From: Fangzhi Zuo <jerry.zuo@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>,
Alexander Chechik <alexander.chechik@amd.com>,
"Dillon Varone" <dillon.varone@amd.com>,
Matthew Stewart <matthew.stewart2@amd.com>
Subject: [PATCH 43/49] drm/amd/display: enforce UCLK pstate support in mode_support
Date: Thu, 23 Jul 2026 16:13:51 -0400 [thread overview]
Message-ID: <20260723201908.373300-44-jerry.zuo@amd.com> (raw)
In-Reply-To: <20260723201908.373300-1-jerry.zuo@amd.com>
From: Alexander Chechik <alexander.chechik@amd.com>
[Why]
mode_support does not require UCLK pstate today, so later PMO optimization
stages can push a plane's VActive latency-hiding margin below zero without
rechecking that the config still supports UCLK pstate. This can blank the
display on high-bandwidth configs.
[How]
Plumb the PMO-selected per-plane pstate method into mode_support and fail
the config only when UCLK pstate is required (method != na) but not
supported. For planes committed to a vactive method, require a non-negative
VActive latency-hiding margin, and skip the check when all streams are
blanked. No-op the PMO DCN42 pstate test (returning false only on the
initial candidate so the optimize/FAMS2 stage-3 setup still runs), since
reserved time is guaranteed by the override and the vactive margin is now
enforced in core mode_support.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Alexander Chechik <alexander.chechik@amd.com>
Signed-off-by: Matthew Stewart <matthew.stewart2@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
---
.../dml21/src/dml2_core/dml2_core_dcn4.c | 8 +++++
.../src/dml2_core/dml2_core_dcn4_calcs.c | 18 +++++++++++
.../dml21/src/dml2_pmo/dml2_pmo_dcn42.c | 32 ++++++-------------
3 files changed, 36 insertions(+), 22 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.c
index 727b01ca18bf..45f1bf5e14bf 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.c
@@ -554,6 +554,14 @@ bool core_dcn4_mode_support(struct dml2_core_mode_support_in_out *in_out)
l->mode_support_ex_params.min_clk_index = in_out->min_clk_index;
l->mode_support_ex_params.out_evaluation_info = &in_out->mode_support_result.cfg_support_info.clean_me_up.support_info;
+ for (i = 0; i < l->svp_expanded_display_cfg.num_planes; i++) {
+ if (i < in_out->display_cfg->display_config.num_planes)
+ core->clean_me_up.mode_lib.ms.uclk_pstate_switch_modes[i] =
+ in_out->display_cfg->stage3.pstate_switch_modes[i];
+ else
+ core->clean_me_up.mode_lib.ms.uclk_pstate_switch_modes[i] = dml2_pstate_method_na;
+ }
+
result = dml2_core_calcs_mode_support_ex(&l->mode_support_ex_params);
in_out->mode_support_result.cfg_support_info.is_supported = result;
diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c
index 110ab26a6a39..da9a1c4e2bdc 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c
@@ -9637,6 +9637,22 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out
DML_LOG_VERBOSE("DML::%s: ROBSupport = %u\n", __func__, mode_lib->ms.support.ROBSupport);
#endif
+ mode_lib->ms.support.global_dram_clock_change_support_required = false;
+
+ if (!display_cfg->overrides.all_streams_blanked) {
+ for (k = 0; k < mode_lib->ms.num_active_planes; k++) {
+ if (mode_lib->ms.uclk_pstate_switch_modes[k] == dml2_pstate_method_na)
+ continue;
+
+ mode_lib->ms.support.global_dram_clock_change_support_required = true;
+
+ if ((mode_lib->ms.uclk_pstate_switch_modes[k] == dml2_pstate_method_vactive ||
+ mode_lib->ms.uclk_pstate_switch_modes[k] == dml2_pstate_method_fw_vactive_drr) &&
+ mode_lib->ms.VActiveLatencyHidingMargin[k] < 0)
+ mode_lib->ms.support.global_dram_clock_change_supported = false;
+ }
+ }
+
/*Mode Support, Voltage State and SOC Configuration*/
{
if (mode_lib->ms.support.ScaleRatioAndTapsSupport
@@ -9683,6 +9699,8 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out
&& mode_lib->ms.support.DCCMetaBufferSizeNotExceeded
&& !mode_lib->ms.support.ExceededMALLSize
&& mode_lib->ms.support.g6_temp_read_support
+ && (mode_lib->ms.support.global_dram_clock_change_supported
+ || !mode_lib->ms.support.global_dram_clock_change_support_required)
&& ((!display_cfg->hostvm_enable && !s->ImmediateFlipRequired) || mode_lib->ms.support.ImmediateFlipSupport)) {
DML_LOG_VERBOSE("DML::%s: mode is supported\n", __func__);
mode_lib->ms.support.ModeSupport = true;
diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn42.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn42.c
index 790aef95caa6..57a6a7d5b740 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn42.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn42.c
@@ -13,8 +13,6 @@
* configurations, ensuring p-state watermark support in the blank period only.
*/
-static const double MIN_VACTIVE_MARGIN_PCT = 0.25; // We need more than non-zero margin because DET buffer granularity can alter vactive latency hiding
-
static const struct dml2_pmo_pstate_strategy dcn42_strategy_list_1_display[] = {
// VBlank only
{
@@ -179,7 +177,7 @@ bool pmo_dcn42_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_i
// Figure out which streams can do vactive, and also build up implicit SVP and FAMS2 meta
for (stream_index = 0; stream_index < display_config->display_config.num_streams; stream_index++) {
- if (dcn4_get_vactive_pstate_margin(display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) >= (int)(MIN_VACTIVE_MARGIN_PCT * pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us))
+ if (dcn4_get_vactive_pstate_margin(display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) >= 0)
dcn42_set_bit_in_bitfield(&s->pmo_dcn4.stream_vactive_capability_mask, stream_index);
}
@@ -259,27 +257,17 @@ bool pmo_dcn42_fams2_optimize_for_pstate_support(struct dml2_pmo_optimize_for_ps
bool pmo_dcn42_test_for_pstate_support(struct dml2_pmo_test_for_pstate_support_in_out *in_out)
{
- const struct dml2_pmo_scratch *s = &in_out->instance->scratch;
- bool p_state_supported = true;
- unsigned int stream_index;
-
- if (s->pmo_dcn4.cur_pstate_candidate < 0)
+ /* Return false on the initial candidate (cur_pstate_candidate == -1) so the
+ * optimization phase runs at least one optimize iteration; otherwise the
+ * FAMS2/stage-3 setup in the optimize callback is skipped.
+ */
+ if (in_out->instance->scratch.pmo_dcn4.cur_pstate_candidate < 0)
return false;
- for (stream_index = 0; stream_index < in_out->base_display_config->display_config.num_streams; stream_index++) {
- if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_vactive) {
- if (dcn4_get_minimum_reserved_time_us_for_planes(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < (int)in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us ||
- dcn4_get_vactive_pstate_margin(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < (int)(MIN_VACTIVE_MARGIN_PCT * in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us)) {
- p_state_supported = false;
- break;
- }
- } else {
- p_state_supported = false;
- break;
- }
- }
-
- return p_state_supported;
+ /* No-op: reserved time is guaranteed by the override and vactive p-state
+ * margin is now enforced in core mode support.
+ */
+ return true;
}
bool pmo_dcn42_initialize(struct dml2_pmo_initialize_in_out *in_out)
--
2.53.0
next prev parent reply other threads:[~2026-07-23 20:20 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-07-23 20:13 [PATCH 00/49] DC Patches July 20th, 2026 Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 01/49] drm/amd/display: share common DM KUnit helpers Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 02/49] drm/amd/display: Port DCN4+ MCIF ARB programming to new format Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 03/49] drm/amd/display: Fix force FRL rate debug setting Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 04/49] drm/amd/display: Add KUnit tests for link_lock and psp SRM helpers Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 05/49] drm/amd/display: Add KUnit tests for HDCP display helpers Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 06/49] drm/amd/display: Add KUnit tests for event_callback Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 07/49] drm/amd/display: Add KUnit tests for event_property_validate Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 08/49] drm/amd/display: Add KUnit tests for watchdog and cpirq events Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 09/49] drm/amd/display: Add KUnit tests for hdcp_destroy Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 10/49] drm/amd/display: Add AV mute wait frames to dce110_set_avmute Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 11/49] drm/amd/display: revert "convert dcn42 GPIO translation to lookup tables" Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 12/49] drm/amd/display: move scaling helper to connector Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 13/49] drm/amd/display: move stutter quirk to quirks file Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 14/49] drm/amd/display: move watermarks table to pp_smu Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 15/49] drm/amd/display: move GPU mem helpers to services Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 16/49] drm/amd/display: add FreeSync/VRR module Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 17/49] drm/amd/display: add cursor module Fangzhi Zuo
2026-08-03 12:48 ` Timur Kristóf
2026-07-23 20:13 ` [PATCH 18/49] drm/amd/display: add KUnit tests for audio component get_eld Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 19/49] drm/amd/display: add KUnit tests for audio commit path Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 20/49] drm/amd/display: Use current mpc pipe in set output transfer func Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 21/49] drm/amd/display: Correct vblank_end calc for fams cmd packet Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 22/49] drm/amd/display: Add KUnit test for native backlight registration Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 23/49] drm/amd/display: Add color transfer-function tests Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 24/49] drm/amd/display: Add atomic " Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 25/49] drm/amd/display: Add CRTC and plane degamma tests Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 26/49] drm/amd/display: Add legacy plane LUT tests Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 27/49] drm/amd/display: Add truncated colorop tests Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 28/49] drm/amd/display: Add colorop LUT programming tests Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 29/49] drm/amd/display: Add KUnit tests for enable_assr Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 30/49] drm/amd/display: Add KUnit tests for update_config Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 31/49] drm/amd/display: Add KUnit tests for hdcp_create_workqueue Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 32/49] drm/amd/display: Add KUnit tests for srm_data_write and srm_data_read Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 33/49] drm/amd/display: Add KUnit tests for HDCP DDC link adapters Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 34/49] drm/amd/display: Add initialized-branch test for psp_set_srm Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 35/49] drm/amd/display: Add deeper event_property_update tests Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 36/49] drm/amd/display: adjust floating point format for gamut remap when needed Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 37/49] drm/amd/display: Reintroduce "convert dcn42 GPIO translation to lookup tables" Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 38/49] drm/amd/display: Prune per-tile Timing from Apple Studio Display Primary Tile Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 39/49] drm/amd/display: Add get replay residency function Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 40/49] drm/amd/display: Fix divide-by-zero in calculate_mcache_setting on zero viewport Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 41/49] drm/amd/display: check if dml21_add_phantom_plane() is successful Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 42/49] drm/amd/display: change dcc_rate from 1 to 2 for log use only Fangzhi Zuo
2026-07-23 20:13 ` Fangzhi Zuo [this message]
2026-07-23 20:13 ` [PATCH 44/49] drm/amd/display: add DalForceMaxDisplayClock debug option to DML2 Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 45/49] drm/amd/display: Fixes for dcn42b_soc_bb.h Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 46/49] drm/amd/display: Fix rounding errors in CalculatePrefetchSchedule Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 47/49] drm/amd/display: plumb PMO per-plane pstate methods into mode_support Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 48/49] drm/amd/display: dispatch compressed FRL cap check inside dml1_frl_cap_chk_inter Fangzhi Zuo
2026-07-23 20:13 ` [PATCH 49/49] drm/amd/display: Promote DC to 3.2.391 Fangzhi Zuo
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=20260723201908.373300-44-jerry.zuo@amd.com \
--to=jerry.zuo@amd.com \
--cc=Chen-Yu.Chen@amd.com \
--cc=PingLei.Lin@amd.com \
--cc=Ray.Wu@amd.com \
--cc=alex.hung@amd.com \
--cc=alexander.chechik@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=aurabindo.pillai@amd.com \
--cc=chiahsuan.chung@amd.com \
--cc=daniel.wheeler@amd.com \
--cc=dillon.varone@amd.com \
--cc=harry.wentland@amd.com \
--cc=ivan.lipski@amd.com \
--cc=matthew.stewart2@amd.com \
--cc=roman.li@amd.com \
--cc=sunpeng.li@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox