Dear John,
In message 4b73d43f0910191545x3127cba5w7fdec3f6382138e3@mail.gmail.com you wrote:
The register map for 5125 does not just change the size of the registers. Some registers change locations. The issue is that the hardware guys decided to "fix" the old broken register access. The 5200, 5121, 5123 had some registers that were:
I always stand fascinated about the inventiveness of these guys; even when just releasing a new chip from one family where one would expect basicly upward-compatibility they find ways not to simplify the design but to make it more complex and wonderful. Nobody else does so much to save our jobs.
So the problem is painful but I believe doable. The problem I never resolved was dealing with this mess in linux where the same binary has to work with both platforms. I decided that the register accesses needed to be done via an offsets array that was populated at run time but I never got around to implementing that.
Heh. I don't envy the guy who has to do this.
Best regards,
Wolfgang Denk