All of lore.kernel.org
 help / color / mirror / Atom feed
From: Shakeel Butt <shakeel.butt@linux.dev>
To: Harry Yoo <harry@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	 Johannes Weiner <hannes@cmpxchg.org>,
	Michal Hocko <mhocko@kernel.org>,
	 Roman Gushchin <roman.gushchin@linux.dev>,
	Muchun Song <muchun.song@linux.dev>,
	 Qi Zheng <qi.zheng@linux.dev>, Alexandre Ghiti <alex@ghiti.fr>,
	 Joshua Hahn <joshua.hahnjy@gmail.com>,
	Meta kernel team <kernel-team@meta.com>,
	linux-mm@kvack.org,  cgroups@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	 kernel test robot <oliver.sang@intel.com>
Subject: Re: [PATCH 4/4] memcg: multi objcg charge support
Date: Thu, 21 May 2026 13:19:50 -0700	[thread overview]
Message-ID: <ag9hFJv_LOJh7IVE@linux.dev> (raw)
In-Reply-To: <5b09f618-3b84-4163-84f9-f3adc0f1cc97@kernel.org>

On Thu, May 21, 2026 at 10:43:11AM +0900, Harry Yoo wrote:
> 
> 
> On 5/21/26 10:05 AM, Shakeel Butt wrote:
> > On Wed, May 20, 2026 at 06:35:30PM +0900, Harry Yoo wrote:
> > > > @@ -3350,19 +3405,45 @@ static void __refill_obj_stock(struct obj_cgroup *objcg,
> > > >    		goto out;
> > > >    	}
> > > > -	stock_nr_bytes = stock->nr_bytes;
> > > > -	if (READ_ONCE(stock->cached_objcg) != objcg) { /* reset if necessary */
> > > > -		drain_obj_stock(stock);
> > > > +	for (i = 0; i < NR_OBJ_STOCK; ++i) {
> > > > +		struct obj_cgroup *cached = READ_ONCE(stock->cached[i]);
> > > > +
> > > > +		if (!cached) {
> > > > +			if (empty_slot == -1)
> > > > +				empty_slot = i;
> > > > +			continue;
> > > > +		}
> > > > +		if (cached == objcg) {
> > > > +			slot = i;
> > > > +			break;
> > > > +		}
> > > > +	}
> > > > +
> > > > +	if (slot == -1) {
> > > > +		slot = empty_slot;
> > > > +		if (slot == -1) {
> > > > +			slot = get_random_u32_below(NR_OBJ_STOCK);
> > > 
> > > It would break kmalloc_nolock() because _get_random_bytes() uses a spinlock.
> > > perhaps prandom_u32_state() should be sufficient in this case.
> 
> s/spinlock/local_lock/

I do see spinlock in crng_make_state() for some code paths.

> 
> > > Is there a reason why it uses random eviction, unlike multi-memcg percpu
> > > charge cache?
> > 
> > Oh I didn't know and actually we are already using get_random_u32_below() in
> > refill_stock(). So, it need fixing as well. That would be a separate patch.
> 
> Ouch, I see.
> > I will explore prandom_u32_state().
> 
> Thanks!
> 
> FYI, SLUB had a similar issue that was recently fixed:
> commit a1e244a9f1778 ("mm/slab: use prandom if !allow_spin").
> 
> It uses prandom if spinning is not allowed when shuffling slab freelist.

The drain does not really need a random number. Fixing an index like 0 makes it
much simpler but it might expose some corner cases. Round robin might be enough
for this though. I will see what would be the easiest way forward.


  reply	other threads:[~2026-05-21 20:20 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-20  5:31 [PATCH 0/4] memcg: shrink obj_stock_pcp and cache multiple objcgs Shakeel Butt
2026-05-20  5:31 ` [PATCH 1/4] memcg: store node_id instead of pglist_data pointer Shakeel Butt
2026-05-20  6:01   ` Harry Yoo
2026-05-20  6:13   ` Muchun Song
2026-05-20  5:31 ` [PATCH 2/4] memcg: uint16_t for nr_bytes in obj_stock_pcp Shakeel Butt
2026-05-20  6:41   ` Harry Yoo
2026-05-20  7:01   ` Harry Yoo
2026-05-21  1:01     ` Shakeel Butt
2026-05-20 13:20   ` David Laight
2026-05-21  1:03     ` Shakeel Butt
2026-05-20  5:31 ` [PATCH 3/4] memcg: int16_t for cached slab stats Shakeel Butt
2026-05-20  7:25   ` Harry Yoo
2026-05-20  5:31 ` [PATCH 4/4] memcg: multi objcg charge support Shakeel Butt
2026-05-20  9:35   ` Harry Yoo
2026-05-21  1:05     ` Shakeel Butt
2026-05-21  1:43       ` Harry Yoo
2026-05-21 20:19         ` Shakeel Butt [this message]
2026-05-21  3:22       ` Joshua Hahn

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ag9hFJv_LOJh7IVE@linux.dev \
    --to=shakeel.butt@linux.dev \
    --cc=akpm@linux-foundation.org \
    --cc=alex@ghiti.fr \
    --cc=cgroups@vger.kernel.org \
    --cc=hannes@cmpxchg.org \
    --cc=harry@kernel.org \
    --cc=joshua.hahnjy@gmail.com \
    --cc=kernel-team@meta.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=mhocko@kernel.org \
    --cc=muchun.song@linux.dev \
    --cc=oliver.sang@intel.com \
    --cc=qi.zheng@linux.dev \
    --cc=roman.gushchin@linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.