Messages in this thread |  | | Date | Wed, 12 Nov 2014 13:52:50 +0100 | From | Peter Zijlstra <> | Subject | Re: [PATCH] sched/cputime: fix clock_nanosleep/clock_gettime inconsistency |
| |
On Wed, Nov 12, 2014 at 01:27:53PM +0100, Stanislaw Gruszka wrote: > On Wed, Nov 12, 2014 at 12:45:38PM +0100, Peter Zijlstra wrote: > > On Wed, Nov 12, 2014 at 12:37:37PM +0100, Peter Zijlstra wrote: > > > > > Which would make the sample run ahead, making the sleep short. So would > > > something like the below not cure things? > > > > Before anyone asks, yes I tried running that 'reproducer' it doesn't. > > I'd be surprised if the patch would help. Issue here happen at start of > cputimer. We set sum_sched_runtime value of cputimer using not yet > accounted threads runtime and then add that runtime values again to > running cputimer on tick, making it's sum_exec_runtime bigger than > actual threads runtime.
Ah yes, reading is hard.
|  |