block: make bi_phys_segments an unsigned int instead of short
[linux-2.6.git] / drivers / md / raid5.c
index 1cfc984..37e5465 100644 (file)
@@ -63,6 +63,7 @@
 #define STRIPE_SHIFT           (PAGE_SHIFT - 9)
 #define STRIPE_SECTORS         (STRIPE_SIZE>>9)
 #define        IO_THRESHOLD            1
+#define BYPASS_THRESHOLD       1
 #define NR_HASH                        (PAGE_SIZE / sizeof(struct hlist_head))
 #define HASH_MASK              (NR_HASH - 1)
 
 #define __inline__
 #endif
 
+#define printk_rl(args...) ((void) (printk_ratelimit() && printk(args)))
+
 #if !RAID6_USE_EMPTY_ZERO_PAGE
 /* In .bss so it's zeroed */
 const char raid6_empty_zero_page[PAGE_SIZE] __attribute__((aligned(256)));
 #endif
 
+/*
+ * We maintain a biased count of active stripes in the bottom 16 bits of
+ * bi_phys_segments, and a count of processed stripes in the upper 16 bits
+ */
+static inline int raid5_bi_phys_segments(struct bio *bio)
+{
+       return bio->bi_phys_segments & 0xffff;
+}
+
+static inline int raid5_bi_hw_segments(struct bio *bio)
+{
+       return (bio->bi_phys_segments >> 16) & 0xffff;
+}
+
+static inline int raid5_dec_bi_phys_segments(struct bio *bio)
+{
+       --bio->bi_phys_segments;
+       return raid5_bi_phys_segments(bio);
+}
+
+static inline int raid5_dec_bi_hw_segments(struct bio *bio)
+{
+       unsigned short val = raid5_bi_hw_segments(bio);
+
+       --val;
+       bio->bi_phys_segments = (val << 16) | raid5_bi_phys_segments(bio);
+       return val;
+}
+
+static inline void raid5_set_bi_hw_segments(struct bio *bio, unsigned int cnt)
+{
+       bio->bi_phys_segments = raid5_bi_phys_segments(bio) || (cnt << 16);
+}
+
 static inline int raid6_next_disk(int disk, int raid_disks)
 {
        disk++;
@@ -112,15 +149,20 @@ static void return_io(struct bio *return_bi)
                return_bi = bi->bi_next;
                bi->bi_next = NULL;
                bi->bi_size = 0;
-               bi->bi_end_io(bi,
-                             test_bit(BIO_UPTODATE, &bi->bi_flags)
-                               ? 0 : -EIO);
+               bio_endio(bi, 0);
                bi = return_bi;
        }
 }
 
 static void print_raid5_conf (raid5_conf_t *conf);
 
+static int stripe_operations_active(struct stripe_head *sh)
+{
+       return sh->check_state || sh->reconstruct_state ||
+              test_bit(STRIPE_BIOFILL_RUN, &sh->state) ||
+              test_bit(STRIPE_COMPUTE_RUN, &sh->state);
+}
+
 static void __release_stripe(raid5_conf_t *conf, struct stripe_head *sh)
 {
        if (atomic_dec_and_test(&sh->count)) {
@@ -140,7 +182,7 @@ static void __release_stripe(raid5_conf_t *conf, struct stripe_head *sh)
                        }
                        md_wakeup_thread(conf->mddev->thread);
                } else {
-                       BUG_ON(sh->ops.pending);
+                       BUG_ON(stripe_operations_active(sh));
                        if (test_and_clear_bit(STRIPE_PREREAD_ACTIVE, &sh->state)) {
                                atomic_dec(&conf->preread_active_stripes);
                                if (atomic_read(&conf->preread_active_stripes) < IO_THRESHOLD)
@@ -242,7 +284,7 @@ static void init_stripe(struct stripe_head *sh, sector_t sector, int pd_idx, int
 
        BUG_ON(atomic_read(&sh->count) != 0);
        BUG_ON(test_bit(STRIPE_HANDLE, &sh->state));
-       BUG_ON(sh->ops.pending || sh->ops.ack || sh->ops.complete);
+       BUG_ON(stripe_operations_active(sh));
 
        CHECK_DEVLOCK();
        pr_debug("init_stripe called, stripe %llu\n",
@@ -343,55 +385,12 @@ static struct stripe_head *get_active_stripe(raid5_conf_t *conf, sector_t sector
        return sh;
 }
 
-/* test_and_ack_op() ensures that we only dequeue an operation once */
-#define test_and_ack_op(op, pend) \
-do {                                                   \
-       if (test_bit(op, &sh->ops.pending) &&           \
-               !test_bit(op, &sh->ops.complete)) {     \
-               if (test_and_set_bit(op, &sh->ops.ack)) \
-                       clear_bit(op, &pend);           \
-               else                                    \
-                       ack++;                          \
-       } else                                          \
-               clear_bit(op, &pend);                   \
-} while (0)
-
-/* find new work to run, do not resubmit work that is already
- * in flight
- */
-static unsigned long get_stripe_work(struct stripe_head *sh)
-{
-       unsigned long pending;
-       int ack = 0;
-
-       pending = sh->ops.pending;
-
-       test_and_ack_op(STRIPE_OP_BIOFILL, pending);
-       test_and_ack_op(STRIPE_OP_COMPUTE_BLK, pending);
-       test_and_ack_op(STRIPE_OP_PREXOR, pending);
-       test_and_ack_op(STRIPE_OP_BIODRAIN, pending);
-       test_and_ack_op(STRIPE_OP_POSTXOR, pending);
-       test_and_ack_op(STRIPE_OP_CHECK, pending);
-       if (test_and_clear_bit(STRIPE_OP_IO, &sh->ops.pending))
-               ack++;
-
-       sh->ops.count -= ack;
-       if (unlikely(sh->ops.count < 0)) {
-               printk(KERN_ERR "pending: %#lx ops.pending: %#lx ops.ack: %#lx "
-                       "ops.complete: %#lx\n", pending, sh->ops.pending,
-                       sh->ops.ack, sh->ops.complete);
-               BUG();
-       }
-
-       return pending;
-}
-
 static void
 raid5_end_read_request(struct bio *bi, int error);
 static void
 raid5_end_write_request(struct bio *bi, int error);
 
-static void ops_run_io(struct stripe_head *sh)
+static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
 {
        raid5_conf_t *conf = sh->raid_conf;
        int i, disks = sh->disks;
@@ -426,14 +425,14 @@ static void ops_run_io(struct stripe_head *sh)
                rcu_read_unlock();
 
                if (rdev) {
-                       if (test_bit(STRIPE_SYNCING, &sh->state) ||
-                               test_bit(STRIPE_EXPAND_SOURCE, &sh->state) ||
-                               test_bit(STRIPE_EXPAND_READY, &sh->state))
+                       if (s->syncing || s->expanding || s->expanded)
                                md_sync_acct(rdev->bdev, STRIPE_SECTORS);
 
+                       set_bit(STRIPE_IO_STARTED, &sh->state);
+
                        bi->bi_bdev = rdev->bdev;
                        pr_debug("%s: for %llu schedule op %ld on disc %d\n",
-                               __FUNCTION__, (unsigned long long)sh->sector,
+                               __func__, (unsigned long long)sh->sector,
                                bi->bi_rw, i);
                        atomic_inc(&sh->count);
                        bi->bi_sector = sh->sector + rdev->data_offset;
@@ -520,42 +519,38 @@ static void ops_complete_biofill(void *stripe_head_ref)
        raid5_conf_t *conf = sh->raid_conf;
        int i;
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
        /* clear completed biofills */
+       spin_lock_irq(&conf->device_lock);
        for (i = sh->disks; i--; ) {
                struct r5dev *dev = &sh->dev[i];
 
                /* acknowledge completion of a biofill operation */
                /* and check if we need to reply to a read request,
                 * new R5_Wantfill requests are held off until
-                * !test_bit(STRIPE_OP_BIOFILL, &sh->ops.pending)
+                * !STRIPE_BIOFILL_RUN
                 */
                if (test_and_clear_bit(R5_Wantfill, &dev->flags)) {
                        struct bio *rbi, *rbi2;
 
-                       /* The access to dev->read is outside of the
-                        * spin_lock_irq(&conf->device_lock), but is protected
-                        * by the STRIPE_OP_BIOFILL pending bit
-                        */
                        BUG_ON(!dev->read);
                        rbi = dev->read;
                        dev->read = NULL;
                        while (rbi && rbi->bi_sector <
                                dev->sector + STRIPE_SECTORS) {
                                rbi2 = r5_next_bio(rbi, dev->sector);
-                               spin_lock_irq(&conf->device_lock);
-                               if (--rbi->bi_phys_segments == 0) {
+                               if (!raid5_dec_bi_phys_segments(rbi)) {
                                        rbi->bi_next = return_bi;
                                        return_bi = rbi;
                                }
-                               spin_unlock_irq(&conf->device_lock);
                                rbi = rbi2;
                        }
                }
        }
-       set_bit(STRIPE_OP_BIOFILL, &sh->ops.complete);
+       spin_unlock_irq(&conf->device_lock);
+       clear_bit(STRIPE_BIOFILL_RUN, &sh->state);
 
        return_io(return_bi);
 
@@ -569,7 +564,7 @@ static void ops_run_biofill(struct stripe_head *sh)
        raid5_conf_t *conf = sh->raid_conf;
        int i;
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
        for (i = sh->disks; i--; ) {
@@ -600,19 +595,20 @@ static void ops_complete_compute5(void *stripe_head_ref)
        int target = sh->ops.target;
        struct r5dev *tgt = &sh->dev[target];
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
        set_bit(R5_UPTODATE, &tgt->flags);
        BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));
        clear_bit(R5_Wantcompute, &tgt->flags);
-       set_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.complete);
+       clear_bit(STRIPE_COMPUTE_RUN, &sh->state);
+       if (sh->check_state == check_state_compute_run)
+               sh->check_state = check_state_compute_result;
        set_bit(STRIPE_HANDLE, &sh->state);
        release_stripe(sh);
 }
 
-static struct dma_async_tx_descriptor *
-ops_run_compute5(struct stripe_head *sh, unsigned long pending)
+static struct dma_async_tx_descriptor *ops_run_compute5(struct stripe_head *sh)
 {
        /* kernel stack size limits the total number of disks */
        int disks = sh->disks;
@@ -625,7 +621,7 @@ ops_run_compute5(struct stripe_head *sh, unsigned long pending)
        int i;
 
        pr_debug("%s: stripe %llu block: %d\n",
-               __FUNCTION__, (unsigned long long)sh->sector, target);
+               __func__, (unsigned long long)sh->sector, target);
        BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));
 
        for (i = disks; i--; )
@@ -642,10 +638,6 @@ ops_run_compute5(struct stripe_head *sh, unsigned long pending)
                        ASYNC_TX_XOR_ZERO_DST, NULL,
                        ops_complete_compute5, sh);
 
-       /* ack now if postxor is not set to be run */
-       if (tx && !test_bit(STRIPE_OP_POSTXOR, &pending))
-               async_tx_ack(tx);
-
        return tx;
 }
 
@@ -653,10 +645,8 @@ static void ops_complete_prexor(void *stripe_head_ref)
 {
        struct stripe_head *sh = stripe_head_ref;
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
-
-       set_bit(STRIPE_OP_PREXOR, &sh->ops.complete);
 }
 
 static struct dma_async_tx_descriptor *
@@ -670,13 +660,13 @@ ops_run_prexor(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
        /* existing parity data subtracted */
        struct page *xor_dest = xor_srcs[count++] = sh->dev[pd_idx].page;
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
        for (i = disks; i--; ) {
                struct r5dev *dev = &sh->dev[i];
                /* Only process blocks that are known to be uptodate */
-               if (dev->towrite && test_bit(R5_Wantprexor, &dev->flags))
+               if (test_bit(R5_Wantdrain, &dev->flags))
                        xor_srcs[count++] = dev->page;
        }
 
@@ -691,33 +681,16 @@ static struct dma_async_tx_descriptor *
 ops_run_biodrain(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
 {
        int disks = sh->disks;
-       int pd_idx = sh->pd_idx, i;
-
-       /* check if prexor is active which means only process blocks
-        * that are part of a read-modify-write (Wantprexor)
-        */
-       int prexor = test_bit(STRIPE_OP_PREXOR, &sh->ops.pending);
+       int i;
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
        for (i = disks; i--; ) {
                struct r5dev *dev = &sh->dev[i];
                struct bio *chosen;
-               int towrite;
-
-               towrite = 0;
-               if (prexor) { /* rmw */
-                       if (dev->towrite &&
-                           test_bit(R5_Wantprexor, &dev->flags))
-                               towrite = 1;
-               } else { /* rcw */
-                       if (i != pd_idx && dev->towrite &&
-                               test_bit(R5_LOCKED, &dev->flags))
-                               towrite = 1;
-               }
 
-               if (towrite) {
+               if (test_and_clear_bit(R5_Wantdrain, &dev->flags)) {
                        struct bio *wbi;
 
                        spin_lock(&sh->lock);
@@ -742,21 +715,9 @@ ops_run_biodrain(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
 static void ops_complete_postxor(void *stripe_head_ref)
 {
        struct stripe_head *sh = stripe_head_ref;
-
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
-               (unsigned long long)sh->sector);
-
-       set_bit(STRIPE_OP_POSTXOR, &sh->ops.complete);
-       set_bit(STRIPE_HANDLE, &sh->state);
-       release_stripe(sh);
-}
-
-static void ops_complete_write(void *stripe_head_ref)
-{
-       struct stripe_head *sh = stripe_head_ref;
        int disks = sh->disks, i, pd_idx = sh->pd_idx;
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
        for (i = disks; i--; ) {
@@ -765,8 +726,14 @@ static void ops_complete_write(void *stripe_head_ref)
                        set_bit(R5_UPTODATE, &dev->flags);
        }
 
-       set_bit(STRIPE_OP_BIODRAIN, &sh->ops.complete);
-       set_bit(STRIPE_OP_POSTXOR, &sh->ops.complete);
+       if (sh->reconstruct_state == reconstruct_state_drain_run)
+               sh->reconstruct_state = reconstruct_state_drain_result;
+       else if (sh->reconstruct_state == reconstruct_state_prexor_drain_run)
+               sh->reconstruct_state = reconstruct_state_prexor_drain_result;
+       else {
+               BUG_ON(sh->reconstruct_state != reconstruct_state_run);
+               sh->reconstruct_state = reconstruct_state_result;
+       }
 
        set_bit(STRIPE_HANDLE, &sh->state);
        release_stripe(sh);
@@ -781,17 +748,17 @@ ops_run_postxor(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
 
        int count = 0, pd_idx = sh->pd_idx, i;
        struct page *xor_dest;
-       int prexor = test_bit(STRIPE_OP_PREXOR, &sh->ops.pending);
+       int prexor = 0;
        unsigned long flags;
-       dma_async_tx_callback callback;
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
        /* check if prexor is active which means only process blocks
         * that are part of a read-modify-write (written)
         */
-       if (prexor) {
+       if (sh->reconstruct_state == reconstruct_state_prexor_drain_run) {
+               prexor = 1;
                xor_dest = xor_srcs[count++] = sh->dev[pd_idx].page;
                for (i = disks; i--; ) {
                        struct r5dev *dev = &sh->dev[i];
@@ -807,10 +774,6 @@ ops_run_postxor(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
                }
        }
 
-       /* check whether this postxor is part of a write */
-       callback = test_bit(STRIPE_OP_BIODRAIN, &sh->ops.pending) ?
-               ops_complete_write : ops_complete_postxor;
-
        /* 1/ if we prexor'd then the dest is reused as a source
         * 2/ if we did not prexor then we are redoing the parity
         * set ASYNC_TX_XOR_DROP_DST and ASYNC_TX_XOR_ZERO_DST
@@ -824,25 +787,20 @@ ops_run_postxor(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
        if (unlikely(count == 1)) {
                flags &= ~(ASYNC_TX_XOR_DROP_DST | ASYNC_TX_XOR_ZERO_DST);
                tx = async_memcpy(xor_dest, xor_srcs[0], 0, 0, STRIPE_SIZE,
-                       flags, tx, callback, sh);
+                       flags, tx, ops_complete_postxor, sh);
        } else
                tx = async_xor(xor_dest, xor_srcs, 0, count, STRIPE_SIZE,
-                       flags, tx, callback, sh);
+                       flags, tx, ops_complete_postxor, sh);
 }
 
 static void ops_complete_check(void *stripe_head_ref)
 {
        struct stripe_head *sh = stripe_head_ref;
-       int pd_idx = sh->pd_idx;
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
-       if (test_and_clear_bit(STRIPE_OP_MOD_DMA_CHECK, &sh->ops.pending) &&
-               sh->ops.zero_sum_result == 0)
-               set_bit(R5_UPTODATE, &sh->dev[pd_idx].flags);
-
-       set_bit(STRIPE_OP_CHECK, &sh->ops.complete);
+       sh->check_state = check_state_check_result;
        set_bit(STRIPE_HANDLE, &sh->state);
        release_stripe(sh);
 }
@@ -857,7 +815,7 @@ static void ops_run_check(struct stripe_head *sh)
        int count = 0, pd_idx = sh->pd_idx, i;
        struct page *xor_dest = xor_srcs[count++] = sh->dev[pd_idx].page;
 
-       pr_debug("%s: stripe %llu\n", __FUNCTION__,
+       pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
        for (i = disks; i--; ) {
@@ -869,46 +827,42 @@ static void ops_run_check(struct stripe_head *sh)
        tx = async_xor_zero_sum(xor_dest, xor_srcs, 0, count, STRIPE_SIZE,
                &sh->ops.zero_sum_result, 0, NULL, NULL, NULL);
 
-       if (tx)
-               set_bit(STRIPE_OP_MOD_DMA_CHECK, &sh->ops.pending);
-       else
-               clear_bit(STRIPE_OP_MOD_DMA_CHECK, &sh->ops.pending);
-
        atomic_inc(&sh->count);
        tx = async_trigger_callback(ASYNC_TX_DEP_ACK | ASYNC_TX_ACK, tx,
                ops_complete_check, sh);
 }
 
-static void raid5_run_ops(struct stripe_head *sh, unsigned long pending)
+static void raid5_run_ops(struct stripe_head *sh, unsigned long ops_request)
 {
        int overlap_clear = 0, i, disks = sh->disks;
        struct dma_async_tx_descriptor *tx = NULL;
 
-       if (test_bit(STRIPE_OP_BIOFILL, &pending)) {
+       if (test_bit(STRIPE_OP_BIOFILL, &ops_request)) {
                ops_run_biofill(sh);
                overlap_clear++;
        }
 
-       if (test_bit(STRIPE_OP_COMPUTE_BLK, &pending))
-               tx = ops_run_compute5(sh, pending);
+       if (test_bit(STRIPE_OP_COMPUTE_BLK, &ops_request)) {
+               tx = ops_run_compute5(sh);
+               /* terminate the chain if postxor is not set to be run */
+               if (tx && !test_bit(STRIPE_OP_POSTXOR, &ops_request))
+                       async_tx_ack(tx);
+       }
 
-       if (test_bit(STRIPE_OP_PREXOR, &pending))
+       if (test_bit(STRIPE_OP_PREXOR, &ops_request))
                tx = ops_run_prexor(sh, tx);
 
-       if (test_bit(STRIPE_OP_BIODRAIN, &pending)) {
+       if (test_bit(STRIPE_OP_BIODRAIN, &ops_request)) {
                tx = ops_run_biodrain(sh, tx);
                overlap_clear++;
        }
 
-       if (test_bit(STRIPE_OP_POSTXOR, &pending))
+       if (test_bit(STRIPE_OP_POSTXOR, &ops_request))
                ops_run_postxor(sh, tx);
 
-       if (test_bit(STRIPE_OP_CHECK, &pending))
+       if (test_bit(STRIPE_OP_CHECK, &ops_request))
                ops_run_check(sh);
 
-       if (test_bit(STRIPE_OP_IO, &pending))
-               ops_run_io(sh);
-
        if (overlap_clear)
                for (i = disks; i--; ) {
                        struct r5dev *dev = &sh->dev[i];
@@ -991,14 +945,16 @@ static int resize_stripes(raid5_conf_t *conf, int newsize)
        struct stripe_head *osh, *nsh;
        LIST_HEAD(newstripes);
        struct disk_info *ndisks;
-       int err = 0;
+       int err;
        struct kmem_cache *sc;
        int i;
 
        if (newsize <= conf->pool_size)
                return 0; /* never bother to shrink */
 
-       md_allow_write(conf->mddev);
+       err = md_allow_write(conf->mddev);
+       if (err)
+               return err;
 
        /* Step 1 */
        sc = kmem_cache_create(conf->cache_name[1-conf->active_name],
@@ -1139,10 +1095,12 @@ static void raid5_end_read_request(struct bio * bi, int error)
                set_bit(R5_UPTODATE, &sh->dev[i].flags);
                if (test_bit(R5_ReadError, &sh->dev[i].flags)) {
                        rdev = conf->disks[i].rdev;
-                       printk(KERN_INFO "raid5:%s: read error corrected (%lu sectors at %llu on %s)\n",
-                              mdname(conf->mddev), STRIPE_SECTORS,
-                              (unsigned long long)sh->sector + rdev->data_offset,
-                              bdevname(rdev->bdev, b));
+                       printk_rl(KERN_INFO "raid5:%s: read error corrected"
+                                 " (%lu sectors at %llu on %s)\n",
+                                 mdname(conf->mddev), STRIPE_SECTORS,
+                                 (unsigned long long)(sh->sector
+                                                      + rdev->data_offset),
+                                 bdevname(rdev->bdev, b));
                        clear_bit(R5_ReadError, &sh->dev[i].flags);
                        clear_bit(R5_ReWrite, &sh->dev[i].flags);
                }
@@ -1156,16 +1114,22 @@ static void raid5_end_read_request(struct bio * bi, int error)
                clear_bit(R5_UPTODATE, &sh->dev[i].flags);
                atomic_inc(&rdev->read_errors);
                if (conf->mddev->degraded)
-                       printk(KERN_WARNING "raid5:%s: read error not correctable (sector %llu on %s).\n",
-                              mdname(conf->mddev),
-                              (unsigned long long)sh->sector + rdev->data_offset,
-                              bdn);
+                       printk_rl(KERN_WARNING
+                                 "raid5:%s: read error not correctable "
+                                 "(sector %llu on %s).\n",
+                                 mdname(conf->mddev),
+                                 (unsigned long long)(sh->sector
+                                                      + rdev->data_offset),
+                                 bdn);
                else if (test_bit(R5_ReWrite, &sh->dev[i].flags))
                        /* Oh, no!!! */
-                       printk(KERN_WARNING "raid5:%s: read error NOT corrected!! (sector %llu on %s).\n",
-                              mdname(conf->mddev),
-                              (unsigned long long)sh->sector + rdev->data_offset,
-                              bdn);
+                       printk_rl(KERN_WARNING
+                                 "raid5:%s: read error NOT corrected!! "
+                                 "(sector %llu on %s).\n",
+                                 mdname(conf->mddev),
+                                 (unsigned long long)(sh->sector
+                                                      + rdev->data_offset),
+                                 bdn);
                else if (atomic_read(&rdev->read_errors)
                         > conf->max_nr_stripes)
                        printk(KERN_WARNING
@@ -1254,12 +1218,12 @@ static void error(mddev_t *mddev, mdk_rdev_t *rdev)
                        /*
                         * if recovery was running, make sure it aborts.
                         */
-                       set_bit(MD_RECOVERY_ERR, &mddev->recovery);
+                       set_bit(MD_RECOVERY_INTR, &mddev->recovery);
                }
                set_bit(Faulty, &rdev->flags);
                printk (KERN_ALERT
-                       "raid5: Disk failure on %s, disabling device."
-                       " Operation continuing on %d devices\n",
+                       "raid5: Disk failure on %s, disabling device.\n"
+                       "raid5: Operation continuing on %d devices.\n",
                        bdevname(rdev->bdev,b), conf->raid_disks - mddev->degraded);
        }
 }
@@ -1689,11 +1653,11 @@ static void compute_block_2(struct stripe_head *sh, int dd_idx1, int dd_idx2)
        }
 }
 
-static int
-handle_write_operations5(struct stripe_head *sh, int rcw, int expand)
+static void
+schedule_reconstruction5(struct stripe_head *sh, struct stripe_head_state *s,
+                        int rcw, int expand)
 {
        int i, pd_idx = sh->pd_idx, disks = sh->disks;
-       int locked = 0;
 
        if (rcw) {
                /* if we are not expanding this is a proper write request, and
@@ -1701,50 +1665,48 @@ handle_write_operations5(struct stripe_head *sh, int rcw, int expand)
                 * stripe cache
                 */
                if (!expand) {
-                       set_bit(STRIPE_OP_BIODRAIN, &sh->ops.pending);
-                       sh->ops.count++;
-               }
+                       sh->reconstruct_state = reconstruct_state_drain_run;
+                       set_bit(STRIPE_OP_BIODRAIN, &s->ops_request);
+               } else
+                       sh->reconstruct_state = reconstruct_state_run;
 
-               set_bit(STRIPE_OP_POSTXOR, &sh->ops.pending);
-               sh->ops.count++;
+               set_bit(STRIPE_OP_POSTXOR, &s->ops_request);
 
                for (i = disks; i--; ) {
                        struct r5dev *dev = &sh->dev[i];
 
                        if (dev->towrite) {
                                set_bit(R5_LOCKED, &dev->flags);
+                               set_bit(R5_Wantdrain, &dev->flags);
                                if (!expand)
                                        clear_bit(R5_UPTODATE, &dev->flags);
-                               locked++;
+                               s->locked++;
                        }
                }
+               if (s->locked + 1 == disks)
+                       if (!test_and_set_bit(STRIPE_FULL_WRITE, &sh->state))
+                               atomic_inc(&sh->raid_conf->pending_full_writes);
        } else {
                BUG_ON(!(test_bit(R5_UPTODATE, &sh->dev[pd_idx].flags) ||
                        test_bit(R5_Wantcompute, &sh->dev[pd_idx].flags)));
 
-               set_bit(STRIPE_OP_PREXOR, &sh->ops.pending);
-               set_bit(STRIPE_OP_BIODRAIN, &sh->ops.pending);
-               set_bit(STRIPE_OP_POSTXOR, &sh->ops.pending);
-
-               sh->ops.count += 3;
+               sh->reconstruct_state = reconstruct_state_prexor_drain_run;
+               set_bit(STRIPE_OP_PREXOR, &s->ops_request);
+               set_bit(STRIPE_OP_BIODRAIN, &s->ops_request);
+               set_bit(STRIPE_OP_POSTXOR, &s->ops_request);
 
                for (i = disks; i--; ) {
                        struct r5dev *dev = &sh->dev[i];
                        if (i == pd_idx)
                                continue;
 
-                       /* For a read-modify write there may be blocks that are
-                        * locked for reading while others are ready to be
-                        * written so we distinguish these blocks by the
-                        * R5_Wantprexor bit
-                        */
                        if (dev->towrite &&
                            (test_bit(R5_UPTODATE, &dev->flags) ||
-                           test_bit(R5_Wantcompute, &dev->flags))) {
-                               set_bit(R5_Wantprexor, &dev->flags);
+                            test_bit(R5_Wantcompute, &dev->flags))) {
+                               set_bit(R5_Wantdrain, &dev->flags);
                                set_bit(R5_LOCKED, &dev->flags);
                                clear_bit(R5_UPTODATE, &dev->flags);
-                               locked++;
+                               s->locked++;
                        }
                }
        }
@@ -1754,13 +1716,11 @@ handle_write_operations5(struct stripe_head *sh, int rcw, int expand)
         */
        set_bit(R5_LOCKED, &sh->dev[pd_idx].flags);
        clear_bit(R5_UPTODATE, &sh->dev[pd_idx].flags);
-       locked++;
+       s->locked++;
 
-       pr_debug("%s: stripe %llu locked: %d pending: %lx\n",
-               __FUNCTION__, (unsigned long long)sh->sector,
-               locked, sh->ops.pending);
-
-       return locked;
+       pr_debug("%s: stripe %llu locked: %d ops_request: %lx\n",
+               __func__, (unsigned long long)sh->sector,
+               s->locked, s->ops_request);
 }
 
 /*
@@ -1799,7 +1759,7 @@ static int add_stripe_bio(struct stripe_head *sh, struct bio *bi, int dd_idx, in
        if (*bip)
                bi->bi_next = *bip;
        *bip = bi;
-       bi->bi_phys_segments ++;
+       bi->bi_phys_segments++;
        spin_unlock_irq(&conf->device_lock);
        spin_unlock(&sh->lock);
 
@@ -1859,7 +1819,7 @@ static int stripe_to_pdidx(sector_t stripe, raid5_conf_t *conf, int disks)
 }
 
 static void
-handle_requests_to_failed_array(raid5_conf_t *conf, struct stripe_head *sh,
+handle_failed_stripe(raid5_conf_t *conf, struct stripe_head *sh,
                                struct stripe_head_state *s, int disks,
                                struct bio **return_bi)
 {
@@ -1893,7 +1853,7 @@ handle_requests_to_failed_array(raid5_conf_t *conf, struct stripe_head *sh,
                        sh->dev[i].sector + STRIPE_SECTORS) {
                        struct bio *nextbi = r5_next_bio(bi, sh->dev[i].sector);
                        clear_bit(BIO_UPTODATE, &bi->bi_flags);
-                       if (--bi->bi_phys_segments == 0) {
+                       if (!raid5_dec_bi_phys_segments(bi)) {
                                md_write_end(conf->mddev);
                                bi->bi_next = *return_bi;
                                *return_bi = bi;
@@ -1908,7 +1868,7 @@ handle_requests_to_failed_array(raid5_conf_t *conf, struct stripe_head *sh,
                       sh->dev[i].sector + STRIPE_SECTORS) {
                        struct bio *bi2 = r5_next_bio(bi, sh->dev[i].sector);
                        clear_bit(BIO_UPTODATE, &bi->bi_flags);
-                       if (--bi->bi_phys_segments == 0) {
+                       if (!raid5_dec_bi_phys_segments(bi)) {
                                md_write_end(conf->mddev);
                                bi->bi_next = *return_bi;
                                *return_bi = bi;
@@ -1932,7 +1892,7 @@ handle_requests_to_failed_array(raid5_conf_t *conf, struct stripe_head *sh,
                                struct bio *nextbi =
                                        r5_next_bio(bi, sh->dev[i].sector);
                                clear_bit(BIO_UPTODATE, &bi->bi_flags);
-                               if (--bi->bi_phys_segments == 0) {
+                               if (!raid5_dec_bi_phys_segments(bi)) {
                                        bi->bi_next = *return_bi;
                                        *return_bi = bi;
                                }
@@ -1945,49 +1905,43 @@ handle_requests_to_failed_array(raid5_conf_t *conf, struct stripe_head *sh,
                                        STRIPE_SECTORS, 0, 0);
        }
 
+       if (test_and_clear_bit(STRIPE_FULL_WRITE, &sh->state))
+               if (atomic_dec_and_test(&conf->pending_full_writes))
+                       md_wakeup_thread(conf->mddev->thread);
 }
 
-/* __handle_issuing_new_read_requests5 - returns 0 if there are no more disks
- * to process
+/* fetch_block5 - checks the given member device to see if its data needs
+ * to be read or computed to satisfy a request.
+ *
+ * Returns 1 when no more member devices need to be checked, otherwise returns
+ * 0 to tell the loop in handle_stripe_fill5 to continue
  */
-static int __handle_issuing_new_read_requests5(struct stripe_head *sh,
-                       struct stripe_head_state *s, int disk_idx, int disks)
+static int fetch_block5(struct stripe_head *sh, struct stripe_head_state *s,
+                       int disk_idx, int disks)
 {
        struct r5dev *dev = &sh->dev[disk_idx];
        struct r5dev *failed_dev = &sh->dev[s->failed_num];
 
-       /* don't schedule compute operations or reads on the parity block while
-        * a check is in flight
-        */
-       if ((disk_idx == sh->pd_idx) &&
-            test_bit(STRIPE_OP_CHECK, &sh->ops.pending))
-               return ~0;
-
        /* is the data in this block needed, and can we get it? */
        if (!test_bit(R5_LOCKED, &dev->flags) &&
-           !test_bit(R5_UPTODATE, &dev->flags) && (dev->toread ||
-           (dev->towrite && !test_bit(R5_OVERWRITE, &dev->flags)) ||
-            s->syncing || s->expanding || (s->failed &&
-            (failed_dev->toread || (failed_dev->towrite &&
-            !test_bit(R5_OVERWRITE, &failed_dev->flags)
-            ))))) {
-               /* 1/ We would like to get this block, possibly by computing it,
-                * but we might not be able to.
-                *
-                * 2/ Since parity check operations potentially make the parity
-                * block !uptodate it will need to be refreshed before any
-                * compute operations on data disks are scheduled.
-                *
-                * 3/ We hold off parity block re-reads until check operations
-                * have quiesced.
+           !test_bit(R5_UPTODATE, &dev->flags) &&
+           (dev->toread ||
+            (dev->towrite && !test_bit(R5_OVERWRITE, &dev->flags)) ||
+            s->syncing || s->expanding ||
+            (s->failed &&
+             (failed_dev->toread ||
+              (failed_dev->towrite &&
+               !test_bit(R5_OVERWRITE, &failed_dev->flags)))))) {
+               /* We would like to get this block, possibly by computing it,
+                * otherwise read it if the backing disk is insync
                 */
                if ((s->uptodate == disks - 1) &&
-                   !test_bit(STRIPE_OP_CHECK, &sh->ops.pending)) {
-                       set_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.pending);
+                   (s->failed && disk_idx == s->failed_num)) {
+                       set_bit(STRIPE_COMPUTE_RUN, &sh->state);
+                       set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
                        set_bit(R5_Wantcompute, &dev->flags);
                        sh->ops.target = disk_idx;
                        s->req_compute = 1;
-                       sh->ops.count++;
                        /* Careful: from this point on 'uptodate' is in the eye
                         * of raid5_run_ops which services 'compute' operations
                         * before writes. R5_Wantcompute flags a block that will
@@ -1995,58 +1949,40 @@ static int __handle_issuing_new_read_requests5(struct stripe_head *sh,
                         * subsequent operation.
                         */
                        s->uptodate++;
-                       return 0; /* uptodate + compute == disks */
-               } else if ((s->uptodate < disks - 1) &&
-                       test_bit(R5_Insync, &dev->flags)) {
-                       /* Note: we hold off compute operations while checks are
-                        * in flight, but we still prefer 'compute' over 'read'
-                        * hence we only read if (uptodate < * disks-1)
-                        */
+                       return 1; /* uptodate + compute == disks */
+               } else if (test_bit(R5_Insync, &dev->flags)) {
                        set_bit(R5_LOCKED, &dev->flags);
                        set_bit(R5_Wantread, &dev->flags);
-                       if (!test_and_set_bit(STRIPE_OP_IO, &sh->ops.pending))
-                               sh->ops.count++;
                        s->locked++;
                        pr_debug("Reading block %d (sync=%d)\n", disk_idx,
                                s->syncing);
                }
        }
 
-       return ~0;
+       return 0;
 }
 
-static void handle_issuing_new_read_requests5(struct stripe_head *sh,
+/**
+ * handle_stripe_fill5 - read or compute data to satisfy pending requests.
+ */
+static void handle_stripe_fill5(struct stripe_head *sh,
                        struct stripe_head_state *s, int disks)
 {
        int i;
 
-       /* Clear completed compute operations.  Parity recovery
-        * (STRIPE_OP_MOD_REPAIR_PD) implies a write-back which is handled
-        * later on in this routine
-        */
-       if (test_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.complete) &&
-               !test_bit(STRIPE_OP_MOD_REPAIR_PD, &sh->ops.pending)) {
-               clear_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.complete);
-               clear_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.ack);
-               clear_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.pending);
-       }
-
        /* look for blocks to read/compute, skip this if a compute
         * is already in flight, or if the stripe contents are in the
         * midst of changing due to a write
         */
-       if (!test_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.pending) &&
-               !test_bit(STRIPE_OP_PREXOR, &sh->ops.pending) &&
-               !test_bit(STRIPE_OP_POSTXOR, &sh->ops.pending)) {
+       if (!test_bit(STRIPE_COMPUTE_RUN, &sh->state) && !sh->check_state &&
+           !sh->reconstruct_state)
                for (i = disks; i--; )
-                       if (__handle_issuing_new_read_requests5(
-                               sh, s, i, disks) == 0)
+                       if (fetch_block5(sh, s, i, disks))
                                break;
-       }
        set_bit(STRIPE_HANDLE, &sh->state);
 }
 
-static void handle_issuing_new_read_requests6(struct stripe_head *sh,
+static void handle_stripe_fill6(struct stripe_head *sh,
                        struct stripe_head_state *s, struct r6_state *r6s,
                        int disks)
 {
@@ -2067,7 +2003,9 @@ static void handle_issuing_new_read_requests6(struct stripe_head *sh,
                        /* we would like to get this block, possibly
                         * by computing it, but we might not be able to
                         */
-                       if (s->uptodate == disks-1) {
+                       if ((s->uptodate == disks - 1) &&
+                           (s->failed && (i == r6s->failed_num[0] ||
+                                          i == r6s->failed_num[1]))) {
                                pr_debug("Computing stripe %llu block %d\n",
                                       (unsigned long long)sh->sector, i);
                                compute_block_1(sh, i, 0);
@@ -2103,12 +2041,12 @@ static void handle_issuing_new_read_requests6(struct stripe_head *sh,
 }
 
 
-/* handle_completed_write_requests
+/* handle_stripe_clean_event
  * any written block on an uptodate or failed drive can be returned.
  * Note that if we 'wrote' to a failed drive, it will be UPTODATE, but
  * never LOCKED, so we don't need to test 'failed' directly.
  */
-static void handle_completed_write_requests(raid5_conf_t *conf,
+static void handle_stripe_clean_event(raid5_conf_t *conf,
        struct stripe_head *sh, int disks, struct bio **return_bi)
 {
        int i;
@@ -2129,7 +2067,7 @@ static void handle_completed_write_requests(raid5_conf_t *conf,
                                while (wbi && wbi->bi_sector <
                                        dev->sector + STRIPE_SECTORS) {
                                        wbi2 = r5_next_bio(wbi, dev->sector);
-                                       if (--wbi->bi_phys_segments == 0) {
+                                       if (!raid5_dec_bi_phys_segments(wbi)) {
                                                md_write_end(conf->mddev);
                                                wbi->bi_next = *return_bi;
                                                *return_bi = wbi;
@@ -2147,9 +2085,13 @@ static void handle_completed_write_requests(raid5_conf_t *conf,
                                                        0);
                        }
                }
+
+       if (test_and_clear_bit(STRIPE_FULL_WRITE, &sh->state))
+               if (atomic_dec_and_test(&conf->pending_full_writes))
+                       md_wakeup_thread(conf->mddev->thread);
 }
 
-static void handle_issuing_new_write_requests5(raid5_conf_t *conf,
+static void handle_stripe_dirtying5(raid5_conf_t *conf,
                struct stripe_head *sh, struct stripe_head_state *s, int disks)
 {
        int rmw = 0, rcw = 0, i;
@@ -2193,9 +2135,6 @@ static void handle_issuing_new_write_requests5(raid5_conf_t *conf,
                                                "%d for r-m-w\n", i);
                                        set_bit(R5_LOCKED, &dev->flags);
                                        set_bit(R5_Wantread, &dev->flags);
-                                       if (!test_and_set_bit(
-                                               STRIPE_OP_IO, &sh->ops.pending))
-                                               sh->ops.count++;
                                        s->locked++;
                                } else {
                                        set_bit(STRIPE_DELAYED, &sh->state);
@@ -2219,9 +2158,6 @@ static void handle_issuing_new_write_requests5(raid5_conf_t *conf,
                                                "%d for Reconstruct\n", i);
                                        set_bit(R5_LOCKED, &dev->flags);
                                        set_bit(R5_Wantread, &dev->flags);
-                                       if (!test_and_set_bit(
-                                               STRIPE_OP_IO, &sh->ops.pending))
-                                               sh->ops.count++;
                                        s->locked++;
                                } else {
                                        set_bit(STRIPE_DELAYED, &sh->state);
@@ -2239,14 +2175,13 @@ static void handle_issuing_new_write_requests5(raid5_conf_t *conf,
         * simultaneously.  If this is not the case then new writes need to be
         * held off until the compute completes.
         */
-       if ((s->req_compute ||
-           !test_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.pending)) &&
-               (s->locked == 0 && (rcw == 0 || rmw == 0) &&
-               !test_bit(STRIPE_BIT_DELAY, &sh->state)))
-               s->locked += handle_write_operations5(sh, rcw == 0, 0);
+       if ((s->req_compute || !test_bit(STRIPE_COMPUTE_RUN, &sh->state)) &&
+           (s->locked == 0 && (rcw == 0 || rmw == 0) &&
+           !test_bit(STRIPE_BIT_DELAY, &sh->state)))
+               schedule_reconstruction5(sh, s, rcw == 0, 0);
 }
 
-static void handle_issuing_new_write_requests6(raid5_conf_t *conf,
+static void handle_stripe_dirtying6(raid5_conf_t *conf,
                struct stripe_head *sh, struct stripe_head_state *s,
                struct r6_state *r6s, int disks)
 {
@@ -2331,6 +2266,9 @@ static void handle_issuing_new_write_requests6(raid5_conf_t *conf,
                                s->locked++;
                                set_bit(R5_Wantwrite, &sh->dev[i].flags);
                        }
+               if (s->locked == disks)
+                       if (!test_and_set_bit(STRIPE_FULL_WRITE, &sh->state))
+                               atomic_inc(&conf->pending_full_writes);
                /* after a RECONSTRUCT_WRITE, the stripe MUST be in-sync */
                set_bit(STRIPE_INSYNC, &sh->state);
 
@@ -2346,84 +2284,86 @@ static void handle_issuing_new_write_requests6(raid5_conf_t *conf,
 static void handle_parity_checks5(raid5_conf_t *conf, struct stripe_head *sh,
                                struct stripe_head_state *s, int disks)
 {
+       struct r5dev *dev = NULL;
+
        set_bit(STRIPE_HANDLE, &sh->state);
-       /* Take one of the following actions:
-        * 1/ start a check parity operation if (uptodate == disks)
-        * 2/ finish a check parity operation and act on the result
-        * 3/ skip to the writeback section if we previously
-        *    initiated a recovery operation
-        */
-       if (s->failed == 0 &&
-           !test_bit(STRIPE_OP_MOD_REPAIR_PD, &sh->ops.pending)) {
-               if (!test_and_set_bit(STRIPE_OP_CHECK, &sh->ops.pending)) {
+
+       switch (sh->check_state) {
+       case check_state_idle:
+               /* start a new check operation if there are no failures */
+               if (s->failed == 0) {
                        BUG_ON(s->uptodate != disks);
+                       sh->check_state = check_state_run;
+                       set_bit(STRIPE_OP_CHECK, &s->ops_request);
                        clear_bit(R5_UPTODATE, &sh->dev[sh->pd_idx].flags);
-                       sh->ops.count++;
                        s->uptodate--;
-               } else if (
-                      test_and_clear_bit(STRIPE_OP_CHECK, &sh->ops.complete)) {
-                       clear_bit(STRIPE_OP_CHECK, &sh->ops.ack);
-                       clear_bit(STRIPE_OP_CHECK, &sh->ops.pending);
-
-                       if (sh->ops.zero_sum_result == 0)
-                               /* parity is correct (on disc,
-                                * not in buffer any more)
-                                */
-                               set_bit(STRIPE_INSYNC, &sh->state);
-                       else {
-                               conf->mddev->resync_mismatches +=
-                                       STRIPE_SECTORS;
-                               if (test_bit(
-                                    MD_RECOVERY_CHECK, &conf->mddev->recovery))
-                                       /* don't try to repair!! */
-                                       set_bit(STRIPE_INSYNC, &sh->state);
-                               else {
-                                       set_bit(STRIPE_OP_COMPUTE_BLK,
-                                               &sh->ops.pending);
-                                       set_bit(STRIPE_OP_MOD_REPAIR_PD,
-                                               &sh->ops.pending);
-                                       set_bit(R5_Wantcompute,
-                                               &sh->dev[sh->pd_idx].flags);
-                                       sh->ops.target = sh->pd_idx;
-                                       sh->ops.count++;
-                                       s->uptodate++;
-                               }
-                       }
+                       break;
                }
-       }
-
-       /* check if we can clear a parity disk reconstruct */
-       if (test_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.complete) &&
-               test_bit(STRIPE_OP_MOD_REPAIR_PD, &sh->ops.pending)) {
-
-               clear_bit(STRIPE_OP_MOD_REPAIR_PD, &sh->ops.pending);
-               clear_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.complete);
-               clear_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.ack);
-               clear_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.pending);
-       }
+               dev = &sh->dev[s->failed_num];
+               /* fall through */
+       case check_state_compute_result:
+               sh->check_state = check_state_idle;
+               if (!dev)
+                       dev = &sh->dev[sh->pd_idx];
+
+               /* check that a write has not made the stripe insync */
+               if (test_bit(STRIPE_INSYNC, &sh->state))
+                       break;
 
-       /* Wait for check parity and compute block operations to complete
-        * before write-back
-        */
-       if (!test_bit(STRIPE_INSYNC, &sh->state) &&
-               !test_bit(STRIPE_OP_CHECK, &sh->ops.pending) &&
-               !test_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.pending)) {
-               struct r5dev *dev;
                /* either failed parity check, or recovery is happening */
-               if (s->failed == 0)
-                       s->failed_num = sh->pd_idx;
-               dev = &sh->dev[s->failed_num];
                BUG_ON(!test_bit(R5_UPTODATE, &dev->flags));
                BUG_ON(s->uptodate != disks);
 
                set_bit(R5_LOCKED, &dev->flags);
+               s->locked++;
                set_bit(R5_Wantwrite, &dev->flags);
-               if (!test_and_set_bit(STRIPE_OP_IO, &sh->ops.pending))
-                       sh->ops.count++;
 
                clear_bit(STRIPE_DEGRADED, &sh->state);
-               s->locked++;
                set_bit(STRIPE_INSYNC, &sh->state);
+               break;
+       case check_state_run:
+               break; /* we will be called again upon completion */
+       case check_state_check_result:
+               sh->check_state = check_state_idle;
+
+               /* if a failure occurred during the check operation, leave
+                * STRIPE_INSYNC not set and let the stripe be handled again
+                */
+               if (s->failed)
+                       break;
+
+               /* handle a successful check operation, if parity is correct
+                * we are done.  Otherwise update the mismatch count and repair
+                * parity if !MD_RECOVERY_CHECK
+                */
+               if (sh->ops.zero_sum_result == 0)
+                       /* parity is correct (on disc,
+                        * not in buffer any more)
+                        */
+                       set_bit(STRIPE_INSYNC, &sh->state);
+               else {
+                       conf->mddev->resync_mismatches += STRIPE_SECTORS;
+                       if (test_bit(MD_RECOVERY_CHECK, &conf->mddev->recovery))
+                               /* don't try to repair!! */
+                               set_bit(STRIPE_INSYNC, &sh->state);
+                       else {
+                               sh->check_state = check_state_compute_run;
+                               set_bit(STRIPE_COMPUTE_RUN, &sh->state);
+                               set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
+                               set_bit(R5_Wantcompute,
+                                       &sh->dev[sh->pd_idx].flags);
+                               sh->ops.target = sh->pd_idx;
+                               s->uptodate++;
+                       }
+               }
+               break;
+       case check_state_compute_run:
+               break;
+       default:
+               printk(KERN_ERR "%s: unknown check_state: %d sector: %llu\n",
+                      __func__, sh->check_state,
+                      (unsigned long long) sh->sector);
+               BUG();
        }
 }
 
@@ -2583,6 +2523,7 @@ static void handle_stripe_expansion(raid5_conf_t *conf, struct stripe_head *sh,
        }
 }
 
+
 /*
  * handle_stripe - do things to a stripe.
  *
@@ -2600,20 +2541,21 @@ static void handle_stripe_expansion(raid5_conf_t *conf, struct stripe_head *sh,
  *
  */
 
-static void handle_stripe5(struct stripe_head *sh)
+static bool handle_stripe5(struct stripe_head *sh)
 {
        raid5_conf_t *conf = sh->raid_conf;
        int disks = sh->disks, i;
        struct bio *return_bi = NULL;
        struct stripe_head_state s;
        struct r5dev *dev;
-       unsigned long pending = 0;
+       mdk_rdev_t *blocked_rdev = NULL;
+       int prexor;
 
        memset(&s, 0, sizeof(s));
-       pr_debug("handling stripe %llu, state=%#lx cnt=%d, pd_idx=%d "
-               "ops=%lx:%lx:%lx\n", (unsigned long long)sh->sector, sh->state,
-               atomic_read(&sh->count), sh->pd_idx,
-               sh->ops.pending, sh->ops.ack, sh->ops.complete);
+       pr_debug("handling stripe %llu, state=%#lx cnt=%d, pd_idx=%d check:%d "
+                "reconstruct:%d\n", (unsigned long long)sh->sector, sh->state,
+                atomic_read(&sh->count), sh->pd_idx, sh->check_state,
+                sh->reconstruct_state);
 
        spin_lock(&sh->lock);
        clear_bit(STRIPE_HANDLE, &sh->state);
@@ -2622,15 +2564,8 @@ static void handle_stripe5(struct stripe_head *sh)
        s.syncing = test_bit(STRIPE_SYNCING, &sh->state);
        s.expanding = test_bit(STRIPE_EXPAND_SOURCE, &sh->state);
        s.expanded = test_bit(STRIPE_EXPAND_READY, &sh->state);
-       /* Now to look around and see what can be done */
-
-       /* clean-up completed biofill operations */
-       if (test_bit(STRIPE_OP_BIOFILL, &sh->ops.complete)) {
-               clear_bit(STRIPE_OP_BIOFILL, &sh->ops.pending);
-               clear_bit(STRIPE_OP_BIOFILL, &sh->ops.ack);
-               clear_bit(STRIPE_OP_BIOFILL, &sh->ops.complete);
-       }
 
+       /* Now to look around and see what can be done */
        rcu_read_lock();
        for (i=disks; i--; ) {
                mdk_rdev_t *rdev;
@@ -2644,10 +2579,10 @@ static void handle_stripe5(struct stripe_head *sh)
                /* maybe we can request a biofill operation
                 *
                 * new wantfill requests are only permitted while
-                * STRIPE_OP_BIOFILL is clear
+                * ops_complete_biofill is guaranteed to be inactive
                 */
                if (test_bit(R5_UPTODATE, &dev->flags) && dev->toread &&
-                       !test_bit(STRIPE_OP_BIOFILL, &sh->ops.pending))
+                   !test_bit(STRIPE_BIOFILL_RUN, &sh->state))
                        set_bit(R5_Wantfill, &dev->flags);
 
                /* now count some things */
@@ -2667,6 +2602,11 @@ static void handle_stripe5(struct stripe_head *sh)
                if (dev->written)
                        s.written++;
                rdev = rcu_dereference(conf->disks[i].rdev);
+               if (blocked_rdev == NULL &&
+                   rdev && unlikely(test_bit(Blocked, &rdev->flags))) {
+                       blocked_rdev = rdev;
+                       atomic_inc(&rdev->nr_pending);
+               }
                if (!rdev || !test_bit(In_sync, &rdev->flags)) {
                        /* The ReadError flag will just be confusing now */
                        clear_bit(R5_ReadError, &dev->flags);
@@ -2681,8 +2621,21 @@ static void handle_stripe5(struct stripe_head *sh)
        }
        rcu_read_unlock();
 
-       if (s.to_fill && !test_and_set_bit(STRIPE_OP_BIOFILL, &sh->ops.pending))
-               sh->ops.count++;
+       if (unlikely(blocked_rdev)) {
+               if (s.syncing || s.expanding || s.expanded ||
+                   s.to_write || s.written) {
+                       set_bit(STRIPE_HANDLE, &sh->state);
+                       goto unlock;
+               }
+               /* There is nothing for the blocked_rdev to block */
+               rdev_dec_pending(blocked_rdev, conf->mddev);
+               blocked_rdev = NULL;
+       }
+
+       if (s.to_fill && !test_bit(STRIPE_BIOFILL_RUN, &sh->state)) {
+               set_bit(STRIPE_OP_BIOFILL, &s.ops_request);
+               set_bit(STRIPE_BIOFILL_RUN, &sh->state);
+       }
 
        pr_debug("locked=%d uptodate=%d to_read=%d"
                " to_write=%d failed=%d failed_num=%d\n",
@@ -2692,8 +2645,7 @@ static void handle_stripe5(struct stripe_head *sh)
         * need to be failed
         */
        if (s.failed > 1 && s.to_read+s.to_write+s.written)
-               handle_requests_to_failed_array(conf, sh, &s, disks,
-                                               &return_bi);
+               handle_failed_stripe(conf, sh, &s, disks, &return_bi);
        if (s.failed > 1 && s.syncing) {
                md_done_sync(conf->mddev, STRIPE_SECTORS,0);
                clear_bit(STRIPE_SYNCING, &sh->state);
@@ -2709,46 +2661,25 @@ static void handle_stripe5(struct stripe_head *sh)
               !test_bit(R5_LOCKED, &dev->flags) &&
               test_bit(R5_UPTODATE, &dev->flags)) ||
               (s.failed == 1 && s.failed_num == sh->pd_idx)))
-               handle_completed_write_requests(conf, sh, disks, &return_bi);
+               handle_stripe_clean_event(conf, sh, disks, &return_bi);
 
        /* Now we might consider reading some blocks, either to check/generate
         * parity, or to satisfy requests
         * or to load a block that is being partially written.
         */
        if (s.to_read || s.non_overwrite ||
-           (s.syncing && (s.uptodate + s.compute < disks)) || s.expanding ||
-           test_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.pending))
-               handle_issuing_new_read_requests5(sh, &s, disks);
+           (s.syncing && (s.uptodate + s.compute < disks)) || s.expanding)
+               handle_stripe_fill5(sh, &s, disks);
 
        /* Now we check to see if any write operations have recently
         * completed
         */
-
-       /* leave prexor set until postxor is done, allows us to distinguish
-        * a rmw from a rcw during biodrain
-        */
-       if (test_bit(STRIPE_OP_PREXOR, &sh->ops.complete) &&
-               test_bit(STRIPE_OP_POSTXOR, &sh->ops.complete)) {
-
-               clear_bit(STRIPE_OP_PREXOR, &sh->ops.complete);
-               clear_bit(STRIPE_OP_PREXOR, &sh->ops.ack);
-               clear_bit(STRIPE_OP_PREXOR, &sh->ops.pending);
-
-               for (i = disks; i--; )
-                       clear_bit(R5_Wantprexor, &sh->dev[i].flags);
-       }
-
-       /* if only POSTXOR is set then this is an 'expand' postxor */
-       if (test_bit(STRIPE_OP_BIODRAIN, &sh->ops.complete) &&
-               test_bit(STRIPE_OP_POSTXOR, &sh->ops.complete)) {
-
-               clear_bit(STRIPE_OP_BIODRAIN, &sh->ops.complete);
-               clear_bit(STRIPE_OP_BIODRAIN, &sh->ops.ack);
-               clear_bit(STRIPE_OP_BIODRAIN, &sh->ops.pending);
-
-               clear_bit(STRIPE_OP_POSTXOR, &sh->ops.complete);
-               clear_bit(STRIPE_OP_POSTXOR, &sh->ops.ack);
-               clear_bit(STRIPE_OP_POSTXOR, &sh->ops.pending);
+       prexor = 0;
+       if (sh->reconstruct_state == reconstruct_state_prexor_drain_result)
+               prexor = 1;
+       if (sh->reconstruct_state == reconstruct_state_drain_result ||
+           sh->reconstruct_state == reconstruct_state_prexor_drain_result) {
+               sh->reconstruct_state = reconstruct_state_idle;
 
                /* All the 'written' buffers and the parity block are ready to
                 * be written back to disk
@@ -2760,9 +2691,8 @@ static void handle_stripe5(struct stripe_head *sh)
                                (i == sh->pd_idx || dev->written)) {
                                pr_debug("Writing block %d\n", i);
                                set_bit(R5_Wantwrite, &dev->flags);
-                               if (!test_and_set_bit(
-                                   STRIPE_OP_IO, &sh->ops.pending))
-                                       sh->ops.count++;
+                               if (prexor)
+                                       continue;
                                if (!test_bit(R5_Insync, &dev->flags) ||
                                    (i == sh->pd_idx && s.failed == 0))
                                        set_bit(STRIPE_INSYNC, &sh->state);
@@ -2782,20 +2712,18 @@ static void handle_stripe5(struct stripe_head *sh)
         * 2/ A 'check' operation is in flight, as it may clobber the parity
         *    block.
         */
-       if (s.to_write && !test_bit(STRIPE_OP_POSTXOR, &sh->ops.pending) &&
-                         !test_bit(STRIPE_OP_CHECK, &sh->ops.pending))
-               handle_issuing_new_write_requests5(conf, sh, &s, disks);
+       if (s.to_write && !sh->reconstruct_state && !sh->check_state)
+               handle_stripe_dirtying5(conf, sh, &s, disks);
 
        /* maybe we need to check and possibly fix the parity for this stripe
         * Any reads will already have been scheduled, so we just see if enough
         * data is available.  The parity check is held off while parity
         * dependent operations are in flight.
         */
-       if ((s.syncing && s.locked == 0 &&
-            !test_bit(STRIPE_OP_COMPUTE_BLK, &sh->ops.pending) &&
-            !test_bit(STRIPE_INSYNC, &sh->state)) ||
-             test_bit(STRIPE_OP_CHECK, &sh->ops.pending) ||
-             test_bit(STRIPE_OP_MOD_REPAIR_PD, &sh->ops.pending))
+       if (sh->check_state ||
+           (s.syncing && s.locked == 0 &&
+            !test_bit(STRIPE_COMPUTE_RUN, &sh->state) &&
+            !test_bit(STRIPE_INSYNC, &sh->state)))
                handle_parity_checks5(conf, sh, &s, disks);
 
        if (s.syncing && s.locked == 0 && test_bit(STRIPE_INSYNC, &sh->state)) {
@@ -2814,71 +2742,64 @@ static void handle_stripe5(struct stripe_head *sh)
                dev = &sh->dev[s.failed_num];
                if (!test_bit(R5_ReWrite, &dev->flags)) {
                        set_bit(R5_Wantwrite, &dev->flags);
-                       if (!test_and_set_bit(STRIPE_OP_IO, &sh->ops.pending))
-                               sh->ops.count++;
                        set_bit(R5_ReWrite, &dev->flags);
                        set_bit(R5_LOCKED, &dev->flags);
                        s.locked++;
                } else {
                        /* let's read it back */
                        set_bit(R5_Wantread, &dev->flags);
-                       if (!test_and_set_bit(STRIPE_OP_IO, &sh->ops.pending))
-                               sh->ops.count++;
                        set_bit(R5_LOCKED, &dev->flags);
                        s.locked++;
                }
        }
 
-       /* Finish postxor operations initiated by the expansion
-        * process
-        */
-       if (test_bit(STRIPE_OP_POSTXOR, &sh->ops.complete) &&
-               !test_bit(STRIPE_OP_BIODRAIN, &sh->ops.pending)) {
-
+       /* Finish reconstruct operations initiated by the expansion process */
+       if (sh->reconstruct_state == reconstruct_state_result) {
+               sh->reconstruct_state = reconstruct_state_idle;
                clear_bit(STRIPE_EXPANDING, &sh->state);
-
-               clear_bit(STRIPE_OP_POSTXOR, &sh->ops.pending);
-               clear_bit(STRIPE_OP_POSTXOR, &sh->ops.ack);
-               clear_bit(STRIPE_OP_POSTXOR, &sh->ops.complete);
-
                for (i = conf->raid_disks; i--; ) {
                        set_bit(R5_Wantwrite, &sh->dev[i].flags);
-                       if (!test_and_set_bit(STRIPE_OP_IO, &sh->ops.pending))
-                               sh->ops.count++;
+                       set_bit(R5_LOCKED, &sh->dev[i].flags);
+                       s.locked++;
                }
        }
 
        if (s.expanded && test_bit(STRIPE_EXPANDING, &sh->state) &&
-               !test_bit(STRIPE_OP_POSTXOR, &sh->ops.pending)) {
+           !sh->reconstruct_state) {
                /* Need to write out all blocks after computing parity */
                sh->disks = conf->raid_disks;
                sh->pd_idx = stripe_to_pdidx(sh->sector, conf,
                        conf->raid_disks);
-               s.locked += handle_write_operations5(sh, 1, 1);
-       } else if (s.expanded &&
-               !test_bit(STRIPE_OP_POSTXOR, &sh->ops.pending)) {
+               schedule_reconstruction5(sh, &s, 1, 1);
+       } else if (s.expanded && !sh->reconstruct_state && s.locked == 0) {
                clear_bit(STRIPE_EXPAND_READY, &sh->state);
                atomic_dec(&conf->reshape_stripes);
                wake_up(&conf->wait_for_overlap);
                md_done_sync(conf->mddev, STRIPE_SECTORS, 1);
        }
 
-       if (s.expanding && s.locked == 0)
+       if (s.expanding && s.locked == 0 &&
+           !test_bit(STRIPE_COMPUTE_RUN, &sh->state))
                handle_stripe_expansion(conf, sh, NULL);
 
-       if (sh->ops.count)
-               pending = get_stripe_work(sh);
-
+ unlock:
        spin_unlock(&sh->lock);
 
-       if (pending)
-               raid5_run_ops(sh, pending);
+       /* wait for this device to become unblocked */
+       if (unlikely(blocked_rdev))
+               md_wait_for_blocked_rdev(blocked_rdev, conf->mddev);
+
+       if (s.ops_request)
+               raid5_run_ops(sh, s.ops_request);
+
+       ops_run_io(sh, &s);
 
        return_io(return_bi);
 
+       return blocked_rdev == NULL;
 }
 
-static void handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
+static bool handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
 {
        raid6_conf_t *conf = sh->raid_conf;
        int disks = sh->disks;
@@ -2887,6 +2808,7 @@ static void handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
        struct stripe_head_state s;
        struct r6_state r6s;
        struct r5dev *dev, *pdev, *qdev;
+       mdk_rdev_t *blocked_rdev = NULL;
 
        r6s.qd_idx = raid6_next_disk(pd_idx, disks);
        pr_debug("handling stripe %llu, state=%#lx cnt=%d, "
@@ -2926,7 +2848,7 @@ static void handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
                                copy_data(0, rbi, dev->page, dev->sector);
                                rbi2 = r5_next_bio(rbi, dev->sector);
                                spin_lock_irq(&conf->device_lock);
-                               if (--rbi->bi_phys_segments == 0) {
+                               if (!raid5_dec_bi_phys_segments(rbi)) {
                                        rbi->bi_next = return_bi;
                                        return_bi = rbi;
                                }
@@ -2950,6 +2872,11 @@ static void handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
                if (dev->written)
                        s.written++;
                rdev = rcu_dereference(conf->disks[i].rdev);
+               if (blocked_rdev == NULL &&
+                   rdev && unlikely(test_bit(Blocked, &rdev->flags))) {
+                       blocked_rdev = rdev;
+                       atomic_inc(&rdev->nr_pending);
+               }
                if (!rdev || !test_bit(In_sync, &rdev->flags)) {
                        /* The ReadError flag will just be confusing now */
                        clear_bit(R5_ReadError, &dev->flags);
@@ -2964,6 +2891,18 @@ static void handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
                        set_bit(R5_Insync, &dev->flags);
        }
        rcu_read_unlock();
+
+       if (unlikely(blocked_rdev)) {
+               if (s.syncing || s.expanding || s.expanded ||
+                   s.to_write || s.written) {
+                       set_bit(STRIPE_HANDLE, &sh->state);
+                       goto unlock;
+               }
+               /* There is nothing for the blocked_rdev to block */
+               rdev_dec_pending(blocked_rdev, conf->mddev);
+               blocked_rdev = NULL;
+       }
+
        pr_debug("locked=%d uptodate=%d to_read=%d"
               " to_write=%d failed=%d failed_num=%d,%d\n",
               s.locked, s.uptodate, s.to_read, s.to_write, s.failed,
@@ -2972,8 +2911,7 @@ static void handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
         * might need to be failed
         */
        if (s.failed > 2 && s.to_read+s.to_write+s.written)
-               handle_requests_to_failed_array(conf, sh, &s, disks,
-                                               &return_bi);
+               handle_failed_stripe(conf, sh, &s, disks, &return_bi);
        if (s.failed > 2 && s.syncing) {
                md_done_sync(conf->mddev, STRIPE_SECTORS,0);
                clear_bit(STRIPE_SYNCING, &sh->state);
@@ -2998,7 +2936,7 @@ static void handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
             ( r6s.q_failed || ((test_bit(R5_Insync, &qdev->flags)
                             && !test_bit(R5_LOCKED, &qdev->flags)
                             && test_bit(R5_UPTODATE, &qdev->flags)))))
-               handle_completed_write_requests(conf, sh, disks, &return_bi);
+               handle_stripe_clean_event(conf, sh, disks, &return_bi);
 
        /* Now we might consider reading some blocks, either to check/generate
         * parity, or to satisfy requests
@@ -3006,11 +2944,11 @@ static void handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
         */
        if (s.to_read || s.non_overwrite || (s.to_write && s.failed) ||
            (s.syncing && (s.uptodate < disks)) || s.expanding)
-               handle_issuing_new_read_requests6(sh, &s, &r6s, disks);
+               handle_stripe_fill6(sh, &s, &r6s, disks);
 
        /* now to consider writing and what else, if anything should be read */
        if (s.to_write)
-               handle_issuing_new_write_requests6(conf, sh, &s, &r6s, disks);
+               handle_stripe_dirtying6(conf, sh, &s, &r6s, disks);
 
        /* maybe we need to check and possibly fix the parity for this stripe
         * Any reads will already have been scheduled, so we just see if enough
@@ -3065,79 +3003,31 @@ static void handle_stripe6(struct stripe_head *sh, struct page *tmp_page)
                md_done_sync(conf->mddev, STRIPE_SECTORS, 1);
        }
 
-       if (s.expanding && s.locked == 0)
+       if (s.expanding && s.locked == 0 &&
+           !test_bit(STRIPE_COMPUTE_RUN, &sh->state))
                handle_stripe_expansion(conf, sh, &r6s);
 
+ unlock:
        spin_unlock(&sh->lock);
 
-       return_io(return_bi);
-
-       for (i=disks; i-- ;) {
-               int rw;
-               struct bio *bi;
-               mdk_rdev_t *rdev;
-               if (test_and_clear_bit(R5_Wantwrite, &sh->dev[i].flags))
-                       rw = WRITE;
-               else if (test_and_clear_bit(R5_Wantread, &sh->dev[i].flags))
-                       rw = READ;
-               else
-                       continue;
-
-               bi = &sh->dev[i].req;
+       /* wait for this device to become unblocked */
+       if (unlikely(blocked_rdev))
+               md_wait_for_blocked_rdev(blocked_rdev, conf->mddev);
 
-               bi->bi_rw = rw;
-               if (rw == WRITE)
-                       bi->bi_end_io = raid5_end_write_request;
-               else
-                       bi->bi_end_io = raid5_end_read_request;
-
-               rcu_read_lock();
-               rdev = rcu_dereference(conf->disks[i].rdev);
-               if (rdev && test_bit(Faulty, &rdev->flags))
-                       rdev = NULL;
-               if (rdev)
-                       atomic_inc(&rdev->nr_pending);
-               rcu_read_unlock();
+       ops_run_io(sh, &s);
 
-               if (rdev) {
-                       if (s.syncing || s.expanding || s.expanded)
-                               md_sync_acct(rdev->bdev, STRIPE_SECTORS);
+       return_io(return_bi);
 
-                       bi->bi_bdev = rdev->bdev;
-                       pr_debug("for %llu schedule op %ld on disc %d\n",
-                               (unsigned long long)sh->sector, bi->bi_rw, i);
-                       atomic_inc(&sh->count);
-                       bi->bi_sector = sh->sector + rdev->data_offset;
-                       bi->bi_flags = 1 << BIO_UPTODATE;
-                       bi->bi_vcnt = 1;
-                       bi->bi_max_vecs = 1;
-                       bi->bi_idx = 0;
-                       bi->bi_io_vec = &sh->dev[i].vec;
-                       bi->bi_io_vec[0].bv_len = STRIPE_SIZE;
-                       bi->bi_io_vec[0].bv_offset = 0;
-                       bi->bi_size = STRIPE_SIZE;
-                       bi->bi_next = NULL;
-                       if (rw == WRITE &&
-                           test_bit(R5_ReWrite, &sh->dev[i].flags))
-                               atomic_add(STRIPE_SECTORS, &rdev->corrected_errors);
-                       generic_make_request(bi);
-               } else {
-                       if (rw == WRITE)
-                               set_bit(STRIPE_DEGRADED, &sh->state);
-                       pr_debug("skip op %ld on disc %d for sector %llu\n",
-                               bi->bi_rw, i, (unsigned long long)sh->sector);
-                       clear_bit(R5_LOCKED, &sh->dev[i].flags);
-                       set_bit(STRIPE_HANDLE, &sh->state);
-               }
-       }
+       return blocked_rdev == NULL;
 }
 
-static void handle_stripe(struct stripe_head *sh, struct page *tmp_page)
+/* returns true if the stripe was handled */
+static bool handle_stripe(struct stripe_head *sh, struct page *tmp_page)
 {
        if (sh->raid_conf->level == 6)
-               handle_stripe6(sh, tmp_page);
+               return handle_stripe6(sh, tmp_page);
        else
-               handle_stripe5(sh);
+               return handle_stripe5(sh);
 }
 
 
@@ -3153,9 +3043,10 @@ static void raid5_activate_delayed(raid5_conf_t *conf)
                        clear_bit(STRIPE_DELAYED, &sh->state);
                        if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
                                atomic_inc(&conf->preread_active_stripes);
-                       list_add_tail(&sh->lru, &conf->handle_list);
+                       list_add_tail(&sh->lru, &conf->hold_list);
                }
-       }
+       } else
+               blk_plug_device(conf->mddev->queue);
 }
 
 static void activate_bit_delay(raid5_conf_t *conf)
@@ -3235,15 +3126,17 @@ static int raid5_congested(void *data, int bits)
 /* We want read requests to align with chunks where possible,
  * but write requests don't need to.
  */
-static int raid5_mergeable_bvec(struct request_queue *q, struct bio *bio, struct bio_vec *biovec)
+static int raid5_mergeable_bvec(struct request_queue *q,
+                               struct bvec_merge_data *bvm,
+                               struct bio_vec *biovec)
 {
        mddev_t *mddev = q->queuedata;
-       sector_t sector = bio->bi_sector + get_start_sect(bio->bi_bdev);
+       sector_t sector = bvm->bi_sector + get_start_sect(bvm->bi_bdev);
        int max;
        unsigned int chunk_sectors = mddev->chunk_size >> 9;
-       unsigned int bio_sectors = bio->bi_size >> 9;
+       unsigned int bio_sectors = bvm->bi_size >> 9;
 
-       if (bio_data_dir(bio) == WRITE)
+       if ((bvm->bi_rw & 1) == WRITE)
                return biovec->bv_len; /* always allow writes to be mergeable */
 
        max =  (chunk_sectors - ((sector & (chunk_sectors - 1)) + bio_sectors)) << 9;
@@ -3296,8 +3189,11 @@ static struct bio *remove_bio_from_retry(raid5_conf_t *conf)
        if(bi) {
                conf->retry_read_aligned_list = bi->bi_next;
                bi->bi_next = NULL;
+               /*
+                * this sets the active strip count to 1 and the processed
+                * strip count to zero (upper 8 bits)
+                */
                bi->bi_phys_segments = 1; /* biased count of active stripes */
-               bi->bi_hw_segments = 0; /* count of processed stripes */
        }
 
        return bi;
@@ -3347,8 +3243,7 @@ static int bio_fits_rdev(struct bio *bi)
        if ((bi->bi_size>>9) > q->max_sectors)
                return 0;
        blk_recount_segments(q, bi);
-       if (bi->bi_phys_segments > q->max_phys_segments ||
-           bi->bi_hw_segments > q->max_hw_segments)
+       if (bi->bi_phys_segments > q->max_phys_segments)
                return 0;
 
        if (q->merge_bvec_fn)
@@ -3430,6 +3325,58 @@ static int chunk_aligned_read(struct request_queue *q, struct bio * raid_bio)
        }
 }
 
+/* __get_priority_stripe - get the next stripe to process
+ *
+ * Full stripe writes are allowed to pass preread active stripes up until
+ * the bypass_threshold is exceeded.  In general the bypass_count
+ * increments when the handle_list is handled before the hold_list; however, it
+ * will not be incremented when STRIPE_IO_STARTED is sampled set signifying a
+ * stripe with in flight i/o.  The bypass_count will be reset when the
+ * head of the hold_list has changed, i.e. the head was promoted to the
+ * handle_list.
+ */
+static struct stripe_head *__get_priority_stripe(raid5_conf_t *conf)
+{
+       struct stripe_head *sh;
+
+       pr_debug("%s: handle: %s hold: %s full_writes: %d bypass_count: %d\n",
+                 __func__,
+                 list_empty(&conf->handle_list) ? "empty" : "busy",
+                 list_empty(&conf->hold_list) ? "empty" : "busy",
+                 atomic_read(&conf->pending_full_writes), conf->bypass_count);
+
+       if (!list_empty(&conf->handle_list)) {
+               sh = list_entry(conf->handle_list.next, typeof(*sh), lru);
+
+               if (list_empty(&conf->hold_list))
+                       conf->bypass_count = 0;
+               else if (!test_bit(STRIPE_IO_STARTED, &sh->state)) {
+                       if (conf->hold_list.next == conf->last_hold)
+                               conf->bypass_count++;
+                       else {
+                               conf->last_hold = conf->hold_list.next;
+                               conf->bypass_count -= conf->bypass_threshold;
+                               if (conf->bypass_count < 0)
+                                       conf->bypass_count = 0;
+                       }
+               }
+       } else if (!list_empty(&conf->hold_list) &&
+                  ((conf->bypass_threshold &&
+                    conf->bypass_count > conf->bypass_threshold) ||
+                   atomic_read(&conf->pending_full_writes) == 0)) {
+               sh = list_entry(conf->hold_list.next,
+                               typeof(*sh), lru);
+               conf->bypass_count -= conf->bypass_threshold;
+               if (conf->bypass_count < 0)
+                       conf->bypass_count = 0;
+       } else
+               return NULL;
+
+       list_del_init(&sh->lru);
+       atomic_inc(&sh->count);
+       BUG_ON(atomic_read(&sh->count) != 1);
+       return sh;
+}
 
 static int make_request(struct request_queue *q, struct bio * bi)
 {
@@ -3545,7 +3492,8 @@ static int make_request(struct request_queue *q, struct bio * bi)
                                goto retry;
                        }
                        finish_wait(&conf->wait_for_overlap, &w);
-                       handle_stripe(sh, NULL);
+                       set_bit(STRIPE_HANDLE, &sh->state);
+                       clear_bit(STRIPE_DELAYED, &sh->state);
                        release_stripe(sh);
                } else {
                        /* cannot get stripe for read-ahead, just give-up */
@@ -3556,16 +3504,14 @@ static int make_request(struct request_queue *q, struct bio * bi)
                        
        }
        spin_lock_irq(&conf->device_lock);
-       remaining = --bi->bi_phys_segments;
+       remaining = raid5_dec_bi_phys_segments(bi);
        spin_unlock_irq(&conf->device_lock);
        if (remaining == 0) {
 
                if ( rw == WRITE )
                        md_write_end(mddev);
 
-               bi->bi_end_io(bi,
-                             test_bit(BIO_UPTODATE, &bi->bi_flags)
-                               ? 0 : -EIO);
+               bio_endio(bi, 0);
        }
        return 0;
 }
@@ -3651,7 +3597,7 @@ static sector_t reshape_request(mddev_t *mddev, sector_t sector_nr, int *skipped
                            j == raid6_next_disk(sh->pd_idx, sh->disks))
                                continue;
                        s = compute_blocknr(sh, j);
-                       if (s < (mddev->array_size<<1)) {
+                       if (s < mddev->array_sectors) {
                                skipped = 1;
                                continue;
                        }
@@ -3694,6 +3640,25 @@ static sector_t reshape_request(mddev_t *mddev, sector_t sector_nr, int *skipped
                release_stripe(sh);
                first_sector += STRIPE_SECTORS;
        }
+       /* If this takes us to the resync_max point where we have to pause,
+        * then we need to write out the superblock.
+        */
+       sector_nr += conf->chunk_size>>9;
+       if (sector_nr >= mddev->resync_max) {
+               /* Cannot proceed until we've updated the superblock... */
+               wait_event(conf->wait_for_overlap,
+                          atomic_read(&conf->reshape_stripes) == 0);
+               mddev->reshape_position = conf->expand_progress;
+               set_bit(MD_CHANGE_DEVS, &mddev->flags);
+               md_wakeup_thread(mddev->thread);
+               wait_event(mddev->sb_wait,
+                          !test_bit(MD_CHANGE_DEVS, &mddev->flags)
+                          || kthread_should_stop());
+               spin_lock_irq(&conf->device_lock);
+               conf->expand_lo = mddev->reshape_position;
+               spin_unlock_irq(&conf->device_lock);
+               wake_up(&conf->wait_for_overlap);
+       }
        return conf->chunk_size>>9;
 }
 
@@ -3730,6 +3695,12 @@ static inline sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *ski
        if (test_bit(MD_RECOVERY_RESHAPE, &mddev->recovery))
                return reshape_request(mddev, sector_nr, skipped);
 
+       /* No need to check resync_max as we never do more than one
+        * stripe, and as resync_max will always be on a chunk boundary,
+        * if the check in md_do_sync didn't fire, there is no chance
+        * of overstepping resync_max here
+        */
+
        /* if there is too many failed drives and we are trying
         * to resync, then assert that we are finished, because there is
         * nothing we can do.
@@ -3749,6 +3720,9 @@ static inline sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *ski
                return sync_blocks * STRIPE_SECTORS; /* keep things rounded to whole stripes */
        }
 
+
+       bitmap_cond_end_sync(mddev->bitmap, sector_nr);
+
        pd_idx = stripe_to_pdidx(sector_nr, conf, raid_disks);
        sh = get_active_stripe(conf, sector_nr, raid_disks, pd_idx, 1);
        if (sh == NULL) {
@@ -3773,7 +3747,9 @@ static inline sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *ski
        clear_bit(STRIPE_INSYNC, &sh->state);
        spin_unlock(&sh->lock);
 
-       handle_stripe(sh, NULL);
+       /* wait for any blocked device to be handled */
+       while(unlikely(!handle_stripe(sh, NULL)))
+               ;
        release_stripe(sh);
 
        return STRIPE_SECTORS;
@@ -3812,7 +3788,7 @@ static int  retry_aligned_read(raid5_conf_t *conf, struct bio *raid_bio)
                     sector += STRIPE_SECTORS,
                     scnt++) {
 
-               if (scnt < raid_bio->bi_hw_segments)
+               if (scnt < raid5_bi_hw_segments(raid_bio))
                        /* already done this stripe */
                        continue;
 
@@ -3820,7 +3796,7 @@ static int  retry_aligned_read(raid5_conf_t *conf, struct bio *raid_bio)
 
                if (!sh) {
                        /* failed to get a stripe - must wait */
-                       raid_bio->bi_hw_segments = scnt;
+                       raid5_set_bi_hw_segments(raid_bio, scnt);
                        conf->retry_read_aligned = raid_bio;
                        return handled;
                }
@@ -3828,7 +3804,7 @@ static int  retry_aligned_read(raid5_conf_t *conf, struct bio *raid_bio)
                set_bit(R5_ReadError, &sh->dev[dd_idx].flags);
                if (!add_stripe_bio(sh, raid_bio, dd_idx, 0)) {
                        release_stripe(sh);
-                       raid_bio->bi_hw_segments = scnt;
+                       raid5_set_bi_hw_segments(raid_bio, scnt);
                        conf->retry_read_aligned = raid_bio;
                        return handled;
                }
@@ -3838,14 +3814,10 @@ static int  retry_aligned_read(raid5_conf_t *conf, struct bio *raid_bio)
                handled++;
        }
        spin_lock_irq(&conf->device_lock);
-       remaining = --raid_bio->bi_phys_segments;
+       remaining = raid5_dec_bi_phys_segments(raid_bio);
        spin_unlock_irq(&conf->device_lock);
-       if (remaining == 0) {
-
-               raid_bio->bi_end_io(raid_bio,
-                             test_bit(BIO_UPTODATE, &raid_bio->bi_flags)
-                               ? 0 : -EIO);
-       }
+       if (remaining == 0)
+               bio_endio(raid_bio, 0);
        if (atomic_dec_and_test(&conf->active_aligned_reads))
                wake_up(&conf->wait_for_stripe);
        return handled;
@@ -3860,7 +3832,7 @@ static int  retry_aligned_read(raid5_conf_t *conf, struct bio *raid_bio)
  * During the scan, completed stripes are saved for us by the interrupt
  * handler, so that they will not have to wait for our next wakeup.
  */
-static void raid5d (mddev_t *mddev)
+static void raid5d(mddev_t *mddev)
 {
        struct stripe_head *sh;
        raid5_conf_t *conf = mddev_to_conf(mddev);
@@ -3873,7 +3845,6 @@ static void raid5d (mddev_t *mddev)
        handled = 0;
        spin_lock_irq(&conf->device_lock);
        while (1) {
-               struct list_head *first;
                struct bio *bio;
 
                if (conf->seq_flush != conf->seq_write) {
@@ -3885,12 +3856,6 @@ static void raid5d (mddev_t *mddev)
                        activate_bit_delay(conf);
                }
 
-               if (list_empty(&conf->handle_list) &&
-                   atomic_read(&conf->preread_active_stripes) < IO_THRESHOLD &&
-                   !blk_queue_plugged(mddev->queue) &&
-                   !list_empty(&conf->delayed_list))
-                       raid5_activate_delayed(conf);
-
                while ((bio = remove_bio_from_retry(conf))) {
                        int ok;
                        spin_unlock_irq(&conf->device_lock);
@@ -3901,17 +3866,10 @@ static void raid5d (mddev_t *mddev)
                        handled++;
                }
 
-               if (list_empty(&conf->handle_list)) {
-                       async_tx_issue_pending_all();
-                       break;
-               }
-
-               first = conf->handle_list.next;
-               sh = list_entry(first, struct stripe_head, lru);
+               sh = __get_priority_stripe(conf);
 
-               list_del_init(first);
-               atomic_inc(&sh->count);
-               BUG_ON(atomic_read(&sh->count)!= 1);
+               if (!sh)
+                       break;
                spin_unlock_irq(&conf->device_lock);
                
                handled++;
@@ -3924,6 +3882,7 @@ static void raid5d (mddev_t *mddev)
 
        spin_unlock_irq(&conf->device_lock);
 
+       async_tx_issue_pending_all();
        unplug_slaves(mddev);
 
        pr_debug("--- raid5d inactive\n");
@@ -3943,15 +3902,15 @@ static ssize_t
 raid5_store_stripe_cache_size(mddev_t *mddev, const char *page, size_t len)
 {
        raid5_conf_t *conf = mddev_to_conf(mddev);
-       char *end;
-       int new;
+       unsigned long new;
+       int err;
+
        if (len >= PAGE_SIZE)
                return -EINVAL;
        if (!conf)
                return -ENODEV;
 
-       new = simple_strtoul(page, &end, 10);
-       if (!*page || (*end && *end != '\n') )
+       if (strict_strtoul(page, 10, &new))
                return -EINVAL;
        if (new <= 16 || new > 32768)
                return -EINVAL;
@@ -3961,7 +3920,9 @@ raid5_store_stripe_cache_size(mddev_t *mddev, const char *page, size_t len)
                else
                        break;
        }
-       md_allow_write(mddev);
+       err = md_allow_write(mddev);
+       if (err)
+               return err;
        while (new > conf->max_nr_stripes) {
                if (grow_one_stripe(conf))
                        conf->max_nr_stripes++;
@@ -3976,6 +3937,40 @@ raid5_stripecache_size = __ATTR(stripe_cache_size, S_IRUGO | S_IWUSR,
                                raid5_store_stripe_cache_size);
 
 static ssize_t
+raid5_show_preread_threshold(mddev_t *mddev, char *page)
+{
+       raid5_conf_t *conf = mddev_to_conf(mddev);
+       if (conf)
+               return sprintf(page, "%d\n", conf->bypass_threshold);
+       else
+               return 0;
+}
+
+static ssize_t
+raid5_store_preread_threshold(mddev_t *mddev, const char *page, size_t len)
+{
+       raid5_conf_t *conf = mddev_to_conf(mddev);
+       unsigned long new;
+       if (len >= PAGE_SIZE)
+               return -EINVAL;
+       if (!conf)
+               return -ENODEV;
+
+       if (strict_strtoul(page, 10, &new))
+               return -EINVAL;
+       if (new > conf->max_nr_stripes)
+               return -EINVAL;
+       conf->bypass_threshold = new;
+       return len;
+}
+
+static struct md_sysfs_entry
+raid5_preread_bypass_threshold = __ATTR(preread_bypass_threshold,
+                                       S_IRUGO | S_IWUSR,
+                                       raid5_show_preread_threshold,
+                                       raid5_store_preread_threshold);
+
+static ssize_t
 stripe_cache_active_show(mddev_t *mddev, char *page)
 {
        raid5_conf_t *conf = mddev_to_conf(mddev);
@@ -3991,6 +3986,7 @@ raid5_stripecache_active = __ATTR_RO(stripe_cache_active);
 static struct attribute *raid5_attrs[] =  {
        &raid5_stripecache_size.attr,
        &raid5_stripecache_active.attr,
+       &raid5_preread_bypass_threshold.attr,
        NULL,
 };
 static struct attribute_group raid5_attrs_group = {
@@ -4092,19 +4088,22 @@ static int run(mddev_t *mddev)
                        goto abort;
        }
        spin_lock_init(&conf->device_lock);
+       mddev->queue->queue_lock = &conf->device_lock;
        init_waitqueue_head(&conf->wait_for_stripe);
        init_waitqueue_head(&conf->wait_for_overlap);
        INIT_LIST_HEAD(&conf->handle_list);
+       INIT_LIST_HEAD(&conf->hold_list);
        INIT_LIST_HEAD(&conf->delayed_list);
        INIT_LIST_HEAD(&conf->bitmap_list);
        INIT_LIST_HEAD(&conf->inactive_list);
        atomic_set(&conf->active_stripes, 0);
        atomic_set(&conf->preread_active_stripes, 0);
        atomic_set(&conf->active_aligned_reads, 0);
+       conf->bypass_threshold = BYPASS_THRESHOLD;
 
        pr_debug("raid5: run(%s) called.\n", mdname(mddev));
 
-       ITERATE_RDEV(mddev,rdev,tmp) {
+       rdev_for_each(rdev, tmp, mddev) {
                raid_disk = rdev->raid_disk;
                if (raid_disk >= conf->raid_disks
                    || raid_disk < 0)
@@ -4119,7 +4118,9 @@ static int run(mddev_t *mddev)
                                " disk %d\n", bdevname(rdev->bdev,b),
                                raid_disk);
                        working_disks++;
-               }
+               } else
+                       /* Cannot rely on bitmap to complete recovery */
+                       conf->fullsync = 1;
        }
 
        /*
@@ -4246,7 +4247,7 @@ static int run(mddev_t *mddev)
        mddev->queue->backing_dev_info.congested_data = mddev;
        mddev->queue->backing_dev_info.congested_fn = raid5_congested;
 
-       mddev->array_size =  mddev->size * (conf->previous_raid_disks -
+       mddev->array_sectors = 2 * mddev->size * (conf->previous_raid_disks -
                                            conf->max_degraded);
 
        blk_queue_merge_bvec(mddev->queue, raid5_mergeable_bvec);
@@ -4396,6 +4397,14 @@ static int raid5_remove_disk(mddev_t *mddev, int number)
                        err = -EBUSY;
                        goto abort;
                }
+               /* Only remove non-faulty devices if recovery
+                * isn't possible.
+                */
+               if (!test_bit(Faulty, &rdev->flags) &&
+                   mddev->degraded <= conf->max_degraded) {
+                       err = -EBUSY;
+                       goto abort;
+               }
                p->rdev = NULL;
                synchronize_rcu();
                if (atomic_read(&rdev->nr_pending)) {
@@ -4413,35 +4422,41 @@ abort:
 static int raid5_add_disk(mddev_t *mddev, mdk_rdev_t *rdev)
 {
        raid5_conf_t *conf = mddev->private;
-       int found = 0;
+       int err = -EEXIST;
        int disk;
        struct disk_info *p;
+       int first = 0;
+       int last = conf->raid_disks - 1;
 
        if (mddev->degraded > conf->max_degraded)
                /* no point adding a device */
-               return 0;
+               return -EINVAL;
+
+       if (rdev->raid_disk >= 0)
+               first = last = rdev->raid_disk;
 
        /*
         * find the disk ... but prefer rdev->saved_raid_disk
         * if possible.
         */
        if (rdev->saved_raid_disk >= 0 &&
+           rdev->saved_raid_disk >= first &&
            conf->disks[rdev->saved_raid_disk].rdev == NULL)
                disk = rdev->saved_raid_disk;
        else
-               disk = 0;
-       for ( ; disk < conf->raid_disks; disk++)
+               disk = first;
+       for ( ; disk <= last ; disk++)
                if ((p=conf->disks + disk)->rdev == NULL) {
                        clear_bit(In_sync, &rdev->flags);
                        rdev->raid_disk = disk;
-                       found = 1;
+                       err = 0;
                        if (rdev->saved_raid_disk != disk)
                                conf->fullsync = 1;
                        rcu_assign_pointer(p->rdev, rdev);
                        break;
                }
        print_raid5_conf(conf);
-       return found;
+       return err;
 }
 
 static int raid5_resize(mddev_t *mddev, sector_t sectors)
@@ -4456,8 +4471,9 @@ static int raid5_resize(mddev_t *mddev, sector_t sectors)
        raid5_conf_t *conf = mddev_to_conf(mddev);
 
        sectors &= ~((sector_t)mddev->chunk_size/512 - 1);
-       mddev->array_size = (sectors * (mddev->raid_disks-conf->max_degraded))>>1;
-       set_capacity(mddev->gendisk, mddev->array_size << 1);
+       mddev->array_sectors = sectors * (mddev->raid_disks
+                                         - conf->max_degraded);
+       set_capacity(mddev->gendisk, mddev->array_sectors);
        mddev->changed = 1;
        if (sectors/2  > mddev->size && mddev->recovery_cp == MaxSector) {
                mddev->recovery_cp = mddev->size << 1;
@@ -4479,6 +4495,9 @@ static int raid5_check_reshape(mddev_t *mddev)
                return -EINVAL; /* Cannot shrink array or change level yet */
        if (mddev->delta_disks == 0)
                return 0; /* nothing to do */
+       if (mddev->bitmap)
+               /* Cannot grow a bitmap yet */
+               return -EBUSY;
 
        /* Can only proceed if there are plenty of stripe_heads.
         * We need a minimum of one full stripe,, and for sensible progress
@@ -4517,7 +4536,7 @@ static int raid5_start_reshape(mddev_t *mddev)
        if (test_bit(MD_RECOVERY_RUNNING, &mddev->recovery))
                return -EBUSY;
 
-       ITERATE_RDEV(mddev, rdev, rtmp)
+       rdev_for_each(rdev, rtmp, mddev)
                if (rdev->raid_disk < 0 &&
                    !test_bit(Faulty, &rdev->flags))
                        spares++;
@@ -4539,10 +4558,10 @@ static int raid5_start_reshape(mddev_t *mddev)
        /* Add some new drives, as many as will fit.
         * We know there are enough to make the newly sized array work.
         */
-       ITERATE_RDEV(mddev, rdev, rtmp)
+       rdev_for_each(rdev, rtmp, mddev)
                if (rdev->raid_disk < 0 &&
                    !test_bit(Faulty, &rdev->flags)) {
-                       if (raid5_add_disk(mddev, rdev)) {
+                       if (raid5_add_disk(mddev, rdev) == 0) {
                                char nm[20];
                                set_bit(In_sync, &rdev->flags);
                                added_devices++;
@@ -4590,15 +4609,16 @@ static void end_reshape(raid5_conf_t *conf)
        struct block_device *bdev;
 
        if (!test_bit(MD_RECOVERY_INTR, &conf->mddev->recovery)) {
-               conf->mddev->array_size = conf->mddev->size *
+               conf->mddev->array_sectors = 2 * conf->mddev->size *
                        (conf->raid_disks - conf->max_degraded);
-               set_capacity(conf->mddev->gendisk, conf->mddev->array_size << 1);
+               set_capacity(conf->mddev->gendisk, conf->mddev->array_sectors);
                conf->mddev->changed = 1;
 
                bdev = bdget_disk(conf->mddev->gendisk, 0);
                if (bdev) {
                        mutex_lock(&bdev->bd_inode->i_mutex);
-                       i_size_write(bdev->bd_inode, (loff_t)conf->mddev->array_size << 10);
+                       i_size_write(bdev->bd_inode,
+                                    (loff_t)conf->mddev->array_sectors << 9);
                        mutex_unlock(&bdev->bd_inode->i_mutex);
                        bdput(bdev);
                }