[RFC PATCH 07/17] netfs: Add a function to extract from an iter into a bvecq

[David Howells]

Add a function to extract a slice of data from an iterator of any type into
a bvec queue chain.

Signed-off-by: David Howells <dhowells at redhat.com>
cc: Paulo Alcantara <pc at manguebit.org>
cc: Matthew Wilcox <willy at infradead.org>
cc: Christoph Hellwig <hch at infradead.org>
cc: Steve French <sfrench at samba.org>
cc: linux-cifs at vger.kernel.org
cc: netfs at lists.linux.dev
cc: linux-fsdevel at vger.kernel.org
---
 fs/netfs/iterator.c   | 122 ++++++++++++++++++++++++++++++++++++++++++
 include/linux/netfs.h |   3 ++
 2 files changed, 125 insertions(+)

diff --git a/fs/netfs/iterator.c b/fs/netfs/iterator.c
index 72a435e5fc6d..faf4f0a3b33d 100644
--- a/fs/netfs/iterator.c
+++ b/fs/netfs/iterator.c
@@ -13,6 +13,128 @@
 #include <linux/netfs.h>
 #include "internal.h"
 
+/**
+ * netfs_extract_iter - Extract the pages from an iterator into a bvecq
+ * @orig: The original iterator
+ * @orig_len: The amount of iterator to copy
+ * @max_segs: Maximum number of contiguous segments
+ * @fpos: Starting file position to label the bvecq with
+ * @_bvecq_head: Where to cache the bvec queue
+ * @extraction_flags: Flags to qualify the request
+ *
+ * Extract the page fragments from the given amount of the source iterator and
+ * build bvec queue that refers to all of those bits.  This allows the original
+ * iterator to disposed of.
+ *
+ * @extraction_flags can have ITER_ALLOW_P2PDMA set to request peer-to-peer DMA be
+ * allowed on the pages extracted.
+ *
+ * On success, the amount of data in the bvec is returned, the original
+ * iterator will have been advanced by the amount extracted.
+ *
+ * The bvecq segments are marked with indications on how to get clean up the
+ * extracted fragments.
+ */
+ssize_t netfs_extract_iter(struct iov_iter *orig, size_t orig_len, size_t max_segs,
+			   unsigned long long fpos, struct bvecq **_bvecq_head,
+			   iov_iter_extraction_t extraction_flags)
+{
+	struct bvecq *bq_tail = NULL;
+	ssize_t ret = 0;
+	size_t segs_per_bq;
+	size_t extracted = 0;
+
+	_enter("{%u,%zx},%zx", orig->iter_type, orig->count, orig_len);
+
+	if (max_segs == 0)
+		max_segs = ULONG_MAX;
+
+	/* We want the biggest pow-of-2 size that has at most 255 segs and that
+	 * won't exceed a 4K page.
+	 */
+	segs_per_bq = (4096 - sizeof(*bq_tail)) / sizeof(bq_tail->__bv[0]);
+	if (segs_per_bq > 255)
+		segs_per_bq = (2048 - sizeof(*bq_tail)) / sizeof(bq_tail->__bv[0]);
+
+	do {
+		struct bvecq *bq;
+		size_t nr_slots = iov_iter_npages(orig, umin(segs_per_bq, max_segs));
+
+		if (WARN_ON(nr_slots == 0 && extracted < orig_len) ||
+		    WARN_ON(nr_slots > max_segs))
+			break;
+		max_segs -= nr_slots;
+
+		bq = netfs_alloc_one_bvecq(nr_slots, GFP_NOFS);
+		if (!bq) {
+			ret = -ENOMEM;
+			break;
+		}
+		bq->free	= user_backed_iter(orig);
+		bq->unpin	= iov_iter_extract_will_pin(orig);
+		bq->prev	= bq_tail;
+		bq->fpos	= fpos + extracted;
+
+		if (bq_tail)
+			bq_tail->next = bq;
+		else
+			*_bvecq_head = bq;
+		bq_tail = bq;
+
+		if (extracted >= orig_len)
+			break;
+
+		/* Put the page list at the end of the bvec list storage.  bvec
+		 * elements are larger than page pointers, so as long as we
+		 * work 0->last, we should be fine.
+		 */
+		struct bio_vec *bv = bq->bv;
+		struct page **pages;
+		size_t bv_size = array_size(bq->max_segs, sizeof(*bv));
+		size_t pg_size = array_size(bq->max_segs, sizeof(*pages));
+
+		pages = (void *)bv + bv_size - pg_size;
+
+		do {
+			unsigned int cur_npages;
+			ssize_t got;
+			size_t offset;
+
+			got = iov_iter_extract_pages(orig, &pages, orig_len - extracted,
+						     bq->max_segs - bq->nr_segs,
+						     extraction_flags, &offset);
+			if (got < 0) {
+				pr_err("Couldn't get user pages (rc=%zd)\n", got);
+				ret = got;
+				break;
+			}
+
+			if (got > orig_len - extracted) {
+				pr_err("get_pages rc=%zd more than %zu\n",
+				       got, orig_len - extracted);
+				break;
+			}
+
+			extracted += got;
+			got += offset;
+			cur_npages = DIV_ROUND_UP(got, PAGE_SIZE);
+
+			for (unsigned int i = 0; i < cur_npages; i++) {
+				size_t len = umin(got, PAGE_SIZE);
+
+				bvec_set_page(&bq->bv[bq->nr_segs],
+					      *pages++, len - offset, offset);
+				bq->nr_segs++;
+				got -= len;
+				offset = 0;
+			}
+		} while (extracted < orig_len && !bvecq_is_full(bq));
+	} while (extracted < orig_len && max_segs > 0);
+
+	return extracted ?: ret;
+}
+EXPORT_SYMBOL_GPL(netfs_extract_iter);
+
 /**
  * netfs_extract_user_iter - Extract the pages from a user iterator into a bvec
  * @orig: The original iterator
diff --git a/include/linux/netfs.h b/include/linux/netfs.h
index f9ad067a0a0c..b146aeaaf6c9 100644
--- a/include/linux/netfs.h
+++ b/include/linux/netfs.h
@@ -448,6 +448,9 @@ void netfs_get_subrequest(struct netfs_io_subrequest *subreq,
 			  enum netfs_sreq_ref_trace what);
 void netfs_put_subrequest(struct netfs_io_subrequest *subreq,
 			  enum netfs_sreq_ref_trace what);
+ssize_t netfs_extract_iter(struct iov_iter *orig, size_t orig_len, size_t max_segs,
+			   unsigned long long fpos, struct bvecq **_bvecq_head,
+			   iov_iter_extraction_t extraction_flags);
 ssize_t netfs_extract_user_iter(struct iov_iter *orig, size_t orig_len,
 				struct iov_iter *new,
 				iov_iter_extraction_t extraction_flags);