Other than Broadwell being fubar (and Ironlake + g4x being special in their own way), there appears to be little fallout from enabling iommu. (The biggest open question is over performance, TLB misses are much more expensive and that impacts meda/CL/GL throughput.) Enabling iommu/dmar makes our CI much more powerful, instead of a random GPU write causing memcorruption which may or may not impact the system, we get a DMAR fault. So once and for all we will be able to ascertain whether those sporadic memcorruption are truly our fault. -Chris _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
