Merge remote branch 'block/for-next'

20 files changed, 756 insertions, 520 deletions

diff --git a/block/blk-core.c b/block/blk-core.c
index f84cce42fc58..5cb6ca99fd7f 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -2114,7 +2114,8 @@ static bool blk_update_bidi_request(struct request *rq, int error,
 	    blk_update_request(rq->next_rq, error, bidi_bytes))
 		return true;
 
-	add_disk_randomness(rq->rq_disk);
+	if (blk_queue_add_random(rq->q))
+		add_disk_randomness(rq->rq_disk);
 
 	return false;
 }
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 306759bbdf1b..001ab18078f5 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -180,26 +180,36 @@ static ssize_t queue_max_hw_sectors_show(struct request_queue *q, char *page)
 	return queue_var_show(max_hw_sectors_kb, (page));
 }
 
-static ssize_t queue_nonrot_show(struct request_queue *q, char *page)
-{
-	return queue_var_show(!blk_queue_nonrot(q), page);
+#define QUEUE_SYSFS_BIT_FNS(name, flag, neg)				\
+static ssize_t								\
+queue_show_##name(struct request_queue *q, char *page)			\
+{									\
+	int bit;							\
+	bit = test_bit(QUEUE_FLAG_##flag, &q->queue_flags);		\
+	return queue_var_show(neg ? !bit : bit, page);			\
+}									\
+static ssize_t								\
+queue_store_##name(struct request_queue *q, const char *page, size_t count) \
+{									\
+	unsigned long val;						\
+	ssize_t ret;							\
+	ret = queue_var_store(&val, page, count);			\
+	if (neg)							\
+		val = !val;						\
+									\
+	spin_lock_irq(q->queue_lock);					\
+	if (val)							\
+		queue_flag_set(QUEUE_FLAG_##flag, q);			\
+	else								\
+		queue_flag_clear(QUEUE_FLAG_##flag, q);			\
+	spin_unlock_irq(q->queue_lock);					\
+	return ret;							\
 }
 
-static ssize_t queue_nonrot_store(struct request_queue *q, const char *page,
-				  size_t count)
-{
-	unsigned long nm;
-	ssize_t ret = queue_var_store(&nm, page, count);
-
-	spin_lock_irq(q->queue_lock);
-	if (nm)
-		queue_flag_clear(QUEUE_FLAG_NONROT, q);
-	else
-		queue_flag_set(QUEUE_FLAG_NONROT, q);
-	spin_unlock_irq(q->queue_lock);
-
-	return ret;
-}
+QUEUE_SYSFS_BIT_FNS(nonrot, NONROT, 1);
+QUEUE_SYSFS_BIT_FNS(random, ADD_RANDOM, 0);
+QUEUE_SYSFS_BIT_FNS(iostats, IO_STAT, 0);
+#undef QUEUE_SYSFS_BIT_FNS
 
 static ssize_t queue_nomerges_show(struct request_queue *q, char *page)
 {
@@ -250,27 +260,6 @@ queue_rq_affinity_store(struct request_queue *q, const char *page, size_t count)
 	return ret;
 }
 
-static ssize_t queue_iostats_show(struct request_queue *q, char *page)
-{
-	return queue_var_show(blk_queue_io_stat(q), page);
-}
-
-static ssize_t queue_iostats_store(struct request_queue *q, const char *page,
-				   size_t count)
-{
-	unsigned long stats;
-	ssize_t ret = queue_var_store(&stats, page, count);
-
-	spin_lock_irq(q->queue_lock);
-	if (stats)
-		queue_flag_set(QUEUE_FLAG_IO_STAT, q);
-	else
-		queue_flag_clear(QUEUE_FLAG_IO_STAT, q);
-	spin_unlock_irq(q->queue_lock);
-
-	return ret;
-}
-
 static struct queue_sysfs_entry queue_requests_entry = {
 	.attr = {.name = "nr_requests", .mode = S_IRUGO | S_IWUSR },
 	.show = queue_requests_show,
@@ -352,8 +341,8 @@ static struct queue_sysfs_entry queue_discard_zeroes_data_entry = {
 
 static struct queue_sysfs_entry queue_nonrot_entry = {
 	.attr = {.name = "rotational", .mode = S_IRUGO | S_IWUSR },
-	.show = queue_nonrot_show,
-	.store = queue_nonrot_store,
+	.show = queue_show_nonrot,
+	.store = queue_store_nonrot,
 };
 
 static struct queue_sysfs_entry queue_nomerges_entry = {
@@ -370,8 +359,14 @@ static struct queue_sysfs_entry queue_rq_affinity_entry = {
 
 static struct queue_sysfs_entry queue_iostats_entry = {
 	.attr = {.name = "iostats", .mode = S_IRUGO | S_IWUSR },
-	.show = queue_iostats_show,
-	.store = queue_iostats_store,
+	.show = queue_show_iostats,
+	.store = queue_store_iostats,
+};
+
+static struct queue_sysfs_entry queue_random_entry = {
+	.attr = {.name = "add_random", .mode = S_IRUGO | S_IWUSR },
+	.show = queue_show_random,
+	.store = queue_store_random,
 };
 
 static struct attribute *default_attrs[] = {
@@ -394,6 +389,7 @@ static struct attribute *default_attrs[] = {
 	&queue_nomerges_entry.attr,
 	&queue_rq_affinity_entry.attr,
 	&queue_iostats_entry.attr,
+	&queue_random_entry.attr,
 	NULL,
 };
 
diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c
index 51ceaee98f9f..10a0268a1f92 100644
--- a/drivers/block/cciss.c
+++ b/drivers/block/cciss.c
@@ -56,16 +56,14 @@
 #include <linux/kthread.h>
 
 #define CCISS_DRIVER_VERSION(maj,min,submin) ((maj<<16)|(min<<8)|(submin))
-#define DRIVER_NAME "HP CISS Driver (v 3.6.20)"
-#define DRIVER_VERSION CCISS_DRIVER_VERSION(3, 6, 20)
+#define DRIVER_NAME "HP CISS Driver (v 3.6.26)"
+#define DRIVER_VERSION CCISS_DRIVER_VERSION(3, 6, 26)
 
 /* Embedded module documentation macros - see modules.h */
 MODULE_AUTHOR("Hewlett-Packard Company");
 MODULE_DESCRIPTION("Driver for HP Smart Array Controllers");
-MODULE_SUPPORTED_DEVICE("HP SA5i SA5i+ SA532 SA5300 SA5312 SA641 SA642 SA6400"
-			" SA6i P600 P800 P400 P400i E200 E200i E500 P700m"
-			" Smart Array G2 Series SAS/SATA Controllers");
-MODULE_VERSION("3.6.20");
+MODULE_SUPPORTED_DEVICE("HP Smart Array Controllers");
+MODULE_VERSION("3.6.26");
 MODULE_LICENSE("GPL");
 
 static int cciss_allow_hpsa;
@@ -107,6 +105,11 @@ static const struct pci_device_id cciss_pci_device_id[] = {
 	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3249},
 	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x324A},
 	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x324B},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3250},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3251},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3252},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3253},
+	{PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3254},
 	{0,}
 };
 
@@ -146,6 +149,11 @@ static struct board_type products[] = {
 	{0x3249103C, "Smart Array P812", &SA5_access},
 	{0x324A103C, "Smart Array P712m", &SA5_access},
 	{0x324B103C, "Smart Array P711m", &SA5_access},
+	{0x3250103C, "Smart Array", &SA5_access},
+	{0x3251103C, "Smart Array", &SA5_access},
+	{0x3252103C, "Smart Array", &SA5_access},
+	{0x3253103C, "Smart Array", &SA5_access},
+	{0x3254103C, "Smart Array", &SA5_access},
 };
 
 /* How long to wait (in milliseconds) for board to go into simple mode */
@@ -167,7 +175,8 @@ static DEFINE_MUTEX(scan_mutex);
 static LIST_HEAD(scan_q);
 
 static void do_cciss_request(struct request_queue *q);
-static irqreturn_t do_cciss_intr(int irq, void *dev_id);
+static irqreturn_t do_cciss_intx(int irq, void *dev_id);
+static irqreturn_t do_cciss_msix_intr(int irq, void *dev_id);
 static int cciss_open(struct block_device *bdev, fmode_t mode);
 static int cciss_release(struct gendisk *disk, fmode_t mode);
 static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
@@ -197,7 +206,6 @@ static int sendcmd_withirq_core(ctlr_info_t *h, CommandList_struct *c,
 	int attempt_retry);
 static int process_sendcmd_error(ctlr_info_t *h, CommandList_struct *c);
 
-static void fail_all_cmds(unsigned long ctlr);
 static int add_to_scan_list(struct ctlr_info *h);
 static int scan_thread(void *data);
 static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c);
@@ -205,6 +213,12 @@ static void cciss_hba_release(struct device *dev);
 static void cciss_device_release(struct device *dev);
 static void cciss_free_gendisk(ctlr_info_t *h, int drv_index);
 static void cciss_free_drive_info(ctlr_info_t *h, int drv_index);
+static inline u32 next_command(ctlr_info_t *h);
+
+/* performant mode helper functions */
+static void  calc_bucket_map(int *bucket, int num_buckets, int nsgs,
+				int *bucket_map);
+static void cciss_put_controller_into_performant_mode(ctlr_info_t *h);
 
 #ifdef CONFIG_PROC_FS
 static void cciss_procinit(int i);
@@ -231,6 +245,16 @@ static const struct block_device_operations cciss_fops = {
 	.revalidate_disk = cciss_revalidate,
 };
 
+/* set_performant_mode: Modify the tag for cciss performant
+ * set bit 0 for pull model, bits 3-1 for block fetch
+ * register number
+ */
+static void set_performant_mode(ctlr_info_t *h, CommandList_struct *c)
+{
+	if (likely(h->transMethod == CFGTBL_Trans_Performant))
+		c->busaddr |= 1 | (h->blockFetchTable[c->Header.SGList] << 1);
+}
+
 /*
  * Enqueuing and dequeuing functions for cmdlists.
  */
@@ -257,6 +281,18 @@ static inline void removeQ(CommandList_struct *c)
 	hlist_del_init(&c->list);
 }
 
+static void enqueue_cmd_and_start_io(ctlr_info_t *h,
+	CommandList_struct *c)
+{
+	unsigned long flags;
+	set_performant_mode(h, c);
+	spin_lock_irqsave(&h->lock, flags);
+	addQ(&h->reqQ, c);
+	h->Qdepth++;
+	start_io(h);
+	spin_unlock_irqrestore(&h->lock, flags);
+}
+
 static void cciss_free_sg_chain_blocks(SGDescriptor_struct **cmd_sg_list,
 	int nr_cmds)
 {
@@ -366,7 +402,7 @@ static void cciss_seq_show_header(struct seq_file *seq)
 		h->product_name,
 		(unsigned long)h->board_id,
 		h->firm_ver[0], h->firm_ver[1], h->firm_ver[2],
-		h->firm_ver[3], (unsigned int)h->intr[SIMPLE_MODE_INT],
+		h->firm_ver[3], (unsigned int)h->intr[PERF_MODE_INT],
 		h->num_luns,
 		h->Qdepth, h->commands_outstanding,
 		h->maxQsinceinit, h->max_outstanding, h->maxSG);
@@ -1377,7 +1413,6 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
 			CommandList_struct *c;
 			char *buff = NULL;
 			u64bit temp64;
-			unsigned long flags;
 			DECLARE_COMPLETION_ONSTACK(wait);
 
 			if (!arg)
@@ -1449,13 +1484,7 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
 			}
 			c->waiting = &wait;
 
-			/* Put the request on the tail of the request queue */
-			spin_lock_irqsave(CCISS_LOCK(ctlr), flags);
-			addQ(&host->reqQ, c);
-			host->Qdepth++;
-			start_io(host);
-			spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);
-
+			enqueue_cmd_and_start_io(host, c);
 			wait_for_completion(&wait);
 
 			/* unlock the buffers from DMA */
@@ -1495,7 +1524,6 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
 			unsigned char **buff = NULL;
 			int *buff_size = NULL;
 			u64bit temp64;
-			unsigned long flags;
 			BYTE sg_used = 0;
 			int status = 0;
 			int i;
@@ -1602,12 +1630,7 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
 				}
 			}
 			c->waiting = &wait;
-			/* Put the request on the tail of the request queue */
-			spin_lock_irqsave(CCISS_LOCK(ctlr), flags);
-			addQ(&host->reqQ, c);
-			host->Qdepth++;
-			start_io(host);
-			spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);
+			enqueue_cmd_and_start_io(host, c);
 			wait_for_completion(&wait);
 			/* unlock the buffers from DMA */
 			for (i = 0; i < sg_used; i++) {
@@ -1729,8 +1752,8 @@ static void cciss_softirq_done(struct request *rq)
 	CommandList_struct *cmd = rq->completion_data;
 	ctlr_info_t *h = hba[cmd->ctlr];
 	SGDescriptor_struct *curr_sg = cmd->SG;
-	unsigned long flags;
 	u64bit temp64;
+	unsigned long flags;
 	int i, ddir;
 	int sg_index = 0;
 
@@ -2679,17 +2702,11 @@ static int sendcmd_withirq_core(ctlr_info_t *h, CommandList_struct *c,
 {
 	DECLARE_COMPLETION_ONSTACK(wait);
 	u64bit buff_dma_handle;
-	unsigned long flags;
 	int return_status = IO_OK;
 
 resend_cmd2:
 	c->waiting = &wait;
-	/* Put the request on the tail of the queue and send it */
-	spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags);
-	addQ(&h->reqQ, c);
-	h->Qdepth++;
-	start_io(h);
-	spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
+	enqueue_cmd_and_start_io(h, c);
 
 	wait_for_completion(&wait);
 
@@ -3132,6 +3149,34 @@ after_error_processing:
 	blk_complete_request(cmd->rq);
 }
 
+static inline u32 cciss_tag_contains_index(u32 tag)
+{
+#define DIRECT_LOOKUP_BIT 0x10
+	return tag & DIRECT_LOOKUP_BIT;
+}
+
+static inline u32 cciss_tag_to_index(u32 tag)
+{
+#define DIRECT_LOOKUP_SHIFT 5
+	return tag >> DIRECT_LOOKUP_SHIFT;
+}
+
+static inline u32 cciss_tag_discard_error_bits(u32 tag)
+{
+#define CCISS_ERROR_BITS 0x03
+	return tag & ~CCISS_ERROR_BITS;
+}
+
+static inline void cciss_mark_tag_indexed(u32 *tag)
+{
+	*tag |= DIRECT_LOOKUP_BIT;
+}
+
+static inline void cciss_set_tag_index(u32 *tag, u32 index)
+{
+	*tag |= (index << DIRECT_LOOKUP_SHIFT);
+}
+
 /*
  * Get a request and submit it to the controller.
  */
@@ -3180,8 +3225,8 @@ static void do_cciss_request(struct request_queue *q)
 	/* got command from pool, so use the command block index instead */
 	/* for direct lookups. */
 	/* The first 2 bits are reserved for controller error reporting. */
-	c->Header.Tag.lower = (c->cmdindex << 3);
-	c->Header.Tag.lower |= 0x04;	/* flag for direct lookup. */
+	cciss_set_tag_index(&c->Header.Tag.lower, c->cmdindex);
+	cciss_mark_tag_indexed(&c->Header.Tag.lower);
 	memcpy(&c->Header.LUN, drv->LunID, sizeof(drv->LunID));
 	c->Request.CDBLen = 10;	/* 12 byte commands not in FW yet; */
 	c->Request.Type.Type = TYPE_CMD;	/* It is a command. */
@@ -3242,9 +3287,12 @@ static void do_cciss_request(struct request_queue *q)
 			blk_rq_sectors(creq), seg, chained);
 #endif				/* CCISS_DEBUG */
 
-	c->Header.SGList = c->Header.SGTotal = seg + chained;
-	if (seg > h->max_cmd_sgentries)
+	c->Header.SGTotal = seg + chained;
+	if (seg <= h->max_cmd_sgentries)
+		c->Header.SGList = c->Header.SGTotal;
+	else
 		c->Header.SGList = h->max_cmd_sgentries;
+	set_performant_mode(h, c);
 
 	if (likely(blk_fs_request(creq))) {
 		if(h->cciss_read == CCISS_READ_10) {
@@ -3313,16 +3361,97 @@ static inline int interrupt_pending(ctlr_info_t *h)
 
 static inline long interrupt_not_for_us(ctlr_info_t *h)
 {
-	return (((h->access.intr_pending(h) == 0) ||
-		 (h->interrupts_enabled == 0)));
+	return !(h->msi_vector || h->msix_vector) &&
+		((h->access.intr_pending(h) == 0) ||
+		(h->interrupts_enabled == 0));
 }
 
-static irqreturn_t do_cciss_intr(int irq, void *dev_id)
+static inline int bad_tag(ctlr_info_t *h, u32 tag_index,
+			u32 raw_tag)
 {
-	ctlr_info_t *h = dev_id;
+	if (unlikely(tag_index >= h->nr_cmds)) {
+		dev_warn(&h->pdev->dev, "bad tag 0x%08x ignored.\n", raw_tag);
+		return 1;
+	}
+	return 0;
+}
+
+static inline void finish_cmd(ctlr_info_t *h, CommandList_struct *c,
+				u32 raw_tag)
+{
+	removeQ(c);
+	if (likely(c->cmd_type == CMD_RWREQ))
+		complete_command(h, c, 0);
+	else if (c->cmd_type == CMD_IOCTL_PEND)
+		complete(c->waiting);
+#ifdef CONFIG_CISS_SCSI_TAPE
+	else if (c->cmd_type == CMD_SCSI)
+		complete_scsi_command(c, 0, raw_tag);
+#endif
+}
+
+static inline u32 next_command(ctlr_info_t *h)
+{
+	u32 a;
+
+	if (unlikely(h->transMethod != CFGTBL_Trans_Performant))
+		return h->access.command_completed(h);
+
+	if ((*(h->reply_pool_head) & 1) == (h->reply_pool_wraparound)) {
+		a = *(h->reply_pool_head); /* Next cmd in ring buffer */
+		(h->reply_pool_head)++;
+		h->commands_outstanding--;
+	} else {
+		a = FIFO_EMPTY;
+	}
+	/* Check for wraparound */
+	if (h->reply_pool_head == (h->reply_pool + h->max_commands)) {
+		h->reply_pool_head = h->reply_pool;
+		h->reply_pool_wraparound ^= 1;
+	}
+	return a;
+}
+
+/* process completion of an indexed ("direct lookup") command */
+static inline u32 process_indexed_cmd(ctlr_info_t *h, u32 raw_tag)
+{
+	u32 tag_index;
 	CommandList_struct *c;
+
+	tag_index = cciss_tag_to_index(raw_tag);
+	if (bad_tag(h, tag_index, raw_tag))
+		return next_command(h);
+	c = h->cmd_pool + tag_index;
+	finish_cmd(h, c, raw_tag);
+	return next_command(h);
+}
+
+/* process completion of a non-indexed command */
+static inline u32 process_nonindexed_cmd(ctlr_info_t *h, u32 raw_tag)
+{
+	u32 tag;
+	CommandList_struct *c = NULL;
+	struct hlist_node *tmp;
+	__u32 busaddr_masked, tag_masked;
+
+	tag = cciss_tag_discard_error_bits(raw_tag);
+	hlist_for_each_entry(c, tmp, &h->cmpQ, list) {
+		busaddr_masked = cciss_tag_discard_error_bits(c->busaddr);
+		tag_masked = cciss_tag_discard_error_bits(tag);
+		if (busaddr_masked == tag_masked) {
+			finish_cmd(h, c, raw_tag);
+			return next_command(h);
+		}
+	}
+	bad_tag(h, h->nr_cmds + 1, raw_tag);
+	return next_command(h);
+}
+
+static irqreturn_t do_cciss_intx(int irq, void *dev_id)
+{
+	ctlr_info_t *h = dev_id;
 	unsigned long flags;
-	__u32 a, a1, a2;
+	u32 raw_tag;
 
 	if (interrupt_not_for_us(h))
 		return IRQ_NONE;
@@ -3332,50 +3461,41 @@ static irqreturn_t do_cciss_intr(int irq, void *dev_id)
 	 */
 	spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags);
 	while (interrupt_pending(h)) {
-		while ((a = get_next_completion(h)) != FIFO_EMPTY) {
-			a1 = a;
-			if ((a & 0x04)) {
-				a2 = (a >> 3);
-				if (a2 >= h->nr_cmds) {
-					printk(KERN_WARNING
-					       "cciss: controller cciss%d failed, stopping.\n",
-					       h->ctlr);
-					spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
-					fail_all_cmds(h->ctlr);
-					return IRQ_HANDLED;
-				}
+		raw_tag = get_next_completion(h);
+		while (raw_tag != FIFO_EMPTY) {
+			if (cciss_tag_contains_index(raw_tag))
+				raw_tag = process_indexed_cmd(h, raw_tag);
+			else
+				raw_tag = process_nonindexed_cmd(h, raw_tag);
+		}
+	}
 
-				c = h->cmd_pool + a2;
-				a = c->busaddr;
+	spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
+	return IRQ_HANDLED;
+}
 
-			} else {
-				struct hlist_node *tmp;
+/* Add a second interrupt handler for MSI/MSI-X mode. In this mode we never
+ * check the interrupt pending register because it is not set.
+ */
+static irqreturn_t do_cciss_msix_intr(int irq, void *dev_id)
+{
+	ctlr_info_t *h = dev_id;
+	unsigned long flags;
+	u32 raw_tag;
 
-				a &= ~3;
-				c = NULL;
-				hlist_for_each_entry(c, tmp, &h->cmpQ, list) {
-					if (c->busaddr == a)
-						break;
-				}
-			}
-			/*
-			 * If we've found the command, take it off the
-			 * completion Q and free it
-			 */
-			if (c && c->busaddr == a) {
-				removeQ(c);
-				if (c->cmd_type == CMD_RWREQ) {
-					complete_command(h, c, 0);
-				} else if (c->cmd_type == CMD_IOCTL_PEND) {
-					complete(c->waiting);
-				}
-#				ifdef CONFIG_CISS_SCSI_TAPE
-				else if (c->cmd_type == CMD_SCSI)
-					complete_scsi_command(c, 0, a1);
-#				endif
-				continue;
-			}
-		}
+	if (interrupt_not_for_us(h))
+		return IRQ_NONE;
+	/*
+	 * If there are completed commands in the completion queue,
+	 * we had better do something about it.
+	 */
+	spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags);
+	raw_tag = get_next_completion(h);
+	while (raw_tag != FIFO_EMPTY) {
+		if (cciss_tag_contains_index(raw_tag))
+			raw_tag = process_indexed_cmd(h, raw_tag);
+		else
+			raw_tag = process_nonindexed_cmd(h, raw_tag);
 	}
 
 	spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
@@ -3630,6 +3750,155 @@ static int find_PCI_BAR_index(struct pci_dev *pdev, unsigned long pci_bar_addr)
 	return -1;
 }
 
+/* Fill in bucket_map[], given nsgs (the max number of
+ * scatter gather elements supported) and bucket[],
+ * which is an array of 8 integers.  The bucket[] array
+ * contains 8 different DMA transfer sizes (in 16
+ * byte increments) which the controller uses to fetch
+ * commands.  This function fills in bucket_map[], which
+ * maps a given number of scatter gather elements to one of
+ * the 8 DMA transfer sizes.  The point of it is to allow the
+ * controller to only do as much DMA as needed to fetch the
+ * command, with the DMA transfer size encoded in the lower
+ * bits of the command address.
+ */
+static void  calc_bucket_map(int bucket[], int num_buckets,
+	int nsgs, int *bucket_map)
+{
+	int i, j, b, size;
+
+	/* even a command with 0 SGs requires 4 blocks */
+#define MINIMUM_TRANSFER_BLOCKS 4
+#define NUM_BUCKETS 8
+	/* Note, bucket_map must have nsgs+1 entries. */
+	for (i = 0; i <= nsgs; i++) {
+		/* Compute size of a command with i SG entries */
+		size = i + MINIMUM_TRANSFER_BLOCKS;
+		b = num_buckets; /* Assume the biggest bucket */
+		/* Find the bucket that is just big enough */
+		for (j = 0; j < 8; j++) {
+			if (bucket[j] >= size) {
+				b = j;
+				break;
+			}
+		}
+		/* for a command with i SG entries, use bucket b. */
+		bucket_map[i] = b;
+	}
+}
+
+static void
+cciss_put_controller_into_performant_mode(ctlr_info_t *h)
+{
+	int l = 0;
+	__u32 trans_support;
+	__u32 trans_offset;
+			/*
+			 *  5 = 1 s/g entry or 4k
+			 *  6 = 2 s/g entry or 8k
+			 *  8 = 4 s/g entry or 16k
+			 * 10 = 6 s/g entry or 24k
+			 */
+	int bft[8] = { 5, 6, 8, 10, 12, 20, 28, MAXSGENTRIES + 4};
+	unsigned long register_value;
+
+	BUILD_BUG_ON(28 > MAXSGENTRIES + 4);
+
+	/* Attempt to put controller into performant mode if supported */
+	/* Does board support performant mode? */
+	trans_support = readl(&(h->cfgtable->TransportSupport));
+	if (!(trans_support & PERFORMANT_MODE))
+		return;
+
+	printk(KERN_WARNING "cciss%d: Placing controller into "
+				"performant mode\n", h->ctlr);
+	/* Performant mode demands commands on a 32 byte boundary
+	 * pci_alloc_consistent aligns on page boundarys already.
+	 * Just need to check if divisible by 32
+	 */
+	if ((sizeof(CommandList_struct) % 32) != 0) {
+		printk(KERN_WARNING "%s %d %s\n",
+			"cciss info: command size[",
+			(int)sizeof(CommandList_struct),
+			"] not divisible by 32, no performant mode..\n");
+		return;
+	}
+
+	/* Performant mode ring buffer and supporting data structures */
+	h->reply_pool = (__u64 *)pci_alloc_consistent(
+		h->pdev, h->max_commands * sizeof(__u64),
+		&(h->reply_pool_dhandle));
+
+	/* Need a block fetch table for performant mode */
+	h->blockFetchTable = kmalloc(((h->maxsgentries+1) *
+		sizeof(__u32)), GFP_KERNEL);
+
+	if ((h->reply_pool == NULL) || (h->blockFetchTable == NULL))
+		goto clean_up;
+
+	h->reply_pool_wraparound = 1; /* spec: init to 1 */
+
+	/* Controller spec: zero out this buffer. */
+	memset(h->reply_pool, 0, h->max_commands * sizeof(__u64));
+	h->reply_pool_head = h->reply_pool;
+
+	trans_offset = readl(&(h->cfgtable->TransMethodOffset));
+	calc_bucket_map(bft, ARRAY_SIZE(bft), h->maxsgentries,
+				h->blockFetchTable);
+	writel(bft[0], &h->transtable->BlockFetch0);
+	writel(bft[1], &h->transtable->BlockFetch1);
+	writel(bft[2], &h->transtable->BlockFetch2);
+	writel(bft[3], &h->transtable->BlockFetch3);
+	writel(bft[4], &h->transtable->BlockFetch4);
+	writel(bft[5], &h->transtable->BlockFetch5);
+	writel(bft[6], &h->transtable->BlockFetch6);
+	writel(bft[7], &h->transtable->BlockFetch7);
+
+	/* size of controller ring buffer */
+	writel(h->max_commands, &h->transtable->RepQSize);
+	writel(1, &h->transtable->RepQCount);
+	writel(0, &h->transtable->RepQCtrAddrLow32);
+	writel(0, &h->transtable->RepQCtrAddrHigh32);
+	writel(h->reply_pool_dhandle, &h->transtable->RepQAddr0Low32);
+	writel(0, &h->transtable->RepQAddr0High32);
+	writel(CFGTBL_Trans_Performant,
+			&(h->cfgtable->HostWrite.TransportRequest));
+
+	h->transMethod = CFGTBL_Trans_Performant;
+	writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL);
+	/* under certain very rare conditions, this can take awhile.
+	 * (e.g.: hot replace a failed 144GB drive in a RAID 5 set right
+	 * as we enter this code.) */
+	for (l = 0; l < MAX_CONFIG_WAIT; l++) {
+		register_value = readl(h->vaddr + SA5_DOORBELL);
+		if (!(register_value & CFGTBL_ChangeReq))
+			break;
+		/* delay and try again */
+		set_current_state(TASK_INTERRUPTIBLE);
+		schedule_timeout(10);
+	}
+	register_value = readl(&(h->cfgtable->TransportActive));
+	if (!(register_value & CFGTBL_Trans_Performant)) {
+		printk(KERN_WARNING "cciss: unable to get board into"
+					" performant mode\n");
+		return;
+	}
+
+	/* Change the access methods to the performant access methods */
+	h->access = SA5_performant_access;
+
+	return;
+clean_up:
+	kfree(h->blockFetchTable);
+	if (h->reply_pool)
+		pci_free_consistent(h->pdev,
+				h->max_commands * sizeof(__u64),
+				h->reply_pool,
+				h->reply_pool_dhandle);
+	return;
+
+} /* cciss_put_controller_into_performant_mode */
+
 /* If MSI/MSI-X is supported by the kernel we will try to enable it on
  * controllers that are capable. If not, we use IO-APIC mode.
  */
@@ -3679,7 +3948,7 @@ static void __devinit cciss_interrupt_mode(ctlr_info_t *c,
 default_int_mode:
 #endif				/* CONFIG_PCI_MSI */
 	/* if we get here we're going to use the default interrupt mode */
-	c->intr[SIMPLE_MODE_INT] = pdev->irq;
+	c->intr[PERF_MODE_INT] = pdev->irq;
 	return;
 }
 
@@ -3691,6 +3960,7 @@ static int __devinit cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
 	__u32 cfg_base_addr;
 	__u64 cfg_base_addr_index;
 	int i, prod_index, err;
+	__u32 trans_offset;
 
 	subsystem_vendor_id = pdev->subsystem_vendor;
 	subsystem_device_id = pdev->subsystem_device;
@@ -3804,11 +4074,16 @@ static int __devinit cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
 	c->cfgtable = remap_pci_mem(pci_resource_start(pdev,
 						       cfg_base_addr_index) +
 				    cfg_offset, sizeof(CfgTable_struct));
+	/* Find performant mode table. */
+	trans_offset = readl(&(c->cfgtable->TransMethodOffset));
+	c->transtable = remap_pci_mem(pci_resource_start(pdev,
+		cfg_base_addr_index) + cfg_offset+trans_offset,
+		sizeof(*c->transtable));
 	c->board_id = board_id;
 
-#ifdef CCISS_DEBUG
-	print_cfg_table(c->cfgtable);
-#endif				/* CCISS_DEBUG */
+	#ifdef CCISS_DEBUG
+		print_cfg_table(c->cfgtable);
+	#endif				/* CCISS_DEBUG */
 
 	/* Some controllers support Zero Memory Raid (ZMR).
 	 * When configured in ZMR mode the number of supported
@@ -3818,7 +4093,7 @@ static int __devinit cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
 	 * are supported on the controller then subtract 4 to
 	 * leave a little room for ioctl calls.
 	 */
-	c->max_commands = readl(&(c->cfgtable->CmdsOutMax));
+	c->max_commands = readl(&(c->cfgtable->MaxPerformantModeCommands));
 	c->maxsgentries = readl(&(c->cfgtable->MaxSGElements));
 
 	/*
@@ -3863,7 +4138,7 @@ static int __devinit cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
 	 * kernels revealed a bug in the refetch if dom0 resides on a P600.
 	 */
 	if(board_id == 0x3225103C) {
-		__u32 dma_prefetch;
+			__u32 dma_prefetch;
 		__u32 dma_refetch;
 		dma_prefetch = readl(c->vaddr + I2O_DMA1_CFG);
 		dma_prefetch |= 0x8000;
@@ -3874,38 +4149,9 @@ static int __devinit cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
 	}
 
 #ifdef CCISS_DEBUG
-	printk("Trying to put board into Simple mode\n");
-#endif				/* CCISS_DEBUG */
-	c->max_commands = readl(&(c->cfgtable->CmdsOutMax));
-	/* Update the field, and then ring the doorbell */
-	writel(CFGTBL_Trans_Simple, &(c->cfgtable->HostWrite.TransportRequest));
-	writel(CFGTBL_ChangeReq, c->vaddr + SA5_DOORBELL);
-
-	/* under certain very rare conditions, this can take awhile.
-	 * (e.g.: hot replace a failed 144GB drive in a RAID 5 set right
-	 * as we enter this code.) */
-	for (i = 0; i < MAX_CONFIG_WAIT; i++) {
-		if (!(readl(c->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq))
-			break;
-		/* delay and try again */
-		set_current_state(TASK_INTERRUPTIBLE);
-		schedule_timeout(msecs_to_jiffies(1));
-	}
-
-#ifdef CCISS_DEBUG
-	printk(KERN_DEBUG "I counter got to %d %x\n", i,
-	       readl(c->vaddr + SA5_DOORBELL));
-#endif				/* CCISS_DEBUG */
-#ifdef CCISS_DEBUG
-	print_cfg_table(c->cfgtable);
+	printk(KERN_WARNING "Trying to put board into Performant mode\n");
 #endif				/* CCISS_DEBUG */
-
-	if (!(readl(&(c->cfgtable->TransportActive)) & CFGTBL_Trans_Simple)) {
-		printk(KERN_WARNING "cciss: unable to get board into"
-		       " simple mode\n");
-		err = -ENODEV;
-		goto err_out_free_res;
-	}
+	cciss_put_controller_into_performant_mode(c);
 	return 0;
 
 err_out_free_res:
@@ -4190,7 +4436,6 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
 	i = alloc_cciss_hba();
 	if (i < 0)
 		return -1;
-
 	hba[i]->busy_initializing = 1;
 	INIT_HLIST_HEAD(&hba[i]->cmpQ);
 	INIT_HLIST_HEAD(&hba[i]->reqQ);
@@ -4238,16 +4483,26 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
 
 	/* make sure the board interrupts are off */
 	hba[i]->access.set_intr_mask(hba[i], CCISS_INTR_OFF);
-	if (request_irq(hba[i]->intr[SIMPLE_MODE_INT], do_cciss_intr,
-			IRQF_DISABLED | IRQF_SHARED, hba[i]->devname, hba[i])) {
-		printk(KERN_ERR "cciss: Unable to get irq %d for %s\n",
-		       hba[i]->intr[SIMPLE_MODE_INT], hba[i]->devname);
-		goto clean2;
+	if (hba[i]->msi_vector || hba[i]->msix_vector) {
+		if (request_irq(hba[i]->intr[PERF_MODE_INT],
+				do_cciss_msix_intr,
+				IRQF_DISABLED, hba[i]->devname, hba[i])) {
+			printk(KERN_ERR "cciss: Unable to get irq %d for %s\n",
+			       hba[i]->intr[PERF_MODE_INT], hba[i]->devname);
+			goto clean2;
+		}
+	} else {
+		if (request_irq(hba[i]->intr[PERF_MODE_INT], do_cciss_intx,
+				IRQF_DISABLED, hba[i]->devname, hba[i])) {
+			printk(KERN_ERR "cciss: Unable to get irq %d for %s\n",
+			       hba[i]->intr[PERF_MODE_INT], hba[i]->devname);
+			goto clean2;
+		}
 	}
 
 	printk(KERN_INFO "%s: <0x%x> at PCI %s IRQ %d%s using DAC\n",
 	       hba[i]->devname, pdev->device, pci_name(pdev),
-	       hba[i]->intr[SIMPLE_MODE_INT], dac ? "" : " not");
+	       hba[i]->intr[PERF_MODE_INT], dac ? "" : " not");
 
 	hba[i]->cmd_pool_bits =
 	    kmalloc(DIV_ROUND_UP(hba[i]->nr_cmds, BITS_PER_LONG)
@@ -4353,7 +4608,7 @@ clean4:
 				    hba[i]->nr_cmds * sizeof(ErrorInfo_struct),
 				    hba[i]->errinfo_pool,
 				    hba[i]->errinfo_pool_dhandle);
-	free_irq(hba[i]->intr[SIMPLE_MODE_INT], hba[i]);
+	free_irq(hba[i]->intr[PERF_MODE_INT], hba[i]);
 clean2:
 	unregister_blkdev(hba[i]->major, hba[i]->devname);
 clean1:
@@ -4395,7 +4650,7 @@ static void cciss_shutdown(struct pci_dev *pdev)
 		printk(KERN_WARNING "cciss%d: Error flushing cache\n",
 			h->ctlr);
 	h->access.set_intr_mask(h, CCISS_INTR_OFF);
-	free_irq(h->intr[2], h);
+	free_irq(h->intr[PERF_MODE_INT], h);
 }
 
 static void __devexit cciss_remove_one(struct pci_dev *pdev)
@@ -4495,7 +4750,6 @@ static int __init cciss_init(void)
 	 * array of them, the size must be a multiple of 8 bytes.
 	 */
 	BUILD_BUG_ON(sizeof(CommandList_struct) % COMMANDLIST_ALIGNMENT);
-
 	printk(KERN_INFO DRIVER_NAME "\n");
 
 	err = bus_register(&cciss_bus_type);
@@ -4542,46 +4796,5 @@ static void __exit cciss_cleanup(void)
 	bus_unregister(&cciss_bus_type);
 }
 
-static void fail_all_cmds(unsigned long ctlr)
-{
-	/* If we get here, the board is apparently dead. */
-	ctlr_info_t *h = hba[ctlr];
-	CommandList_struct *c;
-	unsigned long flags;
-
-	printk(KERN_WARNING "cciss%d: controller not responding.\n", h->ctlr);
-	h->alive = 0;		/* the controller apparently died... */
-
-	spin_lock_irqsave(CCISS_LOCK(ctlr), flags);
-
-	pci_disable_device(h->pdev);	/* Make sure it is really dead. */
-
-	/* move everything off the request queue onto the completed queue */
-	while (!hlist_empty(&h->reqQ)) {
-		c = hlist_entry(h->reqQ.first, CommandList_struct, list);
-		removeQ(c);
-		h->Qdepth--;
-		addQ(&h->cmpQ, c);
-	}
-
-	/* Now, fail everything on the completed queue with a HW error */
-	while (!hlist_empty(&h->cmpQ)) {
-		c = hlist_entry(h->cmpQ.first, CommandList_struct, list);
-		removeQ(c);
-		if (c->cmd_type != CMD_MSG_STALE)
-			c->err_info->CommandStatus = CMD_HARDWARE_ERR;
-		if (c->cmd_type == CMD_RWREQ) {
-			complete_command(h, c, 0);
-		} else if (c->cmd_type == CMD_IOCTL_PEND)
-			complete(c->waiting);
-#ifdef CONFIG_CISS_SCSI_TAPE
-		else if (c->cmd_type == CMD_SCSI)
-			complete_scsi_command(c, 0, 0);
-#endif
-	}
-	spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);
-	return;
-}
-
 module_init(cciss_init);
 module_exit(cciss_cleanup);
diff --git a/drivers/block/cciss.h b/drivers/block/cciss.h
index c5d411174db0..8a9f5b58daa8 100644
--- a/drivers/block/cciss.h
+++ b/drivers/block/cciss.h
@@ -25,7 +25,7 @@ struct access_method {
 	void (*submit_command)(ctlr_info_t *h, CommandList_struct *c);
 	void (*set_intr_mask)(ctlr_info_t *h, unsigned long val);
 	unsigned long (*fifo_full)(ctlr_info_t *h);
-	unsigned long (*intr_pending)(ctlr_info_t *h);
+	bool (*intr_pending)(ctlr_info_t *h);
 	unsigned long (*command_completed)(ctlr_info_t *h);
 };
 typedef struct _drive_info_struct
@@ -85,8 +85,8 @@ struct ctlr_info
 	int	max_cmd_sgentries;
 	SGDescriptor_struct **cmd_sg_list;
 
-#	define DOORBELL_INT	0
-#	define PERF_MODE_INT	1
+#	define PERF_MODE_INT	0
+#	define DOORBELL_INT	1
 #	define SIMPLE_MODE_INT	2
 #	define MEMQ_MODE_INT	3
 	unsigned int intr[4];
@@ -137,10 +137,27 @@ struct ctlr_info
 	struct list_head scan_list;
 	struct completion scan_wait;
 	struct device dev;
+	/*
+	 * Performant mode tables.
+	 */
+	u32 trans_support;
+	u32 trans_offset;
+	struct TransTable_struct *transtable;
+	unsigned long transMethod;
+
+	/*
+	 * Performant mode completion buffer
+	 */
+	u64 *reply_pool;
+	dma_addr_t reply_pool_dhandle;
+	u64 *reply_pool_head;
+	size_t reply_pool_size;
+	unsigned char reply_pool_wraparound;
+	u32 *blockFetchTable;
 };
 
-/*  Defining the diffent access_menthods */
-/*
+/*  Defining the diffent access_methods
+ *
  * Memory mapped FIFO interface (SMART 53xx cards)
  */
 #define SA5_DOORBELL	0x20
@@ -159,6 +176,15 @@ struct ctlr_info
 #define SA5B_INTR_PENDING	0x04
 #define FIFO_EMPTY		0xffffffff	
 #define CCISS_FIRMWARE_READY	0xffff0000 /* value in scratchpad register */
+/* Perf. mode flags */
+#define SA5_PERF_INTR_PENDING	0x04
+#define SA5_PERF_INTR_OFF	0x05
+#define SA5_OUTDB_STATUS_PERF_BIT	0x01
+#define SA5_OUTDB_CLEAR_PERF_BIT	0x01
+#define SA5_OUTDB_CLEAR         0xA0
+#define SA5_OUTDB_CLEAR_PERF_BIT        0x01
+#define SA5_OUTDB_STATUS        0x9C
+
 
 #define  CISS_ERROR_BIT		0x02
 
@@ -170,8 +196,9 @@ struct ctlr_info
 static void SA5_submit_command( ctlr_info_t *h, CommandList_struct *c) 
 {
 #ifdef CCISS_DEBUG
-	 printk("Sending %x - down to controller\n", c->busaddr );
-#endif /* CCISS_DEBUG */ 
+	printk(KERN_WARNING "cciss%d: Sending %08x - down to controller\n",
+			h->ctlr, c->busaddr);
+#endif /* CCISS_DEBUG */
          writel(c->busaddr, h->vaddr + SA5_REQUEST_PORT_OFFSET);
 	 h->commands_outstanding++;
 	 if ( h->commands_outstanding > h->max_outstanding)
@@ -214,6 +241,20 @@ static void SA5B_intr_mask(ctlr_info_t *h, unsigned long val)
                         h->vaddr + SA5_REPLY_INTR_MASK_OFFSET);
         }
 }
+
+/* Performant mode intr_mask */
+static void SA5_performant_intr_mask(ctlr_info_t *h, unsigned long val)
+{
+	if (val) { /* turn on interrupts */
+		h->interrupts_enabled = 1;
+		writel(0, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET);
+	} else {
+		h->interrupts_enabled = 0;
+		writel(SA5_PERF_INTR_OFF,
+				h->vaddr + SA5_REPLY_INTR_MASK_OFFSET);
+	}
+}
+
 /*
  *  Returns true if fifo is full.  
  * 
@@ -250,10 +291,44 @@ static unsigned long SA5_completed(ctlr_info_t *h)
 	return ( register_value); 
 
 }
+
+/* Performant mode command completed */
+static unsigned long SA5_performant_completed(ctlr_info_t *h)
+{
+	unsigned long register_value = FIFO_EMPTY;
+
+	/* flush the controller write of the reply queue by reading
+	 * outbound doorbell status register.
+	 */
+	register_value = readl(h->vaddr + SA5_OUTDB_STATUS);
+	/* msi auto clears the interrupt pending bit. */
+	if (!(h->msi_vector || h->msix_vector)) {
+		writel(SA5_OUTDB_CLEAR_PERF_BIT, h->vaddr + SA5_OUTDB_CLEAR);
+		/* Do a read in order to flush the write to the controller
+		 * (as per spec.)
+		 */
+		register_value = readl(h->vaddr + SA5_OUTDB_STATUS);
+	}
+
+	if ((*(h->reply_pool_head) & 1) == (h->reply_pool_wraparound)) {
+		register_value = *(h->reply_pool_head);
+		(h->reply_pool_head)++;
+		h->commands_outstanding--;
+	} else {
+		register_value = FIFO_EMPTY;
+	}
+	/* Check for wraparound */
+	if (h->reply_pool_head == (h->reply_pool + h->max_commands)) {
+		h->reply_pool_head = h->reply_pool;
+		h->reply_pool_wraparound ^= 1;
+	}
+
+	return register_value;
+}
 /*
  *	Returns true if an interrupt is pending.. 
  */
-static unsigned long SA5_intr_pending(ctlr_info_t *h)
+static bool SA5_intr_pending(ctlr_info_t *h)
 {
 	unsigned long register_value  = 
 		readl(h->vaddr + SA5_INTR_STATUS);
@@ -268,7 +343,7 @@ static unsigned long SA5_intr_pending(ctlr_info_t *h)
 /*
  *      Returns true if an interrupt is pending..
  */
-static unsigned long SA5B_intr_pending(ctlr_info_t *h)
+static bool SA5B_intr_pending(ctlr_info_t *h)
 {
         unsigned long register_value  =
                 readl(h->vaddr + SA5_INTR_STATUS);
@@ -280,6 +355,20 @@ static unsigned long SA5B_intr_pending(ctlr_info_t *h)
         return 0 ;
 }
 
+static bool SA5_performant_intr_pending(ctlr_info_t *h)
+{
+	unsigned long register_value = readl(h->vaddr + SA5_INTR_STATUS);
+
+	if (!register_value)
+		return false;
+
+	if (h->msi_vector || h->msix_vector)
+		return true;
+
+	/* Read outbound doorbell to flush */
+	register_value = readl(h->vaddr + SA5_OUTDB_STATUS);
+	return register_value & SA5_OUTDB_STATUS_PERF_BIT;
+}
 
 static struct access_method SA5_access = {
 	SA5_submit_command,
@@ -297,6 +386,14 @@ static struct access_method SA5B_access = {
         SA5_completed,
 };
 
+static struct access_method SA5_performant_access = {
+	SA5_submit_command,
+	SA5_performant_intr_mask,
+	SA5_fifo_full,
+	SA5_performant_intr_pending,
+	SA5_performant_completed,
+};
+
 struct board_type {
 	__u32	board_id;
 	char	*product_name;
diff --git a/drivers/block/cciss_cmd.h b/drivers/block/cciss_cmd.h
index e624ff959cb6..936b9666da6a 100644
--- a/drivers/block/cciss_cmd.h
+++ b/drivers/block/cciss_cmd.h
@@ -54,6 +54,7 @@
 #define CFGTBL_AccCmds          0x00000001l
 
 #define CFGTBL_Trans_Simple     0x00000002l
+#define CFGTBL_Trans_Performant 0x00000004l
 
 #define CFGTBL_BusType_Ultra2   0x00000001l
 #define CFGTBL_BusType_Ultra3   0x00000002l
@@ -173,12 +174,15 @@ typedef struct _SGDescriptor_struct {
  * PAD_64 can be adjusted independently as needed for 32-bit
  * and 64-bits systems.
  */
-#define COMMANDLIST_ALIGNMENT (8)
+#define COMMANDLIST_ALIGNMENT (32)
 #define IS_64_BIT ((sizeof(long) - 4)/4)
 #define IS_32_BIT (!IS_64_BIT)
 #define PAD_32 (0)
 #define PAD_64 (4)
 #define PADSIZE (IS_32_BIT * PAD_32 + IS_64_BIT * PAD_64)
+#define DIRECT_LOOKUP_BIT 0x10
+#define DIRECT_LOOKUP_SHIFT 5
+
 typedef struct _CommandList_struct {
   CommandListHeader_struct Header;
   RequestBlock_struct      Request;
@@ -195,7 +199,7 @@ typedef struct _CommandList_struct {
   struct completion *waiting;
   int	 retry_count;
   void * scsi_cmd;
-  char   pad[PADSIZE];
+  char pad[PADSIZE];
 } CommandList_struct;
 
 /* Configuration Table Structure */
@@ -209,12 +213,15 @@ typedef struct _HostWrite_struct {
 typedef struct _CfgTable_struct {
   BYTE             Signature[4];
   DWORD            SpecValence;
+#define SIMPLE_MODE	0x02
+#define PERFORMANT_MODE	0x04
+#define MEMQ_MODE	0x08
   DWORD            TransportSupport;
   DWORD            TransportActive;
   HostWrite_struct HostWrite;
   DWORD            CmdsOutMax;
   DWORD            BusTypes;
-  DWORD            Reserved; 
+  DWORD            TransMethodOffset;
   BYTE             ServerName[16];
   DWORD            HeartBeat;
   DWORD            SCSI_Prefetch;
@@ -222,6 +229,25 @@ typedef struct _CfgTable_struct {
   DWORD            MaxLogicalUnits;
   DWORD            MaxPhysicalDrives;
   DWORD            MaxPhysicalDrivesPerLogicalUnit;
+  DWORD            MaxPerformantModeCommands;
 } CfgTable_struct;
+
+struct TransTable_struct {
+  u32 BlockFetch0;
+  u32 BlockFetch1;
+  u32 BlockFetch2;
+  u32 BlockFetch3;
+  u32 BlockFetch4;
+  u32 BlockFetch5;
+  u32 BlockFetch6;
+  u32 BlockFetch7;
+  u32 RepQSize;
+  u32 RepQCount;
+  u32 RepQCtrAddrLow32;
+  u32 RepQCtrAddrHigh32;
+  u32 RepQAddr0Low32;
+  u32 RepQAddr0High32;
+};
+
 #pragma pack()	 
 #endif /* CCISS_CMD_H */
diff --git a/drivers/block/cciss_scsi.c b/drivers/block/cciss_scsi.c
index 3381505c8a6c..8e0a709286df 100644
--- a/drivers/block/cciss_scsi.c
+++ b/drivers/block/cciss_scsi.c
@@ -93,8 +93,8 @@ static struct scsi_host_template cciss_driver_template = {
 
 #pragma pack(1)
 
-#define SCSI_PAD_32 0
-#define SCSI_PAD_64 0
+#define SCSI_PAD_32 8
+#define SCSI_PAD_64 8
 
 struct cciss_scsi_cmd_stack_elem_t {
 	CommandList_struct cmd;
@@ -861,6 +861,7 @@ cciss_scsi_detect(int ctlr)
 	sh->n_io_port = 0;	// I don't think we use these two...
 	sh->this_id = SELF_SCSI_ID;  
 	sh->sg_tablesize = hba[ctlr]->maxsgentries;
+	sh->max_cmd_len = MAX_COMMAND_SIZE;
 
 	((struct cciss_scsi_adapter_data_t *) 
 		hba[ctlr]->scsi_ctlr)->scsi_host = sh;
diff --git a/drivers/block/cpqarray.c b/drivers/block/cpqarray.c
index 91d11631cec9..abb4ec6690fc 100644
--- a/drivers/block/cpqarray.c
+++ b/drivers/block/cpqarray.c
@@ -386,7 +386,7 @@ static void __devexit cpqarray_remove_one_eisa (int i)
 }
 
 /* pdev is NULL for eisa */
-static int __init cpqarray_register_ctlr( int i, struct pci_dev *pdev)
+static int __devinit cpqarray_register_ctlr( int i, struct pci_dev *pdev)
 {
 	struct request_queue *q;
 	int j;
@@ -503,7 +503,7 @@ Enomem4:
 	return -1;
 }
 
-static int __init cpqarray_init_one( struct pci_dev *pdev,
+static int __devinit cpqarray_init_one( struct pci_dev *pdev,
 	const struct pci_device_id *ent)
 {
 	int i;
@@ -740,7 +740,7 @@ __setup("smart2=", cpqarray_setup);
 /*
  * Find an EISA controller's signature.  Set up an hba if we find it.
  */
-static int __init cpqarray_eisa_detect(void)
+static int __devinit cpqarray_eisa_detect(void)
 {
 	int i=0, j;
 	__u32 board_id;
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 6b077f93acc6..7258c95e895e 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -1236,8 +1236,6 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 	/* Last part of the attaching process ... */
 	if (ns.conn >= C_CONNECTED &&
 	    os.disk == D_ATTACHING && ns.disk == D_NEGOTIATING) {
-		kfree(mdev->p_uuid); /* We expect to receive up-to-date UUIDs soon. */
-		mdev->p_uuid = NULL; /* ...to not use the old ones in the mean time */
 		drbd_send_sizes(mdev, 0, 0);  /* to start sync... */
 		drbd_send_uuids(mdev);
 		drbd_send_state(mdev);
diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index 632e3245d1bb..2151f18b21de 100644
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -1114,6 +1114,12 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 		mdev->new_state_tmp.i = ns.i;
 		ns.i = os.i;
 		ns.disk = D_NEGOTIATING;
+
+		/* We expect to receive up-to-date UUIDs soon.
+		   To avoid a race in receive_state, free p_uuid while
+		   holding req_lock. I.e. atomic with the state change */
+		kfree(mdev->p_uuid);
+		mdev->p_uuid = NULL;
 	}
 
 	rv = _drbd_set_state(mdev, ns, CS_VERBOSE, NULL);
diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c
index 90c4038702da..82c30f9f81ca 100644
--- a/drivers/block/floppy.c
+++ b/drivers/block/floppy.c
@@ -514,8 +514,6 @@ static unsigned long fdc_busy;
 static DECLARE_WAIT_QUEUE_HEAD(fdc_wait);
 static DECLARE_WAIT_QUEUE_HEAD(command_done);
 
-#define NO_SIGNAL (!interruptible || !signal_pending(current))
-
 /* Errors during formatting are counted here. */
 static int format_errors;
 
@@ -578,7 +576,7 @@ static void reset_fdc(void);
 #define NEED_1_RECAL	-2
 #define NEED_2_RECAL	-3
 
-static int usage_count;
+static atomic_t usage_count = ATOMIC_INIT(0);
 
 /* buffer related variables */
 static int buffer_track = -1;
@@ -858,36 +856,15 @@ static void set_fdc(int drive)
 }
 
 /* locks the driver */
-static int _lock_fdc(int drive, bool interruptible, int line)
+static int lock_fdc(int drive, bool interruptible)
 {
-	if (!usage_count) {
-		pr_err("Trying to lock fdc while usage count=0 at line %d\n",
-		       line);
+	if (WARN(atomic_read(&usage_count) == 0,
+		 "Trying to lock fdc while usage count=0\n"))
 		return -1;
-	}
-
-	if (test_and_set_bit(0, &fdc_busy)) {
-		DECLARE_WAITQUEUE(wait, current);
-		add_wait_queue(&fdc_wait, &wait);
 
-		for (;;) {
-			set_current_state(TASK_INTERRUPTIBLE);
-
-			if (!test_and_set_bit(0, &fdc_busy))
-				break;
-
-			schedule();
-
-			if (!NO_SIGNAL) {
-				remove_wait_queue(&fdc_wait, &wait);
-				return -EINTR;
-			}
-		}
+	if (wait_event_interruptible(fdc_wait, !test_and_set_bit(0, &fdc_busy)))
+		return -EINTR;
 
-		set_current_state(TASK_RUNNING);
-		remove_wait_queue(&fdc_wait, &wait);
-		flush_scheduled_work();
-	}
 	command_status = FD_COMMAND_NONE;
 
 	__reschedule_timeout(drive, "lock fdc");
@@ -895,11 +872,8 @@ static int _lock_fdc(int drive, bool interruptible, int line)
 	return 0;
 }
 
-#define lock_fdc(drive, interruptible)			\
-	_lock_fdc(drive, interruptible, __LINE__)
-
 /* unlocks the driver */
-static inline void unlock_fdc(void)
+static void unlock_fdc(void)
 {
 	unsigned long flags;
 
@@ -1224,7 +1198,7 @@ static int need_more_output(void)
 /* Set perpendicular mode as required, based on data rate, if supported.
  * 82077 Now tested. 1Mbps data rate only possible with 82077-1.
  */
-static inline void perpendicular_mode(void)
+static void perpendicular_mode(void)
 {
 	unsigned char perp_mode;
 
@@ -2015,25 +1989,10 @@ static int wait_til_done(void (*handler)(void), bool interruptible)
 
 	schedule_bh(handler);
 
-	if (command_status < 2 && NO_SIGNAL) {
-		DECLARE_WAITQUEUE(wait, current);
-
-		add_wait_queue(&command_done, &wait);
-		for (;;) {
-			set_current_state(interruptible ?
-					  TASK_INTERRUPTIBLE :
-					  TASK_UNINTERRUPTIBLE);
-
-			if (command_status >= 2 || !NO_SIGNAL)
-				break;
-
-			is_alive(__func__, "");
-			schedule();
-		}
-
-		set_current_state(TASK_RUNNING);
-		remove_wait_queue(&command_done, &wait);
-	}
+	if (interruptible)
+		wait_event_interruptible(command_done, command_status >= 2);
+	else
+		wait_event(command_done, command_status >= 2);
 
 	if (command_status < 2) {
 		cancel_activity();
@@ -2583,10 +2542,8 @@ static int make_raw_rw_request(void)
 	int tracksize;
 	int ssize;
 
-	if (max_buffer_sectors == 0) {
-		pr_info("VFS: Block I/O scheduled on unopened device\n");
+	if (WARN(max_buffer_sectors == 0, "VFS: Block I/O scheduled on unopened device\n"))
 		return 0;
-	}
 
 	set_fdc((long)current_req->rq_disk->private_data);
 
@@ -2936,19 +2893,16 @@ static void process_fd_request(void)
 
 static void do_fd_request(struct request_queue *q)
 {
-	if (max_buffer_sectors == 0) {
-		pr_info("VFS: %s called on non-open device\n", __func__);
+	if (WARN(max_buffer_sectors == 0,
+		 "VFS: %s called on non-open device\n", __func__))
 		return;
-	}
 
-	if (usage_count == 0) {
-		pr_info("warning: usage count=0, current_req=%p exiting\n",
-			current_req);
-		pr_info("sect=%ld type=%x flags=%x\n",
-			(long)blk_rq_pos(current_req), current_req->cmd_type,
-			current_req->cmd_flags);
+	if (WARN(atomic_read(&usage_count) == 0,
+		 "warning: usage count=0, current_req=%p sect=%ld type=%x flags=%x\n",
+		 current_req, (long)blk_rq_pos(current_req), current_req->cmd_type,
+		 current_req->cmd_flags))
 		return;
-	}
+
 	if (test_bit(0, &fdc_busy)) {
 		/* fdc busy, this new request will be treated when the
 		   current one is done */
@@ -3033,7 +2987,7 @@ static inline int fd_copyin(void __user *param, void *address,
 	return copy_from_user(address, param, size) ? -EFAULT : 0;
 }
 
-static inline const char *drive_name(int type, int drive)
+static const char *drive_name(int type, int drive)
 {
 	struct floppy_struct *floppy;
 
@@ -3103,7 +3057,7 @@ static struct cont_t raw_cmd_cont = {
 	.done		= raw_cmd_done
 };
 
-static inline int raw_cmd_copyout(int cmd, void __user *param,
+static int raw_cmd_copyout(int cmd, void __user *param,
 				  struct floppy_raw_cmd *ptr)
 {
 	int ret;
@@ -3148,7 +3102,7 @@ static void raw_cmd_free(struct floppy_raw_cmd **ptr)
 	}
 }
 
-static inline int raw_cmd_copyin(int cmd, void __user *param,
+static int raw_cmd_copyin(int cmd, void __user *param,
 				 struct floppy_raw_cmd **rcmd)
 {
 	struct floppy_raw_cmd *ptr;
@@ -3266,7 +3220,7 @@ static int invalidate_drive(struct block_device *bdev)
 	return 0;
 }
 
-static inline int set_geometry(unsigned int cmd, struct floppy_struct *g,
+static int set_geometry(unsigned int cmd, struct floppy_struct *g,
 			       int drive, int type, struct block_device *bdev)
 {
 	int cnt;
@@ -3337,7 +3291,7 @@ static inline int set_geometry(unsigned int cmd, struct floppy_struct *g,
 }
 
 /* handle obsolete ioctl's */
-static int ioctl_table[] = {
+static unsigned int ioctl_table[] = {
 	FDCLRPRM,
 	FDSETPRM,
 	FDDEFPRM,
@@ -3365,7 +3319,7 @@ static int ioctl_table[] = {
 	FDTWADDLE
 };
 
-static inline int normalize_ioctl(int *cmd, int *size)
+static int normalize_ioctl(unsigned int *cmd, int *size)
 {
 	int i;
 
@@ -3829,6 +3783,7 @@ static int __floppy_read_block_0(struct block_device *bdev)
 	bio.bi_size = size;
 	bio.bi_bdev = bdev;
 	bio.bi_sector = 0;
+	bio.bi_flags = BIO_QUIET;
 	init_completion(&complete);
 	bio.bi_private = &complete;
 	bio.bi_end_io = floppy_rb0_complete;
@@ -3857,10 +3812,10 @@ static int floppy_revalidate(struct gendisk *disk)
 	if (test_bit(FD_DISK_CHANGED_BIT, &UDRS->flags) ||
 	    test_bit(FD_VERIFY_BIT, &UDRS->flags) ||
 	    test_bit(drive, &fake_change) || NO_GEOM) {
-		if (usage_count == 0) {
-			pr_info("VFS: revalidate called on non-open device.\n");
+		if (WARN(atomic_read(&usage_count) == 0,
+			 "VFS: revalidate called on non-open device.\n"))
 			return -EFAULT;
-		}
+
 		lock_fdc(drive, false);
 		cf = (test_bit(FD_DISK_CHANGED_BIT, &UDRS->flags) ||
 		      test_bit(FD_VERIFY_BIT, &UDRS->flags));
@@ -4126,7 +4081,7 @@ static ssize_t floppy_cmos_show(struct device *dev,
 	return sprintf(buf, "%X\n", UDP->cmos);
 }
 
-DEVICE_ATTR(cmos, S_IRUGO, floppy_cmos_show, NULL);
+static DEVICE_ATTR(cmos, S_IRUGO, floppy_cmos_show, NULL);
 
 static void floppy_device_release(struct device *dev)
 {
@@ -4175,6 +4130,9 @@ static int __init floppy_init(void)
 	int i, unit, drive;
 	int err, dr;
 
+	set_debugt();
+	interruptjiffies = resultjiffies = jiffies;
+
 #if defined(CONFIG_PPC)
 	if (check_legacy_ioport(FDC1))
 		return -ENODEV;
@@ -4353,7 +4311,7 @@ out_unreg_platform_dev:
 	platform_device_unregister(&floppy_device[drive]);
 out_flush_work:
 	flush_scheduled_work();
-	if (usage_count)
+	if (atomic_read(&usage_count))
 		floppy_release_irq_and_dma();
 out_unreg_region:
 	blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256);
@@ -4370,8 +4328,6 @@ out_put_disk:
 	return err;
 }
 
-static DEFINE_SPINLOCK(floppy_usage_lock);
-
 static const struct io_region {
 	int offset;
 	int size;
@@ -4417,14 +4373,8 @@ static void floppy_release_regions(int fdc)
 
 static int floppy_grab_irq_and_dma(void)
 {
-	unsigned long flags;
-
-	spin_lock_irqsave(&floppy_usage_lock, flags);
-	if (usage_count++) {
-		spin_unlock_irqrestore(&floppy_usage_lock, flags);
+	if (atomic_inc_return(&usage_count) > 1)
 		return 0;
-	}
-	spin_unlock_irqrestore(&floppy_usage_lock, flags);
 
 	/*
 	 * We might have scheduled a free_irq(), wait it to
@@ -4435,9 +4385,7 @@ static int floppy_grab_irq_and_dma(void)
 	if (fd_request_irq()) {
 		DPRINT("Unable to grab IRQ%d for the floppy driver\n",
 		       FLOPPY_IRQ);
-		spin_lock_irqsave(&floppy_usage_lock, flags);
-		usage_count--;
-		spin_unlock_irqrestore(&floppy_usage_lock, flags);
+		atomic_dec(&usage_count);
 		return -1;
 	}
 	if (fd_request_dma()) {
@@ -4447,9 +4395,7 @@ static int floppy_grab_irq_and_dma(void)
 			use_virtual_dma = can_use_virtual_dma = 1;
 		if (!(can_use_virtual_dma & 1)) {
 			fd_free_irq();
-			spin_lock_irqsave(&floppy_usage_lock, flags);
-			usage_count--;
-			spin_unlock_irqrestore(&floppy_usage_lock, flags);
+			atomic_dec(&usage_count);
 			return -1;
 		}
 	}
@@ -4484,9 +4430,7 @@ cleanup:
 	fd_free_dma();
 	while (--fdc >= 0)
 		floppy_release_regions(fdc);
-	spin_lock_irqsave(&floppy_usage_lock, flags);
-	usage_count--;
-	spin_unlock_irqrestore(&floppy_usage_lock, flags);
+	atomic_dec(&usage_count);
 	return -1;
 }
 
@@ -4498,14 +4442,10 @@ static void floppy_release_irq_and_dma(void)
 #endif
 	long tmpsize;
 	unsigned long tmpaddr;
-	unsigned long flags;
 
-	spin_lock_irqsave(&floppy_usage_lock, flags);
-	if (--usage_count) {
-		spin_unlock_irqrestore(&floppy_usage_lock, flags);
+	if (!atomic_dec_and_test(&usage_count))
 		return;
-	}
-	spin_unlock_irqrestore(&floppy_usage_lock, flags);
+
 	if (irqdma_allocated) {
 		fd_disable_dma();
 		fd_free_dma();
@@ -4598,7 +4538,7 @@ static void __exit floppy_module_exit(void)
 	del_timer_sync(&fd_timer);
 	blk_cleanup_queue(floppy_queue);
 
-	if (usage_count)
+	if (atomic_read(&usage_count))
 		floppy_release_irq_and_dma();
 
 	/* eject disk, if any */
diff --git a/fs/afs/write.c b/fs/afs/write.c
index 3dab9e9948d0..722743b152d8 100644
--- a/fs/afs/write.c
+++ b/fs/afs/write.c
@@ -680,7 +680,6 @@ int afs_writeback_all(struct afs_vnode *vnode)
 {
 	struct address_space *mapping = vnode->vfs_inode.i_mapping;
 	struct writeback_control wbc = {
-		.bdi		= mapping->backing_dev_info,
 		.sync_mode	= WB_SYNC_ALL,
 		.nr_to_write	= LONG_MAX,
 		.range_cyclic	= 1,
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index a4080c21ec55..d74e6af9b53a 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -2594,7 +2594,6 @@ int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
 		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
 	};
 	struct writeback_control wbc_writepages = {
-		.bdi		= wbc->bdi,
 		.sync_mode	= wbc->sync_mode,
 		.older_than_this = NULL,
 		.nr_to_write	= 64,
@@ -2628,7 +2627,6 @@ int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode,
 		.sync_io = mode == WB_SYNC_ALL,
 	};
 	struct writeback_control wbc_writepages = {
-		.bdi		= inode->i_mapping->backing_dev_info,
 		.sync_mode	= mode,
 		.older_than_this = NULL,
 		.nr_to_write	= nr_pages * 2,
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 1d1088f48bc2..5455009b4142 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -63,24 +63,16 @@ struct bdi_work {
 };
 
 enum {
-	WS_USED_B = 0,
-	WS_ONSTACK_B,
+	WS_INPROGRESS = 0,
+	WS_ONSTACK,
 };
 
-#define WS_USED (1 << WS_USED_B)
-#define WS_ONSTACK (1 << WS_ONSTACK_B)
-
-static inline bool bdi_work_on_stack(struct bdi_work *work)
-{
-	return test_bit(WS_ONSTACK_B, &work->state);
-}
-
 static inline void bdi_work_init(struct bdi_work *work,
 				 struct wb_writeback_args *args)
 {
 	INIT_RCU_HEAD(&work->rcu_head);
 	work->args = *args;
-	work->state = WS_USED;
+	__set_bit(WS_INPROGRESS, &work->state);
 }
 
 /**
@@ -95,43 +87,16 @@ int writeback_in_progress(struct backing_dev_info *bdi)
 	return !list_empty(&bdi->work_list);
 }
 
-static void bdi_work_clear(struct bdi_work *work)
-{
-	clear_bit(WS_USED_B, &work->state);
-	smp_mb__after_clear_bit();
-	/*
-	 * work can have disappeared at this point. bit waitq functions
-	 * should be able to tolerate this, provided bdi_sched_wait does
-	 * not dereference it's pointer argument.
-	*/
-	wake_up_bit(&work->state, WS_USED_B);
-}
-
 static void bdi_work_free(struct rcu_head *head)
 {
 	struct bdi_work *work = container_of(head, struct bdi_work, rcu_head);
 
-	if (!bdi_work_on_stack(work))
-		kfree(work);
-	else
-		bdi_work_clear(work);
-}
-
-static void wb_work_complete(struct bdi_work *work)
-{
-	const enum writeback_sync_modes sync_mode = work->args.sync_mode;
-	int onstack = bdi_work_on_stack(work);
+	clear_bit(WS_INPROGRESS, &work->state);
+	smp_mb__after_clear_bit();
+	wake_up_bit(&work->state, WS_INPROGRESS);
 
-	/*
-	 * For allocated work, we can clear the done/seen bit right here.
-	 * For on-stack work, we need to postpone both the clear and free
-	 * to after the RCU grace period, since the stack could be invalidated
-	 * as soon as bdi_work_clear() has done the wakeup.
-	 */
-	if (!onstack)
-		bdi_work_clear(work);
-	if (sync_mode == WB_SYNC_NONE || onstack)
-		call_rcu(&work->rcu_head, bdi_work_free);
+	if (!test_bit(WS_ONSTACK, &work->state))
+		kfree(work);
 }
 
 static void wb_clear_pending(struct bdi_writeback *wb, struct bdi_work *work)
@@ -147,7 +112,7 @@ static void wb_clear_pending(struct bdi_writeback *wb, struct bdi_work *work)
 		list_del_rcu(&work->list);
 		spin_unlock(&bdi->wb_lock);
 
-		wb_work_complete(work);
+		call_rcu(&work->rcu_head, bdi_work_free);
 	}
 }
 
@@ -185,9 +150,9 @@ static void bdi_queue_work(struct backing_dev_info *bdi, struct bdi_work *work)
  * Used for on-stack allocated work items. The caller needs to wait until
  * the wb threads have acked the work before it's safe to continue.
  */
-static void bdi_wait_on_work_clear(struct bdi_work *work)
+static void bdi_wait_on_work_done(struct bdi_work *work)
 {
-	wait_on_bit(&work->state, WS_USED_B, bdi_sched_wait,
+	wait_on_bit(&work->state, WS_INPROGRESS, bdi_sched_wait,
 		    TASK_UNINTERRUPTIBLE);
 }
 
@@ -213,37 +178,28 @@ static void bdi_alloc_queue_work(struct backing_dev_info *bdi,
 }
 
 /**
- * bdi_sync_writeback - start and wait for writeback
- * @bdi: the backing device to write from
+ * bdi_queue_work_onstack - start and wait for writeback
  * @sb: write inodes from this super_block
  *
  * Description:
- *   This does WB_SYNC_ALL data integrity writeback and waits for the
- *   IO to complete. Callers must hold the sb s_umount semaphore for
+ *   This function initiates writeback and waits for the operation to
+ *   complete. Callers must hold the sb s_umount semaphore for
  *   reading, to avoid having the super disappear before we are done.
  */
-static void bdi_sync_writeback(struct backing_dev_info *bdi,
-			       struct super_block *sb)
+static void bdi_queue_work_onstack(struct wb_writeback_args *args)
 {
-	struct wb_writeback_args args = {
-		.sb		= sb,
-		.sync_mode	= WB_SYNC_ALL,
-		.nr_pages	= LONG_MAX,
-		.range_cyclic	= 0,
-	};
 	struct bdi_work work;
 
-	bdi_work_init(&work, &args);
-	work.state |= WS_ONSTACK;
+	bdi_work_init(&work, args);
+	__set_bit(WS_ONSTACK, &work.state);
 
-	bdi_queue_work(bdi, &work);
-	bdi_wait_on_work_clear(&work);
+	bdi_queue_work(args->sb->s_bdi, &work);
+	bdi_wait_on_work_done(&work);
 }
 
 /**
  * bdi_start_writeback - start writeback
  * @bdi: the backing device to write from
- * @sb: write inodes from this super_block
  * @nr_pages: the number of pages to write
  *
  * Description:
@@ -252,25 +208,34 @@ static void bdi_sync_writeback(struct backing_dev_info *bdi,
  *   completion. Caller need not hold sb s_umount semaphore.
  *
  */
-void bdi_start_writeback(struct backing_dev_info *bdi, struct super_block *sb,
-			 long nr_pages)
+void bdi_start_writeback(struct backing_dev_info *bdi, long nr_pages)
 {
 	struct wb_writeback_args args = {
-		.sb		= sb,
 		.sync_mode	= WB_SYNC_NONE,
 		.nr_pages	= nr_pages,
 		.range_cyclic	= 1,
 	};
 
-	/*
-	 * We treat @nr_pages=0 as the special case to do background writeback,
-	 * ie. to sync pages until the background dirty threshold is reached.
-	 */
-	if (!nr_pages) {
-		args.nr_pages = LONG_MAX;
-		args.for_background = 1;
-	}
+	bdi_alloc_queue_work(bdi, &args);
+}
 
+/**
+ * bdi_start_background_writeback - start background writeback
+ * @bdi: the backing device to write from
+ *
+ * Description:
+ *   This does WB_SYNC_NONE background writeback. The IO is only
+ *   started when this function returns, we make no guarentees on
+ *   completion. Caller need not hold sb s_umount semaphore.
+ */
+void bdi_start_background_writeback(struct backing_dev_info *bdi)
+{
+	struct wb_writeback_args args = {
+		.sync_mode	= WB_SYNC_NONE,
+		.nr_pages	= LONG_MAX,
+		.for_background = 1,
+		.range_cyclic	= 1,
+	};
 	bdi_alloc_queue_work(bdi, &args);
 }
 
@@ -561,75 +526,69 @@ select_queue:
 	return ret;
 }
 
-static void unpin_sb_for_writeback(struct super_block *sb)
-{
-	up_read(&sb->s_umount);
-	put_super(sb);
-}
-
-enum sb_pin_state {
-	SB_PINNED,
-	SB_NOT_PINNED,
-	SB_PIN_FAILED
-};
-
 /*
- * For WB_SYNC_NONE writeback, the caller does not have the sb pinned
+ * For background writeback the caller does not have the sb pinned
  * before calling writeback. So make sure that we do pin it, so it doesn't
  * go away while we are writing inodes from it.
  */
-static enum sb_pin_state pin_sb_for_writeback(struct writeback_control *wbc,
-					      struct super_block *sb)
+static bool pin_sb_for_writeback(struct super_block *sb)
 {
-	/*
-	 * Caller must already hold the ref for this
-	 */
-	if (wbc->sync_mode == WB_SYNC_ALL) {
-		WARN_ON(!rwsem_is_locked(&sb->s_umount));
-		return SB_NOT_PINNED;
-	}
 	spin_lock(&sb_lock);
+	if (list_empty(&sb->s_instances)) {
+		spin_unlock(&sb_lock);
+		return false;
+	}
+
 	sb->s_count++;
+	spin_unlock(&sb_lock);
+
 	if (down_read_trylock(&sb->s_umount)) {
-		if (sb->s_root) {
-			spin_unlock(&sb_lock);
-			return SB_PINNED;
-		}
-		/*
-		 * umounted, drop rwsem again and fall through to failure
-		 */
+		if (sb->s_root)
+			return true;
 		up_read(&sb->s_umount);
 	}
-	sb->s_count--;
-	spin_unlock(&sb_lock);
-	return SB_PIN_FAILED;
+
+	put_super(sb);
+	return false;
 }
 
 /*
  * Write a portion of b_io inodes which belong to @sb.
- * If @wbc->sb != NULL, then find and write all such
+ *
+ * If @only_this_sb is true, then find and write all such
  * inodes. Otherwise write only ones which go sequentially
  * in reverse order.
+ *
  * Return 1, if the caller writeback routine should be
  * interrupted. Otherwise return 0.
  */
-static int writeback_sb_inodes(struct super_block *sb,
-			       struct bdi_writeback *wb,
-			       struct writeback_control *wbc)
+static int writeback_sb_inodes(struct super_block *sb, struct bdi_writeback *wb,
+		struct writeback_control *wbc, bool only_this_sb)
 {
 	while (!list_empty(&wb->b_io)) {
 		long pages_skipped;
 		struct inode *inode = list_entry(wb->b_io.prev,
 						 struct inode, i_list);
-		if (wbc->sb && sb != inode->i_sb) {
-			/* super block given and doesn't
-			   match, skip this inode */
-			redirty_tail(inode);
-			continue;
-		}
-		if (sb != inode->i_sb)
-			/* finish with this superblock */
+
+		if (inode->i_sb != sb) {
+			if (only_this_sb) {
+				/*
+				 * We only want to write back data for this
+				 * superblock, move all inodes not belonging
+				 * to it back onto the dirty list.
+				 */
+				redirty_tail(inode);
+				continue;
+			}
+
+			/*
+			 * The inode belongs to a different superblock.
+			 * Bounce back to the caller to unpin this and
+			 * pin the next superblock.
+			 */
 			return 0;
+		}
+
 		if (inode->i_state & (I_NEW | I_WILL_FREE)) {
 			requeue_io(inode);
 			continue;
@@ -667,8 +626,8 @@ static int writeback_sb_inodes(struct super_block *sb,
 	return 1;
 }
 
-static void writeback_inodes_wb(struct bdi_writeback *wb,
-				struct writeback_control *wbc)
+void writeback_inodes_wb(struct bdi_writeback *wb,
+		struct writeback_control *wbc)
 {
 	int ret = 0;
 
@@ -681,24 +640,14 @@ static void writeback_inodes_wb(struct bdi_writeback *wb,
 		struct inode *inode = list_entry(wb->b_io.prev,
 						 struct inode, i_list);
 		struct super_block *sb = inode->i_sb;
-		enum sb_pin_state state;
 
-		if (wbc->sb && sb != wbc->sb) {
-			/* super block given and doesn't
-			   match, skip this inode */
-			redirty_tail(inode);
-			continue;
-		}
-		state = pin_sb_for_writeback(wbc, sb);
-
-		if (state == SB_PIN_FAILED) {
+		if (!pin_sb_for_writeback(sb)) {
 			requeue_io(inode);
 			continue;
 		}
-		ret = writeback_sb_inodes(sb, wb, wbc);
+		ret = writeback_sb_inodes(sb, wb, wbc, false);
+		drop_super(sb);
 
-		if (state == SB_PINNED)
-			unpin_sb_for_writeback(sb);
 		if (ret)
 			break;
 	}
@@ -706,11 +655,17 @@ static void writeback_inodes_wb(struct bdi_writeback *wb,
 	/* Leave any unwritten inodes on b_io */
 }
 
-void writeback_inodes_wbc(struct writeback_control *wbc)
+static void __writeback_inodes_sb(struct super_block *sb,
+		struct bdi_writeback *wb, struct writeback_control *wbc)
 {
-	struct backing_dev_info *bdi = wbc->bdi;
+	WARN_ON(!rwsem_is_locked(&sb->s_umount));
 
-	writeback_inodes_wb(&bdi->wb, wbc);
+	wbc->wb_start = jiffies; /* livelock avoidance */
+	spin_lock(&inode_lock);
+	if (!wbc->for_kupdate || list_empty(&wb->b_io))
+		queue_io(wb, wbc->older_than_this);
+	writeback_sb_inodes(sb, wb, wbc, true);
+	spin_unlock(&inode_lock);
 }
 
 /*
@@ -751,8 +706,6 @@ static long wb_writeback(struct bdi_writeback *wb,
 			 struct wb_writeback_args *args)
 {
 	struct writeback_control wbc = {
-		.bdi			= wb->bdi,
-		.sb			= args->sb,
 		.sync_mode		= args->sync_mode,
 		.older_than_this	= NULL,
 		.for_kupdate		= args->for_kupdate,
@@ -790,7 +743,10 @@ static long wb_writeback(struct bdi_writeback *wb,
 		wbc.more_io = 0;
 		wbc.nr_to_write = MAX_WRITEBACK_PAGES;
 		wbc.pages_skipped = 0;
-		writeback_inodes_wb(wb, &wbc);
+		if (args->sb)
+			__writeback_inodes_sb(args->sb, wb, &wbc);
+		else
+			writeback_inodes_wb(wb, &wbc);
 		args->nr_pages -= MAX_WRITEBACK_PAGES - wbc.nr_to_write;
 		wrote += MAX_WRITEBACK_PAGES - wbc.nr_to_write;
 
@@ -911,7 +867,7 @@ long wb_do_writeback(struct bdi_writeback *wb, int force_wait)
 		 * If this isn't a data integrity operation, just notify
 		 * that we have seen this work and we are now starting it.
 		 */
-		if (args.sync_mode == WB_SYNC_NONE)
+		if (!test_bit(WS_ONSTACK, &work->state))
 			wb_clear_pending(wb, work);
 
 		wrote += wb_writeback(wb, &args);
@@ -920,7 +876,7 @@ long wb_do_writeback(struct bdi_writeback *wb, int force_wait)
 		 * This is a data integrity writeback, so only do the
 		 * notification when we have completed the work.
 		 */
-		if (args.sync_mode == WB_SYNC_ALL)
+		if (test_bit(WS_ONSTACK, &work->state))
 			wb_clear_pending(wb, work);
 	}
 
@@ -978,42 +934,32 @@ int bdi_writeback_task(struct bdi_writeback *wb)
 }
 
 /*
- * Schedule writeback for all backing devices. This does WB_SYNC_NONE
- * writeback, for integrity writeback see bdi_sync_writeback().
+ * Start writeback of `nr_pages' pages.  If `nr_pages' is zero, write back
+ * the whole world.
  */
-static void bdi_writeback_all(struct super_block *sb, long nr_pages)
+void wakeup_flusher_threads(long nr_pages)
 {
+	struct backing_dev_info *bdi;
 	struct wb_writeback_args args = {
-		.sb		= sb,
-		.nr_pages	= nr_pages,
 		.sync_mode	= WB_SYNC_NONE,
 	};
-	struct backing_dev_info *bdi;
 
-	rcu_read_lock();
+	if (nr_pages) {
+		args.nr_pages = nr_pages;
+	} else {
+		args.nr_pages = global_page_state(NR_FILE_DIRTY) +
+				global_page_state(NR_UNSTABLE_NFS);
+	}
 
+	rcu_read_lock();
 	list_for_each_entry_rcu(bdi, &bdi_list, bdi_list) {
 		if (!bdi_has_dirty_io(bdi))
 			continue;
-
 		bdi_alloc_queue_work(bdi, &args);
 	}
-
 	rcu_read_unlock();
 }
 
-/*
- * Start writeback of `nr_pages' pages.  If `nr_pages' is zero, write back
- * the whole world.
- */
-void wakeup_flusher_threads(long nr_pages)
-{
-	if (nr_pages == 0)
-		nr_pages = global_page_state(NR_FILE_DIRTY) +
-				global_page_state(NR_UNSTABLE_NFS);
-	bdi_writeback_all(NULL, nr_pages);
-}
-
 static noinline void block_dump___mark_inode_dirty(struct inode *inode)
 {
 	if (inode->i_ino || strcmp(inode->i_sb->s_id, "bdev")) {
@@ -1218,12 +1164,17 @@ void writeback_inodes_sb(struct super_block *sb)
 {
 	unsigned long nr_dirty = global_page_state(NR_FILE_DIRTY);
 	unsigned long nr_unstable = global_page_state(NR_UNSTABLE_NFS);
-	long nr_to_write;
+	struct wb_writeback_args args = {
+		.sb		= sb,
+		.sync_mode	= WB_SYNC_NONE,
+	};
 
-	nr_to_write = nr_dirty + nr_unstable +
+	WARN_ON(!rwsem_is_locked(&sb->s_umount));
+
+	args.nr_pages = nr_dirty + nr_unstable +
 			(inodes_stat.nr_inodes - inodes_stat.nr_unused);
 
-	bdi_start_writeback(sb->s_bdi, sb, nr_to_write);
+	bdi_queue_work_onstack(&args);
 }
 EXPORT_SYMBOL(writeback_inodes_sb);
 
@@ -1237,7 +1188,9 @@ EXPORT_SYMBOL(writeback_inodes_sb);
 int writeback_inodes_sb_if_idle(struct super_block *sb)
 {
 	if (!writeback_in_progress(sb->s_bdi)) {
+		down_read(&sb->s_umount);
 		writeback_inodes_sb(sb);
+		up_read(&sb->s_umount);
 		return 1;
 	} else
 		return 0;
@@ -1253,7 +1206,16 @@ EXPORT_SYMBOL(writeback_inodes_sb_if_idle);
  */
 void sync_inodes_sb(struct super_block *sb)
 {
-	bdi_sync_writeback(sb->s_bdi, sb);
+	struct wb_writeback_args args = {
+		.sb		= sb,
+		.sync_mode	= WB_SYNC_ALL,
+		.nr_pages	= LONG_MAX,
+		.range_cyclic	= 0,
+	};
+
+	WARN_ON(!rwsem_is_locked(&sb->s_umount));
+
+	bdi_queue_work_onstack(&args);
 	wait_sb_inodes(sb);
 }
 EXPORT_SYMBOL(sync_inodes_sb);
diff --git a/fs/ubifs/budget.c b/fs/ubifs/budget.c
index 076ca50e9933..c8ff0d1ae5d3 100644
--- a/fs/ubifs/budget.c
+++ b/fs/ubifs/budget.c
@@ -62,7 +62,9 @@
  */
 static void shrink_liability(struct ubifs_info *c, int nr_to_write)
 {
+	down_read(&c->vfs_sb->s_umount);
 	writeback_inodes_sb(c->vfs_sb);
+	up_read(&c->vfs_sb->s_umount);
 }
 
 /**
diff --git a/include/linux/backing-dev.h b/include/linux/backing-dev.h
index aee5f6ce166e..9ae2889096b6 100644
--- a/include/linux/backing-dev.h
+++ b/include/linux/backing-dev.h
@@ -105,8 +105,8 @@ int bdi_register(struct backing_dev_info *bdi, struct device *parent,
 int bdi_register_dev(struct backing_dev_info *bdi, dev_t dev);
 void bdi_unregister(struct backing_dev_info *bdi);
 int bdi_setup_and_register(struct backing_dev_info *, char *, unsigned int);
-void bdi_start_writeback(struct backing_dev_info *bdi, struct super_block *sb,
-				long nr_pages);
+void bdi_start_writeback(struct backing_dev_info *bdi, long nr_pages);
+void bdi_start_background_writeback(struct backing_dev_info *bdi);
 int bdi_writeback_task(struct bdi_writeback *wb);
 int bdi_has_dirty_io(struct backing_dev_info *bdi);
 void bdi_arm_supers_timer(void);
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 09a840264d6f..b8224ea4a5de 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -467,11 +467,13 @@ struct request_queue
 #define QUEUE_FLAG_IO_STAT     15	/* do IO stats */
 #define QUEUE_FLAG_DISCARD     16	/* supports DISCARD */
 #define QUEUE_FLAG_NOXMERGES   17	/* No extended merges */
+#define QUEUE_FLAG_ADD_RANDOM  18	/* Contributes to random pool */
 
 #define QUEUE_FLAG_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
 				 (1 << QUEUE_FLAG_CLUSTER) |		\
 				 (1 << QUEUE_FLAG_STACKABLE)	|	\
-				 (1 << QUEUE_FLAG_SAME_COMP))
+				 (1 << QUEUE_FLAG_SAME_COMP)	|	\
+				 (1 << QUEUE_FLAG_ADD_RANDOM))
 
 static inline int queue_is_locked(struct request_queue *q)
 {
@@ -596,6 +598,7 @@ enum {
 	test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
 #define blk_queue_nonrot(q)	test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
 #define blk_queue_io_stat(q)	test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
+#define blk_queue_add_random(q)	test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
 #define blk_queue_flushing(q)	((q)->ordseq)
 #define blk_queue_stackable(q)	\
 	test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
diff --git a/include/linux/drbd.h b/include/linux/drbd.h
index 30da4ae48972..b8d2516668aa 100644
--- a/include/linux/drbd.h
+++ b/include/linux/drbd.h
@@ -53,7 +53,7 @@
 
 
 extern const char *drbd_buildtag(void);
-#define REL_VERSION "8.3.8rc2"
+#define REL_VERSION "8.3.8"
 #define API_VERSION 88
 #define PRO_VERSION_MIN 86
 #define PRO_VERSION_MAX 94
diff --git a/include/linux/writeback.h b/include/linux/writeback.h
index d63ef8f9609f..c24eca71e80c 100644
--- a/include/linux/writeback.h
+++ b/include/linux/writeback.h
@@ -27,10 +27,6 @@ enum writeback_sync_modes {
  * in a manner such that unspecified fields are set to zero.
  */
 struct writeback_control {
-	struct backing_dev_info *bdi;	/* If !NULL, only write back this
-					   queue */
-	struct super_block *sb;		/* if !NULL, only write inodes from
-					   this super_block */
 	enum writeback_sync_modes sync_mode;
 	unsigned long *older_than_this;	/* If !NULL, only write back inodes
 					   older than this */
@@ -66,7 +62,8 @@ int inode_wait(void *);
 void writeback_inodes_sb(struct super_block *);
 int writeback_inodes_sb_if_idle(struct super_block *);
 void sync_inodes_sb(struct super_block *);
-void writeback_inodes_wbc(struct writeback_control *wbc);
+void writeback_inodes_wb(struct bdi_writeback *wb,
+		struct writeback_control *wbc);
 long wb_do_writeback(struct bdi_writeback *wb, int force_wait);
 void wakeup_flusher_threads(long nr_pages);
 
diff --git a/mm/backing-dev.c b/mm/backing-dev.c
index 660a87a22511..6e0b09a1ec2c 100644
--- a/mm/backing-dev.c
+++ b/mm/backing-dev.c
@@ -340,14 +340,13 @@ int bdi_has_dirty_io(struct backing_dev_info *bdi)
 static void bdi_flush_io(struct backing_dev_info *bdi)
 {
 	struct writeback_control wbc = {
-		.bdi			= bdi,
 		.sync_mode		= WB_SYNC_NONE,
 		.older_than_this	= NULL,
 		.range_cyclic		= 1,
 		.nr_to_write		= 1024,
 	};
 
-	writeback_inodes_wbc(&wbc);
+	writeback_inodes_wb(&bdi->wb, &wbc);
 }
 
 /*
diff --git a/mm/page-writeback.c b/mm/page-writeback.c
index bbd396ac9546..37498ef61548 100644
--- a/mm/page-writeback.c
+++ b/mm/page-writeback.c
@@ -495,7 +495,6 @@ static void balance_dirty_pages(struct address_space *mapping,
 
 	for (;;) {
 		struct writeback_control wbc = {
-			.bdi		= bdi,
 			.sync_mode	= WB_SYNC_NONE,
 			.older_than_this = NULL,
 			.nr_to_write	= write_chunk,
@@ -537,7 +536,7 @@ static void balance_dirty_pages(struct address_space *mapping,
 		 * up.
 		 */
 		if (bdi_nr_reclaimable > bdi_thresh) {
-			writeback_inodes_wbc(&wbc);
+			writeback_inodes_wb(&bdi->wb, &wbc);
 			pages_written += write_chunk - wbc.nr_to_write;
 			get_dirty_limits(&background_thresh, &dirty_thresh,
 				       &bdi_thresh, bdi);
@@ -597,7 +596,7 @@ static void balance_dirty_pages(struct address_space *mapping,
 	    (!laptop_mode && ((global_page_state(NR_FILE_DIRTY)
 			       + global_page_state(NR_UNSTABLE_NFS))
 					  > background_thresh)))
-		bdi_start_writeback(bdi, NULL, 0);
+		bdi_start_background_writeback(bdi);
 }
 
 void set_page_dirty_balance(struct page *page, int page_mkwrite)
@@ -705,9 +704,8 @@ void laptop_mode_timer_fn(unsigned long data)
 	 * We want to write everything out, not just down to the dirty
 	 * threshold
 	 */
-
 	if (bdi_has_dirty_io(&q->backing_dev_info))
-		bdi_start_writeback(&q->backing_dev_info, NULL, nr_pages);
+		bdi_start_writeback(&q->backing_dev_info, nr_pages);
 }
 
 /*