Fixed
Created: Jul 26, 2017
Updated: Dec 3, 2018
Resolved Date: Aug 24, 2017
Found In Version: 9.0.0.8
Fix Version: 9.0.0.10
Severity: Standard
Applicable for: Wind River Linux 9
Component/s: Userspace
Various userspace programs as bash, watch, stress, md5sum, systemd-udevd, etc are crashing with "unhandled signal 4" and "unhandled signal 11" errors during CPU hotplug:
cpuloader-md5su[15047]: unhandled signal 4 at 0000000010052ea4 nip 0000000010052ea4 lr 0000000010052ea4 code 30001
./cpuloader-md5sum-proc.sh: line 11: 15047 Illegal instruction (core dumped) md5sum /proc/interrupts > /dev/null
cpuloader-md5su[19586]: unhandled signal 11 at 00003ffe03534cd0 nip 0000000010073320 lr 0000000010063d88 code 30001
./cpuloader-md5sum-proc.sh: line 11: 19586 Segmentation fault (core dumped) md5sum /proc/interrupts > /dev/null
*** Error in `/bin/bash': malloc(): memory corruption (fast): 0x0000000010182010 ***
systemd-udevd[28140]: unhandled signal 11 at 000001b601800004 nip 000001b601800004 lr 000000002b78a70c code 30001
watch[12007]: unhandled signal 11 at 0000000000000000 nip 0000000000000000 lr 0000000000000000 code 30001
stress: FAIL: [24092] (522) memory corruption at: 0x3fff8bb28010
stress: FAIL: [24052] (394) <-- worker 24092 returned error 1
stress: WARN: [24052] (396) now reaping child worker processes
stress: FAIL: [24052] (451) failed run completed in 95598s
1.
taskset 1 ./hs-main.sh &
2.
taskset 1 ./hs-coreoff.sh 22 &
3.
watch -n 0 "cat /sys/devices/system/cpu/cpu*/online"
4.
wait a few minutes/hours/days (some userspace segfaults are hard to reproduce, for others a few minutes will suffice)
====
Step 3. can be replaced with one of the following steps:
3.
./cpuloader-md5sum-proc.sh 24
or
3.
./cpuloader-md5sum.sh 24
or
3.
./stress --cpu 24 --io 24 --vm 24
or
3.
(use a fileystem with systemd-udevd)
