Commit e7ddd47b authored by Chengming Zhou's avatar Chengming Zhou Committed by Peng Zhang
Browse files

mm/slub: directly load freelist from cpu partial slab in the likely case 

mainline inclusion
from mainline-v6.9-rc1
commit 90b1e56641bbab801e22141c56aa79dc095a3764
category: performance
bugzilla: https://gitee.com/openeuler/kernel/issues/I9CSFJ
CVE: NA

Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=90b1e56641bbab801e22141c56aa79dc095a3764



--------------------------------

The likely case is that we get a usable slab from the cpu partial list,
we can directly load freelist from it and return back, instead of going
the other way that need more work, like reenable interrupt and recheck.

But we need to remove the "VM_BUG_ON(!new.frozen)" in get_freelist()
for reusing it, since cpu partial slab is not frozen. It seems
acceptable since it's only for debug purpose.

And get_freelist() also assumes it can return NULL if the freelist is
empty, which is not possible for the cpu partial slab case, so we
add "VM_BUG_ON(!freelist)" after get_freelist() to make it explicit.

There is some small performance improvement too, which shows by:
perf bench sched messaging -g 5 -t -l 100000

            mm-stable   slub-optimize
Total time      7.473    7.209

Signed-off-by: default avatarChengming Zhou <zhouchengming@bytedance.com>
Reviewed-by: default avatarVlastimil Babka <vbabka@suse.cz>
Signed-off-by: default avatarVlastimil Babka <vbabka@suse.cz>
Signed-off-by: default avatarZhangPeng <zhangpeng362@huawei.com>
parent 5ee630af

mm/slub.c

+11 −10
Original line number Diff line number Diff line
@@ -3004,7 +3004,6 @@ static inline void *get_freelist(struct kmem_cache *s, struct slab *slab) 
		counters = slab->counters;



		new.counters = counters;

		VM_BUG_ON(!new.frozen);



		new.inuse = slab->objects;

		new.frozen = freelist != NULL;
@@ -3176,18 +3175,20 @@ static void *___slab_alloc(struct kmem_cache *s, gfp_t gfpflags, int node, 


		slab = slub_percpu_partial(c);

		slub_set_percpu_partial(c, slab);

		local_unlock_irqrestore(&s->cpu_slab->lock, flags);



		if (likely(node_match(slab, node) &&

			   pfmemalloc_match(slab, gfpflags))) {

			c->slab = slab;

			freelist = get_freelist(s, slab);

			VM_BUG_ON(!freelist);

			stat(s, CPU_PARTIAL_ALLOC);

			goto load_freelist;

		}



		local_unlock_irqrestore(&s->cpu_slab->lock, flags);



		if (unlikely(!node_match(slab, node) ||

			     !pfmemalloc_match(slab, gfpflags))) {

		slab->next = NULL;

		__put_partials(s, slab);

			continue;

		}



		freelist = freeze_slab(s, slab);

		goto retry_load_slab;

	}

#endif