[RFC PATCH 9/9] arm64: implement dynamic shadow call stack for GCC
Ard Biesheuvel
ardb at kernel.org
Wed Oct 13 08:22:43 PDT 2021
Implement support for the shadow call stack on GCC, and in a dynamic
manner, by parsing the unwind tables at init time to locate all
occurrences of PACIASP/AUTIASP, and replacing them with the shadow call
stack push and pop instructions, respectively.
This is useful because the overhead of the shadow call stack is
difficult to justify on hardware that implements pointer authentication
(PAC), and given that the PAC instructions are executed as NOPs on
hardware that doesn't, we can just replace them.
This patch only implements this for the core kernel, but the logic can
be reused for modules without much trouble.
Signed-off-by: Ard Biesheuvel <ardb at kernel.org>
---
Makefile | 4 +-
arch/Kconfig | 4 +-
arch/arm64/Kconfig | 8 +-
arch/arm64/kernel/Makefile | 2 +
arch/arm64/kernel/head.S | 3 +
arch/arm64/kernel/patch-scs.c | 223 ++++++++++++++++++++
6 files changed, 239 insertions(+), 5 deletions(-)
diff --git a/Makefile b/Makefile
index 7cfe4ff36f44..2d94fed93d9d 100644
--- a/Makefile
+++ b/Makefile
@@ -933,8 +933,8 @@ LDFLAGS_vmlinux += --gc-sections
endif
ifdef CONFIG_SHADOW_CALL_STACK
-CC_FLAGS_SCS := -fsanitize=shadow-call-stack
-KBUILD_CFLAGS += $(CC_FLAGS_SCS)
+CC_FLAGS_SCS-$(CONFIG_CC_IS_CLANG) := -fsanitize=shadow-call-stack
+KBUILD_CFLAGS += $(CC_FLAGS_SCS-y)
export CC_FLAGS_SCS
endif
diff --git a/arch/Kconfig b/arch/Kconfig
index 8df1c7102643..21eeec66bf4c 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -596,8 +596,8 @@ config ARCH_SUPPORTS_SHADOW_CALL_STACK
switching.
config SHADOW_CALL_STACK
- bool "Clang Shadow Call Stack"
- depends on CC_IS_CLANG && ARCH_SUPPORTS_SHADOW_CALL_STACK
+ bool "Shadow Call Stack"
+ depends on ARCH_SUPPORTS_SHADOW_CALL_STACK
depends on DYNAMIC_FTRACE_WITH_REGS || !FUNCTION_GRAPH_TRACER
help
This option enables Clang's Shadow Call Stack, which uses a
diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index 742baca09343..6d74822fd386 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -81,7 +81,7 @@ config ARM64
select ARCH_SUPPORTS_DEBUG_PAGEALLOC
select ARCH_SUPPORTS_HUGETLBFS
select ARCH_SUPPORTS_MEMORY_FAILURE
- select ARCH_SUPPORTS_SHADOW_CALL_STACK if CC_HAVE_SHADOW_CALL_STACK
+ select ARCH_SUPPORTS_SHADOW_CALL_STACK if CC_HAVE_SHADOW_CALL_STACK || CC_IS_GCC
select ARCH_SUPPORTS_LTO_CLANG if CPU_LITTLE_ENDIAN
select ARCH_SUPPORTS_LTO_CLANG_THIN
select ARCH_SUPPORTS_CFI_CLANG
@@ -353,6 +353,12 @@ config KASAN_SHADOW_OFFSET
config UNWIND_TABLES
bool
+config UNWIND_PATCH_PAC_INTO_SCS
+ def_bool y
+ depends on CC_IS_GCC && SHADOW_CALL_STACK
+ select UNWIND_TABLES
+ select ARM64_PTR_AUTH_KERNEL
+
source "arch/arm64/Kconfig.platforms"
menu "Kernel Features"
diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index 3f1490bfb938..42b9bd92d51e 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -73,6 +73,8 @@ obj-$(CONFIG_ARM64_PTR_AUTH) += pointer_auth.o
obj-$(CONFIG_ARM64_MTE) += mte.o
obj-y += vdso-wrap.o
obj-$(CONFIG_COMPAT_VDSO) += vdso32-wrap.o
+obj-$(CONFIG_UNWIND_PATCH_PAC_INTO_SCS) += patch-scs.o
+CFLAGS_patch-scs.o += -mbranch-protection=none
obj-y += probes/
head-y := head.o
diff --git a/arch/arm64/kernel/head.S b/arch/arm64/kernel/head.S
index 17962452e31d..5d50d212d3ae 100644
--- a/arch/arm64/kernel/head.S
+++ b/arch/arm64/kernel/head.S
@@ -447,6 +447,9 @@ SYM_FUNC_START_LOCAL(__primary_switched)
bl __pi_memset
dsb ishst // Make zero page visible to PTW
+#ifdef CONFIG_UNWIND_PATCH_PAC_INTO_SCS
+ bl scs_patch_vmlinux
+#endif
#if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS)
bl kasan_early_init
#endif
diff --git a/arch/arm64/kernel/patch-scs.c b/arch/arm64/kernel/patch-scs.c
new file mode 100644
index 000000000000..878a40060550
--- /dev/null
+++ b/arch/arm64/kernel/patch-scs.c
@@ -0,0 +1,223 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2021 - Google LLC
+ * Author: Ard Biesheuvel <ardb at google.com>
+ */
+
+#include <linux/bug.h>
+#include <linux/init.h>
+#include <linux/linkage.h>
+#include <linux/printk.h>
+#include <linux/types.h>
+
+#define DW_CFA_nop 0x00
+#define DW_CFA_set_loc 0x01
+#define DW_CFA_advance_loc1 0x02
+#define DW_CFA_advance_loc2 0x03
+#define DW_CFA_advance_loc4 0x04
+#define DW_CFA_offset_extended 0x05
+#define DW_CFA_restore_extended 0x06
+#define DW_CFA_undefined 0x07
+#define DW_CFA_same_value 0x08
+#define DW_CFA_register 0x09
+#define DW_CFA_remember_state 0x0a
+#define DW_CFA_restore_state 0x0b
+#define DW_CFA_def_cfa 0x0c
+#define DW_CFA_def_cfa_register 0x0d
+#define DW_CFA_def_cfa_offset 0x0e
+#define DW_CFA_def_cfa_expression 0x0f
+#define DW_CFA_expression 0x10
+#define DW_CFA_offset_extended_sf 0x11
+#define DW_CFA_def_cfa_sf 0x12
+#define DW_CFA_def_cfa_offset_sf 0x13
+#define DW_CFA_val_offset 0x14
+#define DW_CFA_val_offset_sf 0x15
+#define DW_CFA_val_expression 0x16
+#define DW_CFA_lo_user 0x1c
+#define DW_CFA_negate_ra_state 0x2d
+#define DW_CFA_GNU_args_size 0x2e
+#define DW_CFA_GNU_negative_offset_extended 0x2f
+#define DW_CFA_hi_user 0x3f
+
+static unsigned long get_uleb128(const u8 **pcur, const u8 *end)
+{
+ const u8 *cur = *pcur;
+ unsigned long value;
+ unsigned int shift;
+
+ for (shift = 0, value = 0; cur < end; shift += 7) {
+ if (shift + 7 > 8 * sizeof(value)
+ && (*cur & 0x7fU) >= (1U << (8 * sizeof(value) - shift))) {
+ cur = end + 1;
+ break;
+ }
+ value |= (unsigned long) (*cur & 0x7f) << shift;
+ if (!(*cur++ & 0x80))
+ break;
+ }
+ *pcur = cur;
+
+ return value;
+}
+
+extern const u8 __eh_frame_start[], __eh_frame_end[];
+
+struct fde_frame {
+ s32 initial_loc;
+ s32 range;
+};
+
+static int scs_patch_loc(u64 loc)
+{
+ u32 insn = le32_to_cpup((void *)loc);
+
+ /*
+ * Sometimes, the unwind data appears to be out of sync, and associates
+ * the DW_CFA_negate_ra_state directive with the ret instruction
+ * following the autiasp, rather than the autiasp itself.
+ */
+ if (insn == 0xd65f03c0) { // ret
+ loc -= 4;
+ insn = le32_to_cpup((void *)loc);
+ }
+
+ switch (insn) {
+ case 0xd503233f: // paciasp
+ *(u32 *)loc = cpu_to_le32(0xf800865e);
+ break;
+ case 0xd50323bf: // autiasp
+ *(u32 *)loc = cpu_to_le32(0xf85f8e5e);
+ break;
+ default:
+ // ignore
+ break;
+ }
+ return 0;
+}
+
+static int noinstr scs_handle_frame(const u8 eh_frame[], u32 size)
+{
+ const struct fde_frame *fde;
+ const u8 *opcode;
+ u64 loc;
+
+ /*
+ * For patching PAC opcodes, we only care about the FDE records, and
+ * not the CIE, which carries the initial CFA directives but they only
+ * pertain to which register is the stack pointer.
+ * TODO this is not 100% true - we need the augmentation string and the
+ * encoding but they are always the same in practice.
+ */
+ if (*(u32 *)eh_frame == 0)
+ return 0;
+
+ fde = (const struct fde_frame *)(eh_frame + 4);
+ loc = (u64)offset_to_ptr(&fde->initial_loc);
+ opcode = (const u8 *)(fde + 1);
+
+ // TODO check augmentation data
+ WARN_ON(*opcode++);
+ size -= sizeof(u32) + sizeof(*fde) + 1;
+
+ /*
+ * Starting from 'loc', apply the CFA opcodes that advance the location
+ * pointer, and identify the locations of the PAC instructions.
+ */
+ do {
+ const u8 *end;
+
+ switch (*opcode & 0xC0) {
+ case 0:
+ // handle DW_CFA_xxx opcodes
+ switch (*opcode) {
+ int ret;
+
+ case DW_CFA_nop:
+ case DW_CFA_remember_state:
+ case DW_CFA_restore_state:
+ break;
+
+ case DW_CFA_advance_loc1:
+ loc += 4 * *++opcode;
+ size--;
+ break;
+
+ case DW_CFA_advance_loc2:
+ loc += 4 * *++opcode;
+ loc += 4 * *++opcode << 8;
+ size -= 2;
+ break;
+
+ case DW_CFA_def_cfa:
+ case DW_CFA_def_cfa_offset:
+ case DW_CFA_def_cfa_register:
+ opcode++;
+ size--;
+ end = opcode + size;
+ get_uleb128(&opcode, end);
+ size = end - opcode;
+ continue;
+
+ case DW_CFA_negate_ra_state:
+ // patch paciasp/autiasp into shadow stack push/pop
+ ret = scs_patch_loc(loc - 4);
+ if (ret)
+ return ret;
+ break;
+
+ default:
+ pr_debug("unhandled opcode: %02x\n", *opcode);
+ return -ENOEXEC;
+ }
+ opcode++;
+ size--;
+ break;
+
+ case 0x40:
+ // advance loc
+ loc += (*opcode++ & 0x3f) * 4;
+ size--;
+ break;
+
+ case 0x80:
+ opcode++;
+ size--;
+ end = opcode + size;
+ get_uleb128(&opcode, end);
+ size = end - opcode;
+ continue;
+
+ default:
+ // ignore
+ opcode++;
+ size--;
+ break;
+ }
+ } while (size > 0);
+
+ return 0;
+}
+
+int noinstr scs_patch(const u8 eh_frame[], int size)
+{
+ const u8 *p = eh_frame;
+
+ while (size > 4) {
+ const u32 *frame_size = (const u32 *)p;
+ int ret;
+
+ if (*frame_size != -1 && *frame_size <= size) {
+ ret = scs_handle_frame(p + 4, *frame_size);
+ if (ret)
+ return ret;
+ p += 4 + *frame_size;
+ size -= 4 + *frame_size;
+ }
+ }
+ return 0;
+}
+
+asmlinkage int noinstr scs_patch_vmlinux(void)
+{
+ return scs_patch(__eh_frame_start, __eh_frame_end - __eh_frame_start);
+}
--
2.30.2
More information about the linux-arm-kernel
mailing list