Messages in this thread | | | Date | Thu, 8 Jan 2009 10:05:02 +0100 | From | Frederic Weisbecker <> | Subject | Re: [PATCH 2/2] tracing/branch-tracer: adapt to the stat tracing API |
| |
On Wed, Jan 07, 2009 at 11:49:49PM -0500, Steven Rostedt wrote: > > > On Sat, 27 Dec 2008, Frederic Weisbecker wrote: > > > This patch adapts the branch tracer to the tracing API. > > This is not really for inclusion but more likely a proof of concept because > > the branch tracer implements two "stat tracing" that were split in two files. > > > > So I added an option to the branch tracer: stat_all_branch. > > If it is set, then trace_stat will output all of the branches entries stats. > > Otherwise, it will print the annotated branches. > > > > Its is a kind of quick trick, waiting for a better solution. > > By default, the annotated branches stat are sorted by incorrect branch prediction > > percentage. > > > > OK, I have not had a chance to look at this code since it was sent when I > was on Holiday. But I do not think that the only way to see the output of > a histogram is to set it in the current_tracer. If I have the branch > tracer on, I would have to run it to see the output of the histogram. > > The histogram starts recording at boot up. Although the branch tracer is > dependent on the histograms, the histogram is not dependent on the branch > tracer. This code is making the histogram dependent on the trace. It took > me a while to figure out how to see my histogram. > > With the new patches (after the compile issue is solved) we could have a > register_histogram or something that would put the file in the trace_stats > directory. This would separate the need of coupling the tracers with the > histogram even when they are not coupled. > > -- Steve
You're right, that's a better and more scalable idea. Should I do it on top of the current patches (when the compile is fixed?). That will be more easy...
Thanks.
| |