From 59228c6631f902fa826dc61321ab377ba8aadec5 Mon Sep 17 00:00:00 2001 From: Alex Sierra Date: Thu, 16 May 2024 17:06:48 -0500 Subject: [PATCH 01/16] drm/amdkfd: clear F8_MODE for gfx950 Default F8_MODE should be OCP format on gfx950. Signed-off-by: Alex Sierra Reviewed-by: Harish Kasiviswanathan Signed-off-by: Amber Lin Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c index 67137e674f1d..210bcc048f4c 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c @@ -64,8 +64,7 @@ static int update_qpd_v9(struct device_queue_manager *dqm, qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT; if (KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 3) || - KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 4) || - KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 5, 0)) + KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 4)) qpd->sh_mem_config |= (1 << SH_MEM_CONFIG__F8_MODE__SHIFT); -- 2.51.0 From 0d1a686b542aaabfcfd254bc7711624d4ec20df0 Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Thu, 6 Mar 2025 12:31:56 -0500 Subject: [PATCH 02/16] drm/amd/amdgpu: Add missing GC 11.5.0 register Adds register needed for debugging purposes. Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/include/asic_reg/gc/gc_11_5_0_offset.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/amd/include/asic_reg/gc/gc_11_5_0_offset.h b/drivers/gpu/drm/amd/include/asic_reg/gc/gc_11_5_0_offset.h index abdb8728156e..d6c02cf815be 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/gc/gc_11_5_0_offset.h +++ b/drivers/gpu/drm/amd/include/asic_reg/gc/gc_11_5_0_offset.h @@ -9478,6 +9478,8 @@ #define regRLC_GFX_IMU_CMD_BASE_IDX 1 #define regGFX_IMU_RLC_STATUS 0x4054 #define regGFX_IMU_RLC_STATUS_BASE_IDX 1 +#define regGFX_IMU_STATUS 0x4055 +#define regGFX_IMU_STATUS_BASE_IDX 1 #define regGFX_IMU_SOC_DATA 0x4059 #define regGFX_IMU_SOC_DATA_BASE_IDX 1 #define regGFX_IMU_SOC_ADDR 0x405a -- 2.51.0 From 68bfdc8dc0a1a7fdd9ab61e69907ae71a6fd3d91 Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Thu, 6 Mar 2025 12:51:24 -0600 Subject: [PATCH 03/16] drm/amd: Keep display off while going into S4 When userspace invokes S4 the flow is: 1) amdgpu_pmops_prepare() 2) amdgpu_pmops_freeze() 3) Create hibernation image 4) amdgpu_pmops_thaw() 5) Write out image to disk 6) Turn off system Then on resume amdgpu_pmops_restore() is called. This flow has a problem that because amdgpu_pmops_thaw() is called it will call amdgpu_device_resume() which will resume all of the GPU. This includes turning the display hardware back on and discovering connectors again. This is an unexpected experience for the display to turn back on. Adjust the flow so that during the S4 sequence display hardware is not turned back on. Reported-by: Xaver Hugl Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/2038 Cc: Muhammad Usama Anjum Tested-by: Muhammad Usama Anjum Acked-by: Alex Deucher Acked-by: Harry Wentland Link: https://lore.kernel.org/r/20250306185124.44780-1-mario.limonciello@amd.com Signed-off-by: Mario Limonciello Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 11 +++++++++-- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 5 +++++ 2 files changed, 14 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 8c1aa9feda53..ce08c428ba4c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -2563,7 +2563,6 @@ static int amdgpu_pmops_freeze(struct device *dev) int r; r = amdgpu_device_suspend(drm_dev, true); - adev->in_s4 = false; if (r) return r; @@ -2575,8 +2574,13 @@ static int amdgpu_pmops_freeze(struct device *dev) static int amdgpu_pmops_thaw(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); + struct amdgpu_device *adev = drm_to_adev(drm_dev); + int r; - return amdgpu_device_resume(drm_dev, true); + r = amdgpu_device_resume(drm_dev, true); + adev->in_s4 = false; + + return r; } static int amdgpu_pmops_poweroff(struct device *dev) @@ -2589,6 +2593,9 @@ static int amdgpu_pmops_poweroff(struct device *dev) static int amdgpu_pmops_restore(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); + struct amdgpu_device *adev = drm_to_adev(drm_dev); + + adev->in_s4 = false; return amdgpu_device_resume(drm_dev, true); } 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 7faf0d0cc23c..bb3ac201c4fb 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -3434,6 +3434,11 @@ static int dm_resume(struct amdgpu_ip_block *ip_block) return 0; } + + /* leave display off for S4 sequence */ + if (adev->in_s4) + return 0; + /* Recreate dc_state - DC invalidates it when setting power state to S3. */ dc_state_release(dm_state->context); dm_state->context = dc_state_create(dm->dc, NULL); -- 2.51.0 From 289e68503a4533b014f8447e2af28ad44c92c221 Mon Sep 17 00:00:00 2001 From: Harish Kasiviswanathan Date: Tue, 14 Jan 2025 14:07:24 -0500 Subject: [PATCH 04/16] drm/amdkfd: Set per-process flags only once cik/vi Set per-process static sh_mem config only once during process initialization. Move all static changes from update_qpd() which is called each time a queue is created to set_cache_memory_policy() which is called once during process initialization. set_cache_memory_policy() is currently defined only for cik and vi family. So this commit only focuses on these two. A separate commit will address other asics. Signed-off-by: Harish Kasiviswanathan Reviewed-by: Amber Lin Signed-off-by: Alex Deucher --- .../drm/amd/amdkfd/kfd_device_queue_manager.c | 39 +--------- .../amd/amdkfd/kfd_device_queue_manager_cik.c | 69 ++++++++++++------ .../amd/amdkfd/kfd_device_queue_manager_vi.c | 71 ++++++++++++------- 3 files changed, 94 insertions(+), 85 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c index f3f2fd6ee65c..d23c6a358d34 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c @@ -2591,14 +2591,6 @@ failed_try_destroy_debugged_queue: return retval; } -/* - * Low bits must be 0000/FFFF as required by HW, high bits must be 0 to - * stay in user mode. - */ -#define APE1_FIXED_BITS_MASK 0xFFFF80000000FFFFULL -/* APE1 limit is inclusive and 64K aligned. */ -#define APE1_LIMIT_ALIGNMENT 0xFFFF - static bool set_cache_memory_policy(struct device_queue_manager *dqm, struct qcm_process_device *qpd, enum cache_policy default_policy, @@ -2613,34 +2605,6 @@ static bool set_cache_memory_policy(struct device_queue_manager *dqm, dqm_lock(dqm); - if (alternate_aperture_size == 0) { - /* base > limit disables APE1 */ - qpd->sh_mem_ape1_base = 1; - qpd->sh_mem_ape1_limit = 0; - } else { - /* - * In FSA64, APE1_Base[63:0] = { 16{SH_MEM_APE1_BASE[31]}, - * SH_MEM_APE1_BASE[31:0], 0x0000 } - * APE1_Limit[63:0] = { 16{SH_MEM_APE1_LIMIT[31]}, - * SH_MEM_APE1_LIMIT[31:0], 0xFFFF } - * Verify that the base and size parameters can be - * represented in this format and convert them. - * Additionally restrict APE1 to user-mode addresses. - */ - - uint64_t base = (uintptr_t)alternate_aperture_base; - uint64_t limit = base + alternate_aperture_size - 1; - - if (limit <= base || (base & APE1_FIXED_BITS_MASK) != 0 || - (limit & APE1_FIXED_BITS_MASK) != APE1_LIMIT_ALIGNMENT) { - retval = false; - goto out; - } - - qpd->sh_mem_ape1_base = base >> 16; - qpd->sh_mem_ape1_limit = limit >> 16; - } - retval = dqm->asic_ops.set_cache_memory_policy( dqm, qpd, @@ -2649,6 +2613,9 @@ static bool set_cache_memory_policy(struct device_queue_manager *dqm, alternate_aperture_base, alternate_aperture_size); + if (retval) + goto out; + if ((dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) && (qpd->vmid != 0)) program_sh_mem_settings(dqm, qpd); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c index d4d95c7f2e5d..32bedef912b3 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c @@ -27,6 +27,14 @@ #include "oss/oss_2_4_sh_mask.h" #include "gca/gfx_7_2_sh_mask.h" +/* + * Low bits must be 0000/FFFF as required by HW, high bits must be 0 to + * stay in user mode. + */ +#define APE1_FIXED_BITS_MASK 0xFFFF80000000FFFFULL +/* APE1 limit is inclusive and 64K aligned. */ +#define APE1_LIMIT_ALIGNMENT 0xFFFF + static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm, struct qcm_process_device *qpd, enum cache_policy default_policy, @@ -84,6 +92,36 @@ static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm, { uint32_t default_mtype; uint32_t ape1_mtype; + unsigned int temp; + bool retval = true; + + if (alternate_aperture_size == 0) { + /* base > limit disables APE1 */ + qpd->sh_mem_ape1_base = 1; + qpd->sh_mem_ape1_limit = 0; + } else { + /* + * In FSA64, APE1_Base[63:0] = { 16{SH_MEM_APE1_BASE[31]}, + * SH_MEM_APE1_BASE[31:0], 0x0000 } + * APE1_Limit[63:0] = { 16{SH_MEM_APE1_LIMIT[31]}, + * SH_MEM_APE1_LIMIT[31:0], 0xFFFF } + * Verify that the base and size parameters can be + * represented in this format and convert them. + * Additionally restrict APE1 to user-mode addresses. + */ + + uint64_t base = (uintptr_t)alternate_aperture_base; + uint64_t limit = base + alternate_aperture_size - 1; + + if (limit <= base || (base & APE1_FIXED_BITS_MASK) != 0 || + (limit & APE1_FIXED_BITS_MASK) != APE1_LIMIT_ALIGNMENT) { + retval = false; + goto out; + } + + qpd->sh_mem_ape1_base = base >> 16; + qpd->sh_mem_ape1_limit = limit >> 16; + } default_mtype = (default_policy == cache_policy_coherent) ? MTYPE_NONCACHED : @@ -97,37 +135,22 @@ static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm, | ALIGNMENT_MODE(SH_MEM_ALIGNMENT_MODE_UNALIGNED) | DEFAULT_MTYPE(default_mtype) | APE1_MTYPE(ape1_mtype); - - return true; -} - -static int update_qpd_cik(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) -{ - struct kfd_process_device *pdd; - unsigned int temp; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - ALIGNMENT_MODE(SH_MEM_ALIGNMENT_MODE_UNALIGNED) | - DEFAULT_MTYPE(MTYPE_NONCACHED) | - APE1_MTYPE(MTYPE_NONCACHED); - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } - /* On dGPU we're always in GPUVM64 addressing mode with 64-bit * aperture addresses. */ - temp = get_sh_mem_bases_nybble_64(pdd); + temp = get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd)); qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp); pr_debug("is32bit process: %d sh_mem_bases nybble: 0x%X and register 0x%X\n", qpd->pqm->process->is_32bit_user_mode, temp, qpd->sh_mem_bases); +out: + return retval; +} + +static int update_qpd_cik(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c index b291ee0fab94..320518f41890 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c @@ -27,6 +27,14 @@ #include "gca/gfx_8_0_sh_mask.h" #include "oss/oss_3_0_sh_mask.h" +/* + * Low bits must be 0000/FFFF as required by HW, high bits must be 0 to + * stay in user mode. + */ +#define APE1_FIXED_BITS_MASK 0xFFFF80000000FFFFULL +/* APE1 limit is inclusive and 64K aligned. */ +#define APE1_LIMIT_ALIGNMENT 0xFFFF + static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm, struct qcm_process_device *qpd, enum cache_policy default_policy, @@ -85,6 +93,36 @@ static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm, { uint32_t default_mtype; uint32_t ape1_mtype; + unsigned int temp; + bool retval = true; + + if (alternate_aperture_size == 0) { + /* base > limit disables APE1 */ + qpd->sh_mem_ape1_base = 1; + qpd->sh_mem_ape1_limit = 0; + } else { + /* + * In FSA64, APE1_Base[63:0] = { 16{SH_MEM_APE1_BASE[31]}, + * SH_MEM_APE1_BASE[31:0], 0x0000 } + * APE1_Limit[63:0] = { 16{SH_MEM_APE1_LIMIT[31]}, + * SH_MEM_APE1_LIMIT[31:0], 0xFFFF } + * Verify that the base and size parameters can be + * represented in this format and convert them. + * Additionally restrict APE1 to user-mode addresses. + */ + + uint64_t base = (uintptr_t)alternate_aperture_base; + uint64_t limit = base + alternate_aperture_size - 1; + + if (limit <= base || (base & APE1_FIXED_BITS_MASK) != 0 || + (limit & APE1_FIXED_BITS_MASK) != APE1_LIMIT_ALIGNMENT) { + retval = false; + goto out; + } + + qpd->sh_mem_ape1_base = base >> 16; + qpd->sh_mem_ape1_limit = limit >> 16; + } default_mtype = (default_policy == cache_policy_coherent) ? MTYPE_UC : @@ -100,40 +138,21 @@ static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm, default_mtype << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT | ape1_mtype << SH_MEM_CONFIG__APE1_MTYPE__SHIFT; - return true; -} - -static int update_qpd_vi(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) -{ - struct kfd_process_device *pdd; - unsigned int temp; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - SH_MEM_ALIGNMENT_MODE_UNALIGNED << - SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT | - MTYPE_UC << - SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT | - MTYPE_UC << - SH_MEM_CONFIG__APE1_MTYPE__SHIFT; - - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } - /* On dGPU we're always in GPUVM64 addressing mode with 64-bit * aperture addresses. */ - temp = get_sh_mem_bases_nybble_64(pdd); + temp = get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd)); qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp); pr_debug("sh_mem_bases nybble: 0x%X and register 0x%X\n", temp, qpd->sh_mem_bases); +out: + return retval; +} +static int update_qpd_vi(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } -- 2.51.0 From 61972cd93af70738a6ad7f93e17cc7f68a01e182 Mon Sep 17 00:00:00 2001 From: Harish Kasiviswanathan Date: Tue, 14 Jan 2025 14:13:35 -0500 Subject: [PATCH 05/16] drm/amdkfd: Set per-process flags only once for gfx9/10/11/12 Define set_cache_memory_policy() for these asics and move all static changes from update_qpd() which is called each time a queue is created to set_cache_memory_policy() which is called once during process initialization Signed-off-by: Harish Kasiviswanathan Reviewed-by: Amber Lin Signed-off-by: Alex Deucher --- .../amd/amdkfd/kfd_device_queue_manager_v10.c | 41 +++++++++++-------- .../amd/amdkfd/kfd_device_queue_manager_v11.c | 41 +++++++++++-------- .../amd/amdkfd/kfd_device_queue_manager_v12.c | 41 +++++++++++-------- .../amd/amdkfd/kfd_device_queue_manager_v9.c | 36 +++++++++++++++- 4 files changed, 107 insertions(+), 52 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c index 245a90dfc2f6..b5f5f141353b 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c @@ -31,10 +31,17 @@ static int update_qpd_v10(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm_v10(struct device_queue_manager *dqm, struct queue *q, struct qcm_process_device *qpd); +static bool set_cache_memory_policy_v10(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size); void device_queue_manager_init_v10( struct device_queue_manager_asic_ops *asic_ops) { + asic_ops->set_cache_memory_policy = set_cache_memory_policy_v10; asic_ops->update_qpd = update_qpd_v10; asic_ops->init_sdma_vm = init_sdma_vm_v10; asic_ops->mqd_manager_init = mqd_manager_init_v10; @@ -49,27 +56,27 @@ static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd) private_base; } -static int update_qpd_v10(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) +static bool set_cache_memory_policy_v10(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size) { - struct kfd_process_device *pdd; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - (SH_MEM_ALIGNMENT_MODE_UNALIGNED << - SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | - (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } - - qpd->sh_mem_bases = compute_sh_mem_bases_64bit(pdd); + qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | + (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); + qpd->sh_mem_ape1_limit = 0; + qpd->sh_mem_ape1_base = 0; + qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd)); pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases); + return true; +} +static int update_qpd_v10(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c index 2e129da7acb4..f436878d0d62 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c @@ -30,10 +30,17 @@ static int update_qpd_v11(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm_v11(struct device_queue_manager *dqm, struct queue *q, struct qcm_process_device *qpd); +static bool set_cache_memory_policy_v11(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size); void device_queue_manager_init_v11( struct device_queue_manager_asic_ops *asic_ops) { + asic_ops->set_cache_memory_policy = set_cache_memory_policy_v11; asic_ops->update_qpd = update_qpd_v11; asic_ops->init_sdma_vm = init_sdma_vm_v11; asic_ops->mqd_manager_init = mqd_manager_init_v11; @@ -48,28 +55,28 @@ static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd) private_base; } -static int update_qpd_v11(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) +static bool set_cache_memory_policy_v11(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size) { - struct kfd_process_device *pdd; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - (SH_MEM_ALIGNMENT_MODE_UNALIGNED << - SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | - (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } + qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | + (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - qpd->sh_mem_bases = compute_sh_mem_bases_64bit(pdd); + qpd->sh_mem_ape1_limit = 0; + qpd->sh_mem_ape1_base = 0; + qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd)); pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases); + return true; +} +static int update_qpd_v11(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c index 4f3295b29dfb..62ca1c8fcbaf 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c @@ -30,10 +30,17 @@ static int update_qpd_v12(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm_v12(struct device_queue_manager *dqm, struct queue *q, struct qcm_process_device *qpd); +static bool set_cache_memory_policy_v12(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size); void device_queue_manager_init_v12( struct device_queue_manager_asic_ops *asic_ops) { + asic_ops->set_cache_memory_policy = set_cache_memory_policy_v12; asic_ops->update_qpd = update_qpd_v12; asic_ops->init_sdma_vm = init_sdma_vm_v12; asic_ops->mqd_manager_init = mqd_manager_init_v12; @@ -48,28 +55,28 @@ static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd) private_base; } -static int update_qpd_v12(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) +static bool set_cache_memory_policy_v12(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size) { - struct kfd_process_device *pdd; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - (SH_MEM_ALIGNMENT_MODE_UNALIGNED << - SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | - (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } + qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | + (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - qpd->sh_mem_bases = compute_sh_mem_bases_64bit(pdd); + qpd->sh_mem_ape1_limit = 0; + qpd->sh_mem_ape1_base = 0; + qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd)); pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases); + return true; +} +static int update_qpd_v12(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c index 210bcc048f4c..3264509408bc 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c @@ -30,10 +30,17 @@ static int update_qpd_v9(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm_v9(struct device_queue_manager *dqm, struct queue *q, struct qcm_process_device *qpd); +static bool set_cache_memory_policy_v9(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size); void device_queue_manager_init_v9( struct device_queue_manager_asic_ops *asic_ops) { + asic_ops->set_cache_memory_policy = set_cache_memory_policy_v9; asic_ops->update_qpd = update_qpd_v9; asic_ops->init_sdma_vm = init_sdma_vm_v9; asic_ops->mqd_manager_init = mqd_manager_init_v9; @@ -48,10 +55,37 @@ static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd) private_base; } +static bool set_cache_memory_policy_v9(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size) +{ + qpd->sh_mem_config = SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT; + + if (dqm->dev->kfd->noretry) + qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT; + + if (KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 3) || + KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 4) || + KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 5, 0)) + qpd->sh_mem_config |= (1 << SH_MEM_CONFIG__F8_MODE__SHIFT); + + qpd->sh_mem_ape1_limit = 0; + qpd->sh_mem_ape1_base = 0; + qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd)); + + pr_debug("sh_mem_bases 0x%X sh_mem_config 0x%X\n", qpd->sh_mem_bases, + qpd->sh_mem_config); + return true; +} + static int update_qpd_v9(struct device_queue_manager *dqm, struct qcm_process_device *qpd) { - struct kfd_process_device *pdd; + struct kfd_process_device *pdd = qpd_to_pdd(qpd); pdd = qpd_to_pdd(qpd); -- 2.51.0 From cf6d949a409e09539477d32dbe7c954e4852e744 Mon Sep 17 00:00:00 2001 From: Harish Kasiviswanathan Date: Tue, 14 Jan 2025 16:02:21 -0500 Subject: [PATCH 06/16] drm/amdkfd: Add support for more per-process flag Add support for more per-process flags starting with option to configure MFMA precision for gfx 9.5 v2: Change flag name to KFD_PROC_FLAG_MFMA_HIGH_PRECISION Remove unused else condition v3: Bump the KFD API version v4: Missed SH_MEM_CONFIG__PRECISION_MODE__SHIFT define. Added it. Signed-off-by: Harish Kasiviswanathan Reviewed-by: Felix Kuehling Reviewed-by: Amber Lin Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdkfd/kfd_chardev.c | 3 ++- drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 6 ++++-- drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h | 6 ++++-- .../gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c | 6 ++++-- .../gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c | 6 ++++-- .../gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c | 6 ++++-- .../gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c | 6 ++++-- .../gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c | 11 +++++++++-- .../gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c | 6 ++++-- .../drm/amd/include/asic_reg/gc/gc_9_4_3_sh_mask.h | 2 ++ include/uapi/linux/kfd_ioctl.h | 8 ++++++-- 11 files changed, 47 insertions(+), 19 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index 8c2e92378b49..1e9dd00620bf 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -606,7 +606,8 @@ static int kfd_ioctl_set_memory_policy(struct file *filep, default_policy, alternate_policy, (void __user *)args->alternate_aperture_base, - args->alternate_aperture_size)) + args->alternate_aperture_size, + args->misc_process_flag)) err = -EINVAL; out: diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c index d23c6a358d34..2afcc1b4856a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c @@ -2596,7 +2596,8 @@ static bool set_cache_memory_policy(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size) + uint64_t alternate_aperture_size, + u32 misc_process_properties) { bool retval = true; @@ -2611,7 +2612,8 @@ static bool set_cache_memory_policy(struct device_queue_manager *dqm, default_policy, alternate_policy, alternate_aperture_base, - alternate_aperture_size); + alternate_aperture_size, + misc_process_properties); if (retval) goto out; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h index 7146e227e2c1..122eb745e9c4 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h @@ -174,7 +174,8 @@ struct device_queue_manager_ops { enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size); + uint64_t alternate_aperture_size, + u32 misc_process_properties); int (*process_termination)(struct device_queue_manager *dqm, struct qcm_process_device *qpd); @@ -210,7 +211,8 @@ struct device_queue_manager_asic_ops { enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size); + uint64_t alternate_aperture_size, + u32 misc_process_properties); void (*init_sdma_vm)(struct device_queue_manager *dqm, struct queue *q, struct qcm_process_device *qpd); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c index 32bedef912b3..0508ef5a41d7 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c @@ -40,7 +40,8 @@ static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size); + uint64_t alternate_aperture_size, + u32 misc_process_properties); static int update_qpd_cik(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm(struct device_queue_manager *dqm, @@ -88,7 +89,8 @@ static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size) + uint64_t alternate_aperture_size, + u32 misc_process_properties) { uint32_t default_mtype; uint32_t ape1_mtype; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c index b5f5f141353b..ba6e3d747ccd 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c @@ -36,7 +36,8 @@ static bool set_cache_memory_policy_v10(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size); + uint64_t alternate_aperture_size, + u32 misc_process_properties); void device_queue_manager_init_v10( struct device_queue_manager_asic_ops *asic_ops) @@ -61,7 +62,8 @@ static bool set_cache_memory_policy_v10(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size) + uint64_t alternate_aperture_size, + u32 misc_process_properties) { qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c index f436878d0d62..8b447d04558f 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c @@ -35,7 +35,8 @@ static bool set_cache_memory_policy_v11(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size); + uint64_t alternate_aperture_size, + u32 misc_process_properties); void device_queue_manager_init_v11( struct device_queue_manager_asic_ops *asic_ops) @@ -60,7 +61,8 @@ static bool set_cache_memory_policy_v11(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size) + uint64_t alternate_aperture_size, + u32 misc_process_properties) { qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c index 62ca1c8fcbaf..3550da3a46f9 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c @@ -35,7 +35,8 @@ static bool set_cache_memory_policy_v12(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size); + uint64_t alternate_aperture_size, + u32 misc_process_properties); void device_queue_manager_init_v12( struct device_queue_manager_asic_ops *asic_ops) @@ -60,7 +61,8 @@ static bool set_cache_memory_policy_v12(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size) + uint64_t alternate_aperture_size, + u32 misc_process_properties) { qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c index 3264509408bc..d794c8172b40 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c @@ -35,7 +35,8 @@ static bool set_cache_memory_policy_v9(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size); + uint64_t alternate_aperture_size, + u32 misc_process_properties); void device_queue_manager_init_v9( struct device_queue_manager_asic_ops *asic_ops) @@ -60,7 +61,8 @@ static bool set_cache_memory_policy_v9(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size) + uint64_t alternate_aperture_size, + u32 misc_process_properties) { qpd->sh_mem_config = SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT; @@ -73,6 +75,11 @@ static bool set_cache_memory_policy_v9(struct device_queue_manager *dqm, KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 5, 0)) qpd->sh_mem_config |= (1 << SH_MEM_CONFIG__F8_MODE__SHIFT); + if (KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 5, 0)) { + if (misc_process_properties & KFD_PROC_FLAG_MFMA_HIGH_PRECISION) + qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__PRECISION_MODE__SHIFT; + } + qpd->sh_mem_ape1_limit = 0; qpd->sh_mem_ape1_base = 0; qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd)); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c index 320518f41890..dad83356e976 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c @@ -40,7 +40,8 @@ static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size); + uint64_t alternate_aperture_size, + u32 misc_process_properties); static int update_qpd_vi(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm(struct device_queue_manager *dqm, @@ -89,7 +90,8 @@ static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm, enum cache_policy default_policy, enum cache_policy alternate_policy, void __user *alternate_aperture_base, - uint64_t alternate_aperture_size) + uint64_t alternate_aperture_size, + u32 misc_process_properties) { uint32_t default_mtype; uint32_t ape1_mtype; diff --git a/drivers/gpu/drm/amd/include/asic_reg/gc/gc_9_4_3_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/gc/gc_9_4_3_sh_mask.h index 2bd9f3f1026f..0122a21c50cf 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/gc/gc_9_4_3_sh_mask.h +++ b/drivers/gpu/drm/amd/include/asic_reg/gc/gc_9_4_3_sh_mask.h @@ -2261,11 +2261,13 @@ #define SH_MEM_CONFIG__ADDRESS_MODE__SHIFT 0x0 #define SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT 0x3 #define SH_MEM_CONFIG__F8_MODE__SHIFT 0x8 +#define SH_MEM_CONFIG__PRECISION_MODE__SHIFT 0x9 #define SH_MEM_CONFIG__RETRY_DISABLE__SHIFT 0xc #define SH_MEM_CONFIG__PRIVATE_NV__SHIFT 0xd #define SH_MEM_CONFIG__ADDRESS_MODE_MASK 0x00000001L #define SH_MEM_CONFIG__ALIGNMENT_MODE_MASK 0x00000018L #define SH_MEM_CONFIG__F8_MODE_MASK 0x00000100L +#define SH_MEM_CONFIG__PRECISION_MODE_MASK 0x00000200L #define SH_MEM_CONFIG__RETRY_DISABLE_MASK 0x00001000L #define SH_MEM_CONFIG__PRIVATE_NV_MASK 0x00002000L //SP_MFMA_PORTD_RD_CONFIG diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux/kfd_ioctl.h index b0160b09987c..1e59344c5673 100644 --- a/include/uapi/linux/kfd_ioctl.h +++ b/include/uapi/linux/kfd_ioctl.h @@ -43,9 +43,10 @@ * - 1.15 - Enable managing mappings in compute VMs with GEM_VA ioctl * - 1.16 - Add contiguous VRAM allocation flag * - 1.17 - Add SDMA queue creation with target SDMA engine ID + * - 1.18 - Rename pad in set_memory_policy_args to misc_process_flag */ #define KFD_IOCTL_MAJOR_VERSION 1 -#define KFD_IOCTL_MINOR_VERSION 17 +#define KFD_IOCTL_MINOR_VERSION 18 struct kfd_ioctl_get_version_args { __u32 major_version; /* from KFD */ @@ -150,6 +151,9 @@ struct kfd_dbg_device_info_entry { #define KFD_IOC_CACHE_POLICY_COHERENT 0 #define KFD_IOC_CACHE_POLICY_NONCOHERENT 1 +/* Misc. per process flags */ +#define KFD_PROC_FLAG_MFMA_HIGH_PRECISION (1 << 0) + struct kfd_ioctl_set_memory_policy_args { __u64 alternate_aperture_base; /* to KFD */ __u64 alternate_aperture_size; /* to KFD */ @@ -157,7 +161,7 @@ struct kfd_ioctl_set_memory_policy_args { __u32 gpu_id; /* to KFD */ __u32 default_policy; /* to KFD */ __u32 alternate_policy; /* to KFD */ - __u32 pad; + __u32 misc_process_flag; /* to KFD */ }; /* -- 2.51.0 From 3855f1d925d4f0971e35ec8c44f62862f78500fd Mon Sep 17 00:00:00 2001 From: =?utf8?q?Marek=20Ol=C5=A1=C3=A1k?= Date: Fri, 7 Mar 2025 09:57:45 -0500 Subject: [PATCH 07/16] drm/amd/display: allow 256B DCC max compressed block sizes on gfx12 MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit The hw supports it. Signed-off-by: Marek Olšák Acked-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 3 ++- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c | 2 +- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index ce08c428ba4c..653f2bc77530 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -122,9 +122,10 @@ * - 3.60.0 - Add AMDGPU_TILING_GFX12_DCC_WRITE_COMPRESS_DISABLE (Vulkan requirement) * - 3.61.0 - Contains fix for RV/PCO compute queues * - 3.62.0 - Add AMDGPU_IDS_FLAGS_MODE_PF, AMDGPU_IDS_FLAGS_MODE_VF & AMDGPU_IDS_FLAGS_MODE_PT + * - 3.63.0 - GFX12 display DCC supports 256B max compressed block size */ #define KMS_DRIVER_MAJOR 3 -#define KMS_DRIVER_MINOR 62 +#define KMS_DRIVER_MINOR 63 #define KMS_DRIVER_PATCHLEVEL 0 /* diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c index 0090e08d5057..3e0f45f1711c 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c @@ -700,7 +700,7 @@ static void amdgpu_dm_plane_add_gfx12_modifiers(struct amdgpu_device *adev, uint64_t mod_4k = ver | AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX12_4K_2D); uint64_t mod_256b = ver | AMD_FMT_MOD_SET(TILE, AMD_FMT_MOD_TILE_GFX12_256B_2D); uint64_t dcc = ver | AMD_FMT_MOD_SET(DCC, 1); - uint8_t max_comp_block[] = {1, 0}; + uint8_t max_comp_block[] = {2, 1, 0}; uint64_t max_comp_block_mod[ARRAY_SIZE(max_comp_block)] = {0}; uint8_t i = 0, j = 0; uint64_t gfx12_modifiers[] = {mod_256k, mod_64k, mod_4k, mod_256b, DRM_FORMAT_MOD_LINEAR}; -- 2.51.0 From 2c01befe4a2707302eb1c97b955d94d66fac7b6f Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 4 Mar 2025 17:57:43 -0500 Subject: [PATCH 08/16] drm/amdgpu/vcn: fix idle work handler for VCN 2.5 VCN 2.5 uses the PG callback to enable VCN DPM which is a global state. As such, we need to make sure all instances are in the same state. v2: switch to a ref count (Lijo) v3: switch to its own idle work handler v4: fix logic in DPG handling Fixes: 4ce4fe27205c ("drm/amdgpu/vcn: use per instance callbacks for idle work handler") Reviewed-by: Boyuan Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c | 120 +++++++++++++++++++++++++- 1 file changed, 116 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index dff1a8859036..ff03436698a4 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -107,6 +107,115 @@ static int amdgpu_ih_clientid_vcns[] = { SOC15_IH_CLIENTID_VCN1 }; +static void vcn_v2_5_idle_work_handler(struct work_struct *work) +{ + struct amdgpu_vcn_inst *vcn_inst = + container_of(work, struct amdgpu_vcn_inst, idle_work.work); + struct amdgpu_device *adev = vcn_inst->adev; + unsigned int fences = 0, fence[AMDGPU_MAX_VCN_INSTANCES] = {0}; + unsigned int i, j; + int r = 0; + + for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + struct amdgpu_vcn_inst *v = &adev->vcn.inst[i]; + + if (adev->vcn.harvest_config & (1 << i)) + continue; + + for (j = 0; j < v->num_enc_rings; ++j) + fence[i] += amdgpu_fence_count_emitted(&v->ring_enc[j]); + + /* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */ + if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG && + !v->using_unified_queue) { + struct dpg_pause_state new_state; + + if (fence[i] || + unlikely(atomic_read(&v->dpg_enc_submission_cnt))) + new_state.fw_based = VCN_DPG_STATE__PAUSE; + else + new_state.fw_based = VCN_DPG_STATE__UNPAUSE; + + v->pause_dpg_mode(v, &new_state); + } + + fence[i] += amdgpu_fence_count_emitted(&v->ring_dec); + fences += fence[i]; + + } + + if (!fences && !atomic_read(&adev->vcn.inst[0].total_submission_cnt)) { + amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN, + AMD_PG_STATE_GATE); + r = amdgpu_dpm_switch_power_profile(adev, PP_SMC_POWER_PROFILE_VIDEO, + false); + if (r) + dev_warn(adev->dev, "(%d) failed to disable video power profile mode\n", r); + } else { + schedule_delayed_work(&adev->vcn.inst[0].idle_work, VCN_IDLE_TIMEOUT); + } +} + +static void vcn_v2_5_ring_begin_use(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + struct amdgpu_vcn_inst *v = &adev->vcn.inst[ring->me]; + int r = 0; + + atomic_inc(&adev->vcn.inst[0].total_submission_cnt); + + if (!cancel_delayed_work_sync(&adev->vcn.inst[0].idle_work)) { + r = amdgpu_dpm_switch_power_profile(adev, PP_SMC_POWER_PROFILE_VIDEO, + true); + if (r) + dev_warn(adev->dev, "(%d) failed to switch to video power profile mode\n", r); + } + + mutex_lock(&adev->vcn.inst[0].vcn_pg_lock); + amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN, + AMD_PG_STATE_UNGATE); + + /* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */ + if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG && + !v->using_unified_queue) { + struct dpg_pause_state new_state; + + if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC) { + atomic_inc(&v->dpg_enc_submission_cnt); + new_state.fw_based = VCN_DPG_STATE__PAUSE; + } else { + unsigned int fences = 0; + unsigned int i; + + for (i = 0; i < v->num_enc_rings; ++i) + fences += amdgpu_fence_count_emitted(&v->ring_enc[i]); + + if (fences || atomic_read(&v->dpg_enc_submission_cnt)) + new_state.fw_based = VCN_DPG_STATE__PAUSE; + else + new_state.fw_based = VCN_DPG_STATE__UNPAUSE; + } + v->pause_dpg_mode(v, &new_state); + } + mutex_unlock(&adev->vcn.inst[0].vcn_pg_lock); +} + +static void vcn_v2_5_ring_end_use(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + + /* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */ + if (ring->adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG && + ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC && + !adev->vcn.inst[ring->me].using_unified_queue) + atomic_dec(&adev->vcn.inst[ring->me].dpg_enc_submission_cnt); + + atomic_dec(&adev->vcn.inst[0].total_submission_cnt); + + schedule_delayed_work(&adev->vcn.inst[0].idle_work, + VCN_IDLE_TIMEOUT); +} + /** * vcn_v2_5_early_init - set function pointers and load microcode * @@ -201,6 +310,9 @@ static int vcn_v2_5_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + /* Override the work func */ + adev->vcn.inst[j].idle_work.work.func = vcn_v2_5_idle_work_handler; + amdgpu_vcn_setup_ucode(adev, j); r = amdgpu_vcn_resume(adev, j); @@ -1661,8 +1773,8 @@ static const struct amdgpu_ring_funcs vcn_v2_5_dec_ring_vm_funcs = { .insert_start = vcn_v2_0_dec_ring_insert_start, .insert_end = vcn_v2_0_dec_ring_insert_end, .pad_ib = amdgpu_ring_generic_pad_ib, - .begin_use = amdgpu_vcn_ring_begin_use, - .end_use = amdgpu_vcn_ring_end_use, + .begin_use = vcn_v2_5_ring_begin_use, + .end_use = vcn_v2_5_ring_end_use, .emit_wreg = vcn_v2_0_dec_ring_emit_wreg, .emit_reg_wait = vcn_v2_0_dec_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, @@ -1759,8 +1871,8 @@ static const struct amdgpu_ring_funcs vcn_v2_5_enc_ring_vm_funcs = { .insert_nop = amdgpu_ring_insert_nop, .insert_end = vcn_v2_0_enc_ring_insert_end, .pad_ib = amdgpu_ring_generic_pad_ib, - .begin_use = amdgpu_vcn_ring_begin_use, - .end_use = amdgpu_vcn_ring_end_use, + .begin_use = vcn_v2_5_ring_begin_use, + .end_use = vcn_v2_5_ring_end_use, .emit_wreg = vcn_v2_0_enc_ring_emit_wreg, .emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, -- 2.51.0 From ed962f8d0603da15c26f1c9ce60cba42607a2768 Mon Sep 17 00:00:00 2001 From: Harish Kasiviswanathan Date: Tue, 11 Feb 2025 12:56:01 -0500 Subject: [PATCH 09/16] drm/amdkfd: Add pm_config_dequeue_wait_counts API Update pm_update_grace_period() to more cleaner pm_config_dequeue_wait_counts(). Previously, grace_period variable was overloaded as a variable and a macro, making it inflexible to configure additional dequeue wait times. pm_config_dequeue_wait_counts() now takes in a cmd / variable. This allows flexibility to update different dequeue wait times. Signed-off-by: Harish Kasiviswanathan Reviewed-by: Jonathan Kim Signed-off-by: Alex Deucher --- .../drm/amd/amdkfd/kfd_device_queue_manager.c | 45 ++++++-------- .../drm/amd/amdkfd/kfd_device_queue_manager.h | 11 +++- .../gpu/drm/amd/amdkfd/kfd_packet_manager.c | 26 +++++++- .../drm/amd/amdkfd/kfd_packet_manager_v9.c | 59 ++++++++++++++----- .../drm/amd/amdkfd/kfd_packet_manager_vi.c | 4 +- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 28 +++++++-- 6 files changed, 123 insertions(+), 50 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c index 2afcc1b4856a..885e0e9cf21b 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c @@ -43,6 +43,8 @@ /* Size of the per-pipe EOP queue */ #define CIK_HPD_EOP_BYTES_LOG2 11 #define CIK_HPD_EOP_BYTES (1U << CIK_HPD_EOP_BYTES_LOG2) +/* See unmap_queues_cpsch() */ +#define USE_DEFAULT_GRACE_PERIOD 0xffffffff static int set_pasid_vmid_mapping(struct device_queue_manager *dqm, u32 pasid, unsigned int vmid); @@ -1746,10 +1748,7 @@ static int initialize_cpsch(struct device_queue_manager *dqm) init_sdma_bitmaps(dqm); - if (dqm->dev->kfd2kgd->get_iq_wait_times) - dqm->dev->kfd2kgd->get_iq_wait_times(dqm->dev->adev, - &dqm->wait_times, - ffs(dqm->dev->xcc_mask) - 1); + update_dqm_wait_times(dqm); return 0; } @@ -1845,25 +1844,11 @@ static int start_cpsch(struct device_queue_manager *dqm) /* clear hang status when driver try to start the hw scheduler */ dqm->sched_running = true; - if (!dqm->dev->kfd->shared_resources.enable_mes) + if (!dqm->dev->kfd->shared_resources.enable_mes) { + if (pm_config_dequeue_wait_counts(&dqm->packet_mgr, + KFD_DEQUEUE_WAIT_INIT, 0 /* unused */)) + dev_err(dev, "Setting optimized dequeue wait failed. Using default values\n"); execute_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0, USE_DEFAULT_GRACE_PERIOD); - - /* Set CWSR grace period to 1x1000 cycle for GFX9.4.3 APU */ - if (amdgpu_emu_mode == 0 && dqm->dev->adev->gmc.is_app_apu && - (KFD_GC_VERSION(dqm->dev) == IP_VERSION(9, 4, 3))) { - uint32_t reg_offset = 0; - uint32_t grace_period = 1; - - retval = pm_update_grace_period(&dqm->packet_mgr, - grace_period); - if (retval) - dev_err(dev, "Setting grace timeout failed\n"); - else if (dqm->dev->kfd2kgd->build_grace_period_packet_info) - /* Update dqm->wait_times maintained in software */ - dqm->dev->kfd2kgd->build_grace_period_packet_info( - dqm->dev->adev, dqm->wait_times, - grace_period, ®_offset, - &dqm->wait_times); } /* setup per-queue reset detection buffer */ @@ -2359,7 +2344,14 @@ static int reset_queues_on_hws_hang(struct device_queue_manager *dqm, bool is_sd return is_sdma ? reset_hung_queues_sdma(dqm) : reset_hung_queues(dqm); } -/* dqm->lock mutex has to be locked before calling this function */ +/* dqm->lock mutex has to be locked before calling this function + * + * @grace_period: If USE_DEFAULT_GRACE_PERIOD then default wait time + * for context switch latency. Lower values are used by debugger + * since context switching are triggered at high frequency. + * This is configured by setting CP_IQ_WAIT_TIME2.SCH_WAVE + * + */ static int unmap_queues_cpsch(struct device_queue_manager *dqm, enum kfd_unmap_queues_filter filter, uint32_t filter_param, @@ -2378,7 +2370,8 @@ static int unmap_queues_cpsch(struct device_queue_manager *dqm, return -EIO; if (grace_period != USE_DEFAULT_GRACE_PERIOD) { - retval = pm_update_grace_period(&dqm->packet_mgr, grace_period); + retval = pm_config_dequeue_wait_counts(&dqm->packet_mgr, + KFD_DEQUEUE_WAIT_SET_SCH_WAVE, grace_period); if (retval) goto out; } @@ -2419,8 +2412,8 @@ static int unmap_queues_cpsch(struct device_queue_manager *dqm, /* We need to reset the grace period value for this device */ if (grace_period != USE_DEFAULT_GRACE_PERIOD) { - if (pm_update_grace_period(&dqm->packet_mgr, - USE_DEFAULT_GRACE_PERIOD)) + if (pm_config_dequeue_wait_counts(&dqm->packet_mgr, + KFD_DEQUEUE_WAIT_RESET, 0 /* unused */)) dev_err(dev, "Failed to reset grace period\n"); } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h index 122eb745e9c4..74a61b5b2f0b 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h @@ -37,7 +37,6 @@ #define KFD_MES_PROCESS_QUANTUM 100000 #define KFD_MES_GANG_QUANTUM 10000 -#define USE_DEFAULT_GRACE_PERIOD 0xffffffff struct device_process_node { struct qcm_process_device *qpd; @@ -360,4 +359,14 @@ static inline int read_sdma_queue_counter(uint64_t __user *q_rptr, uint64_t *val /* SDMA activity counter is stored at queue's RPTR + 0x8 location. */ return get_user(*val, q_rptr + 1); } + +static inline void update_dqm_wait_times(struct device_queue_manager *dqm) +{ + if (dqm->dev->kfd2kgd->get_iq_wait_times) + dqm->dev->kfd2kgd->get_iq_wait_times(dqm->dev->adev, + &dqm->wait_times, + ffs(dqm->dev->xcc_mask) - 1); +} + + #endif /* KFD_DEVICE_QUEUE_MANAGER_H_ */ diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c index 4984b41cd372..3f574d82b5fc 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c @@ -396,14 +396,29 @@ out: return retval; } -int pm_update_grace_period(struct packet_manager *pm, uint32_t grace_period) +/* pm_config_dequeue_wait_counts: Configure dequeue timer Wait Counts + * by writing to CP_IQ_WAIT_TIME2 registers. + * + * @cmd: See emum kfd_config_dequeue_wait_counts_cmd definition + * @value: Depends on the cmd. This parameter is unused for + * KFD_DEQUEUE_WAIT_INIT and KFD_DEQUEUE_WAIT_RESET. For + * KFD_DEQUEUE_WAIT_SET_SCH_WAVE it holds value to be set + * + */ +int pm_config_dequeue_wait_counts(struct packet_manager *pm, + enum kfd_config_dequeue_wait_counts_cmd cmd, + uint32_t value) { struct kfd_node *node = pm->dqm->dev; struct device *dev = node->adev->dev; int retval = 0; uint32_t *buffer, size; - size = pm->pmf->set_grace_period_size; + if (!pm->pmf->config_dequeue_wait_counts || + !pm->pmf->config_dequeue_wait_counts_size) + return 0; + + size = pm->pmf->config_dequeue_wait_counts_size; mutex_lock(&pm->lock); @@ -419,13 +434,18 @@ int pm_update_grace_period(struct packet_manager *pm, uint32_t grace_period) goto out; } - retval = pm->pmf->set_grace_period(pm, buffer, grace_period); + retval = pm->pmf->config_dequeue_wait_counts(pm, buffer, + cmd, value); if (!retval) retval = kq_submit_packet(pm->priv_queue); else kq_rollback_packet(pm->priv_queue); } + /* If default value is modified, cache that value in dqm->wait_times */ + if (!retval && cmd == KFD_DEQUEUE_WAIT_INIT) + update_dqm_wait_times(pm->dqm); + out: mutex_unlock(&pm->lock); return retval; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c index d56525201155..b9c611b249e6 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c @@ -297,23 +297,54 @@ static int pm_map_queues_v9(struct packet_manager *pm, uint32_t *buffer, return 0; } -static int pm_set_grace_period_v9(struct packet_manager *pm, +static inline void pm_build_dequeue_wait_counts_packet_info(struct packet_manager *pm, + uint32_t sch_value, uint32_t *reg_offset, + uint32_t *reg_data) +{ + pm->dqm->dev->kfd2kgd->build_grace_period_packet_info( + pm->dqm->dev->adev, + pm->dqm->wait_times, + sch_value, + reg_offset, + reg_data); +} + +static int pm_config_dequeue_wait_counts_v9(struct packet_manager *pm, uint32_t *buffer, - uint32_t grace_period) + enum kfd_config_dequeue_wait_counts_cmd cmd, + uint32_t value) { struct pm4_mec_write_data_mmio *packet; uint32_t reg_offset = 0; uint32_t reg_data = 0; - pm->dqm->dev->kfd2kgd->build_grace_period_packet_info( - pm->dqm->dev->adev, - pm->dqm->wait_times, - grace_period, - ®_offset, - ®_data); - - if (grace_period == USE_DEFAULT_GRACE_PERIOD) + switch (cmd) { + case KFD_DEQUEUE_WAIT_INIT: + /* Set CWSR grace period to 1x1000 cycle for GFX9.4.3 APU */ + if (amdgpu_emu_mode == 0 && pm->dqm->dev->adev->gmc.is_app_apu && + (KFD_GC_VERSION(pm->dqm->dev) == IP_VERSION(9, 4, 3))) + pm_build_dequeue_wait_counts_packet_info(pm, 1, ®_offset, ®_data); + else + return 0; + break; + case KFD_DEQUEUE_WAIT_RESET: + /* function called only to get reg_offset */ + pm_build_dequeue_wait_counts_packet_info(pm, 0, ®_offset, ®_data); reg_data = pm->dqm->wait_times; + break; + + case KFD_DEQUEUE_WAIT_SET_SCH_WAVE: + /* The CP cannot handle value 0 and it will result in + * an infinite grace period being set so set to 1 to prevent this. + */ + if (!value) + value = 1; + pm_build_dequeue_wait_counts_packet_info(pm, value, ®_offset, ®_data); + break; + default: + pr_err("Invalid dequeue wait cmd\n"); + return -EINVAL; + } packet = (struct pm4_mec_write_data_mmio *)buffer; memset(buffer, 0, sizeof(struct pm4_mec_write_data_mmio)); @@ -415,7 +446,7 @@ const struct packet_manager_funcs kfd_v9_pm_funcs = { .set_resources = pm_set_resources_v9, .map_queues = pm_map_queues_v9, .unmap_queues = pm_unmap_queues_v9, - .set_grace_period = pm_set_grace_period_v9, + .config_dequeue_wait_counts = pm_config_dequeue_wait_counts_v9, .query_status = pm_query_status_v9, .release_mem = NULL, .map_process_size = sizeof(struct pm4_mes_map_process), @@ -423,7 +454,7 @@ const struct packet_manager_funcs kfd_v9_pm_funcs = { .set_resources_size = sizeof(struct pm4_mes_set_resources), .map_queues_size = sizeof(struct pm4_mes_map_queues), .unmap_queues_size = sizeof(struct pm4_mes_unmap_queues), - .set_grace_period_size = sizeof(struct pm4_mec_write_data_mmio), + .config_dequeue_wait_counts_size = sizeof(struct pm4_mec_write_data_mmio), .query_status_size = sizeof(struct pm4_mes_query_status), .release_mem_size = 0, }; @@ -434,7 +465,7 @@ const struct packet_manager_funcs kfd_aldebaran_pm_funcs = { .set_resources = pm_set_resources_v9, .map_queues = pm_map_queues_v9, .unmap_queues = pm_unmap_queues_v9, - .set_grace_period = pm_set_grace_period_v9, + .config_dequeue_wait_counts = pm_config_dequeue_wait_counts_v9, .query_status = pm_query_status_v9, .release_mem = NULL, .map_process_size = sizeof(struct pm4_mes_map_process_aldebaran), @@ -442,7 +473,7 @@ const struct packet_manager_funcs kfd_aldebaran_pm_funcs = { .set_resources_size = sizeof(struct pm4_mes_set_resources), .map_queues_size = sizeof(struct pm4_mes_map_queues), .unmap_queues_size = sizeof(struct pm4_mes_unmap_queues), - .set_grace_period_size = sizeof(struct pm4_mec_write_data_mmio), + .config_dequeue_wait_counts_size = sizeof(struct pm4_mec_write_data_mmio), .query_status_size = sizeof(struct pm4_mes_query_status), .release_mem_size = 0, }; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c index 347c86e1c378..a1de5d7e173a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c @@ -304,7 +304,7 @@ const struct packet_manager_funcs kfd_vi_pm_funcs = { .set_resources = pm_set_resources_vi, .map_queues = pm_map_queues_vi, .unmap_queues = pm_unmap_queues_vi, - .set_grace_period = NULL, + .config_dequeue_wait_counts = NULL, .query_status = pm_query_status_vi, .release_mem = pm_release_mem_vi, .map_process_size = sizeof(struct pm4_mes_map_process), @@ -312,7 +312,7 @@ const struct packet_manager_funcs kfd_vi_pm_funcs = { .set_resources_size = sizeof(struct pm4_mes_set_resources), .map_queues_size = sizeof(struct pm4_mes_map_queues), .unmap_queues_size = sizeof(struct pm4_mes_unmap_queues), - .set_grace_period_size = 0, + .config_dequeue_wait_counts_size = 0, .query_status_size = sizeof(struct pm4_mes_query_status), .release_mem_size = sizeof(struct pm4_mec_release_mem) }; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index bb09c873a9a5..f6aedf69c644 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -1389,6 +1389,24 @@ int pqm_get_queue_checkpoint_info(struct process_queue_manager *pqm, #define KFD_FENCE_COMPLETED (100) #define KFD_FENCE_INIT (10) +/** + * enum kfd_config_dequeue_wait_counts_cmd - Command for configuring + * dequeue wait counts. + * + * @KFD_DEQUEUE_WAIT_INIT: Set optimized dequeue wait counts for a + * certain ASICs. For these ASICs, this is default value used by RESET + * @KFD_DEQUEUE_WAIT_RESET: Reset dequeue wait counts to the optimized value + * for certain ASICs. For others set it to default hardware reset value + * @KFD_DEQUEUE_WAIT_SET_SCH_WAVE: Set context switch latency wait + * + */ +enum kfd_config_dequeue_wait_counts_cmd { + KFD_DEQUEUE_WAIT_INIT = 1, + KFD_DEQUEUE_WAIT_RESET = 2, + KFD_DEQUEUE_WAIT_SET_SCH_WAVE = 3 +}; + + struct packet_manager { struct device_queue_manager *dqm; struct kernel_queue *priv_queue; @@ -1414,8 +1432,8 @@ struct packet_manager_funcs { int (*unmap_queues)(struct packet_manager *pm, uint32_t *buffer, enum kfd_unmap_queues_filter mode, uint32_t filter_param, bool reset); - int (*set_grace_period)(struct packet_manager *pm, uint32_t *buffer, - uint32_t grace_period); + int (*config_dequeue_wait_counts)(struct packet_manager *pm, uint32_t *buffer, + enum kfd_config_dequeue_wait_counts_cmd cmd, uint32_t value); int (*query_status)(struct packet_manager *pm, uint32_t *buffer, uint64_t fence_address, uint64_t fence_value); int (*release_mem)(uint64_t gpu_addr, uint32_t *buffer); @@ -1426,7 +1444,7 @@ struct packet_manager_funcs { int set_resources_size; int map_queues_size; int unmap_queues_size; - int set_grace_period_size; + int config_dequeue_wait_counts_size; int query_status_size; int release_mem_size; }; @@ -1449,7 +1467,9 @@ int pm_send_unmap_queue(struct packet_manager *pm, void pm_release_ib(struct packet_manager *pm); -int pm_update_grace_period(struct packet_manager *pm, uint32_t grace_period); +int pm_config_dequeue_wait_counts(struct packet_manager *pm, + enum kfd_config_dequeue_wait_counts_cmd cmd, + uint32_t wait_counts_config); /* Following PM funcs can be shared among VI and AI */ unsigned int pm_build_pm4_header(unsigned int opcode, size_t packet_size); -- 2.51.0 From 0d9cabc8f591ea1cd97c071b853b75b155c13259 Mon Sep 17 00:00:00 2001 From: Leon Huang Date: Tue, 11 Feb 2025 15:45:43 +0800 Subject: [PATCH 10/16] drm/amd/display: Fix incorrect DPCD configs while Replay/PSR switch [Why] When switching between PSR/Replay, the DPCD config of previous mode is not cleared, resulting in unexpected behavior in TCON. [How] Initialize the DPCD in setup function Reviewed-by: Robin Chen Signed-off-by: Leon Huang Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../link/protocols/link_edp_panel_control.c | 25 ++++++++++++++++--- 1 file changed, 22 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c index e0e3bb865359..1e4adbc764ea 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c @@ -675,6 +675,18 @@ bool edp_setup_psr(struct dc_link *link, if (!link) return false; + //Clear PSR cfg + memset(&psr_configuration, 0, sizeof(psr_configuration)); + dm_helpers_dp_write_dpcd( + link->ctx, + link, + DP_PSR_EN_CFG, + &psr_configuration.raw, + sizeof(psr_configuration.raw)); + + if (link->psr_settings.psr_version == DC_PSR_VERSION_UNSUPPORTED) + return false; + dc = link->ctx->dc; dmcu = dc->res_pool->dmcu; psr = dc->res_pool->psr; @@ -685,9 +697,6 @@ bool edp_setup_psr(struct dc_link *link, if (!dc_get_edp_link_panel_inst(dc, link, &panel_inst)) return false; - - memset(&psr_configuration, 0, sizeof(psr_configuration)); - psr_configuration.bits.ENABLE = 1; psr_configuration.bits.CRC_VERIFICATION = 1; psr_configuration.bits.FRAME_CAPTURE_INDICATION = @@ -950,6 +959,16 @@ bool edp_setup_replay(struct dc_link *link, const struct dc_stream_state *stream if (!link) return false; + //Clear Replay config + dm_helpers_dp_write_dpcd(link->ctx, link, + DP_SINK_PR_ENABLE_AND_CONFIGURATION, + (uint8_t *)&(replay_config.raw), sizeof(uint8_t)); + + if (!(link->replay_settings.config.replay_supported)) + return false; + + link->replay_settings.config.replay_error_status.raw = 0; + dc = link->ctx->dc; replay = dc->res_pool->replay; -- 2.51.0 From 7b1ba19eb15f88e70782642ce2d934211269337b Mon Sep 17 00:00:00 2001 From: Leo Li Date: Thu, 20 Feb 2025 16:20:26 -0500 Subject: [PATCH 11/16] drm/amd/display: Disable unneeded hpd interrupts during dm_init [Why] It seems HPD interrupts are enabled by default for all connectors, even if the hpd source isn't valid. An eDP for example, does not have a valid hpd source (but does have a valid hpdrx source; see construct_phy()). Thus, eDPs should have their hpd interrupt disabled. In the past, this wasn't really an issue. Although the driver gets interrupted, then acks by writing to hw registers, there weren't any subscribed handlers that did anything meaningful (see register_hpd_handlers()). But things changed with the introduction of IPS. s2idle requires that the driver allows IPS for DMUB fw to put hw to sleep. Since register access requires hw to be awake, the driver will block IPS entry to do so. And no IPS means no hw sleep during s2idle. This was the observation on DCN35 systems with an eDP. During suspend, the eDP toggled its hpd pin as part of the panel power down sequence. The driver was then interrupted, and acked by writing to registers, blocking IPS entry. [How] Since DC marks eDP connections as having invalid hpd sources (see construct_phy()), DM should disable them at the hw level. Do so in amdgpu_dm_hpd_init() by disabling all hpd ints first, then selectively enabling ones for connectors that have valid hpd sources. Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Harry Wentland Signed-off-by: Leo Li Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c | 64 +++++++++++++------ 1 file changed, 45 insertions(+), 19 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c index 2b63cbab0e87..b61e210f6246 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c @@ -890,8 +890,16 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) struct drm_device *dev = adev_to_drm(adev); struct drm_connector *connector; struct drm_connector_list_iter iter; + int irq_type; int i; + /* First, clear all hpd and hpdrx interrupts */ + for (i = DC_IRQ_SOURCE_HPD1; i <= DC_IRQ_SOURCE_HPD6RX; i++) { + if (!dc_interrupt_set(adev->dm.dc, i, false)) + drm_err(dev, "Failed to clear hpd(rx) source=%d on init\n", + i); + } + drm_connector_list_iter_begin(dev, &iter); drm_for_each_connector_iter(connector, &iter) { struct amdgpu_dm_connector *amdgpu_dm_connector; @@ -904,10 +912,31 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) dc_link = amdgpu_dm_connector->dc_link; + /* + * Get a base driver irq reference for hpd ints for the lifetime + * of dm. Note that only hpd interrupt types are registered with + * base driver; hpd_rx types aren't. IOW, amdgpu_irq_get/put on + * hpd_rx isn't available. DM currently controls hpd_rx + * explicitly with dc_interrupt_set() + */ if (dc_link->irq_source_hpd != DC_IRQ_SOURCE_INVALID) { - dc_interrupt_set(adev->dm.dc, - dc_link->irq_source_hpd, - true); + irq_type = dc_link->irq_source_hpd - DC_IRQ_SOURCE_HPD1; + /* + * TODO: There's a mismatch between mode_info.num_hpd + * and what bios reports as the # of connectors with hpd + * sources. Since the # of hpd source types registered + * with base driver == mode_info.num_hpd, we have to + * fallback to dc_interrupt_set for the remaining types. + */ + if (irq_type < adev->mode_info.num_hpd) { + if (amdgpu_irq_get(adev, &adev->hpd_irq, irq_type)) + drm_err(dev, "DM_IRQ: Failed get HPD for source=%d)!\n", + dc_link->irq_source_hpd); + } else { + dc_interrupt_set(adev->dm.dc, + dc_link->irq_source_hpd, + true); + } } if (dc_link->irq_source_hpd_rx != DC_IRQ_SOURCE_INVALID) { @@ -917,12 +946,6 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) } } drm_connector_list_iter_end(&iter); - - /* Update reference counts for HPDs */ - for (i = DC_IRQ_SOURCE_HPD1; i <= adev->mode_info.num_hpd; i++) { - if (amdgpu_irq_get(adev, &adev->hpd_irq, i - DC_IRQ_SOURCE_HPD1)) - drm_err(dev, "DM_IRQ: Failed get HPD for source=%d)!\n", i); - } } /** @@ -938,7 +961,7 @@ void amdgpu_dm_hpd_fini(struct amdgpu_device *adev) struct drm_device *dev = adev_to_drm(adev); struct drm_connector *connector; struct drm_connector_list_iter iter; - int i; + int irq_type; drm_connector_list_iter_begin(dev, &iter); drm_for_each_connector_iter(connector, &iter) { @@ -952,9 +975,18 @@ void amdgpu_dm_hpd_fini(struct amdgpu_device *adev) dc_link = amdgpu_dm_connector->dc_link; if (dc_link->irq_source_hpd != DC_IRQ_SOURCE_INVALID) { - dc_interrupt_set(adev->dm.dc, - dc_link->irq_source_hpd, - false); + irq_type = dc_link->irq_source_hpd - DC_IRQ_SOURCE_HPD1; + + /* TODO: See same TODO in amdgpu_dm_hpd_init() */ + if (irq_type < adev->mode_info.num_hpd) { + if (amdgpu_irq_put(adev, &adev->hpd_irq, irq_type)) + drm_err(dev, "DM_IRQ: Failed put HPD for source=%d!\n", + dc_link->irq_source_hpd); + } else { + dc_interrupt_set(adev->dm.dc, + dc_link->irq_source_hpd, + false); + } } if (dc_link->irq_source_hpd_rx != DC_IRQ_SOURCE_INVALID) { @@ -964,10 +996,4 @@ void amdgpu_dm_hpd_fini(struct amdgpu_device *adev) } } drm_connector_list_iter_end(&iter); - - /* Update reference counts for HPDs */ - for (i = DC_IRQ_SOURCE_HPD1; i <= adev->mode_info.num_hpd; i++) { - if (amdgpu_irq_put(adev, &adev->hpd_irq, i - DC_IRQ_SOURCE_HPD1)) - drm_err(dev, "DM_IRQ: Failed put HPD for source=%d!\n", i); - } } -- 2.51.0 From d93b92c976671bf3352ad808c3783d37f62e9a0a Mon Sep 17 00:00:00 2001 From: Joshua Aberback Date: Tue, 25 Feb 2025 16:17:25 -0500 Subject: [PATCH 12/16] drm/amd/display: Add more debug data to dmub_srv [Why] When analyzing some crash dumps, not all of the expected DMUB info was available, so we want to add in-object storage for this data. [How] - dmub_srv_debug (renamed to dmub_timeout_info) is already a member of dmub_diagnostic_data, therefore keep a dmub_diagnostic_data directly in dmub_srv - use dmub_srv->debug when collecting diagnostic info instead of stack object to allow for easy inspection in crash dumps Reviewed-by: Alvin Lee Signed-off-by: Joshua Aberback Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c | 89 +++++++++-------- drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h | 2 +- drivers/gpu/drm/amd/display/dmub/dmub_srv.h | 10 +- .../gpu/drm/amd/display/dmub/src/dmub_dcn20.c | 85 ++++++++-------- .../gpu/drm/amd/display/dmub/src/dmub_dcn20.h | 2 +- .../gpu/drm/amd/display/dmub/src/dmub_dcn31.c | 93 +++++++++--------- .../gpu/drm/amd/display/dmub/src/dmub_dcn31.h | 2 +- .../gpu/drm/amd/display/dmub/src/dmub_dcn32.c | 98 ++++++++++--------- .../gpu/drm/amd/display/dmub/src/dmub_dcn32.h | 2 +- .../gpu/drm/amd/display/dmub/src/dmub_dcn35.c | 93 +++++++++--------- .../gpu/drm/amd/display/dmub/src/dmub_dcn35.h | 2 +- .../drm/amd/display/dmub/src/dmub_dcn401.c | 97 +++++++++--------- .../drm/amd/display/dmub/src/dmub_dcn401.h | 2 +- .../gpu/drm/amd/display/dmub/src/dmub_srv.c | 6 +- 14 files changed, 298 insertions(+), 285 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c index 87b4c2793df3..614e03bfd598 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c @@ -200,10 +200,10 @@ bool dc_dmub_srv_wait_for_idle(struct dc_dmub_srv *dc_dmub_srv, if (status != DMUB_STATUS_OK) { DC_LOG_DEBUG("No reply for DMUB command: status=%d\n", status); - if (!dmub->debug.timeout_occured) { - dmub->debug.timeout_occured = true; - dmub->debug.timeout_cmd = *cmd_list; - dmub->debug.timestamp = dm_get_timestamp(dc_dmub_srv->ctx); + if (!dmub->debug.timeout_info.timeout_occured) { + dmub->debug.timeout_info.timeout_occured = true; + dmub->debug.timeout_info.timeout_cmd = *cmd_list; + dmub->debug.timeout_info.timestamp = dm_get_timestamp(dc_dmub_srv->ctx); } dc_dmub_srv_log_diagnostic_data(dc_dmub_srv); return false; @@ -927,16 +927,15 @@ void dc_dmub_setup_subvp_dmub_command(struct dc *dc, dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT); } -bool dc_dmub_srv_get_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv, struct dmub_diagnostic_data *diag_data) +bool dc_dmub_srv_get_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv) { - if (!dc_dmub_srv || !dc_dmub_srv->dmub || !diag_data) + if (!dc_dmub_srv || !dc_dmub_srv->dmub) return false; - return dmub_srv_get_diagnostic_data(dc_dmub_srv->dmub, diag_data); + return dmub_srv_get_diagnostic_data(dc_dmub_srv->dmub); } void dc_dmub_srv_log_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv) { - struct dmub_diagnostic_data diag_data = {0}; uint32_t i; if (!dc_dmub_srv || !dc_dmub_srv->dmub) { @@ -946,49 +945,49 @@ void dc_dmub_srv_log_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv) DC_LOG_ERROR("%s: DMCUB error - collecting diagnostic data\n", __func__); - if (!dc_dmub_srv_get_diagnostic_data(dc_dmub_srv, &diag_data)) { + if (!dc_dmub_srv_get_diagnostic_data(dc_dmub_srv)) { DC_LOG_ERROR("%s: dc_dmub_srv_get_diagnostic_data failed.", __func__); return; } DC_LOG_DEBUG("DMCUB STATE:"); - DC_LOG_DEBUG(" dmcub_version : %08x", diag_data.dmcub_version); - DC_LOG_DEBUG(" scratch [0] : %08x", diag_data.scratch[0]); - DC_LOG_DEBUG(" scratch [1] : %08x", diag_data.scratch[1]); - DC_LOG_DEBUG(" scratch [2] : %08x", diag_data.scratch[2]); - DC_LOG_DEBUG(" scratch [3] : %08x", diag_data.scratch[3]); - DC_LOG_DEBUG(" scratch [4] : %08x", diag_data.scratch[4]); - DC_LOG_DEBUG(" scratch [5] : %08x", diag_data.scratch[5]); - DC_LOG_DEBUG(" scratch [6] : %08x", diag_data.scratch[6]); - DC_LOG_DEBUG(" scratch [7] : %08x", diag_data.scratch[7]); - DC_LOG_DEBUG(" scratch [8] : %08x", diag_data.scratch[8]); - DC_LOG_DEBUG(" scratch [9] : %08x", diag_data.scratch[9]); - DC_LOG_DEBUG(" scratch [10] : %08x", diag_data.scratch[10]); - DC_LOG_DEBUG(" scratch [11] : %08x", diag_data.scratch[11]); - DC_LOG_DEBUG(" scratch [12] : %08x", diag_data.scratch[12]); - DC_LOG_DEBUG(" scratch [13] : %08x", diag_data.scratch[13]); - DC_LOG_DEBUG(" scratch [14] : %08x", diag_data.scratch[14]); - DC_LOG_DEBUG(" scratch [15] : %08x", diag_data.scratch[15]); + DC_LOG_DEBUG(" dmcub_version : %08x", dc_dmub_srv->dmub->debug.dmcub_version); + DC_LOG_DEBUG(" scratch [0] : %08x", dc_dmub_srv->dmub->debug.scratch[0]); + DC_LOG_DEBUG(" scratch [1] : %08x", dc_dmub_srv->dmub->debug.scratch[1]); + DC_LOG_DEBUG(" scratch [2] : %08x", dc_dmub_srv->dmub->debug.scratch[2]); + DC_LOG_DEBUG(" scratch [3] : %08x", dc_dmub_srv->dmub->debug.scratch[3]); + DC_LOG_DEBUG(" scratch [4] : %08x", dc_dmub_srv->dmub->debug.scratch[4]); + DC_LOG_DEBUG(" scratch [5] : %08x", dc_dmub_srv->dmub->debug.scratch[5]); + DC_LOG_DEBUG(" scratch [6] : %08x", dc_dmub_srv->dmub->debug.scratch[6]); + DC_LOG_DEBUG(" scratch [7] : %08x", dc_dmub_srv->dmub->debug.scratch[7]); + DC_LOG_DEBUG(" scratch [8] : %08x", dc_dmub_srv->dmub->debug.scratch[8]); + DC_LOG_DEBUG(" scratch [9] : %08x", dc_dmub_srv->dmub->debug.scratch[9]); + DC_LOG_DEBUG(" scratch [10] : %08x", dc_dmub_srv->dmub->debug.scratch[10]); + DC_LOG_DEBUG(" scratch [11] : %08x", dc_dmub_srv->dmub->debug.scratch[11]); + DC_LOG_DEBUG(" scratch [12] : %08x", dc_dmub_srv->dmub->debug.scratch[12]); + DC_LOG_DEBUG(" scratch [13] : %08x", dc_dmub_srv->dmub->debug.scratch[13]); + DC_LOG_DEBUG(" scratch [14] : %08x", dc_dmub_srv->dmub->debug.scratch[14]); + DC_LOG_DEBUG(" scratch [15] : %08x", dc_dmub_srv->dmub->debug.scratch[15]); for (i = 0; i < DMUB_PC_SNAPSHOT_COUNT; i++) - DC_LOG_DEBUG(" pc[%d] : %08x", i, diag_data.pc[i]); - DC_LOG_DEBUG(" unk_fault_addr : %08x", diag_data.undefined_address_fault_addr); - DC_LOG_DEBUG(" inst_fault_addr : %08x", diag_data.inst_fetch_fault_addr); - DC_LOG_DEBUG(" data_fault_addr : %08x", diag_data.data_write_fault_addr); - DC_LOG_DEBUG(" inbox1_rptr : %08x", diag_data.inbox1_rptr); - DC_LOG_DEBUG(" inbox1_wptr : %08x", diag_data.inbox1_wptr); - DC_LOG_DEBUG(" inbox1_size : %08x", diag_data.inbox1_size); - DC_LOG_DEBUG(" inbox0_rptr : %08x", diag_data.inbox0_rptr); - DC_LOG_DEBUG(" inbox0_wptr : %08x", diag_data.inbox0_wptr); - DC_LOG_DEBUG(" inbox0_size : %08x", diag_data.inbox0_size); - DC_LOG_DEBUG(" outbox1_rptr : %08x", diag_data.outbox1_rptr); - DC_LOG_DEBUG(" outbox1_wptr : %08x", diag_data.outbox1_wptr); - DC_LOG_DEBUG(" outbox1_size : %08x", diag_data.outbox1_size); - DC_LOG_DEBUG(" is_enabled : %d", diag_data.is_dmcub_enabled); - DC_LOG_DEBUG(" is_soft_reset : %d", diag_data.is_dmcub_soft_reset); - DC_LOG_DEBUG(" is_secure_reset : %d", diag_data.is_dmcub_secure_reset); - DC_LOG_DEBUG(" is_traceport_en : %d", diag_data.is_traceport_en); - DC_LOG_DEBUG(" is_cw0_en : %d", diag_data.is_cw0_enabled); - DC_LOG_DEBUG(" is_cw6_en : %d", diag_data.is_cw6_enabled); + DC_LOG_DEBUG(" pc[%d] : %08x", i, dc_dmub_srv->dmub->debug.pc[i]); + DC_LOG_DEBUG(" unk_fault_addr : %08x", dc_dmub_srv->dmub->debug.undefined_address_fault_addr); + DC_LOG_DEBUG(" inst_fault_addr : %08x", dc_dmub_srv->dmub->debug.inst_fetch_fault_addr); + DC_LOG_DEBUG(" data_fault_addr : %08x", dc_dmub_srv->dmub->debug.data_write_fault_addr); + DC_LOG_DEBUG(" inbox1_rptr : %08x", dc_dmub_srv->dmub->debug.inbox1_rptr); + DC_LOG_DEBUG(" inbox1_wptr : %08x", dc_dmub_srv->dmub->debug.inbox1_wptr); + DC_LOG_DEBUG(" inbox1_size : %08x", dc_dmub_srv->dmub->debug.inbox1_size); + DC_LOG_DEBUG(" inbox0_rptr : %08x", dc_dmub_srv->dmub->debug.inbox0_rptr); + DC_LOG_DEBUG(" inbox0_wptr : %08x", dc_dmub_srv->dmub->debug.inbox0_wptr); + DC_LOG_DEBUG(" inbox0_size : %08x", dc_dmub_srv->dmub->debug.inbox0_size); + DC_LOG_DEBUG(" outbox1_rptr : %08x", dc_dmub_srv->dmub->debug.outbox1_rptr); + DC_LOG_DEBUG(" outbox1_wptr : %08x", dc_dmub_srv->dmub->debug.outbox1_wptr); + DC_LOG_DEBUG(" outbox1_size : %08x", dc_dmub_srv->dmub->debug.outbox1_size); + DC_LOG_DEBUG(" is_enabled : %d", dc_dmub_srv->dmub->debug.is_dmcub_enabled); + DC_LOG_DEBUG(" is_soft_reset : %d", dc_dmub_srv->dmub->debug.is_dmcub_soft_reset); + DC_LOG_DEBUG(" is_secure_reset : %d", dc_dmub_srv->dmub->debug.is_dmcub_secure_reset); + DC_LOG_DEBUG(" is_traceport_en : %d", dc_dmub_srv->dmub->debug.is_traceport_en); + DC_LOG_DEBUG(" is_cw0_en : %d", dc_dmub_srv->dmub->debug.is_cw0_enabled); + DC_LOG_DEBUG(" is_cw6_en : %d", dc_dmub_srv->dmub->debug.is_cw6_enabled); } static bool dc_dmub_should_update_cursor_data(struct pipe_ctx *pipe_ctx) diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h index 10b48198b7a6..a636f4c3f01d 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h @@ -94,7 +94,7 @@ void dc_dmub_srv_clear_inbox0_ack(struct dc_dmub_srv *dmub_srv); void dc_dmub_srv_wait_for_inbox0_ack(struct dc_dmub_srv *dmub_srv); void dc_dmub_srv_send_inbox0_cmd(struct dc_dmub_srv *dmub_srv, union dmub_inbox0_data_register data); -bool dc_dmub_srv_get_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv, struct dmub_diagnostic_data *dmub_oca); +bool dc_dmub_srv_get_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv); void dc_dmub_setup_subvp_dmub_command(struct dc *dc, struct dc_state *context, bool enable); void dc_dmub_srv_log_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv); diff --git a/drivers/gpu/drm/amd/display/dmub/dmub_srv.h b/drivers/gpu/drm/amd/display/dmub/dmub_srv.h index 0787cc3904d6..8ebeb34021e5 100644 --- a/drivers/gpu/drm/amd/display/dmub/dmub_srv.h +++ b/drivers/gpu/drm/amd/display/dmub/dmub_srv.h @@ -313,7 +313,7 @@ struct dmub_srv_hw_params { * @timeout_occured: Indicates a timeout occured on any message from driver to dmub * @timeout_cmd: first cmd sent from driver that timed out - subsequent timeouts are not stored */ -struct dmub_srv_debug { +struct dmub_timeout_info { bool timeout_occured; union dmub_rb_cmd timeout_cmd; unsigned long long timestamp; @@ -340,7 +340,7 @@ struct dmub_diagnostic_data { uint32_t outbox1_wptr; uint32_t outbox1_size; uint32_t gpint_datain0; - struct dmub_srv_debug timeout_info; + struct dmub_timeout_info timeout_info; uint8_t is_dmcub_enabled : 1; uint8_t is_dmcub_soft_reset : 1; uint8_t is_dmcub_secure_reset : 1; @@ -456,7 +456,7 @@ struct dmub_srv_hw_funcs { void (*send_inbox0_cmd)(struct dmub_srv *dmub, union dmub_inbox0_data_register data); uint32_t (*get_current_time)(struct dmub_srv *dmub); - void (*get_diagnostic_data)(struct dmub_srv *dmub, struct dmub_diagnostic_data *dmub_oca); + void (*get_diagnostic_data)(struct dmub_srv *dmub); bool (*should_detect)(struct dmub_srv *dmub); void (*init_reg_offsets)(struct dmub_srv *dmub, struct dc_context *ctx); @@ -545,7 +545,7 @@ struct dmub_srv { struct dmub_visual_confirm_color visual_confirm_color; enum dmub_srv_power_state_type power_state; - struct dmub_srv_debug debug; + struct dmub_diagnostic_data debug; }; /** @@ -901,7 +901,7 @@ enum dmub_status dmub_srv_set_skip_panel_power_sequence(struct dmub_srv *dmub, bool dmub_srv_get_outbox0_msg(struct dmub_srv *dmub, struct dmcub_trace_buf_entry *entry); -bool dmub_srv_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data); +bool dmub_srv_get_diagnostic_data(struct dmub_srv *dmub); bool dmub_srv_should_detect(struct dmub_srv *dmub); 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 e500ca9ae09c..73888c1bea93 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.c @@ -414,63 +414,66 @@ uint32_t dmub_dcn20_get_current_time(struct dmub_srv *dmub) return REG_READ(DMCUB_TIMER_CURRENT); } -void dmub_dcn20_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data) +void dmub_dcn20_get_diagnostic_data(struct dmub_srv *dmub) { uint32_t is_dmub_enabled, is_soft_reset, is_sec_reset; uint32_t is_traceport_enabled, is_cw0_enabled, is_cw6_enabled; + struct dmub_timeout_info timeout = {0}; - if (!dmub || !diag_data) + if (!dmub) return; - memset(diag_data, 0, sizeof(*diag_data)); - - diag_data->dmcub_version = dmub->fw_version; - - diag_data->scratch[0] = REG_READ(DMCUB_SCRATCH0); - diag_data->scratch[1] = REG_READ(DMCUB_SCRATCH1); - diag_data->scratch[2] = REG_READ(DMCUB_SCRATCH2); - diag_data->scratch[3] = REG_READ(DMCUB_SCRATCH3); - diag_data->scratch[4] = REG_READ(DMCUB_SCRATCH4); - diag_data->scratch[5] = REG_READ(DMCUB_SCRATCH5); - diag_data->scratch[6] = REG_READ(DMCUB_SCRATCH6); - diag_data->scratch[7] = REG_READ(DMCUB_SCRATCH7); - diag_data->scratch[8] = REG_READ(DMCUB_SCRATCH8); - diag_data->scratch[9] = REG_READ(DMCUB_SCRATCH9); - diag_data->scratch[10] = REG_READ(DMCUB_SCRATCH10); - diag_data->scratch[11] = REG_READ(DMCUB_SCRATCH11); - diag_data->scratch[12] = REG_READ(DMCUB_SCRATCH12); - diag_data->scratch[13] = REG_READ(DMCUB_SCRATCH13); - diag_data->scratch[14] = REG_READ(DMCUB_SCRATCH14); - diag_data->scratch[15] = REG_READ(DMCUB_SCRATCH15); - - diag_data->undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); - diag_data->inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); - diag_data->data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); - - diag_data->inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); - diag_data->inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); - diag_data->inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); - - diag_data->inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); - diag_data->inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); - diag_data->inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); + /* timeout data filled externally, cache before resetting memory */ + timeout = dmub->debug.timeout_info; + memset(&dmub->debug, 0, sizeof(dmub->debug)); + dmub->debug.timeout_info = timeout; + + dmub->debug.dmcub_version = dmub->fw_version; + + dmub->debug.scratch[0] = REG_READ(DMCUB_SCRATCH0); + dmub->debug.scratch[1] = REG_READ(DMCUB_SCRATCH1); + dmub->debug.scratch[2] = REG_READ(DMCUB_SCRATCH2); + dmub->debug.scratch[3] = REG_READ(DMCUB_SCRATCH3); + dmub->debug.scratch[4] = REG_READ(DMCUB_SCRATCH4); + dmub->debug.scratch[5] = REG_READ(DMCUB_SCRATCH5); + dmub->debug.scratch[6] = REG_READ(DMCUB_SCRATCH6); + dmub->debug.scratch[7] = REG_READ(DMCUB_SCRATCH7); + dmub->debug.scratch[8] = REG_READ(DMCUB_SCRATCH8); + dmub->debug.scratch[9] = REG_READ(DMCUB_SCRATCH9); + dmub->debug.scratch[10] = REG_READ(DMCUB_SCRATCH10); + dmub->debug.scratch[11] = REG_READ(DMCUB_SCRATCH11); + dmub->debug.scratch[12] = REG_READ(DMCUB_SCRATCH12); + dmub->debug.scratch[13] = REG_READ(DMCUB_SCRATCH13); + dmub->debug.scratch[14] = REG_READ(DMCUB_SCRATCH14); + dmub->debug.scratch[15] = REG_READ(DMCUB_SCRATCH15); + + dmub->debug.undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); + dmub->debug.inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); + dmub->debug.data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); + + dmub->debug.inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); + dmub->debug.inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); + dmub->debug.inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); + + dmub->debug.inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); + dmub->debug.inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); + dmub->debug.inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_dmub_enabled); - diag_data->is_dmcub_enabled = is_dmub_enabled; + dmub->debug.is_dmcub_enabled = is_dmub_enabled; REG_GET(DMCUB_CNTL, DMCUB_SOFT_RESET, &is_soft_reset); - diag_data->is_dmcub_soft_reset = is_soft_reset; + dmub->debug.is_dmcub_soft_reset = is_soft_reset; REG_GET(DMCUB_SEC_CNTL, DMCUB_SEC_RESET_STATUS, &is_sec_reset); - diag_data->is_dmcub_secure_reset = is_sec_reset; + dmub->debug.is_dmcub_secure_reset = is_sec_reset; REG_GET(DMCUB_CNTL, DMCUB_TRACEPORT_EN, &is_traceport_enabled); - diag_data->is_traceport_en = is_traceport_enabled; + dmub->debug.is_traceport_en = is_traceport_enabled; REG_GET(DMCUB_REGION3_CW0_TOP_ADDRESS, DMCUB_REGION3_CW0_ENABLE, &is_cw0_enabled); - diag_data->is_cw0_enabled = is_cw0_enabled; + dmub->debug.is_cw0_enabled = is_cw0_enabled; REG_GET(DMCUB_REGION3_CW6_TOP_ADDRESS, DMCUB_REGION3_CW6_ENABLE, &is_cw6_enabled); - diag_data->is_cw6_enabled = is_cw6_enabled; - diag_data->timeout_info = dmub->debug; + dmub->debug.is_cw6_enabled = is_cw6_enabled; } 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 de287b101848..42c1fb4bc73f 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.h +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn20.h @@ -247,6 +247,6 @@ bool dmub_dcn20_use_cached_trace_buffer(struct dmub_srv *dmub); uint32_t dmub_dcn20_get_current_time(struct dmub_srv *dmub); -void dmub_dcn20_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *dmub_oca); +void dmub_dcn20_get_diagnostic_data(struct dmub_srv *dmub); #endif /* _DMUB_DCN20_H_ */ diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c index 9796077885c9..a308bd604677 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c @@ -414,69 +414,72 @@ uint32_t dmub_dcn31_get_current_time(struct dmub_srv *dmub) return REG_READ(DMCUB_TIMER_CURRENT); } -void dmub_dcn31_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data) +void dmub_dcn31_get_diagnostic_data(struct dmub_srv *dmub) { uint32_t is_dmub_enabled, is_soft_reset, is_sec_reset; uint32_t is_traceport_enabled, is_cw0_enabled, is_cw6_enabled; + struct dmub_timeout_info timeout = {0}; - if (!dmub || !diag_data) + if (!dmub) return; - memset(diag_data, 0, sizeof(*diag_data)); - - diag_data->dmcub_version = dmub->fw_version; - - diag_data->scratch[0] = REG_READ(DMCUB_SCRATCH0); - diag_data->scratch[1] = REG_READ(DMCUB_SCRATCH1); - diag_data->scratch[2] = REG_READ(DMCUB_SCRATCH2); - diag_data->scratch[3] = REG_READ(DMCUB_SCRATCH3); - diag_data->scratch[4] = REG_READ(DMCUB_SCRATCH4); - diag_data->scratch[5] = REG_READ(DMCUB_SCRATCH5); - diag_data->scratch[6] = REG_READ(DMCUB_SCRATCH6); - diag_data->scratch[7] = REG_READ(DMCUB_SCRATCH7); - diag_data->scratch[8] = REG_READ(DMCUB_SCRATCH8); - diag_data->scratch[9] = REG_READ(DMCUB_SCRATCH9); - diag_data->scratch[10] = REG_READ(DMCUB_SCRATCH10); - diag_data->scratch[11] = REG_READ(DMCUB_SCRATCH11); - diag_data->scratch[12] = REG_READ(DMCUB_SCRATCH12); - diag_data->scratch[13] = REG_READ(DMCUB_SCRATCH13); - diag_data->scratch[14] = REG_READ(DMCUB_SCRATCH14); - diag_data->scratch[15] = REG_READ(DMCUB_SCRATCH15); - - diag_data->undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); - diag_data->inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); - diag_data->data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); - - diag_data->inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); - diag_data->inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); - diag_data->inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); - - diag_data->inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); - diag_data->inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); - diag_data->inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); - - diag_data->outbox1_rptr = REG_READ(DMCUB_OUTBOX1_RPTR); - diag_data->outbox1_wptr = REG_READ(DMCUB_OUTBOX1_WPTR); - diag_data->outbox1_size = REG_READ(DMCUB_OUTBOX1_SIZE); + /* timeout data filled externally, cache before resetting memory */ + timeout = dmub->debug.timeout_info; + memset(&dmub->debug, 0, sizeof(dmub->debug)); + dmub->debug.timeout_info = timeout; + + dmub->debug.dmcub_version = dmub->fw_version; + + dmub->debug.scratch[0] = REG_READ(DMCUB_SCRATCH0); + dmub->debug.scratch[1] = REG_READ(DMCUB_SCRATCH1); + dmub->debug.scratch[2] = REG_READ(DMCUB_SCRATCH2); + dmub->debug.scratch[3] = REG_READ(DMCUB_SCRATCH3); + dmub->debug.scratch[4] = REG_READ(DMCUB_SCRATCH4); + dmub->debug.scratch[5] = REG_READ(DMCUB_SCRATCH5); + dmub->debug.scratch[6] = REG_READ(DMCUB_SCRATCH6); + dmub->debug.scratch[7] = REG_READ(DMCUB_SCRATCH7); + dmub->debug.scratch[8] = REG_READ(DMCUB_SCRATCH8); + dmub->debug.scratch[9] = REG_READ(DMCUB_SCRATCH9); + dmub->debug.scratch[10] = REG_READ(DMCUB_SCRATCH10); + dmub->debug.scratch[11] = REG_READ(DMCUB_SCRATCH11); + dmub->debug.scratch[12] = REG_READ(DMCUB_SCRATCH12); + dmub->debug.scratch[13] = REG_READ(DMCUB_SCRATCH13); + dmub->debug.scratch[14] = REG_READ(DMCUB_SCRATCH14); + dmub->debug.scratch[15] = REG_READ(DMCUB_SCRATCH15); + + dmub->debug.undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); + dmub->debug.inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); + dmub->debug.data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); + + dmub->debug.inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); + dmub->debug.inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); + dmub->debug.inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); + + dmub->debug.inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); + dmub->debug.inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); + dmub->debug.inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); + + dmub->debug.outbox1_rptr = REG_READ(DMCUB_OUTBOX1_RPTR); + dmub->debug.outbox1_wptr = REG_READ(DMCUB_OUTBOX1_WPTR); + dmub->debug.outbox1_size = REG_READ(DMCUB_OUTBOX1_SIZE); REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_dmub_enabled); - diag_data->is_dmcub_enabled = is_dmub_enabled; + dmub->debug.is_dmcub_enabled = is_dmub_enabled; REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &is_soft_reset); - diag_data->is_dmcub_soft_reset = is_soft_reset; + dmub->debug.is_dmcub_soft_reset = is_soft_reset; REG_GET(DMCUB_SEC_CNTL, DMCUB_SEC_RESET_STATUS, &is_sec_reset); - diag_data->is_dmcub_secure_reset = is_sec_reset; + dmub->debug.is_dmcub_secure_reset = is_sec_reset; REG_GET(DMCUB_CNTL, DMCUB_TRACEPORT_EN, &is_traceport_enabled); - diag_data->is_traceport_en = is_traceport_enabled; + dmub->debug.is_traceport_en = is_traceport_enabled; REG_GET(DMCUB_REGION3_CW0_TOP_ADDRESS, DMCUB_REGION3_CW0_ENABLE, &is_cw0_enabled); - diag_data->is_cw0_enabled = is_cw0_enabled; + dmub->debug.is_cw0_enabled = is_cw0_enabled; REG_GET(DMCUB_REGION3_CW6_TOP_ADDRESS, DMCUB_REGION3_CW6_ENABLE, &is_cw6_enabled); - diag_data->is_cw6_enabled = is_cw6_enabled; - diag_data->timeout_info = dmub->debug; + dmub->debug.is_cw6_enabled = is_cw6_enabled; } bool dmub_dcn31_should_detect(struct dmub_srv *dmub) diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.h b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.h index eccdab4986ce..1c43ef2bca66 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.h +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.h @@ -251,7 +251,7 @@ void dmub_dcn31_set_outbox0_rptr(struct dmub_srv *dmub, uint32_t rptr_offset); uint32_t dmub_dcn31_get_current_time(struct dmub_srv *dmub); -void dmub_dcn31_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data); +void dmub_dcn31_get_diagnostic_data(struct dmub_srv *dmub); bool dmub_dcn31_should_detect(struct dmub_srv *dmub); diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c index 9600b7f858b0..e7056205b050 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c @@ -417,73 +417,75 @@ uint32_t dmub_dcn32_get_current_time(struct dmub_srv *dmub) return REG_READ(DMCUB_TIMER_CURRENT); } -void dmub_dcn32_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data) +void dmub_dcn32_get_diagnostic_data(struct dmub_srv *dmub) { uint32_t is_dmub_enabled, is_soft_reset, is_sec_reset; uint32_t is_traceport_enabled, is_cw0_enabled, is_cw6_enabled; + struct dmub_timeout_info timeout = {0}; - if (!dmub || !diag_data) + if (!dmub) return; - memset(diag_data, 0, sizeof(*diag_data)); - - diag_data->dmcub_version = dmub->fw_version; - - diag_data->scratch[0] = REG_READ(DMCUB_SCRATCH0); - diag_data->scratch[1] = REG_READ(DMCUB_SCRATCH1); - diag_data->scratch[2] = REG_READ(DMCUB_SCRATCH2); - diag_data->scratch[3] = REG_READ(DMCUB_SCRATCH3); - diag_data->scratch[4] = REG_READ(DMCUB_SCRATCH4); - diag_data->scratch[5] = REG_READ(DMCUB_SCRATCH5); - diag_data->scratch[6] = REG_READ(DMCUB_SCRATCH6); - diag_data->scratch[7] = REG_READ(DMCUB_SCRATCH7); - diag_data->scratch[8] = REG_READ(DMCUB_SCRATCH8); - diag_data->scratch[9] = REG_READ(DMCUB_SCRATCH9); - diag_data->scratch[10] = REG_READ(DMCUB_SCRATCH10); - diag_data->scratch[11] = REG_READ(DMCUB_SCRATCH11); - diag_data->scratch[12] = REG_READ(DMCUB_SCRATCH12); - diag_data->scratch[13] = REG_READ(DMCUB_SCRATCH13); - diag_data->scratch[14] = REG_READ(DMCUB_SCRATCH14); - diag_data->scratch[15] = REG_READ(DMCUB_SCRATCH15); - diag_data->scratch[16] = REG_READ(DMCUB_SCRATCH16); - - diag_data->undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); - diag_data->inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); - diag_data->data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); - - diag_data->inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); - diag_data->inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); - diag_data->inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); - - diag_data->inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); - diag_data->inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); - diag_data->inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); - - diag_data->outbox1_rptr = REG_READ(DMCUB_OUTBOX1_RPTR); - diag_data->outbox1_wptr = REG_READ(DMCUB_OUTBOX1_WPTR); - diag_data->outbox1_size = REG_READ(DMCUB_OUTBOX1_SIZE); + /* timeout data filled externally, cache before resetting memory */ + timeout = dmub->debug.timeout_info; + memset(&dmub->debug, 0, sizeof(dmub->debug)); + dmub->debug.timeout_info = timeout; + + dmub->debug.dmcub_version = dmub->fw_version; + + dmub->debug.scratch[0] = REG_READ(DMCUB_SCRATCH0); + dmub->debug.scratch[1] = REG_READ(DMCUB_SCRATCH1); + dmub->debug.scratch[2] = REG_READ(DMCUB_SCRATCH2); + dmub->debug.scratch[3] = REG_READ(DMCUB_SCRATCH3); + dmub->debug.scratch[4] = REG_READ(DMCUB_SCRATCH4); + dmub->debug.scratch[5] = REG_READ(DMCUB_SCRATCH5); + dmub->debug.scratch[6] = REG_READ(DMCUB_SCRATCH6); + dmub->debug.scratch[7] = REG_READ(DMCUB_SCRATCH7); + dmub->debug.scratch[8] = REG_READ(DMCUB_SCRATCH8); + dmub->debug.scratch[9] = REG_READ(DMCUB_SCRATCH9); + dmub->debug.scratch[10] = REG_READ(DMCUB_SCRATCH10); + dmub->debug.scratch[11] = REG_READ(DMCUB_SCRATCH11); + dmub->debug.scratch[12] = REG_READ(DMCUB_SCRATCH12); + dmub->debug.scratch[13] = REG_READ(DMCUB_SCRATCH13); + dmub->debug.scratch[14] = REG_READ(DMCUB_SCRATCH14); + dmub->debug.scratch[15] = REG_READ(DMCUB_SCRATCH15); + dmub->debug.scratch[16] = REG_READ(DMCUB_SCRATCH16); + + dmub->debug.undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); + dmub->debug.inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); + dmub->debug.data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); + + dmub->debug.inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); + dmub->debug.inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); + dmub->debug.inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); + + dmub->debug.inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); + dmub->debug.inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); + dmub->debug.inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); + + dmub->debug.outbox1_rptr = REG_READ(DMCUB_OUTBOX1_RPTR); + dmub->debug.outbox1_wptr = REG_READ(DMCUB_OUTBOX1_WPTR); + dmub->debug.outbox1_size = REG_READ(DMCUB_OUTBOX1_SIZE); REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_dmub_enabled); - diag_data->is_dmcub_enabled = is_dmub_enabled; + dmub->debug.is_dmcub_enabled = is_dmub_enabled; REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &is_soft_reset); - diag_data->is_dmcub_soft_reset = is_soft_reset; + dmub->debug.is_dmcub_soft_reset = is_soft_reset; REG_GET(DMCUB_SEC_CNTL, DMCUB_SEC_RESET_STATUS, &is_sec_reset); - diag_data->is_dmcub_secure_reset = is_sec_reset; + dmub->debug.is_dmcub_secure_reset = is_sec_reset; REG_GET(DMCUB_CNTL, DMCUB_TRACEPORT_EN, &is_traceport_enabled); - diag_data->is_traceport_en = is_traceport_enabled; + dmub->debug.is_traceport_en = is_traceport_enabled; REG_GET(DMCUB_REGION3_CW0_TOP_ADDRESS, DMCUB_REGION3_CW0_ENABLE, &is_cw0_enabled); - diag_data->is_cw0_enabled = is_cw0_enabled; + dmub->debug.is_cw0_enabled = is_cw0_enabled; REG_GET(DMCUB_REGION3_CW6_TOP_ADDRESS, DMCUB_REGION3_CW6_ENABLE, &is_cw6_enabled); - diag_data->is_cw6_enabled = is_cw6_enabled; + dmub->debug.is_cw6_enabled = is_cw6_enabled; - diag_data->gpint_datain0 = REG_READ(DMCUB_GPINT_DATAIN0); - - diag_data->timeout_info = dmub->debug; + dmub->debug.gpint_datain0 = REG_READ(DMCUB_GPINT_DATAIN0); } void dmub_dcn32_configure_dmub_in_system_memory(struct dmub_srv *dmub) { diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.h b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.h index 29c1132951af..1a229450c53d 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.h +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.h @@ -254,7 +254,7 @@ void dmub_dcn32_set_outbox0_rptr(struct dmub_srv *dmub, uint32_t rptr_offset); uint32_t dmub_dcn32_get_current_time(struct dmub_srv *dmub); -void dmub_dcn32_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data); +void dmub_dcn32_get_diagnostic_data(struct dmub_srv *dmub); void dmub_dcn32_configure_dmub_in_system_memory(struct dmub_srv *dmub); void dmub_dcn32_send_inbox0_cmd(struct dmub_srv *dmub, union dmub_inbox0_data_register data); diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c index 01d013a12b94..72a0f078cd1a 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c @@ -462,66 +462,69 @@ uint32_t dmub_dcn35_get_current_time(struct dmub_srv *dmub) return REG_READ(DMCUB_TIMER_CURRENT); } -void dmub_dcn35_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data) +void dmub_dcn35_get_diagnostic_data(struct dmub_srv *dmub) { uint32_t is_dmub_enabled, is_soft_reset; uint32_t is_traceport_enabled, is_cw6_enabled; + struct dmub_timeout_info timeout = {0}; - if (!dmub || !diag_data) + if (!dmub) return; - memset(diag_data, 0, sizeof(*diag_data)); - - diag_data->dmcub_version = dmub->fw_version; - - diag_data->scratch[0] = REG_READ(DMCUB_SCRATCH0); - diag_data->scratch[1] = REG_READ(DMCUB_SCRATCH1); - diag_data->scratch[2] = REG_READ(DMCUB_SCRATCH2); - diag_data->scratch[3] = REG_READ(DMCUB_SCRATCH3); - diag_data->scratch[4] = REG_READ(DMCUB_SCRATCH4); - diag_data->scratch[5] = REG_READ(DMCUB_SCRATCH5); - diag_data->scratch[6] = REG_READ(DMCUB_SCRATCH6); - diag_data->scratch[7] = REG_READ(DMCUB_SCRATCH7); - diag_data->scratch[8] = REG_READ(DMCUB_SCRATCH8); - diag_data->scratch[9] = REG_READ(DMCUB_SCRATCH9); - diag_data->scratch[10] = REG_READ(DMCUB_SCRATCH10); - diag_data->scratch[11] = REG_READ(DMCUB_SCRATCH11); - diag_data->scratch[12] = REG_READ(DMCUB_SCRATCH12); - diag_data->scratch[13] = REG_READ(DMCUB_SCRATCH13); - diag_data->scratch[14] = REG_READ(DMCUB_SCRATCH14); - diag_data->scratch[15] = REG_READ(DMCUB_SCRATCH15); - diag_data->scratch[16] = REG_READ(DMCUB_SCRATCH16); - - diag_data->undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); - diag_data->inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); - diag_data->data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); - - diag_data->inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); - diag_data->inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); - diag_data->inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); - - diag_data->inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); - diag_data->inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); - diag_data->inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); - - diag_data->outbox1_rptr = REG_READ(DMCUB_OUTBOX1_RPTR); - diag_data->outbox1_wptr = REG_READ(DMCUB_OUTBOX1_WPTR); - diag_data->outbox1_size = REG_READ(DMCUB_OUTBOX1_SIZE); + /* timeout data filled externally, cache before resetting memory */ + timeout = dmub->debug.timeout_info; + memset(&dmub->debug, 0, sizeof(dmub->debug)); + dmub->debug.timeout_info = timeout; + + dmub->debug.dmcub_version = dmub->fw_version; + + dmub->debug.scratch[0] = REG_READ(DMCUB_SCRATCH0); + dmub->debug.scratch[1] = REG_READ(DMCUB_SCRATCH1); + dmub->debug.scratch[2] = REG_READ(DMCUB_SCRATCH2); + dmub->debug.scratch[3] = REG_READ(DMCUB_SCRATCH3); + dmub->debug.scratch[4] = REG_READ(DMCUB_SCRATCH4); + dmub->debug.scratch[5] = REG_READ(DMCUB_SCRATCH5); + dmub->debug.scratch[6] = REG_READ(DMCUB_SCRATCH6); + dmub->debug.scratch[7] = REG_READ(DMCUB_SCRATCH7); + dmub->debug.scratch[8] = REG_READ(DMCUB_SCRATCH8); + dmub->debug.scratch[9] = REG_READ(DMCUB_SCRATCH9); + dmub->debug.scratch[10] = REG_READ(DMCUB_SCRATCH10); + dmub->debug.scratch[11] = REG_READ(DMCUB_SCRATCH11); + dmub->debug.scratch[12] = REG_READ(DMCUB_SCRATCH12); + dmub->debug.scratch[13] = REG_READ(DMCUB_SCRATCH13); + dmub->debug.scratch[14] = REG_READ(DMCUB_SCRATCH14); + dmub->debug.scratch[15] = REG_READ(DMCUB_SCRATCH15); + dmub->debug.scratch[16] = REG_READ(DMCUB_SCRATCH16); + + dmub->debug.undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); + dmub->debug.inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); + dmub->debug.data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); + + dmub->debug.inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); + dmub->debug.inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); + dmub->debug.inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); + + dmub->debug.inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); + dmub->debug.inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); + dmub->debug.inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); + + dmub->debug.outbox1_rptr = REG_READ(DMCUB_OUTBOX1_RPTR); + dmub->debug.outbox1_wptr = REG_READ(DMCUB_OUTBOX1_WPTR); + dmub->debug.outbox1_size = REG_READ(DMCUB_OUTBOX1_SIZE); REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_dmub_enabled); - diag_data->is_dmcub_enabled = is_dmub_enabled; + dmub->debug.is_dmcub_enabled = is_dmub_enabled; REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &is_soft_reset); - diag_data->is_dmcub_soft_reset = is_soft_reset; + dmub->debug.is_dmcub_soft_reset = is_soft_reset; REG_GET(DMCUB_CNTL, DMCUB_TRACEPORT_EN, &is_traceport_enabled); - diag_data->is_traceport_en = is_traceport_enabled; + dmub->debug.is_traceport_en = is_traceport_enabled; REG_GET(DMCUB_REGION3_CW6_TOP_ADDRESS, DMCUB_REGION3_CW6_ENABLE, &is_cw6_enabled); - diag_data->is_cw6_enabled = is_cw6_enabled; + dmub->debug.is_cw6_enabled = is_cw6_enabled; - diag_data->gpint_datain0 = REG_READ(DMCUB_GPINT_DATAIN0); - diag_data->timeout_info = dmub->debug; + dmub->debug.gpint_datain0 = REG_READ(DMCUB_GPINT_DATAIN0); } void dmub_dcn35_configure_dmub_in_system_memory(struct dmub_srv *dmub) { diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h index 686e97c00ccc..39fcb7275da5 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h @@ -269,7 +269,7 @@ void dmub_dcn35_set_outbox0_rptr(struct dmub_srv *dmub, uint32_t rptr_offset); uint32_t dmub_dcn35_get_current_time(struct dmub_srv *dmub); -void dmub_dcn35_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data); +void dmub_dcn35_get_diagnostic_data(struct dmub_srv *dmub); void dmub_dcn35_configure_dmub_in_system_memory(struct dmub_srv *dmub); diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c index e1c4fe1c6e3e..e67f7c4784eb 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c @@ -415,72 +415,75 @@ uint32_t dmub_dcn401_get_current_time(struct dmub_srv *dmub) return REG_READ(DMCUB_TIMER_CURRENT); } -void dmub_dcn401_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data) +void dmub_dcn401_get_diagnostic_data(struct dmub_srv *dmub) { uint32_t is_dmub_enabled, is_soft_reset, is_sec_reset; uint32_t is_traceport_enabled, is_cw0_enabled, is_cw6_enabled; + struct dmub_timeout_info timeout = {0}; - if (!dmub || !diag_data) + if (!dmub) return; - memset(diag_data, 0, sizeof(*diag_data)); - - diag_data->dmcub_version = dmub->fw_version; - - diag_data->scratch[0] = REG_READ(DMCUB_SCRATCH0); - diag_data->scratch[1] = REG_READ(DMCUB_SCRATCH1); - diag_data->scratch[2] = REG_READ(DMCUB_SCRATCH2); - diag_data->scratch[3] = REG_READ(DMCUB_SCRATCH3); - diag_data->scratch[4] = REG_READ(DMCUB_SCRATCH4); - diag_data->scratch[5] = REG_READ(DMCUB_SCRATCH5); - diag_data->scratch[6] = REG_READ(DMCUB_SCRATCH6); - diag_data->scratch[7] = REG_READ(DMCUB_SCRATCH7); - diag_data->scratch[8] = REG_READ(DMCUB_SCRATCH8); - diag_data->scratch[9] = REG_READ(DMCUB_SCRATCH9); - diag_data->scratch[10] = REG_READ(DMCUB_SCRATCH10); - diag_data->scratch[11] = REG_READ(DMCUB_SCRATCH11); - diag_data->scratch[12] = REG_READ(DMCUB_SCRATCH12); - diag_data->scratch[13] = REG_READ(DMCUB_SCRATCH13); - diag_data->scratch[14] = REG_READ(DMCUB_SCRATCH14); - diag_data->scratch[15] = REG_READ(DMCUB_SCRATCH15); - diag_data->scratch[16] = REG_READ(DMCUB_SCRATCH16); - - diag_data->undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); - diag_data->inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); - diag_data->data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); - - diag_data->inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); - diag_data->inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); - diag_data->inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); - - diag_data->inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); - diag_data->inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); - diag_data->inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); - - diag_data->outbox1_rptr = REG_READ(DMCUB_OUTBOX1_RPTR); - diag_data->outbox1_wptr = REG_READ(DMCUB_OUTBOX1_WPTR); - diag_data->outbox1_size = REG_READ(DMCUB_OUTBOX1_SIZE); + /* timeout data filled externally, cache before resetting memory */ + timeout = dmub->debug.timeout_info; + memset(&dmub->debug, 0, sizeof(dmub->debug)); + dmub->debug.timeout_info = timeout; + + dmub->debug.dmcub_version = dmub->fw_version; + + dmub->debug.scratch[0] = REG_READ(DMCUB_SCRATCH0); + dmub->debug.scratch[1] = REG_READ(DMCUB_SCRATCH1); + dmub->debug.scratch[2] = REG_READ(DMCUB_SCRATCH2); + dmub->debug.scratch[3] = REG_READ(DMCUB_SCRATCH3); + dmub->debug.scratch[4] = REG_READ(DMCUB_SCRATCH4); + dmub->debug.scratch[5] = REG_READ(DMCUB_SCRATCH5); + dmub->debug.scratch[6] = REG_READ(DMCUB_SCRATCH6); + dmub->debug.scratch[7] = REG_READ(DMCUB_SCRATCH7); + dmub->debug.scratch[8] = REG_READ(DMCUB_SCRATCH8); + dmub->debug.scratch[9] = REG_READ(DMCUB_SCRATCH9); + dmub->debug.scratch[10] = REG_READ(DMCUB_SCRATCH10); + dmub->debug.scratch[11] = REG_READ(DMCUB_SCRATCH11); + dmub->debug.scratch[12] = REG_READ(DMCUB_SCRATCH12); + dmub->debug.scratch[13] = REG_READ(DMCUB_SCRATCH13); + dmub->debug.scratch[14] = REG_READ(DMCUB_SCRATCH14); + dmub->debug.scratch[15] = REG_READ(DMCUB_SCRATCH15); + dmub->debug.scratch[16] = REG_READ(DMCUB_SCRATCH16); + + dmub->debug.undefined_address_fault_addr = REG_READ(DMCUB_UNDEFINED_ADDRESS_FAULT_ADDR); + dmub->debug.inst_fetch_fault_addr = REG_READ(DMCUB_INST_FETCH_FAULT_ADDR); + dmub->debug.data_write_fault_addr = REG_READ(DMCUB_DATA_WRITE_FAULT_ADDR); + + dmub->debug.inbox1_rptr = REG_READ(DMCUB_INBOX1_RPTR); + dmub->debug.inbox1_wptr = REG_READ(DMCUB_INBOX1_WPTR); + dmub->debug.inbox1_size = REG_READ(DMCUB_INBOX1_SIZE); + + dmub->debug.inbox0_rptr = REG_READ(DMCUB_INBOX0_RPTR); + dmub->debug.inbox0_wptr = REG_READ(DMCUB_INBOX0_WPTR); + dmub->debug.inbox0_size = REG_READ(DMCUB_INBOX0_SIZE); + + dmub->debug.outbox1_rptr = REG_READ(DMCUB_OUTBOX1_RPTR); + dmub->debug.outbox1_wptr = REG_READ(DMCUB_OUTBOX1_WPTR); + dmub->debug.outbox1_size = REG_READ(DMCUB_OUTBOX1_SIZE); REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_dmub_enabled); - diag_data->is_dmcub_enabled = is_dmub_enabled; + dmub->debug.is_dmcub_enabled = is_dmub_enabled; REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &is_soft_reset); - diag_data->is_dmcub_soft_reset = is_soft_reset; + dmub->debug.is_dmcub_soft_reset = is_soft_reset; REG_GET(DMCUB_SEC_CNTL, DMCUB_SEC_RESET_STATUS, &is_sec_reset); - diag_data->is_dmcub_secure_reset = is_sec_reset; + dmub->debug.is_dmcub_secure_reset = is_sec_reset; REG_GET(DMCUB_CNTL, DMCUB_TRACEPORT_EN, &is_traceport_enabled); - diag_data->is_traceport_en = is_traceport_enabled; + dmub->debug.is_traceport_en = is_traceport_enabled; REG_GET(DMCUB_REGION3_CW0_TOP_ADDRESS, DMCUB_REGION3_CW0_ENABLE, &is_cw0_enabled); - diag_data->is_cw0_enabled = is_cw0_enabled; + dmub->debug.is_cw0_enabled = is_cw0_enabled; REG_GET(DMCUB_REGION3_CW6_TOP_ADDRESS, DMCUB_REGION3_CW6_ENABLE, &is_cw6_enabled); - diag_data->is_cw6_enabled = is_cw6_enabled; + dmub->debug.is_cw6_enabled = is_cw6_enabled; - diag_data->gpint_datain0 = REG_READ(DMCUB_GPINT_DATAIN0); - diag_data->timeout_info = dmub->debug; + dmub->debug.gpint_datain0 = REG_READ(DMCUB_GPINT_DATAIN0); } void dmub_dcn401_configure_dmub_in_system_memory(struct dmub_srv *dmub) { diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.h b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.h index 31f95b27e227..c35be52676f6 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.h +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.h @@ -264,7 +264,7 @@ void dmub_dcn401_set_outbox0_rptr(struct dmub_srv *dmub, uint32_t rptr_offset); uint32_t dmub_dcn401_get_current_time(struct dmub_srv *dmub); -void dmub_dcn401_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data); +void dmub_dcn401_get_diagnostic_data(struct dmub_srv *dmub); void dmub_dcn401_configure_dmub_in_system_memory(struct dmub_srv *dmub); void dmub_dcn401_send_inbox0_cmd(struct dmub_srv *dmub, union dmub_inbox0_data_register data); 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 6133d25da301..0a2285540daf 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c @@ -1099,11 +1099,11 @@ bool dmub_srv_get_outbox0_msg(struct dmub_srv *dmub, struct dmcub_trace_buf_entr return dmub_rb_out_trace_buffer_front(&dmub->outbox0_rb, (void *)entry); } -bool dmub_srv_get_diagnostic_data(struct dmub_srv *dmub, struct dmub_diagnostic_data *diag_data) +bool dmub_srv_get_diagnostic_data(struct dmub_srv *dmub) { - if (!dmub || !dmub->hw_funcs.get_diagnostic_data || !diag_data) + if (!dmub || !dmub->hw_funcs.get_diagnostic_data) return false; - dmub->hw_funcs.get_diagnostic_data(dmub, diag_data); + dmub->hw_funcs.get_diagnostic_data(dmub); return true; } -- 2.51.0 From 0747acf3311229e22009bec4a9e7fc30c879e842 Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Sat, 22 Feb 2025 23:37:32 -0600 Subject: [PATCH 13/16] drm/amd/display: fix default brightness [Why] To avoid flickering during boot default brightness level set by BIOS should be maintained for as much of the boot as feasible. commit 2fe87f54abdc ("drm/amd/display: Set default brightness according to ACPI") attempted to set the right levels for AC vs DC, but brightness still got reset to maximum level in initialization code for setup_backlight_device(). [How] Remove the hardcoded initialization in setup_backlight_device() and instead program brightness value to match BIOS (AC or DC). This avoids a brightness flicker from kernel changing the value. Userspace may however still change it during boot. Fixes: 2fe87f54abdc ("drm/amd/display: Set default brightness according to ACPI") Acked-by: Wayne Lin Signed-off-by: Mario Limonciello Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- 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 6a54f1cfa125..5ae521d1b74b 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -4986,6 +4986,7 @@ amdgpu_dm_register_backlight_device(struct amdgpu_dm_connector *aconnector) dm->backlight_dev[aconnector->bl_idx] = backlight_device_register(bl_name, aconnector->base.kdev, dm, &amdgpu_dm_backlight_ops, &props); + dm->brightness[aconnector->bl_idx] = props.brightness; if (IS_ERR(dm->backlight_dev[aconnector->bl_idx])) { DRM_ERROR("DM: Backlight registration failed!\n"); @@ -5053,7 +5054,6 @@ static void setup_backlight_device(struct amdgpu_display_manager *dm, aconnector->bl_idx = bl_idx; amdgpu_dm_update_backlight_caps(dm, bl_idx); - dm->brightness[bl_idx] = AMDGPU_MAX_BL_LEVEL; dm->backlight_link[bl_idx] = link; dm->num_of_edps++; -- 2.51.0 From 5e19e2b57b6bb640d68dfc7991e1e182922cf867 Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Sun, 23 Feb 2025 00:04:35 -0600 Subject: [PATCH 14/16] drm/amd/display: Restore correct backlight brightness after a GPU reset [Why] GPU reset will attempt to restore cached state, but brightness doesn't get restored. It will come back at 100% brightness, but userspace thinks it's the previous value. [How] When running resume sequence if GPU is in reset restore brightness to previous value. Acked-by: Wayne Lin Signed-off-by: Mario Limonciello Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 10 ++++++++++ 1 file changed, 10 insertions(+) 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 5ae521d1b74b..bf69cd50712e 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -250,6 +250,10 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, static void handle_hpd_irq_helper(struct amdgpu_dm_connector *aconnector); static void handle_hpd_rx_irq(void *param); +static void amdgpu_dm_backlight_set_level(struct amdgpu_display_manager *dm, + int bl_idx, + u32 user_brightness); + static bool is_timing_unchanged_for_freesync(struct drm_crtc_state *old_crtc_state, struct drm_crtc_state *new_crtc_state); @@ -3432,6 +3436,12 @@ static int dm_resume(struct amdgpu_ip_block *ip_block) mutex_unlock(&dm->dc_lock); + /* set the backlight after a reset */ + for (i = 0; i < dm->num_of_edps; i++) { + if (dm->backlight_dev[i]) + amdgpu_dm_backlight_set_level(dm, i, dm->brightness[i]); + } + return 0; } -- 2.51.0 From 50e0bae34fa6b8b18e13473ddf0bcdab6ab68310 Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Thu, 13 Feb 2025 16:26:22 -0600 Subject: [PATCH 15/16] drm/amd/display: Add and use new dm_prepare_suspend() callback [Why] The displays currently don't get turned off until after other IP blocks have been suspended. However turning off the displays first gives a very visible response that the system is on it's way down. [How] Turn off displays in a prepare_suspend() callback instead when possible. This will help for suspend and hibernate sequences. The shutdown sequence however will not call prepare() so check whether the state has been already saved to decide what to do. Acked-by: Wayne Lin Signed-off-by: Mario Limonciello Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 25 ++++++++++++++++--- 1 file changed, 21 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 bf69cd50712e..4f1cf8afe25f 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -3141,6 +3141,21 @@ static void hpd_rx_irq_work_suspend(struct amdgpu_display_manager *dm) } } +static int dm_prepare_suspend(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + + if (amdgpu_in_reset(adev)) + return 0; + + WARN_ON(adev->dm.cached_state); + adev->dm.cached_state = drm_atomic_helper_suspend(adev_to_drm(adev)); + if (IS_ERR(adev->dm.cached_state)) + return PTR_ERR(adev->dm.cached_state); + + return 0; +} + static int dm_suspend(struct amdgpu_ip_block *ip_block) { struct amdgpu_device *adev = ip_block->adev; @@ -3171,10 +3186,11 @@ static int dm_suspend(struct amdgpu_ip_block *ip_block) return 0; } - WARN_ON(adev->dm.cached_state); - adev->dm.cached_state = drm_atomic_helper_suspend(adev_to_drm(adev)); - if (IS_ERR(adev->dm.cached_state)) - return PTR_ERR(adev->dm.cached_state); + if (!adev->dm.cached_state) { + adev->dm.cached_state = drm_atomic_helper_suspend(adev_to_drm(adev)); + if (IS_ERR(adev->dm.cached_state)) + return PTR_ERR(adev->dm.cached_state); + } s3_handle_hdmi_cec(adev_to_drm(adev), true); @@ -3606,6 +3622,7 @@ static const struct amd_ip_funcs amdgpu_dm_funcs = { .early_fini = amdgpu_dm_early_fini, .hw_init = dm_hw_init, .hw_fini = dm_hw_fini, + .prepare_suspend = dm_prepare_suspend, .suspend = dm_suspend, .resume = dm_resume, .is_idle = dm_is_idle, -- 2.51.0 From bd00b29b5f236dce677089319176dee5872b5a7a Mon Sep 17 00:00:00 2001 From: Danny Wang Date: Thu, 13 Feb 2025 16:18:34 +0800 Subject: [PATCH 16/16] drm/amd/display: Do not enable replay when vtotal update is pending. [Why&How] Vtotal is not applied to HW when handling vsync interrupt. Make sure vtotal is aligned before enable replay. Reviewed-by: Anthony Koo Reviewed-by: Robin Chen Signed-off-by: Danny Wang Signed-off-by: Zhongwei Zhang Signed-off-by: Tom Chung Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 9 +++++++-- .../gpu/drm/amd/display/dc/core/dc_hw_sequencer.c | 15 +++++++++++++++ drivers/gpu/drm/amd/display/dc/dc_hw_types.h | 1 + .../drm/amd/display/dc/hwss/dce110/dce110_hwseq.c | 7 ++----- .../drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c | 7 ++----- .../drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c | 8 ++------ .../drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c | 4 +--- .../drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c | 3 +-- .../drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 10 +++------- .../gpu/drm/amd/display/dc/hwss/hw_sequencer.h | 6 ++++++ 10 files changed, 40 insertions(+), 30 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index e71ea21401f5..5a43e4901cc0 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -453,6 +453,7 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc, if (dc->caps.max_v_total != 0 && (adjust->v_total_max > dc->caps.max_v_total || adjust->v_total_min > dc->caps.max_v_total)) { + stream->adjust.timing_adjust_pending = false; if (adjust->allow_otg_v_count_halt) return set_long_vtotal(dc, stream, adjust); else @@ -466,7 +467,7 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc, dc->hwss.set_drr(&pipe, 1, *adjust); - + stream->adjust.timing_adjust_pending = false; return true; } } @@ -3165,8 +3166,12 @@ static void copy_stream_update_to_stream(struct dc *dc, if (update->vrr_active_fixed) stream->vrr_active_fixed = *update->vrr_active_fixed; - if (update->crtc_timing_adjust) + if (update->crtc_timing_adjust) { + if (stream->adjust.v_total_min != update->crtc_timing_adjust->v_total_min || + stream->adjust.v_total_max != update->crtc_timing_adjust->v_total_max) + stream->adjust.timing_adjust_pending = true; stream->adjust = *update->crtc_timing_adjust; + } if (update->dpms_off) stream->dpms_off = *update->dpms_off; diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c index e0277728268a..52ee2225e132 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c @@ -659,6 +659,21 @@ void set_p_state_switch_method( } } +void set_drr_and_clear_adjust_pending( + struct pipe_ctx *pipe_ctx, + struct dc_stream_state *stream, + struct drr_params *params) +{ + /* params can be null.*/ + if (pipe_ctx && pipe_ctx->stream_res.tg && + pipe_ctx->stream_res.tg->funcs->set_drr) + pipe_ctx->stream_res.tg->funcs->set_drr( + pipe_ctx->stream_res.tg, params); + + if (stream) + stream->adjust.timing_adjust_pending = false; +} + void get_fams2_visual_confirm_color( struct dc *dc, struct dc_state *context, diff --git a/drivers/gpu/drm/amd/display/dc/dc_hw_types.h b/drivers/gpu/drm/amd/display/dc/dc_hw_types.h index 9f3dd8824ed5..d562ddeca512 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_hw_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_hw_types.h @@ -1017,6 +1017,7 @@ struct dc_crtc_timing_adjust { uint32_t v_total_mid; uint32_t v_total_mid_frame_num; uint32_t allow_otg_v_count_halt; + uint8_t timing_adjust_pending; }; diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c index 9c9947fc5d44..bfd734e15731 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c @@ -1658,9 +1658,7 @@ enum dc_status dce110_apply_single_controller_ctx_to_hw( params.vertical_total_min = stream->adjust.v_total_min; params.vertical_total_max = stream->adjust.v_total_max; - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx, stream, ¶ms); // DRR should set trigger event to monitor surface update event if (stream->adjust.v_total_min != 0 && stream->adjust.v_total_max != 0) @@ -2109,8 +2107,7 @@ static void set_drr(struct pipe_ctx **pipe_ctx, struct timing_generator *tg = pipe_ctx[i]->stream_res.tg; if ((tg != NULL) && tg->funcs) { - if (tg->funcs->set_drr) - tg->funcs->set_drr(tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx[i], pipe_ctx[i]->stream, ¶ms); if (adjust.v_total_max != 0 && adjust.v_total_min != 0) if (tg->funcs->set_static_screen_control) tg->funcs->set_static_screen_control( diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c index 301ef36d3d05..912f96323ed6 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c @@ -1113,9 +1113,7 @@ static void dcn10_reset_back_end_for_pipe( 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); - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, NULL); + set_drr_and_clear_adjust_pending(pipe_ctx, pipe_ctx->stream, NULL); if (dc_is_hdmi_tmds_signal(pipe_ctx->stream->signal)) pipe_ctx->stream->link->phy_state.symclk_ref_cnts.otg = 0; } @@ -3218,8 +3216,7 @@ void dcn10_set_drr(struct pipe_ctx **pipe_ctx, struct timing_generator *tg = pipe_ctx[i]->stream_res.tg; if ((tg != NULL) && tg->funcs) { - if (tg->funcs->set_drr) - tg->funcs->set_drr(tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx[i], pipe_ctx[i]->stream, ¶ms); if (adjust.v_total_max != 0 && adjust.v_total_min != 0) if (tg->funcs->set_static_screen_control) tg->funcs->set_static_screen_control( diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c index a5a3e0823e21..926c08e790c1 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c @@ -952,9 +952,7 @@ enum dc_status dcn20_enable_stream_timing( params.vertical_total_max = stream->adjust.v_total_max; params.vertical_total_mid = stream->adjust.v_total_mid; params.vertical_total_mid_frame_num = stream->adjust.v_total_mid_frame_num; - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx, stream, ¶ms); // DRR should set trigger event to monitor surface update event if (stream->adjust.v_total_min != 0 && stream->adjust.v_total_max != 0) @@ -2856,9 +2854,7 @@ void dcn20_reset_back_end_for_pipe( pipe_ctx->stream_res.tg->funcs->set_odm_bypass( pipe_ctx->stream_res.tg, &pipe_ctx->stream->timing); - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, NULL); + set_drr_and_clear_adjust_pending(pipe_ctx, pipe_ctx->stream, NULL); /* TODO - convert symclk_ref_cnts for otg to a bit map to solve * the case where the same symclk is shared across multiple otg * instances diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c index 288e9dd9205d..f38340aa3f15 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c @@ -543,9 +543,7 @@ static void dcn31_reset_back_end_for_pipe( if (dc_is_hdmi_tmds_signal(pipe_ctx->stream->signal)) pipe_ctx->stream->link->phy_state.symclk_ref_cnts.otg = 0; - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, NULL); + set_drr_and_clear_adjust_pending(pipe_ctx, pipe_ctx->stream, NULL); /* DPMS may already disable or */ /* dpms_off status is incorrect due to fastboot diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c index b907ad1acedd..922b8d71cf1a 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c @@ -1473,8 +1473,7 @@ void dcn35_set_drr(struct pipe_ctx **pipe_ctx, num_frames = 2 * (frame_rate % 60); } } - if (tg->funcs->set_drr) - tg->funcs->set_drr(tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx[i], pipe_ctx[i]->stream, ¶ms); if (adjust.v_total_max != 0 && adjust.v_total_min != 0) if (tg->funcs->set_static_screen_control) tg->funcs->set_static_screen_control( diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 39668d8cc13a..8f5da0ded850 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -830,10 +830,7 @@ enum dc_status dcn401_enable_stream_timing( } hws->funcs.wait_for_blank_complete(pipe_ctx->stream_res.opp); - - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx, stream, ¶ms); /* Event triggers and num frames initialized for DRR, but can be * later updated for PSR use. Note DRR trigger events are generated @@ -1820,9 +1817,8 @@ void dcn401_reset_back_end_for_pipe( pipe_ctx->stream_res.tg->funcs->set_odm_bypass( pipe_ctx->stream_res.tg, &pipe_ctx->stream->timing); - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, NULL); + set_drr_and_clear_adjust_pending(pipe_ctx, pipe_ctx->stream, NULL); + /* TODO - convert symclk_ref_cnts for otg to a bit map to solve * the case where the same symclk is shared across multiple otg * instances diff --git a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h index 2b1a2a00648a..c8b5ed834579 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h @@ -46,6 +46,7 @@ struct dce_hwseq; struct link_resource; struct dc_dmub_cmd; struct pg_block_update; +struct drr_params; struct subvp_pipe_control_lock_fast_params { struct dc *dc; @@ -527,6 +528,11 @@ void set_p_state_switch_method( struct dc_state *context, struct pipe_ctx *pipe_ctx); +void set_drr_and_clear_adjust_pending( + struct pipe_ctx *pipe_ctx, + struct dc_stream_state *stream, + struct drr_params *params); + void hwss_execute_sequence(struct dc *dc, struct block_sequence block_sequence[], int num_steps); -- 2.51.0