Update bcachefs sources to 3ca08ab51ec9 bcachefs: six locks: Simplify optimistic spinning

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

.bcachefs_revision

@@ -1 +1 @@
-d464ec667b2b9de097e39d1505b45aafd87a9552
+3ca08ab51ec996180c20105489176b8c4327240c

include/linux/kernel.h

@@ -278,4 +278,7 @@ static inline void dump_stack(void) {}
 #define unsafe_memcpy(dst, src, bytes, justification)		\
 	memcpy(dst, src, bytes)
 
+#define DECLARE_FLEX_ARRAY(TYPE, NAME) \
+	__DECLARE_FLEX_ARRAY(TYPE, NAME)
+
 #endif

include/linux/list.h

@@ -98,4 +98,15 @@ static inline void hlist_del_init(struct hlist_node *n)
 	     pos;							\
 	     pos = hlist_entry_safe((pos)->member.next, typeof(*(pos)), member))
 
+static inline size_t list_count_nodes(struct list_head *head)
+{
+	struct list_head *pos;
+	size_t count = 0;
+
+	list_for_each(pos, head)
+		count++;
+
+	return count;
+}
+
 #endif /* _LIST_LIST_H */

libbcachefs/alloc_background.c

@@ -561,8 +561,8 @@ int bch2_bucket_gens_init(struct bch_fs *c)
 
 		if (have_bucket_gens_key && bkey_cmp(iter.pos, pos)) {
 			ret = commit_do(trans, NULL, NULL,
-					BTREE_INSERT_NOFAIL|
+					BCH_TRANS_COMMIT_no_enospc|
-					BTREE_INSERT_LAZY_RW,
+					BCH_TRANS_COMMIT_lazy_rw,
 				bch2_btree_insert_trans(trans, BTREE_ID_bucket_gens, &g.k_i, 0));
 			if (ret)
 				break;
@@ -581,8 +581,8 @@ int bch2_bucket_gens_init(struct bch_fs *c)
 
 	if (have_bucket_gens_key && !ret)
 		ret = commit_do(trans, NULL, NULL,
-				BTREE_INSERT_NOFAIL|
+				BCH_TRANS_COMMIT_no_enospc|
-				BTREE_INSERT_LAZY_RW,
+				BCH_TRANS_COMMIT_lazy_rw,
 			bch2_btree_insert_trans(trans, BTREE_ID_bucket_gens, &g.k_i, 0));
 
 	bch2_trans_put(trans);
@@ -1267,7 +1267,7 @@ delete:
 	ret =   bch2_btree_delete_extent_at(trans, iter,
 			iter->btree_id == BTREE_ID_freespace ? 1 : 0, 0) ?:
 		bch2_trans_commit(trans, NULL, NULL,
-			BTREE_INSERT_NOFAIL|BTREE_INSERT_LAZY_RW);
+			BCH_TRANS_COMMIT_no_enospc|BCH_TRANS_COMMIT_lazy_rw);
 	goto out;
 }
 
@@ -1422,8 +1422,8 @@ int bch2_check_alloc_info(struct bch_fs *c)
 		}
 
 		ret = bch2_trans_commit(trans, NULL, NULL,
-					BTREE_INSERT_NOFAIL|
+					BCH_TRANS_COMMIT_no_enospc|
-					BTREE_INSERT_LAZY_RW);
+					BCH_TRANS_COMMIT_lazy_rw);
 		if (ret)
 			goto bkey_err;
 
@@ -1453,7 +1453,7 @@ bkey_err:
 	      for_each_btree_key_commit(trans, iter,
 			BTREE_ID_bucket_gens, POS_MIN,
 			BTREE_ITER_PREFETCH, k,
-			NULL, NULL, BTREE_INSERT_NOFAIL|BTREE_INSERT_LAZY_RW,
+			NULL, NULL, BCH_TRANS_COMMIT_no_enospc|BCH_TRANS_COMMIT_lazy_rw,
 		bch2_check_bucket_gens_key(trans, &iter, k));
 err:
 	bch2_trans_put(trans);
@@ -1546,7 +1546,7 @@ int bch2_check_alloc_to_lru_refs(struct bch_fs *c)
 	ret = bch2_trans_run(c,
 		for_each_btree_key_commit(trans, iter, BTREE_ID_alloc,
 				POS_MIN, BTREE_ITER_PREFETCH, k,
-				NULL, NULL, BTREE_INSERT_NOFAIL|BTREE_INSERT_LAZY_RW,
+				NULL, NULL, BCH_TRANS_COMMIT_no_enospc|BCH_TRANS_COMMIT_lazy_rw,
 			bch2_check_alloc_to_lru_ref(trans, &iter)));
 	if (ret)
 		bch_err_fn(c, ret);
@@ -1655,7 +1655,7 @@ write:
 	ret =   bch2_trans_update(trans, &iter, &a->k_i, 0) ?:
 		bch2_trans_commit(trans, NULL, NULL,
 				  BCH_WATERMARK_btree|
-				  BTREE_INSERT_NOFAIL);
+				  BCH_TRANS_COMMIT_no_enospc);
 	if (ret)
 		goto out;
 
@@ -1760,7 +1760,7 @@ static int invalidate_one_bucket(struct btree_trans *trans,
 				BTREE_TRIGGER_BUCKET_INVALIDATE) ?:
 		bch2_trans_commit(trans, NULL, NULL,
 				  BCH_WATERMARK_btree|
-				  BTREE_INSERT_NOFAIL);
+				  BCH_TRANS_COMMIT_no_enospc);
 	if (ret)
 		goto out;
 
@@ -1884,8 +1884,8 @@ int bch2_dev_freespace_init(struct bch_fs *c, struct bch_dev *ca,
 
 			ret =   bch2_bucket_do_index(trans, k, a, true) ?:
 				bch2_trans_commit(trans, NULL, NULL,
-						  BTREE_INSERT_LAZY_RW|
+						  BCH_TRANS_COMMIT_lazy_rw|
-						  BTREE_INSERT_NOFAIL);
+						  BCH_TRANS_COMMIT_no_enospc);
 			if (ret)
 				goto bkey_err;
 
@@ -1905,8 +1905,8 @@ int bch2_dev_freespace_init(struct bch_fs *c, struct bch_dev *ca,
 
 			ret = bch2_btree_insert_trans(trans, BTREE_ID_freespace, freespace, 0) ?:
 				bch2_trans_commit(trans, NULL, NULL,
-						  BTREE_INSERT_LAZY_RW|
+						  BCH_TRANS_COMMIT_lazy_rw|
-						  BTREE_INSERT_NOFAIL);
+						  BCH_TRANS_COMMIT_no_enospc);
 			if (ret)
 				goto bkey_err;
 

libbcachefs/backpointers.c

@@ -5,6 +5,7 @@
 #include "backpointers.h"
 #include "btree_cache.h"
 #include "btree_update.h"
+#include "btree_update_interior.h"
 #include "btree_write_buffer.h"
 #include "error.h"
 
@@ -220,18 +221,22 @@ out:
 static void backpointer_not_found(struct btree_trans *trans,
 				  struct bpos bp_pos,
 				  struct bch_backpointer bp,
-				  struct bkey_s_c k,
+				  struct bkey_s_c k)
-				  const char *thing_it_points_to)
 {
 	struct bch_fs *c = trans->c;
 	struct printbuf buf = PRINTBUF;
 	struct bpos bucket = bp_pos_to_bucket(c, bp_pos);
 
+	/*
+	 * If we're using the btree write buffer, the backpointer we were
+	 * looking at may have already been deleted - failure to find what it
+	 * pointed to is not an error:
+	 */
 	if (likely(!bch2_backpointers_no_use_write_buffer))
 		return;
 
 	prt_printf(&buf, "backpointer doesn't match %s it points to:\n  ",
-		   thing_it_points_to);
+		   bp.level ? "btree node" : "extent");
 	prt_printf(&buf, "bucket: ");
 	bch2_bpos_to_text(&buf, bucket);
 	prt_printf(&buf, "\n  ");
@@ -257,56 +262,37 @@ struct bkey_s_c bch2_backpointer_get_key(struct btree_trans *trans,
 					 struct bch_backpointer bp,
 					 unsigned iter_flags)
 {
-	struct bch_fs *c = trans->c;
+	if (likely(!bp.level)) {
-	struct btree_root *r = bch2_btree_id_root(c, bp.btree_id);
+		struct bch_fs *c = trans->c;
-	struct bpos bucket = bp_pos_to_bucket(c, bp_pos);
+		struct bpos bucket = bp_pos_to_bucket(c, bp_pos);
-	struct bkey_s_c k;
+		struct bkey_s_c k;
-
-	bch2_trans_node_iter_init(trans, iter,
-				  bp.btree_id,
-				  bp.pos,
-				  0,
-				  min(bp.level, r->level),
-				  iter_flags);
-	k = bch2_btree_iter_peek_slot(iter);
-	if (bkey_err(k)) {
-		bch2_trans_iter_exit(trans, iter);
-		return k;
-	}
-
-	if (bp.level == r->level + 1)
-		k = bkey_i_to_s_c(&r->key);
-
-	if (k.k && extent_matches_bp(c, bp.btree_id, bp.level, k, bucket, bp))
-		return k;
-
-	bch2_trans_iter_exit(trans, iter);
-
-	if (unlikely(bch2_backpointers_no_use_write_buffer)) {
-		if (bp.level) {
-			struct btree *b;
-
-			/*
-			 * If a backpointer for a btree node wasn't found, it may be
-			 * because it was overwritten by a new btree node that hasn't
-			 * been written out yet - backpointer_get_node() checks for
-			 * this:
-			 */
-			b = bch2_backpointer_get_node(trans, iter, bp_pos, bp);
-			if (!IS_ERR_OR_NULL(b))
-				return bkey_i_to_s_c(&b->key);
 
+		bch2_trans_node_iter_init(trans, iter,
+					  bp.btree_id,
+					  bp.pos,
+					  0, 0,
+					  iter_flags);
+		k = bch2_btree_iter_peek_slot(iter);
+		if (bkey_err(k)) {
 			bch2_trans_iter_exit(trans, iter);
-
+			return k;
-			if (IS_ERR(b))
-				return bkey_s_c_err(PTR_ERR(b));
-			return bkey_s_c_null;
 		}
 
-		backpointer_not_found(trans, bp_pos, bp, k, "extent");
+		if (k.k && extent_matches_bp(c, bp.btree_id, bp.level, k, bucket, bp))
-	}
+			return k;
 
-	return bkey_s_c_null;
+		bch2_trans_iter_exit(trans, iter);
+		backpointer_not_found(trans, bp_pos, bp, k);
+		return bkey_s_c_null;
+	} else {
+		struct btree *b = bch2_backpointer_get_node(trans, iter, bp_pos, bp);
+
+		if (IS_ERR_OR_NULL(b)) {
+			bch2_trans_iter_exit(trans, iter);
+			return IS_ERR(b) ? bkey_s_c_err(PTR_ERR(b)) : bkey_s_c_null;
+		}
+		return bkey_i_to_s_c(&b->key);
+	}
 }
 
 struct btree *bch2_backpointer_get_node(struct btree_trans *trans,
@@ -327,19 +313,20 @@ struct btree *bch2_backpointer_get_node(struct btree_trans *trans,
 				  bp.level - 1,
 				  0);
 	b = bch2_btree_iter_peek_node(iter);
-	if (IS_ERR(b))
+	if (IS_ERR_OR_NULL(b))
 		goto err;
 
-	if (b && extent_matches_bp(c, bp.btree_id, bp.level,
+	BUG_ON(b->c.level != bp.level - 1);
-				   bkey_i_to_s_c(&b->key),
+
-				   bucket, bp))
+	if (extent_matches_bp(c, bp.btree_id, bp.level,
+			      bkey_i_to_s_c(&b->key),
+			      bucket, bp))
 		return b;
 
-	if (b && btree_node_will_make_reachable(b)) {
+	if (btree_node_will_make_reachable(b)) {
 		b = ERR_PTR(-BCH_ERR_backpointer_to_overwritten_btree_node);
 	} else {
-		backpointer_not_found(trans, bp_pos, bp,
+		backpointer_not_found(trans, bp_pos, bp, bkey_i_to_s_c(&b->key));
-				      bkey_i_to_s_c(&b->key), "btree node");
 		b = NULL;
 	}
 err:
@@ -395,7 +382,7 @@ int bch2_check_btree_backpointers(struct bch_fs *c)
 	ret = bch2_trans_run(c,
 		for_each_btree_key_commit(trans, iter,
 			BTREE_ID_backpointers, POS_MIN, 0, k,
-			NULL, NULL, BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
+			NULL, NULL, BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc,
 		  bch2_check_btree_backpointer(trans, &iter, k)));
 	if (ret)
 		bch_err_fn(c, ret);
@@ -642,8 +629,8 @@ static int bch2_check_extents_to_backpointers_pass(struct btree_trans *trans,
 
 		do {
 			ret = commit_do(trans, NULL, NULL,
-					BTREE_INSERT_LAZY_RW|
+					BCH_TRANS_COMMIT_lazy_rw|
-					BTREE_INSERT_NOFAIL,
+					BCH_TRANS_COMMIT_no_enospc,
 					check_extent_to_backpointers(trans, &iter,
 								bucket_start, bucket_end,
 								&last_flushed));
@@ -657,8 +644,8 @@ static int bch2_check_extents_to_backpointers_pass(struct btree_trans *trans,
 			break;
 
 		ret = commit_do(trans, NULL, NULL,
-				BTREE_INSERT_LAZY_RW|
+				BCH_TRANS_COMMIT_lazy_rw|
-				BTREE_INSERT_NOFAIL,
+				BCH_TRANS_COMMIT_no_enospc,
 				check_btree_root_to_backpointers(trans, btree_id,
 							bucket_start, bucket_end,
 							&last_flushed));
@@ -797,7 +784,8 @@ static int check_one_backpointer(struct btree_trans *trans,
 
 	if (fsck_err_on(!k.k, c,
 			backpointer_to_missing_ptr,
-			"backpointer for missing extent\n  %s",
+			"backpointer for missing %s\n  %s",
+			bp.v->level ? "btree node" : "extent",
 			(bch2_bkey_val_to_text(&buf, c, bp.s_c), buf.buf))) {
 		ret = bch2_btree_delete_at_buffered(trans, BTREE_ID_backpointers, bp.k->p);
 		goto out;
@@ -819,7 +807,7 @@ static int bch2_check_backpointers_to_extents_pass(struct btree_trans *trans,
 
 	return for_each_btree_key_commit(trans, iter, BTREE_ID_backpointers,
 				  POS_MIN, BTREE_ITER_PREFETCH, k,
-				  NULL, NULL, BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
+				  NULL, NULL, BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc,
 		check_one_backpointer(trans, start, end,
 				      bkey_s_c_to_backpointer(k),
 				      &last_flushed_pos));

libbcachefs/bcachefs.h

@@ -401,7 +401,9 @@ BCH_DEBUG_PARAMS_DEBUG()
 	x(journal_flush_write)			\
 	x(journal_noflush_write)		\
 	x(journal_flush_seq)			\
-	x(blocked_journal)			\
+	x(blocked_journal_low_on_space)		\
+	x(blocked_journal_low_on_pin)		\
+	x(blocked_journal_max_in_flight)	\
 	x(blocked_allocate)			\
 	x(blocked_allocate_open_bucket)		\
 	x(nocow_lock_contended)
@@ -617,7 +619,7 @@ struct journal_seq_blacklist_table {
 		u64		start;
 		u64		end;
 		bool		dirty;
-	}			entries[0];
+	}			entries[];
 };
 
 struct journal_keys {

libbcachefs/bcachefs_format.h

@@ -2256,7 +2256,8 @@ LE32_BITMASK(JSET_NO_FLUSH,	struct jset, flags, 5, 6);
 enum btree_id_flags {
 	BTREE_ID_EXTENTS	= BIT(0),
 	BTREE_ID_SNAPSHOTS	= BIT(1),
-	BTREE_ID_DATA		= BIT(2),
+	BTREE_ID_SNAPSHOT_FIELD	= BIT(2),
+	BTREE_ID_DATA		= BIT(3),
 };
 
 #define BCH_BTREE_IDS()								\
@@ -2311,12 +2312,12 @@ enum btree_id_flags {
 	  BIT_ULL(KEY_TYPE_bucket_gens))					\
 	x(snapshot_trees,	15,	0,					\
 	  BIT_ULL(KEY_TYPE_snapshot_tree))					\
-	x(deleted_inodes,	16,	BTREE_ID_SNAPSHOTS,			\
+	x(deleted_inodes,	16,	BTREE_ID_SNAPSHOT_FIELD,		\
 	  BIT_ULL(KEY_TYPE_set))						\
 	x(logged_ops,		17,	0,					\
 	  BIT_ULL(KEY_TYPE_logged_op_truncate)|					\
 	  BIT_ULL(KEY_TYPE_logged_op_finsert))					\
-	x(rebalance_work,	18,	BTREE_ID_SNAPSHOTS,			\
+	x(rebalance_work,	18,	BTREE_ID_SNAPSHOT_FIELD,		\
 	  BIT_ULL(KEY_TYPE_set)|BIT_ULL(KEY_TYPE_cookie))
 
 enum btree_id {

libbcachefs/bkey_methods.c

@@ -186,15 +186,20 @@ int __bch2_bkey_invalid(struct bch_fs *c, struct bkey_s_c k,
 	if (type != BKEY_TYPE_btree) {
 		enum btree_id btree = type - 1;
 
-		bkey_fsck_err_on(!btree_type_has_snapshots(btree) &&
+		if (btree_type_has_snapshots(btree)) {
-				 k.k->p.snapshot, c, err,
+			bkey_fsck_err_on(!k.k->p.snapshot, c, err,
-				 bkey_snapshot_nonzero,
+					 bkey_snapshot_zero,
-				 "nonzero snapshot");
+					 "snapshot == 0");
-
+		} else if (!btree_type_has_snapshot_field(btree)) {
-		bkey_fsck_err_on(btree_type_has_snapshots(btree) &&
+			bkey_fsck_err_on(k.k->p.snapshot, c, err,
-				 !k.k->p.snapshot, c, err,
+					 bkey_snapshot_nonzero,
-				 bkey_snapshot_zero,
+					 "nonzero snapshot");
-				 "snapshot == 0");
+		} else {
+			/*
+			 * btree uses snapshot field but it's not required to be
+			 * nonzero
+			 */
+		}
 
 		bkey_fsck_err_on(bkey_eq(k.k->p, POS_MAX), c, err,
 				 bkey_at_pos_max,

libbcachefs/bkey_methods.h

@@ -93,7 +93,6 @@ static inline int bch2_mark_key(struct btree_trans *trans,
 enum btree_update_flags {
 	__BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE = __BTREE_ITER_FLAGS_END,
 	__BTREE_UPDATE_NOJOURNAL,
-	__BTREE_UPDATE_PREJOURNAL,
 	__BTREE_UPDATE_KEY_CACHE_RECLAIM,
 
 	__BTREE_TRIGGER_NORUN,		/* Don't run triggers at all */
@@ -108,7 +107,6 @@ enum btree_update_flags {
 
 #define BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE (1U << __BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE)
 #define BTREE_UPDATE_NOJOURNAL		(1U << __BTREE_UPDATE_NOJOURNAL)
-#define BTREE_UPDATE_PREJOURNAL		(1U << __BTREE_UPDATE_PREJOURNAL)
 #define BTREE_UPDATE_KEY_CACHE_RECLAIM	(1U << __BTREE_UPDATE_KEY_CACHE_RECLAIM)
 
 #define BTREE_TRIGGER_NORUN		(1U << __BTREE_TRIGGER_NORUN)

libbcachefs/btree_gc.c

@@ -1502,7 +1502,7 @@ static int bch2_gc_alloc_done(struct bch_fs *c, bool metadata_only)
 		ret = for_each_btree_key_commit(trans, iter, BTREE_ID_alloc,
 				POS(ca->dev_idx, ca->mi.first_bucket),
 				BTREE_ITER_SLOTS|BTREE_ITER_PREFETCH, k,
-				NULL, NULL, BTREE_INSERT_LAZY_RW,
+				NULL, NULL, BCH_TRANS_COMMIT_lazy_rw,
 			bch2_alloc_write_key(trans, &iter, k, metadata_only));
 
 		if (ret < 0) {
@@ -1659,7 +1659,7 @@ static int bch2_gc_reflink_done(struct bch_fs *c, bool metadata_only)
 	ret = for_each_btree_key_commit(trans, iter,
 			BTREE_ID_reflink, POS_MIN,
 			BTREE_ITER_PREFETCH, k,
-			NULL, NULL, BTREE_INSERT_NOFAIL,
+			NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 		bch2_gc_write_reflink_key(trans, &iter, k, &idx));
 
 	c->reflink_gc_nr = 0;
@@ -1783,7 +1783,7 @@ static int bch2_gc_stripes_done(struct bch_fs *c, bool metadata_only)
 	ret = for_each_btree_key_commit(trans, iter,
 			BTREE_ID_stripes, POS_MIN,
 			BTREE_ITER_PREFETCH, k,
-			NULL, NULL, BTREE_INSERT_NOFAIL,
+			NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 		bch2_gc_write_stripes_key(trans, &iter, k));
 
 	bch2_trans_put(trans);
@@ -2019,7 +2019,7 @@ int bch2_gc_gens(struct bch_fs *c)
 					BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS,
 					k,
 					NULL, NULL,
-					BTREE_INSERT_NOFAIL,
+					BCH_TRANS_COMMIT_no_enospc,
 				gc_btree_gens_key(trans, &iter, k));
 			if (ret && !bch2_err_matches(ret, EROFS))
 				bch_err_fn(c, ret);
@@ -2032,7 +2032,7 @@ int bch2_gc_gens(struct bch_fs *c)
 			BTREE_ITER_PREFETCH,
 			k,
 			NULL, NULL,
-			BTREE_INSERT_NOFAIL,
+			BCH_TRANS_COMMIT_no_enospc,
 		bch2_alloc_write_oldest_gen(trans, &iter, k));
 	if (ret && !bch2_err_matches(ret, EROFS))
 		bch_err_fn(c, ret);

libbcachefs/btree_io.c

@@ -1801,9 +1801,9 @@ static void btree_node_write_work(struct work_struct *work)
 		ret = bch2_trans_do(c, NULL, NULL, 0,
 			bch2_btree_node_update_key_get_iter(trans, b, &wbio->key,
 					BCH_WATERMARK_reclaim|
-					BTREE_INSERT_JOURNAL_RECLAIM|
+					BCH_TRANS_COMMIT_journal_reclaim|
-					BTREE_INSERT_NOFAIL|
+					BCH_TRANS_COMMIT_no_enospc|
-					BTREE_INSERT_NOCHECK_RW,
+					BCH_TRANS_COMMIT_no_check_rw,
 					!wbio->wbio.failed.nr));
 		if (ret)
 			goto err;

libbcachefs/btree_iter.c

@@ -257,7 +257,7 @@ static void bch2_btree_iter_verify(struct btree_iter *iter)
 
 	BUG_ON(!(iter->flags & __BTREE_ITER_ALL_SNAPSHOTS) &&
 	       (iter->flags & BTREE_ITER_ALL_SNAPSHOTS) &&
-	       !btree_type_has_snapshots(iter->btree_id));
+	       !btree_type_has_snapshot_field(iter->btree_id));
 
 	if (iter->update_path)
 		bch2_btree_path_verify(trans, iter->update_path);
@@ -1214,8 +1214,6 @@ __bch2_btree_path_set_pos(struct btree_trans *trans,
 		   struct btree_path *path, struct bpos new_pos,
 		   bool intent, unsigned long ip, int cmp)
 {
-	unsigned level = path->level;
-
 	bch2_trans_verify_not_in_restart(trans);
 	EBUG_ON(!path->ref);
 
@@ -1231,7 +1229,7 @@ __bch2_btree_path_set_pos(struct btree_trans *trans,
 		goto out;
 	}
 
-	level = btree_path_up_until_good_node(trans, path, cmp);
+	unsigned level = btree_path_up_until_good_node(trans, path, cmp);
 
 	if (btree_path_node(path, level)) {
 		struct btree_path_level *l = &path->l[level];
@@ -2835,8 +2833,9 @@ void *__bch2_trans_kmalloc(struct btree_trans *trans, size_t size)
 
 static inline void check_srcu_held_too_long(struct btree_trans *trans)
 {
-	WARN(time_after(jiffies, trans->srcu_lock_time + HZ * 10),
+	WARN(trans->srcu_held && time_after(jiffies, trans->srcu_lock_time + HZ * 10),
-	     "btree trans held srcu lock (delaying memory reclaim) by more than 10 seconds");
+	     "btree trans held srcu lock (delaying memory reclaim) for %lu seconds",
+	     (jiffies - trans->srcu_lock_time) / HZ);
 }
 
 void bch2_trans_srcu_unlock(struct btree_trans *trans)
@@ -3088,8 +3087,6 @@ void bch2_trans_put(struct btree_trans *trans)
 		srcu_read_unlock(&c->btree_trans_barrier, trans->srcu_idx);
 	}
 
-	bch2_journal_preres_put(&c->journal, &trans->journal_preres);
-
 	kfree(trans->extra_journal_entries.data);
 
 	if (trans->fs_usage_deltas) {

libbcachefs/btree_iter.h

@@ -416,7 +416,7 @@ static inline unsigned __bch2_btree_iter_flags(struct btree_trans *trans,
 		flags |= BTREE_ITER_IS_EXTENTS;
 
 	if (!(flags & __BTREE_ITER_ALL_SNAPSHOTS) &&
-	    !btree_type_has_snapshots(btree_id))
+	    !btree_type_has_snapshot_field(btree_id))
 		flags &= ~BTREE_ITER_ALL_SNAPSHOTS;
 
 	if (!(flags & BTREE_ITER_ALL_SNAPSHOTS) &&

libbcachefs/btree_key_cache.c

@@ -90,10 +90,13 @@ static void bkey_cached_free(struct btree_key_cache *bc,
 	ck->btree_trans_barrier_seq =
 		start_poll_synchronize_srcu(&c->btree_trans_barrier);
 
-	if (ck->c.lock.readers)
+	if (ck->c.lock.readers) {
 		list_move_tail(&ck->list, &bc->freed_pcpu);
-	else
+		bc->nr_freed_pcpu++;
+	} else {
 		list_move_tail(&ck->list, &bc->freed_nonpcpu);
+		bc->nr_freed_nonpcpu++;
+	}
 	atomic_long_inc(&bc->nr_freed);
 
 	kfree(ck->k);
@@ -110,6 +113,8 @@ static void __bkey_cached_move_to_freelist_ordered(struct btree_key_cache *bc,
 {
 	struct bkey_cached *pos;
 
+	bc->nr_freed_nonpcpu++;
+
 	list_for_each_entry_reverse(pos, &bc->freed_nonpcpu, list) {
 		if (ULONG_CMP_GE(ck->btree_trans_barrier_seq,
 				 pos->btree_trans_barrier_seq)) {
@@ -159,6 +164,7 @@ static void bkey_cached_move_to_freelist(struct btree_key_cache *bc,
 #else
 		mutex_lock(&bc->lock);
 		list_move_tail(&ck->list, &bc->freed_nonpcpu);
+		bc->nr_freed_nonpcpu++;
 		mutex_unlock(&bc->lock);
 #endif
 	} else {
@@ -218,6 +224,7 @@ bkey_cached_alloc(struct btree_trans *trans, struct btree_path *path,
 			       f->nr < ARRAY_SIZE(f->objs) / 2) {
 				ck = list_last_entry(&bc->freed_nonpcpu, struct bkey_cached, list);
 				list_del_init(&ck->list);
+				bc->nr_freed_nonpcpu--;
 				f->objs[f->nr++] = ck;
 			}
 
@@ -230,6 +237,7 @@ bkey_cached_alloc(struct btree_trans *trans, struct btree_path *path,
 		if (!list_empty(&bc->freed_nonpcpu)) {
 			ck = list_last_entry(&bc->freed_nonpcpu, struct bkey_cached, list);
 			list_del_init(&ck->list);
+			bc->nr_freed_nonpcpu--;
 		}
 		mutex_unlock(&bc->lock);
 #endif
@@ -649,8 +657,8 @@ static int btree_key_cache_flush_pos(struct btree_trans *trans,
 				  BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE|
 				  BTREE_TRIGGER_NORUN) ?:
 		bch2_trans_commit(trans, NULL, NULL,
-				  BTREE_INSERT_NOCHECK_RW|
+				  BCH_TRANS_COMMIT_no_check_rw|
-				  BTREE_INSERT_NOFAIL|
+				  BCH_TRANS_COMMIT_no_enospc|
 				  (ck->journal.seq == journal_last_seq(j)
 				   ? BCH_WATERMARK_reclaim
 				   : 0)|
@@ -665,7 +673,6 @@ static int btree_key_cache_flush_pos(struct btree_trans *trans,
 		goto out;
 
 	bch2_journal_pin_drop(j, &ck->journal);
-	bch2_journal_preres_put(j, &ck->res);
 
 	BUG_ON(!btree_node_locked(c_iter.path, 0));
 
@@ -728,7 +735,7 @@ int bch2_btree_key_cache_journal_flush(struct journal *j,
 
 	ret = commit_do(trans, NULL, NULL, 0,
 		btree_key_cache_flush_pos(trans, key, seq,
-				BTREE_INSERT_JOURNAL_RECLAIM, false));
+				BCH_TRANS_COMMIT_journal_reclaim, false));
 unlock:
 	srcu_read_unlock(&c->btree_trans_barrier, srcu_idx);
 
@@ -763,18 +770,6 @@ bool bch2_btree_insert_key_cached(struct btree_trans *trans,
 
 	BUG_ON(insert->k.u64s > ck->u64s);
 
-	if (likely(!(flags & BTREE_INSERT_JOURNAL_REPLAY))) {
-		int difference;
-
-		BUG_ON(jset_u64s(insert->k.u64s) > trans->journal_preres.u64s);
-
-		difference = jset_u64s(insert->k.u64s) - ck->res.u64s;
-		if (difference > 0) {
-			trans->journal_preres.u64s	-= difference;
-			ck->res.u64s			+= difference;
-		}
-	}
-
 	bkey_copy(ck->k, insert);
 	ck->valid = true;
 
@@ -852,6 +847,8 @@ static unsigned long bch2_btree_key_cache_scan(struct shrinker *shrink,
 	 * Newest freed entries are at the end of the list - once we hit one
 	 * that's too new to be freed, we can bail out:
 	 */
+	scanned += bc->nr_freed_nonpcpu;
+
 	list_for_each_entry_safe(ck, t, &bc->freed_nonpcpu, list) {
 		if (!poll_state_synchronize_srcu(&c->btree_trans_barrier,
 						 ck->btree_trans_barrier_seq))
@@ -861,13 +858,15 @@ static unsigned long bch2_btree_key_cache_scan(struct shrinker *shrink,
 		six_lock_exit(&ck->c.lock);
 		kmem_cache_free(bch2_key_cache, ck);
 		atomic_long_dec(&bc->nr_freed);
-		scanned++;
 		freed++;
+		bc->nr_freed_nonpcpu--;
 	}
 
 	if (scanned >= nr)
 		goto out;
 
+	scanned += bc->nr_freed_pcpu;
+
 	list_for_each_entry_safe(ck, t, &bc->freed_pcpu, list) {
 		if (!poll_state_synchronize_srcu(&c->btree_trans_barrier,
 						 ck->btree_trans_barrier_seq))
@@ -877,8 +876,8 @@ static unsigned long bch2_btree_key_cache_scan(struct shrinker *shrink,
 		six_lock_exit(&ck->c.lock);
 		kmem_cache_free(bch2_key_cache, ck);
 		atomic_long_dec(&bc->nr_freed);
-		scanned++;
 		freed++;
+		bc->nr_freed_pcpu--;
 	}
 
 	if (scanned >= nr)
@@ -985,6 +984,9 @@ void bch2_fs_btree_key_cache_exit(struct btree_key_cache *bc)
 	}
 #endif
 
+	BUG_ON(list_count_nodes(&bc->freed_pcpu) != bc->nr_freed_pcpu);
+	BUG_ON(list_count_nodes(&bc->freed_nonpcpu) != bc->nr_freed_nonpcpu);
+
 	list_splice(&bc->freed_pcpu,	&items);
 	list_splice(&bc->freed_nonpcpu,	&items);
 
@@ -994,7 +996,6 @@ void bch2_fs_btree_key_cache_exit(struct btree_key_cache *bc)
 		cond_resched();
 
 		bch2_journal_pin_drop(&c->journal, &ck->journal);
-		bch2_journal_preres_put(&c->journal, &ck->res);
 
 		list_del(&ck->list);
 		kfree(ck->k);

libbcachefs/btree_key_cache_types.h

@@ -0,0 +1,34 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _BCACHEFS_BTREE_KEY_CACHE_TYPES_H
+#define _BCACHEFS_BTREE_KEY_CACHE_TYPES_H
+
+struct btree_key_cache_freelist {
+	struct bkey_cached	*objs[16];
+	unsigned		nr;
+};
+
+struct btree_key_cache {
+	struct mutex		lock;
+	struct rhashtable	table;
+	bool			table_init_done;
+
+	struct list_head	freed_pcpu;
+	size_t			nr_freed_pcpu;
+	struct list_head	freed_nonpcpu;
+	size_t			nr_freed_nonpcpu;
+
+	struct shrinker		shrink;
+	unsigned		shrink_iter;
+	struct btree_key_cache_freelist __percpu *pcpu_freed;
+
+	atomic_long_t		nr_freed;
+	atomic_long_t		nr_keys;
+	atomic_long_t		nr_dirty;
+};
+
+struct bkey_cached_key {
+	u32			btree_id;
+	struct bpos		pos;
+} __packed __aligned(4);
+
+#endif /* _BCACHEFS_BTREE_KEY_CACHE_TYPES_H */

libbcachefs/btree_trans_commit.c

@@ -78,6 +78,53 @@ inline void bch2_btree_node_prep_for_write(struct btree_trans *trans,
 		bch2_btree_init_next(trans, b);
 }
 
+static noinline int trans_lock_write_fail(struct btree_trans *trans, struct btree_insert_entry *i)
+{
+	while (--i >= trans->updates) {
+		if (same_leaf_as_prev(trans, i))
+			continue;
+
+		bch2_btree_node_unlock_write(trans, i->path, insert_l(i)->b);
+	}
+
+	trace_and_count(trans->c, trans_restart_would_deadlock_write, trans);
+	return btree_trans_restart(trans, BCH_ERR_transaction_restart_would_deadlock_write);
+}
+
+static inline int bch2_trans_lock_write(struct btree_trans *trans)
+{
+	struct btree_insert_entry *i;
+
+	EBUG_ON(trans->write_locked);
+
+	trans_for_each_update(trans, i) {
+		if (same_leaf_as_prev(trans, i))
+			continue;
+
+		if (bch2_btree_node_lock_write(trans, i->path, &insert_l(i)->b->c))
+			return trans_lock_write_fail(trans, i);
+
+		if (!i->cached)
+			bch2_btree_node_prep_for_write(trans, i->path, insert_l(i)->b);
+	}
+
+	trans->write_locked = true;
+	return 0;
+}
+
+static inline void bch2_trans_unlock_write(struct btree_trans *trans)
+{
+	if (likely(trans->write_locked)) {
+		struct btree_insert_entry *i;
+
+		trans_for_each_update(trans, i)
+			if (!same_leaf_as_prev(trans, i))
+				bch2_btree_node_unlock_write_inlined(trans, i->path,
+								     insert_l(i)->b);
+		trans->write_locked = false;
+	}
+}
+
 /* Inserting into a given leaf node (last stage of insert): */
 
 /* Handle overwrites and do insert, for non extents: */
@@ -269,23 +316,13 @@ static inline void btree_insert_entry_checks(struct btree_trans *trans,
 	BUG_ON(i->level		!= i->path->level);
 	BUG_ON(i->btree_id	!= i->path->btree_id);
 	EBUG_ON(!i->level &&
+		btree_type_has_snapshots(i->btree_id) &&
 		!(i->flags & BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE) &&
 		test_bit(JOURNAL_REPLAY_DONE, &trans->c->journal.flags) &&
 		i->k->k.p.snapshot &&
 		bch2_snapshot_is_internal_node(trans->c, i->k->k.p.snapshot));
 }
 
-static noinline int
-bch2_trans_journal_preres_get_cold(struct btree_trans *trans, unsigned flags,
-				   unsigned long trace_ip)
-{
-	return drop_locks_do(trans,
-		bch2_journal_preres_get(&trans->c->journal,
-			&trans->journal_preres,
-			trans->journal_preres_u64s,
-			(flags & BCH_WATERMARK_MASK)));
-}
-
 static __always_inline int bch2_trans_journal_res_get(struct btree_trans *trans,
 						      unsigned flags)
 {
@@ -320,6 +357,45 @@ static inline int btree_key_can_insert(struct btree_trans *trans,
 	return 0;
 }
 
+noinline static int
+btree_key_can_insert_cached_slowpath(struct btree_trans *trans, unsigned flags,
+				     struct btree_path *path, unsigned new_u64s)
+{
+	struct bch_fs *c = trans->c;
+	struct btree_insert_entry *i;
+	struct bkey_cached *ck = (void *) path->l[0].b;
+	struct bkey_i *new_k;
+	int ret;
+
+	bch2_trans_unlock_write(trans);
+	bch2_trans_unlock(trans);
+
+	new_k = kmalloc(new_u64s * sizeof(u64), GFP_KERNEL);
+	if (!new_k) {
+		bch_err(c, "error allocating memory for key cache key, btree %s u64s %u",
+			bch2_btree_id_str(path->btree_id), new_u64s);
+		return -BCH_ERR_ENOMEM_btree_key_cache_insert;
+	}
+
+	ret =   bch2_trans_relock(trans) ?:
+		bch2_trans_lock_write(trans);
+	if (unlikely(ret)) {
+		kfree(new_k);
+		return ret;
+	}
+
+	memcpy(new_k, ck->k, ck->u64s * sizeof(u64));
+
+	trans_for_each_update(trans, i)
+		if (i->old_v == &ck->k->v)
+			i->old_v = &new_k->v;
+
+	kfree(ck->k);
+	ck->u64s	= new_u64s;
+	ck->k		= new_k;
+	return 0;
+}
+
 static int btree_key_can_insert_cached(struct btree_trans *trans, unsigned flags,
 				       struct btree_path *path, unsigned u64s)
 {
@@ -333,7 +409,7 @@ static int btree_key_can_insert_cached(struct btree_trans *trans, unsigned flags
 
 	if (!test_bit(BKEY_CACHED_DIRTY, &ck->flags) &&
 	    bch2_btree_key_cache_must_wait(c) &&
-	    !(flags & BTREE_INSERT_JOURNAL_RECLAIM))
+	    !(flags & BCH_TRANS_COMMIT_journal_reclaim))
 		return -BCH_ERR_btree_insert_need_journal_reclaim;
 
 	/*
@@ -346,12 +422,9 @@ static int btree_key_can_insert_cached(struct btree_trans *trans, unsigned flags
 		return 0;
 
 	new_u64s	= roundup_pow_of_two(u64s);
-	new_k		= krealloc(ck->k, new_u64s * sizeof(u64), GFP_NOFS);
+	new_k		= krealloc(ck->k, new_u64s * sizeof(u64), GFP_NOWAIT);
-	if (!new_k) {
+	if (unlikely(!new_k))
-		bch_err(c, "error allocating memory for key cache key, btree %s u64s %u",
+		return btree_key_can_insert_cached_slowpath(trans, flags, path, new_u64s);
-			bch2_btree_id_str(path->btree_id), new_u64s);
-		return -BCH_ERR_ENOMEM_btree_key_cache_insert;
-	}
 
 	trans_for_each_update(trans, i)
 		if (i->old_v == &ck->k->v)
@@ -583,6 +656,8 @@ bch2_trans_commit_write_locked(struct btree_trans *trans, unsigned flags,
 			*stopped_at = i;
 			return ret;
 		}
+
+		i->k->k.needs_whiteout = false;
 	}
 
 	if (trans->nr_wb_updates &&
@@ -593,7 +668,7 @@ bch2_trans_commit_write_locked(struct btree_trans *trans, unsigned flags,
 	 * Don't get journal reservation until after we know insert will
 	 * succeed:
 	 */
-	if (likely(!(flags & BTREE_INSERT_JOURNAL_REPLAY))) {
+	if (likely(!(flags & BCH_TRANS_COMMIT_no_journal_res))) {
 		ret = bch2_trans_journal_res_get(trans,
 				(flags & BCH_WATERMARK_MASK)|
 				JOURNAL_RES_GET_NONBLOCK);
@@ -602,8 +677,6 @@ bch2_trans_commit_write_locked(struct btree_trans *trans, unsigned flags,
 
 		if (unlikely(trans->journal_transaction_names))
 			journal_transaction_name(trans);
-	} else {
-		trans->journal_res.seq = c->journal.replay_journal_seq;
 	}
 
 	/*
@@ -612,7 +685,7 @@ bch2_trans_commit_write_locked(struct btree_trans *trans, unsigned flags,
 	 */
 
 	if (IS_ENABLED(CONFIG_BCACHEFS_DEBUG) &&
-	    !(flags & BTREE_INSERT_JOURNAL_REPLAY)) {
+	    !(flags & BCH_TRANS_COMMIT_no_journal_res)) {
 		if (bch2_journal_seq_verify)
 			trans_for_each_update(trans, i)
 				i->k->k.version.lo = trans->journal_res.seq;
@@ -626,7 +699,7 @@ bch2_trans_commit_write_locked(struct btree_trans *trans, unsigned flags,
 		return -BCH_ERR_btree_insert_need_mark_replicas;
 
 	if (trans->nr_wb_updates) {
-		EBUG_ON(flags & BTREE_INSERT_JOURNAL_REPLAY);
+		EBUG_ON(flags & BCH_TRANS_COMMIT_no_journal_res);
 
 		ret = bch2_btree_insert_keys_write_buffer(trans);
 		if (ret)
@@ -663,7 +736,7 @@ bch2_trans_commit_write_locked(struct btree_trans *trans, unsigned flags,
 		trans->journal_res.u64s		-= trans->extra_journal_entries.nr;
 	}
 
-	if (likely(!(flags & BTREE_INSERT_JOURNAL_REPLAY))) {
+	if (likely(!(flags & BCH_TRANS_COMMIT_no_journal_res))) {
 		struct journal *j = &c->journal;
 		struct jset_entry *entry;
 
@@ -705,15 +778,8 @@ bch2_trans_commit_write_locked(struct btree_trans *trans, unsigned flags,
 	}
 
 	trans_for_each_update(trans, i) {
-		i->k->k.needs_whiteout = false;
-
 		if (!i->cached) {
-			u64 seq = trans->journal_res.seq;
+			bch2_btree_insert_key_leaf(trans, i->path, i->k, trans->journal_res.seq);
-
-			if (i->flags & BTREE_UPDATE_PREJOURNAL)
-				seq = i->seq;
-
-			bch2_btree_insert_key_leaf(trans, i->path, i->k, seq);
 		} else if (!i->key_cache_already_flushed)
 			bch2_btree_insert_key_cached(trans, flags, i);
 		else {
@@ -731,37 +797,6 @@ revert_fs_usage:
 	return ret;
 }
 
-static noinline int trans_lock_write_fail(struct btree_trans *trans, struct btree_insert_entry *i)
-{
-	while (--i >= trans->updates) {
-		if (same_leaf_as_prev(trans, i))
-			continue;
-
-		bch2_btree_node_unlock_write(trans, i->path, insert_l(i)->b);
-	}
-
-	trace_and_count(trans->c, trans_restart_would_deadlock_write, trans);
-	return btree_trans_restart(trans, BCH_ERR_transaction_restart_would_deadlock_write);
-}
-
-static inline int trans_lock_write(struct btree_trans *trans)
-{
-	struct btree_insert_entry *i;
-
-	trans_for_each_update(trans, i) {
-		if (same_leaf_as_prev(trans, i))
-			continue;
-
-		if (bch2_btree_node_lock_write(trans, i->path, &insert_l(i)->b->c))
-			return trans_lock_write_fail(trans, i);
-
-		if (!i->cached)
-			bch2_btree_node_prep_for_write(trans, i->path, insert_l(i)->b);
-	}
-
-	return 0;
-}
-
 static noinline void bch2_drop_overwrites_from_journal(struct btree_trans *trans)
 {
 	struct btree_insert_entry *i;
@@ -799,6 +834,12 @@ static noinline int bch2_trans_commit_bkey_invalid(struct btree_trans *trans,
 	return -EINVAL;
 }
 
+static int bch2_trans_commit_journal_pin_flush(struct journal *j,
+				struct journal_entry_pin *_pin, u64 seq)
+{
+	return 0;
+}
+
 /*
  * Get journal reservation, take write locks, and attempt to do btree update(s):
  */
@@ -829,15 +870,7 @@ static inline int do_bch2_trans_commit(struct btree_trans *trans, unsigned flags
 		}
 	}
 
-	ret = bch2_journal_preres_get(&c->journal,
+	ret = bch2_trans_lock_write(trans);
-			&trans->journal_preres, trans->journal_preres_u64s,
-			(flags & BCH_WATERMARK_MASK)|JOURNAL_RES_GET_NONBLOCK);
-	if (unlikely(ret == -BCH_ERR_journal_preres_get_blocked))
-		ret = bch2_trans_journal_preres_get_cold(trans, flags, trace_ip);
-	if (unlikely(ret))
-		return ret;
-
-	ret = trans_lock_write(trans);
 	if (unlikely(ret))
 		return ret;
 
@@ -846,20 +879,19 @@ static inline int do_bch2_trans_commit(struct btree_trans *trans, unsigned flags
 	if (!ret && unlikely(trans->journal_replay_not_finished))
 		bch2_drop_overwrites_from_journal(trans);
 
-	trans_for_each_update(trans, i)
+	bch2_trans_unlock_write(trans);
-		if (!same_leaf_as_prev(trans, i))
-			bch2_btree_node_unlock_write_inlined(trans, i->path,
-							insert_l(i)->b);
 
 	if (!ret && trans->journal_pin)
 		bch2_journal_pin_add(&c->journal, trans->journal_res.seq,
-				     trans->journal_pin, NULL);
+				     trans->journal_pin,
+				     bch2_trans_commit_journal_pin_flush);
 
 	/*
 	 * Drop journal reservation after dropping write locks, since dropping
 	 * the journal reservation may kick off a journal write:
 	 */
-	bch2_journal_res_put(&c->journal, &trans->journal_res);
+	if (likely(!(flags & BCH_TRANS_COMMIT_no_journal_res)))
+		bch2_journal_res_put(&c->journal, &trans->journal_res);
 
 	return ret;
 }
@@ -896,7 +928,7 @@ int bch2_trans_commit_error(struct btree_trans *trans, unsigned flags,
 		 * XXX: this should probably be a separate BTREE_INSERT_NONBLOCK
 		 * flag
 		 */
-		if ((flags & BTREE_INSERT_JOURNAL_RECLAIM) &&
+		if ((flags & BCH_TRANS_COMMIT_journal_reclaim) &&
 		    (flags & BCH_WATERMARK_MASK) != BCH_WATERMARK_reclaim) {
 			ret = -BCH_ERR_journal_reclaim_would_deadlock;
 			break;
@@ -931,7 +963,7 @@ int bch2_trans_commit_error(struct btree_trans *trans, unsigned flags,
 			if (wb->state.nr > wb->size * 3 / 4) {
 				bch2_trans_begin(trans);
 				ret = __bch2_btree_write_buffer_flush(trans,
-						flags|BTREE_INSERT_NOCHECK_RW, true);
+						flags|BCH_TRANS_COMMIT_no_check_rw, true);
 				if (!ret) {
 					trace_and_count(c, trans_restart_write_buffer_flush, trans, _THIS_IP_);
 					ret = btree_trans_restart(trans, BCH_ERR_transaction_restart_write_buffer_flush);
@@ -951,8 +983,7 @@ int bch2_trans_commit_error(struct btree_trans *trans, unsigned flags,
 	BUG_ON(bch2_err_matches(ret, BCH_ERR_transaction_restart) != !!trans->restarted);
 
 	bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOSPC) &&
-				!(flags & BTREE_INSERT_NOWAIT) &&
+				(flags & BCH_TRANS_COMMIT_no_enospc), c,
-				(flags & BTREE_INSERT_NOFAIL), c,
 		"%s: incorrectly got %s\n", __func__, bch2_err_str(ret));
 
 	return ret;
@@ -964,7 +995,7 @@ bch2_trans_commit_get_rw_cold(struct btree_trans *trans, unsigned flags)
 	struct bch_fs *c = trans->c;
 	int ret;
 
-	if (likely(!(flags & BTREE_INSERT_LAZY_RW)) ||
+	if (likely(!(flags & BCH_TRANS_COMMIT_lazy_rw)) ||
 	    test_bit(BCH_FS_STARTED, &c->flags))
 		return -BCH_ERR_erofs_trans_commit;
 
@@ -1002,7 +1033,6 @@ int __bch2_trans_commit(struct btree_trans *trans, unsigned flags)
 	struct bch_fs *c = trans->c;
 	struct btree_insert_entry *i = NULL;
 	struct btree_write_buffered_key *wb;
-	unsigned u64s;
 	int ret = 0;
 
 	if (!trans->nr_updates &&
@@ -1010,9 +1040,6 @@ int __bch2_trans_commit(struct btree_trans *trans, unsigned flags)
 	    !trans->extra_journal_entries.nr)
 		goto out_reset;
 
-	if (flags & BTREE_INSERT_GC_LOCK_HELD)
-		lockdep_assert_held(&c->gc_lock);
-
 	ret = bch2_trans_commit_run_triggers(trans);
 	if (ret)
 		goto out_reset;
@@ -1021,7 +1048,7 @@ int __bch2_trans_commit(struct btree_trans *trans, unsigned flags)
 		struct printbuf buf = PRINTBUF;
 		enum bkey_invalid_flags invalid_flags = 0;
 
-		if (!(flags & BTREE_INSERT_JOURNAL_REPLAY))
+		if (!(flags & BCH_TRANS_COMMIT_no_journal_res))
 			invalid_flags |= BKEY_INVALID_WRITE|BKEY_INVALID_COMMIT;
 
 		if (unlikely(bch2_bkey_invalid(c, bkey_i_to_s_c(i->k),
@@ -1039,7 +1066,7 @@ int __bch2_trans_commit(struct btree_trans *trans, unsigned flags)
 		goto out_reset;
 	}
 
-	if (!(flags & BTREE_INSERT_NOCHECK_RW) &&
+	if (!(flags & BCH_TRANS_COMMIT_no_check_rw) &&
 	    unlikely(!bch2_write_ref_tryget(c, BCH_WRITE_REF_trans))) {
 		ret = bch2_trans_commit_get_rw_cold(trans, flags);
 		if (ret)
@@ -1052,7 +1079,7 @@ int __bch2_trans_commit(struct btree_trans *trans, unsigned flags)
 		bch2_trans_unlock(trans);
 
 		ret = __bch2_btree_write_buffer_flush(trans,
-					flags|BTREE_INSERT_NOCHECK_RW, true);
+					flags|BCH_TRANS_COMMIT_no_check_rw, true);
 		if (!ret) {
 			trace_and_count(c, trans_restart_write_buffer_flush, trans, _THIS_IP_);
 			ret = btree_trans_restart(trans, BCH_ERR_transaction_restart_write_buffer_flush);
@@ -1062,13 +1089,8 @@ int __bch2_trans_commit(struct btree_trans *trans, unsigned flags)
 
 	EBUG_ON(test_bit(BCH_FS_CLEAN_SHUTDOWN, &c->flags));
 
-	memset(&trans->journal_preres, 0, sizeof(trans->journal_preres));
-
 	trans->journal_u64s		= trans->extra_journal_entries.nr;
-	trans->journal_preres_u64s	= 0;
-
 	trans->journal_transaction_names = READ_ONCE(c->opts.journal_transaction_names);
-
 	if (trans->journal_transaction_names)
 		trans->journal_u64s += jset_u64s(JSET_ENTRY_LOG_U64s);
 
@@ -1084,16 +1106,11 @@ int __bch2_trans_commit(struct btree_trans *trans, unsigned flags)
 		if (i->key_cache_already_flushed)
 			continue;
 
-		/* we're going to journal the key being updated: */
-		u64s = jset_u64s(i->k->k.u64s);
-		if (i->cached &&
-		    likely(!(flags & BTREE_INSERT_JOURNAL_REPLAY)))
-			trans->journal_preres_u64s += u64s;
-
 		if (i->flags & BTREE_UPDATE_NOJOURNAL)
 			continue;
 
-		trans->journal_u64s += u64s;
+		/* we're going to journal the key being updated: */
+		trans->journal_u64s += jset_u64s(i->k->k.u64s);
 
 		/* and we're also going to log the overwrite: */
 		if (trans->journal_transaction_names)
@@ -1106,14 +1123,15 @@ int __bch2_trans_commit(struct btree_trans *trans, unsigned flags)
 	if (trans->extra_journal_res) {
 		ret = bch2_disk_reservation_add(c, trans->disk_res,
 				trans->extra_journal_res,
-				(flags & BTREE_INSERT_NOFAIL)
+				(flags & BCH_TRANS_COMMIT_no_enospc)
 				? BCH_DISK_RESERVATION_NOFAIL : 0);
 		if (ret)
 			goto err;
 	}
 retry:
 	bch2_trans_verify_not_in_restart(trans);
-	memset(&trans->journal_res, 0, sizeof(trans->journal_res));
+	if (likely(!(flags & BCH_TRANS_COMMIT_no_journal_res)))
+		memset(&trans->journal_res, 0, sizeof(trans->journal_res));
 
 	ret = do_bch2_trans_commit(trans, flags, &i, _RET_IP_);
 
@@ -1125,9 +1143,7 @@ retry:
 
 	trace_and_count(c, transaction_commit, trans, _RET_IP_);
 out:
-	bch2_journal_preres_put(&c->journal, &trans->journal_preres);
+	if (likely(!(flags & BCH_TRANS_COMMIT_no_check_rw)))
-
-	if (likely(!(flags & BTREE_INSERT_NOCHECK_RW)))
 		bch2_write_ref_put(c, BCH_WRITE_REF_trans);
 out_reset:
 	if (!ret)
@@ -1140,5 +1156,17 @@ err:
 	if (ret)
 		goto out;
 
+	/*
+	 * We might have done another transaction commit in the error path -
+	 * i.e. btree write buffer flush - which will have made use of
+	 * trans->journal_res, but with BCH_TRANS_COMMIT_no_journal_res that is
+	 * how the journal sequence number to pin is passed in - so we must
+	 * restart:
+	 */
+	if (flags & BCH_TRANS_COMMIT_no_journal_res) {
+		ret = -BCH_ERR_transaction_restart_nested;
+		goto out;
+	}
+
 	goto retry;
 }

libbcachefs/btree_types.h

@@ -5,7 +5,7 @@
 #include <linux/list.h>
 #include <linux/rhashtable.h>
 
-//#include "bkey_methods.h"
+#include "btree_key_cache_types.h"
 #include "buckets_types.h"
 #include "darray.h"
 #include "errcode.h"
@@ -322,31 +322,6 @@ struct btree_iter {
 #endif
 };
 
-struct btree_key_cache_freelist {
-	struct bkey_cached	*objs[16];
-	unsigned		nr;
-};
-
-struct btree_key_cache {
-	struct mutex		lock;
-	struct rhashtable	table;
-	bool			table_init_done;
-	struct list_head	freed_pcpu;
-	struct list_head	freed_nonpcpu;
-	struct shrinker		shrink;
-	unsigned		shrink_iter;
-	struct btree_key_cache_freelist __percpu *pcpu_freed;
-
-	atomic_long_t		nr_freed;
-	atomic_long_t		nr_keys;
-	atomic_long_t		nr_dirty;
-};
-
-struct bkey_cached_key {
-	u32			btree_id;
-	struct bpos		pos;
-} __packed __aligned(4);
-
 #define BKEY_CACHED_ACCESSED		0
 #define BKEY_CACHED_DIRTY		1
 
@@ -362,7 +337,6 @@ struct bkey_cached {
 	struct rhash_head	hash;
 	struct list_head	list;
 
-	struct journal_preres	res;
 	struct journal_entry_pin journal;
 	u64			seq;
 
@@ -392,7 +366,6 @@ struct btree_insert_entry {
 	u8			old_btree_u64s;
 	struct bkey_i		*k;
 	struct btree_path	*path;
-	u64			seq;
 	/* key being overwritten: */
 	struct bkey		old_k;
 	const struct bch_val	*old_v;
@@ -441,6 +414,7 @@ struct btree_trans {
 	bool			journal_replay_not_finished:1;
 	bool			is_initial_gc:1;
 	bool			notrace_relock_fail:1;
+	bool			write_locked:1;
 	enum bch_errcode	restarted:16;
 	u32			restart_count;
 	unsigned long		last_begin_ip;
@@ -472,11 +446,9 @@ struct btree_trans {
 	struct journal_entry_pin *journal_pin;
 
 	struct journal_res	journal_res;
-	struct journal_preres	journal_preres;
 	u64			*journal_seq;
 	struct disk_reservation *disk_res;
 	unsigned		journal_u64s;
-	unsigned		journal_preres_u64s;
 	struct replicas_delta_list *fs_usage_deltas;
 };
 
@@ -717,6 +689,17 @@ static inline bool btree_type_has_snapshots(enum btree_id id)
 	return (1U << id) & mask;
 }
 
+static inline bool btree_type_has_snapshot_field(enum btree_id id)
+{
+	const unsigned mask = 0
+#define x(name, nr, flags, ...)	|((!!((flags) & (BTREE_ID_SNAPSHOT_FIELD|BTREE_ID_SNAPSHOTS))) << nr)
+	BCH_BTREE_IDS()
+#undef x
+	;
+
+	return (1U << id) & mask;
+}
+
 static inline bool btree_type_has_ptrs(enum btree_id id)
 {
 	const unsigned mask = 0

libbcachefs/btree_update.c

@@ -380,21 +380,12 @@ bch2_trans_update_by_path(struct btree_trans *trans, struct btree_path *path,
 {
 	struct bch_fs *c = trans->c;
 	struct btree_insert_entry *i, n;
-	u64 seq = 0;
 	int cmp;
 
 	EBUG_ON(!path->should_be_locked);
 	EBUG_ON(trans->nr_updates >= BTREE_ITER_MAX);
 	EBUG_ON(!bpos_eq(k->k.p, path->pos));
 
-	/*
-	 * The transaction journal res hasn't been allocated at this point.
-	 * That occurs at commit time. Reuse the seq field to pass in the seq
-	 * of a prejournaled key.
-	 */
-	if (flags & BTREE_UPDATE_PREJOURNAL)
-		seq = trans->journal_res.seq;
-
 	n = (struct btree_insert_entry) {
 		.flags		= flags,
 		.bkey_type	= __btree_node_type(path->level, path->btree_id),
@@ -403,7 +394,6 @@ bch2_trans_update_by_path(struct btree_trans *trans, struct btree_path *path,
 		.cached		= path->cached,
 		.path		= path,
 		.k		= k,
-		.seq		= seq,
 		.ip_allocated	= ip,
 	};
 
@@ -431,7 +421,6 @@ bch2_trans_update_by_path(struct btree_trans *trans, struct btree_path *path,
 		i->cached	= n.cached;
 		i->k		= n.k;
 		i->path		= n.path;
-		i->seq		= n.seq;
 		i->ip_allocated	= n.ip_allocated;
 	} else {
 		array_insert_item(trans->updates, trans->nr_updates,
@@ -542,18 +531,6 @@ int __must_check bch2_trans_update(struct btree_trans *trans, struct btree_iter
 	return bch2_trans_update_by_path(trans, path, k, flags, _RET_IP_);
 }
 
-/*
- * Add a transaction update for a key that has already been journaled.
- */
-int __must_check bch2_trans_update_seq(struct btree_trans *trans, u64 seq,
-				       struct btree_iter *iter, struct bkey_i *k,
-				       enum btree_update_flags flags)
-{
-	trans->journal_res.seq = seq;
-	return bch2_trans_update(trans, iter, k, flags|BTREE_UPDATE_NOJOURNAL|
-						 BTREE_UPDATE_PREJOURNAL);
-}
-
 int __must_check bch2_trans_update_buffered(struct btree_trans *trans,
 					    enum btree_id btree,
 					    struct bkey_i *k)
@@ -792,7 +769,7 @@ int bch2_btree_delete_range_trans(struct btree_trans *trans, enum btree_id id,
 
 		ret   = bch2_trans_update(trans, &iter, &delete, update_flags) ?:
 			bch2_trans_commit(trans, &disk_res, journal_seq,
-					  BTREE_INSERT_NOFAIL);
+					  BCH_TRANS_COMMIT_no_enospc);
 		bch2_disk_reservation_put(trans->c, &disk_res);
 err:
 		/*
@@ -897,7 +874,7 @@ __bch2_fs_log_msg(struct bch_fs *c, unsigned commit_flags, const char *fmt,
 		ret = __bch2_trans_log_msg(&c->journal.early_journal_entries, fmt, args);
 	} else {
 		ret = bch2_trans_do(c, NULL, NULL,
-			BTREE_INSERT_LAZY_RW|commit_flags,
+			BCH_TRANS_COMMIT_lazy_rw|commit_flags,
 			__bch2_trans_log_msg(&trans->extra_journal_entries, fmt, args));
 	}
 

libbcachefs/btree_update.h

@@ -21,37 +21,28 @@ void bch2_btree_add_journal_pin(struct bch_fs *, struct btree *, u64);
 void bch2_btree_insert_key_leaf(struct btree_trans *, struct btree_path *,
 				struct bkey_i *, u64);
 
-enum btree_insert_flags {
+#define BCH_TRANS_COMMIT_FLAGS()							\
+	x(no_enospc,	"don't check for enospc")					\
+	x(no_check_rw,	"don't attempt to take a ref on c->writes")			\
+	x(lazy_rw,	"go read-write if we haven't yet - only for use in recovery")	\
+	x(no_journal_res, "don't take a journal reservation, instead "			\
+			"pin journal entry referred to by trans->journal_res.seq")	\
+	x(journal_reclaim, "operation required for journal reclaim; may return error"	\
+			"instead of deadlocking if BCH_WATERMARK_reclaim not specified")\
+
+enum __bch_trans_commit_flags {
 	/* First bits for bch_watermark: */
-	__BTREE_INSERT_NOFAIL = BCH_WATERMARK_BITS,
+	__BCH_TRANS_COMMIT_FLAGS_START = BCH_WATERMARK_BITS,
-	__BTREE_INSERT_NOCHECK_RW,
+#define x(n, ...)	__BCH_TRANS_COMMIT_##n,
-	__BTREE_INSERT_LAZY_RW,
+	BCH_TRANS_COMMIT_FLAGS()
-	__BTREE_INSERT_JOURNAL_REPLAY,
+#undef x
-	__BTREE_INSERT_JOURNAL_RECLAIM,
-	__BTREE_INSERT_NOWAIT,
-	__BTREE_INSERT_GC_LOCK_HELD,
-	__BCH_HASH_SET_MUST_CREATE,
-	__BCH_HASH_SET_MUST_REPLACE,
 };
 
-/* Don't check for -ENOSPC: */
+enum bch_trans_commit_flags {
-#define BTREE_INSERT_NOFAIL		BIT(__BTREE_INSERT_NOFAIL)
+#define x(n, ...)	BCH_TRANS_COMMIT_##n = BIT(__BCH_TRANS_COMMIT_##n),
-
+	BCH_TRANS_COMMIT_FLAGS()
-#define BTREE_INSERT_NOCHECK_RW		BIT(__BTREE_INSERT_NOCHECK_RW)
+#undef x
-#define BTREE_INSERT_LAZY_RW		BIT(__BTREE_INSERT_LAZY_RW)
+};
-
-/* Insert is for journal replay - don't get journal reservations: */
-#define BTREE_INSERT_JOURNAL_REPLAY	BIT(__BTREE_INSERT_JOURNAL_REPLAY)
-
-/* Insert is being called from journal reclaim path: */
-#define BTREE_INSERT_JOURNAL_RECLAIM	BIT(__BTREE_INSERT_JOURNAL_RECLAIM)
-
-/* Don't block on allocation failure (for new btree nodes: */
-#define BTREE_INSERT_NOWAIT		BIT(__BTREE_INSERT_NOWAIT)
-#define BTREE_INSERT_GC_LOCK_HELD	BIT(__BTREE_INSERT_GC_LOCK_HELD)
-
-#define BCH_HASH_SET_MUST_CREATE	BIT(__BCH_HASH_SET_MUST_CREATE)
-#define BCH_HASH_SET_MUST_REPLACE	BIT(__BCH_HASH_SET_MUST_REPLACE)
 
 int bch2_btree_delete_extent_at(struct btree_trans *, struct btree_iter *,
 				unsigned, unsigned);

libbcachefs/btree_update_interior.c

@@ -475,9 +475,6 @@ static int bch2_btree_reserve_get(struct btree_trans *trans,
 	/*
 	 * Protects reaping from the btree node cache and using the btree node
 	 * open bucket reserve:
-	 *
-	 * BTREE_INSERT_NOWAIT only applies to btree node allocation, not
-	 * blocking on this lock:
 	 */
 	ret = bch2_btree_cache_cannibalize_lock(c, cl);
 	if (ret)
@@ -487,9 +484,8 @@ static int bch2_btree_reserve_get(struct btree_trans *trans,
 		struct prealloc_nodes *p = as->prealloc_nodes + interior;
 
 		while (p->nr < nr_nodes[interior]) {
-			b = __bch2_btree_node_alloc(trans, &as->disk_res,
+			b = __bch2_btree_node_alloc(trans, &as->disk_res, cl,
-					flags & BTREE_INSERT_NOWAIT ? NULL : cl,
+						    interior, flags);
-					interior, flags);
 			if (IS_ERR(b)) {
 				ret = PTR_ERR(b);
 				goto err;
@@ -513,8 +509,6 @@ static void bch2_btree_update_free(struct btree_update *as, struct btree_trans *
 		up_read(&c->gc_lock);
 	as->took_gc_lock = false;
 
-	bch2_journal_preres_put(&c->journal, &as->journal_preres);
-
 	bch2_journal_pin_drop(&c->journal, &as->journal);
 	bch2_journal_pin_flush(&c->journal, &as->journal);
 	bch2_disk_reservation_put(c, &as->disk_res);
@@ -646,9 +640,9 @@ static void btree_update_nodes_written(struct btree_update *as)
 	 */
 	ret = commit_do(trans, &as->disk_res, &journal_seq,
 			BCH_WATERMARK_reclaim|
-			BTREE_INSERT_NOFAIL|
+			BCH_TRANS_COMMIT_no_enospc|
-			BTREE_INSERT_NOCHECK_RW|
+			BCH_TRANS_COMMIT_no_check_rw|
-			BTREE_INSERT_JOURNAL_RECLAIM,
+			BCH_TRANS_COMMIT_journal_reclaim,
 			btree_update_nodes_written_trans(trans, as));
 	bch2_trans_unlock(trans);
 
@@ -734,8 +728,6 @@ err:
 
 	bch2_journal_pin_drop(&c->journal, &as->journal);
 
-	bch2_journal_preres_put(&c->journal, &as->journal_preres);
-
 	mutex_lock(&c->btree_interior_update_lock);
 	for (i = 0; i < as->nr_new_nodes; i++) {
 		b = as->new_nodes[i];
@@ -818,6 +810,12 @@ static void btree_update_updated_node(struct btree_update *as, struct btree *b)
 	mutex_unlock(&c->btree_interior_update_lock);
 }
 
+static int bch2_update_reparent_journal_pin_flush(struct journal *j,
+				struct journal_entry_pin *_pin, u64 seq)
+{
+	return 0;
+}
+
 static void btree_update_reparent(struct btree_update *as,
 				  struct btree_update *child)
 {
@@ -828,7 +826,8 @@ static void btree_update_reparent(struct btree_update *as,
 	child->b = NULL;
 	child->mode = BTREE_INTERIOR_UPDATING_AS;
 
-	bch2_journal_pin_copy(&c->journal, &as->journal, &child->journal, NULL);
+	bch2_journal_pin_copy(&c->journal, &as->journal, &child->journal,
+			      bch2_update_reparent_journal_pin_flush);
 }
 
 static void btree_update_updated_root(struct btree_update *as, struct btree *b)
@@ -937,6 +936,12 @@ static void bch2_btree_update_get_open_buckets(struct btree_update *as, struct b
 			b->ob.v[--b->ob.nr];
 }
 
+static int bch2_btree_update_will_free_node_journal_pin_flush(struct journal *j,
+				struct journal_entry_pin *_pin, u64 seq)
+{
+	return 0;
+}
+
 /*
  * @b is being split/rewritten: it may have pointers to not-yet-written btree
  * nodes and thus outstanding btree_updates - redirect @b's
@@ -988,11 +993,13 @@ static void bch2_btree_interior_update_will_free_node(struct btree_update *as,
 	 * when the new nodes are persistent and reachable on disk:
 	 */
 	w = btree_current_write(b);
-	bch2_journal_pin_copy(&c->journal, &as->journal, &w->journal, NULL);
+	bch2_journal_pin_copy(&c->journal, &as->journal, &w->journal,
+			      bch2_btree_update_will_free_node_journal_pin_flush);
 	bch2_journal_pin_drop(&c->journal, &w->journal);
 
 	w = btree_prev_write(b);
-	bch2_journal_pin_copy(&c->journal, &as->journal, &w->journal, NULL);
+	bch2_journal_pin_copy(&c->journal, &as->journal, &w->journal,
+			      bch2_btree_update_will_free_node_journal_pin_flush);
 	bch2_journal_pin_drop(&c->journal, &w->journal);
 
 	mutex_unlock(&c->btree_interior_update_lock);
@@ -1042,7 +1049,7 @@ bch2_btree_update_start(struct btree_trans *trans, struct btree_path *path,
 	struct bch_fs *c = trans->c;
 	struct btree_update *as;
 	u64 start_time = local_clock();
-	int disk_res_flags = (flags & BTREE_INSERT_NOFAIL)
+	int disk_res_flags = (flags & BCH_TRANS_COMMIT_no_enospc)
 		? BCH_DISK_RESERVATION_NOFAIL : 0;
 	unsigned nr_nodes[2] = { 0, 0 };
 	unsigned update_level = level;
@@ -1061,7 +1068,7 @@ bch2_btree_update_start(struct btree_trans *trans, struct btree_path *path,
 	flags &= ~BCH_WATERMARK_MASK;
 	flags |= watermark;
 
-	if (flags & BTREE_INSERT_JOURNAL_RECLAIM)
+	if (flags & BCH_TRANS_COMMIT_journal_reclaim)
 		journal_flags |= JOURNAL_RES_GET_NONBLOCK;
 	journal_flags |= watermark;
 
@@ -1087,9 +1094,7 @@ bch2_btree_update_start(struct btree_trans *trans, struct btree_path *path,
 		split = path->l[update_level].b->nr.live_u64s > BTREE_SPLIT_THRESHOLD(c);
 	}
 
-	if (flags & BTREE_INSERT_GC_LOCK_HELD)
+	if (!down_read_trylock(&c->gc_lock)) {
-		lockdep_assert_held(&c->gc_lock);
-	else if (!down_read_trylock(&c->gc_lock)) {
 		ret = drop_locks_do(trans, (down_read(&c->gc_lock), 0));
 		if (ret) {
 			up_read(&c->gc_lock);
@@ -1103,7 +1108,7 @@ bch2_btree_update_start(struct btree_trans *trans, struct btree_path *path,
 	as->c		= c;
 	as->start_time	= start_time;
 	as->mode	= BTREE_INTERIOR_NO_UPDATE;
-	as->took_gc_lock = !(flags & BTREE_INSERT_GC_LOCK_HELD);
+	as->took_gc_lock = true;
 	as->btree_id	= path->btree_id;
 	as->update_level = update_level;
 	INIT_LIST_HEAD(&as->list);
@@ -1129,27 +1134,6 @@ bch2_btree_update_start(struct btree_trans *trans, struct btree_path *path,
 	if (ret)
 		goto err;
 
-	ret = bch2_journal_preres_get(&c->journal, &as->journal_preres,
-				      BTREE_UPDATE_JOURNAL_RES,
-				      journal_flags|JOURNAL_RES_GET_NONBLOCK);
-	if (ret) {
-		if (flags & BTREE_INSERT_JOURNAL_RECLAIM) {
-			ret = -BCH_ERR_journal_reclaim_would_deadlock;
-			goto err;
-		}
-
-		ret = drop_locks_do(trans,
-			bch2_journal_preres_get(&c->journal, &as->journal_preres,
-					      BTREE_UPDATE_JOURNAL_RES,
-					      journal_flags));
-		if (ret == -BCH_ERR_journal_preres_get_blocked) {
-			trace_and_count(c, trans_restart_journal_preres_get, trans, _RET_IP_, journal_flags);
-			ret = btree_trans_restart(trans, BCH_ERR_transaction_restart_journal_preres_get);
-		}
-		if (ret)
-			goto err;
-	}
-
 	ret = bch2_disk_reservation_get(c, &as->disk_res,
 			(nr_nodes[0] + nr_nodes[1]) * btree_sectors(c),
 			c->opts.metadata_replicas,
@@ -1167,7 +1151,7 @@ bch2_btree_update_start(struct btree_trans *trans, struct btree_path *path,
 		 * flag
 		 */
 		if (bch2_err_matches(ret, ENOSPC) &&
-		    (flags & BTREE_INSERT_JOURNAL_RECLAIM) &&
+		    (flags & BCH_TRANS_COMMIT_journal_reclaim) &&
 		    watermark != BCH_WATERMARK_reclaim) {
 			ret = -BCH_ERR_journal_reclaim_would_deadlock;
 			goto err;
@@ -1855,7 +1839,7 @@ int __bch2_foreground_maybe_merge(struct btree_trans *trans,
 
 	parent = btree_node_parent(path, b);
 	as = bch2_btree_update_start(trans, path, level, false,
-				     BTREE_INSERT_NOFAIL|flags);
+				     BCH_TRANS_COMMIT_no_enospc|flags);
 	ret = PTR_ERR_OR_ZERO(as);
 	if (ret)
 		goto err;
@@ -1941,7 +1925,7 @@ int bch2_btree_node_rewrite(struct btree_trans *trans,
 	struct btree_update *as;
 	int ret;
 
-	flags |= BTREE_INSERT_NOFAIL;
+	flags |= BCH_TRANS_COMMIT_no_enospc;
 
 	parent = btree_node_parent(iter->path, b);
 	as = bch2_btree_update_start(trans, iter->path, b->c.level,
@@ -2418,23 +2402,17 @@ void bch2_journal_entry_to_btree_root(struct bch_fs *c, struct jset_entry *entry
 
 struct jset_entry *
 bch2_btree_roots_to_journal_entries(struct bch_fs *c,
-				    struct jset_entry *start,
+				    struct jset_entry *end,
-				    struct jset_entry *end)
+				    unsigned long skip)
 {
-	struct jset_entry *entry;
-	unsigned long have = 0;
 	unsigned i;
 
-	for (entry = start; entry < end; entry = vstruct_next(entry))
-		if (entry->type == BCH_JSET_ENTRY_btree_root)
-			__set_bit(entry->btree_id, &have);
-
 	mutex_lock(&c->btree_root_lock);
 
 	for (i = 0; i < btree_id_nr_alive(c); i++) {
 		struct btree_root *r = bch2_btree_id_root(c, i);
 
-		if (r->alive && !test_bit(i, &have)) {
+		if (r->alive && !test_bit(i, &skip)) {
 			journal_entry_set(end, BCH_JSET_ENTRY_btree_root,
 					  i, r->level, &r->key, r->key.k.u64s);
 			end = vstruct_next(end);

libbcachefs/btree_update_interior.h

@@ -55,7 +55,6 @@ struct btree_update {
 	unsigned			update_level;
 
 	struct disk_reservation		disk_res;
-	struct journal_preres		journal_preres;
 
 	/*
 	 * BTREE_INTERIOR_UPDATING_NODE:
@@ -325,7 +324,7 @@ bool bch2_btree_interior_updates_flush(struct bch_fs *);
 
 void bch2_journal_entry_to_btree_root(struct bch_fs *, struct jset_entry *);
 struct jset_entry *bch2_btree_roots_to_journal_entries(struct bch_fs *,
-					struct jset_entry *, struct jset_entry *);
+					struct jset_entry *, unsigned long);
 
 void bch2_do_pending_node_rewrites(struct bch_fs *);
 void bch2_free_pending_node_rewrites(struct bch_fs *);

libbcachefs/btree_write_buffer.c

@@ -9,9 +9,11 @@
 #include "journal.h"
 #include "journal_reclaim.h"
 
-#include <linux/atomic.h>
 #include <linux/sort.h>
 
+static int bch2_btree_write_buffer_journal_flush(struct journal *,
+				struct journal_entry_pin *, u64);
+
 static int btree_write_buffered_key_cmp(const void *_l, const void *_r)
 {
 	const struct btree_write_buffered_key *l = _l;
@@ -46,6 +48,13 @@ static int bch2_btree_write_buffer_flush_one(struct btree_trans *trans,
 	if (ret)
 		return ret;
 
+	/*
+	 * We can't clone a path that has write locks: unshare it now, before
+	 * set_pos and traverse():
+	 */
+	if (iter->path->ref > 1)
+		iter->path = __bch2_btree_path_make_mut(trans, iter->path, true, _THIS_IP_);
+
 	path = iter->path;
 
 	if (!*write_locked) {
@@ -65,24 +74,18 @@ static int bch2_btree_write_buffer_flush_one(struct btree_trans *trans,
 
 	bch2_btree_insert_key_leaf(trans, path, &wb->k, wb->journal_seq);
 	(*fast)++;
-
-	if (path->ref > 1) {
-		/*
-		 * We can't clone a path that has write locks: if the path is
-		 * shared, unlock before set_pos(), traverse():
-		 */
-		bch2_btree_node_unlock_write(trans, path, path->l[0].b);
-		*write_locked = false;
-	}
 	return 0;
 trans_commit:
-	return  bch2_trans_update_seq(trans, wb->journal_seq, iter, &wb->k,
+	trans->journal_res.seq = wb->journal_seq;
-				      BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE) ?:
+
+	return  bch2_trans_update(trans, iter, &wb->k,
+				  BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE) ?:
 		bch2_trans_commit(trans, NULL, NULL,
 				  commit_flags|
-				  BTREE_INSERT_NOCHECK_RW|
+				  BCH_TRANS_COMMIT_no_check_rw|
-				  BTREE_INSERT_NOFAIL|
+				  BCH_TRANS_COMMIT_no_enospc|
-				  BTREE_INSERT_JOURNAL_RECLAIM);
+				  BCH_TRANS_COMMIT_no_journal_res|
+				  BCH_TRANS_COMMIT_journal_reclaim);
 }
 
 static union btree_write_buffer_state btree_write_buffer_switch(struct btree_write_buffer *wb)
@@ -125,9 +128,11 @@ btree_write_buffered_insert(struct btree_trans *trans,
 	bch2_trans_iter_init(trans, &iter, wb->btree, bkey_start_pos(&wb->k.k),
 			     BTREE_ITER_CACHED|BTREE_ITER_INTENT);
 
+	trans->journal_res.seq = wb->journal_seq;
+
 	ret   = bch2_btree_iter_traverse(&iter) ?:
-		bch2_trans_update_seq(trans, wb->journal_seq, &iter, &wb->k,
+		bch2_trans_update(trans, &iter, &wb->k,
-				      BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE);
+				  BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE);
 	bch2_trans_iter_exit(trans, &iter);
 	return ret;
 }
@@ -151,7 +156,8 @@ int __bch2_btree_write_buffer_flush(struct btree_trans *trans, unsigned commit_f
 	if (!locked && !mutex_trylock(&wb->flush_lock))
 		return 0;
 
-	bch2_journal_pin_copy(j, &pin, &wb->journal_pin, NULL);
+	bch2_journal_pin_copy(j, &pin, &wb->journal_pin,
+			      bch2_btree_write_buffer_journal_flush);
 	bch2_journal_pin_drop(j, &wb->journal_pin);
 
 	s = btree_write_buffer_switch(wb);
@@ -169,7 +175,7 @@ int __bch2_btree_write_buffer_flush(struct btree_trans *trans, unsigned commit_f
 	 * However, since we're not flushing in the order they appear in the
 	 * journal we won't be able to drop our journal pin until everything is
 	 * flushed - which means this could deadlock the journal if we weren't
-	 * passing BTREE_INSERT_JOURNAL_RECLAIM. This causes the update to fail
+	 * passing BCH_TRANS_COMMIT_journal_reclaim. This causes the update to fail
 	 * if it would block taking a journal reservation.
 	 *
 	 * If that happens, simply skip the key so we can optimistically insert
@@ -253,21 +259,14 @@ slowpath:
 		if (!i->journal_seq)
 			continue;
 
-		if (i->journal_seq > pin.seq) {
+		bch2_journal_pin_update(j, i->journal_seq, &pin,
-			struct journal_entry_pin pin2;
+			      bch2_btree_write_buffer_journal_flush);
-
-			memset(&pin2, 0, sizeof(pin2));
-
-			bch2_journal_pin_add(j, i->journal_seq, &pin2, NULL);
-			bch2_journal_pin_drop(j, &pin);
-			bch2_journal_pin_copy(j, &pin, &pin2, NULL);
-			bch2_journal_pin_drop(j, &pin2);
-		}
 
 		ret = commit_do(trans, NULL, NULL,
 				commit_flags|
-				BTREE_INSERT_NOFAIL|
+				BCH_TRANS_COMMIT_no_enospc|
-				BTREE_INSERT_JOURNAL_RECLAIM,
+				BCH_TRANS_COMMIT_no_journal_res|
+				BCH_TRANS_COMMIT_journal_reclaim,
 				btree_write_buffered_insert(trans, i));
 		if (bch2_fs_fatal_err_on(ret, c, "%s: insert error %s", __func__, bch2_err_str(ret)))
 			break;
@@ -297,7 +296,7 @@ static int bch2_btree_write_buffer_journal_flush(struct journal *j,
 	mutex_lock(&wb->flush_lock);
 
 	return bch2_trans_run(c,
-			__bch2_btree_write_buffer_flush(trans, BTREE_INSERT_NOCHECK_RW, true));
+			__bch2_btree_write_buffer_flush(trans, BCH_TRANS_COMMIT_no_check_rw, true));
 }
 
 static inline u64 btree_write_buffer_ref(int idx)

libbcachefs/darray.c

@@ -0,0 +1,21 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/log2.h>
+#include <linux/slab.h>
+#include "darray.h"
+
+int __bch2_darray_resize(darray_void *d, size_t element_size, size_t new_size, gfp_t gfp)
+{
+	if (new_size > d->size) {
+		new_size = roundup_pow_of_two(new_size);
+
+		void *data = krealloc_array(d->data, new_size, element_size, gfp);
+		if (!data)
+			return -ENOMEM;
+
+		d->data	= data;
+		d->size = new_size;
+	}
+
+	return 0;
+}

libbcachefs/darray.h

@@ -8,7 +8,6 @@
  * Inspired by CCAN's darray
  */
 
-#include "util.h"
 #include <linux/slab.h>
 
 #define DARRAY(type)							\
@@ -19,20 +18,25 @@ struct {								\
 
 typedef DARRAY(void) darray_void;
 
+int __bch2_darray_resize(darray_void *, size_t, size_t, gfp_t);
+
+static inline int __darray_resize(darray_void *d, size_t element_size,
+				  size_t new_size, gfp_t gfp)
+{
+	return unlikely(new_size > d->size)
+		? __bch2_darray_resize(d, element_size, new_size, gfp)
+		: 0;
+}
+
+#define darray_resize_gfp(_d, _new_size, _gfp)				\
+	__darray_resize((darray_void *) (_d), sizeof((_d)->data[0]), (_new_size), _gfp)
+
+#define darray_resize(_d, _new_size)					\
+	darray_resize_gfp(_d, _new_size, GFP_KERNEL)
+
 static inline int __darray_make_room(darray_void *d, size_t t_size, size_t more, gfp_t gfp)
 {
-	if (d->nr + more > d->size) {
+	return __darray_resize(d, t_size, d->nr + more, gfp);
-		size_t new_size = roundup_pow_of_two(d->nr + more);
-		void *data = krealloc_array(d->data, new_size, t_size, gfp);
-
-		if (!data)
-			return -ENOMEM;
-
-		d->data	= data;
-		d->size = new_size;
-	}
-
-	return 0;
 }
 
 #define darray_make_room_gfp(_d, _more, _gfp)				\
@@ -41,6 +45,8 @@ static inline int __darray_make_room(darray_void *d, size_t t_size, size_t more,
 #define darray_make_room(_d, _more)					\
 	darray_make_room_gfp(_d, _more, GFP_KERNEL)
 
+#define darray_room(_d)		((_d).size - (_d).nr)
+
 #define darray_top(_d)		((_d).data[(_d).nr])
 
 #define darray_push_gfp(_d, _item, _gfp)				\

libbcachefs/data_update.c

@@ -239,6 +239,34 @@ restart_drop_extra_replicas:
 
 		next_pos = insert->k.p;
 
+		/*
+		 * Check for nonce offset inconsistency:
+		 * This is debug code - we've been seeing this bug rarely, and
+		 * it's been hard to reproduce, so this should give us some more
+		 * information when it does occur:
+		 */
+		struct printbuf err = PRINTBUF;
+		int invalid = bch2_bkey_invalid(c, bkey_i_to_s_c(insert), __btree_node_type(0, m->btree_id), 0, &err);
+		printbuf_exit(&err);
+
+		if (invalid) {
+			struct printbuf buf = PRINTBUF;
+
+			prt_str(&buf, "about to insert invalid key in data update path");
+			prt_str(&buf, "\nold: ");
+			bch2_bkey_val_to_text(&buf, c, old);
+			prt_str(&buf, "\nk:   ");
+			bch2_bkey_val_to_text(&buf, c, k);
+			prt_str(&buf, "\nnew: ");
+			bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(insert));
+
+			bch2_print_string_as_lines(KERN_ERR, buf.buf);
+			printbuf_exit(&buf);
+
+			bch2_fatal_error(c);
+			goto out;
+		}
+
 		ret =   bch2_insert_snapshot_whiteouts(trans, m->btree_id,
 						k.k->p, bkey_start_pos(&insert->k)) ?:
 			bch2_insert_snapshot_whiteouts(trans, m->btree_id,
@@ -250,8 +278,8 @@ restart_drop_extra_replicas:
 				BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE) ?:
 			bch2_trans_commit(trans, &op->res,
 				NULL,
-				BTREE_INSERT_NOCHECK_RW|
+				BCH_TRANS_COMMIT_no_check_rw|
-				BTREE_INSERT_NOFAIL|
+				BCH_TRANS_COMMIT_no_enospc|
 				m->data_opts.btree_insert_flags);
 		if (!ret) {
 			bch2_btree_iter_set_pos(&iter, next_pos);

libbcachefs/dirent.c

@@ -201,7 +201,8 @@ static struct bkey_i_dirent *dirent_create_key(struct btree_trans *trans,
 int bch2_dirent_create(struct btree_trans *trans, subvol_inum dir,
 		       const struct bch_hash_info *hash_info,
 		       u8 type, const struct qstr *name, u64 dst_inum,
-		       u64 *dir_offset, int flags)
+		       u64 *dir_offset,
+		       bch_str_hash_flags_t str_hash_flags)
 {
 	struct bkey_i_dirent *dirent;
 	int ret;
@@ -212,7 +213,7 @@ int bch2_dirent_create(struct btree_trans *trans, subvol_inum dir,
 		return ret;
 
 	ret = bch2_hash_set(trans, bch2_dirent_hash_desc, hash_info,
-			    dir, &dirent->k_i, flags);
+			    dir, &dirent->k_i, str_hash_flags);
 	*dir_offset = dirent->k.p.offset;
 
 	return ret;

libbcachefs/dirent.h

@@ -37,7 +37,8 @@ int bch2_dirent_read_target(struct btree_trans *, subvol_inum,
 
 int bch2_dirent_create(struct btree_trans *, subvol_inum,
 		       const struct bch_hash_info *, u8,
-		       const struct qstr *, u64, u64 *, int);
+		       const struct qstr *, u64, u64 *,
+		       bch_str_hash_flags_t);
 
 static inline unsigned vfs_d_type(unsigned type)
 {

libbcachefs/disk_groups.c

@@ -555,6 +555,7 @@ void bch2_target_to_text(struct printbuf *out, struct bch_fs *c, unsigned v)
 	case TARGET_DEV: {
 		struct bch_dev *ca;
 
+		out->atomic++;
 		rcu_read_lock();
 		ca = t.dev < c->sb.nr_devices
 			? rcu_dereference(c->devs[t.dev])
@@ -570,6 +571,7 @@ void bch2_target_to_text(struct printbuf *out, struct bch_fs *c, unsigned v)
 		}
 
 		rcu_read_unlock();
+		out->atomic--;
 		break;
 	}
 	case TARGET_GROUP:
@@ -580,7 +582,7 @@ void bch2_target_to_text(struct printbuf *out, struct bch_fs *c, unsigned v)
 	}
 }
 
-void bch2_target_to_text_sb(struct printbuf *out, struct bch_sb *sb, unsigned v)
+static void bch2_target_to_text_sb(struct printbuf *out, struct bch_sb *sb, unsigned v)
 {
 	struct target t = target_decode(v);
 

libbcachefs/ec.c

@@ -150,6 +150,7 @@ void bch2_stripe_to_text(struct printbuf *out, struct bch_fs *c,
 		prt_printf(out, " %u:%llu:%u", ptr->dev, b, offset);
 		if (i < nr_data)
 			prt_printf(out, "#%u", stripe_blockcount_get(s, i));
+		prt_printf(out, " gen %u", ptr->gen);
 		if (ptr_stale(ca, ptr))
 			prt_printf(out, " stale");
 	}
@@ -303,16 +304,21 @@ static void ec_validate_checksums(struct bch_fs *c, struct ec_stripe_buf *buf)
 			struct bch_csum got = ec_block_checksum(buf, i, offset);
 
 			if (bch2_crc_cmp(want, got)) {
-				struct printbuf buf2 = PRINTBUF;
+				struct printbuf err = PRINTBUF;
+				struct bch_dev *ca = bch_dev_bkey_exists(c, v->ptrs[i].dev);
 
-				bch2_bkey_val_to_text(&buf2, c, bkey_i_to_s_c(&buf->key));
+				prt_printf(&err, "stripe checksum error: expected %0llx:%0llx got %0llx:%0llx (type %s)\n",
+					   want.hi, want.lo,
+					   got.hi, got.lo,
+					   bch2_csum_types[v->csum_type]);
+				prt_printf(&err, "  for %ps at %u of\n  ", (void *) _RET_IP_, i);
+				bch2_bkey_val_to_text(&err, c, bkey_i_to_s_c(&buf->key));
+				bch_err_ratelimited(ca, "%s", err.buf);
+				printbuf_exit(&err);
 
-				bch_err_ratelimited(c,
-					"stripe checksum error for %ps at %u:%u: csum type %u, expected %llx got %llx\n%s",
-					(void *) _RET_IP_, i, j, v->csum_type,
-					want.lo, got.lo, buf2.buf);
-				printbuf_exit(&buf2);
 				clear_bit(i, buf->valid);
+
+				bch2_io_error(ca, BCH_MEMBER_ERROR_checksum);
 				break;
 			}
 
@@ -475,14 +481,10 @@ err:
 	return ret;
 }
 
-static int get_stripe_key(struct bch_fs *c, u64 idx, struct ec_stripe_buf *stripe)
-{
-	return bch2_trans_run(c, get_stripe_key_trans(trans, idx, stripe));
-}
-
 /* recovery read path: */
-int bch2_ec_read_extent(struct bch_fs *c, struct bch_read_bio *rbio)
+int bch2_ec_read_extent(struct btree_trans *trans, struct bch_read_bio *rbio)
 {
+	struct bch_fs *c = trans->c;
 	struct ec_stripe_buf *buf;
 	struct closure cl;
 	struct bch_stripe *v;
@@ -497,7 +499,7 @@ int bch2_ec_read_extent(struct bch_fs *c, struct bch_read_bio *rbio)
 	if (!buf)
 		return -BCH_ERR_ENOMEM_ec_read_extent;
 
-	ret = get_stripe_key(c, rbio->pick.ec.idx, buf);
+	ret = lockrestart_do(trans, get_stripe_key_trans(trans, rbio->pick.ec.idx, buf));
 	if (ret) {
 		bch_err_ratelimited(c,
 			"error doing reconstruct read: error %i looking up stripe", ret);
@@ -801,7 +803,7 @@ static void ec_stripe_delete_work(struct work_struct *work)
 		if (!idx)
 			break;
 
-		ret = commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+		ret = commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 				ec_stripe_delete(trans, idx));
 		if (ret) {
 			bch_err_fn(c, ret);
@@ -981,8 +983,8 @@ static int ec_stripe_update_bucket(struct btree_trans *trans, struct ec_stripe_b
 
 	while (1) {
 		ret = commit_do(trans, NULL, NULL,
-				BTREE_INSERT_NOCHECK_RW|
+				BCH_TRANS_COMMIT_no_check_rw|
-				BTREE_INSERT_NOFAIL,
+				BCH_TRANS_COMMIT_no_enospc,
 			ec_stripe_update_extent(trans, bucket_pos, bucket.gen,
 						s, &bp_pos));
 		if (ret)
@@ -1119,8 +1121,8 @@ static void ec_stripe_create(struct ec_stripe_new *s)
 	}
 
 	ret = bch2_trans_do(c, &s->res, NULL,
-			    BTREE_INSERT_NOCHECK_RW|
+			    BCH_TRANS_COMMIT_no_check_rw|
-			    BTREE_INSERT_NOFAIL,
+			    BCH_TRANS_COMMIT_no_enospc,
 			    ec_stripe_key_update(trans,
 					bkey_i_to_stripe(&s->new_stripe.key),
 					!s->have_existing_stripe));
@@ -1371,6 +1373,15 @@ ec_new_stripe_head_alloc(struct bch_fs *c, unsigned target,
 			h->nr_active_devs++;
 
 	rcu_read_unlock();
+
+	/*
+	 * If we only have redundancy + 1 devices, we're better off with just
+	 * replication:
+	 */
+	if (h->nr_active_devs < h->redundancy + 2)
+		bch_err(c, "insufficient devices available to create stripe (have %u, need %u) - mismatched bucket sizes?",
+			h->nr_active_devs, h->redundancy + 2);
+
 	list_add(&h->list, &c->ec_stripe_head_list);
 	return h;
 }
@@ -1422,6 +1433,11 @@ __bch2_ec_stripe_head_get(struct btree_trans *trans,
 
 	h = ec_new_stripe_head_alloc(c, target, algo, redundancy, watermark);
 found:
+	if (!IS_ERR_OR_NULL(h) &&
+	    h->nr_active_devs < h->redundancy + 2) {
+		mutex_unlock(&h->lock);
+		h = NULL;
+	}
 	mutex_unlock(&c->ec_stripe_head_lock);
 	return h;
 }
@@ -1679,8 +1695,6 @@ struct ec_stripe_head *bch2_ec_stripe_head_get(struct btree_trans *trans,
 	int ret;
 
 	h = __bch2_ec_stripe_head_get(trans, target, algo, redundancy, watermark);
-	if (!h)
-		bch_err(c, "no stripe head");
 	if (IS_ERR_OR_NULL(h))
 		return h;
 

libbcachefs/ec.h

@@ -199,7 +199,7 @@ struct ec_stripe_head {
 	struct ec_stripe_new	*s;
 };
 
-int bch2_ec_read_extent(struct bch_fs *, struct bch_read_bio *);
+int bch2_ec_read_extent(struct btree_trans *, struct bch_read_bio *);
 
 void *bch2_writepoint_ec_buf(struct bch_fs *, struct write_point *);
 

libbcachefs/errcode.h

@@ -73,7 +73,6 @@
 	x(ENOMEM,			ENOMEM_fsck_add_nlink)			\
 	x(ENOMEM,			ENOMEM_journal_key_insert)		\
 	x(ENOMEM,			ENOMEM_journal_keys_sort)		\
-	x(ENOMEM,			ENOMEM_journal_replay)			\
 	x(ENOMEM,			ENOMEM_read_superblock_clean)		\
 	x(ENOMEM,			ENOMEM_fs_alloc)			\
 	x(ENOMEM,			ENOMEM_fs_name_alloc)			\

libbcachefs/fs-io-pagecache.c

@@ -13,7 +13,7 @@
 
 int bch2_filemap_get_contig_folios_d(struct address_space *mapping,
 				     loff_t start, u64 end,
-				     int fgp_flags, gfp_t gfp,
+				     fgf_t fgp_flags, gfp_t gfp,
 				     folios *fs)
 {
 	struct folio *f;

libbcachefs/fs-io-pagecache.h

@@ -7,7 +7,7 @@
 typedef DARRAY(struct folio *) folios;
 
 int bch2_filemap_get_contig_folios_d(struct address_space *, loff_t,
-				     u64, int, gfp_t, folios *);
+				     u64, fgf_t, gfp_t, folios *);
 int bch2_write_invalidate_inode_pages_range(struct address_space *, loff_t, loff_t);
 
 /*

libbcachefs/fs.c

@@ -93,7 +93,7 @@ retry:
 				BTREE_ITER_INTENT) ?:
 		(set ? set(trans, inode, &inode_u, p) : 0) ?:
 		bch2_inode_write(trans, &iter, &inode_u) ?:
-		bch2_trans_commit(trans, NULL, NULL, BTREE_INSERT_NOFAIL);
+		bch2_trans_commit(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc);
 
 	/*
 	 * the btree node lock protects inode->ei_inode, not ei_update_lock;
@@ -452,7 +452,7 @@ int __bch2_unlink(struct inode *vdir, struct dentry *dentry,
 	bch2_lock_inodes(INODE_UPDATE_LOCK, dir, inode);
 
 	ret = commit_do(trans, NULL, NULL,
-			BTREE_INSERT_NOFAIL,
+			BCH_TRANS_COMMIT_no_enospc,
 		bch2_unlink_trans(trans,
 				  inode_inum(dir), &dir_u,
 				  &inode_u, &dentry->d_name,
@@ -717,7 +717,7 @@ retry:
 
 	ret =   bch2_inode_write(trans, &inode_iter, &inode_u) ?:
 		bch2_trans_commit(trans, NULL, NULL,
-				  BTREE_INSERT_NOFAIL);
+				  BCH_TRANS_COMMIT_no_enospc);
 btree_err:
 	bch2_trans_iter_exit(trans, &inode_iter);
 
@@ -1922,10 +1922,7 @@ out:
 	return dget(sb->s_root);
 
 err_put_super:
-	sb->s_fs_info = NULL;
-	c->vfs_sb = NULL;
 	deactivate_locked_super(sb);
-	bch2_fs_stop(c);
 	return ERR_PTR(bch2_err_class(ret));
 }
 
@@ -1933,11 +1930,8 @@ static void bch2_kill_sb(struct super_block *sb)
 {
 	struct bch_fs *c = sb->s_fs_info;
 
-	if (c)
-		c->vfs_sb = NULL;
 	generic_shutdown_super(sb);
-	if (c)
+	bch2_fs_free(c);
-		bch2_fs_free(c);
 }
 
 static struct file_system_type bcache_fs_type = {

libbcachefs/fsck.c

@@ -208,8 +208,8 @@ static int fsck_write_inode(struct btree_trans *trans,
 			    u32 snapshot)
 {
 	int ret = commit_do(trans, NULL, NULL,
-				  BTREE_INSERT_NOFAIL|
+				  BCH_TRANS_COMMIT_no_enospc|
-				  BTREE_INSERT_LAZY_RW,
+				  BCH_TRANS_COMMIT_lazy_rw,
 				  __write_inode(trans, inode, snapshot));
 	if (ret)
 		bch_err_fn(trans->c, ret);
@@ -354,8 +354,8 @@ static int reattach_inode(struct btree_trans *trans,
 			  u32 inode_snapshot)
 {
 	int ret = commit_do(trans, NULL, NULL,
-				  BTREE_INSERT_LAZY_RW|
+				  BCH_TRANS_COMMIT_lazy_rw|
-				  BTREE_INSERT_NOFAIL,
+				  BCH_TRANS_COMMIT_no_enospc,
 			__reattach_inode(trans, inode, inode_snapshot));
 	bch_err_msg(trans->c, ret, "reattaching inode %llu", inode->bi_inum);
 	return ret;
@@ -757,8 +757,8 @@ static int hash_redo_key(struct btree_trans *trans,
 				       BCH_HASH_SET_MUST_CREATE,
 				       BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE) ?:
 		bch2_trans_commit(trans, NULL, NULL,
-				  BTREE_INSERT_NOFAIL|
+				  BCH_TRANS_COMMIT_no_enospc|
-				  BTREE_INSERT_LAZY_RW);
+				  BCH_TRANS_COMMIT_lazy_rw);
 }
 
 static int hash_check_key(struct btree_trans *trans,
@@ -992,7 +992,7 @@ int bch2_check_inodes(struct bch_fs *c)
 	ret = for_each_btree_key_commit(trans, iter, BTREE_ID_inodes,
 			POS_MIN,
 			BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS, k,
-			NULL, NULL, BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
+			NULL, NULL, BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc,
 		check_inode(trans, &iter, k, &prev, &s, full));
 
 	snapshots_seen_exit(&s);
@@ -1226,7 +1226,7 @@ static int overlapping_extents_found(struct btree_trans *trans,
 				BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE,
 				k1, k2) ?:
 			bch2_trans_commit(trans, &res, NULL,
-				BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL);
+				BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc);
 		bch2_disk_reservation_put(c, &res);
 
 		if (ret)
@@ -1465,7 +1465,7 @@ int bch2_check_extents(struct bch_fs *c)
 			POS(BCACHEFS_ROOT_INO, 0),
 			BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS, k,
 			&res, NULL,
-			BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL, ({
+			BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc, ({
 		bch2_disk_reservation_put(c, &res);
 		check_extent(trans, &iter, k, &w, &s, &extent_ends) ?:
 		check_extent_overbig(trans, &iter, k);
@@ -1494,7 +1494,7 @@ int bch2_check_indirect_extents(struct bch_fs *c)
 			POS_MIN,
 			BTREE_ITER_PREFETCH, k,
 			&res, NULL,
-			BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL, ({
+			BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc, ({
 		bch2_disk_reservation_put(c, &res);
 		check_extent_overbig(trans, &iter, k);
 	}));
@@ -1854,7 +1854,7 @@ int bch2_check_dirents(struct bch_fs *c)
 			BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS,
 			k,
 			NULL, NULL,
-			BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
+			BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc,
 		check_dirent(trans, &iter, k, &hash_info, &dir, &target, &s));
 
 	bch2_trans_put(trans);
@@ -1918,7 +1918,7 @@ int bch2_check_xattrs(struct bch_fs *c)
 			BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS,
 			k,
 			NULL, NULL,
-			BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
+			BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc,
 		check_xattr(trans, &iter, k, &hash_info, &inode)));
 	bch_err_fn(c, ret);
 	return ret;
@@ -1949,8 +1949,8 @@ static int check_root_trans(struct btree_trans *trans)
 		root_subvol.v.snapshot	= cpu_to_le32(snapshot);
 		root_subvol.v.inode	= cpu_to_le64(inum);
 		ret = commit_do(trans, NULL, NULL,
-				      BTREE_INSERT_NOFAIL|
+				      BCH_TRANS_COMMIT_no_enospc|
-				      BTREE_INSERT_LAZY_RW,
+				      BCH_TRANS_COMMIT_lazy_rw,
 			bch2_btree_insert_trans(trans, BTREE_ID_subvolumes,
 					    &root_subvol.k_i, 0));
 		bch_err_msg(c, ret, "writing root subvol");
@@ -1986,8 +1986,8 @@ int bch2_check_root(struct bch_fs *c)
 	int ret;
 
 	ret = bch2_trans_do(c, NULL, NULL,
-			     BTREE_INSERT_NOFAIL|
+			     BCH_TRANS_COMMIT_no_enospc|
-			     BTREE_INSERT_LAZY_RW,
+			     BCH_TRANS_COMMIT_lazy_rw,
 		check_root_trans(trans));
 	bch_err_fn(c, ret);
 	return ret;
@@ -2116,8 +2116,8 @@ static int check_path(struct btree_trans *trans,
 				return 0;
 
 			ret = commit_do(trans, NULL, NULL,
-					      BTREE_INSERT_NOFAIL|
+					      BCH_TRANS_COMMIT_no_enospc|
-					      BTREE_INSERT_LAZY_RW,
+					      BCH_TRANS_COMMIT_lazy_rw,
 					remove_backpointer(trans, inode));
 			if (ret) {
 				bch_err(c, "error removing dirent: %i", ret);
@@ -2398,7 +2398,7 @@ static int check_nlinks_update_hardlinks(struct bch_fs *c,
 		for_each_btree_key_commit(trans, iter, BTREE_ID_inodes,
 				POS(0, range_start),
 				BTREE_ITER_INTENT|BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS, k,
-				NULL, NULL, BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
+				NULL, NULL, BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc,
 			check_nlinks_update_inode(trans, &iter, k, links, &idx, range_end)));
 	if (ret < 0) {
 		bch_err(c, "error in fsck: btree error %i while walking inodes", ret);
@@ -2483,7 +2483,7 @@ int bch2_fix_reflink_p(struct bch_fs *c)
 				BTREE_ID_extents, POS_MIN,
 				BTREE_ITER_INTENT|BTREE_ITER_PREFETCH|
 				BTREE_ITER_ALL_SNAPSHOTS, k,
-				NULL, NULL, BTREE_INSERT_NOFAIL|BTREE_INSERT_LAZY_RW,
+				NULL, NULL, BCH_TRANS_COMMIT_no_enospc|BCH_TRANS_COMMIT_lazy_rw,
 			fix_reflink_p_key(trans, &iter, k)));
 	bch_err_fn(c, ret);
 	return ret;

libbcachefs/inode.c

@@ -830,7 +830,7 @@ static int bch2_inode_delete_keys(struct btree_trans *trans,
 
 		ret = bch2_trans_update(trans, &iter, &delete, 0) ?:
 		      bch2_trans_commit(trans, NULL, NULL,
-					BTREE_INSERT_NOFAIL);
+					BCH_TRANS_COMMIT_no_enospc);
 err:
 		if (ret && !bch2_err_matches(ret, BCH_ERR_transaction_restart))
 			break;
@@ -893,7 +893,7 @@ retry:
 
 	ret   = bch2_trans_update(trans, &iter, &delete.k_i, 0) ?:
 		bch2_trans_commit(trans, NULL, NULL,
-				BTREE_INSERT_NOFAIL);
+				BCH_TRANS_COMMIT_no_enospc);
 err:
 	bch2_trans_iter_exit(trans, &iter);
 	if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
@@ -1057,7 +1057,7 @@ retry:
 
 	ret   = bch2_trans_update(trans, &iter, &delete.k_i, 0) ?:
 		bch2_trans_commit(trans, NULL, NULL,
-				BTREE_INSERT_NOFAIL);
+				BCH_TRANS_COMMIT_no_enospc);
 err:
 	bch2_trans_iter_exit(trans, &iter);
 	if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
@@ -1091,7 +1091,7 @@ static int may_delete_deleted_inode(struct btree_trans *trans,
 
 	ret = bch2_inode_unpack(k, &inode);
 	if (ret)
-		goto err;
+		goto out;
 
 	if (fsck_err_on(S_ISDIR(inode.bi_mode), c,
 			deleted_inode_is_dir,
@@ -1109,38 +1109,45 @@ static int may_delete_deleted_inode(struct btree_trans *trans,
 	    !fsck_err(c,
 		      deleted_inode_but_clean,
 		      "filesystem marked as clean but have deleted inode %llu:%u",
-		      pos.offset, pos.snapshot))
+		      pos.offset, pos.snapshot)) {
-		return 0;
+		ret = 0;
+		goto out;
+	}
 
 	if (bch2_snapshot_is_internal_node(c, pos.snapshot)) {
 		struct bpos new_min_pos;
 
 		ret = bch2_propagate_key_to_snapshot_leaves(trans, inode_iter.btree_id, k, &new_min_pos);
 		if (ret)
-			goto err;
+			goto out;
 
 		inode.bi_flags &= ~BCH_INODE_unlinked;
 
 		ret = bch2_inode_write_flags(trans, &inode_iter, &inode,
-					     BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE);
+					     BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE) ?:
+			bch2_trans_commit(trans, NULL, NULL,
+					  BCH_TRANS_COMMIT_no_enospc|
+					  BCH_TRANS_COMMIT_lazy_rw);
 		bch_err_msg(c, ret, "clearing inode unlinked flag");
 		if (ret)
-			return ret;
+			goto out;
 
 		/*
 		 * We'll need another write buffer flush to pick up the new
 		 * unlinked inodes in the snapshot leaves:
 		 */
 		*need_another_pass = true;
-		return 0;
+		goto out;
 	}
 
-	return 1;
+	ret = 1;
-err:
+out:
 fsck_err:
+	bch2_trans_iter_exit(trans, &inode_iter);
 	return ret;
 delete:
-	return bch2_btree_bit_mod(trans, BTREE_ID_deleted_inodes, pos, false);
+	ret = bch2_btree_bit_mod(trans, BTREE_ID_deleted_inodes, pos, false);
+	goto out;
 }
 
 int bch2_delete_dead_inodes(struct bch_fs *c)

libbcachefs/io_misc.c

@@ -256,7 +256,7 @@ static int __bch2_resume_logged_op_truncate(struct btree_trans *trans,
 	u64 new_i_size = le64_to_cpu(op->v.new_i_size);
 	int ret;
 
-	ret = commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+	ret = commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 			truncate_set_isize(trans, inum, new_i_size));
 	if (ret)
 		goto err;
@@ -378,7 +378,7 @@ case LOGGED_OP_FINSERT_start:
 	op->v.state = LOGGED_OP_FINSERT_shift_extents;
 
 	if (insert) {
-		ret = commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+		ret = commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 				adjust_i_size(trans, inum, src_offset, len) ?:
 				bch2_logged_op_update(trans, &op->k_i));
 		if (ret)
@@ -390,7 +390,7 @@ case LOGGED_OP_FINSERT_start:
 		if (ret && !bch2_err_matches(ret, BCH_ERR_transaction_restart))
 			goto err;
 
-		ret = commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+		ret = commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 				bch2_logged_op_update(trans, &op->k_i));
 	}
 
@@ -455,7 +455,7 @@ case LOGGED_OP_FINSERT_shift_extents:
 			bch2_btree_insert_trans(trans, BTREE_ID_extents, &delete, 0) ?:
 			bch2_btree_insert_trans(trans, BTREE_ID_extents, copy, 0) ?:
 			bch2_logged_op_update(trans, &op->k_i) ?:
-			bch2_trans_commit(trans, &disk_res, NULL, BTREE_INSERT_NOFAIL);
+			bch2_trans_commit(trans, &disk_res, NULL, BCH_TRANS_COMMIT_no_enospc);
 btree_err:
 		bch2_disk_reservation_put(c, &disk_res);
 
@@ -470,12 +470,12 @@ btree_err:
 	op->v.state = LOGGED_OP_FINSERT_finish;
 
 	if (!insert) {
-		ret = commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+		ret = commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 				adjust_i_size(trans, inum, src_offset, shift) ?:
 				bch2_logged_op_update(trans, &op->k_i));
 	} else {
 		/* We need an inode update to update bi_journal_seq for fsync: */
-		ret = commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+		ret = commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 				adjust_i_size(trans, inum, 0, 0) ?:
 				bch2_logged_op_update(trans, &op->k_i));
 	}

libbcachefs/io_read.c

@@ -526,7 +526,7 @@ out:
 
 static noinline void bch2_rbio_narrow_crcs(struct bch_read_bio *rbio)
 {
-	bch2_trans_do(rbio->c, NULL, NULL, BTREE_INSERT_NOFAIL,
+	bch2_trans_do(rbio->c, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 		      __bch2_rbio_narrow_crcs(trans, rbio));
 }
 
@@ -1025,7 +1025,7 @@ get_bio:
 		trans->notrace_relock_fail = true;
 	} else {
 		/* Attempting reconstruct read: */
-		if (bch2_ec_read_extent(c, rbio)) {
+		if (bch2_ec_read_extent(trans, rbio)) {
 			bch2_rbio_error(rbio, READ_RETRY_AVOID, BLK_STS_IOERR);
 			goto out;
 		}

libbcachefs/io_write.c

@@ -202,6 +202,17 @@ static inline int bch2_extent_update_i_size_sectors(struct btree_trans *trans,
 	struct btree_iter iter;
 	struct bkey_i *k;
 	struct bkey_i_inode_v3 *inode;
+	/*
+	 * Crazy performance optimization:
+	 * Every extent update needs to also update the inode: the inode trigger
+	 * will set bi->journal_seq to the journal sequence number of this
+	 * transaction - for fsync.
+	 *
+	 * But if that's the only reason we're updating the inode (we're not
+	 * updating bi_size or bi_sectors), then we don't need the inode update
+	 * to be journalled - if we crash, the bi_journal_seq update will be
+	 * lost, but that's fine.
+	 */
 	unsigned inode_update_flags = BTREE_UPDATE_NOJOURNAL;
 	int ret;
 
@@ -305,8 +316,8 @@ int bch2_extent_update(struct btree_trans *trans,
 						  i_sectors_delta) ?:
 		bch2_trans_update(trans, iter, k, 0) ?:
 		bch2_trans_commit(trans, disk_res, NULL,
-				BTREE_INSERT_NOCHECK_RW|
+				BCH_TRANS_COMMIT_no_check_rw|
-				BTREE_INSERT_NOFAIL);
+				BCH_TRANS_COMMIT_no_enospc);
 	if (unlikely(ret))
 		return ret;
 
@@ -1165,7 +1176,7 @@ static void bch2_nocow_write_convert_unwritten(struct bch_write_op *op)
 		ret = for_each_btree_key_upto_commit(trans, iter, BTREE_ID_extents,
 				     bkey_start_pos(&orig->k), orig->k.p,
 				     BTREE_ITER_INTENT, k,
-				     NULL, NULL, BTREE_INSERT_NOFAIL, ({
+				     NULL, NULL, BCH_TRANS_COMMIT_no_enospc, ({
 			bch2_nocow_write_convert_one_unwritten(trans, &iter, orig, k, op->new_i_size);
 		}));
 

libbcachefs/journal.c

@@ -361,11 +361,6 @@ static int journal_entry_open(struct journal *j)
 	} while ((v = atomic64_cmpxchg(&j->reservations.counter,
 				       old.v, new.v)) != old.v);
 
-	if (j->res_get_blocked_start)
-		bch2_time_stats_update(j->blocked_time,
-				       j->res_get_blocked_start);
-	j->res_get_blocked_start = 0;
-
 	mod_delayed_work(c->io_complete_wq,
 			 &j->write_work,
 			 msecs_to_jiffies(c->opts.journal_flush_delay));
@@ -465,15 +460,12 @@ retry:
 	__journal_entry_close(j, JOURNAL_ENTRY_CLOSED_VAL);
 	ret = journal_entry_open(j);
 
-	if (ret == JOURNAL_ERR_max_in_flight)
+	if (ret == JOURNAL_ERR_max_in_flight) {
+		track_event_change(&c->times[BCH_TIME_blocked_journal_max_in_flight],
+				   &j->max_in_flight_start, true);
 		trace_and_count(c, journal_entry_full, c);
-unlock:
-	if ((ret && ret != JOURNAL_ERR_insufficient_devices) &&
-	    !j->res_get_blocked_start) {
-		j->res_get_blocked_start = local_clock() ?: 1;
-		trace_and_count(c, journal_full, c);
 	}
-
+unlock:
 	can_discard = j->can_discard;
 	spin_unlock(&j->lock);
 
@@ -526,36 +518,6 @@ int bch2_journal_res_get_slowpath(struct journal *j, struct journal_res *res,
 	return ret;
 }
 
-/* journal_preres: */
-
-static bool journal_preres_available(struct journal *j,
-				     struct journal_preres *res,
-				     unsigned new_u64s,
-				     unsigned flags)
-{
-	bool ret = bch2_journal_preres_get_fast(j, res, new_u64s, flags, true);
-
-	if (!ret && mutex_trylock(&j->reclaim_lock)) {
-		bch2_journal_reclaim(j);
-		mutex_unlock(&j->reclaim_lock);
-	}
-
-	return ret;
-}
-
-int __bch2_journal_preres_get(struct journal *j,
-			      struct journal_preres *res,
-			      unsigned new_u64s,
-			      unsigned flags)
-{
-	int ret;
-
-	closure_wait_event(&j->preres_wait,
-		   (ret = bch2_journal_error(j)) ||
-		   journal_preres_available(j, res, new_u64s, flags));
-	return ret;
-}
-
 /* journal_entry_res: */
 
 void bch2_journal_entry_res_resize(struct journal *j,
@@ -1290,6 +1252,7 @@ void __bch2_journal_debug_to_text(struct printbuf *out, struct journal *j)
 	union journal_res_state s;
 	struct bch_dev *ca;
 	unsigned long now = jiffies;
+	u64 nr_writes = j->nr_flush_writes + j->nr_noflush_writes;
 	u64 seq;
 	unsigned i;
 
@@ -1303,21 +1266,23 @@ void __bch2_journal_debug_to_text(struct printbuf *out, struct journal *j)
 	prt_printf(out, "dirty journal entries:\t%llu/%llu\n",	fifo_used(&j->pin), j->pin.size);
 	prt_printf(out, "seq:\t\t\t%llu\n",			journal_cur_seq(j));
 	prt_printf(out, "seq_ondisk:\t\t%llu\n",		j->seq_ondisk);
-	prt_printf(out, "last_seq:\t\t%llu\n",		journal_last_seq(j));
+	prt_printf(out, "last_seq:\t\t%llu\n",			journal_last_seq(j));
 	prt_printf(out, "last_seq_ondisk:\t%llu\n",		j->last_seq_ondisk);
-	prt_printf(out, "flushed_seq_ondisk:\t%llu\n",	j->flushed_seq_ondisk);
+	prt_printf(out, "flushed_seq_ondisk:\t%llu\n",		j->flushed_seq_ondisk);
-	prt_printf(out, "prereserved:\t\t%u/%u\n",		j->prereserved.reserved, j->prereserved.remaining);
+	prt_printf(out, "watermark:\t\t%s\n",			bch2_watermarks[j->watermark]);
-	prt_printf(out, "watermark:\t\t%s\n",		bch2_watermarks[j->watermark]);
+	prt_printf(out, "each entry reserved:\t%u\n",		j->entry_u64s_reserved);
-	prt_printf(out, "each entry reserved:\t%u\n",	j->entry_u64s_reserved);
 	prt_printf(out, "nr flush writes:\t%llu\n",		j->nr_flush_writes);
-	prt_printf(out, "nr noflush writes:\t%llu\n",	j->nr_noflush_writes);
+	prt_printf(out, "nr noflush writes:\t%llu\n",		j->nr_noflush_writes);
-	prt_printf(out, "nr direct reclaim:\t%llu\n",	j->nr_direct_reclaim);
+	prt_printf(out, "average write size:\t");
+	prt_human_readable_u64(out, nr_writes ? div64_u64(j->entry_bytes_written, nr_writes) : 0);
+	prt_newline(out);
+	prt_printf(out, "nr direct reclaim:\t%llu\n",		j->nr_direct_reclaim);
 	prt_printf(out, "nr background reclaim:\t%llu\n",	j->nr_background_reclaim);
 	prt_printf(out, "reclaim kicked:\t\t%u\n",		j->reclaim_kicked);
-	prt_printf(out, "reclaim runs in:\t%u ms\n",	time_after(j->next_reclaim, now)
+	prt_printf(out, "reclaim runs in:\t%u ms\n",		time_after(j->next_reclaim, now)
 	       ? jiffies_to_msecs(j->next_reclaim - jiffies) : 0);
-	prt_printf(out, "current entry sectors:\t%u\n",	j->cur_entry_sectors);
+	prt_printf(out, "current entry sectors:\t%u\n",		j->cur_entry_sectors);
-	prt_printf(out, "current entry error:\t%s\n",	bch2_journal_errors[j->cur_entry_error]);
+	prt_printf(out, "current entry error:\t%s\n",		bch2_journal_errors[j->cur_entry_error]);
 	prt_printf(out, "current entry:\t\t");
 
 	switch (s.cur_entry_offset) {

libbcachefs/journal.h

@@ -395,104 +395,6 @@ out:
 	return 0;
 }
 
-/* journal_preres: */
-
-static inline void journal_set_watermark(struct journal *j)
-{
-	union journal_preres_state s = READ_ONCE(j->prereserved);
-	unsigned watermark = BCH_WATERMARK_stripe;
-
-	if (fifo_free(&j->pin) < j->pin.size / 4)
-		watermark = max_t(unsigned, watermark, BCH_WATERMARK_copygc);
-	if (fifo_free(&j->pin) < j->pin.size / 8)
-		watermark = max_t(unsigned, watermark, BCH_WATERMARK_reclaim);
-
-	if (s.reserved > s.remaining)
-		watermark = max_t(unsigned, watermark, BCH_WATERMARK_copygc);
-	if (!s.remaining)
-		watermark = max_t(unsigned, watermark, BCH_WATERMARK_reclaim);
-
-	if (watermark == j->watermark)
-		return;
-
-	swap(watermark, j->watermark);
-	if (watermark > j->watermark)
-		journal_wake(j);
-}
-
-static inline void bch2_journal_preres_put(struct journal *j,
-					   struct journal_preres *res)
-{
-	union journal_preres_state s = { .reserved = res->u64s };
-
-	if (!res->u64s)
-		return;
-
-	s.v = atomic64_sub_return(s.v, &j->prereserved.counter);
-	res->u64s = 0;
-
-	if (unlikely(s.waiting)) {
-		clear_bit(ilog2((((union journal_preres_state) { .waiting = 1 }).v)),
-			  (unsigned long *) &j->prereserved.v);
-		closure_wake_up(&j->preres_wait);
-	}
-
-	if (s.reserved <= s.remaining && j->watermark)
-		journal_set_watermark(j);
-}
-
-int __bch2_journal_preres_get(struct journal *,
-			struct journal_preres *, unsigned, unsigned);
-
-static inline int bch2_journal_preres_get_fast(struct journal *j,
-					       struct journal_preres *res,
-					       unsigned new_u64s,
-					       unsigned flags,
-					       bool set_waiting)
-{
-	int d = new_u64s - res->u64s;
-	union journal_preres_state old, new;
-	u64 v = atomic64_read(&j->prereserved.counter);
-	enum bch_watermark watermark = flags & BCH_WATERMARK_MASK;
-	int ret;
-
-	do {
-		old.v = new.v = v;
-		ret = 0;
-
-		if (watermark == BCH_WATERMARK_reclaim ||
-		    new.reserved + d < new.remaining) {
-			new.reserved += d;
-			ret = 1;
-		} else if (set_waiting && !new.waiting)
-			new.waiting = true;
-		else
-			return 0;
-	} while ((v = atomic64_cmpxchg(&j->prereserved.counter,
-				       old.v, new.v)) != old.v);
-
-	if (ret)
-		res->u64s += d;
-	return ret;
-}
-
-static inline int bch2_journal_preres_get(struct journal *j,
-					  struct journal_preres *res,
-					  unsigned new_u64s,
-					  unsigned flags)
-{
-	if (new_u64s <= res->u64s)
-		return 0;
-
-	if (bch2_journal_preres_get_fast(j, res, new_u64s, flags, false))
-		return 0;
-
-	if (flags & JOURNAL_RES_GET_NONBLOCK)
-		return -BCH_ERR_journal_preres_get_blocked;
-
-	return __bch2_journal_preres_get(j, res, new_u64s, flags);
-}
-
 /* journal_entry_res: */
 
 void bch2_journal_entry_res_resize(struct journal *,

libbcachefs/journal_io.c

@@ -1079,6 +1079,12 @@ found:
 
 	if (ja->bucket_seq[ja->cur_idx] &&
 	    ja->sectors_free == ca->mi.bucket_size) {
+#if 0
+		/*
+		 * Debug code for ZNS support, where we (probably) want to be
+		 * correlated where we stopped in the journal to the zone write
+		 * points:
+		 */
 		bch_err(c, "ja->sectors_free == ca->mi.bucket_size");
 		bch_err(c, "cur_idx %u/%u", ja->cur_idx, ja->nr);
 		for (i = 0; i < 3; i++) {
@@ -1086,6 +1092,7 @@ found:
 
 			bch_err(c, "bucket_seq[%u] = %llu", idx, ja->bucket_seq[idx]);
 		}
+#endif
 		ja->sectors_free = 0;
 	}
 
@@ -1585,6 +1592,9 @@ static void journal_write_done(struct closure *cl)
 
 	bch2_journal_space_available(j);
 
+	track_event_change(&c->times[BCH_TIME_blocked_journal_max_in_flight],
+			   &j->max_in_flight_start, false);
+
 	closure_wake_up(&w->wait);
 	journal_wake(j);
 
@@ -1678,9 +1688,15 @@ static void do_journal_write(struct closure *cl)
 	continue_at(cl, journal_write_done, c->io_complete_wq);
 }
 
-static void bch2_journal_entries_postprocess(struct bch_fs *c, struct jset *jset)
+static int bch2_journal_write_prep(struct journal *j, struct journal_buf *w)
 {
-	struct jset_entry *i, *next, *prev = NULL;
+	struct bch_fs *c = container_of(j, struct bch_fs, journal);
+	struct jset_entry *start, *end, *i, *next, *prev = NULL;
+	struct jset *jset = w->data;
+	unsigned sectors, bytes, u64s;
+	bool validate_before_checksum = false;
+	unsigned long btree_roots_have = 0;
+	int ret;
 
 	/*
 	 * Simple compaction, dropping empty jset_entries (from journal
@@ -1697,8 +1713,20 @@ static void bch2_journal_entries_postprocess(struct bch_fs *c, struct jset *jset
 		if (!u64s)
 			continue;
 
-		if (i->type == BCH_JSET_ENTRY_btree_root)
+		/*
+		 * New btree roots are set by journalling them; when the journal
+		 * entry gets written we have to propagate them to
+		 * c->btree_roots
+		 *
+		 * But, every journal entry we write has to contain all the
+		 * btree roots (at least for now); so after we copy btree roots
+		 * to c->btree_roots we have to get any missing btree roots and
+		 * add them to this journal entry:
+		 */
+		if (i->type == BCH_JSET_ENTRY_btree_root) {
 			bch2_journal_entry_to_btree_root(c, i);
+			__set_bit(i->btree_id, &btree_roots_have);
+		}
 
 		/* Can we merge with previous entry? */
 		if (prev &&
@@ -1722,85 +1750,10 @@ static void bch2_journal_entries_postprocess(struct bch_fs *c, struct jset *jset
 
 	prev = prev ? vstruct_next(prev) : jset->start;
 	jset->u64s = cpu_to_le32((u64 *) prev - jset->_data);
-}
-
-void bch2_journal_write(struct closure *cl)
-{
-	struct journal *j = container_of(cl, struct journal, io);
-	struct bch_fs *c = container_of(j, struct bch_fs, journal);
-	struct bch_dev *ca;
-	struct journal_buf *w = journal_last_unwritten_buf(j);
-	struct bch_replicas_padded replicas;
-	struct jset_entry *start, *end;
-	struct jset *jset;
-	struct bio *bio;
-	struct printbuf journal_debug_buf = PRINTBUF;
-	bool validate_before_checksum = false;
-	unsigned i, sectors, bytes, u64s, nr_rw_members = 0;
-	int ret;
-
-	BUG_ON(BCH_SB_CLEAN(c->disk_sb.sb));
-
-	journal_buf_realloc(j, w);
-	jset = w->data;
-
-	j->write_start_time = local_clock();
-
-	spin_lock(&j->lock);
-
-	/*
-	 * If the journal is in an error state - we did an emergency shutdown -
-	 * we prefer to continue doing journal writes. We just mark them as
-	 * noflush so they'll never be used, but they'll still be visible by the
-	 * list_journal tool - this helps in debugging.
-	 *
-	 * There's a caveat: the first journal write after marking the
-	 * superblock dirty must always be a flush write, because on startup
-	 * from a clean shutdown we didn't necessarily read the journal and the
-	 * new journal write might overwrite whatever was in the journal
-	 * previously - we can't leave the journal without any flush writes in
-	 * it.
-	 *
-	 * So if we're in an error state, and we're still starting up, we don't
-	 * write anything at all.
-	 */
-	if (!test_bit(JOURNAL_NEED_FLUSH_WRITE, &j->flags) &&
-	    (bch2_journal_error(j) ||
-	     w->noflush ||
-	     (!w->must_flush &&
-	      (jiffies - j->last_flush_write) < msecs_to_jiffies(c->opts.journal_flush_delay) &&
-	      test_bit(JOURNAL_MAY_SKIP_FLUSH, &j->flags)))) {
-		w->noflush = true;
-		SET_JSET_NO_FLUSH(jset, true);
-		jset->last_seq	= 0;
-		w->last_seq	= 0;
-
-		j->nr_noflush_writes++;
-	} else if (!bch2_journal_error(j)) {
-		j->last_flush_write = jiffies;
-		j->nr_flush_writes++;
-		clear_bit(JOURNAL_NEED_FLUSH_WRITE, &j->flags);
-	} else {
-		spin_unlock(&j->lock);
-		goto err;
-	}
-	spin_unlock(&j->lock);
-
-	/*
-	 * New btree roots are set by journalling them; when the journal entry
-	 * gets written we have to propagate them to c->btree_roots
-	 *
-	 * But, every journal entry we write has to contain all the btree roots
-	 * (at least for now); so after we copy btree roots to c->btree_roots we
-	 * have to get any missing btree roots and add them to this journal
-	 * entry:
-	 */
-
-	bch2_journal_entries_postprocess(c, jset);
 
 	start = end = vstruct_last(jset);
 
-	end	= bch2_btree_roots_to_journal_entries(c, jset->start, end);
+	end	= bch2_btree_roots_to_journal_entries(c, end, btree_roots_have);
 
 	bch2_journal_super_entries_add_common(c, &end,
 				le64_to_cpu(jset->seq));
@@ -1816,7 +1769,7 @@ void bch2_journal_write(struct closure *cl)
 		bch2_fs_fatal_error(c, "aieeee! journal write overran available space, %zu > %u (extra %u reserved %u/%u)",
 				    vstruct_bytes(jset), w->sectors << 9,
 				    u64s, w->u64s_reserved, j->entry_u64s_reserved);
-		goto err;
+		return -EINVAL;
 	}
 
 	jset->magic		= cpu_to_le64(jset_magic(c));
@@ -1835,37 +1788,119 @@ void bch2_journal_write(struct closure *cl)
 		validate_before_checksum = true;
 
 	if (validate_before_checksum &&
-	    jset_validate(c, NULL, jset, 0, WRITE))
+	    (ret = jset_validate(c, NULL, jset, 0, WRITE)))
-		goto err;
+		return ret;
 
 	ret = bch2_encrypt(c, JSET_CSUM_TYPE(jset), journal_nonce(jset),
 		    jset->encrypted_start,
 		    vstruct_end(jset) - (void *) jset->encrypted_start);
 	if (bch2_fs_fatal_err_on(ret, c,
 			"error decrypting journal entry: %i", ret))
-		goto err;
+		return ret;
 
 	jset->csum = csum_vstruct(c, JSET_CSUM_TYPE(jset),
 				  journal_nonce(jset), jset);
 
 	if (!validate_before_checksum &&
-	    jset_validate(c, NULL, jset, 0, WRITE))
+	    (ret = jset_validate(c, NULL, jset, 0, WRITE)))
-		goto err;
+		return ret;
 
 	memset((void *) jset + bytes, 0, (sectors << 9) - bytes);
+	return 0;
+}
 
-retry_alloc:
+static int bch2_journal_write_pick_flush(struct journal *j, struct journal_buf *w)
-	spin_lock(&j->lock);
+{
-	ret = journal_write_alloc(j, w);
+	struct bch_fs *c = container_of(j, struct bch_fs, journal);
+	int error = bch2_journal_error(j);
 
-	if (ret && j->can_discard) {
+	/*
-		spin_unlock(&j->lock);
+	 * If the journal is in an error state - we did an emergency shutdown -
-		bch2_journal_do_discards(j);
+	 * we prefer to continue doing journal writes. We just mark them as
-		goto retry_alloc;
+	 * noflush so they'll never be used, but they'll still be visible by the
+	 * list_journal tool - this helps in debugging.
+	 *
+	 * There's a caveat: the first journal write after marking the
+	 * superblock dirty must always be a flush write, because on startup
+	 * from a clean shutdown we didn't necessarily read the journal and the
+	 * new journal write might overwrite whatever was in the journal
+	 * previously - we can't leave the journal without any flush writes in
+	 * it.
+	 *
+	 * So if we're in an error state, and we're still starting up, we don't
+	 * write anything at all.
+	 */
+	if (error && test_bit(JOURNAL_NEED_FLUSH_WRITE, &j->flags))
+		return -EIO;
+
+	if (error ||
+	    w->noflush ||
+	    (!w->must_flush &&
+	     (jiffies - j->last_flush_write) < msecs_to_jiffies(c->opts.journal_flush_delay) &&
+	     test_bit(JOURNAL_MAY_SKIP_FLUSH, &j->flags))) {
+		     w->noflush = true;
+		SET_JSET_NO_FLUSH(w->data, true);
+		w->data->last_seq	= 0;
+		w->last_seq		= 0;
+
+		j->nr_noflush_writes++;
+	} else {
+		j->last_flush_write = jiffies;
+		j->nr_flush_writes++;
+		clear_bit(JOURNAL_NEED_FLUSH_WRITE, &j->flags);
 	}
 
+	return 0;
+}
+
+void bch2_journal_write(struct closure *cl)
+{
+	struct journal *j = container_of(cl, struct journal, io);
+	struct bch_fs *c = container_of(j, struct bch_fs, journal);
+	struct bch_dev *ca;
+	struct journal_buf *w = journal_last_unwritten_buf(j);
+	struct bch_replicas_padded replicas;
+	struct bio *bio;
+	struct printbuf journal_debug_buf = PRINTBUF;
+	unsigned i, nr_rw_members = 0;
+	int ret;
+
+	BUG_ON(BCH_SB_CLEAN(c->disk_sb.sb));
+
+	j->write_start_time = local_clock();
+
+	spin_lock(&j->lock);
+	ret = bch2_journal_write_pick_flush(j, w);
+	spin_unlock(&j->lock);
 	if (ret)
+		goto err;
+
+	journal_buf_realloc(j, w);
+
+	ret = bch2_journal_write_prep(j, w);
+	if (ret)
+		goto err;
+
+	j->entry_bytes_written += vstruct_bytes(w->data);
+
+	while (1) {
+		spin_lock(&j->lock);
+		ret = journal_write_alloc(j, w);
+		if (!ret || !j->can_discard)
+			break;
+
+		spin_unlock(&j->lock);
+		bch2_journal_do_discards(j);
+	}
+
+	if (ret) {
 		__bch2_journal_debug_to_text(&journal_debug_buf, j);
+		spin_unlock(&j->lock);
+		bch_err(c, "Unable to allocate journal write:\n%s",
+			journal_debug_buf.buf);
+		printbuf_exit(&journal_debug_buf);
+		goto err;
+	}
 
 	/*
 	 * write is allocated, no longer need to account for it in
@@ -1880,13 +1915,6 @@ retry_alloc:
 	bch2_journal_space_available(j);
 	spin_unlock(&j->lock);
 
-	if (ret) {
-		bch_err(c, "Unable to allocate journal write:\n%s",
-			journal_debug_buf.buf);
-		printbuf_exit(&journal_debug_buf);
-		goto err;
-	}
-
 	w->devs_written = bch2_bkey_devs(bkey_i_to_s_c(&w->key));
 
 	if (c->opts.nochanges)
@@ -1908,7 +1936,7 @@ retry_alloc:
 	if (ret)
 		goto err;
 
-	if (!JSET_NO_FLUSH(jset) && w->separate_flush) {
+	if (!JSET_NO_FLUSH(w->data) && w->separate_flush) {
 		for_each_rw_member(ca, c, i) {
 			percpu_ref_get(&ca->io_ref);
 

libbcachefs/journal_reclaim.c

@@ -50,16 +50,25 @@ unsigned bch2_journal_dev_buckets_available(struct journal *j,
 	return available;
 }
 
-static void journal_set_remaining(struct journal *j, unsigned u64s_remaining)
+static inline void journal_set_watermark(struct journal *j)
 {
-	union journal_preres_state old, new;
+	struct bch_fs *c = container_of(j, struct bch_fs, journal);
-	u64 v = atomic64_read(&j->prereserved.counter);
+	bool low_on_space = j->space[journal_space_clean].total * 4 <=
+		j->space[journal_space_total].total;
+	bool low_on_pin = fifo_free(&j->pin) < j->pin.size / 4;
+	unsigned watermark = low_on_space || low_on_pin
+		? BCH_WATERMARK_reclaim
+		: BCH_WATERMARK_stripe;
 
-	do {
+	if (track_event_change(&c->times[BCH_TIME_blocked_journal_low_on_space],
-		old.v = new.v = v;
+			       &j->low_on_space_start, low_on_space) ||
-		new.remaining = u64s_remaining;
+	    track_event_change(&c->times[BCH_TIME_blocked_journal_low_on_pin],
-	} while ((v = atomic64_cmpxchg(&j->prereserved.counter,
+			       &j->low_on_pin_start, low_on_pin))
-				       old.v, new.v)) != old.v);
+		trace_and_count(c, journal_full, c);
+
+	swap(watermark, j->watermark);
+	if (watermark > j->watermark)
+		journal_wake(j);
 }
 
 static struct journal_space
@@ -162,7 +171,6 @@ void bch2_journal_space_available(struct journal *j)
 	struct bch_fs *c = container_of(j, struct bch_fs, journal);
 	struct bch_dev *ca;
 	unsigned clean, clean_ondisk, total;
-	s64 u64s_remaining = 0;
 	unsigned max_entry_size	 = min(j->buf[0].buf_size >> 9,
 				       j->buf[1].buf_size >> 9);
 	unsigned i, nr_online = 0, nr_devs_want;
@@ -222,16 +230,10 @@ void bch2_journal_space_available(struct journal *j)
 	else
 		clear_bit(JOURNAL_MAY_SKIP_FLUSH, &j->flags);
 
-	u64s_remaining  = (u64) clean << 6;
+	journal_set_watermark(j);
-	u64s_remaining -= (u64) total << 3;
-	u64s_remaining = max(0LL, u64s_remaining);
-	u64s_remaining /= 4;
-	u64s_remaining = min_t(u64, u64s_remaining, U32_MAX);
 out:
 	j->cur_entry_sectors	= !ret ? j->space[journal_space_discarded].next_entry : 0;
 	j->cur_entry_error	= ret;
-	journal_set_remaining(j, u64s_remaining);
-	journal_set_watermark(j);
 
 	if (!ret)
 		journal_wake(j);
@@ -369,15 +371,36 @@ static enum journal_pin_type journal_pin_type(journal_pin_flush_fn fn)
 		return JOURNAL_PIN_other;
 }
 
-void bch2_journal_pin_set(struct journal *j, u64 seq,
+static inline void bch2_journal_pin_set_locked(struct journal *j, u64 seq,
 			  struct journal_entry_pin *pin,
-			  journal_pin_flush_fn flush_fn)
+			  journal_pin_flush_fn flush_fn,
+			  enum journal_pin_type type)
+{
+	struct journal_entry_pin_list *pin_list = journal_seq_pin(j, seq);
+
+	/*
+	 * flush_fn is how we identify journal pins in debugfs, so must always
+	 * exist, even if it doesn't do anything:
+	 */
+	BUG_ON(!flush_fn);
+
+	atomic_inc(&pin_list->count);
+	pin->seq	= seq;
+	pin->flush	= flush_fn;
+	list_add(&pin->list, &pin_list->list[type]);
+}
+
+void bch2_journal_pin_copy(struct journal *j,
+			   struct journal_entry_pin *dst,
+			   struct journal_entry_pin *src,
+			   journal_pin_flush_fn flush_fn)
 {
-	struct journal_entry_pin_list *pin_list;
 	bool reclaim;
 
 	spin_lock(&j->lock);
 
+	u64 seq = READ_ONCE(src->seq);
+
 	if (seq < journal_last_seq(j)) {
 		/*
 		 * bch2_journal_pin_copy() raced with bch2_journal_pin_drop() on
@@ -389,18 +412,34 @@ void bch2_journal_pin_set(struct journal *j, u64 seq,
 		return;
 	}
 
-	pin_list = journal_seq_pin(j, seq);
+	reclaim = __journal_pin_drop(j, dst);
+
+	bch2_journal_pin_set_locked(j, seq, dst, flush_fn, journal_pin_type(flush_fn));
+
+	if (reclaim)
+		bch2_journal_reclaim_fast(j);
+	spin_unlock(&j->lock);
+
+	/*
+	 * If the journal is currently full,  we might want to call flush_fn
+	 * immediately:
+	 */
+	journal_wake(j);
+}
+
+void bch2_journal_pin_set(struct journal *j, u64 seq,
+			  struct journal_entry_pin *pin,
+			  journal_pin_flush_fn flush_fn)
+{
+	bool reclaim;
+
+	spin_lock(&j->lock);
+
+	BUG_ON(seq < journal_last_seq(j));
 
 	reclaim = __journal_pin_drop(j, pin);
 
-	atomic_inc(&pin_list->count);
+	bch2_journal_pin_set_locked(j, seq, pin, flush_fn, journal_pin_type(flush_fn));
-	pin->seq	= seq;
-	pin->flush	= flush_fn;
-
-	if (flush_fn)
-		list_add(&pin->list, &pin_list->list[journal_pin_type(flush_fn)]);
-	else
-		list_add(&pin->list, &pin_list->flushed);
 
 	if (reclaim)
 		bch2_journal_reclaim_fast(j);
@@ -555,11 +594,6 @@ static u64 journal_seq_to_flush(struct journal *j)
 		/* Try to keep the journal at most half full: */
 		nr_buckets = ja->nr / 2;
 
-		/* And include pre-reservations: */
-		nr_buckets += DIV_ROUND_UP(j->prereserved.reserved,
-					   (ca->mi.bucket_size << 6) -
-					   journal_entry_overhead(j));
-
 		nr_buckets = min(nr_buckets, ja->nr);
 
 		bucket_to_flush = (ja->cur_idx + nr_buckets) % ja->nr;
@@ -638,10 +672,7 @@ static int __bch2_journal_reclaim(struct journal *j, bool direct, bool kicked)
 			       msecs_to_jiffies(c->opts.journal_reclaim_delay)))
 			min_nr = 1;
 
-		if (j->prereserved.reserved * 4 > j->prereserved.remaining)
+		if (j->watermark != BCH_WATERMARK_stripe)
-			min_nr = 1;
-
-		if (fifo_free(&j->pin) <= 32)
 			min_nr = 1;
 
 		if (atomic_read(&c->btree_cache.dirty) * 2 > c->btree_cache.used)
@@ -652,8 +683,6 @@ static int __bch2_journal_reclaim(struct journal *j, bool direct, bool kicked)
 		trace_and_count(c, journal_reclaim_start, c,
 				direct, kicked,
 				min_nr, min_key_cache,
-				j->prereserved.reserved,
-				j->prereserved.remaining,
 				atomic_read(&c->btree_cache.dirty),
 				c->btree_cache.used,
 				atomic_long_read(&c->btree_key_cache.nr_dirty),
@@ -805,6 +834,7 @@ static int journal_flush_done(struct journal *j, u64 seq_to_flush,
 
 bool bch2_journal_flush_pins(struct journal *j, u64 seq_to_flush)
 {
+	/* time_stats this */
 	bool did_work = false;
 
 	if (!test_bit(JOURNAL_STARTED, &j->flags))

libbcachefs/journal_reclaim.h

@@ -47,17 +47,10 @@ static inline void bch2_journal_pin_add(struct journal *j, u64 seq,
 		bch2_journal_pin_set(j, seq, pin, flush_fn);
 }
 
-static inline void bch2_journal_pin_copy(struct journal *j,
+void bch2_journal_pin_copy(struct journal *,
-					 struct journal_entry_pin *dst,
+			   struct journal_entry_pin *,
-					 struct journal_entry_pin *src,
+			   struct journal_entry_pin *,
-					 journal_pin_flush_fn flush_fn)
+			   journal_pin_flush_fn);
-{
-	/* Guard against racing with journal_pin_drop(src): */
-	u64 seq = READ_ONCE(src->seq);
-
-	if (seq)
-		bch2_journal_pin_add(j, seq, dst, flush_fn);
-}
 
 static inline void bch2_journal_pin_update(struct journal *j, u64 seq,
 					   struct journal_entry_pin *pin,

libbcachefs/journal_types.h

@@ -76,14 +76,6 @@ struct journal_res {
 	u64			seq;
 };
 
-/*
- * For reserving space in the journal prior to getting a reservation on a
- * particular journal entry:
- */
-struct journal_preres {
-	unsigned		u64s;
-};
-
 union journal_res_state {
 	struct {
 		atomic64_t	counter;
@@ -104,22 +96,6 @@ union journal_res_state {
 	};
 };
 
-union journal_preres_state {
-	struct {
-		atomic64_t	counter;
-	};
-
-	struct {
-		u64		v;
-	};
-
-	struct {
-		u64		waiting:1,
-				reserved:31,
-				remaining:32;
-	};
-};
-
 /* bytes: */
 #define JOURNAL_ENTRY_SIZE_MIN		(64U << 10) /* 64k */
 #define JOURNAL_ENTRY_SIZE_MAX		(4U  << 20) /* 4M */
@@ -180,8 +156,6 @@ struct journal {
 	union journal_res_state reservations;
 	enum bch_watermark	watermark;
 
-	union journal_preres_state prereserved;
-
 	} __aligned(SMP_CACHE_BYTES);
 
 	unsigned long		flags;
@@ -288,15 +262,18 @@ struct journal {
 
 	unsigned long		last_flush_write;
 
-	u64			res_get_blocked_start;
 	u64			write_start_time;
 
 	u64			nr_flush_writes;
 	u64			nr_noflush_writes;
+	u64			entry_bytes_written;
+
+	u64			low_on_space_start;
+	u64			low_on_pin_start;
+	u64			max_in_flight_start;
 
 	struct bch2_time_stats	*flush_write_time;
 	struct bch2_time_stats	*noflush_write_time;
-	struct bch2_time_stats	*blocked_time;
 	struct bch2_time_stats	*flush_seq_time;
 
 #ifdef CONFIG_DEBUG_LOCK_ALLOC

libbcachefs/logged_ops.c

@@ -85,13 +85,13 @@ static int __bch2_logged_op_start(struct btree_trans *trans, struct bkey_i *k)
 
 int bch2_logged_op_start(struct btree_trans *trans, struct bkey_i *k)
 {
-	return commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+	return commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 			 __bch2_logged_op_start(trans, k));
 }
 
 void bch2_logged_op_finish(struct btree_trans *trans, struct bkey_i *k)
 {
-	int ret = commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+	int ret = commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 			    bch2_btree_delete(trans, BTREE_ID_logged_ops, k->k.p, 0));
 	/*
 	 * This needs to be a fatal error because we've left an unfinished

libbcachefs/lru.c

@@ -155,7 +155,7 @@ int bch2_check_lrus(struct bch_fs *c)
 	ret = bch2_trans_run(c,
 		for_each_btree_key_commit(trans, iter,
 				BTREE_ID_lru, POS_MIN, BTREE_ITER_PREFETCH, k,
-				NULL, NULL, BTREE_INSERT_NOFAIL|BTREE_INSERT_LAZY_RW,
+				NULL, NULL, BCH_TRANS_COMMIT_no_enospc|BCH_TRANS_COMMIT_lazy_rw,
 			bch2_check_lru_key(trans, &iter, k, &last_flushed_pos)));
 	if (ret)
 		bch_err_fn(c, ret);

libbcachefs/migrate.c

@@ -90,7 +90,7 @@ static int bch2_dev_usrdata_drop(struct bch_fs *c, unsigned dev_idx, int flags)
 
 		ret = for_each_btree_key_commit(trans, iter, id, POS_MIN,
 				BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS, k,
-				NULL, NULL, BTREE_INSERT_NOFAIL,
+				NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 			bch2_dev_usrdata_drop_key(trans, &iter, k, dev_idx, flags));
 		if (ret)
 			break;

libbcachefs/move.c

@@ -263,7 +263,7 @@ static int bch2_extent_drop_ptrs(struct btree_trans *trans,
 
 	return bch2_trans_relock(trans) ?:
 		bch2_trans_update(trans, iter, n, BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE) ?:
-		bch2_trans_commit(trans, NULL, NULL, BTREE_INSERT_NOFAIL);
+		bch2_trans_commit(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc);
 }
 
 int bch2_move_extent(struct moving_context *ctxt,

libbcachefs/movinggc.c

@@ -370,6 +370,7 @@ static int bch2_copygc_thread(void *arg)
 			if (min_member_capacity == U64_MAX)
 				min_member_capacity = 128 * 2048;
 
+			bch2_trans_unlock_long(ctxt.trans);
 			bch2_kthread_io_clock_wait(clock, last + (min_member_capacity >> 6),
 					MAX_SCHEDULE_TIMEOUT);
 		}

libbcachefs/rebalance.c

@@ -69,7 +69,7 @@ err:
 
 int bch2_set_rebalance_needs_scan(struct bch_fs *c, u64 inum)
 {
-	int ret = bch2_trans_do(c, NULL, NULL, BTREE_INSERT_NOFAIL|BTREE_INSERT_LAZY_RW,
+	int ret = bch2_trans_do(c, NULL, NULL, BCH_TRANS_COMMIT_no_enospc|BCH_TRANS_COMMIT_lazy_rw,
 			    __bch2_set_rebalance_needs_scan(trans, inum));
 	rebalance_wakeup(c);
 	return ret;
@@ -125,7 +125,7 @@ static int bch2_bkey_clear_needs_rebalance(struct btree_trans *trans,
 
 	extent_entry_drop(bkey_i_to_s(n),
 			  (void *) bch2_bkey_rebalance_opts(bkey_i_to_s_c(n)));
-	return bch2_trans_commit(trans, NULL, NULL, BTREE_INSERT_NOFAIL);
+	return bch2_trans_commit(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc);
 }
 
 static struct bkey_s_c next_rebalance_extent(struct btree_trans *trans,
@@ -273,7 +273,7 @@ static int do_rebalance_scan(struct moving_context *ctxt, u64 inum, u64 cookie)
 	r->state = BCH_REBALANCE_scanning;
 
 	ret = __bch2_move_data(ctxt, r->scan_start, r->scan_end, rebalance_pred, NULL) ?:
-		commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+		commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 			  bch2_clear_rebalance_needs_scan(trans, inum, cookie));
 
 	bch2_move_stats_exit(&r->scan_stats, trans->c);

libbcachefs/recovery.c

@@ -98,6 +98,11 @@ static int bch2_journal_replay_key(struct btree_trans *trans,
 	unsigned update_flags = BTREE_TRIGGER_NORUN;
 	int ret;
 
+	if (k->overwritten)
+		return 0;
+
+	trans->journal_res.seq = k->journal_seq;
+
 	/*
 	 * BTREE_UPDATE_KEY_CACHE_RECLAIM disables key cache lookup/update to
 	 * keep the key cache coherent with the underlying btree. Nothing
@@ -139,27 +144,14 @@ static int journal_sort_seq_cmp(const void *_l, const void *_r)
 static int bch2_journal_replay(struct bch_fs *c)
 {
 	struct journal_keys *keys = &c->journal_keys;
-	struct journal_key **keys_sorted, *k;
+	DARRAY(struct journal_key *) keys_sorted = { 0 };
+	struct journal_key **kp;
 	struct journal *j = &c->journal;
 	u64 start_seq	= c->journal_replay_seq_start;
 	u64 end_seq	= c->journal_replay_seq_start;
-	size_t i;
+	struct btree_trans *trans = bch2_trans_get(c);
 	int ret;
 
-	move_gap(keys->d, keys->nr, keys->size, keys->gap, keys->nr);
-	keys->gap = keys->nr;
-
-	keys_sorted = kvmalloc_array(keys->nr, sizeof(*keys_sorted), GFP_KERNEL);
-	if (!keys_sorted)
-		return -BCH_ERR_ENOMEM_journal_replay;
-
-	for (i = 0; i < keys->nr; i++)
-		keys_sorted[i] = &keys->d[i];
-
-	sort(keys_sorted, keys->nr,
-	     sizeof(keys_sorted[0]),
-	     journal_sort_seq_cmp, NULL);
-
 	if (keys->nr) {
 		ret = bch2_journal_log_msg(c, "Starting journal replay (%zu keys in entries %llu-%llu)",
 					   keys->nr, start_seq, end_seq);
@@ -167,27 +159,61 @@ static int bch2_journal_replay(struct bch_fs *c)
 			goto err;
 	}
 
-	for (i = 0; i < keys->nr; i++) {
+	/*
-		k = keys_sorted[i];
+	 * First, attempt to replay keys in sorted order. This is more
-
+	 * efficient, but some might fail if that would cause a journal
+	 * deadlock.
+	 */
+	for (size_t i = 0; i < keys->nr; i++) {
 		cond_resched();
 
+		struct journal_key *k = keys->d + i;
+
+		ret = commit_do(trans, NULL, NULL,
+				BCH_TRANS_COMMIT_no_enospc|
+				BCH_TRANS_COMMIT_journal_reclaim|
+				(!k->allocated ? BCH_TRANS_COMMIT_no_journal_res : 0),
+			     bch2_journal_replay_key(trans, k));
+		BUG_ON(!ret && !k->overwritten);
+		if (ret) {
+			ret = darray_push(&keys_sorted, k);
+			if (ret)
+				goto err;
+		}
+	}
+
+	/*
+	 * Now, replay any remaining keys in the order in which they appear in
+	 * the journal, unpinning those journal entries as we go:
+	 */
+	sort(keys_sorted.data, keys_sorted.nr,
+	     sizeof(keys_sorted.data[0]),
+	     journal_sort_seq_cmp, NULL);
+
+	darray_for_each(keys_sorted, kp) {
+		cond_resched();
+
+		struct journal_key *k = *kp;
+
 		replay_now_at(j, k->journal_seq);
 
-		ret = bch2_trans_do(c, NULL, NULL,
+		ret = commit_do(trans, NULL, NULL,
-				    BTREE_INSERT_LAZY_RW|
+				BCH_TRANS_COMMIT_no_enospc|
-				    BTREE_INSERT_NOFAIL|
+				(!k->allocated
-				    (!k->allocated
+				 ? BCH_TRANS_COMMIT_no_journal_res|BCH_WATERMARK_reclaim
-				     ? BTREE_INSERT_JOURNAL_REPLAY|BCH_WATERMARK_reclaim
+				 : 0),
-				     : 0),
 			     bch2_journal_replay_key(trans, k));
-		if (ret) {
+		bch_err_msg(c, ret, "while replaying key at btree %s level %u:",
-			bch_err(c, "journal replay: error while replaying key at btree %s level %u: %s",
+			    bch2_btree_id_str(k->btree_id), k->level);
-				bch2_btree_id_str(k->btree_id), k->level, bch2_err_str(ret));
+		if (ret)
 			goto err;
-		}
+
+		BUG_ON(!k->overwritten);
 	}
 
+	bch2_trans_put(trans);
+	trans = NULL;
+
 	replay_now_at(j, j->replay_journal_seq_end);
 	j->replay_journal_seq = 0;
 
@@ -198,10 +224,10 @@ static int bch2_journal_replay(struct bch_fs *c)
 	if (keys->nr && !ret)
 		bch2_journal_log_msg(c, "journal replay finished");
 err:
-	kvfree(keys_sorted);
+	if (trans)
-
+		bch2_trans_put(trans);
-	if (ret)
+	darray_exit(&keys_sorted);
-		bch_err_fn(c, ret);
+	bch_err_fn(c, ret);
 	return ret;
 }
 
@@ -468,7 +494,7 @@ err:
 noinline_for_stack
 static int bch2_fs_upgrade_for_subvolumes(struct bch_fs *c)
 {
-	int ret = bch2_trans_do(c, NULL, NULL, BTREE_INSERT_LAZY_RW,
+	int ret = bch2_trans_do(c, NULL, NULL, BCH_TRANS_COMMIT_lazy_rw,
 				__bch2_fs_upgrade_for_subvolumes(trans));
 	if (ret)
 		bch_err_fn(c, ret);
@@ -489,7 +515,19 @@ static int bch2_check_allocations(struct bch_fs *c)
 
 static int bch2_set_may_go_rw(struct bch_fs *c)
 {
+	struct journal_keys *keys = &c->journal_keys;
+
+	/*
+	 * After we go RW, the journal keys buffer can't be modified (except for
+	 * setting journal_key->overwritten: it will be accessed by multiple
+	 * threads
+	 */
+	move_gap(keys->d, keys->nr, keys->size, keys->gap, keys->nr);
+	keys->gap = keys->nr;
+
 	set_bit(BCH_FS_MAY_GO_RW, &c->flags);
+	if (keys->nr)
+		return bch2_fs_read_write_early(c);
 	return 0;
 }
 

libbcachefs/reflink.c

@@ -390,7 +390,7 @@ s64 bch2_remap_range(struct bch_fs *c,
 			inode_u.bi_size = new_i_size;
 			ret2  = bch2_inode_write(trans, &inode_iter, &inode_u) ?:
 				bch2_trans_commit(trans, NULL, NULL,
-						  BTREE_INSERT_NOFAIL);
+						  BCH_TRANS_COMMIT_no_enospc);
 		}
 
 		bch2_trans_iter_exit(trans, &inode_iter);

libbcachefs/sb-clean.c

@@ -376,7 +376,7 @@ void bch2_fs_mark_clean(struct bch_fs *c)
 
 	entry = sb_clean->start;
 	bch2_journal_super_entries_add_common(c, &entry, 0);
-	entry = bch2_btree_roots_to_journal_entries(c, entry, entry);
+	entry = bch2_btree_roots_to_journal_entries(c, entry, 0);
 	BUG_ON((void *) entry > vstruct_end(&sb_clean->field));
 
 	memset(entry, 0,

libbcachefs/sb-errors.c

@@ -70,7 +70,7 @@ static void bch2_sb_errors_to_text(struct printbuf *out, struct bch_sb *sb,
 		prt_tab(out);
 		prt_u64(out, BCH_SB_ERROR_ENTRY_NR(&e->entries[i]));
 		prt_tab(out);
-		bch2_prt_date_seconds(out, le64_to_cpu(e->entries[i].last_error_time));
+		bch2_prt_datetime(out, le64_to_cpu(e->entries[i].last_error_time));
 		prt_newline(out);
 	}
 }

libbcachefs/sb-members.c

@@ -230,7 +230,7 @@ static void member_to_text(struct printbuf *out,
 	prt_printf(out, "Last mount:");
 	prt_tab(out);
 	if (m.last_mount)
-		bch2_prt_date_seconds(out, le64_to_cpu(m.last_mount));
+		bch2_prt_datetime(out, le64_to_cpu(m.last_mount));
 	else
 		prt_printf(out, "(never)");
 	prt_newline(out);

libbcachefs/snapshot.c

@@ -590,7 +590,7 @@ int bch2_check_snapshot_trees(struct bch_fs *c)
 		for_each_btree_key_commit(trans, iter,
 			BTREE_ID_snapshot_trees, POS_MIN,
 			BTREE_ITER_PREFETCH, k,
-			NULL, NULL, BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
+			NULL, NULL, BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc,
 		check_snapshot_tree(trans, &iter, k)));
 
 	if (ret)
@@ -868,7 +868,7 @@ int bch2_check_snapshots(struct bch_fs *c)
 		for_each_btree_key_reverse_commit(trans, iter,
 			BTREE_ID_snapshots, POS_MAX,
 			BTREE_ITER_PREFETCH, k,
-			NULL, NULL, BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
+			NULL, NULL, BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc,
 		check_snapshot(trans, &iter, k)));
 	if (ret)
 		bch_err_fn(c, ret);
@@ -959,7 +959,7 @@ static int bch2_snapshot_node_delete(struct btree_trans *trans, u32 id)
 					parent_id, id))
 			goto err;
 
-		parent->v.children[i] = le32_to_cpu(child_id);
+		parent->v.children[i] = cpu_to_le32(child_id);
 
 		normalize_snapshot_child_pointers(&parent->v);
 	}
@@ -1449,12 +1449,12 @@ int bch2_delete_dead_snapshots(struct bch_fs *c)
 		ret = for_each_btree_key_commit(trans, iter,
 				id, POS_MIN,
 				BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS, k,
-				&res, NULL, BTREE_INSERT_NOFAIL,
+				&res, NULL, BCH_TRANS_COMMIT_no_enospc,
 			snapshot_delete_key(trans, &iter, k, &deleted, &equiv_seen, &last_pos)) ?:
 		      for_each_btree_key_commit(trans, iter,
 				id, POS_MIN,
 				BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS, k,
-				&res, NULL, BTREE_INSERT_NOFAIL,
+				&res, NULL, BCH_TRANS_COMMIT_no_enospc,
 			move_key_to_correct_snapshot(trans, &iter, k));
 
 		bch2_disk_reservation_put(c, &res);
@@ -1489,7 +1489,7 @@ int bch2_delete_dead_snapshots(struct bch_fs *c)
 	 */
 	ret = for_each_btree_key_commit(trans, iter, BTREE_ID_snapshots, POS_MIN,
 				  BTREE_ITER_INTENT, k,
-				  NULL, NULL, BTREE_INSERT_NOFAIL,
+				  NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 		bch2_fix_child_of_deleted_snapshot(trans, &iter, k, &deleted_interior));
 	if (ret)
 		goto err_create_lock;

libbcachefs/str_hash.h

@@ -15,6 +15,16 @@
 #include <crypto/hash.h>
 #include <crypto/sha2.h>
 
+typedef unsigned __bitwise bch_str_hash_flags_t;
+
+enum bch_str_hash_flags {
+	__BCH_HASH_SET_MUST_CREATE,
+	__BCH_HASH_SET_MUST_REPLACE,
+};
+
+#define BCH_HASH_SET_MUST_CREATE	(__force bch_str_hash_flags_t) BIT(__BCH_HASH_SET_MUST_CREATE)
+#define BCH_HASH_SET_MUST_REPLACE	(__force bch_str_hash_flags_t) BIT(__BCH_HASH_SET_MUST_REPLACE)
+
 static inline enum bch_str_hash_type
 bch2_str_hash_opt_to_type(struct bch_fs *c, enum bch_str_hash_opts opt)
 {
@@ -246,7 +256,7 @@ int bch2_hash_set_snapshot(struct btree_trans *trans,
 			   const struct bch_hash_info *info,
 			   subvol_inum inum, u32 snapshot,
 			   struct bkey_i *insert,
-			   int flags,
+			   bch_str_hash_flags_t str_hash_flags,
 			   int update_flags)
 {
 	struct btree_iter iter, slot = { NULL };
@@ -269,7 +279,7 @@ int bch2_hash_set_snapshot(struct btree_trans *trans,
 		}
 
 		if (!slot.path &&
-		    !(flags & BCH_HASH_SET_MUST_REPLACE))
+		    !(str_hash_flags & BCH_HASH_SET_MUST_REPLACE))
 			bch2_trans_copy_iter(&slot, &iter);
 
 		if (k.k->type != KEY_TYPE_hash_whiteout)
@@ -287,16 +297,16 @@ found:
 	found = true;
 not_found:
 
-	if (!found && (flags & BCH_HASH_SET_MUST_REPLACE)) {
+	if (!found && (str_hash_flags & BCH_HASH_SET_MUST_REPLACE)) {
 		ret = -BCH_ERR_ENOENT_str_hash_set_must_replace;
-	} else if (found && (flags & BCH_HASH_SET_MUST_CREATE)) {
+	} else if (found && (str_hash_flags & BCH_HASH_SET_MUST_CREATE)) {
 		ret = -EEXIST;
 	} else {
 		if (!found && slot.path)
 			swap(iter, slot);
 
 		insert->k.p = iter.pos;
-		ret = bch2_trans_update(trans, &iter, insert, 0);
+		ret = bch2_trans_update(trans, &iter, insert, update_flags);
 	}
 
 	goto out;
@@ -307,7 +317,8 @@ int bch2_hash_set(struct btree_trans *trans,
 		  const struct bch_hash_desc desc,
 		  const struct bch_hash_info *info,
 		  subvol_inum inum,
-		  struct bkey_i *insert, int flags)
+		  struct bkey_i *insert,
+		  bch_str_hash_flags_t str_hash_flags)
 {
 	u32 snapshot;
 	int ret;
@@ -319,7 +330,7 @@ int bch2_hash_set(struct btree_trans *trans,
 	insert->k.p.inode = inum.inum;
 
 	return bch2_hash_set_snapshot(trans, desc, info, inum,
-				      snapshot, insert, flags, 0);
+				      snapshot, insert, str_hash_flags, 0);
 }
 
 static __always_inline

libbcachefs/subvolume.c

@@ -89,7 +89,7 @@ int bch2_check_subvols(struct bch_fs *c)
 	ret = bch2_trans_run(c,
 		for_each_btree_key_commit(trans, iter,
 			BTREE_ID_subvolumes, POS_MIN, BTREE_ITER_PREFETCH, k,
-			NULL, NULL, BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
+			NULL, NULL, BCH_TRANS_COMMIT_lazy_rw|BCH_TRANS_COMMIT_no_enospc,
 		check_subvol(trans, &iter, k)));
 	if (ret)
 		bch_err_fn(c, ret);
@@ -219,7 +219,7 @@ static int bch2_subvolumes_reparent(struct btree_trans *trans, u32 subvolid_to_d
 				   BTREE_ITER_CACHED, &s)) ?:
 		for_each_btree_key_commit(trans, iter,
 				BTREE_ID_subvolumes, POS_MIN, BTREE_ITER_PREFETCH, k,
-				NULL, NULL, BTREE_INSERT_NOFAIL,
+				NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 			bch2_subvolume_reparent(trans, &iter, k,
 					subvolid_to_delete, le32_to_cpu(s.parent)));
 }
@@ -256,7 +256,7 @@ static int __bch2_subvolume_delete(struct btree_trans *trans, u32 subvolid)
 static int bch2_subvolume_delete(struct btree_trans *trans, u32 subvolid)
 {
 	return bch2_subvolumes_reparent(trans, subvolid) ?:
-		commit_do(trans, NULL, NULL, BTREE_INSERT_NOFAIL,
+		commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
 			  __bch2_subvolume_delete(trans, subvolid));
 }
 

libbcachefs/subvolume_types.h

@@ -20,7 +20,7 @@ struct snapshot_t {
 };
 
 struct snapshot_table {
-	struct snapshot_t	s[0];
+	DECLARE_FLEX_ARRAY(struct snapshot_t, s);
 };
 
 typedef struct {

libbcachefs/super-io.c

@@ -1183,7 +1183,7 @@ void bch2_sb_to_text(struct printbuf *out, struct bch_sb *sb,
 	prt_printf(out, "Created:");
 	prt_tab(out);
 	if (sb->time_base_lo)
-		bch2_prt_date_seconds(out, div_u64(le64_to_cpu(sb->time_base_lo), NSEC_PER_SEC));
+		bch2_prt_datetime(out, div_u64(le64_to_cpu(sb->time_base_lo), NSEC_PER_SEC));
 	else
 		prt_printf(out, "(not set)");
 	prt_newline(out);

libbcachefs/super.c

@@ -641,7 +641,9 @@ static int bch2_fs_online(struct bch_fs *c)
 	ret = kobject_add(&c->kobj, NULL, "%pU", c->sb.user_uuid.b) ?:
 	    kobject_add(&c->internal, &c->kobj, "internal") ?:
 	    kobject_add(&c->opts_dir, &c->kobj, "options") ?:
+#ifndef CONFIG_BCACHEFS_NO_LATENCY_ACCT
 	    kobject_add(&c->time_stats, &c->kobj, "time_stats") ?:
+#endif
 	    kobject_add(&c->counters_kobj, &c->kobj, "counters") ?:
 	    bch2_opts_create_sysfs_files(&c->opts_dir);
 	if (ret) {
@@ -750,7 +752,6 @@ static struct bch_fs *bch2_fs_alloc(struct bch_sb *sb, struct bch_opts opts)
 
 	c->journal.flush_write_time	= &c->times[BCH_TIME_journal_flush_write];
 	c->journal.noflush_write_time	= &c->times[BCH_TIME_journal_noflush_write];
-	c->journal.blocked_time		= &c->times[BCH_TIME_blocked_journal];
 	c->journal.flush_seq_time	= &c->times[BCH_TIME_journal_flush_seq];
 
 	bch2_fs_btree_cache_init_early(&c->btree_cache);

libbcachefs/trace.h

@@ -196,10 +196,9 @@ DEFINE_EVENT(bio, journal_write,
 TRACE_EVENT(journal_reclaim_start,
 	TP_PROTO(struct bch_fs *c, bool direct, bool kicked,
 		 u64 min_nr, u64 min_key_cache,
-		 u64 prereserved, u64 prereserved_total,
 		 u64 btree_cache_dirty, u64 btree_cache_total,
 		 u64 btree_key_cache_dirty, u64 btree_key_cache_total),
-	TP_ARGS(c, direct, kicked, min_nr, min_key_cache, prereserved, prereserved_total,
+	TP_ARGS(c, direct, kicked, min_nr, min_key_cache,
 		btree_cache_dirty, btree_cache_total,
 		btree_key_cache_dirty, btree_key_cache_total),
 
@@ -209,8 +208,6 @@ TRACE_EVENT(journal_reclaim_start,
 		__field(bool,		kicked			)
 		__field(u64,		min_nr			)
 		__field(u64,		min_key_cache		)
-		__field(u64,		prereserved		)
-		__field(u64,		prereserved_total	)
 		__field(u64,		btree_cache_dirty	)
 		__field(u64,		btree_cache_total	)
 		__field(u64,		btree_key_cache_dirty	)
@@ -223,22 +220,18 @@ TRACE_EVENT(journal_reclaim_start,
 		__entry->kicked			= kicked;
 		__entry->min_nr			= min_nr;
 		__entry->min_key_cache		= min_key_cache;
-		__entry->prereserved		= prereserved;
-		__entry->prereserved_total	= prereserved_total;
 		__entry->btree_cache_dirty	= btree_cache_dirty;
 		__entry->btree_cache_total	= btree_cache_total;
 		__entry->btree_key_cache_dirty	= btree_key_cache_dirty;
 		__entry->btree_key_cache_total	= btree_key_cache_total;
 	),
 
-	TP_printk("%d,%d direct %u kicked %u min %llu key cache %llu prereserved %llu/%llu btree cache %llu/%llu key cache %llu/%llu",
+	TP_printk("%d,%d direct %u kicked %u min %llu key cache %llu btree cache %llu/%llu key cache %llu/%llu",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  __entry->direct,
 		  __entry->kicked,
 		  __entry->min_nr,
 		  __entry->min_key_cache,
-		  __entry->prereserved,
-		  __entry->prereserved_total,
 		  __entry->btree_cache_dirty,
 		  __entry->btree_cache_total,
 		  __entry->btree_key_cache_dirty,

libbcachefs/util.c

@@ -315,6 +315,57 @@ int bch2_prt_task_backtrace(struct printbuf *out, struct task_struct *task)
 	return ret;
 }
 
+#ifndef __KERNEL__
+#include <time.h>
+void bch2_prt_datetime(struct printbuf *out, time64_t sec)
+{
+	time_t t = sec;
+	char buf[64];
+	ctime_r(&t, buf);
+	prt_str(out, buf);
+}
+#else
+void bch2_prt_datetime(struct printbuf *out, time64_t sec)
+{
+	char buf[64];
+	snprintf(buf, sizeof(buf), "%ptT", &sec);
+	prt_u64(out, sec);
+}
+#endif
+
+static const struct time_unit {
+	const char	*name;
+	u64		nsecs;
+} time_units[] = {
+	{ "ns",		1		 },
+	{ "us",		NSEC_PER_USEC	 },
+	{ "ms",		NSEC_PER_MSEC	 },
+	{ "s",		NSEC_PER_SEC	 },
+	{ "m",          (u64) NSEC_PER_SEC * 60},
+	{ "h",          (u64) NSEC_PER_SEC * 3600},
+	{ "eon",        U64_MAX          },
+};
+
+static const struct time_unit *pick_time_units(u64 ns)
+{
+	const struct time_unit *u;
+
+	for (u = time_units;
+	     u + 1 < time_units + ARRAY_SIZE(time_units) &&
+	     ns >= u[1].nsecs << 1;
+	     u++)
+		;
+
+	return u;
+}
+
+void bch2_pr_time_units(struct printbuf *out, u64 ns)
+{
+	const struct time_unit *u = pick_time_units(ns);
+
+	prt_printf(out, "%llu %s", div_u64(ns, u->nsecs), u->name);
+}
+
 /* time stats: */
 
 #ifndef CONFIG_BCACHEFS_NO_LATENCY_ACCT
@@ -359,6 +410,7 @@ static inline void bch2_time_stats_update_one(struct bch2_time_stats *stats,
 		mean_and_variance_weighted_update(&stats->duration_stats_weighted, duration);
 		stats->max_duration = max(stats->max_duration, duration);
 		stats->min_duration = min(stats->min_duration, duration);
+		stats->total_duration += duration;
 		bch2_quantiles_update(&stats->quantiles, duration);
 	}
 
@@ -372,22 +424,26 @@ static inline void bch2_time_stats_update_one(struct bch2_time_stats *stats,
 	}
 }
 
-static noinline void bch2_time_stats_clear_buffer(struct bch2_time_stats *stats,
+static void __bch2_time_stats_clear_buffer(struct bch2_time_stats *stats,
-						  struct bch2_time_stat_buffer *b)
+					   struct bch2_time_stat_buffer *b)
 {
-	struct bch2_time_stat_buffer_entry *i;
+	for (struct bch2_time_stat_buffer_entry *i = b->entries;
-	unsigned long flags;
-
-	spin_lock_irqsave(&stats->lock, flags);
-	for (i = b->entries;
 	     i < b->entries + ARRAY_SIZE(b->entries);
 	     i++)
 		bch2_time_stats_update_one(stats, i->start, i->end);
-	spin_unlock_irqrestore(&stats->lock, flags);
-
 	b->nr = 0;
 }
 
+static noinline void bch2_time_stats_clear_buffer(struct bch2_time_stats *stats,
+						  struct bch2_time_stat_buffer *b)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&stats->lock, flags);
+	__bch2_time_stats_clear_buffer(stats, b);
+	spin_unlock_irqrestore(&stats->lock, flags);
+}
+
 void __bch2_time_stats_update(struct bch2_time_stats *stats, u64 start, u64 end)
 {
 	unsigned long flags;
@@ -423,40 +479,6 @@ void __bch2_time_stats_update(struct bch2_time_stats *stats, u64 start, u64 end)
 		preempt_enable();
 	}
 }
-#endif
-
-static const struct time_unit {
-	const char	*name;
-	u64		nsecs;
-} time_units[] = {
-	{ "ns",		1		 },
-	{ "us",		NSEC_PER_USEC	 },
-	{ "ms",		NSEC_PER_MSEC	 },
-	{ "s",		NSEC_PER_SEC	 },
-	{ "m",          (u64) NSEC_PER_SEC * 60},
-	{ "h",          (u64) NSEC_PER_SEC * 3600},
-	{ "eon",        U64_MAX          },
-};
-
-static const struct time_unit *pick_time_units(u64 ns)
-{
-	const struct time_unit *u;
-
-	for (u = time_units;
-	     u + 1 < time_units + ARRAY_SIZE(time_units) &&
-	     ns >= u[1].nsecs << 1;
-	     u++)
-		;
-
-	return u;
-}
-
-void bch2_pr_time_units(struct printbuf *out, u64 ns)
-{
-	const struct time_unit *u = pick_time_units(ns);
-
-	prt_printf(out, "%llu %s", div_u64(ns, u->nsecs), u->name);
-}
 
 static void bch2_pr_time_units_aligned(struct printbuf *out, u64 ns)
 {
@@ -467,26 +489,6 @@ static void bch2_pr_time_units_aligned(struct printbuf *out, u64 ns)
 	prt_printf(out, "%s", u->name);
 }
 
-#ifndef __KERNEL__
-#include <time.h>
-void bch2_prt_date_seconds(struct printbuf *out, time64_t sec)
-{
-	time_t t = sec;
-	char buf[64];
-	ctime_r(&t, buf);
-	prt_str(out, buf);
-}
-#else
-void bch2_prt_date_seconds(struct printbuf *out, time64_t sec)
-{
-	char buf[64];
-	snprintf(buf, sizeof(buf), "%ptT", &sec);
-	prt_u64(out, sec);
-}
-#endif
-
-#define TABSTOP_SIZE 12
-
 static inline void pr_name_and_units(struct printbuf *out, const char *name, u64 ns)
 {
 	prt_str(out, name);
@@ -495,12 +497,24 @@ static inline void pr_name_and_units(struct printbuf *out, const char *name, u64
 	prt_newline(out);
 }
 
+#define TABSTOP_SIZE 12
+
 void bch2_time_stats_to_text(struct printbuf *out, struct bch2_time_stats *stats)
 {
 	const struct time_unit *u;
 	s64 f_mean = 0, d_mean = 0;
 	u64 q, last_q = 0, f_stddev = 0, d_stddev = 0;
 	int i;
+
+	if (stats->buffer) {
+		int cpu;
+
+		spin_lock_irq(&stats->lock);
+		for_each_possible_cpu(cpu)
+			__bch2_time_stats_clear_buffer(stats, per_cpu_ptr(stats->buffer, cpu));
+		spin_unlock_irq(&stats->lock);
+	}
+
 	/*
 	 * avoid divide by zero
 	 */
@@ -546,6 +560,7 @@ void bch2_time_stats_to_text(struct printbuf *out, struct bch2_time_stats *stats
 
 	pr_name_and_units(out, "min:", stats->min_duration);
 	pr_name_and_units(out, "max:", stats->max_duration);
+	pr_name_and_units(out, "total:", stats->total_duration);
 
 	prt_printf(out, "mean:");
 	prt_tab(out);
@@ -603,6 +618,9 @@ void bch2_time_stats_to_text(struct printbuf *out, struct bch2_time_stats *stats
 		last_q = q;
 	}
 }
+#else
+void bch2_time_stats_to_text(struct printbuf *out, struct bch2_time_stats *stats) {}
+#endif
 
 void bch2_time_stats_exit(struct bch2_time_stats *stats)
 {

libbcachefs/util.h

@@ -244,7 +244,7 @@ do {									\
 #define prt_bitflags(...)		bch2_prt_bitflags(__VA_ARGS__)
 
 void bch2_pr_time_units(struct printbuf *, u64);
-void bch2_prt_date_seconds(struct printbuf *, time64_t);
+void bch2_prt_datetime(struct printbuf *, time64_t);
 
 #ifdef __KERNEL__
 static inline void uuid_unparse_lower(u8 *uuid, char *out)
@@ -372,8 +372,9 @@ struct bch2_time_stat_buffer {
 struct bch2_time_stats {
 	spinlock_t	lock;
 	/* all fields are in nanoseconds */
-	u64		max_duration;
 	u64             min_duration;
+	u64		max_duration;
+	u64		total_duration;
 	u64             max_freq;
 	u64             min_freq;
 	u64		last_event;
@@ -388,15 +389,39 @@ struct bch2_time_stats {
 
 #ifndef CONFIG_BCACHEFS_NO_LATENCY_ACCT
 void __bch2_time_stats_update(struct bch2_time_stats *stats, u64, u64);
-#else
-static inline void __bch2_time_stats_update(struct bch2_time_stats *stats, u64 start, u64 end) {}
-#endif
 
 static inline void bch2_time_stats_update(struct bch2_time_stats *stats, u64 start)
 {
 	__bch2_time_stats_update(stats, start, local_clock());
 }
 
+static inline bool track_event_change(struct bch2_time_stats *stats,
+				      u64 *start, bool v)
+{
+	if (v != !!*start) {
+		if (!v) {
+			bch2_time_stats_update(stats, *start);
+			*start = 0;
+		} else {
+			*start = local_clock() ?: 1;
+			return true;
+		}
+	}
+
+	return false;
+}
+#else
+static inline void __bch2_time_stats_update(struct bch2_time_stats *stats, u64 start, u64 end) {}
+static inline void bch2_time_stats_update(struct bch2_time_stats *stats, u64 start) {}
+static inline bool track_event_change(struct bch2_time_stats *stats,
+				      u64 *start, bool v)
+{
+	bool ret = v && !*start;
+	*start = v;
+	return ret;
+}
+#endif
+
 void bch2_time_stats_to_text(struct printbuf *, struct bch2_time_stats *);
 
 void bch2_time_stats_exit(struct bch2_time_stats *);