On Sat, Oct 27, 2018 at 01:09:44PM -0700, Vinicius Costa Gomes wrote: > Hi Jirka, > > Jiri Olsa <jo...@redhat.com> writes: > > > On Fri, Oct 26, 2018 at 04:19:52PM -0700, Vinicius Costa Gomes wrote: > >> Hi, > >> > >> Adrian Hunter <adrian.hun...@intel.com> writes: > >> > >> > On 18/10/18 1:55 AM, Arnaldo Carvalho de Melo wrote: > >> >> From: Arnaldo Carvalho de Melo <a...@redhat.com> > >> >> > >> >> David reports that: > >> >> > >> >> <quote> > >> >> Perf has this hack where it uses the kernel symbol map as a backup when > >> >> a symbol can't be found in the user's symbol table(s). > >> > > >> > I don't think this is a complete fix because it exposes new problems. > >> > >> This commit broke function name resolution for 'perf record -g' for me. > >> > >> What I mean is, with this commit applied: > >> > >> $ ./tools/perf/perf record -g -- sleep 1 > >> > >> $ ./tools/perf/perf report > >> > >> 'perf report' doesn't seem to be able to show the function names of the > >> trace. > >> > >> If I revert this commit, function names are resolved fine. > > > > that commit just showed up some places where we have the > > ip resolve wrong.. would attached patch fix it for you? > > Thank you for your patch. > > I can some difference in the output, but I wouldn't say that it's fixed. > > Here are some samples, if it's useful somehow: > > https://gist.github.com/vcgomes/985626705e0968b973e426964f86a4b0 > > The "ping" tests were done running > > $ sudo ./tools/perf/perf record -g -- ping -f -c 1000 127.0.0.1 > > And the "sleep" tests were done running > > $ sudo ./tools/perf/perf record -g -- /bin/sleep 1
ugh, I tried with 'sudo ./perf record -g' and it looks like it matter to callchains if there's a workload Adrian said he's preparing complex patch for this let's wait for his changes thanks, jirka
