From: Hamza Mahfooz <hamza.mahfooz@amd.com>
To: <amd-gfx@lists.freedesktop.org>
Cc: stylon.wang@amd.com, Sunpeng.Li@amd.com, Harry.Wentland@amd.com,
qingqing.zhuo@amd.com, Rodrigo.Siqueira@amd.com,
roman.li@amd.com, Wenjing Liu <wenjing.liu@amd.com>,
Daniel Wheeler <daniel.wheeler@amd.com>,
solomon.chiu@amd.com, jerry.zuo@amd.com,
Aurabindo.Pillai@amd.com, hamza.mahfooz@amd.com,
wayne.lin@amd.com, Jun Lei <jun.lei@amd.com>,
Bhawanpreet.Lakha@amd.com, agustin.gutierrez@amd.com,
pavle.kotarac@amd.com
Subject: [PATCH 13/21] drm/amd/display: switch to new ODM policy for windowed MPO ODM support
Date: Wed, 23 Aug 2023 11:58:15 -0400 [thread overview]
Message-ID: <20230823160347.176991-14-hamza.mahfooz@amd.com> (raw)
In-Reply-To: <20230823160347.176991-1-hamza.mahfooz@amd.com>
From: Wenjing Liu <wenjing.liu@amd.com>
We need to align windowed MPO ODM support on DCN3x with new ODM policy.
Reviewed-by: Jun Lei <jun.lei@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
---
.../drm/amd/display/dc/dcn32/dcn32_resource.c | 84 ++++++++++++++-----
1 file changed, 65 insertions(+), 19 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c
index 88c9dbbacbd3..fd12791995a7 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c
@@ -1885,6 +1885,67 @@ bool dcn32_validate_bandwidth(struct dc *dc,
return out;
}
+static bool should_allow_odm_power_optimization(struct dc *dc,
+ struct dc_state *context)
+{
+ struct dc_stream_state *stream = context->streams[0];
+
+ /*
+ * this debug flag allows us to disable ODM power optimization feature
+ * unconditionally. we force the feature off if this is set to false.
+ */
+ if (!dc->debug.enable_single_display_2to1_odm_policy)
+ return false;
+
+ /* current design and test coverage is only limited to allow ODM power
+ * optimization for single stream. Supporting it for multiple streams
+ * use case would require additional algorithm to decide how to
+ * optimize power consumption when there are not enough free pipes to
+ * allocate for all the streams. This level of optimization would
+ * require multiple attempts of revalidation to make an optimized
+ * decision. Unfortunately We do not support revalidation flow in
+ * current version of DML.
+ */
+ if (context->stream_count != 1)
+ return false;
+
+ /*
+ * Our hardware doesn't support ODM for HDMI TMDS
+ */
+ if (dc_is_hdmi_signal(stream->signal))
+ return false;
+
+ /*
+ * ODM Combine 2:1 requires horizontal timing divisible by 2 so each
+ * ODM segment has the same size.
+ */
+ if (!is_h_timing_divisible_by_2(stream))
+ return false;
+
+ /*
+ * No power benefits if the timing's pixel clock is not high enough to
+ * raise display clock from minimum power state.
+ */
+ if (stream->timing.pix_clk_100hz * 100 <= DCN3_2_VMIN_DISPCLK_HZ)
+ return false;
+
+ /* the new ODM power optimization feature reduces software design
+ * limitation and allows ODM power optimization to be supported even
+ * with presence of overlay planes. The new feature is enabled based on
+ * enable_windowed_mpo_odm flag. If the flag is not set, we limit our
+ * feature scope due to previous software design limitation */
+ if (!dc->config.enable_windowed_mpo_odm) {
+ if (context->stream_status[0].plane_count != 1)
+ return false;
+
+ if (stream->src.width >= 5120 &&
+ stream->src.width > stream->dst.width)
+ return false;
+ }
+
+ return true;
+}
+
int dcn32_populate_dml_pipes_from_context(
struct dc *dc, struct dc_state *context,
display_e2e_pipe_params_st *pipes,
@@ -1895,35 +1956,20 @@ int dcn32_populate_dml_pipes_from_context(
struct pipe_ctx *pipe = NULL;
bool subvp_in_use = false;
struct dc_crtc_timing *timing;
- bool vsr_odm_support = false;
dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
- /* Determine whether we will apply ODM 2to1 policy:
- * Applies to single display and where the number of planes is less than 3.
- * For 3 plane case ( 2 MPO planes ), we will not set the policy for the MPO pipes.
- *
+ /*
* Apply pipe split policy first so we can predict the pipe split correctly
* (dcn32_predict_pipe_split).
*/
for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
if (!res_ctx->pipe_ctx[i].stream)
continue;
- pipe = &res_ctx->pipe_ctx[i];
- timing = &pipe->stream->timing;
-
- pipes[pipe_cnt].pipe.dest.odm_combine_policy = dm_odm_combine_policy_dal;
- vsr_odm_support = (res_ctx->pipe_ctx[i].stream->src.width >= 5120 &&
- res_ctx->pipe_ctx[i].stream->src.width > res_ctx->pipe_ctx[i].stream->dst.width);
- if (context->stream_count == 1 &&
- context->stream_status[0].plane_count == 1 &&
- !dc_is_hdmi_signal(res_ctx->pipe_ctx[i].stream->signal) &&
- is_h_timing_divisible_by_2(res_ctx->pipe_ctx[i].stream) &&
- pipe->stream->timing.pix_clk_100hz * 100 > DCN3_2_VMIN_DISPCLK_HZ &&
- dc->debug.enable_single_display_2to1_odm_policy &&
- !vsr_odm_support) { //excluding 2to1 ODM combine on >= 5k vsr
+ if (should_allow_odm_power_optimization(dc, context))
pipes[pipe_cnt].pipe.dest.odm_combine_policy = dm_odm_combine_policy_2to1;
- }
+ else
+ pipes[pipe_cnt].pipe.dest.odm_combine_policy = dm_odm_combine_policy_dal;
pipe_cnt++;
}
--
2.41.0
next prev parent reply other threads:[~2023-08-23 16:04 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-08-23 15:58 [PATCH 00/21] DC Patches August 23, 2023 Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 01/21] Partially revert "drm/amd/display: update add plane to context logic with a new algorithm" Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 02/21] drm/amd/display: Add support for 1080p SubVP to reduce idle power Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 03/21] drm/amd/display: Add smu write msg id fail retry process Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 04/21] drm/amd/display: update blank state on ODM changes Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 05/21] drm/amd/display: always switch off ODM before committing more streams Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 06/21] drm/amd/display: Add DPIA Link Encoder Assignment Fix Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 07/21] drm/amd/display: add comments to add plane functions Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 08/21] drm/amd/display: rename function to add otg master for stream Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 09/21] drm/amd/display: add new resource interface for acquiring sec opp heads and release pipe Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 10/21] drm/amd/display: add new resource interfaces to update odm mpc slice count Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 11/21] drm/amd/display: add more pipe resource interfaces Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 12/21] drm/amd/display: use new pipe allocation interface in dcn32 fpu Hamza Mahfooz
2023-08-23 15:58 ` Hamza Mahfooz [this message]
2023-08-23 15:58 ` [PATCH 14/21] drm/amd/display: add pipe topology update log Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 15/21] drm/amd/display: Remove wait while locked Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 16/21] drm/amd/display: Skip dmub memory flush when not needed Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 17/21] drm/amd/display: Fix incorrect comment Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 18/21] drm/amd/display: correct z8_watermark 16bit to 20bit mask Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 19/21] drm/amd/display: Add debugfs interface for ODM combine info Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 20/21] drm/amd/display: fix pipe topology logging error Hamza Mahfooz
2023-08-23 15:58 ` [PATCH 21/21] drm/amd/display: 3.2.249 Hamza Mahfooz
2023-08-28 13:21 ` [PATCH 00/21] DC Patches August 23, 2023 Wheeler, Daniel
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=20230823160347.176991-14-hamza.mahfooz@amd.com \
--to=hamza.mahfooz@amd.com \
--cc=Aurabindo.Pillai@amd.com \
--cc=Bhawanpreet.Lakha@amd.com \
--cc=Harry.Wentland@amd.com \
--cc=Rodrigo.Siqueira@amd.com \
--cc=Sunpeng.Li@amd.com \
--cc=agustin.gutierrez@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=daniel.wheeler@amd.com \
--cc=jerry.zuo@amd.com \
--cc=jun.lei@amd.com \
--cc=pavle.kotarac@amd.com \
--cc=qingqing.zhuo@amd.com \
--cc=roman.li@amd.com \
--cc=solomon.chiu@amd.com \
--cc=stylon.wang@amd.com \
--cc=wayne.lin@amd.com \
--cc=wenjing.liu@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