[Bug 17902] [830M missing dvo driver] need support for DVO-LVDS via na2501

[Daniel Vetter]

On Tue, May 29, 2012 at 9:13 PM,  <bugzilla-daemon at freedesktop.org> wrote:
> --- Comment #75 from thor at math.tu-berlin.de 2012-05-29 12:13:12 PDT ---
> Well, I do have a documentation from National Semiconductors, but there is
> nothing that would define frequencies or bandwidths - only the sync pulses can
> be set, and intput to syncs. So I played writing values into all other
> registers with i2cset, but with little success - not any change, actually,
> except for the mentioned register. Ok, and you could block off the screen as
> well by disabling the sync pulses (not much of a surprise, I guess).

Well, most lvds panels have a fixed mode they work at. We fake
different resolutions by using the hw scaler on the gpu's scanout
unit. Still, having a dvo chip without any means to adjust that sounds
a bit strange.

> However, leaving all that aside - I'm no longer able to talk to the chip, upon
> rebooting i2cdetect gave me all blanks on bus #6. I really wonder what is going
> on. Could it be that there is some additional "gating" going on that blocks off
> access to this chip unless another register is set? Is that typical?

We do have some parts of the mmio space that need unlocking, but i2c
devices should always respond. You might need to power cycle the
machine to get it out of limbo though (i.e. yank the battery).
-- 
Daniel Vetter
daniel.vetter at ffwll.ch - +41 (0) 79 364 57 48 - http://blog.ffwll.ch