Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-10 Thread Ingo Molnar
* Frederic Weisbecker wrote: > On Wed, May 07, 2014 at 09:07:14PM +0200, Ingo Molnar wrote: > > > > * Peter Zijlstra wrote: > > > > > On Wed, May 07, 2014 at 06:05:08PM +0200, Frederic Weisbecker wrote: > > > > So we can as well zap these commits and replace them with the > > > >

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-10 Thread Ingo Molnar
* Frederic Weisbecker fweis...@gmail.com wrote: On Wed, May 07, 2014 at 09:07:14PM +0200, Ingo Molnar wrote: * Peter Zijlstra pet...@infradead.org wrote: On Wed, May 07, 2014 at 06:05:08PM +0200, Frederic Weisbecker wrote: So we can as well zap these commits and replace them with

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-09 Thread Frederic Weisbecker
On Wed, May 07, 2014 at 09:07:14PM +0200, Ingo Molnar wrote: > > * Peter Zijlstra wrote: > > > On Wed, May 07, 2014 at 06:05:08PM +0200, Frederic Weisbecker wrote: > > > So we can as well zap these commits and replace them with the > > > irq_work_on() > > > conversion (still preparing that). >

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-09 Thread Frederic Weisbecker
On Wed, May 07, 2014 at 09:07:14PM +0200, Ingo Molnar wrote: * Peter Zijlstra pet...@infradead.org wrote: On Wed, May 07, 2014 at 06:05:08PM +0200, Frederic Weisbecker wrote: So we can as well zap these commits and replace them with the irq_work_on() conversion (still preparing

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Ingo Molnar
* Peter Zijlstra wrote: > On Wed, May 07, 2014 at 06:05:08PM +0200, Frederic Weisbecker wrote: > > So we can as well zap these commits and replace them with the irq_work_on() > > conversion (still preparing that). > > I hear you, but I think Ingo doesn't want to rebase the tree because its >

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Peter Zijlstra
On Wed, May 07, 2014 at 06:05:08PM +0200, Frederic Weisbecker wrote: > So we can as well zap these commits and replace them with the irq_work_on() > conversion (still preparing that). I hear you, but I think Ingo doesn't want to rebase the tree because its public, but maybe he can make an

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Frederic Weisbecker
On Wed, May 07, 2014 at 05:37:36PM +0200, Peter Zijlstra wrote: > On Wed, May 07, 2014 at 05:29:24PM +0200, Frederic Weisbecker wrote: > > On Wed, May 07, 2014 at 05:17:35PM +0200, Peter Zijlstra wrote: > > > On Mon, May 05, 2014 at 05:34:08PM +0200, Frederic Weisbecker wrote: > > > > On Mon, May

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Peter Zijlstra
On Wed, May 07, 2014 at 05:29:24PM +0200, Frederic Weisbecker wrote: > On Wed, May 07, 2014 at 05:17:35PM +0200, Peter Zijlstra wrote: > > On Mon, May 05, 2014 at 05:34:08PM +0200, Frederic Weisbecker wrote: > > > On Mon, May 05, 2014 at 05:12:28PM +0200, Peter Zijlstra wrote: > > > > > Note the

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Frederic Weisbecker
On Wed, May 07, 2014 at 05:17:35PM +0200, Peter Zijlstra wrote: > On Mon, May 05, 2014 at 05:34:08PM +0200, Frederic Weisbecker wrote: > > On Mon, May 05, 2014 at 05:12:28PM +0200, Peter Zijlstra wrote: > > > > Note the current ordering: > > > > > > > > cmpxchg(>pending, 0, 1) get ipi >

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Peter Zijlstra
On Mon, May 05, 2014 at 05:34:08PM +0200, Frederic Weisbecker wrote: > On Mon, May 05, 2014 at 05:12:28PM +0200, Peter Zijlstra wrote: > > > Note the current ordering: > > > > > > cmpxchg(>pending, 0, 1) get ipi > > > csd_lock(qsd->csd) xchg(>pending, 1) > > >

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Peter Zijlstra
On Mon, May 05, 2014 at 05:34:08PM +0200, Frederic Weisbecker wrote: On Mon, May 05, 2014 at 05:12:28PM +0200, Peter Zijlstra wrote: Note the current ordering: cmpxchg(qsd-pending, 0, 1) get ipi csd_lock(qsd-csd) xchg(qsd-pending, 1) send ipi

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Frederic Weisbecker
On Wed, May 07, 2014 at 05:17:35PM +0200, Peter Zijlstra wrote: On Mon, May 05, 2014 at 05:34:08PM +0200, Frederic Weisbecker wrote: On Mon, May 05, 2014 at 05:12:28PM +0200, Peter Zijlstra wrote: Note the current ordering: cmpxchg(qsd-pending, 0, 1) get ipi

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Peter Zijlstra
On Wed, May 07, 2014 at 05:29:24PM +0200, Frederic Weisbecker wrote: On Wed, May 07, 2014 at 05:17:35PM +0200, Peter Zijlstra wrote: On Mon, May 05, 2014 at 05:34:08PM +0200, Frederic Weisbecker wrote: On Mon, May 05, 2014 at 05:12:28PM +0200, Peter Zijlstra wrote: Note the current

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Frederic Weisbecker
On Wed, May 07, 2014 at 05:37:36PM +0200, Peter Zijlstra wrote: On Wed, May 07, 2014 at 05:29:24PM +0200, Frederic Weisbecker wrote: On Wed, May 07, 2014 at 05:17:35PM +0200, Peter Zijlstra wrote: On Mon, May 05, 2014 at 05:34:08PM +0200, Frederic Weisbecker wrote: On Mon, May 05, 2014

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Peter Zijlstra
On Wed, May 07, 2014 at 06:05:08PM +0200, Frederic Weisbecker wrote: So we can as well zap these commits and replace them with the irq_work_on() conversion (still preparing that). I hear you, but I think Ingo doesn't want to rebase the tree because its public, but maybe he can make an

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-07 Thread Ingo Molnar
* Peter Zijlstra pet...@infradead.org wrote: On Wed, May 07, 2014 at 06:05:08PM +0200, Frederic Weisbecker wrote: So we can as well zap these commits and replace them with the irq_work_on() conversion (still preparing that). I hear you, but I think Ingo doesn't want to rebase the tree

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Frederic Weisbecker
On Mon, May 05, 2014 at 05:12:28PM +0200, Peter Zijlstra wrote: > > Note the current ordering: > > > > cmpxchg(>pending, 0, 1) get ipi > > csd_lock(qsd->csd) xchg(>pending, 1) > > send ipi csd_unlock(qsd->csd) > > > > > > So there

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Peter Zijlstra
> Note the current ordering: > > cmpxchg(>pending, 0, 1) get ipi > csd_lock(qsd->csd) xchg(>pending, 1) > send ipi csd_unlock(qsd->csd) > > > So there shouldn't be racing updaters. Also ipi sender shouldn't > race with ipi receiver,

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Frederic Weisbecker
On Mon, May 05, 2014 at 04:58:15PM +0200, Peter Zijlstra wrote: > On Mon, May 05, 2014 at 04:52:59PM +0200, Frederic Weisbecker wrote: > > > Should we instead do irq_work_queue_on() ? > > > > I would really much prefer that yeah. But if we do that, expect some added > > overhead on the local > >

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Frederic Weisbecker
On Mon, May 05, 2014 at 03:31:13PM +0200, Peter Zijlstra wrote: > On Mon, May 05, 2014 at 02:37:06PM +0200, Peter Zijlstra wrote: > > On Wed, Apr 16, 2014 at 12:40:01AM -0700, tip-bot for Frederic Weisbecker > > wrote: > > > Commit-ID: 72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 > > > Gitweb:

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Peter Zijlstra
On Mon, May 05, 2014 at 04:52:59PM +0200, Frederic Weisbecker wrote: > > Should we instead do irq_work_queue_on() ? > > I would really much prefer that yeah. But if we do that, expect some added > overhead on the local > irq_work_queue() path though. irq_work_raise can't use local cmpxchg ops >

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Frederic Weisbecker
On Mon, May 05, 2014 at 02:37:06PM +0200, Peter Zijlstra wrote: > On Wed, Apr 16, 2014 at 12:40:01AM -0700, tip-bot for Frederic Weisbecker > wrote: > > Commit-ID: 72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 > > Gitweb: > > http://git.kernel.org/tip/72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 > >

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Peter Zijlstra
On Mon, May 05, 2014 at 02:37:06PM +0200, Peter Zijlstra wrote: > On Wed, Apr 16, 2014 at 12:40:01AM -0700, tip-bot for Frederic Weisbecker > wrote: > > Commit-ID: 72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 > > Gitweb: > > http://git.kernel.org/tip/72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 > >

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Peter Zijlstra
On Wed, Apr 16, 2014 at 12:40:01AM -0700, tip-bot for Frederic Weisbecker wrote: > Commit-ID: 72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 > Gitweb: http://git.kernel.org/tip/72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 > Author: Frederic Weisbecker > AuthorDate: Tue, 18 Mar 2014 21:12:53 +0100

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Peter Zijlstra
On Wed, Apr 16, 2014 at 12:40:01AM -0700, tip-bot for Frederic Weisbecker wrote: Commit-ID: 72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 Gitweb: http://git.kernel.org/tip/72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 Author: Frederic Weisbecker fweis...@gmail.com AuthorDate: Tue, 18 Mar 2014

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Peter Zijlstra
On Mon, May 05, 2014 at 02:37:06PM +0200, Peter Zijlstra wrote: On Wed, Apr 16, 2014 at 12:40:01AM -0700, tip-bot for Frederic Weisbecker wrote: Commit-ID: 72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 Gitweb: http://git.kernel.org/tip/72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 Author:

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Frederic Weisbecker
On Mon, May 05, 2014 at 02:37:06PM +0200, Peter Zijlstra wrote: On Wed, Apr 16, 2014 at 12:40:01AM -0700, tip-bot for Frederic Weisbecker wrote: Commit-ID: 72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 Gitweb: http://git.kernel.org/tip/72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 Author:

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Peter Zijlstra
On Mon, May 05, 2014 at 04:52:59PM +0200, Frederic Weisbecker wrote: Should we instead do irq_work_queue_on() ? I would really much prefer that yeah. But if we do that, expect some added overhead on the local irq_work_queue() path though. irq_work_raise can't use local cmpxchg ops

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Frederic Weisbecker
On Mon, May 05, 2014 at 03:31:13PM +0200, Peter Zijlstra wrote: On Mon, May 05, 2014 at 02:37:06PM +0200, Peter Zijlstra wrote: On Wed, Apr 16, 2014 at 12:40:01AM -0700, tip-bot for Frederic Weisbecker wrote: Commit-ID: 72aacf0259bb7d53b7a3b5b2f7bf982acaa52b61 Gitweb:

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Frederic Weisbecker
On Mon, May 05, 2014 at 04:58:15PM +0200, Peter Zijlstra wrote: On Mon, May 05, 2014 at 04:52:59PM +0200, Frederic Weisbecker wrote: Should we instead do irq_work_queue_on() ? I would really much prefer that yeah. But if we do that, expect some added overhead on the local

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Peter Zijlstra
Note the current ordering: cmpxchg(qsd-pending, 0, 1) get ipi csd_lock(qsd-csd) xchg(qsd-pending, 1) send ipi csd_unlock(qsd-csd) So there shouldn't be racing updaters. Also ipi sender shouldn't race with ipi receiver, the

Re: [tip:timers/nohz] nohz: Move full nohz kick to its own IPI

2014-05-05 Thread Frederic Weisbecker
On Mon, May 05, 2014 at 05:12:28PM +0200, Peter Zijlstra wrote: Note the current ordering: cmpxchg(qsd-pending, 0, 1) get ipi csd_lock(qsd-csd) xchg(qsd-pending, 1) send ipi csd_unlock(qsd-csd) So there shouldn't be