Date:   Wed, 16 Nov 2022 08:35:27 -0300
From:   Arnaldo Carvalho de Melo <acme@...nel.org>
To:     Ian Rogers <irogers@...gle.com>
Cc:     Weilin Wang <weilin.wang@...el.com>,
        Perry Taylor <perry.taylor@...el.com>,
        Caleb Biggers <caleb.biggers@...el.com>,
        Leo Yan <leo.yan@...aro.org>,
        Adrian Hunter <adrian.hunter@...el.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Mark Rutland <mark.rutland@....com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Jiri Olsa <jolsa@...nel.org>,
        Namhyung Kim <namhyung@...nel.org>,
        Sandipan Das <sandipan.das@....com>,
        Kajol Jain <kjain@...ux.ibm.com>,
        Zhengjun Xing <zhengjun.xing@...ux.intel.com>,
        Kan Liang <kan.liang@...ux.intel.com>,
        Ravi Bangoria <ravi.bangoria@....com>,
        Xin Gao <gaoxin@...rlc.com>, Rob Herring <robh@...nel.org>,
        linux-kernel@...r.kernel.org, linux-perf-users@...r.kernel.org,
        Stephane Eranian <eranian@...gle.com>
Subject: Re: [PATCH v3 10/10] perf list: Add json output option

Em Wed, Nov 16, 2022 at 08:23:07AM -0300, Arnaldo Carvalho de Melo escreveu:
> Em Tue, Nov 15, 2022 at 10:44:54AM -0300, Arnaldo Carvalho de Melo escreveu:
> > Em Mon, Nov 14, 2022 at 01:07:23PM -0800, Ian Rogers escreveu:
> > > Output events and metrics in a json format by overriding the print
> > > callbacks. Currently other command line options aren't supported and
> > > metrics are repeated once per metric group.
> > 
> > Applied the patch with a few fixes and added this to the last cset:
> 
> There is a problem, detected using 'perf test', the last one:
> 
> [root@...co ~]# perf test "trace + vfs_getname"
> 112: Check open filename arg using perf trace + vfs_getname          : FAILED!
> [root@...co ~]# perf test -v "trace + vfs_getname"
> 112: Check open filename arg using perf trace + vfs_getname          :
> --- start ---
> test child forked, pid 611667
> test child finished with -1
> ---- end ----
> Check open filename arg using perf trace + vfs_getname: FAILED!
> [root@...co ~]#
> 
> Verbose didn't help, so I looked at the shell script for that test and
> the problem is here:
> 
> [root@...co ~]# perf list syscalls:sys_enter_open*
> Segmentation fault (core dumped)
> ^C
> [root@...co ~]#
> [root@...co ~]#
> [root@...co ~]# gdb perf
> GNU gdb (GDB) Fedora 12.1-2.fc36
> (gdb) run list syscalls:sys_enter_open*
> Starting program: /root/bin/perf list syscalls:sys_enter_open*
> [Thread debugging using libthread_db enabled]
> Using host libthread_db library "/lib64/libthread_db.so.1".
> [Detaching after fork from child process 611757]
> 
> Program received signal SIGSEGV, Segmentation fault.
> 0x00000000004e742b in __match_glob (str=0x0, pat=0xe44900 "syscalls", ignore_space=ignore_space@...ry=false, case_ins=case_ins@...ry=false) at util/string.c:113
> 113		while (*str && *pat && *pat != '*') {
> Missing separate debuginfos, use: dnf debuginfo-install bzip2-libs-1.0.8-11.fc36.x86_64 cyrus-sasl-lib-2.1.27-18.fc36.x86_64 elfutils-debuginfod-client-0.187-4.fc36.x86_64 elfutils-libelf-0.187-4.fc36.x86_64 elfutils-libs-0.187-4.fc36.x86_64 glib2-2.72.3-1.fc36.x86_64 glibc-2.35-20.fc36.x86_64 keyutils-libs-1.6.1-4.fc36.x86_64 krb5-libs-1.19.2-11.fc36.x86_64 libbabeltrace-1.5.8-9.fc36.x86_64 libbrotli-1.0.9-7.fc36.x86_64 libcap-2.48-4.fc36.x86_64 libcom_err-1.46.5-2.fc36.x86_64 libcurl-7.82.0-8.fc36.x86_64 libevent-2.1.12-6.fc36.x86_64 libgcc-12.2.1-2.fc36.x86_64 libidn2-2.3.4-1.fc36.x86_64 libnghttp2-1.46.0-2.fc36.x86_64 libpsl-0.21.1-5.fc36.x86_64 libselinux-3.3-4.fc36.x86_64 libssh-0.9.6-4.fc36.x86_64 libunistring-1.0-1.fc36.x86_64 libunwind-1.6.2-2.fc36.x86_64 libuuid-2.38-1.fc36.x86_64 libxcrypt-4.4.28-1.fc36.x86_64 libzstd-1.5.2-2.fc36.x86_64 numactl-libs-2.0.14-5.fc36.x86_64 openldap-2.6.3-1.fc36.x86_64 openssl-libs-3.0.5-1.fc36.x86_64 pcre-8.45-1.fc36.1.x86_64 perl-libs-5.34.1-486.fc36.x86_64 popt-1.18-7.fc36.x86_64 python3-libs-3.10.7-1.fc36.x86_64 slang-2.3.2-11.fc36.x86_64 xz-libs-5.2.5-9.fc36.x86_64 zlib-1.2.11-33.fc36.x86_64
> (gdb) bt
> #0  0x00000000004e742b in __match_glob (str=0x0, pat=0xe44900 "syscalls", ignore_space=ignore_space@...ry=false, case_ins=case_ins@...ry=false) at util/string.c:113
> #1  0x00000000004e7830 in strglobmatch (str=<optimized out>, pat=<optimized out>) at util/string.c:172
> #2  0x00000000004241ea in default_print_event (ps=0x7fffffffd370, pmu_name=<optimized out>, topic=0x0, event_name=0x7fffffffc2d0 "alarmtimer:alarmtimer_cancel", event_alias=0x0, scale_unit=<optimized out>, deprecated=<optimized out>, event_type_desc=<optimized out>,
>     desc=<optimized out>, long_desc=<optimized out>, encoding_desc=<optimized out>, metric_name=<optimized out>, metric_expr=<optimized out>) at builtin-list.c:107
> #3  0x00000000004e3224 in print_tracepoint_events (print_cb=print_cb@...ry=0x7fffffffd350, print_state=0x7fffffffd370) at util/print-events.c:94
> #4  0x0000000000425342 in cmd_list (argc=<optimized out>, argv=<optimized out>) at builtin-list.c:521
> #5  0x00000000004b4579 in run_builtin (p=p@...ry=0xdb0490 <commands+240>, argc=argc@...ry=2, argv=argv@...ry=0x7fffffffdbb0) at perf.c:322
> #6  0x000000000040f506 in handle_internal_command (argv=<optimized out>, argc=<optimized out>) at perf.c:376
> #7  run_argv (argv=<optimized out>, argcp=<optimized out>) at perf.c:420
> #8  main (argc=2, argv=0x7fffffffdbb0) at perf.c:550
> (gdb)
> 
> Trying to fix this now.
> 
> Please always run 'perf test' before and after your patches and before
> sending it upstream.

Running as !root on a different machine I get some other interesting
info:

⬢[acme@...lbox perf]$  perf list syscalls:sys_enter_open*
double free or corruption (fasttop)
Aborted (core dumped)
⬢[acme@...lbox perf]$

That is:

        free(ps.pmu_glob);


at the end of cmd_list().

- Arnaldo

Powered by blists - more mailing lists