From a993d319aebb7cce8a10c6e685344b7c2ad5c4c2 Mon Sep 17 00:00:00 2001 From: Zhu Lingshan Date: Wed, 11 Dec 2024 11:51:13 +0800 Subject: [PATCH 01/16] drm/amdkfd: wq_release signals dma_fence only when available kfd_process_wq_release() signals eviction fence by dma_fence_signal() which wanrs if dma_fence is NULL. kfd_process->ef is initialized by kfd_process_device_init_vm() through ioctl. That means the fence is NULL for a new created kfd_process, and close a kfd_process right after open it will trigger the warning. This commit conditionally signals the eviction fence in kfd_process_wq_release() only when it is available. [ 503.660882] WARNING: CPU: 0 PID: 9 at drivers/dma-buf/dma-fence.c:467 dma_fence_signal+0x74/0xa0 [ 503.782940] Workqueue: kfd_process_wq kfd_process_wq_release [amdgpu] [ 503.789640] RIP: 0010:dma_fence_signal+0x74/0xa0 [ 503.877620] Call Trace: [ 503.880066] [ 503.882168] ? __warn+0xcd/0x260 [ 503.885407] ? dma_fence_signal+0x74/0xa0 [ 503.889416] ? report_bug+0x288/0x2d0 [ 503.893089] ? handle_bug+0x53/0xa0 [ 503.896587] ? exc_invalid_op+0x14/0x50 [ 503.900424] ? asm_exc_invalid_op+0x16/0x20 [ 503.904616] ? dma_fence_signal+0x74/0xa0 [ 503.908626] kfd_process_wq_release+0x6b/0x370 [amdgpu] [ 503.914081] process_one_work+0x654/0x10a0 [ 503.918186] worker_thread+0x6c3/0xe70 [ 503.921943] ? srso_alias_return_thunk+0x5/0xfbef5 [ 503.926735] ? srso_alias_return_thunk+0x5/0xfbef5 [ 503.931527] ? __kthread_parkme+0x82/0x140 [ 503.935631] ? __pfx_worker_thread+0x10/0x10 [ 503.939904] kthread+0x2a8/0x380 [ 503.943132] ? __pfx_kthread+0x10/0x10 [ 503.946882] ret_from_fork+0x2d/0x70 [ 503.950458] ? __pfx_kthread+0x10/0x10 [ 503.954210] ret_from_fork_asm+0x1a/0x30 [ 503.958142] [ 503.960328] ---[ end trace 0000000000000000 ]--- Fixes: 967d226eaae8 ("dma-buf: add WARN_ON() illegal dma-fence signaling") Signed-off-by: Zhu Lingshan Reviewed-by: Felix Kuehling Signed-off-by: Alex Deucher (cherry picked from commit 2774ef7625adb5fb9e9265c26a59dca7b8fd171e) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index d0ee173acf82..edfe0b4788f4 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -1160,7 +1160,8 @@ static void kfd_process_wq_release(struct work_struct *work) */ synchronize_rcu(); ef = rcu_access_pointer(p->ef); - dma_fence_signal(ef); + if (ef) + dma_fence_signal(ef); kfd_process_remove_sysfs(p); -- 2.50.1 From 2a238b09bfd04e8155a7a323364bce1c38b28c0f Mon Sep 17 00:00:00 2001 From: Kun Liu Date: Fri, 27 Dec 2024 11:43:22 +0800 Subject: [PATCH 02/16] drm/amd/pm: fix BUG: scheduling while atomic atomic scheduling will be triggered in interrupt handler for AC/DC mode switch as following backtrace. Call Trace: dump_stack_lvl __schedule_bug __schedule schedule schedule_preempt_disabled __mutex_lock smu_cmn_send_smc_msg_with_param smu_v13_0_irq_process amdgpu_irq_dispatch amdgpu_ih_process amdgpu_irq_handler __handle_irq_event_percpu handle_irq_event handle_edge_irq __common_interrupt common_interrupt asm_common_interrupt Reviewed-by: Lijo Lazar Reviewed-by: Kenneth Feng Signed-off-by: Kun Liu Signed-off-by: Alex Deucher (cherry picked from commit 03cc84b102d1a832e8dfc59344346dedcebcdf42) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h | 2 ++ drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c | 12 ++++++------ drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c | 1 + drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c | 1 + 4 files changed, 10 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h index ae3563d71fa0..356d9422b411 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h @@ -303,5 +303,7 @@ int smu_v13_0_set_wbrf_exclusion_ranges(struct smu_context *smu, int smu_v13_0_get_boot_freq_by_index(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t *value); + +void smu_v13_0_interrupt_work(struct smu_context *smu); #endif #endif diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c index 2bfea740dace..2d1e7ebd1bac 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c @@ -1320,11 +1320,11 @@ static int smu_v13_0_set_irq_state(struct amdgpu_device *adev, return 0; } -static int smu_v13_0_ack_ac_dc_interrupt(struct smu_context *smu) +void smu_v13_0_interrupt_work(struct smu_context *smu) { - return smu_cmn_send_smc_msg(smu, - SMU_MSG_ReenableAcDcInterrupt, - NULL); + smu_cmn_send_smc_msg(smu, + SMU_MSG_ReenableAcDcInterrupt, + NULL); } #define THM_11_0__SRCID__THM_DIG_THERM_L2H 0 /* ASIC_TEMP > CG_THERMAL_INT.DIG_THERM_INTH */ @@ -1377,12 +1377,12 @@ static int smu_v13_0_irq_process(struct amdgpu_device *adev, switch (ctxid) { case SMU_IH_INTERRUPT_CONTEXT_ID_AC: dev_dbg(adev->dev, "Switched to AC mode!\n"); - smu_v13_0_ack_ac_dc_interrupt(smu); + schedule_work(&smu->interrupt_work); adev->pm.ac_power = true; break; case SMU_IH_INTERRUPT_CONTEXT_ID_DC: dev_dbg(adev->dev, "Switched to DC mode!\n"); - smu_v13_0_ack_ac_dc_interrupt(smu); + schedule_work(&smu->interrupt_work); adev->pm.ac_power = false; break; case SMU_IH_INTERRUPT_CONTEXT_ID_THERMAL_THROTTLING: diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c index 3aa705aae4c0..2a09b27788e8 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c @@ -3219,6 +3219,7 @@ static const struct pptable_funcs smu_v13_0_0_ppt_funcs = { .is_asic_wbrf_supported = smu_v13_0_0_wbrf_support_check, .enable_uclk_shadow = smu_v13_0_enable_uclk_shadow, .set_wbrf_exclusion_ranges = smu_v13_0_set_wbrf_exclusion_ranges, + .interrupt_work = smu_v13_0_interrupt_work, }; void smu_v13_0_0_set_ppt_funcs(struct smu_context *smu) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c index aabb94796005..55ef18517b0f 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c @@ -2797,6 +2797,7 @@ static const struct pptable_funcs smu_v13_0_7_ppt_funcs = { .is_asic_wbrf_supported = smu_v13_0_7_wbrf_support_check, .enable_uclk_shadow = smu_v13_0_enable_uclk_shadow, .set_wbrf_exclusion_ranges = smu_v13_0_set_wbrf_exclusion_ranges, + .interrupt_work = smu_v13_0_interrupt_work, }; void smu_v13_0_7_set_ppt_funcs(struct smu_context *smu) -- 2.50.1 From 75c8b703e5bded1e33b08fb09b829e7c2c1ed50a Mon Sep 17 00:00:00 2001 From: Arunpravin Paneer Selvam Date: Tue, 10 Dec 2024 12:50:08 +0530 Subject: [PATCH 03/16] drm/amdgpu: Add a lock when accessing the buddy trim function MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit When running YouTube videos and Steam games simultaneously, the tester found a system hang / race condition issue with the multi-display configuration setting. Adding a lock to the buddy allocator's trim function would be the solution. [ 7197.250436] general protection fault, probably for non-canonical address 0xdead000000000108 [ 7197.250447] RIP: 0010:__alloc_range+0x8b/0x340 [amddrm_buddy] [ 7197.250470] Call Trace: [ 7197.250472] [ 7197.250475] ? show_regs+0x6d/0x80 [ 7197.250481] ? die_addr+0x37/0xa0 [ 7197.250483] ? exc_general_protection+0x1db/0x480 [ 7197.250488] ? drm_suballoc_new+0x13c/0x93d [drm_suballoc_helper] [ 7197.250493] ? asm_exc_general_protection+0x27/0x30 [ 7197.250498] ? __alloc_range+0x8b/0x340 [amddrm_buddy] [ 7197.250501] ? __alloc_range+0x109/0x340 [amddrm_buddy] [ 7197.250506] amddrm_buddy_block_trim+0x1b5/0x260 [amddrm_buddy] [ 7197.250511] amdgpu_vram_mgr_new+0x4f5/0x590 [amdgpu] [ 7197.250682] amdttm_resource_alloc+0x46/0xb0 [amdttm] [ 7197.250689] ttm_bo_alloc_resource+0xe4/0x370 [amdttm] [ 7197.250696] amdttm_bo_validate+0x9d/0x180 [amdttm] [ 7197.250701] amdgpu_bo_pin+0x15a/0x2f0 [amdgpu] [ 7197.250831] amdgpu_dm_plane_helper_prepare_fb+0xb2/0x360 [amdgpu] [ 7197.251025] ? try_wait_for_completion+0x59/0x70 [ 7197.251030] drm_atomic_helper_prepare_planes.part.0+0x2f/0x1e0 [ 7197.251035] drm_atomic_helper_prepare_planes+0x5d/0x70 [ 7197.251037] drm_atomic_helper_commit+0x84/0x160 [ 7197.251040] drm_atomic_nonblocking_commit+0x59/0x70 [ 7197.251043] drm_mode_atomic_ioctl+0x720/0x850 [ 7197.251047] ? __pfx_drm_mode_atomic_ioctl+0x10/0x10 [ 7197.251049] drm_ioctl_kernel+0xb9/0x120 [ 7197.251053] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7197.251056] drm_ioctl+0x2d4/0x550 [ 7197.251058] ? __pfx_drm_mode_atomic_ioctl+0x10/0x10 [ 7197.251063] amdgpu_drm_ioctl+0x4e/0x90 [amdgpu] [ 7197.251186] __x64_sys_ioctl+0xa0/0xf0 [ 7197.251190] x64_sys_call+0x143b/0x25c0 [ 7197.251193] do_syscall_64+0x7f/0x180 [ 7197.251197] ? srso_alias_return_thunk+0x5/0xfbef5 [ 7197.251199] ? amdgpu_display_user_framebuffer_create+0x215/0x320 [amdgpu] [ 7197.251329] ? drm_internal_framebuffer_create+0xb7/0x1a0 [ 7197.251332] ? srso_alias_return_thunk+0x5/0xfbef5 Signed-off-by: Arunpravin Paneer Selvam Fixes: 4a5ad08f5377 ("drm/amdgpu: Add address alignment support to DCC buffers") Acked-by: Christian König Signed-off-by: Alex Deucher (cherry picked from commit 3318ba94e56b9183d0304577c74b33b6b01ce516) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index 7d26a962f811..ff5e52025266 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -567,7 +567,6 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, else remaining_size -= size; } - mutex_unlock(&mgr->lock); if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS && adjust_dcc_size) { struct drm_buddy_block *dcc_block; @@ -584,6 +583,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, (u64)vres->base.size, &vres->blocks); } + mutex_unlock(&mgr->lock); vres->base.start = 0; size = max_t(u64, amdgpu_vram_mgr_blocks_size(&vres->blocks), -- 2.50.1 From 77bf21a03a2ad45cf66f73f13154b1669d9cf52a Mon Sep 17 00:00:00 2001 From: Suraj Kandpal Date: Fri, 3 Jan 2025 14:15:17 +0530 Subject: [PATCH 04/16] Revert "drm/i915/hdcp: Don't enable HDCP1.4 directly from check_link" This reverts commit 483f7d94a0453564ad9295288c0242136c5f36a0. This needs to be reverted since HDCP even after updating the connector state HDCP property we don't reenable HDCP until the next commit in which the CP Property is set causing compliance to fail. --v2 -Fix build issue [Dnyaneshwar] Signed-off-by: Suraj Kandpal Reviewed-by: Dnyaneshwar Bhadane Link: https://patchwork.freedesktop.org/patch/msgid/20250103084517.239998-1-suraj.kandpal@intel.com (cherry picked from commit fcf73e20cd1fe60c3ba5f9626f1e8f9cd4511edf) Signed-off-by: Tvrtko Ursulin --- drivers/gpu/drm/i915/display/intel_hdcp.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/i915/display/intel_hdcp.c b/drivers/gpu/drm/i915/display/intel_hdcp.c index f57e4dba2873..8fee26d791f4 100644 --- a/drivers/gpu/drm/i915/display/intel_hdcp.c +++ b/drivers/gpu/drm/i915/display/intel_hdcp.c @@ -1158,9 +1158,15 @@ static int intel_hdcp_check_link(struct intel_connector *connector) goto out; } - intel_hdcp_update_value(connector, - DRM_MODE_CONTENT_PROTECTION_DESIRED, - true); + ret = intel_hdcp1_enable(connector); + if (ret) { + drm_err(display->drm, "Failed to enable hdcp (%d)\n", ret); + intel_hdcp_update_value(connector, + DRM_MODE_CONTENT_PROTECTION_DESIRED, + true); + goto out; + } + out: mutex_unlock(&dig_port->hdcp_mutex); mutex_unlock(&hdcp->mutex); -- 2.50.1 From 9ab4981552930a9c45682d62424ba610edc3992d Mon Sep 17 00:00:00 2001 From: Lucas De Marchi Date: Thu, 2 Jan 2025 16:11:10 -0800 Subject: [PATCH 05/16] drm/xe: Fix tlb invalidation when wedging MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit If GuC fails to load, the driver wedges, but in the process it tries to do stuff that may not be initialized yet. This moves the xe_gt_tlb_invalidation_init() to be done earlier: as its own doc says, it's a software-only initialization and should had been named with the _early() suffix. Move it to be called by xe_gt_init_early(), so the locks and seqno are initialized, avoiding a NULL ptr deref when wedging: xe 0000:03:00.0: [drm] *ERROR* GT0: load failed: status: Reset = 0, BootROM = 0x50, UKernel = 0x00, MIA = 0x00, Auth = 0x01 xe 0000:03:00.0: [drm] *ERROR* GT0: firmware signature verification failed xe 0000:03:00.0: [drm] *ERROR* CRITICAL: Xe has declared device 0000:03:00.0 as wedged. ... BUG: kernel NULL pointer dereference, address: 0000000000000000 #PF: supervisor read access in kernel mode #PF: error_code(0x0000) - not-present page PGD 0 P4D 0 Oops: Oops: 0000 [#1] PREEMPT SMP NOPTI CPU: 9 UID: 0 PID: 3908 Comm: modprobe Tainted: G U W 6.13.0-rc4-xe+ #3 Tainted: [U]=USER, [W]=WARN Hardware name: Intel Corporation Alder Lake Client Platform/AlderLake-S ADP-S DDR5 UDIMM CRB, BIOS ADLSFWI1.R00.3275.A00.2207010640 07/01/2022 RIP: 0010:xe_gt_tlb_invalidation_reset+0x75/0x110 [xe] This can be easily triggered by poking the GuC binary to force a signature failure. There will still be an extra message, xe 0000:03:00.0: [drm] *ERROR* GT0: GuC mmio request 0x4100: no reply 0x4100 but that's better than a NULL ptr deref. Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/3956 Fixes: c9474b726b93 ("drm/xe: Wedge the entire device") Reviewed-by: Matthew Brost Link: https://patchwork.freedesktop.org/patch/msgid/20250103001111.331684-2-lucas.demarchi@intel.com Signed-off-by: Lucas De Marchi (cherry picked from commit 5001ef3af8f2c972d6fd9c5221a8457556f8bea6) Signed-off-by: Thomas Hellström --- drivers/gpu/drm/xe/xe_gt.c | 8 ++++---- drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c | 4 ++-- drivers/gpu/drm/xe/xe_gt_tlb_invalidation.h | 3 ++- 3 files changed, 8 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_gt.c b/drivers/gpu/drm/xe/xe_gt.c index d6744be01a68..94d468d01253 100644 --- a/drivers/gpu/drm/xe/xe_gt.c +++ b/drivers/gpu/drm/xe/xe_gt.c @@ -387,6 +387,10 @@ int xe_gt_init_early(struct xe_gt *gt) xe_force_wake_init_gt(gt, gt_to_fw(gt)); spin_lock_init(>->global_invl_lock); + err = xe_gt_tlb_invalidation_init_early(gt); + if (err) + return err; + return 0; } @@ -588,10 +592,6 @@ int xe_gt_init(struct xe_gt *gt) xe_hw_fence_irq_init(>->fence_irq[i]); } - err = xe_gt_tlb_invalidation_init(gt); - if (err) - return err; - err = xe_gt_pagefault_init(gt); if (err) return err; diff --git a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c index 6146d1776bda..0a0af413770e 100644 --- a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c +++ b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c @@ -106,7 +106,7 @@ static void xe_gt_tlb_fence_timeout(struct work_struct *work) } /** - * xe_gt_tlb_invalidation_init - Initialize GT TLB invalidation state + * xe_gt_tlb_invalidation_init_early - Initialize GT TLB invalidation state * @gt: graphics tile * * Initialize GT TLB invalidation state, purely software initialization, should @@ -114,7 +114,7 @@ static void xe_gt_tlb_fence_timeout(struct work_struct *work) * * Return: 0 on success, negative error code on error. */ -int xe_gt_tlb_invalidation_init(struct xe_gt *gt) +int xe_gt_tlb_invalidation_init_early(struct xe_gt *gt) { gt->tlb_invalidation.seqno = 1; INIT_LIST_HEAD(>->tlb_invalidation.pending_fences); diff --git a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.h b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.h index 00b1c6c01e8d..672acfcdf0d7 100644 --- a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.h +++ b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.h @@ -14,7 +14,8 @@ struct xe_gt; struct xe_guc; struct xe_vma; -int xe_gt_tlb_invalidation_init(struct xe_gt *gt); +int xe_gt_tlb_invalidation_init_early(struct xe_gt *gt); + void xe_gt_tlb_invalidation_reset(struct xe_gt *gt); int xe_gt_tlb_invalidation_ggtt(struct xe_gt *gt); int xe_gt_tlb_invalidation_vma(struct xe_gt *gt, -- 2.50.1 From b84e1cd22f8a8c03b7b1051372560c7017c8be92 Mon Sep 17 00:00:00 2001 From: Rodrigo Vivi Date: Thu, 19 Dec 2024 18:55:36 -0500 Subject: [PATCH 06/16] drm/xe/dg1: Fix power gate sequence. MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit sub-pipe PG is not present on DG1. Setting these bits can disable other power gates and cause GPU hangs on video playbacks. VLK: 16314, 4304 Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13381 Cc: Vinay Belgaumkar Cc: Himal Prasad Ghimiray Reviewed-by: Himal Prasad Ghimiray Link: https://patchwork.freedesktop.org/patch/msgid/20241219235536.454270-1-rodrigo.vivi@intel.com Signed-off-by: Rodrigo Vivi (cherry picked from commit 2f12e9c029315c1400059b2e7fdf53117c09c3a9) Signed-off-by: Thomas Hellström --- drivers/gpu/drm/xe/xe_gt_idle.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_gt_idle.c b/drivers/gpu/drm/xe/xe_gt_idle.c index fd80afeef56a..ffd3ba7f6656 100644 --- a/drivers/gpu/drm/xe/xe_gt_idle.c +++ b/drivers/gpu/drm/xe/xe_gt_idle.c @@ -122,10 +122,12 @@ void xe_gt_idle_enable_pg(struct xe_gt *gt) if (!xe_gt_is_media_type(gt)) gtidle->powergate_enable |= RENDER_POWERGATE_ENABLE; - for (i = XE_HW_ENGINE_VCS0, j = 0; i <= XE_HW_ENGINE_VCS7; ++i, ++j) { - if ((gt->info.engine_mask & BIT(i))) - gtidle->powergate_enable |= (VDN_HCP_POWERGATE_ENABLE(j) | - VDN_MFXVDENC_POWERGATE_ENABLE(j)); + if (xe->info.platform != XE_DG1) { + for (i = XE_HW_ENGINE_VCS0, j = 0; i <= XE_HW_ENGINE_VCS7; ++i, ++j) { + if ((gt->info.engine_mask & BIT(i))) + gtidle->powergate_enable |= (VDN_HCP_POWERGATE_ENABLE(j) | + VDN_MFXVDENC_POWERGATE_ENABLE(j)); + } } fw_ref = xe_force_wake_get(gt_to_fw(gt), XE_FW_GT); -- 2.50.1 From 11510e67d0bd956878ab4ffa03c45766788092c1 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 8 Jan 2025 15:17:12 -0500 Subject: [PATCH 07/16] drm/amdgpu/smu13: update powersave optimizations Only apply when compute profile is selected. This is the only supported configuration. Selecting other profiles can lead to performane degradations. Reviewed-by: Kenneth Feng Signed-off-by: Alex Deucher (cherry picked from commit d477e39532d725b1cdb3c8005c689c74ffbf3b94) Cc: stable@vger.kernel.org # 6.12.x --- drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c index 2a09b27788e8..0551a3311217 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c @@ -2643,11 +2643,12 @@ static int smu_v13_0_0_set_power_profile_mode(struct smu_context *smu, &backend_workload_mask); /* Add optimizations for SMU13.0.0/10. Reuse the power saving profile */ - if ((amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 0) && - ((smu->adev->pm.fw_version == 0x004e6601) || - (smu->adev->pm.fw_version >= 0x004e7300))) || - (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 10) && - smu->adev->pm.fw_version >= 0x00504500)) { + if ((workload_mask & (1 << PP_SMC_POWER_PROFILE_COMPUTE)) && + ((amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 0) && + ((smu->adev->pm.fw_version == 0x004e6601) || + (smu->adev->pm.fw_version >= 0x004e7300))) || + (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 10) && + smu->adev->pm.fw_version >= 0x00504500))) { workload_type = smu_cmn_to_asic_specific_index(smu, CMN2ASIC_MAPPING_WORKLOAD, PP_SMC_POWER_PROFILE_POWERSAVING); -- 2.50.1 From b5cd418f016fb801be413fd52fe4711d2d13018c Mon Sep 17 00:00:00 2001 From: Wayne Lin Date: Tue, 10 Dec 2024 11:17:55 +0800 Subject: [PATCH 08/16] drm/amd/display: Validate mdoe under MST LCT=1 case as well [Why & How] Currently in dm_dp_mst_is_port_support_mode(), when valdidating mode under dsc decoding at the last DP link config, we only validate the case when there is an UFP. However, if the MSTB LCT=1, there is no UFP. Under this case, use root_link_bw_in_kbps as the available bw to compare. Link: https://gitlab.freedesktop.org/drm/amd/-/issues/3720 Fixes: fa57924c76d9 ("drm/amd/display: Refactor function dm_dp_mst_is_port_support_mode()") Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Jerry Zuo Signed-off-by: Wayne Lin Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit a04d9534a8a75b2806c5321c387be450c364b55e) Cc: stable@vger.kernel.org --- .../amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-) 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 6e4359490613..aadaa61ac5ac 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 @@ -1835,11 +1835,15 @@ enum dc_status dm_dp_mst_is_port_support_mode( if (immediate_upstream_port) { virtual_channel_bw_in_kbps = kbps_from_pbn(immediate_upstream_port->full_pbn); virtual_channel_bw_in_kbps = min(root_link_bw_in_kbps, virtual_channel_bw_in_kbps); - if (bw_range.min_kbps > virtual_channel_bw_in_kbps) { - DRM_DEBUG_DRIVER("MST_DSC dsc decode at last link." - "Max dsc compression can't fit into MST available bw\n"); - return DC_FAIL_BANDWIDTH_VALIDATE; - } + } else { + /* For topology LCT 1 case - only one mstb*/ + virtual_channel_bw_in_kbps = root_link_bw_in_kbps; + } + + if (bw_range.min_kbps > virtual_channel_bw_in_kbps) { + DRM_DEBUG_DRIVER("MST_DSC dsc decode at last link." + "Max dsc compression can't fit into MST available bw\n"); + return DC_FAIL_BANDWIDTH_VALIDATE; } } -- 2.50.1 From adb4998f4928a17d91be054218a902ba9f8c1f93 Mon Sep 17 00:00:00 2001 From: Wayne Lin Date: Mon, 9 Dec 2024 15:25:35 +0800 Subject: [PATCH 09/16] drm/amd/display: Reduce accessing remote DPCD overhead [Why] Observed frame rate get dropped by tool like glxgear. Even though the output to monitor is 60Hz, the rendered frame rate drops to 30Hz lower. It's due to code path in some cases will trigger dm_dp_mst_is_port_support_mode() to read out remote Link status to assess the available bandwidth for dsc maniplation. Overhead of keep reading remote DPCD is considerable. [How] Store the remote link BW in mst_local_bw and use end-to-end full_pbn as an indicator to decide whether update the remote link bw or not. Whenever we need the info to assess the BW, visit the stored one first. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3720 Fixes: fa57924c76d9 ("drm/amd/display: Refactor function dm_dp_mst_is_port_support_mode()") Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Jerry Zuo Signed-off-by: Wayne Lin Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit 4a9a918545455a5979c6232fcf61ed3d8f0db3ae) Cc: stable@vger.kernel.org --- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h | 2 ++ .../display/amdgpu_dm/amdgpu_dm_mst_types.c | 34 ++++++++++++++----- 2 files changed, 27 insertions(+), 9 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h index 6464a8378387..2227cd8e4a89 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h @@ -697,6 +697,8 @@ struct amdgpu_dm_connector { struct drm_dp_mst_port *mst_output_port; struct amdgpu_dm_connector *mst_root; struct drm_dp_aux *dsc_aux; + uint32_t mst_local_bw; + uint16_t vc_full_pbn; struct mutex handle_mst_msg_ready; /* TODO see if we can merge with ddc_bus or make a dm_connector */ 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 aadaa61ac5ac..1080075ccb17 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 @@ -155,6 +155,17 @@ amdgpu_dm_mst_connector_late_register(struct drm_connector *connector) return 0; } + +static inline void +amdgpu_dm_mst_reset_mst_connector_setting(struct amdgpu_dm_connector *aconnector) +{ + aconnector->drm_edid = NULL; + aconnector->dsc_aux = NULL; + aconnector->mst_output_port->passthrough_aux = NULL; + aconnector->mst_local_bw = 0; + aconnector->vc_full_pbn = 0; +} + static void amdgpu_dm_mst_connector_early_unregister(struct drm_connector *connector) { @@ -182,9 +193,7 @@ amdgpu_dm_mst_connector_early_unregister(struct drm_connector *connector) dc_sink_release(dc_sink); aconnector->dc_sink = NULL; - aconnector->drm_edid = NULL; - aconnector->dsc_aux = NULL; - port->passthrough_aux = NULL; + amdgpu_dm_mst_reset_mst_connector_setting(aconnector); } aconnector->mst_status = MST_STATUS_DEFAULT; @@ -504,9 +513,7 @@ dm_dp_mst_detect(struct drm_connector *connector, dc_sink_release(aconnector->dc_sink); aconnector->dc_sink = NULL; - aconnector->drm_edid = NULL; - aconnector->dsc_aux = NULL; - port->passthrough_aux = NULL; + amdgpu_dm_mst_reset_mst_connector_setting(aconnector); amdgpu_dm_set_mst_status(&aconnector->mst_status, MST_REMOTE_EDID | MST_ALLOCATE_NEW_PAYLOAD | MST_CLEAR_ALLOCATED_PAYLOAD, @@ -1819,9 +1826,18 @@ enum dc_status dm_dp_mst_is_port_support_mode( struct drm_dp_mst_port *immediate_upstream_port = NULL; uint32_t end_link_bw = 0; - /*Get last DP link BW capability*/ - if (dp_get_link_current_set_bw(&aconnector->mst_output_port->aux, &end_link_bw)) { - if (stream_kbps > end_link_bw) { + /*Get last DP link BW capability. Mode shall be supported by Legacy peer*/ + if (aconnector->mst_output_port->pdt != DP_PEER_DEVICE_DP_LEGACY_CONV && + aconnector->mst_output_port->pdt != DP_PEER_DEVICE_NONE) { + if (aconnector->vc_full_pbn != aconnector->mst_output_port->full_pbn) { + dp_get_link_current_set_bw(&aconnector->mst_output_port->aux, &end_link_bw); + aconnector->vc_full_pbn = aconnector->mst_output_port->full_pbn; + aconnector->mst_local_bw = end_link_bw; + } else { + end_link_bw = aconnector->mst_local_bw; + } + + if (end_link_bw > 0 && stream_kbps > end_link_bw) { DRM_DEBUG_DRIVER("MST_DSC dsc decode at last link." "Mode required bw can't fit into last link\n"); return DC_FAIL_BANDWIDTH_VALIDATE; -- 2.50.1 From 3412860cc4c0c484f53f91b371483e6e4440c3e5 Mon Sep 17 00:00:00 2001 From: Nicholas Susanto Date: Thu, 19 Dec 2024 14:15:37 -0500 Subject: [PATCH 10/16] Revert "drm/amd/display: Enable urgent latency adjustments for DCN35" Revert commit 284f141f5ce5 ("drm/amd/display: Enable urgent latency adjustments for DCN35") [Why & How] Urgent latency increase caused 2.8K OLED monitor caused it to block this panel support P0. Reverting this change does not reintroduce the netflix corruption issue which it fixed. Fixes: 284f141f5ce5 ("drm/amd/display: Enable urgent latency adjustments for DCN35") Reviewed-by: Charlene Liu Signed-off-by: Nicholas Susanto Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit c7ccfc0d4241a834c25a9a9e1e78b388b4445d23) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c index beed7adbbd43..47d785204f29 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c @@ -195,9 +195,9 @@ struct _vcs_dpi_soc_bounding_box_st dcn3_5_soc = { .dcn_downspread_percent = 0.5, .gpuvm_min_page_size_bytes = 4096, .hostvm_min_page_size_bytes = 4096, - .do_urgent_latency_adjustment = 1, + .do_urgent_latency_adjustment = 0, .urgent_latency_adjustment_fabric_clock_component_us = 0, - .urgent_latency_adjustment_fabric_clock_reference_mhz = 3000, + .urgent_latency_adjustment_fabric_clock_reference_mhz = 0, }; void dcn35_build_wm_range_table_fpu(struct clk_mgr *clk_mgr) -- 2.50.1 From ff2e4d874726c549130308b6b46aa0f8a34e04cb Mon Sep 17 00:00:00 2001 From: Leo Li Date: Mon, 9 Dec 2024 12:58:33 -0500 Subject: [PATCH 11/16] drm/amd/display: Do not wait for PSR disable on vbl enable [Why] Outside of a modeset/link configuration change, we should not have to wait for the panel to exit PSR. Depending on the panel and it's state, it may take multiple frames for it to exit PSR. Therefore, waiting in all scenarios may cause perceived stuttering, especially in combination with faster vblank shutdown. [How] PSR1 disable is hooked up to the vblank enable event, and vice versa. In case of vblank enable, do not wait for panel to exit PSR, but still wait in all other cases. We also avoid a call to unnecessarily change power_opts on disable - this ends up sending another command to dmcub fw. When testing against IGT, some crc tests like kms_plane_alpha_blend and amd_hotplug were failing due to CRC timeouts. This was found to be caused by the early return before HW has fully exited PSR1. Fix this by first making sure we grab a vblank reference, then waiting for panel to exit PSR1, before programming hw for CRC generation. Fixes: 58a261bfc967 ("drm/amd/display: use a more lax vblank enable policy for older ASICs") Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3743 Reviewed-by: Tom Chung Signed-off-by: Leo Li Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit aa6713fa2046f4c09bf3013dd1420ae15603ca6f) Cc: stable@vger.kernel.org --- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 4 +-- .../drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c | 25 ++++++++----- .../amd/display/amdgpu_dm/amdgpu_dm_crtc.c | 2 +- .../amd/display/amdgpu_dm/amdgpu_dm_debugfs.c | 2 +- .../drm/amd/display/amdgpu_dm/amdgpu_dm_psr.c | 35 +++++++++++++++++-- .../drm/amd/display/amdgpu_dm/amdgpu_dm_psr.h | 3 +- 6 files changed, 54 insertions(+), 17 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 cd16dae534dc..56b47e02db0b 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -9100,7 +9100,7 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state, acrtc_state->stream->link->psr_settings.psr_dirty_rects_change_timestamp_ns = timestamp_ns; if (acrtc_state->stream->link->psr_settings.psr_allow_active) - amdgpu_dm_psr_disable(acrtc_state->stream); + amdgpu_dm_psr_disable(acrtc_state->stream, true); mutex_unlock(&dm->dc_lock); } } @@ -9270,7 +9270,7 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state, if (acrtc_state->stream->link->replay_settings.replay_allow_active) amdgpu_dm_replay_disable(acrtc_state->stream); if (acrtc_state->stream->link->psr_settings.psr_allow_active) - amdgpu_dm_psr_disable(acrtc_state->stream); + amdgpu_dm_psr_disable(acrtc_state->stream, true); } mutex_unlock(&dm->dc_lock); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c index f936a35fa9eb..0f6ba7b1575d 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c @@ -30,6 +30,7 @@ #include "amdgpu_dm.h" #include "dc.h" #include "amdgpu_securedisplay.h" +#include "amdgpu_dm_psr.h" static const char *const pipe_crc_sources[] = { "none", @@ -224,6 +225,10 @@ int amdgpu_dm_crtc_configure_crc_source(struct drm_crtc *crtc, mutex_lock(&adev->dm.dc_lock); + /* For PSR1, check that the panel has exited PSR */ + if (stream_state->link->psr_settings.psr_version < DC_PSR_VERSION_SU_1) + amdgpu_dm_psr_wait_disable(stream_state); + /* Enable or disable CRTC CRC generation */ if (dm_is_crc_source_crtc(source) || source == AMDGPU_DM_PIPE_CRC_SOURCE_NONE) { if (!dc_stream_configure_crc(stream_state->ctx->dc, @@ -357,6 +362,17 @@ int amdgpu_dm_crtc_set_crc_source(struct drm_crtc *crtc, const char *src_name) } + /* + * Reading the CRC requires the vblank interrupt handler to be + * enabled. Keep a reference until CRC capture stops. + */ + enabled = amdgpu_dm_is_valid_crc_source(cur_crc_src); + if (!enabled && enable) { + ret = drm_crtc_vblank_get(crtc); + if (ret) + goto cleanup; + } + #if defined(CONFIG_DRM_AMD_SECURE_DISPLAY) /* Reset secure_display when we change crc source from debugfs */ amdgpu_dm_set_crc_window_default(crtc, crtc_state->stream); @@ -367,16 +383,7 @@ int amdgpu_dm_crtc_set_crc_source(struct drm_crtc *crtc, const char *src_name) goto cleanup; } - /* - * Reading the CRC requires the vblank interrupt handler to be - * enabled. Keep a reference until CRC capture stops. - */ - enabled = amdgpu_dm_is_valid_crc_source(cur_crc_src); if (!enabled && enable) { - ret = drm_crtc_vblank_get(crtc); - if (ret) - goto cleanup; - if (dm_is_crc_source_dprx(source)) { if (drm_dp_start_crc(aux, crtc)) { DRM_DEBUG_DRIVER("dp start crc failed\n"); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c index 64a041c2af05..3011ea709b91 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c @@ -142,7 +142,7 @@ static void amdgpu_dm_crtc_set_panel_sr_feature( amdgpu_dm_replay_enable(vblank_work->stream, true); } else if (vblank_enabled) { if (link->psr_settings.psr_version < DC_PSR_VERSION_SU_1 && is_sr_active) - amdgpu_dm_psr_disable(vblank_work->stream); + amdgpu_dm_psr_disable(vblank_work->stream, false); } else if (link->psr_settings.psr_feature_enabled && allow_sr_entry && !is_sr_active && !is_crc_window_active) { 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 6a97bb2d9160..8b7c1a428f64 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 @@ -3606,7 +3606,7 @@ static int crc_win_update_set(void *data, u64 val) /* PSR may write to OTG CRC window control register, * so close it before starting secure_display. */ - amdgpu_dm_psr_disable(acrtc->dm_irq_params.stream); + amdgpu_dm_psr_disable(acrtc->dm_irq_params.stream, true); spin_lock_irq(&adev_to_drm(adev)->event_lock); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.c index f40240aafe98..45858bf1523d 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.c @@ -201,14 +201,13 @@ void amdgpu_dm_psr_enable(struct dc_stream_state *stream) * * Return: true if success */ -bool amdgpu_dm_psr_disable(struct dc_stream_state *stream) +bool amdgpu_dm_psr_disable(struct dc_stream_state *stream, bool wait) { - unsigned int power_opt = 0; bool psr_enable = false; DRM_DEBUG_DRIVER("Disabling psr...\n"); - return dc_link_set_psr_allow_active(stream->link, &psr_enable, true, false, &power_opt); + return dc_link_set_psr_allow_active(stream->link, &psr_enable, wait, false, NULL); } /* @@ -251,3 +250,33 @@ bool amdgpu_dm_psr_is_active_allowed(struct amdgpu_display_manager *dm) return allow_active; } + +/** + * amdgpu_dm_psr_wait_disable() - Wait for eDP panel to exit PSR + * @stream: stream state attached to the eDP link + * + * Waits for a max of 500ms for the eDP panel to exit PSR. + * + * Return: true if panel exited PSR, false otherwise. + */ +bool amdgpu_dm_psr_wait_disable(struct dc_stream_state *stream) +{ + enum dc_psr_state psr_state = PSR_STATE0; + struct dc_link *link = stream->link; + int retry_count; + + if (link == NULL) + return false; + + for (retry_count = 0; retry_count <= 1000; retry_count++) { + dc_link_get_psr_state(link, &psr_state); + if (psr_state == PSR_STATE0) + break; + udelay(500); + } + + if (retry_count == 1000) + return false; + + return true; +} diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.h index cd2d45c2b5ef..e2366321a3c1 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_psr.h @@ -34,8 +34,9 @@ void amdgpu_dm_set_psr_caps(struct dc_link *link); void amdgpu_dm_psr_enable(struct dc_stream_state *stream); bool amdgpu_dm_link_setup_psr(struct dc_stream_state *stream); -bool amdgpu_dm_psr_disable(struct dc_stream_state *stream); +bool amdgpu_dm_psr_disable(struct dc_stream_state *stream, bool wait); bool amdgpu_dm_psr_disable_all(struct amdgpu_display_manager *dm); bool amdgpu_dm_psr_is_active_allowed(struct amdgpu_display_manager *dm); +bool amdgpu_dm_psr_wait_disable(struct dc_stream_state *stream); #endif /* AMDGPU_DM_AMDGPU_DM_PSR_H_ */ -- 2.50.1 From 35ca53b7b0f0ffd16c6675fd76abac9409cf83e0 Mon Sep 17 00:00:00 2001 From: Leo Li Date: Wed, 11 Dec 2024 12:06:24 -0500 Subject: [PATCH 12/16] drm/amd/display: Do not elevate mem_type change to full update [Why] There should not be any need to revalidate bandwidth on memory placement change, since the fb is expected to be pinned to DCN-accessable memory before scanout. For APU it's DRAM, and DGPU, it's VRAM. However, async flips + memory type change needs to be rejected. [How] Do not set lock_and_validation_needed on mem_type change. Instead, reject an async_flip request if the crtc's buffer(s) changed mem_type. This may fix stuttering/corruption experienced with PSR SU and PSR1 panels, if the compositor allocates fbs in both VRAM carveout and GTT and flips between them. Fixes: a7c0cad0dc06 ("drm/amd/display: ensure async flips are only accepted for fast updates") Reviewed-by: Tom Chung Signed-off-by: Leo Li Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit 4caacd1671b7a013ad04cd8b6398f002540bdd4d) Cc: stable@vger.kernel.org --- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 29 +++++++++++++++---- 1 file changed, 23 insertions(+), 6 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 56b47e02db0b..dcc5d8ded662 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -11379,6 +11379,25 @@ static int dm_crtc_get_cursor_mode(struct amdgpu_device *adev, return 0; } +static bool amdgpu_dm_crtc_mem_type_changed(struct drm_device *dev, + struct drm_atomic_state *state, + struct drm_crtc_state *crtc_state) +{ + struct drm_plane *plane; + struct drm_plane_state *new_plane_state, *old_plane_state; + + drm_for_each_plane_mask(plane, dev, crtc_state->plane_mask) { + new_plane_state = drm_atomic_get_plane_state(state, plane); + old_plane_state = drm_atomic_get_plane_state(state, plane); + + if (old_plane_state->fb && new_plane_state->fb && + get_mem_type(old_plane_state->fb) != get_mem_type(new_plane_state->fb)) + return true; + } + + return false; +} + /** * amdgpu_dm_atomic_check() - Atomic check implementation for AMDgpu DM. * @@ -11576,10 +11595,6 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, /* Remove exiting planes if they are modified */ for_each_oldnew_plane_in_descending_zpos(state, plane, old_plane_state, new_plane_state) { - if (old_plane_state->fb && new_plane_state->fb && - get_mem_type(old_plane_state->fb) != - get_mem_type(new_plane_state->fb)) - lock_and_validation_needed = true; ret = dm_update_plane_state(dc, state, plane, old_plane_state, @@ -11874,9 +11889,11 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, /* * Only allow async flips for fast updates that don't change - * the FB pitch, the DCC state, rotation, etc. + * the FB pitch, the DCC state, rotation, mem_type, etc. */ - if (new_crtc_state->async_flip && lock_and_validation_needed) { + if (new_crtc_state->async_flip && + (lock_and_validation_needed || + amdgpu_dm_crtc_mem_type_changed(dev, state, new_crtc_state))) { drm_dbg_atomic(crtc->dev, "[CRTC:%d:%s] async flips are only supported for fast updates\n", crtc->base.id, crtc->name); -- 2.50.1 From 470d4f05c77153b97f53554a3c44164361a4d939 Mon Sep 17 00:00:00 2001 From: Yiling Chen Date: Fri, 20 Dec 2024 17:12:54 +0800 Subject: [PATCH 13/16] drm/amd/display: Remove unnecessary eDP power down [why] When first time of link training is fail, eDP would be powered down and would not be powered up for next retry link training. It causes that all of retry link linking would be fail. [how] We has extracted both power up and down sequence from enable/disable link output function before DCN32. We remov eDP power down in dcn32_disable_link_output(). Reviewed-by: Charlene Liu Signed-off-by: Yiling Chen Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit f5860c88cdfe7300d08c1aef881bba0cac369e34) --- .../gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c index fa11f075d1f9..d9136d252ee1 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c @@ -1398,12 +1398,12 @@ void dcn32_disable_link_output(struct dc_link *link, link_hwss->disable_link_output(link, link_res, signal); link->phy_state.symclk_state = SYMCLK_OFF_TX_OFF; - - if (signal == SIGNAL_TYPE_EDP && - link->dc->hwss.edp_power_control && - !link->skip_implict_edp_power_control) - link->dc->hwss.edp_power_control(link, false); - else if (dmcu != NULL && dmcu->funcs->unlock_phy) + /* + * Add the logic to extract BOTH power up and power down sequences + * from enable/disable link output and only call edp panel control + * in enable_link_dp and disable_link_dp once. + */ + if (dmcu != NULL && dmcu->funcs->unlock_phy) dmcu->funcs->unlock_phy(dmcu); dc->link_srv->dp_trace_source_sequence(link, DPCD_SOURCE_SEQ_AFTER_DISABLE_LINK_PHY); -- 2.50.1 From b5c764d6ed556c4e81fbe3fd976da77ec450c08e Mon Sep 17 00:00:00 2001 From: Tom Chung Date: Tue, 1 Oct 2024 17:13:07 +0800 Subject: [PATCH 14/16] drm/amd/display: Use HW lock mgr for PSR1 [Why] Without the dmub hw lock, it may cause the lock timeout issue while do modeset on PSR1 eDP panel. [How] Allow dmub hw lock for PSR1. Reviewed-by: Sun peng Li Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit a2b5a9956269f4c1a09537177f18ab0229fe79f7) --- drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c index bf636b28e3e1..5bb8b78bf250 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c @@ -63,7 +63,8 @@ void dmub_hw_lock_mgr_inbox0_cmd(struct dc_dmub_srv *dmub_srv, bool should_use_dmub_lock(struct dc_link *link) { - if (link->psr_settings.psr_version == DC_PSR_VERSION_SU_1) + if (link->psr_settings.psr_version == DC_PSR_VERSION_SU_1 || + link->psr_settings.psr_version == DC_PSR_VERSION_1) return true; if (link->replay_settings.replay_feature_enabled) -- 2.50.1 From 36b23e3baf9129d5b6c3a3a85b6b7ffb75ae287c Mon Sep 17 00:00:00 2001 From: Alex Hung Date: Tue, 17 Dec 2024 11:51:58 -0700 Subject: [PATCH 15/16] drm/amd/display: Initialize denominator defaults to 1 [WHAT & HOW] Variables, used as denominators and maybe not assigned to other values, should be initialized to non-zero to avoid DIVIDE_BY_ZERO, as reported by Coverity. Reviewed-by: Austin Zheng Reviewed-by: Rodrigo Siqueira Signed-off-by: Alex Hung Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit e2c4c6c10542ccfe4a0830bb6c9fd5b177b7bbb7) --- .../dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c index 601320b1be81..ecfa3c898e09 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c @@ -579,8 +579,8 @@ static void CalculateBytePerPixelAndBlockSizes( { *BytePerPixelDETY = 0; *BytePerPixelDETC = 0; - *BytePerPixelY = 0; - *BytePerPixelC = 0; + *BytePerPixelY = 1; + *BytePerPixelC = 1; if (SourcePixelFormat == dml2_444_64) { *BytePerPixelDETY = 8; -- 2.50.1 From b0a3e840ad287c33a86b5515d606451b7df86ad4 Mon Sep 17 00:00:00 2001 From: Tom Chung Date: Thu, 5 Dec 2024 23:08:28 +0800 Subject: [PATCH 16/16] drm/amd/display: Fix PSR-SU not support but still call the amdgpu_dm_psr_enable [Why] The enum DC_PSR_VERSION_SU_1 of psr_version is 1 and DC_PSR_VERSION_UNSUPPORTED is 0xFFFFFFFF. The original code may has chance trigger the amdgpu_dm_psr_enable() while psr version is set to DC_PSR_VERSION_UNSUPPORTED. [How] Modify the condition to psr->psr_version == DC_PSR_VERSION_SU_1 Reviewed-by: Sun peng Li Signed-off-by: Tom Chung Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher (cherry picked from commit f765e7ce0417f8dc38479b4b495047c397c16902) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index dcc5d8ded662..a5fd19ee89dc 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -8929,7 +8929,7 @@ static void amdgpu_dm_enable_self_refresh(struct amdgpu_crtc *acrtc_attach, (current_ts - psr->psr_dirty_rects_change_timestamp_ns) > 500000000) { if (pr->replay_feature_enabled && !pr->replay_allow_active) amdgpu_dm_replay_enable(acrtc_state->stream, true); - if (psr->psr_version >= DC_PSR_VERSION_SU_1 && + if (psr->psr_version == DC_PSR_VERSION_SU_1 && !psr->psr_allow_active && !aconn->disallow_edp_enter_psr) amdgpu_dm_psr_enable(acrtc_state->stream); } -- 2.50.1