cleanup blk_execute_rq* v2

cleanup blk_execute_rq* v2

[Christoph Hellwig]

Hi Jens,

this series cleans up the blk_execute_rq* helpers.  It simplifies the
plugging mess a bit, fixes the sparse __bitwise warnings and simplifies
the blk_execute_rq_nowait API a bit.

Changes since v1:
 - rebased to the current Linus tree with the nvme driver changes
   merged
 - fixed a trailing whitespace and odd formatting
 - fixed a mising semicolon in ufs

[PATCH 1/3] blk-mq: remove __blk_execute_rq_nowait

[Christoph Hellwig]

We don't want to plug for synchronous execution that where we immediately
wait for the request.  Once that is done not a whole lot of code is
shared, so just remove __blk_execute_rq_nowait.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <kbusch@kernel.org>
---
 block/blk-mq.c | 69 ++++++++++++++++++++++----------------------------
 1 file changed, 30 insertions(+), 39 deletions(-)

diff --git a/block/blk-mq.c b/block/blk-mq.c
index ae116b7556482..31a89d1004b8f 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1203,28 +1203,6 @@ static void blk_add_rq_to_plug(struct blk_plug *plug, struct request *rq)
 	plug->rq_count++;
 }
 
-static void __blk_execute_rq_nowait(struct request *rq, bool at_head,
-		rq_end_io_fn *done, bool use_plug)
-{
-	WARN_ON(irqs_disabled());
-	WARN_ON(!blk_rq_is_passthrough(rq));
-
-	rq->end_io = done;
-
-	blk_account_io_start(rq);
-
-	if (use_plug && current->plug) {
-		blk_add_rq_to_plug(current->plug, rq);
-		return;
-	}
-	/*
-	 * don't check dying flag for MQ because the request won't
-	 * be reused after dying flag is set
-	 */
-	blk_mq_sched_insert_request(rq, at_head, true, false);
-}
-
-
 /**
  * blk_execute_rq_nowait - insert a request to I/O scheduler for execution
  * @rq:		request to insert
@@ -1240,8 +1218,16 @@ static void __blk_execute_rq_nowait(struct request *rq, bool at_head,
  */
 void blk_execute_rq_nowait(struct request *rq, bool at_head, rq_end_io_fn *done)
 {
-	__blk_execute_rq_nowait(rq, at_head, done, true);
+	WARN_ON(irqs_disabled());
+	WARN_ON(!blk_rq_is_passthrough(rq));
 
+	rq->end_io = done;
+
+	blk_account_io_start(rq);
+	if (current->plug)
+		blk_add_rq_to_plug(current->plug, rq);
+	else
+		blk_mq_sched_insert_request(rq, at_head, true, false);
 }
 EXPORT_SYMBOL_GPL(blk_execute_rq_nowait);
 
@@ -1277,27 +1263,32 @@ static void blk_rq_poll_completion(struct request *rq, struct completion *wait)
 blk_status_t blk_execute_rq(struct request *rq, bool at_head)
 {
 	DECLARE_COMPLETION_ONSTACK(wait);
-	unsigned long hang_check;
 
-	/*
-	 * iopoll requires request to be submitted to driver, so can't
-	 * use plug
-	 */
+	WARN_ON(irqs_disabled());
+	WARN_ON(!blk_rq_is_passthrough(rq));
+
 	rq->end_io_data = &wait;
-	__blk_execute_rq_nowait(rq, at_head, blk_end_sync_rq,
-			!blk_rq_is_poll(rq));
+	rq->end_io = blk_end_sync_rq;
 
-	/* Prevent hang_check timer from firing at us during very long I/O */
-	hang_check = sysctl_hung_task_timeout_secs;
+	blk_account_io_start(rq);
+	blk_mq_sched_insert_request(rq, at_head, true, false);
 
-	if (blk_rq_is_poll(rq))
+	if (blk_rq_is_poll(rq)) {
 		blk_rq_poll_completion(rq, &wait);
-	else if (hang_check)
-		while (!wait_for_completion_io_timeout(&wait,
-				hang_check * (HZ/2)))
-			;
-	else
-		wait_for_completion_io(&wait);
+	} else {
+		/*
+		 * Prevent hang_check timer from firing at us during very long
+		 * I/O
+		 */
+		unsigned long hang_check = sysctl_hung_task_timeout_secs;
+
+		if (hang_check)
+			while (!wait_for_completion_io_timeout(&wait,
+					hang_check * (HZ/2)))
+				;
+		else
+			wait_for_completion_io(&wait);
+	}
 
 	return (blk_status_t)(uintptr_t)rq->end_io_data;
 }
-- 
2.30.2

[PATCH 2/3] blk-mq: avoid a mess of casts for blk_end_sync_rq

[Christoph Hellwig]

Instead of trying to cast a __bitwise 32-bit integer to a larger integer
and then a pointer, just allow a struct with the blk_status_t and the
completion on stack and set the end_io_data to that.  Use the
opportunity to move the code to where it belongs and drop rather
confusing comments.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <kbusch@kernel.org>
---
 block/blk-mq.c | 43 ++++++++++++++++++++-----------------------
 1 file changed, 20 insertions(+), 23 deletions(-)

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 31a89d1004b8f..28b3e6db98499 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1151,24 +1151,6 @@ void blk_mq_start_request(struct request *rq)
 }
 EXPORT_SYMBOL(blk_mq_start_request);
 
-/**
- * blk_end_sync_rq - executes a completion event on a request
- * @rq: request to complete
- * @error: end I/O status of the request
- */
-static void blk_end_sync_rq(struct request *rq, blk_status_t error)
-{
-	struct completion *waiting = rq->end_io_data;
-
-	rq->end_io_data = (void *)(uintptr_t)error;
-
-	/*
-	 * complete last, if this is a stack request the process (and thus
-	 * the rq pointer) could be invalid right after this complete()
-	 */
-	complete(waiting);
-}
-
 /*
  * Allow 2x BLK_MAX_REQUEST_COUNT requests on plug queue for multiple
  * queues. This is important for md arrays to benefit from merging
@@ -1231,6 +1213,19 @@ void blk_execute_rq_nowait(struct request *rq, bool at_head, rq_end_io_fn *done)
 }
 EXPORT_SYMBOL_GPL(blk_execute_rq_nowait);
 
+struct blk_rq_wait {
+	struct completion done;
+	blk_status_t ret;
+};
+
+static void blk_end_sync_rq(struct request *rq, blk_status_t ret)
+{
+	struct blk_rq_wait *wait = rq->end_io_data;
+
+	wait->ret = ret;
+	complete(&wait->done);
+}
+
 static bool blk_rq_is_poll(struct request *rq)
 {
 	if (!rq->mq_hctx)
@@ -1262,7 +1257,9 @@ static void blk_rq_poll_completion(struct request *rq, struct completion *wait)
  */
 blk_status_t blk_execute_rq(struct request *rq, bool at_head)
 {
-	DECLARE_COMPLETION_ONSTACK(wait);
+	struct blk_rq_wait wait = {
+		.done = COMPLETION_INITIALIZER_ONSTACK(wait.done),
+	};
 
 	WARN_ON(irqs_disabled());
 	WARN_ON(!blk_rq_is_passthrough(rq));
@@ -1274,7 +1271,7 @@ blk_status_t blk_execute_rq(struct request *rq, bool at_head)
 	blk_mq_sched_insert_request(rq, at_head, true, false);
 
 	if (blk_rq_is_poll(rq)) {
-		blk_rq_poll_completion(rq, &wait);
+		blk_rq_poll_completion(rq, &wait.done);
 	} else {
 		/*
 		 * Prevent hang_check timer from firing at us during very long
@@ -1283,14 +1280,14 @@ blk_status_t blk_execute_rq(struct request *rq, bool at_head)
 		unsigned long hang_check = sysctl_hung_task_timeout_secs;
 
 		if (hang_check)
-			while (!wait_for_completion_io_timeout(&wait,
+			while (!wait_for_completion_io_timeout(&wait.done,
 					hang_check * (HZ/2)))
 				;
 		else
-			wait_for_completion_io(&wait);
+			wait_for_completion_io(&wait.done);
 	}
 
-	return (blk_status_t)(uintptr_t)rq->end_io_data;
+	return wait.ret;
 }
 EXPORT_SYMBOL(blk_execute_rq);
 
-- 
2.30.2

[PATCH 3/3] blk-mq: remove the done argument to blk_execute_rq_nowait

[Christoph Hellwig]

Let the caller set it together with the end_io_data instead of passing
a pointless argument.  Note the the target code did in fact already
set it and then just overrode it again by calling blk_execute_rq_nowait.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Kanchan Joshi <joshi.k@samsung.com>
---
 block/blk-mq.c                     |  5 +----
 drivers/block/sx8.c                |  4 ++--
 drivers/nvme/host/core.c           |  3 ++-
 drivers/nvme/host/ioctl.c          |  3 ++-
 drivers/nvme/host/pci.c            | 10 +++++++---
 drivers/nvme/target/passthru.c     |  3 ++-
 drivers/scsi/scsi_error.c          |  5 +++--
 drivers/scsi/sg.c                  |  3 ++-
 drivers/scsi/st.c                  |  3 ++-
 drivers/scsi/ufs/ufshpb.c          |  6 ++++--
 drivers/target/target_core_pscsi.c |  3 +--
 include/linux/blk-mq.h             |  3 +--
 12 files changed, 29 insertions(+), 22 deletions(-)

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 28b3e6db98499..8e7860268f614 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1189,7 +1189,6 @@ static void blk_add_rq_to_plug(struct blk_plug *plug, struct request *rq)
  * blk_execute_rq_nowait - insert a request to I/O scheduler for execution
  * @rq:		request to insert
  * @at_head:    insert request at head or tail of queue
- * @done:	I/O completion handler
  *
  * Description:
  *    Insert a fully prepared request at the back of the I/O scheduler queue
@@ -1198,13 +1197,11 @@ static void blk_add_rq_to_plug(struct blk_plug *plug, struct request *rq)
  * Note:
  *    This function will invoke @done directly if the queue is dead.
  */
-void blk_execute_rq_nowait(struct request *rq, bool at_head, rq_end_io_fn *done)
+void blk_execute_rq_nowait(struct request *rq, bool at_head)
 {
 	WARN_ON(irqs_disabled());
 	WARN_ON(!blk_rq_is_passthrough(rq));
 
-	rq->end_io = done;
-
 	blk_account_io_start(rq);
 	if (current->plug)
 		blk_add_rq_to_plug(current->plug, rq);
diff --git a/drivers/block/sx8.c b/drivers/block/sx8.c
index b361583944b94..63b4f6431d2e6 100644
--- a/drivers/block/sx8.c
+++ b/drivers/block/sx8.c
@@ -540,7 +540,7 @@ static int carm_array_info (struct carm_host *host, unsigned int array_idx)
 	spin_unlock_irq(&host->lock);
 
 	DPRINTK("blk_execute_rq_nowait, tag == %u\n", rq->tag);
-	blk_execute_rq_nowait(rq, true, NULL);
+	blk_execute_rq_nowait(rq, true);
 
 	return 0;
 
@@ -579,7 +579,7 @@ static int carm_send_special (struct carm_host *host, carm_sspc_t func)
 	crq->msg_bucket = (u32) rc;
 
 	DPRINTK("blk_execute_rq_nowait, tag == %u\n", rq->tag);
-	blk_execute_rq_nowait(rq, true, NULL);
+	blk_execute_rq_nowait(rq, true);
 
 	return 0;
 }
diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 72f7c955c7078..727c12cbe3272 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -1206,9 +1206,10 @@ static void nvme_keep_alive_work(struct work_struct *work)
 	nvme_init_request(rq, &ctrl->ka_cmd);
 
 	rq->timeout = ctrl->kato * HZ;
+	rq->end_io = nvme_keep_alive_end_io;
 	rq->end_io_data = ctrl;
 	rq->rq_flags |= RQF_QUIET;
-	blk_execute_rq_nowait(rq, false, nvme_keep_alive_end_io);
+	blk_execute_rq_nowait(rq, false);
 }
 
 static void nvme_start_keep_alive(struct nvme_ctrl *ctrl)
diff --git a/drivers/nvme/host/ioctl.c b/drivers/nvme/host/ioctl.c
index 096b1b47d750e..a2e89db1cd639 100644
--- a/drivers/nvme/host/ioctl.c
+++ b/drivers/nvme/host/ioctl.c
@@ -453,6 +453,7 @@ static int nvme_uring_cmd_io(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
 			blk_flags);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
+	req->end_io = nvme_uring_cmd_end_io;
 	req->end_io_data = ioucmd;
 
 	/* to free bio on completion, as req->bio will be null at that time */
@@ -461,7 +462,7 @@ static int nvme_uring_cmd_io(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
 	pdu->meta_buffer = nvme_to_user_ptr(d.metadata);
 	pdu->meta_len = d.metadata_len;
 
-	blk_execute_rq_nowait(req, 0, nvme_uring_cmd_end_io);
+	blk_execute_rq_nowait(req, false);
 	return -EIOCBQUEUED;
 }
 
diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index 5a98a7de09642..0403b6d10bb48 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -1438,9 +1438,10 @@ static enum blk_eh_timer_return nvme_timeout(struct request *req, bool reserved)
 	}
 	nvme_init_request(abort_req, &cmd);
 
+	abort_req->end_io = abort_endio;
 	abort_req->end_io_data = NULL;
 	abort_req->rq_flags |= RQF_QUIET;
-	blk_execute_rq_nowait(abort_req, false, abort_endio);
+	blk_execute_rq_nowait(abort_req, false);
 
 	/*
 	 * The aborted req will be completed on receiving the abort req.
@@ -2485,12 +2486,15 @@ static int nvme_delete_queue(struct nvme_queue *nvmeq, u8 opcode)
 		return PTR_ERR(req);
 	nvme_init_request(req, &cmd);
 
+	if (opcode == nvme_admin_delete_cq)
+		req->end_io = nvme_del_cq_end;
+	else
+		req->end_io = nvme_del_queue_end;
 	req->end_io_data = nvmeq;
 
 	init_completion(&nvmeq->delete_done);
 	req->rq_flags |= RQF_QUIET;
-	blk_execute_rq_nowait(req, false, opcode == nvme_admin_delete_cq ?
-			nvme_del_cq_end : nvme_del_queue_end);
+	blk_execute_rq_nowait(req, false);
 	return 0;
 }
 
diff --git a/drivers/nvme/target/passthru.c b/drivers/nvme/target/passthru.c
index 5247c24538eba..3cc4d6709c93c 100644
--- a/drivers/nvme/target/passthru.c
+++ b/drivers/nvme/target/passthru.c
@@ -285,8 +285,9 @@ static void nvmet_passthru_execute_cmd(struct nvmet_req *req)
 		req->p.rq = rq;
 		queue_work(nvmet_wq, &req->p.work);
 	} else {
+		rq->end_io = nvmet_passthru_req_done;
 		rq->end_io_data = req;
-		blk_execute_rq_nowait(rq, false, nvmet_passthru_req_done);
+		blk_execute_rq_nowait(rq, false);
 	}
 
 	if (ns)
diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c
index cdaca13ac1f1c..49ef864df5816 100644
--- a/drivers/scsi/scsi_error.c
+++ b/drivers/scsi/scsi_error.c
@@ -2039,12 +2039,13 @@ static void scsi_eh_lock_door(struct scsi_device *sdev)
 	scmd->cmnd[4] = SCSI_REMOVAL_PREVENT;
 	scmd->cmnd[5] = 0;
 	scmd->cmd_len = COMMAND_SIZE(scmd->cmnd[0]);
+	scmd->allowed = 5;
 
 	req->rq_flags |= RQF_QUIET;
 	req->timeout = 10 * HZ;
-	scmd->allowed = 5;
+	req->end_io = eh_lock_door_done;
 
-	blk_execute_rq_nowait(req, true, eh_lock_door_done);
+	blk_execute_rq_nowait(req, true);
 }
 
 /**
diff --git a/drivers/scsi/sg.c b/drivers/scsi/sg.c
index cbffa712b9f3e..118c7b4a8af2c 100644
--- a/drivers/scsi/sg.c
+++ b/drivers/scsi/sg.c
@@ -831,7 +831,8 @@ sg_common_write(Sg_fd * sfp, Sg_request * srp,
 
 	srp->rq->timeout = timeout;
 	kref_get(&sfp->f_ref); /* sg_rq_end_io() does kref_put(). */
-	blk_execute_rq_nowait(srp->rq, at_head, sg_rq_end_io);
+	srp->rq->end_io = sg_rq_end_io;
+	blk_execute_rq_nowait(srp->rq, at_head);
 	return 0;
 }
 
diff --git a/drivers/scsi/st.c b/drivers/scsi/st.c
index 56a093a90b922..850172a2b8f14 100644
--- a/drivers/scsi/st.c
+++ b/drivers/scsi/st.c
@@ -579,9 +579,10 @@ static int st_scsi_execute(struct st_request *SRpnt, const unsigned char *cmd,
 	memcpy(scmd->cmnd, cmd, scmd->cmd_len);
 	req->timeout = timeout;
 	scmd->allowed = retries;
+	req->end_io = st_scsi_execute_end;
 	req->end_io_data = SRpnt;
 
-	blk_execute_rq_nowait(req, true, st_scsi_execute_end);
+	blk_execute_rq_nowait(req, true);
 	return 0;
 }
 
diff --git a/drivers/scsi/ufs/ufshpb.c b/drivers/scsi/ufs/ufshpb.c
index 588c0329b80ca..8a7809b9728df 100644
--- a/drivers/scsi/ufs/ufshpb.c
+++ b/drivers/scsi/ufs/ufshpb.c
@@ -671,11 +671,12 @@ static void ufshpb_execute_umap_req(struct ufshpb_lu *hpb,
 
 	req->timeout = 0;
 	req->end_io_data = umap_req;
+	req->end_io = ufshpb_umap_req_compl_fn;
 
 	ufshpb_set_unmap_cmd(scmd->cmnd, rgn);
 	scmd->cmd_len = HPB_WRITE_BUFFER_CMD_LENGTH;
 
-	blk_execute_rq_nowait(req, true, ufshpb_umap_req_compl_fn);
+	blk_execute_rq_nowait(req, true);
 
 	hpb->stats.umap_req_cnt++;
 }
@@ -707,6 +708,7 @@ static int ufshpb_execute_map_req(struct ufshpb_lu *hpb,
 	blk_rq_append_bio(req, map_req->bio);
 
 	req->end_io_data = map_req;
+	req->end_io = ufshpb_map_req_compl_fn;
 
 	if (unlikely(last))
 		mem_size = hpb->last_srgn_entries * HPB_ENTRY_SIZE;
@@ -716,7 +718,7 @@ static int ufshpb_execute_map_req(struct ufshpb_lu *hpb,
 				map_req->rb.srgn_idx, mem_size);
 	scmd->cmd_len = HPB_READ_BUFFER_CMD_LENGTH;
 
-	blk_execute_rq_nowait(req, true, ufshpb_map_req_compl_fn);
+	blk_execute_rq_nowait(req, true);
 
 	hpb->stats.map_req_cnt++;
 	return 0;
diff --git a/drivers/target/target_core_pscsi.c b/drivers/target/target_core_pscsi.c
index bb3fb18b2316d..e6a967ddc08ce 100644
--- a/drivers/target/target_core_pscsi.c
+++ b/drivers/target/target_core_pscsi.c
@@ -972,8 +972,7 @@ pscsi_execute_cmd(struct se_cmd *cmd)
 
 	cmd->priv = scmd->cmnd;
 
-	blk_execute_rq_nowait(req, cmd->sam_task_attr == TCM_HEAD_TAG,
-			pscsi_req_done);
+	blk_execute_rq_nowait(req, cmd->sam_task_attr == TCM_HEAD_TAG);
 
 	return 0;
 
diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h
index 9f07061418db0..e2d9daf7e8dd0 100644
--- a/include/linux/blk-mq.h
+++ b/include/linux/blk-mq.h
@@ -969,8 +969,7 @@ int blk_rq_unmap_user(struct bio *);
 int blk_rq_map_kern(struct request_queue *, struct request *, void *,
 		unsigned int, gfp_t);
 int blk_rq_append_bio(struct request *rq, struct bio *bio);
-void blk_execute_rq_nowait(struct request *rq, bool at_head,
-		rq_end_io_fn *end_io);
+void blk_execute_rq_nowait(struct request *rq, bool at_head);
 blk_status_t blk_execute_rq(struct request *rq, bool at_head);
 
 struct req_iterator {
-- 
2.30.2

Re: [PATCH 1/3] blk-mq: remove __blk_execute_rq_nowait

[Chaitanya Kulkarni]

On 5/24/22 05:15, Christoph Hellwig wrote:
> We don't want to plug for synchronous execution that where we immediately
> wait for the request.  Once that is done not a whole lot of code is
> shared, so just remove __blk_execute_rq_nowait.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> Reviewed-by: Keith Busch <kbusch@kernel.org>
> ---
>

Looks good.

Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

-ck

Re: [PATCH 2/3] blk-mq: avoid a mess of casts for blk_end_sync_rq

[Chaitanya Kulkarni]

On 5/24/22 05:15, Christoph Hellwig wrote:
> Instead of trying to cast a __bitwise 32-bit integer to a larger integer
> and then a pointer, just allow a struct with the blk_status_t and the
> completion on stack and set the end_io_data to that.  Use the
> opportunity to move the code to where it belongs and drop rather
> confusing comments.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> Reviewed-by: Keith Busch <kbusch@kernel.org>
> ---

Looks good.

Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

-ck

Re: [PATCH 3/3] blk-mq: remove the done argument to blk_execute_rq_nowait

[Chaitanya Kulkarni]

On 5/24/22 05:15, Christoph Hellwig wrote:
> Let the caller set it together with the end_io_data instead of passing
> a pointless argument.  Note the the target code did in fact already
> set it and then just overrode it again by calling blk_execute_rq_nowait.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> Reviewed-by: Keith Busch <kbusch@kernel.org>
> Reviewed-by: Kanchan Joshi <joshi.k@samsung.com>
> ---

Looks good.

Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>

-ck

Re: [PATCH 1/3] blk-mq: remove __blk_execute_rq_nowait

[Jens Axboe]

On Tue, 24 May 2022 14:15:28 +0200, Christoph Hellwig wrote:
> We don't want to plug for synchronous execution that where we immediately
> wait for the request.  Once that is done not a whole lot of code is
> shared, so just remove __blk_execute_rq_nowait.
> 
> 

Applied, thanks!

[1/3] blk-mq: remove __blk_execute_rq_nowait
      (no commit info)
[2/3] blk-mq: avoid a mess of casts for blk_end_sync_rq
      (no commit info)
[3/3] blk-mq: remove the done argument to blk_execute_rq_nowait
      (no commit info)

Best regards,
-- 
Jens Axboe