On 10/25/2021 10:58 AM, Michael S. Tsirkin wrote:
> virtblk_setup_cmd returns blk_status_t in an int, callers then assign it
> back to a blk_status_t variable. blk_status_t is either u32 or (more
> typically) u8 so it works, but is inelegant and causes sparse warnings.
>
> Pass the status in blk_status_t in a consistent way.
>
> Reported-by: kernel test robot <[email protected]>
> Fixes: b2c5221fd074 ("virtio-blk: avoid preallocating big SGL for data")
> Cc: Max Gurtovoy <[email protected]>
> Signed-off-by: Michael S. Tsirkin <[email protected]>
> ---
> drivers/block/virtio_blk.c | 14 ++++++++------
> 1 file changed, 8 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
> index c336d9bb9105..c7d05ff24084 100644
> --- a/drivers/block/virtio_blk.c
> +++ b/drivers/block/virtio_blk.c
> @@ -208,8 +208,9 @@ static void virtblk_cleanup_cmd(struct request *req)
> kfree(bvec_virt(&req->special_vec));
> }
>
> -static int virtblk_setup_cmd(struct virtio_device *vdev, struct request *req,
> - struct virtblk_req *vbr)
> +static blk_status_t virtblk_setup_cmd(struct virtio_device *vdev,
> + struct request *req,
> + struct virtblk_req *vbr)
> {
> bool unmap = false;
> u32 type;
> @@ -317,14 +318,15 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
> unsigned long flags;
> unsigned int num;
> int qid = hctx->queue_num;
> - int err;
> bool notify = false;
> + blk_status_t status;
> + int err;
>
> BUG_ON(req->nr_phys_segments + 2 > vblk->sg_elems);
>
> - err = virtblk_setup_cmd(vblk->vdev, req, vbr);
> - if (unlikely(err))
> - return err;
> + status = virtblk_setup_cmd(vblk->vdev, req, vbr);
> + if (unlikely(status))
> + return status;
>
Maybe we can compare the returned status to BLK_STS_OK. But I see we
don't do it also in NVMe subsystem so I guess we can assume BLK_STS_OK
== 0 forever.
Jens/Christoph, any thoughts on that ?
Anyway Looks good.
Thanks for the fix,
Reviewed-by: Max Gurtovoy <[email protected]>
> blk_mq_start_request(req);
>
On Mon, Oct 25, 2021 at 11:24:57AM +0300, Max Gurtovoy wrote:
> Maybe we can compare the returned status to BLK_STS_OK. But I see we don't
> do it also in NVMe subsystem so I guess we can assume BLK_STS_OK == 0
> forever.
Jes, BLK_STS_OK == 0 is an intentional allowed short cut. It is not
just a block layer design, but part of how the sparse __bitwise__
annotations work.