]> git.ipfire.org Git - thirdparty/linux.git/commitdiff
vfio: Prevent open_count decrement to negative
authorJacob Pan <jacob.pan@linux.microsoft.com>
Wed, 18 Jun 2025 23:46:18 +0000 (16:46 -0700)
committerAlex Williamson <alex.williamson@redhat.com>
Fri, 11 Jul 2025 20:43:38 +0000 (14:43 -0600)
When vfio_df_close() is called with open_count=0, it triggers a warning in
vfio_assert_device_open() but still decrements open_count to -1. This allows
a subsequent open to incorrectly pass the open_count == 0 check, leading to
unintended behavior, such as setting df->access_granted = true.

For example, running an IOMMUFD compat no-IOMMU device with VFIO tests
(https://github.com/awilliam/tests/blob/master/vfio-noiommu-pci-device-open.c)
results in a warning and a failed VFIO_GROUP_GET_DEVICE_FD ioctl on the first
run, but the second run succeeds incorrectly.

Add checks to avoid decrementing open_count below zero.

Fixes: 05f37e1c03b6 ("vfio: Pass struct vfio_device_file * to vfio_device_open/close()")
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Jacob Pan <jacob.pan@linux.microsoft.com>
Link: https://lore.kernel.org/r/20250618234618.1910456-2-jacob.pan@linux.microsoft.com
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
drivers/vfio/vfio_main.c

index 1fd261efc582d0d95067ef3034f3611cf3c40eea..5046cae052224e25cef601753b4eab9da9571702 100644 (file)
@@ -583,7 +583,8 @@ void vfio_df_close(struct vfio_device_file *df)
 
        lockdep_assert_held(&device->dev_set->lock);
 
-       vfio_assert_device_open(device);
+       if (!vfio_assert_device_open(device))
+               return;
        if (device->open_count == 1)
                vfio_df_device_last_close(df);
        device->open_count--;