Re: perf stat crashes with latest download (git clone -b perf/core)

[Arnaldo Carvalho de Melo <acme@kernel.org>]

Em Thu, Nov 09, 2017 at 03:23:43PM +0100, Thomas-Mich Richter escreveu:
> I have downloaded this code some days ago
> 
>     git clone -b perf/core git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux acme
> 
> and run into this issue which I reproduce on my KVM virtual machine on intel and s390x.
> 
> [root@localhost perf]# ./perf stat  -e '{cpu-clock,instructions}' kill
> kill: not enough arguments
> Segmentation fault (core dumped)
> [root@localhost perf]# 
> 
> When I debug this further using gdb I get this error:

Trying with today's acme/perf/core:

Yeah,

[acme@jouet linuxdev-br]$ perf stat  -e '{cpu-clock,instructions}' kill
kill: not enough arguments
Segmentation fault (core dumped)
[acme@jouet linuxdev-br]$

Trying to bisect it now, thanks for the report!
 
> root@localhost perf]# gdb perf core.4496 
> GNU gdb (GDB) Fedora 8.0.1-26.fc26
> Copyright (C) 2017 Free Software Foundation, Inc.
> ....
> Core was generated by `./perf stat -e {cpu-clock,instructions} kill'.
> Program terminated with signal SIGSEGV, Segmentation fault.
> #0  perf_evsel__read_group (thread=0, cpu=2, leader=0x3bc8110) at util/evsel.c:1377
> 1377		u64 *data = ps->group_data;
> 
> This goes back to commit f7794d525447f1e4e4b2228dd29dba084005e6bf ("perf evsel: Add read_counter()")
> 
> (gdb) where
> #0  perf_evsel__read_group (thread=0, cpu=2, leader=0x3bc8110) at util/evsel.c:1377
> #1  perf_evsel__read_counter (evsel=evsel@entry=0x3bc8110, cpu=cpu@entry=0, thread=thread@entry=0) at util/evsel.c:1407
> #2  0x0000000000442aad in read_counter (counter=0x3bc8110) at builtin-stat.c:360
> #3  read_counters () at builtin-stat.c:395
> #4  0x000000000044735a in __run_perf_stat (argc=argc@entry=1, argv=argv@entry=0x7ffdc05145f0) at builtin-stat.c:786
> #5  0x0000000000448a60 in run_perf_stat (argv=0x7ffdc05145f0, argc=1) at builtin-stat.c:805
> #6  cmd_stat (argc=1, argv=0x7ffdc05145f0) at builtin-stat.c:2838
> #7  0x00000000004a004f in run_builtin (p=p@entry=0xa34ae0 <commands+288>, argc=argc@entry=4, argv=argv@entry=0x7ffdc05145f0)
>     at perf.c:297
> #8  0x00000000004a0351 in handle_internal_command (argc=argc@entry=4, argv=argv@entry=0x7ffdc05145f0) at perf.c:349
> #9  0x000000000042a21a in run_argv (argcp=<synthetic pointer>, argv=<synthetic pointer>) at perf.c:393
> #10 main (argc=<optimized out>, argv=0x7ffdc05145f0) at perf.c:537
> (gdb) list
> 1372	perf_evsel__read_group(struct perf_evsel *leader, int cpu, int thread)
> 1373	{
> 1374		struct perf_stat_evsel *ps = leader->priv;
> 1375		u64 read_format = leader->attr.read_format;
> 1376		int size = perf_evsel__read_size(leader);
> 1377		u64 *data = ps->group_data;
> 1378	
> 1379		if (!(read_format & PERF_FORMAT_ID))
> 1380			return -EINVAL;
> 1381	
> (gdb) p *leader
> $1 = {node = {next = 0x3bc8320, prev = 0x3bb75c0}, evlist = 0x3bb75c0, attr = {type = 1, size = 112, config = 0, {
>       sample_period = 0, sample_freq = 0}, sample_type = 65536, read_format = 15, disabled = 1, inherit = 1, pinned = 0, 
>     exclusive = 0, exclude_user = 0, exclude_kernel = 0, exclude_hv = 0, exclude_idle = 0, mmap = 0, comm = 0, freq = 0, 
>     inherit_stat = 0, enable_on_exec = 1, task = 0, watermark = 0, precise_ip = 0, mmap_data = 0, sample_id_all = 0, 
>     exclude_host = 0, exclude_guest = 1, exclude_callchain_kernel = 0, exclude_callchain_user = 0, mmap2 = 0, comm_exec = 0, 
>     use_clockid = 0, context_switch = 0, write_backward = 0, namespaces = 0, __reserved_1 = 0, {wakeup_events = 0, 
>       wakeup_watermark = 0}, bp_type = 0, {bp_addr = 0, config1 = 0}, {bp_len = 0, config2 = 0}, branch_sample_type = 0, 
>     sample_regs_user = 0, sample_stack_user = 0, clockid = 0, sample_regs_intr = 0, aux_watermark = 0, sample_max_stack = 0, 
>     __reserved_2 = 0}, filter = 0x0, fd = 0x3bc02f0, sample_id = 0x3bc4360, id = 0x3bc43d0, counts = 0x3bc0100, 
>   prev_raw_counts = 0x0, idx = 0, ids = 1, name = 0x3bc82c0 "cpu-clock", scale = 1, unit = 0x7b8643 "", tp_format = 0x0, 
>   id_offset = 0, stats = 0x3bc5530, priv = 0x0, db_id = 0, cgrp = 0x0, handler = 0x0, cpus = 0x3bc5510, own_cpus = 0x0, 
>   threads = 0x3bc54f0, sample_size = 0, id_pos = -1, is_pos = -1, snapshot = false, supported = true, needs_swap = false, 
>   no_aux_samples = false, immediate = false, system_wide = false, tracking = true, per_pkg = false, precise_max = false, 
>   ignore_missing_thread = false, exclude_GH = 0, nr_members = 2, sample_read = 0, per_pkg_mask = 0x0, leader = 0x3bc8110, 
>   group_name = 0x0, cmdline_group_boundary = false, config_terms = {next = 0x3bc8280, prev = 0x3bc8280}, bpf_fd = -1, 
>   auto_merge_stats = false, merged_stat = false, metric_expr = 0x0, metric_name = 0x0, metric_events = 0x0, collect_stat = false, 
>   weak_group = false}
> (
> (gdb) p leader->priv
> $2 = (void *) 0x0
> (gdb)
> 
> 
> The problem is the value of leader->priv which is 0x0 and line 1377 actually does leader->priv->group_data
> which references a NULL pointer.
> 
> -- 
> Thomas Richter, Dept 3303, IBM LTC Boeblingen Germany
> --
> Vorsitzende des Aufsichtsrats: Martina Koederitz 
> Geschäftsführung: Dirk Wittkopp
> Sitz der Gesellschaft: Böblingen / Registergericht: Amtsgericht Stuttgart, HRB 243294
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-perf-users" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html