From: Jason Green <news AT jgreen4 DOT fsnet DOT co DOT uk>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Signal & setitimer
Date: Sun, 07 May 2000 20:18:02 +0100
Organization: Customer of Planet Online
Lines: 105
Message-ID: <o9gbhskd6sm2n0cgidgnnu3q6hvc070qsq@4ax.com>
References: <200005051818 DOT OAA05711 AT indy DOT delorie DOT com>
NNTP-Posting-Host: modem-56.dextroamphetam.dialup.pol.co.uk
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: newsg2.svr.pol.co.uk 957727120 3262 62.136.90.56 (7 May 2000 19:18:40 GMT)
NNTP-Posting-Date: 7 May 2000 19:18:40 GMT
X-Complaints-To: abuse AT theplanet DOT net
X-Newsreader: Forte Agent 1.7/32.534
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Eli Zaretskii <eliz AT delorie DOT com> wrote:

> > From: Frederic Cazenave <cazenave AT radar DOT mcgill DOT ca>

> > I have written a small progranm to test setitmer.
> > It's appear this code run correctly on a unix system but
> > not under DOS.
> > Under UNIX the timer is reactivate every 10 seconds, but
> > under dos the timer is active the first time but after is never
> > reactivate.
> 
> I cannot reproduce this with your test program on my machine.  On what
> OS did you try this, and what version of DJGPP do you use?

I have re-written the test program (pasted below) to prove that
setitimer() is not somehow corrupting the timeout variable, and to
remove any suspicion about alarm(), which is also a libc function.  

I'm not sure exactly what the expected behaviour of this program
*should* be (can someone clarify this) but the actual behaviour is
definitely not consistent.

This is with DJGPP v2.03 under Windows 95, compiled with gcc 2.95.2
using the command: gcc -Wall test.c -o test

In a normal DOS box the program runs (until aborted with Ctrl-C).
Unless I run using redir, in which case it terminates immediately with
exit code 1:

D:\CODE\TEST>test

setitimer0123!!!!4!!!!!!!!!!!!!!!!!!!5!!!!!!!!!!!!!!!!!!!6!!!!!!!!!!!!!!!!!!!7!!
!!!!!!!!!!!!!!!!!8!!!!!!!!!!!!!!!!!!!9
setitimer0123!!!!4!!!!!!!!!!!!!!!!!!!5!!!!!!!!!!!!!!!!!!!6!!!!!!!!!!!!!!!!!!!7!!
!!!!!!!!!!!!!!!!!8!!!!!!!!!!!!!!!!!!!9
setitimer0123!!!!4!!!!!!!!!!!!!!!!!!!5!!!!!!!!!!!!!!!!!!!6!!!!!!!!!!!!!!!!!!!7!!
!!!!!!!!!!!!!!!!!8!!!!!!!!!!!!!!!!^C
Exiting due to signal SIGINT
INTR key Pressed at eip=0000897d

D:\CODE\TEST>redir -x -t test
Exit code: 1
Elapsed time: 0.280 seconds

A couple of times when pressing the Ctrl key, Windows popped up a box
saying "The program encountered a general protection exception".
After this Windows becomes unstable and requires a reboot.

Running the program from UltraEdit gives a different result and the
timing between printing digits varies +/- a fraction of a second:

setitimer0123456789
setitimer0123456789
setitimer0123456789
setitimer0123456

UltraEdit seems to use the programs c:\windows\system\conagent.exe and
c:\windows\system\redir32.exe to create a shell for test.exe and
capture its output. 

Each running of test.exe from UltraEdit leaves instances of Redir32
and Winoldap listed when I do Ctrl-Alt-Del even though the test
program appears to exit gracefully.

This is the test code (modified from the original post):

#include <stdlib.h>
#include <stdio.h>
#include <signal.h>
#include <sys/time.h>
#include <unistd.h>

void handle_sigalrm(int sig) {
 printf("!");
 fflush(stdout);
 signal (SIGALRM,handle_sigalrm);
}

int main(void) {
   static struct itimerval timeout;
   int i;

   signal (SIGALRM,handle_sigalrm);

   while(1)
   {
        timeout.it_interval.tv_sec=0;
        timeout.it_interval.tv_usec = 0;
        timeout.it_value.tv_sec=5;
        timeout.it_value.tv_usec = 0;
        
        printf("\nsetitimer");
        
        setitimer(ITIMER_REAL,&timeout,NULL);

        for (i=0; i<10; i++)
        {
            sleep(1);
            printf("%d", i);
            fflush(stdout);
        }
    }
    
    return(0);
}