linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Zi Yan <ziy@nvidia.com>
To: Qi Zheng <qi.zheng@linux.dev>
Cc: hannes@cmpxchg.org, hughd@google.com, mhocko@suse.com,
	roman.gushchin@linux.dev, shakeel.butt@linux.dev,
	muchun.song@linux.dev, david@redhat.com,
	lorenzo.stoakes@oracle.com, harry.yoo@oracle.com,
	baolin.wang@linux.alibaba.com, Liam.Howlett@oracle.com,
	npache@redhat.com, ryan.roberts@arm.com, dev.jain@arm.com,
	baohua@kernel.org, lance.yang@linux.dev,
	akpm@linux-foundation.org, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org, cgroups@vger.kernel.org,
	Qi Zheng <zhengqi.arch@bytedance.com>
Subject: Re: [PATCH v4 0/4] reparent the THP split queue
Date: Mon, 13 Oct 2025 12:37:32 -0400	[thread overview]
Message-ID: <C3134C16-584F-41D2-88E4-4B94B58C16F2@nvidia.com> (raw)
In-Reply-To: <df9a3e22-caca-4298-b7d8-5334ce5446a0@linux.dev>

On 13 Oct 2025, at 3:23, Qi Zheng wrote:

> Hi Zi,
>
> On 10/11/25 12:25 AM, Zi Yan wrote:
>> On 3 Oct 2025, at 12:53, Qi Zheng wrote:
>
> [snip]
>
>>>
>>
>> Hi Qi,
>>
>> I got CPU soft locks when run "echo 3 | sudo tee /proc/sys/vm/drop_caches"
>> with today's mm-new on a freshly booted system. Reverting Patch 3 (and Patch 4)
>> of your patchset solves the issue.
>>
>> My config file is attached. My kernel relevant kernel parameters are:
>> "cgroup_no_v1=all transparent_hugepage=always thp_shmem=2M:always".
>> The machine is a 8GB 8-core x86_64 VM.
>>
>> The kernel log:
>>
>> [   36.441539] watchdog: BUG: soft lockup - CPU#0 stuck for 26s! [tee:810]
>> [   36.441549] Modules linked in:
>> [   36.441566] CPU: 0 UID: 0 PID: 810 Comm: tee Not tainted 6.17.0-mm-everything-2024-01-29-07-19-no-mglru+ #526 PREEMPT(voluntary)
>> [   36.441570] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.17.0-debian-1.17.0-1 04/01/2014
>> [   36.441574] RIP: 0010:_raw_spin_unlock_irqrestore+0x19/0x40
>
> It seems that the softlockup is not caused by lock contention. I have
> also seen the following stack locally:
>
> [14819.914867][    C2] watchdog: BUG: soft lockup - CPU#2 stuck for 49s! [kswapd0:202]
> [14819.914873][    C2] Modules linked in:
> [14819.914877][    C2] CPU: 2 UID: 0 PID: 202 Comm: kswapd0 Tainted: G           L      6.1
> [14819.914880][    C2] Tainted: [L]=SOFTLOCKUP
> [14819.914881][    C2] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-1 04
> [14819.914883][    C2] RIP: 0010:folios_put_refs+0x34/0x210
> [14819.914889][    C2] Code: 57 41 56 49 89 fe 41 55 41 54 55 53 48 83 ec 18 65 48 8b 05 36 51
> [14819.914891][    C2] RSP: 0000:ffffc900039db8e8 EFLAGS: 00000246
> [14819.914893][    C2] RAX: 0000000000000000 RBX: ffff888109e64330 RCX: ffffffff82848bb8
> [14819.914895][    C2] RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffffc900039db960
> [14819.914896][    C2] RBP: ffff888109e642a0 R08: ffff888100403058 R09: ffff888100402fe8
> [14819.914897][    C2] R10: 0000000000000000 R11: ffffffff82711328 R12: ffff888109e64328
> [14819.914898][    C2] R13: ffffc900039dbb30 R14: ffffc900039db960 R15: ffffc900039dbb30
> [14819.914916][    C2] FS:  0000000000000000(0000) GS:ffff888332ad7000(0000) knlGS:00000000000
> [14819.914918][    C2] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [14819.914919][    C2] CR2: 00007f8d211b3000 CR3: 000000010833c000 CR4: 00000000000006f0
> [14819.914921][    C2] Call Trace:
> [14819.914923][    C2]  <TASK>
> [14819.914927][    C2]  deferred_split_scan+0x30d/0x420
> [14819.914934][    C2]  do_shrink_slab+0x105/0x350
> [14819.914937][    C2]  shrink_slab+0x2fa/0x3f0
> [14819.914939][    C2]  ? try_to_shrink_lruvec+0x1a6/0x270
> [14819.914944][    C2]  shrink_one+0x11d/0x1b0
> [14819.914947][    C2]  shrink_node+0x9a4/0xb90
> [14819.914952][    C2]  balance_pgdat+0x548/0x910
> [14819.914955][    C2]  ? __sched_fork+0x160/0x180
> [14819.914964][    C2]  ? __timer_delete_sync+0x2b/0x40
> [14819.914970][    C2]  kswapd+0x1b9/0x320
> [14819.914973][    C2]  ? __pfx_autoremove_wake_function+0x10/0x10
> [14819.914978][    C2]  ? __pfx_kswapd+0x10/0x10
> [14819.914981][    C2]  kthread+0xfd/0x250
> [14819.914984][    C2]  ? finish_task_switch+0x90/0x270
> [14819.914987][    C2]  ? __pfx_kthread+0x10/0x10
> [14819.914989][    C2]  ? __pfx_kthread+0x10/0x10
> [14819.914991][    C2]  ret_from_fork+0x170/0x190
> [14819.914995][    C2]  ? __pfx_kthread+0x10/0x10
> [14819.914996][    C2]  ret_from_fork_asm+0x1a/0x30
> [14819.915002][    C2]  </TASK>
>
> It seems that the reason is that deferred_split_scan() has been executed
> for too long.
>
> Could you please help test the following changes? I applied them locally
> and no softlockup occurred.
>
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index b5eea2091cdf6..5353c7bd2c9af 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -4286,8 +4286,10 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
>         }
>         folios_put(&fbatch);
>
> -       if (sc->nr_to_scan)
> +       if (sc->nr_to_scan) {
> +               cond_resched();
>                 goto retry;
> +       }
>
>         /*
>          * Stop shrinker if we didn't split any page, but the queue is empty.
>

It does not fix the issue, but only gets rid of the soft lockup warning.
"echo 3 | sudo tee /proc/sys/vm/drop_caches" just runs forever.

Looking at the original code, sc->nr_to_scan was one of the two conditions
on breaking out of split_queue scanning and was never checked again
afterwards. When split_queue size is smaller than nr_to_scan, your code
will retry forever but not the original one. After I added pr_info() to
print sc->nr_to_scan at
1) before retry:,
2) before for (... folio_batch_count();...),
3) before "if (sc->nr_to_scan)",

I see that 1) printed 2, 2) and 3) kept printing 1. It matches my
above guess.

The below patch fixes the issue:

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 43a3c499aec0..d38816a0c117 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -4415,7 +4415,7 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
 	}
 	folios_put(&fbatch);

-	if (sc->nr_to_scan)
+	if (sc->nr_to_scan && !list_empty(&ds_queue->split_queue))
 		goto retry;

 	/*



>
>> [   36.441592] Code: 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 0f 1f 44 00 00 53 48 89 f3 e8 92 68 fd fe 80 e7 02 74 06 fb 0f 1f 44 00 00 <65> ff 0d d0 5f 7e 01 74 06 5b c3 cc cc cc cc 0f 1f 44 00 00 5b c3
>> [   36.441594] RSP: 0018:ffffc900029afb60 EFLAGS: 00000202
>> [   36.441598] RAX: 0000000000000001 RBX: 0000000000000286 RCX: ffff888101168670
>> [   36.441601] RDX: 0000000000000001 RSI: 0000000000000286 RDI: ffff888101168658
>> [   36.441602] RBP: 0000000000000001 R08: ffff88813ba44ec0 R09: 0000000000000000
>> [   36.441603] R10: 00000000000001a8 R11: 0000000000000000 R12: ffff8881011685e0
>> [   36.441604] R13: 0000000000000000 R14: ffff888101168000 R15: ffffc900029afd60
>> [   36.441606] FS:  00007f7fe3655740(0000) GS:ffff8881b7e5d000(0000) knlGS:0000000000000000
>> [   36.441607] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> [   36.441608] CR2: 0000563d4d439bf0 CR3: 000000010873c006 CR4: 0000000000370ef0
>> [   36.441614] Call Trace:
>> [   36.441616]  <TASK>
>> [   36.441619]  deferred_split_scan+0x1e0/0x480
>> [   36.441627]  ? _raw_spin_unlock_irqrestore+0xe/0x40
>> [   36.441630]  ? kvfree_rcu_queue_batch+0x96/0x1c0
>> [   36.441634]  ? do_raw_spin_unlock+0x46/0xd0
>> [   36.441639]  ? kfree_rcu_monitor+0x1da/0x2c0
>> [   36.441641]  ? list_lru_count_one+0x47/0x90
>> [   36.441644]  do_shrink_slab+0x153/0x360
>> [   36.441649]  shrink_slab+0xd3/0x390
>> [   36.441652]  drop_slab+0x7d/0x130
>> [   36.441655]  drop_caches_sysctl_handler+0x98/0xb0
>> [   36.441660]  proc_sys_call_handler+0x1c7/0x2c0
>> [   36.441664]  vfs_write+0x221/0x450
>> [   36.441669]  ksys_write+0x6c/0xe0
>> [   36.441672]  do_syscall_64+0x50/0x200
>> [   36.441675]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
>> [   36.441678] RIP: 0033:0x7f7fe36e7687
>> [   36.441685] Code: 48 89 fa 4c 89 df e8 58 b3 00 00 8b 93 08 03 00 00 59 5e 48 83 f8 fc 74 1a 5b c3 0f 1f 84 00 00 00 00 00 48 8b 44 24 10 0f 05 <5b> c3 0f 1f 80 00 00 00 00 83 e2 39 83 fa 08 75 de e8 23 ff ff ff
>> [   36.441686] RSP: 002b:00007ffdffcbba10 EFLAGS: 00000202 ORIG_RAX: 0000000000000001
>> [   36.441688] RAX: ffffffffffffffda RBX: 00007f7fe3655740 RCX: 00007f7fe36e7687
>> [   36.441689] RDX: 0000000000000002 RSI: 00007ffdffcbbbb0 RDI: 0000000000000003
>> [   36.441690] RBP: 00007ffdffcbbbb0 R08: 0000000000000000 R09: 0000000000000000
>> [   36.441691] R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000000002
>> [   36.441692] R13: 0000558d40be64c0 R14: 00007f7fe383de80 R15: 0000000000000002
>> [   36.441694]  </TASK>
>> [   64.441531] watchdog: BUG: soft lockup - CPU#0 stuck for 53s! [tee:810]
>> [   64.441537] Modules linked in:
>> [   64.441545] CPU: 0 UID: 0 PID: 810 Comm: tee Tainted: G             L      6.17.0-mm-everything-2024-01-29-07-19-no-mglru+ #526 PREEMPT(voluntary)
>> [   64.441548] Tainted: [L]=SOFTLOCKUP
>> [   64.441552] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.17.0-debian-1.17.0-1 04/01/2014
>> [   64.441555] RIP: 0010:_raw_spin_unlock_irqrestore+0x19/0x40
>> [   64.441565] Code: 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 0f 1f 44 00 00 53 48 89 f3 e8 92 68 fd fe 80 e7 02 74 06 fb 0f 1f 44 00 00 <65> ff 0d d0 5f 7e 01 74 06 5b c3 cc cc cc cc 0f 1f 44 00 00 5b c3
>> [   64.441566] RSP: 0018:ffffc900029afb60 EFLAGS: 00000202
>> [   64.441568] RAX: 0000000000000001 RBX: 0000000000000286 RCX: ffff888101168670
>> [   64.441570] RDX: 0000000000000001 RSI: 0000000000000286 RDI: ffff888101168658
>> [   64.441571] RBP: 0000000000000001 R08: ffff88813ba44ec0 R09: 0000000000000000
>> [   64.441572] R10: 00000000000001a8 R11: 0000000000000000 R12: ffff8881011685e0
>> [   64.441573] R13: 0000000000000000 R14: ffff888101168000 R15: ffffc900029afd60
>> [   64.441574] FS:  00007f7fe3655740(0000) GS:ffff8881b7e5d000(0000) knlGS:0000000000000000
>> [   64.441576] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> [   64.441577] CR2: 0000563d4d439bf0 CR3: 000000010873c006 CR4: 0000000000370ef0
>> [   64.441581] Call Trace:
>> [   64.441583]  <TASK>
>> [   64.441591]  deferred_split_scan+0x1e0/0x480
>> [   64.441598]  ? _raw_spin_unlock_irqrestore+0xe/0x40
>> [   64.441599]  ? kvfree_rcu_queue_batch+0x96/0x1c0
>> [   64.441603]  ? do_raw_spin_unlock+0x46/0xd0
>> [   64.441607]  ? kfree_rcu_monitor+0x1da/0x2c0
>> [   64.441610]  ? list_lru_count_one+0x47/0x90
>> [   64.441613]  do_shrink_slab+0x153/0x360
>> [   64.441618]  shrink_slab+0xd3/0x390
>> [   64.441621]  drop_slab+0x7d/0x130
>> [   64.441624]  drop_caches_sysctl_handler+0x98/0xb0
>> [   64.441629]  proc_sys_call_handler+0x1c7/0x2c0
>> [   64.441632]  vfs_write+0x221/0x450
>> [   64.441638]  ksys_write+0x6c/0xe0
>> [   64.441641]  do_syscall_64+0x50/0x200
>> [   64.441645]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
>> [   64.441648] RIP: 0033:0x7f7fe36e7687
>> [   64.441654] Code: 48 89 fa 4c 89 df e8 58 b3 00 00 8b 93 08 03 00 00 59 5e 48 83 f8 fc 74 1a 5b c3 0f 1f 84 00 00 00 00 00 48 8b 44 24 10 0f 05 <5b> c3 0f 1f 80 00 00 00 00 83 e2 39 83 fa 08 75 de e8 23 ff ff ff
>> [   64.441656] RSP: 002b:00007ffdffcbba10 EFLAGS: 00000202 ORIG_RAX: 0000000000000001
>> [   64.441658] RAX: ffffffffffffffda RBX: 00007f7fe3655740 RCX: 00007f7fe36e7687
>> [   64.441659] RDX: 0000000000000002 RSI: 00007ffdffcbbbb0 RDI: 0000000000000003
>> [   64.441660] RBP: 00007ffdffcbbbb0 R08: 0000000000000000 R09: 0000000000000000
>> [   64.441661] R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000000002
>> [   64.441662] R13: 0000558d40be64c0 R14: 00007f7fe383de80 R15: 0000000000000002
>> [   64.441663]  </TASK>
>>
>>
>>
>> --
>> Best Regards,
>> Yan, Zi


--
Best Regards,
Yan, Zi


  reply	other threads:[~2025-10-13 16:37 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-10-03 16:53 Qi Zheng
2025-10-03 16:53 ` [PATCH v4 1/4] mm: thp: replace folio_memcg() with folio_memcg_charged() Qi Zheng
2025-10-03 16:53 ` [PATCH v4 2/4] mm: thp: introduce folio_split_queue_lock and its variants Qi Zheng
2025-10-03 16:53 ` [PATCH v4 3/4] mm: thp: use folio_batch to handle THP splitting in deferred_split_scan() Qi Zheng
2025-10-06 23:16   ` Shakeel Butt
2025-10-13  7:28     ` Qi Zheng
2025-10-14 14:25   ` kernel test robot
2025-10-03 16:53 ` [PATCH v4 4/4] mm: thp: reparent the split queue during memcg offline Qi Zheng
2025-10-03 16:58   ` Zi Yan
2025-10-04  7:52   ` Muchun Song
2025-10-06  6:46   ` David Hildenbrand
2025-10-07 17:56   ` Shakeel Butt
2025-10-13  7:29     ` Qi Zheng
2025-10-10 16:25 ` [PATCH v4 0/4] reparent the THP split queue Zi Yan
2025-10-11  0:51   ` Qi Zheng
2025-10-11 18:28     ` Andrew Morton
2025-10-13  7:23   ` Qi Zheng
2025-10-13 16:37     ` Zi Yan [this message]
2025-10-14  6:49       ` Qi Zheng

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=C3134C16-584F-41D2-88E4-4B94B58C16F2@nvidia.com \
    --to=ziy@nvidia.com \
    --cc=Liam.Howlett@oracle.com \
    --cc=akpm@linux-foundation.org \
    --cc=baohua@kernel.org \
    --cc=baolin.wang@linux.alibaba.com \
    --cc=cgroups@vger.kernel.org \
    --cc=david@redhat.com \
    --cc=dev.jain@arm.com \
    --cc=hannes@cmpxchg.org \
    --cc=harry.yoo@oracle.com \
    --cc=hughd@google.com \
    --cc=lance.yang@linux.dev \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=lorenzo.stoakes@oracle.com \
    --cc=mhocko@suse.com \
    --cc=muchun.song@linux.dev \
    --cc=npache@redhat.com \
    --cc=qi.zheng@linux.dev \
    --cc=roman.gushchin@linux.dev \
    --cc=ryan.roberts@arm.com \
    --cc=shakeel.butt@linux.dev \
    --cc=zhengqi.arch@bytedance.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