Sebastian Hesselbarth sebastian.hesselbarth@gmail.com wrote on Wed [2013-Jul-31 22:21:20 +0200]:
Darren,
I now fully understand the issues of AM335x's LCD controller and your fix for it. I suggest to clarify the comments you added to tilcdc to allow others to understand it more quickly.
Sebastian, Thanks for looking at my proposed changes, you understand this sync stuff very well so I appreciate your input that this is actually an acceptable workaround.
Actually, the LCD controller always aligns vsync to the second edge of hsync, which will never give VESA-compliant sync. The (elegant) workaround you are proposing is to align both rising edges, so at least TDA998x can sync on those with some hskew added. Lucky you that it ignores hsync length but only looks for rising HS/VS edges ;)
Yes we definitely got lucky with this one, good thing the NXP supported that reference pixel position, as I was out of options from the lcd controller side of things to adjust the horizontal position.
Should we prepare a new patch set comprising the following patches?
Russell King: drm/i2c: nxp-tda998x: fix EDID reading on TDA19988 devices drm/i2c: nxp-tda998x: ensure VIP output mux is properly set drm/i2c: nxp-tda998x: fix npix/nline programming drm/i2c: nxp-tda998x: prepare for video input configuration drm/i2c: nxp-tda998x: add video and audio input configuration
Sebastian Hesselbarth: drm/i2c: tda998x: fix sync generation and calculation
Darren Etheridge: drm/i2c/tda998x prepare for tilcdc sync workaround drm/tilcdc fixup mode to workaound sync for tda998x
Or do we keep them separated and possibly resend them if David cannot find them anymore?
I vote we submit a complete series that we can all test, there were quite a lot of versions of things in flight at the same time so I am sure David would appreciate a consolidated version. The only thing I have not tested is audio support, but as the original driver did not have that anyway I don't consider it blocking if it is working for CuBox.
Darren