Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!caip!clyde!burl!ulysses!mhuxr!mhuxt!houxm!ihnp4!inuxc!pur-ee!uiucdcs!uiucuxc!falk
From: falk@uiucuxc.CSO.UIUC.EDU
Newsgroups: net.unix-wizards
Subject: SIGNAL failure FTN->C
Message-ID: <106000021@uiucuxc>
Date: Thu, 28-Aug-86 12:06:00 EDT
Article-I.D.: uiucuxc.106000021
Posted: Thu Aug 28 12:06:00 1986
Date-Received: Sun, 31-Aug-86 10:06:07 EDT
Lines: 28
Nf-ID: #N:uiucuxc:106000021:000:1130
Nf-From: uiucuxc.CSO.UIUC.EDU!falk    Aug 28 11:06:00 1986


{}
problem synopsis: signal handling in a c-function fails with an AST-fault
                  when the c-function was invoked from a fortran program

example:	CALL CFUNC       (in fortran- only executable line in pgm)

		-------
		   .
		   .
		   .
		SIGNAL(SIGALRM,func_addr);
		ALARM(SECONDS);
		   .
		   .
		   .
when the ALARM goes off, an AST fault occurs in the gsignal routine (PUSHL R2,
if I am reading my addresses correctly). gsignal allegedly sets the signal, so
the fault is occuring before it ever gets to func_addr. The c-routine that does
the signal processing, run by itself, does just fine. Anyone know why it breaks
when I surround it with a fortran program? (no editorial comments about *why* I
would want to surround a perfectly reasonable C program with an awful fortran
program, it wasn't *my* choice :-)

Is there some sort of workaround (short of taking out the signal processing)
that I could do to get this beast to work?  I've looked at SIGVEC and SIGSTACK
as a possible means of forcing an alternate stack, but I haven't been able
to get them to work, either.  Any help/hints would be appreciated.