When doing the following test steps, an error was found:
step 1: modprobe virtio_gpu succeeded
# modprobe virtio_gpu <-- OK
step 2: fault injection in virtio_gpu_alloc_vbufs()
# modprobe -r virtio_gpu <-- OK
# ...
CPU: 0 PID: 1714 Comm: modprobe Not tainted 6.1.0-rc7-dirty
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996)
Call Trace:
<TASK>
should_fail_ex.cold+0x1a/0x1f
...
kmem_cache_create+0x12/0x20
virtio_gpu_alloc_vbufs+0x2f/0x90 [virtio_gpu]
virtio_gpu_init.cold+0x659/0xcad [virtio_gpu]
virtio_gpu_probe+0x14f/0x260 [virtio_gpu]
virtio_dev_probe+0x608/0xae0
?...
</TASK>
kmem_cache_create_usercopy(virtio-gpu-vbufs) failed with error -12
step 3: modprobe virtio_gpu failed
# modprobe virtio_gpu <-- failed
failed to find virt queues
virtio_gpu: probe of virtio6 failed with error -2
The root cause of the problem is that the virtqueues are not
stopped on the error handling path when virtio_gpu_alloc_vbufs()
fails in virtio_gpu_init(), resulting in an error "-ENOENT"
returned in the next modprobe call in setup_vq().
virtio_pci_modern_device uses virtqueues to send or
receive message, and "queue_enable" records whether the
queues are available. In vp_modern_find_vqs(), all queues
will be selected and activated, but once queues are enabled
there is no way to go back except reset.
Fix it by reset virtio device on error handling path. After
virtio_find_vqs() succeeded, all virtqueues should be stopped
on error handling path.
Fixes: dc5698e80cf7 ("Add virtio gpu driver.")
Signed-off-by: Li Zetao <lizetao1@xxxxxxxxxx>
---
v1 -> v2: patch is new.
drivers/gpu/drm/virtio/virtgpu_kms.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/gpu/drm/virtio/virtgpu_kms.c b/drivers/gpu/drm/virtio/virtgpu_kms.c
index 27b7f14dae89..1a03e8e13b5b 100644
--- a/drivers/gpu/drm/virtio/virtgpu_kms.c
+++ b/drivers/gpu/drm/virtio/virtgpu_kms.c
@@ -255,6 +255,7 @@ int virtio_gpu_init(struct virtio_device *vdev, struct drm_device *dev)
err_scanouts:
virtio_gpu_free_vbufs(vgdev);
err_vbufs:
+ virtio_reset_device(vgdev->vdev);
vgdev->vdev->config->del_vqs(vgdev->vdev);
err_vqs:
dev->dev_private = NULL;
--
2.25.1
