From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mout01.posteo.de (mout01.posteo.de [185.67.36.65]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 956FF524F for ; Mon, 6 Jan 2025 12:53:41 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=185.67.36.65 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1736168024; cv=none; b=ikCwRpVf4nM+FkOvAowSA+alvkiUAgDZjH4uniTGdwTWThMnM3urR3BTxkhF2STNW3yZNJdQrnSrJ1cVBATNT26jr4J4JftJaZjMKAFTEjMYm2EbBGWFNMWOxNhLovbS88T6Bz35FHrxfO6+wyUkZyLhHTiJEZuzVlXRuzcIQpE= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1736168024; c=relaxed/simple; bh=NwPD9OTbzu7amAfdX3cPOauvw7fdeW5apjUajyr5o4k=; h=From:To:Cc:Subject:In-Reply-To:References:Date:Message-ID: MIME-Version:Content-Type; b=a5oJ26MfoUldWWTOp08bLUn8F8jirXJqdy8ixw7LJvq7lkeavEV3YtI9uYQyQgplWikr+MUUWvyyUD539R2AZFXPqIAW+39S6BtDAwp3VHE0xtyOtsUnZEo7CzzkQebj9luTIb4W1hqnVK5KN0UvsyBQDKEOXP5UK0G3ohR/n48= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=posteo.net; spf=pass smtp.mailfrom=posteo.net; dkim=pass (2048-bit key) header.d=posteo.net header.i=@posteo.net header.b=r7z6+hlt; arc=none smtp.client-ip=185.67.36.65 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=posteo.net Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=posteo.net Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=posteo.net header.i=@posteo.net header.b="r7z6+hlt" Received: from submission (posteo.de [185.67.36.169]) by mout01.posteo.de (Postfix) with ESMTPS id 822DE240027 for ; Mon, 6 Jan 2025 13:53:39 +0100 (CET) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=posteo.net; s=2017; t=1736168019; bh=NwPD9OTbzu7amAfdX3cPOauvw7fdeW5apjUajyr5o4k=; h=From:To:Cc:Subject:Date:Message-ID:MIME-Version:Content-Type: From; b=r7z6+hlt/T/E9O0wr+I1iPvdeBdT83r5+1v8XkkhIei2ukm1bCT9t3mX7Nyqcvms5 eylCh67J7TXlKhgE34SpUFUBlUP58Ya6R1SGXDsloRrmywdq6KA/eK1VsI8qFoxhOk 3hj4feXBs/KxZSG0GQQ9UOlaoM0JkMUXi60WrqAuybQKNuRbiKW/XfHq8uQPBe/VLs qXHfZt/PL8mx8hy+nMJD9dLr51uWDNbgTjEh0ndTo/MnNQ/Z4egfeNMdSOMj5LWy8c 0Q1z4X7jd+VrFiz7PDCuQssUYyedtApktCv3g/fiUu5dAlKGr53m4xa3Bwkr4z2L0+ BMkMrebZEXL1A== Received: from customer (localhost [127.0.0.1]) by submission (posteo.de) with ESMTPSA id 4YRYyT5nydz9rxL; Mon, 6 Jan 2025 13:53:37 +0100 (CET) From: Charalampos Mitrodimas To: Lorenzo Stoakes Cc: Koichiro Den , linux-mm@kvack.org, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, Thomas Gleixner , Peter Zijlstra Subject: Re: [PATCH v2] vmstat: disable vmstat_work on vmstat_cpu_down_prep() In-Reply-To: <7ed97096-859e-46d0-8f27-16a2298a8914@lucifer.local> (Lorenzo Stoakes's message of "Mon, 6 Jan 2025 10:52:37 +0000") References: <20241221033321.4154409-1-koichiro.den@canonical.com> <2q7ge6cgzeowqffyn6w6ed4trhaaumv5ubdgud2tsoolen7wpw@4akuomhbacyh> <7ed97096-859e-46d0-8f27-16a2298a8914@lucifer.local> Date: Mon, 06 Jan 2025 12:53:36 +0000 Message-ID: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain Lorenzo Stoakes writes: > +cc tglx, peterz for insight on CPU hot plug > > On Sat, Jan 04, 2025 at 01:00:17PM +0900, Koichiro Den wrote: >> On Fri, Jan 03, 2025 at 11:33:19PM +0000, Lorenzo Stoakes wrote: >> > On Sat, Dec 21, 2024 at 12:33:20PM +0900, Koichiro Den wrote: >> > > Even after mm/vmstat:online teardown, shepherd may still queue work for >> > > the dying cpu until the cpu is removed from online mask. While it's >> > > quite rare, this means that after unbind_workers() unbinds a per-cpu >> > > kworker, it potentially runs vmstat_update for the dying CPU on an >> > > irrelevant cpu before entering atomic AP states. >> > > When CONFIG_DEBUG_PREEMPT=y, it results in the following error with the >> > > backtrace. >> > > >> > > BUG: using smp_processor_id() in preemptible [00000000] code: \ >> > > kworker/7:3/1702 >> > > caller is refresh_cpu_vm_stats+0x235/0x5f0 >> > > CPU: 0 UID: 0 PID: 1702 Comm: kworker/7:3 Tainted: G >> > > Tainted: [N]=TEST >> > > Workqueue: mm_percpu_wq vmstat_update >> > > Call Trace: >> > > >> > > dump_stack_lvl+0x8d/0xb0 >> > > check_preemption_disabled+0xce/0xe0 >> > > refresh_cpu_vm_stats+0x235/0x5f0 >> > > vmstat_update+0x17/0xa0 >> > > process_one_work+0x869/0x1aa0 >> > > worker_thread+0x5e5/0x1100 >> > > kthread+0x29e/0x380 >> > > ret_from_fork+0x2d/0x70 >> > > ret_from_fork_asm+0x1a/0x30 >> > > >> > > >> > > So, for mm/vmstat:online, disable vmstat_work reliably on teardown and >> > > symmetrically enable it on startup. >> > > >> > > Signed-off-by: Koichiro Den >> > >> > Hi, >> > >> > I observed a warning in my qemu and real hardware, which I bisected to this commit: >> > >> > [ 0.087733] ------------[ cut here ]------------ >> > [ 0.087733] workqueue: work disable count underflowed >> > [ 0.087733] WARNING: CPU: 1 PID: 21 at kernel/workqueue.c:4313 enable_work+0xb5/0xc0 >> > >> > This is: >> > >> > static void work_offqd_enable(struct work_offq_data *offqd) >> > { >> > if (likely(offqd->disable > 0)) >> > offqd->disable--; >> > else >> > WARN_ONCE(true, "workqueue: work disable count underflowed\n"); <-- this line >> > } >> > >> > So (based on this code) presumably an enable is only required if previously >> > disabled, and this code is being called on startup unconditionally without >> > the work having been disabled previously? I'm not hugely familiar with >> > delayed workqueue implementation details. >> > >> > [ 0.087733] Modules linked in: >> > [ 0.087733] CPU: 1 UID: 0 PID: 21 Comm: cpuhp/1 Not tainted 6.13.0-rc4+ #58 >> > [ 0.087733] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Arch Linux 1.16.3-1-1 04/01/2014 >> > [ 0.087733] RIP: 0010:enable_work+0xb5/0xc0 >> > [ 0.087733] Code: 6f b8 01 00 74 0f 31 d2 be 01 00 00 00 eb b5 90 0f 0b 90 eb ca c6 05 60 6f b8 01 01 90 48 c7 c7 b0 a9 6e 82 e8 4c a4 fd ff 90 <0f> 0b 90 90 eb d6 0f 1f 44 00 00 90 90 90 90 90 90 90 90 90 90 90 >> > [ 0.087733] RSP: 0018:ffffc900000cbe30 EFLAGS: 00010092 >> > [ 0.087733] RAX: 0000000000000029 RBX: ffff888263ca9d60 RCX: 0000000000000000 >> > [ 0.087733] RDX: 0000000000000001 RSI: ffffc900000cbce8 RDI: 0000000000000001 >> > [ 0.087733] RBP: ffffc900000cbe30 R08: 00000000ffffdfff R09: ffffffff82b12f08 >> > [ 0.087733] R10: 0000000000000003 R11: 0000000000000002 R12: 00000000000000c4 >> > [ 0.087733] R13: ffffffff81278d90 R14: 0000000000000000 R15: ffff888263c9c648 >> > [ 0.087733] FS: 0000000000000000(0000) GS:ffff888263c80000(0000) knlGS:0000000000000000 >> > [ 0.087733] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 >> > [ 0.087733] CR2: 0000000000000000 CR3: 0000000002a2e000 CR4: 0000000000750ef0 >> > [ 0.087733] PKRU: 55555554 >> > [ 0.087733] Call Trace: >> > [ 0.087733] >> > [ 0.087733] ? enable_work+0xb5/0xc0 >> > [ 0.087733] ? __warn.cold+0x93/0xf2 >> > [ 0.087733] ? enable_work+0xb5/0xc0 >> > [ 0.087733] ? report_bug+0xff/0x140 >> > [ 0.087733] ? handle_bug+0x54/0x90 >> > [ 0.087733] ? exc_invalid_op+0x17/0x70 >> > [ 0.087733] ? asm_exc_invalid_op+0x1a/0x20 >> > [ 0.087733] ? __pfx_vmstat_cpu_online+0x10/0x10 >> > [ 0.087733] ? enable_work+0xb5/0xc0 >> > [ 0.087733] vmstat_cpu_online+0x5c/0x70 >> > [ 0.087733] cpuhp_invoke_callback+0x133/0x440 >> > [ 0.087733] cpuhp_thread_fun+0x95/0x150 >> > [ 0.087733] smpboot_thread_fn+0xd5/0x1d0 >> > [ 0.087734] ? __pfx_smpboot_thread_fn+0x10/0x10 >> > [ 0.087735] kthread+0xc8/0xf0 >> > [ 0.087737] ? __pfx_kthread+0x10/0x10 >> > [ 0.087738] ret_from_fork+0x2c/0x50 >> > [ 0.087739] ? __pfx_kthread+0x10/0x10 >> > [ 0.087740] ret_from_fork_asm+0x1a/0x30 >> > [ 0.087742] >> > [ 0.087742] ---[ end trace 0000000000000000 ]--- >> > >> > >> > > --- >> > > v1: https://lore.kernel.org/all/20241220134234.3809621-1-koichiro.den@canonical.com/ >> > > --- >> > > mm/vmstat.c | 3 ++- >> > > 1 file changed, 2 insertions(+), 1 deletion(-) >> > > >> > > diff --git a/mm/vmstat.c b/mm/vmstat.c >> > > index 4d016314a56c..0889b75cef14 100644 >> > > --- a/mm/vmstat.c >> > > +++ b/mm/vmstat.c >> > > @@ -2148,13 +2148,14 @@ static int vmstat_cpu_online(unsigned int cpu) >> > > if (!node_state(cpu_to_node(cpu), N_CPU)) { >> > > node_set_state(cpu_to_node(cpu), N_CPU); >> > > } >> > > + enable_delayed_work(&per_cpu(vmstat_work, cpu)); >> > >> > Probably needs to be 'if disabled' here, as this is invoked on normal >> > startup when the work won't have been disabled? >> > >> > Had a brief look at code and couldn't see how that could be done >> > however... and one would need to be careful about races... Tricky! >> > >> > > >> > > return 0; >> > > } >> > > >> > > static int vmstat_cpu_down_prep(unsigned int cpu) >> > > { >> > > - cancel_delayed_work_sync(&per_cpu(vmstat_work, cpu)); >> > > + disable_delayed_work_sync(&per_cpu(vmstat_work, cpu)); >> > > return 0; >> > > } >> > > >> > > -- >> > > 2.43.0 >> > > >> > > >> > >> > Let me know if you need any more details, .config etc. >> > >> > I noticed this warning on a real box too (in both cases running akpm's >> > mm-unstable branch), FWIW. >> >> Thank you for the report. I was able to reproduce the warning and now >> wonder how I missed it.. My oversight, apologies. >> >> In my current view, the simplest solution would be to make sure a local >> vmstat_work is disabled until vmstat_cpu_online() runs for the cpu, even >> during boot-up. The following patch suppresses the warning: >> >> diff --git a/mm/vmstat.c b/mm/vmstat.c >> index 0889b75cef14..19ceed5d34bf 100644 >> --- a/mm/vmstat.c >> +++ b/mm/vmstat.c >> @@ -2122,10 +2122,14 @@ static void __init start_shepherd_timer(void) >> { >> int cpu; >> >> - for_each_possible_cpu(cpu) >> + for_each_possible_cpu(cpu) { >> INIT_DEFERRABLE_WORK(per_cpu_ptr(&vmstat_work, cpu), >> vmstat_update); >> >> + /* will be enabled on vmstat_cpu_online */ >> + disable_delayed_work_sync(&per_cpu(vmstat_work, cpu)); >> + } >> + >> schedule_delayed_work(&shepherd, >> round_jiffies_relative(sysctl_stat_interval)); >> } >> >> If you think of a better solution later, please let me know. Otherwise, >> I'll submit a follow-up fix patch with the above diff. > > Thanks, this resolves the problem, but are we sure that _all_ CPUs will > definitely call vmstat_cpu_online()? > > I did a bit of printk output and it seems like this _didn't_ online CPU 0, > presumably the boot CPU which calls this function in the first instance? FWIW with the proposed fix I can see that all CPUs are online, grep -H . /sys/devices/system/cpu/cpu*/online /sys/devices/system/cpu/cpu0/online:1 /sys/devices/system/cpu/cpu1/online:1 /sys/devices/system/cpu/cpu2/online:1 /sys/devices/system/cpu/cpu3/online:1 /sys/devices/system/cpu/cpu4/online:1 /sys/devices/system/cpu/cpu5/online:1 /sys/devices/system/cpu/cpu6/online:1 /sys/devices/system/cpu/cpu7/online:1 > > I also see that init_mm_internals() invokes cpuhp_setup_state_nocalls() > explicitly which does _not_ call the callback, though even if it did this > would be too early as it calls start_shepherd_timer() _after_ this anyway. > > So yeah, unless I'm missing something, I think this patch is broken. > > I have added Thomas and Peter to give some insight on the CPU hotplug side. > > It feels like the patch really needs an 'enable if not already enabled' > call in vmstat_cpu_online(). > >> >> Thanks. >> >> -Koichiro