* [PATCH 01/34] drm/amd/display: Rearchitecture HDMI HPD
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
@ 2018-02-12 17:15 ` Harry Wentland
2018-02-12 17:15 ` [PATCH 02/34] drm/amd/display: VGA black screen from s3 when attached to hook Harry Wentland
` (32 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:15 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: John Barberiz
From: John Barberiz <jbarberi@amd.com>
- Disabled HPD filter and used HPD software timer instead
- Allows DM to disable HPD filtering
Change-Id: I379d6285ffe85e1a98f560d5428839235928e43b
Signed-off-by: John Barberiz <jbarberi@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc_link.c | 32 +++++++++++++++++++++++++++
drivers/gpu/drm/amd/display/dc/dc_link.h | 4 ++++
2 files changed, 36 insertions(+)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
index 50aa9a4521f3..02e1c3b188fd 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
@@ -127,6 +127,8 @@ static bool program_hpd_filter(
int delay_on_connect_in_ms = 0;
int delay_on_disconnect_in_ms = 0;
+ if (link->is_hpd_filter_disabled)
+ return false;
/* Verify feature is supported */
switch (link->connector_signal) {
case SIGNAL_TYPE_DVI_SINGLE_LINK:
@@ -2346,3 +2348,33 @@ void core_link_set_avmute(struct pipe_ctx *pipe_ctx, bool enable)
core_dc->hwss.set_avmute(pipe_ctx, enable);
}
+void dc_link_disable_hpd_filter(struct dc_link *link)
+{
+ struct gpio *hpd;
+
+ if (!link->is_hpd_filter_disabled) {
+ link->is_hpd_filter_disabled = true;
+ /* Obtain HPD handle */
+ hpd = get_hpd_gpio(link->ctx->dc_bios, link->link_id, link->ctx->gpio_service);
+
+ if (!hpd)
+ return;
+
+ /* Setup HPD filtering */
+ if (dal_gpio_open(hpd, GPIO_MODE_INTERRUPT) == GPIO_RESULT_OK) {
+ struct gpio_hpd_config config;
+
+ config.delay_on_connect = 0;
+ config.delay_on_disconnect = 0;
+
+ dal_irq_setup_hpd_filter(hpd, &config);
+
+ dal_gpio_close(hpd);
+ } else {
+ ASSERT_CRITICAL(false);
+ }
+ /* Release HPD handle */
+ dal_gpio_destroy_irq(&hpd);
+ }
+}
+
diff --git a/drivers/gpu/drm/amd/display/dc/dc_link.h b/drivers/gpu/drm/amd/display/dc/dc_link.h
index f11a734da1db..ac0f617b43c9 100644
--- a/drivers/gpu/drm/amd/display/dc/dc_link.h
+++ b/drivers/gpu/drm/amd/display/dc/dc_link.h
@@ -64,6 +64,8 @@ struct dc_link {
enum signal_type connector_signal;
enum dc_irq_source irq_source_hpd;
enum dc_irq_source irq_source_hpd_rx;/* aka DP Short Pulse */
+ bool is_hpd_filter_disabled;
+
/* caps is the same as reported_link_cap. link_traing use
* reported_link_cap. Will clean up. TODO
*/
@@ -195,6 +197,8 @@ bool dc_link_dp_set_test_pattern(
const unsigned char *p_custom_pattern,
unsigned int cust_pattern_size);
+void dc_link_disable_hpd_filter(struct dc_link *link);
+
/*
* DPCD access interfaces
*/
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 02/34] drm/amd/display: VGA black screen from s3 when attached to hook
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
2018-02-12 17:15 ` [PATCH 01/34] drm/amd/display: Rearchitecture HDMI HPD Harry Wentland
@ 2018-02-12 17:15 ` Harry Wentland
2018-02-12 17:15 ` [PATCH 03/34] drm/amd/display: boot up/S4 fix mainlink off before BL Harry Wentland
` (31 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:15 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Hersen Wu
From: Hersen Wu <hersenxs.wu@amd.com>
[Description] For MST, DC already notify MST sink for MST mode, DC stll
check DP SINK DPCD register to see if MST enabled. DP RX firmware may
not handle this properly.
Change-Id: I54cd4d4fc4f9baf8a6f44ad0b50209e50cbbdc92
Signed-off-by: Hersen Wu <hersenxs.wu@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c
index 4ee4c03a6724..604fb0171ee3 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c
@@ -1465,7 +1465,7 @@ void decide_link_settings(struct dc_stream_state *stream,
/* MST doesn't perform link training for now
* TODO: add MST specific link training routine
*/
- if (is_mst_supported(link)) {
+ if (stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST) {
*link_setting = link->verified_link_cap;
return;
}
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 03/34] drm/amd/display: boot up/S4 fix mainlink off before BL.
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
2018-02-12 17:15 ` [PATCH 01/34] drm/amd/display: Rearchitecture HDMI HPD Harry Wentland
2018-02-12 17:15 ` [PATCH 02/34] drm/amd/display: VGA black screen from s3 when attached to hook Harry Wentland
@ 2018-02-12 17:15 ` Harry Wentland
2018-02-12 17:15 ` [PATCH 04/34] drm/amd/display: dal 3.1.33 Harry Wentland
` (30 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:15 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Charlene Liu
From: Charlene Liu <charlene.liu@amd.com>
Change-Id: I0c7e3422f9ba134ffccd50778789caa6d4b6ae30
Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
index 54c933b94093..53243d5bad9a 100644
--- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
@@ -1481,7 +1481,7 @@ void dce110_enable_accelerated_mode(struct dc *dc, struct dc_state *context)
struct dc_link *edp_link = get_link_for_edp(dc);
if (dcb->funcs->get_vga_enabled_displays(dc->ctx->dc_bios) != 0) {
- if (edp_link_to_turnoff) {
+ if (edp_link) {
/*we need turn off backlight before DP_blank and encoder powered down, todo add optimization*/
dc->hwss.edp_backlight_control(edp_link, false);
}
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 04/34] drm/amd/display: dal 3.1.33
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (2 preceding siblings ...)
2018-02-12 17:15 ` [PATCH 03/34] drm/amd/display: boot up/S4 fix mainlink off before BL Harry Wentland
@ 2018-02-12 17:15 ` Harry Wentland
2018-02-12 17:15 ` [PATCH 05/34] drm/amd/display: add force_trigger even to static screen control Harry Wentland
` (29 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:15 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Tony Cheng
From: Tony Cheng <tony.cheng@amd.com>
Change-Id: I559abccee2a41f70731200ec81a4d4b50c6cc739
Signed-off-by: Tony Cheng <tony.cheng@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/dc.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index 03af1a04c6aa..d3ca22fcc166 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -38,7 +38,7 @@
#include "inc/compressor.h"
#include "dml/display_mode_lib.h"
-#define DC_VER "3.1.32"
+#define DC_VER "3.1.33"
#define MAX_SURFACES 3
#define MAX_STREAMS 6
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 05/34] drm/amd/display: add force_trigger even to static screen control
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (3 preceding siblings ...)
2018-02-12 17:15 ` [PATCH 04/34] drm/amd/display: dal 3.1.33 Harry Wentland
@ 2018-02-12 17:15 ` Harry Wentland
2018-02-12 17:15 ` [PATCH 06/34] drm/amd/display: Change blackout time to 0 on CZ/ST Harry Wentland
` (28 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:15 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Charlene Liu
From: Charlene Liu <charlene.liu@amd.com>
Change-Id: I27e1779010662ac63b7b4e9c83aaa4f3ebf1db37
Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/dc.h | 1 +
drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c | 2 ++
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c | 2 ++
3 files changed, 5 insertions(+)
diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index d3ca22fcc166..0c92ddb3d3f5 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -95,6 +95,7 @@ struct dc_surface_dcc_cap {
};
struct dc_static_screen_events {
+ bool force_trigger;
bool cursor_update;
bool surface_update;
bool overlay_update;
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
index 53243d5bad9a..5c387e8d58a2 100644
--- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
@@ -1625,6 +1625,8 @@ static void set_static_screen_control(struct pipe_ctx **pipe_ctx,
value |= 0x80;
if (events->cursor_update)
value |= 0x2;
+ if (events->force_trigger)
+ value |= 0x1;
#if defined(CONFIG_DRM_AMD_DC_FBC)
value |= 0x84;
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
index 5d3dedfbc69e..29dc37fbdb26 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
@@ -2144,6 +2144,8 @@ static void set_static_screen_control(struct pipe_ctx **pipe_ctx,
value |= 0x80;
if (events->cursor_update)
value |= 0x2;
+ if (events->force_trigger)
+ value |= 0x1;
for (i = 0; i < num_pipes; i++)
pipe_ctx[i]->stream_res.tg->funcs->
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 06/34] drm/amd/display: Change blackout time to 0 on CZ/ST
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (4 preceding siblings ...)
2018-02-12 17:15 ` [PATCH 05/34] drm/amd/display: add force_trigger even to static screen control Harry Wentland
@ 2018-02-12 17:15 ` Harry Wentland
2018-02-12 17:15 ` [PATCH 07/34] drm/amd/display: add monitor patch for delay after DP receive power up Harry Wentland
` (27 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:15 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Harry Wentland
These should only be non-0 if big hammer w/a is implemented. Currently
DC doesn't implement it, so leave them 0.
Change-Id: I8bfb01d3bbf02e08628e64e740ef3df5de4ae04a
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c b/drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c
index bb03a9c64d5a..6d38b8f43198 100644
--- a/drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c
+++ b/drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c
@@ -2033,8 +2033,8 @@ void bw_calcs_init(struct bw_calcs_dceip *bw_dceip,
vbios.cursor_width = 32;
vbios.average_compression_rate = 4;
vbios.number_of_request_slots_gmc_reserves_for_dmif_per_channel = 256;
- vbios.blackout_duration = bw_int_to_fixed(18); /* us */
- vbios.maximum_blackout_recovery_time = bw_int_to_fixed(20);
+ vbios.blackout_duration = bw_int_to_fixed(0); /* us */
+ vbios.maximum_blackout_recovery_time = bw_int_to_fixed(0);
dceip.large_cursor = false;
dceip.dmif_request_buffer_size = bw_int_to_fixed(768);
@@ -2366,8 +2366,8 @@ void bw_calcs_init(struct bw_calcs_dceip *bw_dceip,
vbios.cursor_width = 32;
vbios.average_compression_rate = 4;
vbios.number_of_request_slots_gmc_reserves_for_dmif_per_channel = 256;
- vbios.blackout_duration = bw_int_to_fixed(18); /* us */
- vbios.maximum_blackout_recovery_time = bw_int_to_fixed(20);
+ vbios.blackout_duration = bw_int_to_fixed(0); /* us */
+ vbios.maximum_blackout_recovery_time = bw_int_to_fixed(0);
dceip.large_cursor = false;
dceip.dmif_request_buffer_size = bw_int_to_fixed(768);
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 07/34] drm/amd/display: add monitor patch for delay after DP receive power up
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (5 preceding siblings ...)
2018-02-12 17:15 ` [PATCH 06/34] drm/amd/display: Change blackout time to 0 on CZ/ST Harry Wentland
@ 2018-02-12 17:15 ` Harry Wentland
2018-02-12 17:15 ` [PATCH 08/34] drm/amd/display: disable seamless vp adjustment for mirrored surface Harry Wentland
` (26 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:15 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Martin Tsai
From: Martin Tsai <martin.tsai@amd.com>
Change-Id: Ic8ab8775d7a0df9c12ed94ec02da8b257a43a77f
Signed-off-by: Martin Tsai <martin.tsai@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc_link.c | 6 ++++++
drivers/gpu/drm/amd/display/dc/dc_types.h | 1 +
2 files changed, 7 insertions(+)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
index 02e1c3b188fd..9ede0f884f15 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
@@ -1260,6 +1260,12 @@ static enum dc_status enable_link_dp(
pipe_ctx->clock_source->id,
&link_settings);
+ if (stream->sink->edid_caps.panel_patch.dppowerup_delay > 0) {
+ int delay_dp_power_up_in_ms = stream->sink->edid_caps.panel_patch.dppowerup_delay;
+
+ msleep(delay_dp_power_up_in_ms);
+ }
+
panel_mode = dp_get_panel_mode(link);
dpcd_configure_panel_mode(link, panel_mode);
diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h
index fbbe92f7a33a..aa5b90e6beb7 100644
--- a/drivers/gpu/drm/amd/display/dc/dc_types.h
+++ b/drivers/gpu/drm/amd/display/dc/dc_types.h
@@ -195,6 +195,7 @@ union display_content_support {
struct dc_panel_patch {
unsigned int disconnect_delay;
+ unsigned int dppowerup_delay;
};
struct dc_edid_caps {
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 08/34] drm/amd/display: disable seamless vp adjustment for mirrored surface
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (6 preceding siblings ...)
2018-02-12 17:15 ` [PATCH 07/34] drm/amd/display: add monitor patch for delay after DP receive power up Harry Wentland
@ 2018-02-12 17:15 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 09/34] drm/amd/display: dal 3.1.34 Harry Wentland
` (25 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:15 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Dmytro Laktyushkin
From: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Change-Id: I36ccf887310c2d76181ef3cf638ee6af897e5abc
Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc_resource.c | 8 ++++----
1 file changed, 4 insertions(+), 4 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 a215c5b74cbd..a6f8c1f93693 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
@@ -696,7 +696,7 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r
/* Adjust for viewport end clip-off */
- if ((data->viewport.x + data->viewport.width) < (src.x + src.width)) {
+ if ((data->viewport.x + data->viewport.width) < (src.x + src.width) && !flip_horz_scan_dir) {
int vp_clip = src.x + src.width - data->viewport.width - data->viewport.x;
int int_part = dal_fixed31_32_floor(
dal_fixed31_32_sub(data->inits.h, data->ratios.horz));
@@ -704,7 +704,7 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r
int_part = int_part > 0 ? int_part : 0;
data->viewport.width += int_part < vp_clip ? int_part : vp_clip;
}
- if ((data->viewport.y + data->viewport.height) < (src.y + src.height)) {
+ if ((data->viewport.y + data->viewport.height) < (src.y + src.height) && !flip_vert_scan_dir) {
int vp_clip = src.y + src.height - data->viewport.height - data->viewport.y;
int int_part = dal_fixed31_32_floor(
dal_fixed31_32_sub(data->inits.v, data->ratios.vert));
@@ -712,7 +712,7 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r
int_part = int_part > 0 ? int_part : 0;
data->viewport.height += int_part < vp_clip ? int_part : vp_clip;
}
- if ((data->viewport_c.x + data->viewport_c.width) < (src.x + src.width) / vpc_div) {
+ if ((data->viewport_c.x + data->viewport_c.width) < (src.x + src.width) / vpc_div && !flip_horz_scan_dir) {
int vp_clip = (src.x + src.width) / vpc_div -
data->viewport_c.width - data->viewport_c.x;
int int_part = dal_fixed31_32_floor(
@@ -721,7 +721,7 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r
int_part = int_part > 0 ? int_part : 0;
data->viewport_c.width += int_part < vp_clip ? int_part : vp_clip;
}
- if ((data->viewport_c.y + data->viewport_c.height) < (src.y + src.height) / vpc_div) {
+ if ((data->viewport_c.y + data->viewport_c.height) < (src.y + src.height) / vpc_div && !flip_vert_scan_dir) {
int vp_clip = (src.y + src.height) / vpc_div -
data->viewport_c.height - data->viewport_c.y;
int int_part = dal_fixed31_32_floor(
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 09/34] drm/amd/display: dal 3.1.34
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (7 preceding siblings ...)
2018-02-12 17:15 ` [PATCH 08/34] drm/amd/display: disable seamless vp adjustment for mirrored surface Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 10/34] drm/amd/display: Keep eDP stream enabled during boot Harry Wentland
` (24 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Tony Cheng
From: Tony Cheng <tony.cheng@amd.com>
Change-Id: Ic417b7ca8089a4f37c4c4c9c426a4d4fdb4d3151
Signed-off-by: Tony Cheng <tony.cheng@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/dc.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index 0c92ddb3d3f5..0be839b152ec 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -38,7 +38,7 @@
#include "inc/compressor.h"
#include "dml/display_mode_lib.h"
-#define DC_VER "3.1.33"
+#define DC_VER "3.1.34"
#define MAX_SURFACES 3
#define MAX_STREAMS 6
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 10/34] drm/amd/display: Keep eDP stream enabled during boot.
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (8 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 09/34] drm/amd/display: dal 3.1.34 Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 11/34] drm/amd/display: Remove unused DCE80 compressor Harry Wentland
` (23 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Yongqiang Sun
From: Yongqiang Sun <yongqiang.sun@amd.com>
This path fixed specific eDP panel cold boot black screen
due to unnecessary enable link.
Change:
In case of boot up with eDP, if OS is going to set mode
on eDP, keep eDP light up, do not disable and reset corresponding
HW.
This change may affect dce asics and S3/S4 Resume with multi-monitor.
Change-Id: I658328103edd41aad01757e80defff5a2c116ca4
Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc_link.c | 11 +++++-
drivers/gpu/drm/amd/display/dc/dc.h | 2 +
.../amd/display/dc/dce110/dce110_hw_sequencer.c | 45 +++++++++++-----------
3 files changed, 34 insertions(+), 24 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
index 9ede0f884f15..44b12f0405ee 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
@@ -2296,7 +2296,16 @@ void core_link_enable_stream(
{
struct dc *core_dc = pipe_ctx->stream->ctx->dc;
- enum dc_status status = enable_link(state, pipe_ctx);
+ enum dc_status status;
+
+ /* eDP lit up by bios already, no need to enable again. */
+ if (pipe_ctx->stream->signal == SIGNAL_TYPE_EDP &&
+ core_dc->apply_edp_fast_boot_optimization) {
+ core_dc->apply_edp_fast_boot_optimization = false;
+ return;
+ }
+
+ status = enable_link(state, pipe_ctx);
if (status != DC_OK) {
dm_logger_write(pipe_ctx->stream->ctx->logger,
diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index 0be839b152ec..a512b05303e2 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -257,6 +257,8 @@ struct dc {
bool optimized_required;
+ bool apply_edp_fast_boot_optimization;
+
/* FBC compressor */
#if defined(CONFIG_DRM_AMD_DC_FBC)
struct compressor *fbc_compressor;
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
index 5c387e8d58a2..9e97cd74b0ec 100644
--- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
@@ -57,6 +57,8 @@
#include "dce/dce_11_0_sh_mask.h"
#include "custom_float.h"
+#include "atomfirmware.h"
+
/*
* All values are in milliseconds;
* For eDP, after power-up/power/down,
@@ -1425,23 +1427,6 @@ static void disable_vga_and_power_gate_all_controllers(
}
}
-static struct dc_link *get_link_for_edp(
- struct dc *dc)
-{
- int i;
- struct dc_link *link = NULL;
-
- /* check if there is an eDP panel not in use */
- for (i = 0; i < dc->link_count; i++) {
- if (dc->links[i]->local_sink &&
- dc->links[i]->local_sink->sink_signal == SIGNAL_TYPE_EDP) {
- link = dc->links[i];
- break;
- }
- }
-
- return link;
-}
static struct dc_link *get_link_for_edp_not_in_use(
struct dc *dc,
struct dc_state *context)
@@ -1477,13 +1462,27 @@ static struct dc_link *get_link_for_edp_not_in_use(
void dce110_enable_accelerated_mode(struct dc *dc, struct dc_state *context)
{
struct dc_bios *dcb = dc->ctx->dc_bios;
- struct dc_link *edp_link_to_turnoff = get_link_for_edp_not_in_use(dc, context);
- struct dc_link *edp_link = get_link_for_edp(dc);
- if (dcb->funcs->get_vga_enabled_displays(dc->ctx->dc_bios) != 0) {
- if (edp_link) {
- /*we need turn off backlight before DP_blank and encoder powered down, todo add optimization*/
- dc->hwss.edp_backlight_control(edp_link, false);
+ /* vbios already light up eDP, so we can leverage vbios and skip eDP
+ * programming
+ */
+ bool can_eDP_fast_boot_optimize =
+ (dcb->funcs->get_vga_enabled_displays(dc->ctx->dc_bios) == ATOM_DISPLAY_LCD1_ACTIVE);
+
+ /* if OS doesn't light up eDP and eDP link is available, we want to disable */
+ struct dc_link *edp_link_to_turnoff = NULL;
+
+ if (can_eDP_fast_boot_optimize) {
+ edp_link_to_turnoff = get_link_for_edp_not_in_use(dc, context);
+
+ if (!edp_link_to_turnoff)
+ dc->apply_edp_fast_boot_optimization = true;
+ }
+
+ if (!dc->apply_edp_fast_boot_optimization) {
+ if (edp_link_to_turnoff) {
+ /*turn off backlight before DP_blank and encoder powered down*/
+ dc->hwss.edp_backlight_control(edp_link_to_turnoff, false);
}
/*resume from S3, no vbios posting, no need to power down again*/
power_down_all_hw_blocks(dc);
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 11/34] drm/amd/display: Remove unused DCE80 compressor
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (9 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 10/34] drm/amd/display: Keep eDP stream enabled during boot Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 12/34] drm/amd/display: Re-use DCE100 display_power_gating for DCE80 Harry Wentland
` (22 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Harry Wentland
Change-Id: Ia65abc9c0516b8b64adb36061087c5b35d3eb10a
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/dce80/Makefile | 2 +-
.../drm/amd/display/dc/dce80/dce80_compressor.c | 834 ---------------------
.../drm/amd/display/dc/dce80/dce80_compressor.h | 78 --
3 files changed, 1 insertion(+), 913 deletions(-)
delete mode 100644 drivers/gpu/drm/amd/display/dc/dce80/dce80_compressor.c
delete mode 100644 drivers/gpu/drm/amd/display/dc/dce80/dce80_compressor.h
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/Makefile b/drivers/gpu/drm/amd/display/dc/dce80/Makefile
index bc388aa4b2f5..666fcb2bdbba 100644
--- a/drivers/gpu/drm/amd/display/dc/dce80/Makefile
+++ b/drivers/gpu/drm/amd/display/dc/dce80/Makefile
@@ -23,7 +23,7 @@
# Makefile for the 'controller' sub-component of DAL.
# It provides the control and status of HW CRTC block.
-DCE80 = dce80_timing_generator.o dce80_compressor.o dce80_hw_sequencer.o \
+DCE80 = dce80_timing_generator.o dce80_hw_sequencer.o \
dce80_resource.o
AMD_DAL_DCE80 = $(addprefix $(AMDDALPATH)/dc/dce80/,$(DCE80))
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_compressor.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_compressor.c
deleted file mode 100644
index 951f2caba9b3..000000000000
--- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_compressor.c
+++ /dev/null
@@ -1,834 +0,0 @@
-/*
- * Copyright 2012-15 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors: AMD
- *
- */
-
-#include "dm_services.h"
-
-#include "dce/dce_8_0_d.h"
-#include "dce/dce_8_0_sh_mask.h"
-#include "gmc/gmc_7_1_sh_mask.h"
-#include "gmc/gmc_7_1_d.h"
-
-#include "include/logger_interface.h"
-#include "dce80_compressor.h"
-
-#define DCP_REG(reg)\
- (reg + cp80->offsets.dcp_offset)
-#define DMIF_REG(reg)\
- (reg + cp80->offsets.dmif_offset)
-
-static const struct dce80_compressor_reg_offsets reg_offsets[] = {
-{
- .dcp_offset = (mmDCP0_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
- .dmif_offset = (mmDMIF_PG0_DPG_PIPE_DPM_CONTROL
- - mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
-},
-{
- .dcp_offset = (mmDCP1_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
- .dmif_offset = (mmDMIF_PG1_DPG_PIPE_DPM_CONTROL
- - mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
-},
-{
- .dcp_offset = (mmDCP2_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
- .dmif_offset = (mmDMIF_PG2_DPG_PIPE_DPM_CONTROL
- - mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
-},
-{
- .dcp_offset = (mmDCP3_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
- .dmif_offset = (mmDMIF_PG3_DPG_PIPE_DPM_CONTROL
- - mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
-},
-{
- .dcp_offset = (mmDCP4_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
- .dmif_offset = (mmDMIF_PG4_DPG_PIPE_DPM_CONTROL
- - mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
-},
-{
- .dcp_offset = (mmDCP5_GRPH_CONTROL - mmDCP0_GRPH_CONTROL),
- .dmif_offset = (mmDMIF_PG5_DPG_PIPE_DPM_CONTROL
- - mmDMIF_PG0_DPG_PIPE_DPM_CONTROL),
-}
-};
-
-static const uint32_t dce8_one_lpt_channel_max_resolution = 2048 * 1200;
-
-enum fbc_idle_force {
- /* Bit 0 - Display registers updated */
- FBC_IDLE_FORCE_DISPLAY_REGISTER_UPDATE = 0x00000001,
-
- /* Bit 2 - FBC_GRPH_COMP_EN register updated */
- FBC_IDLE_FORCE_GRPH_COMP_EN = 0x00000002,
- /* Bit 3 - FBC_SRC_SEL register updated */
- FBC_IDLE_FORCE_SRC_SEL_CHANGE = 0x00000004,
- /* Bit 4 - FBC_MIN_COMPRESSION register updated */
- FBC_IDLE_FORCE_MIN_COMPRESSION_CHANGE = 0x00000008,
- /* Bit 5 - FBC_ALPHA_COMP_EN register updated */
- FBC_IDLE_FORCE_ALPHA_COMP_EN = 0x00000010,
- /* Bit 6 - FBC_ZERO_ALPHA_CHUNK_SKIP_EN register updated */
- FBC_IDLE_FORCE_ZERO_ALPHA_CHUNK_SKIP_EN = 0x00000020,
- /* Bit 7 - FBC_FORCE_COPY_TO_COMP_BUF register updated */
- FBC_IDLE_FORCE_FORCE_COPY_TO_COMP_BUF = 0x00000040,
-
- /* Bit 24 - Memory write to region 0 defined by MC registers. */
- FBC_IDLE_FORCE_MEMORY_WRITE_TO_REGION0 = 0x01000000,
- /* Bit 25 - Memory write to region 1 defined by MC registers */
- FBC_IDLE_FORCE_MEMORY_WRITE_TO_REGION1 = 0x02000000,
- /* Bit 26 - Memory write to region 2 defined by MC registers */
- FBC_IDLE_FORCE_MEMORY_WRITE_TO_REGION2 = 0x04000000,
- /* Bit 27 - Memory write to region 3 defined by MC registers. */
- FBC_IDLE_FORCE_MEMORY_WRITE_TO_REGION3 = 0x08000000,
-
- /* Bit 28 - Memory write from any client other than MCIF */
- FBC_IDLE_FORCE_MEMORY_WRITE_OTHER_THAN_MCIF = 0x10000000,
- /* Bit 29 - CG statics screen signal is inactive */
- FBC_IDLE_FORCE_CG_STATIC_SCREEN_IS_INACTIVE = 0x20000000,
-};
-
-static uint32_t lpt_size_alignment(struct dce80_compressor *cp80)
-{
- /*LPT_ALIGNMENT (in bytes) = ROW_SIZE * #BANKS * # DRAM CHANNELS. */
- return cp80->base.raw_size * cp80->base.banks_num *
- cp80->base.dram_channels_num;
-}
-
-static uint32_t lpt_memory_control_config(struct dce80_compressor *cp80,
- uint32_t lpt_control)
-{
- /*LPT MC Config */
- if (cp80->base.options.bits.LPT_MC_CONFIG == 1) {
- /* POSSIBLE VALUES for LPT NUM_PIPES (DRAM CHANNELS):
- * 00 - 1 CHANNEL
- * 01 - 2 CHANNELS
- * 02 - 4 OR 6 CHANNELS
- * (Only for discrete GPU, N/A for CZ)
- * 03 - 8 OR 12 CHANNELS
- * (Only for discrete GPU, N/A for CZ) */
- switch (cp80->base.dram_channels_num) {
- case 2:
- set_reg_field_value(
- lpt_control,
- 1,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_NUM_PIPES);
- break;
- case 1:
- set_reg_field_value(
- lpt_control,
- 0,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_NUM_PIPES);
- break;
- default:
- dm_logger_write(
- cp80->base.ctx->logger, LOG_WARNING,
- "%s: Invalid LPT NUM_PIPES!!!",
- __func__);
- break;
- }
-
- /* The mapping for LPT NUM_BANKS is in
- * GRPH_CONTROL.GRPH_NUM_BANKS register field
- * Specifies the number of memory banks for tiling
- * purposes. Only applies to 2D and 3D tiling modes.
- * POSSIBLE VALUES:
- * 00 - DCP_GRPH_NUM_BANKS_2BANK: ADDR_SURF_2_BANK
- * 01 - DCP_GRPH_NUM_BANKS_4BANK: ADDR_SURF_4_BANK
- * 02 - DCP_GRPH_NUM_BANKS_8BANK: ADDR_SURF_8_BANK
- * 03 - DCP_GRPH_NUM_BANKS_16BANK: ADDR_SURF_16_BANK */
- switch (cp80->base.banks_num) {
- case 16:
- set_reg_field_value(
- lpt_control,
- 3,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_NUM_BANKS);
- break;
- case 8:
- set_reg_field_value(
- lpt_control,
- 2,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_NUM_BANKS);
- break;
- case 4:
- set_reg_field_value(
- lpt_control,
- 1,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_NUM_BANKS);
- break;
- case 2:
- set_reg_field_value(
- lpt_control,
- 0,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_NUM_BANKS);
- break;
- default:
- dm_logger_write(
- cp80->base.ctx->logger, LOG_WARNING,
- "%s: Invalid LPT NUM_BANKS!!!",
- __func__);
- break;
- }
-
- /* The mapping is in DMIF_ADDR_CALC.
- * ADDR_CONFIG_PIPE_INTERLEAVE_SIZE register field for
- * Carrizo specifies the memory interleave per pipe.
- * It effectively specifies the location of pipe bits in
- * the memory address.
- * POSSIBLE VALUES:
- * 00 - ADDR_CONFIG_PIPE_INTERLEAVE_256B: 256 byte
- * interleave
- * 01 - ADDR_CONFIG_PIPE_INTERLEAVE_512B: 512 byte
- * interleave
- */
- switch (cp80->base.channel_interleave_size) {
- case 256: /*256B */
- set_reg_field_value(
- lpt_control,
- 0,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_PIPE_INTERLEAVE_SIZE);
- break;
- case 512: /*512B */
- set_reg_field_value(
- lpt_control,
- 1,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_PIPE_INTERLEAVE_SIZE);
- break;
- default:
- dm_logger_write(
- cp80->base.ctx->logger, LOG_WARNING,
- "%s: Invalid LPT INTERLEAVE_SIZE!!!",
- __func__);
- break;
- }
-
- /* The mapping for LOW_POWER_TILING_ROW_SIZE is in
- * DMIF_ADDR_CALC.ADDR_CONFIG_ROW_SIZE register field
- * for Carrizo. Specifies the size of dram row in bytes.
- * This should match up with NOOFCOLS field in
- * MC_ARB_RAMCFG (ROW_SIZE = 4 * 2 ^^ columns).
- * This register DMIF_ADDR_CALC is not used by the
- * hardware as it is only used for addrlib assertions.
- * POSSIBLE VALUES:
- * 00 - ADDR_CONFIG_1KB_ROW: Treat 1KB as DRAM row
- * boundary
- * 01 - ADDR_CONFIG_2KB_ROW: Treat 2KB as DRAM row
- * boundary
- * 02 - ADDR_CONFIG_4KB_ROW: Treat 4KB as DRAM row
- * boundary */
- switch (cp80->base.raw_size) {
- case 4096: /*4 KB */
- set_reg_field_value(
- lpt_control,
- 2,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_ROW_SIZE);
- break;
- case 2048:
- set_reg_field_value(
- lpt_control,
- 1,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_ROW_SIZE);
- break;
- case 1024:
- set_reg_field_value(
- lpt_control,
- 0,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_ROW_SIZE);
- break;
- default:
- dm_logger_write(
- cp80->base.ctx->logger, LOG_WARNING,
- "%s: Invalid LPT ROW_SIZE!!!",
- __func__);
- break;
- }
- } else {
- dm_logger_write(
- cp80->base.ctx->logger, LOG_WARNING,
- "%s: LPT MC Configuration is not provided",
- __func__);
- }
-
- return lpt_control;
-}
-
-static bool is_source_bigger_than_epanel_size(
- struct dce80_compressor *cp80,
- uint32_t source_view_width,
- uint32_t source_view_height)
-{
- if (cp80->base.embedded_panel_h_size != 0 &&
- cp80->base.embedded_panel_v_size != 0 &&
- ((source_view_width * source_view_height) >
- (cp80->base.embedded_panel_h_size *
- cp80->base.embedded_panel_v_size)))
- return true;
-
- return false;
-}
-
-static uint32_t align_to_chunks_number_per_line(
- struct dce80_compressor *cp80,
- uint32_t pixels)
-{
- return 256 * ((pixels + 255) / 256);
-}
-
-static void wait_for_fbc_state_changed(
- struct dce80_compressor *cp80,
- bool enabled)
-{
- uint8_t counter = 0;
- uint32_t addr = mmFBC_STATUS;
- uint32_t value;
-
- while (counter < 10) {
- value = dm_read_reg(cp80->base.ctx, addr);
- if (get_reg_field_value(
- value,
- FBC_STATUS,
- FBC_ENABLE_STATUS) == enabled)
- break;
- udelay(10);
- counter++;
- }
-
- if (counter == 10) {
- dm_logger_write(
- cp80->base.ctx->logger, LOG_WARNING,
- "%s: wait counter exceeded, changes to HW not applied",
- __func__);
- }
-}
-
-void dce80_compressor_power_up_fbc(struct compressor *compressor)
-{
- uint32_t value;
- uint32_t addr;
-
- addr = mmFBC_CNTL;
- value = dm_read_reg(compressor->ctx, addr);
- set_reg_field_value(value, 0, FBC_CNTL, FBC_GRPH_COMP_EN);
- set_reg_field_value(value, 1, FBC_CNTL, FBC_EN);
- set_reg_field_value(value, 2, FBC_CNTL, FBC_COHERENCY_MODE);
- dm_write_reg(compressor->ctx, addr, value);
-
- addr = mmFBC_COMP_MODE;
- value = dm_read_reg(compressor->ctx, addr);
- set_reg_field_value(value, 1, FBC_COMP_MODE, FBC_RLE_EN);
- set_reg_field_value(value, 1, FBC_COMP_MODE, FBC_DPCM4_RGB_EN);
- set_reg_field_value(value, 1, FBC_COMP_MODE, FBC_IND_EN);
- dm_write_reg(compressor->ctx, addr, value);
-
- addr = mmFBC_COMP_CNTL;
- value = dm_read_reg(compressor->ctx, addr);
- set_reg_field_value(value, 1, FBC_COMP_CNTL, FBC_DEPTH_RGB08_EN);
- dm_write_reg(compressor->ctx, addr, value);
- /*FBC_MIN_COMPRESSION 0 ==> 2:1 */
- /* 1 ==> 4:1 */
- /* 2 ==> 8:1 */
- /* 0xF ==> 1:1 */
- set_reg_field_value(value, 0xF, FBC_COMP_CNTL, FBC_MIN_COMPRESSION);
- dm_write_reg(compressor->ctx, addr, value);
- compressor->min_compress_ratio = FBC_COMPRESS_RATIO_1TO1;
-
- value = 0;
- dm_write_reg(compressor->ctx, mmFBC_IND_LUT0, value);
-
- value = 0xFFFFFF;
- dm_write_reg(compressor->ctx, mmFBC_IND_LUT1, value);
-}
-
-void dce80_compressor_enable_fbc(
- struct compressor *compressor,
- uint32_t paths_num,
- struct compr_addr_and_pitch_params *params)
-{
- struct dce80_compressor *cp80 = TO_DCE80_COMPRESSOR(compressor);
-
- if (compressor->options.bits.FBC_SUPPORT &&
- (compressor->options.bits.DUMMY_BACKEND == 0) &&
- (!dce80_compressor_is_fbc_enabled_in_hw(compressor, NULL)) &&
- (!is_source_bigger_than_epanel_size(
- cp80,
- params->source_view_width,
- params->source_view_height))) {
-
- uint32_t addr;
- uint32_t value;
-
- /* Before enabling FBC first need to enable LPT if applicable
- * LPT state should always be changed (enable/disable) while FBC
- * is disabled */
- if (compressor->options.bits.LPT_SUPPORT && (paths_num < 2) &&
- (params->source_view_width *
- params->source_view_height <=
- dce8_one_lpt_channel_max_resolution)) {
- dce80_compressor_enable_lpt(compressor);
- }
-
- addr = mmFBC_CNTL;
- value = dm_read_reg(compressor->ctx, addr);
- set_reg_field_value(value, 1, FBC_CNTL, FBC_GRPH_COMP_EN);
- set_reg_field_value(
- value,
- params->inst,
- FBC_CNTL, FBC_SRC_SEL);
- dm_write_reg(compressor->ctx, addr, value);
-
- /* Keep track of enum controller_id FBC is attached to */
- compressor->is_enabled = true;
- compressor->attached_inst = params->inst;
- cp80->offsets = reg_offsets[params->inst];
-
- /*Toggle it as there is bug in HW */
- set_reg_field_value(value, 0, FBC_CNTL, FBC_GRPH_COMP_EN);
- dm_write_reg(compressor->ctx, addr, value);
- set_reg_field_value(value, 1, FBC_CNTL, FBC_GRPH_COMP_EN);
- dm_write_reg(compressor->ctx, addr, value);
-
- wait_for_fbc_state_changed(cp80, true);
- }
-}
-
-void dce80_compressor_disable_fbc(struct compressor *compressor)
-{
- struct dce80_compressor *cp80 = TO_DCE80_COMPRESSOR(compressor);
-
- if (compressor->options.bits.FBC_SUPPORT &&
- dce80_compressor_is_fbc_enabled_in_hw(compressor, NULL)) {
- uint32_t reg_data;
- /* Turn off compression */
- reg_data = dm_read_reg(compressor->ctx, mmFBC_CNTL);
- set_reg_field_value(reg_data, 0, FBC_CNTL, FBC_GRPH_COMP_EN);
- dm_write_reg(compressor->ctx, mmFBC_CNTL, reg_data);
-
- /* Reset enum controller_id to undefined */
- compressor->attached_inst = 0;
- compressor->is_enabled = false;
-
- /* Whenever disabling FBC make sure LPT is disabled if LPT
- * supported */
- if (compressor->options.bits.LPT_SUPPORT)
- dce80_compressor_disable_lpt(compressor);
-
- wait_for_fbc_state_changed(cp80, false);
- }
-}
-
-bool dce80_compressor_is_fbc_enabled_in_hw(
- struct compressor *compressor,
- uint32_t *inst)
-{
- /* Check the hardware register */
- uint32_t value;
-
- value = dm_read_reg(compressor->ctx, mmFBC_STATUS);
- if (get_reg_field_value(value, FBC_STATUS, FBC_ENABLE_STATUS)) {
- if (inst != NULL)
- *inst = compressor->attached_inst;
- return true;
- }
-
- value = dm_read_reg(compressor->ctx, mmFBC_CNTL);
- if (get_reg_field_value(value, FBC_CNTL, FBC_GRPH_COMP_EN)) {
- if (inst != NULL)
- *inst = compressor->attached_inst;
- return true;
- }
-
- return false;
-}
-
-bool dce80_compressor_is_lpt_enabled_in_hw(struct compressor *compressor)
-{
- /* Check the hardware register */
- uint32_t value = dm_read_reg(compressor->ctx,
- mmLOW_POWER_TILING_CONTROL);
-
- return get_reg_field_value(
- value,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_ENABLE);
-}
-
-void dce80_compressor_program_compressed_surface_address_and_pitch(
- struct compressor *compressor,
- struct compr_addr_and_pitch_params *params)
-{
- struct dce80_compressor *cp80 = TO_DCE80_COMPRESSOR(compressor);
- uint32_t value = 0;
- uint32_t fbc_pitch = 0;
- uint32_t compressed_surf_address_low_part =
- compressor->compr_surface_address.addr.low_part;
-
- /* Clear content first. */
- dm_write_reg(
- compressor->ctx,
- DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS_HIGH),
- 0);
- dm_write_reg(compressor->ctx,
- DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS), 0);
-
- if (compressor->options.bits.LPT_SUPPORT) {
- uint32_t lpt_alignment = lpt_size_alignment(cp80);
-
- if (lpt_alignment != 0) {
- compressed_surf_address_low_part =
- ((compressed_surf_address_low_part
- + (lpt_alignment - 1)) / lpt_alignment)
- * lpt_alignment;
- }
- }
-
- /* Write address, HIGH has to be first. */
- dm_write_reg(compressor->ctx,
- DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS_HIGH),
- compressor->compr_surface_address.addr.high_part);
- dm_write_reg(compressor->ctx,
- DCP_REG(mmGRPH_COMPRESS_SURFACE_ADDRESS),
- compressed_surf_address_low_part);
-
- fbc_pitch = align_to_chunks_number_per_line(
- cp80,
- params->source_view_width);
-
- if (compressor->min_compress_ratio == FBC_COMPRESS_RATIO_1TO1)
- fbc_pitch = fbc_pitch / 8;
- else
- dm_logger_write(
- compressor->ctx->logger, LOG_WARNING,
- "%s: Unexpected DCE8 compression ratio",
- __func__);
-
- /* Clear content first. */
- dm_write_reg(compressor->ctx, DCP_REG(mmGRPH_COMPRESS_PITCH), 0);
-
- /* Write FBC Pitch. */
- set_reg_field_value(
- value,
- fbc_pitch,
- GRPH_COMPRESS_PITCH,
- GRPH_COMPRESS_PITCH);
- dm_write_reg(compressor->ctx, DCP_REG(mmGRPH_COMPRESS_PITCH), value);
-
-}
-
-void dce80_compressor_disable_lpt(struct compressor *compressor)
-{
- struct dce80_compressor *cp80 = TO_DCE80_COMPRESSOR(compressor);
- uint32_t value;
- uint32_t addr;
- uint32_t inx;
-
- /* Disable all pipes LPT Stutter */
- for (inx = 0; inx < 3; inx++) {
- value =
- dm_read_reg(
- compressor->ctx,
- DMIF_REG(mmDPG_PIPE_STUTTER_CONTROL_NONLPTCH));
- set_reg_field_value(
- value,
- 0,
- DPG_PIPE_STUTTER_CONTROL_NONLPTCH,
- STUTTER_ENABLE_NONLPTCH);
- dm_write_reg(
- compressor->ctx,
- DMIF_REG(mmDPG_PIPE_STUTTER_CONTROL_NONLPTCH),
- value);
- }
-
- /* Disable LPT */
- addr = mmLOW_POWER_TILING_CONTROL;
- value = dm_read_reg(compressor->ctx, addr);
- set_reg_field_value(
- value,
- 0,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_ENABLE);
- dm_write_reg(compressor->ctx, addr, value);
-
- /* Clear selection of Channel(s) containing Compressed Surface */
- addr = mmGMCON_LPT_TARGET;
- value = dm_read_reg(compressor->ctx, addr);
- set_reg_field_value(
- value,
- 0xFFFFFFFF,
- GMCON_LPT_TARGET,
- STCTRL_LPT_TARGET);
- dm_write_reg(compressor->ctx, mmGMCON_LPT_TARGET, value);
-}
-
-void dce80_compressor_enable_lpt(struct compressor *compressor)
-{
- struct dce80_compressor *cp80 = TO_DCE80_COMPRESSOR(compressor);
- uint32_t value;
- uint32_t addr;
- uint32_t value_control;
- uint32_t channels;
-
- /* Enable LPT Stutter from Display pipe */
- value = dm_read_reg(compressor->ctx,
- DMIF_REG(mmDPG_PIPE_STUTTER_CONTROL_NONLPTCH));
- set_reg_field_value(
- value,
- 1,
- DPG_PIPE_STUTTER_CONTROL_NONLPTCH,
- STUTTER_ENABLE_NONLPTCH);
- dm_write_reg(compressor->ctx,
- DMIF_REG(mmDPG_PIPE_STUTTER_CONTROL_NONLPTCH), value);
-
- /* Selection of Channel(s) containing Compressed Surface: 0xfffffff
- * will disable LPT.
- * STCTRL_LPT_TARGETn corresponds to channel n. */
- addr = mmLOW_POWER_TILING_CONTROL;
- value_control = dm_read_reg(compressor->ctx, addr);
- channels = get_reg_field_value(value_control,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_MODE);
-
- addr = mmGMCON_LPT_TARGET;
- value = dm_read_reg(compressor->ctx, addr);
- set_reg_field_value(
- value,
- channels + 1, /* not mentioned in programming guide,
- but follow DCE8.1 */
- GMCON_LPT_TARGET,
- STCTRL_LPT_TARGET);
- dm_write_reg(compressor->ctx, addr, value);
-
- /* Enable LPT */
- addr = mmLOW_POWER_TILING_CONTROL;
- value = dm_read_reg(compressor->ctx, addr);
- set_reg_field_value(
- value,
- 1,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_ENABLE);
- dm_write_reg(compressor->ctx, addr, value);
-}
-
-void dce80_compressor_program_lpt_control(
- struct compressor *compressor,
- struct compr_addr_and_pitch_params *params)
-{
- struct dce80_compressor *cp80 = TO_DCE80_COMPRESSOR(compressor);
- uint32_t rows_per_channel;
- uint32_t lpt_alignment;
- uint32_t source_view_width;
- uint32_t source_view_height;
- uint32_t lpt_control = 0;
-
- if (!compressor->options.bits.LPT_SUPPORT)
- return;
-
- lpt_control = dm_read_reg(compressor->ctx,
- mmLOW_POWER_TILING_CONTROL);
-
- /* POSSIBLE VALUES for Low Power Tiling Mode:
- * 00 - Use channel 0
- * 01 - Use Channel 0 and 1
- * 02 - Use Channel 0,1,2,3
- * 03 - reserved */
- switch (compressor->lpt_channels_num) {
- /* case 2:
- * Use Channel 0 & 1 / Not used for DCE 11 */
- case 1:
- /*Use Channel 0 for LPT for DCE 11 */
- set_reg_field_value(
- lpt_control,
- 0,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_MODE);
- break;
- default:
- dm_logger_write(
- compressor->ctx->logger, LOG_WARNING,
- "%s: Invalid selected DRAM channels for LPT!!!",
- __func__);
- break;
- }
-
- lpt_control = lpt_memory_control_config(cp80, lpt_control);
-
- /* Program LOW_POWER_TILING_ROWS_PER_CHAN field which depends on
- * FBC compressed surface pitch.
- * LOW_POWER_TILING_ROWS_PER_CHAN = Roundup ((Surface Height *
- * Surface Pitch) / (Row Size * Number of Channels *
- * Number of Banks)). */
- rows_per_channel = 0;
- lpt_alignment = lpt_size_alignment(cp80);
- source_view_width =
- align_to_chunks_number_per_line(
- cp80,
- params->source_view_width);
- source_view_height = (params->source_view_height + 1) & (~0x1);
-
- if (lpt_alignment != 0) {
- rows_per_channel = source_view_width * source_view_height * 4;
- rows_per_channel =
- (rows_per_channel % lpt_alignment) ?
- (rows_per_channel / lpt_alignment + 1) :
- rows_per_channel / lpt_alignment;
- }
-
- set_reg_field_value(
- lpt_control,
- rows_per_channel,
- LOW_POWER_TILING_CONTROL,
- LOW_POWER_TILING_ROWS_PER_CHAN);
-
- dm_write_reg(compressor->ctx,
- mmLOW_POWER_TILING_CONTROL, lpt_control);
-}
-
-/*
- * DCE 11 Frame Buffer Compression Implementation
- */
-
-void dce80_compressor_set_fbc_invalidation_triggers(
- struct compressor *compressor,
- uint32_t fbc_trigger)
-{
- /* Disable region hit event, FBC_MEMORY_REGION_MASK = 0 (bits 16-19)
- * for DCE 11 regions cannot be used - does not work with S/G
- */
- uint32_t addr = mmFBC_CLIENT_REGION_MASK;
- uint32_t value = dm_read_reg(compressor->ctx, addr);
-
- set_reg_field_value(
- value,
- 0,
- FBC_CLIENT_REGION_MASK,
- FBC_MEMORY_REGION_MASK);
- dm_write_reg(compressor->ctx, addr, value);
-
- /* Setup events when to clear all CSM entries (effectively marking
- * current compressed data invalid)
- * For DCE 11 CSM metadata 11111 means - "Not Compressed"
- * Used as the initial value of the metadata sent to the compressor
- * after invalidation, to indicate that the compressor should attempt
- * to compress all chunks on the current pass. Also used when the chunk
- * is not successfully written to memory.
- * When this CSM value is detected, FBC reads from the uncompressed
- * buffer. Set events according to passed in value, these events are
- * valid for DCE8:
- * - bit 0 - display register updated
- * - bit 28 - memory write from any client except from MCIF
- * - bit 29 - CG static screen signal is inactive
- * In addition, DCE8.1 also needs to set new DCE8.1 specific events
- * that are used to trigger invalidation on certain register changes,
- * for example enabling of Alpha Compression may trigger invalidation of
- * FBC once bit is set. These events are as follows:
- * - Bit 2 - FBC_GRPH_COMP_EN register updated
- * - Bit 3 - FBC_SRC_SEL register updated
- * - Bit 4 - FBC_MIN_COMPRESSION register updated
- * - Bit 5 - FBC_ALPHA_COMP_EN register updated
- * - Bit 6 - FBC_ZERO_ALPHA_CHUNK_SKIP_EN register updated
- * - Bit 7 - FBC_FORCE_COPY_TO_COMP_BUF register updated
- */
- addr = mmFBC_IDLE_FORCE_CLEAR_MASK;
- value = dm_read_reg(compressor->ctx, addr);
- set_reg_field_value(
- value,
- fbc_trigger |
- FBC_IDLE_FORCE_GRPH_COMP_EN |
- FBC_IDLE_FORCE_SRC_SEL_CHANGE |
- FBC_IDLE_FORCE_MIN_COMPRESSION_CHANGE |
- FBC_IDLE_FORCE_ALPHA_COMP_EN |
- FBC_IDLE_FORCE_ZERO_ALPHA_CHUNK_SKIP_EN |
- FBC_IDLE_FORCE_FORCE_COPY_TO_COMP_BUF,
- FBC_IDLE_FORCE_CLEAR_MASK,
- FBC_IDLE_FORCE_CLEAR_MASK);
- dm_write_reg(compressor->ctx, addr, value);
-}
-
-void dce80_compressor_construct(struct dce80_compressor *compressor,
- struct dc_context *ctx)
-{
- struct dc_bios *bp = ctx->dc_bios;
- struct embedded_panel_info panel_info;
-
- compressor->base.options.raw = 0;
- compressor->base.options.bits.FBC_SUPPORT = true;
- compressor->base.options.bits.LPT_SUPPORT = true;
- /* For DCE 11 always use one DRAM channel for LPT */
- compressor->base.lpt_channels_num = 1;
- compressor->base.options.bits.DUMMY_BACKEND = false;
-
- /* Check if this system has more than 1 DRAM channel; if only 1 then LPT
- * should not be supported */
- if (compressor->base.memory_bus_width == 64)
- compressor->base.options.bits.LPT_SUPPORT = false;
-
- compressor->base.options.bits.CLK_GATING_DISABLED = false;
-
- compressor->base.ctx = ctx;
- compressor->base.embedded_panel_h_size = 0;
- compressor->base.embedded_panel_v_size = 0;
- compressor->base.memory_bus_width = ctx->asic_id.vram_width;
- compressor->base.allocated_size = 0;
- compressor->base.preferred_requested_size = 0;
- compressor->base.min_compress_ratio = FBC_COMPRESS_RATIO_INVALID;
- compressor->base.banks_num = 0;
- compressor->base.raw_size = 0;
- compressor->base.channel_interleave_size = 0;
- compressor->base.dram_channels_num = 0;
- compressor->base.lpt_channels_num = 0;
- compressor->base.attached_inst = 0;
- compressor->base.is_enabled = false;
-
- if (BP_RESULT_OK ==
- bp->funcs->get_embedded_panel_info(bp, &panel_info)) {
- compressor->base.embedded_panel_h_size =
- panel_info.lcd_timing.horizontal_addressable;
- compressor->base.embedded_panel_v_size =
- panel_info.lcd_timing.vertical_addressable;
- }
-}
-
-struct compressor *dce80_compressor_create(struct dc_context *ctx)
-{
- struct dce80_compressor *cp80 =
- kzalloc(sizeof(struct dce80_compressor), GFP_KERNEL);
-
- if (!cp80)
- return NULL;
-
- dce80_compressor_construct(cp80, ctx);
- return &cp80->base;
-}
-
-void dce80_compressor_destroy(struct compressor **compressor)
-{
- kfree(TO_DCE80_COMPRESSOR(*compressor));
- *compressor = NULL;
-}
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_compressor.h b/drivers/gpu/drm/amd/display/dc/dce80/dce80_compressor.h
deleted file mode 100644
index cca58b044402..000000000000
--- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_compressor.h
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2012-15 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors: AMD
- *
- */
-
-#ifndef __DC_COMPRESSOR_DCE80_H__
-#define __DC_COMPRESSOR_DCE80_H__
-
-#include "../inc/compressor.h"
-
-#define TO_DCE80_COMPRESSOR(compressor)\
- container_of(compressor, struct dce80_compressor, base)
-
-struct dce80_compressor_reg_offsets {
- uint32_t dcp_offset;
- uint32_t dmif_offset;
-};
-
-struct dce80_compressor {
- struct compressor base;
- struct dce80_compressor_reg_offsets offsets;
-};
-
-struct compressor *dce80_compressor_create(struct dc_context *ctx);
-
-void dce80_compressor_construct(struct dce80_compressor *cp80,
- struct dc_context *ctx);
-
-void dce80_compressor_destroy(struct compressor **cp);
-
-/* FBC RELATED */
-void dce80_compressor_power_up_fbc(struct compressor *cp);
-
-void dce80_compressor_enable_fbc(struct compressor *cp, uint32_t paths_num,
- struct compr_addr_and_pitch_params *params);
-
-void dce80_compressor_disable_fbc(struct compressor *cp);
-
-void dce80_compressor_set_fbc_invalidation_triggers(struct compressor *cp,
- uint32_t fbc_trigger);
-
-void dce80_compressor_program_compressed_surface_address_and_pitch(
- struct compressor *cp,
- struct compr_addr_and_pitch_params *params);
-
-bool dce80_compressor_is_fbc_enabled_in_hw(struct compressor *cp,
- uint32_t *fbc_mapped_crtc_id);
-
-/* LPT RELATED */
-void dce80_compressor_enable_lpt(struct compressor *cp);
-
-void dce80_compressor_disable_lpt(struct compressor *cp);
-
-void dce80_compressor_program_lpt_control(struct compressor *cp,
- struct compr_addr_and_pitch_params *params);
-
-bool dce80_compressor_is_lpt_enabled_in_hw(struct compressor *cp);
-
-#endif
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 12/34] drm/amd/display: Re-use DCE100 display_power_gating for DCE80
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (10 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 11/34] drm/amd/display: Remove unused DCE80 compressor Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 13/34] drm/amd/display: Make couple functions in DCE80 TG static Harry Wentland
` (21 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Harry Wentland
Both functions are the same
Change-Id: Ic5b3aca8aab8f06531cfbab8acb53ff9c235ffaf
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
.../amd/display/dc/dce100/dce100_hw_sequencer.c | 2 +-
.../amd/display/dc/dce100/dce100_hw_sequencer.h | 4 +++
.../drm/amd/display/dc/dce80/dce80_hw_sequencer.c | 38 +---------------------
3 files changed, 6 insertions(+), 38 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.c
index 469af0587604..41f83ecd7469 100644
--- a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.c
@@ -69,7 +69,7 @@ static const struct dce100_hw_seq_reg_offsets reg_offsets[] = {
******************************************************************************/
/***************************PIPE_CONTROL***********************************/
-static bool dce100_enable_display_power_gating(
+bool dce100_enable_display_power_gating(
struct dc *dc,
uint8_t controller_id,
struct dc_bios *dcb,
diff --git a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h
index cb5384ef46c3..c6ec0ed6ec3d 100644
--- a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h
+++ b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h
@@ -38,5 +38,9 @@ void dce100_set_bandwidth(
struct dc_state *context,
bool decrease_allowed);
+bool dce100_enable_display_power_gating(struct dc *dc, uint8_t controller_id,
+ struct dc_bios *dcb,
+ enum pipe_gating_control power_gating);
+
#endif /* __DC_HWSS_DCE100_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c
index ccfcf1c0eeb3..6c6a1a16af19 100644
--- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c
@@ -70,47 +70,11 @@ static const struct dce80_hw_seq_reg_offsets reg_offsets[] = {
/***************************PIPE_CONTROL***********************************/
-static bool dce80_enable_display_power_gating(
- struct dc *dc,
- uint8_t controller_id,
- struct dc_bios *dcb,
- enum pipe_gating_control power_gating)
-{
- enum bp_result bp_result = BP_RESULT_OK;
- enum bp_pipe_control_action cntl;
- struct dc_context *ctx = dc->ctx;
-
- if (power_gating == PIPE_GATING_CONTROL_INIT)
- cntl = ASIC_PIPE_INIT;
- else if (power_gating == PIPE_GATING_CONTROL_ENABLE)
- cntl = ASIC_PIPE_ENABLE;
- else
- cntl = ASIC_PIPE_DISABLE;
-
- if (!(power_gating == PIPE_GATING_CONTROL_INIT && controller_id != 0)){
-
- bp_result = dcb->funcs->enable_disp_power_gating(
- dcb, controller_id + 1, cntl);
-
- /* Revert MASTER_UPDATE_MODE to 0 because bios sets it 2
- * by default when command table is called
- */
- dm_write_reg(ctx,
- HW_REG_CRTC(mmMASTER_UPDATE_MODE, controller_id),
- 0);
- }
-
- if (bp_result == BP_RESULT_OK)
- return true;
- else
- return false;
-}
-
void dce80_hw_sequencer_construct(struct dc *dc)
{
dce110_hw_sequencer_construct(dc);
- dc->hwss.enable_display_power_gating = dce80_enable_display_power_gating;
+ dc->hwss.enable_display_power_gating = dce100_enable_display_power_gating;
dc->hwss.pipe_control_lock = dce_pipe_control_lock;
dc->hwss.set_bandwidth = dce100_set_bandwidth;
}
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 13/34] drm/amd/display: Make couple functions in DCE80 TG static
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (11 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 12/34] drm/amd/display: Re-use DCE100 display_power_gating for DCE80 Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 14/34] drm/amd/display: Use hardware max low point when sampling OTF Harry Wentland
` (20 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Harry Wentland
Change-Id: Iadd754d57ec8e7bd5ce46bba97664b1c4932e395
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
.../amd/display/dc/dce80/dce80_timing_generator.c | 125 +++++++++++----------
.../amd/display/dc/dce80/dce80_timing_generator.h | 6 -
2 files changed, 63 insertions(+), 68 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c
index 265894851493..2934650e0434 100644
--- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c
+++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c
@@ -84,7 +84,7 @@ static const struct dce110_timing_generator_offsets reg_offsets[] = {
#define DCP_REG(reg) (reg + tg110->offsets.dcp)
#define DMIF_REG(reg) (reg + tg110->offsets.dmif)
-void program_pix_dur(struct timing_generator *tg, uint32_t pix_clk_khz)
+static void program_pix_dur(struct timing_generator *tg, uint32_t pix_clk_khz)
{
uint64_t pix_dur;
uint32_t addr = mmDMIF_PG0_DPG_PIPE_ARBITRATION_CONTROL1
@@ -115,6 +115,68 @@ static void program_timing(struct timing_generator *tg,
dce110_tg_program_timing(tg, timing, use_vbios);
}
+static void dce80_timing_generator_enable_advanced_request(
+ struct timing_generator *tg,
+ bool enable,
+ const struct dc_crtc_timing *timing)
+{
+ struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
+ uint32_t addr = CRTC_REG(mmCRTC_START_LINE_CONTROL);
+ uint32_t value = dm_read_reg(tg->ctx, addr);
+
+ if (enable) {
+ set_reg_field_value(
+ value,
+ 0,
+ CRTC_START_LINE_CONTROL,
+ CRTC_LEGACY_REQUESTOR_EN);
+ } else {
+ set_reg_field_value(
+ value,
+ 1,
+ CRTC_START_LINE_CONTROL,
+ CRTC_LEGACY_REQUESTOR_EN);
+ }
+
+ if ((timing->v_sync_width + timing->v_front_porch) <= 3) {
+ set_reg_field_value(
+ value,
+ 3,
+ CRTC_START_LINE_CONTROL,
+ CRTC_ADVANCED_START_LINE_POSITION);
+ set_reg_field_value(
+ value,
+ 0,
+ CRTC_START_LINE_CONTROL,
+ CRTC_PREFETCH_EN);
+ } else {
+ set_reg_field_value(
+ value,
+ 4,
+ CRTC_START_LINE_CONTROL,
+ CRTC_ADVANCED_START_LINE_POSITION);
+ set_reg_field_value(
+ value,
+ 1,
+ CRTC_START_LINE_CONTROL,
+ CRTC_PREFETCH_EN);
+ }
+
+ set_reg_field_value(
+ value,
+ 1,
+ CRTC_START_LINE_CONTROL,
+ CRTC_PROGRESSIVE_START_LINE_EARLY);
+
+ set_reg_field_value(
+ value,
+ 1,
+ CRTC_START_LINE_CONTROL,
+ CRTC_INTERLACE_START_LINE_EARLY);
+
+ dm_write_reg(tg->ctx, addr, value);
+}
+
static const struct timing_generator_funcs dce80_tg_funcs = {
.validate_timing = dce110_tg_validate_timing,
.program_timing = program_timing,
@@ -176,64 +238,3 @@ void dce80_timing_generator_construct(
tg110->min_h_back_porch = 4;
}
-void dce80_timing_generator_enable_advanced_request(
- struct timing_generator *tg,
- bool enable,
- const struct dc_crtc_timing *timing)
-{
- struct dce110_timing_generator *tg110 = DCE110TG_FROM_TG(tg);
- uint32_t addr = CRTC_REG(mmCRTC_START_LINE_CONTROL);
- uint32_t value = dm_read_reg(tg->ctx, addr);
-
- if (enable) {
- set_reg_field_value(
- value,
- 0,
- CRTC_START_LINE_CONTROL,
- CRTC_LEGACY_REQUESTOR_EN);
- } else {
- set_reg_field_value(
- value,
- 1,
- CRTC_START_LINE_CONTROL,
- CRTC_LEGACY_REQUESTOR_EN);
- }
-
- if ((timing->v_sync_width + timing->v_front_porch) <= 3) {
- set_reg_field_value(
- value,
- 3,
- CRTC_START_LINE_CONTROL,
- CRTC_ADVANCED_START_LINE_POSITION);
- set_reg_field_value(
- value,
- 0,
- CRTC_START_LINE_CONTROL,
- CRTC_PREFETCH_EN);
- } else {
- set_reg_field_value(
- value,
- 4,
- CRTC_START_LINE_CONTROL,
- CRTC_ADVANCED_START_LINE_POSITION);
- set_reg_field_value(
- value,
- 1,
- CRTC_START_LINE_CONTROL,
- CRTC_PREFETCH_EN);
- }
-
- set_reg_field_value(
- value,
- 1,
- CRTC_START_LINE_CONTROL,
- CRTC_PROGRESSIVE_START_LINE_EARLY);
-
- set_reg_field_value(
- value,
- 1,
- CRTC_START_LINE_CONTROL,
- CRTC_INTERLACE_START_LINE_EARLY);
-
- dm_write_reg(tg->ctx, addr, value);
-}
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.h b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.h
index 9cebb24c94c8..8ff1b06bcd8b 100644
--- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.h
+++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.h
@@ -36,10 +36,4 @@ void dce80_timing_generator_construct(
uint32_t instance,
const struct dce110_timing_generator_offsets *offsets);
-/******** HW programming ************/
-void dce80_timing_generator_enable_advanced_request(
- struct timing_generator *tg,
- bool enable,
- const struct dc_crtc_timing *timing);
-
#endif /* __DC_TIMING_GENERATOR_DCE80_H__ */
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 14/34] drm/amd/display: Use hardware max low point when sampling OTF
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (12 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 13/34] drm/amd/display: Make couple functions in DCE80 TG static Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 15/34] drm/amd/display: Add color module's gamma helpers to Linux build Harry Wentland
` (19 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
The MAX_LOW_POINT macro should reflect the maximum low point within
hardware. Otherwise, sampling for the hardware points from the output
transfer function (OTF) will be incorrect.
Also, fix usage of MAX_LOW_POINT accordingly.
Change-Id: I6affb8e987ac786c2c9ac8fba4bee5384dc54d07
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
index 9e97cd74b0ec..bc698644d42a 100644
--- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
@@ -409,7 +409,7 @@ static bool convert_to_custom_float(struct pwl_result_data *rgb_resulted,
return true;
}
-#define MAX_LOW_POINT 11
+#define MAX_LOW_POINT 25
#define NUMBER_REGIONS 16
#define NUMBER_SW_SEGMENTS 16
@@ -443,8 +443,8 @@ dce110_translate_regamma_to_hw_format(const struct dc_transfer_func *output_tf,
/* 16 segments
* segments are from 2^-11 to 2^5
*/
- region_start = -MAX_LOW_POINT;
- region_end = NUMBER_REGIONS - MAX_LOW_POINT;
+ region_start = -11;
+ region_end = region_start + NUMBER_REGIONS;
for (i = 0; i < NUMBER_REGIONS; i++)
seg_distr[i] = 4;
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 15/34] drm/amd/display: Add color module's gamma helpers to Linux build
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (13 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 14/34] drm/amd/display: Use hardware max low point when sampling OTF Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 16/34] drm/amd/display: Implement color management Harry Wentland
` (18 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
Also guard includes that we don't need.
Change-Id: I75201e80dd991b8a15e7a082332b4b6c920fb17e
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
---
drivers/gpu/drm/amd/display/Makefile | 3 +-
drivers/gpu/drm/amd/display/modules/color/Makefile | 31 +
.../drm/amd/display/modules/color/color_gamma.c | 1070 ++++++++++++++++++++
.../drm/amd/display/modules/color/color_gamma.h | 45 +
4 files changed, 1148 insertions(+), 1 deletion(-)
create mode 100644 drivers/gpu/drm/amd/display/modules/color/Makefile
create mode 100644 drivers/gpu/drm/amd/display/modules/color/color_gamma.c
create mode 100644 drivers/gpu/drm/amd/display/modules/color/color_gamma.h
diff --git a/drivers/gpu/drm/amd/display/Makefile b/drivers/gpu/drm/amd/display/Makefile
index c27c81cdeed3..3d14478913de 100644
--- a/drivers/gpu/drm/amd/display/Makefile
+++ b/drivers/gpu/drm/amd/display/Makefile
@@ -32,11 +32,12 @@ subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/inc/
subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/inc/hw
subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/modules/inc
subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/modules/freesync
+subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/modules/color
#TODO: remove when Timing Sync feature is complete
subdir-ccflags-y += -DBUILD_FEATURE_TIMING_SYNC=0
-DAL_LIBS = amdgpu_dm dc modules/freesync
+DAL_LIBS = amdgpu_dm dc modules/freesync modules/color
AMD_DAL = $(addsuffix /Makefile, $(addprefix $(FULL_AMD_DISPLAY_PATH)/,$(DAL_LIBS)))
diff --git a/drivers/gpu/drm/amd/display/modules/color/Makefile b/drivers/gpu/drm/amd/display/modules/color/Makefile
new file mode 100644
index 000000000000..65c33a76951a
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/modules/color/Makefile
@@ -0,0 +1,31 @@
+#
+# Copyright 2018 Advanced Micro Devices, Inc.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+#
+# Makefile for the color sub-module of DAL.
+#
+
+MOD_COLOR = color_gamma.o
+
+AMD_DAL_MOD_COLOR = $(addprefix $(AMDDALPATH)/modules/color/,$(MOD_COLOR))
+#$(info ************ DAL COLOR MODULE MAKEFILE ************)
+
+AMD_DISPLAY_FILES += $(AMD_DAL_MOD_COLOR)
diff --git a/drivers/gpu/drm/amd/display/modules/color/color_gamma.c b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c
new file mode 100644
index 000000000000..fde3ae8b12a5
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c
@@ -0,0 +1,1070 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "dc.h"
+#include "opp.h"
+#include "color_gamma.h"
+
+/* MAX_HW_POINTS = NUM_REGIONS * NUM_PTS_IN_REGION */
+#define NUM_PTS_IN_REGION 16
+#define NUM_REGIONS 32
+#define MAX_HW_POINTS 512
+
+static struct hw_x_point coordinates_x[MAX_HW_POINTS + 2];
+static struct fixed31_32 pq_table[MAX_HW_POINTS + 2];
+static bool pq_initialized; /* = false; */
+
+/* one-time setup of X points */
+void setup_x_points_distribution(void)
+{
+ struct fixed31_32 region_size = dal_fixed31_32_from_int(128);
+ int32_t segment;
+ uint32_t seg_offset;
+ uint32_t index;
+ struct fixed31_32 increment;
+
+ coordinates_x[NUM_REGIONS * NUM_PTS_IN_REGION].x = region_size;
+ coordinates_x[NUM_REGIONS * NUM_PTS_IN_REGION + 1].x = region_size;
+
+ for (segment = 6; segment > (6 - NUM_REGIONS); segment--) {
+ region_size = dal_fixed31_32_div_int(region_size, 2);
+ increment = dal_fixed31_32_div_int(region_size,
+ NUM_PTS_IN_REGION);
+ seg_offset = (segment + (NUM_REGIONS - 7)) * NUM_PTS_IN_REGION;
+ coordinates_x[seg_offset].x = region_size;
+
+ for (index = seg_offset + 1;
+ index < seg_offset + NUM_PTS_IN_REGION;
+ index++) {
+ coordinates_x[index].x = dal_fixed31_32_add
+ (coordinates_x[index-1].x, increment);
+ }
+ }
+}
+
+static void compute_pq(struct fixed31_32 in_x, struct fixed31_32 *out_y)
+{
+ /* consts for PQ gamma formula. */
+ const struct fixed31_32 m1 =
+ dal_fixed31_32_from_fraction(159301758, 1000000000);
+ const struct fixed31_32 m2 =
+ dal_fixed31_32_from_fraction(7884375, 100000);
+ const struct fixed31_32 c1 =
+ dal_fixed31_32_from_fraction(8359375, 10000000);
+ const struct fixed31_32 c2 =
+ dal_fixed31_32_from_fraction(188515625, 10000000);
+ const struct fixed31_32 c3 =
+ dal_fixed31_32_from_fraction(186875, 10000);
+
+ struct fixed31_32 l_pow_m1;
+ struct fixed31_32 base;
+
+ if (dal_fixed31_32_lt(in_x, dal_fixed31_32_zero))
+ in_x = dal_fixed31_32_zero;
+
+ l_pow_m1 = dal_fixed31_32_pow(in_x, m1);
+ base = dal_fixed31_32_div(
+ dal_fixed31_32_add(c1,
+ (dal_fixed31_32_mul(c2, l_pow_m1))),
+ dal_fixed31_32_add(dal_fixed31_32_one,
+ (dal_fixed31_32_mul(c3, l_pow_m1))));
+ *out_y = dal_fixed31_32_pow(base, m2);
+}
+
+/* one-time pre-compute PQ values - only for sdr_white_level 80 */
+void precompute_pq(void)
+{
+ int i;
+ struct fixed31_32 x;
+ const struct hw_x_point *coord_x = coordinates_x + 32;
+ struct fixed31_32 scaling_factor =
+ dal_fixed31_32_from_fraction(80, 10000);
+
+ /* pow function has problems with arguments too small */
+ for (i = 0; i < 32; i++)
+ pq_table[i] = dal_fixed31_32_zero;
+
+ for (i = 32; i <= MAX_HW_POINTS; i++) {
+ x = dal_fixed31_32_mul(coord_x->x, scaling_factor);
+ compute_pq(x, &pq_table[i]);
+ ++coord_x;
+ }
+}
+
+struct dividers {
+ struct fixed31_32 divider1;
+ struct fixed31_32 divider2;
+ struct fixed31_32 divider3;
+};
+
+static void build_regamma_coefficients(struct gamma_coefficients *coefficients)
+{
+ /* sRGB should apply 2.4 */
+ static const int32_t numerator01[3] = { 31308, 31308, 31308 };
+ static const int32_t numerator02[3] = { 12920, 12920, 12920 };
+ static const int32_t numerator03[3] = { 55, 55, 55 };
+ static const int32_t numerator04[3] = { 55, 55, 55 };
+ static const int32_t numerator05[3] = { 2400, 2400, 2400 };
+
+ const int32_t *numerator1;
+ const int32_t *numerator2;
+ const int32_t *numerator3;
+ const int32_t *numerator4;
+ const int32_t *numerator5;
+
+ uint32_t i = 0;
+
+ numerator1 = numerator01;
+ numerator2 = numerator02;
+ numerator3 = numerator03;
+ numerator4 = numerator04;
+ numerator5 = numerator05;
+
+ do {
+ coefficients->a0[i] = dal_fixed31_32_from_fraction(
+ numerator1[i], 10000000);
+ coefficients->a1[i] = dal_fixed31_32_from_fraction(
+ numerator2[i], 1000);
+ coefficients->a2[i] = dal_fixed31_32_from_fraction(
+ numerator3[i], 1000);
+ coefficients->a3[i] = dal_fixed31_32_from_fraction(
+ numerator4[i], 1000);
+ coefficients->user_gamma[i] = dal_fixed31_32_from_fraction(
+ numerator5[i], 1000);
+
+ ++i;
+ } while (i != ARRAY_SIZE(coefficients->a0));
+}
+
+static struct fixed31_32 translate_from_linear_space(
+ struct fixed31_32 arg,
+ struct fixed31_32 a0,
+ struct fixed31_32 a1,
+ struct fixed31_32 a2,
+ struct fixed31_32 a3,
+ struct fixed31_32 gamma)
+{
+ const struct fixed31_32 one = dal_fixed31_32_from_int(1);
+
+ if (dal_fixed31_32_lt(one, arg))
+ return one;
+
+ if (dal_fixed31_32_le(arg, dal_fixed31_32_neg(a0)))
+ return dal_fixed31_32_sub(
+ a2,
+ dal_fixed31_32_mul(
+ dal_fixed31_32_add(
+ one,
+ a3),
+ dal_fixed31_32_pow(
+ dal_fixed31_32_neg(arg),
+ dal_fixed31_32_recip(gamma))));
+ else if (dal_fixed31_32_le(a0, arg))
+ return dal_fixed31_32_sub(
+ dal_fixed31_32_mul(
+ dal_fixed31_32_add(
+ one,
+ a3),
+ dal_fixed31_32_pow(
+ arg,
+ dal_fixed31_32_recip(gamma))),
+ a2);
+ else
+ return dal_fixed31_32_mul(
+ arg,
+ a1);
+}
+
+static inline struct fixed31_32 translate_from_linear_space_ex(
+ struct fixed31_32 arg,
+ struct gamma_coefficients *coeff,
+ uint32_t color_index)
+{
+ return translate_from_linear_space(
+ arg,
+ coeff->a0[color_index],
+ coeff->a1[color_index],
+ coeff->a2[color_index],
+ coeff->a3[color_index],
+ coeff->user_gamma[color_index]);
+}
+
+static bool find_software_points(
+ const struct dc_gamma *ramp,
+ const struct gamma_pixel *axis_x,
+ struct fixed31_32 hw_point,
+ enum channel_name channel,
+ uint32_t *index_to_start,
+ uint32_t *index_left,
+ uint32_t *index_right,
+ enum hw_point_position *pos)
+{
+ const uint32_t max_number = ramp->num_entries + 3;
+
+ struct fixed31_32 left, right;
+
+ uint32_t i = *index_to_start;
+
+ while (i < max_number) {
+ if (channel == CHANNEL_NAME_RED) {
+ left = axis_x[i].r;
+
+ if (i < max_number - 1)
+ right = axis_x[i + 1].r;
+ else
+ right = axis_x[max_number - 1].r;
+ } else if (channel == CHANNEL_NAME_GREEN) {
+ left = axis_x[i].g;
+
+ if (i < max_number - 1)
+ right = axis_x[i + 1].g;
+ else
+ right = axis_x[max_number - 1].g;
+ } else {
+ left = axis_x[i].b;
+
+ if (i < max_number - 1)
+ right = axis_x[i + 1].b;
+ else
+ right = axis_x[max_number - 1].b;
+ }
+
+ if (dal_fixed31_32_le(left, hw_point) &&
+ dal_fixed31_32_le(hw_point, right)) {
+ *index_to_start = i;
+ *index_left = i;
+
+ if (i < max_number - 1)
+ *index_right = i + 1;
+ else
+ *index_right = max_number - 1;
+
+ *pos = HW_POINT_POSITION_MIDDLE;
+
+ return true;
+ } else if ((i == *index_to_start) &&
+ dal_fixed31_32_le(hw_point, left)) {
+ *index_to_start = i;
+ *index_left = i;
+ *index_right = i;
+
+ *pos = HW_POINT_POSITION_LEFT;
+
+ return true;
+ } else if ((i == max_number - 1) &&
+ dal_fixed31_32_le(right, hw_point)) {
+ *index_to_start = i;
+ *index_left = i;
+ *index_right = i;
+
+ *pos = HW_POINT_POSITION_RIGHT;
+
+ return true;
+ }
+
+ ++i;
+ }
+
+ return false;
+}
+
+static bool build_custom_gamma_mapping_coefficients_worker(
+ const struct dc_gamma *ramp,
+ struct pixel_gamma_point *coeff,
+ const struct hw_x_point *coordinates_x,
+ const struct gamma_pixel *axis_x,
+ enum channel_name channel,
+ uint32_t number_of_points)
+{
+ uint32_t i = 0;
+
+ while (i <= number_of_points) {
+ struct fixed31_32 coord_x;
+
+ uint32_t index_to_start = 0;
+ uint32_t index_left = 0;
+ uint32_t index_right = 0;
+
+ enum hw_point_position hw_pos;
+
+ struct gamma_point *point;
+
+ struct fixed31_32 left_pos;
+ struct fixed31_32 right_pos;
+
+ /*
+ * TODO: confirm enum in surface_pixel_format
+ * if (pixel_format == PIXEL_FORMAT_FP16)
+ *coord_x = coordinates_x[i].adjusted_x;
+ *else
+ */
+ if (channel == CHANNEL_NAME_RED)
+ coord_x = coordinates_x[i].regamma_y_red;
+ else if (channel == CHANNEL_NAME_GREEN)
+ coord_x = coordinates_x[i].regamma_y_green;
+ else
+ coord_x = coordinates_x[i].regamma_y_blue;
+
+ if (!find_software_points(
+ ramp, axis_x, coord_x, channel,
+ &index_to_start, &index_left, &index_right, &hw_pos)) {
+ BREAK_TO_DEBUGGER();
+ return false;
+ }
+
+ if (index_left >= ramp->num_entries + 3) {
+ BREAK_TO_DEBUGGER();
+ return false;
+ }
+
+ if (index_right >= ramp->num_entries + 3) {
+ BREAK_TO_DEBUGGER();
+ return false;
+ }
+
+ if (channel == CHANNEL_NAME_RED) {
+ point = &coeff[i].r;
+
+ left_pos = axis_x[index_left].r;
+ right_pos = axis_x[index_right].r;
+ } else if (channel == CHANNEL_NAME_GREEN) {
+ point = &coeff[i].g;
+
+ left_pos = axis_x[index_left].g;
+ right_pos = axis_x[index_right].g;
+ } else {
+ point = &coeff[i].b;
+
+ left_pos = axis_x[index_left].b;
+ right_pos = axis_x[index_right].b;
+ }
+
+ if (hw_pos == HW_POINT_POSITION_MIDDLE)
+ point->coeff = dal_fixed31_32_div(
+ dal_fixed31_32_sub(
+ coord_x,
+ left_pos),
+ dal_fixed31_32_sub(
+ right_pos,
+ left_pos));
+ else if (hw_pos == HW_POINT_POSITION_LEFT)
+ point->coeff = dal_fixed31_32_zero;
+ else if (hw_pos == HW_POINT_POSITION_RIGHT)
+ point->coeff = dal_fixed31_32_from_int(2);
+ else {
+ BREAK_TO_DEBUGGER();
+ return false;
+ }
+
+ point->left_index = index_left;
+ point->right_index = index_right;
+ point->pos = hw_pos;
+
+ ++i;
+ }
+
+ return true;
+}
+
+static struct fixed31_32 calculate_mapped_value(
+ struct pwl_float_data *rgb,
+ const struct pixel_gamma_point *coeff,
+ enum channel_name channel,
+ uint32_t max_index)
+{
+ const struct gamma_point *point;
+
+ struct fixed31_32 result;
+
+ if (channel == CHANNEL_NAME_RED)
+ point = &coeff->r;
+ else if (channel == CHANNEL_NAME_GREEN)
+ point = &coeff->g;
+ else
+ point = &coeff->b;
+
+ if ((point->left_index < 0) || (point->left_index > max_index)) {
+ BREAK_TO_DEBUGGER();
+ return dal_fixed31_32_zero;
+ }
+
+ if ((point->right_index < 0) || (point->right_index > max_index)) {
+ BREAK_TO_DEBUGGER();
+ return dal_fixed31_32_zero;
+ }
+
+ if (point->pos == HW_POINT_POSITION_MIDDLE)
+ if (channel == CHANNEL_NAME_RED)
+ result = dal_fixed31_32_add(
+ dal_fixed31_32_mul(
+ point->coeff,
+ dal_fixed31_32_sub(
+ rgb[point->right_index].r,
+ rgb[point->left_index].r)),
+ rgb[point->left_index].r);
+ else if (channel == CHANNEL_NAME_GREEN)
+ result = dal_fixed31_32_add(
+ dal_fixed31_32_mul(
+ point->coeff,
+ dal_fixed31_32_sub(
+ rgb[point->right_index].g,
+ rgb[point->left_index].g)),
+ rgb[point->left_index].g);
+ else
+ result = dal_fixed31_32_add(
+ dal_fixed31_32_mul(
+ point->coeff,
+ dal_fixed31_32_sub(
+ rgb[point->right_index].b,
+ rgb[point->left_index].b)),
+ rgb[point->left_index].b);
+ else if (point->pos == HW_POINT_POSITION_LEFT) {
+ BREAK_TO_DEBUGGER();
+ result = dal_fixed31_32_zero;
+ } else {
+ BREAK_TO_DEBUGGER();
+ result = dal_fixed31_32_one;
+ }
+
+ return result;
+}
+
+static void build_regamma_curve_pq(struct pwl_float_data_ex *rgb_regamma,
+ uint32_t hw_points_num,
+ const struct hw_x_point *coordinate_x,
+ uint32_t sdr_white_level)
+{
+ uint32_t i, start_index;
+
+ struct pwl_float_data_ex *rgb = rgb_regamma;
+ const struct hw_x_point *coord_x = coordinate_x;
+ struct fixed31_32 x;
+ struct fixed31_32 output;
+ struct fixed31_32 scaling_factor =
+ dal_fixed31_32_from_fraction(sdr_white_level, 10000);
+
+ if (!pq_initialized && sdr_white_level == 80) {
+ precompute_pq();
+ pq_initialized = true;
+ }
+
+ /* TODO: start index is from segment 2^-24, skipping first segment
+ * due to x values too small for power calculations
+ */
+ start_index = 32;
+ rgb += start_index;
+ coord_x += start_index;
+
+ /* use coord_x to retrieve coordinates chosen base on given user curve
+ * the x values are exponentially distributed and currently it is hard
+ * coded, the user curve shape is ignored. Need to recalculate coord_x
+ * based on input curve, translation from 256/1025 to 128 PWL points.
+ */
+ for (i = start_index; i <= hw_points_num; i++) {
+ /* Multiply 0.008 as regamma is 0-1 and FP16 input is 0-125.
+ * FP 1.0 = 80nits
+ */
+ if (sdr_white_level == 80) {
+ output = pq_table[i];
+ } else {
+ x = dal_fixed31_32_mul(coord_x->x, scaling_factor);
+ compute_pq(x, &output);
+ }
+
+ /* should really not happen? */
+ if (dal_fixed31_32_lt(output, dal_fixed31_32_zero))
+ output = dal_fixed31_32_zero;
+ else if (dal_fixed31_32_lt(dal_fixed31_32_one, output))
+ output = dal_fixed31_32_one;
+
+ rgb->r = output;
+ rgb->g = output;
+ rgb->b = output;
+
+ ++coord_x;
+ ++rgb;
+ }
+}
+
+static void build_regamma_curve(struct pwl_float_data_ex *rgb_regamma,
+ uint32_t hw_points_num,
+ const struct hw_x_point *coordinate_x)
+{
+ uint32_t i;
+
+ struct gamma_coefficients coeff;
+ struct pwl_float_data_ex *rgb = rgb_regamma;
+ const struct hw_x_point *coord_x = coordinate_x;
+
+ build_regamma_coefficients(&coeff);
+
+ /* Use opp110->regamma.coordinates_x to retrieve
+ * coordinates chosen base on given user curve (future task).
+ * The x values are exponentially distributed and currently
+ * it is hard-coded, the user curve shape is ignored.
+ * The future task is to recalculate opp110-
+ * regamma.coordinates_x based on input/user curve,
+ * translation from 256/1025 to 128 pwl points.
+ */
+
+ i = 0;
+
+ while (i != hw_points_num + 1) {
+ rgb->r = translate_from_linear_space_ex(
+ coord_x->x, &coeff, 0);
+ rgb->g = translate_from_linear_space_ex(
+ coord_x->x, &coeff, 1);
+ rgb->b = translate_from_linear_space_ex(
+ coord_x->x, &coeff, 2);
+
+ ++coord_x;
+ ++rgb;
+ ++i;
+ }
+}
+
+static bool scale_gamma(struct pwl_float_data *pwl_rgb,
+ const struct dc_gamma *ramp,
+ struct dividers dividers)
+{
+ const struct fixed31_32 max_driver = dal_fixed31_32_from_int(0xFFFF);
+ const struct fixed31_32 max_os = dal_fixed31_32_from_int(0xFF00);
+ struct fixed31_32 scaler = max_os;
+ uint32_t i;
+ struct pwl_float_data *rgb = pwl_rgb;
+ struct pwl_float_data *rgb_last = rgb + ramp->num_entries - 1;
+
+ i = 0;
+
+ do {
+ if (dal_fixed31_32_lt(max_os, ramp->entries.red[i]) ||
+ dal_fixed31_32_lt(max_os, ramp->entries.green[i]) ||
+ dal_fixed31_32_lt(max_os, ramp->entries.blue[i])) {
+ scaler = max_driver;
+ break;
+ }
+ ++i;
+ } while (i != ramp->num_entries);
+
+ i = 0;
+
+ do {
+ rgb->r = dal_fixed31_32_div(
+ ramp->entries.red[i], scaler);
+ rgb->g = dal_fixed31_32_div(
+ ramp->entries.green[i], scaler);
+ rgb->b = dal_fixed31_32_div(
+ ramp->entries.blue[i], scaler);
+
+ ++rgb;
+ ++i;
+ } while (i != ramp->num_entries);
+
+ rgb->r = dal_fixed31_32_mul(rgb_last->r,
+ dividers.divider1);
+ rgb->g = dal_fixed31_32_mul(rgb_last->g,
+ dividers.divider1);
+ rgb->b = dal_fixed31_32_mul(rgb_last->b,
+ dividers.divider1);
+
+ ++rgb;
+
+ rgb->r = dal_fixed31_32_mul(rgb_last->r,
+ dividers.divider2);
+ rgb->g = dal_fixed31_32_mul(rgb_last->g,
+ dividers.divider2);
+ rgb->b = dal_fixed31_32_mul(rgb_last->b,
+ dividers.divider2);
+
+ ++rgb;
+
+ rgb->r = dal_fixed31_32_mul(rgb_last->r,
+ dividers.divider3);
+ rgb->g = dal_fixed31_32_mul(rgb_last->g,
+ dividers.divider3);
+ rgb->b = dal_fixed31_32_mul(rgb_last->b,
+ dividers.divider3);
+
+ return true;
+}
+
+static bool scale_gamma_dx(struct pwl_float_data *pwl_rgb,
+ const struct dc_gamma *ramp,
+ struct dividers dividers)
+{
+ uint32_t i;
+ struct fixed31_32 min = dal_fixed31_32_zero;
+ struct fixed31_32 max = dal_fixed31_32_one;
+
+ struct fixed31_32 delta = dal_fixed31_32_zero;
+ struct fixed31_32 offset = dal_fixed31_32_zero;
+
+ for (i = 0 ; i < ramp->num_entries; i++) {
+ if (dal_fixed31_32_lt(ramp->entries.red[i], min))
+ min = ramp->entries.red[i];
+
+ if (dal_fixed31_32_lt(ramp->entries.green[i], min))
+ min = ramp->entries.green[i];
+
+ if (dal_fixed31_32_lt(ramp->entries.blue[i], min))
+ min = ramp->entries.blue[i];
+
+ if (dal_fixed31_32_lt(max, ramp->entries.red[i]))
+ max = ramp->entries.red[i];
+
+ if (dal_fixed31_32_lt(max, ramp->entries.green[i]))
+ max = ramp->entries.green[i];
+
+ if (dal_fixed31_32_lt(max, ramp->entries.blue[i]))
+ max = ramp->entries.blue[i];
+ }
+
+ if (dal_fixed31_32_lt(min, dal_fixed31_32_zero))
+ delta = dal_fixed31_32_neg(min);
+
+ offset = dal_fixed31_32_add(min, max);
+
+ for (i = 0 ; i < ramp->num_entries; i++) {
+ pwl_rgb[i].r = dal_fixed31_32_div(
+ dal_fixed31_32_add(
+ ramp->entries.red[i], delta), offset);
+ pwl_rgb[i].g = dal_fixed31_32_div(
+ dal_fixed31_32_add(
+ ramp->entries.green[i], delta), offset);
+ pwl_rgb[i].b = dal_fixed31_32_div(
+ dal_fixed31_32_add(
+ ramp->entries.blue[i], delta), offset);
+
+ }
+
+ pwl_rgb[i].r = dal_fixed31_32_sub(dal_fixed31_32_mul_int(
+ pwl_rgb[i-1].r, 2), pwl_rgb[i-2].r);
+ pwl_rgb[i].g = dal_fixed31_32_sub(dal_fixed31_32_mul_int(
+ pwl_rgb[i-1].g, 2), pwl_rgb[i-2].g);
+ pwl_rgb[i].b = dal_fixed31_32_sub(dal_fixed31_32_mul_int(
+ pwl_rgb[i-1].b, 2), pwl_rgb[i-2].b);
+ ++i;
+ pwl_rgb[i].r = dal_fixed31_32_sub(dal_fixed31_32_mul_int(
+ pwl_rgb[i-1].r, 2), pwl_rgb[i-2].r);
+ pwl_rgb[i].g = dal_fixed31_32_sub(dal_fixed31_32_mul_int(
+ pwl_rgb[i-1].g, 2), pwl_rgb[i-2].g);
+ pwl_rgb[i].b = dal_fixed31_32_sub(dal_fixed31_32_mul_int(
+ pwl_rgb[i-1].b, 2), pwl_rgb[i-2].b);
+
+ return true;
+}
+
+/*
+ * RS3+ color transform DDI - 1D LUT adjustment is composed with regamma here
+ * Input is evenly distributed in the output color space as specified in
+ * SetTimings
+ *
+ * Interpolation details:
+ * 1D LUT has 4096 values which give curve correction in 0-1 float range
+ * for evenly spaced points in 0-1 range. lut1D[index] gives correction
+ * for index/4095.
+ * First we find index for which:
+ * index/4095 < regamma_y < (index+1)/4095 =>
+ * index < 4095*regamma_y < index + 1
+ * norm_y = 4095*regamma_y, and index is just truncating to nearest integer
+ * lut1 = lut1D[index], lut2 = lut1D[index+1]
+ *
+ *adjustedY is then linearly interpolating regamma Y between lut1 and lut2
+ */
+static void apply_lut_1d(
+ const struct dc_gamma *ramp,
+ uint32_t num_hw_points,
+ struct dc_transfer_func_distributed_points *tf_pts)
+{
+ int i = 0;
+ int color = 0;
+ struct fixed31_32 *regamma_y;
+ struct fixed31_32 norm_y;
+ struct fixed31_32 lut1;
+ struct fixed31_32 lut2;
+ const int max_lut_index = 4095;
+ const struct fixed31_32 max_lut_index_f =
+ dal_fixed31_32_from_int_nonconst(max_lut_index);
+ int32_t index = 0, index_next = 0;
+ struct fixed31_32 index_f;
+ struct fixed31_32 delta_lut;
+ struct fixed31_32 delta_index;
+
+ if (ramp->type != GAMMA_CS_TFM_1D)
+ return; // this is not expected
+
+ for (i = 0; i < num_hw_points; i++) {
+ for (color = 0; color < 3; color++) {
+ if (color == 0)
+ regamma_y = &tf_pts->red[i];
+ else if (color == 1)
+ regamma_y = &tf_pts->green[i];
+ else
+ regamma_y = &tf_pts->blue[i];
+
+ norm_y = dal_fixed31_32_mul(max_lut_index_f,
+ *regamma_y);
+ index = dal_fixed31_32_floor(norm_y);
+ index_f = dal_fixed31_32_from_int_nonconst(index);
+
+ if (index < 0 || index > max_lut_index)
+ continue;
+
+ index_next = (index == max_lut_index) ? index : index+1;
+
+ if (color == 0) {
+ lut1 = ramp->entries.red[index];
+ lut2 = ramp->entries.red[index_next];
+ } else if (color == 1) {
+ lut1 = ramp->entries.green[index];
+ lut2 = ramp->entries.green[index_next];
+ } else {
+ lut1 = ramp->entries.blue[index];
+ lut2 = ramp->entries.blue[index_next];
+ }
+
+ // we have everything now, so interpolate
+ delta_lut = dal_fixed31_32_sub(lut2, lut1);
+ delta_index = dal_fixed31_32_sub(norm_y, index_f);
+
+ *regamma_y = dal_fixed31_32_add(lut1,
+ dal_fixed31_32_mul(delta_index, delta_lut));
+ }
+ }
+}
+
+static void build_evenly_distributed_points(
+ struct gamma_pixel *points,
+ uint32_t numberof_points,
+ struct dividers dividers)
+{
+ struct gamma_pixel *p = points;
+ struct gamma_pixel *p_last = p + numberof_points - 1;
+
+ uint32_t i = 0;
+
+ do {
+ struct fixed31_32 value = dal_fixed31_32_from_fraction(i,
+ numberof_points - 1);
+
+ p->r = value;
+ p->g = value;
+ p->b = value;
+
+ ++p;
+ ++i;
+ } while (i != numberof_points);
+
+ p->r = dal_fixed31_32_div(p_last->r, dividers.divider1);
+ p->g = dal_fixed31_32_div(p_last->g, dividers.divider1);
+ p->b = dal_fixed31_32_div(p_last->b, dividers.divider1);
+
+ ++p;
+
+ p->r = dal_fixed31_32_div(p_last->r, dividers.divider2);
+ p->g = dal_fixed31_32_div(p_last->g, dividers.divider2);
+ p->b = dal_fixed31_32_div(p_last->b, dividers.divider2);
+
+ ++p;
+
+ p->r = dal_fixed31_32_div(p_last->r, dividers.divider3);
+ p->g = dal_fixed31_32_div(p_last->g, dividers.divider3);
+ p->b = dal_fixed31_32_div(p_last->b, dividers.divider3);
+}
+
+static inline void copy_rgb_regamma_to_coordinates_x(
+ struct hw_x_point *coordinates_x,
+ uint32_t hw_points_num,
+ const struct pwl_float_data_ex *rgb_ex)
+{
+ struct hw_x_point *coords = coordinates_x;
+ uint32_t i = 0;
+ const struct pwl_float_data_ex *rgb_regamma = rgb_ex;
+
+ while (i <= hw_points_num) {
+ coords->regamma_y_red = rgb_regamma->r;
+ coords->regamma_y_green = rgb_regamma->g;
+ coords->regamma_y_blue = rgb_regamma->b;
+
+ ++coords;
+ ++rgb_regamma;
+ ++i;
+ }
+}
+
+static bool calculate_interpolated_hardware_curve(
+ const struct dc_gamma *ramp,
+ struct pixel_gamma_point *coeff128,
+ struct pwl_float_data *rgb_user,
+ const struct hw_x_point *coordinates_x,
+ const struct gamma_pixel *axis_x,
+ uint32_t number_of_points,
+ struct dc_transfer_func_distributed_points *tf_pts)
+{
+
+ const struct pixel_gamma_point *coeff = coeff128;
+ uint32_t max_entries = 3 - 1;
+
+ uint32_t i = 0;
+
+ for (i = 0; i < 3; i++) {
+ if (!build_custom_gamma_mapping_coefficients_worker(
+ ramp, coeff128, coordinates_x, axis_x, i,
+ number_of_points))
+ return false;
+ }
+
+ i = 0;
+ max_entries += ramp->num_entries;
+
+ /* TODO: float point case */
+
+ while (i <= number_of_points) {
+ tf_pts->red[i] = calculate_mapped_value(
+ rgb_user, coeff, CHANNEL_NAME_RED, max_entries);
+ tf_pts->green[i] = calculate_mapped_value(
+ rgb_user, coeff, CHANNEL_NAME_GREEN, max_entries);
+ tf_pts->blue[i] = calculate_mapped_value(
+ rgb_user, coeff, CHANNEL_NAME_BLUE, max_entries);
+
+ ++coeff;
+ ++i;
+ }
+
+ return true;
+}
+
+static void build_new_custom_resulted_curve(
+ uint32_t hw_points_num,
+ struct dc_transfer_func_distributed_points *tf_pts)
+{
+ uint32_t i;
+
+ i = 0;
+
+ while (i != hw_points_num + 1) {
+ tf_pts->red[i] = dal_fixed31_32_clamp(
+ tf_pts->red[i], dal_fixed31_32_zero,
+ dal_fixed31_32_one);
+ tf_pts->green[i] = dal_fixed31_32_clamp(
+ tf_pts->green[i], dal_fixed31_32_zero,
+ dal_fixed31_32_one);
+ tf_pts->blue[i] = dal_fixed31_32_clamp(
+ tf_pts->blue[i], dal_fixed31_32_zero,
+ dal_fixed31_32_one);
+
+ ++i;
+ }
+}
+
+static bool map_regamma_hw_to_x_user(
+ const struct dc_gamma *ramp,
+ struct pixel_gamma_point *coeff128,
+ struct pwl_float_data *rgb_user,
+ struct hw_x_point *coords_x,
+ const struct gamma_pixel *axis_x,
+ const struct pwl_float_data_ex *rgb_regamma,
+ uint32_t hw_points_num,
+ struct dc_transfer_func_distributed_points *tf_pts,
+ bool mapUserRamp)
+{
+ /* setup to spare calculated ideal regamma values */
+
+ int i = 0;
+ struct hw_x_point *coords = coords_x;
+ const struct pwl_float_data_ex *regamma = rgb_regamma;
+
+ if (mapUserRamp) {
+ copy_rgb_regamma_to_coordinates_x(coords,
+ hw_points_num,
+ rgb_regamma);
+
+ calculate_interpolated_hardware_curve(
+ ramp, coeff128, rgb_user, coords, axis_x,
+ hw_points_num, tf_pts);
+ } else {
+ /* just copy current rgb_regamma into tf_pts */
+ while (i <= hw_points_num) {
+ tf_pts->red[i] = regamma->r;
+ tf_pts->green[i] = regamma->g;
+ tf_pts->blue[i] = regamma->b;
+
+ ++regamma;
+ ++i;
+ }
+ }
+
+ build_new_custom_resulted_curve(hw_points_num, tf_pts);
+
+ return true;
+}
+
+bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
+ const struct dc_gamma *ramp, bool mapUserRamp)
+{
+ struct dc_transfer_func_distributed_points *tf_pts = &output_tf->tf_pts;
+ struct dividers dividers;
+
+ struct pwl_float_data *rgb_user = NULL;
+ struct pwl_float_data_ex *rgb_regamma = NULL;
+ struct gamma_pixel *axix_x = NULL;
+ struct pixel_gamma_point *coeff128 = NULL;
+ enum dc_transfer_func_predefined tf = TRANSFER_FUNCTION_SRGB;
+ bool ret = false;
+
+ if (output_tf->type == TF_TYPE_BYPASS)
+ return false;
+
+ /* we can use hardcoded curve for plain SRGB TF */
+ if (output_tf->type == TF_TYPE_PREDEFINED &&
+ output_tf->tf == TRANSFER_FUNCTION_SRGB &&
+ (!mapUserRamp && ramp->type == GAMMA_RGB_256))
+ return true;
+
+ output_tf->type = TF_TYPE_DISTRIBUTED_POINTS;
+
+ rgb_user = kzalloc(sizeof(*rgb_user) * (ramp->num_entries + 3),
+ GFP_KERNEL);
+ if (!rgb_user)
+ goto rgb_user_alloc_fail;
+ rgb_regamma = kzalloc(sizeof(*rgb_regamma) * (MAX_HW_POINTS + 3),
+ GFP_KERNEL);
+ if (!rgb_regamma)
+ goto rgb_regamma_alloc_fail;
+ axix_x = kzalloc(sizeof(*axix_x) * (ramp->num_entries + 3),
+ GFP_KERNEL);
+ if (!axix_x)
+ goto axix_x_alloc_fail;
+ coeff128 = kzalloc(sizeof(*coeff128) * (MAX_HW_POINTS + 3), GFP_KERNEL);
+ if (!coeff128)
+ goto coeff128_alloc_fail;
+
+ dividers.divider1 = dal_fixed31_32_from_fraction(3, 2);
+ dividers.divider2 = dal_fixed31_32_from_int(2);
+ dividers.divider3 = dal_fixed31_32_from_fraction(5, 2);
+
+ tf = output_tf->tf;
+
+ build_evenly_distributed_points(
+ axix_x,
+ ramp->num_entries,
+ dividers);
+
+ if (ramp->type == GAMMA_RGB_256 && mapUserRamp)
+ scale_gamma(rgb_user, ramp, dividers);
+ else if (ramp->type == GAMMA_RGB_FLOAT_1024)
+ scale_gamma_dx(rgb_user, ramp, dividers);
+
+ if (tf == TRANSFER_FUNCTION_PQ) {
+ tf_pts->end_exponent = 7;
+ tf_pts->x_point_at_y1_red = 125;
+ tf_pts->x_point_at_y1_green = 125;
+ tf_pts->x_point_at_y1_blue = 125;
+
+ build_regamma_curve_pq(rgb_regamma,
+ MAX_HW_POINTS,
+ coordinates_x,
+ output_tf->sdr_ref_white_level);
+ } else {
+ tf_pts->end_exponent = 0;
+ tf_pts->x_point_at_y1_red = 1;
+ tf_pts->x_point_at_y1_green = 1;
+ tf_pts->x_point_at_y1_blue = 1;
+
+ build_regamma_curve(rgb_regamma,
+ MAX_HW_POINTS,
+ coordinates_x);
+ }
+
+ map_regamma_hw_to_x_user(ramp, coeff128, rgb_user,
+ coordinates_x, axix_x, rgb_regamma,
+ MAX_HW_POINTS, tf_pts,
+ (mapUserRamp || ramp->type != GAMMA_RGB_256) &&
+ ramp->type != GAMMA_CS_TFM_1D);
+
+ if (ramp->type == GAMMA_CS_TFM_1D)
+ apply_lut_1d(ramp, MAX_HW_POINTS, tf_pts);
+
+ ret = true;
+
+ kfree(coeff128);
+coeff128_alloc_fail:
+ kfree(axix_x);
+axix_x_alloc_fail:
+ kfree(rgb_regamma);
+rgb_regamma_alloc_fail:
+ kfree(rgb_user);
+rgb_user_alloc_fail:
+ return ret;
+}
+
+
+/*TODO fix me should be 2*/
+#define _EXTRA_POINTS 3
+
+bool mod_color_calculate_curve(enum dc_transfer_func_predefined trans,
+ struct dc_transfer_func_distributed_points *points)
+{
+ uint32_t i;
+ bool ret = false;
+ struct pwl_float_data_ex *rgb_regamma = NULL;
+
+ if (trans == TRANSFER_FUNCTION_UNITY) {
+ //setup_x_points_distribution(coordinates_x);
+ for (i = 0; i < MAX_HW_POINTS ; i++) {
+ points->red[i] = coordinates_x[i].x;
+ points->green[i] = coordinates_x[i].x;
+ points->blue[i] = coordinates_x[i].x;
+ }
+ ret = true;
+ } else if (trans == TRANSFER_FUNCTION_PQ) {
+ rgb_regamma = kzalloc(sizeof(*rgb_regamma) * (MAX_HW_POINTS +
+ _EXTRA_POINTS), GFP_KERNEL);
+ if (!rgb_regamma)
+ goto rgb_regamma_alloc_fail;
+ //setup_x_points_distribution(coordinates_x);
+ points->end_exponent = 0;
+ points->x_point_at_y1_red = 1;
+ points->x_point_at_y1_green = 1;
+ points->x_point_at_y1_blue = 1;
+
+ build_regamma_curve_pq(rgb_regamma,
+ MAX_HW_POINTS,
+ coordinates_x,
+ 80);
+ for (i = 0; i < MAX_HW_POINTS ; i++) {
+ points->red[i] = rgb_regamma[i].r;
+ points->green[i] = rgb_regamma[i].g;
+ points->blue[i] = rgb_regamma[i].b;
+ }
+ ret = true;
+
+ kfree(rgb_regamma);
+ }
+rgb_regamma_alloc_fail:
+ return ret;
+}
+
+
diff --git a/drivers/gpu/drm/amd/display/modules/color/color_gamma.h b/drivers/gpu/drm/amd/display/modules/color/color_gamma.h
new file mode 100644
index 000000000000..774c6daa1689
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/modules/color/color_gamma.h
@@ -0,0 +1,45 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef COLOR_MOD_COLOR_GAMMA_H_
+#define COLOR_MOD_COLOR_GAMMA_H_
+
+struct dc_transfer_func;
+struct dc_gamma;
+struct dc_transfer_func_distributed_points;
+struct dc_rgb_fixed;
+enum dc_transfer_func_predefined;
+
+void setup_x_points_distribution(void);
+void precompute_pq(void);
+
+bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
+ const struct dc_gamma *ramp, bool mapUserRamp);
+
+bool mod_color_calculate_curve(enum dc_transfer_func_predefined trans,
+ struct dc_transfer_func_distributed_points *points);
+
+
+#endif /* COLOR_MOD_COLOR_GAMMA_H_ */
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 16/34] drm/amd/display: Implement color management
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (14 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 15/34] drm/amd/display: Add color module's gamma helpers to Linux build Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 17/34] drm/amd/display: Hookup color management functions Harry Wentland
` (17 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
Implement color management functionalities within amdgpu_dm_color, and
expose functions within amdgpu_dm.h.
Change-Id: Ib1b85a42b4f535becf67b5b3f386e81ebf0bcc33
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
---
drivers/gpu/drm/amd/display/amdgpu_dm/Makefile | 2 +-
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h | 5 +
.../drm/amd/display/amdgpu_dm/amdgpu_dm_color.c | 218 +++++++++++++++++++++
3 files changed, 224 insertions(+), 1 deletion(-)
create mode 100644 drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile b/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile
index d7accc2071c4..af16973f2c41 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile
@@ -25,7 +25,7 @@
-AMDGPUDM = amdgpu_dm.o amdgpu_dm_irq.o amdgpu_dm_mst_types.o
+AMDGPUDM = amdgpu_dm.o amdgpu_dm_irq.o amdgpu_dm_mst_types.o amdgpu_dm_color.o
ifneq ($(CONFIG_DRM_AMD_DC),)
AMDGPUDM += amdgpu_dm_services.o amdgpu_dm_helpers.o
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
index 124203673207..9c556d85dd0e 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
@@ -269,6 +269,11 @@ void amdgpu_dm_crtc_handle_crc_irq(struct drm_crtc *crtc);
#define amdgpu_dm_crtc_handle_crc_irq(x)
#endif
+int amdgpu_dm_set_degamma_lut(struct drm_crtc_state *crtc_state,
+ struct dc_plane_state *dc_plane_state);
+void amdgpu_dm_set_ctm(struct dm_crtc_state *crtc);
+int amdgpu_dm_set_regamma_lut(struct dm_crtc_state *crtc);
+
extern const struct drm_encoder_helper_funcs amdgpu_dm_encoder_helper_funcs;
#endif /* __AMDGPU_DM_H__ */
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
new file mode 100644
index 000000000000..cc3ee0748a70
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
@@ -0,0 +1,218 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include "amdgpu_mode.h"
+#include "amdgpu_dm.h"
+#include "modules/color/color_gamma.h"
+
+
+#define MAX_LUT_ENTRIES 256
+
+/*
+ * Return true if the given lut is a linear mapping of values, i.e. it acts
+ * like a bypass LUT.
+ *
+ * It is considered linear if the lut represents:
+ * f(a) = (0xFF00/MAX_LUT_ENTRIES-1)a; for integer a in [0, MAX_LUT_ENTRIES)
+ */
+static bool __is_lut_linear(struct drm_color_lut *lut)
+{
+ int i;
+ uint32_t max_os = 0xFF00;
+ uint32_t expected;
+ int delta;
+
+ for (i = 0; i < MAX_LUT_ENTRIES; i++) {
+ /* All color values should equal */
+ if ((lut[i].red != lut[i].green) || (lut[i].green != lut[i].blue))
+ return false;
+
+ expected = i * max_os / (MAX_LUT_ENTRIES-1);
+
+ /* Allow a +/-1 error. */
+ delta = lut[i].red - expected;
+ if (delta < -1 || 1 < delta)
+ return false;
+ }
+ return true;
+}
+
+/**
+ * amdgpu_dm_set_regamma_lut: Set regamma lut for the given CRTC.
+ * @crtc: amdgpu_dm crtc state
+ *
+ * Update the underlying dc_stream_state's output transfer function (OTF) in
+ * preparation for hardware commit. If no lut is specified by user, we default
+ * to SRGB.
+ *
+ * RETURNS:
+ * 0 on success, -ENOMEM if memory cannot be allocated to calculate the OTF.
+ */
+int amdgpu_dm_set_regamma_lut(struct dm_crtc_state *crtc)
+{
+ struct drm_property_blob *blob = crtc->base.gamma_lut;
+ struct dc_stream_state *stream = crtc->stream;
+ struct drm_color_lut *lut;
+ struct dc_gamma *gamma;
+ enum dc_transfer_func_type old_type = stream->out_transfer_func->type;
+
+ uint32_t r, g, b;
+ int i;
+ bool ret;
+
+ if (!blob) {
+ /* By default, use the SRGB predefined curve.*/
+ stream->out_transfer_func->type = TF_TYPE_PREDEFINED;
+ stream->out_transfer_func->tf = TRANSFER_FUNCTION_SRGB;
+ return 0;
+ }
+
+ lut = (struct drm_color_lut *)blob->data;
+
+ if (__is_lut_linear(lut)) {
+ /* Set to bypass if lut is set to linear */
+ stream->out_transfer_func->type = TF_TYPE_BYPASS;
+ stream->out_transfer_func->tf = TRANSFER_FUNCTION_LINEAR;
+ return 0;
+ }
+
+ gamma = dc_create_gamma();
+ if (!gamma)
+ return -ENOMEM;
+
+ gamma->num_entries = MAX_LUT_ENTRIES;
+ gamma->type = GAMMA_RGB_256;
+
+ /* Truncate, and store in dc_gamma for output tf calculation */
+ for (i = 0; i < gamma->num_entries; i++) {
+ r = drm_color_lut_extract(lut[i].red, 16);
+ g = drm_color_lut_extract(lut[i].green, 16);
+ b = drm_color_lut_extract(lut[i].blue, 16);
+
+ gamma->entries.red[i] = dal_fixed31_32_from_int(r);
+ gamma->entries.green[i] = dal_fixed31_32_from_int(g);
+ gamma->entries.blue[i] = dal_fixed31_32_from_int(b);
+ }
+
+ /* Call color module to translate into something DC understands. Namely
+ * a transfer function.
+ */
+ stream->out_transfer_func->type = TF_TYPE_DISTRIBUTED_POINTS;
+ ret = mod_color_calculate_regamma_params(stream->out_transfer_func,
+ gamma, true);
+ dc_gamma_release(&gamma);
+ if (!ret) {
+ stream->out_transfer_func->type = old_type;
+ DRM_ERROR("Out of memory when calculating regamma params\n");
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+/**
+ * amdgpu_dm_set_ctm: Set the color transform matrix for the given CRTC.
+ * @crtc: amdgpu_dm crtc state
+ *
+ * Update the underlying dc_stream_state's gamut remap matrix in preparation
+ * for hardware commit. If no matrix is specified by user, gamut remap will be
+ * disabled.
+ */
+void amdgpu_dm_set_ctm(struct dm_crtc_state *crtc)
+{
+
+ struct drm_property_blob *blob = crtc->base.ctm;
+ struct dc_stream_state *stream = crtc->stream;
+ struct drm_color_ctm *ctm;
+ int i;
+
+ if (!blob) {
+ stream->gamut_remap_matrix.enable_remap = false;
+ return;
+ }
+
+ stream->gamut_remap_matrix.enable_remap = true;
+ ctm = (struct drm_color_ctm *)blob->data;
+ /*
+ * DRM gives a 3x3 matrix, but DC wants 3x4. Assuming we're operating
+ * with homogeneous coordinates, augment the matrix with 0's.
+ *
+ * The format provided is S31.32, which is the same as our fixed31_32.
+ */
+ for (i = 0; i < 12; i++) {
+ /* Skip 4th element */
+ if (i % 4 == 3) {
+ stream->gamut_remap_matrix.matrix[i] = dal_fixed31_32_zero;
+ continue;
+ }
+ /* csc[i] = ctm[i - floor(i/4)] */
+ stream->gamut_remap_matrix.matrix[i].value = ctm->matrix[i - (i/4)];
+ }
+}
+
+
+/**
+ * amdgpu_dm_set_degamma_lut: Set degamma lut for the given CRTC.
+ * @crtc: amdgpu_dm crtc state
+ *
+ * Update the underlying dc_stream_state's input transfer function (ITF) in
+ * preparation for hardware commit. If no lut is specified by user, we default
+ * to SRGB degamma.
+ *
+ * Currently, we only support degamma bypass, or preprogrammed SRGB degamma.
+ * Programmable degamma is not supported, and an attempt to do so will return
+ * -EINVAL.
+ *
+ * RETURNS:
+ * 0 on success, -EINVAL if custom degamma curve is given.
+ */
+int amdgpu_dm_set_degamma_lut(struct drm_crtc_state *crtc_state,
+ struct dc_plane_state *dc_plane_state)
+{
+ struct drm_property_blob *blob = crtc_state->degamma_lut;
+ struct drm_color_lut *lut;
+
+ if (!blob) {
+ /* Default to SRGB */
+ dc_plane_state->in_transfer_func->type = TF_TYPE_PREDEFINED;
+ dc_plane_state->in_transfer_func->tf = TRANSFER_FUNCTION_SRGB;
+ return 0;
+ }
+
+ lut = (struct drm_color_lut *)blob->data;
+ if (__is_lut_linear(lut)) {
+ dc_plane_state->in_transfer_func->type = TF_TYPE_BYPASS;
+ dc_plane_state->in_transfer_func->tf = TRANSFER_FUNCTION_LINEAR;
+ return 0;
+ }
+
+ /* Otherwise, assume SRGB, since programmable degamma is not
+ * supported.
+ */
+ dc_plane_state->in_transfer_func->type = TF_TYPE_PREDEFINED;
+ dc_plane_state->in_transfer_func->tf = TRANSFER_FUNCTION_SRGB;
+ return -EINVAL;
+}
+
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 17/34] drm/amd/display: Hookup color management functions
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (15 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 16/34] drm/amd/display: Implement color management Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 18/34] drm/amd/display: Refactor max color lut entries into a macro Harry Wentland
` (16 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
Hookup new color management functions into amdgpu_dm:
- Notify DRM that we support CRTC color management during CRTC init
- Call color management functions within atomic check to update dc
states in preparation for a commit
Change-Id: I7246a679479c4efbf31fe0d631559b06f64c36ff
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
---
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 65 +++++++++++-----------
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h | 1 +
.../drm/amd/display/amdgpu_dm/amdgpu_dm_color.c | 11 ++++
3 files changed, 44 insertions(+), 33 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index e285147f32d3..9212964a1246 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -474,6 +474,8 @@ static int amdgpu_dm_init(struct amdgpu_device *adev)
DRM_DEBUG_DRIVER("amdgpu: freesync_module init done %p.\n",
adev->dm.freesync_module);
+ amdgpu_dm_init_color_mod();
+
if (amdgpu_dm_initialize_drm_device(adev)) {
DRM_ERROR(
"amdgpu: failed to initialize sw for display support.\n");
@@ -1958,32 +1960,6 @@ static int fill_plane_attributes_from_fb(struct amdgpu_device *adev,
}
-static void fill_gamma_from_crtc_state(const struct drm_crtc_state *crtc_state,
- struct dc_plane_state *plane_state)
-{
- int i;
- struct dc_gamma *gamma;
- struct drm_color_lut *lut =
- (struct drm_color_lut *) crtc_state->gamma_lut->data;
-
- gamma = dc_create_gamma();
-
- if (gamma == NULL) {
- WARN_ON(1);
- return;
- }
-
- gamma->type = GAMMA_RGB_256;
- gamma->num_entries = GAMMA_RGB_256_ENTRIES;
- for (i = 0; i < GAMMA_RGB_256_ENTRIES; i++) {
- gamma->entries.red[i] = dal_fixed31_32_from_int(lut[i].red);
- gamma->entries.green[i] = dal_fixed31_32_from_int(lut[i].green);
- gamma->entries.blue[i] = dal_fixed31_32_from_int(lut[i].blue);
- }
-
- plane_state->gamma_correction = gamma;
-}
-
static int fill_plane_attributes(struct amdgpu_device *adev,
struct dc_plane_state *dc_plane_state,
struct drm_plane_state *plane_state,
@@ -2011,14 +1987,13 @@ static int fill_plane_attributes(struct amdgpu_device *adev,
if (input_tf == NULL)
return -ENOMEM;
- input_tf->type = TF_TYPE_PREDEFINED;
- input_tf->tf = TRANSFER_FUNCTION_SRGB;
-
dc_plane_state->in_transfer_func = input_tf;
- /* In case of gamma set, update gamma value */
- if (crtc_state->gamma_lut)
- fill_gamma_from_crtc_state(crtc_state, dc_plane_state);
+ /*
+ * Always set input transfer function, since plane state is refreshed
+ * every time.
+ */
+ ret = amdgpu_dm_set_degamma_lut(crtc_state, dc_plane_state);
return ret;
}
@@ -3258,6 +3233,7 @@ static int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm,
acrtc->base.enabled = false;
dm->adev->mode_info.crtcs[crtc_index] = acrtc;
+ drm_crtc_enable_color_mgmt(&acrtc->base, 256, true, 256);
drm_mode_crtc_set_gamma_size(&acrtc->base, 256);
return 0;
@@ -4679,6 +4655,30 @@ static int dm_update_crtcs_state(struct dc *dc,
/* Release extra reference */
if (new_stream)
dc_stream_release(new_stream);
+
+ /*
+ * We want to do dc stream updates that do not require a
+ * full modeset below.
+ */
+ if (!enable || !aconnector || modereset_required(new_crtc_state))
+ continue;
+ /*
+ * Given above conditions, the dc state cannot be NULL because:
+ * 1. We're attempting to enable a CRTC. Which has a...
+ * 2. Valid connector attached, and
+ * 3. User does not want to reset it (disable or mark inactive,
+ * which can happen on a CRTC that's already disabled).
+ * => It currently exists.
+ */
+ BUG_ON(dm_new_crtc_state->stream == NULL);
+
+ /* Color managment settings */
+ if (dm_new_crtc_state->base.color_mgmt_changed) {
+ ret = amdgpu_dm_set_regamma_lut(dm_new_crtc_state);
+ if (ret)
+ goto fail;
+ amdgpu_dm_set_ctm(dm_new_crtc_state);
+ }
}
return ret;
@@ -4787,7 +4787,6 @@ static int dm_update_planes_state(struct dc *dc,
if (ret)
return ret;
-
if (!dc_add_plane_to_context(
dc,
dm_new_crtc_state->stream,
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
index 9c556d85dd0e..7e5d5cd40680 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
@@ -269,6 +269,7 @@ void amdgpu_dm_crtc_handle_crc_irq(struct drm_crtc *crtc);
#define amdgpu_dm_crtc_handle_crc_irq(x)
#endif
+void amdgpu_dm_init_color_mod(void);
int amdgpu_dm_set_degamma_lut(struct drm_crtc_state *crtc_state,
struct dc_plane_state *dc_plane_state);
void amdgpu_dm_set_ctm(struct dm_crtc_state *crtc);
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
index cc3ee0748a70..d7bc1b7dd152 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
@@ -27,6 +27,17 @@
#include "amdgpu_dm.h"
#include "modules/color/color_gamma.h"
+/*
+ * Initialize the color module.
+ *
+ * We're not using the full color module, only certain components.
+ * Only call setup functions for components that we need.
+ */
+void amdgpu_dm_init_color_mod(void)
+{
+ setup_x_points_distribution();
+}
+
#define MAX_LUT_ENTRIES 256
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 18/34] drm/amd/display: Refactor max color lut entries into a macro.
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (16 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 17/34] drm/amd/display: Hookup color management functions Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 19/34] drm/amd/display: Expose dither setting functionality to Linux Harry Wentland
` (15 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
Change-Id: If0205a3552cfd74a06511f595e5d1b6be5b655e8
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
---
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 5 +++--
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h | 2 ++
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c | 11 +++++------
3 files changed, 10 insertions(+), 8 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index 9212964a1246..b632be5203b8 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -3233,8 +3233,9 @@ static int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm,
acrtc->base.enabled = false;
dm->adev->mode_info.crtcs[crtc_index] = acrtc;
- drm_crtc_enable_color_mgmt(&acrtc->base, 256, true, 256);
- drm_mode_crtc_set_gamma_size(&acrtc->base, 256);
+ drm_crtc_enable_color_mgmt(&acrtc->base, MAX_COLOR_LUT_ENTRIES,
+ true, MAX_COLOR_LUT_ENTRIES);
+ drm_mode_crtc_set_gamma_size(&acrtc->base, MAX_COLOR_LUT_ENTRIES);
return 0;
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
index 7e5d5cd40680..c40c13a9f614 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
@@ -269,6 +269,8 @@ void amdgpu_dm_crtc_handle_crc_irq(struct drm_crtc *crtc);
#define amdgpu_dm_crtc_handle_crc_irq(x)
#endif
+#define MAX_COLOR_LUT_ENTRIES 256
+
void amdgpu_dm_init_color_mod(void);
int amdgpu_dm_set_degamma_lut(struct drm_crtc_state *crtc_state,
struct dc_plane_state *dc_plane_state);
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
index d7bc1b7dd152..62bb72fe9aa5 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
@@ -39,14 +39,13 @@ void amdgpu_dm_init_color_mod(void)
}
-#define MAX_LUT_ENTRIES 256
-
/*
* Return true if the given lut is a linear mapping of values, i.e. it acts
* like a bypass LUT.
*
* It is considered linear if the lut represents:
- * f(a) = (0xFF00/MAX_LUT_ENTRIES-1)a; for integer a in [0, MAX_LUT_ENTRIES)
+ * f(a) = (0xFF00/MAX_COLOR_LUT_ENTRIES-1)a; for integer a in
+ * [0, MAX_COLOR_LUT_ENTRIES)
*/
static bool __is_lut_linear(struct drm_color_lut *lut)
{
@@ -55,12 +54,12 @@ static bool __is_lut_linear(struct drm_color_lut *lut)
uint32_t expected;
int delta;
- for (i = 0; i < MAX_LUT_ENTRIES; i++) {
+ for (i = 0; i < MAX_COLOR_LUT_ENTRIES; i++) {
/* All color values should equal */
if ((lut[i].red != lut[i].green) || (lut[i].green != lut[i].blue))
return false;
- expected = i * max_os / (MAX_LUT_ENTRIES-1);
+ expected = i * max_os / (MAX_COLOR_LUT_ENTRIES-1);
/* Allow a +/-1 error. */
delta = lut[i].red - expected;
@@ -113,7 +112,7 @@ int amdgpu_dm_set_regamma_lut(struct dm_crtc_state *crtc)
if (!gamma)
return -ENOMEM;
- gamma->num_entries = MAX_LUT_ENTRIES;
+ gamma->num_entries = MAX_COLOR_LUT_ENTRIES;
gamma->type = GAMMA_RGB_256;
/* Truncate, and store in dc_gamma for output tf calculation */
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 19/34] drm/amd/display: Expose dither setting functionality to Linux
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (17 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 18/34] drm/amd/display: Refactor max color lut entries into a macro Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 20/34] drm/amd/display: When enabling CRC, disable dither & enable truncation Harry Wentland
` (14 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
We will need this to disable dither for CRC capture.
Change-Id: I3ae4fcd86c74b5ab05ad16559d92e156c6f389bc
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc.c | 31 ++++++++++++++++++++++++++++++
drivers/gpu/drm/amd/display/dc/dc_stream.h | 3 +++
2 files changed, 34 insertions(+)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c
index a78e0364b1b2..ea5c35166ffa 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
@@ -300,6 +300,37 @@ bool dc_stream_get_crc(struct dc *dc, struct dc_stream_state *stream,
return false;
}
+void dc_stream_set_dither_option(struct dc_stream_state *stream,
+ enum dc_dither_option option)
+{
+ struct bit_depth_reduction_params params;
+ struct dc_link *link = stream->status.link;
+ struct pipe_ctx *pipes = NULL;
+ int i;
+
+ for (i = 0; i < MAX_PIPES; i++) {
+ if (link->dc->current_state->res_ctx.pipe_ctx[i].stream ==
+ stream) {
+ pipes = &link->dc->current_state->res_ctx.pipe_ctx[i];
+ break;
+ }
+ }
+
+ memset(¶ms, 0, sizeof(params));
+ if (!pipes)
+ return;
+ if (option > DITHER_OPTION_MAX)
+ return;
+
+ stream->dither_option = option;
+
+ resource_build_bit_depth_reduction_params(stream,
+ ¶ms);
+ stream->bit_depth_params = params;
+ pipes->stream_res.opp->funcs->
+ opp_program_bit_depth_reduction(pipes->stream_res.opp, ¶ms);
+}
+
void dc_stream_set_static_screen_events(struct dc *dc,
struct dc_stream_state **streams,
int num_streams,
diff --git a/drivers/gpu/drm/amd/display/dc/dc_stream.h b/drivers/gpu/drm/amd/display/dc/dc_stream.h
index 2d2472ba97bf..78a2bbe0b272 100644
--- a/drivers/gpu/drm/amd/display/dc/dc_stream.h
+++ b/drivers/gpu/drm/amd/display/dc/dc_stream.h
@@ -287,6 +287,9 @@ void dc_stream_set_static_screen_events(struct dc *dc,
int num_streams,
const struct dc_static_screen_events *events);
+void dc_stream_set_dither_option(struct dc_stream_state *stream,
+ enum dc_dither_option option);
+
bool dc_stream_adjust_vmin_vmax(struct dc *dc,
struct dc_stream_state **stream,
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 20/34] drm/amd/display: When enabling CRC, disable dither & enable truncation
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (18 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 19/34] drm/amd/display: Expose dither setting functionality to Linux Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 21/34] drm/amd/display: Rework DCE transform bit depth reduction programming Harry Wentland
` (13 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
When user-mode is using 8bpc, the hardware represents it internally
using a higher bit depth. This causes problems when comparing CRCs for
color managment tests.
We need to disable dithering as well, since it makes CRC values
non-deterministic.
It's easy to see why dithering needs to be disabled, The reason why
truncation also needs to be enabled is better described with an example.
Consider the folowing which tests the color transform matrix (CTM):
Expected CRC = FB_A -> Degamma (Bypassed) -> CTM (Bypassed)
|
v
Obtain CRC <- Regamma(Bypassed)
Actual CRC = FB_B -> Degamma (Bypassed) -> CTM (0.5*Identity)
|
v
Obtain CRC <- Regamma(Bypassed)
FB_A contains a solid red color at half intensity (127 @ 8bpc)
FB_B contains a solid red color at full intensity (255 @ 8bpc)
We expect that Expected CRC = Actual CRC, but that's not the case. When
the CTM is applied, the output is at half intensity, but also at a
higher bit depth within hardware. i.e. 255/2 = 127.5: not representable
in 8bpc, but can be at 10bpc. This causes the two CRC's to be different.
The solution is to truncate the output bit depth to the same as input
when enabling CRC capture. Since Linux only supports 8bpc, hard code
that for now.
Change-Id: Ic9277659ec785d655989ba0c6d170b1691525376
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
---
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c | 11 +++++++++--
drivers/gpu/drm/amd/display/dc/core/dc.c | 14 +++++++++++---
2 files changed, 20 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
index e5fb53a56b73..55aa379cfcbe 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
@@ -60,18 +60,25 @@ int amdgpu_dm_crtc_set_crc_source(struct drm_crtc *crtc, const char *src_name,
return -EINVAL;
}
+ /* When enabling CRC, we should also disable dithering. */
if (source == AMDGPU_DM_PIPE_CRC_SOURCE_AUTO) {
if (dc_stream_configure_crc(stream_state->ctx->dc,
stream_state,
- true, true))
+ true, true)) {
crtc_state->crc_enabled = true;
+ dc_stream_set_dither_option(stream_state,
+ DITHER_OPTION_TRUN8);
+ }
else
return -EINVAL;
} else {
if (dc_stream_configure_crc(stream_state->ctx->dc,
stream_state,
- false, false))
+ false, false)) {
crtc_state->crc_enabled = false;
+ dc_stream_set_dither_option(stream_state,
+ DITHER_OPTION_DEFAULT);
+ }
else
return -EINVAL;
}
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c
index ea5c35166ffa..f4ffbf2e2caf 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
@@ -316,7 +316,6 @@ void dc_stream_set_dither_option(struct dc_stream_state *stream,
}
}
- memset(¶ms, 0, sizeof(params));
if (!pipes)
return;
if (option > DITHER_OPTION_MAX)
@@ -324,9 +323,18 @@ void dc_stream_set_dither_option(struct dc_stream_state *stream,
stream->dither_option = option;
- resource_build_bit_depth_reduction_params(stream,
- ¶ms);
+ memset(¶ms, 0, sizeof(params));
+ resource_build_bit_depth_reduction_params(stream, ¶ms);
stream->bit_depth_params = params;
+
+ if (pipes->plane_res.xfm &&
+ pipes->plane_res.xfm->funcs->transform_set_pixel_storage_depth) {
+ pipes->plane_res.xfm->funcs->transform_set_pixel_storage_depth(
+ pipes->plane_res.xfm,
+ pipes->plane_res.scl_data.lb_params.depth,
+ &stream->bit_depth_params);
+ }
+
pipes->stream_res.opp->funcs->
opp_program_bit_depth_reduction(pipes->stream_res.opp, ¶ms);
}
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 21/34] drm/amd/display: Rework DCE transform bit depth reduction programming.
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (19 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 20/34] drm/amd/display: When enabling CRC, disable dither & enable truncation Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 22/34] drm/amd/display: Expose DCE110 CRC functions for DCE8 Harry Wentland
` (12 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
Clear up the logic, and enable programming truncation as a bit reduction
mode.
Change-Id: I72199753222a3f77fb9986a6e36a8bfb1f793e36
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
---
drivers/gpu/drm/amd/display/dc/dce/dce_transform.c | 104 +++++++--------------
1 file changed, 36 insertions(+), 68 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c b/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c
index 6357546b49cd..ad411dac5639 100644
--- a/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c
@@ -618,80 +618,48 @@ static void program_bit_depth_reduction(
enum dc_color_depth depth,
const struct bit_depth_reduction_params *bit_depth_params)
{
- enum dcp_bit_depth_reduction_mode depth_reduction_mode;
- enum dcp_spatial_dither_mode spatial_dither_mode;
- bool frame_random_enable;
- bool rgb_random_enable;
- bool highpass_random_enable;
+ enum dcp_out_trunc_round_depth trunc_round_depth;
+ enum dcp_out_trunc_round_mode trunc_mode;
+ bool spatial_dither_enable;
ASSERT(depth < COLOR_DEPTH_121212); /* Invalid clamp bit depth */
- if (bit_depth_params->flags.SPATIAL_DITHER_ENABLED) {
- depth_reduction_mode = DCP_BIT_DEPTH_REDUCTION_MODE_DITHER;
- frame_random_enable = true;
- rgb_random_enable = true;
- highpass_random_enable = true;
-
- } else {
- depth_reduction_mode = DCP_BIT_DEPTH_REDUCTION_MODE_DISABLED;
- frame_random_enable = false;
- rgb_random_enable = false;
- highpass_random_enable = false;
+ spatial_dither_enable = bit_depth_params->flags.SPATIAL_DITHER_ENABLED;
+ /* Default to 12 bit truncation without rounding */
+ trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_12BIT;
+ trunc_mode = DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE;
+
+ if (bit_depth_params->flags.TRUNCATE_ENABLED) {
+ /* Don't enable dithering if truncation is enabled */
+ spatial_dither_enable = false;
+ trunc_mode = bit_depth_params->flags.TRUNCATE_MODE ?
+ DCP_OUT_TRUNC_ROUND_MODE_ROUND :
+ DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE;
+
+ if (bit_depth_params->flags.TRUNCATE_DEPTH == 0 ||
+ bit_depth_params->flags.TRUNCATE_DEPTH == 1)
+ trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_8BIT;
+ else if (bit_depth_params->flags.TRUNCATE_DEPTH == 2)
+ trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_10BIT;
+ else {
+ /*
+ * Invalid truncate/round depth. Setting here to 12bit
+ * to prevent use-before-initialize errors.
+ */
+ trunc_round_depth = DCP_OUT_TRUNC_ROUND_DEPTH_12BIT;
+ BREAK_TO_DEBUGGER();
+ }
}
- spatial_dither_mode = DCP_SPATIAL_DITHER_MODE_A_AA_A;
-
set_clamp(xfm_dce, depth);
-
- switch (depth_reduction_mode) {
- case DCP_BIT_DEPTH_REDUCTION_MODE_DITHER:
- /* Spatial Dither: Set round/truncate to bypass (12bit),
- * enable Dither (30bpp) */
- set_round(xfm_dce,
- DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE,
- DCP_OUT_TRUNC_ROUND_DEPTH_12BIT);
-
- set_dither(xfm_dce, true, spatial_dither_mode,
- DCP_SPATIAL_DITHER_DEPTH_30BPP, frame_random_enable,
- rgb_random_enable, highpass_random_enable);
- break;
- case DCP_BIT_DEPTH_REDUCTION_MODE_ROUND:
- /* Round: Enable round (10bit), disable Dither */
- set_round(xfm_dce,
- DCP_OUT_TRUNC_ROUND_MODE_ROUND,
- DCP_OUT_TRUNC_ROUND_DEPTH_10BIT);
-
- set_dither(xfm_dce, false, spatial_dither_mode,
- DCP_SPATIAL_DITHER_DEPTH_30BPP, frame_random_enable,
- rgb_random_enable, highpass_random_enable);
- break;
- case DCP_BIT_DEPTH_REDUCTION_MODE_TRUNCATE: /* Truncate */
- /* Truncate: Enable truncate (10bit), disable Dither */
- set_round(xfm_dce,
- DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE,
- DCP_OUT_TRUNC_ROUND_DEPTH_10BIT);
-
- set_dither(xfm_dce, false, spatial_dither_mode,
- DCP_SPATIAL_DITHER_DEPTH_30BPP, frame_random_enable,
- rgb_random_enable, highpass_random_enable);
- break;
-
- case DCP_BIT_DEPTH_REDUCTION_MODE_DISABLED: /* Disabled */
- /* Truncate: Set round/truncate to bypass (12bit),
- * disable Dither */
- set_round(xfm_dce,
- DCP_OUT_TRUNC_ROUND_MODE_TRUNCATE,
- DCP_OUT_TRUNC_ROUND_DEPTH_12BIT);
-
- set_dither(xfm_dce, false, spatial_dither_mode,
- DCP_SPATIAL_DITHER_DEPTH_30BPP, frame_random_enable,
- rgb_random_enable, highpass_random_enable);
- break;
- default:
- /* Invalid DCP Depth reduction mode */
- BREAK_TO_DEBUGGER();
- break;
- }
+ set_round(xfm_dce, trunc_mode, trunc_round_depth);
+ set_dither(xfm_dce,
+ spatial_dither_enable,
+ DCP_SPATIAL_DITHER_MODE_A_AA_A,
+ DCP_SPATIAL_DITHER_DEPTH_30BPP,
+ bit_depth_params->flags.FRAME_RANDOM,
+ bit_depth_params->flags.RGB_RANDOM,
+ bit_depth_params->flags.HIGHPASS_RANDOM);
}
static int dce_transform_get_max_num_of_supported_lines(
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 22/34] drm/amd/display: Expose DCE110 CRC functions for DCE8
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (20 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 21/34] drm/amd/display: Rework DCE transform bit depth reduction programming Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 23/34] drm/amd/display: Skip 2 frames when first reading CRC Harry Wentland
` (11 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
Implement CRC for DCE8. Registers remain the same, so call DCE110 code
directly.
Change-Id: Icddc945b6a216bb96dbc9107696dc08579e6decd
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.c | 8 ++++----
drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.h | 6 ++++++
drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c | 2 ++
3 files changed, 12 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.c
index 078d18c3eee5..be7153924a70 100644
--- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.c
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.c
@@ -2091,8 +2091,8 @@ static bool dce110_is_tg_enabled(struct timing_generator *tg)
return field == 1;
}
-static bool dce110_configure_crc(struct timing_generator *tg,
- const struct crc_params *params)
+bool dce110_configure_crc(struct timing_generator *tg,
+ const struct crc_params *params)
{
uint32_t cntl_addr = 0;
uint32_t addr = 0;
@@ -2168,8 +2168,8 @@ static bool dce110_configure_crc(struct timing_generator *tg,
return true;
}
-static bool dce110_get_crc(struct timing_generator *tg,
- uint32_t *r_cr, uint32_t *g_y, uint32_t *b_cb)
+bool dce110_get_crc(struct timing_generator *tg,
+ uint32_t *r_cr, uint32_t *g_y, uint32_t *b_cb)
{
uint32_t addr = 0;
uint32_t value = 0;
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.h b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.h
index 232747c7c60b..734d4965dab1 100644
--- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.h
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.h
@@ -276,4 +276,10 @@ void dce110_tg_set_colors(struct timing_generator *tg,
bool dce110_arm_vert_intr(
struct timing_generator *tg, uint8_t width);
+bool dce110_configure_crc(struct timing_generator *tg,
+ const struct crc_params *params);
+
+bool dce110_get_crc(struct timing_generator *tg,
+ uint32_t *r_cr, uint32_t *g_y, uint32_t *b_cb);
+
#endif /* __DC_TIMING_GENERATOR_DCE110_H__ */
diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c
index 2934650e0434..3ba4712a35ab 100644
--- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c
+++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_timing_generator.c
@@ -212,6 +212,8 @@ static const struct timing_generator_funcs dce80_tg_funcs = {
/* DCE8.0 overrides */
.enable_advanced_request =
dce80_timing_generator_enable_advanced_request,
+ .configure_crc = dce110_configure_crc,
+ .get_crc = dce110_get_crc,
};
void dce80_timing_generator_construct(
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 23/34] drm/amd/display: Skip 2 frames when first reading CRC
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (21 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 22/34] drm/amd/display: Expose DCE110 CRC functions for DCE8 Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 24/34] drm/amd/display: Remove delay on disconnect patch Harry Wentland
` (10 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
Skipping the first frame will prevent uncoooked values most of the time.
However, in some unlikely cases, the second frame will be uncooked as
well.
Change-Id: Ie811bd27a732a474b575350decc216b6fffb75d6
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
---
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h | 2 +-
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c | 12 ++++++------
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
index c40c13a9f614..5b5fa80916cb 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
@@ -198,7 +198,7 @@ struct dm_crtc_state {
struct drm_crtc_state base;
struct dc_stream_state *stream;
- bool crc_first_skipped;
+ int crc_skip_count;
bool crc_enabled;
};
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
index 55aa379cfcbe..52f2c01349e3 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
@@ -84,8 +84,8 @@ int amdgpu_dm_crtc_set_crc_source(struct drm_crtc *crtc, const char *src_name,
}
*values_cnt = 3;
- /* Reset crc_skipped flag on dm state */
- crtc_state->crc_first_skipped = false;
+ /* Reset crc_skipped on dm state */
+ crtc_state->crc_skip_count = 0;
return 0;
}
@@ -109,11 +109,11 @@ void amdgpu_dm_crtc_handle_crc_irq(struct drm_crtc *crtc)
/*
* Since flipping and crc enablement happen asynchronously, we - more
* often than not - will be returning an 'uncooked' crc on first frame.
- * Probably because hw isn't ready yet. Simply skip the first crc
- * value.
+ * Probably because hw isn't ready yet. For added security, skip the
+ * first two CRC values.
*/
- if (!crtc_state->crc_first_skipped) {
- crtc_state->crc_first_skipped = true;
+ if (crtc_state->crc_skip_count < 2) {
+ crtc_state->crc_skip_count += 1;
return;
}
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 24/34] drm/amd/display: Remove delay on disconnect patch
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (22 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 23/34] drm/amd/display: Skip 2 frames when first reading CRC Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 25/34] drm/amd/display: Make FBC work without fbdev emulation Harry Wentland
` (9 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: John Barberiz
From: John Barberiz <jbarberi@amd.com>
HDMI HPD's generic solution makes the monitor
patch code unnecessary so anything related has
been removed.
Change-Id: I302aaeba6ef805b4dfe44cbe4a8ef449c6c5211e
Signed-off-by: John Barberiz <jbarberi@amd.com>
Reviewed-by: John Barberiz <jbarberi@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc_link.c | 16 ++++------------
drivers/gpu/drm/amd/display/dc/dc_types.h | 1 -
2 files changed, 4 insertions(+), 13 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
index 44b12f0405ee..95955ade4012 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
@@ -50,7 +50,6 @@
dm_logger_write(dc_ctx->logger, LOG_HW_HOTPLUG, \
__VA_ARGS__)
-#define DEFAULT_DELAY_DISCONNECT 100
/*******************************************************************************
* Private structures
******************************************************************************/
@@ -118,7 +117,7 @@ struct gpio *get_hpd_gpio(struct dc_bios *dcb,
* true on success, false otherwise
*/
static bool program_hpd_filter(
- const struct dc_link *link, int default_disconnect_delay)
+ const struct dc_link *link)
{
bool result = false;
@@ -136,7 +135,7 @@ static bool program_hpd_filter(
case SIGNAL_TYPE_HDMI_TYPE_A:
/* Program hpd filter */
delay_on_connect_in_ms = 500;
- delay_on_disconnect_in_ms = default_disconnect_delay;
+ delay_on_disconnect_in_ms = 100;
break;
case SIGNAL_TYPE_DISPLAY_PORT:
case SIGNAL_TYPE_DISPLAY_PORT_MST:
@@ -700,13 +699,6 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason)
dp_hbr_verify_link_cap(link, &link->reported_link_cap);
}
- /* Add delay for certain monitors */
- if (sink->edid_caps.panel_patch.disconnect_delay > 0
- && sink->sink_signal == SIGNAL_TYPE_HDMI_TYPE_A)
- program_hpd_filter(link, sink->edid_caps.panel_patch.disconnect_delay);
- else
- program_hpd_filter(link, DEFAULT_DELAY_DISCONNECT);
-
/* HDMI-DVI Dongle */
if (sink->sink_signal == SIGNAL_TYPE_HDMI_TYPE_A &&
!sink->edid_caps.edid_hdmi)
@@ -1093,7 +1085,7 @@ static bool construct(
* If GPIO isn't programmed correctly HPD might not rise or drain
* fast enough, leading to bounces.
*/
- program_hpd_filter(link, DEFAULT_DELAY_DISCONNECT);
+ program_hpd_filter(link);
return true;
device_tag_fail:
@@ -2018,7 +2010,7 @@ const struct dc_link_status *dc_link_get_status(const struct dc_link *link)
void core_link_resume(struct dc_link *link)
{
if (link->connector_signal != SIGNAL_TYPE_VIRTUAL)
- program_hpd_filter(link, DEFAULT_DELAY_DISCONNECT);
+ program_hpd_filter(link);
}
static struct fixed31_32 get_pbn_per_slot(struct dc_stream_state *stream)
diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h
index aa5b90e6beb7..8811b6f86bff 100644
--- a/drivers/gpu/drm/amd/display/dc/dc_types.h
+++ b/drivers/gpu/drm/amd/display/dc/dc_types.h
@@ -194,7 +194,6 @@ union display_content_support {
};
struct dc_panel_patch {
- unsigned int disconnect_delay;
unsigned int dppowerup_delay;
};
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 25/34] drm/amd/display: Make FBC work without fbdev emulation
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (23 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 24/34] drm/amd/display: Remove delay on disconnect patch Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 26/34] drm/amd/display: Remove duplicate entries from BIOS function table Harry Wentland
` (8 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Roman Li
From: Roman Li <Roman.Li@amd.com>
Previously, FBC dynamic allocation relied on connector modes
populated during dm init. This is only the case if
DRM_FBDEV_EMULATION config flag is enabled.
Moving fbc allocation from dm_late_init() to
amdgpu_dm_connector_get_modes() where actual modes init happens.
Change-Id: Ic2251f00454bfcf926c09de1c2a54019da09675b
Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Shirish Shankarappa <Shirish.S@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 38 +++++++++--------------
1 file changed, 15 insertions(+), 23 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index b632be5203b8..511cd58c16b9 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -348,35 +348,28 @@ static void hotplug_notify_work_func(struct work_struct *work)
#if defined(CONFIG_DRM_AMD_DC_FBC)
/* Allocate memory for FBC compressed data */
-static void amdgpu_dm_fbc_init(struct amdgpu_device *adev)
+static void amdgpu_dm_fbc_init(struct drm_connector *connector)
{
+ struct drm_device *dev = connector->dev;
+ struct amdgpu_device *adev = dev->dev_private;
struct dm_comressor_info *compressor = &adev->dm.compressor;
- struct drm_connector *conn;
- struct drm_device *dev = adev->ddev;
+ struct amdgpu_dm_connector *aconn = to_amdgpu_dm_connector(connector);
+ struct drm_display_mode *mode;
unsigned long max_size = 0;
if (adev->dm.dc->fbc_compressor == NULL)
return;
- if (compressor->bo_ptr)
+ if (aconn->dc_link->connector_signal != SIGNAL_TYPE_EDP)
return;
- drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
-
- /* For eDP connector find a mode requiring max size */
- list_for_each_entry(conn,
- &dev->mode_config.connector_list, head) {
- struct amdgpu_dm_connector *aconn;
+ if (compressor->bo_ptr)
+ return;
- aconn = to_amdgpu_dm_connector(conn);
- if (aconn->dc_link->connector_signal == SIGNAL_TYPE_EDP) {
- struct drm_display_mode *mode;
- list_for_each_entry(mode, &conn->modes, head) {
- if (max_size < mode->hdisplay * mode->vdisplay)
- max_size = mode->htotal * mode->vtotal;
- }
- }
+ list_for_each_entry(mode, &connector->modes, head) {
+ if (max_size < mode->htotal * mode->vtotal)
+ max_size = mode->htotal * mode->vtotal;
}
if (max_size) {
@@ -393,7 +386,6 @@ static void amdgpu_dm_fbc_init(struct amdgpu_device *adev)
}
- drm_modeset_unlock(&dev->mode_config.connection_mutex);
}
#endif
@@ -571,9 +563,6 @@ static int dm_late_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-#if defined(CONFIG_DRM_AMD_DC_FBC)
- amdgpu_dm_fbc_init(adev);
-#endif
return detect_mst_link_for_all_connectors(adev->ddev);
}
@@ -3411,9 +3400,12 @@ static int amdgpu_dm_connector_get_modes(struct drm_connector *connector)
struct edid *edid = amdgpu_dm_connector->edid;
encoder = helper->best_encoder(connector);
-
amdgpu_dm_connector_ddc_get_modes(connector, edid);
amdgpu_dm_connector_add_common_modes(encoder, connector);
+
+#if defined(CONFIG_DRM_AMD_DC_FBC)
+ amdgpu_dm_fbc_init(connector);
+#endif
return amdgpu_dm_connector->num_modes;
}
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 26/34] drm/amd/display: Remove duplicate entries from BIOS function table
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (24 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 25/34] drm/amd/display: Make FBC work without fbdev emulation Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 27/34] drm/amd/display: Tread bad EDID as no EDID Harry Wentland
` (7 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Ken Chalmers
From: Ken Chalmers <ken.chalmers@amd.com>
Identical to the two entries above them.
Change-Id: Iebe722944338e42d3bcccaa5243f2fc1fa925793
Signed-off-by: Ken Chalmers <ken.chalmers@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/bios/bios_parser.c | 4 ----
1 file changed, 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c
index fb3cceec1a7f..69c59e050a96 100644
--- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c
@@ -3795,10 +3795,6 @@ static const struct dc_vbios_funcs vbios_funcs = {
.get_gpio_pin_info = bios_parser_get_gpio_pin_info,
- .get_embedded_panel_info = bios_parser_get_embedded_panel_info,
-
- .get_gpio_pin_info = bios_parser_get_gpio_pin_info,
-
.get_encoder_cap_info = bios_parser_get_encoder_cap_info,
/* bios scratch register communication */
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 27/34] drm/amd/display: Tread bad EDID as no EDID
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (25 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 26/34] drm/amd/display: Remove duplicate entries from BIOS function table Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
[not found] ` <20180212171625.14325-28-harry.wentland-5C7GfCeVMHo@public.gmane.org>
2018-02-12 17:16 ` [PATCH 28/34] drm/amd/display: Add missing Vega defines to dal_asic_id Harry Wentland
` (6 subsequent siblings)
33 siblings, 1 reply; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Eric Yang
From: Eric Yang <Eric.Yang2@amd.com>
We have an mst dock firmware that will emulate an EDID with bad
checksum.
Change-Id: Ice8515c929230366ec4ad7da743b448447c6db7e
Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc.c | 13 +++++++++----
1 file changed, 9 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c
index f4ffbf2e2caf..8583d9e05721 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
@@ -1637,12 +1637,17 @@ struct dc_sink *dc_link_add_remote_sink(
&dc_sink->dc_edid,
&dc_sink->edid_caps);
- if (edid_status != EDID_OK)
- goto fail;
+ /*
+ * Treat device as no EDID device if EDID
+ * parsing fails
+ */
+ if (edid_status != EDID_OK) {
+ dc_sink->dc_edid.length = 0;
+ dm_error("Bad EDID, status%d!\n", edid_status);
+ }
return dc_sink;
-fail:
- dc_link_remove_remote_sink(link, dc_sink);
+
fail_add_sink:
dc_sink_release(dc_sink);
return NULL;
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 28/34] drm/amd/display: Add missing Vega defines to dal_asic_id
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (26 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 27/34] drm/amd/display: Tread bad EDID as no EDID Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 29/34] drm/amd/display: provide an interface to query firmware version Harry Wentland
` (5 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Harry Wentland
Change-Id: I587700a29d97fd975dc297bba2104e1db3703a3e
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/include/dal_asic_id.h | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/gpu/drm/amd/display/include/dal_asic_id.h b/drivers/gpu/drm/amd/display/include/dal_asic_id.h
index 7abe663ecc6e..9831cb5eaa7c 100644
--- a/drivers/gpu/drm/amd/display/include/dal_asic_id.h
+++ b/drivers/gpu/drm/amd/display/include/dal_asic_id.h
@@ -109,6 +109,14 @@
#define ASIC_REV_IS_STONEY(rev) \
((rev >= STONEY_A0) && (rev < CZ_UNKNOWN))
+/* DCE12 */
+
+#define AI_GREENLAND_P_A0 1
+#define AI_GREENLAND_P_A1 2
+
+#define ASICREV_IS_GREENLAND_M(eChipRev) (eChipRev < AI_UNKNOWN)
+#define ASICREV_IS_GREENLAND_P(eChipRev) (eChipRev < AI_UNKNOWN)
+
/* DCN1_0 */
#define INTERNAL_REV_RAVEN_A0 0x00 /* First spin of Raven */
#define RAVEN_A0 0x01
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 29/34] drm/amd/display: provide an interface to query firmware version
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (27 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 28/34] drm/amd/display: Add missing Vega defines to dal_asic_id Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 30/34] drm/amd/display: Set vsc pack revision when DPCD revision is >= 1.2 Harry Wentland
` (4 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Anthony Koo
From: Anthony Koo <Anthony.Koo@amd.com>
Change-Id: I5c59531ebb145a8860358dbfee2d273adf2f09b6
Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc.c | 7 +++++++
drivers/gpu/drm/amd/display/dc/dc.h | 13 +++++++++++++
drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c | 2 +-
drivers/gpu/drm/amd/display/dc/inc/hw/dmcu.h | 7 -------
4 files changed, 21 insertions(+), 8 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c
index 8583d9e05721..77a1bf233c3c 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
@@ -39,6 +39,7 @@
#include "bios_parser_interface.h"
#include "include/irq_service_interface.h"
#include "transform.h"
+#include "dmcu.h"
#include "dpp.h"
#include "timing_generator.h"
#include "virtual/virtual_link_encoder.h"
@@ -609,6 +610,12 @@ struct dc *dc_create(const struct dc_init_data *init_params)
dc->caps.max_audios = dc->res_pool->audio_count;
dc->caps.linear_pitch_alignment = 64;
+ /* Populate versioning information */
+ dc->versions.dc_ver = DC_VER;
+
+ if (dc->res_pool->dmcu != NULL)
+ dc->versions.dmcu_version = dc->res_pool->dmcu->dmcu_version;
+
dc->config = init_params->flags;
dm_logger_write(dc->ctx->logger, LOG_DC,
diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index a512b05303e2..5bb0e5defaf4 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -48,6 +48,18 @@
/*******************************************************************************
* Display Core Interfaces
******************************************************************************/
+struct dmcu_version {
+ unsigned int date;
+ unsigned int month;
+ unsigned int year;
+ unsigned int interface_version;
+};
+
+struct dc_versions {
+ const char *dc_ver;
+ struct dmcu_version dmcu_version;
+};
+
struct dc_caps {
uint32_t max_streams;
uint32_t max_links;
@@ -221,6 +233,7 @@ struct dc_state;
struct resource_pool;
struct dce_hwseq;
struct dc {
+ struct dc_versions versions;
struct dc_caps caps;
struct dc_cap_funcs cap_funcs;
struct dc_config config;
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c
index 48e21ead3142..2ee3d9bf1062 100644
--- a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c
@@ -360,7 +360,7 @@ static void dcn10_get_dmcu_version(struct dmcu *dmcu)
dmcu->dmcu_version.year = ((REG_READ(DMCU_IRAM_RD_DATA) << 8) |
REG_READ(DMCU_IRAM_RD_DATA));
dmcu->dmcu_version.month = REG_READ(DMCU_IRAM_RD_DATA);
- dmcu->dmcu_version.day = REG_READ(DMCU_IRAM_RD_DATA);
+ dmcu->dmcu_version.date = REG_READ(DMCU_IRAM_RD_DATA);
/* Disable write access to IRAM to allow dynamic sleep state */
REG_UPDATE_2(DMCU_RAM_ACCESS_CTRL,
diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dmcu.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dmcu.h
index ce206355461b..de60f940030d 100644
--- a/drivers/gpu/drm/amd/display/dc/inc/hw/dmcu.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dmcu.h
@@ -32,13 +32,6 @@ enum dmcu_state {
DMCU_RUNNING = 1
};
-struct dmcu_version {
- unsigned int day;
- unsigned int month;
- unsigned int year;
- unsigned int interface_version;
-};
-
struct dmcu {
struct dc_context *ctx;
const struct dmcu_funcs *funcs;
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 30/34] drm/amd/display: Set vsc pack revision when DPCD revision is >= 1.2
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (28 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 29/34] drm/amd/display: provide an interface to query firmware version Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 31/34] drm/amd/display: Add logging for aux DPCD access Harry Wentland
` (3 subsequent siblings)
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Tao
From: Tao <xtao@amd.com>
Brightness couldn't change when booting up in DC mode.
It was because "psr_enabled" flag was not set to true before
setting vsc packet revision, causing packet rev setup was skipped.
Now instead of checking the psr flag, it checks if the DPCD_REV >= 1.2
and set the vsc packet revision.
Change-Id: I7a3f1f810e6c8263524356a3751be93e033f9202
Signed-off-by: Tao <xtao@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/core/dc_resource.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
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 a6f8c1f93693..ce0e9e76eb35 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
@@ -35,6 +35,7 @@
#include "core_types.h"
#include "set_mode_types.h"
#include "virtual/virtual_stream_encoder.h"
+#include "dpcd_defs.h"
#include "dce80/dce80_resource.h"
#include "dce100/dce100_resource.h"
@@ -2434,7 +2435,8 @@ static void set_vsc_info_packet(
unsigned int vscPacketRevision = 0;
unsigned int i;
- if (stream->sink->link->psr_enabled) {
+ /*VSC packet set to 2 when DP revision >= 1.2*/
+ if (stream->sink->link->dpcd_caps.dpcd_rev.raw >= DPCD_REV_12) {
vscPacketRevision = 2;
}
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 31/34] drm/amd/display: Add logging for aux DPCD access
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (29 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 30/34] drm/amd/display: Set vsc pack revision when DPCD revision is >= 1.2 Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
[not found] ` <20180212171625.14325-32-harry.wentland-5C7GfCeVMHo@public.gmane.org>
2018-02-12 17:16 ` [PATCH 32/34] drm/amd/display: Remove unused dm_pp_ interfaces Harry Wentland
` (2 subsequent siblings)
33 siblings, 1 reply; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Eric Yang
From: Eric Yang <Eric.Yang2@amd.com>
Add basic logging for DPCD access. Does not print
by default.
Currently only prints first byte of the data accessed.
Technical debt: Need to make it so that the entire
data block accessed is printed. Also need to log
address space that's not DPCD.
Change-Id: I10ef7042c14d70508845ef827ebec2432d8d8176
Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/i2caux/aux_engine.c | 16 ++++++++++++++++
1 file changed, 16 insertions(+)
diff --git a/drivers/gpu/drm/amd/display/dc/i2caux/aux_engine.c b/drivers/gpu/drm/amd/display/dc/i2caux/aux_engine.c
index fc7a7d4ebca5..0b1db48fef36 100644
--- a/drivers/gpu/drm/amd/display/dc/i2caux/aux_engine.c
+++ b/drivers/gpu/drm/amd/display/dc/i2caux/aux_engine.c
@@ -284,6 +284,14 @@ static bool read_command(
msleep(engine->delay);
} while (ctx.operation_succeeded && !ctx.transaction_complete);
+ if (request->payload.address_space ==
+ I2CAUX_TRANSACTION_ADDRESS_SPACE_DPCD) {
+ dm_logger_write(engine->base.ctx->logger, LOG_I2C_AUX, "READ: addr:0x%x value:0x%x Result:%d",
+ request->payload.address,
+ request->payload.data[0],
+ ctx.operation_succeeded);
+ }
+
return ctx.operation_succeeded;
}
@@ -484,6 +492,14 @@ static bool write_command(
msleep(engine->delay);
} while (ctx.operation_succeeded && !ctx.transaction_complete);
+ if (request->payload.address_space ==
+ I2CAUX_TRANSACTION_ADDRESS_SPACE_DPCD) {
+ dm_logger_write(engine->base.ctx->logger, LOG_I2C_AUX, "WRITE: addr:0x%x value:0x%x Result:%d",
+ request->payload.address,
+ request->payload.data[0],
+ ctx.operation_succeeded);
+ }
+
return ctx.operation_succeeded;
}
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 32/34] drm/amd/display: Remove unused dm_pp_ interfaces
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (30 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 31/34] drm/amd/display: Add logging for aux DPCD access Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 33/34] drm/amd/display: De PQ implementation Harry Wentland
2018-02-12 17:16 ` [PATCH 34/34] drm/amd/display: Fix increment when sampling OTF in DCE Harry Wentland
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Harry Wentland
Change-Id: Iaafba5a3a5fd8fee6838503ce579bdb7dd1c632f
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
.../drm/amd/display/amdgpu_dm/amdgpu_dm_services.c | 33 ----------------------
drivers/gpu/drm/amd/display/dc/dm_services.h | 31 --------------------
2 files changed, 64 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_services.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_services.c
index 56e549249134..89342b48be6b 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_services.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_services.c
@@ -71,15 +71,6 @@ bool dm_read_persistent_data(struct dc_context *ctx,
/**** power component interfaces ****/
-bool dm_pp_pre_dce_clock_change(
- struct dc_context *ctx,
- struct dm_pp_gpu_clock_range *requested_state,
- struct dm_pp_gpu_clock_range *actual_state)
-{
- /*TODO*/
- return false;
-}
-
bool dm_pp_apply_display_requirements(
const struct dc_context *ctx,
const struct dm_pp_display_configuration *pp_display_cfg)
@@ -151,30 +142,6 @@ bool dm_pp_apply_display_requirements(
return true;
}
-bool dc_service_get_system_clocks_range(
- const struct dc_context *ctx,
- struct dm_pp_gpu_clock_range *sys_clks)
-{
- struct amdgpu_device *adev = ctx->driver_context;
-
- /* Default values, in case PPLib is not compiled-in. */
- sys_clks->mclk.max_khz = 800000;
- sys_clks->mclk.min_khz = 800000;
-
- sys_clks->sclk.max_khz = 600000;
- sys_clks->sclk.min_khz = 300000;
-
- if (adev->pm.dpm_enabled) {
- sys_clks->mclk.max_khz = amdgpu_dpm_get_mclk(adev, false);
- sys_clks->mclk.min_khz = amdgpu_dpm_get_mclk(adev, true);
-
- sys_clks->sclk.max_khz = amdgpu_dpm_get_sclk(adev, false);
- sys_clks->sclk.min_khz = amdgpu_dpm_get_sclk(adev, true);
- }
-
- return true;
-}
-
static void get_default_clock_levels(
enum dm_pp_clock_type clk_type,
struct dm_pp_clock_levels *clks)
diff --git a/drivers/gpu/drm/amd/display/dc/dm_services.h b/drivers/gpu/drm/amd/display/dc/dm_services.h
index 225b7bfb09a9..22e7ee7dcd26 100644
--- a/drivers/gpu/drm/amd/display/dc/dm_services.h
+++ b/drivers/gpu/drm/amd/display/dc/dm_services.h
@@ -192,37 +192,6 @@ unsigned int generic_reg_wait(const struct dc_context *ctx,
* Power Play (PP) interfaces
**************************************/
-/* DAL calls this function to notify PP about clocks it needs for the Mode Set.
- * This is done *before* it changes DCE clock.
- *
- * If required clock is higher than current, then PP will increase the voltage.
- *
- * If required clock is lower than current, then PP will defer reduction of
- * voltage until the call to dc_service_pp_post_dce_clock_change().
- *
- * \input - Contains clocks needed for Mode Set.
- *
- * \output - Contains clocks adjusted by PP which DAL should use for Mode Set.
- * Valid only if function returns zero.
- *
- * \returns true - call is successful
- * false - call failed
- */
-bool dm_pp_pre_dce_clock_change(
- struct dc_context *ctx,
- struct dm_pp_gpu_clock_range *requested_state,
- struct dm_pp_gpu_clock_range *actual_state);
-
-/* The returned clocks range are 'static' system clocks which will be used for
- * mode validation purposes.
- *
- * \returns true - call is successful
- * false - call failed
- */
-bool dc_service_get_system_clocks_range(
- const struct dc_context *ctx,
- struct dm_pp_gpu_clock_range *sys_clks);
-
/* Gets valid clocks levels from pplib
*
* input: clk_type - display clk / sclk / mem clk
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 33/34] drm/amd/display: De PQ implementation
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (31 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 32/34] drm/amd/display: Remove unused dm_pp_ interfaces Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
2018-02-12 17:16 ` [PATCH 34/34] drm/amd/display: Fix increment when sampling OTF in DCE Harry Wentland
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Vitaly Prosyak
From: Vitaly Prosyak <vitaly.prosyak@amd.com>
Some refactoring and optimizations in color module.
Added de gamma 2.2 & 2.4, also re gamma 2.2.
Added interface for diagnostic for de gamma & de pq.
Change-Id: If7d9309a8ad27765c342adb80177caae04913cf8
Signed-off-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
.../gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c | 30 +-
.../drm/amd/display/modules/color/color_gamma.c | 479 +++++++++++++++++----
.../drm/amd/display/modules/color/color_gamma.h | 8 +
3 files changed, 442 insertions(+), 75 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c
index 2482390c1557..bd3fcdfb79c5 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c
@@ -298,6 +298,32 @@ static void dpp1_cm_get_reg_field(
reg->masks.exp_resion_start_segment = dpp->tf_mask->CM_RGAM_RAMB_EXP_REGION_START_SEGMENT_B;
}
+static void dpp1_cm_get_degamma_reg_field(
+ struct dcn10_dpp *dpp,
+ struct xfer_func_reg *reg)
+{
+ reg->shifts.exp_region0_lut_offset = dpp->tf_shift->CM_DGAM_RAMA_EXP_REGION0_LUT_OFFSET;
+ reg->masks.exp_region0_lut_offset = dpp->tf_mask->CM_DGAM_RAMA_EXP_REGION0_LUT_OFFSET;
+ reg->shifts.exp_region0_num_segments = dpp->tf_shift->CM_DGAM_RAMA_EXP_REGION0_NUM_SEGMENTS;
+ reg->masks.exp_region0_num_segments = dpp->tf_mask->CM_DGAM_RAMA_EXP_REGION0_NUM_SEGMENTS;
+ reg->shifts.exp_region1_lut_offset = dpp->tf_shift->CM_DGAM_RAMA_EXP_REGION1_LUT_OFFSET;
+ reg->masks.exp_region1_lut_offset = dpp->tf_mask->CM_DGAM_RAMA_EXP_REGION1_LUT_OFFSET;
+ reg->shifts.exp_region1_num_segments = dpp->tf_shift->CM_DGAM_RAMA_EXP_REGION1_NUM_SEGMENTS;
+ reg->masks.exp_region1_num_segments = dpp->tf_mask->CM_DGAM_RAMA_EXP_REGION1_NUM_SEGMENTS;
+
+ reg->shifts.field_region_end = dpp->tf_shift->CM_DGAM_RAMB_EXP_REGION_END_B;
+ reg->masks.field_region_end = dpp->tf_mask->CM_DGAM_RAMB_EXP_REGION_END_B;
+ reg->shifts.field_region_end_slope = dpp->tf_shift->CM_DGAM_RAMB_EXP_REGION_END_SLOPE_B;
+ reg->masks.field_region_end_slope = dpp->tf_mask->CM_DGAM_RAMB_EXP_REGION_END_SLOPE_B;
+ reg->shifts.field_region_end_base = dpp->tf_shift->CM_DGAM_RAMB_EXP_REGION_END_BASE_B;
+ reg->masks.field_region_end_base = dpp->tf_mask->CM_DGAM_RAMB_EXP_REGION_END_BASE_B;
+ reg->shifts.field_region_linear_slope = dpp->tf_shift->CM_DGAM_RAMB_EXP_REGION_LINEAR_SLOPE_B;
+ reg->masks.field_region_linear_slope = dpp->tf_mask->CM_DGAM_RAMB_EXP_REGION_LINEAR_SLOPE_B;
+ reg->shifts.exp_region_start = dpp->tf_shift->CM_DGAM_RAMB_EXP_REGION_START_B;
+ reg->masks.exp_region_start = dpp->tf_mask->CM_DGAM_RAMB_EXP_REGION_START_B;
+ reg->shifts.exp_resion_start_segment = dpp->tf_shift->CM_DGAM_RAMB_EXP_REGION_START_SEGMENT_B;
+ reg->masks.exp_resion_start_segment = dpp->tf_mask->CM_DGAM_RAMB_EXP_REGION_START_SEGMENT_B;
+}
void dpp1_cm_set_output_csc_adjustment(
struct dpp *dpp_base,
const uint16_t *regval)
@@ -502,7 +528,7 @@ void dpp1_program_degamma_lutb_settings(
struct dcn10_dpp *dpp = TO_DCN10_DPP(dpp_base);
struct xfer_func_reg gam_regs;
- dpp1_cm_get_reg_field(dpp, &gam_regs);
+ dpp1_cm_get_degamma_reg_field(dpp, &gam_regs);
gam_regs.start_cntl_b = REG(CM_DGAM_RAMB_START_CNTL_B);
gam_regs.start_cntl_g = REG(CM_DGAM_RAMB_START_CNTL_G);
@@ -531,7 +557,7 @@ void dpp1_program_degamma_luta_settings(
struct dcn10_dpp *dpp = TO_DCN10_DPP(dpp_base);
struct xfer_func_reg gam_regs;
- dpp1_cm_get_reg_field(dpp, &gam_regs);
+ dpp1_cm_get_degamma_reg_field(dpp, &gam_regs);
gam_regs.start_cntl_b = REG(CM_DGAM_RAMA_START_CNTL_B);
gam_regs.start_cntl_g = REG(CM_DGAM_RAMA_START_CNTL_G);
diff --git a/drivers/gpu/drm/amd/display/modules/color/color_gamma.c b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c
index fde3ae8b12a5..a5fd14a4016f 100644
--- a/drivers/gpu/drm/amd/display/modules/color/color_gamma.c
+++ b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c
@@ -27,14 +27,21 @@
#include "opp.h"
#include "color_gamma.h"
-/* MAX_HW_POINTS = NUM_REGIONS * NUM_PTS_IN_REGION */
+
#define NUM_PTS_IN_REGION 16
#define NUM_REGIONS 32
-#define MAX_HW_POINTS 512
+#define NUM_DEGAMMA_REGIONS 12
+#define MAX_HW_POINTS (NUM_PTS_IN_REGION*NUM_REGIONS)
+#define MAX_HW_DEGAMMA_POINTS (NUM_PTS_IN_REGION*NUM_DEGAMMA_REGIONS)
static struct hw_x_point coordinates_x[MAX_HW_POINTS + 2];
+static struct hw_x_point degamma_coordinates_x[MAX_HW_DEGAMMA_POINTS + 2];
+
static struct fixed31_32 pq_table[MAX_HW_POINTS + 2];
+static struct fixed31_32 de_pq_table[MAX_HW_DEGAMMA_POINTS + 2];
+
static bool pq_initialized; /* = false; */
+static bool de_pq_initialized; /* = false; */
/* one-time setup of X points */
void setup_x_points_distribution(void)
@@ -45,8 +52,8 @@ void setup_x_points_distribution(void)
uint32_t index;
struct fixed31_32 increment;
- coordinates_x[NUM_REGIONS * NUM_PTS_IN_REGION].x = region_size;
- coordinates_x[NUM_REGIONS * NUM_PTS_IN_REGION + 1].x = region_size;
+ coordinates_x[MAX_HW_POINTS].x = region_size;
+ coordinates_x[MAX_HW_POINTS + 1].x = region_size;
for (segment = 6; segment > (6 - NUM_REGIONS); segment--) {
region_size = dal_fixed31_32_div_int(region_size, 2);
@@ -62,6 +69,26 @@ void setup_x_points_distribution(void)
(coordinates_x[index-1].x, increment);
}
}
+
+ region_size = dal_fixed31_32_from_int(1);
+ degamma_coordinates_x[MAX_HW_DEGAMMA_POINTS].x = region_size;
+ degamma_coordinates_x[MAX_HW_DEGAMMA_POINTS + 1].x = region_size;
+
+ for (segment = -1; segment > -(NUM_DEGAMMA_REGIONS + 1); segment--) {
+ region_size = dal_fixed31_32_div_int(region_size, 2);
+ increment = dal_fixed31_32_div_int(region_size,
+ NUM_PTS_IN_REGION);
+ seg_offset = (segment + NUM_DEGAMMA_REGIONS) * NUM_PTS_IN_REGION;
+ degamma_coordinates_x[seg_offset].x = region_size;
+
+ for (index = seg_offset + 1;
+ index < seg_offset + NUM_PTS_IN_REGION;
+ index++) {
+ degamma_coordinates_x[index].x = dal_fixed31_32_add
+ (degamma_coordinates_x[index-1].x, increment);
+ }
+ }
+
}
static void compute_pq(struct fixed31_32 in_x, struct fixed31_32 *out_y)
@@ -93,6 +120,40 @@ static void compute_pq(struct fixed31_32 in_x, struct fixed31_32 *out_y)
*out_y = dal_fixed31_32_pow(base, m2);
}
+static void compute_de_pq(struct fixed31_32 in_x, struct fixed31_32 *out_y)
+{
+ /* consts for dePQ gamma formula. */
+ const struct fixed31_32 m1 =
+ dal_fixed31_32_from_fraction(159301758, 1000000000);
+ const struct fixed31_32 m2 =
+ dal_fixed31_32_from_fraction(7884375, 100000);
+ const struct fixed31_32 c1 =
+ dal_fixed31_32_from_fraction(8359375, 10000000);
+ const struct fixed31_32 c2 =
+ dal_fixed31_32_from_fraction(188515625, 10000000);
+ const struct fixed31_32 c3 =
+ dal_fixed31_32_from_fraction(186875, 10000);
+
+ struct fixed31_32 l_pow_m1;
+ struct fixed31_32 base, div;
+
+
+ if (dal_fixed31_32_lt(in_x, dal_fixed31_32_zero))
+ in_x = dal_fixed31_32_zero;
+
+ l_pow_m1 = dal_fixed31_32_pow(in_x,
+ dal_fixed31_32_div(dal_fixed31_32_one, m2));
+ base = dal_fixed31_32_sub(l_pow_m1, c1);
+
+ if (dal_fixed31_32_lt(base, dal_fixed31_32_zero))
+ base = dal_fixed31_32_zero;
+
+ div = dal_fixed31_32_sub(c2, dal_fixed31_32_mul(c3, l_pow_m1));
+
+ *out_y = dal_fixed31_32_pow(dal_fixed31_32_div(base, div),
+ dal_fixed31_32_div(dal_fixed31_32_one, m1));
+
+}
/* one-time pre-compute PQ values - only for sdr_white_level 80 */
void precompute_pq(void)
{
@@ -113,46 +174,49 @@ void precompute_pq(void)
}
}
+/* one-time pre-compute dePQ values - only for max pixel value 125 FP16 */
+void precompute_de_pq(void)
+{
+ int i;
+ struct fixed31_32 y;
+ const struct hw_x_point *coord_x = degamma_coordinates_x;
+ struct fixed31_32 scaling_factor = dal_fixed31_32_from_int(125);
+
+
+ for (i = 0; i <= MAX_HW_DEGAMMA_POINTS; i++) {
+ compute_de_pq(coord_x->x, &y);
+ de_pq_table[i] = dal_fixed31_32_mul(y, scaling_factor);
+ ++coord_x;
+ }
+}
struct dividers {
struct fixed31_32 divider1;
struct fixed31_32 divider2;
struct fixed31_32 divider3;
};
-static void build_regamma_coefficients(struct gamma_coefficients *coefficients)
+static void build_coefficients(struct gamma_coefficients *coefficients, bool is_2_4)
{
- /* sRGB should apply 2.4 */
- static const int32_t numerator01[3] = { 31308, 31308, 31308 };
- static const int32_t numerator02[3] = { 12920, 12920, 12920 };
- static const int32_t numerator03[3] = { 55, 55, 55 };
- static const int32_t numerator04[3] = { 55, 55, 55 };
- static const int32_t numerator05[3] = { 2400, 2400, 2400 };
-
- const int32_t *numerator1;
- const int32_t *numerator2;
- const int32_t *numerator3;
- const int32_t *numerator4;
- const int32_t *numerator5;
-
- uint32_t i = 0;
+ static const int32_t numerator01[] = { 31308, 180000};
+ static const int32_t numerator02[] = { 12920, 4500};
+ static const int32_t numerator03[] = { 55, 99};
+ static const int32_t numerator04[] = { 55, 99};
+ static const int32_t numerator05[] = { 2400, 2200};
- numerator1 = numerator01;
- numerator2 = numerator02;
- numerator3 = numerator03;
- numerator4 = numerator04;
- numerator5 = numerator05;
+ uint32_t i = 0;
+ uint32_t index = is_2_4 == true ? 0:1;
do {
coefficients->a0[i] = dal_fixed31_32_from_fraction(
- numerator1[i], 10000000);
+ numerator01[index], 10000000);
coefficients->a1[i] = dal_fixed31_32_from_fraction(
- numerator2[i], 1000);
+ numerator02[index], 1000);
coefficients->a2[i] = dal_fixed31_32_from_fraction(
- numerator3[i], 1000);
+ numerator03[index], 1000);
coefficients->a3[i] = dal_fixed31_32_from_fraction(
- numerator4[i], 1000);
+ numerator04[index], 1000);
coefficients->user_gamma[i] = dal_fixed31_32_from_fraction(
- numerator5[i], 1000);
+ numerator05[index], 1000);
++i;
} while (i != ARRAY_SIZE(coefficients->a0));
@@ -197,6 +261,39 @@ static struct fixed31_32 translate_from_linear_space(
a1);
}
+static struct fixed31_32 translate_to_linear_space(
+ struct fixed31_32 arg,
+ struct fixed31_32 a0,
+ struct fixed31_32 a1,
+ struct fixed31_32 a2,
+ struct fixed31_32 a3,
+ struct fixed31_32 gamma)
+{
+ struct fixed31_32 linear;
+
+ a0 = dal_fixed31_32_mul(a0, a1);
+ if (dal_fixed31_32_le(arg, dal_fixed31_32_neg(a0)))
+
+ linear = dal_fixed31_32_neg(
+ dal_fixed31_32_pow(
+ dal_fixed31_32_div(
+ dal_fixed31_32_sub(a2, arg),
+ dal_fixed31_32_add(
+ dal_fixed31_32_one, a3)), gamma));
+
+ else if (dal_fixed31_32_le(dal_fixed31_32_neg(a0), arg) &&
+ dal_fixed31_32_le(arg, a0))
+ linear = dal_fixed31_32_div(arg, a1);
+ else
+ linear = dal_fixed31_32_pow(
+ dal_fixed31_32_div(
+ dal_fixed31_32_add(a2, arg),
+ dal_fixed31_32_add(
+ dal_fixed31_32_one, a3)), gamma);
+
+ return linear;
+}
+
static inline struct fixed31_32 translate_from_linear_space_ex(
struct fixed31_32 arg,
struct gamma_coefficients *coeff,
@@ -211,6 +308,22 @@ static inline struct fixed31_32 translate_from_linear_space_ex(
coeff->user_gamma[color_index]);
}
+
+static inline struct fixed31_32 translate_to_linear_space_ex(
+ struct fixed31_32 arg,
+ struct gamma_coefficients *coeff,
+ uint32_t color_index)
+{
+ return translate_to_linear_space(
+ arg,
+ coeff->a0[color_index],
+ coeff->a1[color_index],
+ coeff->a2[color_index],
+ coeff->a3[color_index],
+ coeff->user_gamma[color_index]);
+}
+
+
static bool find_software_points(
const struct dc_gamma *ramp,
const struct gamma_pixel *axis_x,
@@ -314,12 +427,6 @@ static bool build_custom_gamma_mapping_coefficients_worker(
struct fixed31_32 left_pos;
struct fixed31_32 right_pos;
- /*
- * TODO: confirm enum in surface_pixel_format
- * if (pixel_format == PIXEL_FORMAT_FP16)
- *coord_x = coordinates_x[i].adjusted_x;
- *else
- */
if (channel == CHANNEL_NAME_RED)
coord_x = coordinates_x[i].regamma_y_red;
else if (channel == CHANNEL_NAME_GREEN)
@@ -451,7 +558,7 @@ static struct fixed31_32 calculate_mapped_value(
return result;
}
-static void build_regamma_curve_pq(struct pwl_float_data_ex *rgb_regamma,
+static void build_pq(struct pwl_float_data_ex *rgb_regamma,
uint32_t hw_points_num,
const struct hw_x_point *coordinate_x,
uint32_t sdr_white_level)
@@ -477,11 +584,6 @@ static void build_regamma_curve_pq(struct pwl_float_data_ex *rgb_regamma,
rgb += start_index;
coord_x += start_index;
- /* use coord_x to retrieve coordinates chosen base on given user curve
- * the x values are exponentially distributed and currently it is hard
- * coded, the user curve shape is ignored. Need to recalculate coord_x
- * based on input curve, translation from 256/1025 to 128 PWL points.
- */
for (i = start_index; i <= hw_points_num; i++) {
/* Multiply 0.008 as regamma is 0-1 and FP16 input is 0-125.
* FP 1.0 = 80nits
@@ -508,37 +610,86 @@ static void build_regamma_curve_pq(struct pwl_float_data_ex *rgb_regamma,
}
}
-static void build_regamma_curve(struct pwl_float_data_ex *rgb_regamma,
+static void build_de_pq(struct pwl_float_data_ex *de_pq,
uint32_t hw_points_num,
const struct hw_x_point *coordinate_x)
{
uint32_t i;
+ struct fixed31_32 output;
+
+ struct pwl_float_data_ex *rgb = de_pq;
+ const struct hw_x_point *coord_x = degamma_coordinates_x;
+ struct fixed31_32 scaling_factor = dal_fixed31_32_from_int(125);
+
+ if (!de_pq_initialized) {
+ precompute_de_pq();
+ de_pq_initialized = true;
+ }
+
+
+ for (i = 0; i <= hw_points_num; i++) {
+ output = de_pq_table[i];
+ /* should really not happen? */
+ if (dal_fixed31_32_lt(output, dal_fixed31_32_zero))
+ output = dal_fixed31_32_zero;
+ else if (dal_fixed31_32_lt(scaling_factor, output))
+ output = scaling_factor;
+
+ rgb->r = output;
+ rgb->g = output;
+ rgb->b = output;
+
+ ++coord_x;
+ ++rgb;
+ }
+}
+
+static void build_regamma(struct pwl_float_data_ex *rgb_regamma,
+ uint32_t hw_points_num,
+ const struct hw_x_point *coordinate_x, bool is_2_4)
+{
+ uint32_t i;
struct gamma_coefficients coeff;
struct pwl_float_data_ex *rgb = rgb_regamma;
const struct hw_x_point *coord_x = coordinate_x;
- build_regamma_coefficients(&coeff);
-
- /* Use opp110->regamma.coordinates_x to retrieve
- * coordinates chosen base on given user curve (future task).
- * The x values are exponentially distributed and currently
- * it is hard-coded, the user curve shape is ignored.
- * The future task is to recalculate opp110-
- * regamma.coordinates_x based on input/user curve,
- * translation from 256/1025 to 128 pwl points.
- */
+ build_coefficients(&coeff, is_2_4);
i = 0;
while (i != hw_points_num + 1) {
+ /*TODO use y vs r,g,b*/
rgb->r = translate_from_linear_space_ex(
coord_x->x, &coeff, 0);
- rgb->g = translate_from_linear_space_ex(
- coord_x->x, &coeff, 1);
- rgb->b = translate_from_linear_space_ex(
- coord_x->x, &coeff, 2);
+ rgb->g = rgb->r;
+ rgb->b = rgb->r;
+ ++coord_x;
+ ++rgb;
+ ++i;
+ }
+}
+
+static void build_degamma(struct pwl_float_data_ex *curve,
+ uint32_t hw_points_num,
+ const struct hw_x_point *coordinate_x, bool is_2_4)
+{
+ uint32_t i;
+ struct gamma_coefficients coeff;
+ struct pwl_float_data_ex *rgb = curve;
+ const struct hw_x_point *coord_x = degamma_coordinates_x;
+
+ build_coefficients(&coeff, is_2_4);
+
+ i = 0;
+
+ while (i != hw_points_num + 1) {
+ /*TODO use y vs r,g,b*/
+ rgb->r = translate_to_linear_space_ex(
+ coord_x->x, &coeff, 0);
+ rgb->g = rgb->r;
+ rgb->b = rgb->r;
++coord_x;
++rgb;
++i;
@@ -921,6 +1072,8 @@ static bool map_regamma_hw_to_x_user(
return true;
}
+#define _EXTRA_POINTS 3
+
bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
const struct dc_gamma *ramp, bool mapUserRamp)
{
@@ -930,7 +1083,7 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
struct pwl_float_data *rgb_user = NULL;
struct pwl_float_data_ex *rgb_regamma = NULL;
struct gamma_pixel *axix_x = NULL;
- struct pixel_gamma_point *coeff128 = NULL;
+ struct pixel_gamma_point *coeff = NULL;
enum dc_transfer_func_predefined tf = TRANSFER_FUNCTION_SRGB;
bool ret = false;
@@ -945,11 +1098,11 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
output_tf->type = TF_TYPE_DISTRIBUTED_POINTS;
- rgb_user = kzalloc(sizeof(*rgb_user) * (ramp->num_entries + 3),
+ rgb_user = kzalloc(sizeof(*rgb_user) * (ramp->num_entries + _EXTRA_POINTS),
GFP_KERNEL);
if (!rgb_user)
goto rgb_user_alloc_fail;
- rgb_regamma = kzalloc(sizeof(*rgb_regamma) * (MAX_HW_POINTS + 3),
+ rgb_regamma = kzalloc(sizeof(*rgb_regamma) * (MAX_HW_POINTS + _EXTRA_POINTS),
GFP_KERNEL);
if (!rgb_regamma)
goto rgb_regamma_alloc_fail;
@@ -957,9 +1110,9 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
GFP_KERNEL);
if (!axix_x)
goto axix_x_alloc_fail;
- coeff128 = kzalloc(sizeof(*coeff128) * (MAX_HW_POINTS + 3), GFP_KERNEL);
- if (!coeff128)
- goto coeff128_alloc_fail;
+ coeff = kzalloc(sizeof(*coeff) * (MAX_HW_POINTS + _EXTRA_POINTS), GFP_KERNEL);
+ if (!coeff)
+ goto coeff_alloc_fail;
dividers.divider1 = dal_fixed31_32_from_fraction(3, 2);
dividers.divider2 = dal_fixed31_32_from_int(2);
@@ -983,7 +1136,7 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
tf_pts->x_point_at_y1_green = 125;
tf_pts->x_point_at_y1_blue = 125;
- build_regamma_curve_pq(rgb_regamma,
+ build_pq(rgb_regamma,
MAX_HW_POINTS,
coordinates_x,
output_tf->sdr_ref_white_level);
@@ -993,12 +1146,12 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
tf_pts->x_point_at_y1_green = 1;
tf_pts->x_point_at_y1_blue = 1;
- build_regamma_curve(rgb_regamma,
+ build_regamma(rgb_regamma,
MAX_HW_POINTS,
- coordinates_x);
+ coordinates_x, tf == TRANSFER_FUNCTION_SRGB ? true:false);
}
- map_regamma_hw_to_x_user(ramp, coeff128, rgb_user,
+ map_regamma_hw_to_x_user(ramp, coeff, rgb_user,
coordinates_x, axix_x, rgb_regamma,
MAX_HW_POINTS, tf_pts,
(mapUserRamp || ramp->type != GAMMA_RGB_256) &&
@@ -1009,8 +1162,8 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
ret = true;
- kfree(coeff128);
-coeff128_alloc_fail:
+ kfree(coeff);
+coeff_alloc_fail:
kfree(axix_x);
axix_x_alloc_fail:
kfree(rgb_regamma);
@@ -1024,6 +1177,98 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
/*TODO fix me should be 2*/
#define _EXTRA_POINTS 3
+bool mod_color_calculate_degamma_params(struct dc_transfer_func *input_tf,
+ const struct dc_gamma *ramp, bool mapUserRamp)
+{
+ struct dc_transfer_func_distributed_points *tf_pts = &input_tf->tf_pts;
+ struct dividers dividers;
+
+ struct pwl_float_data *rgb_user = NULL;
+ struct pwl_float_data_ex *curve = NULL;
+ struct gamma_pixel *axix_x = NULL;
+ struct pixel_gamma_point *coeff = NULL;
+ enum dc_transfer_func_predefined tf = TRANSFER_FUNCTION_SRGB;
+ bool ret = false;
+
+ if (input_tf->type == TF_TYPE_BYPASS)
+ return false;
+
+ /* we can use hardcoded curve for plain SRGB TF */
+ if (input_tf->type == TF_TYPE_PREDEFINED &&
+ input_tf->tf == TRANSFER_FUNCTION_SRGB &&
+ (!mapUserRamp && ramp->type == GAMMA_RGB_256))
+ return true;
+
+ input_tf->type = TF_TYPE_DISTRIBUTED_POINTS;
+
+ rgb_user = kzalloc(sizeof(*rgb_user) * (ramp->num_entries + _EXTRA_POINTS),
+ GFP_KERNEL);
+ if (!rgb_user)
+ goto rgb_user_alloc_fail;
+ curve = kzalloc(sizeof(*curve) * (MAX_HW_DEGAMMA_POINTS + _EXTRA_POINTS),
+ GFP_KERNEL);
+ if (!curve)
+ goto curve_alloc_fail;
+ axix_x = kzalloc(sizeof(*axix_x) * (ramp->num_entries + _EXTRA_POINTS),
+ GFP_KERNEL);
+ if (!axix_x)
+ goto axix_x_alloc_fail;
+ coeff = kzalloc(sizeof(*coeff) * (MAX_HW_DEGAMMA_POINTS + _EXTRA_POINTS), GFP_KERNEL);
+ if (!coeff)
+ goto coeff_alloc_fail;
+
+ dividers.divider1 = dal_fixed31_32_from_fraction(3, 2);
+ dividers.divider2 = dal_fixed31_32_from_int(2);
+ dividers.divider3 = dal_fixed31_32_from_fraction(5, 2);
+
+ tf = input_tf->tf;
+
+ build_evenly_distributed_points(
+ axix_x,
+ ramp->num_entries,
+ dividers);
+
+ if (ramp->type == GAMMA_RGB_256 && mapUserRamp)
+ scale_gamma(rgb_user, ramp, dividers);
+ else if (ramp->type == GAMMA_RGB_FLOAT_1024)
+ scale_gamma_dx(rgb_user, ramp, dividers);
+
+ if (tf == TRANSFER_FUNCTION_PQ)
+ build_de_pq(curve,
+ MAX_HW_DEGAMMA_POINTS,
+ degamma_coordinates_x);
+ else
+ build_degamma(curve,
+ MAX_HW_DEGAMMA_POINTS,
+ degamma_coordinates_x,
+ tf == TRANSFER_FUNCTION_SRGB ? true:false);
+
+ tf_pts->end_exponent = 0;
+ tf_pts->x_point_at_y1_red = 1;
+ tf_pts->x_point_at_y1_green = 1;
+ tf_pts->x_point_at_y1_blue = 1;
+
+ map_regamma_hw_to_x_user(ramp, coeff, rgb_user,
+ degamma_coordinates_x, axix_x, curve,
+ MAX_HW_DEGAMMA_POINTS, tf_pts,
+ mapUserRamp);
+
+ ret = true;
+
+ kfree(coeff);
+coeff_alloc_fail:
+ kfree(axix_x);
+axix_x_alloc_fail:
+ kfree(curve);
+curve_alloc_fail:
+ kfree(rgb_user);
+rgb_user_alloc_fail:
+
+ return ret;
+
+}
+
+
bool mod_color_calculate_curve(enum dc_transfer_func_predefined trans,
struct dc_transfer_func_distributed_points *points)
{
@@ -1032,7 +1277,11 @@ bool mod_color_calculate_curve(enum dc_transfer_func_predefined trans,
struct pwl_float_data_ex *rgb_regamma = NULL;
if (trans == TRANSFER_FUNCTION_UNITY) {
- //setup_x_points_distribution(coordinates_x);
+ points->end_exponent = 0;
+ points->x_point_at_y1_red = 1;
+ points->x_point_at_y1_green = 1;
+ points->x_point_at_y1_blue = 1;
+
for (i = 0; i < MAX_HW_POINTS ; i++) {
points->red[i] = coordinates_x[i].x;
points->green[i] = coordinates_x[i].x;
@@ -1044,16 +1293,38 @@ bool mod_color_calculate_curve(enum dc_transfer_func_predefined trans,
_EXTRA_POINTS), GFP_KERNEL);
if (!rgb_regamma)
goto rgb_regamma_alloc_fail;
- //setup_x_points_distribution(coordinates_x);
+ points->end_exponent = 7;
+ points->x_point_at_y1_red = 125;
+ points->x_point_at_y1_green = 125;
+ points->x_point_at_y1_blue = 125;
+
+
+ build_pq(rgb_regamma,
+ MAX_HW_POINTS,
+ coordinates_x,
+ 80);
+ for (i = 0; i < MAX_HW_POINTS ; i++) {
+ points->red[i] = rgb_regamma[i].r;
+ points->green[i] = rgb_regamma[i].g;
+ points->blue[i] = rgb_regamma[i].b;
+ }
+ ret = true;
+
+ kfree(rgb_regamma);
+ } else if (trans == TRANSFER_FUNCTION_SRGB ||
+ trans == TRANSFER_FUNCTION_BT709) {
+ rgb_regamma = kzalloc(sizeof(*rgb_regamma) * (MAX_HW_POINTS +
+ _EXTRA_POINTS), GFP_KERNEL);
+ if (!rgb_regamma)
+ goto rgb_regamma_alloc_fail;
points->end_exponent = 0;
points->x_point_at_y1_red = 1;
points->x_point_at_y1_green = 1;
points->x_point_at_y1_blue = 1;
- build_regamma_curve_pq(rgb_regamma,
+ build_regamma(rgb_regamma,
MAX_HW_POINTS,
- coordinates_x,
- 80);
+ coordinates_x, trans == TRANSFER_FUNCTION_SRGB ? true:false);
for (i = 0; i < MAX_HW_POINTS ; i++) {
points->red[i] = rgb_regamma[i].r;
points->green[i] = rgb_regamma[i].g;
@@ -1068,3 +1339,65 @@ bool mod_color_calculate_curve(enum dc_transfer_func_predefined trans,
}
+bool mod_color_calculate_degamma_curve(enum dc_transfer_func_predefined trans,
+ struct dc_transfer_func_distributed_points *points)
+{
+ uint32_t i;
+ bool ret = false;
+ struct pwl_float_data_ex *rgb_degamma = NULL;
+
+ if (trans == TRANSFER_FUNCTION_UNITY) {
+
+ for (i = 0; i < MAX_HW_DEGAMMA_POINTS ; i++) {
+ points->red[i] = degamma_coordinates_x[i].x;
+ points->green[i] = degamma_coordinates_x[i].x;
+ points->blue[i] = degamma_coordinates_x[i].x;
+ }
+ ret = true;
+ } else if (trans == TRANSFER_FUNCTION_PQ) {
+ rgb_degamma = kzalloc(sizeof(*rgb_degamma) * (MAX_HW_DEGAMMA_POINTS +
+ _EXTRA_POINTS), GFP_KERNEL);
+ if (!rgb_degamma)
+ goto rgb_degamma_alloc_fail;
+
+
+ build_de_pq(rgb_degamma,
+ MAX_HW_DEGAMMA_POINTS,
+ degamma_coordinates_x);
+ for (i = 0; i < MAX_HW_DEGAMMA_POINTS ; i++) {
+ points->red[i] = rgb_degamma[i].r;
+ points->green[i] = rgb_degamma[i].g;
+ points->blue[i] = rgb_degamma[i].b;
+ }
+ ret = true;
+
+ kfree(rgb_degamma);
+ } else if (trans == TRANSFER_FUNCTION_SRGB ||
+ trans == TRANSFER_FUNCTION_BT709) {
+ rgb_degamma = kzalloc(sizeof(*rgb_degamma) * (MAX_HW_DEGAMMA_POINTS +
+ _EXTRA_POINTS), GFP_KERNEL);
+ if (!rgb_degamma)
+ goto rgb_degamma_alloc_fail;
+
+ build_degamma(rgb_degamma,
+ MAX_HW_DEGAMMA_POINTS,
+ degamma_coordinates_x, trans == TRANSFER_FUNCTION_SRGB ? true:false);
+ for (i = 0; i < MAX_HW_DEGAMMA_POINTS ; i++) {
+ points->red[i] = rgb_degamma[i].r;
+ points->green[i] = rgb_degamma[i].g;
+ points->blue[i] = rgb_degamma[i].b;
+ }
+ ret = true;
+
+ kfree(rgb_degamma);
+ }
+ points->end_exponent = 0;
+ points->x_point_at_y1_red = 1;
+ points->x_point_at_y1_green = 1;
+ points->x_point_at_y1_blue = 1;
+
+rgb_degamma_alloc_fail:
+ return ret;
+}
+
+
diff --git a/drivers/gpu/drm/amd/display/modules/color/color_gamma.h b/drivers/gpu/drm/amd/display/modules/color/color_gamma.h
index 774c6daa1689..b7f9bc27d101 100644
--- a/drivers/gpu/drm/amd/display/modules/color/color_gamma.h
+++ b/drivers/gpu/drm/amd/display/modules/color/color_gamma.h
@@ -34,12 +34,20 @@ enum dc_transfer_func_predefined;
void setup_x_points_distribution(void);
void precompute_pq(void);
+void precompute_de_pq(void);
bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf,
const struct dc_gamma *ramp, bool mapUserRamp);
+bool mod_color_calculate_degamma_params(struct dc_transfer_func *output_tf,
+ const struct dc_gamma *ramp, bool mapUserRamp);
+
bool mod_color_calculate_curve(enum dc_transfer_func_predefined trans,
struct dc_transfer_func_distributed_points *points);
+bool mod_color_calculate_degamma_curve(enum dc_transfer_func_predefined trans,
+ struct dc_transfer_func_distributed_points *points);
+
+
#endif /* COLOR_MOD_COLOR_GAMMA_H_ */
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread* [PATCH 34/34] drm/amd/display: Fix increment when sampling OTF in DCE
[not found] ` <20180212171625.14325-1-harry.wentland-5C7GfCeVMHo@public.gmane.org>
` (32 preceding siblings ...)
2018-02-12 17:16 ` [PATCH 33/34] drm/amd/display: De PQ implementation Harry Wentland
@ 2018-02-12 17:16 ` Harry Wentland
33 siblings, 0 replies; 38+ messages in thread
From: Harry Wentland @ 2018-02-12 17:16 UTC (permalink / raw)
To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW; +Cc: Leo (Sunpeng) Li
From: "Leo (Sunpeng) Li" <sunpeng.li@amd.com>
Previously, the number of software segmets per region was reduced to 16.
This needs to be reflected in the sampling distance (increment) used when
translating to the hardware format.
Change-Id: Ic1bf86d87ad36d9d4801472a114c45c95db4baf3
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
---
drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
index bc698644d42a..0422c72a7579 100644
--- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
@@ -481,7 +481,7 @@ dce110_translate_regamma_to_hw_format(const struct dc_transfer_func *output_tf,
j = 0;
for (k = 0; k < (region_end - region_start); k++) {
- increment = 32 / (1 << seg_distr[k]);
+ increment = NUMBER_SW_SEGMENTS / (1 << seg_distr[k]);
start_index = (region_start + k + MAX_LOW_POINT) *
NUMBER_SW_SEGMENTS;
for (i = start_index; i < start_index + NUMBER_SW_SEGMENTS;
--
2.14.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 38+ messages in thread