[PATCHv4 0/3] mpt3sas: exposing all reply queues

[PATCHv4 0/3] mpt3sas: exposing all reply queues

[Hannes Reinecke]

Hi all,

this is the second part of my patchset to enable scsi multiqueue for the
mpt3sas driver.
While the HBA only has a single mailbox register for submitting commands,
it does have individual receive queues per MSI-X interrupt and as such
does benefit from converting it to full multiqueue support.

On request from Broadcom the patchset has been split in two parts, one
to enable lockless command submission and a secondary one for exposing
all hardware queues to the OS.

As usual, comments and reviews are welcome.

Changes to v1:
- Include reviews from Christoph
- Use reserved commands for ioctl passthrough commands
- Include reviews from Sreekanth

Changes to v2:
- Rework ioctl code to not use blk_mq_busy_iter
- Open-code mpt3sas_scsi_direct_io_(get,set)

Changes to v3:
- Simplify task management code
- Simplify _wait_for_commands_to_complete
- Use first reserved smid for ioctl passthrough

Hannes Reinecke (3):
  mpt3sas: Use 'msix_index' as argument for put_smid functions
  block: Add blk_mq_make_unique_tag()
  mpt3sas: scsi-mq interrupt steering

 block/blk-mq-tag.c                       |  3 +-
 drivers/scsi/mpt3sas/mpt3sas_base.c      | 80 +++++++++++++++++++-------------
 drivers/scsi/mpt3sas/mpt3sas_base.h      | 20 ++++++--
 drivers/scsi/mpt3sas/mpt3sas_config.c    |  2 +-
 drivers/scsi/mpt3sas/mpt3sas_ctl.c       | 21 +++++----
 drivers/scsi/mpt3sas/mpt3sas_scsih.c     | 59 ++++++++++++++++++-----
 drivers/scsi/mpt3sas/mpt3sas_transport.c |  8 ++--
 include/linux/blk-mq.h                   |  5 ++
 8 files changed, 135 insertions(+), 63 deletions(-)

-- 
1.8.5.6

[PATCHv4 1/3] mpt3sas: Use 'msix_index' as argument for put_smid functions

[Hannes Reinecke]

Use msix_index as explicit argument for the various put_smid
callbacks.
No functional change.

Signed-off-by: Hannes Reinecke <hare@suse.com>
---
 drivers/scsi/mpt3sas/mpt3sas_base.c      | 56 +++++++++++++++++---------------
 drivers/scsi/mpt3sas/mpt3sas_base.h      | 18 ++++++++--
 drivers/scsi/mpt3sas/mpt3sas_config.c    |  2 +-
 drivers/scsi/mpt3sas/mpt3sas_ctl.c       | 21 ++++++------
 drivers/scsi/mpt3sas/mpt3sas_scsih.c     | 21 ++++++------
 drivers/scsi/mpt3sas/mpt3sas_transport.c |  8 ++---
 6 files changed, 71 insertions(+), 55 deletions(-)

diff --git a/drivers/scsi/mpt3sas/mpt3sas_base.c b/drivers/scsi/mpt3sas/mpt3sas_base.c
index 966a775..7125296 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_base.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_base.c
@@ -849,7 +849,7 @@ static int mpt3sas_remove_dead_ioc_func(void *arg)
 	ack_request->EventContext = mpi_reply->EventContext;
 	ack_request->VF_ID = 0;  /* TODO */
 	ack_request->VP_ID = 0;
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 
  out:
 
@@ -2306,12 +2306,6 @@ struct scsiio_tracker *
 	return ioc->reply + (phys_addr - (u32)ioc->reply_dma);
 }
 
-static inline u8
-_base_get_msix_index(struct MPT3SAS_ADAPTER *ioc)
-{
-	return ioc->cpu_msix_table[raw_smp_processor_id()];
-}
-
 /**
  * mpt3sas_base_get_smid - obtain a free smid from internal queue
  * @ioc: per adapter object
@@ -2361,7 +2355,7 @@ struct scsiio_tracker *
 
 	smid = tag + 1;
 	request->cb_idx = cb_idx;
-	request->msix_io = _base_get_msix_index(ioc);
+	request->msix_io = mpt3sas_base_get_msix_index(ioc);
 	request->smid = smid;
 	INIT_LIST_HEAD(&request->chain_list);
 	return smid;
@@ -2503,18 +2497,20 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
  * @ioc: per adapter object
  * @smid: system request message index
  * @handle: device handle
+ * @msix_index: MSI-X index
  *
  * Return nothing.
  */
 static void
-_base_put_smid_scsi_io(struct MPT3SAS_ADAPTER *ioc, u16 smid, u16 handle)
+_base_put_smid_scsi_io(struct MPT3SAS_ADAPTER *ioc, u16 smid,
+		       u16 handle, u16 msix_index)
 {
 	Mpi2RequestDescriptorUnion_t descriptor;
 	u64 *request = (u64 *)&descriptor;
 
 
 	descriptor.SCSIIO.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_SCSI_IO;
-	descriptor.SCSIIO.MSIxIndex =  _base_get_msix_index(ioc);
+	descriptor.SCSIIO.MSIxIndex =  cpu_to_le16(msix_index);
 	descriptor.SCSIIO.SMID = cpu_to_le16(smid);
 	descriptor.SCSIIO.DevHandle = cpu_to_le16(handle);
 	descriptor.SCSIIO.LMID = 0;
@@ -2527,19 +2523,20 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
  * @ioc: per adapter object
  * @smid: system request message index
  * @handle: device handle
+ * @msix_index: MSI-X index
  *
  * Return nothing.
  */
 static void
 _base_put_smid_fast_path(struct MPT3SAS_ADAPTER *ioc, u16 smid,
-	u16 handle)
+	u16 handle, u16 msix_index)
 {
 	Mpi2RequestDescriptorUnion_t descriptor;
 	u64 *request = (u64 *)&descriptor;
 
 	descriptor.SCSIIO.RequestFlags =
 	    MPI25_REQ_DESCRIPT_FLAGS_FAST_PATH_SCSI_IO;
-	descriptor.SCSIIO.MSIxIndex = _base_get_msix_index(ioc);
+	descriptor.SCSIIO.MSIxIndex = cpu_to_le16(msix_index);
 	descriptor.SCSIIO.SMID = cpu_to_le16(smid);
 	descriptor.SCSIIO.DevHandle = cpu_to_le16(handle);
 	descriptor.SCSIIO.LMID = 0;
@@ -2563,7 +2560,7 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
 
 	descriptor.HighPriority.RequestFlags =
 	    MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY;
-	descriptor.HighPriority.MSIxIndex =  msix_task;
+	descriptor.HighPriority.MSIxIndex =  cpu_to_le16(msix_task);
 	descriptor.HighPriority.SMID = cpu_to_le16(smid);
 	descriptor.HighPriority.LMID = 0;
 	descriptor.HighPriority.Reserved1 = 0;
@@ -2575,17 +2572,18 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
  * _base_put_smid_default - Default, primarily used for config pages
  * @ioc: per adapter object
  * @smid: system request message index
+ * @msix_index: MSI-X index
  *
  * Return nothing.
  */
 static void
-_base_put_smid_default(struct MPT3SAS_ADAPTER *ioc, u16 smid)
+_base_put_smid_default(struct MPT3SAS_ADAPTER *ioc, u16 smid, u16 msix_index)
 {
 	Mpi2RequestDescriptorUnion_t descriptor;
 	u64 *request = (u64 *)&descriptor;
 
 	descriptor.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE;
-	descriptor.Default.MSIxIndex =  _base_get_msix_index(ioc);
+	descriptor.Default.MSIxIndex =  cpu_to_le16(msix_index);
 	descriptor.Default.SMID = cpu_to_le16(smid);
 	descriptor.Default.LMID = 0;
 	descriptor.Default.DescriptorTypeDependent = 0;
@@ -2599,18 +2597,19 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
 * @ioc: per adapter object
 * @smid: system request message index
 * @handle: device handle, unused in this function, for function type match
+* @msix_index: MSI-X index
 *
 * Return nothing.
 */
 static void
 _base_put_smid_scsi_io_atomic(struct MPT3SAS_ADAPTER *ioc, u16 smid,
-	u16 handle)
+	u16 handle, u16 msix_index)
 {
 	Mpi26AtomicRequestDescriptor_t descriptor;
 	u32 *request = (u32 *)&descriptor;
 
 	descriptor.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_SCSI_IO;
-	descriptor.MSIxIndex = _base_get_msix_index(ioc);
+	descriptor.MSIxIndex = cpu_to_le16(msix_index);
 	descriptor.SMID = cpu_to_le16(smid);
 
 	writel(cpu_to_le32(*request), &ioc->chip->AtomicRequestDescriptorPost);
@@ -2622,17 +2621,18 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
  * @ioc: per adapter object
  * @smid: system request message index
  * @handle: device handle, unused in this function, for function type match
+ * @msix_index: MSI-X index
  * Return nothing
  */
 static void
 _base_put_smid_fast_path_atomic(struct MPT3SAS_ADAPTER *ioc, u16 smid,
-	u16 handle)
+	u16 handle, u16 msix_index)
 {
 	Mpi26AtomicRequestDescriptor_t descriptor;
 	u32 *request = (u32 *)&descriptor;
 
 	descriptor.RequestFlags = MPI25_REQ_DESCRIPT_FLAGS_FAST_PATH_SCSI_IO;
-	descriptor.MSIxIndex = _base_get_msix_index(ioc);
+	descriptor.MSIxIndex = cpu_to_le16(msix_index);
 	descriptor.SMID = cpu_to_le16(smid);
 
 	writel(cpu_to_le32(*request), &ioc->chip->AtomicRequestDescriptorPost);
@@ -2655,7 +2655,7 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
 	u32 *request = (u32 *)&descriptor;
 
 	descriptor.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY;
-	descriptor.MSIxIndex = msix_task;
+	descriptor.MSIxIndex = cpu_to_le16(msix_task);
 	descriptor.SMID = cpu_to_le16(smid);
 
 	writel(cpu_to_le32(*request), &ioc->chip->AtomicRequestDescriptorPost);
@@ -2666,17 +2666,19 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
  * use Atomic Request Descriptor
  * @ioc: per adapter object
  * @smid: system request message index
+ * @msix_index: MSI-X index
  *
  * Return nothing.
  */
 static void
-_base_put_smid_default_atomic(struct MPT3SAS_ADAPTER *ioc, u16 smid)
+_base_put_smid_default_atomic(struct MPT3SAS_ADAPTER *ioc,
+	u16 smid, u16 msix_index)
 {
 	Mpi26AtomicRequestDescriptor_t descriptor;
 	u32 *request = (u32 *)&descriptor;
 
 	descriptor.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE;
-	descriptor.MSIxIndex = _base_get_msix_index(ioc);
+	descriptor.MSIxIndex = cpu_to_le16(msix_index);
 	descriptor.SMID = cpu_to_le16(smid);
 
 	writel(cpu_to_le32(*request), &ioc->chip->AtomicRequestDescriptorPost);
@@ -4182,7 +4184,7 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
 	    mpi_request->Operation == MPI2_SAS_OP_PHY_LINK_RESET)
 		ioc->ioc_link_reset_in_progress = 1;
 	init_completion(&ioc->base_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->base_cmds.done,
 	    msecs_to_jiffies(10000));
 	if ((mpi_request->Operation == MPI2_SAS_OP_PHY_HARD_RESET ||
@@ -4282,7 +4284,7 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
 	ioc->base_cmds.smid = smid;
 	memcpy(request, mpi_request, sizeof(Mpi2SepReply_t));
 	init_completion(&ioc->base_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->base_cmds.done,
 	    msecs_to_jiffies(10000));
 	if (!(ioc->base_cmds.status & MPT3_CMD_COMPLETE)) {
@@ -4696,7 +4698,7 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
 	mpi_request->Function = MPI2_FUNCTION_PORT_ENABLE;
 
 	init_completion(&ioc->port_enable_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->port_enable_cmds.done, 300*HZ);
 	if (!(ioc->port_enable_cmds.status & MPT3_CMD_COMPLETE)) {
 		pr_err(MPT3SAS_FMT "%s: timeout\n",
@@ -4759,7 +4761,7 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
 	memset(mpi_request, 0, sizeof(Mpi2PortEnableRequest_t));
 	mpi_request->Function = MPI2_FUNCTION_PORT_ENABLE;
 
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	return 0;
 }
 
@@ -4878,7 +4880,7 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
 		mpi_request->EventMasks[i] =
 		    cpu_to_le32(ioc->event_masks[i]);
 	init_completion(&ioc->base_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->base_cmds.done, 30*HZ);
 	if (!(ioc->base_cmds.status & MPT3_CMD_COMPLETE)) {
 		pr_err(MPT3SAS_FMT "%s: timeout\n",
diff --git a/drivers/scsi/mpt3sas/mpt3sas_base.h b/drivers/scsi/mpt3sas/mpt3sas_base.h
index 08ddf08..a6dcb67 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_base.h
+++ b/drivers/scsi/mpt3sas/mpt3sas_base.h
@@ -736,8 +736,9 @@ typedef void (*MPT_BUILD_ZERO_LEN_SGE)(struct MPT3SAS_ADAPTER *ioc,
 
 /* To support atomic and non atomic descriptors*/
 typedef void (*PUT_SMID_IO_FP_HIP) (struct MPT3SAS_ADAPTER *ioc, u16 smid,
-	u16 funcdep);
-typedef void (*PUT_SMID_DEFAULT) (struct MPT3SAS_ADAPTER *ioc, u16 smid);
+				    u16 funcdep, u16 msix_index);
+typedef void (*PUT_SMID_DEFAULT) (struct MPT3SAS_ADAPTER *ioc, u16 smid,
+				  u16 msix_index);
 
 /* IOC Facts and Port Facts converted from little endian to cpu */
 union mpi3_version_union {
@@ -1196,7 +1197,7 @@ struct MPT3SAS_ADAPTER {
 	u8		atomic_desc_capable;
 	PUT_SMID_IO_FP_HIP put_smid_scsi_io;
 	PUT_SMID_IO_FP_HIP put_smid_fast_path;
-	PUT_SMID_IO_FP_HIP put_smid_hi_priority;
+	PUT_SMID_DEFAULT put_smid_hi_priority;
 	PUT_SMID_DEFAULT put_smid_default;
 
 };
@@ -1276,6 +1277,17 @@ void mpt3sas_base_update_missing_delay(struct MPT3SAS_ADAPTER *ioc,
 
 int mpt3sas_port_enable(struct MPT3SAS_ADAPTER *ioc);
 
+static inline u8
+mpt3sas_base_get_msix_index(struct MPT3SAS_ADAPTER *ioc)
+{
+	return ioc->cpu_msix_table[raw_smp_processor_id()];
+}
+
+static inline void
+mpt3sas_base_put_smid_default(struct MPT3SAS_ADAPTER *ioc, u16 smid)
+{
+	ioc->put_smid_default(ioc, smid, mpt3sas_base_get_msix_index(ioc));
+}
 
 /* scsih shared API */
 struct scsi_cmnd *mpt3sas_scsih_scsi_lookup_get(struct MPT3SAS_ADAPTER *ioc,
diff --git a/drivers/scsi/mpt3sas/mpt3sas_config.c b/drivers/scsi/mpt3sas/mpt3sas_config.c
index dd62701..cebfd73 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_config.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_config.c
@@ -384,7 +384,7 @@ struct config_request {
 	memcpy(config_request, mpi_request, sizeof(Mpi2ConfigRequest_t));
 	_config_display_some_debug(ioc, smid, "config_request", NULL);
 	init_completion(&ioc->config_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->config_cmds.done, timeout*HZ);
 	if (!(ioc->config_cmds.status & MPT3_CMD_COMPLETE)) {
 		pr_err(MPT3SAS_FMT "%s: timeout\n",
diff --git a/drivers/scsi/mpt3sas/mpt3sas_ctl.c b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
index 7bbc5b1..24695ec 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_ctl.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
@@ -805,9 +805,10 @@ enum block_state {
 		ioc->build_sg(ioc, psge, data_out_dma, data_out_sz,
 		    data_in_dma, data_in_sz);
 		if (mpi_request->Function == MPI2_FUNCTION_SCSI_IO_REQUEST)
-			ioc->put_smid_scsi_io(ioc, smid, device_handle);
+			ioc->put_smid_scsi_io(ioc, smid, device_handle,
+					      mpt3sas_base_get_msix_index(ioc));
 		else
-			ioc->put_smid_default(ioc, smid);
+			mpt3sas_base_put_smid_default(ioc, smid);
 		break;
 	}
 	case MPI2_FUNCTION_SCSI_TASK_MGMT:
@@ -875,7 +876,7 @@ enum block_state {
 		}
 		ioc->build_sg(ioc, psge, data_out_dma, data_out_sz, data_in_dma,
 		    data_in_sz);
-		ioc->put_smid_default(ioc, smid);
+		mpt3sas_base_put_smid_default(ioc, smid);
 		break;
 	}
 	case MPI2_FUNCTION_SATA_PASSTHROUGH:
@@ -890,7 +891,7 @@ enum block_state {
 		}
 		ioc->build_sg(ioc, psge, data_out_dma, data_out_sz, data_in_dma,
 		    data_in_sz);
-		ioc->put_smid_default(ioc, smid);
+		mpt3sas_base_put_smid_default(ioc, smid);
 		break;
 	}
 	case MPI2_FUNCTION_FW_DOWNLOAD:
@@ -898,7 +899,7 @@ enum block_state {
 	{
 		ioc->build_sg(ioc, psge, data_out_dma, data_out_sz, data_in_dma,
 		    data_in_sz);
-		ioc->put_smid_default(ioc, smid);
+		mpt3sas_base_put_smid_default(ioc, smid);
 		break;
 	}
 	case MPI2_FUNCTION_TOOLBOX:
@@ -913,7 +914,7 @@ enum block_state {
 			ioc->build_sg_mpi(ioc, psge, data_out_dma, data_out_sz,
 				data_in_dma, data_in_sz);
 		}
-		ioc->put_smid_default(ioc, smid);
+		mpt3sas_base_put_smid_default(ioc, smid);
 		break;
 	}
 	case MPI2_FUNCTION_SAS_IO_UNIT_CONTROL:
@@ -932,7 +933,7 @@ enum block_state {
 	default:
 		ioc->build_sg_mpi(ioc, psge, data_out_dma, data_out_sz,
 		    data_in_dma, data_in_sz);
-		ioc->put_smid_default(ioc, smid);
+		mpt3sas_base_put_smid_default(ioc, smid);
 		break;
 	}
 
@@ -1521,7 +1522,7 @@ enum block_state {
 			cpu_to_le32(ioc->product_specific[buffer_type][i]);
 
 	init_completion(&ioc->ctl_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->ctl_cmds.done,
 	    MPT3_IOCTL_DEFAULT_TIMEOUT*HZ);
 
@@ -1868,7 +1869,7 @@ enum block_state {
 	mpi_request->VP_ID = 0;
 
 	init_completion(&ioc->ctl_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->ctl_cmds.done,
 	    MPT3_IOCTL_DEFAULT_TIMEOUT*HZ);
 
@@ -2135,7 +2136,7 @@ enum block_state {
 	mpi_request->VP_ID = 0;
 
 	init_completion(&ioc->ctl_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->ctl_cmds.done,
 	    MPT3_IOCTL_DEFAULT_TIMEOUT*HZ);
 
diff --git a/drivers/scsi/mpt3sas/mpt3sas_scsih.c b/drivers/scsi/mpt3sas/mpt3sas_scsih.c
index 8177519..f19f324 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_scsih.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_scsih.c
@@ -3159,7 +3159,7 @@ static struct fw_event_work *dequeue_next_fw_event(struct MPT3SAS_ADAPTER *ioc)
 	mpi_request->Function = MPI2_FUNCTION_SAS_IO_UNIT_CONTROL;
 	mpi_request->Operation = MPI2_SAS_OP_REMOVE_DEVICE;
 	mpi_request->DevHandle = mpi_request_tm->DevHandle;
-	ioc->put_smid_default(ioc, smid_sas_ctrl);
+	ioc->put_smid_default(ioc, smid_sas_ctrl, msix_index);
 
 	return _scsih_check_for_pending_tm(ioc, smid);
 }
@@ -3254,7 +3254,7 @@ static struct fw_event_work *dequeue_next_fw_event(struct MPT3SAS_ADAPTER *ioc)
 	mpi_request->Function = MPI2_FUNCTION_SCSI_TASK_MGMT;
 	mpi_request->DevHandle = cpu_to_le16(handle);
 	mpi_request->TaskType = MPI2_SCSITASKMGMT_TASKTYPE_TARGET_RESET;
-	ioc->put_smid_hi_priority(ioc, smid, 0);
+	ioc->put_smid_hi_priority(ioc, smid, mpt3sas_base_get_msix_index(ioc));
 }
 
 /**
@@ -3346,7 +3346,7 @@ static struct fw_event_work *dequeue_next_fw_event(struct MPT3SAS_ADAPTER *ioc)
 	ack_request->EventContext = event_context;
 	ack_request->VF_ID = 0;  /* TODO */
 	ack_request->VP_ID = 0;
-	ioc->put_smid_default(ioc, smid);
+	ioc->put_smid_default(ioc, smid, mpt3sas_base_get_msix_index(ioc));
 }
 
 /**
@@ -3403,7 +3403,7 @@ static struct fw_event_work *dequeue_next_fw_event(struct MPT3SAS_ADAPTER *ioc)
 	mpi_request->Function = MPI2_FUNCTION_SAS_IO_UNIT_CONTROL;
 	mpi_request->Operation = MPI2_SAS_OP_REMOVE_DEVICE;
 	mpi_request->DevHandle = handle;
-	ioc->put_smid_default(ioc, smid);
+	ioc->put_smid_default(ioc, smid, mpt3sas_base_get_msix_index(ioc));
 }
 
 /**
@@ -3920,6 +3920,7 @@ static inline bool ata_12_16_cmd(struct scsi_cmnd *scmd)
 	u32 mpi_control;
 	u16 smid;
 	u16 handle;
+	u16 msix_task = mpt3sas_base_get_msix_index(ioc);
 
 	if (ioc->logging_level & MPT_DEBUG_SCSI)
 		scsi_print_command(scmd);
@@ -4038,12 +4039,12 @@ static inline bool ata_12_16_cmd(struct scsi_cmnd *scmd)
 		if (sas_target_priv_data->flags & MPT_TARGET_FASTPATH_IO) {
 			mpi_request->IoFlags = cpu_to_le16(scmd->cmd_len |
 			    MPI25_SCSIIO_IOFLAGS_FAST_PATH);
-			ioc->put_smid_fast_path(ioc, smid, handle);
+			ioc->put_smid_fast_path(ioc, smid, handle, msix_task);
 		} else
 			ioc->put_smid_scsi_io(ioc, smid,
-			    le16_to_cpu(mpi_request->DevHandle));
+				le16_to_cpu(mpi_request->DevHandle), msix_task);
 	} else
-		ioc->put_smid_default(ioc, smid);
+		ioc->put_smid_default(ioc, smid, msix_task);
 	return 0;
 
  out:
@@ -4542,7 +4543,7 @@ static inline bool ata_12_16_cmd(struct scsi_cmnd *scmd)
 		mpi_request->DevHandle =
 		    cpu_to_le16(sas_device_priv_data->sas_target->handle);
 		ioc->put_smid_scsi_io(ioc, smid,
-		    sas_device_priv_data->sas_target->handle);
+			sas_device_priv_data->sas_target->handle, msix_index);
 		return 0;
 	}
 	/* turning off TLR */
@@ -6159,7 +6160,7 @@ static inline bool ata_12_16_cmd(struct scsi_cmnd *scmd)
 	    handle, phys_disk_num));
 
 	init_completion(&ioc->scsih_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->scsih_cmds.done, 10*HZ);
 
 	if (!(ioc->scsih_cmds.status & MPT3_CMD_COMPLETE)) {
@@ -8018,7 +8019,7 @@ static inline bool ata_12_16_cmd(struct scsi_cmnd *scmd)
 	if (!ioc->hide_ir_msg)
 		pr_info(MPT3SAS_FMT "IR shutdown (sending)\n", ioc->name);
 	init_completion(&ioc->scsih_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	ioc->put_smid_default(ioc, smid, mpt3sas_base_get_msix_index(ioc));
 	wait_for_completion_timeout(&ioc->scsih_cmds.done, 10*HZ);
 
 	if (!(ioc->scsih_cmds.status & MPT3_CMD_COMPLETE)) {
diff --git a/drivers/scsi/mpt3sas/mpt3sas_transport.c b/drivers/scsi/mpt3sas/mpt3sas_transport.c
index 7f1d578..b74faf1 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_transport.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_transport.c
@@ -392,7 +392,7 @@ struct rep_manu_reply {
 		"report_manufacture - send to sas_addr(0x%016llx)\n",
 		ioc->name, (unsigned long long)sas_address));
 	init_completion(&ioc->transport_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->transport_cmds.done, 10*HZ);
 
 	if (!(ioc->transport_cmds.status & MPT3_CMD_COMPLETE)) {
@@ -1198,7 +1198,7 @@ struct phy_error_log_reply {
 		ioc->name, (unsigned long long)phy->identify.sas_address,
 		phy->number));
 	init_completion(&ioc->transport_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->transport_cmds.done, 10*HZ);
 
 	if (!(ioc->transport_cmds.status & MPT3_CMD_COMPLETE)) {
@@ -1514,7 +1514,7 @@ struct phy_control_reply {
 		ioc->name, (unsigned long long)phy->identify.sas_address,
 		phy->number, phy_operation));
 	init_completion(&ioc->transport_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->transport_cmds.done, 10*HZ);
 
 	if (!(ioc->transport_cmds.status & MPT3_CMD_COMPLETE)) {
@@ -2032,7 +2032,7 @@ struct phy_control_reply {
 		"%s - sending smp request\n", ioc->name, __func__));
 
 	init_completion(&ioc->transport_cmds.done);
-	ioc->put_smid_default(ioc, smid);
+	mpt3sas_base_put_smid_default(ioc, smid);
 	wait_for_completion_timeout(&ioc->transport_cmds.done, 10*HZ);
 
 	if (!(ioc->transport_cmds.status & MPT3_CMD_COMPLETE)) {
-- 
1.8.5.6

[PATCHv4 2/3] block: Add blk_mq_make_unique_tag()

[Hannes Reinecke]

Implement blk_mq_make_unique_tag() to generate a unique tag
value from a given hardware queue and tag value.

Signed-off-by: Hannes Reinecke <hare@suse.com>
---
 block/blk-mq-tag.c     | 3 +--
 include/linux/blk-mq.h | 5 +++++
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/block/blk-mq-tag.c b/block/blk-mq-tag.c
index dcf5ce3..2df6d35 100644
--- a/block/blk-mq-tag.c
+++ b/block/blk-mq-tag.c
@@ -450,8 +450,7 @@ u32 blk_mq_unique_tag(struct request *rq)
 		hwq = hctx->queue_num;
 	}
 
-	return (hwq << BLK_MQ_UNIQUE_TAG_BITS) |
-		(rq->tag & BLK_MQ_UNIQUE_TAG_MASK);
+	return blk_mq_make_unique_tag(hwq, rq->tag);
 }
 EXPORT_SYMBOL(blk_mq_unique_tag);
 
diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h
index 4a2ab5d9..1442c34 100644
--- a/include/linux/blk-mq.h
+++ b/include/linux/blk-mq.h
@@ -212,6 +212,11 @@ static inline u16 blk_mq_unique_tag_to_tag(u32 unique_tag)
 	return unique_tag & BLK_MQ_UNIQUE_TAG_MASK;
 }
 
+static inline u32 blk_mq_make_unique_tag(u16 hwq, u16 tag)
+{
+	return (hwq << BLK_MQ_UNIQUE_TAG_BITS) |
+		(tag & BLK_MQ_UNIQUE_TAG_MASK);
+}
 
 int blk_mq_request_started(struct request *rq);
 void blk_mq_start_request(struct request *rq);
-- 
1.8.5.6

[PATCHv4 3/3] mpt3sas: scsi-mq interrupt steering

[Hannes Reinecke]

The device has several reply queues, so this patch implements
correct scsi-mq interrupt steering to take advantage of full
blk-mq support.
However, as performance results are inconclusive this patch
also adds a module parameter 'enable_scsi_mq' to expose all
completion queues via scsi-mq; default is off.

Signed-off-by: Hannes Reinecke <hare@suse.com>
---
 drivers/scsi/mpt3sas/mpt3sas_base.c  | 26 +++++++++++++++++++-----
 drivers/scsi/mpt3sas/mpt3sas_base.h  |  2 ++
 drivers/scsi/mpt3sas/mpt3sas_scsih.c | 38 ++++++++++++++++++++++++++++++++++--
 3 files changed, 59 insertions(+), 7 deletions(-)

diff --git a/drivers/scsi/mpt3sas/mpt3sas_base.c b/drivers/scsi/mpt3sas/mpt3sas_base.c
index 7125296..b959306 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_base.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_base.c
@@ -2350,12 +2350,24 @@ struct scsiio_tracker *
 	struct scsi_cmnd *scmd)
 {
 	struct scsiio_tracker *request = scsi_cmd_priv(scmd);
-	unsigned int tag = scmd->request->tag;
+	unsigned int tag;
 	u16 smid;
+	u16 hwq = 0;
 
-	smid = tag + 1;
+	if (ioc->shost->use_blk_mq) {
+		u32 unique_tag = blk_mq_unique_tag(scmd->request);
+		hwq = blk_mq_unique_tag_to_hwq(unique_tag);
+		tag = blk_mq_unique_tag_to_tag(unique_tag);
+		smid = (tag * ioc->shost->nr_hw_queues) + hwq + 1;
+	} else {
+		tag = scmd->request->tag;
+		smid = tag + 1;
+	}
 	request->cb_idx = cb_idx;
-	request->msix_io = mpt3sas_base_get_msix_index(ioc);
+	if (ioc->shost->nr_hw_queues > 1)
+		request->msix_io = hwq;
+	else
+		request->msix_io = mpt3sas_base_get_msix_index(ioc);
 	request->smid = smid;
 	INIT_LIST_HEAD(&request->chain_list);
 	return smid;
@@ -3521,9 +3533,13 @@ void mpt3sas_base_clear_st(struct MPT3SAS_ADAPTER *ioc,
 	 * with some internal commands that could be outstanding
 	 */
 	ioc->shost->can_queue = ioc->scsiio_depth - INTERNAL_SCSIIO_CMDS_COUNT;
+	if (ioc->expose_all_queues_flag) {
+		ioc->shost->nr_hw_queues = ioc->reply_queue_count;
+		ioc->shost->can_queue /= ioc->reply_queue_count;
+	}
 	dinitprintk(ioc, pr_info(MPT3SAS_FMT
-		"scsi host: can_queue depth (%d)\n",
-		ioc->name, ioc->shost->can_queue));
+		"scsi host: can_queue depth (%d), nr_hw_queues (%d)\n",
+		ioc->name, ioc->shost->can_queue, ioc->shost->nr_hw_queues));
 
 
 	/* contiguous pool for request and chains, 16 byte align, one extra "
diff --git a/drivers/scsi/mpt3sas/mpt3sas_base.h b/drivers/scsi/mpt3sas/mpt3sas_base.h
index a6dcb67..df2535a 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_base.h
+++ b/drivers/scsi/mpt3sas/mpt3sas_base.h
@@ -819,6 +819,7 @@ struct reply_post_struct {
  * @fw_event_lock:
  * @fw_event_list: list of fw events
  * @aen_event_read_flag: event log was read
+ * @expose_all_queues_flag: expose all reply queues to scsi mq
  * @broadcast_aen_busy: broadcast aen waiting to be serviced
  * @shost_recovery: host reset in progress
  * @ioc_reset_in_progress_lock:
@@ -981,6 +982,7 @@ struct MPT3SAS_ADAPTER {
 
 	 /* misc flags */
 	int		aen_event_read_flag;
+	int		expose_all_queues_flag;
 	u8		broadcast_aen_busy;
 	u16		broadcast_aen_pending;
 	u8		shost_recovery;
diff --git a/drivers/scsi/mpt3sas/mpt3sas_scsih.c b/drivers/scsi/mpt3sas/mpt3sas_scsih.c
index f19f324..0ff507b 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_scsih.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_scsih.c
@@ -54,6 +54,7 @@
 #include <linux/interrupt.h>
 #include <linux/aer.h>
 #include <linux/raid_class.h>
+#include <linux/blk-mq-pci.h>
 #include <asm/unaligned.h>
 
 #include "mpt3sas_base.h"
@@ -147,6 +148,10 @@ static int _scsih_add_device(struct MPT3SAS_ADAPTER *ioc, u16 handle,
 module_param(prot_mask, int, 0);
 MODULE_PARM_DESC(prot_mask, " host protection capabilities mask, def=7 ");
 
+static int enable_scsi_mq;
+module_param(enable_scsi_mq, int, S_IRUGO);
+MODULE_PARM_DESC(enable_scsi_mq,
+	"Enable scsi-mq (single queue=0/all queues=1/default=0)");
 
 /* raid transport support */
 struct raid_template *mpt3sas_raid_template;
@@ -1076,7 +1081,15 @@ struct scsi_cmnd *
 
 	if (smid > 0  &&
 	    smid <= ioc->scsiio_depth - INTERNAL_SCSIIO_CMDS_COUNT) {
-		u32 unique_tag = smid - 1;
+		u32 unique_tag;
+
+		if (ioc->shost->use_blk_mq) {
+			u16 hwq = (smid - 1) % ioc->shost->nr_hw_queues;
+			u16 tag = (smid - 1) / ioc->shost->nr_hw_queues;
+
+			unique_tag = blk_mq_make_unique_tag(hwq, tag);
+		} else
+			unique_tag = smid - 1;
 
 		scmd = scsi_host_find_tag(ioc->shost, unique_tag);
 		if (scmd) {
@@ -4035,6 +4048,11 @@ static inline bool ata_12_16_cmd(struct scsi_cmnd *scmd)
 	if (raid_device && raid_device->direct_io_enabled)
 		mpt3sas_setup_direct_io(ioc, scmd, raid_device, mpi_request);
 
+	if (shost_use_blk_mq(ioc->shost) && (ioc->shost->nr_hw_queues > 1)) {
+		u32 unique_tag = blk_mq_unique_tag(scmd->request);
+
+		msix_task = blk_mq_unique_tag_to_hwq(unique_tag);
+	}
 	if (likely(mpi_request->Function == MPI2_FUNCTION_SCSI_IO_REQUEST)) {
 		if (sas_target_priv_data->flags & MPT_TARGET_FASTPATH_IO) {
 			mpi_request->IoFlags = cpu_to_le16(scmd->cmd_len |
@@ -8435,6 +8453,16 @@ static void sas_device_make_active(struct MPT3SAS_ADAPTER *ioc,
 	return 1;
 }
 
+static int scsih_map_queues(struct Scsi_Host *shost)
+{
+	struct MPT3SAS_ADAPTER *ioc = shost_priv(shost);
+
+	if (shost->nr_hw_queues > 1)
+		return blk_mq_pci_map_queues(&shost->tag_set, ioc->pdev);
+
+	return -EINVAL;
+}
+
 /* shost template for SAS 2.0 HBA devices */
 static struct scsi_host_template mpt2sas_driver_template = {
 	.module				= THIS_MODULE,
@@ -8448,6 +8476,7 @@ static void sas_device_make_active(struct MPT3SAS_ADAPTER *ioc,
 	.slave_destroy			= scsih_slave_destroy,
 	.scan_finished			= scsih_scan_finished,
 	.scan_start			= scsih_scan_start,
+	.map_queues			= scsih_map_queues,
 	.change_queue_depth		= scsih_change_queue_depth,
 	.eh_abort_handler		= scsih_abort,
 	.eh_device_reset_handler	= scsih_dev_reset,
@@ -8714,7 +8743,12 @@ static void sas_device_make_active(struct MPT3SAS_ADAPTER *ioc,
 	shost->max_lun = max_lun;
 	shost->transportt = mpt3sas_transport_template;
 	shost->unique_id = ioc->id;
-
+	if (!shost->use_blk_mq)
+		enable_scsi_mq = 0;
+	else
+		shost->nr_hw_queues = 1;
+	if (enable_scsi_mq)
+		ioc->expose_all_queues_flag = 1;
 	if (max_sectors != 0xFFFF) {
 		if (max_sectors < 64) {
 			shost->max_sectors = 64;
-- 
1.8.5.6