bcache: Btree verify code improvements

Used this fixed code to find and fix the bug fixed by
a4d885097b0ac0cd1337f171f2d4b83e946094d4.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>

drivers/md/bcache/bcache.h

@@ -629,6 +629,7 @@ struct cache_set {
 
 #ifdef CONFIG_BCACHE_DEBUG
 	struct btree		*verify_data;
+	struct bset		*verify_ondisk;
 	struct mutex		verify_lock;
 #endif
 

drivers/md/bcache/bset.c

@@ -1060,9 +1060,6 @@ static void __btree_sort(struct btree *b, struct btree_iter *iter,
 	btree_mergesort(b, out, iter, fixup, remove_stale);
 	b->nsets = start;
 
-	if (!fixup && !start && b->written)
-		bch_btree_verify(b, out);
-
 	if (!start && order == b->page_order) {
 		/*
 		 * Our temporary buffer is the same size as the btree node's

drivers/md/bcache/btree.c

@@ -203,7 +203,7 @@ static uint64_t btree_csum_set(struct btree *b, struct bset *i)
 	return crc ^ 0xffffffffffffffffULL;
 }
 
-static void bch_btree_node_read_done(struct btree *b)
+void bch_btree_node_read_done(struct btree *b)
 {
 	const char *err = "bad btree header";
 	struct bset *i = b->sets[0].data;
@@ -290,7 +290,7 @@ static void btree_node_read_endio(struct bio *bio, int error)
 	closure_put(cl);
 }
 
-void bch_btree_node_read(struct btree *b)
+static void bch_btree_node_read(struct btree *b)
 {
 	uint64_t start_time = local_clock();
 	struct closure cl;
@@ -478,6 +478,13 @@ void bch_btree_node_write(struct btree *b, struct closure *parent)
 
 	bch_btree_sort_lazy(b);
 
+	/*
+	 * do verify if there was more than one set initially (i.e. we did a
+	 * sort) and we sorted down to a single set:
+	 */
+	if (i != b->sets->data && !b->nsets)
+		bch_btree_verify(b);
+
 	if (b->written < btree_blocks(b))
 		bch_bset_init_next(b);
 }
@@ -782,6 +789,8 @@ void bch_btree_cache_free(struct cache_set *c)
 #ifdef CONFIG_BCACHE_DEBUG
 	if (c->verify_data)
 		list_move(&c->verify_data->list, &c->btree_cache);
+
+	free_pages((unsigned long) c->verify_ondisk, ilog2(bucket_pages(c)));
 #endif
 
 	list_splice(&c->btree_cache_freeable,
@@ -822,6 +831,9 @@ int bch_btree_cache_alloc(struct cache_set *c)
 #ifdef CONFIG_BCACHE_DEBUG
 	mutex_init(&c->verify_lock);
 
+	c->verify_ondisk = (void *)
+		__get_free_pages(GFP_KERNEL, ilog2(bucket_pages(c)));
+
 	c->verify_data = mca_bucket_alloc(c, &ZERO_KEY, GFP_KERNEL);
 
 	if (c->verify_data &&

drivers/md/bcache/btree.h

@@ -292,7 +292,7 @@ static inline void rw_unlock(bool w, struct btree *b)
 	(w ? up_write : up_read)(&b->lock);
 }
 
-void bch_btree_node_read(struct btree *);
+void bch_btree_node_read_done(struct btree *);
 void bch_btree_node_write(struct btree *, struct closure *);
 
 void bch_btree_set_root(struct btree *);

drivers/md/bcache/debug.c

@@ -53,18 +53,18 @@ int bch_bkey_to_text(char *buf, size_t size, const struct bkey *k)
 
 #define p(...)	(out += scnprintf(out, end - out, __VA_ARGS__))
 
-	p("%llu:%llu len %llu -> [", KEY_INODE(k), KEY_OFFSET(k), KEY_SIZE(k));
-
-	if (KEY_PTRS(k))
-		while (1) {
-			p("%llu:%llu gen %llu",
-			  PTR_DEV(k, i), PTR_OFFSET(k, i), PTR_GEN(k, i));
-
-			if (++i == KEY_PTRS(k))
-				break;
+	p("%llu:%llu len %llu -> [", KEY_INODE(k), KEY_START(k), KEY_SIZE(k));
 
+	for (i = 0; i < KEY_PTRS(k); i++) {
+		if (i)
 			p(", ");
-		}
+
+		if (PTR_DEV(k, i) == PTR_CHECK_DEV)
+			p("check dev");
+		else
+			p("%llu:%llu gen %llu", PTR_DEV(k, i),
+			  PTR_OFFSET(k, i), PTR_GEN(k, i));
+	}
 
 	p("]");
 
@@ -78,7 +78,7 @@ int bch_bkey_to_text(char *buf, size_t size, const struct bkey *k)
 
 #ifdef CONFIG_BCACHE_DEBUG
 
-static void dump_bset(struct btree *b, struct bset *i)
+static void dump_bset(struct btree *b, struct bset *i, unsigned set)
 {
 	struct bkey *k, *next;
 	unsigned j;
@@ -88,7 +88,7 @@ static void dump_bset(struct btree *b, struct bset *i)
 		next = bkey_next(k);
 
 		bch_bkey_to_text(buf, sizeof(buf), k);
-		printk(KERN_ERR "block %u key %zi/%u: %s", bset_block_offset(b, i),
+		printk(KERN_ERR "b %u k %zi/%u: %s", set,
 		       (uint64_t *) k - i->d, i->keys, buf);
 
 		for (j = 0; j < KEY_PTRS(k); j++) {
@@ -114,50 +114,83 @@ static void bch_dump_bucket(struct btree *b)
 
 	console_lock();
 	for (i = 0; i <= b->nsets; i++)
-		dump_bset(b, b->sets[i].data);
+		dump_bset(b, b->sets[i].data,
+			  bset_block_offset(b, b->sets[i].data));
 	console_unlock();
 }
 
-void bch_btree_verify(struct btree *b, struct bset *new)
+#define for_each_written_bset(b, start, i)				\
+	for (i = (start);						\
+	     (void *) i < (void *) (start) + (KEY_SIZE(&b->key) << 9) &&\
+	     i->seq == (start)->seq;					\
+	     i = (void *) i + set_blocks(i, b->c) * block_bytes(b->c))
+
+void bch_btree_verify(struct btree *b)
 {
 	struct btree *v = b->c->verify_data;
-	struct closure cl;
-	closure_init_stack(&cl);
+	struct bset *ondisk, *sorted, *inmemory;
+	struct bio *bio;
 
-	if (!b->c->verify)
+	if (!b->c->verify || !b->c->verify_ondisk)
 		return;
 
 	down(&b->io_mutex);
 	mutex_lock(&b->c->verify_lock);
 
+	ondisk = b->c->verify_ondisk;
+	sorted = b->c->verify_data->sets->data;
+	inmemory = b->sets->data;
+
 	bkey_copy(&v->key, &b->key);
 	v->written = 0;
 	v->level = b->level;
 
-	bch_btree_node_read(v);
+	bio = bch_bbio_alloc(b->c);
+	bio->bi_bdev		= PTR_CACHE(b->c, &b->key, 0)->bdev;
+	bio->bi_iter.bi_sector	= PTR_OFFSET(&b->key, 0);
+	bio->bi_iter.bi_size	= KEY_SIZE(&v->key) << 9;
+	bch_bio_map(bio, sorted);
 
-	if (new->keys != v->sets[0].data->keys ||
-	    memcmp(new->start,
-		   v->sets[0].data->start,
-		   (void *) end(new) - (void *) new->start)) {
-		unsigned i, j;
+	submit_bio_wait(REQ_META|READ_SYNC, bio);
+	bch_bbio_free(bio, b->c);
+
+	memcpy(ondisk, sorted, KEY_SIZE(&v->key) << 9);
+
+	bch_btree_node_read_done(v);
+	sorted = v->sets->data;
+
+	if (inmemory->keys != sorted->keys ||
+	    memcmp(inmemory->start,
+		   sorted->start,
+		   (void *) end(inmemory) - (void *) inmemory->start)) {
+		struct bset *i;
+		unsigned j;
 
 		console_lock();
 
-		printk(KERN_ERR "*** original memory node:\n");
-		for (i = 0; i <= b->nsets; i++)
-			dump_bset(b, b->sets[i].data);
+		printk(KERN_ERR "*** in memory:\n");
+		dump_bset(b, inmemory, 0);
 
-		printk(KERN_ERR "*** sorted memory node:\n");
-		dump_bset(b, new);
+		printk(KERN_ERR "*** read back in:\n");
+		dump_bset(v, sorted, 0);
 
-		printk(KERN_ERR "*** on disk node:\n");
-		dump_bset(v, v->sets[0].data);
+		for_each_written_bset(b, ondisk, i) {
+			unsigned block = ((void *) i - (void *) ondisk) /
+				block_bytes(b->c);
 
-		for (j = 0; j < new->keys; j++)
-			if (new->d[j] != v->sets[0].data->d[j])
+			printk(KERN_ERR "*** on disk block %u:\n", block);
+			dump_bset(b, i, block);
+		}
+
+		printk(KERN_ERR "*** block %zu not written\n",
+		       ((void *) i - (void *) ondisk) / block_bytes(b->c));
+
+		for (j = 0; j < inmemory->keys; j++)
+			if (inmemory->d[j] != sorted->d[j])
 				break;
 
+		printk(KERN_ERR "b->written %u\n", b->written);
+
 		console_unlock();
 		panic("verify failed at %u\n", j);
 	}

drivers/md/bcache/debug.h

@@ -7,7 +7,7 @@ int bch_bkey_to_text(char *buf, size_t size, const struct bkey *k);
 
 #ifdef CONFIG_BCACHE_DEBUG
 
-void bch_btree_verify(struct btree *, struct bset *);
+void bch_btree_verify(struct btree *);
 void bch_data_verify(struct cached_dev *, struct bio *);
 int __bch_count_data(struct btree *);
 void __bch_check_keys(struct btree *, const char *, ...);
@@ -20,7 +20,7 @@ void bch_btree_iter_next_check(struct btree_iter *);
 
 #else /* DEBUG */
 
-static inline void bch_btree_verify(struct btree *b, struct bset *i) {}
+static inline void bch_btree_verify(struct btree *b) {}
 static inline void bch_data_verify(struct cached_dev *dc, struct bio *bio) {}
 static inline int __bch_count_data(struct btree *b) { return -1; }
 static inline void __bch_check_keys(struct btree *b, const char *fmt, ...) {}