Comment 3 for bug 1248181

I don't remember the order of which the signals hit each other to cause the problem.
If uc_link is a reliable path to the original synchronous interrupt, then that might do.