[PATCH] vfio/pci: Reject invalid MSI-X Table and PBA BIR values

[PATCH] vfio/pci: Reject invalid MSI-X Table and PBA BIR values

[Cédric Le Goater]

The 3-bit MSI-X BIR fields permit values 0-7, but VFIOPCIDevice::bars[]
only contains BAR0-BAR5 (6 entries). An invalid BIR value of 6 or 7
can cause an out-of-bounds array access (CWE-129) in vfio_msix_early_setup().

Add range checks immediately after extracting the BIR values.

Reported-by: Feifan Qian <bea1e@proton.me>
Fixes: 65501a745dba ("vfio: vfio-pci device assignment driver")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/3878
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
 hw/vfio/pci.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index 67d9d9e846446577e8de7a8c588165c6f38c3bee..324e3f5d98206ccb64963ee232f69d3ada422ef5 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -1762,6 +1762,14 @@ static bool vfio_msix_early_setup(VFIOPCIDevice *vdev, Error **errp)
     msix->pba_offset = pba & ~PCI_MSIX_FLAGS_BIRMASK;
     msix->entries = (ctrl & PCI_MSIX_FLAGS_QSIZE) + 1;
 
+    if (msix->table_bar >= PCI_NUM_REGIONS - 1 ||
+        msix->pba_bar >= PCI_NUM_REGIONS - 1) {
+        error_setg(errp, "invalid MSI-X BIR, table_bar=%d pba_bar=%d",
+                   msix->table_bar, msix->pba_bar);
+        g_free(msix);
+        return false;
+    }
+
     ret = vfio_device_get_irq_info(&vdev->vbasedev, VFIO_PCI_MSIX_IRQ_INDEX,
                                    &irq_info);
     if (ret < 0) {
-- 
2.54.0

Re: [PATCH] vfio/pci: Reject invalid MSI-X Table and PBA BIR values

[Alex Williamson]

On Mon,  6 Jul 2026 16:31:24 +0200
Cédric Le Goater <clg@redhat.com> wrote:

> The 3-bit MSI-X BIR fields permit values 0-7, but VFIOPCIDevice::bars[]
> only contains BAR0-BAR5 (6 entries). An invalid BIR value of 6 or 7
> can cause an out-of-bounds array access (CWE-129) in vfio_msix_early_setup().
> 
> Add range checks immediately after extracting the BIR values.

A compliant PCI device used for traditional device assignment is not
susceptible to this without a malicious intermediary.  The primary
exposure seems to be a vfio-user server, maliciously or inadvertently
exposing bogus BIR values, which appears to introduce a DoS vector
without any reliable further exploit.

> Reported-by: Feifan Qian <bea1e@proton.me>
> Fixes: 65501a745dba ("vfio: vfio-pci device assignment driver")
> Resolves: https://gitlab.com/qemu-project/qemu/-/issues/3878
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
> ---
>  hw/vfio/pci.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
> index 67d9d9e846446577e8de7a8c588165c6f38c3bee..324e3f5d98206ccb64963ee232f69d3ada422ef5 100644
> --- a/hw/vfio/pci.c
> +++ b/hw/vfio/pci.c
> @@ -1762,6 +1762,14 @@ static bool vfio_msix_early_setup(VFIOPCIDevice *vdev, Error **errp)
>      msix->pba_offset = pba & ~PCI_MSIX_FLAGS_BIRMASK;
>      msix->entries = (ctrl & PCI_MSIX_FLAGS_QSIZE) + 1;
>  
> +    if (msix->table_bar >= PCI_NUM_REGIONS - 1 ||
> +        msix->pba_bar >= PCI_NUM_REGIONS - 1) {
> +        error_setg(errp, "invalid MSI-X BIR, table_bar=%d pba_bar=%d",
> +                   msix->table_bar, msix->pba_bar);
> +        g_free(msix);
> +        return false;
> +    }
> +
>      ret = vfio_device_get_irq_info(&vdev->vbasedev, VFIO_PCI_MSIX_IRQ_INDEX,
>                                     &irq_info);
>      if (ret < 0) {

Note that the vulnerability is an out-of-bounds index on vdev->bars, so
we could test >= ARRAY_SIZE(vdev->bars), which is sized as
(PCI_NUM_REGIONS - 1).  Either way...

Reviewed-by: Alex Williamson <alex@shazbot.org>