Thomas Munro <thomas.mu...@enterprisedb.com> writes: > Let me try that again with that stupid typo (crc2) fixed...
I didn't like that too much as-is, because it was capable of calling elog(ERROR) without having reset the SIGILL trap first. That's just trouble waiting to happen, so I rearranged to avoid it. I also noticed that we'd been sloppy about making the file safe to compile for both frontend and backend, so I cleaned that up. Also, I had thought that maybe the postmaster should do something to ensure that it sets up the function pointer, so that child processes inherit the correct pointer via fork() and don't need to repeat the test (and then possibly spam the postmaster log). On closer inspection, no new code is needed because ReadControlFile runs a CRC check, but I felt it was worth documenting that. regards, tom lane