[PATCH v6 0/2] kasan: solve redzone overwritten issue at debug

[PATCH v6 0/2] kasan: solve redzone overwritten issue at debug

[yee.lee]

From: Yee Lee <yee.lee@mediatek.com>

Issue: In SLUB debug, hwtag kasan_unpoison() would overwrite the redzone
in those objects with unaligned size.

The first patch Introduces slub_debug_enable_unlikely() to check
the state of debug mode.

The second patch Adds memzero_explict() to separate the initialization for
such condition. The new code path is executed about 1.1% during nromal
booting process. The penalty is acceptable since it only works in debug
mode.

=============
Exp: QEMUv5.2(+mte)/SLUB_debug mode
code path exec : 941/80854 (1.1%)

Changed since v6:
 - Move helper instead of introducing a new one.

Changed since v5:
 - Fix format

Changed since v4:
 - Introduce slub_debug_enable_unlikly() to check the debug state.
 - Include "slab.h" and Add slub_debug_enable_unlikly() to lead
   the condition statement.
 - Add comment block about this new code path in source code.

Changed since v3:
 - Apply IS_ENABLED to wrap codes under SLUB debug mode.
 - Replace memset() by memzero_explict().

---

Marco Elver (1):
  mm: move helper to check slub_debug_enabled

Yee Lee (1):
  kasan: Add memzero int for unaligned size at DEBUG

 mm/kasan/kasan.h | 12 ++++++++++++
 mm/slab.h        | 15 +++++++++++----
 mm/slub.c        | 14 --------------
 3 files changed, 23 insertions(+), 18 deletions(-)

-- 
2.18.0

[PATCH v6 1/2] mm: move helper to check slub_debug_enabled

[yee.lee]

From: Marco Elver <elver@google.com>

Move the helper to check slub_debug_enabled, so that we can confine
the use of #ifdef outside slub.c as well.

Suggested-by: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Yee Lee <yee.lee@mediatek.com>
---
 mm/slab.h | 15 +++++++++++----
 mm/slub.c | 14 --------------
 2 files changed, 11 insertions(+), 18 deletions(-)

diff --git a/mm/slab.h b/mm/slab.h
index 7b60ef2f32c3..67d3272f3368 100644
--- a/mm/slab.h
+++ b/mm/slab.h
@@ -216,10 +216,18 @@ DECLARE_STATIC_KEY_FALSE(slub_debug_enabled);
 #endif
 extern void print_tracking(struct kmem_cache *s, void *object);
 long validate_slab_cache(struct kmem_cache *s);
+static inline bool __slub_debug_enabled(void)
+{
+	return static_branch_unlikely(&slub_debug_enabled);
+}
 #else
 static inline void print_tracking(struct kmem_cache *s, void *object)
 {
 }
+static inline bool __slub_debug_enabled(void)
+{
+	return false;
+}
 #endif
 
 /*
@@ -229,11 +237,10 @@ static inline void print_tracking(struct kmem_cache *s, void *object)
  */
 static inline bool kmem_cache_debug_flags(struct kmem_cache *s, slab_flags_t flags)
 {
-#ifdef CONFIG_SLUB_DEBUG
-	VM_WARN_ON_ONCE(!(flags & SLAB_DEBUG_FLAGS));
-	if (static_branch_unlikely(&slub_debug_enabled))
+	if (IS_ENABLED(CONFIG_SLUB_DEBUG))
+		VM_WARN_ON_ONCE(!(flags & SLAB_DEBUG_FLAGS));
+	if (__slub_debug_enabled())
 		return s->flags & flags;
-#endif
 	return false;
 }
 
diff --git a/mm/slub.c b/mm/slub.c
index 3bc8b940c933..71515c484d1c 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -119,25 +119,11 @@
  */
 
 #ifdef CONFIG_SLUB_DEBUG
-
 #ifdef CONFIG_SLUB_DEBUG_ON
 DEFINE_STATIC_KEY_TRUE(slub_debug_enabled);
 #else
 DEFINE_STATIC_KEY_FALSE(slub_debug_enabled);
 #endif
-
-static inline bool __slub_debug_enabled(void)
-{
-	return static_branch_unlikely(&slub_debug_enabled);
-}
-
-#else		/* CONFIG_SLUB_DEBUG */
-
-static inline bool __slub_debug_enabled(void)
-{
-	return false;
-}
-
 #endif		/* CONFIG_SLUB_DEBUG */
 
 static inline bool kmem_cache_debug(struct kmem_cache *s)
-- 
2.18.0

[PATCH v6 2/2] kasan: Add memzero int for unaligned size at DEBUG

[yee.lee]

From: Yee Lee <yee.lee@mediatek.com>

Issue: when SLUB debug is on, hwtag kasan_unpoison() would overwrite
the redzone of object with unaligned size.

An additional memzero_explicit() path is added to replacing init by
hwtag instruction for those unaligned size at SLUB debug mode.

The penalty is acceptable since they are only enabled in debug mode,
not production builds. A block of comment is added for explanation.

Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Suggested-by: Marco Elver <elver@google.com>
Suggested-by: Andrey Konovalov <andreyknvl@gmail.com>
Signed-off-by: Yee Lee <yee.lee@mediatek.com>
---
 mm/kasan/kasan.h | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index 98e3059bfea4..d739cdd1621a 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -9,6 +9,7 @@
 #ifdef CONFIG_KASAN_HW_TAGS
 
 #include <linux/static_key.h>
+#include "../slab.h"
 
 DECLARE_STATIC_KEY_FALSE(kasan_flag_stacktrace);
 extern bool kasan_flag_async __ro_after_init;
@@ -387,6 +388,17 @@ static inline void kasan_unpoison(const void *addr, size_t size, bool init)
 
 	if (WARN_ON((unsigned long)addr & KASAN_GRANULE_MASK))
 		return;
+	/*
+	 * Explicitly initialize the memory with the precise object size to
+	 * avoid overwriting the SLAB redzone. This disables initialization in
+	 * the arch code and may thus lead to performance penalty. The penalty
+	 * is accepted since SLAB redzones aren't enabled in production builds.
+	 */
+	if (__slub_debug_enabled() &&
+	    init && ((unsigned long)size & KASAN_GRANULE_MASK)) {
+		init = false;
+		memzero_explicit((void *)addr, size);
+	}
 	size = round_up(size, KASAN_GRANULE_SIZE);
 
 	hw_set_mem_tag_range((void *)addr, size, tag, init);
-- 
2.18.0

Re: [PATCH v6 2/2] kasan: Add memzero int for unaligned size at DEBUG

[Marco Elver]

On Mon, 5 Jul 2021 at 12:33, <yee.lee@mediatek.com> wrote:
> From: Yee Lee <yee.lee@mediatek.com>
>
> Issue: when SLUB debug is on, hwtag kasan_unpoison() would overwrite
> the redzone of object with unaligned size.
>
> An additional memzero_explicit() path is added to replacing init by
> hwtag instruction for those unaligned size at SLUB debug mode.
>
> The penalty is acceptable since they are only enabled in debug mode,
> not production builds. A block of comment is added for explanation.
>
> Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
> Cc: Alexander Potapenko <glider@google.com>
> Cc: Dmitry Vyukov <dvyukov@google.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Suggested-by: Marco Elver <elver@google.com>
> Suggested-by: Andrey Konovalov <andreyknvl@gmail.com>
> Signed-off-by: Yee Lee <yee.lee@mediatek.com>

Reviewed-by: Marco Elver <elver@google.com>

Thank you!

> ---
>  mm/kasan/kasan.h | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
>
> diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
> index 98e3059bfea4..d739cdd1621a 100644
> --- a/mm/kasan/kasan.h
> +++ b/mm/kasan/kasan.h
> @@ -9,6 +9,7 @@
>  #ifdef CONFIG_KASAN_HW_TAGS
>
>  #include <linux/static_key.h>
> +#include "../slab.h"
>
>  DECLARE_STATIC_KEY_FALSE(kasan_flag_stacktrace);
>  extern bool kasan_flag_async __ro_after_init;
> @@ -387,6 +388,17 @@ static inline void kasan_unpoison(const void *addr, size_t size, bool init)
>
>         if (WARN_ON((unsigned long)addr & KASAN_GRANULE_MASK))
>                 return;
> +       /*
> +        * Explicitly initialize the memory with the precise object size to
> +        * avoid overwriting the SLAB redzone. This disables initialization in
> +        * the arch code and may thus lead to performance penalty. The penalty
> +        * is accepted since SLAB redzones aren't enabled in production builds.
> +        */
> +       if (__slub_debug_enabled() &&
> +           init && ((unsigned long)size & KASAN_GRANULE_MASK)) {
> +               init = false;
> +               memzero_explicit((void *)addr, size);
> +       }
>         size = round_up(size, KASAN_GRANULE_SIZE);
>
>         hw_set_mem_tag_range((void *)addr, size, tag, init);
> --
> 2.18.0

Re: [PATCH v6 2/2] kasan: Add memzero int for unaligned size at DEBUG

[Andrey Konovalov]

On Mon, Jul 5, 2021 at 12:33 PM <yee.lee@mediatek.com> wrote:
>
> From: Yee Lee <yee.lee@mediatek.com>
>
> Issue: when SLUB debug is on, hwtag kasan_unpoison() would overwrite
> the redzone of object with unaligned size.
>
> An additional memzero_explicit() path is added to replacing init by
> hwtag instruction for those unaligned size at SLUB debug mode.
>
> The penalty is acceptable since they are only enabled in debug mode,
> not production builds. A block of comment is added for explanation.
>
> Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
> Cc: Alexander Potapenko <glider@google.com>
> Cc: Dmitry Vyukov <dvyukov@google.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Suggested-by: Marco Elver <elver@google.com>
> Suggested-by: Andrey Konovalov <andreyknvl@gmail.com>
> Signed-off-by: Yee Lee <yee.lee@mediatek.com>
> ---
>  mm/kasan/kasan.h | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
>
> diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
> index 98e3059bfea4..d739cdd1621a 100644
> --- a/mm/kasan/kasan.h
> +++ b/mm/kasan/kasan.h
> @@ -9,6 +9,7 @@
>  #ifdef CONFIG_KASAN_HW_TAGS
>
>  #include <linux/static_key.h>
> +#include "../slab.h"
>
>  DECLARE_STATIC_KEY_FALSE(kasan_flag_stacktrace);
>  extern bool kasan_flag_async __ro_after_init;
> @@ -387,6 +388,17 @@ static inline void kasan_unpoison(const void *addr, size_t size, bool init)
>
>         if (WARN_ON((unsigned long)addr & KASAN_GRANULE_MASK))
>                 return;
> +       /*
> +        * Explicitly initialize the memory with the precise object size to
> +        * avoid overwriting the SLAB redzone. This disables initialization in
> +        * the arch code and may thus lead to performance penalty. The penalty
> +        * is accepted since SLAB redzones aren't enabled in production builds.
> +        */
> +       if (__slub_debug_enabled() &&

What happened to slub_debug_enabled_unlikely()? Was it renamed? Why? I
didn't receive patch #1 of v6 (nor of v5).

> +           init && ((unsigned long)size & KASAN_GRANULE_MASK)) {
> +               init = false;
> +               memzero_explicit((void *)addr, size);
> +       }
>         size = round_up(size, KASAN_GRANULE_SIZE);
>
>         hw_set_mem_tag_range((void *)addr, size, tag, init);
> --
> 2.18.0
>

Re: [PATCH v6 2/2] kasan: Add memzero int for unaligned size at DEBUG

[Marco Elver]

On Mon, 5 Jul 2021 at 13:12, Andrey Konovalov <andreyknvl@gmail.com> wrote:
[...]
> > +       /*
> > +        * Explicitly initialize the memory with the precise object size to
> > +        * avoid overwriting the SLAB redzone. This disables initialization in
> > +        * the arch code and may thus lead to performance penalty. The penalty
> > +        * is accepted since SLAB redzones aren't enabled in production builds.
> > +        */
> > +       if (__slub_debug_enabled() &&
>
> What happened to slub_debug_enabled_unlikely()? Was it renamed? Why? I
> didn't receive patch #1 of v6 (nor of v5).

Somebody had the same idea with the helper:
https://lkml.kernel.org/r/YOKsC75kJfCZwySD@elver.google.com
and Matthew didn't like the _unlikely() prefix.

Which meant we should just move the existing helper introduced in the
merge window.

Patch 1/2: https://lkml.kernel.org/r/20210705103229.8505-2-yee.lee@mediatek.com

Thanks,
-- Marco

Re: [PATCH v6 2/2] kasan: Add memzero int for unaligned size at DEBUG

[Andrey Konovalov]

On Mon, Jul 5, 2021 at 1:18 PM Marco Elver <elver@google.com> wrote:
>
> On Mon, 5 Jul 2021 at 13:12, Andrey Konovalov <andreyknvl@gmail.com> wrote:
> [...]
> > > +       /*
> > > +        * Explicitly initialize the memory with the precise object size to
> > > +        * avoid overwriting the SLAB redzone. This disables initialization in
> > > +        * the arch code and may thus lead to performance penalty. The penalty
> > > +        * is accepted since SLAB redzones aren't enabled in production builds.
> > > +        */
> > > +       if (__slub_debug_enabled() &&
> >
> > What happened to slub_debug_enabled_unlikely()? Was it renamed? Why? I
> > didn't receive patch #1 of v6 (nor of v5).
>
> Somebody had the same idea with the helper:
> https://lkml.kernel.org/r/YOKsC75kJfCZwySD@elver.google.com
> and Matthew didn't like the _unlikely() prefix.
>
> Which meant we should just move the existing helper introduced in the
> merge window.
>
> Patch 1/2: https://lkml.kernel.org/r/20210705103229.8505-2-yee.lee@mediatek.com

Got it. Thank you, Marco!

Re: [PATCH v6 2/2] kasan: Add memzero int for unaligned size at DEBUG

[Yee Lee]

Thank you, Macro. I thought members in "suggested-by" would be put in
the list as well... And thank you guys for the review these days. 


@Andrew Motom
Hi Andrew, 

Could you help to push the patches? We are dealing with the issue and 
would backport to Android porject right after the action.
Appreciated!


BR,
Yee 


On Mon, 2021-07-05 at 13:23 +0200, Andrey Konovalov wrote:
> On Mon, Jul 5, 2021 at 1:18 PM Marco Elver <elver@google.com> wrote:
> > 
> > On Mon, 5 Jul 2021 at 13:12, Andrey Konovalov <andreyknvl@gmail.com
> > > wrote:
> > [...]
> > > > +       /*
> > > > +        * Explicitly initialize the memory with the precise
> > > > object size to
> > > > +        * avoid overwriting the SLAB redzone. This disables
> > > > initialization in
> > > > +        * the arch code and may thus lead to performance
> > > > penalty. The penalty
> > > > +        * is accepted since SLAB redzones aren't enabled in
> > > > production builds.
> > > > +        */
> > > > +       if (__slub_debug_enabled() &&
> > > 
> > > What happened to slub_debug_enabled_unlikely()? Was it renamed?
> > > Why? I
> > > didn't receive patch #1 of v6 (nor of v5).
> > 
> > Somebody had the same idea with the helper:
> > https://lkml.kernel.org/r/YOKsC75kJfCZwySD@elver.google.com
> > and Matthew didn't like the _unlikely() prefix.
> > 
> > Which meant we should just move the existing helper introduced in
> > the
> > merge window.
> > 
> > Patch 1/2: 
> > https://lkml.kernel.org/r/20210705103229.8505-2-yee.lee@mediatek.com
> 
> Got it. Thank you, Marco!

Re: [PATCH v6 2/2] kasan: Add memzero int for unaligned size at DEBUG

[Andrey Konovalov]

On Mon, Jul 5, 2021 at 12:33 PM <yee.lee@mediatek.com> wrote:
>
> From: Yee Lee <yee.lee@mediatek.com>
>
> Issue: when SLUB debug is on, hwtag kasan_unpoison() would overwrite
> the redzone of object with unaligned size.
>
> An additional memzero_explicit() path is added to replacing init by
> hwtag instruction for those unaligned size at SLUB debug mode.
>
> The penalty is acceptable since they are only enabled in debug mode,
> not production builds. A block of comment is added for explanation.
>
> Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
> Cc: Alexander Potapenko <glider@google.com>
> Cc: Dmitry Vyukov <dvyukov@google.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Suggested-by: Marco Elver <elver@google.com>
> Suggested-by: Andrey Konovalov <andreyknvl@gmail.com>
> Signed-off-by: Yee Lee <yee.lee@mediatek.com>
> ---
>  mm/kasan/kasan.h | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
>
> diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
> index 98e3059bfea4..d739cdd1621a 100644
> --- a/mm/kasan/kasan.h
> +++ b/mm/kasan/kasan.h
> @@ -9,6 +9,7 @@
>  #ifdef CONFIG_KASAN_HW_TAGS
>
>  #include <linux/static_key.h>
> +#include "../slab.h"
>
>  DECLARE_STATIC_KEY_FALSE(kasan_flag_stacktrace);
>  extern bool kasan_flag_async __ro_after_init;
> @@ -387,6 +388,17 @@ static inline void kasan_unpoison(const void *addr, size_t size, bool init)
>
>         if (WARN_ON((unsigned long)addr & KASAN_GRANULE_MASK))
>                 return;
> +       /*
> +        * Explicitly initialize the memory with the precise object size to
> +        * avoid overwriting the SLAB redzone. This disables initialization in
> +        * the arch code and may thus lead to performance penalty. The penalty
> +        * is accepted since SLAB redzones aren't enabled in production builds.
> +        */
> +       if (__slub_debug_enabled() &&
> +           init && ((unsigned long)size & KASAN_GRANULE_MASK)) {
> +               init = false;
> +               memzero_explicit((void *)addr, size);
> +       }
>         size = round_up(size, KASAN_GRANULE_SIZE);
>
>         hw_set_mem_tag_range((void *)addr, size, tag, init);
> --
> 2.18.0
>

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