From: Ingo Molnar <mingo@elte.hu> To: Chris Friesen <cfriesen@nortel.com>, Peter Zijlstra <a.p.zijlstra@chello.nl>, Thomas Gleixner <tglx@linutronix.de>, Steven Rostedt <rostedt@goodmis.org> Cc: David Miller <davem@davemloft.net>, linuxppc-dev@ozlabs.org, paulus@samba.org, netdev@vger.kernel.org Subject: Re: question about softirqs Date: Tue, 12 May 2009 10:12:37 +0200 [thread overview] Message-ID: <20090512081237.GA16403@elte.hu> (raw) In-Reply-To: <4A08C62F.1050105@nortel.com> * Chris Friesen <cfriesen@nortel.com> wrote: > This started out as a thread on the ppc list, but on the > suggestion of DaveM and Paul Mackerras I'm expanding the receiver > list a bit. > > Currently, if a softirq is raised in process context the > TIF_RESCHED_PENDING flag gets set and on return to userspace we > run the scheduler, expecting it to switch to ksoftirqd to handle > the softirqd processing. > > I think I see a possible problem with this. Suppose I have a > SCHED_FIFO task spinning on recvmsg() with MSG_DONTWAIT set. Under > the scenario above, schedule() would re-run the spinning task > rather than ksoftirqd, thus preventing any incoming packets from > being sent up the stack until we get a real hardware > interrupt--which could be a whole jiffy if interrupt mitigation is > enabled in the net device. TIF_RESCHED_PENDING will not be set if a SCHED_FIFO task wakes up a SCHED_OTHER ksoftirqd task. But starvation of ksoftirqd processing will occur. > DaveM pointed out that if we're doing transmits we're likely to > hit local_bh_enable(), which would process the softirq work. > However, I think we may still have a problem in the above rx-only > scenario--or is it too contrived to matter? This could occur, and the problem is really that task priorities do not extend across softirq work processing. This could occur in ordinary SCHED_OTHER tasks as well, if the softirq is bounced to ksoftirqd - which it only should be if there's serious softirq overload - or, as you describe it above, if the softirq is raised in process context: if (!in_interrupt()) wakeup_softirqd(); that's not really clean. We look into eliminating process context use of raise_softirq_irqsoff(). Such code sequence: local_irq_save(flags); ... raise_softirq_irqsoff(nr); ... local_irq_restore(flags); should be converted to something like: local_irq_save(flags); ... raise_softirq_irqsoff(nr); ... local_irq_restore(flags); recheck_softirqs(); If someone does not do proper local_bh_disable()/enable() sequences for micro-optimization reasons, then push the check to after the critcal section - and dont cause extra reschedules by waking up ksoftirqd. raise_softirq_irqsoff() will also be faster. Ingo
WARNING: multiple messages have this Message-ID (diff)
From: Ingo Molnar <mingo@elte.hu> To: Chris Friesen <cfriesen@nortel.com>, Peter Zijlstra <a.p.zijlstra@chello.nl>, Thomas Gleixner <tglx@linutronix.de>, Steven Rostedt <rostedt@goodmis.org> Cc: linuxppc-dev@ozlabs.org, paulus@samba.org, David Miller <davem@davemloft.net>, netdev@vger.kernel.org Subject: Re: question about softirqs Date: Tue, 12 May 2009 10:12:37 +0200 [thread overview] Message-ID: <20090512081237.GA16403@elte.hu> (raw) In-Reply-To: <4A08C62F.1050105@nortel.com> * Chris Friesen <cfriesen@nortel.com> wrote: > This started out as a thread on the ppc list, but on the > suggestion of DaveM and Paul Mackerras I'm expanding the receiver > list a bit. > > Currently, if a softirq is raised in process context the > TIF_RESCHED_PENDING flag gets set and on return to userspace we > run the scheduler, expecting it to switch to ksoftirqd to handle > the softirqd processing. > > I think I see a possible problem with this. Suppose I have a > SCHED_FIFO task spinning on recvmsg() with MSG_DONTWAIT set. Under > the scenario above, schedule() would re-run the spinning task > rather than ksoftirqd, thus preventing any incoming packets from > being sent up the stack until we get a real hardware > interrupt--which could be a whole jiffy if interrupt mitigation is > enabled in the net device. TIF_RESCHED_PENDING will not be set if a SCHED_FIFO task wakes up a SCHED_OTHER ksoftirqd task. But starvation of ksoftirqd processing will occur. > DaveM pointed out that if we're doing transmits we're likely to > hit local_bh_enable(), which would process the softirq work. > However, I think we may still have a problem in the above rx-only > scenario--or is it too contrived to matter? This could occur, and the problem is really that task priorities do not extend across softirq work processing. This could occur in ordinary SCHED_OTHER tasks as well, if the softirq is bounced to ksoftirqd - which it only should be if there's serious softirq overload - or, as you describe it above, if the softirq is raised in process context: if (!in_interrupt()) wakeup_softirqd(); that's not really clean. We look into eliminating process context use of raise_softirq_irqsoff(). Such code sequence: local_irq_save(flags); ... raise_softirq_irqsoff(nr); ... local_irq_restore(flags); should be converted to something like: local_irq_save(flags); ... raise_softirq_irqsoff(nr); ... local_irq_restore(flags); recheck_softirqs(); If someone does not do proper local_bh_disable()/enable() sequences for micro-optimization reasons, then push the check to after the critcal section - and dont cause extra reschedules by waking up ksoftirqd. raise_softirq_irqsoff() will also be faster. Ingo
next prev parent reply other threads:[~2009-05-12 8:13 UTC|newest] Thread overview: 57+ messages / expand[flat|nested] mbox.gz Atom feed top 2009-05-08 22:51 question about softirqs Chris Friesen 2009-05-08 23:05 ` David Miller 2009-05-08 23:34 ` Paul Mackerras 2009-05-08 23:53 ` David Miller 2009-05-09 2:52 ` Benjamin Herrenschmidt 2009-05-09 3:31 ` Paul Mackerras 2009-05-09 6:48 ` David Miller 2009-05-11 18:25 ` Chris Friesen 2009-05-11 23:24 ` David Miller 2009-05-12 0:43 ` Chris Friesen 2009-05-12 8:12 ` Ingo Molnar [this message] 2009-05-12 8:12 ` Ingo Molnar 2009-05-12 9:12 ` Peter Zijlstra 2009-05-12 9:23 ` Ingo Molnar 2009-05-12 9:32 ` Peter Zijlstra 2009-05-12 12:20 ` Steven Rostedt 2009-05-12 12:20 ` Steven Rostedt 2009-05-13 4:45 ` David Miller 2009-05-13 4:44 ` David Miller 2009-05-13 4:44 ` David Miller 2009-05-13 5:15 ` Paul Mackerras 2009-05-13 5:15 ` Paul Mackerras 2009-05-13 5:28 ` David Miller 2009-05-13 5:28 ` David Miller 2009-05-13 5:55 ` Evgeniy Polyakov 2009-05-13 5:55 ` Evgeniy Polyakov 2009-05-12 15:18 ` Chris Friesen 2009-05-12 15:18 ` Chris Friesen 2009-05-13 8:34 ` Andi Kleen 2009-05-13 8:34 ` Andi Kleen 2009-05-13 13:23 ` Chris Friesen 2009-05-13 14:15 ` Andi Kleen 2009-05-13 14:15 ` Andi Kleen 2009-05-13 14:17 ` Thomas Gleixner 2009-05-13 14:17 ` Thomas Gleixner 2009-05-13 14:24 ` Andi Kleen 2009-05-13 14:24 ` Andi Kleen 2009-05-13 14:54 ` Eric Dumazet 2009-05-13 14:54 ` Eric Dumazet 2009-05-13 15:02 ` Andi Kleen 2009-05-13 15:02 ` Andi Kleen 2009-05-13 15:05 ` Chris Friesen 2009-05-13 15:54 ` Thomas Gleixner 2009-05-13 15:54 ` Thomas Gleixner 2009-05-13 16:10 ` Chris Friesen 2009-05-13 17:01 ` Andi Kleen 2009-05-13 19:04 ` Chris Friesen 2009-05-13 19:04 ` Chris Friesen 2009-05-13 19:13 ` Andi Kleen 2009-05-13 19:13 ` Andi Kleen 2009-05-13 19:44 ` Chris Friesen 2009-05-13 19:53 ` Andi Kleen 2009-05-13 19:53 ` Andi Kleen 2009-05-13 20:55 ` Thomas Gleixner 2009-05-13 20:55 ` Thomas Gleixner 2009-05-11 23:34 ` Paul Mackerras 2009-05-09 0:28 ` Chris Friesen
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=20090512081237.GA16403@elte.hu \ --to=mingo@elte.hu \ --cc=a.p.zijlstra@chello.nl \ --cc=cfriesen@nortel.com \ --cc=davem@davemloft.net \ --cc=linuxppc-dev@ozlabs.org \ --cc=netdev@vger.kernel.org \ --cc=paulus@samba.org \ --cc=rostedt@goodmis.org \ --cc=tglx@linutronix.de \ /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.