From: Ryan Roberts <ryan.roberts@arm.com>
To: Matthew Wilcox <willy@infradead.org>
Cc: Zi Yan <ziy@nvidia.com>,
Andrew Morton <akpm@linux-foundation.org>,
linux-mm@kvack.org, Yang Shi <shy828301@gmail.com>,
Huang Ying <ying.huang@intel.com>
Subject: Re: [PATCH v3 10/18] mm: Allow non-hugetlb large folios to be batch processed
Date: Fri, 8 Mar 2024 11:44:35 +0000 [thread overview]
Message-ID: <e8911b30-e96b-486c-a92a-c3513facc12e@arm.com> (raw)
In-Reply-To: <Zen6VDC5B_SN4zpR@casper.infradead.org>
> The thought occurs that we don't need to take the folios off the list.
> I don't know that will fix anything, but this will fix your "running out
> of memory" problem -- I forgot to drop the reference if folio_trylock()
> failed. Of course, I can't call folio_put() inside the lock, so may
> as well move the trylock back to the second loop.
>
> Again, compile-tessted only.
>
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index fd745bcc97ff..4a2ab17f802d 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -3312,7 +3312,7 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
> struct pglist_data *pgdata = NODE_DATA(sc->nid);
> struct deferred_split *ds_queue = &pgdata->deferred_split_queue;
> unsigned long flags;
> - LIST_HEAD(list);
> + struct folio_batch batch;
> struct folio *folio, *next;
> int split = 0;
>
> @@ -3321,36 +3321,31 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
> ds_queue = &sc->memcg->deferred_split_queue;
> #endif
>
> + folio_batch_init(&batch);
> spin_lock_irqsave(&ds_queue->split_queue_lock, flags);
> - /* Take pin on all head pages to avoid freeing them under us */
> + /* Take ref on all folios to avoid freeing them under us */
> list_for_each_entry_safe(folio, next, &ds_queue->split_queue,
> _deferred_list) {
> - if (folio_try_get(folio)) {
> - list_move(&folio->_deferred_list, &list);
> - } else {
> - /* We lost race with folio_put() */
> - list_del_init(&folio->_deferred_list);
> - ds_queue->split_queue_len--;
> + if (!folio_try_get(folio))
> + continue;
> + if (folio_batch_add(&batch, folio) == 0) {
> + --sc->nr_to_scan;
> + break;
> }
> if (!--sc->nr_to_scan)
> break;
> }
> spin_unlock_irqrestore(&ds_queue->split_queue_lock, flags);
>
> - list_for_each_entry_safe(folio, next, &list, _deferred_list) {
> + while ((folio = folio_batch_next(&batch)) != NULL) {
> if (!folio_trylock(folio))
> - goto next;
> - /* split_huge_page() removes page from list on success */
> + continue;
> if (!split_folio(folio))
> split++;
> folio_unlock(folio);
> -next:
> - folio_put(folio);
> }
>
> - spin_lock_irqsave(&ds_queue->split_queue_lock, flags);
> - list_splice_tail(&list, &ds_queue->split_queue);
> - spin_unlock_irqrestore(&ds_queue->split_queue_lock, flags);
> + folios_put(&batch);
>
> /*
> * Stop shrinker if we didn't split any page, but the queue is empty.
OK I've tested this; the good news is that I haven't seen any oopses or memory
leaks. The bad news is that it still takes an absolute age (hours) to complete
the same test that without "mm: Allow non-hugetlb large folios to be batch
processed" took a couple of mins. And during that time, the system is completely
unresponsive - serial terminal doesn't work - can't even break in with sysreq.
And sometimes I see RCU stall warnings.
Dumping all the CPU back traces with gdb, all the cores (except one) are
contending on the the deferred split lock.
A couple of thoughts:
- Since we are now taking a maximum of 15 folios into a batch,
deferred_split_scan() is called much more often (in a tight loop from
do_shrink_slab()). Could it be that we are just trying to take the lock so much
more often now? I don't think it's quite that simple because we take the lock
for every single folio when adding it to the queue, so the dequeing cost should
still be a factor of 15 locks less.
- do_shrink_slab() might be calling deferred_split_scan() in a tight loop with
deferred_split_scan() returning 0 most of the time. If there are still folios on
the deferred split list but deferred_split_scan() was unable to lock any folios
then it will return 0, not SHRINK_STOP, so do_shrink_slab() will keep calling
it, essentially live locking. Has your patch changed the duration of the folio
being locked? I don't think so...
- Ahh, perhaps its as simple as your fix has removed the code that removed the
folio from the deferred split queue if it fails to get a reference? That could
mean we end up returning 0 instead of SHRINK_STOP too. I'll have play.
next prev parent reply other threads:[~2024-03-08 11:44 UTC|newest]
Thread overview: 73+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-02-27 17:42 [PATCH v3 00/18] Rearrange batched folio freeing Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 01/18] mm: Make folios_put() the basis of release_pages() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 02/18] mm: Convert free_unref_page_list() to use folios Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 03/18] mm: Add free_unref_folios() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 04/18] mm: Use folios_put() in __folio_batch_release() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 05/18] memcg: Add mem_cgroup_uncharge_folios() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 06/18] mm: Remove use of folio list from folios_put() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 07/18] mm: Use free_unref_folios() in put_pages_list() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 08/18] mm: use __page_cache_release() in folios_put() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 09/18] mm: Handle large folios in free_unref_folios() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 10/18] mm: Allow non-hugetlb large folios to be batch processed Matthew Wilcox (Oracle)
2024-03-06 13:42 ` Ryan Roberts
2024-03-06 16:09 ` Matthew Wilcox
2024-03-06 16:19 ` Ryan Roberts
2024-03-06 17:41 ` Ryan Roberts
2024-03-06 18:41 ` Zi Yan
2024-03-06 19:55 ` Matthew Wilcox
2024-03-06 21:55 ` Matthew Wilcox
2024-03-07 8:56 ` Ryan Roberts
2024-03-07 13:50 ` Yin, Fengwei
2024-03-07 14:05 ` Re: Matthew Wilcox
2024-03-07 15:24 ` Re: Ryan Roberts
2024-03-07 16:24 ` Re: Ryan Roberts
2024-03-07 23:02 ` Re: Matthew Wilcox
2024-03-08 1:06 ` Re: Yin, Fengwei
2024-03-07 17:33 ` [PATCH v3 10/18] mm: Allow non-hugetlb large folios to be batch processed Matthew Wilcox
2024-03-07 18:35 ` Ryan Roberts
2024-03-07 20:42 ` Matthew Wilcox
2024-03-08 11:44 ` Ryan Roberts [this message]
2024-03-08 12:09 ` Ryan Roberts
2024-03-08 14:21 ` Ryan Roberts
2024-03-08 15:11 ` Matthew Wilcox
2024-03-08 16:03 ` Matthew Wilcox
2024-03-08 17:13 ` Ryan Roberts
2024-03-08 18:09 ` Ryan Roberts
2024-03-08 18:18 ` Matthew Wilcox
2024-03-09 4:34 ` Andrew Morton
2024-03-09 4:52 ` Matthew Wilcox
2024-03-09 8:05 ` Ryan Roberts
2024-03-09 12:33 ` Ryan Roberts
2024-03-10 13:38 ` Matthew Wilcox
2024-03-08 15:33 ` Matthew Wilcox
2024-03-09 6:09 ` Matthew Wilcox
2024-03-09 7:59 ` Ryan Roberts
2024-03-09 8:18 ` Ryan Roberts
2024-03-09 9:38 ` Ryan Roberts
2024-03-10 4:23 ` Matthew Wilcox
2024-03-10 8:23 ` Ryan Roberts
2024-03-10 11:08 ` Matthew Wilcox
2024-03-10 11:01 ` Ryan Roberts
2024-03-10 11:11 ` Matthew Wilcox
2024-03-10 16:31 ` Ryan Roberts
2024-03-10 19:57 ` Matthew Wilcox
2024-03-10 19:59 ` Ryan Roberts
2024-03-10 20:46 ` Matthew Wilcox
2024-03-10 21:52 ` Matthew Wilcox
2024-03-11 9:01 ` Ryan Roberts
2024-03-11 12:26 ` Matthew Wilcox
2024-03-11 12:36 ` Ryan Roberts
2024-03-11 15:50 ` Matthew Wilcox
2024-03-11 16:14 ` Ryan Roberts
2024-03-11 17:49 ` Matthew Wilcox
2024-03-12 11:57 ` Ryan Roberts
2024-03-11 19:26 ` Matthew Wilcox
2024-03-10 11:14 ` Ryan Roberts
2024-02-27 17:42 ` [PATCH v3 11/18] mm: Free folios in a batch in shrink_folio_list() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 12/18] mm: Free folios directly in move_folios_to_lru() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 13/18] memcg: Remove mem_cgroup_uncharge_list() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 14/18] mm: Remove free_unref_page_list() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 15/18] mm: Remove lru_to_page() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 16/18] mm: Convert free_pages_and_swap_cache() to use folios_put() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 17/18] mm: Use a folio in __collapse_huge_page_copy_succeeded() Matthew Wilcox (Oracle)
2024-02-27 17:42 ` [PATCH v3 18/18] mm: Convert free_swap_cache() to take a folio Matthew Wilcox (Oracle)
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=e8911b30-e96b-486c-a92a-c3513facc12e@arm.com \
--to=ryan.roberts@arm.com \
--cc=akpm@linux-foundation.org \
--cc=linux-mm@kvack.org \
--cc=shy828301@gmail.com \
--cc=willy@infradead.org \
--cc=ying.huang@intel.com \
--cc=ziy@nvidia.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).