DC Patches - January 10, 2020
cover DC Patches - January 10, 2020
This DC patchset brings improvements in multiple areas. In summary, we highlight:
* Add OCSC improvements.
* Bug fix on chroma, Freesync dml, and DMUB.
* DMCUB improvements.
* Some code refactor.
Alvin Lee (3):
drm/amd/display: Fix 300Hz Freesync bug drm/amd/display: Don't always set pstate true if dummy latency = 0 drm/amd/display: Enable double buffer for OTG_BLANK
Amanda Liu (1):
drm/amd/display: Clear state after exiting fixed active VRR state
Anthony Koo (1):
drm/amd/display: make PSR static screen entry within 30 ms
Aric Cyr (4):
drm/amd/display: 3.2.65
drm/amd/display: 3.2.66
drm/amd/display: 3.2.67
drm/amd/display: 3.2.68
Charlene Liu (1):
drm/amd/display: rename _lvp to l_vp
Chris Park (1):
drm/amd/display: Add default switch case for DCC
Dmytro Laktyushkin (1):
drm/amd/display: expand dml structs
Eric Yang (1):
drm/amd/display: fix chroma vp wa corner case
George Shen (1):
drm/amd/display: Add w/a to reset PHY before link training in
verify_link_cap
Joseph Gravenor (1):
drm/amd/display: stop doing unnecessary detection when going to D3
Josip Pavic (1):
drm/amd/display: implement fw-driver interface for abm 2.4
Jun Lei (1):
drm/amd/display: fixup DML dependencies
Lewis Huang (2):
drm/amd/display: remove psr state condition when psr exit case drm/amd/display: Add monitor patch for AUO dpcd issue
Martin Leung (1):
drm/amd/display: Adding forgotten hubbub func
Martin Tsai (1):
drm/amd/display: Use udelay to avoid context switch
Michael Strauss (1):
drm/amd/display: Update HDMI hang w/a to apply to all TMDS signals
Nicholas Kazlauskas (3):
drm/amd/display: Flush framebuffer data before passing to DMCUB drm/amd/display: Read inst_fb data back during DMUB loading drm/amd/display: Soft reset DMUIF during DMUB reset
Nikola Cornij (1):
drm/amd/display: Disable secondary link for certain monitors
Noah Abradjian (4):
drm/amd/display: Add double buffering to dcn20 OCSC drm/amd/display: Fix double buffering in dcn2 ICSC drm/amd/display: Double buffer dcn2 Gamut Remap drm/amd/display: Indirect reg read macro with shift and mask
Paul Hsieh (1):
drm/amd/display: reallocate MST payload when link loss
Rodrigo Siqueira (1):
drm/amd/include: Add OCSC registers
Sung Lee (1):
drm/amd/display: Use SMU ClockTable Values for DML Calculations
Wenjing Liu (3):
drm/amd/display: wait for update when setting dpg test pattern drm/amd/display: wait for test pattern after when all pipes are
programmed
drm/amd/display: skip opp blank or unblank if test pattern enabled
Wyatt Wood (3):
drm/amd/display: Driverside changes to support PSR in DMCUB drm/amd/display: DMCUB FW Changes to support PSR drm/amd/display: Fix DMUB PSR command IDs
Yongqiang Sun (3):
drm/amd/display: programing surface flip by dmcub. drm/amd/display: Refactor surface flip programming drm/amd/display: Only program surface flip for video plane via dmcub
abdoulaye berthe (1):
drm/amd/display: store lttpr mode with dpcd
.../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 32 +-
.../amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c | 10 +-
drivers/gpu/drm/amd/display/dc/core/dc.c | 13 +-
drivers/gpu/drm/amd/display/dc/core/dc_link.c | 64 +++-
.../gpu/drm/amd/display/dc/core/dc_link_dp.c | 99 +++++-
.../gpu/drm/amd/display/dc/core/dc_resource.c | 10 +-
drivers/gpu/drm/amd/display/dc/dc.h | 20 +-
drivers/gpu/drm/amd/display/dc/dc_helper.c | 30 ++
drivers/gpu/drm/amd/display/dc/dc_link.h | 4 +
drivers/gpu/drm/amd/display/dc/dc_stream.h | 4 +-
drivers/gpu/drm/amd/display/dc/dc_types.h | 3 +-
drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c | 3 -
drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c | 220 +++++++++++++
drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h | 47 +++
.../display/dc/dce110/dce110_hw_sequencer.c | 35 +-
.../dc/dce110/dce110_timing_generator.c | 11 +-
.../dc/dce110/dce110_timing_generator.h | 3 +-
.../dc/dce120/dce120_timing_generator.c | 11 +-
.../drm/amd/display/dc/dcn10/dcn10_dpp_cm.c | 20 --
.../drm/amd/display/dc/dcn10/dcn10_hubbub.c | 3 +
.../amd/display/dc/dcn10/dcn10_hw_sequencer.c | 31 +-
.../amd/display/dc/dcn10/dcn10_hw_sequencer.h | 2 +-
.../gpu/drm/amd/display/dc/dcn10/dcn10_optc.c | 15 +-
.../gpu/drm/amd/display/dc/dcn10/dcn10_optc.h | 3 +-
.../gpu/drm/amd/display/dc/dcn20/dcn20_dpp.c | 26 +-
.../gpu/drm/amd/display/dc/dcn20/dcn20_dpp.h | 64 +++-
.../drm/amd/display/dc/dcn20/dcn20_dpp_cm.c | 155 ++++++++-
.../drm/amd/display/dc/dcn20/dcn20_hubbub.c | 6 +-
.../drm/amd/display/dc/dcn20/dcn20_hwseq.c | 13 +-
.../gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c | 47 ++-
.../gpu/drm/amd/display/dc/dcn20/dcn20_mpc.h | 22 ++
.../drm/amd/display/dc/dcn20/dcn20_resource.c | 53 +++-
.../gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c | 300 ++++++++++--------
.../drm/amd/display/dc/dcn21/dcn21_resource.c | 28 +-
.../dc/dml/dcn20/display_mode_vba_20.c | 169 +++++-----
.../dc/dml/dcn20/display_mode_vba_20v2.c | 171 +++++-----
.../dc/dml/dcn21/display_mode_vba_21.c | 161 +++++-----
.../amd/display/dc/dml/display_mode_enums.h | 18 +-
.../amd/display/dc/dml/display_mode_structs.h | 7 +-
.../drm/amd/display/dc/dml/display_mode_vba.c | 22 +-
.../drm/amd/display/dc/dml/display_mode_vba.h | 128 +++++---
.../drm/amd/display/dc/dml/dml_common_defs.c | 2 +-
.../drm/amd/display/dc/dml/dml_inline_defs.h | 2 +-
.../gpu/drm/amd/display/dc/inc/core_types.h | 1 +
.../gpu/drm/amd/display/dc/inc/dc_link_dp.h | 2 +
.../display/dc/{calcs => inc}/dcn_calc_math.h | 0
drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h | 20 ++
drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h | 20 ++
.../amd/display/dc/inc/hw/timing_generator.h | 3 +-
.../gpu/drm/amd/display/dc/inc/hw_sequencer.h | 4 +-
.../amd/display/dc/inc/hw_sequencer_private.h | 2 +-
.../gpu/drm/amd/display/dc/inc/reg_helper.h | 13 +
.../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 78 +++--
.../drm/amd/display/dmub/inc/dmub_cmd_dal.h | 15 +-
.../gpu/drm/amd/display/dmub/inc/dmub_rb.h | 29 +-
.../gpu/drm/amd/display/dmub/inc/dmub_srv.h | 19 --
.../gpu/drm/amd/display/dmub/src/dmub_dcn20.c | 2 +
.../gpu/drm/amd/display/dmub/src/dmub_dcn20.h | 6 +-
.../gpu/drm/amd/display/dmub/src/dmub_srv.c | 57 ++--
.../amd/display/modules/freesync/freesync.c | 3 +
.../amd/display/modules/power/power_helpers.c | 7 +-
.../include/asic_reg/dcn/dcn_2_0_0_offset.h | 4 +
.../include/asic_reg/dcn/dcn_2_0_0_sh_mask.h | 9 +-
.../include/asic_reg/dcn/dcn_2_1_0_offset.h | 5 +-
.../include/asic_reg/dcn/dcn_2_1_0_sh_mask.h | 8 +
65 files changed, 1743 insertions(+), 651 deletions(-)
create mode 100644 drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c
create mode 100644 drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h
rename drivers/gpu/drm/amd/display/dc/{calcs => inc}/dcn_calc_math.h (100%)Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 1/43 drm/amd/include: Add OCSC registers
Add registers for handling Post Gamma Color Blending (OCSC), which is useful for conversion from RGB->YUV for HDMI.
--- .../gpu/drm/amd/include/asic_reg/dcn/dcn_2_0_0_offset.h | 4 ++++ .../gpu/drm/amd/include/asic_reg/dcn/dcn_2_0_0_sh_mask.h | 9 ++++++++- .../gpu/drm/amd/include/asic_reg/dcn/dcn_2_1_0_offset.h | 5 ++++- .../gpu/drm/amd/include/asic_reg/dcn/dcn_2_1_0_sh_mask.h | 8 ++++++++ 4 files changed, 24 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_2_0_0_offset.h +4 -0
@@ -8134,6 +8134,10 @@#define mmMPC_OUT5_CSC_C33_C34_B 0x1604#define mmMPC_OUT5_CSC_C33_C34_B_BASE_IDX 2 +#define mmMPC_OCSC_TEST_DEBUG_INDEX 0x163b +#define mmMPC_OCSC_TEST_DEBUG_INDEX_BASE_IDX 2 +#define mmMPC_OCSC_TEST_DEBUG_DATA_BASE_IDX 2 +#define mmMPC_OCSC_TEST_DEBUG_DATA 0x163c// addressBlock: dce_dc_mpc_mpc_dcperfmon_dc_perfmon_dispdec // base address: 0x5964
drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_2_0_0_sh_mask.h +8 -1
@@ -28263,7 +28263,14 @@#define MPC_OUT5_CSC_C33_C34_B__MPC_OCSC_C34_B__SHIFT 0x10#define MPC_OUT5_CSC_C33_C34_B__MPC_OCSC_C33_B_MASK 0x0000FFFFL#define MPC_OUT5_CSC_C33_C34_B__MPC_OCSC_C34_B_MASK 0xFFFF0000L - +//MPC_OCSC_TEST_DEBUG_INDEX +#define MPC_OCSC_TEST_DEBUG_INDEX__MPC_OCSC_TEST_DEBUG_INDEX__SHIFT 0x0 +#define MPC_OCSC_TEST_DEBUG_INDEX__MPC_OCSC_TEST_DEBUG_WRITE_EN__SHIFT 0x8 +#define MPC_OCSC_TEST_DEBUG_INDEX__MPC_OCSC_TEST_DEBUG_INDEX_MASK 0x000000FFL +#define MPC_OCSC_TEST_DEBUG_INDEX__MPC_OCSC_TEST_DEBUG_WRITE_EN_MASK 0x00000100L +//MPC_OCSC_TEST_DEBUG_DATA +#define MPC_OCSC_TEST_DEBUG_DATA__MPC_OCSC_TEST_DEBUG_DATA__SHIFT 0x0 +#define MPC_OCSC_TEST_DEBUG_DATA__MPC_OCSC_TEST_DEBUG_DATA_MASK 0xFFFFFFFFL// addressBlock: dce_dc_mpc_mpc_dcperfmon_dc_perfmon_dispdec //DC_PERFMON17_PERFCOUNTER_CNTL
drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_2_1_0_offset.h +4 -1
@@ -7103,7 +7103,10 @@#define mmMPC_OUT3_CSC_C31_C32_B_BASE_IDX 2#define mmMPC_OUT3_CSC_C33_C34_B 0x15ea#define mmMPC_OUT3_CSC_C33_C34_B_BASE_IDX 2 - +#define mmMPC_OCSC_TEST_DEBUG_INDEX 0x163b +#define mmMPC_OCSC_TEST_DEBUG_INDEX_BASE_IDX 2 +#define mmMPC_OCSC_TEST_DEBUG_DATA_BASE_IDX 2 +#define mmMPC_OCSC_TEST_DEBUG_DATA 0x163c// addressBlock: dce_dc_mpc_mpc_dcperfmon_dc_perfmon_dispdec // base address: 0x5964
drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_2_1_0_sh_mask.h +8 -0
@@ -56634,5 +56634,13 @@#define AZF0INPUTENDPOINT7_AZALIA_F0_CODEC_INPUT_PIN_CONTROL_INFOFRAME__INFOFRAME_BYTE_5_MASK 0x00FF0000L#define AZF0INPUTENDPOINT7_AZALIA_F0_CODEC_INPUT_PIN_CONTROL_INFOFRAME__INFOFRAME_VALID_MASK 0x80000000L +//MPC_OCSC_TEST_DEBUG_INDEX +#define MPC_OCSC_TEST_DEBUG_INDEX__MPC_OCSC_TEST_DEBUG_INDEX__SHIFT 0x0 +#define MPC_OCSC_TEST_DEBUG_INDEX__MPC_OCSC_TEST_DEBUG_WRITE_EN__SHIFT 0x8 +#define MPC_OCSC_TEST_DEBUG_INDEX__MPC_OCSC_TEST_DEBUG_INDEX_MASK 0x000000FFL +#define MPC_OCSC_TEST_DEBUG_INDEX__MPC_OCSC_TEST_DEBUG_WRITE_EN_MASK 0x00000100L +//MPC_OCSC_TEST_DEBUG_DATA +#define MPC_OCSC_TEST_DEBUG_DATA__MPC_OCSC_TEST_DEBUG_DATA__SHIFT 0x0 +#define MPC_OCSC_TEST_DEBUG_DATA__MPC_OCSC_TEST_DEBUG_DATA_MASK 0xFFFFFFFFL#endif
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 2/43 drm/amd/display: Clear state after exiting fixed active VRR state
In reply to Rodrigo SiqueiraFrom: Amanda Liu <[email protected]>
[why]
Upon exiting a fixed active VRR state, the state isn't cleared. This leads to the variable VRR range to be calculated incorrectly.
[how]
Set fixed active state to false when updating vrr params
--- drivers/gpu/drm/amd/display/modules/freesync/freesync.c | 2 ++ 1 file changed, 2 insertions(+)
drivers/gpu/drm/amd/display/modules/freesync/freesync.c +2 -0
@@ -812,6 +812,7 @@ void mod_freesync_build_vrr_params(struct mod_freesync *mod_freesync, 2 * in_out_vrr->min_refresh_in_uhz) in_out_vrr->btr.btr_enabled = false; + in_out_vrr->fixed.fixed_active = false; in_out_vrr->btr.btr_active = false; in_out_vrr->btr.inserted_duration_in_us =0 ; in_out_vrr->btr.frames_to_insert =0 ;
@@ -832,6 +833,7 @@ void mod_freesync_build_vrr_params(struct mod_freesync *mod_freesync, in_out_vrr->adjust.v_total_max = stream->timing.v_total; }else if (in_out_vrr->state ==VRR_STATE_ACTIVE_VARIABLE && refresh_range >=MIN_REFRESH_RANGE_IN_US ) { + in_out_vrr->adjust.v_total_min =calc_v_total_from_refresh (stream, in_out_vrr->max_refresh_in_uhz);
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 3/43 drm/amd/display: 3.2.65
In reply to Rodrigo SiqueiraFrom: Aric Cyr <[email protected]>
--- drivers/gpu/drm/amd/display/dc/dc.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/dc.h +1 -1
@@ -39,7 +39,7 @@#include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#define DC_VER "3.2.64" +#define DC_VER "3.2.65"#define MAX_SURFACES 3#define MAX_PLANES 6
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 4/43 drm/amd/display: fix chroma vp wa corner case
In reply to Rodrigo SiqueiraFrom: Eric Yang <[email protected]>
[Why]
Previous implementation we may have residual chroma address offset if transition from wa enable -> wa disable.
[How]
Clear address offset cache when viewport updates. Also update the vp size check condition to account for rotation angle
--- .../gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c +13 -3
@@ -251,10 +251,20 @@ static void hubp21_apply_PLAT_54186_wa(ROTATION_ANGLE , &rotation_angle, H_MIRROR_EN, &h_mirror_en); -/* apply wa only for NV12 surface with scatter gather enabled with view port > 512 */ +/* reset persistent cached data */ + hubp21->PLAT_54186_wa_chroma_addr_offset =0 ; +/* apply wa only for NV12 surface with scatter gather enabled with viewport > 512 along +* the vertical direction*/ if (address->type !=PLN_ADDR_TYPE_VIDEO_PROGRESSIVE || - address->video_progressive.luma_addr.high_part ==0xf4 - || viewport_c_height <=512 ) + address->video_progressive.luma_addr.high_part ==0xf4 ) +return ; + +if ((rotation_angle ==0 || rotation_angle ==180 ) + && viewport_c_height <=512 ) +return ; + +if ((rotation_angle ==90 || rotation_angle ==270 ) + && viewport_c_width <=512 )return ;switch (rotation_angle) {
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 5/43 drm/amd/display: Fix 300Hz Freesync bug
In reply to Rodrigo SiqueiraFrom: Alvin Lee <[email protected]>
Needed to reprogram vblank_start in dml properly in order to get the correct dlg params to program VTG.
--- drivers/gpu/drm/amd/display/dc/core/dc.c | 1 - drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c | 4 +++- drivers/gpu/drm/amd/display/modules/freesync/freesync.c | 3 ++- 3 files changed, 5 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/display/dc/core/dc.c +0 -1
@@ -287,7 +287,6 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc,struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];if (pipe->stream == stream && pipe->stream_res.tg) { - pipe->stream->adjust = *adjust; dc->hwss.set_drr(&pipe, 1, adjust->v_total_min,
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +3 -1
@@ -1883,12 +1883,14 @@ int dcn20_populate_dml_pipes_from_context(for (i =0 , pipe_cnt =0 ; i < dc->res_pool->pipe_count; i++) {struct dc_crtc_timing *timing = &res_ctx->pipe_ctx[i].stream->timing;unsigned int v_total; +unsigned int front_porch;int output_bpc;if (!res_ctx->pipe_ctx[i].stream)continue ; v_total = timing->v_total; + front_porch = timing->v_front_porch;/* todo: pipes[pipe_cnt].pipe.src.dynamic_metadata_enable = 0; pipes[pipe_cnt].pipe.src.dcc = 0;
@@ -1915,7 +1917,7 @@ int dcn20_populate_dml_pipes_from_context( - timing->h_addressable - timing->h_border_left - timing->h_border_right; - pipes[pipe_cnt].pipe.dest.vblank_start = v_total - timing->v_front_porch; + pipes[pipe_cnt].pipe.dest.vblank_start = v_total - front_porch; pipes[pipe_cnt].pipe.dest.vblank_end = pipes[pipe_cnt].pipe.dest.vblank_start - timing->v_addressable - timing->v_border_top
drivers/gpu/drm/amd/display/modules/freesync/freesync.c +2 -1
@@ -812,11 +812,12 @@ void mod_freesync_build_vrr_params(struct mod_freesync *mod_freesync, 2 * in_out_vrr->min_refresh_in_uhz) in_out_vrr->btr.btr_enabled = false; - in_out_vrr->fixed.fixed_active = false; in_out_vrr->btr.btr_active = false; in_out_vrr->btr.inserted_duration_in_us =0 ; in_out_vrr->btr.frames_to_insert =0 ; in_out_vrr->btr.frame_counter =0 ; + in_out_vrr->fixed.fixed_active = false; + in_out_vrr->fixed.target_refresh_in_uhz =0 ; in_out_vrr->btr.mid_point_in_us = (in_out_vrr->min_duration_in_us +
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 6/43 drm/amd/display: Disable secondary link for certain monitors
In reply to Rodrigo SiqueiraFrom: Nikola Cornij <[email protected]>
[why]
If the specific monitor supports DSC, the secondary link should be disabled, and the other way around, too: if either that monitor or our ASIC doesn't support DSC, the secodary link should be enabled.
[how]
Add a monitor patch and disable secondary link if that monitor is detected and if ASIC supports DSC, or otherwise enable secondary link.
--- drivers/gpu/drm/amd/display/dc/core/dc_resource.c | 10 +++++++--- drivers/gpu/drm/amd/display/dc/dc.h | 2 ++ drivers/gpu/drm/amd/display/dc/dc_types.h | 1 + 3 files changed, 10 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/display/dc/core/dc_resource.c +7 -3
@@ -2066,6 +2066,13 @@ void dc_resource_state_construct( dst_ctx->clk_mgr = dc->clk_mgr; } + +bool dc_resource_is_dsc_encoding_supported (const struct dc *dc) +{ +return dc->res_pool->res_cap->num_dsc >0 ; +} + +/** * dc_validate_global_state() - Determine if HW can support a given state * Checks HW resource availability and bandwidth requirement.
@@ -2897,6 +2904,3 @@ void get_audio_check(struct audio_info *aud_modes, } } - - -
drivers/gpu/drm/amd/display/dc/dc.h +2 -0
@@ -910,6 +910,8 @@ void dc_resource_state_copy_construct_current(void dc_resource_state_destruct (struct dc_state *context); +bool dc_resource_is_dsc_encoding_supported (const struct dc *dc); +/* * TODO update to make it about validation sets * Set up streams and links associated to drive sinks
drivers/gpu/drm/amd/display/dc/dc_types.h +1 -0
@@ -229,6 +229,7 @@ struct dc_panel_patch {unsigned int extra_t12_ms;unsigned int extra_delay_backlight_off;unsigned int extra_t7_ms; +unsigned int manage_secondary_link; };struct dc_edid_caps {
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 7/43 drm/amd/display: Adding forgotten hubbub func
In reply to Rodrigo SiqueiraFrom: Martin Leung <[email protected]>
[why]
While doing seamless boot I made some changes to dcn2 hubbub functions, missed a link
[how]
link hubbub1 func to hubbub2 usage. It has already been successfully linked in dcn1 and 3.
--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubbub.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubbub.c +2 -1
@@ -601,7 +601,8 @@ static const struct hubbub_funcs hubbub2_funcs = { .wm_read_state = hubbub2_wm_read_state, .get_dchub_ref_freq = hubbub2_get_dchub_ref_freq, .program_watermarks = hubbub2_program_watermarks, - .allow_self_refresh_control = hubbub1_allow_self_refresh_control + .is_allow_self_refresh_enabled = hubbub1_is_allow_self_refresh_enabled, + .allow_self_refresh_control = hubbub1_allow_self_refresh_control, }; void hubbub2_construct(struct dcn20_hubbub *hubbub,
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 8/43 drm/amd/display: Flush framebuffer data before passing to DMCUB
In reply to Rodrigo SiqueiraFrom: Nicholas Kazlauskas <[email protected]>
[Why]
There's a data race that can occur between when we update the inbox write pointer vs when the memory for the command actually gets flushed from the map to the framebuffer.
DMCUB can read stale or partially invalid data when this race occurs.
[How]
Before updating the write pointer we can read back all pending commands to ensure that we stall for the writes to be flushed to framebuffer.
We don't need to worry about choosing HDP vs VM flush with this mechanism.
Drop the dmub_srv_cmd_submit() while we're updating this to work correctly since nothing was actually using this API and the caller should be explicit about the API flow here - by doing this on execute we can give some extra time for the flush to finish while preparing other commands.
We should try to avoid writing single commands
because of this overhead.
--- .../gpu/drm/amd/display/dmub/inc/dmub_rb.h | 17 +++++++++++ .../gpu/drm/amd/display/dmub/inc/dmub_srv.h | 19 ------------ .../gpu/drm/amd/display/dmub/src/dmub_srv.c | 30 +++++-------------- 3 files changed, 24 insertions(+), 42 deletions(-)
drivers/gpu/drm/amd/display/dmub/inc/dmub_rb.h +17 -0
@@ -113,6 +113,23 @@ static inline bool dmub_rb_pop_front(struct dmub_rb *rb)return true; } +static inline void dmub_rb_flush_pending (const struct dmub_rb *rb) +{ +uint8_t buf[DMUB_RB_CMD_SIZE ]; +uint32_t rptr = rb->rptr; +uint32_t wptr = rb->wrpt; + +while (rptr != wptr) { +const uint8_t *data = (const uint8_t *)rb->base_address + rptr; + +dmub_memcpy (buf, data,DMUB_RB_CMD_SIZE ); + + rptr +=DMUB_RB_CMD_SIZE ; +if (rptr >= rb->capacity) + rptr %= rb->capacity; + } +} +static inline void dmub_rb_init (struct dmub_rb *rb,struct dmub_rb_init_params *init_params) {
drivers/gpu/drm/amd/display/dmub/inc/dmub_srv.h +0 -19
@@ -444,25 +444,6 @@ enum dmub_status dmub_srv_cmd_queue(struct dmub_srv *dmub,*/ enum dmub_status dmub_srv_cmd_execute (struct dmub_srv *dmub); -/** -* dmub_srv_cmd_submit() - submits a command to the DMUB immediately -* @dmub: the dmub service -* @cmd: the command to submit -* @timeout_us: the maximum number of microseconds to wait -* -* Submits a command to the DMUB with an optional timeout. -* If timeout_us is given then the service will attempt to -* resubmit for the given number of microseconds. -* -* Return: -* DMUB_STATUS_OK - success -* DMUB_STATUS_TIMEOUT - wait for submit timed out -* DMUB_STATUS_INVALID - unspecified error -*/ -enum dmub_status dmub_srv_cmd_submit (struct dmub_srv *dmub, -const struct dmub_cmd_header *cmd, -uint32_t timeout_us); -/** * dmub_srv_wait_for_auto_load() - Waits for firmware auto load to complete * @dmub: the dmub service
drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c +7 -23
@@ -405,33 +405,17 @@ enum dmub_status dmub_srv_cmd_execute(struct dmub_srv *dmub)if (!dmub->hw_init)return DMUB_STATUS_INVALID ; +/** +* Read back all the queued commands to ensure that they've +* been flushed to framebuffer memory. Otherwise DMCUB might +* read back stale, fully invalid or partially invalid data. +*/ +dmub_rb_flush_pending (&dmub->inbox1_rb); + dmub->hw_funcs.set_inbox1_wptr (dmub, dmub->inbox1_rb.wrpt);return DMUB_STATUS_OK ; } -enum dmub_status dmub_srv_cmd_submit (struct dmub_srv *dmub, -const struct dmub_cmd_header *cmd, -uint32_t timeout_us) -{ -uint32_t i =0 ; - -if (!dmub->hw_init) -return DMUB_STATUS_INVALID ; - -for (i =0 ; i <= timeout_us; ++i) { - dmub->inbox1_rb.rptr = dmub->hw_funcs.get_inbox1_rptr (dmub); -if (dmub_rb_push_front (&dmub->inbox1_rb, cmd)) { - dmub->hw_funcs.set_inbox1_wptr (dmub, - dmub->inbox1_rb.wrpt); -return DMUB_STATUS_OK ; - } - -udelay (1 ); - } - -return DMUB_STATUS_TIMEOUT ; -} -enum dmub_status dmub_srv_wait_for_auto_load (struct dmub_srv *dmub,uint32_t timeout_us) {
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 9/43 drm/amd/display: Read inst_fb data back during DMUB loading
In reply to Rodrigo SiqueiraFrom: Nicholas Kazlauskas <[email protected]>
[Why]
The inst firmware isn't necessarily fully flushed to framebuffer memory and the DMCUB can hang due to invalid inst being parsed.
[How]
Like the fix to flush ringbuffer commands before updating the inbox write pointer we need to read back inst memory before writing the CW0 registers.
Add a helper to read 64-byte chunks to avoid a large temporary buffer. Read the remaining leftover bytes if the inst_fb isn't fully aligned.
--- .../gpu/drm/amd/display/dmub/src/dmub_srv.c | 27 +++++++++++++++++++ 1 file changed, 27 insertions(+)
drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c +27 -0
@@ -67,6 +67,26 @@ static inline uint32_t dmub_align(uint32_t val, uint32_t factor)return (val + factor -1 ) / factor * factor; } +static void dmub_flush_buffer_mem (const struct dmub_fb *fb) +{ +const uint8_t *base = (const uint8_t *)fb->cpu_addr; +uint8_t buf[64 ]; +uint32_t pos, end; + +/** +* Read 64-byte chunks since we don't want to store a +* large temporary buffer for this purpose. +*/ + end = fb->size /sizeof (buf) *sizeof (buf); + +for (pos =0 ; pos < end; pos +=sizeof (buf)) +dmub_memcpy (buf, base + pos,sizeof (buf)); + +/* Read anything leftover into the buffer. */ +if (end < fb->size) +dmub_memcpy (buf, base + pos, fb->size - end); +} +static const struct dmub_fw_meta_info *dmub_get_fw_meta_info (const uint8_t *fw_bss_data,uint32_t fw_bss_data_size) {
@@ -329,6 +349,13 @@ enum dmub_status dmub_srv_hw_init(struct dmub_srv *dmub, cw1.region.base =DMUB_CW1_BASE ; cw1.region.top = cw1.region.base + stack_fb->size -1 ; +/** +* Read back all the instruction memory so we don't hang the +* DMCUB when backdoor loading if the write from x86 hasn't been +* flushed yet. This only occurs in backdoor loading. +*/ +dmub_flush_buffer_mem (inst_fb); +if (params->load_inst_const && dmub->hw_funcs.backdoor_load) dmub->hw_funcs.backdoor_load (dmub, &cw0, &cw1); }
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 10/43 drm/amd/display: Soft reset DMUIF during DMUB reset
In reply to Rodrigo SiqueiraFrom: Nicholas Kazlauskas <[email protected]>
[Why]
We need to ensure that the DMUIF in MMHUBBUB is also in reset so we aren't generating requests while the DMCUB is in reset.
[How]
Set DMUIF_SOFT_RESET=1 on reset and DMUIF_SOFT_RESET=0 on reset release.
--- drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c | 2 ++ drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.h | 6 ++++-- 2 files changed, 6 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c +2 -0
@@ -66,10 +66,12 @@ void dmub_dcn20_reset(struct dmub_srv *dmub) {REG_UPDATE (DMCUB_CNTL ,DMCUB_SOFT_RESET ,1 );REG_UPDATE (DMCUB_CNTL ,DMCUB_ENABLE ,0 ); +REG_UPDATE (MMHUBBUB_SOFT_RESET ,DMUIF_SOFT_RESET ,1 ); }void dmub_dcn20_reset_release (struct dmub_srv *dmub) { +REG_UPDATE (MMHUBBUB_SOFT_RESET ,DMUIF_SOFT_RESET ,0 );REG_WRITE (DMCUB_SCRATCH15 , dmub->psp_version &0x001100FF );REG_UPDATE_2 (DMCUB_CNTL ,DMCUB_ENABLE ,1 ,DMCUB_TRACEPORT_EN ,1 );REG_UPDATE (DMCUB_CNTL ,DMCUB_SOFT_RESET ,0 );
drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.h +4 -2
@@ -91,7 +91,8 @@ struct dmub_srv; DMUB_SR(DMCUB_SCRATCH13) \ DMUB_SR(DMCUB_SCRATCH14) \ DMUB_SR(DMCUB_SCRATCH15) \ - DMUB_SR(CC_DC_PIPE_DIS) + DMUB_SR(CC_DC_PIPE_DIS) \ + DMUB_SR(MMHUBBUB_SOFT_RESET) #define DMUB_COMMON_FIELDS() \ DMUB_SF(DMCUB_CNTL, DMCUB_ENABLE) \
@@ -119,7 +120,8 @@ struct dmub_srv; DMUB_SF(DMCUB_REGION3_CW7_TOP_ADDRESS, DMCUB_REGION3_CW7_ENABLE) \ DMUB_SF(DMCUB_REGION4_TOP_ADDRESS, DMCUB_REGION4_TOP_ADDRESS) \ DMUB_SF(DMCUB_REGION4_TOP_ADDRESS, DMCUB_REGION4_ENABLE) \ - DMUB_SF(CC_DC_PIPE_DIS, DC_DMCUB_ENABLE) + DMUB_SF(CC_DC_PIPE_DIS, DC_DMCUB_ENABLE) \ + DMUB_SF(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET)struct dmub_srv_common_reg_offset {#define DMUB_SR (reg) uint32_t reg;
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 11/43 drm/amd/display: Add double buffering to dcn20 OCSC
In reply to Rodrigo SiqueiraFrom: Noah Abradjian <[email protected]>
[Why]
When rapidly adjusting colour properties (e.g. brightness), screen tearing was observed. This was due to overwritten values in OCSC registers. In dcn10, this issue had been fixed by implementing double buffering by alternating OCSC modes.
[How]
Alternate which OCSC registers are used by switching modes each time. This double buffers the CSC writes.
--- .../gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c | 45 ++++++++++++++++--- .../gpu/drm/amd/display/dc/dcn20/dcn20_mpc.h | 16 +++++++ .../drm/amd/display/dc/dcn20/dcn20_resource.c | 1 + .../drm/amd/display/dc/dcn21/dcn21_resource.c | 3 +- 4 files changed, 59 insertions(+), 6 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c +40 -5
@@ -33,6 +33,9 @@#define REG (reg)\ mpc20->mpc_regs->reg +#define IND_REG (index) \ + (index) +#define CTX \ mpc20->base.ctx
@@ -132,19 +135,32 @@ void mpc2_set_output_csc(const uint16_t *regval, enum mpc_output_csc_mode ocsc_mode) { + uint32_t cur_mode;struct dcn20_mpc *mpc20 =TO_DCN20_MPC (mpc);struct color_matrices_reg ocsc_regs; -REG_SET (CSC_MODE [opp_id],0 ,MPC_OCSC_MODE , ocsc_mode); - -if (ocsc_mode ==MPC_OUTPUT_CSC_DISABLE ) +if (ocsc_mode ==MPC_OUTPUT_CSC_DISABLE ) { +REG_SET (CSC_MODE [opp_id],0 ,MPC_OCSC_MODE , ocsc_mode);return ; + }if (regval ==NULL ) {BREAK_TO_DEBUGGER ();return ; } +/* determine which CSC coefficients (A or B) we are using +* currently. select the alternate set to double buffer +* the CSC update so CSC is updated on frame boundary +*/ + cur_mode =IX_REG_READ (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_DATA , +MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_IDX ); + +if (cur_mode !=MPC_OUTPUT_CSC_COEF_A ) + ocsc_mode =MPC_OUTPUT_CSC_COEF_A ; +else + ocsc_mode =MPC_OUTPUT_CSC_COEF_B ; + ocsc_regs.shifts.csc_c11 = mpc20->mpc_shift->MPC_OCSC_C11_A; ocsc_regs.masks.csc_c11 = mpc20->mpc_mask->MPC_OCSC_C11_A; ocsc_regs.shifts.csc_c12 = mpc20->mpc_shift->MPC_OCSC_C12_A;
@@ -157,10 +173,13 @@ void mpc2_set_output_csc( ocsc_regs.csc_c11_c12 =REG (CSC_C11_C12_B [opp_id]); ocsc_regs.csc_c33_c34 =REG (CSC_C33_C34_B [opp_id]); } +cm_helper_program_color_matrices ( mpc20->base.ctx, regval, &ocsc_regs); + +REG_SET (CSC_MODE [opp_id],0 ,MPC_OCSC_MODE , ocsc_mode); } void mpc2_set_ocsc_default(
@@ -169,14 +188,16 @@ void mpc2_set_ocsc_default(enum dc_color_space color_space, enum mpc_output_csc_mode ocsc_mode) { + uint32_t cur_mode;struct dcn20_mpc *mpc20 =TO_DCN20_MPC (mpc);uint32_t arr_size;struct color_matrices_reg ocsc_regs;const uint16_t *regval =NULL ; -REG_SET (CSC_MODE [opp_id],0 ,MPC_OCSC_MODE , ocsc_mode); -if (ocsc_mode ==MPC_OUTPUT_CSC_DISABLE ) +if (ocsc_mode ==MPC_OUTPUT_CSC_DISABLE ) { +REG_SET (CSC_MODE [opp_id],0 ,MPC_OCSC_MODE , ocsc_mode);return ; + } regval =find_color_matrix (color_space, &arr_size);
@@ -185,6 +206,18 @@ void mpc2_set_ocsc_default(return ; } +/* determine which CSC coefficients (A or B) we are using +* currently. select the alternate set to double buffer +* the CSC update so CSC is updated on frame boundary +*/ + cur_mode =IX_REG_READ (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_DATA , +MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_IDX ); + +if (cur_mode !=MPC_OUTPUT_CSC_COEF_A ) + ocsc_mode =MPC_OUTPUT_CSC_COEF_A ; +else + ocsc_mode =MPC_OUTPUT_CSC_COEF_B ; + ocsc_regs.shifts.csc_c11 = mpc20->mpc_shift->MPC_OCSC_C11_A; ocsc_regs.masks.csc_c11 = mpc20->mpc_mask->MPC_OCSC_C11_A; ocsc_regs.shifts.csc_c12 = mpc20->mpc_shift->MPC_OCSC_C12_A;
@@ -203,6 +236,8 @@ void mpc2_set_ocsc_default( mpc20->base.ctx, regval, &ocsc_regs); + +REG_SET (CSC_MODE [opp_id],0 ,MPC_OCSC_MODE , ocsc_mode); } static void mpc2_ogam_get_reg_field(
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.h +16 -0
@@ -80,6 +80,10 @@SRII (DENORM_CLAMP_G_Y ,MPC_OUT , inst),\ SRII(DENORM_CLAMP_B_CB, MPC_OUT, inst) +#define MPC_DBG_REG_LIST_DCN2_0 () \ + SR(MPC_OCSC_TEST_DEBUG_DATA),\ + SR(MPC_OCSC_TEST_DEBUG_INDEX) +#define MPC_REG_VARIABLE_LIST_DCN2_0 \ MPC_COMMON_REG_VARIABLE_LIST \ uint32_t MPCC_TOP_GAIN[MAX_MPCC]; \
@@ -118,6 +122,8 @@uint32_t MPCC_OGAM_LUT_RAM_CONTROL [MAX_MPCC ];\uint32_t MPCC_OGAM_LUT_DATA [MAX_MPCC ];\uint32_t MPCC_OGAM_MODE [MAX_MPCC ];\ +uint32_t MPC_OCSC_TEST_DEBUG_DATA ;\ +uint32_t MPC_OCSC_TEST_DEBUG_INDEX ;\uint32_t CSC_MODE [MAX_OPP ]; \uint32_t CSC_C11_C12_A [MAX_OPP ]; \uint32_t CSC_C33_C34_A [MAX_OPP ]; \
@@ -134,6 +140,8 @@SF (MPCC0_MPCC_TOP_GAIN ,MPCC_TOP_GAIN , mask_sh),\SF (MPCC0_MPCC_BOT_GAIN_INSIDE ,MPCC_BOT_GAIN_INSIDE , mask_sh),\SF (MPCC0_MPCC_BOT_GAIN_OUTSIDE ,MPCC_BOT_GAIN_OUTSIDE , mask_sh),\ +SF (MPC_OCSC_TEST_DEBUG_DATA ,MPC_OCSC_TEST_DEBUG_DATA , mask_sh),\ +SF (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_INDEX , mask_sh),\SF (MPC_OUT0_CSC_MODE ,MPC_OCSC_MODE , mask_sh),\SF (MPC_OUT0_CSC_C11_C12_A ,MPC_OCSC_C11_A , mask_sh),\ SF(MPC_OUT0_CSC_C11_C12_A, MPC_OCSC_C12_A, mask_sh),\
@@ -174,6 +182,12 @@SF (MPC_OUT0_DENORM_CLAMP_B_CB ,MPC_OUT_DENORM_CLAMP_MAX_B_CB , mask_sh),\SF (MPC_OUT0_DENORM_CLAMP_B_CB ,MPC_OUT_DENORM_CLAMP_MIN_B_CB , mask_sh) +/* +* DCN2 MPC_OCSC debug status register: +* +* Field describing current OCSC Mode has index 1 [1..0] + */ +#define MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_IDX 1#define MPC_REG_FIELD_LIST_DCN2_0 (type) \ MPC_REG_FIELD_LIST(type)\
@@ -182,6 +196,8 @@type MPCC_TOP_GAIN ;\type MPCC_BOT_GAIN_INSIDE ;\type MPCC_BOT_GAIN_OUTSIDE ;\ +type MPC_OCSC_TEST_DEBUG_DATA ;\ +type MPC_OCSC_TEST_DEBUG_INDEX ;\type MPC_OCSC_MODE ;\type MPC_OCSC_C11_A ;\type MPC_OCSC_C12_A ;\
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +1 -0
@@ -706,6 +706,7 @@ static const struct dcn20_mpc_registers mpc_regs = {MPC_OUT_MUX_REG_LIST_DCN2_0 (3 ),MPC_OUT_MUX_REG_LIST_DCN2_0 (4 ),MPC_OUT_MUX_REG_LIST_DCN2_0 (5 ), +MPC_DBG_REG_LIST_DCN2_0 () }; static const struct dcn20_mpc_shift mpc_shift = {
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +2 -1
@@ -465,7 +465,8 @@ static const struct dcn20_mpc_registers mpc_regs = {MPC_OUT_MUX_REG_LIST_DCN2_0 (0 ),MPC_OUT_MUX_REG_LIST_DCN2_0 (1 ),MPC_OUT_MUX_REG_LIST_DCN2_0 (2 ), -MPC_OUT_MUX_REG_LIST_DCN2_0 (3 ) +MPC_OUT_MUX_REG_LIST_DCN2_0 (3 ), +MPC_DBG_REG_LIST_DCN2_0 () }; static const struct dcn20_mpc_shift mpc_shift = {
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 12/43 drm/amd/display: store lttpr mode with dpcd
In reply to Rodrigo SiqueiraFrom: abdoulaye berthe <[email protected]>
Make sure that lttpr_caps has the mode set to repeater.
--- drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +11 -2
@@ -1217,24 +1217,33 @@ static void configure_lttpr_mode(struct dc_link *link)uint8_t repeater_cnt;uint32_t aux_interval_address;uint8_t repeater_id; +enum dc_status result =DC_ERROR_UNEXPECTED ;uint8_t repeater_mode =DP_PHY_REPEATER_MODE_TRANSPARENT ;DC_LOG_HW_LINK_TRAINING ("%s\n Set LTTPR to Transparent Mode\n" , __func__); -core_link_write_dpcd (link, + result =core_link_write_dpcd (link,DP_PHY_REPEATER_MODE , (uint8_t *)&repeater_mode,sizeof (repeater_mode)); +if (result ==DC_OK ) { + link->dpcd_caps.lttpr_caps.mode = repeater_mode; + } +if (!link->is_lttpr_mode_transparent) {DC_LOG_HW_LINK_TRAINING ("%s\n Set LTTPR to Non Transparent Mode\n" , __func__); repeater_mode =DP_PHY_REPEATER_MODE_NON_TRANSPARENT ; -core_link_write_dpcd (link, + result =core_link_write_dpcd (link,DP_PHY_REPEATER_MODE , (uint8_t *)&repeater_mode,sizeof (repeater_mode)); +if (result ==DC_OK ) { + link->dpcd_caps.lttpr_caps.mode = repeater_mode; + } + repeater_cnt =convert_to_count (link->dpcd_caps.lttpr_caps.phy_repeater_cnt);for (repeater_id = repeater_cnt; repeater_id >0 ; repeater_id--) { aux_interval_address = DP_TRAINING_AUX_RD_INTERVAL_PHY_REPEATER1 +
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 13/43 drm/amd/display: Don't always set pstate true if dummy latency = 0
In reply to Rodrigo SiqueiraFrom: Alvin Lee <[email protected]>
[Why]
If dummy pstate latency is 0 we should report the actual pstate support, and not that pstate is always supported.
[How]
Don't always program pstate support true
--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +1 -1
@@ -2918,7 +2918,7 @@ bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context,if (context->bw_ctx.dml.soc.dummy_pstate_latency_us ==0 || (voltage_supported && full_pstate_supported)) { - context->bw_ctx.bw.dcn.clk.p_state_change_support = true; + context->bw_ctx.bw.dcn.clk.p_state_change_support = full_pstate_supported; goto restore_dml_state; }
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 14/43 drm/amd/display: Update HDMI hang w/a to apply to all TMDS signals
In reply to Rodrigo SiqueiraFrom: Michael Strauss <[email protected]>
[WHY]
48mhz turn off feature does not work on HDMI and DVI, but the feauture was only blocked on HDMI, this change will apply the same wa on DVI
[HOW]
Apply workaround for all TMDS signal types (HDMI, DVI single/dual link)
--- .../gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c +6 -4
@@ -59,14 +59,16 @@ int rn_get_active_display_cnt_wa( struct dc_state *context) {int i, display_count; -bool hdmi_present = false; +bool tmds_present = false; display_count =0 ;for (i =0 ; i < context->stream_count; i++) {const struct dc_stream_state *stream = context->streams[i]; -if (stream->signal ==SIGNAL_TYPE_HDMI_TYPE_A ) - hdmi_present = true; +if (stream->signal ==SIGNAL_TYPE_HDMI_TYPE_A || + stream->signal ==SIGNAL_TYPE_DVI_SINGLE_LINK || + stream->signal ==SIGNAL_TYPE_DVI_DUAL_LINK ) + tmds_present = true; }for (i =0 ; i < dc->link_count; i++) {
@@ -85,7 +87,7 @@ int rn_get_active_display_cnt_wa( }/* WA for hang on HDMI after display off back back on*/ -if (display_count ==0 && hdmi_present) +if (display_count ==0 && tmds_present) display_count =1 ;return display_count;
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 15/43 drm/amd/display: wait for update when setting dpg test pattern
In reply to Rodrigo SiqueiraFrom: Wenjing Liu <[email protected]>
Test pattern should be applied to hardware when exiting set test pattern function.
--- drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c | 13 +++++++++++++ 1 file changed, 13 insertions(+)
drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +13 -0
@@ -3612,6 +3612,7 @@ static void set_crtc_test_pattern(struct dc_link *link,struct pipe_ctx *odm_pipe;enum controller_dp_color_space controller_color_space;int opp_cnt =1 ; +uint8_t count =0 ;switch (test_pattern_color_space) {case DP_TEST_PATTERN_COLOR_SPACE_RGB :
@@ -3647,6 +3648,12 @@ static void set_crtc_test_pattern(struct dc_link *link,NULL , width, height); +/* wait for dpg to blank pixel data with test pattern */ +for (count =0 ; count <1000 ; count++) +if (odm_opp->funcs->dpg_is_blanked (odm_opp)) +break ; +else +udelay (100 ); } opp->funcs->opp_set_disp_pattern_generator(opp, controller_test_pattern,
@@ -3655,6 +3662,12 @@ static void set_crtc_test_pattern(struct dc_link *link,NULL , width, height); +/* wait for dpg to blank pixel data with test pattern */ +for (count =0 ; count <1000 ; count++) +if (opp->funcs->dpg_is_blanked (opp)) +break ; +else +udelay (100 ); } }break ;
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 16/43 drm/amd/display: stop doing unnecessary detection when going to D3
In reply to Rodrigo SiqueiraFrom: Joseph Gravenor <[email protected]>
[Why]
Don't want to start HW discovery unless we have lost power, as doing rediscovery otherwise is both unnecessary and time consuming. Before this change it takes 40 seconds to go in to suspend, after it takes 27 seconds
[How]
Accelerated mode gets cleared if we lose power. Only do detection if this register is cleared
--- drivers/gpu/drm/amd/display/dc/core/dc.c | 6 ++++++ drivers/gpu/drm/amd/display/dc/dc.h | 1 + 2 files changed, 7 insertions(+)
drivers/gpu/drm/amd/display/dc/core/dc.c +6 -0
@@ -1318,6 +1318,12 @@ bool dc_commit_state(struct dc *dc, struct dc_state *context)return (result ==DC_OK ); } +bool dc_is_hw_initialized (struct dc *dc) +{ +struct dc_bios *dcb = dc->ctx->dc_bios; +return dcb->funcs->is_accelerated_mode (dcb); +} +bool dc_post_update_surfaces_to_stream (struct dc *dc) {int i;
drivers/gpu/drm/amd/display/dc/dc.h +1 -0
@@ -1069,6 +1069,7 @@ unsigned int dc_get_current_backlight_pwm(struct dc *dc);unsigned int dc_get_target_backlight_pwm (struct dc *dc);bool dc_is_dmcu_initialized (struct dc *dc); +bool dc_is_hw_initialized (struct dc *dc);enum dc_status dc_set_clock (struct dc *dc,enum dc_clock_type clock_type,uint32_t clk_khz,uint32_t stepping);void dc_get_clock (struct dc *dc,enum dc_clock_type clock_type,struct dc_clock_config *clock_cfg);
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 17/43 drm/amd/display: expand dml structs
In reply to Rodrigo SiqueiraFrom: Dmytro Laktyushkin <[email protected]>
Add more fields to support upcoming dml versions
--- .../drm/amd/display/dc/dcn20/dcn20_resource.c | 35 +++- .../dc/dml/dcn20/display_mode_vba_20.c | 165 +++++++++-------- .../dc/dml/dcn20/display_mode_vba_20v2.c | 167 +++++++++--------- .../dc/dml/dcn21/display_mode_vba_21.c | 157 ++++++++-------- .../amd/display/dc/dml/display_mode_enums.h | 16 ++ .../amd/display/dc/dml/display_mode_structs.h | 7 +- .../drm/amd/display/dc/dml/display_mode_vba.c | 22 ++- .../drm/amd/display/dc/dml/display_mode_vba.h | 128 ++++++++++---- 8 files changed, 402 insertions(+), 295 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +32 -3
@@ -1858,6 +1858,22 @@ void dcn20_populate_dml_writeback_from_context( } +static int get_num_odm_heads (struct pipe_ctx *pipe) +{ +int odm_head_count =0 ; +struct pipe_ctx *next_pipe = pipe->next_odm_pipe; +while (next_pipe) { + odm_head_count++; + next_pipe = next_pipe->next_odm_pipe; + } + pipe = pipe->prev_odm_pipe; +while (pipe) { + odm_head_count++; + pipe = pipe->prev_odm_pipe; + } +return odm_head_count ? odm_head_count +1 :0 ; +} +int dcn20_populate_dml_pipes_from_context (struct dc *dc,struct dc_state *context,display_e2e_pipe_params_st *pipes) {
@@ -1897,6 +1913,8 @@ int dcn20_populate_dml_pipes_from_context(pipes[pipe_cnt].pipe.src.dcc = 0; pipes[pipe_cnt].pipe.src.vm = 0;*/ + pipes[pipe_cnt].clks_cfg.refclk_mhz = dc->res_pool->ref_clocks.dchub_ref_clock_inKhz /1000.0 ; + pipes[pipe_cnt].dout.dsc_enable = res_ctx->pipe_ctx[i].stream->timing.flags.DSC;/* todo: rotation?*/ pipes[pipe_cnt].dout.dsc_slices = res_ctx->pipe_ctx[i].stream->timing.dsc_cfg.num_slices_h;
@@ -1935,8 +1953,13 @@ int dcn20_populate_dml_pipes_from_context( pipes[pipe_cnt].dout.dp_lanes =4 ; pipes[pipe_cnt].pipe.dest.vtotal_min = res_ctx->pipe_ctx[i].stream->adjust.v_total_min; pipes[pipe_cnt].pipe.dest.vtotal_max = res_ctx->pipe_ctx[i].stream->adjust.v_total_max; - pipes[pipe_cnt].pipe.dest.odm_combine = res_ctx->pipe_ctx[i].prev_odm_pipe - || res_ctx->pipe_ctx[i].next_odm_pipe; +switch (get_num_odm_heads (&res_ctx->pipe_ctx[i])) { +case 2 : + pipes[pipe_cnt].pipe.dest.odm_combine = dm_odm_combine_mode_2to1; +break ; +default : + pipes[pipe_cnt].pipe.dest.odm_combine = dm_odm_combine_mode_disabled; + } pipes[pipe_cnt].pipe.src.hsplit_grp = res_ctx->pipe_ctx[i].pipe_idx;if (res_ctx->pipe_ctx[i].top_pipe && res_ctx->pipe_ctx[i].top_pipe->plane_state == res_ctx->pipe_ctx[i].plane_state)
@@ -2046,6 +2069,9 @@ int dcn20_populate_dml_pipes_from_context(if (pipes[pipe_cnt].pipe.src.viewport_height >1080 ) pipes[pipe_cnt].pipe.src.viewport_height =1080 ; pipes[pipe_cnt].pipe.src.surface_height_y = pipes[pipe_cnt].pipe.src.viewport_height; + pipes[pipe_cnt].pipe.src.surface_width_y = pipes[pipe_cnt].pipe.src.viewport_width; + pipes[pipe_cnt].pipe.src.surface_height_c = pipes[pipe_cnt].pipe.src.viewport_height; + pipes[pipe_cnt].pipe.src.surface_width_c = pipes[pipe_cnt].pipe.src.viewport_width; pipes[pipe_cnt].pipe.src.data_pitch = ((pipes[pipe_cnt].pipe.src.viewport_width +63 ) /64 ) *64 ;/* linear sw only */ pipes[pipe_cnt].pipe.src.source_format = dm_444_32; pipes[pipe_cnt].pipe.dest.recout_width = pipes[pipe_cnt].pipe.src.viewport_width;/*vp_width/hratio*/
@@ -2079,7 +2105,10 @@ int dcn20_populate_dml_pipes_from_context( pipes[pipe_cnt].pipe.src.viewport_width_c = scl->viewport_c.width; pipes[pipe_cnt].pipe.src.viewport_height = scl->viewport.height; pipes[pipe_cnt].pipe.src.viewport_height_c = scl->viewport_c.height; + pipes[pipe_cnt].pipe.src.surface_width_y = pln->plane_size.surface_size.width; pipes[pipe_cnt].pipe.src.surface_height_y = pln->plane_size.surface_size.height; + pipes[pipe_cnt].pipe.src.surface_width_c = pln->plane_size.chroma_size.width; + pipes[pipe_cnt].pipe.src.surface_height_c = pln->plane_size.chroma_size.height;if (pln->format >=SURFACE_PIXEL_FORMAT_VIDEO_BEGIN ) { pipes[pipe_cnt].pipe.src.data_pitch = pln->plane_size.surface_pitch; pipes[pipe_cnt].pipe.src.data_pitch_c = pln->plane_size.chroma_pitch;
@@ -2493,7 +2522,7 @@ int dcn20_validate_apply_pipe_split_flags( split[i] = true;if (dc->debug.force_odm_combine & (1 << pipe->stream_res.tg->inst)) { split[i] = true; - context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_idx] = true; + context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_idx] = dm_odm_combine_mode_2to1; } context->bw_ctx.dml.vba.ODMCombineEnabled[pipe_idx] = context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_idx];
drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20.c +82 -83
470 lines not on this page · raw message
drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20v2.c +83 -84
478 lines not on this page · raw message
drivers/gpu/drm/amd/display/dc/dml/dcn21/display_mode_vba_21.c +78 -79
489 lines not on this page · raw message
drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h +16 -0
@@ -119,6 +119,10 @@ enum mpc_combine_affinity {dm_mpc_never }; +enum RequestType { + REQ_256Bytes, REQ_128BytesNonContiguous, REQ_128BytesContiguous,REQ_NA +}; +enum self_refresh_affinity { dm_try_to_allow_self_refresh_and_mclk_switch, dm_allow_self_refresh_and_mclk_switch,
@@ -165,4 +169,16 @@ enum odm_combine_mode { dm_odm_combine_mode_4to1, }; +enum odm_combine_policy { + dm_odm_combine_policy_dal, + dm_odm_combine_policy_none, + dm_odm_combine_policy_2to1, + dm_odm_combine_policy_4to1, +}; + +enum immediate_flip_requirement { + dm_immediate_flip_not_required, + dm_immediate_flip_required, +}; +#endif
drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h +6 -1
@@ -63,6 +63,7 @@ struct _vcs_dpi_voltage_scaling_st {double dispclk_mhz;double phyclk_mhz;double dppclk_mhz; +double dtbclk_mhz; };struct _vcs_dpi_soc_bounding_box_st {
@@ -214,6 +215,7 @@ struct _vcs_dpi_display_pipe_source_params_st {int source_format;unsigned char dcc;unsigned int dcc_rate; +unsigned int dcc_rate_chroma;unsigned char dcc_use_global;unsigned char vm;bool gpuvm;// gpuvm enabled
@@ -225,7 +227,10 @@ struct _vcs_dpi_display_pipe_source_params_st {int source_scan;int sw_mode;int macro_tile_size; +unsigned int surface_width_y;unsigned int surface_height_y; +unsigned int surface_width_c; +unsigned int surface_height_c;unsigned int viewport_width;unsigned int viewport_height;unsigned int viewport_y_y;
@@ -324,7 +329,7 @@ struct _vcs_dpi_display_pipe_dest_params_st {double pixel_rate_mhz;unsigned char synchronized_vblank_all_planes;unsigned char otg_inst; -unsigned char odm_combine; +unsigned int odm_combine;unsigned char use_maximum_vstartup;unsigned int vtotal_max;unsigned int vtotal_min;
drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.c +15 -7
@@ -264,7 +264,10 @@ static void fetch_socbb_params(struct display_mode_lib *mode_lib) mode_lib->vba.DRAMSpeedPerState[i] = soc->clock_limits[i].dram_speed_mts;//mode_lib->vba.DRAMSpeedPerState[i] = soc->clock_limits[i].dram_speed_mhz; mode_lib->vba.MaxDispclk[i] = soc->clock_limits[i].dispclk_mhz; + mode_lib->vba.DTBCLKPerState[i] = soc->clock_limits[i].dtbclk_mhz; } + mode_lib->vba.MinVoltageLevel =0 ; + mode_lib->vba.MaxVoltageLevel = mode_lib->vba.soc.num_states; mode_lib->vba.DoUrgentLatencyAdjustment = soc->do_urgent_latency_adjustment;
@@ -306,8 +309,6 @@ static void fetch_ip_params(struct display_mode_lib *mode_lib) mode_lib->vba.WritebackInterfaceBufferSize = ip->writeback_interface_buffer_size_kbytes; mode_lib->vba.WritebackLineBufferSize = ip->writeback_line_buffer_buffer_size; - mode_lib->vba.MinVoltageLevel =0 ; - mode_lib->vba.MaxVoltageLevel =5 ; mode_lib->vba.WritebackChromaLineBufferWidth = ip->writeback_chroma_line_buffer_width_pixels;
@@ -423,8 +424,8 @@ static void fetch_pipe_params(struct display_mode_lib *mode_lib) ip->dcc_supported : src->dcc && ip->dcc_supported; mode_lib->vba.DCCRate[mode_lib->vba.NumberOfActivePlanes] = src->dcc_rate;/* TODO: Needs to be set based on src->dcc_rate_luma/chroma */ - mode_lib->vba.DCCRateLuma[mode_lib->vba.NumberOfActivePlanes] =0 ; - mode_lib->vba.DCCRateChroma[mode_lib->vba.NumberOfActivePlanes] =0 ; + mode_lib->vba.DCCRateLuma[mode_lib->vba.NumberOfActivePlanes] = src->dcc_rate; + mode_lib->vba.DCCRateChroma[mode_lib->vba.NumberOfActivePlanes] = src->dcc_rate_chroma; mode_lib->vba.SourcePixelFormat[mode_lib->vba.NumberOfActivePlanes] = (enum source_format_class ) (src->source_format);
@@ -436,8 +437,6 @@ static void fetch_pipe_params(struct display_mode_lib *mode_lib) dst->recout_width;// TODO: or should this be full_recout_width???...maybe only when in hsplit mode? mode_lib->vba.ODMCombineEnabled[mode_lib->vba.NumberOfActivePlanes] = dst->odm_combine; - mode_lib->vba.ODMCombineTypeEnabled[mode_lib->vba.NumberOfActivePlanes] = - dst->odm_combine; mode_lib->vba.OutputFormat[mode_lib->vba.NumberOfActivePlanes] = (enum output_format_class ) (dout->output_format); mode_lib->vba.OutputBpp[mode_lib->vba.NumberOfActivePlanes] =
@@ -590,6 +589,7 @@ static void fetch_pipe_params(struct display_mode_lib *mode_lib)for (k = j +1 ; k < mode_lib->vba.cache_num_pipes; ++k) {display_pipe_source_params_st *src_k = &pipes[k].pipe.src;display_pipe_dest_params_st *dst_k = &pipes[k].pipe.dest; +display_output_params_st *dout_k = &pipes[j].dout;if (src_k->is_hsplit && !visited[k] && src->hsplit_grp == src_k->hsplit_grp) {
@@ -600,12 +600,18 @@ static void fetch_pipe_params(struct display_mode_lib *mode_lib) == dm_horz) { mode_lib->vba.ViewportWidth[mode_lib->vba.NumberOfActivePlanes] += src_k->viewport_width; + mode_lib->vba.ViewportWidthChroma[mode_lib->vba.NumberOfActivePlanes] += + src_k->viewport_width; mode_lib->vba.ScalerRecoutWidth[mode_lib->vba.NumberOfActivePlanes] += dst_k->recout_width; } else { mode_lib->vba.ViewportHeight[mode_lib->vba.NumberOfActivePlanes] += src_k->viewport_height; + mode_lib->vba.ViewportHeightChroma[mode_lib->vba.NumberOfActivePlanes] += + src_k->viewport_height; } + mode_lib->vba.NumberOfDSCSlices[mode_lib->vba.NumberOfActivePlanes] += + dout_k->dsc_slices; visited[k] = true; }
@@ -811,7 +817,9 @@ void ModeSupportAndSystemConfiguration(struct display_mode_lib *mode_lib)unsigned int total_pipes =0 ; mode_lib->vba.VoltageLevel = mode_lib->vba.cache_pipes[0 ].clks_cfg.voltage; - mode_lib->vba.ReturnBW = mode_lib->vba.ReturnBWPerState[mode_lib->vba.VoltageLevel]; + mode_lib->vba.ReturnBW = mode_lib->vba.ReturnBWPerState[mode_lib->vba.VoltageLevel][mode_lib->vba.maxMpcComb]; +if (mode_lib->vba.ReturnBW ==0 ) + mode_lib->vba.ReturnBW = mode_lib->vba.ReturnBWPerState[mode_lib->vba.VoltageLevel][0 ]; mode_lib->vba.FabricAndDRAMBandwidth = mode_lib->vba.FabricAndDRAMBandwidthPerState[mode_lib->vba.VoltageLevel];fetch_socbb_params (mode_lib);
drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h +90 -38
284 lines not on this page · raw message
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 18/43 drm/amd/display: 3.2.66
In reply to Rodrigo SiqueiraFrom: Aric Cyr <[email protected]>
--- drivers/gpu/drm/amd/display/dc/dc.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/dc.h +1 -1
@@ -39,7 +39,7 @@#include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#define DC_VER "3.2.65" +#define DC_VER "3.2.66"#define MAX_SURFACES 3#define MAX_PLANES 6
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 19/43 drm/amd/display: Fix double buffering in dcn2 ICSC
In reply to Rodrigo SiqueiraFrom: Noah Abradjian <[email protected]>
[Why]
When rapidly adjusting video brightness, screen tearing was observed. This was due to overwritten values in ICSC registers. In dcn10, this issue had been fixed by implementing double buffering via alternating ICSC modes. However, the second register set used in dcn1 doesn't exist in dcn2.
[How]
Create new program_input_csc for dcn20.
Use ICSC_B registers instead of COMA registers as second set.
--- .../drm/amd/display/dc/dcn10/dcn10_dpp_cm.c | 20 ----- .../gpu/drm/amd/display/dc/dcn20/dcn20_dpp.c | 24 +++--- .../gpu/drm/amd/display/dc/dcn20/dcn20_dpp.h | 31 ++++++- .../drm/amd/display/dc/dcn20/dcn20_dpp_cm.c | 83 ++++++++++++++++++- .../gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c | 10 ++- .../gpu/drm/amd/display/dc/dcn20/dcn20_mpc.h | 6 +- .../drm/amd/display/dc/dcn20/dcn20_resource.c | 1 + .../drm/amd/display/dc/dcn21/dcn21_resource.c | 7 +- drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h | 20 +++++ 9 files changed, 160 insertions(+), 42 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c +0 -20
@@ -88,26 +88,6 @@ enum dscl_mode_sel {DSCL_MODE_DSCL_BYPASS =6 }; -static const struct dpp_input_csc_matrix dpp_input_csc_matrix[] = { - {COLOR_SPACE_SRGB , - {0x2000 ,0 ,0 ,0 ,0 ,0x2000 ,0 ,0 ,0 ,0 ,0x2000 ,0 } }, - {COLOR_SPACE_SRGB_LIMITED , - {0x2000 ,0 ,0 ,0 ,0 ,0x2000 ,0 ,0 ,0 ,0 ,0x2000 ,0 } }, - {COLOR_SPACE_YCBCR601 , - {0x2cdd ,0x2000 ,0 ,0xe991 ,0xe926 ,0x2000 ,0xf4fd ,0x10ef , -0 ,0x2000 ,0x38b4 ,0xe3a6 } }, - {COLOR_SPACE_YCBCR601_LIMITED , - {0x3353 ,0x2568 ,0 ,0xe400 ,0xe5dc ,0x2568 ,0xf367 ,0x1108 , -0 ,0x2568 ,0x40de ,0xdd3a } }, - {COLOR_SPACE_YCBCR709 , - {0x3265 ,0x2000 ,0 ,0xe6ce ,0xf105 ,0x2000 ,0xfa01 ,0xa7d ,0 , -0x2000 ,0x3b61 ,0xe24f } }, - - {COLOR_SPACE_YCBCR709_LIMITED , - {0x39a6 ,0x2568 ,0 ,0xe0d6 ,0xeedd ,0x2568 ,0xf925 ,0x9a8 ,0 , -0x2568 ,0x43ee ,0xdbb2 } } -}; - static void program_gamut_remap( struct dcn10_dpp *dpp, const uint16_t *regval,
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp.c +12 -12
@@ -104,7 +104,7 @@ static void dpp2_cnv_setup (uint32_t pixel_format =0 ;uint32_t alpha_en =1 ;enum dc_color_space color_space =COLOR_SPACE_SRGB ; -enum dcn10_input_csc_select select =INPUT_CSC_SELECT_BYPASS ; +enum dcn20_input_csc_select select =DCN2_ICSC_SELECT_BYPASS ;bool force_disable_cursor = false;struct out_csc_color_matrix tbl_entry;uint32_t is_2bit =0 ;
@@ -145,25 +145,25 @@ static void dpp2_cnv_setup ( force_disable_cursor = false; pixel_format =65 ; color_space =COLOR_SPACE_YCBCR709 ; - select =INPUT_CSC_SELECT_ICSC ; + select =DCN2_ICSC_SELECT_ICSC_A ;break ;case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb: force_disable_cursor = true; pixel_format =64 ; color_space =COLOR_SPACE_YCBCR709 ; - select =INPUT_CSC_SELECT_ICSC ; + select =DCN2_ICSC_SELECT_ICSC_A ;break ;case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr: force_disable_cursor = true; pixel_format =67 ; color_space =COLOR_SPACE_YCBCR709 ; - select =INPUT_CSC_SELECT_ICSC ; + select =DCN2_ICSC_SELECT_ICSC_A ;break ;case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb: force_disable_cursor = true; pixel_format =66 ; color_space =COLOR_SPACE_YCBCR709 ; - select =INPUT_CSC_SELECT_ICSC ; + select =DCN2_ICSC_SELECT_ICSC_A ;break ;case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616 : pixel_format =22 ;
@@ -177,7 +177,7 @@ static void dpp2_cnv_setup (case SURFACE_PIXEL_FORMAT_VIDEO_AYCrCb8888: pixel_format =12 ; color_space =COLOR_SPACE_YCBCR709 ; - select =INPUT_CSC_SELECT_ICSC ; + select =DCN2_ICSC_SELECT_ICSC_A ;break ;case SURFACE_PIXEL_FORMAT_GRPH_RGB111110_FIX : pixel_format =112 ;
@@ -188,13 +188,13 @@ static void dpp2_cnv_setup (case SURFACE_PIXEL_FORMAT_VIDEO_ACrYCb2101010: pixel_format =114 ; color_space =COLOR_SPACE_YCBCR709 ; - select =INPUT_CSC_SELECT_ICSC ; + select =DCN2_ICSC_SELECT_ICSC_A ; is_2bit =1 ;break ;case SURFACE_PIXEL_FORMAT_VIDEO_CrYCbA1010102: pixel_format =115 ; color_space =COLOR_SPACE_YCBCR709 ; - select =INPUT_CSC_SELECT_ICSC ; + select =DCN2_ICSC_SELECT_ICSC_A ; is_2bit =1 ;break ;case SURFACE_PIXEL_FORMAT_GRPH_RGB111110_FLOAT :
@@ -227,13 +227,13 @@ static void dpp2_cnv_setup ( tbl_entry.color_space = input_color_space;if (color_space >=COLOR_SPACE_YCBCR601 ) - select =INPUT_CSC_SELECT_ICSC ; + select =DCN2_ICSC_SELECT_ICSC_A ;else - select =INPUT_CSC_SELECT_BYPASS ; + select =DCN2_ICSC_SELECT_BYPASS ; -dpp1_program_input_csc (dpp_base, color_space, select, &tbl_entry); +dpp2_program_input_csc (dpp_base, color_space, select, &tbl_entry); }else -dpp1_program_input_csc (dpp_base, color_space, select,NULL ); +dpp2_program_input_csc (dpp_base, color_space, select,NULL );if (force_disable_cursor) { REG_UPDATE(CURSOR_CONTROL,
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp.h +30 -1
@@ -150,6 +150,10 @@SRI (CM_SHAPER_RAMA_REGION_32_33 ,CM , id), \ SRI(CM_SHAPER_LUT_INDEX, CM, id) +#define TF_REG_LIST_DCN20_COMMON_APPEND (id) \ + SRI(CM_ICSC_B_C11_C12, CM, id), \ + SRI(CM_ICSC_B_C33_C34, CM, id) +#define TF_REG_LIST_DCN20 (id) \ TF_REG_LIST_DCN(id), \ TF_REG_LIST_DCN20_COMMON(id), \
@@ -572,6 +576,14 @@TF_SF (DSCL0_OBUF_MEM_PWR_CTRL ,OBUF_MEM_PWR_FORCE , mask_sh),\TF_SF (DSCL0_DSCL_MEM_PWR_CTRL ,LUT_MEM_PWR_FORCE , mask_sh) +/* DPP CM debug status register: +* +* Status index including current ICSC, Gamut Remap Mode is 9 +* ICSC Mode: [5..4] + */ +#define CM_TEST_DEBUG_DATA_STATUS_IDX 9 +#define CM_TEST_DEBUG_DATA_ICSC_MODE_SH 4 +#define CM_TEST_DEBUG_DATA_ICSC_MODE_MASK 0x3#define TF_REG_FIELD_LIST_DCN2_0 (type) \ TF_REG_FIELD_LIST(type) \
@@ -630,11 +642,16 @@ struct dcn2_dpp_mask {uint32_t COLOR_KEYER_RED ; \uint32_t COLOR_KEYER_GREEN ; \uint32_t COLOR_KEYER_BLUE ; \ -uint32_t OBUF_MEM_PWR_CTRL ;\ +uint32_t OBUF_MEM_PWR_CTRL ; \ uint32_t DSCL_MEM_PWR_CTRL +#define DPP_DCN2_REG_VARIABLE_LIST_CM_APPEND \ + uint32_t CM_ICSC_B_C11_C12; \ + uint32_t CM_ICSC_B_C33_C34 +struct dcn2_dpp_registers {DPP_DCN2_REG_VARIABLE_LIST ; +DPP_DCN2_REG_VARIABLE_LIST_CM_APPEND ; };struct dcn20_dpp {
@@ -656,6 +673,12 @@ struct dcn20_dpp {struct pwl_params pwl_data; }; +enum dcn20_input_csc_select { +DCN2_ICSC_SELECT_BYPASS =0 , +DCN2_ICSC_SELECT_ICSC_A =1 , +DCN2_ICSC_SELECT_ICSC_B =2 +}; +void dpp20_read_state (struct dpp *dpp_base,struct dcn_dpp_state *s);
@@ -667,6 +690,12 @@ void dpp2_set_degamma(struct dpp *dpp_base, enum ipp_degamma_mode mode); +void dpp2_program_input_csc ( +struct dpp *dpp_base, +enum dc_color_space color_space, +enum dcn20_input_csc_select input_select, +const struct out_csc_color_matrix *tbl_entry); +bool dpp20_program_blnd_lut (struct dpp *dpp_base,const struct pwl_params *params);
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp_cm.c +80 -3
101 lines not on this page · raw message
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c +8 -2
@@ -154,7 +154,10 @@ void mpc2_set_output_csc(* the CSC update so CSC is updated on frame boundary */ cur_mode =IX_REG_READ (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_DATA , -MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_IDX ); +MPC_OCSC_TEST_DEBUG_DATA_STATUS_IDX ); + +/* Isolate part of reg data we want [1..0] */ + cur_mode = cur_mode &MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_MASK ;if (cur_mode !=MPC_OUTPUT_CSC_COEF_A ) ocsc_mode =MPC_OUTPUT_CSC_COEF_A ;
@@ -211,7 +214,10 @@ void mpc2_set_ocsc_default(* the CSC update so CSC is updated on frame boundary */ cur_mode =IX_REG_READ (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_DATA , -MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_IDX ); +MPC_OCSC_TEST_DEBUG_DATA_STATUS_IDX ); + +/* Isolate part of reg data we want [1..0] */ + cur_mode = cur_mode &MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_MASK ;if (cur_mode !=MPC_OUTPUT_CSC_COEF_A ) ocsc_mode =MPC_OUTPUT_CSC_COEF_A ;
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.h +4 -2
@@ -185,9 +185,11 @@/* * DCN2 MPC_OCSC debug status register: * -* Field describing current OCSC Mode has index 1 [1..0] +* Status index including current OCSC Mode is 1 +* OCSC Mode: [1..0] */ -#define MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_IDX 1 +#define MPC_OCSC_TEST_DEBUG_DATA_STATUS_IDX 1 +#define MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_MASK 0x3#define MPC_REG_FIELD_LIST_DCN2_0 (type) \ MPC_REG_FIELD_LIST(type)\
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +1 -0
@@ -638,6 +638,7 @@ static const struct dce110_aux_registers aux_engine_regs[] = {#define tf_regs (id)\ [id] = {\ TF_REG_LIST_DCN20(id),\ + TF_REG_LIST_DCN20_COMMON_APPEND(id),\ }static const struct dcn2_dpp_registers tf_regs[] = {
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +5 -2
@@ -609,6 +609,7 @@ static const struct dce110_aux_registers aux_engine_regs[] = {#define tf_regs (id)\ [id] = {\ TF_REG_LIST_DCN20(id),\ + TF_REG_LIST_DCN20_COMMON_APPEND(id),\ }static const struct dcn2_dpp_registers tf_regs[] = {
@@ -619,11 +620,13 @@ static const struct dcn2_dpp_registers tf_regs[] = { };static const struct dcn2_dpp_shift tf_shift = { -TF_REG_LIST_SH_MASK_DCN20 (__SHIFT) +TF_REG_LIST_SH_MASK_DCN20 (__SHIFT), +TF_DEBUG_REG_LIST_SH_DCN10 };static const struct dcn2_dpp_mask tf_mask = { -TF_REG_LIST_SH_MASK_DCN20 (_MASK) +TF_REG_LIST_SH_MASK_DCN20 (_MASK), +TF_DEBUG_REG_LIST_MASK_DCN10 };#define stream_enc_regs (id)\
drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h +20 -0
@@ -47,6 +47,26 @@ struct dpp_input_csc_matrix {uint16_t regval[12 ]; }; +static const struct dpp_input_csc_matrix dpp_input_csc_matrix[] = { + {COLOR_SPACE_SRGB , + {0x2000 ,0 ,0 ,0 ,0 ,0x2000 ,0 ,0 ,0 ,0 ,0x2000 ,0 } }, + {COLOR_SPACE_SRGB_LIMITED , + {0x2000 ,0 ,0 ,0 ,0 ,0x2000 ,0 ,0 ,0 ,0 ,0x2000 ,0 } }, + {COLOR_SPACE_YCBCR601 , + {0x2cdd ,0x2000 ,0 ,0xe991 ,0xe926 ,0x2000 ,0xf4fd ,0x10ef , +0 ,0x2000 ,0x38b4 ,0xe3a6 } }, + {COLOR_SPACE_YCBCR601_LIMITED , + {0x3353 ,0x2568 ,0 ,0xe400 ,0xe5dc ,0x2568 ,0xf367 ,0x1108 , +0 ,0x2568 ,0x40de ,0xdd3a } }, + {COLOR_SPACE_YCBCR709 , + {0x3265 ,0x2000 ,0 ,0xe6ce ,0xf105 ,0x2000 ,0xfa01 ,0xa7d ,0 , +0x2000 ,0x3b61 ,0xe24f } }, + + {COLOR_SPACE_YCBCR709_LIMITED , + {0x39a6 ,0x2568 ,0 ,0xe0d6 ,0xeedd ,0x2568 ,0xf925 ,0x9a8 ,0 , +0x2568 ,0x43ee ,0xdbb2 } } +}; +struct dpp_grph_csc_adjustment {struct fixed31_32 temperature_matrix[CSC_TEMPERATURE_MATRIX_SIZE ];enum graphics_gamut_adjust_type gamut_adjust_type;
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 20/43 drm/amd/display: Driverside changes to support PSR in DMCUB
In reply to Rodrigo SiqueiraFrom: Wyatt Wood <[email protected]>
[Why]
Moving PSR from DMCU to DMCUB.
[How]
Add driverside PSR changes required to send inbox messages to fw. These changes are non-functional until the psr structure allocation is uncommented.
--- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 22 +- drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c | 220 ++++++++++++++++++ drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h | 47 ++++ .../drm/amd/display/dc/dcn21/dcn21_resource.c | 5 + .../gpu/drm/amd/display/dc/inc/core_types.h | 1 + .../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 34 ++- 6 files changed, 313 insertions(+), 16 deletions(-) create mode 100644 drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c create mode 100644 drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h
drivers/gpu/drm/amd/display/dc/core/dc_link.c +16 -6
@@ -45,6 +45,7 @@#include "dpcd_defs.h" #include "dmcu.h" #include "hw/clk_mgr.h" +#include "../dce/dmub_psr.h" #define DC_LOGGER_INIT (logger)
@@ -2404,10 +2405,11 @@ bool dc_link_set_psr_allow_active(struct dc_link *link, bool allow_active, bool {struct dc *dc = link->ctx->dc;struct dmcu *dmcu = dc->res_pool->dmcu; +struct dmub_psr *psr = dc->res_pool->psr; - - -if ((dmcu !=NULL && dmcu->funcs->is_dmcu_initialized (dmcu)) && link->psr_feature_enabled) +if ((psr !=NULL ) && link->psr_feature_enabled) + psr->funcs->set_psr_enable (psr, allow_active); +else if ((dmcu !=NULL && dmcu->funcs->is_dmcu_initialized (dmcu)) && link->psr_feature_enabled) dmcu->funcs->set_psr_enable (dmcu, allow_active, wait); link->psr_allow_active = allow_active;
@@ -2419,8 +2421,11 @@ bool dc_link_get_psr_state(const struct dc_link *link, uint32_t *psr_state) {struct dc *dc = link->ctx->dc;struct dmcu *dmcu = dc->res_pool->dmcu; +struct dmub_psr *psr = dc->res_pool->psr; -if (dmcu !=NULL && link->psr_feature_enabled) +if (psr !=NULL && link->psr_feature_enabled) + psr->funcs->get_psr_state (psr_state); +else if (dmcu !=NULL && link->psr_feature_enabled) dmcu->funcs->get_psr_state (dmcu, psr_state);return true;
@@ -2467,6 +2472,7 @@ bool dc_link_setup_psr(struct dc_link *link, {struct dc *dc;struct dmcu *dmcu; +struct dmub_psr *psr;int i;/* updateSinkPsrDpcdConfig*/ union dpcd_psr_configuration psr_configuration;
@@ -2478,8 +2484,9 @@ bool dc_link_setup_psr(struct dc_link *link, dc = link->ctx->dc; dmcu = dc->res_pool->dmcu; + psr = dc->res_pool->psr; -if (!dmcu) +if (!dmcu && !psr)return false;
@@ -2588,7 +2595,10 @@ bool dc_link_setup_psr(struct dc_link *link,*/ psr_context->frame_delay =0 ; - link->psr_feature_enabled = dmcu->funcs->setup_psr (dmcu, link, psr_context); +if (psr) + link->psr_feature_enabled = psr->funcs->setup_psr (psr, link, psr_context); +else + link->psr_feature_enabled = dmcu->funcs->setup_psr (dmcu, link, psr_context);/* psr_enabled == 0 indicates setup_psr did not succeed, but this * should not happen since firmware should be running at this point
drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c new file +220 -0
220 lines not on this page · raw message
drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h new file +47 -0
47 lines not on this page · raw message
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +5 -0
@@ -83,6 +83,7 @@#include "dcn21_resource.h" #include "vm_helper.h" #include "dcn20/dcn20_vmid.h" +#include "../dce/dmub_psr.h" #define SOC_BOUNDING_BOX_VALID false#define DC_LOGGER_INIT (logger)
@@ -1744,6 +1745,10 @@ static bool dcn21_resource_construct( goto create_fail; } +// Leave as NULL to not affect current dmcu psr programming sequence +// Will be uncommented when functionality is confirmed to be working + pool->base.psr =NULL ; + pool->base.abm = dce_abm_create(ctx, &abm_regs, &abm_shift,
drivers/gpu/drm/amd/display/dc/inc/core_types.h +1 -0
@@ -212,6 +212,7 @@ struct resource_pool {struct abm *abm;struct dmcu *dmcu; +struct dmub_psr *psr;const struct resource_funcs *funcs;const struct resource_caps *res_cap;
drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +24 -10
@@ -187,9 +187,28 @@ struct dmub_rb_cmd_dpphy_init { };struct dmub_cmd_psr_copy_settings_data { -uint32_t reg1; -uint32_t reg2; -uint32_t reg3; +uint16_t psr_level; +uint8_t hubp_inst; +uint8_t dpp_inst; +uint8_t mpcc_inst; +uint8_t opp_inst; +uint8_t otg_inst; +uint8_t digfe_inst; +uint8_t digbe_inst; +uint8_t dpphy_inst; +uint8_t aux_inst; +uint8_t hyst_frames; +uint8_t hyst_lines; +uint8_t phy_num; +uint8_t phy_type; +uint8_t aux_repeat; +uint8_t smu_optimizations_en; +uint8_t skip_wait_for_pll_lock; +uint8_t frame_delay; +uint8_t smu_phy_id; +uint8_t num_of_controllers; +uint8_t link_rate; +uint8_t frame_cap_ind; };struct dmub_rb_cmd_psr_copy_settings {
@@ -206,10 +225,6 @@ struct dmub_rb_cmd_psr_set_level {struct dmub_cmd_psr_set_level_data psr_set_level_data; }; -struct dmub_rb_cmd_psr_disable { -struct dmub_cmd_header header; -}; -struct dmub_rb_cmd_psr_enable {struct dmub_cmd_header header; };
@@ -224,8 +239,8 @@ struct dmub_rb_cmd_notify_vblank { };struct dmub_cmd_psr_notify_static_state_data { -uint32_t ss_int;// Which static screen interrupt was triggered -uint32_t ss_enter;// Enter (1) or exit (0) static screen +uint32_t ss_int;// Which static screen interrupt was triggered +uint32_t ss_enter;// Enter (1) or exit (0) static screen };struct dmub_rb_cmd_psr_notify_static_state {
@@ -245,7 +260,6 @@ union dmub_rb_cmd {struct dmub_rb_cmd_dpphy_init dpphy_init;struct dmub_rb_cmd_dig1_transmitter_control dig1_transmitter_control;struct dmub_rb_cmd_psr_enable psr_enable; -struct dmub_rb_cmd_psr_disable psr_disable;struct dmub_rb_cmd_psr_copy_settings psr_copy_settings;struct dmub_rb_cmd_psr_set_level psr_set_level; };
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 21/43 drm/amd/display: Double buffer dcn2 Gamut Remap
In reply to Rodrigo SiqueiraFrom: Noah Abradjian <[email protected]>
[Why]
When rapidly adjusting color temperature, screen tearing was observed. This was due to overwritten values in gamut remap registers. This issue was solved for OCSC and ICSC by alternating between "A" and "B" registers to double buffer the writes.
[How]
Create new set_gamut_remap and program_gamut_remap for dcn20. Alternate which registers are written to by switching modes each time. Also fixes ICSC mode reg read to use proper data offset.
--- .../gpu/drm/amd/display/dc/dcn20/dcn20_dpp.c | 2 +- .../gpu/drm/amd/display/dc/dcn20/dcn20_dpp.h | 29 ++++++- .../drm/amd/display/dc/dcn20/dcn20_dpp_cm.c | 84 ++++++++++++++++++- 3 files changed, 109 insertions(+), 6 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp.c +1 -1
@@ -458,7 +458,7 @@ static struct dpp_funcs dcn20_dpp_funcs = { .dpp_reset = dpp_reset, .dpp_set_scaler = dpp1_dscl_set_scaler_manual_scale, .dpp_get_optimal_number_of_taps = dpp1_get_optimal_number_of_taps, - .dpp_set_gamut_remap = dpp1_cm_set_gamut_remap, + .dpp_set_gamut_remap = dpp2_cm_set_gamut_remap, .dpp_set_csc_adjustment = NULL, .dpp_set_csc_default = NULL, .dpp_program_regamma_pwl = oppn20_dummy_program_regamma_pwl,
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp.h +27 -2
@@ -151,6 +151,12 @@ SRI(CM_SHAPER_LUT_INDEX, CM, id)#define TF_REG_LIST_DCN20_COMMON_APPEND (id) \ + SRI(CM_GAMUT_REMAP_B_C11_C12, CM, id),\ + SRI(CM_GAMUT_REMAP_B_C13_C14, CM, id),\ + SRI(CM_GAMUT_REMAP_B_C21_C22, CM, id),\ + SRI(CM_GAMUT_REMAP_B_C23_C24, CM, id),\ + SRI(CM_GAMUT_REMAP_B_C31_C32, CM, id),\ + SRI(CM_GAMUT_REMAP_B_C33_C34, CM, id),\ SRI(CM_ICSC_B_C11_C12, CM, id), \ SRI(CM_ICSC_B_C33_C34, CM, id)
@@ -579,11 +585,14 @@/* DPP CM debug status register: * * Status index including current ICSC, Gamut Remap Mode is 9 -* ICSC Mode: [5..4] +* ICSC Mode: [4..3] +* Gamut Remap Mode: [10..9] */ #define CM_TEST_DEBUG_DATA_STATUS_IDX 9 -#define CM_TEST_DEBUG_DATA_ICSC_MODE_SH 4 +#define CM_TEST_DEBUG_DATA_ICSC_MODE_SH 3#define CM_TEST_DEBUG_DATA_ICSC_MODE_MASK 0x3 +#define CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE_SH 9 +#define CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE_MASK 0x3#define TF_REG_FIELD_LIST_DCN2_0 (type) \ TF_REG_FIELD_LIST(type) \
@@ -646,6 +655,12 @@ struct dcn2_dpp_mask { uint32_t DSCL_MEM_PWR_CTRL#define DPP_DCN2_REG_VARIABLE_LIST_CM_APPEND \ + uint32_t CM_GAMUT_REMAP_B_C11_C12; \ + uint32_t CM_GAMUT_REMAP_B_C13_C14; \ + uint32_t CM_GAMUT_REMAP_B_C21_C22; \ + uint32_t CM_GAMUT_REMAP_B_C23_C24; \ + uint32_t CM_GAMUT_REMAP_B_C31_C32; \ + uint32_t CM_GAMUT_REMAP_B_C33_C34; \ uint32_t CM_ICSC_B_C11_C12; \ uint32_t CM_ICSC_B_C33_C34
@@ -679,6 +694,12 @@ enum dcn20_input_csc_select { DCN2_ICSC_SELECT_ICSC_B = 2 }; +enum dcn20_gamut_remap_select { +DCN2_GAMUT_REMAP_BYPASS =0 , +DCN2_GAMUT_REMAP_COEF_A =1 , +DCN2_GAMUT_REMAP_COEF_B =2 +}; +void dpp20_read_state (struct dpp *dpp_base,struct dcn_dpp_state *s);
@@ -690,6 +711,10 @@ void dpp2_set_degamma(struct dpp *dpp_base, enum ipp_degamma_mode mode); +void dpp2_cm_set_gamut_remap ( +struct dpp *dpp_base, +const struct dpp_grph_csc_adjustment *adjust); + void dpp2_program_input_csc( struct dpp *dpp_base, enum dc_color_space color_space,
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp_cm.c +81 -3
99 lines not on this page · raw message
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 22/43 drm/amd/display: programing surface flip by dmcub.
In reply to Rodrigo SiqueiraFrom: Yongqiang Sun <[email protected]>
Programming surface flip addresses via dmcub uC for optimizing the data flush.
--- drivers/gpu/drm/amd/display/dc/dc.h | 1 + .../gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c | 116 ++++++++++++++++++ .../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 36 +++++- .../gpu/drm/amd/display/dmub/inc/dmub_rb.h | 18 ++- 4 files changed, 165 insertions(+), 6 deletions(-)
drivers/gpu/drm/amd/display/dc/dc.h +1 -0
@@ -420,6 +420,7 @@ struct dc_debug_options {bool nv12_iflip_vm_wa;bool disable_dram_clock_change_vactive_support;bool validate_dml_output; +bool enable_dmcub_surface_flip; };struct dc_debug_data {
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c +116 -0
134 lines not on this page · raw message
drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +35 -1
@@ -30,12 +30,13 @@#include "dmub_cmd_dal.h" #include "dmub_cmd_vbios.h" #include "atomfirmware.h" - +#include "dc_hw_types.h" #define DMUB_RB_CMD_SIZE 64#define DMUB_RB_MAX_ENTRY 128#define DMUB_RB_SIZE (DMUB_RB_CMD_SIZE * DMUB_RB_MAX_ENTRY)#define REG_SET_MASK 0xFFFF +/* * Command IDs should be treated as stable ABI. * Do not reuse or modify IDs.
@@ -47,6 +48,7 @@ enum dmub_cmd_type {DMUB_CMD__REG_SEQ_FIELD_UPDATE_SEQ =2 ,DMUB_CMD__REG_SEQ_BURST_WRITE =3 ,DMUB_CMD__REG_REG_WAIT =4 , +DMUB_CMD__SURFACE_FLIP =5 ,DMUB_CMD__PSR =64 , DMUB_CMD__VBIOS = 128, };
@@ -145,6 +147,37 @@ struct dmub_rb_cmd_reg_wait {struct dmub_cmd_reg_wait_data reg_wait; }; +#ifndef PHYSICAL_ADDRESS_LOC +#define PHYSICAL_ADDRESS_LOC union large_integer +#endif + +struct dmub_cmd_surface_flip { +uint32_t DCSURF_SURFACE_CONTROL; +uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH; +uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS; +uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH; +uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS; +uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C; +uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS_C; +uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C; +uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS_C; +uint32_t DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH; +uint32_t DCSURF_SECONDARY_META_SURFACE_ADDRESS; +uint32_t DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH; +uint32_t DCSURF_SECONDARY_SURFACE_ADDRESS; +enum dc_plane_addr_type addr_type; +uint8_t hubp_inst; +bool tmz_surface; +bool immediate; +uint8_t vmid; +bool grph_stereo; +}; + +struct dmub_rb_cmd_flip { +struct dmub_cmd_header header; +struct dmub_cmd_surface_flip flip; +}; +struct dmub_cmd_digx_encoder_control_data {union dig_encoder_control_parameters_v1_5 dig; };
@@ -262,6 +295,7 @@ union dmub_rb_cmd {struct dmub_rb_cmd_psr_enable psr_enable;struct dmub_rb_cmd_psr_copy_settings psr_copy_settings;struct dmub_rb_cmd_psr_set_level psr_set_level; +struct dmub_rb_cmd_flip surface_flip; };#pragma pack(pop)
drivers/gpu/drm/amd/display/dmub/inc/dmub_rb.h +13 -5
@@ -73,12 +73,17 @@ static inline bool dmub_rb_full(struct dmub_rb *rb)static inline bool dmub_rb_push_front (struct dmub_rb *rb,const struct dmub_cmd_header *cmd) { -uint8_t *wt_ptr = (uint8_t *)(rb->base_address) + rb->wrpt; +uint64_t volatile *dst = (uint64_t volatile *)(rb->base_address) + rb->wrpt /sizeof (uint64_t ); +const uint64_t *src = (const uint64_t *)cmd; +int i;if (dmub_rb_full (rb))return false; -dmub_memcpy (wt_ptr, cmd,DMUB_RB_CMD_SIZE ); +// copying data +for (i =0 ; i <DMUB_RB_CMD_SIZE /sizeof (uint64_t ); i++) + *dst++ = *src++; + rb->wrpt +=DMUB_RB_CMD_SIZE ;if (rb->wrpt >= rb->capacity)
@@ -115,14 +120,17 @@ static inline bool dmub_rb_pop_front(struct dmub_rb *rb)static inline void dmub_rb_flush_pending (const struct dmub_rb *rb) { -uint8_t buf[DMUB_RB_CMD_SIZE ];uint32_t rptr = rb->rptr;uint32_t wptr = rb->wrpt;while (rptr != wptr) { -const uint8_t *data = (const uint8_t *)rb->base_address + rptr; +uint64_t volatile *data = (uint64_t volatile *)rb->base_address + rptr /sizeof (uint64_t ); +//uint64_t volatile *p = (uint64_t volatile *)data; +uint64_t temp; +int i; -dmub_memcpy (buf, data,DMUB_RB_CMD_SIZE ); +for (i =0 ; i <DMUB_RB_CMD_SIZE /sizeof (uint64_t ); i++) + temp = *data++; rptr +=DMUB_RB_CMD_SIZE ;if (rptr >= rb->capacity)
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 23/43 drm/amd/display: DMCUB FW Changes to support PSR
In reply to Rodrigo SiqueiraFrom: Wyatt Wood <[email protected]>
[Why]
Moving PSR from DMCU to DMCUB.
[How]
Cleanup psr spec files and add PSR hw programming files. No functionality is included in this change.
--- .../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 19 +++++-------------- .../drm/amd/display/dmub/inc/dmub_cmd_dal.h | 6 ++++++ 2 files changed, 11 insertions(+), 14 deletions(-)
drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +5 -14
@@ -262,23 +262,13 @@ struct dmub_rb_cmd_psr_enable {struct dmub_cmd_header header; }; -struct dmub_cmd_psr_notify_vblank_data { -uint32_t vblank_int;// Which vblank interrupt was triggered +struct dmub_cmd_psr_setup_data { +enum psr_version version;// PSR version 1 or 2 }; -struct dmub_rb_cmd_notify_vblank { +struct dmub_rb_cmd_psr_setup {struct dmub_cmd_header header; -struct dmub_cmd_psr_notify_vblank_data psr_notify_vblank_data; -}; - -struct dmub_cmd_psr_notify_static_state_data { -uint32_t ss_int;// Which static screen interrupt was triggered -uint32_t ss_enter;// Enter (1) or exit (0) static screen -}; - -struct dmub_rb_cmd_psr_notify_static_state { -struct dmub_cmd_header header; -struct dmub_cmd_psr_notify_static_state_data psr_notify_static_state_data; +struct dmub_cmd_psr_setup_data psr_setup_data; };union dmub_rb_cmd {
@@ -296,6 +286,7 @@ union dmub_rb_cmd {struct dmub_rb_cmd_psr_copy_settings psr_copy_settings;struct dmub_rb_cmd_psr_set_level psr_set_level;struct dmub_rb_cmd_flip surface_flip; +struct dmub_rb_cmd_psr_setup psr_setup; };#pragma pack(pop)
drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd_dal.h +6 -0
@@ -36,6 +36,12 @@ enum dmub_cmd_psr_type {DMUB_CMD__PSR_DISABLE =1 ,DMUB_CMD__PSR_COPY_SETTINGS =2 ,DMUB_CMD__PSR_SET_LEVEL =3 , + DMUB_CMD__PSR_SETUP = 4, +}; + +enum psr_version { +PSR_VERSION_1 =0x0 , +PSR_VERSION_2 =0x10 , };#endif /* _DMUB_CMD_DAL_H_ */
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 24/43 drm/amd/display: Indirect reg read macro with shift and mask
In reply to Rodrigo SiqueiraFrom: Noah Abradjian <[email protected]>
[Why]
Recent double buffering changes for dcn2 use IX_REG_READ. However, this macro returns the full register value, with the need to manually shift and mask it to retrieve field data.
[How]
Create new IX_REG_GET macro that handles shift and mask. Use this for double buffering reads instead of IX_REG_READ.
--- drivers/gpu/drm/amd/display/dc/dc_helper.c | 30 +++++++++++++++++++ .../gpu/drm/amd/display/dc/dcn20/dcn20_dpp.h | 16 +++++++--- .../drm/amd/display/dc/dcn20/dcn20_dpp_cm.c | 18 ++++------- .../gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c | 16 ++++------ .../gpu/drm/amd/display/dc/dcn20/dcn20_mpc.h | 10 +++++-- .../drm/amd/display/dc/dcn20/dcn20_resource.c | 10 ++++--- .../drm/amd/display/dc/dcn21/dcn21_resource.c | 10 ++++--- .../gpu/drm/amd/display/dc/inc/reg_helper.h | 13 ++++++++ 8 files changed, 86 insertions(+), 37 deletions(-)
drivers/gpu/drm/amd/display/dc/dc_helper.c +30 -0
@@ -552,6 +552,36 @@ uint32_t generic_read_indirect_reg(const struct dc_context *ctx,return value; } +uint32_t generic_indirect_reg_get (const struct dc_context *ctx, +uint32_t addr_index,uint32_t addr_data, +uint32_t index,int n, +uint8_t shift1,uint32_t mask1,uint32_t *field_value1, + ...) +{ +uint32_t shift, mask, *field_value; +uint32_t value =0 ; +int i =1 ; + +va_list ap; + +va_start (ap, field_value1); + + value =generic_read_indirect_reg (ctx, addr_index, addr_data, index); + *field_value1 =get_reg_field_value_ex (value, mask1, shift1); + +while (i < n) { + shift =va_arg (ap, uint32_t); + mask =va_arg (ap, uint32_t); + field_value = va_arg(ap, uint32_t *); + + *field_value =get_reg_field_value_ex (value, mask, shift); + i++; + } + +va_end (ap); + +return value; +} uint32_t generic_indirect_reg_update_ex(const struct dc_context *ctx, uint32_t addr_index, uint32_t addr_data,
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp.h +12 -4
@@ -589,14 +589,22 @@* Gamut Remap Mode: [10..9] */ #define CM_TEST_DEBUG_DATA_STATUS_IDX 9 -#define CM_TEST_DEBUG_DATA_ICSC_MODE_SH 3 -#define CM_TEST_DEBUG_DATA_ICSC_MODE_MASK 0x3 -#define CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE_SH 9 -#define CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE_MASK 0x3 + +#define TF_DEBUG_REG_LIST_SH_DCN20 \ + TF_DEBUG_REG_LIST_SH_DCN10, \ + .CM_TEST_DEBUG_DATA_ICSC_MODE = 3, \ + .CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE = 9 + +#define TF_DEBUG_REG_LIST_MASK_DCN20 \ + TF_DEBUG_REG_LIST_MASK_DCN10, \ + .CM_TEST_DEBUG_DATA_ICSC_MODE = 0x18, \ + .CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE = 0x600#define TF_REG_FIELD_LIST_DCN2_0 (type) \ TF_REG_FIELD_LIST(type) \ type CM_BLNDGAM_LUT_DATA; \ + type CM_TEST_DEBUG_DATA_ICSC_MODE; \ + type CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE; \ type FORMAT_CNV16; \ type CNVC_BYPASS_MSB_ALIGN; \ type CLAMP_POSITIVE; \
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp_cm.c +6 -12
@@ -176,12 +176,9 @@ static void program_gamut_remap(* currently. select the alternate set to double buffer * the update so gamut_remap is updated on frame boundary */ - cur_select =IX_REG_READ (CM_TEST_DEBUG_INDEX ,CM_TEST_DEBUG_DATA , -CM_TEST_DEBUG_DATA_STATUS_IDX ); - -/* IX_REG_READ reads whole reg, so isolate part we want [10..9] */ - cur_select = (cur_select >>CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE_SH ) - &CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE_MASK ; +IX_REG_GET (CM_TEST_DEBUG_INDEX ,CM_TEST_DEBUG_DATA , +CM_TEST_DEBUG_DATA_STATUS_IDX , +CM_TEST_DEBUG_DATA_GAMUT_REMAP_MODE , &cur_select);/* value stored in dbg reg will be 1 greater than mode we want */ if (cur_select !=DCN2_GAMUT_REMAP_COEF_A )
@@ -275,12 +272,9 @@ void dpp2_program_input_csc(* currently. select the alternate set to double buffer * the CSC update so CSC is updated on frame boundary */ - cur_select =IX_REG_READ (CM_TEST_DEBUG_INDEX ,CM_TEST_DEBUG_DATA , -CM_TEST_DEBUG_DATA_STATUS_IDX ); - -/* IX_REG_READ reads whole reg, so isolate part we want [4..3] */ - cur_select = (cur_select >>CM_TEST_DEBUG_DATA_ICSC_MODE_SH ) - &CM_TEST_DEBUG_DATA_ICSC_MODE_MASK ; +IX_REG_GET (CM_TEST_DEBUG_INDEX ,CM_TEST_DEBUG_DATA , +CM_TEST_DEBUG_DATA_STATUS_IDX , +CM_TEST_DEBUG_DATA_ICSC_MODE , &cur_select);if (cur_select !=DCN2_ICSC_SELECT_ICSC_A ) select =DCN2_ICSC_SELECT_ICSC_A ;
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c +6 -10
@@ -153,11 +153,9 @@ void mpc2_set_output_csc(* currently. select the alternate set to double buffer * the CSC update so CSC is updated on frame boundary */ - cur_mode =IX_REG_READ (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_DATA , -MPC_OCSC_TEST_DEBUG_DATA_STATUS_IDX ); - -/* Isolate part of reg data we want [1..0] */ - cur_mode = cur_mode &MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_MASK ; +IX_REG_GET (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_DATA , +MPC_OCSC_TEST_DEBUG_DATA_STATUS_IDX , +MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE , &cur_mode);if (cur_mode !=MPC_OUTPUT_CSC_COEF_A ) ocsc_mode =MPC_OUTPUT_CSC_COEF_A ;
@@ -213,11 +211,9 @@ void mpc2_set_ocsc_default(* currently. select the alternate set to double buffer * the CSC update so CSC is updated on frame boundary */ - cur_mode =IX_REG_READ (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_DATA , -MPC_OCSC_TEST_DEBUG_DATA_STATUS_IDX ); - -/* Isolate part of reg data we want [1..0] */ - cur_mode = cur_mode &MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_MASK ; +IX_REG_GET (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_DATA , +MPC_OCSC_TEST_DEBUG_DATA_STATUS_IDX , +MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE , &cur_mode);if (cur_mode !=MPC_OUTPUT_CSC_COEF_A ) ocsc_mode =MPC_OUTPUT_CSC_COEF_A ;
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.h +7 -3
@@ -140,7 +140,6 @@SF (MPCC0_MPCC_TOP_GAIN ,MPCC_TOP_GAIN , mask_sh),\SF (MPCC0_MPCC_BOT_GAIN_INSIDE ,MPCC_BOT_GAIN_INSIDE , mask_sh),\SF (MPCC0_MPCC_BOT_GAIN_OUTSIDE ,MPCC_BOT_GAIN_OUTSIDE , mask_sh),\ -SF (MPC_OCSC_TEST_DEBUG_DATA ,MPC_OCSC_TEST_DEBUG_DATA , mask_sh),\SF (MPC_OCSC_TEST_DEBUG_INDEX ,MPC_OCSC_TEST_DEBUG_INDEX , mask_sh),\SF (MPC_OUT0_CSC_MODE ,MPC_OCSC_MODE , mask_sh),\ SF(MPC_OUT0_CSC_C11_C12_A, MPC_OCSC_C11_A, mask_sh),\
@@ -189,7 +188,12 @@* OCSC Mode: [1..0] */ #define MPC_OCSC_TEST_DEBUG_DATA_STATUS_IDX 1 -#define MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE_MASK 0x3 + +#define MPC_DEBUG_REG_LIST_SH_DCN20 \ + .MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE = 0 + +#define MPC_DEBUG_REG_LIST_MASK_DCN20 \ + .MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE = 0x3#define MPC_REG_FIELD_LIST_DCN2_0 (type) \ MPC_REG_FIELD_LIST(type)\
@@ -198,7 +202,7 @@type MPCC_TOP_GAIN ;\type MPCC_BOT_GAIN_INSIDE ;\type MPCC_BOT_GAIN_OUTSIDE ;\ -type MPC_OCSC_TEST_DEBUG_DATA ;\ +type MPC_OCSC_TEST_DEBUG_DATA_OCSC_MODE ;\type MPC_OCSC_TEST_DEBUG_INDEX ;\type MPC_OCSC_MODE ;\type MPC_OCSC_C11_A ;\
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +6 -4
@@ -652,12 +652,12 @@ static const struct dcn2_dpp_registers tf_regs[] = {static const struct dcn2_dpp_shift tf_shift = {TF_REG_LIST_SH_MASK_DCN20 (__SHIFT), -TF_DEBUG_REG_LIST_SH_DCN10 +TF_DEBUG_REG_LIST_SH_DCN20 };static const struct dcn2_dpp_mask tf_mask = {TF_REG_LIST_SH_MASK_DCN20 (_MASK), -TF_DEBUG_REG_LIST_MASK_DCN10 +TF_DEBUG_REG_LIST_MASK_DCN20 };#define dwbc_regs_dcn2 (id)\
@@ -711,11 +711,13 @@ static const struct dcn20_mpc_registers mpc_regs = { };static const struct dcn20_mpc_shift mpc_shift = { -MPC_COMMON_MASK_SH_LIST_DCN2_0 (__SHIFT) +MPC_COMMON_MASK_SH_LIST_DCN2_0 (__SHIFT), +MPC_DEBUG_REG_LIST_SH_DCN20 };static const struct dcn20_mpc_mask mpc_mask = { -MPC_COMMON_MASK_SH_LIST_DCN2_0 (_MASK) +MPC_COMMON_MASK_SH_LIST_DCN2_0 (_MASK), +MPC_DEBUG_REG_LIST_MASK_DCN20 };#define tg_regs (id)\
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +6 -4
@@ -471,11 +471,13 @@ static const struct dcn20_mpc_registers mpc_regs = { };static const struct dcn20_mpc_shift mpc_shift = { -MPC_COMMON_MASK_SH_LIST_DCN2_0 (__SHIFT) +MPC_COMMON_MASK_SH_LIST_DCN2_0 (__SHIFT), +MPC_DEBUG_REG_LIST_SH_DCN20 };static const struct dcn20_mpc_mask mpc_mask = { -MPC_COMMON_MASK_SH_LIST_DCN2_0 (_MASK) +MPC_COMMON_MASK_SH_LIST_DCN2_0 (_MASK), +MPC_DEBUG_REG_LIST_MASK_DCN20 };#define hubp_regs (id)\
@@ -622,12 +624,12 @@ static const struct dcn2_dpp_registers tf_regs[] = {static const struct dcn2_dpp_shift tf_shift = {TF_REG_LIST_SH_MASK_DCN20 (__SHIFT), -TF_DEBUG_REG_LIST_SH_DCN10 +TF_DEBUG_REG_LIST_SH_DCN20 };static const struct dcn2_dpp_mask tf_mask = {TF_REG_LIST_SH_MASK_DCN20 (_MASK), -TF_DEBUG_REG_LIST_MASK_DCN10 +TF_DEBUG_REG_LIST_MASK_DCN20 };#define stream_enc_regs (id)\
drivers/gpu/drm/amd/display/dc/inc/reg_helper.h +13 -0
@@ -458,7 +458,14 @@ uint32_t generic_reg_get8(const struct dc_context *ctx, uint32_t addr,#define IX_REG_READ (index_reg_name, data_reg_name, index) \ generic_read_indirect_reg(CTX, REG(index_reg_name), REG(data_reg_name), IND_REG(index)) +#define IX_REG_GET_N (index_reg_name, data_reg_name, index, n, ...) \ + generic_indirect_reg_get(CTX, REG(index_reg_name), REG(data_reg_name), \ + IND_REG(index), \ + n, __VA_ARGS__) +#define IX_REG_GET (index_reg_name, data_reg_name, index, field, val) \ + IX_REG_GET_N(index_reg_name, data_reg_name, index, 1, \ + FN(data_reg_name, field), val)#define IX_REG_UPDATE_N (index_reg_name, data_reg_name, index, n, ...) \ generic_indirect_reg_update_ex(CTX, \
@@ -479,6 +486,12 @@ uint32_t generic_read_indirect_reg(const struct dc_context *ctx, uint32_t addr_index, uint32_t addr_data, uint32_t index); +uint32_t generic_indirect_reg_get (const struct dc_context *ctx, +uint32_t addr_index,uint32_t addr_data, +uint32_t index,int n, +uint8_t shift1,uint32_t mask1,uint32_t *field_value1, + ...); + uint32_t generic_indirect_reg_update_ex(const struct dc_context *ctx, uint32_t addr_index, uint32_t addr_data, uint32_t index, uint32_t reg_val, int n,
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 25/43 drm/amd/display: Refactor surface flip programming
In reply to Rodrigo SiqueiraFrom: Yongqiang Sun <[email protected]>
Rework surface programming for RN to separate preparing parameters and register programming.
--- .../gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c | 324 +++++++----------- .../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 16 +- 2 files changed, 127 insertions(+), 213 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c +118 -206
@@ -690,28 +690,114 @@ void hubp21_validate_dml_output(struct hubp *hubp, dml_dlg_attr->refcyc_per_meta_chunk_flip_l, dlg_attr.refcyc_per_meta_chunk_flip_l); } -static void program_video_progressive_dmcub ( -struct dc_dmub_srv *dmcub, +static void program_surface_flip_and_addr (struct hubp *hubp,struct dmub_rb_cmd_flip *surface_flip) +{ +struct dcn21_hubp *hubp21 =TO_DCN21_HUBP (hubp); + +REG_UPDATE_3 (DCSURF_FLIP_CONTROL , +SURFACE_FLIP_TYPE , surface_flip->flip.flip_params.immediate, +SURFACE_FLIP_MODE_FOR_STEREOSYNC , surface_flip->flip.flip_params.grph_stereo, +SURFACE_FLIP_IN_STEREOSYNC , surface_flip->flip.flip_params.grph_stereo); + +REG_UPDATE (VMID_SETTINGS_0 , +VMID , surface_flip->flip.flip_params.vmid); + +REG_UPDATE_8 (DCSURF_SURFACE_CONTROL , +PRIMARY_SURFACE_TMZ , surface_flip->flip.flip_params.tmz_surface, +PRIMARY_SURFACE_TMZ_C , surface_flip->flip.flip_params.tmz_surface, +PRIMARY_META_SURFACE_TMZ , surface_flip->flip.flip_params.tmz_surface, +PRIMARY_META_SURFACE_TMZ_C , surface_flip->flip.flip_params.tmz_surface, +SECONDARY_SURFACE_TMZ , surface_flip->flip.flip_params.tmz_surface, +SECONDARY_SURFACE_TMZ_C , surface_flip->flip.flip_params.tmz_surface, +SECONDARY_META_SURFACE_TMZ , surface_flip->flip.flip_params.tmz_surface, +SECONDARY_META_SURFACE_TMZ_C , surface_flip->flip.flip_params.tmz_surface); + +REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C ,0 , +PRIMARY_META_SURFACE_ADDRESS_HIGH_C , + surface_flip->flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C); + +REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_C ,0 , +PRIMARY_META_SURFACE_ADDRESS_C , + surface_flip->flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_C); + +REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH ,0 , +PRIMARY_META_SURFACE_ADDRESS_HIGH , + surface_flip->flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH); + +REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS ,0 , +PRIMARY_META_SURFACE_ADDRESS , + surface_flip->flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS); + +REG_SET (DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH ,0 , +SECONDARY_META_SURFACE_ADDRESS_HIGH , + surface_flip->flip.DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH); + +REG_SET (DCSURF_SECONDARY_META_SURFACE_ADDRESS ,0 , +SECONDARY_META_SURFACE_ADDRESS , + surface_flip->flip.DCSURF_SECONDARY_META_SURFACE_ADDRESS); + + +REG_SET (DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH ,0 , +SECONDARY_SURFACE_ADDRESS_HIGH , + surface_flip->flip.DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH); + +REG_SET (DCSURF_SECONDARY_SURFACE_ADDRESS ,0 , +SECONDARY_SURFACE_ADDRESS , + surface_flip->flip.DCSURF_SECONDARY_SURFACE_ADDRESS); + + +REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C ,0 , +PRIMARY_SURFACE_ADDRESS_HIGH_C , + surface_flip->flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C); + +REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_C ,0 , +PRIMARY_SURFACE_ADDRESS_C , + surface_flip->flip.DCSURF_PRIMARY_SURFACE_ADDRESS_C); + +REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH ,0 , +PRIMARY_SURFACE_ADDRESS_HIGH , + surface_flip->flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH); + +REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS ,0 , +PRIMARY_SURFACE_ADDRESS , + surface_flip->flip.DCSURF_PRIMARY_SURFACE_ADDRESS); +} + +void program_surface_flip_and_addr_dmcub (struct hubp *hubp,struct dmub_rb_cmd_flip *surface_flip) +{ +struct dc_dmub_srv *dmcub = hubp->ctx->dmub_srv; +struct dcn21_hubp *hubp21 =TO_DCN21_HUBP (hubp); + +PERF_TRACE ();// TODO: remove after performance is stable. +dc_dmub_srv_cmd_queue (dmcub, &surface_flip->header); +PERF_TRACE ();// TODO: remove after performance is stable. +dc_dmub_srv_cmd_execute (dmcub); +PERF_TRACE ();// TODO: remove after performance is stable. +dc_dmub_srv_wait_idle (dmcub); +PERF_TRACE ();// TODO: remove after performance is stable. +} + +bool hubp21_program_surface_flip_and_addr (struct hubp *hubp,const struct dc_plane_address *address,bool flip_immediate) { -struct dcn21_hubp *hubp21 =TO_DCN21_HUBP (hubp);struct dmub_rb_cmd_flip surface_flip = {0 }; +bool grph_stereo = false; +struct dc_debug_options *debug = &hubp->ctx->dc->debug; +struct dcn21_hubp *hubp21 =TO_DCN21_HUBP (hubp); surface_flip.header.type =DMUB_CMD__SURFACE_FLIP ; - surface_flip.flip.addr_type = address->type; - surface_flip.flip.immediate = flip_immediate; - surface_flip.flip.vmid = address->vmid; - - surface_flip.flip.hubp_inst = hubp->inst; - surface_flip.flip.tmz_surface = address->tmz_surface; + surface_flip.flip.flip_params.vmid = address->vmid; + surface_flip.flip.flip_params.hubp_inst = hubp->inst;switch (address->type) {case PLN_ADDR_TYPE_GRAPHICS : -if (address->grph.addr.quad_part ==0 ) -return ; +if (address->grph.addr.quad_part ==0 ) { +BREAK_TO_DEBUGGER (); +break ; + }if (address->grph.meta_addr.quad_part !=0 ) { surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS =
@@ -728,7 +814,7 @@ static void program_video_progressive_dmcub(case PLN_ADDR_TYPE_VIDEO_PROGRESSIVE :if (address->video_progressive.luma_addr.quad_part ==0 || address->video_progressive.chroma_addr.quad_part ==0 ) -return ; +break ;if (address->video_progressive.luma_meta_addr.quad_part !=0 ) { surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS =
@@ -747,19 +833,24 @@ static void program_video_progressive_dmcub( surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH = address->video_progressive.luma_addr.high_part; - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_C = - address->video_progressive.chroma_addr.low_part; +if (debug->nv12_iflip_vm_wa) { + surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_C = + address->video_progressive.chroma_addr.low_part + hubp21->PLAT_54186_wa_chroma_addr_offset; + }else + surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_C = + address->video_progressive.chroma_addr.low_part; + surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C = address->video_progressive.chroma_addr.high_part;break ;case PLN_ADDR_TYPE_GRPH_STEREO :if (address->grph_stereo.left_addr.quad_part ==0 ) -return ; +break ;if (address->grph_stereo.right_addr.quad_part ==0 ) -return ; +break ; - surface_flip.flip.grph_stereo = true; + grph_stereo = true;if (address->grph_stereo.right_meta_addr.quad_part !=0 ) { surface_flip.flip.DCSURF_SECONDARY_META_SURFACE_ADDRESS =
@@ -786,200 +877,21 @@ static void program_video_progressive_dmcub( address->grph_stereo.right_addr.high_part;break ; - - } - -PERF_TRACE ();// TODO: remove after performance is stable. -dc_dmub_srv_cmd_queue (dmcub, &surface_flip.header); -PERF_TRACE ();// TODO: remove after performance is stable. -dc_dmub_srv_cmd_execute (dmcub); -PERF_TRACE ();// TODO: remove after performance is stable. -dc_dmub_srv_wait_idle (dmcub); -} - -bool hubp21_program_surface_flip_and_addr ( -struct hubp *hubp, -const struct dc_plane_address *address, -bool flip_immediate) -{ -struct dcn21_hubp *hubp21 =TO_DCN21_HUBP (hubp); -struct dc_debug_options *debug = &hubp->ctx->dc->debug; - - -if (hubp->ctx->dc->debug.enable_dmcub_surface_flip) { -program_video_progressive_dmcub (hubp->ctx->dmub_srv, hubp, address, flip_immediate); - hubp->request_address = *address; -return true; - } - -//program flip type -REG_UPDATE (DCSURF_FLIP_CONTROL , -SURFACE_FLIP_TYPE , flip_immediate); - -// Program VMID reg -REG_UPDATE (VMID_SETTINGS_0 , -VMID , address->vmid); - -if (address->type ==PLN_ADDR_TYPE_GRPH_STEREO ) { -REG_UPDATE (DCSURF_FLIP_CONTROL ,SURFACE_FLIP_MODE_FOR_STEREOSYNC ,0x1 ); -REG_UPDATE (DCSURF_FLIP_CONTROL ,SURFACE_FLIP_IN_STEREOSYNC ,0x1 ); - - }else { -// turn off stereo if not in stereo -REG_UPDATE (DCSURF_FLIP_CONTROL ,SURFACE_FLIP_MODE_FOR_STEREOSYNC ,0x0 ); -REG_UPDATE (DCSURF_FLIP_CONTROL ,SURFACE_FLIP_IN_STEREOSYNC ,0x0 ); - } - - - -/* HW automatically latch rest of address register on write to -* DCSURF_PRIMARY_SURFACE_ADDRESS if SURFACE_UPDATE_LOCK is not used -* -* program high first and then the low addr, order matters! -*/ -switch (address->type) { -case PLN_ADDR_TYPE_GRAPHICS : -/* DCN1.0 does not support const color -* TODO: program DCHUBBUB_RET_PATH_DCC_CFGx_0/1 -* base on address->grph.dcc_const_color -* x = 0, 2, 4, 6 for pipe 0, 1, 2, 3 for rgb and luma -* x = 1, 3, 5, 7 for pipe 0, 1, 2, 3 for chroma -*/ - -if (address->grph.addr.quad_part ==0 ) -break ; - -REG_UPDATE_2 (DCSURF_SURFACE_CONTROL , -PRIMARY_SURFACE_TMZ , address->tmz_surface, -PRIMARY_META_SURFACE_TMZ , address->tmz_surface); - -if (address->grph.meta_addr.quad_part !=0 ) { -REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH ,0 , -PRIMARY_META_SURFACE_ADDRESS_HIGH , - address->grph.meta_addr.high_part); - -REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS ,0 , -PRIMARY_META_SURFACE_ADDRESS , - address->grph.meta_addr.low_part); - } - -REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH ,0 , -PRIMARY_SURFACE_ADDRESS_HIGH , - address->grph.addr.high_part); - -REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS ,0 , -PRIMARY_SURFACE_ADDRESS , - address->grph.addr.low_part); -break ; -case PLN_ADDR_TYPE_VIDEO_PROGRESSIVE : -if (address->video_progressive.luma_addr.quad_part ==0 - || address->video_progressive.chroma_addr.quad_part ==0 ) -break ; - -REG_UPDATE_4 (DCSURF_SURFACE_CONTROL , -PRIMARY_SURFACE_TMZ , address->tmz_surface, -PRIMARY_SURFACE_TMZ_C , address->tmz_surface, -PRIMARY_META_SURFACE_TMZ , address->tmz_surface, -PRIMARY_META_SURFACE_TMZ_C , address->tmz_surface); - -if (address->video_progressive.luma_meta_addr.quad_part !=0 ) { -REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C ,0 , -PRIMARY_META_SURFACE_ADDRESS_HIGH_C , - address->video_progressive.chroma_meta_addr.high_part); - -REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_C ,0 , -PRIMARY_META_SURFACE_ADDRESS_C , - address->video_progressive.chroma_meta_addr.low_part); - -REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH ,0 , -PRIMARY_META_SURFACE_ADDRESS_HIGH , - address->video_progressive.luma_meta_addr.high_part); - -REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS ,0 , -PRIMARY_META_SURFACE_ADDRESS , - address->video_progressive.luma_meta_addr.low_part); - } - -REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C ,0 , -PRIMARY_SURFACE_ADDRESS_HIGH_C , - address->video_progressive.chroma_addr.high_part); - -if (debug->nv12_iflip_vm_wa) { -REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_C ,0 , -PRIMARY_SURFACE_ADDRESS_C , - address->video_progressive.chroma_addr.low_part + hubp21->PLAT_54186_wa_chroma_addr_offset); - }else { -REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_C ,0 , -PRIMARY_SURFACE_ADDRESS_C , - address->video_progressive.chroma_addr.low_part); - } - -REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH ,0 , -PRIMARY_SURFACE_ADDRESS_HIGH , - address->video_progressive.luma_addr.high_part); - -REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS ,0 , -PRIMARY_SURFACE_ADDRESS , - address->video_progressive.luma_addr.low_part); -break ; -case PLN_ADDR_TYPE_GRPH_STEREO : -if (address->grph_stereo.left_addr.quad_part ==0 ) -break ; -if (address->grph_stereo.right_addr.quad_part ==0 ) -break ; - -REG_UPDATE_8 (DCSURF_SURFACE_CONTROL , -PRIMARY_SURFACE_TMZ , address->tmz_surface, -PRIMARY_SURFACE_TMZ_C , address->tmz_surface, -PRIMARY_META_SURFACE_TMZ , address->tmz_surface, -PRIMARY_META_SURFACE_TMZ_C , address->tmz_surface, -SECONDARY_SURFACE_TMZ , address->tmz_surface, -SECONDARY_SURFACE_TMZ_C , address->tmz_surface, -SECONDARY_META_SURFACE_TMZ , address->tmz_surface, -SECONDARY_META_SURFACE_TMZ_C , address->tmz_surface); - -if (address->grph_stereo.right_meta_addr.quad_part !=0 ) { - -REG_SET (DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH ,0 , -SECONDARY_META_SURFACE_ADDRESS_HIGH , - address->grph_stereo.right_meta_addr.high_part); - -REG_SET (DCSURF_SECONDARY_META_SURFACE_ADDRESS ,0 , -SECONDARY_META_SURFACE_ADDRESS , - address->grph_stereo.right_meta_addr.low_part); - } -if (address->grph_stereo.left_meta_addr.quad_part !=0 ) { - -REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH ,0 , -PRIMARY_META_SURFACE_ADDRESS_HIGH , - address->grph_stereo.left_meta_addr.high_part); - -REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS ,0 , -PRIMARY_META_SURFACE_ADDRESS , - address->grph_stereo.left_meta_addr.low_part); - } - -REG_SET (DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH ,0 , -SECONDARY_SURFACE_ADDRESS_HIGH , - address->grph_stereo.right_addr.high_part); - -REG_SET (DCSURF_SECONDARY_SURFACE_ADDRESS ,0 , -SECONDARY_SURFACE_ADDRESS , - address->grph_stereo.right_addr.low_part); - -REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH ,0 , -PRIMARY_SURFACE_ADDRESS_HIGH , - address->grph_stereo.left_addr.high_part); - -REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS ,0 , -PRIMARY_SURFACE_ADDRESS , - address->grph_stereo.left_addr.low_part); -break ;default :BREAK_TO_DEBUGGER ();break ; } + surface_flip.flip.flip_params.vmid = address->vmid; + surface_flip.flip.flip_params.grph_stereo = grph_stereo; + surface_flip.flip.flip_params.tmz_surface = address->tmz_surface; + surface_flip.flip.flip_params.immediate = flip_immediate; + +if (hubp->ctx->dc->debug.enable_dmcub_surface_flip) +program_surface_flip_and_addr_dmcub (hubp, &surface_flip); +else +program_surface_flip_and_addr (hubp, &surface_flip); + hubp->request_address = *address;return true;
drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +9 -7
@@ -30,7 +30,7 @@#include "dmub_cmd_dal.h" #include "dmub_cmd_vbios.h" #include "atomfirmware.h" -#include "dc_hw_types.h" +#define DMUB_RB_CMD_SIZE 64#define DMUB_RB_MAX_ENTRY 128#define DMUB_RB_SIZE (DMUB_RB_CMD_SIZE * DMUB_RB_MAX_ENTRY)
@@ -165,12 +165,14 @@ struct dmub_cmd_surface_flip {uint32_t DCSURF_SECONDARY_META_SURFACE_ADDRESS ;uint32_t DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH ;uint32_t DCSURF_SECONDARY_SURFACE_ADDRESS ; -enum dc_plane_addr_type addr_type; -uint8_t hubp_inst; -bool tmz_surface; -bool immediate; -uint8_t vmid; -bool grph_stereo; +struct { +uint8_t hubp_inst :4 ; +uint8_t tmz_surface :1 ; +uint8_t immediate :1 ; +uint8_t vmid :4 ; +uint8_t grph_stereo :1 ; +uint32_t reserved :21 ; + } flip_params; };struct dmub_rb_cmd_flip {
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 26/43 drm/amd/display: Fix DMUB PSR command IDs
In reply to Rodrigo SiqueiraFrom: Wyatt Wood <[email protected]>
[Why]
The DMUB PSR command IDs do not have the correct values.
[How]
Fix the command IDs and cleanup the formatting.
--- .../gpu/drm/amd/display/dmub/inc/dmub_cmd_dal.h | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-)
drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd_dal.h +8 -7
@@ -32,16 +32,17 @@*/ enum dmub_cmd_psr_type { -DMUB_CMD__PSR_ENABLE =0 , -DMUB_CMD__PSR_DISABLE =1 , -DMUB_CMD__PSR_COPY_SETTINGS =2 , -DMUB_CMD__PSR_SET_LEVEL =3 , -DMUB_CMD__PSR_SETUP =4 , +DMUB_CMD__PSR_SETUP =0 , +DMUB_CMD__PSR_COPY_SETTINGS =1 , +DMUB_CMD__PSR_ENABLE =2 , +DMUB_CMD__PSR_DISABLE =3 , +DMUB_CMD__PSR_SET_LEVEL =4 , };enum psr_version { -PSR_VERSION_1 =0x0 , -PSR_VERSION_2 =0x10 , +PSR_VERSION_1 =0x10 ,// PSR Version 1 +PSR_VERSION_2 =0x20 ,// PSR Version 2, includes selective update +PSR_VERSION_2_Y_COORD =0x21 ,// PSR Version 2, includes Y-coordinate support for SU };#endif /* _DMUB_CMD_DAL_H_ */
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 27/43 drm/amd/display: Add w/a to reset PHY before link training in verify_link_cap
In reply to Rodrigo SiqueiraFrom: George Shen <[email protected]>
[Why]
PHY will sometimes be in bad state on hotplugging display from USB-C dongle.
[How]
Add additional calls to disable and then enable PHY before link training starts during verify_link_cap.
--- drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c | 10 ++++++++++ drivers/gpu/drm/amd/display/dc/dc.h | 1 + drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c | 3 ++- 3 files changed, 13 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +10 -0
@@ -1892,6 +1892,16 @@ bool dp_verify_link_cap(/* disable PHY done possible by BIOS, will be done by driver itself */ dp_disable_link_phy (link, link->connector_signal); +/* Temporary Renoir-specific workaround for SWDEV-215184; +* PHY will sometimes be in bad state on hotplugging display from certain USB-C dongle, +* so add extra cycle of enabling and disabling the PHY before first link training. +*/ +if (link->link_enc->features.flags.bits.DP_IS_USB_C && + link->dc->debug.usbc_combo_phy_reset_wa) { +dp_enable_link_phy (link, link->connector_signal, dp_cs_id, cur); +dp_disable_link_phy (link, link->connector_signal); + } + dp_cs_id =get_clock_source_id (link);/* link training starts with the maximum common settings
drivers/gpu/drm/amd/display/dc/dc.h +1 -0
@@ -421,6 +421,7 @@ struct dc_debug_options {bool disable_dram_clock_change_vactive_support;bool validate_dml_output;bool enable_dmcub_surface_flip; +bool usbc_combo_phy_reset_wa; };struct dc_debug_data {
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +2 -1
@@ -835,7 +835,8 @@ static const struct dc_debug_options debug_defaults_drv = { .scl_reset_length10 = true, .sanity_checks = true, .disable_48mhz_pwrdwn = false, - .nv12_iflip_vm_wa = true + .nv12_iflip_vm_wa = true, + .usbc_combo_phy_reset_wa = true };static const struct dc_debug_options debug_defaults_diags = {
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 28/43 drm/amd/display: rename _lvp to l_vp
In reply to Rodrigo SiqueiraFrom: Charlene Liu <[email protected]>
--- .../gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20.c | 4 ++-- .../gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20v2.c | 4 ++-- .../gpu/drm/amd/display/dc/dml/dcn21/display_mode_vba_21.c | 4 ++-- drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h | 2 +- 4 files changed, 7 insertions(+), 7 deletions(-)
drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20.c +2 -2
@@ -937,7 +937,7 @@ static unsigned int CalculateVMAndRowBytes( *MetaRowByte =0 ; } -if (SurfaceTiling == dm_sw_linear || SurfaceTiling == dm_sw_gfx7_2d_thin_gl || SurfaceTiling == dm_sw_gfx7_2d_thin_lvp) { +if (SurfaceTiling == dm_sw_linear || SurfaceTiling == dm_sw_gfx7_2d_thin_gl || SurfaceTiling == dm_sw_gfx7_2d_thin_l_vp) { MacroTileSizeBytes =256 ; MacroTileHeight = BlockHeight256Bytes; } else if (SurfaceTiling == dm_sw_4kb_s || SurfaceTiling == dm_sw_4kb_s_x
@@ -3348,7 +3348,7 @@ void dml20_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l == dm_420_10)) || (((mode_lib->vba.SurfaceTiling[k] == dm_sw_gfx7_2d_thin_gl || mode_lib->vba.SurfaceTiling[k] - == dm_sw_gfx7_2d_thin_lvp) + == dm_sw_gfx7_2d_thin_l_vp) && !((mode_lib->vba.SourcePixelFormat[k] == dm_444_64 || mode_lib->vba.SourcePixelFormat[k]
drivers/gpu/drm/amd/display/dc/dml/dcn20/display_mode_vba_20v2.c +2 -2
@@ -997,7 +997,7 @@ static unsigned int CalculateVMAndRowBytes( *MetaRowByte =0 ; } -if (SurfaceTiling == dm_sw_linear || SurfaceTiling == dm_sw_gfx7_2d_thin_gl || SurfaceTiling == dm_sw_gfx7_2d_thin_lvp) { +if (SurfaceTiling == dm_sw_linear || SurfaceTiling == dm_sw_gfx7_2d_thin_gl || SurfaceTiling == dm_sw_gfx7_2d_thin_l_vp) { MacroTileSizeBytes =256 ; MacroTileHeight = BlockHeight256Bytes; } else if (SurfaceTiling == dm_sw_4kb_s || SurfaceTiling == dm_sw_4kb_s_x
@@ -3385,7 +3385,7 @@ void dml20v2_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode == dm_420_10)) || (((mode_lib->vba.SurfaceTiling[k] == dm_sw_gfx7_2d_thin_gl || mode_lib->vba.SurfaceTiling[k] - == dm_sw_gfx7_2d_thin_lvp) + == dm_sw_gfx7_2d_thin_l_vp) && !((mode_lib->vba.SourcePixelFormat[k] == dm_444_64 || mode_lib->vba.SourcePixelFormat[k]
drivers/gpu/drm/amd/display/dc/dml/dcn21/display_mode_vba_21.c +2 -2
@@ -1338,7 +1338,7 @@ static unsigned int CalculateVMAndRowBytes( *MetaRowByte =0 ; } -if (SurfaceTiling == dm_sw_linear || SurfaceTiling == dm_sw_gfx7_2d_thin_gl || SurfaceTiling == dm_sw_gfx7_2d_thin_lvp) { +if (SurfaceTiling == dm_sw_linear || SurfaceTiling == dm_sw_gfx7_2d_thin_gl || SurfaceTiling == dm_sw_gfx7_2d_thin_l_vp) { MacroTileSizeBytes =256 ; MacroTileHeight = BlockHeight256Bytes; } else if (SurfaceTiling == dm_sw_4kb_s || SurfaceTiling == dm_sw_4kb_s_x
@@ -3453,7 +3453,7 @@ void dml21_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l == dm_420_10)) || (((mode_lib->vba.SurfaceTiling[k] == dm_sw_gfx7_2d_thin_gl || mode_lib->vba.SurfaceTiling[k] - == dm_sw_gfx7_2d_thin_lvp) + == dm_sw_gfx7_2d_thin_l_vp) && !((mode_lib->vba.SourcePixelFormat[k] == dm_444_64 || mode_lib->vba.SourcePixelFormat[k]
drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h +1 -1
@@ -85,7 +85,7 @@ enum dm_swizzle_mode { dm_sw_var_s_x =29 , dm_sw_var_d_x =30 , dm_sw_64kb_r_x, - dm_sw_gfx7_2d_thin_lvp, + dm_sw_gfx7_2d_thin_l_vp, dm_sw_gfx7_2d_thin_gl, }; enum lb_depth {
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 29/43 drm/amd/display: Use SMU ClockTable Values for DML Calculations
In reply to Rodrigo SiqueiraFrom: Sung Lee <[email protected]>
[Why]
DML Initialization was previously done on dcn21_resource initialization. This meant that DML soc struct was populated with hardcoded values.
[How]
Move DML initialization to after updating bounding box, to use clock table values from SMU.
--- drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +4 -0
@@ -1349,6 +1349,10 @@ static void update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_param } dcn2_1_soc.clock_limits[i] = dcn2_1_soc.clock_limits[i -1 ]; dcn2_1_soc.num_states = i; + +// diags does not retrieve proper values from SMU, do not update DML instance for diags +if (!IS_FPGA_MAXIMUS_DC (dc->ctx->dce_environment) && !IS_DIAG_DC (dc->ctx->dce_environment)) +dml_init_instance (&dc->dml, &dcn2_1_soc, &dcn2_1_ip,DML_PROJECT_DCN21 ); }/* Temporary Place holder until we can get them from fuse */
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 30/43 drm/amd/display: Add default switch case for DCC
In reply to Rodrigo SiqueiraFrom: Chris Park <[email protected]>
--- drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c | 3 +++ drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubbub.c | 3 +++ 2 files changed, 6 insertions(+)
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c +3 -0
@@ -930,6 +930,9 @@ static bool hubbub1_get_dcc_compression_cap(struct hubbub *hubbub, output->grph.rgb.max_compressed_blk_size =64 ; output->grph.rgb.independent_64b_blks = true;break ; +default : +ASSERT (false); +break ; } output->capable = true;
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubbub.c +3 -0
@@ -293,6 +293,9 @@ bool hubbub2_get_dcc_compression_cap(struct hubbub *hubbub, output->grph.rgb.max_compressed_blk_size =64 ; output->grph.rgb.independent_64b_blks = true;break ; +default : +ASSERT (false); +break ; } output->capable = true; output->const_color_support = true;
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 31/43 drm/amd/display: implement fw-driver interface for abm 2.4
In reply to Rodrigo SiqueiraFrom: Josip Pavic <[email protected]>
[Why]
IRAM definition needed for versions of DMCU containing ABM 2.4
[How]
Pass ABM 2.3 IRAM definition, which is compatible with ABM 2.4, to DMCU when ABM 2.4 FW is detected
--- drivers/gpu/drm/amd/display/modules/power/power_helpers.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/display/modules/power/power_helpers.c +6 -1
@@ -662,7 +662,11 @@ bool dmcu_load_iram(struct dmcu *dmcu,memset (&ram_table,0 ,sizeof (ram_table)); -if (dmcu->dmcu_version.abm_version ==0x23 ) { +if (dmcu->dmcu_version.abm_version ==0x24 ) { +fill_iram_v_2_3 ((struct iram_table_v_2_2 *)ram_table, params); + result = dmcu->funcs->load_iram ( + dmcu,0 , (char *)(&ram_table),IRAM_RESERVE_AREA_START_V2_2 ); + }else if (dmcu->dmcu_version.abm_version ==0x23 ) {fill_iram_v_2_3 ((struct iram_table_v_2_2 *)ram_table, params); result = dmcu->funcs->load_iram(
@@ -687,3 +691,4 @@ bool dmcu_load_iram(struct dmcu *dmcu,return result; } +
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 32/43 drm/amd/display: remove psr state condition when psr exit case
In reply to Rodrigo SiqueiraFrom: Lewis Huang <[email protected]>
[Why]
DMCU need to wait a vblank to handle psr enter command. When psr exit coming before vblank, the psr exit command will be skip because current psr state is disable.
[How]
remove psr state condition when psr exit case
--- drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c | 3 --- 1 file changed, 3 deletions(-)
drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c +0 -3
@@ -537,9 +537,6 @@ static void dcn10_dmcu_set_psr_enable(struct dmcu *dmcu, bool enable, bool wait)if (dmcu->dmcu_state !=DMCU_RUNNING )return ; -dcn10_get_dmcu_psr_state (dmcu, &psr_state); -if (psr_state ==0 && !enable) -return ;/* waitDMCUReadyForCmd */ REG_WAIT(MASTER_COMM_CNTL_REG, MASTER_COMM_INTERRUPT, 0, dmcu_wait_reg_ready_interval,
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 33/43 drm/amd/display: 3.2.67
In reply to Rodrigo SiqueiraFrom: Aric Cyr <[email protected]>
--- drivers/gpu/drm/amd/display/dc/dc.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/dc.h +1 -1
@@ -39,7 +39,7 @@#include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#define DC_VER "3.2.66" +#define DC_VER "3.2.67"#define MAX_SURFACES 3#define MAX_PLANES 6
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 34/43 drm/amd/display: make PSR static screen entry within 30 ms
In reply to Rodrigo SiqueiraFrom: Anthony Koo <[email protected]>
[Why]
With different refresh rate panels, the PSR entry/exit time is different since it is dependent on 2 frame entry time today
[How]
Make static screen num frame entry time to be calculated such that entry time is within 30 ms instead of fixed num frames.
--- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 32 +++++++++++++---- drivers/gpu/drm/amd/display/dc/core/dc.c | 6 ++-- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 2 +- drivers/gpu/drm/amd/display/dc/dc.h | 13 ++++--- drivers/gpu/drm/amd/display/dc/dc_stream.h | 4 +-- drivers/gpu/drm/amd/display/dc/dc_types.h | 2 +- .../display/dc/dce110/dce110_hw_sequencer.c | 35 +++++++++++-------- .../dc/dce110/dce110_timing_generator.c | 11 ++++-- .../dc/dce110/dce110_timing_generator.h | 3 +- .../dc/dce120/dce120_timing_generator.c | 11 ++++-- .../amd/display/dc/dcn10/dcn10_hw_sequencer.c | 23 ++++++------ .../amd/display/dc/dcn10/dcn10_hw_sequencer.h | 2 +- .../gpu/drm/amd/display/dc/dcn10/dcn10_optc.c | 15 +++++--- .../gpu/drm/amd/display/dc/dcn10/dcn10_optc.h | 3 +- .../drm/amd/display/dc/dcn20/dcn20_hwseq.c | 6 +++- .../amd/display/dc/inc/hw/timing_generator.h | 3 +- .../gpu/drm/amd/display/dc/inc/hw_sequencer.h | 4 +-- .../amd/display/dc/inc/hw_sequencer_private.h | 2 +- 18 files changed, 116 insertions(+), 61 deletions(-)
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +26 -6
@@ -8474,17 +8474,37 @@ static bool amdgpu_dm_link_setup_psr(struct dc_stream_state *stream)bool amdgpu_dm_psr_enable (struct dc_stream_state *stream) {struct dc_link *link = stream->link; -struct dc_static_screen_events triggers = {0 }; +unsigned int vsync_rate_hz =0 ; +struct dc_static_screen_params params = {0 }; +/* Calculate number of static frames before generating interrupt to +* enter PSR. +*/ +unsigned int frame_time_microsec =1000000 / vsync_rate_hz; +// Init fail safe of 2 frames static +unsigned int num_frames_static =2 ;DRM_DEBUG_DRIVER ("Enabling psr...\n" ); - triggers.cursor_update = true; - triggers.overlay_update = true; - triggers.surface_update = true; + vsync_rate_hz =div64_u64 (div64_u64 (( + stream->timing.pix_clk_100hz *100 ), + stream->timing.v_total), + stream->timing.h_total); + +/* Round up +* Calculate number of frames such that at least 30 ms of time has +* passed. +*/ +if (vsync_rate_hz !=0 ) + num_frames_static = (30000 / frame_time_microsec) +1 ; + + params.triggers.cursor_update = true; + params.triggers.overlay_update = true; + params.triggers.surface_update = true; + params.num_frames = num_frames_static; -dc_stream_set_static_screen_events (link->ctx->dc, +dc_stream_set_static_screen_params (link->ctx->dc, &stream,1 , - &triggers); + ¶ms);return dc_link_set_psr_allow_active (link, true, false); }
drivers/gpu/drm/amd/display/dc/core/dc.c +3 -3
@@ -510,10 +510,10 @@ bool dc_stream_program_csc_matrix(struct dc *dc, struct dc_stream_state *stream)return ret; } -void dc_stream_set_static_screen_events (struct dc *dc, +void dc_stream_set_static_screen_params (struct dc *dc,struct dc_stream_state **streams,int num_streams, -const struct dc_static_screen_events *events) +const struct dc_static_screen_params *params) {int i =0 ;int j =0 ;
@@ -532,7 +532,7 @@ void dc_stream_set_static_screen_events(struct dc *dc, } } - dc->hwss.set_static_screen_control (pipes_affected, num_pipes_affected, events); + dc->hwss.set_static_screen_control (pipes_affected, num_pipes_affected, params); }static void dc_destruct (struct dc *dc)
drivers/gpu/drm/amd/display/dc/core/dc_link.c +1 -1
@@ -2542,7 +2542,7 @@ bool dc_link_setup_psr(struct dc_link *link,transmitter_to_phy_id (link->link_enc->transmitter); psr_context->crtcTimingVerticalTotal = stream->timing.v_total; - psr_context->vsyncRateHz =div64_u64 (div64_u64 ((stream-> + psr_context->vsync_rate_hz =div64_u64 (div64_u64 ((stream-> timing.pix_clk_100hz *100 ), stream->timing.v_total), stream->timing.h_total);
drivers/gpu/drm/amd/display/dc/dc.h +8 -5
@@ -157,11 +157,14 @@ struct dc_surface_dcc_cap {bool const_color_support; }; -struct dc_static_screen_events { -bool force_trigger; -bool cursor_update; -bool surface_update; -bool overlay_update; +struct dc_static_screen_params { +struct { +bool force_trigger; +bool cursor_update; +bool surface_update; +bool overlay_update; + } triggers; +unsigned int num_frames; };
drivers/gpu/drm/amd/display/dc/dc_stream.h +2 -2
@@ -439,10 +439,10 @@ bool dc_stream_get_crc(struct dc *dc,uint32_t *g_y, uint32_t *b_cb); -void dc_stream_set_static_screen_events (struct dc *dc, +void dc_stream_set_static_screen_params (struct dc *dc,struct dc_stream_state **stream,int num_streams, -const struct dc_static_screen_events *events); +const struct dc_static_screen_params *params);void dc_stream_set_dyn_expansion (struct dc *dc,struct dc_stream_state *stream,enum dc_dynamic_expansion option);
drivers/gpu/drm/amd/display/dc/dc_types.h +1 -1
@@ -729,7 +729,7 @@ struct psr_context {/* The VSync rate in Hz used to calculate the * step size for smooth brightness feature */ -unsigned int vsyncRateHz; +unsigned int vsync_rate_hz;unsigned int skipPsrWaitForPllLock;unsigned int numberOfControllers;/* Unused, for future use. To indicate that first changed frame from
drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c +21 -14
@@ -1373,9 +1373,13 @@ static enum dc_status apply_single_controller_ctx_to_hw(// DRR should set trigger event to monitor surface update event if (stream->adjust.v_total_min !=0 && stream->adjust.v_total_max !=0 ) event_triggers =0x80 ; +/* Event triggers and num frames initialized for DRR, but can be +* later updated for PSR use. Note DRR trigger events are generated +* regardless of whether num frames met. +*/ if (pipe_ctx->stream_res.tg->funcs->set_static_screen_control) pipe_ctx->stream_res.tg->funcs->set_static_screen_control ( - pipe_ctx->stream_res.tg, event_triggers); + pipe_ctx->stream_res.tg, event_triggers,2 );if (!dc_is_virtual_signal (pipe_ctx->stream->signal)) pipe_ctx->stream_res.stream_enc->funcs->dig_connect_to_otg(
@@ -1706,6 +1710,8 @@ static void set_drr(struct pipe_ctx **pipe_ctx,struct drr_params params = {0 };// DRR should set trigger event to monitor surface update event unsigned int event_triggers =0x80 ; +// Note DRR trigger events are generated regardless of whether num frames met. +unsigned int num_frames =2 ; params.vertical_total_max = vmax; params.vertical_total_min = vmin;
@@ -1721,7 +1727,7 @@ static void set_drr(struct pipe_ctx **pipe_ctx,if (vmax !=0 && vmin !=0 ) pipe_ctx[i]->stream_res.tg->funcs->set_static_screen_control ( pipe_ctx[i]->stream_res.tg, - event_triggers); + event_triggers, num_frames); } }
@@ -1738,30 +1744,31 @@ static void get_position(struct pipe_ctx **pipe_ctx, }static void set_static_screen_control (struct pipe_ctx **pipe_ctx, -int num_pipes,const struct dc_static_screen_events *events) +int num_pipes,const struct dc_static_screen_params *params) {unsigned int i; -unsigned int value =0 ; +unsigned int triggers =0 ; -if (events->overlay_update) - value |=0x100 ; -if (events->surface_update) - value |=0x80 ; -if (events->cursor_update) - value |=0x2 ; -if (events->force_trigger) - value |=0x1 ; +if (params->triggers.overlay_update) + triggers |=0x100 ; +if (params->triggers.surface_update) + triggers |=0x80 ; +if (params->triggers.cursor_update) + triggers |=0x2 ; +if (params->triggers.force_trigger) + triggers |=0x1 ;if (num_pipes) {struct dc *dc = pipe_ctx[0 ]->stream->ctx->dc;if (dc->fbc_compressor) - value |=0x84 ; + triggers |=0x84 ; }for (i =0 ; i < num_pipes; i++) pipe_ctx[i]->stream_res.tg->funcs-> -set_static_screen_control (pipe_ctx[i]->stream_res.tg, value); +set_static_screen_control (pipe_ctx[i]->stream_res.tg, + triggers, params->num_frames); }/*
drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.c +8 -3
@@ -469,22 +469,27 @@ void dce110_timing_generator_set_drr(void dce110_timing_generator_set_static_screen_control (struct timing_generator *tg, -uint32_t value) +uint32_t event_triggers, +uint32_t num_frames) {struct dce110_timing_generator *tg110 =DCE110TG_FROM_TG (tg);uint32_t static_screen_cntl =0 ;uint32_t addr =0 ; +// By register spec, it only takes 8 bit value +if (num_frames >0xFF ) + num_frames =0xFF ; + addr =CRTC_REG (mmCRTC_STATIC_SCREEN_CONTROL); static_screen_cntl =dm_read_reg (tg->ctx, addr);set_reg_field_value (static_screen_cntl, - value, + event_triggers,CRTC_STATIC_SCREEN_CONTROL ,CRTC_STATIC_SCREEN_EVENT_MASK );set_reg_field_value (static_screen_cntl, -2 , + num_frames,CRTC_STATIC_SCREEN_CONTROL ,CRTC_STATIC_SCREEN_FRAME_COUNT );
drivers/gpu/drm/amd/display/dc/dce110/dce110_timing_generator.h +2 -1
@@ -231,7 +231,8 @@ void dce110_timing_generator_set_drr(void dce110_timing_generator_set_static_screen_control (struct timing_generator *tg, -uint32_t value); +uint32_t event_triggers, +uint32_t num_frames); void dce110_timing_generator_get_crtc_scanoutpos( struct timing_generator *tg,
drivers/gpu/drm/amd/display/dc/dce120/dce120_timing_generator.c +8 -3
@@ -819,13 +819,18 @@ void dce120_tg_set_colors(struct timing_generator *tg,static void dce120_timing_generator_set_static_screen_control (struct timing_generator *tg, -uint32_t value) +uint32_t event_triggers, +uint32_t num_frames) {struct dce110_timing_generator *tg110 =DCE110TG_FROM_TG (tg); +// By register spec, it only takes 8 bit value +if (num_frames >0xFF ) + num_frames =0xFF ; +CRTC_REG_UPDATE_2 (CRTC0_CRTC_STATIC_SCREEN_CONTROL , -CRTC_STATIC_SCREEN_EVENT_MASK , value, -CRTC_STATIC_SCREEN_FRAME_COUNT ,2 ); +CRTC_STATIC_SCREEN_EVENT_MASK , event_triggers, +CRTC_STATIC_SCREEN_FRAME_COUNT , num_frames); } void dce120_timing_generator_set_test_pattern(
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +13 -10
@@ -2704,6 +2704,8 @@ void dcn10_set_drr(struct pipe_ctx **pipe_ctx,struct drr_params params = {0 };// DRR set trigger event mapped to OTG_TRIG_A (bit 11) for manual control flow unsigned int event_triggers =0x800 ; +// Note DRR trigger events are generated regardless of whether num frames met. +unsigned int num_frames =2 ; params.vertical_total_max = vmax; params.vertical_total_min = vmin;
@@ -2720,7 +2722,7 @@ void dcn10_set_drr(struct pipe_ctx **pipe_ctx,if (vmax !=0 && vmin !=0 ) pipe_ctx[i]->stream_res.tg->funcs->set_static_screen_control ( pipe_ctx[i]->stream_res.tg, - event_triggers); + event_triggers, num_frames); } }
@@ -2737,21 +2739,22 @@ void dcn10_get_position(struct pipe_ctx **pipe_ctx, }void dcn10_set_static_screen_control (struct pipe_ctx **pipe_ctx, -int num_pipes,const struct dc_static_screen_events *events) +int num_pipes,const struct dc_static_screen_params *params) {unsigned int i; -unsigned int value =0 ; +unsigned int triggers =0 ; -if (events->surface_update) - value |=0x80 ; -if (events->cursor_update) - value |=0x2 ; -if (events->force_trigger) - value |=0x1 ; +if (params->triggers.surface_update) + triggers |=0x80 ; +if (params->triggers.cursor_update) + triggers |=0x2 ; +if (params->triggers.force_trigger) + triggers |=0x1 ;for (i =0 ; i < num_pipes; i++) pipe_ctx[i]->stream_res.tg->funcs-> -set_static_screen_control (pipe_ctx[i]->stream_res.tg, value); +set_static_screen_control (pipe_ctx[i]->stream_res.tg, + triggers, params->num_frames); } static void dcn10_config_stereo_parameters(
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h +1 -1
@@ -132,7 +132,7 @@ void dcn10_get_position(struct pipe_ctx **pipe_ctx,int num_pipes, struct crtc_position *position);void dcn10_set_static_screen_control (struct pipe_ctx **pipe_ctx, -int num_pipes,const struct dc_static_screen_events *events); +int num_pipes,const struct dc_static_screen_params *params);void dcn10_setup_stereo (struct pipe_ctx *pipe_ctx,struct dc *dc);void dce110_set_avmute (struct pipe_ctx *pipe_ctx,bool enable); void dcn10_log_hw_state(struct dc *dc,
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c +10 -5
@@ -789,21 +789,26 @@ void optc1_set_early_control(void optc1_set_static_screen_control (struct timing_generator *optc, -uint32_t value) +uint32_t event_triggers, +uint32_t num_frames) {struct optc *optc1 =DCN10TG_FROM_TG (optc); +// By register spec, it only takes 8 bit value +if (num_frames >0xFF ) + num_frames =0xFF ; +/* Bit 8 is no longer applicable in RV for PSR case, * set bit 8 to 0 if given */ -if ((value &STATIC_SCREEN_EVENT_MASK_RANGETIMING_DOUBLE_BUFFER_UPDATE_EN ) +if ((event_triggers &STATIC_SCREEN_EVENT_MASK_RANGETIMING_DOUBLE_BUFFER_UPDATE_EN ) !=0 ) - value = value & + event_triggers = event_triggers & ~STATIC_SCREEN_EVENT_MASK_RANGETIMING_DOUBLE_BUFFER_UPDATE_EN ;REG_SET_2 (OTG_STATIC_SCREEN_CONTROL ,0 , -OTG_STATIC_SCREEN_EVENT_MASK , value, -OTG_STATIC_SCREEN_FRAME_COUNT ,2 ); +OTG_STATIC_SCREEN_EVENT_MASK , event_triggers, +OTG_STATIC_SCREEN_FRAME_COUNT , num_frames); }void optc1_setup_manual_trigger (struct timing_generator *optc)
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.h +2 -1
@@ -625,7 +625,8 @@ void optc1_set_drr(void optc1_set_static_screen_control (struct timing_generator *optc, -uint32_t value); +uint32_t event_triggers, +uint32_t num_frames);void optc1_program_stereo (struct timing_generator *optc,const struct dc_crtc_timing *timing,struct crtc_stereo_flags *flags);
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +5 -1
@@ -686,9 +686,13 @@ enum dc_status dcn20_enable_stream_timing(// DRR should set trigger event to monitor surface update event if (stream->adjust.v_total_min !=0 && stream->adjust.v_total_max !=0 ) event_triggers =0x80 ; +/* Event triggers and num frames initialized for DRR, but can be +* later updated for PSR use. Note DRR trigger events are generated +* regardless of whether num frames met. +*/ if (pipe_ctx->stream_res.tg->funcs->set_static_screen_control) pipe_ctx->stream_res.tg->funcs->set_static_screen_control ( - pipe_ctx->stream_res.tg, event_triggers); + pipe_ctx->stream_res.tg, event_triggers,2 );/* TODO program crtc source select for non-virtual signal*/ /* TODO program FMT */
drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h +2 -1
@@ -208,7 +208,8 @@ struct timing_generator_funcs { bool enable, const struct dc_crtc_timing *timing);void (*set_drr)(struct timing_generator *tg,const struct drr_params *params);void (*set_static_screen_control)(struct timing_generator *tg, -uint32_t value); +uint32_t event_triggers, +uint32_t num_frames); void (*set_test_pattern)( struct timing_generator *tg, enum controller_dp_test_pattern test_pattern,
drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h +2 -2
@@ -42,7 +42,7 @@ struct dc_state;struct dc_stream_status ;struct dc_writeback_info ;struct dchub_init_data ; -struct dc_static_screen_events ; +struct dc_static_screen_params ;struct resource_pool ;struct dc_phy_addr_space_config ;struct dc_virtual_addr_space_config ;
@@ -102,7 +102,7 @@ struct hw_sequencer_funcs { unsigned int vmid, unsigned int vmid_frame_number);void (*set_static_screen_control)(struct pipe_ctx **pipe_ctx,int num_pipes, -const struct dc_static_screen_events *events); +const struct dc_static_screen_params *events);/* Stream Related */ void (*enable_stream)(struct pipe_ctx *pipe_ctx);
drivers/gpu/drm/amd/display/dc/inc/hw_sequencer_private.h +1 -1
@@ -51,7 +51,7 @@ struct dc_state;struct dc_stream_status ;struct dc_writeback_info ;struct dchub_init_data ; -struct dc_static_screen_events ; +struct dc_static_screen_params ;struct resource_pool ;struct resource_context ;struct stream_resource ;
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 35/43 drm/amd/display: wait for test pattern after when all pipes are programmed
In reply to Rodrigo SiqueiraFrom: Wenjing Liu <[email protected]>
[why]
Currently we wait for test pattern after each pipe is programmed. For ODM combined scenario it will cause test pattern is shown on only half screen for 1 frame. This is not desirable.
[how]
No wait between odm pipe programming, only wait after all pipes are programmed.
--- drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c | 6 ------ 1 file changed, 6 deletions(-)
drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +0 -6
@@ -3658,12 +3658,6 @@ static void set_crtc_test_pattern(struct dc_link *link,NULL , width, height); -/* wait for dpg to blank pixel data with test pattern */ -for (count =0 ; count <1000 ; count++) -if (odm_opp->funcs->dpg_is_blanked (odm_opp)) -break ; -else -udelay (100 ); } opp->funcs->opp_set_disp_pattern_generator(opp, controller_test_pattern,
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 36/43 drm/amd/display: Add monitor patch for AUO dpcd issue
In reply to Rodrigo SiqueiraFrom: Lewis Huang <[email protected]>
[Why]
dpcd cap mismatch in 2200 vs base
[How]
Add monitor patch which using based caps to overwrite 2200
--- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 7 +++ .../gpu/drm/amd/display/dc/core/dc_link_dp.c | 62 +++++++++++++++++++ drivers/gpu/drm/amd/display/dc/dc_link.h | 3 + .../gpu/drm/amd/display/dc/inc/dc_link_dp.h | 2 + 4 files changed, 74 insertions(+)
drivers/gpu/drm/amd/display/dc/core/dc_link.c +7 -0
@@ -3371,3 +3371,10 @@ const struct dc_link_settings *dc_link_get_link_cap(return &link->preferred_link_setting;return &link->verified_link_cap; } + +void dc_link_overwrite_extended_receiver_cap ( +struct dc_link *link) +{ +dp_overwrite_extended_receiver_cap (link); +} +
drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +62 -0
68 lines not on this page · raw message
drivers/gpu/drm/amd/display/dc/dc_link.h +3 -0
@@ -301,6 +301,9 @@ uint32_t dc_link_bandwidth_kbps(const struct dc_link_settings *dc_link_get_link_cap (const struct dc_link *link); +void dc_link_overwrite_extended_receiver_cap ( +struct dc_link *link); + bool dc_submit_i2c( struct dc *dc, uint32_t link_index,
drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h +2 -0
@@ -76,6 +76,8 @@ void dp_enable_mst_on_sink(struct dc_link *link, bool enable);enum dp_panel_mode dp_get_panel_mode (struct dc_link *link);void dp_set_panel_mode (struct dc_link *link,enum dp_panel_mode panel_mode); +bool dp_overwrite_extended_receiver_cap (struct dc_link *link); +void dp_set_fec_ready (struct dc_link *link,bool ready);void dp_set_fec_enable (struct dc_link *link,bool enable);bool dp_set_dsc_enable (struct pipe_ctx *pipe_ctx,bool enable);
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 37/43 drm/amd/display: Enable double buffer for OTG_BLANK
In reply to Rodrigo SiqueiraFrom: Alvin Lee <[email protected]>
[Why]
Currently if seamless boot is enabled, we will skip double buffer enable for OTG_BLANK. However, we need the double buffer enable in order to block global sync signals when OTG becomes blanked (for PSR). Blocking global sync signals prevent pipe from requesting data.
[How]
Move tg_init before seamless boot check.
--- drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +7 -1
@@ -1188,8 +1188,14 @@ void dcn10_init_pipes(struct dc *dc, struct dc_state *context)if (can_apply_seamless_boot && pipe_ctx->stream !=NULL && pipe_ctx->stream_res.tg->funcs->is_tg_enabled ( - pipe_ctx->stream_res.tg)) + pipe_ctx->stream_res.tg)) { +// Enable double buffering for OTG_BLANK no matter if +// seamless boot is enabled or not to suppress global sync +// signals when OTG blanked. This is to prevent pipe from +// requesting data while in PSR. + tg->funcs->tg_init (tg);continue ; + }/* Disable on the current state so the new one isn't cleared. */ pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i];
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 38/43 drm/amd/display: Only program surface flip for video plane via dmcub
In reply to Rodrigo SiqueiraFrom: Yongqiang Sun <[email protected]>
Only need to do surface flip for video plane via dmcub.
--- .../gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c | 130 +++++++++--------- drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h | 20 +++ .../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 19 +-- 3 files changed, 94 insertions(+), 75 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c +68 -62
@@ -690,86 +690,98 @@ void hubp21_validate_dml_output(struct hubp *hubp, dml_dlg_attr->refcyc_per_meta_chunk_flip_l, dlg_attr.refcyc_per_meta_chunk_flip_l); } -static void program_surface_flip_and_addr (struct hubp *hubp,struct dmub_rb_cmd_flip *surface_flip) +static void program_surface_flip_and_addr (struct hubp *hubp,struct surface_flip_registers *flip_regs) {struct dcn21_hubp *hubp21 =TO_DCN21_HUBP (hubp);REG_UPDATE_3 (DCSURF_FLIP_CONTROL , -SURFACE_FLIP_TYPE , surface_flip->flip.flip_params.immediate, -SURFACE_FLIP_MODE_FOR_STEREOSYNC , surface_flip->flip.flip_params.grph_stereo, -SURFACE_FLIP_IN_STEREOSYNC , surface_flip->flip.flip_params.grph_stereo); +SURFACE_FLIP_TYPE , flip_regs->immediate, +SURFACE_FLIP_MODE_FOR_STEREOSYNC , flip_regs->grph_stereo, +SURFACE_FLIP_IN_STEREOSYNC , flip_regs->grph_stereo);REG_UPDATE (VMID_SETTINGS_0 , -VMID , surface_flip->flip.flip_params.vmid); +VMID , flip_regs->vmid);REG_UPDATE_8 (DCSURF_SURFACE_CONTROL , -PRIMARY_SURFACE_TMZ , surface_flip->flip.flip_params.tmz_surface, -PRIMARY_SURFACE_TMZ_C , surface_flip->flip.flip_params.tmz_surface, -PRIMARY_META_SURFACE_TMZ , surface_flip->flip.flip_params.tmz_surface, -PRIMARY_META_SURFACE_TMZ_C , surface_flip->flip.flip_params.tmz_surface, -SECONDARY_SURFACE_TMZ , surface_flip->flip.flip_params.tmz_surface, -SECONDARY_SURFACE_TMZ_C , surface_flip->flip.flip_params.tmz_surface, -SECONDARY_META_SURFACE_TMZ , surface_flip->flip.flip_params.tmz_surface, -SECONDARY_META_SURFACE_TMZ_C , surface_flip->flip.flip_params.tmz_surface); +PRIMARY_SURFACE_TMZ , flip_regs->tmz_surface, +PRIMARY_SURFACE_TMZ_C , flip_regs->tmz_surface, +PRIMARY_META_SURFACE_TMZ , flip_regs->tmz_surface, +PRIMARY_META_SURFACE_TMZ_C , flip_regs->tmz_surface, +SECONDARY_SURFACE_TMZ , flip_regs->tmz_surface, +SECONDARY_SURFACE_TMZ_C , flip_regs->tmz_surface, +SECONDARY_META_SURFACE_TMZ , flip_regs->tmz_surface, +SECONDARY_META_SURFACE_TMZ_C , flip_regs->tmz_surface);REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C ,0 ,PRIMARY_META_SURFACE_ADDRESS_HIGH_C , - surface_flip->flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C); + flip_regs->DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C);REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_C ,0 ,PRIMARY_META_SURFACE_ADDRESS_C , - surface_flip->flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_C); + flip_regs->DCSURF_PRIMARY_META_SURFACE_ADDRESS_C);REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH ,0 ,PRIMARY_META_SURFACE_ADDRESS_HIGH , - surface_flip->flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH); + flip_regs->DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH);REG_SET (DCSURF_PRIMARY_META_SURFACE_ADDRESS ,0 ,PRIMARY_META_SURFACE_ADDRESS , - surface_flip->flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS); + flip_regs->DCSURF_PRIMARY_META_SURFACE_ADDRESS);REG_SET (DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH ,0 ,SECONDARY_META_SURFACE_ADDRESS_HIGH , - surface_flip->flip.DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH); + flip_regs->DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH);REG_SET (DCSURF_SECONDARY_META_SURFACE_ADDRESS ,0 ,SECONDARY_META_SURFACE_ADDRESS , - surface_flip->flip.DCSURF_SECONDARY_META_SURFACE_ADDRESS); + flip_regs->DCSURF_SECONDARY_META_SURFACE_ADDRESS);REG_SET (DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH ,0 ,SECONDARY_SURFACE_ADDRESS_HIGH , - surface_flip->flip.DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH); + flip_regs->DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH);REG_SET (DCSURF_SECONDARY_SURFACE_ADDRESS ,0 ,SECONDARY_SURFACE_ADDRESS , - surface_flip->flip.DCSURF_SECONDARY_SURFACE_ADDRESS); + flip_regs->DCSURF_SECONDARY_SURFACE_ADDRESS);REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C ,0 ,PRIMARY_SURFACE_ADDRESS_HIGH_C , - surface_flip->flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C); + flip_regs->DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C);REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_C ,0 ,PRIMARY_SURFACE_ADDRESS_C , - surface_flip->flip.DCSURF_PRIMARY_SURFACE_ADDRESS_C); + flip_regs->DCSURF_PRIMARY_SURFACE_ADDRESS_C);REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH ,0 ,PRIMARY_SURFACE_ADDRESS_HIGH , - surface_flip->flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH); + flip_regs->DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH);REG_SET (DCSURF_PRIMARY_SURFACE_ADDRESS ,0 ,PRIMARY_SURFACE_ADDRESS , - surface_flip->flip.DCSURF_PRIMARY_SURFACE_ADDRESS); + flip_regs->DCSURF_PRIMARY_SURFACE_ADDRESS); } -void program_surface_flip_and_addr_dmcub (struct hubp *hubp,struct dmub_rb_cmd_flip *surface_flip) +void dmcub_PLAT_54186_wa (struct hubp *hubp,struct surface_flip_registers *flip_regs) {struct dc_dmub_srv *dmcub = hubp->ctx->dmub_srv;struct dcn21_hubp *hubp21 =TO_DCN21_HUBP (hubp); +struct dmub_rb_cmd_PLAT_54186_wa PLAT_54186_wa = {0 }; + + PLAT_54186_wa.header.type =DMUB_CMD__PLAT_54186_WA ; + PLAT_54186_wa.flip.DCSURF_PRIMARY_SURFACE_ADDRESS = flip_regs->DCSURF_PRIMARY_SURFACE_ADDRESS; + PLAT_54186_wa.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_C = flip_regs->DCSURF_PRIMARY_SURFACE_ADDRESS_C; + PLAT_54186_wa.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH = flip_regs->DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH; + PLAT_54186_wa.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C = flip_regs->DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C; + PLAT_54186_wa.flip.flip_params.grph_stereo = flip_regs->grph_stereo; + PLAT_54186_wa.flip.flip_params.hubp_inst = hubp->inst; + PLAT_54186_wa.flip.flip_params.immediate = flip_regs->immediate; + PLAT_54186_wa.flip.flip_params.tmz_surface = flip_regs->tmz_surface; + PLAT_54186_wa.flip.flip_params.vmid = flip_regs->vmid;PERF_TRACE ();// TODO: remove after performance is stable. -dc_dmub_srv_cmd_queue (dmcub, &surface_flip->header); +dc_dmub_srv_cmd_queue (dmcub, &PLAT_54186_wa.header);PERF_TRACE ();// TODO: remove after performance is stable. dc_dmub_srv_cmd_execute (dmcub);PERF_TRACE ();// TODO: remove after performance is stable.
@@ -782,15 +794,11 @@ bool hubp21_program_surface_flip_and_addr( const struct dc_plane_address *address, bool flip_immediate) { -struct dmub_rb_cmd_flip surface_flip = {0 }; -bool grph_stereo = false;struct dc_debug_options *debug = &hubp->ctx->dc->debug;struct dcn21_hubp *hubp21 =TO_DCN21_HUBP (hubp); +struct surface_flip_registers flip_regs = {0 }; - surface_flip.header.type =DMUB_CMD__SURFACE_FLIP ; - - surface_flip.flip.flip_params.vmid = address->vmid; - surface_flip.flip.flip_params.hubp_inst = hubp->inst; + flip_regs.vmid = address->vmid;switch (address->type) {case PLN_ADDR_TYPE_GRAPHICS :
@@ -800,15 +808,15 @@ bool hubp21_program_surface_flip_and_addr( }if (address->grph.meta_addr.quad_part !=0 ) { - surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS = + flip_regs.DCSURF_PRIMARY_META_SURFACE_ADDRESS = address->grph.meta_addr.low_part; - surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH = + flip_regs.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH = address->grph.meta_addr.high_part; } - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS = + flip_regs.DCSURF_PRIMARY_SURFACE_ADDRESS = address->grph.addr.low_part; - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH = + flip_regs.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH = address->grph.addr.high_part;break ;case PLN_ADDR_TYPE_VIDEO_PROGRESSIVE :
@@ -817,30 +825,30 @@ bool hubp21_program_surface_flip_and_addr(break ;if (address->video_progressive.luma_meta_addr.quad_part !=0 ) { - surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS = + flip_regs.DCSURF_PRIMARY_META_SURFACE_ADDRESS = address->video_progressive.luma_meta_addr.low_part; - surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH = + flip_regs.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH = address->video_progressive.luma_meta_addr.high_part; - surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_C = + flip_regs.DCSURF_PRIMARY_META_SURFACE_ADDRESS_C = address->video_progressive.chroma_meta_addr.low_part; - surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C = + flip_regs.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C = address->video_progressive.chroma_meta_addr.high_part; } - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS = + flip_regs.DCSURF_PRIMARY_SURFACE_ADDRESS = address->video_progressive.luma_addr.low_part; - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH = + flip_regs.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH = address->video_progressive.luma_addr.high_part;if (debug->nv12_iflip_vm_wa) { - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_C = + flip_regs.DCSURF_PRIMARY_SURFACE_ADDRESS_C = address->video_progressive.chroma_addr.low_part + hubp21->PLAT_54186_wa_chroma_addr_offset; }else - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_C = + flip_regs.DCSURF_PRIMARY_SURFACE_ADDRESS_C = address->video_progressive.chroma_addr.low_part; - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C = + flip_regs.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C = address->video_progressive.chroma_addr.high_part;break ;
@@ -850,30 +858,30 @@ bool hubp21_program_surface_flip_and_addr(if (address->grph_stereo.right_addr.quad_part ==0 )break ; - grph_stereo = true; + flip_regs.grph_stereo = true;if (address->grph_stereo.right_meta_addr.quad_part !=0 ) { - surface_flip.flip.DCSURF_SECONDARY_META_SURFACE_ADDRESS = + flip_regs.DCSURF_SECONDARY_META_SURFACE_ADDRESS = address->grph_stereo.right_meta_addr.low_part; - surface_flip.flip.DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH = + flip_regs.DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH = address->grph_stereo.right_meta_addr.high_part; }if (address->grph_stereo.left_meta_addr.quad_part !=0 ) { - surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS = + flip_regs.DCSURF_PRIMARY_META_SURFACE_ADDRESS = address->grph_stereo.left_meta_addr.low_part; - surface_flip.flip.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH = + flip_regs.DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH = address->grph_stereo.left_meta_addr.high_part; } - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS = + flip_regs.DCSURF_PRIMARY_SURFACE_ADDRESS = address->grph_stereo.left_addr.low_part; - surface_flip.flip.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH = + flip_regs.DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH = address->grph_stereo.left_addr.high_part; - surface_flip.flip.DCSURF_SECONDARY_SURFACE_ADDRESS = + flip_regs.DCSURF_SECONDARY_SURFACE_ADDRESS = address->grph_stereo.right_addr.low_part; - surface_flip.flip.DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH = + flip_regs.DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH = address->grph_stereo.right_addr.high_part;break ;
@@ -882,15 +890,13 @@ bool hubp21_program_surface_flip_and_addr(break ; } - surface_flip.flip.flip_params.vmid = address->vmid; - surface_flip.flip.flip_params.grph_stereo = grph_stereo; - surface_flip.flip.flip_params.tmz_surface = address->tmz_surface; - surface_flip.flip.flip_params.immediate = flip_immediate; + flip_regs.tmz_surface = address->tmz_surface; + flip_regs.immediate = flip_immediate; -if (hubp->ctx->dc->debug.enable_dmcub_surface_flip) -program_surface_flip_and_addr_dmcub (hubp, &surface_flip); +if (hubp->ctx->dc->debug.enable_dmcub_surface_flip && address->type ==PLN_ADDR_TYPE_VIDEO_PROGRESSIVE ) +dmcub_PLAT_54186_wa (hubp, &flip_regs);else -program_surface_flip_and_addr (hubp, &surface_flip); +program_surface_flip_and_addr (hubp, &flip_regs); hubp->request_address = *address;
drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h +20 -0
@@ -63,6 +63,26 @@ struct hubp {bool power_gated; }; +struct surface_flip_registers { +uint32_t DCSURF_SURFACE_CONTROL; +uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH; +uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS; +uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH; +uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS; +uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C; +uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS_C; +uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C; +uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS_C; +uint32_t DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH; +uint32_t DCSURF_SECONDARY_META_SURFACE_ADDRESS; +uint32_t DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH; +uint32_t DCSURF_SECONDARY_SURFACE_ADDRESS; +bool tmz_surface; +bool immediate; +uint8_t vmid; +bool grph_stereo; +}; +struct hubp_funcs { void (*hubp_setup)( struct hubp *hubp,
drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +6 -13
@@ -48,7 +48,7 @@ enum dmub_cmd_type {DMUB_CMD__REG_SEQ_FIELD_UPDATE_SEQ =2 ,DMUB_CMD__REG_SEQ_BURST_WRITE =3 ,DMUB_CMD__REG_REG_WAIT =4 , -DMUB_CMD__SURFACE_FLIP =5 , +DMUB_CMD__PLAT_54186_WA =5 ,DMUB_CMD__PSR =64 , DMUB_CMD__VBIOS = 128, };
@@ -151,20 +151,12 @@ struct dmub_rb_cmd_reg_wait {#define PHYSICAL_ADDRESS_LOC union large_integer#endif -struct dmub_cmd_surface_flip { +struct dmub_cmd_PLAT_54186_wa {uint32_t DCSURF_SURFACE_CONTROL; -uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH; -uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS;uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH;uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS; -uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C; -uint32_t DCSURF_PRIMARY_META_SURFACE_ADDRESS_C;uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C;uint32_t DCSURF_PRIMARY_SURFACE_ADDRESS_C; -uint32_t DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH; -uint32_t DCSURF_SECONDARY_META_SURFACE_ADDRESS; -uint32_t DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH; -uint32_t DCSURF_SECONDARY_SURFACE_ADDRESS;struct {uint8_t hubp_inst :4 ;uint8_t tmz_surface :1 ;
@@ -173,11 +165,12 @@ struct dmub_cmd_surface_flip { uint8_t grph_stereo : 1; uint32_t reserved : 21; } flip_params; +uint32_t reserved[9 ]; }; -struct dmub_rb_cmd_flip { +struct dmub_rb_cmd_PLAT_54186_wa {struct dmub_cmd_header header; -struct dmub_cmd_surface_flip flip; +struct dmub_cmd_PLAT_54186_wa flip; };struct dmub_cmd_digx_encoder_control_data {
@@ -287,7 +280,7 @@ union dmub_rb_cmd {struct dmub_rb_cmd_psr_enable psr_enable;struct dmub_rb_cmd_psr_copy_settings psr_copy_settings;struct dmub_rb_cmd_psr_set_level psr_set_level; -struct dmub_rb_cmd_flip surface_flip; +struct dmub_rb_cmd_PLAT_54186_wa PLAT_54186_wa;struct dmub_rb_cmd_psr_setup psr_setup; };
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 39/43 drm/amd/display: Use udelay to avoid context switch
In reply to Rodrigo SiqueiraFrom: Martin Tsai <[email protected]>
[why]
The rapid msleep operation causes the white line garbage when DAL check flip pending status in SetVidPnSourceVisibility. To execute this msleep will induce context switch, and longer delay could cause worse garbage situation.
[how]
To replace msleep with udelay.
--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +2 -2
@@ -1642,9 +1642,9 @@ void dcn20_program_front_end_for_ctx(struct hubp *hubp = pipe->plane_res.hubp;int j =0 ; -for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS +for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS *1000 && hubp->funcs->hubp_is_flip_pending (hubp); j++) -msleep (1 ); +udelay (1 ); } }
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switch
Am 10.01.20 um 15:46 schrieb Rodrigo Siqueira:
Quoted 24 lines
From: Martin Tsai <[email protected]>
[why]
The rapid msleep operation causes the white line garbage when DAL check flip pending status in SetVidPnSourceVisibility. To execute this msleep will induce context switch, and longer delay could cause worse garbage situation.[how]
To replace msleep with udelay.Signed-off-by Martin Tsai <[email protected]>Reviewed-by Tony Cheng <[email protected]>Acked-by Harry Wentland <[email protected]>Acked-by Rodrigo Siqueira <[email protected]>--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1647
Quoted 8 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +2
-2@@ -1642,9 +1642,9 @@ void dcn20_program_front_end_for_ctx(struct hubp *hubp = pipe->plane_res.hubp;int j =0 ; -for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS +for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS *1000 && hubp->funcs->hubp_is_flip_pending (hubp); j++) -msleep (1 ); +udelay (1 );
Why not using mdelay() here?
Christian.
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1650
Quoted 3 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0} }
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
RE: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switch
Quoted 44 lines
-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Christian König
Sent: 2020/January/10, Friday 10:02 AM
To: Siqueira, Rodrigo <[email protected]>; amd- [email protected]
Cc: Li, Sun peng (Leo) <[email protected]>; Cheng, Tony <[email protected]>; Tsai, Martin <[email protected]>; Lakha, Bhawanpreet <[email protected]>; Wentland, Harry <[email protected]>
Subject: Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switchAm 10.01.20 um 15:46 schrieb Rodrigo Siqueira:
Quoted 28 lines
From: Martin Tsai <[email protected]>
[why]
The rapid msleep operation causes the white line garbage when DAL check flip pending status in SetVidPnSourceVisibility. To execute this msleep will induce context switch, and longer delay could cause worse garbage situation.[how]
To replace msleep with udelay.Signed-off-by Martin Tsai <[email protected]>Reviewed-by Tony Cheng <[email protected]>Acked-by Harry Wentland <[email protected]>Acked-by Rodrigo Siqueira <[email protected]>--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0@@ -1642,9 +1642,9 @@ void dcn20_program_front_end_for_ctx(struct hubp *hubp = pipe->plane_res.hubp;int j =0 ;
Quoted 8 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1647
Quoted 6 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +2
-2-for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS +for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS *1000 && hubp->funcs-hubp_is_flip_pending (hubp); j++) -msleep (1 ); +udelay (1 );Why not using
mdelay()here?
As far as I know, mdelay() is only defined on Linux side.
This piece of code is shared by both Linux and Windows, so we have to use a function that's available on both platforms.
Zhan
Quoted 10 lines
Christian.
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1650
Quoted 3 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0} }_______________________________________________
amd-gfx mailing list
[email protected]
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switch
Quoted 53 lines · Zhan
Quoted 44 lines
-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Christian König
Sent: 2020/January/10, Friday 10:02 AM
To: Siqueira, Rodrigo <[email protected]>; amd- [email protected]
Cc: Li, Sun peng (Leo) <[email protected]>; Cheng, Tony <[email protected]>; Tsai, Martin <[email protected]>; Lakha, Bhawanpreet <[email protected]>; Wentland, Harry <[email protected]>
Subject: Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switchAm 10.01.20 um 15:46 schrieb Rodrigo Siqueira:
Quoted 28 lines
From: Martin Tsai <[email protected]>
[why]
The rapid msleep operation causes the white line garbage when DAL check flip pending status in SetVidPnSourceVisibility. To execute this msleep will induce context switch, and longer delay could cause worse garbage situation.[how]
To replace msleep with udelay.Signed-off-by Martin Tsai <[email protected]>Reviewed-by Tony Cheng <[email protected]>Acked-by Harry Wentland <[email protected]>Acked-by Rodrigo Siqueira <[email protected]>--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0@@ -1642,9 +1642,9 @@ void dcn20_program_front_end_for_ctx(struct hubp *hubp = pipe->plane_res.hubp;int j =0 ;Quoted 8 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1647
Quoted 6 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +2
-2-for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS +for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS *1000 && hubp->funcs-hubp_is_flip_pending (hubp); j++) -msleep (1 ); +udelay (1 );Why not using
mdelay()here?
Quoted 4 lines
As far as I know,
mdelay()is only defined on Linux side.This piece of code is shared by both Linux and Windows, so we have to use a function that's available on both platforms.
It was used here before so we definitely have it defined for Windows as well.
From the commit description it sounds like msleep wasn't tight enough and longer delays lead to issues here, at least on Windows.
Martin, Tony, do you have more details about this?
Thanks,
Harry
Quoted 12 lines
Zhan
Quoted 10 lines
Christian.
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1650
Quoted 3 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0} }_______________________________________________
amd-gfx mailing list
[email protected]
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switch
Quoted 54 lines · Harry Wentland
Quoted 53 lines · Zhan
Quoted 52 lines
-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Christian König
Sent: 2020/January/10, Friday 10:02 AM
To: Siqueira, Rodrigo <[email protected]>; amd- [email protected]
Cc: Li, Sun peng (Leo) <[email protected]>; Cheng, Tony <[email protected]>; Tsai, Martin <[email protected]>; Lakha, Bhawanpreet <[email protected]>; Wentland, Harry <[email protected]>
Subject: Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switchAm 10.01.20 um 15:46 schrieb Rodrigo Siqueira:
Quoted 25 lines
From: Martin Tsai <[email protected]>
[why]
The rapid msleep operation causes the white line garbage when DAL check flip pending status in SetVidPnSourceVisibility. To execute this msleep will induce context switch, and longer delay could cause worse garbage situation.[how]
To replace msleep with udelay.Signed-off-by Martin Tsai <[email protected]>Reviewed-by Tony Cheng <[email protected]>Acked-by Harry Wentland <[email protected]>Acked-by Rodrigo Siqueira <[email protected]>--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1648
Quoted 9 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +2
-2@@ -1642,9 +1642,9 @@ void dcn20_program_front_end_for_ctx(struct hubp *hubp = pipe->plane_res.hubp;int j =0 ; -for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS +for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS *1000 && hubp->funcs-hubp_is_flip_pending (hubp); j++) -msleep (1 ); +udelay (1 );Why not using
mdelay()here?
Quoted 8 lines
Quoted 4 lines
As far as I know,
mdelay()is only defined on Linux side.This piece of code is shared by both Linux and Windows, so we have to use a function that's available on both platforms.
It was used here before so we definitely have it defined for Windows as well.
Whoops, I misread that.
mdelay is indeed not defined on our other platforms but we could go ahead and define it if needed.
Harry
Quoted 20 lines
From the commit description it sounds like msleep wasn't tight enough and longer delays lead to issues here, at least on Windows.
Martin, Tony, do you have more details about this?
Thanks,
HarryQuoted 12 lines
Zhan
Quoted 10 lines
Christian.
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1650
Quoted 3 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0} }_______________________________________________
amd-gfx mailing list
[email protected]
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switch
Am 10.01.20 um 22:30 schrieb Harry Wentland:
Quoted 56 lines · Harry Wentland
Quoted 55 lines · Zhan
Quoted 54 lines
Quoted 43 lines
-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Christian König
Sent: 2020/January/10, Friday 10:02 AM
To: Siqueira, Rodrigo <[email protected]>; amd- [email protected]
Cc: Li, Sun peng (Leo) <[email protected]>; Cheng, Tony <[email protected]>; Tsai, Martin <[email protected]>; Lakha, Bhawanpreet <[email protected]>; Wentland, Harry <[email protected]>
Subject: Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switchAm 10.01.20 um 15:46 schrieb Rodrigo Siqueira:
Quoted 28 lines
From: Martin Tsai <[email protected]>
[why]
The rapid msleep operation causes the white line garbage when DAL check flip pending status in SetVidPnSourceVisibility. To execute this msleep will induce context switch, and longer delay could cause worse garbage situation.[how]
To replace msleep with udelay.Signed-off-by Martin Tsai <[email protected]>Reviewed-by Tony Cheng <[email protected]>Acked-by Harry Wentland <[email protected]>Acked-by Rodrigo Siqueira <[email protected]>--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0@@ -1642,9 +1642,9 @@ void dcn20_program_front_end_for_ctx(struct hubp *hubp = pipe->plane_res.hubp;int j =0 ;Quoted 7 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1647
Quoted 6 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +2
-2-for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS +for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS *1000 && hubp->funcs-hubp_is_flip_pending (hubp); j++) -msleep (1 ); +udelay (1 );Why not using
mdelay()here?As far as I know,
mdelay()is only defined on Linux side.This piece of code is shared by both Linux and Windows, so we have to use a function that's available on both platforms.
Quoted 7 lines
Quoted 3 lines
It was used here before so we definitely have it defined for Windows as well.
Whoops, I misread that.
mdelay is indeed not defined on our other platforms but we could go ahead and define it if needed.
Yes, indeed. And to repeat other platforms are irrelevant for up streaming.
When changing the code in a certain way for Linux makes sense then we have do to this and need to adjust other platforms accordingly or maintain the delta somehow.
Christian.
Quoted 18 lines
Harry
Quoted 15 lines
From the commit description it sounds like msleep wasn't tight enough
and longer delays lead to issues here, at least on Windows.Martin, Tony, do you have more details about this?
Thanks,
HarryQuoted 7 lines
Zhan
Quoted 5 lines
Christian.
Quoted 3 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0} }
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
_______________________________________________
amd-gfx mailing list
[email protected]_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switch
Quoted 62 lines · Christian König
Am 10.01.20 um 22:30 schrieb Harry Wentland:
Quoted 61 lines · Harry Wentland
Quoted 51 lines · Zhan
Quoted 50 lines
Quoted 50 lines
-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Christian König
Sent: 2020/January/10, Friday 10:02 AM
To: Siqueira, Rodrigo <[email protected]>; amd- [email protected]
Cc: Li, Sun peng (Leo) <[email protected]>; Cheng, Tony <[email protected]>; Tsai, Martin <[email protected]>; Lakha, Bhawanpreet <[email protected]>; Wentland, Harry <[email protected]>
Subject: Re: [PATCH 39/43] drm/amd/display: Use udelay to avoid context switchAm 10.01.20 um 15:46 schrieb Rodrigo Siqueira:
Quoted 25 lines
From: Martin Tsai <[email protected]>
[why]
The rapid msleep operation causes the white line garbage when DAL check flip pending status in SetVidPnSourceVisibility. To execute this msleep will induce context switch, and longer delay could cause worse garbage situation.[how]
To replace msleep with udelay.Signed-off-by Martin Tsai <[email protected]>Reviewed-by Tony Cheng <[email protected]>Acked-by Harry Wentland <[email protected]>Acked-by Rodrigo Siqueira <[email protected]>--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1648
Quoted 9 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +2
-2@@ -1642,9 +1642,9 @@ void dcn20_program_front_end_for_ctx(struct hubp *hubp = pipe->plane_res.hubp;int j =0 ; -for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS +for (j =0 ; j <TIMEOUT_FOR_PIPE_ENABLE_MS *1000 && hubp->funcs-hubp_is_flip_pending (hubp); j++) -msleep (1 ); +udelay (1 );Why not using
mdelay()here?Quoted 7 lines
Quoted 4 lines
As far as I know,
mdelay()is only defined on Linux side.This piece of code is shared by both Linux and Windows, so we have to use a function that's available on both platforms.
It was used here before so we definitely have it defined for Windows as well.
Whoops, I misread that.
Quoted 8 lines
Quoted 2 lines
mdelay is indeed not defined on our other platforms but we could go ahead and define it if needed.
Yes, indeed. And to repeat other platforms are irrelevant for up streaming.
When changing the code in a certain way for Linux makes sense then we have do to this and need to adjust other platforms accordingly or maintain the delta somehow.
Hi,
I already made a patch to define mdelay on other OS platforms. For this patch, I just replace udelay by mdelay.
Best Regards
Quoted 28 lines
Christian.
Quoted 25 lines
Harry
Quoted 18 lines
From the commit description it sounds like msleep wasn't tight enough
and longer delays lead to issues here, at least on Windows.Martin, Tony, do you have more details about this?
Thanks,
HarryQuoted 10 lines
Zhan
Quoted 8 lines
Christian.
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c void dcn20_program_front_end_for_ctx( · line 1650
Quoted 3 lines
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +0
-0} }_______________________________________________
amd-gfx mailing list
[email protected]_______________________________________________
amd-gfx mailing list
[email protected]
https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Famd-gfx&data=02%7C01%7CRodrigo.Siqueira%40amd.com%7Cc13de5453b1b45fe04a108d796b41f24%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637143572686205847&sdata=VgCSlghDomSX5T6v2JtujBb%2F0qwYUNvX0UCDUU0ehB0%3D&reserved=0
Signature · 3 lines
Rodrigo Siqueira Software Engineer, Advanced Micro Devices (AMD) https://siqueira.tech
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 40/43 drm/amd/display: fixup DML dependencies
In reply to Rodrigo SiqueiraFrom: Jun Lei <[email protected]>
[why]
Need to fix DML portability issues to enable SW unit testing around DML
[how]
Move calcs into dc include folder since multiple components reference it Remove relative paths to external dependencies
---
drivers/gpu/drm/amd/display/dc/dml/dml_common_defs.c | 2 +-
drivers/gpu/drm/amd/display/dc/dml/dml_inline_defs.h | 2 +-
drivers/gpu/drm/amd/display/dc/{calcs => inc}/dcn_calc_math.h | 0
3 files changed, 2 insertions(+), 2 deletions(-)
rename drivers/gpu/drm/amd/display/dc/{calcs => inc}/dcn_calc_math.h (100%)drivers/gpu/drm/amd/display/dc/dml/dml_common_defs.c +1 -1
@@ -24,7 +24,7 @@*/ #include "dml_common_defs.h" -#include "../calcs/dcn_calc_math.h" +#include "dcn_calc_math.h" #include "dml_inline_defs.h"
drivers/gpu/drm/amd/display/dc/dml/dml_inline_defs.h +1 -1
@@ -27,7 +27,7 @@#define __DML_INLINE_DEFS_H__#include "dml_common_defs.h" -#include "../calcs/dcn_calc_math.h" +#include "dcn_calc_math.h" #include "dml_logger.h" static inline double dml_min (double a,double b)
drivers/gpu/drm/amd/display/dc/inc/dcn_calc_math.h renamed from drivers/gpu/drm/amd/display/dc/calcs/dcn_calc_math.h +0 -0
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 41/43 drm/amd/display: reallocate MST payload when link loss
In reply to Rodrigo SiqueiraFrom: Paul Hsieh <[email protected]>
[Why]
Try to allocate MST payload but receive HPD short pulse with link loss casue driver allocate payload twice. It cause monitor can't light up successfully.
[How]
When driver receive HPD short pulse with link loss, we need to deallocate payload then allocate payload.
Then we will not allocate payload twice with same sink.
--- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 33 +++++++++++++++++++ .../gpu/drm/amd/display/dc/core/dc_link_dp.c | 7 ++-- drivers/gpu/drm/amd/display/dc/dc_link.h | 1 + 3 files changed, 36 insertions(+), 5 deletions(-)
drivers/gpu/drm/amd/display/dc/core/dc_link.c +33 -0
@@ -2873,6 +2873,39 @@ static enum dc_status deallocate_mst_payload(struct pipe_ctx *pipe_ctx)return DC_OK ; } + +enum dc_status dc_link_reallocate_mst_payload (struct dc_link *link) +{ +int i; +struct pipe_ctx *pipe_ctx; + +// Clear all of MST payload then reallocate +for (i =0 ; i <MAX_PIPES ; i++) { + pipe_ctx = &link->dc->current_state->res_ctx.pipe_ctx[i]; +if (pipe_ctx && pipe_ctx->stream && pipe_ctx->stream->link == link && + pipe_ctx->stream->dpms_off == false && + pipe_ctx->stream->signal ==SIGNAL_TYPE_DISPLAY_PORT_MST ) { +deallocate_mst_payload (pipe_ctx); + } + } + +for (i =0 ; i <MAX_PIPES ; i++) { + pipe_ctx = &link->dc->current_state->res_ctx.pipe_ctx[i]; +if (pipe_ctx && pipe_ctx->stream && pipe_ctx->stream->link == link && + pipe_ctx->stream->dpms_off == false && + pipe_ctx->stream->signal ==SIGNAL_TYPE_DISPLAY_PORT_MST ) { +/* enable/disable PHY will clear connection between BE and FE +* need to restore it. +*/ + link->link_enc->funcs->connect_dig_be_to_fe (link->link_enc, + pipe_ctx->stream_res.stream_enc->id, true); +dc_link_allocate_mst_payload (pipe_ctx); + } + } + +return DC_OK ; +} +#if defined(CONFIG_DRM_AMD_DC_HDCP )static void update_psp_stream_config (struct pipe_ctx *pipe_ctx,bool dpms_off) {
drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +2 -5
@@ -2902,11 +2902,8 @@ bool dc_link_handle_hpd_rx_irq(struct dc_link *link, union hpd_irq_data *out_hpd pipe_ctx, pipe_ctx->stream->signal); -if (pipe_ctx && pipe_ctx->stream && pipe_ctx->stream->link == link && - pipe_ctx->stream->dpms_off == false && - pipe_ctx->stream->signal ==SIGNAL_TYPE_DISPLAY_PORT_MST ) { -dc_link_allocate_mst_payload (pipe_ctx); - } +if (pipe_ctx->stream->signal ==SIGNAL_TYPE_DISPLAY_PORT_MST ) +dc_link_reallocate_mst_payload (link); status = false;if (out_link_loss)
drivers/gpu/drm/amd/display/dc/dc_link.h +1 -0
@@ -205,6 +205,7 @@ enum dc_detect_reason {bool dc_link_detect (struct dc_link *dc_link,enum dc_detect_reason reason);bool dc_link_get_hpd_state (struct dc_link *dc_link);enum dc_status dc_link_allocate_mst_payload (struct pipe_ctx *pipe_ctx); +enum dc_status dc_link_reallocate_mst_payload (struct dc_link *link);/* Notify DC about DP RX Interrupt (aka Short Pulse Interrupt). * Return:
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 42/43 drm/amd/display: 3.2.68
In reply to Rodrigo SiqueiraFrom: Aric Cyr <[email protected]>
--- drivers/gpu/drm/amd/display/dc/dc.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/dc.h +1 -1
@@ -39,7 +39,7 @@#include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#define DC_VER "3.2.67" +#define DC_VER "3.2.68"#define MAX_SURFACES 3#define MAX_PLANES 6
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 43/43 drm/amd/display: skip opp blank or unblank if test pattern enabled
In reply to Rodrigo SiqueiraFrom: Wenjing Liu <[email protected]>
[why]
DPG is used to generate both blank and test automation test pattern. When test automation is running the requested test pattern can be overriden by the blank or unblank call because it is using the same hardware.
[how]
When test pattern is requested, skip blank or unblank.
--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +3 -0
@@ -945,6 +945,9 @@ void dcn20_blank_pixel_data(int width = stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right;int height = stream->timing.v_addressable + stream->timing.v_border_bottom + stream->timing.v_border_top; +if (stream->link->test_pattern_enabled) +return ; +/* get opp dpg blank color */ color_space_to_black_color (dc, color_space, &black_color);
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx