Linux Virtualization - Mar 2023 - [PATCH v3 virtio 1/8] virtio: allow caller to override device id and DMA mask

On Thu, Mar 23, 2023 at 3:11?AM Shannon Nelson <shannon.nelson at amd.com>
wrote:
>
> To allow a bit of flexibility with various virtio based devices, allow
> the caller to specify a different device id and DMA mask.  This adds
> fields to struct XXX to specify an override device id check and a DMA mask.
>
> Signed-off-by: Shannon Nelson <shannon.nelson at amd.com>
> ---
>  drivers/virtio/virtio_pci_modern_dev.c | 36 +++++++++++++++++---------
>  include/linux/virtio_pci_modern.h      |  6 +++++
>  2 files changed, 30 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/virtio/virtio_pci_modern_dev.c
b/drivers/virtio/virtio_pci_modern_dev.c
> index 869cb46bef96..6ad1bb9ae8fa 100644
> --- a/drivers/virtio/virtio_pci_modern_dev.c
> +++ b/drivers/virtio/virtio_pci_modern_dev.c
> @@ -221,18 +221,25 @@ int vp_modern_probe(struct virtio_pci_modern_device
*mdev)
>
>         check_offsets();
>
> -       /* We only own devices >= 0x1000 and <= 0x107f: leave the
rest. */
> -       if (pci_dev->device < 0x1000 || pci_dev->device >
0x107f)
> -               return -ENODEV;
> -
> -       if (pci_dev->device < 0x1040) {
> -               /* Transitional devices: use the PCI subsystem device id as
> -                * virtio device id, same as legacy driver always did.
> -                */
> -               mdev->id.device = pci_dev->subsystem_device;
> +       if (mdev->device_id_check_override) {
> +               err = mdev->device_id_check_override(pci_dev);
> +               if (err)
> +                       return err;
> +               mdev->id.device = pci_dev->device;
While at this, would it be better to let the device_id_check_override
to return the mdev->id.device ?

Others look good.

Thanks
>         } else {
> -               /* Modern devices: simply use PCI device id, but start from
0x1040. */
> -               mdev->id.device = pci_dev->device - 0x1040;
> +               /* We only own devices >= 0x1000 and <= 0x107f: leave
the rest. */
> +               if (pci_dev->device < 0x1000 || pci_dev->device
> 0x107f)
> +                       return -ENODEV;
> +
> +               if (pci_dev->device < 0x1040) {
> +                       /* Transitional devices: use the PCI subsystem
device id as
> +                        * virtio device id, same as legacy driver always
did.
> +                        */
> +                       mdev->id.device = pci_dev->subsystem_device;
> +               } else {
> +                       /* Modern devices: simply use PCI device id, but
start from 0x1040. */
> +                       mdev->id.device = pci_dev->device - 0x1040;
> +               }
>         }
>         mdev->id.vendor = pci_dev->subsystem_vendor;
>
> @@ -260,7 +267,12 @@ int vp_modern_probe(struct virtio_pci_modern_device
*mdev)
>                 return -EINVAL;
>         }
>
> -       err = dma_set_mask_and_coherent(&pci_dev->dev,
DMA_BIT_MASK(64));
> +       if (mdev->dma_mask_override)
> +               err = dma_set_mask_and_coherent(&pci_dev->dev,
> +                                              
mdev->dma_mask_override);
> +       else
> +               err = dma_set_mask_and_coherent(&pci_dev->dev,
> +                                               DMA_BIT_MASK(64));
>         if (err)
>                 err = dma_set_mask_and_coherent(&pci_dev->dev,
>                                                 DMA_BIT_MASK(32));
> diff --git a/include/linux/virtio_pci_modern.h
b/include/linux/virtio_pci_modern.h
> index c4eeb79b0139..84765bbd8dc5 100644
> --- a/include/linux/virtio_pci_modern.h
> +++ b/include/linux/virtio_pci_modern.h
> @@ -38,6 +38,12 @@ struct virtio_pci_modern_device {
>         int modern_bars;
>
>         struct virtio_device_id id;
> +
> +       /* alt. check for vendor virtio device, return 0 or -ERRNO */
> +       int (*device_id_check_override)(struct pci_dev *pdev);
> +
> +       /* alt. mask for devices with limited DMA space */
> +       u64 dma_mask_override;
>  };
>
>  /*
> --
> 2.17.1
>