Biomenace works (or at least used to - I did not test it recently) when a small patch made by Finster is run before the game. It was available only on old private betaboard, therefore I upload it here.

Finster comments:

hell, those games are buggy... they do some weird things with the interrupt table (comparing things that dont make any sense). I have attached a small patch which has to be started in dosbox before one of those games are executed. it's a simple fix writing the value 0000 at the memory address 0000:0006. Works with all three bio menaces. stupid games...

Mirek

Currently there is no possibility to adjust the number of stacks in the dosbox.conf as far as I know. Correct me if I am wrong.

Alas, you are right, as DOSBox does not allow editing of config.sys, period.

I remember about asking about this on VOGONS over a YEAR ago.
It had somethign to do with a comparing something 0:0 that didn't exist and was basically a division of 0.
Biomenace is looking for some memory register that doesn't exist and should not exist but old MS-DOS didn't make any noise about it but if we put something in to fix it there would be less free memory for other games.
Correct me if I'm wrong but I do remember something like that.

Essentially I believe the patch is like a TSR *terminate stay resident* that circumvents this "issue" but we've never considered on adding it because of wasting memory.

/Ieremiou

btw: I think you or somebody else gave me that patch last year when I asked about this issue.

As I understand Finsterr's description, it is *NOT TSR* (!!!) - it just simply writes 0000 to 0000:0006 and ends.

Mirek

> Alas, you are right, as DOSBox does not allow editing of config.sys, period.

You don't need the stacks statement as dosbox doesn't
use the stack for int21. All int21 functions are pretty
much "instant" operations.

The interupt table finsterr mentions doesn't seem to
be intended, as for some reason ES gets zero (some
huge list walking is before it, looks like they search
for maybe a sprite location or so).

Works with the image support.

wd

yes. looks like a null terminated list.
we came that far as well.
never figured out what they sought.

(that list is traversed on sound effects so it seems like queue or so)

Finsterr is perfectly right, when walking the linked
list they don't check for 0 explicitly resulting in
odd checks against the int table.
These checks "work" with dos as the values are below
0x8000, but dosbox has callbacks at 0xc800 which lead
to different code paths.
The fix finsterr provides handles this.

Sorry for confusion 😉
wd