8 files changed, 131 insertions, 138 deletions

diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c
index 95980ad6b27b..70bca955e0de 100644
--- a/drivers/md/bitmap.c
+++ b/drivers/md/bitmap.c
@@ -108,7 +108,7 @@ static unsigned char *bitmap_alloc_page(struct bitmap *bitmap)
 {
         unsigned char *page;
 
-#if INJECT_FAULTS_1
+#ifdef INJECT_FAULTS_1
         page = NULL;
 #else
         page = kmalloc(PAGE_SIZE, GFP_NOIO);
@@ -843,7 +843,7 @@ static int bitmap_init_from_disk(struct bitmap *bitmap, int in_sync)
 
         BUG_ON(!file && !bitmap->offset);
 
-#if INJECT_FAULTS_3
+#ifdef INJECT_FAULTS_3
         outofdate = 1;
 #else
         outofdate = bitmap->flags & BITMAP_STALE;
@@ -1187,7 +1187,7 @@ static int bitmap_start_daemon(struct bitmap *bitmap, mdk_thread_t **ptr,
 
         spin_unlock_irqrestore(&bitmap->lock, flags);
 
-#if INJECT_FATAL_FAULT_2
+#ifdef INJECT_FATAL_FAULT_2
         daemon = NULL;
 #else
         sprintf(namebuf, "%%s_%s", name);
@@ -1345,7 +1345,8 @@ void bitmap_endwrite(struct bitmap *bitmap, sector_t offset, unsigned long secto
         }
 }
 
-int bitmap_start_sync(struct bitmap *bitmap, sector_t offset, int *blocks)
+int bitmap_start_sync(struct bitmap *bitmap, sector_t offset, int *blocks,
+                        int degraded)
 {
         bitmap_counter_t *bmc;
         int rv;
@@ -1362,8 +1363,10 @@ int bitmap_start_sync(struct bitmap *bitmap, sector_t offset, int *blocks)
                         rv = 1;
                 else if (NEEDED(*bmc)) {
                         rv = 1;
-                        *bmc |= RESYNC_MASK;
-                        *bmc &= ~NEEDED_MASK;
+                        if (!degraded) { /* don't set/clear bits if degraded */
+                                *bmc |= RESYNC_MASK;
+                                *bmc &= ~NEEDED_MASK;
+                        }
                 }
         }
         spin_unlock_irq(&bitmap->lock);
@@ -1549,7 +1552,7 @@ int bitmap_create(mddev_t *mddev)
 
         bitmap->syncchunk = ~0UL;
 
-#if INJECT_FATAL_FAULT_1
+#ifdef INJECT_FATAL_FAULT_1
         bitmap->bp = NULL;
 #else
         bitmap->bp = kmalloc(pages * sizeof(*bitmap->bp), GFP_KERNEL);
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index a5a4c0ed8a14..a6d3baa46f61 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -869,11 +869,17 @@ static void suspend_targets(struct dm_table *t, unsigned postsuspend)
 
 void dm_table_presuspend_targets(struct dm_table *t)
 {
+        if (!t)
+                return;
+
         return suspend_targets(t, 0);
 }
 
 void dm_table_postsuspend_targets(struct dm_table *t)
 {
+        if (!t)
+                return;
+
         return suspend_targets(t, 1);
 }
 
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 54fabbf06678..d487d9deb98e 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -55,10 +55,10 @@ union map_info *dm_get_mapinfo(struct bio *bio)
  */
 #define DMF_BLOCK_IO 0
 #define DMF_SUSPENDED 1
-#define DMF_FS_LOCKED 2
 
 struct mapped_device {
-        struct rw_semaphore lock;
+        struct rw_semaphore io_lock;
+        struct semaphore suspend_lock;
         rwlock_t map_lock;
         atomic_t holders;
 
@@ -248,16 +248,16 @@ static inline void free_tio(struct mapped_device *md, struct target_io *tio)
  */
 static int queue_io(struct mapped_device *md, struct bio *bio)
 {
-        down_write(&md->lock);
+        down_write(&md->io_lock);
 
         if (!test_bit(DMF_BLOCK_IO, &md->flags)) {
-                up_write(&md->lock);
+                up_write(&md->io_lock);
                 return 1;
         }
 
         bio_list_add(&md->deferred, bio);
 
-        up_write(&md->lock);
+        up_write(&md->io_lock);
         return 0;               /* deferred successfully */
 }
 
@@ -568,14 +568,14 @@ static int dm_request(request_queue_t *q, struct bio *bio)
         int r;
         struct mapped_device *md = q->queuedata;
 
-        down_read(&md->lock);
+        down_read(&md->io_lock);
 
         /*
          * If we're suspended we have to queue
          * this io for later.
          */
         while (test_bit(DMF_BLOCK_IO, &md->flags)) {
-                up_read(&md->lock);
+                up_read(&md->io_lock);
 
                 if (bio_rw(bio) == READA) {
                         bio_io_error(bio, bio->bi_size);
@@ -594,11 +594,11 @@ static int dm_request(request_queue_t *q, struct bio *bio)
                  * We're in a while loop, because someone could suspend
                  * before we get to the following read lock.
                  */
-                down_read(&md->lock);
+                down_read(&md->io_lock);
         }
 
         __split_bio(md, bio);
-        up_read(&md->lock);
+        up_read(&md->io_lock);
         return 0;
 }
 
@@ -610,7 +610,7 @@ static int dm_flush_all(request_queue_t *q, struct gendisk *disk,
         int ret = -ENXIO;
 
         if (map) {
-                ret = dm_table_flush_all(md->map);
+                ret = dm_table_flush_all(map);
                 dm_table_put(map);
         }
 
@@ -747,7 +747,8 @@ static struct mapped_device *alloc_dev(unsigned int minor, int persistent)
                 goto bad1;
 
         memset(md, 0, sizeof(*md));
-        init_rwsem(&md->lock);
+        init_rwsem(&md->io_lock);
+        init_MUTEX(&md->suspend_lock);
         rwlock_init(&md->map_lock);
         atomic_set(&md->holders, 1);
         atomic_set(&md->event_nr, 0);
@@ -825,18 +826,13 @@ static void event_callback(void *context)
         wake_up(&md->eventq);
 }
 
-static void __set_size(struct gendisk *disk, sector_t size)
+static void __set_size(struct mapped_device *md, sector_t size)
 {
-        struct block_device *bdev;
-
-        set_capacity(disk, size);
-        bdev = bdget_disk(disk, 0);
-        if (bdev) {
-                down(&bdev->bd_inode->i_sem);
-                i_size_write(bdev->bd_inode, (loff_t)size << SECTOR_SHIFT);
-                up(&bdev->bd_inode->i_sem);
-                bdput(bdev);
-        }
+        set_capacity(md->disk, size);
+
+        down(&md->frozen_bdev->bd_inode->i_sem);
+        i_size_write(md->frozen_bdev->bd_inode, (loff_t)size << SECTOR_SHIFT);
+        up(&md->frozen_bdev->bd_inode->i_sem);
 }
 
 static int __bind(struct mapped_device *md, struct dm_table *t)
@@ -845,17 +841,18 @@ static int __bind(struct mapped_device *md, struct dm_table *t)
         sector_t size;
 
         size = dm_table_get_size(t);
-        __set_size(md->disk, size);
+        __set_size(md, size);
         if (size == 0)
                 return 0;
 
+        dm_table_get(t);
+        dm_table_event_callback(t, event_callback, md);
+
         write_lock(&md->map_lock);
         md->map = t;
+        dm_table_set_restrictions(t, q);
         write_unlock(&md->map_lock);
 
-        dm_table_get(t);
-        dm_table_event_callback(md->map, event_callback, md);
-        dm_table_set_restrictions(t, q);
         return 0;
 }
 
@@ -935,7 +932,7 @@ void dm_put(struct mapped_device *md)
         struct dm_table *map = dm_get_table(md);
 
         if (atomic_dec_and_test(&md->holders)) {
-                if (!test_bit(DMF_SUSPENDED, &md->flags) && map) {
+                if (!dm_suspended(md)) {
                         dm_table_presuspend_targets(map);
                         dm_table_postsuspend_targets(map);
                 }
@@ -968,17 +965,17 @@ int dm_swap_table(struct mapped_device *md, struct dm_table *table)
 {
         int r = -EINVAL;
 
-        down_write(&md->lock);
+        down(&md->suspend_lock);
 
         /* device must be suspended */
-        if (!test_bit(DMF_SUSPENDED, &md->flags))
+        if (!dm_suspended(md))
                 goto out;
 
         __unbind(md);
         r = __bind(md, table);
 
 out:
-        up_write(&md->lock);
+        up(&md->suspend_lock);
         return r;
 }
 
@@ -986,16 +983,13 @@ out:
  * Functions to lock and unlock any filesystem running on the
  * device.
  */
-static int __lock_fs(struct mapped_device *md)
+static int lock_fs(struct mapped_device *md)
 {
-        int error = -ENOMEM;
-
-        if (test_and_set_bit(DMF_FS_LOCKED, &md->flags))
-                return 0;
+        int r = -ENOMEM;
 
         md->frozen_bdev = bdget_disk(md->disk, 0);
         if (!md->frozen_bdev) {
-                DMWARN("bdget failed in __lock_fs");
+                DMWARN("bdget failed in lock_fs");
                 goto out;
         }
 
@@ -1003,13 +997,13 @@ static int __lock_fs(struct mapped_device *md)
 
         md->frozen_sb = freeze_bdev(md->frozen_bdev);
         if (IS_ERR(md->frozen_sb)) {
-                error = PTR_ERR(md->frozen_sb);
+                r = PTR_ERR(md->frozen_sb);
                 goto out_bdput;
         }
 
         /* don't bdput right now, we don't want the bdev
          * to go away while it is locked.  We'll bdput
-         * in __unlock_fs
+         * in unlock_fs
          */
         return 0;
 
@@ -1018,15 +1012,11 @@ out_bdput:
         md->frozen_sb = NULL;
         md->frozen_bdev = NULL;
 out:
-        clear_bit(DMF_FS_LOCKED, &md->flags);
-        return error;
+        return r;
 }
 
-static void __unlock_fs(struct mapped_device *md)
+static void unlock_fs(struct mapped_device *md)
 {
-        if (!test_and_clear_bit(DMF_FS_LOCKED, &md->flags))
-                return;
-
         thaw_bdev(md->frozen_bdev, md->frozen_sb);
         bdput(md->frozen_bdev);
 
@@ -1043,50 +1033,37 @@ static void __unlock_fs(struct mapped_device *md)
  */
 int dm_suspend(struct mapped_device *md)
 {
-        struct dm_table *map;
+        struct dm_table *map = NULL;
         DECLARE_WAITQUEUE(wait, current);
-        int error = -EINVAL;
+        int r = -EINVAL;
 
-        /* Flush I/O to the device. */
-        down_read(&md->lock);
-        if (test_bit(DMF_BLOCK_IO, &md->flags))
-                goto out_read_unlock;
+        down(&md->suspend_lock);
+
+        if (dm_suspended(md))
+                goto out;
 
         map = dm_get_table(md);
-        if (map)
-                /* This does not get reverted if there's an error later. */
-                dm_table_presuspend_targets(map);
 
-        error = __lock_fs(md);
-        if (error) {
-                dm_table_put(map);
-                goto out_read_unlock;
-        }
+        /* This does not get reverted if there's an error later. */
+        dm_table_presuspend_targets(map);
 
-        up_read(&md->lock);
+        /* Flush I/O to the device. */
+        r = lock_fs(md);
+        if (r)
+                goto out;
 
         /*
          * First we set the BLOCK_IO flag so no more ios will be mapped.
-         *
-         * If the flag is already set we know another thread is trying to
-         * suspend as well, so we leave the fs locked for this thread.
          */
-        error = -EINVAL;
-        down_write(&md->lock);
-        if (test_and_set_bit(DMF_BLOCK_IO, &md->flags)) {
-                if (map)
-                        dm_table_put(map);
-                goto out_write_unlock;
-        }
+        down_write(&md->io_lock);
+        set_bit(DMF_BLOCK_IO, &md->flags);
 
         add_wait_queue(&md->wait, &wait);
-        up_write(&md->lock);
+        up_write(&md->io_lock);
 
         /* unplug */
-        if (map) {
+        if (map)
                 dm_table_unplug_all(map);
-                dm_table_put(map);
-        }
 
         /*
          * Then we wait for the already mapped ios to
@@ -1102,62 +1079,67 @@ int dm_suspend(struct mapped_device *md)
         }
         set_current_state(TASK_RUNNING);
 
-        down_write(&md->lock);
+        down_write(&md->io_lock);
         remove_wait_queue(&md->wait, &wait);
 
         /* were we interrupted ? */
-        error = -EINTR;
-        if (atomic_read(&md->pending))
-                goto out_unfreeze;
-
-        set_bit(DMF_SUSPENDED, &md->flags);
+        r = -EINTR;
+        if (atomic_read(&md->pending)) {
+                up_write(&md->io_lock);
+                unlock_fs(md);
+                clear_bit(DMF_BLOCK_IO, &md->flags);
+                goto out;
+        }
+        up_write(&md->io_lock);
 
-        map = dm_get_table(md);
-        if (map)
-                dm_table_postsuspend_targets(map);
-        dm_table_put(map);
-        up_write(&md->lock);
+        dm_table_postsuspend_targets(map);
 
-        return 0;
+        set_bit(DMF_SUSPENDED, &md->flags);
 
-out_unfreeze:
-        __unlock_fs(md);
-        clear_bit(DMF_BLOCK_IO, &md->flags);
-out_write_unlock:
-        up_write(&md->lock);
-        return error;
+        r = 0;
 
-out_read_unlock:
-        up_read(&md->lock);
-        return error;
+out:
+        dm_table_put(map);
+        up(&md->suspend_lock);
+        return r;
 }
 
 int dm_resume(struct mapped_device *md)
 {
+        int r = -EINVAL;
         struct bio *def;
-        struct dm_table *map = dm_get_table(md);
+        struct dm_table *map = NULL;
 
-        down_write(&md->lock);
-        if (!map ||
-            !test_bit(DMF_SUSPENDED, &md->flags) ||
-            !dm_table_get_size(map)) {
-                up_write(&md->lock);
-                dm_table_put(map);
-                return -EINVAL;
-        }
+        down(&md->suspend_lock);
+        if (!dm_suspended(md))
+                goto out;
+
+        map = dm_get_table(md);
+        if (!map || !dm_table_get_size(map))
+                goto out;
 
         dm_table_resume_targets(map);
-        clear_bit(DMF_SUSPENDED, &md->flags);
+
+        down_write(&md->io_lock);
         clear_bit(DMF_BLOCK_IO, &md->flags);
 
         def = bio_list_get(&md->deferred);
         __flush_deferred_io(md, def);
-        up_write(&md->lock);
-        __unlock_fs(md);
+        up_write(&md->io_lock);
+
+        unlock_fs(md);
+
+        clear_bit(DMF_SUSPENDED, &md->flags);
+
         dm_table_unplug_all(map);
+
+        r = 0;
+
+out:
         dm_table_put(map);
+        up(&md->suspend_lock);
 
-        return 0;
+        return r;
 }
 
 /*-----------------------------------------------------------------
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 4a0c57db2b67..6580e0fa4a47 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -284,7 +284,7 @@ static mdk_rdev_t * find_rdev(mddev_t * mddev, dev_t dev)
         return NULL;
 }
 
-inline static sector_t calc_dev_sboffset(struct block_device *bdev)
+static inline sector_t calc_dev_sboffset(struct block_device *bdev)
 {
         sector_t size = bdev->bd_inode->i_size >> BLOCK_SIZE_BITS;
         return MD_NEW_SIZE_BLOCKS(size);
diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
index e11dd14d0b43..2120710172c5 100644
--- a/drivers/md/raid0.c
+++ b/drivers/md/raid0.c
@@ -314,16 +314,16 @@ static int raid0_run (mddev_t *mddev)
                 sector_t space = conf->hash_spacing;
                 int round;
                 conf->preshift = 0;
-                if (sizeof(sector_t) > sizeof(unsigned long)) {
+                if (sizeof(sector_t) > sizeof(u32)) {
                         /*shift down space and s so that sector_div will work */
-                        while (space > (sector_t) (~(unsigned long)0)) {
+                        while (space > (sector_t) (~(u32)0)) {
                                 s >>= 1;
                                 space >>= 1;
                                 s += 1; /* force round-up */
                                 conf->preshift++;
                         }
                 }
-                round = sector_div(s, (unsigned long)space) ? 1 : 0;
+                round = sector_div(s, (u32)space) ? 1 : 0;
                 nb_zone = s + round;
         }
         printk("raid0 : nb_zone is %d.\n", nb_zone);
@@ -443,7 +443,7 @@ static int raid0_make_request (request_queue_t *q, struct bio *bio)
                 volatile
 #endif
                 sector_t x = block >> conf->preshift;
-                sector_div(x, (unsigned long)conf->hash_spacing);
+                sector_div(x, (u32)conf->hash_spacing);
                 zone = conf->hash_table[x];
         }
  
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index ff1dbec864af..d3a64a04a6d8 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -1126,21 +1126,19 @@ static sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *skipped, i
                  * only be one in raid1 resync.
                  * We can find the current addess in mddev->curr_resync
                  */
-                if (!conf->fullsync) {
-                        if (mddev->curr_resync < max_sector)
-                                bitmap_end_sync(mddev->bitmap,
-                                                mddev->curr_resync,
+                if (mddev->curr_resync < max_sector) /* aborted */
+                        bitmap_end_sync(mddev->bitmap, mddev->curr_resync,
                                                 &sync_blocks, 1);
-                        bitmap_close_sync(mddev->bitmap);
-                }
-                if (mddev->curr_resync >= max_sector)
+                else /* completed sync */
                         conf->fullsync = 0;
+
+                bitmap_close_sync(mddev->bitmap);
                 close_sync(conf);
                 return 0;
         }
 
-        if (!conf->fullsync &&
-            !bitmap_start_sync(mddev->bitmap, sector_nr, &sync_blocks)) {
+        if (!bitmap_start_sync(mddev->bitmap, sector_nr, &sync_blocks, mddev->degraded) &&
+            !conf->fullsync) {
                 /* We can skip this block, and probably several more */
                 *skipped = 1;
                 return sync_blocks;
@@ -1243,15 +1241,15 @@ static sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *skipped, i
                         len = (max_sector - sector_nr) << 9;
                 if (len == 0)
                         break;
-                if (!conf->fullsync) {
-                        if (sync_blocks == 0) {
-                                if (!bitmap_start_sync(mddev->bitmap,
-                                                       sector_nr, &sync_blocks))
-                                        break;
-                                if (sync_blocks < (PAGE_SIZE>>9))
-                                        BUG();
-                                if (len > (sync_blocks<<9)) len = sync_blocks<<9;
-                        }
+                if (sync_blocks == 0) {
+                        if (!bitmap_start_sync(mddev->bitmap, sector_nr,
+                                        &sync_blocks, mddev->degraded) &&
+                                        !conf->fullsync)
+                                break;
+                        if (sync_blocks < (PAGE_SIZE>>9))
+                                BUG();
+                        if (len > (sync_blocks<<9))
+                                len = sync_blocks<<9;
                 }
 
                 for (i=0 ; i < conf->raid_disks; i++) {
@@ -1264,7 +1262,8 @@ static sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *skipped, i
                                         while (i > 0) {
                                                 i--;
                                                 bio = r1_bio->bios[i];
-                                                if (bio->bi_end_io==NULL) continue;
+                                                if (bio->bi_end_io==NULL)
+                                                        continue;
                                                 /* remove last page from this bio */
                                                 bio->bi_vcnt--;
                                                 bio->bi_size -= len;
@@ -1469,6 +1468,7 @@ static int raid1_resize(mddev_t *mddev, sector_t sectors)
                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
         }
         mddev->size = mddev->array_size;
+        mddev->resync_max_sectors = sectors;
         return 0;
 }
 
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 93a9726cc2d6..4698d5f79575 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -1931,6 +1931,7 @@ static int raid5_resize(mddev_t *mddev, sector_t sectors)
                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
         }
         mddev->size = sectors /2;
+        mddev->resync_max_sectors = sectors;
         return 0;
 }
 
diff --git a/drivers/md/raid6main.c b/drivers/md/raid6main.c
index f62ea1a73d0d..f5ee16805111 100644
--- a/drivers/md/raid6main.c
+++ b/drivers/md/raid6main.c
@@ -2095,6 +2095,7 @@ static int raid6_resize(mddev_t *mddev, sector_t sectors)
                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
         }
         mddev->size = sectors /2;
+        mddev->resync_max_sectors = sectors;
         return 0;
 }