[Bug 1081431] qemu: NMI watchdog: BUG: soft lockup - CPU#0 stuck for 23s! [modprobe:238]
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1081431 http://bugzilla.suse.com/show_bug.cgi?id=1081431#c7 --- Comment #7 from Matwey Kornilov <matwey.kornilov@gmail.com> --- (In reply to Takashi Iwai from comment #6)
Adding Joerg to Cc.
This might be the same cause we've been seeing in some bug reports with the nested KVM.
Long story in short for the reference. I've spent a lot of time to dig out the root for the issue. Finally, I've accidentally found that 4.16-rc3 host kernel has the issue fixed. From the user space point of view, the way to trigger the issue is to use KVM_IOEVENTFD with len set to 0 (it is valid setting). When len is set to non-zero value - everything works as expected. When it is set to zero, then qemu is receiving notification storm from eventfd, and the guest execution is frozen (only interruptions/timers are executed). -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply@novell.com