After f6e7d48 (block: remove BLK_EH_HANDLED), the low-level device driver
is responsible to complete the timed out request and a series of changes
were submitted for various LLDDs to make completions from ->timeout
subsequently. However, adding the completion code in NVMe driver was
skipped with the assumption made in below NVMe LLDD's change.
Commit message of db8c48e (nvme: return BLK_EH_DONE from ->timeout):
NVMe always completes the request before returning from ->timeout, either
by polling for it, or by disabling the controller. Return BLK_EH_DONE so
that the block layer doesn't even try to complete it again.
This does not hold at least for NVMe RDMA host driver. An example scenario
is when the RDMA connection is gone while the controller is being deleted.
In this case, the nvmf_reg_write32() for sending shutdown admin command by
the delete_work could be hung forever if the command is not completed by
the timeout handler.
Stack trace of hang looks like:
kworker/u66:2 D 0 21098 2 0x80000080
Workqueue: nvme-delete-wq nvme_delete_ctrl_work
Call Trace:
__schedule+0x2ab/0x880
schedule+0x36/0x80
schedule_timeout+0x161/0x300
? __next_timer_interrupt+0xe0/0xe0
io_schedule_timeout+0x1e/0x50
wait_for_completion_io_timeout+0x130/0x1a0
? wake_up_q+0x80/0x80
blk_execute_rq+0x6e/0xa0
__nvme_submit_sync_cmd+0x6e/0xe0
nvmf_reg_write32+0x6c/0xc0 [nvme_fabrics]
nvme_shutdown_ctrl+0x56/0x110
nvme_rdma_shutdown_ctrl+0xf8/0x100 [nvme_rdma]
nvme_rdma_delete_ctrl+0x1a/0x20 [nvme_rdma]
nvme_delete_ctrl_work+0x66/0x90
process_one_work+0x179/0x390
worker_thread+0x1da/0x3e0
kthread+0x105/0x140
? max_active_store+0x80/0x80
? kthread_bind+0x20/0x20
ret_from_fork+0x35/0x40
Signed-off-by: Jaesoo Lee <ja...@purestorage.com>
---
drivers/nvme/host/rdma.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c
index d181caf..25319b7 100644
--- a/drivers/nvme/host/rdma.c
+++ b/drivers/nvme/host/rdma.c
@@ -1688,6 +1688,7 @@ static int nvme_rdma_cm_handler(struct rdma_cm_id *cm_id,
/* fail with DNR on cmd timeout */
nvme_req(rq)->status = NVME_SC_ABORT_REQ | NVME_SC_DNR;
+ blk_mq_complete_request(rq);
return BLK_EH_DONE;
}
--
1.9.1
