[PATCH 1/2] mm: slab: fix comment for ARCH_KMALLOC_MINALIGN

[PATCH 1/2] mm: slab: fix comment for ARCH_KMALLOC_MINALIGN

[andrey.konovalov]

From: Andrey Konovalov <andreyknvl@google.com>

The comment next to the ARCH_KMALLOC_MINALIGN definition says that
ARCH_KMALLOC_MINALIGN can be defined in arch headers. This is incorrect:
it's actually ARCH_DMA_MINALIGN that can be defined there.

Fix the comment.

Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
---
 include/linux/slab.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/slab.h b/include/linux/slab.h
index 373b3ef99f4e..8cc1d54e56ad 100644
--- a/include/linux/slab.h
+++ b/include/linux/slab.h
@@ -190,7 +190,7 @@ void kmem_dump_obj(void *object);
 /*
  * Some archs want to perform DMA into kmalloc caches and need a guaranteed
  * alignment larger than the alignment of a 64-bit integer.
- * Setting ARCH_KMALLOC_MINALIGN in arch headers allows that.
+ * Setting ARCH_DMA_MINALIGN in arch headers allows that.
  */
 #if defined(ARCH_DMA_MINALIGN) && ARCH_DMA_MINALIGN > 8
 #define ARCH_KMALLOC_MINALIGN ARCH_DMA_MINALIGN
-- 
2.25.1

[PATCH 2/2] mm: slab: fix comment for __assume_kmalloc_alignment

[andrey.konovalov]

From: Andrey Konovalov <andreyknvl@google.com>

The comment next to the __assume_kmalloc_alignment definition is not
precise: kmalloc relies on kmem_cache_alloc, so kmalloc technically returns
pointers aligned to both ARCH_KMALLOC_MINALIGN and ARCH_SLAB_MINALIGN,
not only to ARCH_KMALLOC_MINALIGN.

(See create_kmalloc_cache()->create_boot_cache()->calculate_alignment()
 for SLAB and SLUB and __do_kmalloc_node() for SLOB.)

Clarify the comment.

The assumption specified by __assume_kmalloc_alignment is still correct,
although it can be made stronger. I'll leave this to a separate patch.

Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
---
 include/linux/slab.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/include/linux/slab.h b/include/linux/slab.h
index 8cc1d54e56ad..06323a4beff0 100644
--- a/include/linux/slab.h
+++ b/include/linux/slab.h
@@ -210,9 +210,9 @@ void kmem_dump_obj(void *object);
 #endif
 
 /*
- * kmalloc and friends return ARCH_KMALLOC_MINALIGN aligned
- * pointers. kmem_cache_alloc and friends return ARCH_SLAB_MINALIGN
- * aligned pointers.
+ * kmem_cache_alloc and friends return pointers aligned to ARCH_SLAB_MINALIGN.
+ * kmalloc and friends return pointers aligned to both ARCH_KMALLOC_MINALIGN
+ * and ARCH_SLAB_MINALIGN, but here we only assume the former alignment.
  */
 #define __assume_kmalloc_alignment __assume_aligned(ARCH_KMALLOC_MINALIGN)
 #define __assume_slab_alignment __assume_aligned(ARCH_SLAB_MINALIGN)
-- 
2.25.1

Re: [PATCH 1/2] mm: slab: fix comment for ARCH_KMALLOC_MINALIGN

[Christoph Lameter]

On Thu, 28 Apr 2022, andrey.konovalov@linux.dev wrote:

> From: Andrey Konovalov <andreyknvl@google.com>
>
> The comment next to the ARCH_KMALLOC_MINALIGN definition says that
> ARCH_KMALLOC_MINALIGN can be defined in arch headers. This is incorrect:
> it's actually ARCH_DMA_MINALIGN that can be defined there.

The section is talking about kmalloc and not general kmem_cache_alloc()
and there was the intention of separating the alignment requirements
between these two groups of caches in the slab allocators.

Re: [PATCH 1/2] mm: slab: fix comment for ARCH_KMALLOC_MINALIGN

[Vlastimil Babka]

On 4/28/22 17:59, andrey.konovalov@linux.dev wrote:
> From: Andrey Konovalov <andreyknvl@google.com>
> 
> The comment next to the ARCH_KMALLOC_MINALIGN definition says that
> ARCH_KMALLOC_MINALIGN can be defined in arch headers. This is incorrect:
> it's actually ARCH_DMA_MINALIGN that can be defined there.
> 
> Fix the comment.
> 
> Signed-off-by: Andrey Konovalov <andreyknvl@google.com>

Thanks, added both patches to the slab tree.

> ---
>  include/linux/slab.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/include/linux/slab.h b/include/linux/slab.h
> index 373b3ef99f4e..8cc1d54e56ad 100644
> --- a/include/linux/slab.h
> +++ b/include/linux/slab.h
> @@ -190,7 +190,7 @@ void kmem_dump_obj(void *object);
>  /*
>   * Some archs want to perform DMA into kmalloc caches and need a guaranteed
>   * alignment larger than the alignment of a 64-bit integer.
> - * Setting ARCH_KMALLOC_MINALIGN in arch headers allows that.
> + * Setting ARCH_DMA_MINALIGN in arch headers allows that.
>   */
>  #if defined(ARCH_DMA_MINALIGN) && ARCH_DMA_MINALIGN > 8
>  #define ARCH_KMALLOC_MINALIGN ARCH_DMA_MINALIGN

Re: [PATCH 1/2] mm: slab: fix comment for ARCH_KMALLOC_MINALIGN

[David Rientjes]

On Thu, 28 Apr 2022, andrey.konovalov@linux.dev wrote:

> From: Andrey Konovalov <andreyknvl@google.com>
> 
> The comment next to the ARCH_KMALLOC_MINALIGN definition says that
> ARCH_KMALLOC_MINALIGN can be defined in arch headers. This is incorrect:
> it's actually ARCH_DMA_MINALIGN that can be defined there.
> 
> Fix the comment.
> 
> Signed-off-by: Andrey Konovalov <andreyknvl@google.com>

Acked-by: David Rientjes <rientjes@google.com>

Re: [PATCH 2/2] mm: slab: fix comment for __assume_kmalloc_alignment

[David Rientjes]

On Thu, 28 Apr 2022, andrey.konovalov@linux.dev wrote:

> From: Andrey Konovalov <andreyknvl@google.com>
> 
> The comment next to the __assume_kmalloc_alignment definition is not
> precise: kmalloc relies on kmem_cache_alloc, so kmalloc technically returns
> pointers aligned to both ARCH_KMALLOC_MINALIGN and ARCH_SLAB_MINALIGN,
> not only to ARCH_KMALLOC_MINALIGN.
> 
> (See create_kmalloc_cache()->create_boot_cache()->calculate_alignment()
>  for SLAB and SLUB and __do_kmalloc_node() for SLOB.)
> 
> Clarify the comment.
> 
> The assumption specified by __assume_kmalloc_alignment is still correct,
> although it can be made stronger. I'll leave this to a separate patch.
> 
> Signed-off-by: Andrey Konovalov <andreyknvl@google.com>

Acked-by: David Rientjes <rientjes@google.com>