[libvirt] [PATCH 0/5] Additional SCSI generated device address checks

[Eric Farman]

On 06/22/2015 05:05 PM, John Ferlan wrote:
> https://bugzilla.redhat.com/show_bug.cgi?id=1210587
>
> These patches will resolve a couple issues with generation of the
> <address type='drive' .../> for a SCSI <disk> and <hostdev>.
>
> The <disk> generation algorithm 'assumes' that when presented with
> <target dev='sda'.../> that it can use controller=0 and unit=0 since
> sda would conceivably be the first device; however, a <hostdev> could
> attempt to assign itself to that address and it doesn't have a target
> device name, so it bypasses the virDomainDiskDefDstDuplicates checks
> that would normally 'catch' two <disk>'s attempting to use the same name.
>
> Likewise, if a <hostdev> occupies an <address> and we attempt to hotplug
> a <disk> without providing an address, the address generation could
> attempt to place the disk on the already existing host device.
>
>
> John Ferlan (5):
>    conf: Enforce SCSI hostdev address type
>    conf: Add 'bus' and 'target' to SCSI address conflict checks
>    conf: Add SCSI hostdev check for disk drive address already in use
>    conf: Refactor virDomainDiskDefParseXML to pass vmdef
>    conf: Check for hostdev conflicts when assign default disk address
>
>   docs/formatdomain.html.in |   4 +-
>   src/conf/domain_conf.c    | 115 ++++++++++++++++++++++++++++++++--------------
>   src/conf/domain_conf.h    |   3 +-
>   src/qemu/qemu_command.c   |   4 +-
>   src/vmx/vmx.c             |  22 +++++----
>   src/vmx/vmx.h             |   3 +-
>   6 files changed, 101 insertions(+), 50 deletions(-)
>

For what it's worth, other than the one doc change suggestion, the rest 
looks (and works) fine to me.

Reviewed-by: Eric Farman <farman at linux.vnet.ibm.com>

  - Eric