On Mon, 2016-03-14 at 12:51 +0100, Maarten Lankhorst wrote:
> Op 14-03-16 om 09:55 schreef Ander Conselvan de Oliveira:
> > Remove the clock calculation from ironlake_crtc_compute_clock() when the
> > encoder compute_config() already set one. The value was just thrown away
> > in that case.
> > 
> > Signed-off-by: Ander Conselvan de Oliveira <
> > ander.conselvan.de.oliveira@intel.com>
> > 
> It was thrown away, but it could still reject based on the limits, which this
> patch changes.
> This might be made more clear in the commit message.

Good point. To be honest, I didn't very this as carefully as I should have
before sending and missed that detail. It turns out that change is safe. To
verify I extracted the relevant code and run it with all possible port clocks we
could have with either the sdvo or the dp encoder setting the clock. See the
attached C file. I was too lazy to actually understand what the
g4x_find_best_dpll() does.

Anyway, I'll send another version with a note about this.

Thanks,
Ander