md: replace STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} with 'reconstruct_states'
authorDan Williams <dan.j.williams@intel.com>
Fri, 27 Jun 2008 22:32:05 +0000 (08:32 +1000)
committerNeil Brown <neilb@notabene.brown>
Fri, 27 Jun 2008 22:32:05 +0000 (08:32 +1000)
From: Dan Williams <dan.j.williams@intel.com>

Track the state of reconstruct operations (recalculating the parity block
usually due to incoming writes, or as part of array expansion)  Reduces the
scope of the STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} flags to only tracking whether
a reconstruct operation has been requested via the ops_request field of struct
stripe_head_state.

This is the final step in the removal of ops.{pending,ack,complete,count}, i.e.
the STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} flags only request an operation and do
not track the state of the operation.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
drivers/md/raid5.c
include/linux/raid/raid5.h

index 835046bf384e035913db9dc88a167d6cf174d6b8..b9159367491a9247f6cf13d3bb4f9ddb2df07af4 100644 (file)
@@ -122,6 +122,13 @@ static void return_io(struct bio *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)) {
@@ -141,7 +148,7 @@ 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)
@@ -243,7 +250,7 @@ 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",
@@ -344,47 +351,6 @@ 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);
-
-       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
@@ -609,7 +575,7 @@ static void ops_complete_compute5(void *stripe_head_ref)
 }
 
 static struct dma_async_tx_descriptor *
-ops_run_compute5(struct stripe_head *sh, unsigned long pending)
+ops_run_compute5(struct stripe_head *sh, unsigned long ops_request)
 {
        /* kernel stack size limits the total number of disks */
        int disks = sh->disks;
@@ -640,7 +606,7 @@ ops_run_compute5(struct stripe_head *sh, unsigned long pending)
                        ops_complete_compute5, sh);
 
        /* ack now if postxor is not set to be run */
-       if (tx && !test_bit(STRIPE_OP_POSTXOR, &pending))
+       if (tx && !test_bit(STRIPE_OP_POSTXOR, &ops_request))
                async_tx_ack(tx);
 
        return tx;
@@ -652,8 +618,6 @@ static void ops_complete_prexor(void *stripe_head_ref)
 
        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 *
@@ -686,7 +650,7 @@ ops_run_prexor(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
 
 static struct dma_async_tx_descriptor *
 ops_run_biodrain(struct stripe_head *sh, struct dma_async_tx_descriptor *tx,
-                unsigned long pending)
+                unsigned long ops_request)
 {
        int disks = sh->disks;
        int pd_idx = sh->pd_idx, i;
@@ -694,7 +658,7 @@ ops_run_biodrain(struct stripe_head *sh, struct dma_async_tx_descriptor *tx,
        /* 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, &pending);
+       int prexor = test_bit(STRIPE_OP_PREXOR, &ops_request);
 
        pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
@@ -744,7 +708,7 @@ static void ops_complete_postxor(void *stripe_head_ref)
        pr_debug("%s: stripe %llu\n", __func__,
                (unsigned long long)sh->sector);
 
-       set_bit(STRIPE_OP_POSTXOR, &sh->ops.complete);
+       sh->reconstruct_state = reconstruct_state_result;
        set_bit(STRIPE_HANDLE, &sh->state);
        release_stripe(sh);
 }
@@ -763,16 +727,14 @@ 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);
-
+       sh->reconstruct_state = reconstruct_state_drain_result;
        set_bit(STRIPE_HANDLE, &sh->state);
        release_stripe(sh);
 }
 
 static void
 ops_run_postxor(struct stripe_head *sh, struct dma_async_tx_descriptor *tx,
-               unsigned long pending)
+               unsigned long ops_request)
 {
        /* kernel stack size limits the total number of disks */
        int disks = sh->disks;
@@ -780,7 +742,7 @@ 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, &pending);
+       int prexor = test_bit(STRIPE_OP_PREXOR, &ops_request);
        unsigned long flags;
        dma_async_tx_callback callback;
 
@@ -807,7 +769,7 @@ 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, &pending) ?
+       callback = test_bit(STRIPE_OP_BIODRAIN, &ops_request) ?
                ops_complete_write : ops_complete_postxor;
 
        /* 1/ if we prexor'd then the dest is reused as a source
@@ -868,8 +830,7 @@ static void ops_run_check(struct stripe_head *sh)
                ops_complete_check, sh);
 }
 
-static void raid5_run_ops(struct stripe_head *sh, unsigned long pending,
-                         unsigned long ops_request)
+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;
@@ -880,18 +841,18 @@ static void raid5_run_ops(struct stripe_head *sh, unsigned long pending,
        }
 
        if (test_bit(STRIPE_OP_COMPUTE_BLK, &ops_request))
-               tx = ops_run_compute5(sh, pending);
+               tx = ops_run_compute5(sh, ops_request);
 
-       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)) {
-               tx = ops_run_biodrain(sh, tx, pending);
+       if (test_bit(STRIPE_OP_BIODRAIN, &ops_request)) {
+               tx = ops_run_biodrain(sh, tx, ops_request);
                overlap_clear++;
        }
 
-       if (test_bit(STRIPE_OP_POSTXOR, &pending))
-               ops_run_postxor(sh, tx, pending);
+       if (test_bit(STRIPE_OP_POSTXOR, &ops_request))
+               ops_run_postxor(sh, tx, ops_request);
 
        if (test_bit(STRIPE_OP_CHECK, &ops_request))
                ops_run_check(sh);
@@ -1684,11 +1645,11 @@ 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
+handle_write_operations5(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
@@ -1696,12 +1657,12 @@ 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];
@@ -1710,21 +1671,20 @@ handle_write_operations5(struct stripe_head *sh, int rcw, int expand)
                                set_bit(R5_LOCKED, &dev->flags);
                                if (!expand)
                                        clear_bit(R5_UPTODATE, &dev->flags);
-                               locked++;
+                               s->locked++;
                        }
                }
-               if (locked + 1 == disks)
+               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_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];
@@ -1742,7 +1702,7 @@ handle_write_operations5(struct stripe_head *sh, int rcw, int expand)
                                set_bit(R5_Wantprexor, &dev->flags);
                                set_bit(R5_LOCKED, &dev->flags);
                                clear_bit(R5_UPTODATE, &dev->flags);
-                               locked++;
+                               s->locked++;
                        }
                }
        }
@@ -1752,13 +1712,11 @@ 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",
+       pr_debug("%s: stripe %llu locked: %d ops_request: %lx\n",
                __func__, (unsigned long long)sh->sector,
-               locked, sh->ops.pending);
-
-       return locked;
+               s->locked, s->ops_request);
 }
 
 /*
@@ -2005,8 +1963,7 @@ static void handle_issuing_new_read_requests5(struct stripe_head *sh,
         * midst of changing due to a write
         */
        if (!test_bit(STRIPE_COMPUTE_RUN, &sh->state) && !sh->check_state &&
-           !test_bit(STRIPE_OP_PREXOR, &sh->ops.pending) &&
-           !test_bit(STRIPE_OP_POSTXOR, &sh->ops.pending)) {
+           !sh->reconstruct_state) {
                for (i = disks; i--; )
                        if (__handle_issuing_new_read_requests5(
                                sh, s, i, disks) == 0)
@@ -2211,7 +2168,7 @@ static void handle_issuing_new_write_requests5(raid5_conf_t *conf,
        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)))
-               s->locked += handle_write_operations5(sh, rcw == 0, 0);
+               handle_write_operations5(sh, s, rcw == 0, 0);
 }
 
 static void handle_issuing_new_write_requests6(raid5_conf_t *conf,
@@ -2581,15 +2538,14 @@ static void handle_stripe5(struct stripe_head *sh)
        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);
@@ -2703,34 +2659,12 @@ static void handle_stripe5(struct stripe_head *sh)
        /* 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
-        */
        prexor = 0;
-       if (test_bit(STRIPE_OP_PREXOR, &sh->ops.complete) &&
-               test_bit(STRIPE_OP_POSTXOR, &sh->ops.complete)) {
-
-               prexor = 1;
-               clear_bit(STRIPE_OP_PREXOR, &sh->ops.complete);
-               clear_bit(STRIPE_OP_PREXOR, &sh->ops.ack);
-               clear_bit(STRIPE_OP_PREXOR, &sh->ops.pending);
-
+       if (sh->reconstruct_state == reconstruct_state_drain_result) {
+               sh->reconstruct_state = reconstruct_state_idle;
                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 += test_and_clear_bit(R5_Wantprexor,
+                                                    &sh->dev[i].flags);
 
                /* All the 'written' buffers and the parity block are ready to
                 * be written back to disk
@@ -2763,8 +2697,7 @@ 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) &&
-           !sh->check_state)
+       if (s.to_write && !sh->reconstruct_state && !sh->check_state)
                handle_issuing_new_write_requests5(conf, sh, &s, disks);
 
        /* maybe we need to check and possibly fix the parity for this stripe
@@ -2805,18 +2738,10 @@ static void handle_stripe5(struct stripe_head *sh)
                }
        }
 
-       /* 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);
                        set_bit(R5_LOCKED, &dev->flags);
@@ -2824,15 +2749,13 @@ static void handle_stripe5(struct stripe_head *sh)
        }
 
        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 &&
-                  s.locked == 0 &&
-               !test_bit(STRIPE_OP_POSTXOR, &sh->ops.pending)) {
+               handle_write_operations5(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);
@@ -2843,9 +2766,6 @@ static void handle_stripe5(struct stripe_head *sh)
            !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);
 
@@ -2853,8 +2773,8 @@ static void handle_stripe5(struct stripe_head *sh)
        if (unlikely(blocked_rdev))
                md_wait_for_blocked_rdev(blocked_rdev, conf->mddev);
 
-       if (pending || s.ops_request)
-               raid5_run_ops(sh, pending, s.ops_request);
+       if (s.ops_request)
+               raid5_run_ops(sh, s.ops_request);
 
        ops_run_io(sh, &s);
 
index 2c96d5fd54bff4e441374f596b0f564c730d6d3d..5f3e674b87dddf9cf5cd38dab9d176cd1d92234e 100644 (file)
@@ -205,19 +205,12 @@ struct stripe_head {
        int                     bm_seq; /* sequence number for bitmap flushes */
        int                     disks;                  /* disks in stripe */
        enum check_states       check_state;
+       enum reconstruct_states reconstruct_state;
        /* stripe_operations
-        * @pending - pending ops flags (set for request->issue->complete)
-        * @ack - submitted ops flags (set for issue->complete)
-        * @complete - completed ops flags (set for complete)
         * @target - STRIPE_OP_COMPUTE_BLK target
-        * @count - raid5_runs_ops is set to run when this is non-zero
         */
        struct stripe_operations {
-               unsigned long      pending;
-               unsigned long      ack;
-               unsigned long      complete;
                int                target;
-               int                count;
                u32                zero_sum_result;
        } ops;
        struct r5dev {