[PATCH mm] kasan, arm64: use ARCH_SLAB_MINALIGN instead of manual aligning

[PATCH mm] kasan, arm64: use ARCH_SLAB_MINALIGN instead of manual aligning

[Andrey Konovalov]

Instead of changing cache->align to be aligned to KASAN_SHADOW_SCALE_SIZE
in kasan_cache_create() we can reuse the ARCH_SLAB_MINALIGN macro.

Suggested-by: Vincenzo Frascino <vincenzo.frascino@arm.com>
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
---
 arch/arm64/include/asm/kasan.h | 4 ++++
 mm/kasan/common.c              | 2 --
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/include/asm/kasan.h b/arch/arm64/include/asm/kasan.h
index b52aacd2c526..ba26150d578d 100644
--- a/arch/arm64/include/asm/kasan.h
+++ b/arch/arm64/include/asm/kasan.h
@@ -36,6 +36,10 @@
 #define KASAN_SHADOW_OFFSET     (KASAN_SHADOW_END - (1ULL << \
 					(64 - KASAN_SHADOW_SCALE_SHIFT)))
 
+#ifdef CONFIG_KASAN_SW_TAGS
+#define ARCH_SLAB_MINALIGN	(1ULL << KASAN_SHADOW_SCALE_SHIFT)
+#endif
+
 void kasan_init(void);
 void kasan_copy_shadow(pgd_t *pgdir);
 asmlinkage void kasan_early_init(void);
diff --git a/mm/kasan/common.c b/mm/kasan/common.c
index 03d5d1374ca7..44390392d4c9 100644
--- a/mm/kasan/common.c
+++ b/mm/kasan/common.c
@@ -298,8 +298,6 @@ void kasan_cache_create(struct kmem_cache *cache, unsigned int *size,
 		return;
 	}
 
-	cache->align = round_up(cache->align, KASAN_SHADOW_SCALE_SIZE);
-
 	*flags |= SLAB_KASAN;
 }
 
-- 
2.20.0.405.gbc1bbc6f85-goog

Re: [PATCH mm] kasan, arm64: use ARCH_SLAB_MINALIGN instead of manual aligning

[Andrew Morton]

On Tue, 18 Dec 2018 14:30:33 +0100 Andrey Konovalov <andreyknvl@google.com> wrote:

> Instead of changing cache->align to be aligned to KASAN_SHADOW_SCALE_SIZE
> in kasan_cache_create() we can reuse the ARCH_SLAB_MINALIGN macro.
> 
> ...
>
> --- a/arch/arm64/include/asm/kasan.h
> +++ b/arch/arm64/include/asm/kasan.h
> @@ -36,6 +36,10 @@
>  #define KASAN_SHADOW_OFFSET     (KASAN_SHADOW_END - (1ULL << \
>  					(64 - KASAN_SHADOW_SCALE_SHIFT)))
>  
> +#ifdef CONFIG_KASAN_SW_TAGS
> +#define ARCH_SLAB_MINALIGN	(1ULL << KASAN_SHADOW_SCALE_SHIFT)
> +#endif
> +
>  void kasan_init(void);
>  void kasan_copy_shadow(pgd_t *pgdir);
>  asmlinkage void kasan_early_init(void);

This looks unreliable.  include/linux/slab.h has

/*
 * Setting ARCH_SLAB_MINALIGN in arch headers allows a different alignment.
 * Intended for arches that get misalignment faults even for 64 bit integer
 * aligned buffers.
 */
#ifndef ARCH_SLAB_MINALIGN
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
#endif

so if a .c file includes arch/arm64/include/asm/kasan.h after
include/linux/slab.h, it can get a macro-redefined warning.  If the .c
file includes those headers in the other order, ARCH_SLAB_MINALIGN will
get a different value compared to other .c files.

Or something like that.

Different architectures define ARCH_SLAB_MINALIGN in different place:

./arch/microblaze/include/asm/page.h:#define ARCH_SLAB_MINALIGN L1_CACHE_BYTES
./arch/arm/include/asm/cache.h:#define ARCH_SLAB_MINALIGN 8
./arch/sh/include/asm/page.h:#define ARCH_SLAB_MINALIGN 8
./arch/c6x/include/asm/cache.h:#define ARCH_SLAB_MINALIGN       L1_CACHE_BYTES
./arch/sparc/include/asm/cache.h:#define ARCH_SLAB_MINALIGN     __alignof__(unsigned long long)
./arch/xtensa/include/asm/processor.h:#define ARCH_SLAB_MINALIGN STACK_ALIGN

which is rather bad of us.

But still.  I think your definition should occur in an arch header file
which is reliably included from slab.h.  And kasan code should get its
definition of ARCH_SLAB_MINALIGN by including slab.h.

Re: [PATCH mm] kasan, arm64: use ARCH_SLAB_MINALIGN instead of manual aligning

[Andrey Konovalov]

On Tue, Dec 18, 2018 at 9:55 PM Andrew Morton <akpm@linux-foundation.org> wrote:
>
> On Tue, 18 Dec 2018 14:30:33 +0100 Andrey Konovalov <andreyknvl@google.com> wrote:
>
> > Instead of changing cache->align to be aligned to KASAN_SHADOW_SCALE_SIZE
> > in kasan_cache_create() we can reuse the ARCH_SLAB_MINALIGN macro.
> >
> > ...
> >
> > --- a/arch/arm64/include/asm/kasan.h
> > +++ b/arch/arm64/include/asm/kasan.h
> > @@ -36,6 +36,10 @@
> >  #define KASAN_SHADOW_OFFSET     (KASAN_SHADOW_END - (1ULL << \
> >                                       (64 - KASAN_SHADOW_SCALE_SHIFT)))
> >
> > +#ifdef CONFIG_KASAN_SW_TAGS
> > +#define ARCH_SLAB_MINALIGN   (1ULL << KASAN_SHADOW_SCALE_SHIFT)
> > +#endif
> > +
> >  void kasan_init(void);
> >  void kasan_copy_shadow(pgd_t *pgdir);
> >  asmlinkage void kasan_early_init(void);
>
> This looks unreliable.  include/linux/slab.h has
>
> /*
>  * Setting ARCH_SLAB_MINALIGN in arch headers allows a different alignment.
>  * Intended for arches that get misalignment faults even for 64 bit integer
>  * aligned buffers.
>  */
> #ifndef ARCH_SLAB_MINALIGN
> #define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
> #endif
>
> so if a .c file includes arch/arm64/include/asm/kasan.h after
> include/linux/slab.h, it can get a macro-redefined warning.  If the .c
> file includes those headers in the other order, ARCH_SLAB_MINALIGN will
> get a different value compared to other .c files.
>
> Or something like that.
>
> Different architectures define ARCH_SLAB_MINALIGN in different place:
>
> ./arch/microblaze/include/asm/page.h:#define ARCH_SLAB_MINALIGN L1_CACHE_BYTES
> ./arch/arm/include/asm/cache.h:#define ARCH_SLAB_MINALIGN 8
> ./arch/sh/include/asm/page.h:#define ARCH_SLAB_MINALIGN 8
> ./arch/c6x/include/asm/cache.h:#define ARCH_SLAB_MINALIGN       L1_CACHE_BYTES
> ./arch/sparc/include/asm/cache.h:#define ARCH_SLAB_MINALIGN     __alignof__(unsigned long long)
> ./arch/xtensa/include/asm/processor.h:#define ARCH_SLAB_MINALIGN STACK_ALIGN
>
> which is rather bad of us.
>
> But still.  I think your definition should occur in an arch header file
> which is reliably included from slab.h.  And kasan code should get its
> definition of ARCH_SLAB_MINALIGN by including slab.h.
>

KASAN code doesn't use this macro directly, so I don't think it needs
to get it's definition.

What do you think about adding #include <linux/kasan.h> into
linux/slab.h? Perhaps with a comment that this is needed to get
definition of ARCH_SLAB_MINALIGN?