lkml.org 
[lkml]   [2015]   [Jul]   [13]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH RFC V3 3/5] perf,tool: partial time support
On Mon, Jul 13, 2015 at 07:01:51PM +0000, Liang, Kan wrote:
> > On Wed, Jul 08, 2015 at 04:44:55AM -0400, kan.liang@intel.com wrote:
> > > From: Kan Liang <kan.liang@intel.com>
> > >
> > > When multiple events are sampled it may not be needed to collect fine
> > > grained time stamps on all events. The sample sites are usually nearby.
> > > It's enough to have time stamps on the regular reference events.
> > > This patchkit adds the ability to turn off time stamps per event. This
> > > in term can reduce sampling overhead and the size of the perf.data.
> >
> > So this patch makes the PERF_SAMPLE_TIME bit set or not independently,
> > right? But AFAIK we sometimes just use first evsel for checking
> > sample_type value, especially for evlist->id_pos. I'm not sure it'll work for
> > all cases of mixed time/notime events..
> >
>
> If the sample_type is different, IDENTIFIER will be set to make sure id is in
> a fixed position. So we don't need to worry about evlist->id_pos here.
>
> I think the only unsupported case should be that the kernel doesn't
> support IDENTIFIER. If so, under my test, perf record will print warning
> " failed to process type". Perf report will also print warning
> " non matching sample_type". So the user can switch back to full time
> support.

Ah, ok then.


> If you think the warning is not good enough, we can show specific
> warning and error out when non IDENTIFIER support is detected.

Well, I think it's good enough. :)

However, perf_evlist__id_hdr_size() seems to be affected also. Are
you sure it's okay too?

Thanks,
Namhyung


\
 
 \ /
  Last update: 2015-07-14 03:21    [W:0.073 / U:0.856 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site