On 03/11/16 16:08, Michael S. Tsirkin wrote: > From: Gerd Hoffmann <kra...@redhat.com> > > Actually fixes linux not finding virtio 1.0 device virtqueues after > reboot. Which is new I think, any chance linux kernel virtio code > became more strict in 4.3? > > Signed-off-by: Gerd Hoffmann <kra...@redhat.com> > Reviewed-by: Michael S. Tsirkin <m...@redhat.com> > Signed-off-by: Michael S. Tsirkin <m...@redhat.com> > Reviewed-by: Fam Zheng <f...@redhat.com> > Tested-by: Fam Zheng <f...@redhat.com> > --- > hw/virtio/virtio-pci.c | 11 ++++------- > 1 file changed, 4 insertions(+), 7 deletions(-) > > diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c > index 440776c..0dadb66 100644 > --- a/hw/virtio/virtio-pci.c > +++ b/hw/virtio/virtio-pci.c > @@ -47,6 +47,7 @@ > > static void virtio_pci_bus_new(VirtioBusState *bus, size_t bus_size, > VirtIOPCIProxy *dev); > +static void virtio_pci_reset(DeviceState *qdev); > > /* virtio device */ > /* DeviceState to VirtIOPCIProxy. For use off data-path. TODO: use QOM. */ > @@ -404,9 +405,7 @@ static void virtio_ioport_write(void *opaque, uint32_t > addr, uint32_t val) > case VIRTIO_PCI_QUEUE_PFN: > pa = (hwaddr)val << VIRTIO_PCI_QUEUE_ADDR_SHIFT; > if (pa == 0) { > - virtio_pci_stop_ioeventfd(proxy); > - virtio_reset(vdev); > - msix_unuse_all_vectors(&proxy->pci_dev); > + virtio_pci_reset(DEVICE(proxy)); > } > else > virtio_queue_set_addr(vdev, vdev->queue_sel, pa); > @@ -432,8 +431,7 @@ static void virtio_ioport_write(void *opaque, uint32_t > addr, uint32_t val) > } > > if (vdev->status == 0) { > - virtio_reset(vdev); > - msix_unuse_all_vectors(&proxy->pci_dev); > + virtio_pci_reset(DEVICE(proxy)); > } > > /* Linux before 2.6.34 drives the device without enabling > @@ -1353,8 +1351,7 @@ static void virtio_pci_common_write(void *opaque, > hwaddr addr, > } > > if (vdev->status == 0) { > - virtio_reset(vdev); > - msix_unuse_all_vectors(&proxy->pci_dev); > + virtio_pci_reset(DEVICE(proxy)); > } > > break; >
I spent about two days hunting this, and was just about to send a patch. Nice to see you've already gotten to it. :) The specific problem was that the queue_enable field was not cleared (for all possible queues of the device) on device reset, only on system reset. Thanks Laszlo