Quoting Ruhl, Michael J (2020-04-17 18:14:24) > Ahh, I just remembered why I did not put the work arounds in the new > helper. > > I discovered that the intel_gmch_probe() code is in the char/agp directory, > so the path is: > > i915 set dma_mask > < gen5 set dma_mask > i915 < gen 5 work arounds > > If I put the workarounds in the set dma mask, I lose the workaround. > > Hmmm, intel_gmch_probe() has "gpu_pdev" to indicate that it is being > inited from i915. I could use this to skip the dma mask setting for this > path. > > Is that reasonable? If this is the straw the breaks the camel's back to pull in the gmch code under gt/, let me add a pile of hay! Honestly, I would set the dma masks as we think they should be here and if they are being overridden in char/agp, so be it. When we do pull the code out, it will remain correct; and overall with the refactoring you are doing the driver will be much more consistent across all gen. -Chris _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
