Messages in this thread | | | Date | Tue, 30 Sep 2008 14:39:06 +0200 | From | "Frédéric Weisbecker" <> | Subject | Re: [PATCH V2 -tip 3/4] Tracing/ftrace: Adapt mmiotrace to the new type of print_line |
| |
2008/9/30 Ingo Molnar <mingo@elte.hu>: > thanks guys! > > Steve's latest trace-ringbuffer code against -tip complicates life a > little bit. > > We dont yet know whether tip/tracing/ring-buffer is ready for wider > consumption, so it's in a separate feature branch. These four patches > conflict with the ring-buffer code quite widely. > > So ... i did an optimistic merge approach: i merged these 4 patches > ontop of the new tip/tracing/ring-buffer code, and have put it into a > new tip/tracing/pipe branch: > > # f19d495: tracing/ftrace: adapt the boot tracer to the new print_line type > # 92261f6: tracing/ftrace: adapt mmiotrace to the new type of print_line > # 157190e: tracing/ftrace: fix pipe breaking > # 92e359a: tracing/ftrace: change the type of the print_line callback > > i'm merging this into tip/master for a bit of testing, but if it breaks > something (or if i havent pushed it out yet) you can merge it locally > too, by doing this ontop of tip/master: > > git-merge tip/tracing/pipe > > it should merge up cleanly and you should be able to check the end > result. (and send me any fixes for merge mistakes) > > ( Once it's all in reasonably well-tested shape it will all show up in > tip/master and you can stop dealing with tip/tracing/* sub-branches.) > > Ingo >
Thanks Ingo.
I'm currently testing tip/master, since tracing/pipe is merged into. It seems there are some locking issues that I hadn't when I tested the patches. Perhaps it's because of the merge with the new ring buffer.
I just launched the sched_switch tracer and it worked well until 4 seconds of execution. And after that it blocked. I tested too a marker message adding and it is never displayed. Worse: I can't break the pipe with Ctrl + C after that.
When I will have some time, I will try to debug all what I find.
| |