On Tue, Feb 06, 2018 at 04:36:42PM +0530, Vidya Srinivas wrote:
> From: Chandra Konduru <chandra.konduru@xxxxxxxxx>
>
> Display WA #0827:
> Switching the plane format from NV12 to RGB and leaving system idle results
> in display underrun and corruption. WA: Set the bit 15 & bit 19 to 1b
> in the CLKGATE_DIS_PSL register for the pipe in which NV12 plane is enabled.
>
> Signed-off-by: Chandra Konduru <chandra.konduru@xxxxxxxxx>
> Signed-off-by: Vidya Srinivas <vidya.srinivas@xxxxxxxxx>
> ---
> drivers/gpu/drm/i915/i915_reg.h | 3 +++
> drivers/gpu/drm/i915/intel_display.c | 16 ++++++++++++++++
> 2 files changed, 19 insertions(+)
>
> diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
> index 8f36023..c4af05e 100644
> --- a/drivers/gpu/drm/i915/i915_reg.h
> +++ b/drivers/gpu/drm/i915/i915_reg.h
> @@ -3822,6 +3822,9 @@ enum {
> #define _CLKGATE_DIS_PSL_A 0x46520
> #define _CLKGATE_DIS_PSL_B 0x46524
> #define _CLKGATE_DIS_PSL_C 0x46528
> +#define DUPS1_GATING_DIS (1 << 15)
> +#define DUPS2_GATING_DIS (1 << 19)
> +#define DUPS3_GATING_DIS (1 << 23)
> #define DPF_GATING_DIS (1 << 10)
> #define DPF_RAM_GATING_DIS (1 << 9)
> #define DPFR_GATING_DIS (1 << 8)
> diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
> index 551c970..94faf3e 100644
> --- a/drivers/gpu/drm/i915/intel_display.c
> +++ b/drivers/gpu/drm/i915/intel_display.c
> @@ -5495,6 +5495,20 @@ static void glk_pipe_scaler_clock_gating_wa(struct drm_i915_private *dev_priv,
> I915_WRITE(CLKGATE_DIS_PSL(pipe), val);
> }
>
> +static void skl_wa_clkgate(struct drm_i915_private *dev_priv,
> + int pipe, int enable)
> +{
> + if (pipe == PIPE_A || pipe == PIPE_B) {
> + if (enable)
> + I915_WRITE(CLKGATE_DIS_PSL(pipe),
> + DUPS1_GATING_DIS | DUPS2_GATING_DIS);
> + else
> + I915_WRITE(CLKGATE_DIS_PSL(pipe),
> + I915_READ(CLKGATE_DIS_PSL(pipe)) &
> + ~(DUPS1_GATING_DIS|DUPS2_GATING_DIS));
> + }
> +}
> +
> static void haswell_crtc_enable(struct intel_crtc_state *pipe_config,
> struct drm_atomic_state *old_state)
> {
> @@ -5599,6 +5613,7 @@ static void haswell_crtc_enable(struct intel_crtc_state *pipe_config,
> intel_wait_for_vblank(dev_priv, hsw_workaround_pipe);
> intel_wait_for_vblank(dev_priv, hsw_workaround_pipe);
> }
> + skl_wa_clkgate(dev_priv, pipe, 1);
> }
>
> static void ironlake_pfit_disable(struct intel_crtc *crtc, bool force)
> @@ -5709,6 +5724,7 @@ static void haswell_crtc_disable(struct intel_crtc_state *old_crtc_state,
> intel_ddi_disable_pipe_clock(intel_crtc->config);
>
> intel_encoders_post_disable(crtc, old_crtc_state, old_state);
> + skl_wa_clkgate(dev_priv, intel_crtc->pipe, 0);
> }
Unless I'm misreading the context of this patch you're applying a workaround,
that by name seems to be for Skylake only, to: Haswell, Broadwell, and gen9+.
Either the name is incorrect, or the application of it.
As per BSpec the workaround seems to be for all of gen9 and only A-stepping of gen10.
I don't see it listed for Haswell or Broadwell.
Cross-referencing the WA-database with Bspec, based on the HSD
link, it seems that this issue *might* be
WaDups1GatingDisableClockGatingForMPO; if this is the case it might
make sense to include that WA name too. At the very least
there should always be a comment mentioning the workaround name/number
and the platform(s) it applies to.
Also, according to the WA database, if the above mentioned issue really
is the same, the WA is *NOT* necessary on GLK (seeing as GLK uses gen10
display this might make sense, though the WA database sometimes contains
mistakes).
Regards, David
> static void i9xx_pfit_enable(struct intel_crtc *crtc)
> --
> 1.9.1
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
