On Wed, May 08, 2013 at 01:07:23PM -0300, Paulo Zanoni wrote: > I also looked briefly to your i-g-t patches. They look fine, but my > concern is that the code inside the Kernel will get out-of-sync with > the code in i-g-t, so we won't really be able to catch regressions. > OTOH, I do have an idea for a different i-g-t test: you read our > registers (WRPLL_1, WRPLL_2, Link M) in order to check which > frequency/r/n/p we're using inside the Kernel, then you look at the > old wrpll_tmds_clock_table and check if the values used on the Kernel > match the values on the table. Maybe this should be incorporated at > testdisplay. Maybe my idea is just a bad idea and should be ignored. > Feel free to do whatever you prefer for the i-g-t patches. Yes, I've had the same thoughts about this as well. I was even pondering pulling enough code in a "kernel compat" library/headers to be able to compile intel_ddi.c and sync it with the kernel one, once in a while. Crazy. I like your idea of adding the wrpll test to testdisplay, not volunteering for that at the moment though :) -- Damien
