Merge branch 'for-3.11/drivers' of git://git.kernel.dk/linux-block

Pull block IO driver bits from Jens Axboe:
 "As I mentioned in the core block pull request, due to real life
  circumstances the driver pull request would be late.  Now it looks
  like -rc2 late...  On the plus side, apart form the rsxx update, these
  are all things that I could argue could go in later in the cycle as
  they are fixes and not features.  So even though things are late, it's
  not ALL bad.

  The pull request contains:

   - Updates to bcache, all bug fixes, from Kent.

   - A pile of drbd bug fixes (no big features this time!).

   - xen blk front/back fixes.

   - rsxx driver updates, some of them deferred form 3.10.  So should be
     well cooked by now"

* 'for-3.11/drivers' of git://git.kernel.dk/linux-block: (63 commits)
  bcache: Allocation kthread fixes
  bcache: Fix GC_SECTORS_USED() calculation
  bcache: Journal replay fix
  bcache: Shutdown fix
  bcache: Fix a sysfs splat on shutdown
  bcache: Advertise that flushes are supported
  bcache: check for allocation failures
  bcache: Fix a dumb race
  bcache: Use standard utility code
  bcache: Update email address
  bcache: Delete fuzz tester
  bcache: Document shrinker reserve better
  bcache: FUA fixes
  drbd: Allow online change of al-stripes and al-stripe-size
  drbd: Constants should be UPPERCASE
  drbd: Ignore the exit code of a fence-peer handler if it returns too late
  drbd: Fix rcu_read_lock balance on error path
  drbd: fix error return code in drbd_init()
  drbd: Do not sleep inside rcu
  bcache: Refresh usage docs
  ...

Documentation/ABI/testing/sysfs-driver-xen-blkback

@@ -0,0 +1,17 @@
+What:           /sys/module/xen_blkback/parameters/max_buffer_pages
+Date:           March 2013
+KernelVersion:  3.11
+Contact:        Roger Pau Monné <roger.pau@citrix.com>
+Description:
+                Maximum number of free pages to keep in each block
+                backend buffer.
+
+What:           /sys/module/xen_blkback/parameters/max_persistent_grants
+Date:           March 2013
+KernelVersion:  3.11
+Contact:        Roger Pau Monné <roger.pau@citrix.com>
+Description:
+                Maximum number of grants to map persistently in
+                blkback. If the frontend tries to use more than
+                max_persistent_grants, the LRU kicks in and starts
+                removing 5% of max_persistent_grants every 100ms.

Documentation/ABI/testing/sysfs-driver-xen-blkfront

@@ -0,0 +1,10 @@
+What:           /sys/module/xen_blkfront/parameters/max
+Date:           June 2013
+KernelVersion:  3.11
+Contact:        Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
+Description:
+                Maximum number of segments that the frontend will negotiate
+                with the backend for indirect descriptors. The default value
+                is 32 - higher value means more potential throughput but more
+                memory usage. The backend picks the minimum of the frontend
+                and its default backend value.

Documentation/bcache.txt

@@ -46,29 +46,33 @@ you format your backing devices and cache device at the same time, you won't
 have to manually attach:
   make-bcache -B /dev/sda /dev/sdb -C /dev/sdc
 
-To make bcache devices known to the kernel, echo them to /sys/fs/bcache/register:
+bcache-tools now ships udev rules, and bcache devices are known to the kernel
+immediately.  Without udev, you can manually register devices like this:
 
   echo /dev/sdb > /sys/fs/bcache/register
   echo /dev/sdc > /sys/fs/bcache/register
 
-To register your bcache devices automatically, you could add something like
+Registering the backing device makes the bcache device show up in /dev; you can
-this to an init script:
+now format it and use it as normal. But the first time using a new bcache
+device, it'll be running in passthrough mode until you attach it to a cache.
+See the section on attaching.
 
-  echo /dev/sd* > /sys/fs/bcache/register_quiet
+The devices show up as:
 
-It'll look for bcache superblocks and ignore everything that doesn't have one.
+  /dev/bcache<N>
 
-Registering the backing device makes the bcache show up in /dev; you can now
+As well as (with udev):
-format it and use it as normal. But the first time using a new bcache device,
-it'll be running in passthrough mode until you attach it to a cache. See the
-section on attaching.
 
-The devices show up at /dev/bcacheN, and can be controlled via sysfs from
+  /dev/bcache/by-uuid/<uuid>
-/sys/block/bcacheN/bcache:
+  /dev/bcache/by-label/<label>
+
+To get started:
 
   mkfs.ext4 /dev/bcache0
   mount /dev/bcache0 /mnt
 
+You can control bcache devices through sysfs at /sys/block/bcache<N>/bcache .
+
 Cache devices are managed as sets; multiple caches per set isn't supported yet
 but will allow for mirroring of metadata and dirty data in the future. Your new
 cache set shows up as /sys/fs/bcache/<UUID>
@@ -80,11 +84,11 @@ must be attached to your cache set to enable caching. Attaching a backing
 device to a cache set is done thusly, with the UUID of the cache set in
 /sys/fs/bcache:
 
-  echo <UUID> > /sys/block/bcache0/bcache/attach
+  echo <CSET-UUID> > /sys/block/bcache0/bcache/attach
 
 This only has to be done once. The next time you reboot, just reregister all
 your bcache devices. If a backing device has data in a cache somewhere, the
-/dev/bcache# device won't be created until the cache shows up - particularly
+/dev/bcache<N> device won't be created until the cache shows up - particularly
 important if you have writeback caching turned on.
 
 If you're booting up and your cache device is gone and never coming back, you
@@ -191,6 +195,9 @@ want for getting the best possible numbers when benchmarking.
 
 SYSFS - BACKING DEVICE:
 
+Available at /sys/block/<bdev>/bcache, /sys/block/bcache*/bcache and
+(if attached) /sys/fs/bcache/<cset-uuid>/bdev*
+
 attach
   Echo the UUID of a cache set to this file to enable caching.
 
@@ -300,6 +307,8 @@ cache_readaheads
 
 SYSFS - CACHE SET:
 
+Available at /sys/fs/bcache/<cset-uuid>
+
 average_key_size
   Average data per key in the btree.
 
@@ -390,6 +399,8 @@ trigger_gc
 
 SYSFS - CACHE DEVICE:
 
+Available at /sys/block/<cdev>/bcache
+
 block_size
   Minimum granularity of writes - should match hardware sector size.
 

MAINTAINERS

@@ -1642,7 +1642,7 @@ S:	Maintained
 F:	drivers/net/hamradio/baycom*
 
 BCACHE (BLOCK LAYER CACHE)
-M:	Kent Overstreet <koverstreet@google.com>
+M:	Kent Overstreet <kmo@daterainc.com>
 L:	linux-bcache@vger.kernel.org
 W:	http://bcache.evilpiepirate.org
 S:	Maintained:
@@ -3346,7 +3346,7 @@ F:	Documentation/firmware_class/
 F:	drivers/base/firmware*.c
 F:	include/linux/firmware.h
 
-FLASHSYSTEM DRIVER (IBM FlashSystem 70/80 PCI SSD Flash Card)
+FLASH ADAPTER DRIVER (IBM Flash Adapter 900GB Full Height PCI Flash Card)
 M:	Joshua Morris <josh.h.morris@us.ibm.com>
 M:	Philip Kelleher <pjk1939@linux.vnet.ibm.com>
 S:	Maintained

drivers/block/Kconfig

@@ -532,11 +532,11 @@ config BLK_DEV_RBD
 	  If unsure, say N.
 
 config BLK_DEV_RSXX
-	tristate "IBM FlashSystem 70/80 PCIe SSD Device Driver"
+	tristate "IBM Flash Adapter 900GB Full Height PCIe Device Driver"
 	depends on PCI
 	help
 	  Device driver for IBM's high speed PCIe SSD
-	  storage devices: FlashSystem-70 and FlashSystem-80.
+	  storage device: Flash Adapter 900GB Full Height.
 
 	  To compile this driver as a module, choose M here: the
 	  module will be called rsxx.

drivers/block/drbd/drbd_actlog.c

@@ -659,6 +659,27 @@ void drbd_al_shrink(struct drbd_conf *mdev)
 	wake_up(&mdev->al_wait);
 }
 
+int drbd_initialize_al(struct drbd_conf *mdev, void *buffer)
+{
+	struct al_transaction_on_disk *al = buffer;
+	struct drbd_md *md = &mdev->ldev->md;
+	sector_t al_base = md->md_offset + md->al_offset;
+	int al_size_4k = md->al_stripes * md->al_stripe_size_4k;
+	int i;
+
+	memset(al, 0, 4096);
+	al->magic = cpu_to_be32(DRBD_AL_MAGIC);
+	al->transaction_type = cpu_to_be16(AL_TR_INITIALIZED);
+	al->crc32c = cpu_to_be32(crc32c(0, al, 4096));
+
+	for (i = 0; i < al_size_4k; i++) {
+		int err = drbd_md_sync_page_io(mdev, mdev->ldev, al_base + i * 8, WRITE);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
 static int w_update_odbm(struct drbd_work *w, int unused)
 {
 	struct update_odbm_work *udw = container_of(w, struct update_odbm_work, w);

drivers/block/drbd/drbd_int.h

@@ -832,6 +832,7 @@ struct drbd_tconn {			/* is a resource from the config file */
 	unsigned susp_nod:1;		/* IO suspended because no data */
 	unsigned susp_fen:1;		/* IO suspended because fence peer handler runs */
 	struct mutex cstate_mutex;	/* Protects graceful disconnects */
+	unsigned int connect_cnt;	/* Inc each time a connection is established */
 
 	unsigned long flags;
 	struct net_conf *net_conf;	/* content protected by rcu */
@@ -1132,6 +1133,7 @@ extern void drbd_mdev_cleanup(struct drbd_conf *mdev);
 void drbd_print_uuids(struct drbd_conf *mdev, const char *text);
 
 extern void conn_md_sync(struct drbd_tconn *tconn);
+extern void drbd_md_write(struct drbd_conf *mdev, void *buffer);
 extern void drbd_md_sync(struct drbd_conf *mdev);
 extern int  drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev);
 extern void drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local);
@@ -1466,8 +1468,16 @@ extern void drbd_suspend_io(struct drbd_conf *mdev);
 extern void drbd_resume_io(struct drbd_conf *mdev);
 extern char *ppsize(char *buf, unsigned long long size);
 extern sector_t drbd_new_dev_size(struct drbd_conf *, struct drbd_backing_dev *, sector_t, int);
-enum determine_dev_size { dev_size_error = -1, unchanged = 0, shrunk = 1, grew = 2 };
+enum determine_dev_size {
-extern enum determine_dev_size drbd_determine_dev_size(struct drbd_conf *, enum dds_flags) __must_hold(local);
+	DS_ERROR_SHRINK = -3,
+	DS_ERROR_SPACE_MD = -2,
+	DS_ERROR = -1,
+	DS_UNCHANGED = 0,
+	DS_SHRUNK = 1,
+	DS_GREW = 2
+};
+extern enum determine_dev_size
+drbd_determine_dev_size(struct drbd_conf *, enum dds_flags, struct resize_parms *) __must_hold(local);
 extern void resync_after_online_grow(struct drbd_conf *);
 extern void drbd_reconsider_max_bio_size(struct drbd_conf *mdev);
 extern enum drbd_state_rv drbd_set_role(struct drbd_conf *mdev,
@@ -1633,6 +1643,7 @@ extern int __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector,
 #define drbd_set_out_of_sync(mdev, sector, size) \
 	__drbd_set_out_of_sync(mdev, sector, size, __FILE__, __LINE__)
 extern void drbd_al_shrink(struct drbd_conf *mdev);
+extern int drbd_initialize_al(struct drbd_conf *, void *);
 
 /* drbd_nl.c */
 /* state info broadcast */

drivers/block/drbd/drbd_main.c

@@ -2762,8 +2762,6 @@ int __init drbd_init(void)
 	/*
 	 * allocate all necessary structs
 	 */
-	err = -ENOMEM;
-
 	init_waitqueue_head(&drbd_pp_wait);
 
 	drbd_proc = NULL; /* play safe for drbd_cleanup */
@@ -2773,6 +2771,7 @@ int __init drbd_init(void)
 	if (err)
 		goto fail;
 
+	err = -ENOMEM;
 	drbd_proc = proc_create_data("drbd", S_IFREG | S_IRUGO , NULL, &drbd_proc_fops, NULL);
 	if (!drbd_proc)	{
 		printk(KERN_ERR "drbd: unable to register proc file\n");
@@ -2803,7 +2802,6 @@ int __init drbd_init(void)
 fail:
 	drbd_cleanup();
 	if (err == -ENOMEM)
-		/* currently always the case */
 		printk(KERN_ERR "drbd: ran out of memory\n");
 	else
 		printk(KERN_ERR "drbd: initialization failure\n");
@@ -2881,34 +2879,14 @@ struct meta_data_on_disk {
 	u8 reserved_u8[4096 - (7*8 + 10*4)];
 } __packed;
 
-/**
+
- * drbd_md_sync() - Writes the meta data super block if the MD_DIRTY flag bit is set
+
- * @mdev:	DRBD device.
+void drbd_md_write(struct drbd_conf *mdev, void *b)
- */
-void drbd_md_sync(struct drbd_conf *mdev)
 {
-	struct meta_data_on_disk *buffer;
+	struct meta_data_on_disk *buffer = b;
 	sector_t sector;
 	int i;
 
-	/* Don't accidentally change the DRBD meta data layout. */
-	BUILD_BUG_ON(UI_SIZE != 4);
-	BUILD_BUG_ON(sizeof(struct meta_data_on_disk) != 4096);
-
-	del_timer(&mdev->md_sync_timer);
-	/* timer may be rearmed by drbd_md_mark_dirty() now. */
-	if (!test_and_clear_bit(MD_DIRTY, &mdev->flags))
-		return;
-
-	/* We use here D_FAILED and not D_ATTACHING because we try to write
-	 * metadata even if we detach due to a disk failure! */
-	if (!get_ldev_if_state(mdev, D_FAILED))
-		return;
-
-	buffer = drbd_md_get_buffer(mdev);
-	if (!buffer)
-		goto out;
-
 	memset(buffer, 0, sizeof(*buffer));
 
 	buffer->la_size_sect = cpu_to_be64(drbd_get_capacity(mdev->this_bdev));
@@ -2937,6 +2915,35 @@ void drbd_md_sync(struct drbd_conf *mdev)
 		dev_err(DEV, "meta data update failed!\n");
 		drbd_chk_io_error(mdev, 1, DRBD_META_IO_ERROR);
 	}
+}
+
+/**
+ * drbd_md_sync() - Writes the meta data super block if the MD_DIRTY flag bit is set
+ * @mdev:	DRBD device.
+ */
+void drbd_md_sync(struct drbd_conf *mdev)
+{
+	struct meta_data_on_disk *buffer;
+
+	/* Don't accidentally change the DRBD meta data layout. */
+	BUILD_BUG_ON(UI_SIZE != 4);
+	BUILD_BUG_ON(sizeof(struct meta_data_on_disk) != 4096);
+
+	del_timer(&mdev->md_sync_timer);
+	/* timer may be rearmed by drbd_md_mark_dirty() now. */
+	if (!test_and_clear_bit(MD_DIRTY, &mdev->flags))
+		return;
+
+	/* We use here D_FAILED and not D_ATTACHING because we try to write
+	 * metadata even if we detach due to a disk failure! */
+	if (!get_ldev_if_state(mdev, D_FAILED))
+		return;
+
+	buffer = drbd_md_get_buffer(mdev);
+	if (!buffer)
+		goto out;
+
+	drbd_md_write(mdev, buffer);
 
 	/* Update mdev->ldev->md.la_size_sect,
 	 * since we updated it on metadata. */

drivers/block/drbd/drbd_nl.c

@@ -417,6 +417,7 @@ static enum drbd_fencing_p highest_fencing_policy(struct drbd_tconn *tconn)
 
 bool conn_try_outdate_peer(struct drbd_tconn *tconn)
 {
+	unsigned int connect_cnt;
 	union drbd_state mask = { };
 	union drbd_state val = { };
 	enum drbd_fencing_p fp;
@@ -428,6 +429,10 @@ bool conn_try_outdate_peer(struct drbd_tconn *tconn)
 		return false;
 	}
 
+	spin_lock_irq(&tconn->req_lock);
+	connect_cnt = tconn->connect_cnt;
+	spin_unlock_irq(&tconn->req_lock);
+
 	fp = highest_fencing_policy(tconn);
 	switch (fp) {
 	case FP_NOT_AVAIL:
@@ -492,8 +497,14 @@ bool conn_try_outdate_peer(struct drbd_tconn *tconn)
 	   here, because we might were able to re-establish the connection in the
 	   meantime. */
 	spin_lock_irq(&tconn->req_lock);
-	if (tconn->cstate < C_WF_REPORT_PARAMS && !test_bit(STATE_SENT, &tconn->flags))
+	if (tconn->cstate < C_WF_REPORT_PARAMS && !test_bit(STATE_SENT, &tconn->flags)) {
+		if (tconn->connect_cnt != connect_cnt)
+			/* In case the connection was established and droped
+			   while the fence-peer handler was running, ignore it */
+			conn_info(tconn, "Ignoring fence-peer exit code\n");
+		else
 			_conn_request_state(tconn, mask, val, CS_VERBOSE);
+	}
 	spin_unlock_irq(&tconn->req_lock);
 
 	return conn_highest_pdsk(tconn) <= D_OUTDATED;
@@ -816,15 +827,20 @@ void drbd_resume_io(struct drbd_conf *mdev)
  * Returns 0 on success, negative return values indicate errors.
  * You should call drbd_md_sync() after calling this function.
  */
-enum determine_dev_size drbd_determine_dev_size(struct drbd_conf *mdev, enum dds_flags flags) __must_hold(local)
+enum determine_dev_size
+drbd_determine_dev_size(struct drbd_conf *mdev, enum dds_flags flags, struct resize_parms *rs) __must_hold(local)
 {
 	sector_t prev_first_sect, prev_size; /* previous meta location */
 	sector_t la_size_sect, u_size;
+	struct drbd_md *md = &mdev->ldev->md;
+	u32 prev_al_stripe_size_4k;
+	u32 prev_al_stripes;
 	sector_t size;
 	char ppb[10];
+	void *buffer;
 
 	int md_moved, la_size_changed;
-	enum determine_dev_size rv = unchanged;
+	enum determine_dev_size rv = DS_UNCHANGED;
 
 	/* race:
 	 * application request passes inc_ap_bio,
@@ -836,6 +852,11 @@ enum determine_dev_size drbd_determine_dev_size(struct drbd_conf *mdev, enum dds
 	 * still lock the act_log to not trigger ASSERTs there.
 	 */
 	drbd_suspend_io(mdev);
+	buffer = drbd_md_get_buffer(mdev); /* Lock meta-data IO */
+	if (!buffer) {
+		drbd_resume_io(mdev);
+		return DS_ERROR;
+	}
 
 	/* no wait necessary anymore, actually we could assert that */
 	wait_event(mdev->al_wait, lc_try_lock(mdev->act_log));
@@ -844,7 +865,17 @@ enum determine_dev_size drbd_determine_dev_size(struct drbd_conf *mdev, enum dds
 	prev_size = mdev->ldev->md.md_size_sect;
 	la_size_sect = mdev->ldev->md.la_size_sect;
 
-	/* TODO: should only be some assert here, not (re)init... */
+	if (rs) {
+		/* rs is non NULL if we should change the AL layout only */
+
+		prev_al_stripes = md->al_stripes;
+		prev_al_stripe_size_4k = md->al_stripe_size_4k;
+
+		md->al_stripes = rs->al_stripes;
+		md->al_stripe_size_4k = rs->al_stripe_size / 4;
+		md->al_size_4k = (u64)rs->al_stripes * rs->al_stripe_size / 4;
+	}
+
 	drbd_md_set_sector_offsets(mdev, mdev->ldev);
 
 	rcu_read_lock();
@@ -852,6 +883,21 @@ enum determine_dev_size drbd_determine_dev_size(struct drbd_conf *mdev, enum dds
 	rcu_read_unlock();
 	size = drbd_new_dev_size(mdev, mdev->ldev, u_size, flags & DDSF_FORCED);
 
+	if (size < la_size_sect) {
+		if (rs && u_size == 0) {
+			/* Remove "rs &&" later. This check should always be active, but
+			   right now the receiver expects the permissive behavior */
+			dev_warn(DEV, "Implicit shrink not allowed. "
+				 "Use --size=%llus for explicit shrink.\n",
+				 (unsigned long long)size);
+			rv = DS_ERROR_SHRINK;
+		}
+		if (u_size > size)
+			rv = DS_ERROR_SPACE_MD;
+		if (rv != DS_UNCHANGED)
+			goto err_out;
+	}
+
 	if (drbd_get_capacity(mdev->this_bdev) != size ||
 	    drbd_bm_capacity(mdev) != size) {
 		int err;
@@ -867,7 +913,7 @@ enum determine_dev_size drbd_determine_dev_size(struct drbd_conf *mdev, enum dds
 				    "Leaving size unchanged at size = %lu KB\n",
 				    (unsigned long)size);
 			}
-			rv = dev_size_error;
+			rv = DS_ERROR;
 		}
 		/* racy, see comments above. */
 		drbd_set_my_capacity(mdev, size);
@@ -875,38 +921,57 @@ enum determine_dev_size drbd_determine_dev_size(struct drbd_conf *mdev, enum dds
 		dev_info(DEV, "size = %s (%llu KB)\n", ppsize(ppb, size>>1),
 		     (unsigned long long)size>>1);
 	}
-	if (rv == dev_size_error)
+	if (rv <= DS_ERROR)
-		goto out;
+		goto err_out;
 
 	la_size_changed = (la_size_sect != mdev->ldev->md.la_size_sect);
 
 	md_moved = prev_first_sect != drbd_md_first_sector(mdev->ldev)
 		|| prev_size	   != mdev->ldev->md.md_size_sect;
 
-	if (la_size_changed || md_moved) {
+	if (la_size_changed || md_moved || rs) {
-		int err;
+		u32 prev_flags;
 
 		drbd_al_shrink(mdev); /* All extents inactive. */
+
+		prev_flags = md->flags;
+		md->flags &= ~MDF_PRIMARY_IND;
+		drbd_md_write(mdev, buffer);
+
 		dev_info(DEV, "Writing the whole bitmap, %s\n",
 			 la_size_changed && md_moved ? "size changed and md moved" :
 			 la_size_changed ? "size changed" : "md moved");
 		/* next line implicitly does drbd_suspend_io()+drbd_resume_io() */
-		err = drbd_bitmap_io(mdev, md_moved ? &drbd_bm_write_all : &drbd_bm_write,
+		drbd_bitmap_io(mdev, md_moved ? &drbd_bm_write_all : &drbd_bm_write,
 			       "size changed", BM_LOCKED_MASK);
-		if (err) {
+		drbd_initialize_al(mdev, buffer);
-			rv = dev_size_error;
+
-			goto out;
+		md->flags = prev_flags;
-		}
+		drbd_md_write(mdev, buffer);
-		drbd_md_mark_dirty(mdev);
+
+		if (rs)
+			dev_info(DEV, "Changed AL layout to al-stripes = %d, al-stripe-size-kB = %d\n",
+				 md->al_stripes, md->al_stripe_size_4k * 4);
 	}
 
 	if (size > la_size_sect)
-		rv = grew;
+		rv = DS_GREW;
 	if (size < la_size_sect)
-		rv = shrunk;
+		rv = DS_SHRUNK;
-out:
+
+	if (0) {
+	err_out:
+		if (rs) {
+			md->al_stripes = prev_al_stripes;
+			md->al_stripe_size_4k = prev_al_stripe_size_4k;
+			md->al_size_4k = (u64)prev_al_stripes * prev_al_stripe_size_4k;
+
+			drbd_md_set_sector_offsets(mdev, mdev->ldev);
+		}
+	}
 	lc_unlock(mdev->act_log);
 	wake_up(&mdev->al_wait);
+	drbd_md_put_buffer(mdev);
 	drbd_resume_io(mdev);
 
 	return rv;
@@ -1607,11 +1672,11 @@ int drbd_adm_attach(struct sk_buff *skb, struct genl_info *info)
 	    !drbd_md_test_flag(mdev->ldev, MDF_CONNECTED_IND))
 		set_bit(USE_DEGR_WFC_T, &mdev->flags);
 
-	dd = drbd_determine_dev_size(mdev, 0);
+	dd = drbd_determine_dev_size(mdev, 0, NULL);
-	if (dd == dev_size_error) {
+	if (dd <= DS_ERROR) {
 		retcode = ERR_NOMEM_BITMAP;
 		goto force_diskless_dec;
-	} else if (dd == grew)
+	} else if (dd == DS_GREW)
 		set_bit(RESYNC_AFTER_NEG, &mdev->flags);
 
 	if (drbd_md_test_flag(mdev->ldev, MDF_FULL_SYNC) ||
@@ -2305,6 +2370,7 @@ int drbd_adm_resize(struct sk_buff *skb, struct genl_info *info)
 	struct drbd_conf *mdev;
 	enum drbd_ret_code retcode;
 	enum determine_dev_size dd;
+	bool change_al_layout = false;
 	enum dds_flags ddsf;
 	sector_t u_size;
 	int err;
@@ -2315,31 +2381,33 @@ int drbd_adm_resize(struct sk_buff *skb, struct genl_info *info)
 	if (retcode != NO_ERROR)
 		goto fail;
 
+	mdev = adm_ctx.mdev;
+	if (!get_ldev(mdev)) {
+		retcode = ERR_NO_DISK;
+		goto fail;
+	}
+
 	memset(&rs, 0, sizeof(struct resize_parms));
+	rs.al_stripes = mdev->ldev->md.al_stripes;
+	rs.al_stripe_size = mdev->ldev->md.al_stripe_size_4k * 4;
 	if (info->attrs[DRBD_NLA_RESIZE_PARMS]) {
 		err = resize_parms_from_attrs(&rs, info);
 		if (err) {
 			retcode = ERR_MANDATORY_TAG;
 			drbd_msg_put_info(from_attrs_err_to_txt(err));
-			goto fail;
+			goto fail_ldev;
 		}
 	}
 
-	mdev = adm_ctx.mdev;
 	if (mdev->state.conn > C_CONNECTED) {
 		retcode = ERR_RESIZE_RESYNC;
-		goto fail;
+		goto fail_ldev;
 	}
 
 	if (mdev->state.role == R_SECONDARY &&
 	    mdev->state.peer == R_SECONDARY) {
 		retcode = ERR_NO_PRIMARY;
-		goto fail;
+		goto fail_ldev;
-	}
-
-	if (!get_ldev(mdev)) {
-		retcode = ERR_NO_DISK;
-		goto fail;
 	}
 
 	if (rs.no_resync && mdev->tconn->agreed_pro_version < 93) {
@@ -2358,6 +2426,28 @@ int drbd_adm_resize(struct sk_buff *skb, struct genl_info *info)
 		}
 	}
 
+	if (mdev->ldev->md.al_stripes != rs.al_stripes ||
+	    mdev->ldev->md.al_stripe_size_4k != rs.al_stripe_size / 4) {
+		u32 al_size_k = rs.al_stripes * rs.al_stripe_size;
+
+		if (al_size_k > (16 * 1024 * 1024)) {
+			retcode = ERR_MD_LAYOUT_TOO_BIG;
+			goto fail_ldev;
+		}
+
+		if (al_size_k < MD_32kB_SECT/2) {
+			retcode = ERR_MD_LAYOUT_TOO_SMALL;
+			goto fail_ldev;
+		}
+
+		if (mdev->state.conn != C_CONNECTED) {
+			retcode = ERR_MD_LAYOUT_CONNECTED;
+			goto fail_ldev;
+		}
+
+		change_al_layout = true;
+	}
+
 	if (mdev->ldev->known_size != drbd_get_capacity(mdev->ldev->backing_bdev))
 		mdev->ldev->known_size = drbd_get_capacity(mdev->ldev->backing_bdev);
 
@@ -2373,16 +2463,22 @@ int drbd_adm_resize(struct sk_buff *skb, struct genl_info *info)
 	}
 
 	ddsf = (rs.resize_force ? DDSF_FORCED : 0) | (rs.no_resync ? DDSF_NO_RESYNC : 0);
-	dd = drbd_determine_dev_size(mdev, ddsf);
+	dd = drbd_determine_dev_size(mdev, ddsf, change_al_layout ? &rs : NULL);
 	drbd_md_sync(mdev);
 	put_ldev(mdev);
-	if (dd == dev_size_error) {
+	if (dd == DS_ERROR) {
 		retcode = ERR_NOMEM_BITMAP;
 		goto fail;
+	} else if (dd == DS_ERROR_SPACE_MD) {
+		retcode = ERR_MD_LAYOUT_NO_FIT;
+		goto fail;
+	} else if (dd == DS_ERROR_SHRINK) {
+		retcode = ERR_IMPLICIT_SHRINK;
+		goto fail;
 	}
 
 	if (mdev->state.conn == C_CONNECTED) {
-		if (dd == grew)
+		if (dd == DS_GREW)
 			set_bit(RESIZE_PENDING, &mdev->flags);
 
 		drbd_send_uuids(mdev);
@@ -2658,7 +2754,6 @@ int nla_put_status_info(struct sk_buff *skb, struct drbd_conf *mdev,
 		const struct sib_info *sib)
 {
 	struct state_info *si = NULL; /* for sizeof(si->member); */
-	struct net_conf *nc;
 	struct nlattr *nla;
 	int got_ldev;
 	int err = 0;
@@ -2688,13 +2783,19 @@ int nla_put_status_info(struct sk_buff *skb, struct drbd_conf *mdev,
 		goto nla_put_failure;
 
 	rcu_read_lock();
-	if (got_ldev)
+	if (got_ldev) {
-		if (disk_conf_to_skb(skb, rcu_dereference(mdev->ldev->disk_conf), exclude_sensitive))
+		struct disk_conf *disk_conf;
-			goto nla_put_failure;
+
+		disk_conf = rcu_dereference(mdev->ldev->disk_conf);
+		err = disk_conf_to_skb(skb, disk_conf, exclude_sensitive);
+	}
+	if (!err) {
+		struct net_conf *nc;
 
 		nc = rcu_dereference(mdev->tconn->net_conf);
 		if (nc)
 			err = net_conf_to_skb(skb, nc, exclude_sensitive);
+	}
 	rcu_read_unlock();
 	if (err)
 		goto nla_put_failure;

drivers/block/drbd/drbd_receiver.c

@@ -1039,6 +1039,8 @@ randomize:
 	rcu_read_lock();
 	idr_for_each_entry(&tconn->volumes, mdev, vnr) {
 		kref_get(&mdev->kref);
+		rcu_read_unlock();
+
 		/* Prevent a race between resync-handshake and
 		 * being promoted to Primary.
 		 *
@@ -1049,8 +1051,6 @@ randomize:
 		mutex_lock(mdev->state_mutex);
 		mutex_unlock(mdev->state_mutex);
 
-		rcu_read_unlock();
-
 		if (discard_my_data)
 			set_bit(DISCARD_MY_DATA, &mdev->flags);
 		else
@@ -3545,7 +3545,7 @@ static int receive_sizes(struct drbd_tconn *tconn, struct packet_info *pi)
 {
 	struct drbd_conf *mdev;
 	struct p_sizes *p = pi->data;
-	enum determine_dev_size dd = unchanged;
+	enum determine_dev_size dd = DS_UNCHANGED;
 	sector_t p_size, p_usize, my_usize;
 	int ldsc = 0; /* local disk size changed */
 	enum dds_flags ddsf;
@@ -3617,9 +3617,9 @@ static int receive_sizes(struct drbd_tconn *tconn, struct packet_info *pi)
 
 	ddsf = be16_to_cpu(p->dds_flags);
 	if (get_ldev(mdev)) {
-		dd = drbd_determine_dev_size(mdev, ddsf);
+		dd = drbd_determine_dev_size(mdev, ddsf, NULL);
 		put_ldev(mdev);
-		if (dd == dev_size_error)
+		if (dd == DS_ERROR)
 			return -EIO;
 		drbd_md_sync(mdev);
 	} else {
@@ -3647,7 +3647,7 @@ static int receive_sizes(struct drbd_tconn *tconn, struct packet_info *pi)
 			drbd_send_sizes(mdev, 0, ddsf);
 		}
 		if (test_and_clear_bit(RESIZE_PENDING, &mdev->flags) ||
-		    (dd == grew && mdev->state.conn == C_CONNECTED)) {
+		    (dd == DS_GREW && mdev->state.conn == C_CONNECTED)) {
 			if (mdev->state.pdsk >= D_INCONSISTENT &&
 			    mdev->state.disk >= D_INCONSISTENT) {
 				if (ddsf & DDSF_NO_RESYNC)

drivers/block/drbd/drbd_state.c

@@ -1115,8 +1115,10 @@ __drbd_set_state(struct drbd_conf *mdev, union drbd_state ns,
 		drbd_thread_restart_nowait(&mdev->tconn->receiver);
 
 	/* Resume AL writing if we get a connection */
-	if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED)
+	if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) {
 		drbd_resume_al(mdev);
+		mdev->tconn->connect_cnt++;
+	}
 
 	/* remember last attach time so request_timer_fn() won't
 	 * kill newly established sessions while we are still trying to thaw

drivers/block/rsxx/core.c

@@ -31,6 +31,8 @@
 #include <linux/slab.h>
 #include <linux/bitops.h>
 #include <linux/delay.h>
+#include <linux/debugfs.h>
+#include <linux/seq_file.h>
 
 #include <linux/genhd.h>
 #include <linux/idr.h>
@@ -39,8 +41,9 @@
 #include "rsxx_cfg.h"
 
 #define NO_LEGACY 0
+#define SYNC_START_TIMEOUT (10 * 60) /* 10 minutes */
 
-MODULE_DESCRIPTION("IBM FlashSystem 70/80 PCIe SSD Device Driver");
+MODULE_DESCRIPTION("IBM Flash Adapter 900GB Full Height Device Driver");
 MODULE_AUTHOR("Joshua Morris/Philip Kelleher, IBM");
 MODULE_LICENSE("GPL");
 MODULE_VERSION(DRIVER_VERSION);
@@ -49,9 +52,282 @@ static unsigned int force_legacy = NO_LEGACY;
 module_param(force_legacy, uint, 0444);
 MODULE_PARM_DESC(force_legacy, "Force the use of legacy type PCI interrupts");
 
+static unsigned int sync_start = 1;
+module_param(sync_start, uint, 0444);
+MODULE_PARM_DESC(sync_start, "On by Default: Driver load will not complete "
+			     "until the card startup has completed.");
+
 static DEFINE_IDA(rsxx_disk_ida);
 static DEFINE_SPINLOCK(rsxx_ida_lock);
 
+/* --------------------Debugfs Setup ------------------- */
+
+struct rsxx_cram {
+	u32 f_pos;
+	u32 offset;
+	void *i_private;
+};
+
+static int rsxx_attr_pci_regs_show(struct seq_file *m, void *p)
+{
+	struct rsxx_cardinfo *card = m->private;
+
+	seq_printf(m, "HWID		0x%08x\n",
+					ioread32(card->regmap + HWID));
+	seq_printf(m, "SCRATCH		0x%08x\n",
+					ioread32(card->regmap + SCRATCH));
+	seq_printf(m, "IER		0x%08x\n",
+					ioread32(card->regmap + IER));
+	seq_printf(m, "IPR		0x%08x\n",
+					ioread32(card->regmap + IPR));
+	seq_printf(m, "CREG_CMD		0x%08x\n",
+					ioread32(card->regmap + CREG_CMD));
+	seq_printf(m, "CREG_ADD		0x%08x\n",
+					ioread32(card->regmap + CREG_ADD));
+	seq_printf(m, "CREG_CNT		0x%08x\n",
+					ioread32(card->regmap + CREG_CNT));
+	seq_printf(m, "CREG_STAT	0x%08x\n",
+					ioread32(card->regmap + CREG_STAT));
+	seq_printf(m, "CREG_DATA0	0x%08x\n",
+					ioread32(card->regmap + CREG_DATA0));
+	seq_printf(m, "CREG_DATA1	0x%08x\n",
+					ioread32(card->regmap + CREG_DATA1));
+	seq_printf(m, "CREG_DATA2	0x%08x\n",
+					ioread32(card->regmap + CREG_DATA2));
+	seq_printf(m, "CREG_DATA3	0x%08x\n",
+					ioread32(card->regmap + CREG_DATA3));
+	seq_printf(m, "CREG_DATA4	0x%08x\n",
+					ioread32(card->regmap + CREG_DATA4));
+	seq_printf(m, "CREG_DATA5	0x%08x\n",
+					ioread32(card->regmap + CREG_DATA5));
+	seq_printf(m, "CREG_DATA6	0x%08x\n",
+					ioread32(card->regmap + CREG_DATA6));
+	seq_printf(m, "CREG_DATA7	0x%08x\n",
+					ioread32(card->regmap + CREG_DATA7));
+	seq_printf(m, "INTR_COAL	0x%08x\n",
+					ioread32(card->regmap + INTR_COAL));
+	seq_printf(m, "HW_ERROR		0x%08x\n",
+					ioread32(card->regmap + HW_ERROR));
+	seq_printf(m, "DEBUG0		0x%08x\n",
+					ioread32(card->regmap + PCI_DEBUG0));
+	seq_printf(m, "DEBUG1		0x%08x\n",
+					ioread32(card->regmap + PCI_DEBUG1));
+	seq_printf(m, "DEBUG2		0x%08x\n",
+					ioread32(card->regmap + PCI_DEBUG2));
+	seq_printf(m, "DEBUG3		0x%08x\n",
+					ioread32(card->regmap + PCI_DEBUG3));
+	seq_printf(m, "DEBUG4		0x%08x\n",
+					ioread32(card->regmap + PCI_DEBUG4));
+	seq_printf(m, "DEBUG5		0x%08x\n",
+					ioread32(card->regmap + PCI_DEBUG5));
+	seq_printf(m, "DEBUG6		0x%08x\n",
+					ioread32(card->regmap + PCI_DEBUG6));
+	seq_printf(m, "DEBUG7		0x%08x\n",
+					ioread32(card->regmap + PCI_DEBUG7));
+	seq_printf(m, "RECONFIG		0x%08x\n",
+					ioread32(card->regmap + PCI_RECONFIG));
+
+	return 0;
+}
+
+static int rsxx_attr_stats_show(struct seq_file *m, void *p)
+{
+	struct rsxx_cardinfo *card = m->private;
+	int i;
+
+	for (i = 0; i < card->n_targets; i++) {
+		seq_printf(m, "Ctrl %d CRC Errors	= %d\n",
+				i, card->ctrl[i].stats.crc_errors);
+		seq_printf(m, "Ctrl %d Hard Errors	= %d\n",
+				i, card->ctrl[i].stats.hard_errors);
+		seq_printf(m, "Ctrl %d Soft Errors	= %d\n",
+				i, card->ctrl[i].stats.soft_errors);
+		seq_printf(m, "Ctrl %d Writes Issued	= %d\n",
+				i, card->ctrl[i].stats.writes_issued);
+		seq_printf(m, "Ctrl %d Writes Failed	= %d\n",
+				i, card->ctrl[i].stats.writes_failed);
+		seq_printf(m, "Ctrl %d Reads Issued	= %d\n",
+				i, card->ctrl[i].stats.reads_issued);
+		seq_printf(m, "Ctrl %d Reads Failed	= %d\n",
+				i, card->ctrl[i].stats.reads_failed);
+		seq_printf(m, "Ctrl %d Reads Retried	= %d\n",
+				i, card->ctrl[i].stats.reads_retried);
+		seq_printf(m, "Ctrl %d Discards Issued	= %d\n",
+				i, card->ctrl[i].stats.discards_issued);
+		seq_printf(m, "Ctrl %d Discards Failed	= %d\n",
+				i, card->ctrl[i].stats.discards_failed);
+		seq_printf(m, "Ctrl %d DMA SW Errors	= %d\n",
+				i, card->ctrl[i].stats.dma_sw_err);
+		seq_printf(m, "Ctrl %d DMA HW Faults	= %d\n",
+				i, card->ctrl[i].stats.dma_hw_fault);
+		seq_printf(m, "Ctrl %d DMAs Cancelled	= %d\n",
+				i, card->ctrl[i].stats.dma_cancelled);
+		seq_printf(m, "Ctrl %d SW Queue Depth	= %d\n",
+				i, card->ctrl[i].stats.sw_q_depth);
+		seq_printf(m, "Ctrl %d HW Queue Depth	= %d\n",
+			i, atomic_read(&card->ctrl[i].stats.hw_q_depth));
+	}
+
+	return 0;
+}
+
+static int rsxx_attr_stats_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, rsxx_attr_stats_show, inode->i_private);
+}
+
+static int rsxx_attr_pci_regs_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, rsxx_attr_pci_regs_show, inode->i_private);
+}
+
+static ssize_t rsxx_cram_read(struct file *fp, char __user *ubuf,
+			      size_t cnt, loff_t *ppos)
+{
+	struct rsxx_cram *info = fp->private_data;
+	struct rsxx_cardinfo *card = info->i_private;
+	char *buf;
+	int st;
+
+	buf = kzalloc(sizeof(*buf) * cnt, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	info->f_pos = (u32)*ppos + info->offset;
+
+	st = rsxx_creg_read(card, CREG_ADD_CRAM + info->f_pos, cnt, buf, 1);
+	if (st)
+		return st;
+
+	st = copy_to_user(ubuf, buf, cnt);
+	if (st)
+		return st;
+
+	info->offset += cnt;
+
+	kfree(buf);
+
+	return cnt;
+}
+
+static ssize_t rsxx_cram_write(struct file *fp, const char __user *ubuf,
+			       size_t cnt, loff_t *ppos)
+{
+	struct rsxx_cram *info = fp->private_data;
+	struct rsxx_cardinfo *card = info->i_private;
+	char *buf;
+	int st;
+
+	buf = kzalloc(sizeof(*buf) * cnt, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	st = copy_from_user(buf, ubuf, cnt);
+	if (st)
+		return st;
+
+	info->f_pos = (u32)*ppos + info->offset;
+
+	st = rsxx_creg_write(card, CREG_ADD_CRAM + info->f_pos, cnt, buf, 1);
+	if (st)
+		return st;
+
+	info->offset += cnt;
+
+	kfree(buf);
+
+	return cnt;
+}
+
+static int rsxx_cram_open(struct inode *inode, struct file *file)
+{
+	struct rsxx_cram *info = kzalloc(sizeof(*info), GFP_KERNEL);
+	if (!info)
+		return -ENOMEM;
+
+	info->i_private = inode->i_private;
+	info->f_pos = file->f_pos;
+	file->private_data = info;
+
+	return 0;
+}
+
+static int rsxx_cram_release(struct inode *inode, struct file *file)
+{
+	struct rsxx_cram *info = file->private_data;
+
+	if (!info)
+		return 0;
+
+	kfree(info);
+	file->private_data = NULL;
+
+	return 0;
+}
+
+static const struct file_operations debugfs_cram_fops = {
+	.owner		= THIS_MODULE,
+	.open		= rsxx_cram_open,
+	.read		= rsxx_cram_read,
+	.write		= rsxx_cram_write,
+	.release	= rsxx_cram_release,
+};
+
+static const struct file_operations debugfs_stats_fops = {
+	.owner		= THIS_MODULE,
+	.open		= rsxx_attr_stats_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+static const struct file_operations debugfs_pci_regs_fops = {
+	.owner		= THIS_MODULE,
+	.open		= rsxx_attr_pci_regs_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+static void rsxx_debugfs_dev_new(struct rsxx_cardinfo *card)
+{
+	struct dentry *debugfs_stats;
+	struct dentry *debugfs_pci_regs;
+	struct dentry *debugfs_cram;
+
+	card->debugfs_dir = debugfs_create_dir(card->gendisk->disk_name, NULL);
+	if (IS_ERR_OR_NULL(card->debugfs_dir))
+		goto failed_debugfs_dir;
+
+	debugfs_stats = debugfs_create_file("stats", S_IRUGO,
+					    card->debugfs_dir, card,
+					    &debugfs_stats_fops);
+	if (IS_ERR_OR_NULL(debugfs_stats))
+		goto failed_debugfs_stats;
+
+	debugfs_pci_regs = debugfs_create_file("pci_regs", S_IRUGO,
+					       card->debugfs_dir, card,
+					       &debugfs_pci_regs_fops);
+	if (IS_ERR_OR_NULL(debugfs_pci_regs))
+		goto failed_debugfs_pci_regs;
+
+	debugfs_cram = debugfs_create_file("cram", S_IRUGO | S_IWUSR,
+					   card->debugfs_dir, card,
+					   &debugfs_cram_fops);
+	if (IS_ERR_OR_NULL(debugfs_cram))
+		goto failed_debugfs_cram;
+
+	return;
+failed_debugfs_cram:
+	debugfs_remove(debugfs_pci_regs);
+failed_debugfs_pci_regs:
+	debugfs_remove(debugfs_stats);
+failed_debugfs_stats:
+	debugfs_remove(card->debugfs_dir);
+failed_debugfs_dir:
+	card->debugfs_dir = NULL;
+}
+
 /*----------------- Interrupt Control & Handling -------------------*/
 
 static void rsxx_mask_interrupts(struct rsxx_cardinfo *card)
@@ -163,12 +439,13 @@ static irqreturn_t rsxx_isr(int irq, void *pdata)
 		}
 
 		if (isr & CR_INTR_CREG) {
-			schedule_work(&card->creg_ctrl.done_work);
+			queue_work(card->creg_ctrl.creg_wq,
+				   &card->creg_ctrl.done_work);
 			handled++;
 		}
 
 		if (isr & CR_INTR_EVENT) {
-			schedule_work(&card->event_work);
+			queue_work(card->event_wq, &card->event_work);
 			rsxx_disable_ier_and_isr(card, CR_INTR_EVENT);
 			handled++;
 		}
@@ -329,7 +606,7 @@ static int rsxx_eeh_frozen(struct pci_dev *dev)
 	int i;
 	int st;
 
-	dev_warn(&dev->dev, "IBM FlashSystem PCI: preparing for slot reset.\n");
+	dev_warn(&dev->dev, "IBM Flash Adapter PCI: preparing for slot reset.\n");
 
 	card->eeh_state = 1;
 	rsxx_mask_interrupts(card);
@@ -367,15 +644,26 @@ static void rsxx_eeh_failure(struct pci_dev *dev)
 {
 	struct rsxx_cardinfo *card = pci_get_drvdata(dev);
 	int i;
+	int cnt = 0;
 
-	dev_err(&dev->dev, "IBM FlashSystem PCI: disabling failed card.\n");
+	dev_err(&dev->dev, "IBM Flash Adapter PCI: disabling failed card.\n");
 
 	card->eeh_state = 1;
+	card->halt = 1;
 
-	for (i = 0; i < card->n_targets; i++)
+	for (i = 0; i < card->n_targets; i++) {
-		del_timer_sync(&card->ctrl[i].activity_timer);
+		spin_lock_bh(&card->ctrl[i].queue_lock);
+		cnt = rsxx_cleanup_dma_queue(&card->ctrl[i],
+					     &card->ctrl[i].queue);
+		spin_unlock_bh(&card->ctrl[i].queue_lock);
 
-	rsxx_eeh_cancel_dmas(card);
+		cnt += rsxx_dma_cancel(&card->ctrl[i]);
+
+		if (cnt)
+			dev_info(CARD_TO_DEV(card),
+				"Freed %d queued DMAs on channel %d\n",
+				cnt, card->ctrl[i].id);
+	}
 }
 
 static int rsxx_eeh_fifo_flush_poll(struct rsxx_cardinfo *card)
@@ -432,7 +720,7 @@ static pci_ers_result_t rsxx_slot_reset(struct pci_dev *dev)
 	int st;
 
 	dev_warn(&dev->dev,
-		"IBM FlashSystem PCI: recovering from slot reset.\n");
+		"IBM Flash Adapter PCI: recovering from slot reset.\n");
 
 	st = pci_enable_device(dev);
 	if (st)
@@ -485,7 +773,7 @@ static pci_ers_result_t rsxx_slot_reset(struct pci_dev *dev)
 				&card->ctrl[i].issue_dma_work);
 	}
 
-	dev_info(&dev->dev, "IBM FlashSystem PCI: recovery complete.\n");
+	dev_info(&dev->dev, "IBM Flash Adapter PCI: recovery complete.\n");
 
 	return PCI_ERS_RESULT_RECOVERED;
 
@@ -528,6 +816,7 @@ static int rsxx_pci_probe(struct pci_dev *dev,
 {
 	struct rsxx_cardinfo *card;
 	int st;
+	unsigned int sync_timeout;
 
 	dev_info(&dev->dev, "PCI-Flash SSD discovered\n");
 
@@ -610,7 +899,11 @@ static int rsxx_pci_probe(struct pci_dev *dev,
 	}
 
 	/************* Setup Processor Command Interface *************/
-	rsxx_creg_setup(card);
+	st = rsxx_creg_setup(card);
+	if (st) {
+		dev_err(CARD_TO_DEV(card), "Failed to setup creg interface.\n");
+		goto failed_creg_setup;
+	}
 
 	spin_lock_irq(&card->irq_lock);
 	rsxx_enable_ier_and_isr(card, CR_INTR_CREG);
@@ -650,6 +943,12 @@ static int rsxx_pci_probe(struct pci_dev *dev,
 	}
 
 	/************* Setup Card Event Handler *************/
+	card->event_wq = create_singlethread_workqueue(DRIVER_NAME"_event");
+	if (!card->event_wq) {
+		dev_err(CARD_TO_DEV(card), "Failed card event setup.\n");
+		goto failed_event_handler;
+	}
+
 	INIT_WORK(&card->event_work, card_event_handler);
 
 	st = rsxx_setup_dev(card);
@@ -676,6 +975,33 @@ static int rsxx_pci_probe(struct pci_dev *dev,
 		if (st)
 			dev_crit(CARD_TO_DEV(card),
 				"Failed issuing card startup\n");
+		if (sync_start) {
+			sync_timeout = SYNC_START_TIMEOUT;
+
+			dev_info(CARD_TO_DEV(card),
+				 "Waiting for card to startup\n");
+
+			do {
+				ssleep(1);
+				sync_timeout--;
+
+				rsxx_get_card_state(card, &card->state);
+			} while (sync_timeout &&
+				(card->state == CARD_STATE_STARTING));
+
+			if (card->state == CARD_STATE_STARTING) {
+				dev_warn(CARD_TO_DEV(card),
+					 "Card startup timed out\n");
+				card->size8 = 0;
+			} else {
+				dev_info(CARD_TO_DEV(card),
+					"card state: %s\n",
+					rsxx_card_state_to_str(card->state));
+				st = rsxx_get_card_size8(card, &card->size8);
+				if (st)
+					card->size8 = 0;
+			}
+		}
 	} else if (card->state == CARD_STATE_GOOD ||
 		   card->state == CARD_STATE_RD_ONLY_FAULT) {
 		st = rsxx_get_card_size8(card, &card->size8);
@@ -685,12 +1011,21 @@ static int rsxx_pci_probe(struct pci_dev *dev,
 
 	rsxx_attach_dev(card);
 
+	/************* Setup Debugfs *************/
+	rsxx_debugfs_dev_new(card);
+
 	return 0;
 
 failed_create_dev:
+	destroy_workqueue(card->event_wq);
+	card->event_wq = NULL;
+failed_event_handler:
 	rsxx_dma_destroy(card);
 failed_dma_setup:
 failed_compatiblity_check:
+	destroy_workqueue(card->creg_ctrl.creg_wq);
+	card->creg_ctrl.creg_wq = NULL;
+failed_creg_setup:
 	spin_lock_irq(&card->irq_lock);
 	rsxx_disable_ier_and_isr(card, CR_INTR_ALL);
 	spin_unlock_irq(&card->irq_lock);
@@ -756,6 +1091,8 @@ static void rsxx_pci_remove(struct pci_dev *dev)
 	/* Prevent work_structs from re-queuing themselves. */
 	card->halt = 1;
 
+	debugfs_remove_recursive(card->debugfs_dir);
+
 	free_irq(dev->irq, card);
 
 	if (!force_legacy)

drivers/block/rsxx/cregs.c

@@ -431,6 +431,15 @@ static int __issue_creg_rw(struct rsxx_cardinfo *card,
 	*hw_stat = completion.creg_status;
 
 	if (completion.st) {
+		/*
+		* This read is needed to verify that there has not been any
+		* extreme errors that might have occurred, i.e. EEH. The
+		* function iowrite32 will not detect EEH errors, so it is
+		* necessary that we recover if such an error is the reason
+		* for the timeout. This is a dummy read.
+		*/
+		ioread32(card->regmap + SCRATCH);
+
 		dev_warn(CARD_TO_DEV(card),
 			"creg command failed(%d x%08x)\n",
 			completion.st, addr);
@@ -727,6 +736,11 @@ int rsxx_creg_setup(struct rsxx_cardinfo *card)
 {
 	card->creg_ctrl.active_cmd = NULL;
 
+	card->creg_ctrl.creg_wq =
+			create_singlethread_workqueue(DRIVER_NAME"_creg");
+	if (!card->creg_ctrl.creg_wq)
+		return -ENOMEM;
+
 	INIT_WORK(&card->creg_ctrl.done_work, creg_cmd_done);
 	mutex_init(&card->creg_ctrl.reset_lock);
 	INIT_LIST_HEAD(&card->creg_ctrl.queue);

drivers/block/rsxx/dev.c

@@ -155,6 +155,7 @@ static void bio_dma_done_cb(struct rsxx_cardinfo *card,
 		atomic_set(&meta->error, 1);
 
 	if (atomic_dec_and_test(&meta->pending_dmas)) {
+		if (!card->eeh_state && card->gendisk)
 			disk_stats_complete(card, meta->bio, meta->start_time);
 
 		bio_endio(meta->bio, atomic_read(&meta->error) ? -EIO : 0);
@@ -170,6 +171,12 @@ static void rsxx_make_request(struct request_queue *q, struct bio *bio)
 
 	might_sleep();
 
+	if (!card)
+		goto req_err;
+
+	if (bio->bi_sector + (bio->bi_size >> 9) > get_capacity(card->gendisk))
+		goto req_err;
+
 	if (unlikely(card->halt)) {
 		st = -EFAULT;
 		goto req_err;
@@ -196,6 +203,7 @@ static void rsxx_make_request(struct request_queue *q, struct bio *bio)
 	atomic_set(&bio_meta->pending_dmas, 0);
 	bio_meta->start_time = jiffies;
 
+	if (!unlikely(card->halt))
 		disk_stats_start(card, bio);
 
 	dev_dbg(CARD_TO_DEV(card), "BIO[%c]: meta: %p addr8: x%llx size: %d\n",
@@ -225,24 +233,6 @@ static bool rsxx_discard_supported(struct rsxx_cardinfo *card)
 	return (pci_rev >= RSXX_DISCARD_SUPPORT);
 }
 
-static unsigned short rsxx_get_logical_block_size(
-					struct rsxx_cardinfo *card)
-{
-	u32 capabilities = 0;
-	int st;
-
-	st = rsxx_get_card_capabilities(card, &capabilities);
-	if (st)
-		dev_warn(CARD_TO_DEV(card),
-			"Failed reading card capabilities register\n");
-
-	/* Earlier firmware did not have support for 512 byte accesses */
-	if (capabilities & CARD_CAP_SUBPAGE_WRITES)
-		return 512;
-	else
-		return RSXX_HW_BLK_SIZE;
-}
-
 int rsxx_attach_dev(struct rsxx_cardinfo *card)
 {
 	mutex_lock(&card->dev_lock);
@@ -305,7 +295,7 @@ int rsxx_setup_dev(struct rsxx_cardinfo *card)
 		return -ENOMEM;
 	}
 
-	blk_size = rsxx_get_logical_block_size(card);
+	blk_size = card->config.data.block_size;
 
 	blk_queue_make_request(card->queue, rsxx_make_request);
 	blk_queue_bounce_limit(card->queue, BLK_BOUNCE_ANY);
@@ -347,6 +337,7 @@ void rsxx_destroy_dev(struct rsxx_cardinfo *card)
 	card->gendisk = NULL;
 
 	blk_cleanup_queue(card->queue);
+	card->queue->queuedata = NULL;
 	unregister_blkdev(card->major, DRIVER_NAME);
 }
 

drivers/block/rsxx/dma.c

@@ -245,6 +245,22 @@ static void rsxx_complete_dma(struct rsxx_dma_ctrl *ctrl,
 	kmem_cache_free(rsxx_dma_pool, dma);
 }
 
+int rsxx_cleanup_dma_queue(struct rsxx_dma_ctrl *ctrl,
+			   struct list_head *q)
+{
+	struct rsxx_dma *dma;
+	struct rsxx_dma *tmp;
+	int cnt = 0;
+
+	list_for_each_entry_safe(dma, tmp, q, list) {
+		list_del(&dma->list);
+		rsxx_complete_dma(ctrl, dma, DMA_CANCELLED);
+		cnt++;
+	}
+
+	return cnt;
+}
+
 static void rsxx_requeue_dma(struct rsxx_dma_ctrl *ctrl,
 				 struct rsxx_dma *dma)
 {
@@ -252,9 +268,10 @@ static void rsxx_requeue_dma(struct rsxx_dma_ctrl *ctrl,
 	 * Requeued DMAs go to the front of the queue so they are issued
 	 * first.
 	 */
-	spin_lock(&ctrl->queue_lock);
+	spin_lock_bh(&ctrl->queue_lock);
+	ctrl->stats.sw_q_depth++;
 	list_add(&dma->list, &ctrl->queue);
-	spin_unlock(&ctrl->queue_lock);
+	spin_unlock_bh(&ctrl->queue_lock);
 }
 
 static void rsxx_handle_dma_error(struct rsxx_dma_ctrl *ctrl,
@@ -329,6 +346,7 @@ static void rsxx_handle_dma_error(struct rsxx_dma_ctrl *ctrl,
 static void dma_engine_stalled(unsigned long data)
 {
 	struct rsxx_dma_ctrl *ctrl = (struct rsxx_dma_ctrl *)data;
+	int cnt;
 
 	if (atomic_read(&ctrl->stats.hw_q_depth) == 0 ||
 	    unlikely(ctrl->card->eeh_state))
@@ -349,18 +367,28 @@ static void dma_engine_stalled(unsigned long data)
 			"DMA channel %d has stalled, faulting interface.\n",
 			ctrl->id);
 		ctrl->card->dma_fault = 1;
+
+		/* Clean up the DMA queue */
+		spin_lock(&ctrl->queue_lock);
+		cnt = rsxx_cleanup_dma_queue(ctrl, &ctrl->queue);
+		spin_unlock(&ctrl->queue_lock);
+
+		cnt += rsxx_dma_cancel(ctrl);
+
+		if (cnt)
+			dev_info(CARD_TO_DEV(ctrl->card),
+				"Freed %d queued DMAs on channel %d\n",
+				cnt, ctrl->id);
 	}
 }
 
-static void rsxx_issue_dmas(struct work_struct *work)
+static void rsxx_issue_dmas(struct rsxx_dma_ctrl *ctrl)
 {
-	struct rsxx_dma_ctrl *ctrl;
 	struct rsxx_dma *dma;
 	int tag;
 	int cmds_pending = 0;
 	struct hw_cmd *hw_cmd_buf;
 
-	ctrl = container_of(work, struct rsxx_dma_ctrl, issue_dma_work);
 	hw_cmd_buf = ctrl->cmd.buf;
 
 	if (unlikely(ctrl->card->halt) ||
@@ -368,22 +396,22 @@ static void rsxx_issue_dmas(struct work_struct *work)
 		return;
 
 	while (1) {
-		spin_lock(&ctrl->queue_lock);
+		spin_lock_bh(&ctrl->queue_lock);
 		if (list_empty(&ctrl->queue)) {
-			spin_unlock(&ctrl->queue_lock);
+			spin_unlock_bh(&ctrl->queue_lock);
 			break;
 		}
-		spin_unlock(&ctrl->queue_lock);
+		spin_unlock_bh(&ctrl->queue_lock);
 
 		tag = pop_tracker(ctrl->trackers);
 		if (tag == -1)
 			break;
 
-		spin_lock(&ctrl->queue_lock);
+		spin_lock_bh(&ctrl->queue_lock);
 		dma = list_entry(ctrl->queue.next, struct rsxx_dma, list);
 		list_del(&dma->list);
 		ctrl->stats.sw_q_depth--;
-		spin_unlock(&ctrl->queue_lock);
+		spin_unlock_bh(&ctrl->queue_lock);
 
 		/*
 		 * This will catch any DMAs that slipped in right before the
@@ -440,9 +468,8 @@ static void rsxx_issue_dmas(struct work_struct *work)
 	}
 }
 
-static void rsxx_dma_done(struct work_struct *work)
+static void rsxx_dma_done(struct rsxx_dma_ctrl *ctrl)
 {
-	struct rsxx_dma_ctrl *ctrl;
 	struct rsxx_dma *dma;
 	unsigned long flags;
 	u16 count;
@@ -450,7 +477,6 @@ static void rsxx_dma_done(struct work_struct *work)
 	u8 tag;
 	struct hw_status *hw_st_buf;
 
-	ctrl = container_of(work, struct rsxx_dma_ctrl, dma_done_work);
 	hw_st_buf = ctrl->status.buf;
 
 	if (unlikely(ctrl->card->halt) ||
@@ -520,33 +546,32 @@ static void rsxx_dma_done(struct work_struct *work)
 	rsxx_enable_ier(ctrl->card, CR_INTR_DMA(ctrl->id));
 	spin_unlock_irqrestore(&ctrl->card->irq_lock, flags);
 
-	spin_lock(&ctrl->queue_lock);
+	spin_lock_bh(&ctrl->queue_lock);
 	if (ctrl->stats.sw_q_depth)
 		queue_work(ctrl->issue_wq, &ctrl->issue_dma_work);
-	spin_unlock(&ctrl->queue_lock);
+	spin_unlock_bh(&ctrl->queue_lock);
 }
 
-static int rsxx_cleanup_dma_queue(struct rsxx_cardinfo *card,
+static void rsxx_schedule_issue(struct work_struct *work)
-				      struct list_head *q)
 {
-	struct rsxx_dma *dma;
+	struct rsxx_dma_ctrl *ctrl;
-	struct rsxx_dma *tmp;
-	int cnt = 0;
 
-	list_for_each_entry_safe(dma, tmp, q, list) {
+	ctrl = container_of(work, struct rsxx_dma_ctrl, issue_dma_work);
-		list_del(&dma->list);
 
-		if (dma->dma_addr)
+	mutex_lock(&ctrl->work_lock);
-			pci_unmap_page(card->dev, dma->dma_addr,
+	rsxx_issue_dmas(ctrl);
-				       get_dma_size(dma),
+	mutex_unlock(&ctrl->work_lock);
-				       (dma->cmd == HW_CMD_BLK_WRITE) ?
-				       PCI_DMA_TODEVICE :
-				       PCI_DMA_FROMDEVICE);
-		kmem_cache_free(rsxx_dma_pool, dma);
-		cnt++;
 }
 
-	return cnt;
+static void rsxx_schedule_done(struct work_struct *work)
+{
+	struct rsxx_dma_ctrl *ctrl;
+
+	ctrl = container_of(work, struct rsxx_dma_ctrl, dma_done_work);
+
+	mutex_lock(&ctrl->work_lock);
+	rsxx_dma_done(ctrl);
+	mutex_unlock(&ctrl->work_lock);
 }
 
 static int rsxx_queue_discard(struct rsxx_cardinfo *card,
@@ -698,10 +723,10 @@ int rsxx_dma_queue_bio(struct rsxx_cardinfo *card,
 
 	for (i = 0; i < card->n_targets; i++) {
 		if (!list_empty(&dma_list[i])) {
-			spin_lock(&card->ctrl[i].queue_lock);
+			spin_lock_bh(&card->ctrl[i].queue_lock);
 			card->ctrl[i].stats.sw_q_depth += dma_cnt[i];
 			list_splice_tail(&dma_list[i], &card->ctrl[i].queue);
-			spin_unlock(&card->ctrl[i].queue_lock);
+			spin_unlock_bh(&card->ctrl[i].queue_lock);
 
 			queue_work(card->ctrl[i].issue_wq,
 				   &card->ctrl[i].issue_dma_work);
@@ -711,8 +736,11 @@ int rsxx_dma_queue_bio(struct rsxx_cardinfo *card,
 	return 0;
 
 bvec_err:
-	for (i = 0; i < card->n_targets; i++)
+	for (i = 0; i < card->n_targets; i++) {
-		rsxx_cleanup_dma_queue(card, &dma_list[i]);
+		spin_lock_bh(&card->ctrl[i].queue_lock);
+		rsxx_cleanup_dma_queue(&card->ctrl[i], &dma_list[i]);
+		spin_unlock_bh(&card->ctrl[i].queue_lock);
+	}
 
 	return st;
 }
@@ -780,6 +808,7 @@ static int rsxx_dma_ctrl_init(struct pci_dev *dev,
 	spin_lock_init(&ctrl->trackers->lock);
 
 	spin_lock_init(&ctrl->queue_lock);
+	mutex_init(&ctrl->work_lock);
 	INIT_LIST_HEAD(&ctrl->queue);
 
 	setup_timer(&ctrl->activity_timer, dma_engine_stalled,
@@ -793,8 +822,8 @@ static int rsxx_dma_ctrl_init(struct pci_dev *dev,
 	if (!ctrl->done_wq)
 		return -ENOMEM;
 
-	INIT_WORK(&ctrl->issue_dma_work, rsxx_issue_dmas);
+	INIT_WORK(&ctrl->issue_dma_work, rsxx_schedule_issue);
-	INIT_WORK(&ctrl->dma_done_work, rsxx_dma_done);
+	INIT_WORK(&ctrl->dma_done_work, rsxx_schedule_done);
 
 	st = rsxx_hw_buffers_init(dev, ctrl);
 	if (st)
@@ -918,13 +947,30 @@ failed_dma_setup:
 	return st;
 }
 
+int rsxx_dma_cancel(struct rsxx_dma_ctrl *ctrl)
+{
+	struct rsxx_dma *dma;
+	int i;
+	int cnt = 0;
+
+	/* Clean up issued DMAs */
+	for (i = 0; i < RSXX_MAX_OUTSTANDING_CMDS; i++) {
+		dma = get_tracker_dma(ctrl->trackers, i);
+		if (dma) {
+			atomic_dec(&ctrl->stats.hw_q_depth);
+			rsxx_complete_dma(ctrl, dma, DMA_CANCELLED);
+			push_tracker(ctrl->trackers, i);
+			cnt++;
+		}
+	}
+
+	return cnt;
+}
 
 void rsxx_dma_destroy(struct rsxx_cardinfo *card)
 {
 	struct rsxx_dma_ctrl *ctrl;
-	struct rsxx_dma *dma;
+	int i;
-	int i, j;
-	int cnt = 0;
 
 	for (i = 0; i < card->n_targets; i++) {
 		ctrl = &card->ctrl[i];
@@ -943,33 +989,11 @@ void rsxx_dma_destroy(struct rsxx_cardinfo *card)
 			del_timer_sync(&ctrl->activity_timer);
 
 		/* Clean up the DMA queue */
-		spin_lock(&ctrl->queue_lock);
+		spin_lock_bh(&ctrl->queue_lock);
-		cnt = rsxx_cleanup_dma_queue(card, &ctrl->queue);
+		rsxx_cleanup_dma_queue(ctrl, &ctrl->queue);
-		spin_unlock(&ctrl->queue_lock);
+		spin_unlock_bh(&ctrl->queue_lock);
 
-		if (cnt)
+		rsxx_dma_cancel(ctrl);
-			dev_info(CARD_TO_DEV(card),
-				"Freed %d queued DMAs on channel %d\n",
-				cnt, i);
-
-		/* Clean up issued DMAs */
-		for (j = 0; j < RSXX_MAX_OUTSTANDING_CMDS; j++) {
-			dma = get_tracker_dma(ctrl->trackers, j);
-			if (dma) {
-				pci_unmap_page(card->dev, dma->dma_addr,
-					       get_dma_size(dma),
-					       (dma->cmd == HW_CMD_BLK_WRITE) ?
-					       PCI_DMA_TODEVICE :
-					       PCI_DMA_FROMDEVICE);
-				kmem_cache_free(rsxx_dma_pool, dma);
-				cnt++;
-			}
-		}
-
-		if (cnt)
-			dev_info(CARD_TO_DEV(card),
-				"Freed %d pending DMAs on channel %d\n",
-				cnt, i);
 
 		vfree(ctrl->trackers);
 
@@ -1013,7 +1037,7 @@ int rsxx_eeh_save_issued_dmas(struct rsxx_cardinfo *card)
 			cnt++;
 		}
 
-		spin_lock(&card->ctrl[i].queue_lock);
+		spin_lock_bh(&card->ctrl[i].queue_lock);
 		list_splice(&issued_dmas[i], &card->ctrl[i].queue);
 
 		atomic_sub(cnt, &card->ctrl[i].stats.hw_q_depth);
@@ -1028,7 +1052,7 @@ int rsxx_eeh_save_issued_dmas(struct rsxx_cardinfo *card)
 					       PCI_DMA_TODEVICE :
 					       PCI_DMA_FROMDEVICE);
 		}
-		spin_unlock(&card->ctrl[i].queue_lock);
+		spin_unlock_bh(&card->ctrl[i].queue_lock);
 	}
 
 	kfree(issued_dmas);
@@ -1036,30 +1060,13 @@ int rsxx_eeh_save_issued_dmas(struct rsxx_cardinfo *card)
 	return 0;
 }
 
-void rsxx_eeh_cancel_dmas(struct rsxx_cardinfo *card)
-{
-	struct rsxx_dma *dma;
-	struct rsxx_dma *tmp;
-	int i;
-
-	for (i = 0; i < card->n_targets; i++) {
-		spin_lock(&card->ctrl[i].queue_lock);
-		list_for_each_entry_safe(dma, tmp, &card->ctrl[i].queue, list) {
-			list_del(&dma->list);
-
-			rsxx_complete_dma(&card->ctrl[i], dma, DMA_CANCELLED);
-		}
-		spin_unlock(&card->ctrl[i].queue_lock);
-	}
-}
-
 int rsxx_eeh_remap_dmas(struct rsxx_cardinfo *card)
 {
 	struct rsxx_dma *dma;
 	int i;
 
 	for (i = 0; i < card->n_targets; i++) {
-		spin_lock(&card->ctrl[i].queue_lock);
+		spin_lock_bh(&card->ctrl[i].queue_lock);
 		list_for_each_entry(dma, &card->ctrl[i].queue, list) {
 			dma->dma_addr = pci_map_page(card->dev, dma->page,
 					dma->pg_off, get_dma_size(dma),
@@ -1067,12 +1074,12 @@ int rsxx_eeh_remap_dmas(struct rsxx_cardinfo *card)
 					PCI_DMA_TODEVICE :
 					PCI_DMA_FROMDEVICE);
 			if (!dma->dma_addr) {
-				spin_unlock(&card->ctrl[i].queue_lock);
+				spin_unlock_bh(&card->ctrl[i].queue_lock);
 				kmem_cache_free(rsxx_dma_pool, dma);
 				return -ENOMEM;
 			}
 		}
-		spin_unlock(&card->ctrl[i].queue_lock);
+		spin_unlock_bh(&card->ctrl[i].queue_lock);
 	}
 
 	return 0;

drivers/block/rsxx/rsxx_priv.h

@@ -39,6 +39,7 @@
 #include <linux/vmalloc.h>
 #include <linux/timer.h>
 #include <linux/ioctl.h>
+#include <linux/delay.h>
 
 #include "rsxx.h"
 #include "rsxx_cfg.h"
@@ -114,6 +115,7 @@ struct rsxx_dma_ctrl {
 	struct timer_list		activity_timer;
 	struct dma_tracker_list		*trackers;
 	struct rsxx_dma_stats		stats;
+	struct mutex			work_lock;
 };
 
 struct rsxx_cardinfo {
@@ -134,6 +136,7 @@ struct rsxx_cardinfo {
 		spinlock_t		lock;
 		bool			active;
 		struct creg_cmd		*active_cmd;
+		struct workqueue_struct	*creg_wq;
 		struct work_struct	done_work;
 		struct list_head	queue;
 		unsigned int		q_depth;
@@ -154,6 +157,7 @@ struct rsxx_cardinfo {
 		int buf_len;
 	} log;
 
+	struct workqueue_struct	*event_wq;
 	struct work_struct	event_work;
 	unsigned int		state;
 	u64			size8;
@@ -181,6 +185,8 @@ struct rsxx_cardinfo {
 
 	int			n_targets;
 	struct rsxx_dma_ctrl	*ctrl;
+
+	struct dentry		*debugfs_dir;
 };
 
 enum rsxx_pci_regmap {
@@ -283,6 +289,7 @@ enum rsxx_creg_addr {
 	CREG_ADD_CAPABILITIES		= 0x80001050,
 	CREG_ADD_LOG			= 0x80002000,
 	CREG_ADD_NUM_TARGETS		= 0x80003000,
+	CREG_ADD_CRAM			= 0xA0000000,
 	CREG_ADD_CONFIG			= 0xB0000000,
 };
 
@@ -372,6 +379,8 @@ typedef void (*rsxx_dma_cb)(struct rsxx_cardinfo *card,
 int rsxx_dma_setup(struct rsxx_cardinfo *card);
 void rsxx_dma_destroy(struct rsxx_cardinfo *card);
 int rsxx_dma_init(void);
+int rsxx_cleanup_dma_queue(struct rsxx_dma_ctrl *ctrl, struct list_head *q);
+int rsxx_dma_cancel(struct rsxx_dma_ctrl *ctrl);
 void rsxx_dma_cleanup(void);
 void rsxx_dma_queue_reset(struct rsxx_cardinfo *card);
 int rsxx_dma_configure(struct rsxx_cardinfo *card);
@@ -382,7 +391,6 @@ int rsxx_dma_queue_bio(struct rsxx_cardinfo *card,
 			   void *cb_data);
 int rsxx_hw_buffers_init(struct pci_dev *dev, struct rsxx_dma_ctrl *ctrl);
 int rsxx_eeh_save_issued_dmas(struct rsxx_cardinfo *card);
-void rsxx_eeh_cancel_dmas(struct rsxx_cardinfo *card);
 int rsxx_eeh_remap_dmas(struct rsxx_cardinfo *card);
 
 /***** cregs.c *****/

drivers/block/xen-blkback/common.h

@@ -50,6 +50,19 @@
 		 __func__, __LINE__, ##args)
 
 
+/*
+ * This is the maximum number of segments that would be allowed in indirect
+ * requests. This value will also be passed to the frontend.
+ */
+#define MAX_INDIRECT_SEGMENTS 256
+
+#define SEGS_PER_INDIRECT_FRAME \
+	(PAGE_SIZE/sizeof(struct blkif_request_segment_aligned))
+#define MAX_INDIRECT_PAGES \
+	((MAX_INDIRECT_SEGMENTS + SEGS_PER_INDIRECT_FRAME - 1)/SEGS_PER_INDIRECT_FRAME)
+#define INDIRECT_PAGES(_segs) \
+	((_segs + SEGS_PER_INDIRECT_FRAME - 1)/SEGS_PER_INDIRECT_FRAME)
+
 /* Not a real protocol.  Used to generate ring structs which contain
  * the elements common to all protocols only.  This way we get a
  * compiler-checkable way to use common struct elements, so we can
@@ -83,12 +96,31 @@ struct blkif_x86_32_request_other {
 	uint64_t       id;           /* private guest value, echoed in resp  */
 } __attribute__((__packed__));
 
+struct blkif_x86_32_request_indirect {
+	uint8_t        indirect_op;
+	uint16_t       nr_segments;
+	uint64_t       id;
+	blkif_sector_t sector_number;
+	blkif_vdev_t   handle;
+	uint16_t       _pad1;
+	grant_ref_t    indirect_grefs[BLKIF_MAX_INDIRECT_PAGES_PER_REQUEST];
+	/*
+	 * The maximum number of indirect segments (and pages) that will
+	 * be used is determined by MAX_INDIRECT_SEGMENTS, this value
+	 * is also exported to the guest (via xenstore
+	 * feature-max-indirect-segments entry), so the frontend knows how
+	 * many indirect segments the backend supports.
+	 */
+	uint64_t       _pad2;        /* make it 64 byte aligned */
+} __attribute__((__packed__));
+
 struct blkif_x86_32_request {
 	uint8_t        operation;    /* BLKIF_OP_???                         */
 	union {
 		struct blkif_x86_32_request_rw rw;
 		struct blkif_x86_32_request_discard discard;
 		struct blkif_x86_32_request_other other;
+		struct blkif_x86_32_request_indirect indirect;
 	} u;
 } __attribute__((__packed__));
 
@@ -127,12 +159,32 @@ struct blkif_x86_64_request_other {
 	uint64_t       id;           /* private guest value, echoed in resp  */
 } __attribute__((__packed__));
 
+struct blkif_x86_64_request_indirect {
+	uint8_t        indirect_op;
+	uint16_t       nr_segments;
+	uint32_t       _pad1;        /* offsetof(blkif_..,u.indirect.id)==8   */
+	uint64_t       id;
+	blkif_sector_t sector_number;
+	blkif_vdev_t   handle;
+	uint16_t       _pad2;
+	grant_ref_t    indirect_grefs[BLKIF_MAX_INDIRECT_PAGES_PER_REQUEST];
+	/*
+	 * The maximum number of indirect segments (and pages) that will
+	 * be used is determined by MAX_INDIRECT_SEGMENTS, this value
+	 * is also exported to the guest (via xenstore
+	 * feature-max-indirect-segments entry), so the frontend knows how
+	 * many indirect segments the backend supports.
+	 */
+	uint32_t       _pad3;        /* make it 64 byte aligned */
+} __attribute__((__packed__));
+
 struct blkif_x86_64_request {
 	uint8_t        operation;    /* BLKIF_OP_???                         */
 	union {
 		struct blkif_x86_64_request_rw rw;
 		struct blkif_x86_64_request_discard discard;
 		struct blkif_x86_64_request_other other;
+		struct blkif_x86_64_request_indirect indirect;
 	} u;
 } __attribute__((__packed__));
 
@@ -182,12 +234,26 @@ struct xen_vbd {
 
 struct backend_info;
 
+/* Number of available flags */
+#define PERSISTENT_GNT_FLAGS_SIZE	2
+/* This persistent grant is currently in use */
+#define PERSISTENT_GNT_ACTIVE		0
+/*
+ * This persistent grant has been used, this flag is set when we remove the
+ * PERSISTENT_GNT_ACTIVE, to know that this grant has been used recently.
+ */
+#define PERSISTENT_GNT_WAS_ACTIVE	1
+
+/* Number of requests that we can fit in a ring */
+#define XEN_BLKIF_REQS			32
 
 struct persistent_gnt {
 	struct page *page;
 	grant_ref_t gnt;
 	grant_handle_t handle;
+	DECLARE_BITMAP(flags, PERSISTENT_GNT_FLAGS_SIZE);
 	struct rb_node node;
+	struct list_head remove_node;
 };
 
 struct xen_blkif {
@@ -219,6 +285,23 @@ struct xen_blkif {
 	/* tree to store persistent grants */
 	struct rb_root		persistent_gnts;
 	unsigned int		persistent_gnt_c;
+	atomic_t		persistent_gnt_in_use;
+	unsigned long           next_lru;
+
+	/* used by the kworker that offload work from the persistent purge */
+	struct list_head	persistent_purge_list;
+	struct work_struct	persistent_purge_work;
+
+	/* buffer of free pages to map grant refs */
+	spinlock_t		free_pages_lock;
+	int			free_pages_num;
+	struct list_head	free_pages;
+
+	/* List of all 'pending_req' available */
+	struct list_head	pending_free;
+	/* And its spinlock. */
+	spinlock_t		pending_free_lock;
+	wait_queue_head_t	pending_free_wq;
 
 	/* statistics */
 	unsigned long		st_print;
@@ -231,6 +314,41 @@ struct xen_blkif {
 	unsigned long long			st_wr_sect;
 
 	wait_queue_head_t	waiting_to_free;
+	/* Thread shutdown wait queue. */
+	wait_queue_head_t	shutdown_wq;
+};
+
+struct seg_buf {
+	unsigned long offset;
+	unsigned int nsec;
+};
+
+struct grant_page {
+	struct page 		*page;
+	struct persistent_gnt	*persistent_gnt;
+	grant_handle_t		handle;
+	grant_ref_t		gref;
+};
+
+/*
+ * Each outstanding request that we've passed to the lower device layers has a
+ * 'pending_req' allocated to it. Each buffer_head that completes decrements
+ * the pendcnt towards zero. When it hits zero, the specified domain has a
+ * response queued for it, with the saved 'id' passed back.
+ */
+struct pending_req {
+	struct xen_blkif	*blkif;
+	u64			id;
+	int			nr_pages;
+	atomic_t		pendcnt;
+	unsigned short		operation;
+	int			status;
+	struct list_head	free_list;
+	struct grant_page	*segments[MAX_INDIRECT_SEGMENTS];
+	/* Indirect descriptors */
+	struct grant_page	*indirect_pages[MAX_INDIRECT_PAGES];
+	struct seg_buf		seg[MAX_INDIRECT_SEGMENTS];
+	struct bio		*biolist[MAX_INDIRECT_SEGMENTS];
 };
 
 
@@ -257,6 +375,7 @@ int xen_blkif_xenbus_init(void);
 
 irqreturn_t xen_blkif_be_int(int irq, void *dev_id);
 int xen_blkif_schedule(void *arg);
+int xen_blkif_purge_persistent(void *arg);
 
 int xen_blkbk_flush_diskcache(struct xenbus_transaction xbt,
 			      struct backend_info *be, int state);
@@ -268,7 +387,7 @@ struct xenbus_device *xen_blkbk_xenbus(struct backend_info *be);
 static inline void blkif_get_x86_32_req(struct blkif_request *dst,
 					struct blkif_x86_32_request *src)
 {
-	int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST;
+	int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST, j;
 	dst->operation = src->operation;
 	switch (src->operation) {
 	case BLKIF_OP_READ:
@@ -291,6 +410,18 @@ static inline void blkif_get_x86_32_req(struct blkif_request *dst,
 		dst->u.discard.sector_number = src->u.discard.sector_number;
 		dst->u.discard.nr_sectors = src->u.discard.nr_sectors;
 		break;
+	case BLKIF_OP_INDIRECT:
+		dst->u.indirect.indirect_op = src->u.indirect.indirect_op;
+		dst->u.indirect.nr_segments = src->u.indirect.nr_segments;
+		dst->u.indirect.handle = src->u.indirect.handle;
+		dst->u.indirect.id = src->u.indirect.id;
+		dst->u.indirect.sector_number = src->u.indirect.sector_number;
+		barrier();
+		j = min(MAX_INDIRECT_PAGES, INDIRECT_PAGES(dst->u.indirect.nr_segments));
+		for (i = 0; i < j; i++)
+			dst->u.indirect.indirect_grefs[i] =
+				src->u.indirect.indirect_grefs[i];
+		break;
 	default:
 		/*
 		 * Don't know how to translate this op. Only get the
@@ -304,7 +435,7 @@ static inline void blkif_get_x86_32_req(struct blkif_request *dst,
 static inline void blkif_get_x86_64_req(struct blkif_request *dst,
 					struct blkif_x86_64_request *src)
 {
-	int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST;
+	int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST, j;
 	dst->operation = src->operation;
 	switch (src->operation) {
 	case BLKIF_OP_READ:
@@ -327,6 +458,18 @@ static inline void blkif_get_x86_64_req(struct blkif_request *dst,
 		dst->u.discard.sector_number = src->u.discard.sector_number;
 		dst->u.discard.nr_sectors = src->u.discard.nr_sectors;
 		break;
+	case BLKIF_OP_INDIRECT:
+		dst->u.indirect.indirect_op = src->u.indirect.indirect_op;
+		dst->u.indirect.nr_segments = src->u.indirect.nr_segments;
+		dst->u.indirect.handle = src->u.indirect.handle;
+		dst->u.indirect.id = src->u.indirect.id;
+		dst->u.indirect.sector_number = src->u.indirect.sector_number;
+		barrier();
+		j = min(MAX_INDIRECT_PAGES, INDIRECT_PAGES(dst->u.indirect.nr_segments));
+		for (i = 0; i < j; i++)
+			dst->u.indirect.indirect_grefs[i] =
+				src->u.indirect.indirect_grefs[i];
+		break;
 	default:
 		/*
 		 * Don't know how to translate this op. Only get the

drivers/block/xen-blkback/xenbus.c

@@ -98,12 +98,17 @@ static void xen_update_blkif_status(struct xen_blkif *blkif)
 		err = PTR_ERR(blkif->xenblkd);
 		blkif->xenblkd = NULL;
 		xenbus_dev_error(blkif->be->dev, err, "start xenblkd");
+		return;
 	}
 }
 
 static struct xen_blkif *xen_blkif_alloc(domid_t domid)
 {
 	struct xen_blkif *blkif;
+	struct pending_req *req, *n;
+	int i, j;
+
+	BUILD_BUG_ON(MAX_INDIRECT_PAGES > BLKIF_MAX_INDIRECT_PAGES_PER_REQUEST);
 
 	blkif = kmem_cache_zalloc(xen_blkif_cachep, GFP_KERNEL);
 	if (!blkif)
@@ -118,8 +123,57 @@ static struct xen_blkif *xen_blkif_alloc(domid_t domid)
 	blkif->st_print = jiffies;
 	init_waitqueue_head(&blkif->waiting_to_free);
 	blkif->persistent_gnts.rb_node = NULL;
+	spin_lock_init(&blkif->free_pages_lock);
+	INIT_LIST_HEAD(&blkif->free_pages);
+	blkif->free_pages_num = 0;
+	atomic_set(&blkif->persistent_gnt_in_use, 0);
+
+	INIT_LIST_HEAD(&blkif->pending_free);
+
+	for (i = 0; i < XEN_BLKIF_REQS; i++) {
+		req = kzalloc(sizeof(*req), GFP_KERNEL);
+		if (!req)
+			goto fail;
+		list_add_tail(&req->free_list,
+		              &blkif->pending_free);
+		for (j = 0; j < MAX_INDIRECT_SEGMENTS; j++) {
+			req->segments[j] = kzalloc(sizeof(*req->segments[0]),
+			                           GFP_KERNEL);
+			if (!req->segments[j])
+				goto fail;
+		}
+		for (j = 0; j < MAX_INDIRECT_PAGES; j++) {
+			req->indirect_pages[j] = kzalloc(sizeof(*req->indirect_pages[0]),
+			                                 GFP_KERNEL);
+			if (!req->indirect_pages[j])
+				goto fail;
+		}
+	}
+	spin_lock_init(&blkif->pending_free_lock);
+	init_waitqueue_head(&blkif->pending_free_wq);
+	init_waitqueue_head(&blkif->shutdown_wq);
 
 	return blkif;
+
+fail:
+	list_for_each_entry_safe(req, n, &blkif->pending_free, free_list) {
+		list_del(&req->free_list);
+		for (j = 0; j < MAX_INDIRECT_SEGMENTS; j++) {
+			if (!req->segments[j])
+				break;
+			kfree(req->segments[j]);
+		}
+		for (j = 0; j < MAX_INDIRECT_PAGES; j++) {
+			if (!req->indirect_pages[j])
+				break;
+			kfree(req->indirect_pages[j]);
+		}
+		kfree(req);
+	}
+
+	kmem_cache_free(xen_blkif_cachep, blkif);
+
+	return ERR_PTR(-ENOMEM);
 }
 
 static int xen_blkif_map(struct xen_blkif *blkif, unsigned long shared_page,
@@ -178,6 +232,7 @@ static void xen_blkif_disconnect(struct xen_blkif *blkif)
 {
 	if (blkif->xenblkd) {
 		kthread_stop(blkif->xenblkd);
+		wake_up(&blkif->shutdown_wq);
 		blkif->xenblkd = NULL;
 	}
 
@@ -198,8 +253,28 @@ static void xen_blkif_disconnect(struct xen_blkif *blkif)
 
 static void xen_blkif_free(struct xen_blkif *blkif)
 {
+	struct pending_req *req, *n;
+	int i = 0, j;
+
 	if (!atomic_dec_and_test(&blkif->refcnt))
 		BUG();
+
+	/* Check that there is no request in use */
+	list_for_each_entry_safe(req, n, &blkif->pending_free, free_list) {
+		list_del(&req->free_list);
+
+		for (j = 0; j < MAX_INDIRECT_SEGMENTS; j++)
+			kfree(req->segments[j]);
+
+		for (j = 0; j < MAX_INDIRECT_PAGES; j++)
+			kfree(req->indirect_pages[j]);
+
+		kfree(req);
+		i++;
+	}
+
+	WARN_ON(i != XEN_BLKIF_REQS);
+
 	kmem_cache_free(xen_blkif_cachep, blkif);
 }
 
@@ -678,6 +753,11 @@ again:
 				 dev->nodename);
 		goto abort;
 	}
+	err = xenbus_printf(xbt, dev->nodename, "feature-max-indirect-segments", "%u",
+			    MAX_INDIRECT_SEGMENTS);
+	if (err)
+		dev_warn(&dev->dev, "writing %s/feature-max-indirect-segments (%d)",
+			 dev->nodename, err);
 
 	err = xenbus_printf(xbt, dev->nodename, "sectors", "%llu",
 			    (unsigned long long)vbd_sz(&be->blkif->vbd));
@@ -704,6 +784,11 @@ again:
 				 dev->nodename);
 		goto abort;
 	}
+	err = xenbus_printf(xbt, dev->nodename, "physical-sector-size", "%u",
+			    bdev_physical_block_size(be->blkif->vbd.bdev));
+	if (err)
+		xenbus_dev_error(dev, err, "writing %s/physical-sector-size",
+				 dev->nodename);
 
 	err = xenbus_transaction_end(xbt, 0);
 	if (err == -EAGAIN)

drivers/block/xen-blkfront.c

@@ -74,12 +74,30 @@ struct grant {
 struct blk_shadow {
 	struct blkif_request req;
 	struct request *request;
-	struct grant *grants_used[BLKIF_MAX_SEGMENTS_PER_REQUEST];
+	struct grant **grants_used;
+	struct grant **indirect_grants;
+	struct scatterlist *sg;
+};
+
+struct split_bio {
+	struct bio *bio;
+	atomic_t pending;
+	int err;
 };
 
 static DEFINE_MUTEX(blkfront_mutex);
 static const struct block_device_operations xlvbd_block_fops;
 
+/*
+ * Maximum number of segments in indirect requests, the actual value used by
+ * the frontend driver is the minimum of this value and the value provided
+ * by the backend driver.
+ */
+
+static unsigned int xen_blkif_max_segments = 32;
+module_param_named(max, xen_blkif_max_segments, int, S_IRUGO);
+MODULE_PARM_DESC(max, "Maximum amount of segments in indirect requests (default is 32)");
+
 #define BLK_RING_SIZE __CONST_RING_SIZE(blkif, PAGE_SIZE)
 
 /*
@@ -98,7 +116,6 @@ struct blkfront_info
 	enum blkif_state connected;
 	int ring_ref;
 	struct blkif_front_ring ring;
-	struct scatterlist sg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
 	unsigned int evtchn, irq;
 	struct request_queue *rq;
 	struct work_struct work;
@@ -114,6 +131,7 @@ struct blkfront_info
 	unsigned int discard_granularity;
 	unsigned int discard_alignment;
 	unsigned int feature_persistent:1;
+	unsigned int max_indirect_segments;
 	int is_ready;
 };
 
@@ -142,6 +160,13 @@ static DEFINE_SPINLOCK(minor_lock);
 
 #define DEV_NAME	"xvd"	/* name in /dev */
 
+#define SEGS_PER_INDIRECT_FRAME \
+	(PAGE_SIZE/sizeof(struct blkif_request_segment_aligned))
+#define INDIRECT_GREFS(_segs) \
+	((_segs + SEGS_PER_INDIRECT_FRAME - 1)/SEGS_PER_INDIRECT_FRAME)
+
+static int blkfront_setup_indirect(struct blkfront_info *info);
+
 static int get_id_from_freelist(struct blkfront_info *info)
 {
 	unsigned long free = info->shadow_free;
@@ -358,7 +383,8 @@ static int blkif_queue_request(struct request *req)
 	struct blkif_request *ring_req;
 	unsigned long id;
 	unsigned int fsect, lsect;
-	int i, ref;
+	int i, ref, n;
+	struct blkif_request_segment_aligned *segments = NULL;
 
 	/*
 	 * Used to store if we are able to queue the request by just using
@@ -369,21 +395,27 @@ static int blkif_queue_request(struct request *req)
 	grant_ref_t gref_head;
 	struct grant *gnt_list_entry = NULL;
 	struct scatterlist *sg;
+	int nseg, max_grefs;
 
 	if (unlikely(info->connected != BLKIF_STATE_CONNECTED))
 		return 1;
 
-	/* Check if we have enought grants to allocate a requests */
+	max_grefs = info->max_indirect_segments ?
-	if (info->persistent_gnts_c < BLKIF_MAX_SEGMENTS_PER_REQUEST) {
+		    info->max_indirect_segments +
+		    INDIRECT_GREFS(info->max_indirect_segments) :
+		    BLKIF_MAX_SEGMENTS_PER_REQUEST;
+
+	/* Check if we have enough grants to allocate a requests */
+	if (info->persistent_gnts_c < max_grefs) {
 		new_persistent_gnts = 1;
 		if (gnttab_alloc_grant_references(
-		    BLKIF_MAX_SEGMENTS_PER_REQUEST - info->persistent_gnts_c,
+		    max_grefs - info->persistent_gnts_c,
 		    &gref_head) < 0) {
 			gnttab_request_free_callback(
 				&info->callback,
 				blkif_restart_queue_callback,
 				info,
-				BLKIF_MAX_SEGMENTS_PER_REQUEST);
+				max_grefs);
 			return 1;
 		}
 	} else
@@ -394,13 +426,39 @@ static int blkif_queue_request(struct request *req)
 	id = get_id_from_freelist(info);
 	info->shadow[id].request = req;
 
+	if (unlikely(req->cmd_flags & (REQ_DISCARD | REQ_SECURE))) {
+		ring_req->operation = BLKIF_OP_DISCARD;
+		ring_req->u.discard.nr_sectors = blk_rq_sectors(req);
+		ring_req->u.discard.id = id;
+		ring_req->u.discard.sector_number = (blkif_sector_t)blk_rq_pos(req);
+		if ((req->cmd_flags & REQ_SECURE) && info->feature_secdiscard)
+			ring_req->u.discard.flag = BLKIF_DISCARD_SECURE;
+		else
+			ring_req->u.discard.flag = 0;
+	} else {
+		BUG_ON(info->max_indirect_segments == 0 &&
+		       req->nr_phys_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST);
+		BUG_ON(info->max_indirect_segments &&
+		       req->nr_phys_segments > info->max_indirect_segments);
+		nseg = blk_rq_map_sg(req->q, req, info->shadow[id].sg);
 		ring_req->u.rw.id = id;
+		if (nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST) {
+			/*
+			 * The indirect operation can only be a BLKIF_OP_READ or
+			 * BLKIF_OP_WRITE
+			 */
+			BUG_ON(req->cmd_flags & (REQ_FLUSH | REQ_FUA));
+			ring_req->operation = BLKIF_OP_INDIRECT;
+			ring_req->u.indirect.indirect_op = rq_data_dir(req) ?
+				BLKIF_OP_WRITE : BLKIF_OP_READ;
+			ring_req->u.indirect.sector_number = (blkif_sector_t)blk_rq_pos(req);
+			ring_req->u.indirect.handle = info->handle;
+			ring_req->u.indirect.nr_segments = nseg;
+		} else {
 			ring_req->u.rw.sector_number = (blkif_sector_t)blk_rq_pos(req);
 			ring_req->u.rw.handle = info->handle;
-
 			ring_req->operation = rq_data_dir(req) ?
 				BLKIF_OP_WRITE : BLKIF_OP_READ;
-
 			if (req->cmd_flags & (REQ_FLUSH | REQ_FUA)) {
 				/*
 				 * Ideally we can do an unordered flush-to-disk. In case the
@@ -411,25 +469,24 @@ static int blkif_queue_request(struct request *req)
 				 */
 				ring_req->operation = info->flush_op;
 			}
-
+			ring_req->u.rw.nr_segments = nseg;
-	if (unlikely(req->cmd_flags & (REQ_DISCARD | REQ_SECURE))) {
+		}
-		/* id, sector_number and handle are set above. */
+		for_each_sg(info->shadow[id].sg, sg, nseg, i) {
-		ring_req->operation = BLKIF_OP_DISCARD;
-		ring_req->u.discard.nr_sectors = blk_rq_sectors(req);
-		if ((req->cmd_flags & REQ_SECURE) && info->feature_secdiscard)
-			ring_req->u.discard.flag = BLKIF_DISCARD_SECURE;
-		else
-			ring_req->u.discard.flag = 0;
-	} else {
-		ring_req->u.rw.nr_segments = blk_rq_map_sg(req->q, req,
-							   info->sg);
-		BUG_ON(ring_req->u.rw.nr_segments >
-		       BLKIF_MAX_SEGMENTS_PER_REQUEST);
-
-		for_each_sg(info->sg, sg, ring_req->u.rw.nr_segments, i) {
 			fsect = sg->offset >> 9;
 			lsect = fsect + (sg->length >> 9) - 1;
 
+			if ((ring_req->operation == BLKIF_OP_INDIRECT) &&
+			    (i % SEGS_PER_INDIRECT_FRAME == 0)) {
+				if (segments)
+					kunmap_atomic(segments);
+
+				n = i / SEGS_PER_INDIRECT_FRAME;
+				gnt_list_entry = get_grant(&gref_head, info);
+				info->shadow[id].indirect_grants[n] = gnt_list_entry;
+				segments = kmap_atomic(pfn_to_page(gnt_list_entry->pfn));
+				ring_req->u.indirect.indirect_grefs[n] = gnt_list_entry->gref;
+			}
+
 			gnt_list_entry = get_grant(&gref_head, info);
 			ref = gnt_list_entry->gref;
 
@@ -441,8 +498,7 @@ static int blkif_queue_request(struct request *req)
 
 				BUG_ON(sg->offset + sg->length > PAGE_SIZE);
 
-				shared_data = kmap_atomic(
+				shared_data = kmap_atomic(pfn_to_page(gnt_list_entry->pfn));
-					pfn_to_page(gnt_list_entry->pfn));
 				bvec_data = kmap_atomic(sg_page(sg));
 
 				/*
@@ -461,14 +517,24 @@ static int blkif_queue_request(struct request *req)
 				kunmap_atomic(bvec_data);
 				kunmap_atomic(shared_data);
 			}
-
+			if (ring_req->operation != BLKIF_OP_INDIRECT) {
 				ring_req->u.rw.seg[i] =
 						(struct blkif_request_segment) {
 							.gref       = ref,
 							.first_sect = fsect,
 							.last_sect  = lsect };
+			} else {
+				n = i % SEGS_PER_INDIRECT_FRAME;
+				segments[n] =
+					(struct blkif_request_segment_aligned) {
+							.gref       = ref,
+							.first_sect = fsect,
+							.last_sect  = lsect };
 			}
 		}
+		if (segments)
+			kunmap_atomic(segments);
+	}
 
 	info->ring.req_prod_pvt++;
 
@@ -542,7 +608,9 @@ wait:
 		flush_requests(info);
 }
 
-static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
+static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size,
+				unsigned int physical_sector_size,
+				unsigned int segments)
 {
 	struct request_queue *rq;
 	struct blkfront_info *info = gd->private_data;
@@ -564,14 +632,15 @@ static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
 
 	/* Hard sector size and max sectors impersonate the equiv. hardware. */
 	blk_queue_logical_block_size(rq, sector_size);
-	blk_queue_max_hw_sectors(rq, 512);
+	blk_queue_physical_block_size(rq, physical_sector_size);
+	blk_queue_max_hw_sectors(rq, (segments * PAGE_SIZE) / 512);
 
 	/* Each segment in a request is up to an aligned page in size. */
 	blk_queue_segment_boundary(rq, PAGE_SIZE - 1);
 	blk_queue_max_segment_size(rq, PAGE_SIZE);
 
 	/* Ensure a merged request will fit in a single I/O ring slot. */
-	blk_queue_max_segments(rq, BLKIF_MAX_SEGMENTS_PER_REQUEST);
+	blk_queue_max_segments(rq, segments);
 
 	/* Make sure buffer addresses are sector-aligned. */
 	blk_queue_dma_alignment(rq, 511);
@@ -588,13 +657,16 @@ static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
 static void xlvbd_flush(struct blkfront_info *info)
 {
 	blk_queue_flush(info->rq, info->feature_flush);
-	printk(KERN_INFO "blkfront: %s: %s: %s %s\n",
+	printk(KERN_INFO "blkfront: %s: %s: %s %s %s %s %s\n",
 	       info->gd->disk_name,
 	       info->flush_op == BLKIF_OP_WRITE_BARRIER ?
 		"barrier" : (info->flush_op == BLKIF_OP_FLUSH_DISKCACHE ?
 		"flush diskcache" : "barrier or flush"),
-	       info->feature_flush ? "enabled" : "disabled",
+	       info->feature_flush ? "enabled;" : "disabled;",
-	       info->feature_persistent ? "using persistent grants" : "");
+	       "persistent grants:",
+	       info->feature_persistent ? "enabled;" : "disabled;",
+	       "indirect descriptors:",
+	       info->max_indirect_segments ? "enabled;" : "disabled;");
 }
 
 static int xen_translate_vdev(int vdevice, int *minor, unsigned int *offset)
@@ -667,7 +739,8 @@ static char *encode_disk_name(char *ptr, unsigned int n)
 
 static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
 			       struct blkfront_info *info,
-			       u16 vdisk_info, u16 sector_size)
+			       u16 vdisk_info, u16 sector_size,
+			       unsigned int physical_sector_size)
 {
 	struct gendisk *gd;
 	int nr_minors = 1;
@@ -734,7 +807,9 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
 	gd->driverfs_dev = &(info->xbdev->dev);
 	set_capacity(gd, capacity);
 
-	if (xlvbd_init_blk_queue(gd, sector_size)) {
+	if (xlvbd_init_blk_queue(gd, sector_size, physical_sector_size,
+				 info->max_indirect_segments ? :
+				 BLKIF_MAX_SEGMENTS_PER_REQUEST)) {
 		del_gendisk(gd);
 		goto release;
 	}
@@ -818,6 +893,7 @@ static void blkif_free(struct blkfront_info *info, int suspend)
 {
 	struct grant *persistent_gnt;
 	struct grant *n;
+	int i, j, segs;
 
 	/* Prevent new requests being issued until we fix things up. */
 	spin_lock_irq(&info->io_lock);
@@ -843,6 +919,47 @@ static void blkif_free(struct blkfront_info *info, int suspend)
 	}
 	BUG_ON(info->persistent_gnts_c != 0);
 
+	for (i = 0; i < BLK_RING_SIZE; i++) {
+		/*
+		 * Clear persistent grants present in requests already
+		 * on the shared ring
+		 */
+		if (!info->shadow[i].request)
+			goto free_shadow;
+
+		segs = info->shadow[i].req.operation == BLKIF_OP_INDIRECT ?
+		       info->shadow[i].req.u.indirect.nr_segments :
+		       info->shadow[i].req.u.rw.nr_segments;
+		for (j = 0; j < segs; j++) {
+			persistent_gnt = info->shadow[i].grants_used[j];
+			gnttab_end_foreign_access(persistent_gnt->gref, 0, 0UL);
+			__free_page(pfn_to_page(persistent_gnt->pfn));
+			kfree(persistent_gnt);
+		}
+
+		if (info->shadow[i].req.operation != BLKIF_OP_INDIRECT)
+			/*
+			 * If this is not an indirect operation don't try to
+			 * free indirect segments
+			 */
+			goto free_shadow;
+
+		for (j = 0; j < INDIRECT_GREFS(segs); j++) {
+			persistent_gnt = info->shadow[i].indirect_grants[j];
+			gnttab_end_foreign_access(persistent_gnt->gref, 0, 0UL);
+			__free_page(pfn_to_page(persistent_gnt->pfn));
+			kfree(persistent_gnt);
+		}
+
+free_shadow:
+		kfree(info->shadow[i].grants_used);
+		info->shadow[i].grants_used = NULL;
+		kfree(info->shadow[i].indirect_grants);
+		info->shadow[i].indirect_grants = NULL;
+		kfree(info->shadow[i].sg);
+		info->shadow[i].sg = NULL;
+	}
+
 	/* No more gnttab callback work. */
 	gnttab_cancel_free_callback(&info->callback);
 	spin_unlock_irq(&info->io_lock);
@@ -867,12 +984,13 @@ static void blkif_completion(struct blk_shadow *s, struct blkfront_info *info,
 			     struct blkif_response *bret)
 {
 	int i = 0;
-	struct bio_vec *bvec;
+	struct scatterlist *sg;
-	struct req_iterator iter;
-	unsigned long flags;
 	char *bvec_data;
 	void *shared_data;
-	unsigned int offset = 0;
+	int nseg;
+
+	nseg = s->req.operation == BLKIF_OP_INDIRECT ?
+		s->req.u.indirect.nr_segments : s->req.u.rw.nr_segments;
 
 	if (bret->operation == BLKIF_OP_READ) {
 		/*
@@ -881,26 +999,29 @@ static void blkif_completion(struct blk_shadow *s, struct blkfront_info *info,
 		 * than PAGE_SIZE, we have to keep track of the current offset,
 		 * to be sure we are copying the data from the right shared page.
 		 */
-		rq_for_each_segment(bvec, s->request, iter) {
+		for_each_sg(s->sg, sg, nseg, i) {
-			BUG_ON((bvec->bv_offset + bvec->bv_len) > PAGE_SIZE);
+			BUG_ON(sg->offset + sg->length > PAGE_SIZE);
-			if (bvec->bv_offset < offset)
-				i++;
-			BUG_ON(i >= s->req.u.rw.nr_segments);
 			shared_data = kmap_atomic(
 				pfn_to_page(s->grants_used[i]->pfn));
-			bvec_data = bvec_kmap_irq(bvec, &flags);
+			bvec_data = kmap_atomic(sg_page(sg));
-			memcpy(bvec_data, shared_data + bvec->bv_offset,
+			memcpy(bvec_data   + sg->offset,
-				bvec->bv_len);
+			       shared_data + sg->offset,
-			bvec_kunmap_irq(bvec_data, &flags);
+			       sg->length);
+			kunmap_atomic(bvec_data);
 			kunmap_atomic(shared_data);
-			offset = bvec->bv_offset + bvec->bv_len;
 		}
 	}
 	/* Add the persistent grant into the list of free grants */
-	for (i = 0; i < s->req.u.rw.nr_segments; i++) {
+	for (i = 0; i < nseg; i++) {
 		list_add(&s->grants_used[i]->node, &info->persistent_gnts);
 		info->persistent_gnts_c++;
 	}
+	if (s->req.operation == BLKIF_OP_INDIRECT) {
+		for (i = 0; i < INDIRECT_GREFS(nseg); i++) {
+			list_add(&s->indirect_grants[i]->node, &info->persistent_gnts);
+			info->persistent_gnts_c++;
+		}
+	}
 }
 
 static irqreturn_t blkif_interrupt(int irq, void *dev_id)
@@ -1034,14 +1155,6 @@ static int setup_blkring(struct xenbus_device *dev,
 	SHARED_RING_INIT(sring);
 	FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
 
-	sg_init_table(info->sg, BLKIF_MAX_SEGMENTS_PER_REQUEST);
-
-	/* Allocate memory for grants */
-	err = fill_grant_buffer(info, BLK_RING_SIZE *
-	                              BLKIF_MAX_SEGMENTS_PER_REQUEST);
-	if (err)
-		goto fail;
-
 	err = xenbus_grant_ring(dev, virt_to_mfn(info->ring.sring));
 	if (err < 0) {
 		free_page((unsigned long)sring);
@@ -1223,13 +1336,84 @@ static int blkfront_probe(struct xenbus_device *dev,
 	return 0;
 }
 
+/*
+ * This is a clone of md_trim_bio, used to split a bio into smaller ones
+ */
+static void trim_bio(struct bio *bio, int offset, int size)
+{
+	/* 'bio' is a cloned bio which we need to trim to match
+	 * the given offset and size.
+	 * This requires adjusting bi_sector, bi_size, and bi_io_vec
+	 */
+	int i;
+	struct bio_vec *bvec;
+	int sofar = 0;
+
+	size <<= 9;
+	if (offset == 0 && size == bio->bi_size)
+		return;
+
+	bio->bi_sector += offset;
+	bio->bi_size = size;
+	offset <<= 9;
+	clear_bit(BIO_SEG_VALID, &bio->bi_flags);
+
+	while (bio->bi_idx < bio->bi_vcnt &&
+	       bio->bi_io_vec[bio->bi_idx].bv_len <= offset) {
+		/* remove this whole bio_vec */
+		offset -= bio->bi_io_vec[bio->bi_idx].bv_len;
+		bio->bi_idx++;
+	}
+	if (bio->bi_idx < bio->bi_vcnt) {
+		bio->bi_io_vec[bio->bi_idx].bv_offset += offset;
+		bio->bi_io_vec[bio->bi_idx].bv_len -= offset;
+	}
+	/* avoid any complications with bi_idx being non-zero*/
+	if (bio->bi_idx) {
+		memmove(bio->bi_io_vec, bio->bi_io_vec+bio->bi_idx,
+			(bio->bi_vcnt - bio->bi_idx) * sizeof(struct bio_vec));
+		bio->bi_vcnt -= bio->bi_idx;
+		bio->bi_idx = 0;
+	}
+	/* Make sure vcnt and last bv are not too big */
+	bio_for_each_segment(bvec, bio, i) {
+		if (sofar + bvec->bv_len > size)
+			bvec->bv_len = size - sofar;
+		if (bvec->bv_len == 0) {
+			bio->bi_vcnt = i;
+			break;
+		}
+		sofar += bvec->bv_len;
+	}
+}
+
+static void split_bio_end(struct bio *bio, int error)
+{
+	struct split_bio *split_bio = bio->bi_private;
+
+	if (error)
+		split_bio->err = error;
+
+	if (atomic_dec_and_test(&split_bio->pending)) {
+		split_bio->bio->bi_phys_segments = 0;
+		bio_endio(split_bio->bio, split_bio->err);
+		kfree(split_bio);
+	}
+	bio_put(bio);
+}
 
 static int blkif_recover(struct blkfront_info *info)
 {
 	int i;
-	struct blkif_request *req;
+	struct request *req, *n;
 	struct blk_shadow *copy;
-	int j;
+	int rc;
+	struct bio *bio, *cloned_bio;
+	struct bio_list bio_list, merge_bio;
+	unsigned int segs, offset;
+	int pending, size;
+	struct split_bio *split_bio;
+	struct list_head requests;
 
 	/* Stage 1: Make a safe copy of the shadow state. */
 	copy = kmemdup(info->shadow, sizeof(info->shadow),
@@ -1244,36 +1428,64 @@ static int blkif_recover(struct blkfront_info *info)
 	info->shadow_free = info->ring.req_prod_pvt;
 	info->shadow[BLK_RING_SIZE-1].req.u.rw.id = 0x0fffffff;
 
-	/* Stage 3: Find pending requests and requeue them. */
+	rc = blkfront_setup_indirect(info);
+	if (rc) {
+		kfree(copy);
+		return rc;
+	}
+
+	segs = info->max_indirect_segments ? : BLKIF_MAX_SEGMENTS_PER_REQUEST;
+	blk_queue_max_segments(info->rq, segs);
+	bio_list_init(&bio_list);
+	INIT_LIST_HEAD(&requests);
 	for (i = 0; i < BLK_RING_SIZE; i++) {
 		/* Not in use? */
 		if (!copy[i].request)
 			continue;
 
-		/* Grab a request slot and copy shadow state into it. */
+		/*
-		req = RING_GET_REQUEST(&info->ring, info->ring.req_prod_pvt);
+		 * Get the bios in the request so we can re-queue them.
-		*req = copy[i].req;
+		 */
-
+		if (copy[i].request->cmd_flags &
-		/* We get a new request id, and must reset the shadow state. */
+		    (REQ_FLUSH | REQ_FUA | REQ_DISCARD | REQ_SECURE)) {
-		req->u.rw.id = get_id_from_freelist(info);
+			/*
-		memcpy(&info->shadow[req->u.rw.id], &copy[i], sizeof(copy[i]));
+			 * Flush operations don't contain bios, so
-
+			 * we need to requeue the whole request
-		if (req->operation != BLKIF_OP_DISCARD) {
+			 */
-		/* Rewrite any grant references invalidated by susp/resume. */
+			list_add(&copy[i].request->queuelist, &requests);
-			for (j = 0; j < req->u.rw.nr_segments; j++)
+			continue;
-				gnttab_grant_foreign_access_ref(
-					req->u.rw.seg[j].gref,
-					info->xbdev->otherend_id,
-					pfn_to_mfn(copy[i].grants_used[j]->pfn),
-					0);
 		}
-		info->shadow[req->u.rw.id].req = *req;
+		merge_bio.head = copy[i].request->bio;
-
+		merge_bio.tail = copy[i].request->biotail;
-		info->ring.req_prod_pvt++;
+		bio_list_merge(&bio_list, &merge_bio);
+		copy[i].request->bio = NULL;
+		blk_put_request(copy[i].request);
 	}
 
 	kfree(copy);
 
+	/*
+	 * Empty the queue, this is important because we might have
+	 * requests in the queue with more segments than what we
+	 * can handle now.
+	 */
+	spin_lock_irq(&info->io_lock);
+	while ((req = blk_fetch_request(info->rq)) != NULL) {
+		if (req->cmd_flags &
+		    (REQ_FLUSH | REQ_FUA | REQ_DISCARD | REQ_SECURE)) {
+			list_add(&req->queuelist, &requests);
+			continue;
+		}
+		merge_bio.head = req->bio;
+		merge_bio.tail = req->biotail;
+		bio_list_merge(&bio_list, &merge_bio);
+		req->bio = NULL;
+		if (req->cmd_flags & (REQ_FLUSH | REQ_FUA))
+			pr_alert("diskcache flush request found!\n");
+		__blk_put_request(info->rq, req);
+	}
+	spin_unlock_irq(&info->io_lock);
+
 	xenbus_switch_state(info->xbdev, XenbusStateConnected);
 
 	spin_lock_irq(&info->io_lock);
@@ -1281,14 +1493,50 @@ static int blkif_recover(struct blkfront_info *info)
 	/* Now safe for us to use the shared ring */
 	info->connected = BLKIF_STATE_CONNECTED;
 
-	/* Send off requeued requests */
-	flush_requests(info);
-
 	/* Kick any other new requests queued since we resumed */
 	kick_pending_request_queues(info);
 
+	list_for_each_entry_safe(req, n, &requests, queuelist) {
+		/* Requeue pending requests (flush or discard) */
+		list_del_init(&req->queuelist);
+		BUG_ON(req->nr_phys_segments > segs);
+		blk_requeue_request(info->rq, req);
+	}
 	spin_unlock_irq(&info->io_lock);
 
+	while ((bio = bio_list_pop(&bio_list)) != NULL) {
+		/* Traverse the list of pending bios and re-queue them */
+		if (bio_segments(bio) > segs) {
+			/*
+			 * This bio has more segments than what we can
+			 * handle, we have to split it.
+			 */
+			pending = (bio_segments(bio) + segs - 1) / segs;
+			split_bio = kzalloc(sizeof(*split_bio), GFP_NOIO);
+			BUG_ON(split_bio == NULL);
+			atomic_set(&split_bio->pending, pending);
+			split_bio->bio = bio;
+			for (i = 0; i < pending; i++) {
+				offset = (i * segs * PAGE_SIZE) >> 9;
+				size = min((unsigned int)(segs * PAGE_SIZE) >> 9,
+					   (unsigned int)(bio->bi_size >> 9) - offset);
+				cloned_bio = bio_clone(bio, GFP_NOIO);
+				BUG_ON(cloned_bio == NULL);
+				trim_bio(cloned_bio, offset, size);
+				cloned_bio->bi_private = split_bio;
+				cloned_bio->bi_end_io = split_bio_end;
+				submit_bio(cloned_bio->bi_rw, cloned_bio);
+			}
+			/*
+			 * Now we have to wait for all those smaller bios to
+			 * end, so we can also end the "parent" bio.
+			 */
+			continue;
+		}
+		/* We don't need to split this bio */
+		submit_bio(bio->bi_rw, bio);
+	}
+
 	return 0;
 }
 
@@ -1308,8 +1556,12 @@ static int blkfront_resume(struct xenbus_device *dev)
 	blkif_free(info, info->connected == BLKIF_STATE_CONNECTED);
 
 	err = talk_to_blkback(dev, info);
-	if (info->connected == BLKIF_STATE_SUSPENDED && !err)
+
-		err = blkif_recover(info);
+	/*
+	 * We have to wait for the backend to switch to
+	 * connected state, since we want to read which
+	 * features it supports.
+	 */
 
 	return err;
 }
@@ -1387,6 +1639,60 @@ static void blkfront_setup_discard(struct blkfront_info *info)
 	kfree(type);
 }
 
+static int blkfront_setup_indirect(struct blkfront_info *info)
+{
+	unsigned int indirect_segments, segs;
+	int err, i;
+
+	err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
+			    "feature-max-indirect-segments", "%u", &indirect_segments,
+			    NULL);
+	if (err) {
+		info->max_indirect_segments = 0;
+		segs = BLKIF_MAX_SEGMENTS_PER_REQUEST;
+	} else {
+		info->max_indirect_segments = min(indirect_segments,
+						  xen_blkif_max_segments);
+		segs = info->max_indirect_segments;
+	}
+
+	err = fill_grant_buffer(info, (segs + INDIRECT_GREFS(segs)) * BLK_RING_SIZE);
+	if (err)
+		goto out_of_memory;
+
+	for (i = 0; i < BLK_RING_SIZE; i++) {
+		info->shadow[i].grants_used = kzalloc(
+			sizeof(info->shadow[i].grants_used[0]) * segs,
+			GFP_NOIO);
+		info->shadow[i].sg = kzalloc(sizeof(info->shadow[i].sg[0]) * segs, GFP_NOIO);
+		if (info->max_indirect_segments)
+			info->shadow[i].indirect_grants = kzalloc(
+				sizeof(info->shadow[i].indirect_grants[0]) *
+				INDIRECT_GREFS(segs),
+				GFP_NOIO);
+		if ((info->shadow[i].grants_used == NULL) ||
+			(info->shadow[i].sg == NULL) ||
+		     (info->max_indirect_segments &&
+		     (info->shadow[i].indirect_grants == NULL)))
+			goto out_of_memory;
+		sg_init_table(info->shadow[i].sg, segs);
+	}
+
+
+	return 0;
+
+out_of_memory:
+	for (i = 0; i < BLK_RING_SIZE; i++) {
+		kfree(info->shadow[i].grants_used);
+		info->shadow[i].grants_used = NULL;
+		kfree(info->shadow[i].sg);
+		info->shadow[i].sg = NULL;
+		kfree(info->shadow[i].indirect_grants);
+		info->shadow[i].indirect_grants = NULL;
+	}
+	return -ENOMEM;
+}
+
 /*
  * Invoked when the backend is finally 'ready' (and has told produced
  * the details about the physical device - #sectors, size, etc).
@@ -1395,6 +1701,7 @@ static void blkfront_connect(struct blkfront_info *info)
 {
 	unsigned long long sectors;
 	unsigned long sector_size;
+	unsigned int physical_sector_size;
 	unsigned int binfo;
 	int err;
 	int barrier, flush, discard, persistent;
@@ -1414,8 +1721,15 @@ static void blkfront_connect(struct blkfront_info *info)
 		set_capacity(info->gd, sectors);
 		revalidate_disk(info->gd);
 
-		/* fall through */
+		return;
 	case BLKIF_STATE_SUSPENDED:
+		/*
+		 * If we are recovering from suspension, we need to wait
+		 * for the backend to announce it's features before
+		 * reconnecting, at least we need to know if the backend
+		 * supports indirect descriptors, and how many.
+		 */
+		blkif_recover(info);
 		return;
 
 	default:
@@ -1437,6 +1751,16 @@ static void blkfront_connect(struct blkfront_info *info)
 		return;
 	}
 
+	/*
+	 * physcial-sector-size is a newer field, so old backends may not
+	 * provide this. Assume physical sector size to be the same as
+	 * sector_size in that case.
+	 */
+	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
+			   "physical-sector-size", "%u", &physical_sector_size);
+	if (err != 1)
+		physical_sector_size = sector_size;
+
 	info->feature_flush = 0;
 	info->flush_op = 0;
 
@@ -1483,7 +1807,15 @@ static void blkfront_connect(struct blkfront_info *info)
 	else
 		info->feature_persistent = persistent;
 
-	err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size);
+	err = blkfront_setup_indirect(info);
+	if (err) {
+		xenbus_dev_fatal(info->xbdev, err, "setup_indirect at %s",
+				 info->xbdev->otherend);
+		return;
+	}
+
+	err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size,
+				  physical_sector_size);
 	if (err) {
 		xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
 				 info->xbdev->otherend);

drivers/md/bcache/alloc.c

@@ -63,7 +63,10 @@
 #include "bcache.h"
 #include "btree.h"
 
+#include <linux/freezer.h>
+#include <linux/kthread.h>
 #include <linux/random.h>
+#include <trace/events/bcache.h>
 
 #define MAX_IN_FLIGHT_DISCARDS		8U
 
@@ -151,7 +154,7 @@ static void discard_finish(struct work_struct *w)
 	mutex_unlock(&ca->set->bucket_lock);
 
 	closure_wake_up(&ca->set->bucket_wait);
-	wake_up(&ca->set->alloc_wait);
+	wake_up_process(ca->alloc_thread);
 
 	closure_put(&ca->set->cl);
 }
@@ -350,38 +353,30 @@ static void invalidate_buckets(struct cache *ca)
 		break;
 	}
 
-	pr_debug("free %zu/%zu free_inc %zu/%zu unused %zu/%zu",
+	trace_bcache_alloc_invalidate(ca);
-		 fifo_used(&ca->free), ca->free.size,
-		 fifo_used(&ca->free_inc), ca->free_inc.size,
-		 fifo_used(&ca->unused), ca->unused.size);
 }
 
 #define allocator_wait(ca, cond)					\
 do {									\
-	DEFINE_WAIT(__wait);						\
-									\
 	while (1) {							\
-		prepare_to_wait(&ca->set->alloc_wait,			\
+		set_current_state(TASK_INTERRUPTIBLE);			\
-				&__wait, TASK_INTERRUPTIBLE);		\
 		if (cond)						\
 			break;						\
 									\
 		mutex_unlock(&(ca)->set->bucket_lock);			\
-		if (test_bit(CACHE_SET_STOPPING_2, &ca->set->flags)) {	\
+		if (kthread_should_stop())				\
-			finish_wait(&ca->set->alloc_wait, &__wait);	\
+			return 0;					\
-			closure_return(cl);				\
-		}							\
 									\
+		try_to_freeze();					\
 		schedule();						\
 		mutex_lock(&(ca)->set->bucket_lock);			\
 	}								\
-									\
+	__set_current_state(TASK_RUNNING);				\
-	finish_wait(&ca->set->alloc_wait, &__wait);			\
 } while (0)
 
-void bch_allocator_thread(struct closure *cl)
+static int bch_allocator_thread(void *arg)
 {
-	struct cache *ca = container_of(cl, struct cache, alloc);
+	struct cache *ca = arg;
 
 	mutex_lock(&ca->set->bucket_lock);
 
@@ -442,7 +437,7 @@ long bch_bucket_alloc(struct cache *ca, unsigned watermark, struct closure *cl)
 {
 	long r = -1;
 again:
-	wake_up(&ca->set->alloc_wait);
+	wake_up_process(ca->alloc_thread);
 
 	if (fifo_used(&ca->free) > ca->watermark[watermark] &&
 	    fifo_pop(&ca->free, r)) {
@@ -476,9 +471,7 @@ again:
 		return r;
 	}
 
-	pr_debug("alloc failure: blocked %i free %zu free_inc %zu unused %zu",
+	trace_bcache_alloc_fail(ca);
-		 atomic_read(&ca->set->prio_blocked), fifo_used(&ca->free),
-		 fifo_used(&ca->free_inc), fifo_used(&ca->unused));
 
 	if (cl) {
 		closure_wait(&ca->set->bucket_wait, cl);
@@ -552,6 +545,17 @@ int bch_bucket_alloc_set(struct cache_set *c, unsigned watermark,
 
 /* Init */
 
+int bch_cache_allocator_start(struct cache *ca)
+{
+	struct task_struct *k = kthread_run(bch_allocator_thread,
+					    ca, "bcache_allocator");
+	if (IS_ERR(k))
+		return PTR_ERR(k);
+
+	ca->alloc_thread = k;
+	return 0;
+}
+
 void bch_cache_allocator_exit(struct cache *ca)
 {
 	struct discard *d;

drivers/md/bcache/bcache.h

@@ -178,7 +178,6 @@
 #define pr_fmt(fmt) "bcache: %s() " fmt "\n", __func__
 
 #include <linux/bio.h>
-#include <linux/blktrace_api.h>
 #include <linux/kobject.h>
 #include <linux/list.h>
 #include <linux/mutex.h>
@@ -388,8 +387,6 @@ struct keybuf_key {
 typedef bool (keybuf_pred_fn)(struct keybuf *, struct bkey *);
 
 struct keybuf {
-	keybuf_pred_fn		*key_predicate;
-
 	struct bkey		last_scanned;
 	spinlock_t		lock;
 
@@ -437,9 +434,12 @@ struct bcache_device {
 
 	/* If nonzero, we're detaching/unregistering from cache set */
 	atomic_t		detaching;
+	int			flush_done;
+
+	uint64_t		nr_stripes;
+	unsigned		stripe_size_bits;
+	atomic_t		*stripe_sectors_dirty;
 
-	atomic_long_t		sectors_dirty;
-	unsigned long		sectors_dirty_gc;
 	unsigned long		sectors_dirty_last;
 	long			sectors_dirty_derivative;
 
@@ -531,6 +531,7 @@ struct cached_dev {
 	unsigned		sequential_merge:1;
 	unsigned		verify:1;
 
+	unsigned		partial_stripes_expensive:1;
 	unsigned		writeback_metadata:1;
 	unsigned		writeback_running:1;
 	unsigned char		writeback_percent;
@@ -565,8 +566,7 @@ struct cache {
 
 	unsigned		watermark[WATERMARK_MAX];
 
-	struct closure		alloc;
+	struct task_struct	*alloc_thread;
-	struct workqueue_struct	*alloc_workqueue;
 
 	struct closure		prio;
 	struct prio_set		*disk_buckets;
@@ -664,13 +664,9 @@ struct gc_stat {
  * CACHE_SET_STOPPING always gets set first when we're closing down a cache set;
  * we'll continue to run normally for awhile with CACHE_SET_STOPPING set (i.e.
  * flushing dirty data).
- *
- * CACHE_SET_STOPPING_2 gets set at the last phase, when it's time to shut down
- * the allocation thread.
  */
 #define CACHE_SET_UNREGISTERING		0
 #define	CACHE_SET_STOPPING		1
-#define	CACHE_SET_STOPPING_2		2
 
 struct cache_set {
 	struct closure		cl;
@@ -703,9 +699,6 @@ struct cache_set {
 	/* For the btree cache */
 	struct shrinker		shrink;
 
-	/* For the allocator itself */
-	wait_queue_head_t	alloc_wait;
-
 	/* For the btree cache and anything allocation related */
 	struct mutex		bucket_lock;
 
@@ -823,10 +816,9 @@ struct cache_set {
 
 	/*
 	 * A btree node on disk could have too many bsets for an iterator to fit
-	 * on the stack - this is a single element mempool for btree_read_work()
+	 * on the stack - have to dynamically allocate them
 	 */
-	struct mutex		fill_lock;
+	mempool_t		*fill_iter;
-	struct btree_iter	*fill_iter;
 
 	/*
 	 * btree_sort() is a merge sort and requires temporary space - single
@@ -834,6 +826,7 @@ struct cache_set {
 	 */
 	struct mutex		sort_lock;
 	struct bset		*sort;
+	unsigned		sort_crit_factor;
 
 	/* List of buckets we're currently writing data to */
 	struct list_head	data_buckets;
@@ -906,8 +899,6 @@ static inline unsigned local_clock_us(void)
 	return local_clock() >> 10;
 }
 
-#define MAX_BSETS		4U
-
 #define BTREE_PRIO		USHRT_MAX
 #define INITIAL_PRIO		32768
 
@@ -1112,23 +1103,6 @@ static inline void __bkey_put(struct cache_set *c, struct bkey *k)
 		atomic_dec_bug(&PTR_BUCKET(c, k, i)->pin);
 }
 
-/* Blktrace macros */
-
-#define blktrace_msg(c, fmt, ...)					\
-do {									\
-	struct request_queue *q = bdev_get_queue(c->bdev);		\
-	if (q)								\
-		blk_add_trace_msg(q, fmt, ##__VA_ARGS__);		\
-} while (0)
-
-#define blktrace_msg_all(s, fmt, ...)					\
-do {									\
-	struct cache *_c;						\
-	unsigned i;							\
-	for_each_cache(_c, (s), i)					\
-		blktrace_msg(_c, fmt, ##__VA_ARGS__);			\
-} while (0)
-
 static inline void cached_dev_put(struct cached_dev *dc)
 {
 	if (atomic_dec_and_test(&dc->count))
@@ -1173,10 +1147,16 @@ static inline uint8_t bucket_disk_gen(struct bucket *b)
 	static struct kobj_attribute ksysfs_##n =			\
 		__ATTR(n, S_IWUSR|S_IRUSR, show, store)
 
-/* Forward declarations */
+static inline void wake_up_allocators(struct cache_set *c)
+{
+	struct cache *ca;
+	unsigned i;
 
-void bch_writeback_queue(struct cached_dev *);
+	for_each_cache(ca, c, i)
-void bch_writeback_add(struct cached_dev *, unsigned);
+		wake_up_process(ca->alloc_thread);
+}
+
+/* Forward declarations */
 
 void bch_count_io_errors(struct cache *, int, const char *);
 void bch_bbio_count_io_errors(struct cache_set *, struct bio *,
@@ -1193,7 +1173,6 @@ void bch_submit_bbio(struct bio *, struct cache_set *, struct bkey *, unsigned);
 uint8_t bch_inc_gen(struct cache *, struct bucket *);
 void bch_rescale_priorities(struct cache_set *, int);
 bool bch_bucket_add_unused(struct cache *, struct bucket *);
-void bch_allocator_thread(struct closure *);
 
 long bch_bucket_alloc(struct cache *, unsigned, struct closure *);
 void bch_bucket_free(struct cache_set *, struct bkey *);
@@ -1241,9 +1220,9 @@ void bch_cache_set_stop(struct cache_set *);
 struct cache_set *bch_cache_set_alloc(struct cache_sb *);
 void bch_btree_cache_free(struct cache_set *);
 int bch_btree_cache_alloc(struct cache_set *);
-void bch_cached_dev_writeback_init(struct cached_dev *);
 void bch_moving_init_cache_set(struct cache_set *);
 
+int bch_cache_allocator_start(struct cache *ca);
 void bch_cache_allocator_exit(struct cache *ca);
 int bch_cache_allocator_init(struct cache *ca);
 

drivers/md/bcache/bset.c

@@ -78,6 +78,7 @@ struct bkey *bch_keylist_pop(struct keylist *l)
 bool __bch_ptr_invalid(struct cache_set *c, int level, const struct bkey *k)
 {
 	unsigned i;
+	char buf[80];
 
 	if (level && (!KEY_PTRS(k) || !KEY_SIZE(k) || KEY_DIRTY(k)))
 		goto bad;
@@ -102,7 +103,8 @@ bool __bch_ptr_invalid(struct cache_set *c, int level, const struct bkey *k)
 
 	return false;
 bad:
-	cache_bug(c, "spotted bad key %s: %s", pkey(k), bch_ptr_status(c, k));
+	bch_bkey_to_text(buf, sizeof(buf), k);
+	cache_bug(c, "spotted bad key %s: %s", buf, bch_ptr_status(c, k));
 	return true;
 }
 
@@ -162,10 +164,16 @@ bool bch_ptr_bad(struct btree *b, const struct bkey *k)
 #ifdef CONFIG_BCACHE_EDEBUG
 bug:
 	mutex_unlock(&b->c->bucket_lock);
+
+	{
+		char buf[80];
+
+		bch_bkey_to_text(buf, sizeof(buf), k);
 		btree_bug(b,
 "inconsistent pointer %s: bucket %zu pin %i prio %i gen %i last_gc %i mark %llu gc_gen %i",
-		  pkey(k), PTR_BUCKET_NR(b->c, k, i), atomic_read(&g->pin),
+			  buf, PTR_BUCKET_NR(b->c, k, i), atomic_read(&g->pin),
 			  g->prio, g->gen, g->last_gc, GC_MARK(g), g->gc_gen);
+	}
 	return true;
 #endif
 }
@@ -1084,33 +1092,39 @@ void bch_btree_sort_into(struct btree *b, struct btree *new)
 	new->sets->size = 0;
 }
 
+#define SORT_CRIT	(4096 / sizeof(uint64_t))
+
 void bch_btree_sort_lazy(struct btree *b)
 {
-	if (b->nsets) {
+	unsigned crit = SORT_CRIT;
-		unsigned i, j, keys = 0, total;
+	int i;
 
-		for (i = 0; i <= b->nsets; i++)
+	/* Don't sort if nothing to do */
-			keys += b->sets[i].data->keys;
+	if (!b->nsets)
+		goto out;
 
-		total = keys;
+	/* If not a leaf node, always sort */
-
+	if (b->level) {
-		for (j = 0; j < b->nsets; j++) {
-			if (keys * 2 < total ||
-			    keys < 1000) {
-				bch_btree_sort_partial(b, j);
-				return;
-			}
-
-			keys -= b->sets[j].data->keys;
-		}
-
-		/* Must sort if b->nsets == 3 or we'll overflow */
-		if (b->nsets >= (MAX_BSETS - 1) - b->level) {
 		bch_btree_sort(b);
 		return;
 	}
+
+	for (i = b->nsets - 1; i >= 0; --i) {
+		crit *= b->c->sort_crit_factor;
+
+		if (b->sets[i].data->keys < crit) {
+			bch_btree_sort_partial(b, i);
+			return;
+		}
 	}
 
+	/* Sort if we'd overflow */
+	if (b->nsets + 1 == MAX_BSETS) {
+		bch_btree_sort(b);
+		return;
+	}
+
+out:
 	bset_build_written_tree(b);
 }
 

drivers/md/bcache/bset.h

@@ -1,6 +1,8 @@
 #ifndef _BCACHE_BSET_H
 #define _BCACHE_BSET_H
 
+#include <linux/slab.h>
+
 /*
  * BKEYS:
  *
@@ -142,6 +144,8 @@
 
 /* Btree key comparison/iteration */
 
+#define MAX_BSETS		4U
+
 struct btree_iter {
 	size_t size, used;
 	struct btree_iter_set {

drivers/md/bcache/btree.c

@@ -24,6 +24,7 @@
 #include "btree.h"
 #include "debug.h"
 #include "request.h"
+#include "writeback.h"
 
 #include <linux/slab.h>
 #include <linux/bitops.h>
@@ -134,44 +135,17 @@ static uint64_t btree_csum_set(struct btree *b, struct bset *i)
 	return crc ^ 0xffffffffffffffffULL;
 }
 
-static void btree_bio_endio(struct bio *bio, int error)
+static void bch_btree_node_read_done(struct btree *b)
 {
-	struct closure *cl = bio->bi_private;
-	struct btree *b = container_of(cl, struct btree, io.cl);
-
-	if (error)
-		set_btree_node_io_error(b);
-
-	bch_bbio_count_io_errors(b->c, bio, error, (bio->bi_rw & WRITE)
-				 ? "writing btree" : "reading btree");
-	closure_put(cl);
-}
-
-static void btree_bio_init(struct btree *b)
-{
-	BUG_ON(b->bio);
-	b->bio = bch_bbio_alloc(b->c);
-
-	b->bio->bi_end_io	= btree_bio_endio;
-	b->bio->bi_private	= &b->io.cl;
-}
-
-void bch_btree_read_done(struct closure *cl)
-{
-	struct btree *b = container_of(cl, struct btree, io.cl);
-	struct bset *i = b->sets[0].data;
-	struct btree_iter *iter = b->c->fill_iter;
 	const char *err = "bad btree header";
-	BUG_ON(b->nsets || b->written);
+	struct bset *i = b->sets[0].data;
+	struct btree_iter *iter;
 
-	bch_bbio_free(b->bio, b->c);
+	iter = mempool_alloc(b->c->fill_iter, GFP_NOWAIT);
-	b->bio = NULL;
+	iter->size = b->c->sb.bucket_size / b->c->sb.block_size;
-
-	mutex_lock(&b->c->fill_lock);
 	iter->used = 0;
 
-	if (btree_node_io_error(b) ||
+	if (!i->seq)
-	    !i->seq)
 		goto err;
 
 	for (;
@@ -228,17 +202,8 @@ void bch_btree_read_done(struct closure *cl)
 	if (b->written < btree_blocks(b))
 		bch_bset_init_next(b);
 out:
-
+	mempool_free(iter, b->c->fill_iter);
-	mutex_unlock(&b->c->fill_lock);
+	return;
-
-	spin_lock(&b->c->btree_read_time_lock);
-	bch_time_stats_update(&b->c->btree_read_time, b->io_start_time);
-	spin_unlock(&b->c->btree_read_time_lock);
-
-	smp_wmb(); /* read_done is our write lock */
-	set_btree_node_read_done(b);
-
-	closure_return(cl);
 err:
 	set_btree_node_io_error(b);
 	bch_cache_set_error(b->c, "%s at bucket %zu, block %zu, %u keys",
@@ -247,48 +212,69 @@ err:
 	goto out;
 }
 
-void bch_btree_read(struct btree *b)
+static void btree_node_read_endio(struct bio *bio, int error)
 {
-	BUG_ON(b->nsets || b->written);
+	struct closure *cl = bio->bi_private;
+	closure_put(cl);
+}
 
-	if (!closure_trylock(&b->io.cl, &b->c->cl))
+void bch_btree_node_read(struct btree *b)
-		BUG();
+{
+	uint64_t start_time = local_clock();
+	struct closure cl;
+	struct bio *bio;
 
-	b->io_start_time = local_clock();
+	trace_bcache_btree_read(b);
 
-	btree_bio_init(b);
+	closure_init_stack(&cl);
-	b->bio->bi_rw	= REQ_META|READ_SYNC;
-	b->bio->bi_size	= KEY_SIZE(&b->key) << 9;
 
-	bch_bio_map(b->bio, b->sets[0].data);
+	bio = bch_bbio_alloc(b->c);
+	bio->bi_rw	= REQ_META|READ_SYNC;
+	bio->bi_size	= KEY_SIZE(&b->key) << 9;
+	bio->bi_end_io	= btree_node_read_endio;
+	bio->bi_private	= &cl;
 
-	pr_debug("%s", pbtree(b));
+	bch_bio_map(bio, b->sets[0].data);
-	trace_bcache_btree_read(b->bio);
-	bch_submit_bbio(b->bio, b->c, &b->key, 0);
 
-	continue_at(&b->io.cl, bch_btree_read_done, system_wq);
+	bch_submit_bbio(bio, b->c, &b->key, 0);
+	closure_sync(&cl);
+
+	if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
+		set_btree_node_io_error(b);
+
+	bch_bbio_free(bio, b->c);
+
+	if (btree_node_io_error(b))
+		goto err;
+
+	bch_btree_node_read_done(b);
+
+	spin_lock(&b->c->btree_read_time_lock);
+	bch_time_stats_update(&b->c->btree_read_time, start_time);
+	spin_unlock(&b->c->btree_read_time_lock);
+
+	return;
+err:
+	bch_cache_set_error(b->c, "io error reading bucket %lu",
+			    PTR_BUCKET_NR(b->c, &b->key, 0));
 }
 
 static void btree_complete_write(struct btree *b, struct btree_write *w)
 {
 	if (w->prio_blocked &&
 	    !atomic_sub_return(w->prio_blocked, &b->c->prio_blocked))
-		wake_up(&b->c->alloc_wait);
+		wake_up_allocators(b->c);
 
 	if (w->journal) {
 		atomic_dec_bug(w->journal);
 		__closure_wake_up(&b->c->journal.wait);
 	}
 
-	if (w->owner)
-		closure_put(w->owner);
-
 	w->prio_blocked	= 0;
 	w->journal	= NULL;
-	w->owner	= NULL;
 }
 
-static void __btree_write_done(struct closure *cl)
+static void __btree_node_write_done(struct closure *cl)
 {
 	struct btree *b = container_of(cl, struct btree, io.cl);
 	struct btree_write *w = btree_prev_write(b);
@@ -304,7 +290,7 @@ static void __btree_write_done(struct closure *cl)
 	closure_return(cl);
 }
 
-static void btree_write_done(struct closure *cl)
+static void btree_node_write_done(struct closure *cl)
 {
 	struct btree *b = container_of(cl, struct btree, io.cl);
 	struct bio_vec *bv;
@@ -313,10 +299,22 @@ static void btree_write_done(struct closure *cl)
 	__bio_for_each_segment(bv, b->bio, n, 0)
 		__free_page(bv->bv_page);
 
-	__btree_write_done(cl);
+	__btree_node_write_done(cl);
 }
 
-static void do_btree_write(struct btree *b)
+static void btree_node_write_endio(struct bio *bio, int error)
+{
+	struct closure *cl = bio->bi_private;
+	struct btree *b = container_of(cl, struct btree, io.cl);
+
+	if (error)
+		set_btree_node_io_error(b);
+
+	bch_bbio_count_io_errors(b->c, bio, error, "writing btree");
+	closure_put(cl);
+}
+
+static void do_btree_node_write(struct btree *b)
 {
 	struct closure *cl = &b->io.cl;
 	struct bset *i = b->sets[b->nsets].data;
@@ -325,15 +323,34 @@ static void do_btree_write(struct btree *b)
 	i->version	= BCACHE_BSET_VERSION;
 	i->csum		= btree_csum_set(b, i);
 
-	btree_bio_init(b);
+	BUG_ON(b->bio);
-	b->bio->bi_rw	= REQ_META|WRITE_SYNC;
+	b->bio = bch_bbio_alloc(b->c);
+
+	b->bio->bi_end_io	= btree_node_write_endio;
+	b->bio->bi_private	= &b->io.cl;
+	b->bio->bi_rw		= REQ_META|WRITE_SYNC|REQ_FUA;
 	b->bio->bi_size		= set_blocks(i, b->c) * block_bytes(b->c);
 	bch_bio_map(b->bio, i);
 
+	/*
+	 * If we're appending to a leaf node, we don't technically need FUA -
+	 * this write just needs to be persisted before the next journal write,
+	 * which will be marked FLUSH|FUA.
+	 *
+	 * Similarly if we're writing a new btree root - the pointer is going to
+	 * be in the next journal entry.
+	 *
+	 * But if we're writing a new btree node (that isn't a root) or
+	 * appending to a non leaf btree node, we need either FUA or a flush
+	 * when we write the parent with the new pointer. FUA is cheaper than a
+	 * flush, and writes appending to leaf nodes aren't blocking anything so
+	 * just make all btree node writes FUA to keep things sane.
+	 */
+
 	bkey_copy(&k.key, &b->key);
 	SET_PTR_OFFSET(&k.key, 0, PTR_OFFSET(&k.key, 0) + bset_offset(b, i));
 
-	if (!bch_bio_alloc_pages(b->bio, GFP_NOIO)) {
+	if (!bio_alloc_pages(b->bio, GFP_NOIO)) {
 		int j;
 		struct bio_vec *bv;
 		void *base = (void *) ((unsigned long) i & ~(PAGE_SIZE - 1));
@@ -342,40 +359,41 @@ static void do_btree_write(struct btree *b)
 			memcpy(page_address(bv->bv_page),
 			       base + j * PAGE_SIZE, PAGE_SIZE);
 
-		trace_bcache_btree_write(b->bio);
 		bch_submit_bbio(b->bio, b->c, &k.key, 0);
 
-		continue_at(cl, btree_write_done, NULL);
+		continue_at(cl, btree_node_write_done, NULL);
 	} else {
 		b->bio->bi_vcnt = 0;
 		bch_bio_map(b->bio, i);
 
-		trace_bcache_btree_write(b->bio);
 		bch_submit_bbio(b->bio, b->c, &k.key, 0);
 
 		closure_sync(cl);
-		__btree_write_done(cl);
+		__btree_node_write_done(cl);
 	}
 }
 
-static void __btree_write(struct btree *b)
+void bch_btree_node_write(struct btree *b, struct closure *parent)
 {
 	struct bset *i = b->sets[b->nsets].data;
 
-	BUG_ON(current->bio_list);
+	trace_bcache_btree_write(b);
+
+	BUG_ON(current->bio_list);
+	BUG_ON(b->written >= btree_blocks(b));
+	BUG_ON(b->written && !i->keys);
+	BUG_ON(b->sets->data->seq != i->seq);
+	bch_check_key_order(b, i);
 
-	closure_lock(&b->io, &b->c->cl);
 	cancel_delayed_work(&b->work);
 
+	/* If caller isn't waiting for write, parent refcount is cache set */
+	closure_lock(&b->io, parent ?: &b->c->cl);
+
 	clear_bit(BTREE_NODE_dirty,	 &b->flags);
 	change_bit(BTREE_NODE_write_idx, &b->flags);
 
-	bch_check_key_order(b, i);
+	do_btree_node_write(b);
-	BUG_ON(b->written && !i->keys);
-
-	do_btree_write(b);
-
-	pr_debug("%s block %i keys %i", pbtree(b), b->written, i->keys);
 
 	b->written += set_blocks(i, b->c);
 	atomic_long_add(set_blocks(i, b->c) * b->c->sb.block_size,
@@ -387,37 +405,31 @@ static void __btree_write(struct btree *b)
 		bch_bset_init_next(b);
 }
 
-static void btree_write_work(struct work_struct *w)
+static void btree_node_write_work(struct work_struct *w)
 {
 	struct btree *b = container_of(to_delayed_work(w), struct btree, work);
 
-	down_write(&b->lock);
+	rw_lock(true, b, b->level);
 
 	if (btree_node_dirty(b))
-		__btree_write(b);
+		bch_btree_node_write(b, NULL);
-	up_write(&b->lock);
+	rw_unlock(true, b);
 }
 
-void bch_btree_write(struct btree *b, bool now, struct btree_op *op)
+static void bch_btree_leaf_dirty(struct btree *b, struct btree_op *op)
 {
 	struct bset *i = b->sets[b->nsets].data;
 	struct btree_write *w = btree_current_write(b);
 
-	BUG_ON(b->written &&
+	BUG_ON(!b->written);
-	       (b->written >= btree_blocks(b) ||
+	BUG_ON(!i->keys);
-		i->seq != b->sets[0].data->seq ||
+
-		!i->keys));
+	if (!btree_node_dirty(b))
+		queue_delayed_work(btree_io_wq, &b->work, 30 * HZ);
 
-	if (!btree_node_dirty(b)) {
 	set_btree_node_dirty(b);
-		queue_delayed_work(btree_io_wq, &b->work,
-				   msecs_to_jiffies(30000));
-	}
 
-	w->prio_blocked += b->prio_blocked;
+	if (op && op->journal) {
-	b->prio_blocked = 0;
-
-	if (op && op->journal && !b->level) {
 		if (w->journal &&
 		    journal_pin_cmp(b->c, w, op)) {
 			atomic_dec_bug(w->journal);
@@ -430,23 +442,10 @@ void bch_btree_write(struct btree *b, bool now, struct btree_op *op)
 		}
 	}
 
-	if (current->bio_list)
-		return;
-
 	/* Force write if set is too big */
-	if (now ||
+	if (set_bytes(i) > PAGE_SIZE - 48 &&
-	    b->level ||
+	    !current->bio_list)
-	    set_bytes(i) > PAGE_SIZE - 48) {
+		bch_btree_node_write(b, NULL);
-		if (op && now) {
-			/* Must wait on multiple writes */
-			BUG_ON(w->owner);
-			w->owner = &op->cl;
-			closure_get(&op->cl);
-		}
-
-		__btree_write(b);
-	}
-	BUG_ON(!b->written);
 }
 
 /*
@@ -559,7 +558,7 @@ static struct btree *mca_bucket_alloc(struct cache_set *c,
 	init_rwsem(&b->lock);
 	lockdep_set_novalidate_class(&b->lock);
 	INIT_LIST_HEAD(&b->list);
-	INIT_DELAYED_WORK(&b->work, btree_write_work);
+	INIT_DELAYED_WORK(&b->work, btree_node_write_work);
 	b->c = c;
 	closure_init_unlocked(&b->io);
 
@@ -582,7 +581,7 @@ static int mca_reap(struct btree *b, struct closure *cl, unsigned min_order)
 	BUG_ON(btree_node_dirty(b) && !b->sets[0].data);
 
 	if (cl && btree_node_dirty(b))
-		bch_btree_write(b, true, NULL);
+		bch_btree_node_write(b, NULL);
 
 	if (cl)
 		closure_wait_event_async(&b->io.wait, cl,
@@ -623,6 +622,13 @@ static int bch_mca_shrink(struct shrinker *shrink, struct shrink_control *sc)
 	else if (!mutex_trylock(&c->bucket_lock))
 		return -1;
 
+	/*
+	 * It's _really_ critical that we don't free too many btree nodes - we
+	 * have to always leave ourselves a reserve. The reserve is how we
+	 * guarantee that allocating memory for a new btree node can always
+	 * succeed, so that inserting keys into the btree can always succeed and
+	 * IO can always make forward progress:
+	 */
 	nr /= c->btree_pages;
 	nr = min_t(unsigned long, nr, mca_can_free(c));
 
@@ -766,6 +772,8 @@ static struct btree *mca_cannibalize(struct cache_set *c, struct bkey *k,
 	int ret = -ENOMEM;
 	struct btree *i;
 
+	trace_bcache_btree_cache_cannibalize(c);
+
 	if (!cl)
 		return ERR_PTR(-ENOMEM);
 
@@ -784,7 +792,6 @@ static struct btree *mca_cannibalize(struct cache_set *c, struct bkey *k,
 		return ERR_PTR(-EAGAIN);
 	}
 
-	/* XXX: tracepoint */
 	c->try_harder = cl;
 	c->try_harder_start = local_clock();
 retry:
@@ -905,6 +912,9 @@ retry:
 	b = mca_find(c, k);
 
 	if (!b) {
+		if (current->bio_list)
+			return ERR_PTR(-EAGAIN);
+
 		mutex_lock(&c->bucket_lock);
 		b = mca_alloc(c, k, level, &op->cl);
 		mutex_unlock(&c->bucket_lock);
@@ -914,7 +924,7 @@ retry:
 		if (IS_ERR(b))
 			return b;
 
-		bch_btree_read(b);
+		bch_btree_node_read(b);
 
 		if (!write)
 			downgrade_write(&b->lock);
@@ -937,14 +947,11 @@ retry:
 	for (; i <= b->nsets; i++)
 		prefetch(b->sets[i].data);
 
-	if (!closure_wait_event(&b->io.wait, &op->cl,
+	if (btree_node_io_error(b)) {
-				btree_node_read_done(b))) {
 		rw_unlock(write, b);
-		b = ERR_PTR(-EAGAIN);
+		return ERR_PTR(-EIO);
-	} else if (btree_node_io_error(b)) {
+	}
-		rw_unlock(write, b);
+
-		b = ERR_PTR(-EIO);
-	} else
 	BUG_ON(!b->written);
 
 	return b;
@@ -959,7 +966,7 @@ static void btree_node_prefetch(struct cache_set *c, struct bkey *k, int level)
 	mutex_unlock(&c->bucket_lock);
 
 	if (!IS_ERR_OR_NULL(b)) {
-		bch_btree_read(b);
+		bch_btree_node_read(b);
 		rw_unlock(true, b);
 	}
 }
@@ -970,24 +977,19 @@ static void btree_node_free(struct btree *b, struct btree_op *op)
 {
 	unsigned i;
 
+	trace_bcache_btree_node_free(b);
+
 	/*
 	 * The BUG_ON() in btree_node_get() implies that we must have a write
 	 * lock on parent to free or even invalidate a node
 	 */
 	BUG_ON(op->lock <= b->level);
 	BUG_ON(b == b->c->root);
-	pr_debug("bucket %s", pbtree(b));
 
 	if (btree_node_dirty(b))
 		btree_complete_write(b, btree_current_write(b));
 	clear_bit(BTREE_NODE_dirty, &b->flags);
 
-	if (b->prio_blocked &&
-	    !atomic_sub_return(b->prio_blocked, &b->c->prio_blocked))
-		wake_up(&b->c->alloc_wait);
-
-	b->prio_blocked = 0;
-
 	cancel_delayed_work(&b->work);
 
 	mutex_lock(&b->c->bucket_lock);
@@ -1028,17 +1030,20 @@ retry:
 		goto retry;
 	}
 
-	set_btree_node_read_done(b);
 	b->accessed = 1;
 	bch_bset_init_next(b);
 
 	mutex_unlock(&c->bucket_lock);
+
+	trace_bcache_btree_node_alloc(b);
 	return b;
 err_free:
 	bch_bucket_free(c, &k.key);
 	__bkey_put(c, &k.key);
 err:
 	mutex_unlock(&c->bucket_lock);
+
+	trace_bcache_btree_node_alloc_fail(b);
 	return b;
 }
 
@@ -1137,11 +1142,8 @@ static int btree_gc_mark_node(struct btree *b, unsigned *keys,
 		gc->nkeys++;
 
 		gc->data += KEY_SIZE(k);
-		if (KEY_DIRTY(k)) {
+		if (KEY_DIRTY(k))
 			gc->dirty += KEY_SIZE(k);
-			if (d)
-				d->sectors_dirty_gc += KEY_SIZE(k);
-		}
 	}
 
 	for (t = b->sets; t <= &b->sets[b->nsets]; t++)
@@ -1166,14 +1168,11 @@ static struct btree *btree_gc_alloc(struct btree *b, struct bkey *k,
 
 	if (!IS_ERR_OR_NULL(n)) {
 		swap(b, n);
+		__bkey_put(b->c, &b->key);
 
 		memcpy(k->ptr, b->key.ptr,
 		       sizeof(uint64_t) * KEY_PTRS(&b->key));
 
-		__bkey_put(b->c, &b->key);
-		atomic_inc(&b->c->prio_blocked);
-		b->prio_blocked++;
-
 		btree_node_free(n, op);
 		up_write(&n->lock);
 	}
@@ -1278,7 +1277,7 @@ static void btree_gc_coalesce(struct btree *b, struct btree_op *op,
 	btree_node_free(r->b, op);
 	up_write(&r->b->lock);
 
-	pr_debug("coalesced %u nodes", nodes);
+	trace_bcache_btree_gc_coalesce(nodes);
 
 	gc->nodes--;
 	nodes--;
@@ -1293,14 +1292,9 @@ static int btree_gc_recurse(struct btree *b, struct btree_op *op,
 	void write(struct btree *r)
 	{
 		if (!r->written)
-			bch_btree_write(r, true, op);
+			bch_btree_node_write(r, &op->cl);
-		else if (btree_node_dirty(r)) {
+		else if (btree_node_dirty(r))
-			BUG_ON(btree_current_write(r)->owner);
+			bch_btree_node_write(r, writes);
-			btree_current_write(r)->owner = writes;
-			closure_get(writes);
-
-			bch_btree_write(r, true, NULL);
-		}
 
 		up_write(&r->lock);
 	}
@@ -1386,9 +1380,7 @@ static int bch_btree_gc_root(struct btree *b, struct btree_op *op,
 		ret = btree_gc_recurse(b, op, writes, gc);
 
 	if (!b->written || btree_node_dirty(b)) {
-		atomic_inc(&b->c->prio_blocked);
+		bch_btree_node_write(b, n ? &op->cl : NULL);
-		b->prio_blocked++;
-		bch_btree_write(b, true, n ? op : NULL);
 	}
 
 	if (!IS_ERR_OR_NULL(n)) {
@@ -1405,7 +1397,6 @@ static void btree_gc_start(struct cache_set *c)
 {
 	struct cache *ca;
 	struct bucket *b;
-	struct bcache_device **d;
 	unsigned i;
 
 	if (!c->gc_mark_valid)
@@ -1419,15 +1410,11 @@ static void btree_gc_start(struct cache_set *c)
 	for_each_cache(ca, c, i)
 		for_each_bucket(b, ca) {
 			b->gc_gen = b->gen;
-			if (!atomic_read(&b->pin))
+			if (!atomic_read(&b->pin)) {
 				SET_GC_MARK(b, GC_MARK_RECLAIMABLE);
+				SET_GC_SECTORS_USED(b, 0);
+			}
 		}
-
-	for (d = c->devices;
-	     d < c->devices + c->nr_uuids;
-	     d++)
-		if (*d)
-			(*d)->sectors_dirty_gc = 0;
 
 	mutex_unlock(&c->bucket_lock);
 }
@@ -1437,7 +1424,6 @@ size_t bch_btree_gc_finish(struct cache_set *c)
 	size_t available = 0;
 	struct bucket *b;
 	struct cache *ca;
-	struct bcache_device **d;
 	unsigned i;
 
 	mutex_lock(&c->bucket_lock);
@@ -1480,22 +1466,6 @@ size_t bch_btree_gc_finish(struct cache_set *c)
 		}
 	}
 
-	for (d = c->devices;
-	     d < c->devices + c->nr_uuids;
-	     d++)
-		if (*d) {
-			unsigned long last =
-				atomic_long_read(&((*d)->sectors_dirty));
-			long difference = (*d)->sectors_dirty_gc - last;
-
-			pr_debug("sectors dirty off by %li", difference);
-
-			(*d)->sectors_dirty_last += difference;
-
-			atomic_long_set(&((*d)->sectors_dirty),
-					(*d)->sectors_dirty_gc);
-		}
-
 	mutex_unlock(&c->bucket_lock);
 	return available;
 }
@@ -1508,10 +1478,9 @@ static void bch_btree_gc(struct closure *cl)
 	struct gc_stat stats;
 	struct closure writes;
 	struct btree_op op;
-
 	uint64_t start_time = local_clock();
-	trace_bcache_gc_start(c->sb.set_uuid);
+
-	blktrace_msg_all(c, "Starting gc");
+	trace_bcache_gc_start(c);
 
 	memset(&stats, 0, sizeof(struct gc_stat));
 	closure_init_stack(&writes);
@@ -1520,14 +1489,14 @@ static void bch_btree_gc(struct closure *cl)
 
 	btree_gc_start(c);
 
+	atomic_inc(&c->prio_blocked);
+
 	ret = btree_root(gc_root, c, &op, &writes, &stats);
 	closure_sync(&op.cl);
 	closure_sync(&writes);
 
 	if (ret) {
-		blktrace_msg_all(c, "Stopped gc");
 		pr_warn("gc failed!");
-
 		continue_at(cl, bch_btree_gc, bch_gc_wq);
 	}
 
@@ -1537,6 +1506,9 @@ static void bch_btree_gc(struct closure *cl)
 
 	available = bch_btree_gc_finish(c);
 
+	atomic_dec(&c->prio_blocked);
+	wake_up_allocators(c);
+
 	bch_time_stats_update(&c->btree_gc_time, start_time);
 
 	stats.key_bytes *= sizeof(uint64_t);
@@ -1544,10 +1516,8 @@ static void bch_btree_gc(struct closure *cl)
 	stats.data	<<= 9;
 	stats.in_use	= (c->nbuckets - available) * 100 / c->nbuckets;
 	memcpy(&c->gc_stats, &stats, sizeof(struct gc_stat));
-	blktrace_msg_all(c, "Finished gc");
 
-	trace_bcache_gc_end(c->sb.set_uuid);
+	trace_bcache_gc_end(c);
-	wake_up(&c->alloc_wait);
 
 	continue_at(cl, bch_moving_gc, bch_gc_wq);
 }
@@ -1654,14 +1624,14 @@ static bool fix_overlapping_extents(struct btree *b,
 				    struct btree_iter *iter,
 				    struct btree_op *op)
 {
-	void subtract_dirty(struct bkey *k, int sectors)
+	void subtract_dirty(struct bkey *k, uint64_t offset, int sectors)
 	{
-		struct bcache_device *d = b->c->devices[KEY_INODE(k)];
+		if (KEY_DIRTY(k))
-
+			bcache_dev_sectors_dirty_add(b->c, KEY_INODE(k),
-		if (KEY_DIRTY(k) && d)
+						     offset, -sectors);
-			atomic_long_sub(sectors, &d->sectors_dirty);
 	}
 
+	uint64_t old_offset;
 	unsigned old_size, sectors_found = 0;
 
 	while (1) {
@@ -1673,6 +1643,7 @@ static bool fix_overlapping_extents(struct btree *b,
 		if (bkey_cmp(k, &START_KEY(insert)) <= 0)
 			continue;
 
+		old_offset = KEY_START(k);
 		old_size = KEY_SIZE(k);
 
 		/*
@@ -1728,7 +1699,7 @@ static bool fix_overlapping_extents(struct btree *b,
 
 			struct bkey *top;
 
-			subtract_dirty(k, KEY_SIZE(insert));
+			subtract_dirty(k, KEY_START(insert), KEY_SIZE(insert));
 
 			if (bkey_written(b, k)) {
 				/*
@@ -1775,7 +1746,7 @@ static bool fix_overlapping_extents(struct btree *b,
 			}
 		}
 
-		subtract_dirty(k, old_size - KEY_SIZE(k));
+		subtract_dirty(k, old_offset, old_size - KEY_SIZE(k));
 	}
 
 check_failed:
@@ -1798,7 +1769,7 @@ static bool btree_insert_key(struct btree *b, struct btree_op *op,
 {
 	struct bset *i = b->sets[b->nsets].data;
 	struct bkey *m, *prev;
-	const char *status = "insert";
+	unsigned status = BTREE_INSERT_STATUS_INSERT;
 
 	BUG_ON(bkey_cmp(k, &b->key) > 0);
 	BUG_ON(b->level && !KEY_PTRS(k));
@@ -1831,17 +1802,17 @@ static bool btree_insert_key(struct btree *b, struct btree_op *op,
 			goto insert;
 
 		/* prev is in the tree, if we merge we're done */
-		status = "back merging";
+		status = BTREE_INSERT_STATUS_BACK_MERGE;
 		if (prev &&
 		    bch_bkey_try_merge(b, prev, k))
 			goto merged;
 
-		status = "overwrote front";
+		status = BTREE_INSERT_STATUS_OVERWROTE;
 		if (m != end(i) &&
 		    KEY_PTRS(m) == KEY_PTRS(k) && !KEY_SIZE(m))
 			goto copy;
 
-		status = "front merge";
+		status = BTREE_INSERT_STATUS_FRONT_MERGE;
 		if (m != end(i) &&
 		    bch_bkey_try_merge(b, k, m))
 			goto copy;
@@ -1851,21 +1822,21 @@ static bool btree_insert_key(struct btree *b, struct btree_op *op,
 insert:	shift_keys(b, m, k);
 copy:	bkey_copy(m, k);
 merged:
-	bch_check_keys(b, "%s for %s at %s: %s", status,
+	if (KEY_DIRTY(k))
-		       op_type(op), pbtree(b), pkey(k));
+		bcache_dev_sectors_dirty_add(b->c, KEY_INODE(k),
-	bch_check_key_order_msg(b, i, "%s for %s at %s: %s", status,
+					     KEY_START(k), KEY_SIZE(k));
-				op_type(op), pbtree(b), pkey(k));
+
+	bch_check_keys(b, "%u for %s", status, op_type(op));
 
 	if (b->level && !KEY_OFFSET(k))
-		b->prio_blocked++;
+		btree_current_write(b)->prio_blocked++;
 
-	pr_debug("%s for %s at %s: %s", status,
+	trace_bcache_btree_insert_key(b, k, op->type, status);
-		 op_type(op), pbtree(b), pkey(k));
 
 	return true;
 }
 
-bool bch_btree_insert_keys(struct btree *b, struct btree_op *op)
+static bool bch_btree_insert_keys(struct btree *b, struct btree_op *op)
 {
 	bool ret = false;
 	struct bkey *k;
@@ -1896,7 +1867,7 @@ bool bch_btree_insert_check_key(struct btree *b, struct btree_op *op,
 	    should_split(b))
 		goto out;
 
-	op->replace = KEY(op->inode, bio_end(bio), bio_sectors(bio));
+	op->replace = KEY(op->inode, bio_end_sector(bio), bio_sectors(bio));
 
 	SET_KEY_PTRS(&op->replace, 1);
 	get_random_bytes(&op->replace.ptr[0], sizeof(uint64_t));
@@ -1907,7 +1878,6 @@ bool bch_btree_insert_check_key(struct btree *b, struct btree_op *op,
 
 	BUG_ON(op->type != BTREE_INSERT);
 	BUG_ON(!btree_insert_key(b, op, &tmp.k));
-	bch_btree_write(b, false, NULL);
 	ret = true;
 out:
 	downgrade_write(&b->lock);
@@ -1929,12 +1899,11 @@ static int btree_split(struct btree *b, struct btree_op *op)
 
 	split = set_blocks(n1->sets[0].data, n1->c) > (btree_blocks(b) * 4) / 5;
 
-	pr_debug("%ssplitting at %s keys %i", split ? "" : "not ",
-		 pbtree(b), n1->sets[0].data->keys);
-
 	if (split) {
 		unsigned keys = 0;
 
+		trace_bcache_btree_node_split(b, n1->sets[0].data->keys);
+
 		n2 = bch_btree_node_alloc(b->c, b->level, &op->cl);
 		if (IS_ERR(n2))
 			goto err_free1;
@@ -1967,18 +1936,21 @@ static int btree_split(struct btree *b, struct btree_op *op)
 		bkey_copy_key(&n2->key, &b->key);
 
 		bch_keylist_add(&op->keys, &n2->key);
-		bch_btree_write(n2, true, op);
+		bch_btree_node_write(n2, &op->cl);
 		rw_unlock(true, n2);
-	} else
+	} else {
+		trace_bcache_btree_node_compact(b, n1->sets[0].data->keys);
+
 		bch_btree_insert_keys(n1, op);
+	}
 
 	bch_keylist_add(&op->keys, &n1->key);
-	bch_btree_write(n1, true, op);
+	bch_btree_node_write(n1, &op->cl);
 
 	if (n3) {
 		bkey_copy_key(&n3->key, &MAX_KEY);
 		bch_btree_insert_keys(n3, op);
-		bch_btree_write(n3, true, op);
+		bch_btree_node_write(n3, &op->cl);
 
 		closure_sync(&op->cl);
 		bch_btree_set_root(n3);
@@ -2082,8 +2054,12 @@ static int bch_btree_insert_recurse(struct btree *b, struct btree_op *op,
 
 		BUG_ON(write_block(b) != b->sets[b->nsets].data);
 
-		if (bch_btree_insert_keys(b, op))
+		if (bch_btree_insert_keys(b, op)) {
-			bch_btree_write(b, false, op);
+			if (!b->level)
+				bch_btree_leaf_dirty(b, op);
+			else
+				bch_btree_node_write(b, &op->cl);
+		}
 	}
 
 	return 0;
@@ -2140,6 +2116,11 @@ int bch_btree_insert(struct btree_op *op, struct cache_set *c)
 void bch_btree_set_root(struct btree *b)
 {
 	unsigned i;
+	struct closure cl;
+
+	closure_init_stack(&cl);
+
+	trace_bcache_btree_set_root(b);
 
 	BUG_ON(!b->written);
 
@@ -2153,8 +2134,8 @@ void bch_btree_set_root(struct btree *b)
 	b->c->root = b;
 	__bkey_put(b->c, &b->key);
 
-	bch_journal_meta(b->c, NULL);
+	bch_journal_meta(b->c, &cl);
-	pr_debug("%s for %pf", pbtree(b), __builtin_return_address(0));
+	closure_sync(&cl);
 }
 
 /* Cache lookup */
@@ -2215,9 +2196,6 @@ static int submit_partial_cache_hit(struct btree *b, struct btree_op *op,
 					 KEY_OFFSET(k) - bio->bi_sector);
 
 		n = bch_bio_split(bio, sectors, GFP_NOIO, s->d->bio_split);
-		if (!n)
-			return -EAGAIN;
-
 		if (n == bio)
 			op->lookup_done = true;
 
@@ -2240,7 +2218,6 @@ static int submit_partial_cache_hit(struct btree *b, struct btree_op *op,
 		n->bi_end_io	= bch_cache_read_endio;
 		n->bi_private	= &s->cl;
 
-		trace_bcache_cache_hit(n);
 		__bch_submit_bbio(n, b->c);
 	}
 
@@ -2257,9 +2234,6 @@ int bch_btree_search_recurse(struct btree *b, struct btree_op *op)
 	struct btree_iter iter;
 	bch_btree_iter_init(b, &iter, &KEY(op->inode, bio->bi_sector, 0));
 
-	pr_debug("at %s searching for %u:%llu", pbtree(b), op->inode,
-		 (uint64_t) bio->bi_sector);
-
 	do {
 		k = bch_btree_iter_next_filter(&iter, b, bch_ptr_bad);
 		if (!k) {
@@ -2303,7 +2277,8 @@ static inline int keybuf_nonoverlapping_cmp(struct keybuf_key *l,
 }
 
 static int bch_btree_refill_keybuf(struct btree *b, struct btree_op *op,
-				   struct keybuf *buf, struct bkey *end)
+				   struct keybuf *buf, struct bkey *end,
+				   keybuf_pred_fn *pred)
 {
 	struct btree_iter iter;
 	bch_btree_iter_init(b, &iter, &buf->last_scanned);
@@ -2322,11 +2297,9 @@ static int bch_btree_refill_keybuf(struct btree *b, struct btree_op *op,
 			if (bkey_cmp(&buf->last_scanned, end) >= 0)
 				break;
 
-			if (buf->key_predicate(buf, k)) {
+			if (pred(buf, k)) {
 				struct keybuf_key *w;
 
-				pr_debug("%s", pkey(k));
-
 				spin_lock(&buf->lock);
 
 				w = array_alloc(&buf->freelist);
@@ -2343,7 +2316,7 @@ static int bch_btree_refill_keybuf(struct btree *b, struct btree_op *op,
 			if (!k)
 				break;
 
-			btree(refill_keybuf, k, b, op, buf, end);
+			btree(refill_keybuf, k, b, op, buf, end, pred);
 			/*
 			 * Might get an error here, but can't really do anything
 			 * and it'll get logged elsewhere. Just read what we
@@ -2361,7 +2334,7 @@ static int bch_btree_refill_keybuf(struct btree *b, struct btree_op *op,
 }
 
 void bch_refill_keybuf(struct cache_set *c, struct keybuf *buf,
-			  struct bkey *end)
+		       struct bkey *end, keybuf_pred_fn *pred)
 {
 	struct bkey start = buf->last_scanned;
 	struct btree_op op;
@@ -2369,7 +2342,7 @@ void bch_refill_keybuf(struct cache_set *c, struct keybuf *buf,
 
 	cond_resched();
 
-	btree_root(refill_keybuf, c, &op, buf, end);
+	btree_root(refill_keybuf, c, &op, buf, end, pred);
 	closure_sync(&op.cl);
 
 	pr_debug("found %s keys from %llu:%llu to %llu:%llu",
@@ -2455,7 +2428,8 @@ struct keybuf_key *bch_keybuf_next(struct keybuf *buf)
 
 struct keybuf_key *bch_keybuf_next_rescan(struct cache_set *c,
 					     struct keybuf *buf,
-					     struct bkey *end)
+					     struct bkey *end,
+					     keybuf_pred_fn *pred)
 {
 	struct keybuf_key *ret;
 
@@ -2469,15 +2443,14 @@ struct keybuf_key *bch_keybuf_next_rescan(struct cache_set *c,
 			break;
 		}
 
-		bch_refill_keybuf(c, buf, end);
+		bch_refill_keybuf(c, buf, end, pred);
 	}
 
 	return ret;
 }
 
-void bch_keybuf_init(struct keybuf *buf, keybuf_pred_fn *fn)
+void bch_keybuf_init(struct keybuf *buf)
 {
-	buf->key_predicate	= fn;
 	buf->last_scanned	= MAX_KEY;
 	buf->keys		= RB_ROOT;
 

drivers/md/bcache/btree.h

@@ -102,7 +102,6 @@
 #include "debug.h"
 
 struct btree_write {
-	struct closure		*owner;
 	atomic_t		*journal;
 
 	/* If btree_split() frees a btree node, it writes a new pointer to that
@@ -142,16 +141,12 @@ struct btree {
 	 */
 	struct bset_tree	sets[MAX_BSETS];
 
-	/* Used to refcount bio splits, also protects b->bio */
+	/* For outstanding btree writes, used as a lock - protects write_idx */
 	struct closure_with_waitlist	io;
 
-	/* Gets transferred to w->prio_blocked - see the comment there */
-	int			prio_blocked;
-
 	struct list_head	list;
 	struct delayed_work	work;
 
-	uint64_t		io_start_time;
 	struct btree_write	writes[2];
 	struct bio		*bio;
 };
@@ -164,13 +159,11 @@ static inline void set_btree_node_ ## flag(struct btree *b)		\
 {	set_bit(BTREE_NODE_ ## flag, &b->flags); }			\
 
 enum btree_flags {
-	BTREE_NODE_read_done,
 	BTREE_NODE_io_error,
 	BTREE_NODE_dirty,
 	BTREE_NODE_write_idx,
 };
 
-BTREE_FLAG(read_done);
 BTREE_FLAG(io_error);
 BTREE_FLAG(dirty);
 BTREE_FLAG(write_idx);
@@ -278,6 +271,13 @@ struct btree_op {
 	BKEY_PADDED(replace);
 };
 
+enum {
+	BTREE_INSERT_STATUS_INSERT,
+	BTREE_INSERT_STATUS_BACK_MERGE,
+	BTREE_INSERT_STATUS_OVERWROTE,
+	BTREE_INSERT_STATUS_FRONT_MERGE,
+};
+
 void bch_btree_op_init_stack(struct btree_op *);
 
 static inline void rw_lock(bool w, struct btree *b, int level)
@@ -293,9 +293,7 @@ static inline void rw_unlock(bool w, struct btree *b)
 #ifdef CONFIG_BCACHE_EDEBUG
 	unsigned i;
 
-	if (w &&
+	if (w && b->key.ptr[0])
-	    b->key.ptr[0] &&
-	    btree_node_read_done(b))
 		for (i = 0; i <= b->nsets; i++)
 			bch_check_key_order(b, b->sets[i].data);
 #endif
@@ -370,9 +368,8 @@ static inline bool should_split(struct btree *b)
 		 > btree_blocks(b));
 }
 
-void bch_btree_read_done(struct closure *);
+void bch_btree_node_read(struct btree *);
-void bch_btree_read(struct btree *);
+void bch_btree_node_write(struct btree *, struct closure *);
-void bch_btree_write(struct btree *b, bool now, struct btree_op *op);
 
 void bch_cannibalize_unlock(struct cache_set *, struct closure *);
 void bch_btree_set_root(struct btree *);
@@ -380,7 +377,6 @@ struct btree *bch_btree_node_alloc(struct cache_set *, int, struct closure *);
 struct btree *bch_btree_node_get(struct cache_set *, struct bkey *,
 				int, struct btree_op *);
 
-bool bch_btree_insert_keys(struct btree *, struct btree_op *);
 bool bch_btree_insert_check_key(struct btree *, struct btree_op *,
 				   struct bio *);
 int bch_btree_insert(struct btree_op *, struct cache_set *);
@@ -393,13 +389,14 @@ void bch_moving_gc(struct closure *);
 int bch_btree_check(struct cache_set *, struct btree_op *);
 uint8_t __bch_btree_mark_key(struct cache_set *, int, struct bkey *);
 
-void bch_keybuf_init(struct keybuf *, keybuf_pred_fn *);
+void bch_keybuf_init(struct keybuf *);
-void bch_refill_keybuf(struct cache_set *, struct keybuf *, struct bkey *);
+void bch_refill_keybuf(struct cache_set *, struct keybuf *, struct bkey *,
+		       keybuf_pred_fn *);
 bool bch_keybuf_check_overlapping(struct keybuf *, struct bkey *,
 				  struct bkey *);
 void bch_keybuf_del(struct keybuf *, struct keybuf_key *);
 struct keybuf_key *bch_keybuf_next(struct keybuf *);
-struct keybuf_key *bch_keybuf_next_rescan(struct cache_set *,
+struct keybuf_key *bch_keybuf_next_rescan(struct cache_set *, struct keybuf *,
-					  struct keybuf *, struct bkey *);
+					  struct bkey *, keybuf_pred_fn *);
 
 #endif

drivers/md/bcache/closure.c

@@ -66,16 +66,18 @@ static inline void closure_put_after_sub(struct closure *cl, int flags)
 		} else {
 			struct closure *parent = cl->parent;
 			struct closure_waitlist *wait = closure_waitlist(cl);
+			closure_fn *destructor = cl->fn;
 
 			closure_debug_destroy(cl);
 
+			smp_mb();
 			atomic_set(&cl->remaining, -1);
 
 			if (wait)
 				closure_wake_up(wait);
 
-			if (cl->fn)
+			if (destructor)
-				cl->fn(cl);
+				destructor(cl);
 
 			if (parent)
 				closure_put(parent);

drivers/md/bcache/debug.c

@@ -47,11 +47,10 @@ const char *bch_ptr_status(struct cache_set *c, const struct bkey *k)
 	return "";
 }
 
-struct keyprint_hack bch_pkey(const struct bkey *k)
+int bch_bkey_to_text(char *buf, size_t size, const struct bkey *k)
 {
 	unsigned i = 0;
-	struct keyprint_hack r;
+	char *out = buf, *end = buf + size;
-	char *out = r.s, *end = r.s + KEYHACK_SIZE;
 
 #define p(...)	(out += scnprintf(out, end - out, __VA_ARGS__))
 
@@ -75,16 +74,14 @@ struct keyprint_hack bch_pkey(const struct bkey *k)
 	if (KEY_CSUM(k))
 		p(" cs%llu %llx", KEY_CSUM(k), k->ptr[1]);
 #undef p
-	return r;
+	return out - buf;
 }
 
-struct keyprint_hack bch_pbtree(const struct btree *b)
+int bch_btree_to_text(char *buf, size_t size, const struct btree *b)
 {
-	struct keyprint_hack r;
+	return scnprintf(buf, size, "%zu level %i/%i",
-
+			 PTR_BUCKET_NR(b->c, &b->key, 0),
-	snprintf(r.s, 40, "%zu level %i/%i", PTR_BUCKET_NR(b->c, &b->key, 0),
 			 b->level, b->c->root ? b->c->root->level : -1);
-	return r;
 }
 
 #if defined(CONFIG_BCACHE_DEBUG) || defined(CONFIG_BCACHE_EDEBUG)
@@ -100,10 +97,12 @@ static void dump_bset(struct btree *b, struct bset *i)
 {
 	struct bkey *k;
 	unsigned j;
+	char buf[80];
 
 	for (k = i->start; k < end(i); k = bkey_next(k)) {
+		bch_bkey_to_text(buf, sizeof(buf), k);
 		printk(KERN_ERR "block %zu key %zi/%u: %s", index(i, b),
-		       (uint64_t *) k - i->d, i->keys, pkey(k));
+		       (uint64_t *) k - i->d, i->keys, buf);
 
 		for (j = 0; j < KEY_PTRS(k); j++) {
 			size_t n = PTR_BUCKET_NR(b->c, k, j);
@@ -144,7 +143,7 @@ void bch_btree_verify(struct btree *b, struct bset *new)
 	v->written = 0;
 	v->level = b->level;
 
-	bch_btree_read(v);
+	bch_btree_node_read(v);
 	closure_wait_event(&v->io.wait, &cl,
 			   atomic_read(&b->io.cl.remaining) == -1);
 
@@ -200,7 +199,7 @@ void bch_data_verify(struct search *s)
 	if (!check)
 		return;
 
-	if (bch_bio_alloc_pages(check, GFP_NOIO))
+	if (bio_alloc_pages(check, GFP_NOIO))
 		goto out_put;
 
 	check->bi_rw		= READ_SYNC;
@@ -252,6 +251,7 @@ static void vdump_bucket_and_panic(struct btree *b, const char *fmt,
 				   va_list args)
 {
 	unsigned i;
+	char buf[80];
 
 	console_lock();
 
@@ -262,7 +262,8 @@ static void vdump_bucket_and_panic(struct btree *b, const char *fmt,
 
 	console_unlock();
 
-	panic("at %s\n", pbtree(b));
+	bch_btree_to_text(buf, sizeof(buf), b);
+	panic("at %s\n", buf);
 }
 
 void bch_check_key_order_msg(struct btree *b, struct bset *i,
@@ -337,6 +338,7 @@ static ssize_t bch_dump_read(struct file *file, char __user *buf,
 {
 	struct dump_iterator *i = file->private_data;
 	ssize_t ret = 0;
+	char kbuf[80];
 
 	while (size) {
 		struct keybuf_key *w;
@@ -355,11 +357,12 @@ static ssize_t bch_dump_read(struct file *file, char __user *buf,
 		if (i->bytes)
 			break;
 
-		w = bch_keybuf_next_rescan(i->c, &i->keys, &MAX_KEY);
+		w = bch_keybuf_next_rescan(i->c, &i->keys, &MAX_KEY, dump_pred);
 		if (!w)
 			break;
 
-		i->bytes = snprintf(i->buf, PAGE_SIZE, "%s\n", pkey(&w->key));
+		bch_bkey_to_text(kbuf, sizeof(kbuf), &w->key);
+		i->bytes = snprintf(i->buf, PAGE_SIZE, "%s\n", kbuf);
 		bch_keybuf_del(&i->keys, w);
 	}
 
@@ -377,7 +380,7 @@ static int bch_dump_open(struct inode *inode, struct file *file)
 
 	file->private_data = i;
 	i->c = c;
-	bch_keybuf_init(&i->keys, dump_pred);
+	bch_keybuf_init(&i->keys);
 	i->keys.last_scanned = KEY(0, 0, 0);
 
 	return 0;
@@ -409,142 +412,6 @@ void bch_debug_init_cache_set(struct cache_set *c)
 
 #endif
 
-/* Fuzz tester has rotted: */
-#if 0
-
-static ssize_t btree_fuzz(struct kobject *k, struct kobj_attribute *a,
-			  const char *buffer, size_t size)
-{
-	void dump(struct btree *b)
-	{
-		struct bset *i;
-
-		for (i = b->sets[0].data;
-		     index(i, b) < btree_blocks(b) &&
-		     i->seq == b->sets[0].data->seq;
-		     i = ((void *) i) + set_blocks(i, b->c) * block_bytes(b->c))
-			dump_bset(b, i);
-	}
-
-	struct cache_sb *sb;
-	struct cache_set *c;
-	struct btree *all[3], *b, *fill, *orig;
-	int j;
-
-	struct btree_op op;
-	bch_btree_op_init_stack(&op);
-
-	sb = kzalloc(sizeof(struct cache_sb), GFP_KERNEL);
-	if (!sb)
-		return -ENOMEM;
-
-	sb->bucket_size = 128;
-	sb->block_size = 4;
-
-	c = bch_cache_set_alloc(sb);
-	if (!c)
-		return -ENOMEM;
-
-	for (j = 0; j < 3; j++) {
-		BUG_ON(list_empty(&c->btree_cache));
-		all[j] = list_first_entry(&c->btree_cache, struct btree, list);
-		list_del_init(&all[j]->list);
-
-		all[j]->key = KEY(0, 0, c->sb.bucket_size);
-		bkey_copy_key(&all[j]->key, &MAX_KEY);
-	}
-
-	b = all[0];
-	fill = all[1];
-	orig = all[2];
-
-	while (1) {
-		for (j = 0; j < 3; j++)
-			all[j]->written = all[j]->nsets = 0;
-
-		bch_bset_init_next(b);
-
-		while (1) {
-			struct bset *i = write_block(b);
-			struct bkey *k = op.keys.top;
-			unsigned rand;
-
-			bkey_init(k);
-			rand = get_random_int();
-
-			op.type = rand & 1
-				? BTREE_INSERT
-				: BTREE_REPLACE;
-			rand >>= 1;
-
-			SET_KEY_SIZE(k, bucket_remainder(c, rand));
-			rand >>= c->bucket_bits;
-			rand &= 1024 * 512 - 1;
-			rand += c->sb.bucket_size;
-			SET_KEY_OFFSET(k, rand);
-#if 0
-			SET_KEY_PTRS(k, 1);
-#endif
-			bch_keylist_push(&op.keys);
-			bch_btree_insert_keys(b, &op);
-
-			if (should_split(b) ||
-			    set_blocks(i, b->c) !=
-			    __set_blocks(i, i->keys + 15, b->c)) {
-				i->csum = csum_set(i);
-
-				memcpy(write_block(fill),
-				       i, set_bytes(i));
-
-				b->written += set_blocks(i, b->c);
-				fill->written = b->written;
-				if (b->written == btree_blocks(b))
-					break;
-
-				bch_btree_sort_lazy(b);
-				bch_bset_init_next(b);
-			}
-		}
-
-		memcpy(orig->sets[0].data,
-		       fill->sets[0].data,
-		       btree_bytes(c));
-
-		bch_btree_sort(b);
-		fill->written = 0;
-		bch_btree_read_done(&fill->io.cl);
-
-		if (b->sets[0].data->keys != fill->sets[0].data->keys ||
-		    memcmp(b->sets[0].data->start,
-			   fill->sets[0].data->start,
-			   b->sets[0].data->keys * sizeof(uint64_t))) {
-			struct bset *i = b->sets[0].data;
-			struct bkey *k, *l;
-
-			for (k = i->start,
-			     l = fill->sets[0].data->start;
-			     k < end(i);
-			     k = bkey_next(k), l = bkey_next(l))
-				if (bkey_cmp(k, l) ||
-				    KEY_SIZE(k) != KEY_SIZE(l))
-					pr_err("key %zi differs: %s != %s",
-					       (uint64_t *) k - i->d,
-					       pkey(k), pkey(l));
-
-			for (j = 0; j < 3; j++) {
-				pr_err("**** Set %i ****", j);
-				dump(all[j]);
-			}
-			panic("\n");
-		}
-
-		pr_info("fuzz complete: %i keys", b->sets[0].data->keys);
-	}
-}
-
-kobj_attribute_write(fuzz, btree_fuzz);
-#endif
-
 void bch_debug_exit(void)
 {
 	if (!IS_ERR_OR_NULL(debug))
@@ -554,11 +421,6 @@ void bch_debug_exit(void)
 int __init bch_debug_init(struct kobject *kobj)
 {
 	int ret = 0;
-#if 0
-	ret = sysfs_create_file(kobj, &ksysfs_fuzz.attr);
-	if (ret)
-		return ret;
-#endif
 
 	debug = debugfs_create_dir("bcache", NULL);
 	return ret;

drivers/md/bcache/debug.h

@@ -3,15 +3,8 @@
 
 /* Btree/bkey debug printing */
 
-#define KEYHACK_SIZE 80
+int bch_bkey_to_text(char *buf, size_t size, const struct bkey *k);
-struct keyprint_hack {
+int bch_btree_to_text(char *buf, size_t size, const struct btree *b);
-	char s[KEYHACK_SIZE];
-};
-
-struct keyprint_hack bch_pkey(const struct bkey *k);
-struct keyprint_hack bch_pbtree(const struct btree *b);
-#define pkey(k)		(&bch_pkey(k).s[0])
-#define pbtree(b)	(&bch_pbtree(b).s[0])
 
 #ifdef CONFIG_BCACHE_EDEBUG
 

drivers/md/bcache/io.c

@@ -9,6 +9,8 @@
 #include "bset.h"
 #include "debug.h"
 
+#include <linux/blkdev.h>
+
 static void bch_bi_idx_hack_endio(struct bio *bio, int error)
 {
 	struct bio *p = bio->bi_private;
@@ -66,13 +68,6 @@ static void bch_generic_make_request_hack(struct bio *bio)
  * The newly allocated bio will point to @bio's bi_io_vec, if the split was on a
  * bvec boundry; it is the caller's responsibility to ensure that @bio is not
  * freed before the split.
- *
- * If bch_bio_split() is running under generic_make_request(), it's not safe to
- * allocate more than one bio from the same bio set. Therefore, if it is running
- * under generic_make_request() it masks out __GFP_WAIT when doing the
- * allocation. The caller must check for failure if there's any possibility of
- * it being called from under generic_make_request(); it is then the caller's
- * responsibility to retry from a safe context (by e.g. punting to workqueue).
  */
 struct bio *bch_bio_split(struct bio *bio, int sectors,
 			  gfp_t gfp, struct bio_set *bs)
@@ -83,20 +78,13 @@ struct bio *bch_bio_split(struct bio *bio, int sectors,
 
 	BUG_ON(sectors <= 0);
 
-	/*
-	 * If we're being called from underneath generic_make_request() and we
-	 * already allocated any bios from this bio set, we risk deadlock if we
-	 * use the mempool. So instead, we possibly fail and let the caller punt
-	 * to workqueue or somesuch and retry in a safe context.
-	 */
-	if (current->bio_list)
-		gfp &= ~__GFP_WAIT;
-
 	if (sectors >= bio_sectors(bio))
 		return bio;
 
 	if (bio->bi_rw & REQ_DISCARD) {
 		ret = bio_alloc_bioset(gfp, 1, bs);
+		if (!ret)
+			return NULL;
 		idx = 0;
 		goto out;
 	}
@@ -160,17 +148,18 @@ static unsigned bch_bio_max_sectors(struct bio *bio)
 	struct request_queue *q = bdev_get_queue(bio->bi_bdev);
 	unsigned max_segments = min_t(unsigned, BIO_MAX_PAGES,
 				      queue_max_segments(q));
-	struct bio_vec *bv, *end = bio_iovec(bio) +
-		min_t(int, bio_segments(bio), max_segments);
 
 	if (bio->bi_rw & REQ_DISCARD)
 		return min(ret, q->limits.max_discard_sectors);
 
 	if (bio_segments(bio) > max_segments ||
 	    q->merge_bvec_fn) {
+		struct bio_vec *bv;
+		int i, seg = 0;
+
 		ret = 0;
 
-		for (bv = bio_iovec(bio); bv < end; bv++) {
+		bio_for_each_segment(bv, bio, i) {
 			struct bvec_merge_data bvm = {
 				.bi_bdev	= bio->bi_bdev,
 				.bi_sector	= bio->bi_sector,
@@ -178,10 +167,14 @@ static unsigned bch_bio_max_sectors(struct bio *bio)
 				.bi_rw		= bio->bi_rw,
 			};
 
+			if (seg == max_segments)
+				break;
+
 			if (q->merge_bvec_fn &&
 			    q->merge_bvec_fn(q, &bvm, bv) < (int) bv->bv_len)
 				break;
 
+			seg++;
 			ret += bv->bv_len >> 9;
 		}
 	}
@@ -218,30 +211,10 @@ static void bch_bio_submit_split_endio(struct bio *bio, int error)
 	closure_put(cl);
 }
 
-static void __bch_bio_submit_split(struct closure *cl)
-{
-	struct bio_split_hook *s = container_of(cl, struct bio_split_hook, cl);
-	struct bio *bio = s->bio, *n;
-
-	do {
-		n = bch_bio_split(bio, bch_bio_max_sectors(bio),
-				  GFP_NOIO, s->p->bio_split);
-		if (!n)
-			continue_at(cl, __bch_bio_submit_split, system_wq);
-
-		n->bi_end_io	= bch_bio_submit_split_endio;
-		n->bi_private	= cl;
-
-		closure_get(cl);
-		bch_generic_make_request_hack(n);
-	} while (n != bio);
-
-	continue_at(cl, bch_bio_submit_split_done, NULL);
-}
-
 void bch_generic_make_request(struct bio *bio, struct bio_split_pool *p)
 {
 	struct bio_split_hook *s;
+	struct bio *n;
 
 	if (!bio_has_data(bio) && !(bio->bi_rw & REQ_DISCARD))
 		goto submit;
@@ -250,6 +223,7 @@ void bch_generic_make_request(struct bio *bio, struct bio_split_pool *p)
 		goto submit;
 
 	s = mempool_alloc(p->bio_split_hook, GFP_NOIO);
+	closure_init(&s->cl, NULL);
 
 	s->bio		= bio;
 	s->p		= p;
@@ -257,8 +231,18 @@ void bch_generic_make_request(struct bio *bio, struct bio_split_pool *p)
 	s->bi_private	= bio->bi_private;
 	bio_get(bio);
 
-	closure_call(&s->cl, __bch_bio_submit_split, NULL, NULL);
+	do {
-	return;
+		n = bch_bio_split(bio, bch_bio_max_sectors(bio),
+				  GFP_NOIO, s->p->bio_split);
+
+		n->bi_end_io	= bch_bio_submit_split_endio;
+		n->bi_private	= &s->cl;
+
+		closure_get(&s->cl);
+		bch_generic_make_request_hack(n);
+	} while (n != bio);
+
+	continue_at(&s->cl, bch_bio_submit_split_done, NULL);
 submit:
 	bch_generic_make_request_hack(bio);
 }

drivers/md/bcache/journal.c

@@ -9,6 +9,8 @@
 #include "debug.h"
 #include "request.h"
 
+#include <trace/events/bcache.h>
+
 /*
  * Journal replay/recovery:
  *
@@ -182,9 +184,14 @@ bsearch:
 		pr_debug("starting binary search, l %u r %u", l, r);
 
 		while (l + 1 < r) {
-			m = (l + r) >> 1;
+			seq = list_entry(list->prev, struct journal_replay,
+					 list)->j.seq;
 
-			if (read_bucket(m))
+			m = (l + r) >> 1;
+			read_bucket(m);
+
+			if (seq != list_entry(list->prev, struct journal_replay,
+					      list)->j.seq)
 				l = m;
 			else
 				r = m;
@@ -300,7 +307,8 @@ int bch_journal_replay(struct cache_set *s, struct list_head *list,
 		for (k = i->j.start;
 		     k < end(&i->j);
 		     k = bkey_next(k)) {
-			pr_debug("%s", pkey(k));
+			trace_bcache_journal_replay_key(k);
+
 			bkey_copy(op->keys.top, k);
 			bch_keylist_push(&op->keys);
 
@@ -384,7 +392,7 @@ out:
 		return;
 found:
 	if (btree_node_dirty(best))
-		bch_btree_write(best, true, NULL);
+		bch_btree_node_write(best, NULL);
 	rw_unlock(true, best);
 }
 
@@ -617,7 +625,7 @@ static void journal_write_unlocked(struct closure *cl)
 		bio_reset(bio);
 		bio->bi_sector	= PTR_OFFSET(k, i);
 		bio->bi_bdev	= ca->bdev;
-		bio->bi_rw	= REQ_WRITE|REQ_SYNC|REQ_META|REQ_FLUSH;
+		bio->bi_rw	= REQ_WRITE|REQ_SYNC|REQ_META|REQ_FLUSH|REQ_FUA;
 		bio->bi_size	= sectors << 9;
 
 		bio->bi_end_io	= journal_write_endio;
@@ -712,7 +720,8 @@ void bch_journal(struct closure *cl)
 	spin_lock(&c->journal.lock);
 
 	if (journal_full(&c->journal)) {
-		/* XXX: tracepoint */
+		trace_bcache_journal_full(c);
+
 		closure_wait(&c->journal.wait, cl);
 
 		journal_reclaim(c);
@@ -728,13 +737,15 @@ void bch_journal(struct closure *cl)
 
 	if (b * c->sb.block_size > PAGE_SECTORS << JSET_BITS ||
 	    b > c->journal.blocks_free) {
-		/* XXX: If we were inserting so many keys that they won't fit in
+		trace_bcache_journal_entry_full(c);
+
+		/*
+		 * XXX: If we were inserting so many keys that they won't fit in
 		 * an _empty_ journal write, we'll deadlock. For now, handle
 		 * this in bch_keylist_realloc() - but something to think about.
 		 */
 		BUG_ON(!w->data->keys);
 
-		/* XXX: tracepoint */
 		BUG_ON(!closure_wait(&w->wait, cl));
 
 		closure_flush(&c->journal.io);

drivers/md/bcache/movinggc.c

@@ -9,6 +9,8 @@
 #include "debug.h"
 #include "request.h"
 
+#include <trace/events/bcache.h>
+
 struct moving_io {
 	struct keybuf_key	*w;
 	struct search		s;
@@ -44,14 +46,14 @@ static void write_moving_finish(struct closure *cl)
 {
 	struct moving_io *io = container_of(cl, struct moving_io, s.cl);
 	struct bio *bio = &io->bio.bio;
-	struct bio_vec *bv = bio_iovec_idx(bio, bio->bi_vcnt);
+	struct bio_vec *bv;
+	int i;
 
-	while (bv-- != bio->bi_io_vec)
+	bio_for_each_segment_all(bv, bio, i)
 		__free_page(bv->bv_page);
 
-	pr_debug("%s %s", io->s.op.insert_collision
+	if (io->s.op.insert_collision)
-		 ? "collision moving" : "moved",
+		trace_bcache_gc_copy_collision(&io->w->key);
-		 pkey(&io->w->key));
 
 	bch_keybuf_del(&io->s.op.c->moving_gc_keys, io->w);
 
@@ -94,8 +96,6 @@ static void write_moving(struct closure *cl)
 	struct moving_io *io = container_of(s, struct moving_io, s);
 
 	if (!s->error) {
-		trace_bcache_write_moving(&io->bio.bio);
-
 		moving_init(io);
 
 		io->bio.bio.bi_sector	= KEY_START(&io->w->key);
@@ -122,7 +122,6 @@ static void read_moving_submit(struct closure *cl)
 	struct moving_io *io = container_of(s, struct moving_io, s);
 	struct bio *bio = &io->bio.bio;
 
-	trace_bcache_read_moving(bio);
 	bch_submit_bbio(bio, s->op.c, &io->w->key, 0);
 
 	continue_at(cl, write_moving, bch_gc_wq);
@@ -138,7 +137,8 @@ static void read_moving(struct closure *cl)
 	/* XXX: if we error, background writeback could stall indefinitely */
 
 	while (!test_bit(CACHE_SET_STOPPING, &c->flags)) {
-		w = bch_keybuf_next_rescan(c, &c->moving_gc_keys, &MAX_KEY);
+		w = bch_keybuf_next_rescan(c, &c->moving_gc_keys,
+					   &MAX_KEY, moving_pred);
 		if (!w)
 			break;
 
@@ -159,10 +159,10 @@ static void read_moving(struct closure *cl)
 		bio->bi_rw	= READ;
 		bio->bi_end_io	= read_moving_endio;
 
-		if (bch_bio_alloc_pages(bio, GFP_KERNEL))
+		if (bio_alloc_pages(bio, GFP_KERNEL))
 			goto err;
 
-		pr_debug("%s", pkey(&w->key));
+		trace_bcache_gc_copy(&w->key);
 
 		closure_call(&io->s.cl, read_moving_submit, NULL, &c->gc.cl);
 
@@ -250,5 +250,5 @@ void bch_moving_gc(struct closure *cl)
 
 void bch_moving_init_cache_set(struct cache_set *c)
 {
-	bch_keybuf_init(&c->moving_gc_keys, moving_pred);
+	bch_keybuf_init(&c->moving_gc_keys);
 }

drivers/md/bcache/request.c

@@ -10,6 +10,7 @@
 #include "btree.h"
 #include "debug.h"
 #include "request.h"
+#include "writeback.h"
 
 #include <linux/cgroup.h>
 #include <linux/module.h>
@@ -21,8 +22,6 @@
 
 #define CUTOFF_CACHE_ADD	95
 #define CUTOFF_CACHE_READA	90
-#define CUTOFF_WRITEBACK	50
-#define CUTOFF_WRITEBACK_SYNC	75
 
 struct kmem_cache *bch_search_cache;
 
@@ -489,6 +488,12 @@ static void bch_insert_data_loop(struct closure *cl)
 		bch_queue_gc(op->c);
 	}
 
+	/*
+	 * Journal writes are marked REQ_FLUSH; if the original write was a
+	 * flush, it'll wait on the journal write.
+	 */
+	bio->bi_rw &= ~(REQ_FLUSH|REQ_FUA);
+
 	do {
 		unsigned i;
 		struct bkey *k;
@@ -510,10 +515,6 @@ static void bch_insert_data_loop(struct closure *cl)
 			goto err;
 
 		n = bch_bio_split(bio, KEY_SIZE(k), GFP_NOIO, split);
-		if (!n) {
-			__bkey_put(op->c, k);
-			continue_at(cl, bch_insert_data_loop, bcache_wq);
-		}
 
 		n->bi_end_io	= bch_insert_data_endio;
 		n->bi_private	= cl;
@@ -530,10 +531,9 @@ static void bch_insert_data_loop(struct closure *cl)
 		if (KEY_CSUM(k))
 			bio_csum(n, k);
 
-		pr_debug("%s", pkey(k));
+		trace_bcache_cache_insert(k);
 		bch_keylist_push(&op->keys);
 
-		trace_bcache_cache_insert(n, n->bi_sector, n->bi_bdev);
 		n->bi_rw |= REQ_WRITE;
 		bch_submit_bbio(n, op->c, k, 0);
 	} while (n != bio);
@@ -716,7 +716,7 @@ static struct search *search_alloc(struct bio *bio, struct bcache_device *d)
 	s->task			= current;
 	s->orig_bio		= bio;
 	s->write		= (bio->bi_rw & REQ_WRITE) != 0;
-	s->op.flush_journal	= (bio->bi_rw & REQ_FLUSH) != 0;
+	s->op.flush_journal	= (bio->bi_rw & (REQ_FLUSH|REQ_FUA)) != 0;
 	s->op.skip		= (bio->bi_rw & REQ_DISCARD) != 0;
 	s->recoverable		= 1;
 	s->start_time		= jiffies;
@@ -784,11 +784,8 @@ static void request_read_error(struct closure *cl)
 	int i;
 
 	if (s->recoverable) {
-		/* The cache read failed, but we can retry from the backing
+		/* Retry from the backing device: */
-		 * device.
+		trace_bcache_read_retry(s->orig_bio);
-		 */
-		pr_debug("recovering at sector %llu",
-			 (uint64_t) s->orig_bio->bi_sector);
 
 		s->error = 0;
 		bv = s->bio.bio.bi_io_vec;
@@ -806,7 +803,6 @@ static void request_read_error(struct closure *cl)
 
 		/* XXX: invalidate cache */
 
-		trace_bcache_read_retry(&s->bio.bio);
 		closure_bio_submit(&s->bio.bio, &s->cl, s->d);
 	}
 
@@ -827,53 +823,13 @@ static void request_read_done(struct closure *cl)
 	 */
 
 	if (s->op.cache_bio) {
-		struct bio_vec *src, *dst;
-		unsigned src_offset, dst_offset, bytes;
-		void *dst_ptr;
-
 		bio_reset(s->op.cache_bio);
 		s->op.cache_bio->bi_sector	= s->cache_miss->bi_sector;
 		s->op.cache_bio->bi_bdev	= s->cache_miss->bi_bdev;
 		s->op.cache_bio->bi_size	= s->cache_bio_sectors << 9;
 		bch_bio_map(s->op.cache_bio, NULL);
 
-		src = bio_iovec(s->op.cache_bio);
+		bio_copy_data(s->cache_miss, s->op.cache_bio);
-		dst = bio_iovec(s->cache_miss);
-		src_offset = src->bv_offset;
-		dst_offset = dst->bv_offset;
-		dst_ptr = kmap(dst->bv_page);
-
-		while (1) {
-			if (dst_offset == dst->bv_offset + dst->bv_len) {
-				kunmap(dst->bv_page);
-				dst++;
-				if (dst == bio_iovec_idx(s->cache_miss,
-						s->cache_miss->bi_vcnt))
-					break;
-
-				dst_offset = dst->bv_offset;
-				dst_ptr = kmap(dst->bv_page);
-			}
-
-			if (src_offset == src->bv_offset + src->bv_len) {
-				src++;
-				if (src == bio_iovec_idx(s->op.cache_bio,
-						 s->op.cache_bio->bi_vcnt))
-					BUG();
-
-				src_offset = src->bv_offset;
-			}
-
-			bytes = min(dst->bv_offset + dst->bv_len - dst_offset,
-				    src->bv_offset + src->bv_len - src_offset);
-
-			memcpy(dst_ptr + dst_offset,
-			       page_address(src->bv_page) + src_offset,
-			       bytes);
-
-			src_offset	+= bytes;
-			dst_offset	+= bytes;
-		}
 
 		bio_put(s->cache_miss);
 		s->cache_miss = NULL;
@@ -899,6 +855,7 @@ static void request_read_done_bh(struct closure *cl)
 	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
 
 	bch_mark_cache_accounting(s, !s->cache_miss, s->op.skip);
+	trace_bcache_read(s->orig_bio, !s->cache_miss, s->op.skip);
 
 	if (s->error)
 		continue_at_nobarrier(cl, request_read_error, bcache_wq);
@@ -917,9 +874,6 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 	struct bio *miss;
 
 	miss = bch_bio_split(bio, sectors, GFP_NOIO, s->d->bio_split);
-	if (!miss)
-		return -EAGAIN;
-
 	if (miss == bio)
 		s->op.lookup_done = true;
 
@@ -938,8 +892,9 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 		reada = min(dc->readahead >> 9,
 			    sectors - bio_sectors(miss));
 
-		if (bio_end(miss) + reada > bdev_sectors(miss->bi_bdev))
+		if (bio_end_sector(miss) + reada > bdev_sectors(miss->bi_bdev))
-			reada = bdev_sectors(miss->bi_bdev) - bio_end(miss);
+			reada = bdev_sectors(miss->bi_bdev) -
+				bio_end_sector(miss);
 	}
 
 	s->cache_bio_sectors = bio_sectors(miss) + reada;
@@ -963,13 +918,12 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 		goto out_put;
 
 	bch_bio_map(s->op.cache_bio, NULL);
-	if (bch_bio_alloc_pages(s->op.cache_bio, __GFP_NOWARN|GFP_NOIO))
+	if (bio_alloc_pages(s->op.cache_bio, __GFP_NOWARN|GFP_NOIO))
 		goto out_put;
 
 	s->cache_miss = miss;
 	bio_get(s->op.cache_bio);
 
-	trace_bcache_cache_miss(s->orig_bio);
 	closure_bio_submit(s->op.cache_bio, &s->cl, s->d);
 
 	return ret;
@@ -1002,24 +956,13 @@ static void cached_dev_write_complete(struct closure *cl)
 	cached_dev_bio_complete(cl);
 }
 
-static bool should_writeback(struct cached_dev *dc, struct bio *bio)
-{
-	unsigned threshold = (bio->bi_rw & REQ_SYNC)
-		? CUTOFF_WRITEBACK_SYNC
-		: CUTOFF_WRITEBACK;
-
-	return !atomic_read(&dc->disk.detaching) &&
-		cache_mode(dc, bio) == CACHE_MODE_WRITEBACK &&
-		dc->disk.c->gc_stats.in_use < threshold;
-}
-
 static void request_write(struct cached_dev *dc, struct search *s)
 {
 	struct closure *cl = &s->cl;
 	struct bio *bio = &s->bio.bio;
 	struct bkey start, end;
 	start = KEY(dc->disk.id, bio->bi_sector, 0);
-	end = KEY(dc->disk.id, bio_end(bio), 0);
+	end = KEY(dc->disk.id, bio_end_sector(bio), 0);
 
 	bch_keybuf_check_overlapping(&s->op.c->moving_gc_keys, &start, &end);
 
@@ -1034,22 +977,37 @@ static void request_write(struct cached_dev *dc, struct search *s)
 	if (bio->bi_rw & REQ_DISCARD)
 		goto skip;
 
+	if (should_writeback(dc, s->orig_bio,
+			     cache_mode(dc, bio),
+			     s->op.skip)) {
+		s->op.skip = false;
+		s->writeback = true;
+	}
+
 	if (s->op.skip)
 		goto skip;
 
-	if (should_writeback(dc, s->orig_bio))
+	trace_bcache_write(s->orig_bio, s->writeback, s->op.skip);
-		s->writeback = true;
 
 	if (!s->writeback) {
 		s->op.cache_bio = bio_clone_bioset(bio, GFP_NOIO,
 						   dc->disk.bio_split);
 
-		trace_bcache_writethrough(s->orig_bio);
+		closure_bio_submit(bio, cl, s->d);
+	} else {
+		bch_writeback_add(dc);
+
+		if (s->op.flush_journal) {
+			/* Also need to send a flush to the backing device */
+			s->op.cache_bio = bio_clone_bioset(bio, GFP_NOIO,
+							   dc->disk.bio_split);
+
+			bio->bi_size = 0;
+			bio->bi_vcnt = 0;
 			closure_bio_submit(bio, cl, s->d);
 		} else {
 			s->op.cache_bio = bio;
-		trace_bcache_writeback(s->orig_bio);
+		}
-		bch_writeback_add(dc, bio_sectors(bio));
 	}
 out:
 	closure_call(&s->op.cl, bch_insert_data, NULL, cl);
@@ -1058,7 +1016,6 @@ skip:
 	s->op.skip = true;
 	s->op.cache_bio = s->orig_bio;
 	bio_get(s->op.cache_bio);
-	trace_bcache_write_skip(s->orig_bio);
 
 	if ((bio->bi_rw & REQ_DISCARD) &&
 	    !blk_queue_discard(bdev_get_queue(dc->bdev)))
@@ -1088,9 +1045,10 @@ static void request_nodata(struct cached_dev *dc, struct search *s)
 
 /* Cached devices - read & write stuff */
 
-int bch_get_congested(struct cache_set *c)
+unsigned bch_get_congested(struct cache_set *c)
 {
 	int i;
+	long rand;
 
 	if (!c->congested_read_threshold_us &&
 	    !c->congested_write_threshold_us)
@@ -1106,7 +1064,13 @@ int bch_get_congested(struct cache_set *c)
 
 	i += CONGESTED_MAX;
 
-	return i <= 0 ? 1 : fract_exp_two(i, 6);
+	if (i > 0)
+		i = fract_exp_two(i, 6);
+
+	rand = get_random_int();
+	i -= bitmap_weight(&rand, BITS_PER_LONG);
+
+	return i > 0 ? i : 1;
 }
 
 static void add_sequential(struct task_struct *t)
@@ -1126,10 +1090,8 @@ static void check_should_skip(struct cached_dev *dc, struct search *s)
 {
 	struct cache_set *c = s->op.c;
 	struct bio *bio = &s->bio.bio;
-
-	long rand;
-	int cutoff = bch_get_congested(c);
 	unsigned mode = cache_mode(dc, bio);
+	unsigned sectors, congested = bch_get_congested(c);
 
 	if (atomic_read(&dc->disk.detaching) ||
 	    c->gc_stats.in_use > CUTOFF_CACHE_ADD ||
@@ -1147,17 +1109,14 @@ static void check_should_skip(struct cached_dev *dc, struct search *s)
 		goto skip;
 	}
 
-	if (!cutoff) {
+	if (!congested && !dc->sequential_cutoff)
-		cutoff = dc->sequential_cutoff >> 9;
-
-		if (!cutoff)
 		goto rescale;
 
-		if (mode == CACHE_MODE_WRITEBACK &&
+	if (!congested &&
+	    mode == CACHE_MODE_WRITEBACK &&
 	    (bio->bi_rw & REQ_WRITE) &&
 	    (bio->bi_rw & REQ_SYNC))
 		goto rescale;
-	}
 
 	if (dc->sequential_merge) {
 		struct io *i;
@@ -1177,7 +1136,7 @@ found:
 		if (i->sequential + bio->bi_size > i->sequential)
 			i->sequential	+= bio->bi_size;
 
-		i->last			 = bio_end(bio);
+		i->last			 = bio_end_sector(bio);
 		i->jiffies		 = jiffies + msecs_to_jiffies(5000);
 		s->task->sequential_io	 = i->sequential;
 
@@ -1192,12 +1151,19 @@ found:
 		add_sequential(s->task);
 	}
 
-	rand = get_random_int();
+	sectors = max(s->task->sequential_io,
-	cutoff -= bitmap_weight(&rand, BITS_PER_LONG);
+		      s->task->sequential_io_avg) >> 9;
 
-	if (cutoff <= (int) (max(s->task->sequential_io,
+	if (dc->sequential_cutoff &&
-				 s->task->sequential_io_avg) >> 9))
+	    sectors >= dc->sequential_cutoff >> 9) {
+		trace_bcache_bypass_sequential(s->orig_bio);
 		goto skip;
+	}
+
+	if (congested && sectors >= congested) {
+		trace_bcache_bypass_congested(s->orig_bio);
+		goto skip;
+	}
 
 rescale:
 	bch_rescale_priorities(c, bio_sectors(bio));
@@ -1288,29 +1254,24 @@ void bch_cached_dev_request_init(struct cached_dev *dc)
 static int flash_dev_cache_miss(struct btree *b, struct search *s,
 				struct bio *bio, unsigned sectors)
 {
+	struct bio_vec *bv;
+	int i;
+
 	/* Zero fill bio */
 
-	while (bio->bi_idx != bio->bi_vcnt) {
+	bio_for_each_segment(bv, bio, i) {
-		struct bio_vec *bv = bio_iovec(bio);
 		unsigned j = min(bv->bv_len >> 9, sectors);
 
 		void *p = kmap(bv->bv_page);
 		memset(p + bv->bv_offset, 0, j << 9);
 		kunmap(bv->bv_page);
 
-		bv->bv_len	-= j << 9;
-		bv->bv_offset	+= j << 9;
-
-		if (bv->bv_len)
-			return 0;
-
-		bio->bi_sector	+= j;
-		bio->bi_size	-= j << 9;
-
-		bio->bi_idx++;
 		sectors	-= j;
 	}
 
+	bio_advance(bio, min(sectors << 9, bio->bi_size));
+
+	if (!bio->bi_size)
 		s->op.lookup_done = true;
 
 	return 0;
@@ -1339,7 +1300,7 @@ static void flash_dev_make_request(struct request_queue *q, struct bio *bio)
 	} else if (bio_has_data(bio) || s->op.skip) {
 		bch_keybuf_check_overlapping(&s->op.c->moving_gc_keys,
 					&KEY(d->id, bio->bi_sector, 0),
-					     &KEY(d->id, bio_end(bio), 0));
+					&KEY(d->id, bio_end_sector(bio), 0));
 
 		s->writeback	= true;
 		s->op.cache_bio	= bio;

drivers/md/bcache/request.h

@@ -30,7 +30,7 @@ struct search {
 };
 
 void bch_cache_read_endio(struct bio *, int);
-int bch_get_congested(struct cache_set *);
+unsigned bch_get_congested(struct cache_set *);
 void bch_insert_data(struct closure *cl);
 void bch_btree_insert_async(struct closure *);
 void bch_cache_read_endio(struct bio *, int);

drivers/md/bcache/super.c

@@ -10,10 +10,13 @@
 #include "btree.h"
 #include "debug.h"
 #include "request.h"
+#include "writeback.h"
 
+#include <linux/blkdev.h>
 #include <linux/buffer_head.h>
 #include <linux/debugfs.h>
 #include <linux/genhd.h>
+#include <linux/kthread.h>
 #include <linux/module.h>
 #include <linux/random.h>
 #include <linux/reboot.h>
@@ -342,6 +345,7 @@ static void uuid_io(struct cache_set *c, unsigned long rw,
 	struct closure *cl = &c->uuid_write.cl;
 	struct uuid_entry *u;
 	unsigned i;
+	char buf[80];
 
 	BUG_ON(!parent);
 	closure_lock(&c->uuid_write, parent);
@@ -362,8 +366,8 @@ static void uuid_io(struct cache_set *c, unsigned long rw,
 			break;
 	}
 
-	pr_debug("%s UUIDs at %s", rw & REQ_WRITE ? "wrote" : "read",
+	bch_bkey_to_text(buf, sizeof(buf), k);
-		 pkey(&c->uuid_bucket));
+	pr_debug("%s UUIDs at %s", rw & REQ_WRITE ? "wrote" : "read", buf);
 
 	for (u = c->uuids; u < c->uuids + c->nr_uuids; u++)
 		if (!bch_is_zero(u->uuid, 16))
@@ -543,7 +547,6 @@ void bch_prio_write(struct cache *ca)
 
 	pr_debug("free %zu, free_inc %zu, unused %zu", fifo_used(&ca->free),
 		 fifo_used(&ca->free_inc), fifo_used(&ca->unused));
-	blktrace_msg(ca, "Starting priorities: " buckets_free(ca));
 
 	for (i = prio_buckets(ca) - 1; i >= 0; --i) {
 		long bucket;
@@ -704,6 +707,7 @@ static void bcache_device_detach(struct bcache_device *d)
 		atomic_set(&d->detaching, 0);
 	}
 
+	if (!d->flush_done)
 		bcache_device_unlink(d);
 
 	d->c->devices[d->id] = NULL;
@@ -743,13 +747,35 @@ static void bcache_device_free(struct bcache_device *d)
 		mempool_destroy(d->unaligned_bvec);
 	if (d->bio_split)
 		bioset_free(d->bio_split);
+	if (is_vmalloc_addr(d->stripe_sectors_dirty))
+		vfree(d->stripe_sectors_dirty);
+	else
+		kfree(d->stripe_sectors_dirty);
 
 	closure_debug_destroy(&d->cl);
 }
 
-static int bcache_device_init(struct bcache_device *d, unsigned block_size)
+static int bcache_device_init(struct bcache_device *d, unsigned block_size,
+			      sector_t sectors)
 {
 	struct request_queue *q;
+	size_t n;
+
+	if (!d->stripe_size_bits)
+		d->stripe_size_bits = 31;
+
+	d->nr_stripes = round_up(sectors, 1 << d->stripe_size_bits) >>
+		d->stripe_size_bits;
+
+	if (!d->nr_stripes || d->nr_stripes > SIZE_MAX / sizeof(atomic_t))
+		return -ENOMEM;
+
+	n = d->nr_stripes * sizeof(atomic_t);
+	d->stripe_sectors_dirty = n < PAGE_SIZE << 6
+		? kzalloc(n, GFP_KERNEL)
+		: vzalloc(n);
+	if (!d->stripe_sectors_dirty)
+		return -ENOMEM;
 
 	if (!(d->bio_split = bioset_create(4, offsetof(struct bbio, bio))) ||
 	    !(d->unaligned_bvec = mempool_create_kmalloc_pool(1,
@@ -759,6 +785,7 @@ static int bcache_device_init(struct bcache_device *d, unsigned block_size)
 	    !(q = blk_alloc_queue(GFP_KERNEL)))
 		return -ENOMEM;
 
+	set_capacity(d->disk, sectors);
 	snprintf(d->disk->disk_name, DISK_NAME_LEN, "bcache%i", bcache_minor);
 
 	d->disk->major		= bcache_major;
@@ -781,6 +808,8 @@ static int bcache_device_init(struct bcache_device *d, unsigned block_size)
 	set_bit(QUEUE_FLAG_NONROT,	&d->disk->queue->queue_flags);
 	set_bit(QUEUE_FLAG_DISCARD,	&d->disk->queue->queue_flags);
 
+	blk_queue_flush(q, REQ_FLUSH|REQ_FUA);
+
 	return 0;
 }
 
@@ -800,6 +829,17 @@ static void calc_cached_dev_sectors(struct cache_set *c)
 void bch_cached_dev_run(struct cached_dev *dc)
 {
 	struct bcache_device *d = &dc->disk;
+	char buf[SB_LABEL_SIZE + 1];
+	char *env[] = {
+		"DRIVER=bcache",
+		kasprintf(GFP_KERNEL, "CACHED_UUID=%pU", dc->sb.uuid),
+		NULL,
+		NULL,
+	};
+
+	memcpy(buf, dc->sb.label, SB_LABEL_SIZE);
+	buf[SB_LABEL_SIZE] = '\0';
+	env[2] = kasprintf(GFP_KERNEL, "CACHED_LABEL=%s", buf);
 
 	if (atomic_xchg(&dc->running, 1))
 		return;
@@ -816,10 +856,12 @@ void bch_cached_dev_run(struct cached_dev *dc)
 
 	add_disk(d->disk);
 	bd_link_disk_holder(dc->bdev, dc->disk.disk);
-#if 0
+	/* won't show up in the uevent file, use udevadm monitor -e instead
-	char *env[] = { "SYMLINK=label" , NULL };
+	 * only class / kset properties are persistent */
 	kobject_uevent_env(&disk_to_dev(d->disk)->kobj, KOBJ_CHANGE, env);
-#endif
+	kfree(env[1]);
+	kfree(env[2]);
+
 	if (sysfs_create_link(&d->kobj, &disk_to_dev(d->disk)->kobj, "dev") ||
 	    sysfs_create_link(&disk_to_dev(d->disk)->kobj, &d->kobj, "bcache"))
 		pr_debug("error creating sysfs link");
@@ -960,6 +1002,7 @@ int bch_cached_dev_attach(struct cached_dev *dc, struct cache_set *c)
 	atomic_set(&dc->count, 1);
 
 	if (BDEV_STATE(&dc->sb) == BDEV_STATE_DIRTY) {
+		bch_sectors_dirty_init(dc);
 		atomic_set(&dc->has_dirty, 1);
 		atomic_inc(&dc->count);
 		bch_writeback_queue(dc);
@@ -1014,6 +1057,14 @@ static void cached_dev_flush(struct closure *cl)
 	struct cached_dev *dc = container_of(cl, struct cached_dev, disk.cl);
 	struct bcache_device *d = &dc->disk;
 
+	mutex_lock(&bch_register_lock);
+	d->flush_done = 1;
+
+	if (d->c)
+		bcache_device_unlink(d);
+
+	mutex_unlock(&bch_register_lock);
+
 	bch_cache_accounting_destroy(&dc->accounting);
 	kobject_del(&d->kobj);
 
@@ -1045,7 +1096,8 @@ static int cached_dev_init(struct cached_dev *dc, unsigned block_size)
 		hlist_add_head(&io->hash, dc->io_hash + RECENT_IO);
 	}
 
-	ret = bcache_device_init(&dc->disk, block_size);
+	ret = bcache_device_init(&dc->disk, block_size,
+			 dc->bdev->bd_part->nr_sects - dc->sb.data_offset);
 	if (ret)
 		return ret;
 
@@ -1144,11 +1196,10 @@ static int flash_dev_run(struct cache_set *c, struct uuid_entry *u)
 
 	kobject_init(&d->kobj, &bch_flash_dev_ktype);
 
-	if (bcache_device_init(d, block_bytes(c)))
+	if (bcache_device_init(d, block_bytes(c), u->sectors))
 		goto err;
 
 	bcache_device_attach(d, c, u - c->uuids);
-	set_capacity(d->disk, u->sectors);
 	bch_flash_dev_request_init(d);
 	add_disk(d->disk);
 
@@ -1255,9 +1306,10 @@ static void cache_set_free(struct closure *cl)
 	free_pages((unsigned long) c->uuids, ilog2(bucket_pages(c)));
 	free_pages((unsigned long) c->sort, ilog2(bucket_pages(c)));
 
-	kfree(c->fill_iter);
 	if (c->bio_split)
 		bioset_free(c->bio_split);
+	if (c->fill_iter)
+		mempool_destroy(c->fill_iter);
 	if (c->bio_meta)
 		mempool_destroy(c->bio_meta);
 	if (c->search)
@@ -1278,11 +1330,9 @@ static void cache_set_free(struct closure *cl)
 static void cache_set_flush(struct closure *cl)
 {
 	struct cache_set *c = container_of(cl, struct cache_set, caching);
+	struct cache *ca;
 	struct btree *b;
-
+	unsigned i;
-	/* Shut down allocator threads */
-	set_bit(CACHE_SET_STOPPING_2, &c->flags);
-	wake_up(&c->alloc_wait);
 
 	bch_cache_accounting_destroy(&c->accounting);
 
@@ -1295,7 +1345,11 @@ static void cache_set_flush(struct closure *cl)
 	/* Should skip this if we're unregistering because of an error */
 	list_for_each_entry(b, &c->btree_cache, list)
 		if (btree_node_dirty(b))
-			bch_btree_write(b, true, NULL);
+			bch_btree_node_write(b, NULL);
+
+	for_each_cache(ca, c, i)
+		if (ca->alloc_thread)
+			kthread_stop(ca->alloc_thread);
 
 	closure_return(cl);
 }
@@ -1303,18 +1357,22 @@ static void cache_set_flush(struct closure *cl)
 static void __cache_set_unregister(struct closure *cl)
 {
 	struct cache_set *c = container_of(cl, struct cache_set, caching);
-	struct cached_dev *dc, *t;
+	struct cached_dev *dc;
 	size_t i;
 
 	mutex_lock(&bch_register_lock);
 
-	if (test_bit(CACHE_SET_UNREGISTERING, &c->flags))
-		list_for_each_entry_safe(dc, t, &c->cached_devs, list)
-			bch_cached_dev_detach(dc);
-
 	for (i = 0; i < c->nr_uuids; i++)
-		if (c->devices[i] && UUID_FLASH_ONLY(&c->uuids[i]))
+		if (c->devices[i]) {
+			if (!UUID_FLASH_ONLY(&c->uuids[i]) &&
+			    test_bit(CACHE_SET_UNREGISTERING, &c->flags)) {
+				dc = container_of(c->devices[i],
+						  struct cached_dev, disk);
+				bch_cached_dev_detach(dc);
+			} else {
 				bcache_device_stop(c->devices[i]);
+			}
+		}
 
 	mutex_unlock(&bch_register_lock);
 
@@ -1373,9 +1431,9 @@ struct cache_set *bch_cache_set_alloc(struct cache_sb *sb)
 		c->btree_pages = max_t(int, c->btree_pages / 4,
 				       BTREE_MAX_PAGES);
 
-	init_waitqueue_head(&c->alloc_wait);
+	c->sort_crit_factor = int_sqrt(c->btree_pages);
+
 	mutex_init(&c->bucket_lock);
-	mutex_init(&c->fill_lock);
 	mutex_init(&c->sort_lock);
 	spin_lock_init(&c->sort_time_lock);
 	closure_init_unlocked(&c->sb_write);
@@ -1401,8 +1459,8 @@ struct cache_set *bch_cache_set_alloc(struct cache_sb *sb)
 	    !(c->bio_meta = mempool_create_kmalloc_pool(2,
 				sizeof(struct bbio) + sizeof(struct bio_vec) *
 				bucket_pages(c))) ||
+	    !(c->fill_iter = mempool_create_kmalloc_pool(1, iter_size)) ||
 	    !(c->bio_split = bioset_create(4, offsetof(struct bbio, bio))) ||
-	    !(c->fill_iter = kmalloc(iter_size, GFP_KERNEL)) ||
 	    !(c->sort = alloc_bucket_pages(GFP_KERNEL, c)) ||
 	    !(c->uuids = alloc_bucket_pages(GFP_KERNEL, c)) ||
 	    bch_journal_alloc(c) ||
@@ -1410,8 +1468,6 @@ struct cache_set *bch_cache_set_alloc(struct cache_sb *sb)
 	    bch_open_buckets_alloc(c))
 		goto err;
 
-	c->fill_iter->size = sb->bucket_size / sb->block_size;
-
 	c->congested_read_threshold_us	= 2000;
 	c->congested_write_threshold_us	= 20000;
 	c->error_limit	= 8 << IO_ERROR_SHIFT;
@@ -1496,9 +1552,10 @@ static void run_cache_set(struct cache_set *c)
 		 */
 		bch_journal_next(&c->journal);
 
+		err = "error starting allocator thread";
 		for_each_cache(ca, c, i)
-			closure_call(&ca->alloc, bch_allocator_thread,
+			if (bch_cache_allocator_start(ca))
-				     system_wq, &c->cl);
+				goto err;
 
 		/*
 		 * First place it's safe to allocate: btree_check() and
@@ -1531,17 +1588,16 @@ static void run_cache_set(struct cache_set *c)
 
 		bch_btree_gc_finish(c);
 
+		err = "error starting allocator thread";
 		for_each_cache(ca, c, i)
-			closure_call(&ca->alloc, bch_allocator_thread,
+			if (bch_cache_allocator_start(ca))
-				     ca->alloc_workqueue, &c->cl);
+				goto err;
 
 		mutex_lock(&c->bucket_lock);
 		for_each_cache(ca, c, i)
 			bch_prio_write(ca);
 		mutex_unlock(&c->bucket_lock);
 
-		wake_up(&c->alloc_wait);
-
 		err = "cannot allocate new UUID bucket";
 		if (__uuid_write(c))
 			goto err_unlock_gc;
@@ -1552,7 +1608,7 @@ static void run_cache_set(struct cache_set *c)
 			goto err_unlock_gc;
 
 		bkey_copy_key(&c->root->key, &MAX_KEY);
-		bch_btree_write(c->root, true, &op);
+		bch_btree_node_write(c->root, &op.cl);
 
 		bch_btree_set_root(c->root);
 		rw_unlock(true, c->root);
@@ -1673,9 +1729,6 @@ void bch_cache_release(struct kobject *kobj)
 
 	bio_split_pool_free(&ca->bio_split_hook);
 
-	if (ca->alloc_workqueue)
-		destroy_workqueue(ca->alloc_workqueue);
-
 	free_pages((unsigned long) ca->disk_buckets, ilog2(bucket_pages(ca)));
 	kfree(ca->prio_buckets);
 	vfree(ca->buckets);
@@ -1723,7 +1776,6 @@ static int cache_alloc(struct cache_sb *sb, struct cache *ca)
 	    !(ca->prio_buckets	= kzalloc(sizeof(uint64_t) * prio_buckets(ca) *
 					  2, GFP_KERNEL)) ||
 	    !(ca->disk_buckets	= alloc_bucket_pages(GFP_KERNEL, ca)) ||
-	    !(ca->alloc_workqueue = alloc_workqueue("bch_allocator", 0, 1)) ||
 	    bio_split_pool_init(&ca->bio_split_hook))
 		return -ENOMEM;
 
@@ -1786,6 +1838,36 @@ static ssize_t register_bcache(struct kobject *, struct kobj_attribute *,
 kobj_attribute_write(register,		register_bcache);
 kobj_attribute_write(register_quiet,	register_bcache);
 
+static bool bch_is_open_backing(struct block_device *bdev) {
+	struct cache_set *c, *tc;
+	struct cached_dev *dc, *t;
+
+	list_for_each_entry_safe(c, tc, &bch_cache_sets, list)
+		list_for_each_entry_safe(dc, t, &c->cached_devs, list)
+			if (dc->bdev == bdev)
+				return true;
+	list_for_each_entry_safe(dc, t, &uncached_devices, list)
+		if (dc->bdev == bdev)
+			return true;
+	return false;
+}
+
+static bool bch_is_open_cache(struct block_device *bdev) {
+	struct cache_set *c, *tc;
+	struct cache *ca;
+	unsigned i;
+
+	list_for_each_entry_safe(c, tc, &bch_cache_sets, list)
+		for_each_cache(ca, c, i)
+			if (ca->bdev == bdev)
+				return true;
+	return false;
+}
+
+static bool bch_is_open(struct block_device *bdev) {
+	return bch_is_open_cache(bdev) || bch_is_open_backing(bdev);
+}
+
 static ssize_t register_bcache(struct kobject *k, struct kobj_attribute *attr,
 			       const char *buffer, size_t size)
 {
@@ -1810,8 +1892,13 @@ static ssize_t register_bcache(struct kobject *k, struct kobj_attribute *attr,
 				  FMODE_READ|FMODE_WRITE|FMODE_EXCL,
 				  sb);
 	if (IS_ERR(bdev)) {
-		if (bdev == ERR_PTR(-EBUSY))
+		if (bdev == ERR_PTR(-EBUSY)) {
+			bdev = lookup_bdev(strim(path));
+			if (!IS_ERR(bdev) && bch_is_open(bdev))
+				err = "device already registered";
+			else
 				err = "device busy";
+		}
 		goto err;
 	}
 

drivers/md/bcache/sysfs.c

@@ -9,7 +9,9 @@
 #include "sysfs.h"
 #include "btree.h"
 #include "request.h"
+#include "writeback.h"
 
+#include <linux/blkdev.h>
 #include <linux/sort.h>
 
 static const char * const cache_replacement_policies[] = {
@@ -79,6 +81,9 @@ rw_attribute(writeback_rate_p_term_inverse);
 rw_attribute(writeback_rate_d_smooth);
 read_attribute(writeback_rate_debug);
 
+read_attribute(stripe_size);
+read_attribute(partial_stripes_expensive);
+
 rw_attribute(synchronous);
 rw_attribute(journal_delay_ms);
 rw_attribute(discard);
@@ -127,7 +132,7 @@ SHOW(__bch_cached_dev)
 		char derivative[20];
 		char target[20];
 		bch_hprint(dirty,
-		       atomic_long_read(&dc->disk.sectors_dirty) << 9);
+			   bcache_dev_sectors_dirty(&dc->disk) << 9);
 		bch_hprint(derivative,	dc->writeback_rate_derivative << 9);
 		bch_hprint(target,	dc->writeback_rate_target << 9);
 
@@ -143,7 +148,10 @@ SHOW(__bch_cached_dev)
 	}
 
 	sysfs_hprint(dirty_data,
-		     atomic_long_read(&dc->disk.sectors_dirty) << 9);
+		     bcache_dev_sectors_dirty(&dc->disk) << 9);
+
+	sysfs_hprint(stripe_size,	(1 << dc->disk.stripe_size_bits) << 9);
+	var_printf(partial_stripes_expensive,	"%u");
 
 	var_printf(sequential_merge,	"%i");
 	var_hprint(sequential_cutoff);
@@ -170,6 +178,7 @@ STORE(__cached_dev)
 					     disk.kobj);
 	unsigned v = size;
 	struct cache_set *c;
+	struct kobj_uevent_env *env;
 
 #define d_strtoul(var)		sysfs_strtoul(var, dc->var)
 #define d_strtoi_h(var)		sysfs_hatoi(var, dc->var)
@@ -214,6 +223,7 @@ STORE(__cached_dev)
 	}
 
 	if (attr == &sysfs_label) {
+		/* note: endlines are preserved */
 		memcpy(dc->sb.label, buf, SB_LABEL_SIZE);
 		bch_write_bdev_super(dc, NULL);
 		if (dc->disk.c) {
@@ -221,6 +231,15 @@ STORE(__cached_dev)
 			       buf, SB_LABEL_SIZE);
 			bch_uuid_write(dc->disk.c);
 		}
+		env = kzalloc(sizeof(struct kobj_uevent_env), GFP_KERNEL);
+		if (!env)
+			return -ENOMEM;
+		add_uevent_var(env, "DRIVER=bcache");
+		add_uevent_var(env, "CACHED_UUID=%pU", dc->sb.uuid),
+		add_uevent_var(env, "CACHED_LABEL=%s", buf);
+		kobject_uevent_env(
+			&disk_to_dev(dc->disk.disk)->kobj, KOBJ_CHANGE, env->envp);
+		kfree(env);
 	}
 
 	if (attr == &sysfs_attach) {
@@ -284,6 +303,8 @@ static struct attribute *bch_cached_dev_files[] = {
 	&sysfs_writeback_rate_d_smooth,
 	&sysfs_writeback_rate_debug,
 	&sysfs_dirty_data,
+	&sysfs_stripe_size,
+	&sysfs_partial_stripes_expensive,
 	&sysfs_sequential_cutoff,
 	&sysfs_sequential_merge,
 	&sysfs_clear_stats,
@@ -665,12 +686,10 @@ SHOW(__bch_cache)
 		int cmp(const void *l, const void *r)
 		{	return *((uint16_t *) r) - *((uint16_t *) l); }
 
-		/* Number of quantiles we compute */
-		const unsigned nq = 31;
-
 		size_t n = ca->sb.nbuckets, i, unused, btree;
 		uint64_t sum = 0;
-		uint16_t q[nq], *p, *cached;
+		/* Compute 31 quantiles */
+		uint16_t q[31], *p, *cached;
 		ssize_t ret;
 
 		cached = p = vmalloc(ca->sb.nbuckets * sizeof(uint16_t));
@@ -703,12 +722,13 @@ SHOW(__bch_cache)
 		if (n)
 			do_div(sum, n);
 
-		for (i = 0; i < nq; i++)
+		for (i = 0; i < ARRAY_SIZE(q); i++)
-			q[i] = INITIAL_PRIO - cached[n * (i + 1) / (nq + 1)];
+			q[i] = INITIAL_PRIO - cached[n * (i + 1) /
+				(ARRAY_SIZE(q) + 1)];
 
 		vfree(p);
 
-		ret = snprintf(buf, PAGE_SIZE,
+		ret = scnprintf(buf, PAGE_SIZE,
 				"Unused:		%zu%%\n"
 				"Metadata:	%zu%%\n"
 				"Average:	%llu\n"
@@ -716,13 +736,15 @@ SHOW(__bch_cache)
 				"Quantiles:	[",
 				unused * 100 / (size_t) ca->sb.nbuckets,
 				btree * 100 / (size_t) ca->sb.nbuckets, sum,
-			       n * ca->sb.bucket_size / (nq + 1));
+				n * ca->sb.bucket_size / (ARRAY_SIZE(q) + 1));
 
-		for (i = 0; i < nq && ret < (ssize_t) PAGE_SIZE; i++)
+		for (i = 0; i < ARRAY_SIZE(q); i++)
-			ret += snprintf(buf + ret, PAGE_SIZE - ret,
+			ret += scnprintf(buf + ret, PAGE_SIZE - ret,
-					i < nq - 1 ? "%u " : "%u]\n", q[i]);
+					 "%u ", q[i]);
+		ret--;
+
+		ret += scnprintf(buf + ret, PAGE_SIZE - ret, "]\n");
 
-		buf[PAGE_SIZE - 1] = '\0';
 		return ret;
 	}
 

drivers/md/bcache/trace.c

@@ -2,6 +2,7 @@
 #include "btree.h"
 #include "request.h"
 
+#include <linux/blktrace_api.h>
 #include <linux/module.h>
 
 #define CREATE_TRACE_POINTS
@@ -9,18 +10,44 @@
 
 EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_request_start);
 EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_request_end);
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_passthrough);
+
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_cache_hit);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_bypass_sequential);
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_cache_miss);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_bypass_congested);
+
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_read);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_write);
 EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_read_retry);
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_writethrough);
+
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_writeback);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_cache_insert);
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_write_skip);
+
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_journal_replay_key);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_journal_write);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_journal_full);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_journal_entry_full);
+
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_cache_cannibalize);
+
 EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_read);
 EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_write);
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_write_dirty);
+
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_read_dirty);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_node_alloc);
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_journal_write);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_node_alloc_fail);
-EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_cache_insert);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_node_free);
+
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_gc_coalesce);
 EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_gc_start);
 EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_gc_end);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_gc_copy);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_gc_copy_collision);
+
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_insert_key);
+
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_node_split);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_node_compact);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_btree_set_root);
+
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_alloc_invalidate);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_alloc_fail);
+
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_writeback);
+EXPORT_TRACEPOINT_SYMBOL_GPL(bcache_writeback_collision);

drivers/md/bcache/util.c

@@ -228,23 +228,6 @@ start:		bv->bv_len	= min_t(size_t, PAGE_SIZE - bv->bv_offset,
 	}
 }
 
-int bch_bio_alloc_pages(struct bio *bio, gfp_t gfp)
-{
-	int i;
-	struct bio_vec *bv;
-
-	bio_for_each_segment(bv, bio, i) {
-		bv->bv_page = alloc_page(gfp);
-		if (!bv->bv_page) {
-			while (bv-- != bio->bi_io_vec + bio->bi_idx)
-				__free_page(bv->bv_page);
-			return -ENOMEM;
-		}
-	}
-
-	return 0;
-}
-
 /*
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group (Any
  * use permitted, subject to terms of PostgreSQL license; see.)

drivers/md/bcache/util.h

@@ -15,8 +15,6 @@
 
 struct closure;
 
-#include <trace/events/bcache.h>
-
 #ifdef CONFIG_BCACHE_EDEBUG
 
 #define atomic_dec_bug(v)	BUG_ON(atomic_dec_return(v) < 0)
@@ -566,12 +564,8 @@ static inline unsigned fract_exp_two(unsigned x, unsigned fract_bits)
 	return x;
 }
 
-#define bio_end(bio)	((bio)->bi_sector + bio_sectors(bio))
-
 void bch_bio_map(struct bio *bio, void *base);
 
-int bch_bio_alloc_pages(struct bio *bio, gfp_t gfp);
-
 static inline sector_t bdev_sectors(struct block_device *bdev)
 {
 	return bdev->bd_inode->i_size >> 9;

drivers/md/bcache/writeback.c

@@ -9,6 +9,9 @@
 #include "bcache.h"
 #include "btree.h"
 #include "debug.h"
+#include "writeback.h"
+
+#include <trace/events/bcache.h>
 
 static struct workqueue_struct *dirty_wq;
 
@@ -36,7 +39,7 @@ static void __update_writeback_rate(struct cached_dev *dc)
 
 	int change = 0;
 	int64_t error;
-	int64_t dirty = atomic_long_read(&dc->disk.sectors_dirty);
+	int64_t dirty = bcache_dev_sectors_dirty(&dc->disk);
 	int64_t derivative = dirty - dc->disk.sectors_dirty_last;
 
 	dc->disk.sectors_dirty_last = dirty;
@@ -105,6 +108,31 @@ static bool dirty_pred(struct keybuf *buf, struct bkey *k)
 	return KEY_DIRTY(k);
 }
 
+static bool dirty_full_stripe_pred(struct keybuf *buf, struct bkey *k)
+{
+	uint64_t stripe;
+	unsigned nr_sectors = KEY_SIZE(k);
+	struct cached_dev *dc = container_of(buf, struct cached_dev,
+					     writeback_keys);
+	unsigned stripe_size = 1 << dc->disk.stripe_size_bits;
+
+	if (!KEY_DIRTY(k))
+		return false;
+
+	stripe = KEY_START(k) >> dc->disk.stripe_size_bits;
+	while (1) {
+		if (atomic_read(dc->disk.stripe_sectors_dirty + stripe) !=
+		    stripe_size)
+			return false;
+
+		if (nr_sectors <= stripe_size)
+			return true;
+
+		nr_sectors -= stripe_size;
+		stripe++;
+	}
+}
+
 static void dirty_init(struct keybuf_key *w)
 {
 	struct dirty_io *io = w->private;
@@ -149,7 +177,22 @@ static void refill_dirty(struct closure *cl)
 		searched_from_start = true;
 	}
 
-	bch_refill_keybuf(dc->disk.c, buf, &end);
+	if (dc->partial_stripes_expensive) {
+		uint64_t i;
+
+		for (i = 0; i < dc->disk.nr_stripes; i++)
+			if (atomic_read(dc->disk.stripe_sectors_dirty + i) ==
+			    1 << dc->disk.stripe_size_bits)
+				goto full_stripes;
+
+		goto normal_refill;
+full_stripes:
+		bch_refill_keybuf(dc->disk.c, buf, &end,
+				  dirty_full_stripe_pred);
+	} else {
+normal_refill:
+		bch_refill_keybuf(dc->disk.c, buf, &end, dirty_pred);
+	}
 
 	if (bkey_cmp(&buf->last_scanned, &end) >= 0 && searched_from_start) {
 		/* Searched the entire btree  - delay awhile */
@@ -181,10 +224,8 @@ void bch_writeback_queue(struct cached_dev *dc)
 	}
 }
 
-void bch_writeback_add(struct cached_dev *dc, unsigned sectors)
+void bch_writeback_add(struct cached_dev *dc)
 {
-	atomic_long_add(sectors, &dc->disk.sectors_dirty);
-
 	if (!atomic_read(&dc->has_dirty) &&
 	    !atomic_xchg(&dc->has_dirty, 1)) {
 		atomic_inc(&dc->count);
@@ -203,6 +244,34 @@ void bch_writeback_add(struct cached_dev *dc, unsigned sectors)
 	}
 }
 
+void bcache_dev_sectors_dirty_add(struct cache_set *c, unsigned inode,
+				  uint64_t offset, int nr_sectors)
+{
+	struct bcache_device *d = c->devices[inode];
+	unsigned stripe_size, stripe_offset;
+	uint64_t stripe;
+
+	if (!d)
+		return;
+
+	stripe_size = 1 << d->stripe_size_bits;
+	stripe = offset >> d->stripe_size_bits;
+	stripe_offset = offset & (stripe_size - 1);
+
+	while (nr_sectors) {
+		int s = min_t(unsigned, abs(nr_sectors),
+			      stripe_size - stripe_offset);
+
+		if (nr_sectors < 0)
+			s = -s;
+
+		atomic_add(s, d->stripe_sectors_dirty + stripe);
+		nr_sectors -= s;
+		stripe_offset = 0;
+		stripe++;
+	}
+}
+
 /* Background writeback - IO loop */
 
 static void dirty_io_destructor(struct closure *cl)
@@ -216,9 +285,10 @@ static void write_dirty_finish(struct closure *cl)
 	struct dirty_io *io = container_of(cl, struct dirty_io, cl);
 	struct keybuf_key *w = io->bio.bi_private;
 	struct cached_dev *dc = io->dc;
-	struct bio_vec *bv = bio_iovec_idx(&io->bio, io->bio.bi_vcnt);
+	struct bio_vec *bv;
+	int i;
 
-	while (bv-- != io->bio.bi_io_vec)
+	bio_for_each_segment_all(bv, &io->bio, i)
 		__free_page(bv->bv_page);
 
 	/* This is kind of a dumb way of signalling errors. */
@@ -236,10 +306,12 @@ static void write_dirty_finish(struct closure *cl)
 		for (i = 0; i < KEY_PTRS(&w->key); i++)
 			atomic_inc(&PTR_BUCKET(dc->disk.c, &w->key, i)->pin);
 
-		pr_debug("clearing %s", pkey(&w->key));
 		bch_btree_insert(&op, dc->disk.c);
 		closure_sync(&op.cl);
 
+		if (op.insert_collision)
+			trace_bcache_writeback_collision(&w->key);
+
 		atomic_long_inc(op.insert_collision
 				? &dc->disk.c->writeback_keys_failed
 				: &dc->disk.c->writeback_keys_done);
@@ -275,7 +347,6 @@ static void write_dirty(struct closure *cl)
 	io->bio.bi_bdev		= io->dc->bdev;
 	io->bio.bi_end_io	= dirty_endio;
 
-	trace_bcache_write_dirty(&io->bio);
 	closure_bio_submit(&io->bio, cl, &io->dc->disk);
 
 	continue_at(cl, write_dirty_finish, dirty_wq);
@@ -296,7 +367,6 @@ static void read_dirty_submit(struct closure *cl)
 {
 	struct dirty_io *io = container_of(cl, struct dirty_io, cl);
 
-	trace_bcache_read_dirty(&io->bio);
 	closure_bio_submit(&io->bio, cl, &io->dc->disk);
 
 	continue_at(cl, write_dirty, dirty_wq);
@@ -349,10 +419,10 @@ static void read_dirty(struct closure *cl)
 		io->bio.bi_rw		= READ;
 		io->bio.bi_end_io	= read_dirty_endio;
 
-		if (bch_bio_alloc_pages(&io->bio, GFP_KERNEL))
+		if (bio_alloc_pages(&io->bio, GFP_KERNEL))
 			goto err_free;
 
-		pr_debug("%s", pkey(&w->key));
+		trace_bcache_writeback(&w->key);
 
 		closure_call(&io->cl, read_dirty_submit, NULL, &dc->disk.cl);
 
@@ -375,12 +445,49 @@ err:
 	refill_dirty(cl);
 }
 
+/* Init */
+
+static int bch_btree_sectors_dirty_init(struct btree *b, struct btree_op *op,
+					struct cached_dev *dc)
+{
+	struct bkey *k;
+	struct btree_iter iter;
+
+	bch_btree_iter_init(b, &iter, &KEY(dc->disk.id, 0, 0));
+	while ((k = bch_btree_iter_next_filter(&iter, b, bch_ptr_bad)))
+		if (!b->level) {
+			if (KEY_INODE(k) > dc->disk.id)
+				break;
+
+			if (KEY_DIRTY(k))
+				bcache_dev_sectors_dirty_add(b->c, dc->disk.id,
+							     KEY_START(k),
+							     KEY_SIZE(k));
+		} else {
+			btree(sectors_dirty_init, k, b, op, dc);
+			if (KEY_INODE(k) > dc->disk.id)
+				break;
+
+			cond_resched();
+		}
+
+	return 0;
+}
+
+void bch_sectors_dirty_init(struct cached_dev *dc)
+{
+	struct btree_op op;
+
+	bch_btree_op_init_stack(&op);
+	btree_root(sectors_dirty_init, dc->disk.c, &op, dc);
+}
+
 void bch_cached_dev_writeback_init(struct cached_dev *dc)
 {
 	closure_init_unlocked(&dc->writeback);
 	init_rwsem(&dc->writeback_lock);
 
-	bch_keybuf_init(&dc->writeback_keys, dirty_pred);
+	bch_keybuf_init(&dc->writeback_keys);
 
 	dc->writeback_metadata		= true;
 	dc->writeback_running		= true;

drivers/md/bcache/writeback.h

@@ -0,0 +1,64 @@
+#ifndef _BCACHE_WRITEBACK_H
+#define _BCACHE_WRITEBACK_H
+
+#define CUTOFF_WRITEBACK	40
+#define CUTOFF_WRITEBACK_SYNC	70
+
+static inline uint64_t bcache_dev_sectors_dirty(struct bcache_device *d)
+{
+	uint64_t i, ret = 0;
+
+	for (i = 0; i < d->nr_stripes; i++)
+		ret += atomic_read(d->stripe_sectors_dirty + i);
+
+	return ret;
+}
+
+static inline bool bcache_dev_stripe_dirty(struct bcache_device *d,
+					   uint64_t offset,
+					   unsigned nr_sectors)
+{
+	uint64_t stripe = offset >> d->stripe_size_bits;
+
+	while (1) {
+		if (atomic_read(d->stripe_sectors_dirty + stripe))
+			return true;
+
+		if (nr_sectors <= 1 << d->stripe_size_bits)
+			return false;
+
+		nr_sectors -= 1 << d->stripe_size_bits;
+		stripe++;
+	}
+}
+
+static inline bool should_writeback(struct cached_dev *dc, struct bio *bio,
+				    unsigned cache_mode, bool would_skip)
+{
+	unsigned in_use = dc->disk.c->gc_stats.in_use;
+
+	if (cache_mode != CACHE_MODE_WRITEBACK ||
+	    atomic_read(&dc->disk.detaching) ||
+	    in_use > CUTOFF_WRITEBACK_SYNC)
+		return false;
+
+	if (dc->partial_stripes_expensive &&
+	    bcache_dev_stripe_dirty(&dc->disk, bio->bi_sector,
+				    bio_sectors(bio)))
+		return true;
+
+	if (would_skip)
+		return false;
+
+	return bio->bi_rw & REQ_SYNC ||
+		in_use <= CUTOFF_WRITEBACK;
+}
+
+void bcache_dev_sectors_dirty_add(struct cache_set *, unsigned, uint64_t, int);
+void bch_writeback_queue(struct cached_dev *);
+void bch_writeback_add(struct cached_dev *);
+
+void bch_sectors_dirty_init(struct cached_dev *dc);
+void bch_cached_dev_writeback_init(struct cached_dev *);
+
+#endif

include/linux/drbd.h

@@ -177,7 +177,11 @@ enum drbd_ret_code {
 	ERR_NEED_APV_100	= 163,
 	ERR_NEED_ALLOW_TWO_PRI  = 164,
 	ERR_MD_UNCLEAN          = 165,
-
+	ERR_MD_LAYOUT_CONNECTED = 166,
+	ERR_MD_LAYOUT_TOO_BIG   = 167,
+	ERR_MD_LAYOUT_TOO_SMALL = 168,
+	ERR_MD_LAYOUT_NO_FIT    = 169,
+	ERR_IMPLICIT_SHRINK     = 170,
 	/* insert new ones above this line */
 	AFTER_LAST_ERR_CODE
 };

include/linux/drbd_genl.h

@@ -181,6 +181,8 @@ GENL_struct(DRBD_NLA_RESIZE_PARMS, 7, resize_parms,
 	__u64_field(1, DRBD_GENLA_F_MANDATORY,	resize_size)
 	__flg_field(2, DRBD_GENLA_F_MANDATORY,	resize_force)
 	__flg_field(3, DRBD_GENLA_F_MANDATORY,	no_resync)
+	__u32_field_def(4, 0 /* OPTIONAL */, al_stripes, DRBD_AL_STRIPES_DEF)
+	__u32_field_def(5, 0 /* OPTIONAL */, al_stripe_size, DRBD_AL_STRIPE_SIZE_DEF)
 )
 
 GENL_struct(DRBD_NLA_STATE_INFO, 8, state_info,

include/linux/drbd_limits.h

@@ -215,4 +215,13 @@
 #define DRBD_ALWAYS_ASBP_DEF	0
 #define DRBD_USE_RLE_DEF	1
 
+#define DRBD_AL_STRIPES_MIN     1
+#define DRBD_AL_STRIPES_MAX     1024
+#define DRBD_AL_STRIPES_DEF     1
+#define DRBD_AL_STRIPES_SCALE   '1'
+
+#define DRBD_AL_STRIPE_SIZE_MIN   4
+#define DRBD_AL_STRIPE_SIZE_MAX   16777216
+#define DRBD_AL_STRIPE_SIZE_DEF   32
+#define DRBD_AL_STRIPE_SIZE_SCALE 'k' /* kilobytes */
 #endif

include/trace/events/bcache.h

@@ -9,9 +9,7 @@
 struct search;
 
 DECLARE_EVENT_CLASS(bcache_request,
-
 	TP_PROTO(struct search *s, struct bio *bio),
-
 	TP_ARGS(s, bio),
 
 	TP_STRUCT__entry(
@@ -22,7 +20,6 @@ DECLARE_EVENT_CLASS(bcache_request,
 		__field(dev_t,		orig_sector		)
 		__field(unsigned int,	nr_sector		)
 		__array(char,		rwbs,	6		)
-		__array(char,		comm,	TASK_COMM_LEN	)
 	),
 
 	TP_fast_assign(
@@ -33,36 +30,66 @@ DECLARE_EVENT_CLASS(bcache_request,
 		__entry->orig_sector	= bio->bi_sector - 16;
 		__entry->nr_sector	= bio->bi_size >> 9;
 		blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size);
-		memcpy(__entry->comm, current->comm, TASK_COMM_LEN);
 	),
 
-	TP_printk("%d,%d %s %llu + %u [%s] (from %d,%d @ %llu)",
+	TP_printk("%d,%d %s %llu + %u (from %d,%d @ %llu)",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
-		  __entry->rwbs,
+		  __entry->rwbs, (unsigned long long)__entry->sector,
-		  (unsigned long long)__entry->sector,
+		  __entry->nr_sector, __entry->orig_major, __entry->orig_minor,
-		  __entry->nr_sector, __entry->comm,
-		  __entry->orig_major, __entry->orig_minor,
 		  (unsigned long long)__entry->orig_sector)
 );
 
+DECLARE_EVENT_CLASS(bkey,
+	TP_PROTO(struct bkey *k),
+	TP_ARGS(k),
+
+	TP_STRUCT__entry(
+		__field(u32,	size				)
+		__field(u32,	inode				)
+		__field(u64,	offset				)
+		__field(bool,	dirty				)
+	),
+
+	TP_fast_assign(
+		__entry->inode	= KEY_INODE(k);
+		__entry->offset	= KEY_OFFSET(k);
+		__entry->size	= KEY_SIZE(k);
+		__entry->dirty	= KEY_DIRTY(k);
+	),
+
+	TP_printk("%u:%llu len %u dirty %u", __entry->inode,
+		  __entry->offset, __entry->size, __entry->dirty)
+);
+
+DECLARE_EVENT_CLASS(btree_node,
+	TP_PROTO(struct btree *b),
+	TP_ARGS(b),
+
+	TP_STRUCT__entry(
+		__field(size_t,		bucket			)
+	),
+
+	TP_fast_assign(
+		__entry->bucket	= PTR_BUCKET_NR(b->c, &b->key, 0);
+	),
+
+	TP_printk("bucket %zu", __entry->bucket)
+);
+
+/* request.c */
+
 DEFINE_EVENT(bcache_request, bcache_request_start,
-
 	TP_PROTO(struct search *s, struct bio *bio),
-
 	TP_ARGS(s, bio)
 );
 
 DEFINE_EVENT(bcache_request, bcache_request_end,
-
 	TP_PROTO(struct search *s, struct bio *bio),
-
 	TP_ARGS(s, bio)
 );
 
 DECLARE_EVENT_CLASS(bcache_bio,
-
 	TP_PROTO(struct bio *bio),
-
 	TP_ARGS(bio),
 
 	TP_STRUCT__entry(
@@ -70,7 +97,6 @@ DECLARE_EVENT_CLASS(bcache_bio,
 		__field(sector_t,	sector			)
 		__field(unsigned int,	nr_sector		)
 		__array(char,		rwbs,	6		)
-		__array(char,		comm,	TASK_COMM_LEN	)
 	),
 
 	TP_fast_assign(
@@ -78,191 +104,328 @@ DECLARE_EVENT_CLASS(bcache_bio,
 		__entry->sector		= bio->bi_sector;
 		__entry->nr_sector	= bio->bi_size >> 9;
 		blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size);
-		memcpy(__entry->comm, current->comm, TASK_COMM_LEN);
 	),
 
-	TP_printk("%d,%d  %s %llu + %u [%s]",
+	TP_printk("%d,%d  %s %llu + %u",
-		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs,
-		  __entry->rwbs,
+		  (unsigned long long)__entry->sector, __entry->nr_sector)
-		  (unsigned long long)__entry->sector,
-		  __entry->nr_sector, __entry->comm)
 );
 
-
+DEFINE_EVENT(bcache_bio, bcache_bypass_sequential,
-DEFINE_EVENT(bcache_bio, bcache_passthrough,
-
 	TP_PROTO(struct bio *bio),
-
 	TP_ARGS(bio)
 );
 
-DEFINE_EVENT(bcache_bio, bcache_cache_hit,
+DEFINE_EVENT(bcache_bio, bcache_bypass_congested,
-
 	TP_PROTO(struct bio *bio),
-
 	TP_ARGS(bio)
 );
 
-DEFINE_EVENT(bcache_bio, bcache_cache_miss,
+TRACE_EVENT(bcache_read,
-
+	TP_PROTO(struct bio *bio, bool hit, bool bypass),
-	TP_PROTO(struct bio *bio),
+	TP_ARGS(bio, hit, bypass),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_read_retry,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_writethrough,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_writeback,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_write_skip,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_btree_read,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_btree_write,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_write_dirty,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_read_dirty,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_write_moving,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_read_moving,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DEFINE_EVENT(bcache_bio, bcache_journal_write,
-
-	TP_PROTO(struct bio *bio),
-
-	TP_ARGS(bio)
-);
-
-DECLARE_EVENT_CLASS(bcache_cache_bio,
-
-	TP_PROTO(struct bio *bio,
-		 sector_t orig_sector,
-		 struct block_device* orig_bdev),
-
-	TP_ARGS(bio, orig_sector, orig_bdev),
 
 	TP_STRUCT__entry(
 		__field(dev_t,		dev			)
-		__field(dev_t,		orig_dev		)
 		__field(sector_t,	sector			)
-		__field(sector_t,	orig_sector		)
 		__field(unsigned int,	nr_sector		)
 		__array(char,		rwbs,	6		)
-		__array(char,		comm,	TASK_COMM_LEN	)
+		__field(bool,		cache_hit		)
+		__field(bool,		bypass			)
 	),
 
 	TP_fast_assign(
 		__entry->dev		= bio->bi_bdev->bd_dev;
-		__entry->orig_dev	= orig_bdev->bd_dev;
 		__entry->sector		= bio->bi_sector;
-		__entry->orig_sector	= orig_sector;
 		__entry->nr_sector	= bio->bi_size >> 9;
 		blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size);
-		memcpy(__entry->comm, current->comm, TASK_COMM_LEN);
+		__entry->cache_hit = hit;
+		__entry->bypass = bypass;
 	),
 
-	TP_printk("%d,%d  %s %llu + %u [%s] (from %d,%d %llu)",
+	TP_printk("%d,%d  %s %llu + %u hit %u bypass %u",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
-		  __entry->rwbs,
+		  __entry->rwbs, (unsigned long long)__entry->sector,
-		  (unsigned long long)__entry->sector,
+		  __entry->nr_sector, __entry->cache_hit, __entry->bypass)
-		  __entry->nr_sector, __entry->comm,
-		  MAJOR(__entry->orig_dev), MINOR(__entry->orig_dev),
-		  (unsigned long long)__entry->orig_sector)
 );
 
-DEFINE_EVENT(bcache_cache_bio, bcache_cache_insert,
+TRACE_EVENT(bcache_write,
-
+	TP_PROTO(struct bio *bio, bool writeback, bool bypass),
-	TP_PROTO(struct bio *bio,
+	TP_ARGS(bio, writeback, bypass),
-		 sector_t orig_sector,
-		 struct block_device *orig_bdev),
-
-	TP_ARGS(bio, orig_sector, orig_bdev)
-);
-
-DECLARE_EVENT_CLASS(bcache_gc,
-
-	TP_PROTO(uint8_t *uuid),
-
-	TP_ARGS(uuid),
 
 	TP_STRUCT__entry(
-		__field(uint8_t *,	uuid)
+		__field(dev_t,		dev			)
+		__field(sector_t,	sector			)
+		__field(unsigned int,	nr_sector		)
+		__array(char,		rwbs,	6		)
+		__field(bool,		writeback		)
+		__field(bool,		bypass			)
 	),
 
 	TP_fast_assign(
-		__entry->uuid		= uuid;
+		__entry->dev		= bio->bi_bdev->bd_dev;
+		__entry->sector		= bio->bi_sector;
+		__entry->nr_sector	= bio->bi_size >> 9;
+		blk_fill_rwbs(__entry->rwbs, bio->bi_rw, bio->bi_size);
+		__entry->writeback = writeback;
+		__entry->bypass = bypass;
+	),
+
+	TP_printk("%d,%d  %s %llu + %u hit %u bypass %u",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->rwbs, (unsigned long long)__entry->sector,
+		  __entry->nr_sector, __entry->writeback, __entry->bypass)
+);
+
+DEFINE_EVENT(bcache_bio, bcache_read_retry,
+	TP_PROTO(struct bio *bio),
+	TP_ARGS(bio)
+);
+
+DEFINE_EVENT(bkey, bcache_cache_insert,
+	TP_PROTO(struct bkey *k),
+	TP_ARGS(k)
+);
+
+/* Journal */
+
+DECLARE_EVENT_CLASS(cache_set,
+	TP_PROTO(struct cache_set *c),
+	TP_ARGS(c),
+
+	TP_STRUCT__entry(
+		__array(char,		uuid,	16 )
+	),
+
+	TP_fast_assign(
+		memcpy(__entry->uuid, c->sb.set_uuid, 16);
 	),
 
 	TP_printk("%pU", __entry->uuid)
 );
 
-
+DEFINE_EVENT(bkey, bcache_journal_replay_key,
-DEFINE_EVENT(bcache_gc, bcache_gc_start,
+	TP_PROTO(struct bkey *k),
-
+	TP_ARGS(k)
-	     TP_PROTO(uint8_t *uuid),
-
-	     TP_ARGS(uuid)
 );
 
-DEFINE_EVENT(bcache_gc, bcache_gc_end,
+DEFINE_EVENT(cache_set, bcache_journal_full,
+	TP_PROTO(struct cache_set *c),
+	TP_ARGS(c)
+);
 
-	     TP_PROTO(uint8_t *uuid),
+DEFINE_EVENT(cache_set, bcache_journal_entry_full,
+	TP_PROTO(struct cache_set *c),
+	TP_ARGS(c)
+);
 
-	     TP_ARGS(uuid)
+DEFINE_EVENT(bcache_bio, bcache_journal_write,
+	TP_PROTO(struct bio *bio),
+	TP_ARGS(bio)
+);
+
+/* Btree */
+
+DEFINE_EVENT(cache_set, bcache_btree_cache_cannibalize,
+	TP_PROTO(struct cache_set *c),
+	TP_ARGS(c)
+);
+
+DEFINE_EVENT(btree_node, bcache_btree_read,
+	TP_PROTO(struct btree *b),
+	TP_ARGS(b)
+);
+
+TRACE_EVENT(bcache_btree_write,
+	TP_PROTO(struct btree *b),
+	TP_ARGS(b),
+
+	TP_STRUCT__entry(
+		__field(size_t,		bucket			)
+		__field(unsigned,	block			)
+		__field(unsigned,	keys			)
+	),
+
+	TP_fast_assign(
+		__entry->bucket	= PTR_BUCKET_NR(b->c, &b->key, 0);
+		__entry->block	= b->written;
+		__entry->keys	= b->sets[b->nsets].data->keys;
+	),
+
+	TP_printk("bucket %zu", __entry->bucket)
+);
+
+DEFINE_EVENT(btree_node, bcache_btree_node_alloc,
+	TP_PROTO(struct btree *b),
+	TP_ARGS(b)
+);
+
+DEFINE_EVENT(btree_node, bcache_btree_node_alloc_fail,
+	TP_PROTO(struct btree *b),
+	TP_ARGS(b)
+);
+
+DEFINE_EVENT(btree_node, bcache_btree_node_free,
+	TP_PROTO(struct btree *b),
+	TP_ARGS(b)
+);
+
+TRACE_EVENT(bcache_btree_gc_coalesce,
+	TP_PROTO(unsigned nodes),
+	TP_ARGS(nodes),
+
+	TP_STRUCT__entry(
+		__field(unsigned,	nodes			)
+	),
+
+	TP_fast_assign(
+		__entry->nodes	= nodes;
+	),
+
+	TP_printk("coalesced %u nodes", __entry->nodes)
+);
+
+DEFINE_EVENT(cache_set, bcache_gc_start,
+	TP_PROTO(struct cache_set *c),
+	TP_ARGS(c)
+);
+
+DEFINE_EVENT(cache_set, bcache_gc_end,
+	TP_PROTO(struct cache_set *c),
+	TP_ARGS(c)
+);
+
+DEFINE_EVENT(bkey, bcache_gc_copy,
+	TP_PROTO(struct bkey *k),
+	TP_ARGS(k)
+);
+
+DEFINE_EVENT(bkey, bcache_gc_copy_collision,
+	TP_PROTO(struct bkey *k),
+	TP_ARGS(k)
+);
+
+TRACE_EVENT(bcache_btree_insert_key,
+	TP_PROTO(struct btree *b, struct bkey *k, unsigned op, unsigned status),
+	TP_ARGS(b, k, op, status),
+
+	TP_STRUCT__entry(
+		__field(u64,	btree_node			)
+		__field(u32,	btree_level			)
+		__field(u32,	inode				)
+		__field(u64,	offset				)
+		__field(u32,	size				)
+		__field(u8,	dirty				)
+		__field(u8,	op				)
+		__field(u8,	status				)
+	),
+
+	TP_fast_assign(
+		__entry->btree_node = PTR_BUCKET_NR(b->c, &b->key, 0);
+		__entry->btree_level = b->level;
+		__entry->inode	= KEY_INODE(k);
+		__entry->offset	= KEY_OFFSET(k);
+		__entry->size	= KEY_SIZE(k);
+		__entry->dirty	= KEY_DIRTY(k);
+		__entry->op = op;
+		__entry->status = status;
+	),
+
+	TP_printk("%u for %u at %llu(%u): %u:%llu len %u dirty %u",
+		  __entry->status, __entry->op,
+		  __entry->btree_node, __entry->btree_level,
+		  __entry->inode, __entry->offset,
+		  __entry->size, __entry->dirty)
+);
+
+DECLARE_EVENT_CLASS(btree_split,
+	TP_PROTO(struct btree *b, unsigned keys),
+	TP_ARGS(b, keys),
+
+	TP_STRUCT__entry(
+		__field(size_t,		bucket			)
+		__field(unsigned,	keys			)
+	),
+
+	TP_fast_assign(
+		__entry->bucket	= PTR_BUCKET_NR(b->c, &b->key, 0);
+		__entry->keys	= keys;
+	),
+
+	TP_printk("bucket %zu keys %u", __entry->bucket, __entry->keys)
+);
+
+DEFINE_EVENT(btree_split, bcache_btree_node_split,
+	TP_PROTO(struct btree *b, unsigned keys),
+	TP_ARGS(b, keys)
+);
+
+DEFINE_EVENT(btree_split, bcache_btree_node_compact,
+	TP_PROTO(struct btree *b, unsigned keys),
+	TP_ARGS(b, keys)
+);
+
+DEFINE_EVENT(btree_node, bcache_btree_set_root,
+	TP_PROTO(struct btree *b),
+	TP_ARGS(b)
+);
+
+/* Allocator */
+
+TRACE_EVENT(bcache_alloc_invalidate,
+	TP_PROTO(struct cache *ca),
+	TP_ARGS(ca),
+
+	TP_STRUCT__entry(
+		__field(unsigned,	free			)
+		__field(unsigned,	free_inc		)
+		__field(unsigned,	free_inc_size		)
+		__field(unsigned,	unused			)
+	),
+
+	TP_fast_assign(
+		__entry->free		= fifo_used(&ca->free);
+		__entry->free_inc	= fifo_used(&ca->free_inc);
+		__entry->free_inc_size	= ca->free_inc.size;
+		__entry->unused		= fifo_used(&ca->unused);
+	),
+
+	TP_printk("free %u free_inc %u/%u unused %u", __entry->free,
+		  __entry->free_inc, __entry->free_inc_size, __entry->unused)
+);
+
+TRACE_EVENT(bcache_alloc_fail,
+	TP_PROTO(struct cache *ca),
+	TP_ARGS(ca),
+
+	TP_STRUCT__entry(
+		__field(unsigned,	free			)
+		__field(unsigned,	free_inc		)
+		__field(unsigned,	unused			)
+		__field(unsigned,	blocked			)
+	),
+
+	TP_fast_assign(
+		__entry->free		= fifo_used(&ca->free);
+		__entry->free_inc	= fifo_used(&ca->free_inc);
+		__entry->unused		= fifo_used(&ca->unused);
+		__entry->blocked	= atomic_read(&ca->set->prio_blocked);
+	),
+
+	TP_printk("free %u free_inc %u unused %u blocked %u", __entry->free,
+		  __entry->free_inc, __entry->unused, __entry->blocked)
+);
+
+/* Background writeback */
+
+DEFINE_EVENT(bkey, bcache_writeback,
+	TP_PROTO(struct bkey *k),
+	TP_ARGS(k)
+);
+
+DEFINE_EVENT(bkey, bcache_writeback_collision,
+	TP_PROTO(struct bkey *k),
+	TP_ARGS(k)
 );
 
 #endif /* _TRACE_BCACHE_H */

include/xen/interface/io/blkif.h

@@ -102,6 +102,30 @@ typedef uint64_t blkif_sector_t;
  */
 #define BLKIF_OP_DISCARD           5
 
+/*
+ * Recognized if "feature-max-indirect-segments" in present in the backend
+ * xenbus info. The "feature-max-indirect-segments" node contains the maximum
+ * number of segments allowed by the backend per request. If the node is
+ * present, the frontend might use blkif_request_indirect structs in order to
+ * issue requests with more than BLKIF_MAX_SEGMENTS_PER_REQUEST (11). The
+ * maximum number of indirect segments is fixed by the backend, but the
+ * frontend can issue requests with any number of indirect segments as long as
+ * it's less than the number provided by the backend. The indirect_grefs field
+ * in blkif_request_indirect should be filled by the frontend with the
+ * grant references of the pages that are holding the indirect segments.
+ * This pages are filled with an array of blkif_request_segment_aligned
+ * that hold the information about the segments. The number of indirect
+ * pages to use is determined by the maximum number of segments
+ * a indirect request contains. Every indirect page can contain a maximum
+ * of 512 segments (PAGE_SIZE/sizeof(blkif_request_segment_aligned)),
+ * so to calculate the number of indirect pages to use we have to do
+ * ceil(indirect_segments/512).
+ *
+ * If a backend does not recognize BLKIF_OP_INDIRECT, it should *not*
+ * create the "feature-max-indirect-segments" node!
+ */
+#define BLKIF_OP_INDIRECT          6
+
 /*
  * Maximum scatter/gather segments per request.
  * This is carefully chosen so that sizeof(struct blkif_ring) <= PAGE_SIZE.
@@ -109,6 +133,16 @@ typedef uint64_t blkif_sector_t;
  */
 #define BLKIF_MAX_SEGMENTS_PER_REQUEST 11
 
+#define BLKIF_MAX_INDIRECT_PAGES_PER_REQUEST 8
+
+struct blkif_request_segment_aligned {
+	grant_ref_t gref;        /* reference to I/O buffer frame        */
+	/* @first_sect: first sector in frame to transfer (inclusive).   */
+	/* @last_sect: last sector in frame to transfer (inclusive).     */
+	uint8_t     first_sect, last_sect;
+	uint16_t    _pad; /* padding to make it 8 bytes, so it's cache-aligned */
+} __attribute__((__packed__));
+
 struct blkif_request_rw {
 	uint8_t        nr_segments;  /* number of segments                   */
 	blkif_vdev_t   handle;       /* only for read/write requests         */
@@ -147,12 +181,31 @@ struct blkif_request_other {
 	uint64_t     id;           /* private guest value, echoed in resp  */
 } __attribute__((__packed__));
 
+struct blkif_request_indirect {
+	uint8_t        indirect_op;
+	uint16_t       nr_segments;
+#ifdef CONFIG_X86_64
+	uint32_t       _pad1;        /* offsetof(blkif_...,u.indirect.id) == 8 */
+#endif
+	uint64_t       id;
+	blkif_sector_t sector_number;
+	blkif_vdev_t   handle;
+	uint16_t       _pad2;
+	grant_ref_t    indirect_grefs[BLKIF_MAX_INDIRECT_PAGES_PER_REQUEST];
+#ifdef CONFIG_X86_64
+	uint32_t      _pad3;         /* make it 64 byte aligned */
+#else
+	uint64_t      _pad3;         /* make it 64 byte aligned */
+#endif
+} __attribute__((__packed__));
+
 struct blkif_request {
 	uint8_t        operation;    /* BLKIF_OP_???                         */
 	union {
 		struct blkif_request_rw rw;
 		struct blkif_request_discard discard;
 		struct blkif_request_other other;
+		struct blkif_request_indirect indirect;
 	} u;
 } __attribute__((__packed__));
 

include/xen/interface/io/ring.h

@@ -188,6 +188,11 @@ struct __name##_back_ring {						\
 #define RING_REQUEST_CONS_OVERFLOW(_r, _cons)				\
     (((_cons) - (_r)->rsp_prod_pvt) >= RING_SIZE(_r))
 
+/* Ill-behaved frontend determination: Can there be this many requests? */
+#define RING_REQUEST_PROD_OVERFLOW(_r, _prod)               \
+    (((_prod) - (_r)->rsp_prod_pvt) > RING_SIZE(_r))
+
+
 #define RING_PUSH_REQUESTS(_r) do {					\
     wmb(); /* back sees requests /before/ updated producer index */	\
     (_r)->sring->req_prod = (_r)->req_prod_pvt;				\