bcachefs: Minor device removal fixes

 - We weren't clearing the LRU btree
 - bch2_alloc_read() runs before bch2_check_alloc_key() deletes alloc
   keys for devices/buckets that don't exists, so it needs to check for
   that
 - bch2_check_lrus() needs to check that buckets exists
 - improve some error messages
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

fs/bcachefs/alloc_background.c

@@ -446,6 +446,13 @@ int bch2_alloc_read(struct bch_fs *c)
 
 	for_each_btree_key(&trans, iter, BTREE_ID_alloc, POS_MIN,
 			   BTREE_ITER_PREFETCH, k, ret) {
+		/*
+		 * Not a fsck error because this is checked/repaired by
+		 * bch2_check_alloc_key() which runs later:
+		 */
+		if (!bch2_dev_bucket_exists(c, k.k->p))
+			continue;
+
 		ca = bch_dev_bkey_exists(c, k.k->p.inode);
 		bch2_alloc_to_v4(k, &a);
 
@@ -614,7 +621,8 @@ static int bch2_check_alloc_key(struct btree_trans *trans,
 		return ret;
 
 	if (fsck_err_on(!bch2_dev_bucket_exists(c, alloc_k.k->p), c,
-			"alloc key for invalid device or bucket"))
+			"alloc key for invalid device:bucket %llu:%llu",
+			alloc_k.k->p.inode, alloc_k.k->p.offset))
 		return bch2_btree_delete_at(trans, alloc_iter, 0);
 
 	ca = bch_dev_bkey_exists(c, alloc_k.k->p.inode);
@@ -727,9 +735,8 @@ static int bch2_check_discard_freespace_key(struct btree_trans *trans,
 	bch2_trans_iter_init(trans, &alloc_iter, BTREE_ID_alloc, pos, 0);
 
 	if (fsck_err_on(!bch2_dev_bucket_exists(c, pos), c,
-			"%llu:%llu set in %s btree but device or bucket does not exist",
-			pos.inode, pos.offset,
-			bch2_btree_ids[iter->btree_id]))
+			"entry in %s btree for nonexistant dev:bucket %llu:%llu",
+			bch2_btree_ids[iter->btree_id], pos.inode, pos.offset))
 		goto delete;
 
 	k = bch2_btree_iter_peek_slot(&alloc_iter);

fs/bcachefs/lru.c

@@ -133,7 +133,7 @@ static int bch2_check_lru_key(struct btree_trans *trans,
 	struct bch_alloc_v4 a;
 	struct printbuf buf1 = PRINTBUF;
 	struct printbuf buf2 = PRINTBUF;
-	u64 idx;
+	struct bpos alloc_pos;
 	int ret;
 
 	lru_k = bch2_btree_iter_peek(lru_iter);
@@ -144,10 +144,15 @@ static int bch2_check_lru_key(struct btree_trans *trans,
 	if (ret)
 		return ret;
 
-	idx = le64_to_cpu(bkey_s_c_to_lru(lru_k).v->idx);
+	alloc_pos = POS(lru_k.k->p.inode,
+			le64_to_cpu(bkey_s_c_to_lru(lru_k).v->idx));
 
-	bch2_trans_iter_init(trans, &iter, BTREE_ID_alloc,
-			     POS(lru_k.k->p.inode, idx), 0);
+	if (fsck_err_on(!bch2_dev_bucket_exists(c, alloc_pos), c,
+			"lru key points to nonexistent device:bucket %llu:%llu",
+			alloc_pos.inode, alloc_pos.offset))
+		return bch2_btree_delete_at(trans, lru_iter, 0);
+
+	bch2_trans_iter_init(trans, &iter, BTREE_ID_alloc, alloc_pos, 0);
 	k = bch2_btree_iter_peek_slot(&iter);
 	ret = bkey_err(k);
 	if (ret)

fs/bcachefs/super.c

@@ -1423,11 +1423,17 @@ static int bch2_dev_remove_alloc(struct bch_fs *c, struct bch_dev *ca)
 	struct bpos end		= POS(ca->dev_idx, U64_MAX);
 	int ret;
 
-	ret =   bch2_btree_delete_range(c, BTREE_ID_alloc, start, end,
+	/*
+	 * We clear the LRU and need_discard btrees first so that we don't race
+	 * with bch2_do_invalidates() and bch2_do_discards()
+	 */
+	ret =   bch2_btree_delete_range(c, BTREE_ID_lru, start, end,
+					BTREE_TRIGGER_NORUN, NULL) ?:
+		bch2_btree_delete_range(c, BTREE_ID_need_discard, start, end,
 					BTREE_TRIGGER_NORUN, NULL) ?:
 		bch2_btree_delete_range(c, BTREE_ID_freespace, start, end,
 					BTREE_TRIGGER_NORUN, NULL) ?:
-		bch2_btree_delete_range(c, BTREE_ID_need_discard, start, end,
+		bch2_btree_delete_range(c, BTREE_ID_alloc, start, end,
 					BTREE_TRIGGER_NORUN, NULL);
 	if (ret)
 		bch_err(c, "error %i removing dev alloc info", ret);
@@ -1462,19 +1468,19 @@ int bch2_dev_remove(struct bch_fs *c, struct bch_dev *ca, int flags)
 		goto err;
 	}
 
-	ret = bch2_journal_flush_device_pins(&c->journal, ca->dev_idx);
+	ret = bch2_dev_remove_alloc(c, ca);
 	if (ret) {
-		bch_err(ca, "Remove failed: error %i flushing journal", ret);
+		bch_err(ca, "Remove failed, error deleting alloc info");
 		goto err;
 	}
 
-	ret = bch2_dev_remove_alloc(c, ca);
+	ret = bch2_journal_flush_device_pins(&c->journal, ca->dev_idx);
 	if (ret) {
-		bch_err(ca, "Remove failed, error deleting alloc info");
+		bch_err(ca, "Remove failed: error %i flushing journal", ret);
 		goto err;
 	}
 
-	ret = bch2_journal_error(&c->journal);
+	ret = bch2_journal_flush(&c->journal);
 	if (ret) {
 		bch_err(ca, "Remove failed, journal error");
 		goto err;