[Intel-gfx] [PATCH] drm/i915: Enable 5.4Ghz (HBR2) link rate for Displayport 1.2-capable devices

[Daniel Vetter]

On Fri, Jan 17, 2014 at 4:06 AM, Todd Previte <tprevite at gmail.com> wrote:
> For HSW+ platforms, enable the 5.4Ghz (HBR2) link rate for devices that support it. The
> sink device must report that is supports Displayport 1.2 and the HBR2 bit rate in the
> DPCD in order to use HBR2.

sob line missing.

> ---
>  drivers/gpu/drm/i915/intel_dp.c | 21 +++++++++++++++------
>  1 file changed, 15 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
> index 7df5085..f92d1c0 100644
> --- a/drivers/gpu/drm/i915/intel_dp.c
> +++ b/drivers/gpu/drm/i915/intel_dp.c
> @@ -102,7 +102,10 @@ intel_dp_max_link_bw(struct intel_dp *intel_dp)
>         case DP_LINK_BW_2_7:
>                 break;
>         case DP_LINK_BW_5_4: /* 1.2 capable displays may advertise higher bw */
> -               max_link_bw = DP_LINK_BW_2_7;
> +        if (intel_dp->dpcd[DP_DPCD_REV] == 0x12)
> +            max_link_bw = DP_LINK_BW_5_4;
> +        else
> +            max_link_bw = DP_LINK_BW_2_7;

Is this really required, i.e. do we have dp 1.1 machines in the wild
which advertise 5.4 but can't? In any case you also need to have a
IS_HSW || IS_BDW check here, since only those two platforms support
5.4 GHz.

>                 break;
>         default:
>                 WARN(1, "invalid max DP link bw val %x, using 1.62Gbps\n",
> @@ -805,9 +808,10 @@ intel_dp_compute_config(struct intel_encoder *encoder,
>         struct intel_connector *intel_connector = intel_dp->attached_connector;
>         int lane_count, clock;
>         int max_lane_count = drm_dp_max_lane_count(intel_dp->dpcd);
> -       int max_clock = intel_dp_max_link_bw(intel_dp) == DP_LINK_BW_2_7 ? 1 : 0;
> +    /* Conveniently, the link BW constants become indices with a shift...*/
> +    int max_clock = intel_dp_max_link_bw(intel_dp) >> 3;
>         int bpp, mode_rate;
> -       static int bws[2] = { DP_LINK_BW_1_62, DP_LINK_BW_2_7 };
> +       static int bws[] = { DP_LINK_BW_1_62, DP_LINK_BW_2_7, DP_LINK_BW_5_4 };
>         int link_avail, link_clock;
>
>         if (HAS_PCH_SPLIT(dev) && !HAS_DDI(dev) && port != PORT_A)
> @@ -2621,10 +2625,15 @@ intel_dp_complete_link_train(struct intel_dp *intel_dp)
>         bool channel_eq = false;
>         int tries, cr_tries;
>         uint32_t DP = intel_dp->DP;
> +    uint32_t training_pattern = DP_TRAINING_PATTERN_2;
> +
> +    /* Training Pattern 3 for HBR2 */
> +    if (intel_dp->link_bw == DP_LINK_BW_5_4)
> +        training_pattern = DP_TRAINING_PATTERN_3;

Hm, I've thought that with 5.4 we're supposed to do both pattern 2&3.
But tbh I didn't bother to dig out the spec ;-) Does it hurt not to?
Also, is there any harm in using pattern 3 for all dp 1.2 capable
sinks? I've thought that it should give us more reliable link
training, so might be beneficial not just for 5.4 mode.
-Daniel

>         /* channel equalization */
>         if (!intel_dp_set_link_train(intel_dp, &DP,
> -                                    DP_TRAINING_PATTERN_2 |
> +                                    training_pattern |
>                                      DP_LINK_SCRAMBLING_DISABLE)) {
>                 DRM_ERROR("failed to start channel equalization\n");
>                 return;
> @@ -2652,7 +2661,7 @@ intel_dp_complete_link_train(struct intel_dp *intel_dp)
>                 if (!drm_dp_clock_recovery_ok(link_status, intel_dp->lane_count)) {
>                         intel_dp_start_link_train(intel_dp);
>                         intel_dp_set_link_train(intel_dp, &DP,
> -                                               DP_TRAINING_PATTERN_2 |
> +                                               training_pattern |
>                                                 DP_LINK_SCRAMBLING_DISABLE);
>                         cr_tries++;
>                         continue;
> @@ -2668,7 +2677,7 @@ intel_dp_complete_link_train(struct intel_dp *intel_dp)
>                         intel_dp_link_down(intel_dp);
>                         intel_dp_start_link_train(intel_dp);
>                         intel_dp_set_link_train(intel_dp, &DP,
> -                                               DP_TRAINING_PATTERN_2 |
> +                                               training_pattern |
>                                                 DP_LINK_SCRAMBLING_DISABLE);
>                         tries = 0;
>                         cr_tries++;
> --
> 1.8.1.2
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx



-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch