Re: [Qemu-devel] [PATCH 2/5] pci: Add interface names to hybrid PCI devices

[Marcel Apfelbaum]

Hi Eduardo,

On 25/08/2017 22:18, Eduardo Habkost wrote:
> On Wed, Aug 23, 2017 at 07:14:42PM -0300, Eduardo Habkost wrote:
> > The following devices support both PCIe and legacy PCI, by
> > including special code to handle the QEMU_PCI_CAP_EXPRESS flag:
> >
> > * vfio-pci (is_express=1, but legacy PCI handled by
> >    vfio_populate_device())
> > * vmxnet3 (is_express=0, but PCIe handled by vmxnet3_realize())
> > * pvscsi (is_express=0, but PCIe handled by pvscsi_realize())
> > * virtio-pci (is_express=0, but PCIe handled by
> >    virtio_pci_dc_realize(), and additional legacy PCI code at
> >    virtio_pci_realize())
>
> Oh, the rules are even messier than that: QEMU_PCI_CAP_EXPRESS
> controls _some_ of the code that makes a device become a PCI
> Express device, but not every case.
>
> In addition to vmxnet3, pvscsi and virtio-pci, PCIe caps
> initialization is conditional on hcd-xhci (see below).
>
> This means xhci is also a hybrid device.  But it doesn't seem to
> clear QEMU_PCI_CAP_EXPRESS.  Doesn't it mean pci_config_size() is
> broken if xhci is plugged to a legacy PCI bus?  How does it
> affect migration?

If this is the case we reserve more config space than needed.
Other than wasted space it should be OK, including migration.

Thanks,
Marcel

> Full list of pci_endpoint_cap*_init() calls:
>
> hw/usb/hcd-xhci.c-    if (pci_bus_is_express(dev->bus) ||
> hw/usb/hcd-xhci.c-        xhci_get_flag(xhci, XHCI_FLAG_FORCE_PCIE_ENDCAP)) {
> hw/usb/hcd-xhci.c:        ret = pcie_endpoint_cap_init(dev, 0xa0);
>
> hw/net/vmxnet3.c-    if (pci_is_express(pci_dev)) {
> hw/net/vmxnet3.c-        if (pci_bus_is_express(pci_dev->bus)) {
> hw/net/vmxnet3.c:            pcie_endpoint_cap_init(pci_dev, 
> VMXNET3_EXP_EP_OFFSET);
> hw/net/vmxnet3.c-        }
>
> hw/scsi/megasas.c-    if (pci_is_express(dev)) {
> hw/scsi/megasas.c:        pcie_endpoint_cap_init(dev, 0xa0);
> hw/scsi/megasas.c-    }
>
> hw/scsi/vmw_pvscsi.c-    if (pci_is_express(pci_dev) && 
> pci_bus_is_express(pci_dev->bus)) {
> hw/scsi/vmw_pvscsi.c:        pcie_endpoint_cap_init(pci_dev, 
> PVSCSI_EXP_EP_OFFSET);
> hw/scsi/vmw_pvscsi.c-    }
>
> hw/virtio/virtio-pci.c-    if (pcie_port && pci_is_express(pci_dev)) {
> hw/virtio/virtio-pci.c-        int pos;
> hw/virtio/virtio-pci.c-
> hw/virtio/virtio-pci.c:        pos = pcie_endpoint_cap_init(pci_dev, 0);
>
>
> Note: on megasas, pci_endpoitn_cap_init() is conditional on
> pci_is_express(), but pci_is_express() will never be false on
> megasas-gen2 (because QEMU_PCI_CAP_EXPRESS is always set).
>
>
> Full list of unconditional pci_endpoint_cap*init() calls:
>
> hw/block/nvme.c:    pcie_endpoint_cap_init(&n->parent_obj, 0x80);
>
> hw/net/e1000e.c:    if (pcie_endpoint_cap_v1_init(pci_dev, e1000e_pcie_offset) 
> < 0) {
>
> hw/pci-bridge/pcie_root_port.c:    rc = pcie_cap_init(d, rpc->exp_offset, 
> PCI_EXP_TYPE_ROOT_PORT,
> hw/pci-bridge/pcie_root_port.c-                       p->port, errp);
>
> hw/pci-bridge/xio3130_downstream.c:    rc = pcie_cap_init(d, 
> XIO3130_EXP_OFFSET, PCI_EXP_TYPE_DOWNSTREAM,
> hw/pci-bridge/xio3130_downstream.c-                       p->port, errp);
>
> hw/pci-bridge/xio3130_upstream.c:    rc = pcie_cap_init(d, XIO3130_EXP_OFFSET, 
> PCI_EXP_TYPE_UPSTREAM,
> hw/pci-bridge/xio3130_upstream.c-                       p->port, errp);
>
> hw/pci-host/xilinx-pcie.c:    if (pcie_endpoint_cap_v1_init(dev, 0x80) < 0) {
>
>
>
> > Signed-off-by: Eduardo Habkost <address@hidden>
> > ---
> >   hw/net/vmxnet3.c       | 5 +++++
> >   hw/scsi/vmw_pvscsi.c   | 2 ++
> >   hw/vfio/pci.c          | 5 +++++
> >   hw/virtio/virtio-pci.c | 5 +++++
> >   4 files changed, 17 insertions(+)
> >
> > diff --git a/hw/net/vmxnet3.c b/hw/net/vmxnet3.c
> > index a19a7a3..61feacf 100644
> > --- a/hw/net/vmxnet3.c
> > +++ b/hw/net/vmxnet3.c
> > @@ -2651,6 +2651,11 @@ static const TypeInfo vmxnet3_info = {
> >       .instance_size = sizeof(VMXNET3State),
> >       .class_init    = vmxnet3_class_init,
> >       .instance_init = vmxnet3_instance_init,
> > +    .interfaces = (InterfaceInfo[]) {
> > +        { INTERFACE_PCIE_DEVICE },
> > +        { INTERFACE_LEGACY_PCI_DEVICE },
> > +        { }
> > +    },
> >   };
> >   
> >   static void vmxnet3_register_types(void)
> > diff --git a/hw/scsi/vmw_pvscsi.c b/hw/scsi/vmw_pvscsi.c
> > index 77d8b6f..40dfffa 100644
> > --- a/hw/scsi/vmw_pvscsi.c
> > +++ b/hw/scsi/vmw_pvscsi.c
> > @@ -1300,6 +1300,8 @@ static const TypeInfo pvscsi_info = {
> >       .class_init    = pvscsi_class_init,
> >       .interfaces = (InterfaceInfo[]) {
> >           { TYPE_HOTPLUG_HANDLER },
> > +        { INTERFACE_PCIE_DEVICE },
> > +        { INTERFACE_LEGACY_PCI_DEVICE },
> >           { }
> >       }
> >   };
> > diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
> > index 31e1edf..2b21391 100644
> > --- a/hw/vfio/pci.c
> > +++ b/hw/vfio/pci.c
> > @@ -3023,6 +3023,11 @@ static const TypeInfo vfio_pci_dev_info = {
> >       .class_init = vfio_pci_dev_class_init,
> >       .instance_init = vfio_instance_init,
> >       .instance_finalize = vfio_instance_finalize,
> > +    .interfaces = (InterfaceInfo[]) {
> > +        { INTERFACE_PCIE_DEVICE },
> > +        { INTERFACE_LEGACY_PCI_DEVICE },
> > +        { }
> > +    },
> >   };
> >   
> >   static void register_vfio_pci_dev_type(void)
> > diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> > index 8b0d6b6..8c0b6bf 100644
> > --- a/hw/virtio/virtio-pci.c
> > +++ b/hw/virtio/virtio-pci.c
> > @@ -1958,6 +1958,11 @@ static const TypeInfo virtio_pci_info = {
> >       .class_init    = virtio_pci_class_init,
> >       .class_size    = sizeof(VirtioPCIClass),
> >       .abstract      = true,
> > +    .interfaces = (InterfaceInfo[]) {
> > +        { INTERFACE_PCIE_DEVICE },
> > +        { INTERFACE_LEGACY_PCI_DEVICE },
> > +        { }
> > +    },
> >   };
> >   
> >   /* virtio-blk-pci */
> > --
> > 2.9.4