On Wed, 7 Aug 1996, Neal Becker wrote:
> gcc -c -g -O3 -I. deliver.c
> deliver.c: In function `deliver_local':
> deliver.c:573: `RLIMIT_CORE' undeclared (first use this function)
> deliver.c:573: (Each undeclared identifier is reported only once
> deliver.c:573: for each function it appears in.)
Oops. Philip goofs again. Apologies. I spent too many years working on
an operating system that existed in only one variation (MVS); Unix
continually catches me out with its infinite variety. It appears that
HP-UX does not have the RLIMIT_CORE option for setrlimit(). I've just
tried to check on the other OS supported by Exim, and those I can see at
the moment all seem to have it. (I can't get at a Linux or a BSDI system
just at present.) The following patch just cuts out the call on systems
that don't have it, and also fixes a misunderstanding on my part about
setrlimit in general - though that won't actually matter on other OS.
[Just had a report of other problems with 0.55, so there may even be
0.56 before I go away - something I was trying to avoid...]
*** exim-0.55/src/deliver.c Wed Aug 7 14:48:41 1996
--- deliver.c Thu Aug 8 09:46:19 1996
***************
*** 562,567 ****
--- 562,578 ----
if ((pid = fork()) == 0)
{
+ /* Prevent core dumps, as we don't want them in users' home directories.
+ HP-UX doesn't have RLIMIT_CORE; I don't know how to do this in that
+ system. */
+
+ #ifdef RLIMIT_CORE
+ struct rlimit rl;
+ rl.rlim_cur = 0;
+ rl.rlim_max = 0;
+ setrlimit(RLIMIT_CORE, &rl);
+ #endif
+
/* Ignore SIGINT and SIGTERM during delivery (SIGHUP is ignored throughout
exim, except when being a daemon). */
***************
*** 568,577 ****
signal(SIGINT, SIG_IGN);
signal(SIGTERM, SIG_IGN);
- /* Prevent core dumps, as we don't want them in users' home directories. */
-
- setrlimit(RLIMIT_CORE, 0);
-
/* Close the unwanted half of the pipe, and set the required gid/uid, first
regaining root privilege if necessary. Note that mac_sete{g,u}id expands
to a failure on systems without the sete{g,u}id functions, but on such
--- 579,584 ----
--
Philip Hazel University Computing Service,
ph10@??? New Museums Site, Cambridge CB2 3QG,
P.Hazel@??? England. Phone: +44 1223 334714
