bell@eniac.seas.upenn.edu (Mike Bell) (06/29/89)
This is from Barry Twycross at Du Pont Pixel systems in England really
I was writting a DDP socket listener running under multifinder, and I
assumed that the interupt handler would take care to make sure that your
CurrentA5 would be set up correctly, but it dosn't. I seem to get other
programs A5 value occasionally (this is good for crashing lightspeed C
or the debugger out from underneath of you as I was writing into thier
global space.)
The fix I use is to store my CurrentA5 processor relative in the code,
but this is really against my principles.
********************************************************************************
Mike Bell CSnet: BELLMA%ERVX01@dupont.com
Senior Engineer Applelink: D2747
DuPont Electronic Imaging
Core Technology Group
Can YOUR mac play FOOTBALL ????
********************************************************************************tim@hoptoad.uucp (Tim Maroney) (07/12/89)
In article <12515@netnews.upenn.edu> bell@eniac.seas.upenn.edu (Mike Bell) writes: >This is from Barry Twycross at Du Pont Pixel systems in England really > >I was writting a DDP socket listener running under multifinder, and I >assumed that the interupt handler would take care to make sure that your >CurrentA5 would be set up correctly, but it dosn't. I seem to get other >programs A5 value occasionally (this is good for crashing lightspeed C >or the debugger out from underneath of you as I was writing into thier >global space.) > >The fix I use is to store my CurrentA5 processor relative in the code, >but this is really against my principles. I've been writing DDP socket listeners for something like five years now, and I have yet to find another solution. Way back when, Switcher had the same problem. It is not only in socket listeners, but in all interrupt driven code -- if you install your own interrupt handlers, you can't assume that you will only be called when your application controls CurrentA5. There's no place else to store your A5 except in PC-relative code space. -- Tim Maroney, Mac Software Consultant, sun!hoptoad!tim, tim@toad.com Postal: 424 Tehama, SF CA 94103; Phone: (415) 495-2934 "Superhero stories could best be described as entertainment that externalizes childhood power fantasies." -- Timothy Fay on rec.arts.comics
ts@cup.portal.com (Tim W Smith) (07/15/89)
I'd like to see a slight change in the interface to protocol handlers to address this problem. The change would be that register A5, instead of being defined as "available" when the protocol handler is called, would instead be defined as contiaing the value it contained when the protocol handler was attached. Tim Smith
tim@hoptoad.uucp (Tim Maroney) (07/16/89)
In article <20453@cup.portal.com> ts@cup.portal.com (Tim W Smith) writes: >I'd like to see a slight change in the interface to protocol handlers to >address this problem. The change would be that register A5, instead of >being defined as "available" when the protocol handler is called, would >instead be defined as contiaing the value it contained when the protocol >handler was attached. I'd like this change to be made retroactively -- unfortunately, as much sense as the proposed change makes, you'd have to work even harder than you do now. You'd have to check the driver version and go to your A5-saving code if the new one were not installed, skipping it if the new one were installed. Most people would just leave out the version test and go on saving and restoring A5 as at present -- it's easier. So it's really not worth changing the driver. It would have been nice if this had been in the definition from the start, though. -- Tim Maroney, Mac Software Consultant, sun!hoptoad!tim, tim@toad.com "Strong men tremble when they hear it. They've got cause enough to fear it; It's even blacker than they smear it! No one mentions -- my name." - Bill Sykes, "Oliver"