Hi,
On Fri, Sep 08, 2017 at 03:50:08PM +0800, Chen-Yu Tsai wrote:
Hi everyone,
Previously I added support for two display pipelines for the sun4i-drm driver. At the time we did not have support for downstream encoders to actually test the second pipeline, nor having both pipelines active at the same time.
With HDMI encoder support now merged and support for it on sun6i in progress, we are now able to test both pipelines being active at the same time, with LCD output from one and HDMI from the other. Testing has revealed some more issues, such as component probing order. Also, we found out that muxing between the display backend and TCON was possible, and required for the second pipeline to work as intended. Last, the number of CRTCs provided to drm_vblank_init needs to be increased for vblanks to work properly.
This patch series does a few things:
Fix endpoint IDs so they conform to what the DT binding stipulates.
Change the order the individual components are queued up.
Fix how the TCON gets its ID when the backend-TCON mux is present.
Support the input mux in the TCON (which selects which backend to use on the A31/A31s).
Fix the drm_vblank_init call and pass the correct number of crtcs.
Add the cross pipeline connections between the DRCs and TCONs on the A31, conforming to the DT binding and what the hardware is capable of.
More details are available in each individual commit. Please have a look.
I've applied all the commits, with: patch 1 in our arm-soc fixes branch patch 2-7 in drm-misc-next patch 8 in sunxi/dt-for-4.15.
I have a minor comment on one patch that can be fixed in a followup patch, so please have a look at my other mail.
Thanks for this great work, Maxime