Initialize ref_blkno output at the start of block_dirty_ref()

scoutfs_block_dirty_ref() only sets *ref_blkno on the path that
allocates a new cow block. The early return paths leave it untouched,
so a caller that reads it back gets whatever value was on the stack.

This is harmless with the current callers. dirty_alloc_blocks() bails
out before calling here when the blocks are already dirty, so the
early return on an already-dirty block is never reached, and the
error paths return a negative value that callers check before using
ref_blkno.

Signed-off-by: Auke Kok <auke.kok@versity.com>

kmod/src/block.c

@@ -836,6 +836,9 @@ int scoutfs_block_dirty_ref(struct super_block *sb, struct scoutfs_alloc *alloc,
 	int ret;
 	int err;
 
+	if (ref_blkno)
+		*ref_blkno = 0;
+
 	/* read existing referenced block, if any */
 	blkno = le64_to_cpu(ref->blkno);
 	if (blkno) {