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
Subject: Re: [PATCH 2/9] drm/i915/display: Add set_context_latency to crtc_state
Date: Tue, 23 Sep 2025 17:11:51 +0300 [thread overview]
Message-ID: <aNKqp_1BzJ0WfsKO@intel.com> (raw)
In-Reply-To: <20250923131043.2628282-3-ankit.k.nautiyal@intel.com>
On Tue, Sep 23, 2025 at 06:40:36PM +0530, Ankit Nautiyal wrote:
> 'Set context latency' (SCL, Window W2) is defined as the number of lines
> before the double buffering point, which are required to complete
> programming of the registers, typically when DSB is used to program the
> display registers.
>
> Since we are not using this window for programming the registers, this
> is mostly set to 0, unless there is a requirement for few cases related
> to PSR/PR where the 'set context latency' should be at least 1.
>
> Currently we are using the 'set context latency' (if required) implicitly
> by moving the vblank start by the required amount and then measuring the
> delay i.e. the difference between undelayed vblank start and delayed vblank
> start.
>
> Since our guardband matches the vblank length, this was not a problem as
> the difference between the undelayed vblank and delayed vblank was at
> the most equal to the 'set context latency' lines.
>
> However, if we want to optimize the guardband, the difference between the
> undelayed and the delayed vblank will be large and we cannot use this
> difference as the 'set context latency' lines.
>
> To make way for this optimization of guardband, formally introduce the
> 'set context latency' or SCL and track it as a new member
> `set_context_latency` of the structure intel_crtc_state.
>
> Eventually, all references of vblank delay where we mean to use set
> context latency will be replaced by this new `set_context_latency`
> member.
>
> Note: for TGL the TRANS_SET_CONTEXT_LATENCY doesn't exist to account for
> the SCL. However, the VBLANK_START-VACTIVE diffence plays an identical role
> here ie. it can be used to create the SCL window ahead of the undelayed
> vblank.
>
> While readback there is no specific register to read out the SCL, so use
> the difference between vblank start and vactive to populate the new
> member for TGL.
>
> v2:
> - Use u16 for set_context_latency. (Ville)
> - s/vblank_delay/set_context_latency. (Ville)
> - Meld the changes for TGL with this change. (Ville)
>
> Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
> ---
> .../drm/i915/display/intel_crtc_state_dump.c | 5 +-
> drivers/gpu/drm/i915/display/intel_display.c | 53 ++++++++++++-------
> .../drm/i915/display/intel_display_types.h | 3 ++
> 3 files changed, 41 insertions(+), 20 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_crtc_state_dump.c b/drivers/gpu/drm/i915/display/intel_crtc_state_dump.c
> index 0c7f91046996..a14bcda4446c 100644
> --- a/drivers/gpu/drm/i915/display/intel_crtc_state_dump.c
> +++ b/drivers/gpu/drm/i915/display/intel_crtc_state_dump.c
> @@ -289,10 +289,11 @@ void intel_crtc_state_dump(const struct intel_crtc_state *pipe_config,
> drm_printf(&p, "scanline offset: %d\n",
> intel_crtc_scanline_offset(pipe_config));
>
> - drm_printf(&p, "vblank delay: %d, framestart delay: %d, MSA timing delay: %d\n",
> + drm_printf(&p, "vblank delay: %d, framestart delay: %d, MSA timing delay: %d set context latency: %d\n",
> pipe_config->hw.adjusted_mode.crtc_vblank_start -
> pipe_config->hw.adjusted_mode.crtc_vdisplay,
> - pipe_config->framestart_delay, pipe_config->msa_timing_delay);
> + pipe_config->framestart_delay, pipe_config->msa_timing_delay,
> + pipe_config->set_context_latency);
>
> drm_printf(&p, "vrr: %s, fixed rr: %s, vmin: %d, vmax: %d, flipline: %d, pipeline full: %d, guardband: %d vsync start: %d, vsync end: %d\n",
> str_yes_no(pipe_config->vrr.enable),
> diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> index 679c2a9baaee..aceafe4478d9 100644
> --- a/drivers/gpu/drm/i915/display/intel_display.c
> +++ b/drivers/gpu/drm/i915/display/intel_display.c
> @@ -2361,39 +2361,44 @@ static int intel_crtc_compute_pipe_mode(struct intel_crtc_state *crtc_state)
> return 0;
> }
>
> -static int intel_crtc_vblank_delay(const struct intel_crtc_state *crtc_state)
> +static int intel_crtc_set_context_latency(struct intel_crtc_state *crtc_state)
> {
> struct intel_display *display = to_intel_display(crtc_state);
> - int vblank_delay = 0;
> + int set_context_latency = 0;
>
> if (!HAS_DSB(display))
> return 0;
>
> - vblank_delay = max(vblank_delay, intel_psr_min_set_context_latency(crtc_state));
> + set_context_latency = max(set_context_latency,
> + intel_psr_min_set_context_latency(crtc_state));
>
> - return vblank_delay;
> + return set_context_latency;
> }
>
> -static int intel_crtc_compute_vblank_delay(struct intel_atomic_state *state,
> - struct intel_crtc *crtc)
> +static int intel_crtc_compute_set_context_latency(struct intel_atomic_state *state,
> + struct intel_crtc *crtc)
> {
> struct intel_display *display = to_intel_display(state);
> struct intel_crtc_state *crtc_state =
> intel_atomic_get_new_crtc_state(state, crtc);
> struct drm_display_mode *adjusted_mode =
> &crtc_state->hw.adjusted_mode;
> - int vblank_delay, max_vblank_delay;
> + int set_context_latency, max_vblank_delay;
> +
> + set_context_latency = intel_crtc_set_context_latency(crtc_state);
>
> - vblank_delay = intel_crtc_vblank_delay(crtc_state);
> max_vblank_delay = adjusted_mode->crtc_vblank_end - adjusted_mode->crtc_vblank_start - 1;
>
> - if (vblank_delay > max_vblank_delay) {
> - drm_dbg_kms(display->drm, "[CRTC:%d:%s] vblank delay (%d) exceeds max (%d)\n",
> - crtc->base.base.id, crtc->base.name, vblank_delay, max_vblank_delay);
> + if (set_context_latency > max_vblank_delay) {
> + drm_dbg_kms(display->drm, "[CRTC:%d:%s] set context latency (%d) exceeds max (%d)\n",
> + crtc->base.base.id, crtc->base.name,
> + set_context_latency,
> + max_vblank_delay);
> return -EINVAL;
> }
>
> - adjusted_mode->crtc_vblank_start += vblank_delay;
> + crtc_state->set_context_latency = set_context_latency;
> + adjusted_mode->crtc_vblank_start += set_context_latency;
>
> return 0;
> }
> @@ -2405,7 +2410,7 @@ static int intel_crtc_compute_config(struct intel_atomic_state *state,
> intel_atomic_get_new_crtc_state(state, crtc);
> int ret;
>
> - ret = intel_crtc_compute_vblank_delay(state, crtc);
> + ret = intel_crtc_compute_set_context_latency(state, crtc);
> if (ret)
> return ret;
>
> @@ -2617,7 +2622,7 @@ static void intel_set_transcoder_timings(const struct intel_crtc_state *crtc_sta
> if (DISPLAY_VER(display) >= 13) {
> intel_de_write(display,
> TRANS_SET_CONTEXT_LATENCY(display, cpu_transcoder),
> - crtc_vblank_start - crtc_vdisplay);
> + crtc_state->set_context_latency);
>
> /*
> * VBLANK_START not used by hw, just clear it
> @@ -2707,7 +2712,7 @@ static void intel_set_transcoder_timings_lrr(const struct intel_crtc_state *crtc
> if (DISPLAY_VER(display) >= 13) {
> intel_de_write(display,
> TRANS_SET_CONTEXT_LATENCY(display, cpu_transcoder),
> - crtc_vblank_start - crtc_vdisplay);
> + crtc_state->set_context_latency);
>
> /*
> * VBLANK_START not used by hw, just clear it
> @@ -2820,11 +2825,21 @@ static void intel_get_transcoder_timings(struct intel_crtc *crtc,
> adjusted_mode->crtc_vblank_end += 1;
> }
>
> - if (DISPLAY_VER(display) >= 13 && !transcoder_is_dsi(cpu_transcoder))
> - adjusted_mode->crtc_vblank_start =
> - adjusted_mode->crtc_vdisplay +
> + if (DISPLAY_VER(display) >= 13 && !transcoder_is_dsi(cpu_transcoder)) {
> + pipe_config->set_context_latency =
> intel_de_read(display,
> TRANS_SET_CONTEXT_LATENCY(display, cpu_transcoder));
> + adjusted_mode->crtc_vblank_start =
> + adjusted_mode->crtc_vdisplay +
> + pipe_config->set_context_latency;
> + } else if (DISPLAY_VER(display) == 12) {
> + /*
> + * There is no specific register for SCL for TGL.
> + * Derive the value from the difference between Vblank start and Vactive.
> + */
Importantly it is the *hardware* that derives it from those register
values. I think we should state that explicitly so that people don't
the wrong impression that this is just some random software thing.
Apart from that:
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> + pipe_config->set_context_latency =
> + adjusted_mode->crtc_vblank_start - adjusted_mode->crtc_vdisplay;
> + }
>
> if (DISPLAY_VER(display) >= 30)
> pipe_config->min_hblank = intel_de_read(display,
> @@ -5387,6 +5402,8 @@ intel_pipe_config_compare(const struct intel_crtc_state *current_config,
> PIPE_CONF_CHECK_I(vrr.guardband);
> }
>
> + PIPE_CONF_CHECK_I(set_context_latency);
> +
> #undef PIPE_CONF_CHECK_X
> #undef PIPE_CONF_CHECK_I
> #undef PIPE_CONF_CHECK_LLI
> diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h
> index 358ab922d7a7..029c47743f8b 100644
> --- a/drivers/gpu/drm/i915/display/intel_display_types.h
> +++ b/drivers/gpu/drm/i915/display/intel_display_types.h
> @@ -1341,6 +1341,9 @@ struct intel_crtc_state {
>
> /* LOBF flag */
> bool has_lobf;
> +
> + /* W2 window or 'set context latency' lines */
> + u16 set_context_latency;
> };
>
> enum intel_pipe_crc_source {
> --
> 2.45.2
--
Ville Syrjälä
Intel
next prev parent reply other threads:[~2025-09-23 14:11 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-23 13:10 [PATCH 0/9] Introduce set_context_latency and refactor VRR/DSB timing logic Ankit Nautiyal
2025-09-23 13:10 ` [PATCH 1/9] drm/i915/psr: s/intel_psr_min_vblank_delay/intel_psr_min_set_context_latency Ankit Nautiyal
2025-09-23 13:10 ` [PATCH 2/9] drm/i915/display: Add set_context_latency to crtc_state Ankit Nautiyal
2025-09-23 14:11 ` Ville Syrjälä [this message]
2025-09-23 13:10 ` [PATCH 3/9] drm/i915/vrr: Use set_context_latency instead of intel_vrr_real_vblank_delay() Ankit Nautiyal
2025-09-23 13:10 ` [PATCH 4/9] drm/i915/vrr: Use SCL for computing guardband Ankit Nautiyal
2025-09-23 14:12 ` Ville Syrjälä
2025-09-23 13:10 ` [PATCH 5/9] drm/i915/vrr: s/intel_vrr_vblank_delay/intel_vrr_scl_delay Ankit Nautiyal
2025-09-23 14:13 ` Ville Syrjälä
2025-09-24 9:33 ` Nautiyal, Ankit K
2025-09-24 9:49 ` Ville Syrjälä
2025-09-23 13:10 ` [PATCH 6/9] drm/i915/dsb: s/intel_dsb_wait_vblank_delay/intel_dsb_wait_for_delayed_vblank Ankit Nautiyal
2025-09-23 17:21 ` Ville Syrjälä
2025-09-23 13:10 ` [PATCH 7/9] drm/i915/display: Wait for scl start instead of dsb_wait_vblanks Ankit Nautiyal
2025-09-23 14:32 ` Ville Syrjälä
2025-09-23 13:10 ` [PATCH 8/9] drm/i915/reg_defs: Add REG_FIELD_MAX wrapper for FIELD_MAX() Ankit Nautiyal
2025-09-23 17:27 ` Ville Syrjälä
2025-09-23 13:10 ` [PATCH 9/9] drm/i915/vrr: Clamp guardband as per hardware and timing constraints Ankit Nautiyal
2025-09-23 17:25 ` Ville Syrjälä
2025-09-23 13:32 ` ✓ CI.KUnit: success for Introduce set_context_latency and refactor VRR/DSB timing logic (rev2) Patchwork
2025-09-23 13:47 ` ✗ CI.checksparse: warning " Patchwork
2025-09-23 14:16 ` ✓ Xe.CI.BAT: success " Patchwork
2025-09-23 15:48 ` ✗ Xe.CI.Full: failure " Patchwork
-- strict thread matches above, loose matches on Subject: below --
2025-09-24 10:51 [PATCH 0/9] Introduce set_context_latency and refactor VRR/DSB timing logic Ankit Nautiyal
2025-09-24 10:51 ` [PATCH 2/9] drm/i915/display: Add set_context_latency to crtc_state 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=aNKqp_1BzJ0WfsKO@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 \
/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).