Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site allegra.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!jpl
From: jpl@allegra.UUCP (John P. Linderman)
Newsgroups: net.bugs.uucp
Subject: Re: Re: hung line help needed
Message-ID: <2923@allegra.UUCP>
Date: Sun, 25-Nov-84 11:54:06 EST
Article-I.D.: allegra.2923
Posted: Sun Nov 25 11:54:06 1984
Date-Received: Mon, 26-Nov-84 02:04:09 EST
References: <85@daemon.UUCP> <33700001@trsvax.UUCP>
Organization: AT&T Bell Laboratories, Murray Hill
Lines: 37


    Could you pass the code (or diffs) along for the timeout fix you've added?

	{microsoft,ctvax}!trsvax!ron

I think I can do better.  Kirk Smith pointed out a bug in sleep(3) that
causes alarms to be masked off and never rearmed.  Since uucico does
both sleeps and alarms, his fix (enclosed below) looks like it will cure
the disease whereas mine merely treated the symptoms.

    From: pur-ee!ef:ks (Kirk Smith)
    Received: by ef.ECN; Mon, 27 Aug 84 10:32:00 est (4.12/5.20)
    Message-Id: <8408271532.AA03198@ef.ECN>
    Subject: sleep(3) and SIGALRM
    Index: 	/usr/src/lib/libc/gen/sleep.c 4.2BSD
    Apparently-To: ucbvax!4bsd-bugs

    Description:
	    After a call to sleep(), the signal SIGALRM remains blocked.

    Repeat-By:
	    main()
	    {
		    sleep(1);
		    alarm(1);
	    }
	    This program will never terminate.

    Fix:
	    In sleep.c add the following line to set the mask back to the old
	    mask before returning as follows:

		(void) sigvec(SIGALRM, &ovec, (struct sigvec *)0);
		(void) setitimer(ITIMER_REAL, &oitv, (struct itimerval *)0);
	+++	(void) sigsetmask(omask);
	}

John P. Linderman  (Not alarmed by the slip in sleep)  allegra!jpl