Re: process wakeups

["Daniel P. Berrange" <berrange redhat com>]

On Tue, Jul 15, 2008 at 10:22:09AM -0700, Ulrich Drepper wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Daniel P. Berrange wrote:
> >>   uid | poll select  epoll itimer  futex nanosle signal| process
> >>  2397 |   16      0      0      0      0       0      0| libvirtd
> > 
> > This isn't libvirtd's fault - we don't use any timeouts, and only
> > break out of poll() if there's real work todo.
> 
> The script counts returns from poll() where no file descriptor is ready.
>  This is no count of all poll() calls.

Sorry I wasn't being clear. The call to poll() with a 0 second timeout
is an artifact of the way the DBus/Avahi/event loop integration works.
It gets a wakeup on a file descriptor, processes some messages and then
does one iteration through the main loop with a 0 second timeout to
flush out the queues, before finally going back to sleep for events on
its file descriptors. As such the poll() with timeout is harmless - the
process was already awake at this point. Its the wakeup from irrelevant
DBus signals that's the real problem that needs to be addressed - once
that's solved the poll() will timeout won't be done.

Daniel
-- 
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|