Xref: news-dnh.mv.net comp.os.msdos.djgpp:137
Path: news-dnh.mv.net!mv!news.sprintlink.net!gatech!news.mathworks.com!news2.near.net!das-news2.harvard.edu!oitnews.harvard.edu!newsfeed.rice.edu!uw-beaver!uw-coco!nwfocus.wa.com!news.telebyte.com!usenet
From: jhunter AT kendaco DOT telebyte DOT com (Joel Hunter)
Newsgroups: comp.os.msdos.djgpp
Subject: IRQ0 & DPMI
Date: Fri, 02 Jun 1995 02:52:01 GMT
Organization: Telebyte North West Silverdale, Wa
Lines: 14
Nntp-Posting-Host: kendaco.telebyte.com
To: djgpp AT sun DOT soe DOT clarkson DOT edu
Dj-Gateway: from newsgroup comp.os.msdos.djgpp

  Just a real quick question, so I understand completely.  Does the
default protected-mode IRQ0 handler chain down to the real-mode
handler, or does GO32 setup protected-mode code to do the same job?
(It seems strange to think that we would be switching to and fro 18.2
times a second and still our programs are so fast! :-)
  Also, I have one other question, and please forgive me if this is
not DJGPP-specific.  In the DPMI 0.9 specs it says that protected-mode
hardware interrupt procedures must also set the interrupt flag before
leaving, just like in real mode.  I have been doing this (I think...)
by calling DPMI function 0x901 at the end of my interrupt routine, but
it seems to cease calling the handler after a few thousand calls.  If
I remove the call to 0x901, it works ok.  Is this unusual behavior or
am I missing something? (I'm running under DPMI.)