From: Chenyu Chen <chen-yu.chen@amd.com>
To: <amd-gfx@lists.freedesktop.org>
Cc: Harry Wentland <harry.wentland@amd.com>,
Leo Li <sunpeng.li@amd.com>,
Aurabindo Pillai <aurabindo.pillai@amd.com>,
Roman Li <roman.li@amd.com>, Wayne Lin <wayne.lin@amd.com>,
Tom Chung <chiahsuan.chung@amd.com>,
"Fangzhi Zuo" <jerry.zuo@amd.com>,
Dan Wheeler <daniel.wheeler@amd.com>, Ray Wu <Ray.Wu@amd.com>,
Ivan Lipski <ivan.lipski@amd.com>, Alex Hung <alex.hung@amd.com>,
James Lin <PingLei.Lin@amd.com>,
Chenyu Chen <Chen-Yu.Chen@amd.com>,
Alvin Lee <Alvin.Lee2@amd.com>, Aric Cyr <aric.cyr@amd.com>,
Taimur Hassan <syed.hassan@amd.com>,
Chenyu Chen <chen-yu.chen@amd.com>
Subject: [PATCH 02/66] drm/amd/display: Update LLS and UPSP programming paths
Date: Tue, 8 Sep 2026 19:30:55 +0800 [thread overview]
Message-ID: <20260908113338.2433445-3-chen-yu.chen@amd.com> (raw)
In-Reply-To: <20260908113338.2433445-1-chen-yu.chen@amd.com>
From: Alvin Lee <Alvin.Lee2@amd.com>
[Description]
- UPSP should only be enabled when we want linear scaling for 420/422
- In DC for non-FP16 formats that are told to scale in non-linear, program
bypass for PRE_GAM since these formats are already non-linear (i.e.,
YUV420, YUV422, RGB, etc.)
Reviewed-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Taimur Hassan <syed.hassan@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Signed-off-by: Chenyu Chen <chen-yu.chen@amd.com>
---
.../gpu/drm/amd/display/dc/core/dc_resource.c | 18 ++++++++++++----
.../drm/amd/display/dc/dpp/dcn50/dcn50_dpp.c | 20 +++++++++++++-----
.../drm/amd/display/dc/dpp/dcn50/dcn50_dpp.h | 3 ++-
.../amd/display/dc/hwss/dcn32/dcn32_hwseq.c | 21 ++++++++++++++++++-
drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h | 3 ++-
drivers/gpu/drm/amd/display/dc/inc/resource.h | 3 ++-
6 files changed, 55 insertions(+), 13 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
index f9b8e9474b4c..593c1fd537ae 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
@@ -1557,11 +1557,21 @@ void resource_build_test_pattern_params(struct resource_context *res_ctx,
}
}
-enum upsp_mode resource_is_upsp_required(enum surface_pixel_format format)
+enum upsp_mode resource_is_upsp_required(enum surface_pixel_format format,
+ enum dc_scaling_linearity scaling_linearity)
{
- if (format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN && format <= SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb) //420 Formats
+ bool scaling_in_linear = (scaling_linearity == DC_SCALING_LINEARITY_LINEAR);
+ bool is_420_format = (format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN &&
+ format <= SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb);
+ bool is_422_format = (format > SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb &&
+ format < SURFACE_PIXEL_FORMAT_SUBSAMPLE_END);
+
+ /* UPSP (chroma upsampling) is only needed when subsampled YUV is scaled in
+ * linear space.
+ */
+ if (scaling_in_linear && is_420_format)
return UPSP_HORIZONTAL_VERTICAL_UPSAMPLING;
- if (format > SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb && format < SURFACE_PIXEL_FORMAT_SUBSAMPLE_END) //422 Formats
+ if (scaling_in_linear && is_422_format)
return UPSP_HORIZONTAL_UPSAMPLING_ONLY;
return UPSP_BYPASS;
}
@@ -1611,7 +1621,7 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx)
pipe_ctx->plane_res.scl_data.lb_params.depth = LB_PIXEL_DEPTH_30BPP;
pipe_ctx->plane_res.scl_data.lb_params.alpha_en = plane_state->per_pixel_alpha;
- pipe_ctx->plane_res.scl_data.upsp = resource_is_upsp_required(plane_state->format);
+ pipe_ctx->plane_res.scl_data.upsp = resource_is_upsp_required(plane_state->format, plane_state->scaling_linearity);
// Convert pipe_ctx to respective input params for SPL
translate_SPL_in_params_from_pipe_ctx(pipe_ctx, spl_in);
diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn50/dcn50_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn50/dcn50_dpp.c
index 04cab7d3a5c2..38739ed048fa 100644
--- a/drivers/gpu/drm/amd/display/dc/dpp/dcn50/dcn50_dpp.c
+++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn50/dcn50_dpp.c
@@ -25,17 +25,27 @@
void dpp50_set_pregam_state(
struct dpp *dpp_base,
enum dc_transfer_func_predefined tr,
- enum dc_scaling_linearity scaling)
+ enum dc_scaling_linearity scaling,
+ bool source_is_linear)
{
struct dcn50_dpp *dpp = TO_DCN50_DPP(dpp_base);
enum pregam_mode pre_degam_en = PREGAM_DEGAM;
enum degam_lut degamma_lut_selection = 0;
if (scaling == DC_SCALING_LINEARITY_SOURCE) {
- //If scaling in non-linear, apply regamma
- REG_SET_2(PRE_GAM, 0,
- PRE_GAM_MODE, PREGAM_REGAM,
- PRE_REGAM_SELECT, REGAM_20);
+ if (source_is_linear) {
+ //Linear source data (e.g. FP16) must be de-linearized
+ //(apply regamma) before scaling in source/non-linear space.
+ REG_SET_2(PRE_GAM, 0,
+ PRE_GAM_MODE, PREGAM_REGAM,
+ PRE_REGAM_SELECT, REGAM_20);
+ } else {
+ //Non-linear source data (e.g. YUV) is already in source
+ //space, so scale directly with no pregam conversion.
+ REG_SET_2(PRE_GAM, 0,
+ PRE_GAM_MODE, PREGAM_BYPASS,
+ PRE_DEGAM_SELECT, 0);
+ }
} else {
//If scaling in linear, apply degamma based on TF
switch (tr) {
diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn50/dcn50_dpp.h b/drivers/gpu/drm/amd/display/dc/dpp/dcn50/dcn50_dpp.h
index 7f5d3fa6bbef..c81e910ae095 100644
--- a/drivers/gpu/drm/amd/display/dc/dpp/dcn50/dcn50_dpp.h
+++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn50/dcn50_dpp.h
@@ -82,6 +82,7 @@ void dpp50_dpp_setup(
void dpp50_set_pregam_state(
struct dpp *dpp_base,
enum dc_transfer_func_predefined tr,
- enum dc_scaling_linearity scaling);
+ enum dc_scaling_linearity scaling,
+ bool source_is_linear);
#endif /* __DCN50_DPP_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c
index 79be5c385280..430966af3808 100644
--- a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c
@@ -528,6 +528,24 @@ bool dcn32_set_mcm_luts(struct dc *dc, struct dpp *dpp, struct hubp *hubp,
return result;
}
+/*
+ * FP16 / 64bpp 16161616 surfaces store pixel data in linear light. These are
+ * the only source formats that must be de-linearized before scaling in
+ * source/non-linear space; every other format is already non-linear.
+ */
+static bool is_source_pixel_format_linear(enum surface_pixel_format format)
+{
+ switch (format) {
+ case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
+ case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616:
+ case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F:
+ case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
+ return true;
+ default:
+ return false;
+ }
+}
+
bool dcn32_set_input_transfer_func(struct set_input_transfer_func_params *params)
{
struct dce_hwseq *hws = params->dc->hwseq;
@@ -547,7 +565,8 @@ bool dcn32_set_input_transfer_func(struct set_input_transfer_func_params *params
tf = plane_state->in_transfer_func.tf;
if (dpp->funcs->dpp_set_pregam_state)
- dpp->funcs->dpp_set_pregam_state(dpp, tf, plane_state->scaling_linearity);
+ dpp->funcs->dpp_set_pregam_state(dpp, tf, plane_state->scaling_linearity,
+ is_source_pixel_format_linear(plane_state->format));
else
dpp->funcs->dpp_set_pre_degam(dpp, tf);
diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h
index b31b144d2e8f..930b5533256f 100644
--- a/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h
@@ -378,7 +378,8 @@ struct dpp_funcs {
void (*dpp_set_pregam_state)(
struct dpp *dpp_base,
enum dc_transfer_func_predefined tr,
- enum dc_scaling_linearity scaling);
+ enum dc_scaling_linearity scaling,
+ bool source_is_linear);
void (*dpp_program_upsp)(
struct dpp *dpp_base,
const struct dscl_prog_data *dscl_prog_data);
diff --git a/drivers/gpu/drm/amd/display/dc/inc/resource.h b/drivers/gpu/drm/amd/display/dc/inc/resource.h
index 8e21aaac06b3..9c6794f8dec0 100644
--- a/drivers/gpu/drm/amd/display/dc/inc/resource.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/resource.h
@@ -123,7 +123,8 @@ void resource_build_test_pattern_params(
struct resource_context *res_ctx,
struct pipe_ctx *pipe_ctx);
-enum upsp_mode resource_is_upsp_required(enum surface_pixel_format format);
+enum upsp_mode resource_is_upsp_required(enum surface_pixel_format format,
+ enum dc_scaling_linearity scaling_linearity);
bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx);
--
2.43.0
next prev parent reply other threads:[~2026-09-08 11:34 UTC|newest]
Thread overview: 67+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-08 11:30 [PATCH 00/66] DC Patches Sep 14 2026 Chenyu Chen
2026-09-08 11:30 ` [PATCH 01/66] drm/amd/display: Decouple cursor offload hwss executors from pipe context Chenyu Chen
2026-09-08 11:30 ` Chenyu Chen [this message]
2026-09-08 11:30 ` [PATCH 03/66] drm/amd/display: Refactor RMCM into a separate module Chenyu Chen
2026-09-08 11:30 ` [PATCH 04/66] drm/amd/display: Remove SDPIF_PORT_CONTROL programming for DCN31/35/42 Chenyu Chen
2026-09-08 11:30 ` [PATCH 05/66] drm/amd/display: Test sink stream creation Chenyu Chen
2026-09-08 11:30 ` [PATCH 06/66] drm/amd/display: Test connector init helper Chenyu Chen
2026-09-08 11:31 ` [PATCH 07/66] drm/amd/display: Test HDMI connector init Chenyu Chen
2026-09-08 11:31 ` [PATCH 08/66] drm/amd/display: Test FreeSync caps update Chenyu Chen
2026-09-08 11:31 ` [PATCH 09/66] drm/amd/display: Test connector init Chenyu Chen
2026-09-08 11:31 ` [PATCH 10/66] drm/amd/display: Test forced atomic commit Chenyu Chen
2026-09-08 11:31 ` [PATCH 11/66] drm/amd/display: Test DCC reject for multi-plane format Chenyu Chen
2026-09-08 11:31 ` [PATCH 12/66] drm/amd/display: Test modifier list growth failure Chenyu Chen
2026-09-08 11:31 ` [PATCH 13/66] drm/amd/display: Test pre-GFX9 plane buffer attributes Chenyu Chen
2026-09-08 11:31 ` [PATCH 14/66] drm/amd/display: Test accepted plane atomic check Chenyu Chen
2026-09-08 11:31 ` [PATCH 15/66] drm/amd/display: Test cursor update without DC stream Chenyu Chen
2026-09-08 11:31 ` [PATCH 16/66] drm/amd/display: Test panic flush DCC teardown Chenyu Chen
2026-09-08 11:31 ` [PATCH 17/66] drm/amd/display: Test optional plane property creation Chenyu Chen
2026-09-08 11:31 ` [PATCH 18/66] drm/amd/display: Add option for certain panels to disable FEC Chenyu Chen
2026-09-08 11:31 ` [PATCH 19/66] drm/amd/display: Build MST DSC helpers for KUnit Chenyu Chen
2026-09-08 11:31 ` [PATCH 20/66] drm/amd/display: Test oversized AUX transfer Chenyu Chen
2026-09-08 11:31 ` [PATCH 21/66] drm/amd/display: Test MST connector creation Chenyu Chen
2026-09-08 11:31 ` [PATCH 22/66] drm/amd/display: Test link bandwidth readback Chenyu Chen
2026-09-08 11:31 ` [PATCH 23/66] drm/amd/display: Test cascaded Panamera check Chenyu Chen
2026-09-08 11:31 ` [PATCH 24/66] drm/amd/display: Test DSC caps validation Chenyu Chen
2026-09-08 11:31 ` [PATCH 25/66] drm/amd/display: Test MST port mode support Chenyu Chen
2026-09-08 11:31 ` [PATCH 26/66] drm/amd/display: Test FRL bandwidth lookup Chenyu Chen
2026-09-08 11:31 ` [PATCH 27/66] drm/amd/display: Test DSC precompute helpers Chenyu Chen
2026-09-08 11:31 ` [PATCH 28/66] drm/amd/display: Test DSC recompute check Chenyu Chen
2026-09-08 11:31 ` [PATCH 29/66] drm/amd/display: Test DSC config computation Chenyu Chen
2026-09-08 11:31 ` [PATCH 30/66] drm/amd/display: Test per-link DSC configs Chenyu Chen
2026-09-08 11:31 ` [PATCH 31/66] drm/amd/display: Add urgent assertion counter probe Chenyu Chen
2026-09-08 11:31 ` [PATCH 32/66] drm/amd/display: Add debug option to force optional UCLK support Chenyu Chen
2026-09-08 11:31 ` [PATCH 33/66] drm/amd/display: Honor forced RGB pixel encoding Chenyu Chen
2026-09-08 11:31 ` [PATCH 34/66] drm/amd/display: Add Replay cumulative residency query Chenyu Chen
2026-09-08 11:31 ` [PATCH 35/66] drm/amd/display: Force DSC to 8bpp for MST DP tunneling over USB4 Chenyu Chen
2026-09-08 11:31 ` [PATCH 36/66] drm/amd/display: Force DSC to 8bpp for SST " Chenyu Chen
2026-09-08 11:31 ` [PATCH 37/66] drm/amd/display: Fix peak bandwidth measurement sequence Chenyu Chen
2026-09-08 11:31 ` [PATCH 38/66] drm/amd/display: Add instance field to struct mpc Chenyu Chen
2026-09-08 11:31 ` [PATCH 39/66] drm/amd/display: Enable back alt-ch Chenyu Chen
2026-09-08 11:31 ` [PATCH 40/66] drm/amd/display: Decouple HUBP_UPDATE_PLANE_ADDR from pipe_ctx Chenyu Chen
2026-09-08 11:31 ` [PATCH 41/66] drm/amd/display: Cleanup DMUB command submission interfaces Chenyu Chen
2026-09-08 11:31 ` [PATCH 42/66] drm/amd/display: Enable power gating on dcn42b Chenyu Chen
2026-09-08 11:31 ` [PATCH 43/66] drm/amd/display: Bound DSC power gating loop by num_dsc Chenyu Chen
2026-09-08 11:31 ` [PATCH 44/66] drm/amd/display: Add lock-free memory pool Chenyu Chen
2026-09-08 11:31 ` [PATCH 45/66] drm/amd/display: Rename lock_and_validation_needed to needs_dc_state_realloc Chenyu Chen
2026-09-08 11:31 ` [PATCH 46/66] drm/amd/display: Attach only plane updates that actually changed Chenyu Chen
2026-09-08 11:31 ` [PATCH 47/66] drm/amd/display: Request DMUB HW cursor offload Chenyu Chen
2026-09-08 11:31 ` [PATCH 48/66] drm/amd/display: Send stream_update to DC only when it changed Chenyu Chen
2026-09-08 11:31 ` [PATCH 49/66] drm/amd/display: Drop dead update_type param from update_planes_and_stream_adapter Chenyu Chen
2026-09-08 11:31 ` [PATCH 50/66] drm/amd/display: Flush ISM work before releasing the stream Chenyu Chen
2026-09-08 11:31 ` [PATCH 51/66] drm/amd/display: Cap DML2.1 vmin ODM combine at 2:1 for eDP Chenyu Chen
2026-09-08 11:31 ` [PATCH 52/66] drm/amd/display: Add is_odm_enabled callback to skip init_odm on active ODM pipes Chenyu Chen
2026-09-08 11:31 ` [PATCH 53/66] drm/amd/display: Program DCC as part of address update Chenyu Chen
2026-09-08 11:31 ` [PATCH 54/66] drm/amd/display: Add instance field to struct dccg Chenyu Chen
2026-09-08 11:31 ` [PATCH 55/66] drm/amd/display: Add SPDX license identifier to dcn30_dpp_cm.c Chenyu Chen
2026-09-08 11:31 ` [PATCH 56/66] drm/amd/display: Remove MALL capabilities from DCN42B Chenyu Chen
2026-09-08 11:31 ` [PATCH 57/66] drm/amd/display: Remove MALL capabilities from DCN42B bounding box Chenyu Chen
2026-09-08 11:31 ` [PATCH 58/66] drm/amd/display: Atomize IRQ register read/modify/write ops Chenyu Chen
2026-09-08 11:31 ` [PATCH 59/66] drm/amd/display: Return success status from check_mode_supported Chenyu Chen
2026-09-08 11:31 ` [PATCH 60/66] drm/amd/display: Add condition to skip MALL calculations if there is no MALL Chenyu Chen
2026-09-08 11:31 ` [PATCH 61/66] drm/amd/display: Fix HDMI FRL audio enable Chenyu Chen
2026-09-08 11:31 ` [PATCH 62/66] drm/amd/display: Cast DP DTO pixel clock math to avoid overflow and narrowing Chenyu Chen
2026-09-08 11:31 ` [PATCH 63/66] drm/amd/display: Add inbox0 HW lock helpers for DCN35 Chenyu Chen
2026-09-08 11:31 ` [PATCH 64/66] drm/amd/display: Unify fast update classification paths Chenyu Chen
2026-09-08 11:31 ` [PATCH 65/66] drm/amd/display: Use unsigned types for FRL cap check params and HPO read_state Chenyu Chen
2026-09-08 11:31 ` [PATCH 66/66] drm/amd/display: Promote DC to 3.2.398 Chenyu Chen
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=20260908113338.2433445-3-chen-yu.chen@amd.com \
--to=chen-yu.chen@amd.com \
--cc=Alvin.Lee2@amd.com \
--cc=PingLei.Lin@amd.com \
--cc=Ray.Wu@amd.com \
--cc=alex.hung@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=aric.cyr@amd.com \
--cc=aurabindo.pillai@amd.com \
--cc=chiahsuan.chung@amd.com \
--cc=daniel.wheeler@amd.com \
--cc=harry.wentland@amd.com \
--cc=ivan.lipski@amd.com \
--cc=jerry.zuo@amd.com \
--cc=roman.li@amd.com \
--cc=sunpeng.li@amd.com \
--cc=syed.hassan@amd.com \
--cc=wayne.lin@amd.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.