From: "Govindapillai, Vinod" <vinod.govindapillai@intel.com>
To: "intel-xe@lists.freedesktop.org" <intel-xe@lists.freedesktop.org>,
"intel-gfx@lists.freedesktop.org"
<intel-gfx@lists.freedesktop.org>,
"Deak, Imre" <imre.deak@intel.com>
Subject: Re: [PATCH 28/50] drm/i915/dp: Unify detect and compute time DSC mode BW validation
Date: Fri, 12 Dec 2025 14:29:45 +0000 [thread overview]
Message-ID: <229e34421cda9c4bdc1eb76ccc947b370a138a22.camel@intel.com> (raw)
In-Reply-To: <20251127175023.1522538-29-imre.deak@intel.com>
On Thu, 2025-11-27 at 19:50 +0200, Imre Deak wrote:
> Atm, a DP DSC video mode's required BW vs. the available BW is
> determined by calculating the maximum compressed BPP value allowed by
> the available BW. Doing that using a closed-form formula as it's done
> atm (vs. an iterative way) is problematic, since the overhead of the
> required BW itself depends on the BPP value being calculated. Instead
> of
> that calculate the required BW for the minimum compressed BPP value
> supported both by the source and the sink and check this BW against
> the
> available BW. This change also aligns the BW calculation during mode
> validation with how this is done during state computation,
> calculating
> the required effective data rate with the corresponding BW overhead.
>
> Signed-off-by: Imre Deak <imre.deak@intel.com>
> ---
> drivers/gpu/drm/i915/display/intel_dp.c | 137 ++++------------------
> --
> drivers/gpu/drm/i915/display/intel_dp.h | 8 --
> 2 files changed, 18 insertions(+), 127 deletions(-)
>
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
> diff --git a/drivers/gpu/drm/i915/display/intel_dp.c
> b/drivers/gpu/drm/i915/display/intel_dp.c
> index b40edf4febcb7..8b601994bb138 100644
> --- a/drivers/gpu/drm/i915/display/intel_dp.c
> +++ b/drivers/gpu/drm/i915/display/intel_dp.c
> @@ -895,49 +895,6 @@ static int align_max_vesa_compressed_bpp_x16(int
> max_link_bpp_x16)
> return 0;
> }
>
> -static u32 intel_dp_dsc_nearest_valid_bpp(struct intel_display
> *display, u32 bpp, u32 pipe_bpp)
> -{
> - u32 bits_per_pixel = bpp;
> -
> - /* Error out if the max bpp is less than smallest allowed
> valid bpp */
> - if (bits_per_pixel < valid_dsc_bpp[0]) {
> - drm_dbg_kms(display->drm, "Unsupported BPP %u, min
> %u\n",
> - bits_per_pixel, valid_dsc_bpp[0]);
> - return 0;
> - }
> -
> - /* From XE_LPD onwards we support from bpc upto uncompressed
> bpp-1 BPPs */
> - if (DISPLAY_VER(display) >= 13) {
> - bits_per_pixel = min(bits_per_pixel, pipe_bpp - 1);
> -
> - /*
> - * According to BSpec, 27 is the max DSC output bpp,
> - * 8 is the min DSC output bpp.
> - * While we can still clamp higher bpp values to 27,
> saving bandwidth,
> - * if it is required to oompress up to bpp < 8,
> means we can't do
> - * that and probably means we can't fit the required
> mode, even with
> - * DSC enabled.
> - */
> - if (bits_per_pixel < 8) {
> - drm_dbg_kms(display->drm,
> - "Unsupported BPP %u, min 8\n",
> - bits_per_pixel);
> - return 0;
> - }
> - bits_per_pixel = min_t(u32, bits_per_pixel, 27);
> - } else {
> - int link_bpp_x16 = fxp_q4_from_int(bits_per_pixel);
> -
> - /* Find the nearest match in the array of known BPPs
> from VESA */
> - link_bpp_x16 =
> align_max_vesa_compressed_bpp_x16(link_bpp_x16);
> -
> - drm_WARN_ON(display->drm,
> fxp_q4_to_frac(link_bpp_x16));
> - bits_per_pixel = fxp_q4_to_int(link_bpp_x16);
> - }
> -
> - return bits_per_pixel;
> -}
> -
> static int bigjoiner_interface_bits(struct intel_display *display)
> {
> return DISPLAY_VER(display) >= 14 ? 36 : 24;
> @@ -1001,64 +958,6 @@ u32 get_max_compressed_bpp_with_joiner(struct
> intel_display *display,
> return max_bpp;
> }
>
> -/* TODO: return a bpp_x16 value */
> -u16 intel_dp_dsc_get_max_compressed_bpp(struct intel_display
> *display,
> - u32 link_clock, u32
> lane_count,
> - u32 mode_clock, u32
> mode_hdisplay,
> - int num_joined_pipes,
> - enum intel_output_format
> output_format,
> - u32 pipe_bpp,
> - u32 timeslots)
> -{
> - u32 bits_per_pixel, joiner_max_bpp;
> -
> - /*
> - * Available Link Bandwidth(Kbits/sec) = (NumberOfLanes)*
> - * (LinkSymbolClock)* 8 * (TimeSlots / 64)
> - * for SST -> TimeSlots is 64(i.e all TimeSlots that are
> available)
> - * for MST -> TimeSlots has to be calculated, based on mode
> requirements
> - *
> - * Due to FEC overhead, the available bw is reduced to
> 97.2261%.
> - * To support the given mode:
> - * Bandwidth required should be <= Available link Bandwidth
> * FEC Overhead
> - * =>ModeClock * bits_per_pixel <= Available Link Bandwidth
> * FEC Overhead
> - * =>bits_per_pixel <= Available link Bandwidth * FEC
> Overhead / ModeClock
> - * =>bits_per_pixel <= (NumberOfLanes * LinkSymbolClock) * 8
> (TimeSlots / 64) /
> - * (ModeClock / FEC Overhead)
> - * =>bits_per_pixel <= (NumberOfLanes * LinkSymbolClock *
> TimeSlots) /
> - * (ModeClock / FEC Overhead * 8)
> - */
> - bits_per_pixel = ((link_clock * lane_count) * timeslots) /
> - (intel_dp_mode_to_fec_clock(mode_clock) *
> 8);
> -
> - /* Bandwidth required for 420 is half, that of 444 format */
> - if (output_format == INTEL_OUTPUT_FORMAT_YCBCR420)
> - bits_per_pixel *= 2;
> -
> - /*
> - * According to DSC 1.2a Section 4.1.1 Table 4.1 the maximum
> - * supported PPS value can be 63.9375 and with the further
> - * mention that for 420, 422 formats, bpp should be
> programmed double
> - * the target bpp restricting our target bpp to be 31.9375
> at max.
> - */
> - if (output_format == INTEL_OUTPUT_FORMAT_YCBCR420)
> - bits_per_pixel = min_t(u32, bits_per_pixel, 31);
> -
> - drm_dbg_kms(display->drm, "Max link bpp is %u for %u
> timeslots "
> - "total bw %u pixel clock %u\n",
> - bits_per_pixel, timeslots,
> - (link_clock * lane_count * 8),
> -
> intel_dp_mode_to_fec_clock(mode_clock));
> -
> - joiner_max_bpp = get_max_compressed_bpp_with_joiner(display,
> mode_clock,
> -
> mode_hdisplay, num_joined_pipes);
> - bits_per_pixel = min(bits_per_pixel, joiner_max_bpp);
> -
> - bits_per_pixel = intel_dp_dsc_nearest_valid_bpp(display,
> bits_per_pixel, pipe_bpp);
> -
> - return bits_per_pixel;
> -}
> -
> u8 intel_dp_dsc_get_slice_count(const struct intel_connector
> *connector,
> int mode_clock, int mode_hdisplay,
> int num_joined_pipes)
> @@ -2707,26 +2606,26 @@ bool intel_dp_mode_valid_with_dsc(struct
> intel_connector *connector,
> enum intel_output_format
> output_format,
> int pipe_bpp, unsigned long
> bw_overhead_flags)
> {
> - struct intel_display *display = to_intel_display(connector);
> - int dsc_max_compressed_bpp;
> - int dsc_slice_count;
> + struct intel_dp *intel_dp = intel_attached_dp(connector);
> + int min_bpp_x16 = compute_min_compressed_bpp_x16(connector,
> output_format);
> + int max_bpp_x16 = compute_max_compressed_bpp_x16(connector,
> + mode_clock,
> mode_hdisplay,
> +
> num_joined_pipes,
> +
> output_format,
> + pipe_bpp,
> INT_MAX);
> + int dsc_slice_count =
> intel_dp_dsc_get_slice_count(connector,
> +
> mode_clock,
> +
> mode_hdisplay,
> +
> num_joined_pipes);
>
> - dsc_max_compressed_bpp =
> - intel_dp_dsc_get_max_compressed_bpp(display,
> - link_clock,
> - lane_count,
> - mode_clock,
> - mode_hdisplay,
> -
> num_joined_pipes,
> - output_format,
> - pipe_bpp, 64);
> - dsc_slice_count =
> - intel_dp_dsc_get_slice_count(connector,
> - mode_clock,
> - mode_hdisplay,
> - num_joined_pipes);
> + if (min_bpp_x16 <= 0 || min_bpp_x16 > max_bpp_x16)
> + return false;
>
> - return dsc_max_compressed_bpp && dsc_slice_count;
> + return is_bw_sufficient_for_dsc_config(intel_dp,
> + link_clock,
> lane_count,
> + mode_clock,
> mode_hdisplay,
> + dsc_slice_count,
> min_bpp_x16,
> + bw_overhead_flags);
> }
>
> /*
> diff --git a/drivers/gpu/drm/i915/display/intel_dp.h
> b/drivers/gpu/drm/i915/display/intel_dp.h
> index 0ec7baec7a8e8..25bfbfd291b0a 100644
> --- a/drivers/gpu/drm/i915/display/intel_dp.h
> +++ b/drivers/gpu/drm/i915/display/intel_dp.h
> @@ -143,14 +143,6 @@ bool intel_digital_port_connected(struct
> intel_encoder *encoder);
> bool intel_digital_port_connected_locked(struct intel_encoder
> *encoder);
> int intel_dp_dsc_compute_max_bpp(const struct intel_connector
> *connector,
> u8 dsc_max_bpc);
> -u16 intel_dp_dsc_get_max_compressed_bpp(struct intel_display
> *display,
> - u32 link_clock, u32
> lane_count,
> - u32 mode_clock, u32
> mode_hdisplay,
> - int num_joined_pipes,
> - enum intel_output_format
> output_format,
> - u32 pipe_bpp,
> - u32 timeslots);
> -
> bool intel_dp_mode_valid_with_dsc(struct intel_connector *connector,
> int link_clock, int lane_count,
> int mode_clock, int mode_hdisplay,
next prev parent reply other threads:[~2025-12-12 14:30 UTC|newest]
Thread overview: 137+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-11-27 17:49 [PATCH 00/50] drm/i915/dp: Clean up link BW/DSC slice config computation Imre Deak
2025-11-27 17:49 ` [PATCH 01/50] drm/dp: Parse all DSC slice count caps for eDP 1.5 Imre Deak
2025-12-08 11:24 ` Luca Coelho
2025-12-08 12:36 ` Imre Deak
2025-11-27 17:49 ` [PATCH 02/50] drm/dp: Add drm_dp_dsc_sink_slice_count_mask() Imre Deak
2025-12-09 8:48 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 03/50] drm/i915/dp: Fix DSC sink's slice count capability check Imre Deak
2025-12-09 8:51 ` Luca Coelho
2025-12-09 9:53 ` Imre Deak
2025-12-09 11:14 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 04/50] drm/i915/dp: Return a fixed point BPP value from intel_dp_output_bpp() Imre Deak
2025-12-09 9:10 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 05/50] drm/i915/dp: Use a mode's crtc_clock vs. clock during state computation Imre Deak
2025-12-09 12:51 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 06/50] drm/i915/dp: Factor out intel_dp_link_bw_overhead() Imre Deak
2025-12-09 12:52 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 07/50] drm/i915/dp: Fix BW check in is_bw_sufficient_for_dsc_config() Imre Deak
2025-12-09 12:53 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 08/50] drm/i915/dp: Use the effective data rate for DP BW calculation Imre Deak
2025-12-10 12:48 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 09/50] drm/i915/dp: Use the effective data rate for DP compressed " Imre Deak
2025-12-10 12:50 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 10/50] drm/i915/dp: Account with MST, SSC BW overhead for uncompressed DP-MST stream BW Imre Deak
2025-12-10 13:08 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 11/50] drm/i915/dp: Account with DSC BW overhead for compressed DP-SST " Imre Deak
2025-12-10 13:39 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 12/50] drm/i915/dp: Account with pipe joiner max compressed BPP limit for DP-MST and eDP Imre Deak
2025-12-10 14:29 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 13/50] drm/i915/dp: Drop unused timeslots param from dsc_compute_link_config() Imre Deak
2025-12-10 14:31 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 14/50] drm/i915/dp: Factor out align_max_sink_dsc_input_bpp() Imre Deak
2025-12-12 15:41 ` Govindapillai, Vinod
2025-12-15 7:46 ` Luca Coelho
2025-12-15 11:53 ` Imre Deak
2025-12-15 12:02 ` Luca Coelho
2025-12-15 12:33 ` Imre Deak
2025-11-27 17:49 ` [PATCH 15/50] drm/i915/dp: Factor out align_max_vesa_compressed_bpp_x16() Imre Deak
2025-12-12 15:46 ` Govindapillai, Vinod
2025-12-15 7:49 ` Luca Coelho
2025-12-15 12:00 ` Imre Deak
2025-12-15 12:08 ` Luca Coelho
2025-12-15 12:24 ` Imre Deak
2025-11-27 17:49 ` [PATCH 16/50] drm/i915/dp: Fail state computation for invalid min/max link BPP values Imre Deak
2025-12-12 15:48 ` Govindapillai, Vinod
2025-12-15 7:51 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 17/50] drm/i915/dp: Fail state computation for invalid max throughput BPP value Imre Deak
2025-12-12 15:51 ` Govindapillai, Vinod
2025-12-15 7:51 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 18/50] drm/i915/dp: Fail state computation for invalid max sink compressed " Imre Deak
2025-12-12 15:52 ` Govindapillai, Vinod
2025-12-15 7:52 ` Luca Coelho
2025-11-27 17:49 ` [PATCH 19/50] drm/i915/dp: Fail state computation for invalid DSC source input BPP values Imre Deak
2025-12-11 8:29 ` Govindapillai, Vinod
2025-11-27 17:49 ` [PATCH 20/50] drm/i915/dp: Align min/max DSC input BPPs to sink caps Imre Deak
2025-12-11 8:51 ` Govindapillai, Vinod
2025-11-27 17:49 ` [PATCH 21/50] drm/i915/dp: Align min/max compressed BPPs when calculating BPP limits Imre Deak
2025-12-12 9:17 ` Govindapillai, Vinod
2025-12-12 11:09 ` Imre Deak
2025-11-27 17:49 ` [PATCH 22/50] drm/i915/dp: Drop intel_dp parameter from intel_dp_compute_config_link_bpp_limits() Imre Deak
2025-12-12 9:23 ` Govindapillai, Vinod
2025-11-27 17:49 ` [PATCH 23/50] drm/i915/dp: Pass intel_output_format to intel_dp_dsc_sink_{min_max}_compressed_bpp() Imre Deak
2025-12-12 9:27 ` Govindapillai, Vinod
2025-11-27 17:49 ` [PATCH 24/50] drm/i915/dp: Pass mode clock to dsc_throughput_quirk_max_bpp_x16() Imre Deak
2025-12-12 9:31 ` Govindapillai, Vinod
2025-11-27 17:49 ` [PATCH 25/50] drm/i915/dp: Factor out compute_min_compressed_bpp_x16() Imre Deak
2025-12-12 9:39 ` Govindapillai, Vinod
2025-12-12 11:01 ` Imre Deak
2025-12-12 11:41 ` Govindapillai, Vinod
2025-11-27 17:49 ` [PATCH 26/50] drm/i915/dp: Factor out compute_max_compressed_bpp_x16() Imre Deak
2025-12-12 9:50 ` Govindapillai, Vinod
2025-11-27 17:50 ` [PATCH 27/50] drm/i915/dp: Add intel_dp_mode_valid_with_dsc() Imre Deak
2025-12-12 11:43 ` Govindapillai, Vinod
2025-11-27 17:50 ` [PATCH 28/50] drm/i915/dp: Unify detect and compute time DSC mode BW validation Imre Deak
2025-12-12 14:29 ` Govindapillai, Vinod [this message]
2025-11-27 17:50 ` [PATCH 29/50] drm/i915/dp: Use helpers to align min/max compressed BPPs Imre Deak
2025-12-12 14:34 ` Govindapillai, Vinod
2025-12-12 14:39 ` Govindapillai, Vinod
2025-11-27 17:50 ` [PATCH 30/50] drm/i915/dp: Simplify computing DSC BPPs for eDP Imre Deak
2025-12-12 14:45 ` Govindapillai, Vinod
2025-11-27 17:50 ` [PATCH 31/50] drm/i915/dp: Simplify computing DSC BPPs for DP-SST Imre Deak
2025-12-12 14:59 ` Govindapillai, Vinod
2025-12-12 18:41 ` Imre Deak
2025-11-27 17:50 ` [PATCH 32/50] drm/i915/dp: Simplify computing forced DSC BPP " Imre Deak
2025-12-12 15:21 ` Govindapillai, Vinod
2025-11-27 17:50 ` [PATCH 33/50] drm/i915/dp: Unify computing compressed BPP for DP-SST and eDP Imre Deak
2025-12-12 15:38 ` Govindapillai, Vinod
2025-11-27 17:50 ` [PATCH 34/50] drm/i915/dp: Simplify eDP vs. DP compressed BPP computation Imre Deak
2025-12-12 15:39 ` Govindapillai, Vinod
2025-11-27 17:50 ` [PATCH 35/50] drm/i915/dp: Simplify computing the DSC compressed BPP for DP-MST Imre Deak
2025-12-08 13:08 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 36/50] drm/i915/dsc: Track the detaild DSC slice configuration Imre Deak
2025-12-09 8:24 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 37/50] drm/i915/dsc: Track the DSC stream count in the DSC slice config state Imre Deak
2025-12-09 8:28 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 38/50] drm/i915/dsi: Move initialization of DSI DSC streams-per-pipe to fill_dsc() Imre Deak
2025-12-09 8:47 ` Hogander, Jouni
2025-12-09 10:38 ` Imre Deak
2025-12-09 11:37 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 39/50] drm/i915/dsi: Track the detailed DSC slice configuration Imre Deak
2025-12-09 12:43 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 40/50] drm/i915/dp: " Imre Deak
2025-12-09 14:06 ` Hogander, Jouni
2025-12-09 14:30 ` Imre Deak
2025-12-09 17:50 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 41/50] drm/i915/dsc: Switch to using intel_dsc_line_slice_count() Imre Deak
2025-12-09 17:14 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 42/50] drm/i915/dp: Factor out intel_dp_dsc_min_slice_count() Imre Deak
2025-12-09 17:26 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 43/50] drm/i915/dp: Use int for DSC slice count variables Imre Deak
2025-12-09 17:30 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 44/50] drm/i915/dp: Rename test_slice_count to slices_per_line Imre Deak
2025-12-09 17:34 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 45/50] drm/i915/dp: Simplify the DSC slice config loop's slices-per-pipe iteration Imre Deak
2025-12-10 12:38 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 46/50] drm/i915/dsc: Add intel_dsc_get_slice_config() Imre Deak
2025-12-10 14:06 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 47/50] drm/i915/dsi: Use intel_dsc_get_slice_config() Imre Deak
2025-12-10 14:44 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 48/50] drm/i915/dp: Unify DP and eDP slice count computation Imre Deak
2025-12-11 6:48 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 49/50] drm/i915/dp: Add intel_dp_dsc_get_slice_config() Imre Deak
2025-12-11 6:55 ` Hogander, Jouni
2025-12-11 9:52 ` Imre Deak
2025-12-12 18:17 ` [PATCH v2 " Imre Deak
2025-12-15 6:06 ` Hogander, Jouni
2025-11-27 17:50 ` [PATCH 50/50] drm/i915/dp: Use intel_dp_dsc_get_slice_config() Imre Deak
2025-12-11 6:59 ` Hogander, Jouni
2025-12-11 10:23 ` Imre Deak
2025-12-12 18:03 ` Imre Deak
2025-12-12 18:17 ` [PATCH v2 " Imre Deak
2025-11-28 16:20 ` [CI 09/50] drm/i915/dp: Use the effective data rate for DP compressed BW calculation Imre Deak
2025-12-12 13:23 ` Govindapillai, Vinod
2025-11-28 18:48 ` ✗ i915.CI.BAT: failure for drm/i915/dp: Clean up link BW/DSC slice config computation Patchwork
2025-11-28 20:49 ` Imre Deak
2025-12-01 9:46 ` ✗ i915.CI.Full: " Patchwork
2025-12-12 20:01 ` ✓ i915.CI.BAT: success for drm/i915/dp: Clean up link BW/DSC slice config computation (rev3) Patchwork
2025-12-13 4:00 ` ✓ i915.CI.Full: " Patchwork
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=229e34421cda9c4bdc1eb76ccc947b370a138a22.camel@intel.com \
--to=vinod.govindapillai@intel.com \
--cc=imre.deak@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).