On 29/05/18 17:55, Shawn Lin wrote:

[+ Ulf, Adrian, linux-mmc, as linux-mmc and folks are missing in the
recipient list.]

On 2018/5/29 21:52, Christoph Hellwig wrote:

quoted

By completing the request entirely in the driver we can remove the
BLK_EH_HANDLED return value and thus the split responsibility between the
driver and the block layer that has been causing trouble.

[While this keeps existing behavior it seems to mismatch the comment,
  maintainers please chime in!]

The comment means the timeout doesn't happen for the current in-flight
request which command queue is doing, and the request is finished
previously(correct me if I'm wrong). So mmc defers the complete stuff to
blk_timeout_work to help. Maybe the comment should gone as well.

Reviewed-by: Shawn Lin <shawn.lin@rock-chips.com>

quoted

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
---
  drivers/mmc/core/queue.c | 5 +++--
  1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/mmc/core/queue.c b/drivers/mmc/core/queue.c
index 56e9a803db21..648eb6743ed5 100644
--- a/drivers/mmc/core/queue.c
+++ b/drivers/mmc/core/queue.c

@@ -111,8 +111,9 @@ static enum blk_eh_timer_return

mmc_cqe_timed_out(struct request *req)
                  __mmc_cqe_recovery_notifier(mq);
              return BLK_EH_RESET_TIMER;
          }
-        /* No timeout */
-        return BLK_EH_HANDLED;
+        /* No timeout (XXX: huh? comment doesn't make much sense) */
+        blk_mq_complete_request(req);

It seems the block layer timeout handler no longer castrates the completion
path, in which case blk_mq_complete_request() is not needed here.

The "No timeout" comment can be removed.

quoted

+        return BLK_EH_DONE;
      default:
          /* Timeout is handled by mmc core */
          return BLK_EH_RESET_TIMER;