Re: improve the kmem_cache_alloc_bulk API
"Vlastimil Babka (SUSE)" <[email protected]>
| Newsgroups | org.kernel.vger.io-uring,org.freedesktop.lists.dri-devel,org.kernel.vger.bpf,org.kernel.vger.linux-arm-msm,org.kernel.vger.linux-kernel,org.kernel.vger.netdev,org.kvack.linux-mm |
|---|---|
| Message-ID | <[email protected]> |
On 5/28/26 11:05, Christoph Hellwig wrote:
> On Wed, May 27, 2026 at 04:07:12PM +0200, Vlastimil Babka (SUSE) wrote:
>> commit 46dea1744498 ("slab: refill sheaves from all nodes") from this January.
>> Previously it was just interrupts enabled.
>>
>> > but by requiring separate functions so I somehow doubt that was meant.
>>
>> Yeah, it's expressed by the _nolock variants. But slab propagates it internally
>> by the gfp flags, and since 46dea1744498 it affects kmem_cache_alloc_bulk().
>
> So what are the GFP flags that affect spinning? I can't find anything
> related to that in either Documentation/core-api/memory-allocation.rst or
> include/linux/gfp_types.h.
See gfpflags_allow_spinning(). The usage by slab and page allocator users
was meant to be via _nolock() variants, not by removing the flags when
calling the normal functions, hence it's not documented. But the abstraction
has now leaked in the kmem_cache_alloc_bulk() case, hm. I think instead of
documenting that, we should adjust the implementation in slab, if possible.
I'll try to look into that.