Re: [PATCH] slob: add size header to all allocations

[Vlastimil Babka]

On 10/24/21 12:43, Hyeonggon Yoo wrote:
>> >> 
>> >> The main reason for this change is to simplify SLOB a little bit, make
>> >> it a bit easier to debug whenever something goes wrong.
>> >>
>> > 
>> > It seems acceptable But I wonder it is worth to increase memory usage
>> > to allow freeing kmem_cache_alloc-ed objects by kfree()?
>> 
>> Not for the reason above, but for providing a useful API guarantee
>> regardless of selected slab allocator IMHO yes.
>> 
> 
> Mm.. that means some callers free kmem_cache_alloc-ed object using
> kfree, and SLAB/SLUB already support that, and SLOB doesn't.

Exactly. Finding that out started this whole thread.

> In what situations is freeing using kfree needed?
> Wouldn't this make code confusing?

XFS seems to have good reasons - at some common freeing place objects can
appears from multiple caches, and it would be expensive to track their cache
just to free them. See
https://lore.kernel.org/all/20210930044202.GP2361455@xxxxxxxxxxxxxxxxxxx/

IMHO it really makes sense to support this from API point of view.
kmem_cache_alloc() is basically a more specific version of the generic
kmalloc(). It makes sense if the generic kind of free, that is kfree() works
on those objects too.

>> > Thanks,
>> > Hyeonggon
>> > 
>> >> meminfo right after the system boot, without the patch:
>> >> Slab:              35500 kB
>> >> 
>> >> the same, with the patch:
>> >> Slab:              36396 kB
>> >> 
>> > 
>> 
>