• Type: Improvement
    • Resolution: Unresolved
    • Priority: Major
    • None
    • Affects Version/s: virtio 1.0 csprd02
    • None
    • Environment:

      "Michael S. Tsirkin" <[email protected]>

    • Hide

      Require that devices have revision <= 1.
      Require that drivers not match devices with revision > 1.
      http://mid.gmane.org/[email protected]

      Show
      Require that devices have revision <= 1. Require that drivers not match devices with revision > 1. http://mid.gmane.org/[email protected]
    • Hide

      it turns out our PCI Discovery is even messier than I thought,
      and we will have to either stick to using the subsystem ID or
      allocate a new ID range with cleaner allocation rules.

      So after all, the interface version that we reserved wasn't helpful even
      for 0.x -> 1.x and even for linux.
      Self-NACK this patch, and I'll close the issue.

      Show
      it turns out our PCI Discovery is even messier than I thought, and we will have to either stick to using the subsystem ID or allocate a new ID range with cleaner allocation rules. So after all, the interface version that we reserved wasn't helpful even for 0.x -> 1.x and even for linux. Self-NACK this patch, and I'll close the issue.

      If the proposed change to VIRTIO-102 is accepted,
      devices can use all 16 bits of the subsystem device id for versioning. This makes it unnecessary to reserve revision ID for this purpose. Let's use it for the interface version again:
      it served us in the 0.X -> 1.X transition, will be handy for 1.X -> 2.X as well.

            Assignee:
            Unassigned
            Reporter:
            Michael Tsirkin
            Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

              Created:
              Updated: