| Message ID | 20230602121653.80017-17-yi.l.liu@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Add vfio_device cdev for iommufd support | expand |
On Fri, Jun 02, 2023 at 05:16:45AM -0700, Yi Liu wrote: > This avoids endless vfio_device refcount increasement by userspace, > which would keep blocking the vfio_unregister_group_dev(). > > Tested-by: Yanting Jiang <yanting.jiang@intel.com> > Tested-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com> > Tested-by: Terrence Xu <terrence.xu@intel.com> > Signed-off-by: Yi Liu <yi.l.liu@intel.com> > --- > drivers/vfio/vfio_main.c | 8 ++++++-- > 1 file changed, 6 insertions(+), 2 deletions(-) Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> It looks OK, at least I couldn't find a reason why the group list would need to continue to be valid while we are waiting for the registration lock to release. Jason
diff --git a/drivers/vfio/vfio_main.c b/drivers/vfio/vfio_main.c index df4f3e37268d..f00ba7603351 100644 --- a/drivers/vfio/vfio_main.c +++ b/drivers/vfio/vfio_main.c @@ -332,6 +332,12 @@ void vfio_unregister_group_dev(struct vfio_device *device) bool interrupted = false; long rc; + /* + * Prevent new device opened by userspace via the + * VFIO_GROUP_GET_DEVICE_FD in the group path. + */ + vfio_device_group_unregister(device); + vfio_device_put_registration(device); rc = try_wait_for_completion(&device->comp); while (rc <= 0) { @@ -355,8 +361,6 @@ void vfio_unregister_group_dev(struct vfio_device *device) } } - vfio_device_group_unregister(device); - /* Balances device_add in register path */ device_del(&device->device);