Received: from PACIFIC-CARRIER-ANNEX.MIT.EDU by po7.MIT.EDU (5.61/4.7) id AA09888; Fri, 1 Dec 95 05:21:15 EST
Received: from JIMI.MIT.EDU by MIT.EDU with SMTP
	id AA16223; Fri, 1 Dec 95 05:07:38 EST
Received: by jimi.MIT.EDU (5.57/4.7) id AA16874; Fri, 1 Dec 95 05:08:43 -0500
Message-Id: <9512011008.AA16874@jimi.MIT.EDU>
To: "SethMeister G." <consp05@bingsuns.cc.binghamton.edu>
Cc: pthreads@MIT.EDU
Subject: Re: BUG IN PTHREADS 1_60BETA4 
In-Reply-To: Your message of "Tue, 28 Nov 1995 17:48:43 EST."
             <199511282248.RAA06886@bingsun2-gw> 
Date: Fri, 01 Dec 1995 05:08:42 EST
From: Christopher Provenzano  <proven@MIT.EDU>


> Hi there,
> 
>   I just wanted to report 2 bugs in the pthreads library:
> 
>    (1)  In the scanf stdio library function, the mutex held while input 
>    was performed was locked TWICE once before the stream manip, and once 
>    after (it should be unlocked after, but it is NOT ).
> 
>    (2) This one is NASTY - On Solaris 2.4, fork() does not work properly 
> 100% of the time.  eg: if one tries the following code:
> 
>   for ( int i = 0; i < 100; i++ )
>     if ( fork() == 0 ) exit(0);
> 
>   Sometimes, the process LOCKS UP totally.
> 
> 
> Please send me an acknowledgement once you receive this.  Thank you.

OK, I got them and have fixed 1 and will try and fix 2. Number 2 is not a 
lockup but a bug in Solaris where if the parent gets a SIGCHLD and tries to
reinstall the signal handler and does not do a wait() in the signal handler
then the kernel will never mark the signal as handled and will reissue the
signal when the handler returns.

CAP