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

* 'for-linus' of git://git.kernel.dk/linux-2.6-block: (27 commits)
  block: remove unused copy_io_context()
  Documentation: remove anticipatory scheduler info
  block: remove REQ_HARDBARRIER
  ioprio: rcu_read_lock/unlock protect find_task_by_vpid call (V2)
  ioprio: fix RCU locking around task dereference
  block: ioctl: fix information leak to userland
  block: read i_size with i_size_read()
  cciss: fix proc warning on attempt to remove non-existant directory
  bio: take care not overflow page count when mapping/copying user data
  block: limit vec count in bio_kmalloc() and bio_alloc_map_data()
  block: take care not to overflow when calculating total iov length
  block: check for proper length of iov entries in blk_rq_map_user_iov()
  cciss: remove controllers supported by hpsa
  cciss: use usleep_range not msleep for small sleeps
  cciss: limit commands allocated on reset_devices
  cciss: Use kernel provided PCI state save and restore functions
  cciss: fix board status waiting code
  drbd: Removed checks for REQ_HARDBARRIER on incomming BIOs
  drbd: REQ_HARDBARRIER -> REQ_FUA transition for meta data accesses
  drbd: Removed the BIO_RW_BARRIER support form the receiver/epoch code
  ...

Documentation/block/switching-sched.txt

@@ -16,7 +16,7 @@ you can do so by typing:
 As of the Linux 2.6.10 kernel, it is now possible to change the
 IO scheduler for a given block device on the fly (thus making it possible,
 for instance, to set the CFQ scheduler for the system default, but
-set a specific device to use the anticipatory or noop schedulers - which
+set a specific device to use the deadline or noop schedulers - which
 can improve that device's throughput).
 
 To set a specific scheduler, simply do this:
@@ -31,7 +31,7 @@ a "cat /sys/block/DEV/queue/scheduler" - the list of valid names
 will be displayed, with the currently selected scheduler in brackets:
 
 # cat /sys/block/hda/queue/scheduler
-noop anticipatory deadline [cfq]
-# echo anticipatory > /sys/block/hda/queue/scheduler
+noop deadline [cfq]
+# echo deadline > /sys/block/hda/queue/scheduler
 # cat /sys/block/hda/queue/scheduler
-noop [anticipatory] deadline cfq
+noop [deadline] cfq

Documentation/kernel-parameters.txt

@@ -706,7 +706,7 @@ and is between 256 and 4096 characters. It is defined in the file
 			arch/x86/kernel/cpu/cpufreq/elanfreq.c.
 
 	elevator=	[IOSCHED]
-			Format: {"anticipatory" | "cfq" | "deadline" | "noop"}
+			Format: {"cfq" | "deadline" | "noop"}
 			See Documentation/block/as-iosched.txt and
 			Documentation/block/deadline-iosched.txt for details.
 

Documentation/rbtree.txt

@@ -21,8 +21,8 @@ three rotations, respectively, to balance the tree), with slightly slower
 To quote Linux Weekly News:
 
     There are a number of red-black trees in use in the kernel.
-    The anticipatory, deadline, and CFQ I/O schedulers all employ
-    rbtrees to track requests; the packet CD/DVD driver does the same.
+    The deadline and CFQ I/O schedulers employ rbtrees to
+    track requests; the packet CD/DVD driver does the same.
     The high-resolution timer code uses an rbtree to organize outstanding
     timer requests.  The ext3 filesystem tracks directory entries in a
     red-black tree.  Virtual memory areas (VMAs) are tracked with red-black

block/blk-core.c

@@ -1194,13 +1194,6 @@ static int __make_request(struct request_queue *q, struct bio *bio)
 	int where = ELEVATOR_INSERT_SORT;
 	int rw_flags;
 
-	/* REQ_HARDBARRIER is no more */
-	if (WARN_ONCE(bio->bi_rw & REQ_HARDBARRIER,
-		"block: HARDBARRIER is deprecated, use FLUSH/FUA instead\n")) {
-		bio_endio(bio, -EOPNOTSUPP);
-		return 0;
-	}
-
 	/*
 	 * low level driver can indicate that it wants pages above a
 	 * certain limit bounced to low memory (ie for highmem, or even
@@ -1351,7 +1344,7 @@ static void handle_bad_sector(struct bio *bio)
 			bdevname(bio->bi_bdev, b),
 			bio->bi_rw,
 			(unsigned long long)bio->bi_sector + bio_sectors(bio),
-			(long long)(bio->bi_bdev->bd_inode->i_size >> 9));
+			(long long)(i_size_read(bio->bi_bdev->bd_inode) >> 9));
 
 	set_bit(BIO_EOF, &bio->bi_flags);
 }
@@ -1404,7 +1397,7 @@ static inline int bio_check_eod(struct bio *bio, unsigned int nr_sectors)
 		return 0;
 
 	/* Test device or partition size, when known. */
-	maxsector = bio->bi_bdev->bd_inode->i_size >> 9;
+	maxsector = i_size_read(bio->bi_bdev->bd_inode) >> 9;
 	if (maxsector) {
 		sector_t sector = bio->bi_sector;
 

block/blk-ioc.c

@@ -153,20 +153,6 @@ struct io_context *get_io_context(gfp_t gfp_flags, int node)
 }
 EXPORT_SYMBOL(get_io_context);
 
-void copy_io_context(struct io_context **pdst, struct io_context **psrc)
-{
-	struct io_context *src = *psrc;
-	struct io_context *dst = *pdst;
-
-	if (src) {
-		BUG_ON(atomic_long_read(&src->refcount) == 0);
-		atomic_long_inc(&src->refcount);
-		put_io_context(dst);
-		*pdst = src;
-	}
-}
-EXPORT_SYMBOL(copy_io_context);
-
 static int __init blk_ioc_init(void)
 {
 	iocontext_cachep = kmem_cache_create("blkdev_ioc",

block/blk-map.c

@@ -205,6 +205,8 @@ int blk_rq_map_user_iov(struct request_queue *q, struct request *rq,
 			unaligned = 1;
 			break;
 		}
+		if (!iov[i].iov_len)
+			return -EINVAL;
 	}
 
 	if (unaligned || (q->dma_pad_mask & len) || map_data)

block/compat_ioctl.c

@@ -744,13 +744,13 @@ long compat_blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg)
 		bdi->ra_pages = (arg * 512) / PAGE_CACHE_SIZE;
 		return 0;
 	case BLKGETSIZE:
-		size = bdev->bd_inode->i_size;
+		size = i_size_read(bdev->bd_inode);
 		if ((size >> 9) > ~0UL)
 			return -EFBIG;
 		return compat_put_ulong(arg, size >> 9);
 
 	case BLKGETSIZE64_32:
-		return compat_put_u64(arg, bdev->bd_inode->i_size);
+		return compat_put_u64(arg, i_size_read(bdev->bd_inode));
 
 	case BLKTRACESETUP32:
 	case BLKTRACESTART: /* compatible */

block/elevator.c

@@ -429,7 +429,7 @@ void elv_dispatch_sort(struct request_queue *q, struct request *rq)
 	q->nr_sorted--;
 
 	boundary = q->end_sector;
-	stop_flags = REQ_SOFTBARRIER | REQ_HARDBARRIER | REQ_STARTED;
+	stop_flags = REQ_SOFTBARRIER | REQ_STARTED;
 	list_for_each_prev(entry, &q->queue_head) {
 		struct request *pos = list_entry_rq(entry);
 
@@ -691,7 +691,7 @@ void elv_insert(struct request_queue *q, struct request *rq, int where)
 void __elv_add_request(struct request_queue *q, struct request *rq, int where,
 		       int plug)
 {
-	if (rq->cmd_flags & (REQ_SOFTBARRIER | REQ_HARDBARRIER)) {
+	if (rq->cmd_flags & REQ_SOFTBARRIER) {
 		/* barriers are scheduling boundary, update end_sector */
 		if (rq->cmd_type == REQ_TYPE_FS ||
 		    (rq->cmd_flags & REQ_DISCARD)) {

block/ioctl.c

@@ -125,7 +125,7 @@ static int blk_ioctl_discard(struct block_device *bdev, uint64_t start,
 	start >>= 9;
 	len >>= 9;
 
-	if (start + len > (bdev->bd_inode->i_size >> 9))
+	if (start + len > (i_size_read(bdev->bd_inode) >> 9))
 		return -EINVAL;
 	if (secure)
 		flags |= BLKDEV_DISCARD_SECURE;
@@ -242,6 +242,7 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
 		 * We need to set the startsect first, the driver may
 		 * want to override it.
 		 */
+		memset(&geo, 0, sizeof(geo));
 		geo.start = get_start_sect(bdev);
 		ret = disk->fops->getgeo(bdev, &geo);
 		if (ret)
@@ -307,12 +308,12 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
 		ret = blkdev_reread_part(bdev);
 		break;
 	case BLKGETSIZE:
-		size = bdev->bd_inode->i_size;
+		size = i_size_read(bdev->bd_inode);
 		if ((size >> 9) > ~0UL)
 			return -EFBIG;
 		return put_ulong(arg, size >> 9);
 	case BLKGETSIZE64:
-		return put_u64(arg, bdev->bd_inode->i_size);
+		return put_u64(arg, i_size_read(bdev->bd_inode));
 	case BLKTRACESTART:
 	case BLKTRACESTOP:
 	case BLKTRACESETUP:

block/scsi_ioctl.c

@@ -321,33 +321,47 @@ static int sg_io(struct request_queue *q, struct gendisk *bd_disk,
 	if (hdr->iovec_count) {
 		const int size = sizeof(struct sg_iovec) * hdr->iovec_count;
 		size_t iov_data_len;
-		struct sg_iovec *iov;
+		struct sg_iovec *sg_iov;
+		struct iovec *iov;
+		int i;
 
-		iov = kmalloc(size, GFP_KERNEL);
-		if (!iov) {
+		sg_iov = kmalloc(size, GFP_KERNEL);
+		if (!sg_iov) {
 			ret = -ENOMEM;
 			goto out;
 		}
 
-		if (copy_from_user(iov, hdr->dxferp, size)) {
-			kfree(iov);
+		if (copy_from_user(sg_iov, hdr->dxferp, size)) {
+			kfree(sg_iov);
 			ret = -EFAULT;
 			goto out;
 		}
 
+		/*
+		 * Sum up the vecs, making sure they don't overflow
+		 */
+		iov = (struct iovec *) sg_iov;
+		iov_data_len = 0;
+		for (i = 0; i < hdr->iovec_count; i++) {
+			if (iov_data_len + iov[i].iov_len < iov_data_len) {
+				kfree(sg_iov);
+				ret = -EINVAL;
+				goto out;
+			}
+			iov_data_len += iov[i].iov_len;
+		}
+
 		/* SG_IO howto says that the shorter of the two wins */
-		iov_data_len = iov_length((struct iovec *)iov,
-					  hdr->iovec_count);
 		if (hdr->dxfer_len < iov_data_len) {
-			hdr->iovec_count = iov_shorten((struct iovec *)iov,
+			hdr->iovec_count = iov_shorten(iov,
 						       hdr->iovec_count,
 						       hdr->dxfer_len);
 			iov_data_len = hdr->dxfer_len;
 		}
 
-		ret = blk_rq_map_user_iov(q, rq, NULL, iov, hdr->iovec_count,
+		ret = blk_rq_map_user_iov(q, rq, NULL, sg_iov, hdr->iovec_count,
 					  iov_data_len, GFP_KERNEL);
-		kfree(iov);
+		kfree(sg_iov);
 	} else if (hdr->dxfer_len)
 		ret = blk_rq_map_user(q, rq, NULL, hdr->dxferp, hdr->dxfer_len,
 				      GFP_KERNEL);

drivers/block/aoe/aoeblk.c

@@ -180,9 +180,6 @@ aoeblk_make_request(struct request_queue *q, struct bio *bio)
 		BUG();
 		bio_endio(bio, -ENXIO);
 		return 0;
-	} else if (bio->bi_rw & REQ_HARDBARRIER) {
-		bio_endio(bio, -EOPNOTSUPP);
-		return 0;
 	} else if (bio->bi_io_vec == NULL) {
 		printk(KERN_ERR "aoe: bi_io_vec is NULL\n");
 		BUG();

drivers/block/cciss.c

@@ -113,6 +113,8 @@ static struct board_type products[] = {
 	{0x409D0E11, "Smart Array 6400 EM", &SA5_access},
 	{0x40910E11, "Smart Array 6i", &SA5_access},
 	{0x3225103C, "Smart Array P600", &SA5_access},
+	{0x3223103C, "Smart Array P800", &SA5_access},
+	{0x3234103C, "Smart Array P400", &SA5_access},
 	{0x3235103C, "Smart Array P400i", &SA5_access},
 	{0x3211103C, "Smart Array E200i", &SA5_access},
 	{0x3212103C, "Smart Array E200", &SA5_access},
@@ -3753,7 +3755,7 @@ static void __devinit cciss_wait_for_mode_change_ack(ctlr_info_t *h)
 	for (i = 0; i < MAX_CONFIG_WAIT; i++) {
 		if (!(readl(h->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq))
 			break;
-		msleep(10);
+		usleep_range(10000, 20000);
 	}
 }
 
@@ -3937,10 +3939,9 @@ static int __devinit cciss_lookup_board_id(struct pci_dev *pdev, u32 *board_id)
 	*board_id = ((subsystem_device_id << 16) & 0xffff0000) |
 			subsystem_vendor_id;
 
-	for (i = 0; i < ARRAY_SIZE(products); i++) {
+	for (i = 0; i < ARRAY_SIZE(products); i++)
 		if (*board_id == products[i].board_id)
 			return i;
-	}
 	dev_warn(&pdev->dev, "unrecognized board ID: 0x%08x, ignoring.\n",
 		*board_id);
 	return -ENODEV;
@@ -3971,18 +3972,31 @@ static int __devinit cciss_pci_find_memory_BAR(struct pci_dev *pdev,
 	return -ENODEV;
 }
 
-static int __devinit cciss_wait_for_board_ready(ctlr_info_t *h)
+static int __devinit cciss_wait_for_board_state(struct pci_dev *pdev,
+	void __iomem *vaddr, int wait_for_ready)
+#define BOARD_READY 1
+#define BOARD_NOT_READY 0
 {
-	int i;
+	int i, iterations;
 	u32 scratchpad;
 
-	for (i = 0; i < CCISS_BOARD_READY_ITERATIONS; i++) {
-		scratchpad = readl(h->vaddr + SA5_SCRATCHPAD_OFFSET);
+	if (wait_for_ready)
+		iterations = CCISS_BOARD_READY_ITERATIONS;
+	else
+		iterations = CCISS_BOARD_NOT_READY_ITERATIONS;
+
+	for (i = 0; i < iterations; i++) {
+		scratchpad = readl(vaddr + SA5_SCRATCHPAD_OFFSET);
+		if (wait_for_ready) {
 			if (scratchpad == CCISS_FIRMWARE_READY)
 				return 0;
+		} else {
+			if (scratchpad != CCISS_FIRMWARE_READY)
+				return 0;
+		}
 		msleep(CCISS_BOARD_READY_POLL_INTERVAL_MSECS);
 	}
-	dev_warn(&h->pdev->dev, "board not ready, timed out.\n");
+	dev_warn(&pdev->dev, "board not ready, timed out.\n");
 	return -ENODEV;
 }
 
@@ -4031,6 +4045,11 @@ static int __devinit cciss_find_cfgtables(ctlr_info_t *h)
 static void __devinit cciss_get_max_perf_mode_cmds(struct ctlr_info *h)
 {
 	h->max_commands = readl(&(h->cfgtable->MaxPerformantModeCommands));
+
+	/* Limit commands in memory limited kdump scenario. */
+	if (reset_devices && h->max_commands > 32)
+		h->max_commands = 32;
+
 	if (h->max_commands < 16) {
 		dev_warn(&h->pdev->dev, "Controller reports "
 			"max supported commands of %d, an obvious lie. "
@@ -4148,7 +4167,7 @@ static int __devinit cciss_pci_init(ctlr_info_t *h)
 		err = -ENOMEM;
 		goto err_out_free_res;
 	}
-	err = cciss_wait_for_board_ready(h);
+	err = cciss_wait_for_board_state(h->pdev, h->vaddr, BOARD_READY);
 	if (err)
 		goto err_out_free_res;
 	err = cciss_find_cfgtables(h);
@@ -4313,36 +4332,6 @@ static __devinit int cciss_message(struct pci_dev *pdev, unsigned char opcode, u
 #define cciss_soft_reset_controller(p) cciss_message(p, 1, 0)
 #define cciss_noop(p) cciss_message(p, 3, 0)
 
-static __devinit int cciss_reset_msi(struct pci_dev *pdev)
-{
-/* the #defines are stolen from drivers/pci/msi.h. */
-#define msi_control_reg(base)		(base + PCI_MSI_FLAGS)
-#define PCI_MSIX_FLAGS_ENABLE		(1 << 15)
-
-	int pos;
-	u16 control = 0;
-
-	pos = pci_find_capability(pdev, PCI_CAP_ID_MSI);
-	if (pos) {
-		pci_read_config_word(pdev, msi_control_reg(pos), &control);
-		if (control & PCI_MSI_FLAGS_ENABLE) {
-			dev_info(&pdev->dev, "resetting MSI\n");
-			pci_write_config_word(pdev, msi_control_reg(pos), control & ~PCI_MSI_FLAGS_ENABLE);
-		}
-	}
-
-	pos = pci_find_capability(pdev, PCI_CAP_ID_MSIX);
-	if (pos) {
-		pci_read_config_word(pdev, msi_control_reg(pos), &control);
-		if (control & PCI_MSIX_FLAGS_ENABLE) {
-			dev_info(&pdev->dev, "resetting MSI-X\n");
-			pci_write_config_word(pdev, msi_control_reg(pos), control & ~PCI_MSIX_FLAGS_ENABLE);
-		}
-	}
-
-	return 0;
-}
-
 static int cciss_controller_hard_reset(struct pci_dev *pdev,
 	void * __iomem vaddr, bool use_doorbell)
 {
@@ -4397,17 +4386,17 @@ static int cciss_controller_hard_reset(struct pci_dev *pdev,
  * states or using the doorbell register. */
 static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev)
 {
-	u16 saved_config_space[32];
 	u64 cfg_offset;
 	u32 cfg_base_addr;
 	u64 cfg_base_addr_index;
 	void __iomem *vaddr;
 	unsigned long paddr;
 	u32 misc_fw_support, active_transport;
-	int rc, i;
+	int rc;
 	CfgTable_struct __iomem *cfgtable;
 	bool use_doorbell;
 	u32 board_id;
+	u16 command_register;
 
 	/* For controllers as old a the p600, this is very nearly
 	 * the same thing as
@@ -4417,14 +4406,6 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev)
 	 * pci_set_power_state(pci_dev, PCI_D0);
 	 * pci_restore_state(pci_dev);
 	 *
-	 * but we can't use these nice canned kernel routines on
-	 * kexec, because they also check the MSI/MSI-X state in PCI
-	 * configuration space and do the wrong thing when it is
-	 * set/cleared.  Also, the pci_save/restore_state functions
-	 * violate the ordering requirements for restoring the
-	 * configuration space from the CCISS document (see the
-	 * comment below).  So we roll our own ....
-	 *
 	 * For controllers newer than the P600, the pci power state
 	 * method of resetting doesn't work so we have another way
 	 * using the doorbell register.
@@ -4443,8 +4424,13 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev)
 		return -ENODEV;
 	}
 
-	for (i = 0; i < 32; i++)
-		pci_read_config_word(pdev, 2*i, &saved_config_space[i]);
+	/* Save the PCI command register */
+	pci_read_config_word(pdev, 4, &command_register);
+	/* Turn the board off.  This is so that later pci_restore_state()
+	 * won't turn the board on before the rest of config space is ready.
+	 */
+	pci_disable_device(pdev);
+	pci_save_state(pdev);
 
 	/* find the first memory BAR, so we can find the cfg table */
 	rc = cciss_pci_find_memory_BAR(pdev, &paddr);
@@ -4479,26 +4465,32 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev)
 	rc = cciss_controller_hard_reset(pdev, vaddr, use_doorbell);
 	if (rc)
 		goto unmap_cfgtable;
-
-	/* Restore the PCI configuration space.  The Open CISS
-	 * Specification says, "Restore the PCI Configuration
-	 * Registers, offsets 00h through 60h. It is important to
-	 * restore the command register, 16-bits at offset 04h,
-	 * last. Do not restore the configuration status register,
-	 * 16-bits at offset 06h."  Note that the offset is 2*i.
-	 */
-	for (i = 0; i < 32; i++) {
-		if (i == 2 || i == 3)
-			continue;
-		pci_write_config_word(pdev, 2*i, saved_config_space[i]);
+	pci_restore_state(pdev);
+	rc = pci_enable_device(pdev);
+	if (rc) {
+		dev_warn(&pdev->dev, "failed to enable device.\n");
+		goto unmap_cfgtable;
 	}
-	wmb();
-	pci_write_config_word(pdev, 4, saved_config_space[2]);
+	pci_write_config_word(pdev, 4, command_register);
 
 	/* Some devices (notably the HP Smart Array 5i Controller)
 	   need a little pause here */
 	msleep(CCISS_POST_RESET_PAUSE_MSECS);
 
+	/* Wait for board to become not ready, then ready. */
+	dev_info(&pdev->dev, "Waiting for board to become ready.\n");
+	rc = cciss_wait_for_board_state(pdev, vaddr, BOARD_NOT_READY);
+	if (rc) /* Don't bail, might be E500, etc. which can't be reset */
+		dev_warn(&pdev->dev,
+			"failed waiting for board to become not ready\n");
+	rc = cciss_wait_for_board_state(pdev, vaddr, BOARD_READY);
+	if (rc) {
+		dev_warn(&pdev->dev,
+			"failed waiting for board to become ready\n");
+		goto unmap_cfgtable;
+	}
+	dev_info(&pdev->dev, "board ready.\n");
+
 	/* Controller should be in simple mode at this point.  If it's not,
 	 * It means we're on one of those controllers which doesn't support
 	 * the doorbell reset method and on which the PCI power management reset
@@ -4539,8 +4531,6 @@ static __devinit int cciss_init_reset_devices(struct pci_dev *pdev)
 		return 0; /* just try to do the kdump anyhow. */
 	if (rc)
 		return -ENODEV;
-	if (cciss_reset_msi(pdev))
-		return -ENODEV;
 
 	/* Now try to get the controller to respond to a no-op */
 	for (i = 0; i < CCISS_POST_RESET_NOOP_RETRIES; i++) {
@@ -4936,6 +4926,7 @@ static void __exit cciss_cleanup(void)
 		}
 	}
 	kthread_stop(cciss_scan_thread);
+	if (proc_cciss)
 		remove_proc_entry("driver/cciss", NULL);
 	bus_unregister(&cciss_bus_type);
 }

drivers/block/cciss.h

@@ -200,10 +200,14 @@ struct ctlr_info
  * the above.
  */
 #define CCISS_BOARD_READY_WAIT_SECS (120)
+#define CCISS_BOARD_NOT_READY_WAIT_SECS (10)
 #define CCISS_BOARD_READY_POLL_INTERVAL_MSECS (100)
 #define CCISS_BOARD_READY_ITERATIONS \
 	((CCISS_BOARD_READY_WAIT_SECS * 1000) / \
 		CCISS_BOARD_READY_POLL_INTERVAL_MSECS)
+#define CCISS_BOARD_NOT_READY_ITERATIONS \
+	((CCISS_BOARD_NOT_READY_WAIT_SECS * 1000) / \
+		CCISS_BOARD_READY_POLL_INTERVAL_MSECS)
 #define CCISS_POST_RESET_PAUSE_MSECS (3000)
 #define CCISS_POST_RESET_NOOP_INTERVAL_MSECS (1000)
 #define CCISS_POST_RESET_NOOP_RETRIES (12)

drivers/block/drbd/drbd_actlog.c

@@ -78,11 +78,10 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 	init_completion(&md_io.event);
 	md_io.error = 0;
 
-	if ((rw & WRITE) && !test_bit(MD_NO_BARRIER, &mdev->flags))
-		rw |= REQ_HARDBARRIER;
+	if ((rw & WRITE) && !test_bit(MD_NO_FUA, &mdev->flags))
+		rw |= REQ_FUA;
 	rw |= REQ_UNPLUG | REQ_SYNC;
 
- retry:
 	bio = bio_alloc(GFP_NOIO, 1);
 	bio->bi_bdev = bdev->md_bdev;
 	bio->bi_sector = sector;
@@ -100,17 +99,6 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 	wait_for_completion(&md_io.event);
 	ok = bio_flagged(bio, BIO_UPTODATE) && md_io.error == 0;
 
-	/* check for unsupported barrier op.
-	 * would rather check on EOPNOTSUPP, but that is not reliable.
-	 * don't try again for ANY return value != 0 */
-	if (unlikely((bio->bi_rw & REQ_HARDBARRIER) && !ok)) {
-		/* Try again with no barrier */
-		dev_warn(DEV, "Barriers not supported on meta data device - disabling\n");
-		set_bit(MD_NO_BARRIER, &mdev->flags);
-		rw &= ~REQ_HARDBARRIER;
-		bio_put(bio);
-		goto retry;
-	}
  out:
 	bio_put(bio);
 	return ok;
@@ -284,18 +272,32 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused)
 	u32 xor_sum = 0;
 
 	if (!get_ldev(mdev)) {
-		dev_err(DEV, "get_ldev() failed in w_al_write_transaction\n");
+		dev_err(DEV,
+			"disk is %s, cannot start al transaction (-%d +%d)\n",
+			drbd_disk_str(mdev->state.disk), evicted, new_enr);
 		complete(&((struct update_al_work *)w)->event);
 		return 1;
 	}
 	/* do we have to do a bitmap write, first?
 	 * TODO reduce maximum latency:
 	 * submit both bios, then wait for both,
-	 * instead of doing two synchronous sector writes. */
+	 * instead of doing two synchronous sector writes.
+	 * 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);
 
-	mutex_lock(&mdev->md_io_mutex); /* protects md_io_page, al_tr_cycle, ... */
+	/* The bitmap write may have failed, causing a state change. */
+	if (mdev->state.disk < D_INCONSISTENT) {
+		dev_err(DEV,
+			"disk is %s, cannot write al transaction (-%d +%d)\n",
+			drbd_disk_str(mdev->state.disk), evicted, new_enr);
+		complete(&((struct update_al_work *)w)->event);
+		put_ldev(mdev);
+		return 1;
+	}
+
+	mutex_lock(&mdev->md_io_mutex); /* protects md_io_buffer, al_tr_cycle, ... */
 	buffer = (struct al_transaction *)page_address(mdev->md_io_page);
 
 	buffer->magic = __constant_cpu_to_be32(DRBD_MAGIC);
@@ -739,7 +741,7 @@ void drbd_al_apply_to_bm(struct drbd_conf *mdev)
 	unsigned int enr;
 	unsigned long add = 0;
 	char ppb[10];
-	int i;
+	int i, tmp;
 
 	wait_event(mdev->al_wait, lc_try_lock(mdev->act_log));
 
@@ -747,7 +749,9 @@ void drbd_al_apply_to_bm(struct drbd_conf *mdev)
 		enr = lc_element_by_index(mdev->act_log, i)->lc_number;
 		if (enr == LC_FREE)
 			continue;
-		add += drbd_bm_ALe_set_all(mdev, enr);
+		tmp = drbd_bm_ALe_set_all(mdev, enr);
+		dynamic_dev_dbg(DEV, "AL: set %d bits in extent %u\n", tmp, enr);
+		add += tmp;
 	}
 
 	lc_unlock(mdev->act_log);

drivers/block/drbd/drbd_int.h

@@ -116,7 +116,7 @@ struct drbd_conf;
 
 #define ERR_IF(exp) if (({						\
 	int _b = (exp) != 0;						\
-	if (_b) dev_err(DEV, "%s: (%s) in %s:%d\n",	\
+	if (_b) dev_err(DEV, "ASSERT FAILED: %s: (%s) in %s:%d\n",	\
 			__func__, #exp, __FILE__, __LINE__);		\
 	_b;								\
 	}))
@@ -749,17 +749,12 @@ struct drbd_epoch {
 
 /* drbd_epoch flag bits */
 enum {
-	DE_BARRIER_IN_NEXT_EPOCH_ISSUED,
-	DE_BARRIER_IN_NEXT_EPOCH_DONE,
-	DE_CONTAINS_A_BARRIER,
 	DE_HAVE_BARRIER_NUMBER,
-	DE_IS_FINISHING,
 };
 
 enum epoch_event {
 	EV_PUT,
 	EV_GOT_BARRIER_NR,
-	EV_BARRIER_DONE,
 	EV_BECAME_LAST,
 	EV_CLEANUP = 32, /* used as flag */
 };
@@ -801,11 +796,6 @@ enum {
 	__EE_CALL_AL_COMPLETE_IO,
 	__EE_MAY_SET_IN_SYNC,
 
-	/* This epoch entry closes an epoch using a barrier.
-	 * On sucessful completion, the epoch is released,
-	 * and the P_BARRIER_ACK send. */
-	__EE_IS_BARRIER,
-
 	/* In case a barrier failed,
 	 * we need to resubmit without the barrier flag. */
 	__EE_RESUBMITTED,
@@ -820,7 +810,6 @@ enum {
 };
 #define EE_CALL_AL_COMPLETE_IO (1<<__EE_CALL_AL_COMPLETE_IO)
 #define EE_MAY_SET_IN_SYNC     (1<<__EE_MAY_SET_IN_SYNC)
-#define EE_IS_BARRIER          (1<<__EE_IS_BARRIER)
 #define	EE_RESUBMITTED         (1<<__EE_RESUBMITTED)
 #define EE_WAS_ERROR           (1<<__EE_WAS_ERROR)
 #define EE_HAS_DIGEST          (1<<__EE_HAS_DIGEST)
@@ -843,16 +832,15 @@ enum {
 				 * Gets cleared when the state.conn
 				 * goes into C_CONNECTED state. */
 	WRITE_BM_AFTER_RESYNC,	/* A kmalloc() during resync failed */
-	NO_BARRIER_SUPP,	/* underlying block device doesn't implement barriers */
 	CONSIDER_RESYNC,
 
-	MD_NO_BARRIER,		/* meta data device does not support barriers,
-				   so don't even try */
+	MD_NO_FUA,		/* Users wants us to not use FUA/FLUSH on meta data dev */
 	SUSPEND_IO,		/* suspend application io */
 	BITMAP_IO,		/* suspend application io;
 				   once no more io in flight, start bitmap io */
 	BITMAP_IO_QUEUED,       /* Started bitmap IO */
-	GO_DISKLESS,		/* Disk failed, local_cnt reached zero, we are going diskless */
+	GO_DISKLESS,		/* Disk is being detached, on io-error or admin request. */
+	WAS_IO_ERROR,		/* Local disk failed returned IO error */
 	RESYNC_AFTER_NEG,       /* Resync after online grow after the attach&negotiate finished. */
 	NET_CONGESTED,		/* The data socket is congested */
 
@@ -947,7 +935,6 @@ enum write_ordering_e {
 	WO_none,
 	WO_drain_io,
 	WO_bdev_flush,
-	WO_bio_barrier
 };
 
 struct fifo_buffer {
@@ -1281,6 +1268,7 @@ extern int drbd_bmio_set_n_write(struct drbd_conf *mdev);
 extern int drbd_bmio_clear_n_write(struct drbd_conf *mdev);
 extern int drbd_bitmap_io(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *), char *why);
 extern void drbd_go_diskless(struct drbd_conf *mdev);
+extern void drbd_ldev_destroy(struct drbd_conf *mdev);
 
 
 /* Meta data layout
@@ -1798,17 +1786,17 @@ static inline void __drbd_chk_io_error_(struct drbd_conf *mdev, int forcedetach,
 	case EP_PASS_ON:
 		if (!forcedetach) {
 			if (__ratelimit(&drbd_ratelimit_state))
-				dev_err(DEV, "Local IO failed in %s."
-					     "Passing error on...\n", where);
+				dev_err(DEV, "Local IO failed in %s.\n", where);
 			break;
 		}
 		/* NOTE fall through to detach case if forcedetach set */
 	case EP_DETACH:
 	case EP_CALL_HELPER:
+		set_bit(WAS_IO_ERROR, &mdev->flags);
 		if (mdev->state.disk > D_FAILED) {
 			_drbd_set_state(_NS(mdev, disk, D_FAILED), CS_HARD, NULL);
-			dev_err(DEV, "Local IO failed in %s."
-				     "Detaching...\n", where);
+			dev_err(DEV,
+				"Local IO failed in %s. Detaching...\n", where);
 		}
 		break;
 	}
@@ -1874,7 +1862,7 @@ static inline sector_t drbd_md_last_sector(struct drbd_backing_dev *bdev)
 static inline sector_t drbd_get_capacity(struct block_device *bdev)
 {
 	/* return bdev ? get_capacity(bdev->bd_disk) : 0; */
-	return bdev ? bdev->bd_inode->i_size >> 9 : 0;
+	return bdev ? i_size_read(bdev->bd_inode) >> 9 : 0;
 }
 
 /**
@@ -2127,7 +2115,11 @@ static inline void put_ldev(struct drbd_conf *mdev)
 	__release(local);
 	D_ASSERT(i >= 0);
 	if (i == 0) {
+		if (mdev->state.disk == D_DISKLESS)
+			/* even internal references gone, safe to destroy */
+			drbd_ldev_destroy(mdev);
 		if (mdev->state.disk == D_FAILED)
+			/* all application IO references gone. */
 			drbd_go_diskless(mdev);
 		wake_up(&mdev->misc_wait);
 	}
@@ -2138,6 +2130,10 @@ static inline int _get_ldev_if_state(struct drbd_conf *mdev, enum drbd_disk_stat
 {
 	int io_allowed;
 
+	/* never get a reference while D_DISKLESS */
+	if (mdev->state.disk == D_DISKLESS)
+		return 0;
+
 	atomic_inc(&mdev->local_cnt);
 	io_allowed = (mdev->state.disk >= mins);
 	if (!io_allowed)
@@ -2406,12 +2402,12 @@ static inline void drbd_md_flush(struct drbd_conf *mdev)
 {
 	int r;
 
-	if (test_bit(MD_NO_BARRIER, &mdev->flags))
+	if (test_bit(MD_NO_FUA, &mdev->flags))
 		return;
 
 	r = blkdev_issue_flush(mdev->ldev->md_bdev, GFP_KERNEL, NULL);
 	if (r) {
-		set_bit(MD_NO_BARRIER, &mdev->flags);
+		set_bit(MD_NO_FUA, &mdev->flags);
 		dev_err(DEV, "meta data flush failed with status %d, disabling md-flushes\n", r);
 	}
 }

drivers/block/drbd/drbd_main.c

@@ -835,6 +835,15 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state
 	    ns.conn != C_UNCONNECTED && ns.conn != C_DISCONNECTING && ns.conn <= C_TEAR_DOWN)
 		ns.conn = os.conn;
 
+	/* we cannot fail (again) if we already detached */
+	if (ns.disk == D_FAILED && os.disk == D_DISKLESS)
+		ns.disk = D_DISKLESS;
+
+	/* if we are only D_ATTACHING yet,
+	 * we can (and should) go directly to D_DISKLESS. */
+	if (ns.disk == D_FAILED && os.disk == D_ATTACHING)
+		ns.disk = D_DISKLESS;
+
 	/* After C_DISCONNECTING only C_STANDALONE may follow */
 	if (os.conn == C_DISCONNECTING && ns.conn != C_STANDALONE)
 		ns.conn = os.conn;
@@ -1056,7 +1065,15 @@ int __drbd_set_state(struct drbd_conf *mdev,
 	    !test_and_set_bit(CONFIG_PENDING, &mdev->flags))
 		set_bit(DEVICE_DYING, &mdev->flags);
 
-	mdev->state.i = ns.i;
+	/* if we are going -> D_FAILED or D_DISKLESS, grab one extra reference
+	 * on the ldev here, to be sure the transition -> D_DISKLESS resp.
+	 * drbd_ldev_destroy() won't happen before our corresponding
+	 * after_state_ch works run, where we put_ldev again. */
+	if ((os.disk != D_FAILED && ns.disk == D_FAILED) ||
+	    (os.disk != D_DISKLESS && ns.disk == D_DISKLESS))
+		atomic_inc(&mdev->local_cnt);
+
+	mdev->state = ns;
 	wake_up(&mdev->misc_wait);
 	wake_up(&mdev->state_wait);
 
@@ -1268,7 +1285,6 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 			if (test_bit(NEW_CUR_UUID, &mdev->flags)) {
 				drbd_uuid_new_current(mdev);
 				clear_bit(NEW_CUR_UUID, &mdev->flags);
-				drbd_md_sync(mdev);
 			}
 			spin_lock_irq(&mdev->req_lock);
 			_drbd_set_state(_NS(mdev, susp_fen, 0), CS_VERBOSE, NULL);
@@ -1365,35 +1381,46 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 	    os.disk > D_INCONSISTENT && ns.disk == D_INCONSISTENT)
 		drbd_queue_bitmap_io(mdev, &drbd_bmio_set_n_write, NULL, "set_n_write from invalidate");
 
-	/* first half of local IO error */
-	if (os.disk > D_FAILED && ns.disk == D_FAILED) {
-		enum drbd_io_error_p eh = EP_PASS_ON;
+	/* first half of local IO error, failure to attach,
+	 * or administrative detach */
+	if (os.disk != D_FAILED && ns.disk == D_FAILED) {
+		enum drbd_io_error_p eh;
+		int was_io_error;
+		/* corresponding get_ldev was in __drbd_set_state, to serialize
+		 * our cleanup here with the transition to D_DISKLESS,
+		 * so it is safe to dreference ldev here. */
+		eh = mdev->ldev->dc.on_io_error;
+		was_io_error = test_and_clear_bit(WAS_IO_ERROR, &mdev->flags);
+
+		/* current state still has to be D_FAILED,
+		 * there is only one way out: to D_DISKLESS,
+		 * and that may only happen after our put_ldev below. */
+		if (mdev->state.disk != D_FAILED)
+			dev_err(DEV,
+				"ASSERT FAILED: disk is %s during detach\n",
+				drbd_disk_str(mdev->state.disk));
 
 		if (drbd_send_state(mdev))
-			dev_warn(DEV, "Notified peer that my disk is broken.\n");
+			dev_warn(DEV, "Notified peer that I am detaching my disk\n");
 		else
-			dev_err(DEV, "Sending state for drbd_io_error() failed\n");
+			dev_err(DEV, "Sending state for detaching disk failed\n");
 
 		drbd_rs_cancel_all(mdev);
 
-		if (get_ldev_if_state(mdev, D_FAILED)) {
-			eh = mdev->ldev->dc.on_io_error;
+		/* In case we want to get something to stable storage still,
+		 * this may be the last chance.
+		 * Following put_ldev may transition to D_DISKLESS. */
+		drbd_md_sync(mdev);
 		put_ldev(mdev);
-		}
-		if (eh == EP_CALL_HELPER)
-			drbd_khelper(mdev, "local-io-error");
-	}
-
 
-	/* second half of local IO error handling,
-	 * after local_cnt references have reached zero: */
-	if (os.disk == D_FAILED && ns.disk == D_DISKLESS) {
-		mdev->rs_total = 0;
-		mdev->rs_failed = 0;
-		atomic_set(&mdev->rs_pending_cnt, 0);
+		if (was_io_error && eh == EP_CALL_HELPER)
+			drbd_khelper(mdev, "local-io-error");
 	}
 
-	if (os.disk > D_DISKLESS && ns.disk == D_DISKLESS) {
+        /* second half of local IO error, failure to attach,
+         * or administrative detach,
+         * after local_cnt references have reached zero again */
+        if (os.disk != D_DISKLESS && ns.disk == D_DISKLESS) {
                 /* We must still be diskless,
                  * re-attach has to be serialized with this! */
                 if (mdev->state.disk != D_DISKLESS)
@@ -1401,27 +1428,17 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
                                 "ASSERT FAILED: disk is %s while going diskless\n",
                                 drbd_disk_str(mdev->state.disk));
 
-		/* we cannot assert local_cnt == 0 here, as get_ldev_if_state
-		 * will inc/dec it frequently. Since we became D_DISKLESS, no
-		 * one has touched the protected members anymore, though, so we
-		 * are safe to free them here. */
+                mdev->rs_total = 0;
+                mdev->rs_failed = 0;
+                atomic_set(&mdev->rs_pending_cnt, 0);
+
 		if (drbd_send_state(mdev))
-			dev_warn(DEV, "Notified peer that I detached my disk.\n");
+			dev_warn(DEV, "Notified peer that I'm now diskless.\n");
 		else
-			dev_err(DEV, "Sending state for detach failed\n");
-
-		lc_destroy(mdev->resync);
-		mdev->resync = NULL;
-		lc_destroy(mdev->act_log);
-		mdev->act_log = NULL;
-		__no_warn(local,
-			drbd_free_bc(mdev->ldev);
-			mdev->ldev = NULL;);
-
-		if (mdev->md_io_tmpp) {
-			__free_page(mdev->md_io_tmpp);
-			mdev->md_io_tmpp = NULL;
-		}
+			dev_err(DEV, "Sending state for being diskless failed\n");
+		/* corresponding get_ldev in __drbd_set_state
+		 * this may finaly trigger drbd_ldev_destroy. */
+		put_ldev(mdev);
 	}
 
 	/* Disks got bigger while they were detached */
@@ -2772,11 +2789,6 @@ void drbd_init_set_defaults(struct drbd_conf *mdev)
 
 	drbd_set_defaults(mdev);
 
-	/* for now, we do NOT yet support it,
-	 * even though we start some framework
-	 * to eventually support barriers */
-	set_bit(NO_BARRIER_SUPP, &mdev->flags);
-
 	atomic_set(&mdev->ap_bio_cnt, 0);
 	atomic_set(&mdev->ap_pending_cnt, 0);
 	atomic_set(&mdev->rs_pending_cnt, 0);
@@ -2842,7 +2854,7 @@ void drbd_init_set_defaults(struct drbd_conf *mdev)
 	drbd_thread_init(mdev, &mdev->asender, drbd_asender);
 
 	mdev->agreed_pro_version = PRO_VERSION_MAX;
-	mdev->write_ordering = WO_bio_barrier;
+	mdev->write_ordering = WO_bdev_flush;
 	mdev->resync_wenr = LC_FREE;
 }
 
@@ -2899,7 +2911,6 @@ void drbd_mdev_cleanup(struct drbd_conf *mdev)
 	D_ASSERT(list_empty(&mdev->resync_work.list));
 	D_ASSERT(list_empty(&mdev->unplug_work.list));
 	D_ASSERT(list_empty(&mdev->go_diskless.list));
-
 }
 
 
@@ -3660,6 +3671,8 @@ void drbd_uuid_new_current(struct drbd_conf *mdev) __must_hold(local)
 
 	get_random_bytes(&val, sizeof(u64));
 	_drbd_uuid_set(mdev, UI_CURRENT, val);
+	/* get it to stable storage _now_ */
+	drbd_md_sync(mdev);
 }
 
 void drbd_uuid_set_bm(struct drbd_conf *mdev, u64 val) __must_hold(local)
@@ -3756,19 +3769,31 @@ static int w_bitmap_io(struct drbd_conf *mdev, struct drbd_work *w, int unused)
 	return 1;
 }
 
+void drbd_ldev_destroy(struct drbd_conf *mdev)
+{
+	lc_destroy(mdev->resync);
+	mdev->resync = NULL;
+	lc_destroy(mdev->act_log);
+	mdev->act_log = NULL;
+	__no_warn(local,
+		drbd_free_bc(mdev->ldev);
+		mdev->ldev = NULL;);
+
+	if (mdev->md_io_tmpp) {
+		__free_page(mdev->md_io_tmpp);
+		mdev->md_io_tmpp = NULL;
+	}
+	clear_bit(GO_DISKLESS, &mdev->flags);
+}
+
 static int w_go_diskless(struct drbd_conf *mdev, struct drbd_work *w, int unused)
 {
 	D_ASSERT(mdev->state.disk == D_FAILED);
 	/* we cannot assert local_cnt == 0 here, as get_ldev_if_state will
 	 * inc/dec it frequently. Once we are D_DISKLESS, no one will touch
-	 * the protected members anymore, though, so in the after_state_ch work
-	 * it will be safe to free them. */
+	 * the protected members anymore, though, so once put_ldev reaches zero
+	 * again, it will be safe to free them. */
 	drbd_force_state(mdev, NS(disk, D_DISKLESS));
-	/* We need to wait for return of references checked out while we still
-	 * have been D_FAILED, though (drbd_md_sync, bitmap io). */
-	wait_event(mdev->misc_wait, !atomic_read(&mdev->local_cnt));
-
-	clear_bit(GO_DISKLESS, &mdev->flags);
 	return 1;
 }
 
@@ -3777,9 +3802,6 @@ void drbd_go_diskless(struct drbd_conf *mdev)
 	D_ASSERT(mdev->state.disk == D_FAILED);
 	if (!test_and_set_bit(GO_DISKLESS, &mdev->flags))
 		drbd_queue_work(&mdev->data.work, &mdev->go_diskless);
-		/* don't drbd_queue_work_front,
-		 * we need to serialize with the after_state_ch work
-		 * of the -> D_FAILED transition. */
 }
 
 /**

drivers/block/drbd/drbd_nl.c

@@ -870,6 +870,11 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 		retcode = ERR_DISK_CONFIGURED;
 		goto fail;
 	}
+	/* It may just now have detached because of IO error.  Make sure
+	 * drbd_ldev_destroy is done already, we may end up here very fast,
+	 * e.g. if someone calls attach from the on-io-error handler,
+	 * to realize a "hot spare" feature (not that I'd recommend that) */
+	wait_event(mdev->misc_wait, !atomic_read(&mdev->local_cnt));
 
 	/* allocation not in the IO path, cqueue thread context */
 	nbc = kzalloc(sizeof(struct drbd_backing_dev), GFP_KERNEL);
@@ -1098,9 +1103,9 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	/* Reset the "barriers don't work" bits here, then force meta data to
 	 * be written, to ensure we determine if barriers are supported. */
 	if (nbc->dc.no_md_flush)
-		set_bit(MD_NO_BARRIER, &mdev->flags);
+		set_bit(MD_NO_FUA, &mdev->flags);
 	else
-		clear_bit(MD_NO_BARRIER, &mdev->flags);
+		clear_bit(MD_NO_FUA, &mdev->flags);
 
 	/* Point of no return reached.
 	 * Devices and memory are no longer released by error cleanup below.
@@ -1112,8 +1117,8 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	nbc = NULL;
 	resync_lru = NULL;
 
-	mdev->write_ordering = WO_bio_barrier;
-	drbd_bump_write_ordering(mdev, WO_bio_barrier);
+	mdev->write_ordering = WO_bdev_flush;
+	drbd_bump_write_ordering(mdev, WO_bdev_flush);
 
 	if (drbd_md_test_flag(mdev->ldev, MDF_CRASHED_PRIMARY))
 		set_bit(CRASHED_PRIMARY, &mdev->flags);
@@ -1262,7 +1267,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
  force_diskless_dec:
 	put_ldev(mdev);
  force_diskless:
-	drbd_force_state(mdev, NS(disk, D_DISKLESS));
+	drbd_force_state(mdev, NS(disk, D_FAILED));
 	drbd_md_sync(mdev);
  release_bdev2_fail:
 	if (nbc)
@@ -1285,10 +1290,19 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	return 0;
 }
 
+/* Detaching the disk is a process in multiple stages.  First we need to lock
+ * out application IO, in-flight IO, IO stuck in drbd_al_begin_io.
+ * Then we transition to D_DISKLESS, and wait for put_ldev() to return all
+ * internal references as well.
+ * Only then we have finally detached. */
 static int drbd_nl_detach(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp,
 			  struct drbd_nl_cfg_reply *reply)
 {
+	drbd_suspend_io(mdev); /* so no-one is stuck in drbd_al_begin_io */
 	reply->ret_code = drbd_request_state(mdev, NS(disk, D_DISKLESS));
+	if (mdev->state.disk == D_DISKLESS)
+		wait_event(mdev->misc_wait, !atomic_read(&mdev->local_cnt));
+	drbd_resume_io(mdev);
 	return 0;
 }
 
@@ -1953,7 +1967,6 @@ static int drbd_nl_resume_io(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	if (test_bit(NEW_CUR_UUID, &mdev->flags)) {
 		drbd_uuid_new_current(mdev);
 		clear_bit(NEW_CUR_UUID, &mdev->flags);
-		drbd_md_sync(mdev);
 	}
 	drbd_suspend_io(mdev);
 	reply->ret_code = drbd_request_state(mdev, NS3(susp, 0, susp_nod, 0, susp_fen, 0));

drivers/block/drbd/drbd_proc.c

@@ -158,7 +158,6 @@ static int drbd_seq_show(struct seq_file *seq, void *v)
 		[WO_none] = 'n',
 		[WO_drain_io] = 'd',
 		[WO_bdev_flush] = 'f',
-		[WO_bio_barrier] = 'b',
 	};
 
 	seq_printf(seq, "version: " REL_VERSION " (api:%d/proto:%d-%d)\n%s\n",

drivers/block/drbd/drbd_req.c

@@ -258,7 +258,7 @@ void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m)
 		if (!hlist_unhashed(&req->colision))
 			hlist_del(&req->colision);
 		else
-			D_ASSERT((s & RQ_NET_MASK) == 0);
+			D_ASSERT((s & (RQ_NET_MASK & ~RQ_NET_DONE)) == 0);
 
 		/* for writes we need to do some extra housekeeping */
 		if (rw == WRITE)
@@ -813,6 +813,7 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio)
 			     mdev->state.conn >= C_CONNECTED));
 
 	if (!(local || remote) && !is_susp(mdev->state)) {
+		if (__ratelimit(&drbd_ratelimit_state))
 			dev_err(DEV, "IO ERROR: neither local nor remote disk\n");
 		goto fail_free_complete;
 	}
@@ -942,12 +943,21 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio)
 	if (local) {
 		req->private_bio->bi_bdev = mdev->ldev->backing_bdev;
 
+		/* State may have changed since we grabbed our reference on the
+		 * mdev->ldev member. Double check, and short-circuit to endio.
+		 * In case the last activity log transaction failed to get on
+		 * stable storage, and this is a WRITE, we may not even submit
+		 * this bio. */
+		if (get_ldev(mdev)) {
 			if (FAULT_ACTIVE(mdev, rw == WRITE ? DRBD_FAULT_DT_WR
 					     : rw == READ  ? DRBD_FAULT_DT_RD
 					     :               DRBD_FAULT_DT_RA))
 				bio_endio(req->private_bio, -EIO);
 			else
 				generic_make_request(req->private_bio);
+			put_ldev(mdev);
+		} else
+			bio_endio(req->private_bio, -EIO);
 	}
 
 	/* we need to plug ALWAYS since we possibly need to kick lo_dev.
@@ -1022,20 +1032,6 @@ int drbd_make_request_26(struct request_queue *q, struct bio *bio)
 		return 0;
 	}
 
-	/* Reject barrier requests if we know the underlying device does
-	 * not support them.
-	 * XXX: Need to get this info from peer as well some how so we
-	 * XXX: reject if EITHER side/data/metadata area does not support them.
-	 *
-	 * because of those XXX, this is not yet enabled,
-	 * i.e. in drbd_init_set_defaults we set the NO_BARRIER_SUPP bit.
-	 */
-	if (unlikely(bio->bi_rw & REQ_HARDBARRIER) && test_bit(NO_BARRIER_SUPP, &mdev->flags)) {
-		/* dev_warn(DEV, "Rejecting barrier request as underlying device does not support\n"); */
-		bio_endio(bio, -EOPNOTSUPP);
-		return 0;
-	}
-
 	/*
 	 * what we "blindly" assume:
 	 */

drivers/block/drbd/drbd_worker.c

@@ -102,12 +102,6 @@ void drbd_endio_read_sec_final(struct drbd_epoch_entry *e) __releases(local)
 	put_ldev(mdev);
 }
 
-static int is_failed_barrier(int ee_flags)
-{
-	return (ee_flags & (EE_IS_BARRIER|EE_WAS_ERROR|EE_RESUBMITTED))
-			== (EE_IS_BARRIER|EE_WAS_ERROR);
-}
-
 /* writes on behalf of the partner, or resync writes,
  * "submitted" by the receiver, final stage.  */
 static void drbd_endio_write_sec_final(struct drbd_epoch_entry *e) __releases(local)
@@ -119,21 +113,6 @@ static void drbd_endio_write_sec_final(struct drbd_epoch_entry *e) __releases(lo
 	int is_syncer_req;
 	int do_al_complete_io;
 
-	/* if this is a failed barrier request, disable use of barriers,
-	 * and schedule for resubmission */
-	if (is_failed_barrier(e->flags)) {
-		drbd_bump_write_ordering(mdev, WO_bdev_flush);
-		spin_lock_irqsave(&mdev->req_lock, flags);
-		list_del(&e->w.list);
-		e->flags = (e->flags & ~EE_WAS_ERROR) | EE_RESUBMITTED;
-		e->w.cb = w_e_reissue;
-		/* put_ldev actually happens below, once we come here again. */
-		__release(local);
-		spin_unlock_irqrestore(&mdev->req_lock, flags);
-		drbd_queue_work(&mdev->data.work, &e->w);
-		return;
-	}
-
 	D_ASSERT(e->block_id != ID_VACANT);
 
 	/* after we moved e to done_ee,
@@ -925,7 +904,7 @@ int drbd_resync_finished(struct drbd_conf *mdev)
 	drbd_md_sync(mdev);
 
 	if (test_and_clear_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags)) {
-		dev_warn(DEV, "Writing the whole bitmap, due to failed kmalloc\n");
+		dev_info(DEV, "Writing the whole bitmap\n");
 		drbd_queue_bitmap_io(mdev, &drbd_bm_write, NULL, "write from resync_finished");
 	}
 

drivers/block/loop.c

@@ -481,12 +481,6 @@ static int do_bio_filebacked(struct loop_device *lo, struct bio *bio)
 	if (bio_rw(bio) == WRITE) {
 		struct file *file = lo->lo_backing_file;
 
-		/* REQ_HARDBARRIER is deprecated */
-		if (bio->bi_rw & REQ_HARDBARRIER) {
-			ret = -EOPNOTSUPP;
-			goto out;
-		}
-
 		if (bio->bi_rw & REQ_FLUSH) {
 			ret = vfs_fsync(file, 0);
 			if (unlikely(ret && ret != -EINVAL)) {

drivers/block/xen-blkfront.c

@@ -289,8 +289,6 @@ static int blkif_queue_request(struct request *req)
 
 	ring_req->operation = rq_data_dir(req) ?
 		BLKIF_OP_WRITE : BLKIF_OP_READ;
-	if (req->cmd_flags & REQ_HARDBARRIER)
-		ring_req->operation = BLKIF_OP_WRITE_BARRIER;
 
 	ring_req->nr_segments = blk_rq_map_sg(req->q, req, info->sg);
 	BUG_ON(ring_req->nr_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST);

drivers/md/md.c

@@ -706,7 +706,7 @@ static struct mdk_personality *find_pers(int level, char *clevel)
 /* return the offset of the super block in 512byte sectors */
 static inline sector_t calc_dev_sboffset(struct block_device *bdev)
 {
-	sector_t num_sectors = bdev->bd_inode->i_size / 512;
+	sector_t num_sectors = i_size_read(bdev->bd_inode) / 512;
 	return MD_NEW_SIZE_SECTORS(num_sectors);
 }
 
@@ -1386,7 +1386,7 @@ static int super_1_load(mdk_rdev_t *rdev, mdk_rdev_t *refdev, int minor_version)
 	 */
 	switch(minor_version) {
 	case 0:
-		sb_start = rdev->bdev->bd_inode->i_size >> 9;
+		sb_start = i_size_read(rdev->bdev->bd_inode) >> 9;
 		sb_start -= 8*2;
 		sb_start &= ~(sector_t)(4*2-1);
 		break;
@@ -1472,7 +1472,7 @@ static int super_1_load(mdk_rdev_t *rdev, mdk_rdev_t *refdev, int minor_version)
 			ret = 0;
 	}
 	if (minor_version)
-		rdev->sectors = (rdev->bdev->bd_inode->i_size >> 9) -
+		rdev->sectors = (i_size_read(rdev->bdev->bd_inode) >> 9) -
 			le64_to_cpu(sb->data_offset);
 	else
 		rdev->sectors = rdev->sb_start;
@@ -1680,7 +1680,7 @@ super_1_rdev_size_change(mdk_rdev_t *rdev, sector_t num_sectors)
 		return 0; /* component must fit device */
 	if (rdev->sb_start < rdev->data_offset) {
 		/* minor versions 1 and 2; superblock before data */
-		max_sectors = rdev->bdev->bd_inode->i_size >> 9;
+		max_sectors = i_size_read(rdev->bdev->bd_inode) >> 9;
 		max_sectors -= rdev->data_offset;
 		if (!num_sectors || num_sectors > max_sectors)
 			num_sectors = max_sectors;
@@ -1690,7 +1690,7 @@ super_1_rdev_size_change(mdk_rdev_t *rdev, sector_t num_sectors)
 	} else {
 		/* minor version 0; superblock after data */
 		sector_t sb_start;
-		sb_start = (rdev->bdev->bd_inode->i_size >> 9) - 8*2;
+		sb_start = (i_size_read(rdev->bdev->bd_inode) >> 9) - 8*2;
 		sb_start &= ~(sector_t)(4*2 - 1);
 		max_sectors = rdev->sectors + sb_start - rdev->sb_start;
 		if (!num_sectors || num_sectors > max_sectors)
@@ -2584,7 +2584,7 @@ rdev_size_store(mdk_rdev_t *rdev, const char *buf, size_t len)
 			if (!sectors)
 				return -EBUSY;
 		} else if (!sectors)
-			sectors = (rdev->bdev->bd_inode->i_size >> 9) -
+			sectors = (i_size_read(rdev->bdev->bd_inode) >> 9) -
 				rdev->data_offset;
 	}
 	if (sectors < my_mddev->dev_sectors)
@@ -2797,7 +2797,7 @@ static mdk_rdev_t *md_import_device(dev_t newdev, int super_format, int super_mi
 
 	kobject_init(&rdev->kobj, &rdev_ktype);
 
-	size = rdev->bdev->bd_inode->i_size >> BLOCK_SIZE_BITS;
+	size = i_size_read(rdev->bdev->bd_inode) >> BLOCK_SIZE_BITS;
 	if (!size) {
 		printk(KERN_WARNING 
 			"md: %s has zero or unknown size, marking faulty!\n",
@@ -5235,7 +5235,7 @@ static int add_new_disk(mddev_t * mddev, mdu_disk_info_t *info)
 
 		if (!mddev->persistent) {
 			printk(KERN_INFO "md: nonpersistent superblock ...\n");
-			rdev->sb_start = rdev->bdev->bd_inode->i_size / 512;
+			rdev->sb_start = i_size_read(rdev->bdev->bd_inode) / 512;
 		} else
 			rdev->sb_start = calc_dev_sboffset(rdev->bdev);
 		rdev->sectors = rdev->sb_start;
@@ -5306,7 +5306,7 @@ static int hot_add_disk(mddev_t * mddev, dev_t dev)
 	if (mddev->persistent)
 		rdev->sb_start = calc_dev_sboffset(rdev->bdev);
 	else
-		rdev->sb_start = rdev->bdev->bd_inode->i_size / 512;
+		rdev->sb_start = i_size_read(rdev->bdev->bd_inode) / 512;
 
 	rdev->sectors = rdev->sb_start;
 

drivers/scsi/scsi_error.c

@@ -320,17 +320,9 @@ static int scsi_check_sense(struct scsi_cmnd *scmd)
 				    "changed. The Linux SCSI layer does not "
 				    "automatically adjust these parameters.\n");
 
-		if (scmd->request->cmd_flags & REQ_HARDBARRIER)
 		/*
-			 * barrier requests should always retry on UA
-			 * otherwise block will get a spurious error
-			 */
-			return NEEDS_RETRY;
-		else
-			/*
-			 * for normal (non barrier) commands, pass the
-			 * UA upwards for a determination in the
-			 * completion functions
+		 * Pass the UA upwards for a determination in the completion
+		 * functions.
 		 */
 		return SUCCESS;
 

drivers/usb/storage/uas.c

@@ -331,9 +331,6 @@ static struct urb *uas_alloc_cmd_urb(struct uas_dev_info *devinfo, gfp_t gfp,
 
 	iu->iu_id = IU_ID_COMMAND;
 	iu->tag = cpu_to_be16(stream_id);
-	if (sdev->ordered_tags && (cmnd->request->cmd_flags & REQ_HARDBARRIER))
-		iu->prio_attr = UAS_ORDERED_TAG;
-	else
 	iu->prio_attr = UAS_SIMPLE_TAG;
 	iu->len = len;
 	int_to_scsilun(sdev->lun, &iu->lun);

fs/bio.c

@@ -370,6 +370,9 @@ struct bio *bio_kmalloc(gfp_t gfp_mask, int nr_iovecs)
 {
 	struct bio *bio;
 
+	if (nr_iovecs > UIO_MAXIOV)
+		return NULL;
+
 	bio = kmalloc(sizeof(struct bio) + nr_iovecs * sizeof(struct bio_vec),
 		      gfp_mask);
 	if (unlikely(!bio))
@@ -697,8 +700,12 @@ static void bio_free_map_data(struct bio_map_data *bmd)
 static struct bio_map_data *bio_alloc_map_data(int nr_segs, int iov_count,
 					       gfp_t gfp_mask)
 {
-	struct bio_map_data *bmd = kmalloc(sizeof(*bmd), gfp_mask);
+	struct bio_map_data *bmd;
 
+	if (iov_count > UIO_MAXIOV)
+		return NULL;
+
+	bmd = kmalloc(sizeof(*bmd), gfp_mask);
 	if (!bmd)
 		return NULL;
 
@@ -827,6 +834,12 @@ struct bio *bio_copy_user_iov(struct request_queue *q,
 		end = (uaddr + iov[i].iov_len + PAGE_SIZE - 1) >> PAGE_SHIFT;
 		start = uaddr >> PAGE_SHIFT;
 
+		/*
+		 * Overflow, abort
+		 */
+		if (end < start)
+			return ERR_PTR(-EINVAL);
+
 		nr_pages += end - start;
 		len += iov[i].iov_len;
 	}
@@ -955,6 +968,12 @@ static struct bio *__bio_map_user_iov(struct request_queue *q,
 		unsigned long end = (uaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
 		unsigned long start = uaddr >> PAGE_SHIFT;
 
+		/*
+		 * Overflow, abort
+		 */
+		if (end < start)
+			return ERR_PTR(-EINVAL);
+
 		nr_pages += end - start;
 		/*
 		 * buffer must be aligned to at least hardsector size for now

fs/ioprio.c

@@ -111,12 +111,14 @@ SYSCALL_DEFINE3(ioprio_set, int, which, int, who, int, ioprio)
 	read_lock(&tasklist_lock);
 	switch (which) {
 		case IOPRIO_WHO_PROCESS:
+			rcu_read_lock();
 			if (!who)
 				p = current;
 			else
 				p = find_task_by_vpid(who);
 			if (p)
 				ret = set_task_ioprio(p, ioprio);
+			rcu_read_unlock();
 			break;
 		case IOPRIO_WHO_PGRP:
 			if (!who)
@@ -139,7 +141,12 @@ SYSCALL_DEFINE3(ioprio_set, int, which, int, who, int, ioprio)
 				break;
 
 			do_each_thread(g, p) {
-				if (__task_cred(p)->uid != who)
+				int match;
+
+				rcu_read_lock();
+				match = __task_cred(p)->uid == who;
+				rcu_read_unlock();
+				if (!match)
 					continue;
 				ret = set_task_ioprio(p, ioprio);
 				if (ret)
@@ -200,12 +207,14 @@ SYSCALL_DEFINE2(ioprio_get, int, which, int, who)
 	read_lock(&tasklist_lock);
 	switch (which) {
 		case IOPRIO_WHO_PROCESS:
+			rcu_read_lock();
 			if (!who)
 				p = current;
 			else
 				p = find_task_by_vpid(who);
 			if (p)
 				ret = get_task_ioprio(p);
+			rcu_read_unlock();
 			break;
 		case IOPRIO_WHO_PGRP:
 			if (!who)
@@ -232,7 +241,12 @@ SYSCALL_DEFINE2(ioprio_get, int, which, int, who)
 				break;
 
 			do_each_thread(g, p) {
-				if (__task_cred(p)->uid != user->uid)
+				int match;
+
+				rcu_read_lock();
+				match = __task_cred(p)->uid == user->uid;
+				rcu_read_unlock();
+				if (!match)
 					continue;
 				tmpio = get_task_ioprio(p);
 				if (tmpio < 0)

include/linux/bio.h

@@ -66,10 +66,6 @@
 #define bio_offset(bio)		bio_iovec((bio))->bv_offset
 #define bio_segments(bio)	((bio)->bi_vcnt - (bio)->bi_idx)
 #define bio_sectors(bio)	((bio)->bi_size >> 9)
-#define bio_empty_barrier(bio) \
-	((bio->bi_rw & REQ_HARDBARRIER) && \
-	 !bio_has_data(bio) && \
-	 !(bio->bi_rw & REQ_DISCARD))
 
 static inline unsigned int bio_cur_bytes(struct bio *bio)
 {

include/linux/blk_types.h

@@ -122,7 +122,6 @@ enum rq_flag_bits {
 	__REQ_FAILFAST_TRANSPORT, /* no driver retries of transport errors */
 	__REQ_FAILFAST_DRIVER,	/* no driver retries of driver errors */
 
-	__REQ_HARDBARRIER,	/* may not be passed by drive either */
 	__REQ_SYNC,		/* request is sync (sync write or read) */
 	__REQ_META,		/* metadata io request */
 	__REQ_DISCARD,		/* request to discard sectors */
@@ -159,7 +158,6 @@ enum rq_flag_bits {
 #define REQ_FAILFAST_DEV	(1 << __REQ_FAILFAST_DEV)
 #define REQ_FAILFAST_TRANSPORT	(1 << __REQ_FAILFAST_TRANSPORT)
 #define REQ_FAILFAST_DRIVER	(1 << __REQ_FAILFAST_DRIVER)
-#define REQ_HARDBARRIER		(1 << __REQ_HARDBARRIER)
 #define REQ_SYNC		(1 << __REQ_SYNC)
 #define REQ_META		(1 << __REQ_META)
 #define REQ_DISCARD		(1 << __REQ_DISCARD)
@@ -168,8 +166,8 @@ enum rq_flag_bits {
 #define REQ_FAILFAST_MASK \
 	(REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT | REQ_FAILFAST_DRIVER)
 #define REQ_COMMON_MASK \
-	(REQ_WRITE | REQ_FAILFAST_MASK | REQ_HARDBARRIER | REQ_SYNC | \
-	 REQ_META | REQ_DISCARD | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
+	(REQ_WRITE | REQ_FAILFAST_MASK | REQ_SYNC | REQ_META | REQ_DISCARD | \
+	 REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
 #define REQ_CLONE_MASK		REQ_COMMON_MASK
 
 #define REQ_UNPLUG		(1 << __REQ_UNPLUG)

include/linux/blkdev.h

@@ -552,8 +552,7 @@ static inline void blk_clear_queue_full(struct request_queue *q, int sync)
  * it already be started by driver.
  */
 #define RQ_NOMERGE_FLAGS	\
-	(REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER | \
-	 REQ_FLUSH | REQ_FUA)
+	(REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
 #define rq_mergeable(rq)	\
 	(!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
 	 (((rq)->cmd_flags & REQ_DISCARD) || \

include/linux/drbd.h

@@ -53,7 +53,7 @@
 
 
 extern const char *drbd_buildtag(void);
-#define REL_VERSION "8.3.9rc2"
+#define REL_VERSION "8.3.9"
 #define API_VERSION 88
 #define PRO_VERSION_MIN 86
 #define PRO_VERSION_MAX 95

include/linux/iocontext.h

@@ -76,7 +76,6 @@ int put_io_context(struct io_context *ioc);
 void exit_io_context(struct task_struct *task);
 struct io_context *get_io_context(gfp_t gfp_flags, int node);
 struct io_context *alloc_io_context(gfp_t gfp_flags, int node);
-void copy_io_context(struct io_context **pdst, struct io_context **psrc);
 #else
 static inline void exit_io_context(struct task_struct *task)
 {

kernel/trace/blktrace.c

@@ -168,7 +168,6 @@ static int act_log_check(struct blk_trace *bt, u32 what, sector_t sector,
 static const u32 ddir_act[2] = { BLK_TC_ACT(BLK_TC_READ),
 				 BLK_TC_ACT(BLK_TC_WRITE) };
 
-#define BLK_TC_HARDBARRIER	BLK_TC_BARRIER
 #define BLK_TC_RAHEAD		BLK_TC_AHEAD
 
 /* The ilog2() calls fall out because they're constant */
@@ -196,7 +195,6 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 		return;
 
 	what |= ddir_act[rw & WRITE];
-	what |= MASK_TC_BIT(rw, HARDBARRIER);
 	what |= MASK_TC_BIT(rw, SYNC);
 	what |= MASK_TC_BIT(rw, RAHEAD);
 	what |= MASK_TC_BIT(rw, META);
@@ -1807,8 +1805,6 @@ void blk_fill_rwbs(char *rwbs, u32 rw, int bytes)
 
 	if (rw & REQ_RAHEAD)
 		rwbs[i++] = 'A';
-	if (rw & REQ_HARDBARRIER)
-		rwbs[i++] = 'B';
 	if (rw & REQ_SYNC)
 		rwbs[i++] = 'S';
 	if (rw & REQ_META)