Re: [PATCH] slab, slub: remove size disparity on debug kernel

[Shakeel Butt]

On Wed, Mar 14, 2018 at 1:43 AM, Vladimir Davydov
<vdavydov.dev@xxxxxxxxx> wrote:
> On Tue, Mar 13, 2018 at 10:36:52AM -0700, Shakeel Butt wrote:
>> On Tue, Mar 13, 2018 at 10:19 AM, Christopher Lameter <cl@xxxxxxxxx> wrote:
>> > On Tue, 13 Mar 2018, Shakeel Butt wrote:
>> >
>> >> However for SLUB in debug kernel, the sizes were same. On further
>> >> inspection it is found that SLUB always use kmem_cache.object_size to
>> >> measure the kmem_cache.size while SLAB use the given kmem_cache.size. In
>> >> the debug kernel the slab's size can be larger than its object_size.
>> >> Thus in the creation of non-root slab, the SLAB uses the root's size as
>> >> base to calculate the non-root slab's size and thus non-root slab's size
>> >> can be larger than the root slab's size. For SLUB, the non-root slab's
>> >> size is measured based on the root's object_size and thus the size will
>> >> remain same for root and non-root slab.
>> >
>> > Note that the object_size and size may differ for SLUB based on kernel
>> > parameters and slab configuration. For SLAB these are compilation options.
>> >
>>
>> Thanks for the explanation.
>>
>> >> @@ -379,7 +379,7 @@ struct kmem_cache *find_mergeable(unsigned int size, unsigned int align,
>> >>  }
>> >>
>> >>  static struct kmem_cache *create_cache(const char *name,
>> >> -             unsigned int object_size, unsigned int size, unsigned int align,
>> >> +             unsigned int object_size, unsigned int align,
>> >>               slab_flags_t flags, unsigned int useroffset,
>> >
>> > Why was both the size and object_size passed during cache creation in the
>> > first place? From the flags etc the slab logic should be able to compute
>> > the actual bytes required for each object and its metadata.
>> >
>>
>> +Vladimir
>>
>> I think it was introduced by 794b1248be4e7 ("memcg, slab: separate
>> memcg vs root cache creation paths") but I could not find out the
>> reason.
>
> This was a mistake - I missed that __kmem_cache_create() overwrites
> kmem_cache->size. Thanks for fixing this.

Thanks for confirming.

Andrew, can you please add following line to the patch commit message.

Fixes: 794b1248be4e ("memcg, slab: separate memcg vs root cache creation paths")