DC Patches October 5, 2020
cover DC Patches October 5, 2020
This DC patchset brings improvements in multiple areas. In summary, we have:
* Fixes in Freesync, audio, ABM
* Improvements in i2c, p-state
* Added HDMI remote sink validation
------------------
Alvin Lee (3):
drm/amd/display: Don't allow pstate if no support in blank drm/amd/display: Program meta addresses correctly drm/amd/display: Only flush inst_fb if backdoor loading
Anthony Koo (1):
drm/amd/display: [FW Promotion] Release 0.0.36
Aric Cyr (4):
drm/amd/display: 3.2.105
drm/amd/display: Check for flip pending before locking pipes drm/amd/display: FreeSync not active near lower bound of non-LFC
monitor range
drm/amd/display: 3.2.106
Charlene Liu (1):
drm/amd/display: Add i2c speed arbitration for dc_i2c and hdcp_i2c
Chris Park (1):
drm/amd/display: Change to correct unit on audio rate
Dmytro Laktyushkin (1):
drm/amd/display: Fix OPTC_DATA_FORMAT programming
Fangzhi Zuo (1):
drm/amd/display: HDMI remote sink need mode validation for Linux
Felipe (1):
drm/amd/display: Fixed comments (uniform style)
Joshua Aberback (2):
drm/amd/display: Force enable pstate on driver unload drm/amd/display: Copy WM values from set A to other sets in hw_init
Rodrigo Siqueira (1):
drm/amd/display: Avoid set zero in the requested clk
Sung Lee (1):
drm/amd/display: Hardcode 45W Bounding Box for DCN2.1 Diags
Yongqiang Sun (1):
drm/amd/display: Block ABM in case of eDP ODM
.../dc/clk_mgr/dce112/dce112_clk_mgr.c | 3 +- drivers/gpu/drm/amd/display/dc/core/dc.c | 18 +- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 2 +- drivers/gpu/drm/amd/display/dc/dc.h | 6 +- drivers/gpu/drm/amd/display/dc/dc_stream.h | 1 + .../gpu/drm/amd/display/dc/dce/dce_i2c_hw.c | 9 +- .../amd/display/dc/dce100/dce100_resource.c | 1 + .../amd/display/dc/dce110/dce110_resource.c | 3 +- .../amd/display/dc/dce112/dce112_resource.c | 1 + .../amd/display/dc/dce120/dce120_resource.c | 1 + .../drm/amd/display/dc/dce80/dce80_resource.c | 3 + .../amd/display/dc/dcn10/dcn10_hw_sequencer.c | 164 +++++++++--------- .../amd/display/dc/dcn10/dcn10_hw_sequencer.h | 2 +- .../gpu/drm/amd/display/dc/dcn10/dcn10_optc.c | 11 ++ .../drm/amd/display/dc/dcn10/dcn10_resource.c | 1 + .../drm/amd/display/dc/dcn20/dcn20_hwseq.c | 2 + .../gpu/drm/amd/display/dc/dcn20/dcn20_optc.c | 8 - .../drm/amd/display/dc/dcn20/dcn20_resource.c | 6 +- .../drm/amd/display/dc/dcn21/dcn21_hwseq.c | 15 ++ .../drm/amd/display/dc/dcn21/dcn21_hwseq.h | 2 + .../gpu/drm/amd/display/dc/dcn21/dcn21_init.c | 1 + .../drm/amd/display/dc/dcn21/dcn21_resource.c | 120 ++++--------- .../drm/amd/display/dc/dcn30/dcn30_hubbub.c | 54 ++++++ .../drm/amd/display/dc/dcn30/dcn30_hubbub.h | 5 + .../drm/amd/display/dc/dcn30/dcn30_hwseq.c | 16 ++ .../drm/amd/display/dc/dcn30/dcn30_hwseq.h | 2 + .../gpu/drm/amd/display/dc/dcn30/dcn30_init.c | 1 + .../gpu/drm/amd/display/dc/dcn30/dcn30_optc.c | 8 - .../drm/amd/display/dc/dcn30/dcn30_resource.c | 1 + .../dc/dml/dcn30/display_mode_vba_30.c | 2 +- .../gpu/drm/amd/display/dc/inc/hw/dchubbub.h | 6 + .../gpu/drm/amd/display/dc/inc/hw_sequencer.h | 7 +- .../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 4 +- .../gpu/drm/amd/display/dmub/src/dmub_srv.c | 17 +- .../amd/display/modules/freesync/freesync.c | 55 +++--- 35 files changed, 323 insertions(+), 235 deletions(-)
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 1/18 drm/amd/display: 3.2.105
From: 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
@@ -42,7 +42,7 @@#include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#define DC_VER "3.2.104" +#define DC_VER "3.2.105"#define MAX_SURFACES 3#define MAX_PLANES 6
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 2/18 drm/amd/display: Don't allow pstate if no support in blank
In reply to Eryk BrolFrom: Alvin Lee <[email protected]>
[Why]
We will hang if we report switch in VACTIVE but not in VBLANK and DPG_EN = 1
[How]
Block switch in ACTIVE if not supported in BLANK
--- drivers/gpu/drm/amd/display/dc/dml/dcn30/display_mode_vba_30.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/dml/dcn30/display_mode_vba_30.c +1 -1
@@ -5558,7 +5558,7 @@ static void CalculateWatermarksAndDRAMSpeedChangeSupport( } } -if (mode_lib->vba.MinActiveDRAMClockChangeMargin >0 ) { +if (mode_lib->vba.MinActiveDRAMClockChangeMargin >0 && PrefetchMode ==0 ) { *DRAMClockChangeSupport = dm_dram_clock_change_vactive; }else if (((mode_lib->vba.SynchronizedVBlank == true || mode_lib->vba.TotalNumberOfActiveOTG ==1 || SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank >0 ) && PrefetchMode ==0 )) { *DRAMClockChangeSupport = dm_dram_clock_change_vblank;
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 3/18 drm/amd/display: Check for flip pending before locking pipes
In reply to Eryk BrolFrom: Aric Cyr <[email protected]>
[Why]
When running games or benchmarking with v-sync disabled, disabling a plane (which is v-sync) can cause underflow. This is caused by flips pending before pipe locking being applied after locks are released and pipes could have been re-arranged or disconnected. This could potentially apply a flip on incorrect pipe. Also, previous logic of always locking pipes was unnecessary.
[How]
Only lock the pipes when there is a pipe being disabled to increase efficiency. Before the pipes are locked, check that any pending flips are cleared to ensure the flips are applied to the correct pipe.
--- drivers/gpu/drm/amd/display/dc/core/dc.c | 11 +- .../amd/display/dc/dcn10/dcn10_hw_sequencer.c | 156 ++++++++++-------- .../amd/display/dc/dcn10/dcn10_hw_sequencer.h | 2 +- .../gpu/drm/amd/display/dc/inc/hw_sequencer.h | 2 +- 4 files changed, 88 insertions(+), 83 deletions(-)
drivers/gpu/drm/amd/display/dc/core/dc.c +2 -9
@@ -2383,7 +2383,6 @@ static void commit_planes_for_stream(struct dc *dc,enum surface_update_type update_type, struct dc_state *context) { - bool mpcc_disconnected = false; int i, j;struct pipe_ctx *top_pipe_to_program =NULL ;
@@ -2414,14 +2413,8 @@ static void commit_planes_for_stream(struct dc *dc,context_clock_trace (dc, context); } -if (update_type !=UPDATE_TYPE_FAST && dc->hwss.interdependent_update_lock && - dc->hwss.disconnect_pipes && dc->hwss.wait_for_pending_cleared){ - dc->hwss.interdependent_update_lock (dc, context, true); - mpcc_disconnected = dc->hwss.disconnect_pipes (dc, context); - dc->hwss.interdependent_update_lock (dc, context, false); -if (mpcc_disconnected) - dc->hwss.wait_for_pending_cleared (dc, context); - } +if (update_type !=UPDATE_TYPE_FAST && dc->hwss.interdependent_update_lock && dc->hwss.wait_for_pending_cleared) + dc->hwss.disconnect_pipes (dc, context);for (j =0 ; j < dc->res_pool->pipe_count; j++) {struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j];
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +84 -72
204 lines not on this page · raw message
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h +1 -1
@@ -194,7 +194,7 @@ void dcn10_get_surface_visual_confirm_color(void dcn10_get_hdr_visual_confirm_color (struct pipe_ctx *pipe_ctx,struct tg_color *color); -bool dcn10_disconnect_pipes ( +void dcn10_disconnect_pipes (struct dc *dc,struct dc_state *context);
drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h +1 -1
@@ -67,7 +67,7 @@ struct hw_sequencer_funcs { int num_planes, struct dc_state *context);void (*program_front_end_for_ctx)(struct dc *dc,struct dc_state *context); -bool (*disconnect_pipes)(struct dc *dc, +void (*disconnect_pipes)(struct dc *dc,struct dc_state *context);void (*wait_for_pending_cleared)(struct dc *dc,struct dc_state *context);
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 4/18 drm/amd/display: Change to correct unit on audio rate
In reply to Eryk BrolFrom: Chris Park <[email protected]>
[Why]
Formula uses kHz in their formula while our driver operates with Hz.
[How]
Divide audio rate by 1000 on the initial variable that is entered into formula.
--- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +2 -3
@@ -150,7 +150,6 @@ struct _vcs_dpi_ip_params_st dcn2_0_ip = { .dispclk_delay_subtotal = 87, // .dcfclk_cstate_latency = 10, // SRExitTime .max_inter_dcn_tile_repeaters = 8, - .xfc_supported = true, .xfc_fill_bw_overhead_percent = 10.0, .xfc_fill_constant_bytes = 0,
@@ -2203,9 +2202,9 @@ int dcn20_populate_dml_pipes_from_context(/* todo: default max for now, until there is logic reflecting this in dc*/ pipes[pipe_cnt].dout.output_bpc =12 ;#if defined(CONFIG_DRM_AMD_DC_DCN3_0 ) -/*fill up the audio sample rate*/ +/*fill up the audio sample rate (unit in kHz)*/ get_audio_check (&res_ctx->pipe_ctx[i].stream->audio_info, &aud_check); - pipes[pipe_cnt].dout.max_audio_sample_rate = aud_check.max_audiosample_rate; + pipes[pipe_cnt].dout.max_audio_sample_rate = aud_check.max_audiosample_rate /1000 ;#endif /* * For graphic plane, cursor number is 1, nv12 is 0
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 5/18 drm/amd/display: Fixed comments (uniform style)
In reply to Eryk BrolFrom: Felipe <[email protected]>
[WHY]
This change was implemented because the comment style was not uniform across the file. In some lines comments were initiated with // and in others they were in between /* ... */.
Additionally, the style for multi-line comments was also not uniform and some comment lines were missing the space between the opening /* and the first word of the comment.
[HOW]
All comments are now in between /*.../*, multi line comments also use /*...*/ and for every comment there is now a space between the opening /* and the first word of the comment.
--- .../amd/display/modules/freesync/freesync.c | 45 ++++++++++--------- 1 file changed, 23 insertions(+), 22 deletions(-)
drivers/gpu/drm/amd/display/modules/freesync/freesync.c +23 -22
@@ -41,9 +41,9 @@#define BTR_MAX_MARGIN 2500/* Threshold to change BTR multiplier (to avoid frequent changes) */ #define BTR_DRIFT_MARGIN 2000 -/*Threshold to exit fixed refresh rate*/ +/* Threshold to exit fixed refresh rate */ #define FIXED_REFRESH_EXIT_MARGIN_IN_HZ 4 -/* Number of consecutive frames to check before entering/exiting fixed refresh*/ +/* Number of consecutive frames to check before entering/exiting fixed refresh */ #define FIXED_REFRESH_ENTER_FRAME_COUNT 5#define FIXED_REFRESH_EXIT_FRAME_COUNT 5
@@ -85,7 +85,7 @@ void mod_freesync_destroy(struct mod_freesync *mod_freesync)kfree (core_freesync); } -#if 0 /* unused currently */ +#if 0 /* Unused currently */ static unsigned int calc_refresh_in_uhz_from_duration (unsigned int duration_in_ns) {
@@ -184,7 +184,7 @@ static void update_v_total_for_static_ramp(bool ramp_direction_is_up = (current_duration_in_us > target_duration_in_us) ? true : false; -/* Calc ratio between new and current frame duration with 3 digit */ +/* Calculate ratio between new and current frame duration with 3 digit */ unsigned int frame_duration_ratio = div64_u64(1000000, (1000 + div64_u64(((unsigned long long)( STATIC_SCREEN_RAMP_DELTA_REFRESH_RATE_PER_FRAME) *
@@ -204,10 +204,10 @@ static void update_v_total_for_static_ramp(/* Going to a higher refresh rate (lower frame duration) */ if (ramp_direction_is_up) { -/* reduce frame duration */ +/* Reduce frame duration */ current_duration_in_us -= ramp_rate_interpolated; -/* adjust for frame duration below min */ +/* Adjust for frame duration below min */ if (current_duration_in_us <= target_duration_in_us) { in_out_vrr->fixed.ramping_active = false; in_out_vrr->fixed.ramping_done = true;
@@ -217,10 +217,10 @@ static void update_v_total_for_static_ramp( }/* Going to a lower refresh rate (larger frame duration) */ }else { -/* increase frame duration */ +/* Increase frame duration */ current_duration_in_us += ramp_rate_interpolated; -/* adjust for frame duration above max */ +/* Adjust for frame duration above max */ if (current_duration_in_us >= target_duration_in_us) { in_out_vrr->fixed.ramping_active = false; in_out_vrr->fixed.ramping_done = true;
@@ -289,7 +289,7 @@ static void apply_below_the_range(struct core_freesync *core_freesync, }else {/* Calculate number of midPoint frames that could fit within -* the render time interval- take ceil of this value +* the render time interval - take ceil of this value */ mid_point_frames_ceil = (last_render_time_in_us + in_out_vrr->btr.mid_point_in_us - 1) /
@@ -306,7 +306,7 @@ static void apply_below_the_range(struct core_freesync *core_freesync, }/* Calculate number of midPoint frames that could fit within -* the render time interval- take floor of this value +* the render time interval - take floor of this value */ mid_point_frames_floor = last_render_time_in_us / in_out_vrr->btr.mid_point_in_us;
@@ -559,7 +559,7 @@ static void build_vrr_infopacket_data_v1(const struct mod_vrr_params *vrr,*/ infopacket->sb[8 ] = (unsigned char )((vrr->max_refresh_in_uhz +500000 ) /1000000 ); -//FreeSync HDR +/* FreeSync HDR */ infopacket->sb[9 ] =0 ; infopacket->sb[10 ] =0 ; }
@@ -897,15 +897,15 @@ void mod_freesync_build_vrr_params(struct mod_freesync *mod_freesync, min_refresh_in_uhz = in_config->min_refresh_in_uhz; max_refresh_in_uhz = in_config->max_refresh_in_uhz; -// Full range may be larger than current video timing, so cap at nominal +/* Full range may be larger than current video timing, so cap at nominal */ if (max_refresh_in_uhz > nominal_field_rate_in_uhz) max_refresh_in_uhz = nominal_field_rate_in_uhz; -// Full range may be larger than current video timing, so cap at nominal +/* Full range may be larger than current video timing, so cap at nominal */ if (min_refresh_in_uhz > max_refresh_in_uhz) min_refresh_in_uhz = max_refresh_in_uhz; -// If a monitor reports exactly max refresh of 2x of min, enforce it on nominal +/* If a monitor reports exactly max refresh of 2x of min, enforce it on nominal */ rounded_nominal_in_uhz =div_u64 (nominal_field_rate_in_uhz +50000 ,100000 ) *100000 ; if (in_config->max_refresh_in_uhz == (2 * in_config->min_refresh_in_uhz) &&
@@ -1042,7 +1042,7 @@ void mod_freesync_handle_preflip(struct mod_freesync *mod_freesync, last_render_time_in_us = curr_time_stamp_in_us - plane->time.prev_update_time_in_us; -// Sum off all entries except oldest one +/* Sum off all entries except oldest one */ for (i =0 ; i <DC_PLANE_UPDATE_TIMES_MAX ; i++) { average_render_time_in_us += plane->time.time_elapsed_in_us[i];
@@ -1050,7 +1050,7 @@ void mod_freesync_handle_preflip(struct mod_freesync *mod_freesync, average_render_time_in_us -= plane->time.time_elapsed_in_us[oldest_index]; -// Add render time for current flip +/* Add render time for current flip */ average_render_time_in_us += last_render_time_in_us; average_render_time_in_us /=DC_PLANE_UPDATE_TIMES_MAX ;
@@ -1125,8 +1125,9 @@ void mod_freesync_handle_v_update(struct mod_freesync *mod_freesync,if (in_out_vrr->state ==VRR_STATE_ACTIVE_VARIABLE ) in_out_vrr->fixed.ramping_active = false; -/* Gradual Static Screen Ramping Logic */ -/* Execute if ramp is active and user enabled freesync static screen*/ +/* Gradual Static Screen Ramping Logic +* Execute if ramp is active and user enabled freesync static screen +*/ if (in_out_vrr->state ==VRR_STATE_ACTIVE_FIXED && in_out_vrr->fixed.ramping_active) { update_v_total_for_static_ramp(
@@ -1214,20 +1215,20 @@ bool mod_freesync_is_valid_range(uint32_t min_refresh_cap_in_uhz, min_refresh_cap_in_uhz /=1000000 ; max_refresh_cap_in_uhz /=1000000 ; -// Check nominal is within range +/* Check nominal is within range */ if (nominal_field_rate_in_uhz > max_refresh_cap_in_uhz || nominal_field_rate_in_uhz < min_refresh_cap_in_uhz)return false; -// If nominal is less than max, limit the max allowed refresh rate +/* If nominal is less than max, limit the max allowed refresh rate */ if (nominal_field_rate_in_uhz < max_refresh_cap_in_uhz) max_refresh_cap_in_uhz = nominal_field_rate_in_uhz; -// Check min is within range +/* Check min is within range */ if (min_refresh_cap_in_uhz > max_refresh_cap_in_uhz)return false; -// For variable range, check for at least 10 Hz range +/* For variable range, check for at least 10 Hz range */ if (nominal_field_rate_in_uhz - min_refresh_cap_in_uhz <10 )return false;
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 6/18 drm/amd/display: Force enable pstate on driver unload
In reply to Eryk BrolFrom: Joshua Aberback <[email protected]>
[Why]
During driver unload, it is expected that p-state switching is supported. If it's not supported, PMFW will hang due to a forced p-state switch. Even if the current timing does not support p-state normally, we still want to force allow because the worst that can happen is underflow. This will match Navi10 behaviour.
[How]
- new hubbub func to control the force pstate register
- force allow when releasing display ownership
- registers are inaccessible after due to m_cgs.hwNotAvailable
- explicitly disable force signal during hw_init
- if driver is disabled and re-enabled, register not cleared otherwise
Also, remove DCN3 part of dcn10_init_hw, we will not be going back to it.
--- drivers/gpu/drm/amd/display/dc/core/dc.c | 7 +++++++ drivers/gpu/drm/amd/display/dc/dc.h | 3 +++ .../drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c | 6 ------ drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hubbub.c | 11 +++++++++++ drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hubbub.h | 3 +++ drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.c | 13 +++++++++++++ drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.h | 2 ++ drivers/gpu/drm/amd/display/dc/dcn30/dcn30_init.c | 1 + drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h | 4 ++++ drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h | 3 +++ 10 files changed, 47 insertions(+), 6 deletions(-)
drivers/gpu/drm/amd/display/dc/core/dc.c +7 -0
@@ -3097,4 +3097,11 @@ bool dc_is_plane_eligible_for_idle_optimizaitons(struct dc *dc, {return false; } + +/* cleanup on driver unload */ +void dc_hardware_release (struct dc *dc) +{ +if (dc->hwss.hardware_release) + dc->hwss.hardware_release (dc); +}#endif
drivers/gpu/drm/amd/display/dc/dc.h +3 -0
@@ -1267,6 +1267,9 @@ void dc_unlock_memory_clock_frequency(struct dc *dc);*/ void dc_lock_memory_clock_frequency (struct dc *dc); +/* cleanup on driver unload */ +void dc_hardware_release (struct dc *dc); +#endif bool dc_set_psr_allow_active (struct dc *dc,bool enable);
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +0 -6
@@ -1434,12 +1434,6 @@ void dcn10_init_hw(struct dc *dc)if (dc->clk_mgr->funcs->notify_wm_ranges) dc->clk_mgr->funcs->notify_wm_ranges (dc->clk_mgr); - -#ifdef CONFIG_DRM_AMD_DC_DCN3_0 -if (dc->clk_mgr->funcs->set_hard_max_memclk) - dc->clk_mgr->funcs->set_hard_max_memclk (dc->clk_mgr); -#endif - }/* In headless boot cases, DIG may be turned
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hubbub.c +11 -0
@@ -384,6 +384,16 @@ void hubbub3_force_wm_propagate_to_pipes(struct hubbub *hubbub) DCHUBBUB_ARB_VM_ROW_URGENCY_WATERMARK_A, prog_wm_value); } +void hubbub3_force_pstate_change_control (struct hubbub *hubbub, +bool force,bool allow) +{ +struct dcn20_hubbub *hubbub1 =TO_DCN20_HUBBUB (hubbub); + +REG_UPDATE_2 (DCHUBBUB_ARB_DRAM_STATE_CNTL , +DCHUBBUB_ARB_ALLOW_PSTATE_CHANGE_FORCE_VALUE , allow, +DCHUBBUB_ARB_ALLOW_PSTATE_CHANGE_FORCE_ENABLE , force); +} +static const struct hubbub_funcs hubbub30_funcs = { .update_dchub = hubbub2_update_dchub, .init_dchub_sys_ctx = hubbub3_init_dchub_sys_ctx,
@@ -397,6 +407,7 @@ static const struct hubbub_funcs hubbub30_funcs = { .allow_self_refresh_control = hubbub1_allow_self_refresh_control, .is_allow_self_refresh_enabled = hubbub1_is_allow_self_refresh_enabled, .force_wm_propagate_to_pipes = hubbub3_force_wm_propagate_to_pipes, + .force_pstate_change_control = hubbub3_force_pstate_change_control, }; void hubbub3_construct(struct dcn20_hubbub *hubbub3,
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hubbub.h +3 -0
@@ -116,4 +116,7 @@ bool hubbub3_program_watermarks(unsigned int refclk_mhz, bool safe_to_lower); +void hubbub3_force_pstate_change_control (struct hubbub *hubbub, +bool force,bool allow); +#endif
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.c +13 -0
@@ -622,6 +622,10 @@ void dcn30_init_hw(struct dc *dc)if (dc->clk_mgr->funcs->set_hard_max_memclk) dc->clk_mgr->funcs->set_hard_max_memclk (dc->clk_mgr); + +if (dc->res_pool->hubbub->funcs->force_pstate_change_control) + dc->res_pool->hubbub->funcs->force_pstate_change_control ( + dc->res_pool->hubbub, false, false); }void dcn30_set_avmute (struct pipe_ctx *pipe_ctx,bool enable)
@@ -711,3 +715,12 @@ bool dcn30_apply_idle_power_optimizations(struct dc *dc, bool enable)return true; } + +void dcn30_hardware_release (struct dc *dc) +{ +/* if pstate unsupported, force it supported */ +if (!dc->clk_mgr->clks.p_state_change_support && + dc->res_pool->hubbub->funcs->force_pstate_change_control) + dc->res_pool->hubbub->funcs->force_pstate_change_control ( + dc->res_pool->hubbub, true, true); +}
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.h +2 -0
@@ -67,4 +67,6 @@ void dcn30_program_dmdata_engine(struct pipe_ctx *pipe_ctx);bool dcn30_apply_idle_power_optimizations (struct dc *dc,bool enable); +void dcn30_hardware_release (struct dc *dc); +#endif /* __DC_HWSS_DCN30_H__ */
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_init.c +1 -0
@@ -93,6 +93,7 @@ static const struct hw_sequencer_funcs dcn30_funcs = { .apply_idle_power_optimizations = dcn30_apply_idle_power_optimizations, .set_backlight_level = dcn21_set_backlight_level, .set_abm_immediate_disable = dcn21_set_abm_immediate_disable, + .hardware_release = dcn30_hardware_release, };static const struct hwseq_private_funcs dcn30_private_funcs = {
drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h +4 -0
@@ -152,6 +152,10 @@ struct hubbub_funcs {void (*apply_DEDCN21_147_wa)(struct hubbub *hubbub);void (*force_wm_propagate_to_pipes)(struct hubbub *hubbub); +#if defined(CONFIG_DRM_AMD_DC_DCN3_0 ) + +void (*force_pstate_change_control)(struct hubbub *hubbub,bool force,bool allow); +#endif };struct hubbub {
drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h +3 -0
@@ -50,6 +50,9 @@ struct dpp;struct dce_hwseq ;struct hw_sequencer_funcs { +#ifdef CONFIG_DRM_AMD_DC_DCN3_0 +void (*hardware_release)(struct dc *dc); +#endif /* Embedded Display Related */ void (*edp_power_control)(struct dc_link *link,bool enable);void (*edp_wait_for_hpd_ready)(struct dc_link *link,bool power_up);
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 7/18 drm/amd/display: Copy WM values from set A to other sets in hw_init
In reply to Eryk BrolFrom: Joshua Aberback <[email protected]>
[Why]
When we transfer the WM range table to SMU, they can perform a watermark switch right away. This can be a problem if we're in not in accelerated mode during hw_init as SMU may initiate a dummy p-state change before the rest of the watermarks are programmed. Watermark set A is defined to be sufficient for all cases, so we can copy the values from set A to all other sets, avoiding any issues from SMU doing WM switches.
[How]
- new hubbub func init_watermarks
- copy register values from set A to all other sets
- call init_watermarks before calling notify_wm_ranges
--- .../drm/amd/display/dc/dcn30/dcn30_hubbub.c | 43 +++++++++++++++++++ .../drm/amd/display/dc/dcn30/dcn30_hubbub.h | 2 + .../gpu/drm/amd/display/dc/inc/hw/dchubbub.h | 2 + 3 files changed, 47 insertions(+)
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hubbub.c +43 -0
@@ -394,6 +394,48 @@ void hubbub3_force_pstate_change_control(struct hubbub *hubbub, DCHUBBUB_ARB_ALLOW_PSTATE_CHANGE_FORCE_ENABLE, force); } +/* Copy values from WM set A to all other sets */ +void hubbub3_init_watermarks (struct hubbub *hubbub) +{ +struct dcn20_hubbub *hubbub1 =TO_DCN20_HUBBUB (hubbub); +uint32_t reg; + + reg =REG_READ (DCHUBBUB_ARB_DATA_URGENCY_WATERMARK_A ); +REG_WRITE (DCHUBBUB_ARB_DATA_URGENCY_WATERMARK_B , reg); +REG_WRITE (DCHUBBUB_ARB_DATA_URGENCY_WATERMARK_C , reg); +REG_WRITE (DCHUBBUB_ARB_DATA_URGENCY_WATERMARK_D , reg); + + reg =REG_READ (DCHUBBUB_ARB_FRAC_URG_BW_FLIP_A ); +REG_WRITE (DCHUBBUB_ARB_FRAC_URG_BW_FLIP_B , reg); +REG_WRITE (DCHUBBUB_ARB_FRAC_URG_BW_FLIP_C , reg); +REG_WRITE (DCHUBBUB_ARB_FRAC_URG_BW_FLIP_D , reg); + + reg =REG_READ (DCHUBBUB_ARB_FRAC_URG_BW_NOM_A ); +REG_WRITE (DCHUBBUB_ARB_FRAC_URG_BW_NOM_B , reg); +REG_WRITE (DCHUBBUB_ARB_FRAC_URG_BW_NOM_C , reg); +REG_WRITE (DCHUBBUB_ARB_FRAC_URG_BW_NOM_D , reg); + + reg =REG_READ (DCHUBBUB_ARB_REFCYC_PER_TRIP_TO_MEMORY_A ); +REG_WRITE (DCHUBBUB_ARB_REFCYC_PER_TRIP_TO_MEMORY_B , reg); +REG_WRITE (DCHUBBUB_ARB_REFCYC_PER_TRIP_TO_MEMORY_C , reg); +REG_WRITE (DCHUBBUB_ARB_REFCYC_PER_TRIP_TO_MEMORY_D , reg); + + reg =REG_READ (DCHUBBUB_ARB_ALLOW_SR_ENTER_WATERMARK_A ); +REG_WRITE (DCHUBBUB_ARB_ALLOW_SR_ENTER_WATERMARK_B , reg); +REG_WRITE (DCHUBBUB_ARB_ALLOW_SR_ENTER_WATERMARK_C , reg); +REG_WRITE (DCHUBBUB_ARB_ALLOW_SR_ENTER_WATERMARK_D , reg); + + reg =REG_READ (DCHUBBUB_ARB_ALLOW_SR_EXIT_WATERMARK_A ); +REG_WRITE (DCHUBBUB_ARB_ALLOW_SR_EXIT_WATERMARK_B , reg); +REG_WRITE (DCHUBBUB_ARB_ALLOW_SR_EXIT_WATERMARK_C , reg); +REG_WRITE (DCHUBBUB_ARB_ALLOW_SR_EXIT_WATERMARK_D , reg); + + reg =REG_READ (DCHUBBUB_ARB_ALLOW_DRAM_CLK_CHANGE_WATERMARK_A ); +REG_WRITE (DCHUBBUB_ARB_ALLOW_DRAM_CLK_CHANGE_WATERMARK_B , reg); +REG_WRITE (DCHUBBUB_ARB_ALLOW_DRAM_CLK_CHANGE_WATERMARK_C , reg); +REG_WRITE (DCHUBBUB_ARB_ALLOW_DRAM_CLK_CHANGE_WATERMARK_D , reg); +} +static const struct hubbub_funcs hubbub30_funcs = { .update_dchub = hubbub2_update_dchub, .init_dchub_sys_ctx = hubbub3_init_dchub_sys_ctx,
@@ -408,6 +450,7 @@ static const struct hubbub_funcs hubbub30_funcs = { .is_allow_self_refresh_enabled = hubbub1_is_allow_self_refresh_enabled, .force_wm_propagate_to_pipes = hubbub3_force_wm_propagate_to_pipes, .force_pstate_change_control = hubbub3_force_pstate_change_control, + .init_watermarks = hubbub3_init_watermarks, }; void hubbub3_construct(struct dcn20_hubbub *hubbub3,
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hubbub.h +2 -0
@@ -119,4 +119,6 @@ bool hubbub3_program_watermarks(void hubbub3_force_pstate_change_control (struct hubbub *hubbub,bool force,bool allow); +void hubbub3_init_watermarks (struct hubbub *hubbub); +#endif
drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h +2 -0
@@ -155,6 +155,8 @@ struct hubbub_funcs {#if defined(CONFIG_DRM_AMD_DC_DCN3_0 )void (*force_pstate_change_control)(struct hubbub *hubbub,bool force,bool allow); + +void (*init_watermarks)(struct hubbub *hubbub);#endif };
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 8/18 drm/amd/display: Block ABM in case of eDP ODM
In reply to Eryk BrolFrom: Yongqiang Sun <[email protected]>
[Why]
enable ODM on eDP panel with ABM will result in color difference on the panel due to only one ABM module to set one pipe.
[How]
Block ABM in case of ODM enabled on eDP.
--- drivers/gpu/drm/amd/display/dc/dc_stream.h | 1 + .../gpu/drm/amd/display/dc/dcn21/dcn21_hwseq.c | 15 +++++++++++++++ .../gpu/drm/amd/display/dc/dcn21/dcn21_hwseq.h | 2 ++ drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c | 1 + .../gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.c | 3 +++ drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h | 2 ++ 6 files changed, 24 insertions(+)
drivers/gpu/drm/amd/display/dc/dc_stream.h +1 -0
@@ -45,6 +45,7 @@ struct dc_stream_status {int audio_inst;struct timing_sync_info timing_sync_info;struct dc_plane_state *plane_states[MAX_SURFACE_NUM ]; +bool is_abm_supported; };// TODO: References to this needs to be removed..
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hwseq.c +15 -0
@@ -223,3 +223,18 @@ bool dcn21_set_backlight_level(struct pipe_ctx *pipe_ctx,return true; } +bool dcn21_is_abm_supported (struct dc *dc, +struct dc_state *context,struct dc_stream_state *stream) +{ +int i; + +for (i =0 ; i < dc->res_pool->pipe_count; i++) { +struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; + +if (pipe_ctx->stream == stream && + (pipe_ctx->prev_odm_pipe ==NULL && pipe_ctx->next_odm_pipe ==NULL )) +return true; + } +return false; +} +
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hwseq.h +2 -0
@@ -52,5 +52,7 @@ void dcn21_set_abm_immediate_disable(struct pipe_ctx *pipe_ctx);bool dcn21_set_backlight_level (struct pipe_ctx *pipe_ctx,uint32_t backlight_pwm_u16_16,uint32_t frame_ramp); +bool dcn21_is_abm_supported (struct dc *dc, +struct dc_state *context,struct dc_stream_state *stream);#endif /* __DC_HWSS_DCN21_H__ */
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c +1 -0
@@ -98,6 +98,7 @@ static const struct hw_sequencer_funcs dcn21_funcs = { #ifndef TRIM_FSFT .optimize_timing_for_fsft = dcn20_optimize_timing_for_fsft, #endif + .is_abm_supported = dcn21_is_abm_supported, }; static const struct hwseq_private_funcs dcn21_private_funcs = {
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.c +3 -0
@@ -617,6 +617,9 @@ void dcn30_init_hw(struct dc *dc)if (hws->funcs.enable_power_gating_plane) hws->funcs.enable_power_gating_plane (dc->hwseq, true); +if (!dcb->funcs->is_accelerated_mode (dcb) && dc->res_pool->hubbub->funcs->init_watermarks) + dc->res_pool->hubbub->funcs->init_watermarks (dc->res_pool->hubbub); +if (dc->clk_mgr->funcs->notify_wm_ranges) dc->clk_mgr->funcs->notify_wm_ranges (dc->clk_mgr);
drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h +2 -0
@@ -223,6 +223,8 @@ struct hw_sequencer_funcs {bool (*apply_idle_power_optimizations)(struct dc *dc,bool enable);#endif +bool (*is_abm_supported)(struct dc *dc, +struct dc_state *context,struct dc_stream_state *stream); }; void color_space_to_black_color(
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 9/18 drm/amd/display: Add i2c speed arbitration for dc_i2c and hdcp_i2c
In reply to Eryk BrolFrom: Charlene Liu <[email protected]>
[why]
HDCP 1.4 failed on SL8800 SW w/a test driver use.
[how]
Slow down the HW i2c speed when used by HW i2c.
This request: each acquired_i2c_engine setup the i2c speed needed and sets the I2c engine for HDCP use at release_engine.
This covers SW using HW i2c engine and HDCP using HW i2c engine. for dmcu using HW i2c engine, needs add similar logic in dmcu fw.
--- drivers/gpu/drm/amd/display/dc/dc.h | 1 + drivers/gpu/drm/amd/display/dc/dce/dce_i2c_hw.c | 9 +++++++-- drivers/gpu/drm/amd/display/dc/dce100/dce100_resource.c | 1 + drivers/gpu/drm/amd/display/dc/dce110/dce110_resource.c | 3 ++- drivers/gpu/drm/amd/display/dc/dce112/dce112_resource.c | 1 + drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c | 1 + drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c | 3 +++ drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c | 1 + drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c | 1 + drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c | 1 + drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c | 1 + 11 files changed, 20 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/display/dc/dc.h +1 -0
@@ -152,6 +152,7 @@ struct dc_caps {uint32_t max_planes;uint32_t max_downscale_ratio;uint32_t i2c_speed_in_khz; +uint32_t i2c_speed_in_khz_hdcp;uint32_t dmdata_alloc_size;unsigned int max_cursor_size;unsigned int max_video_width;
drivers/gpu/drm/amd/display/dc/dce/dce_i2c_hw.c +7 -2
@@ -299,8 +299,12 @@ static bool setup_engine(/* we have checked I2c not used by DMCU, set SW use I2C REQ to 1 to indicate SW using it*/ REG_UPDATE (DC_I2C_ARBITRATION ,DC_I2C_SW_USE_I2C_REG_REQ ,1 ); +/*set SW requested I2c speed to default, if API calls in it will be override later*/ +set_speed (dce_i2c_hw, dce_i2c_hw->ctx->dc->caps.i2c_speed_in_khz); +if (dce_i2c_hw->setup_limit !=0 ) i2c_setup_limit = dce_i2c_hw->setup_limit; +/* Program pin select */ REG_UPDATE_6(DC_I2C_CONTROL, DC_I2C_GO, 0,
@@ -339,8 +343,6 @@ static void release_engine( {bool safe_to_reset; -/* Restore original HW engine speed */ -set_speed (dce_i2c_hw, dce_i2c_hw->default_speed);/* Reset HW engine */ {
@@ -360,6 +362,9 @@ static void release_engine(/* HW I2c engine - clock gating feature */ if (!dce_i2c_hw->engine_keep_power_up_count)REG_UPDATE_N (SETUP ,1 ,FN (SETUP ,DC_I2C_DDC1_ENABLE ),0 ); + +/*for HW HDCP Ri polling failure w/a test*/ +set_speed (dce_i2c_hw, dce_i2c_hw->ctx->dc->caps.i2c_speed_in_khz_hdcp);/* Release I2C after reset, so HW or DMCU could use it */ REG_UPDATE_2 (DC_I2C_ARBITRATION ,DC_I2C_SW_DONE_USING_I2C_REG ,1 ,DC_I2C_SW_USE_I2C_REG_REQ ,0 );
drivers/gpu/drm/amd/display/dc/dce100/dce100_resource.c +1 -0
@@ -1071,6 +1071,7 @@ static bool dce100_resource_construct( pool->base.timing_generator_count = pool->base.res_cap->num_timing_generator; dc->caps.max_downscale_ratio =200 ; dc->caps.i2c_speed_in_khz =40 ; + dc->caps.i2c_speed_in_khz =40 ; dc->caps.max_cursor_size =128 ; dc->caps.dual_link_dvi = true; dc->caps.disable_dp_clk_share = true;
drivers/gpu/drm/amd/display/dc/dce110/dce110_resource.c +2 -1
@@ -1372,7 +1372,8 @@ static bool dce110_resource_construct( pool->base.underlay_pipe_index = pool->base.pipe_count; pool->base.timing_generator_count = pool->base.res_cap->num_timing_generator; dc->caps.max_downscale_ratio =150 ; - dc->caps.i2c_speed_in_khz =100 ; + dc->caps.i2c_speed_in_khz =40 ; + dc->caps.i2c_speed_in_khz_hdcp =40 ; dc->caps.max_cursor_size =128 ; dc->caps.is_apu = true; dc->caps.extended_aux_timeout_support = false;
drivers/gpu/drm/amd/display/dc/dce112/dce112_resource.c +1 -0
@@ -1240,6 +1240,7 @@ static bool dce112_resource_construct( pool->base.timing_generator_count = pool->base.res_cap->num_timing_generator; dc->caps.max_downscale_ratio =200 ; dc->caps.i2c_speed_in_khz =100 ; + dc->caps.i2c_speed_in_khz_hdcp =100 ;/*1.4 w/a not applied by default*/ dc->caps.max_cursor_size =128 ; dc->caps.dual_link_dvi = true; dc->caps.extended_aux_timeout_support = false;
drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c +1 -0
@@ -1080,6 +1080,7 @@ static bool dce120_resource_construct( dc->caps.max_downscale_ratio =200 ; dc->caps.i2c_speed_in_khz =100 ; + dc->caps.i2c_speed_in_khz_hdcp =100 ;/*1.4 w/a not applied by default*/ dc->caps.max_cursor_size =128 ; dc->caps.dual_link_dvi = true; dc->caps.psp_setup_panel_mode = true;
drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c +3 -0
@@ -970,6 +970,7 @@ static bool dce80_construct( pool->base.timing_generator_count = res_cap.num_timing_generator; dc->caps.max_downscale_ratio =200 ; dc->caps.i2c_speed_in_khz =40 ; + dc->caps.i2c_speed_in_khz_hdcp =40 ; dc->caps.max_cursor_size =128 ; dc->caps.dual_link_dvi = true; dc->caps.extended_aux_timeout_support = false;
@@ -1168,6 +1169,7 @@ static bool dce81_construct( pool->base.timing_generator_count = res_cap_81.num_timing_generator; dc->caps.max_downscale_ratio =200 ; dc->caps.i2c_speed_in_khz =40 ; + dc->caps.i2c_speed_in_khz_hdcp =40 ; dc->caps.max_cursor_size =128 ; dc->caps.is_apu = true;
@@ -1365,6 +1367,7 @@ static bool dce83_construct( pool->base.timing_generator_count = res_cap_83.num_timing_generator; dc->caps.max_downscale_ratio =200 ; dc->caps.i2c_speed_in_khz =40 ; + dc->caps.i2c_speed_in_khz_hdcp =40 ; dc->caps.max_cursor_size =128 ; dc->caps.is_apu = true;
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c +1 -0
@@ -1416,6 +1416,7 @@ static bool dcn10_resource_construct( dc->caps.max_video_width =3840 ; dc->caps.max_downscale_ratio =200 ; dc->caps.i2c_speed_in_khz =100 ; + dc->caps.i2c_speed_in_khz_hdcp =100 ;/*1.4 w/a not applied by default*/ dc->caps.max_cursor_size =256 ; dc->caps.max_slave_planes =1 ; dc->caps.is_apu = true;
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +1 -0
@@ -3801,6 +3801,7 @@ static bool dcn20_resource_construct( dc->caps.max_downscale_ratio =200 ; dc->caps.i2c_speed_in_khz =100 ; + dc->caps.i2c_speed_in_khz_hdcp =100 ;/*1.4 w/a not applied by default*/ dc->caps.max_cursor_size =256 ; dc->caps.dmdata_alloc_size =2048 ;
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +1 -0
@@ -1808,6 +1808,7 @@ static bool dcn21_resource_construct( dc->caps.max_downscale_ratio =200 ; dc->caps.i2c_speed_in_khz =100 ; + dc->caps.i2c_speed_in_khz_hdcp =5 ;/*1.4 w/a applied by default*/ dc->caps.max_cursor_size =256 ; dc->caps.dmdata_alloc_size =2048 ;
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c +1 -0
@@ -2519,6 +2519,7 @@ static bool dcn30_resource_construct( pool->base.mpcc_count = pool->base.res_cap->num_timing_generator; dc->caps.max_downscale_ratio =600 ; dc->caps.i2c_speed_in_khz =100 ; + dc->caps.i2c_speed_in_khz_hdcp =100 ;/*1.4 w/a not applied by default*/ dc->caps.max_cursor_size =256 ; dc->caps.dmdata_alloc_size =2048 ;
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 10/18 drm/amd/display: HDMI remote sink need mode validation for Linux
In reply to Eryk BrolFrom: Fangzhi Zuo <[email protected]>
[Why]
Currently mode validation is bypassed if remote sink exists. That leads to mode set issue when a BW bottle neck exists in the link path, e.g., a DP-to-HDMI converter that only supports HDMI 1.4.
Any invalid mode passed to Linux user space will cause the modeset failure due to limitation of Linux user space implementation.
[How]
Mode validation is skipped only if in edid override. For real remote sink, clock limit check should be done for HDMI remote sink.
Have HDMI related remote sink going through mode validation to elimiate modes which pixel clock exceeds BW limitation.
--- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/core/dc_link.c +1 -1
@@ -2441,7 +2441,7 @@ enum dc_status dc_link_validate_mode_timing(/* A hack to avoid failing any modes for EDID override feature on * topology change such as lower quality cable for DP or different dongle */ -if (link->remote_sinks[0 ]) +if (link->remote_sinks[0 ] && link->remote_sinks[0 ]->sink_signal ==SIGNAL_TYPE_VIRTUAL )return DC_OK ;/* Passive Dongle */
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 11/18 drm/amd/display: Fix OPTC_DATA_FORMAT programming
In reply to Eryk BrolFrom: Dmytro Laktyushkin <[email protected]>
This should be programmed with timing rather than with odm.
--- drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c | 11 +++++++++++ drivers/gpu/drm/amd/display/dc/dcn20/dcn20_optc.c | 8 -------- drivers/gpu/drm/amd/display/dc/dcn30/dcn30_optc.c | 8 -------- 3 files changed, 11 insertions(+), 16 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c +11 -0
@@ -288,6 +288,17 @@ void optc1_program_timing(if (optc1_is_two_pixels_per_containter (&patched_crtc_timing) || optc1->opp_count ==2 ) h_div =H_TIMING_DIV_BY2 ; +if (REG (OPTC_DATA_FORMAT_CONTROL )) { +uint32_t data_fmt =0 ; + +if (patched_crtc_timing.pixel_encoding ==PIXEL_ENCODING_YCBCR422 ) + data_fmt =1 ; +else if (patched_crtc_timing.pixel_encoding ==PIXEL_ENCODING_YCBCR420 ) + data_fmt =2 ; + +REG_UPDATE (OPTC_DATA_FORMAT_CONTROL ,OPTC_DATA_FORMAT , data_fmt); + } +#if defined(CONFIG_DRM_AMD_DC_DCN3_0 )if (optc1->tg_mask->OTG_H_TIMING_DIV_MODE !=0 ) {if (optc1->opp_count ==4 )
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_optc.c +0 -8
@@ -239,7 +239,6 @@ void optc2_set_odm_combine(struct timing_generator *optc, int *opp_id, int opp_cint mpcc_hactive = (timing->h_addressable + timing->h_border_left + timing->h_border_right) / opp_cnt;uint32_t memory_mask; -uint32_t data_fmt =0 ;ASSERT (opp_cnt ==2 );
@@ -262,13 +261,6 @@ void optc2_set_odm_combine(struct timing_generator *optc, int *opp_id, int opp_cREG_SET (OPTC_MEMORY_CONFIG ,0 ,OPTC_MEM_SEL , memory_mask); -if (timing->pixel_encoding ==PIXEL_ENCODING_YCBCR422 ) - data_fmt =1 ; -else if (timing->pixel_encoding ==PIXEL_ENCODING_YCBCR420 ) - data_fmt =2 ; - -REG_UPDATE (OPTC_DATA_FORMAT_CONTROL ,OPTC_DATA_FORMAT , data_fmt); - REG_SET_3(OPTC_DATA_SOURCE_SELECT, 0, OPTC_NUM_OF_INPUT_SEGMENT, 1, OPTC_SEG0_SRC_SEL, opp_id[0],
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_optc.c +0 -8
@@ -209,7 +209,6 @@ static void optc3_set_odm_combine(struct timing_generator *optc, int *opp_id, inint mpcc_hactive = (timing->h_addressable + timing->h_border_left + timing->h_border_right) / opp_cnt;uint32_t memory_mask =0 ; -uint32_t data_fmt =0 ;/* TODO: In pseudocode but does not affect maximus, delete comment if we dont need on asic * REG_SET(OTG_GLOBAL_CONTROL2, 0, GLOBAL_UPDATE_LOCK_EN, 1);
@@ -240,13 +239,6 @@ static void optc3_set_odm_combine(struct timing_generator *optc, int *opp_id, inREG_SET (OPTC_MEMORY_CONFIG ,0 ,OPTC_MEM_SEL , memory_mask); -if (timing->pixel_encoding ==PIXEL_ENCODING_YCBCR422 ) - data_fmt =1 ; -else if (timing->pixel_encoding ==PIXEL_ENCODING_YCBCR420 ) - data_fmt =2 ; - -REG_UPDATE (OPTC_DATA_FORMAT_CONTROL ,OPTC_DATA_FORMAT , data_fmt); -if (opp_cnt ==2 ) { REG_SET_3(OPTC_DATA_SOURCE_SELECT, 0, OPTC_NUM_OF_INPUT_SEGMENT, 1,
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 12/18 drm/amd/display: FreeSync not active near lower bound of non-LFC monitor range
In reply to Eryk BrolFrom: Aric Cyr <[email protected]>
[Why]
On narrow range monitors without LFC, a margin prevents good utilization of the available range.
[How]
Decrease the margin for exiting fixed mode and fix the frame counter to reset if a non-consecutive render is found.
--- .../gpu/drm/amd/display/modules/freesync/freesync.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/display/modules/freesync/freesync.c +6 -4
@@ -42,10 +42,10 @@/* Threshold to change BTR multiplier (to avoid frequent changes) */ #define BTR_DRIFT_MARGIN 2000/* Threshold to exit fixed refresh rate */ -#define FIXED_REFRESH_EXIT_MARGIN_IN_HZ 4 +#define FIXED_REFRESH_EXIT_MARGIN_IN_HZ 1/* Number of consecutive frames to check before entering/exiting fixed refresh */ #define FIXED_REFRESH_ENTER_FRAME_COUNT 5 -#define FIXED_REFRESH_EXIT_FRAME_COUNT 5 +#define FIXED_REFRESH_EXIT_FRAME_COUNT 10struct core_freesync {struct mod_freesync public;
@@ -420,7 +420,8 @@ static void apply_fixed_refresh(struct core_freesync *core_freesync, in_out_vrr->fixed.target_refresh_in_uhz =0 ; update = true; } - } + }else + in_out_vrr->fixed.frame_counter =0 ; }else if (last_render_time_in_us > max_render_time_in_us) {/* Enter Fixed Refresh mode */ if (!in_out_vrr->fixed.fixed_active) {
@@ -434,7 +435,8 @@ static void apply_fixed_refresh(struct core_freesync *core_freesync, in_out_vrr->max_refresh_in_uhz; update = true; } - } + }else + in_out_vrr->fixed.frame_counter =0 ; }if (update) {
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 13/18 drm/amd/display: Hardcode 45W Bounding Box for DCN2.1 Diags
In reply to Eryk BrolFrom: Sung Lee <[email protected]>
[WHY]
Currently construction of clock limits gets skipped for diags. This logic would be useful to get tested in diags.
[HOW]
Copy existing states to clk_table such that update_bw_bounding_box logic gets used to fill the table.
--- .../drm/amd/display/dc/dcn21/dcn21_resource.c | 119 ++++++------------ 1 file changed, 36 insertions(+), 83 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +36 -83
@@ -1393,87 +1393,46 @@ static void update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_paramunsigned int i, closest_clk_lvl;int j; -// Default clock levels are used for diags, which may lead to overclocking. -if (!IS_DIAG_DC (dc->ctx->dce_environment)) { - dcn2_1_ip.max_num_otg = pool->base.res_cap->num_timing_generator; - dcn2_1_ip.max_num_dpp = pool->base.pipe_count; - dcn2_1_soc.num_chans = bw_params->num_channels; - -ASSERT (clk_table->num_entries); -for (i =0 ; i < clk_table->num_entries; i++) { -/* loop backwards*/ -for (closest_clk_lvl =0 , j = dcn2_1_soc.num_states -1 ; j >=0 ; j--) { -if ((unsigned int ) dcn2_1_soc.clock_limits[j].dcfclk_mhz <= clk_table->entries[i].dcfclk_mhz) { - closest_clk_lvl = j; -break ; - } + dcn2_1_ip.max_num_otg = pool->base.res_cap->num_timing_generator; + dcn2_1_ip.max_num_dpp = pool->base.pipe_count; + dcn2_1_soc.num_chans = bw_params->num_channels; + +ASSERT (clk_table->num_entries); +for (i =0 ; i < clk_table->num_entries; i++) { +/* loop backwards*/ +for (closest_clk_lvl =0 , j = dcn2_1_soc.num_states -1 ; j >=0 ; j--) { +if ((unsigned int ) dcn2_1_soc.clock_limits[j].dcfclk_mhz <= clk_table->entries[i].dcfclk_mhz) { + closest_clk_lvl = j; +break ; } - - clock_limits[i].state = i; - clock_limits[i].dcfclk_mhz = clk_table->entries[i].dcfclk_mhz; - clock_limits[i].fabricclk_mhz = clk_table->entries[i].fclk_mhz; - clock_limits[i].socclk_mhz = clk_table->entries[i].socclk_mhz; - clock_limits[i].dram_speed_mts = clk_table->entries[i].memclk_mhz *2 ; - - clock_limits[i].dispclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dispclk_mhz; - clock_limits[i].dppclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dppclk_mhz; - clock_limits[i].dram_bw_per_chan_gbps = dcn2_1_soc.clock_limits[closest_clk_lvl].dram_bw_per_chan_gbps; - clock_limits[i].dscclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dscclk_mhz; - clock_limits[i].dtbclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dtbclk_mhz; - clock_limits[i].phyclk_d18_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].phyclk_d18_mhz; - clock_limits[i].phyclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].phyclk_mhz; - } -for (i =0 ; i < clk_table->num_entries; i++) - dcn2_1_soc.clock_limits[i] = clock_limits[i]; -if (clk_table->num_entries) { - dcn2_1_soc.num_states = clk_table->num_entries; -/* duplicate last level */ - dcn2_1_soc.clock_limits[dcn2_1_soc.num_states] = dcn2_1_soc.clock_limits[dcn2_1_soc.num_states -1 ]; - dcn2_1_soc.clock_limits[dcn2_1_soc.num_states].state = dcn2_1_soc.num_states; } + + clock_limits[i].state = i; + clock_limits[i].dcfclk_mhz = clk_table->entries[i].dcfclk_mhz; + clock_limits[i].fabricclk_mhz = clk_table->entries[i].fclk_mhz; + clock_limits[i].socclk_mhz = clk_table->entries[i].socclk_mhz; + clock_limits[i].dram_speed_mts = clk_table->entries[i].memclk_mhz *2 ; + + clock_limits[i].dispclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dispclk_mhz; + clock_limits[i].dppclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dppclk_mhz; + clock_limits[i].dram_bw_per_chan_gbps = dcn2_1_soc.clock_limits[closest_clk_lvl].dram_bw_per_chan_gbps; + clock_limits[i].dscclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dscclk_mhz; + clock_limits[i].dtbclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dtbclk_mhz; + clock_limits[i].phyclk_d18_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].phyclk_d18_mhz; + clock_limits[i].phyclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].phyclk_mhz; + } +for (i =0 ; i < clk_table->num_entries; i++) + dcn2_1_soc.clock_limits[i] = clock_limits[i]; +if (clk_table->num_entries) { + dcn2_1_soc.num_states = clk_table->num_entries; +/* duplicate last level */ + dcn2_1_soc.clock_limits[dcn2_1_soc.num_states] = dcn2_1_soc.clock_limits[dcn2_1_soc.num_states -1 ]; + dcn2_1_soc.clock_limits[dcn2_1_soc.num_states].state = dcn2_1_soc.num_states; }dml_init_instance (&dc->dml, &dcn2_1_soc, &dcn2_1_ip,DML_PROJECT_DCN21 ); } -/* Temporary Place holder until we can get them from fuse */ -static struct dpm_clocks dummy_clocks = { - .DcfClocks = { - {.Freq =400 , .Vol =1 }, - {.Freq =483 , .Vol =1 }, - {.Freq =602 , .Vol =1 }, - {.Freq =738 , .Vol =1 } }, - .SocClocks = { - {.Freq =300 , .Vol =1 }, - {.Freq =400 , .Vol =1 }, - {.Freq =400 , .Vol =1 }, - {.Freq =400 , .Vol =1 } }, - .FClocks = { - {.Freq =400 , .Vol =1 }, - {.Freq =800 , .Vol =1 }, - {.Freq =1067 , .Vol =1 }, - {.Freq =1600 , .Vol =1 } }, - .MemClocks = { - {.Freq =800 , .Vol =1 }, - {.Freq =1600 , .Vol =1 }, - {.Freq =1067 , .Vol =1 }, - {.Freq =1600 , .Vol =1 } }, - -}; - -static enum pp_smu_status dummy_set_wm_ranges (struct pp_smu *pp, -struct pp_smu_wm_range_sets *ranges) -{ -return PP_SMU_RESULT_OK ; -} - -static enum pp_smu_status dummy_get_dpm_clock_table (struct pp_smu *pp, -struct dpm_clocks *clock_table) -{ - *clock_table = dummy_clocks; -return PP_SMU_RESULT_OK ; -} -static struct pp_smu_funcs *dcn21_pp_smu_create (struct dc_context *ctx) {struct pp_smu_funcs *pp_smu =kzalloc (sizeof (*pp_smu),GFP_KERNEL );
@@ -1481,17 +1440,11 @@ static struct pp_smu_funcs *dcn21_pp_smu_create(struct dc_context *ctx)if (!pp_smu)return pp_smu; -if (IS_FPGA_MAXIMUS_DC (ctx->dce_environment) ||IS_DIAG_DC (ctx->dce_environment)) { - pp_smu->ctx.ver =PP_SMU_VER_RN ; - pp_smu->rn_funcs.get_dpm_clock_table = dummy_get_dpm_clock_table; - pp_smu->rn_funcs.set_wm_ranges = dummy_set_wm_ranges; - }else { +dm_pp_get_funcs (ctx, pp_smu); -dm_pp_get_funcs (ctx, pp_smu); +if (pp_smu->ctx.ver !=PP_SMU_VER_RN ) + pp_smu =memset (pp_smu,0 ,sizeof (struct pp_smu_funcs )); -if (pp_smu->ctx.ver !=PP_SMU_VER_RN ) - pp_smu =memset (pp_smu,0 ,sizeof (struct pp_smu_funcs )); - }return pp_smu; }
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 14/18 drm/amd/display: Program meta addresses correctly
In reply to Eryk BrolFrom: Alvin Lee <[email protected]>
[Why]
When forcing 3D mode in DAL, we set the right address to be the same as the left address. We need to do the same for the meta addresses.
[How]
Program right meta to be same as left meta.
--- drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c | 2 ++ drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 2 ++ 2 files changed, 4 insertions(+)
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +2 -0
@@ -1535,6 +1535,8 @@ static bool patch_address_for_sbs_tb_stereo( plane_state->address.type =PLN_ADDR_TYPE_GRPH_STEREO ; plane_state->address.grph_stereo.right_addr = plane_state->address.grph_stereo.left_addr; + plane_state->address.grph_stereo.right_meta_addr = + plane_state->address.grph_stereo.left_meta_addr; } }return false;
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +2 -0
@@ -2034,6 +2034,8 @@ static bool patch_address_for_sbs_tb_stereo( plane_state->address.type =PLN_ADDR_TYPE_GRPH_STEREO ; plane_state->address.grph_stereo.right_addr = plane_state->address.grph_stereo.left_addr; + plane_state->address.grph_stereo.right_meta_addr = + plane_state->address.grph_stereo.left_meta_addr; }return false; }
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 15/18 drm/amd/display: Only flush inst_fb if backdoor loading
In reply to Eryk BrolFrom: Alvin Lee <[email protected]>
[Why]
DAL resume from BACO time is longer if we always flush inst_fb
[How]
Check if backdoor loading to flush inst_fb
--- drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-)
drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c +9 -8
@@ -404,15 +404,16 @@ 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) { +/** +* 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); + dmub->hw_funcs.backdoor_load (dmub, &cw0, &cw1); + } -if (params->load_inst_const && dmub->hw_funcs.backdoor_load) - dmub->hw_funcs.backdoor_load (dmub, &cw0, &cw1); }if (dmub->hw_funcs.reset)
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 16/18 drm/amd/display: [FW Promotion] Release 0.0.36
In reply to Eryk BrolFrom: Anthony Koo <[email protected]>
--- drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +2 -2
@@ -36,10 +36,10 @@/* Firmware versioning. */ #ifdef DMUB_EXPOSE_VERSION -#define DMUB_FW_VERSION_GIT_HASH 0x9cf8f05fe +#define DMUB_FW_VERSION_GIT_HASH 0x959f9d769#define DMUB_FW_VERSION_MAJOR 0#define DMUB_FW_VERSION_MINOR 0 -#define DMUB_FW_VERSION_REVISION 35 +#define DMUB_FW_VERSION_REVISION 36#define DMUB_FW_VERSION_TEST 0#define DMUB_FW_VERSION_VBIOS 0#define DMUB_FW_VERSION_HOTFIX 0
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 17/18 drm/amd/display: 3.2.106
In reply to Eryk BrolFrom: 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
@@ -42,7 +42,7 @@#include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#define DC_VER "3.2.105" +#define DC_VER "3.2.106"#define MAX_SURFACES 3#define MAX_PLANES 6
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 18/18 drm/amd/display: Avoid set zero in the requested clk
In reply to Eryk BrolFrom: Rodrigo Siqueira <[email protected]>
[Why]
Sometimes CRTCs can be disabled due to display unplugging or temporarily transition in the userspace; in these circumstances, DCE tries to set the minimum clock threshold. When we have this situation, the function bw_calcs is invoked with number_of_displays set to zero, making DCE set dispclk_khz and sclk_khz to zero. For these reasons, we have seen some ATOM bios errors that look like:
[drm:atom_op_jump [amdgpu]] *ERROR* atombios stuck in loop for more than 5secs aborting
[drm:amdgpu_atom_execute_table_locked [amdgpu]] *ERROR* atombios stuck executing EA8A (len 761, WS 0, PS 0) @ 0xEABA
[How]
This error happens due to an attempt to optimize the bandwidth using the sclk, and the dispclk clock set to zero. Technically we handle this in the function dce112_set_clock, but we are not considering the case that this value is set to zero. This commit fixes this issue by ensuring that we never set a minimum value below the minimum clock threshold.
--- drivers/gpu/drm/amd/display/dc/clk_mgr/dce112/dce112_clk_mgr.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/gpu/drm/amd/display/dc/clk_mgr/dce112/dce112_clk_mgr.c +1 -2
@@ -79,8 +79,7 @@ int dce112_set_clock(struct clk_mgr *clk_mgr_base, int requested_clk_khz)memset (&dce_clk_params,0 ,sizeof (dce_clk_params));/* Make sure requested clock isn't lower than minimum threshold*/ -if (requested_clk_khz >0 ) - requested_clk_khz =max (requested_clk_khz, + requested_clk_khz =max (requested_clk_khz, clk_mgr_dce->base.dentist_vco_freq_khz /62 ); dce_clk_params.target_clock_frequency = requested_clk_khz;
Signature · 6 lines
2.25.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx