From 450d007d199e632a1a4c4b91302deacd7d56815f Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Thu, 14 Feb 2019 23:46:19 +0100 Subject: gpu: drm: radeon: Set DPM_FLAG_NEVER_SKIP when enabling PM-runtime MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit On HP ProBook 4540s, if PM-runtime is enabled in the radeon driver and the direct-complete optimization is used for the radeon device during system-wide suspend, the system doesn't resume. Preventing direct-complete from being used with the radeon device by setting the DPM_FLAG_NEVER_SKIP driver flag for it makes the problem go away, which indicates that direct-complete is not safe for the radeon driver in general and should not be used with it (at least for now). This fixes a regression introduced by commit c62ec4610c40 ("PM / core: Fix direct_complete handling for devices with no callbacks") which allowed direct-complete to be applied to devices without PM callbacks (again) which in turn unlocked direct-complete for radeon on HP ProBook 4540s. Fixes: c62ec4610c40 ("PM / core: Fix direct_complete handling for devices with no callbacks") Link: https://bugzilla.kernel.org/show_bug.cgi?id=201519 Reported-by: Ярослав Семченко Tested-by: Ярослав Семченко Signed-off-by: Rafael J. Wysocki Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/radeon/radeon_kms.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index dec1e081f529..6a8fb6fd183c 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -172,6 +172,7 @@ int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags) } if (radeon_is_px(dev)) { + dev_pm_set_driver_flags(dev->dev, DPM_FLAG_NEVER_SKIP); pm_runtime_use_autosuspend(dev->dev); pm_runtime_set_autosuspend_delay(dev->dev, 5000); pm_runtime_set_active(dev->dev); -- cgit v1.2.3 From d33158530660bc89be3cc870a2152e4e9a76cac7 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Mon, 18 Feb 2019 17:11:38 -0500 Subject: drm/amdgpu: Set DPM_FLAG_NEVER_SKIP when enabling PM-runtime Based on a similar patch from Rafael for radeon. When using ATPX to control dGPU power, the state is not retained across suspend and resume cycles by default. This can probably be loosened for Hybrid Graphics (_PR3) laptops where I think the state is properly retained. Fixes: c62ec4610c40 ("PM / core: Fix direct_complete handling for devices with no callbacks") Cc: Rafael J. Wysocki Acked-by: Rafael J. Wysocki Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index bc62bf41b7e9..5dc349173e4f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -212,6 +212,7 @@ int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags) } if (amdgpu_device_is_px(dev)) { + dev_pm_set_driver_flags(dev->dev, DPM_FLAG_NEVER_SKIP); pm_runtime_use_autosuspend(dev->dev); pm_runtime_set_autosuspend_delay(dev->dev, 5000); pm_runtime_set_active(dev->dev); -- cgit v1.2.3 From 9db97d8aa8f8a518c421196a504dbfc942ef8d40 Mon Sep 17 00:00:00 2001 From: shaoyunl Date: Fri, 15 Feb 2019 11:05:04 -0500 Subject: drm/amdgpu: Update sdma golden setting for vega20 According to hardware engineer, WRITE_BURST_LENGTH [9:8] in register SDMA0_CHICKEN_BITS need to change to 3 for better performance Signed-off-by: shaoyunl Acked-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index 6811a5d05b27..aa2f71cc1eba 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -128,7 +128,7 @@ static const struct soc15_reg_golden golden_settings_sdma0_4_2_init[] = { static const struct soc15_reg_golden golden_settings_sdma0_4_2[] = { - SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831d07), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831f07), SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xffffffff, 0x3f000100), SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0000773f, 0x00004002), SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002), @@ -158,7 +158,7 @@ static const struct soc15_reg_golden golden_settings_sdma0_4_2[] = }; static const struct soc15_reg_golden golden_settings_sdma1_4_2[] = { - SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831d07), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831f07), SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CLK_CTRL, 0xffffffff, 0x3f000100), SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0000773f, 0x00004002), SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002), -- cgit v1.2.3 From d2f0b53bda3193874f3905bc839888f895d1c0cf Mon Sep 17 00:00:00 2001 From: "Leo (Hanghong) Ma" Date: Thu, 24 Jan 2019 15:07:52 -0500 Subject: drm/amd/display: Fix MST reboot/poweroff sequence [Why] drm_dp_mst_topology_mgr_suspend() is added into the new reboot sequence, which disables the UP request at the beginning. Therefore sideband messages are blocked. [How] Finish MST sideband message transaction before UP request is suppressed. Signed-off-by: Leo (Hanghong) Ma Reviewed-by: Roman Li Acked-by: Leo Li Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'drivers/gpu') 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 0b392bfca284..5296b8f3e0ab 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -786,12 +786,13 @@ static int dm_suspend(void *handle) struct amdgpu_display_manager *dm = &adev->dm; int ret = 0; + WARN_ON(adev->dm.cached_state); + adev->dm.cached_state = drm_atomic_helper_suspend(adev->ddev); + s3_handle_mst(adev->ddev, true); amdgpu_dm_irq_suspend(adev); - WARN_ON(adev->dm.cached_state); - adev->dm.cached_state = drm_atomic_helper_suspend(adev->ddev); dc_set_power_state(dm->dc, DC_ACPI_CM_POWER_STATE_D3); -- cgit v1.2.3 From 8852ae9a82498207c15262b6294d14aea1796966 Mon Sep 17 00:00:00 2001 From: Roman Li Date: Mon, 28 Jan 2019 10:59:34 -0500 Subject: drm/amd/display: Raise dispclk value for dce11 [Why] The visual corruption due to low display clock value. Observed on Carrizo 4K@60Hz. [How] There was earlier patch for dce_update_clocks: Adding +15% workaround also to to dce11_update_clocks Signed-off-by: Roman Li Reviewed-by: Nicholas Kazlauskas Acked-by: Leo Li Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c index 19801bdba0d2..7a72ee46f14b 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c @@ -662,6 +662,11 @@ static void dce11_update_clocks(struct clk_mgr *clk_mgr, { struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); struct dm_pp_power_level_change_request level_change_req; + int patched_disp_clk = context->bw.dce.dispclk_khz; + + /*TODO: W/A for dal3 linux, investigate why this works */ + if (!clk_mgr_dce->dfs_bypass_active) + patched_disp_clk = patched_disp_clk * 115 / 100; level_change_req.power_level = dce_get_required_clocks_state(clk_mgr, context); /* get max clock state from PPLIB */ @@ -671,9 +676,9 @@ static void dce11_update_clocks(struct clk_mgr *clk_mgr, clk_mgr_dce->cur_min_clks_state = level_change_req.power_level; } - if (should_set_clock(safe_to_lower, context->bw.dce.dispclk_khz, clk_mgr->clks.dispclk_khz)) { - context->bw.dce.dispclk_khz = dce_set_clock(clk_mgr, context->bw.dce.dispclk_khz); - clk_mgr->clks.dispclk_khz = context->bw.dce.dispclk_khz; + if (should_set_clock(safe_to_lower, patched_disp_clk, clk_mgr->clks.dispclk_khz)) { + context->bw.dce.dispclk_khz = dce_set_clock(clk_mgr, patched_disp_clk); + clk_mgr->clks.dispclk_khz = patched_disp_clk; } dce11_pplib_apply_display_requirements(clk_mgr->ctx->dc, context); } -- cgit v1.2.3 From d179b88deb3bf6fed4991a31fd6f0f2cad21fab5 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Fri, 15 Feb 2019 12:30:19 +0000 Subject: drm/i915/fbdev: Actually configure untiled displays If we skipped all the connectors that were not part of a tile, we would leave conn_seq=0 and conn_configured=0, convincing ourselves that we had stagnated in our configuration attempts. Avoid this situation by starting conn_seq=ALL_CONNECTORS, and repeating until we find no more connectors to configure. Fixes: 754a76591b12 ("drm/i915/fbdev: Stop repeating tile configuration on stagnation") Reported-by: Maarten Lankhorst Signed-off-by: Chris Wilson Cc: Maarten Lankhorst Reviewed-by: Maarten Lankhorst Link: https://patchwork.freedesktop.org/patch/msgid/20190215123019.32283-1-chris@chris-wilson.co.uk Cc: # v3.19+ (cherry picked from commit d9b308b1f8a1acc0c3279f443d4fe0f9f663252e) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/intel_fbdev.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/i915/intel_fbdev.c b/drivers/gpu/drm/i915/intel_fbdev.c index 7f365ac0b549..4ee16b264dbe 100644 --- a/drivers/gpu/drm/i915/intel_fbdev.c +++ b/drivers/gpu/drm/i915/intel_fbdev.c @@ -336,8 +336,8 @@ static bool intel_fb_initial_config(struct drm_fb_helper *fb_helper, bool *enabled, int width, int height) { struct drm_i915_private *dev_priv = to_i915(fb_helper->dev); - unsigned long conn_configured, conn_seq, mask; unsigned int count = min(fb_helper->connector_count, BITS_PER_LONG); + unsigned long conn_configured, conn_seq; int i, j; bool *save_enabled; bool fallback = true, ret = true; @@ -355,10 +355,9 @@ static bool intel_fb_initial_config(struct drm_fb_helper *fb_helper, drm_modeset_backoff(&ctx); memcpy(save_enabled, enabled, count); - mask = GENMASK(count - 1, 0); + conn_seq = GENMASK(count - 1, 0); conn_configured = 0; retry: - conn_seq = conn_configured; for (i = 0; i < count; i++) { struct drm_fb_helper_connector *fb_conn; struct drm_connector *connector; @@ -371,7 +370,8 @@ retry: if (conn_configured & BIT(i)) continue; - if (conn_seq == 0 && !connector->has_tile) + /* First pass, only consider tiled connectors */ + if (conn_seq == GENMASK(count - 1, 0) && !connector->has_tile) continue; if (connector->status == connector_status_connected) @@ -475,8 +475,10 @@ retry: conn_configured |= BIT(i); } - if ((conn_configured & mask) != mask && conn_configured != conn_seq) + if (conn_configured != conn_seq) { /* repeat until no more are found */ + conn_seq = conn_configured; goto retry; + } /* * If the BIOS didn't enable everything it could, fall back to have the -- cgit v1.2.3 From 0921c41e19028314830b33daa681e46b46477c5e Mon Sep 17 00:00:00 2001 From: Nicholas Kazlauskas Date: Fri, 1 Feb 2019 09:36:59 -0500 Subject: drm/amd/display: Fix negative cursor pos programming [Why] If the cursor pos passed from DM is less than the plane_state->dst_rect top left corner then the unsigned cursor pos wraps around to a large positive number since cursor pos is a u32. There was an attempt to guard against this in hubp1_cursor_set_position by checking the src_x_offset and src_y_offset and offseting the cursor hotspot within hubp1_cursor_set_position. However, the cursor position itself is still being programmed incorrectly as a large value. This manifests itself visually as the cursor disappearing or containing strange artifacts near the middle of the screen on raven. [How] Don't subtract the destination rect top left corner from the pos but add it to the hotspot instead. This happens before the pos gets passed into hubp1_cursor_set_position. This achieves the same result but avoids the subtraction wrap around. With this fix the original cursor programming logic can be used again. Signed-off-by: Nicholas Kazlauskas Reviewed-by: Charlene Liu Acked-by: Leo Li Acked-by: Murton Liu Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/gpu') 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 58a12ddf12f3..41883c981789 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 @@ -2658,8 +2658,8 @@ static void dcn10_set_cursor_position(struct pipe_ctx *pipe_ctx) .mirror = pipe_ctx->plane_state->horizontal_mirror }; - pos_cpy.x -= pipe_ctx->plane_state->dst_rect.x; - pos_cpy.y -= pipe_ctx->plane_state->dst_rect.y; + pos_cpy.x_hotspot += pipe_ctx->plane_state->dst_rect.x; + pos_cpy.y_hotspot += pipe_ctx->plane_state->dst_rect.y; if (pipe_ctx->plane_state->address.type == PLN_ADDR_TYPE_VIDEO_PROGRESSIVE) -- cgit v1.2.3 From 9f7ddbea2bb826a2147309f735726a8b09950944 Mon Sep 17 00:00:00 2001 From: Bhawanpreet Lakha Date: Tue, 5 Feb 2019 13:55:20 -0500 Subject: drm/amd/display: fix optimize_bandwidth func pointer for dce80 [Why] optimize_bandwidth was using dce100_prepare_bandwidth this is incorrect [How] change it to dce100_optimize_bandwidth Signed-off-by: Bhawanpreet Lakha Reviewed-by: Charlene Liu Acked-by: Leo Li Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h | 4 ++++ drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c | 2 +- 2 files changed, 5 insertions(+), 1 deletion(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h index acd418515346..a6b80fdaa666 100644 --- a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h @@ -37,6 +37,10 @@ void dce100_prepare_bandwidth( struct dc *dc, struct dc_state *context); +void dce100_optimize_bandwidth( + struct dc *dc, + struct dc_state *context); + bool dce100_enable_display_power_gating(struct dc *dc, uint8_t controller_id, struct dc_bios *dcb, enum pipe_gating_control power_gating); diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c index a60a90e68d91..c4543178ba20 100644 --- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c @@ -77,6 +77,6 @@ void dce80_hw_sequencer_construct(struct dc *dc) dc->hwss.enable_display_power_gating = dce100_enable_display_power_gating; dc->hwss.pipe_control_lock = dce_pipe_control_lock; dc->hwss.prepare_bandwidth = dce100_prepare_bandwidth; - dc->hwss.optimize_bandwidth = dce100_prepare_bandwidth; + dc->hwss.optimize_bandwidth = dce100_optimize_bandwidth; } -- cgit v1.2.3 From 4ece61a22be5ab5d49cc5fc20a19a0afa24a019d Mon Sep 17 00:00:00 2001 From: Bhawanpreet Lakha Date: Tue, 5 Feb 2019 14:03:52 -0500 Subject: drm/amd/display: set clocks to 0 on suspend on dce80 [Why] When a dce80 asic was suspended, the clocks were not set to 0. Upon resume, the new clock was compared to the existing clock, they were found to be the same, and so the clock was not set. This resulted in a blackscreen. [How] In atomic commit, check to see if there are any active pipes. If no, set clocks to 0 Signed-off-by: Bhawanpreet Lakha Reviewed-by: Nicholas Kazlauskas Acked-by: Leo Li Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c | 19 ++++++++++++++++--- 1 file changed, 16 insertions(+), 3 deletions(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c index cdd1d6b7b9f2..4e9ea50141bd 100644 --- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c @@ -790,9 +790,22 @@ bool dce80_validate_bandwidth( struct dc *dc, struct dc_state *context) { - /* TODO implement when needed but for now hardcode max value*/ - context->bw.dce.dispclk_khz = 681000; - context->bw.dce.yclk_khz = 250000 * MEMORY_TYPE_MULTIPLIER_CZ; + int i; + bool at_least_one_pipe = false; + + for (i = 0; i < dc->res_pool->pipe_count; i++) { + if (context->res_ctx.pipe_ctx[i].stream) + at_least_one_pipe = true; + } + + if (at_least_one_pipe) { + /* TODO implement when needed but for now hardcode max value*/ + context->bw.dce.dispclk_khz = 681000; + context->bw.dce.yclk_khz = 250000 * MEMORY_TYPE_MULTIPLIER_CZ; + } else { + context->bw.dce.dispclk_khz = 0; + context->bw.dce.yclk_khz = 0; + } return true; } -- cgit v1.2.3 From a213c2c7e235cfc0e0a161a558f7fdf2fb3a624a Mon Sep 17 00:00:00 2001 From: Christian König Date: Wed, 20 Feb 2019 15:16:06 +0100 Subject: drm/amdgpu: disable bulk moves for now MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The changes to fix those are two invasive for backporting. Just disable the feature in 4.20 and 5.0. Acked-by: Alex Deucher Signed-off-by: Christian König Cc: [4.20+] Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 7c108e687683..698bcb8ce61d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -638,12 +638,14 @@ void amdgpu_vm_move_to_lru_tail(struct amdgpu_device *adev, struct ttm_bo_global *glob = adev->mman.bdev.glob; struct amdgpu_vm_bo_base *bo_base; +#if 0 if (vm->bulk_moveable) { spin_lock(&glob->lru_lock); ttm_bo_bulk_move_lru_tail(&vm->lru_bulk_move); spin_unlock(&glob->lru_lock); return; } +#endif memset(&vm->lru_bulk_move, 0, sizeof(vm->lru_bulk_move)); -- cgit v1.2.3 From d63716658ac16c515d1223a9fbf5edbf76b1b333 Mon Sep 17 00:00:00 2001 From: Mario Kleiner Date: Wed, 30 Jan 2019 06:04:46 +0100 Subject: drm/amd/display: Use vrr friendly pageflip throttling in DC. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit In VRR mode, keep track of the vblank count of the last completed pageflip in amdgpu_crtc->last_flip_vblank, as recorded in the pageflip completion handler after each completed flip. Use that count to prevent mmio programming a new pageflip within the same vblank in which the last pageflip completed, iow. to throttle pageflips to at most one flip per video frame, while at the same time allowing to request a flip not only before start of vblank, but also anywhere within vblank. The old logic did the same, and made sense for regular fixed refresh rate flipping, but in vrr mode it prevents requesting a flip anywhere inside the possibly huge vblank, thereby reducing framerate in vrr mode instead of improving it, by delaying a slightly delayed flip requests up to a maximum vblank duration + 1 scanout duration. This would limit VRR usefulness to only help applications with a very high GPU demand, which can submit the flip request before start of vblank, but then have to wait long for fences to complete. With this method a flip can be both requested and - after fences have completed - executed, ie. it doesn't matter if the request (amdgpu_dm_do_flip()) gets delayed until deep into the extended vblank due to cpu execution delays. This also allows clients which want to regulate framerate within the vrr range a much more fine-grained control of flip timing, a feature that might be useful for video playback, and is very useful for neuroscience/vision research applications. In regular non-VRR mode, retain the old flip submission behavior. This to keep flip scheduling for fullscreen X11/GLX OpenGL clients intact, if they use the GLX_OML_sync_control extensions glXSwapBufferMscOML(, ..., target_msc,...) function with a specific target_msc target vblank count. glXSwapBuffersMscOML() or DRI3/Present PresentPixmap() will not flip at the proper target_msc for a non-zero target_msc if VRR mode is active with this patch. They'd often flip one frame too early. However, this limitation should not matter much in VRR mode, as scheduling based on vblank counts is pretty futile/unusable under variable refresh duration anyway, so no real extra harm is done. Signed-off-by: Mario Kleiner Cc: Nicholas Kazlauskas Cc: Harry Wentland Cc: Alex Deucher Cc: Michel Dänzer Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h | 1 + drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 27 +++++++++++++++++++---- 2 files changed, 24 insertions(+), 4 deletions(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index aadd0fa42e43..3aa42c64484a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -405,6 +405,7 @@ struct amdgpu_crtc { struct amdgpu_flip_work *pflip_works; enum amdgpu_flip_status pflip_status; int deferred_flip_completion; + u64 last_flip_vblank; /* pll sharing */ struct amdgpu_atom_ss ss; bool ss_enabled; 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 5296b8f3e0ab..636d14a60952 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -303,12 +303,11 @@ static void dm_pflip_high_irq(void *interrupt_params) return; } + /* Update to correct count(s) if racing with vblank irq */ + amdgpu_crtc->last_flip_vblank = drm_crtc_accurate_vblank_count(&amdgpu_crtc->base); /* wake up userspace */ if (amdgpu_crtc->event) { - /* Update to correct count(s) if racing with vblank irq */ - drm_crtc_accurate_vblank_count(&amdgpu_crtc->base); - drm_crtc_send_vblank_event(&amdgpu_crtc->base, amdgpu_crtc->event); /* page flip completed. clean up */ @@ -4828,6 +4827,8 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state, to_dm_crtc_state(drm_atomic_get_old_crtc_state(state, pcrtc)); int planes_count = 0; unsigned long flags; + u64 last_flip_vblank; + bool vrr_active = acrtc_state->freesync_config.state == VRR_STATE_ACTIVE_VARIABLE; /* update planes when needed */ for_each_oldnew_plane_in_state(state, plane, old_plane_state, new_plane_state, i) { @@ -4859,6 +4860,16 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state, /* In commit tail framework this cannot happen */ WARN_ON(1); } + + /* For variable refresh rate mode only: + * Get vblank of last completed flip to avoid > 1 vrr flips per + * video frame by use of throttling, but allow flip programming + * anywhere in the possibly large variable vrr vblank interval + * for fine-grained flip timing control and more opportunity to + * avoid stutter on late submission of amdgpu_dm_do_flip() calls. + */ + last_flip_vblank = acrtc_attach->last_flip_vblank; + spin_unlock_irqrestore(&crtc->dev->event_lock, flags); if (!pflip_needed || plane->type == DRM_PLANE_TYPE_OVERLAY) { @@ -4882,10 +4893,18 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state, if (plane->type == DRM_PLANE_TYPE_PRIMARY) drm_crtc_vblank_get(crtc); + /* Use old throttling in non-vrr fixed refresh rate mode + * to keep flip scheduling based on target vblank counts + * working in a backwards compatible way, e.g., clients + * using GLX_OML_sync_control extension. + */ + if (!vrr_active) + last_flip_vblank = drm_crtc_vblank_count(crtc); + amdgpu_dm_do_flip( crtc, fb, - (uint32_t)drm_crtc_vblank_count(crtc) + *wait_for_vblank, + (uint32_t) last_flip_vblank + *wait_for_vblank, dc_state); } -- cgit v1.2.3 From 2216322919c8608a448d7ebc560a845238a5d6b6 Mon Sep 17 00:00:00 2001 From: Nicholas Kazlauskas Date: Mon, 7 Jan 2019 12:41:46 -0500 Subject: drm: Block fb changes for async plane updates The prepare_fb call always happens on new_plane_state. The drm_atomic_helper_cleanup_planes checks to see if plane state pointer has changed when deciding to call cleanup_fb on either the new_plane_state or the old_plane_state. For a non-async atomic commit the state pointer is swapped, so this helper calls prepare_fb on the new_plane_state and cleanup_fb on the old_plane_state. This makes sense, since we want to prepare the framebuffer we are going to use and cleanup the the framebuffer we are no longer using. For the async atomic update helpers this differs. The async atomic update helpers perform in-place updates on the existing state. They call drm_atomic_helper_cleanup_planes but the state pointer is not swapped. This means that prepare_fb is called on the new_plane_state and cleanup_fb is called on the new_plane_state (not the old). In the case where old_plane_state->fb == new_plane_state->fb then there should be no behavioral difference between an async update and a non-async commit. But there are issues that arise when old_plane_state->fb != new_plane_state->fb. The first is that the new_plane_state->fb is immediately cleaned up after it has been prepared, so we're using a fb that we shouldn't be. The second occurs during a sequence of async atomic updates and non-async regular atomic commits. Suppose there are two framebuffers being interleaved in a double-buffering scenario, fb1 and fb2: - Async update, oldfb = NULL, newfb = fb1, prepare fb1, cleanup fb1 - Async update, oldfb = fb1, newfb = fb2, prepare fb2, cleanup fb2 - Non-async commit, oldfb = fb2, newfb = fb1, prepare fb1, cleanup fb2 We call cleanup_fb on fb2 twice in this example scenario, and any further use will result in use-after-free. The simple fix to this problem is to block framebuffer changes in the drm_atomic_helper_async_check function for now. v2: Move check by itself, add a FIXME (Daniel) Cc: Daniel Vetter Cc: Harry Wentland Cc: Andrey Grodzovsky Cc: # v4.14+ Fixes: fef9df8b5945 ("drm/atomic: initial support for asynchronous plane update") Signed-off-by: Nicholas Kazlauskas Acked-by: Andrey Grodzovsky Acked-by: Harry Wentland Reviewed-by: Daniel Vetter Signed-off-by: Harry Wentland Link: https://patchwork.freedesktop.org/patch/275364/ Signed-off-by: Dave Airlie --- drivers/gpu/drm/drm_atomic_helper.c | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c index 54e2ae614dcc..f4290f6b0c38 100644 --- a/drivers/gpu/drm/drm_atomic_helper.c +++ b/drivers/gpu/drm/drm_atomic_helper.c @@ -1602,6 +1602,15 @@ int drm_atomic_helper_async_check(struct drm_device *dev, old_plane_state->crtc != new_plane_state->crtc) return -EINVAL; + /* + * FIXME: Since prepare_fb and cleanup_fb are always called on + * the new_plane_state for async updates we need to block framebuffer + * changes. This prevents use of a fb that's been cleaned up and + * double cleanups from occuring. + */ + if (old_plane_state->fb != new_plane_state->fb) + return -EINVAL; + funcs = plane->helper_private; if (!funcs->atomic_async_update) return -EINVAL; -- cgit v1.2.3 From 17fb465f16027ea22225b282295f5b8af19992e0 Mon Sep 17 00:00:00 2001 From: Alistair Francis Date: Thu, 21 Feb 2019 00:33:03 +0000 Subject: drm/bochs: Fix the ID mismatch error When running RISC-V QEMU with the Bochs device attached via PCIe the probe of the Bochs device fails with: [drm:bochs_hw_init] *ERROR* ID mismatch This was introduced by this commit: 7780eb9ce8 bochs: convert to drm_dev_register To fix the error we ensure that pci_enable_device() is called before bochs_load(). Fixes: 7780eb9ce80f ("bochs: convert to drm_dev_register") Signed-off-by: Alistair Francis Reported-by: David Abdurachmanov Link: http://patchwork.freedesktop.org/patch/msgid/20190221003231.31625-1-alistair.francis@wdc.com Signed-off-by: Gerd Hoffmann Signed-off-by: Dave Airlie --- drivers/gpu/drm/bochs/bochs_drv.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/bochs/bochs_drv.c b/drivers/gpu/drm/bochs/bochs_drv.c index f3dd66ae990a..aa35007262cd 100644 --- a/drivers/gpu/drm/bochs/bochs_drv.c +++ b/drivers/gpu/drm/bochs/bochs_drv.c @@ -154,6 +154,10 @@ static int bochs_pci_probe(struct pci_dev *pdev, if (IS_ERR(dev)) return PTR_ERR(dev); + ret = pci_enable_device(pdev); + if (ret) + goto err_free_dev; + dev->pdev = pdev; pci_set_drvdata(pdev, dev); -- cgit v1.2.3