I pulled the EVEN/ODD BIOSes from the EEPROM chips on the motherboard, combined them, and sure enough, 819A wasn't EB, it was still 74. And E2C0 wasn't FF, it was BF. I'm not sure how this happened because I posted the correct files to the forum. There must have been some mixup during the renaming and movement to my XP machine, which does the EEPROM programming.

I re-edited the combined BIOS and split it again, then wrote the EEPROMs, and now the mouse works fine. This is terribly embarrassing. I guess sometimes the hand is quicker than the eye!

Well if nothing else I'm grateful you made the thread because otherwise I would have continued to assume that this old BIOS core had no mouse support!

Like you, I also thought the 286 BIOSes were too old, but apparently not!

Over the next week, I will be modding about ten 286-thru-486 boards for BIOS PS/2 support, so there's likely to still be some support tickets issued. I have a 386 board that uses an AWARD 4.20 BIOS. Has there been any effort to support those? If I recall correctly, MODBIN couldn't mod that BIOS.

If you link to the BIOS image I or someone else can look, especially if it can be tested under emulation.
By the way, in talking directly with maxtherabbit, it's possible there is a conflict between this and XT-IDE Universal BIOS if you run that--we suspect it's some issue between the EBDA (that steals 1 KB conventional mem) and the XT-IDE also wanting that 1KB.

jakethompson1 wrote on 2025-03-11, 03:52:

If you link to the BIOS image I or someone else can look, especially if it can be tested under emulation.
By the way, in talking directly with maxtherabbit, it's possible there is a conflict between this and XT-IDE Universal BIOS if you run that--we suspect it's some issue between the EBDA (that steals 1 KB conventional mem) and the XT-IDE also wanting that 1KB.

I will keep this information in mind. I generally do not use XT-IDE on any fixed 286/386 system, though I sometimes use it on testbeds. For fixed systems, I prefer to use SCSI, either by way of SCSI2SD, or with a CF-SCSI-ACARD adaptor.

jakethompson1 wrote on 2025-03-11, 03:52:

By the way, in talking directly with maxtherabbit, it's possible there is a conflict between this and XT-IDE Universal BIOS if you run that--we suspect it's some issue between the EBDA (that steals 1 KB conventional mem) and the XT-IDE also wanting that 1KB.

In my opinion, a confluct would mean a bug in either the mainboard BIOS or the XUB: The EBDA segment is stored at 40:06 or 40:0E, the space that would traditionally be the base address of LPT4. The end of user-available conventional memory is stored at 40:13. BIOS extensions are supposed to either enlarge the EBDA by decreasing both values, and moving the data of the original EBDA downwards, or just decrease 40:13, allocating stuff before the EBDA. In either case, 40:06 (or 0E?) keeps pointing to the EBDA required for PS/2 support, and the XUB needs to store the address of its parameters somewhere else, e.g. in the interrupt vectors 41 and 46 pointing to hard disk parameters.