Il 09/01/2013 21:40, Anthony Liguori ha scritto: >> > Yes, but this can only be done after looking at all PCI devices that >> > have buses below (of which the bridge is a special case). And Anthony >> > also metioned using a new method of PCIDevice instead of device_reset. > Sorry, what's the bug here? Is this a reset bug with cold reset or with > a form of warm reset?
Warm reset (virtio status register reset) of virtio-scsi wasn't propagated down to the SCSI bus. I fixed it using qdev_reset_all, but now I'll do it manually in the HBA. I'm fine with that as long as all buses move away from device_reset. Paolo
