On Tue, Jul 21, 2009 at 02:09:38AM -0400, Laine Stump wrote:
These patches correspond to patches 2-6 in the set I sent yesterday (the 1st and 2nd had no changes, so they stand as-is). They incorporate all changes brought up in review of the first set. In particular:
1) When looking for a cached virInterface object for an interface, I now check that the mac address being searched for matches the one in the virInterface (case insensitive comparison). If it doesn't match, I create a new object rather than trying to modify the existing object with the same interface name, as that could have led to "improper behavior".
2) in virInterfaceFindByMACString() I now follow danpb's suggestion and use continue, rather than getting fancy with booleans.
3) testinterfaceCreate() and testInterfaceDestroy() now check if the interface has already been created/destroyed, and won't do the same operation twice in a row.
4) in the netcf backend driver, I now "roundtrip" the xml data in both interfaceDefineXML() and interfaceGetXMLDesc().
Okay I have applied everything, congrats ! I hope I didn't forgot anything ... I also applied my old patch to extend .syms, and had to cleanup a couple of things like superflous lines at end of module. One thing we we may have to revisit is how netcf NETCF_ENOMEM is handled, currently we just convert it to VIR_ERR_NO_MEMORY (which gave some syntax-check fun), but we should raise a proper virReportOOMError() instead, code around src/interface_driver.c line 72 may need some tweaking. But with the test driver commited in make check now works properly :-) Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel@veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/