Re: [PATCH] hw/pci: prevent hotplug of devices on pcie-root-ports with wrong devfn address

[Ani Sinha]

> On 15-Jun-2023, at 2:22 AM, Michael S. Tsirkin <mst@redhat.com> wrote:
> 
> On Wed, Jun 14, 2023 at 10:31:40PM +0530, Ani Sinha wrote:
>> 
>> 
>>> On 14-Jun-2023, at 6:31 PM, Igor Mammedov <imammedo@redhat.com> wrote:
>>> 
>>> On Wed, 14 Jun 2023 18:01:50 +0530
>>> Ani Sinha <anisinha@redhat.com> wrote:
>>> 
>>>> PCIE root ports only allow one device on slot 0/function 0. When 
>>>> hotplugging a
>>>> device on a pcie root port, make sure that the device address passed is
>>>> always 0x00 that represents slot 0 and function 0. Any other slot value and
>>>> function value would be illegal on a root port.
>>>> 
>>>> CC: jusual@redhat.com
>>>> CC: imammedo@redhat.com
>>>> Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2128929
>>>> Signed-off-by: Ani Sinha <anisinha@redhat.com>
>>>> ---
>>>> hw/pci/pci.c | 9 +++++++++
>>>> 1 file changed, 9 insertions(+)
>>>> 
>>>> Note:
>>>> I tested this with both native and acpi hotplug enabled on pcie. The
>>>> check seems to work on both.
>>>> 
>>>> (qemu)  netdev_add socket,id=hostnet1,listen=:1234
>>>> (qemu) device_add 
>>>> e1000e,netdev=hostnet1,mac=00:11:22:33:44:03,id=net1,bus=pci.6,addr=0x2.0x5
>>>> Error: PCI: slot 2 function 5 is not valid for e1000e
>>>> (qemu) device_add 
>>>> e1000e,netdev=hostnet1,mac=00:11:22:33:44:03,id=net1,bus=pci.6,addr=0x0.0
>>>> (qemu) info network
>>>> net1: index=0,type=nic,model=e1000e,macaddr=00:11:22:33:44:03
>>>> \ hostnet1: index=0,type=socket,
>>>> (qemu) device_del net1
>>>> (qemu) info network
>>>> hostnet1: index=0,type=socket,
>>>> 
>>>> diff --git a/hw/pci/pci.c b/hw/pci/pci.c
>>>> index bf38905b7d..bc39322dfd 100644
>>>> --- a/hw/pci/pci.c
>>>> +++ b/hw/pci/pci.c
>>>> @@ -64,6 +64,7 @@ bool pci_available = true;
>>>> static char *pcibus_get_dev_path(DeviceState *dev);
>>>> static char *pcibus_get_fw_dev_path(DeviceState *dev);
>>>> static void pcibus_reset(BusState *qbus);
>>>> +static bool pcie_has_upstream_port(PCIDevice *dev);
>>>> 
>>>> static Property pci_props[] = {
>>>>    DEFINE_PROP_PCI_DEVFN("addr", PCIDevice, devfn, -1),
>>>> @@ -1189,6 +1190,14 @@ static PCIDevice *do_pci_register_device(PCIDevice 
>>>> *pci_dev,
>>>>                   name);
>>>> 
>>>>       return NULL;
>>>> +    } else if (dev->hotplugged &&
>>> 
>>>> +               !pci_is_vf(pci_dev) &&
>>>> +               pcie_has_upstream_port(pci_dev) 
>>> can you clarify what above 2 lines are here for?
>>> 
>>> you can't hotplug anything into upstream port (only downstream)
>> 
>> This code checks if the hot plugged device is being plugged into a pcie root 
>> port. For pcie root ports, only slot 0 can be used. Please see 
>> 3f1e1478db2d67098d9 . The conditional check just above this hunk therefore 
>> checks for unoccupied function 0 in slot 0 if the device is plugged into an 
>> upstream pcie root port. I think for hot plugs, only function 0 is allowed.
> 
> Not really. The way we hotplug a multifunction device is by first
> populating all functions != 0, then finally pupulating function 0
> which is when the event triggers.

Let me explain better ..

This check currently already exists:
    } else if (dev->hotplugged &&
               !pci_is_vf(pci_dev) &&
               pci_get_function_0(pci_dev)) {
        error_setg(errp, "PCI: slot %d function 0 already occupied by %s,"
                   " new func %s cannot be exposed to guest.",
                   PCI_SLOT(pci_get_function_0(pci_dev)->devfn),
                   pci_get_function_0(pci_dev)->name,
                   name);

       return NULL;
    }

This code checks for unoccupied function 0. If you look into 
pci_get_function_0(), it checks for slot 0 function 0 for devices that are 
plugged into pcie root port.

> 
>> 
>> As for vf, I am not sure. I believe for vfs the restriction is not valid. I 
>> just kept the check based on the conditional just prior to it.
>> 
>>> and why _vf() is here also not clear.
>>> 
>>>> && devfn) {
>>>> +        /* For an upstream PCIE port, we only support one device at slot 
>>>> 0 */
>>>> +        error_setg(errp, "PCI: slot %d function %d is not valid for %s",
>>>> +                   PCI_SLOT(devfn), PCI_FUNC(devfn), name);
>>>> +        return NULL;
>>>> +
>>>>    }
>>>> 
>>>>    pci_dev->devfn = devfn;