Xiaoran Wang <fanfuxiaoran@gmail.com> writes:
>> Yeah, that's the big problem. I don't have any confidence at all
>> that this would detect misuse. It'd require that the old stack
>> frame gets overwritten, which might not happen for a long time,
>> and it'd require that somebody eventually do a longjmp, which again
>> might not happen for a long time --- and when those did happen, the
>> error would be detected in someplace far away from the actual bug,
>> with little evidence remaining to help you localize it.
> Exactly, it cannot tell you which PG_TRY left the invalid sigjmp_buf,
> but to implement that is easy I think, recording the line num maybe.
I don't think you get to assume that the canary word gets overwritten
but debug data a few bytes away survives.
regards, tom lane
