KVM: shrink halt polling even more for invalid wakeups
authorChristian Borntraeger <borntraeger@de.ibm.com>
Tue, 17 May 2016 08:49:22 +0000 (10:49 +0200)
committerPaolo Bonzini <pbonzini@redhat.com>
Wed, 18 May 2016 16:04:22 +0000 (18:04 +0200)
commit2086d3200dc9966c96a6c319a1214a94f00223f8
treed894bf017417f0349d8fda96c86dc414aec875e2
parentc4a8de357ef1916c38e4419c0bafad86ca4fbcf7
KVM: shrink halt polling even more for invalid wakeups

commit 3491caf2755e ("KVM: halt_polling: provide a way to qualify
 wakeups during poll") added more aggressive shrinking of the
polling interval if the wakeup did not match some criteria. This
still allows to keep polling enabled if the polling time was
smaller that the current max poll time (block_ns <= vcpu->halt_poll_ns).
Performance measurement shows that even more aggressive shrinking
(shrink polling on any invalid wakeup) reduces absolute and relative
(to the workload) CPU usage even further.

Cc: David Matlack <dmatlack@google.com>
Cc: Wanpeng Li <kernellwp@gmail.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
virt/kvm/kvm_main.c