Re: [PATCH RFC 05/19] slab: add sheaves to most caches

[Vlastimil Babka]

On 10/27/25 01:24, Harry Yoo wrote:
> On Thu, Oct 23, 2025 at 03:52:27PM +0200, Vlastimil Babka wrote:
>> In the first step to replace cpu (partial) slabs with sheaves, enable
>> sheaves for almost all caches. Treat args->sheaf_capacity as a minimum,
>> and calculate sheaf capacity with a formula that roughly follows the
>> formula for number of objects in cpu partial slabs in set_cpu_partial().
> 
> Should we scale sheaf capacity not only based on object size but also
> on the number of CPUs, like calculate_order() does?

We can try that as a follow-up, right now it's trying to roughly match the
pre-existing amount of caching so that bots hopefully won't report
regressions just because it became smaller (like we've already seen for
maple nodes).

>> This should achieve roughly similar contention on the barn spin lock as
>> there's currently for node list_lock without sheaves, to make
>> benchmarking results comparable. It can be further tuned later.
>> 
>> Don't enable sheaves for kmalloc caches yet, as that needs further
>> changes to bootstraping.
>> 
>> Signed-off-by: Vlastimil Babka <vbabka@xxxxxxx>
>> ---
>