[PATCH 00/19] hpsa driver updates May 8, 2014

[PATCH 00/19] hpsa driver updates May 8, 2014

[Stephen M. Cameron]

Not a lot of extensive changes in this set, some new PCI IDs,
some small bug fixes, quieting some noisy messages, allowing 
more reply queues, setting irq affinity hints, and some minor
locking improvments.

 drivers/scsi/hpsa.c     |  266 +++++++++++++++++++++++++++++------------------
 drivers/scsi/hpsa.h     |   42 ++++---
 drivers/scsi/hpsa_cmd.h |   47 +++-----
 3 files changed, 203 insertions(+), 152 deletions(-)

-- 
-- steve

[PATCH 01/19] hpsa: add new Smart Array PCI IDs (May 2014)

[Stephen M. Cameron]

From: Joe Handzik <joseph.t.handzik@hp.com>

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Joe Handzik <joseph.t.handzik@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Cc: stable@vger.kernel.org
---
 drivers/scsi/hpsa.c |   12 ++++++++++++
 1 files changed, 12 insertions(+), 0 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 9a6e4a2..fda6cf1 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -115,9 +115,15 @@ static const struct pci_device_id hpsa_pci_device_id[] = {
 	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21C3},
 	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21C4},
 	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21C5},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21C6},
 	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21C7},
 	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21C8},
 	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21C9},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21CA},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21CB},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21CC},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21CD},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSI,     0x103C, 0x21CE},
 	{PCI_VENDOR_ID_HP_3PAR, 0x0075, 0x1590, 0x0076},
 	{PCI_VENDOR_ID_HP_3PAR, 0x0075, 0x1590, 0x0087},
 	{PCI_VENDOR_ID_HP_3PAR, 0x0075, 0x1590, 0x007D},
@@ -165,9 +171,15 @@ static struct board_type products[] = {
 	{0x21C3103C, "Smart Array", &SA5_access},
 	{0x21C4103C, "Smart Array", &SA5_access},
 	{0x21C5103C, "Smart Array", &SA5_access},
+	{0x21C6103C, "Smart Array", &SA5_access},
 	{0x21C7103C, "Smart Array", &SA5_access},
 	{0x21C8103C, "Smart Array", &SA5_access},
 	{0x21C9103C, "Smart Array", &SA5_access},
+	{0x21CA103C, "Smart Array", &SA5_access},
+	{0x21CB103C, "Smart Array", &SA5_access},
+	{0x21CC103C, "Smart Array", &SA5_access},
+	{0x21CD103C, "Smart Array", &SA5_access},
+	{0x21CE103C, "Smart Array", &SA5_access},
 	{0x00761590, "HP Storage P1224 Array Controller", &SA5_access},
 	{0x00871590, "HP Storage P1224e Array Controller", &SA5_access},
 	{0x007D1590, "HP Storage P1228 Array Controller", &SA5_access},

[PATCH 02/19] Checking for a NULL return from a kzalloc call in hpsa_get_pdisk_of_ioaccel2.

[Stephen M. Cameron]

From: Joe Handzik <joseph.t.handzik@hp.com>

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Joe Handzik <joseph.t.handzik@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Cc: stable@vger.kernel.org
---
 drivers/scsi/hpsa.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index fda6cf1..ef66905 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -2848,6 +2848,8 @@ static int hpsa_get_pdisk_of_ioaccel2(struct ctlr_info *h,
 
 	/* Get the list of physical devices */
 	physicals = kzalloc(reportsize, GFP_KERNEL);
+	if (physicals == NULL)
+		return 0;
 	if (hpsa_scsi_do_report_phys_luns(h, (struct ReportLUNdata *) physicals,
 		reportsize, extended)) {
 		dev_err(&h->pdev->dev,

[PATCH 03/19] hpsa: remove unused fields from struct ctlr_info

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

The fields "major", "max_outstanding", and "usage_count"
of struct ctlr_info were not used for anything.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |    2 --
 drivers/scsi/hpsa.h |    3 ---
 2 files changed, 0 insertions(+), 5 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index ef66905..7ba4e30 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -5481,8 +5481,6 @@ static void start_io(struct ctlr_info *h)
 		 * condition.
 		 */
 		h->commands_outstanding++;
-		if (h->commands_outstanding > h->max_outstanding)
-			h->max_outstanding = h->commands_outstanding;
 
 		/* Tell the controller execute command */
 		spin_unlock_irqrestore(&h->lock, flags);
diff --git a/drivers/scsi/hpsa.h b/drivers/scsi/hpsa.h
index 44235a2..b988e04 100644
--- a/drivers/scsi/hpsa.h
+++ b/drivers/scsi/hpsa.h
@@ -115,11 +115,8 @@ struct ctlr_info {
 	int 	nr_cmds; /* Number of commands allowed on this controller */
 	struct CfgTable __iomem *cfgtable;
 	int	interrupts_enabled;
-	int	major;
 	int 	max_commands;
 	int	commands_outstanding;
-	int 	max_outstanding; /* Debug */
-	int	usage_count;  /* number of opens all all minor devices */
 #	define PERF_MODE_INT	0
 #	define DOORBELL_INT	1
 #	define SIMPLE_MODE_INT	2

[PATCH 04/19] hpsa: allow passthru ioctls to work with bidirectional commands

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

Treat the the data direction bits as a bit mask allowing both
READ and WRITE at the same time instead of testing for equality
to see if it's a exclusively a READ or a WRITE.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |    8 ++++----
 1 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 7ba4e30..1df9a8a 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -4956,7 +4956,7 @@ static int hpsa_passthru_ioctl(struct ctlr_info *h, void __user *argp)
 		buff = kmalloc(iocommand.buf_size, GFP_KERNEL);
 		if (buff == NULL)
 			return -EFAULT;
-		if (iocommand.Request.Type.Direction == XFER_WRITE) {
+		if (iocommand.Request.Type.Direction & XFER_WRITE) {
 			/* Copy the data into the buffer we created */
 			if (copy_from_user(buff, iocommand.buf,
 				iocommand.buf_size)) {
@@ -5019,7 +5019,7 @@ static int hpsa_passthru_ioctl(struct ctlr_info *h, void __user *argp)
 		rc = -EFAULT;
 		goto out;
 	}
-	if (iocommand.Request.Type.Direction == XFER_READ &&
+	if ((iocommand.Request.Type.Direction & XFER_READ) &&
 		iocommand.buf_size > 0) {
 		/* Copy the data out of the buffer we created */
 		if (copy_to_user(iocommand.buf, buff, iocommand.buf_size)) {
@@ -5096,7 +5096,7 @@ static int hpsa_big_passthru_ioctl(struct ctlr_info *h, void __user *argp)
 			status = -ENOMEM;
 			goto cleanup1;
 		}
-		if (ioc->Request.Type.Direction == XFER_WRITE) {
+		if (ioc->Request.Type.Direction & XFER_WRITE) {
 			if (copy_from_user(buff[sg_used], data_ptr, sz)) {
 				status = -ENOMEM;
 				goto cleanup1;
@@ -5148,7 +5148,7 @@ static int hpsa_big_passthru_ioctl(struct ctlr_info *h, void __user *argp)
 		status = -EFAULT;
 		goto cleanup0;
 	}
-	if (ioc->Request.Type.Direction == XFER_READ && ioc->buf_size > 0) {
+	if ((ioc->Request.Type.Direction & XFER_READ) && ioc->buf_size > 0) {
 		/* Copy the data out of the buffer we created */
 		BYTE __user *ptr = ioc->buf;
 		for (i = 0; i < sg_used; i++) {

[PATCH 05/19] hpsa: change doorbell reset delay to ten seconds

[Stephen M. Cameron]

From: Justin Lindley <justin.lindley@hp.com>

After 3.22 firmware, PMC firmware guys tell us the
previous 5 second delay after a reset now needs to
be 10 secs to avoid a PCIe error due to the driver
looking at the controller too soon after the reset.

Signed-off-by: Justin Lindley <justin.lindley@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 1df9a8a..2220b32 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -5831,12 +5831,12 @@ static int hpsa_controller_hard_reset(struct pci_dev *pdev,
 		dev_info(&pdev->dev, "using doorbell to reset controller\n");
 		writel(use_doorbell, vaddr + SA5_DOORBELL);
 
-		/* PMC hardware guys tell us we need a 5 second delay after
+		/* PMC hardware guys tell us we need a 10 second delay after
 		 * doorbell reset and before any attempt to talk to the board
 		 * at all to ensure that this actually works and doesn't fall
 		 * over in some weird corner cases.
 		 */
-		msleep(5000);
+		msleep(10000);
 	} else { /* Try to do it the PCI power state way */
 
 		/* Quoting from the Open CISS Specification: "The Power

[PATCH 06/19] hpsa: use gcc aligned attribute instead of manually padding structs

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c     |    3 ---
 drivers/scsi/hpsa_cmd.h |   33 ++++++---------------------------
 2 files changed, 6 insertions(+), 30 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 2220b32..b1ecfd8 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -6925,7 +6925,6 @@ reinit_after_soft_reset:
 	 * the 5 lower bits of the address are used by the hardware. and by
 	 * the driver.  See comments in hpsa.h for more info.
 	 */
-#define COMMANDLIST_ALIGNMENT 128
 	BUILD_BUG_ON(sizeof(struct CommandList) % COMMANDLIST_ALIGNMENT);
 	h = kzalloc(sizeof(*h), GFP_KERNEL);
 	if (!h)
@@ -7398,7 +7397,6 @@ static int hpsa_alloc_ioaccel_cmd_and_bft(struct ctlr_info *h)
 	 * because the 7 lower bits of the address are used by the
 	 * hardware.
 	 */
-#define IOACCEL1_COMMANDLIST_ALIGNMENT 128
 	BUILD_BUG_ON(sizeof(struct io_accel1_cmd) %
 			IOACCEL1_COMMANDLIST_ALIGNMENT);
 	h->ioaccel_cmd_pool =
@@ -7436,7 +7434,6 @@ static int ioaccel2_alloc_cmds_and_bft(struct ctlr_info *h)
 	if (h->ioaccel_maxsg > IOACCEL2_MAXSGENTRIES)
 		h->ioaccel_maxsg = IOACCEL2_MAXSGENTRIES;
 
-#define IOACCEL2_COMMANDLIST_ALIGNMENT 128
 	BUILD_BUG_ON(sizeof(struct io_accel2_cmd) %
 			IOACCEL2_COMMANDLIST_ALIGNMENT);
 	h->ioaccel2_cmd_pool =
diff --git a/drivers/scsi/hpsa_cmd.h b/drivers/scsi/hpsa_cmd.h
index b5cc705..9737ac2 100644
--- a/drivers/scsi/hpsa_cmd.h
+++ b/drivers/scsi/hpsa_cmd.h
@@ -375,6 +375,7 @@ struct ctlr_info; /* defined in hpsa.h */
  *        or a bus address.
  */
 
+#define COMMANDLIST_ALIGNMENT 128
 struct CommandList {
 	struct CommandListHeader Header;
 	struct RequestBlock      Request;
@@ -389,21 +390,7 @@ struct CommandList {
 	struct list_head list;
 	struct completion *waiting;
 	void   *scsi_cmd;
-
-/* on 64 bit architectures, to get this to be 32-byte-aligned
- * it so happens we need PAD_64 bytes of padding, on 32 bit systems,
- * we need PAD_32 bytes of padding (see below).   This does that.
- * If it happens that 64 bit and 32 bit systems need different
- * padding, PAD_32 and PAD_64 can be set independently, and.
- * the code below will do the right thing.
- */
-#define IS_32_BIT ((8 - sizeof(long))/4)
-#define IS_64_BIT (!IS_32_BIT)
-#define PAD_32 (40)
-#define PAD_64 (12)
-#define COMMANDLIST_PAD (IS_32_BIT * PAD_32 + IS_64_BIT * PAD_64)
-	u8 pad[COMMANDLIST_PAD];
-};
+} __attribute__((aligned(COMMANDLIST_ALIGNMENT)));
 
 /* Max S/G elements in I/O accelerator command */
 #define IOACCEL1_MAXSGENTRIES           24
@@ -413,6 +400,7 @@ struct CommandList {
  * Structure for I/O accelerator (mode 1) commands.
  * Note that this structure must be 128-byte aligned in size.
  */
+#define IOACCEL1_COMMANDLIST_ALIGNMENT 128
 struct io_accel1_cmd {
 	u16 dev_handle;			/* 0x00 - 0x01 */
 	u8  reserved1;			/* 0x02 */
@@ -440,12 +428,7 @@ struct io_accel1_cmd {
 	struct vals32 host_addr;	/* 0x70 - 0x77 */
 	u8  CISS_LUN[8];		/* 0x78 - 0x7F */
 	struct SGDescriptor SG[IOACCEL1_MAXSGENTRIES];
-#define IOACCEL1_PAD_64 0
-#define IOACCEL1_PAD_32 0
-#define IOACCEL1_PAD (IS_32_BIT * IOACCEL1_PAD_32 + \
-			IS_64_BIT * IOACCEL1_PAD_64)
-	u8 pad[IOACCEL1_PAD];
-};
+} __attribute__((aligned(IOACCEL1_COMMANDLIST_ALIGNMENT)));
 
 #define IOACCEL1_FUNCTION_SCSIIO        0x00
 #define IOACCEL1_SGLOFFSET              32
@@ -510,14 +493,11 @@ struct io_accel2_scsi_response {
 	u8 sense_data_buff[32];		/* sense/response data buffer */
 };
 
-#define IOACCEL2_64_PAD 76
-#define IOACCEL2_32_PAD 76
-#define IOACCEL2_PAD (IS_32_BIT * IOACCEL2_32_PAD + \
-			IS_64_BIT * IOACCEL2_64_PAD)
 /*
  * Structure for I/O accelerator (mode 2 or m2) commands.
  * Note that this structure must be 128-byte aligned in size.
  */
+#define IOACCEL2_COMMANDLIST_ALIGNMENT 128
 struct io_accel2_cmd {
 	u8  IU_type;			/* IU Type */
 	u8  direction;			/* direction, memtype, and encryption */
@@ -544,8 +524,7 @@ struct io_accel2_cmd {
 	u32 tweak_upper;		/* Encryption tweak, upper 4 bytes */
 	struct ioaccel2_sg_element sg[IOACCEL2_MAXSGENTRIES];
 	struct io_accel2_scsi_response error_data;
-	u8 pad[IOACCEL2_PAD];
-};
+} __attribute__((aligned(IOACCEL2_COMMANDLIST_ALIGNMENT)));
 
 /*
  * defines for Mode 2 command struct

[PATCH 07/19] hpsa: remove dev_dbg() calls from hot paths

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

They are not completely free of cost when disabled and
when enabled emitting debug output for every command
submitted produces far too much output to be useful.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.h |    5 -----
 1 files changed, 0 insertions(+), 5 deletions(-)

diff --git a/drivers/scsi/hpsa.h b/drivers/scsi/hpsa.h
index b988e04..da67c07 100644
--- a/drivers/scsi/hpsa.h
+++ b/drivers/scsi/hpsa.h
@@ -342,8 +342,6 @@ struct offline_device_entry {
 static void SA5_submit_command(struct ctlr_info *h,
 	struct CommandList *c)
 {
-	dev_dbg(&h->pdev->dev, "Sending %x, tag = %x\n", c->busaddr,
-		c->Header.Tag.lower);
 	writel(c->busaddr, h->vaddr + SA5_REQUEST_PORT_OFFSET);
 	(void) readl(h->vaddr + SA5_SCRATCHPAD_OFFSET);
 }
@@ -351,8 +349,6 @@ static void SA5_submit_command(struct ctlr_info *h,
 static void SA5_submit_command_ioaccel2(struct ctlr_info *h,
 	struct CommandList *c)
 {
-	dev_dbg(&h->pdev->dev, "Sending %x, tag = %x\n", c->busaddr,
-		c->Header.Tag.lower);
 	if (c->cmd_type == CMD_IOACCEL2)
 		writel(c->busaddr, h->vaddr + IOACCEL2_INBOUND_POSTQ_32);
 	else
@@ -474,7 +470,6 @@ static bool SA5_intr_pending(struct ctlr_info *h)
 {
 	unsigned long register_value  =
 		readl(h->vaddr + SA5_INTR_STATUS);
-	dev_dbg(&h->pdev->dev, "intr_pending %lx\n", register_value);
 	return register_value & SA5_INTR_PENDING;
 }
 

[PATCH 08/19] hpsa: choose number of reply queues more intelligently.

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

No sense having 8 or 16 reply queues if you only have 4 cpus,
and likewise no sense limiting to 8 reply queues if you have
many more cpus.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c     |    2 ++
 drivers/scsi/hpsa_cmd.h |    2 +-
 2 files changed, 3 insertions(+), 1 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index b1ecfd8..b903e86 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -6157,6 +6157,8 @@ static void hpsa_interrupt_mode(struct ctlr_info *h)
 	if (pci_find_capability(h->pdev, PCI_CAP_ID_MSIX)) {
 		dev_info(&h->pdev->dev, "MSIX\n");
 		h->msix_vector = MAX_REPLY_QUEUES;
+		if (h->msix_vector > num_online_cpus())
+			h->msix_vector = num_online_cpus();
 		err = pci_enable_msix(h->pdev, hpsa_msix_entries,
 				      h->msix_vector);
 		if (err > 0) {
diff --git a/drivers/scsi/hpsa_cmd.h b/drivers/scsi/hpsa_cmd.h
index 9737ac2..55215d3 100644
--- a/drivers/scsi/hpsa_cmd.h
+++ b/drivers/scsi/hpsa_cmd.h
@@ -615,7 +615,7 @@ struct TransTable_struct {
 	u32            RepQCount;
 	u32            RepQCtrAddrLow32;
 	u32            RepQCtrAddrHigh32;
-#define MAX_REPLY_QUEUES 8
+#define MAX_REPLY_QUEUES 64
 	struct vals32  RepQAddr[MAX_REPLY_QUEUES];
 };
 

[PATCH 09/19] hpsa: allocate reply queues individually

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

Now that we can allocate more than 4 reply queues (up to 64)
we shouldn't try to make them share the same allocation but
should allocate them separately.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |   53 +++++++++++++++++++++++++++++++--------------------
 drivers/scsi/hpsa.h |   13 ++++++-------
 2 files changed, 38 insertions(+), 28 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index b903e86..9c44f26 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -695,7 +695,7 @@ static inline void addQ(struct list_head *list, struct CommandList *c)
 static inline u32 next_command(struct ctlr_info *h, u8 q)
 {
 	u32 a;
-	struct reply_pool *rq = &h->reply_queue[q];
+	struct reply_queue_buffer *rq = &h->reply_queue[q];
 	unsigned long flags;
 
 	if (h->transMethod & CFGTBL_Trans_io_accel1)
@@ -6700,6 +6700,20 @@ static void hpsa_free_irqs_and_disable_msix(struct ctlr_info *h)
 #endif /* CONFIG_PCI_MSI */
 }
 
+static void hpsa_free_reply_queues(struct ctlr_info *h)
+{
+	int i;
+
+	for (i = 0; i < h->nreply_queues; i++) {
+		if (!h->reply_queue[i].head)
+			continue;
+		pci_free_consistent(h->pdev, h->reply_queue_size,
+			h->reply_queue[i].head, h->reply_queue[i].busaddr);
+		h->reply_queue[i].head = NULL;
+		h->reply_queue[i].busaddr = 0;
+	}
+}
+
 static void hpsa_undo_allocations_after_kdump_soft_reset(struct ctlr_info *h)
 {
 	hpsa_free_irqs_and_disable_msix(h);
@@ -6707,8 +6721,7 @@ static void hpsa_undo_allocations_after_kdump_soft_reset(struct ctlr_info *h)
 	hpsa_free_cmd_pool(h);
 	kfree(h->ioaccel1_blockFetchTable);
 	kfree(h->blockFetchTable);
-	pci_free_consistent(h->pdev, h->reply_pool_size,
-		h->reply_pool, h->reply_pool_dhandle);
+	hpsa_free_reply_queues(h);
 	if (h->vaddr)
 		iounmap(h->vaddr);
 	if (h->transtable)
@@ -7157,8 +7170,7 @@ static void hpsa_remove_one(struct pci_dev *pdev)
 	pci_free_consistent(h->pdev,
 		h->nr_cmds * sizeof(struct ErrorInfo),
 		h->errinfo_pool, h->errinfo_pool_dhandle);
-	pci_free_consistent(h->pdev, h->reply_pool_size,
-		h->reply_pool, h->reply_pool_dhandle);
+	hpsa_free_reply_queues(h);
 	kfree(h->cmd_pool_bits);
 	kfree(h->blockFetchTable);
 	kfree(h->ioaccel1_blockFetchTable);
@@ -7271,7 +7283,8 @@ static void hpsa_enter_performant_mode(struct ctlr_info *h, u32 trans_support)
 	 */
 
 	/* Controller spec: zero out this buffer. */
-	memset(h->reply_pool, 0, h->reply_pool_size);
+	for (i = 0; i < h->nreply_queues; i++)
+		memset(h->reply_queue[i].head, 0, h->reply_queue_size);
 
 	bft[7] = SG_ENTRIES_IN_CMD + 4;
 	calc_bucket_map(bft, ARRAY_SIZE(bft),
@@ -7287,8 +7300,7 @@ static void hpsa_enter_performant_mode(struct ctlr_info *h, u32 trans_support)
 
 	for (i = 0; i < h->nreply_queues; i++) {
 		writel(0, &h->transtable->RepQAddr[i].upper);
-		writel(h->reply_pool_dhandle +
-			(h->max_commands * sizeof(u64) * i),
+		writel(h->reply_queue[i].busaddr,
 			&h->transtable->RepQAddr[i].lower);
 	}
 
@@ -7336,8 +7348,10 @@ static void hpsa_enter_performant_mode(struct ctlr_info *h, u32 trans_support)
 				h->ioaccel1_blockFetchTable);
 
 		/* initialize all reply queue entries to unused */
-		memset(h->reply_pool, (u8) IOACCEL_MODE1_REPLY_UNUSED,
-				h->reply_pool_size);
+		for (i = 0; i < h->nreply_queues; i++)
+			memset(h->reply_queue[i].head,
+				(u8) IOACCEL_MODE1_REPLY_UNUSED,
+				h->reply_queue_size);
 
 		/* set all the constant fields in the accelerator command
 		 * frames once at init time to save CPU cycles later.
@@ -7493,16 +7507,17 @@ static void hpsa_put_ctlr_into_performant_mode(struct ctlr_info *h)
 		}
 	}
 
-	/* TODO, check that this next line h->nreply_queues is correct */
 	h->nreply_queues = h->msix_vector > 0 ? h->msix_vector : 1;
 	hpsa_get_max_perf_mode_cmds(h);
 	/* Performant mode ring buffer and supporting data structures */
-	h->reply_pool_size = h->max_commands * sizeof(u64) * h->nreply_queues;
-	h->reply_pool = pci_alloc_consistent(h->pdev, h->reply_pool_size,
-				&(h->reply_pool_dhandle));
+	h->reply_queue_size = h->max_commands * sizeof(u64);
 
 	for (i = 0; i < h->nreply_queues; i++) {
-		h->reply_queue[i].head = &h->reply_pool[h->max_commands * i];
+		h->reply_queue[i].head = pci_alloc_consistent(h->pdev,
+						h->reply_queue_size,
+						&(h->reply_queue[i].busaddr));
+		if (!h->reply_queue[i].head)
+			goto clean_up;
 		h->reply_queue[i].size = h->max_commands;
 		h->reply_queue[i].wraparound = 1;  /* spec: init to 1 */
 		h->reply_queue[i].current_entry = 0;
@@ -7511,18 +7526,14 @@ static void hpsa_put_ctlr_into_performant_mode(struct ctlr_info *h)
 	/* Need a block fetch table for performant mode */
 	h->blockFetchTable = kmalloc(((SG_ENTRIES_IN_CMD + 1) *
 				sizeof(u32)), GFP_KERNEL);
-
-	if ((h->reply_pool == NULL)
-		|| (h->blockFetchTable == NULL))
+	if (!h->blockFetchTable)
 		goto clean_up;
 
 	hpsa_enter_performant_mode(h, trans_support);
 	return;
 
 clean_up:
-	if (h->reply_pool)
-		pci_free_consistent(h->pdev, h->reply_pool_size,
-			h->reply_pool, h->reply_pool_dhandle);
+	hpsa_free_reply_queues(h);
 	kfree(h->blockFetchTable);
 }
 
diff --git a/drivers/scsi/hpsa.h b/drivers/scsi/hpsa.h
index da67c07..fa63576 100644
--- a/drivers/scsi/hpsa.h
+++ b/drivers/scsi/hpsa.h
@@ -57,11 +57,12 @@ struct hpsa_scsi_dev_t {
 
 };
 
-struct reply_pool {
+struct reply_queue_buffer {
 	u64 *head;
 	size_t size;
 	u8 wraparound;
 	u32 current_entry;
+	dma_addr_t busaddr;
 };
 
 #pragma pack(1)
@@ -173,11 +174,9 @@ struct ctlr_info {
 	/*
 	 * Performant mode completion buffers
 	 */
-	u64 *reply_pool;
-	size_t reply_pool_size;
-	struct reply_pool reply_queue[MAX_REPLY_QUEUES];
+	size_t reply_queue_size;
+	struct reply_queue_buffer reply_queue[MAX_REPLY_QUEUES];
 	u8 nreply_queues;
-	dma_addr_t reply_pool_dhandle;
 	u32 *blockFetchTable;
 	u32 *ioaccel1_blockFetchTable;
 	u32 *ioaccel2_blockFetchTable;
@@ -391,7 +390,7 @@ static void SA5_performant_intr_mask(struct ctlr_info *h, unsigned long val)
 
 static unsigned long SA5_performant_completed(struct ctlr_info *h, u8 q)
 {
-	struct reply_pool *rq = &h->reply_queue[q];
+	struct reply_queue_buffer *rq = &h->reply_queue[q];
 	unsigned long flags, register_value = FIFO_EMPTY;
 
 	/* msi auto clears the interrupt pending bit. */
@@ -506,7 +505,7 @@ static bool SA5_ioaccel_mode1_intr_pending(struct ctlr_info *h)
 static unsigned long SA5_ioaccel_mode1_completed(struct ctlr_info *h, u8 q)
 {
 	u64 register_value;
-	struct reply_pool *rq = &h->reply_queue[q];
+	struct reply_queue_buffer *rq = &h->reply_queue[q];
 	unsigned long flags;
 
 	BUG_ON(q >= h->nreply_queues);

[PATCH 10/19] hpsa: set irq affinity hints to route MSI-X vectors across CPUs

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |   17 ++++++++++++++++-
 1 files changed, 16 insertions(+), 1 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 9c44f26..e8090e2 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -6608,6 +6608,17 @@ static void hpsa_free_cmd_pool(struct ctlr_info *h)
 			h->ioaccel_cmd_pool, h->ioaccel_cmd_pool_dhandle);
 }
 
+static void hpsa_irq_affinity_hints(struct ctlr_info *h)
+{
+	int i, cpu, rc;
+
+	cpu = cpumask_first(cpu_online_mask);
+	for (i = 0; i < h->msix_vector; i++) {
+		rc = irq_set_affinity_hint(h->intr[i], get_cpu_mask(cpu));
+		cpu = cpumask_next(cpu, cpu_online_mask);
+	}
+}
+
 static int hpsa_request_irq(struct ctlr_info *h,
 	irqreturn_t (*msixhandler)(int, void *),
 	irqreturn_t (*intxhandler)(int, void *))
@@ -6627,6 +6638,7 @@ static int hpsa_request_irq(struct ctlr_info *h,
 			rc = request_irq(h->intr[i], msixhandler,
 					0, h->devname,
 					&h->q[i]);
+		hpsa_irq_affinity_hints(h);
 	} else {
 		/* Use single reply pool */
 		if (h->msix_vector > 0 || h->msi_vector) {
@@ -6678,12 +6690,15 @@ static void free_irqs(struct ctlr_info *h)
 	if (!h->msix_vector || h->intr_mode != PERF_MODE_INT) {
 		/* Single reply queue, only one irq to free */
 		i = h->intr_mode;
+		irq_set_affinity_hint(h->intr[i], NULL);
 		free_irq(h->intr[i], &h->q[i]);
 		return;
 	}
 
-	for (i = 0; i < h->msix_vector; i++)
+	for (i = 0; i < h->msix_vector; i++) {
+		irq_set_affinity_hint(h->intr[i], NULL);
 		free_irq(h->intr[i], &h->q[i]);
+	}
 }
 
 static void hpsa_free_irqs_and_disable_msix(struct ctlr_info *h)

[PATCH 11/19] hpsa: use per-cpu variable for lockup_detected

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

Avoid excessive locking by using per-cpu variable for lockup_detected

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |   80 ++++++++++++++++++++++++++++++++++-----------------
 drivers/scsi/hpsa.h |    2 +
 2 files changed, 54 insertions(+), 28 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index e8090e2..acb88e5 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -48,6 +48,7 @@
 #include <linux/bitmap.h>
 #include <linux/atomic.h>
 #include <linux/jiffies.h>
+#include <linux/percpu.h>
 #include <asm/div64.h>
 #include "hpsa_cmd.h"
 #include "hpsa.h"
@@ -1991,20 +1992,26 @@ static inline void hpsa_scsi_do_simple_cmd_core(struct ctlr_info *h,
 	wait_for_completion(&wait);
 }
 
+static u32 lockup_detected(struct ctlr_info *h)
+{
+	int cpu;
+	u32 rc, *lockup_detected;
+
+	cpu = get_cpu();
+	lockup_detected = per_cpu_ptr(h->lockup_detected, cpu);
+	rc = *lockup_detected;
+	put_cpu();
+	return rc;
+}
+
 static void hpsa_scsi_do_simple_cmd_core_if_no_lockup(struct ctlr_info *h,
 	struct CommandList *c)
 {
-	unsigned long flags;
-
 	/* If controller lockup detected, fake a hardware error. */
-	spin_lock_irqsave(&h->lock, flags);
-	if (unlikely(h->lockup_detected)) {
-		spin_unlock_irqrestore(&h->lock, flags);
+	if (unlikely(lockup_detected(h)))
 		c->err_info->CommandStatus = CMD_HARDWARE_ERR;
-	} else {
-		spin_unlock_irqrestore(&h->lock, flags);
+	else
 		hpsa_scsi_do_simple_cmd_core(h, c);
-	}
 }
 
 #define MAX_DRIVER_CMD_RETRIES 25
@@ -3964,7 +3971,6 @@ static int hpsa_scsi_queue_command_lck(struct scsi_cmnd *cmd,
 	struct hpsa_scsi_dev_t *dev;
 	unsigned char scsi3addr[8];
 	struct CommandList *c;
-	unsigned long flags;
 	int rc = 0;
 
 	/* Get the ptr to our adapter structure out of cmd->host. */
@@ -3977,14 +3983,11 @@ static int hpsa_scsi_queue_command_lck(struct scsi_cmnd *cmd,
 	}
 	memcpy(scsi3addr, dev->scsi3addr, sizeof(scsi3addr));
 
-	spin_lock_irqsave(&h->lock, flags);
-	if (unlikely(h->lockup_detected)) {
-		spin_unlock_irqrestore(&h->lock, flags);
+	if (unlikely(lockup_detected(h))) {
 		cmd->result = DID_ERROR << 16;
 		done(cmd);
 		return 0;
 	}
-	spin_unlock_irqrestore(&h->lock, flags);
 	c = cmd_alloc(h);
 	if (c == NULL) {			/* trouble... */
 		dev_err(&h->pdev->dev, "cmd_alloc returned NULL!\n");
@@ -4096,16 +4099,13 @@ static int do_not_scan_if_controller_locked_up(struct ctlr_info *h)
 	 * we can prevent new rescan threads from piling up on a
 	 * locked up controller.
 	 */
-	spin_lock_irqsave(&h->lock, flags);
-	if (unlikely(h->lockup_detected)) {
-		spin_unlock_irqrestore(&h->lock, flags);
+	if (unlikely(lockup_detected(h))) {
 		spin_lock_irqsave(&h->scan_lock, flags);
 		h->scan_finished = 1;
 		wake_up_all(&h->scan_wait_queue);
 		spin_unlock_irqrestore(&h->scan_lock, flags);
 		return 1;
 	}
-	spin_unlock_irqrestore(&h->lock, flags);
 	return 0;
 }
 
@@ -6761,16 +6761,38 @@ static void fail_all_cmds_on_list(struct ctlr_info *h, struct list_head *list)
 	}
 }
 
+static void set_lockup_detected_for_all_cpus(struct ctlr_info *h, u32 value)
+{
+	int i, cpu;
+
+	cpu = cpumask_first(cpu_online_mask);
+	for (i = 0; i < num_online_cpus(); i++) {
+		u32 *lockup_detected;
+		lockup_detected = per_cpu_ptr(h->lockup_detected, cpu);
+		*lockup_detected = value;
+		cpu = cpumask_next(cpu, cpu_online_mask);
+	}
+	wmb(); /* be sure the per-cpu variables are out to memory */
+}
+
 static void controller_lockup_detected(struct ctlr_info *h)
 {
 	unsigned long flags;
+	u32 lockup_detected;
 
 	h->access.set_intr_mask(h, HPSA_INTR_OFF);
 	spin_lock_irqsave(&h->lock, flags);
-	h->lockup_detected = readl(h->vaddr + SA5_SCRATCHPAD_OFFSET);
+	lockup_detected = readl(h->vaddr + SA5_SCRATCHPAD_OFFSET);
+	if (!lockup_detected) {
+		/* no heartbeat, but controller gave us a zero. */
+		dev_warn(&h->pdev->dev,
+			"lockup detected but scratchpad register is zero\n");
+		lockup_detected = 0xffffffff;
+	}
+	set_lockup_detected_for_all_cpus(h, lockup_detected);
 	spin_unlock_irqrestore(&h->lock, flags);
 	dev_warn(&h->pdev->dev, "Controller lockup detected: 0x%08x\n",
-			h->lockup_detected);
+			lockup_detected);
 	pci_disable_device(h->pdev);
 	spin_lock_irqsave(&h->lock, flags);
 	fail_all_cmds_on_list(h, &h->cmpQ);
@@ -6905,7 +6927,7 @@ static void hpsa_monitor_ctlr_worker(struct work_struct *work)
 	struct ctlr_info *h = container_of(to_delayed_work(work),
 					struct ctlr_info, monitor_ctlr_work);
 	detect_controller_lockup(h);
-	if (h->lockup_detected)
+	if (lockup_detected(h))
 		return;
 
 	if (hpsa_ctlr_needs_rescan(h) || hpsa_offline_devices_ready(h)) {
@@ -6969,6 +6991,13 @@ reinit_after_soft_reset:
 	spin_lock_init(&h->offline_device_lock);
 	spin_lock_init(&h->scan_lock);
 	spin_lock_init(&h->passthru_count_lock);
+
+	/* Allocate and clear per-cpu variable lockup_detected */
+	h->lockup_detected = alloc_percpu(u32);
+	if (!h->lockup_detected)
+		goto clean1;
+	set_lockup_detected_for_all_cpus(h, 0);
+
 	rc = hpsa_pci_init(h);
 	if (rc != 0)
 		goto clean1;
@@ -7092,6 +7121,8 @@ clean4:
 	free_irqs(h);
 clean2:
 clean1:
+	if (h->lockup_detected)
+		free_percpu(h->lockup_detected);
 	kfree(h);
 	return rc;
 }
@@ -7100,16 +7131,10 @@ static void hpsa_flush_cache(struct ctlr_info *h)
 {
 	char *flush_buf;
 	struct CommandList *c;
-	unsigned long flags;
 
 	/* Don't bother trying to flush the cache if locked up */
-	spin_lock_irqsave(&h->lock, flags);
-	if (unlikely(h->lockup_detected)) {
-		spin_unlock_irqrestore(&h->lock, flags);
+	if (unlikely(lockup_detected(h)))
 		return;
-	}
-	spin_unlock_irqrestore(&h->lock, flags);
-
 	flush_buf = kzalloc(4, GFP_KERNEL);
 	if (!flush_buf)
 		return;
@@ -7193,6 +7218,7 @@ static void hpsa_remove_one(struct pci_dev *pdev)
 	kfree(h->hba_inquiry_data);
 	pci_disable_device(pdev);
 	pci_release_regions(pdev);
+	free_percpu(h->lockup_detected);
 	kfree(h);
 }
 
diff --git a/drivers/scsi/hpsa.h b/drivers/scsi/hpsa.h
index fa63576..8e8310e 100644
--- a/drivers/scsi/hpsa.h
+++ b/drivers/scsi/hpsa.h
@@ -191,7 +191,7 @@ struct ctlr_info {
 	u64 last_heartbeat_timestamp;
 	u32 heartbeat_sample_interval;
 	atomic_t firmware_flash_in_progress;
-	u32 lockup_detected;
+	u32 *lockup_detected;
 	struct delayed_work monitor_ctlr_work;
 	int remove_in_progress;
 	u32 fifo_recently_full;

[PATCH 12/19] hpsa: avoid unnecessary readl on every command submission

[Stephen M. Cameron]

From: root <root@rhel6u2e-s1.ldev.net>

for controllers which support either of the ioaccel transport methods.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |    7 +++++++
 drivers/scsi/hpsa.h |   15 ++++++++++++++-
 2 files changed, 21 insertions(+), 1 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index acb88e5..ad73017 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -7323,6 +7323,13 @@ static void hpsa_enter_performant_mode(struct ctlr_info *h, u32 trans_support)
 	 * 10 = 6 s/g entry or 24k
 	 */
 
+	/* If the controller supports either ioaccel method then
+	 * we can also use the RAID stack submit path that does not
+	 * perform the superfluous readl() after each command submission.
+	 */
+	if (trans_support & (CFGTBL_Trans_io_accel1 | CFGTBL_Trans_io_accel2))
+		access = SA5_performant_access_no_read;
+
 	/* Controller spec: zero out this buffer. */
 	for (i = 0; i < h->nreply_queues; i++)
 		memset(h->reply_queue[i].head, 0, h->reply_queue_size);
diff --git a/drivers/scsi/hpsa.h b/drivers/scsi/hpsa.h
index 8e8310e..1855469 100644
--- a/drivers/scsi/hpsa.h
+++ b/drivers/scsi/hpsa.h
@@ -345,6 +345,12 @@ static void SA5_submit_command(struct ctlr_info *h,
 	(void) readl(h->vaddr + SA5_SCRATCHPAD_OFFSET);
 }
 
+static void SA5_submit_command_no_read(struct ctlr_info *h,
+	struct CommandList *c)
+{
+	writel(c->busaddr, h->vaddr + SA5_REQUEST_PORT_OFFSET);
+}
+
 static void SA5_submit_command_ioaccel2(struct ctlr_info *h,
 	struct CommandList *c)
 {
@@ -352,7 +358,6 @@ static void SA5_submit_command_ioaccel2(struct ctlr_info *h,
 		writel(c->busaddr, h->vaddr + IOACCEL2_INBOUND_POSTQ_32);
 	else
 		writel(c->busaddr, h->vaddr + SA5_REQUEST_PORT_OFFSET);
-	(void) readl(h->vaddr + SA5_SCRATCHPAD_OFFSET);
 }
 
 /*
@@ -563,6 +568,14 @@ static struct access_method SA5_performant_access = {
 	SA5_performant_completed,
 };
 
+static struct access_method SA5_performant_access_no_read = {
+	SA5_submit_command_no_read,
+	SA5_performant_intr_mask,
+	SA5_fifo_full,
+	SA5_performant_intr_pending,
+	SA5_performant_completed,
+};
+
 struct board_type {
 	u32	board_id;
 	char	*product_name;

[PATCH 13/19] hpsa: Rearrange start_io to avoid one unlock/lock sequence in main io path

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |   24 ++++++++++++++++--------
 1 files changed, 16 insertions(+), 8 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index ad73017..c6ed5ea 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -194,7 +194,8 @@ static int number_of_controllers;
 static irqreturn_t do_hpsa_intr_intx(int irq, void *dev_id);
 static irqreturn_t do_hpsa_intr_msi(int irq, void *dev_id);
 static int hpsa_ioctl(struct scsi_device *dev, int cmd, void *arg);
-static void start_io(struct ctlr_info *h);
+static void lock_and_start_io(struct ctlr_info *h);
+static void start_io(struct ctlr_info *h, unsigned long *flags);
 
 #ifdef CONFIG_COMPAT
 static int hpsa_compat_ioctl(struct scsi_device *dev, int cmd, void *arg);
@@ -845,8 +846,8 @@ static void enqueue_cmd_and_start_io(struct ctlr_info *h,
 	spin_lock_irqsave(&h->lock, flags);
 	addQ(&h->reqQ, c);
 	h->Qdepth++;
+	start_io(h, &flags);
 	spin_unlock_irqrestore(&h->lock, flags);
-	start_io(h);
 }
 
 static inline void removeQ(struct CommandList *c)
@@ -5452,13 +5453,12 @@ static void __iomem *remap_pci_mem(ulong base, ulong size)
 
 /* Takes cmds off the submission queue and sends them to the hardware,
  * then puts them on the queue of cmds waiting for completion.
+ * Assumes h->lock is held
  */
-static void start_io(struct ctlr_info *h)
+static void start_io(struct ctlr_info *h, unsigned long *flags)
 {
 	struct CommandList *c;
-	unsigned long flags;
 
-	spin_lock_irqsave(&h->lock, flags);
 	while (!list_empty(&h->reqQ)) {
 		c = list_entry(h->reqQ.next, struct CommandList, list);
 		/* can't do anything if fifo is full */
@@ -5483,10 +5483,18 @@ static void start_io(struct ctlr_info *h)
 		h->commands_outstanding++;
 
 		/* Tell the controller execute command */
-		spin_unlock_irqrestore(&h->lock, flags);
+		spin_unlock_irqrestore(&h->lock, *flags);
 		h->access.submit_command(h, c);
-		spin_lock_irqsave(&h->lock, flags);
+		spin_lock_irqsave(&h->lock, *flags);
 	}
+}
+
+static void lock_and_start_io(struct ctlr_info *h)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&h->lock, flags);
+	start_io(h, &flags);
 	spin_unlock_irqrestore(&h->lock, flags);
 }
 
@@ -5554,7 +5562,7 @@ static inline void finish_cmd(struct CommandList *c)
 	else if (c->cmd_type == CMD_IOCTL_PEND)
 		complete(c->waiting);
 	if (unlikely(io_may_be_stalled))
-		start_io(h);
+		lock_and_start_io(h);
 }
 
 static inline u32 hpsa_tag_contains_index(u32 tag)

[PATCH 14/19] hpsa: define extended_report_lun_entry data structure

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c     |   21 ++++++++-------------
 drivers/scsi/hpsa_cmd.h |   12 +++++++++++-
 2 files changed, 19 insertions(+), 14 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index c6ed5ea..96436bb 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -2869,26 +2869,20 @@ static int hpsa_get_pdisk_of_ioaccel2(struct ctlr_info *h,
 	nphysicals = be32_to_cpu(*((__be32 *)physicals->LUNListLength)) /
 							responsesize;
 
-
 	/* find ioaccel2 handle in list of physicals: */
 	for (i = 0; i < nphysicals; i++) {
+		struct ext_report_lun_entry *entry = &physicals->LUN[i];
+
 		/* handle is in bytes 28-31 of each lun */
-		if (memcmp(&((struct ReportExtendedLUNdata *)
-				physicals)->LUN[i][20], &find, 4) != 0) {
+		if (entry->ioaccel_handle != find)
 			continue; /* didn't match */
-		}
 		found = 1;
-		memcpy(scsi3addr, &((struct ReportExtendedLUNdata *)
-					physicals)->LUN[i][0], 8);
+		memcpy(scsi3addr, entry->lunid, 8);
 		if (h->raid_offload_debug > 0)
 			dev_info(&h->pdev->dev,
-				"%s: Searched h=0x%08x, Found h=0x%08x, scsiaddr 0x%02x%02x%02x%02x%02x%02x%02x%02x\n",
+				"%s: Searched h=0x%08x, Found h=0x%08x, scsiaddr 0x%8phN\n",
 				__func__, find,
-				((struct ReportExtendedLUNdata *)
-					physicals)->LUN[i][20],
-				scsi3addr[0], scsi3addr[1], scsi3addr[2],
-				scsi3addr[3], scsi3addr[4], scsi3addr[5],
-				scsi3addr[6], scsi3addr[7]);
+				entry->ioaccel_handle, scsi3addr);
 		break; /* found it */
 	}
 
@@ -2973,7 +2967,8 @@ u8 *figure_lunaddrbytes(struct ctlr_info *h, int raid_ctlr_position, int i,
 		return RAID_CTLR_LUNID;
 
 	if (i < logicals_start)
-		return &physdev_list->LUN[i - (raid_ctlr_position == 0)][0];
+		return &physdev_list->LUN[i -
+				(raid_ctlr_position == 0)].lunid[0];
 
 	if (i < last_device)
 		return &logdev_list->LUN[i - nphysicals -
diff --git a/drivers/scsi/hpsa_cmd.h b/drivers/scsi/hpsa_cmd.h
index 55215d3..e394542 100644
--- a/drivers/scsi/hpsa_cmd.h
+++ b/drivers/scsi/hpsa_cmd.h
@@ -238,11 +238,21 @@ struct ReportLUNdata {
 	u8 LUN[HPSA_MAX_LUN][8];
 };
 
+struct ext_report_lun_entry {
+	u8 lunid[8];
+	u8 wwid[8];
+	u8 device_type;
+	u8 device_flags;
+	u8 lun_count; /* multi-lun device, how many luns */
+	u8 redundant_paths;
+	u32 ioaccel_handle; /* ioaccel1 only uses lower 16 bits */
+};
+
 struct ReportExtendedLUNdata {
 	u8 LUNListLength[4];
 	u8 extended_response_flag;
 	u8 reserved[3];
-	u8 LUN[HPSA_MAX_LUN][24];
+	struct ext_report_lun_entry LUN[HPSA_MAX_LUN];
 };
 
 struct SenseSubsystem_info {

[PATCH 15/19] hpsa: kill annoying messages about SSD Smart Path retries

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

There's nothing the user can or should do about these messages,
the commands are retried down the normal RAID path, and the
messages just flood the logs and sap performance.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |   10 ----------
 1 files changed, 0 insertions(+), 10 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 96436bb..8359884 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -1653,16 +1653,6 @@ static void process_ioaccel2_completion(struct ctlr_info *h,
 	if (is_logical_dev_addr_mode(dev->scsi3addr) &&
 		c2->error_data.serv_response ==
 			IOACCEL2_SERV_RESPONSE_FAILURE) {
-		if (c2->error_data.status ==
-			IOACCEL2_STATUS_SR_IOACCEL_DISABLED)
-			dev_warn(&h->pdev->dev,
-				"%s: Path is unavailable, retrying on standard path.\n",
-				"HP SSD Smart Path");
-		else
-			dev_warn(&h->pdev->dev,
-				"%s: Error 0x%02x, retrying on standard path.\n",
-				"HP SSD Smart Path", c2->error_data.status);
-
 		dev->offload_enabled = 0;
 		h->drv_req_rescan = 1;	/* schedule controller for a rescan */
 		cmd->result = DID_SOFT_ERROR << 16;

[PATCH 16/19] hpsa: fix event filtering to prevent excessive rescans with old firmware

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

CTLR_STATE_CHANGE_EVENT and CTLR_STATE_CHANGE_EVENT_REDUNDANT_CNTRL
do not require rescans to be initiated.  Current firmware filters out
these events already, but some out of date firmware doesn't, so the
driver needs to filter them out too.  Without this change and with out
of date firmware you may see the driver spending a lot of time
scanning devices unnecessarily on some Smart Arrays.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.h |    4 +---
 1 files changed, 1 insertions(+), 3 deletions(-)

diff --git a/drivers/scsi/hpsa.h b/drivers/scsi/hpsa.h
index 1855469..18f5093 100644
--- a/drivers/scsi/hpsa.h
+++ b/drivers/scsi/hpsa.h
@@ -228,11 +228,9 @@ struct ctlr_info {
 #define CTLR_STATE_CHANGE_EVENT_AIO_CONFIG_CHANGE	(1 << 31)
 
 #define RESCAN_REQUIRED_EVENT_BITS \
-		(CTLR_STATE_CHANGE_EVENT | \
-		CTLR_ENCLOSURE_HOT_PLUG_EVENT | \
+		(CTLR_ENCLOSURE_HOT_PLUG_EVENT | \
 		CTLR_STATE_CHANGE_EVENT_PHYSICAL_DRV | \
 		CTLR_STATE_CHANGE_EVENT_LOGICAL_DRV | \
-		CTLR_STATE_CHANGE_EVENT_REDUNDANT_CNTRL | \
 		CTLR_STATE_CHANGE_EVENT_AIO_ENABLED_DISABLED | \
 		CTLR_STATE_CHANGE_EVENT_AIO_CONFIG_CHANGE)
 	spinlock_t offline_device_lock;

[PATCH 17/19] hpsa: remove bad unlikely annotation from device list updating code

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 8359884..9d4a0bd 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -3060,7 +3060,7 @@ static void hpsa_update_scsi_devices(struct ctlr_info *h, int hostno)
 		ndev_allocated++;
 	}
 
-	if (unlikely(is_scsi_rev_5(h)))
+	if (is_scsi_rev_5(h))
 		raid_ctlr_position = 0;
 	else
 		raid_ctlr_position = nphysicals + nlogicals;

[PATCH 18/19] hpsa: report check condition even if no sense data present for ioaccel2 mode

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

It shouldn't happen that we get a check condition with no sense data, but if it
does, we shouldn't just drop the check condition on the floor.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |    7 +++++--
 1 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 9d4a0bd..3446b8d 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -1556,9 +1556,13 @@ static int handle_ioaccel_mode2_error(struct ctlr_info *h,
 			dev_warn(&h->pdev->dev,
 				"%s: task complete with check condition.\n",
 				"HP SSD Smart Path");
+			cmd->result |= SAM_STAT_CHECK_CONDITION;
 			if (c2->error_data.data_present !=
-					IOACCEL2_SENSE_DATA_PRESENT)
+					IOACCEL2_SENSE_DATA_PRESENT) {
+				memset(cmd->sense_buffer, 0,
+					SCSI_SENSE_BUFFERSIZE);
 				break;
+			}
 			/* copy the sense data */
 			data_len = c2->error_data.sense_data_len;
 			if (data_len > SCSI_SENSE_BUFFERSIZE)
@@ -1568,7 +1572,6 @@ static int handle_ioaccel_mode2_error(struct ctlr_info *h,
 					sizeof(c2->error_data.sense_data_buff);
 			memcpy(cmd->sense_buffer,
 				c2->error_data.sense_data_buff, data_len);
-			cmd->result |= SAM_STAT_CHECK_CONDITION;
 			retry = 1;
 			break;
 		case IOACCEL2_STATUS_SR_TASK_COMP_BUSY:

[PATCH 19/19] hpsa: remove messages about volume status VPD inquiry page not supported

[Stephen M. Cameron]

From: Stephen M. Cameron <scameron@beardog.cce.hp.com>

They are annoying and do not help anyone.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
---
 drivers/scsi/hpsa.c |   12 +++---------
 1 files changed, 3 insertions(+), 9 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 3446b8d..2fe8705 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -2516,27 +2516,21 @@ static int hpsa_get_volume_status(struct ctlr_info *h,
 		return HPSA_VPD_LV_STATUS_UNSUPPORTED;
 
 	/* Does controller have VPD for logical volume status? */
-	if (!hpsa_vpd_page_supported(h, scsi3addr, HPSA_VPD_LV_STATUS)) {
-		dev_warn(&h->pdev->dev, "Logical volume status VPD page is unsupported.\n");
+	if (!hpsa_vpd_page_supported(h, scsi3addr, HPSA_VPD_LV_STATUS))
 		goto exit_failed;
-	}
 
 	/* Get the size of the VPD return buffer */
 	rc = hpsa_scsi_do_inquiry(h, scsi3addr, VPD_PAGE | HPSA_VPD_LV_STATUS,
 					buf, HPSA_VPD_HEADER_SZ);
-	if (rc != 0) {
-		dev_warn(&h->pdev->dev, "Logical volume status VPD inquiry failed.\n");
+	if (rc != 0)
 		goto exit_failed;
-	}
 	size = buf[3];
 
 	/* Now get the whole VPD buffer */
 	rc = hpsa_scsi_do_inquiry(h, scsi3addr, VPD_PAGE | HPSA_VPD_LV_STATUS,
 					buf, size + HPSA_VPD_HEADER_SZ);
-	if (rc != 0) {
-		dev_warn(&h->pdev->dev, "Logical volume status VPD inquiry failed.\n");
+	if (rc != 0)
 		goto exit_failed;
-	}
 	status = buf[4]; /* status byte */
 
 	kfree(buf);

Re: [PATCH 10/19] hpsa: set irq affinity hints to route MSI-X vectors across CPUs

[Govindarajulu Varadarajan]

On Thu, 8 May 2014, Stephen M. Cameron wrote:

> From: Stephen M. Cameron <scameron@beardog.cce.hp.com>
>
> Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
> ---
> drivers/scsi/hpsa.c |   17 ++++++++++++++++-
> 1 files changed, 16 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
> index 9c44f26..e8090e2 100644
> --- a/drivers/scsi/hpsa.c
> +++ b/drivers/scsi/hpsa.c
> @@ -6608,6 +6608,17 @@ static void hpsa_free_cmd_pool(struct ctlr_info *h)
> 			h->ioaccel_cmd_pool, h->ioaccel_cmd_pool_dhandle);
> }
>
> +static void hpsa_irq_affinity_hints(struct ctlr_info *h)
> +{
> +	int i, cpu, rc;
> +
> +	cpu = cpumask_first(cpu_online_mask);

May be consider setting cpu affinity hint based on numa locality of pcie device
and cpu? Instead of blindly assigning all online cpu in order.

Some thing like

 	cpu = cpumask_first(cpumask_of_node(numa_node))

> +	for (i = 0; i < h->msix_vector; i++) {
> +		rc = irq_set_affinity_hint(h->intr[i], get_cpu_mask(cpu));

You are not using return value rc. Why declare it?

Thanks
Govind
> +		cpu = cpumask_next(cpu, cpu_online_mask);
> +	}
> +}
> +
> static int hpsa_request_irq(struct ctlr_info *h,
> 	irqreturn_t (*msixhandler)(int, void *),
> 	irqreturn_t (*intxhandler)(int, void *))
> @@ -6627,6 +6638,7 @@ static int hpsa_request_irq(struct ctlr_info *h,
> 			rc = request_irq(h->intr[i], msixhandler,
> 					0, h->devname,
> 					&h->q[i]);
> +		hpsa_irq_affinity_hints(h);
> 	} else {
> 		/* Use single reply pool */
> 		if (h->msix_vector > 0 || h->msi_vector) {
> @@ -6678,12 +6690,15 @@ static void free_irqs(struct ctlr_info *h)
> 	if (!h->msix_vector || h->intr_mode != PERF_MODE_INT) {
> 		/* Single reply queue, only one irq to free */
> 		i = h->intr_mode;
> +		irq_set_affinity_hint(h->intr[i], NULL);
> 		free_irq(h->intr[i], &h->q[i]);
> 		return;
> 	}
>
> -	for (i = 0; i < h->msix_vector; i++)
> +	for (i = 0; i < h->msix_vector; i++) {
> +		irq_set_affinity_hint(h->intr[i], NULL);
> 		free_irq(h->intr[i], &h->q[i]);
> +	}
> }
>
> static void hpsa_free_irqs_and_disable_msix(struct ctlr_info *h)
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

RE: [PATCH 10/19] hpsa: set irq affinity hints to route MSI-X vectors across CPUs

[Elliott, Robert (Server Storage)]

> -----Original Message-----
> From: Govindarajulu Varadarajan [mailto:gvaradar@cisco.com]
> Sent: Thursday, 08 May, 2014 4:12 PM
> To: Stephen M. Cameron
> Cc: james.bottomley@hansenpartnership.com; Lindley, Justin;
> martin.petersen@oracle.com; linux-scsi@vger.kernel.org;
> stephenmcameron@gmail.com; Handzik, Joe; thenzl@redhat.com;
> michael.miller@canonical.com; Elliott, Robert (Server Storage); Teel, Scott
> Stacy
> Subject: Re: [PATCH 10/19] hpsa: set irq affinity hints to route MSI-X
> vectors across CPUs
> 
> 
> 
> On Thu, 8 May 2014, Stephen M. Cameron wrote:
> 
> > From: Stephen M. Cameron <scameron@beardog.cce.hp.com>
> >
> > Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
> > ---
> > drivers/scsi/hpsa.c |   17 ++++++++++++++++-
> > 1 files changed, 16 insertions(+), 1 deletions(-)
> >
> > diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
> > index 9c44f26..e8090e2 100644
> > --- a/drivers/scsi/hpsa.c
> > +++ b/drivers/scsi/hpsa.c
> > @@ -6608,6 +6608,17 @@ static void hpsa_free_cmd_pool(struct ctlr_info *h)
> > 			h->ioaccel_cmd_pool, h->ioaccel_cmd_pool_dhandle);
> > }
> >
> > +static void hpsa_irq_affinity_hints(struct ctlr_info *h)
> > +{
> > +	int i, cpu, rc;
> > +
> > +	cpu = cpumask_first(cpu_online_mask);
> 
> Maybe consider setting cpu affinity hint based on numa locality 
> of pcie device and cpu? Instead of blindly assigning all online 
> cpu in order.
> 
> Something like
> 
>  	cpu = cpumask_first(cpumask_of_node(numa_node))
> 
> > +	for (i = 0; i < h->msix_vector; i++) {
> > +		rc = irq_set_affinity_hint(h->intr[i], get_cpu_mask(cpu));
> 
> You are not using return value rc. Why declare it?
> 
> Thanks
> Govind
> > +		cpu = cpumask_next(cpu, cpu_online_mask);
> > +	}
> > +}
> > +

We think matching the locality of the application is more 
important - delivering the message write carrying the MSI-X
interrupt across the CPU-to-CPU interconnect is less expensive 
than delivering it to the closest CPU and doing hardirq 
processing there, which causes cache traffic between that 
CPU and the CPU used for:
a) submission by the application and driver stack;
b) softirq processing; and
c) completion processing by the application.

That presumes the application thread is not wandering
to a different CPU after submission before completion.
If so, any choice is bound to be wrong sometimes.

If the block layer queue has rq_affinity=2, the softirq is
handled on the submitting CPU core; if rq_affinity=1, it
is handled on any core sharing the L3 cache (i.e., 
the same physical CPU socket).

With hyperthreading, the hyperthreaded sibling of the
submitting CPU core should be equally good at hardirq 
and softirq processing.  Is there an 
architecture-independent call in linux to report 
the CPUs sharing the same L1 (and L2) cache (i.e.,
the siblings)?  cpus_share_cache(), used by the block 
layer for rq_affinity=1, just reports CPUs sharing the 
same last level (L3) cache domain.  That still results 
in L2 traffic between cores inside a socket.

---
Rob Elliott    HP Server Storage

Re: [PATCH 00/19] hpsa driver updates May 8, 2014

[Christoph Hellwig]

The whole series looks good to me,

Reviewed-by: Christoph Hellwig <hch@lst.de>