Path: utzoo!utgpu!water!watmath!clyde!rutgers!sri-spam!ames!amdcad!sun!pitstop!texsun!texsun.central-relay.sun.com!convex!convexe!authorplaceholder
From: loafman@convexe.UUCP
Newsgroups: comp.sys.ibm.pc
Subject: Re: MSC 4.0 interrupt (^C) handler help
Message-ID: <63200025@convexe>
Date: 29 Jan 88 15:43:00 GMT
References: <682@silver.bacs.indiana.edu>
Lines: 41
Nf-ID: #R:silver.bacs.indiana.edu:-68200:convexe:63200025:000:2085
Nf-From: convexe.UUCP!loafman    Jan 29 09:43:00 1988


/* Written  7:25 pm  Jan 27, 1988 by creps@silver.Sun.COM in convexe:comp.sys.ibm.pc */
/* ---------- "MSC 4.0 interrupt (^C) handler help" ---------- */

   I'm trying to get an interrupt handler to work in MSC 4.0 that will
handle ^C's at any time. The manual isn't a lot of help, giving only
one example with the signal routine. I wrote a test program which is
basically for(i=1;i<100000;i++) if (i % 10000 == 0) fprintf("%d",i);
and it seems to work, except that it waits until it's ready to print the
next number before calling the interrupt routine. However, it does seem
as if it is storing the interrupt until output is done, and not just
waiting until the output before it even recognizes the interrupt.

...remainder deleted...

/* End of text from convexe:comp.sys.ibm.pc */

Steve,

Your assessment is correct, MSDOS is waiting until the I/O is done (or
until the first i/o system call is done) before recognizing the
interrupt.  Until then, it is just queued (logically) into the first
keyboard buffer position.  If you want more frequent looks at the
state of Ctrl-C, then do a 'break on' in either autoexec.bat, or do
the appropriate system call.  This will cause MSDOS to look at the
state of Ctrl-C on every system call, and not just those specific to
i/o.  Basically Ctrl-C handling is polled, not interrupt driven.

There is a 'user settable' interrupt, 0x1B I think, that gets called
as soon as the keyboard handler sees the Ctrl-C.  If you write your
own interrupt routine, you could do something intelligent then, such
as setting a global flag or such.  Keep it simple though.  Keyboard
interrupt routines are hard to debug without a hardware based
debugger.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Kenneth W. Loafman @ CONVEX Computer Corp, Dallas, Texas | All opinions
Voice:	  work: (214) 952-0829  home: (214) 964-1250     | are my own,
USPSnail: 1705 Coit Rd #2101, Plano, Tx 75075		 | of course.
UUCP:	  ihnp4!convex!loafman				 |        ...KWL
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~