Messages in this thread | | | Date | Sat, 26 Jul 2008 17:26:50 +0200 | From | Ingo Molnar <> | Subject | Re: [PATCH 1/2] ftrace: single CPU tracers use CPU clock |
| |
* Steven Rostedt <rostedt@goodmis.org> wrote:
> > other tracers care too - for example to have the right chronology of > > trace events. The scheduler cares too. What kind of worst-case > > cross-CPU effects have you observed? > > 1: bash-3498 [01] 5459.824565: 0:140:R + 7971:120:R > 2: <idle>-0 [00] 5459.824836: 0:140:R ==> 7971:120:R > 3: bash-3498 [01] 5459.824984: 3498:120:S ==> 0:140:R > 4: <idle>-0 [01] 5459.825342: 0:140:R ==> 7971:120:R > 5: ls-7971 [00] 5459.825380: 7971:120:R + 3: 0:S > 6: ls-7971 [00] 5459.825384: 7971:120:R ==> 3: 0:R > 7: migration/0-3 [00] 5459.825401: 3: 0:S ==> 0:140:R > 8: ls-7971 [01] 5459.825565: 7971:120:R + 598:115:S > > The above was from my tutorial at OLS. I added the annotated numbers > at the beginning of the line. This is the sched_switch tracer.
so it's off by 60 usecs. That matters to the chronology of SMP events, and to the irqsoff and preemptoff tracers - if we update the clock incorrectly - i.e. if we modify the clock across CPUs so that a running CPU can see an 'involuntary' jump in local time.
that should be fixed then - and all tracers (and the scheduler) will improve - instead of having this special-case.
Ingo
| |