bcachefs: Fixes for startup on very full filesystems

 - Always pass BTREE_INSERT_USE_RESERVE when writing alloc btree keys
 - Don't strand buckest on the copygc freelist until after recovery is
   done and we're starting copygc.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

fs/bcachefs/alloc_background.c

@@ -315,7 +315,9 @@ static int bch2_alloc_write_key(struct btree_trans *trans,
 	bch2_trans_update(trans, iter, &a->k_i,
 			  BTREE_TRIGGER_NORUN);
 	ret = bch2_trans_commit(trans, NULL, NULL,
-				BTREE_INSERT_NOFAIL|flags);
+				BTREE_INSERT_NOFAIL|
+				BTREE_INSERT_USE_RESERVE|
+				flags);
 err:
 	if (ret == -EINTR)
 		goto retry;
@@ -1033,7 +1035,16 @@ static int push_invalidated_bucket(struct bch_fs *c, struct bch_dev *ca, size_t
 		set_current_state(TASK_INTERRUPTIBLE);
 
 		spin_lock(&c->freelist_lock);
-		for (i = 0; i < RESERVE_NR; i++)
+		for (i = 0; i < RESERVE_NR; i++) {
+
+			/*
+			 * Don't strand buckets on the copygc freelist until
+			 * after recovery is finished:
+			 */
+			if (!test_bit(BCH_FS_STARTED, &c->flags) &&
+			    i == RESERVE_MOVINGGC)
+				continue;
+
 			if (fifo_push(&ca->free[i], bucket)) {
 				fifo_pop(&ca->free_inc, bucket);
 
@@ -1043,6 +1054,7 @@ static int push_invalidated_bucket(struct bch_fs *c, struct bch_dev *ca, size_t
 				spin_unlock(&c->freelist_lock);
 				goto out;
 			}
+		}
 
 		if (ca->allocator_state != ALLOCATOR_BLOCKED_FULL) {
 			ca->allocator_state = ALLOCATOR_BLOCKED_FULL;

fs/bcachefs/super.c

@@ -860,6 +860,8 @@ int bch2_fs_start(struct bch_fs *c)
 	if (bch2_fs_init_fault("fs_start"))
 		goto err;
 
+	set_bit(BCH_FS_STARTED, &c->flags);
+
 	if (c->opts.read_only || c->opts.nochanges) {
 		bch2_fs_read_only(c);
 	} else {
@@ -871,7 +873,6 @@ int bch2_fs_start(struct bch_fs *c)
 			goto err;
 	}
 
-	set_bit(BCH_FS_STARTED, &c->flags);
 	print_mount_opts(c);
 	ret = 0;
 out: