search

open-power-host-os / linux

Linux kernel source tree
Other
3 stars 4 forks source link

cpu hotplug-unplug in parallel with suspend resume.. guest becomes unresponsive.. #16

Closed bssrikanth closed 6 years ago

bssrikanth commented 7 years ago

I tried cpu hotplug-unplug in parallel with suspend resume.. guest becomes unresponsive.. i. On host run terminal 1: for i in {1..100};do sleep 2;virsh suspend srikanth_1710_Cdrom;sleep 2;virsh resume srikanth_1710_Cdrom;done terminal 2: for i in {1..100};do sleep 5;virsh setvcpus srikanth_1710_Cdrom 4 --live;sleep 5;virsh setvcpus srikanth_1710_Cdrom 2 --live;done

ii. After/while both above commands complete, will start seeing below messages on guest console:

[ 1209.368538] INFO: task jbd2/dm-0-8:1088 blocked for more than 120 seconds. [ 1209.368633] Not tainted 4.12.0-11-generic #12-Ubuntu [ 1209.368694] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.

And the guest goes unresponsive..

I could'nt run sosreport on guest since it was unresponsive..

I tested same scenario as mentioned in previous comment with latest hostos guest.. I am hitting similar stall issues there as well... I am attaching the traces of hostos here.. Guest is hung.. so could not get sosreports...

So these stall issues present on Ubuntu 17.10 [kernel: 4.13.0-11] and HostOS guest [kernel: 4.13.0-4.rel ]

In an internal comment I will post machine access details and guest details..

Mirrored with LTC bug #159338
bssrikanth commented 7 years ago

Stall_hostosguest.txt

bssrikanth commented 7 years ago

@Sam has fix for this issue, patch @ http://www.spinics.net/lists/kvm-ppc/msg12981.html .. This bug is track verification of the issue when HostOS gets the above patch

cdeadmin commented 6 years ago

------- Comment From bssrikanth@in.ibm.com 2017-12-04 04:54:15 EDT------- Tested on 4.14 rel branch.. issue seems to be fixed from the tests conducted. Closing.