All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 0/2] drm/msm/dp: make the eDP 1.4 LINK_RATE_SET path work
@ 2026-11-20 19:20 Jesse Casco
  2026-11-20 19:20 ` [PATCH 1/2] drm/msm/dp: carry the eDP rate-set selection into the link parameters Jesse Casco
  2026-11-20 19:20 ` [PATCH 2/2] drm/msm/dp: stop the rate-set fallback at the lowest table entry Jesse Casco
  0 siblings, 2 replies; 3+ messages in thread
From: Jesse Casco @ 2026-11-20 19:20 UTC (permalink / raw)
  To: Rob Clark, Dmitry Baryshkov
  Cc: Abhinav Kumar, Jessica Zhang, Sean Paul, Marijn Suijten,
	David Airlie, Simona Vetter, Jim Martin, linux-arm-msm, dri-devel,
	freedreno

The eDP 1.4 link-rate table support picks a LINK_RATE_SET index in
dp_panel, but the selection never reaches link training: only rate and
num_lanes are copied into link_params, so LINK_RATE_SET is never
written and the rate-set fallback branch can't run. Patch 1 carries
the fields through. Patch 2 fixes the fallback once it is reachable,
so it stops at the lowest table entry instead of dropping to a standard
rate the sink was not told about.

I found this while bringing up an ASUS Zenbook A16 (Glymur), whose eDP
1.4 panel lists 1.62, 2.7 and 5.4 Gbps in SUPPORTED_LINK_RATES. At boot
the firmware has left LINK_BW_SET at 0x14, so dp_panel picks the
LINK_BW_SET path. When the compositor starts and the link caps are read
again, LINK_BW_SET reads 0 and dp_panel selects the rate table, but
without patch 1 that selection never reaches link training. With the
series the link trains through LINK_RATE_SET (5.4 Gbps, 4 lanes).

With a test-only change that fails the first clock-recovery attempt,
the fallback moves to index 1 and the panel trains at 2.7 Gbps,
reading back LINK_RATE_SET=0x01. The lowest entry of this panel's table
is RBR, so patch 2 does not change behavior here.

I tested on the A16 with a next-20261103-based kernel; the patches
apply unchanged to next-20261119. They touch dp_ctrl.c alongside
"drm/msm/dp: don't fall back to a link that can't carry the eDP
mode" [1], and the two apply in either order.

Jim, Cc'ing you as mentioned off-list: Omarchy's arm64 kernel carries
patch 1 as 2367 with your 2368-2371 on top, and your 2369 covers what
patch 2 does. I'm happy to drop patch 2 if you post yours.

[1] https://lore.kernel.org/r/20261115000021.78869-1-jesse.casco@gmail.com

Jesse Casco (2):
  drm/msm/dp: carry the eDP rate-set selection into the link parameters
  drm/msm/dp: stop the rate-set fallback at the lowest table entry

 drivers/gpu/drm/msm/dp/dp_ctrl.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)


base-commit: adcad7f4e6332cff3d6d7165f41584899f598534
-- 
2.56.1


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2026-11-20 19:20 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-20 19:20 [PATCH 0/2] drm/msm/dp: make the eDP 1.4 LINK_RATE_SET path work Jesse Casco
2026-11-20 19:20 ` [PATCH 1/2] drm/msm/dp: carry the eDP rate-set selection into the link parameters Jesse Casco
2026-11-20 19:20 ` [PATCH 2/2] drm/msm/dp: stop the rate-set fallback at the lowest table entry Jesse Casco

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.