One backlog for new requirements, another backlog for incidents. Track your time working on the different backlogs (or even per-tasks) and pretty soon you will have some numbers to show the boss why new stuff isn't happening snap-snap.