2022-10-10 17:47:12

by Michael S. Tsirkin

Subject: [PATCH RFC] pci: fix device presence detection for VFs

virtio uses the same driver for VFs and PFs. Accordingly,
pci_device_is_present is used to detect device presence. This function
isn't currently working properly for VFs since it attempts reading
device and vendor ID. Result is device marked broken incorrectly. As
VFs are present if and only if PF is present, just return the value for
that device.

Reported-by: gongwei <[email protected]>
Signed-off-by: Michael S. Tsirkin <[email protected]>

---

Warning - compile tested only. gongwei could you help test and report
please?

drivers/pci/pci.c | 3 +++
1 file changed, 3 insertions(+)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 95bc329e74c0..ba29b8e2f3c1 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -6462,6 +6462,9 @@ bool pci_device_is_present(struct pci_dev *pdev)
{
u32 v;

+ if (pdev->is_virtfn)
+ return pci_device_is_present(pdev->physfn);
+
if (pci_dev_is_disconnected(pdev))
return false;
return pci_bus_read_dev_vendor_id(pdev->bus, pdev->devfn, &v, 0);
--
MST

2022-10-11 21:07:11

by Michael S. Tsirkin

Subject: Re: [PATCH RFC] pci: fix device presence detection for VFs

On Mon, Oct 10, 2022 at 01:12:20PM -0400, Michael S. Tsirkin wrote:
> virtio uses the same driver for VFs and PFs. Accordingly,
> pci_device_is_present is used to detect device presence. This function
> isn't currently working properly for VFs since it attempts reading
> device and vendor ID. Result is device marked broken incorrectly. As
> VFs are present if and only if PF is present, just return the value for
> that device.
>
> Reported-by: gongwei <[email protected]>
> Signed-off-by: Michael S. Tsirkin <[email protected]>
>
> ---
>
> Warning - compile tested only. gongwei could you help test and report
> please?


Just got a report in private email that it was tested and works well.
Will repost as non-RFC now.


> drivers/pci/pci.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 95bc329e74c0..ba29b8e2f3c1 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -6462,6 +6462,9 @@ bool pci_device_is_present(struct pci_dev *pdev)
> {
> u32 v;
>
> + if (pdev->is_virtfn)
> + return pci_device_is_present(pdev->physfn);
> +
> if (pci_dev_is_disconnected(pdev))
> return false;
> return pci_bus_read_dev_vendor_id(pdev->bus, pdev->devfn, &v, 0);
> --
> MST

2022-10-14 11:49:23

by Wei Gong

Subject: Re: [PATCH RFC] pci: fix device presence detection for VFs

On Mon, Oct 10, 2022 at 01:12:20PM -0400, Michael S. Tsirkin wrote:
> virtio uses the same driver for VFs and PFs. Accordingly,
> pci_device_is_present is used to detect device presence. This function
> isn't currently working properly for VFs since it attempts reading
> device and vendor ID. Result is device marked broken incorrectly. As
> VFs are present if and only if PF is present, just return the value for
> that device.
>
> Reported-by: gongwei <[email protected]>
> Signed-off-by: Michael S. Tsirkin <[email protected]>
Tested-by: Wei Gong <[email protected]>
>
> ---
>
> Warning - compile tested only. gongwei could you help test and report
> please?
I have done the test and it works well
>
> drivers/pci/pci.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 95bc329e74c0..ba29b8e2f3c1 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -6462,6 +6462,9 @@ bool pci_device_is_present(struct pci_dev *pdev)
> {
> u32 v;
>
> + if (pdev->is_virtfn)
> + return pci_device_is_present(pdev->physfn);
> +
> if (pci_dev_is_disconnected(pdev))
> return false;
> return pci_bus_read_dev_vendor_id(pdev->bus, pdev->devfn, &v, 0);
> --
> MST
>

2022-10-26 04:02:46

by Wei Gong

Subject: Re: [PATCH RFC] pci: fix device presence detection for VFs

On Mon, Oct 10, 2022 at 01:12:20PM -0400, Michael S. Tsirkin wrote:
> virtio uses the same driver for VFs and PFs. Accordingly,
> pci_device_is_present is used to detect device presence. This function
> isn't currently working properly for VFs since it attempts reading
> device and vendor ID. Result is device marked broken incorrectly. As
> VFs are present if and only if PF is present, just return the value for
> that device.
>
> Reported-by: gongwei <[email protected]>
> Signed-off-by: Michael S. Tsirkin <[email protected]>

Tested-by: Wei Gong <[email protected]>

>
> ---
>
> Warning - compile tested only. gongwei could you help test and report
> please?

Tested and working well. Any plans for next steps?

>
> drivers/pci/pci.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 95bc329e74c0..ba29b8e2f3c1 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -6462,6 +6462,9 @@ bool pci_device_is_present(struct pci_dev *pdev)
> {
> u32 v;
>
> + if (pdev->is_virtfn)
> + return pci_device_is_present(pdev->physfn);
> +
> if (pci_dev_is_disconnected(pdev))
> return false;
> return pci_bus_read_dev_vendor_id(pdev->bus, pdev->devfn, &v, 0);
> --
> MST
>