Re: [PATCH v2] mm: don't be stuck to rmap lock on reclaim path

[Minchan Kim <minchan@kernel.org>]

On Mon, May 09, 2022 at 11:54:49AM -0700, Andrew Morton wrote:
> On Mon,  9 May 2022 08:47:10 -0700 Minchan Kim <minchan@kernel.org> wrote:
> 
> > The rmap locks(i_mmap_rwsem and anon_vma->root->rwsem) could be
> > contented under memory pressure if processes keep working on
> > their vmas(e.g., fork, mmap, munmap). It makes reclaim path
> > stuck. In our real workload traces, we see kswapd is waiting the
> > lock for 300ms+(a sec as worst case) and it makes other processes
> > entering direct reclaim, which were also stuck on the lock.
> > 
> > This patch makes LRU aging path try_lock mode like shink_page_list
> > so the reclaim context will keep working with next LRU pages
> > without being stuck.
> > 
> > Since this patch introduces a new "contended" field as out-param
> > along with try_lock in-param in rmap_walk_control, it's not
> > immutable any longer if the try_lock is set so remove const
> > keywords on rmap related functions. Since rmap walking is already
> > expensive operation, I doubt the const would help sizable benefit(
> > And we didn't have it until 5.17).
> 
> Some quantitative testing results would be helpful.  Demonstrate
> the benefits of the patch?

In a heavy app workload in Android, trace shows following statistics.
It removes almost of lock contention from those rmap.

Before:

   max_dur(ms)  min_dur(ms)  max-min(dur)ms  avg_dur(ms)  sum_dur(ms)  count blocked_function
         1632            0            1631   151.542173        31672    209  page_lock_anon_vma_read
          601            0             601   145.544681        28817    198  rmap_walk_file

After:

   max_dur(ms)  min_dur(ms)  max-min(dur)ms  avg_dur(ms)  sum_dur(ms)  count blocked_function
          NaN          NaN              NaN          NaN          NaN    0.0             NaN
            0            0                0     0.127645            1     12  rmap_walk_file


I will include this data in the description.