Re: [PATCH v5 3/6] perf record: Force enable --timestamp-filename when --switch-output is provided

[Wangnan (F)]

On 2016/4/20 15:59, Jiri Olsa wrote:
> On Mon, Apr 18, 2016 at 02:55:29PM +0000, Wang Nan wrote:
> > Without this patch, the last output doesn't have timestamp appended if
> > --timestamp-filename is not explicitly provided. For example:
> >
> >    # perf record -a --switch-output &
> >    [1] 11224
> >    # kill -s SIGUSR2 11224
> >    [ perf record: dump data: Woken up 1 times ]
> >    # [ perf record: Dump perf.data.2015122622372823 ]
> >
> >    # fg
> >    perf record -a --switch-output
> >    ^C[ perf record: Woken up 1 times to write data ]
> >    [ perf record: Captured and wrote 0.027 MB perf.data (540 samples) ]
> >
> >    # ls -l
> >    total 836
> >    -rw------- 1 root root  33256 Dec 26 22:37 perf.data   <---- *Odd*
> >    -rw------- 1 root root 817156 Dec 26 22:37 perf.data.2015122622372823
> I'm getting this:
>
> [root@krava perf]# ./perf record -a --switch-output &
> [root@krava perf]# kill -s SIGUSR2 18974
> [ perf record: dump data: Woken up 4 times ]
> [ perf record: Dump perf.data.2016042009574314 ]
>
> [root@krava perf]# ./perf report -i perf.data.2016042009574314
> perf: Segmentation fault
> -------- backtrace --------
> ./perf[0x552b0b]
> /lib64/libc.so.6(+0x34a50)[0x7f711b434a50]
> /lib64/libc.so.6(strlen+0x2a)[0x7f711b48b33a]
> ./perf(perf_hpp__reset_sort_width+0x4f)[0x4e9b1f]
> ./perf[0x54b113]
> ./perf(perf_evlist__tui_browse_hists+0x91)[0x551361]
> ./perf(cmd_report+0x1a34)[0x434b44]
> ./perf[0x485681]
> ./perf(main+0x672)[0x424382]
> /lib64/libc.so.6(__libc_start_main+0xf0)[0x7f711b420700]
> ./perf(_start+0x29)[0x4244a9]
> [0x0]

Can't reproduce... Can you get reproduce problem without this patch (3/6)?

My local tree is based on newest perf/core (commit ccd62a8 "perf trace:
Fix build when DWARF unwind isn't available"). Could you please check your
source?

Thank you.

> thanks,
> jirka