ZSTDnit_frameHeader, ZSTDnit_blockHeader, ZSTDnit_block, ZSTDnit_lastBlock, ZSTDnit_checksum, ZSTDnit_skippableFrame } ZSTD_nextInputType_e;
+ZSTDLIB_API ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx);
+
+
+
+
+/* ============================ */
+/*       Block level API       */
+/* ============================ */
+
+/*!
+    Block functions produce and decode raw zstd blocks, without frame metadata.
+    Frame metadata cost is typically ~12 bytes, which can be non-negligible for very small blocks (< 100 bytes).
+    But users will have to take in charge needed metadata to regenerate data, such as compressed and content sizes.
+
+    A few rules to respect :
+    - Compressing and decompressing require a context structure
+      + Use ZSTD_createCCtx() and ZSTD_createDCtx()
+    - It is necessary to init context before starting
+      + compression : any ZSTD_compressBegin*() variant, including with dictionary
+      + decompression : any ZSTD_decompressBegin*() variant, including with dictionary
+      + copyCCtx() and copyDCtx() can be used too
+    - Block size is limited, it must be <= ZSTD_getBlockSize() <= ZSTD_BLOCKSIZE_MAX == 128 KB
+      + If input is larger than a block size, it's necessary to split input data into multiple blocks
+      + For inputs larger than a single block, consider using regular ZSTD_compress() instead.
+        Frame metadata is not that costly, and quickly becomes negligible as source size grows larger than a block.
+    - When a block is considered not compressible enough, ZSTD_compressBlock() result will be 0 (zero) !
+      ===> In which case, nothing is produced into `dst` !
+      + User __must__ test for such outcome and deal directly with uncompressed data
+      + A block cannot be declared incompressible if ZSTD_compressBlock() return value was != 0.
+        Doing so would mess up with statistics history, leading to potential data corruption.
+      + ZSTD_decompressBlock() _doesn't accept uncompressed data as input_ !!
+      + In case of multiple successive blocks, should some of them be uncompressed,
+        decoder must be informed of their existence in order to follow proper history.
+        Use ZSTD_insertBlock() for such a case.
+*/
+
+/*=====   Raw zstd block functions  =====*/
+ZSTDLIB_API size_t ZSTD_getBlockSize   (const ZSTD_CCtx* cctx);
+ZSTDLIB_API size_t ZSTD_compressBlock  (ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTDLIB_API size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTDLIB_API size_t ZSTD_insertBlock    (ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize);  /*< insert uncompressed block into `dctx` history. Useful for multi-blocks decompression. */
+
+
+#endif   /* ZSTD_H_ZSTD_STATIC_LINKING_ONLY */
+
diff --git a/lib/zstd/Makefile b/lib/zstd/Makefile
index 099c26c378f8..528d3a870bb9 100644
--- a/lib/zstd/Makefile
+++ b/lib/zstd/Makefile
@@ -10,11 +10,14 @@
 # ################################################################
 obj-$(CONFIG_ZSTD_DECOMPRESS) += zstd_decompress.o
 
-ccflags-y += -O3
-
 zstd_decompress-y := \
-		decompress.o \
+		zstd_decompress_module.o \
+		common/debug.o \
 		common/entropy_common.o \
+		common/error_private.o \
 		common/fse_decompress.o \
 		common/zstd_common.o \
-		decompress/huf_decompress.o
+		decompress/huf_decompress.o \
+		decompress/zstd_ddict.o \
+		decompress/zstd_decompress.o \
+		decompress/zstd_decompress_block.o \
diff --git a/lib/zstd/bitstream.h b/lib/zstd/bitstream.h
deleted file mode 100644
index a826b99e1d63..000000000000
--- a/lib/zstd/bitstream.h
+++ /dev/null
@@ -1,374 +0,0 @@
-/*
- * bitstream
- * Part of FSE library
- * header file (to include)
- * Copyright (C) 2013-2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-#ifndef BITSTREAM_H_MODULE
-#define BITSTREAM_H_MODULE
-
-/*
-*  This API consists of small unitary functions, which must be inlined for best performance.
-*  Since link-time-optimization is not available for all compilers,
-*  these functions are defined into a .h to be included.
-*/
-
-/*-****************************************
-*  Dependencies
-******************************************/
-#include "error_private.h" /* error codes and messages */
-#include "mem.h"	   /* unaligned access routines */
-
-/*=========================================
-*  Target specific
-=========================================*/
-#define STREAM_ACCUMULATOR_MIN_32 25
-#define STREAM_ACCUMULATOR_MIN_64 57
-#define STREAM_ACCUMULATOR_MIN ((U32)(ZSTD_32bits() ? STREAM_ACCUMULATOR_MIN_32 : STREAM_ACCUMULATOR_MIN_64))
-
-/*-******************************************
-*  bitStream encoding API (write forward)
-********************************************/
-/* bitStream can mix input from multiple sources.
-*  A critical property of these streams is that they encode and decode in **reverse** direction.
-*  So the first bit sequence you add will be the last to be read, like a LIFO stack.
-*/
-typedef struct {
-	size_t bitContainer;
-	int bitPos;
-	char *startPtr;
-	char *ptr;
-	char *endPtr;
-} BIT_CStream_t;
-
-ZSTD_STATIC size_t BIT_initCStream(BIT_CStream_t *bitC, void *dstBuffer, size_t dstCapacity);
-ZSTD_STATIC void BIT_addBits(BIT_CStream_t *bitC, size_t value, unsigned nbBits);
-ZSTD_STATIC void BIT_flushBits(BIT_CStream_t *bitC);
-ZSTD_STATIC size_t BIT_closeCStream(BIT_CStream_t *bitC);
-
-/* Start with initCStream, providing the size of buffer to write into.
-*  bitStream will never write outside of this buffer.
-*  `dstCapacity` must be >= sizeof(bitD->bitContainer), otherwise @return will be an error code.
-*
-*  bits are first added to a local register.
-*  Local register is size_t, hence 64-bits on 64-bits systems, or 32-bits on 32-bits systems.
-*  Writing data into memory is an explicit operation, performed by the flushBits function.
-*  Hence keep track how many bits are potentially stored into local register to avoid register overflow.
-*  After a flushBits, a maximum of 7 bits might still be stored into local register.
-*
-*  Avoid storing elements of more than 24 bits if you want compatibility with 32-bits bitstream readers.
-*
-*  Last operation is to close the bitStream.
-*  The function returns the final size of CStream in bytes.
-*  If data couldn't fit into `dstBuffer`, it will return a 0 ( == not storable)
-*/
-
-/*-********************************************
-*  bitStream decoding API (read backward)
-**********************************************/
-typedef struct {
-	size_t bitContainer;
-	unsigned bitsConsumed;
-	const char *ptr;
-	const char *start;
-} BIT_DStream_t;
-
-typedef enum {
-	BIT_DStream_unfinished = 0,
-	BIT_DStream_endOfBuffer = 1,
-	BIT_DStream_completed = 2,
-	BIT_DStream_overflow = 3
-} BIT_DStream_status; /* result of BIT_reloadDStream() */
-/* 1,2,4,8 would be better for bitmap combinations, but slows down performance a bit ... :( */
-
-ZSTD_STATIC size_t BIT_initDStream(BIT_DStream_t *bitD, const void *srcBuffer, size_t srcSize);
-ZSTD_STATIC size_t BIT_readBits(BIT_DStream_t *bitD, unsigned nbBits);
-ZSTD_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t *bitD);
-ZSTD_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t *bitD);
-
-/* Start by invoking BIT_initDStream().
-*  A chunk of the bitStream is then stored into a local register.
-*  Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
-*  You can then retrieve bitFields stored into the local register, **in reverse order**.
-*  Local register is explicitly reloaded from memory by the BIT_reloadDStream() method.
-*  A reload guarantee a minimum of ((8*sizeof(bitD->bitContainer))-7) bits when its result is BIT_DStream_unfinished.
-*  Otherwise, it can be less than that, so proceed accordingly.
-*  Checking if DStream has reached its end can be performed with BIT_endOfDStream().
-*/
-
-/*-****************************************
-*  unsafe API
-******************************************/
-ZSTD_STATIC void BIT_addBitsFast(BIT_CStream_t *bitC, size_t value, unsigned nbBits);
-/* faster, but works only if value is "clean", meaning all high bits above nbBits are 0 */
-
-ZSTD_STATIC void BIT_flushBitsFast(BIT_CStream_t *bitC);
-/* unsafe version; does not check buffer overflow */
-
-ZSTD_STATIC size_t BIT_readBitsFast(BIT_DStream_t *bitD, unsigned nbBits);
-/* faster, but works only if nbBits >= 1 */
-
-/*-**************************************************************
-*  Internal functions
-****************************************************************/
-ZSTD_STATIC unsigned BIT_highbit32(register U32 val) { return 31 - __builtin_clz(val); }
-
-/*=====    Local Constants   =====*/
-static const unsigned BIT_mask[] = {0,       1,       3,       7,	0xF,      0x1F,     0x3F,     0x7F,      0xFF,
-				    0x1FF,   0x3FF,   0x7FF,   0xFFF,    0x1FFF,   0x3FFF,   0x7FFF,   0xFFFF,    0x1FFFF,
-				    0x3FFFF, 0x7FFFF, 0xFFFFF, 0x1FFFFF, 0x3FFFFF, 0x7FFFFF, 0xFFFFFF, 0x1FFFFFF, 0x3FFFFFF}; /* up to 26 bits */
-
-/*-**************************************************************
-*  bitStream encoding
-****************************************************************/
-/*! BIT_initCStream() :
- *  `dstCapacity` must be > sizeof(void*)
- *  @return : 0 if success,
-			  otherwise an error code (can be tested using ERR_isError() ) */
-ZSTD_STATIC size_t BIT_initCStream(BIT_CStream_t *bitC, void *startPtr, size_t dstCapacity)
-{
-	bitC->bitContainer = 0;
-	bitC->bitPos = 0;
-	bitC->startPtr = (char *)startPtr;
-	bitC->ptr = bitC->startPtr;
-	bitC->endPtr = bitC->startPtr + dstCapacity - sizeof(bitC->ptr);
-	if (dstCapacity <= sizeof(bitC->ptr))
-		return ERROR(dstSize_tooSmall);
-	return 0;
-}
-
-/*! BIT_addBits() :
-	can add up to 26 bits into `bitC`.
-	Does not check for register overflow ! */
-ZSTD_STATIC void BIT_addBits(BIT_CStream_t *bitC, size_t value, unsigned nbBits)
-{
-	bitC->bitContainer |= (value & BIT_mask[nbBits]) << bitC->bitPos;
-	bitC->bitPos += nbBits;
-}
-
-/*! BIT_addBitsFast() :
- *  works only if `value` is _clean_, meaning all high bits above nbBits are 0 */
-ZSTD_STATIC void BIT_addBitsFast(BIT_CStream_t *bitC, size_t value, unsigned nbBits)
-{
-	bitC->bitContainer |= value << bitC->bitPos;
-	bitC->bitPos += nbBits;
-}
-
-/*! BIT_flushBitsFast() :
- *  unsafe version; does not check buffer overflow */
-ZSTD_STATIC void BIT_flushBitsFast(BIT_CStream_t *bitC)
-{
-	size_t const nbBytes = bitC->bitPos >> 3;
-	ZSTD_writeLEST(bitC->ptr, bitC->bitContainer);
-	bitC->ptr += nbBytes;
-	bitC->bitPos &= 7;
-	bitC->bitContainer >>= nbBytes * 8; /* if bitPos >= sizeof(bitContainer)*8 --> undefined behavior */
-}
-
-/*! BIT_flushBits() :
- *  safe version; check for buffer overflow, and prevents it.
- *  note : does not signal buffer overflow. This will be revealed later on using BIT_closeCStream() */
-ZSTD_STATIC void BIT_flushBits(BIT_CStream_t *bitC)
-{
-	size_t const nbBytes = bitC->bitPos >> 3;
-	ZSTD_writeLEST(bitC->ptr, bitC->bitContainer);
-	bitC->ptr += nbBytes;
-	if (bitC->ptr > bitC->endPtr)
-		bitC->ptr = bitC->endPtr;
-	bitC->bitPos &= 7;
-	bitC->bitContainer >>= nbBytes * 8; /* if bitPos >= sizeof(bitContainer)*8 --> undefined behavior */
-}
-
-/*! BIT_closeCStream() :
- *  @return : size of CStream, in bytes,
-			  or 0 if it could not fit into dstBuffer */
-ZSTD_STATIC size_t BIT_closeCStream(BIT_CStream_t *bitC)
-{
-	BIT_addBitsFast(bitC, 1, 1); /* endMark */
-	BIT_flushBits(bitC);
-
-	if (bitC->ptr >= bitC->endPtr)
-		return 0; /* doesn't fit within authorized budget : cancel */
-
-	return (bitC->ptr - bitC->startPtr) + (bitC->bitPos > 0);
-}
-
-/*-********************************************************
-* bitStream decoding
-**********************************************************/
-/*! BIT_initDStream() :
-*   Initialize a BIT_DStream_t.
-*   `bitD` : a pointer to an already allocated BIT_DStream_t structure.
-*   `srcSize` must be the *exact* size of the bitStream, in bytes.
-*   @return : size of stream (== srcSize) or an errorCode if a problem is detected
-*/
-ZSTD_STATIC size_t BIT_initDStream(BIT_DStream_t *bitD, const void *srcBuffer, size_t srcSize)
-{
-	if (srcSize < 1) {
-		memset(bitD, 0, sizeof(*bitD));
-		return ERROR(srcSize_wrong);
-	}
-
-	if (srcSize >= sizeof(bitD->bitContainer)) { /* normal case */
-		bitD->start = (const char *)srcBuffer;
-		bitD->ptr = (const char *)srcBuffer + srcSize - sizeof(bitD->bitContainer);
-		bitD->bitContainer = ZSTD_readLEST(bitD->ptr);
-		{
-			BYTE const lastByte = ((const BYTE *)srcBuffer)[srcSize - 1];
-			bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0; /* ensures bitsConsumed is always set */
-			if (lastByte == 0)
-				return ERROR(GENERIC); /* endMark not present */
-		}
-	} else {
-		bitD->start = (const char *)srcBuffer;
-		bitD->ptr = bitD->start;
-		bitD->bitContainer = *(const BYTE *)(bitD->start);
-		switch (srcSize) {
-		case 7: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[6]) << (sizeof(bitD->bitContainer) * 8 - 16);
-		case 6: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[5]) << (sizeof(bitD->bitContainer) * 8 - 24);
-		case 5: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[4]) << (sizeof(bitD->bitContainer) * 8 - 32);
-		case 4: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[3]) << 24;
-		case 3: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[2]) << 16;
-		case 2: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[1]) << 8;
-		default:;
-		}
-		{
-			BYTE const lastByte = ((const BYTE *)srcBuffer)[srcSize - 1];
-			bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;
-			if (lastByte == 0)
-				return ERROR(GENERIC); /* endMark not present */
-		}
-		bitD->bitsConsumed += (U32)(sizeof(bitD->bitContainer) - srcSize) * 8;
-	}
-
-	return srcSize;
-}
-
-ZSTD_STATIC size_t BIT_getUpperBits(size_t bitContainer, U32 const start) { return bitContainer >> start; }
-
-ZSTD_STATIC size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 const nbBits) { return (bitContainer >> start) & BIT_mask[nbBits]; }
-
-ZSTD_STATIC size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits) { return bitContainer & BIT_mask[nbBits]; }
-
-/*! BIT_lookBits() :
- *  Provides next n bits from local register.
- *  local register is not modified.
- *  On 32-bits, maxNbBits==24.
- *  On 64-bits, maxNbBits==56.
- *  @return : value extracted
- */
-ZSTD_STATIC size_t BIT_lookBits(const BIT_DStream_t *bitD, U32 nbBits)
-{
-	U32 const bitMask = sizeof(bitD->bitContainer) * 8 - 1;
-	return ((bitD->bitContainer << (bitD->bitsConsumed & bitMask)) >> 1) >> ((bitMask - nbBits) & bitMask);
-}
-
-/*! BIT_lookBitsFast() :
-*   unsafe version; only works only if nbBits >= 1 */
-ZSTD_STATIC size_t BIT_lookBitsFast(const BIT_DStream_t *bitD, U32 nbBits)
-{
-	U32 const bitMask = sizeof(bitD->bitContainer) * 8 - 1;
-	return (bitD->bitContainer << (bitD->bitsConsumed & bitMask)) >> (((bitMask + 1) - nbBits) & bitMask);
-}
-
-ZSTD_STATIC void BIT_skipBits(BIT_DStream_t *bitD, U32 nbBits) { bitD->bitsConsumed += nbBits; }
-
-/*! BIT_readBits() :
- *  Read (consume) next n bits from local register and update.
- *  Pay attention to not read more than nbBits contained into local register.
- *  @return : extracted value.
- */
-ZSTD_STATIC size_t BIT_readBits(BIT_DStream_t *bitD, U32 nbBits)
-{
-	size_t const value = BIT_lookBits(bitD, nbBits);
-	BIT_skipBits(bitD, nbBits);
-	return value;
-}
-
-/*! BIT_readBitsFast() :
-*   unsafe version; only works only if nbBits >= 1 */
-ZSTD_STATIC size_t BIT_readBitsFast(BIT_DStream_t *bitD, U32 nbBits)
-{
-	size_t const value = BIT_lookBitsFast(bitD, nbBits);
-	BIT_skipBits(bitD, nbBits);
-	return value;
-}
-
-/*! BIT_reloadDStream() :
-*   Refill `bitD` from buffer previously set in BIT_initDStream() .
-*   This function is safe, it guarantees it will not read beyond src buffer.
-*   @return : status of `BIT_DStream_t` internal register.
-			  if status == BIT_DStream_unfinished, internal register is filled with >= (sizeof(bitD->bitContainer)*8 - 7) bits */
-ZSTD_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t *bitD)
-{
-	if (bitD->bitsConsumed > (sizeof(bitD->bitContainer) * 8)) /* should not happen => corruption detected */
-		return BIT_DStream_overflow;
-
-	if (bitD->ptr >= bitD->start + sizeof(bitD->bitContainer)) {
-		bitD->ptr -= bitD->bitsConsumed >> 3;
-		bitD->bitsConsumed &= 7;
-		bitD->bitContainer = ZSTD_readLEST(bitD->ptr);
-		return BIT_DStream_unfinished;
-	}
-	if (bitD->ptr == bitD->start) {
-		if (bitD->bitsConsumed < sizeof(bitD->bitContainer) * 8)
-			return BIT_DStream_endOfBuffer;
-		return BIT_DStream_completed;
-	}
-	{
-		U32 nbBytes = bitD->bitsConsumed >> 3;
-		BIT_DStream_status result = BIT_DStream_unfinished;
-		if (bitD->ptr - nbBytes < bitD->start) {
-			nbBytes = (U32)(bitD->ptr - bitD->start); /* ptr > start */
-			result = BIT_DStream_endOfBuffer;
-		}
-		bitD->ptr -= nbBytes;
-		bitD->bitsConsumed -= nbBytes * 8;
-		bitD->bitContainer = ZSTD_readLEST(bitD->ptr); /* reminder : srcSize > sizeof(bitD) */
-		return result;
-	}
-}
-
-/*! BIT_endOfDStream() :
-*   @return Tells if DStream has exactly reached its end (all bits consumed).
-*/
-ZSTD_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t *DStream)
-{
-	return ((DStream->ptr == DStream->start) && (DStream->bitsConsumed == sizeof(DStream->bitContainer) * 8));
-}
-
-#endif /* BITSTREAM_H_MODULE */
diff --git a/lib/zstd/common/bitstream.h b/lib/zstd/common/bitstream.h
new file mode 100644
index 000000000000..28248abe8612
--- /dev/null
+++ b/lib/zstd/common/bitstream.h
@@ -0,0 +1,437 @@
+/* ******************************************************************
+ * bitstream
+ * Part of FSE library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+#ifndef BITSTREAM_H_MODULE
+#define BITSTREAM_H_MODULE
+
+/*
+*  This API consists of small unitary functions, which must be inlined for best performance.
+*  Since link-time-optimization is not available for all compilers,
+*  these functions are defined into a .h to be included.
+*/
+
+/*-****************************************
+*  Dependencies
+******************************************/
+#include "mem.h"            /* unaligned access routines */
+#include "compiler.h"       /* UNLIKELY() */
+#include "debug.h"          /* assert(), DEBUGLOG(), RAWLOG() */
+#include "error_private.h"  /* error codes and messages */
+
+
+/*=========================================
+*  Target specific
+=========================================*/
+
+#define STREAM_ACCUMULATOR_MIN_32  25
+#define STREAM_ACCUMULATOR_MIN_64  57
+#define STREAM_ACCUMULATOR_MIN    ((U32)(MEM_32bits() ? STREAM_ACCUMULATOR_MIN_32 : STREAM_ACCUMULATOR_MIN_64))
+
+
+/*-******************************************
+*  bitStream encoding API (write forward)
+********************************************/
+/* bitStream can mix input from multiple sources.
+ * A critical property of these streams is that they encode and decode in **reverse** direction.
+ * So the first bit sequence you add will be the last to be read, like a LIFO stack.
+ */
+typedef struct {
+    size_t bitContainer;
+    unsigned bitPos;
+    char*  startPtr;
+    char*  ptr;
+    char*  endPtr;
+} BIT_CStream_t;
+
+MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC, void* dstBuffer, size_t dstCapacity);
+MEM_STATIC void   BIT_addBits(BIT_CStream_t* bitC, size_t value, unsigned nbBits);
+MEM_STATIC void   BIT_flushBits(BIT_CStream_t* bitC);
+MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC);
+
+/* Start with initCStream, providing the size of buffer to write into.
+*  bitStream will never write outside of this buffer.
+*  `dstCapacity` must be >= sizeof(bitD->bitContainer), otherwise @return will be an error code.
+*
+*  bits are first added to a local register.
+*  Local register is size_t, hence 64-bits on 64-bits systems, or 32-bits on 32-bits systems.
+*  Writing data into memory is an explicit operation, performed by the flushBits function.
+*  Hence keep track how many bits are potentially stored into local register to avoid register overflow.
+*  After a flushBits, a maximum of 7 bits might still be stored into local register.
+*
+*  Avoid storing elements of more than 24 bits if you want compatibility with 32-bits bitstream readers.
+*
+*  Last operation is to close the bitStream.
+*  The function returns the final size of CStream in bytes.
+*  If data couldn't fit into `dstBuffer`, it will return a 0 ( == not storable)
+*/
+
+
+/*-********************************************
+*  bitStream decoding API (read backward)
+**********************************************/
+typedef struct {
+    size_t   bitContainer;
+    unsigned bitsConsumed;
+    const char* ptr;
+    const char* start;
+    const char* limitPtr;
+} BIT_DStream_t;
+
+typedef enum { BIT_DStream_unfinished = 0,
+               BIT_DStream_endOfBuffer = 1,
+               BIT_DStream_completed = 2,
+               BIT_DStream_overflow = 3 } BIT_DStream_status;  /* result of BIT_reloadDStream() */
+               /* 1,2,4,8 would be better for bitmap combinations, but slows down performance a bit ... :( */
+
+MEM_STATIC size_t   BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize);
+MEM_STATIC size_t   BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits);
+MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD);
+MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* bitD);
+
+
+/* Start by invoking BIT_initDStream().
+*  A chunk of the bitStream is then stored into a local register.
+*  Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
+*  You can then retrieve bitFields stored into the local register, **in reverse order**.
+*  Local register is explicitly reloaded from memory by the BIT_reloadDStream() method.
+*  A reload guarantee a minimum of ((8*sizeof(bitD->bitContainer))-7) bits when its result is BIT_DStream_unfinished.
+*  Otherwise, it can be less than that, so proceed accordingly.
+*  Checking if DStream has reached its end can be performed with BIT_endOfDStream().
+*/
+
+
+/*-****************************************
+*  unsafe API
+******************************************/
+MEM_STATIC void BIT_addBitsFast(BIT_CStream_t* bitC, size_t value, unsigned nbBits);
+/* faster, but works only if value is "clean", meaning all high bits above nbBits are 0 */
+
+MEM_STATIC void BIT_flushBitsFast(BIT_CStream_t* bitC);
+/* unsafe version; does not check buffer overflow */
+
+MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits);
+/* faster, but works only if nbBits >= 1 */
+
+
+
+/*-**************************************************************
+*  Internal functions
+****************************************************************/
+MEM_STATIC unsigned BIT_highbit32 (U32 val)
+{
+    assert(val != 0);
+    {
+#   if (__GNUC__ >= 3)   /* Use GCC Intrinsic */
+        return __builtin_clz (val) ^ 31;
+#   else   /* Software version */
+        static const unsigned DeBruijnClz[32] = { 0,  9,  1, 10, 13, 21,  2, 29,
+                                                 11, 14, 16, 18, 22, 25,  3, 30,
+                                                  8, 12, 20, 28, 15, 17, 24,  7,
+                                                 19, 27, 23,  6, 26,  5,  4, 31 };
+        U32 v = val;
+        v |= v >> 1;
+        v |= v >> 2;
+        v |= v >> 4;
+        v |= v >> 8;
+        v |= v >> 16;
+        return DeBruijnClz[ (U32) (v * 0x07C4ACDDU) >> 27];
+#   endif
+    }
+}
+
+/*=====    Local Constants   =====*/
+static const unsigned BIT_mask[] = {
+    0,          1,         3,         7,         0xF,       0x1F,
+    0x3F,       0x7F,      0xFF,      0x1FF,     0x3FF,     0x7FF,
+    0xFFF,      0x1FFF,    0x3FFF,    0x7FFF,    0xFFFF,    0x1FFFF,
+    0x3FFFF,    0x7FFFF,   0xFFFFF,   0x1FFFFF,  0x3FFFFF,  0x7FFFFF,
+    0xFFFFFF,   0x1FFFFFF, 0x3FFFFFF, 0x7FFFFFF, 0xFFFFFFF, 0x1FFFFFFF,
+    0x3FFFFFFF, 0x7FFFFFFF}; /* up to 31 bits */
+#define BIT_MASK_SIZE (sizeof(BIT_mask) / sizeof(BIT_mask[0]))
+
+/*-**************************************************************
+*  bitStream encoding
+****************************************************************/
+/*! BIT_initCStream() :
+ *  `dstCapacity` must be > sizeof(size_t)
+ *  @return : 0 if success,
+ *            otherwise an error code (can be tested using ERR_isError()) */
+MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC,
+                                  void* startPtr, size_t dstCapacity)
+{
+    bitC->bitContainer = 0;
+    bitC->bitPos = 0;
+    bitC->startPtr = (char*)startPtr;
+    bitC->ptr = bitC->startPtr;
+    bitC->endPtr = bitC->startPtr + dstCapacity - sizeof(bitC->bitContainer);
+    if (dstCapacity <= sizeof(bitC->bitContainer)) return ERROR(dstSize_tooSmall);
+    return 0;
+}
+
+/*! BIT_addBits() :
+ *  can add up to 31 bits into `bitC`.
+ *  Note : does not check for register overflow ! */
+MEM_STATIC void BIT_addBits(BIT_CStream_t* bitC,
+                            size_t value, unsigned nbBits)
+{
+    DEBUG_STATIC_ASSERT(BIT_MASK_SIZE == 32);
+    assert(nbBits < BIT_MASK_SIZE);
+    assert(nbBits + bitC->bitPos < sizeof(bitC->bitContainer) * 8);
+    bitC->bitContainer |= (value & BIT_mask[nbBits]) << bitC->bitPos;
+    bitC->bitPos += nbBits;
+}
+
+/*! BIT_addBitsFast() :
+ *  works only if `value` is _clean_,
+ *  meaning all high bits above nbBits are 0 */
+MEM_STATIC void BIT_addBitsFast(BIT_CStream_t* bitC,
+                                size_t value, unsigned nbBits)
+{
+    assert((value>>nbBits) == 0);
+    assert(nbBits + bitC->bitPos < sizeof(bitC->bitContainer) * 8);
+    bitC->bitContainer |= value << bitC->bitPos;
+    bitC->bitPos += nbBits;
+}
+
+/*! BIT_flushBitsFast() :
+ *  assumption : bitContainer has not overflowed
+ *  unsafe version; does not check buffer overflow */
+MEM_STATIC void BIT_flushBitsFast(BIT_CStream_t* bitC)
+{
+    size_t const nbBytes = bitC->bitPos >> 3;
+    assert(bitC->bitPos < sizeof(bitC->bitContainer) * 8);
+    assert(bitC->ptr <= bitC->endPtr);
+    MEM_writeLEST(bitC->ptr, bitC->bitContainer);
+    bitC->ptr += nbBytes;
+    bitC->bitPos &= 7;
+    bitC->bitContainer >>= nbBytes*8;
+}
+
+/*! BIT_flushBits() :
+ *  assumption : bitContainer has not overflowed
+ *  safe version; check for buffer overflow, and prevents it.
+ *  note : does not signal buffer overflow.
+ *  overflow will be revealed later on using BIT_closeCStream() */
+MEM_STATIC void BIT_flushBits(BIT_CStream_t* bitC)
+{
+    size_t const nbBytes = bitC->bitPos >> 3;
+    assert(bitC->bitPos < sizeof(bitC->bitContainer) * 8);
+    assert(bitC->ptr <= bitC->endPtr);
+    MEM_writeLEST(bitC->ptr, bitC->bitContainer);
+    bitC->ptr += nbBytes;
+    if (bitC->ptr > bitC->endPtr) bitC->ptr = bitC->endPtr;
+    bitC->bitPos &= 7;
+    bitC->bitContainer >>= nbBytes*8;
+}
+
+/*! BIT_closeCStream() :
+ *  @return : size of CStream, in bytes,
+ *            or 0 if it could not fit into dstBuffer */
+MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC)
+{
+    BIT_addBitsFast(bitC, 1, 1);   /* endMark */
+    BIT_flushBits(bitC);
+    if (bitC->ptr >= bitC->endPtr) return 0; /* overflow detected */
+    return (bitC->ptr - bitC->startPtr) + (bitC->bitPos > 0);
+}
+
+
+/*-********************************************************
+*  bitStream decoding
+**********************************************************/
+/*! BIT_initDStream() :
+ *  Initialize a BIT_DStream_t.
+ * `bitD` : a pointer to an already allocated BIT_DStream_t structure.
+ * `srcSize` must be the *exact* size of the bitStream, in bytes.
+ * @return : size of stream (== srcSize), or an errorCode if a problem is detected
+ */
+MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize)
+{
+    if (srcSize < 1) { ZSTD_memset(bitD, 0, sizeof(*bitD)); return ERROR(srcSize_wrong); }
+
+    bitD->start = (const char*)srcBuffer;
+    bitD->limitPtr = bitD->start + sizeof(bitD->bitContainer);
+
+    if (srcSize >=  sizeof(bitD->bitContainer)) {  /* normal case */
+        bitD->ptr   = (const char*)srcBuffer + srcSize - sizeof(bitD->bitContainer);
+        bitD->bitContainer = MEM_readLEST(bitD->ptr);
+        { BYTE const lastByte = ((const BYTE*)srcBuffer)[srcSize-1];
+          bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;  /* ensures bitsConsumed is always set */
+          if (lastByte == 0) return ERROR(GENERIC); /* endMark not present */ }
+    } else {
+        bitD->ptr   = bitD->start;
+        bitD->bitContainer = *(const BYTE*)(bitD->start);
+        switch(srcSize)
+        {
+        case 7: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[6]) << (sizeof(bitD->bitContainer)*8 - 16);
+                ZSTD_FALLTHROUGH;
+
+        case 6: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[5]) << (sizeof(bitD->bitContainer)*8 - 24);
+                ZSTD_FALLTHROUGH;
+
+        case 5: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[4]) << (sizeof(bitD->bitContainer)*8 - 32);
+                ZSTD_FALLTHROUGH;
+
+        case 4: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[3]) << 24;
+                ZSTD_FALLTHROUGH;
+
+        case 3: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[2]) << 16;
+                ZSTD_FALLTHROUGH;
+
+        case 2: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[1]) <<  8;
+                ZSTD_FALLTHROUGH;
+
+        default: break;
+        }
+        {   BYTE const lastByte = ((const BYTE*)srcBuffer)[srcSize-1];
+            bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;
+            if (lastByte == 0) return ERROR(corruption_detected);  /* endMark not present */
+        }
+        bitD->bitsConsumed += (U32)(sizeof(bitD->bitContainer) - srcSize)*8;
+    }
+
+    return srcSize;
+}
+
+MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getUpperBits(size_t bitContainer, U32 const start)
+{
+    return bitContainer >> start;
+}
+
+MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 const nbBits)
+{
+    U32 const regMask = sizeof(bitContainer)*8 - 1;
+    /* if start > regMask, bitstream is corrupted, and result is undefined */
+    assert(nbBits < BIT_MASK_SIZE);
+    return (bitContainer >> (start & regMask)) & BIT_mask[nbBits];
+}
+
+MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
+{
+    assert(nbBits < BIT_MASK_SIZE);
+    return bitContainer & BIT_mask[nbBits];
+}
+
+/*! BIT_lookBits() :
+ *  Provides next n bits from local register.
+ *  local register is not modified.
+ *  On 32-bits, maxNbBits==24.
+ *  On 64-bits, maxNbBits==56.
+ * @return : value extracted */
+MEM_STATIC  FORCE_INLINE_ATTR size_t BIT_lookBits(const BIT_DStream_t*  bitD, U32 nbBits)
+{
+    /* arbitrate between double-shift and shift+mask */
+#if 1
+    /* if bitD->bitsConsumed + nbBits > sizeof(bitD->bitContainer)*8,
+     * bitstream is likely corrupted, and result is undefined */
+    return BIT_getMiddleBits(bitD->bitContainer, (sizeof(bitD->bitContainer)*8) - bitD->bitsConsumed - nbBits, nbBits);
+#else
+    /* this code path is slower on my os-x laptop */
+    U32 const regMask = sizeof(bitD->bitContainer)*8 - 1;
+    return ((bitD->bitContainer << (bitD->bitsConsumed & regMask)) >> 1) >> ((regMask-nbBits) & regMask);
+#endif
+}
+
+/*! BIT_lookBitsFast() :
+ *  unsafe version; only works if nbBits >= 1 */
+MEM_STATIC size_t BIT_lookBitsFast(const BIT_DStream_t* bitD, U32 nbBits)
+{
+    U32 const regMask = sizeof(bitD->bitContainer)*8 - 1;
+    assert(nbBits >= 1);
+    return (bitD->bitContainer << (bitD->bitsConsumed & regMask)) >> (((regMask+1)-nbBits) & regMask);
+}
+
+MEM_STATIC FORCE_INLINE_ATTR void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
+{
+    bitD->bitsConsumed += nbBits;
+}
+
+/*! BIT_readBits() :
+ *  Read (consume) next n bits from local register and update.
+ *  Pay attention to not read more than nbBits contained into local register.
+ * @return : extracted value. */
+MEM_STATIC FORCE_INLINE_ATTR size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits)
+{
+    size_t const value = BIT_lookBits(bitD, nbBits);
+    BIT_skipBits(bitD, nbBits);
+    return value;
+}
+
+/*! BIT_readBitsFast() :
+ *  unsafe version; only works only if nbBits >= 1 */
+MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits)
+{
+    size_t const value = BIT_lookBitsFast(bitD, nbBits);
+    assert(nbBits >= 1);
+    BIT_skipBits(bitD, nbBits);
+    return value;
+}
+
+/*! BIT_reloadDStreamFast() :
+ *  Similar to BIT_reloadDStream(), but with two differences:
+ *  1. bitsConsumed <= sizeof(bitD->bitContainer)*8 must hold!
+ *  2. Returns BIT_DStream_overflow when bitD->ptr < bitD->limitPtr, at this
+ *     point you must use BIT_reloadDStream() to reload.
+ */
+MEM_STATIC BIT_DStream_status BIT_reloadDStreamFast(BIT_DStream_t* bitD)
+{
+    if (UNLIKELY(bitD->ptr < bitD->limitPtr))
+        return BIT_DStream_overflow;
+    assert(bitD->bitsConsumed <= sizeof(bitD->bitContainer)*8);
+    bitD->ptr -= bitD->bitsConsumed >> 3;
+    bitD->bitsConsumed &= 7;
+    bitD->bitContainer = MEM_readLEST(bitD->ptr);
+    return BIT_DStream_unfinished;
+}
+
+/*! BIT_reloadDStream() :
+ *  Refill `bitD` from buffer previously set in BIT_initDStream() .
+ *  This function is safe, it guarantees it will not read beyond src buffer.
+ * @return : status of `BIT_DStream_t` internal register.
+ *           when status == BIT_DStream_unfinished, internal register is filled with at least 25 or 57 bits */
+MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
+{
+    if (bitD->bitsConsumed > (sizeof(bitD->bitContainer)*8))  /* overflow detected, like end of stream */
+        return BIT_DStream_overflow;
+
+    if (bitD->ptr >= bitD->limitPtr) {
+        return BIT_reloadDStreamFast(bitD);
+    }
+    if (bitD->ptr == bitD->start) {
+        if (bitD->bitsConsumed < sizeof(bitD->bitContainer)*8) return BIT_DStream_endOfBuffer;
+        return BIT_DStream_completed;
+    }
+    /* start < ptr < limitPtr */
+    {   U32 nbBytes = bitD->bitsConsumed >> 3;
+        BIT_DStream_status result = BIT_DStream_unfinished;
+        if (bitD->ptr - nbBytes < bitD->start) {
+            nbBytes = (U32)(bitD->ptr - bitD->start);  /* ptr > start */
+            result = BIT_DStream_endOfBuffer;
+        }
+        bitD->ptr -= nbBytes;
+        bitD->bitsConsumed -= nbBytes*8;
+        bitD->bitContainer = MEM_readLEST(bitD->ptr);   /* reminder : srcSize > sizeof(bitD->bitContainer), otherwise bitD->ptr == bitD->start */
+        return result;
+    }
+}
+
+/*! BIT_endOfDStream() :
+ * @return : 1 if DStream has _exactly_ reached its end (all bits consumed).
+ */
+MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* DStream)
+{
+    return ((DStream->ptr == DStream->start) && (DStream->bitsConsumed == sizeof(DStream->bitContainer)*8));
+}
+
+
+#endif /* BITSTREAM_H_MODULE */
diff --git a/lib/zstd/common/compiler.h b/lib/zstd/common/compiler.h
new file mode 100644
index 000000000000..f5a9c70a228a
--- /dev/null
+++ b/lib/zstd/common/compiler.h
@@ -0,0 +1,177 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_COMPILER_H
+#define ZSTD_COMPILER_H
+
+/*-*******************************************************
+*  Compiler specifics
+*********************************************************/
+/* force inlining */
+
+#if !defined(ZSTD_NO_INLINE)
+#if (defined(__GNUC__) && !defined(__STRICT_ANSI__)) || defined(__cplusplus) || defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L   /* C99 */
+#  define INLINE_KEYWORD inline
+#else
+#  define INLINE_KEYWORD
+#endif
+
+#define FORCE_INLINE_ATTR __attribute__((always_inline))
+
+#else
+
+#define INLINE_KEYWORD
+#define FORCE_INLINE_ATTR
+
+#endif
+
+/*
+  On MSVC qsort requires that functions passed into it use the __cdecl calling conversion(CC).
+  This explictly marks such functions as __cdecl so that the code will still compile
+  if a CC other than __cdecl has been made the default.
+*/
+#define WIN_CDECL
+
+/*
+ * FORCE_INLINE_TEMPLATE is used to define C "templates", which take constant
+ * parameters. They must be inlined for the compiler to eliminate the constant
+ * branches.
+ */
+#define FORCE_INLINE_TEMPLATE static INLINE_KEYWORD FORCE_INLINE_ATTR
+/*
+ * HINT_INLINE is used to help the compiler generate better code. It is *not*
+ * used for "templates", so it can be tweaked based on the compilers
+ * performance.
+ *
+ * gcc-4.8 and gcc-4.9 have been shown to benefit from leaving off the
+ * always_inline attribute.
+ *
+ * clang up to 5.0.0 (trunk) benefit tremendously from the always_inline
+ * attribute.
+ */
+#if !defined(__clang__) && defined(__GNUC__) && __GNUC__ >= 4 && __GNUC_MINOR__ >= 8 && __GNUC__ < 5
+#  define HINT_INLINE static INLINE_KEYWORD
+#else
+#  define HINT_INLINE static INLINE_KEYWORD FORCE_INLINE_ATTR
+#endif
+
+/* UNUSED_ATTR tells the compiler it is okay if the function is unused. */
+#define UNUSED_ATTR __attribute__((unused))
+
+/* force no inlining */
+#define FORCE_NOINLINE static __attribute__((__noinline__))
+
+
+/* target attribute */
+#ifndef __has_attribute
+  #define __has_attribute(x) 0  /* Compatibility with non-clang compilers. */
+#endif
+#define TARGET_ATTRIBUTE(target) __attribute__((__target__(target)))
+
+/* Enable runtime BMI2 dispatch based on the CPU.
+ * Enabled for clang & gcc >=4.8 on x86 when BMI2 isn't enabled by default.
+ */
+#ifndef DYNAMIC_BMI2
+  #if ((defined(__clang__) && __has_attribute(__target__)) \
+      || (defined(__GNUC__) \
+          && (__GNUC__ >= 5 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 8)))) \
+      && (defined(__x86_64__) || defined(_M_X86)) \
+      && !defined(__BMI2__)
+  #  define DYNAMIC_BMI2 1
+  #else
+  #  define DYNAMIC_BMI2 0
+  #endif
+#endif
+
+/* prefetch
+ * can be disabled, by declaring NO_PREFETCH build macro */
+#if ( (__GNUC__ >= 4) || ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) )
+#  define PREFETCH_L1(ptr)  __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */)
+#  define PREFETCH_L2(ptr)  __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */)
+#elif defined(__aarch64__)
+#  define PREFETCH_L1(ptr)  __asm__ __volatile__("prfm pldl1keep, %0" ::"Q"(*(ptr)))
+#  define PREFETCH_L2(ptr)  __asm__ __volatile__("prfm pldl2keep, %0" ::"Q"(*(ptr)))
+#else
+#  define PREFETCH_L1(ptr) (void)(ptr)  /* disabled */
+#  define PREFETCH_L2(ptr) (void)(ptr)  /* disabled */
+#endif  /* NO_PREFETCH */
+
+#define CACHELINE_SIZE 64
+
+#define PREFETCH_AREA(p, s)  {            \
+    const char* const _ptr = (const char*)(p);  \
+    size_t const _size = (size_t)(s);     \
+    size_t _pos;                          \
+    for (_pos=0; _pos<_size; _pos+=CACHELINE_SIZE) {  \
+        PREFETCH_L2(_ptr + _pos);         \
+    }                                     \
+}
+
+/* vectorization
+ * older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax */
+#if !defined(__INTEL_COMPILER) && !defined(__clang__) && defined(__GNUC__)
+#  if (__GNUC__ == 4 && __GNUC_MINOR__ > 3) || (__GNUC__ >= 5)
+#    define DONT_VECTORIZE __attribute__((optimize("no-tree-vectorize")))
+#  else
+#    define DONT_VECTORIZE _Pragma("GCC optimize(\"no-tree-vectorize\")")
+#  endif
+#else
+#  define DONT_VECTORIZE
+#endif
+
+/* Tell the compiler that a branch is likely or unlikely.
+ * Only use these macros if it causes the compiler to generate better code.
+ * If you can remove a LIKELY/UNLIKELY annotation without speed changes in gcc
+ * and clang, please do.
+ */
+#define LIKELY(x) (__builtin_expect((x), 1))
+#define UNLIKELY(x) (__builtin_expect((x), 0))
+
+/* disable warnings */
+
+/*Like DYNAMIC_BMI2 but for compile time determination of BMI2 support*/
+
+
+/* compat. with non-clang compilers */
+#ifndef __has_builtin
+#  define __has_builtin(x) 0
+#endif
+
+/* compat. with non-clang compilers */
+#ifndef __has_feature
+#  define __has_feature(x) 0
+#endif
+
+/* C-language Attributes are added in C23. */
+#if defined(__STDC_VERSION__) && (__STDC_VERSION__ > 201710L) && defined(__has_c_attribute)
+# define ZSTD_HAS_C_ATTRIBUTE(x) __has_c_attribute(x)
+#else
+# define ZSTD_HAS_C_ATTRIBUTE(x) 0
+#endif
+
+/* Only use C++ attributes in C++. Some compilers report support for C++
+ * attributes when compiling with C.
+ */
+#define ZSTD_HAS_CPP_ATTRIBUTE(x) 0
+
+/* Define ZSTD_FALLTHROUGH macro for annotating switch case with the 'fallthrough' attribute.
+ * - C23: https://en.cppreference.com/w/c/language/attributes/fallthrough
+ * - CPP17: https://en.cppreference.com/w/cpp/language/attributes/fallthrough
+ * - Else: __attribute__((__fallthrough__))
+ */
+#define ZSTD_FALLTHROUGH fallthrough
+
+/* detects whether we are being compiled under msan */
+
+
+/* detects whether we are being compiled under asan */
+
+
+#endif /* ZSTD_COMPILER_H */
diff --git a/lib/zstd/common/cpu.h b/lib/zstd/common/cpu.h
new file mode 100644
index 000000000000..0db7b42407ee
--- /dev/null
+++ b/lib/zstd/common/cpu.h
@@ -0,0 +1,194 @@
+/*
+ * Copyright (c) Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_COMMON_CPU_H
+#define ZSTD_COMMON_CPU_H
+
+/*
+ * Implementation taken from folly/CpuId.h
+ * https://github.com/facebook/folly/blob/master/folly/CpuId.h
+ */
+
+#include "mem.h"
+
+
+typedef struct {
+    U32 f1c;
+    U32 f1d;
+    U32 f7b;
+    U32 f7c;
+} ZSTD_cpuid_t;
+
+MEM_STATIC ZSTD_cpuid_t ZSTD_cpuid(void) {
+    U32 f1c = 0;
+    U32 f1d = 0;
+    U32 f7b = 0;
+    U32 f7c = 0;
+#if defined(__i386__) && defined(__PIC__) && !defined(__clang__) && defined(__GNUC__)
+    /* The following block like the normal cpuid branch below, but gcc
+     * reserves ebx for use of its pic register so we must specially
+     * handle the save and restore to avoid clobbering the register
+     */
+    U32 n;
+    __asm__(
+        "pushl %%ebx\n\t"
+        "cpuid\n\t"
+        "popl %%ebx\n\t"
+        : "=a"(n)
+        : "a"(0)
+        : "ecx", "edx");
+    if (n >= 1) {
+      U32 f1a;
+      __asm__(
+          "pushl %%ebx\n\t"
+          "cpuid\n\t"
+          "popl %%ebx\n\t"
+          : "=a"(f1a), "=c"(f1c), "=d"(f1d)
+          : "a"(1));
+    }
+    if (n >= 7) {
+      __asm__(
+          "pushl %%ebx\n\t"
+          "cpuid\n\t"
+          "movl %%ebx, %%eax\n\t"
+          "popl %%ebx"
+          : "=a"(f7b), "=c"(f7c)
+          : "a"(7), "c"(0)
+          : "edx");
+    }
+#elif defined(__x86_64__) || defined(_M_X64) || defined(__i386__)
+    U32 n;
+    __asm__("cpuid" : "=a"(n) : "a"(0) : "ebx", "ecx", "edx");
+    if (n >= 1) {
+      U32 f1a;
+      __asm__("cpuid" : "=a"(f1a), "=c"(f1c), "=d"(f1d) : "a"(1) : "ebx");
+    }
+    if (n >= 7) {
+      U32 f7a;
+      __asm__("cpuid"
+              : "=a"(f7a), "=b"(f7b), "=c"(f7c)
+              : "a"(7), "c"(0)
+              : "edx");
+    }
+#endif
+    {
+        ZSTD_cpuid_t cpuid;
+        cpuid.f1c = f1c;
+        cpuid.f1d = f1d;
+        cpuid.f7b = f7b;
+        cpuid.f7c = f7c;
+        return cpuid;
+    }
+}
+
+#define X(name, r, bit)                                                        \
+  MEM_STATIC int ZSTD_cpuid_##name(ZSTD_cpuid_t const cpuid) {                 \
+    return ((cpuid.r) & (1U << bit)) != 0;                                     \
+  }
+
+/* cpuid(1): Processor Info and Feature Bits. */
+#define C(name, bit) X(name, f1c, bit)
+  C(sse3, 0)
+  C(pclmuldq, 1)
+  C(dtes64, 2)
+  C(monitor, 3)
+  C(dscpl, 4)
+  C(vmx, 5)
+  C(smx, 6)
+  C(eist, 7)
+  C(tm2, 8)
+  C(ssse3, 9)
+  C(cnxtid, 10)
+  C(fma, 12)
+  C(cx16, 13)
+  C(xtpr, 14)
+  C(pdcm, 15)
+  C(pcid, 17)
+  C(dca, 18)
+  C(sse41, 19)
+  C(sse42, 20)
+  C(x2apic, 21)
+  C(movbe, 22)
+  C(popcnt, 23)
+  C(tscdeadline, 24)
+  C(aes, 25)
+  C(xsave, 26)
+  C(osxsave, 27)
+  C(avx, 28)
+  C(f16c, 29)
+  C(rdrand, 30)
+#undef C
+#define D(name, bit) X(name, f1d, bit)
+  D(fpu, 0)
+  D(vme, 1)
+  D(de, 2)
+  D(pse, 3)
+  D(tsc, 4)
+  D(msr, 5)
+  D(pae, 6)
+  D(mce, 7)
+  D(cx8, 8)
+  D(apic, 9)
+  D(sep, 11)
+  D(mtrr, 12)
+  D(pge, 13)
+  D(mca, 14)
+  D(cmov, 15)
+  D(pat, 16)
+  D(pse36, 17)
+  D(psn, 18)
+  D(clfsh, 19)
+  D(ds, 21)
+  D(acpi, 22)
+  D(mmx, 23)
+  D(fxsr, 24)
+  D(sse, 25)
+  D(sse2, 26)
+  D(ss, 27)
+  D(htt, 28)
+  D(tm, 29)
+  D(pbe, 31)
+#undef D
+
+/* cpuid(7): Extended Features. */
+#define B(name, bit) X(name, f7b, bit)
+  B(bmi1, 3)
+  B(hle, 4)
+  B(avx2, 5)
+  B(smep, 7)
+  B(bmi2, 8)
+  B(erms, 9)
+  B(invpcid, 10)
+  B(rtm, 11)
+  B(mpx, 14)
+  B(avx512f, 16)
+  B(avx512dq, 17)
+  B(rdseed, 18)
+  B(adx, 19)
+  B(smap, 20)
+  B(avx512ifma, 21)
+  B(pcommit, 22)
+  B(clflushopt, 23)
+  B(clwb, 24)
+  B(avx512pf, 26)
+  B(avx512er, 27)
+  B(avx512cd, 28)
+  B(sha, 29)
+  B(avx512bw, 30)
+  B(avx512vl, 31)
+#undef B
+#define C(name, bit) X(name, f7c, bit)
+  C(prefetchwt1, 0)
+  C(avx512vbmi, 1)
+#undef C
+
+#undef X
+
+#endif /* ZSTD_COMMON_CPU_H */
diff --git a/lib/zstd/common/debug.c b/lib/zstd/common/debug.c
new file mode 100644
index 000000000000..bb863c9ea616
--- /dev/null
+++ b/lib/zstd/common/debug.c
@@ -0,0 +1,24 @@
+/* ******************************************************************
+ * debug
+ * Part of FSE library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+
+/*
+ * This module only hosts one global variable
+ * which can be used to dynamically influence the verbosity of traces,
+ * such as DEBUGLOG and RAWLOG
+ */
+
+#include "debug.h"
+
+int g_debuglevel = DEBUGLEVEL;
diff --git a/lib/zstd/common/debug.h b/lib/zstd/common/debug.h
new file mode 100644
index 000000000000..6dd88d1fbd02
--- /dev/null
+++ b/lib/zstd/common/debug.h
@@ -0,0 +1,101 @@
+/* ******************************************************************
+ * debug
+ * Part of FSE library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+
+/*
+ * The purpose of this header is to enable debug functions.
+ * They regroup assert(), DEBUGLOG() and RAWLOG() for run-time,
+ * and DEBUG_STATIC_ASSERT() for compile-time.
+ *
+ * By default, DEBUGLEVEL==0, which means run-time debug is disabled.
+ *
+ * Level 1 enables assert() only.
+ * Starting level 2, traces can be generated and pushed to stderr.
+ * The higher the level, the more verbose the traces.
+ *
+ * It's possible to dynamically adjust level using variable g_debug_level,
+ * which is only declared if DEBUGLEVEL>=2,
+ * and is a global variable, not multi-thread protected (use with care)
+ */
+
+#ifndef DEBUG_H_12987983217
+#define DEBUG_H_12987983217
+
+
+
+/* static assert is triggered at compile time, leaving no runtime artefact.
+ * static assert only works with compile-time constants.
+ * Also, this variant can only be used inside a function. */
+#define DEBUG_STATIC_ASSERT(c) (void)sizeof(char[(c) ? 1 : -1])
+
+
+/* DEBUGLEVEL is expected to be defined externally,
+ * typically through compiler command line.
+ * Value must be a number. */
+#ifndef DEBUGLEVEL
+#  define DEBUGLEVEL 0
+#endif
+
+
+/* recommended values for DEBUGLEVEL :
+ * 0 : release mode, no debug, all run-time checks disabled
+ * 1 : enables assert() only, no display
+ * 2 : reserved, for currently active debug path
+ * 3 : events once per object lifetime (CCtx, CDict, etc.)
+ * 4 : events once per frame
+ * 5 : events once per block
+ * 6 : events once per sequence (verbose)
+ * 7+: events at every position (*very* verbose)
+ *
+ * It's generally inconvenient to output traces > 5.
+ * In which case, it's possible to selectively trigger high verbosity levels
+ * by modifying g_debug_level.
+ */
+
+#if (DEBUGLEVEL>=1)
+#  define ZSTD_DEPS_NEED_ASSERT
+#  include "zstd_deps.h"
+#else
+#  ifndef assert   /* assert may be already defined, due to prior #include <assert.h> */
+#    define assert(condition) ((void)0)   /* disable assert (default) */
+#  endif
+#endif
+
+#if (DEBUGLEVEL>=2)
+#  define ZSTD_DEPS_NEED_IO
+#  include "zstd_deps.h"
+extern int g_debuglevel; /* the variable is only declared,
+                            it actually lives in debug.c,
+                            and is shared by the whole process.
+                            It's not thread-safe.
+                            It's useful when enabling very verbose levels
+                            on selective conditions (such as position in src) */
+
+#  define RAWLOG(l, ...) {                                       \
+                if (l<=g_debuglevel) {                           \
+                    ZSTD_DEBUG_PRINT(__VA_ARGS__);               \
+            }   }
+#  define DEBUGLOG(l, ...) {                                     \
+                if (l<=g_debuglevel) {                           \
+                    ZSTD_DEBUG_PRINT(__FILE__ ": " __VA_ARGS__); \
+                    ZSTD_DEBUG_PRINT(" \n");                     \
+            }   }
+#else
+#  define RAWLOG(l, ...)      {}    /* disabled */
+#  define DEBUGLOG(l, ...)    {}    /* disabled */
+#endif
+
+
+
+#endif /* DEBUG_H_12987983217 */
diff --git a/lib/zstd/common/entropy_common.c b/lib/zstd/common/entropy_common.c
index 2b0a643c32c4..53b47a2b52ff 100644
--- a/lib/zstd/common/entropy_common.c
+++ b/lib/zstd/common/entropy_common.c
@@ -1,243 +1,357 @@
-/*
+/* ******************************************************************
  * Common functions of New Generation Entropy library
- * Copyright (C) 2016, Yann Collet.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *  You can contact the author at :
+ *  - FSE+HUF source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *  - Public forum : https://groups.google.com/forum/#!forum/lz4c
  *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
 
 /* *************************************
 *  Dependencies
 ***************************************/
-#include "error_private.h" /* ERR_*, ERROR */
+#include "mem.h"
+#include "error_private.h"       /* ERR_*, ERROR */
+#define FSE_STATIC_LINKING_ONLY  /* FSE_MIN_TABLELOG */
 #include "fse.h"
+#define HUF_STATIC_LINKING_ONLY  /* HUF_TABLELOG_ABSOLUTEMAX */
 #include "huf.h"
-#include "mem.h"
+
 
 /*===   Version   ===*/
 unsigned FSE_versionNumber(void) { return FSE_VERSION_NUMBER; }
 
+
 /*===   Error Management   ===*/
 unsigned FSE_isError(size_t code) { return ERR_isError(code); }
+const char* FSE_getErrorName(size_t code) { return ERR_getErrorName(code); }
 
 unsigned HUF_isError(size_t code) { return ERR_isError(code); }
+const char* HUF_getErrorName(size_t code) { return ERR_getErrorName(code); }
+
 
 /*-**************************************************************
 *  FSE NCount encoding-decoding
 ****************************************************************/
-size_t FSE_readNCount(short *normalizedCounter, unsigned *maxSVPtr, unsigned *tableLogPtr, const void *headerBuffer, size_t hbSize)
+static U32 FSE_ctz(U32 val)
+{
+    assert(val != 0);
+    {
+#   if (__GNUC__ >= 3)   /* GCC Intrinsic */
+        return __builtin_ctz(val);
+#   else   /* Software version */
+        U32 count = 0;
+        while ((val & 1) == 0) {
+            val >>= 1;
+            ++count;
+        }
+        return count;
+#   endif
+    }
+}
+
+FORCE_INLINE_TEMPLATE
+size_t FSE_readNCount_body(short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+                           const void* headerBuffer, size_t hbSize)
+{
+    const BYTE* const istart = (const BYTE*) headerBuffer;
+    const BYTE* const iend = istart + hbSize;
+    const BYTE* ip = istart;
+    int nbBits;
+    int remaining;
+    int threshold;
+    U32 bitStream;
+    int bitCount;
+    unsigned charnum = 0;
+    unsigned const maxSV1 = *maxSVPtr + 1;
+    int previous0 = 0;
+
+    if (hbSize < 8) {
+        /* This function only works when hbSize >= 8 */
+        char buffer[8] = {0};
+        ZSTD_memcpy(buffer, headerBuffer, hbSize);
+        {   size_t const countSize = FSE_readNCount(normalizedCounter, maxSVPtr, tableLogPtr,
+                                                    buffer, sizeof(buffer));
+            if (FSE_isError(countSize)) return countSize;
+            if (countSize > hbSize) return ERROR(corruption_detected);
+            return countSize;
+    }   }
+    assert(hbSize >= 8);
+
+    /* init */
+    ZSTD_memset(normalizedCounter, 0, (*maxSVPtr+1) * sizeof(normalizedCounter[0]));   /* all symbols not present in NCount have a frequency of 0 */
+    bitStream = MEM_readLE32(ip);
+    nbBits = (bitStream & 0xF) + FSE_MIN_TABLELOG;   /* extract tableLog */
+    if (nbBits > FSE_TABLELOG_ABSOLUTE_MAX) return ERROR(tableLog_tooLarge);
+    bitStream >>= 4;
+    bitCount = 4;
+    *tableLogPtr = nbBits;
+    remaining = (1<<nbBits)+1;
+    threshold = 1<<nbBits;
+    nbBits++;
+
+    for (;;) {
+        if (previous0) {
+            /* Count the number of repeats. Each time the
+             * 2-bit repeat code is 0b11 there is another
+             * repeat.
+             * Avoid UB by setting the high bit to 1.
+             */
+            int repeats = FSE_ctz(~bitStream | 0x80000000) >> 1;
+            while (repeats >= 12) {
+                charnum += 3 * 12;
+                if (LIKELY(ip <= iend-7)) {
+                    ip += 3;
+                } else {
+                    bitCount -= (int)(8 * (iend - 7 - ip));
+                    bitCount &= 31;
+                    ip = iend - 4;
+                }
+                bitStream = MEM_readLE32(ip) >> bitCount;
+                repeats = FSE_ctz(~bitStream | 0x80000000) >> 1;
+            }
+            charnum += 3 * repeats;
+            bitStream >>= 2 * repeats;
+            bitCount += 2 * repeats;
+
+            /* Add the final repeat which isn't 0b11. */
+            assert((bitStream & 3) < 3);
+            charnum += bitStream & 3;
+            bitCount += 2;
+
+            /* This is an error, but break and return an error
+             * at the end, because returning out of a loop makes
+             * it harder for the compiler to optimize.
+             */
+            if (charnum >= maxSV1) break;
+
+            /* We don't need to set the normalized count to 0
+             * because we already memset the whole buffer to 0.
+             */
+
+            if (LIKELY(ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) {
+                assert((bitCount >> 3) <= 3); /* For first condition to work */
+                ip += bitCount>>3;
+                bitCount &= 7;
+            } else {
+                bitCount -= (int)(8 * (iend - 4 - ip));
+                bitCount &= 31;
+                ip = iend - 4;
+            }
+            bitStream = MEM_readLE32(ip) >> bitCount;
+        }
+        {
+            int const max = (2*threshold-1) - remaining;
+            int count;
+
+            if ((bitStream & (threshold-1)) < (U32)max) {
+                count = bitStream & (threshold-1);
+                bitCount += nbBits-1;
+            } else {
+                count = bitStream & (2*threshold-1);
+                if (count >= threshold) count -= max;
+                bitCount += nbBits;
+            }
+
+            count--;   /* extra accuracy */
+            /* When it matters (small blocks), this is a
+             * predictable branch, because we don't use -1.
+             */
+            if (count >= 0) {
+                remaining -= count;
+            } else {
+                assert(count == -1);
+                remaining += count;
+            }
+            normalizedCounter[charnum++] = (short)count;
+            previous0 = !count;
+
+            assert(threshold > 1);
+            if (remaining < threshold) {
+                /* This branch can be folded into the
+                 * threshold update condition because we
+                 * know that threshold > 1.
+                 */
+                if (remaining <= 1) break;
+                nbBits = BIT_highbit32(remaining) + 1;
+                threshold = 1 << (nbBits - 1);
+            }
+            if (charnum >= maxSV1) break;
+
+            if (LIKELY(ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) {
+                ip += bitCount>>3;
+                bitCount &= 7;
+            } else {
+                bitCount -= (int)(8 * (iend - 4 - ip));
+                bitCount &= 31;
+                ip = iend - 4;
+            }
+            bitStream = MEM_readLE32(ip) >> bitCount;
+    }   }
+    if (remaining != 1) return ERROR(corruption_detected);
+    /* Only possible when there are too many zeros. */
+    if (charnum > maxSV1) return ERROR(maxSymbolValue_tooSmall);
+    if (bitCount > 32) return ERROR(corruption_detected);
+    *maxSVPtr = charnum-1;
+
+    ip += (bitCount+7)>>3;
+    return ip-istart;
+}
+
+/* Avoids the FORCE_INLINE of the _body() function. */
+static size_t FSE_readNCount_body_default(
+        short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+        const void* headerBuffer, size_t hbSize)
 {
-	const BYTE *const istart = (const BYTE *)headerBuffer;
-	const BYTE *const iend = istart + hbSize;
-	const BYTE *ip = istart;
-	int nbBits;
-	int remaining;
-	int threshold;
-	U32 bitStream;
-	int bitCount;
-	unsigned charnum = 0;
-	int previous0 = 0;
-
-	if (hbSize < 4)
-		return ERROR(srcSize_wrong);
-	bitStream = ZSTD_readLE32(ip);
-	nbBits = (bitStream & 0xF) + FSE_MIN_TABLELOG; /* extract tableLog */
-	if (nbBits > FSE_TABLELOG_ABSOLUTE_MAX)
-		return ERROR(tableLog_tooLarge);
-	bitStream >>= 4;
-	bitCount = 4;
-	*tableLogPtr = nbBits;
-	remaining = (1 << nbBits) + 1;
-	threshold = 1 << nbBits;
-	nbBits++;
-
-	while ((remaining > 1) & (charnum <= *maxSVPtr)) {
-		if (previous0) {
-			unsigned n0 = charnum;
-			while ((bitStream & 0xFFFF) == 0xFFFF) {
-				n0 += 24;
-				if (ip < iend - 5) {
-					ip += 2;
-					bitStream = ZSTD_readLE32(ip) >> bitCount;
-				} else {
-					bitStream >>= 16;
-					bitCount += 16;
-				}
-			}
-			while ((bitStream & 3) == 3) {
-				n0 += 3;
-				bitStream >>= 2;
-				bitCount += 2;
-			}
-			n0 += bitStream & 3;
-			bitCount += 2;
-			if (n0 > *maxSVPtr)
-				return ERROR(maxSymbolValue_tooSmall);
-			while (charnum < n0)
-				normalizedCounter[charnum++] = 0;
-			if ((ip <= iend - 7) || (ip + (bitCount >> 3) <= iend - 4)) {
-				ip += bitCount >> 3;
-				bitCount &= 7;
-				bitStream = ZSTD_readLE32(ip) >> bitCount;
-			} else {
-				bitStream >>= 2;
-			}
-		}
-		{
-			int const max = (2 * threshold - 1) - remaining;
-			int count;
-
-			if ((bitStream & (threshold - 1)) < (U32)max) {
-				count = bitStream & (threshold - 1);
-				bitCount += nbBits - 1;
-			} else {
-				count = bitStream & (2 * threshold - 1);
-				if (count >= threshold)
-					count -= max;
-				bitCount += nbBits;
-			}
-
-			count--;				 /* extra accuracy */
-			remaining -= count < 0 ? -count : count; /* -1 means +1 */
-			normalizedCounter[charnum++] = (short)count;
-			previous0 = !count;
-			while (remaining < threshold) {
-				nbBits--;
-				threshold >>= 1;
-			}
-
-			if ((ip <= iend - 7) || (ip + (bitCount >> 3) <= iend - 4)) {
-				ip += bitCount >> 3;
-				bitCount &= 7;
-			} else {
-				bitCount -= (int)(8 * (iend - 4 - ip));
-				ip = iend - 4;
-			}
-			bitStream = ZSTD_readLE32(ip) >> (bitCount & 31);
-		}
-	} /* while ((remaining>1) & (charnum<=*maxSVPtr)) */
-	if (remaining != 1)
-		return ERROR(corruption_detected);
-	if (bitCount > 32)
-		return ERROR(corruption_detected);
-	*maxSVPtr = charnum - 1;
-
-	ip += (bitCount + 7) >> 3;
-	return ip - istart;
+    return FSE_readNCount_body(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize);
 }
 
+#if DYNAMIC_BMI2
+TARGET_ATTRIBUTE("bmi2") static size_t FSE_readNCount_body_bmi2(
+        short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+        const void* headerBuffer, size_t hbSize)
+{
+    return FSE_readNCount_body(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize);
+}
+#endif
+
+size_t FSE_readNCount_bmi2(
+        short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+        const void* headerBuffer, size_t hbSize, int bmi2)
+{
+#if DYNAMIC_BMI2
+    if (bmi2) {
+        return FSE_readNCount_body_bmi2(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize);
+    }
+#endif
+    (void)bmi2;
+    return FSE_readNCount_body_default(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize);
+}
+
+size_t FSE_readNCount(
+        short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+        const void* headerBuffer, size_t hbSize)
+{
+    return FSE_readNCount_bmi2(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize, /* bmi2 */ 0);
+}
+
+
 /*! HUF_readStats() :
-	Read compact Huffman tree, saved by HUF_writeCTable().
-	`huffWeight` is destination buffer.
-	`rankStats` is assumed to be a table of at least HUF_TABLELOG_MAX U32.
-	@return : size read from `src` , or an error Code .
-	Note : Needed by HUF_readCTable() and HUF_readDTableX?() .
+    Read compact Huffman tree, saved by HUF_writeCTable().
+    `huffWeight` is destination buffer.
+    `rankStats` is assumed to be a table of at least HUF_TABLELOG_MAX U32.
+    @return : size read from `src` , or an error Code .
+    Note : Needed by HUF_readCTable() and HUF_readDTableX?() .
 */
-size_t HUF_readStats_wksp(BYTE *huffWeight, size_t hwSize, U32 *rankStats, U32 *nbSymbolsPtr, U32 *tableLogPtr, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
+size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                     U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize)
+{
+    U32 wksp[HUF_READ_STATS_WORKSPACE_SIZE_U32];
+    return HUF_readStats_wksp(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, wksp, sizeof(wksp), /* bmi2 */ 0);
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_readStats_body(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                   U32* nbSymbolsPtr, U32* tableLogPtr,
+                   const void* src, size_t srcSize,
+                   void* workSpace, size_t wkspSize,
+                   int bmi2)
+{
+    U32 weightTotal;
+    const BYTE* ip = (const BYTE*) src;
+    size_t iSize;
+    size_t oSize;
+
+    if (!srcSize) return ERROR(srcSize_wrong);
+    iSize = ip[0];
+    /* ZSTD_memset(huffWeight, 0, hwSize);   *//* is not necessary, even though some analyzer complain ... */
+
+    if (iSize >= 128) {  /* special header */
+        oSize = iSize - 127;
+        iSize = ((oSize+1)/2);
+        if (iSize+1 > srcSize) return ERROR(srcSize_wrong);
+        if (oSize >= hwSize) return ERROR(corruption_detected);
+        ip += 1;
+        {   U32 n;
+            for (n=0; n<oSize; n+=2) {
+                huffWeight[n]   = ip[n/2] >> 4;
+                huffWeight[n+1] = ip[n/2] & 15;
+    }   }   }
+    else  {   /* header compressed with FSE (normal case) */
+        if (iSize+1 > srcSize) return ERROR(srcSize_wrong);
+        /* max (hwSize-1) values decoded, as last one is implied */
+        oSize = FSE_decompress_wksp_bmi2(huffWeight, hwSize-1, ip+1, iSize, 6, workSpace, wkspSize, bmi2);
+        if (FSE_isError(oSize)) return oSize;
+    }
+
+    /* collect weight stats */
+    ZSTD_memset(rankStats, 0, (HUF_TABLELOG_MAX + 1) * sizeof(U32));
+    weightTotal = 0;
+    {   U32 n; for (n=0; n<oSize; n++) {
+            if (huffWeight[n] >= HUF_TABLELOG_MAX) return ERROR(corruption_detected);
+            rankStats[huffWeight[n]]++;
+            weightTotal += (1 << huffWeight[n]) >> 1;
+    }   }
+    if (weightTotal == 0) return ERROR(corruption_detected);
+
+    /* get last non-null symbol weight (implied, total must be 2^n) */
+    {   U32 const tableLog = BIT_highbit32(weightTotal) + 1;
+        if (tableLog > HUF_TABLELOG_MAX) return ERROR(corruption_detected);
+        *tableLogPtr = tableLog;
+        /* determine last weight */
+        {   U32 const total = 1 << tableLog;
+            U32 const rest = total - weightTotal;
+            U32 const verif = 1 << BIT_highbit32(rest);
+            U32 const lastWeight = BIT_highbit32(rest) + 1;
+            if (verif != rest) return ERROR(corruption_detected);    /* last value must be a clean power of 2 */
+            huffWeight[oSize] = (BYTE)lastWeight;
+            rankStats[lastWeight]++;
+    }   }
+
+    /* check tree construction validity */
+    if ((rankStats[1] < 2) || (rankStats[1] & 1)) return ERROR(corruption_detected);   /* by construction : at least 2 elts of rank 1, must be even */
+
+    /* results */
+    *nbSymbolsPtr = (U32)(oSize+1);
+    return iSize+1;
+}
+
+/* Avoids the FORCE_INLINE of the _body() function. */
+static size_t HUF_readStats_body_default(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                     U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize,
+                     void* workSpace, size_t wkspSize)
+{
+    return HUF_readStats_body(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize, 0);
+}
+
+#if DYNAMIC_BMI2
+static TARGET_ATTRIBUTE("bmi2") size_t HUF_readStats_body_bmi2(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                     U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize,
+                     void* workSpace, size_t wkspSize)
+{
+    return HUF_readStats_body(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize, 1);
+}
+#endif
+
+size_t HUF_readStats_wksp(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                     U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize,
+                     void* workSpace, size_t wkspSize,
+                     int bmi2)
 {
-	U32 weightTotal;
-	const BYTE *ip = (const BYTE *)src;
-	size_t iSize;
-	size_t oSize;
-
-	if (!srcSize)
-		return ERROR(srcSize_wrong);
-	iSize = ip[0];
-	/* memset(huffWeight, 0, hwSize);   */ /* is not necessary, even though some analyzer complain ... */
-
-	if (iSize >= 128) { /* special header */
-		oSize = iSize - 127;
-		iSize = ((oSize + 1) / 2);
-		if (iSize + 1 > srcSize)
-			return ERROR(srcSize_wrong);
-		if (oSize >= hwSize)
-			return ERROR(corruption_detected);
-		ip += 1;
-		{
-			U32 n;
-			for (n = 0; n < oSize; n += 2) {
-				huffWeight[n] = ip[n / 2] >> 4;
-				huffWeight[n + 1] = ip[n / 2] & 15;
-			}
-		}
-	} else {						 /* header compressed with FSE (normal case) */
-		if (iSize + 1 > srcSize)
-			return ERROR(srcSize_wrong);
-		oSize = FSE_decompress_wksp(huffWeight, hwSize - 1, ip + 1, iSize, 6, workspace, workspaceSize); /* max (hwSize-1) values decoded, as last one is implied */
-		if (FSE_isError(oSize))
-			return oSize;
-	}
-
-	/* collect weight stats */
-	memset(rankStats, 0, (HUF_TABLELOG_MAX + 1) * sizeof(U32));
-	weightTotal = 0;
-	{
-		U32 n;
-		for (n = 0; n < oSize; n++) {
-			if (huffWeight[n] >= HUF_TABLELOG_MAX)
-				return ERROR(corruption_detected);
-			rankStats[huffWeight[n]]++;
-			weightTotal += (1 << huffWeight[n]) >> 1;
-		}
-	}
-	if (weightTotal == 0)
-		return ERROR(corruption_detected);
-
-	/* get last non-null symbol weight (implied, total must be 2^n) */
-	{
-		U32 const tableLog = BIT_highbit32(weightTotal) + 1;
-		if (tableLog > HUF_TABLELOG_MAX)
-			return ERROR(corruption_detected);
-		*tableLogPtr = tableLog;
-		/* determine last weight */
-		{
-			U32 const total = 1 << tableLog;
-			U32 const rest = total - weightTotal;
-			U32 const verif = 1 << BIT_highbit32(rest);
-			U32 const lastWeight = BIT_highbit32(rest) + 1;
-			if (verif != rest)
-				return ERROR(corruption_detected); /* last value must be a clean power of 2 */
-			huffWeight[oSize] = (BYTE)lastWeight;
-			rankStats[lastWeight]++;
-		}
-	}
-
-	/* check tree construction validity */
-	if ((rankStats[1] < 2) || (rankStats[1] & 1))
-		return ERROR(corruption_detected); /* by construction : at least 2 elts of rank 1, must be even */
-
-	/* results */
-	*nbSymbolsPtr = (U32)(oSize + 1);
-	return iSize + 1;
+#if DYNAMIC_BMI2
+    if (bmi2) {
+        return HUF_readStats_body_bmi2(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize);
+    }
+#endif
+    (void)bmi2;
+    return HUF_readStats_body_default(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize);
 }
diff --git a/lib/zstd/common/error_private.c b/lib/zstd/common/error_private.c
new file mode 100644
index 000000000000..6d1135f8c373
--- /dev/null
+++ b/lib/zstd/common/error_private.c
@@ -0,0 +1,56 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* The purpose of this file is to have a single list of error strings embedded in binary */
+
+#include "error_private.h"
+
+const char* ERR_getErrorString(ERR_enum code)
+{
+#ifdef ZSTD_STRIP_ERROR_STRINGS
+    (void)code;
+    return "Error strings stripped";
+#else
+    static const char* const notErrorCode = "Unspecified error code";
+    switch( code )
+    {
+    case PREFIX(no_error): return "No error detected";
+    case PREFIX(GENERIC):  return "Error (generic)";
+    case PREFIX(prefix_unknown): return "Unknown frame descriptor";
+    case PREFIX(version_unsupported): return "Version not supported";
+    case PREFIX(frameParameter_unsupported): return "Unsupported frame parameter";
+    case PREFIX(frameParameter_windowTooLarge): return "Frame requires too much memory for decoding";
+    case PREFIX(corruption_detected): return "Corrupted block detected";
+    case PREFIX(checksum_wrong): return "Restored data doesn't match checksum";
+    case PREFIX(parameter_unsupported): return "Unsupported parameter";
+    case PREFIX(parameter_outOfBound): return "Parameter is out of bound";
+    case PREFIX(init_missing): return "Context should be init first";
+    case PREFIX(memory_allocation): return "Allocation error : not enough memory";
+    case PREFIX(workSpace_tooSmall): return "workSpace buffer is not large enough";
+    case PREFIX(stage_wrong): return "Operation not authorized at current processing stage";
+    case PREFIX(tableLog_tooLarge): return "tableLog requires too much memory : unsupported";
+    case PREFIX(maxSymbolValue_tooLarge): return "Unsupported max Symbol Value : too large";
+    case PREFIX(maxSymbolValue_tooSmall): return "Specified maxSymbolValue is too small";
+    case PREFIX(dictionary_corrupted): return "Dictionary is corrupted";
+    case PREFIX(dictionary_wrong): return "Dictionary mismatch";
+    case PREFIX(dictionaryCreation_failed): return "Cannot create Dictionary from provided samples";
+    case PREFIX(dstSize_tooSmall): return "Destination buffer is too small";
+    case PREFIX(srcSize_wrong): return "Src size is incorrect";
+    case PREFIX(dstBuffer_null): return "Operation on NULL destination buffer";
+        /* following error codes are not stable and may be removed or changed in a future version */
+    case PREFIX(frameIndex_tooLarge): return "Frame index is too large";
+    case PREFIX(seekableIO): return "An I/O error occurred when reading/seeking";
+    case PREFIX(dstBuffer_wrong): return "Destination buffer is wrong";
+    case PREFIX(srcBuffer_wrong): return "Source buffer is wrong";
+    case PREFIX(maxCode):
+    default: return notErrorCode;
+    }
+#endif
+}
diff --git a/lib/zstd/common/error_private.h b/lib/zstd/common/error_private.h
new file mode 100644
index 000000000000..bb60835cac30
--- /dev/null
+++ b/lib/zstd/common/error_private.h
@@ -0,0 +1,69 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* Note : this module is expected to remain private, do not expose it */
+
+#ifndef ERROR_H_MODULE
+#define ERROR_H_MODULE
+
+#ifdef __PBL__
+#define ZSTD_STRIP_ERROR_STRINGS
+#endif
+
+
+/* ****************************************
+*  Dependencies
+******************************************/
+#include "zstd_deps.h"    /* size_t */
+#include <linux/zstd_errors.h>  /* enum list */
+
+
+/* ****************************************
+*  Compiler-specific
+******************************************/
+#define ERR_STATIC static __attribute__((unused))
+
+
+/*-****************************************
+*  Customization (error_public.h)
+******************************************/
+typedef ZSTD_ErrorCode ERR_enum;
+#define PREFIX(name) ZSTD_error_##name
+
+
+/*-****************************************
+*  Error codes handling
+******************************************/
+#undef ERROR   /* already defined on Visual Studio */
+#define ERROR(name) ZSTD_ERROR(name)
+#define ZSTD_ERROR(name) ((size_t)-PREFIX(name))
+
+ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); }
+
+ERR_STATIC ERR_enum ERR_getErrorCode(size_t code) { if (!ERR_isError(code)) return (ERR_enum)0; return (ERR_enum) (0-code); }
+
+/* check and forward error code */
+#define CHECK_V_F(e, f) size_t const e = f; if (ERR_isError(e)) return e
+#define CHECK_F(f)   { CHECK_V_F(_var_err__, f); }
+
+
+/*-****************************************
+*  Error Strings
+******************************************/
+
+const char* ERR_getErrorString(ERR_enum code);   /* error_private.c */
+
+ERR_STATIC const char* ERR_getErrorName(size_t code)
+{
+    return ERR_getErrorString(ERR_getErrorCode(code));
+}
+
+
+#endif /* ERROR_H_MODULE */
diff --git a/lib/zstd/common/fse.h b/lib/zstd/common/fse.h
new file mode 100644
index 000000000000..0bb174c2c367
--- /dev/null
+++ b/lib/zstd/common/fse.h
@@ -0,0 +1,710 @@
+/* ******************************************************************
+ * FSE : Finite State Entropy codec
+ * Public Prototypes declaration
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+
+#ifndef FSE_H
+#define FSE_H
+
+
+/*-*****************************************
+*  Dependencies
+******************************************/
+#include "zstd_deps.h"    /* size_t, ptrdiff_t */
+
+
+/*-*****************************************
+*  FSE_PUBLIC_API : control library symbols visibility
+******************************************/
+#if defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) && defined(__GNUC__) && (__GNUC__ >= 4)
+#  define FSE_PUBLIC_API __attribute__ ((visibility ("default")))
+#elif defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1)   /* Visual expected */
+#  define FSE_PUBLIC_API __declspec(dllexport)
+#elif defined(FSE_DLL_IMPORT) && (FSE_DLL_IMPORT==1)
+#  define FSE_PUBLIC_API __declspec(dllimport) /* It isn't required but allows to generate better code, saving a function pointer load from the IAT and an indirect jump.*/
+#else
+#  define FSE_PUBLIC_API
+#endif
+
+/*------   Version   ------*/
+#define FSE_VERSION_MAJOR    0
+#define FSE_VERSION_MINOR    9
+#define FSE_VERSION_RELEASE  0
+
+#define FSE_LIB_VERSION FSE_VERSION_MAJOR.FSE_VERSION_MINOR.FSE_VERSION_RELEASE
+#define FSE_QUOTE(str) #str
+#define FSE_EXPAND_AND_QUOTE(str) FSE_QUOTE(str)
+#define FSE_VERSION_STRING FSE_EXPAND_AND_QUOTE(FSE_LIB_VERSION)
+
+#define FSE_VERSION_NUMBER  (FSE_VERSION_MAJOR *100*100 + FSE_VERSION_MINOR *100 + FSE_VERSION_RELEASE)
+FSE_PUBLIC_API unsigned FSE_versionNumber(void);   /*< library version number; to be used when checking dll version */
+
+
+/*-****************************************
+*  FSE simple functions
+******************************************/
+/*! FSE_compress() :
+    Compress content of buffer 'src', of size 'srcSize', into destination buffer 'dst'.
+    'dst' buffer must be already allocated. Compression runs faster is dstCapacity >= FSE_compressBound(srcSize).
+    @return : size of compressed data (<= dstCapacity).
+    Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
+                     if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression instead.
+                     if FSE_isError(return), compression failed (more details using FSE_getErrorName())
+*/
+FSE_PUBLIC_API size_t FSE_compress(void* dst, size_t dstCapacity,
+                             const void* src, size_t srcSize);
+
+/*! FSE_decompress():
+    Decompress FSE data from buffer 'cSrc', of size 'cSrcSize',
+    into already allocated destination buffer 'dst', of size 'dstCapacity'.
+    @return : size of regenerated data (<= maxDstSize),
+              or an error code, which can be tested using FSE_isError() .
+
+    ** Important ** : FSE_decompress() does not decompress non-compressible nor RLE data !!!
+    Why ? : making this distinction requires a header.
+    Header management is intentionally delegated to the user layer, which can better manage special cases.
+*/
+FSE_PUBLIC_API size_t FSE_decompress(void* dst,  size_t dstCapacity,
+                               const void* cSrc, size_t cSrcSize);
+
+
+/*-*****************************************
+*  Tool functions
+******************************************/
+FSE_PUBLIC_API size_t FSE_compressBound(size_t size);       /* maximum compressed size */
+
+/* Error Management */
+FSE_PUBLIC_API unsigned    FSE_isError(size_t code);        /* tells if a return value is an error code */
+FSE_PUBLIC_API const char* FSE_getErrorName(size_t code);   /* provides error code string (useful for debugging) */
+
+
+/*-*****************************************
+*  FSE advanced functions
+******************************************/
+/*! FSE_compress2() :
+    Same as FSE_compress(), but allows the selection of 'maxSymbolValue' and 'tableLog'
+    Both parameters can be defined as '0' to mean : use default value
+    @return : size of compressed data
+    Special values : if return == 0, srcData is not compressible => Nothing is stored within cSrc !!!
+                     if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression.
+                     if FSE_isError(return), it's an error code.
+*/
+FSE_PUBLIC_API size_t FSE_compress2 (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
+
+
+/*-*****************************************
+*  FSE detailed API
+******************************************/
+/*!
+FSE_compress() does the following:
+1. count symbol occurrence from source[] into table count[] (see hist.h)
+2. normalize counters so that sum(count[]) == Power_of_2 (2^tableLog)
+3. save normalized counters to memory buffer using writeNCount()
+4. build encoding table 'CTable' from normalized counters
+5. encode the data stream using encoding table 'CTable'
+
+FSE_decompress() does the following:
+1. read normalized counters with readNCount()
+2. build decoding table 'DTable' from normalized counters
+3. decode the data stream using decoding table 'DTable'
+
+The following API allows targeting specific sub-functions for advanced tasks.
+For example, it's possible to compress several blocks using the same 'CTable',
+or to save and provide normalized distribution using external method.
+*/
+
+/* *** COMPRESSION *** */
+
+/*! FSE_optimalTableLog():
+    dynamically downsize 'tableLog' when conditions are met.
+    It saves CPU time, by using smaller tables, while preserving or even improving compression ratio.
+    @return : recommended tableLog (necessarily <= 'maxTableLog') */
+FSE_PUBLIC_API unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
+
+/*! FSE_normalizeCount():
+    normalize counts so that sum(count[]) == Power_of_2 (2^tableLog)
+    'normalizedCounter' is a table of short, of minimum size (maxSymbolValue+1).
+    useLowProbCount is a boolean parameter which trades off compressed size for
+    faster header decoding. When it is set to 1, the compressed data will be slightly
+    smaller. And when it is set to 0, FSE_readNCount() and FSE_buildDTable() will be
+    faster. If you are compressing a small amount of data (< 2 KB) then useLowProbCount=0
+    is a good default, since header deserialization makes a big speed difference.
+    Otherwise, useLowProbCount=1 is a good default, since the speed difference is small.
+    @return : tableLog,
+              or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_normalizeCount(short* normalizedCounter, unsigned tableLog,
+                    const unsigned* count, size_t srcSize, unsigned maxSymbolValue, unsigned useLowProbCount);
+
+/*! FSE_NCountWriteBound():
+    Provides the maximum possible size of an FSE normalized table, given 'maxSymbolValue' and 'tableLog'.
+    Typically useful for allocation purpose. */
+FSE_PUBLIC_API size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog);
+
+/*! FSE_writeNCount():
+    Compactly save 'normalizedCounter' into 'buffer'.
+    @return : size of the compressed table,
+              or an errorCode, which can be tested using FSE_isError(). */
+FSE_PUBLIC_API size_t FSE_writeNCount (void* buffer, size_t bufferSize,
+                                 const short* normalizedCounter,
+                                 unsigned maxSymbolValue, unsigned tableLog);
+
+/*! Constructor and Destructor of FSE_CTable.
+    Note that FSE_CTable size depends on 'tableLog' and 'maxSymbolValue' */
+typedef unsigned FSE_CTable;   /* don't allocate that. It's only meant to be more restrictive than void* */
+FSE_PUBLIC_API FSE_CTable* FSE_createCTable (unsigned maxSymbolValue, unsigned tableLog);
+FSE_PUBLIC_API void        FSE_freeCTable (FSE_CTable* ct);
+
+/*! FSE_buildCTable():
+    Builds `ct`, which must be already allocated, using FSE_createCTable().
+    @return : 0, or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_buildCTable(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);
+
+/*! FSE_compress_usingCTable():
+    Compress `src` using `ct` into `dst` which must be already allocated.
+    @return : size of compressed data (<= `dstCapacity`),
+              or 0 if compressed data could not fit into `dst`,
+              or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_compress_usingCTable (void* dst, size_t dstCapacity, const void* src, size_t srcSize, const FSE_CTable* ct);
+
+/*!
+Tutorial :
+----------
+The first step is to count all symbols. FSE_count() does this job very fast.
+Result will be saved into 'count', a table of unsigned int, which must be already allocated, and have 'maxSymbolValuePtr[0]+1' cells.
+'src' is a table of bytes of size 'srcSize'. All values within 'src' MUST be <= maxSymbolValuePtr[0]
+maxSymbolValuePtr[0] will be updated, with its real value (necessarily <= original value)
+FSE_count() will return the number of occurrence of the most frequent symbol.
+This can be used to know if there is a single symbol within 'src', and to quickly evaluate its compressibility.
+If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
+
+The next step is to normalize the frequencies.
+FSE_normalizeCount() will ensure that sum of frequencies is == 2 ^'tableLog'.
+It also guarantees a minimum of 1 to any Symbol with frequency >= 1.
+You can use 'tableLog'==0 to mean "use default tableLog value".
+If you are unsure of which tableLog value to use, you can ask FSE_optimalTableLog(),
+which will provide the optimal valid tableLog given sourceSize, maxSymbolValue, and a user-defined maximum (0 means "default").
+
+The result of FSE_normalizeCount() will be saved into a table,
+called 'normalizedCounter', which is a table of signed short.
+'normalizedCounter' must be already allocated, and have at least 'maxSymbolValue+1' cells.
+The return value is tableLog if everything proceeded as expected.
+It is 0 if there is a single symbol within distribution.
+If there is an error (ex: invalid tableLog value), the function will return an ErrorCode (which can be tested using FSE_isError()).
+
+'normalizedCounter' can be saved in a compact manner to a memory area using FSE_writeNCount().
+'buffer' must be already allocated.
+For guaranteed success, buffer size must be at least FSE_headerBound().
+The result of the function is the number of bytes written into 'buffer'.
+If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError(); ex : buffer size too small).
+
+'normalizedCounter' can then be used to create the compression table 'CTable'.
+The space required by 'CTable' must be already allocated, using FSE_createCTable().
+You can then use FSE_buildCTable() to fill 'CTable'.
+If there is an error, both functions will return an ErrorCode (which can be tested using FSE_isError()).
+
+'CTable' can then be used to compress 'src', with FSE_compress_usingCTable().
+Similar to FSE_count(), the convention is that 'src' is assumed to be a table of char of size 'srcSize'
+The function returns the size of compressed data (without header), necessarily <= `dstCapacity`.
+If it returns '0', compressed data could not fit into 'dst'.
+If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
+*/
+
+
+/* *** DECOMPRESSION *** */
+
+/*! FSE_readNCount():
+    Read compactly saved 'normalizedCounter' from 'rBuffer'.
+    @return : size read from 'rBuffer',
+              or an errorCode, which can be tested using FSE_isError().
+              maxSymbolValuePtr[0] and tableLogPtr[0] will also be updated with their respective values */
+FSE_PUBLIC_API size_t FSE_readNCount (short* normalizedCounter,
+                           unsigned* maxSymbolValuePtr, unsigned* tableLogPtr,
+                           const void* rBuffer, size_t rBuffSize);
+
+/*! FSE_readNCount_bmi2():
+ * Same as FSE_readNCount() but pass bmi2=1 when your CPU supports BMI2 and 0 otherwise.
+ */
+FSE_PUBLIC_API size_t FSE_readNCount_bmi2(short* normalizedCounter,
+                           unsigned* maxSymbolValuePtr, unsigned* tableLogPtr,
+                           const void* rBuffer, size_t rBuffSize, int bmi2);
+
+/*! Constructor and Destructor of FSE_DTable.
+    Note that its size depends on 'tableLog' */
+typedef unsigned FSE_DTable;   /* don't allocate that. It's just a way to be more restrictive than void* */
+FSE_PUBLIC_API FSE_DTable* FSE_createDTable(unsigned tableLog);
+FSE_PUBLIC_API void        FSE_freeDTable(FSE_DTable* dt);
+
+/*! FSE_buildDTable():
+    Builds 'dt', which must be already allocated, using FSE_createDTable().
+    return : 0, or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_buildDTable (FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);
+
+/*! FSE_decompress_usingDTable():
+    Decompress compressed source `cSrc` of size `cSrcSize` using `dt`
+    into `dst` which must be already allocated.
+    @return : size of regenerated data (necessarily <= `dstCapacity`),
+              or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_decompress_usingDTable(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, const FSE_DTable* dt);
+
+/*!
+Tutorial :
+----------
+(Note : these functions only decompress FSE-compressed blocks.
+ If block is uncompressed, use memcpy() instead
+ If block is a single repeated byte, use memset() instead )
+
+The first step is to obtain the normalized frequencies of symbols.
+This can be performed by FSE_readNCount() if it was saved using FSE_writeNCount().
+'normalizedCounter' must be already allocated, and have at least 'maxSymbolValuePtr[0]+1' cells of signed short.
+In practice, that means it's necessary to know 'maxSymbolValue' beforehand,
+or size the table to handle worst case situations (typically 256).
+FSE_readNCount() will provide 'tableLog' and 'maxSymbolValue'.
+The result of FSE_readNCount() is the number of bytes read from 'rBuffer'.
+Note that 'rBufferSize' must be at least 4 bytes, even if useful information is less than that.
+If there is an error, the function will return an error code, which can be tested using FSE_isError().
+
+The next step is to build the decompression tables 'FSE_DTable' from 'normalizedCounter'.
+This is performed by the function FSE_buildDTable().
+The space required by 'FSE_DTable' must be already allocated using FSE_createDTable().
+If there is an error, the function will return an error code, which can be tested using FSE_isError().
+
+`FSE_DTable` can then be used to decompress `cSrc`, with FSE_decompress_usingDTable().
+`cSrcSize` must be strictly correct, otherwise decompression will fail.
+FSE_decompress_usingDTable() result will tell how many bytes were regenerated (<=`dstCapacity`).
+If there is an error, the function will return an error code, which can be tested using FSE_isError(). (ex: dst buffer too small)
+*/
+
+#endif  /* FSE_H */
+
+#if !defined(FSE_H_FSE_STATIC_LINKING_ONLY)
+#define FSE_H_FSE_STATIC_LINKING_ONLY
+
+/* *** Dependency *** */
+#include "bitstream.h"
+
+
+/* *****************************************
+*  Static allocation
+*******************************************/
+/* FSE buffer bounds */
+#define FSE_NCOUNTBOUND 512
+#define FSE_BLOCKBOUND(size) ((size) + ((size)>>7) + 4 /* fse states */ + sizeof(size_t) /* bitContainer */)
+#define FSE_COMPRESSBOUND(size) (FSE_NCOUNTBOUND + FSE_BLOCKBOUND(size))   /* Macro version, useful for static allocation */
+
+/* It is possible to statically allocate FSE CTable/DTable as a table of FSE_CTable/FSE_DTable using below macros */
+#define FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue)   (1 + (1<<((maxTableLog)-1)) + (((maxSymbolValue)+1)*2))
+#define FSE_DTABLE_SIZE_U32(maxTableLog)                   (1 + (1<<(maxTableLog)))
+
+/* or use the size to malloc() space directly. Pay attention to alignment restrictions though */
+#define FSE_CTABLE_SIZE(maxTableLog, maxSymbolValue)   (FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) * sizeof(FSE_CTable))
+#define FSE_DTABLE_SIZE(maxTableLog)                   (FSE_DTABLE_SIZE_U32(maxTableLog) * sizeof(FSE_DTable))
+
+
+/* *****************************************
+ *  FSE advanced API
+ ***************************************** */
+
+unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus);
+/*< same as FSE_optimalTableLog(), which used `minus==2` */
+
+/* FSE_compress_wksp() :
+ * Same as FSE_compress2(), but using an externally allocated scratch buffer (`workSpace`).
+ * FSE_COMPRESS_WKSP_SIZE_U32() provides the minimum size required for `workSpace` as a table of FSE_CTable.
+ */
+#define FSE_COMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue)   ( FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) + ((maxTableLog > 12) ? (1 << (maxTableLog - 2)) : 1024) )
+size_t FSE_compress_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
+
+size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits);
+/*< build a fake FSE_CTable, designed for a flat distribution, where each symbol uses nbBits */
+
+size_t FSE_buildCTable_rle (FSE_CTable* ct, unsigned char symbolValue);
+/*< build a fake FSE_CTable, designed to compress always the same symbolValue */
+
+/* FSE_buildCTable_wksp() :
+ * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
+ * `wkspSize` must be >= `FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(maxSymbolValue, tableLog)` of `unsigned`.
+ */
+#define FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(maxSymbolValue, tableLog) (maxSymbolValue + 2 + (1ull << (tableLog - 2)))
+#define FSE_BUILD_CTABLE_WORKSPACE_SIZE(maxSymbolValue, tableLog) (sizeof(unsigned) * FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(maxSymbolValue, tableLog))
+size_t FSE_buildCTable_wksp(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
+
+#define FSE_BUILD_DTABLE_WKSP_SIZE(maxTableLog, maxSymbolValue) (sizeof(short) * (maxSymbolValue + 1) + (1ULL << maxTableLog) + 8)
+#define FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) ((FSE_BUILD_DTABLE_WKSP_SIZE(maxTableLog, maxSymbolValue) + sizeof(unsigned) - 1) / sizeof(unsigned))
+FSE_PUBLIC_API size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
+/*< Same as FSE_buildDTable(), using an externally allocated `workspace` produced with `FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxSymbolValue)` */
+
+size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits);
+/*< build a fake FSE_DTable, designed to read a flat distribution where each symbol uses nbBits */
+
+size_t FSE_buildDTable_rle (FSE_DTable* dt, unsigned char symbolValue);
+/*< build a fake FSE_DTable, designed to always generate the same symbolValue */
+
+#define FSE_DECOMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) (FSE_DTABLE_SIZE_U32(maxTableLog) + FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) + (FSE_MAX_SYMBOL_VALUE + 1) / 2 + 1)
+#define FSE_DECOMPRESS_WKSP_SIZE(maxTableLog, maxSymbolValue) (FSE_DECOMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) * sizeof(unsigned))
+size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize);
+/*< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DECOMPRESS_WKSP_SIZE_U32(maxLog, maxSymbolValue)` */
+
+size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize, int bmi2);
+/*< Same as FSE_decompress_wksp() but with dynamic BMI2 support. Pass 1 if your CPU supports BMI2 or 0 if it doesn't. */
+
+typedef enum {
+   FSE_repeat_none,  /*< Cannot use the previous table */
+   FSE_repeat_check, /*< Can use the previous table but it must be checked */
+   FSE_repeat_valid  /*< Can use the previous table and it is assumed to be valid */
+ } FSE_repeat;
+
+/* *****************************************
+*  FSE symbol compression API
+*******************************************/
+/*!
+   This API consists of small unitary functions, which highly benefit from being inlined.
+   Hence their body are included in next section.
+*/
+typedef struct {
+    ptrdiff_t   value;
+    const void* stateTable;
+    const void* symbolTT;
+    unsigned    stateLog;
+} FSE_CState_t;
+
+static void FSE_initCState(FSE_CState_t* CStatePtr, const FSE_CTable* ct);
+
+static void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* CStatePtr, unsigned symbol);
+
+static void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* CStatePtr);
+
+/*<
+These functions are inner components of FSE_compress_usingCTable().
+They allow the creation of custom streams, mixing multiple tables and bit sources.
+
+A key property to keep in mind is that encoding and decoding are done **in reverse direction**.
+So the first symbol you will encode is the last you will decode, like a LIFO stack.
+
+You will need a few variables to track your CStream. They are :
+
+FSE_CTable    ct;         // Provided by FSE_buildCTable()
+BIT_CStream_t bitStream;  // bitStream tracking structure
+FSE_CState_t  state;      // State tracking structure (can have several)
+
+
+The first thing to do is to init bitStream and state.
+    size_t errorCode = BIT_initCStream(&bitStream, dstBuffer, maxDstSize);
+    FSE_initCState(&state, ct);
+
+Note that BIT_initCStream() can produce an error code, so its result should be tested, using FSE_isError();
+You can then encode your input data, byte after byte.
+FSE_encodeSymbol() outputs a maximum of 'tableLog' bits at a time.
+Remember decoding will be done in reverse direction.
+    FSE_encodeByte(&bitStream, &state, symbol);
+
+At any time, you can also add any bit sequence.
+Note : maximum allowed nbBits is 25, for compatibility with 32-bits decoders
+    BIT_addBits(&bitStream, bitField, nbBits);
+
+The above methods don't commit data to memory, they just store it into local register, for speed.
+Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
+Writing data to memory is a manual operation, performed by the flushBits function.
+    BIT_flushBits(&bitStream);
+
+Your last FSE encoding operation shall be to flush your last state value(s).
+    FSE_flushState(&bitStream, &state);
+
+Finally, you must close the bitStream.
+The function returns the size of CStream in bytes.
+If data couldn't fit into dstBuffer, it will return a 0 ( == not compressible)
+If there is an error, it returns an errorCode (which can be tested using FSE_isError()).
+    size_t size = BIT_closeCStream(&bitStream);
+*/
+
+
+/* *****************************************
+*  FSE symbol decompression API
+*******************************************/
+typedef struct {
+    size_t      state;
+    const void* table;   /* precise table may vary, depending on U16 */
+} FSE_DState_t;
+
+
+static void     FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, const FSE_DTable* dt);
+
+static unsigned char FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD);
+
+static unsigned FSE_endOfDState(const FSE_DState_t* DStatePtr);
+
+/*<
+Let's now decompose FSE_decompress_usingDTable() into its unitary components.
+You will decode FSE-encoded symbols from the bitStream,
+and also any other bitFields you put in, **in reverse order**.
+
+You will need a few variables to track your bitStream. They are :
+
+BIT_DStream_t DStream;    // Stream context
+FSE_DState_t  DState;     // State context. Multiple ones are possible
+FSE_DTable*   DTablePtr;  // Decoding table, provided by FSE_buildDTable()
+
+The first thing to do is to init the bitStream.
+    errorCode = BIT_initDStream(&DStream, srcBuffer, srcSize);
+
+You should then retrieve your initial state(s)
+(in reverse flushing order if you have several ones) :
+    errorCode = FSE_initDState(&DState, &DStream, DTablePtr);
+
+You can then decode your data, symbol after symbol.
+For information the maximum number of bits read by FSE_decodeSymbol() is 'tableLog'.
+Keep in mind that symbols are decoded in reverse order, like a LIFO stack (last in, first out).
+    unsigned char symbol = FSE_decodeSymbol(&DState, &DStream);
+
+You can retrieve any bitfield you eventually stored into the bitStream (in reverse order)
+Note : maximum allowed nbBits is 25, for 32-bits compatibility
+    size_t bitField = BIT_readBits(&DStream, nbBits);
+
+All above operations only read from local register (which size depends on size_t).
+Refueling the register from memory is manually performed by the reload method.
+    endSignal = FSE_reloadDStream(&DStream);
+
+BIT_reloadDStream() result tells if there is still some more data to read from DStream.
+BIT_DStream_unfinished : there is still some data left into the DStream.
+BIT_DStream_endOfBuffer : Dstream reached end of buffer. Its container may no longer be completely filled.
+BIT_DStream_completed : Dstream reached its exact end, corresponding in general to decompression completed.
+BIT_DStream_tooFar : Dstream went too far. Decompression result is corrupted.
+
+When reaching end of buffer (BIT_DStream_endOfBuffer), progress slowly, notably if you decode multiple symbols per loop,
+to properly detect the exact end of stream.
+After each decoded symbol, check if DStream is fully consumed using this simple test :
+    BIT_reloadDStream(&DStream) >= BIT_DStream_completed
+
+When it's done, verify decompression is fully completed, by checking both DStream and the relevant states.
+Checking if DStream has reached its end is performed by :
+    BIT_endOfDStream(&DStream);
+Check also the states. There might be some symbols left there, if some high probability ones (>50%) are possible.
+    FSE_endOfDState(&DState);
+*/
+
+
+/* *****************************************
+*  FSE unsafe API
+*******************************************/
+static unsigned char FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD);
+/* faster, but works only if nbBits is always >= 1 (otherwise, result will be corrupted) */
+
+
+/* *****************************************
+*  Implementation of inlined functions
+*******************************************/
+typedef struct {
+    int deltaFindState;
+    U32 deltaNbBits;
+} FSE_symbolCompressionTransform; /* total 8 bytes */
+
+MEM_STATIC void FSE_initCState(FSE_CState_t* statePtr, const FSE_CTable* ct)
+{
+    const void* ptr = ct;
+    const U16* u16ptr = (const U16*) ptr;
+    const U32 tableLog = MEM_read16(ptr);
+    statePtr->value = (ptrdiff_t)1<<tableLog;
+    statePtr->stateTable = u16ptr+2;
+    statePtr->symbolTT = ct + 1 + (tableLog ? (1<<(tableLog-1)) : 1);
+    statePtr->stateLog = tableLog;
+}
+
+
+/*! FSE_initCState2() :
+*   Same as FSE_initCState(), but the first symbol to include (which will be the last to be read)
+*   uses the smallest state value possible, saving the cost of this symbol */
+MEM_STATIC void FSE_initCState2(FSE_CState_t* statePtr, const FSE_CTable* ct, U32 symbol)
+{
+    FSE_initCState(statePtr, ct);
+    {   const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol];
+        const U16* stateTable = (const U16*)(statePtr->stateTable);
+        U32 nbBitsOut  = (U32)((symbolTT.deltaNbBits + (1<<15)) >> 16);
+        statePtr->value = (nbBitsOut << 16) - symbolTT.deltaNbBits;
+        statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
+    }
+}
+
+MEM_STATIC void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* statePtr, unsigned symbol)
+{
+    FSE_symbolCompressionTransform const symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol];
+    const U16* const stateTable = (const U16*)(statePtr->stateTable);
+    U32 const nbBitsOut  = (U32)((statePtr->value + symbolTT.deltaNbBits) >> 16);
+    BIT_addBits(bitC, statePtr->value, nbBitsOut);
+    statePtr->value = stateTable[ (statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
+}
+
+MEM_STATIC void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* statePtr)
+{
+    BIT_addBits(bitC, statePtr->value, statePtr->stateLog);
+    BIT_flushBits(bitC);
+}
+
+
+/* FSE_getMaxNbBits() :
+ * Approximate maximum cost of a symbol, in bits.
+ * Fractional get rounded up (i.e : a symbol with a normalized frequency of 3 gives the same result as a frequency of 2)
+ * note 1 : assume symbolValue is valid (<= maxSymbolValue)
+ * note 2 : if freq[symbolValue]==0, @return a fake cost of tableLog+1 bits */
+MEM_STATIC U32 FSE_getMaxNbBits(const void* symbolTTPtr, U32 symbolValue)
+{
+    const FSE_symbolCompressionTransform* symbolTT = (const FSE_symbolCompressionTransform*) symbolTTPtr;
+    return (symbolTT[symbolValue].deltaNbBits + ((1<<16)-1)) >> 16;
+}
+
+/* FSE_bitCost() :
+ * Approximate symbol cost, as fractional value, using fixed-point format (accuracyLog fractional bits)
+ * note 1 : assume symbolValue is valid (<= maxSymbolValue)
+ * note 2 : if freq[symbolValue]==0, @return a fake cost of tableLog+1 bits */
+MEM_STATIC U32 FSE_bitCost(const void* symbolTTPtr, U32 tableLog, U32 symbolValue, U32 accuracyLog)
+{
+    const FSE_symbolCompressionTransform* symbolTT = (const FSE_symbolCompressionTransform*) symbolTTPtr;
+    U32 const minNbBits = symbolTT[symbolValue].deltaNbBits >> 16;
+    U32 const threshold = (minNbBits+1) << 16;
+    assert(tableLog < 16);
+    assert(accuracyLog < 31-tableLog);  /* ensure enough room for renormalization double shift */
+    {   U32 const tableSize = 1 << tableLog;
+        U32 const deltaFromThreshold = threshold - (symbolTT[symbolValue].deltaNbBits + tableSize);
+        U32 const normalizedDeltaFromThreshold = (deltaFromThreshold << accuracyLog) >> tableLog;   /* linear interpolation (very approximate) */
+        U32 const bitMultiplier = 1 << accuracyLog;
+        assert(symbolTT[symbolValue].deltaNbBits + tableSize <= threshold);
+        assert(normalizedDeltaFromThreshold <= bitMultiplier);
+        return (minNbBits+1)*bitMultiplier - normalizedDeltaFromThreshold;
+    }
+}
+
+
+/* ======    Decompression    ====== */
+
+typedef struct {
+    U16 tableLog;
+    U16 fastMode;
+} FSE_DTableHeader;   /* sizeof U32 */
+
+typedef struct
+{
+    unsigned short newState;
+    unsigned char  symbol;
+    unsigned char  nbBits;
+} FSE_decode_t;   /* size == U32 */
+
+MEM_STATIC void FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, const FSE_DTable* dt)
+{
+    const void* ptr = dt;
+    const FSE_DTableHeader* const DTableH = (const FSE_DTableHeader*)ptr;
+    DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
+    BIT_reloadDStream(bitD);
+    DStatePtr->table = dt + 1;
+}
+
+MEM_STATIC BYTE FSE_peekSymbol(const FSE_DState_t* DStatePtr)
+{
+    FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
+    return DInfo.symbol;
+}
+
+MEM_STATIC void FSE_updateState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
+{
+    FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
+    U32 const nbBits = DInfo.nbBits;
+    size_t const lowBits = BIT_readBits(bitD, nbBits);
+    DStatePtr->state = DInfo.newState + lowBits;
+}
+
+MEM_STATIC BYTE FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
+{
+    FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
+    U32 const nbBits = DInfo.nbBits;
+    BYTE const symbol = DInfo.symbol;
+    size_t const lowBits = BIT_readBits(bitD, nbBits);
+
+    DStatePtr->state = DInfo.newState + lowBits;
+    return symbol;
+}
+
+/*! FSE_decodeSymbolFast() :
+    unsafe, only works if no symbol has a probability > 50% */
+MEM_STATIC BYTE FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
+{
+    FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
+    U32 const nbBits = DInfo.nbBits;
+    BYTE const symbol = DInfo.symbol;
+    size_t const lowBits = BIT_readBitsFast(bitD, nbBits);
+
+    DStatePtr->state = DInfo.newState + lowBits;
+    return symbol;
+}
+
+MEM_STATIC unsigned FSE_endOfDState(const FSE_DState_t* DStatePtr)
+{
+    return DStatePtr->state == 0;
+}
+
+
+
+#ifndef FSE_COMMONDEFS_ONLY
+
+/* **************************************************************
+*  Tuning parameters
+****************************************************************/
+/*!MEMORY_USAGE :
+*  Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB; etc.)
+*  Increasing memory usage improves compression ratio
+*  Reduced memory usage can improve speed, due to cache effect
+*  Recommended max value is 14, for 16KB, which nicely fits into Intel x86 L1 cache */
+#ifndef FSE_MAX_MEMORY_USAGE
+#  define FSE_MAX_MEMORY_USAGE 14
+#endif
+#ifndef FSE_DEFAULT_MEMORY_USAGE
+#  define FSE_DEFAULT_MEMORY_USAGE 13
+#endif
+#if (FSE_DEFAULT_MEMORY_USAGE > FSE_MAX_MEMORY_USAGE)
+#  error "FSE_DEFAULT_MEMORY_USAGE must be <= FSE_MAX_MEMORY_USAGE"
+#endif
+
+/*!FSE_MAX_SYMBOL_VALUE :
+*  Maximum symbol value authorized.
+*  Required for proper stack allocation */
+#ifndef FSE_MAX_SYMBOL_VALUE
+#  define FSE_MAX_SYMBOL_VALUE 255
+#endif
+
+/* **************************************************************
+*  template functions type & suffix
+****************************************************************/
+#define FSE_FUNCTION_TYPE BYTE
+#define FSE_FUNCTION_EXTENSION
+#define FSE_DECODE_TYPE FSE_decode_t
+
+
+#endif   /* !FSE_COMMONDEFS_ONLY */
+
+
+/* ***************************************************************
+*  Constants
+*****************************************************************/
+#define FSE_MAX_TABLELOG  (FSE_MAX_MEMORY_USAGE-2)
+#define FSE_MAX_TABLESIZE (1U<<FSE_MAX_TABLELOG)
+#define FSE_MAXTABLESIZE_MASK (FSE_MAX_TABLESIZE-1)
+#define FSE_DEFAULT_TABLELOG (FSE_DEFAULT_MEMORY_USAGE-2)
+#define FSE_MIN_TABLELOG 5
+
+#define FSE_TABLELOG_ABSOLUTE_MAX 15
+#if FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX
+#  error "FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX is not supported"
+#endif
+
+#define FSE_TABLESTEP(tableSize) (((tableSize)>>1) + ((tableSize)>>3) + 3)
+
+
+#endif /* FSE_STATIC_LINKING_ONLY */
+
+
diff --git a/lib/zstd/common/fse_decompress.c b/lib/zstd/common/fse_decompress.c
index a84300e5a013..2c8bbe3e4c14 100644
--- a/lib/zstd/common/fse_decompress.c
+++ b/lib/zstd/common/fse_decompress.c
@@ -1,72 +1,37 @@
-/*
+/* ******************************************************************
  * FSE : Finite State Entropy decoder
- * Copyright (C) 2013-2015, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *  You can contact the author at :
+ *  - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *  - Public forum : https://groups.google.com/forum/#!forum/lz4c
  *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
 
-/* **************************************************************
-*  Compiler specifics
-****************************************************************/
-#define FORCE_INLINE static __always_inline
 
 /* **************************************************************
 *  Includes
 ****************************************************************/
+#include "debug.h"      /* assert */
 #include "bitstream.h"
+#include "compiler.h"
+#define FSE_STATIC_LINKING_ONLY
 #include "fse.h"
-#include <linux/compiler.h>
-#include <linux/kernel.h>
-#include <linux/string.h> /* memcpy, memset */
+#include "error_private.h"
+#define ZSTD_DEPS_NEED_MALLOC
+#include "zstd_deps.h"
+
 
 /* **************************************************************
 *  Error Management
 ****************************************************************/
 #define FSE_isError ERR_isError
-#define FSE_STATIC_ASSERT(c)                                   \
-	{                                                      \
-		enum { FSE_static_assert = 1 / (int)(!!(c)) }; \
-	} /* use only *after* variable declarations */
-
-/* check and forward error code */
-#define CHECK_F(f)                  \
-	{                           \
-		size_t const e = f; \
-		if (FSE_isError(e)) \
-			return e;   \
-	}
+#define FSE_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c)   /* use only *after* variable declarations */
+
 
 /* **************************************************************
 *  Templates
@@ -79,254 +44,347 @@
 
 /* safety checks */
 #ifndef FSE_FUNCTION_EXTENSION
-#error "FSE_FUNCTION_EXTENSION must be defined"
+#  error "FSE_FUNCTION_EXTENSION must be defined"
 #endif
 #ifndef FSE_FUNCTION_TYPE
-#error "FSE_FUNCTION_TYPE must be defined"
+#  error "FSE_FUNCTION_TYPE must be defined"
 #endif
 
 /* Function names */
-#define FSE_CAT(X, Y) X##Y
-#define FSE_FUNCTION_NAME(X, Y) FSE_CAT(X, Y)
-#define FSE_TYPE_NAME(X, Y) FSE_CAT(X, Y)
+#define FSE_CAT(X,Y) X##Y
+#define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y)
+#define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y)
+
 
 /* Function templates */
+FSE_DTable* FSE_createDTable (unsigned tableLog)
+{
+    if (tableLog > FSE_TABLELOG_ABSOLUTE_MAX) tableLog = FSE_TABLELOG_ABSOLUTE_MAX;
+    return (FSE_DTable*)ZSTD_malloc( FSE_DTABLE_SIZE_U32(tableLog) * sizeof (U32) );
+}
 
-size_t FSE_buildDTable_wksp(FSE_DTable *dt, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workspace, size_t workspaceSize)
+void FSE_freeDTable (FSE_DTable* dt)
 {
-	void *const tdPtr = dt + 1; /* because *dt is unsigned, 32-bits aligned on 32-bits */
-	FSE_DECODE_TYPE *const tableDecode = (FSE_DECODE_TYPE *)(tdPtr);
-	U16 *symbolNext = (U16 *)workspace;
-
-	U32 const maxSV1 = maxSymbolValue + 1;
-	U32 const tableSize = 1 << tableLog;
-	U32 highThreshold = tableSize - 1;
-
-	/* Sanity Checks */
-	if (workspaceSize < sizeof(U16) * (FSE_MAX_SYMBOL_VALUE + 1))
-		return ERROR(tableLog_tooLarge);
-	if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE)
-		return ERROR(maxSymbolValue_tooLarge);
-	if (tableLog > FSE_MAX_TABLELOG)
-		return ERROR(tableLog_tooLarge);
-
-	/* Init, lay down lowprob symbols */
-	{
-		FSE_DTableHeader DTableH;
-		DTableH.tableLog = (U16)tableLog;
-		DTableH.fastMode = 1;
-		{
-			S16 const largeLimit = (S16)(1 << (tableLog - 1));
-			U32 s;
-			for (s = 0; s < maxSV1; s++) {
-				if (normalizedCounter[s] == -1) {
-					tableDecode[highThreshold--].symbol = (FSE_FUNCTION_TYPE)s;
-					symbolNext[s] = 1;
-				} else {
-					if (normalizedCounter[s] >= largeLimit)
-						DTableH.fastMode = 0;
-					symbolNext[s] = normalizedCounter[s];
-				}
-			}
-		}
-		memcpy(dt, &DTableH, sizeof(DTableH));
-	}
-
-	/* Spread symbols */
-	{
-		U32 const tableMask = tableSize - 1;
-		U32 const step = FSE_TABLESTEP(tableSize);
-		U32 s, position = 0;
-		for (s = 0; s < maxSV1; s++) {
-			int i;
-			for (i = 0; i < normalizedCounter[s]; i++) {
-				tableDecode[position].symbol = (FSE_FUNCTION_TYPE)s;
-				position = (position + step) & tableMask;
-				while (position > highThreshold)
-					position = (position + step) & tableMask; /* lowprob area */
-			}
-		}
-		if (position != 0)
-			return ERROR(GENERIC); /* position must reach all cells once, otherwise normalizedCounter is incorrect */
-	}
-
-	/* Build Decoding table */
-	{
-		U32 u;
-		for (u = 0; u < tableSize; u++) {
-			FSE_FUNCTION_TYPE const symbol = (FSE_FUNCTION_TYPE)(tableDecode[u].symbol);
-			U16 nextState = symbolNext[symbol]++;
-			tableDecode[u].nbBits = (BYTE)(tableLog - BIT_highbit32((U32)nextState));
-			tableDecode[u].newState = (U16)((nextState << tableDecode[u].nbBits) - tableSize);
-		}
-	}
-
-	return 0;
+    ZSTD_free(dt);
 }
 
+static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
+{
+    void* const tdPtr = dt+1;   /* because *dt is unsigned, 32-bits aligned on 32-bits */
+    FSE_DECODE_TYPE* const tableDecode = (FSE_DECODE_TYPE*) (tdPtr);
+    U16* symbolNext = (U16*)workSpace;
+    BYTE* spread = (BYTE*)(symbolNext + maxSymbolValue + 1);
+
+    U32 const maxSV1 = maxSymbolValue + 1;
+    U32 const tableSize = 1 << tableLog;
+    U32 highThreshold = tableSize-1;
+
+    /* Sanity Checks */
+    if (FSE_BUILD_DTABLE_WKSP_SIZE(tableLog, maxSymbolValue) > wkspSize) return ERROR(maxSymbolValue_tooLarge);
+    if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE) return ERROR(maxSymbolValue_tooLarge);
+    if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge);
+
+    /* Init, lay down lowprob symbols */
+    {   FSE_DTableHeader DTableH;
+        DTableH.tableLog = (U16)tableLog;
+        DTableH.fastMode = 1;
+        {   S16 const largeLimit= (S16)(1 << (tableLog-1));
+            U32 s;
+            for (s=0; s<maxSV1; s++) {
+                if (normalizedCounter[s]==-1) {
+                    tableDecode[highThreshold--].symbol = (FSE_FUNCTION_TYPE)s;
+                    symbolNext[s] = 1;
+                } else {
+                    if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
+                    symbolNext[s] = normalizedCounter[s];
+        }   }   }
+        ZSTD_memcpy(dt, &DTableH, sizeof(DTableH));
+    }
+
+    /* Spread symbols */
+    if (highThreshold == tableSize - 1) {
+        size_t const tableMask = tableSize-1;
+        size_t const step = FSE_TABLESTEP(tableSize);
+        /* First lay down the symbols in order.
+         * We use a uint64_t to lay down 8 bytes at a time. This reduces branch
+         * misses since small blocks generally have small table logs, so nearly
+         * all symbols have counts <= 8. We ensure we have 8 bytes at the end of
+         * our buffer to handle the over-write.
+         */
+        {
+            U64 const add = 0x0101010101010101ull;
+            size_t pos = 0;
+            U64 sv = 0;
+            U32 s;
+            for (s=0; s<maxSV1; ++s, sv += add) {
+                int i;
+                int const n = normalizedCounter[s];
+                MEM_write64(spread + pos, sv);
+                for (i = 8; i < n; i += 8) {
+                    MEM_write64(spread + pos + i, sv);
+                }
+                pos += n;
+            }
+        }
+        /* Now we spread those positions across the table.
+         * The benefit of doing it in two stages is that we avoid the the
+         * variable size inner loop, which caused lots of branch misses.
+         * Now we can run through all the positions without any branch misses.
+         * We unroll the loop twice, since that is what emperically worked best.
+         */
+        {
+            size_t position = 0;
+            size_t s;
+            size_t const unroll = 2;
+            assert(tableSize % unroll == 0); /* FSE_MIN_TABLELOG is 5 */
+            for (s = 0; s < (size_t)tableSize; s += unroll) {
+                size_t u;
+                for (u = 0; u < unroll; ++u) {
+                    size_t const uPosition = (position + (u * step)) & tableMask;
+                    tableDecode[uPosition].symbol = spread[s + u];
+                }
+                position = (position + (unroll * step)) & tableMask;
+            }
+            assert(position == 0);
+        }
+    } else {
+        U32 const tableMask = tableSize-1;
+        U32 const step = FSE_TABLESTEP(tableSize);
+        U32 s, position = 0;
+        for (s=0; s<maxSV1; s++) {
+            int i;
+            for (i=0; i<normalizedCounter[s]; i++) {
+                tableDecode[position].symbol = (FSE_FUNCTION_TYPE)s;
+                position = (position + step) & tableMask;
+                while (position > highThreshold) position = (position + step) & tableMask;   /* lowprob area */
+        }   }
+        if (position!=0) return ERROR(GENERIC);   /* position must reach all cells once, otherwise normalizedCounter is incorrect */
+    }
+
+    /* Build Decoding table */
+    {   U32 u;
+        for (u=0; u<tableSize; u++) {
+            FSE_FUNCTION_TYPE const symbol = (FSE_FUNCTION_TYPE)(tableDecode[u].symbol);
+            U32 const nextState = symbolNext[symbol]++;
+            tableDecode[u].nbBits = (BYTE) (tableLog - BIT_highbit32(nextState) );
+            tableDecode[u].newState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
+    }   }
+
+    return 0;
+}
+
+size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
+{
+    return FSE_buildDTable_internal(dt, normalizedCounter, maxSymbolValue, tableLog, workSpace, wkspSize);
+}
+
+
+#ifndef FSE_COMMONDEFS_ONLY
+
 /*-*******************************************************
 *  Decompression (Byte symbols)
 *********************************************************/
-size_t FSE_buildDTable_rle(FSE_DTable *dt, BYTE symbolValue)
+size_t FSE_buildDTable_rle (FSE_DTable* dt, BYTE symbolValue)
 {
-	void *ptr = dt;
-	FSE_DTableHeader *const DTableH = (FSE_DTableHeader *)ptr;
-	void *dPtr = dt + 1;
-	FSE_decode_t *const cell = (FSE_decode_t *)dPtr;
+    void* ptr = dt;
+    FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
+    void* dPtr = dt + 1;
+    FSE_decode_t* const cell = (FSE_decode_t*)dPtr;
 
-	DTableH->tableLog = 0;
-	DTableH->fastMode = 0;
+    DTableH->tableLog = 0;
+    DTableH->fastMode = 0;
 
-	cell->newState = 0;
-	cell->symbol = symbolValue;
-	cell->nbBits = 0;
+    cell->newState = 0;
+    cell->symbol = symbolValue;
+    cell->nbBits = 0;
 
-	return 0;
+    return 0;
 }
 
-size_t FSE_buildDTable_raw(FSE_DTable *dt, unsigned nbBits)
+
+size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits)
 {
-	void *ptr = dt;
-	FSE_DTableHeader *const DTableH = (FSE_DTableHeader *)ptr;
-	void *dPtr = dt + 1;
-	FSE_decode_t *const dinfo = (FSE_decode_t *)dPtr;
-	const unsigned tableSize = 1 << nbBits;
-	const unsigned tableMask = tableSize - 1;
-	const unsigned maxSV1 = tableMask + 1;
-	unsigned s;
-
-	/* Sanity checks */
-	if (nbBits < 1)
-		return ERROR(GENERIC); /* min size */
-
-	/* Build Decoding Table */
-	DTableH->tableLog = (U16)nbBits;
-	DTableH->fastMode = 1;
-	for (s = 0; s < maxSV1; s++) {
-		dinfo[s].newState = 0;
-		dinfo[s].symbol = (BYTE)s;
-		dinfo[s].nbBits = (BYTE)nbBits;
-	}
-
-	return 0;
+    void* ptr = dt;
+    FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
+    void* dPtr = dt + 1;
+    FSE_decode_t* const dinfo = (FSE_decode_t*)dPtr;
+    const unsigned tableSize = 1 << nbBits;
+    const unsigned tableMask = tableSize - 1;
+    const unsigned maxSV1 = tableMask+1;
+    unsigned s;
+
+    /* Sanity checks */
+    if (nbBits < 1) return ERROR(GENERIC);         /* min size */
+
+    /* Build Decoding Table */
+    DTableH->tableLog = (U16)nbBits;
+    DTableH->fastMode = 1;
+    for (s=0; s<maxSV1; s++) {
+        dinfo[s].newState = 0;
+        dinfo[s].symbol = (BYTE)s;
+        dinfo[s].nbBits = (BYTE)nbBits;
+    }
+
+    return 0;
 }
 
-FORCE_INLINE size_t FSE_decompress_usingDTable_generic(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const FSE_DTable *dt,
-						       const unsigned fast)
+FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
+          void* dst, size_t maxDstSize,
+    const void* cSrc, size_t cSrcSize,
+    const FSE_DTable* dt, const unsigned fast)
 {
-	BYTE *const ostart = (BYTE *)dst;
-	BYTE *op = ostart;
-	BYTE *const omax = op + maxDstSize;
-	BYTE *const olimit = omax - 3;
+    BYTE* const ostart = (BYTE*) dst;
+    BYTE* op = ostart;
+    BYTE* const omax = op + maxDstSize;
+    BYTE* const olimit = omax-3;
 
-	BIT_DStream_t bitD;
-	FSE_DState_t state1;
-	FSE_DState_t state2;
+    BIT_DStream_t bitD;
+    FSE_DState_t state1;
+    FSE_DState_t state2;
 
-	/* Init */
-	CHECK_F(BIT_initDStream(&bitD, cSrc, cSrcSize));
+    /* Init */
+    CHECK_F(BIT_initDStream(&bitD, cSrc, cSrcSize));
 
-	FSE_initDState(&state1, &bitD, dt);
-	FSE_initDState(&state2, &bitD, dt);
+    FSE_initDState(&state1, &bitD, dt);
+    FSE_initDState(&state2, &bitD, dt);
 
 #define FSE_GETSYMBOL(statePtr) fast ? FSE_decodeSymbolFast(statePtr, &bitD) : FSE_decodeSymbol(statePtr, &bitD)
 
-	/* 4 symbols per loop */
-	for (; (BIT_reloadDStream(&bitD) == BIT_DStream_unfinished) & (op < olimit); op += 4) {
-		op[0] = FSE_GETSYMBOL(&state1);
-
-		if (FSE_MAX_TABLELOG * 2 + 7 > sizeof(bitD.bitContainer) * 8) /* This test must be static */
-			BIT_reloadDStream(&bitD);
-
-		op[1] = FSE_GETSYMBOL(&state2);
-
-		if (FSE_MAX_TABLELOG * 4 + 7 > sizeof(bitD.bitContainer) * 8) /* This test must be static */
-		{
-			if (BIT_reloadDStream(&bitD) > BIT_DStream_unfinished) {
-				op += 2;
-				break;
-			}
-		}
-
-		op[2] = FSE_GETSYMBOL(&state1);
-
-		if (FSE_MAX_TABLELOG * 2 + 7 > sizeof(bitD.bitContainer) * 8) /* This test must be static */
-			BIT_reloadDStream(&bitD);
-
-		op[3] = FSE_GETSYMBOL(&state2);
-	}
-
-	/* tail */
-	/* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */
-	while (1) {
-		if (op > (omax - 2))
-			return ERROR(dstSize_tooSmall);
-		*op++ = FSE_GETSYMBOL(&state1);
-		if (BIT_reloadDStream(&bitD) == BIT_DStream_overflow) {
-			*op++ = FSE_GETSYMBOL(&state2);
-			break;
-		}
-
-		if (op > (omax - 2))
-			return ERROR(dstSize_tooSmall);
-		*op++ = FSE_GETSYMBOL(&state2);
-		if (BIT_reloadDStream(&bitD) == BIT_DStream_overflow) {
-			*op++ = FSE_GETSYMBOL(&state1);
-			break;
-		}
-	}
-
-	return op - ostart;
+    /* 4 symbols per loop */
+    for ( ; (BIT_reloadDStream(&bitD)==BIT_DStream_unfinished) & (op<olimit) ; op+=4) {
+        op[0] = FSE_GETSYMBOL(&state1);
+
+        if (FSE_MAX_TABLELOG*2+7 > sizeof(bitD.bitContainer)*8)    /* This test must be static */
+            BIT_reloadDStream(&bitD);
+
+        op[1] = FSE_GETSYMBOL(&state2);
+
+        if (FSE_MAX_TABLELOG*4+7 > sizeof(bitD.bitContainer)*8)    /* This test must be static */
+            { if (BIT_reloadDStream(&bitD) > BIT_DStream_unfinished) { op+=2; break; } }
+
+        op[2] = FSE_GETSYMBOL(&state1);
+
+        if (FSE_MAX_TABLELOG*2+7 > sizeof(bitD.bitContainer)*8)    /* This test must be static */
+            BIT_reloadDStream(&bitD);
+
+        op[3] = FSE_GETSYMBOL(&state2);
+    }
+
+    /* tail */
+    /* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */
+    while (1) {
+        if (op>(omax-2)) return ERROR(dstSize_tooSmall);
+        *op++ = FSE_GETSYMBOL(&state1);
+        if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
+            *op++ = FSE_GETSYMBOL(&state2);
+            break;
+        }
+
+        if (op>(omax-2)) return ERROR(dstSize_tooSmall);
+        *op++ = FSE_GETSYMBOL(&state2);
+        if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
+            *op++ = FSE_GETSYMBOL(&state1);
+            break;
+    }   }
+
+    return op-ostart;
+}
+
+
+size_t FSE_decompress_usingDTable(void* dst, size_t originalSize,
+                            const void* cSrc, size_t cSrcSize,
+                            const FSE_DTable* dt)
+{
+    const void* ptr = dt;
+    const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
+    const U32 fastMode = DTableH->fastMode;
+
+    /* select fast mode (static) */
+    if (fastMode) return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 1);
+    return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 0);
 }
 
-size_t FSE_decompress_usingDTable(void *dst, size_t originalSize, const void *cSrc, size_t cSrcSize, const FSE_DTable *dt)
+
+size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
+{
+    return FSE_decompress_wksp_bmi2(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, /* bmi2 */ 0);
+}
+
+typedef struct {
+    short ncount[FSE_MAX_SYMBOL_VALUE + 1];
+    FSE_DTable dtable[1]; /* Dynamically sized */
+} FSE_DecompressWksp;
+
+
+FORCE_INLINE_TEMPLATE size_t FSE_decompress_wksp_body(
+        void* dst, size_t dstCapacity,
+        const void* cSrc, size_t cSrcSize,
+        unsigned maxLog, void* workSpace, size_t wkspSize,
+        int bmi2)
 {
-	const void *ptr = dt;
-	const FSE_DTableHeader *DTableH = (const FSE_DTableHeader *)ptr;
-	const U32 fastMode = DTableH->fastMode;
-
-	/* select fast mode (static) */
-	if (fastMode)
-		return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 1);
-	return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 0);
+    const BYTE* const istart = (const BYTE*)cSrc;
+    const BYTE* ip = istart;
+    unsigned tableLog;
+    unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
+    FSE_DecompressWksp* const wksp = (FSE_DecompressWksp*)workSpace;
+
+    DEBUG_STATIC_ASSERT((FSE_MAX_SYMBOL_VALUE + 1) % 2 == 0);
+    if (wkspSize < sizeof(*wksp)) return ERROR(GENERIC);
+
+    /* normal FSE decoding mode */
+    {
+        size_t const NCountLength = FSE_readNCount_bmi2(wksp->ncount, &maxSymbolValue, &tableLog, istart, cSrcSize, bmi2);
+        if (FSE_isError(NCountLength)) return NCountLength;
+        if (tableLog > maxLog) return ERROR(tableLog_tooLarge);
+        assert(NCountLength <= cSrcSize);
+        ip += NCountLength;
+        cSrcSize -= NCountLength;
+    }
+
+    if (FSE_DECOMPRESS_WKSP_SIZE(tableLog, maxSymbolValue) > wkspSize) return ERROR(tableLog_tooLarge);
+    workSpace = wksp->dtable + FSE_DTABLE_SIZE_U32(tableLog);
+    wkspSize -= sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog);
+
+    CHECK_F( FSE_buildDTable_internal(wksp->dtable, wksp->ncount, maxSymbolValue, tableLog, workSpace, wkspSize) );
+
+    {
+        const void* ptr = wksp->dtable;
+        const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
+        const U32 fastMode = DTableH->fastMode;
+
+        /* select fast mode (static) */
+        if (fastMode) return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, wksp->dtable, 1);
+        return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, wksp->dtable, 0);
+    }
 }
 
-size_t FSE_decompress_wksp(void *dst, size_t dstCapacity, const void *cSrc, size_t cSrcSize, unsigned maxLog, void *workspace, size_t workspaceSize)
+/* Avoids the FORCE_INLINE of the _body() function. */
+static size_t FSE_decompress_wksp_body_default(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
 {
-	const BYTE *const istart = (const BYTE *)cSrc;
-	const BYTE *ip = istart;
-	unsigned tableLog;
-	unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
-	size_t NCountLength;
-
-	FSE_DTable *dt;
-	short *counting;
-	size_t spaceUsed32 = 0;
-
-	FSE_STATIC_ASSERT(sizeof(FSE_DTable) == sizeof(U32));
-
-	dt = (FSE_DTable *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += FSE_DTABLE_SIZE_U32(maxLog);
-	counting = (short *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(sizeof(short) * (FSE_MAX_SYMBOL_VALUE + 1), sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	/* normal FSE decoding mode */
-	NCountLength = FSE_readNCount(counting, &maxSymbolValue, &tableLog, istart, cSrcSize);
-	if (FSE_isError(NCountLength))
-		return NCountLength;
-	// if (NCountLength >= cSrcSize) return ERROR(srcSize_wrong);   /* too small input size; supposed to be already checked in NCountLength, only remaining
-	// case : NCountLength==cSrcSize */
-	if (tableLog > maxLog)
-		return ERROR(tableLog_tooLarge);
-	ip += NCountLength;
-	cSrcSize -= NCountLength;
-
-	CHECK_F(FSE_buildDTable_wksp(dt, counting, maxSymbolValue, tableLog, workspace, workspaceSize));
-
-	return FSE_decompress_usingDTable(dst, dstCapacity, ip, cSrcSize, dt); /* always return, even if it is an error code */
+    return FSE_decompress_wksp_body(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, 0);
 }
+
+#if DYNAMIC_BMI2
+TARGET_ATTRIBUTE("bmi2") static size_t FSE_decompress_wksp_body_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
+{
+    return FSE_decompress_wksp_body(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, 1);
+}
+#endif
+
+size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize, int bmi2)
+{
+#if DYNAMIC_BMI2
+    if (bmi2) {
+        return FSE_decompress_wksp_body_bmi2(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
+    }
+#endif
+    (void)bmi2;
+    return FSE_decompress_wksp_body_default(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
+}
+
+
+typedef FSE_DTable DTable_max_t[FSE_DTABLE_SIZE_U32(FSE_MAX_TABLELOG)];
+
+
+
+#endif   /* FSE_COMMONDEFS_ONLY */
diff --git a/lib/zstd/common/huf.h b/lib/zstd/common/huf.h
new file mode 100644
index 000000000000..88c5586646aa
--- /dev/null
+++ b/lib/zstd/common/huf.h
@@ -0,0 +1,356 @@
+/* ******************************************************************
+ * huff0 huffman codec,
+ * part of Finite State Entropy library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+
+#ifndef HUF_H_298734234
+#define HUF_H_298734234
+
+/* *** Dependencies *** */
+#include "zstd_deps.h"    /* size_t */
+
+
+/* *** library symbols visibility *** */
+/* Note : when linking with -fvisibility=hidden on gcc, or by default on Visual,
+ *        HUF symbols remain "private" (internal symbols for library only).
+ *        Set macro FSE_DLL_EXPORT to 1 if you want HUF symbols visible on DLL interface */
+#if defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) && defined(__GNUC__) && (__GNUC__ >= 4)
+#  define HUF_PUBLIC_API __attribute__ ((visibility ("default")))
+#elif defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1)   /* Visual expected */
+#  define HUF_PUBLIC_API __declspec(dllexport)
+#elif defined(FSE_DLL_IMPORT) && (FSE_DLL_IMPORT==1)
+#  define HUF_PUBLIC_API __declspec(dllimport)  /* not required, just to generate faster code (saves a function pointer load from IAT and an indirect jump) */
+#else
+#  define HUF_PUBLIC_API
+#endif
+
+
+/* ========================== */
+/* ***  simple functions  *** */
+/* ========================== */
+
+/* HUF_compress() :
+ *  Compress content from buffer 'src', of size 'srcSize', into buffer 'dst'.
+ * 'dst' buffer must be already allocated.
+ *  Compression runs faster if `dstCapacity` >= HUF_compressBound(srcSize).
+ * `srcSize` must be <= `HUF_BLOCKSIZE_MAX` == 128 KB.
+ * @return : size of compressed data (<= `dstCapacity`).
+ *  Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
+ *                   if HUF_isError(return), compression failed (more details using HUF_getErrorName())
+ */
+HUF_PUBLIC_API size_t HUF_compress(void* dst, size_t dstCapacity,
+                             const void* src, size_t srcSize);
+
+/* HUF_decompress() :
+ *  Decompress HUF data from buffer 'cSrc', of size 'cSrcSize',
+ *  into already allocated buffer 'dst', of minimum size 'dstSize'.
+ * `originalSize` : **must** be the ***exact*** size of original (uncompressed) data.
+ *  Note : in contrast with FSE, HUF_decompress can regenerate
+ *         RLE (cSrcSize==1) and uncompressed (cSrcSize==dstSize) data,
+ *         because it knows size to regenerate (originalSize).
+ * @return : size of regenerated data (== originalSize),
+ *           or an error code, which can be tested using HUF_isError()
+ */
+HUF_PUBLIC_API size_t HUF_decompress(void* dst,  size_t originalSize,
+                               const void* cSrc, size_t cSrcSize);
+
+
+/* ***   Tool functions *** */
+#define HUF_BLOCKSIZE_MAX (128 * 1024)                  /*< maximum input size for a single block compressed with HUF_compress */
+HUF_PUBLIC_API size_t HUF_compressBound(size_t size);   /*< maximum compressed size (worst case) */
+
+/* Error Management */
+HUF_PUBLIC_API unsigned    HUF_isError(size_t code);       /*< tells if a return value is an error code */
+HUF_PUBLIC_API const char* HUF_getErrorName(size_t code);  /*< provides error code string (useful for debugging) */
+
+
+/* ***   Advanced function   *** */
+
+/* HUF_compress2() :
+ *  Same as HUF_compress(), but offers control over `maxSymbolValue` and `tableLog`.
+ * `maxSymbolValue` must be <= HUF_SYMBOLVALUE_MAX .
+ * `tableLog` must be `<= HUF_TABLELOG_MAX` . */
+HUF_PUBLIC_API size_t HUF_compress2 (void* dst, size_t dstCapacity,
+                               const void* src, size_t srcSize,
+                               unsigned maxSymbolValue, unsigned tableLog);
+
+/* HUF_compress4X_wksp() :
+ *  Same as HUF_compress2(), but uses externally allocated `workSpace`.
+ * `workspace` must have minimum alignment of 4, and be at least as large as HUF_WORKSPACE_SIZE */
+#define HUF_WORKSPACE_SIZE ((6 << 10) + 256)
+#define HUF_WORKSPACE_SIZE_U32 (HUF_WORKSPACE_SIZE / sizeof(U32))
+HUF_PUBLIC_API size_t HUF_compress4X_wksp (void* dst, size_t dstCapacity,
+                                     const void* src, size_t srcSize,
+                                     unsigned maxSymbolValue, unsigned tableLog,
+                                     void* workSpace, size_t wkspSize);
+
+#endif   /* HUF_H_298734234 */
+
+/* ******************************************************************
+ *  WARNING !!
+ *  The following section contains advanced and experimental definitions
+ *  which shall never be used in the context of a dynamic library,
+ *  because they are not guaranteed to remain stable in the future.
+ *  Only consider them in association with static linking.
+ * *****************************************************************/
+#if !defined(HUF_H_HUF_STATIC_LINKING_ONLY)
+#define HUF_H_HUF_STATIC_LINKING_ONLY
+
+/* *** Dependencies *** */
+#include "mem.h"   /* U32 */
+#define FSE_STATIC_LINKING_ONLY
+#include "fse.h"
+
+
+/* *** Constants *** */
+#define HUF_TABLELOG_MAX      12      /* max runtime value of tableLog (due to static allocation); can be modified up to HUF_ABSOLUTEMAX_TABLELOG */
+#define HUF_TABLELOG_DEFAULT  11      /* default tableLog value when none specified */
+#define HUF_SYMBOLVALUE_MAX  255
+
+#define HUF_TABLELOG_ABSOLUTEMAX  15  /* absolute limit of HUF_MAX_TABLELOG. Beyond that value, code does not work */
+#if (HUF_TABLELOG_MAX > HUF_TABLELOG_ABSOLUTEMAX)
+#  error "HUF_TABLELOG_MAX is too large !"
+#endif
+
+
+/* ****************************************
+*  Static allocation
+******************************************/
+/* HUF buffer bounds */
+#define HUF_CTABLEBOUND 129
+#define HUF_BLOCKBOUND(size) (size + (size>>8) + 8)   /* only true when incompressible is pre-filtered with fast heuristic */
+#define HUF_COMPRESSBOUND(size) (HUF_CTABLEBOUND + HUF_BLOCKBOUND(size))   /* Macro version, useful for static allocation */
+
+/* static allocation of HUF's Compression Table */
+/* this is a private definition, just exposed for allocation and strict aliasing purpose. never EVER access its members directly */
+struct HUF_CElt_s {
+  U16  val;
+  BYTE nbBits;
+};   /* typedef'd to HUF_CElt */
+typedef struct HUF_CElt_s HUF_CElt;   /* consider it an incomplete type */
+#define HUF_CTABLE_SIZE_U32(maxSymbolValue)   ((maxSymbolValue)+1)   /* Use tables of U32, for proper alignment */
+#define HUF_CTABLE_SIZE(maxSymbolValue)       (HUF_CTABLE_SIZE_U32(maxSymbolValue) * sizeof(U32))
+#define HUF_CREATE_STATIC_CTABLE(name, maxSymbolValue) \
+    HUF_CElt name[HUF_CTABLE_SIZE_U32(maxSymbolValue)] /* no final ; */
+
+/* static allocation of HUF's DTable */
+typedef U32 HUF_DTable;
+#define HUF_DTABLE_SIZE(maxTableLog)   (1 + (1<<(maxTableLog)))
+#define HUF_CREATE_STATIC_DTABLEX1(DTable, maxTableLog) \
+        HUF_DTable DTable[HUF_DTABLE_SIZE((maxTableLog)-1)] = { ((U32)((maxTableLog)-1) * 0x01000001) }
+#define HUF_CREATE_STATIC_DTABLEX2(DTable, maxTableLog) \
+        HUF_DTable DTable[HUF_DTABLE_SIZE(maxTableLog)] = { ((U32)(maxTableLog) * 0x01000001) }
+
+
+/* ****************************************
+*  Advanced decompression functions
+******************************************/
+size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< single-symbol decoder */
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< double-symbols decoder */
+#endif
+
+size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< decodes RLE and uncompressed */
+size_t HUF_decompress4X_hufOnly(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< considers RLE and uncompressed as errors */
+size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /*< considers RLE and uncompressed as errors */
+size_t HUF_decompress4X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< single-symbol decoder */
+size_t HUF_decompress4X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /*< single-symbol decoder */
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress4X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< double-symbols decoder */
+size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /*< double-symbols decoder */
+#endif
+
+
+/* ****************************************
+ *  HUF detailed API
+ * ****************************************/
+
+/*! HUF_compress() does the following:
+ *  1. count symbol occurrence from source[] into table count[] using FSE_count() (exposed within "fse.h")
+ *  2. (optional) refine tableLog using HUF_optimalTableLog()
+ *  3. build Huffman table from count using HUF_buildCTable()
+ *  4. save Huffman table to memory buffer using HUF_writeCTable()
+ *  5. encode the data stream using HUF_compress4X_usingCTable()
+ *
+ *  The following API allows targeting specific sub-functions for advanced tasks.
+ *  For example, it's possible to compress several blocks using the same 'CTable',
+ *  or to save and regenerate 'CTable' using external methods.
+ */
+unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
+size_t HUF_buildCTable (HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue, unsigned maxNbBits);   /* @return : maxNbBits; CTable and count can overlap. In which case, CTable will overwrite count content */
+size_t HUF_writeCTable (void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog);
+size_t HUF_writeCTable_wksp(void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog, void* workspace, size_t workspaceSize);
+size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
+size_t HUF_estimateCompressedSize(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
+int HUF_validateCTable(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
+
+typedef enum {
+   HUF_repeat_none,  /*< Cannot use the previous table */
+   HUF_repeat_check, /*< Can use the previous table but it must be checked. Note : The previous table must have been constructed by HUF_compress{1, 4}X_repeat */
+   HUF_repeat_valid  /*< Can use the previous table and it is assumed to be valid */
+ } HUF_repeat;
+/* HUF_compress4X_repeat() :
+ *  Same as HUF_compress4X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
+ *  If it uses hufTable it does not modify hufTable or repeat.
+ *  If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
+ *  If preferRepeat then the old table will always be used if valid. */
+size_t HUF_compress4X_repeat(void* dst, size_t dstSize,
+                       const void* src, size_t srcSize,
+                       unsigned maxSymbolValue, unsigned tableLog,
+                       void* workSpace, size_t wkspSize,    /*< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
+                       HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);
+
+/* HUF_buildCTable_wksp() :
+ *  Same as HUF_buildCTable(), but using externally allocated scratch buffer.
+ * `workSpace` must be aligned on 4-bytes boundaries, and its size must be >= HUF_CTABLE_WORKSPACE_SIZE.
+ */
+#define HUF_CTABLE_WORKSPACE_SIZE_U32 (2*HUF_SYMBOLVALUE_MAX +1 +1)
+#define HUF_CTABLE_WORKSPACE_SIZE (HUF_CTABLE_WORKSPACE_SIZE_U32 * sizeof(unsigned))
+size_t HUF_buildCTable_wksp (HUF_CElt* tree,
+                       const unsigned* count, U32 maxSymbolValue, U32 maxNbBits,
+                             void* workSpace, size_t wkspSize);
+
+/*! HUF_readStats() :
+ *  Read compact Huffman tree, saved by HUF_writeCTable().
+ * `huffWeight` is destination buffer.
+ * @return : size read from `src` , or an error Code .
+ *  Note : Needed by HUF_readCTable() and HUF_readDTableXn() . */
+size_t HUF_readStats(BYTE* huffWeight, size_t hwSize,
+                     U32* rankStats, U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize);
+
+/*! HUF_readStats_wksp() :
+ * Same as HUF_readStats() but takes an external workspace which must be
+ * 4-byte aligned and its size must be >= HUF_READ_STATS_WORKSPACE_SIZE.
+ * If the CPU has BMI2 support, pass bmi2=1, otherwise pass bmi2=0.
+ */
+#define HUF_READ_STATS_WORKSPACE_SIZE_U32 FSE_DECOMPRESS_WKSP_SIZE_U32(6, HUF_TABLELOG_MAX-1)
+#define HUF_READ_STATS_WORKSPACE_SIZE (HUF_READ_STATS_WORKSPACE_SIZE_U32 * sizeof(unsigned))
+size_t HUF_readStats_wksp(BYTE* huffWeight, size_t hwSize,
+                          U32* rankStats, U32* nbSymbolsPtr, U32* tableLogPtr,
+                          const void* src, size_t srcSize,
+                          void* workspace, size_t wkspSize,
+                          int bmi2);
+
+/* HUF_readCTable() :
+ *  Loading a CTable saved with HUF_writeCTable() */
+size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize, unsigned *hasZeroWeights);
+
+/* HUF_getNbBits() :
+ *  Read nbBits from CTable symbolTable, for symbol `symbolValue` presumed <= HUF_SYMBOLVALUE_MAX
+ *  Note 1 : is not inlined, as HUF_CElt definition is private
+ *  Note 2 : const void* used, so that it can provide a statically allocated table as argument (which uses type U32) */
+U32 HUF_getNbBits(const void* symbolTable, U32 symbolValue);
+
+/*
+ * HUF_decompress() does the following:
+ * 1. select the decompression algorithm (X1, X2) based on pre-computed heuristics
+ * 2. build Huffman table from save, using HUF_readDTableX?()
+ * 3. decode 1 or 4 segments in parallel using HUF_decompress?X?_usingDTable()
+ */
+
+/* HUF_selectDecoder() :
+ *  Tells which decoder is likely to decode faster,
+ *  based on a set of pre-computed metrics.
+ * @return : 0==HUF_decompress4X1, 1==HUF_decompress4X2 .
+ *  Assumption : 0 < dstSize <= 128 KB */
+U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize);
+
+/*
+ *  The minimum workspace size for the `workSpace` used in
+ *  HUF_readDTableX1_wksp() and HUF_readDTableX2_wksp().
+ *
+ *  The space used depends on HUF_TABLELOG_MAX, ranging from ~1500 bytes when
+ *  HUF_TABLE_LOG_MAX=12 to ~1850 bytes when HUF_TABLE_LOG_MAX=15.
+ *  Buffer overflow errors may potentially occur if code modifications result in
+ *  a required workspace size greater than that specified in the following
+ *  macro.
+ */
+#define HUF_DECOMPRESS_WORKSPACE_SIZE ((2 << 10) + (1 << 9))
+#define HUF_DECOMPRESS_WORKSPACE_SIZE_U32 (HUF_DECOMPRESS_WORKSPACE_SIZE / sizeof(U32))
+
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_readDTableX1 (HUF_DTable* DTable, const void* src, size_t srcSize);
+size_t HUF_readDTableX1_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_readDTableX2 (HUF_DTable* DTable, const void* src, size_t srcSize);
+size_t HUF_readDTableX2_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
+#endif
+
+size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress4X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress4X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+
+
+/* ====================== */
+/* single stream variants */
+/* ====================== */
+
+size_t HUF_compress1X (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
+size_t HUF_compress1X_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);  /*< `workSpace` must be a table of at least HUF_WORKSPACE_SIZE_U32 unsigned */
+size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
+/* HUF_compress1X_repeat() :
+ *  Same as HUF_compress1X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
+ *  If it uses hufTable it does not modify hufTable or repeat.
+ *  If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
+ *  If preferRepeat then the old table will always be used if valid. */
+size_t HUF_compress1X_repeat(void* dst, size_t dstSize,
+                       const void* src, size_t srcSize,
+                       unsigned maxSymbolValue, unsigned tableLog,
+                       void* workSpace, size_t wkspSize,   /*< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
+                       HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);
+
+size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /* single-symbol decoder */
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /* double-symbol decoder */
+#endif
+
+size_t HUF_decompress1X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);
+size_t HUF_decompress1X_DCtx_wksp (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress1X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< single-symbol decoder */
+size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /*< single-symbol decoder */
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress1X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< double-symbols decoder */
+size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /*< double-symbols decoder */
+#endif
+
+size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);   /*< automatic selection of sing or double symbol decoder, based on DTable */
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress1X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress1X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+
+/* BMI2 variants.
+ * If the CPU has BMI2 support, pass bmi2=1, otherwise pass bmi2=0.
+ */
+size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);
+#endif
+size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
+size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_readDTableX1_wksp_bmi2(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int bmi2);
+#endif
+
+#endif /* HUF_STATIC_LINKING_ONLY */
+
diff --git a/lib/zstd/common/mem.h b/lib/zstd/common/mem.h
new file mode 100644
index 000000000000..dcdd586a9fd9
--- /dev/null
+++ b/lib/zstd/common/mem.h
@@ -0,0 +1,259 @@
+/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef MEM_H_MODULE
+#define MEM_H_MODULE
+
+/*-****************************************
+*  Dependencies
+******************************************/
+#include <asm/unaligned.h>  /* get_unaligned, put_unaligned* */
+#include <linux/compiler.h>  /* inline */
+#include <linux/swab.h>  /* swab32, swab64 */
+#include <linux/types.h>  /* size_t, ptrdiff_t */
+#include "debug.h"  /* DEBUG_STATIC_ASSERT */
+
+/*-****************************************
+*  Compiler specifics
+******************************************/
+#define MEM_STATIC static inline
+
+/*-**************************************************************
+*  Basic Types
+*****************************************************************/
+typedef uint8_t  BYTE;
+typedef uint16_t U16;
+typedef int16_t  S16;
+typedef uint32_t U32;
+typedef int32_t  S32;
+typedef uint64_t U64;
+typedef int64_t  S64;
+
+/*-**************************************************************
+*  Memory I/O API
+*****************************************************************/
+/*=== Static platform detection ===*/
+MEM_STATIC unsigned MEM_32bits(void);
+MEM_STATIC unsigned MEM_64bits(void);
+MEM_STATIC unsigned MEM_isLittleEndian(void);
+
+/*=== Native unaligned read/write ===*/
+MEM_STATIC U16 MEM_read16(const void* memPtr);
+MEM_STATIC U32 MEM_read32(const void* memPtr);
+MEM_STATIC U64 MEM_read64(const void* memPtr);
+MEM_STATIC size_t MEM_readST(const void* memPtr);
+
+MEM_STATIC void MEM_write16(void* memPtr, U16 value);
+MEM_STATIC void MEM_write32(void* memPtr, U32 value);
+MEM_STATIC void MEM_write64(void* memPtr, U64 value);
+
+/*=== Little endian unaligned read/write ===*/
+MEM_STATIC U16 MEM_readLE16(const void* memPtr);
+MEM_STATIC U32 MEM_readLE24(const void* memPtr);
+MEM_STATIC U32 MEM_readLE32(const void* memPtr);
+MEM_STATIC U64 MEM_readLE64(const void* memPtr);
+MEM_STATIC size_t MEM_readLEST(const void* memPtr);
+
+MEM_STATIC void MEM_writeLE16(void* memPtr, U16 val);
+MEM_STATIC void MEM_writeLE24(void* memPtr, U32 val);
+MEM_STATIC void MEM_writeLE32(void* memPtr, U32 val32);
+MEM_STATIC void MEM_writeLE64(void* memPtr, U64 val64);
+MEM_STATIC void MEM_writeLEST(void* memPtr, size_t val);
+
+/*=== Big endian unaligned read/write ===*/
+MEM_STATIC U32 MEM_readBE32(const void* memPtr);
+MEM_STATIC U64 MEM_readBE64(const void* memPtr);
+MEM_STATIC size_t MEM_readBEST(const void* memPtr);
+
+MEM_STATIC void MEM_writeBE32(void* memPtr, U32 val32);
+MEM_STATIC void MEM_writeBE64(void* memPtr, U64 val64);
+MEM_STATIC void MEM_writeBEST(void* memPtr, size_t val);
+
+/*=== Byteswap ===*/
+MEM_STATIC U32 MEM_swap32(U32 in);
+MEM_STATIC U64 MEM_swap64(U64 in);
+MEM_STATIC size_t MEM_swapST(size_t in);
+
+/*-**************************************************************
+*  Memory I/O Implementation
+*****************************************************************/
+MEM_STATIC unsigned MEM_32bits(void)
+{
+    return sizeof(size_t) == 4;
+}
+
+MEM_STATIC unsigned MEM_64bits(void)
+{
+    return sizeof(size_t) == 8;
+}
+
+#if defined(__LITTLE_ENDIAN)
+#define MEM_LITTLE_ENDIAN 1
+#else
+#define MEM_LITTLE_ENDIAN 0
+#endif
+
+MEM_STATIC unsigned MEM_isLittleEndian(void)
+{
+    return MEM_LITTLE_ENDIAN;
+}
+
+MEM_STATIC U16 MEM_read16(const void *memPtr)
+{
+    return get_unaligned((const U16 *)memPtr);
+}
+
+MEM_STATIC U32 MEM_read32(const void *memPtr)
+{
+    return get_unaligned((const U32 *)memPtr);
+}
+
+MEM_STATIC U64 MEM_read64(const void *memPtr)
+{
+    return get_unaligned((const U64 *)memPtr);
+}
+
+MEM_STATIC size_t MEM_readST(const void *memPtr)
+{
+    return get_unaligned((const size_t *)memPtr);
+}
+
+MEM_STATIC void MEM_write16(void *memPtr, U16 value)
+{
+    put_unaligned(value, (U16 *)memPtr);
+}
+
+MEM_STATIC void MEM_write32(void *memPtr, U32 value)
+{
+    put_unaligned(value, (U32 *)memPtr);
+}
+
+MEM_STATIC void MEM_write64(void *memPtr, U64 value)
+{
+    put_unaligned(value, (U64 *)memPtr);
+}
+
+/*=== Little endian r/w ===*/
+
+MEM_STATIC U16 MEM_readLE16(const void *memPtr)
+{
+    return get_unaligned_le16(memPtr);
+}
+
+MEM_STATIC void MEM_writeLE16(void *memPtr, U16 val)
+{
+    put_unaligned_le16(val, memPtr);
+}
+
+MEM_STATIC U32 MEM_readLE24(const void *memPtr)
+{
+    return MEM_readLE16(memPtr) + (((const BYTE *)memPtr)[2] << 16);
+}
+
+MEM_STATIC void MEM_writeLE24(void *memPtr, U32 val)
+{
+	MEM_writeLE16(memPtr, (U16)val);
+	((BYTE *)memPtr)[2] = (BYTE)(val >> 16);
+}
+
+MEM_STATIC U32 MEM_readLE32(const void *memPtr)
+{
+    return get_unaligned_le32(memPtr);
+}
+
+MEM_STATIC void MEM_writeLE32(void *memPtr, U32 val32)
+{
+    put_unaligned_le32(val32, memPtr);
+}
+
+MEM_STATIC U64 MEM_readLE64(const void *memPtr)
+{
+    return get_unaligned_le64(memPtr);
+}
+
+MEM_STATIC void MEM_writeLE64(void *memPtr, U64 val64)
+{
+    put_unaligned_le64(val64, memPtr);
+}
+
+MEM_STATIC size_t MEM_readLEST(const void *memPtr)
+{
+	if (MEM_32bits())
+		return (size_t)MEM_readLE32(memPtr);
+	else
+		return (size_t)MEM_readLE64(memPtr);
+}
+
+MEM_STATIC void MEM_writeLEST(void *memPtr, size_t val)
+{
+	if (MEM_32bits())
+		MEM_writeLE32(memPtr, (U32)val);
+	else
+		MEM_writeLE64(memPtr, (U64)val);
+}
+
+/*=== Big endian r/w ===*/
+
+MEM_STATIC U32 MEM_readBE32(const void *memPtr)
+{
+    return get_unaligned_be32(memPtr);
+}
+
+MEM_STATIC void MEM_writeBE32(void *memPtr, U32 val32)
+{
+    put_unaligned_be32(val32, memPtr);
+}
+
+MEM_STATIC U64 MEM_readBE64(const void *memPtr)
+{
+    return get_unaligned_be64(memPtr);
+}
+
+MEM_STATIC void MEM_writeBE64(void *memPtr, U64 val64)
+{
+    put_unaligned_be64(val64, memPtr);
+}
+
+MEM_STATIC size_t MEM_readBEST(const void *memPtr)
+{
+	if (MEM_32bits())
+		return (size_t)MEM_readBE32(memPtr);
+	else
+		return (size_t)MEM_readBE64(memPtr);
+}
+
+MEM_STATIC void MEM_writeBEST(void *memPtr, size_t val)
+{
+	if (MEM_32bits())
+		MEM_writeBE32(memPtr, (U32)val);
+	else
+		MEM_writeBE64(memPtr, (U64)val);
+}
+
+MEM_STATIC U32 MEM_swap32(U32 in)
+{
+    return swab32(in);
+}
+
+MEM_STATIC U64 MEM_swap64(U64 in)
+{
+    return swab64(in);
+}
+
+MEM_STATIC size_t MEM_swapST(size_t in)
+{
+    if (MEM_32bits())
+        return (size_t)MEM_swap32((U32)in);
+    else
+        return (size_t)MEM_swap64((U64)in);
+}
+
+#endif /* MEM_H_MODULE */
diff --git a/lib/zstd/common/zstd_common.c b/lib/zstd/common/zstd_common.c
index a282624ee155..3d7e35b309b5 100644
--- a/lib/zstd/common/zstd_common.c
+++ b/lib/zstd/common/zstd_common.c
@@ -1,75 +1,83 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
  */
 
+
+
 /*-*************************************
 *  Dependencies
 ***************************************/
+#define ZSTD_DEPS_NEED_MALLOC
+#include "zstd_deps.h"   /* ZSTD_malloc, ZSTD_calloc, ZSTD_free, ZSTD_memset */
 #include "error_private.h"
-#include "zstd_internal.h" /* declaration of ZSTD_isError, ZSTD_getErrorName, ZSTD_getErrorCode, ZSTD_getErrorString, ZSTD_versionNumber */
-#include <linux/kernel.h>
+#include "zstd_internal.h"
 
-/*=**************************************************************
-*  Custom allocator
-****************************************************************/
 
-#define stack_push(stack, size)                                 \
-	({                                                      \
-		void *const ptr = ZSTD_PTR_ALIGN((stack)->ptr); \
-		(stack)->ptr = (char *)ptr + (size);            \
-		(stack)->ptr <= (stack)->end ? ptr : NULL;      \
-	})
+/*-****************************************
+*  Version
+******************************************/
+unsigned ZSTD_versionNumber(void) { return ZSTD_VERSION_NUMBER; }
 
-ZSTD_customMem ZSTD_initStack(void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem stackMem = {ZSTD_stackAlloc, ZSTD_stackFree, workspace};
-	ZSTD_stack *stack = (ZSTD_stack *)workspace;
-	/* Verify preconditions */
-	if (!workspace || workspaceSize < sizeof(ZSTD_stack) || workspace != ZSTD_PTR_ALIGN(workspace)) {
-		ZSTD_customMem error = {NULL, NULL, NULL};
-		return error;
-	}
-	/* Initialize the stack */
-	stack->ptr = workspace;
-	stack->end = (char *)workspace + workspaceSize;
-	stack_push(stack, sizeof(ZSTD_stack));
-	return stackMem;
-}
+const char* ZSTD_versionString(void) { return ZSTD_VERSION_STRING; }
 
-void *ZSTD_stackAllocAll(void *opaque, size_t *size)
-{
-	ZSTD_stack *stack = (ZSTD_stack *)opaque;
-	*size = (BYTE const *)stack->end - (BYTE *)ZSTD_PTR_ALIGN(stack->ptr);
-	return stack_push(stack, *size);
-}
 
-void *ZSTD_stackAlloc(void *opaque, size_t size)
+/*-****************************************
+*  ZSTD Error Management
+******************************************/
+#undef ZSTD_isError   /* defined within zstd_internal.h */
+/*! ZSTD_isError() :
+ *  tells if a return value is an error code
+ *  symbol is required for external callers */
+unsigned ZSTD_isError(size_t code) { return ERR_isError(code); }
+
+/*! ZSTD_getErrorName() :
+ *  provides error code string from function result (useful for debugging) */
+const char* ZSTD_getErrorName(size_t code) { return ERR_getErrorName(code); }
+
+/*! ZSTD_getError() :
+ *  convert a `size_t` function result into a proper ZSTD_errorCode enum */
+ZSTD_ErrorCode ZSTD_getErrorCode(size_t code) { return ERR_getErrorCode(code); }
+
+/*! ZSTD_getErrorString() :
+ *  provides error code string from enum */
+const char* ZSTD_getErrorString(ZSTD_ErrorCode code) { return ERR_getErrorString(code); }
+
+
+
+/*=**************************************************************
+*  Custom allocator
+****************************************************************/
+void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem)
 {
-	ZSTD_stack *stack = (ZSTD_stack *)opaque;
-	return stack_push(stack, size);
+    if (customMem.customAlloc)
+        return customMem.customAlloc(customMem.opaque, size);
+    return ZSTD_malloc(size);
 }
-void ZSTD_stackFree(void *opaque, void *address)
+
+void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem)
 {
-	(void)opaque;
-	(void)address;
+    if (customMem.customAlloc) {
+        /* calloc implemented as malloc+memset;
+         * not as efficient as calloc, but next best guess for custom malloc */
+        void* const ptr = customMem.customAlloc(customMem.opaque, size);
+        ZSTD_memset(ptr, 0, size);
+        return ptr;
+    }
+    return ZSTD_calloc(1, size);
 }
 
-void *ZSTD_malloc(size_t size, ZSTD_customMem customMem) { return customMem.customAlloc(customMem.opaque, size); }
-
-void ZSTD_free(void *ptr, ZSTD_customMem customMem)
+void ZSTD_customFree(void* ptr, ZSTD_customMem customMem)
 {
-	if (ptr != NULL)
-		customMem.customFree(customMem.opaque, ptr);
+    if (ptr!=NULL) {
+        if (customMem.customFree)
+            customMem.customFree(customMem.opaque, ptr);
+        else
+            ZSTD_free(ptr);
+    }
 }
diff --git a/lib/zstd/common/zstd_deps.h b/lib/zstd/common/zstd_deps.h
new file mode 100644
index 000000000000..7a5bf44839c9
--- /dev/null
+++ b/lib/zstd/common/zstd_deps.h
@@ -0,0 +1,125 @@
+/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */
+/*
+ * Copyright (c) Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/*
+ * This file provides common libc dependencies that zstd requires.
+ * The purpose is to allow replacing this file with a custom implementation
+ * to compile zstd without libc support.
+ */
+
+/* Need:
+ * NULL
+ * INT_MAX
+ * UINT_MAX
+ * ZSTD_memcpy()
+ * ZSTD_memset()
+ * ZSTD_memmove()
+ */
+#ifndef ZSTD_DEPS_COMMON
+#define ZSTD_DEPS_COMMON
+
+#include <linux/limits.h>
+#include <linux/stddef.h>
+
+#define ZSTD_memcpy(d,s,n) __builtin_memcpy((d),(s),(n))
+#define ZSTD_memmove(d,s,n) __builtin_memmove((d),(s),(n))
+#define ZSTD_memset(d,s,n) __builtin_memset((d),(s),(n))
+
+#endif /* ZSTD_DEPS_COMMON */
+
+/*
+ * Define malloc as always failing. That means the user must
+ * either use ZSTD_customMem or statically allocate memory.
+ * Need:
+ * ZSTD_malloc()
+ * ZSTD_free()
+ * ZSTD_calloc()
+ */
+#ifdef ZSTD_DEPS_NEED_MALLOC
+#ifndef ZSTD_DEPS_MALLOC
+#define ZSTD_DEPS_MALLOC
+
+#define ZSTD_malloc(s) ({ (void)(s); NULL; })
+#define ZSTD_free(p) ((void)(p))
+#define ZSTD_calloc(n,s) ({ (void)(n); (void)(s); NULL; })
+
+#endif /* ZSTD_DEPS_MALLOC */
+#endif /* ZSTD_DEPS_NEED_MALLOC */
+
+/*
+ * Provides 64-bit math support.
+ * Need:
+ * U64 ZSTD_div64(U64 dividend, U32 divisor)
+ */
+#ifdef ZSTD_DEPS_NEED_MATH64
+#ifndef ZSTD_DEPS_MATH64
+#define ZSTD_DEPS_MATH64
+
+#include <linux/math64.h>
+
+static uint64_t ZSTD_div64(uint64_t dividend, uint32_t divisor) {
+  return div_u64(dividend, divisor);
+}
+
+#endif /* ZSTD_DEPS_MATH64 */
+#endif /* ZSTD_DEPS_NEED_MATH64 */
+
+/*
+ * This is only requested when DEBUGLEVEL >= 1, meaning
+ * it is disabled in production.
+ * Need:
+ * assert()
+ */
+#ifdef ZSTD_DEPS_NEED_ASSERT
+#ifndef ZSTD_DEPS_ASSERT
+#define ZSTD_DEPS_ASSERT
+
+#include <linux/kernel.h>
+
+#define assert(x) WARN_ON((x))
+
+#endif /* ZSTD_DEPS_ASSERT */
+#endif /* ZSTD_DEPS_NEED_ASSERT */
+
+/*
+ * This is only requested when DEBUGLEVEL >= 2, meaning
+ * it is disabled in production.
+ * Need:
+ * ZSTD_DEBUG_PRINT()
+ */
+#ifdef ZSTD_DEPS_NEED_IO
+#ifndef ZSTD_DEPS_IO
+#define ZSTD_DEPS_IO
+
+#include <linux/printk.h>
+
+#define ZSTD_DEBUG_PRINT(...) pr_debug(__VA_ARGS__)
+
+#endif /* ZSTD_DEPS_IO */
+#endif /* ZSTD_DEPS_NEED_IO */
+
+/*
+ * Only requested when MSAN is enabled.
+ * Need:
+ * intptr_t
+ */
+#ifdef ZSTD_DEPS_NEED_STDINT
+#ifndef ZSTD_DEPS_STDINT
+#define ZSTD_DEPS_STDINT
+
+/*
+ * The Linux Kernel doesn't provide intptr_t, only uintptr_t, which
+ * is an unsigned long.
+ */
+typedef long intptr_t;
+
+#endif /* ZSTD_DEPS_STDINT */
+#endif /* ZSTD_DEPS_NEED_STDINT */
diff --git a/lib/zstd/common/zstd_internal.h b/lib/zstd/common/zstd_internal.h
new file mode 100644
index 000000000000..fc6f3a9b40c0
--- /dev/null
+++ b/lib/zstd/common/zstd_internal.h
@@ -0,0 +1,450 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_CCOMMON_H_MODULE
+#define ZSTD_CCOMMON_H_MODULE
+
+/* this module contains definitions which must be identical
+ * across compression, decompression and dictBuilder.
+ * It also contains a few functions useful to at least 2 of them
+ * and which benefit from being inlined */
+
+/*-*************************************
+*  Dependencies
+***************************************/
+#include "compiler.h"
+#include "mem.h"
+#include "debug.h"                 /* assert, DEBUGLOG, RAWLOG, g_debuglevel */
+#include "error_private.h"
+#define ZSTD_STATIC_LINKING_ONLY
+#include <linux/zstd.h>
+#define FSE_STATIC_LINKING_ONLY
+#include "fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "huf.h"
+#include <linux/xxhash.h>                /* XXH_reset, update, digest */
+#define ZSTD_TRACE 0
+
+
+/* ---- static assert (debug) --- */
+#define ZSTD_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c)
+#define ZSTD_isError ERR_isError   /* for inlining */
+#define FSE_isError  ERR_isError
+#define HUF_isError  ERR_isError
+
+
+/*-*************************************
+*  shared macros
+***************************************/
+#undef MIN
+#undef MAX
+#define MIN(a,b) ((a)<(b) ? (a) : (b))
+#define MAX(a,b) ((a)>(b) ? (a) : (b))
+
+/*
+ * Ignore: this is an internal helper.
+ *
+ * This is a helper function to help force C99-correctness during compilation.
+ * Under strict compilation modes, variadic macro arguments can't be empty.
+ * However, variadic function arguments can be. Using a function therefore lets
+ * us statically check that at least one (string) argument was passed,
+ * independent of the compilation flags.
+ */
+static INLINE_KEYWORD UNUSED_ATTR
+void _force_has_format_string(const char *format, ...) {
+  (void)format;
+}
+
+/*
+ * Ignore: this is an internal helper.
+ *
+ * We want to force this function invocation to be syntactically correct, but
+ * we don't want to force runtime evaluation of its arguments.
+ */
+#define _FORCE_HAS_FORMAT_STRING(...) \
+  if (0) { \
+    _force_has_format_string(__VA_ARGS__); \
+  }
+
+/*
+ * Return the specified error if the condition evaluates to true.
+ *
+ * In debug modes, prints additional information.
+ * In order to do that (particularly, printing the conditional that failed),
+ * this can't just wrap RETURN_ERROR().
+ */
+#define RETURN_ERROR_IF(cond, err, ...) \
+  if (cond) { \
+    RAWLOG(3, "%s:%d: ERROR!: check %s failed, returning %s", \
+           __FILE__, __LINE__, ZSTD_QUOTE(cond), ZSTD_QUOTE(ERROR(err))); \
+    _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
+    RAWLOG(3, ": " __VA_ARGS__); \
+    RAWLOG(3, "\n"); \
+    return ERROR(err); \
+  }
+
+/*
+ * Unconditionally return the specified error.
+ *
+ * In debug modes, prints additional information.
+ */
+#define RETURN_ERROR(err, ...) \
+  do { \
+    RAWLOG(3, "%s:%d: ERROR!: unconditional check failed, returning %s", \
+           __FILE__, __LINE__, ZSTD_QUOTE(ERROR(err))); \
+    _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
+    RAWLOG(3, ": " __VA_ARGS__); \
+    RAWLOG(3, "\n"); \
+    return ERROR(err); \
+  } while(0);
+
+/*
+ * If the provided expression evaluates to an error code, returns that error code.
+ *
+ * In debug modes, prints additional information.
+ */
+#define FORWARD_IF_ERROR(err, ...) \
+  do { \
+    size_t const err_code = (err); \
+    if (ERR_isError(err_code)) { \
+      RAWLOG(3, "%s:%d: ERROR!: forwarding error in %s: %s", \
+             __FILE__, __LINE__, ZSTD_QUOTE(err), ERR_getErrorName(err_code)); \
+      _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
+      RAWLOG(3, ": " __VA_ARGS__); \
+      RAWLOG(3, "\n"); \
+      return err_code; \
+    } \
+  } while(0);
+
+
+/*-*************************************
+*  Common constants
+***************************************/
+#define ZSTD_OPT_NUM    (1<<12)
+
+#define ZSTD_REP_NUM      3                 /* number of repcodes */
+#define ZSTD_REP_MOVE     (ZSTD_REP_NUM-1)
+static UNUSED_ATTR const U32 repStartValue[ZSTD_REP_NUM] = { 1, 4, 8 };
+
+#define KB *(1 <<10)
+#define MB *(1 <<20)
+#define GB *(1U<<30)
+
+#define BIT7 128
+#define BIT6  64
+#define BIT5  32
+#define BIT4  16
+#define BIT1   2
+#define BIT0   1
+
+#define ZSTD_WINDOWLOG_ABSOLUTEMIN 10
+static UNUSED_ATTR const size_t ZSTD_fcs_fieldSize[4] = { 0, 2, 4, 8 };
+static UNUSED_ATTR const size_t ZSTD_did_fieldSize[4] = { 0, 1, 2, 4 };
+
+#define ZSTD_FRAMEIDSIZE 4   /* magic number size */
+
+#define ZSTD_BLOCKHEADERSIZE 3   /* C standard doesn't allow `static const` variable to be init using another `static const` variable */
+static UNUSED_ATTR const size_t ZSTD_blockHeaderSize = ZSTD_BLOCKHEADERSIZE;
+typedef enum { bt_raw, bt_rle, bt_compressed, bt_reserved } blockType_e;
+
+#define ZSTD_FRAMECHECKSUMSIZE 4
+
+#define MIN_SEQUENCES_SIZE 1 /* nbSeq==0 */
+#define MIN_CBLOCK_SIZE (1 /*litCSize*/ + 1 /* RLE or RAW */ + MIN_SEQUENCES_SIZE /* nbSeq==0 */)   /* for a non-null block */
+
+#define HufLog 12
+typedef enum { set_basic, set_rle, set_compressed, set_repeat } symbolEncodingType_e;
+
+#define LONGNBSEQ 0x7F00
+
+#define MINMATCH 3
+
+#define Litbits  8
+#define MaxLit ((1<<Litbits) - 1)
+#define MaxML   52
+#define MaxLL   35
+#define DefaultMaxOff 28
+#define MaxOff  31
+#define MaxSeq MAX(MaxLL, MaxML)   /* Assumption : MaxOff < MaxLL,MaxML */
+#define MLFSELog    9
+#define LLFSELog    9
+#define OffFSELog   8
+#define MaxFSELog  MAX(MAX(MLFSELog, LLFSELog), OffFSELog)
+
+#define ZSTD_MAX_HUF_HEADER_SIZE 128 /* header + <= 127 byte tree description */
+/* Each table cannot take more than #symbols * FSELog bits */
+#define ZSTD_MAX_FSE_HEADERS_SIZE (((MaxML + 1) * MLFSELog + (MaxLL + 1) * LLFSELog + (MaxOff + 1) * OffFSELog + 7) / 8)
+
+static UNUSED_ATTR const U32 LL_bits[MaxLL+1] = {
+     0, 0, 0, 0, 0, 0, 0, 0,
+     0, 0, 0, 0, 0, 0, 0, 0,
+     1, 1, 1, 1, 2, 2, 3, 3,
+     4, 6, 7, 8, 9,10,11,12,
+    13,14,15,16
+};
+static UNUSED_ATTR const S16 LL_defaultNorm[MaxLL+1] = {
+     4, 3, 2, 2, 2, 2, 2, 2,
+     2, 2, 2, 2, 2, 1, 1, 1,
+     2, 2, 2, 2, 2, 2, 2, 2,
+     2, 3, 2, 1, 1, 1, 1, 1,
+    -1,-1,-1,-1
+};
+#define LL_DEFAULTNORMLOG 6  /* for static allocation */
+static UNUSED_ATTR const U32 LL_defaultNormLog = LL_DEFAULTNORMLOG;
+
+static UNUSED_ATTR const U32 ML_bits[MaxML+1] = {
+     0, 0, 0, 0, 0, 0, 0, 0,
+     0, 0, 0, 0, 0, 0, 0, 0,
+     0, 0, 0, 0, 0, 0, 0, 0,
+     0, 0, 0, 0, 0, 0, 0, 0,
+     1, 1, 1, 1, 2, 2, 3, 3,
+     4, 4, 5, 7, 8, 9,10,11,
+    12,13,14,15,16
+};
+static UNUSED_ATTR const S16 ML_defaultNorm[MaxML+1] = {
+     1, 4, 3, 2, 2, 2, 2, 2,
+     2, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1,-1,-1,
+    -1,-1,-1,-1,-1
+};
+#define ML_DEFAULTNORMLOG 6  /* for static allocation */
+static UNUSED_ATTR const U32 ML_defaultNormLog = ML_DEFAULTNORMLOG;
+
+static UNUSED_ATTR const S16 OF_defaultNorm[DefaultMaxOff+1] = {
+     1, 1, 1, 1, 1, 1, 2, 2,
+     2, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1, 1, 1,
+    -1,-1,-1,-1,-1
+};
+#define OF_DEFAULTNORMLOG 5  /* for static allocation */
+static UNUSED_ATTR const U32 OF_defaultNormLog = OF_DEFAULTNORMLOG;
+
+
+/*-*******************************************
+*  Shared functions to include for inlining
+*********************************************/
+static void ZSTD_copy8(void* dst, const void* src) {
+    ZSTD_memcpy(dst, src, 8);
+}
+
+#define COPY8(d,s) { ZSTD_copy8(d,s); d+=8; s+=8; }
+static void ZSTD_copy16(void* dst, const void* src) {
+    ZSTD_memcpy(dst, src, 16);
+}
+#define COPY16(d,s) { ZSTD_copy16(d,s); d+=16; s+=16; }
+
+#define WILDCOPY_OVERLENGTH 32
+#define WILDCOPY_VECLEN 16
+
+typedef enum {
+    ZSTD_no_overlap,
+    ZSTD_overlap_src_before_dst
+    /*  ZSTD_overlap_dst_before_src, */
+} ZSTD_overlap_e;
+
+/*! ZSTD_wildcopy() :
+ *  Custom version of ZSTD_memcpy(), can over read/write up to WILDCOPY_OVERLENGTH bytes (if length==0)
+ *  @param ovtype controls the overlap detection
+ *         - ZSTD_no_overlap: The source and destination are guaranteed to be at least WILDCOPY_VECLEN bytes apart.
+ *         - ZSTD_overlap_src_before_dst: The src and dst may overlap, but they MUST be at least 8 bytes apart.
+ *           The src buffer must be before the dst buffer.
+ */
+MEM_STATIC FORCE_INLINE_ATTR
+void ZSTD_wildcopy(void* dst, const void* src, ptrdiff_t length, ZSTD_overlap_e const ovtype)
+{
+    ptrdiff_t diff = (BYTE*)dst - (const BYTE*)src;
+    const BYTE* ip = (const BYTE*)src;
+    BYTE* op = (BYTE*)dst;
+    BYTE* const oend = op + length;
+
+    assert(diff >= 8 || (ovtype == ZSTD_no_overlap && diff <= -WILDCOPY_VECLEN));
+
+    if (ovtype == ZSTD_overlap_src_before_dst && diff < WILDCOPY_VECLEN) {
+        /* Handle short offset copies. */
+        do {
+            COPY8(op, ip)
+        } while (op < oend);
+    } else {
+        assert(diff >= WILDCOPY_VECLEN || diff <= -WILDCOPY_VECLEN);
+        /* Separate out the first COPY16() call because the copy length is
+         * almost certain to be short, so the branches have different
+         * probabilities. Since it is almost certain to be short, only do
+         * one COPY16() in the first call. Then, do two calls per loop since
+         * at that point it is more likely to have a high trip count.
+         */
+#ifdef __aarch64__
+        do {
+            COPY16(op, ip);
+        }
+        while (op < oend);
+#else
+        ZSTD_copy16(op, ip);
+        if (16 >= length) return;
+        op += 16;
+        ip += 16;
+        do {
+            COPY16(op, ip);
+            COPY16(op, ip);
+        }
+        while (op < oend);
+#endif
+    }
+}
+
+MEM_STATIC size_t ZSTD_limitCopy(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+    size_t const length = MIN(dstCapacity, srcSize);
+    if (length > 0) {
+        ZSTD_memcpy(dst, src, length);
+    }
+    return length;
+}
+
+/* define "workspace is too large" as this number of times larger than needed */
+#define ZSTD_WORKSPACETOOLARGE_FACTOR 3
+
+/* when workspace is continuously too large
+ * during at least this number of times,
+ * context's memory usage is considered wasteful,
+ * because it's sized to handle a worst case scenario which rarely happens.
+ * In which case, resize it down to free some memory */
+#define ZSTD_WORKSPACETOOLARGE_MAXDURATION 128
+
+/* Controls whether the input/output buffer is buffered or stable. */
+typedef enum {
+    ZSTD_bm_buffered = 0,  /* Buffer the input/output */
+    ZSTD_bm_stable = 1     /* ZSTD_inBuffer/ZSTD_outBuffer is stable */
+} ZSTD_bufferMode_e;
+
+
+/*-*******************************************
+*  Private declarations
+*********************************************/
+typedef struct seqDef_s {
+    U32 offset;         /* Offset code of the sequence */
+    U16 litLength;
+    U16 matchLength;
+} seqDef;
+
+typedef struct {
+    seqDef* sequencesStart;
+    seqDef* sequences;      /* ptr to end of sequences */
+    BYTE* litStart;
+    BYTE* lit;              /* ptr to end of literals */
+    BYTE* llCode;
+    BYTE* mlCode;
+    BYTE* ofCode;
+    size_t maxNbSeq;
+    size_t maxNbLit;
+
+    /* longLengthPos and longLengthID to allow us to represent either a single litLength or matchLength
+     * in the seqStore that has a value larger than U16 (if it exists). To do so, we increment
+     * the existing value of the litLength or matchLength by 0x10000.
+     */
+    U32   longLengthID;   /* 0 == no longLength; 1 == Represent the long literal; 2 == Represent the long match; */
+    U32   longLengthPos;  /* Index of the sequence to apply long length modification to */
+} seqStore_t;
+
+typedef struct {
+    U32 litLength;
+    U32 matchLength;
+} ZSTD_sequenceLength;
+
+/*
+ * Returns the ZSTD_sequenceLength for the given sequences. It handles the decoding of long sequences
+ * indicated by longLengthPos and longLengthID, and adds MINMATCH back to matchLength.
+ */
+MEM_STATIC ZSTD_sequenceLength ZSTD_getSequenceLength(seqStore_t const* seqStore, seqDef const* seq)
+{
+    ZSTD_sequenceLength seqLen;
+    seqLen.litLength = seq->litLength;
+    seqLen.matchLength = seq->matchLength + MINMATCH;
+    if (seqStore->longLengthPos == (U32)(seq - seqStore->sequencesStart)) {
+        if (seqStore->longLengthID == 1) {
+            seqLen.litLength += 0xFFFF;
+        }
+        if (seqStore->longLengthID == 2) {
+            seqLen.matchLength += 0xFFFF;
+        }
+    }
+    return seqLen;
+}
+
+/*
+ * Contains the compressed frame size and an upper-bound for the decompressed frame size.
+ * Note: before using `compressedSize`, check for errors using ZSTD_isError().
+ *       similarly, before using `decompressedBound`, check for errors using:
+ *          `decompressedBound != ZSTD_CONTENTSIZE_ERROR`
+ */
+typedef struct {
+    size_t compressedSize;
+    unsigned long long decompressedBound;
+} ZSTD_frameSizeInfo;   /* decompress & legacy */
+
+const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx);   /* compress & dictBuilder */
+void ZSTD_seqToCodes(const seqStore_t* seqStorePtr);   /* compress, dictBuilder, decodeCorpus (shouldn't get its definition from here) */
+
+/* custom memory allocation functions */
+void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem);
+void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem);
+void ZSTD_customFree(void* ptr, ZSTD_customMem customMem);
+
+
+MEM_STATIC U32 ZSTD_highbit32(U32 val)   /* compress, dictBuilder, decodeCorpus */
+{
+    assert(val != 0);
+    {
+#   if (__GNUC__ >= 3)   /* GCC Intrinsic */
+        return __builtin_clz (val) ^ 31;
+#   else   /* Software version */
+        static const U32 DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 };
+        U32 v = val;
+        v |= v >> 1;
+        v |= v >> 2;
+        v |= v >> 4;
+        v |= v >> 8;
+        v |= v >> 16;
+        return DeBruijnClz[(v * 0x07C4ACDDU) >> 27];
+#   endif
+    }
+}
+
+
+/* ZSTD_invalidateRepCodes() :
+ * ensures next compression will not use repcodes from previous block.
+ * Note : only works with regular variant;
+ *        do not use with extDict variant ! */
+void ZSTD_invalidateRepCodes(ZSTD_CCtx* cctx);   /* zstdmt, adaptive_compression (shouldn't get this definition from here) */
+
+
+typedef struct {
+    blockType_e blockType;
+    U32 lastBlock;
+    U32 origSize;
+} blockProperties_t;   /* declared here for decompress and fullbench */
+
+/*! ZSTD_getcBlockSize() :
+ *  Provides the size of compressed block from block header `src` */
+/* Used by: decompress, fullbench (does not get its definition from here) */
+size_t ZSTD_getcBlockSize(const void* src, size_t srcSize,
+                          blockProperties_t* bpPtr);
+
+/*! ZSTD_decodeSeqHeaders() :
+ *  decode sequence header from src */
+/* Used by: decompress, fullbench (does not get its definition from here) */
+size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr,
+                       const void* src, size_t srcSize);
+
+
+
+#endif   /* ZSTD_CCOMMON_H_MODULE */
diff --git a/lib/zstd/decompress.c b/lib/zstd/decompress.c
deleted file mode 100644
index 19bf71288111..000000000000
--- a/lib/zstd/decompress.c
+++ /dev/null
@@ -1,2528 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-/* ***************************************************************
-*  Tuning parameters
-*****************************************************************/
-/*!
-*  MAXWINDOWSIZE_DEFAULT :
-*  maximum window size accepted by DStream, by default.
-*  Frames requiring more memory will be rejected.
-*/
-#ifndef ZSTD_MAXWINDOWSIZE_DEFAULT
-#define ZSTD_MAXWINDOWSIZE_DEFAULT ((1 << ZSTD_WINDOWLOG_MAX) + 1) /* defined within zstd.h */
-#endif
-
-/*-*******************************************************
-*  Dependencies
-*********************************************************/
-#include "fse.h"
-#include "huf.h"
-#include "mem.h" /* low level memory routines */
-#include "zstd_internal.h"
-#include <linux/kernel.h>
-#include <module.h>
-#include <linux/string.h> /* memcpy, memmove, memset */
-
-#define ZSTD_PREFETCH(ptr) __builtin_prefetch(ptr, 0, 0)
-
-/*-*************************************
-*  Macros
-***************************************/
-#define ZSTD_isError ERR_isError /* for inlining */
-#define FSE_isError ERR_isError
-#define HUF_isError ERR_isError
-
-/*_*******************************************************
-*  Memory operations
-**********************************************************/
-static void ZSTD_copy4(void *dst, const void *src) { memcpy(dst, src, 4); }
-
-/*-*************************************************************
-*   Context management
-***************************************************************/
-typedef enum {
-	ZSTDds_getFrameHeaderSize,
-	ZSTDds_decodeFrameHeader,
-	ZSTDds_decodeBlockHeader,
-	ZSTDds_decompressBlock,
-	ZSTDds_decompressLastBlock,
-	ZSTDds_checkChecksum,
-	ZSTDds_decodeSkippableHeader,
-	ZSTDds_skipFrame
-} ZSTD_dStage;
-
-typedef struct {
-	FSE_DTable LLTable[FSE_DTABLE_SIZE_U32(LLFSELog)];
-	FSE_DTable OFTable[FSE_DTABLE_SIZE_U32(OffFSELog)];
-	FSE_DTable MLTable[FSE_DTABLE_SIZE_U32(MLFSELog)];
-	HUF_DTable hufTable[HUF_DTABLE_SIZE(HufLog)]; /* can accommodate HUF_decompress4X */
-	U64 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32 / 2];
-	U32 rep[ZSTD_REP_NUM];
-} ZSTD_entropyTables_t;
-
-struct ZSTD_DCtx_s {
-	const FSE_DTable *LLTptr;
-	const FSE_DTable *MLTptr;
-	const FSE_DTable *OFTptr;
-	const HUF_DTable *HUFptr;
-	ZSTD_entropyTables_t entropy;
-	const void *previousDstEnd; /* detect continuity */
-	const void *base;	   /* start of curr segment */
-	const void *vBase;	  /* virtual start of previous segment if it was just before curr one */
-	const void *dictEnd;	/* end of previous segment */
-	size_t expected;
-	ZSTD_frameParams fParams;
-	blockType_e bType; /* used in ZSTD_decompressContinue(), to transfer blockType between header decoding and block decoding stages */
-	ZSTD_dStage stage;
-	U32 litEntropy;
-	U32 fseEntropy;
-	struct xxh64_state xxhState;
-	size_t headerSize;
-	U32 dictID;
-	const BYTE *litPtr;
-	ZSTD_customMem customMem;
-	size_t litSize;
-	size_t rleSize;
-	BYTE litBuffer[ZSTD_BLOCKSIZE_ABSOLUTEMAX + WILDCOPY_OVERLENGTH];
-	BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX];
-}; /* typedef'd to ZSTD_DCtx within "zstd.h" */
-
-size_t ZSTD_DCtxWorkspaceBound(void) { return ZSTD_ALIGN(sizeof(ZSTD_stack)) + ZSTD_ALIGN(sizeof(ZSTD_DCtx)); }
-
-size_t ZSTD_decompressBegin(ZSTD_DCtx *dctx)
-{
-	dctx->expected = ZSTD_frameHeaderSize_prefix;
-	dctx->stage = ZSTDds_getFrameHeaderSize;
-	dctx->previousDstEnd = NULL;
-	dctx->base = NULL;
-	dctx->vBase = NULL;
-	dctx->dictEnd = NULL;
-	dctx->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001); /* cover both little and big endian */
-	dctx->litEntropy = dctx->fseEntropy = 0;
-	dctx->dictID = 0;
-	ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.rep) == sizeof(repStartValue));
-	memcpy(dctx->entropy.rep, repStartValue, sizeof(repStartValue)); /* initial repcodes */
-	dctx->LLTptr = dctx->entropy.LLTable;
-	dctx->MLTptr = dctx->entropy.MLTable;
-	dctx->OFTptr = dctx->entropy.OFTable;
-	dctx->HUFptr = dctx->entropy.hufTable;
-	return 0;
-}
-
-static ZSTD_DCtx *ZSTD_createDCtx_advanced(ZSTD_customMem customMem)
-{
-	ZSTD_DCtx *dctx;
-
-	if (!customMem.customAlloc || !customMem.customFree)
-		return NULL;
-
-	dctx = (ZSTD_DCtx *)ZSTD_malloc(sizeof(ZSTD_DCtx), customMem);
-	if (!dctx)
-		return NULL;
-	memcpy(&dctx->customMem, &customMem, sizeof(customMem));
-	ZSTD_decompressBegin(dctx);
-	return dctx;
-}
-
-ZSTD_DCtx *ZSTD_initDCtx(void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
-	return ZSTD_createDCtx_advanced(stackMem);
-}
-
-size_t ZSTD_freeDCtx(ZSTD_DCtx *dctx)
-{
-	if (dctx == NULL)
-		return 0; /* support free on NULL */
-	ZSTD_free(dctx, dctx->customMem);
-	return 0; /* reserved as a potential error code in the future */
-}
-
-void ZSTD_copyDCtx(ZSTD_DCtx *dstDCtx, const ZSTD_DCtx *srcDCtx)
-{
-	size_t const workSpaceSize = (ZSTD_BLOCKSIZE_ABSOLUTEMAX + WILDCOPY_OVERLENGTH) + ZSTD_frameHeaderSize_max;
-	memcpy(dstDCtx, srcDCtx, sizeof(ZSTD_DCtx) - workSpaceSize); /* no need to copy workspace */
-}
-
-static void ZSTD_refDDict(ZSTD_DCtx *dstDCtx, const ZSTD_DDict *ddict);
-
-/*-*************************************************************
-*   Decompression section
-***************************************************************/
-
-/*! ZSTD_isFrame() :
- *  Tells if the content of `buffer` starts with a valid Frame Identifier.
- *  Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
- *  Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
- *  Note 3 : Skippable Frame Identifiers are considered valid. */
-unsigned ZSTD_isFrame(const void *buffer, size_t size)
-{
-	if (size < 4)
-		return 0;
-	{
-		U32 const magic = ZSTD_readLE32(buffer);
-		if (magic == ZSTD_MAGICNUMBER)
-			return 1;
-		if ((magic & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START)
-			return 1;
-	}
-	return 0;
-}
-
-/** ZSTD_frameHeaderSize() :
-*   srcSize must be >= ZSTD_frameHeaderSize_prefix.
-*   @return : size of the Frame Header */
-static size_t ZSTD_frameHeaderSize(const void *src, size_t srcSize)
-{
-	if (srcSize < ZSTD_frameHeaderSize_prefix)
-		return ERROR(srcSize_wrong);
-	{
-		BYTE const fhd = ((const BYTE *)src)[4];
-		U32 const dictID = fhd & 3;
-		U32 const singleSegment = (fhd >> 5) & 1;
-		U32 const fcsId = fhd >> 6;
-		return ZSTD_frameHeaderSize_prefix + !singleSegment + ZSTD_did_fieldSize[dictID] + ZSTD_fcs_fieldSize[fcsId] + (singleSegment && !fcsId);
-	}
-}
-
-/** ZSTD_getFrameParams() :
-*   decode Frame Header, or require larger `srcSize`.
-*   @return : 0, `fparamsPtr` is correctly filled,
-*            >0, `srcSize` is too small, result is expected `srcSize`,
-*             or an error code, which can be tested using ZSTD_isError() */
-size_t ZSTD_getFrameParams(ZSTD_frameParams *fparamsPtr, const void *src, size_t srcSize)
-{
-	const BYTE *ip = (const BYTE *)src;
-
-	if (srcSize < ZSTD_frameHeaderSize_prefix)
-		return ZSTD_frameHeaderSize_prefix;
-	if (ZSTD_readLE32(src) != ZSTD_MAGICNUMBER) {
-		if ((ZSTD_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
-			if (srcSize < ZSTD_skippableHeaderSize)
-				return ZSTD_skippableHeaderSize; /* magic number + skippable frame length */
-			memset(fparamsPtr, 0, sizeof(*fparamsPtr));
-			fparamsPtr->frameContentSize = ZSTD_readLE32((const char *)src + 4);
-			fparamsPtr->windowSize = 0; /* windowSize==0 means a frame is skippable */
-			return 0;
-		}
-		return ERROR(prefix_unknown);
-	}
-
-	/* ensure there is enough `srcSize` to fully read/decode frame header */
-	{
-		size_t const fhsize = ZSTD_frameHeaderSize(src, srcSize);
-		if (srcSize < fhsize)
-			return fhsize;
-	}
-
-	{
-		BYTE const fhdByte = ip[4];
-		size_t pos = 5;
-		U32 const dictIDSizeCode = fhdByte & 3;
-		U32 const checksumFlag = (fhdByte >> 2) & 1;
-		U32 const singleSegment = (fhdByte >> 5) & 1;
-		U32 const fcsID = fhdByte >> 6;
-		U32 const windowSizeMax = 1U << ZSTD_WINDOWLOG_MAX;
-		U32 windowSize = 0;
-		U32 dictID = 0;
-		U64 frameContentSize = 0;
-		if ((fhdByte & 0x08) != 0)
-			return ERROR(frameParameter_unsupported); /* reserved bits, which must be zero */
-		if (!singleSegment) {
-			BYTE const wlByte = ip[pos++];
-			U32 const windowLog = (wlByte >> 3) + ZSTD_WINDOWLOG_ABSOLUTEMIN;
-			if (windowLog > ZSTD_WINDOWLOG_MAX)
-				return ERROR(frameParameter_windowTooLarge); /* avoids issue with 1 << windowLog */
-			windowSize = (1U << windowLog);
-			windowSize += (windowSize >> 3) * (wlByte & 7);
-		}
-
-		switch (dictIDSizeCode) {
-		default: /* impossible */
-		case 0: break;
-		case 1:
-			dictID = ip[pos];
-			pos++;
-			break;
-		case 2:
-			dictID = ZSTD_readLE16(ip + pos);
-			pos += 2;
-			break;
-		case 3:
-			dictID = ZSTD_readLE32(ip + pos);
-			pos += 4;
-			break;
-		}
-		switch (fcsID) {
-		default: /* impossible */
-		case 0:
-			if (singleSegment)
-				frameContentSize = ip[pos];
-			break;
-		case 1: frameContentSize = ZSTD_readLE16(ip + pos) + 256; break;
-		case 2: frameContentSize = ZSTD_readLE32(ip + pos); break;
-		case 3: frameContentSize = ZSTD_readLE64(ip + pos); break;
-		}
-		if (!windowSize)
-			windowSize = (U32)frameContentSize;
-		if (windowSize > windowSizeMax)
-			return ERROR(frameParameter_windowTooLarge);
-		fparamsPtr->frameContentSize = frameContentSize;
-		fparamsPtr->windowSize = windowSize;
-		fparamsPtr->dictID = dictID;
-		fparamsPtr->checksumFlag = checksumFlag;
-	}
-	return 0;
-}
-
-/** ZSTD_getFrameContentSize() :
-*   compatible with legacy mode
-*   @return : decompressed size of the single frame pointed to be `src` if known, otherwise
-*             - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
-*             - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small) */
-unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize)
-{
-	{
-		ZSTD_frameParams fParams;
-		if (ZSTD_getFrameParams(&fParams, src, srcSize) != 0)
-			return ZSTD_CONTENTSIZE_ERROR;
-		if (fParams.windowSize == 0) {
-			/* Either skippable or empty frame, size == 0 either way */
-			return 0;
-		} else if (fParams.frameContentSize != 0) {
-			return fParams.frameContentSize;
-		} else {
-			return ZSTD_CONTENTSIZE_UNKNOWN;
-		}
-	}
-}
-
-/** ZSTD_findDecompressedSize() :
- *  compatible with legacy mode
- *  `srcSize` must be the exact length of some number of ZSTD compressed and/or
- *      skippable frames
- *  @return : decompressed size of the frames contained */
-unsigned long long ZSTD_findDecompressedSize(const void *src, size_t srcSize)
-{
-	{
-		unsigned long long totalDstSize = 0;
-		while (srcSize >= ZSTD_frameHeaderSize_prefix) {
-			const U32 magicNumber = ZSTD_readLE32(src);
-
-			if ((magicNumber & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
-				size_t skippableSize;
-				if (srcSize < ZSTD_skippableHeaderSize)
-					return ERROR(srcSize_wrong);
-				skippableSize = ZSTD_readLE32((const BYTE *)src + 4) + ZSTD_skippableHeaderSize;
-				if (srcSize < skippableSize) {
-					return ZSTD_CONTENTSIZE_ERROR;
-				}
-
-				src = (const BYTE *)src + skippableSize;
-				srcSize -= skippableSize;
-				continue;
-			}
-
-			{
-				unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
-				if (ret >= ZSTD_CONTENTSIZE_ERROR)
-					return ret;
-
-				/* check for overflow */
-				if (totalDstSize + ret < totalDstSize)
-					return ZSTD_CONTENTSIZE_ERROR;
-				totalDstSize += ret;
-			}
-			{
-				size_t const frameSrcSize = ZSTD_findFrameCompressedSize(src, srcSize);
-				if (ZSTD_isError(frameSrcSize)) {
-					return ZSTD_CONTENTSIZE_ERROR;
-				}
-
-				src = (const BYTE *)src + frameSrcSize;
-				srcSize -= frameSrcSize;
-			}
-		}
-
-		if (srcSize) {
-			return ZSTD_CONTENTSIZE_ERROR;
-		}
-
-		return totalDstSize;
-	}
-}
-
-/** ZSTD_decodeFrameHeader() :
-*   `headerSize` must be the size provided by ZSTD_frameHeaderSize().
-*   @return : 0 if success, or an error code, which can be tested using ZSTD_isError() */
-static size_t ZSTD_decodeFrameHeader(ZSTD_DCtx *dctx, const void *src, size_t headerSize)
-{
-	size_t const result = ZSTD_getFrameParams(&(dctx->fParams), src, headerSize);
-	if (ZSTD_isError(result))
-		return result; /* invalid header */
-	if (result > 0)
-		return ERROR(srcSize_wrong); /* headerSize too small */
-	if (dctx->fParams.dictID && (dctx->dictID != dctx->fParams.dictID))
-		return ERROR(dictionary_wrong);
-	if (dctx->fParams.checksumFlag)
-		xxh64_reset(&dctx->xxhState, 0);
-	return 0;
-}
-
-typedef struct {
-	blockType_e blockType;
-	U32 lastBlock;
-	U32 origSize;
-} blockProperties_t;
-
-/*! ZSTD_getcBlockSize() :
-*   Provides the size of compressed block from block header `src` */
-static size_t ZSTD_getcBlockSize(const void *src, size_t srcSize, blockProperties_t *bpPtr)
-{
-	if (srcSize < ZSTD_blockHeaderSize)
-		return ERROR(srcSize_wrong);
-	{
-		U32 const cBlockHeader = ZSTD_readLE24(src);
-		U32 const cSize = cBlockHeader >> 3;
-		bpPtr->lastBlock = cBlockHeader & 1;
-		bpPtr->blockType = (blockType_e)((cBlockHeader >> 1) & 3);
-		bpPtr->origSize = cSize; /* only useful for RLE */
-		if (bpPtr->blockType == bt_rle)
-			return 1;
-		if (bpPtr->blockType == bt_reserved)
-			return ERROR(corruption_detected);
-		return cSize;
-	}
-}
-
-static size_t ZSTD_copyRawBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	if (srcSize > dstCapacity)
-		return ERROR(dstSize_tooSmall);
-	memcpy(dst, src, srcSize);
-	return srcSize;
-}
-
-static size_t ZSTD_setRleBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize, size_t regenSize)
-{
-	if (srcSize != 1)
-		return ERROR(srcSize_wrong);
-	if (regenSize > dstCapacity)
-		return ERROR(dstSize_tooSmall);
-	memset(dst, *(const BYTE *)src, regenSize);
-	return regenSize;
-}
-
-/*! ZSTD_decodeLiteralsBlock() :
-	@return : nb of bytes read from src (< srcSize ) */
-static size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx *dctx, const void *src, size_t srcSize) /* note : srcSize < BLOCKSIZE */
-{
-	if (srcSize < MIN_CBLOCK_SIZE)
-		return ERROR(corruption_detected);
-
-	{
-		const BYTE *const istart = (const BYTE *)src;
-		symbolEncodingType_e const litEncType = (symbolEncodingType_e)(istart[0] & 3);
-
-		switch (litEncType) {
-		case set_repeat:
-			if (dctx->litEntropy == 0)
-				return ERROR(dictionary_corrupted);
-		/* fall-through */
-		case set_compressed:
-			if (srcSize < 5)
-				return ERROR(corruption_detected); /* srcSize >= MIN_CBLOCK_SIZE == 3; here we need up to 5 for case 3 */
-			{
-				size_t lhSize, litSize, litCSize;
-				U32 singleStream = 0;
-				U32 const lhlCode = (istart[0] >> 2) & 3;
-				U32 const lhc = ZSTD_readLE32(istart);
-				switch (lhlCode) {
-				case 0:
-				case 1:
-				default: /* note : default is impossible, since lhlCode into [0..3] */
-					/* 2 - 2 - 10 - 10 */
-					singleStream = !lhlCode;
-					lhSize = 3;
-					litSize = (lhc >> 4) & 0x3FF;
-					litCSize = (lhc >> 14) & 0x3FF;
-					break;
-				case 2:
-					/* 2 - 2 - 14 - 14 */
-					lhSize = 4;
-					litSize = (lhc >> 4) & 0x3FFF;
-					litCSize = lhc >> 18;
-					break;
-				case 3:
-					/* 2 - 2 - 18 - 18 */
-					lhSize = 5;
-					litSize = (lhc >> 4) & 0x3FFFF;
-					litCSize = (lhc >> 22) + (istart[4] << 10);
-					break;
-				}
-				if (litSize > ZSTD_BLOCKSIZE_ABSOLUTEMAX)
-					return ERROR(corruption_detected);
-				if (litCSize + lhSize > srcSize)
-					return ERROR(corruption_detected);
-
-				if (HUF_isError(
-					(litEncType == set_repeat)
-					    ? (singleStream ? HUF_decompress1X_usingDTable(dctx->litBuffer, litSize, istart + lhSize, litCSize, dctx->HUFptr)
-							    : HUF_decompress4X_usingDTable(dctx->litBuffer, litSize, istart + lhSize, litCSize, dctx->HUFptr))
-					    : (singleStream
-						   ? HUF_decompress1X2_DCtx_wksp(dctx->entropy.hufTable, dctx->litBuffer, litSize, istart + lhSize, litCSize,
-										 dctx->entropy.workspace, sizeof(dctx->entropy.workspace))
-						   : HUF_decompress4X_hufOnly_wksp(dctx->entropy.hufTable, dctx->litBuffer, litSize, istart + lhSize, litCSize,
-										   dctx->entropy.workspace, sizeof(dctx->entropy.workspace)))))
-					return ERROR(corruption_detected);
-
-				dctx->litPtr = dctx->litBuffer;
-				dctx->litSize = litSize;
-				dctx->litEntropy = 1;
-				if (litEncType == set_compressed)
-					dctx->HUFptr = dctx->entropy.hufTable;
-				memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
-				return litCSize + lhSize;
-			}
-
-		case set_basic: {
-			size_t litSize, lhSize;
-			U32 const lhlCode = ((istart[0]) >> 2) & 3;
-			switch (lhlCode) {
-			case 0:
-			case 2:
-			default: /* note : default is impossible, since lhlCode into [0..3] */
-				lhSize = 1;
-				litSize = istart[0] >> 3;
-				break;
-			case 1:
-				lhSize = 2;
-				litSize = ZSTD_readLE16(istart) >> 4;
-				break;
-			case 3:
-				lhSize = 3;
-				litSize = ZSTD_readLE24(istart) >> 4;
-				break;
-			}
-
-			if (lhSize + litSize + WILDCOPY_OVERLENGTH > srcSize) { /* risk reading beyond src buffer with wildcopy */
-				if (litSize + lhSize > srcSize)
-					return ERROR(corruption_detected);
-				memcpy(dctx->litBuffer, istart + lhSize, litSize);
-				dctx->litPtr = dctx->litBuffer;
-				dctx->litSize = litSize;
-				memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
-				return lhSize + litSize;
-			}
-			/* direct reference into compressed stream */
-			dctx->litPtr = istart + lhSize;
-			dctx->litSize = litSize;
-			return lhSize + litSize;
-		}
-
-		case set_rle: {
-			U32 const lhlCode = ((istart[0]) >> 2) & 3;
-			size_t litSize, lhSize;
-			switch (lhlCode) {
-			case 0:
-			case 2:
-			default: /* note : default is impossible, since lhlCode into [0..3] */
-				lhSize = 1;
-				litSize = istart[0] >> 3;
-				break;
-			case 1:
-				lhSize = 2;
-				litSize = ZSTD_readLE16(istart) >> 4;
-				break;
-			case 3:
-				lhSize = 3;
-				litSize = ZSTD_readLE24(istart) >> 4;
-				if (srcSize < 4)
-					return ERROR(corruption_detected); /* srcSize >= MIN_CBLOCK_SIZE == 3; here we need lhSize+1 = 4 */
-				break;
-			}
-			if (litSize > ZSTD_BLOCKSIZE_ABSOLUTEMAX)
-				return ERROR(corruption_detected);
-			memset(dctx->litBuffer, istart[lhSize], litSize + WILDCOPY_OVERLENGTH);
-			dctx->litPtr = dctx->litBuffer;
-			dctx->litSize = litSize;
-			return lhSize + 1;
-		}
-		default:
-			return ERROR(corruption_detected); /* impossible */
-		}
-	}
-}
-
-typedef union {
-	FSE_decode_t realData;
-	U32 alignedBy4;
-} FSE_decode_t4;
-
-static const FSE_decode_t4 LL_defaultDTable[(1 << LL_DEFAULTNORMLOG) + 1] = {
-    {{LL_DEFAULTNORMLOG, 1, 1}}, /* header : tableLog, fastMode, fastMode */
-    {{0, 0, 4}},		 /* 0 : base, symbol, bits */
-    {{16, 0, 4}},
-    {{32, 1, 5}},
-    {{0, 3, 5}},
-    {{0, 4, 5}},
-    {{0, 6, 5}},
-    {{0, 7, 5}},
-    {{0, 9, 5}},
-    {{0, 10, 5}},
-    {{0, 12, 5}},
-    {{0, 14, 6}},
-    {{0, 16, 5}},
-    {{0, 18, 5}},
-    {{0, 19, 5}},
-    {{0, 21, 5}},
-    {{0, 22, 5}},
-    {{0, 24, 5}},
-    {{32, 25, 5}},
-    {{0, 26, 5}},
-    {{0, 27, 6}},
-    {{0, 29, 6}},
-    {{0, 31, 6}},
-    {{32, 0, 4}},
-    {{0, 1, 4}},
-    {{0, 2, 5}},
-    {{32, 4, 5}},
-    {{0, 5, 5}},
-    {{32, 7, 5}},
-    {{0, 8, 5}},
-    {{32, 10, 5}},
-    {{0, 11, 5}},
-    {{0, 13, 6}},
-    {{32, 16, 5}},
-    {{0, 17, 5}},
-    {{32, 19, 5}},
-    {{0, 20, 5}},
-    {{32, 22, 5}},
-    {{0, 23, 5}},
-    {{0, 25, 4}},
-    {{16, 25, 4}},
-    {{32, 26, 5}},
-    {{0, 28, 6}},
-    {{0, 30, 6}},
-    {{48, 0, 4}},
-    {{16, 1, 4}},
-    {{32, 2, 5}},
-    {{32, 3, 5}},
-    {{32, 5, 5}},
-    {{32, 6, 5}},
-    {{32, 8, 5}},
-    {{32, 9, 5}},
-    {{32, 11, 5}},
-    {{32, 12, 5}},
-    {{0, 15, 6}},
-    {{32, 17, 5}},
-    {{32, 18, 5}},
-    {{32, 20, 5}},
-    {{32, 21, 5}},
-    {{32, 23, 5}},
-    {{32, 24, 5}},
-    {{0, 35, 6}},
-    {{0, 34, 6}},
-    {{0, 33, 6}},
-    {{0, 32, 6}},
-}; /* LL_defaultDTable */
-
-static const FSE_decode_t4 ML_defaultDTable[(1 << ML_DEFAULTNORMLOG) + 1] = {
-    {{ML_DEFAULTNORMLOG, 1, 1}}, /* header : tableLog, fastMode, fastMode */
-    {{0, 0, 6}},		 /* 0 : base, symbol, bits */
-    {{0, 1, 4}},
-    {{32, 2, 5}},
-    {{0, 3, 5}},
-    {{0, 5, 5}},
-    {{0, 6, 5}},
-    {{0, 8, 5}},
-    {{0, 10, 6}},
-    {{0, 13, 6}},
-    {{0, 16, 6}},
-    {{0, 19, 6}},
-    {{0, 22, 6}},
-    {{0, 25, 6}},
-    {{0, 28, 6}},
-    {{0, 31, 6}},
-    {{0, 33, 6}},
-    {{0, 35, 6}},
-    {{0, 37, 6}},
-    {{0, 39, 6}},
-    {{0, 41, 6}},
-    {{0, 43, 6}},
-    {{0, 45, 6}},
-    {{16, 1, 4}},
-    {{0, 2, 4}},
-    {{32, 3, 5}},
-    {{0, 4, 5}},
-    {{32, 6, 5}},
-    {{0, 7, 5}},
-    {{0, 9, 6}},
-    {{0, 12, 6}},
-    {{0, 15, 6}},
-    {{0, 18, 6}},
-    {{0, 21, 6}},
-    {{0, 24, 6}},
-    {{0, 27, 6}},
-    {{0, 30, 6}},
-    {{0, 32, 6}},
-    {{0, 34, 6}},
-    {{0, 36, 6}},
-    {{0, 38, 6}},
-    {{0, 40, 6}},
-    {{0, 42, 6}},
-    {{0, 44, 6}},
-    {{32, 1, 4}},
-    {{48, 1, 4}},
-    {{16, 2, 4}},
-    {{32, 4, 5}},
-    {{32, 5, 5}},
-    {{32, 7, 5}},
-    {{32, 8, 5}},
-    {{0, 11, 6}},
-    {{0, 14, 6}},
-    {{0, 17, 6}},
-    {{0, 20, 6}},
-    {{0, 23, 6}},
-    {{0, 26, 6}},
-    {{0, 29, 6}},
-    {{0, 52, 6}},
-    {{0, 51, 6}},
-    {{0, 50, 6}},
-    {{0, 49, 6}},
-    {{0, 48, 6}},
-    {{0, 47, 6}},
-    {{0, 46, 6}},
-}; /* ML_defaultDTable */
-
-static const FSE_decode_t4 OF_defaultDTable[(1 << OF_DEFAULTNORMLOG) + 1] = {
-    {{OF_DEFAULTNORMLOG, 1, 1}}, /* header : tableLog, fastMode, fastMode */
-    {{0, 0, 5}},		 /* 0 : base, symbol, bits */
-    {{0, 6, 4}},
-    {{0, 9, 5}},
-    {{0, 15, 5}},
-    {{0, 21, 5}},
-    {{0, 3, 5}},
-    {{0, 7, 4}},
-    {{0, 12, 5}},
-    {{0, 18, 5}},
-    {{0, 23, 5}},
-    {{0, 5, 5}},
-    {{0, 8, 4}},
-    {{0, 14, 5}},
-    {{0, 20, 5}},
-    {{0, 2, 5}},
-    {{16, 7, 4}},
-    {{0, 11, 5}},
-    {{0, 17, 5}},
-    {{0, 22, 5}},
-    {{0, 4, 5}},
-    {{16, 8, 4}},
-    {{0, 13, 5}},
-    {{0, 19, 5}},
-    {{0, 1, 5}},
-    {{16, 6, 4}},
-    {{0, 10, 5}},
-    {{0, 16, 5}},
-    {{0, 28, 5}},
-    {{0, 27, 5}},
-    {{0, 26, 5}},
-    {{0, 25, 5}},
-    {{0, 24, 5}},
-}; /* OF_defaultDTable */
-
-/*! ZSTD_buildSeqTable() :
-	@return : nb bytes read from src,
-			  or an error code if it fails, testable with ZSTD_isError()
-*/
-static size_t ZSTD_buildSeqTable(FSE_DTable *DTableSpace, const FSE_DTable **DTablePtr, symbolEncodingType_e type, U32 max, U32 maxLog, const void *src,
-				 size_t srcSize, const FSE_decode_t4 *defaultTable, U32 flagRepeatTable, void *workspace, size_t workspaceSize)
-{
-	const void *const tmpPtr = defaultTable; /* bypass strict aliasing */
-	switch (type) {
-	case set_rle:
-		if (!srcSize)
-			return ERROR(srcSize_wrong);
-		if ((*(const BYTE *)src) > max)
-			return ERROR(corruption_detected);
-		FSE_buildDTable_rle(DTableSpace, *(const BYTE *)src);
-		*DTablePtr = DTableSpace;
-		return 1;
-	case set_basic: *DTablePtr = (const FSE_DTable *)tmpPtr; return 0;
-	case set_repeat:
-		if (!flagRepeatTable)
-			return ERROR(corruption_detected);
-		return 0;
-	default: /* impossible */
-	case set_compressed: {
-		U32 tableLog;
-		S16 *norm = (S16 *)workspace;
-		size_t const spaceUsed32 = ALIGN(sizeof(S16) * (MaxSeq + 1), sizeof(U32)) >> 2;
-
-		if ((spaceUsed32 << 2) > workspaceSize)
-			return ERROR(GENERIC);
-		workspace = (U32 *)workspace + spaceUsed32;
-		workspaceSize -= (spaceUsed32 << 2);
-		{
-			size_t const headerSize = FSE_readNCount(norm, &max, &tableLog, src, srcSize);
-			if (FSE_isError(headerSize))
-				return ERROR(corruption_detected);
-			if (tableLog > maxLog)
-				return ERROR(corruption_detected);
-			FSE_buildDTable_wksp(DTableSpace, norm, max, tableLog, workspace, workspaceSize);
-			*DTablePtr = DTableSpace;
-			return headerSize;
-		}
-	}
-	}
-}
-
-static size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx *dctx, int *nbSeqPtr, const void *src, size_t srcSize)
-{
-	const BYTE *const istart = (const BYTE *const)src;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *ip = istart;
-
-	/* check */
-	if (srcSize < MIN_SEQUENCES_SIZE)
-		return ERROR(srcSize_wrong);
-
-	/* SeqHead */
-	{
-		int nbSeq = *ip++;
-		if (!nbSeq) {
-			*nbSeqPtr = 0;
-			return 1;
-		}
-		if (nbSeq > 0x7F) {
-			if (nbSeq == 0xFF) {
-				if (ip + 2 > iend)
-					return ERROR(srcSize_wrong);
-				nbSeq = ZSTD_readLE16(ip) + LONGNBSEQ, ip += 2;
-			} else {
-				if (ip >= iend)
-					return ERROR(srcSize_wrong);
-				nbSeq = ((nbSeq - 0x80) << 8) + *ip++;
-			}
-		}
-		*nbSeqPtr = nbSeq;
-	}
-
-	/* FSE table descriptors */
-	if (ip + 4 > iend)
-		return ERROR(srcSize_wrong); /* minimum possible size */
-	{
-		symbolEncodingType_e const LLtype = (symbolEncodingType_e)(*ip >> 6);
-		symbolEncodingType_e const OFtype = (symbolEncodingType_e)((*ip >> 4) & 3);
-		symbolEncodingType_e const MLtype = (symbolEncodingType_e)((*ip >> 2) & 3);
-		ip++;
-
-		/* Build DTables */
-		{
-			size_t const llhSize = ZSTD_buildSeqTable(dctx->entropy.LLTable, &dctx->LLTptr, LLtype, MaxLL, LLFSELog, ip, iend - ip,
-								  LL_defaultDTable, dctx->fseEntropy, dctx->entropy.workspace, sizeof(dctx->entropy.workspace));
-			if (ZSTD_isError(llhSize))
-				return ERROR(corruption_detected);
-			ip += llhSize;
-		}
-		{
-			size_t const ofhSize = ZSTD_buildSeqTable(dctx->entropy.OFTable, &dctx->OFTptr, OFtype, MaxOff, OffFSELog, ip, iend - ip,
-								  OF_defaultDTable, dctx->fseEntropy, dctx->entropy.workspace, sizeof(dctx->entropy.workspace));
-			if (ZSTD_isError(ofhSize))
-				return ERROR(corruption_detected);
-			ip += ofhSize;
-		}
-		{
-			size_t const mlhSize = ZSTD_buildSeqTable(dctx->entropy.MLTable, &dctx->MLTptr, MLtype, MaxML, MLFSELog, ip, iend - ip,
-								  ML_defaultDTable, dctx->fseEntropy, dctx->entropy.workspace, sizeof(dctx->entropy.workspace));
-			if (ZSTD_isError(mlhSize))
-				return ERROR(corruption_detected);
-			ip += mlhSize;
-		}
-	}
-
-	return ip - istart;
-}
-
-typedef struct {
-	size_t litLength;
-	size_t matchLength;
-	size_t offset;
-	const BYTE *match;
-} seq_t;
-
-typedef struct {
-	BIT_DStream_t DStream;
-	FSE_DState_t stateLL;
-	FSE_DState_t stateOffb;
-	FSE_DState_t stateML;
-	size_t prevOffset[ZSTD_REP_NUM];
-	const BYTE *base;
-	size_t pos;
-	uPtrDiff gotoDict;
-} seqState_t;
-
-FORCE_NOINLINE
-size_t ZSTD_execSequenceLast7(BYTE *op, BYTE *const oend, seq_t sequence, const BYTE **litPtr, const BYTE *const litLimit, const BYTE *const base,
-			      const BYTE *const vBase, const BYTE *const dictEnd)
-{
-	BYTE *const oLitEnd = op + sequence.litLength;
-	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
-	BYTE *const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
-	BYTE *const oend_w = oend - WILDCOPY_OVERLENGTH;
-	const BYTE *const iLitEnd = *litPtr + sequence.litLength;
-	const BYTE *match = oLitEnd - sequence.offset;
-
-	/* check */
-	if (oMatchEnd > oend)
-		return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
-	if (iLitEnd > litLimit)
-		return ERROR(corruption_detected); /* over-read beyond lit buffer */
-	if (oLitEnd <= oend_w)
-		return ERROR(GENERIC); /* Precondition */
-
-	/* copy literals */
-	if (op < oend_w) {
-		ZSTD_wildcopy(op, *litPtr, oend_w - op);
-		*litPtr += oend_w - op;
-		op = oend_w;
-	}
-	while (op < oLitEnd)
-		*op++ = *(*litPtr)++;
-
-	/* copy Match */
-	if (sequence.offset > (size_t)(oLitEnd - base)) {
-		/* offset beyond prefix */
-		if (sequence.offset > (size_t)(oLitEnd - vBase))
-			return ERROR(corruption_detected);
-		match = dictEnd - (base - match);
-		if (match + sequence.matchLength <= dictEnd) {
-			memmove(oLitEnd, match, sequence.matchLength);
-			return sequenceLength;
-		}
-		/* span extDict & currPrefixSegment */
-		{
-			size_t const length1 = dictEnd - match;
-			memmove(oLitEnd, match, length1);
-			op = oLitEnd + length1;
-			sequence.matchLength -= length1;
-			match = base;
-		}
-	}
-	while (op < oMatchEnd)
-		*op++ = *match++;
-	return sequenceLength;
-}
-
-static seq_t ZSTD_decodeSequence(seqState_t *seqState)
-{
-	seq_t seq;
-
-	U32 const llCode = FSE_peekSymbol(&seqState->stateLL);
-	U32 const mlCode = FSE_peekSymbol(&seqState->stateML);
-	U32 const ofCode = FSE_peekSymbol(&seqState->stateOffb); /* <= maxOff, by table construction */
-
-	U32 const llBits = LL_bits[llCode];
-	U32 const mlBits = ML_bits[mlCode];
-	U32 const ofBits = ofCode;
-	U32 const totalBits = llBits + mlBits + ofBits;
-
-	static const U32 LL_base[MaxLL + 1] = {0,  1,  2,  3,  4,  5,  6,  7,  8,    9,     10,    11,    12,    13,     14,     15,     16,     18,
-					       20, 22, 24, 28, 32, 40, 48, 64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000, 0x2000, 0x4000, 0x8000, 0x10000};
-
-	static const U32 ML_base[MaxML + 1] = {3,  4,  5,  6,  7,  8,  9,  10,   11,    12,    13,    14,    15,     16,     17,     18,     19,     20,
-					       21, 22, 23, 24, 25, 26, 27, 28,   29,    30,    31,    32,    33,     34,     35,     37,     39,     41,
-					       43, 47, 51, 59, 67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803, 0x1003, 0x2003, 0x4003, 0x8003, 0x10003};
-
-	static const U32 OF_base[MaxOff + 1] = {0,       1,	1,	5,	0xD,      0x1D,      0x3D,      0x7D,      0xFD,     0x1FD,
-						0x3FD,   0x7FD,    0xFFD,    0x1FFD,   0x3FFD,   0x7FFD,    0xFFFD,    0x1FFFD,   0x3FFFD,  0x7FFFD,
-						0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD, 0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD};
-
-	/* sequence */
-	{
-		size_t offset;
-		if (!ofCode)
-			offset = 0;
-		else {
-			offset = OF_base[ofCode] + BIT_readBitsFast(&seqState->DStream, ofBits); /* <=  (ZSTD_WINDOWLOG_MAX-1) bits */
-			if (ZSTD_32bits())
-				BIT_reloadDStream(&seqState->DStream);
-		}
-
-		if (ofCode <= 1) {
-			offset += (llCode == 0);
-			if (offset) {
-				size_t temp = (offset == 3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
-				temp += !temp; /* 0 is not valid; input is corrupted; force offset to 1 */
-				if (offset != 1)
-					seqState->prevOffset[2] = seqState->prevOffset[1];
-				seqState->prevOffset[1] = seqState->prevOffset[0];
-				seqState->prevOffset[0] = offset = temp;
-			} else {
-				offset = seqState->prevOffset[0];
-			}
-		} else {
-			seqState->prevOffset[2] = seqState->prevOffset[1];
-			seqState->prevOffset[1] = seqState->prevOffset[0];
-			seqState->prevOffset[0] = offset;
-		}
-		seq.offset = offset;
-	}
-
-	seq.matchLength = ML_base[mlCode] + ((mlCode > 31) ? BIT_readBitsFast(&seqState->DStream, mlBits) : 0); /* <=  16 bits */
-	if (ZSTD_32bits() && (mlBits + llBits > 24))
-		BIT_reloadDStream(&seqState->DStream);
-
-	seq.litLength = LL_base[llCode] + ((llCode > 15) ? BIT_readBitsFast(&seqState->DStream, llBits) : 0); /* <=  16 bits */
-	if (ZSTD_32bits() || (totalBits > 64 - 7 - (LLFSELog + MLFSELog + OffFSELog)))
-		BIT_reloadDStream(&seqState->DStream);
-
-	/* ANS state update */
-	FSE_updateState(&seqState->stateLL, &seqState->DStream); /* <=  9 bits */
-	FSE_updateState(&seqState->stateML, &seqState->DStream); /* <=  9 bits */
-	if (ZSTD_32bits())
-		BIT_reloadDStream(&seqState->DStream);		   /* <= 18 bits */
-	FSE_updateState(&seqState->stateOffb, &seqState->DStream); /* <=  8 bits */
-
-	seq.match = NULL;
-
-	return seq;
-}
-
-FORCE_INLINE
-size_t ZSTD_execSequence(BYTE *op, BYTE *const oend, seq_t sequence, const BYTE **litPtr, const BYTE *const litLimit, const BYTE *const base,
-			 const BYTE *const vBase, const BYTE *const dictEnd)
-{
-	BYTE *const oLitEnd = op + sequence.litLength;
-	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
-	BYTE *const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
-	BYTE *const oend_w = oend - WILDCOPY_OVERLENGTH;
-	const BYTE *const iLitEnd = *litPtr + sequence.litLength;
-	const BYTE *match = oLitEnd - sequence.offset;
-
-	/* check */
-	if (oMatchEnd > oend)
-		return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
-	if (iLitEnd > litLimit)
-		return ERROR(corruption_detected); /* over-read beyond lit buffer */
-	if (oLitEnd > oend_w)
-		return ZSTD_execSequenceLast7(op, oend, sequence, litPtr, litLimit, base, vBase, dictEnd);
-
-	/* copy Literals */
-	ZSTD_copy8(op, *litPtr);
-	if (sequence.litLength > 8)
-		ZSTD_wildcopy(op + 8, (*litPtr) + 8,
-			      sequence.litLength - 8); /* note : since oLitEnd <= oend-WILDCOPY_OVERLENGTH, no risk of overwrite beyond oend */
-	op = oLitEnd;
-	*litPtr = iLitEnd; /* update for next sequence */
-
-	/* copy Match */
-	if (sequence.offset > (size_t)(oLitEnd - base)) {
-		/* offset beyond prefix */
-		if (sequence.offset > (size_t)(oLitEnd - vBase))
-			return ERROR(corruption_detected);
-		match = dictEnd + (match - base);
-		if (match + sequence.matchLength <= dictEnd) {
-			memmove(oLitEnd, match, sequence.matchLength);
-			return sequenceLength;
-		}
-		/* span extDict & currPrefixSegment */
-		{
-			size_t const length1 = dictEnd - match;
-			memmove(oLitEnd, match, length1);
-			op = oLitEnd + length1;
-			sequence.matchLength -= length1;
-			match = base;
-			if (op > oend_w || sequence.matchLength < MINMATCH) {
-				U32 i;
-				for (i = 0; i < sequence.matchLength; ++i)
-					op[i] = match[i];
-				return sequenceLength;
-			}
-		}
-	}
-	/* Requirement: op <= oend_w && sequence.matchLength >= MINMATCH */
-
-	/* match within prefix */
-	if (sequence.offset < 8) {
-		/* close range match, overlap */
-		static const U32 dec32table[] = {0, 1, 2, 1, 4, 4, 4, 4};   /* added */
-		static const int dec64table[] = {8, 8, 8, 7, 8, 9, 10, 11}; /* subtracted */
-		int const sub2 = dec64table[sequence.offset];
-		op[0] = match[0];
-		op[1] = match[1];
-		op[2] = match[2];
-		op[3] = match[3];
-		match += dec32table[sequence.offset];
-		ZSTD_copy4(op + 4, match);
-		match -= sub2;
-	} else {
-		ZSTD_copy8(op, match);
-	}
-	op += 8;
-	match += 8;
-
-	if (oMatchEnd > oend - (16 - MINMATCH)) {
-		if (op < oend_w) {
-			ZSTD_wildcopy(op, match, oend_w - op);
-			match += oend_w - op;
-			op = oend_w;
-		}
-		while (op < oMatchEnd)
-			*op++ = *match++;
-	} else {
-		ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength - 8); /* works even if matchLength < 8 */
-	}
-	return sequenceLength;
-}
-
-static size_t ZSTD_decompressSequences(ZSTD_DCtx *dctx, void *dst, size_t maxDstSize, const void *seqStart, size_t seqSize)
-{
-	const BYTE *ip = (const BYTE *)seqStart;
-	const BYTE *const iend = ip + seqSize;
-	BYTE *const ostart = (BYTE * const)dst;
-	BYTE *const oend = ostart + maxDstSize;
-	BYTE *op = ostart;
-	const BYTE *litPtr = dctx->litPtr;
-	const BYTE *const litEnd = litPtr + dctx->litSize;
-	const BYTE *const base = (const BYTE *)(dctx->base);
-	const BYTE *const vBase = (const BYTE *)(dctx->vBase);
-	const BYTE *const dictEnd = (const BYTE *)(dctx->dictEnd);
-	int nbSeq;
-
-	/* Build Decoding Tables */
-	{
-		size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, seqSize);
-		if (ZSTD_isError(seqHSize))
-			return seqHSize;
-		ip += seqHSize;
-	}
-
-	/* Regen sequences */
-	if (nbSeq) {
-		seqState_t seqState;
-		dctx->fseEntropy = 1;
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				seqState.prevOffset[i] = dctx->entropy.rep[i];
-		}
-		CHECK_E(BIT_initDStream(&seqState.DStream, ip, iend - ip), corruption_detected);
-		FSE_initDState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
-		FSE_initDState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
-		FSE_initDState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
-
-		for (; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && nbSeq;) {
-			nbSeq--;
-			{
-				seq_t const sequence = ZSTD_decodeSequence(&seqState);
-				size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litEnd, base, vBase, dictEnd);
-				if (ZSTD_isError(oneSeqSize))
-					return oneSeqSize;
-				op += oneSeqSize;
-			}
-		}
-
-		/* check if reached exact end */
-		if (nbSeq)
-			return ERROR(corruption_detected);
-		/* save reps for next block */
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]);
-		}
-	}
-
-	/* last literal segment */
-	{
-		size_t const lastLLSize = litEnd - litPtr;
-		if (lastLLSize > (size_t)(oend - op))
-			return ERROR(dstSize_tooSmall);
-		memcpy(op, litPtr, lastLLSize);
-		op += lastLLSize;
-	}
-
-	return op - ostart;
-}
-
-FORCE_INLINE seq_t ZSTD_decodeSequenceLong_generic(seqState_t *seqState, int const longOffsets)
-{
-	seq_t seq;
-
-	U32 const llCode = FSE_peekSymbol(&seqState->stateLL);
-	U32 const mlCode = FSE_peekSymbol(&seqState->stateML);
-	U32 const ofCode = FSE_peekSymbol(&seqState->stateOffb); /* <= maxOff, by table construction */
-
-	U32 const llBits = LL_bits[llCode];
-	U32 const mlBits = ML_bits[mlCode];
-	U32 const ofBits = ofCode;
-	U32 const totalBits = llBits + mlBits + ofBits;
-
-	static const U32 LL_base[MaxLL + 1] = {0,  1,  2,  3,  4,  5,  6,  7,  8,    9,     10,    11,    12,    13,     14,     15,     16,     18,
-					       20, 22, 24, 28, 32, 40, 48, 64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000, 0x2000, 0x4000, 0x8000, 0x10000};
-
-	static const U32 ML_base[MaxML + 1] = {3,  4,  5,  6,  7,  8,  9,  10,   11,    12,    13,    14,    15,     16,     17,     18,     19,     20,
-					       21, 22, 23, 24, 25, 26, 27, 28,   29,    30,    31,    32,    33,     34,     35,     37,     39,     41,
-					       43, 47, 51, 59, 67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803, 0x1003, 0x2003, 0x4003, 0x8003, 0x10003};
-
-	static const U32 OF_base[MaxOff + 1] = {0,       1,	1,	5,	0xD,      0x1D,      0x3D,      0x7D,      0xFD,     0x1FD,
-						0x3FD,   0x7FD,    0xFFD,    0x1FFD,   0x3FFD,   0x7FFD,    0xFFFD,    0x1FFFD,   0x3FFFD,  0x7FFFD,
-						0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD, 0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD};
-
-	/* sequence */
-	{
-		size_t offset;
-		if (!ofCode)
-			offset = 0;
-		else {
-			if (longOffsets) {
-				int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN);
-				offset = OF_base[ofCode] + (BIT_readBitsFast(&seqState->DStream, ofBits - extraBits) << extraBits);
-				if (ZSTD_32bits() || extraBits)
-					BIT_reloadDStream(&seqState->DStream);
-				if (extraBits)
-					offset += BIT_readBitsFast(&seqState->DStream, extraBits);
-			} else {
-				offset = OF_base[ofCode] + BIT_readBitsFast(&seqState->DStream, ofBits); /* <=  (ZSTD_WINDOWLOG_MAX-1) bits */
-				if (ZSTD_32bits())
-					BIT_reloadDStream(&seqState->DStream);
-			}
-		}
-
-		if (ofCode <= 1) {
-			offset += (llCode == 0);
-			if (offset) {
-				size_t temp = (offset == 3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
-				temp += !temp; /* 0 is not valid; input is corrupted; force offset to 1 */
-				if (offset != 1)
-					seqState->prevOffset[2] = seqState->prevOffset[1];
-				seqState->prevOffset[1] = seqState->prevOffset[0];
-				seqState->prevOffset[0] = offset = temp;
-			} else {
-				offset = seqState->prevOffset[0];
-			}
-		} else {
-			seqState->prevOffset[2] = seqState->prevOffset[1];
-			seqState->prevOffset[1] = seqState->prevOffset[0];
-			seqState->prevOffset[0] = offset;
-		}
-		seq.offset = offset;
-	}
-
-	seq.matchLength = ML_base[mlCode] + ((mlCode > 31) ? BIT_readBitsFast(&seqState->DStream, mlBits) : 0); /* <=  16 bits */
-	if (ZSTD_32bits() && (mlBits + llBits > 24))
-		BIT_reloadDStream(&seqState->DStream);
-
-	seq.litLength = LL_base[llCode] + ((llCode > 15) ? BIT_readBitsFast(&seqState->DStream, llBits) : 0); /* <=  16 bits */
-	if (ZSTD_32bits() || (totalBits > 64 - 7 - (LLFSELog + MLFSELog + OffFSELog)))
-		BIT_reloadDStream(&seqState->DStream);
-
-	{
-		size_t const pos = seqState->pos + seq.litLength;
-		seq.match = seqState->base + pos - seq.offset; /* single memory segment */
-		if (seq.offset > pos)
-			seq.match += seqState->gotoDict; /* separate memory segment */
-		seqState->pos = pos + seq.matchLength;
-	}
-
-	/* ANS state update */
-	FSE_updateState(&seqState->stateLL, &seqState->DStream); /* <=  9 bits */
-	FSE_updateState(&seqState->stateML, &seqState->DStream); /* <=  9 bits */
-	if (ZSTD_32bits())
-		BIT_reloadDStream(&seqState->DStream);		   /* <= 18 bits */
-	FSE_updateState(&seqState->stateOffb, &seqState->DStream); /* <=  8 bits */
-
-	return seq;
-}
-
-static seq_t ZSTD_decodeSequenceLong(seqState_t *seqState, unsigned const windowSize)
-{
-	if (ZSTD_highbit32(windowSize) > STREAM_ACCUMULATOR_MIN) {
-		return ZSTD_decodeSequenceLong_generic(seqState, 1);
-	} else {
-		return ZSTD_decodeSequenceLong_generic(seqState, 0);
-	}
-}
-
-FORCE_INLINE
-size_t ZSTD_execSequenceLong(BYTE *op, BYTE *const oend, seq_t sequence, const BYTE **litPtr, const BYTE *const litLimit, const BYTE *const base,
-			     const BYTE *const vBase, const BYTE *const dictEnd)
-{
-	BYTE *const oLitEnd = op + sequence.litLength;
-	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
-	BYTE *const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
-	BYTE *const oend_w = oend - WILDCOPY_OVERLENGTH;
-	const BYTE *const iLitEnd = *litPtr + sequence.litLength;
-	const BYTE *match = sequence.match;
-
-	/* check */
-	if (oMatchEnd > oend)
-		return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
-	if (iLitEnd > litLimit)
-		return ERROR(corruption_detected); /* over-read beyond lit buffer */
-	if (oLitEnd > oend_w)
-		return ZSTD_execSequenceLast7(op, oend, sequence, litPtr, litLimit, base, vBase, dictEnd);
-
-	/* copy Literals */
-	ZSTD_copy8(op, *litPtr);
-	if (sequence.litLength > 8)
-		ZSTD_wildcopy(op + 8, (*litPtr) + 8,
-			      sequence.litLength - 8); /* note : since oLitEnd <= oend-WILDCOPY_OVERLENGTH, no risk of overwrite beyond oend */
-	op = oLitEnd;
-	*litPtr = iLitEnd; /* update for next sequence */
-
-	/* copy Match */
-	if (sequence.offset > (size_t)(oLitEnd - base)) {
-		/* offset beyond prefix */
-		if (sequence.offset > (size_t)(oLitEnd - vBase))
-			return ERROR(corruption_detected);
-		if (match + sequence.matchLength <= dictEnd) {
-			memmove(oLitEnd, match, sequence.matchLength);
-			return sequenceLength;
-		}
-		/* span extDict & currPrefixSegment */
-		{
-			size_t const length1 = dictEnd - match;
-			memmove(oLitEnd, match, length1);
-			op = oLitEnd + length1;
-			sequence.matchLength -= length1;
-			match = base;
-			if (op > oend_w || sequence.matchLength < MINMATCH) {
-				U32 i;
-				for (i = 0; i < sequence.matchLength; ++i)
-					op[i] = match[i];
-				return sequenceLength;
-			}
-		}
-	}
-	/* Requirement: op <= oend_w && sequence.matchLength >= MINMATCH */
-
-	/* match within prefix */
-	if (sequence.offset < 8) {
-		/* close range match, overlap */
-		static const U32 dec32table[] = {0, 1, 2, 1, 4, 4, 4, 4};   /* added */
-		static const int dec64table[] = {8, 8, 8, 7, 8, 9, 10, 11}; /* subtracted */
-		int const sub2 = dec64table[sequence.offset];
-		op[0] = match[0];
-		op[1] = match[1];
-		op[2] = match[2];
-		op[3] = match[3];
-		match += dec32table[sequence.offset];
-		ZSTD_copy4(op + 4, match);
-		match -= sub2;
-	} else {
-		ZSTD_copy8(op, match);
-	}
-	op += 8;
-	match += 8;
-
-	if (oMatchEnd > oend - (16 - MINMATCH)) {
-		if (op < oend_w) {
-			ZSTD_wildcopy(op, match, oend_w - op);
-			match += oend_w - op;
-			op = oend_w;
-		}
-		while (op < oMatchEnd)
-			*op++ = *match++;
-	} else {
-		ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength - 8); /* works even if matchLength < 8 */
-	}
-	return sequenceLength;
-}
-
-static size_t ZSTD_decompressSequencesLong(ZSTD_DCtx *dctx, void *dst, size_t maxDstSize, const void *seqStart, size_t seqSize)
-{
-	const BYTE *ip = (const BYTE *)seqStart;
-	const BYTE *const iend = ip + seqSize;
-	BYTE *const ostart = (BYTE * const)dst;
-	BYTE *const oend = ostart + maxDstSize;
-	BYTE *op = ostart;
-	const BYTE *litPtr = dctx->litPtr;
-	const BYTE *const litEnd = litPtr + dctx->litSize;
-	const BYTE *const base = (const BYTE *)(dctx->base);
-	const BYTE *const vBase = (const BYTE *)(dctx->vBase);
-	const BYTE *const dictEnd = (const BYTE *)(dctx->dictEnd);
-	unsigned const windowSize = dctx->fParams.windowSize;
-	int nbSeq;
-
-	/* Build Decoding Tables */
-	{
-		size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, seqSize);
-		if (ZSTD_isError(seqHSize))
-			return seqHSize;
-		ip += seqHSize;
-	}
-
-	/* Regen sequences */
-	if (nbSeq) {
-#define STORED_SEQS 4
-#define STOSEQ_MASK (STORED_SEQS - 1)
-#define ADVANCED_SEQS 4
-		seq_t *sequences = (seq_t *)dctx->entropy.workspace;
-		int const seqAdvance = MIN(nbSeq, ADVANCED_SEQS);
-		seqState_t seqState;
-		int seqNb;
-		ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.workspace) >= sizeof(seq_t) * STORED_SEQS);
-		dctx->fseEntropy = 1;
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				seqState.prevOffset[i] = dctx->entropy.rep[i];
-		}
-		seqState.base = base;
-		seqState.pos = (size_t)(op - base);
-		seqState.gotoDict = (uPtrDiff)dictEnd - (uPtrDiff)base; /* cast to avoid undefined behaviour */
-		CHECK_E(BIT_initDStream(&seqState.DStream, ip, iend - ip), corruption_detected);
-		FSE_initDState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
-		FSE_initDState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
-		FSE_initDState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
-
-		/* prepare in advance */
-		for (seqNb = 0; (BIT_reloadDStream(&seqState.DStream) <= BIT_DStream_completed) && seqNb < seqAdvance; seqNb++) {
-			sequences[seqNb] = ZSTD_decodeSequenceLong(&seqState, windowSize);
-		}
-		if (seqNb < seqAdvance)
-			return ERROR(corruption_detected);
-
-		/* decode and decompress */
-		for (; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && seqNb < nbSeq; seqNb++) {
-			seq_t const sequence = ZSTD_decodeSequenceLong(&seqState, windowSize);
-			size_t const oneSeqSize =
-			    ZSTD_execSequenceLong(op, oend, sequences[(seqNb - ADVANCED_SEQS) & STOSEQ_MASK], &litPtr, litEnd, base, vBase, dictEnd);
-			if (ZSTD_isError(oneSeqSize))
-				return oneSeqSize;
-			ZSTD_PREFETCH(sequence.match);
-			sequences[seqNb & STOSEQ_MASK] = sequence;
-			op += oneSeqSize;
-		}
-		if (seqNb < nbSeq)
-			return ERROR(corruption_detected);
-
-		/* finish queue */
-		seqNb -= seqAdvance;
-		for (; seqNb < nbSeq; seqNb++) {
-			size_t const oneSeqSize = ZSTD_execSequenceLong(op, oend, sequences[seqNb & STOSEQ_MASK], &litPtr, litEnd, base, vBase, dictEnd);
-			if (ZSTD_isError(oneSeqSize))
-				return oneSeqSize;
-			op += oneSeqSize;
-		}
-
-		/* save reps for next block */
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]);
-		}
-	}
-
-	/* last literal segment */
-	{
-		size_t const lastLLSize = litEnd - litPtr;
-		if (lastLLSize > (size_t)(oend - op))
-			return ERROR(dstSize_tooSmall);
-		memcpy(op, litPtr, lastLLSize);
-		op += lastLLSize;
-	}
-
-	return op - ostart;
-}
-
-static size_t ZSTD_decompressBlock_internal(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{ /* blockType == blockCompressed */
-	const BYTE *ip = (const BYTE *)src;
-
-	if (srcSize >= ZSTD_BLOCKSIZE_ABSOLUTEMAX)
-		return ERROR(srcSize_wrong);
-
-	/* Decode literals section */
-	{
-		size_t const litCSize = ZSTD_decodeLiteralsBlock(dctx, src, srcSize);
-		if (ZSTD_isError(litCSize))
-			return litCSize;
-		ip += litCSize;
-		srcSize -= litCSize;
-	}
-	if (sizeof(size_t) > 4) /* do not enable prefetching on 32-bits x86, as it's performance detrimental */
-				/* likely because of register pressure */
-				/* if that's the correct cause, then 32-bits ARM should be affected differently */
-				/* it would be good to test this on ARM real hardware, to see if prefetch version improves speed */
-		if (dctx->fParams.windowSize > (1 << 23))
-			return ZSTD_decompressSequencesLong(dctx, dst, dstCapacity, ip, srcSize);
-	return ZSTD_decompressSequences(dctx, dst, dstCapacity, ip, srcSize);
-}
-
-static void ZSTD_checkContinuity(ZSTD_DCtx *dctx, const void *dst)
-{
-	if (dst != dctx->previousDstEnd) { /* not contiguous */
-		dctx->dictEnd = dctx->previousDstEnd;
-		dctx->vBase = (const char *)dst - ((const char *)(dctx->previousDstEnd) - (const char *)(dctx->base));
-		dctx->base = dst;
-		dctx->previousDstEnd = dst;
-	}
-}
-
-size_t ZSTD_decompressBlock(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	size_t dSize;
-	ZSTD_checkContinuity(dctx, dst);
-	dSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize);
-	dctx->previousDstEnd = (char *)dst + dSize;
-	return dSize;
-}
-
-/** ZSTD_insertBlock() :
-	insert `src` block into `dctx` history. Useful to track uncompressed blocks. */
-size_t ZSTD_insertBlock(ZSTD_DCtx *dctx, const void *blockStart, size_t blockSize)
-{
-	ZSTD_checkContinuity(dctx, blockStart);
-	dctx->previousDstEnd = (const char *)blockStart + blockSize;
-	return blockSize;
-}
-
-static size_t ZSTD_generateNxBytes(void *dst, size_t dstCapacity, BYTE byte, size_t length)
-{
-	if (length > dstCapacity)
-		return ERROR(dstSize_tooSmall);
-	memset(dst, byte, length);
-	return length;
-}
-
-/** ZSTD_findFrameCompressedSize() :
- *  compatible with legacy mode
- *  `src` must point to the start of a ZSTD frame, ZSTD legacy frame, or skippable frame
- *  `srcSize` must be at least as large as the frame contained
- *  @return : the compressed size of the frame starting at `src` */
-size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize)
-{
-	if (srcSize >= ZSTD_skippableHeaderSize && (ZSTD_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
-		return ZSTD_skippableHeaderSize + ZSTD_readLE32((const BYTE *)src + 4);
-	} else {
-		const BYTE *ip = (const BYTE *)src;
-		const BYTE *const ipstart = ip;
-		size_t remainingSize = srcSize;
-		ZSTD_frameParams fParams;
-
-		size_t const headerSize = ZSTD_frameHeaderSize(ip, remainingSize);
-		if (ZSTD_isError(headerSize))
-			return headerSize;
-
-		/* Frame Header */
-		{
-			size_t const ret = ZSTD_getFrameParams(&fParams, ip, remainingSize);
-			if (ZSTD_isError(ret))
-				return ret;
-			if (ret > 0)
-				return ERROR(srcSize_wrong);
-		}
-
-		ip += headerSize;
-		remainingSize -= headerSize;
-
-		/* Loop on each block */
-		while (1) {
-			blockProperties_t blockProperties;
-			size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
-			if (ZSTD_isError(cBlockSize))
-				return cBlockSize;
-
-			if (ZSTD_blockHeaderSize + cBlockSize > remainingSize)
-				return ERROR(srcSize_wrong);
-
-			ip += ZSTD_blockHeaderSize + cBlockSize;
-			remainingSize -= ZSTD_blockHeaderSize + cBlockSize;
-
-			if (blockProperties.lastBlock)
-				break;
-		}
-
-		if (fParams.checksumFlag) { /* Frame content checksum */
-			if (remainingSize < 4)
-				return ERROR(srcSize_wrong);
-			ip += 4;
-			remainingSize -= 4;
-		}
-
-		return ip - ipstart;
-	}
-}
-
-/*! ZSTD_decompressFrame() :
-*   @dctx must be properly initialized */
-static size_t ZSTD_decompressFrame(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void **srcPtr, size_t *srcSizePtr)
-{
-	const BYTE *ip = (const BYTE *)(*srcPtr);
-	BYTE *const ostart = (BYTE * const)dst;
-	BYTE *const oend = ostart + dstCapacity;
-	BYTE *op = ostart;
-	size_t remainingSize = *srcSizePtr;
-
-	/* check */
-	if (remainingSize < ZSTD_frameHeaderSize_min + ZSTD_blockHeaderSize)
-		return ERROR(srcSize_wrong);
-
-	/* Frame Header */
-	{
-		size_t const frameHeaderSize = ZSTD_frameHeaderSize(ip, ZSTD_frameHeaderSize_prefix);
-		if (ZSTD_isError(frameHeaderSize))
-			return frameHeaderSize;
-		if (remainingSize < frameHeaderSize + ZSTD_blockHeaderSize)
-			return ERROR(srcSize_wrong);
-		CHECK_F(ZSTD_decodeFrameHeader(dctx, ip, frameHeaderSize));
-		ip += frameHeaderSize;
-		remainingSize -= frameHeaderSize;
-	}
-
-	/* Loop on each block */
-	while (1) {
-		size_t decodedSize;
-		blockProperties_t blockProperties;
-		size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
-		if (ZSTD_isError(cBlockSize))
-			return cBlockSize;
-
-		ip += ZSTD_blockHeaderSize;
-		remainingSize -= ZSTD_blockHeaderSize;
-		if (cBlockSize > remainingSize)
-			return ERROR(srcSize_wrong);
-
-		switch (blockProperties.blockType) {
-		case bt_compressed: decodedSize = ZSTD_decompressBlock_internal(dctx, op, oend - op, ip, cBlockSize); break;
-		case bt_raw: decodedSize = ZSTD_copyRawBlock(op, oend - op, ip, cBlockSize); break;
-		case bt_rle: decodedSize = ZSTD_generateNxBytes(op, oend - op, *ip, blockProperties.origSize); break;
-		case bt_reserved:
-		default: return ERROR(corruption_detected);
-		}
-
-		if (ZSTD_isError(decodedSize))
-			return decodedSize;
-		if (dctx->fParams.checksumFlag)
-			xxh64_update(&dctx->xxhState, op, decodedSize);
-		op += decodedSize;
-		ip += cBlockSize;
-		remainingSize -= cBlockSize;
-		if (blockProperties.lastBlock)
-			break;
-	}
-
-	if (dctx->fParams.checksumFlag) { /* Frame content checksum verification */
-		U32 const checkCalc = (U32)xxh64_digest(&dctx->xxhState);
-		U32 checkRead;
-		if (remainingSize < 4)
-			return ERROR(checksum_wrong);
-		checkRead = ZSTD_readLE32(ip);
-		if (checkRead != checkCalc)
-			return ERROR(checksum_wrong);
-		ip += 4;
-		remainingSize -= 4;
-	}
-
-	/* Allow caller to get size read */
-	*srcPtr = ip;
-	*srcSizePtr = remainingSize;
-	return op - ostart;
-}
-
-static const void *ZSTD_DDictDictContent(const ZSTD_DDict *ddict);
-static size_t ZSTD_DDictDictSize(const ZSTD_DDict *ddict);
-
-static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize,
-					const ZSTD_DDict *ddict)
-{
-	void *const dststart = dst;
-
-	if (ddict) {
-		if (dict) {
-			/* programmer error, these two cases should be mutually exclusive */
-			return ERROR(GENERIC);
-		}
-
-		dict = ZSTD_DDictDictContent(ddict);
-		dictSize = ZSTD_DDictDictSize(ddict);
-	}
-
-	while (srcSize >= ZSTD_frameHeaderSize_prefix) {
-		U32 magicNumber;
-
-		magicNumber = ZSTD_readLE32(src);
-		if (magicNumber != ZSTD_MAGICNUMBER) {
-			if ((magicNumber & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
-				size_t skippableSize;
-				if (srcSize < ZSTD_skippableHeaderSize)
-					return ERROR(srcSize_wrong);
-				skippableSize = ZSTD_readLE32((const BYTE *)src + 4) + ZSTD_skippableHeaderSize;
-				if (srcSize < skippableSize) {
-					return ERROR(srcSize_wrong);
-				}
-
-				src = (const BYTE *)src + skippableSize;
-				srcSize -= skippableSize;
-				continue;
-			} else {
-				return ERROR(prefix_unknown);
-			}
-		}
-
-		if (ddict) {
-			/* we were called from ZSTD_decompress_usingDDict */
-			ZSTD_refDDict(dctx, ddict);
-		} else {
-			/* this will initialize correctly with no dict if dict == NULL, so
-			 * use this in all cases but ddict */
-			CHECK_F(ZSTD_decompressBegin_usingDict(dctx, dict, dictSize));
-		}
-		ZSTD_checkContinuity(dctx, dst);
-
-		{
-			const size_t res = ZSTD_decompressFrame(dctx, dst, dstCapacity, &src, &srcSize);
-			if (ZSTD_isError(res))
-				return res;
-			/* don't need to bounds check this, ZSTD_decompressFrame will have
-			 * already */
-			dst = (BYTE *)dst + res;
-			dstCapacity -= res;
-		}
-	}
-
-	if (srcSize)
-		return ERROR(srcSize_wrong); /* input not entirely consumed */
-
-	return (BYTE *)dst - (BYTE *)dststart;
-}
-
-size_t ZSTD_decompress_usingDict(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize)
-{
-	return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, dict, dictSize, NULL);
-}
-
-size_t ZSTD_decompressDCtx(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	return ZSTD_decompress_usingDict(dctx, dst, dstCapacity, src, srcSize, NULL, 0);
-}
-
-/*-**************************************
-*   Advanced Streaming Decompression API
-*   Bufferless and synchronous
-****************************************/
-size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx *dctx) { return dctx->expected; }
-
-ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx *dctx)
-{
-	switch (dctx->stage) {
-	default: /* should not happen */
-	case ZSTDds_getFrameHeaderSize:
-	case ZSTDds_decodeFrameHeader: return ZSTDnit_frameHeader;
-	case ZSTDds_decodeBlockHeader: return ZSTDnit_blockHeader;
-	case ZSTDds_decompressBlock: return ZSTDnit_block;
-	case ZSTDds_decompressLastBlock: return ZSTDnit_lastBlock;
-	case ZSTDds_checkChecksum: return ZSTDnit_checksum;
-	case ZSTDds_decodeSkippableHeader:
-	case ZSTDds_skipFrame: return ZSTDnit_skippableFrame;
-	}
-}
-
-int ZSTD_isSkipFrame(ZSTD_DCtx *dctx) { return dctx->stage == ZSTDds_skipFrame; } /* for zbuff */
-
-/** ZSTD_decompressContinue() :
-*   @return : nb of bytes generated into `dst` (necessarily <= `dstCapacity)
-*             or an error code, which can be tested using ZSTD_isError() */
-size_t ZSTD_decompressContinue(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	/* Sanity check */
-	if (srcSize != dctx->expected)
-		return ERROR(srcSize_wrong);
-	if (dstCapacity)
-		ZSTD_checkContinuity(dctx, dst);
-
-	switch (dctx->stage) {
-	case ZSTDds_getFrameHeaderSize:
-		if (srcSize != ZSTD_frameHeaderSize_prefix)
-			return ERROR(srcSize_wrong);					/* impossible */
-		if ((ZSTD_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) { /* skippable frame */
-			memcpy(dctx->headerBuffer, src, ZSTD_frameHeaderSize_prefix);
-			dctx->expected = ZSTD_skippableHeaderSize - ZSTD_frameHeaderSize_prefix; /* magic number + skippable frame length */
-			dctx->stage = ZSTDds_decodeSkippableHeader;
-			return 0;
-		}
-		dctx->headerSize = ZSTD_frameHeaderSize(src, ZSTD_frameHeaderSize_prefix);
-		if (ZSTD_isError(dctx->headerSize))
-			return dctx->headerSize;
-		memcpy(dctx->headerBuffer, src, ZSTD_frameHeaderSize_prefix);
-		if (dctx->headerSize > ZSTD_frameHeaderSize_prefix) {
-			dctx->expected = dctx->headerSize - ZSTD_frameHeaderSize_prefix;
-			dctx->stage = ZSTDds_decodeFrameHeader;
-			return 0;
-		}
-		dctx->expected = 0; /* not necessary to copy more */
-
-	case ZSTDds_decodeFrameHeader:
-		memcpy(dctx->headerBuffer + ZSTD_frameHeaderSize_prefix, src, dctx->expected);
-		CHECK_F(ZSTD_decodeFrameHeader(dctx, dctx->headerBuffer, dctx->headerSize));
-		dctx->expected = ZSTD_blockHeaderSize;
-		dctx->stage = ZSTDds_decodeBlockHeader;
-		return 0;
-
-	case ZSTDds_decodeBlockHeader: {
-		blockProperties_t bp;
-		size_t const cBlockSize = ZSTD_getcBlockSize(src, ZSTD_blockHeaderSize, &bp);
-		if (ZSTD_isError(cBlockSize))
-			return cBlockSize;
-		dctx->expected = cBlockSize;
-		dctx->bType = bp.blockType;
-		dctx->rleSize = bp.origSize;
-		if (cBlockSize) {
-			dctx->stage = bp.lastBlock ? ZSTDds_decompressLastBlock : ZSTDds_decompressBlock;
-			return 0;
-		}
-		/* empty block */
-		if (bp.lastBlock) {
-			if (dctx->fParams.checksumFlag) {
-				dctx->expected = 4;
-				dctx->stage = ZSTDds_checkChecksum;
-			} else {
-				dctx->expected = 0; /* end of frame */
-				dctx->stage = ZSTDds_getFrameHeaderSize;
-			}
-		} else {
-			dctx->expected = 3; /* go directly to next header */
-			dctx->stage = ZSTDds_decodeBlockHeader;
-		}
-		return 0;
-	}
-	case ZSTDds_decompressLastBlock:
-	case ZSTDds_decompressBlock: {
-		size_t rSize;
-		switch (dctx->bType) {
-		case bt_compressed: rSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize); break;
-		case bt_raw: rSize = ZSTD_copyRawBlock(dst, dstCapacity, src, srcSize); break;
-		case bt_rle: rSize = ZSTD_setRleBlock(dst, dstCapacity, src, srcSize, dctx->rleSize); break;
-		case bt_reserved: /* should never happen */
-		default: return ERROR(corruption_detected);
-		}
-		if (ZSTD_isError(rSize))
-			return rSize;
-		if (dctx->fParams.checksumFlag)
-			xxh64_update(&dctx->xxhState, dst, rSize);
-
-		if (dctx->stage == ZSTDds_decompressLastBlock) { /* end of frame */
-			if (dctx->fParams.checksumFlag) {	/* another round for frame checksum */
-				dctx->expected = 4;
-				dctx->stage = ZSTDds_checkChecksum;
-			} else {
-				dctx->expected = 0; /* ends here */
-				dctx->stage = ZSTDds_getFrameHeaderSize;
-			}
-		} else {
-			dctx->stage = ZSTDds_decodeBlockHeader;
-			dctx->expected = ZSTD_blockHeaderSize;
-			dctx->previousDstEnd = (char *)dst + rSize;
-		}
-		return rSize;
-	}
-	case ZSTDds_checkChecksum: {
-		U32 const h32 = (U32)xxh64_digest(&dctx->xxhState);
-		U32 const check32 = ZSTD_readLE32(src); /* srcSize == 4, guaranteed by dctx->expected */
-		if (check32 != h32)
-			return ERROR(checksum_wrong);
-		dctx->expected = 0;
-		dctx->stage = ZSTDds_getFrameHeaderSize;
-		return 0;
-	}
-	case ZSTDds_decodeSkippableHeader: {
-		memcpy(dctx->headerBuffer + ZSTD_frameHeaderSize_prefix, src, dctx->expected);
-		dctx->expected = ZSTD_readLE32(dctx->headerBuffer + 4);
-		dctx->stage = ZSTDds_skipFrame;
-		return 0;
-	}
-	case ZSTDds_skipFrame: {
-		dctx->expected = 0;
-		dctx->stage = ZSTDds_getFrameHeaderSize;
-		return 0;
-	}
-	default:
-		return ERROR(GENERIC); /* impossible */
-	}
-}
-
-static size_t ZSTD_refDictContent(ZSTD_DCtx *dctx, const void *dict, size_t dictSize)
-{
-	dctx->dictEnd = dctx->previousDstEnd;
-	dctx->vBase = (const char *)dict - ((const char *)(dctx->previousDstEnd) - (const char *)(dctx->base));
-	dctx->base = dict;
-	dctx->previousDstEnd = (const char *)dict + dictSize;
-	return 0;
-}
-
-/* ZSTD_loadEntropy() :
- * dict : must point at beginning of a valid zstd dictionary
- * @return : size of entropy tables read */
-static size_t ZSTD_loadEntropy(ZSTD_entropyTables_t *entropy, const void *const dict, size_t const dictSize)
-{
-	const BYTE *dictPtr = (const BYTE *)dict;
-	const BYTE *const dictEnd = dictPtr + dictSize;
-
-	if (dictSize <= 8)
-		return ERROR(dictionary_corrupted);
-	dictPtr += 8; /* skip header = magic + dictID */
-
-	{
-		size_t const hSize = HUF_readDTableX4_wksp(entropy->hufTable, dictPtr, dictEnd - dictPtr, entropy->workspace, sizeof(entropy->workspace));
-		if (HUF_isError(hSize))
-			return ERROR(dictionary_corrupted);
-		dictPtr += hSize;
-	}
-
-	{
-		short offcodeNCount[MaxOff + 1];
-		U32 offcodeMaxValue = MaxOff, offcodeLog;
-		size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, dictEnd - dictPtr);
-		if (FSE_isError(offcodeHeaderSize))
-			return ERROR(dictionary_corrupted);
-		if (offcodeLog > OffFSELog)
-			return ERROR(dictionary_corrupted);
-		CHECK_E(FSE_buildDTable_wksp(entropy->OFTable, offcodeNCount, offcodeMaxValue, offcodeLog, entropy->workspace, sizeof(entropy->workspace)), dictionary_corrupted);
-		dictPtr += offcodeHeaderSize;
-	}
-
-	{
-		short matchlengthNCount[MaxML + 1];
-		unsigned matchlengthMaxValue = MaxML, matchlengthLog;
-		size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd - dictPtr);
-		if (FSE_isError(matchlengthHeaderSize))
-			return ERROR(dictionary_corrupted);
-		if (matchlengthLog > MLFSELog)
-			return ERROR(dictionary_corrupted);
-		CHECK_E(FSE_buildDTable_wksp(entropy->MLTable, matchlengthNCount, matchlengthMaxValue, matchlengthLog, entropy->workspace, sizeof(entropy->workspace)), dictionary_corrupted);
-		dictPtr += matchlengthHeaderSize;
-	}
-
-	{
-		short litlengthNCount[MaxLL + 1];
-		unsigned litlengthMaxValue = MaxLL, litlengthLog;
-		size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd - dictPtr);
-		if (FSE_isError(litlengthHeaderSize))
-			return ERROR(dictionary_corrupted);
-		if (litlengthLog > LLFSELog)
-			return ERROR(dictionary_corrupted);
-		CHECK_E(FSE_buildDTable_wksp(entropy->LLTable, litlengthNCount, litlengthMaxValue, litlengthLog, entropy->workspace, sizeof(entropy->workspace)), dictionary_corrupted);
-		dictPtr += litlengthHeaderSize;
-	}
-
-	if (dictPtr + 12 > dictEnd)
-		return ERROR(dictionary_corrupted);
-	{
-		int i;
-		size_t const dictContentSize = (size_t)(dictEnd - (dictPtr + 12));
-		for (i = 0; i < 3; i++) {
-			U32 const rep = ZSTD_readLE32(dictPtr);
-			dictPtr += 4;
-			if (rep == 0 || rep >= dictContentSize)
-				return ERROR(dictionary_corrupted);
-			entropy->rep[i] = rep;
-		}
-	}
-
-	return dictPtr - (const BYTE *)dict;
-}
-
-static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx *dctx, const void *dict, size_t dictSize)
-{
-	if (dictSize < 8)
-		return ZSTD_refDictContent(dctx, dict, dictSize);
-	{
-		U32 const magic = ZSTD_readLE32(dict);
-		if (magic != ZSTD_DICT_MAGIC) {
-			return ZSTD_refDictContent(dctx, dict, dictSize); /* pure content mode */
-		}
-	}
-	dctx->dictID = ZSTD_readLE32((const char *)dict + 4);
-
-	/* load entropy tables */
-	{
-		size_t const eSize = ZSTD_loadEntropy(&dctx->entropy, dict, dictSize);
-		if (ZSTD_isError(eSize))
-			return ERROR(dictionary_corrupted);
-		dict = (const char *)dict + eSize;
-		dictSize -= eSize;
-	}
-	dctx->litEntropy = dctx->fseEntropy = 1;
-
-	/* reference dictionary content */
-	return ZSTD_refDictContent(dctx, dict, dictSize);
-}
-
-size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx *dctx, const void *dict, size_t dictSize)
-{
-	CHECK_F(ZSTD_decompressBegin(dctx));
-	if (dict && dictSize)
-		CHECK_E(ZSTD_decompress_insertDictionary(dctx, dict, dictSize), dictionary_corrupted);
-	return 0;
-}
-
-/* ======   ZSTD_DDict   ====== */
-
-struct ZSTD_DDict_s {
-	void *dictBuffer;
-	const void *dictContent;
-	size_t dictSize;
-	ZSTD_entropyTables_t entropy;
-	U32 dictID;
-	U32 entropyPresent;
-	ZSTD_customMem cMem;
-}; /* typedef'd to ZSTD_DDict within "zstd.h" */
-
-size_t ZSTD_DDictWorkspaceBound(void) { return ZSTD_ALIGN(sizeof(ZSTD_stack)) + ZSTD_ALIGN(sizeof(ZSTD_DDict)); }
-
-static const void *ZSTD_DDictDictContent(const ZSTD_DDict *ddict) { return ddict->dictContent; }
-
-static size_t ZSTD_DDictDictSize(const ZSTD_DDict *ddict) { return ddict->dictSize; }
-
-static void ZSTD_refDDict(ZSTD_DCtx *dstDCtx, const ZSTD_DDict *ddict)
-{
-	ZSTD_decompressBegin(dstDCtx); /* init */
-	if (ddict) {		       /* support refDDict on NULL */
-		dstDCtx->dictID = ddict->dictID;
-		dstDCtx->base = ddict->dictContent;
-		dstDCtx->vBase = ddict->dictContent;
-		dstDCtx->dictEnd = (const BYTE *)ddict->dictContent + ddict->dictSize;
-		dstDCtx->previousDstEnd = dstDCtx->dictEnd;
-		if (ddict->entropyPresent) {
-			dstDCtx->litEntropy = 1;
-			dstDCtx->fseEntropy = 1;
-			dstDCtx->LLTptr = ddict->entropy.LLTable;
-			dstDCtx->MLTptr = ddict->entropy.MLTable;
-			dstDCtx->OFTptr = ddict->entropy.OFTable;
-			dstDCtx->HUFptr = ddict->entropy.hufTable;
-			dstDCtx->entropy.rep[0] = ddict->entropy.rep[0];
-			dstDCtx->entropy.rep[1] = ddict->entropy.rep[1];
-			dstDCtx->entropy.rep[2] = ddict->entropy.rep[2];
-		} else {
-			dstDCtx->litEntropy = 0;
-			dstDCtx->fseEntropy = 0;
-		}
-	}
-}
-
-static size_t ZSTD_loadEntropy_inDDict(ZSTD_DDict *ddict)
-{
-	ddict->dictID = 0;
-	ddict->entropyPresent = 0;
-	if (ddict->dictSize < 8)
-		return 0;
-	{
-		U32 const magic = ZSTD_readLE32(ddict->dictContent);
-		if (magic != ZSTD_DICT_MAGIC)
-			return 0; /* pure content mode */
-	}
-	ddict->dictID = ZSTD_readLE32((const char *)ddict->dictContent + 4);
-
-	/* load entropy tables */
-	CHECK_E(ZSTD_loadEntropy(&ddict->entropy, ddict->dictContent, ddict->dictSize), dictionary_corrupted);
-	ddict->entropyPresent = 1;
-	return 0;
-}
-
-static ZSTD_DDict *ZSTD_createDDict_advanced(const void *dict, size_t dictSize, unsigned byReference, ZSTD_customMem customMem)
-{
-	if (!customMem.customAlloc || !customMem.customFree)
-		return NULL;
-
-	{
-		ZSTD_DDict *const ddict = (ZSTD_DDict *)ZSTD_malloc(sizeof(ZSTD_DDict), customMem);
-		if (!ddict)
-			return NULL;
-		ddict->cMem = customMem;
-
-		if ((byReference) || (!dict) || (!dictSize)) {
-			ddict->dictBuffer = NULL;
-			ddict->dictContent = dict;
-		} else {
-			void *const internalBuffer = ZSTD_malloc(dictSize, customMem);
-			if (!internalBuffer) {
-				ZSTD_freeDDict(ddict);
-				return NULL;
-			}
-			memcpy(internalBuffer, dict, dictSize);
-			ddict->dictBuffer = internalBuffer;
-			ddict->dictContent = internalBuffer;
-		}
-		ddict->dictSize = dictSize;
-		ddict->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001); /* cover both little and big endian */
-		/* parse dictionary content */
-		{
-			size_t const errorCode = ZSTD_loadEntropy_inDDict(ddict);
-			if (ZSTD_isError(errorCode)) {
-				ZSTD_freeDDict(ddict);
-				return NULL;
-			}
-		}
-
-		return ddict;
-	}
-}
-
-/*! ZSTD_initDDict() :
-*   Create a digested dictionary, to start decompression without startup delay.
-*   `dict` content is copied inside DDict.
-*   Consequently, `dict` can be released after `ZSTD_DDict` creation */
-ZSTD_DDict *ZSTD_initDDict(const void *dict, size_t dictSize, void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
-	return ZSTD_createDDict_advanced(dict, dictSize, 1, stackMem);
-}
-
-size_t ZSTD_freeDDict(ZSTD_DDict *ddict)
-{
-	if (ddict == NULL)
-		return 0; /* support free on NULL */
-	{
-		ZSTD_customMem const cMem = ddict->cMem;
-		ZSTD_free(ddict->dictBuffer, cMem);
-		ZSTD_free(ddict, cMem);
-		return 0;
-	}
-}
-
-/*! ZSTD_getDictID_fromDict() :
- *  Provides the dictID stored within dictionary.
- *  if @return == 0, the dictionary is not conformant with Zstandard specification.
- *  It can still be loaded, but as a content-only dictionary. */
-unsigned ZSTD_getDictID_fromDict(const void *dict, size_t dictSize)
-{
-	if (dictSize < 8)
-		return 0;
-	if (ZSTD_readLE32(dict) != ZSTD_DICT_MAGIC)
-		return 0;
-	return ZSTD_readLE32((const char *)dict + 4);
-}
-
-/*! ZSTD_getDictID_fromDDict() :
- *  Provides the dictID of the dictionary loaded into `ddict`.
- *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
- *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
-unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict *ddict)
-{
-	if (ddict == NULL)
-		return 0;
-	return ZSTD_getDictID_fromDict(ddict->dictContent, ddict->dictSize);
-}
-
-/*! ZSTD_getDictID_fromFrame() :
- *  Provides the dictID required to decompressed the frame stored within `src`.
- *  If @return == 0, the dictID could not be decoded.
- *  This could for one of the following reasons :
- *  - The frame does not require a dictionary to be decoded (most common case).
- *  - The frame was built with dictID intentionally removed. Whatever dictionary is necessary is a hidden information.
- *    Note : this use case also happens when using a non-conformant dictionary.
- *  - `srcSize` is too small, and as a result, the frame header could not be decoded (only possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`).
- *  - This is not a Zstandard frame.
- *  When identifying the exact failure cause, it's possible to used ZSTD_getFrameParams(), which will provide a more precise error code. */
-unsigned ZSTD_getDictID_fromFrame(const void *src, size_t srcSize)
-{
-	ZSTD_frameParams zfp = {0, 0, 0, 0};
-	size_t const hError = ZSTD_getFrameParams(&zfp, src, srcSize);
-	if (ZSTD_isError(hError))
-		return 0;
-	return zfp.dictID;
-}
-
-/*! ZSTD_decompress_usingDDict() :
-*   Decompression using a pre-digested Dictionary
-*   Use dictionary without significant overhead. */
-size_t ZSTD_decompress_usingDDict(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const ZSTD_DDict *ddict)
-{
-	/* pass content and size in case legacy frames are encountered */
-	return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, NULL, 0, ddict);
-}
-
-/*=====================================
-*   Streaming decompression
-*====================================*/
-
-typedef enum { zdss_init, zdss_loadHeader, zdss_read, zdss_load, zdss_flush } ZSTD_dStreamStage;
-
-/* *** Resource management *** */
-struct ZSTD_DStream_s {
-	ZSTD_DCtx *dctx;
-	ZSTD_DDict *ddictLocal;
-	const ZSTD_DDict *ddict;
-	ZSTD_frameParams fParams;
-	ZSTD_dStreamStage stage;
-	char *inBuff;
-	size_t inBuffSize;
-	size_t inPos;
-	size_t maxWindowSize;
-	char *outBuff;
-	size_t outBuffSize;
-	size_t outStart;
-	size_t outEnd;
-	size_t blockSize;
-	BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX]; /* tmp buffer to store frame header */
-	size_t lhSize;
-	ZSTD_customMem customMem;
-	void *legacyContext;
-	U32 previousLegacyVersion;
-	U32 legacyVersion;
-	U32 hostageByte;
-}; /* typedef'd to ZSTD_DStream within "zstd.h" */
-
-size_t ZSTD_DStreamWorkspaceBound(size_t maxWindowSize)
-{
-	size_t const blockSize = MIN(maxWindowSize, ZSTD_BLOCKSIZE_ABSOLUTEMAX);
-	size_t const inBuffSize = blockSize;
-	size_t const outBuffSize = maxWindowSize + blockSize + WILDCOPY_OVERLENGTH * 2;
-	return ZSTD_DCtxWorkspaceBound() + ZSTD_ALIGN(sizeof(ZSTD_DStream)) + ZSTD_ALIGN(inBuffSize) + ZSTD_ALIGN(outBuffSize);
-}
-
-static ZSTD_DStream *ZSTD_createDStream_advanced(ZSTD_customMem customMem)
-{
-	ZSTD_DStream *zds;
-
-	if (!customMem.customAlloc || !customMem.customFree)
-		return NULL;
-
-	zds = (ZSTD_DStream *)ZSTD_malloc(sizeof(ZSTD_DStream), customMem);
-	if (zds == NULL)
-		return NULL;
-	memset(zds, 0, sizeof(ZSTD_DStream));
-	memcpy(&zds->customMem, &customMem, sizeof(ZSTD_customMem));
-	zds->dctx = ZSTD_createDCtx_advanced(customMem);
-	if (zds->dctx == NULL) {
-		ZSTD_freeDStream(zds);
-		return NULL;
-	}
-	zds->stage = zdss_init;
-	zds->maxWindowSize = ZSTD_MAXWINDOWSIZE_DEFAULT;
-	return zds;
-}
-
-ZSTD_DStream *ZSTD_initDStream(size_t maxWindowSize, void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
-	ZSTD_DStream *zds = ZSTD_createDStream_advanced(stackMem);
-	if (!zds) {
-		return NULL;
-	}
-
-	zds->maxWindowSize = maxWindowSize;
-	zds->stage = zdss_loadHeader;
-	zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
-	ZSTD_freeDDict(zds->ddictLocal);
-	zds->ddictLocal = NULL;
-	zds->ddict = zds->ddictLocal;
-	zds->legacyVersion = 0;
-	zds->hostageByte = 0;
-
-	{
-		size_t const blockSize = MIN(zds->maxWindowSize, ZSTD_BLOCKSIZE_ABSOLUTEMAX);
-		size_t const neededOutSize = zds->maxWindowSize + blockSize + WILDCOPY_OVERLENGTH * 2;
-
-		zds->inBuff = (char *)ZSTD_malloc(blockSize, zds->customMem);
-		zds->inBuffSize = blockSize;
-		zds->outBuff = (char *)ZSTD_malloc(neededOutSize, zds->customMem);
-		zds->outBuffSize = neededOutSize;
-		if (zds->inBuff == NULL || zds->outBuff == NULL) {
-			ZSTD_freeDStream(zds);
-			return NULL;
-		}
-	}
-	return zds;
-}
-
-ZSTD_DStream *ZSTD_initDStream_usingDDict(size_t maxWindowSize, const ZSTD_DDict *ddict, void *workspace, size_t workspaceSize)
-{
-	ZSTD_DStream *zds = ZSTD_initDStream(maxWindowSize, workspace, workspaceSize);
-	if (zds) {
-		zds->ddict = ddict;
-	}
-	return zds;
-}
-
-size_t ZSTD_freeDStream(ZSTD_DStream *zds)
-{
-	if (zds == NULL)
-		return 0; /* support free on null */
-	{
-		ZSTD_customMem const cMem = zds->customMem;
-		ZSTD_freeDCtx(zds->dctx);
-		zds->dctx = NULL;
-		ZSTD_freeDDict(zds->ddictLocal);
-		zds->ddictLocal = NULL;
-		ZSTD_free(zds->inBuff, cMem);
-		zds->inBuff = NULL;
-		ZSTD_free(zds->outBuff, cMem);
-		zds->outBuff = NULL;
-		ZSTD_free(zds, cMem);
-		return 0;
-	}
-}
-
-/* *** Initialization *** */
-
-size_t ZSTD_DStreamInSize(void) { return ZSTD_BLOCKSIZE_ABSOLUTEMAX + ZSTD_blockHeaderSize; }
-size_t ZSTD_DStreamOutSize(void) { return ZSTD_BLOCKSIZE_ABSOLUTEMAX; }
-
-size_t ZSTD_resetDStream(ZSTD_DStream *zds)
-{
-	zds->stage = zdss_loadHeader;
-	zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
-	zds->legacyVersion = 0;
-	zds->hostageByte = 0;
-	return ZSTD_frameHeaderSize_prefix;
-}
-
-/* *****   Decompression   ***** */
-
-ZSTD_STATIC size_t ZSTD_limitCopy(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	size_t const length = MIN(dstCapacity, srcSize);
-	memcpy(dst, src, length);
-	return length;
-}
-
-size_t ZSTD_decompressStream(ZSTD_DStream *zds, ZSTD_outBuffer *output, ZSTD_inBuffer *input)
-{
-	const char *const istart = (const char *)(input->src) + input->pos;
-	const char *const iend = (const char *)(input->src) + input->size;
-	const char *ip = istart;
-	char *const ostart = (char *)(output->dst) + output->pos;
-	char *const oend = (char *)(output->dst) + output->size;
-	char *op = ostart;
-	U32 someMoreWork = 1;
-
-	while (someMoreWork) {
-		switch (zds->stage) {
-		case zdss_init:
-			ZSTD_resetDStream(zds); /* transparent reset on starting decoding a new frame */
-						/* fall-through */
-
-		case zdss_loadHeader: {
-			size_t const hSize = ZSTD_getFrameParams(&zds->fParams, zds->headerBuffer, zds->lhSize);
-			if (ZSTD_isError(hSize))
-				return hSize;
-			if (hSize != 0) {				   /* need more input */
-				size_t const toLoad = hSize - zds->lhSize; /* if hSize!=0, hSize > zds->lhSize */
-				if (toLoad > (size_t)(iend - ip)) {	/* not enough input to load full header */
-					memcpy(zds->headerBuffer + zds->lhSize, ip, iend - ip);
-					zds->lhSize += iend - ip;
-					input->pos = input->size;
-					return (MAX(ZSTD_frameHeaderSize_min, hSize) - zds->lhSize) +
-					       ZSTD_blockHeaderSize; /* remaining header bytes + next block header */
-				}
-				memcpy(zds->headerBuffer + zds->lhSize, ip, toLoad);
-				zds->lhSize = hSize;
-				ip += toLoad;
-				break;
-			}
-
-			/* check for single-pass mode opportunity */
-			if (zds->fParams.frameContentSize && zds->fParams.windowSize /* skippable frame if == 0 */
-			    && (U64)(size_t)(oend - op) >= zds->fParams.frameContentSize) {
-				size_t const cSize = ZSTD_findFrameCompressedSize(istart, iend - istart);
-				if (cSize <= (size_t)(iend - istart)) {
-					size_t const decompressedSize = ZSTD_decompress_usingDDict(zds->dctx, op, oend - op, istart, cSize, zds->ddict);
-					if (ZSTD_isError(decompressedSize))
-						return decompressedSize;
-					ip = istart + cSize;
-					op += decompressedSize;
-					zds->dctx->expected = 0;
-					zds->stage = zdss_init;
-					someMoreWork = 0;
-					break;
-				}
-			}
-
-			/* Consume header */
-			ZSTD_refDDict(zds->dctx, zds->ddict);
-			{
-				size_t const h1Size = ZSTD_nextSrcSizeToDecompress(zds->dctx); /* == ZSTD_frameHeaderSize_prefix */
-				CHECK_F(ZSTD_decompressContinue(zds->dctx, NULL, 0, zds->headerBuffer, h1Size));
-				{
-					size_t const h2Size = ZSTD_nextSrcSizeToDecompress(zds->dctx);
-					CHECK_F(ZSTD_decompressContinue(zds->dctx, NULL, 0, zds->headerBuffer + h1Size, h2Size));
-				}
-			}
-
-			zds->fParams.windowSize = MAX(zds->fParams.windowSize, 1U << ZSTD_WINDOWLOG_ABSOLUTEMIN);
-			if (zds->fParams.windowSize > zds->maxWindowSize)
-				return ERROR(frameParameter_windowTooLarge);
-
-			/* Buffers are preallocated, but double check */
-			{
-				size_t const blockSize = MIN(zds->maxWindowSize, ZSTD_BLOCKSIZE_ABSOLUTEMAX);
-				size_t const neededOutSize = zds->maxWindowSize + blockSize + WILDCOPY_OVERLENGTH * 2;
-				if (zds->inBuffSize < blockSize) {
-					return ERROR(GENERIC);
-				}
-				if (zds->outBuffSize < neededOutSize) {
-					return ERROR(GENERIC);
-				}
-				zds->blockSize = blockSize;
-			}
-			zds->stage = zdss_read;
-		}
-		/* pass-through */
-
-		case zdss_read: {
-			size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds->dctx);
-			if (neededInSize == 0) { /* end of frame */
-				zds->stage = zdss_init;
-				someMoreWork = 0;
-				break;
-			}
-			if ((size_t)(iend - ip) >= neededInSize) { /* decode directly from src */
-				const int isSkipFrame = ZSTD_isSkipFrame(zds->dctx);
-				size_t const decodedSize = ZSTD_decompressContinue(zds->dctx, zds->outBuff + zds->outStart,
-										   (isSkipFrame ? 0 : zds->outBuffSize - zds->outStart), ip, neededInSize);
-				if (ZSTD_isError(decodedSize))
-					return decodedSize;
-				ip += neededInSize;
-				if (!decodedSize && !isSkipFrame)
-					break; /* this was just a header */
-				zds->outEnd = zds->outStart + decodedSize;
-				zds->stage = zdss_flush;
-				break;
-			}
-			if (ip == iend) {
-				someMoreWork = 0;
-				break;
-			} /* no more input */
-			zds->stage = zdss_load;
-			/* pass-through */
-		}
-
-		case zdss_load: {
-			size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds->dctx);
-			size_t const toLoad = neededInSize - zds->inPos; /* should always be <= remaining space within inBuff */
-			size_t loadedSize;
-			if (toLoad > zds->inBuffSize - zds->inPos)
-				return ERROR(corruption_detected); /* should never happen */
-			loadedSize = ZSTD_limitCopy(zds->inBuff + zds->inPos, toLoad, ip, iend - ip);
-			ip += loadedSize;
-			zds->inPos += loadedSize;
-			if (loadedSize < toLoad) {
-				someMoreWork = 0;
-				break;
-			} /* not enough input, wait for more */
-
-			/* decode loaded input */
-			{
-				const int isSkipFrame = ZSTD_isSkipFrame(zds->dctx);
-				size_t const decodedSize = ZSTD_decompressContinue(zds->dctx, zds->outBuff + zds->outStart, zds->outBuffSize - zds->outStart,
-										   zds->inBuff, neededInSize);
-				if (ZSTD_isError(decodedSize))
-					return decodedSize;
-				zds->inPos = 0; /* input is consumed */
-				if (!decodedSize && !isSkipFrame) {
-					zds->stage = zdss_read;
-					break;
-				} /* this was just a header */
-				zds->outEnd = zds->outStart + decodedSize;
-				zds->stage = zdss_flush;
-				/* pass-through */
-			}
-		}
-
-		case zdss_flush: {
-			size_t const toFlushSize = zds->outEnd - zds->outStart;
-			size_t const flushedSize = ZSTD_limitCopy(op, oend - op, zds->outBuff + zds->outStart, toFlushSize);
-			op += flushedSize;
-			zds->outStart += flushedSize;
-			if (flushedSize == toFlushSize) { /* flush completed */
-				zds->stage = zdss_read;
-				if (zds->outStart + zds->blockSize > zds->outBuffSize)
-					zds->outStart = zds->outEnd = 0;
-				break;
-			}
-			/* cannot complete flush */
-			someMoreWork = 0;
-			break;
-		}
-		default:
-			return ERROR(GENERIC); /* impossible */
-		}
-	}
-
-	/* result */
-	input->pos += (size_t)(ip - istart);
-	output->pos += (size_t)(op - ostart);
-	{
-		size_t nextSrcSizeHint = ZSTD_nextSrcSizeToDecompress(zds->dctx);
-		if (!nextSrcSizeHint) {			    /* frame fully decoded */
-			if (zds->outEnd == zds->outStart) { /* output fully flushed */
-				if (zds->hostageByte) {
-					if (input->pos >= input->size) {
-						zds->stage = zdss_read;
-						return 1;
-					}	     /* can't release hostage (not present) */
-					input->pos++; /* release hostage */
-				}
-				return 0;
-			}
-			if (!zds->hostageByte) { /* output not fully flushed; keep last byte as hostage; will be released when all output is flushed */
-				input->pos--;    /* note : pos > 0, otherwise, impossible to finish reading last block */
-				zds->hostageByte = 1;
-			}
-			return 1;
-		}
-		nextSrcSizeHint += ZSTD_blockHeaderSize * (ZSTD_nextInputType(zds->dctx) == ZSTDnit_block); /* preload header of next block */
-		if (zds->inPos > nextSrcSizeHint)
-			return ERROR(GENERIC); /* should never happen */
-		nextSrcSizeHint -= zds->inPos; /* already loaded*/
-		return nextSrcSizeHint;
-	}
-}
-
-EXPORT_SYMBOL(ZSTD_DCtxWorkspaceBound);
-EXPORT_SYMBOL(ZSTD_initDCtx);
-EXPORT_SYMBOL(ZSTD_decompressDCtx);
-EXPORT_SYMBOL(ZSTD_decompress_usingDict);
-
-EXPORT_SYMBOL(ZSTD_DDictWorkspaceBound);
-EXPORT_SYMBOL(ZSTD_initDDict);
-EXPORT_SYMBOL(ZSTD_decompress_usingDDict);
-
-EXPORT_SYMBOL(ZSTD_DStreamWorkspaceBound);
-EXPORT_SYMBOL(ZSTD_initDStream);
-EXPORT_SYMBOL(ZSTD_initDStream_usingDDict);
-EXPORT_SYMBOL(ZSTD_resetDStream);
-EXPORT_SYMBOL(ZSTD_decompressStream);
-EXPORT_SYMBOL(ZSTD_DStreamInSize);
-EXPORT_SYMBOL(ZSTD_DStreamOutSize);
-
-EXPORT_SYMBOL(ZSTD_findFrameCompressedSize);
-EXPORT_SYMBOL(ZSTD_getFrameContentSize);
-EXPORT_SYMBOL(ZSTD_findDecompressedSize);
-
-EXPORT_SYMBOL(ZSTD_isFrame);
-EXPORT_SYMBOL(ZSTD_getDictID_fromDict);
-EXPORT_SYMBOL(ZSTD_getDictID_fromDDict);
-EXPORT_SYMBOL(ZSTD_getDictID_fromFrame);
-
-EXPORT_SYMBOL(ZSTD_getFrameParams);
-EXPORT_SYMBOL(ZSTD_decompressBegin);
-EXPORT_SYMBOL(ZSTD_decompressBegin_usingDict);
-EXPORT_SYMBOL(ZSTD_copyDCtx);
-EXPORT_SYMBOL(ZSTD_nextSrcSizeToDecompress);
-EXPORT_SYMBOL(ZSTD_decompressContinue);
-EXPORT_SYMBOL(ZSTD_nextInputType);
-
-EXPORT_SYMBOL(ZSTD_decompressBlock);
-EXPORT_SYMBOL(ZSTD_insertBlock);
-
-MODULE_LICENSE("Dual BSD/GPL");
-MODULE_DESCRIPTION("Zstd Decompressor");
diff --git a/lib/zstd/decompress/huf_decompress.c b/lib/zstd/decompress/huf_decompress.c
index 6526482047dc..5105e59ac04a 100644
--- a/lib/zstd/decompress/huf_decompress.c
+++ b/lib/zstd/decompress/huf_decompress.c
@@ -1,960 +1,1206 @@
-/*
- * Huffman decoder, part of New Generation Entropy library
- * Copyright (C) 2013-2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
+/* ******************************************************************
+ * huff0 huffman decoder,
+ * part of Finite State Entropy library
+ * Copyright (c) Yann Collet, Facebook, Inc.
  *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
+ *  You can contact the author at :
+ *  - FSE+HUF source repository : https://github.com/Cyan4973/FiniteStateEntropy
  *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
 
 /* **************************************************************
-*  Compiler specifics
+*  Dependencies
 ****************************************************************/
-#define FORCE_INLINE static __always_inline
+#include "../common/zstd_deps.h"  /* ZSTD_memcpy, ZSTD_memset */
+#include "../common/compiler.h"
+#include "../common/bitstream.h"  /* BIT_* */
+#include "../common/fse.h"        /* to compress headers */
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include "../common/error_private.h"
 
 /* **************************************************************
-*  Dependencies
+*  Macros
 ****************************************************************/
-#include "bitstream.h" /* BIT_* */
-#include "fse.h"       /* header compression */
-#include "huf.h"
-#include <linux/compiler.h>
-#include <linux/kernel.h>
-#include <linux/string.h> /* memcpy, memset */
+
+/* These two optional macros force the use one way or another of the two
+ * Huffman decompression implementations. You can't force in both directions
+ * at the same time.
+ */
+#if defined(HUF_FORCE_DECOMPRESS_X1) && \
+    defined(HUF_FORCE_DECOMPRESS_X2)
+#error "Cannot force the use of the X1 and X2 decoders at the same time!"
+#endif
+
 
 /* **************************************************************
 *  Error Management
 ****************************************************************/
-#define HUF_STATIC_ASSERT(c)                                   \
-	{                                                      \
-		enum { HUF_static_assert = 1 / (int)(!!(c)) }; \
-	} /* use only *after* variable declarations */
+#define HUF_isError ERR_isError
+
+
+/* **************************************************************
+*  Byte alignment for workSpace management
+****************************************************************/
+#define HUF_ALIGN(x, a)         HUF_ALIGN_MASK((x), (a) - 1)
+#define HUF_ALIGN_MASK(x, mask) (((x) + (mask)) & ~(mask))
+
+
+/* **************************************************************
+*  BMI2 Variant Wrappers
+****************************************************************/
+#if DYNAMIC_BMI2
+
+#define HUF_DGEN(fn)                                                        \
+                                                                            \
+    static size_t fn##_default(                                             \
+                  void* dst,  size_t dstSize,                               \
+            const void* cSrc, size_t cSrcSize,                              \
+            const HUF_DTable* DTable)                                       \
+    {                                                                       \
+        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
+    }                                                                       \
+                                                                            \
+    static TARGET_ATTRIBUTE("bmi2") size_t fn##_bmi2(                       \
+                  void* dst,  size_t dstSize,                               \
+            const void* cSrc, size_t cSrcSize,                              \
+            const HUF_DTable* DTable)                                       \
+    {                                                                       \
+        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
+    }                                                                       \
+                                                                            \
+    static size_t fn(void* dst, size_t dstSize, void const* cSrc,           \
+                     size_t cSrcSize, HUF_DTable const* DTable, int bmi2)   \
+    {                                                                       \
+        if (bmi2) {                                                         \
+            return fn##_bmi2(dst, dstSize, cSrc, cSrcSize, DTable);         \
+        }                                                                   \
+        return fn##_default(dst, dstSize, cSrc, cSrcSize, DTable);          \
+    }
+
+#else
+
+#define HUF_DGEN(fn)                                                        \
+    static size_t fn(void* dst, size_t dstSize, void const* cSrc,           \
+                     size_t cSrcSize, HUF_DTable const* DTable, int bmi2)   \
+    {                                                                       \
+        (void)bmi2;                                                         \
+        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
+    }
+
+#endif
+
 
 /*-***************************/
 /*  generic DTableDesc       */
 /*-***************************/
+typedef struct { BYTE maxTableLog; BYTE tableType; BYTE tableLog; BYTE reserved; } DTableDesc;
 
-typedef struct {
-	BYTE maxTableLog;
-	BYTE tableType;
-	BYTE tableLog;
-	BYTE reserved;
-} DTableDesc;
-
-static DTableDesc HUF_getDTableDesc(const HUF_DTable *table)
+static DTableDesc HUF_getDTableDesc(const HUF_DTable* table)
 {
-	DTableDesc dtd;
-	memcpy(&dtd, table, sizeof(dtd));
-	return dtd;
+    DTableDesc dtd;
+    ZSTD_memcpy(&dtd, table, sizeof(dtd));
+    return dtd;
 }
 
+
+#ifndef HUF_FORCE_DECOMPRESS_X2
+
 /*-***************************/
 /*  single-symbol decoding   */
 /*-***************************/
+typedef struct { BYTE byte; BYTE nbBits; } HUF_DEltX1;   /* single-symbol decoding */
+
+/*
+ * Packs 4 HUF_DEltX1 structs into a U64. This is used to lay down 4 entries at
+ * a time.
+ */
+static U64 HUF_DEltX1_set4(BYTE symbol, BYTE nbBits) {
+    U64 D4;
+    if (MEM_isLittleEndian()) {
+        D4 = symbol + (nbBits << 8);
+    } else {
+        D4 = (symbol << 8) + nbBits;
+    }
+    D4 *= 0x0001000100010001ULL;
+    return D4;
+}
 
 typedef struct {
-	BYTE byte;
-	BYTE nbBits;
-} HUF_DEltX2; /* single-symbol decoding */
-
-size_t HUF_readDTableX2_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
-{
-	U32 tableLog = 0;
-	U32 nbSymbols = 0;
-	size_t iSize;
-	void *const dtPtr = DTable + 1;
-	HUF_DEltX2 *const dt = (HUF_DEltX2 *)dtPtr;
-
-	U32 *rankVal;
-	BYTE *huffWeight;
-	size_t spaceUsed32 = 0;
-
-	rankVal = (U32 *)workspace + spaceUsed32;
-	spaceUsed32 += HUF_TABLELOG_ABSOLUTEMAX + 1;
-	huffWeight = (BYTE *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	HUF_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUF_DTable));
-	/* memset(huffWeight, 0, sizeof(huffWeight)); */ /* is not necessary, even though some analyzer complain ... */
-
-	iSize = HUF_readStats_wksp(huffWeight, HUF_SYMBOLVALUE_MAX + 1, rankVal, &nbSymbols, &tableLog, src, srcSize, workspace, workspaceSize);
-	if (HUF_isError(iSize))
-		return iSize;
-
-	/* Table header */
-	{
-		DTableDesc dtd = HUF_getDTableDesc(DTable);
-		if (tableLog > (U32)(dtd.maxTableLog + 1))
-			return ERROR(tableLog_tooLarge); /* DTable too small, Huffman tree cannot fit in */
-		dtd.tableType = 0;
-		dtd.tableLog = (BYTE)tableLog;
-		memcpy(DTable, &dtd, sizeof(dtd));
-	}
-
-	/* Calculate starting value for each rank */
-	{
-		U32 n, nextRankStart = 0;
-		for (n = 1; n < tableLog + 1; n++) {
-			U32 const curr = nextRankStart;
-			nextRankStart += (rankVal[n] << (n - 1));
-			rankVal[n] = curr;
-		}
-	}
-
-	/* fill DTable */
-	{
-		U32 n;
-		for (n = 0; n < nbSymbols; n++) {
-			U32 const w = huffWeight[n];
-			U32 const length = (1 << w) >> 1;
-			U32 u;
-			HUF_DEltX2 D;
-			D.byte = (BYTE)n;
-			D.nbBits = (BYTE)(tableLog + 1 - w);
-			for (u = rankVal[w]; u < rankVal[w] + length; u++)
-				dt[u] = D;
-			rankVal[w] += length;
-		}
-	}
-
-	return iSize;
-}
-
-static BYTE HUF_decodeSymbolX2(BIT_DStream_t *Dstream, const HUF_DEltX2 *dt, const U32 dtLog)
-{
-	size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */
-	BYTE const c = dt[val].byte;
-	BIT_skipBits(Dstream, dt[val].nbBits);
-	return c;
-}
-
-#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) *ptr++ = HUF_decodeSymbolX2(DStreamPtr, dt, dtLog)
-
-#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr)         \
-	if (ZSTD_64bits() || (HUF_TABLELOG_MAX <= 12)) \
-	HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
+        U32 rankVal[HUF_TABLELOG_ABSOLUTEMAX + 1];
+        U32 rankStart[HUF_TABLELOG_ABSOLUTEMAX + 1];
+        U32 statsWksp[HUF_READ_STATS_WORKSPACE_SIZE_U32];
+        BYTE symbols[HUF_SYMBOLVALUE_MAX + 1];
+        BYTE huffWeight[HUF_SYMBOLVALUE_MAX + 1];
+} HUF_ReadDTableX1_Workspace;
 
-#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
-	if (ZSTD_64bits())                     \
-	HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
 
-FORCE_INLINE size_t HUF_decodeStreamX2(BYTE *p, BIT_DStream_t *const bitDPtr, BYTE *const pEnd, const HUF_DEltX2 *const dt, const U32 dtLog)
+size_t HUF_readDTableX1_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize)
+{
+    return HUF_readDTableX1_wksp_bmi2(DTable, src, srcSize, workSpace, wkspSize, /* bmi2 */ 0);
+}
+
+size_t HUF_readDTableX1_wksp_bmi2(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int bmi2)
+{
+    U32 tableLog = 0;
+    U32 nbSymbols = 0;
+    size_t iSize;
+    void* const dtPtr = DTable + 1;
+    HUF_DEltX1* const dt = (HUF_DEltX1*)dtPtr;
+    HUF_ReadDTableX1_Workspace* wksp = (HUF_ReadDTableX1_Workspace*)workSpace;
+
+    DEBUG_STATIC_ASSERT(HUF_DECOMPRESS_WORKSPACE_SIZE >= sizeof(*wksp));
+    if (sizeof(*wksp) > wkspSize) return ERROR(tableLog_tooLarge);
+
+    DEBUG_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUF_DTable));
+    /* ZSTD_memset(huffWeight, 0, sizeof(huffWeight)); */   /* is not necessary, even though some analyzer complain ... */
+
+    iSize = HUF_readStats_wksp(wksp->huffWeight, HUF_SYMBOLVALUE_MAX + 1, wksp->rankVal, &nbSymbols, &tableLog, src, srcSize, wksp->statsWksp, sizeof(wksp->statsWksp), bmi2);
+    if (HUF_isError(iSize)) return iSize;
+
+    /* Table header */
+    {   DTableDesc dtd = HUF_getDTableDesc(DTable);
+        if (tableLog > (U32)(dtd.maxTableLog+1)) return ERROR(tableLog_tooLarge);   /* DTable too small, Huffman tree cannot fit in */
+        dtd.tableType = 0;
+        dtd.tableLog = (BYTE)tableLog;
+        ZSTD_memcpy(DTable, &dtd, sizeof(dtd));
+    }
+
+    /* Compute symbols and rankStart given rankVal:
+     *
+     * rankVal already contains the number of values of each weight.
+     *
+     * symbols contains the symbols ordered by weight. First are the rankVal[0]
+     * weight 0 symbols, followed by the rankVal[1] weight 1 symbols, and so on.
+     * symbols[0] is filled (but unused) to avoid a branch.
+     *
+     * rankStart contains the offset where each rank belongs in the DTable.
+     * rankStart[0] is not filled because there are no entries in the table for
+     * weight 0.
+     */
+    {
+        int n;
+        int nextRankStart = 0;
+        int const unroll = 4;
+        int const nLimit = (int)nbSymbols - unroll + 1;
+        for (n=0; n<(int)tableLog+1; n++) {
+            U32 const curr = nextRankStart;
+            nextRankStart += wksp->rankVal[n];
+            wksp->rankStart[n] = curr;
+        }
+        for (n=0; n < nLimit; n += unroll) {
+            int u;
+            for (u=0; u < unroll; ++u) {
+                size_t const w = wksp->huffWeight[n+u];
+                wksp->symbols[wksp->rankStart[w]++] = (BYTE)(n+u);
+            }
+        }
+        for (; n < (int)nbSymbols; ++n) {
+            size_t const w = wksp->huffWeight[n];
+            wksp->symbols[wksp->rankStart[w]++] = (BYTE)n;
+        }
+    }
+
+    /* fill DTable
+     * We fill all entries of each weight in order.
+     * That way length is a constant for each iteration of the outter loop.
+     * We can switch based on the length to a different inner loop which is
+     * optimized for that particular case.
+     */
+    {
+        U32 w;
+        int symbol=wksp->rankVal[0];
+        int rankStart=0;
+        for (w=1; w<tableLog+1; ++w) {
+            int const symbolCount = wksp->rankVal[w];
+            int const length = (1 << w) >> 1;
+            int uStart = rankStart;
+            BYTE const nbBits = (BYTE)(tableLog + 1 - w);
+            int s;
+            int u;
+            switch (length) {
+            case 1:
+                for (s=0; s<symbolCount; ++s) {
+                    HUF_DEltX1 D;
+                    D.byte = wksp->symbols[symbol + s];
+                    D.nbBits = nbBits;
+                    dt[uStart] = D;
+                    uStart += 1;
+                }
+                break;
+            case 2:
+                for (s=0; s<symbolCount; ++s) {
+                    HUF_DEltX1 D;
+                    D.byte = wksp->symbols[symbol + s];
+                    D.nbBits = nbBits;
+                    dt[uStart+0] = D;
+                    dt[uStart+1] = D;
+                    uStart += 2;
+                }
+                break;
+            case 4:
+                for (s=0; s<symbolCount; ++s) {
+                    U64 const D4 = HUF_DEltX1_set4(wksp->symbols[symbol + s], nbBits);
+                    MEM_write64(dt + uStart, D4);
+                    uStart += 4;
+                }
+                break;
+            case 8:
+                for (s=0; s<symbolCount; ++s) {
+                    U64 const D4 = HUF_DEltX1_set4(wksp->symbols[symbol + s], nbBits);
+                    MEM_write64(dt + uStart, D4);
+                    MEM_write64(dt + uStart + 4, D4);
+                    uStart += 8;
+                }
+                break;
+            default:
+                for (s=0; s<symbolCount; ++s) {
+                    U64 const D4 = HUF_DEltX1_set4(wksp->symbols[symbol + s], nbBits);
+                    for (u=0; u < length; u += 16) {
+                        MEM_write64(dt + uStart + u + 0, D4);
+                        MEM_write64(dt + uStart + u + 4, D4);
+                        MEM_write64(dt + uStart + u + 8, D4);
+                        MEM_write64(dt + uStart + u + 12, D4);
+                    }
+                    assert(u == length);
+                    uStart += length;
+                }
+                break;
+            }
+            symbol += symbolCount;
+            rankStart += symbolCount * length;
+        }
+    }
+    return iSize;
+}
+
+FORCE_INLINE_TEMPLATE BYTE
+HUF_decodeSymbolX1(BIT_DStream_t* Dstream, const HUF_DEltX1* dt, const U32 dtLog)
+{
+    size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */
+    BYTE const c = dt[val].byte;
+    BIT_skipBits(Dstream, dt[val].nbBits);
+    return c;
+}
+
+#define HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr) \
+    *ptr++ = HUF_decodeSymbolX1(DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX1_1(ptr, DStreamPtr)  \
+    if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
+        HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr)
+
+#define HUF_DECODE_SYMBOLX1_2(ptr, DStreamPtr) \
+    if (MEM_64bits()) \
+        HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr)
+
+HINT_INLINE size_t
+HUF_decodeStreamX1(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX1* const dt, const U32 dtLog)
+{
+    BYTE* const pStart = p;
+
+    /* up to 4 symbols at a time */
+    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-3)) {
+        HUF_DECODE_SYMBOLX1_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX1_1(p, bitDPtr);
+        HUF_DECODE_SYMBOLX1_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX1_0(p, bitDPtr);
+    }
+
+    /* [0-3] symbols remaining */
+    if (MEM_32bits())
+        while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd))
+            HUF_DECODE_SYMBOLX1_0(p, bitDPtr);
+
+    /* no more data to retrieve from bitstream, no need to reload */
+    while (p < pEnd)
+        HUF_DECODE_SYMBOLX1_0(p, bitDPtr);
+
+    return pEnd-pStart;
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress1X1_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
 {
-	BYTE *const pStart = p;
+    BYTE* op = (BYTE*)dst;
+    BYTE* const oend = op + dstSize;
+    const void* dtPtr = DTable + 1;
+    const HUF_DEltX1* const dt = (const HUF_DEltX1*)dtPtr;
+    BIT_DStream_t bitD;
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+    U32 const dtLog = dtd.tableLog;
 
-	/* up to 4 symbols at a time */
-	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd - 4)) {
-		HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
-		HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
-		HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
-		HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
-	}
-
-	/* closer to the end */
-	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd))
-		HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
-
-	/* no more data to retrieve from bitstream, hence no need to reload */
-	while (p < pEnd)
-		HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
-
-	return pEnd - pStart;
-}
-
-static size_t HUF_decompress1X2_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	BYTE *op = (BYTE *)dst;
-	BYTE *const oend = op + dstSize;
-	const void *dtPtr = DTable + 1;
-	const HUF_DEltX2 *const dt = (const HUF_DEltX2 *)dtPtr;
-	BIT_DStream_t bitD;
-	DTableDesc const dtd = HUF_getDTableDesc(DTable);
-	U32 const dtLog = dtd.tableLog;
-
-	{
-		size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
-		if (HUF_isError(errorCode))
-			return errorCode;
-	}
-
-	HUF_decodeStreamX2(op, &bitD, oend, dt, dtLog);
-
-	/* check */
-	if (!BIT_endOfDStream(&bitD))
-		return ERROR(corruption_detected);
-
-	return dstSize;
-}
-
-size_t HUF_decompress1X2_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	if (dtd.tableType != 0)
-		return ERROR(GENERIC);
-	return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable *DCtx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	const BYTE *ip = (const BYTE *)cSrc;
-
-	size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize, workspace, workspaceSize);
-	if (HUF_isError(hSize))
-		return hSize;
-	if (hSize >= cSrcSize)
-		return ERROR(srcSize_wrong);
-	ip += hSize;
-	cSrcSize -= hSize;
-
-	return HUF_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx);
-}
-
-static size_t HUF_decompress4X2_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	/* Check */
-	if (cSrcSize < 10)
-		return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
-
-	{
-		const BYTE *const istart = (const BYTE *)cSrc;
-		BYTE *const ostart = (BYTE *)dst;
-		BYTE *const oend = ostart + dstSize;
-		const void *const dtPtr = DTable + 1;
-		const HUF_DEltX2 *const dt = (const HUF_DEltX2 *)dtPtr;
-
-		/* Init */
-		BIT_DStream_t bitD1;
-		BIT_DStream_t bitD2;
-		BIT_DStream_t bitD3;
-		BIT_DStream_t bitD4;
-		size_t const length1 = ZSTD_readLE16(istart);
-		size_t const length2 = ZSTD_readLE16(istart + 2);
-		size_t const length3 = ZSTD_readLE16(istart + 4);
-		size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
-		const BYTE *const istart1 = istart + 6; /* jumpTable */
-		const BYTE *const istart2 = istart1 + length1;
-		const BYTE *const istart3 = istart2 + length2;
-		const BYTE *const istart4 = istart3 + length3;
-		const size_t segmentSize = (dstSize + 3) / 4;
-		BYTE *const opStart2 = ostart + segmentSize;
-		BYTE *const opStart3 = opStart2 + segmentSize;
-		BYTE *const opStart4 = opStart3 + segmentSize;
-		BYTE *op1 = ostart;
-		BYTE *op2 = opStart2;
-		BYTE *op3 = opStart3;
-		BYTE *op4 = opStart4;
-		U32 endSignal;
-		DTableDesc const dtd = HUF_getDTableDesc(DTable);
-		U32 const dtLog = dtd.tableLog;
-
-		if (length4 > cSrcSize)
-			return ERROR(corruption_detected); /* overflow */
-		{
-			size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-
-		/* 16-32 symbols per loop (4-8 symbols per stream) */
-		endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-		for (; (endSignal == BIT_DStream_unfinished) && (op4 < (oend - 7));) {
-			HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
-			HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
-			HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
-			HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
-			HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
-			HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
-			HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
-			HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
-			HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
-			HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
-			HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
-			HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
-			HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
-			HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
-			HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
-			HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
-			endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-		}
-
-		/* check corruption */
-		if (op1 > opStart2)
-			return ERROR(corruption_detected);
-		if (op2 > opStart3)
-			return ERROR(corruption_detected);
-		if (op3 > opStart4)
-			return ERROR(corruption_detected);
-		/* note : op4 supposed already verified within main loop */
-
-		/* finish bitStreams one by one */
-		HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog);
-		HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog);
-		HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog);
-		HUF_decodeStreamX2(op4, &bitD4, oend, dt, dtLog);
-
-		/* check */
-		endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
-		if (!endSignal)
-			return ERROR(corruption_detected);
-
-		/* decoded size */
-		return dstSize;
-	}
-}
-
-size_t HUF_decompress4X2_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	if (dtd.tableType != 0)
-		return ERROR(GENERIC);
-	return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	const BYTE *ip = (const BYTE *)cSrc;
-
-	size_t const hSize = HUF_readDTableX2_wksp(dctx, cSrc, cSrcSize, workspace, workspaceSize);
-	if (HUF_isError(hSize))
-		return hSize;
-	if (hSize >= cSrcSize)
-		return ERROR(srcSize_wrong);
-	ip += hSize;
-	cSrcSize -= hSize;
-
-	return HUF_decompress4X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx);
+    CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );
+
+    HUF_decodeStreamX1(op, &bitD, oend, dt, dtLog);
+
+    if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
+
+    return dstSize;
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress4X1_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    /* Check */
+    if (cSrcSize < 10) return ERROR(corruption_detected);  /* strict minimum : jump table + 1 byte per stream */
+
+    {   const BYTE* const istart = (const BYTE*) cSrc;
+        BYTE* const ostart = (BYTE*) dst;
+        BYTE* const oend = ostart + dstSize;
+        BYTE* const olimit = oend - 3;
+        const void* const dtPtr = DTable + 1;
+        const HUF_DEltX1* const dt = (const HUF_DEltX1*)dtPtr;
+
+        /* Init */
+        BIT_DStream_t bitD1;
+        BIT_DStream_t bitD2;
+        BIT_DStream_t bitD3;
+        BIT_DStream_t bitD4;
+        size_t const length1 = MEM_readLE16(istart);
+        size_t const length2 = MEM_readLE16(istart+2);
+        size_t const length3 = MEM_readLE16(istart+4);
+        size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
+        const BYTE* const istart1 = istart + 6;  /* jumpTable */
+        const BYTE* const istart2 = istart1 + length1;
+        const BYTE* const istart3 = istart2 + length2;
+        const BYTE* const istart4 = istart3 + length3;
+        const size_t segmentSize = (dstSize+3) / 4;
+        BYTE* const opStart2 = ostart + segmentSize;
+        BYTE* const opStart3 = opStart2 + segmentSize;
+        BYTE* const opStart4 = opStart3 + segmentSize;
+        BYTE* op1 = ostart;
+        BYTE* op2 = opStart2;
+        BYTE* op3 = opStart3;
+        BYTE* op4 = opStart4;
+        DTableDesc const dtd = HUF_getDTableDesc(DTable);
+        U32 const dtLog = dtd.tableLog;
+        U32 endSignal = 1;
+
+        if (length4 > cSrcSize) return ERROR(corruption_detected);   /* overflow */
+        CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
+        CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
+        CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
+        CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );
+
+        /* up to 16 symbols per loop (4 symbols per stream) in 64-bit mode */
+        for ( ; (endSignal) & (op4 < olimit) ; ) {
+            HUF_DECODE_SYMBOLX1_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX1_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX1_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX1_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX1_1(op1, &bitD1);
+            HUF_DECODE_SYMBOLX1_1(op2, &bitD2);
+            HUF_DECODE_SYMBOLX1_1(op3, &bitD3);
+            HUF_DECODE_SYMBOLX1_1(op4, &bitD4);
+            HUF_DECODE_SYMBOLX1_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX1_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX1_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX1_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX1_0(op1, &bitD1);
+            HUF_DECODE_SYMBOLX1_0(op2, &bitD2);
+            HUF_DECODE_SYMBOLX1_0(op3, &bitD3);
+            HUF_DECODE_SYMBOLX1_0(op4, &bitD4);
+            endSignal &= BIT_reloadDStreamFast(&bitD1) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD2) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD3) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD4) == BIT_DStream_unfinished;
+        }
+
+        /* check corruption */
+        /* note : should not be necessary : op# advance in lock step, and we control op4.
+         *        but curiously, binary generated by gcc 7.2 & 7.3 with -mbmi2 runs faster when >=1 test is present */
+        if (op1 > opStart2) return ERROR(corruption_detected);
+        if (op2 > opStart3) return ERROR(corruption_detected);
+        if (op3 > opStart4) return ERROR(corruption_detected);
+        /* note : op4 supposed already verified within main loop */
+
+        /* finish bitStreams one by one */
+        HUF_decodeStreamX1(op1, &bitD1, opStart2, dt, dtLog);
+        HUF_decodeStreamX1(op2, &bitD2, opStart3, dt, dtLog);
+        HUF_decodeStreamX1(op3, &bitD3, opStart4, dt, dtLog);
+        HUF_decodeStreamX1(op4, &bitD4, oend,     dt, dtLog);
+
+        /* check */
+        { U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
+          if (!endCheck) return ERROR(corruption_detected); }
+
+        /* decoded size */
+        return dstSize;
+    }
+}
+
+
+typedef size_t (*HUF_decompress_usingDTable_t)(void *dst, size_t dstSize,
+                                               const void *cSrc,
+                                               size_t cSrcSize,
+                                               const HUF_DTable *DTable);
+
+HUF_DGEN(HUF_decompress1X1_usingDTable_internal)
+HUF_DGEN(HUF_decompress4X1_usingDTable_internal)
+
+
+
+size_t HUF_decompress1X1_usingDTable(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    if (dtd.tableType != 0) return ERROR(GENERIC);
+    return HUF_decompress1X1_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+}
+
+size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
+
+    size_t const hSize = HUF_readDTableX1_wksp(DCtx, cSrc, cSrcSize, workSpace, wkspSize);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress1X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
+}
+
+
+size_t HUF_decompress4X1_usingDTable(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    if (dtd.tableType != 0) return ERROR(GENERIC);
+    return HUF_decompress4X1_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+}
+
+static size_t HUF_decompress4X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize, int bmi2)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
+
+    size_t const hSize = HUF_readDTableX1_wksp_bmi2(dctx, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress4X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
 }
 
+size_t HUF_decompress4X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize)
+{
+    return HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, 0);
+}
+
+
+#endif /* HUF_FORCE_DECOMPRESS_X2 */
+
+
+#ifndef HUF_FORCE_DECOMPRESS_X1
+
 /* *************************/
 /* double-symbols decoding */
 /* *************************/
-typedef struct {
-	U16 sequence;
-	BYTE nbBits;
-	BYTE length;
-} HUF_DEltX4; /* double-symbols decoding */
 
-typedef struct {
-	BYTE symbol;
-	BYTE weight;
-} sortedSymbol_t;
+typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX2;  /* double-symbols decoding */
+typedef struct { BYTE symbol; BYTE weight; } sortedSymbol_t;
+typedef U32 rankValCol_t[HUF_TABLELOG_MAX + 1];
+typedef rankValCol_t rankVal_t[HUF_TABLELOG_MAX];
+
 
-/* HUF_fillDTableX4Level2() :
+/* HUF_fillDTableX2Level2() :
  * `rankValOrigin` must be a table of at least (HUF_TABLELOG_MAX + 1) U32 */
-static void HUF_fillDTableX4Level2(HUF_DEltX4 *DTable, U32 sizeLog, const U32 consumed, const U32 *rankValOrigin, const int minWeight,
-				   const sortedSymbol_t *sortedSymbols, const U32 sortedListSize, U32 nbBitsBaseline, U16 baseSeq)
-{
-	HUF_DEltX4 DElt;
-	U32 rankVal[HUF_TABLELOG_MAX + 1];
-
-	/* get pre-calculated rankVal */
-	memcpy(rankVal, rankValOrigin, sizeof(rankVal));
-
-	/* fill skipped values */
-	if (minWeight > 1) {
-		U32 i, skipSize = rankVal[minWeight];
-		ZSTD_writeLE16(&(DElt.sequence), baseSeq);
-		DElt.nbBits = (BYTE)(consumed);
-		DElt.length = 1;
-		for (i = 0; i < skipSize; i++)
-			DTable[i] = DElt;
-	}
-
-	/* fill DTable */
-	{
-		U32 s;
-		for (s = 0; s < sortedListSize; s++) { /* note : sortedSymbols already skipped */
-			const U32 symbol = sortedSymbols[s].symbol;
-			const U32 weight = sortedSymbols[s].weight;
-			const U32 nbBits = nbBitsBaseline - weight;
-			const U32 length = 1 << (sizeLog - nbBits);
-			const U32 start = rankVal[weight];
-			U32 i = start;
-			const U32 end = start + length;
-
-			ZSTD_writeLE16(&(DElt.sequence), (U16)(baseSeq + (symbol << 8)));
-			DElt.nbBits = (BYTE)(nbBits + consumed);
-			DElt.length = 2;
-			do {
-				DTable[i++] = DElt;
-			} while (i < end); /* since length >= 1 */
-
-			rankVal[weight] += length;
-		}
-	}
-}
-
-typedef U32 rankVal_t[HUF_TABLELOG_MAX][HUF_TABLELOG_MAX + 1];
-typedef U32 rankValCol_t[HUF_TABLELOG_MAX + 1];
+static void HUF_fillDTableX2Level2(HUF_DEltX2* DTable, U32 sizeLog, const U32 consumed,
+                           const U32* rankValOrigin, const int minWeight,
+                           const sortedSymbol_t* sortedSymbols, const U32 sortedListSize,
+                           U32 nbBitsBaseline, U16 baseSeq, U32* wksp, size_t wkspSize)
+{
+    HUF_DEltX2 DElt;
+    U32* rankVal = wksp;
+
+    assert(wkspSize >= HUF_TABLELOG_MAX + 1);
+    (void)wkspSize;
+    /* get pre-calculated rankVal */
+    ZSTD_memcpy(rankVal, rankValOrigin, sizeof(U32) * (HUF_TABLELOG_MAX + 1));
+
+    /* fill skipped values */
+    if (minWeight>1) {
+        U32 i, skipSize = rankVal[minWeight];
+        MEM_writeLE16(&(DElt.sequence), baseSeq);
+        DElt.nbBits   = (BYTE)(consumed);
+        DElt.length   = 1;
+        for (i = 0; i < skipSize; i++)
+            DTable[i] = DElt;
+    }
+
+    /* fill DTable */
+    {   U32 s; for (s=0; s<sortedListSize; s++) {   /* note : sortedSymbols already skipped */
+            const U32 symbol = sortedSymbols[s].symbol;
+            const U32 weight = sortedSymbols[s].weight;
+            const U32 nbBits = nbBitsBaseline - weight;
+            const U32 length = 1 << (sizeLog-nbBits);
+            const U32 start = rankVal[weight];
+            U32 i = start;
+            const U32 end = start + length;
+
+            MEM_writeLE16(&(DElt.sequence), (U16)(baseSeq + (symbol << 8)));
+            DElt.nbBits = (BYTE)(nbBits + consumed);
+            DElt.length = 2;
+            do { DTable[i++] = DElt; } while (i<end);   /* since length >= 1 */
+
+            rankVal[weight] += length;
+    }   }
+}
 
-static void HUF_fillDTableX4(HUF_DEltX4 *DTable, const U32 targetLog, const sortedSymbol_t *sortedList, const U32 sortedListSize, const U32 *rankStart,
-			     rankVal_t rankValOrigin, const U32 maxWeight, const U32 nbBitsBaseline)
-{
-	U32 rankVal[HUF_TABLELOG_MAX + 1];
-	const int scaleLog = nbBitsBaseline - targetLog; /* note : targetLog >= srcLog, hence scaleLog <= 1 */
-	const U32 minBits = nbBitsBaseline - maxWeight;
-	U32 s;
-
-	memcpy(rankVal, rankValOrigin, sizeof(rankVal));
-
-	/* fill DTable */
-	for (s = 0; s < sortedListSize; s++) {
-		const U16 symbol = sortedList[s].symbol;
-		const U32 weight = sortedList[s].weight;
-		const U32 nbBits = nbBitsBaseline - weight;
-		const U32 start = rankVal[weight];
-		const U32 length = 1 << (targetLog - nbBits);
-
-		if (targetLog - nbBits >= minBits) { /* enough room for a second symbol */
-			U32 sortedRank;
-			int minWeight = nbBits + scaleLog;
-			if (minWeight < 1)
-				minWeight = 1;
-			sortedRank = rankStart[minWeight];
-			HUF_fillDTableX4Level2(DTable + start, targetLog - nbBits, nbBits, rankValOrigin[nbBits], minWeight, sortedList + sortedRank,
-					       sortedListSize - sortedRank, nbBitsBaseline, symbol);
-		} else {
-			HUF_DEltX4 DElt;
-			ZSTD_writeLE16(&(DElt.sequence), symbol);
-			DElt.nbBits = (BYTE)(nbBits);
-			DElt.length = 1;
-			{
-				U32 const end = start + length;
-				U32 u;
-				for (u = start; u < end; u++)
-					DTable[u] = DElt;
-			}
-		}
-		rankVal[weight] += length;
-	}
-}
-
-size_t HUF_readDTableX4_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
-{
-	U32 tableLog, maxW, sizeOfSort, nbSymbols;
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	U32 const maxTableLog = dtd.maxTableLog;
-	size_t iSize;
-	void *dtPtr = DTable + 1; /* force compiler to avoid strict-aliasing */
-	HUF_DEltX4 *const dt = (HUF_DEltX4 *)dtPtr;
-	U32 *rankStart;
-
-	rankValCol_t *rankVal;
-	U32 *rankStats;
-	U32 *rankStart0;
-	sortedSymbol_t *sortedSymbol;
-	BYTE *weightList;
-	size_t spaceUsed32 = 0;
-
-	HUF_STATIC_ASSERT((sizeof(rankValCol_t) & 3) == 0);
-
-	rankVal = (rankValCol_t *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += (sizeof(rankValCol_t) * HUF_TABLELOG_MAX) >> 2;
-	rankStats = (U32 *)workspace + spaceUsed32;
-	spaceUsed32 += HUF_TABLELOG_MAX + 1;
-	rankStart0 = (U32 *)workspace + spaceUsed32;
-	spaceUsed32 += HUF_TABLELOG_MAX + 2;
-	sortedSymbol = (sortedSymbol_t *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(sizeof(sortedSymbol_t) * (HUF_SYMBOLVALUE_MAX + 1), sizeof(U32)) >> 2;
-	weightList = (BYTE *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	rankStart = rankStart0 + 1;
-	memset(rankStats, 0, sizeof(U32) * (2 * HUF_TABLELOG_MAX + 2 + 1));
-
-	HUF_STATIC_ASSERT(sizeof(HUF_DEltX4) == sizeof(HUF_DTable)); /* if compiler fails here, assertion is wrong */
-	if (maxTableLog > HUF_TABLELOG_MAX)
-		return ERROR(tableLog_tooLarge);
-	/* memset(weightList, 0, sizeof(weightList)); */ /* is not necessary, even though some analyzer complain ... */
-
-	iSize = HUF_readStats_wksp(weightList, HUF_SYMBOLVALUE_MAX + 1, rankStats, &nbSymbols, &tableLog, src, srcSize, workspace, workspaceSize);
-	if (HUF_isError(iSize))
-		return iSize;
-
-	/* check result */
-	if (tableLog > maxTableLog)
-		return ERROR(tableLog_tooLarge); /* DTable can't fit code depth */
-
-	/* find maxWeight */
-	for (maxW = tableLog; rankStats[maxW] == 0; maxW--) {
-	} /* necessarily finds a solution before 0 */
-
-	/* Get start index of each weight */
-	{
-		U32 w, nextRankStart = 0;
-		for (w = 1; w < maxW + 1; w++) {
-			U32 curr = nextRankStart;
-			nextRankStart += rankStats[w];
-			rankStart[w] = curr;
-		}
-		rankStart[0] = nextRankStart; /* put all 0w symbols at the end of sorted list*/
-		sizeOfSort = nextRankStart;
-	}
-
-	/* sort symbols by weight */
-	{
-		U32 s;
-		for (s = 0; s < nbSymbols; s++) {
-			U32 const w = weightList[s];
-			U32 const r = rankStart[w]++;
-			sortedSymbol[r].symbol = (BYTE)s;
-			sortedSymbol[r].weight = (BYTE)w;
-		}
-		rankStart[0] = 0; /* forget 0w symbols; this is beginning of weight(1) */
-	}
-
-	/* Build rankVal */
-	{
-		U32 *const rankVal0 = rankVal[0];
-		{
-			int const rescale = (maxTableLog - tableLog) - 1; /* tableLog <= maxTableLog */
-			U32 nextRankVal = 0;
-			U32 w;
-			for (w = 1; w < maxW + 1; w++) {
-				U32 curr = nextRankVal;
-				nextRankVal += rankStats[w] << (w + rescale);
-				rankVal0[w] = curr;
-			}
-		}
-		{
-			U32 const minBits = tableLog + 1 - maxW;
-			U32 consumed;
-			for (consumed = minBits; consumed < maxTableLog - minBits + 1; consumed++) {
-				U32 *const rankValPtr = rankVal[consumed];
-				U32 w;
-				for (w = 1; w < maxW + 1; w++) {
-					rankValPtr[w] = rankVal0[w] >> consumed;
-				}
-			}
-		}
-	}
-
-	HUF_fillDTableX4(dt, maxTableLog, sortedSymbol, sizeOfSort, rankStart0, rankVal, maxW, tableLog + 1);
-
-	dtd.tableLog = (BYTE)maxTableLog;
-	dtd.tableType = 1;
-	memcpy(DTable, &dtd, sizeof(dtd));
-	return iSize;
-}
-
-static U32 HUF_decodeSymbolX4(void *op, BIT_DStream_t *DStream, const HUF_DEltX4 *dt, const U32 dtLog)
-{
-	size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
-	memcpy(op, dt + val, 2);
-	BIT_skipBits(DStream, dt[val].nbBits);
-	return dt[val].length;
-}
-
-static U32 HUF_decodeLastSymbolX4(void *op, BIT_DStream_t *DStream, const HUF_DEltX4 *dt, const U32 dtLog)
-{
-	size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
-	memcpy(op, dt + val, 1);
-	if (dt[val].length == 1)
-		BIT_skipBits(DStream, dt[val].nbBits);
-	else {
-		if (DStream->bitsConsumed < (sizeof(DStream->bitContainer) * 8)) {
-			BIT_skipBits(DStream, dt[val].nbBits);
-			if (DStream->bitsConsumed > (sizeof(DStream->bitContainer) * 8))
-				/* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
-				DStream->bitsConsumed = (sizeof(DStream->bitContainer) * 8);
-		}
-	}
-	return 1;
-}
-
-#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
-
-#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr)         \
-	if (ZSTD_64bits() || (HUF_TABLELOG_MAX <= 12)) \
-	ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
-
-#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \
-	if (ZSTD_64bits())                     \
-	ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
-
-FORCE_INLINE size_t HUF_decodeStreamX4(BYTE *p, BIT_DStream_t *bitDPtr, BYTE *const pEnd, const HUF_DEltX4 *const dt, const U32 dtLog)
-{
-	BYTE *const pStart = p;
-
-	/* up to 8 symbols at a time */
-	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd - (sizeof(bitDPtr->bitContainer) - 1))) {
-		HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
-		HUF_DECODE_SYMBOLX4_1(p, bitDPtr);
-		HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
-		HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
-	}
-
-	/* closer to end : up to 2 symbols at a time */
-	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd - 2))
-		HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
-
-	while (p <= pEnd - 2)
-		HUF_DECODE_SYMBOLX4_0(p, bitDPtr); /* no need to reload : reached the end of DStream */
-
-	if (p < pEnd)
-		p += HUF_decodeLastSymbolX4(p, bitDPtr, dt, dtLog);
-
-	return p - pStart;
-}
-
-static size_t HUF_decompress1X4_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	BIT_DStream_t bitD;
-
-	/* Init */
-	{
-		size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
-		if (HUF_isError(errorCode))
-			return errorCode;
-	}
-
-	/* decode */
-	{
-		BYTE *const ostart = (BYTE *)dst;
-		BYTE *const oend = ostart + dstSize;
-		const void *const dtPtr = DTable + 1; /* force compiler to not use strict-aliasing */
-		const HUF_DEltX4 *const dt = (const HUF_DEltX4 *)dtPtr;
-		DTableDesc const dtd = HUF_getDTableDesc(DTable);
-		HUF_decodeStreamX4(ostart, &bitD, oend, dt, dtd.tableLog);
-	}
-
-	/* check */
-	if (!BIT_endOfDStream(&bitD))
-		return ERROR(corruption_detected);
-
-	/* decoded size */
-	return dstSize;
-}
-
-size_t HUF_decompress1X4_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	if (dtd.tableType != 1)
-		return ERROR(GENERIC);
-	return HUF_decompress1X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable *DCtx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	const BYTE *ip = (const BYTE *)cSrc;
-
-	size_t const hSize = HUF_readDTableX4_wksp(DCtx, cSrc, cSrcSize, workspace, workspaceSize);
-	if (HUF_isError(hSize))
-		return hSize;
-	if (hSize >= cSrcSize)
-		return ERROR(srcSize_wrong);
-	ip += hSize;
-	cSrcSize -= hSize;
-
-	return HUF_decompress1X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx);
-}
-
-static size_t HUF_decompress4X4_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	if (cSrcSize < 10)
-		return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
-
-	{
-		const BYTE *const istart = (const BYTE *)cSrc;
-		BYTE *const ostart = (BYTE *)dst;
-		BYTE *const oend = ostart + dstSize;
-		const void *const dtPtr = DTable + 1;
-		const HUF_DEltX4 *const dt = (const HUF_DEltX4 *)dtPtr;
-
-		/* Init */
-		BIT_DStream_t bitD1;
-		BIT_DStream_t bitD2;
-		BIT_DStream_t bitD3;
-		BIT_DStream_t bitD4;
-		size_t const length1 = ZSTD_readLE16(istart);
-		size_t const length2 = ZSTD_readLE16(istart + 2);
-		size_t const length3 = ZSTD_readLE16(istart + 4);
-		size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
-		const BYTE *const istart1 = istart + 6; /* jumpTable */
-		const BYTE *const istart2 = istart1 + length1;
-		const BYTE *const istart3 = istart2 + length2;
-		const BYTE *const istart4 = istart3 + length3;
-		size_t const segmentSize = (dstSize + 3) / 4;
-		BYTE *const opStart2 = ostart + segmentSize;
-		BYTE *const opStart3 = opStart2 + segmentSize;
-		BYTE *const opStart4 = opStart3 + segmentSize;
-		BYTE *op1 = ostart;
-		BYTE *op2 = opStart2;
-		BYTE *op3 = opStart3;
-		BYTE *op4 = opStart4;
-		U32 endSignal;
-		DTableDesc const dtd = HUF_getDTableDesc(DTable);
-		U32 const dtLog = dtd.tableLog;
-
-		if (length4 > cSrcSize)
-			return ERROR(corruption_detected); /* overflow */
-		{
-			size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-
-		/* 16-32 symbols per loop (4-8 symbols per stream) */
-		endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-		for (; (endSignal == BIT_DStream_unfinished) & (op4 < (oend - (sizeof(bitD4.bitContainer) - 1)));) {
-			HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
-			HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
-			HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
-			HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
-			HUF_DECODE_SYMBOLX4_1(op1, &bitD1);
-			HUF_DECODE_SYMBOLX4_1(op2, &bitD2);
-			HUF_DECODE_SYMBOLX4_1(op3, &bitD3);
-			HUF_DECODE_SYMBOLX4_1(op4, &bitD4);
-			HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
-			HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
-			HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
-			HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
-			HUF_DECODE_SYMBOLX4_0(op1, &bitD1);
-			HUF_DECODE_SYMBOLX4_0(op2, &bitD2);
-			HUF_DECODE_SYMBOLX4_0(op3, &bitD3);
-			HUF_DECODE_SYMBOLX4_0(op4, &bitD4);
-
-			endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-		}
-
-		/* check corruption */
-		if (op1 > opStart2)
-			return ERROR(corruption_detected);
-		if (op2 > opStart3)
-			return ERROR(corruption_detected);
-		if (op3 > opStart4)
-			return ERROR(corruption_detected);
-		/* note : op4 already verified within main loop */
-
-		/* finish bitStreams one by one */
-		HUF_decodeStreamX4(op1, &bitD1, opStart2, dt, dtLog);
-		HUF_decodeStreamX4(op2, &bitD2, opStart3, dt, dtLog);
-		HUF_decodeStreamX4(op3, &bitD3, opStart4, dt, dtLog);
-		HUF_decodeStreamX4(op4, &bitD4, oend, dt, dtLog);
-
-		/* check */
-		{
-			U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
-			if (!endCheck)
-				return ERROR(corruption_detected);
-		}
-
-		/* decoded size */
-		return dstSize;
-	}
-}
-
-size_t HUF_decompress4X4_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	if (dtd.tableType != 1)
-		return ERROR(GENERIC);
-	return HUF_decompress4X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	const BYTE *ip = (const BYTE *)cSrc;
-
-	size_t hSize = HUF_readDTableX4_wksp(dctx, cSrc, cSrcSize, workspace, workspaceSize);
-	if (HUF_isError(hSize))
-		return hSize;
-	if (hSize >= cSrcSize)
-		return ERROR(srcSize_wrong);
-	ip += hSize;
-	cSrcSize -= hSize;
-
-	return HUF_decompress4X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx);
-}
-
-/* ********************************/
-/* Generic decompression selector */
-/* ********************************/
-
-size_t HUF_decompress1X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc const dtd = HUF_getDTableDesc(DTable);
-	return dtd.tableType ? HUF_decompress1X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable)
-			     : HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress4X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc const dtd = HUF_getDTableDesc(DTable);
-	return dtd.tableType ? HUF_decompress4X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable)
-			     : HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable);
+
+static void HUF_fillDTableX2(HUF_DEltX2* DTable, const U32 targetLog,
+                           const sortedSymbol_t* sortedList, const U32 sortedListSize,
+                           const U32* rankStart, rankVal_t rankValOrigin, const U32 maxWeight,
+                           const U32 nbBitsBaseline, U32* wksp, size_t wkspSize)
+{
+    U32* rankVal = wksp;
+    const int scaleLog = nbBitsBaseline - targetLog;   /* note : targetLog >= srcLog, hence scaleLog <= 1 */
+    const U32 minBits  = nbBitsBaseline - maxWeight;
+    U32 s;
+
+    assert(wkspSize >= HUF_TABLELOG_MAX + 1);
+    wksp += HUF_TABLELOG_MAX + 1;
+    wkspSize -= HUF_TABLELOG_MAX + 1;
+
+    ZSTD_memcpy(rankVal, rankValOrigin, sizeof(U32) * (HUF_TABLELOG_MAX + 1));
+
+    /* fill DTable */
+    for (s=0; s<sortedListSize; s++) {
+        const U16 symbol = sortedList[s].symbol;
+        const U32 weight = sortedList[s].weight;
+        const U32 nbBits = nbBitsBaseline - weight;
+        const U32 start = rankVal[weight];
+        const U32 length = 1 << (targetLog-nbBits);
+
+        if (targetLog-nbBits >= minBits) {   /* enough room for a second symbol */
+            U32 sortedRank;
+            int minWeight = nbBits + scaleLog;
+            if (minWeight < 1) minWeight = 1;
+            sortedRank = rankStart[minWeight];
+            HUF_fillDTableX2Level2(DTable+start, targetLog-nbBits, nbBits,
+                           rankValOrigin[nbBits], minWeight,
+                           sortedList+sortedRank, sortedListSize-sortedRank,
+                           nbBitsBaseline, symbol, wksp, wkspSize);
+        } else {
+            HUF_DEltX2 DElt;
+            MEM_writeLE16(&(DElt.sequence), symbol);
+            DElt.nbBits = (BYTE)(nbBits);
+            DElt.length = 1;
+            {   U32 const end = start + length;
+                U32 u;
+                for (u = start; u < end; u++) DTable[u] = DElt;
+        }   }
+        rankVal[weight] += length;
+    }
 }
 
 typedef struct {
-	U32 tableTime;
-	U32 decode256Time;
-} algo_time_t;
-static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, quad */] = {
+    rankValCol_t rankVal[HUF_TABLELOG_MAX];
+    U32 rankStats[HUF_TABLELOG_MAX + 1];
+    U32 rankStart0[HUF_TABLELOG_MAX + 2];
+    sortedSymbol_t sortedSymbol[HUF_SYMBOLVALUE_MAX + 1];
+    BYTE weightList[HUF_SYMBOLVALUE_MAX + 1];
+    U32 calleeWksp[HUF_READ_STATS_WORKSPACE_SIZE_U32];
+} HUF_ReadDTableX2_Workspace;
+
+size_t HUF_readDTableX2_wksp(HUF_DTable* DTable,
+                       const void* src, size_t srcSize,
+                             void* workSpace, size_t wkspSize)
+{
+    U32 tableLog, maxW, sizeOfSort, nbSymbols;
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    U32 const maxTableLog = dtd.maxTableLog;
+    size_t iSize;
+    void* dtPtr = DTable+1;   /* force compiler to avoid strict-aliasing */
+    HUF_DEltX2* const dt = (HUF_DEltX2*)dtPtr;
+    U32 *rankStart;
+
+    HUF_ReadDTableX2_Workspace* const wksp = (HUF_ReadDTableX2_Workspace*)workSpace;
+
+    if (sizeof(*wksp) > wkspSize) return ERROR(GENERIC);
+
+    rankStart = wksp->rankStart0 + 1;
+    ZSTD_memset(wksp->rankStats, 0, sizeof(wksp->rankStats));
+    ZSTD_memset(wksp->rankStart0, 0, sizeof(wksp->rankStart0));
+
+    DEBUG_STATIC_ASSERT(sizeof(HUF_DEltX2) == sizeof(HUF_DTable));   /* if compiler fails here, assertion is wrong */
+    if (maxTableLog > HUF_TABLELOG_MAX) return ERROR(tableLog_tooLarge);
+    /* ZSTD_memset(weightList, 0, sizeof(weightList)); */  /* is not necessary, even though some analyzer complain ... */
+
+    iSize = HUF_readStats_wksp(wksp->weightList, HUF_SYMBOLVALUE_MAX + 1, wksp->rankStats, &nbSymbols, &tableLog, src, srcSize, wksp->calleeWksp, sizeof(wksp->calleeWksp), /* bmi2 */ 0);
+    if (HUF_isError(iSize)) return iSize;
+
+    /* check result */
+    if (tableLog > maxTableLog) return ERROR(tableLog_tooLarge);   /* DTable can't fit code depth */
+
+    /* find maxWeight */
+    for (maxW = tableLog; wksp->rankStats[maxW]==0; maxW--) {}  /* necessarily finds a solution before 0 */
+
+    /* Get start index of each weight */
+    {   U32 w, nextRankStart = 0;
+        for (w=1; w<maxW+1; w++) {
+            U32 curr = nextRankStart;
+            nextRankStart += wksp->rankStats[w];
+            rankStart[w] = curr;
+        }
+        rankStart[0] = nextRankStart;   /* put all 0w symbols at the end of sorted list*/
+        sizeOfSort = nextRankStart;
+    }
+
+    /* sort symbols by weight */
+    {   U32 s;
+        for (s=0; s<nbSymbols; s++) {
+            U32 const w = wksp->weightList[s];
+            U32 const r = rankStart[w]++;
+            wksp->sortedSymbol[r].symbol = (BYTE)s;
+            wksp->sortedSymbol[r].weight = (BYTE)w;
+        }
+        rankStart[0] = 0;   /* forget 0w symbols; this is beginning of weight(1) */
+    }
+
+    /* Build rankVal */
+    {   U32* const rankVal0 = wksp->rankVal[0];
+        {   int const rescale = (maxTableLog-tableLog) - 1;   /* tableLog <= maxTableLog */
+            U32 nextRankVal = 0;
+            U32 w;
+            for (w=1; w<maxW+1; w++) {
+                U32 curr = nextRankVal;
+                nextRankVal += wksp->rankStats[w] << (w+rescale);
+                rankVal0[w] = curr;
+        }   }
+        {   U32 const minBits = tableLog+1 - maxW;
+            U32 consumed;
+            for (consumed = minBits; consumed < maxTableLog - minBits + 1; consumed++) {
+                U32* const rankValPtr = wksp->rankVal[consumed];
+                U32 w;
+                for (w = 1; w < maxW+1; w++) {
+                    rankValPtr[w] = rankVal0[w] >> consumed;
+    }   }   }   }
+
+    HUF_fillDTableX2(dt, maxTableLog,
+                   wksp->sortedSymbol, sizeOfSort,
+                   wksp->rankStart0, wksp->rankVal, maxW,
+                   tableLog+1,
+                   wksp->calleeWksp, sizeof(wksp->calleeWksp) / sizeof(U32));
+
+    dtd.tableLog = (BYTE)maxTableLog;
+    dtd.tableType = 1;
+    ZSTD_memcpy(DTable, &dtd, sizeof(dtd));
+    return iSize;
+}
+
+
+FORCE_INLINE_TEMPLATE U32
+HUF_decodeSymbolX2(void* op, BIT_DStream_t* DStream, const HUF_DEltX2* dt, const U32 dtLog)
+{
+    size_t const val = BIT_lookBitsFast(DStream, dtLog);   /* note : dtLog >= 1 */
+    ZSTD_memcpy(op, dt+val, 2);
+    BIT_skipBits(DStream, dt[val].nbBits);
+    return dt[val].length;
+}
+
+FORCE_INLINE_TEMPLATE U32
+HUF_decodeLastSymbolX2(void* op, BIT_DStream_t* DStream, const HUF_DEltX2* dt, const U32 dtLog)
+{
+    size_t const val = BIT_lookBitsFast(DStream, dtLog);   /* note : dtLog >= 1 */
+    ZSTD_memcpy(op, dt+val, 1);
+    if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits);
+    else {
+        if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) {
+            BIT_skipBits(DStream, dt[val].nbBits);
+            if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8))
+                /* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
+                DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8);
+    }   }
+    return 1;
+}
+
+#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \
+    ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \
+    if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
+        ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
+    if (MEM_64bits()) \
+        ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
+
+HINT_INLINE size_t
+HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd,
+                const HUF_DEltX2* const dt, const U32 dtLog)
+{
+    BYTE* const pStart = p;
+
+    /* up to 8 symbols at a time */
+    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-(sizeof(bitDPtr->bitContainer)-1))) {
+        HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
+        HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
+    }
+
+    /* closer to end : up to 2 symbols at a time */
+    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd-2))
+        HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
+
+    while (p <= pEnd-2)
+        HUF_DECODE_SYMBOLX2_0(p, bitDPtr);   /* no need to reload : reached the end of DStream */
+
+    if (p < pEnd)
+        p += HUF_decodeLastSymbolX2(p, bitDPtr, dt, dtLog);
+
+    return p-pStart;
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress1X2_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    BIT_DStream_t bitD;
+
+    /* Init */
+    CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );
+
+    /* decode */
+    {   BYTE* const ostart = (BYTE*) dst;
+        BYTE* const oend = ostart + dstSize;
+        const void* const dtPtr = DTable+1;   /* force compiler to not use strict-aliasing */
+        const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
+        DTableDesc const dtd = HUF_getDTableDesc(DTable);
+        HUF_decodeStreamX2(ostart, &bitD, oend, dt, dtd.tableLog);
+    }
+
+    /* check */
+    if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
+
+    /* decoded size */
+    return dstSize;
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress4X2_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    if (cSrcSize < 10) return ERROR(corruption_detected);   /* strict minimum : jump table + 1 byte per stream */
+
+    {   const BYTE* const istart = (const BYTE*) cSrc;
+        BYTE* const ostart = (BYTE*) dst;
+        BYTE* const oend = ostart + dstSize;
+        BYTE* const olimit = oend - (sizeof(size_t)-1);
+        const void* const dtPtr = DTable+1;
+        const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
+
+        /* Init */
+        BIT_DStream_t bitD1;
+        BIT_DStream_t bitD2;
+        BIT_DStream_t bitD3;
+        BIT_DStream_t bitD4;
+        size_t const length1 = MEM_readLE16(istart);
+        size_t const length2 = MEM_readLE16(istart+2);
+        size_t const length3 = MEM_readLE16(istart+4);
+        size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
+        const BYTE* const istart1 = istart + 6;  /* jumpTable */
+        const BYTE* const istart2 = istart1 + length1;
+        const BYTE* const istart3 = istart2 + length2;
+        const BYTE* const istart4 = istart3 + length3;
+        size_t const segmentSize = (dstSize+3) / 4;
+        BYTE* const opStart2 = ostart + segmentSize;
+        BYTE* const opStart3 = opStart2 + segmentSize;
+        BYTE* const opStart4 = opStart3 + segmentSize;
+        BYTE* op1 = ostart;
+        BYTE* op2 = opStart2;
+        BYTE* op3 = opStart3;
+        BYTE* op4 = opStart4;
+        U32 endSignal = 1;
+        DTableDesc const dtd = HUF_getDTableDesc(DTable);
+        U32 const dtLog = dtd.tableLog;
+
+        if (length4 > cSrcSize) return ERROR(corruption_detected);   /* overflow */
+        CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
+        CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
+        CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
+        CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );
+
+        /* 16-32 symbols per loop (4-8 symbols per stream) */
+        for ( ; (endSignal) & (op4 < olimit); ) {
+#if defined(__clang__) && (defined(__x86_64__) || defined(__i386__))
+            HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
+            endSignal &= BIT_reloadDStreamFast(&bitD1) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD2) == BIT_DStream_unfinished;
+            HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
+            endSignal &= BIT_reloadDStreamFast(&bitD3) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD4) == BIT_DStream_unfinished;
+#else
+            HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
+            endSignal = (U32)LIKELY((U32)
+                        (BIT_reloadDStreamFast(&bitD1) == BIT_DStream_unfinished)
+                      & (BIT_reloadDStreamFast(&bitD2) == BIT_DStream_unfinished)
+                      & (BIT_reloadDStreamFast(&bitD3) == BIT_DStream_unfinished)
+                      & (BIT_reloadDStreamFast(&bitD4) == BIT_DStream_unfinished));
+#endif
+        }
+
+        /* check corruption */
+        if (op1 > opStart2) return ERROR(corruption_detected);
+        if (op2 > opStart3) return ERROR(corruption_detected);
+        if (op3 > opStart4) return ERROR(corruption_detected);
+        /* note : op4 already verified within main loop */
+
+        /* finish bitStreams one by one */
+        HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog);
+        HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog);
+        HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog);
+        HUF_decodeStreamX2(op4, &bitD4, oend,     dt, dtLog);
+
+        /* check */
+        { U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
+          if (!endCheck) return ERROR(corruption_detected); }
+
+        /* decoded size */
+        return dstSize;
+    }
+}
+
+HUF_DGEN(HUF_decompress1X2_usingDTable_internal)
+HUF_DGEN(HUF_decompress4X2_usingDTable_internal)
+
+size_t HUF_decompress1X2_usingDTable(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    if (dtd.tableType != 1) return ERROR(GENERIC);
+    return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+}
+
+size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
+
+    size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize,
+                                               workSpace, wkspSize);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
+}
+
+
+size_t HUF_decompress4X2_usingDTable(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    if (dtd.tableType != 1) return ERROR(GENERIC);
+    return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+}
+
+static size_t HUF_decompress4X2_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize, int bmi2)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
+
+    size_t hSize = HUF_readDTableX2_wksp(dctx, cSrc, cSrcSize,
+                                         workSpace, wkspSize);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress4X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
+}
+
+size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize)
+{
+    return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, /* bmi2 */ 0);
+}
+
+
+#endif /* HUF_FORCE_DECOMPRESS_X1 */
+
+
+/* ***********************************/
+/* Universal decompression selectors */
+/* ***********************************/
+
+size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize,
+                                    const void* cSrc, size_t cSrcSize,
+                                    const HUF_DTable* DTable)
+{
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#else
+    return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
+                           HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#endif
+}
+
+size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize,
+                                    const void* cSrc, size_t cSrcSize,
+                                    const HUF_DTable* DTable)
+{
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#else
+    return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
+                           HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#endif
+}
+
+
+#if !defined(HUF_FORCE_DECOMPRESS_X1) && !defined(HUF_FORCE_DECOMPRESS_X2)
+typedef struct { U32 tableTime; U32 decode256Time; } algo_time_t;
+static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, quad */] =
+{
     /* single, double, quad */
-    {{0, 0}, {1, 1}, {2, 2}},		     /* Q==0 : impossible */
-    {{0, 0}, {1, 1}, {2, 2}},		     /* Q==1 : impossible */
-    {{38, 130}, {1313, 74}, {2151, 38}},     /* Q == 2 : 12-18% */
-    {{448, 128}, {1353, 74}, {2238, 41}},    /* Q == 3 : 18-25% */
-    {{556, 128}, {1353, 74}, {2238, 47}},    /* Q == 4 : 25-32% */
-    {{714, 128}, {1418, 74}, {2436, 53}},    /* Q == 5 : 32-38% */
-    {{883, 128}, {1437, 74}, {2464, 61}},    /* Q == 6 : 38-44% */
-    {{897, 128}, {1515, 75}, {2622, 68}},    /* Q == 7 : 44-50% */
-    {{926, 128}, {1613, 75}, {2730, 75}},    /* Q == 8 : 50-56% */
-    {{947, 128}, {1729, 77}, {3359, 77}},    /* Q == 9 : 56-62% */
-    {{1107, 128}, {2083, 81}, {4006, 84}},   /* Q ==10 : 62-69% */
-    {{1177, 128}, {2379, 87}, {4785, 88}},   /* Q ==11 : 69-75% */
-    {{1242, 128}, {2415, 93}, {5155, 84}},   /* Q ==12 : 75-81% */
-    {{1349, 128}, {2644, 106}, {5260, 106}}, /* Q ==13 : 81-87% */
-    {{1455, 128}, {2422, 124}, {4174, 124}}, /* Q ==14 : 87-93% */
-    {{722, 128}, {1891, 145}, {1936, 146}},  /* Q ==15 : 93-99% */
+    {{0,0}, {1,1}, {2,2}},  /* Q==0 : impossible */
+    {{0,0}, {1,1}, {2,2}},  /* Q==1 : impossible */
+    {{  38,130}, {1313, 74}, {2151, 38}},   /* Q == 2 : 12-18% */
+    {{ 448,128}, {1353, 74}, {2238, 41}},   /* Q == 3 : 18-25% */
+    {{ 556,128}, {1353, 74}, {2238, 47}},   /* Q == 4 : 25-32% */
+    {{ 714,128}, {1418, 74}, {2436, 53}},   /* Q == 5 : 32-38% */
+    {{ 883,128}, {1437, 74}, {2464, 61}},   /* Q == 6 : 38-44% */
+    {{ 897,128}, {1515, 75}, {2622, 68}},   /* Q == 7 : 44-50% */
+    {{ 926,128}, {1613, 75}, {2730, 75}},   /* Q == 8 : 50-56% */
+    {{ 947,128}, {1729, 77}, {3359, 77}},   /* Q == 9 : 56-62% */
+    {{1107,128}, {2083, 81}, {4006, 84}},   /* Q ==10 : 62-69% */
+    {{1177,128}, {2379, 87}, {4785, 88}},   /* Q ==11 : 69-75% */
+    {{1242,128}, {2415, 93}, {5155, 84}},   /* Q ==12 : 75-81% */
+    {{1349,128}, {2644,106}, {5260,106}},   /* Q ==13 : 81-87% */
+    {{1455,128}, {2422,124}, {4174,124}},   /* Q ==14 : 87-93% */
+    {{ 722,128}, {1891,145}, {1936,146}},   /* Q ==15 : 93-99% */
 };
+#endif
+
+/* HUF_selectDecoder() :
+ *  Tells which decoder is likely to decode faster,
+ *  based on a set of pre-computed metrics.
+ * @return : 0==HUF_decompress4X1, 1==HUF_decompress4X2 .
+ *  Assumption : 0 < dstSize <= 128 KB */
+U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize)
+{
+    assert(dstSize > 0);
+    assert(dstSize <= 128*1024);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dstSize;
+    (void)cSrcSize;
+    return 0;
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dstSize;
+    (void)cSrcSize;
+    return 1;
+#else
+    /* decoder timing evaluation */
+    {   U32 const Q = (cSrcSize >= dstSize) ? 15 : (U32)(cSrcSize * 16 / dstSize);   /* Q < 16 */
+        U32 const D256 = (U32)(dstSize >> 8);
+        U32 const DTime0 = algoTime[Q][0].tableTime + (algoTime[Q][0].decode256Time * D256);
+        U32 DTime1 = algoTime[Q][1].tableTime + (algoTime[Q][1].decode256Time * D256);
+        DTime1 += DTime1 >> 3;  /* advantage to algorithm using less memory, to reduce cache eviction */
+        return DTime1 < DTime0;
+    }
+#endif
+}
+
+
+size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst,
+                                     size_t dstSize, const void* cSrc,
+                                     size_t cSrcSize, void* workSpace,
+                                     size_t wkspSize)
+{
+    /* validation checks */
+    if (dstSize == 0) return ERROR(dstSize_tooSmall);
+    if (cSrcSize == 0) return ERROR(corruption_detected);
+
+    {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
+#else
+        return algoNb ? HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                            cSrcSize, workSpace, wkspSize):
+                        HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
+#endif
+    }
+}
+
+size_t HUF_decompress1X_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                  const void* cSrc, size_t cSrcSize,
+                                  void* workSpace, size_t wkspSize)
+{
+    /* validation checks */
+    if (dstSize == 0) return ERROR(dstSize_tooSmall);
+    if (cSrcSize > dstSize) return ERROR(corruption_detected);   /* invalid */
+    if (cSrcSize == dstSize) { ZSTD_memcpy(dst, cSrc, dstSize); return dstSize; }   /* not compressed */
+    if (cSrcSize == 1) { ZSTD_memset(dst, *(const BYTE*)cSrc, dstSize); return dstSize; }   /* RLE */
+
+    {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize);
+#else
+        return algoNb ? HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize):
+                        HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize);
+#endif
+    }
+}
+
+
+size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
+{
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#else
+    return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
+                           HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#endif
+}
+
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
 
-/** HUF_selectDecoder() :
-*   Tells which decoder is likely to decode faster,
-*   based on a set of pre-determined metrics.
-*   @return : 0==HUF_decompress4X2, 1==HUF_decompress4X4 .
-*   Assumption : 0 < cSrcSize < dstSize <= 128 KB */
-U32 HUF_selectDecoder(size_t dstSize, size_t cSrcSize)
-{
-	/* decoder timing evaluation */
-	U32 const Q = (U32)(cSrcSize * 16 / dstSize); /* Q < 16 since dstSize > cSrcSize */
-	U32 const D256 = (U32)(dstSize >> 8);
-	U32 const DTime0 = algoTime[Q][0].tableTime + (algoTime[Q][0].decode256Time * D256);
-	U32 DTime1 = algoTime[Q][1].tableTime + (algoTime[Q][1].decode256Time * D256);
-	DTime1 += DTime1 >> 3; /* advantage to algorithm using less memory, for cache eviction */
-
-	return DTime1 < DTime0;
-}
-
-typedef size_t (*decompressionAlgo)(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize);
-
-size_t HUF_decompress4X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	/* validation checks */
-	if (dstSize == 0)
-		return ERROR(dstSize_tooSmall);
-	if (cSrcSize > dstSize)
-		return ERROR(corruption_detected); /* invalid */
-	if (cSrcSize == dstSize) {
-		memcpy(dst, cSrc, dstSize);
-		return dstSize;
-	} /* not compressed */
-	if (cSrcSize == 1) {
-		memset(dst, *(const BYTE *)cSrc, dstSize);
-		return dstSize;
-	} /* RLE */
-
-	{
-		U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
-		return algoNb ? HUF_decompress4X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize)
-			      : HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize);
-	}
-}
-
-size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	/* validation checks */
-	if (dstSize == 0)
-		return ERROR(dstSize_tooSmall);
-	if ((cSrcSize >= dstSize) || (cSrcSize <= 1))
-		return ERROR(corruption_detected); /* invalid */
-
-	{
-		U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
-		return algoNb ? HUF_decompress4X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize)
-			      : HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize);
-	}
-}
-
-size_t HUF_decompress1X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	/* validation checks */
-	if (dstSize == 0)
-		return ERROR(dstSize_tooSmall);
-	if (cSrcSize > dstSize)
-		return ERROR(corruption_detected); /* invalid */
-	if (cSrcSize == dstSize) {
-		memcpy(dst, cSrc, dstSize);
-		return dstSize;
-	} /* not compressed */
-	if (cSrcSize == 1) {
-		memset(dst, *(const BYTE *)cSrc, dstSize);
-		return dstSize;
-	} /* RLE */
-
-	{
-		U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
-		return algoNb ? HUF_decompress1X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize)
-			      : HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize);
-	}
+    size_t const hSize = HUF_readDTableX1_wksp_bmi2(dctx, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress1X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
 }
+#endif
+
+size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
+{
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#else
+    return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
+                           HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#endif
+}
+
+size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2)
+{
+    /* validation checks */
+    if (dstSize == 0) return ERROR(dstSize_tooSmall);
+    if (cSrcSize == 0) return ERROR(corruption_detected);
+
+    {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+#else
+        return algoNb ? HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2) :
+                        HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+#endif
+    }
+}
+
diff --git a/lib/zstd/decompress/zstd_ddict.c b/lib/zstd/decompress/zstd_ddict.c
new file mode 100644
index 000000000000..dbbc7919de53
--- /dev/null
+++ b/lib/zstd/decompress/zstd_ddict.c
@@ -0,0 +1,241 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* zstd_ddict.c :
+ * concentrates all logic that needs to know the internals of ZSTD_DDict object */
+
+/*-*******************************************************
+*  Dependencies
+*********************************************************/
+#include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
+#include "../common/cpu.h"         /* bmi2 */
+#include "../common/mem.h"         /* low level memory routines */
+#define FSE_STATIC_LINKING_ONLY
+#include "../common/fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include "zstd_decompress_internal.h"
+#include "zstd_ddict.h"
+
+
+
+
+/*-*******************************************************
+*  Types
+*********************************************************/
+struct ZSTD_DDict_s {
+    void* dictBuffer;
+    const void* dictContent;
+    size_t dictSize;
+    ZSTD_entropyDTables_t entropy;
+    U32 dictID;
+    U32 entropyPresent;
+    ZSTD_customMem cMem;
+};  /* typedef'd to ZSTD_DDict within "zstd.h" */
+
+const void* ZSTD_DDict_dictContent(const ZSTD_DDict* ddict)
+{
+    assert(ddict != NULL);
+    return ddict->dictContent;
+}
+
+size_t ZSTD_DDict_dictSize(const ZSTD_DDict* ddict)
+{
+    assert(ddict != NULL);
+    return ddict->dictSize;
+}
+
+void ZSTD_copyDDictParameters(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
+{
+    DEBUGLOG(4, "ZSTD_copyDDictParameters");
+    assert(dctx != NULL);
+    assert(ddict != NULL);
+    dctx->dictID = ddict->dictID;
+    dctx->prefixStart = ddict->dictContent;
+    dctx->virtualStart = ddict->dictContent;
+    dctx->dictEnd = (const BYTE*)ddict->dictContent + ddict->dictSize;
+    dctx->previousDstEnd = dctx->dictEnd;
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    dctx->dictContentBeginForFuzzing = dctx->prefixStart;
+    dctx->dictContentEndForFuzzing = dctx->previousDstEnd;
+#endif
+    if (ddict->entropyPresent) {
+        dctx->litEntropy = 1;
+        dctx->fseEntropy = 1;
+        dctx->LLTptr = ddict->entropy.LLTable;
+        dctx->MLTptr = ddict->entropy.MLTable;
+        dctx->OFTptr = ddict->entropy.OFTable;
+        dctx->HUFptr = ddict->entropy.hufTable;
+        dctx->entropy.rep[0] = ddict->entropy.rep[0];
+        dctx->entropy.rep[1] = ddict->entropy.rep[1];
+        dctx->entropy.rep[2] = ddict->entropy.rep[2];
+    } else {
+        dctx->litEntropy = 0;
+        dctx->fseEntropy = 0;
+    }
+}
+
+
+static size_t
+ZSTD_loadEntropy_intoDDict(ZSTD_DDict* ddict,
+                           ZSTD_dictContentType_e dictContentType)
+{
+    ddict->dictID = 0;
+    ddict->entropyPresent = 0;
+    if (dictContentType == ZSTD_dct_rawContent) return 0;
+
+    if (ddict->dictSize < 8) {
+        if (dictContentType == ZSTD_dct_fullDict)
+            return ERROR(dictionary_corrupted);   /* only accept specified dictionaries */
+        return 0;   /* pure content mode */
+    }
+    {   U32 const magic = MEM_readLE32(ddict->dictContent);
+        if (magic != ZSTD_MAGIC_DICTIONARY) {
+            if (dictContentType == ZSTD_dct_fullDict)
+                return ERROR(dictionary_corrupted);   /* only accept specified dictionaries */
+            return 0;   /* pure content mode */
+        }
+    }
+    ddict->dictID = MEM_readLE32((const char*)ddict->dictContent + ZSTD_FRAMEIDSIZE);
+
+    /* load entropy tables */
+    RETURN_ERROR_IF(ZSTD_isError(ZSTD_loadDEntropy(
+            &ddict->entropy, ddict->dictContent, ddict->dictSize)),
+        dictionary_corrupted, "");
+    ddict->entropyPresent = 1;
+    return 0;
+}
+
+
+static size_t ZSTD_initDDict_internal(ZSTD_DDict* ddict,
+                                      const void* dict, size_t dictSize,
+                                      ZSTD_dictLoadMethod_e dictLoadMethod,
+                                      ZSTD_dictContentType_e dictContentType)
+{
+    if ((dictLoadMethod == ZSTD_dlm_byRef) || (!dict) || (!dictSize)) {
+        ddict->dictBuffer = NULL;
+        ddict->dictContent = dict;
+        if (!dict) dictSize = 0;
+    } else {
+        void* const internalBuffer = ZSTD_customMalloc(dictSize, ddict->cMem);
+        ddict->dictBuffer = internalBuffer;
+        ddict->dictContent = internalBuffer;
+        if (!internalBuffer) return ERROR(memory_allocation);
+        ZSTD_memcpy(internalBuffer, dict, dictSize);
+    }
+    ddict->dictSize = dictSize;
+    ddict->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001);  /* cover both little and big endian */
+
+    /* parse dictionary content */
+    FORWARD_IF_ERROR( ZSTD_loadEntropy_intoDDict(ddict, dictContentType) , "");
+
+    return 0;
+}
+
+ZSTD_DDict* ZSTD_createDDict_advanced(const void* dict, size_t dictSize,
+                                      ZSTD_dictLoadMethod_e dictLoadMethod,
+                                      ZSTD_dictContentType_e dictContentType,
+                                      ZSTD_customMem customMem)
+{
+    if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
+
+    {   ZSTD_DDict* const ddict = (ZSTD_DDict*) ZSTD_customMalloc(sizeof(ZSTD_DDict), customMem);
+        if (ddict == NULL) return NULL;
+        ddict->cMem = customMem;
+        {   size_t const initResult = ZSTD_initDDict_internal(ddict,
+                                            dict, dictSize,
+                                            dictLoadMethod, dictContentType);
+            if (ZSTD_isError(initResult)) {
+                ZSTD_freeDDict(ddict);
+                return NULL;
+        }   }
+        return ddict;
+    }
+}
+
+/*! ZSTD_createDDict() :
+*   Create a digested dictionary, to start decompression without startup delay.
+*   `dict` content is copied inside DDict.
+*   Consequently, `dict` can be released after `ZSTD_DDict` creation */
+ZSTD_DDict* ZSTD_createDDict(const void* dict, size_t dictSize)
+{
+    ZSTD_customMem const allocator = { NULL, NULL, NULL };
+    return ZSTD_createDDict_advanced(dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto, allocator);
+}
+
+/*! ZSTD_createDDict_byReference() :
+ *  Create a digested dictionary, to start decompression without startup delay.
+ *  Dictionary content is simply referenced, it will be accessed during decompression.
+ *  Warning : dictBuffer must outlive DDict (DDict must be freed before dictBuffer) */
+ZSTD_DDict* ZSTD_createDDict_byReference(const void* dictBuffer, size_t dictSize)
+{
+    ZSTD_customMem const allocator = { NULL, NULL, NULL };
+    return ZSTD_createDDict_advanced(dictBuffer, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto, allocator);
+}
+
+
+const ZSTD_DDict* ZSTD_initStaticDDict(
+                                void* sBuffer, size_t sBufferSize,
+                                const void* dict, size_t dictSize,
+                                ZSTD_dictLoadMethod_e dictLoadMethod,
+                                ZSTD_dictContentType_e dictContentType)
+{
+    size_t const neededSpace = sizeof(ZSTD_DDict)
+                             + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize);
+    ZSTD_DDict* const ddict = (ZSTD_DDict*)sBuffer;
+    assert(sBuffer != NULL);
+    assert(dict != NULL);
+    if ((size_t)sBuffer & 7) return NULL;   /* 8-aligned */
+    if (sBufferSize < neededSpace) return NULL;
+    if (dictLoadMethod == ZSTD_dlm_byCopy) {
+        ZSTD_memcpy(ddict+1, dict, dictSize);  /* local copy */
+        dict = ddict+1;
+    }
+    if (ZSTD_isError( ZSTD_initDDict_internal(ddict,
+                                              dict, dictSize,
+                                              ZSTD_dlm_byRef, dictContentType) ))
+        return NULL;
+    return ddict;
+}
+
+
+size_t ZSTD_freeDDict(ZSTD_DDict* ddict)
+{
+    if (ddict==NULL) return 0;   /* support free on NULL */
+    {   ZSTD_customMem const cMem = ddict->cMem;
+        ZSTD_customFree(ddict->dictBuffer, cMem);
+        ZSTD_customFree(ddict, cMem);
+        return 0;
+    }
+}
+
+/*! ZSTD_estimateDDictSize() :
+ *  Estimate amount of memory that will be needed to create a dictionary for decompression.
+ *  Note : dictionary created by reference using ZSTD_dlm_byRef are smaller */
+size_t ZSTD_estimateDDictSize(size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod)
+{
+    return sizeof(ZSTD_DDict) + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize);
+}
+
+size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict)
+{
+    if (ddict==NULL) return 0;   /* support sizeof on NULL */
+    return sizeof(*ddict) + (ddict->dictBuffer ? ddict->dictSize : 0) ;
+}
+
+/*! ZSTD_getDictID_fromDDict() :
+ *  Provides the dictID of the dictionary loaded into `ddict`.
+ *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
+ *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
+unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict* ddict)
+{
+    if (ddict==NULL) return 0;
+    return ZSTD_getDictID_fromDict(ddict->dictContent, ddict->dictSize);
+}
diff --git a/lib/zstd/decompress/zstd_ddict.h b/lib/zstd/decompress/zstd_ddict.h
new file mode 100644
index 000000000000..8c1a79d666f8
--- /dev/null
+++ b/lib/zstd/decompress/zstd_ddict.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+
+#ifndef ZSTD_DDICT_H
+#define ZSTD_DDICT_H
+
+/*-*******************************************************
+ *  Dependencies
+ *********************************************************/
+#include "../common/zstd_deps.h"   /* size_t */
+#include <linux/zstd.h>     /* ZSTD_DDict, and several public functions */
+
+
+/*-*******************************************************
+ *  Interface
+ *********************************************************/
+
+/* note: several prototypes are already published in `zstd.h` :
+ * ZSTD_createDDict()
+ * ZSTD_createDDict_byReference()
+ * ZSTD_createDDict_advanced()
+ * ZSTD_freeDDict()
+ * ZSTD_initStaticDDict()
+ * ZSTD_sizeof_DDict()
+ * ZSTD_estimateDDictSize()
+ * ZSTD_getDictID_fromDict()
+ */
+
+const void* ZSTD_DDict_dictContent(const ZSTD_DDict* ddict);
+size_t ZSTD_DDict_dictSize(const ZSTD_DDict* ddict);
+
+void ZSTD_copyDDictParameters(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);
+
+
+
+#endif /* ZSTD_DDICT_H */
diff --git a/lib/zstd/decompress/zstd_decompress.c b/lib/zstd/decompress/zstd_decompress.c
new file mode 100644
index 000000000000..b4d81d84479a
--- /dev/null
+++ b/lib/zstd/decompress/zstd_decompress.c
@@ -0,0 +1,2085 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+
+/* ***************************************************************
+*  Tuning parameters
+*****************************************************************/
+/*!
+ * HEAPMODE :
+ * Select how default decompression function ZSTD_decompress() allocates its context,
+ * on stack (0), or into heap (1, default; requires malloc()).
+ * Note that functions with explicit context such as ZSTD_decompressDCtx() are unaffected.
+ */
+#ifndef ZSTD_HEAPMODE
+#  define ZSTD_HEAPMODE 1
+#endif
+
+/*!
+*  LEGACY_SUPPORT :
+*  if set to 1+, ZSTD_decompress() can decode older formats (v0.1+)
+*/
+
+/*!
+ *  MAXWINDOWSIZE_DEFAULT :
+ *  maximum window size accepted by DStream __by default__.
+ *  Frames requiring more memory will be rejected.
+ *  It's possible to set a different limit using ZSTD_DCtx_setMaxWindowSize().
+ */
+#ifndef ZSTD_MAXWINDOWSIZE_DEFAULT
+#  define ZSTD_MAXWINDOWSIZE_DEFAULT (((U32)1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT) + 1)
+#endif
+
+/*!
+ *  NO_FORWARD_PROGRESS_MAX :
+ *  maximum allowed nb of calls to ZSTD_decompressStream()
+ *  without any forward progress
+ *  (defined as: no byte read from input, and no byte flushed to output)
+ *  before triggering an error.
+ */
+#ifndef ZSTD_NO_FORWARD_PROGRESS_MAX
+#  define ZSTD_NO_FORWARD_PROGRESS_MAX 16
+#endif
+
+
+/*-*******************************************************
+*  Dependencies
+*********************************************************/
+#include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
+#include "../common/cpu.h"         /* bmi2 */
+#include "../common/mem.h"         /* low level memory routines */
+#define FSE_STATIC_LINKING_ONLY
+#include "../common/fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include <linux/xxhash.h> /* xxh64_reset, xxh64_update, xxh64_digest, XXH64 */
+#include "../common/zstd_internal.h"  /* blockProperties_t */
+#include "zstd_decompress_internal.h"   /* ZSTD_DCtx */
+#include "zstd_ddict.h"  /* ZSTD_DDictDictContent */
+#include "zstd_decompress_block.h"   /* ZSTD_decompressBlock_internal */
+
+
+
+
+/* ***********************************
+ * Multiple DDicts Hashset internals *
+ *************************************/
+
+#define DDICT_HASHSET_MAX_LOAD_FACTOR_COUNT_MULT 4
+#define DDICT_HASHSET_MAX_LOAD_FACTOR_SIZE_MULT 3   /* These two constants represent SIZE_MULT/COUNT_MULT load factor without using a float.
+                                                     * Currently, that means a 0.75 load factor.
+                                                     * So, if count * COUNT_MULT / size * SIZE_MULT != 0, then we've exceeded
+                                                     * the load factor of the ddict hash set.
+                                                     */
+
+#define DDICT_HASHSET_TABLE_BASE_SIZE 64
+#define DDICT_HASHSET_RESIZE_FACTOR 2
+
+/* Hash function to determine starting position of dict insertion within the table
+ * Returns an index between [0, hashSet->ddictPtrTableSize]
+ */
+static size_t ZSTD_DDictHashSet_getIndex(const ZSTD_DDictHashSet* hashSet, U32 dictID) {
+    const U64 hash = xxh64(&dictID, sizeof(U32), 0);
+    /* DDict ptr table size is a multiple of 2, use size - 1 as mask to get index within [0, hashSet->ddictPtrTableSize) */
+    return hash & (hashSet->ddictPtrTableSize - 1);
+}
+
+/* Adds DDict to a hashset without resizing it.
+ * If inserting a DDict with a dictID that already exists in the set, replaces the one in the set.
+ * Returns 0 if successful, or a zstd error code if something went wrong.
+ */
+static size_t ZSTD_DDictHashSet_emplaceDDict(ZSTD_DDictHashSet* hashSet, const ZSTD_DDict* ddict) {
+    const U32 dictID = ZSTD_getDictID_fromDDict(ddict);
+    size_t idx = ZSTD_DDictHashSet_getIndex(hashSet, dictID);
+    const size_t idxRangeMask = hashSet->ddictPtrTableSize - 1;
+    RETURN_ERROR_IF(hashSet->ddictPtrCount == hashSet->ddictPtrTableSize, GENERIC, "Hash set is full!");
+    DEBUGLOG(4, "Hashed index: for dictID: %u is %zu", dictID, idx);
+    while (hashSet->ddictPtrTable[idx] != NULL) {
+        /* Replace existing ddict if inserting ddict with same dictID */
+        if (ZSTD_getDictID_fromDDict(hashSet->ddictPtrTable[idx]) == dictID) {
+            DEBUGLOG(4, "DictID already exists, replacing rather than adding");
+            hashSet->ddictPtrTable[idx] = ddict;
+            return 0;
+        }
+        idx &= idxRangeMask;
+        idx++;
+    }
+    DEBUGLOG(4, "Final idx after probing for dictID %u is: %zu", dictID, idx);
+    hashSet->ddictPtrTable[idx] = ddict;
+    hashSet->ddictPtrCount++;
+    return 0;
+}
+
+/* Expands hash table by factor of DDICT_HASHSET_RESIZE_FACTOR and
+ * rehashes all values, allocates new table, frees old table.
+ * Returns 0 on success, otherwise a zstd error code.
+ */
+static size_t ZSTD_DDictHashSet_expand(ZSTD_DDictHashSet* hashSet, ZSTD_customMem customMem) {
+    size_t newTableSize = hashSet->ddictPtrTableSize * DDICT_HASHSET_RESIZE_FACTOR;
+    const ZSTD_DDict** newTable = (const ZSTD_DDict**)ZSTD_customCalloc(sizeof(ZSTD_DDict*) * newTableSize, customMem);
+    const ZSTD_DDict** oldTable = hashSet->ddictPtrTable;
+    size_t oldTableSize = hashSet->ddictPtrTableSize;
+    size_t i;
+
+    DEBUGLOG(4, "Expanding DDict hash table! Old size: %zu new size: %zu", oldTableSize, newTableSize);
+    RETURN_ERROR_IF(!newTable, memory_allocation, "Expanded hashset allocation failed!");
+    hashSet->ddictPtrTable = newTable;
+    hashSet->ddictPtrTableSize = newTableSize;
+    hashSet->ddictPtrCount = 0;
+    for (i = 0; i < oldTableSize; ++i) {
+        if (oldTable[i] != NULL) {
+            FORWARD_IF_ERROR(ZSTD_DDictHashSet_emplaceDDict(hashSet, oldTable[i]), "");
+        }
+    }
+    ZSTD_customFree((void*)oldTable, customMem);
+    DEBUGLOG(4, "Finished re-hash");
+    return 0;
+}
+
+/* Fetches a DDict with the given dictID
+ * Returns the ZSTD_DDict* with the requested dictID. If it doesn't exist, then returns NULL.
+ */
+static const ZSTD_DDict* ZSTD_DDictHashSet_getDDict(ZSTD_DDictHashSet* hashSet, U32 dictID) {
+    size_t idx = ZSTD_DDictHashSet_getIndex(hashSet, dictID);
+    const size_t idxRangeMask = hashSet->ddictPtrTableSize - 1;
+    DEBUGLOG(4, "Hashed index: for dictID: %u is %zu", dictID, idx);
+    for (;;) {
+        size_t currDictID = ZSTD_getDictID_fromDDict(hashSet->ddictPtrTable[idx]);
+        if (currDictID == dictID || currDictID == 0) {
+            /* currDictID == 0 implies a NULL ddict entry */
+            break;
+        } else {
+            idx &= idxRangeMask;    /* Goes to start of table when we reach the end */
+            idx++;
+        }
+    }
+    DEBUGLOG(4, "Final idx after probing for dictID %u is: %zu", dictID, idx);
+    return hashSet->ddictPtrTable[idx];
+}
+
+/* Allocates space for and returns a ddict hash set
+ * The hash set's ZSTD_DDict* table has all values automatically set to NULL to begin with.
+ * Returns NULL if allocation failed.
+ */
+static ZSTD_DDictHashSet* ZSTD_createDDictHashSet(ZSTD_customMem customMem) {
+    ZSTD_DDictHashSet* ret = (ZSTD_DDictHashSet*)ZSTD_customMalloc(sizeof(ZSTD_DDictHashSet), customMem);
+    DEBUGLOG(4, "Allocating new hash set");
+    if (!ret)
+        return NULL;
+    ret->ddictPtrTable = (const ZSTD_DDict**)ZSTD_customCalloc(DDICT_HASHSET_TABLE_BASE_SIZE * sizeof(ZSTD_DDict*), customMem);
+    if (!ret->ddictPtrTable) {
+        ZSTD_customFree(ret, customMem);
+        return NULL;
+    }
+    ret->ddictPtrTableSize = DDICT_HASHSET_TABLE_BASE_SIZE;
+    ret->ddictPtrCount = 0;
+    return ret;
+}
+
+/* Frees the table of ZSTD_DDict* within a hashset, then frees the hashset itself.
+ * Note: The ZSTD_DDict* within the table are NOT freed.
+ */
+static void ZSTD_freeDDictHashSet(ZSTD_DDictHashSet* hashSet, ZSTD_customMem customMem) {
+    DEBUGLOG(4, "Freeing ddict hash set");
+    if (hashSet && hashSet->ddictPtrTable) {
+        ZSTD_customFree((void*)hashSet->ddictPtrTable, customMem);
+    }
+    if (hashSet) {
+        ZSTD_customFree(hashSet, customMem);
+    }
+}
+
+/* Public function: Adds a DDict into the ZSTD_DDictHashSet, possibly triggering a resize of the hash set.
+ * Returns 0 on success, or a ZSTD error.
+ */
+static size_t ZSTD_DDictHashSet_addDDict(ZSTD_DDictHashSet* hashSet, const ZSTD_DDict* ddict, ZSTD_customMem customMem) {
+    DEBUGLOG(4, "Adding dict ID: %u to hashset with - Count: %zu Tablesize: %zu", ZSTD_getDictID_fromDDict(ddict), hashSet->ddictPtrCount, hashSet->ddictPtrTableSize);
+    if (hashSet->ddictPtrCount * DDICT_HASHSET_MAX_LOAD_FACTOR_COUNT_MULT / hashSet->ddictPtrTableSize * DDICT_HASHSET_MAX_LOAD_FACTOR_SIZE_MULT != 0) {
+        FORWARD_IF_ERROR(ZSTD_DDictHashSet_expand(hashSet, customMem), "");
+    }
+    FORWARD_IF_ERROR(ZSTD_DDictHashSet_emplaceDDict(hashSet, ddict), "");
+    return 0;
+}
+
+/*-*************************************************************
+*   Context management
+***************************************************************/
+size_t ZSTD_sizeof_DCtx (const ZSTD_DCtx* dctx)
+{
+    if (dctx==NULL) return 0;   /* support sizeof NULL */
+    return sizeof(*dctx)
+           + ZSTD_sizeof_DDict(dctx->ddictLocal)
+           + dctx->inBuffSize + dctx->outBuffSize;
+}
+
+size_t ZSTD_estimateDCtxSize(void) { return sizeof(ZSTD_DCtx); }
+
+
+static size_t ZSTD_startingInputLength(ZSTD_format_e format)
+{
+    size_t const startingInputLength = ZSTD_FRAMEHEADERSIZE_PREFIX(format);
+    /* only supports formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless */
+    assert( (format == ZSTD_f_zstd1) || (format == ZSTD_f_zstd1_magicless) );
+    return startingInputLength;
+}
+
+static void ZSTD_DCtx_resetParameters(ZSTD_DCtx* dctx)
+{
+    assert(dctx->streamStage == zdss_init);
+    dctx->format = ZSTD_f_zstd1;
+    dctx->maxWindowSize = ZSTD_MAXWINDOWSIZE_DEFAULT;
+    dctx->outBufferMode = ZSTD_bm_buffered;
+    dctx->forceIgnoreChecksum = ZSTD_d_validateChecksum;
+    dctx->refMultipleDDicts = ZSTD_rmd_refSingleDDict;
+}
+
+static void ZSTD_initDCtx_internal(ZSTD_DCtx* dctx)
+{
+    dctx->staticSize  = 0;
+    dctx->ddict       = NULL;
+    dctx->ddictLocal  = NULL;
+    dctx->dictEnd     = NULL;
+    dctx->ddictIsCold = 0;
+    dctx->dictUses = ZSTD_dont_use;
+    dctx->inBuff      = NULL;
+    dctx->inBuffSize  = 0;
+    dctx->outBuffSize = 0;
+    dctx->streamStage = zdss_init;
+    dctx->legacyContext = NULL;
+    dctx->previousLegacyVersion = 0;
+    dctx->noForwardProgress = 0;
+    dctx->oversizedDuration = 0;
+    dctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
+    dctx->ddictSet = NULL;
+    ZSTD_DCtx_resetParameters(dctx);
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    dctx->dictContentEndForFuzzing = NULL;
+#endif
+}
+
+ZSTD_DCtx* ZSTD_initStaticDCtx(void *workspace, size_t workspaceSize)
+{
+    ZSTD_DCtx* const dctx = (ZSTD_DCtx*) workspace;
+
+    if ((size_t)workspace & 7) return NULL;  /* 8-aligned */
+    if (workspaceSize < sizeof(ZSTD_DCtx)) return NULL;  /* minimum size */
+
+    ZSTD_initDCtx_internal(dctx);
+    dctx->staticSize = workspaceSize;
+    dctx->inBuff = (char*)(dctx+1);
+    return dctx;
+}
+
+ZSTD_DCtx* ZSTD_createDCtx_advanced(ZSTD_customMem customMem)
+{
+    if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
+
+    {   ZSTD_DCtx* const dctx = (ZSTD_DCtx*)ZSTD_customMalloc(sizeof(*dctx), customMem);
+        if (!dctx) return NULL;
+        dctx->customMem = customMem;
+        ZSTD_initDCtx_internal(dctx);
+        return dctx;
+    }
+}
+
+ZSTD_DCtx* ZSTD_createDCtx(void)
+{
+    DEBUGLOG(3, "ZSTD_createDCtx");
+    return ZSTD_createDCtx_advanced(ZSTD_defaultCMem);
+}
+
+static void ZSTD_clearDict(ZSTD_DCtx* dctx)
+{
+    ZSTD_freeDDict(dctx->ddictLocal);
+    dctx->ddictLocal = NULL;
+    dctx->ddict = NULL;
+    dctx->dictUses = ZSTD_dont_use;
+}
+
+size_t ZSTD_freeDCtx(ZSTD_DCtx* dctx)
+{
+    if (dctx==NULL) return 0;   /* support free on NULL */
+    RETURN_ERROR_IF(dctx->staticSize, memory_allocation, "not compatible with static DCtx");
+    {   ZSTD_customMem const cMem = dctx->customMem;
+        ZSTD_clearDict(dctx);
+        ZSTD_customFree(dctx->inBuff, cMem);
+        dctx->inBuff = NULL;
+        if (dctx->ddictSet) {
+            ZSTD_freeDDictHashSet(dctx->ddictSet, cMem);
+            dctx->ddictSet = NULL;
+        }
+        ZSTD_customFree(dctx, cMem);
+        return 0;
+    }
+}
+
+/* no longer useful */
+void ZSTD_copyDCtx(ZSTD_DCtx* dstDCtx, const ZSTD_DCtx* srcDCtx)
+{
+    size_t const toCopy = (size_t)((char*)(&dstDCtx->inBuff) - (char*)dstDCtx);
+    ZSTD_memcpy(dstDCtx, srcDCtx, toCopy);  /* no need to copy workspace */
+}
+
+/* Given a dctx with a digested frame params, re-selects the correct ZSTD_DDict based on
+ * the requested dict ID from the frame. If there exists a reference to the correct ZSTD_DDict, then
+ * accordingly sets the ddict to be used to decompress the frame.
+ *
+ * If no DDict is found, then no action is taken, and the ZSTD_DCtx::ddict remains as-is.
+ *
+ * ZSTD_d_refMultipleDDicts must be enabled for this function to be called.
+ */
+static void ZSTD_DCtx_selectFrameDDict(ZSTD_DCtx* dctx) {
+    assert(dctx->refMultipleDDicts && dctx->ddictSet);
+    DEBUGLOG(4, "Adjusting DDict based on requested dict ID from frame");
+    if (dctx->ddict) {
+        const ZSTD_DDict* frameDDict = ZSTD_DDictHashSet_getDDict(dctx->ddictSet, dctx->fParams.dictID);
+        if (frameDDict) {
+            DEBUGLOG(4, "DDict found!");
+            ZSTD_clearDict(dctx);
+            dctx->dictID = dctx->fParams.dictID;
+            dctx->ddict = frameDDict;
+            dctx->dictUses = ZSTD_use_indefinitely;
+        }
+    }
+}
+
+
+/*-*************************************************************
+ *   Frame header decoding
+ ***************************************************************/
+
+/*! ZSTD_isFrame() :
+ *  Tells if the content of `buffer` starts with a valid Frame Identifier.
+ *  Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
+ *  Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
+ *  Note 3 : Skippable Frame Identifiers are considered valid. */
+unsigned ZSTD_isFrame(const void* buffer, size_t size)
+{
+    if (size < ZSTD_FRAMEIDSIZE) return 0;
+    {   U32 const magic = MEM_readLE32(buffer);
+        if (magic == ZSTD_MAGICNUMBER) return 1;
+        if ((magic & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) return 1;
+    }
+    return 0;
+}
+
+/* ZSTD_frameHeaderSize_internal() :
+ *  srcSize must be large enough to reach header size fields.
+ *  note : only works for formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless.
+ * @return : size of the Frame Header
+ *           or an error code, which can be tested with ZSTD_isError() */
+static size_t ZSTD_frameHeaderSize_internal(const void* src, size_t srcSize, ZSTD_format_e format)
+{
+    size_t const minInputSize = ZSTD_startingInputLength(format);
+    RETURN_ERROR_IF(srcSize < minInputSize, srcSize_wrong, "");
+
+    {   BYTE const fhd = ((const BYTE*)src)[minInputSize-1];
+        U32 const dictID= fhd & 3;
+        U32 const singleSegment = (fhd >> 5) & 1;
+        U32 const fcsId = fhd >> 6;
+        return minInputSize + !singleSegment
+             + ZSTD_did_fieldSize[dictID] + ZSTD_fcs_fieldSize[fcsId]
+             + (singleSegment && !fcsId);
+    }
+}
+
+/* ZSTD_frameHeaderSize() :
+ *  srcSize must be >= ZSTD_frameHeaderSize_prefix.
+ * @return : size of the Frame Header,
+ *           or an error code (if srcSize is too small) */
+size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize)
+{
+    return ZSTD_frameHeaderSize_internal(src, srcSize, ZSTD_f_zstd1);
+}
+
+
+/* ZSTD_getFrameHeader_advanced() :
+ *  decode Frame Header, or require larger `srcSize`.
+ *  note : only works for formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless
+ * @return : 0, `zfhPtr` is correctly filled,
+ *          >0, `srcSize` is too small, value is wanted `srcSize` amount,
+ *           or an error code, which can be tested using ZSTD_isError() */
+size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format)
+{
+    const BYTE* ip = (const BYTE*)src;
+    size_t const minInputSize = ZSTD_startingInputLength(format);
+
+    ZSTD_memset(zfhPtr, 0, sizeof(*zfhPtr));   /* not strictly necessary, but static analyzer do not understand that zfhPtr is only going to be read only if return value is zero, since they are 2 different signals */
+    if (srcSize < minInputSize) return minInputSize;
+    RETURN_ERROR_IF(src==NULL, GENERIC, "invalid parameter");
+
+    if ( (format != ZSTD_f_zstd1_magicless)
+      && (MEM_readLE32(src) != ZSTD_MAGICNUMBER) ) {
+        if ((MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+            /* skippable frame */
+            if (srcSize < ZSTD_SKIPPABLEHEADERSIZE)
+                return ZSTD_SKIPPABLEHEADERSIZE; /* magic number + frame length */
+            ZSTD_memset(zfhPtr, 0, sizeof(*zfhPtr));
+            zfhPtr->frameContentSize = MEM_readLE32((const char *)src + ZSTD_FRAMEIDSIZE);
+            zfhPtr->frameType = ZSTD_skippableFrame;
+            return 0;
+        }
+        RETURN_ERROR(prefix_unknown, "");
+    }
+
+    /* ensure there is enough `srcSize` to fully read/decode frame header */
+    {   size_t const fhsize = ZSTD_frameHeaderSize_internal(src, srcSize, format);
+        if (srcSize < fhsize) return fhsize;
+        zfhPtr->headerSize = (U32)fhsize;
+    }
+
+    {   BYTE const fhdByte = ip[minInputSize-1];
+        size_t pos = minInputSize;
+        U32 const dictIDSizeCode = fhdByte&3;
+        U32 const checksumFlag = (fhdByte>>2)&1;
+        U32 const singleSegment = (fhdByte>>5)&1;
+        U32 const fcsID = fhdByte>>6;
+        U64 windowSize = 0;
+        U32 dictID = 0;
+        U64 frameContentSize = ZSTD_CONTENTSIZE_UNKNOWN;
+        RETURN_ERROR_IF((fhdByte & 0x08) != 0, frameParameter_unsupported,
+                        "reserved bits, must be zero");
+
+        if (!singleSegment) {
+            BYTE const wlByte = ip[pos++];
+            U32 const windowLog = (wlByte >> 3) + ZSTD_WINDOWLOG_ABSOLUTEMIN;
+            RETURN_ERROR_IF(windowLog > ZSTD_WINDOWLOG_MAX, frameParameter_windowTooLarge, "");
+            windowSize = (1ULL << windowLog);
+            windowSize += (windowSize >> 3) * (wlByte&7);
+        }
+        switch(dictIDSizeCode)
+        {
+            default:
+                assert(0);  /* impossible */
+                ZSTD_FALLTHROUGH;
+            case 0 : break;
+            case 1 : dictID = ip[pos]; pos++; break;
+            case 2 : dictID = MEM_readLE16(ip+pos); pos+=2; break;
+            case 3 : dictID = MEM_readLE32(ip+pos); pos+=4; break;
+        }
+        switch(fcsID)
+        {
+            default:
+                assert(0);  /* impossible */
+                ZSTD_FALLTHROUGH;
+            case 0 : if (singleSegment) frameContentSize = ip[pos]; break;
+            case 1 : frameContentSize = MEM_readLE16(ip+pos)+256; break;
+            case 2 : frameContentSize = MEM_readLE32(ip+pos); break;
+            case 3 : frameContentSize = MEM_readLE64(ip+pos); break;
+        }
+        if (singleSegment) windowSize = frameContentSize;
+
+        zfhPtr->frameType = ZSTD_frame;
+        zfhPtr->frameContentSize = frameContentSize;
+        zfhPtr->windowSize = windowSize;
+        zfhPtr->blockSizeMax = (unsigned) MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
+        zfhPtr->dictID = dictID;
+        zfhPtr->checksumFlag = checksumFlag;
+    }
+    return 0;
+}
+
+/* ZSTD_getFrameHeader() :
+ *  decode Frame Header, or require larger `srcSize`.
+ *  note : this function does not consume input, it only reads it.
+ * @return : 0, `zfhPtr` is correctly filled,
+ *          >0, `srcSize` is too small, value is wanted `srcSize` amount,
+ *           or an error code, which can be tested using ZSTD_isError() */
+size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize)
+{
+    return ZSTD_getFrameHeader_advanced(zfhPtr, src, srcSize, ZSTD_f_zstd1);
+}
+
+
+/* ZSTD_getFrameContentSize() :
+ *  compatible with legacy mode
+ * @return : decompressed size of the single frame pointed to be `src` if known, otherwise
+ *         - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
+ *         - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small) */
+unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize)
+{
+    {   ZSTD_frameHeader zfh;
+        if (ZSTD_getFrameHeader(&zfh, src, srcSize) != 0)
+            return ZSTD_CONTENTSIZE_ERROR;
+        if (zfh.frameType == ZSTD_skippableFrame) {
+            return 0;
+        } else {
+            return zfh.frameContentSize;
+    }   }
+}
+
+static size_t readSkippableFrameSize(void const* src, size_t srcSize)
+{
+    size_t const skippableHeaderSize = ZSTD_SKIPPABLEHEADERSIZE;
+    U32 sizeU32;
+
+    RETURN_ERROR_IF(srcSize < ZSTD_SKIPPABLEHEADERSIZE, srcSize_wrong, "");
+
+    sizeU32 = MEM_readLE32((BYTE const*)src + ZSTD_FRAMEIDSIZE);
+    RETURN_ERROR_IF((U32)(sizeU32 + ZSTD_SKIPPABLEHEADERSIZE) < sizeU32,
+                    frameParameter_unsupported, "");
+    {
+        size_t const skippableSize = skippableHeaderSize + sizeU32;
+        RETURN_ERROR_IF(skippableSize > srcSize, srcSize_wrong, "");
+        return skippableSize;
+    }
+}
+
+/* ZSTD_findDecompressedSize() :
+ *  compatible with legacy mode
+ *  `srcSize` must be the exact length of some number of ZSTD compressed and/or
+ *      skippable frames
+ *  @return : decompressed size of the frames contained */
+unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)
+{
+    unsigned long long totalDstSize = 0;
+
+    while (srcSize >= ZSTD_startingInputLength(ZSTD_f_zstd1)) {
+        U32 const magicNumber = MEM_readLE32(src);
+
+        if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+            size_t const skippableSize = readSkippableFrameSize(src, srcSize);
+            if (ZSTD_isError(skippableSize)) {
+                return ZSTD_CONTENTSIZE_ERROR;
+            }
+            assert(skippableSize <= srcSize);
+
+            src = (const BYTE *)src + skippableSize;
+            srcSize -= skippableSize;
+            continue;
+        }
+
+        {   unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
+            if (ret >= ZSTD_CONTENTSIZE_ERROR) return ret;
+
+            /* check for overflow */
+            if (totalDstSize + ret < totalDstSize) return ZSTD_CONTENTSIZE_ERROR;
+            totalDstSize += ret;
+        }
+        {   size_t const frameSrcSize = ZSTD_findFrameCompressedSize(src, srcSize);
+            if (ZSTD_isError(frameSrcSize)) {
+                return ZSTD_CONTENTSIZE_ERROR;
+            }
+
+            src = (const BYTE *)src + frameSrcSize;
+            srcSize -= frameSrcSize;
+        }
+    }  /* while (srcSize >= ZSTD_frameHeaderSize_prefix) */
+
+    if (srcSize) return ZSTD_CONTENTSIZE_ERROR;
+
+    return totalDstSize;
+}
+
+/* ZSTD_getDecompressedSize() :
+ *  compatible with legacy mode
+ * @return : decompressed size if known, 0 otherwise
+             note : 0 can mean any of the following :
+                   - frame content is empty
+                   - decompressed size field is not present in frame header
+                   - frame header unknown / not supported
+                   - frame header not complete (`srcSize` too small) */
+unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize)
+{
+    unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
+    ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_ERROR < ZSTD_CONTENTSIZE_UNKNOWN);
+    return (ret >= ZSTD_CONTENTSIZE_ERROR) ? 0 : ret;
+}
+
+
+/* ZSTD_decodeFrameHeader() :
+ * `headerSize` must be the size provided by ZSTD_frameHeaderSize().
+ * If multiple DDict references are enabled, also will choose the correct DDict to use.
+ * @return : 0 if success, or an error code, which can be tested using ZSTD_isError() */
+static size_t ZSTD_decodeFrameHeader(ZSTD_DCtx* dctx, const void* src, size_t headerSize)
+{
+    size_t const result = ZSTD_getFrameHeader_advanced(&(dctx->fParams), src, headerSize, dctx->format);
+    if (ZSTD_isError(result)) return result;    /* invalid header */
+    RETURN_ERROR_IF(result>0, srcSize_wrong, "headerSize too small");
+
+    /* Reference DDict requested by frame if dctx references multiple ddicts */
+    if (dctx->refMultipleDDicts == ZSTD_rmd_refMultipleDDicts && dctx->ddictSet) {
+        ZSTD_DCtx_selectFrameDDict(dctx);
+    }
+
+#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    /* Skip the dictID check in fuzzing mode, because it makes the search
+     * harder.
+     */
+    RETURN_ERROR_IF(dctx->fParams.dictID && (dctx->dictID != dctx->fParams.dictID),
+                    dictionary_wrong, "");
+#endif
+    dctx->validateChecksum = (dctx->fParams.checksumFlag && !dctx->forceIgnoreChecksum) ? 1 : 0;
+    if (dctx->validateChecksum) xxh64_reset(&dctx->xxhState, 0);
+    dctx->processedCSize += headerSize;
+    return 0;
+}
+
+static ZSTD_frameSizeInfo ZSTD_errorFrameSizeInfo(size_t ret)
+{
+    ZSTD_frameSizeInfo frameSizeInfo;
+    frameSizeInfo.compressedSize = ret;
+    frameSizeInfo.decompressedBound = ZSTD_CONTENTSIZE_ERROR;
+    return frameSizeInfo;
+}
+
+static ZSTD_frameSizeInfo ZSTD_findFrameSizeInfo(const void* src, size_t srcSize)
+{
+    ZSTD_frameSizeInfo frameSizeInfo;
+    ZSTD_memset(&frameSizeInfo, 0, sizeof(ZSTD_frameSizeInfo));
+
+
+    if ((srcSize >= ZSTD_SKIPPABLEHEADERSIZE)
+        && (MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+        frameSizeInfo.compressedSize = readSkippableFrameSize(src, srcSize);
+        assert(ZSTD_isError(frameSizeInfo.compressedSize) ||
+               frameSizeInfo.compressedSize <= srcSize);
+        return frameSizeInfo;
+    } else {
+        const BYTE* ip = (const BYTE*)src;
+        const BYTE* const ipstart = ip;
+        size_t remainingSize = srcSize;
+        size_t nbBlocks = 0;
+        ZSTD_frameHeader zfh;
+
+        /* Extract Frame Header */
+        {   size_t const ret = ZSTD_getFrameHeader(&zfh, src, srcSize);
+            if (ZSTD_isError(ret))
+                return ZSTD_errorFrameSizeInfo(ret);
+            if (ret > 0)
+                return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
+        }
+
+        ip += zfh.headerSize;
+        remainingSize -= zfh.headerSize;
+
+        /* Iterate over each block */
+        while (1) {
+            blockProperties_t blockProperties;
+            size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
+            if (ZSTD_isError(cBlockSize))
+                return ZSTD_errorFrameSizeInfo(cBlockSize);
+
+            if (ZSTD_blockHeaderSize + cBlockSize > remainingSize)
+                return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
+
+            ip += ZSTD_blockHeaderSize + cBlockSize;
+            remainingSize -= ZSTD_blockHeaderSize + cBlockSize;
+            nbBlocks++;
+
+            if (blockProperties.lastBlock) break;
+        }
+
+        /* Final frame content checksum */
+        if (zfh.checksumFlag) {
+            if (remainingSize < 4)
+                return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
+            ip += 4;
+        }
+
+        frameSizeInfo.compressedSize = (size_t)(ip - ipstart);
+        frameSizeInfo.decompressedBound = (zfh.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN)
+                                        ? zfh.frameContentSize
+                                        : nbBlocks * zfh.blockSizeMax;
+        return frameSizeInfo;
+    }
+}
+
+/* ZSTD_findFrameCompressedSize() :
+ *  compatible with legacy mode
+ *  `src` must point to the start of a ZSTD frame, ZSTD legacy frame, or skippable frame
+ *  `srcSize` must be at least as large as the frame contained
+ *  @return : the compressed size of the frame starting at `src` */
+size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize)
+{
+    ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
+    return frameSizeInfo.compressedSize;
+}
+
+/* ZSTD_decompressBound() :
+ *  compatible with legacy mode
+ *  `src` must point to the start of a ZSTD frame or a skippeable frame
+ *  `srcSize` must be at least as large as the frame contained
+ *  @return : the maximum decompressed size of the compressed source
+ */
+unsigned long long ZSTD_decompressBound(const void* src, size_t srcSize)
+{
+    unsigned long long bound = 0;
+    /* Iterate over each frame */
+    while (srcSize > 0) {
+        ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
+        size_t const compressedSize = frameSizeInfo.compressedSize;
+        unsigned long long const decompressedBound = frameSizeInfo.decompressedBound;
+        if (ZSTD_isError(compressedSize) || decompressedBound == ZSTD_CONTENTSIZE_ERROR)
+            return ZSTD_CONTENTSIZE_ERROR;
+        assert(srcSize >= compressedSize);
+        src = (const BYTE*)src + compressedSize;
+        srcSize -= compressedSize;
+        bound += decompressedBound;
+    }
+    return bound;
+}
+
+
+/*-*************************************************************
+ *   Frame decoding
+ ***************************************************************/
+
+/* ZSTD_insertBlock() :
+ *  insert `src` block into `dctx` history. Useful to track uncompressed blocks. */
+size_t ZSTD_insertBlock(ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize)
+{
+    DEBUGLOG(5, "ZSTD_insertBlock: %u bytes", (unsigned)blockSize);
+    ZSTD_checkContinuity(dctx, blockStart, blockSize);
+    dctx->previousDstEnd = (const char*)blockStart + blockSize;
+    return blockSize;
+}
+
+
+static size_t ZSTD_copyRawBlock(void* dst, size_t dstCapacity,
+                          const void* src, size_t srcSize)
+{
+    DEBUGLOG(5, "ZSTD_copyRawBlock");
+    RETURN_ERROR_IF(srcSize > dstCapacity, dstSize_tooSmall, "");
+    if (dst == NULL) {
+        if (srcSize == 0) return 0;
+        RETURN_ERROR(dstBuffer_null, "");
+    }
+    ZSTD_memcpy(dst, src, srcSize);
+    return srcSize;
+}
+
+static size_t ZSTD_setRleBlock(void* dst, size_t dstCapacity,
+                               BYTE b,
+                               size_t regenSize)
+{
+    RETURN_ERROR_IF(regenSize > dstCapacity, dstSize_tooSmall, "");
+    if (dst == NULL) {
+        if (regenSize == 0) return 0;
+        RETURN_ERROR(dstBuffer_null, "");
+    }
+    ZSTD_memset(dst, b, regenSize);
+    return regenSize;
+}
+
+static void ZSTD_DCtx_trace_end(ZSTD_DCtx const* dctx, U64 uncompressedSize, U64 compressedSize, unsigned streaming)
+{
+    (void)dctx;
+    (void)uncompressedSize;
+    (void)compressedSize;
+    (void)streaming;
+}
+
+
+/*! ZSTD_decompressFrame() :
+ * @dctx must be properly initialized
+ *  will update *srcPtr and *srcSizePtr,
+ *  to make *srcPtr progress by one frame. */
+static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
+                                   void* dst, size_t dstCapacity,
+                             const void** srcPtr, size_t *srcSizePtr)
+{
+    const BYTE* const istart = (const BYTE*)(*srcPtr);
+    const BYTE* ip = istart;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = dstCapacity != 0 ? ostart + dstCapacity : ostart;
+    BYTE* op = ostart;
+    size_t remainingSrcSize = *srcSizePtr;
+
+    DEBUGLOG(4, "ZSTD_decompressFrame (srcSize:%i)", (int)*srcSizePtr);
+
+    /* check */
+    RETURN_ERROR_IF(
+        remainingSrcSize < ZSTD_FRAMEHEADERSIZE_MIN(dctx->format)+ZSTD_blockHeaderSize,
+        srcSize_wrong, "");
+
+    /* Frame Header */
+    {   size_t const frameHeaderSize = ZSTD_frameHeaderSize_internal(
+                ip, ZSTD_FRAMEHEADERSIZE_PREFIX(dctx->format), dctx->format);
+        if (ZSTD_isError(frameHeaderSize)) return frameHeaderSize;
+        RETURN_ERROR_IF(remainingSrcSize < frameHeaderSize+ZSTD_blockHeaderSize,
+                        srcSize_wrong, "");
+        FORWARD_IF_ERROR( ZSTD_decodeFrameHeader(dctx, ip, frameHeaderSize) , "");
+        ip += frameHeaderSize; remainingSrcSize -= frameHeaderSize;
+    }
+
+    /* Loop on each block */
+    while (1) {
+        size_t decodedSize;
+        blockProperties_t blockProperties;
+        size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSrcSize, &blockProperties);
+        if (ZSTD_isError(cBlockSize)) return cBlockSize;
+
+        ip += ZSTD_blockHeaderSize;
+        remainingSrcSize -= ZSTD_blockHeaderSize;
+        RETURN_ERROR_IF(cBlockSize > remainingSrcSize, srcSize_wrong, "");
+
+        switch(blockProperties.blockType)
+        {
+        case bt_compressed:
+            decodedSize = ZSTD_decompressBlock_internal(dctx, op, (size_t)(oend-op), ip, cBlockSize, /* frame */ 1);
+            break;
+        case bt_raw :
+            decodedSize = ZSTD_copyRawBlock(op, (size_t)(oend-op), ip, cBlockSize);
+            break;
+        case bt_rle :
+            decodedSize = ZSTD_setRleBlock(op, (size_t)(oend-op), *ip, blockProperties.origSize);
+            break;
+        case bt_reserved :
+        default:
+            RETURN_ERROR(corruption_detected, "invalid block type");
+        }
+
+        if (ZSTD_isError(decodedSize)) return decodedSize;
+        if (dctx->validateChecksum)
+            xxh64_update(&dctx->xxhState, op, decodedSize);
+        if (decodedSize != 0)
+            op += decodedSize;
+        assert(ip != NULL);
+        ip += cBlockSize;
+        remainingSrcSize -= cBlockSize;
+        if (blockProperties.lastBlock) break;
+    }
+
+    if (dctx->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN) {
+        RETURN_ERROR_IF((U64)(op-ostart) != dctx->fParams.frameContentSize,
+                        corruption_detected, "");
+    }
+    if (dctx->fParams.checksumFlag) { /* Frame content checksum verification */
+        RETURN_ERROR_IF(remainingSrcSize<4, checksum_wrong, "");
+        if (!dctx->forceIgnoreChecksum) {
+            U32 const checkCalc = (U32)xxh64_digest(&dctx->xxhState);
+            U32 checkRead;
+            checkRead = MEM_readLE32(ip);
+            RETURN_ERROR_IF(checkRead != checkCalc, checksum_wrong, "");
+        }
+        ip += 4;
+        remainingSrcSize -= 4;
+    }
+    ZSTD_DCtx_trace_end(dctx, (U64)(op-ostart), (U64)(ip-istart), /* streaming */ 0);
+    /* Allow caller to get size read */
+    *srcPtr = ip;
+    *srcSizePtr = remainingSrcSize;
+    return (size_t)(op-ostart);
+}
+
+static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
+                                        void* dst, size_t dstCapacity,
+                                  const void* src, size_t srcSize,
+                                  const void* dict, size_t dictSize,
+                                  const ZSTD_DDict* ddict)
+{
+    void* const dststart = dst;
+    int moreThan1Frame = 0;
+
+    DEBUGLOG(5, "ZSTD_decompressMultiFrame");
+    assert(dict==NULL || ddict==NULL);  /* either dict or ddict set, not both */
+
+    if (ddict) {
+        dict = ZSTD_DDict_dictContent(ddict);
+        dictSize = ZSTD_DDict_dictSize(ddict);
+    }
+
+    while (srcSize >= ZSTD_startingInputLength(dctx->format)) {
+
+
+        {   U32 const magicNumber = MEM_readLE32(src);
+            DEBUGLOG(4, "reading magic number %08X (expecting %08X)",
+                        (unsigned)magicNumber, ZSTD_MAGICNUMBER);
+            if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+                size_t const skippableSize = readSkippableFrameSize(src, srcSize);
+                FORWARD_IF_ERROR(skippableSize, "readSkippableFrameSize failed");
+                assert(skippableSize <= srcSize);
+
+                src = (const BYTE *)src + skippableSize;
+                srcSize -= skippableSize;
+                continue;
+        }   }
+
+        if (ddict) {
+            /* we were called from ZSTD_decompress_usingDDict */
+            FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDDict(dctx, ddict), "");
+        } else {
+            /* this will initialize correctly with no dict if dict == NULL, so
+             * use this in all cases but ddict */
+            FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDict(dctx, dict, dictSize), "");
+        }
+        ZSTD_checkContinuity(dctx, dst, dstCapacity);
+
+        {   const size_t res = ZSTD_decompressFrame(dctx, dst, dstCapacity,
+                                                    &src, &srcSize);
+            RETURN_ERROR_IF(
+                (ZSTD_getErrorCode(res) == ZSTD_error_prefix_unknown)
+             && (moreThan1Frame==1),
+                srcSize_wrong,
+                "At least one frame successfully completed, "
+                "but following bytes are garbage: "
+                "it's more likely to be a srcSize error, "
+                "specifying more input bytes than size of frame(s). "
+                "Note: one could be unlucky, it might be a corruption error instead, "
+                "happening right at the place where we expect zstd magic bytes. "
+                "But this is _much_ less likely than a srcSize field error.");
+            if (ZSTD_isError(res)) return res;
+            assert(res <= dstCapacity);
+            if (res != 0)
+                dst = (BYTE*)dst + res;
+            dstCapacity -= res;
+        }
+        moreThan1Frame = 1;
+    }  /* while (srcSize >= ZSTD_frameHeaderSize_prefix) */
+
+    RETURN_ERROR_IF(srcSize, srcSize_wrong, "input not entirely consumed");
+
+    return (size_t)((BYTE*)dst - (BYTE*)dststart);
+}
+
+size_t ZSTD_decompress_usingDict(ZSTD_DCtx* dctx,
+                                 void* dst, size_t dstCapacity,
+                           const void* src, size_t srcSize,
+                           const void* dict, size_t dictSize)
+{
+    return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, dict, dictSize, NULL);
+}
+
+
+static ZSTD_DDict const* ZSTD_getDDict(ZSTD_DCtx* dctx)
+{
+    switch (dctx->dictUses) {
+    default:
+        assert(0 /* Impossible */);
+        ZSTD_FALLTHROUGH;
+    case ZSTD_dont_use:
+        ZSTD_clearDict(dctx);
+        return NULL;
+    case ZSTD_use_indefinitely:
+        return dctx->ddict;
+    case ZSTD_use_once:
+        dctx->dictUses = ZSTD_dont_use;
+        return dctx->ddict;
+    }
+}
+
+size_t ZSTD_decompressDCtx(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+    return ZSTD_decompress_usingDDict(dctx, dst, dstCapacity, src, srcSize, ZSTD_getDDict(dctx));
+}
+
+
+size_t ZSTD_decompress(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+#if defined(ZSTD_HEAPMODE) && (ZSTD_HEAPMODE>=1)
+    size_t regenSize;
+    ZSTD_DCtx* const dctx = ZSTD_createDCtx();
+    RETURN_ERROR_IF(dctx==NULL, memory_allocation, "NULL pointer!");
+    regenSize = ZSTD_decompressDCtx(dctx, dst, dstCapacity, src, srcSize);
+    ZSTD_freeDCtx(dctx);
+    return regenSize;
+#else   /* stack mode */
+    ZSTD_DCtx dctx;
+    ZSTD_initDCtx_internal(&dctx);
+    return ZSTD_decompressDCtx(&dctx, dst, dstCapacity, src, srcSize);
+#endif
+}
+
+
+/*-**************************************
+*   Advanced Streaming Decompression API
+*   Bufferless and synchronous
+****************************************/
+size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx) { return dctx->expected; }
+
+/*
+ * Similar to ZSTD_nextSrcSizeToDecompress(), but when when a block input can be streamed,
+ * we allow taking a partial block as the input. Currently only raw uncompressed blocks can
+ * be streamed.
+ *
+ * For blocks that can be streamed, this allows us to reduce the latency until we produce
+ * output, and avoid copying the input.
+ *
+ * @param inputSize - The total amount of input that the caller currently has.
+ */
+static size_t ZSTD_nextSrcSizeToDecompressWithInputSize(ZSTD_DCtx* dctx, size_t inputSize) {
+    if (!(dctx->stage == ZSTDds_decompressBlock || dctx->stage == ZSTDds_decompressLastBlock))
+        return dctx->expected;
+    if (dctx->bType != bt_raw)
+        return dctx->expected;
+    return MIN(MAX(inputSize, 1), dctx->expected);
+}
+
+ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx) {
+    switch(dctx->stage)
+    {
+    default:   /* should not happen */
+        assert(0);
+        ZSTD_FALLTHROUGH;
+    case ZSTDds_getFrameHeaderSize:
+        ZSTD_FALLTHROUGH;
+    case ZSTDds_decodeFrameHeader:
+        return ZSTDnit_frameHeader;
+    case ZSTDds_decodeBlockHeader:
+        return ZSTDnit_blockHeader;
+    case ZSTDds_decompressBlock:
+        return ZSTDnit_block;
+    case ZSTDds_decompressLastBlock:
+        return ZSTDnit_lastBlock;
+    case ZSTDds_checkChecksum:
+        return ZSTDnit_checksum;
+    case ZSTDds_decodeSkippableHeader:
+        ZSTD_FALLTHROUGH;
+    case ZSTDds_skipFrame:
+        return ZSTDnit_skippableFrame;
+    }
+}
+
+static int ZSTD_isSkipFrame(ZSTD_DCtx* dctx) { return dctx->stage == ZSTDds_skipFrame; }
+
+/* ZSTD_decompressContinue() :
+ *  srcSize : must be the exact nb of bytes expected (see ZSTD_nextSrcSizeToDecompress())
+ *  @return : nb of bytes generated into `dst` (necessarily <= `dstCapacity)
+ *            or an error code, which can be tested using ZSTD_isError() */
+size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+    DEBUGLOG(5, "ZSTD_decompressContinue (srcSize:%u)", (unsigned)srcSize);
+    /* Sanity check */
+    RETURN_ERROR_IF(srcSize != ZSTD_nextSrcSizeToDecompressWithInputSize(dctx, srcSize), srcSize_wrong, "not allowed");
+    ZSTD_checkContinuity(dctx, dst, dstCapacity);
+
+    dctx->processedCSize += srcSize;
+
+    switch (dctx->stage)
+    {
+    case ZSTDds_getFrameHeaderSize :
+        assert(src != NULL);
+        if (dctx->format == ZSTD_f_zstd1) {  /* allows header */
+            assert(srcSize >= ZSTD_FRAMEIDSIZE);  /* to read skippable magic number */
+            if ((MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {        /* skippable frame */
+                ZSTD_memcpy(dctx->headerBuffer, src, srcSize);
+                dctx->expected = ZSTD_SKIPPABLEHEADERSIZE - srcSize;  /* remaining to load to get full skippable frame header */
+                dctx->stage = ZSTDds_decodeSkippableHeader;
+                return 0;
+        }   }
+        dctx->headerSize = ZSTD_frameHeaderSize_internal(src, srcSize, dctx->format);
+        if (ZSTD_isError(dctx->headerSize)) return dctx->headerSize;
+        ZSTD_memcpy(dctx->headerBuffer, src, srcSize);
+        dctx->expected = dctx->headerSize - srcSize;
+        dctx->stage = ZSTDds_decodeFrameHeader;
+        return 0;
+
+    case ZSTDds_decodeFrameHeader:
+        assert(src != NULL);
+        ZSTD_memcpy(dctx->headerBuffer + (dctx->headerSize - srcSize), src, srcSize);
+        FORWARD_IF_ERROR(ZSTD_decodeFrameHeader(dctx, dctx->headerBuffer, dctx->headerSize), "");
+        dctx->expected = ZSTD_blockHeaderSize;
+        dctx->stage = ZSTDds_decodeBlockHeader;
+        return 0;
+
+    case ZSTDds_decodeBlockHeader:
+        {   blockProperties_t bp;
+            size_t const cBlockSize = ZSTD_getcBlockSize(src, ZSTD_blockHeaderSize, &bp);
+            if (ZSTD_isError(cBlockSize)) return cBlockSize;
+            RETURN_ERROR_IF(cBlockSize > dctx->fParams.blockSizeMax, corruption_detected, "Block Size Exceeds Maximum");
+            dctx->expected = cBlockSize;
+            dctx->bType = bp.blockType;
+            dctx->rleSize = bp.origSize;
+            if (cBlockSize) {
+                dctx->stage = bp.lastBlock ? ZSTDds_decompressLastBlock : ZSTDds_decompressBlock;
+                return 0;
+            }
+            /* empty block */
+            if (bp.lastBlock) {
+                if (dctx->fParams.checksumFlag) {
+                    dctx->expected = 4;
+                    dctx->stage = ZSTDds_checkChecksum;
+                } else {
+                    dctx->expected = 0; /* end of frame */
+                    dctx->stage = ZSTDds_getFrameHeaderSize;
+                }
+            } else {
+                dctx->expected = ZSTD_blockHeaderSize;  /* jump to next header */
+                dctx->stage = ZSTDds_decodeBlockHeader;
+            }
+            return 0;
+        }
+
+    case ZSTDds_decompressLastBlock:
+    case ZSTDds_decompressBlock:
+        DEBUGLOG(5, "ZSTD_decompressContinue: case ZSTDds_decompressBlock");
+        {   size_t rSize;
+            switch(dctx->bType)
+            {
+            case bt_compressed:
+                DEBUGLOG(5, "ZSTD_decompressContinue: case bt_compressed");
+                rSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, /* frame */ 1);
+                dctx->expected = 0;  /* Streaming not supported */
+                break;
+            case bt_raw :
+                assert(srcSize <= dctx->expected);
+                rSize = ZSTD_copyRawBlock(dst, dstCapacity, src, srcSize);
+                FORWARD_IF_ERROR(rSize, "ZSTD_copyRawBlock failed");
+                assert(rSize == srcSize);
+                dctx->expected -= rSize;
+                break;
+            case bt_rle :
+                rSize = ZSTD_setRleBlock(dst, dstCapacity, *(const BYTE*)src, dctx->rleSize);
+                dctx->expected = 0;  /* Streaming not supported */
+                break;
+            case bt_reserved :   /* should never happen */
+            default:
+                RETURN_ERROR(corruption_detected, "invalid block type");
+            }
+            FORWARD_IF_ERROR(rSize, "");
+            RETURN_ERROR_IF(rSize > dctx->fParams.blockSizeMax, corruption_detected, "Decompressed Block Size Exceeds Maximum");
+            DEBUGLOG(5, "ZSTD_decompressContinue: decoded size from block : %u", (unsigned)rSize);
+            dctx->decodedSize += rSize;
+            if (dctx->validateChecksum) xxh64_update(&dctx->xxhState, dst, rSize);
+            dctx->previousDstEnd = (char*)dst + rSize;
+
+            /* Stay on the same stage until we are finished streaming the block. */
+            if (dctx->expected > 0) {
+                return rSize;
+            }
+
+            if (dctx->stage == ZSTDds_decompressLastBlock) {   /* end of frame */
+                DEBUGLOG(4, "ZSTD_decompressContinue: decoded size from frame : %u", (unsigned)dctx->decodedSize);
+                RETURN_ERROR_IF(
+                    dctx->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
+                 && dctx->decodedSize != dctx->fParams.frameContentSize,
+                    corruption_detected, "");
+                if (dctx->fParams.checksumFlag) {  /* another round for frame checksum */
+                    dctx->expected = 4;
+                    dctx->stage = ZSTDds_checkChecksum;
+                } else {
+                    ZSTD_DCtx_trace_end(dctx, dctx->decodedSize, dctx->processedCSize, /* streaming */ 1);
+                    dctx->expected = 0;   /* ends here */
+                    dctx->stage = ZSTDds_getFrameHeaderSize;
+                }
+            } else {
+                dctx->stage = ZSTDds_decodeBlockHeader;
+                dctx->expected = ZSTD_blockHeaderSize;
+            }
+            return rSize;
+        }
+
+    case ZSTDds_checkChecksum:
+        assert(srcSize == 4);  /* guaranteed by dctx->expected */
+        {
+            if (dctx->validateChecksum) {
+                U32 const h32 = (U32)xxh64_digest(&dctx->xxhState);
+                U32 const check32 = MEM_readLE32(src);
+                DEBUGLOG(4, "ZSTD_decompressContinue: checksum : calculated %08X :: %08X read", (unsigned)h32, (unsigned)check32);
+                RETURN_ERROR_IF(check32 != h32, checksum_wrong, "");
+            }
+            ZSTD_DCtx_trace_end(dctx, dctx->decodedSize, dctx->processedCSize, /* streaming */ 1);
+            dctx->expected = 0;
+            dctx->stage = ZSTDds_getFrameHeaderSize;
+            return 0;
+        }
+
+    case ZSTDds_decodeSkippableHeader:
+        assert(src != NULL);
+        assert(srcSize <= ZSTD_SKIPPABLEHEADERSIZE);
+        ZSTD_memcpy(dctx->headerBuffer + (ZSTD_SKIPPABLEHEADERSIZE - srcSize), src, srcSize);   /* complete skippable header */
+        dctx->expected = MEM_readLE32(dctx->headerBuffer + ZSTD_FRAMEIDSIZE);   /* note : dctx->expected can grow seriously large, beyond local buffer size */
+        dctx->stage = ZSTDds_skipFrame;
+        return 0;
+
+    case ZSTDds_skipFrame:
+        dctx->expected = 0;
+        dctx->stage = ZSTDds_getFrameHeaderSize;
+        return 0;
+
+    default:
+        assert(0);   /* impossible */
+        RETURN_ERROR(GENERIC, "impossible to reach");   /* some compiler require default to do something */
+    }
+}
+
+
+static size_t ZSTD_refDictContent(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    dctx->dictEnd = dctx->previousDstEnd;
+    dctx->virtualStart = (const char*)dict - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart));
+    dctx->prefixStart = dict;
+    dctx->previousDstEnd = (const char*)dict + dictSize;
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    dctx->dictContentBeginForFuzzing = dctx->prefixStart;
+    dctx->dictContentEndForFuzzing = dctx->previousDstEnd;
+#endif
+    return 0;
+}
+
+/*! ZSTD_loadDEntropy() :
+ *  dict : must point at beginning of a valid zstd dictionary.
+ * @return : size of entropy tables read */
+size_t
+ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
+                  const void* const dict, size_t const dictSize)
+{
+    const BYTE* dictPtr = (const BYTE*)dict;
+    const BYTE* const dictEnd = dictPtr + dictSize;
+
+    RETURN_ERROR_IF(dictSize <= 8, dictionary_corrupted, "dict is too small");
+    assert(MEM_readLE32(dict) == ZSTD_MAGIC_DICTIONARY);   /* dict must be valid */
+    dictPtr += 8;   /* skip header = magic + dictID */
+
+    ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, OFTable) == offsetof(ZSTD_entropyDTables_t, LLTable) + sizeof(entropy->LLTable));
+    ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, MLTable) == offsetof(ZSTD_entropyDTables_t, OFTable) + sizeof(entropy->OFTable));
+    ZSTD_STATIC_ASSERT(sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable) >= HUF_DECOMPRESS_WORKSPACE_SIZE);
+    {   void* const workspace = &entropy->LLTable;   /* use fse tables as temporary workspace; implies fse tables are grouped together */
+        size_t const workspaceSize = sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable);
+#ifdef HUF_FORCE_DECOMPRESS_X1
+        /* in minimal huffman, we always use X1 variants */
+        size_t const hSize = HUF_readDTableX1_wksp(entropy->hufTable,
+                                                dictPtr, dictEnd - dictPtr,
+                                                workspace, workspaceSize);
+#else
+        size_t const hSize = HUF_readDTableX2_wksp(entropy->hufTable,
+                                                dictPtr, (size_t)(dictEnd - dictPtr),
+                                                workspace, workspaceSize);
+#endif
+        RETURN_ERROR_IF(HUF_isError(hSize), dictionary_corrupted, "");
+        dictPtr += hSize;
+    }
+
+    {   short offcodeNCount[MaxOff+1];
+        unsigned offcodeMaxValue = MaxOff, offcodeLog;
+        size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, (size_t)(dictEnd-dictPtr));
+        RETURN_ERROR_IF(FSE_isError(offcodeHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(offcodeMaxValue > MaxOff, dictionary_corrupted, "");
+        RETURN_ERROR_IF(offcodeLog > OffFSELog, dictionary_corrupted, "");
+        ZSTD_buildFSETable( entropy->OFTable,
+                            offcodeNCount, offcodeMaxValue,
+                            OF_base, OF_bits,
+                            offcodeLog,
+                            entropy->workspace, sizeof(entropy->workspace),
+                            /* bmi2 */0);
+        dictPtr += offcodeHeaderSize;
+    }
+
+    {   short matchlengthNCount[MaxML+1];
+        unsigned matchlengthMaxValue = MaxML, matchlengthLog;
+        size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, (size_t)(dictEnd-dictPtr));
+        RETURN_ERROR_IF(FSE_isError(matchlengthHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(matchlengthMaxValue > MaxML, dictionary_corrupted, "");
+        RETURN_ERROR_IF(matchlengthLog > MLFSELog, dictionary_corrupted, "");
+        ZSTD_buildFSETable( entropy->MLTable,
+                            matchlengthNCount, matchlengthMaxValue,
+                            ML_base, ML_bits,
+                            matchlengthLog,
+                            entropy->workspace, sizeof(entropy->workspace),
+                            /* bmi2 */ 0);
+        dictPtr += matchlengthHeaderSize;
+    }
+
+    {   short litlengthNCount[MaxLL+1];
+        unsigned litlengthMaxValue = MaxLL, litlengthLog;
+        size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, (size_t)(dictEnd-dictPtr));
+        RETURN_ERROR_IF(FSE_isError(litlengthHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(litlengthMaxValue > MaxLL, dictionary_corrupted, "");
+        RETURN_ERROR_IF(litlengthLog > LLFSELog, dictionary_corrupted, "");
+        ZSTD_buildFSETable( entropy->LLTable,
+                            litlengthNCount, litlengthMaxValue,
+                            LL_base, LL_bits,
+                            litlengthLog,
+                            entropy->workspace, sizeof(entropy->workspace),
+                            /* bmi2 */ 0);
+        dictPtr += litlengthHeaderSize;
+    }
+
+    RETURN_ERROR_IF(dictPtr+12 > dictEnd, dictionary_corrupted, "");
+    {   int i;
+        size_t const dictContentSize = (size_t)(dictEnd - (dictPtr+12));
+        for (i=0; i<3; i++) {
+            U32 const rep = MEM_readLE32(dictPtr); dictPtr += 4;
+            RETURN_ERROR_IF(rep==0 || rep > dictContentSize,
+                            dictionary_corrupted, "");
+            entropy->rep[i] = rep;
+    }   }
+
+    return (size_t)(dictPtr - (const BYTE*)dict);
+}
+
+static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    if (dictSize < 8) return ZSTD_refDictContent(dctx, dict, dictSize);
+    {   U32 const magic = MEM_readLE32(dict);
+        if (magic != ZSTD_MAGIC_DICTIONARY) {
+            return ZSTD_refDictContent(dctx, dict, dictSize);   /* pure content mode */
+    }   }
+    dctx->dictID = MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE);
+
+    /* load entropy tables */
+    {   size_t const eSize = ZSTD_loadDEntropy(&dctx->entropy, dict, dictSize);
+        RETURN_ERROR_IF(ZSTD_isError(eSize), dictionary_corrupted, "");
+        dict = (const char*)dict + eSize;
+        dictSize -= eSize;
+    }
+    dctx->litEntropy = dctx->fseEntropy = 1;
+
+    /* reference dictionary content */
+    return ZSTD_refDictContent(dctx, dict, dictSize);
+}
+
+size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx)
+{
+    assert(dctx != NULL);
+    dctx->expected = ZSTD_startingInputLength(dctx->format);  /* dctx->format must be properly set */
+    dctx->stage = ZSTDds_getFrameHeaderSize;
+    dctx->processedCSize = 0;
+    dctx->decodedSize = 0;
+    dctx->previousDstEnd = NULL;
+    dctx->prefixStart = NULL;
+    dctx->virtualStart = NULL;
+    dctx->dictEnd = NULL;
+    dctx->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001);  /* cover both little and big endian */
+    dctx->litEntropy = dctx->fseEntropy = 0;
+    dctx->dictID = 0;
+    dctx->bType = bt_reserved;
+    ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.rep) == sizeof(repStartValue));
+    ZSTD_memcpy(dctx->entropy.rep, repStartValue, sizeof(repStartValue));  /* initial repcodes */
+    dctx->LLTptr = dctx->entropy.LLTable;
+    dctx->MLTptr = dctx->entropy.MLTable;
+    dctx->OFTptr = dctx->entropy.OFTable;
+    dctx->HUFptr = dctx->entropy.hufTable;
+    return 0;
+}
+
+size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    FORWARD_IF_ERROR( ZSTD_decompressBegin(dctx) , "");
+    if (dict && dictSize)
+        RETURN_ERROR_IF(
+            ZSTD_isError(ZSTD_decompress_insertDictionary(dctx, dict, dictSize)),
+            dictionary_corrupted, "");
+    return 0;
+}
+
+
+/* ======   ZSTD_DDict   ====== */
+
+size_t ZSTD_decompressBegin_usingDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
+{
+    DEBUGLOG(4, "ZSTD_decompressBegin_usingDDict");
+    assert(dctx != NULL);
+    if (ddict) {
+        const char* const dictStart = (const char*)ZSTD_DDict_dictContent(ddict);
+        size_t const dictSize = ZSTD_DDict_dictSize(ddict);
+        const void* const dictEnd = dictStart + dictSize;
+        dctx->ddictIsCold = (dctx->dictEnd != dictEnd);
+        DEBUGLOG(4, "DDict is %s",
+                    dctx->ddictIsCold ? "~cold~" : "hot!");
+    }
+    FORWARD_IF_ERROR( ZSTD_decompressBegin(dctx) , "");
+    if (ddict) {   /* NULL ddict is equivalent to no dictionary */
+        ZSTD_copyDDictParameters(dctx, ddict);
+    }
+    return 0;
+}
+
+/*! ZSTD_getDictID_fromDict() :
+ *  Provides the dictID stored within dictionary.
+ *  if @return == 0, the dictionary is not conformant with Zstandard specification.
+ *  It can still be loaded, but as a content-only dictionary. */
+unsigned ZSTD_getDictID_fromDict(const void* dict, size_t dictSize)
+{
+    if (dictSize < 8) return 0;
+    if (MEM_readLE32(dict) != ZSTD_MAGIC_DICTIONARY) return 0;
+    return MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE);
+}
+
+/*! ZSTD_getDictID_fromFrame() :
+ *  Provides the dictID required to decompress frame stored within `src`.
+ *  If @return == 0, the dictID could not be decoded.
+ *  This could for one of the following reasons :
+ *  - The frame does not require a dictionary (most common case).
+ *  - The frame was built with dictID intentionally removed.
+ *    Needed dictionary is a hidden information.
+ *    Note : this use case also happens when using a non-conformant dictionary.
+ *  - `srcSize` is too small, and as a result, frame header could not be decoded.
+ *    Note : possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`.
+ *  - This is not a Zstandard frame.
+ *  When identifying the exact failure cause, it's possible to use
+ *  ZSTD_getFrameHeader(), which will provide a more precise error code. */
+unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize)
+{
+    ZSTD_frameHeader zfp = { 0, 0, 0, ZSTD_frame, 0, 0, 0 };
+    size_t const hError = ZSTD_getFrameHeader(&zfp, src, srcSize);
+    if (ZSTD_isError(hError)) return 0;
+    return zfp.dictID;
+}
+
+
+/*! ZSTD_decompress_usingDDict() :
+*   Decompression using a pre-digested Dictionary
+*   Use dictionary without significant overhead. */
+size_t ZSTD_decompress_usingDDict(ZSTD_DCtx* dctx,
+                                  void* dst, size_t dstCapacity,
+                            const void* src, size_t srcSize,
+                            const ZSTD_DDict* ddict)
+{
+    /* pass content and size in case legacy frames are encountered */
+    return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize,
+                                     NULL, 0,
+                                     ddict);
+}
+
+
+/*=====================================
+*   Streaming decompression
+*====================================*/
+
+ZSTD_DStream* ZSTD_createDStream(void)
+{
+    DEBUGLOG(3, "ZSTD_createDStream");
+    return ZSTD_createDStream_advanced(ZSTD_defaultCMem);
+}
+
+ZSTD_DStream* ZSTD_initStaticDStream(void *workspace, size_t workspaceSize)
+{
+    return ZSTD_initStaticDCtx(workspace, workspaceSize);
+}
+
+ZSTD_DStream* ZSTD_createDStream_advanced(ZSTD_customMem customMem)
+{
+    return ZSTD_createDCtx_advanced(customMem);
+}
+
+size_t ZSTD_freeDStream(ZSTD_DStream* zds)
+{
+    return ZSTD_freeDCtx(zds);
+}
+
+
+/* ***  Initialization  *** */
+
+size_t ZSTD_DStreamInSize(void)  { return ZSTD_BLOCKSIZE_MAX + ZSTD_blockHeaderSize; }
+size_t ZSTD_DStreamOutSize(void) { return ZSTD_BLOCKSIZE_MAX; }
+
+size_t ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx* dctx,
+                                   const void* dict, size_t dictSize,
+                                         ZSTD_dictLoadMethod_e dictLoadMethod,
+                                         ZSTD_dictContentType_e dictContentType)
+{
+    RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+    ZSTD_clearDict(dctx);
+    if (dict && dictSize != 0) {
+        dctx->ddictLocal = ZSTD_createDDict_advanced(dict, dictSize, dictLoadMethod, dictContentType, dctx->customMem);
+        RETURN_ERROR_IF(dctx->ddictLocal == NULL, memory_allocation, "NULL pointer!");
+        dctx->ddict = dctx->ddictLocal;
+        dctx->dictUses = ZSTD_use_indefinitely;
+    }
+    return 0;
+}
+
+size_t ZSTD_DCtx_loadDictionary_byReference(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    return ZSTD_DCtx_loadDictionary_advanced(dctx, dict, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto);
+}
+
+size_t ZSTD_DCtx_loadDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    return ZSTD_DCtx_loadDictionary_advanced(dctx, dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto);
+}
+
+size_t ZSTD_DCtx_refPrefix_advanced(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType)
+{
+    FORWARD_IF_ERROR(ZSTD_DCtx_loadDictionary_advanced(dctx, prefix, prefixSize, ZSTD_dlm_byRef, dictContentType), "");
+    dctx->dictUses = ZSTD_use_once;
+    return 0;
+}
+
+size_t ZSTD_DCtx_refPrefix(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize)
+{
+    return ZSTD_DCtx_refPrefix_advanced(dctx, prefix, prefixSize, ZSTD_dct_rawContent);
+}
+
+
+/* ZSTD_initDStream_usingDict() :
+ * return : expected size, aka ZSTD_startingInputLength().
+ * this function cannot fail */
+size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize)
+{
+    DEBUGLOG(4, "ZSTD_initDStream_usingDict");
+    FORWARD_IF_ERROR( ZSTD_DCtx_reset(zds, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_DCtx_loadDictionary(zds, dict, dictSize) , "");
+    return ZSTD_startingInputLength(zds->format);
+}
+
+/* note : this variant can't fail */
+size_t ZSTD_initDStream(ZSTD_DStream* zds)
+{
+    DEBUGLOG(4, "ZSTD_initDStream");
+    return ZSTD_initDStream_usingDDict(zds, NULL);
+}
+
+/* ZSTD_initDStream_usingDDict() :
+ * ddict will just be referenced, and must outlive decompression session
+ * this function cannot fail */
+size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* dctx, const ZSTD_DDict* ddict)
+{
+    FORWARD_IF_ERROR( ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_DCtx_refDDict(dctx, ddict) , "");
+    return ZSTD_startingInputLength(dctx->format);
+}
+
+/* ZSTD_resetDStream() :
+ * return : expected size, aka ZSTD_startingInputLength().
+ * this function cannot fail */
+size_t ZSTD_resetDStream(ZSTD_DStream* dctx)
+{
+    FORWARD_IF_ERROR(ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only), "");
+    return ZSTD_startingInputLength(dctx->format);
+}
+
+
+size_t ZSTD_DCtx_refDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
+{
+    RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+    ZSTD_clearDict(dctx);
+    if (ddict) {
+        dctx->ddict = ddict;
+        dctx->dictUses = ZSTD_use_indefinitely;
+        if (dctx->refMultipleDDicts == ZSTD_rmd_refMultipleDDicts) {
+            if (dctx->ddictSet == NULL) {
+                dctx->ddictSet = ZSTD_createDDictHashSet(dctx->customMem);
+                if (!dctx->ddictSet) {
+                    RETURN_ERROR(memory_allocation, "Failed to allocate memory for hash set!");
+                }
+            }
+            assert(!dctx->staticSize);  /* Impossible: ddictSet cannot have been allocated if static dctx */
+            FORWARD_IF_ERROR(ZSTD_DDictHashSet_addDDict(dctx->ddictSet, ddict, dctx->customMem), "");
+        }
+    }
+    return 0;
+}
+
+/* ZSTD_DCtx_setMaxWindowSize() :
+ * note : no direct equivalence in ZSTD_DCtx_setParameter,
+ * since this version sets windowSize, and the other sets windowLog */
+size_t ZSTD_DCtx_setMaxWindowSize(ZSTD_DCtx* dctx, size_t maxWindowSize)
+{
+    ZSTD_bounds const bounds = ZSTD_dParam_getBounds(ZSTD_d_windowLogMax);
+    size_t const min = (size_t)1 << bounds.lowerBound;
+    size_t const max = (size_t)1 << bounds.upperBound;
+    RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+    RETURN_ERROR_IF(maxWindowSize < min, parameter_outOfBound, "");
+    RETURN_ERROR_IF(maxWindowSize > max, parameter_outOfBound, "");
+    dctx->maxWindowSize = maxWindowSize;
+    return 0;
+}
+
+size_t ZSTD_DCtx_setFormat(ZSTD_DCtx* dctx, ZSTD_format_e format)
+{
+    return ZSTD_DCtx_setParameter(dctx, ZSTD_d_format, (int)format);
+}
+
+ZSTD_bounds ZSTD_dParam_getBounds(ZSTD_dParameter dParam)
+{
+    ZSTD_bounds bounds = { 0, 0, 0 };
+    switch(dParam) {
+        case ZSTD_d_windowLogMax:
+            bounds.lowerBound = ZSTD_WINDOWLOG_ABSOLUTEMIN;
+            bounds.upperBound = ZSTD_WINDOWLOG_MAX;
+            return bounds;
+        case ZSTD_d_format:
+            bounds.lowerBound = (int)ZSTD_f_zstd1;
+            bounds.upperBound = (int)ZSTD_f_zstd1_magicless;
+            ZSTD_STATIC_ASSERT(ZSTD_f_zstd1 < ZSTD_f_zstd1_magicless);
+            return bounds;
+        case ZSTD_d_stableOutBuffer:
+            bounds.lowerBound = (int)ZSTD_bm_buffered;
+            bounds.upperBound = (int)ZSTD_bm_stable;
+            return bounds;
+        case ZSTD_d_forceIgnoreChecksum:
+            bounds.lowerBound = (int)ZSTD_d_validateChecksum;
+            bounds.upperBound = (int)ZSTD_d_ignoreChecksum;
+            return bounds;
+        case ZSTD_d_refMultipleDDicts:
+            bounds.lowerBound = (int)ZSTD_rmd_refSingleDDict;
+            bounds.upperBound = (int)ZSTD_rmd_refMultipleDDicts;
+            return bounds;
+        default:;
+    }
+    bounds.error = ERROR(parameter_unsupported);
+    return bounds;
+}
+
+/* ZSTD_dParam_withinBounds:
+ * @return 1 if value is within dParam bounds,
+ * 0 otherwise */
+static int ZSTD_dParam_withinBounds(ZSTD_dParameter dParam, int value)
+{
+    ZSTD_bounds const bounds = ZSTD_dParam_getBounds(dParam);
+    if (ZSTD_isError(bounds.error)) return 0;
+    if (value < bounds.lowerBound) return 0;
+    if (value > bounds.upperBound) return 0;
+    return 1;
+}
+
+#define CHECK_DBOUNDS(p,v) {                \
+    RETURN_ERROR_IF(!ZSTD_dParam_withinBounds(p, v), parameter_outOfBound, ""); \
+}
+
+size_t ZSTD_DCtx_getParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int* value)
+{
+    switch (param) {
+        case ZSTD_d_windowLogMax:
+            *value = (int)ZSTD_highbit32((U32)dctx->maxWindowSize);
+            return 0;
+        case ZSTD_d_format:
+            *value = (int)dctx->format;
+            return 0;
+        case ZSTD_d_stableOutBuffer:
+            *value = (int)dctx->outBufferMode;
+            return 0;
+        case ZSTD_d_forceIgnoreChecksum:
+            *value = (int)dctx->forceIgnoreChecksum;
+            return 0;
+        case ZSTD_d_refMultipleDDicts:
+            *value = (int)dctx->refMultipleDDicts;
+            return 0;
+        default:;
+    }
+    RETURN_ERROR(parameter_unsupported, "");
+}
+
+size_t ZSTD_DCtx_setParameter(ZSTD_DCtx* dctx, ZSTD_dParameter dParam, int value)
+{
+    RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+    switch(dParam) {
+        case ZSTD_d_windowLogMax:
+            if (value == 0) value = ZSTD_WINDOWLOG_LIMIT_DEFAULT;
+            CHECK_DBOUNDS(ZSTD_d_windowLogMax, value);
+            dctx->maxWindowSize = ((size_t)1) << value;
+            return 0;
+        case ZSTD_d_format:
+            CHECK_DBOUNDS(ZSTD_d_format, value);
+            dctx->format = (ZSTD_format_e)value;
+            return 0;
+        case ZSTD_d_stableOutBuffer:
+            CHECK_DBOUNDS(ZSTD_d_stableOutBuffer, value);
+            dctx->outBufferMode = (ZSTD_bufferMode_e)value;
+            return 0;
+        case ZSTD_d_forceIgnoreChecksum:
+            CHECK_DBOUNDS(ZSTD_d_forceIgnoreChecksum, value);
+            dctx->forceIgnoreChecksum = (ZSTD_forceIgnoreChecksum_e)value;
+            return 0;
+        case ZSTD_d_refMultipleDDicts:
+            CHECK_DBOUNDS(ZSTD_d_refMultipleDDicts, value);
+            if (dctx->staticSize != 0) {
+                RETURN_ERROR(parameter_unsupported, "Static dctx does not support multiple DDicts!");
+            }
+            dctx->refMultipleDDicts = (ZSTD_refMultipleDDicts_e)value;
+            return 0;
+        default:;
+    }
+    RETURN_ERROR(parameter_unsupported, "");
+}
+
+size_t ZSTD_DCtx_reset(ZSTD_DCtx* dctx, ZSTD_ResetDirective reset)
+{
+    if ( (reset == ZSTD_reset_session_only)
+      || (reset == ZSTD_reset_session_and_parameters) ) {
+        dctx->streamStage = zdss_init;
+        dctx->noForwardProgress = 0;
+    }
+    if ( (reset == ZSTD_reset_parameters)
+      || (reset == ZSTD_reset_session_and_parameters) ) {
+        RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+        ZSTD_clearDict(dctx);
+        ZSTD_DCtx_resetParameters(dctx);
+    }
+    return 0;
+}
+
+
+size_t ZSTD_sizeof_DStream(const ZSTD_DStream* dctx)
+{
+    return ZSTD_sizeof_DCtx(dctx);
+}
+
+size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize)
+{
+    size_t const blockSize = (size_t) MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
+    unsigned long long const neededRBSize = windowSize + blockSize + (WILDCOPY_OVERLENGTH * 2);
+    unsigned long long const neededSize = MIN(frameContentSize, neededRBSize);
+    size_t const minRBSize = (size_t) neededSize;
+    RETURN_ERROR_IF((unsigned long long)minRBSize != neededSize,
+                    frameParameter_windowTooLarge, "");
+    return minRBSize;
+}
+
+size_t ZSTD_estimateDStreamSize(size_t windowSize)
+{
+    size_t const blockSize = MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
+    size_t const inBuffSize = blockSize;  /* no block can be larger */
+    size_t const outBuffSize = ZSTD_decodingBufferSize_min(windowSize, ZSTD_CONTENTSIZE_UNKNOWN);
+    return ZSTD_estimateDCtxSize() + inBuffSize + outBuffSize;
+}
+
+size_t ZSTD_estimateDStreamSize_fromFrame(const void* src, size_t srcSize)
+{
+    U32 const windowSizeMax = 1U << ZSTD_WINDOWLOG_MAX;   /* note : should be user-selectable, but requires an additional parameter (or a dctx) */
+    ZSTD_frameHeader zfh;
+    size_t const err = ZSTD_getFrameHeader(&zfh, src, srcSize);
+    if (ZSTD_isError(err)) return err;
+    RETURN_ERROR_IF(err>0, srcSize_wrong, "");
+    RETURN_ERROR_IF(zfh.windowSize > windowSizeMax,
+                    frameParameter_windowTooLarge, "");
+    return ZSTD_estimateDStreamSize((size_t)zfh.windowSize);
+}
+
+
+/* *****   Decompression   ***** */
+
+static int ZSTD_DCtx_isOverflow(ZSTD_DStream* zds, size_t const neededInBuffSize, size_t const neededOutBuffSize)
+{
+    return (zds->inBuffSize + zds->outBuffSize) >= (neededInBuffSize + neededOutBuffSize) * ZSTD_WORKSPACETOOLARGE_FACTOR;
+}
+
+static void ZSTD_DCtx_updateOversizedDuration(ZSTD_DStream* zds, size_t const neededInBuffSize, size_t const neededOutBuffSize)
+{
+    if (ZSTD_DCtx_isOverflow(zds, neededInBuffSize, neededOutBuffSize))
+        zds->oversizedDuration++;
+    else
+        zds->oversizedDuration = 0;
+}
+
+static int ZSTD_DCtx_isOversizedTooLong(ZSTD_DStream* zds)
+{
+    return zds->oversizedDuration >= ZSTD_WORKSPACETOOLARGE_MAXDURATION;
+}
+
+/* Checks that the output buffer hasn't changed if ZSTD_obm_stable is used. */
+static size_t ZSTD_checkOutBuffer(ZSTD_DStream const* zds, ZSTD_outBuffer const* output)
+{
+    ZSTD_outBuffer const expect = zds->expectedOutBuffer;
+    /* No requirement when ZSTD_obm_stable is not enabled. */
+    if (zds->outBufferMode != ZSTD_bm_stable)
+        return 0;
+    /* Any buffer is allowed in zdss_init, this must be the same for every other call until
+     * the context is reset.
+     */
+    if (zds->streamStage == zdss_init)
+        return 0;
+    /* The buffer must match our expectation exactly. */
+    if (expect.dst == output->dst && expect.pos == output->pos && expect.size == output->size)
+        return 0;
+    RETURN_ERROR(dstBuffer_wrong, "ZSTD_d_stableOutBuffer enabled but output differs!");
+}
+
+/* Calls ZSTD_decompressContinue() with the right parameters for ZSTD_decompressStream()
+ * and updates the stage and the output buffer state. This call is extracted so it can be
+ * used both when reading directly from the ZSTD_inBuffer, and in buffered input mode.
+ * NOTE: You must break after calling this function since the streamStage is modified.
+ */
+static size_t ZSTD_decompressContinueStream(
+            ZSTD_DStream* zds, char** op, char* oend,
+            void const* src, size_t srcSize) {
+    int const isSkipFrame = ZSTD_isSkipFrame(zds);
+    if (zds->outBufferMode == ZSTD_bm_buffered) {
+        size_t const dstSize = isSkipFrame ? 0 : zds->outBuffSize - zds->outStart;
+        size_t const decodedSize = ZSTD_decompressContinue(zds,
+                zds->outBuff + zds->outStart, dstSize, src, srcSize);
+        FORWARD_IF_ERROR(decodedSize, "");
+        if (!decodedSize && !isSkipFrame) {
+            zds->streamStage = zdss_read;
+        } else {
+            zds->outEnd = zds->outStart + decodedSize;
+            zds->streamStage = zdss_flush;
+        }
+    } else {
+        /* Write directly into the output buffer */
+        size_t const dstSize = isSkipFrame ? 0 : (size_t)(oend - *op);
+        size_t const decodedSize = ZSTD_decompressContinue(zds, *op, dstSize, src, srcSize);
+        FORWARD_IF_ERROR(decodedSize, "");
+        *op += decodedSize;
+        /* Flushing is not needed. */
+        zds->streamStage = zdss_read;
+        assert(*op <= oend);
+        assert(zds->outBufferMode == ZSTD_bm_stable);
+    }
+    return 0;
+}
+
+size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inBuffer* input)
+{
+    const char* const src = (const char*)input->src;
+    const char* const istart = input->pos != 0 ? src + input->pos : src;
+    const char* const iend = input->size != 0 ? src + input->size : src;
+    const char* ip = istart;
+    char* const dst = (char*)output->dst;
+    char* const ostart = output->pos != 0 ? dst + output->pos : dst;
+    char* const oend = output->size != 0 ? dst + output->size : dst;
+    char* op = ostart;
+    U32 someMoreWork = 1;
+
+    DEBUGLOG(5, "ZSTD_decompressStream");
+    RETURN_ERROR_IF(
+        input->pos > input->size,
+        srcSize_wrong,
+        "forbidden. in: pos: %u   vs size: %u",
+        (U32)input->pos, (U32)input->size);
+    RETURN_ERROR_IF(
+        output->pos > output->size,
+        dstSize_tooSmall,
+        "forbidden. out: pos: %u   vs size: %u",
+        (U32)output->pos, (U32)output->size);
+    DEBUGLOG(5, "input size : %u", (U32)(input->size - input->pos));
+    FORWARD_IF_ERROR(ZSTD_checkOutBuffer(zds, output), "");
+
+    while (someMoreWork) {
+        switch(zds->streamStage)
+        {
+        case zdss_init :
+            DEBUGLOG(5, "stage zdss_init => transparent reset ");
+            zds->streamStage = zdss_loadHeader;
+            zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
+            zds->legacyVersion = 0;
+            zds->hostageByte = 0;
+            zds->expectedOutBuffer = *output;
+            ZSTD_FALLTHROUGH;
+
+        case zdss_loadHeader :
+            DEBUGLOG(5, "stage zdss_loadHeader (srcSize : %u)", (U32)(iend - ip));
+            {   size_t const hSize = ZSTD_getFrameHeader_advanced(&zds->fParams, zds->headerBuffer, zds->lhSize, zds->format);
+                if (zds->refMultipleDDicts && zds->ddictSet) {
+                    ZSTD_DCtx_selectFrameDDict(zds);
+                }
+                DEBUGLOG(5, "header size : %u", (U32)hSize);
+                if (ZSTD_isError(hSize)) {
+                    return hSize;   /* error */
+                }
+                if (hSize != 0) {   /* need more input */
+                    size_t const toLoad = hSize - zds->lhSize;   /* if hSize!=0, hSize > zds->lhSize */
+                    size_t const remainingInput = (size_t)(iend-ip);
+                    assert(iend >= ip);
+                    if (toLoad > remainingInput) {   /* not enough input to load full header */
+                        if (remainingInput > 0) {
+                            ZSTD_memcpy(zds->headerBuffer + zds->lhSize, ip, remainingInput);
+                            zds->lhSize += remainingInput;
+                        }
+                        input->pos = input->size;
+                        return (MAX((size_t)ZSTD_FRAMEHEADERSIZE_MIN(zds->format), hSize) - zds->lhSize) + ZSTD_blockHeaderSize;   /* remaining header bytes + next block header */
+                    }
+                    assert(ip != NULL);
+                    ZSTD_memcpy(zds->headerBuffer + zds->lhSize, ip, toLoad); zds->lhSize = hSize; ip += toLoad;
+                    break;
+            }   }
+
+            /* check for single-pass mode opportunity */
+            if (zds->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
+                && zds->fParams.frameType != ZSTD_skippableFrame
+                && (U64)(size_t)(oend-op) >= zds->fParams.frameContentSize) {
+                size_t const cSize = ZSTD_findFrameCompressedSize(istart, (size_t)(iend-istart));
+                if (cSize <= (size_t)(iend-istart)) {
+                    /* shortcut : using single-pass mode */
+                    size_t const decompressedSize = ZSTD_decompress_usingDDict(zds, op, (size_t)(oend-op), istart, cSize, ZSTD_getDDict(zds));
+                    if (ZSTD_isError(decompressedSize)) return decompressedSize;
+                    DEBUGLOG(4, "shortcut to single-pass ZSTD_decompress_usingDDict()")
+                    ip = istart + cSize;
+                    op += decompressedSize;
+                    zds->expected = 0;
+                    zds->streamStage = zdss_init;
+                    someMoreWork = 0;
+                    break;
+            }   }
+
+            /* Check output buffer is large enough for ZSTD_odm_stable. */
+            if (zds->outBufferMode == ZSTD_bm_stable
+                && zds->fParams.frameType != ZSTD_skippableFrame
+                && zds->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
+                && (U64)(size_t)(oend-op) < zds->fParams.frameContentSize) {
+                RETURN_ERROR(dstSize_tooSmall, "ZSTD_obm_stable passed but ZSTD_outBuffer is too small");
+            }
+
+            /* Consume header (see ZSTDds_decodeFrameHeader) */
+            DEBUGLOG(4, "Consume header");
+            FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDDict(zds, ZSTD_getDDict(zds)), "");
+
+            if ((MEM_readLE32(zds->headerBuffer) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {  /* skippable frame */
+                zds->expected = MEM_readLE32(zds->headerBuffer + ZSTD_FRAMEIDSIZE);
+                zds->stage = ZSTDds_skipFrame;
+            } else {
+                FORWARD_IF_ERROR(ZSTD_decodeFrameHeader(zds, zds->headerBuffer, zds->lhSize), "");
+                zds->expected = ZSTD_blockHeaderSize;
+                zds->stage = ZSTDds_decodeBlockHeader;
+            }
+
+            /* control buffer memory usage */
+            DEBUGLOG(4, "Control max memory usage (%u KB <= max %u KB)",
+                        (U32)(zds->fParams.windowSize >>10),
+                        (U32)(zds->maxWindowSize >> 10) );
+            zds->fParams.windowSize = MAX(zds->fParams.windowSize, 1U << ZSTD_WINDOWLOG_ABSOLUTEMIN);
+            RETURN_ERROR_IF(zds->fParams.windowSize > zds->maxWindowSize,
+                            frameParameter_windowTooLarge, "");
+
+            /* Adapt buffer sizes to frame header instructions */
+            {   size_t const neededInBuffSize = MAX(zds->fParams.blockSizeMax, 4 /* frame checksum */);
+                size_t const neededOutBuffSize = zds->outBufferMode == ZSTD_bm_buffered
+                        ? ZSTD_decodingBufferSize_min(zds->fParams.windowSize, zds->fParams.frameContentSize)
+                        : 0;
+
+                ZSTD_DCtx_updateOversizedDuration(zds, neededInBuffSize, neededOutBuffSize);
+
+                {   int const tooSmall = (zds->inBuffSize < neededInBuffSize) || (zds->outBuffSize < neededOutBuffSize);
+                    int const tooLarge = ZSTD_DCtx_isOversizedTooLong(zds);
+
+                    if (tooSmall || tooLarge) {
+                        size_t const bufferSize = neededInBuffSize + neededOutBuffSize;
+                        DEBUGLOG(4, "inBuff  : from %u to %u",
+                                    (U32)zds->inBuffSize, (U32)neededInBuffSize);
+                        DEBUGLOG(4, "outBuff : from %u to %u",
+                                    (U32)zds->outBuffSize, (U32)neededOutBuffSize);
+                        if (zds->staticSize) {  /* static DCtx */
+                            DEBUGLOG(4, "staticSize : %u", (U32)zds->staticSize);
+                            assert(zds->staticSize >= sizeof(ZSTD_DCtx));  /* controlled at init */
+                            RETURN_ERROR_IF(
+                                bufferSize > zds->staticSize - sizeof(ZSTD_DCtx),
+                                memory_allocation, "");
+                        } else {
+                            ZSTD_customFree(zds->inBuff, zds->customMem);
+                            zds->inBuffSize = 0;
+                            zds->outBuffSize = 0;
+                            zds->inBuff = (char*)ZSTD_customMalloc(bufferSize, zds->customMem);
+                            RETURN_ERROR_IF(zds->inBuff == NULL, memory_allocation, "");
+                        }
+                        zds->inBuffSize = neededInBuffSize;
+                        zds->outBuff = zds->inBuff + zds->inBuffSize;
+                        zds->outBuffSize = neededOutBuffSize;
+            }   }   }
+            zds->streamStage = zdss_read;
+            ZSTD_FALLTHROUGH;
+
+        case zdss_read:
+            DEBUGLOG(5, "stage zdss_read");
+            {   size_t const neededInSize = ZSTD_nextSrcSizeToDecompressWithInputSize(zds, (size_t)(iend - ip));
+                DEBUGLOG(5, "neededInSize = %u", (U32)neededInSize);
+                if (neededInSize==0) {  /* end of frame */
+                    zds->streamStage = zdss_init;
+                    someMoreWork = 0;
+                    break;
+                }
+                if ((size_t)(iend-ip) >= neededInSize) {  /* decode directly from src */
+                    FORWARD_IF_ERROR(ZSTD_decompressContinueStream(zds, &op, oend, ip, neededInSize), "");
+                    ip += neededInSize;
+                    /* Function modifies the stage so we must break */
+                    break;
+            }   }
+            if (ip==iend) { someMoreWork = 0; break; }   /* no more input */
+            zds->streamStage = zdss_load;
+            ZSTD_FALLTHROUGH;
+
+        case zdss_load:
+            {   size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds);
+                size_t const toLoad = neededInSize - zds->inPos;
+                int const isSkipFrame = ZSTD_isSkipFrame(zds);
+                size_t loadedSize;
+                /* At this point we shouldn't be decompressing a block that we can stream. */
+                assert(neededInSize == ZSTD_nextSrcSizeToDecompressWithInputSize(zds, iend - ip));
+                if (isSkipFrame) {
+                    loadedSize = MIN(toLoad, (size_t)(iend-ip));
+                } else {
+                    RETURN_ERROR_IF(toLoad > zds->inBuffSize - zds->inPos,
+                                    corruption_detected,
+                                    "should never happen");
+                    loadedSize = ZSTD_limitCopy(zds->inBuff + zds->inPos, toLoad, ip, (size_t)(iend-ip));
+                }
+                ip += loadedSize;
+                zds->inPos += loadedSize;
+                if (loadedSize < toLoad) { someMoreWork = 0; break; }   /* not enough input, wait for more */
+
+                /* decode loaded input */
+                zds->inPos = 0;   /* input is consumed */
+                FORWARD_IF_ERROR(ZSTD_decompressContinueStream(zds, &op, oend, zds->inBuff, neededInSize), "");
+                /* Function modifies the stage so we must break */
+                break;
+            }
+        case zdss_flush:
+            {   size_t const toFlushSize = zds->outEnd - zds->outStart;
+                size_t const flushedSize = ZSTD_limitCopy(op, (size_t)(oend-op), zds->outBuff + zds->outStart, toFlushSize);
+                op += flushedSize;
+                zds->outStart += flushedSize;
+                if (flushedSize == toFlushSize) {  /* flush completed */
+                    zds->streamStage = zdss_read;
+                    if ( (zds->outBuffSize < zds->fParams.frameContentSize)
+                      && (zds->outStart + zds->fParams.blockSizeMax > zds->outBuffSize) ) {
+                        DEBUGLOG(5, "restart filling outBuff from beginning (left:%i, needed:%u)",
+                                (int)(zds->outBuffSize - zds->outStart),
+                                (U32)zds->fParams.blockSizeMax);
+                        zds->outStart = zds->outEnd = 0;
+                    }
+                    break;
+            }   }
+            /* cannot complete flush */
+            someMoreWork = 0;
+            break;
+
+        default:
+            assert(0);    /* impossible */
+            RETURN_ERROR(GENERIC, "impossible to reach");   /* some compiler require default to do something */
+    }   }
+
+    /* result */
+    input->pos = (size_t)(ip - (const char*)(input->src));
+    output->pos = (size_t)(op - (char*)(output->dst));
+
+    /* Update the expected output buffer for ZSTD_obm_stable. */
+    zds->expectedOutBuffer = *output;
+
+    if ((ip==istart) && (op==ostart)) {  /* no forward progress */
+        zds->noForwardProgress ++;
+        if (zds->noForwardProgress >= ZSTD_NO_FORWARD_PROGRESS_MAX) {
+            RETURN_ERROR_IF(op==oend, dstSize_tooSmall, "");
+            RETURN_ERROR_IF(ip==iend, srcSize_wrong, "");
+            assert(0);
+        }
+    } else {
+        zds->noForwardProgress = 0;
+    }
+    {   size_t nextSrcSizeHint = ZSTD_nextSrcSizeToDecompress(zds);
+        if (!nextSrcSizeHint) {   /* frame fully decoded */
+            if (zds->outEnd == zds->outStart) {  /* output fully flushed */
+                if (zds->hostageByte) {
+                    if (input->pos >= input->size) {
+                        /* can't release hostage (not present) */
+                        zds->streamStage = zdss_read;
+                        return 1;
+                    }
+                    input->pos++;  /* release hostage */
+                }   /* zds->hostageByte */
+                return 0;
+            }  /* zds->outEnd == zds->outStart */
+            if (!zds->hostageByte) { /* output not fully flushed; keep last byte as hostage; will be released when all output is flushed */
+                input->pos--;   /* note : pos > 0, otherwise, impossible to finish reading last block */
+                zds->hostageByte=1;
+            }
+            return 1;
+        }  /* nextSrcSizeHint==0 */
+        nextSrcSizeHint += ZSTD_blockHeaderSize * (ZSTD_nextInputType(zds) == ZSTDnit_block);   /* preload header of next block */
+        assert(zds->inPos <= nextSrcSizeHint);
+        nextSrcSizeHint -= zds->inPos;   /* part already loaded*/
+        return nextSrcSizeHint;
+    }
+}
+
+size_t ZSTD_decompressStream_simpleArgs (
+                            ZSTD_DCtx* dctx,
+                            void* dst, size_t dstCapacity, size_t* dstPos,
+                      const void* src, size_t srcSize, size_t* srcPos)
+{
+    ZSTD_outBuffer output = { dst, dstCapacity, *dstPos };
+    ZSTD_inBuffer  input  = { src, srcSize, *srcPos };
+    /* ZSTD_compress_generic() will check validity of dstPos and srcPos */
+    size_t const cErr = ZSTD_decompressStream(dctx, &output, &input);
+    *dstPos = output.pos;
+    *srcPos = input.pos;
+    return cErr;
+}
diff --git a/lib/zstd/decompress/zstd_decompress_block.c b/lib/zstd/decompress/zstd_decompress_block.c
new file mode 100644
index 000000000000..2d101d9a842e
--- /dev/null
+++ b/lib/zstd/decompress/zstd_decompress_block.c
@@ -0,0 +1,1540 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* zstd_decompress_block :
+ * this module takes care of decompressing _compressed_ block */
+
+/*-*******************************************************
+*  Dependencies
+*********************************************************/
+#include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
+#include "../common/compiler.h"    /* prefetch */
+#include "../common/cpu.h"         /* bmi2 */
+#include "../common/mem.h"         /* low level memory routines */
+#define FSE_STATIC_LINKING_ONLY
+#include "../common/fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include "../common/zstd_internal.h"
+#include "zstd_decompress_internal.h"   /* ZSTD_DCtx */
+#include "zstd_ddict.h"  /* ZSTD_DDictDictContent */
+#include "zstd_decompress_block.h"
+
+/*_*******************************************************
+*  Macros
+**********************************************************/
+
+/* These two optional macros force the use one way or another of the two
+ * ZSTD_decompressSequences implementations. You can't force in both directions
+ * at the same time.
+ */
+#if defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+#error "Cannot force the use of the short and the long ZSTD_decompressSequences variants!"
+#endif
+
+
+/*_*******************************************************
+*  Memory operations
+**********************************************************/
+static void ZSTD_copy4(void* dst, const void* src) { ZSTD_memcpy(dst, src, 4); }
+
+
+/*-*************************************************************
+ *   Block decoding
+ ***************************************************************/
+
+/*! ZSTD_getcBlockSize() :
+ *  Provides the size of compressed block from block header `src` */
+size_t ZSTD_getcBlockSize(const void* src, size_t srcSize,
+                          blockProperties_t* bpPtr)
+{
+    RETURN_ERROR_IF(srcSize < ZSTD_blockHeaderSize, srcSize_wrong, "");
+
+    {   U32 const cBlockHeader = MEM_readLE24(src);
+        U32 const cSize = cBlockHeader >> 3;
+        bpPtr->lastBlock = cBlockHeader & 1;
+        bpPtr->blockType = (blockType_e)((cBlockHeader >> 1) & 3);
+        bpPtr->origSize = cSize;   /* only useful for RLE */
+        if (bpPtr->blockType == bt_rle) return 1;
+        RETURN_ERROR_IF(bpPtr->blockType == bt_reserved, corruption_detected, "");
+        return cSize;
+    }
+}
+
+
+/* Hidden declaration for fullbench */
+size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
+                          const void* src, size_t srcSize);
+/*! ZSTD_decodeLiteralsBlock() :
+ * @return : nb of bytes read from src (< srcSize )
+ *  note : symbol not declared but exposed for fullbench */
+size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
+                          const void* src, size_t srcSize)   /* note : srcSize < BLOCKSIZE */
+{
+    DEBUGLOG(5, "ZSTD_decodeLiteralsBlock");
+    RETURN_ERROR_IF(srcSize < MIN_CBLOCK_SIZE, corruption_detected, "");
+
+    {   const BYTE* const istart = (const BYTE*) src;
+        symbolEncodingType_e const litEncType = (symbolEncodingType_e)(istart[0] & 3);
+
+        switch(litEncType)
+        {
+        case set_repeat:
+            DEBUGLOG(5, "set_repeat flag : re-using stats from previous compressed literals block");
+            RETURN_ERROR_IF(dctx->litEntropy==0, dictionary_corrupted, "");
+            ZSTD_FALLTHROUGH;
+
+        case set_compressed:
+            RETURN_ERROR_IF(srcSize < 5, corruption_detected, "srcSize >= MIN_CBLOCK_SIZE == 3; here we need up to 5 for case 3");
+            {   size_t lhSize, litSize, litCSize;
+                U32 singleStream=0;
+                U32 const lhlCode = (istart[0] >> 2) & 3;
+                U32 const lhc = MEM_readLE32(istart);
+                size_t hufSuccess;
+                switch(lhlCode)
+                {
+                case 0: case 1: default:   /* note : default is impossible, since lhlCode into [0..3] */
+                    /* 2 - 2 - 10 - 10 */
+                    singleStream = !lhlCode;
+                    lhSize = 3;
+                    litSize  = (lhc >> 4) & 0x3FF;
+                    litCSize = (lhc >> 14) & 0x3FF;
+                    break;
+                case 2:
+                    /* 2 - 2 - 14 - 14 */
+                    lhSize = 4;
+                    litSize  = (lhc >> 4) & 0x3FFF;
+                    litCSize = lhc >> 18;
+                    break;
+                case 3:
+                    /* 2 - 2 - 18 - 18 */
+                    lhSize = 5;
+                    litSize  = (lhc >> 4) & 0x3FFFF;
+                    litCSize = (lhc >> 22) + ((size_t)istart[4] << 10);
+                    break;
+                }
+                RETURN_ERROR_IF(litSize > ZSTD_BLOCKSIZE_MAX, corruption_detected, "");
+                RETURN_ERROR_IF(litCSize + lhSize > srcSize, corruption_detected, "");
+
+                /* prefetch huffman table if cold */
+                if (dctx->ddictIsCold && (litSize > 768 /* heuristic */)) {
+                    PREFETCH_AREA(dctx->HUFptr, sizeof(dctx->entropy.hufTable));
+                }
+
+                if (litEncType==set_repeat) {
+                    if (singleStream) {
+                        hufSuccess = HUF_decompress1X_usingDTable_bmi2(
+                            dctx->litBuffer, litSize, istart+lhSize, litCSize,
+                            dctx->HUFptr, dctx->bmi2);
+                    } else {
+                        hufSuccess = HUF_decompress4X_usingDTable_bmi2(
+                            dctx->litBuffer, litSize, istart+lhSize, litCSize,
+                            dctx->HUFptr, dctx->bmi2);
+                    }
+                } else {
+                    if (singleStream) {
+#if defined(HUF_FORCE_DECOMPRESS_X2)
+                        hufSuccess = HUF_decompress1X_DCtx_wksp(
+                            dctx->entropy.hufTable, dctx->litBuffer, litSize,
+                            istart+lhSize, litCSize, dctx->workspace,
+                            sizeof(dctx->workspace));
+#else
+                        hufSuccess = HUF_decompress1X1_DCtx_wksp_bmi2(
+                            dctx->entropy.hufTable, dctx->litBuffer, litSize,
+                            istart+lhSize, litCSize, dctx->workspace,
+                            sizeof(dctx->workspace), dctx->bmi2);
+#endif
+                    } else {
+                        hufSuccess = HUF_decompress4X_hufOnly_wksp_bmi2(
+                            dctx->entropy.hufTable, dctx->litBuffer, litSize,
+                            istart+lhSize, litCSize, dctx->workspace,
+                            sizeof(dctx->workspace), dctx->bmi2);
+                    }
+                }
+
+                RETURN_ERROR_IF(HUF_isError(hufSuccess), corruption_detected, "");
+
+                dctx->litPtr = dctx->litBuffer;
+                dctx->litSize = litSize;
+                dctx->litEntropy = 1;
+                if (litEncType==set_compressed) dctx->HUFptr = dctx->entropy.hufTable;
+                ZSTD_memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
+                return litCSize + lhSize;
+            }
+
+        case set_basic:
+            {   size_t litSize, lhSize;
+                U32 const lhlCode = ((istart[0]) >> 2) & 3;
+                switch(lhlCode)
+                {
+                case 0: case 2: default:   /* note : default is impossible, since lhlCode into [0..3] */
+                    lhSize = 1;
+                    litSize = istart[0] >> 3;
+                    break;
+                case 1:
+                    lhSize = 2;
+                    litSize = MEM_readLE16(istart) >> 4;
+                    break;
+                case 3:
+                    lhSize = 3;
+                    litSize = MEM_readLE24(istart) >> 4;
+                    break;
+                }
+
+                if (lhSize+litSize+WILDCOPY_OVERLENGTH > srcSize) {  /* risk reading beyond src buffer with wildcopy */
+                    RETURN_ERROR_IF(litSize+lhSize > srcSize, corruption_detected, "");
+                    ZSTD_memcpy(dctx->litBuffer, istart+lhSize, litSize);
+                    dctx->litPtr = dctx->litBuffer;
+                    dctx->litSize = litSize;
+                    ZSTD_memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
+                    return lhSize+litSize;
+                }
+                /* direct reference into compressed stream */
+                dctx->litPtr = istart+lhSize;
+                dctx->litSize = litSize;
+                return lhSize+litSize;
+            }
+
+        case set_rle:
+            {   U32 const lhlCode = ((istart[0]) >> 2) & 3;
+                size_t litSize, lhSize;
+                switch(lhlCode)
+                {
+                case 0: case 2: default:   /* note : default is impossible, since lhlCode into [0..3] */
+                    lhSize = 1;
+                    litSize = istart[0] >> 3;
+                    break;
+                case 1:
+                    lhSize = 2;
+                    litSize = MEM_readLE16(istart) >> 4;
+                    break;
+                case 3:
+                    lhSize = 3;
+                    litSize = MEM_readLE24(istart) >> 4;
+                    RETURN_ERROR_IF(srcSize<4, corruption_detected, "srcSize >= MIN_CBLOCK_SIZE == 3; here we need lhSize+1 = 4");
+                    break;
+                }
+                RETURN_ERROR_IF(litSize > ZSTD_BLOCKSIZE_MAX, corruption_detected, "");
+                ZSTD_memset(dctx->litBuffer, istart[lhSize], litSize + WILDCOPY_OVERLENGTH);
+                dctx->litPtr = dctx->litBuffer;
+                dctx->litSize = litSize;
+                return lhSize+1;
+            }
+        default:
+            RETURN_ERROR(corruption_detected, "impossible");
+        }
+    }
+}
+
+/* Default FSE distribution tables.
+ * These are pre-calculated FSE decoding tables using default distributions as defined in specification :
+ * https://github.com/facebook/zstd/blob/release/doc/zstd_compression_format.md#default-distributions
+ * They were generated programmatically with following method :
+ * - start from default distributions, present in /lib/common/zstd_internal.h
+ * - generate tables normally, using ZSTD_buildFSETable()
+ * - printout the content of tables
+ * - pretify output, report below, test with fuzzer to ensure it's correct */
+
+/* Default FSE distribution table for Literal Lengths */
+static const ZSTD_seqSymbol LL_defaultDTable[(1<<LL_DEFAULTNORMLOG)+1] = {
+     {  1,  1,  1, LL_DEFAULTNORMLOG},  /* header : fastMode, tableLog */
+     /* nextState, nbAddBits, nbBits, baseVal */
+     {  0,  0,  4,    0},  { 16,  0,  4,    0},
+     { 32,  0,  5,    1},  {  0,  0,  5,    3},
+     {  0,  0,  5,    4},  {  0,  0,  5,    6},
+     {  0,  0,  5,    7},  {  0,  0,  5,    9},
+     {  0,  0,  5,   10},  {  0,  0,  5,   12},
+     {  0,  0,  6,   14},  {  0,  1,  5,   16},
+     {  0,  1,  5,   20},  {  0,  1,  5,   22},
+     {  0,  2,  5,   28},  {  0,  3,  5,   32},
+     {  0,  4,  5,   48},  { 32,  6,  5,   64},
+     {  0,  7,  5,  128},  {  0,  8,  6,  256},
+     {  0, 10,  6, 1024},  {  0, 12,  6, 4096},
+     { 32,  0,  4,    0},  {  0,  0,  4,    1},
+     {  0,  0,  5,    2},  { 32,  0,  5,    4},
+     {  0,  0,  5,    5},  { 32,  0,  5,    7},
+     {  0,  0,  5,    8},  { 32,  0,  5,   10},
+     {  0,  0,  5,   11},  {  0,  0,  6,   13},
+     { 32,  1,  5,   16},  {  0,  1,  5,   18},
+     { 32,  1,  5,   22},  {  0,  2,  5,   24},
+     { 32,  3,  5,   32},  {  0,  3,  5,   40},
+     {  0,  6,  4,   64},  { 16,  6,  4,   64},
+     { 32,  7,  5,  128},  {  0,  9,  6,  512},
+     {  0, 11,  6, 2048},  { 48,  0,  4,    0},
+     { 16,  0,  4,    1},  { 32,  0,  5,    2},
+     { 32,  0,  5,    3},  { 32,  0,  5,    5},
+     { 32,  0,  5,    6},  { 32,  0,  5,    8},
+     { 32,  0,  5,    9},  { 32,  0,  5,   11},
+     { 32,  0,  5,   12},  {  0,  0,  6,   15},
+     { 32,  1,  5,   18},  { 32,  1,  5,   20},
+     { 32,  2,  5,   24},  { 32,  2,  5,   28},
+     { 32,  3,  5,   40},  { 32,  4,  5,   48},
+     {  0, 16,  6,65536},  {  0, 15,  6,32768},
+     {  0, 14,  6,16384},  {  0, 13,  6, 8192},
+};   /* LL_defaultDTable */
+
+/* Default FSE distribution table for Offset Codes */
+static const ZSTD_seqSymbol OF_defaultDTable[(1<<OF_DEFAULTNORMLOG)+1] = {
+    {  1,  1,  1, OF_DEFAULTNORMLOG},  /* header : fastMode, tableLog */
+    /* nextState, nbAddBits, nbBits, baseVal */
+    {  0,  0,  5,    0},     {  0,  6,  4,   61},
+    {  0,  9,  5,  509},     {  0, 15,  5,32765},
+    {  0, 21,  5,2097149},   {  0,  3,  5,    5},
+    {  0,  7,  4,  125},     {  0, 12,  5, 4093},
+    {  0, 18,  5,262141},    {  0, 23,  5,8388605},
+    {  0,  5,  5,   29},     {  0,  8,  4,  253},
+    {  0, 14,  5,16381},     {  0, 20,  5,1048573},
+    {  0,  2,  5,    1},     { 16,  7,  4,  125},
+    {  0, 11,  5, 2045},     {  0, 17,  5,131069},
+    {  0, 22,  5,4194301},   {  0,  4,  5,   13},
+    { 16,  8,  4,  253},     {  0, 13,  5, 8189},
+    {  0, 19,  5,524285},    {  0,  1,  5,    1},
+    { 16,  6,  4,   61},     {  0, 10,  5, 1021},
+    {  0, 16,  5,65533},     {  0, 28,  5,268435453},
+    {  0, 27,  5,134217725}, {  0, 26,  5,67108861},
+    {  0, 25,  5,33554429},  {  0, 24,  5,16777213},
+};   /* OF_defaultDTable */
+
+
+/* Default FSE distribution table for Match Lengths */
+static const ZSTD_seqSymbol ML_defaultDTable[(1<<ML_DEFAULTNORMLOG)+1] = {
+    {  1,  1,  1, ML_DEFAULTNORMLOG},  /* header : fastMode, tableLog */
+    /* nextState, nbAddBits, nbBits, baseVal */
+    {  0,  0,  6,    3},  {  0,  0,  4,    4},
+    { 32,  0,  5,    5},  {  0,  0,  5,    6},
+    {  0,  0,  5,    8},  {  0,  0,  5,    9},
+    {  0,  0,  5,   11},  {  0,  0,  6,   13},
+    {  0,  0,  6,   16},  {  0,  0,  6,   19},
+    {  0,  0,  6,   22},  {  0,  0,  6,   25},
+    {  0,  0,  6,   28},  {  0,  0,  6,   31},
+    {  0,  0,  6,   34},  {  0,  1,  6,   37},
+    {  0,  1,  6,   41},  {  0,  2,  6,   47},
+    {  0,  3,  6,   59},  {  0,  4,  6,   83},
+    {  0,  7,  6,  131},  {  0,  9,  6,  515},
+    { 16,  0,  4,    4},  {  0,  0,  4,    5},
+    { 32,  0,  5,    6},  {  0,  0,  5,    7},
+    { 32,  0,  5,    9},  {  0,  0,  5,   10},
+    {  0,  0,  6,   12},  {  0,  0,  6,   15},
+    {  0,  0,  6,   18},  {  0,  0,  6,   21},
+    {  0,  0,  6,   24},  {  0,  0,  6,   27},
+    {  0,  0,  6,   30},  {  0,  0,  6,   33},
+    {  0,  1,  6,   35},  {  0,  1,  6,   39},
+    {  0,  2,  6,   43},  {  0,  3,  6,   51},
+    {  0,  4,  6,   67},  {  0,  5,  6,   99},
+    {  0,  8,  6,  259},  { 32,  0,  4,    4},
+    { 48,  0,  4,    4},  { 16,  0,  4,    5},
+    { 32,  0,  5,    7},  { 32,  0,  5,    8},
+    { 32,  0,  5,   10},  { 32,  0,  5,   11},
+    {  0,  0,  6,   14},  {  0,  0,  6,   17},
+    {  0,  0,  6,   20},  {  0,  0,  6,   23},
+    {  0,  0,  6,   26},  {  0,  0,  6,   29},
+    {  0,  0,  6,   32},  {  0, 16,  6,65539},
+    {  0, 15,  6,32771},  {  0, 14,  6,16387},
+    {  0, 13,  6, 8195},  {  0, 12,  6, 4099},
+    {  0, 11,  6, 2051},  {  0, 10,  6, 1027},
+};   /* ML_defaultDTable */
+
+
+static void ZSTD_buildSeqTable_rle(ZSTD_seqSymbol* dt, U32 baseValue, U32 nbAddBits)
+{
+    void* ptr = dt;
+    ZSTD_seqSymbol_header* const DTableH = (ZSTD_seqSymbol_header*)ptr;
+    ZSTD_seqSymbol* const cell = dt + 1;
+
+    DTableH->tableLog = 0;
+    DTableH->fastMode = 0;
+
+    cell->nbBits = 0;
+    cell->nextState = 0;
+    assert(nbAddBits < 255);
+    cell->nbAdditionalBits = (BYTE)nbAddBits;
+    cell->baseValue = baseValue;
+}
+
+
+/* ZSTD_buildFSETable() :
+ * generate FSE decoding table for one symbol (ll, ml or off)
+ * cannot fail if input is valid =>
+ * all inputs are presumed validated at this stage */
+FORCE_INLINE_TEMPLATE
+void ZSTD_buildFSETable_body(ZSTD_seqSymbol* dt,
+            const short* normalizedCounter, unsigned maxSymbolValue,
+            const U32* baseValue, const U32* nbAdditionalBits,
+            unsigned tableLog, void* wksp, size_t wkspSize)
+{
+    ZSTD_seqSymbol* const tableDecode = dt+1;
+    U32 const maxSV1 = maxSymbolValue + 1;
+    U32 const tableSize = 1 << tableLog;
+
+    U16* symbolNext = (U16*)wksp;
+    BYTE* spread = (BYTE*)(symbolNext + MaxSeq + 1);
+    U32 highThreshold = tableSize - 1;
+
+
+    /* Sanity Checks */
+    assert(maxSymbolValue <= MaxSeq);
+    assert(tableLog <= MaxFSELog);
+    assert(wkspSize >= ZSTD_BUILD_FSE_TABLE_WKSP_SIZE);
+    (void)wkspSize;
+    /* Init, lay down lowprob symbols */
+    {   ZSTD_seqSymbol_header DTableH;
+        DTableH.tableLog = tableLog;
+        DTableH.fastMode = 1;
+        {   S16 const largeLimit= (S16)(1 << (tableLog-1));
+            U32 s;
+            for (s=0; s<maxSV1; s++) {
+                if (normalizedCounter[s]==-1) {
+                    tableDecode[highThreshold--].baseValue = s;
+                    symbolNext[s] = 1;
+                } else {
+                    if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
+                    assert(normalizedCounter[s]>=0);
+                    symbolNext[s] = (U16)normalizedCounter[s];
+        }   }   }
+        ZSTD_memcpy(dt, &DTableH, sizeof(DTableH));
+    }
+
+    /* Spread symbols */
+    assert(tableSize <= 512);
+    /* Specialized symbol spreading for the case when there are
+     * no low probability (-1 count) symbols. When compressing
+     * small blocks we avoid low probability symbols to hit this
+     * case, since header decoding speed matters more.
+     */
+    if (highThreshold == tableSize - 1) {
+        size_t const tableMask = tableSize-1;
+        size_t const step = FSE_TABLESTEP(tableSize);
+        /* First lay down the symbols in order.
+         * We use a uint64_t to lay down 8 bytes at a time. This reduces branch
+         * misses since small blocks generally have small table logs, so nearly
+         * all symbols have counts <= 8. We ensure we have 8 bytes at the end of
+         * our buffer to handle the over-write.
+         */
+        {
+            U64 const add = 0x0101010101010101ull;
+            size_t pos = 0;
+            U64 sv = 0;
+            U32 s;
+            for (s=0; s<maxSV1; ++s, sv += add) {
+                int i;
+                int const n = normalizedCounter[s];
+                MEM_write64(spread + pos, sv);
+                for (i = 8; i < n; i += 8) {
+                    MEM_write64(spread + pos + i, sv);
+                }
+                pos += n;
+            }
+        }
+        /* Now we spread those positions across the table.
+         * The benefit of doing it in two stages is that we avoid the the
+         * variable size inner loop, which caused lots of branch misses.
+         * Now we can run through all the positions without any branch misses.
+         * We unroll the loop twice, since that is what emperically worked best.
+         */
+        {
+            size_t position = 0;
+            size_t s;
+            size_t const unroll = 2;
+            assert(tableSize % unroll == 0); /* FSE_MIN_TABLELOG is 5 */
+            for (s = 0; s < (size_t)tableSize; s += unroll) {
+                size_t u;
+                for (u = 0; u < unroll; ++u) {
+                    size_t const uPosition = (position + (u * step)) & tableMask;
+                    tableDecode[uPosition].baseValue = spread[s + u];
+                }
+                position = (position + (unroll * step)) & tableMask;
+            }
+            assert(position == 0);
+        }
+    } else {
+        U32 const tableMask = tableSize-1;
+        U32 const step = FSE_TABLESTEP(tableSize);
+        U32 s, position = 0;
+        for (s=0; s<maxSV1; s++) {
+            int i;
+            int const n = normalizedCounter[s];
+            for (i=0; i<n; i++) {
+                tableDecode[position].baseValue = s;
+                position = (position + step) & tableMask;
+                while (position > highThreshold) position = (position + step) & tableMask;   /* lowprob area */
+        }   }
+        assert(position == 0); /* position must reach all cells once, otherwise normalizedCounter is incorrect */
+    }
+
+    /* Build Decoding table */
+    {
+        U32 u;
+        for (u=0; u<tableSize; u++) {
+            U32 const symbol = tableDecode[u].baseValue;
+            U32 const nextState = symbolNext[symbol]++;
+            tableDecode[u].nbBits = (BYTE) (tableLog - BIT_highbit32(nextState) );
+            tableDecode[u].nextState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
+            assert(nbAdditionalBits[symbol] < 255);
+            tableDecode[u].nbAdditionalBits = (BYTE)nbAdditionalBits[symbol];
+            tableDecode[u].baseValue = baseValue[symbol];
+        }
+    }
+}
+
+/* Avoids the FORCE_INLINE of the _body() function. */
+static void ZSTD_buildFSETable_body_default(ZSTD_seqSymbol* dt,
+            const short* normalizedCounter, unsigned maxSymbolValue,
+            const U32* baseValue, const U32* nbAdditionalBits,
+            unsigned tableLog, void* wksp, size_t wkspSize)
+{
+    ZSTD_buildFSETable_body(dt, normalizedCounter, maxSymbolValue,
+            baseValue, nbAdditionalBits, tableLog, wksp, wkspSize);
+}
+
+#if DYNAMIC_BMI2
+TARGET_ATTRIBUTE("bmi2") static void ZSTD_buildFSETable_body_bmi2(ZSTD_seqSymbol* dt,
+            const short* normalizedCounter, unsigned maxSymbolValue,
+            const U32* baseValue, const U32* nbAdditionalBits,
+            unsigned tableLog, void* wksp, size_t wkspSize)
+{
+    ZSTD_buildFSETable_body(dt, normalizedCounter, maxSymbolValue,
+            baseValue, nbAdditionalBits, tableLog, wksp, wkspSize);
+}
+#endif
+
+void ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
+            const short* normalizedCounter, unsigned maxSymbolValue,
+            const U32* baseValue, const U32* nbAdditionalBits,
+            unsigned tableLog, void* wksp, size_t wkspSize, int bmi2)
+{
+#if DYNAMIC_BMI2
+    if (bmi2) {
+        ZSTD_buildFSETable_body_bmi2(dt, normalizedCounter, maxSymbolValue,
+                baseValue, nbAdditionalBits, tableLog, wksp, wkspSize);
+        return;
+    }
+#endif
+    (void)bmi2;
+    ZSTD_buildFSETable_body_default(dt, normalizedCounter, maxSymbolValue,
+            baseValue, nbAdditionalBits, tableLog, wksp, wkspSize);
+}
+
+
+/*! ZSTD_buildSeqTable() :
+ * @return : nb bytes read from src,
+ *           or an error code if it fails */
+static size_t ZSTD_buildSeqTable(ZSTD_seqSymbol* DTableSpace, const ZSTD_seqSymbol** DTablePtr,
+                                 symbolEncodingType_e type, unsigned max, U32 maxLog,
+                                 const void* src, size_t srcSize,
+                                 const U32* baseValue, const U32* nbAdditionalBits,
+                                 const ZSTD_seqSymbol* defaultTable, U32 flagRepeatTable,
+                                 int ddictIsCold, int nbSeq, U32* wksp, size_t wkspSize,
+                                 int bmi2)
+{
+    switch(type)
+    {
+    case set_rle :
+        RETURN_ERROR_IF(!srcSize, srcSize_wrong, "");
+        RETURN_ERROR_IF((*(const BYTE*)src) > max, corruption_detected, "");
+        {   U32 const symbol = *(const BYTE*)src;
+            U32 const baseline = baseValue[symbol];
+            U32 const nbBits = nbAdditionalBits[symbol];
+            ZSTD_buildSeqTable_rle(DTableSpace, baseline, nbBits);
+        }
+        *DTablePtr = DTableSpace;
+        return 1;
+    case set_basic :
+        *DTablePtr = defaultTable;
+        return 0;
+    case set_repeat:
+        RETURN_ERROR_IF(!flagRepeatTable, corruption_detected, "");
+        /* prefetch FSE table if used */
+        if (ddictIsCold && (nbSeq > 24 /* heuristic */)) {
+            const void* const pStart = *DTablePtr;
+            size_t const pSize = sizeof(ZSTD_seqSymbol) * (SEQSYMBOL_TABLE_SIZE(maxLog));
+            PREFETCH_AREA(pStart, pSize);
+        }
+        return 0;
+    case set_compressed :
+        {   unsigned tableLog;
+            S16 norm[MaxSeq+1];
+            size_t const headerSize = FSE_readNCount(norm, &max, &tableLog, src, srcSize);
+            RETURN_ERROR_IF(FSE_isError(headerSize), corruption_detected, "");
+            RETURN_ERROR_IF(tableLog > maxLog, corruption_detected, "");
+            ZSTD_buildFSETable(DTableSpace, norm, max, baseValue, nbAdditionalBits, tableLog, wksp, wkspSize, bmi2);
+            *DTablePtr = DTableSpace;
+            return headerSize;
+        }
+    default :
+        assert(0);
+        RETURN_ERROR(GENERIC, "impossible");
+    }
+}
+
+size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr,
+                             const void* src, size_t srcSize)
+{
+    const BYTE* const istart = (const BYTE*)src;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* ip = istart;
+    int nbSeq;
+    DEBUGLOG(5, "ZSTD_decodeSeqHeaders");
+
+    /* check */
+    RETURN_ERROR_IF(srcSize < MIN_SEQUENCES_SIZE, srcSize_wrong, "");
+
+    /* SeqHead */
+    nbSeq = *ip++;
+    if (!nbSeq) {
+        *nbSeqPtr=0;
+        RETURN_ERROR_IF(srcSize != 1, srcSize_wrong, "");
+        return 1;
+    }
+    if (nbSeq > 0x7F) {
+        if (nbSeq == 0xFF) {
+            RETURN_ERROR_IF(ip+2 > iend, srcSize_wrong, "");
+            nbSeq = MEM_readLE16(ip) + LONGNBSEQ;
+            ip+=2;
+        } else {
+            RETURN_ERROR_IF(ip >= iend, srcSize_wrong, "");
+            nbSeq = ((nbSeq-0x80)<<8) + *ip++;
+        }
+    }
+    *nbSeqPtr = nbSeq;
+
+    /* FSE table descriptors */
+    RETURN_ERROR_IF(ip+1 > iend, srcSize_wrong, ""); /* minimum possible size: 1 byte for symbol encoding types */
+    {   symbolEncodingType_e const LLtype = (symbolEncodingType_e)(*ip >> 6);
+        symbolEncodingType_e const OFtype = (symbolEncodingType_e)((*ip >> 4) & 3);
+        symbolEncodingType_e const MLtype = (symbolEncodingType_e)((*ip >> 2) & 3);
+        ip++;
+
+        /* Build DTables */
+        {   size_t const llhSize = ZSTD_buildSeqTable(dctx->entropy.LLTable, &dctx->LLTptr,
+                                                      LLtype, MaxLL, LLFSELog,
+                                                      ip, iend-ip,
+                                                      LL_base, LL_bits,
+                                                      LL_defaultDTable, dctx->fseEntropy,
+                                                      dctx->ddictIsCold, nbSeq,
+                                                      dctx->workspace, sizeof(dctx->workspace),
+                                                      dctx->bmi2);
+            RETURN_ERROR_IF(ZSTD_isError(llhSize), corruption_detected, "ZSTD_buildSeqTable failed");
+            ip += llhSize;
+        }
+
+        {   size_t const ofhSize = ZSTD_buildSeqTable(dctx->entropy.OFTable, &dctx->OFTptr,
+                                                      OFtype, MaxOff, OffFSELog,
+                                                      ip, iend-ip,
+                                                      OF_base, OF_bits,
+                                                      OF_defaultDTable, dctx->fseEntropy,
+                                                      dctx->ddictIsCold, nbSeq,
+                                                      dctx->workspace, sizeof(dctx->workspace),
+                                                      dctx->bmi2);
+            RETURN_ERROR_IF(ZSTD_isError(ofhSize), corruption_detected, "ZSTD_buildSeqTable failed");
+            ip += ofhSize;
+        }
+
+        {   size_t const mlhSize = ZSTD_buildSeqTable(dctx->entropy.MLTable, &dctx->MLTptr,
+                                                      MLtype, MaxML, MLFSELog,
+                                                      ip, iend-ip,
+                                                      ML_base, ML_bits,
+                                                      ML_defaultDTable, dctx->fseEntropy,
+                                                      dctx->ddictIsCold, nbSeq,
+                                                      dctx->workspace, sizeof(dctx->workspace),
+                                                      dctx->bmi2);
+            RETURN_ERROR_IF(ZSTD_isError(mlhSize), corruption_detected, "ZSTD_buildSeqTable failed");
+            ip += mlhSize;
+        }
+    }
+
+    return ip-istart;
+}
+
+
+typedef struct {
+    size_t litLength;
+    size_t matchLength;
+    size_t offset;
+    const BYTE* match;
+} seq_t;
+
+typedef struct {
+    size_t state;
+    const ZSTD_seqSymbol* table;
+} ZSTD_fseState;
+
+typedef struct {
+    BIT_DStream_t DStream;
+    ZSTD_fseState stateLL;
+    ZSTD_fseState stateOffb;
+    ZSTD_fseState stateML;
+    size_t prevOffset[ZSTD_REP_NUM];
+    const BYTE* prefixStart;
+    const BYTE* dictEnd;
+    size_t pos;
+} seqState_t;
+
+/*! ZSTD_overlapCopy8() :
+ *  Copies 8 bytes from ip to op and updates op and ip where ip <= op.
+ *  If the offset is < 8 then the offset is spread to at least 8 bytes.
+ *
+ *  Precondition: *ip <= *op
+ *  Postcondition: *op - *op >= 8
+ */
+HINT_INLINE void ZSTD_overlapCopy8(BYTE** op, BYTE const** ip, size_t offset) {
+    assert(*ip <= *op);
+    if (offset < 8) {
+        /* close range match, overlap */
+        static const U32 dec32table[] = { 0, 1, 2, 1, 4, 4, 4, 4 };   /* added */
+        static const int dec64table[] = { 8, 8, 8, 7, 8, 9,10,11 };   /* subtracted */
+        int const sub2 = dec64table[offset];
+        (*op)[0] = (*ip)[0];
+        (*op)[1] = (*ip)[1];
+        (*op)[2] = (*ip)[2];
+        (*op)[3] = (*ip)[3];
+        *ip += dec32table[offset];
+        ZSTD_copy4(*op+4, *ip);
+        *ip -= sub2;
+    } else {
+        ZSTD_copy8(*op, *ip);
+    }
+    *ip += 8;
+    *op += 8;
+    assert(*op - *ip >= 8);
+}
+
+/*! ZSTD_safecopy() :
+ *  Specialized version of memcpy() that is allowed to READ up to WILDCOPY_OVERLENGTH past the input buffer
+ *  and write up to 16 bytes past oend_w (op >= oend_w is allowed).
+ *  This function is only called in the uncommon case where the sequence is near the end of the block. It
+ *  should be fast for a single long sequence, but can be slow for several short sequences.
+ *
+ *  @param ovtype controls the overlap detection
+ *         - ZSTD_no_overlap: The source and destination are guaranteed to be at least WILDCOPY_VECLEN bytes apart.
+ *         - ZSTD_overlap_src_before_dst: The src and dst may overlap and may be any distance apart.
+ *           The src buffer must be before the dst buffer.
+ */
+static void ZSTD_safecopy(BYTE* op, BYTE* const oend_w, BYTE const* ip, ptrdiff_t length, ZSTD_overlap_e ovtype) {
+    ptrdiff_t const diff = op - ip;
+    BYTE* const oend = op + length;
+
+    assert((ovtype == ZSTD_no_overlap && (diff <= -8 || diff >= 8 || op >= oend_w)) ||
+           (ovtype == ZSTD_overlap_src_before_dst && diff >= 0));
+
+    if (length < 8) {
+        /* Handle short lengths. */
+        while (op < oend) *op++ = *ip++;
+        return;
+    }
+    if (ovtype == ZSTD_overlap_src_before_dst) {
+        /* Copy 8 bytes and ensure the offset >= 8 when there can be overlap. */
+        assert(length >= 8);
+        ZSTD_overlapCopy8(&op, &ip, diff);
+        assert(op - ip >= 8);
+        assert(op <= oend);
+    }
+
+    if (oend <= oend_w) {
+        /* No risk of overwrite. */
+        ZSTD_wildcopy(op, ip, length, ovtype);
+        return;
+    }
+    if (op <= oend_w) {
+        /* Wildcopy until we get close to the end. */
+        assert(oend > oend_w);
+        ZSTD_wildcopy(op, ip, oend_w - op, ovtype);
+        ip += oend_w - op;
+        op = oend_w;
+    }
+    /* Handle the leftovers. */
+    while (op < oend) *op++ = *ip++;
+}
+
+/* ZSTD_execSequenceEnd():
+ * This version handles cases that are near the end of the output buffer. It requires
+ * more careful checks to make sure there is no overflow. By separating out these hard
+ * and unlikely cases, we can speed up the common cases.
+ *
+ * NOTE: This function needs to be fast for a single long sequence, but doesn't need
+ * to be optimized for many small sequences, since those fall into ZSTD_execSequence().
+ */
+FORCE_NOINLINE
+size_t ZSTD_execSequenceEnd(BYTE* op,
+                            BYTE* const oend, seq_t sequence,
+                            const BYTE** litPtr, const BYTE* const litLimit,
+                            const BYTE* const prefixStart, const BYTE* const virtualStart, const BYTE* const dictEnd)
+{
+    BYTE* const oLitEnd = op + sequence.litLength;
+    size_t const sequenceLength = sequence.litLength + sequence.matchLength;
+    const BYTE* const iLitEnd = *litPtr + sequence.litLength;
+    const BYTE* match = oLitEnd - sequence.offset;
+    BYTE* const oend_w = oend - WILDCOPY_OVERLENGTH;
+
+    /* bounds checks : careful of address space overflow in 32-bit mode */
+    RETURN_ERROR_IF(sequenceLength > (size_t)(oend - op), dstSize_tooSmall, "last match must fit within dstBuffer");
+    RETURN_ERROR_IF(sequence.litLength > (size_t)(litLimit - *litPtr), corruption_detected, "try to read beyond literal buffer");
+    assert(op < op + sequenceLength);
+    assert(oLitEnd < op + sequenceLength);
+
+    /* copy literals */
+    ZSTD_safecopy(op, oend_w, *litPtr, sequence.litLength, ZSTD_no_overlap);
+    op = oLitEnd;
+    *litPtr = iLitEnd;
+
+    /* copy Match */
+    if (sequence.offset > (size_t)(oLitEnd - prefixStart)) {
+        /* offset beyond prefix */
+        RETURN_ERROR_IF(sequence.offset > (size_t)(oLitEnd - virtualStart), corruption_detected, "");
+        match = dictEnd - (prefixStart-match);
+        if (match + sequence.matchLength <= dictEnd) {
+            ZSTD_memmove(oLitEnd, match, sequence.matchLength);
+            return sequenceLength;
+        }
+        /* span extDict & currentPrefixSegment */
+        {   size_t const length1 = dictEnd - match;
+            ZSTD_memmove(oLitEnd, match, length1);
+            op = oLitEnd + length1;
+            sequence.matchLength -= length1;
+            match = prefixStart;
+    }   }
+    ZSTD_safecopy(op, oend_w, match, sequence.matchLength, ZSTD_overlap_src_before_dst);
+    return sequenceLength;
+}
+
+HINT_INLINE
+size_t ZSTD_execSequence(BYTE* op,
+                         BYTE* const oend, seq_t sequence,
+                         const BYTE** litPtr, const BYTE* const litLimit,
+                         const BYTE* const prefixStart, const BYTE* const virtualStart, const BYTE* const dictEnd)
+{
+    BYTE* const oLitEnd = op + sequence.litLength;
+    size_t const sequenceLength = sequence.litLength + sequence.matchLength;
+    BYTE* const oMatchEnd = op + sequenceLength;   /* risk : address space overflow (32-bits) */
+    BYTE* const oend_w = oend - WILDCOPY_OVERLENGTH;   /* risk : address space underflow on oend=NULL */
+    const BYTE* const iLitEnd = *litPtr + sequence.litLength;
+    const BYTE* match = oLitEnd - sequence.offset;
+
+    assert(op != NULL /* Precondition */);
+    assert(oend_w < oend /* No underflow */);
+    /* Handle edge cases in a slow path:
+     *   - Read beyond end of literals
+     *   - Match end is within WILDCOPY_OVERLIMIT of oend
+     *   - 32-bit mode and the match length overflows
+     */
+    if (UNLIKELY(
+            iLitEnd > litLimit ||
+            oMatchEnd > oend_w ||
+            (MEM_32bits() && (size_t)(oend - op) < sequenceLength + WILDCOPY_OVERLENGTH)))
+        return ZSTD_execSequenceEnd(op, oend, sequence, litPtr, litLimit, prefixStart, virtualStart, dictEnd);
+
+    /* Assumptions (everything else goes into ZSTD_execSequenceEnd()) */
+    assert(op <= oLitEnd /* No overflow */);
+    assert(oLitEnd < oMatchEnd /* Non-zero match & no overflow */);
+    assert(oMatchEnd <= oend /* No underflow */);
+    assert(iLitEnd <= litLimit /* Literal length is in bounds */);
+    assert(oLitEnd <= oend_w /* Can wildcopy literals */);
+    assert(oMatchEnd <= oend_w /* Can wildcopy matches */);
+
+    /* Copy Literals:
+     * Split out litLength <= 16 since it is nearly always true. +1.6% on gcc-9.
+     * We likely don't need the full 32-byte wildcopy.
+     */
+    assert(WILDCOPY_OVERLENGTH >= 16);
+    ZSTD_copy16(op, (*litPtr));
+    if (UNLIKELY(sequence.litLength > 16)) {
+        ZSTD_wildcopy(op+16, (*litPtr)+16, sequence.litLength-16, ZSTD_no_overlap);
+    }
+    op = oLitEnd;
+    *litPtr = iLitEnd;   /* update for next sequence */
+
+    /* Copy Match */
+    if (sequence.offset > (size_t)(oLitEnd - prefixStart)) {
+        /* offset beyond prefix -> go into extDict */
+        RETURN_ERROR_IF(UNLIKELY(sequence.offset > (size_t)(oLitEnd - virtualStart)), corruption_detected, "");
+        match = dictEnd + (match - prefixStart);
+        if (match + sequence.matchLength <= dictEnd) {
+            ZSTD_memmove(oLitEnd, match, sequence.matchLength);
+            return sequenceLength;
+        }
+        /* span extDict & currentPrefixSegment */
+        {   size_t const length1 = dictEnd - match;
+            ZSTD_memmove(oLitEnd, match, length1);
+            op = oLitEnd + length1;
+            sequence.matchLength -= length1;
+            match = prefixStart;
+    }   }
+    /* Match within prefix of 1 or more bytes */
+    assert(op <= oMatchEnd);
+    assert(oMatchEnd <= oend_w);
+    assert(match >= prefixStart);
+    assert(sequence.matchLength >= 1);
+
+    /* Nearly all offsets are >= WILDCOPY_VECLEN bytes, which means we can use wildcopy
+     * without overlap checking.
+     */
+    if (LIKELY(sequence.offset >= WILDCOPY_VECLEN)) {
+        /* We bet on a full wildcopy for matches, since we expect matches to be
+         * longer than literals (in general). In silesia, ~10% of matches are longer
+         * than 16 bytes.
+         */
+        ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength, ZSTD_no_overlap);
+        return sequenceLength;
+    }
+    assert(sequence.offset < WILDCOPY_VECLEN);
+
+    /* Copy 8 bytes and spread the offset to be >= 8. */
+    ZSTD_overlapCopy8(&op, &match, sequence.offset);
+
+    /* If the match length is > 8 bytes, then continue with the wildcopy. */
+    if (sequence.matchLength > 8) {
+        assert(op < oMatchEnd);
+        ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength-8, ZSTD_overlap_src_before_dst);
+    }
+    return sequenceLength;
+}
+
+static void
+ZSTD_initFseState(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD, const ZSTD_seqSymbol* dt)
+{
+    const void* ptr = dt;
+    const ZSTD_seqSymbol_header* const DTableH = (const ZSTD_seqSymbol_header*)ptr;
+    DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
+    DEBUGLOG(6, "ZSTD_initFseState : val=%u using %u bits",
+                (U32)DStatePtr->state, DTableH->tableLog);
+    BIT_reloadDStream(bitD);
+    DStatePtr->table = dt + 1;
+}
+
+FORCE_INLINE_TEMPLATE void
+ZSTD_updateFseState(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD)
+{
+    ZSTD_seqSymbol const DInfo = DStatePtr->table[DStatePtr->state];
+    U32 const nbBits = DInfo.nbBits;
+    size_t const lowBits = BIT_readBits(bitD, nbBits);
+    DStatePtr->state = DInfo.nextState + lowBits;
+}
+
+FORCE_INLINE_TEMPLATE void
+ZSTD_updateFseStateWithDInfo(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD, ZSTD_seqSymbol const DInfo)
+{
+    U32 const nbBits = DInfo.nbBits;
+    size_t const lowBits = BIT_readBits(bitD, nbBits);
+    DStatePtr->state = DInfo.nextState + lowBits;
+}
+
+/* We need to add at most (ZSTD_WINDOWLOG_MAX_32 - 1) bits to read the maximum
+ * offset bits. But we can only read at most (STREAM_ACCUMULATOR_MIN_32 - 1)
+ * bits before reloading. This value is the maximum number of bytes we read
+ * after reloading when we are decoding long offsets.
+ */
+#define LONG_OFFSETS_MAX_EXTRA_BITS_32                       \
+    (ZSTD_WINDOWLOG_MAX_32 > STREAM_ACCUMULATOR_MIN_32       \
+        ? ZSTD_WINDOWLOG_MAX_32 - STREAM_ACCUMULATOR_MIN_32  \
+        : 0)
+
+typedef enum { ZSTD_lo_isRegularOffset, ZSTD_lo_isLongOffset=1 } ZSTD_longOffset_e;
+typedef enum { ZSTD_p_noPrefetch=0, ZSTD_p_prefetch=1 } ZSTD_prefetch_e;
+
+FORCE_INLINE_TEMPLATE seq_t
+ZSTD_decodeSequence(seqState_t* seqState, const ZSTD_longOffset_e longOffsets, const ZSTD_prefetch_e prefetch)
+{
+    seq_t seq;
+    ZSTD_seqSymbol const llDInfo = seqState->stateLL.table[seqState->stateLL.state];
+    ZSTD_seqSymbol const mlDInfo = seqState->stateML.table[seqState->stateML.state];
+    ZSTD_seqSymbol const ofDInfo = seqState->stateOffb.table[seqState->stateOffb.state];
+    U32 const llBase = llDInfo.baseValue;
+    U32 const mlBase = mlDInfo.baseValue;
+    U32 const ofBase = ofDInfo.baseValue;
+    BYTE const llBits = llDInfo.nbAdditionalBits;
+    BYTE const mlBits = mlDInfo.nbAdditionalBits;
+    BYTE const ofBits = ofDInfo.nbAdditionalBits;
+    BYTE const totalBits = llBits+mlBits+ofBits;
+
+    /* sequence */
+    {   size_t offset;
+        if (ofBits > 1) {
+            ZSTD_STATIC_ASSERT(ZSTD_lo_isLongOffset == 1);
+            ZSTD_STATIC_ASSERT(LONG_OFFSETS_MAX_EXTRA_BITS_32 == 5);
+            assert(ofBits <= MaxOff);
+            if (MEM_32bits() && longOffsets && (ofBits >= STREAM_ACCUMULATOR_MIN_32)) {
+                U32 const extraBits = ofBits - MIN(ofBits, 32 - seqState->DStream.bitsConsumed);
+                offset = ofBase + (BIT_readBitsFast(&seqState->DStream, ofBits - extraBits) << extraBits);
+                BIT_reloadDStream(&seqState->DStream);
+                if (extraBits) offset += BIT_readBitsFast(&seqState->DStream, extraBits);
+                assert(extraBits <= LONG_OFFSETS_MAX_EXTRA_BITS_32);   /* to avoid another reload */
+            } else {
+                offset = ofBase + BIT_readBitsFast(&seqState->DStream, ofBits/*>0*/);   /* <=  (ZSTD_WINDOWLOG_MAX-1) bits */
+                if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream);
+            }
+            seqState->prevOffset[2] = seqState->prevOffset[1];
+            seqState->prevOffset[1] = seqState->prevOffset[0];
+            seqState->prevOffset[0] = offset;
+        } else {
+            U32 const ll0 = (llBase == 0);
+            if (LIKELY((ofBits == 0))) {
+                if (LIKELY(!ll0))
+                    offset = seqState->prevOffset[0];
+                else {
+                    offset = seqState->prevOffset[1];
+                    seqState->prevOffset[1] = seqState->prevOffset[0];
+                    seqState->prevOffset[0] = offset;
+                }
+            } else {
+                offset = ofBase + ll0 + BIT_readBitsFast(&seqState->DStream, 1);
+                {   size_t temp = (offset==3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
+                    temp += !temp;   /* 0 is not valid; input is corrupted; force offset to 1 */
+                    if (offset != 1) seqState->prevOffset[2] = seqState->prevOffset[1];
+                    seqState->prevOffset[1] = seqState->prevOffset[0];
+                    seqState->prevOffset[0] = offset = temp;
+        }   }   }
+        seq.offset = offset;
+    }
+
+    seq.matchLength = mlBase;
+    if (mlBits > 0)
+        seq.matchLength += BIT_readBitsFast(&seqState->DStream, mlBits/*>0*/);
+
+    if (MEM_32bits() && (mlBits+llBits >= STREAM_ACCUMULATOR_MIN_32-LONG_OFFSETS_MAX_EXTRA_BITS_32))
+        BIT_reloadDStream(&seqState->DStream);
+    if (MEM_64bits() && UNLIKELY(totalBits >= STREAM_ACCUMULATOR_MIN_64-(LLFSELog+MLFSELog+OffFSELog)))
+        BIT_reloadDStream(&seqState->DStream);
+    /* Ensure there are enough bits to read the rest of data in 64-bit mode. */
+    ZSTD_STATIC_ASSERT(16+LLFSELog+MLFSELog+OffFSELog < STREAM_ACCUMULATOR_MIN_64);
+
+    seq.litLength = llBase;
+    if (llBits > 0)
+        seq.litLength += BIT_readBitsFast(&seqState->DStream, llBits/*>0*/);
+
+    if (MEM_32bits())
+        BIT_reloadDStream(&seqState->DStream);
+
+    DEBUGLOG(6, "seq: litL=%u, matchL=%u, offset=%u",
+                (U32)seq.litLength, (U32)seq.matchLength, (U32)seq.offset);
+
+    if (prefetch == ZSTD_p_prefetch) {
+        size_t const pos = seqState->pos + seq.litLength;
+        const BYTE* const matchBase = (seq.offset > pos) ? seqState->dictEnd : seqState->prefixStart;
+        seq.match = matchBase + pos - seq.offset;  /* note : this operation can overflow when seq.offset is really too large, which can only happen when input is corrupted.
+                                                    * No consequence though : no memory access will occur, offset is only used for prefetching */
+        seqState->pos = pos + seq.matchLength;
+    }
+
+    /* ANS state update
+     * gcc-9.0.0 does 2.5% worse with ZSTD_updateFseStateWithDInfo().
+     * clang-9.2.0 does 7% worse with ZSTD_updateFseState().
+     * Naturally it seems like ZSTD_updateFseStateWithDInfo() should be the
+     * better option, so it is the default for other compilers. But, if you
+     * measure that it is worse, please put up a pull request.
+     */
+    {
+#if !defined(__clang__)
+        const int kUseUpdateFseState = 1;
+#else
+        const int kUseUpdateFseState = 0;
+#endif
+        if (kUseUpdateFseState) {
+            ZSTD_updateFseState(&seqState->stateLL, &seqState->DStream);    /* <=  9 bits */
+            ZSTD_updateFseState(&seqState->stateML, &seqState->DStream);    /* <=  9 bits */
+            if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream);    /* <= 18 bits */
+            ZSTD_updateFseState(&seqState->stateOffb, &seqState->DStream);  /* <=  8 bits */
+        } else {
+            ZSTD_updateFseStateWithDInfo(&seqState->stateLL, &seqState->DStream, llDInfo);    /* <=  9 bits */
+            ZSTD_updateFseStateWithDInfo(&seqState->stateML, &seqState->DStream, mlDInfo);    /* <=  9 bits */
+            if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream);    /* <= 18 bits */
+            ZSTD_updateFseStateWithDInfo(&seqState->stateOffb, &seqState->DStream, ofDInfo);  /* <=  8 bits */
+        }
+    }
+
+    return seq;
+}
+
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+MEM_STATIC int ZSTD_dictionaryIsActive(ZSTD_DCtx const* dctx, BYTE const* prefixStart, BYTE const* oLitEnd)
+{
+    size_t const windowSize = dctx->fParams.windowSize;
+    /* No dictionary used. */
+    if (dctx->dictContentEndForFuzzing == NULL) return 0;
+    /* Dictionary is our prefix. */
+    if (prefixStart == dctx->dictContentBeginForFuzzing) return 1;
+    /* Dictionary is not our ext-dict. */
+    if (dctx->dictEnd != dctx->dictContentEndForFuzzing) return 0;
+    /* Dictionary is not within our window size. */
+    if ((size_t)(oLitEnd - prefixStart) >= windowSize) return 0;
+    /* Dictionary is active. */
+    return 1;
+}
+
+MEM_STATIC void ZSTD_assertValidSequence(
+        ZSTD_DCtx const* dctx,
+        BYTE const* op, BYTE const* oend,
+        seq_t const seq,
+        BYTE const* prefixStart, BYTE const* virtualStart)
+{
+#if DEBUGLEVEL >= 1
+    size_t const windowSize = dctx->fParams.windowSize;
+    size_t const sequenceSize = seq.litLength + seq.matchLength;
+    BYTE const* const oLitEnd = op + seq.litLength;
+    DEBUGLOG(6, "Checking sequence: litL=%u matchL=%u offset=%u",
+            (U32)seq.litLength, (U32)seq.matchLength, (U32)seq.offset);
+    assert(op <= oend);
+    assert((size_t)(oend - op) >= sequenceSize);
+    assert(sequenceSize <= ZSTD_BLOCKSIZE_MAX);
+    if (ZSTD_dictionaryIsActive(dctx, prefixStart, oLitEnd)) {
+        size_t const dictSize = (size_t)((char const*)dctx->dictContentEndForFuzzing - (char const*)dctx->dictContentBeginForFuzzing);
+        /* Offset must be within the dictionary. */
+        assert(seq.offset <= (size_t)(oLitEnd - virtualStart));
+        assert(seq.offset <= windowSize + dictSize);
+    } else {
+        /* Offset must be within our window. */
+        assert(seq.offset <= windowSize);
+    }
+#else
+    (void)dctx, (void)op, (void)oend, (void)seq, (void)prefixStart, (void)virtualStart;
+#endif
+}
+#endif
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
+FORCE_INLINE_TEMPLATE size_t
+DONT_VECTORIZE
+ZSTD_decompressSequences_body( ZSTD_DCtx* dctx,
+                               void* dst, size_t maxDstSize,
+                         const void* seqStart, size_t seqSize, int nbSeq,
+                         const ZSTD_longOffset_e isLongOffset,
+                         const int frame)
+{
+    const BYTE* ip = (const BYTE*)seqStart;
+    const BYTE* const iend = ip + seqSize;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + maxDstSize;
+    BYTE* op = ostart;
+    const BYTE* litPtr = dctx->litPtr;
+    const BYTE* const litEnd = litPtr + dctx->litSize;
+    const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart);
+    const BYTE* const vBase = (const BYTE*) (dctx->virtualStart);
+    const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd);
+    DEBUGLOG(5, "ZSTD_decompressSequences_body");
+    (void)frame;
+
+    /* Regen sequences */
+    if (nbSeq) {
+        seqState_t seqState;
+        size_t error = 0;
+        dctx->fseEntropy = 1;
+        { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) seqState.prevOffset[i] = dctx->entropy.rep[i]; }
+        RETURN_ERROR_IF(
+            ERR_isError(BIT_initDStream(&seqState.DStream, ip, iend-ip)),
+            corruption_detected, "");
+        ZSTD_initFseState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
+        ZSTD_initFseState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
+        ZSTD_initFseState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
+        assert(dst != NULL);
+
+        ZSTD_STATIC_ASSERT(
+                BIT_DStream_unfinished < BIT_DStream_completed &&
+                BIT_DStream_endOfBuffer < BIT_DStream_completed &&
+                BIT_DStream_completed < BIT_DStream_overflow);
+
+#if defined(__x86_64__)
+        /* Align the decompression loop to 32 + 16 bytes.
+         *
+         * zstd compiled with gcc-9 on an Intel i9-9900k shows 10% decompression
+         * speed swings based on the alignment of the decompression loop. This
+         * performance swing is caused by parts of the decompression loop falling
+         * out of the DSB. The entire decompression loop should fit in the DSB,
+         * when it can't we get much worse performance. You can measure if you've
+         * hit the good case or the bad case with this perf command for some
+         * compressed file test.zst:
+         *
+         *   perf stat -e cycles -e instructions -e idq.all_dsb_cycles_any_uops \
+         *             -e idq.all_mite_cycles_any_uops -- ./zstd -tq test.zst
+         *
+         * If you see most cycles served out of the MITE you've hit the bad case.
+         * If you see most cycles served out of the DSB you've hit the good case.
+         * If it is pretty even then you may be in an okay case.
+         *
+         * I've been able to reproduce this issue on the following CPUs:
+         *   - Kabylake: Macbook Pro (15-inch, 2019) 2.4 GHz Intel Core i9
+         *               Use Instruments->Counters to get DSB/MITE cycles.
+         *               I never got performance swings, but I was able to
+         *               go from the good case of mostly DSB to half of the
+         *               cycles served from MITE.
+         *   - Coffeelake: Intel i9-9900k
+         *
+         * I haven't been able to reproduce the instability or DSB misses on any
+         * of the following CPUS:
+         *   - Haswell
+         *   - Broadwell: Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GH
+         *   - Skylake
+         *
+         * If you are seeing performance stability this script can help test.
+         * It tests on 4 commits in zstd where I saw performance change.
+         *
+         *   https://gist.github.com/terrelln/9889fc06a423fd5ca6e99351564473f4
+         */
+        __asm__(".p2align 5");
+        __asm__("nop");
+        __asm__(".p2align 4");
+#endif
+        for ( ; ; ) {
+            seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset, ZSTD_p_noPrefetch);
+            size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litEnd, prefixStart, vBase, dictEnd);
+#if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
+            assert(!ZSTD_isError(oneSeqSize));
+            if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequence, prefixStart, vBase);
+#endif
+            DEBUGLOG(6, "regenerated sequence size : %u", (U32)oneSeqSize);
+            BIT_reloadDStream(&(seqState.DStream));
+            op += oneSeqSize;
+            /* gcc and clang both don't like early returns in this loop.
+             * Instead break and check for an error at the end of the loop.
+             */
+            if (UNLIKELY(ZSTD_isError(oneSeqSize))) {
+                error = oneSeqSize;
+                break;
+            }
+            if (UNLIKELY(!--nbSeq)) break;
+        }
+
+        /* check if reached exact end */
+        DEBUGLOG(5, "ZSTD_decompressSequences_body: after decode loop, remaining nbSeq : %i", nbSeq);
+        if (ZSTD_isError(error)) return error;
+        RETURN_ERROR_IF(nbSeq, corruption_detected, "");
+        RETURN_ERROR_IF(BIT_reloadDStream(&seqState.DStream) < BIT_DStream_completed, corruption_detected, "");
+        /* save reps for next block */
+        { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]); }
+    }
+
+    /* last literal segment */
+    {   size_t const lastLLSize = litEnd - litPtr;
+        RETURN_ERROR_IF(lastLLSize > (size_t)(oend-op), dstSize_tooSmall, "");
+        if (op != NULL) {
+            ZSTD_memcpy(op, litPtr, lastLLSize);
+            op += lastLLSize;
+        }
+    }
+
+    return op-ostart;
+}
+
+static size_t
+ZSTD_decompressSequences_default(ZSTD_DCtx* dctx,
+                                 void* dst, size_t maxDstSize,
+                           const void* seqStart, size_t seqSize, int nbSeq,
+                           const ZSTD_longOffset_e isLongOffset,
+                           const int frame)
+{
+    return ZSTD_decompressSequences_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
+FORCE_INLINE_TEMPLATE size_t
+ZSTD_decompressSequencesLong_body(
+                               ZSTD_DCtx* dctx,
+                               void* dst, size_t maxDstSize,
+                         const void* seqStart, size_t seqSize, int nbSeq,
+                         const ZSTD_longOffset_e isLongOffset,
+                         const int frame)
+{
+    const BYTE* ip = (const BYTE*)seqStart;
+    const BYTE* const iend = ip + seqSize;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + maxDstSize;
+    BYTE* op = ostart;
+    const BYTE* litPtr = dctx->litPtr;
+    const BYTE* const litEnd = litPtr + dctx->litSize;
+    const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart);
+    const BYTE* const dictStart = (const BYTE*) (dctx->virtualStart);
+    const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd);
+    (void)frame;
+
+    /* Regen sequences */
+    if (nbSeq) {
+#define STORED_SEQS 4
+#define STORED_SEQS_MASK (STORED_SEQS-1)
+#define ADVANCED_SEQS 4
+        seq_t sequences[STORED_SEQS];
+        int const seqAdvance = MIN(nbSeq, ADVANCED_SEQS);
+        seqState_t seqState;
+        int seqNb;
+        dctx->fseEntropy = 1;
+        { int i; for (i=0; i<ZSTD_REP_NUM; i++) seqState.prevOffset[i] = dctx->entropy.rep[i]; }
+        seqState.prefixStart = prefixStart;
+        seqState.pos = (size_t)(op-prefixStart);
+        seqState.dictEnd = dictEnd;
+        assert(dst != NULL);
+        assert(iend >= ip);
+        RETURN_ERROR_IF(
+            ERR_isError(BIT_initDStream(&seqState.DStream, ip, iend-ip)),
+            corruption_detected, "");
+        ZSTD_initFseState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
+        ZSTD_initFseState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
+        ZSTD_initFseState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
+
+        /* prepare in advance */
+        for (seqNb=0; (BIT_reloadDStream(&seqState.DStream) <= BIT_DStream_completed) && (seqNb<seqAdvance); seqNb++) {
+            sequences[seqNb] = ZSTD_decodeSequence(&seqState, isLongOffset, ZSTD_p_prefetch);
+            PREFETCH_L1(sequences[seqNb].match); PREFETCH_L1(sequences[seqNb].match + sequences[seqNb].matchLength - 1); /* note : it's safe to invoke PREFETCH() on any memory address, including invalid ones */
+        }
+        RETURN_ERROR_IF(seqNb<seqAdvance, corruption_detected, "");
+
+        /* decode and decompress */
+        for ( ; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && (seqNb<nbSeq) ; seqNb++) {
+            seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset, ZSTD_p_prefetch);
+            size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequences[(seqNb-ADVANCED_SEQS) & STORED_SEQS_MASK], &litPtr, litEnd, prefixStart, dictStart, dictEnd);
+#if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
+            assert(!ZSTD_isError(oneSeqSize));
+            if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequences[(seqNb-ADVANCED_SEQS) & STORED_SEQS_MASK], prefixStart, dictStart);
+#endif
+            if (ZSTD_isError(oneSeqSize)) return oneSeqSize;
+            PREFETCH_L1(sequence.match); PREFETCH_L1(sequence.match + sequence.matchLength - 1); /* note : it's safe to invoke PREFETCH() on any memory address, including invalid ones */
+            sequences[seqNb & STORED_SEQS_MASK] = sequence;
+            op += oneSeqSize;
+        }
+        RETURN_ERROR_IF(seqNb<nbSeq, corruption_detected, "");
+
+        /* finish queue */
+        seqNb -= seqAdvance;
+        for ( ; seqNb<nbSeq ; seqNb++) {
+            size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequences[seqNb&STORED_SEQS_MASK], &litPtr, litEnd, prefixStart, dictStart, dictEnd);
+#if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
+            assert(!ZSTD_isError(oneSeqSize));
+            if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequences[seqNb&STORED_SEQS_MASK], prefixStart, dictStart);
+#endif
+            if (ZSTD_isError(oneSeqSize)) return oneSeqSize;
+            op += oneSeqSize;
+        }
+
+        /* save reps for next block */
+        { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]); }
+    }
+
+    /* last literal segment */
+    {   size_t const lastLLSize = litEnd - litPtr;
+        RETURN_ERROR_IF(lastLLSize > (size_t)(oend-op), dstSize_tooSmall, "");
+        if (op != NULL) {
+            ZSTD_memcpy(op, litPtr, lastLLSize);
+            op += lastLLSize;
+        }
+    }
+
+    return op-ostart;
+}
+
+static size_t
+ZSTD_decompressSequencesLong_default(ZSTD_DCtx* dctx,
+                                 void* dst, size_t maxDstSize,
+                           const void* seqStart, size_t seqSize, int nbSeq,
+                           const ZSTD_longOffset_e isLongOffset,
+                           const int frame)
+{
+    return ZSTD_decompressSequencesLong_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */
+
+
+
+#if DYNAMIC_BMI2
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
+static TARGET_ATTRIBUTE("bmi2") size_t
+DONT_VECTORIZE
+ZSTD_decompressSequences_bmi2(ZSTD_DCtx* dctx,
+                                 void* dst, size_t maxDstSize,
+                           const void* seqStart, size_t seqSize, int nbSeq,
+                           const ZSTD_longOffset_e isLongOffset,
+                           const int frame)
+{
+    return ZSTD_decompressSequences_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
+static TARGET_ATTRIBUTE("bmi2") size_t
+ZSTD_decompressSequencesLong_bmi2(ZSTD_DCtx* dctx,
+                                 void* dst, size_t maxDstSize,
+                           const void* seqStart, size_t seqSize, int nbSeq,
+                           const ZSTD_longOffset_e isLongOffset,
+                           const int frame)
+{
+    return ZSTD_decompressSequencesLong_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */
+
+#endif /* DYNAMIC_BMI2 */
+
+typedef size_t (*ZSTD_decompressSequences_t)(
+                            ZSTD_DCtx* dctx,
+                            void* dst, size_t maxDstSize,
+                            const void* seqStart, size_t seqSize, int nbSeq,
+                            const ZSTD_longOffset_e isLongOffset,
+                            const int frame);
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
+static size_t
+ZSTD_decompressSequences(ZSTD_DCtx* dctx, void* dst, size_t maxDstSize,
+                   const void* seqStart, size_t seqSize, int nbSeq,
+                   const ZSTD_longOffset_e isLongOffset,
+                   const int frame)
+{
+    DEBUGLOG(5, "ZSTD_decompressSequences");
+#if DYNAMIC_BMI2
+    if (dctx->bmi2) {
+        return ZSTD_decompressSequences_bmi2(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+    }
+#endif
+  return ZSTD_decompressSequences_default(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */
+
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
+/* ZSTD_decompressSequencesLong() :
+ * decompression function triggered when a minimum share of offsets is considered "long",
+ * aka out of cache.
+ * note : "long" definition seems overloaded here, sometimes meaning "wider than bitstream register", and sometimes meaning "farther than memory cache distance".
+ * This function will try to mitigate main memory latency through the use of prefetching */
+static size_t
+ZSTD_decompressSequencesLong(ZSTD_DCtx* dctx,
+                             void* dst, size_t maxDstSize,
+                             const void* seqStart, size_t seqSize, int nbSeq,
+                             const ZSTD_longOffset_e isLongOffset,
+                             const int frame)
+{
+    DEBUGLOG(5, "ZSTD_decompressSequencesLong");
+#if DYNAMIC_BMI2
+    if (dctx->bmi2) {
+        return ZSTD_decompressSequencesLong_bmi2(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+    }
+#endif
+  return ZSTD_decompressSequencesLong_default(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */
+
+
+
+#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+/* ZSTD_getLongOffsetsShare() :
+ * condition : offTable must be valid
+ * @return : "share" of long offsets (arbitrarily defined as > (1<<23))
+ *           compared to maximum possible of (1<<OffFSELog) */
+static unsigned
+ZSTD_getLongOffsetsShare(const ZSTD_seqSymbol* offTable)
+{
+    const void* ptr = offTable;
+    U32 const tableLog = ((const ZSTD_seqSymbol_header*)ptr)[0].tableLog;
+    const ZSTD_seqSymbol* table = offTable + 1;
+    U32 const max = 1 << tableLog;
+    U32 u, total = 0;
+    DEBUGLOG(5, "ZSTD_getLongOffsetsShare: (tableLog=%u)", tableLog);
+
+    assert(max <= (1 << OffFSELog));  /* max not too large */
+    for (u=0; u<max; u++) {
+        if (table[u].nbAdditionalBits > 22) total += 1;
+    }
+
+    assert(tableLog <= OffFSELog);
+    total <<= (OffFSELog - tableLog);  /* scale to OffFSELog */
+
+    return total;
+}
+#endif
+
+size_t
+ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
+                              void* dst, size_t dstCapacity,
+                        const void* src, size_t srcSize, const int frame)
+{   /* blockType == blockCompressed */
+    const BYTE* ip = (const BYTE*)src;
+    /* isLongOffset must be true if there are long offsets.
+     * Offsets are long if they are larger than 2^STREAM_ACCUMULATOR_MIN.
+     * We don't expect that to be the case in 64-bit mode.
+     * In block mode, window size is not known, so we have to be conservative.
+     * (note: but it could be evaluated from current-lowLimit)
+     */
+    ZSTD_longOffset_e const isLongOffset = (ZSTD_longOffset_e)(MEM_32bits() && (!frame || (dctx->fParams.windowSize > (1ULL << STREAM_ACCUMULATOR_MIN))));
+    DEBUGLOG(5, "ZSTD_decompressBlock_internal (size : %u)", (U32)srcSize);
+
+    RETURN_ERROR_IF(srcSize >= ZSTD_BLOCKSIZE_MAX, srcSize_wrong, "");
+
+    /* Decode literals section */
+    {   size_t const litCSize = ZSTD_decodeLiteralsBlock(dctx, src, srcSize);
+        DEBUGLOG(5, "ZSTD_decodeLiteralsBlock : %u", (U32)litCSize);
+        if (ZSTD_isError(litCSize)) return litCSize;
+        ip += litCSize;
+        srcSize -= litCSize;
+    }
+
+    /* Build Decoding Tables */
+    {
+        /* These macros control at build-time which decompressor implementation
+         * we use. If neither is defined, we do some inspection and dispatch at
+         * runtime.
+         */
+#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+        int usePrefetchDecoder = dctx->ddictIsCold;
+#endif
+        int nbSeq;
+        size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, srcSize);
+        if (ZSTD_isError(seqHSize)) return seqHSize;
+        ip += seqHSize;
+        srcSize -= seqHSize;
+
+        RETURN_ERROR_IF(dst == NULL && nbSeq > 0, dstSize_tooSmall, "NULL not handled");
+
+#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+        if ( !usePrefetchDecoder
+          && (!frame || (dctx->fParams.windowSize > (1<<24)))
+          && (nbSeq>ADVANCED_SEQS) ) {  /* could probably use a larger nbSeq limit */
+            U32 const shareLongOffsets = ZSTD_getLongOffsetsShare(dctx->OFTptr);
+            U32 const minShare = MEM_64bits() ? 7 : 20; /* heuristic values, correspond to 2.73% and 7.81% */
+            usePrefetchDecoder = (shareLongOffsets >= minShare);
+        }
+#endif
+
+        dctx->ddictIsCold = 0;
+
+#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+        if (usePrefetchDecoder)
+#endif
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
+            return ZSTD_decompressSequencesLong(dctx, dst, dstCapacity, ip, srcSize, nbSeq, isLongOffset, frame);
+#endif
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
+        /* else */
+        return ZSTD_decompressSequences(dctx, dst, dstCapacity, ip, srcSize, nbSeq, isLongOffset, frame);
+#endif
+    }
+}
+
+
+void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst, size_t dstSize)
+{
+    if (dst != dctx->previousDstEnd && dstSize > 0) {   /* not contiguous */
+        dctx->dictEnd = dctx->previousDstEnd;
+        dctx->virtualStart = (const char*)dst - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart));
+        dctx->prefixStart = dst;
+        dctx->previousDstEnd = dst;
+    }
+}
+
+
+size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx,
+                            void* dst, size_t dstCapacity,
+                      const void* src, size_t srcSize)
+{
+    size_t dSize;
+    ZSTD_checkContinuity(dctx, dst, dstCapacity);
+    dSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, /* frame */ 0);
+    dctx->previousDstEnd = (char*)dst + dSize;
+    return dSize;
+}
diff --git a/lib/zstd/decompress/zstd_decompress_block.h b/lib/zstd/decompress/zstd_decompress_block.h
new file mode 100644
index 000000000000..e7f5f6689459
--- /dev/null
+++ b/lib/zstd/decompress/zstd_decompress_block.h
@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+
+#ifndef ZSTD_DEC_BLOCK_H
+#define ZSTD_DEC_BLOCK_H
+
+/*-*******************************************************
+ *  Dependencies
+ *********************************************************/
+#include "../common/zstd_deps.h"   /* size_t */
+#include <linux/zstd.h>    /* DCtx, and some public functions */
+#include "../common/zstd_internal.h"  /* blockProperties_t, and some public functions */
+#include "zstd_decompress_internal.h"  /* ZSTD_seqSymbol */
+
+
+/* ===   Prototypes   === */
+
+/* note: prototypes already published within `zstd.h` :
+ * ZSTD_decompressBlock()
+ */
+
+/* note: prototypes already published within `zstd_internal.h` :
+ * ZSTD_getcBlockSize()
+ * ZSTD_decodeSeqHeaders()
+ */
+
+
+/* ZSTD_decompressBlock_internal() :
+ * decompress block, starting at `src`,
+ * into destination buffer `dst`.
+ * @return : decompressed block size,
+ *           or an error code (which can be tested using ZSTD_isError())
+ */
+size_t ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
+                               void* dst, size_t dstCapacity,
+                         const void* src, size_t srcSize, const int frame);
+
+/* ZSTD_buildFSETable() :
+ * generate FSE decoding table for one symbol (ll, ml or off)
+ * this function must be called with valid parameters only
+ * (dt is large enough, normalizedCounter distribution total is a power of 2, max is within range, etc.)
+ * in which case it cannot fail.
+ * The workspace must be 4-byte aligned and at least ZSTD_BUILD_FSE_TABLE_WKSP_SIZE bytes, which is
+ * defined in zstd_decompress_internal.h.
+ * Internal use only.
+ */
+void ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
+             const short* normalizedCounter, unsigned maxSymbolValue,
+             const U32* baseValue, const U32* nbAdditionalBits,
+                   unsigned tableLog, void* wksp, size_t wkspSize,
+                   int bmi2);
+
+
+#endif /* ZSTD_DEC_BLOCK_H */
diff --git a/lib/zstd/decompress/zstd_decompress_internal.h b/lib/zstd/decompress/zstd_decompress_internal.h
new file mode 100644
index 000000000000..4b9052f68755
--- /dev/null
+++ b/lib/zstd/decompress/zstd_decompress_internal.h
@@ -0,0 +1,202 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+
+/* zstd_decompress_internal:
+ * objects and definitions shared within lib/decompress modules */
+
+ #ifndef ZSTD_DECOMPRESS_INTERNAL_H
+ #define ZSTD_DECOMPRESS_INTERNAL_H
+
+
+/*-*******************************************************
+ *  Dependencies
+ *********************************************************/
+#include "../common/mem.h"             /* BYTE, U16, U32 */
+#include "../common/zstd_internal.h"   /* ZSTD_seqSymbol */
+
+
+
+/*-*******************************************************
+ *  Constants
+ *********************************************************/
+static UNUSED_ATTR const U32 LL_base[MaxLL+1] = {
+                 0,    1,    2,     3,     4,     5,     6,      7,
+                 8,    9,   10,    11,    12,    13,    14,     15,
+                16,   18,   20,    22,    24,    28,    32,     40,
+                48,   64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000,
+                0x2000, 0x4000, 0x8000, 0x10000 };
+
+static UNUSED_ATTR const U32 OF_base[MaxOff+1] = {
+                 0,        1,       1,       5,     0xD,     0x1D,     0x3D,     0x7D,
+                 0xFD,   0x1FD,   0x3FD,   0x7FD,   0xFFD,   0x1FFD,   0x3FFD,   0x7FFD,
+                 0xFFFD, 0x1FFFD, 0x3FFFD, 0x7FFFD, 0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD,
+                 0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD, 0x1FFFFFFD, 0x3FFFFFFD, 0x7FFFFFFD };
+
+static UNUSED_ATTR const U32 OF_bits[MaxOff+1] = {
+                     0,  1,  2,  3,  4,  5,  6,  7,
+                     8,  9, 10, 11, 12, 13, 14, 15,
+                    16, 17, 18, 19, 20, 21, 22, 23,
+                    24, 25, 26, 27, 28, 29, 30, 31 };
+
+static UNUSED_ATTR const U32 ML_base[MaxML+1] = {
+                     3,  4,  5,    6,     7,     8,     9,    10,
+                    11, 12, 13,   14,    15,    16,    17,    18,
+                    19, 20, 21,   22,    23,    24,    25,    26,
+                    27, 28, 29,   30,    31,    32,    33,    34,
+                    35, 37, 39,   41,    43,    47,    51,    59,
+                    67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803,
+                    0x1003, 0x2003, 0x4003, 0x8003, 0x10003 };
+
+
+/*-*******************************************************
+ *  Decompression types
+ *********************************************************/
+ typedef struct {
+     U32 fastMode;
+     U32 tableLog;
+ } ZSTD_seqSymbol_header;
+
+ typedef struct {
+     U16  nextState;
+     BYTE nbAdditionalBits;
+     BYTE nbBits;
+     U32  baseValue;
+ } ZSTD_seqSymbol;
+
+ #define SEQSYMBOL_TABLE_SIZE(log)   (1 + (1 << (log)))
+
+#define ZSTD_BUILD_FSE_TABLE_WKSP_SIZE (sizeof(S16) * (MaxSeq + 1) + (1u << MaxFSELog) + sizeof(U64))
+#define ZSTD_BUILD_FSE_TABLE_WKSP_SIZE_U32 ((ZSTD_BUILD_FSE_TABLE_WKSP_SIZE + sizeof(U32) - 1) / sizeof(U32))
+
+typedef struct {
+    ZSTD_seqSymbol LLTable[SEQSYMBOL_TABLE_SIZE(LLFSELog)];    /* Note : Space reserved for FSE Tables */
+    ZSTD_seqSymbol OFTable[SEQSYMBOL_TABLE_SIZE(OffFSELog)];   /* is also used as temporary workspace while building hufTable during DDict creation */
+    ZSTD_seqSymbol MLTable[SEQSYMBOL_TABLE_SIZE(MLFSELog)];    /* and therefore must be at least HUF_DECOMPRESS_WORKSPACE_SIZE large */
+    HUF_DTable hufTable[HUF_DTABLE_SIZE(HufLog)];  /* can accommodate HUF_decompress4X */
+    U32 rep[ZSTD_REP_NUM];
+    U32 workspace[ZSTD_BUILD_FSE_TABLE_WKSP_SIZE_U32];
+} ZSTD_entropyDTables_t;
+
+typedef enum { ZSTDds_getFrameHeaderSize, ZSTDds_decodeFrameHeader,
+               ZSTDds_decodeBlockHeader, ZSTDds_decompressBlock,
+               ZSTDds_decompressLastBlock, ZSTDds_checkChecksum,
+               ZSTDds_decodeSkippableHeader, ZSTDds_skipFrame } ZSTD_dStage;
+
+typedef enum { zdss_init=0, zdss_loadHeader,
+               zdss_read, zdss_load, zdss_flush } ZSTD_dStreamStage;
+
+typedef enum {
+    ZSTD_use_indefinitely = -1,  /* Use the dictionary indefinitely */
+    ZSTD_dont_use = 0,           /* Do not use the dictionary (if one exists free it) */
+    ZSTD_use_once = 1            /* Use the dictionary once and set to ZSTD_dont_use */
+} ZSTD_dictUses_e;
+
+/* Hashset for storing references to multiple ZSTD_DDict within ZSTD_DCtx */
+typedef struct {
+    const ZSTD_DDict** ddictPtrTable;
+    size_t ddictPtrTableSize;
+    size_t ddictPtrCount;
+} ZSTD_DDictHashSet;
+
+struct ZSTD_DCtx_s
+{
+    const ZSTD_seqSymbol* LLTptr;
+    const ZSTD_seqSymbol* MLTptr;
+    const ZSTD_seqSymbol* OFTptr;
+    const HUF_DTable* HUFptr;
+    ZSTD_entropyDTables_t entropy;
+    U32 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];   /* space needed when building huffman tables */
+    const void* previousDstEnd;   /* detect continuity */
+    const void* prefixStart;      /* start of current segment */
+    const void* virtualStart;     /* virtual start of previous segment if it was just before current one */
+    const void* dictEnd;          /* end of previous segment */
+    size_t expected;
+    ZSTD_frameHeader fParams;
+    U64 processedCSize;
+    U64 decodedSize;
+    blockType_e bType;            /* used in ZSTD_decompressContinue(), store blockType between block header decoding and block decompression stages */
+    ZSTD_dStage stage;
+    U32 litEntropy;
+    U32 fseEntropy;
+    struct xxh64_state xxhState;
+    size_t headerSize;
+    ZSTD_format_e format;
+    ZSTD_forceIgnoreChecksum_e forceIgnoreChecksum;   /* User specified: if == 1, will ignore checksums in compressed frame. Default == 0 */
+    U32 validateChecksum;         /* if == 1, will validate checksum. Is == 1 if (fParams.checksumFlag == 1) and (forceIgnoreChecksum == 0). */
+    const BYTE* litPtr;
+    ZSTD_customMem customMem;
+    size_t litSize;
+    size_t rleSize;
+    size_t staticSize;
+    int bmi2;                     /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */
+
+    /* dictionary */
+    ZSTD_DDict* ddictLocal;
+    const ZSTD_DDict* ddict;     /* set by ZSTD_initDStream_usingDDict(), or ZSTD_DCtx_refDDict() */
+    U32 dictID;
+    int ddictIsCold;             /* if == 1 : dictionary is "new" for working context, and presumed "cold" (not in cpu cache) */
+    ZSTD_dictUses_e dictUses;
+    ZSTD_DDictHashSet* ddictSet;                    /* Hash set for multiple ddicts */
+    ZSTD_refMultipleDDicts_e refMultipleDDicts;     /* User specified: if == 1, will allow references to multiple DDicts. Default == 0 (disabled) */
+
+    /* streaming */
+    ZSTD_dStreamStage streamStage;
+    char*  inBuff;
+    size_t inBuffSize;
+    size_t inPos;
+    size_t maxWindowSize;
+    char*  outBuff;
+    size_t outBuffSize;
+    size_t outStart;
+    size_t outEnd;
+    size_t lhSize;
+    void* legacyContext;
+    U32 previousLegacyVersion;
+    U32 legacyVersion;
+    U32 hostageByte;
+    int noForwardProgress;
+    ZSTD_bufferMode_e outBufferMode;
+    ZSTD_outBuffer expectedOutBuffer;
+
+    /* workspace */
+    BYTE litBuffer[ZSTD_BLOCKSIZE_MAX + WILDCOPY_OVERLENGTH];
+    BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX];
+
+    size_t oversizedDuration;
+
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    void const* dictContentBeginForFuzzing;
+    void const* dictContentEndForFuzzing;
+#endif
+
+    /* Tracing */
+};  /* typedef'd to ZSTD_DCtx within "zstd.h" */
+
+
+/*-*******************************************************
+ *  Shared internal functions
+ *********************************************************/
+
+/*! ZSTD_loadDEntropy() :
+ *  dict : must point at beginning of a valid zstd dictionary.
+ * @return : size of dictionary header (size of magic number + dict ID + entropy tables) */
+size_t ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
+                   const void* const dict, size_t const dictSize);
+
+/*! ZSTD_checkContinuity() :
+ *  check if next `dst` follows previous position, where decompression ended.
+ *  If yes, do nothing (continue on current segment).
+ *  If not, classify previous segment as "external dictionary", and start a new segment.
+ *  This function cannot fail. */
+void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst, size_t dstSize);
+
+
+#endif /* ZSTD_DECOMPRESS_INTERNAL_H */
diff --git a/lib/zstd/decompress_sources.h b/lib/zstd/decompress_sources.h
new file mode 100644
index 000000000000..0fbec508f285
--- /dev/null
+++ b/lib/zstd/decompress_sources.h
@@ -0,0 +1,28 @@
+/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */
+/*
+ * Copyright (c) Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/*
+ * This file includes every .c file needed for decompression.
+ * It is used by lib/decompress_unzstd.c to include the decompression
+ * source into the translation-unit, so it can be used for kernel
+ * decompression.
+ */
+
+#include "common/debug.c"
+#include "common/entropy_common.c"
+#include "common/error_private.c"
+#include "common/fse_decompress.c"
+#include "common/zstd_common.c"
+#include "decompress/huf_decompress.c"
+#include "decompress/zstd_ddict.c"
+#include "decompress/zstd_decompress.c"
+#include "decompress/zstd_decompress_block.c"
+#include "zstd_decompress_module.c"
diff --git a/lib/zstd/error_private.h b/lib/zstd/error_private.h
deleted file mode 100644
index 1a60b31f706c..000000000000
--- a/lib/zstd/error_private.h
+++ /dev/null
@@ -1,53 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-/* Note : this module is expected to remain private, do not expose it */
-
-#ifndef ERROR_H_MODULE
-#define ERROR_H_MODULE
-
-/* ****************************************
-*  Dependencies
-******************************************/
-#include <linux/types.h> /* size_t */
-#include <linux/zstd.h>  /* enum list */
-
-/* ****************************************
-*  Compiler-specific
-******************************************/
-#define ERR_STATIC static __attribute__((unused))
-
-/*-****************************************
-*  Customization (error_public.h)
-******************************************/
-typedef ZSTD_ErrorCode ERR_enum;
-#define PREFIX(name) ZSTD_error_##name
-
-/*-****************************************
-*  Error codes handling
-******************************************/
-#define ERROR(name) ((size_t)-PREFIX(name))
-
-ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); }
-
-ERR_STATIC ERR_enum ERR_getErrorCode(size_t code)
-{
-	if (!ERR_isError(code))
-		return (ERR_enum)0;
-	return (ERR_enum)(0 - code);
-}
-
-#endif /* ERROR_H_MODULE */
diff --git a/lib/zstd/fse.h b/lib/zstd/fse.h
deleted file mode 100644
index 7460ab04b191..000000000000
--- a/lib/zstd/fse.h
+++ /dev/null
@@ -1,575 +0,0 @@
-/*
- * FSE : Finite State Entropy codec
- * Public Prototypes declaration
- * Copyright (C) 2013-2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-#ifndef FSE_H
-#define FSE_H
-
-/*-*****************************************
-*  Dependencies
-******************************************/
-#include <linux/types.h> /* size_t, ptrdiff_t */
-
-/*-*****************************************
-*  FSE_PUBLIC_API : control library symbols visibility
-******************************************/
-#define FSE_PUBLIC_API
-
-/*------   Version   ------*/
-#define FSE_VERSION_MAJOR 0
-#define FSE_VERSION_MINOR 9
-#define FSE_VERSION_RELEASE 0
-
-#define FSE_LIB_VERSION FSE_VERSION_MAJOR.FSE_VERSION_MINOR.FSE_VERSION_RELEASE
-#define FSE_QUOTE(str) #str
-#define FSE_EXPAND_AND_QUOTE(str) FSE_QUOTE(str)
-#define FSE_VERSION_STRING FSE_EXPAND_AND_QUOTE(FSE_LIB_VERSION)
-
-#define FSE_VERSION_NUMBER (FSE_VERSION_MAJOR * 100 * 100 + FSE_VERSION_MINOR * 100 + FSE_VERSION_RELEASE)
-FSE_PUBLIC_API unsigned FSE_versionNumber(void); /**< library version number; to be used when checking dll version */
-
-/*-*****************************************
-*  Tool functions
-******************************************/
-FSE_PUBLIC_API size_t FSE_compressBound(size_t size); /* maximum compressed size */
-
-/* Error Management */
-FSE_PUBLIC_API unsigned FSE_isError(size_t code); /* tells if a return value is an error code */
-
-/*-*****************************************
-*  FSE detailed API
-******************************************/
-/*!
-FSE_compress() does the following:
-1. count symbol occurrence from source[] into table count[]
-2. normalize counters so that sum(count[]) == Power_of_2 (2^tableLog)
-3. save normalized counters to memory buffer using writeNCount()
-4. build encoding table 'CTable' from normalized counters
-5. encode the data stream using encoding table 'CTable'
-
-FSE_decompress() does the following:
-1. read normalized counters with readNCount()
-2. build decoding table 'DTable' from normalized counters
-3. decode the data stream using decoding table 'DTable'
-
-The following API allows targeting specific sub-functions for advanced tasks.
-For example, it's possible to compress several blocks using the same 'CTable',
-or to save and provide normalized distribution using external method.
-*/
-
-/* *** COMPRESSION *** */
-/*! FSE_optimalTableLog():
-	dynamically downsize 'tableLog' when conditions are met.
-	It saves CPU time, by using smaller tables, while preserving or even improving compression ratio.
-	@return : recommended tableLog (necessarily <= 'maxTableLog') */
-FSE_PUBLIC_API unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
-
-/*! FSE_normalizeCount():
-	normalize counts so that sum(count[]) == Power_of_2 (2^tableLog)
-	'normalizedCounter' is a table of short, of minimum size (maxSymbolValue+1).
-	@return : tableLog,
-			  or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_normalizeCount(short *normalizedCounter, unsigned tableLog, const unsigned *count, size_t srcSize, unsigned maxSymbolValue);
-
-/*! FSE_NCountWriteBound():
-	Provides the maximum possible size of an FSE normalized table, given 'maxSymbolValue' and 'tableLog'.
-	Typically useful for allocation purpose. */
-FSE_PUBLIC_API size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog);
-
-/*! FSE_writeNCount():
-	Compactly save 'normalizedCounter' into 'buffer'.
-	@return : size of the compressed table,
-			  or an errorCode, which can be tested using FSE_isError(). */
-FSE_PUBLIC_API size_t FSE_writeNCount(void *buffer, size_t bufferSize, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);
-
-/*! Constructor and Destructor of FSE_CTable.
-	Note that FSE_CTable size depends on 'tableLog' and 'maxSymbolValue' */
-typedef unsigned FSE_CTable; /* don't allocate that. It's only meant to be more restrictive than void* */
-
-/*! FSE_compress_usingCTable():
-	Compress `src` using `ct` into `dst` which must be already allocated.
-	@return : size of compressed data (<= `dstCapacity`),
-			  or 0 if compressed data could not fit into `dst`,
-			  or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_compress_usingCTable(void *dst, size_t dstCapacity, const void *src, size_t srcSize, const FSE_CTable *ct);
-
-/*!
-Tutorial :
-----------
-The first step is to count all symbols. FSE_count() does this job very fast.
-Result will be saved into 'count', a table of unsigned int, which must be already allocated, and have 'maxSymbolValuePtr[0]+1' cells.
-'src' is a table of bytes of size 'srcSize'. All values within 'src' MUST be <= maxSymbolValuePtr[0]
-maxSymbolValuePtr[0] will be updated, with its real value (necessarily <= original value)
-FSE_count() will return the number of occurrence of the most frequent symbol.
-This can be used to know if there is a single symbol within 'src', and to quickly evaluate its compressibility.
-If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
-
-The next step is to normalize the frequencies.
-FSE_normalizeCount() will ensure that sum of frequencies is == 2 ^'tableLog'.
-It also guarantees a minimum of 1 to any Symbol with frequency >= 1.
-You can use 'tableLog'==0 to mean "use default tableLog value".
-If you are unsure of which tableLog value to use, you can ask FSE_optimalTableLog(),
-which will provide the optimal valid tableLog given sourceSize, maxSymbolValue, and a user-defined maximum (0 means "default").
-
-The result of FSE_normalizeCount() will be saved into a table,
-called 'normalizedCounter', which is a table of signed short.
-'normalizedCounter' must be already allocated, and have at least 'maxSymbolValue+1' cells.
-The return value is tableLog if everything proceeded as expected.
-It is 0 if there is a single symbol within distribution.
-If there is an error (ex: invalid tableLog value), the function will return an ErrorCode (which can be tested using FSE_isError()).
-
-'normalizedCounter' can be saved in a compact manner to a memory area using FSE_writeNCount().
-'buffer' must be already allocated.
-For guaranteed success, buffer size must be at least FSE_headerBound().
-The result of the function is the number of bytes written into 'buffer'.
-If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError(); ex : buffer size too small).
-
-'normalizedCounter' can then be used to create the compression table 'CTable'.
-The space required by 'CTable' must be already allocated, using FSE_createCTable().
-You can then use FSE_buildCTable() to fill 'CTable'.
-If there is an error, both functions will return an ErrorCode (which can be tested using FSE_isError()).
-
-'CTable' can then be used to compress 'src', with FSE_compress_usingCTable().
-Similar to FSE_count(), the convention is that 'src' is assumed to be a table of char of size 'srcSize'
-The function returns the size of compressed data (without header), necessarily <= `dstCapacity`.
-If it returns '0', compressed data could not fit into 'dst'.
-If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
-*/
-
-/* *** DECOMPRESSION *** */
-
-/*! FSE_readNCount():
-	Read compactly saved 'normalizedCounter' from 'rBuffer'.
-	@return : size read from 'rBuffer',
-			  or an errorCode, which can be tested using FSE_isError().
-			  maxSymbolValuePtr[0] and tableLogPtr[0] will also be updated with their respective values */
-FSE_PUBLIC_API size_t FSE_readNCount(short *normalizedCounter, unsigned *maxSymbolValuePtr, unsigned *tableLogPtr, const void *rBuffer, size_t rBuffSize);
-
-/*! Constructor and Destructor of FSE_DTable.
-	Note that its size depends on 'tableLog' */
-typedef unsigned FSE_DTable; /* don't allocate that. It's just a way to be more restrictive than void* */
-
-/*! FSE_buildDTable():
-	Builds 'dt', which must be already allocated, using FSE_createDTable().
-	return : 0, or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_buildDTable_wksp(FSE_DTable *dt, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workspace, size_t workspaceSize);
-
-/*! FSE_decompress_usingDTable():
-	Decompress compressed source `cSrc` of size `cSrcSize` using `dt`
-	into `dst` which must be already allocated.
-	@return : size of regenerated data (necessarily <= `dstCapacity`),
-			  or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_decompress_usingDTable(void *dst, size_t dstCapacity, const void *cSrc, size_t cSrcSize, const FSE_DTable *dt);
-
-/*!
-Tutorial :
-----------
-(Note : these functions only decompress FSE-compressed blocks.
- If block is uncompressed, use memcpy() instead
- If block is a single repeated byte, use memset() instead )
-
-The first step is to obtain the normalized frequencies of symbols.
-This can be performed by FSE_readNCount() if it was saved using FSE_writeNCount().
-'normalizedCounter' must be already allocated, and have at least 'maxSymbolValuePtr[0]+1' cells of signed short.
-In practice, that means it's necessary to know 'maxSymbolValue' beforehand,
-or size the table to handle worst case situations (typically 256).
-FSE_readNCount() will provide 'tableLog' and 'maxSymbolValue'.
-The result of FSE_readNCount() is the number of bytes read from 'rBuffer'.
-Note that 'rBufferSize' must be at least 4 bytes, even if useful information is less than that.
-If there is an error, the function will return an error code, which can be tested using FSE_isError().
-
-The next step is to build the decompression tables 'FSE_DTable' from 'normalizedCounter'.
-This is performed by the function FSE_buildDTable().
-The space required by 'FSE_DTable' must be already allocated using FSE_createDTable().
-If there is an error, the function will return an error code, which can be tested using FSE_isError().
-
-`FSE_DTable` can then be used to decompress `cSrc`, with FSE_decompress_usingDTable().
-`cSrcSize` must be strictly correct, otherwise decompression will fail.
-FSE_decompress_usingDTable() result will tell how many bytes were regenerated (<=`dstCapacity`).
-If there is an error, the function will return an error code, which can be tested using FSE_isError(). (ex: dst buffer too small)
-*/
-
-/* *** Dependency *** */
-#include "bitstream.h"
-
-/* *****************************************
-*  Static allocation
-*******************************************/
-/* FSE buffer bounds */
-#define FSE_NCOUNTBOUND 512
-#define FSE_BLOCKBOUND(size) (size + (size >> 7))
-#define FSE_COMPRESSBOUND(size) (FSE_NCOUNTBOUND + FSE_BLOCKBOUND(size)) /* Macro version, useful for static allocation */
-
-/* It is possible to statically allocate FSE CTable/DTable as a table of FSE_CTable/FSE_DTable using below macros */
-#define FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) (1 + (1 << (maxTableLog - 1)) + ((maxSymbolValue + 1) * 2))
-#define FSE_DTABLE_SIZE_U32(maxTableLog) (1 + (1 << maxTableLog))
-
-/* *****************************************
-*  FSE advanced API
-*******************************************/
-/* FSE_count_wksp() :
- * Same as FSE_count(), but using an externally provided scratch buffer.
- * `workSpace` size must be table of >= `1024` unsigned
- */
-size_t FSE_count_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *source, size_t sourceSize, unsigned *workSpace);
-
-/* FSE_countFast_wksp() :
- * Same as FSE_countFast(), but using an externally provided scratch buffer.
- * `workSpace` must be a table of minimum `1024` unsigned
- */
-size_t FSE_countFast_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *src, size_t srcSize, unsigned *workSpace);
-
-/*! FSE_count_simple
- * Same as FSE_countFast(), but does not use any additional memory (not even on stack).
- * This function is unsafe, and will segfault if any value within `src` is `> *maxSymbolValuePtr` (presuming it's also the size of `count`).
-*/
-size_t FSE_count_simple(unsigned *count, unsigned *maxSymbolValuePtr, const void *src, size_t srcSize);
-
-unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus);
-/**< same as FSE_optimalTableLog(), which used `minus==2` */
-
-size_t FSE_buildCTable_raw(FSE_CTable *ct, unsigned nbBits);
-/**< build a fake FSE_CTable, designed for a flat distribution, where each symbol uses nbBits */
-
-size_t FSE_buildCTable_rle(FSE_CTable *ct, unsigned char symbolValue);
-/**< build a fake FSE_CTable, designed to compress always the same symbolValue */
-
-/* FSE_buildCTable_wksp() :
- * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
- * `wkspSize` must be >= `(1<<tableLog)`.
- */
-size_t FSE_buildCTable_wksp(FSE_CTable *ct, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workSpace, size_t wkspSize);
-
-size_t FSE_buildDTable_raw(FSE_DTable *dt, unsigned nbBits);
-/**< build a fake FSE_DTable, designed to read a flat distribution where each symbol uses nbBits */
-
-size_t FSE_buildDTable_rle(FSE_DTable *dt, unsigned char symbolValue);
-/**< build a fake FSE_DTable, designed to always generate the same symbolValue */
-
-size_t FSE_decompress_wksp(void *dst, size_t dstCapacity, const void *cSrc, size_t cSrcSize, unsigned maxLog, void *workspace, size_t workspaceSize);
-/**< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DTABLE_SIZE_U32(maxLog)` */
-
-/* *****************************************
-*  FSE symbol compression API
-*******************************************/
-/*!
-   This API consists of small unitary functions, which highly benefit from being inlined.
-   Hence their body are included in next section.
-*/
-typedef struct {
-	ptrdiff_t value;
-	const void *stateTable;
-	const void *symbolTT;
-	unsigned stateLog;
-} FSE_CState_t;
-
-static void FSE_initCState(FSE_CState_t *CStatePtr, const FSE_CTable *ct);
-
-static void FSE_encodeSymbol(BIT_CStream_t *bitC, FSE_CState_t *CStatePtr, unsigned symbol);
-
-static void FSE_flushCState(BIT_CStream_t *bitC, const FSE_CState_t *CStatePtr);
-
-/**<
-These functions are inner components of FSE_compress_usingCTable().
-They allow the creation of custom streams, mixing multiple tables and bit sources.
-
-A key property to keep in mind is that encoding and decoding are done **in reverse direction**.
-So the first symbol you will encode is the last you will decode, like a LIFO stack.
-
-You will need a few variables to track your CStream. They are :
-
-FSE_CTable    ct;         // Provided by FSE_buildCTable()
-BIT_CStream_t bitStream;  // bitStream tracking structure
-FSE_CState_t  state;      // State tracking structure (can have several)
-
-
-The first thing to do is to init bitStream and state.
-	size_t errorCode = BIT_initCStream(&bitStream, dstBuffer, maxDstSize);
-	FSE_initCState(&state, ct);
-
-Note that BIT_initCStream() can produce an error code, so its result should be tested, using FSE_isError();
-You can then encode your input data, byte after byte.
-FSE_encodeSymbol() outputs a maximum of 'tableLog' bits at a time.
-Remember decoding will be done in reverse direction.
-	FSE_encodeByte(&bitStream, &state, symbol);
-
-At any time, you can also add any bit sequence.
-Note : maximum allowed nbBits is 25, for compatibility with 32-bits decoders
-	BIT_addBits(&bitStream, bitField, nbBits);
-
-The above methods don't commit data to memory, they just store it into local register, for speed.
-Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
-Writing data to memory is a manual operation, performed by the flushBits function.
-	BIT_flushBits(&bitStream);
-
-Your last FSE encoding operation shall be to flush your last state value(s).
-	FSE_flushState(&bitStream, &state);
-
-Finally, you must close the bitStream.
-The function returns the size of CStream in bytes.
-If data couldn't fit into dstBuffer, it will return a 0 ( == not compressible)
-If there is an error, it returns an errorCode (which can be tested using FSE_isError()).
-	size_t size = BIT_closeCStream(&bitStream);
-*/
-
-/* *****************************************
-*  FSE symbol decompression API
-*******************************************/
-typedef struct {
-	size_t state;
-	const void *table; /* precise table may vary, depending on U16 */
-} FSE_DState_t;
-
-static void FSE_initDState(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD, const FSE_DTable *dt);
-
-static unsigned char FSE_decodeSymbol(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD);
-
-static unsigned FSE_endOfDState(const FSE_DState_t *DStatePtr);
-
-/**<
-Let's now decompose FSE_decompress_usingDTable() into its unitary components.
-You will decode FSE-encoded symbols from the bitStream,
-and also any other bitFields you put in, **in reverse order**.
-
-You will need a few variables to track your bitStream. They are :
-
-BIT_DStream_t DStream;    // Stream context
-FSE_DState_t  DState;     // State context. Multiple ones are possible
-FSE_DTable*   DTablePtr;  // Decoding table, provided by FSE_buildDTable()
-
-The first thing to do is to init the bitStream.
-	errorCode = BIT_initDStream(&DStream, srcBuffer, srcSize);
-
-You should then retrieve your initial state(s)
-(in reverse flushing order if you have several ones) :
-	errorCode = FSE_initDState(&DState, &DStream, DTablePtr);
-
-You can then decode your data, symbol after symbol.
-For information the maximum number of bits read by FSE_decodeSymbol() is 'tableLog'.
-Keep in mind that symbols are decoded in reverse order, like a LIFO stack (last in, first out).
-	unsigned char symbol = FSE_decodeSymbol(&DState, &DStream);
-
-You can retrieve any bitfield you eventually stored into the bitStream (in reverse order)
-Note : maximum allowed nbBits is 25, for 32-bits compatibility
-	size_t bitField = BIT_readBits(&DStream, nbBits);
-
-All above operations only read from local register (which size depends on size_t).
-Refueling the register from memory is manually performed by the reload method.
-	endSignal = FSE_reloadDStream(&DStream);
-
-BIT_reloadDStream() result tells if there is still some more data to read from DStream.
-BIT_DStream_unfinished : there is still some data left into the DStream.
-BIT_DStream_endOfBuffer : Dstream reached end of buffer. Its container may no longer be completely filled.
-BIT_DStream_completed : Dstream reached its exact end, corresponding in general to decompression completed.
-BIT_DStream_tooFar : Dstream went too far. Decompression result is corrupted.
-
-When reaching end of buffer (BIT_DStream_endOfBuffer), progress slowly, notably if you decode multiple symbols per loop,
-to properly detect the exact end of stream.
-After each decoded symbol, check if DStream is fully consumed using this simple test :
-	BIT_reloadDStream(&DStream) >= BIT_DStream_completed
-
-When it's done, verify decompression is fully completed, by checking both DStream and the relevant states.
-Checking if DStream has reached its end is performed by :
-	BIT_endOfDStream(&DStream);
-Check also the states. There might be some symbols left there, if some high probability ones (>50%) are possible.
-	FSE_endOfDState(&DState);
-*/
-
-/* *****************************************
-*  FSE unsafe API
-*******************************************/
-static unsigned char FSE_decodeSymbolFast(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD);
-/* faster, but works only if nbBits is always >= 1 (otherwise, result will be corrupted) */
-
-/* *****************************************
-*  Implementation of inlined functions
-*******************************************/
-typedef struct {
-	int deltaFindState;
-	U32 deltaNbBits;
-} FSE_symbolCompressionTransform; /* total 8 bytes */
-
-ZSTD_STATIC void FSE_initCState(FSE_CState_t *statePtr, const FSE_CTable *ct)
-{
-	const void *ptr = ct;
-	const U16 *u16ptr = (const U16 *)ptr;
-	const U32 tableLog = ZSTD_read16(ptr);
-	statePtr->value = (ptrdiff_t)1 << tableLog;
-	statePtr->stateTable = u16ptr + 2;
-	statePtr->symbolTT = ((const U32 *)ct + 1 + (tableLog ? (1 << (tableLog - 1)) : 1));
-	statePtr->stateLog = tableLog;
-}
-
-/*! FSE_initCState2() :
-*   Same as FSE_initCState(), but the first symbol to include (which will be the last to be read)
-*   uses the smallest state value possible, saving the cost of this symbol */
-ZSTD_STATIC void FSE_initCState2(FSE_CState_t *statePtr, const FSE_CTable *ct, U32 symbol)
-{
-	FSE_initCState(statePtr, ct);
-	{
-		const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform *)(statePtr->symbolTT))[symbol];
-		const U16 *stateTable = (const U16 *)(statePtr->stateTable);
-		U32 nbBitsOut = (U32)((symbolTT.deltaNbBits + (1 << 15)) >> 16);
-		statePtr->value = (nbBitsOut << 16) - symbolTT.deltaNbBits;
-		statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
-	}
-}
-
-ZSTD_STATIC void FSE_encodeSymbol(BIT_CStream_t *bitC, FSE_CState_t *statePtr, U32 symbol)
-{
-	const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform *)(statePtr->symbolTT))[symbol];
-	const U16 *const stateTable = (const U16 *)(statePtr->stateTable);
-	U32 nbBitsOut = (U32)((statePtr->value + symbolTT.deltaNbBits) >> 16);
-	BIT_addBits(bitC, statePtr->value, nbBitsOut);
-	statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
-}
-
-ZSTD_STATIC void FSE_flushCState(BIT_CStream_t *bitC, const FSE_CState_t *statePtr)
-{
-	BIT_addBits(bitC, statePtr->value, statePtr->stateLog);
-	BIT_flushBits(bitC);
-}
-
-/* ======    Decompression    ====== */
-
-typedef struct {
-	U16 tableLog;
-	U16 fastMode;
-} FSE_DTableHeader; /* sizeof U32 */
-
-typedef struct {
-	unsigned short newState;
-	unsigned char symbol;
-	unsigned char nbBits;
-} FSE_decode_t; /* size == U32 */
-
-ZSTD_STATIC void FSE_initDState(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD, const FSE_DTable *dt)
-{
-	const void *ptr = dt;
-	const FSE_DTableHeader *const DTableH = (const FSE_DTableHeader *)ptr;
-	DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
-	BIT_reloadDStream(bitD);
-	DStatePtr->table = dt + 1;
-}
-
-ZSTD_STATIC BYTE FSE_peekSymbol(const FSE_DState_t *DStatePtr)
-{
-	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
-	return DInfo.symbol;
-}
-
-ZSTD_STATIC void FSE_updateState(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD)
-{
-	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
-	U32 const nbBits = DInfo.nbBits;
-	size_t const lowBits = BIT_readBits(bitD, nbBits);
-	DStatePtr->state = DInfo.newState + lowBits;
-}
-
-ZSTD_STATIC BYTE FSE_decodeSymbol(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD)
-{
-	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
-	U32 const nbBits = DInfo.nbBits;
-	BYTE const symbol = DInfo.symbol;
-	size_t const lowBits = BIT_readBits(bitD, nbBits);
-
-	DStatePtr->state = DInfo.newState + lowBits;
-	return symbol;
-}
-
-/*! FSE_decodeSymbolFast() :
-	unsafe, only works if no symbol has a probability > 50% */
-ZSTD_STATIC BYTE FSE_decodeSymbolFast(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD)
-{
-	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
-	U32 const nbBits = DInfo.nbBits;
-	BYTE const symbol = DInfo.symbol;
-	size_t const lowBits = BIT_readBitsFast(bitD, nbBits);
-
-	DStatePtr->state = DInfo.newState + lowBits;
-	return symbol;
-}
-
-ZSTD_STATIC unsigned FSE_endOfDState(const FSE_DState_t *DStatePtr) { return DStatePtr->state == 0; }
-
-/* **************************************************************
-*  Tuning parameters
-****************************************************************/
-/*!MEMORY_USAGE :
-*  Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB; etc.)
-*  Increasing memory usage improves compression ratio
-*  Reduced memory usage can improve speed, due to cache effect
-*  Recommended max value is 14, for 16KB, which nicely fits into Intel x86 L1 cache */
-#ifndef FSE_MAX_MEMORY_USAGE
-#define FSE_MAX_MEMORY_USAGE 14
-#endif
-#ifndef FSE_DEFAULT_MEMORY_USAGE
-#define FSE_DEFAULT_MEMORY_USAGE 13
-#endif
-
-/*!FSE_MAX_SYMBOL_VALUE :
-*  Maximum symbol value authorized.
-*  Required for proper stack allocation */
-#ifndef FSE_MAX_SYMBOL_VALUE
-#define FSE_MAX_SYMBOL_VALUE 255
-#endif
-
-/* **************************************************************
-*  template functions type & suffix
-****************************************************************/
-#define FSE_FUNCTION_TYPE BYTE
-#define FSE_FUNCTION_EXTENSION
-#define FSE_DECODE_TYPE FSE_decode_t
-
-/* ***************************************************************
-*  Constants
-*****************************************************************/
-#define FSE_MAX_TABLELOG (FSE_MAX_MEMORY_USAGE - 2)
-#define FSE_MAX_TABLESIZE (1U << FSE_MAX_TABLELOG)
-#define FSE_MAXTABLESIZE_MASK (FSE_MAX_TABLESIZE - 1)
-#define FSE_DEFAULT_TABLELOG (FSE_DEFAULT_MEMORY_USAGE - 2)
-#define FSE_MIN_TABLELOG 5
-
-#define FSE_TABLELOG_ABSOLUTE_MAX 15
-#if FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX
-#error "FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX is not supported"
-#endif
-
-#define FSE_TABLESTEP(tableSize) ((tableSize >> 1) + (tableSize >> 3) + 3)
-
-#endif /* FSE_H */
diff --git a/lib/zstd/huf.h b/lib/zstd/huf.h
deleted file mode 100644
index 2143da28d952..000000000000
--- a/lib/zstd/huf.h
+++ /dev/null
@@ -1,212 +0,0 @@
-/*
- * Huffman coder, part of New Generation Entropy library
- * header file
- * Copyright (C) 2013-2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-#ifndef HUF_H_298734234
-#define HUF_H_298734234
-
-/* *** Dependencies *** */
-#include <linux/types.h> /* size_t */
-
-/* ***   Tool functions *** */
-#define HUF_BLOCKSIZE_MAX (128 * 1024) /**< maximum input size for a single block compressed with HUF_compress */
-size_t HUF_compressBound(size_t size); /**< maximum compressed size (worst case) */
-
-/* Error Management */
-unsigned HUF_isError(size_t code); /**< tells if a return value is an error code */
-
-/* ***   Advanced function   *** */
-
-/** HUF_compress4X_wksp() :
-*   Same as HUF_compress2(), but uses externally allocated `workSpace`, which must be a table of >= 1024 unsigned */
-size_t HUF_compress4X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
-			   size_t wkspSize); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
-
-/* *** Dependencies *** */
-#include "mem.h" /* U32 */
-
-/* *** Constants *** */
-#define HUF_TABLELOG_MAX 12     /* max configured tableLog (for static allocation); can be modified up to HUF_ABSOLUTEMAX_TABLELOG */
-#define HUF_TABLELOG_DEFAULT 11 /* tableLog by default, when not specified */
-#define HUF_SYMBOLVALUE_MAX 255
-
-#define HUF_TABLELOG_ABSOLUTEMAX 15 /* absolute limit of HUF_MAX_TABLELOG. Beyond that value, code does not work */
-#if (HUF_TABLELOG_MAX > HUF_TABLELOG_ABSOLUTEMAX)
-#error "HUF_TABLELOG_MAX is too large !"
-#endif
-
-/* ****************************************
-*  Static allocation
-******************************************/
-/* HUF buffer bounds */
-#define HUF_CTABLEBOUND 129
-#define HUF_BLOCKBOUND(size) (size + (size >> 8) + 8)			 /* only true if incompressible pre-filtered with fast heuristic */
-#define HUF_COMPRESSBOUND(size) (HUF_CTABLEBOUND + HUF_BLOCKBOUND(size)) /* Macro version, useful for static allocation */
-
-/* static allocation of HUF's Compression Table */
-#define HUF_CREATE_STATIC_CTABLE(name, maxSymbolValue) \
-	U32 name##hb[maxSymbolValue + 1];              \
-	void *name##hv = &(name##hb);                  \
-	HUF_CElt *name = (HUF_CElt *)(name##hv) /* no final ; */
-
-/* static allocation of HUF's DTable */
-typedef U32 HUF_DTable;
-#define HUF_DTABLE_SIZE(maxTableLog) (1 + (1 << (maxTableLog)))
-#define HUF_CREATE_STATIC_DTABLEX2(DTable, maxTableLog) HUF_DTable DTable[HUF_DTABLE_SIZE((maxTableLog)-1)] = {((U32)((maxTableLog)-1) * 0x01000001)}
-#define HUF_CREATE_STATIC_DTABLEX4(DTable, maxTableLog) HUF_DTable DTable[HUF_DTABLE_SIZE(maxTableLog)] = {((U32)(maxTableLog)*0x01000001)}
-
-/* The workspace must have alignment at least 4 and be at least this large */
-#define HUF_COMPRESS_WORKSPACE_SIZE (6 << 10)
-#define HUF_COMPRESS_WORKSPACE_SIZE_U32 (HUF_COMPRESS_WORKSPACE_SIZE / sizeof(U32))
-
-/* The workspace must have alignment at least 4 and be at least this large */
-#define HUF_DECOMPRESS_WORKSPACE_SIZE (3 << 10)
-#define HUF_DECOMPRESS_WORKSPACE_SIZE_U32 (HUF_DECOMPRESS_WORKSPACE_SIZE / sizeof(U32))
-
-/* ****************************************
-*  Advanced decompression functions
-******************************************/
-size_t HUF_decompress4X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize); /**< decodes RLE and uncompressed */
-size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				size_t workspaceSize);							       /**< considers RLE and uncompressed as errors */
-size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				   size_t workspaceSize); /**< single-symbol decoder */
-size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				   size_t workspaceSize); /**< double-symbols decoder */
-
-/* ****************************************
-*  HUF detailed API
-******************************************/
-/*!
-HUF_compress() does the following:
-1. count symbol occurrence from source[] into table count[] using FSE_count()
-2. (optional) refine tableLog using HUF_optimalTableLog()
-3. build Huffman table from count using HUF_buildCTable()
-4. save Huffman table to memory buffer using HUF_writeCTable_wksp()
-5. encode the data stream using HUF_compress4X_usingCTable()
-
-The following API allows targeting specific sub-functions for advanced tasks.
-For example, it's possible to compress several blocks using the same 'CTable',
-or to save and regenerate 'CTable' using external methods.
-*/
-/* FSE_count() : find it within "fse.h" */
-unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
-typedef struct HUF_CElt_s HUF_CElt; /* incomplete type */
-size_t HUF_writeCTable_wksp(void *dst, size_t maxDstSize, const HUF_CElt *CTable, unsigned maxSymbolValue, unsigned huffLog, void *workspace, size_t workspaceSize);
-size_t HUF_compress4X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable);
-
-typedef enum {
-	HUF_repeat_none,  /**< Cannot use the previous table */
-	HUF_repeat_check, /**< Can use the previous table but it must be checked. Note : The previous table must have been constructed by HUF_compress{1,
-			     4}X_repeat */
-	HUF_repeat_valid  /**< Can use the previous table and it is asumed to be valid */
-} HUF_repeat;
-/** HUF_compress4X_repeat() :
-*   Same as HUF_compress4X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
-*   If it uses hufTable it does not modify hufTable or repeat.
-*   If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
-*   If preferRepeat then the old table will always be used if valid. */
-size_t HUF_compress4X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
-			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat,
-			     int preferRepeat); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
-
-/** HUF_buildCTable_wksp() :
- *  Same as HUF_buildCTable(), but using externally allocated scratch buffer.
- *  `workSpace` must be aligned on 4-bytes boundaries, and be at least as large as a table of 1024 unsigned.
- */
-size_t HUF_buildCTable_wksp(HUF_CElt *tree, const U32 *count, U32 maxSymbolValue, U32 maxNbBits, void *workSpace, size_t wkspSize);
-
-/*! HUF_readStats() :
-	Read compact Huffman tree, saved by HUF_writeCTable().
-	`huffWeight` is destination buffer.
-	@return : size read from `src` , or an error Code .
-	Note : Needed by HUF_readCTable() and HUF_readDTableXn() . */
-size_t HUF_readStats_wksp(BYTE *huffWeight, size_t hwSize, U32 *rankStats, U32 *nbSymbolsPtr, U32 *tableLogPtr, const void *src, size_t srcSize,
-			  void *workspace, size_t workspaceSize);
-
-/** HUF_readCTable() :
-*   Loading a CTable saved with HUF_writeCTable() */
-size_t HUF_readCTable_wksp(HUF_CElt *CTable, unsigned maxSymbolValue, const void *src, size_t srcSize, void *workspace, size_t workspaceSize);
-
-/*
-HUF_decompress() does the following:
-1. select the decompression algorithm (X2, X4) based on pre-computed heuristics
-2. build Huffman table from save, using HUF_readDTableXn()
-3. decode 1 or 4 segments in parallel using HUF_decompressSXn_usingDTable
-*/
-
-/** HUF_selectDecoder() :
-*   Tells which decoder is likely to decode faster,
-*   based on a set of pre-determined metrics.
-*   @return : 0==HUF_decompress4X2, 1==HUF_decompress4X4 .
-*   Assumption : 0 < cSrcSize < dstSize <= 128 KB */
-U32 HUF_selectDecoder(size_t dstSize, size_t cSrcSize);
-
-size_t HUF_readDTableX2_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize);
-size_t HUF_readDTableX4_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize);
-
-size_t HUF_decompress4X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-size_t HUF_decompress4X2_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-size_t HUF_decompress4X4_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-
-/* single stream variants */
-
-size_t HUF_compress1X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
-			   size_t wkspSize); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
-size_t HUF_compress1X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable);
-/** HUF_compress1X_repeat() :
-*   Same as HUF_compress1X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
-*   If it uses hufTable it does not modify hufTable or repeat.
-*   If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
-*   If preferRepeat then the old table will always be used if valid. */
-size_t HUF_compress1X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
-			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat,
-			     int preferRepeat); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
-
-size_t HUF_decompress1X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize);
-size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				   size_t workspaceSize); /**< single-symbol decoder */
-size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				   size_t workspaceSize); /**< double-symbols decoder */
-
-size_t HUF_decompress1X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize,
-				    const HUF_DTable *DTable); /**< automatic selection of sing or double symbol decoder, based on DTable */
-size_t HUF_decompress1X2_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-size_t HUF_decompress1X4_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-
-#endif /* HUF_H_298734234 */
diff --git a/lib/zstd/mem.h b/lib/zstd/mem.h
deleted file mode 100644
index 3a0f34c8706c..000000000000
--- a/lib/zstd/mem.h
+++ /dev/null
@@ -1,151 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-#ifndef MEM_H_MODULE
-#define MEM_H_MODULE
-
-/*-****************************************
-*  Dependencies
-******************************************/
-#include <asm/unaligned.h>
-#include <linux/string.h> /* memcpy */
-#include <linux/types.h>  /* size_t, ptrdiff_t */
-
-/*-****************************************
-*  Compiler specifics
-******************************************/
-#define ZSTD_STATIC static __inline __attribute__((unused))
-
-/*-**************************************************************
-*  Basic Types
-*****************************************************************/
-typedef uint8_t BYTE;
-typedef uint16_t U16;
-typedef int16_t S16;
-typedef uint32_t U32;
-typedef int32_t S32;
-typedef uint64_t U64;
-typedef int64_t S64;
-typedef ptrdiff_t iPtrDiff;
-typedef uintptr_t uPtrDiff;
-
-/*-**************************************************************
-*  Memory I/O
-*****************************************************************/
-ZSTD_STATIC unsigned ZSTD_32bits(void) { return sizeof(size_t) == 4; }
-ZSTD_STATIC unsigned ZSTD_64bits(void) { return sizeof(size_t) == 8; }
-
-#if defined(__LITTLE_ENDIAN)
-#define ZSTD_LITTLE_ENDIAN 1
-#else
-#define ZSTD_LITTLE_ENDIAN 0
-#endif
-
-ZSTD_STATIC unsigned ZSTD_isLittleEndian(void) { return ZSTD_LITTLE_ENDIAN; }
-
-ZSTD_STATIC U16 ZSTD_read16(const void *memPtr) { return get_unaligned((const U16 *)memPtr); }
-
-ZSTD_STATIC U32 ZSTD_read32(const void *memPtr) { return get_unaligned((const U32 *)memPtr); }
-
-ZSTD_STATIC U64 ZSTD_read64(const void *memPtr) { return get_unaligned((const U64 *)memPtr); }
-
-ZSTD_STATIC size_t ZSTD_readST(const void *memPtr) { return get_unaligned((const size_t *)memPtr); }
-
-ZSTD_STATIC void ZSTD_write16(void *memPtr, U16 value) { put_unaligned(value, (U16 *)memPtr); }
-
-ZSTD_STATIC void ZSTD_write32(void *memPtr, U32 value) { put_unaligned(value, (U32 *)memPtr); }
-
-ZSTD_STATIC void ZSTD_write64(void *memPtr, U64 value) { put_unaligned(value, (U64 *)memPtr); }
-
-/*=== Little endian r/w ===*/
-
-ZSTD_STATIC U16 ZSTD_readLE16(const void *memPtr) { return get_unaligned_le16(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeLE16(void *memPtr, U16 val) { put_unaligned_le16(val, memPtr); }
-
-ZSTD_STATIC U32 ZSTD_readLE24(const void *memPtr) { return ZSTD_readLE16(memPtr) + (((const BYTE *)memPtr)[2] << 16); }
-
-ZSTD_STATIC void ZSTD_writeLE24(void *memPtr, U32 val)
-{
-	ZSTD_writeLE16(memPtr, (U16)val);
-	((BYTE *)memPtr)[2] = (BYTE)(val >> 16);
-}
-
-ZSTD_STATIC U32 ZSTD_readLE32(const void *memPtr) { return get_unaligned_le32(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeLE32(void *memPtr, U32 val32) { put_unaligned_le32(val32, memPtr); }
-
-ZSTD_STATIC U64 ZSTD_readLE64(const void *memPtr) { return get_unaligned_le64(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeLE64(void *memPtr, U64 val64) { put_unaligned_le64(val64, memPtr); }
-
-ZSTD_STATIC size_t ZSTD_readLEST(const void *memPtr)
-{
-	if (ZSTD_32bits())
-		return (size_t)ZSTD_readLE32(memPtr);
-	else
-		return (size_t)ZSTD_readLE64(memPtr);
-}
-
-ZSTD_STATIC void ZSTD_writeLEST(void *memPtr, size_t val)
-{
-	if (ZSTD_32bits())
-		ZSTD_writeLE32(memPtr, (U32)val);
-	else
-		ZSTD_writeLE64(memPtr, (U64)val);
-}
-
-/*=== Big endian r/w ===*/
-
-ZSTD_STATIC U32 ZSTD_readBE32(const void *memPtr) { return get_unaligned_be32(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeBE32(void *memPtr, U32 val32) { put_unaligned_be32(val32, memPtr); }
-
-ZSTD_STATIC U64 ZSTD_readBE64(const void *memPtr) { return get_unaligned_be64(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeBE64(void *memPtr, U64 val64) { put_unaligned_be64(val64, memPtr); }
-
-ZSTD_STATIC size_t ZSTD_readBEST(const void *memPtr)
-{
-	if (ZSTD_32bits())
-		return (size_t)ZSTD_readBE32(memPtr);
-	else
-		return (size_t)ZSTD_readBE64(memPtr);
-}
-
-ZSTD_STATIC void ZSTD_writeBEST(void *memPtr, size_t val)
-{
-	if (ZSTD_32bits())
-		ZSTD_writeBE32(memPtr, (U32)val);
-	else
-		ZSTD_writeBE64(memPtr, (U64)val);
-}
-
-/* function safe only for comparisons */
-ZSTD_STATIC U32 ZSTD_readMINMATCH(const void *memPtr, U32 length)
-{
-	switch (length) {
-	default:
-	case 4: return ZSTD_read32(memPtr);
-	case 3:
-		if (ZSTD_isLittleEndian())
-			return ZSTD_read32(memPtr) << 8;
-		else
-			return ZSTD_read32(memPtr) >> 8;
-	}
-}
-
-#endif /* MEM_H_MODULE */
diff --git a/lib/zstd/zstd_decompress_module.c b/lib/zstd/zstd_decompress_module.c
new file mode 100644
index 000000000000..6fec1f178de7
--- /dev/null
+++ b/lib/zstd/zstd_decompress_module.c
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause
+/*
+ * Copyright (c) Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#include <linux/kernel.h>
+#include <linux/export.h>
+#include <linux/string.h>
+#include <linux/zstd.h>
+
+#include "common/zstd_deps.h"
+
+/* Common symbols. zstd_compress must depend on zstd_decompress. */
+
+unsigned int zstd_is_error(size_t code)
+{
+	return ZSTD_isError(code);
+}
+EXPORT_SYMBOL(zstd_is_error);
+
+zstd_error_code zstd_get_error_code(size_t code)
+{
+	return ZSTD_getErrorCode(code);
+}
+EXPORT_SYMBOL(zstd_get_error_code);
+
+const char *zstd_get_error_name(size_t code)
+{
+	return ZSTD_getErrorName(code);
+}
+EXPORT_SYMBOL(zstd_get_error_name);
+
+/* Decompression symbols. */
+
+size_t zstd_dctx_workspace_bound(void)
+{
+	return ZSTD_estimateDCtxSize();
+}
+EXPORT_SYMBOL(zstd_dctx_workspace_bound);
+
+zstd_dctx *zstd_init_dctx(void *workspace, size_t workspace_size)
+{
+	if (workspace == NULL)
+		return NULL;
+	return ZSTD_initStaticDCtx(workspace, workspace_size);
+}
+EXPORT_SYMBOL(zstd_init_dctx);
+
+size_t zstd_decompress_dctx(zstd_dctx *dctx, void *dst, size_t dst_capacity,
+	const void *src, size_t src_size)
+{
+	return ZSTD_decompressDCtx(dctx, dst, dst_capacity, src, src_size);
+}
+EXPORT_SYMBOL(zstd_decompress_dctx);
+
+size_t zstd_dstream_workspace_bound(size_t max_window_size)
+{
+	return ZSTD_estimateDStreamSize(max_window_size);
+}
+EXPORT_SYMBOL(zstd_dstream_workspace_bound);
+
+zstd_dstream *zstd_init_dstream(size_t max_window_size, void *workspace,
+	size_t workspace_size)
+{
+	if (workspace == NULL)
+		return NULL;
+	(void)max_window_size;
+	return ZSTD_initStaticDStream(workspace, workspace_size);
+}
+EXPORT_SYMBOL(zstd_init_dstream);
+
+size_t zstd_reset_dstream(zstd_dstream *dstream)
+{
+	return ZSTD_resetDStream(dstream);
+}
+EXPORT_SYMBOL(zstd_reset_dstream);
+
+size_t zstd_decompress_stream(zstd_dstream *dstream, zstd_out_buffer *output,
+	zstd_in_buffer *input)
+{
+	return ZSTD_decompressStream(dstream, output, input);
+}
+EXPORT_SYMBOL(zstd_decompress_stream);
+
+size_t zstd_find_frame_compressed_size(const void *src, size_t src_size)
+{
+	return ZSTD_findFrameCompressedSize(src, src_size);
+}
+EXPORT_SYMBOL(zstd_find_frame_compressed_size);
+
+size_t zstd_get_frame_header(zstd_frame_header *header, const void *src,
+	size_t src_size)
+{
+	return ZSTD_getFrameHeader(header, src, src_size);
+}
+EXPORT_SYMBOL(zstd_get_frame_header);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_DESCRIPTION("Zstd Decompressor");
diff --git a/lib/zstd/zstd_internal.h b/lib/zstd/zstd_internal.h
deleted file mode 100644
index 1a79fab9e13a..000000000000
--- a/lib/zstd/zstd_internal.h
+++ /dev/null
@@ -1,263 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-#ifndef ZSTD_CCOMMON_H_MODULE
-#define ZSTD_CCOMMON_H_MODULE
-
-/*-*******************************************************
-*  Compiler specifics
-*********************************************************/
-#define FORCE_INLINE static __always_inline
-#define FORCE_NOINLINE static noinline
-
-/*-*************************************
-*  Dependencies
-***************************************/
-#include "error_private.h"
-#include "mem.h"
-#include <linux/compiler.h>
-#include <linux/kernel.h>
-#include <linux/xxhash.h>
-#include <linux/zstd.h>
-
-/*-*************************************
-*  shared macros
-***************************************/
-#define MIN(a, b) ((a) < (b) ? (a) : (b))
-#define MAX(a, b) ((a) > (b) ? (a) : (b))
-#define CHECK_F(f)                       \
-	{                                \
-		size_t const errcod = f; \
-		if (ERR_isError(errcod)) \
-			return errcod;   \
-	} /* check and Forward error code */
-#define CHECK_E(f, e)                    \
-	{                                \
-		size_t const errcod = f; \
-		if (ERR_isError(errcod)) \
-			return ERROR(e); \
-	} /* check and send Error code */
-#define ZSTD_STATIC_ASSERT(c)                                   \
-	{                                                       \
-		enum { ZSTD_static_assert = 1 / (int)(!!(c)) }; \
-	}
-
-/*-*************************************
-*  Common constants
-***************************************/
-#define ZSTD_OPT_NUM (1 << 12)
-#define ZSTD_DICT_MAGIC 0xEC30A437 /* v0.7+ */
-
-#define ZSTD_REP_NUM 3		      /* number of repcodes */
-#define ZSTD_REP_CHECK (ZSTD_REP_NUM) /* number of repcodes to check by the optimal parser */
-#define ZSTD_REP_MOVE (ZSTD_REP_NUM - 1)
-#define ZSTD_REP_MOVE_OPT (ZSTD_REP_NUM)
-static const U32 repStartValue[ZSTD_REP_NUM] = {1, 4, 8};
-
-#define KB *(1 << 10)
-#define MB *(1 << 20)
-#define GB *(1U << 30)
-
-#define BIT7 128
-#define BIT6 64
-#define BIT5 32
-#define BIT4 16
-#define BIT1 2
-#define BIT0 1
-
-#define ZSTD_WINDOWLOG_ABSOLUTEMIN 10
-static const size_t ZSTD_fcs_fieldSize[4] = {0, 2, 4, 8};
-static const size_t ZSTD_did_fieldSize[4] = {0, 1, 2, 4};
-
-#define ZSTD_BLOCKHEADERSIZE 3 /* C standard doesn't allow `static const` variable to be init using another `static const` variable */
-static const size_t ZSTD_blockHeaderSize = ZSTD_BLOCKHEADERSIZE;
-typedef enum { bt_raw, bt_rle, bt_compressed, bt_reserved } blockType_e;
-
-#define MIN_SEQUENCES_SIZE 1									  /* nbSeq==0 */
-#define MIN_CBLOCK_SIZE (1 /*litCSize*/ + 1 /* RLE or RAW */ + MIN_SEQUENCES_SIZE /* nbSeq==0 */) /* for a non-null block */
-
-#define HufLog 12
-typedef enum { set_basic, set_rle, set_compressed, set_repeat } symbolEncodingType_e;
-
-#define LONGNBSEQ 0x7F00
-
-#define MINMATCH 3
-#define EQUAL_READ32 4
-
-#define Litbits 8
-#define MaxLit ((1 << Litbits) - 1)
-#define MaxML 52
-#define MaxLL 35
-#define MaxOff 28
-#define MaxSeq MAX(MaxLL, MaxML) /* Assumption : MaxOff < MaxLL,MaxML */
-#define MLFSELog 9
-#define LLFSELog 9
-#define OffFSELog 8
-
-static const U32 LL_bits[MaxLL + 1] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16};
-static const S16 LL_defaultNorm[MaxLL + 1] = {4, 3, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 2, 1, 1, 1, 1, 1, -1, -1, -1, -1};
-#define LL_DEFAULTNORMLOG 6 /* for static allocation */
-static const U32 LL_defaultNormLog = LL_DEFAULTNORMLOG;
-
-static const U32 ML_bits[MaxML + 1] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  0,  0,  0,  0,  0,  0, 0,
-				       0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 4, 4, 5, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16};
-static const S16 ML_defaultNorm[MaxML + 1] = {1, 4, 3, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  1,  1,  1,  1,  1,  1, 1,
-					      1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, -1, -1, -1, -1, -1, -1, -1};
-#define ML_DEFAULTNORMLOG 6 /* for static allocation */
-static const U32 ML_defaultNormLog = ML_DEFAULTNORMLOG;
-
-static const S16 OF_defaultNorm[MaxOff + 1] = {1, 1, 1, 1, 1, 1, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, -1, -1, -1, -1, -1};
-#define OF_DEFAULTNORMLOG 5 /* for static allocation */
-static const U32 OF_defaultNormLog = OF_DEFAULTNORMLOG;
-
-/*-*******************************************
-*  Shared functions to include for inlining
-*********************************************/
-ZSTD_STATIC void ZSTD_copy8(void *dst, const void *src) {
-	memcpy(dst, src, 8);
-}
-/*! ZSTD_wildcopy() :
-*   custom version of memcpy(), can copy up to 7 bytes too many (8 bytes if length==0) */
-#define WILDCOPY_OVERLENGTH 8
-ZSTD_STATIC void ZSTD_wildcopy(void *dst, const void *src, ptrdiff_t length)
-{
-	const BYTE* ip = (const BYTE*)src;
-	BYTE* op = (BYTE*)dst;
-	BYTE* const oend = op + length;
-	/* Work around https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81388.
-	 * Avoid the bad case where the loop only runs once by handling the
-	 * special case separately. This doesn't trigger the bug because it
-	 * doesn't involve pointer/integer overflow.
-	 */
-	if (length <= 8)
-		return ZSTD_copy8(dst, src);
-	do {
-		ZSTD_copy8(op, ip);
-		op += 8;
-		ip += 8;
-	} while (op < oend);
-}
-
-/*-*******************************************
-*  Private interfaces
-*********************************************/
-typedef struct ZSTD_stats_s ZSTD_stats_t;
-
-typedef struct {
-	U32 off;
-	U32 len;
-} ZSTD_match_t;
-
-typedef struct {
-	U32 price;
-	U32 off;
-	U32 mlen;
-	U32 litlen;
-	U32 rep[ZSTD_REP_NUM];
-} ZSTD_optimal_t;
-
-typedef struct seqDef_s {
-	U32 offset;
-	U16 litLength;
-	U16 matchLength;
-} seqDef;
-
-typedef struct {
-	seqDef *sequencesStart;
-	seqDef *sequences;
-	BYTE *litStart;
-	BYTE *lit;
-	BYTE *llCode;
-	BYTE *mlCode;
-	BYTE *ofCode;
-	U32 longLengthID; /* 0 == no longLength; 1 == Lit.longLength; 2 == Match.longLength; */
-	U32 longLengthPos;
-	/* opt */
-	ZSTD_optimal_t *priceTable;
-	ZSTD_match_t *matchTable;
-	U32 *matchLengthFreq;
-	U32 *litLengthFreq;
-	U32 *litFreq;
-	U32 *offCodeFreq;
-	U32 matchLengthSum;
-	U32 matchSum;
-	U32 litLengthSum;
-	U32 litSum;
-	U32 offCodeSum;
-	U32 log2matchLengthSum;
-	U32 log2matchSum;
-	U32 log2litLengthSum;
-	U32 log2litSum;
-	U32 log2offCodeSum;
-	U32 factor;
-	U32 staticPrices;
-	U32 cachedPrice;
-	U32 cachedLitLength;
-	const BYTE *cachedLiterals;
-} seqStore_t;
-
-const seqStore_t *ZSTD_getSeqStore(const ZSTD_CCtx *ctx);
-void ZSTD_seqToCodes(const seqStore_t *seqStorePtr);
-int ZSTD_isSkipFrame(ZSTD_DCtx *dctx);
-
-/*= Custom memory allocation functions */
-typedef void *(*ZSTD_allocFunction)(void *opaque, size_t size);
-typedef void (*ZSTD_freeFunction)(void *opaque, void *address);
-typedef struct {
-	ZSTD_allocFunction customAlloc;
-	ZSTD_freeFunction customFree;
-	void *opaque;
-} ZSTD_customMem;
-
-void *ZSTD_malloc(size_t size, ZSTD_customMem customMem);
-void ZSTD_free(void *ptr, ZSTD_customMem customMem);
-
-/*====== stack allocation  ======*/
-
-typedef struct {
-	void *ptr;
-	const void *end;
-} ZSTD_stack;
-
-#define ZSTD_ALIGN(x) ALIGN(x, sizeof(size_t))
-#define ZSTD_PTR_ALIGN(p) PTR_ALIGN(p, sizeof(size_t))
-
-ZSTD_customMem ZSTD_initStack(void *workspace, size_t workspaceSize);
-
-void *ZSTD_stackAllocAll(void *opaque, size_t *size);
-void *ZSTD_stackAlloc(void *opaque, size_t size);
-void ZSTD_stackFree(void *opaque, void *address);
-
-/*======  common function  ======*/
-
-ZSTD_STATIC U32 ZSTD_highbit32(U32 val) { return 31 - __builtin_clz(val); }
-
-/* hidden functions */
-
-/* ZSTD_invalidateRepCodes() :
- * ensures next compression will not use repcodes from previous block.
- * Note : only works with regular variant;
- *        do not use with extDict variant ! */
-void ZSTD_invalidateRepCodes(ZSTD_CCtx *cctx);
-
-size_t ZSTD_freeCCtx(ZSTD_CCtx *cctx);
-size_t ZSTD_freeDCtx(ZSTD_DCtx *dctx);
-size_t ZSTD_freeCDict(ZSTD_CDict *cdict);
-size_t ZSTD_freeDDict(ZSTD_DDict *cdict);
-size_t ZSTD_freeCStream(ZSTD_CStream *zcs);
-size_t ZSTD_freeDStream(ZSTD_DStream *zds);
-
-#endif /* ZSTD_CCOMMON_H_MODULE */
diff --git a/lib/zstd/zstd_opt.h b/lib/zstd/zstd_opt.h
deleted file mode 100644
index 55e1b4cba808..000000000000
--- a/lib/zstd/zstd_opt.h
+++ /dev/null
@@ -1,1014 +0,0 @@
-/**
- * Copyright (c) 2016-present, Przemyslaw Skibinski, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-/* Note : this file is intended to be included within zstd_compress.c */
-
-#ifndef ZSTD_OPT_H_91842398743
-#define ZSTD_OPT_H_91842398743
-
-#define ZSTD_LITFREQ_ADD 2
-#define ZSTD_FREQ_DIV 4
-#define ZSTD_MAX_PRICE (1 << 30)
-
-/*-*************************************
-*  Price functions for optimal parser
-***************************************/
-FORCE_INLINE void ZSTD_setLog2Prices(seqStore_t *ssPtr)
-{
-	ssPtr->log2matchLengthSum = ZSTD_highbit32(ssPtr->matchLengthSum + 1);
-	ssPtr->log2litLengthSum = ZSTD_highbit32(ssPtr->litLengthSum + 1);
-	ssPtr->log2litSum = ZSTD_highbit32(ssPtr->litSum + 1);
-	ssPtr->log2offCodeSum = ZSTD_highbit32(ssPtr->offCodeSum + 1);
-	ssPtr->factor = 1 + ((ssPtr->litSum >> 5) / ssPtr->litLengthSum) + ((ssPtr->litSum << 1) / (ssPtr->litSum + ssPtr->matchSum));
-}
-
-ZSTD_STATIC void ZSTD_rescaleFreqs(seqStore_t *ssPtr, const BYTE *src, size_t srcSize)
-{
-	unsigned u;
-
-	ssPtr->cachedLiterals = NULL;
-	ssPtr->cachedPrice = ssPtr->cachedLitLength = 0;
-	ssPtr->staticPrices = 0;
-
-	if (ssPtr->litLengthSum == 0) {
-		if (srcSize <= 1024)
-			ssPtr->staticPrices = 1;
-
-		for (u = 0; u <= MaxLit; u++)
-			ssPtr->litFreq[u] = 0;
-		for (u = 0; u < srcSize; u++)
-			ssPtr->litFreq[src[u]]++;
-
-		ssPtr->litSum = 0;
-		ssPtr->litLengthSum = MaxLL + 1;
-		ssPtr->matchLengthSum = MaxML + 1;
-		ssPtr->offCodeSum = (MaxOff + 1);
-		ssPtr->matchSum = (ZSTD_LITFREQ_ADD << Litbits);
-
-		for (u = 0; u <= MaxLit; u++) {
-			ssPtr->litFreq[u] = 1 + (ssPtr->litFreq[u] >> ZSTD_FREQ_DIV);
-			ssPtr->litSum += ssPtr->litFreq[u];
-		}
-		for (u = 0; u <= MaxLL; u++)
-			ssPtr->litLengthFreq[u] = 1;
-		for (u = 0; u <= MaxML; u++)
-			ssPtr->matchLengthFreq[u] = 1;
-		for (u = 0; u <= MaxOff; u++)
-			ssPtr->offCodeFreq[u] = 1;
-	} else {
-		ssPtr->matchLengthSum = 0;
-		ssPtr->litLengthSum = 0;
-		ssPtr->offCodeSum = 0;
-		ssPtr->matchSum = 0;
-		ssPtr->litSum = 0;
-
-		for (u = 0; u <= MaxLit; u++) {
-			ssPtr->litFreq[u] = 1 + (ssPtr->litFreq[u] >> (ZSTD_FREQ_DIV + 1));
-			ssPtr->litSum += ssPtr->litFreq[u];
-		}
-		for (u = 0; u <= MaxLL; u++) {
-			ssPtr->litLengthFreq[u] = 1 + (ssPtr->litLengthFreq[u] >> (ZSTD_FREQ_DIV + 1));
-			ssPtr->litLengthSum += ssPtr->litLengthFreq[u];
-		}
-		for (u = 0; u <= MaxML; u++) {
-			ssPtr->matchLengthFreq[u] = 1 + (ssPtr->matchLengthFreq[u] >> ZSTD_FREQ_DIV);
-			ssPtr->matchLengthSum += ssPtr->matchLengthFreq[u];
-			ssPtr->matchSum += ssPtr->matchLengthFreq[u] * (u + 3);
-		}
-		ssPtr->matchSum *= ZSTD_LITFREQ_ADD;
-		for (u = 0; u <= MaxOff; u++) {
-			ssPtr->offCodeFreq[u] = 1 + (ssPtr->offCodeFreq[u] >> ZSTD_FREQ_DIV);
-			ssPtr->offCodeSum += ssPtr->offCodeFreq[u];
-		}
-	}
-
-	ZSTD_setLog2Prices(ssPtr);
-}
-
-FORCE_INLINE U32 ZSTD_getLiteralPrice(seqStore_t *ssPtr, U32 litLength, const BYTE *literals)
-{
-	U32 price, u;
-
-	if (ssPtr->staticPrices)
-		return ZSTD_highbit32((U32)litLength + 1) + (litLength * 6);
-
-	if (litLength == 0)
-		return ssPtr->log2litLengthSum - ZSTD_highbit32(ssPtr->litLengthFreq[0] + 1);
-
-	/* literals */
-	if (ssPtr->cachedLiterals == literals) {
-		U32 const additional = litLength - ssPtr->cachedLitLength;
-		const BYTE *literals2 = ssPtr->cachedLiterals + ssPtr->cachedLitLength;
-		price = ssPtr->cachedPrice + additional * ssPtr->log2litSum;
-		for (u = 0; u < additional; u++)
-			price -= ZSTD_highbit32(ssPtr->litFreq[literals2[u]] + 1);
-		ssPtr->cachedPrice = price;
-		ssPtr->cachedLitLength = litLength;
-	} else {
-		price = litLength * ssPtr->log2litSum;
-		for (u = 0; u < litLength; u++)
-			price -= ZSTD_highbit32(ssPtr->litFreq[literals[u]] + 1);
-
-		if (litLength >= 12) {
-			ssPtr->cachedLiterals = literals;
-			ssPtr->cachedPrice = price;
-			ssPtr->cachedLitLength = litLength;
-		}
-	}
-
-	/* literal Length */
-	{
-		const BYTE LL_deltaCode = 19;
-		const BYTE llCode = (litLength > 63) ? (BYTE)ZSTD_highbit32(litLength) + LL_deltaCode : LL_Code[litLength];
-		price += LL_bits[llCode] + ssPtr->log2litLengthSum - ZSTD_highbit32(ssPtr->litLengthFreq[llCode] + 1);
-	}
-
-	return price;
-}
-
-FORCE_INLINE U32 ZSTD_getPrice(seqStore_t *seqStorePtr, U32 litLength, const BYTE *literals, U32 offset, U32 matchLength, const int ultra)
-{
-	/* offset */
-	U32 price;
-	BYTE const offCode = (BYTE)ZSTD_highbit32(offset + 1);
-
-	if (seqStorePtr->staticPrices)
-		return ZSTD_getLiteralPrice(seqStorePtr, litLength, literals) + ZSTD_highbit32((U32)matchLength + 1) + 16 + offCode;
-
-	price = offCode + seqStorePtr->log2offCodeSum - ZSTD_highbit32(seqStorePtr->offCodeFreq[offCode] + 1);
-	if (!ultra && offCode >= 20)
-		price += (offCode - 19) * 2;
-
-	/* match Length */
-	{
-		const BYTE ML_deltaCode = 36;
-		const BYTE mlCode = (matchLength > 127) ? (BYTE)ZSTD_highbit32(matchLength) + ML_deltaCode : ML_Code[matchLength];
-		price += ML_bits[mlCode] + seqStorePtr->log2matchLengthSum - ZSTD_highbit32(seqStorePtr->matchLengthFreq[mlCode] + 1);
-	}
-
-	return price + ZSTD_getLiteralPrice(seqStorePtr, litLength, literals) + seqStorePtr->factor;
-}
-
-ZSTD_STATIC void ZSTD_updatePrice(seqStore_t *seqStorePtr, U32 litLength, const BYTE *literals, U32 offset, U32 matchLength)
-{
-	U32 u;
-
-	/* literals */
-	seqStorePtr->litSum += litLength * ZSTD_LITFREQ_ADD;
-	for (u = 0; u < litLength; u++)
-		seqStorePtr->litFreq[literals[u]] += ZSTD_LITFREQ_ADD;
-
-	/* literal Length */
-	{
-		const BYTE LL_deltaCode = 19;
-		const BYTE llCode = (litLength > 63) ? (BYTE)ZSTD_highbit32(litLength) + LL_deltaCode : LL_Code[litLength];
-		seqStorePtr->litLengthFreq[llCode]++;
-		seqStorePtr->litLengthSum++;
-	}
-
-	/* match offset */
-	{
-		BYTE const offCode = (BYTE)ZSTD_highbit32(offset + 1);
-		seqStorePtr->offCodeSum++;
-		seqStorePtr->offCodeFreq[offCode]++;
-	}
-
-	/* match Length */
-	{
-		const BYTE ML_deltaCode = 36;
-		const BYTE mlCode = (matchLength > 127) ? (BYTE)ZSTD_highbit32(matchLength) + ML_deltaCode : ML_Code[matchLength];
-		seqStorePtr->matchLengthFreq[mlCode]++;
-		seqStorePtr->matchLengthSum++;
-	}
-
-	ZSTD_setLog2Prices(seqStorePtr);
-}
-
-#define SET_PRICE(pos, mlen_, offset_, litlen_, price_)           \
-	{                                                         \
-		while (last_pos < pos) {                          \
-			opt[last_pos + 1].price = ZSTD_MAX_PRICE; \
-			last_pos++;                               \
-		}                                                 \
-		opt[pos].mlen = mlen_;                            \
-		opt[pos].off = offset_;                           \
-		opt[pos].litlen = litlen_;                        \
-		opt[pos].price = price_;                          \
-	}
-
-/* Update hashTable3 up to ip (excluded)
-   Assumption : always within prefix (i.e. not within extDict) */
-FORCE_INLINE
-U32 ZSTD_insertAndFindFirstIndexHash3(ZSTD_CCtx *zc, const BYTE *ip)
-{
-	U32 *const hashTable3 = zc->hashTable3;
-	U32 const hashLog3 = zc->hashLog3;
-	const BYTE *const base = zc->base;
-	U32 idx = zc->nextToUpdate3;
-	const U32 target = zc->nextToUpdate3 = (U32)(ip - base);
-	const size_t hash3 = ZSTD_hash3Ptr(ip, hashLog3);
-
-	while (idx < target) {
-		hashTable3[ZSTD_hash3Ptr(base + idx, hashLog3)] = idx;
-		idx++;
-	}
-
-	return hashTable3[hash3];
-}
-
-/*-*************************************
-*  Binary Tree search
-***************************************/
-static U32 ZSTD_insertBtAndGetAllMatches(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, U32 nbCompares, const U32 mls, U32 extDict,
-					 ZSTD_match_t *matches, const U32 minMatchLen)
-{
-	const BYTE *const base = zc->base;
-	const U32 curr = (U32)(ip - base);
-	const U32 hashLog = zc->params.cParams.hashLog;
-	const size_t h = ZSTD_hashPtr(ip, hashLog, mls);
-	U32 *const hashTable = zc->hashTable;
-	U32 matchIndex = hashTable[h];
-	U32 *const bt = zc->chainTable;
-	const U32 btLog = zc->params.cParams.chainLog - 1;
-	const U32 btMask = (1U << btLog) - 1;
-	size_t commonLengthSmaller = 0, commonLengthLarger = 0;
-	const BYTE *const dictBase = zc->dictBase;
-	const U32 dictLimit = zc->dictLimit;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-	const BYTE *const prefixStart = base + dictLimit;
-	const U32 btLow = btMask >= curr ? 0 : curr - btMask;
-	const U32 windowLow = zc->lowLimit;
-	U32 *smallerPtr = bt + 2 * (curr & btMask);
-	U32 *largerPtr = bt + 2 * (curr & btMask) + 1;
-	U32 matchEndIdx = curr + 8;
-	U32 dummy32; /* to be nullified at the end */
-	U32 mnum = 0;
-
-	const U32 minMatch = (mls == 3) ? 3 : 4;
-	size_t bestLength = minMatchLen - 1;
-
-	if (minMatch == 3) { /* HC3 match finder */
-		U32 const matchIndex3 = ZSTD_insertAndFindFirstIndexHash3(zc, ip);
-		if (matchIndex3 > windowLow && (curr - matchIndex3 < (1 << 18))) {
-			const BYTE *match;
-			size_t currMl = 0;
-			if ((!extDict) || matchIndex3 >= dictLimit) {
-				match = base + matchIndex3;
-				if (match[bestLength] == ip[bestLength])
-					currMl = ZSTD_count(ip, match, iLimit);
-			} else {
-				match = dictBase + matchIndex3;
-				if (ZSTD_readMINMATCH(match, MINMATCH) ==
-				    ZSTD_readMINMATCH(ip, MINMATCH)) /* assumption : matchIndex3 <= dictLimit-4 (by table construction) */
-					currMl = ZSTD_count_2segments(ip + MINMATCH, match + MINMATCH, iLimit, dictEnd, prefixStart) + MINMATCH;
-			}
-
-			/* save best solution */
-			if (currMl > bestLength) {
-				bestLength = currMl;
-				matches[mnum].off = ZSTD_REP_MOVE_OPT + curr - matchIndex3;
-				matches[mnum].len = (U32)currMl;
-				mnum++;
-				if (currMl > ZSTD_OPT_NUM)
-					goto update;
-				if (ip + currMl == iLimit)
-					goto update; /* best possible, and avoid read overflow*/
-			}
-		}
-	}
-
-	hashTable[h] = curr; /* Update Hash Table */
-
-	while (nbCompares-- && (matchIndex > windowLow)) {
-		U32 *nextPtr = bt + 2 * (matchIndex & btMask);
-		size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
-		const BYTE *match;
-
-		if ((!extDict) || (matchIndex + matchLength >= dictLimit)) {
-			match = base + matchIndex;
-			if (match[matchLength] == ip[matchLength]) {
-				matchLength += ZSTD_count(ip + matchLength + 1, match + matchLength + 1, iLimit) + 1;
-			}
-		} else {
-			match = dictBase + matchIndex;
-			matchLength += ZSTD_count_2segments(ip + matchLength, match + matchLength, iLimit, dictEnd, prefixStart);
-			if (matchIndex + matchLength >= dictLimit)
-				match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
-		}
-
-		if (matchLength > bestLength) {
-			if (matchLength > matchEndIdx - matchIndex)
-				matchEndIdx = matchIndex + (U32)matchLength;
-			bestLength = matchLength;
-			matches[mnum].off = ZSTD_REP_MOVE_OPT + curr - matchIndex;
-			matches[mnum].len = (U32)matchLength;
-			mnum++;
-			if (matchLength > ZSTD_OPT_NUM)
-				break;
-			if (ip + matchLength == iLimit) /* equal : no way to know if inf or sup */
-				break;			/* drop, to guarantee consistency (miss a little bit of compression) */
-		}
-
-		if (match[matchLength] < ip[matchLength]) {
-			/* match is smaller than curr */
-			*smallerPtr = matchIndex;	  /* update smaller idx */
-			commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
-			if (matchIndex <= btLow) {
-				smallerPtr = &dummy32;
-				break;
-			}			  /* beyond tree size, stop the search */
-			smallerPtr = nextPtr + 1; /* new "smaller" => larger of match */
-			matchIndex = nextPtr[1];  /* new matchIndex larger than previous (closer to curr) */
-		} else {
-			/* match is larger than curr */
-			*largerPtr = matchIndex;
-			commonLengthLarger = matchLength;
-			if (matchIndex <= btLow) {
-				largerPtr = &dummy32;
-				break;
-			} /* beyond tree size, stop the search */
-			largerPtr = nextPtr;
-			matchIndex = nextPtr[0];
-		}
-	}
-
-	*smallerPtr = *largerPtr = 0;
-
-update:
-	zc->nextToUpdate = (matchEndIdx > curr + 8) ? matchEndIdx - 8 : curr + 1;
-	return mnum;
-}
-
-/** Tree updater, providing best match */
-static U32 ZSTD_BtGetAllMatches(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, const U32 maxNbAttempts, const U32 mls, ZSTD_match_t *matches,
-				const U32 minMatchLen)
-{
-	if (ip < zc->base + zc->nextToUpdate)
-		return 0; /* skipped area */
-	ZSTD_updateTree(zc, ip, iLimit, maxNbAttempts, mls);
-	return ZSTD_insertBtAndGetAllMatches(zc, ip, iLimit, maxNbAttempts, mls, 0, matches, minMatchLen);
-}
-
-static U32 ZSTD_BtGetAllMatches_selectMLS(ZSTD_CCtx *zc, /* Index table will be updated */
-					  const BYTE *ip, const BYTE *const iHighLimit, const U32 maxNbAttempts, const U32 matchLengthSearch,
-					  ZSTD_match_t *matches, const U32 minMatchLen)
-{
-	switch (matchLengthSearch) {
-	case 3: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 3, matches, minMatchLen);
-	default:
-	case 4: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 4, matches, minMatchLen);
-	case 5: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 5, matches, minMatchLen);
-	case 7:
-	case 6: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 6, matches, minMatchLen);
-	}
-}
-
-/** Tree updater, providing best match */
-static U32 ZSTD_BtGetAllMatches_extDict(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, const U32 maxNbAttempts, const U32 mls,
-					ZSTD_match_t *matches, const U32 minMatchLen)
-{
-	if (ip < zc->base + zc->nextToUpdate)
-		return 0; /* skipped area */
-	ZSTD_updateTree_extDict(zc, ip, iLimit, maxNbAttempts, mls);
-	return ZSTD_insertBtAndGetAllMatches(zc, ip, iLimit, maxNbAttempts, mls, 1, matches, minMatchLen);
-}
-
-static U32 ZSTD_BtGetAllMatches_selectMLS_extDict(ZSTD_CCtx *zc, /* Index table will be updated */
-						  const BYTE *ip, const BYTE *const iHighLimit, const U32 maxNbAttempts, const U32 matchLengthSearch,
-						  ZSTD_match_t *matches, const U32 minMatchLen)
-{
-	switch (matchLengthSearch) {
-	case 3: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 3, matches, minMatchLen);
-	default:
-	case 4: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 4, matches, minMatchLen);
-	case 5: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 5, matches, minMatchLen);
-	case 7:
-	case 6: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 6, matches, minMatchLen);
-	}
-}
-
-/*-*******************************
-*  Optimal parser
-*********************************/
-FORCE_INLINE
-void ZSTD_compressBlock_opt_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const int ultra)
-{
-	seqStore_t *seqStorePtr = &(ctx->seqStore);
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - 8;
-	const BYTE *const base = ctx->base;
-	const BYTE *const prefixStart = base + ctx->dictLimit;
-
-	const U32 maxSearches = 1U << ctx->params.cParams.searchLog;
-	const U32 sufficient_len = ctx->params.cParams.targetLength;
-	const U32 mls = ctx->params.cParams.searchLength;
-	const U32 minMatch = (ctx->params.cParams.searchLength == 3) ? 3 : 4;
-
-	ZSTD_optimal_t *opt = seqStorePtr->priceTable;
-	ZSTD_match_t *matches = seqStorePtr->matchTable;
-	const BYTE *inr;
-	U32 offset, rep[ZSTD_REP_NUM];
-
-	/* init */
-	ctx->nextToUpdate3 = ctx->nextToUpdate;
-	ZSTD_rescaleFreqs(seqStorePtr, (const BYTE *)src, srcSize);
-	ip += (ip == prefixStart);
-	{
-		U32 i;
-		for (i = 0; i < ZSTD_REP_NUM; i++)
-			rep[i] = ctx->rep[i];
-	}
-
-	/* Match Loop */
-	while (ip < ilimit) {
-		U32 cur, match_num, last_pos, litlen, price;
-		U32 u, mlen, best_mlen, best_off, litLength;
-		memset(opt, 0, sizeof(ZSTD_optimal_t));
-		last_pos = 0;
-		litlen = (U32)(ip - anchor);
-
-		/* check repCode */
-		{
-			U32 i, last_i = ZSTD_REP_CHECK + (ip == anchor);
-			for (i = (ip == anchor); i < last_i; i++) {
-				const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : rep[i];
-				if ((repCur > 0) && (repCur < (S32)(ip - prefixStart)) &&
-				    (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(ip - repCur, minMatch))) {
-					mlen = (U32)ZSTD_count(ip + minMatch, ip + minMatch - repCur, iend) + minMatch;
-					if (mlen > sufficient_len || mlen >= ZSTD_OPT_NUM) {
-						best_mlen = mlen;
-						best_off = i;
-						cur = 0;
-						last_pos = 1;
-						goto _storeSequence;
-					}
-					best_off = i - (ip == anchor);
-					do {
-						price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
-						if (mlen > last_pos || price < opt[mlen].price)
-							SET_PRICE(mlen, mlen, i, litlen, price); /* note : macro modifies last_pos */
-						mlen--;
-					} while (mlen >= minMatch);
-				}
-			}
-		}
-
-		match_num = ZSTD_BtGetAllMatches_selectMLS(ctx, ip, iend, maxSearches, mls, matches, minMatch);
-
-		if (!last_pos && !match_num) {
-			ip++;
-			continue;
-		}
-
-		if (match_num && (matches[match_num - 1].len > sufficient_len || matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
-			best_mlen = matches[match_num - 1].len;
-			best_off = matches[match_num - 1].off;
-			cur = 0;
-			last_pos = 1;
-			goto _storeSequence;
-		}
-
-		/* set prices using matches at position = 0 */
-		best_mlen = (last_pos) ? last_pos : minMatch;
-		for (u = 0; u < match_num; u++) {
-			mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
-			best_mlen = matches[u].len;
-			while (mlen <= best_mlen) {
-				price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
-				if (mlen > last_pos || price < opt[mlen].price)
-					SET_PRICE(mlen, mlen, matches[u].off, litlen, price); /* note : macro modifies last_pos */
-				mlen++;
-			}
-		}
-
-		if (last_pos < minMatch) {
-			ip++;
-			continue;
-		}
-
-		/* initialize opt[0] */
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				opt[0].rep[i] = rep[i];
-		}
-		opt[0].mlen = 1;
-		opt[0].litlen = litlen;
-
-		/* check further positions */
-		for (cur = 1; cur <= last_pos; cur++) {
-			inr = ip + cur;
-
-			if (opt[cur - 1].mlen == 1) {
-				litlen = opt[cur - 1].litlen + 1;
-				if (cur > litlen) {
-					price = opt[cur - litlen].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - litlen);
-				} else
-					price = ZSTD_getLiteralPrice(seqStorePtr, litlen, anchor);
-			} else {
-				litlen = 1;
-				price = opt[cur - 1].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - 1);
-			}
-
-			if (cur > last_pos || price <= opt[cur].price)
-				SET_PRICE(cur, 1, 0, litlen, price);
-
-			if (cur == last_pos)
-				break;
-
-			if (inr > ilimit) /* last match must start at a minimum distance of 8 from oend */
-				continue;
-
-			mlen = opt[cur].mlen;
-			if (opt[cur].off > ZSTD_REP_MOVE_OPT) {
-				opt[cur].rep[2] = opt[cur - mlen].rep[1];
-				opt[cur].rep[1] = opt[cur - mlen].rep[0];
-				opt[cur].rep[0] = opt[cur].off - ZSTD_REP_MOVE_OPT;
-			} else {
-				opt[cur].rep[2] = (opt[cur].off > 1) ? opt[cur - mlen].rep[1] : opt[cur - mlen].rep[2];
-				opt[cur].rep[1] = (opt[cur].off > 0) ? opt[cur - mlen].rep[0] : opt[cur - mlen].rep[1];
-				opt[cur].rep[0] =
-				    ((opt[cur].off == ZSTD_REP_MOVE_OPT) && (mlen != 1)) ? (opt[cur - mlen].rep[0] - 1) : (opt[cur - mlen].rep[opt[cur].off]);
-			}
-
-			best_mlen = minMatch;
-			{
-				U32 i, last_i = ZSTD_REP_CHECK + (mlen != 1);
-				for (i = (opt[cur].mlen != 1); i < last_i; i++) { /* check rep */
-					const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (opt[cur].rep[0] - 1) : opt[cur].rep[i];
-					if ((repCur > 0) && (repCur < (S32)(inr - prefixStart)) &&
-					    (ZSTD_readMINMATCH(inr, minMatch) == ZSTD_readMINMATCH(inr - repCur, minMatch))) {
-						mlen = (U32)ZSTD_count(inr + minMatch, inr + minMatch - repCur, iend) + minMatch;
-
-						if (mlen > sufficient_len || cur + mlen >= ZSTD_OPT_NUM) {
-							best_mlen = mlen;
-							best_off = i;
-							last_pos = cur + 1;
-							goto _storeSequence;
-						}
-
-						best_off = i - (opt[cur].mlen != 1);
-						if (mlen > best_mlen)
-							best_mlen = mlen;
-
-						do {
-							if (opt[cur].mlen == 1) {
-								litlen = opt[cur].litlen;
-								if (cur > litlen) {
-									price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, inr - litlen,
-															best_off, mlen - MINMATCH, ultra);
-								} else
-									price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
-							} else {
-								litlen = 0;
-								price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, best_off, mlen - MINMATCH, ultra);
-							}
-
-							if (cur + mlen > last_pos || price <= opt[cur + mlen].price)
-								SET_PRICE(cur + mlen, mlen, i, litlen, price);
-							mlen--;
-						} while (mlen >= minMatch);
-					}
-				}
-			}
-
-			match_num = ZSTD_BtGetAllMatches_selectMLS(ctx, inr, iend, maxSearches, mls, matches, best_mlen);
-
-			if (match_num > 0 && (matches[match_num - 1].len > sufficient_len || cur + matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
-				best_mlen = matches[match_num - 1].len;
-				best_off = matches[match_num - 1].off;
-				last_pos = cur + 1;
-				goto _storeSequence;
-			}
-
-			/* set prices using matches at position = cur */
-			for (u = 0; u < match_num; u++) {
-				mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
-				best_mlen = matches[u].len;
-
-				while (mlen <= best_mlen) {
-					if (opt[cur].mlen == 1) {
-						litlen = opt[cur].litlen;
-						if (cur > litlen)
-							price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, ip + cur - litlen,
-													matches[u].off - 1, mlen - MINMATCH, ultra);
-						else
-							price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
-					} else {
-						litlen = 0;
-						price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, matches[u].off - 1, mlen - MINMATCH, ultra);
-					}
-
-					if (cur + mlen > last_pos || (price < opt[cur + mlen].price))
-						SET_PRICE(cur + mlen, mlen, matches[u].off, litlen, price);
-
-					mlen++;
-				}
-			}
-		}
-
-		best_mlen = opt[last_pos].mlen;
-		best_off = opt[last_pos].off;
-		cur = last_pos - best_mlen;
-
-	/* store sequence */
-_storeSequence: /* cur, last_pos, best_mlen, best_off have to be set */
-		opt[0].mlen = 1;
-
-		while (1) {
-			mlen = opt[cur].mlen;
-			offset = opt[cur].off;
-			opt[cur].mlen = best_mlen;
-			opt[cur].off = best_off;
-			best_mlen = mlen;
-			best_off = offset;
-			if (mlen > cur)
-				break;
-			cur -= mlen;
-		}
-
-		for (u = 0; u <= last_pos;) {
-			u += opt[u].mlen;
-		}
-
-		for (cur = 0; cur < last_pos;) {
-			mlen = opt[cur].mlen;
-			if (mlen == 1) {
-				ip++;
-				cur++;
-				continue;
-			}
-			offset = opt[cur].off;
-			cur += mlen;
-			litLength = (U32)(ip - anchor);
-
-			if (offset > ZSTD_REP_MOVE_OPT) {
-				rep[2] = rep[1];
-				rep[1] = rep[0];
-				rep[0] = offset - ZSTD_REP_MOVE_OPT;
-				offset--;
-			} else {
-				if (offset != 0) {
-					best_off = (offset == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : (rep[offset]);
-					if (offset != 1)
-						rep[2] = rep[1];
-					rep[1] = rep[0];
-					rep[0] = best_off;
-				}
-				if (litLength == 0)
-					offset--;
-			}
-
-			ZSTD_updatePrice(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
-			ZSTD_storeSeq(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
-			anchor = ip = ip + mlen;
-		}
-	} /* for (cur=0; cur < last_pos; ) */
-
-	/* Save reps for next block */
-	{
-		int i;
-		for (i = 0; i < ZSTD_REP_NUM; i++)
-			ctx->repToConfirm[i] = rep[i];
-	}
-
-	/* Last Literals */
-	{
-		size_t const lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-FORCE_INLINE
-void ZSTD_compressBlock_opt_extDict_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const int ultra)
-{
-	seqStore_t *seqStorePtr = &(ctx->seqStore);
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - 8;
-	const BYTE *const base = ctx->base;
-	const U32 lowestIndex = ctx->lowLimit;
-	const U32 dictLimit = ctx->dictLimit;
-	const BYTE *const prefixStart = base + dictLimit;
-	const BYTE *const dictBase = ctx->dictBase;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-
-	const U32 maxSearches = 1U << ctx->params.cParams.searchLog;
-	const U32 sufficient_len = ctx->params.cParams.targetLength;
-	const U32 mls = ctx->params.cParams.searchLength;
-	const U32 minMatch = (ctx->params.cParams.searchLength == 3) ? 3 : 4;
-
-	ZSTD_optimal_t *opt = seqStorePtr->priceTable;
-	ZSTD_match_t *matches = seqStorePtr->matchTable;
-	const BYTE *inr;
-
-	/* init */
-	U32 offset, rep[ZSTD_REP_NUM];
-	{
-		U32 i;
-		for (i = 0; i < ZSTD_REP_NUM; i++)
-			rep[i] = ctx->rep[i];
-	}
-
-	ctx->nextToUpdate3 = ctx->nextToUpdate;
-	ZSTD_rescaleFreqs(seqStorePtr, (const BYTE *)src, srcSize);
-	ip += (ip == prefixStart);
-
-	/* Match Loop */
-	while (ip < ilimit) {
-		U32 cur, match_num, last_pos, litlen, price;
-		U32 u, mlen, best_mlen, best_off, litLength;
-		U32 curr = (U32)(ip - base);
-		memset(opt, 0, sizeof(ZSTD_optimal_t));
-		last_pos = 0;
-		opt[0].litlen = (U32)(ip - anchor);
-
-		/* check repCode */
-		{
-			U32 i, last_i = ZSTD_REP_CHECK + (ip == anchor);
-			for (i = (ip == anchor); i < last_i; i++) {
-				const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : rep[i];
-				const U32 repIndex = (U32)(curr - repCur);
-				const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
-				const BYTE *const repMatch = repBase + repIndex;
-				if ((repCur > 0 && repCur <= (S32)curr) &&
-				    (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
-				    && (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch))) {
-					/* repcode detected we should take it */
-					const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
-					mlen = (U32)ZSTD_count_2segments(ip + minMatch, repMatch + minMatch, iend, repEnd, prefixStart) + minMatch;
-
-					if (mlen > sufficient_len || mlen >= ZSTD_OPT_NUM) {
-						best_mlen = mlen;
-						best_off = i;
-						cur = 0;
-						last_pos = 1;
-						goto _storeSequence;
-					}
-
-					best_off = i - (ip == anchor);
-					litlen = opt[0].litlen;
-					do {
-						price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
-						if (mlen > last_pos || price < opt[mlen].price)
-							SET_PRICE(mlen, mlen, i, litlen, price); /* note : macro modifies last_pos */
-						mlen--;
-					} while (mlen >= minMatch);
-				}
-			}
-		}
-
-		match_num = ZSTD_BtGetAllMatches_selectMLS_extDict(ctx, ip, iend, maxSearches, mls, matches, minMatch); /* first search (depth 0) */
-
-		if (!last_pos && !match_num) {
-			ip++;
-			continue;
-		}
-
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				opt[0].rep[i] = rep[i];
-		}
-		opt[0].mlen = 1;
-
-		if (match_num && (matches[match_num - 1].len > sufficient_len || matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
-			best_mlen = matches[match_num - 1].len;
-			best_off = matches[match_num - 1].off;
-			cur = 0;
-			last_pos = 1;
-			goto _storeSequence;
-		}
-
-		best_mlen = (last_pos) ? last_pos : minMatch;
-
-		/* set prices using matches at position = 0 */
-		for (u = 0; u < match_num; u++) {
-			mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
-			best_mlen = matches[u].len;
-			litlen = opt[0].litlen;
-			while (mlen <= best_mlen) {
-				price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
-				if (mlen > last_pos || price < opt[mlen].price)
-					SET_PRICE(mlen, mlen, matches[u].off, litlen, price);
-				mlen++;
-			}
-		}
-
-		if (last_pos < minMatch) {
-			ip++;
-			continue;
-		}
-
-		/* check further positions */
-		for (cur = 1; cur <= last_pos; cur++) {
-			inr = ip + cur;
-
-			if (opt[cur - 1].mlen == 1) {
-				litlen = opt[cur - 1].litlen + 1;
-				if (cur > litlen) {
-					price = opt[cur - litlen].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - litlen);
-				} else
-					price = ZSTD_getLiteralPrice(seqStorePtr, litlen, anchor);
-			} else {
-				litlen = 1;
-				price = opt[cur - 1].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - 1);
-			}
-
-			if (cur > last_pos || price <= opt[cur].price)
-				SET_PRICE(cur, 1, 0, litlen, price);
-
-			if (cur == last_pos)
-				break;
-
-			if (inr > ilimit) /* last match must start at a minimum distance of 8 from oend */
-				continue;
-
-			mlen = opt[cur].mlen;
-			if (opt[cur].off > ZSTD_REP_MOVE_OPT) {
-				opt[cur].rep[2] = opt[cur - mlen].rep[1];
-				opt[cur].rep[1] = opt[cur - mlen].rep[0];
-				opt[cur].rep[0] = opt[cur].off - ZSTD_REP_MOVE_OPT;
-			} else {
-				opt[cur].rep[2] = (opt[cur].off > 1) ? opt[cur - mlen].rep[1] : opt[cur - mlen].rep[2];
-				opt[cur].rep[1] = (opt[cur].off > 0) ? opt[cur - mlen].rep[0] : opt[cur - mlen].rep[1];
-				opt[cur].rep[0] =
-				    ((opt[cur].off == ZSTD_REP_MOVE_OPT) && (mlen != 1)) ? (opt[cur - mlen].rep[0] - 1) : (opt[cur - mlen].rep[opt[cur].off]);
-			}
-
-			best_mlen = minMatch;
-			{
-				U32 i, last_i = ZSTD_REP_CHECK + (mlen != 1);
-				for (i = (mlen != 1); i < last_i; i++) {
-					const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (opt[cur].rep[0] - 1) : opt[cur].rep[i];
-					const U32 repIndex = (U32)(curr + cur - repCur);
-					const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
-					const BYTE *const repMatch = repBase + repIndex;
-					if ((repCur > 0 && repCur <= (S32)(curr + cur)) &&
-					    (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
-					    && (ZSTD_readMINMATCH(inr, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch))) {
-						/* repcode detected */
-						const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
-						mlen = (U32)ZSTD_count_2segments(inr + minMatch, repMatch + minMatch, iend, repEnd, prefixStart) + minMatch;
-
-						if (mlen > sufficient_len || cur + mlen >= ZSTD_OPT_NUM) {
-							best_mlen = mlen;
-							best_off = i;
-							last_pos = cur + 1;
-							goto _storeSequence;
-						}
-
-						best_off = i - (opt[cur].mlen != 1);
-						if (mlen > best_mlen)
-							best_mlen = mlen;
-
-						do {
-							if (opt[cur].mlen == 1) {
-								litlen = opt[cur].litlen;
-								if (cur > litlen) {
-									price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, inr - litlen,
-															best_off, mlen - MINMATCH, ultra);
-								} else
-									price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
-							} else {
-								litlen = 0;
-								price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, best_off, mlen - MINMATCH, ultra);
-							}
-
-							if (cur + mlen > last_pos || price <= opt[cur + mlen].price)
-								SET_PRICE(cur + mlen, mlen, i, litlen, price);
-							mlen--;
-						} while (mlen >= minMatch);
-					}
-				}
-			}
-
-			match_num = ZSTD_BtGetAllMatches_selectMLS_extDict(ctx, inr, iend, maxSearches, mls, matches, minMatch);
-
-			if (match_num > 0 && (matches[match_num - 1].len > sufficient_len || cur + matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
-				best_mlen = matches[match_num - 1].len;
-				best_off = matches[match_num - 1].off;
-				last_pos = cur + 1;
-				goto _storeSequence;
-			}
-
-			/* set prices using matches at position = cur */
-			for (u = 0; u < match_num; u++) {
-				mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
-				best_mlen = matches[u].len;
-
-				while (mlen <= best_mlen) {
-					if (opt[cur].mlen == 1) {
-						litlen = opt[cur].litlen;
-						if (cur > litlen)
-							price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, ip + cur - litlen,
-													matches[u].off - 1, mlen - MINMATCH, ultra);
-						else
-							price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
-					} else {
-						litlen = 0;
-						price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, matches[u].off - 1, mlen - MINMATCH, ultra);
-					}
-
-					if (cur + mlen > last_pos || (price < opt[cur + mlen].price))
-						SET_PRICE(cur + mlen, mlen, matches[u].off, litlen, price);
-
-					mlen++;
-				}
-			}
-		} /* for (cur = 1; cur <= last_pos; cur++) */
-
-		best_mlen = opt[last_pos].mlen;
-		best_off = opt[last_pos].off;
-		cur = last_pos - best_mlen;
-
-	/* store sequence */
-_storeSequence: /* cur, last_pos, best_mlen, best_off have to be set */
-		opt[0].mlen = 1;
-
-		while (1) {
-			mlen = opt[cur].mlen;
-			offset = opt[cur].off;
-			opt[cur].mlen = best_mlen;
-			opt[cur].off = best_off;
-			best_mlen = mlen;
-			best_off = offset;
-			if (mlen > cur)
-				break;
-			cur -= mlen;
-		}
-
-		for (u = 0; u <= last_pos;) {
-			u += opt[u].mlen;
-		}
-
-		for (cur = 0; cur < last_pos;) {
-			mlen = opt[cur].mlen;
-			if (mlen == 1) {
-				ip++;
-				cur++;
-				continue;
-			}
-			offset = opt[cur].off;
-			cur += mlen;
-			litLength = (U32)(ip - anchor);
-
-			if (offset > ZSTD_REP_MOVE_OPT) {
-				rep[2] = rep[1];
-				rep[1] = rep[0];
-				rep[0] = offset - ZSTD_REP_MOVE_OPT;
-				offset--;
-			} else {
-				if (offset != 0) {
-					best_off = (offset == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : (rep[offset]);
-					if (offset != 1)
-						rep[2] = rep[1];
-					rep[1] = rep[0];
-					rep[0] = best_off;
-				}
-
-				if (litLength == 0)
-					offset--;
-			}
-
-			ZSTD_updatePrice(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
-			ZSTD_storeSeq(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
-			anchor = ip = ip + mlen;
-		}
-	} /* for (cur=0; cur < last_pos; ) */
-
-	/* Save reps for next block */
-	{
-		int i;
-		for (i = 0; i < ZSTD_REP_NUM; i++)
-			ctx->repToConfirm[i] = rep[i];
-	}
-
-	/* Last Literals */
-	{
-		size_t lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-#endif /* ZSTD_OPT_H_91842398743 */
-- 
2.30.2




More information about the barebox mailing list