On Fri, Apr 24, 2020 at 03:25:39PM +0900, Chirantan Ekbote wrote:
> virtiofs device implementations are allowed to provide more than one
> request queue. In this case `fsvq->fud` would not be initialized,
> leading to a nullptr dereference later during driver initialization.
>
> Make sure that `fsvq->fud` is initialized for all request queues even if
> the driver doesn't use them.
>
> Signed-off-by: Chirantan Ekbote <chirantan@xxxxxxxxxxxx>
> ---
> fs/fuse/virtio_fs.c | 7 +++++--
> 1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/fs/fuse/virtio_fs.c b/fs/fuse/virtio_fs.c
> index bade747689033..d3c38222a7e4e 100644
> --- a/fs/fuse/virtio_fs.c
> +++ b/fs/fuse/virtio_fs.c
> @@ -1066,10 +1066,13 @@ static int virtio_fs_fill_super(struct super_block *sb)
> }
>
> err = -ENOMEM;
> - /* Allocate fuse_dev for hiprio and notification queues */
> - for (i = 0; i < VQ_REQUEST; i++) {
> + /* Allocate fuse_dev for all queues except the first request queue. */
> + for (i = 0; i < fs->nvqs; i++) {
> struct virtio_fs_vq *fsvq = &fs->vqs[i];
>
> + if (i == VQ_REQUEST)
> + continue;
> +
These special conditions of initializing fuse device for one queue
fusing fill_super_common() and rest of the queues outside of it, are
bothering me. I am proposing a separate patch where all fuse device
initialization/cleanup is done by the caller. It makes code look
cleaner and easier to understand.
Vivek
> fsvq->fud = fuse_dev_alloc();
> if (!fsvq->fud)
> goto err_free_fuse_devs;
> --
> 2.26.2.303.gf8c07b1a785-goog
>
