From: "Ville Syrjälä" <ville.syrjala@linux.intel.com>
To: Imre Deak <imre.deak@intel.com>
Cc: intel-gfx@lists.freedesktop.org, intel-xe@lists.freedesktop.org,
dri-devel@lists.freedesktop.org
Subject: Re: [PATCH v5 2/6] drm/dp: Add helpers to query the branch DSC max throughput/line-width
Date: Mon, 29 Sep 2025 13:12:15 +0300 [thread overview]
Message-ID: <aNpbf4HJDrtbsuex@intel.com> (raw)
In-Reply-To: <20250929063644.533890-1-imre.deak@intel.com>
On Mon, Sep 29, 2025 at 09:36:44AM +0300, Imre Deak wrote:
> Add helpers to query the DP DSC sink device's per-slice throughput as
> well as a DSC branch device's overall throughput and line-width
> capabilities.
>
> v2 (Ville):
> - Rename pixel_clock to peak_pixel_rate, document what the value means
> in case of MST tiled displays.
> - Fix name of drm_dp_dsc_branch_max_slice_throughput() to
> drm_dp_dsc_sink_max_slice_throughput().
> v3:
> - Fix the DSC branch device minimum valid line width value from 2560
> to 5120 pixels.
> - Fix drm_dp_dsc_sink_max_slice_throughput()'s pixel_clock parameter
> name to peak_pixel_rate in header file.
> - Add handling for throughput mode 0 granular delta, defined by DP
> Standard v2.1a.
>
> Cc: dri-devel@lists.freedesktop.org
> Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Signed-off-by: Imre Deak <imre.deak@intel.com>
> ---
> drivers/gpu/drm/display/drm_dp_helper.c | 156 ++++++++++++++++++++++++
> include/drm/display/drm_dp.h | 3 +
> include/drm/display/drm_dp_helper.h | 5 +
> 3 files changed, 164 insertions(+)
>
> diff --git a/drivers/gpu/drm/display/drm_dp_helper.c b/drivers/gpu/drm/display/drm_dp_helper.c
> index 1c74fe9459ad9..6084ca7fbe263 100644
> --- a/drivers/gpu/drm/display/drm_dp_helper.c
> +++ b/drivers/gpu/drm/display/drm_dp_helper.c
> @@ -2844,6 +2844,162 @@ int drm_dp_dsc_sink_supported_input_bpcs(const u8 dsc_dpcd[DP_DSC_RECEIVER_CAP_S
> }
> EXPORT_SYMBOL(drm_dp_dsc_sink_supported_input_bpcs);
>
> +/*
> + * See DP Standard v2.1a 2.8.4 Minimum Slices/Display, Table 2-159 and
> + * Appendix L.1 Derivation of Slice Count Requirements.
> + */
> +static int dsc_sink_min_slice_throughput(int peak_pixel_rate)
> +{
> + if (peak_pixel_rate >= 4800000)
> + return 600000;
> + else if (peak_pixel_rate >= 2700000)
> + return 400000;
> + else
> + return 340000;
> +}
> +
> +/**
> + * drm_dp_dsc_sink_max_slice_throughput() - Get a DSC sink's maximum pixel throughput per slice
> + * @dsc_dpcd: DSC sink's capabilities from DPCD
> + * @peak_pixel_rate: Cumulative peak pixel rate in kHz
> + * @is_rgb_yuv444: The mode is either RGB or YUV444
> + *
> + * Return the DSC sink device's maximum pixel throughput per slice, based on
> + * the device's @dsc_dpcd capabilities, the @peak_pixel_rate of the transferred
> + * stream(s) and whether the output format @is_rgb_yuv444 or yuv422/yuv420.
> + *
> + * Note that @peak_pixel_rate is the total pixel rate transferred to the same
> + * DSC/display sink. For instance to calculate a tile's slice count of an MST
> + * multi-tiled display sink (not considering here the required
> + * rounding/alignment of slice count)::
> + *
> + * @peak_pixel_rate = tile_pixel_rate * tile_count
> + * total_slice_count = @peak_pixel_rate / drm_dp_dsc_sink_max_slice_throughput(@peak_pixel_rate)
> + * tile_slice_count = total_slice_count / tile_count
> + *
> + * Returns:
> + * The maximum pixel throughput per slice supported by the DSC sink device
> + * in kPixels/sec.
> + */
> +int drm_dp_dsc_sink_max_slice_throughput(const u8 dsc_dpcd[DP_DSC_RECEIVER_CAP_SIZE],
> + int peak_pixel_rate, bool is_rgb_yuv444)
> +{
> + int throughput;
> + int delta = 0;
> + int base;
> +
> + throughput = dsc_dpcd[DP_DSC_PEAK_THROUGHPUT - DP_DSC_SUPPORT];
> +
> + if (is_rgb_yuv444) {
> + throughput = (throughput & DP_DSC_THROUGHPUT_MODE_0_MASK) >>
> + DP_DSC_THROUGHPUT_MODE_0_SHIFT;
> +
> + delta = ((dsc_dpcd[DP_DSC_RC_BUF_BLK_SIZE - DP_DSC_SUPPORT]) &
> + DP_DSC_THROUGHPUT_MODE_0_DELTA_MASK) >>
> + DP_DSC_THROUGHPUT_MODE_0_DELTA_SHIFT; /* in units of 2 MPixels/sec */
> + delta *= 2000;
That thing really is very poorly documented. The DSC spec has a
note about possibly needing extra throughput at the right edges
of slices due to some alignment reasons. Maybe this is for that?
Dunno.
And the fact that there is a delta just for the 4:4:4 throughput
is also rather weird.
> + } else {
> + throughput = (throughput & DP_DSC_THROUGHPUT_MODE_1_MASK) >>
> + DP_DSC_THROUGHPUT_MODE_1_SHIFT;
> + }
> +
> + switch (throughput) {
> + case 0:
> + return dsc_sink_min_slice_throughput(peak_pixel_rate);
> + case 1:
> + base = 340000;
> + break;
> + case 2 ... 14:
> + base = 400000 + 50000 * (throughput - 2);
> + break;
> + case 15:
> + base = 170000;
> + break;
> + default:
> + WARN(1, "Missing case %d\n", throughput);
> + base = 340000;
> + break;
nit: that looks like an impossible situation. Could either nuke
the default case, or if the compiler/static checks are dumb
then get rid of the 2...14 case instead.
Anyways, for the series
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> + }
> +
> + return base + delta;
> +}
> +EXPORT_SYMBOL(drm_dp_dsc_sink_max_slice_throughput);
> +
> +static u8 dsc_branch_dpcd_cap(const u8 dpcd[DP_DSC_BRANCH_CAP_SIZE], int reg)
> +{
> + return dpcd[reg - DP_DSC_BRANCH_OVERALL_THROUGHPUT_0];
> +}
> +
> +/**
> + * drm_dp_dsc_branch_max_overall_throughput() - Branch device's max overall DSC pixel throughput
> + * @dsc_branch_dpcd: DSC branch capabilities from DPCD
> + * @is_rgb_yuv444: The mode is either RGB or YUV444
> + *
> + * Return the branch device's maximum overall DSC pixel throughput, based on
> + * the device's DPCD DSC branch capabilities, and whether the output
> + * format @is_rgb_yuv444 or yuv422/yuv420.
> + *
> + * Returns:
> + * - 0: The maximum overall throughput capability is not indicated by
> + * the device separately and it must be determined from the per-slice
> + * max throughput (see @drm_dp_dsc_branch_slice_max_throughput())
> + * and the maximum slice count supported by the device.
> + * - > 0: The maximum overall DSC pixel throughput supported by the branch
> + * device in kPixels/sec.
> + */
> +int drm_dp_dsc_branch_max_overall_throughput(const u8 dsc_branch_dpcd[DP_DSC_BRANCH_CAP_SIZE],
> + bool is_rgb_yuv444)
> +{
> + int throughput;
> +
> + if (is_rgb_yuv444)
> + throughput = dsc_branch_dpcd_cap(dsc_branch_dpcd,
> + DP_DSC_BRANCH_OVERALL_THROUGHPUT_0);
> + else
> + throughput = dsc_branch_dpcd_cap(dsc_branch_dpcd,
> + DP_DSC_BRANCH_OVERALL_THROUGHPUT_1);
> +
> + switch (throughput) {
> + case 0:
> + return 0;
> + case 1:
> + return 680000;
> + default:
> + return 600000 + 50000 * throughput;
> + }
> +}
> +EXPORT_SYMBOL(drm_dp_dsc_branch_max_overall_throughput);
> +
> +/**
> + * drm_dp_dsc_branch_max_line_width() - Branch device's max DSC line width
> + * @dsc_branch_dpcd: DSC branch capabilities from DPCD
> + *
> + * Return the branch device's maximum overall DSC line width, based on
> + * the device's @dsc_branch_dpcd capabilities.
> + *
> + * Returns:
> + * - 0: The maximum line width is not indicated by the device
> + * separately and it must be determined from the maximum
> + * slice count and slice-width supported by the device.
> + * - %-EINVAL: The device indicates an invalid maximum line width
> + * (< 5120 pixels).
> + * - >= 5120: The maximum line width in pixels.
> + */
> +int drm_dp_dsc_branch_max_line_width(const u8 dsc_branch_dpcd[DP_DSC_BRANCH_CAP_SIZE])
> +{
> + int line_width = dsc_branch_dpcd_cap(dsc_branch_dpcd, DP_DSC_BRANCH_MAX_LINE_WIDTH);
> +
> + switch (line_width) {
> + case 0:
> + return 0;
> + case 1 ... 15:
> + return -EINVAL;
> + default:
> + return line_width * 320;
> + }
> +}
> +EXPORT_SYMBOL(drm_dp_dsc_branch_max_line_width);
> +
> static int drm_dp_read_lttpr_regs(struct drm_dp_aux *aux,
> const u8 dpcd[DP_RECEIVER_CAP_SIZE], int address,
> u8 *buf, int buf_size)
> diff --git a/include/drm/display/drm_dp.h b/include/drm/display/drm_dp.h
> index cf318e3ddb5c5..e4eebabab9759 100644
> --- a/include/drm/display/drm_dp.h
> +++ b/include/drm/display/drm_dp.h
> @@ -258,6 +258,8 @@
> # define DP_DSC_RC_BUF_BLK_SIZE_4 0x1
> # define DP_DSC_RC_BUF_BLK_SIZE_16 0x2
> # define DP_DSC_RC_BUF_BLK_SIZE_64 0x3
> +# define DP_DSC_THROUGHPUT_MODE_0_DELTA_SHIFT 3 /* DP 2.1a, in units of 2 MPixels/sec */
> +# define DP_DSC_THROUGHPUT_MODE_0_DELTA_MASK (0x1f << DP_DSC_THROUGHPUT_MODE_0_DELTA_SHIFT)
>
> #define DP_DSC_RC_BUF_SIZE 0x063
>
> @@ -1686,6 +1688,7 @@ enum drm_dp_phy {
> #define DP_BRANCH_OUI_HEADER_SIZE 0xc
> #define DP_RECEIVER_CAP_SIZE 0xf
> #define DP_DSC_RECEIVER_CAP_SIZE 0x10 /* DSC Capabilities 0x60 through 0x6F */
> +#define DP_DSC_BRANCH_CAP_SIZE 3
> #define EDP_PSR_RECEIVER_CAP_SIZE 2
> #define EDP_DISPLAY_CTL_CAP_SIZE 5
> #define DP_LTTPR_COMMON_CAP_SIZE 8
> diff --git a/include/drm/display/drm_dp_helper.h b/include/drm/display/drm_dp_helper.h
> index e438c44409952..df2f24b950e4c 100644
> --- a/include/drm/display/drm_dp_helper.h
> +++ b/include/drm/display/drm_dp_helper.h
> @@ -211,6 +211,11 @@ u8 drm_dp_dsc_sink_max_slice_count(const u8 dsc_dpcd[DP_DSC_RECEIVER_CAP_SIZE],
> u8 drm_dp_dsc_sink_line_buf_depth(const u8 dsc_dpcd[DP_DSC_RECEIVER_CAP_SIZE]);
> int drm_dp_dsc_sink_supported_input_bpcs(const u8 dsc_dpc[DP_DSC_RECEIVER_CAP_SIZE],
> u8 dsc_bpc[3]);
> +int drm_dp_dsc_sink_max_slice_throughput(const u8 dsc_dpcd[DP_DSC_RECEIVER_CAP_SIZE],
> + int peak_pixel_rate, bool is_rgb_yuv444);
> +int drm_dp_dsc_branch_max_overall_throughput(const u8 dsc_branch_dpcd[DP_DSC_BRANCH_CAP_SIZE],
> + bool is_rgb_yuv444);
> +int drm_dp_dsc_branch_max_line_width(const u8 dsc_branch_dpcd[DP_DSC_BRANCH_CAP_SIZE]);
>
> static inline bool
> drm_dp_sink_supports_dsc(const u8 dsc_dpcd[DP_DSC_RECEIVER_CAP_SIZE])
> --
> 2.49.1
--
Ville Syrjälä
Intel
next prev parent reply other threads:[~2025-09-29 10:12 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-26 21:12 [PATCH v4 0/6] drm/i915/dp: Work around a DSC pixel throughput issue Imre Deak
2025-09-26 21:12 ` [PATCH v4 1/6] drm/dp: Add quirk for Synaptics DSC throughput link-bpp limit Imre Deak
2025-09-26 21:12 ` [PATCH v4 2/6] drm/dp: Add helpers to query the branch DSC max throughput/line-width Imre Deak
2025-09-29 6:36 ` [PATCH v5 " Imre Deak
2025-09-29 9:00 ` Dmitry Baryshkov
2025-09-29 10:10 ` Imre Deak
2025-09-30 5:30 ` Dmitry Baryshkov
2025-09-30 11:38 ` Imre Deak
2025-09-30 20:04 ` Dmitry Baryshkov
2025-09-30 21:03 ` Imre Deak
2025-09-29 10:12 ` Ville Syrjälä [this message]
2025-09-29 10:47 ` Imre Deak
2025-09-26 21:12 ` [PATCH v4 3/6] drm/i915/dp: Calculate DSC slice count based on per-slice peak throughput Imre Deak
2025-09-26 21:12 ` [PATCH v4 4/6] drm/i915/dp: Pass DPCD device descriptor to intel_dp_get_dsc_sink_cap() Imre Deak
2025-09-26 21:12 ` [PATCH v4 5/6] drm/i915/dp: Verify branch devices' overall pixel throughput/line width Imre Deak
2025-09-26 21:12 ` [PATCH v4 6/6] drm/i915/dp: Handle Synaptics DSC throughput link-bpp quirk Imre Deak
2025-09-26 22:04 ` ✓ i915.CI.BAT: success for drm/i915/dp: Work around a DSC pixel throughput issue (rev5) Patchwork
2025-09-27 3:08 ` ✓ i915.CI.Full: " Patchwork
2025-09-29 7:24 ` ✓ i915.CI.BAT: success for drm/i915/dp: Work around a DSC pixel throughput issue (rev6) Patchwork
2025-09-29 9:02 ` ✗ i915.CI.Full: failure " Patchwork
2025-09-29 21:47 ` [PATCH v4 0/6] drm/i915/dp: Work around a DSC pixel throughput issue Sharma, Swati2
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=aNpbf4HJDrtbsuex@intel.com \
--to=ville.syrjala@linux.intel.com \
--cc=dri-devel@lists.freedesktop.org \
--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).