> Then I edited the diff to move the failing hunk from the end of the patch > file to the beginning and it worked fine. But I would find it hard to > believe it was reflecting a bug in a utility as simple as patch. Well, a gdb session could tell.