Re: [PATCH v4 08/22] slab: make percpu sheaves compatible with kmalloc_nolock()/kfree_nolock()

[Vlastimil Babka]

On 1/27/26 18:36, Liam R. Howlett wrote:
> * Vlastimil Babka <vbabka@xxxxxxx> [260123 01:54]:
>> Before we enable percpu sheaves for kmalloc caches, we need to make sure
>> kmalloc_nolock() and kfree_nolock() will continue working properly and
>> not spin when not allowed to.
>> 
>> Percpu sheaves themselves use local_trylock() so they are already
>> compatible. We just need to be careful with the barn->lock spin_lock.
>> Pass a new allow_spin parameter where necessary to use
>> spin_trylock_irqsave().
>> 
>> In kmalloc_nolock_noprof() we can now attempt alloc_from_pcs() safely,
>> for now it will always fail until we enable sheaves for kmalloc caches
>> next. Similarly in kfree_nolock() we can attempt free_to_pcs().
>> 
>> Reviewed-by: Suren Baghdasaryan <surenb@xxxxxxxxxx>
>> Reviewed-by: Harry Yoo <harry.yoo@xxxxxxxxxx>
>> Reviewed-by: Hao Li <hao.li@xxxxxxxxx>
>> Signed-off-by: Vlastimil Babka <vbabka@xxxxxxx>
> 
> I'd rather have a helper that _mayspin() or something.  That way, when
> we know we need to or don't want to, we avoid extra instructions.  We
> can also avoid passing true/false and complicating the interface when
> it's not necessary.

Can you elaborate a bit more how the helper looks like and it's used? I
don't follow immediately. We do have the helper for checking gfp flags to
extract mayspin, but that wouldn't avoid extra instructions so you must be
thinking of something else.

> The way it is written works as well, though.
> 
> Reviewed-by: Liam R. Howlett <Liam.Howlett@xxxxxxxxxx>

Thanks!