Linux perf regression in 4.4.0-42 and newer
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
linux (Ubuntu) |
Fix Released
|
Medium
|
Joseph Salisbury | ||
Xenial |
Fix Released
|
Medium
|
Joseph Salisbury |
Bug Description
There is a regression in the 4.4.0-42 kernel update that is still present in the latest 4.4.0 update. I have verified that this appears in Ubuntu 16.04 and Ubuntu 14.04 (with ubuntu-xenial kernel installed) amd64. I also suspect this happens in default Ubuntu 16.10 kernel as well, but i have not tested.
4.4.0-38 did not have this issue.
in 4.4.0-42 and newer, perf records incorrect timestamps which can be verified by running this command:
$ perf record -a -e sched:sched_-k CLOCK_MONOTONIC switch stress -c 10000 -t 10
stress: info: [25560] dispatching hogs: 10000 cpu, 0 io, 0 vm, 0 hdd
stress: info: [25560] successful run completed in 6s
[ perf record: Woken up 16 times to write data ]
Warning:
78487 out of order events recorded.
Note that the warning about out of order events is a symptom of this problem and is not expected to happen. This does not happen(correct behavior) if i run 4.4.0-38 and older, or if i run a fully patched upstream 4.4.35 kernel instead.
This regression has already been fixed upstream but has not been included in ubuntu-xenial kernel yet.
Upstream fix is here:
https:/
Changed in linux (Ubuntu): | |
status: | Fix Committed → Fix Released |
Changed in linux (Ubuntu Xenial): | |
status: | Fix Committed → Fix Released |
This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:
apport-collect 1645449
and then change the status of the bug to 'Confirmed'.
If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.
This change has been made by an automated script, maintained by the Ubuntu Kernel Team.