From: Tejun Heo <tj@kernel.org> To: Peter Zijlstra <peterz@infradead.org> Cc: Christoph Lameter <cl@gentwo.org>, Sasha Levin <sasha.levin@oracle.com>, akpm@linux-foundation.org, Gilad Ben-Yossef <gilad@benyossef.com>, Thomas Gleixner <tglx@linutronix.de>, John Stultz <johnstul@us.ibm.com>, Mike Frysinger <vapier@gentoo.org>, Minchan Kim <minchan.kim@gmail.com>, Hakan Akkan <hakanakkan@gmail.com>, Max Krasnyansky <maxk@qualcomm.com>, Frederic Weisbecker <fweisbec@gmail.com>, "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>, linux-kernel@vger.kernel.org, linux-mm@kvack.org, hughd@google.com, viresh.kumar@linaro.org, hpa@zytor.com, mingo@kernel.org, Lai Jiangshan <laijs@cn.fujitsu.com> Subject: Re: vmstat: On demand vmstat workers V8 Date: Tue, 29 Jul 2014 09:12:26 -0400 [thread overview] Message-ID: <20140729131226.GS7462@htj.dyndns.org> (raw) In-Reply-To: <20140729122303.GA3935@laptop> (cc'ing Lai) Hello, On Tue, Jul 29, 2014 at 02:23:03PM +0200, Peter Zijlstra wrote: > > It's because we don't distinguish work items which are per-cpu for > > optimization and per-cpu for correctness and can't automatically flush > > / cancel / block per-cpu work items when a cpu goes down. I like the > > idea of distingushing them but it's gonna take a lot of auditing. > > Just force flush on unplug and fix those that complain. No auditing > needed for that. I'm not sure that's a viable way forward. It's not like we can readily trigger the problematic cases which can lead to long pauses during cpu down. Besides, we need the distinction at the API level, which is the whole point of this. The best way probably is converting all the correctness ones (these are the minorities) over to queue_work_on() so that the per-cpu requirement is explicit. > > Any work item usage which requires per-cpu for correctness should > > implement cpu down hook to flush in-flight work items and block > > further issuance. This hasn't changed from the beginning and was > > necessary even before cmwq. > > I think before cmwq we'd run into the broken affinity warning in the > scheduler. That and work items silently not executed if queued on a downed cpu. IIRC, we also had quite a few broken ones which were per-cpu but w/o cpu down handling which just happened to work most of the time because queueing itself was per-cpu in most cases and we didn't do cpu on/offlining as often back then. During cmwq conversion, I just allowed them as I didn't want to add cpu down hooks for all of the many per-cpu workqueue usages. The lack of the distinction between the two sets has always been there. I agree this can be improved, but at least for now, please add cpu down hooks. We need them right now and they'll be helpful when later separating out the correctness ones. Thanks. -- tejun
WARNING: multiple messages have this Message-ID (diff)
From: Tejun Heo <tj@kernel.org> To: Peter Zijlstra <peterz@infradead.org> Cc: Christoph Lameter <cl@gentwo.org>, Sasha Levin <sasha.levin@oracle.com>, akpm@linux-foundation.org, Gilad Ben-Yossef <gilad@benyossef.com>, Thomas Gleixner <tglx@linutronix.de>, John Stultz <johnstul@us.ibm.com>, Mike Frysinger <vapier@gentoo.org>, Minchan Kim <minchan.kim@gmail.com>, Hakan Akkan <hakanakkan@gmail.com>, Max Krasnyansky <maxk@qualcomm.com>, Frederic Weisbecker <fweisbec@gmail.com>, "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>, linux-kernel@vger.kernel.org, linux-mm@kvack.org, hughd@google.com, viresh.kumar@linaro.org, hpa@zytor.com, mingo@kernel.org, Lai Jiangshan <laijs@cn.fujitsu.com> Subject: Re: vmstat: On demand vmstat workers V8 Date: Tue, 29 Jul 2014 09:12:26 -0400 [thread overview] Message-ID: <20140729131226.GS7462@htj.dyndns.org> (raw) In-Reply-To: <20140729122303.GA3935@laptop> (cc'ing Lai) Hello, On Tue, Jul 29, 2014 at 02:23:03PM +0200, Peter Zijlstra wrote: > > It's because we don't distinguish work items which are per-cpu for > > optimization and per-cpu for correctness and can't automatically flush > > / cancel / block per-cpu work items when a cpu goes down. I like the > > idea of distingushing them but it's gonna take a lot of auditing. > > Just force flush on unplug and fix those that complain. No auditing > needed for that. I'm not sure that's a viable way forward. It's not like we can readily trigger the problematic cases which can lead to long pauses during cpu down. Besides, we need the distinction at the API level, which is the whole point of this. The best way probably is converting all the correctness ones (these are the minorities) over to queue_work_on() so that the per-cpu requirement is explicit. > > Any work item usage which requires per-cpu for correctness should > > implement cpu down hook to flush in-flight work items and block > > further issuance. This hasn't changed from the beginning and was > > necessary even before cmwq. > > I think before cmwq we'd run into the broken affinity warning in the > scheduler. That and work items silently not executed if queued on a downed cpu. IIRC, we also had quite a few broken ones which were per-cpu but w/o cpu down handling which just happened to work most of the time because queueing itself was per-cpu in most cases and we didn't do cpu on/offlining as often back then. During cmwq conversion, I just allowed them as I didn't want to add cpu down hooks for all of the many per-cpu workqueue usages. The lack of the distinction between the two sets has always been there. I agree this can be improved, but at least for now, please add cpu down hooks. We need them right now and they'll be helpful when later separating out the correctness ones. Thanks. -- tejun -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2014-07-29 13:12 UTC|newest] Thread overview: 73+ messages / expand[flat|nested] mbox.gz Atom feed top 2014-07-10 14:04 vmstat: On demand vmstat workers V8 Christoph Lameter 2014-07-10 14:04 ` Christoph Lameter 2014-07-11 13:20 ` Frederic Weisbecker 2014-07-11 13:20 ` Frederic Weisbecker 2014-07-11 13:56 ` Christoph Lameter 2014-07-11 13:56 ` Christoph Lameter 2014-07-11 13:58 ` Frederic Weisbecker 2014-07-11 13:58 ` Frederic Weisbecker 2014-07-11 15:17 ` Christoph Lameter 2014-07-11 15:17 ` Christoph Lameter 2014-07-11 15:19 ` Frederic Weisbecker 2014-07-11 15:19 ` Frederic Weisbecker 2014-07-11 15:22 ` Christoph Lameter 2014-07-11 15:22 ` Christoph Lameter 2014-07-14 20:10 ` Hugh Dickins 2014-07-14 20:10 ` Hugh Dickins 2014-07-14 20:51 ` Christoph Lameter 2014-07-14 20:51 ` Christoph Lameter 2014-07-30 3:04 ` Lai Jiangshan 2014-07-30 3:04 ` Lai Jiangshan 2014-07-26 2:22 ` Sasha Levin 2014-07-26 2:22 ` Sasha Levin 2014-07-28 18:55 ` Christoph Lameter 2014-07-28 18:55 ` Christoph Lameter 2014-07-28 21:54 ` Andrew Morton 2014-07-28 21:54 ` Andrew Morton 2014-07-28 22:00 ` Sasha Levin 2014-07-28 22:00 ` Sasha Levin 2014-07-29 15:17 ` Christoph Lameter 2014-07-29 15:17 ` Christoph Lameter 2014-07-29 7:56 ` Peter Zijlstra 2014-07-29 12:05 ` Tejun Heo 2014-07-29 12:05 ` Tejun Heo 2014-07-29 12:23 ` Peter Zijlstra 2014-07-29 12:23 ` Peter Zijlstra 2014-07-29 13:12 ` Tejun Heo [this message] 2014-07-29 13:12 ` Tejun Heo 2014-07-29 15:10 ` Christoph Lameter 2014-07-29 15:10 ` Christoph Lameter 2014-07-29 15:14 ` Tejun Heo 2014-07-29 15:14 ` Tejun Heo 2014-07-29 15:26 ` Christoph Lameter 2014-07-29 15:26 ` Christoph Lameter 2014-07-29 15:39 ` Christoph Lameter 2014-07-29 15:39 ` Christoph Lameter 2014-07-29 15:47 ` Sasha Levin 2014-07-29 15:47 ` Sasha Levin 2014-07-29 15:59 ` Christoph Lameter 2014-07-29 15:59 ` Christoph Lameter 2014-07-30 3:11 ` Lai Jiangshan 2014-07-30 3:11 ` Lai Jiangshan 2014-07-30 14:34 ` Christoph Lameter 2014-07-30 14:34 ` Christoph Lameter 2014-07-29 15:22 ` Christoph Lameter 2014-07-29 15:22 ` Christoph Lameter 2014-07-29 15:43 ` Sasha Levin 2014-07-29 15:43 ` Sasha Levin 2014-08-04 21:37 ` Sasha Levin 2014-08-04 21:37 ` Sasha Levin 2014-08-05 14:51 ` Christoph Lameter 2014-08-05 14:51 ` Christoph Lameter 2014-08-05 22:25 ` Sasha Levin 2014-08-05 22:25 ` Sasha Levin 2014-08-06 14:12 ` Christoph Lameter 2014-08-06 14:12 ` Christoph Lameter 2014-08-07 1:50 ` Sasha Levin 2014-08-07 1:50 ` Sasha Levin 2014-07-30 2:57 ` Lai Jiangshan 2014-07-30 2:57 ` Lai Jiangshan 2014-07-30 14:45 ` Christoph Lameter 2014-07-30 14:45 ` Christoph Lameter 2014-07-31 0:52 ` Lai Jiangshan 2014-07-31 0:52 ` Lai Jiangshan
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=20140729131226.GS7462@htj.dyndns.org \ --to=tj@kernel.org \ --cc=akpm@linux-foundation.org \ --cc=cl@gentwo.org \ --cc=fweisbec@gmail.com \ --cc=gilad@benyossef.com \ --cc=hakanakkan@gmail.com \ --cc=hpa@zytor.com \ --cc=hughd@google.com \ --cc=johnstul@us.ibm.com \ --cc=laijs@cn.fujitsu.com \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-mm@kvack.org \ --cc=maxk@qualcomm.com \ --cc=minchan.kim@gmail.com \ --cc=mingo@kernel.org \ --cc=paulmck@linux.vnet.ibm.com \ --cc=peterz@infradead.org \ --cc=sasha.levin@oracle.com \ --cc=tglx@linutronix.de \ --cc=vapier@gentoo.org \ --cc=viresh.kumar@linaro.org \ /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: linkBe 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.