On Thu, May 11, 2017 at 7:23 AM, Steven Rostedt <rost...@goodmis.org> wrote: [..] >> To fix the suggestion I made in the very beginning of this email, I >> suggest a combination of trace_find_cmdline type of approach to save >> the tgid of all current threads available, and then using >> trace_sched_fork to find out what we missed (such as threads that were >> added and removed.) I think this approach will be robust and not >> affect any of the existing ftrace users (since we can make it a print >> option). Thoughts? >> >> Update: actually I find that someone already wrote an out-of-tree >> patch for Android long time back during 3.10 kernel days that added a >> 'print-tgid' option that saved tgid in the trace_save_cmdline function >> and then find it later during output time: >> https://android.googlesource.com/kernel/msm.git/+/0c5363f6a0a89952b0d0072ff4e4c3bd042add9a%5E%21/ >> It will be nice if we can have a more upstream ready version of this >> patch or some other approach that you think works best. Thanks! > > Well, make that patch upstream friendly and post it to LMKL (Ccing me > of course). And we'll talk about that. >
Sure, will do. Thanks. Regards, Joel