I wasn't getting any output on a monitor connected to an AM625 BeaglePlay board, and I tasked a Cursor agent based on Claude Opus 4.6 to conduct a root cause analysis.
It correctly figured out that it was due to the IT66121 bridge assuming that the sink type was HDMI and sending AVI infoframes, but this can confuse DVI monitors. This patch series fixes the issue, the changes are fairly trivial and I take full responsibility of the contribution. An Assisted-by tag is added to the patches in conformance with the AI Coding Assistants [0] policy. Patch #1 is just a cleanup to avoid doing direct calls to container_of() and use a helper function which is the convention in most drivers. Patch by storing the connector sink type in the bridge state, to determine whether the mode should be HDMI or DVI. The patches were tested both using a DVI monitor and a HDMI-to-DVI adapter and a HDMI monitor, to ensure that there are no regressions for HDMI mode. [0]: https://docs.kernel.org/process/coding-assistants.html Javier Martinez Canillas (3): drm/bridge: it66121: Add bridge_to_it66121_ctx() helper drm/bridge: it66121: Add bridge-private atomic state drm/bridge: it66121: Select HDMI or DVI mode based on sink type drivers/gpu/drm/bridge/ite-it66121.c | 137 +++++++++++++++++++-------- 1 file changed, 100 insertions(+), 37 deletions(-) -- 2.54.0 base-commit: 19d584a634fe999786acfb0ac5289710cc84a5f6 branch: drm-misc-next
