virtio-pci: Set the QEMU_PCI_CAP_EXPRESS capability early in its DeviceClass realize...
commit0560b0e97df3da43651158c799c6d889f27529c3
authorShmulik Ladkani <shmulik.ladkani@ravellosystems.com>
Wed, 2 Dec 2015 17:49:07 +0000 (2 19:49 +0200)
committerMichael S. Tsirkin <mst@redhat.com>
Wed, 2 Dec 2015 19:51:33 +0000 (2 21:51 +0200)
tree223c7402f4a51de7d2114999a039b76746f593a1
parent11380b36196c483ff5c7f800b0f7af6aa53b5657
virtio-pci: Set the QEMU_PCI_CAP_EXPRESS capability early in its DeviceClass realize method

In 1811e64 'hw/virtio: Add PCIe capability to virtio devices', the
QEMU_PCI_CAP_EXPRESS capability was added to virtio's pci_dev, within
'virtio_pci_realize' - the pci device object realization method.

This occurs to late, as 'pci_qdev_realize' (DeviceClass.realize of
TYPE_PCI_DEVICE) has already been called, without knowing that the
device instance is indeed an "express" instance, thus allocating
insufficient pci config space.

As a result, device may crash upon attempt to write to the PCIE config
space.

Fix, by arming the QEMU_PCI_CAP_EXPRESS capability early in virtio-pci's
own DeviceClass realize method.

This also makes code cleaner, as 'virtio_pci_realize' may now access the
'pci_is_express' predicate when needed.

Signed-off-by: Shmulik Ladkani <shmulik.ladkani@ravellosystems.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Tested-by: Marcel Apfelbaum <marcel@redhat.com>
hw/virtio/virtio-pci.c
hw/virtio/virtio-pci.h