I wrote:
>> Hmm. That seems like a problem, all right, but I don't believe the
>> claim that it applies only to the poll() code path. On many platforms
>> the select() path would have the same issue.
> BTW, doesn't win32_latch.c have the identical problem? I don't see it
> updating the timeout after a signal, either.
I've committed a patch for this, but the win32_latch version is entirely
innocent of any testing. The buildfarm will probably show any mistake
there, but if someone would like to try Jeff's test case on Windows just
to make sure the autovac launcher wakes up on time, it wouldn't hurt.
regards, tom lane