Re: [PATCH v11 00/21] Synopsys DisplayPort Controller improvements for Rockchip platforms

[Igor Paunovic <royalnet026@gmail.com>]

Hi Sebastian,

The Tested-by tag is at the bottom.  Short version: v11 works here on
an Orange Pi 5 Plus over USB-C DP AltMode.  It needs your usbdp v13
series underneath to probe at all, and above 2560x1440@144 it needs a
VOP2 ACLK change that has nothing to do with your patches.

One caveat first, so that the next paragraph is not misread: dp0 is
disabled in the mainline RK3588 DTS, so none of this is reachable on
an untouched tree.  I enable it with a local board DT patch.

With dp0 enabled but without usbdp v13, dw-dp never finishes probing.
It stops at "failed to get follow-up bridge" (dw-dp.c), because
mainline phy-rockchip-usbdp has no drm_aux_bridge_register() at all;
that call arrives with usbdp v13 patch 16/35 ("phy: rockchip: usbdp:
Register DP aux bridge").  The component bind then never completes, so
both HDMI outputs stay dark as well - the symptom is much wider than
the cause, which cost me a boot to work out.

Hardware: Samsung Odyssey G70B behind a USB-C to HDMI 2.1 protocol
converter.  DPCD extended revision 1.4, the link trains at HBR2 x4,
VSC SDP colorimetry is supported.  The converter ships a malformed
EDID, so a boot script feeds a corrected one through the debugfs
edid_override and kicks fusb302 once; without that the 4K modes are
dropped and AUX stays dead.  Both are adapter problems, not driver
ones.

What I saw:

  - DP-1 comes up connected, the EDID reads through the converter,
    the link trains, and KWin drives the output.
  - At the stock VOP2 ACLK rate, modes up to 2560x1440@144 are clean.
    I left the machine at 2560x1440@143.99 for more than eight hours
    with no DRM or VOP2 errors.
  - 3840x2160@120 comes up, but the picture is unusable until the
    VOP2 ACLK is raised.  With it raised the picture is correct
    immediately.

That last point is a platform issue rather than a dw-dp one, so I have
sent it as its own thread to dri-devel and linux-rockchip with you on
Cc: "drm/rockchip: vop2: ACLK_VOP pinned at 500 MHz starves DP 4K120
on RK3588".

Two small things I noticed while I was in there:

Kconfig: ROCKCHIP_DW_DP selects nothing, and DRM_DW_DP does not select
DRM_BRIDGE_CONNECTOR either, while ROCKCHIP_ANALOGIX_DP,
ROCKCHIP_CDN_DP, ROCKCHIP_DW_HDMI_QP, ROCKCHIP_LVDS and ROCKCHIP_RGB
all do.  Since dw_dp-rockchip.c calls drm_bridge_connector_init(), a
configuration with only ROCKCHIP_DW_DP enabled fails to link.  This
predates your series; I am happy to send the one-liner if you agree it
is worth fixing.

sound-dai-cells: the dt-bindings patch in this series ("dt-bindings:
display: rockchip: dw-dp: Fix sound DAI cells") notes that the
mainline rk3588 DTS still sets 0 and will be fixed up separately.  Is
that DTS patch already queued somewhere?  I would rather not duplicate
it.

DP audio: it does work here, but on my rockchip-devel based daily
kernel, which carries the v9 generation of the audio patch - a
simple-audio-card with cpu &spdif_tx2 and codec <&dp0 1>, picked up by
PipeWire unaided, playing over the same cable as the video.  I have
not retested audio on the v11 tree, so the tag below covers video
only.

Tested on drm-misc-next of 2026-08-07 (dc2f9f7fed1a) with this series,
usbdp v13, two local board DT patches (enable DP AltMode on the USB-C
connector; pin DCLK_VOP2 to V0PLL), and a one-line local change to
clk.c so that I could set clock rates from debugfs.

Thanks for the series.

Tested-by: Igor Paunovic <royalnet026@gmail.com>