From c00b452db6d24c51d32572f5e61c646fbcdc3e89 Mon Sep 17 00:00:00 2001 From: Harry Wentland Date: Tue, 21 Jan 2020 16:12:45 -0500 Subject: [PATCH 01/69] drm/amd/display: Retrain dongles when SINK_COUNT becomes non-zero [WHY] Two years ago the patch referenced by the Fixes tag stopped running dp_verify_link_cap_with_retries during DP detection when the reason for the detection was a short-pulse interrupt. This effectively meant that we were no longer doing the verify_link_cap training on active dongles when their SINK_COUNT changed from 0 to 1. A year ago this was partly remedied with this commit: 80adaebd2d41 drm/amd/display: Don't skip link training for empty dongle This made sure that we trained the dongle on initial hotplug (without connected downstream devices). This is all fine and dandy if it weren't for the fact that there are some dongles on the market that don't like link training when SINK_COUNT is 0 These dongles will in fact indicate a SINK_COUNT of 0 immediately after hotplug, even when a downstream device is connected, and then trigger a shortpulse interrupt indicating a SINK_COUNT change to 1. In order to play nicely we will need our policy to not link train an active DP dongle when SINK_COUNT is 0 but ensure we train it when the SINK_COUNT changes to 1. [HOW] Call dp_verify_link_cap_with_retries on detection even when the detection is triggered from a short pulse interrupt. With this change we can also revert this commit which we'll do in a separate follow-up change: 80adaebd2d41 drm/amd/display: Don't skip link training for empty dongle Fixes: 0301ccbaf67d drm/amd/display: DP Compliance 400.1.1 failure Suggested-by: Louis Li Tested-by: Louis Li Cc: Wenjing Liu Cc: Hersen Wu Cc: Eric Yang Reviewed-by: Wenjing Liu Signed-off-by: Harry Wentland --- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c index 3a0e3ff906226..90656279028c9 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c @@ -969,8 +969,7 @@ static bool dc_link_detect_helper(struct dc_link *link, same_edid = is_same_edid(&prev_sink->dc_edid, &sink->dc_edid); if (link->connector_signal == SIGNAL_TYPE_DISPLAY_PORT && - sink_caps.transaction_type == DDC_TRANSACTION_TYPE_I2C_OVER_AUX && - reason != DETECT_REASON_HPDRX) { + sink_caps.transaction_type == DDC_TRANSACTION_TYPE_I2C_OVER_AUX) { /* * TODO debug why Dell 2413 doesn't like * two link trainings From 5adcbdbec82377e6d0f6f16994c2d61e7c65a301 Mon Sep 17 00:00:00 2001 From: Harry Wentland Date: Tue, 21 Jan 2020 16:29:54 -0500 Subject: [PATCH 02/69] Revert "drm/amd/display: Don't skip link training for empty dongle" This reverts commit 80adaebd2d411b7d6872a097634848a71eb13d20. [WHY] This change was working around a regression that occured in this patch: 0301ccbaf67d drm/amd/display: DP Compliance 400.1.1 failure With the fix to run verify_link_cap when the SINK_COUNT of dongles becomes non-zero this change is no longer needed. Cc: Louis Li Cc: Wenjing Liu Cc: Hersen Wu Cc: Eric Yang Reviewed-by: Wenjing Liu Signed-off-by: Harry Wentland --- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 10 ++-------- 1 file changed, 2 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c index 90656279028c9..64223e7551bd6 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c @@ -851,18 +851,12 @@ static bool dc_link_detect_helper(struct dc_link *link, if (memcmp(&link->dpcd_caps, &prev_dpcd_caps, sizeof(struct dpcd_caps))) same_dpcd = false; } - /* Active dongle plug in without display or downstream unplug*/ + /* Active dongle downstream unplug*/ if (link->type == dc_connection_active_dongle && link->dpcd_caps.sink_count.bits.SINK_COUNT == 0) { - if (prev_sink != NULL) { + if (prev_sink != NULL) /* Downstream unplug */ dc_sink_release(prev_sink); - } else { - /* Empty dongle plug in */ - dp_verify_link_cap_with_retries(link, - &link->reported_link_cap, - LINK_TRAINING_MAX_VERIFY_RETRY); - } return true; } From 2e52b5948d046013d8937d1c6f2fb9a8a7c9e506 Mon Sep 17 00:00:00 2001 From: Dor Askayo Date: Sat, 4 Jan 2020 14:22:15 +0200 Subject: [PATCH 03/69] drm/amd/display: do not allocate display_mode_lib unnecessarily This allocation isn't required and can fail when resuming from suspend. Bug: https://gitlab.freedesktop.org/drm/amd/issues/1009 Signed-off-by: Dor Askayo Reviewed-by: Leo Li --- drivers/gpu/drm/amd/display/dc/core/dc.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 977068b908e55..5f9390d8a3128 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -2537,12 +2537,7 @@ void dc_set_power_state( enum dc_acpi_cm_power_state power_state) { struct kref refcount; - struct display_mode_lib *dml = kzalloc(sizeof(struct display_mode_lib), - GFP_KERNEL); - - ASSERT(dml); - if (!dml) - return; + struct display_mode_lib *dml; switch (power_state) { case DC_ACPI_CM_POWER_STATE_D0: @@ -2567,6 +2562,12 @@ void dc_set_power_state( * clean state, and dc hw programming optimizations will not * cause any trouble. */ + dml = kzalloc(sizeof(struct display_mode_lib), + GFP_KERNEL); + + ASSERT(dml); + if (!dml) + return; /* Preserve refcount */ refcount = dc->current_state->refcount; @@ -2580,10 +2581,10 @@ void dc_set_power_state( dc->current_state->refcount = refcount; dc->current_state->bw_ctx.dml = *dml; + kfree(dml); + break; } - - kfree(dml); } void dc_resume(struct dc *dc) From 7c5792b007002b47e0d42066a604f7593bc4dbe5 Mon Sep 17 00:00:00 2001 From: Colin Ian King Date: Fri, 17 Jan 2020 13:33:05 +0000 Subject: [PATCH 04/69] drm/amd/display: fix for-loop with incorrectly sized loop counter (v2) A for-loop is iterating from 0 up to 1000 however the loop variable count is a u8 and hence not large enough. Fix this by making count an int. Also remove the redundant initialization of count since this is never used and add { } on the loop statement make the loop block clearer. v2: drop useless else (Walter Harms) Addresses-Coverity: ("Operands don't affect result") Fixes: ed581a0ace44 ("drm/amd/display: wait for update when setting dpg test pattern") Signed-off-by: Colin Ian King Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c index 89deae776ffad..ddf42667822ff 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c @@ -3699,7 +3699,7 @@ static void set_crtc_test_pattern(struct dc_link *link, struct pipe_ctx *odm_pipe; enum controller_dp_color_space controller_color_space; int opp_cnt = 1; - uint8_t count = 0; + int count; switch (test_pattern_color_space) { case DP_TEST_PATTERN_COLOR_SPACE_RGB: @@ -3744,11 +3744,11 @@ static void set_crtc_test_pattern(struct dc_link *link, width, height); /* wait for dpg to blank pixel data with test pattern */ - for (count = 0; count < 1000; count++) + for (count = 0; count < 1000; count++) { if (opp->funcs->dpg_is_blanked(opp)) break; - else - udelay(100); + udelay(100); + } } #endif } From 2c4d9ca8bec7420117ab2960b28384cfd065ccfd Mon Sep 17 00:00:00 2001 From: Harry Wentland Date: Tue, 21 Jan 2020 16:29:54 -0500 Subject: [PATCH 05/69] Revert "drm/amd/display: Don't skip link training for empty dongle" This reverts commit 80adaebd2d411b7d6872a097634848a71eb13d20. [WHY] This change was working around a regression that occured in this patch: 0301ccbaf67d drm/amd/display: DP Compliance 400.1.1 failure With the fix to run verify_link_cap when the SINK_COUNT of dongles becomes non-zero this change is no longer needed. Change-Id: Iff11cfde28272ad730698801adeacbe0f8066315 Cc: Louis Li Cc: Wenjing Liu Cc: Hersen Wu Cc: Eric Yang Reviewed-by: Wenjing Liu Signed-off-by: Harry Wentland From ece575ea247d3d8ac976bd37ef6f5675c075ec73 Mon Sep 17 00:00:00 2001 From: Colin Ian King Date: Tue, 28 Jan 2020 11:28:27 +0000 Subject: [PATCH 06/69] drm/amd/display: fix spelling mistake link_integiry_check -> link_integrity_check There is a spelling mistake on the struct field name link_integiry_check, fix this by renaming it. Signed-off-by: Colin Ian King Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h | 2 +- .../gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c | 8 ++++---- .../gpu/drm/amd/display/modules/hdcp/hdcp1_transition.c | 4 ++-- 3 files changed, 7 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h index f98d3d9ecb6df..af78e4f1be686 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h @@ -63,7 +63,7 @@ struct mod_hdcp_transition_input_hdcp1 { uint8_t hdcp_capable_dp; uint8_t binfo_read_dp; uint8_t r0p_available_dp; - uint8_t link_integiry_check; + uint8_t link_integrity_check; uint8_t reauth_request_check; uint8_t stream_encryption_dp; }; diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c index 04845e43df15c..37670db648556 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c @@ -283,8 +283,8 @@ static enum mod_hdcp_status wait_for_ready(struct mod_hdcp *hdcp, hdcp, "bstatus_read")) goto out; if (!mod_hdcp_execute_and_set(check_link_integrity_dp, - &input->link_integiry_check, &status, - hdcp, "link_integiry_check")) + &input->link_integrity_check, &status, + hdcp, "link_integrity_check")) goto out; if (!mod_hdcp_execute_and_set(check_no_reauthentication_request_dp, &input->reauth_request_check, &status, @@ -431,8 +431,8 @@ static enum mod_hdcp_status authenticated_dp(struct mod_hdcp *hdcp, hdcp, "bstatus_read")) goto out; if (!mod_hdcp_execute_and_set(check_link_integrity_dp, - &input->link_integiry_check, &status, - hdcp, "link_integiry_check")) + &input->link_integrity_check, &status, + hdcp, "link_integrity_check")) goto out; if (!mod_hdcp_execute_and_set(check_no_reauthentication_request_dp, &input->reauth_request_check, &status, diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_transition.c b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_transition.c index 21ebc62bb9d9e..76edcbe51f71c 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_transition.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_transition.c @@ -241,7 +241,7 @@ enum mod_hdcp_status mod_hdcp_hdcp1_dp_transition(struct mod_hdcp *hdcp, } break; case D1_A4_AUTHENTICATED: - if (input->link_integiry_check != PASS || + if (input->link_integrity_check != PASS || input->reauth_request_check != PASS) { /* 1A-07: restart hdcp on a link integrity failure */ fail_and_restart_in_ms(0, &status, output); @@ -249,7 +249,7 @@ enum mod_hdcp_status mod_hdcp_hdcp1_dp_transition(struct mod_hdcp *hdcp, } break; case D1_A6_WAIT_FOR_READY: - if (input->link_integiry_check == FAIL || + if (input->link_integrity_check == FAIL || input->reauth_request_check == FAIL) { fail_and_restart_in_ms(0, &status, output); break; From 7478c608b197b0833266db965d3a7f9cde1b74d1 Mon Sep 17 00:00:00 2001 From: Zhan Liu Date: Tue, 28 Jan 2020 16:38:53 -0500 Subject: [PATCH 07/69] drm/amd/display: Move drm_dp_mst_atomic_check() to the front of dc_validate_global_state() [Why] Need to do atomic check first, then validate global state. If not, when connecting both MST and HDMI displays and set a bad mode via xrandr, system will hang. [How] Move drm_dp_mst_atomic_check() to the front of dc_validate_global_state(). Change-Id: I913fa3a58852eee8f739a291591254f4be7ba446 Signed-off-by: Zhan Liu Reviewed-by: Mikita Lipski --- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 22 ++++++++++++------- 1 file changed, 14 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 43a5e988ed041..fa7fd631cb51d 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -9307,6 +9307,20 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, goto fail; #endif #endif +#endif + + /* + * Perform validation of MST topology in the state: + * We need to perform MST atomic check before calling + * dc_validate_global_state(), or there is a chance + * to get stuck in an infinite loop and hang eventually. + */ +#ifdef HAVE_DRM_DP_MST_ATOMIC_CHECK +#if defined(HAVE_DRM_DP_MST_ATOMIC_ENABLE_DSC) + ret = drm_dp_mst_atomic_check(state); + if (ret) + goto fail; +#endif #endif if (dc_validate_global_state(dc, dm_state->context, false) != DC_OK) { @@ -9339,14 +9353,6 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, } #endif } - /* Perform validation of MST topology in the state*/ -#ifdef HAVE_DRM_DP_MST_ATOMIC_CHECK -#if defined(HAVE_DRM_DP_MST_ATOMIC_ENABLE_DSC) - ret = drm_dp_mst_atomic_check(state); - if (ret) - goto fail; -#endif -#endif /* Store the overall update type for use later in atomic check. */ #if !defined(for_each_new_crtc_in_state) From 3c88f42d228d20c67902dfb73378767cab939729 Mon Sep 17 00:00:00 2001 From: Nicholas Kazlauskas Date: Thu, 30 Jan 2020 13:29:05 -0500 Subject: [PATCH 08/69] drm/amd/display: Only enable cursor on pipes that need it [Why] In current code we're essentially drawing the cursor on every pipe that contains it. This only works when the planes have the same scaling for src to dest rect, otherwise we'll get "double cursor" where one cursor is incorrectly filtered and offset from the real position. [How] Without dedicated cursor planes on DCN we require at least one pipe that matches the scaling of the current timing. This is an optimization and workaround for the most common case where the top-most plane is not scaled but the bottom-most plane is scaled. Whenever a pipe has a parent pipe in the blending tree whose recout fully contains the current pipe we can disable the pipe. This only applies when the pipe is actually visible of course. Signed-off-by: Nicholas Kazlauskas Reviewed-by: Harry Wentland --- .../amd/display/dc/dcn10/dcn10_hw_sequencer.c | 30 +++++++++++++++++++ 1 file changed, 30 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c index 5a5dbb60d60b4..8a92abf48d51a 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c @@ -2929,6 +2929,33 @@ void dcn10_update_dchub(struct dce_hwseq *hws, struct dchub_init_data *dh_data) hubbub->funcs->update_dchub(hubbub, dh_data); } +static bool dcn10_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx) +{ + struct pipe_ctx *test_pipe; + const struct rect *r1 = &pipe_ctx->plane_res.scl_data.recout, *r2; + int r1_r = r1->x + r1->width, r1_b = r1->y + r1->height, r2_r, r2_b; + + /** + * Disable the cursor if there's another pipe above this with a + * plane that contains this pipe's viewport to prevent double cursor + * and incorrect scaling artifacts. + */ + for (test_pipe = pipe_ctx->top_pipe; test_pipe; + test_pipe = test_pipe->top_pipe) { + if (!test_pipe->plane_state->visible) + continue; + + r2 = &test_pipe->plane_res.scl_data.recout; + r2_r = r2->x + r2->width; + r2_b = r2->y + r2->height; + + if (r1->x >= r2->x && r1->y >= r2->y && r1_r <= r2_r && r1_b <= r2_b) + return true; + } + + return false; +} + void dcn10_set_cursor_position(struct pipe_ctx *pipe_ctx) { struct dc_cursor_position pos_cpy = pipe_ctx->stream->cursor_position; @@ -2974,6 +3001,9 @@ void dcn10_set_cursor_position(struct pipe_ctx *pipe_ctx) == PLN_ADDR_TYPE_VIDEO_PROGRESSIVE) pos_cpy.enable = false; + if (pos_cpy.enable && dcn10_can_pipe_disable_cursor(pipe_ctx)) + pos_cpy.enable = false; + // Swap axis and mirror horizontally if (param.rotation == ROTATION_ANGLE_90) { uint32_t temp_x = pos_cpy.x; From 3f655901df2bb6e65f15bccc541f941ded8c8a65 Mon Sep 17 00:00:00 2001 From: Mikita Lipski Date: Fri, 31 Jan 2020 09:51:23 -0500 Subject: [PATCH 09/69] drm/amd/display: Fix a typo when computing dsc configuration [why] Remove a backslash symbol accidentally left in increase bpp function when computing mst dsc configuration. Signed-off-by: Mikita Lipski Reviewed-by: Zhan Liu --- 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 0e9e93d172b31..34d5e871cab0a 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 @@ -736,7 +736,7 @@ static void increase_dsc_bpp(struct drm_atomic_state *state, if (drm_dp_atomic_find_vcpi_slots(state, params[next_index].port->mgr, params[next_index].port, - vars[next_index].pbn,\ + vars[next_index].pbn, dm_mst_get_pbn_divider(dc_link)) < 0) return; if (!drm_dp_mst_atomic_check(state)) { From d764465e2ef5b51c8ff11a64cf35dc4522952440 Mon Sep 17 00:00:00 2001 From: Bhawanpreet Lakha Date: Fri, 6 Dec 2019 13:16:08 -0500 Subject: [PATCH 10/69] drm/amd/display: Fix HW/SW state mismatch [Why] When we disable a connector we don't explicitly remove it from the module so the display is still cached(SW) in the hdcp_module. SST: no issues because we can only have 1 display per link MST: We have x displays per link, now if we disable 1 we don't remove it from the module so the module has x display cached(SW). If we try to enable HDCP, psp verification will fail because we are reporting x displays while the HW only has x-1 display enabled [How] Check the callback for when we disable stream and call remove display. Signed-off-by: Bhawanpreet Lakha Reviewed-by: Rodrigo Siqueira --- .../amd/display/amdgpu_dm/amdgpu_dm_hdcp.c | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c index ae329335dfcc7..0acd3409dd6ce 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c @@ -135,6 +135,20 @@ void hdcp_update_display(struct hdcp_workqueue *hdcp_work, mutex_unlock(&hdcp_w->mutex); } +static void hdcp_remove_display(struct hdcp_workqueue *hdcp_work, + unsigned int link_index, + struct amdgpu_dm_connector *aconnector) +{ + struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index]; + + mutex_lock(&hdcp_w->mutex); + hdcp_w->aconnector = aconnector; + + mod_hdcp_remove_display(&hdcp_w->hdcp, aconnector->base.index, &hdcp_w->output); + + process_output(hdcp_w); + mutex_unlock(&hdcp_w->mutex); +} void hdcp_reset_display(struct hdcp_workqueue *hdcp_work, unsigned int link_index) { struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index]; @@ -303,6 +317,11 @@ static void update_config(void *handle, struct cp_psp_stream_config *config) memset(link, 0, sizeof(*link)); display->index = aconnector->base.index; + + if (config->dpms_off) { + hdcp_remove_display(hdcp_work, link_index, aconnector); + return; + } display->state = MOD_HDCP_DISPLAY_ACTIVE; if (aconnector->dc_sink != NULL) From ca2591d6c7cccf205d678edac69001708e8ce379 Mon Sep 17 00:00:00 2001 From: Bhawanpreet Lakha Date: Thu, 12 Dec 2019 16:28:23 -0500 Subject: [PATCH 11/69] drm/amd/display: Pass amdgpu_device instead of psp_context [Why] We need this to create sysfs (followup patch) [How] Change the parameter Signed-off-by: Bhawanpreet Lakha Reviewed-by: Rodrigo Siqueira Reviewed-by: Harry Wentland --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 2 +- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c | 4 ++-- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h | 3 ++- 3 files changed, 5 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index fa7fd631cb51d..464ad24e7b086 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -1004,7 +1004,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) #ifdef CONFIG_DRM_AMD_DC_HDCP if (adev->asic_type >= CHIP_RAVEN) { - adev->dm.hdcp_workqueue = hdcp_create_workqueue(&adev->psp, &init_params.cp_psp, adev->dm.dc); + adev->dm.hdcp_workqueue = hdcp_create_workqueue(adev, &init_params.cp_psp, adev->dm.dc); if (!adev->dm.hdcp_workqueue) DRM_ERROR("amdgpu: failed to initialize hdcp_workqueue.\n"); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c index 0acd3409dd6ce..a269916f7dd6c 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c @@ -338,7 +338,7 @@ static void update_config(void *handle, struct cp_psp_stream_config *config) hdcp_update_display(hdcp_work, link_index, aconnector, DRM_MODE_HDCP_CONTENT_TYPE0, false); } -struct hdcp_workqueue *hdcp_create_workqueue(void *psp_context, struct cp_psp *cp_psp, struct dc *dc) +struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev, struct cp_psp *cp_psp, struct dc *dc) { int max_caps = dc->caps.max_links; @@ -360,7 +360,7 @@ struct hdcp_workqueue *hdcp_create_workqueue(void *psp_context, struct cp_psp *c INIT_DELAYED_WORK(&hdcp_work[i].watchdog_timer_dwork, event_watchdog_timer); INIT_DELAYED_WORK(&hdcp_work[i].property_validate_dwork, event_property_validate); - hdcp_work[i].hdcp.config.psp.handle = psp_context; + hdcp_work[i].hdcp.config.psp.handle = &adev->psp; hdcp_work[i].hdcp.config.ddc.handle = dc_get_link_at_index(dc, i); hdcp_work[i].hdcp.config.ddc.funcs.write_i2c = lp_write_i2c; hdcp_work[i].hdcp.config.ddc.funcs.read_i2c = lp_read_i2c; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h index 6abde86bce4aa..331b50825510d 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h @@ -30,6 +30,7 @@ #include "hdcp.h" #include "dc.h" #include "dm_cp_psp.h" +#include "amdgpu.h" struct mod_hdcp; struct mod_hdcp_link; @@ -64,6 +65,6 @@ void hdcp_reset_display(struct hdcp_workqueue *work, unsigned int link_index); void hdcp_handle_cpirq(struct hdcp_workqueue *work, unsigned int link_index); void hdcp_destroy(struct hdcp_workqueue *work); -struct hdcp_workqueue *hdcp_create_workqueue(void *psp_context, struct cp_psp *cp_psp, struct dc *dc); +struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev, struct cp_psp *cp_psp, struct dc *dc); #endif /* AMDGPU_DM_AMDGPU_DM_HDCP_H_ */ From 70d50bb215eb53b950a7fc183ff348ed92411838 Mon Sep 17 00:00:00 2001 From: Bhawanpreet Lakha Date: Thu, 12 Dec 2019 16:00:01 -0500 Subject: [PATCH 12/69] drm/amd/display: update psp interface header [Why] We need to support SRM(System Renewability Message) As per hdcp spec (5.Renewability) SRM needs to be storage in a non-volatile memory. PSP owns the checking of SRM but doesn't have the ability to store it in a non-volatile memory. So we need the kernel driver to facilitate it using the interface provided by PSP [How] Add the interface to the header file, so the driver can use them v2: update commit description Signed-off-by: Bhawanpreet Lakha Reviewed-by: Rodrigo Siqueira Reviewed-by: Harry Wentland --- .../drm/amd/display/modules/hdcp/hdcp_psp.h | 26 ++++++++++++++++++- 1 file changed, 25 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_psp.h b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_psp.h index 82a5e997d573f..d5cb3f46606ff 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_psp.h +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_psp.h @@ -117,6 +117,8 @@ struct ta_dtm_shared_memory { int psp_cmd_submit_buf(struct psp_context *psp, struct amdgpu_firmware_info *ucode, struct psp_gfx_cmd_resp *cmd, uint64_t fence_mc_addr); +enum { PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE = 5120 }; + enum ta_hdcp_command { TA_HDCP_COMMAND__INITIALIZE, TA_HDCP_COMMAND__HDCP1_CREATE_SESSION, @@ -134,7 +136,10 @@ enum ta_hdcp_command { TA_HDCP_COMMAND__UNUSED_3, TA_HDCP_COMMAND__HDCP2_CREATE_SESSION_V2, TA_HDCP_COMMAND__HDCP2_PREPARE_PROCESS_AUTHENTICATION_MSG_V2, - TA_HDCP_COMMAND__HDCP2_ENABLE_DP_STREAM_ENCRYPTION + TA_HDCP_COMMAND__HDCP2_ENABLE_DP_STREAM_ENCRYPTION, + TA_HDCP_COMMAND__HDCP_DESTROY_ALL_SESSIONS, + TA_HDCP_COMMAND__HDCP_SET_SRM, + TA_HDCP_COMMAND__HDCP_GET_SRM }; enum ta_hdcp2_msg_id { @@ -415,6 +420,22 @@ struct ta_hdcp_cmd_hdcp2_enable_dp_stream_encryption_input { uint32_t display_handle; }; +struct ta_hdcp_cmd_set_srm_input { + uint32_t srm_buf_size; + uint8_t srm_buf[PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE]; +}; + +struct ta_hdcp_cmd_set_srm_output { + uint8_t valid_signature; + uint32_t srm_version; +}; + +struct ta_hdcp_cmd_get_srm_output { + uint32_t srm_version; + uint32_t srm_buf_size; + uint8_t srm_buf[PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE]; +}; + /**********************************************************/ /* Common input structure for HDCP callbacks */ union ta_hdcp_cmd_input { @@ -432,6 +453,7 @@ union ta_hdcp_cmd_input { struct ta_hdcp_cmd_hdcp2_process_prepare_authentication_message_input_v2 hdcp2_prepare_process_authentication_message_v2; struct ta_hdcp_cmd_hdcp2_enable_dp_stream_encryption_input hdcp2_enable_dp_stream_encryption; + struct ta_hdcp_cmd_set_srm_input hdcp_set_srm; }; /* Common output structure for HDCP callbacks */ @@ -444,6 +466,8 @@ union ta_hdcp_cmd_output { struct ta_hdcp_cmd_hdcp2_create_session_output_v2 hdcp2_create_session_v2; struct ta_hdcp_cmd_hdcp2_process_prepare_authentication_message_output_v2 hdcp2_prepare_process_authentication_message_v2; + struct ta_hdcp_cmd_set_srm_output hdcp_set_srm; + struct ta_hdcp_cmd_get_srm_output hdcp_get_srm; }; /**********************************************************/ From f7eaa044ed816b9bf725fea67717f0e7ed5eb17f Mon Sep 17 00:00:00 2001 From: Bhawanpreet Lakha Date: Thu, 12 Dec 2019 16:36:32 -0500 Subject: [PATCH 13/69] drm/amd/display: Add sysfs interface for set/get srm [Why] PSP doesn't have the ability to store SRM in a non-volatile memory. And since the kernel cannot write to the storage directly, we need usermode to facilitate this As per spec the SRM needs to be persistent so this interface is to be called by the usermode anytime the system goes down/powers on *boot/resume: load from storage *shutdown/suspend: save to storage [How] Provide a sysfs interface so that the usermode can set/get srm at the right times save to storage: call "cat /sys/class/drm/card0/device/hdcp_srm > file" after boot and resume -driver calls psp_get_srm() to get the stored srm and outputs it load from storage: call "cat file > /sys/class/drm/card0/device/hdcp_srm" before shutdown and suspend -driver reads the file from sysfs and calls psp_set_srm() to send the SRM to PSP v2: -update commit description -add comment about sysfs file handling in the code Signed-off-by: Bhawanpreet Lakha Reviewed-by: Rodrigo Siqueira Reviewed-by: Harry Wentland --- .../amd/display/amdgpu_dm/amdgpu_dm_hdcp.c | 179 +++++++++++++++++- .../amd/display/amdgpu_dm/amdgpu_dm_hdcp.h | 6 + 2 files changed, 183 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c index a269916f7dd6c..074243a2f808a 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c @@ -28,6 +28,8 @@ #include "amdgpu_dm.h" #include "dm_helpers.h" #include +#include "hdcp_psp.h" + static bool lp_write_i2c(void *handle, uint32_t address, const uint8_t *data, uint32_t size) @@ -67,6 +69,16 @@ lp_read_dpcd(void *handle, uint32_t address, uint8_t *data, uint32_t size) return dm_helpers_dp_read_dpcd(link->ctx, link, address, data, size); } +static uint8_t *psp_get_srm(struct psp_context *psp, uint32_t *srm_version, uint32_t *srm_size) +{ + return NULL; +} + +static int psp_set_srm(struct psp_context *psp, uint8_t *srm, uint32_t srm_size, uint32_t *srm_version) +{ + return 0; +} + static void process_output(struct hdcp_workqueue *hdcp_work) { struct mod_hdcp_output output = hdcp_work->output; @@ -88,6 +100,18 @@ static void process_output(struct hdcp_workqueue *hdcp_work) schedule_delayed_work(&hdcp_work->property_validate_dwork, msecs_to_jiffies(0)); } +static void link_lock(struct hdcp_workqueue *work, bool lock) +{ + + int i = 0; + + for (i = 0; i < work->max_link; i++) { + if (lock) + mutex_lock(&work[i].mutex); + else + mutex_unlock(&work[i].mutex); + } +} void hdcp_update_display(struct hdcp_workqueue *hdcp_work, unsigned int link_index, struct amdgpu_dm_connector *aconnector, @@ -302,7 +326,8 @@ void hdcp_destroy(struct hdcp_workqueue *hdcp_work) } kfree(hdcp_work); - + kfree(hdcp_work->srm); + kfree(hdcp_work->srm_temp); } static void update_config(void *handle, struct cp_psp_stream_config *config) @@ -338,6 +363,139 @@ static void update_config(void *handle, struct cp_psp_stream_config *config) hdcp_update_display(hdcp_work, link_index, aconnector, DRM_MODE_HDCP_CONTENT_TYPE0, false); } + +/* NOTE: From the usermodes prospective you only need to call write *ONCE*, the kernel + * will automatically call once or twice depending on the size + * + * call: "cat file > /sys/class/drm/card0/device/hdcp_srm" from usermode no matter what the size is + * + * The kernel can only send PAGE_SIZE at once and since MAX_SRM_FILE(5120) > PAGE_SIZE(4096), + * srm_data_write can be called multiple times. + * + * sysfs interface doesn't tell us the size we will get so we are sending partial SRMs to psp and on + * the last call we will send the full SRM. PSP will fail on every call before the last. + * + * This means we don't know if the SRM is good until the last call. And because of this limitation we + * cannot throw errors early as it will stop the kernel from writing to sysfs + * + * Example 1: + * Good SRM size = 5096 + * first call to write 4096 -> PSP fails + * Second call to write 1000 -> PSP Pass -> SRM is set + * + * Example 2: + * Bad SRM size = 4096 + * first call to write 4096 -> PSP fails (This is the same as above, but we don't know if this + * is the last call) + * + * Solution?: + * 1: Parse the SRM? -> It is signed so we don't know the EOF + * 2: We can have another sysfs that passes the size before calling set. -> simpler solution + * below + * + * Easy Solution: + * Always call get after Set to verify if set was successful. + * +----------------------+ + * | Why it works: | + * +----------------------+ + * PSP will only update its srm if its older than the one we are trying to load. + * Always do set first than get. + * -if we try to "1. SET" a older version PSP will reject it and we can "2. GET" the newer + * version and save it + * + * -if we try to "1. SET" a newer version PSP will accept it and we can "2. GET" the + * same(newer) version back and save it + * + * -if we try to "1. SET" a newer version and PSP rejects it. That means the format is + * incorrect/corrupted and we should correct our SRM by getting it from PSP + */ +static ssize_t srm_data_write(struct file *filp, struct kobject *kobj, struct bin_attribute *bin_attr, char *buffer, + loff_t pos, size_t count) +{ + struct hdcp_workqueue *work; + uint32_t srm_version = 0; + + work = container_of(bin_attr, struct hdcp_workqueue, attr); + link_lock(work, true); + + memcpy(work->srm_temp + pos, buffer, count); + + if (!psp_set_srm(work->hdcp.config.psp.handle, work->srm_temp, pos + count, &srm_version)) { + DRM_DEBUG_DRIVER("HDCP SRM SET version 0x%X", srm_version); + memcpy(work->srm, work->srm_temp, pos + count); + work->srm_size = pos + count; + work->srm_version = srm_version; + } + + + link_lock(work, false); + + return count; +} + +static ssize_t srm_data_read(struct file *filp, struct kobject *kobj, struct bin_attribute *bin_attr, char *buffer, + loff_t pos, size_t count) +{ + struct hdcp_workqueue *work; + uint8_t *srm = NULL; + uint32_t srm_version; + uint32_t srm_size; + size_t ret = count; + + work = container_of(bin_attr, struct hdcp_workqueue, attr); + + link_lock(work, true); + + srm = psp_get_srm(work->hdcp.config.psp.handle, &srm_version, &srm_size); + + if (!srm) + return -EINVAL; + + if (pos >= srm_size) + ret = 0; + + if (srm_size - pos < count) { + memcpy(buffer, srm + pos, srm_size - pos); + ret = srm_size - pos; + goto ret; + } + + memcpy(buffer, srm + pos, count); + +ret: + link_lock(work, false); + return ret; +} + +/* From the hdcp spec (5.Renewability) SRM needs to be stored in a non-volatile memory. + * + * For example, + * if Application "A" sets the SRM (ver 2) and we reboot/suspend and later when Application "B" + * needs to use HDCP, the version in PSP should be SRM(ver 2). So SRM should be persistent + * across boot/reboots/suspend/resume/shutdown + * + * Currently when the system goes down (suspend/shutdown) the SRM is cleared from PSP. For HDCP we need + * to make the SRM persistent. + * + * -PSP owns the checking of SRM but doesn't have the ability to store it in a non-volatile memory. + * -The kernel cannot write to the file systems. + * -So we need usermode to do this for us, which is why an interface for usermode is needed + * + * + * + * Usermode can read/write to/from PSP using the sysfs interface + * For example: + * to save SRM from PSP to storage : cat /sys/class/drm/card0/device/hdcp_srm > srmfile + * to load from storage to PSP: cat srmfile > /sys/class/drm/card0/device/hdcp_srm + */ +static const struct bin_attribute data_attr = { + .attr = {.name = "hdcp_srm", .mode = 0664}, + .size = PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE, /* Limit SRM size */ + .write = srm_data_write, + .read = srm_data_read, +}; + + struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev, struct cp_psp *cp_psp, struct dc *dc) { @@ -348,10 +506,19 @@ struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev, struct if (hdcp_work == NULL) goto fail_alloc_context; + hdcp_work->srm = kcalloc(PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE, sizeof(*hdcp_work->srm), GFP_KERNEL); + + if (hdcp_work->srm == NULL) + goto fail_alloc_context; + + hdcp_work->srm_temp = kcalloc(PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE, sizeof(*hdcp_work->srm_temp), GFP_KERNEL); + + if (hdcp_work->srm_temp == NULL) + goto fail_alloc_context; + hdcp_work->max_link = max_caps; for (i = 0; i < max_caps; i++) { - mutex_init(&hdcp_work[i].mutex); INIT_WORK(&hdcp_work[i].cpirq_work, event_cpirq); @@ -371,10 +538,18 @@ struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev, struct cp_psp->funcs.update_stream_config = update_config; cp_psp->handle = hdcp_work; + /* File created at /sys/class/drm/card0/device/hdcp_srm*/ + hdcp_work[0].attr = data_attr; + + if (sysfs_create_bin_file(&adev->dev->kobj, &hdcp_work[0].attr)) + DRM_WARN("Failed to create device file hdcp_srm"); + return hdcp_work; fail_alloc_context: kfree(hdcp_work); + kfree(hdcp_work->srm); + kfree(hdcp_work->srm_temp); return NULL; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h index 331b50825510d..5159b3a5e5b03 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.h @@ -53,6 +53,12 @@ struct hdcp_workqueue { enum mod_hdcp_encryption_status encryption_status; uint8_t max_link; + + uint8_t *srm; + uint8_t *srm_temp; + uint32_t srm_version; + uint32_t srm_size; + struct bin_attribute attr; }; void hdcp_update_display(struct hdcp_workqueue *hdcp_work, From eabf7849b66c9b654cb1602d9b50ec5a4a898dd8 Mon Sep 17 00:00:00 2001 From: Bhawanpreet Lakha Date: Wed, 4 Dec 2019 11:54:24 -0500 Subject: [PATCH 14/69] drm/amd/display: Load srm before enabling HDCP [Why] we need to load SRM before we start HDCP. Because for S3 case the sysfs call will be after we have already enabled HDCP, so we might not be using the latest SRM [How] Set srm before starting HDCP. Signed-off-by: Bhawanpreet Lakha Reviewed-by: Rodrigo Siqueira Reviewed-by: Harry Wentland --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c index 074243a2f808a..e6a89cd7ed57b 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c @@ -136,6 +136,13 @@ void hdcp_update_display(struct hdcp_workqueue *hdcp_work, hdcp_w->link.adjust.hdcp2.force_type = MOD_HDCP_FORCE_TYPE_0; if (enable_encryption) { + /* Explicitly set the saved SRM as sysfs call will be after we already enabled hdcp + * (s3 resume case) + */ + if (hdcp_work->srm_size > 0) + psp_set_srm(hdcp_work->hdcp.config.psp.handle, hdcp_work->srm, hdcp_work->srm_size, + &hdcp_work->srm_version); + display->adjust.disable = 0; if (content_type == DRM_MODE_HDCP_CONTENT_TYPE0) hdcp_w->link.adjust.hdcp2.force_type = MOD_HDCP_FORCE_TYPE_0; From 37b1ba4b5720def878fe2c917ece91913f876069 Mon Sep 17 00:00:00 2001 From: Bhawanpreet Lakha Date: Thu, 12 Dec 2019 16:38:38 -0500 Subject: [PATCH 15/69] drm/amd/display: call psp set/get interfaces Call the cmd ids for set/get srm according to the sysfs call v2: Use define for the magic number Signed-off-by: Bhawanpreet Lakha Reviewed-by: Rodrigo Siqueira Reviewed-by: Harry Wentland --- .../amd/display/amdgpu_dm/amdgpu_dm_hdcp.c | 50 ++++++++++++++++++- 1 file changed, 49 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c index e6a89cd7ed57b..1768a33b1dc3a 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c @@ -30,6 +30,11 @@ #include #include "hdcp_psp.h" +/* + * If the SRM version being loaded is less than or equal to the + * currently loaded SRM, psp will return 0xFFFF as the version + */ +#define PSP_SRM_VERSION_MAX 0xFFFF static bool lp_write_i2c(void *handle, uint32_t address, const uint8_t *data, uint32_t size) @@ -71,11 +76,54 @@ lp_read_dpcd(void *handle, uint32_t address, uint8_t *data, uint32_t size) static uint8_t *psp_get_srm(struct psp_context *psp, uint32_t *srm_version, uint32_t *srm_size) { - return NULL; + + struct ta_hdcp_shared_memory *hdcp_cmd; + + if (!psp->hdcp_context.hdcp_initialized) { + DRM_WARN("Failed to get hdcp srm. HDCP TA is not initialized."); + return NULL; + } + + hdcp_cmd = (struct ta_hdcp_shared_memory *)psp->hdcp_context.hdcp_shared_buf; + memset(hdcp_cmd, 0, sizeof(struct ta_hdcp_shared_memory)); + + hdcp_cmd->cmd_id = TA_HDCP_COMMAND__HDCP_GET_SRM; + psp_hdcp_invoke(psp, hdcp_cmd->cmd_id); + + if (hdcp_cmd->hdcp_status != TA_HDCP_STATUS__SUCCESS) + return NULL; + + *srm_version = hdcp_cmd->out_msg.hdcp_get_srm.srm_version; + *srm_size = hdcp_cmd->out_msg.hdcp_get_srm.srm_buf_size; + + + return hdcp_cmd->out_msg.hdcp_get_srm.srm_buf; } static int psp_set_srm(struct psp_context *psp, uint8_t *srm, uint32_t srm_size, uint32_t *srm_version) { + + struct ta_hdcp_shared_memory *hdcp_cmd; + + if (!psp->hdcp_context.hdcp_initialized) { + DRM_WARN("Failed to get hdcp srm. HDCP TA is not initialized."); + return -EINVAL; + } + + hdcp_cmd = (struct ta_hdcp_shared_memory *)psp->hdcp_context.hdcp_shared_buf; + memset(hdcp_cmd, 0, sizeof(struct ta_hdcp_shared_memory)); + + memcpy(hdcp_cmd->in_msg.hdcp_set_srm.srm_buf, srm, srm_size); + hdcp_cmd->in_msg.hdcp_set_srm.srm_buf_size = srm_size; + hdcp_cmd->cmd_id = TA_HDCP_COMMAND__HDCP_SET_SRM; + + psp_hdcp_invoke(psp, hdcp_cmd->cmd_id); + + if (hdcp_cmd->hdcp_status != TA_HDCP_STATUS__SUCCESS || hdcp_cmd->out_msg.hdcp_set_srm.valid_signature != 1 || + hdcp_cmd->out_msg.hdcp_set_srm.srm_version == PSP_SRM_VERSION_MAX) + return -EINVAL; + + *srm_version = hdcp_cmd->out_msg.hdcp_set_srm.srm_version; return 0; } From 2c1b356f69fe9dc46c381882c59c6ec5101b7a90 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 28 Jan 2020 14:39:45 -0500 Subject: [PATCH 16/69] drm/amdgpu/display: handle multiple numbers of fclks in dcn_calcs.c (v2) We might get different numbers of clocks from powerplay depending on what the OEM has populated. v2: add assert for at least one level Bug: https://gitlab.freedesktop.org/drm/amd/issues/963 Reviewed-by: Nicholas Kazlauskas Signed-off-by: Alex Deucher --- .../gpu/drm/amd/display/dc/calcs/dcn_calcs.c | 34 +++++++++++++------ 1 file changed, 23 insertions(+), 11 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c index 22e51b39256e8..304c35ff0e536 100644 --- a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c @@ -1439,6 +1439,7 @@ void dcn_bw_update_from_pplib(struct dc *dc) struct dc_context *ctx = dc->ctx; struct dm_pp_clock_levels_with_voltage fclks = {0}, dcfclks = {0}; bool res; + unsigned vmin0p65_idx, vmid0p72_idx, vnom0p8_idx, vmax0p9_idx; /* TODO: This is not the proper way to obtain fabric_and_dram_bandwidth, should be min(fclk, memclk) */ res = dm_pp_get_clock_levels_by_type_with_voltage( @@ -1450,17 +1451,28 @@ void dcn_bw_update_from_pplib(struct dc *dc) res = verify_clock_values(&fclks); if (res) { - ASSERT(fclks.num_levels >= 3); - dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 = 32 * (fclks.data[0].clocks_in_khz / 1000.0) / 1000.0; - dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72 = dc->dcn_soc->number_of_channels * - (fclks.data[fclks.num_levels - (fclks.num_levels > 2 ? 3 : 2)].clocks_in_khz / 1000.0) - * ddr4_dram_factor_single_Channel / 1000.0; - dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8 = dc->dcn_soc->number_of_channels * - (fclks.data[fclks.num_levels - 2].clocks_in_khz / 1000.0) - * ddr4_dram_factor_single_Channel / 1000.0; - dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = dc->dcn_soc->number_of_channels * - (fclks.data[fclks.num_levels - 1].clocks_in_khz / 1000.0) - * ddr4_dram_factor_single_Channel / 1000.0; + ASSERT(fclks.num_levels); + + vmin0p65_idx = 0; + vmid0p72_idx = fclks.num_levels - + (fclks.num_levels > 2 ? 3 : (fclks.num_levels > 1 ? 2 : 1)); + vnom0p8_idx = fclks.num_levels - (fclks.num_levels > 1 ? 2 : 1); + vmax0p9_idx = fclks.num_levels - 1; + + dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 = + 32 * (fclks.data[vmin0p65_idx].clocks_in_khz / 1000.0) / 1000.0; + dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72 = + dc->dcn_soc->number_of_channels * + (fclks.data[vmid0p72_idx].clocks_in_khz / 1000.0) + * ddr4_dram_factor_single_Channel / 1000.0; + dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8 = + dc->dcn_soc->number_of_channels * + (fclks.data[vnom0p8_idx].clocks_in_khz / 1000.0) + * ddr4_dram_factor_single_Channel / 1000.0; + dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = + dc->dcn_soc->number_of_channels * + (fclks.data[vmax0p9_idx].clocks_in_khz / 1000.0) + * ddr4_dram_factor_single_Channel / 1000.0; } else BREAK_TO_DEBUGGER(); From f26dbc3bff18fcba2771e1afbbc54d086ed2bec1 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Mon, 23 Dec 2019 16:51:42 +0800 Subject: [PATCH 17/69] drm/amdgpu: move xgmi init/fini to xgmi_add/remove_device call For sriov, psp ip block has to be initialized before ih block for the dynamic register programming interface that needed for vf ih ring buffer. On the other hand, current psp ip block hw_init function will initialize xgmi session which actaully depends on interrupt to return session context. This results an empty xgmi ta session id and later failures on all the xgmi ta cmd invoked from vf. xgmi ta session initialization has to be done after ih ip block hw_init call. to unify xgmi session init/fini for both bare-metal sriov virtualization use scenario, move xgmi ta init to xgmi_add_device call, and accordingly terminate xgmi ta session in xgmi_remove_device call. The existing suspend/resume sequence will not be changed. Change-Id: I40776af28e45d2d5ed9f87b28983069c746f2f2e Signed-off-by: Hawking Zhang Reviewed-by: Frank Min --- drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c | 18 ++---------------- drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h | 2 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 11 ++++++++++- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h | 2 +- 4 files changed, 15 insertions(+), 18 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index 3a1570dafe348..939a114605c09 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -558,7 +558,7 @@ int psp_xgmi_invoke(struct psp_context *psp, uint32_t ta_cmd_id) return psp_ta_invoke(psp, ta_cmd_id, psp->xgmi_context.session_id); } -static int psp_xgmi_terminate(struct psp_context *psp) +int psp_xgmi_terminate(struct psp_context *psp) { int ret; @@ -579,7 +579,7 @@ static int psp_xgmi_terminate(struct psp_context *psp) return 0; } -static int psp_xgmi_initialize(struct psp_context *psp) +int psp_xgmi_initialize(struct psp_context *psp) { struct ta_xgmi_shared_memory *xgmi_cmd; int ret; @@ -1420,16 +1420,6 @@ static int psp_load_fw(struct amdgpu_device *adev) return ret; } - if (adev->gmc.xgmi.num_physical_nodes > 1) { - ret = psp_xgmi_initialize(psp); - /* Warning the XGMI seesion initialize failure - * Instead of stop driver initialization - */ - if (ret) - dev_err(psp->adev->dev, - "XGMI: Failed to initialize XGMI session\n"); - } - if (psp->adev->psp.ta_fw) { ret = psp_ras_initialize(psp); if (ret) @@ -1494,10 +1484,6 @@ static int psp_hw_fini(void *handle) void *tmr_buf; void **pptr; - if (adev->gmc.xgmi.num_physical_nodes > 1 && - psp->xgmi_context.initialized == 1) - psp_xgmi_terminate(psp); - if (psp->adev->psp.ta_fw) { psp_ras_terminate(psp); psp_dtm_terminate(psp); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index 611021514c525..c77e1abb538aa 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -362,6 +362,8 @@ int psp_gpu_reset(struct amdgpu_device *adev); int psp_update_vcn_sram(struct amdgpu_device *adev, int inst_idx, uint64_t cmd_gpu_addr, int cmd_size); +int psp_xgmi_initialize(struct psp_context *psp); +int psp_xgmi_terminate(struct psp_context *psp); int psp_xgmi_invoke(struct psp_context *psp, uint32_t ta_cmd_id); int psp_ras_invoke(struct psp_context *psp, uint32_t ta_cmd_id); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 202599fa6253d..c6a1306eb6196 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -365,6 +365,13 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev) return 0; if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP)) { + ret = psp_xgmi_initialize(&adev->psp); + if (ret) { + dev_err(adev->dev, + "XGMI: Failed to initialize xgmi session\n"); + return ret; + } + ret = psp_xgmi_get_hive_id(&adev->psp, &adev->gmc.xgmi.hive_id); if (ret) { dev_err(adev->dev, @@ -451,7 +458,7 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev) return ret; } -void amdgpu_xgmi_remove_device(struct amdgpu_device *adev) +int amdgpu_xgmi_remove_device(struct amdgpu_device *adev) { struct amdgpu_hive_info *hive; @@ -471,6 +478,8 @@ void amdgpu_xgmi_remove_device(struct amdgpu_device *adev) amdgpu_xgmi_sysfs_rem_dev_info(adev, hive); mutex_unlock(&hive->hive_lock); } + + return psp_xgmi_terminate(&adev->psp); } int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h index 74011fbc22519..c62a4acf4c147 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h @@ -40,7 +40,7 @@ struct amdgpu_hive_info { struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev, int lock); int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev); int amdgpu_xgmi_add_device(struct amdgpu_device *adev); -void amdgpu_xgmi_remove_device(struct amdgpu_device *adev); +int amdgpu_xgmi_remove_device(struct amdgpu_device *adev); int amdgpu_xgmi_set_pstate(struct amdgpu_device *adev, int pstate); int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev, struct amdgpu_device *peer_adev); From a86dc760b4aba776ad0f25bec10bf42e9ef145c5 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Sat, 25 Jan 2020 11:51:41 -0500 Subject: [PATCH 18/69] drm/amdgpu/navi: fix index for OD MCLK You can only adjust the max mclk, not the min. Bug: https://gitlab.freedesktop.org/drm/amd/issues/1020 Reviewed-by: Evan Quan Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/powerplay/navi10_ppt.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c index e59cd2c74dc9a..7c3629f644193 100644 --- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c @@ -845,7 +845,7 @@ static int navi10_print_clk_levels(struct smu_context *smu, if (!navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_UCLK_MAX)) break; size += sprintf(buf + size, "OD_MCLK:\n"); - size += sprintf(buf + size, "0: %uMHz\n", od_table->UclkFmax); + size += sprintf(buf + size, "1: %uMHz\n", od_table->UclkFmax); break; case SMU_OD_VDDC_CURVE: if (!smu->od_enabled || !od_table || !od_settings) From c46477b9786d1a3c08ff9f3c75525c65f8cbe3c8 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Sat, 25 Jan 2020 11:27:06 -0500 Subject: [PATCH 19/69] drm/amdgpu/navi10: add OD_RANGE for navi overclocking So users can see the range of valid values. Bug: https://gitlab.freedesktop.org/drm/amd/issues/1020 Reviewed-by: Evan Quan Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/powerplay/navi10_ppt.c | 59 ++++++++++++++++++++++ 1 file changed, 59 insertions(+) diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c index 7c3629f644193..a0a6dc2c190a0 100644 --- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c @@ -739,6 +739,15 @@ static inline bool navi10_od_feature_is_supported(struct smu_11_0_overdrive_tabl return od_table->cap[feature]; } +static void navi10_od_setting_get_range(struct smu_11_0_overdrive_table *od_table, + enum SMU_11_0_ODSETTING_ID setting, + uint32_t *min, uint32_t *max) +{ + if (min) + *min = od_table->min[setting]; + if (max) + *max = od_table->max[setting]; +} static int navi10_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) @@ -757,6 +766,7 @@ static int navi10_print_clk_levels(struct smu_context *smu, OverDriveTable_t *od_table = (OverDriveTable_t *)table_context->overdrive_table; struct smu_11_0_overdrive_table *od_settings = smu->od_settings; + uint32_t min_value, max_value; switch (clk_type) { case SMU_GFXCLK: @@ -869,6 +879,55 @@ static int navi10_print_clk_levels(struct smu_context *smu, } size += sprintf(buf + size, "%d: %uMHz @ %umV\n", i, curve_settings[0], curve_settings[1] / NAVI10_VOLTAGE_SCALE); } + break; + case SMU_OD_RANGE: + if (!smu->od_enabled || !od_table || !od_settings) + break; + size = sprintf(buf, "%s:\n", "OD_RANGE"); + + if (navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_GFXCLK_LIMITS)) { + navi10_od_setting_get_range(od_settings, SMU_11_0_ODSETTING_GFXCLKFMIN, + &min_value, NULL); + navi10_od_setting_get_range(od_settings, SMU_11_0_ODSETTING_GFXCLKFMAX, + NULL, &max_value); + size += sprintf(buf + size, "SCLK: %7uMhz %10uMhz\n", + min_value, max_value); + } + + if (navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_UCLK_MAX)) { + navi10_od_setting_get_range(od_settings, SMU_11_0_ODSETTING_UCLKFMAX, + &min_value, &max_value); + size += sprintf(buf + size, "MCLK: %7uMhz %10uMhz\n", + min_value, max_value); + } + + if (navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_GFXCLK_CURVE)) { + navi10_od_setting_get_range(od_settings, SMU_11_0_ODSETTING_VDDGFXCURVEFREQ_P1, + &min_value, &max_value); + size += sprintf(buf + size, "VDDC_CURVE_SCLK[0]: %7uMhz %10uMhz\n", + min_value, max_value); + navi10_od_setting_get_range(od_settings, SMU_11_0_ODSETTING_VDDGFXCURVEVOLTAGE_P1, + &min_value, &max_value); + size += sprintf(buf + size, "VDDC_CURVE_VOLT[0]: %7dmV %11dmV\n", + min_value, max_value); + navi10_od_setting_get_range(od_settings, SMU_11_0_ODSETTING_VDDGFXCURVEFREQ_P2, + &min_value, &max_value); + size += sprintf(buf + size, "VDDC_CURVE_SCLK[1]: %7uMhz %10uMhz\n", + min_value, max_value); + navi10_od_setting_get_range(od_settings, SMU_11_0_ODSETTING_VDDGFXCURVEVOLTAGE_P2, + &min_value, &max_value); + size += sprintf(buf + size, "VDDC_CURVE_VOLT[1]: %7dmV %11dmV\n", + min_value, max_value); + navi10_od_setting_get_range(od_settings, SMU_11_0_ODSETTING_VDDGFXCURVEFREQ_P3, + &min_value, &max_value); + size += sprintf(buf + size, "VDDC_CURVE_SCLK[2]: %7uMhz %10uMhz\n", + min_value, max_value); + navi10_od_setting_get_range(od_settings, SMU_11_0_ODSETTING_VDDGFXCURVEVOLTAGE_P3, + &min_value, &max_value); + size += sprintf(buf + size, "VDDC_CURVE_VOLT[2]: %7dmV %11dmV\n", + min_value, max_value); + } + break; default: break; From dba7d917031287a73d37f7a42bed89f17a923c23 Mon Sep 17 00:00:00 2001 From: Matt Coffin Date: Sat, 25 Jan 2020 13:04:05 -0500 Subject: [PATCH 20/69] drm/amdgpu/smu_v11_0: Correct behavior of restoring default tables (v2) Previously, the syfs functionality for restoring the default powerplay table was sourcing it's information from the currently-staged powerplay table. This patch adds a step to cache the first overdrive table that we see on boot, so that it can be used later to "restore" the powerplay table v2: sqaush my original with Matt's fix Bug: https://gitlab.freedesktop.org/drm/amd/issues/1020 Signed-off-by: Matt Coffin Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher --- .../gpu/drm/amd/powerplay/inc/amdgpu_smu.h | 1 + drivers/gpu/drm/amd/powerplay/navi10_ppt.c | 7 +++++ drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 6 ++++ drivers/gpu/drm/amd/powerplay/vega20_ppt.c | 28 ++++++------------- 4 files changed, 22 insertions(+), 20 deletions(-) diff --git a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h index 3237eb1ff7082..97b6714e83e67 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +++ b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h @@ -273,6 +273,7 @@ struct smu_table_context uint8_t thermal_controller_type; void *overdrive_table; + void *boot_overdrive_table; }; struct smu_dpm_context { diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c index a0a6dc2c190a0..272e58510d0dd 100644 --- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c @@ -2065,6 +2065,13 @@ static int navi10_od_edit_dpm_table(struct smu_context *smu, enum PP_OD_DPM_TABL return ret; od_table->UclkFmax = input[1]; break; + case PP_OD_RESTORE_DEFAULT_TABLE: + if (!(table_context->overdrive_table && table_context->boot_overdrive_table)) { + pr_err("Overdrive table was not initialized!\n"); + return -EINVAL; + } + memcpy(table_context->overdrive_table, table_context->boot_overdrive_table, sizeof(OverDriveTable_t)); + break; case PP_OD_COMMIT_DPM_TABLE: navi10_dump_od_table(od_table); ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, (void *)od_table, true); diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c index 3924c8a550e1e..28071eea4dc41 100644 --- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c +++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c @@ -1885,6 +1885,12 @@ int smu_v11_0_set_default_od_settings(struct smu_context *smu, bool initialize, pr_err("Failed to export overdrive table!\n"); return ret; } + if (!table_context->boot_overdrive_table) { + table_context->boot_overdrive_table = kmemdup(table_context->overdrive_table, overdrive_table_size, GFP_KERNEL); + if (!table_context->boot_overdrive_table) { + return -ENOMEM; + } + } } ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, table_context->overdrive_table, true); if (ret) { diff --git a/drivers/gpu/drm/amd/powerplay/vega20_ppt.c b/drivers/gpu/drm/amd/powerplay/vega20_ppt.c index 38febd5ca4daa..4ad8d6c14ee5e 100644 --- a/drivers/gpu/drm/amd/powerplay/vega20_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/vega20_ppt.c @@ -1706,22 +1706,11 @@ static int vega20_set_default_od_settings(struct smu_context *smu, struct smu_table_context *table_context = &smu->smu_table; int ret; - if (initialize) { - if (table_context->overdrive_table) - return -EINVAL; - - table_context->overdrive_table = kzalloc(sizeof(OverDriveTable_t), GFP_KERNEL); - - if (!table_context->overdrive_table) - return -ENOMEM; - - ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, - table_context->overdrive_table, false); - if (ret) { - pr_err("Failed to export over drive table!\n"); - return ret; - } + ret = smu_v11_0_set_default_od_settings(smu, initialize, sizeof(OverDriveTable_t)); + if (ret) + return ret; + if (initialize) { ret = vega20_set_default_od8_setttings(smu); if (ret) return ret; @@ -2778,12 +2767,11 @@ static int vega20_odn_edit_dpm_table(struct smu_context *smu, break; case PP_OD_RESTORE_DEFAULT_TABLE: - ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, table_context->overdrive_table, false); - if (ret) { - pr_err("Failed to export over drive table!\n"); - return ret; + if (!(table_context->overdrive_table && table_context->boot_overdrive_table)) { + pr_err("Overdrive table was not initialized!\n"); + return -EINVAL; } - + memcpy(table_context->overdrive_table, table_context->boot_overdrive_table, sizeof(OverDriveTable_t)); break; case PP_OD_COMMIT_DPM_TABLE: From 5cda3d8655fc746e9b7722beae1bdc527ad55d7f Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Sat, 25 Jan 2020 13:30:45 -0500 Subject: [PATCH 21/69] drm/amdgpu: fetch default VDDC curve voltages (v2) Ask the SMU for the default VDDC curve voltage values. This properly reports the VDDC values in the OD interface. v2: only update if the original values are 0 Bug: https://gitlab.freedesktop.org/drm/amd/issues/1020 Reviewed-by: Evan Quan Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/powerplay/navi10_ppt.c | 50 +++++++++++++++++++++- 1 file changed, 49 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c index 272e58510d0dd..2c8c4cbce5488 100644 --- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c @@ -121,6 +121,8 @@ static struct smu_11_0_cmn2aisc_mapping navi10_message_map[SMU_MSG_MAX_COUNT] = MSG_MAP(ArmD3, PPSMC_MSG_ArmD3), MSG_MAP(DAL_DISABLE_DUMMY_PSTATE_CHANGE,PPSMC_MSG_DALDisableDummyPstateChange), MSG_MAP(DAL_ENABLE_DUMMY_PSTATE_CHANGE, PPSMC_MSG_DALEnableDummyPstateChange), + MSG_MAP(GetVoltageByDpm, PPSMC_MSG_GetVoltageByDpm), + MSG_MAP(GetVoltageByDpmOverdrive, PPSMC_MSG_GetVoltageByDpmOverdrive), }; static struct smu_11_0_cmn2aisc_mapping navi10_clk_map[SMU_CLK_COUNT] = { @@ -1934,6 +1936,28 @@ static int navi10_od_setting_check_range(struct smu_11_0_overdrive_table *od_tab return 0; } +static int navi10_overdrive_get_gfx_clk_base_voltage(struct smu_context *smu, + uint16_t *voltage, + uint32_t freq) +{ + uint32_t param = (freq & 0xFFFF) | (PPCLK_GFXCLK << 16); + uint32_t value = 0; + int ret; + + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_GetVoltageByDpm, + param); + if (ret) { + pr_err("[GetBaseVoltage] failed to get GFXCLK AVFS voltage from SMU!"); + return ret; + } + + smu_read_smc_arg(smu, &value); + *voltage = (uint16_t)value; + + return 0; +} + static int navi10_setup_od_limits(struct smu_context *smu) { struct smu_11_0_overdrive_table *overdrive_table = NULL; struct smu_11_0_powerplay_table *powerplay_table = NULL; @@ -1960,16 +1984,40 @@ static int navi10_set_default_od_settings(struct smu_context *smu, bool initiali if (ret) return ret; + od_table = (OverDriveTable_t *)smu->smu_table.overdrive_table; if (initialize) { ret = navi10_setup_od_limits(smu); if (ret) { pr_err("Failed to retrieve board OD limits\n"); return ret; } + if (od_table) { + if (!od_table->GfxclkVolt1) { + ret = navi10_overdrive_get_gfx_clk_base_voltage(smu, + &od_table->GfxclkVolt1, + od_table->GfxclkFreq1); + if (ret) + od_table->GfxclkVolt1 = 0; + } + + if (!od_table->GfxclkVolt2) { + ret = navi10_overdrive_get_gfx_clk_base_voltage(smu, + &od_table->GfxclkVolt2, + od_table->GfxclkFreq2); + if (ret) + od_table->GfxclkVolt2 = 0; + } + if (!od_table->GfxclkVolt3) { + ret = navi10_overdrive_get_gfx_clk_base_voltage(smu, + &od_table->GfxclkVolt3, + od_table->GfxclkFreq3); + if (ret) + od_table->GfxclkVolt3 = 0; + } + } } - od_table = (OverDriveTable_t *)smu->smu_table.overdrive_table; if (od_table) { navi10_dump_od_table(od_table); } From 5f2113185ef072b3df6eb567f34aefefe6c25fdc Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 28 Jan 2020 13:19:51 -0500 Subject: [PATCH 22/69] drm/amdgpu/smu10: fix smu10_get_clock_by_type_with_latency Only send non-0 clocks to DC for validation. This mirrors what the windows driver does. Bug: https://gitlab.freedesktop.org/drm/amd/issues/963 Reviewed-by: Evan Quan Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c index 4e8ab139bb3bd..273126cfc37de 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c @@ -1026,12 +1026,15 @@ static int smu10_get_clock_by_type_with_latency(struct pp_hwmgr *hwmgr, clocks->num_levels = 0; for (i = 0; i < pclk_vol_table->count; i++) { - clocks->data[i].clocks_in_khz = pclk_vol_table->entries[i].clk * 10; - clocks->data[i].latency_in_us = latency_required ? - smu10_get_mem_latency(hwmgr, - pclk_vol_table->entries[i].clk) : - 0; - clocks->num_levels++; + if (pclk_vol_table->entries[i].clk) { + clocks->data[clocks->num_levels].clocks_in_khz = + pclk_vol_table->entries[i].clk * 10; + clocks->data[clocks->num_levels].latency_in_us = latency_required ? + smu10_get_mem_latency(hwmgr, + pclk_vol_table->entries[i].clk) : + 0; + clocks->num_levels++; + } } return 0; From 0c87d6981cf486b1e89794f4565bdd1b851e36f4 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 29 Jan 2020 12:42:57 -0500 Subject: [PATCH 23/69] drm/amdgpu/smu10: fix smu10_get_clock_by_type_with_voltage Cull out 0 clocks to avoid a warning in DC. Bug: https://gitlab.freedesktop.org/drm/amd/issues/963 Reviewed-by: Evan Quan Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c index 273126cfc37de..689072a312a7f 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c @@ -1080,9 +1080,11 @@ static int smu10_get_clock_by_type_with_voltage(struct pp_hwmgr *hwmgr, clocks->num_levels = 0; for (i = 0; i < pclk_vol_table->count; i++) { - clocks->data[i].clocks_in_khz = pclk_vol_table->entries[i].clk * 10; - clocks->data[i].voltage_in_mv = pclk_vol_table->entries[i].vol; - clocks->num_levels++; + if (pclk_vol_table->entries[i].clk) { + clocks->data[clocks->num_levels].clocks_in_khz = pclk_vol_table->entries[i].clk * 10; + clocks->data[clocks->num_levels].voltage_in_mv = pclk_vol_table->entries[i].vol; + clocks->num_levels++; + } } return 0; From 5aa0e19769df2db6c7f5762a3ddd8426c93e7853 Mon Sep 17 00:00:00 2001 From: John Clements Date: Fri, 17 Jan 2020 12:11:26 +0800 Subject: [PATCH 24/69] Revert "drm/amdgpu: added support to get mGPU DRAM base" This reverts commit 12c740dcec0936a92b96b34f94240a75207c1796. Reverting change to accidental overwrite of another change Signed-off-by: John Clements --- drivers/gpu/drm/amd/amdgpu/amdgpu_df.h | 3 - drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 20 ------- drivers/gpu/drm/amd/amdgpu/df_v3_6.c | 59 +------------------ .../amd/include/asic_reg/df/df_3_6_offset.h | 3 - .../amd/include/asic_reg/df/df_3_6_sh_mask.h | 8 --- 5 files changed, 1 insertion(+), 92 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h index 057f6ea645d7e..61a26c15c8dd5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h @@ -52,9 +52,6 @@ struct amdgpu_df_funcs { uint64_t (*get_fica)(struct amdgpu_device *adev, uint32_t ficaa_val); void (*set_fica)(struct amdgpu_device *adev, uint32_t ficaa_val, uint32_t ficadl_val, uint32_t ficadh_val); - uint64_t (*get_dram_base_addr)(struct amdgpu_device *adev, - uint32_t df_inst); - uint32_t (*get_df_inst_id)(struct amdgpu_device *adev); }; struct amdgpu_df { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 87f57878dbd50..92e9c73a68204 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -742,20 +742,6 @@ int amdgpu_ras_error_query(struct amdgpu_device *adev, return 0; } -uint64_t get_xgmi_relative_phy_addr(struct amdgpu_device *adev, uint64_t addr) -{ - uint32_t df_inst_id; - - if ((!adev->df.funcs) || - (!adev->df.funcs->get_df_inst_id) || - (!adev->df.funcs->get_dram_base_addr)) - return addr; - - df_inst_id = adev->df.funcs->get_df_inst_id(adev); - - return addr + adev->df.funcs->get_dram_base_addr(adev, df_inst_id); -} - /* wrapper of psp_ras_trigger_error */ int amdgpu_ras_error_inject(struct amdgpu_device *adev, struct ras_inject_if *info) @@ -773,12 +759,6 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, if (!obj) return -EINVAL; - /* Calculate XGMI relative offset */ - if (adev->gmc.xgmi.num_physical_nodes > 1) { - block_info.address = get_xgmi_relative_phy_addr(adev, - block_info.address); - } - switch (info->head.block) { case AMDGPU_RAS_BLOCK__GFX: if (adev->gfx.funcs->ras_error_inject) diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c index d8cfd7f53e1fa..f51326598a8c0 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c @@ -27,9 +27,6 @@ #include "df/df_3_6_offset.h" #include "df/df_3_6_sh_mask.h" -#define DF_3_6_SMN_REG_INST_DIST 0x8 -#define DF_3_6_INST_CNT 8 - static u32 df_v3_6_channel_number[] = {1, 2, 0, 4, 0, 8, 0, 16, 32, 0, 0, 0, 2, 4, 8}; @@ -686,58 +683,6 @@ static void df_v3_6_pmc_get_count(struct amdgpu_device *adev, } } -static uint64_t df_v3_6_get_dram_base_addr(struct amdgpu_device *adev, - uint32_t df_inst) -{ - uint32_t base_addr_reg_val = 0; - uint64_t base_addr = 0; - - base_addr_reg_val = RREG32_PCIE(smnDF_CS_UMC_AON0_DramBaseAddress0 + - df_inst * DF_3_6_SMN_REG_INST_DIST); - - if (REG_GET_FIELD(base_addr_reg_val, - DF_CS_UMC_AON0_DramBaseAddress0, - AddrRngVal) == 0) { - DRM_WARN("address range not valid"); - return 0; - } - - base_addr = REG_GET_FIELD(base_addr_reg_val, - DF_CS_UMC_AON0_DramBaseAddress0, - DramBaseAddr); - - return base_addr << 28; -} - -static uint32_t df_v3_6_get_df_inst_id(struct amdgpu_device *adev) -{ - uint32_t xgmi_node_id = 0; - uint32_t df_inst_id = 0; - - /* Walk through DF dst nodes to find current XGMI node */ - for (df_inst_id = 0; df_inst_id < DF_3_6_INST_CNT; df_inst_id++) { - - xgmi_node_id = RREG32_PCIE(smnDF_CS_UMC_AON0_DramLimitAddress0 + - df_inst_id * DF_3_6_SMN_REG_INST_DIST); - xgmi_node_id = REG_GET_FIELD(xgmi_node_id, - DF_CS_UMC_AON0_DramLimitAddress0, - DstFabricID); - - /* TODO: establish reason dest fabric id is offset by 7 */ - xgmi_node_id = xgmi_node_id >> 7; - - if (adev->gmc.xgmi.physical_node_id == xgmi_node_id) - break; - } - - if (df_inst_id == DF_3_6_INST_CNT) { - DRM_WARN("cant match df dst id with gpu node"); - return 0; - } - - return df_inst_id; -} - const struct amdgpu_df_funcs df_v3_6_funcs = { .sw_init = df_v3_6_sw_init, .sw_fini = df_v3_6_sw_fini, @@ -751,7 +696,5 @@ const struct amdgpu_df_funcs df_v3_6_funcs = { .pmc_stop = df_v3_6_pmc_stop, .pmc_get_count = df_v3_6_pmc_get_count, .get_fica = df_v3_6_get_fica, - .set_fica = df_v3_6_set_fica, - .get_dram_base_addr = df_v3_6_get_dram_base_addr, - .get_df_inst_id = df_v3_6_get_df_inst_id + .set_fica = df_v3_6_set_fica }; diff --git a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h index bb2c9c7a18dff..87c84691b5be9 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h +++ b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h @@ -71,7 +71,4 @@ #define smnDF_PIE_AON_FabricIndirectConfigAccessDataLo3 0x1d098UL #define smnDF_PIE_AON_FabricIndirectConfigAccessDataHi3 0x1d09cUL -#define smnDF_CS_UMC_AON0_DramBaseAddress0 0x1c110UL -#define smnDF_CS_UMC_AON0_DramLimitAddress0 0x1c114UL - #endif diff --git a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h index 7afa87c7ff543..65e9f756e86e0 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h +++ b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h @@ -53,12 +53,4 @@ #define DF_CS_UMC_AON0_DramBaseAddress0__IntLvAddrSel_MASK 0x00000E00L #define DF_CS_UMC_AON0_DramBaseAddress0__DramBaseAddr_MASK 0xFFFFF000L -//DF_CS_UMC_AON0_DramLimitAddress0 -#define DF_CS_UMC_AON0_DramLimitAddress0__DstFabricID__SHIFT 0x0 -#define DF_CS_UMC_AON0_DramLimitAddress0__AllowReqIO__SHIFT 0xa -#define DF_CS_UMC_AON0_DramLimitAddress0__DramLimitAddr__SHIFT 0xc -#define DF_CS_UMC_AON0_DramLimitAddress0__DstFabricID_MASK 0x000003FFL -#define DF_CS_UMC_AON0_DramLimitAddress0__AllowReqIO_MASK 0x00000400L -#define DF_CS_UMC_AON0_DramLimitAddress0__DramLimitAddr_MASK 0xFFFFF000L - #endif From 85e45949ac2849957e94f20c016d0e28b1d03db3 Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Wed, 5 Feb 2020 10:26:21 +0800 Subject: [PATCH 25/69] Revert "drm/ttm: flush the fence on the bo after we individualize the reservation object" MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This reverts commit 2ef084f6e3253a2b4047fbbc798089bba0a2e1c6. The commit: “2ef084f6e325 drm/ttm: flush the fence on the bo after we individualize the reservation object“ is to fix the large VRAM oOo issue, but introduced some other side effect, so revert it. Change-Id: I9cd81635b0e9a5a05fb8e48142ceb9dd16ca0224 Signed-off-by: Stanley.Yang --- drivers/gpu/drm/ttm/ttm_bo.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index dde14cf427c2f..ee1da57f3b407 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -509,10 +509,8 @@ static void ttm_bo_cleanup_refs_or_queue(struct ttm_buffer_object *bo) dma_resv_unlock(amdkcl_ttm_resvp(bo)); } - if (amdkcl_ttm_resvp(bo) != &amdkcl_ttm_resv(bo)) { - ttm_bo_flush_all_fences(bo); + if (amdkcl_ttm_resvp(bo) != &amdkcl_ttm_resv(bo)) dma_resv_unlock(&amdkcl_ttm_resv(bo)); - } error: kref_get(&bo->list_kref); From ea104b6913a08d697dcc7b8bb4a3f0f4b81f7841 Mon Sep 17 00:00:00 2001 From: Nirmoy Das Date: Tue, 4 Feb 2020 15:46:07 +0100 Subject: [PATCH 26/69] drm/amdgpu: fix empty return on non-void function This fixes empty return on non-void function, amdgpu_xgmi_remove_device Fixes: b80574252499e (drm/amdgpu: move xgmi init/fini to xgmi_add/remove_device call) Signed-off-by: Nirmoy Das Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index c6a1306eb6196..fbc41926de08f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -463,11 +463,11 @@ int amdgpu_xgmi_remove_device(struct amdgpu_device *adev) struct amdgpu_hive_info *hive; if (!adev->gmc.xgmi.supported) - return; + return -EINVAL; hive = amdgpu_get_xgmi_hive(adev, 1); if (!hive) - return; + return -EINVAL; if (!(hive->number_devices--)) { amdgpu_xgmi_sysfs_destroy(adev, hive); From 4cfd764b4507ee84484b38367611d544ba88060e Mon Sep 17 00:00:00 2001 From: Amber Lin Date: Thu, 30 Jan 2020 00:08:05 -0500 Subject: [PATCH 27/69] drm/amdkfd: Add queue information to sysfs Provide compute queues information in sysfs under /sys/class/kfd/kfd/proc. The format is /sys/class/kfd/kfd/proc//queues//XX where XX are size, type, and gpuid three files to represent queue size, queue type, and the GPU this queue uses. folder and files underneath are generated when a queue is created. They are removed when the queue is destroyed. Change-Id: Id7f2d7a11ce6d465c121624f34be1777d0897a94 Signed-off-by: Amber Lin Reviewed-by: Felix Kuehling --- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 7 ++ drivers/gpu/drm/amd/amdkfd/kfd_process.c | 90 +++++++++++++++++++ .../amd/amdkfd/kfd_process_queue_manager.c | 2 + 3 files changed, 99 insertions(+) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index a2c47392ef0a7..82b3c31e99f8a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -43,6 +43,7 @@ #include #include #include +#include #include "amd_shared.h" @@ -570,6 +571,9 @@ struct queue { struct kfd_process *process; struct kfd_dev *device; void *gws; + + /* procfs */ + struct kobject kobj; }; /* @@ -864,6 +868,7 @@ struct kfd_process { /* Kobj for our procfs */ struct kobject *kobj; + struct kobject *kobj_queues; struct attribute attr_pasid; /* Keep track cwsr init */ @@ -986,6 +991,8 @@ extern struct device *kfd_device; /* KFD's procfs */ void kfd_procfs_init(void); void kfd_procfs_shutdown(void); +int kfd_procfs_add_queue(struct queue *q); +void kfd_procfs_del_queue(struct queue *q); /* Topology */ int kfd_topology_init(void); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index d5d4660221af9..08960008d8c8e 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -148,6 +148,88 @@ void kfd_procfs_shutdown(void) } } +static ssize_t kfd_procfs_queue_show(struct kobject *kobj, + struct attribute *attr, char *buffer) +{ + struct queue *q = container_of(kobj, struct queue, kobj); + + if (!strcmp(attr->name, "size")) + return snprintf(buffer, PAGE_SIZE, "%llu", + q->properties.queue_size); + else if (!strcmp(attr->name, "type")) + return snprintf(buffer, PAGE_SIZE, "%d", q->properties.type); + else if (!strcmp(attr->name, "gpuid")) + return snprintf(buffer, PAGE_SIZE, "%u", q->device->id); + else + pr_err("Invalid attribute"); + + return 0; +} + +static struct attribute attr_queue_size = { + .name = "size", + .mode = KFD_SYSFS_FILE_MODE +}; + +static struct attribute attr_queue_type = { + .name = "type", + .mode = KFD_SYSFS_FILE_MODE +}; + +static struct attribute attr_queue_gpuid = { + .name = "gpuid", + .mode = KFD_SYSFS_FILE_MODE +}; + +static struct attribute *procfs_queue_attrs[] = { + &attr_queue_size, + &attr_queue_type, + &attr_queue_gpuid, + NULL +}; + +static const struct sysfs_ops procfs_queue_ops = { + .show = kfd_procfs_queue_show, +}; + +static struct kobj_type procfs_queue_type = { + .sysfs_ops = &procfs_queue_ops, + .default_attrs = procfs_queue_attrs, +}; + +int kfd_procfs_add_queue(struct queue *q) +{ + struct kfd_process *proc; + int ret; + + if (!q || !q->process) + return -EINVAL; + proc = q->process; + + /* Create proc//queues/ folder */ + if (!proc->kobj_queues) + return -EFAULT; + ret = kobject_init_and_add(&q->kobj, &procfs_queue_type, + proc->kobj_queues, "%u", q->properties.queue_id); + if (ret < 0) { + pr_warn("Creating proc//queues/%u failed", + q->properties.queue_id); + kobject_put(&q->kobj); + return ret; + } + + return 0; +} + +void kfd_procfs_del_queue(struct queue *q) +{ + if (!q) + return; + + kobject_del(&q->kobj); + kobject_put(&q->kobj); +} + int kfd_process_create_wq(void) { if (!kfd_process_wq) @@ -348,6 +430,11 @@ struct kfd_process *kfd_create_process(struct task_struct *thread) if (ret) pr_warn("Creating pasid for pid %d failed", (int)process->lead_thread->pid); + + process->kobj_queues = kobject_create_and_add("queues", + process->kobj); + if (!process->kobj_queues) + pr_warn("Creating KFD proc/queues folder failed"); } out: if (!IS_ERR(process)) @@ -510,6 +597,9 @@ static void kfd_process_wq_release(struct work_struct *work) /* Remove the procfs files */ if (p->kobj) { sysfs_remove_file(p->kobj, &p->attr_pasid); + kobject_del(p->kobj_queues); + kobject_put(p->kobj_queues); + p->kobj_queues = NULL; kobject_del(p->kobj); kobject_put(p->kobj); p->kobj = NULL; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c index 92ab625c1d345..cf7554971b366 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c @@ -323,6 +323,7 @@ int pqm_create_queue(struct process_queue_manager *pqm, if (q) { pr_debug("PQM done creating queue\n"); q->properties.is_new = true; + kfd_procfs_add_queue(q); print_queue_properties(&q->properties); } @@ -379,6 +380,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid) } if (pqn->q) { + kfd_procfs_del_queue(pqn->q); dqm = pqn->q->device->dqm; retval = dqm->ops.destroy_queue(dqm, &pdd->qpd, pqn->q); if (retval) { From 49935167b5337429ddd46a0e667f28add69de69e Mon Sep 17 00:00:00 2001 From: Wyatt Wood Date: Thu, 9 Jan 2020 14:29:13 -0500 Subject: [PATCH 28/69] drm/amd/display: Add set psr version message [Why] Must know psr version during runtime. [How] Add set psr version message structures. Signed-off-by: Wyatt Wood Reviewed-by: Nicholas Kazlauskas Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 6 +- drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c | 62 +++++++++++++------ drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h | 9 +-- .../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 10 +-- .../drm/amd/display/dmub/inc/dmub_cmd_dal.h | 4 +- 5 files changed, 57 insertions(+), 34 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c index 64223e7551bd6..8bbe45bfb12ad 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c @@ -2408,7 +2408,7 @@ bool dc_link_set_psr_allow_active(struct dc_link *link, bool allow_active, bool struct dmub_psr *psr = dc->res_pool->psr; if ((psr != NULL) && link->psr_feature_enabled) - psr->funcs->set_psr_enable(psr, allow_active); + psr->funcs->psr_enable(psr, allow_active); else if ((dmcu != NULL && dmcu->funcs->is_dmcu_initialized(dmcu)) && link->psr_feature_enabled) dmcu->funcs->set_psr_enable(dmcu, allow_active, wait); @@ -2424,7 +2424,7 @@ bool dc_link_get_psr_state(const struct dc_link *link, uint32_t *psr_state) struct dmub_psr *psr = dc->res_pool->psr; if (psr != NULL && link->psr_feature_enabled) - psr->funcs->get_psr_state(psr_state); + psr->funcs->psr_get_state(psr_state); else if (dmcu != NULL && link->psr_feature_enabled) dmcu->funcs->get_psr_state(dmcu, psr_state); @@ -2596,7 +2596,7 @@ bool dc_link_setup_psr(struct dc_link *link, psr_context->frame_delay = 0; if (psr) - link->psr_feature_enabled = psr->funcs->setup_psr(psr, link, psr_context); + link->psr_feature_enabled = psr->funcs->psr_copy_settings(psr, link, psr_context); else link->psr_feature_enabled = dmcu->funcs->setup_psr(dmcu, link, psr_context); diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c index 225955ec6d392..bdf80b09277e6 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c @@ -36,16 +36,39 @@ /** * Get PSR state from firmware. */ -static void dmub_get_psr_state(uint32_t *psr_state) +static void dmub_psr_get_state(uint32_t *psr_state) { // Not yet implemented // Trigger GPINT interrupt from firmware } +static void dmub_psr_set_version(struct dmub_psr *dmub, struct dc_stream_state *stream) +{ + //stream->psr_version; + union dmub_rb_cmd cmd; + struct dc_context *dc = dmub->ctx; + + cmd.psr_set_version.header.type = DMUB_CMD__PSR; + cmd.psr_set_version.header.sub_type = DMUB_CMD__PSR_SET_VERSION; + + if (stream->psr_version == 0x0) + return; + else if (stream->psr_version == 0x1) + cmd.psr_set_version.psr_set_version_data.version = PSR_VERSION_1; + else if (stream->psr_version == 0x2) + cmd.psr_set_version.psr_set_version_data.version = PSR_VERSION_2; + + cmd.psr_enable.header.payload_bytes = sizeof(struct dmub_cmd_psr_set_version_data); + + dc_dmub_srv_cmd_queue(dc->dmub_srv, &cmd.psr_enable.header); + dc_dmub_srv_cmd_execute(dc->dmub_srv); + dc_dmub_srv_wait_idle(dc->dmub_srv); +} + /** * Enable/Disable PSR. */ -static void dmub_set_psr_enable(struct dmub_psr *dmub, bool enable) +static void dmub_psr_enable(struct dmub_psr *dmub, bool enable) { union dmub_rb_cmd cmd; struct dc_context *dc = dmub->ctx; @@ -67,13 +90,13 @@ static void dmub_set_psr_enable(struct dmub_psr *dmub, bool enable) /** * Set PSR level. */ -static void dmub_set_psr_level(struct dmub_psr *dmub, uint16_t psr_level) +static void dmub_psr_set_level(struct dmub_psr *dmub, uint16_t psr_level) { union dmub_rb_cmd cmd; uint32_t psr_state = 0; struct dc_context *dc = dmub->ctx; - dmub_get_psr_state(&psr_state); + dmub_psr_get_state(&psr_state); if (psr_state == 0) return; @@ -91,7 +114,7 @@ static void dmub_set_psr_level(struct dmub_psr *dmub, uint16_t psr_level) /** * Setup PSR by programming phy registers and sending psr hw context values to firmware. */ -static bool dmub_setup_psr(struct dmub_psr *dmub, +static bool dmub_psr_copy_settings(struct dmub_psr *dmub, struct dc_link *link, struct psr_context *psr_context) { @@ -104,18 +127,16 @@ static bool dmub_setup_psr(struct dmub_psr *dmub, for (int i = 0; i < MAX_PIPES; i++) { if (res_ctx && - res_ctx->pipe_ctx[i].stream && - res_ctx->pipe_ctx[i].stream->link && - res_ctx->pipe_ctx[i].stream->link == link && - res_ctx->pipe_ctx[i].stream->link->connector_signal == SIGNAL_TYPE_EDP) { + res_ctx->pipe_ctx[i].stream && + res_ctx->pipe_ctx[i].stream->link && + res_ctx->pipe_ctx[i].stream->link == link && + res_ctx->pipe_ctx[i].stream->link->connector_signal == SIGNAL_TYPE_EDP) { pipe_ctx = &res_ctx->pipe_ctx[i]; break; } } - if (!pipe_ctx || - !&pipe_ctx->plane_res || - !&pipe_ctx->stream_res) + if (!pipe_ctx || !&pipe_ctx->plane_res || !&pipe_ctx->stream_res) return false; // Program DP DPHY fast training registers @@ -161,13 +182,13 @@ static bool dmub_setup_psr(struct dmub_psr *dmub, copy_settings_data->hyst_lines = psr_context->hyst_lines; copy_settings_data->phy_type = psr_context->phyType; copy_settings_data->aux_repeat = psr_context->aux_repeats; - copy_settings_data->smu_optimizations_en = psr_context->allow_smu_optimizations; - copy_settings_data->skip_wait_for_pll_lock = psr_context->skipPsrWaitForPllLock; + copy_settings_data->smu_optimizations_en = psr_context->allow_smu_optimizations; + copy_settings_data->skip_wait_for_pll_lock = psr_context->skipPsrWaitForPllLock; copy_settings_data->frame_delay = psr_context->frame_delay; copy_settings_data->smu_phy_id = psr_context->smuPhyId; - copy_settings_data->num_of_controllers = psr_context->numberOfControllers; + copy_settings_data->num_of_controllers = psr_context->numberOfControllers; copy_settings_data->frame_cap_ind = psr_context->psrFrameCaptureIndicationReq; - copy_settings_data->phy_num = psr_context->frame_delay & 0x7; + copy_settings_data->phy_num = psr_context->frame_delay & 0x7; copy_settings_data->link_rate = psr_context->frame_delay & 0xF; dc_dmub_srv_cmd_queue(dc->dmub_srv, &cmd.psr_copy_settings.header); @@ -178,10 +199,11 @@ static bool dmub_setup_psr(struct dmub_psr *dmub, } static const struct dmub_psr_funcs psr_funcs = { - .set_psr_enable = dmub_set_psr_enable, - .setup_psr = dmub_setup_psr, - .get_psr_state = dmub_get_psr_state, - .set_psr_level = dmub_set_psr_level, + .psr_set_version = dmub_psr_set_version, + .psr_copy_settings = dmub_psr_copy_settings, + .psr_enable = dmub_psr_enable, + .psr_get_state = dmub_psr_get_state, + .psr_set_level = dmub_psr_set_level, }; /** diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h index 229958de3035b..4fe8b53d9551d 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h +++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.h @@ -34,10 +34,11 @@ struct dmub_psr { }; struct dmub_psr_funcs { - void (*set_psr_enable)(struct dmub_psr *dmub, bool enable); - bool (*setup_psr)(struct dmub_psr *dmub, struct dc_link *link, struct psr_context *psr_context); - void (*get_psr_state)(uint32_t *psr_state); - void (*set_psr_level)(struct dmub_psr *dmub, uint16_t psr_level); + void (*psr_set_version)(struct dmub_psr *dmub, struct dc_stream_state *stream); + bool (*psr_copy_settings)(struct dmub_psr *dmub, struct dc_link *link, struct psr_context *psr_context); + void (*psr_enable)(struct dmub_psr *dmub, bool enable); + void (*psr_get_state)(uint32_t *psr_state); + void (*psr_set_level)(struct dmub_psr *dmub, uint16_t psr_level); }; struct dmub_psr *dmub_psr_create(struct dc_context *ctx); diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h index cd9532b4f14df..8f518e7d8b4ec 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h @@ -257,13 +257,13 @@ struct dmub_rb_cmd_psr_enable { struct dmub_cmd_header header; }; -struct dmub_cmd_psr_setup_data { +struct dmub_cmd_psr_set_version_data { enum psr_version version; // PSR version 1 or 2 }; -struct dmub_rb_cmd_psr_setup { +struct dmub_rb_cmd_psr_set_version { struct dmub_cmd_header header; - struct dmub_cmd_psr_setup_data psr_setup_data; + struct dmub_cmd_psr_set_version_data psr_set_version_data; }; union dmub_rb_cmd { @@ -277,11 +277,11 @@ union dmub_rb_cmd { struct dmub_rb_cmd_enable_disp_power_gating enable_disp_power_gating; struct dmub_rb_cmd_dpphy_init dpphy_init; struct dmub_rb_cmd_dig1_transmitter_control dig1_transmitter_control; - struct dmub_rb_cmd_psr_enable psr_enable; + struct dmub_rb_cmd_psr_set_version psr_set_version; struct dmub_rb_cmd_psr_copy_settings psr_copy_settings; + struct dmub_rb_cmd_psr_enable psr_enable; struct dmub_rb_cmd_psr_set_level psr_set_level; struct dmub_rb_cmd_PLAT_54186_wa PLAT_54186_wa; - struct dmub_rb_cmd_psr_setup psr_setup; }; #pragma pack(pop) diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd_dal.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd_dal.h index 7b69eb37f762c..ce793f47f234c 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd_dal.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd_dal.h @@ -32,7 +32,7 @@ */ enum dmub_cmd_psr_type { - DMUB_CMD__PSR_SETUP = 0, + DMUB_CMD__PSR_SET_VERSION = 0, DMUB_CMD__PSR_COPY_SETTINGS = 1, DMUB_CMD__PSR_ENABLE = 2, DMUB_CMD__PSR_DISABLE = 3, @@ -42,7 +42,7 @@ enum dmub_cmd_psr_type { enum psr_version { PSR_VERSION_1 = 0x10, // PSR Version 1 PSR_VERSION_2 = 0x20, // PSR Version 2, includes selective update - PSR_VERSION_2_Y_COORD = 0x21, // PSR Version 2, includes Y-coordinate support for SU + PSR_VERSION_2_1 = 0x21, // PSR Version 2, includes Y-coordinate support for SU }; #endif /* _DMUB_CMD_DAL_H_ */ From e5eafcba165e1f1e4005d71170821a53a81c78fb Mon Sep 17 00:00:00 2001 From: George Shen Date: Tue, 14 Jan 2020 17:11:15 -0500 Subject: [PATCH 29/69] drm/amd/display: Move USB-C workaround to after parameter variables are set [Why] The call to dp_enable_link_phy are using default/invalid values for clock id and link settings. [How] Move workaround code to after its parameter variables are determined. Signed-off-by: George Shen Reviewed-by: Tony Cheng Acked-by: Bhawanpreet Lakha --- .../gpu/drm/amd/display/dc/core/dc_link_dp.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c index ddf42667822ff..1587438bffade 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c @@ -1906,6 +1906,16 @@ bool dp_verify_link_cap( /* disable PHY done possible by BIOS, will be done by driver itself */ dp_disable_link_phy(link, link->connector_signal); + dp_cs_id = get_clock_source_id(link); + + /* link training starts with the maximum common settings + * supported by both sink and ASIC. + */ + initial_link_settings = get_common_supported_link_settings( + *known_limit_link_setting, + max_link_cap); + cur_link_setting = initial_link_settings; + /* Temporary Renoir-specific workaround for SWDEV-215184; * PHY will sometimes be in bad state on hotplugging display from certain USB-C dongle, * so add extra cycle of enabling and disabling the PHY before first link training. @@ -1916,15 +1926,6 @@ bool dp_verify_link_cap( dp_disable_link_phy(link, link->connector_signal); } - dp_cs_id = get_clock_source_id(link); - - /* link training starts with the maximum common settings - * supported by both sink and ASIC. - */ - initial_link_settings = get_common_supported_link_settings( - *known_limit_link_setting, - max_link_cap); - cur_link_setting = initial_link_settings; do { skip_video_pattern = true; From c773ed280884a5ea5ea4ec8e1dbc99e8e40fc1d9 Mon Sep 17 00:00:00 2001 From: Joseph Gravenor Date: Wed, 15 Jan 2020 13:27:06 -0500 Subject: [PATCH 30/69] drm/amd/display: remove invalid dc_is_hw_initialized function [why/how] We found out that the register we read actually gets reset by SMU after we loose power, meaning this always returns true Signed-off-by: Joseph Gravenor Reviewed-by: Eric Yang Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc.c | 6 ------ drivers/gpu/drm/amd/display/dc/dc.h | 1 - 2 files changed, 7 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 5f9390d8a3128..135f146a9ff0c 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -1338,12 +1338,6 @@ bool dc_commit_state(struct dc *dc, struct dc_state *context) return (result == DC_OK); } -bool dc_is_hw_initialized(struct dc *dc) -{ - struct dc_bios *dcb = dc->ctx->dc_bios; - return dcb->funcs->is_accelerated_mode(dcb); -} - bool dc_post_update_surfaces_to_stream(struct dc *dc) { int i; diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 4ceb64d50bb3f..f3277eacc5542 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -1125,7 +1125,6 @@ unsigned int dc_get_current_backlight_pwm(struct dc *dc); unsigned int dc_get_target_backlight_pwm(struct dc *dc); bool dc_is_dmcu_initialized(struct dc *dc); -bool dc_is_hw_initialized(struct dc *dc); enum dc_status dc_set_clock(struct dc *dc, enum dc_clock_type clock_type, uint32_t clk_khz, uint32_t stepping); void dc_get_clock(struct dc *dc, enum dc_clock_type clock_type, struct dc_clock_config *clock_cfg); From fa9938c3925254ecb3da10d570185698fb1aeec0 Mon Sep 17 00:00:00 2001 From: Roman Li Date: Tue, 14 Jan 2020 13:56:08 -0500 Subject: [PATCH 31/69] drm/amd/display: Fix psr static frames calculation [Why] Driver crash with psr feature enabled due to divide-by-zero error. This is a regression after rework to calculate static screen frame number entry time. [How] Correct order of operations to avoid divide-by-zero. Signed-off-by: Roman Li Reviewed-by: Zhan Liu Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 464ad24e7b086..86516d9205d03 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -9566,7 +9566,6 @@ bool amdgpu_dm_psr_enable(struct dc_stream_state *stream) /* Calculate number of static frames before generating interrupt to * enter PSR. */ - unsigned int frame_time_microsec = 1000000 / vsync_rate_hz; // Init fail safe of 2 frames static unsigned int num_frames_static = 2; @@ -9581,8 +9580,10 @@ bool amdgpu_dm_psr_enable(struct dc_stream_state *stream) * Calculate number of frames such that at least 30 ms of time has * passed. */ - if (vsync_rate_hz != 0) + if (vsync_rate_hz != 0) { + unsigned int frame_time_microsec = 1000000 / vsync_rate_hz; num_frames_static = (30000 / frame_time_microsec) + 1; + } params.triggers.cursor_update = true; params.triggers.overlay_update = true; From 34741f50e14c8e3fb91becae6dfbebd93f6a98ad Mon Sep 17 00:00:00 2001 From: Sung Lee Date: Wed, 15 Jan 2020 11:55:06 -0500 Subject: [PATCH 32/69] drm/amd/display: Do not set optimized_require to false after plane disable [WHY] The optimized_require flag is needed to set watermarks and clocks lower in certain conditions. This flag is set to true and then set to false while programming front end in dcn20. [HOW] Do not set the flag to false while disabling plane. Signed-off-by: Sung Lee Reviewed-by: Tony Cheng Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c index cce7d4bd8771c..e825ee71d8fc9 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c @@ -578,7 +578,6 @@ void dcn20_plane_atomic_disable(struct dc *dc, struct pipe_ctx *pipe_ctx) dpp->funcs->dpp_dppclk_control(dpp, false, false); hubp->power_gated = true; - dc->optimized_required = false; /* We're powering off, no need to optimize */ hws->funcs.plane_atomic_power_down(dc, pipe_ctx->plane_res.dpp, From 719bb85649a1f5c2edab2df48c51713d2566016e Mon Sep 17 00:00:00 2001 From: Anthony Koo Date: Tue, 14 Jan 2020 16:23:31 -0500 Subject: [PATCH 33/69] drm/amd/display: Split program front end part that occur outside lock [Why] Eventually want to lock at a higher level in stack. To do this, we need to be able to isolate the parts that need to be done after pipe unlock. [How] Split out programming that is done post unlock. Change-Id: I88e04100e14d7483716905103ab264331d58bbde Signed-off-by: Anthony Koo Reviewed-by: Aric Cyr Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc.c | 26 ++++++++++--- .../display/dc/dce110/dce110_hw_sequencer.c | 6 +++ .../amd/display/dc/dcn10/dcn10_hw_sequencer.c | 39 ++++++++++++++++--- .../amd/display/dc/dcn10/dcn10_hw_sequencer.h | 3 ++ .../gpu/drm/amd/display/dc/dcn10/dcn10_init.c | 1 + .../drm/amd/display/dc/dcn20/dcn20_hwseq.c | 11 +++++- .../drm/amd/display/dc/dcn20/dcn20_hwseq.h | 3 ++ .../gpu/drm/amd/display/dc/dcn20/dcn20_init.c | 1 + .../gpu/drm/amd/display/dc/dcn21/dcn21_init.c | 1 + .../gpu/drm/amd/display/dc/inc/hw_sequencer.h | 2 + 10 files changed, 80 insertions(+), 13 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 135f146a9ff0c..fee424fadca0e 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -802,13 +802,17 @@ static void disable_dangling_plane(struct dc *dc, struct dc_state *context) #if defined(CONFIG_DRM_AMD_DC_DCN2_0) disable_all_writeback_pipes_for_stream(dc, old_stream, dangling_context); #endif - if (dc->hwss.apply_ctx_for_surface) + if (dc->hwss.apply_ctx_for_surface) { dc->hwss.apply_ctx_for_surface(dc, old_stream, 0, dangling_context); - } + dc->hwss.post_unlock_program_front_end(dc, dangling_context); + } #if defined(CONFIG_DRM_AMD_DC_DCN2_0) - if (dc->hwss.program_front_end_for_ctx) - dc->hwss.program_front_end_for_ctx(dc, dangling_context); + if (dc->hwss.program_front_end_for_ctx) { + dc->hwss.program_front_end_for_ctx(dc, dangling_context); + dc->hwss.post_unlock_program_front_end(dc, dangling_context); + } #endif + } } current_ctx = dc->current_state; @@ -1237,6 +1241,7 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c dc, context->streams[i], context->stream_status[i].plane_count, context); /* use new pipe config in new context */ + dc->hwss.post_unlock_program_front_end(dc, context); } /* Program hardware */ @@ -1257,20 +1262,25 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c /* Program all planes within new context*/ #if defined(CONFIG_DRM_AMD_DC_DCN2_0) - if (dc->hwss.program_front_end_for_ctx) + if (dc->hwss.program_front_end_for_ctx) { dc->hwss.program_front_end_for_ctx(dc, context); + dc->hwss.post_unlock_program_front_end(dc, context); + } #endif + for (i = 0; i < context->stream_count; i++) { const struct dc_link *link = context->streams[i]->link; if (!context->streams[i]->mode_changed) continue; - if (dc->hwss.apply_ctx_for_surface) + if (dc->hwss.apply_ctx_for_surface) { dc->hwss.apply_ctx_for_surface( dc, context->streams[i], context->stream_status[i].plane_count, context); + dc->hwss.post_unlock_program_front_end(dc, context); + } /* * enable stereo @@ -2221,6 +2231,7 @@ static void commit_planes_for_stream(struct dc *dc, dc->hwss.program_front_end_for_ctx(dc, context); #endif + dc->hwss.post_unlock_program_front_end(dc, context); return; } @@ -2363,6 +2374,9 @@ static void commit_planes_for_stream(struct dc *dc, dc->hwss.pipe_control_lock(dc, top_pipe_to_program, false); } + if (update_type != UPDATE_TYPE_FAST) + dc->hwss.post_unlock_program_front_end(dc, context); + // Fire manual trigger only when bottom plane is flipped 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/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c index 66ea9b009b8bc..dcf8a2ac48ab0 100644 --- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c @@ -2625,6 +2625,11 @@ static void dce110_apply_ctx_for_surface( enable_fbc(dc, context); } +static void dce110_post_unlock_program_front_end( + struct dc *dc, + struct dc_state *context) +{ +} static void dce110_power_down_fe(struct dc *dc, struct pipe_ctx *pipe_ctx) { struct dce_hwseq *hws = dc->hwseq; @@ -2726,6 +2731,7 @@ static const struct hw_sequencer_funcs dce110_funcs = { .init_hw = init_hw, .apply_ctx_to_hw = dce110_apply_ctx_to_hw, .apply_ctx_for_surface = dce110_apply_ctx_for_surface, + .post_unlock_program_front_end = dce110_post_unlock_program_front_end, .update_plane_addr = update_plane_addr, .update_pending_status = dce110_update_pending_status, .enable_accelerated_mode = dce110_enable_accelerated_mode, diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c index 8a92abf48d51a..4bca4cf38cf1d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c @@ -2528,7 +2528,6 @@ void dcn10_apply_ctx_for_surface( int i; struct timing_generator *tg; uint32_t underflow_check_delay_us; - bool removed_pipe[4] = { false }; bool interdependent_update = false; struct pipe_ctx *top_pipe_to_program = dcn10_find_top_pipe_for_stream(dc, context, stream); @@ -2568,6 +2567,9 @@ void dcn10_apply_ctx_for_surface( struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; struct pipe_ctx *old_pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; + + pipe_ctx->update_flags.raw = 0; + /* * Powergate reused pipes that are not powergated * fairly hacky right now, using opp_id as indicator @@ -2587,7 +2589,7 @@ void dcn10_apply_ctx_for_surface( old_pipe_ctx->stream_res.tg == tg) { hws->funcs.plane_atomic_disconnect(dc, old_pipe_ctx); - removed_pipe[i] = true; + pipe_ctx->update_flags.bits.disable = 1; DC_LOG_DC("Reset mpcc for pipe %d\n", old_pipe_ctx->pipe_idx); @@ -2620,16 +2622,41 @@ void dcn10_apply_ctx_for_surface( dcn10_lock_all_pipes(dc, context, false); else dcn10_pipe_control_lock(dc, top_pipe_to_program, false); +} + +void dcn10_post_unlock_program_front_end( + struct dc *dc, + struct dc_state *context) +{ + int i, j; + + DC_LOGGER_INIT(dc->ctx->logger); - if (num_planes == 0) - false_optc_underflow_wa(dc, stream, tg); + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; + + if (!pipe_ctx->top_pipe && + !pipe_ctx->prev_odm_pipe && + pipe_ctx->stream) { + struct dc_stream_status *stream_status = NULL; + struct timing_generator *tg = pipe_ctx->stream_res.tg; + + for (j = 0; j < context->stream_count; j++) { + if (pipe_ctx->stream == context->streams[j]) + stream_status = &context->stream_status[j]; + } + + if (context->stream_status[i].plane_count == 0) + false_optc_underflow_wa(dc, pipe_ctx->stream, tg); + } + } for (i = 0; i < dc->res_pool->pipe_count; i++) - if (removed_pipe[i]) + if (context->res_ctx.pipe_ctx[i].update_flags.bits.disable) dc->hwss.disable_plane(dc, &dc->current_state->res_ctx.pipe_ctx[i]); for (i = 0; i < dc->res_pool->pipe_count; i++) - if (removed_pipe[i]) { + if (context->res_ctx.pipe_ctx[i].update_flags.bits.disable) { dc->hwss.optimize_bandwidth(dc, context); break; } diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h index 4d20f6586bb57..b523f0b8dc23f 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h @@ -75,6 +75,9 @@ void dcn10_apply_ctx_for_surface( const struct dc_stream_state *stream, int num_planes, struct dc_state *context); +void dcn10_post_unlock_program_front_end( + struct dc *dc, + struct dc_state *context); void dcn10_hubp_pg_control( struct dce_hwseq *hws, unsigned int hubp_inst, diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_init.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_init.c index e7e5352ec424b..681db997a5323 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_init.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_init.c @@ -32,6 +32,7 @@ static const struct hw_sequencer_funcs dcn10_funcs = { .init_hw = dcn10_init_hw, .apply_ctx_to_hw = dce110_apply_ctx_to_hw, .apply_ctx_for_surface = dcn10_apply_ctx_for_surface, + .post_unlock_program_front_end = dcn10_post_unlock_program_front_end, .update_plane_addr = dcn10_update_plane_addr, .update_dchub = dcn10_update_dchub, .update_pending_status = dcn10_update_pending_status, diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c index e825ee71d8fc9..4b38864aad4d6 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c @@ -1557,7 +1557,6 @@ void dcn20_program_front_end_for_ctx( struct dc *dc, struct dc_state *context) { - const unsigned int TIMEOUT_FOR_PIPE_ENABLE_MS = 100; int i; struct dce_hwseq *hws = dc->hwseq; bool pipe_locked[MAX_PIPES] = {false}; @@ -1632,6 +1631,16 @@ void dcn20_program_front_end_for_ctx( if (!pipe_ctx->update_flags.bits.enable) dc->hwss.pipe_control_lock(dc, &dc->current_state->res_ctx.pipe_ctx[i], false); } +} + +void dcn20_post_unlock_program_front_end( + struct dc *dc, + struct dc_state *context) +{ + int i; + const unsigned int TIMEOUT_FOR_PIPE_ENABLE_MS = 100; + + DC_LOGGER_INIT(dc->ctx->logger); for (i = 0; i < dc->res_pool->pipe_count; i++) if (context->res_ctx.pipe_ctx[i].update_flags.bits.disable) diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h index 1373cd8875dda..59fdbdcb4f8e4 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h @@ -35,6 +35,9 @@ bool dcn20_set_shaper_3dlut( void dcn20_program_front_end_for_ctx( struct dc *dc, struct dc_state *context); +void dcn20_post_unlock_program_front_end( + struct dc *dc, + struct dc_state *context); void dcn20_update_plane_addr(const struct dc *dc, struct pipe_ctx *pipe_ctx); void dcn20_update_mpcc(struct dc *dc, struct pipe_ctx *pipe_ctx); bool dcn20_set_input_transfer_func(struct dc *dc, struct pipe_ctx *pipe_ctx, diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_init.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_init.c index 7ac145ef165f8..65ff1e264e1b8 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_init.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_init.c @@ -33,6 +33,7 @@ static const struct hw_sequencer_funcs dcn20_funcs = { .apply_ctx_to_hw = dce110_apply_ctx_to_hw, .apply_ctx_for_surface = NULL, .program_front_end_for_ctx = dcn20_program_front_end_for_ctx, + .post_unlock_program_front_end = dcn20_post_unlock_program_front_end, .update_plane_addr = dcn20_update_plane_addr, .update_dchub = dcn10_update_dchub, .update_pending_status = dcn10_update_pending_status, diff --git a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c index 45e79a8b70708..fd56eda52f03e 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c +++ b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c @@ -34,6 +34,7 @@ static const struct hw_sequencer_funcs dcn21_funcs = { .apply_ctx_to_hw = dce110_apply_ctx_to_hw, .apply_ctx_for_surface = NULL, .program_front_end_for_ctx = dcn20_program_front_end_for_ctx, + .post_unlock_program_front_end = dcn20_post_unlock_program_front_end, .update_plane_addr = dcn20_update_plane_addr, .update_dchub = dcn10_update_dchub, .update_pending_status = dcn10_update_pending_status, diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h index 052dc7533fadf..3b8ce17839ee3 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h @@ -72,6 +72,8 @@ struct hw_sequencer_funcs { void (*program_front_end_for_ctx)(struct dc *dc, struct dc_state *context); #endif + void (*post_unlock_program_front_end)(struct dc *dc, + struct dc_state *context); void (*update_plane_addr)(const struct dc *dc, struct pipe_ctx *pipe_ctx); void (*update_dchub)(struct dce_hwseq *hws, From 22de6799baf607a917b83a2c9958e6e9e9984f50 Mon Sep 17 00:00:00 2001 From: Anthony Koo Date: Tue, 14 Jan 2020 17:02:09 -0500 Subject: [PATCH 34/69] drm/amd/display: Indicate dsc updates explicitly [Why] DSC updates only set type to FULL UPDATE, but doesn't flag the change [How] Add DSC flag update flag Change-Id: I46dbd8c0ecd8dcc03b29acca5300adeea9f792a7 Signed-off-by: Anthony Koo Reviewed-by: Aric Cyr Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc.c | 23 +++++++++++++--------- drivers/gpu/drm/amd/display/dc/dc_stream.h | 1 + 2 files changed, 15 insertions(+), 9 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index fee424fadca0e..49644d479e1ce 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -1760,16 +1760,17 @@ static enum surface_update_type check_update_surfaces_for_stream( if (stream_update->wb_update) su_flags->bits.wb_update = 1; #endif + +#if defined(CONFIG_DRM_AMD_DC_DSC_SUPPORT) + if (stream_update->dsc_config) + su_flags->bits.dsc_changed = 1; +#endif + if (su_flags->raw != 0) overall_type = UPDATE_TYPE_FULL; if (stream_update->output_csc_transform || stream_update->output_color_space) su_flags->bits.out_csc = 1; - -#if defined(CONFIG_DRM_AMD_DC_DSC_SUPPORT) - if (stream_update->dsc_config) - overall_type = UPDATE_TYPE_FULL; -#endif } for (i = 0 ; i < surface_count; i++) { @@ -1804,8 +1805,11 @@ enum surface_update_type dc_check_update_surfaces_for_stream( type = check_update_surfaces_for_stream(dc, updates, surface_count, stream_update, stream_status); if (type == UPDATE_TYPE_FULL) { - if (stream_update) + if (stream_update) { + uint32_t dsc_changed = stream_update->stream->update_flags.bits.dsc_changed; stream_update->stream->update_flags.raw = 0xFFFFFFFF; + stream_update->stream->update_flags.bits.dsc_changed = dsc_changed; + } for (i = 0; i < surface_count; i++) updates[i].surface->update_flags.raw = 0xFFFFFFFF; } @@ -2131,6 +2135,10 @@ static void commit_planes_do_stream_update(struct dc *dc, #endif } + /* Full fe update*/ + if (update_type == UPDATE_TYPE_FAST) + continue; + #if defined(CONFIG_DRM_AMD_DC_DSC_SUPPORT) if (stream_update->dsc_config && dc->hwss.pipe_control_lock_global) { dc->hwss.pipe_control_lock_global(dc, pipe_ctx, true); @@ -2138,9 +2146,6 @@ static void commit_planes_do_stream_update(struct dc *dc, dc->hwss.pipe_control_lock_global(dc, pipe_ctx, false); } #endif - /* Full fe update*/ - if (update_type == UPDATE_TYPE_FAST) - continue; if (stream_update->dpms_off) { dc->hwss.pipe_control_lock(dc, pipe_ctx, true); diff --git a/drivers/gpu/drm/amd/display/dc/dc_stream.h b/drivers/gpu/drm/amd/display/dc/dc_stream.h index daf95d2971c03..98e6f3f65a459 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_stream.h +++ b/drivers/gpu/drm/amd/display/dc/dc_stream.h @@ -124,6 +124,7 @@ union stream_update_flags { #if defined(CONFIG_DRM_AMD_DC_DCN2_0) uint32_t wb_update:1; #endif + uint32_t dsc_changed : 1; } bits; uint32_t raw; From cdb6f54cf8510c25a4798af5a112baf2c1c33ee4 Mon Sep 17 00:00:00 2001 From: Eric Bernstein Date: Wed, 15 Jan 2020 14:41:23 -0500 Subject: [PATCH 35/69] drm/amd/display: Fix various issues found by compiler warning as errors [Why] Diagnostics team reported various issues found when enabling warnings as errors [How] Fix implicit conversions Signed-off-by: Eric Bernstein Reviewed-by: Aric Cyr Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c | 2 +- drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h | 2 +- drivers/gpu/drm/amd/display/dc/inc/hw/dwb.h | 3 +-- 3 files changed, 3 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c index 1587438bffade..b94c8a14571c8 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c @@ -3700,7 +3700,7 @@ static void set_crtc_test_pattern(struct dc_link *link, struct pipe_ctx *odm_pipe; enum controller_dp_color_space controller_color_space; int opp_cnt = 1; - int count; + uint16_t count = 0; switch (test_pattern_color_space) { case DP_TEST_PATTERN_COLOR_SPACE_RGB: diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h index 15d2429f04d74..8531531f94a91 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h +++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h @@ -191,7 +191,7 @@ struct _vcs_dpi_ip_params_st { unsigned int min_vblank_lines; unsigned int dppclk_delay_subtotal; unsigned int dispclk_delay_subtotal; - unsigned int dcfclk_cstate_latency; + double dcfclk_cstate_latency; unsigned int dppclk_delay_scl; unsigned int dppclk_delay_scl_lb_only; unsigned int dppclk_delay_cnvc_formatter; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dwb.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dwb.h index bd69580b9da92..87ddd7921113a 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dwb.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dwb.h @@ -25,16 +25,15 @@ #ifndef __DC_DWBC_H__ #define __DC_DWBC_H__ +#include "dal_types.h" #include "dc_hw_types.h" - #define DWB_SW_V2 1 #define DWB_MCIF_BUF_COUNT 4 /* forward declaration of mcif_wb struct */ struct mcif_wb; -enum dce_version; enum dwb_sw_version { dwb_ver_1_0 = 1, From d02f0be6125436a493635c28a5221a4ec875e76b Mon Sep 17 00:00:00 2001 From: Nicholas Kazlauskas Date: Wed, 15 Jan 2020 15:17:29 -0500 Subject: [PATCH 36/69] drm/amd/display: Add GPINT handler interface [Why] The General Purpose Interrupt is used on the DMCUB to pass lightweight commands via a register to the DMCUB. This is limited to 32-bit command and 32-bit response. This will be used for shutting down the firmware in a clean manner. [How] Add the command IDs and the data register to correctly format the commands. Add the interface functions to dmub_srv for sending and receiving the commands. Signed-off-by: Nicholas Kazlauskas Reviewed-by: Tony Cheng Acked-by: Bhawanpreet Lakha --- .../drm/amd/display/dmub/inc/dmub_gpint_cmd.h | 68 +++++++++++++++++++ .../gpu/drm/amd/display/dmub/inc/dmub_srv.h | 48 +++++++++++++ .../gpu/drm/amd/display/dmub/src/dmub_dcn20.c | 22 ++++++ .../gpu/drm/amd/display/dmub/src/dmub_dcn20.h | 9 +++ .../gpu/drm/amd/display/dmub/src/dmub_srv.c | 50 ++++++++++++++ 5 files changed, 197 insertions(+) create mode 100644 drivers/gpu/drm/amd/display/dmub/inc/dmub_gpint_cmd.h diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_gpint_cmd.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_gpint_cmd.h new file mode 100644 index 0000000000000..5398ed6b35d1f --- /dev/null +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_gpint_cmd.h @@ -0,0 +1,68 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef _DMUB_GPINT_CMD_H_ +#define _DMUB_GPINT_CMD_H_ + +#include "dmub_types.h" + +/** + * The register format for sending a command via the GPINT. + */ +union dmub_gpint_data_register { + struct { + uint32_t param : 16; + uint32_t command_code : 12; + uint32_t status : 4; + } bits; + uint32_t all; +}; + +/** + * The shifts and masks below may alternatively be used to format and read + * the command register bits. + */ + +#define DMUB_GPINT_DATA_PARAM_MASK 0xFFFF +#define DMUB_GPINT_DATA_PARAM_SHIFT 0 + +#define DMUB_GPINT_DATA_COMMAND_CODE_MASK 0xFFF +#define DMUB_GPINT_DATA_COMMAND_CODE_SHIFT 16 + +#define DMUB_GPINT_DATA_STATUS_MASK 0xF +#define DMUB_GPINT_DATA_STATUS_SHIFT 28 + +/* + * Command IDs should be treated as stable ABI. + * Do not reuse or modify IDs. + */ + +enum dmub_gpint_command { + DMUB_GPINT__INVALID_COMMAND = 0, + DMUB_GPINT__GET_FW_VERSION = 1, + DMUB_GPINT__STOP_FW = 2, +}; + +#endif /* _DMUB_GPINT_CMD_H_ */ diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_srv.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_srv.h index f8917594036ab..e619fa9cf53ac 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_srv.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_srv.h @@ -66,6 +66,7 @@ #include "dmub_types.h" #include "dmub_cmd.h" +#include "dmub_gpint_cmd.h" #include "dmub_rb.h" #if defined(__cplusplus) @@ -262,6 +263,14 @@ struct dmub_srv_hw_funcs { bool (*is_phy_init)(struct dmub_srv *dmub); bool (*is_auto_load_done)(struct dmub_srv *dmub); + + void (*set_gpint)(struct dmub_srv *dmub, + union dmub_gpint_data_register reg); + + bool (*is_gpint_acked)(struct dmub_srv *dmub, + union dmub_gpint_data_register reg); + + uint32_t (*get_gpint_response)(struct dmub_srv *dmub); }; /** @@ -516,6 +525,45 @@ enum dmub_status dmub_srv_wait_for_phy_init(struct dmub_srv *dmub, enum dmub_status dmub_srv_wait_for_idle(struct dmub_srv *dmub, uint32_t timeout_us); +/** + * dmub_srv_send_gpint_command() - Sends a GPINT based command. + * @dmub: the dmub service + * @command_code: the command code to send + * @param: the command parameter to send + * @timeout_us: the maximum number of microseconds to wait + * + * Sends a command via the general purpose interrupt (GPINT). + * Waits for the number of microseconds specified by timeout_us + * for the command ACK before returning. + * + * Can be called after software initialization. + * + * Return: + * DMUB_STATUS_OK - success + * DMUB_STATUS_TIMEOUT - wait for ACK timed out + * DMUB_STATUS_INVALID - unspecified error + */ +enum dmub_status +dmub_srv_send_gpint_command(struct dmub_srv *dmub, + enum dmub_gpint_command command_code, + uint16_t param, uint32_t timeout_us); + +/** + * dmub_srv_get_gpint_response() - Queries the GPINT response. + * @dmub: the dmub service + * @response: the response for the last GPINT + * + * Returns the response code for the last GPINT interrupt. + * + * Can be called after software initialization. + * + * Return: + * DMUB_STATUS_OK - success + * DMUB_STATUS_INVALID - unspecified error + */ +enum dmub_status dmub_srv_get_gpint_response(struct dmub_srv *dmub, + uint32_t *response); + #if defined(__cplusplus) } #endif diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c index b2ca8e0dbac92..7c1604c2221cc 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c @@ -217,3 +217,25 @@ bool dmub_dcn20_is_supported(struct dmub_srv *dmub) return supported; } + +void dmub_dcn20_set_gpint(struct dmub_srv *dmub, + union dmub_gpint_data_register reg) +{ + REG_WRITE(DMCUB_GPINT_DATAIN1, reg.all); +} + +bool dmub_dcn20_is_gpint_acked(struct dmub_srv *dmub, + union dmub_gpint_data_register reg) +{ + union dmub_gpint_data_register test; + + reg.bits.status = 0; + test.all = REG_READ(DMCUB_GPINT_DATAIN1); + + return test.all == reg.all; +} + +uint32_t dmub_dcn20_get_gpint_response(struct dmub_srv *dmub) +{ + return REG_READ(DMCUB_SCRATCH7); +} diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.h b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.h index 04b0fa13153dc..7f046c73927e7 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.h +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.h @@ -91,6 +91,7 @@ struct dmub_srv; DMUB_SR(DMCUB_SCRATCH13) \ DMUB_SR(DMCUB_SCRATCH14) \ DMUB_SR(DMCUB_SCRATCH15) \ + DMUB_SR(DMCUB_GPINT_DATAIN1) \ DMUB_SR(CC_DC_PIPE_DIS) \ DMUB_SR(MMHUBBUB_SOFT_RESET) \ DMUB_SR(DCN_VM_FB_LOCATION_BASE) \ @@ -183,4 +184,12 @@ bool dmub_dcn20_is_hw_init(struct dmub_srv *dmub); bool dmub_dcn20_is_supported(struct dmub_srv *dmub); +void dmub_dcn20_set_gpint(struct dmub_srv *dmub, + union dmub_gpint_data_register reg); + +bool dmub_dcn20_is_gpint_acked(struct dmub_srv *dmub, + union dmub_gpint_data_register reg); + +uint32_t dmub_dcn20_get_gpint_response(struct dmub_srv *dmub); + #endif /* _DMUB_DCN20_H_ */ diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c index 85a518bf8a76b..45be185ef3129 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c @@ -126,6 +126,9 @@ static bool dmub_srv_hw_setup(struct dmub_srv *dmub, enum dmub_asic asic) funcs->set_inbox1_wptr = dmub_dcn20_set_inbox1_wptr; funcs->is_supported = dmub_dcn20_is_supported; funcs->is_hw_init = dmub_dcn20_is_hw_init; + funcs->set_gpint = dmub_dcn20_set_gpint; + funcs->is_gpint_acked = dmub_dcn20_is_gpint_acked; + funcs->get_gpint_response = dmub_dcn20_get_gpint_response; if (asic == DMUB_ASIC_DCN21) { dmub->regs = &dmub_srv_dcn21_regs; @@ -522,3 +525,50 @@ enum dmub_status dmub_srv_wait_for_idle(struct dmub_srv *dmub, return DMUB_STATUS_TIMEOUT; } + +enum dmub_status +dmub_srv_send_gpint_command(struct dmub_srv *dmub, + enum dmub_gpint_command command_code, + uint16_t param, uint32_t timeout_us) +{ + union dmub_gpint_data_register reg; + uint32_t i; + + if (!dmub->sw_init) + return DMUB_STATUS_INVALID; + + if (!dmub->hw_funcs.set_gpint) + return DMUB_STATUS_INVALID; + + if (!dmub->hw_funcs.is_gpint_acked) + return DMUB_STATUS_INVALID; + + reg.bits.status = 1; + reg.bits.command_code = command_code; + reg.bits.param = param; + + dmub->hw_funcs.set_gpint(dmub, reg); + + for (i = 0; i < timeout_us; ++i) { + if (dmub->hw_funcs.is_gpint_acked(dmub, reg)) + return DMUB_STATUS_OK; + } + + return DMUB_STATUS_TIMEOUT; +} + +enum dmub_status dmub_srv_get_gpint_response(struct dmub_srv *dmub, + uint32_t *response) +{ + *response = 0; + + if (!dmub->sw_init) + return DMUB_STATUS_INVALID; + + if (!dmub->hw_funcs.get_gpint_response) + return DMUB_STATUS_INVALID; + + *response = dmub->hw_funcs.get_gpint_response(dmub); + + return DMUB_STATUS_OK; +} From 183f7bbc4df1cfc884f7c7096fbb4c39bdec0e97 Mon Sep 17 00:00:00 2001 From: Nicholas Kazlauskas Date: Thu, 16 Jan 2020 09:16:18 -0500 Subject: [PATCH 37/69] drm/amd/display: Wait for clean shutdown in DMCUB reset [Why] The DMCUB may be currently executing commands when the reset is triggered. Before issuing a reset we should first wait for the DMCUB to finish its work. [How] Send the GPINT command for halting the firmware before reset. Get the ack for the command then wait for the scratch register to become the correct value. We want this to take under ~40us or so at most before we force reset to cover PHY delay sequence max time. Each register read will be at least ~1-3us so don't bother using udelay. Signed-off-by: Nicholas Kazlauskas Reviewed-by: Tony Cheng Acked-by: Bhawanpreet Lakha --- .../drm/amd/display/dmub/inc/dmub_gpint_cmd.h | 6 ++++ .../gpu/drm/amd/display/dmub/src/dmub_dcn20.c | 36 +++++++++++++++++++ 2 files changed, 42 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_gpint_cmd.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_gpint_cmd.h index 5398ed6b35d1f..e13685917dab6 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_gpint_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_gpint_cmd.h @@ -65,4 +65,10 @@ enum dmub_gpint_command { DMUB_GPINT__STOP_FW = 2, }; +/** + * Command responses. + */ + +#define DMUB_GPINT__STOP_FW_RESPONSE 0xDEADDEAD + #endif /* _DMUB_GPINT_CMD_H_ */ diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c index 7c1604c2221cc..479f17bb3800f 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c @@ -77,6 +77,42 @@ static inline void dmub_dcn20_translate_addr(const union dmub_addr *addr_in, void dmub_dcn20_reset(struct dmub_srv *dmub) { + union dmub_gpint_data_register cmd; + const uint32_t timeout = 30; + uint32_t in_reset, scratch, i; + + REG_GET(DMCUB_CNTL, DMCUB_SOFT_RESET, &in_reset); + + if (in_reset == 0) { + cmd.bits.status = 1; + cmd.bits.command_code = DMUB_GPINT__STOP_FW; + cmd.bits.param = 0; + + dmub->hw_funcs.set_gpint(dmub, cmd); + + /** + * Timeout covers both the ACK and the wait + * for remaining work to finish. + * + * This is mostly bound by the PHY disable sequence. + * Each register check will be greater than 1us, so + * don't bother using udelay. + */ + + for (i = 0; i < timeout; ++i) { + if (dmub->hw_funcs.is_gpint_acked(dmub, cmd)) + break; + } + + for (i = 0; i < timeout; ++i) { + scratch = dmub->hw_funcs.get_gpint_response(dmub); + if (scratch == DMUB_GPINT__STOP_FW_RESPONSE) + break; + } + + /* Force reset in case we timed out, DMCUB is likely hung. */ + } + REG_UPDATE(DMCUB_CNTL, DMCUB_SOFT_RESET, 1); REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); From 6bac2a46a3d6651f56180e48cd10b5cbf5dc708b Mon Sep 17 00:00:00 2001 From: Anthony Koo Date: Tue, 14 Jan 2020 17:04:24 -0500 Subject: [PATCH 38/69] drm/amd/display: Added locking for atomic update stream and update planes [Why] Screen flickering when HDR switches between FP16 and ARGB2101010 [How] Moved pipe_control_lock so stream update and plane update occur atomically Change-Id: I1457f64bdba61c93964f83d8be14a55b8fd85146 Signed-off-by: Anthony Koo Signed-off-by: Lucy Li Reviewed-by: Aric Cyr Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc.c | 85 +++++++++++---- .../display/dc/dce110/dce110_hw_sequencer.c | 23 +--- .../amd/display/dc/dcn10/dcn10_hw_sequencer.c | 32 +----- .../amd/display/dc/dcn10/dcn10_hw_sequencer.h | 4 + .../gpu/drm/amd/display/dc/dcn10/dcn10_init.c | 1 + .../drm/amd/display/dc/dcn20/dcn20_hwseq.c | 101 ++++++++---------- .../drm/amd/display/dc/dcn20/dcn20_hwseq.h | 4 - .../gpu/drm/amd/display/dc/dcn20/dcn20_init.c | 2 +- .../gpu/drm/amd/display/dc/dcn21/dcn21_init.c | 2 +- .../gpu/drm/amd/display/dc/inc/hw_sequencer.h | 4 +- 10 files changed, 124 insertions(+), 134 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 49644d479e1ce..b78813671c027 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -775,6 +775,29 @@ static bool disable_all_writeback_pipes_for_stream( } #endif +void apply_ctx_interdependent_lock(struct dc *dc, struct dc_state *context, struct dc_stream_state *stream, bool lock) +{ + int i = 0; + + /* Checks if interdependent update function pointer is NULL or not, takes care of DCE110 case */ + if (dc->hwss.interdependent_update_lock) + dc->hwss.interdependent_update_lock(dc, context, lock); + else { + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; + struct pipe_ctx *old_pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; + + // Copied conditions that were previously in dce110_apply_ctx_for_surface + if (stream == pipe_ctx->stream) { + if (!pipe_ctx->top_pipe && + (pipe_ctx->plane_state || old_pipe_ctx->plane_state)) + dc->hwss.pipe_control_lock(dc, pipe_ctx, lock); + break; + } + } + } +} + static void disable_dangling_plane(struct dc *dc, struct dc_state *context) { int i, j; @@ -803,12 +826,16 @@ static void disable_dangling_plane(struct dc *dc, struct dc_state *context) disable_all_writeback_pipes_for_stream(dc, old_stream, dangling_context); #endif if (dc->hwss.apply_ctx_for_surface) { + apply_ctx_interdependent_lock(dc, dc->current_state, old_stream, true); dc->hwss.apply_ctx_for_surface(dc, old_stream, 0, dangling_context); + apply_ctx_interdependent_lock(dc, dc->current_state, old_stream, false); dc->hwss.post_unlock_program_front_end(dc, dangling_context); } #if defined(CONFIG_DRM_AMD_DC_DCN2_0) if (dc->hwss.program_front_end_for_ctx) { + dc->hwss.interdependent_update_lock(dc, dc->current_state, true); dc->hwss.program_front_end_for_ctx(dc, dangling_context); + dc->hwss.interdependent_update_lock(dc, dc->current_state, false); dc->hwss.post_unlock_program_front_end(dc, dangling_context); } #endif @@ -1232,17 +1259,19 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c /* re-program planes for existing stream, in case we need to * free up plane resource for later use */ - if (dc->hwss.apply_ctx_for_surface) + if (dc->hwss.apply_ctx_for_surface) { for (i = 0; i < context->stream_count; i++) { if (context->streams[i]->mode_changed) continue; - + apply_ctx_interdependent_lock(dc, context, context->streams[i], true); dc->hwss.apply_ctx_for_surface( dc, context->streams[i], context->stream_status[i].plane_count, context); /* use new pipe config in new context */ + apply_ctx_interdependent_lock(dc, context, context->streams[i], false); dc->hwss.post_unlock_program_front_end(dc, context); } + } /* Program hardware */ for (i = 0; i < dc->res_pool->pipe_count; i++) { @@ -1263,7 +1292,9 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c /* Program all planes within new context*/ #if defined(CONFIG_DRM_AMD_DC_DCN2_0) if (dc->hwss.program_front_end_for_ctx) { + dc->hwss.interdependent_update_lock(dc, context, true); dc->hwss.program_front_end_for_ctx(dc, context); + dc->hwss.interdependent_update_lock(dc, context, false); dc->hwss.post_unlock_program_front_end(dc, context); } #endif @@ -1275,10 +1306,12 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c continue; if (dc->hwss.apply_ctx_for_surface) { + apply_ctx_interdependent_lock(dc, context, context->streams[i], true); dc->hwss.apply_ctx_for_surface( dc, context->streams[i], context->stream_status[i].plane_count, context); + apply_ctx_interdependent_lock(dc, context, context->streams[i], false); dc->hwss.post_unlock_program_front_end(dc, context); } @@ -2140,16 +2173,11 @@ static void commit_planes_do_stream_update(struct dc *dc, continue; #if defined(CONFIG_DRM_AMD_DC_DSC_SUPPORT) - if (stream_update->dsc_config && dc->hwss.pipe_control_lock_global) { - dc->hwss.pipe_control_lock_global(dc, pipe_ctx, true); + if (stream_update->dsc_config) dp_update_dsc_config(pipe_ctx); - dc->hwss.pipe_control_lock_global(dc, pipe_ctx, false); - } #endif if (stream_update->dpms_off) { - dc->hwss.pipe_control_lock(dc, pipe_ctx, true); - if (*stream_update->dpms_off) { core_link_disable_stream(pipe_ctx); /* for dpms, keep acquired resources*/ @@ -2163,8 +2191,6 @@ static void commit_planes_do_stream_update(struct dc *dc, core_link_enable_stream(dc->current_state, pipe_ctx); } - - dc->hwss.pipe_control_lock(dc, pipe_ctx, false); } if (stream_update->abm_level && pipe_ctx->stream_res.abm) { @@ -2220,6 +2246,27 @@ static void commit_planes_for_stream(struct dc *dc, context_clock_trace(dc, context); } + for (j = 0; j < dc->res_pool->pipe_count; j++) { + struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j]; + + if (!pipe_ctx->top_pipe && + !pipe_ctx->prev_odm_pipe && + pipe_ctx->stream && + pipe_ctx->stream == stream) { + top_pipe_to_program = pipe_ctx; + } + } + + if ((update_type != UPDATE_TYPE_FAST) && dc->hwss.interdependent_update_lock) + dc->hwss.interdependent_update_lock(dc, context, true); + else + /* Lock the top pipe while updating plane addrs, since freesync requires + * plane addr update event triggers to be synchronized. + * top_pipe_to_program is expected to never be NULL + */ + dc->hwss.pipe_control_lock(dc, top_pipe_to_program, true); + + // Stream updates if (stream_update) commit_planes_do_stream_update(dc, stream, stream_update, update_type, context); @@ -2236,6 +2283,11 @@ static void commit_planes_for_stream(struct dc *dc, dc->hwss.program_front_end_for_ctx(dc, context); #endif + if ((update_type != UPDATE_TYPE_FAST) && dc->hwss.interdependent_update_lock) + dc->hwss.interdependent_update_lock(dc, context, false); + else + dc->hwss.pipe_control_lock(dc, top_pipe_to_program, false); + dc->hwss.post_unlock_program_front_end(dc, context); return; } @@ -2274,8 +2326,6 @@ static void commit_planes_for_stream(struct dc *dc, pipe_ctx->stream == stream) { struct dc_stream_status *stream_status = NULL; - top_pipe_to_program = pipe_ctx; - if (!pipe_ctx->plane_state) continue; @@ -2324,12 +2374,6 @@ static void commit_planes_for_stream(struct dc *dc, // Update Type FAST, Surface updates if (update_type == UPDATE_TYPE_FAST) { - /* Lock the top pipe while updating plane addrs, since freesync requires - * plane addr update event triggers to be synchronized. - * top_pipe_to_program is expected to never be NULL - */ - dc->hwss.pipe_control_lock(dc, top_pipe_to_program, true); - #if defined(CONFIG_DRM_AMD_DC_DCN2_0) if (dc->hwss.set_flip_control_gsl) for (i = 0; i < surface_count; i++) { @@ -2375,9 +2419,12 @@ static void commit_planes_for_stream(struct dc *dc, dc->hwss.update_plane_addr(dc, pipe_ctx); } } + } + if ((update_type != UPDATE_TYPE_FAST) && dc->hwss.interdependent_update_lock) + dc->hwss.interdependent_update_lock(dc, context, false); + else dc->hwss.pipe_control_lock(dc, top_pipe_to_program, false); - } if (update_type != UPDATE_TYPE_FAST) dc->hwss.post_unlock_program_front_end(dc, context); diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c index dcf8a2ac48ab0..064970b71d121 100644 --- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c @@ -2578,17 +2578,6 @@ static void dce110_apply_ctx_for_surface( if (dc->fbc_compressor) dc->fbc_compressor->funcs->disable_fbc(dc->fbc_compressor); - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; - struct pipe_ctx *old_pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; - - if (stream == pipe_ctx->stream) { - if (!pipe_ctx->top_pipe && - (pipe_ctx->plane_state || old_pipe_ctx->plane_state)) - dc->hwss.pipe_control_lock(dc, pipe_ctx, true); - } - } - for (i = 0; i < dc->res_pool->pipe_count; i++) { struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; @@ -2611,16 +2600,6 @@ static void dce110_apply_ctx_for_surface( } - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; - struct pipe_ctx *old_pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; - - if ((stream == pipe_ctx->stream) && - (!pipe_ctx->top_pipe) && - (pipe_ctx->plane_state || old_pipe_ctx->plane_state)) - dc->hwss.pipe_control_lock(dc, pipe_ctx, false); - } - if (dc->fbc_compressor) enable_fbc(dc, context); } @@ -2630,6 +2609,7 @@ static void dce110_post_unlock_program_front_end( struct dc_state *context) { } + static void dce110_power_down_fe(struct dc *dc, struct pipe_ctx *pipe_ctx) { struct dce_hwseq *hws = dc->hwseq; @@ -2746,6 +2726,7 @@ static const struct hw_sequencer_funcs dce110_funcs = { .disable_audio_stream = dce110_disable_audio_stream, .disable_plane = dce110_power_down_fe, .pipe_control_lock = dce_pipe_control_lock, + .interdependent_update_lock = NULL, .prepare_bandwidth = dce110_prepare_bandwidth, .optimize_bandwidth = dce110_optimize_bandwidth, .set_drr = set_drr, diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c index 4bca4cf38cf1d..5dd89d0dc6417 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c @@ -84,7 +84,7 @@ void print_microsec(struct dc_context *dc_ctx, us_x10 % frac); } -static void dcn10_lock_all_pipes(struct dc *dc, +void dcn10_lock_all_pipes(struct dc *dc, struct dc_state *context, bool lock) { @@ -95,6 +95,7 @@ static void dcn10_lock_all_pipes(struct dc *dc, for (i = 0; i < dc->res_pool->pipe_count; i++) { pipe_ctx = &context->res_ctx.pipe_ctx[i]; tg = pipe_ctx->stream_res.tg; + /* * Only lock the top pipe's tg to prevent redundant * (un)locking. Also skip if pipe is disabled. @@ -105,9 +106,9 @@ static void dcn10_lock_all_pipes(struct dc *dc, continue; if (lock) - tg->funcs->lock(tg); + dc->hwss.pipe_control_lock(dc, pipe_ctx, true); else - tg->funcs->unlock(tg); + dc->hwss.pipe_control_lock(dc, pipe_ctx, false); } } @@ -1588,7 +1589,7 @@ void dcn10_pipe_control_lock( /* use TG master update lock to lock everything on the TG * therefore only top pipe need to lock */ - if (pipe->top_pipe) + if (!pipe || pipe->top_pipe) return; if (dc->debug.sanity_checks) @@ -2546,11 +2547,6 @@ void dcn10_apply_ctx_for_surface( if (underflow_check_delay_us != 0xFFFFFFFF && hws->funcs.did_underflow_occur) ASSERT(hws->funcs.did_underflow_occur(dc, top_pipe_to_program)); - if (interdependent_update) - dcn10_lock_all_pipes(dc, context, true); - else - dcn10_pipe_control_lock(dc, top_pipe_to_program, true); - if (underflow_check_delay_us != 0xFFFFFFFF) udelay(underflow_check_delay_us); @@ -2570,19 +2566,6 @@ void dcn10_apply_ctx_for_surface( pipe_ctx->update_flags.raw = 0; - /* - * Powergate reused pipes that are not powergated - * fairly hacky right now, using opp_id as indicator - * TODO: After move dc_post to dc_update, this will - * be removed. - */ - if (pipe_ctx->plane_state && !old_pipe_ctx->plane_state) { - if (old_pipe_ctx->stream_res.tg == tg && - old_pipe_ctx->plane_res.hubp && - old_pipe_ctx->plane_res.hubp->opp_id != OPP_ID_INVALID) - dc->hwss.disable_plane(dc, old_pipe_ctx); - } - if ((!pipe_ctx->plane_state || pipe_ctx->stream_res.tg != old_pipe_ctx->stream_res.tg) && old_pipe_ctx->plane_state && @@ -2617,11 +2600,6 @@ void dcn10_apply_ctx_for_surface( &pipe_ctx->dlg_regs, &pipe_ctx->ttu_regs); } - - if (interdependent_update) - dcn10_lock_all_pipes(dc, context, false); - else - dcn10_pipe_control_lock(dc, top_pipe_to_program, false); } void dcn10_post_unlock_program_front_end( diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h index b523f0b8dc23f..16a50e05ffbfa 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h @@ -70,6 +70,10 @@ void dcn10_reset_hw_ctx_wrap( struct dc *dc, struct dc_state *context); void dcn10_disable_plane(struct dc *dc, struct pipe_ctx *pipe_ctx); +void dcn10_lock_all_pipes( + struct dc *dc, + struct dc_state *context, + bool lock); void dcn10_apply_ctx_for_surface( struct dc *dc, const struct dc_stream_state *stream, diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_init.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_init.c index 681db997a5323..b88ef9703b2bf 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_init.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_init.c @@ -50,6 +50,7 @@ static const struct hw_sequencer_funcs dcn10_funcs = { .disable_audio_stream = dce110_disable_audio_stream, .disable_plane = dcn10_disable_plane, .pipe_control_lock = dcn10_pipe_control_lock, + .interdependent_update_lock = dcn10_lock_all_pipes, .prepare_bandwidth = dcn10_prepare_bandwidth, .optimize_bandwidth = dcn10_optimize_bandwidth, .set_drr = dcn10_set_drr, diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c index 4b38864aad4d6..e4e27ec4891b5 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c @@ -1094,40 +1094,18 @@ void dcn20_enable_plane( // } } - -void dcn20_pipe_control_lock_global( - struct dc *dc, - struct pipe_ctx *pipe, - bool lock) -{ - if (lock) { - pipe->stream_res.tg->funcs->lock_doublebuffer_enable( - pipe->stream_res.tg); - pipe->stream_res.tg->funcs->lock(pipe->stream_res.tg); - } else { - pipe->stream_res.tg->funcs->unlock(pipe->stream_res.tg); - pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, - CRTC_STATE_VACTIVE); - pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, - CRTC_STATE_VBLANK); - pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, - CRTC_STATE_VACTIVE); - pipe->stream_res.tg->funcs->lock_doublebuffer_disable( - pipe->stream_res.tg); - } -} - void dcn20_pipe_control_lock( struct dc *dc, struct pipe_ctx *pipe, bool lock) { bool flip_immediate = false; + bool dig_update_required = false; /* use TG master update lock to lock everything on the TG * therefore only top pipe need to lock */ - if (pipe->top_pipe) + if (!pipe || pipe->top_pipe) return; if (pipe->plane_state != NULL) @@ -1160,6 +1138,19 @@ void dcn20_pipe_control_lock( (!flip_immediate && pipe->stream_res.gsl_group > 0)) dcn20_setup_gsl_group_as_lock(dc, pipe, flip_immediate); + if (pipe->stream && pipe->stream->update_flags.bits.dsc_changed) + dig_update_required = true; + + /* Need double buffer lock mode in order to synchronize front end pipe + * updates with dig updates. + */ + if (dig_update_required) { + if (lock) { + pipe->stream_res.tg->funcs->lock_doublebuffer_enable( + pipe->stream_res.tg); + } + } + if (pipe->plane_state != NULL && pipe->plane_state->triplebuffer_flips) { if (lock) pipe->stream_res.tg->funcs->triplebuffer_lock(pipe->stream_res.tg); @@ -1171,6 +1162,19 @@ void dcn20_pipe_control_lock( else pipe->stream_res.tg->funcs->unlock(pipe->stream_res.tg); } + + if (dig_update_required) { + if (!lock) { + pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, + CRTC_STATE_VACTIVE); + pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, + CRTC_STATE_VBLANK); + pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, + CRTC_STATE_VACTIVE); + pipe->stream_res.tg->funcs->lock_doublebuffer_disable( + pipe->stream_res.tg); + } + } } static void dcn20_detect_pipe_changes(struct pipe_ctx *old_pipe, struct pipe_ctx *new_pipe) @@ -1542,26 +1546,28 @@ static void dcn20_program_pipe( } } -static bool does_pipe_need_lock(struct pipe_ctx *pipe) -{ - if ((pipe->plane_state && pipe->plane_state->update_flags.raw) - || pipe->update_flags.raw) - return true; - if (pipe->bottom_pipe) - return does_pipe_need_lock(pipe->bottom_pipe); - - return false; -} - void dcn20_program_front_end_for_ctx( struct dc *dc, struct dc_state *context) { int i; struct dce_hwseq *hws = dc->hwseq; - bool pipe_locked[MAX_PIPES] = {false}; DC_LOGGER_INIT(dc->ctx->logger); + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; + + if (!pipe_ctx->top_pipe && !pipe_ctx->prev_odm_pipe && pipe_ctx->plane_state) { + ASSERT(!pipe_ctx->plane_state->triplebuffer_flips); + if (dc->hwss.program_triplebuffer != NULL && + !dc->debug.disable_tri_buf) { + /*turn off triple buffer for full update*/ + dc->hwss.program_triplebuffer( + dc, pipe_ctx, pipe_ctx->plane_state->triplebuffer_flips); + } + } + } + /* Carry over GSL groups in case the context is changing. */ for (i = 0; i < dc->res_pool->pipe_count; i++) if (context->res_ctx.pipe_ctx[i].stream == dc->current_state->res_ctx.pipe_ctx[i].stream) @@ -1572,17 +1578,6 @@ void dcn20_program_front_end_for_ctx( for (i = 0; i < dc->res_pool->pipe_count; i++) dcn20_detect_pipe_changes(&dc->current_state->res_ctx.pipe_ctx[i], &context->res_ctx.pipe_ctx[i]); - for (i = 0; i < dc->res_pool->pipe_count; i++) - if (!context->res_ctx.pipe_ctx[i].top_pipe && - does_pipe_need_lock(&context->res_ctx.pipe_ctx[i])) { - struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; - - if (pipe_ctx->update_flags.bits.tg_changed || pipe_ctx->update_flags.bits.enable) - dc->hwss.pipe_control_lock(dc, pipe_ctx, true); - if (!pipe_ctx->update_flags.bits.enable) - dc->hwss.pipe_control_lock(dc, &dc->current_state->res_ctx.pipe_ctx[i], true); - pipe_locked[i] = true; - } /* OTG blank before disabling all front ends */ for (i = 0; i < dc->res_pool->pipe_count; i++) @@ -1620,17 +1615,6 @@ void dcn20_program_front_end_for_ctx( hws->funcs.program_all_writeback_pipes_in_tree(dc, pipe->stream, context); } } - - /* Unlock all locked pipes */ - for (i = 0; i < dc->res_pool->pipe_count; i++) - if (pipe_locked[i]) { - struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; - - if (pipe_ctx->update_flags.bits.tg_changed || pipe_ctx->update_flags.bits.enable) - dc->hwss.pipe_control_lock(dc, pipe_ctx, false); - if (!pipe_ctx->update_flags.bits.enable) - dc->hwss.pipe_control_lock(dc, &dc->current_state->res_ctx.pipe_ctx[i], false); - } } void dcn20_post_unlock_program_front_end( @@ -1670,7 +1654,6 @@ void dcn20_post_unlock_program_front_end( dc->res_pool->hubbub->funcs->apply_DEDCN21_147_wa(dc->res_pool->hubbub); } - void dcn20_prepare_bandwidth( struct dc *dc, struct dc_state *context) diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h index 59fdbdcb4f8e4..7d952dab58c5b 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h @@ -61,10 +61,6 @@ void dcn20_pipe_control_lock( struct dc *dc, struct pipe_ctx *pipe, bool lock); -void dcn20_pipe_control_lock_global( - struct dc *dc, - struct pipe_ctx *pipe, - bool lock); void dcn20_prepare_bandwidth( struct dc *dc, struct dc_state *context); diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_init.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_init.c index 65ff1e264e1b8..0176e4cadcfbc 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_init.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_init.c @@ -51,7 +51,7 @@ static const struct hw_sequencer_funcs dcn20_funcs = { .disable_audio_stream = dce110_disable_audio_stream, .disable_plane = dcn20_disable_plane, .pipe_control_lock = dcn20_pipe_control_lock, - .pipe_control_lock_global = dcn20_pipe_control_lock_global, + .interdependent_update_lock = dcn10_lock_all_pipes, .prepare_bandwidth = dcn20_prepare_bandwidth, .optimize_bandwidth = dcn20_optimize_bandwidth, .update_bandwidth = dcn20_update_bandwidth, diff --git a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c index fd56eda52f03e..b6902f7432169 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c +++ b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c @@ -52,7 +52,7 @@ static const struct hw_sequencer_funcs dcn21_funcs = { .disable_audio_stream = dce110_disable_audio_stream, .disable_plane = dcn20_disable_plane, .pipe_control_lock = dcn20_pipe_control_lock, - .pipe_control_lock_global = dcn20_pipe_control_lock_global, + .interdependent_update_lock = dcn10_lock_all_pipes, .prepare_bandwidth = dcn20_prepare_bandwidth, .optimize_bandwidth = dcn20_optimize_bandwidth, .update_bandwidth = dcn20_update_bandwidth, diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h index 3b8ce17839ee3..cca730e22bf99 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h @@ -88,10 +88,10 @@ struct hw_sequencer_funcs { void (*update_pending_status)(struct pipe_ctx *pipe_ctx); /* Pipe Lock Related */ - void (*pipe_control_lock_global)(struct dc *dc, - struct pipe_ctx *pipe, bool lock); void (*pipe_control_lock)(struct dc *dc, struct pipe_ctx *pipe, bool lock); + void (*interdependent_update_lock)(struct dc *dc, + struct dc_state *context, bool lock); #if defined(CONFIG_DRM_AMD_DC_DCN2_0) void (*set_flip_control_gsl)(struct pipe_ctx *pipe_ctx, bool flip_immediate); From 07c8b2398295fe10a50d9b7331186259a62131ed Mon Sep 17 00:00:00 2001 From: Aric Cyr Date: Sat, 18 Jan 2020 14:03:55 -0500 Subject: [PATCH 39/69] drm/amd/display: Fix GSL acquire [Why] After locking refactor GSL is not acquired properly resulting in immediate flip issues. [How] Do not copy old GSL state anymore since GSL is acquired earlier now. Signed-off-by: Aric Cyr Reviewed-by: Anthony Koo Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 6 ------ 1 file changed, 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c index e4e27ec4891b5..d7cc7ccdaed64 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c @@ -1568,12 +1568,6 @@ void dcn20_program_front_end_for_ctx( } } - /* Carry over GSL groups in case the context is changing. */ - for (i = 0; i < dc->res_pool->pipe_count; i++) - if (context->res_ctx.pipe_ctx[i].stream == dc->current_state->res_ctx.pipe_ctx[i].stream) - context->res_ctx.pipe_ctx[i].stream_res.gsl_group = - dc->current_state->res_ctx.pipe_ctx[i].stream_res.gsl_group; - /* Set pipe update flags and lock pipes */ for (i = 0; i < dc->res_pool->pipe_count; i++) dcn20_detect_pipe_changes(&dc->current_state->res_ctx.pipe_ctx[i], From 9bbedd787aa5a4092131ca1e4118735d71c2bbf5 Mon Sep 17 00:00:00 2001 From: Aric Cyr Date: Sat, 18 Jan 2020 22:50:59 -0500 Subject: [PATCH 40/69] drm/amd/display: remove unused variable Signed-off-by: Aric Cyr Reviewed-by: Aric Cyr Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/dc.h | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index f3277eacc5542..440b5d599fc85 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -705,7 +705,6 @@ struct dc_3dlut { struct kref refcount; struct tetrahedral_params lut_3d; struct fixed31_32 hdr_multiplier; - bool initialized; /*remove after diag fix*/ union dc_3dlut_state state; struct dc_context *ctx; }; From 160489b81f6a3f9894cb6a8ed341efe0b59172aa Mon Sep 17 00:00:00 2001 From: Aric Cyr Date: Mon, 20 Jan 2020 08:15:27 -0500 Subject: [PATCH 41/69] drm/amd/display: 3.2.70 Signed-off-by: Aric Cyr Reviewed-by: Aric Cyr Acked-by: Bhawanpreet Lakha --- 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 440b5d599fc85..88578a1d868d5 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -39,7 +39,7 @@ #include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#define DC_VER "3.2.69" +#define DC_VER "3.2.70" #define MAX_SURFACES 3 #define MAX_PLANES 6 From 1d2061fc855b76a63c041530f824842b052852fb Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Wed, 5 Feb 2020 18:16:12 +0800 Subject: [PATCH 42/69] Revert "Revert "drm/amdgpu: added support to get mGPU DRAM base"" This reverts commit 5aa0e19769df2db6c7f5762a3ddd8426c93e7853. Signed-off-by: Stanley.Yang --- drivers/gpu/drm/amd/amdgpu/amdgpu_df.h | 3 + drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 20 +++++++ drivers/gpu/drm/amd/amdgpu/df_v3_6.c | 59 ++++++++++++++++++- .../amd/include/asic_reg/df/df_3_6_offset.h | 3 + .../amd/include/asic_reg/df/df_3_6_sh_mask.h | 8 +++ 5 files changed, 92 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h index 61a26c15c8dd5..057f6ea645d7e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h @@ -52,6 +52,9 @@ struct amdgpu_df_funcs { uint64_t (*get_fica)(struct amdgpu_device *adev, uint32_t ficaa_val); void (*set_fica)(struct amdgpu_device *adev, uint32_t ficaa_val, uint32_t ficadl_val, uint32_t ficadh_val); + uint64_t (*get_dram_base_addr)(struct amdgpu_device *adev, + uint32_t df_inst); + uint32_t (*get_df_inst_id)(struct amdgpu_device *adev); }; struct amdgpu_df { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 92e9c73a68204..87f57878dbd50 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -742,6 +742,20 @@ int amdgpu_ras_error_query(struct amdgpu_device *adev, return 0; } +uint64_t get_xgmi_relative_phy_addr(struct amdgpu_device *adev, uint64_t addr) +{ + uint32_t df_inst_id; + + if ((!adev->df.funcs) || + (!adev->df.funcs->get_df_inst_id) || + (!adev->df.funcs->get_dram_base_addr)) + return addr; + + df_inst_id = adev->df.funcs->get_df_inst_id(adev); + + return addr + adev->df.funcs->get_dram_base_addr(adev, df_inst_id); +} + /* wrapper of psp_ras_trigger_error */ int amdgpu_ras_error_inject(struct amdgpu_device *adev, struct ras_inject_if *info) @@ -759,6 +773,12 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, if (!obj) return -EINVAL; + /* Calculate XGMI relative offset */ + if (adev->gmc.xgmi.num_physical_nodes > 1) { + block_info.address = get_xgmi_relative_phy_addr(adev, + block_info.address); + } + switch (info->head.block) { case AMDGPU_RAS_BLOCK__GFX: if (adev->gfx.funcs->ras_error_inject) diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c index f51326598a8c0..d8cfd7f53e1fa 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c @@ -27,6 +27,9 @@ #include "df/df_3_6_offset.h" #include "df/df_3_6_sh_mask.h" +#define DF_3_6_SMN_REG_INST_DIST 0x8 +#define DF_3_6_INST_CNT 8 + static u32 df_v3_6_channel_number[] = {1, 2, 0, 4, 0, 8, 0, 16, 32, 0, 0, 0, 2, 4, 8}; @@ -683,6 +686,58 @@ static void df_v3_6_pmc_get_count(struct amdgpu_device *adev, } } +static uint64_t df_v3_6_get_dram_base_addr(struct amdgpu_device *adev, + uint32_t df_inst) +{ + uint32_t base_addr_reg_val = 0; + uint64_t base_addr = 0; + + base_addr_reg_val = RREG32_PCIE(smnDF_CS_UMC_AON0_DramBaseAddress0 + + df_inst * DF_3_6_SMN_REG_INST_DIST); + + if (REG_GET_FIELD(base_addr_reg_val, + DF_CS_UMC_AON0_DramBaseAddress0, + AddrRngVal) == 0) { + DRM_WARN("address range not valid"); + return 0; + } + + base_addr = REG_GET_FIELD(base_addr_reg_val, + DF_CS_UMC_AON0_DramBaseAddress0, + DramBaseAddr); + + return base_addr << 28; +} + +static uint32_t df_v3_6_get_df_inst_id(struct amdgpu_device *adev) +{ + uint32_t xgmi_node_id = 0; + uint32_t df_inst_id = 0; + + /* Walk through DF dst nodes to find current XGMI node */ + for (df_inst_id = 0; df_inst_id < DF_3_6_INST_CNT; df_inst_id++) { + + xgmi_node_id = RREG32_PCIE(smnDF_CS_UMC_AON0_DramLimitAddress0 + + df_inst_id * DF_3_6_SMN_REG_INST_DIST); + xgmi_node_id = REG_GET_FIELD(xgmi_node_id, + DF_CS_UMC_AON0_DramLimitAddress0, + DstFabricID); + + /* TODO: establish reason dest fabric id is offset by 7 */ + xgmi_node_id = xgmi_node_id >> 7; + + if (adev->gmc.xgmi.physical_node_id == xgmi_node_id) + break; + } + + if (df_inst_id == DF_3_6_INST_CNT) { + DRM_WARN("cant match df dst id with gpu node"); + return 0; + } + + return df_inst_id; +} + const struct amdgpu_df_funcs df_v3_6_funcs = { .sw_init = df_v3_6_sw_init, .sw_fini = df_v3_6_sw_fini, @@ -696,5 +751,7 @@ const struct amdgpu_df_funcs df_v3_6_funcs = { .pmc_stop = df_v3_6_pmc_stop, .pmc_get_count = df_v3_6_pmc_get_count, .get_fica = df_v3_6_get_fica, - .set_fica = df_v3_6_set_fica + .set_fica = df_v3_6_set_fica, + .get_dram_base_addr = df_v3_6_get_dram_base_addr, + .get_df_inst_id = df_v3_6_get_df_inst_id }; diff --git a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h index 87c84691b5be9..bb2c9c7a18dff 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h +++ b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h @@ -71,4 +71,7 @@ #define smnDF_PIE_AON_FabricIndirectConfigAccessDataLo3 0x1d098UL #define smnDF_PIE_AON_FabricIndirectConfigAccessDataHi3 0x1d09cUL +#define smnDF_CS_UMC_AON0_DramBaseAddress0 0x1c110UL +#define smnDF_CS_UMC_AON0_DramLimitAddress0 0x1c114UL + #endif diff --git a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h index 65e9f756e86e0..7afa87c7ff543 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h +++ b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h @@ -53,4 +53,12 @@ #define DF_CS_UMC_AON0_DramBaseAddress0__IntLvAddrSel_MASK 0x00000E00L #define DF_CS_UMC_AON0_DramBaseAddress0__DramBaseAddr_MASK 0xFFFFF000L +//DF_CS_UMC_AON0_DramLimitAddress0 +#define DF_CS_UMC_AON0_DramLimitAddress0__DstFabricID__SHIFT 0x0 +#define DF_CS_UMC_AON0_DramLimitAddress0__AllowReqIO__SHIFT 0xa +#define DF_CS_UMC_AON0_DramLimitAddress0__DramLimitAddr__SHIFT 0xc +#define DF_CS_UMC_AON0_DramLimitAddress0__DstFabricID_MASK 0x000003FFL +#define DF_CS_UMC_AON0_DramLimitAddress0__AllowReqIO_MASK 0x00000400L +#define DF_CS_UMC_AON0_DramLimitAddress0__DramLimitAddr_MASK 0xFFFFF000L + #endif From 795c62273954a82323d229cd195512074298bbc2 Mon Sep 17 00:00:00 2001 From: Peikang Zhang Date: Mon, 20 Jan 2020 09:25:15 -0500 Subject: [PATCH 43/69] drm/amd/display: dc_get_vmid_use_vector() crashes when get called [Why] int i can go out of boundary which will cause crash [How] Fixed the maximum value of i to avoid i going out of boundary Signed-off-by: Peikang Zhang Reviewed-by: Jun Lei Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c b/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c index a96d8de9380e6..f2b39ec35c898 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c @@ -62,7 +62,7 @@ int dc_get_vmid_use_vector(struct dc *dc) int i; int in_use = 0; - for (i = 0; i < dc->vm_helper->num_vmid; i++) + for (i = 0; i < MAX_HUBP; i++) in_use |= dc->vm_helper->hubp_vmid_usage[i].vmid_usage[0] | dc->vm_helper->hubp_vmid_usage[i].vmid_usage[1]; return in_use; From 449569da4ddc3108e90db33e85c48c33c8f6fe4e Mon Sep 17 00:00:00 2001 From: Roman Li Date: Thu, 30 Jan 2020 15:43:07 -0500 Subject: [PATCH 44/69] drm/amd/display: remove early break in interdependent_lock [Why] The break in apply_ctx_interdependent_lock() may potentially lead to early break from the loop leaving update plane unlocked [How] Remove break Signed-off-by: Roman Li Reviewed-by: Anthony Koo --- drivers/gpu/drm/amd/display/dc/core/dc.c | 1 - 1 file changed, 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 b78813671c027..c4e5bcb3e5413 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -792,7 +792,6 @@ void apply_ctx_interdependent_lock(struct dc *dc, struct dc_state *context, stru if (!pipe_ctx->top_pipe && (pipe_ctx->plane_state || old_pipe_ctx->plane_state)) dc->hwss.pipe_control_lock(dc, pipe_ctx, lock); - break; } } } From 4fd9a523417a4736838d4bd417ac8f67ff20640c Mon Sep 17 00:00:00 2001 From: Peikang Zhang Date: Wed, 15 Jan 2020 15:45:10 -0500 Subject: [PATCH 45/69] drm/amd/display: Update hubbub description comment Description for DCHUBBUB_TEST_DEBUG_DATA is changed to avoid any future confusions. Signed-off-by: Peikang Zhang Reviewed-by: Aric Cyr Acked-by: Bhawanpreet Lakha --- .../drm/amd/display/dc/dcn10/dcn10_hubbub.c | 63 +------------------ 1 file changed, 3 insertions(+), 60 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c index f36a0d8cedfe1..870735deaca71 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c @@ -147,8 +147,9 @@ bool hubbub1_verify_allow_pstate_change_high( forced_pstate_allow = false; } - /* RV2: - * dchubbubdebugind, at: 0xB + /* The following table only applies to DCN1 and DCN2, + * for newer DCNs, need to consult with HW IP folks to read RTL + * HUBBUB:DCHUBBUB_TEST_ARB_DEBUG10 DCHUBBUBDEBUGIND:0xB * description * 0: Pipe0 Plane0 Allow Pstate Change * 1: Pipe0 Plane1 Allow Pstate Change @@ -181,64 +182,6 @@ bool hubbub1_verify_allow_pstate_change_high( * 28: WB0 Allow Pstate Change * 29: WB1 Allow Pstate Change * 30: Arbiter's allow_pstate_change - * 31: SOC pstate change request" - */ - /*DCN2.x: - HUBBUB:DCHUBBUB_TEST_ARB_DEBUG10 DCHUBBUBDEBUGIND:0xB - 0: Pipe0 Plane0 Allow P-state Change - 1: Pipe0 Plane1 Allow P-state Change - 2: Pipe0 Cursor0 Allow P-state Change - 3: Pipe0 Cursor1 Allow P-state Change - 4: Pipe1 Plane0 Allow P-state Change - 5: Pipe1 Plane1 Allow P-state Change - 6: Pipe1 Cursor0 Allow P-state Change - 7: Pipe1 Cursor1 Allow P-state Change - 8: Pipe2 Plane0 Allow P-state Change - 9: Pipe2 Plane1 Allow P-state Change - 10: Pipe2 Cursor0 Allow P-state Change - 11: Pipe2 Cursor1 Allow P-state Change - 12: Pipe3 Plane0 Allow P-state Change - 13: Pipe3 Plane1 Allow P-state Change - 14: Pipe3 Cursor0 Allow P-state Change - 15: Pipe3 Cursor1 Allow P-state Change - 16: Pipe4 Plane0 Allow P-state Change - 17: Pipe4 Plane1 Allow P-state Change - 18: Pipe4 Cursor0 Allow P-state Change - 19: Pipe4 Cursor1 Allow P-state Change - 20: Pipe5 Plane0 Allow P-state Change - 21: Pipe5 Plane1 Allow P-state Change - 22: Pipe5 Cursor0 Allow P-state Change - 23: Pipe5 Cursor1 Allow P-state Change - 24: Pipe6 Plane0 Allow P-state Change - 25: Pipe6 Plane1 Allow P-state Change - 26: Pipe6 Cursor0 Allow P-state Change - 27: Pipe6 Cursor1 Allow P-state Change - 28: WB0 Allow P-state Change - 29: WB1 Allow P-state Change - 30: Arbiter`s Allow P-state Change - 31: SOC P-state Change request - */ - /* RV1: - * dchubbubdebugind, at: 0x7 - * description "3-0: Pipe0 cursor0 QOS - * 7-4: Pipe1 cursor0 QOS - * 11-8: Pipe2 cursor0 QOS - * 15-12: Pipe3 cursor0 QOS - * 16: Pipe0 Plane0 Allow Pstate Change - * 17: Pipe1 Plane0 Allow Pstate Change - * 18: Pipe2 Plane0 Allow Pstate Change - * 19: Pipe3 Plane0 Allow Pstate Change - * 20: Pipe0 Plane1 Allow Pstate Change - * 21: Pipe1 Plane1 Allow Pstate Change - * 22: Pipe2 Plane1 Allow Pstate Change - * 23: Pipe3 Plane1 Allow Pstate Change - * 24: Pipe0 cursor0 Allow Pstate Change - * 25: Pipe1 cursor0 Allow Pstate Change - * 26: Pipe2 cursor0 Allow Pstate Change - * 27: Pipe3 cursor0 Allow Pstate Change - * 28: WB0 Allow Pstate Change - * 29: WB1 Allow Pstate Change - * 30: Arbiter's allow_pstate_change * 31: SOC pstate change request */ From fdf60552bfb8e5dd6e672ed5c05021d31b75c700 Mon Sep 17 00:00:00 2001 From: Charlene Liu Date: Fri, 10 Jan 2020 14:52:49 -0500 Subject: [PATCH 46/69] drm/amd/display: add stream_enc_inst for PSP HDCP inst use [why] new HW engine mapping requirment use in PSP [how] report stream_enc_inst Signed-off-by: Charlene Liu Reviewed-by: Wenjing Liu Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 3 ++- drivers/gpu/drm/amd/display/dc/core/dc_resource.c | 2 +- drivers/gpu/drm/amd/display/dc/dcn10/dcn10_stream_encoder.c | 1 + drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.c | 1 + drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h | 1 + 5 files changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c index 8bbe45bfb12ad..b21c176ccad9f 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c @@ -2929,7 +2929,8 @@ static void update_psp_stream_config(struct pipe_ctx *pipe_ctx, bool dpms_off) memset(&config, 0, sizeof(config)); config.otg_inst = (uint8_t) pipe_ctx->stream_res.tg->inst; - config.stream_enc_inst = (uint8_t) pipe_ctx->stream_res.stream_enc->id; + /*stream_enc_inst*/ + config.stream_enc_inst = (uint8_t) pipe_ctx->stream_res.stream_enc->stream_enc_inst; config.link_enc_inst = pipe_ctx->stream->link->link_enc_hw_inst; config.dpms_off = dpms_off; config.dm_stream_ctx = pipe_ctx->stream->dm_stream_context; diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index 9410614a1cb1c..50740aab0e7ad 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -2046,7 +2046,7 @@ enum dc_status resource_map_pool_resources( for (i = 0; i < context->stream_count; i++) if (context->streams[i] == stream) { context->stream_status[i].primary_otg_inst = pipe_ctx->stream_res.tg->inst; - context->stream_status[i].stream_enc_inst = pipe_ctx->stream_res.stream_enc->id; + context->stream_status[i].stream_enc_inst = pipe_ctx->stream_res.stream_enc->stream_enc_inst; context->stream_status[i].audio_inst = pipe_ctx->stream_res.audio ? pipe_ctx->stream_res.audio->inst : -1; diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_stream_encoder.c index 376c4264d295b..7eba9333c3285 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_stream_encoder.c @@ -1667,5 +1667,6 @@ void dcn10_stream_encoder_construct( enc1->regs = regs; enc1->se_shift = se_shift; enc1->se_mask = se_mask; + enc1->base.stream_enc_inst = eng_id - ENGINE_ID_DIGA; } diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.c index 98c5bd90ae5ab..3011cdf7fd2b2 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.c @@ -622,5 +622,6 @@ void dcn20_stream_encoder_construct( enc1->regs = regs; enc1->se_shift = se_shift; enc1->se_mask = se_mask; + enc1->base.stream_enc_inst = eng_id - ENGINE_ID_DIGA; } diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h b/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h index c0b93d51ca8d4..1d35aa1edb451 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h @@ -107,6 +107,7 @@ struct stream_encoder { struct dc_context *ctx; struct dc_bios *bp; enum engine_id id; + uint32_t stream_enc_inst; }; #ifdef CONFIG_DRM_AMD_DC_DSC_SUPPORT From 313f865ebf4855c57d4f55ef969f5c329ae3f42f Mon Sep 17 00:00:00 2001 From: Sung Lee Date: Mon, 20 Jan 2020 18:58:45 -0500 Subject: [PATCH 47/69] drm/amd/display: Use dcfclk to populate watermark ranges [WHY & HOW] Previously drain clk was unconstrained and fill clk was constrained on fclk. We want to change it to fill clk unconstrained and drain clock constrained to dcfclk. Signed-off-by: Sung Lee Reviewed-by: Tony Cheng Acked-by: Bhawanpreet Lakha --- .../drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c index 7ae4c06232dd2..034a5852a4168 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c @@ -412,19 +412,19 @@ void build_watermark_ranges(struct clk_bw_params *bw_params, struct pp_smu_wm_ra ranges->reader_wm_sets[num_valid_sets].wm_inst = bw_params->wm_table.entries[i].wm_inst; ranges->reader_wm_sets[num_valid_sets].wm_type = bw_params->wm_table.entries[i].wm_type; - /* We will not select WM based on dcfclk, so leave it as unconstrained */ - ranges->reader_wm_sets[num_valid_sets].min_drain_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN; - ranges->reader_wm_sets[num_valid_sets].max_drain_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX; - /* fclk wil be used to select WM*/ + /* We will not select WM based on fclk, so leave it as unconstrained */ + ranges->reader_wm_sets[num_valid_sets].min_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN; + ranges->reader_wm_sets[num_valid_sets].max_fill_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX; + /* dcfclk wil be used to select WM*/ if (ranges->reader_wm_sets[num_valid_sets].wm_type == WM_TYPE_PSTATE_CHG) { if (i == 0) - ranges->reader_wm_sets[num_valid_sets].min_fill_clk_mhz = 0; + ranges->reader_wm_sets[num_valid_sets].min_drain_clk_mhz = 0; else { /* add 1 to make it non-overlapping with next lvl */ - ranges->reader_wm_sets[num_valid_sets].min_fill_clk_mhz = bw_params->clk_table.entries[i - 1].fclk_mhz + 1; + ranges->reader_wm_sets[num_valid_sets].min_drain_clk_mhz = bw_params->clk_table.entries[i - 1].dcfclk_mhz + 1; } - ranges->reader_wm_sets[num_valid_sets].max_fill_clk_mhz = bw_params->clk_table.entries[i].fclk_mhz; + ranges->reader_wm_sets[num_valid_sets].max_drain_clk_mhz = bw_params->clk_table.entries[i].dcfclk_mhz; } else { /* unconstrained for memory retraining */ From 1a229156422172f70115d6ec29a7f814c3fb0913 Mon Sep 17 00:00:00 2001 From: Aric Cyr Date: Tue, 21 Jan 2020 22:50:13 -0500 Subject: [PATCH 48/69] drm/amd/display: Check engine is not NULL before acquiring [Why] Engine can be NULL in some cases, so we must not acquire it. [How] Check for NULL engine before acquiring. Signed-off-by: Aric Cyr Reviewed-by: Harry Wentland Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/dce/dce_aux.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_aux.c b/drivers/gpu/drm/amd/display/dc/dce/dce_aux.c index f1a5d2c6aa378..68c4049cbc2ad 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_aux.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_aux.c @@ -400,7 +400,7 @@ static bool acquire( { enum gpio_result result; - if (!is_engine_available(engine)) + if ((engine == NULL) || !is_engine_available(engine)) return false; result = dal_ddc_open(ddc, GPIO_MODE_HARDWARE, From 9027045bce8e5e1c12ba48893e01fc28767b0e60 Mon Sep 17 00:00:00 2001 From: Wenjing Liu Date: Wed, 22 Jan 2020 11:54:54 -0500 Subject: [PATCH 49/69] drm/amd/display: decouple global lock out of pipe control lock [why] hwss should not guess what type of pipe lock is needed. The caller of the lock function should know the right type of pipe lock. Decouple the setup of global lock outside of pipe control lock logic. Signed-off-by: Wenjing Liu Reviewed-by: Anthony Koo Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc.c | 20 ++++++++++++++ .../drm/amd/display/dc/dcn20/dcn20_hwseq.c | 27 ------------------- 2 files changed, 20 insertions(+), 27 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index c4e5bcb3e5413..b6aa0509624b0 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -2256,6 +2256,11 @@ static void commit_planes_for_stream(struct dc *dc, } } + if ((update_type != UPDATE_TYPE_FAST) && stream->update_flags.bits.dsc_changed) + if (top_pipe_to_program->stream_res.tg->funcs->lock_doublebuffer_enable) + top_pipe_to_program->stream_res.tg->funcs->lock_doublebuffer_enable( + top_pipe_to_program->stream_res.tg); + if ((update_type != UPDATE_TYPE_FAST) && dc->hwss.interdependent_update_lock) dc->hwss.interdependent_update_lock(dc, context, true); else @@ -2425,6 +2430,21 @@ static void commit_planes_for_stream(struct dc *dc, else dc->hwss.pipe_control_lock(dc, top_pipe_to_program, false); + if ((update_type != UPDATE_TYPE_FAST) && stream->update_flags.bits.dsc_changed) + if (top_pipe_to_program->stream_res.tg->funcs->lock_doublebuffer_enable) { + top_pipe_to_program->stream_res.tg->funcs->wait_for_state( + top_pipe_to_program->stream_res.tg, + CRTC_STATE_VACTIVE); + top_pipe_to_program->stream_res.tg->funcs->wait_for_state( + top_pipe_to_program->stream_res.tg, + CRTC_STATE_VBLANK); + top_pipe_to_program->stream_res.tg->funcs->wait_for_state( + top_pipe_to_program->stream_res.tg, + CRTC_STATE_VACTIVE); + top_pipe_to_program->stream_res.tg->funcs->lock_doublebuffer_disable( + top_pipe_to_program->stream_res.tg); + } + if (update_type != UPDATE_TYPE_FAST) dc->hwss.post_unlock_program_front_end(dc, context); diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c index d7cc7ccdaed64..e7a82d5955b91 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c @@ -1100,7 +1100,6 @@ void dcn20_pipe_control_lock( bool lock) { bool flip_immediate = false; - bool dig_update_required = false; /* use TG master update lock to lock everything on the TG * therefore only top pipe need to lock @@ -1138,19 +1137,6 @@ void dcn20_pipe_control_lock( (!flip_immediate && pipe->stream_res.gsl_group > 0)) dcn20_setup_gsl_group_as_lock(dc, pipe, flip_immediate); - if (pipe->stream && pipe->stream->update_flags.bits.dsc_changed) - dig_update_required = true; - - /* Need double buffer lock mode in order to synchronize front end pipe - * updates with dig updates. - */ - if (dig_update_required) { - if (lock) { - pipe->stream_res.tg->funcs->lock_doublebuffer_enable( - pipe->stream_res.tg); - } - } - if (pipe->plane_state != NULL && pipe->plane_state->triplebuffer_flips) { if (lock) pipe->stream_res.tg->funcs->triplebuffer_lock(pipe->stream_res.tg); @@ -1162,19 +1148,6 @@ void dcn20_pipe_control_lock( else pipe->stream_res.tg->funcs->unlock(pipe->stream_res.tg); } - - if (dig_update_required) { - if (!lock) { - pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, - CRTC_STATE_VACTIVE); - pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, - CRTC_STATE_VBLANK); - pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, - CRTC_STATE_VACTIVE); - pipe->stream_res.tg->funcs->lock_doublebuffer_disable( - pipe->stream_res.tg); - } - } } static void dcn20_detect_pipe_changes(struct pipe_ctx *old_pipe, struct pipe_ctx *new_pipe) From c481110b5b574c2864d78c2a6835640a2910d15b Mon Sep 17 00:00:00 2001 From: Eric Yang Date: Mon, 20 Jan 2020 12:56:43 -0500 Subject: [PATCH 50/69] drm/amd/display: fix inputting clk lvl into dml for RN [Why] Previous logic is only good for 15W parts. Other configuration need a smarter logic to match clk levels with pp table in the fuse. [How] Cache all 8 DPM level's clock data, find lvl that match each pstate in the pp table and build input into DML base on that Signed-off-by: Eric Yang Signed-off-by: Sung Lee Reviewed-by: Tony Cheng Acked-by: Bhawanpreet Lakha --- .../drm/amd/display/dc/dcn21/dcn21_resource.c | 165 +++++++++++++----- .../amd/display/dc/dml/display_mode_structs.h | 3 +- 2 files changed, 127 insertions(+), 41 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c index effc3d209388e..13b2f25fd7c9c 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c @@ -165,10 +165,10 @@ struct _vcs_dpi_soc_bounding_box_st dcn2_1_soc = { .clock_limits = { { .state = 0, - .dcfclk_mhz = 304.0, - .fabricclk_mhz = 600.0, - .dispclk_mhz = 618.0, - .dppclk_mhz = 440.0, + .dcfclk_mhz = 400.0, + .fabricclk_mhz = 400.0, + .dispclk_mhz = 600.0, + .dppclk_mhz = 400.00, .phyclk_mhz = 600.0, .socclk_mhz = 278.0, .dscclk_mhz = 205.67, @@ -176,10 +176,10 @@ struct _vcs_dpi_soc_bounding_box_st dcn2_1_soc = { }, { .state = 1, - .dcfclk_mhz = 304.0, - .fabricclk_mhz = 600.0, - .dispclk_mhz = 618.0, - .dppclk_mhz = 618.0, + .dcfclk_mhz = 464.52, + .fabricclk_mhz = 800.0, + .dispclk_mhz = 654.55, + .dppclk_mhz = 626.09, .phyclk_mhz = 600.0, .socclk_mhz = 278.0, .dscclk_mhz = 205.67, @@ -187,32 +187,65 @@ struct _vcs_dpi_soc_bounding_box_st dcn2_1_soc = { }, { .state = 2, - .dcfclk_mhz = 608.0, - .fabricclk_mhz = 1066.0, - .dispclk_mhz = 888.0, - .dppclk_mhz = 888.0, - .phyclk_mhz = 810.0, + .dcfclk_mhz = 514.29, + .fabricclk_mhz = 933.0, + .dispclk_mhz = 757.89, + .dppclk_mhz = 685.71, + .phyclk_mhz = 600.0, .socclk_mhz = 278.0, .dscclk_mhz = 287.67, - .dram_speed_mts = 2133.0, + .dram_speed_mts = 1866.0, }, { .state = 3, - .dcfclk_mhz = 676.0, - .fabricclk_mhz = 1600.0, - .dispclk_mhz = 1015.0, - .dppclk_mhz = 1015.0, - .phyclk_mhz = 810.0, + .dcfclk_mhz = 576.00, + .fabricclk_mhz = 1067.0, + .dispclk_mhz = 847.06, + .dppclk_mhz = 757.89, + .phyclk_mhz = 600.0, .socclk_mhz = 715.0, .dscclk_mhz = 318.334, - .dram_speed_mts = 4266.0, + .dram_speed_mts = 2134.0, }, { .state = 4, - .dcfclk_mhz = 810.0, + .dcfclk_mhz = 626.09, + .fabricclk_mhz = 1200.0, + .dispclk_mhz = 900.00, + .dppclk_mhz = 847.06, + .phyclk_mhz = 810.0, + .socclk_mhz = 953.0, + .dscclk_mhz = 489.0, + .dram_speed_mts = 2400.0, + }, + { + .state = 5, + .dcfclk_mhz = 685.71, + .fabricclk_mhz = 1333.0, + .dispclk_mhz = 1028.57, + .dppclk_mhz = 960.00, + .phyclk_mhz = 810.0, + .socclk_mhz = 278.0, + .dscclk_mhz = 287.67, + .dram_speed_mts = 2666.0, + }, + { + .state = 6, + .dcfclk_mhz = 757.89, + .fabricclk_mhz = 1467.0, + .dispclk_mhz = 1107.69, + .dppclk_mhz = 1028.57, + .phyclk_mhz = 810.0, + .socclk_mhz = 715.0, + .dscclk_mhz = 318.334, + .dram_speed_mts = 3200.0, + }, + { + .state = 7, + .dcfclk_mhz = 847.06, .fabricclk_mhz = 1600.0, .dispclk_mhz = 1395.0, - .dppclk_mhz = 1285.0, + .dppclk_mhz = 1285.00, .phyclk_mhz = 1325.0, .socclk_mhz = 953.0, .dscclk_mhz = 489.0, @@ -220,8 +253,8 @@ struct _vcs_dpi_soc_bounding_box_st dcn2_1_soc = { }, /*Extra state, no dispclk ramping*/ { - .state = 5, - .dcfclk_mhz = 810.0, + .state = 8, + .dcfclk_mhz = 847.06, .fabricclk_mhz = 1600.0, .dispclk_mhz = 1395.0, .dppclk_mhz = 1285.0, @@ -268,7 +301,7 @@ struct _vcs_dpi_soc_bounding_box_st dcn2_1_soc = { .xfc_bus_transport_time_us = 4, .xfc_xbuf_latency_tolerance_us = 4, .use_urgent_burst_bw = 1, - .num_states = 5 + .num_states = 9 }; #ifndef MAX @@ -1347,26 +1380,78 @@ static void update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_param { struct dcn21_resource_pool *pool = TO_DCN21_RES_POOL(dc->res_pool); struct clk_limit_table *clk_table = &bw_params->clk_table; - int i; + unsigned int i, j, k; + int closest_clk_lvl; + + // diags does not retrieve proper values from SMU + // cap states to 5 and make state 5 the max state + if (IS_FPGA_MAXIMUS_DC(dc->ctx->dce_environment) || IS_DIAG_DC(dc->ctx->dce_environment)) { + dcn2_1_soc.num_states = 5; + + dcn2_1_soc.clock_limits[5].state = 5; + dcn2_1_soc.clock_limits[5].dcfclk_mhz = 810.0; + dcn2_1_soc.clock_limits[5].fabricclk_mhz = 1600.0; + dcn2_1_soc.clock_limits[5].dispclk_mhz = 1395.0; + dcn2_1_soc.clock_limits[5].dppclk_mhz = 1285.0; + dcn2_1_soc.clock_limits[5].phyclk_mhz = 1325.0; + dcn2_1_soc.clock_limits[5].socclk_mhz = 953.0; + dcn2_1_soc.clock_limits[5].dscclk_mhz = 489.0; + dcn2_1_soc.clock_limits[5].dram_speed_mts = 4266.0; + } else { + 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; + + /* Vmin: leave lowest DCN clocks, override with dcfclk, fclk, memclk from fuse */ + dcn2_1_soc.clock_limits[0].state = 0; + dcn2_1_soc.clock_limits[0].dcfclk_mhz = clk_table->entries[0].dcfclk_mhz; + dcn2_1_soc.clock_limits[0].fabricclk_mhz = clk_table->entries[0].fclk_mhz; + dcn2_1_soc.clock_limits[0].socclk_mhz = clk_table->entries[0].socclk_mhz; + dcn2_1_soc.clock_limits[0].dram_speed_mts = clk_table->entries[0].memclk_mhz * 2; + + /* + * Other levels: find cloest DCN clocks that fit the given clock limit using dcfclk + * as indicater + */ - 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; + closest_clk_lvl = -1; + /* index currently being filled */ + k = 1; + for (i = 1; i < clk_table->num_entries; i++) { + /* loop backwards, skip duplicate state, +1 because SMU has precision issue */ + for (j = dcn2_1_soc.num_states - 2; j >= k; j--) { + if ((unsigned int) dcn2_1_soc.clock_limits[j].dcfclk_mhz <= clk_table->entries[i].dcfclk_mhz) { + closest_clk_lvl = j; + break; + } + } - for (i = 0; i < clk_table->num_entries; i++) { + /* if found a lvl that fits, use the DCN clks from it, if not, go to next clk limit*/ + if (closest_clk_lvl != -1) { + dcn2_1_soc.clock_limits[k].state = i; + dcn2_1_soc.clock_limits[k].dcfclk_mhz = clk_table->entries[i].dcfclk_mhz; + dcn2_1_soc.clock_limits[k].fabricclk_mhz = clk_table->entries[i].fclk_mhz; + dcn2_1_soc.clock_limits[k].socclk_mhz = clk_table->entries[i].socclk_mhz; + dcn2_1_soc.clock_limits[k].dram_speed_mts = clk_table->entries[i].memclk_mhz * 2; + + dcn2_1_soc.clock_limits[k].dispclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dispclk_mhz; + dcn2_1_soc.clock_limits[k].dppclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dppclk_mhz; + dcn2_1_soc.clock_limits[k].dram_bw_per_chan_gbps = dcn2_1_soc.clock_limits[closest_clk_lvl].dram_bw_per_chan_gbps; + dcn2_1_soc.clock_limits[k].dscclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dscclk_mhz; + dcn2_1_soc.clock_limits[k].dtbclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].dtbclk_mhz; + dcn2_1_soc.clock_limits[k].phyclk_d18_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].phyclk_d18_mhz; + dcn2_1_soc.clock_limits[k].phyclk_mhz = dcn2_1_soc.clock_limits[closest_clk_lvl].phyclk_mhz; + k++; + } + } - dcn2_1_soc.clock_limits[i].state = i; - dcn2_1_soc.clock_limits[i].dcfclk_mhz = clk_table->entries[i].dcfclk_mhz; - dcn2_1_soc.clock_limits[i].fabricclk_mhz = clk_table->entries[i].fclk_mhz; - dcn2_1_soc.clock_limits[i].socclk_mhz = clk_table->entries[i].socclk_mhz; - dcn2_1_soc.clock_limits[i].dram_speed_mts = clk_table->entries[i].memclk_mhz * 2; + /* duplicate last level */ + dcn2_1_soc.clock_limits[k] = dcn2_1_soc.clock_limits[k - 1]; + dcn2_1_soc.clock_limits[k].state = k; + dcn2_1_soc.num_states = k + 1; } - dcn2_1_soc.clock_limits[i] = dcn2_1_soc.clock_limits[i - 1]; - dcn2_1_soc.num_states = i; - // diags does not retrieve proper values from SMU, do not update DML instance for diags - if (!IS_FPGA_MAXIMUS_DC(dc->ctx->dce_environment) && !IS_DIAG_DC(dc->ctx->dce_environment)) - dml_init_instance(&dc->dml, &dcn2_1_soc, &dcn2_1_ip, DML_PROJECT_DCN21); + dml_init_instance(&dc->dml, &dcn2_1_soc, &dcn2_1_ip, DML_PROJECT_DCN21); } /* Temporary Place holder until we can get them from fuse */ diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h index 8531531f94a91..0922f4037cadd 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h +++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h @@ -25,7 +25,7 @@ #ifndef __DISPLAY_MODE_STRUCTS_H__ #define __DISPLAY_MODE_STRUCTS_H__ -#define MAX_CLOCK_LIMIT_STATES 8 +#define MAX_CLOCK_LIMIT_STATES 9 typedef struct _vcs_dpi_voltage_scaling_st voltage_scaling_st; typedef struct _vcs_dpi_soc_bounding_box_st soc_bounding_box_st; @@ -61,6 +61,7 @@ struct _vcs_dpi_voltage_scaling_st { double dram_speed_mts; double fabricclk_mhz; double dispclk_mhz; + double dram_bw_per_chan_gbps; double phyclk_mhz; double dppclk_mhz; double dtbclk_mhz; From 87e78a77991d7c7d7291ed8a804f42de2e585b82 Mon Sep 17 00:00:00 2001 From: Yongqiang Sun Date: Wed, 22 Jan 2020 09:24:40 -0500 Subject: [PATCH 51/69] drm/amd/display: Check hyperV flag in DC. [Why] hyperV flag should be passed from dm to DC, and override the nv12 flip workaround flag. [How] Add flag to phy address config struct and pass the value in dm. Signed-off-by: Yongqiang Sun Reviewed-by: Tony Cheng Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c | 3 +++ drivers/gpu/drm/amd/display/dc/dc.h | 1 + 2 files changed, 4 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c b/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c index f2b39ec35c898..64cf24a9ab08e 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_vm_helper.c @@ -47,6 +47,9 @@ int dc_setup_system_context(struct dc *dc, struct dc_phy_addr_space_config *pa_c */ memcpy(&dc->vm_pa_config, pa_config, sizeof(struct dc_phy_addr_space_config)); dc->vm_pa_config.valid = true; + + if (pa_config->is_hvm_enabled == 0) + dc->debug.nv12_iflip_vm_wa = false; } return num_vmids; diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 88578a1d868d5..8be4266a7c954 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -466,6 +466,7 @@ struct dc_phy_addr_space_config { } gart_config; bool valid; + bool is_hvm_enabled; uint64_t page_table_default_page_addr; }; From e492493957c3f59999d71800ceb4ea7024d266e0 Mon Sep 17 00:00:00 2001 From: Sung Lee Date: Thu, 23 Jan 2020 15:31:43 -0500 Subject: [PATCH 52/69] drm/amd/display: Add wm ranges to clk_mgr [WHY & HOW] Having watermark ranges saved inside clk_mgr to be available for debug at all times would be useful. Add it to the clk_mgr_internal struct for reference. Only populated for Renoir, unused for other asics. Change-Id: I2982d0047816186b86ffdbaccb6a74617fd0938c Signed-off-by: Sung Lee Reviewed-by: Yongqiang Sun Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c | 5 ++--- drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h | 2 ++ 2 files changed, 4 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c index 034a5852a4168..8ecb98c29eb93 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c @@ -459,16 +459,15 @@ void build_watermark_ranges(struct clk_bw_params *bw_params, struct pp_smu_wm_ra static void rn_notify_wm_ranges(struct clk_mgr *clk_mgr_base) { struct dc_debug_options *debug = &clk_mgr_base->ctx->dc->debug; - struct pp_smu_wm_range_sets ranges = {0}; struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base); struct pp_smu_funcs *pp_smu = clk_mgr->pp_smu; if (!debug->disable_pplib_wm_range) { - build_watermark_ranges(clk_mgr_base->bw_params, &ranges); + build_watermark_ranges(clk_mgr_base->bw_params, &clk_mgr_base->ranges); /* Notify PP Lib/SMU which Watermarks to use for which clock ranges */ if (pp_smu && pp_smu->rn_funcs.set_wm_ranges) - pp_smu->rn_funcs.set_wm_ranges(&pp_smu->rn_funcs.pp_smu, &ranges); + pp_smu->rn_funcs.set_wm_ranges(&pp_smu->rn_funcs.pp_smu, &clk_mgr_base->ranges); } } diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h index c10cb4b54fae3..06084f847210d 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h @@ -27,6 +27,7 @@ #define __DAL_CLK_MGR_H__ #include "dc.h" +#include "dm_pp_smu.h" #define DCN_MINIMUM_DISPCLK_Khz 100000 #define DCN_MINIMUM_DPPCLK_Khz 100000 @@ -199,6 +200,7 @@ struct clk_mgr { struct clk_state_registers_and_bypass boot_snapshot; struct clk_bw_params *bw_params; #endif + struct pp_smu_wm_range_sets ranges; }; /* forward declarations */ From ad760b43b356f59ffbf318ec965351b6cab6e233 Mon Sep 17 00:00:00 2001 From: David Galiffi Date: Tue, 21 Jan 2020 16:29:52 -0500 Subject: [PATCH 53/69] drm/amd/display: Use uint64_t logger_mask instead of uint32_t [WHY] enum dc_log_type has more than 32 entries. User cannot set larger entries, like LOG_DSC. Logs from LOG_GAMMA_DEBUG where being printed even though flag was not enabled, because dal_logger_should_log check erroneously passed. [HOW] Change struct dal_logger.mask and struct dc_init_data.mask to uint64_t. Change-Id: Ic532330c9daca0ca1bd588b28e6142ef546f945e Signed-off-by: David Galiffi Reviewed-by: Tony Cheng Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/dc.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 8be4266a7c954..3349e55de664b 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -584,7 +584,8 @@ struct dc_init_data { struct dc_reg_helper_state *dmub_offload; struct dc_config flags; - uint32_t log_mask; + uint64_t log_mask; + #ifdef CONFIG_DRM_AMD_DC_DCN2_0 /** * gpu_info FW provided soc bounding box struct or 0 if not From b2906b1dda1c8f4bf2ea3dc9a09e300ecc978f03 Mon Sep 17 00:00:00 2001 From: Yongqiang Sun Date: Thu, 23 Jan 2020 16:30:15 -0500 Subject: [PATCH 54/69] drm/amd/display: Limit minimum DPPCLK to 100MHz. [Why] Underflow is observed when plug in a 4K@60 monitor with 1366x768 eDP due to DPPCLK is too low. [How] Limit minimum DPPCLK to 100MHz. Signed-off-by: Yongqiang Sun Reviewed-by: Eric Yang Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c index 8ecb98c29eb93..5d82ec1f1ce54 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c @@ -151,6 +151,12 @@ void rn_update_clocks(struct clk_mgr *clk_mgr_base, rn_vbios_smu_set_min_deep_sleep_dcfclk(clk_mgr, clk_mgr_base->clks.dcfclk_deep_sleep_khz); } + // workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow. + if (!IS_DIAG_DC(dc->ctx->dce_environment)) { + if (new_clocks->dppclk_khz < 100000) + new_clocks->dppclk_khz = 100000; + } + if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) { if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz) dpp_clock_lowered = true; From cb0fe1b61b2cf6949bb2fdf5f66af4f958b0bbc1 Mon Sep 17 00:00:00 2001 From: Aric Cyr Date: Mon, 27 Jan 2020 10:12:36 -0500 Subject: [PATCH 55/69] drm/amd/display: 3.2.71 Signed-off-by: Aric Cyr Reviewed-by: Aric Cyr Acked-by: Bhawanpreet Lakha --- 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 3349e55de664b..440098d145451 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -39,7 +39,7 @@ #include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#define DC_VER "3.2.70" +#define DC_VER "3.2.71" #define MAX_SURFACES 3 #define MAX_PLANES 6 From 2f836e04c7a965ed24f0f06e34e48357eb3749c3 Mon Sep 17 00:00:00 2001 From: Nicholas Kazlauskas Date: Wed, 15 Jan 2020 14:39:52 -0500 Subject: [PATCH 56/69] drm/amd/display: Add DMUB tracebuffer debugfs [Why] The DMUB tracebuffer is useful for understanding DMCUB execution state. [How] Add a "show" attribute debugfs so we can loop through the buffer and print the entries. The structs for the entry format are defined in the debugfs since the tracebuffer header no longer exists in the DMUB service. Change-Id: Ic346d081a53277b4decdd8ead70b802adea6ed61 Signed-off-by: Nicholas Kazlauskas Reviewed-by: Hersen Wu Acked-by: Bhawanpreet Lakha --- .../amd/display/amdgpu_dm/amdgpu_dm_debugfs.c | 61 +++++++++++++++++++ 1 file changed, 61 insertions(+) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c index 6c9713f1c9214..3ca2ea6b33ad2 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c @@ -32,6 +32,19 @@ #include "amdgpu_dm.h" #include "amdgpu_dm_debugfs.h" #include "dm_helpers.h" +#include "dmub/inc/dmub_srv.h" + +struct dmub_debugfs_trace_header { + uint32_t entry_count; + uint32_t reserved[3]; +}; + +struct dmub_debugfs_trace_entry { + uint32_t trace_code; + uint32_t tick_count; + uint32_t param0; + uint32_t param1; +}; /* function description * get/ set DP configuration: lane_count, link_rate, spread_spectrum @@ -675,6 +688,50 @@ static ssize_t dp_phy_test_pattern_debugfs_write(struct file *f, const char __us return bytes_from_user; } +/** + * Returns the DMCUB tracebuffer contents. + * Example usage: cat /sys/kernel/debug/dri/0/amdgpu_dm_dmub_tracebuffer + */ +static int dmub_tracebuffer_show(struct seq_file *m, void *data) +{ + struct amdgpu_device *adev = m->private; + struct dmub_srv_fb_info *fb_info = adev->dm.dmub_fb_info; + struct dmub_debugfs_trace_entry *entries; + uint8_t *tbuf_base; + uint32_t tbuf_size, max_entries, num_entries, i; + + if (!fb_info) + return 0; + + tbuf_base = (uint8_t *)fb_info->fb[DMUB_WINDOW_5_TRACEBUFF].cpu_addr; + if (!tbuf_base) + return 0; + + tbuf_size = fb_info->fb[DMUB_WINDOW_5_TRACEBUFF].size; + max_entries = (tbuf_size - sizeof(struct dmub_debugfs_trace_header)) / + sizeof(struct dmub_debugfs_trace_entry); + + num_entries = + ((struct dmub_debugfs_trace_header *)tbuf_base)->entry_count; + + num_entries = min(num_entries, max_entries); + + entries = (struct dmub_debugfs_trace_entry + *)(tbuf_base + + sizeof(struct dmub_debugfs_trace_header)); + + for (i = 0; i < num_entries; ++i) { + struct dmub_debugfs_trace_entry *entry = &entries[i]; + + seq_printf(m, + "trace_code=%u tick_count=%u param0=%u param1=%u\n", + entry->trace_code, entry->tick_count, entry->param0, + entry->param1); + } + + return 0; +} + /* * Returns the current and maximum output bpc for the connector. * Example usage: cat /sys/kernel/debug/dri/0/DP-1/output_bpc @@ -881,6 +938,7 @@ static ssize_t dp_dpcd_data_read(struct file *f, char __user *buf, } #ifdef DEFINE_SHOW_ATTRIBUTE +DEFINE_SHOW_ATTRIBUTE(dmub_tracebuffer); DEFINE_SHOW_ATTRIBUTE(output_bpc); DEFINE_SHOW_ATTRIBUTE(vrr_range); #endif @@ -1209,6 +1267,9 @@ int dtn_debugfs_init(struct amdgpu_device *adev) #ifdef DEFINE_DEBUGFS_ATTRIBUTE debugfs_create_file_unsafe("amdgpu_dm_visual_confirm", 0644, root, adev, &visual_confirm_fops); + + debugfs_create_file_unsafe("amdgpu_dm_dmub_tracebuffer", 0644, root, + adev, &dmub_tracebuffer_fops); #endif return 0; From 78b93b560494326802ab59ca315f40caa456d62c Mon Sep 17 00:00:00 2001 From: Jing Zhou Date: Tue, 21 Jan 2020 18:05:35 +0800 Subject: [PATCH 57/69] drm/amd/display: external monitor abm enabled in modern standby [why] Resume from modern standby, edp stream disabled but abm keep enabled. External monitor select OTG source 0 which ABM enabled. [how] Disable abm before disable crtc when reset path mode not call core link disable stream. Signed-off-by: Jing Zhou Reviewed-by: Anthony Koo Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c | 4 ++++ drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 4 ++++ 2 files changed, 8 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c index 5dd89d0dc6417..8c7592e43ca0d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c @@ -911,6 +911,10 @@ static void dcn10_reset_back_end_for_pipe( * parent pipe. */ if (pipe_ctx->top_pipe == NULL) { + + if (pipe_ctx->stream_res.abm) + pipe_ctx->stream_res.abm->funcs->set_abm_immediate_disable(pipe_ctx->stream_res.abm); + pipe_ctx->stream_res.tg->funcs->disable_crtc(pipe_ctx->stream_res.tg); pipe_ctx->stream_res.tg->funcs->enable_optc_clock(pipe_ctx->stream_res.tg, false); diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c index e7a82d5955b91..db17df21fddcf 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c @@ -2023,6 +2023,10 @@ static void dcn20_reset_back_end_for_pipe( * parent pipe. */ if (pipe_ctx->top_pipe == NULL) { + + if (pipe_ctx->stream_res.abm) + pipe_ctx->stream_res.abm->funcs->set_abm_immediate_disable(pipe_ctx->stream_res.abm); + pipe_ctx->stream_res.tg->funcs->disable_crtc(pipe_ctx->stream_res.tg); pipe_ctx->stream_res.tg->funcs->enable_optc_clock(pipe_ctx->stream_res.tg, false); From 00610879e5583182f2fef47a155bc724fc36bea3 Mon Sep 17 00:00:00 2001 From: Hersen Wu Date: Tue, 14 Jan 2020 15:39:07 -0500 Subject: [PATCH 58/69] drm/amd/display: linux enable oled panel support dc part [Why] old panel has been enabled for window driver but not linux. [How] enable oled panel support for linux. this patch is dc part. Change-Id: Id181d51188d24ed7a86c857c9b39ba17a7685df1 Signed-off-by: Hersen Wu Reviewed-by: Harry Wentland Reviewed-by: Hersen Wu Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/core/dc_link.c | 32 +++- .../gpu/drm/amd/display/dc/core/dc_link_dp.c | 168 ++++++++++++++++++ drivers/gpu/drm/amd/display/dc/dc.h | 14 ++ drivers/gpu/drm/amd/display/dc/dc_dp_types.h | 34 ++++ drivers/gpu/drm/amd/display/dc/dc_link.h | 16 ++ .../display/dc/dce110/dce110_hw_sequencer.c | 14 ++ .../gpu/drm/amd/display/dc/inc/dc_link_dp.h | 2 + 7 files changed, 279 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c index b21c176ccad9f..ac39b19ab4a92 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c @@ -599,6 +599,9 @@ static bool detect_dp( if (sink_caps->transaction_type == DDC_TRANSACTION_TYPE_I2C_OVER_AUX) { sink_caps->signal = SIGNAL_TYPE_DISPLAY_PORT; + + dpcd_set_source_specific_data(link); + if (!detect_dp_sink_caps(link)) return false; @@ -769,8 +772,16 @@ static bool dc_link_detect_helper(struct dc_link *link, if ((link->connector_signal == SIGNAL_TYPE_LVDS || link->connector_signal == SIGNAL_TYPE_EDP) && - link->local_sink) + link->local_sink) { + + // need to re-write OUI and brightness in resume case + if (link->connector_signal == SIGNAL_TYPE_EDP) { + dpcd_set_source_specific_data(link); + dc_link_set_default_brightness_aux(link); //TODO: use cached + } + return true; + } if (false == dc_link_detect_sink(link, &new_connection_type)) { BREAK_TO_DEBUGGER(); @@ -818,6 +829,10 @@ static bool dc_link_detect_helper(struct dc_link *link, } case SIGNAL_TYPE_EDP: { + read_current_link_settings_on_detect(link); + + dpcd_set_source_specific_data(link); + detect_edp_sink_caps(link); read_current_link_settings_on_detect(link); sink_caps.transaction_type = DDC_TRANSACTION_TYPE_I2C_OVER_AUX; @@ -1495,6 +1510,7 @@ static enum dc_status enable_link_dp( #endif int i; bool apply_seamless_boot_optimization = false; + uint32_t bl_oled_enable_delay = 50; // in ms // check for seamless boot for (i = 0; i < state->stream_count; i++) { @@ -1518,6 +1534,9 @@ static enum dc_status enable_link_dp( if (state->clk_mgr && !apply_seamless_boot_optimization) state->clk_mgr->funcs->update_clocks(state->clk_mgr, state, false); + // during mode switch we do DP_SET_POWER off then on, and OUI is lost + dpcd_set_source_specific_data(link); + skip_video_pattern = true; if (link_settings.link_rate == LINK_RATE_LOW) @@ -1543,6 +1562,17 @@ static enum dc_status enable_link_dp( dp_set_fec_enable(link, fec_enable); #endif + + // during mode set we do DP_SET_POWER off then on, aux writes are lost + if (link->dpcd_sink_ext_caps.bits.oled == 1 || + link->dpcd_sink_ext_caps.bits.sdr_aux_backlight_control == 1 || + link->dpcd_sink_ext_caps.bits.hdr_aux_backlight_control == 1) { + dc_link_set_default_brightness_aux(link); // TODO: use cached if known + if (link->dpcd_sink_ext_caps.bits.oled == 1) + msleep(bl_oled_enable_delay); + dc_link_backlight_enable_aux(link, true); + } + return status; } diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c index b94c8a14571c8..fbcdb8f89ca8b 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c @@ -24,6 +24,14 @@ #define DP_REPEATER_CONFIGURATION_AND_STATUS_SIZE 0x50 +#define DP_SOURCE_TABLE_REVISION 0x310 +#define DP_SOURCE_PAYLOAD_SIZE 0x311 +#define DP_SOURCE_SINK_CAP 0x317 +#define DP_SOURCE_BACKLIGHT_LEVEL 0x320 +#define DP_SOURCE_BACKLIGHT_CURRENT_PEAK 0x326 +#define DP_SOURCE_BACKLIGHT_CONTROL 0x32E +#define DP_SOURCE_BACKLIGHT_ENABLE 0x32F + /* maximum pre emphasis level allowed for each voltage swing level*/ static const enum dc_pre_emphasis voltage_swing_to_pre_emphasis[] = { PRE_EMPHASIS_LEVEL3, @@ -3180,6 +3188,23 @@ static void dp_wa_power_up_0010FA(struct dc_link *link, uint8_t *dpcd_data, link->wa_flags.dp_keep_receiver_powered = false; } +/* Read additional sink caps defined in source specific DPCD area + * This function currently only reads from SinkCapability address (DP_SOURCE_SINK_CAP) + */ +static bool dpcd_read_sink_ext_caps(struct dc_link *link) +{ + uint8_t dpcd_data; + + if (!link) + return false; + + if (core_link_read_dpcd(link, DP_SOURCE_SINK_CAP, &dpcd_data, 1) != DC_OK) + return false; + + link->dpcd_sink_ext_caps.raw = dpcd_data; + return true; +} + static bool retrieve_link_cap(struct dc_link *link) { /* DP_ADAPTER_CAP - DP_DPCD_REV + 1 == 16 and also DP_DSC_BITS_PER_PIXEL_INC - DP_DSC_SUPPORT + 1 == 16, @@ -3454,6 +3479,9 @@ static bool retrieve_link_cap(struct dc_link *link) } #endif + if (!dpcd_read_sink_ext_caps(link)) + link->dpcd_sink_ext_caps.raw = 0; + /* Connectivity log: detection */ CONN_DATA_DETECT(link, dpcd_data, sizeof(dpcd_data), "Rx Caps: "); @@ -3606,6 +3634,8 @@ void detect_edp_sink_caps(struct dc_link *link) } } link->verified_link_cap = link->reported_link_cap; + + dc_link_set_default_brightness_aux(link); } void dc_link_dp_enable_hpd(const struct dc_link *link) @@ -4171,3 +4201,141 @@ void dp_set_fec_enable(struct dc_link *link, bool enable) } #endif +void dpcd_set_source_specific_data(struct dc_link *link) +{ + struct dpcd_amd_signature amd_signature; + const uint32_t post_oui_delay = 30; // 30ms + + amd_signature.AMD_IEEE_TxSignature_byte1 = 0x0; + amd_signature.AMD_IEEE_TxSignature_byte2 = 0x0; + amd_signature.AMD_IEEE_TxSignature_byte3 = 0x1A; + amd_signature.device_id_byte1 = + (uint8_t)(link->ctx->asic_id.chip_id); + amd_signature.device_id_byte2 = + (uint8_t)(link->ctx->asic_id.chip_id >> 8); + memset(&amd_signature.zero, 0, 4); + amd_signature.dce_version = + (uint8_t)(link->ctx->dce_version); + amd_signature.dal_version_byte1 = 0x0; // needed? where to get? + amd_signature.dal_version_byte2 = 0x0; // needed? where to get? + + core_link_write_dpcd(link, DP_SOURCE_OUI, + (uint8_t *)(&amd_signature), + sizeof(amd_signature)); + + // Sink may need to configure internals based on vendor, so allow some + // time before proceeding with possibly vendor specific transactions + msleep(post_oui_delay); +} + +bool dc_link_set_backlight_level_nits(struct dc_link *link, + bool isHDR, + uint32_t backlight_millinits, + uint32_t transition_time_in_ms) +{ + struct dpcd_source_backlight_set dpcd_backlight_set; + uint8_t backlight_control = isHDR ? 1 : 0; + + if (!link || (link->connector_signal != SIGNAL_TYPE_EDP && + link->connector_signal != SIGNAL_TYPE_DISPLAY_PORT)) + return false; + + // OLEDs have no PWM, they can only use AUX + if (link->dpcd_sink_ext_caps.bits.oled == 1) + backlight_control = 1; + + *(uint32_t *)&dpcd_backlight_set.backlight_level_millinits = backlight_millinits; + *(uint16_t *)&dpcd_backlight_set.backlight_transition_time_ms = (uint16_t)transition_time_in_ms; + + + if (core_link_write_dpcd(link, DP_SOURCE_BACKLIGHT_LEVEL, + (uint8_t *)(&dpcd_backlight_set), + sizeof(dpcd_backlight_set)) != DC_OK) + return false; + + if (core_link_write_dpcd(link, DP_SOURCE_BACKLIGHT_CONTROL, + &backlight_control, 1) != DC_OK) + return false; + + return true; +} + +bool dc_link_get_backlight_level_nits(struct dc_link *link, + uint32_t *backlight_millinits_avg, + uint32_t *backlight_millinits_peak) +{ + union dpcd_source_backlight_get dpcd_backlight_get; + + memset(&dpcd_backlight_get, 0, sizeof(union dpcd_source_backlight_get)); + + if (!link || (link->connector_signal != SIGNAL_TYPE_EDP && + link->connector_signal != SIGNAL_TYPE_DISPLAY_PORT)) + return false; + + if (!core_link_read_dpcd(link, DP_SOURCE_BACKLIGHT_CURRENT_PEAK, + dpcd_backlight_get.raw, + sizeof(union dpcd_source_backlight_get))) + return false; + + *backlight_millinits_avg = + dpcd_backlight_get.bytes.backlight_millinits_avg; + *backlight_millinits_peak = + dpcd_backlight_get.bytes.backlight_millinits_peak; + + /* On non-supported panels dpcd_read usually succeeds with 0 returned */ + if (*backlight_millinits_avg == 0 || + *backlight_millinits_avg > *backlight_millinits_peak) + return false; + + return true; +} + +bool dc_link_backlight_enable_aux(struct dc_link *link, bool enable) +{ + uint8_t backlight_enable = enable ? 1 : 0; + + if (!link || (link->connector_signal != SIGNAL_TYPE_EDP && + link->connector_signal != SIGNAL_TYPE_DISPLAY_PORT)) + return false; + + if (core_link_write_dpcd(link, DP_SOURCE_BACKLIGHT_ENABLE, + &backlight_enable, 1) != DC_OK) + return false; + + return true; +} + +// we read default from 0x320 because we expect BIOS wrote it there +// regular get_backlight_nit reads from panel set at 0x326 +bool dc_link_read_default_bl_aux(struct dc_link *link, uint32_t *backlight_millinits) +{ + if (!link || (link->connector_signal != SIGNAL_TYPE_EDP && + link->connector_signal != SIGNAL_TYPE_DISPLAY_PORT)) + return false; + + if (!core_link_read_dpcd(link, DP_SOURCE_BACKLIGHT_LEVEL, + (uint8_t *) backlight_millinits, + sizeof(uint32_t))) + return false; + + return true; +} + +bool dc_link_set_default_brightness_aux(struct dc_link *link) +{ + uint32_t default_backlight; + + if (link && + (link->dpcd_sink_ext_caps.bits.hdr_aux_backlight_control == 1 || + link->dpcd_sink_ext_caps.bits.sdr_aux_backlight_control == 1)) { + if (!dc_link_read_default_bl_aux(link, &default_backlight)) + default_backlight = 150000; + // if < 5 nits or > 5000, it might be wrong readback + if (default_backlight < 5000 || default_backlight > 5000000) + default_backlight = 150000; // + + return dc_link_set_backlight_level_nits(link, true, + default_backlight, 0); + } + return false; +} diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 440098d145451..0c438a6cf4d2d 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -1026,6 +1026,20 @@ struct dpcd_caps { }; +union dpcd_sink_ext_caps { + struct { + /* 0 - Sink supports backlight adjust via PWM during SDR/HDR mode + * 1 - Sink supports backlight adjust via AUX during SDR/HDR mode. + */ + uint8_t sdr_aux_backlight_control : 1; + uint8_t hdr_aux_backlight_control : 1; + uint8_t reserved_1 : 2; + uint8_t oled : 1; + uint8_t reserved : 3; + } bits; + uint8_t raw; +}; + #include "dc_link.h" /******************************************************************************* diff --git a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h index 28234d8fdb2c5..c6a2cee1019f6 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h @@ -434,6 +434,40 @@ struct dp_sink_hw_fw_revision { uint8_t ieee_fw_rev[2]; }; +struct dpcd_amd_signature { + uint8_t AMD_IEEE_TxSignature_byte1; + uint8_t AMD_IEEE_TxSignature_byte2; + uint8_t AMD_IEEE_TxSignature_byte3; + uint8_t device_id_byte1; + uint8_t device_id_byte2; + uint8_t zero[4]; + uint8_t dce_version; + uint8_t dal_version_byte1; + uint8_t dal_version_byte2; +}; + +struct dpcd_source_backlight_set { + struct { + uint8_t byte0; + uint8_t byte1; + uint8_t byte2; + uint8_t byte3; + } backlight_level_millinits; + + struct { + uint8_t byte0; + uint8_t byte1; + } backlight_transition_time_ms; +}; + +union dpcd_source_backlight_get { + struct { + uint32_t backlight_millinits_peak; /* 326h */ + uint32_t backlight_millinits_avg; /* 32Ah */ + } bytes; + uint8_t raw[8]; +}; + /*DPCD register of DP receiver capability field bits-*/ union edp_configuration_cap { struct { diff --git a/drivers/gpu/drm/amd/display/dc/dc_link.h b/drivers/gpu/drm/amd/display/dc/dc_link.h index 16d984f752427..725ad43d2de12 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_link.h +++ b/drivers/gpu/drm/amd/display/dc/dc_link.h @@ -130,6 +130,7 @@ struct dc_link { enum edp_revision edp_revision; bool psr_feature_enabled; bool psr_allow_active; + union dpcd_sink_ext_caps dpcd_sink_ext_caps; /* MST record stream using this link */ struct link_flags { @@ -182,6 +183,21 @@ bool dc_link_set_backlight_level(const struct dc_link *dc_link, uint32_t backlight_pwm_u16_16, uint32_t frame_ramp); +/* Set/get nits-based backlight level of an embedded panel (eDP, LVDS). */ +bool dc_link_set_backlight_level_nits(struct dc_link *link, + bool isHDR, + uint32_t backlight_millinits, + uint32_t transition_time_in_ms); + +bool dc_link_get_backlight_level_nits(struct dc_link *link, + uint32_t *backlight_millinits, + uint32_t *backlight_millinits_peak); + +bool dc_link_backlight_enable_aux(struct dc_link *link, bool enable); + +bool dc_link_read_default_bl_aux(struct dc_link *link, uint32_t *backlight_millinits); +bool dc_link_set_default_brightness_aux(struct dc_link *link); + int dc_link_get_backlight_level(const struct dc_link *dc_link); bool dc_link_set_abm_disable(const struct dc_link *dc_link); diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c index 064970b71d121..69c6589dfbe49 100644 --- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c @@ -71,6 +71,8 @@ #define PANEL_POWER_UP_TIMEOUT 300 #define PANEL_POWER_DOWN_TIMEOUT 500 #define HPD_CHECK_INTERVAL 10 +#define OLED_POST_T7_DELAY 100 +#define OLED_PRE_T11_DELAY 150 #define CTX \ hws->ctx @@ -936,9 +938,21 @@ void dce110_edp_backlight_control( if (cntl.action == TRANSMITTER_CONTROL_BACKLIGHT_ON) edp_receiver_ready_T7(link); link_transmitter_control(ctx->dc_bios, &cntl); + + if (enable && link->dpcd_sink_ext_caps.bits.oled) + msleep(OLED_POST_T7_DELAY); + + if (link->dpcd_sink_ext_caps.bits.oled || + link->dpcd_sink_ext_caps.bits.hdr_aux_backlight_control == 1 || + link->dpcd_sink_ext_caps.bits.sdr_aux_backlight_control == 1) + dc_link_backlight_enable_aux(link, enable); + /*edp 1.2*/ if (cntl.action == TRANSMITTER_CONTROL_BACKLIGHT_OFF) edp_receiver_ready_T9(link); + + if (!enable && link->dpcd_sink_ext_caps.bits.oled) + msleep(OLED_PRE_T11_DELAY); } void dce110_enable_audio_stream(struct pipe_ctx *pipe_ctx) diff --git a/drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h b/drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h index a7b6abbb5a730..8e3eb24e08187 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h @@ -78,6 +78,8 @@ void dp_set_panel_mode(struct dc_link *link, enum dp_panel_mode panel_mode); bool dp_overwrite_extended_receiver_cap(struct dc_link *link); +void dpcd_set_source_specific_data(struct dc_link *link); + #ifdef CONFIG_DRM_AMD_DC_DSC_SUPPORT void dp_set_fec_ready(struct dc_link *link, bool ready); void dp_set_fec_enable(struct dc_link *link, bool enable); From 8657f4e0ade9e16498ace9fab5a1cf35e0a78ffc Mon Sep 17 00:00:00 2001 From: Tony Cheng Date: Tue, 28 Jan 2020 16:00:22 +0800 Subject: [PATCH 59/69] drm/amd/display: fix workaround for incorrect double buffer register for DLG ADL and TTU [Why] these registers should have been double buffered. SW workaround we will have SW program the more aggressive (lower) values whenever we are upating this register, so we will not have underflow at expense of less optimzal request pattern. [How] there is a driver bug where we don't check for 0, which is uninitialzed HW default. since 0 is smaller than any value we need to program, driver end up with not programming these registers Signed-off-by: Tony Cheng Reviewed-by: Yongqiang Sun Acked-by: Bhawanpreet Lakha --- .../gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c | 35 +++++++++++++------ 1 file changed, 25 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c index cf09b93357280..aa7b0e7eb945d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hubp.c @@ -79,32 +79,47 @@ void apply_DEDCN21_142_wa_for_hostvm_deadline( struct _vcs_dpi_display_dlg_regs_st *dlg_attr) { struct dcn21_hubp *hubp21 = TO_DCN21_HUBP(hubp); - uint32_t cur_value; + uint32_t refcyc_per_vm_group_vblank; + uint32_t refcyc_per_vm_req_vblank; + uint32_t refcyc_per_vm_group_flip; + uint32_t refcyc_per_vm_req_flip; + const uint32_t uninitialized_hw_default = 0; - REG_GET(VBLANK_PARAMETERS_5, REFCYC_PER_VM_GROUP_VBLANK, &cur_value); - if (cur_value > dlg_attr->refcyc_per_vm_group_vblank) + REG_GET(VBLANK_PARAMETERS_5, + REFCYC_PER_VM_GROUP_VBLANK, &refcyc_per_vm_group_vblank); + + if (refcyc_per_vm_group_vblank == uninitialized_hw_default || + refcyc_per_vm_group_vblank > dlg_attr->refcyc_per_vm_group_vblank) REG_SET(VBLANK_PARAMETERS_5, 0, REFCYC_PER_VM_GROUP_VBLANK, dlg_attr->refcyc_per_vm_group_vblank); REG_GET(VBLANK_PARAMETERS_6, - REFCYC_PER_VM_REQ_VBLANK, - &cur_value); - if (cur_value > dlg_attr->refcyc_per_vm_req_vblank) + REFCYC_PER_VM_REQ_VBLANK, &refcyc_per_vm_req_vblank); + + if (refcyc_per_vm_req_vblank == uninitialized_hw_default || + refcyc_per_vm_req_vblank > dlg_attr->refcyc_per_vm_req_vblank) REG_SET(VBLANK_PARAMETERS_6, 0, REFCYC_PER_VM_REQ_VBLANK, dlg_attr->refcyc_per_vm_req_vblank); - REG_GET(FLIP_PARAMETERS_3, REFCYC_PER_VM_GROUP_FLIP, &cur_value); - if (cur_value > dlg_attr->refcyc_per_vm_group_flip) + REG_GET(FLIP_PARAMETERS_3, + REFCYC_PER_VM_GROUP_FLIP, &refcyc_per_vm_group_flip); + + if (refcyc_per_vm_group_flip == uninitialized_hw_default || + refcyc_per_vm_group_flip > dlg_attr->refcyc_per_vm_group_flip) REG_SET(FLIP_PARAMETERS_3, 0, REFCYC_PER_VM_GROUP_FLIP, dlg_attr->refcyc_per_vm_group_flip); - REG_GET(FLIP_PARAMETERS_4, REFCYC_PER_VM_REQ_FLIP, &cur_value); - if (cur_value > dlg_attr->refcyc_per_vm_req_flip) + REG_GET(FLIP_PARAMETERS_4, + REFCYC_PER_VM_REQ_FLIP, &refcyc_per_vm_req_flip); + + if (refcyc_per_vm_req_flip == uninitialized_hw_default || + refcyc_per_vm_req_flip > dlg_attr->refcyc_per_vm_req_flip) REG_SET(FLIP_PARAMETERS_4, 0, REFCYC_PER_VM_REQ_FLIP, dlg_attr->refcyc_per_vm_req_flip); REG_SET(FLIP_PARAMETERS_5, 0, REFCYC_PER_PTE_GROUP_FLIP_C, dlg_attr->refcyc_per_pte_group_flip_c); + REG_SET(FLIP_PARAMETERS_6, 0, REFCYC_PER_META_CHUNK_FLIP_C, dlg_attr->refcyc_per_meta_chunk_flip_c); } From 9f63ce2aa3e1575e853a6d415eec13c7677df5b1 Mon Sep 17 00:00:00 2001 From: Isabel Zhang Date: Mon, 27 Jan 2020 10:57:16 -0500 Subject: [PATCH 60/69] drm/amd/display: Add initialitions for PLL2 clock source [Why] Starting from 14nm, the PLL is built into the PHY and the PLL is mapped to PHY on 1 to 1 basis. In the code, the DP port is mapped to a PLL that was not initialized. This causes DP to HDMI dongle to not light up the display. [How] Initializations added for PLL2 when creating resources. Signed-off-by: Isabel Zhang Reviewed-by: Eric Yang Acked-by: Bhawanpreet Lakha --- drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c index 13b2f25fd7c9c..5c2c31b985600 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c @@ -59,6 +59,7 @@ #include "dcn20/dcn20_dccg.h" #include "dcn21_hubbub.h" #include "dcn10/dcn10_resource.h" +#include "dce110/dce110_resource.h" #include "dcn20/dcn20_dwb.h" #include "dcn20/dcn20_mmhubbub.h" @@ -898,6 +899,7 @@ static const struct dc_debug_options debug_defaults_diags = { enum dcn20_clk_src_array_id { DCN20_CLK_SRC_PLL0, DCN20_CLK_SRC_PLL1, + DCN20_CLK_SRC_PLL2, DCN20_CLK_SRC_TOTAL_DCN21 }; @@ -1817,6 +1819,10 @@ static bool dcn21_resource_construct( dcn21_clock_source_create(ctx, ctx->dc_bios, CLOCK_SOURCE_COMBO_PHY_PLL1, &clk_src_regs[1], false); + pool->base.clock_sources[DCN20_CLK_SRC_PLL2] = + dcn21_clock_source_create(ctx, ctx->dc_bios, + CLOCK_SOURCE_COMBO_PHY_PLL2, + &clk_src_regs[2], false); pool->base.clk_src_count = DCN20_CLK_SRC_TOTAL_DCN21; From 13fe5f48b2c7925b64055be07eaffd476cce108c Mon Sep 17 00:00:00 2001 From: Joseph Gravenor Date: Tue, 28 Jan 2020 10:38:22 -0500 Subject: [PATCH 61/69] drm/amd/display: turn off the mst hub before we do detection [why] not turning off the mst hub before detection on reboot causes us to not be able to light up displays with mst hook [how] on hw init, see if any displays are lit up. if so, turn them off Change-Id: Ie6804e7a44bf9a9fc83fc159076a2e664f99b17e Signed-off-by: Joseph Gravenor Reviewed-by: Eric Yang Acked-by: Bhawanpreet Lakha --- .../amd/display/dc/dcn10/dcn10_hw_sequencer.c | 22 +++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c index 8c7592e43ca0d..9ce15d561ab7b 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c @@ -48,8 +48,8 @@ #include "dc_link_dp.h" #include "dccg.h" #include "clk_mgr.h" - - +#include "link_hwss.h" +#include "dpcd_defs.h" #ifdef CONFIG_DRM_AMD_DC_DSC_SUPPORT #include "dsc.h" #endif @@ -1334,6 +1334,24 @@ void dcn10_init_hw(struct dc *dc) hws->funcs.dsc_pg_control(hws, res_pool->dscs[i]->inst, false); #endif + /* we want to turn off all dp displays before doing detection */ + if (dc->config.power_down_display_on_boot) { + uint8_t dpcd_power_state = '\0'; + enum dc_status status = DC_ERROR_UNEXPECTED; + + for (i = 0; i < dc->link_count; i++) { + if (dc->links[i]->connector_signal != SIGNAL_TYPE_DISPLAY_PORT) { + continue; + } + /* if any of the displays are lit up turn them off */ + status = core_link_read_dpcd(dc->links[i], DP_SET_POWER, + &dpcd_power_state, sizeof(dpcd_power_state)); + if (status == DC_OK && dpcd_power_state == DP_POWER_STATE_D0) { + dp_receiver_power_ctrl(dc->links[i], false); + } + } + } + /* If taking control over from VBIOS, we may want to optimize our first * mode set, so we need to skip powering down pipes until we know which * pipes we want to use. From faaaf68959e2d0e5b6563bc02d9a060376784873 Mon Sep 17 00:00:00 2001 From: Takashi Iwai Date: Thu, 28 Nov 2019 09:27:14 +0100 Subject: [PATCH 62/69] drm/amd/display: Fix wrongly passed static prefix MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Currently, gcc spews a warning as: drivers/gpu/drm/amd/amdgpu/../display/dc/dcn10/dcn10_hubbub.c: In function ‘hubbub1_verify_allow_pstate_change_high’: ./include/drm/drm_print.h:316:2: warning: ‘debug_data’ may be used uninitialized in this function [-Wmaybe-uninitialized] This is because the code checks against a static value although it's basically a constant and guaranteed to be set. This patch changes the type prefix from static to const for addressing the compile warning above and also for letting the compiler optimize better. Fixes: 62d591a8e00c ("drm/amd/display: create new files for hubbub functions") Reviewed-by: Harry Wentland Signed-off-by: Takashi Iwai Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c index 870735deaca71..3e851713cf8de 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c @@ -128,8 +128,8 @@ bool hubbub1_verify_allow_pstate_change_high( * pstate takes around ~100us on linux. Unknown currently as to * why it takes that long on linux */ - static unsigned int pstate_wait_timeout_us = 200; - static unsigned int pstate_wait_expected_timeout_us = 40; + const unsigned int pstate_wait_timeout_us = 200; + const unsigned int pstate_wait_expected_timeout_us = 40; static unsigned int max_sampled_pstate_wait_us; /* data collection */ static bool forced_pstate_allow; /* help with revert wa */ From 172a1ff92840a05446e4eea52dc68528609b8fb4 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Wed, 5 Feb 2020 21:38:27 +0300 Subject: [PATCH 63/69] drm/amd/display: Fix some use after free bugs These frees need to be re-ordered so that we don't dereference "hdcp_work" right after it's freed. Also in hdcp_create_workqueue() there is a problem that "hdcp_work" can be NULL if the allocation fails so it would lead to a NULL dereference in the cleanup code. Fixes: 9aeb8a134a0a ("drm/amd/display: Add sysfs interface for set/get srm") Reviewed-by: Bhawanpreet Lakha Signed-off-by: Dan Carpenter Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c index 1768a33b1dc3a..c4fd148bf6e0f 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c @@ -380,9 +380,9 @@ void hdcp_destroy(struct hdcp_workqueue *hdcp_work) cancel_delayed_work_sync(&hdcp_work[i].watchdog_timer_dwork); } - kfree(hdcp_work); kfree(hdcp_work->srm); kfree(hdcp_work->srm_temp); + kfree(hdcp_work); } static void update_config(void *handle, struct cp_psp_stream_config *config) @@ -555,11 +555,12 @@ struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev, struct { int max_caps = dc->caps.max_links; - struct hdcp_workqueue *hdcp_work = kzalloc(max_caps*sizeof(*hdcp_work), GFP_KERNEL); + struct hdcp_workqueue *hdcp_work; int i = 0; + hdcp_work = kcalloc(max_caps, sizeof(*hdcp_work), GFP_KERNEL); if (hdcp_work == NULL) - goto fail_alloc_context; + return NULL; hdcp_work->srm = kcalloc(PSP_HDCP_SRM_FIRST_GEN_MAX_SIZE, sizeof(*hdcp_work->srm), GFP_KERNEL); @@ -602,9 +603,9 @@ struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev, struct return hdcp_work; fail_alloc_context: - kfree(hdcp_work); kfree(hdcp_work->srm); kfree(hdcp_work->srm_temp); + kfree(hdcp_work); return NULL; From cecd508e07ef1ce27fcdcf3632e25d10a41aaca6 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 4 Feb 2020 09:07:19 -0500 Subject: [PATCH 64/69] drm/amdgpu: update default voltage for boot od table for navi1x It needed to be updated as well so it will show the proper values if you reset to the defaults. Bug: https://gitlab.freedesktop.org/drm/amd/issues/1020 Reviewed-by: Evan Quan Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/powerplay/navi10_ppt.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c index 2c8c4cbce5488..19a9846b730e1 100644 --- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c @@ -1977,7 +1977,7 @@ static int navi10_setup_od_limits(struct smu_context *smu) { } static int navi10_set_default_od_settings(struct smu_context *smu, bool initialize) { - OverDriveTable_t *od_table; + OverDriveTable_t *od_table, *boot_od_table; int ret = 0; ret = smu_v11_0_set_default_od_settings(smu, initialize, sizeof(OverDriveTable_t)); @@ -1985,6 +1985,7 @@ static int navi10_set_default_od_settings(struct smu_context *smu, bool initiali return ret; od_table = (OverDriveTable_t *)smu->smu_table.overdrive_table; + boot_od_table = (OverDriveTable_t *)smu->smu_table.boot_overdrive_table; if (initialize) { ret = navi10_setup_od_limits(smu); if (ret) { @@ -1998,6 +1999,8 @@ static int navi10_set_default_od_settings(struct smu_context *smu, bool initiali od_table->GfxclkFreq1); if (ret) od_table->GfxclkVolt1 = 0; + if (boot_od_table) + boot_od_table->GfxclkVolt1 = od_table->GfxclkVolt1; } if (!od_table->GfxclkVolt2) { @@ -2006,6 +2009,8 @@ static int navi10_set_default_od_settings(struct smu_context *smu, bool initiali od_table->GfxclkFreq2); if (ret) od_table->GfxclkVolt2 = 0; + if (boot_od_table) + boot_od_table->GfxclkVolt2 = od_table->GfxclkVolt2; } if (!od_table->GfxclkVolt3) { @@ -2014,6 +2019,8 @@ static int navi10_set_default_od_settings(struct smu_context *smu, bool initiali od_table->GfxclkFreq3); if (ret) od_table->GfxclkVolt3 = 0; + if (boot_od_table) + boot_od_table->GfxclkVolt3 = od_table->GfxclkVolt3; } } } From 3d8f2599c042d8766c1c4d5d85eb67ec2a4974ce Mon Sep 17 00:00:00 2001 From: Lyude Paul Date: Fri, 24 Jan 2020 14:10:46 -0500 Subject: [PATCH 65/69] drm/amd/dm/mst: Ignore payload update failures Disabling a display on MST can potentially happen after the entire MST topology has been removed, which means that we can't communicate with the topology at all in this scenario. Likewise, this also means that we can't properly update payloads on the topology and as such, it's a good idea to ignore payload update failures when disabling displays. Currently, amdgpu makes the mistake of halting the payload update process when any payload update failures occur, resulting in leaving DC's local copies of the payload tables out of date. This ends up causing problems with hotplugging MST topologies, and causes modesets on the second hotplug to fail like so: [drm] Failed to updateMST allocation table forpipe idx:1 ------------[ cut here ]------------ WARNING: CPU: 5 PID: 1511 at drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_link.c:2677 update_mst_stream_alloc_table+0x11e/0x130 [amdgpu] Modules linked in: cdc_ether usbnet fuse xt_conntrack nf_conntrack nf_defrag_ipv6 libcrc32c nf_defrag_ipv4 ipt_REJECT nf_reject_ipv4 nft_counter nft_compat nf_tables nfnetlink tun bridge stp llc sunrpc vfat fat wmi_bmof uvcvideo snd_hda_codec_realtek snd_hda_codec_generic snd_hda_codec_hdmi videobuf2_vmalloc snd_hda_intel videobuf2_memops videobuf2_v4l2 snd_intel_dspcfg videobuf2_common crct10dif_pclmul snd_hda_codec videodev crc32_pclmul snd_hwdep snd_hda_core ghash_clmulni_intel snd_seq mc joydev pcspkr snd_seq_device snd_pcm sp5100_tco k10temp i2c_piix4 snd_timer thinkpad_acpi ledtrig_audio snd wmi soundcore video i2c_scmi acpi_cpufreq ip_tables amdgpu(O) rtsx_pci_sdmmc amd_iommu_v2 gpu_sched mmc_core i2c_algo_bit ttm drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops cec drm crc32c_intel serio_raw hid_multitouch r8152 mii nvme r8169 nvme_core rtsx_pci pinctrl_amd CPU: 5 PID: 1511 Comm: gnome-shell Tainted: G O 5.5.0-rc7Lyude-Test+ #4 Hardware name: LENOVO FA495SIT26/FA495SIT26, BIOS R12ET22W(0.22 ) 01/31/2019 RIP: 0010:update_mst_stream_alloc_table+0x11e/0x130 [amdgpu] Code: 28 00 00 00 75 2b 48 8d 65 e0 5b 41 5c 41 5d 41 5e 5d c3 0f b6 06 49 89 1c 24 41 88 44 24 08 0f b6 46 01 41 88 44 24 09 eb 93 <0f> 0b e9 2f ff ff ff e8 a6 82 a3 c2 66 0f 1f 44 00 00 0f 1f 44 00 RSP: 0018:ffffac428127f5b0 EFLAGS: 00010202 RAX: 0000000000000002 RBX: ffff8d1e166eee80 RCX: 0000000000000000 RDX: ffffac428127f668 RSI: ffff8d1e166eee80 RDI: ffffac428127f610 RBP: ffffac428127f640 R08: ffffffffc03d94a8 R09: 0000000000000000 R10: ffff8d1e24b02000 R11: ffffac428127f5b0 R12: ffff8d1e1b83d000 R13: ffff8d1e1bea0b08 R14: 0000000000000002 R15: 0000000000000002 FS: 00007fab23ffcd80(0000) GS:ffff8d1e28b40000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f151f1711e8 CR3: 00000005997c0000 CR4: 00000000003406e0 Call Trace: ? mutex_lock+0xe/0x30 dc_link_allocate_mst_payload+0x9a/0x210 [amdgpu] ? dm_read_reg_func+0x39/0xb0 [amdgpu] ? core_link_enable_stream+0x656/0x730 [amdgpu] core_link_enable_stream+0x656/0x730 [amdgpu] dce110_apply_ctx_to_hw+0x58e/0x5d0 [amdgpu] ? dcn10_verify_allow_pstate_change_high+0x1d/0x280 [amdgpu] ? dcn10_wait_for_mpcc_disconnect+0x3c/0x130 [amdgpu] dc_commit_state+0x292/0x770 [amdgpu] ? add_timer+0x101/0x1f0 ? ttm_bo_put+0x1a1/0x2f0 [ttm] amdgpu_dm_atomic_commit_tail+0xb59/0x1ff0 [amdgpu] ? amdgpu_move_blit.constprop.0+0xb8/0x1f0 [amdgpu] ? amdgpu_bo_move+0x16d/0x2b0 [amdgpu] ? ttm_bo_handle_move_mem+0x118/0x570 [ttm] ? ttm_bo_validate+0x134/0x150 [ttm] ? dm_plane_helper_prepare_fb+0x1b9/0x2a0 [amdgpu] ? _cond_resched+0x15/0x30 ? wait_for_completion_timeout+0x38/0x160 ? _cond_resched+0x15/0x30 ? wait_for_completion_interruptible+0x33/0x190 commit_tail+0x94/0x130 [drm_kms_helper] drm_atomic_helper_commit+0x113/0x140 [drm_kms_helper] drm_atomic_helper_set_config+0x70/0xb0 [drm_kms_helper] drm_mode_setcrtc+0x194/0x6a0 [drm] ? _cond_resched+0x15/0x30 ? mutex_lock+0xe/0x30 ? drm_mode_getcrtc+0x180/0x180 [drm] drm_ioctl_kernel+0xaa/0xf0 [drm] drm_ioctl+0x208/0x390 [drm] ? drm_mode_getcrtc+0x180/0x180 [drm] amdgpu_drm_ioctl+0x49/0x80 [amdgpu] do_vfs_ioctl+0x458/0x6d0 ksys_ioctl+0x5e/0x90 __x64_sys_ioctl+0x16/0x20 do_syscall_64+0x55/0x1b0 entry_SYSCALL_64_after_hwframe+0x44/0xa9 RIP: 0033:0x7fab2121f87b Code: 0f 1e fa 48 8b 05 0d 96 2c 00 64 c7 00 26 00 00 00 48 c7 c0 ff ff ff ff c3 66 0f 1f 44 00 00 f3 0f 1e fa b8 10 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d dd 95 2c 00 f7 d8 64 89 01 48 RSP: 002b:00007ffd045f9068 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 RAX: ffffffffffffffda RBX: 00007ffd045f90a0 RCX: 00007fab2121f87b RDX: 00007ffd045f90a0 RSI: 00000000c06864a2 RDI: 000000000000000b RBP: 00007ffd045f90a0 R08: 0000000000000000 R09: 000055dbd2985d10 R10: 000055dbd2196280 R11: 0000000000000246 R12: 00000000c06864a2 R13: 000000000000000b R14: 0000000000000000 R15: 000055dbd2196280 ---[ end trace 6ea888c24d2059cd ]--- Note as well, I have only been able to reproduce this on setups with 2 MST displays. Changes since v1: * Don't return false when part 1 or part 2 of updating the payloads fails, we don't want to abort at any step of the process even if things fail Reviewed-by: Mikita Lipski Signed-off-by: Lyude Paul Acked-by: Harry Wentland Cc: stable@vger.kernel.org Signed-off-by: Alex Deucher --- .../drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c | 13 ++++--------- 1 file changed, 4 insertions(+), 9 deletions(-) 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 9f87c54b362fd..a5bfbb1b1fc5f 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 @@ -278,7 +278,8 @@ bool dm_helpers_dp_mst_write_payload_allocation_table( drm_dp_mst_reset_vcpi_slots(mst_mgr, mst_port); } - ret = drm_dp_update_payload_part1(mst_mgr); + /* It's OK for this to fail */ + drm_dp_update_payload_part1(mst_mgr); /* mst_mgr->->payloads are VC payload notify MST branch using DPCD or * AUX message. The sequence is slot 1-63 allocated sequence for each @@ -287,9 +288,6 @@ bool dm_helpers_dp_mst_write_payload_allocation_table( get_payload_table(aconnector, proposed_table); - if (ret) - return false; - return true; } @@ -347,7 +345,6 @@ bool dm_helpers_dp_mst_send_payload_allocation( struct amdgpu_dm_connector *aconnector; struct drm_dp_mst_topology_mgr *mst_mgr; struct drm_dp_mst_port *mst_port; - int ret; aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context; @@ -361,10 +358,8 @@ bool dm_helpers_dp_mst_send_payload_allocation( if (!mst_mgr->mst_state) return false; - ret = drm_dp_update_payload_part2(mst_mgr); - - if (ret) - return false; + /* It's OK for this to fail */ + drm_dp_update_payload_part2(mst_mgr); if (!enable) drm_dp_mst_deallocate_vcpi(mst_mgr, mst_port); From fc12621b275a75dffbb4a933d23767ebe381b0f3 Mon Sep 17 00:00:00 2001 From: Jack Zhang Date: Wed, 5 Feb 2020 16:34:27 +0800 Subject: [PATCH 66/69] drm/amdgpu/sriov Don't send msg when smu suspend For sriov and pp_onevf_mode, do not send message to set smu status, because smu doesn't support these messages under VF. Besides, it should skip smu_suspend when pp_onevf_mode is disabled. Signed-off-by: Jack Zhang Acked-by: Alex Deucher Acked-by: Evan Quan --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 15 ++++++++------- drivers/gpu/drm/amd/powerplay/amdgpu_smu.c | 21 +++++++++++++-------- 2 files changed, 21 insertions(+), 15 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 4ff7ce3f8a907..2d1f8d4338a71 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -2353,15 +2353,16 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) } adev->ip_blocks[i].status.hw = false; /* handle putting the SMC in the appropriate state */ - if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) { - r = amdgpu_dpm_set_mp1_state(adev, adev->mp1_state); - if (r) { - DRM_ERROR("SMC failed to set mp1 state %d, %d\n", - adev->mp1_state, r); - return r; + if(!amdgpu_sriov_vf(adev)){ + if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) { + r = amdgpu_dpm_set_mp1_state(adev, adev->mp1_state); + if (r) { + DRM_ERROR("SMC failed to set mp1 state %d, %d\n", + adev->mp1_state, r); + return r; + } } } - adev->ip_blocks[i].status.hw = false; } diff --git a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c index 99ad4ddbe12f0..a6d7b5f3aa823 100644 --- a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c @@ -1461,21 +1461,26 @@ static int smu_suspend(void *handle) struct smu_context *smu = &adev->smu; bool baco_feature_is_enabled = false; + if (amdgpu_sriov_vf(adev)&& !amdgpu_sriov_is_pp_one_vf(adev)) + return 0; + if (!smu->pm_enabled) return 0; if(!smu->is_apu) baco_feature_is_enabled = smu_feature_is_enabled(smu, SMU_FEATURE_BACO_BIT); - ret = smu_system_features_control(smu, false); - if (ret) - return ret; - - if (baco_feature_is_enabled) { - ret = smu_feature_set_enabled(smu, SMU_FEATURE_BACO_BIT, true); - if (ret) { - pr_warn("set BACO feature enabled failed, return %d\n", ret); + if(!amdgpu_sriov_vf(adev)) { + ret = smu_system_features_control(smu, false); + if (ret) return ret; + + if (baco_feature_is_enabled) { + ret = smu_feature_set_enabled(smu, SMU_FEATURE_BACO_BIT, true); + if (ret) { + pr_warn("set BACO feature enabled failed, return %d\n", ret); + return ret; + } } } From 19a31a705d3b58bb5ffdd3ff98816638134765a8 Mon Sep 17 00:00:00 2001 From: James Zhu Date: Wed, 5 Feb 2020 09:20:22 -0500 Subject: [PATCH 67/69] drm/amdgpu/vcn2.5: fix DPG mode power off issue on instance 1 Support pause_state for multiple instance, and it will fix vcn2.5 DPG mode power off issue on instance 1. Signed-off-by: James Zhu Reviewed-by: Leo Liu --- drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h | 3 +-- drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c | 14 ++++++++------ drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c | 6 +++--- drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c | 6 +++--- 4 files changed, 15 insertions(+), 14 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h index d6deb0eb1e15a..6fe057329de2b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h @@ -179,6 +179,7 @@ struct amdgpu_vcn_inst { struct amdgpu_irq_src irq; struct amdgpu_vcn_reg external; struct amdgpu_bo *dpg_sram_bo; + struct dpg_pause_state pause_state; void *dpg_sram_cpu_addr; uint64_t dpg_sram_gpu_addr; uint32_t *dpg_sram_curr_addr; @@ -190,8 +191,6 @@ struct amdgpu_vcn { const struct firmware *fw; /* VCN firmware */ unsigned num_enc_rings; enum amd_powergating_state cur_state; - struct dpg_pause_state pause_state; - bool indirect_sram; uint8_t num_vcn_inst; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c index 1a24fadd30e2d..71f61afdc6551 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c @@ -1207,9 +1207,10 @@ static int vcn_v1_0_pause_dpg_mode(struct amdgpu_device *adev, struct amdgpu_ring *ring; /* pause/unpause if state is changed */ - if (adev->vcn.pause_state.fw_based != new_state->fw_based) { + if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) { DRM_DEBUG("dpg pause state changed %d:%d -> %d:%d", - adev->vcn.pause_state.fw_based, adev->vcn.pause_state.jpeg, + adev->vcn.inst[inst_idx].pause_state.fw_based, + adev->vcn.inst[inst_idx].pause_state.jpeg, new_state->fw_based, new_state->jpeg); reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) & @@ -1258,13 +1259,14 @@ static int vcn_v1_0_pause_dpg_mode(struct amdgpu_device *adev, reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK; WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data); } - adev->vcn.pause_state.fw_based = new_state->fw_based; + adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based; } /* pause/unpause if state is changed */ - if (adev->vcn.pause_state.jpeg != new_state->jpeg) { + if (adev->vcn.inst[inst_idx].pause_state.jpeg != new_state->jpeg) { DRM_DEBUG("dpg pause state changed %d:%d -> %d:%d", - adev->vcn.pause_state.fw_based, adev->vcn.pause_state.jpeg, + adev->vcn.inst[inst_idx].pause_state.fw_based, + adev->vcn.inst[inst_idx].pause_state.jpeg, new_state->fw_based, new_state->jpeg); reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) & @@ -1318,7 +1320,7 @@ static int vcn_v1_0_pause_dpg_mode(struct amdgpu_device *adev, reg_data &= ~UVD_DPG_PAUSE__JPEG_PAUSE_DPG_REQ_MASK; WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data); } - adev->vcn.pause_state.jpeg = new_state->jpeg; + adev->vcn.inst[inst_idx].pause_state.jpeg = new_state->jpeg; } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c index 4f7216788f113..c387c81f86958 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c @@ -1137,9 +1137,9 @@ static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev, int ret_code; /* pause/unpause if state is changed */ - if (adev->vcn.pause_state.fw_based != new_state->fw_based) { + if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) { DRM_DEBUG("dpg pause state changed %d -> %d", - adev->vcn.pause_state.fw_based, new_state->fw_based); + adev->vcn.inst[inst_idx].pause_state.fw_based, new_state->fw_based); reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) & (~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK); @@ -1185,7 +1185,7 @@ static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev, reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK; WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data); } - adev->vcn.pause_state.fw_based = new_state->fw_based; + adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based; } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index 70fae7977f8f4..97ab44c2f250a 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -1367,9 +1367,9 @@ static int vcn_v2_5_pause_dpg_mode(struct amdgpu_device *adev, int ret_code; /* pause/unpause if state is changed */ - if (adev->vcn.pause_state.fw_based != new_state->fw_based) { + if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) { DRM_DEBUG("dpg pause state changed %d -> %d", - adev->vcn.pause_state.fw_based, new_state->fw_based); + adev->vcn.inst[inst_idx].pause_state.fw_based, new_state->fw_based); reg_data = RREG32_SOC15(UVD, inst_idx, mmUVD_DPG_PAUSE) & (~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK); @@ -1414,7 +1414,7 @@ static int vcn_v2_5_pause_dpg_mode(struct amdgpu_device *adev, reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK; WREG32_SOC15(UVD, inst_idx, mmUVD_DPG_PAUSE, reg_data); } - adev->vcn.pause_state.fw_based = new_state->fw_based; + adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based; } return 0; From 12c27639dc831648dcf5fe4bc690d32faef08549 Mon Sep 17 00:00:00 2001 From: Daniel Kolesa Date: Thu, 6 Feb 2020 20:14:35 +0100 Subject: [PATCH 68/69] amdgpu: Prevent build errors regarding soft/hard-float FP ABI tags On PowerPC, the compiler will tag object files with whether they use hard or soft float FP ABI and whether they use 64 or 128-bit long double ABI. On systems with 64-bit long double ABI, a tag will get emitted whenever a double is used, as on those systems a long double is the same as a double. This will prevent linkage as other files are being compiled with hard-float. On ppc64, this code will never actually get used for the time being, as the only currently existing hardware using it are the Renoir APUs. Therefore, until this is testable and can be fixed properly, at least make sure the build will not fail. Signed-off-by: Daniel Kolesa Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/clk_mgr/Makefile | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/Makefile b/drivers/gpu/drm/amd/display/dc/clk_mgr/Makefile index b864869cc7e3e..6fa7422c51da5 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/Makefile +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/Makefile @@ -91,6 +91,12 @@ ifdef CONFIG_DRM_AMD_DC_DCN2_1 ############################################################################### CLK_MGR_DCN21 = rn_clk_mgr.o rn_clk_mgr_vbios_smu.o +# prevent build errors regarding soft-float vs hard-float FP ABI tags +# this code is currently unused on ppc64, as it applies to Renoir APUs only +ifdef CONFIG_PPC64 +CFLAGS_$(AMDDALPATH)/dc/clk_mgr/dcn21/rn_clk_mgr.o := $(call cc-option,-mno-gnu-attribute) +endif + AMD_DAL_CLK_MGR_DCN21 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn21/,$(CLK_MGR_DCN21)) AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN21) From 8c913492ffea6a18370385e355f10a73bc014c96 Mon Sep 17 00:00:00 2001 From: Jonathan Kim Date: Thu, 6 Feb 2020 11:57:11 -0500 Subject: [PATCH 69/69] drm/amdgpu: fix amdgpu pmu to use hwc->config instead of hwc->conf hwc->conf was designated specifically for AMD APU IOMMU purposes. This could cause problems in performance and/or function since APU IOMMU implementation is elsewhere. Also hwc->conf and hwc->config are different members of an anonymous union so hwc->conf aliases as hw->last_tag. Change-Id: I5d0c8396bc5a11dd69262a6a5423633d9ca6df3d Signed-off-by: Jonathan Kim Reviewed-by: Felix Kuehling --- drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c | 27 +------------------------ 1 file changed, 1 insertion(+), 26 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c index 0536d3b6979ac..63e2bbf1afbcd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c @@ -52,11 +52,7 @@ static int amdgpu_perf_event_init(struct perf_event *event) return -ENOENT; /* update the hw_perf_event struct with config data */ -#if defined(HAVE_HW_PERF_EVENT_CONF_MEMBER) - hwc->conf = event->attr.config; -#else hwc->config = event->attr.config; -#endif return 0; } @@ -78,13 +74,9 @@ static void amdgpu_perf_start(struct perf_event *event, int flags) switch (pe->pmu_perf_type) { case PERF_TYPE_AMDGPU_DF: if (!(flags & PERF_EF_RELOAD)) -#if defined(HAVE_HW_PERF_EVENT_CONF_MEMBER) - pe->adev->df.funcs->pmc_start(pe->adev, hwc->conf, 1); - pe->adev->df.funcs->pmc_start(pe->adev, hwc->conf, 0); -#else pe->adev->df.funcs->pmc_start(pe->adev, hwc->config, 1); + pe->adev->df.funcs->pmc_start(pe->adev, hwc->config, 0); -#endif break; default: break; @@ -109,13 +101,8 @@ static void amdgpu_perf_read(struct perf_event *event) switch (pe->pmu_perf_type) { case PERF_TYPE_AMDGPU_DF: -#if defined(HAVE_HW_PERF_EVENT_CONF_MEMBER) - pe->adev->df.funcs->pmc_get_count(pe->adev, hwc->conf, - &count); -#else pe->adev->df.funcs->pmc_get_count(pe->adev, hwc->config, &count); -#endif break; default: count = 0; @@ -139,11 +126,7 @@ static void amdgpu_perf_stop(struct perf_event *event, int flags) switch (pe->pmu_perf_type) { case PERF_TYPE_AMDGPU_DF: -#if defined(HAVE_HW_PERF_EVENT_CONF_MEMBER) - pe->adev->df.funcs->pmc_stop(pe->adev, hwc->conf, 0); -#else pe->adev->df.funcs->pmc_stop(pe->adev, hwc->config, 0); -#endif break; default: break; @@ -173,11 +156,7 @@ static int amdgpu_perf_add(struct perf_event *event, int flags) switch (pe->pmu_perf_type) { case PERF_TYPE_AMDGPU_DF: -#if defined(HAVE_HW_PERF_EVENT_CONF_MEMBER) - retval = pe->adev->df.funcs->pmc_start(pe->adev, hwc->conf, 1); -#else retval = pe->adev->df.funcs->pmc_start(pe->adev, hwc->config, 1); -#endif break; default: return 0; @@ -205,11 +184,7 @@ static void amdgpu_perf_del(struct perf_event *event, int flags) switch (pe->pmu_perf_type) { case PERF_TYPE_AMDGPU_DF: -#if defined(HAVE_HW_PERF_EVENT_CONF_MEMBER) - pe->adev->df.funcs->pmc_stop(pe->adev, hwc->conf, 1); -#else pe->adev->df.funcs->pmc_stop(pe->adev, hwc->config, 1); -#endif break; default: break;