intel-xe.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: "Ville Syrjälä" <ville.syrjala@linux.intel.com>
To: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Cc: intel-gfx@lists.freedesktop.org, intel-xe@lists.freedesktop.org,
	Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Subject: Re: [PATCH 15/15] drm/i915/vrr: Fix seamless_mn drrs for PTL
Date: Thu, 11 Sep 2025 17:41:30 +0300	[thread overview]
Message-ID: <aMLfmlp0u6KKpSwa@intel.com> (raw)
In-Reply-To: <20250911024554.692469-16-ankit.k.nautiyal@intel.com>

On Thu, Sep 11, 2025 at 08:15:54AM +0530, Ankit Nautiyal wrote:
> With VRR timing generator always on, the fixed refresh rate is achieved
> by setting vrr.flipline and vrr.vmax as the vtotal for the desired mode.
> 
> This creates a problem for seamless_mn drrs feature, where user can
> seamlessly set a lower mode on the supporting panels. With VRR timing
> generator, the vrr.flipline and vrr.vmax are set to vtotal, but that
> corresponds to the higher mode.
> 
> To fix this, re-compute the vrr timings when seamless_mn drrs is in
> picture. At the same time make sure that the vrr.guardband is set as
> per the highest mode for such panels, so that switching between higher
> to lower mode, does not change the vrr.guardband.
> 
> v2: Add a new member `use_highest_mode` to vrr struct to help set the
> vrr timings for highest mode for the seamless_mn drrs case.
> v3:
> -Modify existing function to compute fixed refresh rate timings instead
> of adding a new function. (Mitul)
> -Tweak computation for scaling the vtotal and use DIV_ROUND_UP_ULL.
> -Improve documentation.
> 
> Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
> Reviewed-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
> ---
>  .../drm/i915/display/intel_display_types.h    |  2 +
>  drivers/gpu/drm/i915/display/intel_dp.c       |  2 +-
>  drivers/gpu/drm/i915/display/intel_dp.h       |  1 +
>  drivers/gpu/drm/i915/display/intel_vrr.c      | 90 ++++++++++++++++++-
>  4 files changed, 90 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h
> index 358ab922d7a7..9796c7b855d0 100644
> --- a/drivers/gpu/drm/i915/display/intel_display_types.h
> +++ b/drivers/gpu/drm/i915/display/intel_display_types.h
> @@ -1321,6 +1321,8 @@ struct intel_crtc_state {
>  		u8 pipeline_full;
>  		u16 flipline, vmin, vmax, guardband;
>  		u32 vsync_end, vsync_start;
> +		/* Indicates VRR timing is scaled to highest mode for seamless M/N */
> +		bool use_highest_mode;
>  	} vrr;
>  
>  	/* Content Match Refresh Rate state */
> diff --git a/drivers/gpu/drm/i915/display/intel_dp.c b/drivers/gpu/drm/i915/display/intel_dp.c
> index f74ac98062d4..5c29c696c83e 100644
> --- a/drivers/gpu/drm/i915/display/intel_dp.c
> +++ b/drivers/gpu/drm/i915/display/intel_dp.c
> @@ -1742,7 +1742,7 @@ static int intel_dp_max_bpp(struct intel_dp *intel_dp,
>  	return bpp;
>  }
>  
> -static bool has_seamless_m_n(struct intel_connector *connector)
> +bool has_seamless_m_n(struct intel_connector *connector)
>  {
>  	struct intel_display *display = to_intel_display(connector);
>  
> diff --git a/drivers/gpu/drm/i915/display/intel_dp.h b/drivers/gpu/drm/i915/display/intel_dp.h
> index d222749b191c..6da0196c23d1 100644
> --- a/drivers/gpu/drm/i915/display/intel_dp.h
> +++ b/drivers/gpu/drm/i915/display/intel_dp.h
> @@ -217,5 +217,6 @@ void intel_dp_dpcd_set_probe(struct intel_dp *intel_dp, bool force_on_external);
>  bool intel_dp_in_hdr_mode(const struct drm_connector_state *conn_state);
>  int intel_dp_compute_sdp_latency(const struct intel_crtc_state *crtc_state,
>  				 bool assume_all_enabled);
> +bool has_seamless_m_n(struct intel_connector *connector);
>  
>  #endif /* __INTEL_DP_H__ */
> diff --git a/drivers/gpu/drm/i915/display/intel_vrr.c b/drivers/gpu/drm/i915/display/intel_vrr.c
> index 002165026a20..80bbe4b1ef7f 100644
> --- a/drivers/gpu/drm/i915/display/intel_vrr.c
> +++ b/drivers/gpu/drm/i915/display/intel_vrr.c
> @@ -11,6 +11,7 @@
>  #include "intel_display_regs.h"
>  #include "intel_display_types.h"
>  #include "intel_dp.h"
> +#include "intel_panel.h"
>  #include "intel_vrr.h"
>  #include "intel_vrr_regs.h"
>  #include "skl_scaler.h"
> @@ -299,6 +300,16 @@ void intel_vrr_set_fixed_rr_timings(const struct intel_crtc_state *crtc_state)
>  	if (!intel_vrr_possible(crtc_state))
>  		return;
>  
> +	if (crtc_state->vrr.use_highest_mode) {
> +		intel_de_write(display, TRANS_VRR_VMIN(display, cpu_transcoder),
> +			       crtc_state->vrr.vmin - 1);
> +		intel_de_write(display, TRANS_VRR_VMAX(display, cpu_transcoder),
> +			       crtc_state->vrr.vmax - 1);
> +		intel_de_write(display, TRANS_VRR_FLIPLINE(display, cpu_transcoder),
> +			       crtc_state->vrr.flipline - 1);
> +		return;
> +	}
> +
>  	intel_de_write(display, TRANS_VRR_VMIN(display, cpu_transcoder),
>  		       intel_vrr_fixed_rr_vmin(crtc_state) - 1);
>  	intel_de_write(display, TRANS_VRR_VMAX(display, cpu_transcoder),
> @@ -307,15 +318,69 @@ void intel_vrr_set_fixed_rr_timings(const struct intel_crtc_state *crtc_state)
>  		       intel_vrr_fixed_rr_flipline(crtc_state) - 1);
>  }
>  
> +static bool needs_seamless_m_n_timings(struct intel_crtc_state *crtc_state,
> +				       struct intel_connector *connector)
> +{
> +	if (!has_seamless_m_n(connector) || crtc_state->joiner_pipes)
> +		return false;
> +
> +	return true;
> +}
> +
> +static int intel_vrr_scale_vtotal_for_seamless_m_n(struct intel_crtc_state *crtc_state,
> +						   struct intel_connector *connector)
> +{
> +	const struct drm_display_mode *highest_mode = intel_panel_highest_mode(connector);
> +	const struct drm_display_mode *adjusted_mode = &crtc_state->hw.adjusted_mode;
> +	int vtotal = adjusted_mode->crtc_vtotal;
> +
> +	/*
> +	 * For panels with seamless_m_n drrs, the user can seamlessly switch to
> +	 * a lower mode, which has a lower clock. This works with legacy timing
> +	 * generator, but not with the VRR timing generator.
> +	 *
> +	 * The VRR timing generator requires flipline and vmax to be equal for
> +	 * fixed refresh rate operation. The default fixed RR computation sets
> +	 * these to the current mode's vtotal. However, when switching to a
> +	 * lower clock mode, this would result in a higher refresh rate than
> +	 * desired.
> +	 *
> +	 * To simulate the lower refresh rate correctly, we scale the vtotal
> +	 * based on the ratio of the highest mode's clock to the current mode's
> +	 * clock.
> +	 *
> +	 * When switching to a higher clock mode, the current vtotal already
> +	 * results in the desired refresh rate, so no scaling is needed.
> +	 *
> +	 * So compute the scaled vtotal if required, and update vrr.vmin to
> +	 * the scaled value. Also, set vrr.use_highest_mode to indicate that
> +	 * VRR timings are based on the highest mode.
> +	 */
> +	if (highest_mode && adjusted_mode->crtc_clock < highest_mode->clock) {
> +		vtotal = DIV_ROUND_UP_ULL(vtotal * highest_mode->clock,
> +					  adjusted_mode->crtc_clock);
> +		crtc_state->vrr.vmin = vtotal;
> +		crtc_state->vrr.use_highest_mode = true;
> +	}

I have no idea what is happening here. I think the only thing we should
be aiming for is a fixed guardband length, but this is now doing all kinds
of otehr massaging of VRR parameters. Why?

> +
> +	return vtotal;
> +}
> +
>  static
> -void intel_vrr_compute_fixed_rr_timings(struct intel_crtc_state *crtc_state)
> +void intel_vrr_compute_fixed_rr_timings(struct intel_crtc_state *crtc_state,
> +					struct intel_connector *connector)
>  {
> +	int vtotal = crtc_state->hw.adjusted_mode.crtc_vtotal;
> +
> +	if (needs_seamless_m_n_timings(crtc_state, connector))
> +		vtotal = intel_vrr_scale_vtotal_for_seamless_m_n(crtc_state, connector);
> +
>  	/*
>  	 * For fixed rr,  vmin = vmax = flipline.
>  	 * vmin is already set to crtc_vtotal set vmax and flipline the same.
>  	 */
> -	crtc_state->vrr.vmax = crtc_state->hw.adjusted_mode.crtc_vtotal;
> -	crtc_state->vrr.flipline = crtc_state->hw.adjusted_mode.crtc_vtotal;
> +	crtc_state->vrr.vmax = vtotal;
> +	crtc_state->vrr.flipline = vtotal;
>  }
>  
>  static
> @@ -397,7 +462,7 @@ intel_vrr_compute_config(struct intel_crtc_state *crtc_state,
>  	else if (is_cmrr_frac_required(crtc_state) && is_edp)
>  		intel_vrr_compute_cmrr_timings(crtc_state);
>  	else
> -		intel_vrr_compute_fixed_rr_timings(crtc_state);
> +		intel_vrr_compute_fixed_rr_timings(crtc_state, connector);
>  
>  	/*
>  	 * flipline determines the min vblank length the hardware will
> @@ -876,6 +941,7 @@ int intel_vrr_compute_guardband(struct intel_crtc_state *crtc_state,
>  {
>  	const struct drm_display_mode *adjusted_mode = &crtc_state->hw.adjusted_mode;
>  	struct intel_display *display = to_intel_display(crtc_state);
> +	const struct drm_display_mode *highest_mode;
>  	int dsc_prefill_time = 0;
>  	int psr2_pr_latency = 0;
>  	int scaler_prefill_time;
> @@ -888,6 +954,22 @@ int intel_vrr_compute_guardband(struct intel_crtc_state *crtc_state,
>  	int guardband;
>  	int pm_delay;
>  
> +	/*
> +	 * For seamless m_n the clock is changed while other modeline
> +	 * parameters are same. In that case the linetime_us will change,
> +	 * causing the guardband to change, and the seamless switch to
> +	 * lower mode would not take place.
> +	 * To avoid this, take the highest mode where panel supports
> +	 * seamless drrs and make guardband equal to the vblank length
> +	 * for the highest mode.
> +	 */
> +	highest_mode = intel_panel_highest_mode(connector);
> +	if (needs_seamless_m_n_timings(crtc_state, connector) && highest_mode) {
> +		guardband = highest_mode->vtotal - highest_mode->vdisplay;
> +
> +		return guardband;
> +	}
> +
>  	linetime_us = DIV_ROUND_UP(adjusted_mode->crtc_htotal * 1000,
>  				   adjusted_mode->crtc_clock);
>  
> -- 
> 2.45.2

-- 
Ville Syrjälä
Intel

  reply	other threads:[~2025-09-11 14:41 UTC|newest]

Thread overview: 47+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-09-11  2:45 [PATCH 00/15] Optimize vrr.guardband and fix LRR Ankit Nautiyal
2025-09-11  2:45 ` [PATCH 01/15] drm/i915/vrr: Use crtc_vsync_start/end for computing vrr.vsync_start/end Ankit Nautiyal
2025-09-11  2:45 ` [PATCH 02/15] drm/i915/skl_watermark: Fix the scaling factor for chroma subsampling Ankit Nautiyal
2025-09-11  2:45 ` [PATCH 03/15] drm/i915/skl_watermark: Pass linetime as argument to latency helpers Ankit Nautiyal
2025-09-11 13:58   ` Ville Syrjälä
2025-09-14  6:00     ` Nautiyal, Ankit K
2025-09-11  2:45 ` [PATCH 04/15] drm/i915/skl_scaler: Introduce helper for chroma downscale factor Ankit Nautiyal
2025-09-11  2:45 ` [PATCH 05/15] drm/i915/display: Extract helpers to set dsc/scaler prefill latencies Ankit Nautiyal
2025-09-11 14:01   ` Ville Syrjälä
2025-09-14  6:02     ` Nautiyal, Ankit K
2025-09-11  2:45 ` [PATCH 06/15] drm/i915/dp: Add SDP latency computation helper Ankit Nautiyal
2025-09-11 14:14   ` Ville Syrjälä
2025-09-14  6:03     ` Nautiyal, Ankit K
2025-09-11  2:45 ` [PATCH 07/15] drm/i915/alpm: Add function to compute max link-wake latency Ankit Nautiyal
2025-09-11  2:45 ` [PATCH 08/15] drm/i915/vrr: Use vrr.sync_start for getting vtotal Ankit Nautiyal
2025-09-11  2:45 ` [PATCH 09/15] drm/i915/display: Add guardband check for feature latencies Ankit Nautiyal
2025-09-11  2:45 ` [PATCH 10/15] drm/i915/skl_watermark: Remove redundant latency checks from vblank validation Ankit Nautiyal
2025-09-11 14:22   ` Ville Syrjälä
2025-09-14  6:04     ` Nautiyal, Ankit K
2025-09-11  2:45 ` [PATCH 11/15] drm/i915/display: Use vrr.guardband to derive vblank_start Ankit Nautiyal
2025-09-11 14:25   ` Ville Syrjälä
2025-09-14  5:59     ` Nautiyal, Ankit K
2025-09-15 12:32       ` Ville Syrjälä
2025-09-16 14:30         ` Nautiyal, Ankit K
2025-09-16 14:38           ` Nautiyal, Ankit K
2025-09-16 18:56             ` Ville Syrjälä
2025-09-17 10:38               ` Nautiyal, Ankit K
2025-09-17 12:36                 ` Ville Syrjälä
2025-09-17 10:51               ` Ville Syrjälä
2025-09-17 12:07                 ` Shankar, Uma
2025-09-17 20:51                 ` Ville Syrjälä
2025-09-17 21:12                   ` Ville Syrjälä
2025-09-11  2:45 ` [PATCH 12/15] drm/i915/vrr: Introduce helper to compute min static guardband Ankit Nautiyal
2025-09-11  2:45 ` [PATCH 13/15] drm/i915/display: Use optimized guardband to set vblank start Ankit Nautiyal
2025-09-11  2:45 ` [PATCH 14/15] drm/i915/panel: Refactor helper to get highest fixed mode Ankit Nautiyal
2025-09-11 14:37   ` Ville Syrjälä
2025-09-14  6:08     ` Nautiyal, Ankit K
2025-09-11  2:45 ` [PATCH 15/15] drm/i915/vrr: Fix seamless_mn drrs for PTL Ankit Nautiyal
2025-09-11 14:41   ` Ville Syrjälä [this message]
2025-09-14  6:07     ` Nautiyal, Ankit K
2025-09-15 13:25       ` Ville Syrjälä
2025-09-11  3:11 ` ✓ CI.KUnit: success for Optimize vrr.guardband and fix LRR (rev11) Patchwork
2025-09-11  3:47 ` ✓ Xe.CI.BAT: " Patchwork
2025-09-11  9:27 ` ✓ Xe.CI.Full: " Patchwork
2025-09-12 14:03 ` [PATCH 00/15] Optimize vrr.guardband and fix LRR Ville Syrjälä
2025-09-14  6:24   ` Nautiyal, Ankit K
  -- strict thread matches above, loose matches on Subject: below --
2025-08-04 13:24 Ankit Nautiyal
2025-08-04 13:24 ` [PATCH 15/15] drm/i915/vrr: Fix seamless_mn drrs for PTL Ankit Nautiyal

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=aMLfmlp0u6KKpSwa@intel.com \
    --to=ville.syrjala@linux.intel.com \
    --cc=ankit.k.nautiyal@intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=mitulkumar.ajitkumar.golani@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).