efa@iesd.auc.dk (Erik F. Andersen) (10/05/90)
I hope anyone will take the time to help me out with some programming problems I have.
a) I want to write a resident routine that can intercept the commands entered from the DOS prompt, and process the command if it is defined in the routine, or otherwise return controle to COMMAND.COM. What I think I need to know, is the following:
1) Where in memory does DOS store the string entered?
2) What interrupt is issued after I hit return?
3) Maby it is better if my routine is invoked, after COMMAND.COM detects that the commands entered are not in it's library, and before the error: COMMAND NOT FOUND is written on the screen! How can I do this - if it's possible?
b) When I in TP (v. 5.0) write sub-routines to certain hardware interrupts, like INT08 to redifine caracters written on to the screen, strange things start to happen. So far the only way I have been able to do this, is through assemble-language, where I never have any problems. I have often written sub-routines in TP to other BIOS-routines, like INT09 and INT 1C, without any trouble. If anyone think they know, what I am doing wrong, please let me know.
FROM: Erik F. Andersen
efa@iesd.auc.dk abcscnuk@Twg-S5.uucp (Naoto Kimura (ACM)) (10/08/90)
In article <EFA.90Oct5133312@student.iesd.auc.dk> efa@iesd.auc.dk (Erik F. Andersen) writes: > > I hope anyone will take the time to help me out with some > programming problems I have. > > a) I want to write a resident routine that can intercept the > commands entered from the DOS prompt, and process the command > if it is defined in the routine, or otherwise return controle > to COMMAND.COM. What I think I need to know, is the following: > > 1) Where in memory does DOS store the string entered? > 2) What interrupt is issued after I hit return? > 3) Maybe it is better if my routine is invoked, after COMMAND.COM > detects that the commands entered are not in it's library, and > before the error: COMMAND NOT FOUND is written on the screen! > How can I do this - if it's possible? How about writing a program that reads the input, and then performs the initial processing, then passes to the modified command line COMMAND.COM if further processing by the command processor is needed ? Aside from this technique, you can't really do it w/o patching the command processor in memory -- which is ugly and non-portable, since it would heavily depend on the version of DOS you've got. Command lines and other strings for most of the DOS services offered in versions 2.0 and later are in the form of ASCIIZ strings -- zero-byte terminated character sequences. > b) When I in TP (v. 5.0) write sub-routines to certain hardware > interrupts, like INT08 to redifine caracters written on to the > screen, strange things start to happen. So far the only way I > have been able to do this, is through assembly language, where > I never have any problems. I have often written subroutines in > TP to other BIOS-routines, like INT09 and INT 1C, without any > trouble. If anyone think they know, what I am doing wrong, > please let me know. Do you make sure that you send a EOI to the interrupt processor ? Make sure that your processing doesn't take more than 1/18.2 seconds, because that's the frequency of the clock interrupt. Make sure that you also don't take too long in your INT 1C processing routine, as this routine would be called in the system's INT 08 routine, before it sends a EOI to the interrupt processor. Another thing to watch out for is making inadvertent calls to DOS routines -- DOS isn't reentrant, so you have to make sure that you don't accidentally call upon DOS services when DOS is trying to process a request (this would be quite probable if you are writing interrupt service routines for hardware interrupt). > >FROM: Erik F. Andersen > efa@iesd.auc.dk //-n-\\ Naoto Kimura _____---=======---_____ (abcscnuk@csuna.csun.edu) ====____\ /.. ..\ /____==== // ---\__O__/--- \\ Enterprise... Surrender or we'll \_\ /_/ send back your *&^$% tribbles !!