On Wed, Dec 03, 2008 at 10:32:58AM +0100, Daniel Veillard wrote:
On Sun, Nov 30, 2008 at 11:55:51PM +0000, Daniel P. Berrange wrote:
> This patch makes the network driver thread safe, following the pattern of
> the earlier patches
> @@ -95,6 +107,7 @@ networkAutostartConfigs(struct network_d
> unsigned int i;
Hum, you're not locking the driver itself during the loop ?
If a concurrent call were to remove one of the network object in
the meantime, that would lead to potentially random memory accesses
no ?
There are only two callers of this method, and they both already hold the
driver lock - networkStartup() and networkReload().
[...]
> cleanup:
> + if (network)
> + virNetworkObjUnlock(network);
like for domain, I would suggest the unlocking accessor to accept NULL
and drop all the if constructs
I've thought about doing that, but wasn't really sure whether it was good
idea or not. I thought it might hide bugs, but on balance it would certainly
simplify alot of code. If people generally think its worth allowing NULL
in the Unlock routine, i'll make it so.
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 :|