Messages in this thread | | | Date | Wed, 15 Feb 2012 11:47:33 -0200 | From | Arnaldo Carvalho de Melo <> | Subject | Re: [PATCH] Perf: bug fix: distinguish between rename and exec |
| |
Em Wed, Feb 15, 2012 at 01:48:33PM +0100, Peter Zijlstra escreveu: > I really dislike changing generic code purely for the purpose of > instrumentation like this. Better to pull perf_event_comm() out of here > if you want to change semantics. > > Personally I couldn't care less about renames, I think they're a waste > of time, so I'm ok with the simple patch moving the perf_event_comm() > into setup_new_exec() and possibly renaming it to perf_event_exec(). > > Acme, do you care about renames?
I like your idea of keeping the semantics of PERF_RECORD_COMM and introducing a PERF_RECORD_EXEC, just have to think about how to handle that in a way that the tools detect that we have PERF_RECORD_EXEC...
Humm, will be yet another fallback for setting an perf_event_attr bit, just like with .sample_id_all and .exclude_{guest,host}...
That together with the per class errnos + __strerror() method will allow to move all the event creation finally to perf_evlist__open() where all this gets nicely hidden away from poor tools.
We can then even have an ui__evlist_perror() method that does all the ui__warning calls, etc.
So, yes, from a tooling perspective, I want to be notified of renames and being able to stop relying on PERF_RECORD_COMM to call map_groups__flush and instead do it at PERF_RECORD_EXEC seems a bonus.
- Arnaldo
| |