Re: [PATCH v5 00/15] drm/tidss: Add BeagleY-AI display support (and some more)

From: Tomi Valkeinen

Date: Tue Nov 17 2026 - 02:54:59 EST


Hi,

On 03/11/2026 13:34, Tomi Valkeinen wrote:
This series aims to add display support for all display interfaces on
BeagleY-AI board. More generally, it adds display support for TI AM62P,
J722S, and related SoCs.

The main topics to highlight:

- The data and sync signal polarities require accessing a register
(DPI0_CLK_CTRL) in the CTRL_MMR block. This has already been discussed
e.g. in
https://lore.kernel.org/all/20250730-fix-edge-handling-v1-0-1bdfb3fe7922%40bootlin.com/.
This topic is still open, and it also touches already upstreamed SoCs
with DSS (AM625). In other words, with this series, we don't have full
polarity support, and DPI works exactly like upstream AM625's does.

- In the SoC, the two OLDI TX instances are shared between the two DSS
instances. This has been a bit tricky to solve neatly. In this series
the result is that in the DT we have three OLDI TX nodes, two under
DSS0, one under DSS1, representing the OLDIs that the DSS instance
can use.

- I'm no audio expert, but based on the original dts from Robert Nelson,
I did get HDMI audio working, but I see "davinci-mcasp
2b10000.audio-controller: Sample-rate is off by 400 PPM".

With this series one can use the HDMI output on BeagleY-AI. I have also
tested the DSI output with Raspberry Pi's 7" DSI display, and OLDI
output with an oscilloscope (I don't have a suitable OLDI panel) with
extra DT overlays.

Tomi

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@xxxxxxxxxxxxxxxx>
---
Changes in v5:
- Add Rb tags from Aradhya
- New cleanup patch: "drm/tidss: Use the AM625 features for AM62A7"
- Ensure tidss_oldi_create_pd_devs() is only called for primary OLDI
- Add missing cbass ranges for DSS1 and DSI (DSS was already there) to k3-am62p.dtsi
- Link to v4: https://patch.msgid.link/20261029-beagley-ai-display-v4-0-dc7f2b532478@xxxxxxxxxxxxxxxx

Changes in v4:
- Based on today's drm-misc-next
- Dropped the CTRL MMR related code. This means the DPI polarities are
not quite right (the same as AM625)
- Dropped the commit to convert OLDI to an aux driver. This version uses
a new (unmerged) PM dev_pm_domain_attach_by_node() to manage the OLDI
power domains.
- Dropped a bunch of commits which were merged via "[PATCH 00/10]
drm/tidss: Preparation for AM62P and BeagleY-AI display support"
series.
- Dropped tested-by tags, as the series has changed quite a bit.
- Includes new fixes and cleanup patches:
- vp_bus_type removal:
- "Add helper to identify the AM65x OLDI video port"
- "Stop using vp_bus_type for directly connected panels"
- "Drop the tied off video port check from mode validation"
- "Remove the unused vp_bus_type feat"
- OLDI:
- "Drop unnecessary forward declaration"
- "Fix wrong return value from tidss_oldi_mode_valid()"
- "Fix serial clock reference leak"
- "Fix cleanup on a failed init"
- "Move the setup of one OLDI TX to its own function"
- "Add power domain support"
- Link to v3: https://patch.msgid.link/20260529-beagley-ai-display-v3-0-7fefdc5d1adf@xxxxxxxxxxxxxxxx

Changes in v3:
- Base on "[PATCH 0/6] arm64: ti: Use syscon for the Control Module"
- Drop quotation marks from Rb tags
- Drop Tested-by tags for the bindings patches
- "drm/tidss: Add external data and sync signal edge configuration": Fix
error handling
- "drm/tidss: Add support for DPIENABLE bit": set/unset DPIENABLE in
prepare/unprepare
- "drm/tidss: oldi: Convert OLDI to an aux driver": Improve cleanups
(of_node_put, clk_put, device_link_del)
- Dropped "arm64: dts: ti: k3-am62p-j722s-common-main: Make main_conf a syscon"
- "arm64: dts: ti: k3-am62p-j722s-common-main: Add support for DSS": Add
missing clocks and ti,oldi-io-ctrl for oldi1 node. Drop unused
main-gpio0-ioexp-intr-default-pins node. Fix MCASP1_AXR0 pin
direction. Fix DSS videoport numbering.
- Link to v2: https://lore.kernel.org/r/20260513-beagley-ai-display-v2-0-9e9bcefde6bc@xxxxxxxxxxxxxxxx

Changes in v2:
- Add Rb and Tb tags
- Drop "dt-bindings: mfd: syscon: Add ti,am625-dss-dpi0-clk-ctrl
compatible", which is now implemented with the below changes.
- New patch "dt-bindings: display: ti,am65x-dss: Add ti,dpi-io-ctrl"
- Modify "drm/tidss: Add external data and sync signal edge
configuration" to use the 'ti,dpi-io-ctrl' with offset.
- New patch "arm64: dts: ti: k3-am62p-j722s-common-main: Make main_conf
a syscon".
- Modify "arm64: dts: ti: k3-am62p-j722s-common-main: Add support for
DSS" to set 'ti,dpi-io-ctrl' properties
- Drop generic desc in DT bindings for power-domains property
- Fix pm_runtime_get_sync() return value check
- Link to v1: https://lore.kernel.org/r/20260420-beagley-ai-display-v1-0-f628543dfd14@xxxxxxxxxxxxxxxx

---
Andrew Davis (1):
arm64: dts: ti: beagley-ai: Enable HDMI display and audio

Jayesh Choudhary (1):
arm64: dts: ti: k3-am62p-j722s-common-main: Add support for DSS

Swamil Jain (1):
drm/tidss: Add support for AM62P display subsystem

Tomi Valkeinen (12):
dt-bindings: display: ti,am65x-dss: Add AM62P DSS
drm/tidss: Add helper to identify the AM65x OLDI video port
drm/tidss: Stop using vp_bus_type for directly connected panels
drm/tidss: Drop the tied off video port check from mode validation
drm/tidss: Remove the unused vp_bus_type feat
drm/tidss: Use the AM625 features for AM62A7
drm/tidss: oldi: Drop unnecessary forward declaration
drm/tidss: oldi: Fix wrong return value from tidss_oldi_mode_valid()
drm/tidss: oldi: Fix serial clock reference leak
drm/tidss: oldi: Fix cleanup on a failed init
drm/tidss: oldi: Move the setup of one OLDI TX to its own function
drm/tidss: oldi: Add power domain support

I'll pick patches 2-11 (inclusive) to drm-misc-next. They stand alone and do not depend on the PD change we need for the oldi.

Tomi