The cdclock sanitization patch reviewed and merged at - http://patchwork.freedesktop.org/patch/msgid/1445344992-14658-1-git-send-email-shobhit.kumar@xxxxxxxxx made the assumptions that DPLL should not be enabled when pre-os does not enable display and if it does then verify that the cdclock is corectly programmed as well. The BIOS was actually enabling DPLL as well while not following BSPEC sequence and writing cdclk register directly. I was working with BIOS team to correct this and found that due to a WA needed where audio codec will not be enumerated in OS if BIOS did not program the audio verbs which needed PG2 and DPLL enabling. More discussion revealed the following logic - 1. BIOS puts max cdclk for the platform in CDCLK_CTL. VBIOS/GOP reads that value and then programs cdclk to desired value. 2. It also then sets SWF18 to indicate to the OS that it has enabled display. Used for fastmodeset actually in windows. 3. It also sets SWF06 with this max cdclock(from what bios programmed in CDCLK_CTL) for OS to know. This patch uses point 2 above while sanitizing the cdclk. We can also update our logic for deciding max cdclock based on SWF06 if pre-os enables else directly from CDCLK_CTL (no pre-os display). That is not part of this patch. Shobhit Kumar (1): drm/i915/skl: While sanitizing cdclock check the SWF18 as well drivers/gpu/drm/i915/i915_reg.h | 3 +++ drivers/gpu/drm/i915/intel_display.c | 8 ++++++++ 2 files changed, 11 insertions(+) -- 2.4.3 _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
