Re: [GIT PULL v2] printk: Make it usable on nohz cpus
2012/12/8 Ingo Molnar : > > * Frederic Weisbecker wrote: > >> Ingo, >> >> Please pull the printk support in dynticks mode patches that can >> be found at: >> >> git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git >> tags/printk-dynticks-for-mingo-v2 >> >> for you to fetch changes up to 74876a98a87a115254b3a66a14b27320b7f0acaa: >> >> printk: Wake up klogd using irq_work (2012-11-18 01:01:49 +0100) >> >> It is based on v3.7-rc4. >> >> Changes since previous pull request include support for irq >> work flush on CPU offlining and acks from Steve. The rest >> hasn't changed except some comment fix. > > Since this changes kernel/printk.c it needs Linus's ack. > > I looked through the older submissions but found no good summary > of these changes: it would be nice if you could write up a good > high level description of these changes - why has printk based > kernel message logging become problematic on nohz, what are the > symptoms to users, and what are the solution alternatives you > found and please justify the irq_work extension variant you > picked. > > I more or less accept the fact that fixes are needed here, but > the linecount appears a bit high. It's possibly unavoidable, but > would be nice to have a discussion of it, as printk is something > we really, really want to keep as simple as possible. Sorry, I focused too much on details and it indeed lacks a detailed general explanation. printk() is problematic for the full dynticks implementation that we are working on because it depends on the tick to stay periodic in order to wake up the potential readers sleeping on the syslog() syscall. printk() can be called about anywhere (sort of), so when a new message is written in the buffer while there are pending readers, printk wakes them up asynchronously instead of doing it in place in order to avoid some gory locking scenarios with the scheduler locks. This asynchronous wake up is handled by a hook on the timer tick: printk_tick(). This is called every timer interrupt. So if we stop the tick outside idle, and some printk() is called while the CPU runs in full dynticks mode while we have pending readers, those may not be woken for a while. Hence the user may miss some important message. So if we want to enter in full dynticks mode safely, I fear we don't have much choices. We need to find some event driven rather than periodic driven way to perform the asynchronous wake up. The irq work subsystem is a very good fit for that because: * it can typically raise self-IPIs (some archs haven't implemented that yet but this will be a requirement for full dynticks mode) * it's light, simple and standalone. This is what we want for an API used by printk() * this is lockless, which is a requirement for printk at this stage. Also irq work subsystem has its own hook on the timer tick in case the arch can't perform self-IPIs. So if the tick is not stopped while printk() is called, we can even avoid the IPI. This may be desirable if we have situations with lots of printk() in a short period of time: this can avoid an IPI storm. Also this is an immediate benefit for mainline because we can now remove the ad-hoc printk_tick() timer hook. One less function call and per-cpu check from the timer interrupt is a win. What do you guys think? -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [GIT PULL v2] printk: Make it usable on nohz cpus
* Frederic Weisbecker wrote: > Ingo, > > Please pull the printk support in dynticks mode patches that can > be found at: > > git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git > tags/printk-dynticks-for-mingo-v2 > > for you to fetch changes up to 74876a98a87a115254b3a66a14b27320b7f0acaa: > > printk: Wake up klogd using irq_work (2012-11-18 01:01:49 +0100) > > It is based on v3.7-rc4. > > Changes since previous pull request include support for irq > work flush on CPU offlining and acks from Steve. The rest > hasn't changed except some comment fix. Since this changes kernel/printk.c it needs Linus's ack. I looked through the older submissions but found no good summary of these changes: it would be nice if you could write up a good high level description of these changes - why has printk based kernel message logging become problematic on nohz, what are the symptoms to users, and what are the solution alternatives you found and please justify the irq_work extension variant you picked. I more or less accept the fact that fixes are needed here, but the linecount appears a bit high. It's possibly unavoidable, but would be nice to have a discussion of it, as printk is something we really, really want to keep as simple as possible. Thanks, Ingo -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[GIT PULL v2] printk: Make it usable on nohz cpus
Ingo, Please pull the printk support in dynticks mode patches that can be found at: git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git tags/printk-dynticks-for-mingo-v2 for you to fetch changes up to 74876a98a87a115254b3a66a14b27320b7f0acaa: printk: Wake up klogd using irq_work (2012-11-18 01:01:49 +0100) It is based on v3.7-rc4. Changes since previous pull request include support for irq work flush on CPU offlining and acks from Steve. The rest hasn't changed except some comment fix. Thanks. Support for printk in dynticks mode: * Fix two races in irq work claiming * Generalize irq_work support to all archs * Don't stop tick with irq works pending. This fix is generally useful and concerns archs that can't raise self IPIs. * Flush irq works before CPU offlining. * Introduce "lazy" irq works that can wait for the next tick to be executed, unless it's stopped. * Implement klogd wake up using irq work. This removes the ad-hoc printk_tick()/printk_needs_cpu() hooks and make it working even in dynticks mode. Signed-off-by: Frederic Weisbecker Frederic Weisbecker (7): irq_work: Fix racy IRQ_WORK_BUSY flag setting irq_work: Fix racy check on work pending flag irq_work: Remove CONFIG_HAVE_IRQ_WORK nohz: Add API to check tick state irq_work: Don't stop the tick with pending works irq_work: Make self-IPIs optable printk: Wake up klogd using irq_work Steven Rostedt (2): irq_work: Flush work on CPU_DYING irq_work: Warn if there's still work on cpu_down arch/alpha/Kconfig |1 - arch/arm/Kconfig|1 - arch/arm64/Kconfig |1 - arch/blackfin/Kconfig |1 - arch/frv/Kconfig|1 - arch/hexagon/Kconfig|1 - arch/mips/Kconfig |1 - arch/parisc/Kconfig |1 - arch/powerpc/Kconfig|1 - arch/s390/Kconfig |1 - arch/sh/Kconfig |1 - arch/sparc/Kconfig |1 - arch/x86/Kconfig|1 - drivers/staging/iio/trigger/Kconfig |1 - include/linux/irq_work.h| 20 ++ include/linux/printk.h |3 - include/linux/tick.h| 17 - init/Kconfig|5 +- kernel/irq_work.c | 131 ++- kernel/printk.c | 36 +- kernel/time/tick-sched.c|7 +- kernel/timer.c |1 - 22 files changed, 161 insertions(+), 73 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
