From faee3edfcff750fe5dd760177f52978aefefeb70 Mon Sep 17 00:00:00 2001 From: Alvin Lee Date: Mon, 9 Sep 2024 16:24:05 -0400 Subject: [PATCH 01/16] drm/amd/display: Wait for all pending cleared before full update [Description] Before every full update we must wait for all pending updates to be cleared - this is particularly important for minimal transitions because if we don't wait for pending cleared, it will be as if there was no minimal transition at all. In OTG we must read 3 different status registers for pending cleared, one specifically for OTG updates, one specifically for OPTC updates, and the last for surface related updates. Reviewed-by: Dillon Varone Signed-off-by: Alvin Lee Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../drm/amd/display/dc/core/dc_hw_sequencer.c | 9 +++- .../amd/display/dc/hwss/dcn20/dcn20_hwseq.c | 4 +- .../amd/display/dc/hwss/dcn30/dcn30_hwseq.c | 27 +++++++++++ .../amd/display/dc/hwss/dcn30/dcn30_hwseq.h | 2 + .../amd/display/dc/hwss/dcn30/dcn30_init.c | 3 +- .../amd/display/dc/hwss/dcn301/dcn301_init.c | 1 + .../amd/display/dc/hwss/dcn32/dcn32_init.c | 1 + .../amd/display/dc/hwss/dcn401/dcn401_init.c | 1 + .../drm/amd/display/dc/hwss/hw_sequencer.h | 1 + .../amd/display/dc/inc/hw/timing_generator.h | 4 +- .../amd/display/dc/optc/dcn10/dcn10_optc.h | 9 ++++ .../amd/display/dc/optc/dcn20/dcn20_optc.h | 7 ++- .../amd/display/dc/optc/dcn30/dcn30_optc.c | 45 +++++++++++++++++++ .../amd/display/dc/optc/dcn30/dcn30_optc.h | 13 +++++- .../amd/display/dc/optc/dcn301/dcn301_optc.c | 3 ++ .../amd/display/dc/optc/dcn31/dcn31_optc.h | 9 +++- .../amd/display/dc/optc/dcn314/dcn314_optc.h | 9 +++- .../amd/display/dc/optc/dcn32/dcn32_optc.c | 16 ++----- .../amd/display/dc/optc/dcn32/dcn32_optc.h | 7 ++- .../amd/display/dc/optc/dcn35/dcn35_optc.h | 6 ++- .../amd/display/dc/optc/dcn401/dcn401_optc.c | 4 +- .../amd/display/dc/optc/dcn401/dcn401_optc.h | 6 ++- .../dc/resource/dcn32/dcn32_resource.h | 3 +- .../dc/resource/dcn401/dcn401_resource.h | 5 ++- 24 files changed, 161 insertions(+), 34 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c index 7ee2be8f82c4..2cb9253c9bde 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c @@ -1071,8 +1071,13 @@ void hwss_wait_for_outstanding_hw_updates(struct dc *dc, struct dc_state *dc_con if (!pipe_ctx->stream) continue; - if (pipe_ctx->stream_res.tg->funcs->wait_drr_doublebuffer_pending_clear) - pipe_ctx->stream_res.tg->funcs->wait_drr_doublebuffer_pending_clear(pipe_ctx->stream_res.tg); + /* For full update we must wait for all double buffer updates, not just DRR updates. This + * is particularly important for minimal transitions. Only check for OTG_MASTER pipes, + * as non-OTG Master pipes share the same OTG as + */ + if (resource_is_pipe_type(pipe_ctx, OTG_MASTER) && dc->hwss.wait_for_all_pending_updates) { + dc->hwss.wait_for_all_pending_updates(pipe_ctx); + } hubp = pipe_ctx->plane_res.hubp; if (!hubp) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c index a80c08582932..b383ed8cb4d4 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c @@ -2255,9 +2255,9 @@ void dcn20_post_unlock_program_front_end( struct timing_generator *tg = pipe->stream_res.tg; - if (tg->funcs->get_double_buffer_pending) { + if (tg->funcs->get_optc_double_buffer_pending) { for (j = 0; j < TIMEOUT_FOR_PIPE_ENABLE_US / polling_interval_us - && tg->funcs->get_double_buffer_pending(tg); j++) + && tg->funcs->get_optc_double_buffer_pending(tg); j++) udelay(polling_interval_us); } } diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c index bded33575493..bf9b5daea2af 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c @@ -1185,3 +1185,30 @@ void dcn30_prepare_bandwidth(struct dc *dc, if (!dc->clk_mgr->clks.fw_based_mclk_switching) dc_dmub_srv_p_state_delegate(dc, false, context); } + +void dcn30_wait_for_all_pending_updates(const struct pipe_ctx *pipe_ctx) +{ + struct timing_generator *tg = pipe_ctx->stream_res.tg; + bool pending_updates = false; + unsigned int i; + + if (tg && tg->funcs->is_tg_enabled(tg)) { + // Poll for 100ms maximum + for (i = 0; i < 100000; i++) { + pending_updates = false; + if (tg->funcs->get_optc_double_buffer_pending) + pending_updates |= tg->funcs->get_optc_double_buffer_pending(tg); + + if (tg->funcs->get_otg_double_buffer_pending) + pending_updates |= tg->funcs->get_otg_double_buffer_pending(tg); + + if (tg->funcs->get_pipe_update_pending && pipe_ctx->plane_state) + pending_updates |= tg->funcs->get_pipe_update_pending(tg); + + if (!pending_updates) + break; + + udelay(1); + } + } +} diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.h index 6a153e7ce910..4b90b781c4f2 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.h @@ -96,4 +96,6 @@ void dcn30_set_hubp_blank(const struct dc *dc, void dcn30_prepare_bandwidth(struct dc *dc, struct dc_state *context); +void dcn30_wait_for_all_pending_updates(const struct pipe_ctx *pipe_ctx); + #endif /* __DC_HWSS_DCN30_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_init.c index 2a8dc40d2847..0e8d32e3dbae 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_init.c @@ -108,7 +108,8 @@ static const struct hw_sequencer_funcs dcn30_funcs = { .set_disp_pattern_generator = dcn30_set_disp_pattern_generator, .get_dcc_en_bits = dcn10_get_dcc_en_bits, .update_visual_confirm_color = dcn10_update_visual_confirm_color, - .is_abm_supported = dcn21_is_abm_supported + .is_abm_supported = dcn21_is_abm_supported, + .wait_for_all_pending_updates = dcn30_wait_for_all_pending_updates, }; static const struct hwseq_private_funcs dcn30_private_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn301/dcn301_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn301/dcn301_init.c index 93e49d87a67c..780ce4c064aa 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn301/dcn301_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn301/dcn301_init.c @@ -107,6 +107,7 @@ static const struct hw_sequencer_funcs dcn301_funcs = { .optimize_pwr_state = dcn21_optimize_pwr_state, .exit_optimized_pwr_state = dcn21_exit_optimized_pwr_state, .update_visual_confirm_color = dcn10_update_visual_confirm_color, + .wait_for_all_pending_updates = dcn30_wait_for_all_pending_updates, }; static const struct hwseq_private_funcs dcn301_private_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_init.c index 3422b564ae98..8e0946fd5b7f 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_init.c @@ -121,6 +121,7 @@ static const struct hw_sequencer_funcs dcn32_funcs = { .is_pipe_topology_transition_seamless = dcn32_is_pipe_topology_transition_seamless, .calculate_pix_rate_divider = dcn32_calculate_pix_rate_divider, .program_outstanding_updates = dcn32_program_outstanding_updates, + .wait_for_all_pending_updates = dcn30_wait_for_all_pending_updates, }; static const struct hwseq_private_funcs dcn32_private_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c index a2ca07235c83..73a632b5ff89 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c @@ -100,6 +100,7 @@ static const struct hw_sequencer_funcs dcn401_funcs = { .fams2_update_config = dcn401_fams2_update_config, .fams2_global_control_lock_fast = dcn401_fams2_global_control_lock_fast, .program_outstanding_updates = dcn401_program_outstanding_updates, + .wait_for_all_pending_updates = dcn30_wait_for_all_pending_updates, }; static const struct hwseq_private_funcs dcn401_private_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h index ac9205625623..b8c47e4c51c1 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h @@ -462,6 +462,7 @@ struct hw_sequencer_funcs { void (*program_outstanding_updates)(struct dc *dc, struct dc_state *context); void (*setup_hpo_hw_control)(const struct dce_hwseq *hws, bool enable); + void (*wait_for_all_pending_updates)(const struct pipe_ctx *pipe_ctx); }; void color_space_to_black_color( diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h index 3d4c8bd42b49..4e08e80eafe8 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h @@ -342,7 +342,9 @@ struct timing_generator_funcs { void (*wait_drr_doublebuffer_pending_clear)(struct timing_generator *tg); void (*set_long_vtotal)(struct timing_generator *optc, const struct long_vtotal_params *params); void (*wait_odm_doublebuffer_pending_clear)(struct timing_generator *tg); - bool (*get_double_buffer_pending)(struct timing_generator *tg); + bool (*get_optc_double_buffer_pending)(struct timing_generator *tg); + bool (*get_otg_double_buffer_pending)(struct timing_generator *tg); + bool (*get_pipe_update_pending)(struct timing_generator *tg); }; #endif diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h index b7a57f98553d..40757f20d73f 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h @@ -202,6 +202,7 @@ struct dcn_optc_registers { uint32_t OPTC_CLOCK_CONTROL; uint32_t OPTC_WIDTH_CONTROL2; uint32_t OTG_PSTATE_REGISTER; + uint32_t OTG_PIPE_UPDATE_STATUS; }; #define TG_COMMON_MASK_SH_LIST_DCN(mask_sh)\ @@ -566,6 +567,12 @@ struct dcn_optc_registers { type OTG_H_TIMING_DIV_MODE_DB_UPDATE_PENDING;\ type OPTC_DOUBLE_BUFFER_PENDING;\ +#define TG_REG_FIELD_LIST_DCN2_0(type) \ + type OTG_FLIP_PENDING;\ + type OTG_DC_REG_UPDATE_PENDING;\ + type OTG_CURSOR_UPDATE_PENDING;\ + type OTG_VUPDATE_KEEPOUT_STATUS;\ + #define TG_REG_FIELD_LIST_DCN3_2(type) \ type OTG_H_TIMING_DIV_MODE_MANUAL; @@ -600,6 +607,7 @@ struct dcn_optc_registers { struct dcn_optc_shift { TG_REG_FIELD_LIST(uint8_t) + TG_REG_FIELD_LIST_DCN2_0(uint8_t) TG_REG_FIELD_LIST_DCN3_2(uint8_t) TG_REG_FIELD_LIST_DCN3_5(uint8_t) TG_REG_FIELD_LIST_DCN401(uint8_t) @@ -607,6 +615,7 @@ struct dcn_optc_shift { struct dcn_optc_mask { TG_REG_FIELD_LIST(uint32_t) + TG_REG_FIELD_LIST_DCN2_0(uint32_t) TG_REG_FIELD_LIST_DCN3_2(uint32_t) TG_REG_FIELD_LIST_DCN3_5(uint32_t) TG_REG_FIELD_LIST_DCN401(uint32_t) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.h index 364034b19028..928e110b95fb 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.h @@ -43,7 +43,8 @@ SRI(OPTC_MEMORY_CONFIG, ODM, inst),\ SR(DWB_SOURCE_SELECT),\ SRI(OTG_MANUAL_FLOW_CONTROL, OTG, inst), \ - SRI(OTG_DRR_CONTROL, OTG, inst) + SRI(OTG_DRR_CONTROL, OTG, inst),\ + SRI(OTG_PIPE_UPDATE_STATUS, OTG, inst) #define TG_COMMON_MASK_SH_LIST_DCN2_0(mask_sh)\ TG_COMMON_MASK_SH_LIST_DCN(mask_sh),\ @@ -53,6 +54,10 @@ SF(OTG0_OTG_GLOBAL_CONTROL2, GLOBAL_UPDATE_LOCK_EN, mask_sh),\ SF(OTG0_OTG_GLOBAL_CONTROL2, DIG_UPDATE_LOCATION, mask_sh),\ SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_RANGE_TIMING_DBUF_UPDATE_MODE, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh),\ SF(OTG0_OTG_GSL_WINDOW_X, OTG_GSL_WINDOW_START_X, mask_sh),\ SF(OTG0_OTG_GSL_WINDOW_X, OTG_GSL_WINDOW_END_X, mask_sh), \ SF(OTG0_OTG_GSL_WINDOW_Y, OTG_GSL_WINDOW_START_Y, mask_sh),\ diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c index abcd03d78668..4c95c0958612 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c @@ -271,6 +271,48 @@ void optc3_set_odm_combine(struct timing_generator *optc, int *opp_id, int opp_c optc1->opp_count = opp_cnt; } +/* OTG status register that indicates OPTC update is pending */ +bool optc3_get_optc_double_buffer_pending(struct timing_generator *optc) +{ + struct optc *optc1 = DCN10TG_FROM_TG(optc); + uint32_t update_pending = 0; + + REG_GET(OPTC_INPUT_GLOBAL_CONTROL, + OPTC_DOUBLE_BUFFER_PENDING, + &update_pending); + + return (update_pending == 1); +} + +/* OTG status register that indicates OTG update is pending */ +bool optc3_get_otg_update_pending(struct timing_generator *optc) +{ + struct optc *optc1 = DCN10TG_FROM_TG(optc); + uint32_t update_pending = 0; + + REG_GET(OTG_DOUBLE_BUFFER_CONTROL, + OTG_UPDATE_PENDING, + &update_pending); + + return (update_pending == 1); +} + +/* OTG status register that indicates surface update is pending */ +bool optc3_get_pipe_update_pending(struct timing_generator *optc) +{ + struct optc *optc1 = DCN10TG_FROM_TG(optc); + uint32_t flip_pending = 0; + uint32_t dc_update_pending = 0; + + REG_GET_2(OTG_PIPE_UPDATE_STATUS, + OTG_FLIP_PENDING, + &flip_pending, + OTG_DC_REG_UPDATE_PENDING, + &dc_update_pending); + + return (flip_pending == 1 || dc_update_pending == 1); +} + /** * optc3_set_timing_double_buffer() - DRR double buffering control * @@ -375,6 +417,9 @@ static struct timing_generator_funcs dcn30_tg_funcs = { .get_hw_timing = optc1_get_hw_timing, .wait_drr_doublebuffer_pending_clear = optc3_wait_drr_doublebuffer_pending_clear, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, + .get_optc_double_buffer_pending = optc3_get_optc_double_buffer_pending, + .get_otg_double_buffer_pending = optc3_get_otg_update_pending, + .get_pipe_update_pending = optc3_get_pipe_update_pending, }; void dcn30_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.h index bda974d432ea..e2303f9eaf13 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.h @@ -109,7 +109,8 @@ SRI(OPTC_BYTES_PER_PIXEL, ODM, inst),\ SRI(OPTC_WIDTH_CONTROL, ODM, inst),\ SRI(OPTC_MEMORY_CONFIG, ODM, inst),\ - SR(DWB_SOURCE_SELECT) + SR(DWB_SOURCE_SELECT),\ + SRI(OTG_PIPE_UPDATE_STATUS, OTG, inst) #define DCN30_VTOTAL_REGS_SF(mask_sh) @@ -209,6 +210,7 @@ SF(ODM0_OPTC_INPUT_CLOCK_CONTROL, OPTC_INPUT_CLK_GATE_DIS, mask_sh),\ SF(ODM0_OPTC_INPUT_GLOBAL_CONTROL, OPTC_UNDERFLOW_OCCURRED_STATUS, mask_sh),\ SF(ODM0_OPTC_INPUT_GLOBAL_CONTROL, OPTC_UNDERFLOW_CLEAR, mask_sh),\ + SF(ODM0_OPTC_INPUT_GLOBAL_CONTROL, OPTC_DOUBLE_BUFFER_PENDING, mask_sh),\ SF(VTG0_CONTROL, VTG0_ENABLE, mask_sh),\ SF(VTG0_CONTROL, VTG0_FP2, mask_sh),\ SF(VTG0_CONTROL, VTG0_VCOUNT_INIT, mask_sh),\ @@ -319,7 +321,11 @@ SF(OTG0_OTG_DRR_V_TOTAL_CHANGE, OTG_DRR_V_TOTAL_CHANGE_LIMIT, mask_sh),\ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE, mask_sh),\ SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_PENDING, mask_sh),\ - SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_MODE, mask_sh) + SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_MODE, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh),\ void dcn30_timing_generator_init(struct optc *optc1); @@ -356,4 +362,7 @@ void optc3_set_odm_combine(struct timing_generator *optc, int *opp_id, int opp_c void optc3_wait_drr_doublebuffer_pending_clear(struct timing_generator *optc); void optc3_tg_init(struct timing_generator *optc); void optc3_set_vtotal_min_max(struct timing_generator *optc, int vtotal_min, int vtotal_max); +bool optc3_get_optc_double_buffer_pending(struct timing_generator *optc); +bool optc3_get_otg_update_pending(struct timing_generator *optc); +bool optc3_get_pipe_update_pending(struct timing_generator *optc); #endif /* __DC_OPTC_DCN30_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c index 1a22ae89fb55..d7a45ef2d01b 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c @@ -169,6 +169,9 @@ static struct timing_generator_funcs dcn30_tg_funcs = { .get_hw_timing = optc1_get_hw_timing, .wait_drr_doublebuffer_pending_clear = optc3_wait_drr_doublebuffer_pending_clear, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, + .get_optc_double_buffer_pending = optc3_get_optc_double_buffer_pending, + .get_otg_double_buffer_pending = optc3_get_otg_update_pending, + .get_pipe_update_pending = optc3_get_pipe_update_pending, }; void dcn301_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h index 30b81a448ce2..fbbe86d00c2e 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h @@ -99,7 +99,8 @@ SRI(OPTC_MEMORY_CONFIG, ODM, inst),\ SRI(OTG_CRC_CNTL2, OTG, inst),\ SR(DWB_SOURCE_SELECT),\ - SRI(OTG_DRR_CONTROL, OTG, inst) + SRI(OTG_DRR_CONTROL, OTG, inst),\ + SRI(OTG_PIPE_UPDATE_STATUS, OTG, inst) #define OPTC_COMMON_MASK_SH_LIST_DCN3_1(mask_sh)\ SF(OTG0_OTG_VSTARTUP_PARAM, VSTARTUP_START, mask_sh),\ @@ -254,7 +255,11 @@ SF(OTG0_OTG_CRC_CNTL2, OTG_CRC_DATA_STREAM_COMBINE_MODE, mask_sh),\ SF(OTG0_OTG_CRC_CNTL2, OTG_CRC_DATA_STREAM_SPLIT_MODE, mask_sh),\ SF(OTG0_OTG_CRC_CNTL2, OTG_CRC_DATA_FORMAT, mask_sh),\ - SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh) + SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh),\ void dcn31_timing_generator_init(struct optc *optc1); diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.h index 99c098e76116..0ff72b97b465 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.h @@ -98,7 +98,8 @@ SRI(OPTC_BYTES_PER_PIXEL, ODM, inst),\ SRI(OPTC_WIDTH_CONTROL, ODM, inst),\ SRI(OPTC_MEMORY_CONFIG, ODM, inst),\ - SRI(OTG_DRR_CONTROL, OTG, inst) + SRI(OTG_DRR_CONTROL, OTG, inst),\ + SRI(OTG_PIPE_UPDATE_STATUS, OTG, inst) #define OPTC_COMMON_MASK_SH_LIST_DCN3_14(mask_sh)\ SF(OTG0_OTG_VSTARTUP_PARAM, VSTARTUP_START, mask_sh),\ @@ -248,7 +249,11 @@ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE, mask_sh),\ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE_MANUAL, mask_sh),\ SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_MODE, mask_sh),\ - SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh) + SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh),\ void dcn314_timing_generator_init(struct optc *optc1); diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c index 00094f0e8470..c217f653b3c8 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c @@ -297,18 +297,6 @@ static void optc32_set_drr( optc32_setup_manual_trigger(optc); } -bool optc32_get_double_buffer_pending(struct timing_generator *optc) -{ - struct optc *optc1 = DCN10TG_FROM_TG(optc); - uint32_t update_pending = 0; - - REG_GET(OPTC_INPUT_GLOBAL_CONTROL, - OPTC_DOUBLE_BUFFER_PENDING, - &update_pending); - - return (update_pending == 1); -} - static struct timing_generator_funcs dcn32_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, @@ -373,7 +361,9 @@ static struct timing_generator_funcs dcn32_tg_funcs = { .setup_manual_trigger = optc2_setup_manual_trigger, .get_hw_timing = optc1_get_hw_timing, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, - .get_double_buffer_pending = optc32_get_double_buffer_pending, + .get_optc_double_buffer_pending = optc3_get_optc_double_buffer_pending, + .get_otg_double_buffer_pending = optc3_get_otg_update_pending, + .get_pipe_update_pending = optc3_get_pipe_update_pending, }; void dcn32_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.h index 665d7c52f67c..0b0964a9da74 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.h @@ -177,7 +177,11 @@ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE, mask_sh),\ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE_MANUAL, mask_sh),\ SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_MODE, mask_sh),\ - SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh) + SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh) void dcn32_timing_generator_init(struct optc *optc1); void optc32_set_h_timing_div_manual_mode(struct timing_generator *optc, bool manual_mode); @@ -185,6 +189,5 @@ void optc32_get_odm_combine_segments(struct timing_generator *tg, int *odm_combi void optc32_set_odm_bypass(struct timing_generator *optc, const struct dc_crtc_timing *dc_crtc_timing); void optc32_wait_odm_doublebuffer_pending_clear(struct timing_generator *tg); -bool optc32_get_double_buffer_pending(struct timing_generator *optc); #endif /* __DC_OPTC_DCN32_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.h index d077e2392379..be749ab41dce 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.h @@ -67,7 +67,11 @@ SF(OTG0_OTG_CRC1_WINDOWB_Y_CONTROL_READBACK, OTG_CRC1_WINDOWB_Y_END_READBACK, mask_sh),\ SF(OPTC_CLOCK_CONTROL, OPTC_FGCG_REP_DIS, mask_sh),\ SF(OTG0_OTG_V_COUNT_STOP_CONTROL, OTG_V_COUNT_STOP, mask_sh),\ - SF(OTG0_OTG_V_COUNT_STOP_CONTROL2, OTG_V_COUNT_STOP_TIMER, mask_sh) + SF(OTG0_OTG_V_COUNT_STOP_CONTROL2, OTG_V_COUNT_STOP_TIMER, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh) void dcn35_timing_generator_init(struct optc *optc1); diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c index a5d6a7dca554..db670fc17264 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c @@ -493,7 +493,9 @@ static struct timing_generator_funcs dcn401_tg_funcs = { .setup_manual_trigger = optc2_setup_manual_trigger, .get_hw_timing = optc1_get_hw_timing, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, - .get_double_buffer_pending = optc32_get_double_buffer_pending, + .get_optc_double_buffer_pending = optc3_get_optc_double_buffer_pending, + .get_otg_double_buffer_pending = optc3_get_otg_update_pending, + .get_pipe_update_pending = optc3_get_pipe_update_pending, }; void dcn401_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.h index bb13a645802d..1be89571986f 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.h @@ -159,7 +159,11 @@ SF(OTG0_OTG_PSTATE_REGISTER, OTG_PSTATE_KEEPOUT_START, mask_sh),\ SF(OTG0_OTG_PSTATE_REGISTER, OTG_PSTATE_EXTEND, mask_sh),\ SF(OTG0_OTG_PSTATE_REGISTER, OTG_UNBLANK, mask_sh),\ - SF(OTG0_OTG_PSTATE_REGISTER, OTG_PSTATE_ALLOW_WIDTH_MIN, mask_sh) + SF(OTG0_OTG_PSTATE_REGISTER, OTG_PSTATE_ALLOW_WIDTH_MIN, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh) void dcn401_timing_generator_init(struct optc *optc1); diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h index 7901792afb7b..86c6e5e8c42e 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h @@ -1054,7 +1054,8 @@ unsigned int dcn32_calculate_mall_ways_from_bytes(const struct dc *dc, unsigned SRI_ARR(OPTC_BYTES_PER_PIXEL, ODM, inst), \ SRI_ARR(OPTC_WIDTH_CONTROL, ODM, inst), \ SRI_ARR(OPTC_MEMORY_CONFIG, ODM, inst), \ - SRI_ARR(OTG_DRR_CONTROL, OTG, inst) + SRI_ARR(OTG_DRR_CONTROL, OTG, inst), \ + SRI_ARR(OTG_PIPE_UPDATE_STATUS, OTG, inst) /* HUBP */ diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h index 514d1ce20df9..bdafa7496cea 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h @@ -536,8 +536,9 @@ void dcn401_prepare_mcache_programming(struct dc *dc, struct dc_state *context); SRI_ARR(OPTC_WIDTH_CONTROL, ODM, inst), \ SRI_ARR(OPTC_WIDTH_CONTROL2, ODM, inst), \ SRI_ARR(OPTC_MEMORY_CONFIG, ODM, inst), \ - SRI_ARR(OTG_DRR_CONTROL, OTG, inst), \ - SRI_ARR(OTG_PSTATE_REGISTER, OTG, inst) + SRI_ARR(OTG_DRR_CONTROL, OTG, inst), \ + SRI_ARR(OTG_PSTATE_REGISTER, OTG, inst), \ + SRI_ARR(OTG_PIPE_UPDATE_STATUS, OTG, inst) /* HUBBUB */ #define HUBBUB_REG_LIST_DCN4_01_RI(id) \ -- 2.50.1 From 3fc98044eb33f1ed11cc708253b115e4325cc167 Mon Sep 17 00:00:00 2001 From: Zhongwei Date: Tue, 10 Sep 2024 13:28:34 +0800 Subject: [PATCH 02/16] drm/amd/display: Monitor patch to call blank_stream() before otg off [Why] Turning off OTG before DIG is on and backlight is on, might cause: DIG FIFO underflow. EDP output unexpected video data. That might violate EDP spec. EDP spec requires black light should be off before validate video data is turned off. Currently garbage issue only occurs in one type of EDP to MIPI converter. [How] Add monitor patch to call blank_stream() before disable_crtc(). That will be no impact for current sequence. If there are more types of EDP panels meeting this issue later, we might remove this monitor patch and fix the current sequence. Reviewed-by: Nicholas Kazlauskas Signed-off-by: Zhongwei Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc_types.h | 1 + drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c | 5 +++++ 2 files changed, 6 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index fd6dca735714..2bbafd1cdce4 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -178,6 +178,7 @@ struct dc_panel_patch { unsigned int skip_avmute; unsigned int mst_start_top_delay; unsigned int remove_sink_ext_caps; + uint8_t blankstream_before_otg_off; }; struct dc_edid_caps { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c index 3d4b31bd9946..bfc78a42bc2a 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c @@ -517,6 +517,11 @@ static void dcn31_reset_back_end_for_pipe( dc->hwss.set_abm_immediate_disable(pipe_ctx); + if ((!pipe_ctx->stream->dpms_off || pipe_ctx->stream->link->link_status.link_active) + && pipe_ctx->stream->sink && pipe_ctx->stream->sink->edid_caps.panel_patch.blankstream_before_otg_off) { + dc->hwss.blank_stream(pipe_ctx); + } + pipe_ctx->stream_res.tg->funcs->set_dsc_config( pipe_ctx->stream_res.tg, OPTC_DSC_DISABLED, 0, 0); -- 2.50.1 From 4641169a8c95d9efc35d2d3c55c3948f3b375ff9 Mon Sep 17 00:00:00 2001 From: Fangzhi Zuo Date: Wed, 4 Sep 2024 16:56:45 -0400 Subject: [PATCH 03/16] drm/amd/display: Fix incorrect DSC recompute trigger A stream without dsc_aux should not be eliminated from the dsc determination. Whether it needs a dsc recompute depends on whether its mode has changed or not. Eliminating such a no-dsc stream from the dsc determination policy will end up with inconsistencies in the new dc_state when compared to the current dc_state, triggering a dsc recompute that should not have happened. Reviewed-by: Rodrigo Siqueira Signed-off-by: Fangzhi Zuo Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c index a08e8a0b696c..68bd8e6b948d 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c @@ -1302,7 +1302,7 @@ static bool is_dsc_need_re_compute( continue; aconnector = (struct amdgpu_dm_connector *) stream->dm_stream_context; - if (!aconnector || !aconnector->dsc_aux) + if (!aconnector) continue; stream_on_link[new_stream_on_link_num] = aconnector; -- 2.50.1 From 3924f200f199e604bc3a9f965b6dd7617d2bbe6a Mon Sep 17 00:00:00 2001 From: Sung Lee Date: Mon, 9 Sep 2024 16:12:55 -0400 Subject: [PATCH 04/16] drm/amd/display: Clean up triple buffer enablement code [WHY] Triple buffer enablement currently does not work properly [HOW] Allow triple buffer enablement to happen properly on fast updates Reviewed-by: Aric Cyr Signed-off-by: Sung Lee Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 5c39390ecbd5..9d427a3d1efb 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -3677,13 +3677,14 @@ static void commit_planes_for_stream_fast(struct dc *dc, if (!pipe_ctx->plane_state) continue; - if (should_update_pipe_for_plane(context, pipe_ctx, plane_state)) + if (!should_update_pipe_for_plane(context, pipe_ctx, plane_state)) continue; + pipe_ctx->plane_state->triplebuffer_flips = false; if (update_type == UPDATE_TYPE_FAST && - dc->hwss.program_triplebuffer != NULL && - !pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) { - /*triple buffer for VUpdate only*/ + dc->hwss.program_triplebuffer != NULL && + !pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) { + /*triple buffer for VUpdate only*/ pipe_ctx->plane_state->triplebuffer_flips = true; } } @@ -3920,19 +3921,20 @@ static void commit_planes_for_stream(struct dc *dc, struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j]; if (!pipe_ctx->plane_state) continue; - if (should_update_pipe_for_plane(context, pipe_ctx, plane_state)) + if (!should_update_pipe_for_plane(context, pipe_ctx, plane_state)) continue; pipe_ctx->plane_state->triplebuffer_flips = false; if (update_type == UPDATE_TYPE_FAST && - dc->hwss.program_triplebuffer != NULL && - !pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) { - /*triple buffer for VUpdate only*/ - pipe_ctx->plane_state->triplebuffer_flips = true; + dc->hwss.program_triplebuffer != NULL && + !pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) { + /*triple buffer for VUpdate only*/ + pipe_ctx->plane_state->triplebuffer_flips = true; } } if (update_type == UPDATE_TYPE_FULL) { /* force vsync flip when reconfiguring pipes to prevent underflow */ plane_state->flip_immediate = false; + plane_state->triplebuffer_flips = false; } } @@ -3953,7 +3955,6 @@ static void commit_planes_for_stream(struct dc *dc, continue; ASSERT(!pipe_ctx->plane_state->triplebuffer_flips); - if (dc->hwss.program_triplebuffer != NULL && dc->debug.enable_tri_buf) { /*turn off triple buffer for full update*/ dc->hwss.program_triplebuffer( @@ -4028,7 +4029,7 @@ static void commit_planes_for_stream(struct dc *dc, /*program triple buffer after lock based on flip type*/ if (dc->hwss.program_triplebuffer != NULL && dc->debug.enable_tri_buf) { - /*only enable triplebuffer for fast_update*/ + /*only enable triplebuffer for fast_update*/ dc->hwss.program_triplebuffer( dc, pipe_ctx, pipe_ctx->plane_state->triplebuffer_flips); } -- 2.50.1 From 83762925c94c9053de9aedf0a880907a655ec394 Mon Sep 17 00:00:00 2001 From: Shunlu Zhang Date: Tue, 3 Sep 2024 01:08:01 -0400 Subject: [PATCH 05/16] drm/amd/display: Change dc_surface_update cm2_params to const [WHY] cm2_params is used to set update_flags. It's value is not intended to be modified. [WHAT] Change the declaration of cm2_params to be a constant variable Reviewed-by: Tao Huang Reviewed-by: Ariel Bernstein Signed-off-by: Shunlu Zhang Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 3992ad73165b..7b164e8fa104 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -1513,7 +1513,7 @@ struct dc_surface_update { * change cm2_params.component_settings: Full update * change cm2_params.cm2_luts: Fast update */ - struct dc_cm2_parameters *cm2_params; + const struct dc_cm2_parameters *cm2_params; const struct dc_csc_transform *cursor_csc_color_matrix; unsigned int sdr_white_level_nits; }; -- 2.50.1 From 0e719a40bf5234e66bbd0f1537e30709ab112b5f Mon Sep 17 00:00:00 2001 From: Austin Zheng Date: Tue, 10 Sep 2024 13:46:50 -0400 Subject: [PATCH 06/16] drm/amd/display: Set Pipe Unlock Order Outside of HWSEQ [Why] Current pipe unlock order set within dcn401_interdependent_update_lock. Separate the logic from the functionality to allow for unit testing and make it easier to debug. [How] Add a flag to indicate if a pipe unlock order should be set. Create function to determine unlock order. Indicate which pipes should be unlocked first using array stored in dc scratch memory. Pipes indicated in array can be unlocked in any order. Reviewed-by: Alvin Lee Signed-off-by: Austin Zheng Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 42 +++++++++++++++++++ drivers/gpu/drm/amd/display/dc/dc.h | 2 + .../amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 32 ++++++-------- .../dc/resource/dcn401/dcn401_resource.c | 1 + 4 files changed, 57 insertions(+), 20 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 9d427a3d1efb..f87f12a7be90 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -1876,6 +1876,41 @@ void dc_z10_save_init(struct dc *dc) dc->hwss.z10_save_init(dc); } +/* Set a pipe unlock order based on the change in DET allocation and stores it in dc scratch memory + * Prevents over allocation of DET during unlock process + * e.g. 2 pipe config with different streams with a max of 20 DET segments + * Before: After: + * - Pipe0: 10 DET segments - Pipe0: 12 DET segments + * - Pipe1: 10 DET segments - Pipe1: 8 DET segments + * If Pipe0 gets updated first, 22 DET segments will be allocated + */ +static void determine_pipe_unlock_order(struct dc *dc, struct dc_state *context) +{ + unsigned int i = 0; + struct pipe_ctx *pipe = NULL; + struct timing_generator *tg = NULL; + + if (!dc->config.set_pipe_unlock_order) + return; + + memset(dc->scratch.pipes_to_unlock_first, 0, sizeof(dc->scratch.pipes_to_unlock_first)); + for (i = 0; i < dc->res_pool->pipe_count; i++) { + pipe = &context->res_ctx.pipe_ctx[i]; + tg = pipe->stream_res.tg; + + if (!resource_is_pipe_type(pipe, OTG_MASTER) || + !tg->funcs->is_tg_enabled(tg) || + dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) { + continue; + } + + if (resource_calculate_det_for_stream(context, pipe) < + resource_calculate_det_for_stream(dc->current_state, &dc->current_state->res_ctx.pipe_ctx[i])) { + dc->scratch.pipes_to_unlock_first[i] = true; + } + } +} + /** * dc_commit_state_no_check - Apply context to the hardware * @@ -1974,6 +2009,7 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c context->streams[i]->update_flags.bits.dsc_changed = prev_dsc_changed; } + determine_pipe_unlock_order(dc, context); /* Program all planes within new context*/ if (dc->res_pool->funcs->prepare_mcache_programming) dc->res_pool->funcs->prepare_mcache_programming(dc, context); @@ -3625,6 +3661,10 @@ static void commit_planes_for_stream_fast(struct dc *dc, struct pipe_ctx *top_pipe_to_program = NULL; struct dc_stream_status *stream_status = NULL; bool should_offload_fams2_flip = false; + bool should_lock_all_pipes = (update_type != UPDATE_TYPE_FAST); + + if (should_lock_all_pipes) + determine_pipe_unlock_order(dc, context); if (dc->debug.fams2_config.bits.enable && dc->debug.fams2_config.bits.enable_offload_flip && @@ -3743,6 +3783,8 @@ static void commit_planes_for_stream(struct dc *dc, bool subvp_curr_use = false; uint8_t current_stream_mask = 0; + if (should_lock_all_pipes) + determine_pipe_unlock_order(dc, context); // Once we apply the new subvp context to hardware it won't be in the // dc->current_state anymore, so we have to cache it before we apply // the new SubVP context diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 7b164e8fa104..d3b6a389fece 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -463,6 +463,7 @@ struct dc_config { unsigned int enable_fpo_flicker_detection; bool disable_hbr_audio_dp2; bool consolidated_dpia_dp_lt; + bool set_pipe_unlock_order; }; enum visual_confirm { @@ -1461,6 +1462,7 @@ struct dc { struct dc_scratch_space current_state; struct dc_scratch_space new_state; struct dc_stream_state temp_stream; // Used so we don't need to allocate stream on the stack + bool pipes_to_unlock_first[MAX_PIPES]; /* Any of the pipes indicated here should be unlocked first */ } scratch; struct dml2_configuration_options dml2_options; diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 0b743669f23b..29fcb4ed6646 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -1705,7 +1705,6 @@ void dcn401_interdependent_update_lock(struct dc *dc, unsigned int i = 0; struct pipe_ctx *pipe = NULL; struct timing_generator *tg = NULL; - bool pipe_unlocked[MAX_PIPES] = {0}; if (lock) { for (i = 0; i < dc->res_pool->pipe_count; i++) { @@ -1719,43 +1718,36 @@ void dcn401_interdependent_update_lock(struct dc *dc, dc->hwss.pipe_control_lock(dc, pipe, true); } } else { - /* Unlock pipes based on the change in DET allocation instead of pipe index - * Prevents over allocation of DET during unlock process - * e.g. 2 pipe config with different streams with a max of 20 DET segments - * Before: After: - * - Pipe0: 10 DET segments - Pipe0: 12 DET segments - * - Pipe1: 10 DET segments - Pipe1: 8 DET segments - * If Pipe0 gets updated first, 22 DET segments will be allocated - */ + /* Need to free DET being used first and have pipe update, then unlock the remaining pipes*/ for (i = 0; i < dc->res_pool->pipe_count; i++) { pipe = &context->res_ctx.pipe_ctx[i]; tg = pipe->stream_res.tg; - int current_pipe_idx = i; if (!resource_is_pipe_type(pipe, OTG_MASTER) || !tg->funcs->is_tg_enabled(tg) || dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) { - pipe_unlocked[i] = true; continue; } - // If the same stream exists in old context, ensure the OTG_MASTER pipes for the same stream get compared - struct pipe_ctx *old_otg_master = resource_get_otg_master_for_stream(&dc->current_state->res_ctx, pipe->stream); - - if (old_otg_master) - current_pipe_idx = old_otg_master->pipe_idx; - if (resource_calculate_det_for_stream(context, pipe) < - resource_calculate_det_for_stream(dc->current_state, &dc->current_state->res_ctx.pipe_ctx[current_pipe_idx])) { + if (dc->scratch.pipes_to_unlock_first[i]) { dc->hwss.pipe_control_lock(dc, pipe, false); - pipe_unlocked[i] = true; dcn401_wait_for_det_buffer_update(dc, context, pipe); } } + /* Unlocking the rest of the pipes */ for (i = 0; i < dc->res_pool->pipe_count; i++) { - if (pipe_unlocked[i]) + if (dc->scratch.pipes_to_unlock_first[i]) continue; + pipe = &context->res_ctx.pipe_ctx[i]; + tg = pipe->stream_res.tg; + if (!resource_is_pipe_type(pipe, OTG_MASTER) || + !tg->funcs->is_tg_enabled(tg) || + dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) { + continue; + } + dc->hwss.pipe_control_lock(dc, pipe, false); } } diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c index 9d56fbdcd06a..f2653a86d3e7 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c @@ -1867,6 +1867,7 @@ static bool dcn401_resource_construct( dc->config.prefer_easf = true; dc->config.dc_mode_clk_limit_support = true; dc->config.enable_windowed_mpo_odm = true; + dc->config.set_pipe_unlock_order = true; /* Need to ensure DET gets freed before allocating */ /* read VBIOS LTTPR caps */ { if (ctx->dc_bios->funcs->get_lttpr_caps) { -- 2.50.1 From 4f01a68751194d05280d659a65758c09e4af04d6 Mon Sep 17 00:00:00 2001 From: Michael Strauss Date: Tue, 3 Sep 2024 16:23:11 -0400 Subject: [PATCH 07/16] drm/amd/display: Block UHBR Based On USB-C PD Cable ID [WHY] Currently the absence of UHBR cable ID caps from USB-C PD does not block UHBR rates. In situations where DPCD reports valid UHBR capability but USB-C PD does not, such as using a USB-C to DP1.4 dongle connected to a native DP2.1 cable, link loss and lightup failures can be seen as a result. Additionally, in edge cases where a platform supports cable ID but DMUB doesn't correctly return cable ID caps, driver currently also allows UHBR. [HOW] Block UHBR rates over DP alt mode if cable id indicates no UHBR support. Additionally, block UHBR rates if a cable ID supported platform receives no reply from a DMUB cable id query. Reviewed-by: Wenjing Liu Reviewed-by: Ovidiu Bunea Signed-off-by: Michael Strauss Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../dc/link/protocols/link_dp_capability.c | 22 ++++++++++++++----- 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c index d78c8ec4de79..fe4aa2c158ea 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c @@ -1409,7 +1409,8 @@ static bool get_usbc_cable_id(struct dc_link *link, union dp_cable_id *cable_id) if (!link->ctx->dmub_srv || link->ep_type != DISPLAY_ENDPOINT_PHY || - link->link_enc->features.flags.bits.DP_IS_USB_C == 0) + link->link_enc->features.flags.bits.DP_IS_USB_C == 0 || + link->link_enc->features.flags.bits.IS_DP2_CAPABLE == 0) return false; memset(&cmd, 0, sizeof(cmd)); @@ -1422,7 +1423,9 @@ static bool get_usbc_cable_id(struct dc_link *link, union dp_cable_id *cable_id) cable_id->raw = cmd.cable_id.data.output_raw; DC_LOG_DC("usbc_cable_id = %d.\n", cable_id->raw); } - return cmd.cable_id.header.ret_status == 1; + + ASSERT(cmd.cable_id.header.ret_status); + return true; } static void retrieve_cable_id(struct dc_link *link) @@ -2103,6 +2106,8 @@ struct dc_link_settings dp_get_max_link_cap(struct dc_link *link) /* get max link encoder capability */ if (link_enc) link_enc->funcs->get_max_link_cap(link_enc, &max_link_cap); + else + return max_link_cap; /* Lower link settings based on sink's link cap */ if (link->reported_link_cap.lane_count < max_link_cap.lane_count) @@ -2136,10 +2141,15 @@ struct dc_link_settings dp_get_max_link_cap(struct dc_link *link) */ cable_max_link_rate = get_cable_max_link_rate(link); - if (!link->dc->debug.ignore_cable_id && - cable_max_link_rate != LINK_RATE_UNKNOWN) { - if (cable_max_link_rate < max_link_cap.link_rate) - max_link_cap.link_rate = cable_max_link_rate; + if (!link->dc->debug.ignore_cable_id) { + if (cable_max_link_rate != LINK_RATE_UNKNOWN) + // cable max link rate known + max_link_cap.link_rate = MIN(max_link_cap.link_rate, cable_max_link_rate); + else if (link_enc->funcs->is_in_alt_mode && link_enc->funcs->is_in_alt_mode(link_enc)) + // cable max link rate ambiguous, DP alt mode, limit to HBR3 + max_link_cap.link_rate = MIN(max_link_cap.link_rate, LINK_RATE_HIGH3); + //else {} + // cable max link rate ambiguous, DP, do nothing if (!link->dpcd_caps.cable_id.bits.UHBR13_5_CAPABILITY && link->dpcd_caps.cable_id.bits.CABLE_TYPE >= 2) -- 2.50.1 From 1c6b16ebf5eb2bc5740be9e37b3a69f1dfe1dded Mon Sep 17 00:00:00 2001 From: Charlene Liu Date: Wed, 11 Sep 2024 19:45:09 -0400 Subject: [PATCH 08/16] drm/amd/display: avoid set dispclk to 0 [why] set dispclk to 0 cause stability issue. Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Nicholas Kazlauskas Signed-off-by: Charlene Liu Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c index da9101b83e8c..70abd32ce2ad 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c @@ -766,6 +766,7 @@ static const struct dc_debug_options debug_defaults_drv = { .disable_dmub_reallow_idle = false, .static_screen_wait_frames = 2, .notify_dpia_hr_bw = true, + .min_disp_clk_khz = 50000, }; static const struct dc_panel_config panel_config_defaults = { -- 2.50.1 From a6f59c0445540f07bef226b33c999f2e63c6dfa2 Mon Sep 17 00:00:00 2001 From: Charlene Liu Date: Wed, 11 Sep 2024 20:35:39 -0400 Subject: [PATCH 09/16] drm/amd/display: correct register Clock Gater incorrectly disabled [why] The "dpp35_dppclk_control" routine is incorrectly disabling the register clock gater when the DPP is enabled. The "DISPCLK_R_GATE_DISABLE" should never be set to 1 in the normal operating mode. This will disable the clock gater and the DPPCLK register clock branch will always be running. As a consequence, the dynamic power will be higher than expected. Reviewed-by: Alvin Lee Signed-off-by: Charlene Liu Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c index 8473c694bfdc..9f885a03eec6 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c @@ -50,13 +50,11 @@ void dpp35_dppclk_control( DPPCLK_RATE_CONTROL, dppclk_div, DPP_CLOCK_ENABLE, 1); else - REG_UPDATE_2(DPP_CONTROL, - DPP_CLOCK_ENABLE, 1, - DISPCLK_R_GATE_DISABLE, 1); + REG_UPDATE(DPP_CONTROL, + DPP_CLOCK_ENABLE, 1); } else - REG_UPDATE_2(DPP_CONTROL, - DPP_CLOCK_ENABLE, 0, - DISPCLK_R_GATE_DISABLE, 0); + REG_UPDATE(DPP_CONTROL, + DPP_CLOCK_ENABLE, 0); } void dpp35_program_bias_and_scale_fcnv( -- 2.50.1 From 83e0a4a946050fa4a7e8aa7919ee097cae787e6d Mon Sep 17 00:00:00 2001 From: Samson Tam Date: Tue, 10 Sep 2024 01:02:16 -0400 Subject: [PATCH 10/16] drm/amd/display: fix static analysis warnings [Why & How] Fix static analysis warnings in SPL library Reviewed-by: Alvin Lee Signed-off-by: Samson Tam Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../gpu/drm/amd/display/dc/spl/dc_spl_types.h | 4 +-- .../gpu/drm/amd/display/dc/spl/spl_debug.h | 33 +++++++++++-------- .../drm/amd/display/dc/spl/spl_fixpt31_32.c | 32 +++++++++--------- .../drm/amd/display/dc/spl/spl_fixpt31_32.h | 17 ++++------ .../gpu/drm/amd/display/dc/spl/spl_os_types.h | 3 +- 5 files changed, 45 insertions(+), 44 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h b/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h index 2a74ff5fdfdb..48e4217555f8 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h +++ b/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h @@ -5,10 +5,8 @@ #ifndef __DC_SPL_TYPES_H__ #define __DC_SPL_TYPES_H__ +#include "spl_debug.h" #include "spl_os_types.h" // swap -#ifndef SPL_ASSERT -#define SPL_ASSERT(_bool) ((void *)0) -#endif #include "spl_fixpt31_32.h" // fixed31_32 and related functions #include "spl_custom_float.h" // custom float and related functions diff --git a/drivers/gpu/drm/amd/display/dc/spl/spl_debug.h b/drivers/gpu/drm/amd/display/dc/spl/spl_debug.h index 5696dafd0894..a6f6132df241 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/spl_debug.h +++ b/drivers/gpu/drm/amd/display/dc/spl/spl_debug.h @@ -5,21 +5,26 @@ #ifndef SPL_DEBUG_H #define SPL_DEBUG_H -#ifdef SPL_ASSERT -#undef SPL_ASSERT -#endif -#define SPL_ASSERT(b) +#if defined(CONFIG_HAVE_KGDB) || defined(CONFIG_KGDB) +#define SPL_ASSERT_CRITICAL(expr) do { \ + if (WARN_ON(!(expr))) { \ + kgdb_breakpoint(); \ + } \ +} while (0) +#else +#define SPL_ASSERT_CRITICAL(expr) do { \ + if (WARN_ON(!(expr))) { \ + ; \ + } \ +} while (0) +#endif /* CONFIG_HAVE_KGDB || CONFIG_KGDB */ -#define SPL_ASSERT_CRITICAL(expr) do {if (expr)/* Do nothing */; } while (0) +#if defined(CONFIG_DEBUG_KERNEL_DC) +#define SPL_ASSERT(expr) SPL_ASSERT_CRITICAL(expr) +#else +#define SPL_ASSERT(expr) WARN_ON(!(expr)) +#endif /* CONFIG_DEBUG_KERNEL_DC */ -#ifdef SPL_DALMSG -#undef SPL_DALMSG -#endif -#define SPL_DALMSG(b) - -#ifdef SPL_DAL_ASSERT_MSG -#undef SPL_DAL_ASSERT_MSG -#endif -#define SPL_DAL_ASSERT_MSG(b, m) +#define SPL_BREAK_TO_DEBUGGER() SPL_ASSERT(0) #endif // SPL_DEBUG_H diff --git a/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.c b/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.c index a95565df5487..5fd79d9c67e2 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.c +++ b/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.c @@ -29,7 +29,7 @@ static inline unsigned long long complete_integer_division_u64( { unsigned long long result; - ASSERT(divisor); + SPL_ASSERT(divisor); result = spl_div64_u64_rem(dividend, divisor, remainder); @@ -63,7 +63,7 @@ struct spl_fixed31_32 spl_fixpt_from_fraction(long long numerator, long long den unsigned long long res_value = complete_integer_division_u64( arg1_value, arg2_value, &remainder); - ASSERT(res_value <= LONG_MAX); + SPL_ASSERT(res_value <= (unsigned long long)LONG_MAX); /* determine fractional part */ { @@ -85,7 +85,7 @@ struct spl_fixed31_32 spl_fixpt_from_fraction(long long numerator, long long den { unsigned long long summand = (remainder << 1) >= arg2_value; - ASSERT(res_value <= LLONG_MAX - summand); + SPL_ASSERT(res_value <= (unsigned long long)LLONG_MAX - summand); res_value += summand; } @@ -118,19 +118,19 @@ struct spl_fixed31_32 spl_fixpt_mul(struct spl_fixed31_32 arg1, struct spl_fixed res.value = arg1_int * arg2_int; - ASSERT(res.value <= (long long)LONG_MAX); + SPL_ASSERT(res.value <= (long long)LONG_MAX); res.value <<= FIXED31_32_BITS_PER_FRACTIONAL_PART; tmp = arg1_int * arg2_fra; - ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); + SPL_ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); res.value += tmp; tmp = arg2_int * arg1_fra; - ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); + SPL_ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); res.value += tmp; @@ -139,7 +139,7 @@ struct spl_fixed31_32 spl_fixpt_mul(struct spl_fixed31_32 arg1, struct spl_fixed tmp = (tmp >> FIXED31_32_BITS_PER_FRACTIONAL_PART) + (tmp >= (unsigned long long)spl_fixpt_half.value); - ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); + SPL_ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); res.value += tmp; @@ -163,17 +163,17 @@ struct spl_fixed31_32 spl_fixpt_sqr(struct spl_fixed31_32 arg) res.value = arg_int * arg_int; - ASSERT(res.value <= (long long)LONG_MAX); + SPL_ASSERT(res.value <= (long long)LONG_MAX); res.value <<= FIXED31_32_BITS_PER_FRACTIONAL_PART; tmp = arg_int * arg_fra; - ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); + SPL_ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); res.value += tmp; - ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); + SPL_ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); res.value += tmp; @@ -182,7 +182,7 @@ struct spl_fixed31_32 spl_fixpt_sqr(struct spl_fixed31_32 arg) tmp = (tmp >> FIXED31_32_BITS_PER_FRACTIONAL_PART) + (tmp >= (unsigned long long)spl_fixpt_half.value); - ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); + SPL_ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); res.value += tmp; @@ -196,7 +196,7 @@ struct spl_fixed31_32 spl_fixpt_recip(struct spl_fixed31_32 arg) * Good idea to use Newton's method */ - ASSERT(arg.value); + SPL_ASSERT(arg.value); return spl_fixpt_from_fraction( spl_fixpt_one.value, @@ -295,7 +295,7 @@ static struct spl_fixed31_32 fixed31_32_exp_from_taylor_series(struct spl_fixed3 n + 1); /* TODO find correct res */ - ASSERT(spl_fixpt_lt(arg, spl_fixpt_one)); + SPL_ASSERT(spl_fixpt_lt(arg, spl_fixpt_one)); do res = spl_fixpt_add( @@ -337,9 +337,9 @@ struct spl_fixed31_32 spl_fixpt_exp(struct spl_fixed31_32 arg) spl_fixpt_ln2, m)); - ASSERT(m != 0); + SPL_ASSERT(m != 0); - ASSERT(spl_fixpt_lt( + SPL_ASSERT(spl_fixpt_lt( spl_fixpt_abs(r), spl_fixpt_one)); @@ -364,7 +364,7 @@ struct spl_fixed31_32 spl_fixpt_log(struct spl_fixed31_32 arg) struct spl_fixed31_32 error; - ASSERT(arg.value > 0); + SPL_ASSERT(arg.value > 0); /* TODO if arg is negative, return NaN */ /* TODO if arg is zero, return -INF */ diff --git a/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.h b/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.h index 8a045e2f8699..ed2647f9a099 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.h +++ b/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.h @@ -5,11 +5,8 @@ #ifndef __SPL_FIXED31_32_H__ #define __SPL_FIXED31_32_H__ -#include "os_types.h" +#include "spl_debug.h" #include "spl_os_types.h" // swap -#ifndef ASSERT -#define ASSERT(_bool) ((void *)0) -#endif #ifndef LLONG_MAX #define LLONG_MAX 9223372036854775807ll @@ -194,7 +191,7 @@ static inline struct spl_fixed31_32 spl_fixpt_clamp( */ static inline struct spl_fixed31_32 spl_fixpt_shl(struct spl_fixed31_32 arg, unsigned char shift) { - ASSERT(((arg.value >= 0) && (arg.value <= LLONG_MAX >> shift)) || + SPL_ASSERT(((arg.value >= 0) && (arg.value <= LLONG_MAX >> shift)) || ((arg.value < 0) && (arg.value >= ~(LLONG_MAX >> shift)))); arg.value = arg.value << shift; @@ -231,7 +228,7 @@ static inline struct spl_fixed31_32 spl_fixpt_add(struct spl_fixed31_32 arg1, st { struct spl_fixed31_32 res; - ASSERT(((arg1.value >= 0) && (LLONG_MAX - arg1.value >= arg2.value)) || + SPL_ASSERT(((arg1.value >= 0) && (LLONG_MAX - arg1.value >= arg2.value)) || ((arg1.value < 0) && (LLONG_MIN - arg1.value <= arg2.value))); res.value = arg1.value + arg2.value; @@ -256,7 +253,7 @@ static inline struct spl_fixed31_32 spl_fixpt_sub(struct spl_fixed31_32 arg1, st { struct spl_fixed31_32 res; - ASSERT(((arg2.value >= 0) && (LLONG_MIN + arg2.value <= arg1.value)) || + SPL_ASSERT(((arg2.value >= 0) && (LLONG_MIN + arg2.value <= arg1.value)) || ((arg2.value < 0) && (LLONG_MAX + arg2.value >= arg1.value))); res.value = arg1.value - arg2.value; @@ -448,7 +445,7 @@ static inline int spl_fixpt_round(struct spl_fixed31_32 arg) const long long summand = spl_fixpt_half.value; - ASSERT(LLONG_MAX - (long long)arg_value >= summand); + SPL_ASSERT(LLONG_MAX - (long long)arg_value >= summand); arg_value += summand; @@ -469,7 +466,7 @@ static inline int spl_fixpt_ceil(struct spl_fixed31_32 arg) const long long summand = spl_fixpt_one.value - spl_fixpt_epsilon.value; - ASSERT(LLONG_MAX - (long long)arg_value >= summand); + SPL_ASSERT(LLONG_MAX - (long long)arg_value >= summand); arg_value += summand; @@ -504,7 +501,7 @@ static inline struct spl_fixed31_32 spl_fixpt_truncate(struct spl_fixed31_32 arg bool negative = arg.value < 0; if (frac_bits >= FIXED31_32_BITS_PER_FRACTIONAL_PART) { - ASSERT(frac_bits == FIXED31_32_BITS_PER_FRACTIONAL_PART); + SPL_ASSERT(frac_bits == FIXED31_32_BITS_PER_FRACTIONAL_PART); return arg; } diff --git a/drivers/gpu/drm/amd/display/dc/spl/spl_os_types.h b/drivers/gpu/drm/amd/display/dc/spl/spl_os_types.h index 709706ed4f2c..2e6ba71960ac 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/spl_os_types.h +++ b/drivers/gpu/drm/amd/display/dc/spl/spl_os_types.h @@ -6,6 +6,8 @@ #ifndef _SPL_OS_TYPES_H_ #define _SPL_OS_TYPES_H_ +#include "spl_debug.h" + #include #include #include @@ -18,7 +20,6 @@ * general debug capabilities * */ -#define SPL_BREAK_TO_DEBUGGER() ASSERT(0) static inline uint64_t spl_div_u64_rem(uint64_t dividend, uint32_t divisor, uint32_t *remainder) { -- 2.50.1 From b8d046985c2dc41a0e264a391da4606099f8d44f Mon Sep 17 00:00:00 2001 From: Austin Zheng Date: Tue, 10 Sep 2024 16:41:20 -0400 Subject: [PATCH 11/16] drm/amd/display: Update Interface to Check UCLK DPM [Why] Videos using YUV420 format may result in high power being used. Disabling MPO may result in lower power usage. Update interface that can be used to check power profile of a dc_state. [How] Allow pstate switching in VBlank as last entry in strategy candidates. Add helper functions that can be used to determine power level: -get power profile after a dc_state has undergone full validation Reviewed-by: Aric Cyr Signed-off-by: Austin Zheng Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 9 ++++++++- .../dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c | 1 + drivers/gpu/drm/amd/display/dc/inc/core_types.h | 4 ++++ .../dc/resource/dcn315/dcn315_resource.c | 6 ++++++ .../dc/resource/dcn401/dcn401_resource.c | 17 +++++++++++++++++ 5 files changed, 36 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index f87f12a7be90..c271fbaaf896 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -6018,8 +6018,15 @@ void dc_set_edp_power(const struct dc *dc, struct dc_link *edp_link, struct dc_power_profile dc_get_power_profile_for_dc_state(const struct dc_state *context) { struct dc_power_profile profile = { 0 }; + struct dc *dc = NULL; - profile.power_level += !context->bw_ctx.bw.dcn.clk.p_state_change_support; + if (!context || !context->clk_mgr || !context->clk_mgr->ctx || !context->clk_mgr->ctx->dc) + return profile; + + dc = context->clk_mgr->ctx->dc; + + if (dc->res_pool->funcs->get_power_profile) + profile.power_level = dc->res_pool->funcs->get_power_profile(context); return profile; } diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c index 1cf9015e854a..2e9c59e9e0c1 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c @@ -1798,6 +1798,7 @@ bool pmo_dcn4_fams2_init_for_pstate_support(struct dml2_pmo_init_for_pstate_supp } if (s->pmo_dcn4.num_pstate_candidates > 0) { + s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.num_pstate_candidates-1].allow_state_increase = true; s->pmo_dcn4.cur_pstate_candidate = -1; return true; } else { diff --git a/drivers/gpu/drm/amd/display/dc/inc/core_types.h b/drivers/gpu/drm/amd/display/dc/inc/core_types.h index bfb8b8502d20..8597e866bfe6 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/core_types.h +++ b/drivers/gpu/drm/amd/display/dc/inc/core_types.h @@ -215,6 +215,10 @@ struct resource_funcs { void (*get_panel_config_defaults)(struct dc_panel_config *panel_config); void (*build_pipe_pix_clk_params)(struct pipe_ctx *pipe_ctx); + /* + * Get indicator of power from a context that went through full validation + */ + int (*get_power_profile)(const struct dc_state *context); }; struct audio_support{ diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c index 3f4b9dba4112..f6b840f046a5 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c @@ -1812,6 +1812,11 @@ static void dcn315_get_panel_config_defaults(struct dc_panel_config *panel_confi *panel_config = panel_config_defaults; } +static int dcn315_get_power_profile(const struct dc_state *context) +{ + return !context->bw_ctx.bw.dcn.clk.p_state_change_support; +} + static struct dc_cap_funcs cap_funcs = { .get_dcc_compression_cap = dcn20_get_dcc_compression_cap }; @@ -1840,6 +1845,7 @@ static struct resource_funcs dcn315_res_pool_funcs = { .update_bw_bounding_box = dcn315_update_bw_bounding_box, .patch_unknown_plane_state = dcn20_patch_unknown_plane_state, .get_panel_config_defaults = dcn315_get_panel_config_defaults, + .get_power_profile = dcn315_get_power_profile, }; static bool dcn315_resource_construct( diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c index f2653a86d3e7..59184abab1a7 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c @@ -1688,6 +1688,22 @@ static void dcn401_build_pipe_pix_clk_params(struct pipe_ctx *pipe_ctx) } } +static int dcn401_get_power_profile(const struct dc_state *context) +{ + int uclk_mhz = context->bw_ctx.bw.dcn.clk.dramclk_khz / 1000; + int dpm_level = 0; + + for (int i = 0; i < context->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_memclk_levels; i++) { + if (context->clk_mgr->bw_params->clk_table.entries[i].memclk_mhz == 0 || + uclk_mhz < context->clk_mgr->bw_params->clk_table.entries[i].memclk_mhz) + break; + if (uclk_mhz > context->clk_mgr->bw_params->clk_table.entries[i].memclk_mhz) + dpm_level++; + } + + return dpm_level; +} + static struct resource_funcs dcn401_res_pool_funcs = { .destroy = dcn401_destroy_resource_pool, .link_enc_create = dcn401_link_encoder_create, @@ -1714,6 +1730,7 @@ static struct resource_funcs dcn401_res_pool_funcs = { .prepare_mcache_programming = dcn401_prepare_mcache_programming, .build_pipe_pix_clk_params = dcn401_build_pipe_pix_clk_params, .calculate_mall_ways_from_bytes = dcn32_calculate_mall_ways_from_bytes, + .get_power_profile = dcn401_get_power_profile, }; static uint32_t read_pipe_fuses(struct dc_context *ctx) -- 2.50.1 From e4dc767f1f38fd350775ea182cab2193292f1d98 Mon Sep 17 00:00:00 2001 From: Ilya Bakoulin Date: Thu, 12 Sep 2024 14:31:13 -0400 Subject: [PATCH 12/16] drm/amd/display: Fix cursor visual confirm update [Why/How] Certain transitions from HW to SW cursor can be missed, which will result in black visual confirm instead of expected blue/red. Need to add handling for cursor visual confirm to dc_update_visual_confirm_color. Reviewed-by: Alvin Lee Signed-off-by: Ilya Bakoulin Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 2 ++ .../drm/amd/display/dc/core/dc_hw_sequencer.c | 17 +++++++++++++++++ drivers/gpu/drm/amd/display/dc/core/dc_stream.c | 11 +---------- .../gpu/drm/amd/display/dc/hwss/hw_sequencer.h | 4 ++++ 4 files changed, 24 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index c271fbaaf896..4c33e3b14c4f 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -1157,6 +1157,8 @@ static void dc_update_visual_confirm_color(struct dc *dc, struct dc_state *conte get_surface_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color)); else if (dc->debug.visual_confirm == VISUAL_CONFIRM_SWIZZLE) get_surface_tile_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color)); + else if (dc->debug.visual_confirm == VISUAL_CONFIRM_HW_CURSOR) + get_cursor_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color)); else { if (dc->ctx->dce_version < DCN_VERSION_2_0) color_space_to_black_color( diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c index 2cb9253c9bde..2fdcf8d59b9f 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c @@ -497,6 +497,23 @@ void get_mclk_switch_visual_confirm_color( } } +void get_cursor_visual_confirm_color( + struct pipe_ctx *pipe_ctx, + struct tg_color *color) +{ + uint32_t color_value = MAX_TG_COLOR_VALUE; + + if (pipe_ctx->stream && pipe_ctx->stream->cursor_position.enable) { + color->color_r_cr = color_value; + color->color_g_y = 0; + color->color_b_cb = 0; + } else { + color->color_r_cr = 0; + color->color_g_y = 0; + color->color_b_cb = color_value; + } +} + void set_p_state_switch_method( struct dc *dc, struct dc_state *context, diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_stream.c b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c index 9a406d74c0dd..5d233c09d239 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_stream.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c @@ -421,7 +421,6 @@ bool dc_stream_program_cursor_position( /* apply/update visual confirm */ if (dc->debug.visual_confirm == VISUAL_CONFIRM_HW_CURSOR) { /* update software state */ - uint32_t color_value = MAX_TG_COLOR_VALUE; int i; for (i = 0; i < dc->res_pool->pipe_count; i++) { @@ -429,15 +428,7 @@ bool dc_stream_program_cursor_position( /* adjust visual confirm color for all pipes with current stream */ if (stream == pipe_ctx->stream) { - if (stream->cursor_position.enable) { - pipe_ctx->visual_confirm_color.color_r_cr = color_value; - pipe_ctx->visual_confirm_color.color_g_y = 0; - pipe_ctx->visual_confirm_color.color_b_cb = 0; - } else { - pipe_ctx->visual_confirm_color.color_r_cr = 0; - pipe_ctx->visual_confirm_color.color_g_y = 0; - pipe_ctx->visual_confirm_color.color_b_cb = color_value; - } + get_cursor_visual_confirm_color(pipe_ctx, &(pipe_ctx->visual_confirm_color)); /* programming hardware */ if (pipe_ctx->plane_state) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h index b8c47e4c51c1..2f56c36e4251 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h @@ -505,6 +505,10 @@ void get_mclk_switch_visual_confirm_color( struct pipe_ctx *pipe_ctx, struct tg_color *color); +void get_cursor_visual_confirm_color( + struct pipe_ctx *pipe_ctx, + struct tg_color *color); + void set_p_state_switch_method( struct dc *dc, struct dc_state *context, -- 2.50.1 From ce5057885ff70d0b61f5c332639e685b7ea720e8 Mon Sep 17 00:00:00 2001 From: Joshua Aberback Date: Thu, 12 Sep 2024 18:47:22 -0400 Subject: [PATCH 13/16] drm/amd/display: Clip rect size changes should be full updates [Why] In cases where an MPO plane is being dragged around partially off-screen, it is possible to get a flip where the only scaling parameters to change are the clip rect size and position. Currently, clip rect size changes are considered medium updates, which can result in the clip rect being used for HW programming being larger than the clip rect that was used for the last DML validation. This can lead to mismatches in different parts of the pipe and can result in a p-state hang. [How] - consider clip rect size changes scaling changes, therefore full updates - refactor get_scaling_info_update_type for clarity - remove clip_size_change update flag Clip rect size changes were previously demoted from full updates as an optimization when the MPO + ODM policy changed to always pre-allocate MPO pipes, but it created the issue described above. Personally testing this use case, the performance feels fine with full update spam, and we expect this is a fairly infrequent use case. If the performance needs to be optimized in the future, consider reworking the entire update type logic to run a DML pass and determine the update type based on what DML says will actually change. Reviewed-by: Dillon Varone Signed-off-by: Joshua Aberback Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 45 ++++++++----------- drivers/gpu/drm/amd/display/dc/dc.h | 1 - .../amd/display/dc/hwss/dcn20/dcn20_hwseq.c | 2 - 3 files changed, 19 insertions(+), 29 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 4c33e3b14c4f..9c8e98682c4e 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -2515,41 +2515,35 @@ static enum surface_update_type get_scaling_info_update_type( if (!u->scaling_info) return UPDATE_TYPE_FAST; - if (u->scaling_info->dst_rect.width != u->surface->dst_rect.width + if (u->scaling_info->src_rect.width != u->surface->src_rect.width + || u->scaling_info->src_rect.height != u->surface->src_rect.height + || u->scaling_info->dst_rect.width != u->surface->dst_rect.width || u->scaling_info->dst_rect.height != u->surface->dst_rect.height + || u->scaling_info->clip_rect.width != u->surface->clip_rect.width + || u->scaling_info->clip_rect.height != u->surface->clip_rect.height || u->scaling_info->scaling_quality.integer_scaling != - u->surface->scaling_quality.integer_scaling - ) { + u->surface->scaling_quality.integer_scaling) { update_flags->bits.scaling_change = 1; + if (u->scaling_info->src_rect.width > u->surface->src_rect.width + || u->scaling_info->src_rect.height > u->surface->src_rect.height) + /* Making src rect bigger requires a bandwidth change */ + update_flags->bits.clock_change = 1; + if ((u->scaling_info->dst_rect.width < u->surface->dst_rect.width || u->scaling_info->dst_rect.height < u->surface->dst_rect.height) && (u->scaling_info->dst_rect.width < u->surface->src_rect.width || u->scaling_info->dst_rect.height < u->surface->src_rect.height)) /* Making dst rect smaller requires a bandwidth change */ update_flags->bits.bandwidth_change = 1; - } - - if (u->scaling_info->src_rect.width != u->surface->src_rect.width - || u->scaling_info->src_rect.height != u->surface->src_rect.height) { - update_flags->bits.scaling_change = 1; - if (u->scaling_info->src_rect.width > u->surface->src_rect.width - || u->scaling_info->src_rect.height > u->surface->src_rect.height) - /* Making src rect bigger requires a bandwidth change */ - update_flags->bits.clock_change = 1; + if (u->scaling_info->src_rect.width > dc->caps.max_optimizable_video_width && + (u->scaling_info->clip_rect.width > u->surface->clip_rect.width || + u->scaling_info->clip_rect.height > u->surface->clip_rect.height)) + /* Changing clip size of a large surface may result in MPC slice count change */ + update_flags->bits.bandwidth_change = 1; } - if (u->scaling_info->src_rect.width > dc->caps.max_optimizable_video_width && - (u->scaling_info->clip_rect.width > u->surface->clip_rect.width || - u->scaling_info->clip_rect.height > u->surface->clip_rect.height)) - /* Changing clip size of a large surface may result in MPC slice count change */ - update_flags->bits.bandwidth_change = 1; - - if (u->scaling_info->clip_rect.width != u->surface->clip_rect.width || - u->scaling_info->clip_rect.height != u->surface->clip_rect.height) - update_flags->bits.clip_size_change = 1; - if (u->scaling_info->src_rect.x != u->surface->src_rect.x || u->scaling_info->src_rect.y != u->surface->src_rect.y || u->scaling_info->clip_rect.x != u->surface->clip_rect.x @@ -2558,13 +2552,13 @@ static enum surface_update_type get_scaling_info_update_type( || u->scaling_info->dst_rect.y != u->surface->dst_rect.y) update_flags->bits.position_change = 1; + /* process every update flag before returning */ if (update_flags->bits.clock_change || update_flags->bits.bandwidth_change || update_flags->bits.scaling_change) return UPDATE_TYPE_FULL; - if (update_flags->bits.position_change || - update_flags->bits.clip_size_change) + if (update_flags->bits.position_change) return UPDATE_TYPE_MED; return UPDATE_TYPE_FAST; @@ -3263,8 +3257,7 @@ static bool update_planes_and_stream_state(struct dc *dc, if (update_type != UPDATE_TYPE_MED) continue; - if (surface->update_flags.bits.clip_size_change || - surface->update_flags.bits.position_change) { + if (surface->update_flags.bits.position_change) { for (j = 0; j < dc->res_pool->pipe_count; j++) { struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j]; diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index d3b6a389fece..6d60f7597f88 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -1254,7 +1254,6 @@ union surface_update_flags { uint32_t rotation_change:1; uint32_t swizzle_change:1; uint32_t scaling_change:1; - uint32_t clip_size_change: 1; uint32_t position_change:1; uint32_t in_transfer_func_change:1; uint32_t input_csc_change:1; diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c index b383ed8cb4d4..e89499536c46 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c @@ -1732,7 +1732,6 @@ static void dcn20_update_dchubp_dpp( if (pipe_ctx->update_flags.bits.scaler || plane_state->update_flags.bits.scaling_change || plane_state->update_flags.bits.position_change || - plane_state->update_flags.bits.clip_size_change || plane_state->update_flags.bits.per_pixel_alpha_change || pipe_ctx->stream->update_flags.bits.scaling) { pipe_ctx->plane_res.scl_data.lb_params.alpha_en = pipe_ctx->plane_state->per_pixel_alpha; @@ -1745,7 +1744,6 @@ static void dcn20_update_dchubp_dpp( if (pipe_ctx->update_flags.bits.viewport || (context == dc->current_state && plane_state->update_flags.bits.position_change) || (context == dc->current_state && plane_state->update_flags.bits.scaling_change) || - (context == dc->current_state && plane_state->update_flags.bits.clip_size_change) || (context == dc->current_state && pipe_ctx->stream->update_flags.bits.scaling)) { hubp->funcs->mem_program_viewport( -- 2.50.1 From 4d4257280d7957727998ef90ccc7b69c7cca8376 Mon Sep 17 00:00:00 2001 From: Alex Hung Date: Fri, 6 Sep 2024 11:39:18 -0600 Subject: [PATCH 14/16] drm/amd/display: Add HDR workaround for specific eDP [WHY & HOW] Some eDP panels suffer from flicking when HDR is enabled in KDE. This quirk works around it by skipping VSC that is incompatible with eDP panels. Link: https://gitlab.freedesktop.org/drm/amd/-/issues/3151 Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Rodrigo Siqueira Signed-off-by: Alex Hung Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 11 ++++++++++- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c | 4 ++++ drivers/gpu/drm/amd/display/dc/dc_types.h | 1 + 3 files changed, 15 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 6e79028c5d78..7a826f20c38a 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -6735,12 +6735,21 @@ create_stream_for_sink(struct drm_connector *connector, if (stream->signal == SIGNAL_TYPE_DISPLAY_PORT || stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST || stream->signal == SIGNAL_TYPE_EDP) { + const struct dc_edid_caps *edid_caps; + unsigned int disable_colorimetry = 0; + + if (aconnector->dc_sink) { + edid_caps = &aconnector->dc_sink->edid_caps; + disable_colorimetry = edid_caps->panel_patch.disable_colorimetry; + } + // // should decide stream support vsc sdp colorimetry capability // before building vsc info packet // stream->use_vsc_sdp_for_colorimetry = stream->link->dpcd_caps.dpcd_rev.raw >= 0x14 && - stream->link->dpcd_caps.dprx_feature.bits.VSC_SDP_COLORIMETRY_SUPPORTED; + stream->link->dpcd_caps.dprx_feature.bits.VSC_SDP_COLORIMETRY_SUPPORTED && + !disable_colorimetry; if (stream->out_transfer_func.tf == TRANSFER_FUNCTION_GAMMA22) tf = TRANSFER_FUNC_GAMMA_22; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index 50109d13d967..069e0195e50a 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -73,6 +73,10 @@ static void apply_edid_quirks(struct edid *edid, struct dc_edid_caps *edid_caps) DRM_DEBUG_DRIVER("Clearing DPCD 0x317 on monitor with panel id %X\n", panel_id); edid_caps->panel_patch.remove_sink_ext_caps = true; break; + case drm_edid_encode_panel_id('S', 'D', 'C', 0x4154): + DRM_DEBUG_DRIVER("Disabling VSC on monitor with panel id %X\n", panel_id); + edid_caps->panel_patch.disable_colorimetry = true; + break; default: return; } diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index 2bbafd1cdce4..b0b7102fdbc7 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -178,6 +178,7 @@ struct dc_panel_patch { unsigned int skip_avmute; unsigned int mst_start_top_delay; unsigned int remove_sink_ext_caps; + unsigned int disable_colorimetry; uint8_t blankstream_before_otg_off; }; -- 2.50.1 From 82f8b2cfd7db83924ff9515273f4781fa53648ab Mon Sep 17 00:00:00 2001 From: Austin Zheng Date: Thu, 12 Sep 2024 15:34:14 -0400 Subject: [PATCH 15/16] drm/amd/display: Wait For DET Update Should Use Current State [Why] Current state should be used when waiting for DET update instead of new context. For any streams decreasing in DET, pipes used in the current state should be checked since those pipes need to free their DET before DET can be reallocated. [How] Pass in current_state instead of context. Use pipe from current_state instead of context. This assumes that pipe in the current_state is an OTG_MASTER pipe if the pipe in the context is an OTG_MASTER pipe. Reviewed-by: Dillon Varone Signed-off-by: Austin Zheng Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 9 +++++++-- .../gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h | 2 +- 2 files changed, 8 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 29fcb4ed6646..805e7b52df29 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -1669,7 +1669,7 @@ void dcn401_hardware_release(struct dc *dc) } } -void dcn401_wait_for_det_buffer_update(struct dc *dc, struct dc_state *context, struct pipe_ctx *otg_master) +void dcn401_wait_for_det_buffer_update_under_otg_master(struct dc *dc, struct dc_state *context, struct pipe_ctx *otg_master) { struct pipe_ctx *opp_heads[MAX_PIPES]; struct pipe_ctx *dpp_pipes[MAX_PIPES]; @@ -1695,6 +1695,9 @@ void dcn401_wait_for_det_buffer_update(struct dc *dc, struct dc_state *context, hubbub->funcs->wait_for_det_update) hubbub->funcs->wait_for_det_update(hubbub, dpp_pipe->plane_res.hubp->inst); } + } else { + if (hubbub && opp_heads[slice_idx]->plane_res.hubp && hubbub->funcs->wait_for_det_update) + hubbub->funcs->wait_for_det_update(hubbub, opp_heads[slice_idx]->plane_res.hubp->inst); } } } @@ -1730,8 +1733,10 @@ void dcn401_interdependent_update_lock(struct dc *dc, } if (dc->scratch.pipes_to_unlock_first[i]) { + struct pipe_ctx *old_pipe = &dc->current_state->res_ctx.pipe_ctx[i]; dc->hwss.pipe_control_lock(dc, pipe, false); - dcn401_wait_for_det_buffer_update(dc, context, pipe); + /* Assumes pipe of the same index in current_state is also an OTG_MASTER pipe*/ + dcn401_wait_for_det_buffer_update_under_otg_master(dc, dc->current_state, old_pipe); } } diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h index a27e62081685..1e8189bb447e 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h @@ -81,7 +81,7 @@ void dcn401_hardware_release(struct dc *dc); void dcn401_update_odm(struct dc *dc, struct dc_state *context, struct pipe_ctx *otg_master); void adjust_hotspot_between_slices_for_2x_magnify(uint32_t cursor_width, struct dc_cursor_position *pos_cpy); -void dcn401_wait_for_det_buffer_update(struct dc *dc, struct dc_state *context, struct pipe_ctx *otg_master); +void dcn401_wait_for_det_buffer_update_under_otg_master(struct dc *dc, struct dc_state *context, struct pipe_ctx *otg_master); void dcn401_interdependent_update_lock(struct dc *dc, struct dc_state *context, bool lock); void dcn401_program_outstanding_updates(struct dc *dc, struct dc_state *context); #endif /* __DC_HWSS_DCN401_H__ */ -- 2.50.1 From 181db30bcfed097ecc680539b1eabe935c11f57f Mon Sep 17 00:00:00 2001 From: Leo Li Date: Wed, 11 Sep 2024 17:27:08 -0400 Subject: [PATCH 16/16] drm/amd/display: Enable idle workqueue for more IPS modes [Why] There are more IPS modes other than DMUB_IPS_ENABLE that enables IPS. We need to enable the hotplug detect idle workqueue for those modes as well. [How] Modify the if condition to initialize the workqueue in all IPS modes except for DMUB_IPS_DISABLE_ALL. Fixes: 65444581a4ae ("drm/amd/display: Determine IPS mode by ASIC and PMFW versions") Signed-off-by: Leo Li Reviewed-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 7a826f20c38a..5109146ef2ba 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -2026,7 +2026,8 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) DRM_ERROR("amdgpu: failed to initialize vblank_workqueue.\n"); } - if (adev->dm.dc->caps.ips_support && adev->dm.dc->config.disable_ips == DMUB_IPS_ENABLE) + if (adev->dm.dc->caps.ips_support && + adev->dm.dc->config.disable_ips != DMUB_IPS_DISABLE_ALL) adev->dm.idle_workqueue = idle_create_workqueue(adev); if (adev->dm.dc->caps.max_links > 0 && adev->family >= AMDGPU_FAMILY_RV) { -- 2.50.1