Messages in this thread | | | Date | Fri, 29 May 2009 15:14:38 +0200 (CEST) | From | Thomas Gleixner <> | Subject | Re: [patch 2/2] keep on ticking if oprofile is active |
| |
On Fri, 29 May 2009, Martin Schwidefsky wrote: > On Thu, 28 May 2009 22:29:38 +0200 (CEST) > Thomas Gleixner <tglx@linutronix.de> wrote: > > > On Thu, 28 May 2009, Martin Schwidefsky wrote: > > > From: Martin Schwidefsky <schwidefsky@de.ibm.com> > > > > > > On a NOHZ system with oprofile enabled the timer tick should not be > > > stopped when a cpu goes idle. Oprofile needs the pt_regs structure > > > of the interrupt and allocates memory in the ring buffer for each > > > sample. Current a maximum of 1 tick is accounted with oprofile if a > > > cpu sleeps for a longer period of time. This does bad things to the > > > percentages in the oprofile output. To postpone the oprofile tick to > > > tick_nohz_restart_sched_tick analog to the in kernel profiler is not > > > possible as there is no pt_regs structure in the context the > > > tick_nohz_restart_sched_tick function is called and it is not a good > > > idea to create hundreds of samples at once. > > > > Sigh. That's stupid. > > What is stupid, the bug or the fix?
The bug :)
> > OTOH, thinking more about the patch itself it might be even useful > > for things aside oprofile. Runtime switching from and to nohz mode > > for debugging or evaluation purposes comes to my mind. That would > > need some sysfs interface, but that's not too hard to do. > > That should be no problem. We used to have the hz_timer system control > with the old no-tick solution on s390. > > > So yeah, I think we should satisfy oprofile needs and utilize it further. > > Ok, so you are in principle fine with the patch?
Yup.
tglx
| |