Re: [PATCH v1 kvmtool 6/7] vfio/pci: Print an error when offset is outside of the MSIX table or PBA

[Andre Przywara <andre.przywara@arm.com>]

On Mon, 13 Sep 2021 16:44:12 +0100
Alexandru Elisei <alexandru.elisei@arm.com> wrote:

> Now that we keep track of the real size of MSIX table and PBA, print an
> error when the guest tries to write to an offset which is not inside the
> correct regions.

What is the actual purpose of this message? Is there anything the user can
do about it? Shouldn't we either abort the guest or inject an error
condition back into KVM or the guest instead?

Cheers,
Andre

> 
> Signed-off-by: Alexandru Elisei <alexandru.elisei@arm.com>
> ---
>  vfio/pci.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/vfio/pci.c b/vfio/pci.c
> index 7781868..a6d0408 100644
> --- a/vfio/pci.c
> +++ b/vfio/pci.c
> @@ -249,6 +249,11 @@ static void vfio_pci_msix_pba_access(struct kvm_cpu
> *vcpu, u64 addr, u8 *data, u64 offset = addr - pba->guest_phys_addr;
>  	struct vfio_device *vdev = container_of(pdev, struct
> vfio_device, pci); 
> +	if (offset >= pba->size) {
> +		vfio_dev_err(vdev, "access outside of the MSIX PBA");
> +		return;
> +	}
> +
>  	if (is_write)
>  		return;
>  
> @@ -269,6 +274,10 @@ static void vfio_pci_msix_table_access(struct
> kvm_cpu *vcpu, u64 addr, u8 *data, struct vfio_device *vdev =
> container_of(pdev, struct vfio_device, pci); 
>  	u64 offset = addr - pdev->msix_table.guest_phys_addr;
> +	if (offset >= pdev->msix_table.size) {
> +		vfio_dev_err(vdev, "access outside of the MSI-X table");
> +		return;
> +	}
>  
>  	size_t vector = offset / PCI_MSIX_ENTRY_SIZE;
>  	off_t field = offset % PCI_MSIX_ENTRY_SIZE;