Regarding testing on AmigaOS, I have so far tested the X-Surf 100 (with x-surf-100.device) on AmigaOS3.9, AmigaOS4 and AmigaOS4.1 with successful result.
I tried your suggestion of patching ariadne_ii.device 43.12 to use the X-Surf 100 autoconfig ids (changed "7265 203c 0000 0877 d281" at offset 2E72 to "7232 203c 0000 1212 d281" as suggested). However, upon opening this patched device the amiga locks up both under AmigaOS3.9 and MorphOS, so perhaps the registers begin at a different offset and have a slightly different layout or spacing?
I also tried patching the ariadne_ii.device with the "Frank Mariak xsurfpatch -> x-surf.device" applied to use the X-Surf 100 autoconfig ids in a similar manner. I found sortof the same code, patched in another way there (the last ADD.L was replace with a MOVE.Q with the product value, Code:
MOVEQ #$65, D1
MOVE.L #$1212, D0
MOVEQ #$17, D1
Which I changed to Code:
MOVEQ #$65, D1
MOVE.L #$1212, D0
MOVEQ #100, D1
As hex: Changed "7265 203c 0000 1212 7217" at offset 1bcc to "7265 203c 0000 1212 7264"
This didn't work either, however I am not sure the patch-attempt I did was correct as I don't get consistent behaviour between AmigaOS3.9 and MorphOS with this one. Maybe there is another check in the code for the manufacturer/id. The result is that in AmigaOS3.9 it locks up the amiga, but under MorphOS it fails opening the device.
Thanks for reaching out to people trying to help me out, very appreciated! My german is not strong, do I understand him correctly if he suspects something with the storage for the interrupt routine?
If so, maybe something is perhaps not 100% right which it, but it gets away with under AmigaOS, like not allocating enough space? Just started thinking of the scsi.device bug with TLSFMem :D.
[ Edited by patrik 13.12.2015 - 00:10 ]