Dan Smith wrote:
# HG changeset patch # User Dan Smith <danms@us.ibm.com> # Date 1199395653 28800 # Node ID 65a07575d2cfc4dfd3c1d57e0a0828c88dacf540 # Parent 3a838dfd165b5f721687a2bb0b2e03bb0c8192b7 Add preliminary XML-parsing support for Xen FV domains
This adds a type switch for XenPV, XenFV, KVM to the struct domain. It also adds parsing of 'graphics' and 'emulator' devices. Finally, amidst the compulsory changes to xmlgen.c are modifications to generate a proper <os> block for XenFV domains. More work will need to be done here, but this is an example of how the domain->type switch will be used to select different generation routines.
Signed-off-by: Dan Smith <danms@us.ibm.com>
+ +static int get_graphics_device(virDomainPtr dom, struct virt_device **dev) +{ + char *xml; + int ret; + struct virt_device *list = NULL; + + xml = virDomainGetXMLDesc(dom, 0); + if (!xml) + return 0;
I suppose this speaks well for the quality of the patch when this is all I can find to complain about, but isn't that a spot where we typically use "if (xml == NULL)" for the sake of clarity? Using "(!xml)" for the conditional tends to make me think of xml as an int. -- -Jay