Update bcachefs sources to c887148ebf99 thread_with_file: add f_ops.flush

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

.bcachefs_revision

@@ -1 +1 @@
-9a555a741e807275c320807babd3f42efb8fee90
+c887148ebf9989ce8bdf6f814d4342ba5bf465fa

libbcachefs/alloc_background.c

@@ -29,6 +29,8 @@
 #include <linux/sched/task.h>
 #include <linux/sort.h>
 
+static void bch2_discard_one_bucket_fast(struct bch_fs *c, struct bpos bucket);
+
 /* Persistent alloc info: */
 
 static const unsigned BCH_ALLOC_V1_FIELD_BYTES[] = {
@@ -866,14 +868,14 @@ int bch2_trigger_alloc(struct btree_trans *trans,
 #define statechange(expr)		!eval_state(old_a, expr) && eval_state(new_a, expr)
 #define bucket_flushed(a)		(!a->journal_seq || a->journal_seq <= c->journal.flushed_seq_ondisk)
 
-		if (statechange(a->data_type == BCH_DATA_free &&
+		if (statechange(a->data_type == BCH_DATA_free) &&
-				bucket_flushed(a)))
+		    bucket_flushed(new_a))
 			closure_wake_up(&c->freelist_wait);
 
-		if (statechange(a->data_type == BCH_DATA_need_discard &&
+		if (statechange(a->data_type == BCH_DATA_need_discard) &&
-				bucket_flushed(a)) &&
+		    !bch2_bucket_is_open(c, new.k->p.inode, new.k->p.offset) &&
-		    !bch2_bucket_is_open(c, new.k->p.inode, new.k->p.offset))
+		    bucket_flushed(new_a))
-			bch2_do_discards(c);
+			bch2_discard_one_bucket_fast(c, new.k->p);
 
 		if (statechange(a->data_type == BCH_DATA_cached) &&
 		    !bch2_bucket_is_open(c, new.k->p.inode, new.k->p.offset) &&
@@ -1609,6 +1611,36 @@ int bch2_check_alloc_to_lru_refs(struct bch_fs *c)
 	return ret;
 }
 
+static int discard_in_flight_add(struct bch_fs *c, struct bpos bucket)
+{
+	int ret;
+
+	mutex_lock(&c->discard_buckets_in_flight_lock);
+	darray_for_each(c->discard_buckets_in_flight, i)
+		if (bkey_eq(*i, bucket)) {
+			ret = -EEXIST;
+			goto out;
+		}
+
+	ret = darray_push(&c->discard_buckets_in_flight, bucket);
+out:
+	mutex_unlock(&c->discard_buckets_in_flight_lock);
+	return ret;
+}
+
+static void discard_in_flight_remove(struct bch_fs *c, struct bpos bucket)
+{
+	mutex_lock(&c->discard_buckets_in_flight_lock);
+	darray_for_each(c->discard_buckets_in_flight, i)
+		if (bkey_eq(*i, bucket)) {
+			darray_remove_item(&c->discard_buckets_in_flight, i);
+			goto found;
+		}
+	BUG();
+found:
+	mutex_unlock(&c->discard_buckets_in_flight_lock);
+}
+
 struct discard_buckets_state {
 	u64		seen;
 	u64		open;
@@ -1647,6 +1679,7 @@ static int bch2_discard_one_bucket(struct btree_trans *trans,
 	struct bch_dev *ca;
 	struct bkey_i_alloc_v4 *a;
 	struct printbuf buf = PRINTBUF;
+	bool discard_locked = false;
 	int ret = 0;
 
 	ca = bch_dev_bkey_exists(c, pos.inode);
@@ -1714,6 +1747,11 @@ static int bch2_discard_one_bucket(struct btree_trans *trans,
 		goto out;
 	}
 
+	if (discard_in_flight_add(c, SPOS(iter.pos.inode, iter.pos.offset, true)))
+		goto out;
+
+	discard_locked = true;
+
 	if (!bkey_eq(*discard_pos_done, iter.pos) &&
 	    ca->mi.discard && !c->opts.nochanges) {
 		/*
@@ -1745,6 +1783,8 @@ write:
 	count_event(c, bucket_discard);
 	s->discarded++;
 out:
+	if (discard_locked)
+		discard_in_flight_remove(c, iter.pos);
 	s->seen++;
 	bch2_trans_iter_exit(trans, &iter);
 	percpu_ref_put(&ca->io_ref);
@@ -1784,6 +1824,92 @@ void bch2_do_discards(struct bch_fs *c)
 		bch2_write_ref_put(c, BCH_WRITE_REF_discard);
 }
 
+static int bch2_clear_bucket_needs_discard(struct btree_trans *trans, struct bpos bucket)
+{
+	struct btree_iter iter;
+	bch2_trans_iter_init(trans, &iter, BTREE_ID_alloc, bucket, BTREE_ITER_INTENT);
+	struct bkey_s_c k = bch2_btree_iter_peek_slot(&iter);
+	int ret = bkey_err(k);
+	if (ret)
+		goto err;
+
+	struct bkey_i_alloc_v4 *a = bch2_alloc_to_v4_mut(trans, k);
+	ret = PTR_ERR_OR_ZERO(a);
+	if (ret)
+		goto err;
+
+	SET_BCH_ALLOC_V4_NEED_DISCARD(&a->v, false);
+	a->v.data_type = alloc_data_type(a->v, a->v.data_type);
+
+	ret = bch2_trans_update(trans, &iter, &a->k_i, 0);
+err:
+	bch2_trans_iter_exit(trans, &iter);
+	return ret;
+}
+
+static void bch2_do_discards_fast_work(struct work_struct *work)
+{
+	struct bch_fs *c = container_of(work, struct bch_fs, discard_fast_work);
+
+	while (1) {
+		bool got_bucket = false;
+		struct bpos bucket;
+		struct bch_dev *ca;
+
+		mutex_lock(&c->discard_buckets_in_flight_lock);
+		darray_for_each(c->discard_buckets_in_flight, i) {
+			if (i->snapshot)
+				continue;
+
+			ca = bch_dev_bkey_exists(c, i->inode);
+
+			if (!percpu_ref_tryget(&ca->io_ref)) {
+				darray_remove_item(&c->discard_buckets_in_flight, i);
+				continue;
+			}
+
+			got_bucket = true;
+			bucket = *i;
+			i->snapshot = true;
+			break;
+		}
+		mutex_unlock(&c->discard_buckets_in_flight_lock);
+
+		if (!got_bucket)
+			break;
+
+		blkdev_issue_discard(ca->disk_sb.bdev,
+				     bucket.offset * ca->mi.bucket_size,
+				     ca->mi.bucket_size,
+				     GFP_KERNEL);
+
+		int ret = bch2_trans_do(c, NULL, NULL,
+					BCH_WATERMARK_btree|
+					BCH_TRANS_COMMIT_no_enospc,
+					bch2_clear_bucket_needs_discard(trans, bucket));
+		bch_err_fn(c, ret);
+
+		percpu_ref_put(&ca->io_ref);
+		discard_in_flight_remove(c, bucket);
+
+		if (ret)
+			break;
+	}
+
+	bch2_write_ref_put(c, BCH_WRITE_REF_discard_fast);
+}
+
+static void bch2_discard_one_bucket_fast(struct bch_fs *c, struct bpos bucket)
+{
+	struct bch_dev *ca = bch_dev_bkey_exists(c, bucket.inode);
+
+	if (!percpu_ref_is_dying(&ca->io_ref) &&
+	    !discard_in_flight_add(c, bucket) &&
+	    bch2_write_ref_tryget(c, BCH_WRITE_REF_discard_fast) &&
+	    !queue_work(c->write_ref_wq, &c->discard_fast_work))
+		bch2_write_ref_put(c, BCH_WRITE_REF_discard_fast);
+}
+
 static int invalidate_one_bucket(struct btree_trans *trans,
 				 struct btree_iter *lru_iter,
 				 struct bkey_s_c lru_k,
@@ -2215,9 +2341,16 @@ void bch2_dev_allocator_add(struct bch_fs *c, struct bch_dev *ca)
 			set_bit(ca->dev_idx, c->rw_devs[i].d);
 }
 
+void bch2_fs_allocator_background_exit(struct bch_fs *c)
+{
+	darray_exit(&c->discard_buckets_in_flight);
+}
+
 void bch2_fs_allocator_background_init(struct bch_fs *c)
 {
 	spin_lock_init(&c->freelist_lock);
+	mutex_init(&c->discard_buckets_in_flight_lock);
 	INIT_WORK(&c->discard_work, bch2_do_discards_work);
+	INIT_WORK(&c->discard_fast_work, bch2_do_discards_fast_work);
 	INIT_WORK(&c->invalidate_work, bch2_do_invalidates_work);
 }

libbcachefs/alloc_background.h

@@ -269,6 +269,7 @@ u64 bch2_min_rw_member_capacity(struct bch_fs *);
 void bch2_dev_allocator_remove(struct bch_fs *, struct bch_dev *);
 void bch2_dev_allocator_add(struct bch_fs *, struct bch_dev *);
 
+void bch2_fs_allocator_background_exit(struct bch_fs *);
 void bch2_fs_allocator_background_init(struct bch_fs *);
 
 #endif /* _BCACHEFS_ALLOC_BACKGROUND_H */

libbcachefs/bcachefs.h

@@ -708,6 +708,7 @@ struct btree_trans_buf {
 	x(reflink)							\
 	x(fallocate)							\
 	x(discard)							\
+	x(discard_fast)							\
 	x(invalidate)							\
 	x(delete_dead_snapshots)					\
 	x(snapshot_delete_pagecache)					\
@@ -943,8 +944,11 @@ struct bch_fs {
 	unsigned		write_points_nr;
 
 	struct buckets_waiting_for_journal buckets_waiting_for_journal;
-	struct work_struct	discard_work;
 	struct work_struct	invalidate_work;
+	struct work_struct	discard_work;
+	struct mutex		discard_buckets_in_flight_lock;
+	DARRAY(struct bpos)	discard_buckets_in_flight;
+	struct work_struct	discard_fast_work;
 
 	/* GARBAGE COLLECTION */
 	struct task_struct	*gc_thread;

libbcachefs/bcachefs_format.h

@@ -189,7 +189,11 @@ struct bversion {
 	__u32		hi;
 	__u64		lo;
 #endif
-} __packed __aligned(4);
+} __packed
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+__aligned(4)
+#endif
+;
 
 struct bkey {
 	/* Size of combined key and value, in u64s */

libbcachefs/btree_locking.c

@@ -747,7 +747,8 @@ void bch2_trans_downgrade(struct btree_trans *trans)
 		return;
 
 	trans_for_each_path(trans, path, i)
-		bch2_btree_path_downgrade(trans, path);
+		if (path->ref)
+			bch2_btree_path_downgrade(trans, path);
 }
 
 int bch2_trans_relock(struct btree_trans *trans)

libbcachefs/btree_write_buffer.c

@@ -590,7 +590,9 @@ static int bch2_journal_keys_to_write_buffer(struct bch_fs *c, struct journal_bu
 		entry->type = BCH_JSET_ENTRY_btree_keys;
 	}
 
+	spin_lock(&c->journal.lock);
 	buf->need_flush_to_write_buffer = false;
+	spin_unlock(&c->journal.lock);
 out:
 	bch2_journal_keys_to_write_buffer_end(c, &dst);
 	return ret;

libbcachefs/chardev.c

@@ -155,14 +155,28 @@ static void bch2_fsck_thread_exit(struct thread_with_stdio *_thr)
 	kfree(thr);
 }
 
-static void bch2_fsck_offline_thread_fn(struct thread_with_stdio *stdio)
+static int bch2_fsck_offline_thread_fn(struct thread_with_stdio *stdio)
 {
 	struct fsck_thread *thr = container_of(stdio, struct fsck_thread, thr);
 	struct bch_fs *c = bch2_fs_open(thr->devs, thr->nr_devs, thr->opts);
 
-	thr->thr.thr.ret = PTR_ERR_OR_ZERO(c);
+	if (IS_ERR(c))
-	if (!thr->thr.thr.ret)
+		return PTR_ERR(c);
-		bch2_fs_stop(c);
+
+	int ret = 0;
+	if (test_bit(BCH_FS_errors_fixed, &c->flags))
+		ret |= 1;
+	if (test_bit(BCH_FS_error, &c->flags))
+		ret |= 4;
+
+	bch2_fs_stop(c);
+
+	if (ret & 1)
+		stdio_redirect_printf(&stdio->stdio, false, "%s: errors fixed\n", c->name);
+	if (ret & 4)
+		stdio_redirect_printf(&stdio->stdio, false, "%s: still has errors\n", c->name);
+
+	return ret;
 }
 
 static const struct thread_with_stdio_ops bch2_offline_fsck_ops = {
@@ -763,7 +777,7 @@ static long bch2_ioctl_disk_resize_journal(struct bch_fs *c,
 	return ret;
 }
 
-static void bch2_fsck_online_thread_fn(struct thread_with_stdio *stdio)
+static int bch2_fsck_online_thread_fn(struct thread_with_stdio *stdio)
 {
 	struct fsck_thread *thr = container_of(stdio, struct fsck_thread, thr);
 	struct bch_fs *c = thr->c;
@@ -795,6 +809,7 @@ static void bch2_fsck_online_thread_fn(struct thread_with_stdio *stdio)
 
 	up(&c->online_fsck_mutex);
 	bch2_ro_ref_put(c);
+	return ret;
 }
 
 static const struct thread_with_stdio_ops bch2_online_fsck_ops = {

libbcachefs/fsck.c

@@ -342,6 +342,27 @@ static int remove_backpointer(struct btree_trans *trans,
 	return ret;
 }
 
+static int reattach_subvol(struct btree_trans *trans, struct bkey_s_c_subvolume s)
+{
+	struct bch_fs *c = trans->c;
+
+	struct bch_inode_unpacked inode;
+	int ret = bch2_inode_find_by_inum_trans(trans,
+				(subvol_inum) { s.k->p.offset, le64_to_cpu(s.v->inode) },
+				&inode);
+	if (ret)
+		return ret;
+
+	ret = remove_backpointer(trans, &inode);
+	bch_err_msg(c, ret, "removing dirent");
+	if (ret)
+		return ret;
+
+	ret = reattach_inode(trans, &inode, le32_to_cpu(s.v->snapshot));
+	bch_err_msg(c, ret, "reattaching inode %llu", inode.bi_inum);
+	return ret;
+}
+
 struct snapshots_seen_entry {
 	u32				id;
 	u32				equiv;
@@ -2111,6 +2132,107 @@ int bch2_check_root(struct bch_fs *c)
 	return ret;
 }
 
+typedef DARRAY(u32) darray_u32;
+
+static bool darray_u32_has(darray_u32 *d, u32 v)
+{
+	darray_for_each(*d, i)
+		if (*i == v)
+			return true;
+	return false;
+}
+
+/*
+ * We've checked that inode backpointers point to valid dirents; here, it's
+ * sufficient to check that the subvolume root has a dirent:
+ */
+static int subvol_has_dirent(struct btree_trans *trans, struct bkey_s_c_subvolume s)
+{
+	struct bch_inode_unpacked inode;
+	int ret = bch2_inode_find_by_inum_trans(trans,
+				(subvol_inum) { s.k->p.offset, le64_to_cpu(s.v->inode) },
+				&inode);
+	if (ret)
+		return ret;
+
+	return inode.bi_dir != 0;
+}
+
+static int check_subvol_path(struct btree_trans *trans, struct btree_iter *iter, struct bkey_s_c k)
+{
+	struct bch_fs *c = trans->c;
+	struct btree_iter parent_iter = {};
+	darray_u32 subvol_path = {};
+	struct printbuf buf = PRINTBUF;
+	int ret = 0;
+
+	if (k.k->type != KEY_TYPE_subvolume)
+		return 0;
+
+	while (k.k->p.offset != BCACHEFS_ROOT_SUBVOL) {
+		ret = darray_push(&subvol_path, k.k->p.offset);
+		if (ret)
+			goto err;
+
+		struct bkey_s_c_subvolume s = bkey_s_c_to_subvolume(k);
+
+		ret = subvol_has_dirent(trans, s);
+		if (ret < 0)
+			break;
+
+		if (fsck_err_on(!ret,
+				c, subvol_unreachable,
+				"unreachable subvolume %s",
+				(bch2_bkey_val_to_text(&buf, c, s.s_c),
+				 buf.buf))) {
+			ret = reattach_subvol(trans, s);
+			break;
+		}
+
+		u32 parent = le32_to_cpu(s.v->fs_path_parent);
+
+		if (darray_u32_has(&subvol_path, parent)) {
+			if (fsck_err(c, subvol_loop, "subvolume loop"))
+				ret = reattach_subvol(trans, s);
+			break;
+		}
+
+		bch2_trans_iter_exit(trans, &parent_iter);
+		bch2_trans_iter_init(trans, &parent_iter,
+				     BTREE_ID_subvolumes, POS(0, parent), 0);
+		k = bch2_btree_iter_peek_slot(&parent_iter);
+		ret = bkey_err(k);
+		if (ret)
+			goto err;
+
+		if (fsck_err_on(k.k->type != KEY_TYPE_subvolume,
+				c, subvol_unreachable,
+				"unreachable subvolume %s",
+				(bch2_bkey_val_to_text(&buf, c, s.s_c),
+				 buf.buf))) {
+			ret = reattach_subvol(trans, s);
+			break;
+		}
+	}
+fsck_err:
+err:
+	printbuf_exit(&buf);
+	darray_exit(&subvol_path);
+	bch2_trans_iter_exit(trans, &parent_iter);
+	return ret;
+}
+
+int bch2_check_subvolume_structure(struct bch_fs *c)
+{
+	int ret = bch2_trans_run(c,
+		for_each_btree_key_commit(trans, iter,
+				BTREE_ID_subvolumes, POS_MIN, BTREE_ITER_PREFETCH, k,
+				NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
+			check_subvol_path(trans, &iter, k)));
+	bch_err_fn(c, ret);
+	return ret;
+}
+
 struct pathbuf_entry {
 	u64	inum;
 	u32	snapshot;
@@ -2127,22 +2249,9 @@ static bool path_is_dup(pathbuf *p, u64 inum, u32 snapshot)
 	return false;
 }
 
-static int path_down(struct bch_fs *c, pathbuf *p,
-		     u64 inum, u32 snapshot)
-{
-	int ret = darray_push(p, ((struct pathbuf_entry) {
-		.inum		= inum,
-		.snapshot	= snapshot,
-	}));
-
-	if (ret)
-		bch_err(c, "fsck: error allocating memory for pathbuf, size %zu",
-			p->size);
-	return ret;
-}
-
 /*
- * Check that a given inode is reachable from the root:
+ * Check that a given inode is reachable from its subvolume root - we already
+ * verified subvolume connectivity:
  *
  * XXX: we should also be verifying that inodes are in the right subvolumes
  */
@@ -2159,8 +2268,7 @@ static int check_path(struct btree_trans *trans, pathbuf *p, struct bkey_s_c ino
 
 	BUG_ON(bch2_inode_unpack(inode_k, &inode));
 
-	while (!(inode.bi_inum == BCACHEFS_ROOT_INO &&
+	while (!inode.bi_subvol) {
-		 inode.bi_subvol == BCACHEFS_ROOT_SUBVOL)) {
 		struct btree_iter dirent_iter;
 		struct bkey_s_c_dirent d;
 		u32 parent_snapshot = snapshot;
@@ -2191,11 +2299,12 @@ static int check_path(struct btree_trans *trans, pathbuf *p, struct bkey_s_c ino
 		if (!S_ISDIR(inode.bi_mode))
 			break;
 
-		ret = path_down(c, p, inode.bi_inum, snapshot);
+		ret = darray_push(p, ((struct pathbuf_entry) {
-		if (ret) {
+			.inum		= inode.bi_inum,
-			bch_err(c, "memory allocation failure");
+			.snapshot	= snapshot,
+		}));
+		if (ret)
 			return ret;
-		}
 
 		snapshot = parent_snapshot;
 
@@ -2222,18 +2331,15 @@ static int check_path(struct btree_trans *trans, pathbuf *p, struct bkey_s_c ino
 				pr_err("%llu:%u", i->inum, i->snapshot);
 			pr_err("%llu:%u", inode.bi_inum, snapshot);
 
-			if (!fsck_err(c, dir_loop, "directory structure loop"))
+			if (fsck_err(c, dir_loop, "directory structure loop")) {
-				return 0;
+				ret = remove_backpointer(trans, &inode);
-
-			ret = remove_backpointer(trans, &inode);
-			if (ret && !bch2_err_matches(ret, BCH_ERR_transaction_restart))
 				bch_err_msg(c, ret, "removing dirent");
-			if (ret)
+				if (ret)
-				break;
+					break;
 
-			ret = reattach_inode(trans, &inode, snapshot);
+				ret = reattach_inode(trans, &inode, snapshot);
-			if (ret && !bch2_err_matches(ret, BCH_ERR_transaction_restart))
 				bch_err_msg(c, ret, "reattaching inode %llu", inode.bi_inum);
+			}
 			break;
 		}
 	}

libbcachefs/fsck.h

@@ -8,6 +8,7 @@ int bch2_check_indirect_extents(struct bch_fs *);
 int bch2_check_dirents(struct bch_fs *);
 int bch2_check_xattrs(struct bch_fs *);
 int bch2_check_root(struct bch_fs *);
+int bch2_check_subvolume_structure(struct bch_fs *);
 int bch2_check_directory_structure(struct bch_fs *);
 int bch2_check_nlinks(struct bch_fs *);
 int bch2_fix_reflink_p(struct bch_fs *);

libbcachefs/io_write.c

@@ -530,7 +530,8 @@ static void __bch2_write_index(struct bch_write_op *op)
 
 			bch_err_inum_offset_ratelimited(c,
 				insert->k.p.inode, insert->k.p.offset << 9,
-				"write error while doing btree update: %s",
+				"%s write error while doing btree update: %s",
+				op->flags & BCH_WRITE_MOVE ? "move" : "user",
 				bch2_err_str(ret));
 		}
 
@@ -1067,7 +1068,8 @@ do_write:
 	*_dst = dst;
 	return more;
 csum_err:
-	bch_err(c, "error verifying existing checksum while rewriting existing data (memory corruption?)");
+	bch_err(c, "%s writ error: error verifying existing checksum while rewriting existing data (memory corruption?)",
+		op->flags & BCH_WRITE_MOVE ? "move" : "user");
 	ret = -EIO;
 err:
 	if (to_wbio(dst)->bounce)
@@ -1169,7 +1171,8 @@ static void bch2_nocow_write_convert_unwritten(struct bch_write_op *op)
 
 			bch_err_inum_offset_ratelimited(c,
 				insert->k.p.inode, insert->k.p.offset << 9,
-				"write error while doing btree update: %s",
+				"%s write error while doing btree update: %s",
+				op->flags & BCH_WRITE_MOVE ? "move" : "user",
 				bch2_err_str(ret));
 		}
 
@@ -1449,7 +1452,9 @@ err:
 					bch_err_inum_offset_ratelimited(c,
 						op->pos.inode,
 						op->pos.offset << 9,
-						"%s(): error: %s", __func__, bch2_err_str(ret));
+						"%s(): %s error: %s", __func__,
+						op->flags & BCH_WRITE_MOVE ? "move" : "user",
+						bch2_err_str(ret));
 				op->error = ret;
 				break;
 			}
@@ -1573,7 +1578,8 @@ CLOSURE_CALLBACK(bch2_write)
 		bch_err_inum_offset_ratelimited(c,
 			op->pos.inode,
 			op->pos.offset << 9,
-			"misaligned write");
+			"%s write error: misaligned write",
+			op->flags & BCH_WRITE_MOVE ? "move" : "user");
 		op->error = -EIO;
 		goto err;
 	}

libbcachefs/journal.c

@@ -53,33 +53,48 @@ static void bch2_journal_buf_to_text(struct printbuf *out, struct journal *j, u6
 	unsigned i = seq & JOURNAL_BUF_MASK;
 	struct journal_buf *buf = j->buf + i;
 
-	prt_printf(out, "seq:");
+	prt_str(out, "seq:");
 	prt_tab(out);
 	prt_printf(out, "%llu", seq);
 	prt_newline(out);
 	printbuf_indent_add(out, 2);
 
-	prt_printf(out, "refcount:");
+	prt_str(out, "refcount:");
 	prt_tab(out);
 	prt_printf(out, "%u", journal_state_count(s, i));
 	prt_newline(out);
 
-	prt_printf(out, "size:");
+	prt_str(out, "size:");
 	prt_tab(out);
 	prt_human_readable_u64(out, vstruct_bytes(buf->data));
 	prt_newline(out);
 
-	prt_printf(out, "expires");
+	prt_str(out, "expires:");
 	prt_tab(out);
 	prt_printf(out, "%li jiffies", buf->expires - jiffies);
 	prt_newline(out);
 
+	prt_str(out, "flags:");
+	prt_tab(out);
+	if (buf->noflush)
+		prt_str(out, "noflush ");
+	if (buf->must_flush)
+		prt_str(out, "must_flush ");
+	if (buf->separate_flush)
+		prt_str(out, "separate_flush ");
+	if (buf->need_flush_to_write_buffer)
+		prt_str(out, "need_flush_to_write_buffer ");
+	if (buf->need_flush_to_write_buffer)
+		prt_str(out, "need_flush_to_write_buffer ");
 	if (buf->write_done)
-		prt_printf(out, "write done\n");
+		prt_str(out, "write done ");
-	else if (buf->write_allocated)
+	if (buf->write_started)
-		prt_printf(out, "write allocated\n");
+		prt_str(out, "write started ");
-	else if (buf->write_started)
+	if (buf->write_allocated)
-		prt_printf(out, "write started\n");
+		prt_str(out, "write allocated ");
+	if (buf->write_done)
+		prt_str(out, "write done");
+	prt_newline(out);
 
 	printbuf_indent_sub(out, 2);
 }
@@ -1420,6 +1435,7 @@ void __bch2_journal_debug_to_text(struct printbuf *out, struct journal *j)
 	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)
 	       ? jiffies_to_msecs(j->next_reclaim - jiffies) : 0);
+	prt_printf(out, "blocked:\t\t%u\n",			j->blocked);
 	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:\t\t");

libbcachefs/journal_io.c

@@ -1830,7 +1830,10 @@ static int bch2_journal_write_prep(struct journal *j, struct journal_buf *w)
 
 	if (wb.wb)
 		bch2_journal_keys_to_write_buffer_end(c, &wb);
+
+	spin_lock(&c->journal.lock);
 	w->need_flush_to_write_buffer = false;
+	spin_unlock(&c->journal.lock);
 
 	start = end = vstruct_last(jset);
 
@@ -1948,12 +1951,20 @@ CLOSURE_CALLBACK(bch2_journal_write)
 	unsigned nr_rw_members = 0;
 	int ret;
 
+	for_each_rw_member(c, ca)
+		nr_rw_members++;
+
 	BUG_ON(BCH_SB_CLEAN(c->disk_sb.sb));
+	BUG_ON(!w->write_started);
 	BUG_ON(w->write_allocated);
+	BUG_ON(w->write_done);
 
 	j->write_start_time = local_clock();
 
 	spin_lock(&j->lock);
+	if (nr_rw_members > 1)
+		w->separate_flush = true;
+
 	ret = bch2_journal_write_pick_flush(j, w);
 	spin_unlock(&j->lock);
 	if (ret)
@@ -2008,12 +2019,6 @@ CLOSURE_CALLBACK(bch2_journal_write)
 	if (c->opts.nochanges)
 		goto no_io;
 
-	for_each_rw_member(c, ca)
-		nr_rw_members++;
-
-	if (nr_rw_members > 1)
-		w->separate_flush = true;
-
 	/*
 	 * Mark journal replicas before we submit the write to guarantee
 	 * recovery will find the journal entries after a crash.

libbcachefs/journal_reclaim.c

@@ -887,9 +887,11 @@ int bch2_journal_flush_device_pins(struct journal *j, int dev_idx)
 					 journal_seq_pin(j, seq)->devs);
 		seq++;
 
-		spin_unlock(&j->lock);
+		if (replicas.e.nr_devs) {
-		ret = bch2_mark_replicas(c, &replicas.e);
+			spin_unlock(&j->lock);
-		spin_lock(&j->lock);
+			ret = bch2_mark_replicas(c, &replicas.e);
+			spin_lock(&j->lock);
+		}
 	}
 	spin_unlock(&j->lock);
 err:

libbcachefs/rebalance.c

@@ -412,11 +412,11 @@ void bch2_rebalance_status_to_text(struct printbuf *out, struct bch_fs *c)
 		u64 now = atomic64_read(&c->io_clock[WRITE].now);
 
 		prt_str(out, "io wait duration:  ");
-		bch2_prt_human_readable_s64(out, r->wait_iotime_end - r->wait_iotime_start);
+		bch2_prt_human_readable_s64(out, (r->wait_iotime_end - r->wait_iotime_start) << 9);
 		prt_newline(out);
 
 		prt_str(out, "io wait remaining: ");
-		bch2_prt_human_readable_s64(out, r->wait_iotime_end - now);
+		bch2_prt_human_readable_s64(out, (r->wait_iotime_end - now) << 9);
 		prt_newline(out);
 
 		prt_str(out, "duration waited:   ");

libbcachefs/recovery_types.h

@@ -44,6 +44,7 @@
 	x(check_dirents,			27, PASS_FSCK)			\
 	x(check_xattrs,				28, PASS_FSCK)			\
 	x(check_root,				29, PASS_ONLINE|PASS_FSCK)	\
+	x(check_subvolume_structure,		36, PASS_ONLINE|PASS_FSCK)	\
 	x(check_directory_structure,		30, PASS_ONLINE|PASS_FSCK)	\
 	x(check_nlinks,				31, PASS_FSCK)			\
 	x(delete_dead_inodes,			32, PASS_FSCK|PASS_UNCLEAN)	\

libbcachefs/sb-errors_types.h

@@ -262,7 +262,9 @@
 	x(subvol_fs_path_parent_wrong,				254)	\
 	x(subvol_root_fs_path_parent_nonzero,			255)	\
 	x(subvol_children_not_set,				256)	\
-	x(subvol_children_bad,					257)
+	x(subvol_children_bad,					257)	\
+	x(subvol_loop,						258)	\
+	x(subvol_unreachable,					259)
 
 enum bch_sb_error_id {
 #define x(t, n) BCH_FSCK_ERR_##t = n,