From:	CRDGW2::CRDGW2::MRGATE::"SMTP::CRVAX.SRI.COM::RELAY-INFO-VAX" 29-AUG-1989 08:47
To:	MRGATE::"ARISIA::EVERHART"
Subj:	SNAP & WATCH, user beware!

Message-Id:  <8908291241.AA21340@crdgw1.ge.com>
Received: From KL.SRI.COM by CRVAX.SRI.COM with TCP; Tue, 29 AUG 89 05:18:46 PDT
Received: from drcvax.af.mil by KL.SRI.COM with TCP; Tue, 29 Aug 89 05:08:41 PDT
Date: 29 Aug 89 07:54:00 EST
From: "Daniel J. Graham" <graham@drcvax.af.mil>
Subject: SNAP & WATCH, user beware!
To: "info-vax" <info-vax@kl.sri.com>

Hi all,

Before I send off part 3 of SNAP, I want to clearly state a difficulty that 
some kind soul pointed out to me.  

As you know, the old MACRO device SETIPL for privileged code has been 
replaced with what are called "spinlocks."  There are two places in this 
program where MACRO routines used to use SETIPL calls:  LIB_DEFAULT_DIR, 
the macro routine that SNAP uses to find out the default directory of the 
observed process, it must execute an AST within the context of the user's 
process to translate SYS$DISK.  The second instance is in the program 
WATCH, which observes another's terminal input/output.  

Both of these routines have had some spinlocks implemented, but they still 
have SETIPL calls too.  I am no MACRO whix, and can't debug them further 
than they have already been.  Other programmers, far better than I have 
attempted to do so, with no success.  We just can't seem to get them to 
operate without those generated warnings.

This is both a caution and a plea:  Many have used SMAP/WATCH on MP 
systems, with no crashes.  This does NOT mean that it cannot cause one.  If 
two competing processes attempt to grap a structure, the system will most 
likely crash.  Because it hasen't happened, doesn't mean it won't.  I 
repeat, however, that I k,now of no crashes of MP systems yet...

Now the plea:  Is there anyone willing to take the two modules, 
LIB_DEFAULT_DIR.MAR and WATCH.MAR and try to exorcise the SETIPLs?
If so, just wait a day or so and all parts of SNAP with the routines will 
have been posted.  I wrote neither of the MACRO routines, and have no pride 
relative to their being modified.  I'll accept anyone's assistance.

Have fun all, part 3 of 8 follows in a little while.

Dan Graham


