[Bug 219399] System panics after several hours of 14-threads-compilation orgies using poudriere on AMD Ryzen...

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399

--- Comment #32 from Nils Beyer <nbe at renzel.net> ---
Looks good; if you check via the command "top", you should see CPU usage. And
please check with the command "mount" whether the TMPFS is really mounted to
"/tmp/ryzen_stress_test"; the output should look something like this:
------------------------------------------------------------------------
zroot on / (zfs, local, noatime, nfsv4acls)
devfs on /dev (devfs, local, multilabel)
tmpfs on /tmp/buildworld (tmpfs, local, nosuid, mounted by nbe)
------------------------------------------------------------------------

Last line with "tmpfs" is important.

If the bug is triggered, you will get ".log" files in your home dir. You can
look in to "nohup.out" to check the errorcodes of each loop. Errorcode zero (0)
means all okay, Errorcode two (2) probably means Ryzen "bug", like here:
------------------------------------------------------------------------
umount: /tmp/ryzen_stress_test: not a file system root directory
mkdir: /tmp/ryzen_stress_test: File exists
Mon Jul  3 16:10:32 CEST 2017 begin
Mon Jul  3 16:23:53 CEST 2017 end - errorcode 2
------------------------------------------------------------------------

5MB log file means that the buildworld/buildkernel is not finished yet. Let it
run; it can take several hours to trigger that error. After 48h, if all your
loops end with zero errorcode, then your system really seems stable now...

-- 
You are receiving this mail because:
You are the assignee for the bug.