https://bugzilla.novell.com/show_bug.cgi?id=386714 User yxu@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c2 --- Comment #2 from Yi Xu 2008-05-05 07:57:34 MST --- Created an attachment (id=212333) --> (https://bugzilla.novell.com/attachment.cgi?id=212333) output of autotest-cpu_hotplug, kernel oops included -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User gregkh@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c3 Greg Kroah-Hartman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO Info Provider| |yxu@novell.com --- Comment #3 from Greg Kroah-Hartman 2008-05-05 11:42:03 MST --- The first too messages are not "oopses", they are kernel debug messages showing an internal problem. The last one is a problem, probably caused by the first two debug messages showing a logic error :( Is this being run on hardware that supports CPU hotplug? Does it work properly on older kernel releases? (10.3, SLE10 SP2, etc.) Where can we get the source of this program? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User yxu@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c5 --- Comment #5 from Yi Xu 2008-05-06 03:43:37 MST --- Created an attachment (id=212592) --> (https://bugzilla.novell.com/attachment.cgi?id=212592) source of autotest-cpu_hotplug -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User yxu@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c7 --- Comment #7 from Yi Xu 2008-05-07 02:47:28 MST --- My consideration came from this page: http://www.mjmwired.net/kernel/Documentation/cpu-hotplug.txt 134 Q: What architectures support CPU hotplug? 135 A: As of 2.6.14, the following architectures support CPU hotplug. 136 137 i386 (Intel), ppc, ppc64, parisc, s390, ia64 and x86_64 138 139 Q: How to test if hotplug is supported on the newly built kernel? 140 A: You should now notice an entry in sysfs. 141 142 Check if sysfs is mounted, using the "mount" command. You should notice 143 an entry as shown below in the output. 144 145 .... 146 none on /sys type sysfs (rw) 147 .... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 Yi Xu changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW Info Provider|yxu@novell.com | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User yxu@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c10 --- Comment #10 from Yi Xu 2008-05-08 04:17:06 MST --- colfax:~ # cd /sys/devices/system/cpu/ colfax:/sys/devices/system/cpu # ll total 0 drwxr-xr-x 6 root root 0 2008-05-08 11:36 cpu0 drwxr-xr-x 4 root root 0 2008-05-08 11:36 cpu1 drwxr-xr-x 5 root root 0 2008-05-08 11:36 cpu2 drwxr-xr-x 4 root root 0 2008-05-08 11:36 cpu3 drwxr-xr-x 2 root root 0 2008-05-08 11:36 cpuidle -rw-r--r-- 1 root root 4096 2008-05-08 12:12 sched_mc_power_savings colfax:/sys/devices/system/cpu # cat sched_mc_power_savings 0 colfax:/sys/devices/system/cpu # ls cpu* cpu0: cache cpufreq cpuidle crash_notes topology cpu1: cache cpufreq crash_notes online topology cpu2: cache cpufreq crash_notes online topology cpu3: cache cpufreq crash_notes online topology cpuidle: current_driver current_governor_ro colfax:/sys/devices/system/cpu # There's "online" file under each cpu directory except cpu0. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User gregkh@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c12 Greg Kroah-Hartman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED QAContact|kernel-maintainers@forge.provo.novell.com |qa@suse.de Resolution| |FIXED --- Comment #12 from Greg Kroah-Hartman 2008-05-20 12:59:44 MST --- This should be fixed now in the next kernel update. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User yxu@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c14 --- Comment #14 from Yi Xu 2008-10-16 07:42:26 MDT --- There wasn't interesting information on the console log. The machine hangs immediately by calling autotest/bin/autotest /usr/lib/autotest/tests/cpu_hotplug/control -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User jeffm@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c16 Jeff Mahoney changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |NEEDINFO Info Provider| |yxu@novell.com --- Comment #16 from Jeff Mahoney 2008-10-17 13:52:17 MDT --- Is the log level up high enough? We really need to see an oops for this. Preferably with an strace or something showing exactly what is causing the failure. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User yxu@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c18 Yi Xu changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|kernel-maintainers@forge.provo.novell.com |bnc-team-screening@forge.provo.novell.com Status|NEEDINFO |NEW Info Provider|yxu@novell.com | --- Comment #18 from Yi Xu 2008-10-20 05:28:55 MDT --- The host freezes, no output to console. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User scott.scriven@hp.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c20 Scott Scriven changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |scott.scriven@hp.com, achiang@hp.com, | |rick.hester@hp.com --- Comment #20 from Scott Scriven 2008-11-20 15:57:54 MST --- I've just encountered several issues related to CPU hotplug and SLES 11b4 on ia64. What I've seen so far: 1) SLES allows offlining CPU0 on ia64, but I don't think the platform supports that. The immediate result is that the clock stops (kernel clock, not realtime clock), so no sleep() calls return. When I try to bring the CPU back online, the system stops responding entirely. I have a console log for this. To reproduce: echo 0 > /sys/devices/system/cpu/cpu0/online (try commands such as "sleep 1") echo 1 > /sys/devices/system/cpu/cpu0/online 2) After about an hour of twiddling CPUs, I get a kernel oops and panic. I have a console log of this, with the log level set to 8. To reproduce: run "cpuoffline.py 5", which I'll attach. 3) While twiddling other CPUs on and off, SLES periodically stops responding to pings, and other services (ssh) sometimes stop responding too. This suggests to me that perhaps processes and/or interrupts aren't migrating away from offlined CPUs. The end result is about 4% packet loss, in batches of about 40 seconds per outage. To reproduce: ping the system from outside while running cpuoffline.py My test hardware so far is a HP rx6600 with Montvale CPUs -- 2 sockets, 2 cores each, threading enabled, so 8 logical CPUs total. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User scott.scriven@hp.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c22 --- Comment #22 from Scott Scriven 2008-11-20 16:08:08 MST --- Created an attachment (id=254121) --> (https://bugzilla.novell.com/attachment.cgi?id=254121) log from kernel oops and panic while offlining CPU 7 While trying to take CPU 7 offline: Unable to handle kernel NULL pointer dereference (address 0000000000000040) kstop7[12161]: Oops 8813272891392 [1] .. This is failure #2 from comment 20. It seems to take less than an hour to trigger. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User scott.scriven@hp.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c24 --- Comment #24 from Scott Scriven 2008-11-21 20:56:11 MST --- For what it's worth, I tried this today on SLES 10SP2, and I don't see any problems there. It won't let me turn off CPU0, and it flips the other CPUs on and off for hours without incident. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User jeffm@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c26 Jeff Mahoney changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO Info Provider| |scott.scriven@hp.com --- Comment #26 from Jeff Mahoney 2008-11-25 09:02:47 MST --- Scott, can you get the rest of the log for comment #7? It's tainted W, which means there was a WARN_ON higher up. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=386714 User scott.scriven@hp.com added comment https://bugzilla.novell.com/show_bug.cgi?id=386714#c28 Scott Scriven changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW Info Provider|scott.scriven@hp.com | --- Comment #28 from Scott Scriven 2008-11-25 16:42:32 MST --- Created an attachment (id=255492) --> (https://bugzilla.novell.com/attachment.cgi?id=255492) oopses.tar.bz2 I haven't gotten the exact same behavior as before, but I did produce two untainted crashes. Both are attached, and include the full output from boot until crash. Also for curiosity, I included some ping logs where an outside system tries to talk to the test box during the test. If it's of interest, I also have snapshots of /proc/interrupts every 2 seconds during the first run (cpu7). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.