On 08.06.2020 13:21, Paul Durrant wrote: >> From: Jan Beulich <jbeul...@suse.com> >> Sent: 08 June 2020 12:10 >> >> As said in the other thread, I think the change here will mask >> a problem elsewhere (presumably in qemu). I'm therefore unsure whether >> we want to apply it right away, rather than first understanding the >> root cause of Marek's specific problem. >> > > I think it ought to be applied right away since an emulator could die at any > time and we don't really want the vcpu constantly bouncing in this case. > Also, thinking of using emulators other than qemu, it's not necessarily a bug > if they deregister at an arbitrary time...
Oh, sure, I fully agree we need the change regardless of the possible qemu misbehavior. But if we commit this change right away, will there be anyone still caring about / looking into the qemu side issue? Jan
