Jonathan Campbell

Simple question: If you disable the emulator INT 33h, enable PS/2 AUX or serial mouse emulation and load a mouse driver TSR instead, does it crash?

Actually __dpmi_yield() makes a specific INT 2Fh call. https://www.delorie.com/djgpp/doc/libc/libc_279.html First check: Can you see if the INT 2Fh real mode vector is non-NULL (not 0000:0000)? Second check: Is there anything...

One function to look at might be DOS_MultiplexFunctions which has the majority of INT 2Fh handling code. Anything else that calls DOS_AddMultiplexHandler should be examined as well.

Testing with DOSBox-X on Mac OS, m_broken doesn't appear to segfault at all unless I move the cursor to any edge of the screen, but it takes a fair amount...

I then tried attaching a Logitch mouse (the gamer type) that I know has a high report rate and tried again. m_works doesn't have any issues, m_broken only segfaults if...

Which leads me to another question: How much stack does your program allocate for itself? Perhaps there is something with Allegro that uses just enough stack that the additional stack...

I just had another thought, because this might not be obvious if you're used to modern OSes: Does your INT 33h mouse handler reside in locked memory? I mean page...

Not only do newer cards add XGA commands but some bitfield widths change slightly as well. S3 virge completely changes most of the XGA registers and commands. You may borrow...

My collection of S3 datasheets might also help here. http://hackipedia.org/browse.cgi/Computer/Platform/PC%2c%20IBM%20compatible/Video/VGA/SVGA/S3%20Graphics%2c%20Ltd

I'm using the S3 864 drivers from the Vogons driver library and I know of one case where I'm not certain whether the driver or DOSBox-X is at fault here....