Re: [PATCH v2 2/7] mm: vmscan: make global slab shrink lockless

[Sultan Alsawaf <sultan@kerneltoast.com>]

On Thu, Feb 23, 2023 at 10:39:17AM -0800, Paul E. McKenney wrote:
> On Thu, Feb 23, 2023 at 10:24:47AM -0800, Sultan Alsawaf wrote:
> > On Thu, Feb 23, 2023 at 09:27:20PM +0800, Qi Zheng wrote:
> > > The shrinker_rwsem is a global lock in shrinkers subsystem,
> > > it is easy to cause blocking in the following cases:
> > > 
> > > a. the write lock of shrinker_rwsem was held for too long.
> > >    For example, there are many memcgs in the system, which
> > >    causes some paths to hold locks and traverse it for too
> > >    long. (e.g. expand_shrinker_info())
> > > b. the read lock of shrinker_rwsem was held for too long,
> > >    and a writer came at this time. Then this writer will be
> > >    forced to wait and block all subsequent readers.
> > >    For example:
> > >    - be scheduled when the read lock of shrinker_rwsem is
> > >      held in do_shrink_slab()
> > >    - some shrinker are blocked for too long. Like the case
> > >      mentioned in the patchset[1].
> > > 
> > > Therefore, many times in history ([2],[3],[4],[5]), some
> > > people wanted to replace shrinker_rwsem reader with SRCU,
> > > but they all gave up because SRCU was not unconditionally
> > > enabled.
> > > 
> > > But now, since commit 1cd0bd06093c ("rcu: Remove CONFIG_SRCU"),
> > > the SRCU is unconditionally enabled. So it's time to use
> > > SRCU to protect readers who previously held shrinker_rwsem.
> > > 
> > > [1]. https://lore.kernel.org/lkml/20191129214541.3110-1-ptikhomirov@virtuozzo.com/
> > > [2]. https://lore.kernel.org/all/1437080113.3596.2.camel@stgolabs.net/
> > > [3]. https://lore.kernel.org/lkml/1510609063-3327-1-git-send-email-penguin-kernel@I-love.SAKURA.ne.jp/
> > > [4]. https://lore.kernel.org/lkml/153365347929.19074.12509495712735843805.stgit@localhost.localdomain/
> > > [5]. https://lore.kernel.org/lkml/20210927074823.5825-1-sultan@kerneltoast.com/
> > > 
> > > Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
> > > ---
> > >  mm/vmscan.c | 27 +++++++++++----------------
> > >  1 file changed, 11 insertions(+), 16 deletions(-)
> > > 
> > > diff --git a/mm/vmscan.c b/mm/vmscan.c
> > > index 9f895ca6216c..02987a6f95d1 100644
> > > --- a/mm/vmscan.c
> > > +++ b/mm/vmscan.c
> > > @@ -202,6 +202,7 @@ static void set_task_reclaim_state(struct task_struct *task,
> > >  
> > >  LIST_HEAD(shrinker_list);
> > >  DECLARE_RWSEM(shrinker_rwsem);
> > > +DEFINE_SRCU(shrinker_srcu);
> > >  
> > >  #ifdef CONFIG_MEMCG
> > >  static int shrinker_nr_max;
> > > @@ -706,7 +707,7 @@ void free_prealloced_shrinker(struct shrinker *shrinker)
> > >  void register_shrinker_prepared(struct shrinker *shrinker)
> > >  {
> > >  	down_write(&shrinker_rwsem);
> > > -	list_add_tail(&shrinker->list, &shrinker_list);
> > > +	list_add_tail_rcu(&shrinker->list, &shrinker_list);
> > >  	shrinker->flags |= SHRINKER_REGISTERED;
> > >  	shrinker_debugfs_add(shrinker);
> > >  	up_write(&shrinker_rwsem);
> > > @@ -760,13 +761,15 @@ void unregister_shrinker(struct shrinker *shrinker)
> > >  		return;
> > >  
> > >  	down_write(&shrinker_rwsem);
> > > -	list_del(&shrinker->list);
> > > +	list_del_rcu(&shrinker->list);
> > >  	shrinker->flags &= ~SHRINKER_REGISTERED;
> > >  	if (shrinker->flags & SHRINKER_MEMCG_AWARE)
> > >  		unregister_memcg_shrinker(shrinker);
> > >  	debugfs_entry = shrinker_debugfs_remove(shrinker);
> > >  	up_write(&shrinker_rwsem);
> > >  
> > > +	synchronize_srcu(&shrinker_srcu);
> > > +
> > >  	debugfs_remove_recursive(debugfs_entry);
> > >  
> > >  	kfree(shrinker->nr_deferred);
> > > @@ -786,6 +789,7 @@ void synchronize_shrinkers(void)
> > >  {
> > >  	down_write(&shrinker_rwsem);
> > >  	up_write(&shrinker_rwsem);
> > > +	synchronize_srcu(&shrinker_srcu);
> > >  }
> > >  EXPORT_SYMBOL(synchronize_shrinkers);
> > >  
> > > @@ -996,6 +1000,7 @@ static unsigned long shrink_slab(gfp_t gfp_mask, int nid,
> > >  {
> > >  	unsigned long ret, freed = 0;
> > >  	struct shrinker *shrinker;
> > > +	int srcu_idx;
> > >  
> > >  	/*
> > >  	 * The root memcg might be allocated even though memcg is disabled
> > > @@ -1007,10 +1012,10 @@ static unsigned long shrink_slab(gfp_t gfp_mask, int nid,
> > >  	if (!mem_cgroup_disabled() && !mem_cgroup_is_root(memcg))
> > >  		return shrink_slab_memcg(gfp_mask, nid, memcg, priority);
> > >  
> > > -	if (!down_read_trylock(&shrinker_rwsem))
> > > -		goto out;
> > > +	srcu_idx = srcu_read_lock(&shrinker_srcu);
> > >  
> > > -	list_for_each_entry(shrinker, &shrinker_list, list) {
> > > +	list_for_each_entry_srcu(shrinker, &shrinker_list, list,
> > > +				 srcu_read_lock_held(&shrinker_srcu)) {
> > >  		struct shrink_control sc = {
> > >  			.gfp_mask = gfp_mask,
> > >  			.nid = nid,
> > > @@ -1021,19 +1026,9 @@ static unsigned long shrink_slab(gfp_t gfp_mask, int nid,
> > >  		if (ret == SHRINK_EMPTY)
> > >  			ret = 0;
> > >  		freed += ret;
> > > -		/*
> > > -		 * Bail out if someone want to register a new shrinker to
> > > -		 * prevent the registration from being stalled for long periods
> > > -		 * by parallel ongoing shrinking.
> > > -		 */
> > > -		if (rwsem_is_contended(&shrinker_rwsem)) {
> > > -			freed = freed ? : 1;
> > > -			break;
> > > -		}
> > >  	}
> > >  
> > > -	up_read(&shrinker_rwsem);
> > > -out:
> > > +	srcu_read_unlock(&shrinker_srcu, srcu_idx);
> > >  	cond_resched();
> > >  	return freed;
> > >  }
> > > -- 
> > > 2.20.1
> > > 
> > > 
> > 
> > Hi Qi,
> > 
> > A different problem I realized after my old attempt to use SRCU was that the
> > unregister_shrinker() path became quite slow due to the heavy synchronize_srcu()
> > call. Both register_shrinker() *and* unregister_shrinker() are called frequently
> > these days, and SRCU is too unfair to the unregister path IMO.
> > 
> > Although I never got around to submitting it, I made a non-SRCU solution [1]
> > that uses fine-grained locking instead, which is fair to both the register path
> > and unregister path. (The patch I've linked is a version of this adapted to an
> > older 4.14 kernel FYI, but it can be reworked for the current kernel.)
> > 
> > What do you think about the fine-grained locking approach?
> 
> Another approach is to use synchronize_srcu_expedited(), which avoids
> the sleeps that are otherwise used to encourage sharing of grace periods
> among concurrent requests.  It might be possible to use call_srcu(),
> but I don't claim to know the shrinker code well enough to say for sure.

Hi Paul,

I don't believe call_srcu() can be used since shrinker users need to be
guaranteed that their shrinkers aren't in use after unregister_shrinker().

Using synchronize_srcu_expedited() sounds like it'd definitely help, though
unregistering a single shrinker would ultimately still require waiting for all
shrinkers to finish running before the grace period can elapse. There can be
many shrinkers and they're not very fast I think.

Thanks,
Sultan

> 
> 							Thanx, Paul
> 
> > Thanks,
> > Sultan
> > 
> > [1] https://github.com/kerneltoast/android_kernel_google_floral/commit/012378f3173a82d2333d3ae7326691544301e76a
>