Yeah I noticed this as well. I think I nopped the jne 😀
I spend some time on it a while ago and wasn't able who would lower the counter placed there.
The counter is said there just before the dma transfer is started
(set a bpm on the memory location and you will see)

(Sorry, I could not send you email, so I ask here...)
Hi
I read a post titled 'HGC(Hercules Monochrome) WIP report'.
There I saw some source code fragment taiken7 posted,
but I still do not know what to do, how to do.
I really hope to play oldies which only support Hercules Graphic Card.

So I have two questions...

1. When I run the program, it checks my graphic card.
Yes, my graphic card is SVGA.
How can I cheat the program as if I have HGC instead of SVGA ?

2. How can I access video memory ?
Frankly speaking, I don't know much about DOS game programming.
I only know 80x86 instructions and some DOS interrupts.
So please introduce me some good books or good websites 😀
(I went to bookstore, but there was no book about obsolete
DOS game programming... I really want to get some information
about HGC, CGA, DMA - Sound in DOS.)

And I will really appreciate it if you let me get DOSBox version
in which I can play hercules games...

Bye

the next version of dosbox will have a configfile option which is called machine
there you can set it lower then vga.

(NOPing, now why didnt I think of that .. ! .. Thats an even better
solution than I was using since it only requires 2 debugs
(Intro / Ingame) as compared to per sample.
Nice one Qbix .. )

Hmm.. after further analysis it appears the PIC is masked
just before running the sample.

I get the impression what the program has done is set some
sort of watchdog timer interrupt. (Which returns CX=0).
Dosbox might be missing the timeslice OR its taking a really long time.

But thats as far as I can take things at the moment,
..hopefully Fizzban will take a look..

Loop seems wrong. Instruction above it moves data from 17A8:3E95 to CX.
That place in memory is set to 1 while sound is playing and when sound
blaster runs IRQ, it's set to 0.
CX is pushed at the beginning and popped at the end of interrupt, with a value 1 (i.e it is unchanged). Then interrupt ends and returns to 035b:91c9 (JNE $-4 instruction in a loop).

Either interrupt returns to wrong instruction (wrong eip) or something is wrong with that JNE.
Maybe I'll look what exactly happens in DOS if I can get SBLive to work with the game.

The process hangs independent of the output medium.

So you could run it from PC speaker and it will still authentically crash.
(I think you're right about the return address being incorrect, if it were
just 4 bytes earlier then it would "set CX=0"..);

There are 2 seperate but identical code segments matching the
loop (one for the intro, one for in-game "phone calls"/"Nice shooting")
.. so it seems consistent. Wrong, but consistent.

might also possibly be problem in i386 decode instruction buffer -
that JNE ($-4) might be $ relative to before the instruction.. I will have
to look it up...

No, it uses sound blaster for digtized sound even if pc speaker is used for the rest.
At least in my version. If I disable sblaster in dosbox.conf game doesn't hang.
Unfortunately, I couldn't get digitized sound in DOS, so I can't even check if it runs properly there.

Btw., to get it working properly, find string 0B CD 75 FC in both shell.exe and gb.exe and replace FC with F8. (It is that JNE's operand 😀

edit:
I tested it with vdmsound; game has same problem.