From: Sasha Levin <sashal@kernel.org>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org
Cc: Wenjing Liu <wenjing.liu@amd.com>,
Dillon Varone <dillon.varone@amd.com>,
Zaeem Mohamed <zaeem.mohamed@amd.com>,
Daniel Wheeler <daniel.wheeler@amd.com>,
Alex Deucher <alexander.deucher@amd.com>,
Sasha Levin <sashal@kernel.org>,
harry.wentland@amd.com, sunpeng.li@amd.com,
Rodrigo.Siqueira@amd.com, christian.koenig@amd.com,
Xinhui.Pan@amd.com, airlied@gmail.com, daniel@ffwll.ch,
jun.lei@amd.com, hamza.mahfooz@amd.com, chiahsuan.chung@amd.com,
alvin.lee2@amd.com, george.shen@amd.com, alex.hung@amd.com,
charlene.liu@amd.com, mario.limonciello@amd.com,
allen.pan@amd.com, chaitanya.dhere@amd.com,
amd-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org
Subject: [PATCH AUTOSEL 6.10 18/34] drm/amd/display: reduce ODM slice count to initial new dc state only when needed
Date: Sun, 28 Jul 2024 11:40:42 -0400 [thread overview]
Message-ID: <20240728154230.2046786-18-sashal@kernel.org> (raw)
In-Reply-To: <20240728154230.2046786-1-sashal@kernel.org>
From: Wenjing Liu <wenjing.liu@amd.com>
[ Upstream commit 9a29c4adb0997be6ba3dd92dfba14ea75a8c6ce4 ]
[why]
We need to decrease ODM slice when adding or removing planes because MPO
support takes precedence over dynamic ODM combine. However there is a case where
we remove ODM combine even for ODM combine required timing in the initial new
dc state. This is normally okay because ODM will be added back after we pass DML
bandwidth validation. However since we remove ODM combine in the initial new
state, the previous ODM pipe allocation is lost. This may cause the new plane to
take away the original secondary OPP head pipe that is still required in the new
state.
For a timing that requires ODM 2:1 but optimized with ODM 4:1, if we add an MPO
plane, we will not have enough pipe to preserve ODM 4:1. In this case we should
reduce ODM slice count then try to add the MPO plane again. By reducing, we are
gradually remove 1 ODM slice from right most side one at a time until we have
enough free pipes for the new plane. If we remove ODM combine entirely, we could
use the pipe at ODM slice index 1 as a DPP pipe for the new plane. But ODM slice
1 is still needed as the timing requires ODM 2:1. This transition is not
seamless and user will see corruption on the screen.
[how]
Remove single ODM slice one at time until we have enough pipes for a new plane.
Remove previous logic to always remove ODM combine entirely.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
.../gpu/drm/amd/display/dc/core/dc_resource.c | 45 +++++++------
.../gpu/drm/amd/display/dc/core/dc_state.c | 67 +++++++++++++------
2 files changed, 71 insertions(+), 41 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 15819416a2f36..a2ca66a268c2d 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
@@ -2508,6 +2508,17 @@ static void remove_hpo_dp_link_enc_from_ctx(struct resource_context *res_ctx,
}
}
+static int get_num_of_free_pipes(const struct resource_pool *pool, const struct dc_state *context)
+{
+ int i;
+ int count = 0;
+
+ for (i = 0; i < pool->pipe_count; i++)
+ if (resource_is_pipe_type(&context->res_ctx.pipe_ctx[i], FREE_PIPE))
+ count++;
+ return count;
+}
+
enum dc_status resource_add_otg_master_for_stream_output(struct dc_state *new_ctx,
const struct resource_pool *pool,
struct dc_stream_state *stream)
@@ -2641,37 +2652,33 @@ static bool acquire_secondary_dpp_pipes_and_add_plane(
struct dc_state *cur_ctx,
struct resource_pool *pool)
{
- struct pipe_ctx *opp_head_pipe, *sec_pipe, *tail_pipe;
+ struct pipe_ctx *sec_pipe, *tail_pipe;
+ struct pipe_ctx *opp_heads[MAX_PIPES];
+ int opp_head_count;
+ int i;
if (!pool->funcs->acquire_free_pipe_as_secondary_dpp_pipe) {
ASSERT(0);
return false;
}
- opp_head_pipe = otg_master_pipe;
- while (opp_head_pipe) {
+ opp_head_count = resource_get_opp_heads_for_otg_master(otg_master_pipe,
+ &new_ctx->res_ctx, opp_heads);
+ if (get_num_of_free_pipes(pool, new_ctx) < opp_head_count)
+ /* not enough free pipes */
+ return false;
+
+ for (i = 0; i < opp_head_count; i++) {
sec_pipe = pool->funcs->acquire_free_pipe_as_secondary_dpp_pipe(
cur_ctx,
new_ctx,
pool,
- opp_head_pipe);
- if (!sec_pipe) {
- /* try tearing down MPCC combine */
- int pipe_idx = acquire_first_split_pipe(
- &new_ctx->res_ctx, pool,
- otg_master_pipe->stream);
-
- if (pipe_idx >= 0)
- sec_pipe = &new_ctx->res_ctx.pipe_ctx[pipe_idx];
- }
-
- if (!sec_pipe)
- return false;
-
+ opp_heads[i]);
+ ASSERT(sec_pipe);
sec_pipe->plane_state = plane_state;
/* establish pipe relationship */
- tail_pipe = get_tail_pipe(opp_head_pipe);
+ tail_pipe = get_tail_pipe(opp_heads[i]);
tail_pipe->bottom_pipe = sec_pipe;
sec_pipe->top_pipe = tail_pipe;
sec_pipe->bottom_pipe = NULL;
@@ -2682,8 +2689,6 @@ static bool acquire_secondary_dpp_pipes_and_add_plane(
} else {
sec_pipe->prev_odm_pipe = NULL;
}
-
- opp_head_pipe = opp_head_pipe->next_odm_pipe;
}
return true;
}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_state.c b/drivers/gpu/drm/amd/display/dc/core/dc_state.c
index 76bb05f4d6bf3..52a1cfc5feed8 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_state.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_state.c
@@ -437,6 +437,19 @@ enum dc_status dc_state_remove_stream(
return DC_OK;
}
+static void remove_mpc_combine_for_stream(const struct dc *dc,
+ struct dc_state *new_ctx,
+ const struct dc_state *cur_ctx,
+ struct dc_stream_status *status)
+{
+ int i;
+
+ for (i = 0; i < status->plane_count; i++)
+ resource_update_pipes_for_plane_with_slice_count(
+ new_ctx, cur_ctx, dc->res_pool,
+ status->plane_states[i], 1);
+}
+
bool dc_state_add_plane(
const struct dc *dc,
struct dc_stream_state *stream,
@@ -447,8 +460,12 @@ bool dc_state_add_plane(
struct pipe_ctx *otg_master_pipe;
struct dc_stream_status *stream_status = NULL;
bool added = false;
+ int odm_slice_count;
+ int i;
stream_status = dc_state_get_stream_status(state, stream);
+ otg_master_pipe = resource_get_otg_master_for_stream(
+ &state->res_ctx, stream);
if (stream_status == NULL) {
dm_error("Existing stream not found; failed to attach surface!\n");
goto out;
@@ -456,22 +473,39 @@ bool dc_state_add_plane(
dm_error("Surface: can not attach plane_state %p! Maximum is: %d\n",
plane_state, MAX_SURFACE_NUM);
goto out;
+ } else if (!otg_master_pipe) {
+ goto out;
}
- if (stream_status->plane_count == 0 && dc->config.enable_windowed_mpo_odm)
- /* ODM combine could prevent us from supporting more planes
- * we will reset ODM slice count back to 1 when all planes have
- * been removed to maximize the amount of planes supported when
- * new planes are added.
- */
- resource_update_pipes_for_stream_with_slice_count(
- state, dc->current_state, dc->res_pool, stream, 1);
+ added = resource_append_dpp_pipes_for_plane_composition(state,
+ dc->current_state, pool, otg_master_pipe, plane_state);
- otg_master_pipe = resource_get_otg_master_for_stream(
- &state->res_ctx, stream);
- if (otg_master_pipe)
+ if (!added) {
+ /* try to remove MPC combine to free up pipes */
+ for (i = 0; i < state->stream_count; i++)
+ remove_mpc_combine_for_stream(dc, state,
+ dc->current_state,
+ &state->stream_status[i]);
added = resource_append_dpp_pipes_for_plane_composition(state,
- dc->current_state, pool, otg_master_pipe, plane_state);
+ dc->current_state, pool,
+ otg_master_pipe, plane_state);
+ }
+
+ if (!added) {
+ /* try to decrease ODM slice count gradually to free up pipes */
+ odm_slice_count = resource_get_odm_slice_count(otg_master_pipe);
+ for (i = odm_slice_count - 1; i > 0; i--) {
+ resource_update_pipes_for_stream_with_slice_count(state,
+ dc->current_state, dc->res_pool, stream,
+ i);
+ added = resource_append_dpp_pipes_for_plane_composition(
+ state,
+ dc->current_state, pool,
+ otg_master_pipe, plane_state);
+ if (added)
+ break;
+ }
+ }
if (added) {
stream_status->plane_states[stream_status->plane_count] =
@@ -531,15 +565,6 @@ bool dc_state_remove_plane(
stream_status->plane_states[stream_status->plane_count] = NULL;
- if (stream_status->plane_count == 0 && dc->config.enable_windowed_mpo_odm)
- /* ODM combine could prevent us from supporting more planes
- * we will reset ODM slice count back to 1 when all planes have
- * been removed to maximize the amount of planes supported when
- * new planes are added.
- */
- resource_update_pipes_for_stream_with_slice_count(
- state, dc->current_state, dc->res_pool, stream, 1);
-
return true;
}
--
2.43.0
next prev parent reply other threads:[~2024-07-28 15:44 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-07-28 15:40 [PATCH AUTOSEL 6.10 01/34] drm/xe/preempt_fence: enlarge the fence critical section Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 02/34] drm/amd/display: Handle HPD_IRQ for internal link Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 03/34] drm/amd/display: Add delay to improve LTTPR UHBR interop Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 04/34] drm/amdgpu: fix potential resource leak warning Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 05/34] drm/amdgpu/pm: Fix the param type of set_power_profile_mode Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 06/34] drm/amd/amdkfd: Fix a resource leak in svm_range_validate_and_map() Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 07/34] drm/xe/xe_guc_submit: Fix exec queue stop race condition Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 08/34] drm/amdgpu/pm: Fix the null pointer dereference for smu7 Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 09/34] drm/amdgpu: Fix the null pointer dereference to ras_manager Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 10/34] drm/amdgpu/pm: Fix the null pointer dereference in apply_state_adjust_rules Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 11/34] drm/admgpu: fix dereferencing null pointer context Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 12/34] drm/amdgpu: Add lock around VF RLCG interface Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 13/34] drm/amd/pm: Fix the null pointer dereference for vega10_hwmgr Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 14/34] drm/amd/display: Add null checks for 'stream' and 'plane' before dereferencing Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 15/34] media: amphion: Remove lock in s_ctrl callback Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 16/34] drm/amd/display: Add NULL check for 'afb' before dereferencing in amdgpu_dm_plane_handle_cursor_update Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 17/34] drm/amd/display: Wake DMCUB before sending a command for replay feature Sasha Levin
2024-07-28 15:40 ` Sasha Levin [this message]
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 19/34] drm/amd/display: Don't refer to dc_sink in is_dsc_need_re_compute Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 20/34] drm/amd/display: remove dpp pipes on failure to update pipe params Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 21/34] drm/amd/display: Add null checker before passing variables Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 22/34] media: i2c: ov5647: replacing of_node_put with __free(device_node) Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 23/34] media: uvcvideo: Ignore empty TS packets Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 24/34] media: uvcvideo: Fix the bandwdith quirk on USB 3.x Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 25/34] media: uvcvideo: Remove mappings form uvc_device_info Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 26/34] drm/panic: depends on !VT_CONSOLE Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 27/34] drm/amd/display: Fix NULL pointer dereference for DTN log in DCN401 Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 28/34] media: xc2028: avoid use-after-free in load_firmware_cb() Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 29/34] ext4: fix uninitialized variable in ext4_inlinedir_to_tree Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 30/34] jbd2: avoid memleak in jbd2_journal_write_metadata_buffer Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 31/34] drm/amd/display: Fix null pointer deref in dcn20_resource.c Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 32/34] s390/sclp: Prevent release of buffer in I/O Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 33/34] ext4: sanity check for NULL pointer after ext4_force_shutdown Sasha Levin
2024-07-28 15:40 ` [PATCH AUTOSEL 6.10 34/34] SUNRPC: Fix a race to wake a sync task Sasha Levin
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=20240728154230.2046786-18-sashal@kernel.org \
--to=sashal@kernel.org \
--cc=Rodrigo.Siqueira@amd.com \
--cc=Xinhui.Pan@amd.com \
--cc=airlied@gmail.com \
--cc=alex.hung@amd.com \
--cc=alexander.deucher@amd.com \
--cc=allen.pan@amd.com \
--cc=alvin.lee2@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=chaitanya.dhere@amd.com \
--cc=charlene.liu@amd.com \
--cc=chiahsuan.chung@amd.com \
--cc=christian.koenig@amd.com \
--cc=daniel.wheeler@amd.com \
--cc=daniel@ffwll.ch \
--cc=dillon.varone@amd.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=george.shen@amd.com \
--cc=hamza.mahfooz@amd.com \
--cc=harry.wentland@amd.com \
--cc=jun.lei@amd.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mario.limonciello@amd.com \
--cc=stable@vger.kernel.org \
--cc=sunpeng.li@amd.com \
--cc=wenjing.liu@amd.com \
--cc=zaeem.mohamed@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