[Intel-gfx] [PATCH] drm/i915: force bpp for eDP panels

[Daniel Vetter]

We've had our fair share of woes already which showed that we can't
rely on the bpc limits in the EDID for eDP panels without risking
black screens. So now we limit the depth by what the BIOS recommends
in the VBT:

commit 2f4f649a69a9eb51f6e98130e19dd90a260a4145
Author: Jani Nikula <jani.nikula at intel.com>
Date:   Mon Nov 12 14:33:44 2012 +0200

    drm/i915: do not ignore eDP bpc settings from vbt

But that's not enough, since at least the panel on my ASUS Zenbook
Prime here is also unhappy if the bpc is too low. Hence just take the
firmware value and dither to get what flimsy panels want.

Like before we ensure that we don't change the bpp if the firmware
doesn't provide a value, see

commit 9a30a61f3516871c5c638fd7c025fbaa11ddf7fe
Author: Jani Nikula <jani.nikula at intel.com>
Date:   Mon Nov 12 14:33:45 2012 +0200

    drm/i915: do not default to 18 bpp for eDP if missing from VBT

v2: Apparently there are some horribly broken eDP panels around which
only work if the DP link is set up as if we want to driver a 24bpp
mode, but still only work if the data is feed at 18bpp. See

commit 57c219633275c7e7413f8bc7be250dc092887458
Author: Daniel Vetter <daniel.vetter at ffwll.ch>
Date:   Thu Apr 4 17:19:37 2013 +0200

    drm/i915: revert eDP bpp clamping code changes

for the gory details.

Adjust the patch accordingly and update all the relevant comments.

Cc: Jani Nikula <jani.nikula at intel.com>
Cc: Paulo Zanoni <przanoni at gmail.com>
Reviewed-by: Jani Nikula <jani.nikula at intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter at ffwll.ch>
---
 drivers/gpu/drm/i915/intel_dp.c | 26 +++++++++++++++++++++-----
 1 file changed, 21 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
index b1a3a64..d602886 100644
--- a/drivers/gpu/drm/i915/intel_dp.c
+++ b/drivers/gpu/drm/i915/intel_dp.c
@@ -745,6 +745,25 @@ intel_dp_compute_config(struct intel_encoder *encoder,
 	/* Walk through all bpp values. Luckily they're all nicely spaced with 2
 	 * bpc in between. */
 	bpp = min_t(int, 8*3, pipe_config->pipe_bpp);
+
+	/*
+	 * eDP panels are really fickle, try to enfore the bpp the firmware
+	 * recomments. This means we'll up-dither 16bpp framebuffers on
+	 * high-depth panels.
+	 */
+	if (is_edp(intel_dp) && dev_priv->edp.bpp) {
+		DRM_DEBUG_KMS("forcing bpp for eDP panel to BIOS-provided %i\n",
+			      dev_priv->edp.bpp);
+		/*
+		 * But for added hilarity some of the 18bpp panels only work if
+		 * we set up the pipe with enough bw for a 24bpp mode, since
+		 * that's what the bios apparently does. So start the bw
+		 * calculation with at least 24 bpp (down-dither is still
+		 * possible).
+		 */
+		bpp = max_t(int, dev_priv->edp.bpp, 24);
+	}
+
 	for (; bpp >= 6*3; bpp -= 2*3) {
 		mode_rate = intel_dp_link_required(target_clock, bpp);
 
@@ -795,11 +814,8 @@ found:
 			       &pipe_config->dp_m_n);
 
 	/*
-	 * XXX: We have a strange regression where using the vbt edp bpp value
-	 * for the link bw computation results in black screens, the panel only
-	 * works when we do the computation at the usual 24bpp (but still
-	 * requires us to use 18bpp). Until that's fully debugged, stay
-	 * bug-for-bug compatible with the old code.
+	 * Since we might have picked a too high bpp for the bw calculation,
+	 * clamp it down again to make even the most fickle eDP panel happy.
 	 */
 	if (is_edp(intel_dp) && dev_priv->edp.bpp) {
 		DRM_DEBUG_KMS("clamping display bpc (was %d) to eDP (%d)\n",
-- 
1.7.11.7