[513] in Zephyr_Bugs
zhm: Solaris signal handling code
daemon@ATHENA.MIT.EDU (epeisach@MIT.EDU)
Thu Jan 6 11:30:25 1994
From: epeisach@MIT.EDU
Date: Thu, 6 Jan 1994 11:30:17 -0500
To: bugs@MIT.EDU, bug-zephyr@MIT.EDU
I discovered this problem when porting zhm to another machine....
I needed the POSIX sigaction handling and noticed that when I compiled
with optimization, zhm was screwed (complaingin could not grow signal
stack, etc). Without optimization all was ok....
The problem was that the sa_flags was never set to something useful
(like 0) and probably had the SA_ONSTACK bit set causing problems...
If this has not been seen on solaris, then I am suprised.
Has zhm ever died mysteriously on Solaris? If so, this patch is probably
required...
*** 1.52 1993/07/26 08:29:36
--- zhm.c 1994/01/06 16:26:38
***************
*** 56,62 ****
--- 56,65 ----
char *PidFile = PIDFILE;
extern char *sbrk();
+ #ifndef POSIX
extern long time();
+ #endif
+
void init_hm(), detach(), handle_timeout(), resend_notices(), die_gracefully();
***************
*** 300,306 ****
struct servent *sp;
Code_t ret;
FILE *fp;
! #ifdef SOLARIS
struct sigaction sa;
#endif
--- 303,309 ----
struct servent *sp;
Code_t ret;
FILE *fp;
! #ifdef POSIX
struct sigaction sa;
#endif
***************
*** 398,406 ****
send_boot_notice(HM_BOOT);
deactivated = 0;
! #ifdef SOLARIS
sa.sa_handler = (void (*)()) set_sig_type;
sigemptyset(&sa.sa_mask);
sigaction(SIGHUP, &sa,0);
sigaction(SIGALRM, &sa, 0);
sigaction(SIGTERM, &sa,0);
--- 401,410 ----
send_boot_notice(HM_BOOT);
deactivated = 0;
! #if defined(POSIX)
sa.sa_handler = (void (*)()) set_sig_type;
sigemptyset(&sa.sa_mask);
+ sa.sa_flags = 0;
sigaction(SIGHUP, &sa,0);
sigaction(SIGALRM, &sa, 0);
sigaction(SIGTERM, &sa,0);
