Em Wed, Oct 18, 2017 at 07:54:13PM +0800, 禹舟键 escreveu: > Hi, Arnaldo > I have done it as you said in v1 patch, but David had told me to separate > the patch :"Makes all those related functions receive the FILE pointer" to > two patches. Below is his opinion. > > On 9/13/17 9:10 AM, yuzhoujian wrote: > > @@ -1621,8 +1634,12 @@ static int process_comm_event(struct perf_tool > *tool, > > sample->tid = event->comm.tid; > > sample->pid = event->comm.pid; > > } > > - print_sample_start(sample, thread, evsel); > > - perf_event__fprintf(event, stdout); > > + if (tool->orientation_output == false) > > + fp = stdout; > > + else > > + fp = orientation_file; > > + fprint_sample_start(sample, thread, evsel, fp); > > + perf_event__fprintf(event, fp); > > ret = 0; > > out: > > thread__put(thread); > > "The subject of this patch is replacing printf and stdout with fprintf > and a given fp. Please keep it to that one change. Meaning the above > setting of fp something other than stdout should be a separate patch.
I have done exactly as described in my and David's suggestion, this "orientation_file" change is not in my patch. > And it would be best to have the fp selection in a helper, versus the > same change in so many places. " > > So, what should I do?? Please consider continuing from the other suggestion I made in a recent message, about how to use evsel->priv to store the perf evsel FILE pointer. - Arnaldo
