Messages in this thread | | | Subject | Re: [PATCH v3 0/5] KVM: optimize userspace exits with a new ioctl | From | Paolo Bonzini <> | Date | Mon, 7 Sep 2015 14:33:11 +0200 |
| |
On 02/09/2015 12:31, Christian Borntraeger wrote: > As far as I can see this should also work for s390 (when we implement > REQ_EXIT handling) > > To double check my understanding: these improvements come with a changed > userspace that does not use KVM_SET_SIGNAL_MASK and therefore this > conditional is false > if (vcpu->sigset_active) > sigprocmask(SIG_SETMASK, &vcpu->sigset, &sigsaved); > Correct? > > That also means we improve exits that go to userspace but lightweight exits > that stay inside the kernel are not affected.
Yes, both observations are correct.
Paolo
| |