View Issue Details

IDProjectCategoryView StatusLast Update
0014961CentOS-7kernelpublic2018-06-17 16:07
Reporterhiboma 
PrioritynormalSeverityminorReproducibilityalways
Status confirmedResolutionopen 
PlatformVirtualBoxOSCentOSOS Version7.5.1804
Product Version7.5.1804 
Target VersionFixed in Version 
Summary0014961: perf probe / perf record cause segmentation fault
DescriptionWhen I test `perf probe` and `perf record`, `perf record` cause segmentation fault involving other processes.

```
[root@localhost ~]# perf probe -x /usr/lib64/libc.so.6 malloc
Added new event:
  probe_libc:malloc (on malloc in /usr/lib64/libc-2.17.so)

You can now use it in all perf tools, such as:

        perf record -e probe_libc:malloc -aR sleep 1

[root@localhost ~]# perf record -e probe_libc:malloc -aR sleep 1
perf: Segmentation fault
Segmentation fault
```

dmesg shows logs like following.

```
[ 466.624783] auditd[566]: segfault at 1 ip 0000000000000001 sp 00007ffc56afd6e8 error 14 in auditd[565224d17000+1c000]
[ 466.625022] systemd-journal[431]: segfault at fbad8000 ip 00000000fbad8000 sp 00007ffee27dbba8 error 14 in systemd-journald[55dea8145000+50000]
[ 466.625303] audit: netlink_unicast sending to audit_pid=566 returned error: -111
[ 466.625315] systemd[1]: segfault at 558dce407c40 ip 0000558dce407c40 sp 00007fff8fef7e98 error 15
[ 466.677708] audit: audit_lost=1 audit_rate_limit=0 audit_backlog_limit=8192
[ 466.677710] audit: audit_pid=566 reset
[ 466.677729] type=1701 audit(1529239965.344:78): auid=4294967295 uid=0 gid=0 ses=4294967295 subj=system_u:system_r:init_t:s0 pid=1091 comm="systemd" reason="memory violation" sig=11
```
Steps To Reproduce$ perf probe -x /usr/lib64/libc.so.6 malloc
$ perf record -e probe_libc:malloc -aR sleep 1
Additional InformationI can reproduce this problem with following kernels.

* 3.10.0-862.3.3.el7.x86_64
* 3.10.0-862.3.2.el7.x86_64
* 3.10.0-862.2.3.el7.x86_64

With 3.10.0-862.el7.x86_64, perf record works fine.
TagsNo tags attached.
abrt_hash
URL

Activities

toracat

toracat

2018-06-17 15:35

manager   ~0032103

I was able to reproduce the issue with RHEL 7.5.

The submitter's original article has some more info:

https://hiboma.hatenadiary.jp/entry/2018/06/17/223416 (Japanese 日本語 :)
toracat

toracat

2018-06-17 16:07

manager   ~0032104

I test-installed the latest mainline kernel (ELRepo's kernel-ml-4.17.2-1.el7.elrepo). It does not have this problem.

Issue History

Date Modified Username Field Change
2018-06-17 13:45 hiboma New Issue
2018-06-17 15:35 toracat Status new => confirmed
2018-06-17 15:35 toracat Note Added: 0032103
2018-06-17 16:07 toracat Note Added: 0032104