libsas: remove task_collector mode

The task_collector mode (or "latency_injector", (C) Dan Willians) is an
optional I/O path in libsas that queues up scsi commands instead of
directly sending it to the hardware.  It generall increases latencies
to in the optiomal case slightly reduce mmio traffic to the hardware.

Only the obsolete aic94xx driver and the mvsas driver allowed to use
it without recompiling the kernel, and most drivers didn't support it
at all.

Remove the giant blob of code to allow better optimizations for scsi-mq
in the future.

Signed-off-by: Christoph Hellwig <[email protected]>
Reviewed-by: Hannes Reinecke <[email protected]>
Acked-by: Dan Williams <[email protected]>

Author: Christoph Hellwig

Documentation/scsi/libsas.txt

@@ -226,9 +226,6 @@ static int register_sas_ha(struct my_sas_ha *my_ha)
 	my_ha->sas_ha.lldd_dev_found = my_dev_found;
 	my_ha->sas_ha.lldd_dev_gone = my_dev_gone;
 
-	my_ha->sas_ha.lldd_max_execute_num = lldd_max_execute_num; (1)
-
-	my_ha->sas_ha.lldd_queue_size = ha_can_queue;
 	my_ha->sas_ha.lldd_execute_task = my_execute_task;
 
 	my_ha->sas_ha.lldd_abort_task     = my_abort_task;
@@ -247,28 +244,6 @@ static int register_sas_ha(struct my_sas_ha *my_ha)
 	return sas_register_ha(&my_ha->sas_ha);
 }
 
-(1) This is normally a LLDD parameter, something of the
-lines of a task collector.  What it tells the SAS Layer is
-whether the SAS layer should run in Direct Mode (default:
-value 0 or 1) or Task Collector Mode (value greater than 1).
-
-In Direct Mode, the SAS Layer calls Execute Task as soon as
-it has a command to send to the SDS, _and_ this is a single
-command, i.e. not linked.
-
-Some hardware (e.g. aic94xx) has the capability to DMA more
-than one task at a time (interrupt) from host memory.  Task
-Collector Mode is an optional feature for HAs which support
-this in their hardware.  (Again, it is completely optional
-even if your hardware supports it.)
-
-In Task Collector Mode, the SAS Layer would do _natural_
-coalescing of tasks and at the appropriate moment it would
-call your driver to DMA more than one task in a single HA
-interrupt. DMBS may want to use this by insmod/modprobe
-setting the lldd_max_execute_num to something greater than
-1.
-
 (2) SAS 1.1 does not define I_T Nexus Reset TMF.
 
 Events
@@ -325,71 +300,22 @@ PHYE_SPINUP_HOLD -- SATA is present, COMWAKE not sent.
 
 The Execute Command SCSI RPC:
 
-	int (*lldd_execute_task)(struct sas_task *, int num,
-				 unsigned long gfp_flags);
+	int (*lldd_execute_task)(struct sas_task *, gfp_t gfp_flags);
 
-Used to queue a task to the SAS LLDD.  @task is the tasks to
-be executed.  @num should be the number of tasks being
-queued at this function call (they are linked listed via
-task::list), @gfp_mask should be the gfp_mask defining the
-context of the caller.
+Used to queue a task to the SAS LLDD.  @task is the task to be executed.
+@gfp_mask is the gfp_mask defining the context of the caller.
 
 This function should implement the Execute Command SCSI RPC,
-or if you're sending a SCSI Task as linked commands, you
-should also use this function.
 
-That is, when lldd_execute_task() is called, the command(s)
+That is, when lldd_execute_task() is called, the command
 go out on the transport *immediately*.  There is *no*
 queuing of any sort and at any level in a SAS LLDD.
 
-The use of task::list is two-fold, one for linked commands,
-the other discussed below.
-
-It is possible to queue up more than one task at a time, by
-initializing the list element of struct sas_task, and
-passing the number of tasks enlisted in this manner in num.
-
 Returns: -SAS_QUEUE_FULL, -ENOMEM, nothing was queued;
 	 0, the task(s) were queued.
 
-If you want to pass num > 1, then either
-A) you're the only caller of this function and keep track
-   of what you've queued to the LLDD, or
-B) you know what you're doing and have a strategy of
-   retrying.
-
-As opposed to queuing one task at a time (function call),
-batch queuing of tasks, by having num > 1, greatly
-simplifies LLDD code, sequencer code, and _hardware design_,
-and has some performance advantages in certain situations
-(DBMS).
-
-The LLDD advertises if it can take more than one command at
-a time at lldd_execute_task(), by setting the
-lldd_max_execute_num parameter (controlled by "collector"
-module parameter in aic94xx SAS LLDD).
-
-You should leave this to the default 1, unless you know what
-you're doing.
-
-This is a function of the LLDD, to which the SAS layer can
-cater to.
-
-int lldd_queue_size
-	The host adapter's queue size.  This is the maximum
-number of commands the lldd can have pending to domain
-devices on behalf of all upper layers submitting through
-lldd_execute_task().
-
-You really want to set this to something (much) larger than
-1.
-
-This _really_ has absolutely nothing to do with queuing.
-There is no queuing in SAS LLDDs.
-
 struct sas_task {
 	dev -- the device this task is destined to
-	list -- must be initialized (INIT_LIST_HEAD)
 	task_proto -- _one_ of enum sas_proto
 	scatter -- pointer to scatter gather list array
 	num_scatter -- number of elements in scatter

drivers/scsi/aic94xx/aic94xx.h

@@ -78,7 +78,7 @@ void asd_dev_gone(struct domain_device *dev);
 
 void asd_invalidate_edb(struct asd_ascb *ascb, int edb_id);
 
-int  asd_execute_task(struct sas_task *, int num, gfp_t gfp_flags);
+int  asd_execute_task(struct sas_task *task, gfp_t gfp_flags);
 
 void asd_set_dmamode(struct domain_device *dev);
 

drivers/scsi/aic94xx/aic94xx_hwi.c

@@ -1200,8 +1200,7 @@ static void asd_start_scb_timers(struct list_head *list)
  * Case A: we can send the whole batch at once.  Increment "pending"
  * in the beginning of this function, when it is checked, in order to
  * eliminate races when this function is called by multiple processes.
- * Case B: should never happen if the managing layer considers
- * lldd_queue_size.
+ * Case B: should never happen.
  */
 int asd_post_ascb_list(struct asd_ha_struct *asd_ha, struct asd_ascb *ascb,
 		       int num)

drivers/scsi/aic94xx/aic94xx_init.c

@@ -49,14 +49,6 @@ MODULE_PARM_DESC(use_msi, "\n"
 	"\tEnable(1) or disable(0) using PCI MSI.\n"
 	"\tDefault: 0");
 
-static int lldd_max_execute_num = 0;
-module_param_named(collector, lldd_max_execute_num, int, S_IRUGO);
-MODULE_PARM_DESC(collector, "\n"
-	"\tIf greater than one, tells the SAS Layer to run in Task Collector\n"
-	"\tMode.  If 1 or 0, tells the SAS Layer to run in Direct Mode.\n"
-	"\tThe aic94xx SAS LLDD supports both modes.\n"
-	"\tDefault: 0 (Direct Mode).\n");
-
 static struct scsi_transport_template *aic94xx_transport_template;
 static int asd_scan_finished(struct Scsi_Host *, unsigned long);
 static void asd_scan_start(struct Scsi_Host *);
@@ -711,9 +703,6 @@ static int asd_register_sas_ha(struct asd_ha_struct *asd_ha)
 	asd_ha->sas_ha.sas_port= sas_ports;
 	asd_ha->sas_ha.num_phys= ASD_MAX_PHYS;
 
-	asd_ha->sas_ha.lldd_queue_size = asd_ha->seq.can_queue;
-	asd_ha->sas_ha.lldd_max_execute_num = lldd_max_execute_num;
-
 	return sas_register_ha(&asd_ha->sas_ha);
 }
 

drivers/scsi/aic94xx/aic94xx_task.c

@@ -543,8 +543,7 @@ static int asd_can_queue(struct asd_ha_struct *asd_ha, int num)
 	return res;
 }
 
-int asd_execute_task(struct sas_task *task, const int num,
-		     gfp_t gfp_flags)
+int asd_execute_task(struct sas_task *task, gfp_t gfp_flags)
 {
 	int res = 0;
 	LIST_HEAD(alist);
@@ -553,11 +552,11 @@ int asd_execute_task(struct sas_task *task, const int num,
 	struct asd_ha_struct *asd_ha = task->dev->port->ha->lldd_ha;
 	unsigned long flags;
 
-	res = asd_can_queue(asd_ha, num);
+	res = asd_can_queue(asd_ha, 1);
 	if (res)
 		return res;
 
-	res = num;
+	res = 1;
 	ascb = asd_ascb_alloc_list(asd_ha, &res, gfp_flags);
 	if (res) {
 		res = -ENOMEM;
@@ -568,7 +567,7 @@ int asd_execute_task(struct sas_task *task, const int num,
 	list_for_each_entry(a, &alist, list) {
 		a->uldd_task = t;
 		t->lldd_task = a;
-		t = list_entry(t->list.next, struct sas_task, list);
+		break;
 	}
 	list_for_each_entry(a, &alist, list) {
 		t = a->uldd_task;
@@ -601,7 +600,7 @@ int asd_execute_task(struct sas_task *task, const int num,
 	}
 	list_del_init(&alist);
 
-	res = asd_post_ascb_list(asd_ha, ascb, num);
+	res = asd_post_ascb_list(asd_ha, ascb, 1);
 	if (unlikely(res)) {
 		a = NULL;
 		__list_add(&alist, ascb->list.prev, &ascb->list);
@@ -639,6 +638,6 @@ int asd_execute_task(struct sas_task *task, const int num,
 out_err:
 	if (ascb)
 		asd_ascb_free_list(ascb);
-	asd_can_dequeue(asd_ha, num);
+	asd_can_dequeue(asd_ha, 1);
 	return res;
 }

drivers/scsi/isci/init.c

@@ -260,8 +260,6 @@ static int isci_register_sas_ha(struct isci_host *isci_host)
 	sas_ha->sas_port = sas_ports;
 	sas_ha->num_phys = SCI_MAX_PHYS;
 
-	sas_ha->lldd_queue_size = ISCI_CAN_QUEUE_VAL;
-	sas_ha->lldd_max_execute_num = 1;
 	sas_ha->strict_wide_ports = 1;
 
 	sas_register_ha(sas_ha);

drivers/scsi/isci/task.c

@@ -117,104 +117,97 @@ static inline int isci_device_io_ready(struct isci_remote_device *idev,
  *    functions. This function is called by libsas to send a task down to
  *    hardware.
  * @task: This parameter specifies the SAS task to send.
- * @num: This parameter specifies the number of tasks to queue.
  * @gfp_flags: This parameter specifies the context of this call.
  *
  * status, zero indicates success.
  */
-int isci_task_execute_task(struct sas_task *task, int num, gfp_t gfp_flags)
+int isci_task_execute_task(struct sas_task *task, gfp_t gfp_flags)
 {
 	struct isci_host *ihost = dev_to_ihost(task->dev);
 	struct isci_remote_device *idev;
 	unsigned long flags;
+	enum sci_status status = SCI_FAILURE;
 	bool io_ready;
 	u16 tag;
 
-	dev_dbg(&ihost->pdev->dev, "%s: num=%d\n", __func__, num);
+	spin_lock_irqsave(&ihost->scic_lock, flags);
+	idev = isci_lookup_device(task->dev);
+	io_ready = isci_device_io_ready(idev, task);
+	tag = isci_alloc_tag(ihost);
+	spin_unlock_irqrestore(&ihost->scic_lock, flags);
 
-	for_each_sas_task(num, task) {
-		enum sci_status status = SCI_FAILURE;
+	dev_dbg(&ihost->pdev->dev,
+		"task: %p, dev: %p idev: %p:%#lx cmd = %p\n",
+		task, task->dev, idev, idev ? idev->flags : 0,
+		task->uldd_task);
 
-		spin_lock_irqsave(&ihost->scic_lock, flags);
-		idev = isci_lookup_device(task->dev);
-		io_ready = isci_device_io_ready(idev, task);
-		tag = isci_alloc_tag(ihost);
-		spin_unlock_irqrestore(&ihost->scic_lock, flags);
+	if (!idev) {
+		isci_task_refuse(ihost, task, SAS_TASK_UNDELIVERED,
+				 SAS_DEVICE_UNKNOWN);
+	} else if (!io_ready || tag == SCI_CONTROLLER_INVALID_IO_TAG) {
+		/* Indicate QUEUE_FULL so that the scsi midlayer
+		 * retries.
+		  */
+		isci_task_refuse(ihost, task, SAS_TASK_COMPLETE,
+				 SAS_QUEUE_FULL);
+	} else {
+		/* There is a device and it's ready for I/O. */
+		spin_lock_irqsave(&task->task_state_lock, flags);
 
-		dev_dbg(&ihost->pdev->dev,
-			"task: %p, num: %d dev: %p idev: %p:%#lx cmd = %p\n",
-			task, num, task->dev, idev, idev ? idev->flags : 0,
-			task->uldd_task);
-
-		if (!idev) {
-			isci_task_refuse(ihost, task, SAS_TASK_UNDELIVERED,
-					 SAS_DEVICE_UNKNOWN);
-		} else if (!io_ready || tag == SCI_CONTROLLER_INVALID_IO_TAG) {
-			/* Indicate QUEUE_FULL so that the scsi midlayer
-			 * retries.
-			  */
-			isci_task_refuse(ihost, task, SAS_TASK_COMPLETE,
-					 SAS_QUEUE_FULL);
+		if (task->task_state_flags & SAS_TASK_STATE_ABORTED) {
+			/* The I/O was aborted. */
+			spin_unlock_irqrestore(&task->task_state_lock, flags);
+
+			isci_task_refuse(ihost, task,
+					 SAS_TASK_UNDELIVERED,
+					 SAM_STAT_TASK_ABORTED);
 		} else {
-			/* There is a device and it's ready for I/O. */
-			spin_lock_irqsave(&task->task_state_lock, flags);
-
-			if (task->task_state_flags & SAS_TASK_STATE_ABORTED) {
-				/* The I/O was aborted. */
-				spin_unlock_irqrestore(&task->task_state_lock,
-						       flags);
-
-				isci_task_refuse(ihost, task,
-						 SAS_TASK_UNDELIVERED,
-						 SAM_STAT_TASK_ABORTED);
-			} else {
-				task->task_state_flags |= SAS_TASK_AT_INITIATOR;
+			task->task_state_flags |= SAS_TASK_AT_INITIATOR;
+			spin_unlock_irqrestore(&task->task_state_lock, flags);
+
+			/* build and send the request. */
+			status = isci_request_execute(ihost, idev, task, tag);
+
+			if (status != SCI_SUCCESS) {
+				spin_lock_irqsave(&task->task_state_lock, flags);
+				/* Did not really start this command. */
+				task->task_state_flags &= ~SAS_TASK_AT_INITIATOR;
 				spin_unlock_irqrestore(&task->task_state_lock, flags);
 
-				/* build and send the request. */
-				status = isci_request_execute(ihost, idev, task, tag);
-
-				if (status != SCI_SUCCESS) {
-
-					spin_lock_irqsave(&task->task_state_lock, flags);
-					/* Did not really start this command. */
-					task->task_state_flags &= ~SAS_TASK_AT_INITIATOR;
-					spin_unlock_irqrestore(&task->task_state_lock, flags);
-
-					if (test_bit(IDEV_GONE, &idev->flags)) {
-
-						/* Indicate that the device
-						 * is gone.
-						 */
-						isci_task_refuse(ihost, task,
-							SAS_TASK_UNDELIVERED,
-							SAS_DEVICE_UNKNOWN);
-					} else {
-						/* Indicate QUEUE_FULL so that
-						 * the scsi midlayer retries.
-						 * If the request failed for
-						 * remote device reasons, it
-						 * gets returned as
-						 * SAS_TASK_UNDELIVERED next
-						 * time through.
-						 */
-						isci_task_refuse(ihost, task,
-							SAS_TASK_COMPLETE,
-							SAS_QUEUE_FULL);
-					}
+				if (test_bit(IDEV_GONE, &idev->flags)) {
+					/* Indicate that the device
+					 * is gone.
+					 */
+					isci_task_refuse(ihost, task,
+						SAS_TASK_UNDELIVERED,
+						SAS_DEVICE_UNKNOWN);
+				} else {
+					/* Indicate QUEUE_FULL so that
+					 * the scsi midlayer retries.
+					 * If the request failed for
+					 * remote device reasons, it
+					 * gets returned as
+					 * SAS_TASK_UNDELIVERED next
+					 * time through.
+					 */
+					isci_task_refuse(ihost, task,
+						SAS_TASK_COMPLETE,
+						SAS_QUEUE_FULL);
 				}
 			}
 		}
-		if (status != SCI_SUCCESS && tag != SCI_CONTROLLER_INVALID_IO_TAG) {
-			spin_lock_irqsave(&ihost->scic_lock, flags);
-			/* command never hit the device, so just free
-			 * the tci and skip the sequence increment
-			 */
-			isci_tci_free(ihost, ISCI_TAG_TCI(tag));
-			spin_unlock_irqrestore(&ihost->scic_lock, flags);
-		}
-		isci_put_device(idev);
 	}
+
+	if (status != SCI_SUCCESS && tag != SCI_CONTROLLER_INVALID_IO_TAG) {
+		spin_lock_irqsave(&ihost->scic_lock, flags);
+		/* command never hit the device, so just free
+		 * the tci and skip the sequence increment
+		 */
+		isci_tci_free(ihost, ISCI_TAG_TCI(tag));
+		spin_unlock_irqrestore(&ihost->scic_lock, flags);
+	}
+
+	isci_put_device(idev);
 	return 0;
 }