[v2] kasan: RISC-V support for KASAN_SW_TAGS using pointer masking

[PATCH v2 0/9] kasan: RISC-V support for KASAN_SW_TAGS using pointer masking

Posted by Samuel Holland 1 month ago

This series implements support for software tag-based KASAN using the
RISC-V pointer masking extension[1], which supports 7 and/or 16-bit
tags. This implementation uses 7-bit tags, so it is compatible with
either hardware mode. Patch 4 adds supports for KASAN_SW_TAGS with tag
widths other than 8 bits.

Pointer masking is an optional ISA extension, and it must be enabled
using an SBI call to firmware on each CPU. If the SBI call fails on the
boot CPU, KASAN is globally disabled. Patch 2 adds support for boot-time
disabling of KASAN_SW_TAGS, and patch 3 adds support for runtime control
of stack tagging.

Patch 1 is an optimization that could be applied separately. It is
included here because it affects the selection of KASAN_SHADOW_OFFSET.

This implementation currently passes the KASAN KUnit test suite:

  # kasan: pass:64 fail:0 skip:9 total:73
  # Totals: pass:64 fail:0 skip:9 total:73
  ok 1 kasan

One workaround is required to pass the vmalloc_percpu test. I have to
shrink the initial percpu area to force the use of a KASAN-tagged percpu
area in the test (depending on .config, this workaround is also needed
on arm64 without this series applied, so it is not a new issue):

diff --git a/include/linux/percpu.h b/include/linux/percpu.h
index b6321fc49159..26b97c79ad7c 100644
--- a/include/linux/percpu.h
+++ b/include/linux/percpu.h
@@ -43,7 +43,7 @@
 #ifdef CONFIG_RANDOM_KMALLOC_CACHES
 #define PERCPU_DYNAMIC_SIZE_SHIFT      12
 #else
-#define PERCPU_DYNAMIC_SIZE_SHIFT      10
+#define PERCPU_DYNAMIC_SIZE_SHIFT      8
 #endif

When running with hardware or firmware that doesn't support pointer
masking, the kernel still boots successfully:

  kasan: test: Can't run KASAN tests with KASAN disabled
      # kasan:     # failed to initialize (-1)
  not ok 1 kasan

This series can be tested by applying patch series to LLVM[2] and
QEMU[3], and using the master branch of OpenSBI[4].

[1]: https://github.com/riscv/riscv-j-extension/raw/d70011dde6c2/zjpm-spec.pdf
[2]: https://github.com/SiFiveHolland/llvm-project/commits/up/riscv64-kernel-hwasan
[3]: https://lore.kernel.org/qemu-devel/20240511101053.1875596-1-me@deliversmonkey.space/
[4]: https://github.com/riscv-software-src/opensbi/commit/1cb234b1c9ed

Changes in v2:
 - Improve the explanation for how KASAN_SHADOW_END is derived
 - Update the range check in kasan_non_canonical_hook()
 - Split the generic and RISC-V parts of stack tag generation control
   to avoid breaking bisectability
 - Add a patch to call kasan_non_canonical_hook() on riscv
 - Fix build error with KASAN_GENERIC
 - Use symbolic definitons for SBI firmware features call
 - Update indentation in scripts/Makefile.kasan
 - Use kasan_params to set hwasan-generate-tags-with-calls=1

Clément Léger (1):
  riscv: Add SBI Firmware Features extension definitions

Samuel Holland (8):
  kasan: sw_tags: Use arithmetic shift for shadow computation
  kasan: sw_tags: Check kasan_flag_enabled at runtime
  kasan: sw_tags: Support outline stack tag generation
  kasan: sw_tags: Support tag widths less than 8 bits
  riscv: mm: Log potential KASAN shadow alias
  riscv: Do not rely on KASAN to define the memory layout
  riscv: Align the sv39 linear map to 16 GiB
  riscv: Implement KASAN_SW_TAGS

 Documentation/arch/riscv/vm-layout.rst | 10 ++---
 Documentation/dev-tools/kasan.rst      | 14 +++---
 arch/arm64/Kconfig                     | 10 ++---
 arch/arm64/include/asm/kasan.h         |  6 ++-
 arch/arm64/include/asm/memory.h        | 17 ++++++-
 arch/arm64/include/asm/uaccess.h       |  1 +
 arch/arm64/mm/kasan_init.c             |  7 ++-
 arch/riscv/Kconfig                     |  4 +-
 arch/riscv/include/asm/cache.h         |  4 ++
 arch/riscv/include/asm/kasan.h         | 29 +++++++++++-
 arch/riscv/include/asm/page.h          | 21 +++++++--
 arch/riscv/include/asm/pgtable.h       |  6 +++
 arch/riscv/include/asm/sbi.h           | 28 ++++++++++++
 arch/riscv/include/asm/tlbflush.h      |  4 +-
 arch/riscv/kernel/setup.c              |  6 +++
 arch/riscv/kernel/smpboot.c            |  8 +++-
 arch/riscv/lib/Makefile                |  2 +
 arch/riscv/lib/kasan_sw_tags.S         | 61 ++++++++++++++++++++++++++
 arch/riscv/mm/fault.c                  |  3 ++
 arch/riscv/mm/init.c                   |  2 +-
 arch/riscv/mm/kasan_init.c             | 32 +++++++++++++-
 arch/riscv/mm/physaddr.c               |  4 ++
 include/linux/kasan-enabled.h          | 15 +++----
 include/linux/kasan-tags.h             | 13 +++---
 include/linux/kasan.h                  | 10 ++++-
 mm/kasan/hw_tags.c                     | 10 -----
 mm/kasan/kasan.h                       |  2 +
 mm/kasan/report.c                      | 22 ++++++++--
 mm/kasan/sw_tags.c                     |  9 ++++
 mm/kasan/tags.c                        | 10 +++++
 scripts/Makefile.kasan                 |  5 +++
 scripts/gdb/linux/mm.py                |  5 ++-
 32 files changed, 313 insertions(+), 67 deletions(-)
 create mode 100644 arch/riscv/lib/kasan_sw_tags.S

-- 
2.45.1

[PATCH v2 1/9] kasan: sw_tags: Use arithmetic shift for shadow computation

Posted by Samuel Holland 1 month ago

Currently, kasan_mem_to_shadow() uses a logical right shift, which turns
canonical kernel addresses into non-canonical addresses by clearing the
high KASAN_SHADOW_SCALE_SHIFT bits. The value of KASAN_SHADOW_OFFSET is
then chosen so that the addition results in a canonical address for the
shadow memory.

For KASAN_GENERIC, this shift/add combination is ABI with the compiler,
because KASAN_SHADOW_OFFSET is used in compiler-generated inline tag
checks[1], which must only attempt to dereference canonical addresses.

However, for KASAN_SW_TAGS we have some freedom to change the algorithm
without breaking the ABI. Because TBI is enabled for kernel addresses,
the top bits of shadow memory addresses computed during tag checks are
irrelevant, and so likewise are the top bits of KASAN_SHADOW_OFFSET.
This is demonstrated by the fact that LLVM uses a logical right shift
in the tag check fast path[2] but a sbfx (signed bitfield extract)
instruction in the slow path[3] without causing any issues.

Using an arithmetic shift in kasan_mem_to_shadow() provides a number of
benefits:

1) The memory layout is easier to understand. KASAN_SHADOW_OFFSET
becomes a canonical memory address, and the shifted pointer becomes a
negative offset, so KASAN_SHADOW_OFFSET == KASAN_SHADOW_END regardless
of the shift amount or the size of the virtual address space.

2) KASAN_SHADOW_OFFSET becomes a simpler constant, requiring only one
instruction to load instead of two. Since it must be loaded in each
function with a tag check, this decreases kernel text size by 0.5%.

3) This shift and the sign extension from kasan_reset_tag() can be
combined into a single sbfx instruction. When this same algorithm change
is applied to the compiler, it removes an instruction from each inline
tag check, further reducing kernel text size by an additional 4.6%.

These benefits extend to other architectures as well. On RISC-V, where
the baseline ISA does not shifted addition or have an equivalent to the
sbfx instruction, loading KASAN_SHADOW_OFFSET is reduced from 3 to 2
instructions, and kasan_mem_to_shadow(kasan_reset_tag(addr)) similarly
combines two consecutive right shifts.

Link: https://github.com/llvm/llvm-project/blob/llvmorg-20-init/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp#L1316 [1]
Link: https://github.com/llvm/llvm-project/blob/llvmorg-20-init/llvm/lib/Transforms/Instrumentation/HWAddressSanitizer.cpp#L895 [2]
Link: https://github.com/llvm/llvm-project/blob/llvmorg-20-init/llvm/lib/Target/AArch64/AArch64AsmPrinter.cpp#L669 [3]
Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

Changes in v2:
 - Improve the explanation for how KASAN_SHADOW_END is derived
 - Update the range check in kasan_non_canonical_hook()

 arch/arm64/Kconfig              | 10 +++++-----
 arch/arm64/include/asm/memory.h | 17 +++++++++++++++--
 arch/arm64/mm/kasan_init.c      |  7 +++++--
 include/linux/kasan.h           | 10 ++++++++--
 mm/kasan/report.c               | 22 ++++++++++++++++++----
 scripts/gdb/linux/mm.py         |  5 +++--
 6 files changed, 54 insertions(+), 17 deletions(-)

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index fd9df6dcc593..6a326908c941 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -418,11 +418,11 @@ config KASAN_SHADOW_OFFSET
 	default 0xdffffe0000000000 if ARM64_VA_BITS_42 && !KASAN_SW_TAGS
 	default 0xdfffffc000000000 if ARM64_VA_BITS_39 && !KASAN_SW_TAGS
 	default 0xdffffff800000000 if ARM64_VA_BITS_36 && !KASAN_SW_TAGS
-	default 0xefff800000000000 if (ARM64_VA_BITS_48 || (ARM64_VA_BITS_52 && !ARM64_16K_PAGES)) && KASAN_SW_TAGS
-	default 0xefffc00000000000 if (ARM64_VA_BITS_47 || ARM64_VA_BITS_52) && ARM64_16K_PAGES && KASAN_SW_TAGS
-	default 0xeffffe0000000000 if ARM64_VA_BITS_42 && KASAN_SW_TAGS
-	default 0xefffffc000000000 if ARM64_VA_BITS_39 && KASAN_SW_TAGS
-	default 0xeffffff800000000 if ARM64_VA_BITS_36 && KASAN_SW_TAGS
+	default 0xffff800000000000 if (ARM64_VA_BITS_48 || (ARM64_VA_BITS_52 && !ARM64_16K_PAGES)) && KASAN_SW_TAGS
+	default 0xffffc00000000000 if (ARM64_VA_BITS_47 || ARM64_VA_BITS_52) && ARM64_16K_PAGES && KASAN_SW_TAGS
+	default 0xfffffe0000000000 if ARM64_VA_BITS_42 && KASAN_SW_TAGS
+	default 0xffffffc000000000 if ARM64_VA_BITS_39 && KASAN_SW_TAGS
+	default 0xfffffff800000000 if ARM64_VA_BITS_36 && KASAN_SW_TAGS
 	default 0xffffffffffffffff
 
 config UNWIND_TABLES
diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h
index 0480c61dbb4f..a93fc9dc16f3 100644
--- a/arch/arm64/include/asm/memory.h
+++ b/arch/arm64/include/asm/memory.h
@@ -80,7 +80,8 @@
  * where KASAN_SHADOW_SCALE_SHIFT is the order of the number of bits that map
  * to a single shadow byte and KASAN_SHADOW_OFFSET is a constant that offsets
  * the mapping. Note that KASAN_SHADOW_OFFSET does not point to the start of
- * the shadow memory region.
+ * the shadow memory region, since not all possible addresses have shadow
+ * memory allocated for them.
  *
  * Based on this mapping, we define two constants:
  *
@@ -89,7 +90,15 @@
  *
  * KASAN_SHADOW_END is defined first as the shadow address that corresponds to
  * the upper bound of possible virtual kernel memory addresses UL(1) << 64
- * according to the mapping formula.
+ * according to the mapping formula. For Generic KASAN, the address in the
+ * mapping formula is treated as unsigned (part of the compiler's ABI), so the
+ * end of the shadow memory region is at a large positive offset from
+ * KASAN_SHADOW_OFFSET. For Software Tag-Based KASAN, the address in the
+ * formula is treated as signed. Since all kernel addresses are negative, they
+ * map to shadow memory below KASAN_SHADOW_OFFSET, making KASAN_SHADOW_OFFSET
+ * itself the end of the shadow memory region. (User pointers are positive and
+ * would map to shadow memory above KASAN_SHADOW_OFFSET, but shadow memory is
+ * not allocated for them.)
  *
  * KASAN_SHADOW_START is defined second based on KASAN_SHADOW_END. The shadow
  * memory start must map to the lowest possible kernel virtual memory address
@@ -100,7 +109,11 @@
  */
 #if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS)
 #define KASAN_SHADOW_OFFSET	_AC(CONFIG_KASAN_SHADOW_OFFSET, UL)
+#ifdef CONFIG_KASAN_GENERIC
 #define KASAN_SHADOW_END	((UL(1) << (64 - KASAN_SHADOW_SCALE_SHIFT)) + KASAN_SHADOW_OFFSET)
+#else
+#define KASAN_SHADOW_END	KASAN_SHADOW_OFFSET
+#endif
 #define _KASAN_SHADOW_START(va)	(KASAN_SHADOW_END - (UL(1) << ((va) - KASAN_SHADOW_SCALE_SHIFT)))
 #define KASAN_SHADOW_START	_KASAN_SHADOW_START(vabits_actual)
 #define PAGE_END		KASAN_SHADOW_START
diff --git a/arch/arm64/mm/kasan_init.c b/arch/arm64/mm/kasan_init.c
index b65a29440a0c..6836e571555c 100644
--- a/arch/arm64/mm/kasan_init.c
+++ b/arch/arm64/mm/kasan_init.c
@@ -198,8 +198,11 @@ static bool __init root_level_aligned(u64 addr)
 /* The early shadow maps everything to a single page of zeroes */
 asmlinkage void __init kasan_early_init(void)
 {
-	BUILD_BUG_ON(KASAN_SHADOW_OFFSET !=
-		KASAN_SHADOW_END - (1UL << (64 - KASAN_SHADOW_SCALE_SHIFT)));
+	if (IS_ENABLED(CONFIG_KASAN_GENERIC))
+		BUILD_BUG_ON(KASAN_SHADOW_OFFSET !=
+			KASAN_SHADOW_END - (1UL << (64 - KASAN_SHADOW_SCALE_SHIFT)));
+	else
+		BUILD_BUG_ON(KASAN_SHADOW_OFFSET != KASAN_SHADOW_END);
 	BUILD_BUG_ON(!IS_ALIGNED(_KASAN_SHADOW_START(VA_BITS), SHADOW_ALIGN));
 	BUILD_BUG_ON(!IS_ALIGNED(_KASAN_SHADOW_START(VA_BITS_MIN), SHADOW_ALIGN));
 	BUILD_BUG_ON(!IS_ALIGNED(KASAN_SHADOW_END, SHADOW_ALIGN));
diff --git a/include/linux/kasan.h b/include/linux/kasan.h
index 00a3bf7c0d8f..03b440658817 100644
--- a/include/linux/kasan.h
+++ b/include/linux/kasan.h
@@ -58,8 +58,14 @@ int kasan_populate_early_shadow(const void *shadow_start,
 #ifndef kasan_mem_to_shadow
 static inline void *kasan_mem_to_shadow(const void *addr)
 {
-	return (void *)((unsigned long)addr >> KASAN_SHADOW_SCALE_SHIFT)
-		+ KASAN_SHADOW_OFFSET;
+	void *scaled;
+
+	if (IS_ENABLED(CONFIG_KASAN_GENERIC))
+		scaled = (void *)((unsigned long)addr >> KASAN_SHADOW_SCALE_SHIFT);
+	else
+		scaled = (void *)((long)addr >> KASAN_SHADOW_SCALE_SHIFT);
+
+	return KASAN_SHADOW_OFFSET + scaled;
 }
 #endif
 
diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index b48c768acc84..c08097715686 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c
@@ -644,15 +644,29 @@ void kasan_report_async(void)
  */
 void kasan_non_canonical_hook(unsigned long addr)
 {
+	unsigned long max_shadow_size = BIT(BITS_PER_LONG - KASAN_SHADOW_SCALE_SHIFT);
 	unsigned long orig_addr;
 	const char *bug_type;
 
 	/*
-	 * All addresses that came as a result of the memory-to-shadow mapping
-	 * (even for bogus pointers) must be >= KASAN_SHADOW_OFFSET.
+	 * With the default kasan_mem_to_shadow() algorithm, all addresses
+	 * returned by the memory-to-shadow mapping (even for bogus pointers)
+	 * must be within a certain displacement from KASAN_SHADOW_OFFSET.
+	 *
+	 * For Generic KASAN, the displacement is unsigned, so
+	 * KASAN_SHADOW_OFFSET is the smallest possible shadow address. For
+	 * Software Tag-Based KASAN, the displacement is signed, so
+	 * KASAN_SHADOW_OFFSET is the center of the range.
 	 */
-	if (addr < KASAN_SHADOW_OFFSET)
-		return;
+	if (IS_ENABLED(CONFIG_KASAN_GENERIC)) {
+		if (addr < KASAN_SHADOW_OFFSET ||
+		    addr >= KASAN_SHADOW_OFFSET + max_shadow_size)
+			return;
+	} else {
+		if (addr < KASAN_SHADOW_OFFSET - max_shadow_size / 2 ||
+		    addr >= KASAN_SHADOW_OFFSET + max_shadow_size / 2)
+			return;
+	}
 
 	orig_addr = (unsigned long)kasan_shadow_to_mem((void *)addr);
 
diff --git a/scripts/gdb/linux/mm.py b/scripts/gdb/linux/mm.py
index 7571aebbe650..2e63f3dedd53 100644
--- a/scripts/gdb/linux/mm.py
+++ b/scripts/gdb/linux/mm.py
@@ -110,12 +110,13 @@ class aarch64_page_ops():
         self.KERNEL_END = gdb.parse_and_eval("_end")
 
         if constants.LX_CONFIG_KASAN_GENERIC or constants.LX_CONFIG_KASAN_SW_TAGS:
+            self.KASAN_SHADOW_OFFSET = constants.LX_CONFIG_KASAN_SHADOW_OFFSET
             if constants.LX_CONFIG_KASAN_GENERIC:
                 self.KASAN_SHADOW_SCALE_SHIFT = 3
+                self.KASAN_SHADOW_END = (1 << (64 - self.KASAN_SHADOW_SCALE_SHIFT)) + self.KASAN_SHADOW_OFFSET
             else:
                 self.KASAN_SHADOW_SCALE_SHIFT = 4
-            self.KASAN_SHADOW_OFFSET = constants.LX_CONFIG_KASAN_SHADOW_OFFSET
-            self.KASAN_SHADOW_END = (1 << (64 - self.KASAN_SHADOW_SCALE_SHIFT)) + self.KASAN_SHADOW_OFFSET
+                self.KASAN_SHADOW_END = self.KASAN_SHADOW_OFFSET
             self.PAGE_END = self.KASAN_SHADOW_END - (1 << (self.vabits_actual - self.KASAN_SHADOW_SCALE_SHIFT))
         else:
             self.PAGE_END = self._PAGE_END(self.VA_BITS_MIN)
-- 
2.45.1

Re: [PATCH v2 1/9] kasan: sw_tags: Use arithmetic shift for shadow computation

Posted by Andrey Konovalov 1 month ago

On Tue, Oct 22, 2024 at 3:59 AM Samuel Holland
<samuel.holland@sifive.com> wrote:
>
> Currently, kasan_mem_to_shadow() uses a logical right shift, which turns
> canonical kernel addresses into non-canonical addresses by clearing the
> high KASAN_SHADOW_SCALE_SHIFT bits. The value of KASAN_SHADOW_OFFSET is
> then chosen so that the addition results in a canonical address for the
> shadow memory.
>
> For KASAN_GENERIC, this shift/add combination is ABI with the compiler,
> because KASAN_SHADOW_OFFSET is used in compiler-generated inline tag
> checks[1], which must only attempt to dereference canonical addresses.
>
> However, for KASAN_SW_TAGS we have some freedom to change the algorithm
> without breaking the ABI. Because TBI is enabled for kernel addresses,
> the top bits of shadow memory addresses computed during tag checks are
> irrelevant, and so likewise are the top bits of KASAN_SHADOW_OFFSET.
> This is demonstrated by the fact that LLVM uses a logical right shift
> in the tag check fast path[2] but a sbfx (signed bitfield extract)
> instruction in the slow path[3] without causing any issues.
>
> Using an arithmetic shift in kasan_mem_to_shadow() provides a number of
> benefits:
>
> 1) The memory layout is easier to understand. KASAN_SHADOW_OFFSET
> becomes a canonical memory address, and the shifted pointer becomes a
> negative offset, so KASAN_SHADOW_OFFSET == KASAN_SHADOW_END regardless
> of the shift amount or the size of the virtual address space.
>
> 2) KASAN_SHADOW_OFFSET becomes a simpler constant, requiring only one
> instruction to load instead of two. Since it must be loaded in each
> function with a tag check, this decreases kernel text size by 0.5%.
>
> 3) This shift and the sign extension from kasan_reset_tag() can be
> combined into a single sbfx instruction. When this same algorithm change
> is applied to the compiler, it removes an instruction from each inline
> tag check, further reducing kernel text size by an additional 4.6%.
>
> These benefits extend to other architectures as well. On RISC-V, where
> the baseline ISA does not shifted addition or have an equivalent to the
> sbfx instruction, loading KASAN_SHADOW_OFFSET is reduced from 3 to 2
> instructions, and kasan_mem_to_shadow(kasan_reset_tag(addr)) similarly
> combines two consecutive right shifts.
>
> Link: https://github.com/llvm/llvm-project/blob/llvmorg-20-init/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp#L1316 [1]
> Link: https://github.com/llvm/llvm-project/blob/llvmorg-20-init/llvm/lib/Transforms/Instrumentation/HWAddressSanitizer.cpp#L895 [2]
> Link: https://github.com/llvm/llvm-project/blob/llvmorg-20-init/llvm/lib/Target/AArch64/AArch64AsmPrinter.cpp#L669 [3]
> Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
> ---
>
> Changes in v2:
>  - Improve the explanation for how KASAN_SHADOW_END is derived
>  - Update the range check in kasan_non_canonical_hook()
>
>  arch/arm64/Kconfig              | 10 +++++-----
>  arch/arm64/include/asm/memory.h | 17 +++++++++++++++--
>  arch/arm64/mm/kasan_init.c      |  7 +++++--
>  include/linux/kasan.h           | 10 ++++++++--
>  mm/kasan/report.c               | 22 ++++++++++++++++++----
>  scripts/gdb/linux/mm.py         |  5 +++--
>  6 files changed, 54 insertions(+), 17 deletions(-)
>
> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> index fd9df6dcc593..6a326908c941 100644
> --- a/arch/arm64/Kconfig
> +++ b/arch/arm64/Kconfig
> @@ -418,11 +418,11 @@ config KASAN_SHADOW_OFFSET
>         default 0xdffffe0000000000 if ARM64_VA_BITS_42 && !KASAN_SW_TAGS
>         default 0xdfffffc000000000 if ARM64_VA_BITS_39 && !KASAN_SW_TAGS
>         default 0xdffffff800000000 if ARM64_VA_BITS_36 && !KASAN_SW_TAGS
> -       default 0xefff800000000000 if (ARM64_VA_BITS_48 || (ARM64_VA_BITS_52 && !ARM64_16K_PAGES)) && KASAN_SW_TAGS
> -       default 0xefffc00000000000 if (ARM64_VA_BITS_47 || ARM64_VA_BITS_52) && ARM64_16K_PAGES && KASAN_SW_TAGS
> -       default 0xeffffe0000000000 if ARM64_VA_BITS_42 && KASAN_SW_TAGS
> -       default 0xefffffc000000000 if ARM64_VA_BITS_39 && KASAN_SW_TAGS
> -       default 0xeffffff800000000 if ARM64_VA_BITS_36 && KASAN_SW_TAGS
> +       default 0xffff800000000000 if (ARM64_VA_BITS_48 || (ARM64_VA_BITS_52 && !ARM64_16K_PAGES)) && KASAN_SW_TAGS
> +       default 0xffffc00000000000 if (ARM64_VA_BITS_47 || ARM64_VA_BITS_52) && ARM64_16K_PAGES && KASAN_SW_TAGS
> +       default 0xfffffe0000000000 if ARM64_VA_BITS_42 && KASAN_SW_TAGS
> +       default 0xffffffc000000000 if ARM64_VA_BITS_39 && KASAN_SW_TAGS
> +       default 0xfffffff800000000 if ARM64_VA_BITS_36 && KASAN_SW_TAGS
>         default 0xffffffffffffffff
>
>  config UNWIND_TABLES
> diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h
> index 0480c61dbb4f..a93fc9dc16f3 100644
> --- a/arch/arm64/include/asm/memory.h
> +++ b/arch/arm64/include/asm/memory.h
> @@ -80,7 +80,8 @@
>   * where KASAN_SHADOW_SCALE_SHIFT is the order of the number of bits that map
>   * to a single shadow byte and KASAN_SHADOW_OFFSET is a constant that offsets
>   * the mapping. Note that KASAN_SHADOW_OFFSET does not point to the start of
> - * the shadow memory region.
> + * the shadow memory region, since not all possible addresses have shadow
> + * memory allocated for them.

I'm not sure this addition makes sense: the original statement was to
point out that KASAN_SHADOW_OFFSET and KASAN_SHADOW_START are
different values. Even if we were to map shadow for userspace,
KASAN_SHADOW_OFFSET would still be a weird offset value for Generic
KASAN.

>   *
>   * Based on this mapping, we define two constants:
>   *
> @@ -89,7 +90,15 @@
>   *
>   * KASAN_SHADOW_END is defined first as the shadow address that corresponds to
>   * the upper bound of possible virtual kernel memory addresses UL(1) << 64
> - * according to the mapping formula.
> + * according to the mapping formula. For Generic KASAN, the address in the
> + * mapping formula is treated as unsigned (part of the compiler's ABI), so the
> + * end of the shadow memory region is at a large positive offset from
> + * KASAN_SHADOW_OFFSET. For Software Tag-Based KASAN, the address in the
> + * formula is treated as signed. Since all kernel addresses are negative, they
> + * map to shadow memory below KASAN_SHADOW_OFFSET, making KASAN_SHADOW_OFFSET
> + * itself the end of the shadow memory region. (User pointers are positive and
> + * would map to shadow memory above KASAN_SHADOW_OFFSET, but shadow memory is
> + * not allocated for them.)

This looks good!

>   *
>   * KASAN_SHADOW_START is defined second based on KASAN_SHADOW_END. The shadow
>   * memory start must map to the lowest possible kernel virtual memory address
> @@ -100,7 +109,11 @@
>   */
>  #if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS)
>  #define KASAN_SHADOW_OFFSET    _AC(CONFIG_KASAN_SHADOW_OFFSET, UL)
> +#ifdef CONFIG_KASAN_GENERIC
>  #define KASAN_SHADOW_END       ((UL(1) << (64 - KASAN_SHADOW_SCALE_SHIFT)) + KASAN_SHADOW_OFFSET)
> +#else
> +#define KASAN_SHADOW_END       KASAN_SHADOW_OFFSET
> +#endif
>  #define _KASAN_SHADOW_START(va)        (KASAN_SHADOW_END - (UL(1) << ((va) - KASAN_SHADOW_SCALE_SHIFT)))
>  #define KASAN_SHADOW_START     _KASAN_SHADOW_START(vabits_actual)
>  #define PAGE_END               KASAN_SHADOW_START
> diff --git a/arch/arm64/mm/kasan_init.c b/arch/arm64/mm/kasan_init.c
> index b65a29440a0c..6836e571555c 100644
> --- a/arch/arm64/mm/kasan_init.c
> +++ b/arch/arm64/mm/kasan_init.c
> @@ -198,8 +198,11 @@ static bool __init root_level_aligned(u64 addr)
>  /* The early shadow maps everything to a single page of zeroes */
>  asmlinkage void __init kasan_early_init(void)
>  {
> -       BUILD_BUG_ON(KASAN_SHADOW_OFFSET !=
> -               KASAN_SHADOW_END - (1UL << (64 - KASAN_SHADOW_SCALE_SHIFT)));
> +       if (IS_ENABLED(CONFIG_KASAN_GENERIC))
> +               BUILD_BUG_ON(KASAN_SHADOW_OFFSET !=
> +                       KASAN_SHADOW_END - (1UL << (64 - KASAN_SHADOW_SCALE_SHIFT)));
> +       else
> +               BUILD_BUG_ON(KASAN_SHADOW_OFFSET != KASAN_SHADOW_END);
>         BUILD_BUG_ON(!IS_ALIGNED(_KASAN_SHADOW_START(VA_BITS), SHADOW_ALIGN));
>         BUILD_BUG_ON(!IS_ALIGNED(_KASAN_SHADOW_START(VA_BITS_MIN), SHADOW_ALIGN));
>         BUILD_BUG_ON(!IS_ALIGNED(KASAN_SHADOW_END, SHADOW_ALIGN));
> diff --git a/include/linux/kasan.h b/include/linux/kasan.h
> index 00a3bf7c0d8f..03b440658817 100644
> --- a/include/linux/kasan.h
> +++ b/include/linux/kasan.h
> @@ -58,8 +58,14 @@ int kasan_populate_early_shadow(const void *shadow_start,
>  #ifndef kasan_mem_to_shadow
>  static inline void *kasan_mem_to_shadow(const void *addr)
>  {
> -       return (void *)((unsigned long)addr >> KASAN_SHADOW_SCALE_SHIFT)
> -               + KASAN_SHADOW_OFFSET;
> +       void *scaled;
> +
> +       if (IS_ENABLED(CONFIG_KASAN_GENERIC))
> +               scaled = (void *)((unsigned long)addr >> KASAN_SHADOW_SCALE_SHIFT);
> +       else
> +               scaled = (void *)((long)addr >> KASAN_SHADOW_SCALE_SHIFT);
> +
> +       return KASAN_SHADOW_OFFSET + scaled;
>  }
>  #endif
>
> diff --git a/mm/kasan/report.c b/mm/kasan/report.c
> index b48c768acc84..c08097715686 100644
> --- a/mm/kasan/report.c
> +++ b/mm/kasan/report.c
> @@ -644,15 +644,29 @@ void kasan_report_async(void)
>   */
>  void kasan_non_canonical_hook(unsigned long addr)
>  {
> +       unsigned long max_shadow_size = BIT(BITS_PER_LONG - KASAN_SHADOW_SCALE_SHIFT);
>         unsigned long orig_addr;
>         const char *bug_type;
>
>         /*
> -        * All addresses that came as a result of the memory-to-shadow mapping
> -        * (even for bogus pointers) must be >= KASAN_SHADOW_OFFSET.
> +        * With the default kasan_mem_to_shadow() algorithm, all addresses
> +        * returned by the memory-to-shadow mapping (even for bogus pointers)
> +        * must be within a certain displacement from KASAN_SHADOW_OFFSET.
> +        *
> +        * For Generic KASAN, the displacement is unsigned, so
> +        * KASAN_SHADOW_OFFSET is the smallest possible shadow address. For

This part of the comment doesn't seem correct: KASAN_SHADOW_OFFSET is
still a weird offset value for Generic KASAN, not the smallest
possible shadow address.

> +        * Software Tag-Based KASAN, the displacement is signed, so
> +        * KASAN_SHADOW_OFFSET is the center of the range.
>          */
> -       if (addr < KASAN_SHADOW_OFFSET)
> -               return;
> +       if (IS_ENABLED(CONFIG_KASAN_GENERIC)) {
> +               if (addr < KASAN_SHADOW_OFFSET ||
> +                   addr >= KASAN_SHADOW_OFFSET + max_shadow_size)
> +                       return;
> +       } else {
> +               if (addr < KASAN_SHADOW_OFFSET - max_shadow_size / 2 ||
> +                   addr >= KASAN_SHADOW_OFFSET + max_shadow_size / 2)
> +                       return;

Hm, I might be wrong, but I think this check does not work.

Let's say we have non-canonical address 0x4242424242424242 and number
of VA bits is 48.

Then:

KASAN_SHADOW_OFFSET == 0xffff800000000000
kasan_mem_to_shadow(0x4242424242424242) == 0x0423a42424242424
max_shadow_size == 0x1000000000000000
KASAN_SHADOW_OFFSET - max_shadow_size / 2 == 0xf7ff800000000000
KASAN_SHADOW_OFFSET + max_shadow_size / 2 == 0x07ff800000000000 (overflows)

0x0423a42424242424 is < than 0xf7ff800000000000, so the function will
wrongly return.

> +       }
>
>         orig_addr = (unsigned long)kasan_shadow_to_mem((void *)addr);
>

Just to double-check: kasan_shadow_to_mem() and addr_has_metadata()
don't need any changes, right?

> diff --git a/scripts/gdb/linux/mm.py b/scripts/gdb/linux/mm.py
> index 7571aebbe650..2e63f3dedd53 100644
> --- a/scripts/gdb/linux/mm.py
> +++ b/scripts/gdb/linux/mm.py
> @@ -110,12 +110,13 @@ class aarch64_page_ops():
>          self.KERNEL_END = gdb.parse_and_eval("_end")
>
>          if constants.LX_CONFIG_KASAN_GENERIC or constants.LX_CONFIG_KASAN_SW_TAGS:
> +            self.KASAN_SHADOW_OFFSET = constants.LX_CONFIG_KASAN_SHADOW_OFFSET
>              if constants.LX_CONFIG_KASAN_GENERIC:
>                  self.KASAN_SHADOW_SCALE_SHIFT = 3
> +                self.KASAN_SHADOW_END = (1 << (64 - self.KASAN_SHADOW_SCALE_SHIFT)) + self.KASAN_SHADOW_OFFSET
>              else:
>                  self.KASAN_SHADOW_SCALE_SHIFT = 4
> -            self.KASAN_SHADOW_OFFSET = constants.LX_CONFIG_KASAN_SHADOW_OFFSET
> -            self.KASAN_SHADOW_END = (1 << (64 - self.KASAN_SHADOW_SCALE_SHIFT)) + self.KASAN_SHADOW_OFFSET
> +                self.KASAN_SHADOW_END = self.KASAN_SHADOW_OFFSET
>              self.PAGE_END = self.KASAN_SHADOW_END - (1 << (self.vabits_actual - self.KASAN_SHADOW_SCALE_SHIFT))
>          else:
>              self.PAGE_END = self._PAGE_END(self.VA_BITS_MIN)
> --
> 2.45.1
>

Could you also check that everything works when CONFIG_KASAN_SW_TAGS +
CONFIG_KASAN_OUTLINE? I think it should, be makes sense to confirm.

Thank you!

[PATCH v2 2/9] kasan: sw_tags: Check kasan_flag_enabled at runtime

Posted by Samuel Holland 1 month ago

On RISC-V, the ISA extension required to dereference tagged pointers is
optional, and the interface to enable pointer masking requires firmware
support. Therefore, we must detect at runtime if sw_tags is usable on a
given machine. Reuse the logic from hw_tags to dynamically enable KASAN.

This commit makes no functional change to the KASAN_HW_TAGS code path.

Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com>
Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

(no changes since v1)

 include/linux/kasan-enabled.h | 15 +++++----------
 mm/kasan/hw_tags.c            | 10 ----------
 mm/kasan/tags.c               | 10 ++++++++++
 3 files changed, 15 insertions(+), 20 deletions(-)

diff --git a/include/linux/kasan-enabled.h b/include/linux/kasan-enabled.h
index 6f612d69ea0c..648bda9495b7 100644
--- a/include/linux/kasan-enabled.h
+++ b/include/linux/kasan-enabled.h
@@ -4,7 +4,7 @@
 
 #include <linux/static_key.h>
 
-#ifdef CONFIG_KASAN_HW_TAGS
+#if defined(CONFIG_KASAN_SW_TAGS) || defined(CONFIG_KASAN_HW_TAGS)
 
 DECLARE_STATIC_KEY_FALSE(kasan_flag_enabled);
 
@@ -13,23 +13,18 @@ static __always_inline bool kasan_enabled(void)
 	return static_branch_likely(&kasan_flag_enabled);
 }
 
-static inline bool kasan_hw_tags_enabled(void)
-{
-	return kasan_enabled();
-}
-
-#else /* CONFIG_KASAN_HW_TAGS */
+#else /* CONFIG_KASAN_SW_TAGS || CONFIG_KASAN_HW_TAGS */
 
 static inline bool kasan_enabled(void)
 {
 	return IS_ENABLED(CONFIG_KASAN);
 }
 
+#endif /* CONFIG_KASAN_SW_TAGS || CONFIG_KASAN_HW_TAGS */
+
 static inline bool kasan_hw_tags_enabled(void)
 {
-	return false;
+	return IS_ENABLED(CONFIG_KASAN_HW_TAGS) && kasan_enabled();
 }
 
-#endif /* CONFIG_KASAN_HW_TAGS */
-
 #endif /* LINUX_KASAN_ENABLED_H */
diff --git a/mm/kasan/hw_tags.c b/mm/kasan/hw_tags.c
index 9958ebc15d38..c3beeb94efa5 100644
--- a/mm/kasan/hw_tags.c
+++ b/mm/kasan/hw_tags.c
@@ -43,13 +43,6 @@ static enum kasan_arg kasan_arg __ro_after_init;
 static enum kasan_arg_mode kasan_arg_mode __ro_after_init;
 static enum kasan_arg_vmalloc kasan_arg_vmalloc __initdata;
 
-/*
- * Whether KASAN is enabled at all.
- * The value remains false until KASAN is initialized by kasan_init_hw_tags().
- */
-DEFINE_STATIC_KEY_FALSE(kasan_flag_enabled);
-EXPORT_SYMBOL(kasan_flag_enabled);
-
 /*
  * Whether the selected mode is synchronous, asynchronous, or asymmetric.
  * Defaults to KASAN_MODE_SYNC.
@@ -257,9 +250,6 @@ void __init kasan_init_hw_tags(void)
 
 	kasan_init_tags();
 
-	/* KASAN is now initialized, enable it. */
-	static_branch_enable(&kasan_flag_enabled);
-
 	pr_info("KernelAddressSanitizer initialized (hw-tags, mode=%s, vmalloc=%s, stacktrace=%s)\n",
 		kasan_mode_info(),
 		kasan_vmalloc_enabled() ? "on" : "off",
diff --git a/mm/kasan/tags.c b/mm/kasan/tags.c
index d65d48b85f90..c111d98961ed 100644
--- a/mm/kasan/tags.c
+++ b/mm/kasan/tags.c
@@ -32,6 +32,13 @@ enum kasan_arg_stacktrace {
 
 static enum kasan_arg_stacktrace kasan_arg_stacktrace __initdata;
 
+/*
+ * Whether KASAN is enabled at all.
+ * The value remains false until KASAN is initialized by kasan_init_tags().
+ */
+DEFINE_STATIC_KEY_FALSE(kasan_flag_enabled);
+EXPORT_SYMBOL(kasan_flag_enabled);
+
 /* Whether to collect alloc/free stack traces. */
 DEFINE_STATIC_KEY_TRUE(kasan_flag_stacktrace);
 
@@ -92,6 +99,9 @@ void __init kasan_init_tags(void)
 		if (WARN_ON(!stack_ring.entries))
 			static_branch_disable(&kasan_flag_stacktrace);
 	}
+
+	/* KASAN is now initialized, enable it. */
+	static_branch_enable(&kasan_flag_enabled);
 }
 
 static void save_stack_info(struct kmem_cache *cache, void *object,
-- 
2.45.1

[PATCH v2 3/9] kasan: sw_tags: Support outline stack tag generation

Posted by Samuel Holland 1 month ago

This allows stack tagging to be disabled at runtime by tagging all
stack objects with the match-all tag. This is necessary on RISC-V,
where a kernel with KASAN_SW_TAGS enabled is expected to boot on
hardware without pointer masking support.

Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

Changes in v2:
 - Split the generic and RISC-V parts of stack tag generation control
   to avoid breaking bisectability

 mm/kasan/kasan.h   | 2 ++
 mm/kasan/sw_tags.c | 9 +++++++++
 2 files changed, 11 insertions(+)

diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index f438a6cdc964..72da5ddcceaa 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -636,6 +636,8 @@ void *__asan_memset(void *addr, int c, ssize_t len);
 void *__asan_memmove(void *dest, const void *src, ssize_t len);
 void *__asan_memcpy(void *dest, const void *src, ssize_t len);
 
+u8 __hwasan_generate_tag(void);
+
 void __hwasan_load1_noabort(void *);
 void __hwasan_store1_noabort(void *);
 void __hwasan_load2_noabort(void *);
diff --git a/mm/kasan/sw_tags.c b/mm/kasan/sw_tags.c
index 220b5d4c6876..32435d33583a 100644
--- a/mm/kasan/sw_tags.c
+++ b/mm/kasan/sw_tags.c
@@ -70,6 +70,15 @@ u8 kasan_random_tag(void)
 	return (u8)(state % (KASAN_TAG_MAX + 1));
 }
 
+u8 __hwasan_generate_tag(void)
+{
+	if (!kasan_enabled())
+		return KASAN_TAG_KERNEL;
+
+	return kasan_random_tag();
+}
+EXPORT_SYMBOL(__hwasan_generate_tag);
+
 bool kasan_check_range(const void *addr, size_t size, bool write,
 			unsigned long ret_ip)
 {
-- 
2.45.1

Re: [PATCH v2 3/9] kasan: sw_tags: Support outline stack tag generation

Posted by Andrey Konovalov 1 month ago

On Tue, Oct 22, 2024 at 3:59 AM Samuel Holland
<samuel.holland@sifive.com> wrote:
>
> This allows stack tagging to be disabled at runtime by tagging all
> stack objects with the match-all tag. This is necessary on RISC-V,
> where a kernel with KASAN_SW_TAGS enabled is expected to boot on
> hardware without pointer masking support.
>
> Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
> ---
>
> Changes in v2:
>  - Split the generic and RISC-V parts of stack tag generation control
>    to avoid breaking bisectability
>
>  mm/kasan/kasan.h   | 2 ++
>  mm/kasan/sw_tags.c | 9 +++++++++
>  2 files changed, 11 insertions(+)
>
> diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
> index f438a6cdc964..72da5ddcceaa 100644
> --- a/mm/kasan/kasan.h
> +++ b/mm/kasan/kasan.h
> @@ -636,6 +636,8 @@ void *__asan_memset(void *addr, int c, ssize_t len);
>  void *__asan_memmove(void *dest, const void *src, ssize_t len);
>  void *__asan_memcpy(void *dest, const void *src, ssize_t len);
>
> +u8 __hwasan_generate_tag(void);
> +
>  void __hwasan_load1_noabort(void *);
>  void __hwasan_store1_noabort(void *);
>  void __hwasan_load2_noabort(void *);
> diff --git a/mm/kasan/sw_tags.c b/mm/kasan/sw_tags.c
> index 220b5d4c6876..32435d33583a 100644
> --- a/mm/kasan/sw_tags.c
> +++ b/mm/kasan/sw_tags.c
> @@ -70,6 +70,15 @@ u8 kasan_random_tag(void)
>         return (u8)(state % (KASAN_TAG_MAX + 1));
>  }
>
> +u8 __hwasan_generate_tag(void)
> +{
> +       if (!kasan_enabled())
> +               return KASAN_TAG_KERNEL;
> +
> +       return kasan_random_tag();
> +}
> +EXPORT_SYMBOL(__hwasan_generate_tag);
> +
>  bool kasan_check_range(const void *addr, size_t size, bool write,
>                         unsigned long ret_ip)
>  {
> --
> 2.45.1
>

Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com>

[PATCH v2 4/9] kasan: sw_tags: Support tag widths less than 8 bits

Posted by Samuel Holland 1 month ago

Allow architectures to override KASAN_TAG_KERNEL in asm/kasan.h. This
is needed on RISC-V, which supports 57-bit virtual addresses and 7-bit
pointer tags. For consistency, move the arm64 MTE definition of
KASAN_TAG_MIN to asm/kasan.h, since it is also architecture-dependent;
RISC-V's equivalent extension is expected to support 7-bit hardware
memory tags.

Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com>
Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

(no changes since v1)

 arch/arm64/include/asm/kasan.h   |  6 ++++--
 arch/arm64/include/asm/uaccess.h |  1 +
 include/linux/kasan-tags.h       | 13 ++++++++-----
 3 files changed, 13 insertions(+), 7 deletions(-)

diff --git a/arch/arm64/include/asm/kasan.h b/arch/arm64/include/asm/kasan.h
index e1b57c13f8a4..4ab419df8b93 100644
--- a/arch/arm64/include/asm/kasan.h
+++ b/arch/arm64/include/asm/kasan.h
@@ -6,8 +6,10 @@
 
 #include <linux/linkage.h>
 #include <asm/memory.h>
-#include <asm/mte-kasan.h>
-#include <asm/pgtable-types.h>
+
+#ifdef CONFIG_KASAN_HW_TAGS
+#define KASAN_TAG_MIN			0xF0 /* minimum value for random tags */
+#endif
 
 #define arch_kasan_set_tag(addr, tag)	__tag_set(addr, tag)
 #define arch_kasan_reset_tag(addr)	__tag_reset(addr)
diff --git a/arch/arm64/include/asm/uaccess.h b/arch/arm64/include/asm/uaccess.h
index 1aa4ecb73429..8f700a7dd2cd 100644
--- a/arch/arm64/include/asm/uaccess.h
+++ b/arch/arm64/include/asm/uaccess.h
@@ -22,6 +22,7 @@
 #include <asm/cpufeature.h>
 #include <asm/mmu.h>
 #include <asm/mte.h>
+#include <asm/mte-kasan.h>
 #include <asm/ptrace.h>
 #include <asm/memory.h>
 #include <asm/extable.h>
diff --git a/include/linux/kasan-tags.h b/include/linux/kasan-tags.h
index 4f85f562512c..e07c896f95d3 100644
--- a/include/linux/kasan-tags.h
+++ b/include/linux/kasan-tags.h
@@ -2,13 +2,16 @@
 #ifndef _LINUX_KASAN_TAGS_H
 #define _LINUX_KASAN_TAGS_H
 
+#include <asm/kasan.h>
+
+#ifndef KASAN_TAG_KERNEL
 #define KASAN_TAG_KERNEL	0xFF /* native kernel pointers tag */
-#define KASAN_TAG_INVALID	0xFE /* inaccessible memory tag */
-#define KASAN_TAG_MAX		0xFD /* maximum value for random tags */
+#endif
+
+#define KASAN_TAG_INVALID	(KASAN_TAG_KERNEL - 1) /* inaccessible memory tag */
+#define KASAN_TAG_MAX		(KASAN_TAG_KERNEL - 2) /* maximum value for random tags */
 
-#ifdef CONFIG_KASAN_HW_TAGS
-#define KASAN_TAG_MIN		0xF0 /* minimum value for random tags */
-#else
+#ifndef KASAN_TAG_MIN
 #define KASAN_TAG_MIN		0x00 /* minimum value for random tags */
 #endif
 
-- 
2.45.1

Re: [PATCH v2 4/9] kasan: sw_tags: Support tag widths less than 8 bits

Posted by kernel test robot 1 month ago

Hi Samuel,

kernel test robot noticed the following build errors:

[auto build test ERROR on akpm-mm/mm-everything]
[also build test ERROR on arm64/for-next/core masahiroy-kbuild/for-next masahiroy-kbuild/fixes linus/master v6.12-rc4 next-20241022]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Samuel-Holland/kasan-sw_tags-Use-arithmetic-shift-for-shadow-computation/20241022-100129
base:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything
patch link:    https://lore.kernel.org/r/20241022015913.3524425-5-samuel.holland%40sifive.com
patch subject: [PATCH v2 4/9] kasan: sw_tags: Support tag widths less than 8 bits
config: um-allnoconfig (https://download.01.org/0day-ci/archive/20241023/202410230319.eQozBGh7-lkp@intel.com/config)
compiler: clang version 17.0.6 (https://github.com/llvm/llvm-project 6009708b4367171ccdbf4b5905cb6a803753fe18)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20241023/202410230319.eQozBGh7-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202410230319.eQozBGh7-lkp@intel.com/

All errors (new ones prefixed by >>):

   In file included from arch/um/kernel/asm-offsets.c:1:
   In file included from arch/x86/um/shared/sysdep/kernel-offsets.h:5:
   In file included from include/linux/crypto.h:17:
   In file included from include/linux/slab.h:234:
   In file included from include/linux/kasan.h:7:
   In file included from include/linux/kasan-tags.h:5:
>> arch/um/include/asm/kasan.h:19:2: error: "KASAN_SHADOW_SIZE is not defined for this sub-architecture"
      19 | #error "KASAN_SHADOW_SIZE is not defined for this sub-architecture"
         |  ^
   1 error generated.
   make[3]: *** [scripts/Makefile.build:102: arch/um/kernel/asm-offsets.s] Error 1
   make[3]: Target 'prepare' not remade because of errors.
   make[2]: *** [Makefile:1203: prepare0] Error 2
   make[2]: Target 'prepare' not remade because of errors.
   make[1]: *** [Makefile:224: __sub-make] Error 2
   make[1]: Target 'prepare' not remade because of errors.
   make: *** [Makefile:224: __sub-make] Error 2
   make: Target 'prepare' not remade because of errors.


vim +19 arch/um/include/asm/kasan.h

5b301409e8bc5d7 Patricia Alfonso 2022-07-01  12  
5b301409e8bc5d7 Patricia Alfonso 2022-07-01  13  #ifdef CONFIG_X86_64
5b301409e8bc5d7 Patricia Alfonso 2022-07-01  14  #define KASAN_HOST_USER_SPACE_END_ADDR 0x00007fffffffffffUL
5b301409e8bc5d7 Patricia Alfonso 2022-07-01  15  /* KASAN_SHADOW_SIZE is the size of total address space divided by 8 */
5b301409e8bc5d7 Patricia Alfonso 2022-07-01  16  #define KASAN_SHADOW_SIZE ((KASAN_HOST_USER_SPACE_END_ADDR + 1) >> \
5b301409e8bc5d7 Patricia Alfonso 2022-07-01  17  			KASAN_SHADOW_SCALE_SHIFT)
5b301409e8bc5d7 Patricia Alfonso 2022-07-01  18  #else
5b301409e8bc5d7 Patricia Alfonso 2022-07-01 @19  #error "KASAN_SHADOW_SIZE is not defined for this sub-architecture"
5b301409e8bc5d7 Patricia Alfonso 2022-07-01  20  #endif /* CONFIG_X86_64 */
5b301409e8bc5d7 Patricia Alfonso 2022-07-01  21  

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH v2 4/9] kasan: sw_tags: Support tag widths less than 8 bits

Posted by kernel test robot 1 month ago

Hi Samuel,

kernel test robot noticed the following build errors:

[auto build test ERROR on akpm-mm/mm-everything]
[also build test ERROR on arm64/for-next/core masahiroy-kbuild/for-next masahiroy-kbuild/fixes linus/master v6.12-rc4 next-20241022]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Samuel-Holland/kasan-sw_tags-Use-arithmetic-shift-for-shadow-computation/20241022-100129
base:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything
patch link:    https://lore.kernel.org/r/20241022015913.3524425-5-samuel.holland%40sifive.com
patch subject: [PATCH v2 4/9] kasan: sw_tags: Support tag widths less than 8 bits
config: sh-allmodconfig (https://download.01.org/0day-ci/archive/20241023/202410230354.sjewoFxA-lkp@intel.com/config)
compiler: sh4-linux-gcc (GCC) 14.1.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20241023/202410230354.sjewoFxA-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202410230354.sjewoFxA-lkp@intel.com/

All errors (new ones prefixed by >>):

   In file included from include/linux/kasan.h:7,
                    from include/linux/mm.h:31,
                    from arch/sh/kernel/asm-offsets.c:14:
>> include/linux/kasan-tags.h:5:10: fatal error: asm/kasan.h: No such file or directory
       5 | #include <asm/kasan.h>
         |          ^~~~~~~~~~~~~
   compilation terminated.
   make[3]: *** [scripts/Makefile.build:102: arch/sh/kernel/asm-offsets.s] Error 1
   make[3]: Target 'prepare' not remade because of errors.
   make[2]: *** [Makefile:1203: prepare0] Error 2
   make[2]: Target 'prepare' not remade because of errors.
   make[1]: *** [Makefile:224: __sub-make] Error 2
   make[1]: Target 'prepare' not remade because of errors.
   make: *** [Makefile:224: __sub-make] Error 2
   make: Target 'prepare' not remade because of errors.


vim +5 include/linux/kasan-tags.h

     4	
   > 5	#include <asm/kasan.h>
     6	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

[PATCH v2 5/9] riscv: mm: Log potential KASAN shadow alias

Posted by Samuel Holland 1 month ago

When KASAN is enabled, shadow memory is allocated and mapped for all
legitimate kernel addresses, but not for the entire address space. As a
result, the kernel can fault when accessing a shadow address computed
from a bogus pointer. This can be confusing, because the shadow address
computed for (e.g.) NULL looks nothing like a NULL pointer. To assist
debugging, if the faulting address might be the result of a KASAN shadow
memory address computation, report the range of original memory
addresses that would map to the faulting address.

Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

Changes in v2:
 - New patch for v2

 arch/riscv/mm/fault.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/arch/riscv/mm/fault.c b/arch/riscv/mm/fault.c
index a9f2b4af8f3f..dae1131221b7 100644
--- a/arch/riscv/mm/fault.c
+++ b/arch/riscv/mm/fault.c
@@ -8,6 +8,7 @@
 
 
 #include <linux/mm.h>
+#include <linux/kasan.h>
 #include <linux/kernel.h>
 #include <linux/interrupt.h>
 #include <linux/perf_event.h>
@@ -30,6 +31,8 @@ static void die_kernel_fault(const char *msg, unsigned long addr,
 	pr_alert("Unable to handle kernel %s at virtual address " REG_FMT "\n", msg,
 		addr);
 
+	kasan_non_canonical_hook(addr);
+
 	bust_spinlocks(0);
 	die(regs, "Oops");
 	make_task_dead(SIGKILL);
-- 
2.45.1

Re: [PATCH v2 5/9] riscv: mm: Log potential KASAN shadow alias

Posted by Alexandre Ghiti 2 weeks, 6 days ago

Hi Samuel,

On 22/10/2024 03:57, Samuel Holland wrote:
> When KASAN is enabled, shadow memory is allocated and mapped for all
> legitimate kernel addresses, but not for the entire address space. As a
> result, the kernel can fault when accessing a shadow address computed
> from a bogus pointer. This can be confusing, because the shadow address
> computed for (e.g.) NULL looks nothing like a NULL pointer. To assist
> debugging, if the faulting address might be the result of a KASAN shadow
> memory address computation, report the range of original memory
> addresses that would map to the faulting address.
>
> Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
> ---
>
> Changes in v2:
>   - New patch for v2
>
>   arch/riscv/mm/fault.c | 3 +++
>   1 file changed, 3 insertions(+)
>
> diff --git a/arch/riscv/mm/fault.c b/arch/riscv/mm/fault.c
> index a9f2b4af8f3f..dae1131221b7 100644
> --- a/arch/riscv/mm/fault.c
> +++ b/arch/riscv/mm/fault.c
> @@ -8,6 +8,7 @@
>   
>   
>   #include <linux/mm.h>
> +#include <linux/kasan.h>
>   #include <linux/kernel.h>
>   #include <linux/interrupt.h>
>   #include <linux/perf_event.h>
> @@ -30,6 +31,8 @@ static void die_kernel_fault(const char *msg, unsigned long addr,
>   	pr_alert("Unable to handle kernel %s at virtual address " REG_FMT "\n", msg,
>   		addr);
>   
> +	kasan_non_canonical_hook(addr);
> +
>   	bust_spinlocks(0);
>   	die(regs, "Oops");
>   	make_task_dead(SIGKILL);


That's nice, I used to do that by hand :)

Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>

Thanks,

Alex

[PATCH v2 6/9] riscv: Do not rely on KASAN to define the memory layout

Posted by Samuel Holland 1 month ago

Commit 66673099f734 ("riscv: mm: Pre-allocate vmemmap/direct map/kasan
PGD entries") used the start of the KASAN shadow memory region to
represent the end of the linear map, since the two memory regions were
immediately adjacent. This is no longer the case for Sv39; commit
5c8405d763dc ("riscv: Extend sv39 linear mapping max size to 128G")
introduced a 4 GiB hole between the regions. Introducing KASAN_SW_TAGS
will cut the size of the shadow memory region in half, creating an even
larger hole.

Avoid wasting PGD entries on this hole by using the size of the linear
map (KERN_VIRT_SIZE) to compute PAGE_END.

Since KASAN_SHADOW_START/KASAN_SHADOW_END are used inside an IS_ENABLED
block, it's not possible to completely hide the constants when KASAN is
disabled, so provide dummy definitions for that case.

Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

(no changes since v1)

 arch/riscv/include/asm/kasan.h | 11 +++++++++--
 arch/riscv/mm/init.c           |  2 +-
 2 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/arch/riscv/include/asm/kasan.h b/arch/riscv/include/asm/kasan.h
index e6a0071bdb56..a4e92ce9fa31 100644
--- a/arch/riscv/include/asm/kasan.h
+++ b/arch/riscv/include/asm/kasan.h
@@ -6,6 +6,8 @@
 
 #ifndef __ASSEMBLY__
 
+#ifdef CONFIG_KASAN
+
 /*
  * The following comment was copied from arm64:
  * KASAN_SHADOW_START: beginning of the kernel virtual addresses.
@@ -33,13 +35,18 @@
 #define KASAN_SHADOW_START	((KASAN_SHADOW_END - KASAN_SHADOW_SIZE) & PGDIR_MASK)
 #define KASAN_SHADOW_END	MODULES_LOWEST_VADDR
 
-#ifdef CONFIG_KASAN
 #define KASAN_SHADOW_OFFSET	_AC(CONFIG_KASAN_SHADOW_OFFSET, UL)
 
 void kasan_init(void);
 asmlinkage void kasan_early_init(void);
 void kasan_swapper_init(void);
 
-#endif
+#else /* CONFIG_KASAN */
+
+#define KASAN_SHADOW_START	MODULES_LOWEST_VADDR
+#define KASAN_SHADOW_END	MODULES_LOWEST_VADDR
+
+#endif /* CONFIG_KASAN */
+
 #endif
 #endif /* __ASM_KASAN_H */
diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c
index 0e8c20adcd98..1f9bb95c2169 100644
--- a/arch/riscv/mm/init.c
+++ b/arch/riscv/mm/init.c
@@ -1494,7 +1494,7 @@ static void __init preallocate_pgd_pages_range(unsigned long start, unsigned lon
 	panic("Failed to pre-allocate %s pages for %s area\n", lvl, area);
 }
 
-#define PAGE_END KASAN_SHADOW_START
+#define PAGE_END (PAGE_OFFSET + KERN_VIRT_SIZE)
 
 void __init pgtable_cache_init(void)
 {
-- 
2.45.1

Re: [PATCH v2 6/9] riscv: Do not rely on KASAN to define the memory layout

Posted by Alexandre Ghiti 2 weeks, 6 days ago

On 22/10/2024 03:57, Samuel Holland wrote:
> Commit 66673099f734 ("riscv: mm: Pre-allocate vmemmap/direct map/kasan
> PGD entries") used the start of the KASAN shadow memory region to
> represent the end of the linear map, since the two memory regions were
> immediately adjacent. This is no longer the case for Sv39; commit
> 5c8405d763dc ("riscv: Extend sv39 linear mapping max size to 128G")
> introduced a 4 GiB hole between the regions. Introducing KASAN_SW_TAGS
> will cut the size of the shadow memory region in half, creating an even
> larger hole.
>
> Avoid wasting PGD entries on this hole by using the size of the linear
> map (KERN_VIRT_SIZE) to compute PAGE_END.
>
> Since KASAN_SHADOW_START/KASAN_SHADOW_END are used inside an IS_ENABLED
> block, it's not possible to completely hide the constants when KASAN is
> disabled, so provide dummy definitions for that case.
>
> Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
> ---
>
> (no changes since v1)
>
>   arch/riscv/include/asm/kasan.h | 11 +++++++++--
>   arch/riscv/mm/init.c           |  2 +-
>   2 files changed, 10 insertions(+), 3 deletions(-)
>
> diff --git a/arch/riscv/include/asm/kasan.h b/arch/riscv/include/asm/kasan.h
> index e6a0071bdb56..a4e92ce9fa31 100644
> --- a/arch/riscv/include/asm/kasan.h
> +++ b/arch/riscv/include/asm/kasan.h
> @@ -6,6 +6,8 @@
>   
>   #ifndef __ASSEMBLY__
>   
> +#ifdef CONFIG_KASAN
> +
>   /*
>    * The following comment was copied from arm64:
>    * KASAN_SHADOW_START: beginning of the kernel virtual addresses.
> @@ -33,13 +35,18 @@
>   #define KASAN_SHADOW_START	((KASAN_SHADOW_END - KASAN_SHADOW_SIZE) & PGDIR_MASK)
>   #define KASAN_SHADOW_END	MODULES_LOWEST_VADDR
>   
> -#ifdef CONFIG_KASAN
>   #define KASAN_SHADOW_OFFSET	_AC(CONFIG_KASAN_SHADOW_OFFSET, UL)
>   
>   void kasan_init(void);
>   asmlinkage void kasan_early_init(void);
>   void kasan_swapper_init(void);
>   
> -#endif
> +#else /* CONFIG_KASAN */
> +
> +#define KASAN_SHADOW_START	MODULES_LOWEST_VADDR
> +#define KASAN_SHADOW_END	MODULES_LOWEST_VADDR
> +
> +#endif /* CONFIG_KASAN */
> +
>   #endif
>   #endif /* __ASM_KASAN_H */
> diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c
> index 0e8c20adcd98..1f9bb95c2169 100644
> --- a/arch/riscv/mm/init.c
> +++ b/arch/riscv/mm/init.c
> @@ -1494,7 +1494,7 @@ static void __init preallocate_pgd_pages_range(unsigned long start, unsigned lon
>   	panic("Failed to pre-allocate %s pages for %s area\n", lvl, area);
>   }
>   
> -#define PAGE_END KASAN_SHADOW_START
> +#define PAGE_END (PAGE_OFFSET + KERN_VIRT_SIZE)
>   
>   void __init pgtable_cache_init(void)
>   {


Looks good and cleaner, you can add:

Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>

Thanks,

Alex

[PATCH v2 7/9] riscv: Align the sv39 linear map to 16 GiB

Posted by Samuel Holland 1 month ago

The KASAN implementation on RISC-V requires the shadow memory for the
vmemmap and linear map regions to be aligned to a PMD boundary (1 GiB).
For KASAN_GENERIC (KASAN_SHADOW_SCALE_SHIFT == 3), this enforces 8 GiB
alignment for the memory regions themselves. KASAN_SW_TAGS uses 16-byte
granules (KASAN_SHADOW_SCALE_SHIFT == 4), so now the memory regions must
be aligned to a 16 GiB boundary.

Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

(no changes since v1)

 Documentation/arch/riscv/vm-layout.rst | 10 +++++-----
 arch/riscv/include/asm/page.h          |  2 +-
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/Documentation/arch/riscv/vm-layout.rst b/Documentation/arch/riscv/vm-layout.rst
index eabec99b5852..c0778c421b34 100644
--- a/Documentation/arch/riscv/vm-layout.rst
+++ b/Documentation/arch/riscv/vm-layout.rst
@@ -47,11 +47,11 @@ RISC-V Linux Kernel SV39
                                                               | Kernel-space virtual memory, shared between all processes:
   ____________________________________________________________|___________________________________________________________
                     |            |                  |         |
-   ffffffc4fea00000 | -236    GB | ffffffc4feffffff |    6 MB | fixmap
-   ffffffc4ff000000 | -236    GB | ffffffc4ffffffff |   16 MB | PCI io
-   ffffffc500000000 | -236    GB | ffffffc5ffffffff |    4 GB | vmemmap
-   ffffffc600000000 | -232    GB | ffffffd5ffffffff |   64 GB | vmalloc/ioremap space
-   ffffffd600000000 | -168    GB | fffffff5ffffffff |  128 GB | direct mapping of all physical memory
+   ffffffc2fea00000 | -244    GB | ffffffc2feffffff |    6 MB | fixmap
+   ffffffc2ff000000 | -244    GB | ffffffc2ffffffff |   16 MB | PCI io
+   ffffffc300000000 | -244    GB | ffffffc3ffffffff |    4 GB | vmemmap
+   ffffffc400000000 | -240    GB | ffffffd3ffffffff |   64 GB | vmalloc/ioremap space
+   ffffffd400000000 | -176    GB | fffffff3ffffffff |  128 GB | direct mapping of all physical memory
                     |            |                  |         |
    fffffff700000000 |  -36    GB | fffffffeffffffff |   32 GB | kasan
   __________________|____________|__________________|_________|____________________________________________________________
diff --git a/arch/riscv/include/asm/page.h b/arch/riscv/include/asm/page.h
index 32d308a3355f..6e2f79cf77c5 100644
--- a/arch/riscv/include/asm/page.h
+++ b/arch/riscv/include/asm/page.h
@@ -37,7 +37,7 @@
  * define the PAGE_OFFSET value for SV48 and SV39.
  */
 #define PAGE_OFFSET_L4		_AC(0xffffaf8000000000, UL)
-#define PAGE_OFFSET_L3		_AC(0xffffffd600000000, UL)
+#define PAGE_OFFSET_L3		_AC(0xffffffd400000000, UL)
 #else
 #define PAGE_OFFSET		_AC(CONFIG_PAGE_OFFSET, UL)
 #endif /* CONFIG_64BIT */
-- 
2.45.1

Re: [PATCH v2 7/9] riscv: Align the sv39 linear map to 16 GiB

Posted by Alexandre Ghiti 2 weeks, 6 days ago

On 22/10/2024 03:57, Samuel Holland wrote:
> The KASAN implementation on RISC-V requires the shadow memory for the
> vmemmap and linear map regions to be aligned to a PMD boundary (1 GiB).


PUD boundary


> For KASAN_GENERIC (KASAN_SHADOW_SCALE_SHIFT == 3), this enforces 8 GiB
> alignment for the memory regions themselves. KASAN_SW_TAGS uses 16-byte
> granules (KASAN_SHADOW_SCALE_SHIFT == 4), so now the memory regions must
> be aligned to a 16 GiB boundary.
>
> Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
> ---
>
> (no changes since v1)
>
>   Documentation/arch/riscv/vm-layout.rst | 10 +++++-----
>   arch/riscv/include/asm/page.h          |  2 +-
>   2 files changed, 6 insertions(+), 6 deletions(-)
>
> diff --git a/Documentation/arch/riscv/vm-layout.rst b/Documentation/arch/riscv/vm-layout.rst
> index eabec99b5852..c0778c421b34 100644
> --- a/Documentation/arch/riscv/vm-layout.rst
> +++ b/Documentation/arch/riscv/vm-layout.rst
> @@ -47,11 +47,11 @@ RISC-V Linux Kernel SV39
>                                                                 | Kernel-space virtual memory, shared between all processes:
>     ____________________________________________________________|___________________________________________________________
>                       |            |                  |         |
> -   ffffffc4fea00000 | -236    GB | ffffffc4feffffff |    6 MB | fixmap
> -   ffffffc4ff000000 | -236    GB | ffffffc4ffffffff |   16 MB | PCI io
> -   ffffffc500000000 | -236    GB | ffffffc5ffffffff |    4 GB | vmemmap
> -   ffffffc600000000 | -232    GB | ffffffd5ffffffff |   64 GB | vmalloc/ioremap space
> -   ffffffd600000000 | -168    GB | fffffff5ffffffff |  128 GB | direct mapping of all physical memory
> +   ffffffc2fea00000 | -244    GB | ffffffc2feffffff |    6 MB | fixmap
> +   ffffffc2ff000000 | -244    GB | ffffffc2ffffffff |   16 MB | PCI io
> +   ffffffc300000000 | -244    GB | ffffffc3ffffffff |    4 GB | vmemmap
> +   ffffffc400000000 | -240    GB | ffffffd3ffffffff |   64 GB | vmalloc/ioremap space
> +   ffffffd400000000 | -176    GB | fffffff3ffffffff |  128 GB | direct mapping of all physical memory
>                       |            |                  |         |
>      fffffff700000000 |  -36    GB | fffffffeffffffff |   32 GB | kasan
>     __________________|____________|__________________|_________|____________________________________________________________
> diff --git a/arch/riscv/include/asm/page.h b/arch/riscv/include/asm/page.h
> index 32d308a3355f..6e2f79cf77c5 100644
> --- a/arch/riscv/include/asm/page.h
> +++ b/arch/riscv/include/asm/page.h
> @@ -37,7 +37,7 @@
>    * define the PAGE_OFFSET value for SV48 and SV39.
>    */
>   #define PAGE_OFFSET_L4		_AC(0xffffaf8000000000, UL)
> -#define PAGE_OFFSET_L3		_AC(0xffffffd600000000, UL)
> +#define PAGE_OFFSET_L3		_AC(0xffffffd400000000, UL)
>   #else
>   #define PAGE_OFFSET		_AC(CONFIG_PAGE_OFFSET, UL)
>   #endif /* CONFIG_64BIT */


Other than the nit above (that I think should be fixed though), you can add:

Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>

Thanks,

Alex

[PATCH v2 8/9] riscv: Add SBI Firmware Features extension definitions

Posted by Samuel Holland 1 month ago

From: Clément Léger <cleger@rivosinc.com>

Add necessary SBI definitions to use the FWFT extension.

[Samuel: Add SBI_FWFT_POINTER_MASKING_PMLEN]

Signed-off-by: Clément Léger <cleger@rivosinc.com>
Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

Changes in v2:
 - New patch for v2

 arch/riscv/include/asm/sbi.h | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/arch/riscv/include/asm/sbi.h b/arch/riscv/include/asm/sbi.h
index 98f631b051db..4a35c6ffe49f 100644
--- a/arch/riscv/include/asm/sbi.h
+++ b/arch/riscv/include/asm/sbi.h
@@ -34,6 +34,7 @@ enum sbi_ext_id {
 	SBI_EXT_PMU = 0x504D55,
 	SBI_EXT_DBCN = 0x4442434E,
 	SBI_EXT_STA = 0x535441,
+	SBI_EXT_FWFT = 0x46574654,
 
 	/* Experimentals extensions must lie within this range */
 	SBI_EXT_EXPERIMENTAL_START = 0x08000000,
@@ -281,6 +282,33 @@ struct sbi_sta_struct {
 
 #define SBI_SHMEM_DISABLE		-1
 
+/* SBI function IDs for FW feature extension */
+#define SBI_EXT_FWFT_SET		0x0
+#define SBI_EXT_FWFT_GET		0x1
+
+enum sbi_fwft_feature_t {
+	SBI_FWFT_MISALIGNED_EXC_DELEG		= 0x0,
+	SBI_FWFT_LANDING_PAD			= 0x1,
+	SBI_FWFT_SHADOW_STACK			= 0x2,
+	SBI_FWFT_DOUBLE_TRAP			= 0x3,
+	SBI_FWFT_PTE_AD_HW_UPDATING		= 0x4,
+	SBI_FWFT_POINTER_MASKING_PMLEN		= 0x5,
+	SBI_FWFT_LOCAL_RESERVED_START		= 0x6,
+	SBI_FWFT_LOCAL_RESERVED_END		= 0x3fffffff,
+	SBI_FWFT_LOCAL_PLATFORM_START		= 0x40000000,
+	SBI_FWFT_LOCAL_PLATFORM_END		= 0x7fffffff,
+
+	SBI_FWFT_GLOBAL_RESERVED_START		= 0x80000000,
+	SBI_FWFT_GLOBAL_RESERVED_END		= 0xbfffffff,
+	SBI_FWFT_GLOBAL_PLATFORM_START		= 0xc0000000,
+	SBI_FWFT_GLOBAL_PLATFORM_END		= 0xffffffff,
+};
+
+#define SBI_FWFT_GLOBAL_FEATURE_BIT		(1 << 31)
+#define SBI_FWFT_PLATFORM_FEATURE_BIT		(1 << 30)
+
+#define SBI_FWFT_SET_FLAG_LOCK			(1 << 0)
+
 /* SBI spec version fields */
 #define SBI_SPEC_VERSION_DEFAULT	0x1
 #define SBI_SPEC_VERSION_MAJOR_SHIFT	24
-- 
2.45.1

[PATCH v2 9/9] riscv: Implement KASAN_SW_TAGS

Posted by Samuel Holland 1 month ago

Implement support for software tag-based KASAN using the RISC-V pointer
masking extension, which supports 7 and/or 16-bit tags. This implemen-
tation uses 7-bit tags, so it is compatible with either hardware mode.

Pointer masking is an optional ISA extension, and it must be enabled
using an SBI call to firmware on each CPU. This SBI call must be made
very early in smp_callin(), as dereferencing any tagged pointers before
that point will crash the kernel. If the SBI call fails on the boot CPU,
then KASAN is globally disabled, and the kernel boots normally (unless
stack tagging is enabled). If the SBI call fails on any other CPU, that
CPU is excluded from the system.

When pointer masking is enabled for the kernel's privilege mode, it must
be more careful about accepting tagged pointers from userspace.
Normally, __access_ok() accepts tagged aliases of kernel memory as long
as the MSB is zero, since those addresses cannot be dereferenced -- they
will cause a page fault in the uaccess routines. But when the kernel is
using pointer masking, those addresses are dereferenceable, so
__access_ok() must specifically check the most-significant non-tag bit
instead of the MSB.

Pointer masking does not apply to the operands of fence instructions, so
software is responsible for untagging those addresses.

Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

Changes in v2:
 - Fix build error with KASAN_GENERIC
 - Use symbolic definitons for SBI firmware features call
 - Update indentation in scripts/Makefile.kasan
 - Use kasan_params to set hwasan-generate-tags-with-calls=1

 Documentation/dev-tools/kasan.rst | 14 ++++---
 arch/riscv/Kconfig                |  4 +-
 arch/riscv/include/asm/cache.h    |  4 ++
 arch/riscv/include/asm/kasan.h    | 20 ++++++++++
 arch/riscv/include/asm/page.h     | 19 ++++++++--
 arch/riscv/include/asm/pgtable.h  |  6 +++
 arch/riscv/include/asm/tlbflush.h |  4 +-
 arch/riscv/kernel/setup.c         |  6 +++
 arch/riscv/kernel/smpboot.c       |  8 +++-
 arch/riscv/lib/Makefile           |  2 +
 arch/riscv/lib/kasan_sw_tags.S    | 61 +++++++++++++++++++++++++++++++
 arch/riscv/mm/kasan_init.c        | 32 +++++++++++++++-
 arch/riscv/mm/physaddr.c          |  4 ++
 scripts/Makefile.kasan            |  5 +++
 14 files changed, 174 insertions(+), 15 deletions(-)
 create mode 100644 arch/riscv/lib/kasan_sw_tags.S

diff --git a/Documentation/dev-tools/kasan.rst b/Documentation/dev-tools/kasan.rst
index d7de44f5339d..6548aebac57f 100644
--- a/Documentation/dev-tools/kasan.rst
+++ b/Documentation/dev-tools/kasan.rst
@@ -22,8 +22,8 @@ architectures, but it has significant performance and memory overheads.
 
 Software Tag-Based KASAN or SW_TAGS KASAN, enabled with CONFIG_KASAN_SW_TAGS,
 can be used for both debugging and dogfood testing, similar to userspace HWASan.
-This mode is only supported for arm64, but its moderate memory overhead allows
-using it for testing on memory-restricted devices with real workloads.
+This mode is only supported on arm64 and riscv, but its moderate memory overhead
+allows using it for testing on memory-restricted devices with real workloads.
 
 Hardware Tag-Based KASAN or HW_TAGS KASAN, enabled with CONFIG_KASAN_HW_TAGS,
 is the mode intended to be used as an in-field memory bug detector or as a
@@ -340,12 +340,14 @@ Software Tag-Based KASAN
 ~~~~~~~~~~~~~~~~~~~~~~~~
 
 Software Tag-Based KASAN uses a software memory tagging approach to checking
-access validity. It is currently only implemented for the arm64 architecture.
+access validity. It is currently only implemented for the arm64 and riscv
+architectures.
 
 Software Tag-Based KASAN uses the Top Byte Ignore (TBI) feature of arm64 CPUs
-to store a pointer tag in the top byte of kernel pointers. It uses shadow memory
-to store memory tags associated with each 16-byte memory cell (therefore, it
-dedicates 1/16th of the kernel memory for shadow memory).
+or the pointer masking (Sspm) feature of RISC-V CPUs to store a pointer tag in
+the top byte of kernel pointers. It uses shadow memory to store memory tags
+associated with each 16-byte memory cell (therefore, it dedicates 1/16th of the
+kernel memory for shadow memory).
 
 On each memory allocation, Software Tag-Based KASAN generates a random tag, tags
 the allocated memory with this tag, and embeds the same tag into the returned
diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
index 62545946ecf4..d08b99f6bf76 100644
--- a/arch/riscv/Kconfig
+++ b/arch/riscv/Kconfig
@@ -121,6 +121,7 @@ config RISCV
 	select HAVE_ARCH_JUMP_LABEL if !XIP_KERNEL
 	select HAVE_ARCH_JUMP_LABEL_RELATIVE if !XIP_KERNEL
 	select HAVE_ARCH_KASAN if MMU && 64BIT
+	select HAVE_ARCH_KASAN_SW_TAGS if MMU && 64BIT
 	select HAVE_ARCH_KASAN_VMALLOC if MMU && 64BIT
 	select HAVE_ARCH_KFENCE if MMU && 64BIT
 	select HAVE_ARCH_KGDB if !XIP_KERNEL
@@ -291,7 +292,8 @@ config PAGE_OFFSET
 
 config KASAN_SHADOW_OFFSET
 	hex
-	depends on KASAN_GENERIC
+	depends on KASAN
+	default 0xffffffff00000000 if KASAN_SW_TAGS
 	default 0xdfffffff00000000 if 64BIT
 	default 0xffffffff if 32BIT
 
diff --git a/arch/riscv/include/asm/cache.h b/arch/riscv/include/asm/cache.h
index 570e9d8acad1..232288a060c6 100644
--- a/arch/riscv/include/asm/cache.h
+++ b/arch/riscv/include/asm/cache.h
@@ -16,6 +16,10 @@
 #define ARCH_KMALLOC_MINALIGN	(8)
 #endif
 
+#ifdef CONFIG_KASAN_SW_TAGS
+#define ARCH_SLAB_MINALIGN	(1ULL << KASAN_SHADOW_SCALE_SHIFT)
+#endif
+
 /*
  * RISC-V requires the stack pointer to be 16-byte aligned, so ensure that
  * the flat loader aligns it accordingly.
diff --git a/arch/riscv/include/asm/kasan.h b/arch/riscv/include/asm/kasan.h
index a4e92ce9fa31..f6b378ba936d 100644
--- a/arch/riscv/include/asm/kasan.h
+++ b/arch/riscv/include/asm/kasan.h
@@ -25,7 +25,11 @@
  *      KASAN_SHADOW_OFFSET = KASAN_SHADOW_END -
  *                              (1ULL << (64 - KASAN_SHADOW_SCALE_SHIFT))
  */
+#if defined(CONFIG_KASAN_GENERIC)
 #define KASAN_SHADOW_SCALE_SHIFT	3
+#elif defined(CONFIG_KASAN_SW_TAGS)
+#define KASAN_SHADOW_SCALE_SHIFT	4
+#endif
 
 #define KASAN_SHADOW_SIZE	(UL(1) << ((VA_BITS - 1) - KASAN_SHADOW_SCALE_SHIFT))
 /*
@@ -37,6 +41,14 @@
 
 #define KASAN_SHADOW_OFFSET	_AC(CONFIG_KASAN_SHADOW_OFFSET, UL)
 
+#ifdef CONFIG_KASAN_SW_TAGS
+#define KASAN_TAG_KERNEL	0x7f /* native kernel pointers tag */
+#endif
+
+#define arch_kasan_set_tag(addr, tag)	__tag_set(addr, tag)
+#define arch_kasan_reset_tag(addr)	__tag_reset(addr)
+#define arch_kasan_get_tag(addr)	__tag_get(addr)
+
 void kasan_init(void);
 asmlinkage void kasan_early_init(void);
 void kasan_swapper_init(void);
@@ -48,5 +60,13 @@ void kasan_swapper_init(void);
 
 #endif /* CONFIG_KASAN */
 
+#ifdef CONFIG_KASAN_SW_TAGS
+bool kasan_boot_cpu_enabled(void);
+int kasan_cpu_enable(void);
+#else
+static inline bool kasan_boot_cpu_enabled(void) { return false; }
+static inline int kasan_cpu_enable(void) { return 0; }
+#endif
+
 #endif
 #endif /* __ASM_KASAN_H */
diff --git a/arch/riscv/include/asm/page.h b/arch/riscv/include/asm/page.h
index 6e2f79cf77c5..43c625a4894d 100644
--- a/arch/riscv/include/asm/page.h
+++ b/arch/riscv/include/asm/page.h
@@ -89,6 +89,16 @@ typedef struct page *pgtable_t;
 #define PTE_FMT "%08lx"
 #endif
 
+#ifdef CONFIG_KASAN_SW_TAGS
+#define __tag_set(addr, tag)	((void *)((((u64)(addr) << 7) >> 7) | ((u64)(tag) << 57)))
+#define __tag_reset(addr)	((void *)((s64)((u64)(addr) << 7) >> 7))
+#define __tag_get(addr)		((u8)((u64)(addr) >> 57))
+#else
+#define __tag_set(addr, tag)	(addr)
+#define __tag_reset(addr)	(addr)
+#define __tag_get(addr)		0
+#endif
+
 #if defined(CONFIG_64BIT) && defined(CONFIG_MMU)
 /*
  * We override this value as its generic definition uses __pa too early in
@@ -168,7 +178,7 @@ phys_addr_t linear_mapping_va_to_pa(unsigned long x);
 #endif
 
 #define __va_to_pa_nodebug(x)	({						\
-	unsigned long _x = x;							\
+	unsigned long _x = (unsigned long)__tag_reset(x);			\
 	is_linear_mapping(_x) ?							\
 		linear_mapping_va_to_pa(_x) : kernel_mapping_va_to_pa(_x);	\
 	})
@@ -192,7 +202,10 @@ extern phys_addr_t __phys_addr_symbol(unsigned long x);
 #define pfn_to_virt(pfn)	(__va(pfn_to_phys(pfn)))
 
 #define virt_to_page(vaddr)	(pfn_to_page(virt_to_pfn(vaddr)))
-#define page_to_virt(page)	(pfn_to_virt(page_to_pfn(page)))
+#define page_to_virt(page)	({						\
+	__typeof__(page) __page = page;						\
+	__tag_set(pfn_to_virt(page_to_pfn(__page)), page_kasan_tag(__page));	\
+})
 
 #define page_to_phys(page)	(pfn_to_phys(page_to_pfn(page)))
 #define phys_to_page(paddr)	(pfn_to_page(phys_to_pfn(paddr)))
@@ -209,7 +222,7 @@ static __always_inline void *pfn_to_kaddr(unsigned long pfn)
 #endif /* __ASSEMBLY__ */
 
 #define virt_addr_valid(vaddr)	({						\
-	unsigned long _addr = (unsigned long)vaddr;				\
+	unsigned long _addr = (unsigned long)__tag_reset(vaddr);		\
 	(unsigned long)(_addr) >= PAGE_OFFSET && pfn_valid(virt_to_pfn(_addr));	\
 })
 
diff --git a/arch/riscv/include/asm/pgtable.h b/arch/riscv/include/asm/pgtable.h
index e79f15293492..ae6fa9dba0fc 100644
--- a/arch/riscv/include/asm/pgtable.h
+++ b/arch/riscv/include/asm/pgtable.h
@@ -916,7 +916,13 @@ static inline pte_t pte_swp_clear_exclusive(pte_t pte)
  */
 #ifdef CONFIG_64BIT
 #define TASK_SIZE_64	(PGDIR_SIZE * PTRS_PER_PGD / 2)
+/*
+ * When pointer masking is enabled for the kernel's privilege mode,
+ * __access_ok() must reject tagged aliases of kernel memory.
+ */
+#ifndef CONFIG_KASAN_SW_TAGS
 #define TASK_SIZE_MAX	LONG_MAX
+#endif
 
 #ifdef CONFIG_COMPAT
 #define TASK_SIZE_32	(_AC(0x80000000, UL) - PAGE_SIZE)
diff --git a/arch/riscv/include/asm/tlbflush.h b/arch/riscv/include/asm/tlbflush.h
index 72e559934952..68b3a85c6960 100644
--- a/arch/riscv/include/asm/tlbflush.h
+++ b/arch/riscv/include/asm/tlbflush.h
@@ -31,14 +31,14 @@ static inline void local_flush_tlb_all_asid(unsigned long asid)
 /* Flush one page from local TLB */
 static inline void local_flush_tlb_page(unsigned long addr)
 {
-	ALT_SFENCE_VMA_ADDR(addr);
+	ALT_SFENCE_VMA_ADDR(__tag_reset(addr));
 }
 
 static inline void local_flush_tlb_page_asid(unsigned long addr,
 					     unsigned long asid)
 {
 	if (asid != FLUSH_TLB_NO_ASID)
-		ALT_SFENCE_VMA_ADDR_ASID(addr, asid);
+		ALT_SFENCE_VMA_ADDR_ASID(__tag_reset(addr), asid);
 	else
 		local_flush_tlb_page(addr);
 }
diff --git a/arch/riscv/kernel/setup.c b/arch/riscv/kernel/setup.c
index a2cde65b69e9..fdc72edc4857 100644
--- a/arch/riscv/kernel/setup.c
+++ b/arch/riscv/kernel/setup.c
@@ -299,6 +299,12 @@ void __init setup_arch(char **cmdline_p)
 	riscv_user_isa_enable();
 }
 
+void __init smp_prepare_boot_cpu(void)
+{
+	if (kasan_boot_cpu_enabled())
+		kasan_init_sw_tags();
+}
+
 bool arch_cpu_is_hotpluggable(int cpu)
 {
 	return cpu_has_hotplug(cpu);
diff --git a/arch/riscv/kernel/smpboot.c b/arch/riscv/kernel/smpboot.c
index 0f8f1c95ac38..a1cc555691b0 100644
--- a/arch/riscv/kernel/smpboot.c
+++ b/arch/riscv/kernel/smpboot.c
@@ -29,6 +29,7 @@
 #include <asm/cacheflush.h>
 #include <asm/cpu_ops.h>
 #include <asm/irq.h>
+#include <asm/kasan.h>
 #include <asm/mmu_context.h>
 #include <asm/numa.h>
 #include <asm/tlbflush.h>
@@ -210,7 +211,11 @@ void __init smp_cpus_done(unsigned int max_cpus)
 asmlinkage __visible void smp_callin(void)
 {
 	struct mm_struct *mm = &init_mm;
-	unsigned int curr_cpuid = smp_processor_id();
+	unsigned int curr_cpuid;
+
+	/* Must be called first, before referencing any dynamic allocations */
+	if (kasan_boot_cpu_enabled() && kasan_cpu_enable())
+		return;
 
 	if (has_vector()) {
 		/*
@@ -225,6 +230,7 @@ asmlinkage __visible void smp_callin(void)
 	mmgrab(mm);
 	current->active_mm = mm;
 
+	curr_cpuid = smp_processor_id();
 	store_cpu_topology(curr_cpuid);
 	notify_cpu_starting(curr_cpuid);
 
diff --git a/arch/riscv/lib/Makefile b/arch/riscv/lib/Makefile
index 8eec6b69a875..ae36616fe1f5 100644
--- a/arch/riscv/lib/Makefile
+++ b/arch/riscv/lib/Makefile
@@ -20,3 +20,5 @@ lib-$(CONFIG_RISCV_ISA_ZBC)	+= crc32.o
 obj-$(CONFIG_FUNCTION_ERROR_INJECTION) += error-inject.o
 lib-$(CONFIG_RISCV_ISA_V)	+= xor.o
 lib-$(CONFIG_RISCV_ISA_V)	+= riscv_v_helpers.o
+
+obj-$(CONFIG_KASAN_SW_TAGS) += kasan_sw_tags.o
diff --git a/arch/riscv/lib/kasan_sw_tags.S b/arch/riscv/lib/kasan_sw_tags.S
new file mode 100644
index 000000000000..f7d3e0acba6a
--- /dev/null
+++ b/arch/riscv/lib/kasan_sw_tags.S
@@ -0,0 +1,61 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2020 Google LLC
+ * Copyright (C) 2024 SiFive
+ */
+
+#include <linux/linkage.h>
+
+/*
+ * Report a tag mismatch detected by tag-based KASAN.
+ *
+ * A compiler-generated thunk calls this with a custom calling convention.
+ * Upon entry to this function, the following registers have been modified:
+ *
+ *   x1/ra:     clobbered by call to this function
+ *   x2/sp:     decremented by 256
+ *   x6/t1:     tag from shadow memory
+ *   x7/t2:     tag from pointer
+ *   x10/a0:    fault address
+ *   x11/a1:    fault description
+ *   x28/t3:    clobbered by thunk
+ *   x29/t4:    clobbered by thunk
+ *   x30/t5:    clobbered by thunk
+ *   x31/t6:    clobbered by thunk
+ *
+ * The caller has decremented the SP by 256 bytes, and stored the following
+ * registers in slots on the stack according to their number (sp + 8 * xN):
+ *
+ *   x1/ra:     return address to user code
+ *   x8/s0/fp:  saved value from user code
+ *   x10/a0:    saved value from user code
+ *   x11/a1:    saved value from user code
+ */
+SYM_CODE_START(__hwasan_tag_mismatch)
+	/* Store the remaining unclobbered caller-saved regs */
+	sd	t0, (8 *  5)(sp)
+	sd	a2, (8 * 12)(sp)
+	sd	a3, (8 * 13)(sp)
+	sd	a4, (8 * 14)(sp)
+	sd	a5, (8 * 15)(sp)
+	sd	a6, (8 * 16)(sp)
+	sd	a7, (8 * 17)(sp)
+
+	/* a0 and a1 are already set by the thunk */
+	ld	a2, (8 *  1)(sp)
+	call	kasan_tag_mismatch
+
+	ld	ra, (8 *  1)(sp)
+	ld	t0, (8 *  5)(sp)
+	ld	a0, (8 * 10)(sp)
+	ld	a1, (8 * 11)(sp)
+	ld	a2, (8 * 12)(sp)
+	ld	a3, (8 * 13)(sp)
+	ld	a4, (8 * 14)(sp)
+	ld	a5, (8 * 15)(sp)
+	ld	a6, (8 * 16)(sp)
+	ld	a7, (8 * 17)(sp)
+	addi	sp, sp, 256
+	ret
+SYM_CODE_END(__hwasan_tag_mismatch)
+EXPORT_SYMBOL(__hwasan_tag_mismatch)
diff --git a/arch/riscv/mm/kasan_init.c b/arch/riscv/mm/kasan_init.c
index c301c8d291d2..50f0e7a03cc8 100644
--- a/arch/riscv/mm/kasan_init.c
+++ b/arch/riscv/mm/kasan_init.c
@@ -11,6 +11,10 @@
 #include <asm/fixmap.h>
 #include <asm/pgalloc.h>
 
+#ifdef CONFIG_KASAN_SW_TAGS
+static bool __kasan_boot_cpu_enabled __ro_after_init;
+#endif
+
 /*
  * Kasan shadow region must lie at a fixed address across sv39, sv48 and sv57
  * which is right before the kernel.
@@ -323,8 +327,11 @@ asmlinkage void __init kasan_early_init(void)
 {
 	uintptr_t i;
 
-	BUILD_BUG_ON(KASAN_SHADOW_OFFSET !=
-		KASAN_SHADOW_END - (1UL << (64 - KASAN_SHADOW_SCALE_SHIFT)));
+	if (IS_ENABLED(CONFIG_KASAN_GENERIC))
+		BUILD_BUG_ON(KASAN_SHADOW_OFFSET !=
+			KASAN_SHADOW_END - (1UL << (64 - KASAN_SHADOW_SCALE_SHIFT)));
+	else
+		BUILD_BUG_ON(KASAN_SHADOW_OFFSET != KASAN_SHADOW_END);
 
 	for (i = 0; i < PTRS_PER_PTE; ++i)
 		set_pte(kasan_early_shadow_pte + i,
@@ -356,6 +363,10 @@ asmlinkage void __init kasan_early_init(void)
 				 KASAN_SHADOW_START, KASAN_SHADOW_END);
 
 	local_flush_tlb_all();
+
+#ifdef CONFIG_KASAN_SW_TAGS
+	__kasan_boot_cpu_enabled = !kasan_cpu_enable();
+#endif
 }
 
 void __init kasan_swapper_init(void)
@@ -534,3 +545,20 @@ void __init kasan_init(void)
 	csr_write(CSR_SATP, PFN_DOWN(__pa(swapper_pg_dir)) | satp_mode);
 	local_flush_tlb_all();
 }
+
+#ifdef CONFIG_KASAN_SW_TAGS
+bool kasan_boot_cpu_enabled(void)
+{
+	return __kasan_boot_cpu_enabled;
+}
+
+int kasan_cpu_enable(void)
+{
+	struct sbiret ret;
+
+	ret = sbi_ecall(SBI_EXT_FWFT, SBI_EXT_FWFT_SET,
+			SBI_FWFT_POINTER_MASKING_PMLEN, 7, 0, 0, 0, 0);
+
+	return sbi_err_map_linux_errno(ret.error);
+}
+#endif
diff --git a/arch/riscv/mm/physaddr.c b/arch/riscv/mm/physaddr.c
index 18706f457da7..6d1cf6ffd54e 100644
--- a/arch/riscv/mm/physaddr.c
+++ b/arch/riscv/mm/physaddr.c
@@ -8,6 +8,8 @@
 
 phys_addr_t __virt_to_phys(unsigned long x)
 {
+	x = __tag_reset(x);
+
 	/*
 	 * Boundary checking aginst the kernel linear mapping space.
 	 */
@@ -24,6 +26,8 @@ phys_addr_t __phys_addr_symbol(unsigned long x)
 	unsigned long kernel_start = kernel_map.virt_addr;
 	unsigned long kernel_end = kernel_start + kernel_map.size;
 
+	x = __tag_reset(x);
+
 	/*
 	 * Boundary checking aginst the kernel image mapping.
 	 * __pa_symbol should only be used on kernel symbol addresses.
diff --git a/scripts/Makefile.kasan b/scripts/Makefile.kasan
index 693dbbebebba..72a8c9d5fb0e 100644
--- a/scripts/Makefile.kasan
+++ b/scripts/Makefile.kasan
@@ -91,6 +91,11 @@ ifeq ($(call clang-min-version, 150000)$(call gcc-min-version, 130000),y)
 	kasan_params += hwasan-kernel-mem-intrinsic-prefix=1
 endif
 
+# RISC-V requires dynamically determining if stack tagging can be enabled.
+ifdef CONFIG_RISCV
+	kasan_params += hwasan-generate-tags-with-calls=1
+endif
+
 endif # CONFIG_KASAN_SW_TAGS
 
 # Add all as-supported KASAN LLVM parameters requested by the configuration.
-- 
2.45.1

Re: [PATCH v2 9/9] riscv: Implement KASAN_SW_TAGS

Posted by Andrey Konovalov 1 month ago

On Tue, Oct 22, 2024 at 3:59 AM Samuel Holland
<samuel.holland@sifive.com> wrote:
>
> Implement support for software tag-based KASAN using the RISC-V pointer
> masking extension, which supports 7 and/or 16-bit tags. This implemen-
> tation uses 7-bit tags, so it is compatible with either hardware mode.
>
> Pointer masking is an optional ISA extension, and it must be enabled
> using an SBI call to firmware on each CPU. This SBI call must be made
> very early in smp_callin(), as dereferencing any tagged pointers before
> that point will crash the kernel. If the SBI call fails on the boot CPU,
> then KASAN is globally disabled, and the kernel boots normally (unless
> stack tagging is enabled). If the SBI call fails on any other CPU, that
> CPU is excluded from the system.
>
> When pointer masking is enabled for the kernel's privilege mode, it must
> be more careful about accepting tagged pointers from userspace.
> Normally, __access_ok() accepts tagged aliases of kernel memory as long
> as the MSB is zero, since those addresses cannot be dereferenced -- they
> will cause a page fault in the uaccess routines. But when the kernel is
> using pointer masking, those addresses are dereferenceable, so
> __access_ok() must specifically check the most-significant non-tag bit
> instead of the MSB.
>
> Pointer masking does not apply to the operands of fence instructions, so
> software is responsible for untagging those addresses.
>
> Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
> ---
>
> Changes in v2:
>  - Fix build error with KASAN_GENERIC
>  - Use symbolic definitons for SBI firmware features call
>  - Update indentation in scripts/Makefile.kasan
>  - Use kasan_params to set hwasan-generate-tags-with-calls=1
>
>  Documentation/dev-tools/kasan.rst | 14 ++++---
>  arch/riscv/Kconfig                |  4 +-
>  arch/riscv/include/asm/cache.h    |  4 ++
>  arch/riscv/include/asm/kasan.h    | 20 ++++++++++
>  arch/riscv/include/asm/page.h     | 19 ++++++++--
>  arch/riscv/include/asm/pgtable.h  |  6 +++
>  arch/riscv/include/asm/tlbflush.h |  4 +-
>  arch/riscv/kernel/setup.c         |  6 +++
>  arch/riscv/kernel/smpboot.c       |  8 +++-
>  arch/riscv/lib/Makefile           |  2 +
>  arch/riscv/lib/kasan_sw_tags.S    | 61 +++++++++++++++++++++++++++++++
>  arch/riscv/mm/kasan_init.c        | 32 +++++++++++++++-
>  arch/riscv/mm/physaddr.c          |  4 ++
>  scripts/Makefile.kasan            |  5 +++
>  14 files changed, 174 insertions(+), 15 deletions(-)
>  create mode 100644 arch/riscv/lib/kasan_sw_tags.S
>
> diff --git a/Documentation/dev-tools/kasan.rst b/Documentation/dev-tools/kasan.rst
> index d7de44f5339d..6548aebac57f 100644
> --- a/Documentation/dev-tools/kasan.rst
> +++ b/Documentation/dev-tools/kasan.rst
> @@ -22,8 +22,8 @@ architectures, but it has significant performance and memory overheads.
>
>  Software Tag-Based KASAN or SW_TAGS KASAN, enabled with CONFIG_KASAN_SW_TAGS,
>  can be used for both debugging and dogfood testing, similar to userspace HWASan.
> -This mode is only supported for arm64, but its moderate memory overhead allows
> -using it for testing on memory-restricted devices with real workloads.
> +This mode is only supported on arm64 and riscv, but its moderate memory overhead
> +allows using it for testing on memory-restricted devices with real workloads.
>
>  Hardware Tag-Based KASAN or HW_TAGS KASAN, enabled with CONFIG_KASAN_HW_TAGS,
>  is the mode intended to be used as an in-field memory bug detector or as a
> @@ -340,12 +340,14 @@ Software Tag-Based KASAN
>  ~~~~~~~~~~~~~~~~~~~~~~~~
>
>  Software Tag-Based KASAN uses a software memory tagging approach to checking
> -access validity. It is currently only implemented for the arm64 architecture.
> +access validity. It is currently only implemented for the arm64 and riscv
> +architectures.
>
>  Software Tag-Based KASAN uses the Top Byte Ignore (TBI) feature of arm64 CPUs
> -to store a pointer tag in the top byte of kernel pointers. It uses shadow memory
> -to store memory tags associated with each 16-byte memory cell (therefore, it
> -dedicates 1/16th of the kernel memory for shadow memory).
> +or the pointer masking (Sspm) feature of RISC-V CPUs to store a pointer tag in
> +the top byte of kernel pointers. It uses shadow memory to store memory tags
> +associated with each 16-byte memory cell (therefore, it dedicates 1/16th of the
> +kernel memory for shadow memory).
>
>  On each memory allocation, Software Tag-Based KASAN generates a random tag, tags
>  the allocated memory with this tag, and embeds the same tag into the returned
> diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
> index 62545946ecf4..d08b99f6bf76 100644
> --- a/arch/riscv/Kconfig
> +++ b/arch/riscv/Kconfig
> @@ -121,6 +121,7 @@ config RISCV
>         select HAVE_ARCH_JUMP_LABEL if !XIP_KERNEL
>         select HAVE_ARCH_JUMP_LABEL_RELATIVE if !XIP_KERNEL
>         select HAVE_ARCH_KASAN if MMU && 64BIT
> +       select HAVE_ARCH_KASAN_SW_TAGS if MMU && 64BIT
>         select HAVE_ARCH_KASAN_VMALLOC if MMU && 64BIT
>         select HAVE_ARCH_KFENCE if MMU && 64BIT
>         select HAVE_ARCH_KGDB if !XIP_KERNEL
> @@ -291,7 +292,8 @@ config PAGE_OFFSET
>
>  config KASAN_SHADOW_OFFSET
>         hex
> -       depends on KASAN_GENERIC
> +       depends on KASAN
> +       default 0xffffffff00000000 if KASAN_SW_TAGS
>         default 0xdfffffff00000000 if 64BIT
>         default 0xffffffff if 32BIT
>
> diff --git a/arch/riscv/include/asm/cache.h b/arch/riscv/include/asm/cache.h
> index 570e9d8acad1..232288a060c6 100644
> --- a/arch/riscv/include/asm/cache.h
> +++ b/arch/riscv/include/asm/cache.h
> @@ -16,6 +16,10 @@
>  #define ARCH_KMALLOC_MINALIGN  (8)
>  #endif
>
> +#ifdef CONFIG_KASAN_SW_TAGS
> +#define ARCH_SLAB_MINALIGN     (1ULL << KASAN_SHADOW_SCALE_SHIFT)
> +#endif
> +
>  /*
>   * RISC-V requires the stack pointer to be 16-byte aligned, so ensure that
>   * the flat loader aligns it accordingly.
> diff --git a/arch/riscv/include/asm/kasan.h b/arch/riscv/include/asm/kasan.h
> index a4e92ce9fa31..f6b378ba936d 100644
> --- a/arch/riscv/include/asm/kasan.h
> +++ b/arch/riscv/include/asm/kasan.h
> @@ -25,7 +25,11 @@
>   *      KASAN_SHADOW_OFFSET = KASAN_SHADOW_END -
>   *                              (1ULL << (64 - KASAN_SHADOW_SCALE_SHIFT))
>   */
> +#if defined(CONFIG_KASAN_GENERIC)
>  #define KASAN_SHADOW_SCALE_SHIFT       3
> +#elif defined(CONFIG_KASAN_SW_TAGS)
> +#define KASAN_SHADOW_SCALE_SHIFT       4
> +#endif
>
>  #define KASAN_SHADOW_SIZE      (UL(1) << ((VA_BITS - 1) - KASAN_SHADOW_SCALE_SHIFT))
>  /*
> @@ -37,6 +41,14 @@
>
>  #define KASAN_SHADOW_OFFSET    _AC(CONFIG_KASAN_SHADOW_OFFSET, UL)
>
> +#ifdef CONFIG_KASAN_SW_TAGS
> +#define KASAN_TAG_KERNEL       0x7f /* native kernel pointers tag */
> +#endif
> +
> +#define arch_kasan_set_tag(addr, tag)  __tag_set(addr, tag)
> +#define arch_kasan_reset_tag(addr)     __tag_reset(addr)
> +#define arch_kasan_get_tag(addr)       __tag_get(addr)
> +
>  void kasan_init(void);
>  asmlinkage void kasan_early_init(void);
>  void kasan_swapper_init(void);
> @@ -48,5 +60,13 @@ void kasan_swapper_init(void);
>
>  #endif /* CONFIG_KASAN */
>
> +#ifdef CONFIG_KASAN_SW_TAGS
> +bool kasan_boot_cpu_enabled(void);
> +int kasan_cpu_enable(void);
> +#else
> +static inline bool kasan_boot_cpu_enabled(void) { return false; }
> +static inline int kasan_cpu_enable(void) { return 0; }
> +#endif
> +
>  #endif
>  #endif /* __ASM_KASAN_H */
> diff --git a/arch/riscv/include/asm/page.h b/arch/riscv/include/asm/page.h
> index 6e2f79cf77c5..43c625a4894d 100644
> --- a/arch/riscv/include/asm/page.h
> +++ b/arch/riscv/include/asm/page.h
> @@ -89,6 +89,16 @@ typedef struct page *pgtable_t;
>  #define PTE_FMT "%08lx"
>  #endif
>
> +#ifdef CONFIG_KASAN_SW_TAGS
> +#define __tag_set(addr, tag)   ((void *)((((u64)(addr) << 7) >> 7) | ((u64)(tag) << 57)))
> +#define __tag_reset(addr)      ((void *)((s64)((u64)(addr) << 7) >> 7))
> +#define __tag_get(addr)                ((u8)((u64)(addr) >> 57))
> +#else
> +#define __tag_set(addr, tag)   (addr)
> +#define __tag_reset(addr)      (addr)
> +#define __tag_get(addr)                0
> +#endif
> +
>  #if defined(CONFIG_64BIT) && defined(CONFIG_MMU)
>  /*
>   * We override this value as its generic definition uses __pa too early in
> @@ -168,7 +178,7 @@ phys_addr_t linear_mapping_va_to_pa(unsigned long x);
>  #endif
>
>  #define __va_to_pa_nodebug(x)  ({                                              \
> -       unsigned long _x = x;                                                   \
> +       unsigned long _x = (unsigned long)__tag_reset(x);                       \
>         is_linear_mapping(_x) ?                                                 \
>                 linear_mapping_va_to_pa(_x) : kernel_mapping_va_to_pa(_x);      \
>         })
> @@ -192,7 +202,10 @@ extern phys_addr_t __phys_addr_symbol(unsigned long x);
>  #define pfn_to_virt(pfn)       (__va(pfn_to_phys(pfn)))
>
>  #define virt_to_page(vaddr)    (pfn_to_page(virt_to_pfn(vaddr)))
> -#define page_to_virt(page)     (pfn_to_virt(page_to_pfn(page)))
> +#define page_to_virt(page)     ({                                              \
> +       __typeof__(page) __page = page;                                         \
> +       __tag_set(pfn_to_virt(page_to_pfn(__page)), page_kasan_tag(__page));    \
> +})
>
>  #define page_to_phys(page)     (pfn_to_phys(page_to_pfn(page)))
>  #define phys_to_page(paddr)    (pfn_to_page(phys_to_pfn(paddr)))
> @@ -209,7 +222,7 @@ static __always_inline void *pfn_to_kaddr(unsigned long pfn)
>  #endif /* __ASSEMBLY__ */
>
>  #define virt_addr_valid(vaddr) ({                                              \
> -       unsigned long _addr = (unsigned long)vaddr;                             \
> +       unsigned long _addr = (unsigned long)__tag_reset(vaddr);                \
>         (unsigned long)(_addr) >= PAGE_OFFSET && pfn_valid(virt_to_pfn(_addr)); \
>  })
>
> diff --git a/arch/riscv/include/asm/pgtable.h b/arch/riscv/include/asm/pgtable.h
> index e79f15293492..ae6fa9dba0fc 100644
> --- a/arch/riscv/include/asm/pgtable.h
> +++ b/arch/riscv/include/asm/pgtable.h
> @@ -916,7 +916,13 @@ static inline pte_t pte_swp_clear_exclusive(pte_t pte)
>   */
>  #ifdef CONFIG_64BIT
>  #define TASK_SIZE_64   (PGDIR_SIZE * PTRS_PER_PGD / 2)
> +/*
> + * When pointer masking is enabled for the kernel's privilege mode,
> + * __access_ok() must reject tagged aliases of kernel memory.
> + */
> +#ifndef CONFIG_KASAN_SW_TAGS
>  #define TASK_SIZE_MAX  LONG_MAX
> +#endif
>
>  #ifdef CONFIG_COMPAT
>  #define TASK_SIZE_32   (_AC(0x80000000, UL) - PAGE_SIZE)
> diff --git a/arch/riscv/include/asm/tlbflush.h b/arch/riscv/include/asm/tlbflush.h
> index 72e559934952..68b3a85c6960 100644
> --- a/arch/riscv/include/asm/tlbflush.h
> +++ b/arch/riscv/include/asm/tlbflush.h
> @@ -31,14 +31,14 @@ static inline void local_flush_tlb_all_asid(unsigned long asid)
>  /* Flush one page from local TLB */
>  static inline void local_flush_tlb_page(unsigned long addr)
>  {
> -       ALT_SFENCE_VMA_ADDR(addr);
> +       ALT_SFENCE_VMA_ADDR(__tag_reset(addr));
>  }
>
>  static inline void local_flush_tlb_page_asid(unsigned long addr,
>                                              unsigned long asid)
>  {
>         if (asid != FLUSH_TLB_NO_ASID)
> -               ALT_SFENCE_VMA_ADDR_ASID(addr, asid);
> +               ALT_SFENCE_VMA_ADDR_ASID(__tag_reset(addr), asid);
>         else
>                 local_flush_tlb_page(addr);
>  }
> diff --git a/arch/riscv/kernel/setup.c b/arch/riscv/kernel/setup.c
> index a2cde65b69e9..fdc72edc4857 100644
> --- a/arch/riscv/kernel/setup.c
> +++ b/arch/riscv/kernel/setup.c
> @@ -299,6 +299,12 @@ void __init setup_arch(char **cmdline_p)
>         riscv_user_isa_enable();
>  }
>
> +void __init smp_prepare_boot_cpu(void)
> +{
> +       if (kasan_boot_cpu_enabled())
> +               kasan_init_sw_tags();
> +}
> +
>  bool arch_cpu_is_hotpluggable(int cpu)
>  {
>         return cpu_has_hotplug(cpu);
> diff --git a/arch/riscv/kernel/smpboot.c b/arch/riscv/kernel/smpboot.c
> index 0f8f1c95ac38..a1cc555691b0 100644
> --- a/arch/riscv/kernel/smpboot.c
> +++ b/arch/riscv/kernel/smpboot.c
> @@ -29,6 +29,7 @@
>  #include <asm/cacheflush.h>
>  #include <asm/cpu_ops.h>
>  #include <asm/irq.h>
> +#include <asm/kasan.h>
>  #include <asm/mmu_context.h>
>  #include <asm/numa.h>
>  #include <asm/tlbflush.h>
> @@ -210,7 +211,11 @@ void __init smp_cpus_done(unsigned int max_cpus)
>  asmlinkage __visible void smp_callin(void)
>  {
>         struct mm_struct *mm = &init_mm;
> -       unsigned int curr_cpuid = smp_processor_id();
> +       unsigned int curr_cpuid;
> +
> +       /* Must be called first, before referencing any dynamic allocations */
> +       if (kasan_boot_cpu_enabled() && kasan_cpu_enable())
> +               return;
>
>         if (has_vector()) {
>                 /*
> @@ -225,6 +230,7 @@ asmlinkage __visible void smp_callin(void)
>         mmgrab(mm);
>         current->active_mm = mm;
>
> +       curr_cpuid = smp_processor_id();
>         store_cpu_topology(curr_cpuid);
>         notify_cpu_starting(curr_cpuid);
>
> diff --git a/arch/riscv/lib/Makefile b/arch/riscv/lib/Makefile
> index 8eec6b69a875..ae36616fe1f5 100644
> --- a/arch/riscv/lib/Makefile
> +++ b/arch/riscv/lib/Makefile
> @@ -20,3 +20,5 @@ lib-$(CONFIG_RISCV_ISA_ZBC)   += crc32.o
>  obj-$(CONFIG_FUNCTION_ERROR_INJECTION) += error-inject.o
>  lib-$(CONFIG_RISCV_ISA_V)      += xor.o
>  lib-$(CONFIG_RISCV_ISA_V)      += riscv_v_helpers.o
> +
> +obj-$(CONFIG_KASAN_SW_TAGS) += kasan_sw_tags.o
> diff --git a/arch/riscv/lib/kasan_sw_tags.S b/arch/riscv/lib/kasan_sw_tags.S
> new file mode 100644
> index 000000000000..f7d3e0acba6a
> --- /dev/null
> +++ b/arch/riscv/lib/kasan_sw_tags.S
> @@ -0,0 +1,61 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +/*
> + * Copyright (C) 2020 Google LLC
> + * Copyright (C) 2024 SiFive
> + */
> +
> +#include <linux/linkage.h>
> +
> +/*
> + * Report a tag mismatch detected by tag-based KASAN.
> + *
> + * A compiler-generated thunk calls this with a custom calling convention.
> + * Upon entry to this function, the following registers have been modified:
> + *
> + *   x1/ra:     clobbered by call to this function
> + *   x2/sp:     decremented by 256
> + *   x6/t1:     tag from shadow memory
> + *   x7/t2:     tag from pointer
> + *   x10/a0:    fault address
> + *   x11/a1:    fault description
> + *   x28/t3:    clobbered by thunk
> + *   x29/t4:    clobbered by thunk
> + *   x30/t5:    clobbered by thunk
> + *   x31/t6:    clobbered by thunk
> + *
> + * The caller has decremented the SP by 256 bytes, and stored the following
> + * registers in slots on the stack according to their number (sp + 8 * xN):
> + *
> + *   x1/ra:     return address to user code
> + *   x8/s0/fp:  saved value from user code
> + *   x10/a0:    saved value from user code
> + *   x11/a1:    saved value from user code
> + */
> +SYM_CODE_START(__hwasan_tag_mismatch)
> +       /* Store the remaining unclobbered caller-saved regs */
> +       sd      t0, (8 *  5)(sp)
> +       sd      a2, (8 * 12)(sp)
> +       sd      a3, (8 * 13)(sp)
> +       sd      a4, (8 * 14)(sp)
> +       sd      a5, (8 * 15)(sp)
> +       sd      a6, (8 * 16)(sp)
> +       sd      a7, (8 * 17)(sp)
> +
> +       /* a0 and a1 are already set by the thunk */
> +       ld      a2, (8 *  1)(sp)
> +       call    kasan_tag_mismatch
> +
> +       ld      ra, (8 *  1)(sp)
> +       ld      t0, (8 *  5)(sp)
> +       ld      a0, (8 * 10)(sp)
> +       ld      a1, (8 * 11)(sp)
> +       ld      a2, (8 * 12)(sp)
> +       ld      a3, (8 * 13)(sp)
> +       ld      a4, (8 * 14)(sp)
> +       ld      a5, (8 * 15)(sp)
> +       ld      a6, (8 * 16)(sp)
> +       ld      a7, (8 * 17)(sp)
> +       addi    sp, sp, 256
> +       ret
> +SYM_CODE_END(__hwasan_tag_mismatch)
> +EXPORT_SYMBOL(__hwasan_tag_mismatch)
> diff --git a/arch/riscv/mm/kasan_init.c b/arch/riscv/mm/kasan_init.c
> index c301c8d291d2..50f0e7a03cc8 100644
> --- a/arch/riscv/mm/kasan_init.c
> +++ b/arch/riscv/mm/kasan_init.c
> @@ -11,6 +11,10 @@
>  #include <asm/fixmap.h>
>  #include <asm/pgalloc.h>
>
> +#ifdef CONFIG_KASAN_SW_TAGS
> +static bool __kasan_boot_cpu_enabled __ro_after_init;
> +#endif
> +
>  /*
>   * Kasan shadow region must lie at a fixed address across sv39, sv48 and sv57
>   * which is right before the kernel.
> @@ -323,8 +327,11 @@ asmlinkage void __init kasan_early_init(void)
>  {
>         uintptr_t i;
>
> -       BUILD_BUG_ON(KASAN_SHADOW_OFFSET !=
> -               KASAN_SHADOW_END - (1UL << (64 - KASAN_SHADOW_SCALE_SHIFT)));
> +       if (IS_ENABLED(CONFIG_KASAN_GENERIC))
> +               BUILD_BUG_ON(KASAN_SHADOW_OFFSET !=
> +                       KASAN_SHADOW_END - (1UL << (64 - KASAN_SHADOW_SCALE_SHIFT)));
> +       else
> +               BUILD_BUG_ON(KASAN_SHADOW_OFFSET != KASAN_SHADOW_END);
>
>         for (i = 0; i < PTRS_PER_PTE; ++i)
>                 set_pte(kasan_early_shadow_pte + i,
> @@ -356,6 +363,10 @@ asmlinkage void __init kasan_early_init(void)
>                                  KASAN_SHADOW_START, KASAN_SHADOW_END);
>
>         local_flush_tlb_all();
> +
> +#ifdef CONFIG_KASAN_SW_TAGS
> +       __kasan_boot_cpu_enabled = !kasan_cpu_enable();
> +#endif
>  }
>
>  void __init kasan_swapper_init(void)
> @@ -534,3 +545,20 @@ void __init kasan_init(void)
>         csr_write(CSR_SATP, PFN_DOWN(__pa(swapper_pg_dir)) | satp_mode);
>         local_flush_tlb_all();
>  }
> +
> +#ifdef CONFIG_KASAN_SW_TAGS
> +bool kasan_boot_cpu_enabled(void)
> +{
> +       return __kasan_boot_cpu_enabled;
> +}
> +
> +int kasan_cpu_enable(void)
> +{
> +       struct sbiret ret;
> +
> +       ret = sbi_ecall(SBI_EXT_FWFT, SBI_EXT_FWFT_SET,
> +                       SBI_FWFT_POINTER_MASKING_PMLEN, 7, 0, 0, 0, 0);
> +
> +       return sbi_err_map_linux_errno(ret.error);
> +}
> +#endif
> diff --git a/arch/riscv/mm/physaddr.c b/arch/riscv/mm/physaddr.c
> index 18706f457da7..6d1cf6ffd54e 100644
> --- a/arch/riscv/mm/physaddr.c
> +++ b/arch/riscv/mm/physaddr.c
> @@ -8,6 +8,8 @@
>
>  phys_addr_t __virt_to_phys(unsigned long x)
>  {
> +       x = __tag_reset(x);
> +
>         /*
>          * Boundary checking aginst the kernel linear mapping space.
>          */
> @@ -24,6 +26,8 @@ phys_addr_t __phys_addr_symbol(unsigned long x)
>         unsigned long kernel_start = kernel_map.virt_addr;
>         unsigned long kernel_end = kernel_start + kernel_map.size;
>
> +       x = __tag_reset(x);
> +
>         /*
>          * Boundary checking aginst the kernel image mapping.
>          * __pa_symbol should only be used on kernel symbol addresses.
> diff --git a/scripts/Makefile.kasan b/scripts/Makefile.kasan
> index 693dbbebebba..72a8c9d5fb0e 100644
> --- a/scripts/Makefile.kasan
> +++ b/scripts/Makefile.kasan
> @@ -91,6 +91,11 @@ ifeq ($(call clang-min-version, 150000)$(call gcc-min-version, 130000),y)
>         kasan_params += hwasan-kernel-mem-intrinsic-prefix=1
>  endif
>
> +# RISC-V requires dynamically determining if stack tagging can be enabled.

Please be more explicit here, something like: "RISC-V requires
dynamically determining if stack tagging can be enabled and thus
cannot allow the compiler to generate tags via inlined code."


> +ifdef CONFIG_RISCV
> +       kasan_params += hwasan-generate-tags-with-calls=1
> +endif
> +
>  endif # CONFIG_KASAN_SW_TAGS
>
>  # Add all as-supported KASAN LLVM parameters requested by the configuration.
> --
> 2.45.1
>