BACKPORT: UPSTREAM: mm: fs: invalidate bh_lrus for only cold path

kernel test robot reported the regression of fio.write_iops[1] with [2].

Since lru_add_drain is called frequently, invalidate bh_lrus there could
increase bh_lrus cache miss ratio, which needs more IO in the end.

This patch moves the bh_lrus invalidation from the hot path( e.g.,
zap_page_range, pagevec_release) to cold path(i.e., lru_add_drain_all,
lru_cache_disable).

"Xing, Zhengjun" confirmed

: I test the patch, the regression reduced to -2.9%.

[1] https://lore.kernel.org/lkml/20210520083144.GD14190@xsang-OptiPlex-9020/
[2] 8cc621d2f4, mm: fs: invalidate BH LRU during page migration

Bug: 194673488
Link: https://lkml.kernel.org/r/20210907212347.1977686-1-minchan@kernel.org
(cherry picked from commit 243418e392)
[Chris: resolved conflicts due to Minchan's AOSP LRU commits]
Signed-off-by: Minchan Kim <minchan@kernel.org>
Reported-by: kernel test robot <oliver.sang@intel.com>
Reviewed-by: Chris Goldsworthy <cgoldswo@codeaurora.org>
Tested-by: "Xing, Zhengjun" <zhengjun.xing@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Goldsworthy <quic_cgoldswo@quicinc.com>
Change-Id: Icc5e456b058df516480b4378853464d6d7b43505
This commit is contained in:
Minchan Kim
2021-09-24 18:19:40 -07:00
committed by Chris Goldsworthy
parent 49af2e35d5
commit a9ac6ae90e
3 changed files with 24 additions and 7 deletions

View File

@@ -1454,12 +1454,16 @@ void invalidate_bh_lrus(void)
}
EXPORT_SYMBOL_GPL(invalidate_bh_lrus);
void invalidate_bh_lrus_cpu(int cpu)
/*
* It's called from workqueue context so we need a bh_lru_lock to close
* the race with preemption/irq.
*/
void invalidate_bh_lrus_cpu(void)
{
struct bh_lru *b;
bh_lru_lock();
b = per_cpu_ptr(&bh_lrus, cpu);
b = this_cpu_ptr(&bh_lrus);
__invalidate_bh_lrus(b);
bh_lru_unlock();
}