Re: [Qemu-devel] [PATCH 2/2] pci: add check for pcie root ports and downstream ports

[Marcel Apfelbaum <marcel.a@redhat.com>]

On Wed, 2014-08-20 at 03:20 +0000, Gonglei (Arei) wrote:
> Hi,
> 
> > > Right now, ARI Forwarding dose not support in QEMU.
> > I would replace the above sentence with "ARI Forwarding is not supported".
> > 
> OK.
> 
> > By the way, there is some support for ARI, I don't know if
> > is enabled yet. I'll have a look.
> > 
> MST had pointed out the pcie_ari_init(), but not completed.
> 
> > > According to PCIe spec section 7.3.1, only slot 0 with
> > > the device attached to logic bus representing the link from
> > > downstream ports and root ports.
> > >
> > > So, adding check about slot 0 for PCIe downstream ports and
> > > root ports, which avoid useless operation, both hotplug and
> > > coldplug.
> > >
> > > Signed-off-by: Gonglei <arei.gonglei@huawei.com>
> > > ---
> > >  hw/pci/pci.c | 41 +++++++++++++++++++++++++++++++++++++++++
> > >  1 file changed, 41 insertions(+)
> > >
> > > diff --git a/hw/pci/pci.c b/hw/pci/pci.c
> > > index 351d320..f2d267f 100644
> > > --- a/hw/pci/pci.c
> > > +++ b/hw/pci/pci.c
> > > @@ -773,6 +773,42 @@ static int pci_init_multifunction(PCIBus *bus,
> > PCIDevice *dev)
> > >      return 0;
> > >  }
> > >
> > > +static int pci_check_pcie_port(PCIBus *bus, PCIDevice *dev)
> > > +{
> > > +    Object *obj = OBJECT(bus);
> > > +
> > > +    if (!strcmp(object_get_typename(obj), TYPE_PCIE_BUS)) {
> > Maybe there is another way to check that this is a PCIe bus?
> > 
> Yes. Paolo has said that object_dynamic_cast() is appropriated.
> 
> > > +        DeviceState *parent = qbus_get_parent(BUS(obj));
> > > +        const char *name = object_get_typename(OBJECT(parent));
> > > +
> > > +        /*
> > > +         * Root ports and downstream ports of switches are the hot
> > > +         * pluggable ports in a PCI Express hierarchy.
> > > +         * PCI Express supports chip-to-chip interconnect, a PCIe link can
> > > +         * only connect one pci device/Switch/EndPoint or PCI-bridge.
> > > +         *
> > > +         * 7.3. Configuration Transaction Rules (PCI Express specification
> > 3.0)
> > > +         * 7.3.1. Device Number
> > > +         *
> > > +         * Downstream Ports that do not have ARI Forwarding enabled
> > must
> > > +         * associate only Device 0 with the device attached to the Logical
> > Bus
> > > +         * representing the Link from the Port.
> > > +         *
> > > +         * Right now, ARI Forwarding dose not support. So, only slot 0 is
> > As above, maybe replace it with "ARI Forwarding is not supported"
> > 
> OK.
> 
> > > +         * supported, regardless of hotplug or coldplug.
> > > +         */
> > > +        if (!strcmp(name, "ioh3420") || !strcmp(name,
> > "xio3130-downstream")) {
> > Please use port_type flag from extended configuration space, don't use device
> > names.
> > If you need help for this, let me know.
> > 
> Yes, please. I appreciate very much that you can help me.
Sure,

I checked and we already have the pcie_cap_get_type function that returns the port type.

    port_type = pcie_cap_get_type(dev);
    if (port_type == PCI_EXP_TYPE_DOWNSTREAM ||
        port_type == PCI_EXP_TYPE_ROOT_PORT) {
        ...
    }

Thanks,
Marcel

> 
> > > +            if (PCI_SLOT(dev->devfn) != 0) {
> > > +                error_report("Unsupported PCI slot %d for %s ports,
> > only "
> > > +                             "supported slot 0",
> > PCI_SLOT(dev->devfn), name);
> > > +                return -1;
> > > +            }
> > > +        }
> > > +    }
> > > +
> > > +    return 0;
> > > +}
> > > +
> > >  static void pci_config_alloc(PCIDevice *pci_dev)
> > >  {
> > >      int config_size = pci_config_size(pci_dev);
> > > @@ -871,6 +907,11 @@ static PCIDevice
> > *do_pci_register_device(PCIDevice *pci_dev, PCIBus *bus,
> > >          return NULL;
> > >      }
> > >
> > > +    if (pci_check_pcie_port(bus, pci_dev)) {
> > > +        do_pci_unregister_device(pci_dev);
> > > +        return NULL;
> > > +    }
> > It is possible to move the above check earlier in do_pci_register_device
> > function?
> > Maybe move it into the first if statement(s)?
> > 
> Agreed, Thanks, Marcel.
> 
> Best regards,
> -Gonglei