Messages in this thread | | | Date | Tue, 03 Nov 2009 16:38:46 +0200 | From | Avi Kivity <> | Subject | Re: [PATCH 03/11] Handle asynchronous page fault in a PV guest. |
| |
On 11/03/2009 04:32 PM, Marcelo Tosatti wrote: > Any attempt to access the swapped out data will cause a #PF vmexit, > since the translation is marked as not present. If there's swapin in > progress, you wait for that swapin, otherwise start swapin and wait. > > Its not as efficient as paravirt because you have to wait for a timer > interrupt and the guest scheduler to decide to taskswitch, but OTOH its > transparent. >
With a dyntick guest the timer interrupt will come at the end of the time slice, likely after the page has been swapped in. That leaves smp reschedule interrupts and non-dyntick guests.
An advantage is that there is one code path for apf and non-apf. Another is that interrupts are processed, improving timekeeping and maybe responsiveness.
-- error compiling committee.c: too many arguments to function
| |