Merge branch 'for-2.6.39/drivers' of git://git.kernel.dk/linux-2.6-block

* 'for-2.6.39/drivers' of git://git.kernel.dk/linux-2.6-block: (122 commits)
  cciss: fix lost command issue
  drbd: need include for bitops functions declarations
  Revert "cciss: Add missing allocation in scsi_cmd_stack_setup and  corresponding deallocation"
  cciss: fix missed command status value CMD_UNABORTABLE
  cciss: remove unnecessary casts
  cciss: Mask off error bits of c->busaddr in cmd_special_free when calling pci_free_consistent
  cciss: Inform controller we are using 32-bit tags.
  cciss: hoist tag masking out of loop
  cciss: Add missing allocation in scsi_cmd_stack_setup and  corresponding deallocation
  cciss: export resettable host attribute
  drbd: drop code present under #ifdef which is relevant to 2.6.28 and below
  drbd: Fixed handling of read errors on a 'VerifyS' node
  drbd: Fixed handling of read errors on a 'VerifyT' node
  drbd: Implemented real timeout checking for request processing time
  drbd: Remove unused function atodb_endio()
  drbd: improve log message if received sector offset exceeds local capacity
  drbd: kill dead code
  drbd: don't BUG_ON, if bio_add_page of a single page to an empty bio fails
  drbd: Removed left over, now wrong comments
  drbd: serialize admin requests for new verify run with pending bitmap io
  ...

21 files changed, 2267 insertions, 1403 deletions

diff --git a/Documentation/ABI/testing/sysfs-bus-pci-devices-cciss b/Documentation/ABI/testing/sysfs-bus-pci-devices-cciss
index 4f29e5f1ebf..f5bb0a3bb8c 100644
--- a/Documentation/ABI/testing/sysfs-bus-pci-devices-cciss
+++ b/Documentation/ABI/testing/sysfs-bus-pci-devices-cciss
@@ -59,3 +59,15 @@ Kernel Version: 2.6.31
 Contact:	iss_storagedev@hp.com
 Description:	Displays the usage count (number of opens) of logical drive Y
 		of controller X.
+
+Where:		/sys/bus/pci/devices/<dev>/ccissX/resettable
+Date:		February 2011
+Kernel Version:	2.6.38
+Contact:	iss_storagedev@hp.com
+Description:	Value of 1 indicates the controller can honor the reset_devices
+		kernel parameter.  Value of 0 indicates reset_devices cannot be
+		honored.  This is to allow, for example, kexec tools to be able
+		to warn the user if they designate an unresettable device as
+		a dump device, as kdump requires resetting the device in order
+		to work reliably.
+
diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c
index 35658f445fc..9bf13988f1a 100644
--- a/drivers/block/cciss.c
+++ b/drivers/block/cciss.c
@@ -193,7 +193,7 @@ static int __devinit cciss_find_cfg_addrs(struct pci_dev *pdev,
 	u64 *cfg_offset);
 static int __devinit cciss_pci_find_memory_BAR(struct pci_dev *pdev,
 	unsigned long *memory_bar);
-
+static inline u32 cciss_tag_discard_error_bits(ctlr_info_t *h, u32 tag);
 
 /* performant mode helper functions */
 static void  calc_bucket_map(int *bucket, int num_buckets, int nsgs,
@@ -231,7 +231,7 @@ static const struct block_device_operations cciss_fops = {
  */
 static void set_performant_mode(ctlr_info_t *h, CommandList_struct *c)
 {
-	if (likely(h->transMethod == CFGTBL_Trans_Performant))
+	if (likely(h->transMethod & CFGTBL_Trans_Performant))
 		c->busaddr |= 1 | (h->blockFetchTable[c->Header.SGList] << 1);
 }
 
@@ -556,6 +556,44 @@ static void __devinit cciss_procinit(ctlr_info_t *h)
 #define to_hba(n) container_of(n, struct ctlr_info, dev)
 #define to_drv(n) container_of(n, drive_info_struct, dev)
 
+/* List of controllers which cannot be reset on kexec with reset_devices */
+static u32 unresettable_controller[] = {
+	0x324a103C, /* Smart Array P712m */
+	0x324b103C, /* SmartArray P711m */
+	0x3223103C, /* Smart Array P800 */
+	0x3234103C, /* Smart Array P400 */
+	0x3235103C, /* Smart Array P400i */
+	0x3211103C, /* Smart Array E200i */
+	0x3212103C, /* Smart Array E200 */
+	0x3213103C, /* Smart Array E200i */
+	0x3214103C, /* Smart Array E200i */
+	0x3215103C, /* Smart Array E200i */
+	0x3237103C, /* Smart Array E500 */
+	0x323D103C, /* Smart Array P700m */
+	0x409C0E11, /* Smart Array 6400 */
+	0x409D0E11, /* Smart Array 6400 EM */
+};
+
+static int ctlr_is_resettable(struct ctlr_info *h)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(unresettable_controller); i++)
+		if (unresettable_controller[i] == h->board_id)
+			return 0;
+	return 1;
+}
+
+static ssize_t host_show_resettable(struct device *dev,
+				    struct device_attribute *attr,
+				    char *buf)
+{
+	struct ctlr_info *h = to_hba(dev);
+
+	return snprintf(buf, 20, "%d\n", ctlr_is_resettable(h));
+}
+static DEVICE_ATTR(resettable, S_IRUGO, host_show_resettable, NULL);
+
 static ssize_t host_store_rescan(struct device *dev,
 				 struct device_attribute *attr,
 				 const char *buf, size_t count)
@@ -741,6 +779,7 @@ static DEVICE_ATTR(usage_count, S_IRUGO, cciss_show_usage_count, NULL);
 
 static struct attribute *cciss_host_attrs[] = {
 	&dev_attr_rescan.attr,
+	&dev_attr_resettable.attr,
 	NULL
 };
 
@@ -973,8 +1012,8 @@ static void cmd_special_free(ctlr_info_t *h, CommandList_struct *c)
 	temp64.val32.upper = c->ErrDesc.Addr.upper;
 	pci_free_consistent(h->pdev, sizeof(ErrorInfo_struct),
 			    c->err_info, (dma_addr_t) temp64.val);
-	pci_free_consistent(h->pdev, sizeof(CommandList_struct),
-			    c, (dma_addr_t) c->busaddr);
+	pci_free_consistent(h->pdev, sizeof(CommandList_struct), c,
+		(dma_addr_t) cciss_tag_discard_error_bits(h, (u32) c->busaddr));
 }
 
 static inline ctlr_info_t *get_host(struct gendisk *disk)
@@ -1490,8 +1529,7 @@ static int cciss_bigpassthru(ctlr_info_t *h, void __user *argp)
 		return -EINVAL;
 	if (!capable(CAP_SYS_RAWIO))
 		return -EPERM;
-	ioc = (BIG_IOCTL_Command_struct *)
-	    kmalloc(sizeof(*ioc), GFP_KERNEL);
+	ioc = kmalloc(sizeof(*ioc), GFP_KERNEL);
 	if (!ioc) {
 		status = -ENOMEM;
 		goto cleanup1;
@@ -2653,6 +2691,10 @@ static int process_sendcmd_error(ctlr_info_t *h, CommandList_struct *c)
 			c->Request.CDB[0]);
 		return_status = IO_NEEDS_RETRY;
 		break;
+	case CMD_UNABORTABLE:
+		dev_warn(&h->pdev->dev, "cmd unabortable\n");
+		return_status = IO_ERROR;
+		break;
 	default:
 		dev_warn(&h->pdev->dev, "cmd 0x%02x returned "
 		       "unknown status %x\n", c->Request.CDB[0],
@@ -3103,6 +3145,13 @@ static inline void complete_command(ctlr_info_t *h, CommandList_struct *cmd,
 			(cmd->rq->cmd_type == REQ_TYPE_BLOCK_PC) ?
 				DID_PASSTHROUGH : DID_ERROR);
 		break;
+	case CMD_UNABORTABLE:
+		dev_warn(&h->pdev->dev, "cmd %p unabortable\n", cmd);
+		rq->errors = make_status_bytes(SAM_STAT_GOOD,
+			cmd->err_info->CommandStatus, DRIVER_OK,
+			cmd->rq->cmd_type == REQ_TYPE_BLOCK_PC ?
+				DID_PASSTHROUGH : DID_ERROR);
+		break;
 	default:
 		dev_warn(&h->pdev->dev, "cmd %p returned "
 		       "unknown status %x\n", cmd,
@@ -3136,10 +3185,13 @@ static inline u32 cciss_tag_to_index(u32 tag)
 	return tag >> DIRECT_LOOKUP_SHIFT;
 }
 
-static inline u32 cciss_tag_discard_error_bits(u32 tag)
+static inline u32 cciss_tag_discard_error_bits(ctlr_info_t *h, u32 tag)
 {
-#define CCISS_ERROR_BITS 0x03
-	return tag & ~CCISS_ERROR_BITS;
+#define CCISS_PERF_ERROR_BITS ((1 << DIRECT_LOOKUP_SHIFT) - 1)
+#define CCISS_SIMPLE_ERROR_BITS 0x03
+	if (likely(h->transMethod & CFGTBL_Trans_Performant))
+		return tag & ~CCISS_PERF_ERROR_BITS;
+	return tag & ~CCISS_SIMPLE_ERROR_BITS;
 }
 
 static inline void cciss_mark_tag_indexed(u32 *tag)
@@ -3359,7 +3411,7 @@ static inline u32 next_command(ctlr_info_t *h)
 {
 	u32 a;
 
-	if (unlikely(h->transMethod != CFGTBL_Trans_Performant))
+	if (unlikely(!(h->transMethod & CFGTBL_Trans_Performant)))
 		return h->access.command_completed(h);
 
 	if ((*(h->reply_pool_head) & 1) == (h->reply_pool_wraparound)) {
@@ -3394,14 +3446,12 @@ static inline u32 process_indexed_cmd(ctlr_info_t *h, u32 raw_tag)
 /* 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;
 	__u32 busaddr_masked, tag_masked;
 
-	tag = cciss_tag_discard_error_bits(raw_tag);
+	tag_masked = cciss_tag_discard_error_bits(h, raw_tag);
 	list_for_each_entry(c, &h->cmpQ, list) {
-		busaddr_masked = cciss_tag_discard_error_bits(c->busaddr);
-		tag_masked = cciss_tag_discard_error_bits(tag);
+		busaddr_masked = cciss_tag_discard_error_bits(h, c->busaddr);
 		if (busaddr_masked == tag_masked) {
 			finish_cmd(h, c, raw_tag);
 			return next_command(h);
@@ -3753,7 +3803,8 @@ static void __devinit cciss_wait_for_mode_change_ack(ctlr_info_t *h)
 	}
 }
 
-static __devinit void cciss_enter_performant_mode(ctlr_info_t *h)
+static __devinit void cciss_enter_performant_mode(ctlr_info_t *h,
+	u32 use_short_tags)
 {
 	/* This is a bit complicated.  There are 8 registers on
 	 * the controller which we write to to tell it 8 different
@@ -3808,7 +3859,7 @@ static __devinit void cciss_enter_performant_mode(ctlr_info_t *h)
 	writel(0, &h->transtable->RepQCtrAddrHigh32);
 	writel(h->reply_pool_dhandle, &h->transtable->RepQAddr0Low32);
 	writel(0, &h->transtable->RepQAddr0High32);
-	writel(CFGTBL_Trans_Performant,
+	writel(CFGTBL_Trans_Performant | use_short_tags,
 			&(h->cfgtable->HostWrite.TransportRequest));
 
 	writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL);
@@ -3855,7 +3906,8 @@ static void __devinit cciss_put_controller_into_performant_mode(ctlr_info_t *h)
 	if ((h->reply_pool == NULL) || (h->blockFetchTable == NULL))
 		goto clean_up;
 
-	cciss_enter_performant_mode(h);
+	cciss_enter_performant_mode(h,
+		trans_support & CFGTBL_Trans_use_short_tags);
 
 	/* Change the access methods to the performant access methods */
 	h->access = SA5_performant_access;
diff --git a/drivers/block/cciss.h b/drivers/block/cciss.h
index 579f7491849..554bbd907d1 100644
--- a/drivers/block/cciss.h
+++ b/drivers/block/cciss.h
@@ -222,6 +222,7 @@ static void SA5_submit_command( ctlr_info_t *h, CommandList_struct *c)
 			h->ctlr, c->busaddr);
 #endif /* CCISS_DEBUG */
          writel(c->busaddr, h->vaddr + SA5_REQUEST_PORT_OFFSET);
+	readl(h->vaddr + SA5_REQUEST_PORT_OFFSET);
 	 h->commands_outstanding++;
 	 if ( h->commands_outstanding > h->max_outstanding)
 		h->max_outstanding = h->commands_outstanding;
diff --git a/drivers/block/cciss_cmd.h b/drivers/block/cciss_cmd.h
index 35463d2f0ee..cd441bef031 100644
--- a/drivers/block/cciss_cmd.h
+++ b/drivers/block/cciss_cmd.h
@@ -56,6 +56,7 @@
 
 #define CFGTBL_Trans_Simple     0x00000002l
 #define CFGTBL_Trans_Performant 0x00000004l
+#define CFGTBL_Trans_use_short_tags 0x20000000l
 
 #define CFGTBL_BusType_Ultra2   0x00000001l
 #define CFGTBL_BusType_Ultra3   0x00000002l
diff --git a/drivers/block/cciss_scsi.c b/drivers/block/cciss_scsi.c
index 727d0225b7d..df793803f5a 100644
--- a/drivers/block/cciss_scsi.c
+++ b/drivers/block/cciss_scsi.c
@@ -824,13 +824,18 @@ static void complete_scsi_command(CommandList_struct *c, int timeout,
 			break;
 			case CMD_UNSOLICITED_ABORT:
 				cmd->result = DID_ABORT << 16;
-				dev_warn(&h->pdev->dev, "%p aborted do to an "
+				dev_warn(&h->pdev->dev, "%p aborted due to an "
 					"unsolicited abort\n", c);
 			break;
 			case CMD_TIMEOUT:
 				cmd->result = DID_TIME_OUT << 16;
 				dev_warn(&h->pdev->dev, "%p timedout\n", c);
 			break;
+			case CMD_UNABORTABLE:
+				cmd->result = DID_ERROR << 16;
+				dev_warn(&h->pdev->dev, "c %p command "
+					"unabortable\n", c);
+			break;
 			default:
 				cmd->result = DID_ERROR << 16;
 				dev_warn(&h->pdev->dev,
@@ -1007,11 +1012,15 @@ cciss_scsi_interpret_error(ctlr_info_t *h, CommandList_struct *c)
 		break;
 		case CMD_UNSOLICITED_ABORT:
 			dev_warn(&h->pdev->dev,
-				"%p aborted do to an unsolicited abort\n", c);
+				"%p aborted due to an unsolicited abort\n", c);
 		break;
 		case CMD_TIMEOUT:
 			dev_warn(&h->pdev->dev, "%p timedout\n", c);
 		break;
+		case CMD_UNABORTABLE:
+			dev_warn(&h->pdev->dev,
+				"%p unabortable\n", c);
+		break;
 		default:
 			dev_warn(&h->pdev->dev,
 				"%p returned unknown status %x\n",
diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c
index aca302492ff..2a1642bc451 100644
--- a/drivers/block/drbd/drbd_actlog.c
+++ b/drivers/block/drbd/drbd_actlog.c
@@ -92,7 +92,7 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 	bio->bi_end_io = drbd_md_io_complete;
 	bio->bi_rw = rw;
 
-	if (FAULT_ACTIVE(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD))
+	if (drbd_insert_fault(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD))
 		bio_endio(bio, -EIO);
 	else
 		submit_bio(rw, bio);
@@ -176,13 +176,17 @@ static struct lc_element *_al_get(struct drbd_conf *mdev, unsigned int enr)
 	struct lc_element *al_ext;
 	struct lc_element *tmp;
 	unsigned long     al_flags = 0;
+	int wake;
 
 	spin_lock_irq(&mdev->al_lock);
 	tmp = lc_find(mdev->resync, enr/AL_EXT_PER_BM_SECT);
 	if (unlikely(tmp != NULL)) {
 		struct bm_extent  *bm_ext = lc_entry(tmp, struct bm_extent, lce);
 		if (test_bit(BME_NO_WRITES, &bm_ext->flags)) {
+			wake = !test_and_set_bit(BME_PRIORITY, &bm_ext->flags);
 			spin_unlock_irq(&mdev->al_lock);
+			if (wake)
+				wake_up(&mdev->al_wait);
 			return NULL;
 		}
 	}
@@ -258,6 +262,33 @@ void drbd_al_complete_io(struct drbd_conf *mdev, sector_t sector)
 	spin_unlock_irqrestore(&mdev->al_lock, flags);
 }
 
+#if (PAGE_SHIFT + 3) < (AL_EXTENT_SHIFT - BM_BLOCK_SHIFT)
+/* Currently BM_BLOCK_SHIFT, BM_EXT_SHIFT and AL_EXTENT_SHIFT
+ * are still coupled, or assume too much about their relation.
+ * Code below will not work if this is violated.
+ * Will be cleaned up with some followup patch.
+ */
+# error FIXME
+#endif
+
+static unsigned int al_extent_to_bm_page(unsigned int al_enr)
+{
+	return al_enr >>
+		/* bit to page */
+		((PAGE_SHIFT + 3) -
+		/* al extent number to bit */
+		 (AL_EXTENT_SHIFT - BM_BLOCK_SHIFT));
+}
+
+static unsigned int rs_extent_to_bm_page(unsigned int rs_enr)
+{
+	return rs_enr >>
+		/* bit to page */
+		((PAGE_SHIFT + 3) -
+		/* al extent number to bit */
+		 (BM_EXT_SHIFT - BM_BLOCK_SHIFT));
+}
+
 int
 w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused)
 {
@@ -285,7 +316,7 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused)
 	 * For now, we must not write the transaction,
 	 * if we cannot write out the bitmap of the evicted extent. */
 	if (mdev->state.conn < C_CONNECTED && evicted != LC_FREE)
-		drbd_bm_write_sect(mdev, evicted/AL_EXT_PER_BM_SECT);
+		drbd_bm_write_page(mdev, al_extent_to_bm_page(evicted));
 
 	/* The bitmap write may have failed, causing a state change. */
 	if (mdev->state.disk < D_INCONSISTENT) {
@@ -334,7 +365,7 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused)
 		+ mdev->ldev->md.al_offset + mdev->al_tr_pos;
 
 	if (!drbd_md_sync_page_io(mdev, mdev->ldev, sector, WRITE))
-		drbd_chk_io_error(mdev, 1, TRUE);
+		drbd_chk_io_error(mdev, 1, true);
 
 	if (++mdev->al_tr_pos >
 	    div_ceil(mdev->act_log->nr_elements, AL_EXTENTS_PT))
@@ -511,225 +542,6 @@ cancel:
 	return 1;
 }
 
-static void atodb_endio(struct bio *bio, int error)
-{
-	struct drbd_atodb_wait *wc = bio->bi_private;
-	struct drbd_conf *mdev = wc->mdev;
-	struct page *page;
-	int uptodate = bio_flagged(bio, BIO_UPTODATE);
-
-	/* strange behavior of some lower level drivers...
-	 * fail the request by clearing the uptodate flag,
-	 * but do not return any error?! */
-	if (!error && !uptodate)
-		error = -EIO;
-
-	drbd_chk_io_error(mdev, error, TRUE);
-	if (error && wc->error == 0)
-		wc->error = error;
-
-	if (atomic_dec_and_test(&wc->count))
-		complete(&wc->io_done);
-
-	page = bio->bi_io_vec[0].bv_page;
-	put_page(page);
-	bio_put(bio);
-	mdev->bm_writ_cnt++;
-	put_ldev(mdev);
-}
-
-/* sector to word */
-#define S2W(s)	((s)<<(BM_EXT_SHIFT-BM_BLOCK_SHIFT-LN2_BPL))
-
-/* activity log to on disk bitmap -- prepare bio unless that sector
- * is already covered by previously prepared bios */
-static int atodb_prepare_unless_covered(struct drbd_conf *mdev,
-					struct bio **bios,
-					unsigned int enr,
-					struct drbd_atodb_wait *wc) __must_hold(local)
-{
-	struct bio *bio;
-	struct page *page;
-	sector_t on_disk_sector;
-	unsigned int page_offset = PAGE_SIZE;
-	int offset;
-	int i = 0;
-	int err = -ENOMEM;
-
-	/* We always write aligned, full 4k blocks,
-	 * so we can ignore the logical_block_size (for now) */
-	enr &= ~7U;
-	on_disk_sector = enr + mdev->ldev->md.md_offset
-			     + mdev->ldev->md.bm_offset;
-
-	D_ASSERT(!(on_disk_sector & 7U));
-
-	/* Check if that enr is already covered by an already created bio.
-	 * Caution, bios[] is not NULL terminated,
-	 * but only initialized to all NULL.
-	 * For completely scattered activity log,
-	 * the last invocation iterates over all bios,
-	 * and finds the last NULL entry.
-	 */
-	while ((bio = bios[i])) {
-		if (bio->bi_sector == on_disk_sector)
-			return 0;
-		i++;
-	}
-	/* bios[i] == NULL, the next not yet used slot */
-
-	/* GFP_KERNEL, we are not in the write-out path */
-	bio = bio_alloc(GFP_KERNEL, 1);
-	if (bio == NULL)
-		return -ENOMEM;
-
-	if (i > 0) {
-		const struct bio_vec *prev_bv = bios[i-1]->bi_io_vec;
-		page_offset = prev_bv->bv_offset + prev_bv->bv_len;
-		page = prev_bv->bv_page;
-	}
-	if (page_offset == PAGE_SIZE) {
-		page = alloc_page(__GFP_HIGHMEM);
-		if (page == NULL)
-			goto out_bio_put;
-		page_offset = 0;
-	} else {
-		get_page(page);
-	}
-
-	offset = S2W(enr);
-	drbd_bm_get_lel(mdev, offset,
-			min_t(size_t, S2W(8), drbd_bm_words(mdev) - offset),
-			kmap(page) + page_offset);
-	kunmap(page);
-
-	bio->bi_private = wc;
-	bio->bi_end_io = atodb_endio;
-	bio->bi_bdev = mdev->ldev->md_bdev;
-	bio->bi_sector = on_disk_sector;
-
-	if (bio_add_page(bio, page, 4096, page_offset) != 4096)
-		goto out_put_page;
-
-	atomic_inc(&wc->count);
-	/* we already know that we may do this...
-	 * get_ldev_if_state(mdev,D_ATTACHING);
-	 * just get the extra reference, so that the local_cnt reflects
-	 * the number of pending IO requests DRBD at its backing device.
-	 */
-	atomic_inc(&mdev->local_cnt);
-
-	bios[i] = bio;
-
-	return 0;
-
-out_put_page:
-	err = -EINVAL;
-	put_page(page);
-out_bio_put:
-	bio_put(bio);
-	return err;
-}
-
-/**
- * drbd_al_to_on_disk_bm() -  * Writes bitmap parts covered by active AL extents
- * @mdev:	DRBD device.
- *
- * Called when we detach (unconfigure) local storage,
- * or when we go from R_PRIMARY to R_SECONDARY role.
- */
-void drbd_al_to_on_disk_bm(struct drbd_conf *mdev)
-{
-	int i, nr_elements;
-	unsigned int enr;
-	struct bio **bios;
-	struct drbd_atodb_wait wc;
-
-	ERR_IF (!get_ldev_if_state(mdev, D_ATTACHING))
-		return; /* sorry, I don't have any act_log etc... */
-
-	wait_event(mdev->al_wait, lc_try_lock(mdev->act_log));
-
-	nr_elements = mdev->act_log->nr_elements;
-
-	/* GFP_KERNEL, we are not in anyone's write-out path */
-	bios = kzalloc(sizeof(struct bio *) * nr_elements, GFP_KERNEL);
-	if (!bios)
-		goto submit_one_by_one;
-
-	atomic_set(&wc.count, 0);
-	init_completion(&wc.io_done);
-	wc.mdev = mdev;
-	wc.error = 0;
-
-	for (i = 0; i < nr_elements; i++) {
-		enr = lc_element_by_index(mdev->act_log, i)->lc_number;
-		if (enr == LC_FREE)
-			continue;
-		/* next statement also does atomic_inc wc.count and local_cnt */
-		if (atodb_prepare_unless_covered(mdev, bios,
-						enr/AL_EXT_PER_BM_SECT,
-						&wc))
-			goto free_bios_submit_one_by_one;
-	}
-
-	/* unnecessary optimization? */
-	lc_unlock(mdev->act_log);
-	wake_up(&mdev->al_wait);
-
-	/* all prepared, submit them */
-	for (i = 0; i < nr_elements; i++) {
-		if (bios[i] == NULL)
-			break;
-		if (FAULT_ACTIVE(mdev, DRBD_FAULT_MD_WR)) {
-			bios[i]->bi_rw = WRITE;
-			bio_endio(bios[i], -EIO);
-		} else {
-			submit_bio(WRITE, bios[i]);
-		}
-	}
-
-	/* always (try to) flush bitmap to stable storage */
-	drbd_md_flush(mdev);
-
-	/* In case we did not submit a single IO do not wait for
-	 * them to complete. ( Because we would wait forever here. )
-	 *
-	 * In case we had IOs and they are already complete, there
-	 * is not point in waiting anyways.
-	 * Therefore this if () ... */
-	if (atomic_read(&wc.count))
-		wait_for_completion(&wc.io_done);
-
-	put_ldev(mdev);
-
-	kfree(bios);
-	return;
-
- free_bios_submit_one_by_one:
-	/* free everything by calling the endio callback directly. */
-	for (i = 0; i < nr_elements && bios[i]; i++)
-		bio_endio(bios[i], 0);
-
-	kfree(bios);
-
- submit_one_by_one:
-	dev_warn(DEV, "Using the slow drbd_al_to_on_disk_bm()\n");
-
-	for (i = 0; i < mdev->act_log->nr_elements; i++) {
-		enr = lc_element_by_index(mdev->act_log, i)->lc_number;
-		if (enr == LC_FREE)
-			continue;
-		/* Really slow: if we have al-extents 16..19 active,
-		 * sector 4 will be written four times! Synchronous! */
-		drbd_bm_write_sect(mdev, enr/AL_EXT_PER_BM_SECT);
-	}
-
-	lc_unlock(mdev->act_log);
-	wake_up(&mdev->al_wait);
-	put_ldev(mdev);
-}
-
 /**
  * drbd_al_apply_to_bm() - Sets the bitmap to diry(1) where covered ba active AL extents
  * @mdev:	DRBD device.
@@ -809,7 +621,7 @@ static int w_update_odbm(struct drbd_conf *mdev, struct drbd_work *w, int unused
 		return 1;
 	}
 
-	drbd_bm_write_sect(mdev, udw->enr);
+	drbd_bm_write_page(mdev, rs_extent_to_bm_page(udw->enr));
 	put_ldev(mdev);
 
 	kfree(udw);
@@ -889,7 +701,6 @@ static void drbd_try_clear_on_disk_bm(struct drbd_conf *mdev, sector_t sector,
 				dev_warn(DEV, "Kicking resync_lru element enr=%u "
 				     "out with rs_failed=%d\n",
 				     ext->lce.lc_number, ext->rs_failed);
-				set_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags);
 			}
 			ext->rs_left = rs_left;
 			ext->rs_failed = success ? 0 : count;
@@ -908,7 +719,6 @@ static void drbd_try_clear_on_disk_bm(struct drbd_conf *mdev, sector_t sector,
 				drbd_queue_work_front(&mdev->data.work, &udw->w);
 			} else {
 				dev_warn(DEV, "Could not kmalloc an udw\n");
-				set_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags);
 			}
 		}
 	} else {
@@ -919,6 +729,22 @@ static void drbd_try_clear_on_disk_bm(struct drbd_conf *mdev, sector_t sector,
 	}
 }
 
+void drbd_advance_rs_marks(struct drbd_conf *mdev, unsigned long still_to_go)
+{
+	unsigned long now = jiffies;
+	unsigned long last = mdev->rs_mark_time[mdev->rs_last_mark];
+	int next = (mdev->rs_last_mark + 1) % DRBD_SYNC_MARKS;
+	if (time_after_eq(now, last + DRBD_SYNC_MARK_STEP)) {
+		if (mdev->rs_mark_left[mdev->rs_last_mark] != still_to_go &&
+		    mdev->state.conn != C_PAUSED_SYNC_T &&
+		    mdev->state.conn != C_PAUSED_SYNC_S) {
+			mdev->rs_mark_time[next] = now;
+			mdev->rs_mark_left[next] = still_to_go;
+			mdev->rs_last_mark = next;
+		}
+	}
+}
+
 /* clear the bit corresponding to the piece of storage in question:
  * size byte of data starting from sector.  Only clear a bits of the affected
  * one ore more _aligned_ BM_BLOCK_SIZE blocks.
@@ -936,7 +762,7 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size,
 	int wake_up = 0;
 	unsigned long flags;
 
-	if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_SEGMENT_SIZE) {
+	if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_BIO_SIZE) {
 		dev_err(DEV, "drbd_set_in_sync: sector=%llus size=%d nonsense!\n",
 				(unsigned long long)sector, size);
 		return;
@@ -969,21 +795,9 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size,
 	 */
 	count = drbd_bm_clear_bits(mdev, sbnr, ebnr);
 	if (count && get_ldev(mdev)) {
-		unsigned long now = jiffies;
-		unsigned long last = mdev->rs_mark_time[mdev->rs_last_mark];
-		int next = (mdev->rs_last_mark + 1) % DRBD_SYNC_MARKS;
-		if (time_after_eq(now, last + DRBD_SYNC_MARK_STEP)) {
-			unsigned long tw = drbd_bm_total_weight(mdev);
-			if (mdev->rs_mark_left[mdev->rs_last_mark] != tw &&
-			    mdev->state.conn != C_PAUSED_SYNC_T &&
-			    mdev->state.conn != C_PAUSED_SYNC_S) {
-				mdev->rs_mark_time[next] = now;
-				mdev->rs_mark_left[next] = tw;
-				mdev->rs_last_mark = next;
-			}
-		}
+		drbd_advance_rs_marks(mdev, drbd_bm_total_weight(mdev));
 		spin_lock_irqsave(&mdev->al_lock, flags);
-		drbd_try_clear_on_disk_bm(mdev, sector, count, TRUE);
+		drbd_try_clear_on_disk_bm(mdev, sector, count, true);
 		spin_unlock_irqrestore(&mdev->al_lock, flags);
 
 		/* just wake_up unconditional now, various lc_chaged(),
@@ -998,27 +812,27 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size,
 /*
  * this is intended to set one request worth of data out of sync.
  * affects at least 1 bit,
- * and at most 1+DRBD_MAX_SEGMENT_SIZE/BM_BLOCK_SIZE bits.
+ * and at most 1+DRBD_MAX_BIO_SIZE/BM_BLOCK_SIZE bits.
  *
  * called by tl_clear and drbd_send_dblock (==drbd_make_request).
  * so this can be _any_ process.
  */
-void __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size,
+int __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size,
 			    const char *file, const unsigned int line)
 {
 	unsigned long sbnr, ebnr, lbnr, flags;
 	sector_t esector, nr_sectors;
-	unsigned int enr, count;
+	unsigned int enr, count = 0;
 	struct lc_element *e;
 
-	if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_SEGMENT_SIZE) {
+	if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_BIO_SIZE) {
 		dev_err(DEV, "sector: %llus, size: %d\n",
 			(unsigned long long)sector, size);
-		return;
+		return 0;
 	}
 
 	if (!get_ldev(mdev))
-		return; /* no disk, no metadata, no bitmap to set bits in */
+		return 0; /* no disk, no metadata, no bitmap to set bits in */
 
 	nr_sectors = drbd_get_capacity(mdev->this_bdev);
 	esector = sector + (size >> 9) - 1;
@@ -1048,6 +862,8 @@ void __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size,
 
 out:
 	put_ldev(mdev);
+
+	return count;
 }
 
 static
@@ -1128,7 +944,10 @@ int drbd_rs_begin_io(struct drbd_conf *mdev, sector_t sector)
 	unsigned int enr = BM_SECT_TO_EXT(sector);
 	struct bm_extent *bm_ext;
 	int i, sig;
+	int sa = 200; /* Step aside 200 times, then grab the extent and let app-IO wait.
+			 200 times -> 20 seconds. */
 
+retry:
 	sig = wait_event_interruptible(mdev->al_wait,
 			(bm_ext = _bme_get(mdev, enr)));
 	if (sig)
@@ -1139,16 +958,25 @@ int drbd_rs_begin_io(struct drbd_conf *mdev, sector_t sector)
 
 	for (i = 0; i < AL_EXT_PER_BM_SECT; i++) {
 		sig = wait_event_interruptible(mdev->al_wait,
-				!_is_in_al(mdev, enr * AL_EXT_PER_BM_SECT + i));
-		if (sig) {
+					       !_is_in_al(mdev, enr * AL_EXT_PER_BM_SECT + i) ||
+					       test_bit(BME_PRIORITY, &bm_ext->flags));
+
+		if (sig || (test_bit(BME_PRIORITY, &bm_ext->flags) && sa)) {
 			spin_lock_irq(&mdev->al_lock);
 			if (lc_put(mdev->resync, &bm_ext->lce) == 0) {
-				clear_bit(BME_NO_WRITES, &bm_ext->flags);
+				bm_ext->flags = 0; /* clears BME_NO_WRITES and eventually BME_PRIORITY */
 				mdev->resync_locked--;
 				wake_up(&mdev->al_wait);
 			}
 			spin_unlock_irq(&mdev->al_lock);
-			return -EINTR;
+			if (sig)
+				return -EINTR;
+			if (schedule_timeout_interruptible(HZ/10))
+				return -EINTR;
+			if (sa && --sa == 0)
+				dev_warn(DEV,"drbd_rs_begin_io() stepped aside for 20sec."
+					 "Resync stalled?\n");
+			goto retry;
 		}
 	}
 	set_bit(BME_LOCKED, &bm_ext->flags);
@@ -1291,8 +1119,7 @@ void drbd_rs_complete_io(struct drbd_conf *mdev, sector_t sector)
 	}
 
 	if (lc_put(mdev->resync, &bm_ext->lce) == 0) {
-		clear_bit(BME_LOCKED, &bm_ext->flags);
-		clear_bit(BME_NO_WRITES, &bm_ext->flags);
+		bm_ext->flags = 0; /* clear BME_LOCKED, BME_NO_WRITES and BME_PRIORITY */
 		mdev->resync_locked--;
 		wake_up(&mdev->al_wait);
 	}
@@ -1383,7 +1210,7 @@ void drbd_rs_failed_io(struct drbd_conf *mdev, sector_t sector, int size)
 	sector_t esector, nr_sectors;
 	int wake_up = 0;
 
-	if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_SEGMENT_SIZE) {
+	if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_BIO_SIZE) {
 		dev_err(DEV, "drbd_rs_failed_io: sector=%llus size=%d nonsense!\n",
 				(unsigned long long)sector, size);
 		return;
@@ -1420,7 +1247,7 @@ void drbd_rs_failed_io(struct drbd_conf *mdev, sector_t sector, int size)
 		mdev->rs_failed += count;
 
 		if (get_ldev(mdev)) {
-			drbd_try_clear_on_disk_bm(mdev, sector, count, FALSE);
+			drbd_try_clear_on_disk_bm(mdev, sector, count, false);
 			put_ldev(mdev);
 		}
 
diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c
index 0645ca829a9..f0ae63d2df6 100644
--- a/drivers/block/drbd/drbd_bitmap.c
+++ b/drivers/block/drbd/drbd_bitmap.c
@@ -28,18 +28,58 @@
 #include <linux/drbd.h>
 #include <linux/slab.h>
 #include <asm/kmap_types.h>
+
+#include <asm-generic/bitops/le.h>
+
 #include "drbd_int.h"
 
+
 /* OPAQUE outside this file!
  * interface defined in drbd_int.h
 
  * convention:
  * function name drbd_bm_... => used elsewhere, "public".
  * function name      bm_... => internal to implementation, "private".
+ */
+
+
+/*
+ * LIMITATIONS:
+ * We want to support >= peta byte of backend storage, while for now still using
+ * a granularity of one bit per 4KiB of storage.
+ * 1 << 50		bytes backend storage (1 PiB)
+ * 1 << (50 - 12)	bits needed
+ *	38 --> we need u64 to index and count bits
+ * 1 << (38 - 3)	bitmap bytes needed
+ *	35 --> we still need u64 to index and count bytes
+ *			(that's 32 GiB of bitmap for 1 PiB storage)
+ * 1 << (35 - 2)	32bit longs needed
+ *	33 --> we'd even need u64 to index and count 32bit long words.
+ * 1 << (35 - 3)	64bit longs needed
+ *	32 --> we could get away with a 32bit unsigned int to index and count
+ *	64bit long words, but I rather stay with unsigned long for now.
+ *	We probably should neither count nor point to bytes or long words
+ *	directly, but either by bitnumber, or by page index and offset.
+ * 1 << (35 - 12)
+ *	22 --> we need that much 4KiB pages of bitmap.
+ *	1 << (22 + 3) --> on a 64bit arch,
+ *	we need 32 MiB to store the array of page pointers.
+ *
+ * Because I'm lazy, and because the resulting patch was too large, too ugly
+ * and still incomplete, on 32bit we still "only" support 16 TiB (minus some),
+ * (1 << 32) bits * 4k storage.
+ *
 
- * Note that since find_first_bit returns int, at the current granularity of
- * the bitmap (4KB per byte), this implementation "only" supports up to
- * 1<<(32+12) == 16 TB...
+ * bitmap storage and IO:
+ *	Bitmap is stored little endian on disk, and is kept little endian in
+ *	core memory. Currently we still hold the full bitmap in core as long
+ *	as we are "attached" to a local disk, which at 32 GiB for 1PiB storage
+ *	seems excessive.
+ *
+ *	We plan to reduce the amount of in-core bitmap pages by pageing them in
+ *	and out against their on-disk location as necessary, but need to make
+ *	sure we don't cause too much meta data IO, and must not deadlock in
+ *	tight memory situations. This needs some