As promised, I have an update on the lab machine I left running ksm_refcnt_overflow.sh for a week straight.
The machine was running 4.15.0-116-generic from -proposed:
$ uname -rv 4.15.0-116-generic #117-Ubuntu SMP Fri Aug 28 16:04:22 UTC 2020 $ uptime 04:36:14 up 7 days, 1 min, 1 user, load average: 3.47, 3.14, 2.97
In that time it has created and destroyed 32,950 virtual machines:
$ virsh list Id Name State ---------------------------------------------------- 32945 instance-0 running 32946 instance-1 running 32947 instance-2 running 32948 instance-3 running 32949 instance-4 running
If we look at the current value of the reference counter, it is still set to 1:
$ cat /proc/zero_page_refcount Zero Page Refcount: 0x1 or 1
I checked /var/log/kern.log, /var/log/syslog and journalctl, there are no oops messages, and the KVM subsystem is stable.
I am shutting the lab machine down now, as I am convinced the patch is stable. This SRU is still verified.
As promised, I have an update on the lab machine I left running ksm_refcnt_ overflow. sh for a week straight.
The machine was running 4.15.0-116-generic from -proposed:
$ uname -rv
4.15.0-116-generic #117-Ubuntu SMP Fri Aug 28 16:04:22 UTC 2020
$ uptime
04:36:14 up 7 days, 1 min, 1 user, load average: 3.47, 3.14, 2.97
In that time it has created and destroyed 32,950 virtual machines:
$ virsh list ------- ------- ------- ------- ------- ------- ---
Id Name State
-------
32945 instance-0 running
32946 instance-1 running
32947 instance-2 running
32948 instance-3 running
32949 instance-4 running
If we look at the current value of the reference counter, it is still set to 1:
$ cat /proc/zero_ page_refcount
Zero Page Refcount: 0x1 or 1
I checked /var/log/kern.log, /var/log/syslog and journalctl, there are no oops messages, and the KVM subsystem is stable.
I am shutting the lab machine down now, as I am convinced the patch is stable. This SRU is still verified.