First post, by Kahenraz
- Rank
- l33t
I've been having problems where I would encounter strange behavior and bad opcode errors. I managed to solve this issue by removing xhdd.sys from my config.sys.
The issue appears to be caused by XHDD, which is excellent when UDMA is available. But I think there is a bug when this feature is not present; in this case, when I have only a PIO4 disk connected.
I tested this on both an Intel and SiS IDE controller which supports ATA33 in BIOS. When the PIO4 disk is connected, XHDD reports an "I.D./mode error" but I think it has already set itself up with a cache. The cache provides a performance boost when observing burst transfers but provides no benefits for sustained reads and writes, as expected. I don't know what the issue is specifically but it is 100% reproducible when loading Doom.
My config.sys for testing is very basic and loads himem.sys but not emm386. There are no other memory managers present. The system is unstable even if I swap himem for himemx. I have tested and reproduced the same issue on a SiS chipset Socket 7 with an AMD processor and an Intel chipset Slot 1 with a Pentium 2.
The lengthy crash method which reports and invalid opcode is output from killer.exe; the system will instead just hard lock without this TSR installed or in some cases even with it loaded.
Has anyone else had any problem similar to this with XHDD?