------- You are receiving this mail because: -------
You are on the CC list for the bug.
http://bugs.exim.org/show_bug.cgi?id=972
--- Comment #10 from Phil Pennock <pdp@???> 2010-09-07 02:37:25 ---
So Exim uses signals in a couple of places. Where it uses setitimer(), it
masks out all signals except SIGALRM (which should be safe, for a short period
of time) so that only the alarm will get through. nanosleep is not specified
to use any particular mechanism, so we can make rough guesses (let through
SIGALRM) I don't know enough about the portability of this assumption to be
entirely comfortable.
So I'd rather stick to the known-good known-portable code which has only ever
had issues with one buggy kernel, and then work around the buggy kernel, rather
than risk breakages on all the non-buggy platforms by making unwarranted
assumptions.
Per:
http://www.gnu.org/software/hello/manual/gnulib/nanosleep.html
we should be fairly safe on current platforms, for our usage, with nanosleep,
so if someone else feels we should switch, I'll not object.
--
Configure bugmail:
http://bugs.exim.org/userprefs.cgi?tab=email
