On Thu, 20 Oct 2011 23:18:06 -0700
Keith Packard <kei...@keithp.com> wrote:
> On Wed, 19 Oct 2011 08:12:08 -0700, Jesse Barnes <jbar...@virtuousgeek.org>
> wrote:
>
> > /* PCH eDP needs FDI, but CPU eDP does not */
> > - if (!has_edp_encoder ||
> > intel_encoder_is_pch_edp(&has_edp_encoder->base)) {
> > + if (!intel_crtc->no_pll &&
> > + (!has_edp_encoder ||
> > + intel_encoder_is_pch_edp(&has_edp_encoder->base))) {
> > I915_WRITE(PCH_FP0(pipe), fp);
> > I915_WRITE(PCH_DPLL(pipe), dpll & ~DPLL_VCO_ENABLE);
> >
> > POSTING_READ(PCH_DPLL(pipe));
> > udelay(150);
> > + } else {
> > + if (dpll == (I915_READ(PCH_DPLL(0)) & 0x7fffffff) &&
> > + fp == I915_READ(PCH_FP0(0))) {
> > + intel_crtc->use_pll_a = true;
> > + DRM_DEBUG_KMS("using pipe a dpll\n");
> > + } else if (dpll == (I915_READ(PCH_DPLL(1)) & 0x7fffffff) &&
> > + fp == I915_READ(PCH_FP0(1))) {
> > + intel_crtc->use_pll_a = false;
> > + DRM_DEBUG_KMS("using pipe b dpll\n");
> > + } else {
> > + DRM_DEBUG_KMS("no matching PLL configuration for pipe
> > 2\n");
> > + return -EINVAL;
> > + }
>
> This hunk breaks eDP nicely -- you end up in the 'else' clause still
> looking for a DPLL to use.
>
> Here's my suggested replacement:
>
> /* PCH eDP needs FDI, but CPU eDP does not */
> - if (!has_edp_encoder ||
> intel_encoder_is_pch_edp(&has_edp_encoder->base)) {
> - I915_WRITE(PCH_FP0(pipe), fp);
> - I915_WRITE(PCH_DPLL(pipe), dpll & ~DPLL_VCO_ENABLE);
> -
> - POSTING_READ(PCH_DPLL(pipe));
> - udelay(150);
> + if (!intel_crtc->no_pll) {
> + if (!has_edp_encoder ||
> + intel_encoder_is_pch_edp(&has_edp_encoder->base)) {
> + I915_WRITE(PCH_FP0(pipe), fp);
> + I915_WRITE(PCH_DPLL(pipe), dpll & ~DPLL_VCO_ENABLE);
> +
> + POSTING_READ(PCH_DPLL(pipe));
> + udelay(150);
> + }
> + } else {
> + if (dpll == (I915_READ(PCH_DPLL(0)) & 0x7fffffff) &&
> + fp == I915_READ(PCH_FP0(0))) {
> + intel_crtc->use_pll_a = true;
> + DRM_DEBUG_KMS("using pipe a dpll\n");
> + } else if (dpll == (I915_READ(PCH_DPLL(1)) & 0x7fffffff) &&
> + fp == I915_READ(PCH_FP0(1))) {
> + intel_crtc->use_pll_a = false;
> + DRM_DEBUG_KMS("using pipe b dpll\n");
> + } else {
> + DRM_DEBUG_KMS("no matching PLL configuration for pipe
> 2\n");
> + return -EINVAL;
> + }
> }
Ah of course. Looks like your patch should work; I'll test on my CPU
eDP system to confirm.
--
Jesse Barnes, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
