DS2x86 progress

April 2nd, 2012, 00:29 Posted By: wraggster

via http://dsx86.patrickaalto.com/DSblog.html

[h=4]DS2x86 runs Windows 95![/h]Well, not really, but that was too good a title to pass by on April Fool's Day. :-) What I am actually working on, is support for Windows 3.11. The problem with it in the current DS2x86 was that it crashed immediately when switching to protected mode. This was caused by it storing the Global Descriptor Table into virtual memory before switching to protected mode. I did not have support for that in DS2x86. It is actually a bit of a chicken and egg situation, as virtual memory needs the GDT table to be correct for page fault handling. Windows 3.11 actually accesses virtual memory using the page tables from real mode before going to protected mode and activating paging! Adding support for such allowed it to continue to protected mode correctly.
After fixing that problem, I needed to improve pushad opcode behaviour, add better support for accessing CPU debug registers, add support for reading secondary DMA controller ports, etc. All of these were pretty easy and straghtforward, so it was nice to see some real progress without having to hunt for bugs for a change. After those I did run into a bigger unsupported issue, namely Virtual 8086 mode. Windows 3.11 seems to use VM86 mode when calling some real-mode interrupt vectors from protected mode, and until now I have not had any support for VM86. Or to be more precise, I have checked that if the processor is supposed to be in VM86 mode, I simply drop into the debugger. However, to support Windows 3.11, I need to add proper support for VM86 mode.
So, for the last couple of days I have been working on adding the VM86 mode handling into DS2x86. Many opcodes need only minor changes, but the actual going into VM86 mode and back using the IRET opcode is a bit complex, so adding support for that will still take some time. I have also found some weird behaviour in Windows 3.11 on DS2x86, so there seem to be something else besides the VM86 mode also missing on broken. So, I don't expect Windows 3.11 working in the next version yet, but we shall see.
I have also checked Alien Legacy, which seems to loop displaying the sound selection menu. I originally thought the problem is in the key input routine, but it seems that the keyboard interrupt INT 16 is never even called by Alien Legacy in DS2x86, even though it is called in DOSBox, so this needs some further studying and debugging. I hope to get that at least working in the next version.
Thanks for the debug logs and error reports again, I plan to fix some issues in those as well for the next version.

There are 0 comments - Join In and Discuss Here