bcache: Convert try_wait to wait_queue_head_t

We never waited on c->try_wait asynchronously, so just use the standard
primitives.

Signed-off-by: Kent Overstreet <kmo@daterainc.com>

diff --git a/drivers/md/bcache/bcache.h b/drivers/md/bcache/bcache.h
index 10ce0c825fcebd5908349cbd65f810bd43fcb995..c1c44191afb1a633fb42be7f23efaa083b92570e 100644 (file)
--- a/drivers/md/bcache/bcache.h
+++ b/drivers/md/bcache/bcache.h
@@ -735,8 +735,8 @@ struct cache_set {
         * basically a lock for this that we can wait on asynchronously. The
         * btree_root() macro releases the lock when it returns.
         */
-       struct closure          *try_harder;
-       struct closure_waitlist try_wait;
+       struct task_struct      *try_harder;
+       wait_queue_head_t       try_wait;
        uint64_t                try_harder_start;
 
        /*

diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c
index 01c0612bc007b510f016b40d4ad0b2bfb2581717..18913d6fb413ae6f472db2e68b4fe61ca6404deb 100644 (file)
--- a/drivers/md/bcache/btree.c
+++ b/drivers/md/bcache/btree.c
@@ -437,7 +437,7 @@ static void bch_btree_leaf_dirty(struct btree *b, struct btree_op *op)
 
        set_btree_node_dirty(b);
 
-       if (op && op->journal) {
+       if (op->journal) {
                if (w->journal &&
                    journal_pin_cmp(b->c, w, op)) {
                        atomic_dec_bug(w->journal);
@@ -574,34 +574,35 @@ static struct btree *mca_bucket_alloc(struct cache_set *c,
        return b;
 }
 
-static int mca_reap(struct btree *b, struct closure *cl, unsigned min_order)
+static int mca_reap(struct btree *b, unsigned min_order, bool flush)
 {
+       struct closure cl;
+
+       closure_init_stack(&cl);
        lockdep_assert_held(&b->c->bucket_lock);
 
        if (!down_write_trylock(&b->lock))
                return -ENOMEM;
 
-       if (b->page_order < min_order) {
+       BUG_ON(btree_node_dirty(b) && !b->sets[0].data);
+
+       if (b->page_order < min_order ||
+           (!flush &&
+            (btree_node_dirty(b) ||
+             atomic_read(&b->io.cl.remaining) != -1))) {
                rw_unlock(true, b);
                return -ENOMEM;
        }
 
-       BUG_ON(btree_node_dirty(b) && !b->sets[0].data);
-
-       if (cl && btree_node_dirty(b))
-               bch_btree_node_write(b, NULL);
-
-       if (cl)
-               closure_wait_event_async(&b->io.wait, cl,
-                        atomic_read(&b->io.cl.remaining) == -1);
-
-       if (btree_node_dirty(b) ||
-           !closure_is_unlocked(&b->io.cl) ||
-           work_pending(&b->work.work)) {
-               rw_unlock(true, b);
-               return -EAGAIN;
+       if (btree_node_dirty(b)) {
+               bch_btree_node_write(b, &cl);
+               closure_sync(&cl);
        }
 
+       /* wait for any in flight btree write */
+       closure_wait_event_sync(&b->io.wait, &cl,
+               atomic_read(&b->io.cl.remaining) == -1);
+
        return 0;
 }
 
@@ -646,7 +647,7 @@ static int bch_mca_shrink(struct shrinker *shrink, struct shrink_control *sc)
                        break;
 
                if (++i > 3 &&
-                   !mca_reap(b, NULL, 0)) {
+                   !mca_reap(b, 0, false)) {
                        mca_data_free(b);
                        rw_unlock(true, b);
                        --nr;
@@ -665,7 +666,7 @@ static int bch_mca_shrink(struct shrinker *shrink, struct shrink_control *sc)
                list_rotate_left(&c->btree_cache);
 
                if (!b->accessed &&
-                   !mca_reap(b, NULL, 0)) {
+                   !mca_reap(b, 0, false)) {
                        mca_bucket_free(b);
                        mca_data_free(b);
                        rw_unlock(true, b);
@@ -774,52 +775,27 @@ out:
        return b;
 }
 
-static struct btree *mca_cannibalize(struct cache_set *c, struct bkey *k,
-                                    int level, struct closure *cl)
+static struct btree *mca_cannibalize(struct cache_set *c, struct bkey *k)
 {
-       int ret = -ENOMEM;
-       struct btree *i;
+       struct btree *b;
 
        trace_bcache_btree_cache_cannibalize(c);
 
-       if (!cl)
-               return ERR_PTR(-ENOMEM);
-
-       /*
-        * Trying to free up some memory - i.e. reuse some btree nodes - may
-        * require initiating IO to flush the dirty part of the node. If we're
-        * running under generic_make_request(), that IO will never finish and
-        * we would deadlock. Returning -EAGAIN causes the cache lookup code to
-        * punt to workqueue and retry.
-        */
-       if (current->bio_list)
-               return ERR_PTR(-EAGAIN);
-
-       if (c->try_harder && c->try_harder != cl) {
-               closure_wait_event_async(&c->try_wait, cl, !c->try_harder);
-               return ERR_PTR(-EAGAIN);
-       }
+       if (!c->try_harder) {
+               c->try_harder = current;
+               c->try_harder_start = local_clock();
+       } else if (c->try_harder != current)
+               return ERR_PTR(-ENOSPC);
 
-       c->try_harder = cl;
-       c->try_harder_start = local_clock();
-retry:
-       list_for_each_entry_reverse(i, &c->btree_cache, list) {
-               int r = mca_reap(i, cl, btree_order(k));
-               if (!r)
-                       return i;
-               if (r != -ENOMEM)
-                       ret = r;
-       }
+       list_for_each_entry_reverse(b, &c->btree_cache, list)
+               if (!mca_reap(b, btree_order(k), false))
+                       return b;
 
-       if (ret == -EAGAIN &&
-           closure_blocking(cl)) {
-               mutex_unlock(&c->bucket_lock);
-               closure_sync(cl);
-               mutex_lock(&c->bucket_lock);
-               goto retry;
-       }
+       list_for_each_entry_reverse(b, &c->btree_cache, list)
+               if (!mca_reap(b, btree_order(k), true))
+                       return b;
 
-       return ERR_PTR(ret);
+       return ERR_PTR(-ENOMEM);
 }
 
 /*
@@ -830,18 +806,19 @@ retry:
  */
 void bch_cannibalize_unlock(struct cache_set *c, struct closure *cl)
 {
-       if (c->try_harder == cl) {
+       if (c->try_harder == current) {
                bch_time_stats_update(&c->try_harder_time, c->try_harder_start);
                c->try_harder = NULL;
-               __closure_wake_up(&c->try_wait);
+               wake_up(&c->try_wait);
        }
 }
 
-static struct btree *mca_alloc(struct cache_set *c, struct bkey *k,
-                              int level, struct closure *cl)
+static struct btree *mca_alloc(struct cache_set *c, struct bkey *k, int level)
 {
        struct btree *b;
 
+       BUG_ON(current->bio_list);
+
        lockdep_assert_held(&c->bucket_lock);
 
        if (mca_find(c, k))
@@ -851,14 +828,14 @@ static struct btree *mca_alloc(struct cache_set *c, struct bkey *k,
         * the list. Check if there's any freed nodes there:
         */
        list_for_each_entry(b, &c->btree_cache_freeable, list)
-               if (!mca_reap(b, NULL, btree_order(k)))
+               if (!mca_reap(b, btree_order(k), false))
                        goto out;
 
        /* We never free struct btree itself, just the memory that holds the on
         * disk node. Check the freed list before allocating a new one:
         */
        list_for_each_entry(b, &c->btree_cache_freed, list)
-               if (!mca_reap(b, NULL, 0)) {
+               if (!mca_reap(b, 0, false)) {
                        mca_data_alloc(b, k, __GFP_NOWARN|GFP_NOIO);
                        if (!b->sets[0].data)
                                goto err;
@@ -892,7 +869,7 @@ err:
        if (b)
                rw_unlock(true, b);
 
-       b = mca_cannibalize(c, k, level, cl);
+       b = mca_cannibalize(c, k);
        if (!IS_ERR(b))
                goto out;
 
@@ -910,10 +887,9 @@ err:
  * level and op->lock.
  */
 struct btree *bch_btree_node_get(struct cache_set *c, struct bkey *k,
-                                int level, struct btree_op *op)
+                                int level, bool write)
 {
        int i = 0;
-       bool write = level <= op->lock;
        struct btree *b;
 
        BUG_ON(level < 0);
@@ -925,7 +901,7 @@ retry:
                        return ERR_PTR(-EAGAIN);
 
                mutex_lock(&c->bucket_lock);
-               b = mca_alloc(c, k, level, &op->cl);
+               b = mca_alloc(c, k, level);
                mutex_unlock(&c->bucket_lock);
 
                if (!b)
@@ -971,7 +947,7 @@ static void btree_node_prefetch(struct cache_set *c, struct bkey *k, int level)
        struct btree *b;
 
        mutex_lock(&c->bucket_lock);
-       b = mca_alloc(c, k, level, NULL);
+       b = mca_alloc(c, k, level);
        mutex_unlock(&c->bucket_lock);
 
        if (!IS_ERR_OR_NULL(b)) {
@@ -982,17 +958,12 @@ static void btree_node_prefetch(struct cache_set *c, struct bkey *k, int level)
 
 /* Btree alloc */
 
-static void btree_node_free(struct btree *b, struct btree_op *op)
+static void btree_node_free(struct btree *b)
 {
        unsigned i;
 
        trace_bcache_btree_node_free(b);
 
-       /*
-        * The BUG_ON() in btree_node_get() implies that we must have a write
-        * lock on parent to free or even invalidate a node
-        */
-       BUG_ON(op->lock <= b->level);
        BUG_ON(b == b->c->root);
 
        if (btree_node_dirty(b))
@@ -1028,7 +999,7 @@ retry:
 
        SET_KEY_SIZE(&k.key, c->btree_pages * PAGE_SECTORS);
 
-       b = mca_alloc(c, &k.key, level, cl);
+       b = mca_alloc(c, &k.key, level);
        if (IS_ERR(b))
                goto err_free;
 
@@ -1164,8 +1135,7 @@ static int btree_gc_mark_node(struct btree *b, unsigned *keys,
        return stale;
 }
 
-static struct btree *btree_gc_alloc(struct btree *b, struct bkey *k,
-                                   struct btree_op *op)
+static struct btree *btree_gc_alloc(struct btree *b, struct bkey *k)
 {
        /*
         * We block priorities from being written for the duration of garbage
@@ -1182,7 +1152,7 @@ static struct btree *btree_gc_alloc(struct btree *b, struct bkey *k,
                memcpy(k->ptr, b->key.ptr,
                       sizeof(uint64_t) * KEY_PTRS(&b->key));
 
-               btree_node_free(n, op);
+               btree_node_free(n);
                up_write(&n->lock);
        }
 
@@ -1202,8 +1172,8 @@ struct gc_merge_info {
        unsigned        keys;
 };
 
-static void btree_gc_coalesce(struct btree *b, struct btree_op *op,
-                             struct gc_stat *gc, struct gc_merge_info *r)
+static void btree_gc_coalesce(struct btree *b, struct gc_stat *gc,
+                             struct gc_merge_info *r)
 {
        unsigned nodes = 0, keys = 0, blocks;
        int i;
@@ -1219,7 +1189,7 @@ static void btree_gc_coalesce(struct btree *b, struct btree_op *op,
 
        for (i = nodes - 1; i >= 0; --i) {
                if (r[i].b->written)
-                       r[i].b = btree_gc_alloc(r[i].b, r[i].k, op);
+                       r[i].b = btree_gc_alloc(r[i].b, r[i].k);
 
                if (r[i].b->written)
                        return;
@@ -1283,7 +1253,7 @@ static void btree_gc_coalesce(struct btree *b, struct btree_op *op,
                r[i - 1].keys   = n2->keys;
        }
 
-       btree_node_free(r->b, op);
+       btree_node_free(r->b);
        up_write(&r->b->lock);
 
        trace_bcache_btree_gc_coalesce(nodes);
@@ -1315,7 +1285,7 @@ static int btree_gc_recurse(struct btree *b, struct btree_op *op,
        memset(r, 0, sizeof(r));
 
        while ((r->k = bch_next_recurse_key(b, &b->c->gc_done))) {
-               r->b = bch_btree_node_get(b->c, r->k, b->level - 1, op);
+               r->b = bch_btree_node_get(b->c, r->k, b->level - 1, true);
 
                if (IS_ERR(r->b)) {
                        ret = PTR_ERR(r->b);
@@ -1328,7 +1298,7 @@ static int btree_gc_recurse(struct btree *b, struct btree_op *op,
                if (!b->written &&
                    (r->b->level || stale > 10 ||
                     b->c->gc_always_rewrite))
-                       r->b = btree_gc_alloc(r->b, r->k, op);
+                       r->b = btree_gc_alloc(r->b, r->k);
 
                if (r->b->level)
                        ret = btree_gc_recurse(r->b, op, writes, gc);
@@ -1341,7 +1311,7 @@ static int btree_gc_recurse(struct btree *b, struct btree_op *op,
                bkey_copy_key(&b->c->gc_done, r->k);
 
                if (!b->written)
-                       btree_gc_coalesce(b, op, gc, r);
+                       btree_gc_coalesce(b, gc, r);
 
                if (r[GC_MERGE_NODES - 1].b)
                        write(r[GC_MERGE_NODES - 1].b);
@@ -1395,7 +1365,7 @@ static int bch_btree_gc_root(struct btree *b, struct btree_op *op,
        if (!IS_ERR_OR_NULL(n)) {
                closure_sync(&op->cl);
                bch_btree_set_root(b);
-               btree_node_free(n, op);
+               btree_node_free(n);
                rw_unlock(true, b);
        }
 
@@ -1992,18 +1962,18 @@ static int btree_split(struct btree *b, struct btree_op *op,
        }
 
        rw_unlock(true, n1);
-       btree_node_free(b, op);
+       btree_node_free(b);
 
        bch_time_stats_update(&b->c->btree_split_time, start_time);
 
        return 0;
 err_free2:
        __bkey_put(n2->c, &n2->key);
-       btree_node_free(n2, op);
+       btree_node_free(n2);
        rw_unlock(true, n2);
 err_free1:
        __bkey_put(n1->c, &n1->key);
-       btree_node_free(n1, op);
+       btree_node_free(n1);
        rw_unlock(true, n1);
 err:
        if (n3 == ERR_PTR(-EAGAIN) ||

diff --git a/drivers/md/bcache/btree.h b/drivers/md/bcache/btree.h
index 17b7a4e39c7e07df5b9966a663b3e9834619191c..72794ab8e8e5038a40faa2ade443a64915269699 100644 (file)
--- a/drivers/md/bcache/btree.h
+++ b/drivers/md/bcache/btree.h
@@ -326,7 +326,7 @@ static inline void rw_unlock(bool w, struct btree *b)
 ({                                                                     \
        int _r, l = (b)->level - 1;                                     \
        bool _w = l <= (op)->lock;                                      \
-       struct btree *_child = bch_btree_node_get((b)->c, key, l, op);  \
+       struct btree *_child = bch_btree_node_get((b)->c, key, l, _w);  \
        if (!IS_ERR(_child)) {                                          \
                _child->parent = (b);                                   \
                _r = bch_btree_ ## fn(_child, op, ##__VA_ARGS__);       \
@@ -356,6 +356,11 @@ static inline void rw_unlock(bool w, struct btree *b)
                }                                                       \
                rw_unlock(_w, _b);                                      \
                bch_cannibalize_unlock(c, &(op)->cl);                   \
+               if (_r == -ENOSPC) {                                    \
+                       wait_event((c)->try_wait,                       \
+                                  !(c)->try_harder);                   \
+                       _r = -EINTR;                                    \
+               }                                                       \
        } while (_r == -EINTR);                                         \
                                                                        \
        _r;                                                             \
@@ -375,8 +380,7 @@ void bch_btree_node_write(struct btree *, struct closure *);
 void bch_cannibalize_unlock(struct cache_set *, struct closure *);
 void bch_btree_set_root(struct btree *);
 struct btree *bch_btree_node_alloc(struct cache_set *, int, struct closure *);
-struct btree *bch_btree_node_get(struct cache_set *, struct bkey *,
-                               int, struct btree_op *);
+struct btree *bch_btree_node_get(struct cache_set *, struct bkey *, int, bool);
 
 int bch_btree_insert_check_key(struct btree *, struct btree_op *,
                               struct bkey *);

diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
index d3169c0652f8cd67e1ae1a478eb687c157ed8ae3..9a164cd4058c94925e9955e863eaef66e791f1e4 100644 (file)
--- a/drivers/md/bcache/super.c
+++ b/drivers/md/bcache/super.c
@@ -1436,12 +1436,14 @@ struct cache_set *bch_cache_set_alloc(struct cache_sb *sb)
 
        c->sort_crit_factor = int_sqrt(c->btree_pages);
 
-       mutex_init(&c->bucket_lock);
-       mutex_init(&c->sort_lock);
-       spin_lock_init(&c->sort_time_lock);
        closure_init_unlocked(&c->sb_write);
+       mutex_init(&c->bucket_lock);
+       init_waitqueue_head(&c->try_wait);
        closure_init_unlocked(&c->uuid_write);
+       spin_lock_init(&c->sort_time_lock);
+       mutex_init(&c->sort_lock);
        spin_lock_init(&c->btree_read_time_lock);
+
        bch_moving_init_cache_set(c);
 
        INIT_LIST_HEAD(&c->list);
@@ -1529,7 +1531,7 @@ static void run_cache_set(struct cache_set *c)
                        goto err;
 
                err = "error reading btree root";
-               c->root = bch_btree_node_get(c, k, j->btree_level, &op);
+               c->root = bch_btree_node_get(c, k, j->btree_level, true);
                if (IS_ERR_OR_NULL(c->root))
                        goto err;