On 01/30/2011 01:10 AM, Osier Yang wrote:
This patch is race with:
http://www.redhat.com/archives/libvir-list/2011-January/msg00681.html
, if patch above works, but it was not pushed yet, will it be
pushed?
I've been kind of sitting on that patch - it's an optimization rather
than a correctness issue, and Paolo correctly pointed out that we can
further optimize by also teaching the outgoing side to pass an fd to the
monitor (since the monitor connection is a Unix socket) rather than the
current exec: used on the outgoing side. Since I have only written the
-incoming side so far, I've deferred pushing until the outgoing side is
also written.
I'm okay if yours goes in first.
Meanwhile, both your patch and my (deferred) patch conflict with
Daniel's refactoring into qemu_process and qemu_migration. Fun stuff,
merge conflicts :)
--
Eric Blake eblake(a)redhat.com +1-801-349-2682
Libvirt virtualization library
http://libvirt.org