On 31/05/13 22:14, Osier Yang wrote:
On 31/05/13 22:09, Eric Blake wrote:
On 05/31/2013 04:09 AM, Osier Yang wrote:
With unknown good reasons, the attribute "bus" of scsi device address is always set to 0, same for attribute "target". (See virDomainDiskDefAssignAddress).
Though we might need to change the algorithm to honor "bus" and "target" too, that's a different issue. The address generator for scsi host device in this patch just follows the unknown good reasons, only considering the "controller" and "unit". It walks through all scsi controllers and their units, to see if the address $controller:0:0:$unit can be used (if not used by any disk or scsi host device yet), if found one, it sits on it, otherwise, it creates a new controller (actually the controller is implicitly created by someone else), and sits on $new_controller:0:0:0 instead. ACK.
src/conf/domain_conf.c | 202 ++++++++++++++++++--- .../qemuxml2argv-hostdev-scsi-autogen-address.xml | 95 ++++++++++ ...qemuxml2xmlout-hostdev-scsi-autogen-address.xml | 106 +++++++++++ tests/qemuxml2xmltest.c | 2 + 4 files changed, 375 insertions(+), 30 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-hostdev-scsi-autogen-address.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-hostdev-scsi-autogen-address.xml
Relatively big to be pushing this late after freeze, but it can be argued that this is a bug fix and worth including in 1.0.6.
I hope it could be included in 1.0.6. The scsi host device can work without it, but the address should be specified manually, which is not quite good, and better to have them together in one release.
Pushed as-is, John is a bit not fine with the comment, but it's trivial, and I do think it's not the time to change it to produce an inconsistent comment in virDomainDiskDefAssignAddress. As I said, I have to ensure the 7 is used for the controller itself indeed for narrow scsi bus, and it should be a independant patch, as it's for different purpose. Thanks. Osier