On 18/06/2019 21:08, John Harrison wrote:
Tvrtko, does this look plausible?
It seems to work for me in that it passes on ICL with the new read-only
registers. I'm not sure if there is a valid way to detect whether the
registers are actually readable though. How would the test know what is
a valid value? If one assumes that one gets back zero from an invalid
read, how does one know that the read-only register is not supposed to
return zero at this point anyway?
Or is it worth just putting in a test for non-zero and if we do find a
register that can validly return zero then we special case that one and
ignore it?
I was thinking we just read the register and then verify it is unchanged
after every existing write to it.
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
