diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
107 files changed, 3725 insertions, 1071 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 39976c7b100c..6fde9a9d5e2b 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -55,7 +55,8 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \ amdgpu_vf_error.o amdgpu_sched.o amdgpu_debugfs.o amdgpu_ids.o \ amdgpu_gmc.o amdgpu_mmhub.o amdgpu_xgmi.o amdgpu_csa.o amdgpu_ras.o amdgpu_vm_cpu.o \ amdgpu_vm_sdma.o amdgpu_discovery.o amdgpu_ras_eeprom.o amdgpu_nbio.o \ - amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o + amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o \ + amdgpu_fw_attestation.o amdgpu-$(CONFIG_PERF_EVENTS) += amdgpu_pmu.o @@ -69,7 +70,8 @@ amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce amdgpu-y += \ vi.o mxgpu_vi.o nbio_v6_1.o soc15.o emu_soc.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o \ vega20_reg_init.o nbio_v7_4.o nbio_v2_3.o nv.o navi10_reg_init.o navi14_reg_init.o \ - arct_reg_init.o navi12_reg_init.o mxgpu_nv.o sienna_cichlid_reg_init.o + arct_reg_init.o navi12_reg_init.o mxgpu_nv.o sienna_cichlid_reg_init.o vangogh_reg_init.o \ + nbio_v7_2.o dimgrey_cavefish_reg_init.o # add DF block amdgpu-y += \ @@ -81,7 +83,7 @@ amdgpu-y += \ gmc_v7_0.o \ gmc_v8_0.o \ gfxhub_v1_0.o mmhub_v1_0.o gmc_v9_0.o gfxhub_v1_1.o mmhub_v9_4.o \ - gfxhub_v2_0.o mmhub_v2_0.o gmc_v10_0.o gfxhub_v2_1.o + gfxhub_v2_0.o mmhub_v2_0.o gmc_v10_0.o gfxhub_v2_1.o mmhub_v2_3.o # add UMC block amdgpu-y += \ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 87f095dc385c..183b09d71b64 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -623,6 +623,8 @@ struct amdgpu_asic_funcs { bool (*supports_baco)(struct amdgpu_device *adev); /* pre asic_init quirks */ void (*pre_asic_init)(struct amdgpu_device *adev); + /* enter/exit umd stable pstate */ + int (*update_umd_stable_pstate)(struct amdgpu_device *adev, bool enter); }; /* @@ -723,6 +725,45 @@ struct amd_powerplay { const struct amd_pm_funcs *pp_funcs; }; +/* polaris10 kickers */ +#define ASICID_IS_P20(did, rid) (((did == 0x67DF) && \ + ((rid == 0xE3) || \ + (rid == 0xE4) || \ + (rid == 0xE5) || \ + (rid == 0xE7) || \ + (rid == 0xEF))) || \ + ((did == 0x6FDF) && \ + ((rid == 0xE7) || \ + (rid == 0xEF) || \ + (rid == 0xFF)))) + +#define ASICID_IS_P30(did, rid) ((did == 0x67DF) && \ + ((rid == 0xE1) || \ + (rid == 0xF7))) + +/* polaris11 kickers */ +#define ASICID_IS_P21(did, rid) (((did == 0x67EF) && \ + ((rid == 0xE0) || \ + (rid == 0xE5))) || \ + ((did == 0x67FF) && \ + ((rid == 0xCF) || \ + (rid == 0xEF) || \ + (rid == 0xFF)))) + +#define ASICID_IS_P31(did, rid) ((did == 0x67EF) && \ + ((rid == 0xE2))) + +/* polaris12 kickers */ +#define ASICID_IS_P23(did, rid) (((did == 0x6987) && \ + ((rid == 0xC0) || \ + (rid == 0xC1) || \ + (rid == 0xC3) || \ + (rid == 0xC7))) || \ + ((did == 0x6981) && \ + ((rid == 0x00) || \ + (rid == 0x01) || \ + (rid == 0x10)))) + #define AMDGPU_RESET_MAGIC_NUM 64 #define AMDGPU_MAX_DF_PERFMONS 4 struct amdgpu_device { @@ -1165,6 +1206,8 @@ int emu_soc_asic_init(struct amdgpu_device *adev); #define amdgpu_asic_get_pcie_replay_count(adev) ((adev)->asic_funcs->get_pcie_replay_count((adev))) #define amdgpu_asic_supports_baco(adev) (adev)->asic_funcs->supports_baco((adev)) #define amdgpu_asic_pre_asic_init(adev) (adev)->asic_funcs->pre_asic_init((adev)) +#define amdgpu_asic_update_umd_stable_pstate(adev, enter) \ + ((adev)->asic_funcs->update_umd_stable_pstate ? (adev)->asic_funcs->update_umd_stable_pstate((adev), (enter)) : 0) #define amdgpu_inc_vram_lost(adev) atomic_inc(&((adev)->vram_lost_counter)); @@ -1294,19 +1337,6 @@ bool amdgpu_device_load_pci_state(struct pci_dev *pdev); #include "amdgpu_object.h" -/* used by df_v3_6.c and amdgpu_pmu.c */ -#define AMDGPU_PMU_ATTR(_name, _object) \ -static ssize_t \ -_name##_show(struct device *dev, \ - struct device_attribute *attr, \ - char *page) \ -{ \ - BUILD_BUG_ON(sizeof(_object) >= PAGE_SIZE - 1); \ - return sprintf(page, _object "\n"); \ -} \ - \ -static struct device_attribute pmu_attr_##_name = __ATTR_RO(_name) - static inline bool amdgpu_is_tmz(struct amdgpu_device *adev) { return adev->gmc.tmz_enabled; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c index d3e51d361179..1400957034a1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c @@ -36,17 +36,17 @@ #include "acp_gfx_if.h" -#define ACP_TILE_ON_MASK 0x03 -#define ACP_TILE_OFF_MASK 0x02 -#define ACP_TILE_ON_RETAIN_REG_MASK 0x1f -#define ACP_TILE_OFF_RETAIN_REG_MASK 0x20 +#define ACP_TILE_ON_MASK 0x03 +#define ACP_TILE_OFF_MASK 0x02 +#define ACP_TILE_ON_RETAIN_REG_MASK 0x1f +#define ACP_TILE_OFF_RETAIN_REG_MASK 0x20 -#define ACP_TILE_P1_MASK 0x3e -#define ACP_TILE_P2_MASK 0x3d -#define ACP_TILE_DSP0_MASK 0x3b -#define ACP_TILE_DSP1_MASK 0x37 +#define ACP_TILE_P1_MASK 0x3e +#define ACP_TILE_P2_MASK 0x3d +#define ACP_TILE_DSP0_MASK 0x3b +#define ACP_TILE_DSP1_MASK 0x37 -#define ACP_TILE_DSP2_MASK 0x2f +#define ACP_TILE_DSP2_MASK 0x2f #define ACP_DMA_REGS_END 0x146c0 #define ACP_I2S_PLAY_REGS_START 0x14840 @@ -75,8 +75,8 @@ #define mmACP_CONTROL 0x5131 #define mmACP_STATUS 0x5133 #define mmACP_SOFT_RESET 0x5134 -#define ACP_CONTROL__ClkEn_MASK 0x1 -#define ACP_SOFT_RESET__SoftResetAud_MASK 0x100 +#define ACP_CONTROL__ClkEn_MASK 0x1 +#define ACP_SOFT_RESET__SoftResetAud_MASK 0x100 #define ACP_SOFT_RESET__SoftResetAudDone_MASK 0x1000000 #define ACP_CLOCK_EN_TIME_OUT_VALUE 0x000000FF #define ACP_SOFT_RESET_DONE_TIME_OUT_VALUE 0x000000FF diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index 0544460653b9..db96d69eb45e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -390,23 +390,17 @@ void amdgpu_amdkfd_get_local_mem_info(struct kgd_dev *kgd, struct kfd_local_mem_info *mem_info) { struct amdgpu_device *adev = (struct amdgpu_device *)kgd; - uint64_t address_mask = adev->dev->dma_mask ? ~*adev->dev->dma_mask : - ~((1ULL << 32) - 1); - resource_size_t aper_limit = adev->gmc.aper_base + adev->gmc.aper_size; memset(mem_info, 0, sizeof(*mem_info)); - if (!(adev->gmc.aper_base & address_mask || aper_limit & address_mask)) { - mem_info->local_mem_size_public = adev->gmc.visible_vram_size; - mem_info->local_mem_size_private = adev->gmc.real_vram_size - - adev->gmc.visible_vram_size; - } else { - mem_info->local_mem_size_public = 0; - mem_info->local_mem_size_private = adev->gmc.real_vram_size; - } + + mem_info->local_mem_size_public = adev->gmc.visible_vram_size; + mem_info->local_mem_size_private = adev->gmc.real_vram_size - + adev->gmc.visible_vram_size; + mem_info->vram_width = adev->gmc.vram_width; - pr_debug("Address base: %pap limit %pap public 0x%llx private 0x%llx\n", - &adev->gmc.aper_base, &aper_limit, + pr_debug("Address base: %pap public 0x%llx private 0x%llx\n", + &adev->gmc.aper_base, mem_info->local_mem_size_public, mem_info->local_mem_size_private); @@ -648,6 +642,13 @@ void amdgpu_amdkfd_set_compute_idle(struct kgd_dev *kgd, bool idle) { struct amdgpu_device *adev = (struct amdgpu_device *)kgd; + /* Temp workaround to fix the soft hang observed in certain compute + * applications if GFXOFF is enabled. + */ + if (adev->asic_type == CHIP_SIENNA_CICHLID) { + pr_debug("GFXOFF is %s\n", idle ? "enabled" : "disabled"); + amdgpu_gfx_off_ctrl(adev, idle); + } amdgpu_dpm_switch_power_profile(adev, PP_SMC_POWER_PROFILE_COMPUTE, !idle); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c index 1afa8f122e7d..604757a1e440 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c @@ -304,4 +304,5 @@ const struct kfd2kgd_calls arcturus_kfd2kgd = { kgd_gfx_v9_get_atc_vmid_pasid_mapping_info, .set_vm_context_page_table_base = kgd_gfx_v9_set_vm_context_page_table_base, + .get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c index 43b18863a8b8..7d6c0013af35 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c @@ -799,7 +799,7 @@ static void get_wave_count(struct amdgpu_device *adev, int queue_idx, * * Reading registers referenced above involves programming GRBM appropriately */ -static void kgd_gfx_v9_get_cu_occupancy(struct kgd_dev *kgd, int pasid, +void kgd_gfx_v9_get_cu_occupancy(struct kgd_dev *kgd, int pasid, int *pasid_wave_cnt, int *max_waves_per_cu) { int qidx; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h index fc8934b86d93..e64deba8900f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h @@ -63,3 +63,5 @@ bool kgd_gfx_v9_get_atc_vmid_pasid_mapping_info(struct kgd_dev *kgd, void kgd_gfx_v9_set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid, uint64_t page_table_base); +void kgd_gfx_v9_get_cu_occupancy(struct kgd_dev *kgd, int pasid, + int *pasid_wave_cnt, int *max_waves_per_cu); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 054a1c2d5054..1755386470e6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -239,8 +239,7 @@ static int amdgpu_amdkfd_remove_eviction_fence(struct amdgpu_bo *bo, if (!old) return 0; - new = kmalloc(offsetof(typeof(*new), shared[old->shared_max]), - GFP_KERNEL); + new = kmalloc(struct_size(new, shared, old->shared_max), GFP_KERNEL); if (!new) return -ENOMEM; @@ -1115,19 +1114,19 @@ void amdgpu_amdkfd_gpuvm_destroy_process_vm(struct kgd_dev *kgd, void *vm) void amdgpu_amdkfd_gpuvm_release_process_vm(struct kgd_dev *kgd, void *vm) { struct amdgpu_device *adev = get_amdgpu_device(kgd); - struct amdgpu_vm *avm = (struct amdgpu_vm *)vm; + struct amdgpu_vm *avm = (struct amdgpu_vm *)vm; if (WARN_ON(!kgd || !vm)) - return; + return; - pr_debug("Releasing process vm %p\n", vm); + pr_debug("Releasing process vm %p\n", vm); - /* The original pasid of amdgpu vm has already been - * released during making a amdgpu vm to a compute vm - * The current pasid is managed by kfd and will be - * released on kfd process destroy. Set amdgpu pasid - * to 0 to avoid duplicate release. - */ + /* The original pasid of amdgpu vm has already been + * released during making a amdgpu vm to a compute vm + * The current pasid is managed by kfd and will be + * released on kfd process destroy. Set amdgpu pasid + * to 0 to avoid duplicate release. + */ amdgpu_vm_release_compute(adev, avm); } @@ -1288,7 +1287,7 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu( struct ttm_validate_buffer *bo_list_entry; unsigned int mapped_to_gpu_memory; int ret; - bool is_imported = 0; + bool is_imported = false; mutex_lock(&mem->lock); mapped_to_gpu_memory = mem->mapped_to_gpu_memory; @@ -1695,7 +1694,7 @@ int amdgpu_amdkfd_gpuvm_import_dmabuf(struct kgd_dev *kgd, INIT_LIST_HEAD(&(*mem)->bo_va_list); mutex_init(&(*mem)->lock); - + (*mem)->alloc_flags = ((bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) ? KFD_IOC_ALLOC_MEM_FLAGS_VRAM : KFD_IOC_ALLOC_MEM_FLAGS_GTT) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c index 469352e2d6ec..86add0f4ea4d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c @@ -1401,7 +1401,7 @@ static ATOM_VOLTAGE_OBJECT_V3 *amdgpu_atombios_lookup_voltage_object_v3(ATOM_VOL { u32 size = le16_to_cpu(v3->sHeader.usStructureSize); u32 offset = offsetof(ATOM_VOLTAGE_OBJECT_INFO_V3_1, asVoltageObj[0]); - u8 *start = (u8*)v3; + u8 *start = (u8 *)v3; while (offset < size) { ATOM_VOLTAGE_OBJECT_V3 *vo = (ATOM_VOLTAGE_OBJECT_V3 *)(start + offset); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c index b4df6460e45a..306077884a67 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c @@ -70,7 +70,7 @@ int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev) struct atom_context *ctx = adev->mode_info.atom_context; int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, vram_usagebyfirmware); - struct vram_usagebyfirmware_v2_1 * firmware_usage; + struct vram_usagebyfirmware_v2_1 *firmware_usage; uint32_t start_addr, size; uint16_t data_offset; int usage_bytes = 0; @@ -149,6 +149,10 @@ static int convert_atom_mem_type_to_vram_type(struct amdgpu_device *adev, case LpDdr4MemType: vram_type = AMDGPU_VRAM_TYPE_DDR4; break; + case Ddr5MemType: + case LpDdr5MemType: + vram_type = AMDGPU_VRAM_TYPE_DDR5; + break; default: vram_type = AMDGPU_VRAM_TYPE_UNKNOWN; break; @@ -544,6 +548,7 @@ int amdgpu_mem_train_support(struct amdgpu_device *adev) case HW_REV(11, 0, 5): case HW_REV(11, 0, 7): case HW_REV(11, 0, 11): + case HW_REV(11, 0, 12): ret = 1; break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index 78ac6dbe70d8..f1a050379190 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -352,17 +352,10 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device, break; case CHIP_POLARIS11: if (type == CGS_UCODE_ID_SMU) { - if (((adev->pdev->device == 0x67ef) && - ((adev->pdev->revision == 0xe0) || - (adev->pdev->revision == 0xe5))) || - ((adev->pdev->device == 0x67ff) && - ((adev->pdev->revision == 0xcf) || - (adev->pdev->revision == 0xef) || - (adev->pdev->revision == 0xff)))) { + if (ASICID_IS_P21(adev->pdev->device, adev->pdev->revision)) { info->is_kicker = true; strcpy(fw_name, "amdgpu/polaris11_k_smc.bin"); - } else if ((adev->pdev->device == 0x67ef) && - (adev->pdev->revision == 0xe2)) { + } else if (ASICID_IS_P31(adev->pdev->device, adev->pdev->revision)) { info->is_kicker = true; strcpy(fw_name, "amdgpu/polaris11_k2_smc.bin"); } else { @@ -374,21 +367,10 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device, break; case CHIP_POLARIS10: if (type == CGS_UCODE_ID_SMU) { - if (((adev->pdev->device == 0x67df) && - ((adev->pdev->revision == 0xe0) || - (adev->pdev->revision == 0xe3) || - (adev->pdev->revision == 0xe4) || - (adev->pdev->revision == 0xe5) || - (adev->pdev->revision == 0xe7) || - (adev->pdev->revision == 0xef))) || - ((adev->pdev->device == 0x6fdf) && - ((adev->pdev->revision == 0xef) || - (adev->pdev->revision == 0xff)))) { + if (ASICID_IS_P20(adev->pdev->device, adev->pdev->revision)) { info->is_kicker = true; strcpy(fw_name, "amdgpu/polaris10_k_smc.bin"); - } else if ((adev->pdev->device == 0x67df) && - ((adev->pdev->revision == 0xe1) || - (adev->pdev->revision == 0xf7))) { + } else if (ASICID_IS_P30(adev->pdev->device, adev->pdev->revision)) { info->is_kicker = true; strcpy(fw_name, "amdgpu/polaris10_k2_smc.bin"); } else { @@ -399,13 +381,7 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device, } break; case CHIP_POLARIS12: - if (((adev->pdev->device == 0x6987) && - ((adev->pdev->revision == 0xc0) || - (adev->pdev->revision == 0xc3))) || - ((adev->pdev->device == 0x6981) && - ((adev->pdev->revision == 0x00) || - (adev->pdev->revision == 0x01) || - (adev->pdev->revision == 0x10)))) { + if (ASICID_IS_P23(adev->pdev->device, adev->pdev->revision)) { info->is_kicker = true; strcpy(fw_name, "amdgpu/polaris12_k_smc.bin"); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 8466558d0d93..8d2878e950da 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -326,7 +326,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev, increment_us = time_us - adev->mm_stats.last_update_us; adev->mm_stats.last_update_us = time_us; adev->mm_stats.accum_us = min(adev->mm_stats.accum_us + increment_us, - us_upper_bound); + us_upper_bound); /* This prevents the short period of low performance when the VRAM * usage is low and the driver is in debt or doesn't have enough @@ -1460,7 +1460,7 @@ int amdgpu_cs_fence_to_handle_ioctl(struct drm_device *dev, void *data, dma_fence_put(fence); if (r) return r; - r = drm_syncobj_get_fd(syncobj, (int*)&info->out.handle); + r = drm_syncobj_get_fd(syncobj, (int *)&info->out.handle); drm_syncobj_put(syncobj); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index c80d8339f58c..0350205c4897 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -100,7 +100,7 @@ static int amdgpu_ctx_init_entity(struct amdgpu_ctx *ctx, u32 hw_ip, enum drm_sched_priority priority; int r; - entity = kcalloc(1, offsetof(typeof(*entity), fences[amdgpu_sched_jobs]), + entity = kzalloc(struct_size(entity, fences, amdgpu_sched_jobs), GFP_KERNEL); if (!entity) return -ENOMEM; @@ -450,7 +450,7 @@ int amdgpu_ctx_put(struct amdgpu_ctx *ctx) void amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, struct drm_sched_entity *entity, - struct dma_fence *fence, uint64_t* handle) + struct dma_fence *fence, uint64_t *handle) { struct amdgpu_ctx_entity *centity = to_amdgpu_ctx_entity(entity); uint64_t seq = centity->sequence; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c index 065937482239..5c1f3725c741 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c @@ -35,6 +35,7 @@ #include "amdgpu_dm_debugfs.h" #include "amdgpu_ras.h" #include "amdgpu_rap.h" +#include "amdgpu_fw_attestation.h" /** * amdgpu_debugfs_add_files - Add simple debugfs entries @@ -169,14 +170,14 @@ static void amdgpu_debugfs_autodump_init(struct amdgpu_device *adev) * * Bit 62: Indicates a GRBM bank switch is needed * Bit 61: Indicates a SRBM bank switch is needed (implies bit 62 is - * zero) + * zero) * Bits 24..33: The SE or ME selector if needed * Bits 34..43: The SH (or SA) or PIPE selector if needed * Bits 44..53: The INSTANCE (or CU/WGP) or QUEUE selector if needed * * Bit 23: Indicates that the PM power gating lock should be held - * This is necessary to read registers that might be - * unreliable during a power gating transistion. + * This is necessary to read registers that might be + * unreliable during a power gating transistion. * * The lower bits are the BYTE offset of the register to read. This * allows reading multiple registers in a single call and having @@ -864,7 +865,7 @@ static ssize_t amdgpu_debugfs_wave_read(struct file *f, char __user *buf, { struct amdgpu_device *adev = f->f_inode->i_private; int r, x; - ssize_t result=0; + ssize_t result = 0; uint32_t offset, se, sh, cu, wave, simd, data[32]; if (size & 3 || *pos & 3) @@ -1210,7 +1211,7 @@ static const char *debugfs_regs_names[] = { /** * amdgpu_debugfs_regs_init - Initialize debugfs entries that provide - * register access. + * register access. * * @adev: The device to attach the debugfs entries to */ @@ -1338,11 +1339,41 @@ static int amdgpu_debugfs_evict_gtt(struct seq_file *m, void *data) return 0; } +static int amdgpu_debugfs_vm_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *)m->private; + struct drm_device *dev = node->minor->dev; + struct drm_file *file; + int r; + + r = mutex_lock_interruptible(&dev->filelist_mutex); + if (r) + return r; + + list_for_each_entry(file, &dev->filelist, lhead) { + struct amdgpu_fpriv *fpriv = file->driver_priv; + struct amdgpu_vm *vm = &fpriv->vm; + + seq_printf(m, "pid:%d\tProcess:%s ----------\n", + vm->task_info.pid, vm->task_info.process_name); + r = amdgpu_bo_reserve(vm->root.base.bo, true); + if (r) + break; + amdgpu_debugfs_vm_bo_info(vm, m); + amdgpu_bo_unreserve(vm->root.base.bo); + } + + mutex_unlock(&dev->filelist_mutex); + + return r; +} + static const struct drm_info_list amdgpu_debugfs_list[] = { {"amdgpu_vbios", amdgpu_debugfs_get_vbios_dump}, {"amdgpu_test_ib", &amdgpu_debugfs_test_ib}, {"amdgpu_evict_vram", &amdgpu_debugfs_evict_vram}, {"amdgpu_evict_gtt", &amdgpu_debugfs_evict_gtt}, + {"amdgpu_vm_info", &amdgpu_debugfs_vm_info}, }; static void amdgpu_ib_preempt_fences_swap(struct amdgpu_ring *ring, @@ -1638,6 +1669,8 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev) amdgpu_rap_debugfs_init(adev); + amdgpu_fw_attestation_debugfs_init(adev); + return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_list, ARRAY_SIZE(amdgpu_debugfs_list)); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 37da3537ba2e..7560b05e4ac1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -80,6 +80,8 @@ MODULE_FIRMWARE("amdgpu/renoir_gpu_info.bin"); MODULE_FIRMWARE("amdgpu/navi10_gpu_info.bin"); MODULE_FIRMWARE("amdgpu/navi14_gpu_info.bin"); MODULE_FIRMWARE("amdgpu/navi12_gpu_info.bin"); +MODULE_FIRMWARE("amdgpu/vangogh_gpu_info.bin"); +MODULE_FIRMWARE("amdgpu/green_sardine_gpu_info.bin"); #define AMDGPU_RESUME_MS 2000 @@ -114,6 +116,8 @@ const char *amdgpu_asic_name[] = { "NAVI12", "SIENNA_CICHLID", "NAVY_FLOUNDER", + "VANGOGH", + "DIMGREY_CAVEFISH", "LAST", }; @@ -239,9 +243,11 @@ bool amdgpu_device_supports_baco(struct drm_device *dev) return amdgpu_asic_supports_baco(adev); } +/* + * VRAM access helper functions + */ + /** - * VRAM access helper functions. - * * amdgpu_device_vram_access - read/write a buffer in vram * * @adev: amdgpu_device pointer @@ -705,7 +711,7 @@ void amdgpu_device_indirect_wreg64(struct amdgpu_device *adev, /** * amdgpu_invalid_rreg - dummy reg read function * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * @reg: offset of register * * Dummy register read function. Used for register blocks @@ -722,7 +728,7 @@ static uint32_t amdgpu_invalid_rreg(struct amdgpu_device *adev, uint32_t reg) /** * amdgpu_invalid_wreg - dummy reg write function * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * @reg: offset of register * @v: value to write to the register * @@ -739,7 +745,7 @@ static void amdgpu_invalid_wreg(struct amdgpu_device *adev, uint32_t reg, uint32 /** * amdgpu_invalid_rreg64 - dummy 64 bit reg read function * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * @reg: offset of register * * Dummy register read function. Used for register blocks @@ -756,7 +762,7 @@ static uint64_t amdgpu_invalid_rreg64(struct amdgpu_device *adev, uint32_t reg) /** * amdgpu_invalid_wreg64 - dummy reg write function * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * @reg: offset of register * @v: value to write to the register * @@ -773,7 +779,7 @@ static void amdgpu_invalid_wreg64(struct amdgpu_device *adev, uint32_t reg, uint /** * amdgpu_block_invalid_rreg - dummy reg read function * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * @block: offset of instance * @reg: offset of register * @@ -793,7 +799,7 @@ static uint32_t amdgpu_block_invalid_rreg(struct amdgpu_device *adev, /** * amdgpu_block_invalid_wreg - dummy reg write function * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * @block: offset of instance * @reg: offset of register * @v: value to write to the register @@ -813,7 +819,7 @@ static void amdgpu_block_invalid_wreg(struct amdgpu_device *adev, /** * amdgpu_device_asic_init - Wrapper for atom asic_init * - * @dev: drm_device pointer + * @adev: amdgpu_device pointer * * Does any asic specific work and then calls atom asic init. */ @@ -827,7 +833,7 @@ static int amdgpu_device_asic_init(struct amdgpu_device *adev) /** * amdgpu_device_vram_scratch_init - allocate the VRAM scratch page * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * * Allocates a scratch page of VRAM for use by various things in the * driver. @@ -844,7 +850,7 @@ static int amdgpu_device_vram_scratch_init(struct amdgpu_device *adev) /** * amdgpu_device_vram_scratch_fini - Free the VRAM scratch page * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * * Frees the VRAM scratch page. */ @@ -1370,13 +1376,6 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev) amdgpu_gmc_tmz_set(adev); - if (amdgpu_num_kcq == -1) { - amdgpu_num_kcq = 8; - } else if (amdgpu_num_kcq > 8 || amdgpu_num_kcq < 0) { - amdgpu_num_kcq = 8; - dev_warn(adev->dev, "set kernel compute queue number to 8 due to invalid parameter provided by user\n"); - } - amdgpu_gmc_noretry_set(adev); return 0; @@ -1783,6 +1782,7 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev) case CHIP_VEGA20: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: default: return 0; case CHIP_VEGA10: @@ -1803,7 +1803,10 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev) chip_name = "arcturus"; break; case CHIP_RENOIR: - chip_name = "renoir"; + if (adev->apu_flags & AMD_APU_IS_RENOIR) + chip_name = "renoir"; + else + chip_name = "green_sardine"; break; case CHIP_NAVI10: chip_name = "navi10"; @@ -1814,6 +1817,9 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev) case CHIP_NAVI12: chip_name = "navi12"; break; + case CHIP_VANGOGH: + chip_name = "vangogh"; + break; } snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_gpu_info.bin", chip_name); @@ -1988,7 +1994,12 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: - adev->family = AMDGPU_FAMILY_NV; + case CHIP_DIMGREY_CAVEFISH: + case CHIP_VANGOGH: + if (adev->asic_type == CHIP_VANGOGH) + adev->family = AMDGPU_FAMILY_VGH; + else + adev->family = AMDGPU_FAMILY_NV; r = nv_set_ip_blocks(adev); if (r) @@ -2993,10 +3004,10 @@ bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type) case CHIP_NAVI14: case CHIP_NAVI12: case CHIP_RENOIR: -#endif -#if defined(CONFIG_DRM_AMD_DC_DCN3_0) case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: + case CHIP_VANGOGH: #endif return amdgpu_dc != 0; #endif @@ -3011,7 +3022,7 @@ bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type) /** * amdgpu_device_has_dc_support - check if dc is supported * - * @adev: amdgpu_device_pointer + * @adev: amdgpu_device pointer * * Returns true for supported, false for not supported */ @@ -4045,7 +4056,7 @@ static int amdgpu_device_recover_vram(struct amdgpu_device *adev) /** * amdgpu_device_reset_sriov - reset ASIC for SR-IOV vf * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * @from_hypervisor: request from hypervisor * * do VF FLR and reinitialize Asic @@ -4100,7 +4111,7 @@ error: /** * amdgpu_device_has_job_running - check if there is any job in mirror list * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * * check if there is any job in mirror list */ @@ -4128,7 +4139,7 @@ bool amdgpu_device_has_job_running(struct amdgpu_device *adev) /** * amdgpu_device_should_recover_gpu - check if we should try GPU recovery * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * * Check amdgpu_gpu_recovery and SRIOV status to see if we should try to recover * a hung GPU. @@ -4477,7 +4488,7 @@ static int amdgpu_device_suspend_display_audio(struct amdgpu_device *adev) /** * amdgpu_device_gpu_recover - reset the asic and recover scheduler * - * @adev: amdgpu device pointer + * @adev: amdgpu_device pointer * @job: which job trigger hang * * Attempt to reset the GPU if it has hung (all asics). @@ -4497,7 +4508,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, bool need_emergency_restart = false; bool audio_suspended = false; - /** + /* * Special case: RAS triggered and full reset isn't supported */ need_emergency_restart = amdgpu_ras_need_emergency_restart(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h index 373cdebe0e2f..52488bb45112 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h @@ -44,11 +44,11 @@ struct amdgpu_df_funcs { void (*enable_ecc_force_par_wr_rmw)(struct amdgpu_device *adev, bool enable); int (*pmc_start)(struct amdgpu_device *adev, uint64_t config, - int is_add); + int counter_idx, int is_add); int (*pmc_stop)(struct amdgpu_device *adev, uint64_t config, - int is_remove); + int counter_idx, int is_remove); void (*pmc_get_count)(struct amdgpu_device *adev, uint64_t config, - uint64_t *count); + int counter_idx, uint64_t *count); uint64_t (*get_fica)(struct amdgpu_device *adev, uint32_t ficaa_val); void (*set_fica)(struct amdgpu_device *adev, uint32_t ficaa_val, uint32_t ficadl_val, uint32_t ficadh_val); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index b25faaee6f0e..2e8a8b57639f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -38,6 +38,7 @@ #include <drm/drm_edid.h> #include <drm/drm_gem_framebuffer_helper.h> #include <drm/drm_fb_helper.h> +#include <drm/drm_fourcc.h> #include <drm/drm_vblank.h> static void amdgpu_display_flip_callback(struct dma_fence *f, @@ -524,6 +525,10 @@ uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev, (adev->apu_flags & AMD_APU_IS_PICASSO)) domain |= AMDGPU_GEM_DOMAIN_GTT; break; + case CHIP_RENOIR: + domain |= AMDGPU_GEM_DOMAIN_GTT; + break; + default: break; } @@ -533,6 +538,154 @@ uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev, return domain; } +static int convert_tiling_flags_to_modifier(struct amdgpu_framebuffer *afb) +{ + struct amdgpu_device *adev = drm_to_adev(afb->base.dev); + uint64_t modifier = 0; + + if (!afb->tiling_flags || !AMDGPU_TILING_GET(afb->tiling_flags, SWIZZLE_MODE)) { + modifier = DRM_FORMAT_MOD_LINEAR; + } else { + int swizzle = AMDGPU_TILING_GET(afb->tiling_flags, SWIZZLE_MODE); + bool has_xor = swizzle >= 16; + int block_size_bits; + int version; + int pipe_xor_bits = 0; + int bank_xor_bits = 0; + int packers = 0; + uint32_t dcc_offset = AMDGPU_TILING_GET(afb->tiling_flags, DCC_OFFSET_256B); + + switch (swizzle >> 2) { + case 0: /* 256B */ + block_size_bits = 8; + break; + case 1: /* 4KiB */ + case 5: /* 4KiB _X */ + block_size_bits = 12; + break; + case 2: /* 64KiB */ + case 4: /* 64 KiB _T */ + case 6: /* 64 KiB _X */ + block_size_bits = 16; + break; + default: + /* RESERVED or VAR */ + return -EINVAL; + } + + if (adev->asic_type >= CHIP_SIENNA_CICHLID) + version = AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS; + else if (adev->family == AMDGPU_FAMILY_NV) + version = AMD_FMT_MOD_TILE_VER_GFX10; + else + version = AMD_FMT_MOD_TILE_VER_GFX9; + + switch (swizzle & 3) { + case 0: /* Z microtiling */ + return -EINVAL; + case 1: /* S microtiling */ + if (!has_xor) + version = AMD_FMT_MOD_TILE_VER_GFX9; + break; + case 2: + if (!has_xor && afb->base.format->cpp[0] != 4) + version = AMD_FMT_MOD_TILE_VER_GFX9; + break; + case 3: + break; + } + + if (has_xor) { + switch (version) { + case AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS: + pipe_xor_bits = min(block_size_bits - 8, + ilog2(adev->gfx.config.gb_addr_config_fields.num_pipes)); + packers = min(block_size_bits - 8 - pipe_xor_bits, + ilog2(adev->gfx.config.gb_addr_config_fields.num_pkrs)); + break; + case AMD_FMT_MOD_TILE_VER_GFX10: + pipe_xor_bits = min(block_size_bits - 8, + ilog2(adev->gfx.config.gb_addr_config_fields.num_pipes)); + break; + case AMD_FMT_MOD_TILE_VER_GFX9: + pipe_xor_bits = min(block_size_bits - 8, + ilog2(adev->gfx.config.gb_addr_config_fields.num_pipes) + + ilog2(adev->gfx.config.gb_addr_config_fields.num_se)); + bank_xor_bits = min(block_size_bits - 8 - pipe_xor_bits, + ilog2(adev->gfx.config.gb_addr_config_fields.num_banks)); + break; + } + } + + modifier = AMD_FMT_MOD | + AMD_FMT_MOD_SET(TILE, AMDGPU_TILING_GET(afb->tiling_flags, SWIZZLE_MODE)) | + AMD_FMT_MOD_SET(TILE_VERSION, version) | + AMD_FMT_MOD_SET(PIPE_XOR_BITS, pipe_xor_bits) | + AMD_FMT_MOD_SET(BANK_XOR_BITS, bank_xor_bits) | + AMD_FMT_MOD_SET(PACKERS, packers); + + if (dcc_offset != 0) { + bool dcc_i64b = AMDGPU_TILING_GET(afb->tiling_flags, DCC_INDEPENDENT_64B) != 0; + bool dcc_i128b = version >= AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS; + + /* Enable constant encode on RAVEN2 and later. */ + bool dcc_constant_encode = adev->asic_type > CHIP_RAVEN || + (adev->asic_type == CHIP_RAVEN && + adev->external_rev_id >= 0x81); + + int max_cblock_size = dcc_i64b ? AMD_FMT_MOD_DCC_BLOCK_64B : + dcc_i128b ? AMD_FMT_MOD_DCC_BLOCK_128B : + AMD_FMT_MOD_DCC_BLOCK_256B; + + modifier |= AMD_FMT_MOD_SET(DCC, 1) | + AMD_FMT_MOD_SET(DCC_CONSTANT_ENCODE, dcc_constant_encode) | + AMD_FMT_MOD_SET(DCC_INDEPENDENT_64B, dcc_i64b) | + AMD_FMT_MOD_SET(DCC_INDEPENDENT_128B, dcc_i128b) | + AMD_FMT_MOD_SET(DCC_MAX_COMPRESSED_BLOCK, max_cblock_size); + + afb->base.offsets[1] = dcc_offset * 256 + afb->base.offsets[0]; + afb->base.pitches[1] = AMDGPU_TILING_GET(afb->tiling_flags, DCC_PITCH_MAX) + 1; + } + } + + afb->base.modifier = modifier; + afb->base.flags |= DRM_MODE_FB_MODIFIERS; + return 0; +} + +static int amdgpu_display_get_fb_info(const struct amdgpu_framebuffer *amdgpu_fb, + uint64_t *tiling_flags, bool *tmz_surface) +{ + struct amdgpu_bo *rbo; + int r; + + if (!amdgpu_fb) { + *tiling_flags = 0; + *tmz_surface = false; + return 0; + } + + rbo = gem_to_amdgpu_bo(amdgpu_fb->base.obj[0]); + r = amdgpu_bo_reserve(rbo, false); + + if (unlikely(r)) { + /* Don't show error message when returning -ERESTARTSYS */ + if (r != -ERESTARTSYS) + DRM_ERROR("Unable to reserve buffer: %d\n", r); + return r; + } + + if (tiling_flags) + amdgpu_bo_get_tiling_flags(rbo, tiling_flags); + + if (tmz_surface) + *tmz_surface = amdgpu_bo_encrypted(rbo); + + amdgpu_bo_unreserve(rbo); + + return r; +} + int amdgpu_display_framebuffer_init(struct drm_device *dev, struct amdgpu_framebuffer *rfb, const struct drm_mode_fb_cmd2 *mode_cmd, @@ -542,11 +695,25 @@ int amdgpu_display_framebuffer_init(struct drm_device *dev, rfb->base.obj[0] = obj; drm_helper_mode_fill_fb_struct(dev, &rfb->base, mode_cmd); ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs); - if (ret) { - rfb->base.obj[0] = NULL; - return ret; + if (ret) + goto fail; + + ret = amdgpu_display_get_fb_info(rfb, &rfb->tiling_flags, &rfb->tmz_surface); + if (ret) + goto fail; + + if (dev->mode_config.allow_fb_modifiers && + !(rfb->base.flags & DRM_MODE_FB_MODIFIERS)) { + ret = convert_tiling_flags_to_modifier(rfb); + if (ret) + goto fail; } + return 0; + +fail: + rfb->base.obj[0] = NULL; + return ret; } struct drm_framebuffer * diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 8b30915aa972..26cf369003e1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -94,16 +94,16 @@ #define KMS_DRIVER_MINOR 40 #define KMS_DRIVER_PATCHLEVEL 0 -int amdgpu_vram_limit = 0; -int amdgpu_vis_vram_limit = 0; +int amdgpu_vram_limit; +int amdgpu_vis_vram_limit; int amdgpu_gart_size = -1; /* auto */ int amdgpu_gtt_size = -1; /* auto */ int amdgpu_moverate = -1; /* auto */ -int amdgpu_benchmarking = 0; -int amdgpu_testing = 0; +int amdgpu_benchmarking; +int amdgpu_testing; int amdgpu_audio = -1; -int amdgpu_disp_priority = 0; -int amdgpu_hw_i2c = 0; +int amdgpu_disp_priority; +int amdgpu_hw_i2c; int amdgpu_pcie_gen2 = -1; int amdgpu_msi = -1; char amdgpu_lockup_timeout[AMDGPU_MAX_TIMEOUT_PARAM_LENGTH]; @@ -113,19 +113,19 @@ int amdgpu_aspm = -1; int amdgpu_runtime_pm = -1; uint amdgpu_ip_block_mask = 0xffffffff; int amdgpu_bapm = -1; -int amdgpu_deep_color = 0; +int amdgpu_deep_color; int amdgpu_vm_size = -1; int amdgpu_vm_fragment_size = -1; int amdgpu_vm_block_size = -1; -int amdgpu_vm_fault_stop = 0; -int amdgpu_vm_debug = 0; +int amdgpu_vm_fault_stop; +int amdgpu_vm_debug; int amdgpu_vm_update_mode = -1; -int amdgpu_exp_hw_support = 0; +int amdgpu_exp_hw_support; int amdgpu_dc = -1; int amdgpu_sched_jobs = 32; int amdgpu_sched_hw_submission = 2; -uint amdgpu_pcie_gen_cap = 0; -uint amdgpu_pcie_lane_cap = 0; +uint amdgpu_pcie_gen_cap; +uint amdgpu_pcie_lane_cap; uint amdgpu_cg_mask = 0xffffffff; uint amdgpu_pg_mask = 0xffffffff; uint amdgpu_sdma_phase_quantum = 32; @@ -133,23 +133,31 @@ char *amdgpu_disable_cu = NULL; char *amdgpu_virtual_display = NULL; /* OverDrive(bit 14) disabled by default*/ uint amdgpu_pp_feature_mask = 0xffffbfff; -uint amdgpu_force_long_training = 0; -int amdgpu_job_hang_limit = 0; +uint amdgpu_force_long_training; +int amdgpu_job_hang_limit; int amdgpu_lbpw = -1; int amdgpu_compute_multipipe = -1; int amdgpu_gpu_recovery = -1; /* auto */ -int amdgpu_emu_mode = 0; -uint amdgpu_smu_memory_pool_size = 0; -/* FBC (bit 0) disabled by default*/ -uint amdgpu_dc_feature_mask = 0; -uint amdgpu_dc_debug_mask = 0; +int amdgpu_emu_mode; +uint amdgpu_smu_memory_pool_size; +/* + * FBC (bit 0) disabled by default + * MULTI_MON_PP_MCLK_SWITCH (bit 1) enabled by default + * - With this, for multiple monitors in sync(e.g. with the same model), + * mclk switching will be allowed. And the mclk will be not foced to the + * highest. That helps saving some idle power. + * DISABLE_FRACTIONAL_PWM (bit 2) disabled by default + * PSR (bit 3) disabled by default + */ +uint amdgpu_dc_feature_mask = 2; +uint amdgpu_dc_debug_mask; int amdgpu_async_gfx_ring = 1; -int amdgpu_mcbp = 0; +int amdgpu_mcbp; int amdgpu_discovery = -1; -int amdgpu_mes = 0; +int amdgpu_mes; int amdgpu_noretry = -1; int amdgpu_force_asic_type = -1; -int amdgpu_tmz = 0; +int amdgpu_tmz; int amdgpu_reset_method = -1; /* auto */ int amdgpu_num_kcq = -1; @@ -271,7 +279,7 @@ module_param_string(lockup_timeout, amdgpu_lockup_timeout, sizeof(amdgpu_lockup_ /** * DOC: dpm (int) * Override for dynamic power management setting - * (0 = disable, 1 = enable, 2 = enable sw smu driver for vega20) + * (0 = disable, 1 = enable) * The default is -1 (auto). */ MODULE_PARM_DESC(dpm, "DPM support (1 = enable, 0 = disable, -1 = auto)"); @@ -764,7 +772,7 @@ module_param_named(dcdebugmask, amdgpu_dc_debug_mask, uint, 0444); * Defaults to 0, or disabled. Userspace can still override this level later * after boot. */ -uint amdgpu_dm_abm_level = 0; +uint amdgpu_dm_abm_level; MODULE_PARM_DESC(abmlevel, "ABM level (0 = off (default), 1-4 = backlight reduction level) "); module_param_named(abmlevel, amdgpu_dm_abm_level, uint, 0444); @@ -1089,6 +1097,9 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x73AE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, {0x1002, 0x73BF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, + /* Van Gogh */ + {0x1002, 0x163F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VANGOGH|AMD_IS_APU}, + {0, 0, 0} }; @@ -1495,7 +1506,7 @@ static const struct file_operations amdgpu_driver_kms_fops = { int amdgpu_file_to_fpriv(struct file *filp, struct amdgpu_fpriv **fpriv) { - struct drm_file *file; + struct drm_file *file; if (!filp) return -EINVAL; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c index e2c2eb45a793..0bf7d36c6686 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c @@ -207,6 +207,7 @@ static int amdgpufb_create(struct drm_fb_helper *helper, int ret; unsigned long tmp; + memset(&mode_cmd, 0, sizeof(mode_cmd)); mode_cmd.width = sizes->surface_width; mode_cmd.height = sizes->surface_height; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.c new file mode 100644 index 000000000000..c6947d6c7ff5 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.c @@ -0,0 +1,142 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include <linux/debugfs.h> +#include <linux/firmware.h> +#include <linux/dma-mapping.h> + +#include "amdgpu.h" +#include "amdgpu_psp.h" +#include "amdgpu_ucode.h" +#include "soc15_common.h" + +#define FW_ATTESTATION_DB_COOKIE 0x143b6a37 +#define FW_ATTESTATION_RECORD_VALID 1 +#define FW_ATTESTATION_MAX_SIZE 4096 + +typedef struct FW_ATT_DB_HEADER +{ + uint32_t AttDbVersion; /* version of the fwar feature */ + uint32_t AttDbCookie; /* cookie as an extra check for corrupt data */ +} FW_ATT_DB_HEADER; + +typedef struct FW_ATT_RECORD +{ + uint16_t AttFwIdV1; /* Legacy FW Type field */ + uint16_t AttFwIdV2; /* V2 FW ID field */ + uint32_t AttFWVersion; /* FW Version */ + uint16_t AttFWActiveFunctionID; /* The VF ID (only in VF Attestation Table) */ + uint16_t AttSource; /* FW source indicator */ + uint16_t RecordValid; /* Indicates whether the record is a valid entry */ + uint8_t AttFwTaId; /* Ta ID (only in TA Attestation Table) */ + uint8_t Reserved; +} FW_ATT_RECORD; + +static ssize_t amdgpu_fw_attestation_debugfs_read(struct file *f, + char __user *buf, + size_t size, + loff_t *pos) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private; + uint64_t records_addr = 0; + uint64_t vram_pos = 0; + FW_ATT_DB_HEADER fw_att_hdr = {0}; + FW_ATT_RECORD fw_att_record = {0}; + + if (size < sizeof(FW_ATT_RECORD)) { + DRM_WARN("FW attestation input buffer not enough memory"); + return -EINVAL; + } + + if ((*pos + sizeof(FW_ATT_DB_HEADER)) >= FW_ATTESTATION_MAX_SIZE) { + DRM_WARN("FW attestation out of bounds"); + return 0; + } + + if (psp_get_fw_attestation_records_addr(&adev->psp, &records_addr)) { + DRM_WARN("Failed to get FW attestation record address"); + return -EINVAL; + } + + vram_pos = records_addr - adev->gmc.vram_start; + + if (*pos == 0) { + amdgpu_device_vram_access(adev, + vram_pos, + (uint32_t*)&fw_att_hdr, + sizeof(FW_ATT_DB_HEADER), + false); + + if (fw_att_hdr.AttDbCookie != FW_ATTESTATION_DB_COOKIE) { + DRM_WARN("Invalid FW attestation cookie"); + return -EINVAL; + } + + DRM_INFO("FW attestation version = 0x%X", fw_att_hdr.AttDbVersion); + } + + amdgpu_device_vram_access(adev, + vram_pos + sizeof(FW_ATT_DB_HEADER) + *pos, + (uint32_t*)&fw_att_record, + sizeof(FW_ATT_RECORD), + false); + + if (fw_att_record.RecordValid != FW_ATTESTATION_RECORD_VALID) + return 0; + + if (copy_to_user(buf, (void*)&fw_att_record, sizeof(FW_ATT_RECORD))) + return -EINVAL; + + *pos += sizeof(FW_ATT_RECORD); + + return sizeof(FW_ATT_RECORD); +} + +static const struct file_operations amdgpu_fw_attestation_debugfs_ops = { + .owner = THIS_MODULE, + .read = amdgpu_fw_attestation_debugfs_read, + .write = NULL, + .llseek = default_llseek +}; + +static int amdgpu_is_fw_attestation_supported(struct amdgpu_device *adev) +{ + if (adev->asic_type >= CHIP_SIENNA_CICHLID) + return 1; + + return 0; +} + +void amdgpu_fw_attestation_debugfs_init(struct amdgpu_device *adev) +{ +#if defined(CONFIG_DEBUG_FS) + if (!amdgpu_is_fw_attestation_supported(adev)) + return; + + debugfs_create_file("amdgpu_fw_attestation", + S_IRUSR, + adev_to_drm(adev)->primary->debugfs_root, + adev, + &amdgpu_fw_attestation_debugfs_ops); +#endif +}
\ No newline at end of file diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.h new file mode 100644 index 000000000000..90af4fe58c99 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fw_attestation.h @@ -0,0 +1,30 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * + */ +#ifndef _AMDGPU_FW_ATTESTATION_H +#define _AMDGPU_FW_ATTESTATION_H + +#include "amdgpu.h" + +void amdgpu_fw_attestation_debugfs_init(struct amdgpu_device *adev); +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index 8ea6fc745769..df94db199532 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -849,67 +849,6 @@ int amdgpu_mode_dumb_create(struct drm_file *file_priv, } #if defined(CONFIG_DEBUG_FS) - -#define amdgpu_debugfs_gem_bo_print_flag(m, bo, flag) \ - if (bo->flags & (AMDGPU_GEM_CREATE_ ## flag)) { \ - seq_printf((m), " " #flag); \ - } - -static int amdgpu_debugfs_gem_bo_info(int id, void *ptr, void *data) -{ - struct drm_gem_object *gobj = ptr; - struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj); - struct seq_file *m = data; - - struct dma_buf_attachment *attachment; - struct dma_buf *dma_buf; - unsigned domain; - const char *placement; - unsigned pin_count; - - domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type); - switch (domain) { - case AMDGPU_GEM_DOMAIN_VRAM: - placement = "VRAM"; - break; - case AMDGPU_GEM_DOMAIN_GTT: - placement = " GTT"; - break; - case AMDGPU_GEM_DOMAIN_CPU: - default: - placement = " CPU"; - break; - } - seq_printf(m, "\t0x%08x: %12ld byte %s", - id, amdgpu_bo_size(bo), placement); - - pin_count = READ_ONCE(bo->tbo.pin_count); - if (pin_count) - seq_printf(m, " pin count %d", pin_count); - - dma_buf = READ_ONCE(bo->tbo.base.dma_buf); - attachment = READ_ONCE(bo->tbo.base.import_attach); - - if (attachment) - seq_printf(m, " imported from %p%s", dma_buf, - attachment->peer2peer ? " P2P" : ""); - else if (dma_buf) - seq_printf(m, " exported as %p", dma_buf); - - amdgpu_debugfs_gem_bo_print_flag(m, bo, CPU_ACCESS_REQUIRED); - amdgpu_debugfs_gem_bo_print_flag(m, bo, NO_CPU_ACCESS); - amdgpu_debugfs_gem_bo_print_flag(m, bo, CPU_GTT_USWC); - amdgpu_debugfs_gem_bo_print_flag(m, bo, VRAM_CLEARED); - amdgpu_debugfs_gem_bo_print_flag(m, bo, SHADOW); - amdgpu_debugfs_gem_bo_print_flag(m, bo, VRAM_CONTIGUOUS); - amdgpu_debugfs_gem_bo_print_flag(m, bo, VM_ALWAYS_VALID); - amdgpu_debugfs_gem_bo_print_flag(m, bo, EXPLICIT_SYNC); - - seq_printf(m, "\n"); - - return 0; -} - static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data) { struct drm_info_node *node = (struct drm_info_node *)m->private; @@ -923,6 +862,8 @@ static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data) list_for_each_entry(file, &dev->filelist, lhead) { struct task_struct *task; + struct drm_gem_object *gobj; + int id; /* * Although we have a valid reference on file->pid, that does @@ -937,7 +878,11 @@ static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data) rcu_read_unlock(); spin_lock(&file->table_lock); - idr_for_each(&file->object_idr, amdgpu_debugfs_gem_bo_info, m); + idr_for_each_entry(&file->object_idr, gobj, id) { + struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj); + + amdgpu_bo_print_info(id, bo, m); + } spin_unlock(&file->table_lock); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 8c9bacfdbc30..e584f48f3b54 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -804,3 +804,14 @@ failed_undo: failed_kiq_write: dev_err(adev->dev, "failed to write reg:%x\n", reg); } + +int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev) +{ + if (amdgpu_num_kcq == -1) { + return 8; + } else if (amdgpu_num_kcq > 8 || amdgpu_num_kcq < 0) { + dev_warn(adev->dev, "set kernel compute queue number to 8 due to invalid parameter provided by user\n"); + return 8; + } + return amdgpu_num_kcq; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index 258498cbf1eb..786eb4aa7314 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h @@ -218,6 +218,7 @@ struct amdgpu_gfx_funcs { void (*reset_ras_error_count) (struct amdgpu_device *adev); void (*init_spm_golden)(struct amdgpu_device *adev); void (*query_ras_error_status) (struct amdgpu_device *adev); + void (*update_perfmon_mgcg)(struct amdgpu_device *adev, bool enable); }; struct sq_work { @@ -392,4 +393,5 @@ int amdgpu_gfx_cp_ecc_error_irq(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry); uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg); void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v); +int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index cc86f431a3d4..056cb87d09ea 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -389,6 +389,7 @@ void amdgpu_gmc_tmz_set(struct amdgpu_device *adev) case CHIP_NAVI10: case CHIP_NAVI14: case CHIP_NAVI12: + case CHIP_VANGOGH: /* Don't enable it by default yet. */ if (amdgpu_tmz < 1) { @@ -421,20 +422,30 @@ void amdgpu_gmc_noretry_set(struct amdgpu_device *adev) struct amdgpu_gmc *gmc = &adev->gmc; switch (adev->asic_type) { - case CHIP_RAVEN: - /* Raven currently has issues with noretry - * regardless of what we decide for other - * asics, we should leave raven with - * noretry = 0 until we root cause the - * issues. + case CHIP_VEGA20: + case CHIP_NAVI10: + case CHIP_NAVI14: + case CHIP_SIENNA_CICHLID: + case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: + /* + * noretry = 0 will cause kfd page fault tests fail + * for some ASICs, so set default to 1 for these ASICs. */ if (amdgpu_noretry == -1) - gmc->noretry = 0; + gmc->noretry = 1; else gmc->noretry = amdgpu_noretry; break; + case CHIP_RAVEN: default: - /* default this to 0 for now, but we may want + /* Raven currently has issues with noretry + * regardless of what we decide for other + * asics, we should leave raven with + * noretry = 0 until we root cause the + * issues. + * + * default this to 0 for now, but we may want * to change this in the future for certain * GPUs as it can increase performance in * certain cases. diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c index 1721739def84..8980329cded0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c @@ -81,8 +81,8 @@ static const struct ttm_resource_manager_func amdgpu_gtt_mgr_func; /** * amdgpu_gtt_mgr_init - init GTT manager and DRM MM * - * @man: TTM memory type manager - * @p_size: maximum size of GTT + * @adev: amdgpu_device pointer + * @gtt_size: maximum size of GTT * * Allocate and initialize the GTT manager. */ @@ -123,7 +123,7 @@ int amdgpu_gtt_mgr_init(struct amdgpu_device *adev, uint64_t gtt_size) /** * amdgpu_gtt_mgr_fini - free and destroy GTT manager * - * @man: TTM memory type manager + * @adev: amdgpu_device pointer * * Destroy and free the GTT manager, returns -EBUSY if ranges are still * allocated inside it. diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index efda38349a03..918d4e9c6461 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -325,6 +325,10 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info, fw_info->ver = adev->dm.dmcub_fw_version; fw_info->feature = 0; break; + case AMDGPU_INFO_FW_TOC: + fw_info->ver = adev->psp.toc_fw_version; + fw_info->feature = adev->psp.toc_feature_version; + break; default: return -EINVAL; } @@ -1466,6 +1470,13 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data) seq_printf(m, "DMCUB feature version: %u, firmware version: 0x%08x\n", fw_info.feature, fw_info.ver); + /* TOC */ + query_fw.fw_type = AMDGPU_INFO_FW_TOC; + ret = amdgpu_firmware_info(&fw_info, &query_fw, adev); + if (ret) + return ret; + seq_printf(m, "TOC feature version: %u, firmware version: 0x%08x\n", + fw_info.feature, fw_info.ver); seq_printf(m, "VBIOS version: %s\n", ctx->vbios_version); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index a04decb934b0..319cb19e1b99 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -302,6 +302,9 @@ struct amdgpu_display_funcs { struct amdgpu_framebuffer { struct drm_framebuffer base; + uint64_t tiling_flags; + bool tmz_surface; + /* caching for later use */ uint64_t address; }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h index edaac242ff85..e62cc0e1a5ad 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h @@ -53,6 +53,8 @@ struct amdgpu_nbio_funcs { u32 (*get_hdp_flush_done_offset)(struct amdgpu_device *adev); u32 (*get_pcie_index_offset)(struct amdgpu_device *adev); u32 (*get_pcie_data_offset)(struct amdgpu_device *adev); + u32 (*get_pcie_port_index_offset)(struct amdgpu_device *adev); + u32 (*get_pcie_port_data_offset)(struct amdgpu_device *adev); u32 (*get_rev_id)(struct amdgpu_device *adev); void (*mc_access_enable)(struct amdgpu_device *adev, bool enable); void (*hdp_flush)(struct amdgpu_device *adev, struct amdgpu_ring *ring); @@ -85,6 +87,8 @@ struct amdgpu_nbio_funcs { void (*query_ras_error_count)(struct amdgpu_device *adev, void *ras_error_status); int (*ras_late_init)(struct amdgpu_device *adev); + void (*enable_aspm)(struct amdgpu_device *adev, + bool enable); }; struct amdgpu_nbio { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c index e1f64ef8c765..c6c9723d3d8a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c @@ -1045,6 +1045,7 @@ static const char *amdgpu_vram_names[] = { "DDR3", "DDR4", "GDDR6", + "DDR5" }; /** @@ -1507,3 +1508,77 @@ uint32_t amdgpu_bo_get_preferred_pin_domain(struct amdgpu_device *adev, } return domain; } + +#if defined(CONFIG_DEBUG_FS) +#define amdgpu_bo_print_flag(m, bo, flag) \ + do { \ + if (bo->flags & (AMDGPU_GEM_CREATE_ ## flag)) { \ + seq_printf((m), " " #flag); \ + } \ + } while (0) + +/** + * amdgpu_debugfs_print_bo_info - print BO info in debugfs file + * + * @id: Index or Id of the BO + * @bo: Requested BO for printing info + * @m: debugfs file + * + * Print BO information in debugfs file + * + * Returns: + * Size of the BO in bytes. + */ +u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m) +{ + struct dma_buf_attachment *attachment; + struct dma_buf *dma_buf; + unsigned int domain; + const char *placement; + unsigned int pin_count; + u64 size; + + domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type); + switch (domain) { + case AMDGPU_GEM_DOMAIN_VRAM: + placement = "VRAM"; + break; + case AMDGPU_GEM_DOMAIN_GTT: + placement = " GTT"; + break; + case AMDGPU_GEM_DOMAIN_CPU: + default: + placement = " CPU"; + break; + } + + size = amdgpu_bo_size(bo); + seq_printf(m, "\t\t0x%08x: %12lld byte %s", + id, size, placement); + + pin_count = READ_ONCE(bo->tbo.pin_count); + if (pin_count) + seq_printf(m, " pin count %d", pin_count); + + dma_buf = READ_ONCE(bo->tbo.base.dma_buf); + attachment = READ_ONCE(bo->tbo.base.import_attach); + + if (attachment) + seq_printf(m, " imported from %p", dma_buf); + else if (dma_buf) + seq_printf(m, " exported as %p", dma_buf); + + amdgpu_bo_print_flag(m, bo, CPU_ACCESS_REQUIRED); + amdgpu_bo_print_flag(m, bo, NO_CPU_ACCESS); + amdgpu_bo_print_flag(m, bo, CPU_GTT_USWC); + amdgpu_bo_print_flag(m, bo, VRAM_CLEARED); + amdgpu_bo_print_flag(m, bo, SHADOW); + amdgpu_bo_print_flag(m, bo, VRAM_CONTIGUOUS); + amdgpu_bo_print_flag(m, bo, VM_ALWAYS_VALID); + amdgpu_bo_print_flag(m, bo, EXPLICIT_SYNC); + + seq_puts(m, "\n"); + + return size; +} +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index 132e5f955180..a5a7992b5773 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -329,6 +329,7 @@ void amdgpu_sa_bo_free(struct amdgpu_device *adev, #if defined(CONFIG_DEBUG_FS) void amdgpu_sa_bo_dump_debug_info(struct amdgpu_sa_manager *sa_manager, struct seq_file *m); +u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m); #endif int amdgpu_debugfs_sa_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c index 69af462db34d..19c0a3655228 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c @@ -19,17 +19,29 @@ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. * - * Author: Jonathan Kim <jonathan.kim@amd.com> - * */ #include <linux/perf_event.h> #include <linux/init.h> #include "amdgpu.h" #include "amdgpu_pmu.h" -#include "df_v3_6.h" #define PMU_NAME_SIZE 32 +#define NUM_FORMATS_AMDGPU_PMU 4 +#define NUM_FORMATS_DF_VEGA20 3 +#define NUM_EVENTS_DF_VEGA20 8 +#define NUM_EVENT_TYPES_VEGA20 1 +#define NUM_EVENTS_VEGA20_XGMI 2 +#define NUM_EVENTS_VEGA20_MAX NUM_EVENTS_VEGA20_XGMI +#define NUM_EVENT_TYPES_ARCTURUS 1 +#define NUM_EVENTS_ARCTURUS_XGMI 6 +#define NUM_EVENTS_ARCTURUS_MAX NUM_EVENTS_ARCTURUS_XGMI + +struct amdgpu_pmu_event_attribute { + struct device_attribute attr; + const char *event_str; + unsigned int type; +}; /* record to keep track of pmu entry per pmu type per device */ struct amdgpu_pmu_entry { @@ -37,11 +49,162 @@ struct amdgpu_pmu_entry { struct amdgpu_device *adev; struct pmu pmu; unsigned int pmu_perf_type; + char *pmu_type_name; + char *pmu_file_prefix; + struct attribute_group fmt_attr_group; + struct amdgpu_pmu_event_attribute *fmt_attr; + struct attribute_group evt_attr_group; + struct amdgpu_pmu_event_attribute *evt_attr; }; +static ssize_t amdgpu_pmu_event_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct amdgpu_pmu_event_attribute *amdgpu_pmu_attr; + + amdgpu_pmu_attr = container_of(attr, struct amdgpu_pmu_event_attribute, + attr); + + if (!amdgpu_pmu_attr->type) + return sprintf(buf, "%s\n", amdgpu_pmu_attr->event_str); + + return sprintf(buf, "%s,type=0x%x\n", + amdgpu_pmu_attr->event_str, amdgpu_pmu_attr->type); +} + static LIST_HEAD(amdgpu_pmu_list); +struct amdgpu_pmu_attr { + const char *name; + const char *config; +}; + +struct amdgpu_pmu_type { + const unsigned int type; + const unsigned int num_of_type; +}; + +struct amdgpu_pmu_config { + struct amdgpu_pmu_attr *formats; + unsigned int num_formats; + struct amdgpu_pmu_attr *events; + unsigned int num_events; + struct amdgpu_pmu_type *types; + unsigned int num_types; +}; + +/* + * Events fall under two categories: + * - PMU typed + * Events in /sys/bus/event_source/devices/amdgpu_<pmu_type>_<dev_num> have + * performance counter operations handled by one IP <pmu_type>. Formats and + * events should be defined by <pmu_type>_<asic_type>_formats and + * <pmu_type>_<asic_type>_events respectively. + * + * - Event config typed + * Events in /sys/bus/event_source/devices/amdgpu_<dev_num> have performance + * counter operations that can be handled by multiple IPs dictated by their + * "type" format field. Formats and events should be defined by + * amdgpu_pmu_formats and <asic_type>_events respectively. Format field + * "type" is generated in amdgpu_pmu_event_show and defined in + * <asic_type>_event_config_types. + */ + +static struct amdgpu_pmu_attr amdgpu_pmu_formats[NUM_FORMATS_AMDGPU_PMU] = { + { .name = "event", .config = "config:0-7" }, + { .name = "instance", .config = "config:8-15" }, + { .name = "umask", .config = "config:16-23"}, + { .name = "type", .config = "config:56-63"} +}; + +/* Vega20 events */ +static struct amdgpu_pmu_attr vega20_events[NUM_EVENTS_VEGA20_MAX] = { + { .name = "xgmi_link0_data_outbound", + .config = "event=0x7,instance=0x46,umask=0x2" }, + { .name = "xgmi_link1_data_outbound", + .config = "event=0x7,instance=0x47,umask=0x2" } +}; + +static struct amdgpu_pmu_type vega20_types[NUM_EVENT_TYPES_VEGA20] = { + { .type = AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI, + .num_of_type = NUM_EVENTS_VEGA20_XGMI } +}; + +static struct amdgpu_pmu_config vega20_config = { + .formats = amdgpu_pmu_formats, + .num_formats = ARRAY_SIZE(amdgpu_pmu_formats), + .events = vega20_events, + .num_events = ARRAY_SIZE(vega20_events), + .types = vega20_types, + .num_types = ARRAY_SIZE(vega20_types) +}; + +/* Vega20 data fabric (DF) events */ +static struct amdgpu_pmu_attr df_vega20_formats[NUM_FORMATS_DF_VEGA20] = { + { .name = "event", .config = "config:0-7" }, + { .name = "instance", .config = "config:8-15" }, + { .name = "umask", .config = "config:16-23"} +}; + +static struct amdgpu_pmu_attr df_vega20_events[NUM_EVENTS_DF_VEGA20] = { + { .name = "cake0_pcsout_txdata", + .config = "event=0x7,instance=0x46,umask=0x2" }, + { .name = "cake1_pcsout_txdata", + .config = "event=0x7,instance=0x47,umask=0x2" }, + { .name = "cake0_pcsout_txmeta", + .config = "event=0x7,instance=0x46,umask=0x4" }, + { .name = "cake1_pcsout_txmeta", + .config = "event=0x7,instance=0x47,umask=0x4" }, + { .name = "cake0_ftiinstat_reqalloc", + .config = "event=0xb,instance=0x46,umask=0x4" }, + { .name = "cake1_ftiinstat_reqalloc", + .config = "event=0xb,instance=0x47,umask=0x4" }, + { .name = "cake0_ftiinstat_rspalloc", + .config = "event=0xb,instance=0x46,umask=0x8" }, + { .name = "cake1_ftiinstat_rspalloc", + .config = "event=0xb,instance=0x47,umask=0x8" } +}; + +static struct amdgpu_pmu_config df_vega20_config = { + .formats = df_vega20_formats, + .num_formats = ARRAY_SIZE(df_vega20_formats), + .events = df_vega20_events, + .num_events = ARRAY_SIZE(df_vega20_events), + .types = NULL, + .num_types = 0 +}; + +/* Arcturus events */ +static struct amdgpu_pmu_attr arcturus_events[NUM_EVENTS_ARCTURUS_MAX] = { + { .name = "xgmi_link0_data_outbound", + .config = "event=0x7,instance=0x4b,umask=0x2" }, + { .name = "xgmi_link1_data_outbound", + .config = "event=0x7,instance=0x4c,umask=0x2" }, + { .name = "xgmi_link2_data_outbound", + .config = "event=0x7,instance=0x4d,umask=0x2" }, + { .name = "xgmi_link3_data_outbound", + .config = "event=0x7,instance=0x4e,umask=0x2" }, + { .name = "xgmi_link4_data_outbound", + .config = "event=0x7,instance=0x4f,umask=0x2" }, + { .name = "xgmi_link5_data_outbound", + .config = "event=0x7,instance=0x50,umask=0x2" } +}; + +static struct amdgpu_pmu_type arcturus_types[NUM_EVENT_TYPES_ARCTURUS] = { + { .type = AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI, + .num_of_type = NUM_EVENTS_ARCTURUS_XGMI } +}; + +static struct amdgpu_pmu_config arcturus_config = { + .formats = amdgpu_pmu_formats, + .num_formats = ARRAY_SIZE(amdgpu_pmu_formats), + .events = arcturus_events, + .num_events = ARRAY_SIZE(arcturus_events), + .types = arcturus_types, + .num_types = ARRAY_SIZE(arcturus_types) +}; + /* initialize perf counter */ static int amdgpu_perf_event_init(struct perf_event *event) { @@ -53,6 +216,7 @@ static int amdgpu_perf_event_init(struct perf_event *event) /* update the hw_perf_event struct with config data */ hwc->config = event->attr.config; + hwc->config_base = AMDGPU_PMU_PERF_TYPE_NONE; return 0; } @@ -64,6 +228,7 @@ static void amdgpu_perf_start(struct perf_event *event, int flags) struct amdgpu_pmu_entry *pe = container_of(event->pmu, struct amdgpu_pmu_entry, pmu); + int target_cntr = 0; if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED))) return; @@ -71,19 +236,27 @@ static void amdgpu_perf_start(struct perf_event *event, int flags) WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE)); hwc->state = 0; - switch (pe->pmu_perf_type) { - case PERF_TYPE_AMDGPU_DF: - if (!(flags & PERF_EF_RELOAD)) - pe->adev->df.funcs->pmc_start(pe->adev, hwc->config, 1); + switch (hwc->config_base) { + case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF: + case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI: + if (!(flags & PERF_EF_RELOAD)) { + target_cntr = pe->adev->df.funcs->pmc_start(pe->adev, + hwc->config, 0 /* unused */, + 1 /* add counter */); + if (target_cntr < 0) + break; + + hwc->idx = target_cntr; + } - pe->adev->df.funcs->pmc_start(pe->adev, hwc->config, 0); + pe->adev->df.funcs->pmc_start(pe->adev, hwc->config, + hwc->idx, 0); break; default: break; } perf_event_update_userpage(event); - } /* read perf counter */ @@ -93,16 +266,16 @@ static void amdgpu_perf_read(struct perf_event *event) struct amdgpu_pmu_entry *pe = container_of(event->pmu, struct amdgpu_pmu_entry, pmu); - u64 count, prev; do { prev = local64_read(&hwc->prev_count); - switch (pe->pmu_perf_type) { - case PERF_TYPE_AMDGPU_DF: - pe->adev->df.funcs->pmc_get_count(pe->adev, hwc->config, - &count); + switch (hwc->config_base) { + case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF: + case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI: + pe->adev->df.funcs->pmc_get_count(pe->adev, + hwc->config, hwc->idx, &count); break; default: count = 0; @@ -124,9 +297,11 @@ static void amdgpu_perf_stop(struct perf_event *event, int flags) if (hwc->state & PERF_HES_UPTODATE) return; - switch (pe->pmu_perf_type) { - case PERF_TYPE_AMDGPU_DF: - pe->adev->df.funcs->pmc_stop(pe->adev, hwc->config, 0); + switch (hwc->config_base) { + case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF: + case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI: + pe->adev->df.funcs->pmc_stop(pe->adev, hwc->config, hwc->idx, + 0); break; default: break; @@ -142,22 +317,39 @@ static void amdgpu_perf_stop(struct perf_event *event, int flags) hwc->state |= PERF_HES_UPTODATE; } -/* add perf counter */ +/* add perf counter */ static int amdgpu_perf_add(struct perf_event *event, int flags) { struct hw_perf_event *hwc = &event->hw; - int retval; - + int retval = 0, target_cntr; struct amdgpu_pmu_entry *pe = container_of(event->pmu, struct amdgpu_pmu_entry, pmu); + switch (pe->pmu_perf_type) { + case AMDGPU_PMU_PERF_TYPE_DF: + hwc->config_base = AMDGPU_PMU_EVENT_CONFIG_TYPE_DF; + break; + case AMDGPU_PMU_PERF_TYPE_ALL: + hwc->config_base = (hwc->config >> + AMDGPU_PMU_EVENT_CONFIG_TYPE_SHIFT) & + AMDGPU_PMU_EVENT_CONFIG_TYPE_MASK; + break; + } + event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED; - switch (pe->pmu_perf_type) { - case PERF_TYPE_AMDGPU_DF: - retval = pe->adev->df.funcs->pmc_start(pe->adev, - hwc->config, 1); + switch (hwc->config_base) { + case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF: + case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI: + target_cntr = pe->adev->df.funcs->pmc_start(pe->adev, + hwc->config, 0 /* unused */, + 1 /* add counter */); + if (target_cntr < 0) + retval = target_cntr; + else + hwc->idx = target_cntr; + break; default: return 0; @@ -170,7 +362,6 @@ static int amdgpu_perf_add(struct perf_event *event, int flags) amdgpu_perf_start(event, PERF_EF_RELOAD); return retval; - } /* delete perf counter */ @@ -183,9 +374,11 @@ static void amdgpu_perf_del(struct perf_event *event, int flags) amdgpu_perf_stop(event, PERF_EF_UPDATE); - switch (pe->pmu_perf_type) { - case PERF_TYPE_AMDGPU_DF: - pe->adev->df.funcs->pmc_stop(pe->adev, hwc->config, 1); + switch (hwc->config_base) { + case AMDGPU_PMU_EVENT_CONFIG_TYPE_DF: + case AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI: + pe->adev->df.funcs->pmc_stop(pe->adev, hwc->config, hwc->idx, + 1); break; default: break; @@ -194,25 +387,92 @@ static void amdgpu_perf_del(struct perf_event *event, int flags) perf_event_update_userpage(event); } -/* vega20 pmus */ +static void amdgpu_pmu_create_event_attrs_by_type( + struct attribute_group *attr_group, + struct amdgpu_pmu_event_attribute *pmu_attr, + struct amdgpu_pmu_attr events[], + int s_offset, + int e_offset, + unsigned int type) +{ + int i; + + pmu_attr += s_offset; + + for (i = s_offset; i < e_offset; i++) { + attr_group->attrs[i] = &pmu_attr->attr.attr; + sysfs_attr_init(&pmu_attr->attr.attr); + pmu_attr->attr.attr.name = events[i].name; + pmu_attr->attr.attr.mode = 0444; + pmu_attr->attr.show = amdgpu_pmu_event_show; + pmu_attr->event_str = events[i].config; + pmu_attr->type = type; + pmu_attr++; + } +} -/* init pmu tracking per pmu type */ -static int init_pmu_by_type(struct amdgpu_device *adev, - const struct attribute_group *attr_groups[], - char *pmu_type_name, char *pmu_file_prefix, - unsigned int pmu_perf_type, - unsigned int num_counters) +static void amdgpu_pmu_create_attrs(struct attribute_group *attr_group, + struct amdgpu_pmu_event_attribute *pmu_attr, + struct amdgpu_pmu_attr events[], + int num_events) { - char pmu_name[PMU_NAME_SIZE]; - struct amdgpu_pmu_entry *pmu_entry; - int ret = 0; + amdgpu_pmu_create_event_attrs_by_type(attr_group, pmu_attr, events, 0, + num_events, AMDGPU_PMU_EVENT_CONFIG_TYPE_NONE); +} - pmu_entry = kzalloc(sizeof(struct amdgpu_pmu_entry), GFP_KERNEL); - if (!pmu_entry) +static int amdgpu_pmu_alloc_pmu_attrs( + struct attribute_group *fmt_attr_group, + struct amdgpu_pmu_event_attribute **fmt_attr, + struct attribute_group *evt_attr_group, + struct amdgpu_pmu_event_attribute **evt_attr, + struct amdgpu_pmu_config *config) +{ + *fmt_attr = kcalloc(config->num_formats, sizeof(**fmt_attr), + GFP_KERNEL); + + if (!(*fmt_attr)) return -ENOMEM; - pmu_entry->adev = adev; + fmt_attr_group->attrs = kcalloc(config->num_formats + 1, + sizeof(*fmt_attr_group->attrs), GFP_KERNEL); + + if (!fmt_attr_group->attrs) + goto err_fmt_attr_grp; + + *evt_attr = kcalloc(config->num_events, sizeof(**evt_attr), GFP_KERNEL); + + if (!(*evt_attr)) + goto err_evt_attr; + + evt_attr_group->attrs = kcalloc(config->num_events + 1, + sizeof(*evt_attr_group->attrs), GFP_KERNEL); + + if (!evt_attr_group->attrs) + goto err_evt_attr_grp; + + return 0; +err_evt_attr_grp: + kfree(*evt_attr); +err_evt_attr: + kfree(fmt_attr_group->attrs); +err_fmt_attr_grp: + kfree(*fmt_attr); + return -ENOMEM; +} + +/* init pmu tracking per pmu type */ +static int init_pmu_entry_by_type_and_add(struct amdgpu_pmu_entry *pmu_entry, + struct amdgpu_pmu_config *config) +{ + const struct attribute_group *attr_groups[] = { + &pmu_entry->fmt_attr_group, + &pmu_entry->evt_attr_group, + NULL + }; + char pmu_name[PMU_NAME_SIZE]; + int ret = 0, total_num_events = 0; + pmu_entry->pmu = (struct pmu){ .event_init = amdgpu_perf_event_init, .add = amdgpu_perf_add, @@ -223,59 +483,178 @@ static int init_pmu_by_type(struct amdgpu_device *adev, .task_ctx_nr = perf_invalid_context, }; - pmu_entry->pmu.attr_groups = attr_groups; - pmu_entry->pmu_perf_type = pmu_perf_type; - snprintf(pmu_name, PMU_NAME_SIZE, "%s_%d", - pmu_file_prefix, adev_to_drm(adev)->primary->index); + ret = amdgpu_pmu_alloc_pmu_attrs(&pmu_entry->fmt_attr_group, + &pmu_entry->fmt_attr, + &pmu_entry->evt_attr_group, + &pmu_entry->evt_attr, + config); + + if (ret) + goto err_out; + + amdgpu_pmu_create_attrs(&pmu_entry->fmt_attr_group, pmu_entry->fmt_attr, + config->formats, config->num_formats); + + if (pmu_entry->pmu_perf_type == AMDGPU_PMU_PERF_TYPE_ALL) { + int i; + + for (i = 0; i < config->num_types; i++) { + amdgpu_pmu_create_event_attrs_by_type( + &pmu_entry->evt_attr_group, + pmu_entry->evt_attr, + config->events, + total_num_events, + total_num_events + + config->types[i].num_of_type, + config->types[i].type); + total_num_events += config->types[i].num_of_type; + } + } else { + amdgpu_pmu_create_attrs(&pmu_entry->evt_attr_group, + pmu_entry->evt_attr, + config->events, config->num_events); + total_num_events = config->num_events; + } + + pmu_entry->pmu.attr_groups = kmemdup(attr_groups, sizeof(attr_groups), + GFP_KERNEL); + + if (!pmu_entry->pmu.attr_groups) + goto err_attr_group; + + snprintf(pmu_name, PMU_NAME_SIZE, "%s_%d", pmu_entry->pmu_file_prefix, + adev_to_drm(pmu_entry->adev)->primary->index); ret = perf_pmu_register(&pmu_entry->pmu, pmu_name, -1); - if (ret) { - kfree(pmu_entry); - pr_warn("Error initializing AMDGPU %s PMUs.\n", pmu_type_name); - return ret; - } + if (ret) + goto err_register; + + if (pmu_entry->pmu_perf_type != AMDGPU_PMU_PERF_TYPE_ALL) + pr_info("Detected AMDGPU %s Counters. # of Counters = %d.\n", + pmu_entry->pmu_type_name, total_num_events); + else + pr_info("Detected AMDGPU %d Perf Events.\n", total_num_events); - pr_info("Detected AMDGPU %s Counters. # of Counters = %d.\n", - pmu_type_name, num_counters); list_add_tail(&pmu_entry->entry, &amdgpu_pmu_list); return 0; +err_register: + kfree(pmu_entry->pmu.attr_groups); +err_attr_group: + kfree(pmu_entry->fmt_attr_group.attrs); + kfree(pmu_entry->fmt_attr); + kfree(pmu_entry->evt_attr_group.attrs); + kfree(pmu_entry->evt_attr); +err_out: + pr_warn("Error initializing AMDGPU %s PMUs.\n", + pmu_entry->pmu_type_name); + return ret; +} + +/* destroy all pmu data associated with target device */ +void amdgpu_pmu_fini(struct amdgpu_device *adev) +{ + struct amdgpu_pmu_entry *pe, *temp; + + list_for_each_entry_safe(pe, temp, &amdgpu_pmu_list, entry) { + if (pe->adev != adev) + continue; + list_del(&pe->entry); + perf_pmu_unregister(&pe->pmu); + kfree(pe->pmu.attr_groups); + kfree(pe->fmt_attr_group.attrs); + kfree(pe->fmt_attr); + kfree(pe->evt_attr_group.attrs); + kfree(pe->evt_attr); + kfree(pe); + } +} + +static struct amdgpu_pmu_entry *create_pmu_entry(struct amdgpu_device *adev, + unsigned int pmu_type, + char *pmu_type_name, + char *pmu_file_prefix) +{ + struct amdgpu_pmu_entry *pmu_entry; + + pmu_entry = kzalloc(sizeof(struct amdgpu_pmu_entry), GFP_KERNEL); + + if (!pmu_entry) + return pmu_entry; + + pmu_entry->adev = adev; + pmu_entry->fmt_attr_group.name = "format"; + pmu_entry->fmt_attr_group.attrs = NULL; + pmu_entry->evt_attr_group.name = "events"; + pmu_entry->evt_attr_group.attrs = NULL; + pmu_entry->pmu_perf_type = pmu_type; + pmu_entry->pmu_type_name = pmu_type_name; + pmu_entry->pmu_file_prefix = pmu_file_prefix; + + return pmu_entry; } /* init amdgpu_pmu */ int amdgpu_pmu_init(struct amdgpu_device *adev) { int ret = 0; + struct amdgpu_pmu_entry *pmu_entry, *pmu_entry_df; switch (adev->asic_type) { case CHIP_VEGA20: - /* init df */ - ret = init_pmu_by_type(adev, df_v3_6_attr_groups, - "DF", "amdgpu_df", PERF_TYPE_AMDGPU_DF, - DF_V3_6_MAX_COUNTERS); + pmu_entry_df = create_pmu_entry(adev, AMDGPU_PMU_PERF_TYPE_DF, + "DF", "amdgpu_df"); - /* other pmu types go here*/ - break; - default: - return 0; - } + if (!pmu_entry_df) + return -ENOMEM; - return 0; -} + ret = init_pmu_entry_by_type_and_add(pmu_entry_df, + &df_vega20_config); + if (ret) { + kfree(pmu_entry_df); + return ret; + } -/* destroy all pmu data associated with target device */ -void amdgpu_pmu_fini(struct amdgpu_device *adev) -{ - struct amdgpu_pmu_entry *pe, *temp; + pmu_entry = create_pmu_entry(adev, AMDGPU_PMU_PERF_TYPE_ALL, + "", "amdgpu"); - list_for_each_entry_safe(pe, temp, &amdgpu_pmu_list, entry) { - if (pe->adev == adev) { - list_del(&pe->entry); - perf_pmu_unregister(&pe->pmu); - kfree(pe); + if (!pmu_entry) { + amdgpu_pmu_fini(adev); + return -ENOMEM; + } + + ret = init_pmu_entry_by_type_and_add(pmu_entry, + &vega20_config); + + if (ret) { + kfree(pmu_entry); + amdgpu_pmu_fini(adev); + return ret; + } + + break; + case CHIP_ARCTURUS: + pmu_entry = create_pmu_entry(adev, AMDGPU_PMU_PERF_TYPE_ALL, + "", "amdgpu"); + if (!pmu_entry) + return -ENOMEM; + + ret = init_pmu_entry_by_type_and_add(pmu_entry, + &arcturus_config); + + if (ret) { + kfree(pmu_entry); + return -ENOMEM; } + + break; + + default: + return 0; } + + return ret; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.h index 7dddb7160a11..6882dc48c5d8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.h @@ -19,18 +19,38 @@ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. * - * Author: Jonathan Kim <jonathan.kim@amd.com> - * */ #ifndef _AMDGPU_PMU_H_ #define _AMDGPU_PMU_H_ +/* PMU types. */ enum amdgpu_pmu_perf_type { - PERF_TYPE_AMDGPU_DF = 0, - PERF_TYPE_AMDGPU_MAX + AMDGPU_PMU_PERF_TYPE_NONE = 0, + AMDGPU_PMU_PERF_TYPE_DF, + AMDGPU_PMU_PERF_TYPE_ALL }; +/* + * PMU type AMDGPU_PMU_PERF_TYPE_ALL can hold events of different "type" + * configurations. Event config types are parsed from the 64-bit raw + * config (See EVENT_CONFIG_TYPE_SHIFT and EVENT_CONFIG_TYPE_MASK) and + * are registered into the HW perf events config_base. + * + * PMU types with only a single event configuration type + * (non-AMDGPU_PMU_PERF_TYPE_ALL) have their event config type auto generated + * when the performance counter is added. + */ +enum amdgpu_pmu_event_config_type { + AMDGPU_PMU_EVENT_CONFIG_TYPE_NONE = 0, + AMDGPU_PMU_EVENT_CONFIG_TYPE_DF, + AMDGPU_PMU_EVENT_CONFIG_TYPE_XGMI, + AMDGPU_PMU_EVENT_CONFIG_TYPE_MAX +}; + +#define AMDGPU_PMU_EVENT_CONFIG_TYPE_SHIFT 56 +#define AMDGPU_PMU_EVENT_CONFIG_TYPE_MASK 0xff + int amdgpu_pmu_init(struct amdgpu_device *adev); void amdgpu_pmu_fini(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index 96a9699f87ba..2b0a2b93994b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -100,6 +100,8 @@ static int psp_early_init(void *handle) case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: psp_v11_0_set_psp_funcs(psp); psp->autoload_supported = true; break; @@ -288,6 +290,8 @@ psp_cmd_submit_buf(struct psp_context *psp, skip_unsupport = (psp->cmd_buf_mem->resp.status == TEE_ERROR_NOT_SUPPORTED || psp->cmd_buf_mem->resp.status == PSP_ERR_UNKNOWN_COMMAND) && amdgpu_sriov_vf(psp->adev); + memcpy((void*)&cmd->resp, (void*)&psp->cmd_buf_mem->resp, sizeof(struct psp_gfx_resp)); + /* In some cases, psp response status is not 0 even there is no * problem while the command is submitted. Some version of PSP FW * doesn't write 0 to that field. @@ -308,9 +312,6 @@ psp_cmd_submit_buf(struct psp_context *psp, } } - /* get xGMI session id from response buffer */ - cmd->resp.session_id = psp->cmd_buf_mem->resp.session_id; - if (ucode) { ucode->tmr_mc_addr_lo = psp->cmd_buf_mem->resp.fw_addr_lo; ucode->tmr_mc_addr_hi = psp->cmd_buf_mem->resp.fw_addr_hi; @@ -509,6 +510,37 @@ static int psp_tmr_terminate(struct psp_context *psp) return 0; } +int psp_get_fw_attestation_records_addr(struct psp_context *psp, + uint64_t *output_ptr) +{ + int ret; + struct psp_gfx_cmd_resp *cmd; + + if (!output_ptr) + return -EINVAL; + + if (amdgpu_sriov_vf(psp->adev)) + return 0; + + cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL); + if (!cmd) + return -ENOMEM; + + cmd->cmd_id = GFX_CMD_ID_GET_FW_ATTESTATION; + + ret = psp_cmd_submit_buf(psp, NULL, cmd, + psp->fence_buf_mc_addr); + + if (!ret) { + *output_ptr = ((uint64_t)cmd->resp.uresp.fwar_db_info.fwar_db_addr_lo) + + ((uint64_t)cmd->resp.uresp.fwar_db_info.fwar_db_addr_hi << 32); + } + + kfree(cmd); + + return ret; +} + static void psp_prep_asd_load_cmd_buf(struct psp_gfx_cmd_resp *cmd, uint64_t asd_mc, uint32_t size) { @@ -624,14 +656,14 @@ static void psp_prep_ta_load_cmd_buf(struct psp_gfx_cmd_resp *cmd, uint64_t ta_shared_mc, uint32_t ta_shared_size) { - cmd->cmd_id = GFX_CMD_ID_LOAD_TA; + cmd->cmd_id = GFX_CMD_ID_LOAD_TA; cmd->cmd.cmd_load_ta.app_phy_addr_lo = lower_32_bits(ta_bin_mc); - cmd->cmd.cmd_load_ta.app_phy_addr_hi = upper_32_bits(ta_bin_mc); - cmd->cmd.cmd_load_ta.app_len = ta_bin_size; + cmd->cmd.cmd_load_ta.app_phy_addr_hi = upper_32_bits(ta_bin_mc); + cmd->cmd.cmd_load_ta.app_len = ta_bin_size; cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo = lower_32_bits(ta_shared_mc); cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi = upper_32_bits(ta_shared_mc); - cmd->cmd.cmd_load_ta.cmd_buf_len = ta_shared_size; + cmd->cmd.cmd_load_ta.cmd_buf_len = ta_shared_size; } static int psp_xgmi_init_shared_buf(struct psp_context *psp) @@ -655,9 +687,9 @@ static void psp_prep_ta_invoke_cmd_buf(struct psp_gfx_cmd_resp *cmd, uint32_t ta_cmd_id, uint32_t session_id) { - cmd->cmd_id = GFX_CMD_ID_INVOKE_CMD; - cmd->cmd.cmd_invoke_cmd.session_id = session_id; - cmd->cmd.cmd_invoke_cmd.ta_cmd_id = ta_cmd_id; + cmd->cmd_id = GFX_CMD_ID_INVOKE_CMD; + cmd->cmd.cmd_invoke_cmd.session_id = session_id; + cmd->cmd.cmd_invoke_cmd.ta_cmd_id = ta_cmd_id; } static int psp_ta_invoke(struct psp_context *psp, @@ -806,7 +838,7 @@ int psp_xgmi_get_hive_id(struct psp_context *psp, uint64_t *hive_id) struct ta_xgmi_shared_memory *xgmi_cmd; int ret; - xgmi_cmd = (struct ta_xgmi_shared_memory*)psp->xgmi_context.xgmi_shared_buf; + xgmi_cmd = (struct ta_xgmi_shared_memory *)psp->xgmi_context.xgmi_shared_buf; memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory)); xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_HIVE_ID; @@ -826,7 +858,7 @@ int psp_xgmi_get_node_id(struct psp_context *psp, uint64_t *node_id) struct ta_xgmi_shared_memory *xgmi_cmd; int ret; - xgmi_cmd = (struct ta_xgmi_shared_memory*)psp->xgmi_context.xgmi_shared_buf; + xgmi_cmd = (struct ta_xgmi_shared_memory *)psp->xgmi_context.xgmi_shared_buf; memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory)); xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_NODE_ID; @@ -854,7 +886,7 @@ int psp_xgmi_get_topology_info(struct psp_context *psp, if (!topology || topology->num_nodes > TA_XGMI__MAX_CONNECTED_NODES) return -EINVAL; - xgmi_cmd = (struct ta_xgmi_shared_memory*)psp->xgmi_context.xgmi_shared_buf; + xgmi_cmd = (struct ta_xgmi_shared_memory *)psp->xgmi_context.xgmi_shared_buf; memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory)); /* Fill in the shared memory with topology information as input */ @@ -898,7 +930,7 @@ int psp_xgmi_set_topology_info(struct psp_context *psp, if (!topology || topology->num_nodes > TA_XGMI__MAX_CONNECTED_NODES) return -EINVAL; - xgmi_cmd = (struct ta_xgmi_shared_memory*)psp->xgmi_context.xgmi_shared_buf; + xgmi_cmd = (struct ta_xgmi_shared_memory *)psp->xgmi_context.xgmi_shared_buf; memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory)); topology_info_input = &xgmi_cmd->xgmi_in_message.get_topology_info; @@ -962,7 +994,7 @@ static int psp_ras_load(struct psp_context *psp) ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr); - ras_cmd = (struct ta_ras_shared_memory*)psp->ras.ras_shared_buf; + ras_cmd = (struct ta_ras_shared_memory *)psp->ras.ras_shared_buf; if (!ret) { psp->ras.session_id = cmd->resp.session_id; @@ -1884,7 +1916,7 @@ static int psp_execute_np_fw_load(struct psp_context *psp, static int psp_load_smu_fw(struct psp_context *psp) { int ret; - struct amdgpu_device* adev = psp->adev; + struct amdgpu_device *adev = psp->adev; struct amdgpu_firmware_info *ucode = &adev->firmware.ucode[AMDGPU_UCODE_ID_SMC]; struct amdgpu_ras *ras = psp->ras.ras; @@ -1950,7 +1982,7 @@ static int psp_np_fw_load(struct psp_context *psp) { int i, ret; struct amdgpu_firmware_info *ucode; - struct amdgpu_device* adev = psp->adev; + struct amdgpu_device *adev = psp->adev; if (psp->autoload_supported && !psp->pmfw_centralized_cstate_management) { @@ -1974,8 +2006,8 @@ static int psp_np_fw_load(struct psp_context *psp) continue; if (psp->autoload_supported && - (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) && + (adev->asic_type >= CHIP_SIENNA_CICHLID && + adev->asic_type <= CHIP_DIMGREY_CAVEFISH) && (ucode->ucode_id == AMDGPU_UCODE_ID_SDMA1 || ucode->ucode_id == AMDGPU_UCODE_ID_SDMA2 || ucode->ucode_id == AMDGPU_UCODE_ID_SDMA3)) @@ -2390,7 +2422,7 @@ int psp_init_asd_microcode(struct psp_context *psp, const char *chip_name) { struct amdgpu_device *adev = psp->adev; - char fw_name[30]; + char fw_name[PSP_FW_NAME_LEN]; const struct psp_firmware_header_v1_0 *asd_hdr; int err = 0; @@ -2422,11 +2454,47 @@ out: return err; } -int psp_init_sos_microcode(struct psp_context *psp, +int psp_init_toc_microcode(struct psp_context *psp, const char *chip_name) { struct amdgpu_device *adev = psp->adev; char fw_name[30]; + const struct psp_firmware_header_v1_0 *toc_hdr; + int err = 0; + + if (!chip_name) { + dev_err(adev->dev, "invalid chip name for toc microcode\n"); + return -EINVAL; + } + + snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_toc.bin", chip_name); + err = request_firmware(&adev->psp.toc_fw, fw_name, adev->dev); + if (err) + goto out; + + err = amdgpu_ucode_validate(adev->psp.toc_fw); + if (err) + goto out; + + toc_hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.toc_fw->data; + adev->psp.toc_fw_version = le32_to_cpu(toc_hdr->header.ucode_version); + adev->psp.toc_feature_version = le32_to_cpu(toc_hdr->ucode_feature_version); + adev->psp.toc_bin_size = le32_to_cpu(toc_hdr->header.ucode_size_bytes); + adev->psp.toc_start_addr = (uint8_t *)toc_hdr + + le32_to_cpu(toc_hdr->header.ucode_array_offset_bytes); + return 0; +out: + dev_err(adev->dev, "fail to request/validate toc microcode\n"); + release_firmware(adev->psp.toc_fw); + adev->psp.toc_fw = NULL; + return err; +} + +int psp_init_sos_microcode(struct psp_context *psp, + const char *chip_name) +{ + struct amdgpu_device *adev = psp->adev; + char fw_name[PSP_FW_NAME_LEN]; const struct psp_firmware_header_v1_0 *sos_hdr; const struct psp_firmware_header_v1_1 *sos_hdr_v1_1; const struct psp_firmware_header_v1_2 *sos_hdr_v1_2; @@ -2520,10 +2588,11 @@ int parse_ta_bin_descriptor(struct psp_context *psp, switch (desc->fw_type) { case TA_FW_TYPE_PSP_ASD: - psp->asd_fw_version = le32_to_cpu(desc->fw_version); + psp->asd_fw_version = le32_to_cpu(desc->fw_version); psp->asd_feature_version = le32_to_cpu(desc->fw_version); - psp->asd_ucode_size = le32_to_cpu(desc->size_bytes); + psp->asd_ucode_size = le32_to_cpu(desc->size_bytes); psp->asd_start_addr = ucode_start_addr; + psp->asd_fw = psp->ta_fw; break; case TA_FW_TYPE_PSP_XGMI: psp->ta_xgmi_ucode_version = le32_to_cpu(desc->fw_version); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index 919d2fb7427b..da250bc1ac57 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -41,6 +41,7 @@ #define PSP_DTM_SHARED_MEM_SIZE 0x4000 #define PSP_RAP_SHARED_MEM_SIZE 0x4000 #define PSP_SHARED_MEM_SIZE 0x4000 +#define PSP_FW_NAME_LEN 0x24 struct psp_context; struct psp_xgmi_node_info; @@ -253,6 +254,11 @@ struct psp_context uint32_t asd_ucode_size; uint8_t *asd_start_addr; + /* toc firmware */ + const struct firmware *toc_fw; + uint32_t toc_fw_version; + uint32_t toc_feature_version; + /* fence buffer */ struct amdgpu_bo *fence_buf_bo; uint64_t fence_buf_mc_addr; @@ -386,8 +392,12 @@ int psp_ring_cmd_submit(struct psp_context *psp, int index); int psp_init_asd_microcode(struct psp_context *psp, const char *chip_name); +int psp_init_toc_microcode(struct psp_context *psp, + const char *chip_name); int psp_init_sos_microcode(struct psp_context *psp, const char *chip_name); int psp_init_ta_microcode(struct psp_context *psp, const char *chip_name); +int psp_get_fw_attestation_records_addr(struct psp_context *psp, + uint64_t *output_ptr); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 4e36551ab50b..b96267068a72 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -80,6 +80,8 @@ enum amdgpu_ras_retire_page_reservation { atomic_t amdgpu_ras_in_intr = ATOMIC_INIT(0); +static bool amdgpu_ras_check_bad_page_unlock(struct amdgpu_ras *con, + uint64_t addr); static bool amdgpu_ras_check_bad_page(struct amdgpu_device *adev, uint64_t addr); @@ -516,9 +518,9 @@ struct ras_manager *amdgpu_ras_find_obj(struct amdgpu_device *adev, /* obj end */ static void amdgpu_ras_parse_status_code(struct amdgpu_device *adev, - const char* invoke_type, - const char* block_name, - enum ta_ras_status ret) + const char* invoke_type, + const char* block_name, + enum ta_ras_status ret) { switch (ret) { case TA_RAS_STATUS__SUCCESS: @@ -607,7 +609,7 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev, if (!con) return -EINVAL; - info = kzalloc(sizeof(union ta_ras_cmd_input), GFP_KERNEL); + info = kzalloc(sizeof(union ta_ras_cmd_input), GFP_KERNEL); if (!info) return -ENOMEM; @@ -953,7 +955,7 @@ static char *amdgpu_ras_badpage_flags_str(unsigned int flags) case AMDGPU_RAS_RETIRE_PAGE_FAULT: default: return "F"; - }; + } } /** @@ -1551,10 +1553,12 @@ static int amdgpu_ras_badpages_read(struct amdgpu_device *adev, .size = AMDGPU_GPU_PAGE_SIZE, .flags = AMDGPU_RAS_RETIRE_PAGE_RESERVED, }; - - if (data->last_reserved <= i) + ret = amdgpu_vram_mgr_query_page_status( + ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM), + data->bps[i].retired_page); + if (ret == -EBUSY) (*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_PENDING; - else if (data->bps_bo[i] == NULL) + else if (ret == -ENOENT) (*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_FAULT; } @@ -1606,12 +1610,9 @@ static int amdgpu_ras_realloc_eh_data_space(struct amdgpu_device *adev, unsigned int new_space = old_space + pages; unsigned int align_space = ALIGN(new_space, 512); void *bps = kmalloc(align_space * sizeof(*data->bps), GFP_KERNEL); - struct amdgpu_bo **bps_bo = - kmalloc(align_space * sizeof(*data->bps_bo), GFP_KERNEL); - if (!bps || !bps_bo) { + if (!bps) { kfree(bps); - kfree(bps_bo); return -ENOMEM; } @@ -1620,14 +1621,8 @@ static int amdgpu_ras_realloc_eh_data_space(struct amdgpu_device *adev, data->count * sizeof(*data->bps)); kfree(data->bps); } - if (data->bps_bo) { - memcpy(bps_bo, data->bps_bo, - data->count * sizeof(*data->bps_bo)); - kfree(data->bps_bo); - } data->bps = bps; - data->bps_bo = bps_bo; data->space_left += align_space - old_space; return 0; } @@ -1639,6 +1634,7 @@ int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev, struct amdgpu_ras *con = amdgpu_ras_get_context(adev); struct ras_err_handler_data *data; int ret = 0; + uint32_t i; if (!con || !con->eh_data || !bps || pages <= 0) return 0; @@ -1648,16 +1644,26 @@ int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev, if (!data) goto out; - if (data->space_left <= pages) - if (amdgpu_ras_realloc_eh_data_space(adev, data, pages)) { + for (i = 0; i < pages; i++) { + if (amdgpu_ras_check_bad_page_unlock(con, + bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT)) + continue; + + if (!data->space_left && + amdgpu_ras_realloc_eh_data_space(adev, data, 256)) { ret = -ENOMEM; goto out; } - memcpy(&data->bps[data->count], bps, pages * sizeof(*data->bps)); - data->count += pages; - data->space_left -= pages; + amdgpu_vram_mgr_reserve_range( + ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM), + bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT, + AMDGPU_GPU_PAGE_SIZE); + memcpy(&data->bps[data->count], &bps[i], sizeof(*data->bps)); + data->count++; + data->space_left--; + } out: mutex_unlock(&con->recovery_lock); @@ -1668,7 +1674,7 @@ out: * write error record array to eeprom, the function should be * protected by recovery_lock */ -static int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev) +int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); struct ras_err_handler_data *data; @@ -1730,6 +1736,20 @@ out: return ret; } +static bool amdgpu_ras_check_bad_page_unlock(struct amdgpu_ras *con, + uint64_t addr) +{ + struct ras_err_handler_data *data = con->eh_data; + int i; + + addr >>= AMDGPU_GPU_PAGE_SHIFT; + for (i = 0; i < data->count; i++) + if (addr == data->bps[i].retired_page) + return true; + + return false; +} + /* * check if an address belongs to bad page * @@ -1739,26 +1759,13 @@ static bool amdgpu_ras_check_bad_page(struct amdgpu_device *adev, uint64_t addr) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - struct ras_err_handler_data *data; - int i; bool ret = false; if (!con || !con->eh_data) return ret; mutex_lock(&con->recovery_lock); - data = con->eh_data; - if (!data) - goto out; - - addr >>= AMDGPU_GPU_PAGE_SHIFT; - for (i = 0; i < data->count; i++) - if (addr == data->bps[i].retired_page) { - ret = true; - goto out; - } - -out: + ret = amdgpu_ras_check_bad_page_unlock(con, addr); mutex_unlock(&con->recovery_lock); return ret; } @@ -1804,80 +1811,6 @@ static void amdgpu_ras_validate_threshold(struct amdgpu_device *adev, } } -/* called in gpu recovery/init */ -int amdgpu_ras_reserve_bad_pages(struct amdgpu_device *adev) -{ - struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - struct ras_err_handler_data *data; - uint64_t bp; - struct amdgpu_bo *bo = NULL; - int i, ret = 0; - - /* Not reserve bad page when amdgpu_bad_page_threshold == 0. */ - if (!con || !con->eh_data || (amdgpu_bad_page_threshold == 0)) - return 0; - - mutex_lock(&con->recovery_lock); - data = con->eh_data; - if (!data) - goto out; - /* reserve vram at driver post stage. */ - for (i = data->last_reserved; i < data->count; i++) { - bp = data->bps[i].retired_page; - - /* There are two cases of reserve error should be ignored: - * 1) a ras bad page has been allocated (used by someone); - * 2) a ras bad page has been reserved (duplicate error injection - * for one page); - */ - if (amdgpu_bo_create_kernel_at(adev, bp << AMDGPU_GPU_PAGE_SHIFT, - AMDGPU_GPU_PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &bo, NULL)) - dev_warn(adev->dev, "RAS WARN: reserve vram for " - "retired page %llx fail\n", bp); - - data->bps_bo[i] = bo; - data->last_reserved = i + 1; - bo = NULL; - } - - /* continue to save bad pages to eeprom even reesrve_vram fails */ - ret = amdgpu_ras_save_bad_pages(adev); -out: - mutex_unlock(&con->recovery_lock); - return ret; -} - -/* called when driver unload */ -static int amdgpu_ras_release_bad_pages(struct amdgpu_device *adev) -{ - struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - struct ras_err_handler_data *data; - struct amdgpu_bo *bo; - int i; - - if (!con || !con->eh_data) - return 0; - - mutex_lock(&con->recovery_lock); - data = con->eh_data; - if (!data) - goto out; - - for (i = data->last_reserved - 1; i >= 0; i--) { - bo = data->bps_bo[i]; - - amdgpu_bo_free_kernel(&bo, NULL, NULL); - - data->bps_bo[i] = bo; - data->last_reserved = i; - } -out: - mutex_unlock(&con->recovery_lock); - return 0; -} - int amdgpu_ras_recovery_init(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); @@ -1917,18 +1850,12 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev) ret = amdgpu_ras_load_bad_pages(adev); if (ret) goto free; - ret = amdgpu_ras_reserve_bad_pages(adev); - if (ret) - goto release; } return 0; -release: - amdgpu_ras_release_bad_pages(adev); free: kfree((*data)->bps); - kfree((*data)->bps_bo); kfree(*data); con->eh_data = NULL; out: @@ -1956,12 +1883,10 @@ static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev) return 0; cancel_work_sync(&con->recovery_work); - amdgpu_ras_release_bad_pages(adev); mutex_lock(&con->recovery_lock); con->eh_data = NULL; kfree(data->bps); - kfree(data->bps_bo); kfree(data); mutex_unlock(&con->recovery_lock); @@ -2156,7 +2081,7 @@ void amdgpu_ras_late_fini(struct amdgpu_device *adev, amdgpu_ras_sysfs_remove(adev, ras_block); if (ih_info->cb) - amdgpu_ras_interrupt_remove_handler(adev, ih_info); + amdgpu_ras_interrupt_remove_handler(adev, ih_info); amdgpu_ras_feature_enable(adev, ras_block, 0); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 6b8d7bb83bb3..4667cce38582 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -33,7 +33,6 @@ #define AMDGPU_RAS_FLAG_INIT_BY_VBIOS (0x1 << 0) #define AMDGPU_RAS_FLAG_INIT_NEED_RESET (0x1 << 1) -#define AMDGPU_RAS_FLAG_SKIP_BAD_PAGE_RESV (0x1 << 2) enum amdgpu_ras_block { AMDGPU_RAS_BLOCK__UMC = 0, @@ -363,14 +362,10 @@ struct ras_err_data { struct ras_err_handler_data { /* point to bad page records array */ struct eeprom_table_record *bps; - /* point to reserved bo array */ - struct amdgpu_bo **bps_bo; /* the count of entries */ int count; /* the space can place new entries */ int space_left; - /* last reserved entry's index + 1 */ - int last_reserved; }; typedef int (*ras_ih_cb)(struct amdgpu_device *adev, @@ -506,22 +501,12 @@ bool amdgpu_ras_check_err_threshold(struct amdgpu_device *adev); int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev, struct eeprom_table_record *bps, int pages); -int amdgpu_ras_reserve_bad_pages(struct amdgpu_device *adev); +int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev); static inline int amdgpu_ras_reset_gpu(struct amdgpu_device *adev) { struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); - /* - * Save bad page to eeprom before gpu reset, i2c may be unstable - * in gpu reset. - * - * Also, exclude the case when ras recovery issuer is - * eeprom page write itself. - */ - if (!(ras->flags & AMDGPU_RAS_FLAG_SKIP_BAD_PAGE_RESV) && in_task()) - amdgpu_ras_reserve_bad_pages(adev); - if (atomic_cmpxchg(&ras->in_recovery, 0, 1) == 0) schedule_work(&ras->recovery_work); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index 0e64c39a2372..1dd040166c63 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -27,9 +27,9 @@ #include <linux/bits.h> #include "atom.h" -#define EEPROM_I2C_TARGET_ADDR_VEGA20 0xA0 -#define EEPROM_I2C_TARGET_ADDR_ARCTURUS 0xA8 -#define EEPROM_I2C_TARGET_ADDR_ARCTURUS_D342 0xA0 +#define EEPROM_I2C_TARGET_ADDR_VEGA20 0xA0 +#define EEPROM_I2C_TARGET_ADDR_ARCTURUS 0xA8 +#define EEPROM_I2C_TARGET_ADDR_ARCTURUS_D342 0xA0 /* * The 2 macros bellow represent the actual size in bytes that @@ -124,11 +124,11 @@ static void __decode_table_header_from_buff(struct amdgpu_ras_eeprom_table_heade { uint32_t *pp = (uint32_t *)buff; - hdr->header = le32_to_cpu(pp[0]); - hdr->version = le32_to_cpu(pp[1]); + hdr->header = le32_to_cpu(pp[0]); + hdr->version = le32_to_cpu(pp[1]); hdr->first_rec_offset = le32_to_cpu(pp[2]); - hdr->tbl_size = le32_to_cpu(pp[3]); - hdr->checksum = le32_to_cpu(pp[4]); + hdr->tbl_size = le32_to_cpu(pp[3]); + hdr->checksum = le32_to_cpu(pp[4]); } static int __update_table_header(struct amdgpu_ras_eeprom_control *control, @@ -149,7 +149,11 @@ static int __update_table_header(struct amdgpu_ras_eeprom_control *control, msg.addr = control->i2c_address; + /* i2c may be unstable in gpu reset */ + down_read(&adev->reset_sem); ret = i2c_transfer(&adev->pm.smu_i2c, &msg, 1); + up_read(&adev->reset_sem); + if (ret < 1) DRM_ERROR("Failed to write EEPROM table header, ret:%d", ret); @@ -475,7 +479,6 @@ int amdgpu_ras_eeprom_process_recods(struct amdgpu_ras_eeprom_control *control, int i, ret = 0; struct i2c_msg *msgs, *msg; unsigned char *buffs, *buff; - bool sched_ras_recovery = false; struct eeprom_table_record *record; struct amdgpu_device *adev = to_amdgpu_device(control); struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); @@ -513,7 +516,6 @@ int amdgpu_ras_eeprom_process_recods(struct amdgpu_ras_eeprom_control *control, "Saved bad pages(%d) reaches threshold value(%d).\n", control->num_recs + num, ras->bad_page_cnt_threshold); control->tbl_hdr.header = EEPROM_TABLE_HDR_BAD; - sched_ras_recovery = true; } /* In case of overflow just start from beginning to not lose newest records */ @@ -557,7 +559,11 @@ int amdgpu_ras_eeprom_process_recods(struct amdgpu_ras_eeprom_control *control, control->next_addr += EEPROM_TABLE_RECORD_SIZE; } + /* i2c may be unstable in gpu reset */ + down_read(&adev->reset_sem); ret = i2c_transfer(&adev->pm.smu_i2c, msgs, num); + up_read(&adev->reset_sem); + if (ret < 1) { DRM_ERROR("Failed to process EEPROM table records, ret:%d", ret); @@ -595,20 +601,6 @@ int amdgpu_ras_eeprom_process_recods(struct amdgpu_ras_eeprom_control *control, __update_tbl_checksum(control, records, num, old_hdr_byte_sum); __update_table_header(control, buffs); - - if (sched_ras_recovery) { - /* - * Before scheduling ras recovery, assert the related - * flag first, which shall bypass common bad page - * reservation execution in amdgpu_ras_reset_gpu. - */ - amdgpu_ras_get_context(adev)->flags |= - AMDGPU_RAS_FLAG_SKIP_BAD_PAGE_RESV; - - dev_warn(adev->dev, "Conduct ras recovery due to bad " - "page threshold reached.\n"); - amdgpu_ras_reset_gpu(adev); - } } else if (!__validate_tbl_checksum(control, records, num)) { DRM_WARN("EEPROM Table checksum mismatch!"); /* TODO Uncomment when EEPROM read/write is relliable */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c index 15ee13c3bd9e..2697b250dc32 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c @@ -396,7 +396,7 @@ static ssize_t amdgpu_debugfs_ring_read(struct file *f, char __user *buf, return result; value = ring->ring[(*pos - 12)/4]; - r = put_user(value, (uint32_t*)buf); + r = put_user(value, (uint32_t *)buf); if (r) return r; buf += 4; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c index 0bd1d4ffc19e..524d10b21041 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c @@ -75,7 +75,7 @@ int amdgpu_sa_bo_manager_init(struct amdgpu_device *adev, } void amdgpu_sa_bo_manager_fini(struct amdgpu_device *adev, - struct amdgpu_sa_manager *sa_manager) + struct amdgpu_sa_manager *sa_manager) { struct amdgpu_sa_bo *sa_bo, *tmp; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c index 250a309e4dee..de91d29c9d96 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c @@ -126,7 +126,7 @@ int amdgpu_sdma_ras_late_init(struct amdgpu_device *adev, goto free; } - return 0; + return 0; late_fini: amdgpu_ras_late_fini(adev, adev->sdma.ras_if, ih_info); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c index 2f4d5ca9894f..6042b3b81a4c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c @@ -157,10 +157,10 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev) i, *vram_start, gart_start, (unsigned long long) (gart_addr - adev->gmc.gart_start + - (void*)gart_start - gtt_map), + (void *)gart_start - gtt_map), (unsigned long long) (vram_addr - adev->gmc.vram_start + - (void*)gart_start - gtt_map)); + (void *)gart_start - gtt_map)); amdgpu_bo_kunmap(vram_obj); goto out_lclean_unpin; } @@ -203,10 +203,10 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev) i, *gart_start, vram_start, (unsigned long long) (vram_addr - adev->gmc.vram_start + - (void*)vram_start - vram_map), + (void *)vram_start - vram_map), (unsigned long long) (gart_addr - adev->gmc.gart_start + - (void*)vram_start - vram_map)); + (void *)vram_start - vram_map)); amdgpu_bo_kunmap(gtt_obj[i]); goto out_lclean_unpin; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index c01c060e4ac5..12aa35ac0eb4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -70,10 +70,10 @@ static void amdgpu_ttm_backend_unbind(struct ttm_bo_device *bdev, static int amdgpu_ttm_init_on_chip(struct amdgpu_device *adev, unsigned int type, - uint64_t size) + uint64_t size_in_page) { return ttm_range_man_init(&adev->mman.bdev, type, - false, size >> PAGE_SHIFT); + false, size_in_page); } /** @@ -1008,7 +1008,7 @@ void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages) } /** - * amdgpu_ttm_tt_pin_userptr - prepare the sg table with the user pages + * amdgpu_ttm_tt_pin_userptr - prepare the sg table with the user pages * * Called by amdgpu_ttm_backend_bind() **/ @@ -1198,7 +1198,7 @@ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo) { struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev); struct ttm_operation_ctx ctx = { false, false }; - struct amdgpu_ttm_tt *gtt = (void*)bo->ttm; + struct amdgpu_ttm_tt *gtt = (void *)bo->ttm; struct ttm_resource tmp; struct ttm_placement placement; struct ttm_place placements; @@ -1449,7 +1449,7 @@ int amdgpu_ttm_tt_set_userptr(struct ttm_buffer_object *bo, return -ENOMEM; } - gtt = (void*)bo->ttm; + gtt = (void *)bo->ttm; gtt->userptr = addr; gtt->userflags = flags; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index a87951b2f06d..bdca2970173e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -37,10 +37,17 @@ #define AMDGPU_POISON 0xd0bed0be +struct amdgpu_vram_reservation { + struct list_head node; + struct drm_mm_node mm_node; +}; + struct amdgpu_vram_mgr { struct ttm_resource_manager manager; struct drm_mm mm; spinlock_t lock; + struct list_head reservations_pending; + struct list_head reserved_pages; atomic64_t usage; atomic64_t vis_usage; }; @@ -54,7 +61,6 @@ struct amdgpu_gtt_mgr { struct amdgpu_mman { struct ttm_bo_device bdev; - bool mem_global_referenced; bool initialized; void __iomem *aper_base_kaddr; @@ -119,6 +125,10 @@ void amdgpu_vram_mgr_free_sgt(struct amdgpu_device *adev, struct sg_table *sgt); uint64_t amdgpu_vram_mgr_usage(struct ttm_resource_manager *man); uint64_t amdgpu_vram_mgr_vis_usage(struct ttm_resource_manager *man); +int amdgpu_vram_mgr_reserve_range(struct ttm_resource_manager *man, + uint64_t start, uint64_t size); +int amdgpu_vram_mgr_query_page_status(struct ttm_resource_manager *man, + uint64_t start); int amdgpu_ttm_init(struct amdgpu_device *adev); void amdgpu_ttm_late_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c index b313ce4c3e97..2b7c90b7a712 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c @@ -391,6 +391,8 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type) case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: if (!load_type) return AMDGPU_FW_LOAD_DIRECT; else @@ -586,8 +588,8 @@ static int amdgpu_ucode_patch_jt(struct amdgpu_firmware_info *ucode, { const struct gfx_firmware_header_v1_0 *header = NULL; const struct common_firmware_header *comm_hdr = NULL; - uint8_t* src_addr = NULL; - uint8_t* dst_addr = NULL; + uint8_t *src_addr = NULL; + uint8_t *dst_addr = NULL; if (NULL == ucode->fw) return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index 262baf0f61ea..a2975c8092a9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -126,10 +126,11 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, err_data->ue_count); if ((amdgpu_bad_page_threshold != 0) && - err_data->err_addr_cnt && + err_data->err_addr_cnt) { amdgpu_ras_add_bad_pages(adev, err_data->err_addr, - err_data->err_addr_cnt)) - dev_warn(adev->dev, "Failed to add ras bad page!\n"); + err_data->err_addr_cnt); + amdgpu_ras_save_bad_pages(adev); + } amdgpu_ras_reset_gpu(adev); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index f3b7287e84c4..7e19a6656715 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -37,24 +37,30 @@ #define FIRMWARE_RAVEN "amdgpu/raven_vcn.bin" #define FIRMWARE_PICASSO "amdgpu/picasso_vcn.bin" #define FIRMWARE_RAVEN2 "amdgpu/raven2_vcn.bin" -#define FIRMWARE_ARCTURUS "amdgpu/arcturus_vcn.bin" -#define FIRMWARE_RENOIR "amdgpu/renoir_vcn.bin" -#define FIRMWARE_NAVI10 "amdgpu/navi10_vcn.bin" -#define FIRMWARE_NAVI14 "amdgpu/navi14_vcn.bin" -#define FIRMWARE_NAVI12 "amdgpu/navi12_vcn.bin" -#define FIRMWARE_SIENNA_CICHLID "amdgpu/sienna_cichlid_vcn.bin" -#define FIRMWARE_NAVY_FLOUNDER "amdgpu/navy_flounder_vcn.bin" +#define FIRMWARE_ARCTURUS "amdgpu/arcturus_vcn.bin" +#define FIRMWARE_RENOIR "amdgpu/renoir_vcn.bin" +#define FIRMWARE_GREEN_SARDINE "amdgpu/green_sardine_vcn.bin" +#define FIRMWARE_NAVI10 "amdgpu/navi10_vcn.bin" +#define FIRMWARE_NAVI14 "amdgpu/navi14_vcn.bin" +#define FIRMWARE_NAVI12 "amdgpu/navi12_vcn.bin" +#define FIRMWARE_SIENNA_CICHLID "amdgpu/sienna_cichlid_vcn.bin" +#define FIRMWARE_NAVY_FLOUNDER "amdgpu/navy_flounder_vcn.bin" +#define FIRMWARE_VANGOGH "amdgpu/vangogh_vcn.bin" +#define FIRMWARE_DIMGREY_CAVEFISH "amdgpu/dimgrey_cavefish_vcn.bin" MODULE_FIRMWARE(FIRMWARE_RAVEN); MODULE_FIRMWARE(FIRMWARE_PICASSO); MODULE_FIRMWARE(FIRMWARE_RAVEN2); MODULE_FIRMWARE(FIRMWARE_ARCTURUS); MODULE_FIRMWARE(FIRMWARE_RENOIR); +MODULE_FIRMWARE(FIRMWARE_GREEN_SARDINE); MODULE_FIRMWARE(FIRMWARE_NAVI10); MODULE_FIRMWARE(FIRMWARE_NAVI14); MODULE_FIRMWARE(FIRMWARE_NAVI12); MODULE_FIRMWARE(FIRMWARE_SIENNA_CICHLID); MODULE_FIRMWARE(FIRMWARE_NAVY_FLOUNDER); +MODULE_FIRMWARE(FIRMWARE_VANGOGH); +MODULE_FIRMWARE(FIRMWARE_DIMGREY_CAVEFISH); static void amdgpu_vcn_idle_work_handler(struct work_struct *work); @@ -89,7 +95,11 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) adev->vcn.indirect_sram = true; break; case CHIP_RENOIR: - fw_name = FIRMWARE_RENOIR; + if (adev->apu_flags & AMD_APU_IS_RENOIR) + fw_name = FIRMWARE_RENOIR; + else + fw_name = FIRMWARE_GREEN_SARDINE; + if ((adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) && (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)) adev->vcn.indirect_sram = true; @@ -124,6 +134,15 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)) adev->vcn.indirect_sram = true; break; + case CHIP_VANGOGH: + fw_name = FIRMWARE_VANGOGH; + break; + case CHIP_DIMGREY_CAVEFISH: + fw_name = FIRMWARE_DIMGREY_CAVEFISH; + if ((adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) && + (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)) + adev->vcn.indirect_sram = true; + break; default: return -EINVAL; } @@ -313,6 +332,7 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work) container_of(work, struct amdgpu_device, vcn.idle_work.work); unsigned int fences = 0, fence[AMDGPU_MAX_VCN_INSTANCES] = {0}; unsigned int i, j; + int r = 0; for (j = 0; j < adev->vcn.num_vcn_inst; ++j) { if (adev->vcn.harvest_config & (1 << j)) @@ -339,8 +359,13 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work) } if (!fences && !atomic_read(&adev->vcn.total_submission_cnt)) { + amdgpu_gfx_off_ctrl(adev, true); 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.idle_work, VCN_IDLE_TIMEOUT); } @@ -349,9 +374,17 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work) void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; + int r = 0; atomic_inc(&adev->vcn.total_submission_cnt); - cancel_delayed_work_sync(&adev->vcn.idle_work); + + if (!cancel_delayed_work_sync(&adev->vcn.idle_work)) { + amdgpu_gfx_off_ctrl(adev, false); + 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.vcn_pg_lock); amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.c index 7f7097931c6f..f9d3d79f68b1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.c @@ -59,7 +59,7 @@ void amdgpu_vf_error_trans_all(struct amdgpu_device *adev) return; } /* - TODO: Enable these code when pv2vf_info is merged + TODO: Enable these code when pv2vf_info is merged AMDGPU_FW_VRAM_PF2VF_READ (adev, feature_flags, &pf2vf_flags); if (!(pf2vf_flags & AMDGIM_FEATURE_ERROR_LOG_COLLECT)) { return; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 38b59a4fc04c..0768c8686983 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -300,7 +300,7 @@ static void amdgpu_vm_bo_relocated(struct amdgpu_vm_bo_base *vm_bo) static void amdgpu_vm_bo_done(struct amdgpu_vm_bo_base *vm_bo) { spin_lock(&vm_bo->vm->invalidated_lock); - list_del_init(&vm_bo->vm_status); + list_move(&vm_bo->vm_status, &vm_bo->vm->done); spin_unlock(&vm_bo->vm->invalidated_lock); } @@ -1570,7 +1570,8 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params, /** * amdgpu_vm_bo_update_mapping - update a mapping in the vm page table * - * @adev: amdgpu_device pointer + * @adev: amdgpu_device pointer of the VM + * @bo_adev: amdgpu_device pointer of the mapped BO * @vm: requested vm * @immediate: immediate submission in a page fault * @unlocked: unlocked invalidation during MM callback @@ -1578,7 +1579,8 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params, * @start: start of mapped range * @last: last mapped entry * @flags: flags for the entries - * @addr: addr to set the area to + * @offset: offset into nodes and pages_addr + * @nodes: array of drm_mm_nodes with the MC addresses * @pages_addr: DMA addresses to use for mapping * @fence: optional resulting fence * @@ -1588,15 +1590,18 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params, * 0 for success, -EINVAL for failure. */ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev, + struct amdgpu_device *bo_adev, struct amdgpu_vm *vm, bool immediate, bool unlocked, struct dma_resv *resv, uint64_t start, uint64_t last, - uint64_t flags, uint64_t addr, + uint64_t flags, uint64_t offset, + struct drm_mm_node *nodes, dma_addr_t *pages_addr, struct dma_fence **fence) { struct amdgpu_vm_update_params params; enum amdgpu_sync_mode sync_mode; + uint64_t pfn; int r; memset(¶ms, 0, sizeof(params)); @@ -1614,6 +1619,14 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev, else sync_mode = AMDGPU_SYNC_EXPLICIT; + pfn = offset >> PAGE_SHIFT; + if (nodes) { + while (pfn >= nodes->size) { + pfn -= nodes->size; + ++nodes; + } + } + amdgpu_vm_eviction_lock(vm); if (vm->evicting) { r = -EBUSY; @@ -1632,105 +1645,47 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev, if (r) goto error_unlock; - r = amdgpu_vm_update_ptes(¶ms, start, last + 1, addr, flags); - if (r) - goto error_unlock; - - r = vm->update_funcs->commit(¶ms, fence); - -error_unlock: - amdgpu_vm_eviction_unlock(vm); - return r; -} - -/** - * amdgpu_vm_bo_split_mapping - split a mapping into smaller chunks - * - * @adev: amdgpu_device pointer - * @resv: fences we need to sync to - * @pages_addr: DMA addresses to use for mapping - * @vm: requested vm - * @mapping: mapped range and flags to use for the update - * @flags: HW flags for the mapping - * @bo_adev: amdgpu_device pointer that bo actually been allocated - * @nodes: array of drm_mm_nodes with the MC addresses - * @fence: optional resulting fence - * - * Split the mapping into smaller chunks so that each update fits - * into a SDMA IB. - * - * Returns: - * 0 for success, -EINVAL for failure. - */ -static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev, - struct dma_resv *resv, - dma_addr_t *pages_addr, - struct amdgpu_vm *vm, - struct amdgpu_bo_va_mapping *mapping, - uint64_t flags, - struct amdgpu_device *bo_adev, - struct drm_mm_node *nodes, - struct dma_fence **fence) -{ - unsigned min_linear_pages = 1 << adev->vm_manager.fragment_size; - uint64_t pfn, start = mapping->start; - int r; - - /* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here - * but in case of something, we filter the flags in first place - */ - if (!(mapping->flags & AMDGPU_PTE_READABLE)) - flags &= ~AMDGPU_PTE_READABLE; - if (!(mapping->flags & AMDGPU_PTE_WRITEABLE)) - flags &= ~AMDGPU_PTE_WRITEABLE; - - /* Apply ASIC specific mapping flags */ - amdgpu_gmc_get_vm_pte(adev, mapping, &flags); - - trace_amdgpu_vm_bo_update(mapping); - - pfn = mapping->offset >> PAGE_SHIFT; - if (nodes) { - while (pfn >= nodes->size) { - pfn -= nodes->size; - ++nodes; - } - } - do { - dma_addr_t *dma_addr = NULL; - uint64_t max_entries; - uint64_t addr, last; + uint64_t tmp, num_entries, addr; + - max_entries = mapping->last - start + 1; + num_entries = last - start + 1; if (nodes) { addr = nodes->start << PAGE_SHIFT; - max_entries = min((nodes->size - pfn) * - AMDGPU_GPU_PAGES_IN_CPU_PAGE, max_entries); + num_entries = min((nodes->size - pfn) * + AMDGPU_GPU_PAGES_IN_CPU_PAGE, num_entries); } else { addr = 0; } if (pages_addr) { - uint64_t count; + bool contiguous = true; - for (count = 1; - count < max_entries / AMDGPU_GPU_PAGES_IN_CPU_PAGE; - ++count) { - uint64_t idx = pfn + count; + if (num_entries > AMDGPU_GPU_PAGES_IN_CPU_PAGE) { + uint64_t count; - if (pages_addr[idx] != - (pages_addr[idx - 1] + PAGE_SIZE)) - break; + contiguous = pages_addr[pfn + 1] == + pages_addr[pfn] + PAGE_SIZE; + + tmp = num_entries / + AMDGPU_GPU_PAGES_IN_CPU_PAGE; + for (count = 2; count < tmp; ++count) { + uint64_t idx = pfn + count; + + if (contiguous != (pages_addr[idx] == + pages_addr[idx - 1] + PAGE_SIZE)) + break; + } + num_entries = count * + AMDGPU_GPU_PAGES_IN_CPU_PAGE; } - if (count < min_linear_pages) { + if (!contiguous) { addr = pfn << PAGE_SHIFT; - dma_addr = pages_addr; + params.pages_addr = pages_addr; } else { addr = pages_addr[pfn]; - max_entries = count * - AMDGPU_GPU_PAGES_IN_CPU_PAGE; + params.pages_addr = NULL; } } else if (flags & (AMDGPU_PTE_VALID | AMDGPU_PTE_PRT)) { @@ -1738,23 +1693,25 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev, addr += pfn << PAGE_SHIFT; } - last = start + max_entries - 1; - r = amdgpu_vm_bo_update_mapping(adev, vm, false, false, resv, - start, last, flags, addr, - dma_addr, fence); + tmp = start + num_entries; + r = amdgpu_vm_update_ptes(¶ms, start, tmp, addr, flags); if (r) - return r; + goto error_unlock; - pfn += (last - start + 1) / AMDGPU_GPU_PAGES_IN_CPU_PAGE; + pfn += num_entries / AMDGPU_GPU_PAGES_IN_CPU_PAGE; if (nodes && nodes->size == pfn) { pfn = 0; ++nodes; } - start = last + 1; + start = tmp; - } while (unlikely(start != mapping->last + 1)); + } while (unlikely(start != last + 1)); - return 0; + r = vm->update_funcs->commit(¶ms, fence); + +error_unlock: + amdgpu_vm_eviction_unlock(vm); + return r; } /** @@ -1832,9 +1789,26 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va, } list_for_each_entry(mapping, &bo_va->invalids, list) { - r = amdgpu_vm_bo_split_mapping(adev, resv, pages_addr, vm, - mapping, flags, bo_adev, nodes, - last_update); + uint64_t update_flags = flags; + + /* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here + * but in case of something, we filter the flags in first place + */ + if (!(mapping->flags & AMDGPU_PTE_READABLE)) + update_flags &= ~AMDGPU_PTE_READABLE; + if (!(mapping->flags & AMDGPU_PTE_WRITEABLE)) + update_flags &= ~AMDGPU_PTE_WRITEABLE; + + /* Apply ASIC specific mapping flags */ + amdgpu_gmc_get_vm_pte(adev, mapping, &update_flags); + + trace_amdgpu_vm_bo_update(mapping); + + r = amdgpu_vm_bo_update_mapping(adev, bo_adev, vm, false, false, + resv, mapping->start, + mapping->last, update_flags, + mapping->offset, nodes, + pages_addr, last_update); if (r) return r; } @@ -2042,9 +2016,10 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev, mapping->start < AMDGPU_GMC_HOLE_START) init_pte_value = AMDGPU_PTE_DEFAULT_ATC; - r = amdgpu_vm_bo_update_mapping(adev, vm, false, false, resv, - mapping->start, mapping->last, - init_pte_value, 0, NULL, &f); + r = amdgpu_vm_bo_update_mapping(adev, adev, vm, false, false, + resv, mapping->start, + mapping->last, init_pte_value, + 0, NULL, NULL, &f); amdgpu_vm_free_mapping(adev, vm, mapping, f); if (r) { dma_fence_put(f); @@ -2163,7 +2138,7 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev, /** - * amdgpu_vm_bo_insert_mapping - insert a new mapping + * amdgpu_vm_bo_insert_map - insert a new mapping * * @adev: amdgpu_device pointer * @bo_va: bo_va to store the address @@ -2820,7 +2795,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, INIT_LIST_HEAD(&vm->invalidated); spin_lock_init(&vm->invalidated_lock); INIT_LIST_HEAD(&vm->freed); - + INIT_LIST_HEAD(&vm->done); /* create scheduler entities for page table updates */ r = drm_sched_entity_init(&vm->immediate, DRM_SCHED_PRIORITY_NORMAL, @@ -3372,8 +3347,9 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, value = 0; } - r = amdgpu_vm_bo_update_mapping(adev, vm, true, false, NULL, addr, - addr + 1, flags, value, NULL, NULL); + r = amdgpu_vm_bo_update_mapping(adev, adev, vm, true, false, NULL, addr, + addr, flags, value, NULL, NULL, + NULL); if (r) goto error_unlock; @@ -3389,3 +3365,99 @@ error_unref: return false; } + +#if defined(CONFIG_DEBUG_FS) +/** + * amdgpu_debugfs_vm_bo_info - print BO info for the VM + * + * @vm: Requested VM for printing BO info + * @m: debugfs file + * + * Print BO information in debugfs file for the VM + */ +void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m) +{ + struct amdgpu_bo_va *bo_va, *tmp; + u64 total_idle = 0; + u64 total_evicted = 0; + u64 total_relocated = 0; + u64 total_moved = 0; + u64 total_invalidated = 0; + u64 total_done = 0; + unsigned int total_idle_objs = 0; + unsigned int total_evicted_objs = 0; + unsigned int total_relocated_objs = 0; + unsigned int total_moved_objs = 0; + unsigned int total_invalidated_objs = 0; + unsigned int total_done_objs = 0; + unsigned int id = 0; + + seq_puts(m, "\tIdle BOs:\n"); + list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status) { + if (!bo_va->base.bo) + continue; + total_idle += amdgpu_bo_print_info(id++, bo_va->base.bo, m); + } + total_idle_objs = id; + id = 0; + + seq_puts(m, "\tEvicted BOs:\n"); + list_for_each_entry_safe(bo_va, tmp, &vm->evicted, base.vm_status) { + if (!bo_va->base.bo) + continue; + total_evicted += amdgpu_bo_print_info(id++, bo_va->base.bo, m); + } + total_evicted_objs = id; + id = 0; + + seq_puts(m, "\tRelocated BOs:\n"); + list_for_each_entry_safe(bo_va, tmp, &vm->relocated, base.vm_status) { + if (!bo_va->base.bo) + continue; + total_relocated += amdgpu_bo_print_info(id++, bo_va->base.bo, m); + } + total_relocated_objs = id; + id = 0; + + seq_puts(m, "\tMoved BOs:\n"); + list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) { + if (!bo_va->base.bo) + continue; + total_moved += amdgpu_bo_print_info(id++, bo_va->base.bo, m); + } + total_moved_objs = id; + id = 0; + + seq_puts(m, "\tInvalidated BOs:\n"); + spin_lock(&vm->invalidated_lock); + list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status) { + if (!bo_va->base.bo) + continue; + total_invalidated += amdgpu_bo_print_info(id++, bo_va->base.bo, m); + } + total_invalidated_objs = id; + id = 0; + + seq_puts(m, "\tDone BOs:\n"); + list_for_each_entry_safe(bo_va, tmp, &vm->done, base.vm_status) { + if (!bo_va->base.bo) + continue; + total_done += amdgpu_bo_print_info(id++, bo_va->base.bo, m); + } + spin_unlock(&vm->invalidated_lock); + total_done_objs = id; + + seq_printf(m, "\tTotal idle size: %12lld\tobjs:\t%d\n", total_idle, + total_idle_objs); + seq_printf(m, "\tTotal evicted size: %12lld\tobjs:\t%d\n", total_evicted, + total_evicted_objs); + seq_printf(m, "\tTotal relocated size: %12lld\tobjs:\t%d\n", total_relocated, + total_relocated_objs); + seq_printf(m, "\tTotal moved size: %12lld\tobjs:\t%d\n", total_moved, + total_moved_objs); + seq_printf(m, "\tTotal invalidated size: %12lld\tobjs:\t%d\n", total_invalidated, + total_invalidated_objs); + seq_printf(m, "\tTotal done size: %12lld\tobjs:\t%d\n", total_done, + total_done_objs); +} +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 58c83a7ad0fd..ca2344beed81 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -76,6 +76,9 @@ struct amdgpu_bo_list_entry; /* PTE is handled as PDE for VEGA10 (Translate Further) */ #define AMDGPU_PTE_TF (1ULL << 56) +/* MALL noalloc for sienna_cichlid, reserved for older ASICs */ +#define AMDGPU_PTE_NOALLOC (1ULL << 58) + /* PDE Block Fragment Size for VEGA10 */ #define AMDGPU_PDE_BFS(a) ((uint64_t)a << 59) @@ -274,6 +277,9 @@ struct amdgpu_vm { /* BO mappings freed, but not yet updated in the PT */ struct list_head freed; + /* BOs which are invalidated, has been updated in the PTs */ + struct list_head done; + /* contains the page directory */ struct amdgpu_vm_pt root; struct dma_fence *last_update; @@ -441,4 +447,8 @@ void amdgpu_vm_move_to_lru_tail(struct amdgpu_device *adev, struct amdgpu_vm *vm); void amdgpu_vm_del_from_lru_notify(struct ttm_buffer_object *bo); +#if defined(CONFIG_DEBUG_FS) +void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m); +#endif + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index 7747be644dd0..c99c2180785f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -168,8 +168,7 @@ static const struct ttm_resource_manager_func amdgpu_vram_mgr_func; /** * amdgpu_vram_mgr_init - init VRAM manager and DRM MM * - * @man: TTM memory type manager - * @p_size: maximum size of VRAM + * @adev: amdgpu_device pointer * * Allocate and initialize the VRAM manager. */ @@ -185,6 +184,8 @@ int amdgpu_vram_mgr_init(struct amdgpu_device *adev) drm_mm_init(&mgr->mm, 0, man->size); spin_lock_init(&mgr->lock); + INIT_LIST_HEAD(&mgr->reservations_pending); + INIT_LIST_HEAD(&mgr->reserved_pages); /* Add the two VRAM-related sysfs files */ ret = sysfs_create_files(&adev->dev->kobj, amdgpu_vram_mgr_attributes); @@ -199,7 +200,7 @@ int amdgpu_vram_mgr_init(struct amdgpu_device *adev) /** * amdgpu_vram_mgr_fini - free and destroy VRAM manager * - * @man: TTM memory type manager + * @adev: amdgpu_device pointer * * Destroy and free the VRAM manager, returns -EBUSY if ranges are still * allocated inside it. @@ -209,6 +210,7 @@ void amdgpu_vram_mgr_fini(struct amdgpu_device *adev) struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr; struct ttm_resource_manager *man = &mgr->manager; int ret; + struct amdgpu_vram_reservation *rsv, *temp; ttm_resource_manager_set_used(man, false); @@ -217,6 +219,13 @@ void amdgpu_vram_mgr_fini(struct amdgpu_device *adev) return; spin_lock(&mgr->lock); + list_for_each_entry_safe(rsv, temp, &mgr->reservations_pending, node) + kfree(rsv); + + list_for_each_entry_safe(rsv, temp, &mgr->reserved_pages, node) { + drm_mm_remove_node(&rsv->mm_node); + kfree(rsv); + } drm_mm_takedown(&mgr->mm); spin_unlock(&mgr->lock); @@ -229,7 +238,7 @@ void amdgpu_vram_mgr_fini(struct amdgpu_device *adev) /** * amdgpu_vram_mgr_vis_size - Calculate visible node size * - * @adev: amdgpu device structure + * @adev: amdgpu_device pointer * @node: MM node structure * * Calculate how many bytes of the MM node are inside visible VRAM @@ -275,6 +284,101 @@ u64 amdgpu_vram_mgr_bo_visible_size(struct amdgpu_bo *bo) return usage; } +static void amdgpu_vram_mgr_do_reserve(struct ttm_resource_manager *man) +{ + struct amdgpu_vram_mgr *mgr = to_vram_mgr(man); + struct amdgpu_device *adev = to_amdgpu_device(mgr); + struct drm_mm *mm = &mgr->mm; + struct amdgpu_vram_reservation *rsv, *temp; + uint64_t vis_usage; + + list_for_each_entry_safe(rsv, temp, &mgr->reservations_pending, node) { + if (drm_mm_reserve_node(mm, &rsv->mm_node)) + continue; + + dev_dbg(adev->dev, "Reservation 0x%llx - %lld, Succeeded\n", + rsv->mm_node.start, rsv->mm_node.size); + + vis_usage = amdgpu_vram_mgr_vis_size(adev, &rsv->mm_node); + atomic64_add(vis_usage, &mgr->vis_usage); + atomic64_add(rsv->mm_node.size << PAGE_SHIFT, &mgr->usage); + list_move(&rsv->node, &mgr->reserved_pages); + } +} + +/** + * amdgpu_vram_mgr_reserve_range - Reserve a range from VRAM + * + * @man: TTM memory type manager + * @start: start address of the range in VRAM + * @size: size of the range + * + * Reserve memory from start addess with the specified size in VRAM + */ +int amdgpu_vram_mgr_reserve_range(struct ttm_resource_manager *man, + uint64_t start, uint64_t size) +{ + struct amdgpu_vram_mgr *mgr = to_vram_mgr(man); + struct amdgpu_vram_reservation *rsv; + + rsv = kzalloc(sizeof(*rsv), GFP_KERNEL); + if (!rsv) + return -ENOMEM; + + INIT_LIST_HEAD(&rsv->node); + rsv->mm_node.start = start >> PAGE_SHIFT; + rsv->mm_node.size = size >> PAGE_SHIFT; + + spin_lock(&mgr->lock); + list_add_tail(&mgr->reservations_pending, &rsv->node); + amdgpu_vram_mgr_do_reserve(man); + spin_unlock(&mgr->lock); + + return 0; +} + +/** + * amdgpu_vram_mgr_query_page_status - query the reservation status + * + * @man: TTM memory type manager + * @start: start address of a page in VRAM + * + * Returns: + * -EBUSY: the page is still hold and in pending list + * 0: the page has been reserved + * -ENOENT: the input page is not a reservation + */ +int amdgpu_vram_mgr_query_page_status(struct ttm_resource_manager *man, + uint64_t start) +{ + struct amdgpu_vram_mgr *mgr = to_vram_mgr(man); + struct amdgpu_vram_reservation *rsv; + int ret; + + spin_lock(&mgr->lock); + + list_for_each_entry(rsv, &mgr->reservations_pending, node) { + if ((rsv->mm_node.start <= start) && + (start < (rsv->mm_node.start + rsv->mm_node.size))) { + ret = -EBUSY; + goto out; + } + } + + list_for_each_entry(rsv, &mgr->reserved_pages, node) { + if ((rsv->mm_node.start <= start) && + (start < (rsv->mm_node.start + rsv->mm_node.size))) { + ret = 0; + goto out; + } + } + + ret = -ENOENT; +out: + spin_unlock(&mgr->lock); + return ret; +} + /** * amdgpu_vram_mgr_virt_start - update virtual start address * @@ -445,6 +549,7 @@ static void amdgpu_vram_mgr_del(struct ttm_resource_manager *man, vis_usage += amdgpu_vram_mgr_vis_size(adev, nodes); ++nodes; } + amdgpu_vram_mgr_do_reserve(man); spin_unlock(&mgr->lock); atomic64_sub(usage, &mgr->usage); @@ -529,7 +634,7 @@ error_free: } /** - * amdgpu_vram_mgr_alloc_sgt - allocate and fill a sg table + * amdgpu_vram_mgr_free_sgt - allocate and fill a sg table * * @adev: amdgpu device pointer * @sgt: sg table to free diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 1162913c8bf4..54c358e71f9a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -246,7 +246,7 @@ static ssize_t amdgpu_xgmi_show_error(struct device *dev, adev->df.funcs->set_fica(adev, ficaa_pie_status_in, 0, 0); - return snprintf(buf, PAGE_SIZE, "%d\n", error_count); + return snprintf(buf, PAGE_SIZE, "%u\n", error_count); } diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c index 939eca63b094..66c183ddd43e 100644 --- a/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c +++ b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c @@ -74,6 +74,7 @@ int athub_v2_1_set_clockgating(struct amdgpu_device *adev, switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: athub_v2_1_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE ? true : false); athub_v2_1_update_medium_grain_light_sleep(adev, diff --git a/drivers/gpu/drm/amd/amdgpu/atom.c b/drivers/gpu/drm/amd/amdgpu/atom.c index 4cfc786699c7..515890f4f5a0 100644 --- a/drivers/gpu/drm/amd/amdgpu/atom.c +++ b/drivers/gpu/drm/amd/amdgpu/atom.c @@ -66,13 +66,13 @@ typedef struct { bool abort; } atom_exec_context; -int amdgpu_atom_debug = 0; -static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params); -int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t * params); +int amdgpu_atom_debug; +static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t *params); +int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params); static uint32_t atom_arg_mask[8] = - { 0xFFFFFFFF, 0xFFFF, 0xFFFF00, 0xFFFF0000, 0xFF, 0xFF00, 0xFF0000, -0xFF000000 }; + { 0xFFFFFFFF, 0xFFFF, 0xFFFF00, 0xFFFF0000, 0xFF, 0xFF00, 0xFF0000, + 0xFF000000 }; static int atom_arg_shift[8] = { 0, 0, 8, 16, 0, 8, 16, 24 }; static int atom_dst_to_src[8][4] = { @@ -88,7 +88,7 @@ static int atom_dst_to_src[8][4] = { }; static int atom_def_dst[8] = { 0, 0, 1, 2, 0, 1, 2, 3 }; -static int debug_depth = 0; +static int debug_depth; #ifdef ATOM_DEBUG static void debug_print_spaces(int n) { @@ -1201,7 +1201,7 @@ static struct { atom_op_div32, ATOM_ARG_WS}, }; -static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params) +static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t *params) { int base = CU16(ctx->cmd_table + 4 + 2 * index); int len, ws, ps, ptr; @@ -1262,7 +1262,7 @@ free: return ret; } -int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t * params) +int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params) { int r; @@ -1388,8 +1388,8 @@ void amdgpu_atom_destroy(struct atom_context *ctx) } bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index, - uint16_t * size, uint8_t * frev, uint8_t * crev, - uint16_t * data_start) + uint16_t *size, uint8_t *frev, uint8_t *crev, + uint16_t *data_start) { int offset = index * 2 + 4; int idx = CU16(ctx->data_table + offset); @@ -1408,8 +1408,8 @@ bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index, return true; } -bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index, uint8_t * frev, - uint8_t * crev) +bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index, uint8_t *frev, + uint8_t *crev) { int offset = index * 2 + 4; int idx = CU16(ctx->cmd_table + offset); diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c index 8339c8c3a328..fa817ebff980 100644 --- a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c +++ b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c @@ -499,10 +499,8 @@ int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder) } else { return ATOM_ENCODER_MODE_DVI; } - break; case DRM_MODE_CONNECTOR_LVDS: return ATOM_ENCODER_MODE_LVDS; - break; case DRM_MODE_CONNECTOR_DisplayPort: dig_connector = amdgpu_connector->con_priv; if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) || @@ -519,20 +517,16 @@ int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder) } else { return ATOM_ENCODER_MODE_DVI; } - break; case DRM_MODE_CONNECTOR_eDP: return ATOM_ENCODER_MODE_DP; case DRM_MODE_CONNECTOR_DVIA: case DRM_MODE_CONNECTOR_VGA: return ATOM_ENCODER_MODE_CRT; - break; case DRM_MODE_CONNECTOR_Composite: case DRM_MODE_CONNECTOR_SVIDEO: case DRM_MODE_CONNECTOR_9PinDIN: /* fix me */ return ATOM_ENCODER_MODE_TV; - /*return ATOM_ENCODER_MODE_CV;*/ - break; } } diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c b/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c index 09a538465ffd..af0335535f82 100644 --- a/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c +++ b/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c @@ -159,7 +159,7 @@ u32 amdgpu_atombios_i2c_func(struct i2c_adapter *adap) return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL; } -void amdgpu_atombios_i2c_channel_trans(struct amdgpu_device* adev, u8 slave_addr, u8 line_number, u8 offset, u8 data) +void amdgpu_atombios_i2c_channel_trans(struct amdgpu_device *adev, u8 slave_addr, u8 line_number, u8 offset, u8 data) { PROCESS_I2C_CHANNEL_TRANSACTION_PS_ALLOCATION args; int index = GetIndexIntoMasterTable(COMMAND, ProcessI2cChannelTransaction); diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index 03ff8bd1fee8..5442df094102 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik.c +++ b/drivers/gpu/drm/amd/amdgpu/cik.c @@ -1336,11 +1336,13 @@ cik_asic_reset_method(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_BONAIRE: - case CHIP_HAWAII: /* disable baco reset until it works */ /* smu7_asic_get_baco_capability(adev, &baco_reset); */ baco_reset = false; break; + case CHIP_HAWAII: + baco_reset = cik_asic_supports_baco(adev); + break; default: baco_reset = false; break; diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c index 20f108818b2b..1a6494ea5091 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c @@ -195,7 +195,7 @@ static void cik_sdma_ring_set_wptr(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], - (lower_32_bits(ring->wptr) << 2) & 0x3fffc); + (lower_32_bits(ring->wptr) << 2) & 0x3fffc); } static void cik_sdma_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) @@ -1071,22 +1071,19 @@ static int cik_sdma_soft_reset(void *handle) { u32 srbm_soft_reset = 0; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - u32 tmp = RREG32(mmSRBM_STATUS2); + u32 tmp; - if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) { - /* sdma0 */ - tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET); - tmp |= SDMA0_F32_CNTL__HALT_MASK; - WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp); - srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK; - } - if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK) { - /* sdma1 */ - tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET); - tmp |= SDMA0_F32_CNTL__HALT_MASK; - WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp); - srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK; - } + /* sdma0 */ + tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET); + tmp |= SDMA0_F32_CNTL__HALT_MASK; + WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp); + srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK; + + /* sdma1 */ + tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET); + tmp |= SDMA0_F32_CNTL__HALT_MASK; + WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp); + srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK; if (srbm_soft_reset) { tmp = RREG32(mmSRBM_SOFT_RESET); diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c index 5963cbe0d455..da240f8fafcf 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c @@ -2202,22 +2202,18 @@ static int dce_v10_0_pick_dig_encoder(struct drm_encoder *encoder) return 1; else return 0; - break; case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: if (dig->linkb) return 3; else return 2; - break; case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: if (dig->linkb) return 5; else return 4; - break; case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3: return 6; - break; default: DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id); return 0; @@ -2677,7 +2673,7 @@ static int dce_v10_0_crtc_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, int x, int y, enum mode_set_atomic state) { - return dce_v10_0_crtc_do_set_base(crtc, fb, x, y, 1); + return dce_v10_0_crtc_do_set_base(crtc, fb, x, y, 1); } static const struct drm_crtc_helper_funcs dce_v10_0_crtc_helper_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c index 1954472c8e8f..c62c56a69fda 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c @@ -2235,22 +2235,18 @@ static int dce_v11_0_pick_dig_encoder(struct drm_encoder *encoder) return 1; else return 0; - break; case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: if (dig->linkb) return 3; else return 2; - break; case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: if (dig->linkb) return 5; else return 4; - break; case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3: return 6; - break; default: DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id); return 0; @@ -2304,19 +2300,16 @@ static u32 dce_v11_0_pick_pll(struct drm_crtc *crtc) return ATOM_COMBOPHY_PLL1; else return ATOM_COMBOPHY_PLL0; - break; case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: if (dig->linkb) return ATOM_COMBOPHY_PLL3; else return ATOM_COMBOPHY_PLL2; - break; case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: if (dig->linkb) return ATOM_COMBOPHY_PLL5; else return ATOM_COMBOPHY_PLL4; - break; default: DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id); return ATOM_PPLL_INVALID; @@ -2785,7 +2778,7 @@ static int dce_v11_0_crtc_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, int x, int y, enum mode_set_atomic state) { - return dce_v11_0_crtc_do_set_base(crtc, fb, x, y, 1); + return dce_v11_0_crtc_do_set_base(crtc, fb, x, y, 1); } static const struct drm_crtc_helper_funcs dce_v11_0_crtc_helper_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c index 3a44753a80d1..943976349346 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c @@ -2567,7 +2567,7 @@ static int dce_v6_0_crtc_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, int x, int y, enum mode_set_atomic state) { - return dce_v6_0_crtc_do_set_base(crtc, fb, x, y, 1); + return dce_v6_0_crtc_do_set_base(crtc, fb, x, y, 1); } static const struct drm_crtc_helper_funcs dce_v6_0_crtc_helper_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c index 3603e5f13077..7973183fa335 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c @@ -2498,7 +2498,7 @@ static void dce_v8_0_crtc_disable(struct drm_crtc *crtc) case ATOM_PPLL2: /* disable the ppll */ amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id, - 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss); + 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss); break; case ATOM_PPLL0: /* disable the ppll */ @@ -2585,7 +2585,7 @@ static int dce_v8_0_crtc_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, int x, int y, enum mode_set_atomic state) { - return dce_v8_0_crtc_do_set_base(crtc, fb, x, y, 1); + return dce_v8_0_crtc_do_set_base(crtc, fb, x, y, 1); } static const struct drm_crtc_helper_funcs dce_v8_0_crtc_helper_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/df_v1_7.c b/drivers/gpu/drm/amd/amdgpu/df_v1_7.c index d6aca1c08068..2d01ac0d4c11 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v1_7.c +++ b/drivers/gpu/drm/amd/amdgpu/df_v1_7.c @@ -41,7 +41,7 @@ static void df_v1_7_sw_fini(struct amdgpu_device *adev) } static void df_v1_7_enable_broadcast_mode(struct amdgpu_device *adev, - bool enable) + bool enable) { u32 tmp; diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c index 7b89fd2aa44a..6b4b30a8dce5 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c @@ -30,71 +30,17 @@ #define DF_3_6_SMN_REG_INST_DIST 0x8 #define DF_3_6_INST_CNT 8 -static u32 df_v3_6_channel_number[] = {1, 2, 0, 4, 0, 8, 0, - 16, 32, 0, 0, 0, 2, 4, 8}; - -/* init df format attrs */ -AMDGPU_PMU_ATTR(event, "config:0-7"); -AMDGPU_PMU_ATTR(instance, "config:8-15"); -AMDGPU_PMU_ATTR(umask, "config:16-23"); - -/* df format attributes */ -static struct attribute *df_v3_6_format_attrs[] = { - &pmu_attr_event.attr, - &pmu_attr_instance.attr, - &pmu_attr_umask.attr, - NULL -}; - -/* df format attribute group */ -static struct attribute_group df_v3_6_format_attr_group = { - .name = "format", - .attrs = df_v3_6_format_attrs, -}; - -/* df event attrs */ -AMDGPU_PMU_ATTR(cake0_pcsout_txdata, - "event=0x7,instance=0x46,umask=0x2"); -AMDGPU_PMU_ATTR(cake1_pcsout_txdata, - "event=0x7,instance=0x47,umask=0x2"); -AMDGPU_PMU_ATTR(cake0_pcsout_txmeta, - "event=0x7,instance=0x46,umask=0x4"); -AMDGPU_PMU_ATTR(cake1_pcsout_txmeta, - "event=0x7,instance=0x47,umask=0x4"); -AMDGPU_PMU_ATTR(cake0_ftiinstat_reqalloc, - "event=0xb,instance=0x46,umask=0x4"); -AMDGPU_PMU_ATTR(cake1_ftiinstat_reqalloc, - "event=0xb,instance=0x47,umask=0x4"); -AMDGPU_PMU_ATTR(cake0_ftiinstat_rspalloc, - "event=0xb,instance=0x46,umask=0x8"); -AMDGPU_PMU_ATTR(cake1_ftiinstat_rspalloc, - "event=0xb,instance=0x47,umask=0x8"); - -/* df event attributes */ -static struct attribute *df_v3_6_event_attrs[] = { - &pmu_attr_cake0_pcsout_txdata.attr, - &pmu_attr_cake1_pcsout_txdata.attr, - &pmu_attr_cake0_pcsout_txmeta.attr, - &pmu_attr_cake1_pcsout_txmeta.attr, - &pmu_attr_cake0_ftiinstat_reqalloc.attr, - &pmu_attr_cake1_ftiinstat_reqalloc.attr, - &pmu_attr_cake0_ftiinstat_rspalloc.attr, - &pmu_attr_cake1_ftiinstat_rspalloc.attr, - NULL -}; +/* Defined in global_features.h as FTI_PERFMON_VISIBLE */ +#define DF_V3_6_MAX_COUNTERS 4 -/* df event attribute group */ -static struct attribute_group df_v3_6_event_attr_group = { - .name = "events", - .attrs = df_v3_6_event_attrs -}; +/* get flags from df perfmon config */ +#define DF_V3_6_GET_EVENT(x) (x & 0xFFUL) +#define DF_V3_6_GET_INSTANCE(x) ((x >> 8) & 0xFFUL) +#define DF_V3_6_GET_UNITMASK(x) ((x >> 16) & 0xFFUL) +#define DF_V3_6_PERFMON_OVERFLOW 0xFFFFFFFFFFFFULL -/* df event attr groups */ -const struct attribute_group *df_v3_6_attr_groups[] = { - &df_v3_6_format_attr_group, - &df_v3_6_event_attr_group, - NULL -}; +static u32 df_v3_6_channel_number[] = {1, 2, 0, 4, 0, 8, 0, + 16, 32, 0, 0, 0, 2, 4, 8}; static uint64_t df_v3_6_get_fica(struct amdgpu_device *adev, uint32_t ficaa_val) @@ -391,33 +337,28 @@ static void df_v3_6_get_clockgating_state(struct amdgpu_device *adev, } /* get assigned df perfmon ctr as int */ -static int df_v3_6_pmc_config_2_cntr(struct amdgpu_device *adev, - uint64_t config) +static bool df_v3_6_pmc_has_counter(struct amdgpu_device *adev, + uint64_t config, + int counter_idx) { - int i; - for (i = 0; i < DF_V3_6_MAX_COUNTERS; i++) { - if ((config & 0x0FFFFFFUL) == - adev->df_perfmon_config_assign_mask[i]) - return i; - } + return ((config & 0x0FFFFFFUL) == + adev->df_perfmon_config_assign_mask[counter_idx]); - return -EINVAL; } /* get address based on counter assignment */ static void df_v3_6_pmc_get_addr(struct amdgpu_device *adev, uint64_t config, + int counter_idx, int is_ctrl, uint32_t *lo_base_addr, uint32_t *hi_base_addr) { - int target_cntr = df_v3_6_pmc_config_2_cntr(adev, config); - - if (target_cntr < 0) + if (!df_v3_6_pmc_has_counter(adev, config, counter_idx)) return; - switch (target_cntr) { + switch (counter_idx) { case 0: *lo_base_addr = is_ctrl ? smnPerfMonCtlLo4 : smnPerfMonCtrLo4; @@ -443,15 +384,18 @@ static void df_v3_6_pmc_get_addr(struct amdgpu_device *adev, /* get read counter address */ static void df_v3_6_pmc_get_read_settings(struct amdgpu_device *adev, uint64_t config, + int counter_idx, uint32_t *lo_base_addr, uint32_t *hi_base_addr) { - df_v3_6_pmc_get_addr(adev, config, 0, lo_base_addr, hi_base_addr); + df_v3_6_pmc_get_addr(adev, config, counter_idx, 0, lo_base_addr, + hi_base_addr); } /* get control counter settings i.e. address and values to set */ static int df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev, uint64_t config, + int counter_idx, uint32_t *lo_base_addr, uint32_t *hi_base_addr, uint32_t *lo_val, @@ -462,7 +406,8 @@ static int df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev, uint32_t eventsel, instance, unitmask; uint32_t instance_10, instance_5432, instance_76; - df_v3_6_pmc_get_addr(adev, config, 1, lo_base_addr, hi_base_addr); + df_v3_6_pmc_get_addr(adev, config, counter_idx, 1, lo_base_addr, + hi_base_addr); if ((*lo_base_addr == 0) || (*hi_base_addr == 0)) { DRM_ERROR("[DF PMC] addressing not retrieved! Lo: %x, Hi: %x", @@ -492,18 +437,13 @@ static int df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev, static int df_v3_6_pmc_add_cntr(struct amdgpu_device *adev, uint64_t config) { - int i, target_cntr; - - target_cntr = df_v3_6_pmc_config_2_cntr(adev, config); - - if (target_cntr >= 0) - return 0; + int i; for (i = 0; i < DF_V3_6_MAX_COUNTERS; i++) { if (adev->df_perfmon_config_assign_mask[i] == 0U) { adev->df_perfmon_config_assign_mask[i] = config & 0x0FFFFFFUL; - return 0; + return i; } } @@ -512,59 +452,50 @@ static int df_v3_6_pmc_add_cntr(struct amdgpu_device *adev, #define DEFERRED_ARM_MASK (1 << 31) static int df_v3_6_pmc_set_deferred(struct amdgpu_device *adev, - uint64_t config, bool is_deferred) + int counter_idx, uint64_t config, + bool is_deferred) { - int target_cntr; - target_cntr = df_v3_6_pmc_config_2_cntr(adev, config); - - if (target_cntr < 0) + if (!df_v3_6_pmc_has_counter(adev, config, counter_idx)) return -EINVAL; if (is_deferred) - adev->df_perfmon_config_assign_mask[target_cntr] |= + adev->df_perfmon_config_assign_mask[counter_idx] |= DEFERRED_ARM_MASK; else - adev->df_perfmon_config_assign_mask[target_cntr] &= + adev->df_perfmon_config_assign_mask[counter_idx] &= ~DEFERRED_ARM_MASK; return 0; } static bool df_v3_6_pmc_is_deferred(struct amdgpu_device *adev, + int counter_idx, uint64_t config) { - int target_cntr; - - target_cntr = df_v3_6_pmc_config_2_cntr(adev, config); - - /* - * we never get target_cntr < 0 since this funciton is only called in - * pmc_count for now but we should check anyways. - */ - return (target_cntr >= 0 && - (adev->df_perfmon_config_assign_mask[target_cntr] - & DEFERRED_ARM_MASK)); + return (df_v3_6_pmc_has_counter(adev, config, counter_idx) && + (adev->df_perfmon_config_assign_mask[counter_idx] + & DEFERRED_ARM_MASK)); } /* release performance counter */ static void df_v3_6_pmc_release_cntr(struct amdgpu_device *adev, - uint64_t config) + uint64_t config, + int counter_idx) { - int target_cntr = df_v3_6_pmc_config_2_cntr(adev, config); - - if (target_cntr >= 0) - adev->df_perfmon_config_assign_mask[target_cntr] = 0ULL; + if (df_v3_6_pmc_has_counter(adev, config, counter_idx)) + adev->df_perfmon_config_assign_mask[counter_idx] = 0ULL; } static void df_v3_6_reset_perfmon_cntr(struct amdgpu_device *adev, - uint64_t config) + uint64_t config, + int counter_idx) { uint32_t lo_base_addr = 0, hi_base_addr = 0; - df_v3_6_pmc_get_read_settings(adev, config, &lo_base_addr, + df_v3_6_pmc_get_read_settings(adev, config, counter_idx, &lo_base_addr, &hi_base_addr); if ((lo_base_addr == 0) || (hi_base_addr == 0)) @@ -573,21 +504,22 @@ static void df_v3_6_reset_perfmon_cntr(struct amdgpu_device *adev, df_v3_6_perfmon_wreg(adev, lo_base_addr, 0, hi_base_addr, 0); } +/* return available counter if is_add == 1 otherwise return error status. */ static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config, - int is_add) + int counter_idx, int is_add) { uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val; int err = 0, ret = 0; switch (adev->asic_type) { case CHIP_VEGA20: + case CHIP_ARCTURUS: if (is_add) return df_v3_6_pmc_add_cntr(adev, config); - df_v3_6_reset_perfmon_cntr(adev, config); - ret = df_v3_6_pmc_get_ctrl_settings(adev, config, + counter_idx, &lo_base_addr, &hi_base_addr, &lo_val, @@ -604,7 +536,8 @@ static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config, hi_val); if (err) - ret = df_v3_6_pmc_set_deferred(adev, config, true); + ret = df_v3_6_pmc_set_deferred(adev, config, + counter_idx, true); break; default: @@ -615,15 +548,17 @@ static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config, } static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config, - int is_remove) + int counter_idx, int is_remove) { uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val; int ret = 0; switch (adev->asic_type) { case CHIP_VEGA20: + case CHIP_ARCTURUS: ret = df_v3_6_pmc_get_ctrl_settings(adev, config, + counter_idx, &lo_base_addr, &hi_base_addr, &lo_val, @@ -635,8 +570,8 @@ static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config, if (is_remove) { - df_v3_6_reset_perfmon_cntr(adev, config); - df_v3_6_pmc_release_cntr(adev, config); + df_v3_6_reset_perfmon_cntr(adev, config, counter_idx); + df_v3_6_pmc_release_cntr(adev, config, counter_idx); } break; @@ -649,6 +584,7 @@ static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config, static void df_v3_6_pmc_get_count(struct amdgpu_device *adev, uint64_t config, + int counter_idx, uint64_t *count) { uint32_t lo_base_addr = 0, hi_base_addr = 0, lo_val = 0, hi_val = 0; @@ -656,14 +592,15 @@ static void df_v3_6_pmc_get_count(struct amdgpu_device *adev, switch (adev->asic_type) { case CHIP_VEGA20: - df_v3_6_pmc_get_read_settings(adev, config, &lo_base_addr, - &hi_base_addr); + case CHIP_ARCTURUS: + df_v3_6_pmc_get_read_settings(adev, config, counter_idx, + &lo_base_addr, &hi_base_addr); if ((lo_base_addr == 0) || (hi_base_addr == 0)) return; /* rearm the counter or throw away count value on failure */ - if (df_v3_6_pmc_is_deferred(adev, config)) { + if (df_v3_6_pmc_is_deferred(adev, config, counter_idx)) { int rearm_err = df_v3_6_perfmon_arm_with_status(adev, lo_base_addr, lo_val, hi_base_addr, hi_val); @@ -671,7 +608,8 @@ static void df_v3_6_pmc_get_count(struct amdgpu_device *adev, if (rearm_err) return; - df_v3_6_pmc_set_deferred(adev, config, false); + df_v3_6_pmc_set_deferred(adev, config, counter_idx, + false); } df_v3_6_perfmon_rreg(adev, lo_base_addr, &lo_val, diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.h b/drivers/gpu/drm/amd/amdgpu/df_v3_6.h index 76998541bc30..2505c7ef258a 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v3_6.h +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.h @@ -35,15 +35,6 @@ enum DF_V3_6_MGCG { DF_V3_6_MGCG_ENABLE_63_CYCLE_DELAY = 15 }; -/* Defined in global_features.h as FTI_PERFMON_VISIBLE */ -#define DF_V3_6_MAX_COUNTERS 4 - -/* get flags from df perfmon config */ -#define DF_V3_6_GET_EVENT(x) (x & 0xFFUL) -#define DF_V3_6_GET_INSTANCE(x) ((x >> 8) & 0xFFUL) -#define DF_V3_6_GET_UNITMASK(x) ((x >> 16) & 0xFFUL) -#define DF_V3_6_PERFMON_OVERFLOW 0xFFFFFFFFFFFFULL - extern const struct attribute_group *df_v3_6_attr_groups[]; extern const struct amdgpu_df_funcs df_v3_6_funcs; diff --git a/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c b/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c new file mode 100755 index 000000000000..e9f177e9e3cf --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/dimgrey_cavefish_reg_init.c @@ -0,0 +1,54 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "nv.h" + +#include "soc15_common.h" +#include "soc15_hw_ip.h" +#include "dimgrey_cavefish_ip_offset.h" + +int dimgrey_cavefish_reg_base_init(struct amdgpu_device *adev) +{ + /* HW has more IP blocks, only initialize the block needed by driver */ + uint32_t i; + for (i = 0 ; i < MAX_INSTANCE ; ++i) { + adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i])); + adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i])); + adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i])); + adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i])); + adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i])); + adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i])); + adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i])); + adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(VCN0_BASE.instance[i])); + adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i])); + adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DCN_BASE.instance[i])); + adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i])); + adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i])); + adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i])); + adev->reg_offset[SDMA2_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i])); + adev->reg_offset[SDMA3_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i])); + adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i])); + adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i])); + } + return 0; +} diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 56fdbe626d30..8c3bad3dfc01 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -99,6 +99,23 @@ #define mmGCR_GENERAL_CNTL_Sienna_Cichlid 0x1580 #define mmGCR_GENERAL_CNTL_Sienna_Cichlid_BASE_IDX 0 +#define mmSPI_CONFIG_CNTL_1_Vangogh 0x2441 +#define mmSPI_CONFIG_CNTL_1_Vangogh_BASE_IDX 1 +#define mmVGT_TF_MEMORY_BASE_HI_Vangogh 0x2261 +#define mmVGT_TF_MEMORY_BASE_HI_Vangogh_BASE_IDX 1 +#define mmVGT_HS_OFFCHIP_PARAM_Vangogh 0x224f +#define mmVGT_HS_OFFCHIP_PARAM_Vangogh_BASE_IDX 1 +#define mmVGT_TF_RING_SIZE_Vangogh 0x224e +#define mmVGT_TF_RING_SIZE_Vangogh_BASE_IDX 1 +#define mmVGT_GSVS_RING_SIZE_Vangogh 0x2241 +#define mmVGT_GSVS_RING_SIZE_Vangogh_BASE_IDX 1 +#define mmVGT_TF_MEMORY_BASE_Vangogh 0x2250 +#define mmVGT_TF_MEMORY_BASE_Vangogh_BASE_IDX 1 +#define mmVGT_ESGS_RING_SIZE_Vangogh 0x2240 +#define mmVGT_ESGS_RING_SIZE_Vangogh_BASE_IDX 1 +#define mmSPI_CONFIG_CNTL_Vangogh 0x2440 +#define mmSPI_CONFIG_CNTL_Vangogh_BASE_IDX 1 + #define mmCP_HYP_PFP_UCODE_ADDR 0x5814 #define mmCP_HYP_PFP_UCODE_ADDR_BASE_IDX 1 #define mmCP_HYP_PFP_UCODE_DATA 0x5815 @@ -112,6 +129,13 @@ #define mmCP_HYP_ME_UCODE_DATA 0x5817 #define mmCP_HYP_ME_UCODE_DATA_BASE_IDX 1 +#define mmCPG_PSP_DEBUG 0x5c10 +#define mmCPG_PSP_DEBUG_BASE_IDX 1 +#define mmCPC_PSP_DEBUG 0x5c11 +#define mmCPC_PSP_DEBUG_BASE_IDX 1 +#define CPC_PSP_DEBUG__GPA_OVERRIDE_MASK 0x00000008L +#define CPG_PSP_DEBUG__GPA_OVERRIDE_MASK 0x00000008L + //CC_GC_SA_UNIT_DISABLE #define mmCC_GC_SA_UNIT_DISABLE 0x0fe9 #define mmCC_GC_SA_UNIT_DISABLE_BASE_IDX 0 @@ -128,6 +152,9 @@ #define PA_SC_ENHANCE_3__FORCE_PBB_WORKLOAD_MODE_TO_ZERO__SHIFT 0x3 #define PA_SC_ENHANCE_3__FORCE_PBB_WORKLOAD_MODE_TO_ZERO_MASK 0x00000008L +#define mmCGTT_SPI_CS_CLK_CTRL 0x507c +#define mmCGTT_SPI_CS_CLK_CTRL_BASE_IDX 1 + MODULE_FIRMWARE("amdgpu/navi10_ce.bin"); MODULE_FIRMWARE("amdgpu/navi10_pfp.bin"); MODULE_FIRMWARE("amdgpu/navi10_me.bin"); @@ -168,6 +195,20 @@ MODULE_FIRMWARE("amdgpu/navy_flounder_mec.bin"); MODULE_FIRMWARE("amdgpu/navy_flounder_mec2.bin"); MODULE_FIRMWARE("amdgpu/navy_flounder_rlc.bin"); +MODULE_FIRMWARE("amdgpu/vangogh_ce.bin"); +MODULE_FIRMWARE("amdgpu/vangogh_pfp.bin"); +MODULE_FIRMWARE("amdgpu/vangogh_me.bin"); +MODULE_FIRMWARE("amdgpu/vangogh_mec.bin"); +MODULE_FIRMWARE("amdgpu/vangogh_mec2.bin"); +MODULE_FIRMWARE("amdgpu/vangogh_rlc.bin"); + +MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_ce.bin"); +MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_pfp.bin"); +MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_me.bin"); +MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_mec.bin"); +MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_mec2.bin"); +MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_rlc.bin"); + static const struct soc15_reg_golden golden_settings_gc_10_1[] = { SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_4, 0xffffffff, 0x00400014), @@ -3094,6 +3135,7 @@ static const struct soc15_reg_golden golden_settings_gc_rlc_spm_10_1_2_nv12[] = static const struct soc15_reg_golden golden_settings_gc_10_3[] = { + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0x78000000, 0x78000100), SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_PS_CLK_CTRL, 0xff7f0fff, 0x78000100), SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0xff7f0fff, 0x30000100), SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0xff7f0fff, 0x7e000100), @@ -3177,7 +3219,74 @@ static const struct soc15_reg_golden golden_settings_gc_10_3_2[] = SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER9_SELECT, 0xf0f001ff, 0x00000000), SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000), SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffbfffff, 0x00a00000), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff) + SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff), + + /* This is not in GDB yet. Don't remove it. It fixes a GPU hang on Navy Flounder. */ + SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG, 0x00000020, 0x00000020), +}; + +static const struct soc15_reg_golden golden_settings_gc_10_3_vangogh[] = +{ + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0xff7f0fff, 0x30000100), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0xff7f0fff, 0x7e000100), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0x000000ff, 0x000000e4), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000200), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_EXCEPTION_CONTROL, 0x7fff0f1f, 0x00b80000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1807ff, 0x00000142), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x1ff1ffff, 0x00000500), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0x000000ff, 0x000000e4), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_0, 0x77777777, 0x32103210), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_1, 0x77777777, 0x32103210), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x00f80988), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf17fffff, 0x01200007), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0xffffffbf, 0x00000020), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1_Vangogh, 0xffffffff, 0x00070103), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQG_CONFIG, 0x000017ff, 0x00001000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00400000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000000ff), +}; + +static const struct soc15_reg_golden golden_settings_gc_10_3_4[] = +{ + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0x30000000, 0x30000100), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0x7e000000, 0x7e000100), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0x00000280, 0x00000280), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0x07800000, 0x00800000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x00001d00, 0x00000500), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_PC_CNTL, 0x003c0000, 0x00280400), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0x40000000, 0x580f1008), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0x00040000, 0x00f80988), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0x01000000, 0x01200007), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0x00000800, 0x00000820), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_CONFIG, 0x0000001f, 0x00180070), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER0_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER1_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER10_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER11_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER12_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER13_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER14_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER15_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER2_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER3_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER4_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER5_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER6_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER7_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER8_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER9_SELECT, 0xf0f001ff, 0x00000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0x01030000, 0x01030000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0x03a00000, 0x00a00000) }; #define DEFAULT_SH_MEM_CONFIG \ @@ -3192,7 +3301,7 @@ static void gfx_v10_0_set_irq_funcs(struct amdgpu_device *adev); static void gfx_v10_0_set_gds_init(struct amdgpu_device *adev); static void gfx_v10_0_set_rlc_funcs(struct amdgpu_device *adev); static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev, - struct amdgpu_cu_info *cu_info); + struct amdgpu_cu_info *cu_info); static uint64_t gfx_v10_0_get_gpu_clock_counter(struct amdgpu_device *adev); static void gfx_v10_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance); @@ -3389,7 +3498,16 @@ static void gfx_v10_0_init_golden_registers(struct amdgpu_device *adev) golden_settings_gc_10_3_2, (const u32)ARRAY_SIZE(golden_settings_gc_10_3_2)); break; - + case CHIP_VANGOGH: + soc15_program_register_sequence(adev, + golden_settings_gc_10_3_vangogh, + (const u32)ARRAY_SIZE(golden_settings_gc_10_3_vangogh)); + break; + case CHIP_DIMGREY_CAVEFISH: + soc15_program_register_sequence(adev, + golden_settings_gc_10_3_4, + (const u32)ARRAY_SIZE(golden_settings_gc_10_3_4)); + break; default: break; } @@ -3573,6 +3691,8 @@ static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev) break; case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: adev->gfx.cp_fw_write_wait = true; break; default: @@ -3640,7 +3760,7 @@ static void gfx_v10_0_check_gfxoff_flag(struct amdgpu_device *adev) if (!gfx_v10_0_navi10_gfxoff_should_enable(adev)) adev->pm.pp_feature &= ~PP_GFXOFF_MASK; break; - case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: adev->pm.pp_feature &= ~PP_GFXOFF_MASK; break; default: @@ -3685,6 +3805,12 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev) case CHIP_NAVY_FLOUNDER: chip_name = "navy_flounder"; break; + case CHIP_VANGOGH: + chip_name = "vangogh"; + break; + case CHIP_DIMGREY_CAVEFISH: + chip_name = "dimgrey_cavefish"; + break; default: BUG(); } @@ -4200,11 +4326,26 @@ static void gfx_v10_0_read_wave_vgprs(struct amdgpu_device *adev, uint32_t simd, } static void gfx_v10_0_select_me_pipe_q(struct amdgpu_device *adev, - u32 me, u32 pipe, u32 q, u32 vm) - { - nv_grbm_select(adev, me, pipe, q, vm); - } + u32 me, u32 pipe, u32 q, u32 vm) +{ + nv_grbm_select(adev, me, pipe, q, vm); +} +static void gfx_v10_0_update_perfmon_mgcg(struct amdgpu_device *adev, + bool enable) +{ + uint32_t data, def; + + data = def = RREG32_SOC15(GC, 0, mmRLC_PERFMON_CLK_CNTL); + + if (enable) + data |= RLC_PERFMON_CLK_CNTL__PERFMON_CLOCK_STATE_MASK; + else + data &= ~RLC_PERFMON_CLK_CNTL__PERFMON_CLOCK_STATE_MASK; + + if (data != def) + WREG32_SOC15(GC, 0, mmRLC_PERFMON_CLK_CNTL, data); +} static const struct amdgpu_gfx_funcs gfx_v10_0_gfx_funcs = { .get_gpu_clock_counter = &gfx_v10_0_get_gpu_clock_counter, @@ -4214,6 +4355,7 @@ static const struct amdgpu_gfx_funcs gfx_v10_0_gfx_funcs = { .read_wave_vgprs = &gfx_v10_0_read_wave_vgprs, .select_me_pipe_q = &gfx_v10_0_select_me_pipe_q, .init_spm_golden = &gfx_v10_0_init_spm_golden_registers, + .update_perfmon_mgcg = &gfx_v10_0_update_perfmon_mgcg, }; static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev) @@ -4235,6 +4377,8 @@ static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev) break; case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: adev->gfx.config.max_hw_contexts = 8; adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; adev->gfx.config.sc_prim_fifo_size_backend = 0x100; @@ -4358,6 +4502,8 @@ static int gfx_v10_0_sw_init(void *handle) break; case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: adev->gfx.me.num_me = 1; adev->gfx.me.num_pipe_per_me = 1; adev->gfx.me.num_queue_per_pipe = 1; @@ -4617,8 +4763,7 @@ static u32 gfx_v10_0_init_pa_sc_tile_steering_override(struct amdgpu_device *ade /* for ASICs that integrates GFX v10.3 * pa_sc_tile_steering_override should be set to 0 */ - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) + if (adev->asic_type >= CHIP_SIENNA_CICHLID) return 0; /* init num_sc */ @@ -5855,6 +6000,8 @@ static void gfx_v10_0_cp_gfx_set_doorbell(struct amdgpu_device *adev, switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER, DOORBELL_RANGE_LOWER_Sienna_Cichlid, ring->doorbell_index); WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_RANGE_LOWER, tmp); @@ -5988,6 +6135,8 @@ static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: WREG32_SOC15(GC, 0, mmCP_MEC_CNTL_Sienna_Cichlid, 0); break; default: @@ -5998,6 +6147,8 @@ static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: WREG32_SOC15(GC, 0, mmCP_MEC_CNTL_Sienna_Cichlid, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK)); @@ -6092,6 +6243,8 @@ static void gfx_v10_0_kiq_setting(struct amdgpu_ring *ring) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: tmp = RREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS_Sienna_Cichlid); tmp &= 0xffffff00; tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue); @@ -6800,6 +6953,7 @@ static bool gfx_v10_0_check_grbm_cam_remapping(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: data = RREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_Sienna_Cichlid); WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_Sienna_Cichlid, 0); WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_UMD, pattern); @@ -6812,6 +6966,8 @@ static bool gfx_v10_0_check_grbm_cam_remapping(struct amdgpu_device *adev) return false; } break; + case CHIP_VANGOGH: + return true; default: data = RREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE); WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE, 0); @@ -6839,6 +6995,8 @@ static void gfx_v10_0_setup_grbm_cam_remapping(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: /* mmVGT_TF_RING_SIZE_UMD -> mmVGT_TF_RING_SIZE */ data = (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_RING_SIZE_UMD) << GRBM_CAM_DATA__CAM_ADDR__SHIFT) | @@ -6954,6 +7112,18 @@ static void gfx_v10_0_setup_grbm_cam_remapping(struct amdgpu_device *adev) WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data); } +static void gfx_v10_0_disable_gpa_mode(struct amdgpu_device *adev) +{ + uint32_t data; + data = RREG32_SOC15(GC, 0, mmCPC_PSP_DEBUG); + data |= CPC_PSP_DEBUG__GPA_OVERRIDE_MASK; + WREG32_SOC15(GC, 0, mmCPC_PSP_DEBUG, data); + + data = RREG32_SOC15(GC, 0, mmCPG_PSP_DEBUG); + data |= CPG_PSP_DEBUG__GPA_OVERRIDE_MASK; + WREG32_SOC15(GC, 0, mmCPG_PSP_DEBUG, data); +} + static int gfx_v10_0_hw_init(void *handle) { int r; @@ -6968,7 +7138,7 @@ static int gfx_v10_0_hw_init(void *handle) * loaded firstly, so in direct type, it has to load smc ucode * here before rlc. */ - if (adev->smu.ppt_funcs != NULL) { + if (adev->smu.ppt_funcs != NULL && !(adev->flags & AMD_IS_APU)) { r = smu_load_microcode(&adev->smu); if (r) return r; @@ -6979,6 +7149,7 @@ static int gfx_v10_0_hw_init(void *handle) return r; } } + gfx_v10_0_disable_gpa_mode(adev); } /* if GRBM CAM not remapped, set up the remapping */ @@ -7136,6 +7307,8 @@ static int gfx_v10_0_soft_reset(void *handle) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: if (REG_GET_FIELD(tmp, GRBM_STATUS2, RLC_BUSY_Sienna_Cichlid)) grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, @@ -7235,13 +7408,16 @@ static int gfx_v10_0_early_init(void *handle) break; case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: adev->gfx.num_gfx_rings = GFX10_NUM_GFX_RINGS_Sienna_Cichlid; break; default: break; } - adev->gfx.num_compute_rings = amdgpu_num_kcq; + adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev), + AMDGPU_MAX_COMPUTE_RINGS); gfx_v10_0_set_kiq_pm4_funcs(adev); gfx_v10_0_set_ring_funcs(adev); @@ -7288,6 +7464,8 @@ static void gfx_v10_0_set_safe_mode(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE_Sienna_Cichlid, data); /* wait for RLC_SAFE_MODE */ @@ -7320,6 +7498,8 @@ static void gfx_v10_0_unset_safe_mode(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE_Sienna_Cichlid, data); break; default: @@ -7344,6 +7524,7 @@ static void gfx_v10_0_update_medium_grain_clock_gating(struct amdgpu_device *ade /* 1 - RLC_CGTT_MGCG_OVERRIDE */ def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE); data &= ~(RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK | + RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK | RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK | RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK | RLC_CGTT_MGCG_OVERRIDE__ENABLE_CGTS_LEGACY_MASK); @@ -7480,12 +7661,50 @@ static void gfx_v10_0_update_coarse_grain_clock_gating(struct amdgpu_device *ade } } +static void gfx_v10_0_update_fine_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + uint32_t def, data; + + if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_FGCG)) { + def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE); + /* unset FGCG override */ + data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK; + /* update FGCG override bits */ + if (def != data) + WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data); + + def = data = RREG32_SOC15(GC, 0, mmRLC_CLK_CNTL); + /* unset RLC SRAM CLK GATER override */ + data &= ~RLC_CLK_CNTL__RLC_SRAM_CLK_GATER_OVERRIDE_MASK; + /* update RLC SRAM CLK GATER override bits */ + if (def != data) + WREG32_SOC15(GC, 0, mmRLC_CLK_CNTL, data); + } else { + def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE); + /* reset FGCG bits */ + data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK; + /* disable FGCG*/ + if (def != data) + WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data); + + def = data = RREG32_SOC15(GC, 0, mmRLC_CLK_CNTL); + /* reset RLC SRAM CLK GATER bits */ + data |= RLC_CLK_CNTL__RLC_SRAM_CLK_GATER_OVERRIDE_MASK; + /* disable RLC SRAM CLK*/ + if (def != data) + WREG32_SOC15(GC, 0, mmRLC_CLK_CNTL, data); + } +} + static int gfx_v10_0_update_gfx_clock_gating(struct amdgpu_device *adev, bool enable) { amdgpu_gfx_rlc_enter_safe_mode(adev); if (enable) { + /* enable FGCG firstly*/ + gfx_v10_0_update_fine_grain_clock_gating(adev, enable); /* CGCG/CGLS should be enabled after MGCG/MGLS * === MGCG + MGLS === */ @@ -7503,6 +7722,8 @@ static int gfx_v10_0_update_gfx_clock_gating(struct amdgpu_device *adev, gfx_v10_0_update_3d_clock_gating(adev, enable); /* === MGCG + MGLS === */ gfx_v10_0_update_medium_grain_clock_gating(adev, enable); + /* disable fgcg at last*/ + gfx_v10_0_update_fine_grain_clock_gating(adev, enable); } if (adev->cg_flags & @@ -7564,6 +7785,27 @@ static bool gfx_v10_0_is_rlcg_access_range(struct amdgpu_device *adev, u32 offse return gfx_v10_0_check_rlcg_range(adev, offset, NULL, 0); } +static void gfx_v10_cntl_power_gating(struct amdgpu_device *adev, bool enable) +{ + u32 data = RREG32_SOC15(GC, 0, mmRLC_PG_CNTL); + + if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) + data |= RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK; + else + data &= ~RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK; + + WREG32_SOC15(GC, 0, mmRLC_PG_CNTL, data); +} + +static void gfx_v10_cntl_pg(struct amdgpu_device *adev, bool enable) +{ + amdgpu_gfx_rlc_enter_safe_mode(adev); + + gfx_v10_cntl_power_gating(adev, enable); + + amdgpu_gfx_rlc_exit_safe_mode(adev); +} + static const struct amdgpu_rlc_funcs gfx_v10_0_rlc_funcs = { .is_rlc_enabled = gfx_v10_0_is_rlc_enabled, .set_safe_mode = gfx_v10_0_set_safe_mode, @@ -7609,8 +7851,12 @@ static int gfx_v10_0_set_powergating_state(void *handle, case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: amdgpu_gfx_off_ctrl(adev, enable); break; + case CHIP_VANGOGH: + gfx_v10_cntl_pg(adev, enable); + break; default: break; } @@ -7631,6 +7877,8 @@ static int gfx_v10_0_set_clockgating_state(void *handle, case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: gfx_v10_0_update_gfx_clock_gating(adev, state == AMD_CG_STATE_GATE); break; @@ -7645,6 +7893,11 @@ static void gfx_v10_0_get_clockgating_state(void *handle, u32 *flags) struct amdgpu_device *adev = (struct amdgpu_device *)handle; int data; + /* AMD_CG_SUPPORT_GFX_FGCG */ + data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE)); + if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK)) + *flags |= AMD_CG_SUPPORT_GFX_FGCG; + /* AMD_CG_SUPPORT_GFX_MGCG */ data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE)); if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK)) @@ -8733,6 +8986,8 @@ static void gfx_v10_0_set_rlc_funcs(struct amdgpu_device *adev) case CHIP_NAVI14: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: adev->gfx.rlc.funcs = &gfx_v10_0_rlc_funcs; break; case CHIP_NAVI12: diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c index 79c52c7a02e3..671c46ebeced 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c @@ -3064,7 +3064,8 @@ static int gfx_v6_0_early_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; adev->gfx.num_gfx_rings = GFX6_NUM_GFX_RINGS; - adev->gfx.num_compute_rings = GFX6_NUM_COMPUTE_RINGS; + adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev), + GFX6_NUM_COMPUTE_RINGS); adev->gfx.funcs = &gfx_v6_0_gfx_funcs; adev->gfx.rlc.funcs = &gfx_v6_0_rlc_funcs; gfx_v6_0_set_ring_funcs(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index 04eaf3a8fddb..cb07bc21dcbe 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -4238,7 +4238,8 @@ static int gfx_v7_0_early_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; adev->gfx.num_gfx_rings = GFX7_NUM_GFX_RINGS; - adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS; + adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev), + AMDGPU_MAX_COMPUTE_RINGS); adev->gfx.funcs = &gfx_v7_0_gfx_funcs; adev->gfx.rlc.funcs = &gfx_v7_0_rlc_funcs; gfx_v7_0_set_ring_funcs(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 94b7e0531d09..c3fff49e6514 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -729,8 +729,13 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev); static void gfx_v8_0_ring_emit_ce_meta(struct amdgpu_ring *ring); static void gfx_v8_0_ring_emit_de_meta(struct amdgpu_ring *ring); +#define CG_ACLK_CNTL__ACLK_DIVIDER_MASK 0x0000007fL +#define CG_ACLK_CNTL__ACLK_DIVIDER__SHIFT 0x00000000L + static void gfx_v8_0_init_golden_registers(struct amdgpu_device *adev) { + uint32_t data; + switch (adev->asic_type) { case CHIP_TOPAZ: amdgpu_device_program_register_sequence(adev, @@ -790,11 +795,14 @@ static void gfx_v8_0_init_golden_registers(struct amdgpu_device *adev) amdgpu_device_program_register_sequence(adev, polaris10_golden_common_all, ARRAY_SIZE(polaris10_golden_common_all)); - WREG32_SMC(ixCG_ACLK_CNTL, 0x0000001C); - if (adev->pdev->revision == 0xc7 && + data = RREG32_SMC(ixCG_ACLK_CNTL); + data &= ~CG_ACLK_CNTL__ACLK_DIVIDER_MASK; + data |= 0x18 << CG_ACLK_CNTL__ACLK_DIVIDER__SHIFT; + WREG32_SMC(ixCG_ACLK_CNTL, data); + if ((adev->pdev->device == 0x67DF) && (adev->pdev->revision == 0xc7) && ((adev->pdev->subsystem_device == 0xb37 && adev->pdev->subsystem_vendor == 0x1002) || (adev->pdev->subsystem_device == 0x4a8 && adev->pdev->subsystem_vendor == 0x1043) || - (adev->pdev->subsystem_device == 0x9480 && adev->pdev->subsystem_vendor == 0x1682))) { + (adev->pdev->subsystem_device == 0x9480 && adev->pdev->subsystem_vendor == 0x1680))) { amdgpu_atombios_i2c_channel_trans(adev, 0x10, 0x96, 0x1E, 0xDD); amdgpu_atombios_i2c_channel_trans(adev, 0x10, 0x96, 0x1F, 0xD0); } @@ -5058,7 +5066,7 @@ static int gfx_v8_0_pre_soft_reset(void *handle) gfx_v8_0_cp_compute_enable(adev, false); } - return 0; + return 0; } static int gfx_v8_0_soft_reset(void *handle) @@ -5295,7 +5303,8 @@ static int gfx_v8_0_early_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; adev->gfx.num_gfx_rings = GFX8_NUM_GFX_RINGS; - adev->gfx.num_compute_rings = amdgpu_num_kcq; + adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev), + AMDGPU_MAX_COMPUTE_RINGS); adev->gfx.funcs = &gfx_v8_0_gfx_funcs; gfx_v8_0_set_ring_funcs(adev); gfx_v8_0_set_irq_funcs(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 6959aebae6d4..3d6fb5a514c8 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -117,6 +117,13 @@ MODULE_FIRMWARE("amdgpu/renoir_mec.bin"); MODULE_FIRMWARE("amdgpu/renoir_mec2.bin"); MODULE_FIRMWARE("amdgpu/renoir_rlc.bin"); +MODULE_FIRMWARE("amdgpu/green_sardine_ce.bin"); +MODULE_FIRMWARE("amdgpu/green_sardine_pfp.bin"); +MODULE_FIRMWARE("amdgpu/green_sardine_me.bin"); +MODULE_FIRMWARE("amdgpu/green_sardine_mec.bin"); +MODULE_FIRMWARE("amdgpu/green_sardine_mec2.bin"); +MODULE_FIRMWARE("amdgpu/green_sardine_rlc.bin"); + #define mmTCP_CHAN_STEER_0_ARCT 0x0b03 #define mmTCP_CHAN_STEER_0_ARCT_BASE_IDX 0 #define mmTCP_CHAN_STEER_1_ARCT 0x0b04 @@ -787,7 +794,7 @@ static void gfx_v9_0_set_irq_funcs(struct amdgpu_device *adev); static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev); static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev); static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev, - struct amdgpu_cu_info *cu_info); + struct amdgpu_cu_info *cu_info); static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev); static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring); static u64 gfx_v9_0_ring_get_rptr_compute(struct amdgpu_ring *ring); @@ -1630,7 +1637,10 @@ static int gfx_v9_0_init_microcode(struct amdgpu_device *adev) chip_name = "arcturus"; break; case CHIP_RENOIR: - chip_name = "renoir"; + if (adev->apu_flags & AMD_APU_IS_RENOIR) + chip_name = "renoir"; + else + chip_name = "green_sardine"; break; default: BUG(); @@ -4623,7 +4633,8 @@ static int gfx_v9_0_early_init(void *handle) adev->gfx.num_gfx_rings = 0; else adev->gfx.num_gfx_rings = GFX9_NUM_GFX_RINGS; - adev->gfx.num_compute_rings = amdgpu_num_kcq; + adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev), + AMDGPU_MAX_COMPUTE_RINGS); gfx_v9_0_set_kiq_pm4_funcs(adev); gfx_v9_0_set_ring_funcs(adev); gfx_v9_0_set_irq_funcs(adev); @@ -5167,7 +5178,7 @@ static void gfx_v9_0_ring_set_wptr_gfx(struct amdgpu_ring *ring) if (ring->use_doorbell) { /* XXX check if swapping is necessary on BE */ - atomic64_set((atomic64_t*)&adev->wb.wb[ring->wptr_offs], ring->wptr); + atomic64_set((atomic64_t *)&adev->wb.wb[ring->wptr_offs], ring->wptr); WDOORBELL64(ring->doorbell_index, ring->wptr); } else { WREG32_SOC15(GC, 0, mmCP_RB0_WPTR, lower_32_bits(ring->wptr)); @@ -5353,7 +5364,7 @@ static void gfx_v9_0_ring_set_wptr_compute(struct amdgpu_ring *ring) /* XXX check if swapping is necessary on BE */ if (ring->use_doorbell) { - atomic64_set((atomic64_t*)&adev->wb.wb[ring->wptr_offs], ring->wptr); + atomic64_set((atomic64_t *)&adev->wb.wb[ring->wptr_offs], ring->wptr); WDOORBELL64(ring->doorbell_index, ring->wptr); } else{ BUG(); /* only DOORBELL method supported on gfx9 now */ diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c index fad887a66886..d83577b77a66 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c @@ -22,6 +22,7 @@ */ #include "amdgpu.h" #include "gfxhub_v1_0.h" +#include "gfxhub_v1_1.h" #include "gc/gc_9_0_offset.h" #include "gc/gc_9_0_sh_mask.h" @@ -412,4 +413,5 @@ const struct amdgpu_gfxhub_funcs gfxhub_v1_0_funcs = { .gart_disable = gfxhub_v1_0_gart_disable, .set_fault_enable_default = gfxhub_v1_0_set_fault_enable_default, .init = gfxhub_v1_0_init, + .get_xgmi_info = gfxhub_v1_1_get_xgmi_info, }; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c index 1e24b6d51e41..c0ab71df0d90 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c @@ -21,7 +21,6 @@ * */ #include "amdgpu.h" -#include "gfxhub_v1_0.h" #include "gfxhub_v1_1.h" #include "gc/gc_9_2_1_offset.h" @@ -29,7 +28,7 @@ #include "soc15_common.h" -static int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev) +int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev) { u32 xgmi_lfb_cntl = RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_CNTL); u32 max_region = @@ -67,13 +66,3 @@ static int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev) return 0; } - -const struct amdgpu_gfxhub_funcs gfxhub_v1_1_funcs = { - .get_mc_fb_offset = gfxhub_v1_0_get_mc_fb_offset, - .setup_vm_pt_regs = gfxhub_v1_0_setup_vm_pt_regs, - .gart_enable = gfxhub_v1_0_gart_enable, - .gart_disable = gfxhub_v1_0_gart_disable, - .set_fault_enable_default = gfxhub_v1_0_set_fault_enable_default, - .init = gfxhub_v1_0_init, - .get_xgmi_info = gfxhub_v1_1_get_xgmi_info, -}; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h index ae5759ffbee3..d753cf28a0a6 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h @@ -24,6 +24,6 @@ #ifndef __GFXHUB_V1_1_H__ #define __GFXHUB_V1_1_H__ -extern const struct amdgpu_gfxhub_funcs gfxhub_v1_1_funcs; +int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index dbc8b76b9b78..d9399324be47 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -46,6 +46,7 @@ #include "gfxhub_v2_0.h" #include "gfxhub_v2_1.h" #include "mmhub_v2_0.h" +#include "mmhub_v2_3.h" #include "athub_v2_0.h" #include "athub_v2_1.h" @@ -106,7 +107,8 @@ static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev, * be updated to avoid reading an incorrect value due to * the new fast GRBM interface. */ - if (entry->vmid_src == AMDGPU_GFXHUB_0) + if ((entry->vmid_src == AMDGPU_GFXHUB_0) && + (adev->asic_type < CHIP_SIENNA_CICHLID)) RREG32(hub->vm_l2_pro_fault_status); status = RREG32(hub->vm_l2_pro_fault_status); @@ -145,7 +147,7 @@ static const struct amdgpu_irq_src_funcs gmc_v10_0_ecc_funcs = { .process = amdgpu_umc_process_ecc_irq, }; - static void gmc_v10_0_set_irq_funcs(struct amdgpu_device *adev) +static void gmc_v10_0_set_irq_funcs(struct amdgpu_device *adev) { adev->gmc.vm_fault.num_types = 1; adev->gmc.vm_fault.funcs = &gmc_v10_0_irq_funcs; @@ -231,7 +233,8 @@ static void gmc_v10_0_flush_vm_hub(struct amdgpu_device *adev, uint32_t vmid, * Issue a dummy read to wait for the ACK register to be cleared * to avoid a false ACK due to the new fast GRBM interface. */ - if (vmhub == AMDGPU_GFXHUB_0) + if ((vmhub == AMDGPU_GFXHUB_0) && + (adev->asic_type < CHIP_SIENNA_CICHLID)) RREG32_NO_KIQ(hub->vm_inv_eng0_req + hub->eng_distance * eng); /* Wait for ACK with a delay.*/ @@ -483,7 +486,8 @@ static void gmc_v10_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid /* * PTE format on NAVI 10: * 63:59 reserved - * 58:57 reserved + * 58 reserved and for sienna_cichlid is used for MALL noalloc + * 57 reserved * 56 F * 55 L * 54 reserved @@ -631,7 +635,14 @@ static void gmc_v10_0_set_umc_funcs(struct amdgpu_device *adev) static void gmc_v10_0_set_mmhub_funcs(struct amdgpu_device *adev) { - adev->mmhub.funcs = &mmhub_v2_0_funcs; + switch (adev->asic_type) { + case CHIP_VANGOGH: + adev->mmhub.funcs = &mmhub_v2_3_funcs; + break; + default: + adev->mmhub.funcs = &mmhub_v2_0_funcs; + break; + } } static void gmc_v10_0_set_gfxhub_funcs(struct amdgpu_device *adev) @@ -639,6 +650,8 @@ static void gmc_v10_0_set_gfxhub_funcs(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: adev->gfxhub.funcs = &gfxhub_v2_1_funcs; break; default: @@ -733,6 +746,13 @@ static int gmc_v10_0_mc_init(struct amdgpu_device *adev) adev->gmc.aper_base = pci_resource_start(adev->pdev, 0); adev->gmc.aper_size = pci_resource_len(adev->pdev, 0); +#ifdef CONFIG_X86_64 + if (adev->flags & AMD_IS_APU) { + adev->gmc.aper_base = adev->gfxhub.funcs->get_mc_fb_offset(adev); + adev->gmc.aper_size = adev->gmc.real_vram_size; + } +#endif + /* In case the PCI BAR is larger than the actual amount of vram */ adev->gmc.visible_vram_size = adev->gmc.aper_size; if (adev->gmc.visible_vram_size > adev->gmc.real_vram_size) @@ -746,6 +766,8 @@ static int gmc_v10_0_mc_init(struct amdgpu_device *adev) case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: default: adev->gmc.gart_size = 512ULL << 20; break; @@ -790,7 +812,10 @@ static int gmc_v10_0_sw_init(void *handle) spin_lock_init(&adev->gmc.invalidate_lock); - if (adev->asic_type == CHIP_SIENNA_CICHLID && amdgpu_emu_mode == 1) { + if ((adev->flags & AMD_IS_APU) && amdgpu_emu_mode == 1) { + adev->gmc.vram_type = AMDGPU_VRAM_TYPE_DDR4; + adev->gmc.vram_width = 64; + } else if (amdgpu_emu_mode == 1) { adev->gmc.vram_type = AMDGPU_VRAM_TYPE_GDDR6; adev->gmc.vram_width = 1 * 128; /* numchan * chansize */ } else { @@ -808,6 +833,8 @@ static int gmc_v10_0_sw_init(void *handle) case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: adev->num_vmhubs = 2; /* * To fulfill 4-level page support, @@ -921,6 +948,8 @@ static void gmc_v10_0_init_golden_registers(struct amdgpu_device *adev) case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: break; default: break; @@ -1081,8 +1110,8 @@ static int gmc_v10_0_set_clockgating_state(void *handle, if (r) return r; - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) + if (adev->asic_type >= CHIP_SIENNA_CICHLID && + adev->asic_type <= CHIP_DIMGREY_CAVEFISH) return athub_v2_1_set_clockgating(adev, state); else return athub_v2_0_set_clockgating(adev, state); @@ -1094,8 +1123,8 @@ static void gmc_v10_0_get_clockgating_state(void *handle, u32 *flags) adev->mmhub.funcs->get_clockgating(adev, flags); - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) + if (adev->asic_type >= CHIP_SIENNA_CICHLID && + adev->asic_type <= CHIP_DIMGREY_CAVEFISH) athub_v2_1_get_clockgating(adev, flags); else athub_v2_0_get_clockgating(adev, flags); diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index 9ab65ca7df77..0f32a8002c3d 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -230,36 +230,20 @@ static int gmc_v8_0_init_microcode(struct amdgpu_device *adev) chip_name = "tonga"; break; case CHIP_POLARIS11: - if (((adev->pdev->device == 0x67ef) && - ((adev->pdev->revision == 0xe0) || - (adev->pdev->revision == 0xe5))) || - ((adev->pdev->device == 0x67ff) && - ((adev->pdev->revision == 0xcf) || - (adev->pdev->revision == 0xef) || - (adev->pdev->revision == 0xff)))) - chip_name = "polaris11_k"; - else if ((adev->pdev->device == 0x67ef) && - (adev->pdev->revision == 0xe2)) + if (ASICID_IS_P21(adev->pdev->device, adev->pdev->revision) || + ASICID_IS_P31(adev->pdev->device, adev->pdev->revision)) chip_name = "polaris11_k"; else chip_name = "polaris11"; break; case CHIP_POLARIS10: - if ((adev->pdev->device == 0x67df) && - ((adev->pdev->revision == 0xe1) || - (adev->pdev->revision == 0xf7))) + if (ASICID_IS_P30(adev->pdev->device, adev->pdev->revision)) chip_name = "polaris10_k"; else chip_name = "polaris10"; break; case CHIP_POLARIS12: - if (((adev->pdev->device == 0x6987) && - ((adev->pdev->revision == 0xc0) || - (adev->pdev->revision == 0xc3))) || - ((adev->pdev->device == 0x6981) && - ((adev->pdev->revision == 0x00) || - (adev->pdev->revision == 0x01) || - (adev->pdev->revision == 0x10)))) + if (ASICID_IS_P23(adev->pdev->device, adev->pdev->revision)) chip_name = "polaris12_k"; else chip_name = "polaris12"; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 3ebbddb63705..0c3421d587e8 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -510,15 +510,16 @@ static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev, } static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, - struct amdgpu_irq_src *source, - struct amdgpu_iv_entry *entry) + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) { - struct amdgpu_vmhub *hub; bool retry_fault = !!(entry->src_data[1] & 0x80); uint32_t status = 0, cid = 0, rw = 0; - u64 addr; - char hub_name[10]; + struct amdgpu_task_info task_info; + struct amdgpu_vmhub *hub; const char *mmhub_cid; + const char *hub_name; + u64 addr; addr = (u64)entry->src_data[0] << 12; addr |= ((u64)entry->src_data[1] & 0xf) << 44; @@ -527,105 +528,103 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, entry->timestamp)) return 1; /* This also prevents sending it to KFD */ + /* If it's the first fault for this address, process it normally */ + if (retry_fault && !in_interrupt() && + amdgpu_vm_handle_fault(adev, entry->pasid, addr)) + return 1; /* This also prevents sending it to KFD */ + + if (!printk_ratelimit()) + return 0; + if (entry->client_id == SOC15_IH_CLIENTID_VMC) { - snprintf(hub_name, sizeof(hub_name), "mmhub0"); + hub_name = "mmhub0"; hub = &adev->vmhub[AMDGPU_MMHUB_0]; } else if (entry->client_id == SOC15_IH_CLIENTID_VMC1) { - snprintf(hub_name, sizeof(hub_name), "mmhub1"); + hub_name = "mmhub1"; hub = &adev->vmhub[AMDGPU_MMHUB_1]; } else { - snprintf(hub_name, sizeof(hub_name), "gfxhub0"); + hub_name = "gfxhub0"; hub = &adev->vmhub[AMDGPU_GFXHUB_0]; } - /* If it's the first fault for this address, process it normally */ - if (retry_fault && !in_interrupt() && - amdgpu_vm_handle_fault(adev, entry->pasid, addr)) - return 1; /* This also prevents sending it to KFD */ + memset(&task_info, 0, sizeof(struct amdgpu_task_info)); + amdgpu_vm_get_task_info(adev, entry->pasid, &task_info); - if (!amdgpu_sriov_vf(adev)) { - /* - * Issue a dummy read to wait for the status register to - * be updated to avoid reading an incorrect value due to - * the new fast GRBM interface. - */ - if (entry->vmid_src == AMDGPU_GFXHUB_0) - RREG32(hub->vm_l2_pro_fault_status); - - status = RREG32(hub->vm_l2_pro_fault_status); - cid = REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, CID); - rw = REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, RW); - WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1); - } + dev_err(adev->dev, + "[%s] %s page fault (src_id:%u ring:%u vmid:%u " + "pasid:%u, for process %s pid %d thread %s pid %d)\n", + hub_name, retry_fault ? "retry" : "no-retry", + entry->src_id, entry->ring_id, entry->vmid, + entry->pasid, task_info.process_name, task_info.tgid, + task_info.task_name, task_info.pid); + dev_err(adev->dev, " in page starting at address 0x%016llx from client %d\n", + addr, entry->client_id); - if (printk_ratelimit()) { - struct amdgpu_task_info task_info; - - memset(&task_info, 0, sizeof(struct amdgpu_task_info)); - amdgpu_vm_get_task_info(adev, entry->pasid, &task_info); - - dev_err(adev->dev, - "[%s] %s page fault (src_id:%u ring:%u vmid:%u " - "pasid:%u, for process %s pid %d thread %s pid %d)\n", - hub_name, retry_fault ? "retry" : "no-retry", - entry->src_id, entry->ring_id, entry->vmid, - entry->pasid, task_info.process_name, task_info.tgid, - task_info.task_name, task_info.pid); - dev_err(adev->dev, " in page starting at address 0x%016llx from client %d\n", - addr, entry->client_id); - if (!amdgpu_sriov_vf(adev)) { - dev_err(adev->dev, - "VM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", - status); - if (hub == &adev->vmhub[AMDGPU_GFXHUB_0]) { - dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n", - cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : gfxhub_client_ids[cid], - cid); - } else { - switch (adev->asic_type) { - case CHIP_VEGA10: - mmhub_cid = mmhub_client_ids_vega10[cid][rw]; - break; - case CHIP_VEGA12: - mmhub_cid = mmhub_client_ids_vega12[cid][rw]; - break; - case CHIP_VEGA20: - mmhub_cid = mmhub_client_ids_vega20[cid][rw]; - break; - case CHIP_ARCTURUS: - mmhub_cid = mmhub_client_ids_arcturus[cid][rw]; - break; - case CHIP_RAVEN: - mmhub_cid = mmhub_client_ids_raven[cid][rw]; - break; - case CHIP_RENOIR: - mmhub_cid = mmhub_client_ids_renoir[cid][rw]; - break; - default: - mmhub_cid = NULL; - break; - } - dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n", - mmhub_cid ? mmhub_cid : "unknown", cid); - } - dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n", - REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS)); - dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n", - REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR)); - dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n", - REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS)); - dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n", - REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR)); - dev_err(adev->dev, "\t RW: 0x%x\n", rw); + if (amdgpu_sriov_vf(adev)) + return 0; + + /* + * Issue a dummy read to wait for the status register to + * be updated to avoid reading an incorrect value due to + * the new fast GRBM interface. + */ + if (entry->vmid_src == AMDGPU_GFXHUB_0) + RREG32(hub->vm_l2_pro_fault_status); + + status = RREG32(hub->vm_l2_pro_fault_status); + cid = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, CID); + rw = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, RW); + WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1); + + + dev_err(adev->dev, + "VM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", + status); + if (hub == &adev->vmhub[AMDGPU_GFXHUB_0]) { + dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n", + cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : + gfxhub_client_ids[cid], + cid); + } else { + switch (adev->asic_type) { + case CHIP_VEGA10: + mmhub_cid = mmhub_client_ids_vega10[cid][rw]; + break; + case CHIP_VEGA12: + mmhub_cid = mmhub_client_ids_vega12[cid][rw]; + break; + case CHIP_VEGA20: + mmhub_cid = mmhub_client_ids_vega20[cid][rw]; + break; + case CHIP_ARCTURUS: + mmhub_cid = mmhub_client_ids_arcturus[cid][rw]; + break; + case CHIP_RAVEN: + mmhub_cid = mmhub_client_ids_raven[cid][rw]; + break; + case CHIP_RENOIR: + mmhub_cid = mmhub_client_ids_renoir[cid][rw]; + break; + default: + mmhub_cid = NULL; + break; } + dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n", + mmhub_cid ? mmhub_cid : "unknown", cid); } - + dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n", + REG_GET_FIELD(status, + VM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS)); + dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n", + REG_GET_FIELD(status, + VM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR)); + dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n", + REG_GET_FIELD(status, + VM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS)); + dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n", + REG_GET_FIELD(status, + VM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR)); + dev_err(adev->dev, "\t RW: 0x%x\n", rw); return 0; } @@ -1166,15 +1165,7 @@ static void gmc_v9_0_set_mmhub_funcs(struct amdgpu_device *adev) static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev) { - switch (adev->asic_type) { - case CHIP_ARCTURUS: - case CHIP_VEGA20: - adev->gfxhub.funcs = &gfxhub_v1_1_funcs; - break; - default: - adev->gfxhub.funcs = &gfxhub_v1_0_funcs; - break; - } + adev->gfxhub.funcs = &gfxhub_v1_0_funcs; } static int gmc_v9_0_early_init(void *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c index 94caf5204c8b..6b80dcea80ec 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c @@ -32,19 +32,19 @@ #include "vcn/vcn_2_0_0_sh_mask.h" #include "ivsrcid/vcn/irqsrcs_vcn_2_0.h" -#define mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET 0x1bfff +#define mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET 0x1bfff #define mmUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET 0x4029 #define mmUVD_JPEG_GPCOM_DATA0_INTERNAL_OFFSET 0x402a #define mmUVD_JPEG_GPCOM_DATA1_INTERNAL_OFFSET 0x402b #define mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40ea -#define mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40eb +#define mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40eb #define mmUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET 0x40cf #define mmUVD_LMI_JPEG_VMID_INTERNAL_OFFSET 0x40d1 -#define mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40e8 +#define mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40e8 #define mmUVD_LMI_JRBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40e9 #define mmUVD_JRBC_IB_SIZE_INTERNAL_OFFSET 0x4082 #define mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40ec -#define mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40ed +#define mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40ed #define mmUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET 0x4085 #define mmUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET 0x4084 #define mmUVD_JRBC_STATUS_INTERNAL_OFFSET 0x4089 @@ -247,7 +247,7 @@ static int jpeg_v2_0_disable_power_gating(struct amdgpu_device *adev) return 0; } -static int jpeg_v2_0_enable_power_gating(struct amdgpu_device* adev) +static int jpeg_v2_0_enable_power_gating(struct amdgpu_device *adev) { if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) { uint32_t data; @@ -274,7 +274,7 @@ static int jpeg_v2_0_enable_power_gating(struct amdgpu_device* adev) return 0; } -static void jpeg_v2_0_disable_clock_gating(struct amdgpu_device* adev) +static void jpeg_v2_0_disable_clock_gating(struct amdgpu_device *adev) { uint32_t data; @@ -297,7 +297,7 @@ static void jpeg_v2_0_disable_clock_gating(struct amdgpu_device* adev) WREG32_SOC15(JPEG, 0, mmJPEG_CGC_GATE, data); } -static void jpeg_v2_0_enable_clock_gating(struct amdgpu_device* adev) +static void jpeg_v2_0_enable_clock_gating(struct amdgpu_device *adev) { uint32_t data; diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c index 845306f63cdb..c6724a0e0c43 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c @@ -247,7 +247,7 @@ static int jpeg_v2_5_resume(void *handle) return r; } -static void jpeg_v2_5_disable_clock_gating(struct amdgpu_device* adev, int inst) +static void jpeg_v2_5_disable_clock_gating(struct amdgpu_device *adev, int inst) { uint32_t data; @@ -276,7 +276,7 @@ static void jpeg_v2_5_disable_clock_gating(struct amdgpu_device* adev, int inst) WREG32_SOC15(JPEG, inst, mmJPEG_CGC_CTRL, data); } -static void jpeg_v2_5_enable_clock_gating(struct amdgpu_device* adev, int inst) +static void jpeg_v2_5_enable_clock_gating(struct amdgpu_device *adev, int inst) { uint32_t data; diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c index 3a0dff53654d..e8fbb2a0de34 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c @@ -213,7 +213,7 @@ static int jpeg_v3_0_resume(void *handle) return r; } -static void jpeg_v3_0_disable_clock_gating(struct amdgpu_device* adev) +static void jpeg_v3_0_disable_clock_gating(struct amdgpu_device *adev) { uint32_t data = 0; @@ -243,7 +243,7 @@ static void jpeg_v3_0_disable_clock_gating(struct amdgpu_device* adev) WREG32_SOC15(JPEG, 0, mmJPEG_CGC_CTRL, data); } -static void jpeg_v3_0_enable_clock_gating(struct amdgpu_device* adev) +static void jpeg_v3_0_enable_clock_gating(struct amdgpu_device *adev) { uint32_t data = 0; @@ -286,7 +286,7 @@ static int jpeg_v3_0_disable_static_power_gating(struct amdgpu_device *adev) return 0; } -static int jpeg_v3_0_enable_static_power_gating(struct amdgpu_device* adev) +static int jpeg_v3_0_enable_static_power_gating(struct amdgpu_device *adev) { /* enable anti hang mechanism */ WREG32_P(SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_POWER_STATUS), diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c index 1c22d8393b21..985e454463e1 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c @@ -46,7 +46,7 @@ static void mes_v10_1_ring_set_wptr(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; if (ring->use_doorbell) { - atomic64_set((atomic64_t*)&adev->wb.wb[ring->wptr_offs], + atomic64_set((atomic64_t *)&adev->wb.wb[ring->wptr_offs], ring->wptr); WDOORBELL64(ring->doorbell_index, ring->wptr); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c index f84701c562bf..0309d84c887d 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c @@ -409,7 +409,7 @@ static void mmhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev, bool CRASH_ON_NO_RETRY_FAULT, 1); tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL, CRASH_ON_RETRY_FAULT, 1); - } + } WREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL, tmp); } @@ -712,7 +712,7 @@ static int mmhub_v1_0_get_ras_error_count(struct amdgpu_device *adev, uint32_t sec_cnt, ded_cnt; for (i = 0; i < ARRAY_SIZE(mmhub_v1_0_ras_fields); i++) { - if(mmhub_v1_0_ras_fields[i].reg_offset != reg->reg_offset) + if (mmhub_v1_0_ras_fields[i].reg_offset != reg->reg_offset) continue; sec_cnt = (value & diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c index 2063700f0bc6..4ac8ac0c56c8 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c @@ -543,6 +543,7 @@ static void mmhub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *ad switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: def = data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid); def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_Sienna_Cichlid); break; @@ -576,6 +577,7 @@ static void mmhub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *ad switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: if (def != data) WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid, data); if (def1 != data1) @@ -598,6 +600,7 @@ static void mmhub_v2_0_update_medium_grain_light_sleep(struct amdgpu_device *ade switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: def = data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid); break; default: @@ -614,6 +617,7 @@ static void mmhub_v2_0_update_medium_grain_light_sleep(struct amdgpu_device *ade switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid, data); break; default: @@ -635,6 +639,7 @@ static int mmhub_v2_0_set_clockgating(struct amdgpu_device *adev, case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: mmhub_v2_0_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE); mmhub_v2_0_update_medium_grain_light_sleep(adev, @@ -657,6 +662,7 @@ static void mmhub_v2_0_get_clockgating(struct amdgpu_device *adev, u32 *flags) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid); data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_Sienna_Cichlid); break; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c new file mode 100644 index 000000000000..3a248c8cd0b9 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c @@ -0,0 +1,589 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "amdgpu.h" +#include "mmhub_v2_3.h" + +#include "mmhub/mmhub_2_3_0_offset.h" +#include "mmhub/mmhub_2_3_0_sh_mask.h" +#include "mmhub/mmhub_2_3_0_default.h" +#include "navi10_enum.h" + +#include "soc15_common.h" + +static const char *mmhub_client_ids_vangogh[][2] = { + [0][0] = "MP0", + [1][0] = "MP1", + [2][0] = "DCEDMC", + [3][0] = "DCEVGA", + [13][0] = "UTCL2", + [26][0] = "OSS", + [27][0] = "HDP", + [28][0] = "VCN", + [29][0] = "VCNU", + [30][0] = "JPEG", + [0][1] = "MP0", + [1][1] = "MP1", + [2][1] = "DCEDMC", + [3][1] = "DCEVGA", + [4][1] = "DCEDWB", + [5][1] = "XDP", + [26][1] = "OSS", + [27][1] = "HDP", + [28][1] = "VCN", + [29][1] = "VCNU", + [30][1] = "JPEG", +}; + +static uint32_t mmhub_v2_3_get_invalidate_req(unsigned int vmid, + uint32_t flush_type) +{ + u32 req = 0; + + /* invalidate using legacy mode on vmid*/ + req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, + PER_VMID_INVALIDATE_REQ, 1 << vmid); + req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type); + req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1); + req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1); + req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1); + req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1); + req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1); + req = REG_SET_FIELD(req, MMVM_INVALIDATE_ENG0_REQ, + CLEAR_PROTECTION_FAULT_STATUS_ADDR, 0); + + return req; +} + +static void +mmhub_v2_3_print_l2_protection_fault_status(struct amdgpu_device *adev, + uint32_t status) +{ + uint32_t cid, rw; + const char *mmhub_cid = NULL; + + cid = REG_GET_FIELD(status, + MMVM_L2_PROTECTION_FAULT_STATUS, CID); + rw = REG_GET_FIELD(status, + MMVM_L2_PROTECTION_FAULT_STATUS, RW); + + dev_err(adev->dev, + "MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", + status); + switch (adev->asic_type) { + case CHIP_VANGOGH: + mmhub_cid = mmhub_client_ids_vangogh[cid][rw]; + break; + default: + mmhub_cid = NULL; + break; + } + dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n", + mmhub_cid ? mmhub_cid : "unknown", cid); + dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n", + REG_GET_FIELD(status, + MMVM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS)); + dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n", + REG_GET_FIELD(status, + MMVM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR)); + dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n", + REG_GET_FIELD(status, + MMVM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS)); + dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n", + REG_GET_FIELD(status, + MMVM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR)); + dev_err(adev->dev, "\t RW: 0x%x\n", rw); +} + +static void mmhub_v2_3_setup_vm_pt_regs(struct amdgpu_device *adev, + uint32_t vmid, + uint64_t page_table_base) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + + WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32, + hub->ctx_addr_distance * vmid, lower_32_bits(page_table_base)); + + WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32, + hub->ctx_addr_distance * vmid, upper_32_bits(page_table_base)); +} + +static void mmhub_v2_3_init_gart_aperture_regs(struct amdgpu_device *adev) +{ + uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo); + + mmhub_v2_3_setup_vm_pt_regs(adev, 0, pt_base); + + WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, + (u32)(adev->gmc.gart_start >> 12)); + WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, + (u32)(adev->gmc.gart_start >> 44)); + + WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, + (u32)(adev->gmc.gart_end >> 12)); + WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32, + (u32)(adev->gmc.gart_end >> 44)); +} + +static void mmhub_v2_3_init_system_aperture_regs(struct amdgpu_device *adev) +{ + uint64_t value; + uint32_t tmp; + + /* Disable AGP. */ + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_BASE, 0); + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_TOP, 0); + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_BOT, 0x00FFFFFF); + + /* Program the system aperture low logical page number. */ + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_LOW_ADDR, + adev->gmc.vram_start >> 18); + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR, + adev->gmc.vram_end >> 18); + + /* Set default page address. */ + value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start + + adev->vm_manager.vram_base_offset; + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, + (u32)(value >> 12)); + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, + (u32)(value >> 44)); + + /* Program "protection fault". */ + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32, + (u32)(adev->dummy_page_addr >> 12)); + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32, + (u32)((u64)adev->dummy_page_addr >> 44)); + + tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL2); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL2, + ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1); + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL2, tmp); +} + +static void mmhub_v2_3_init_tlb_regs(struct amdgpu_device *adev) +{ + uint32_t tmp; + + /* Setup TLB control */ + tmp = RREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL); + + tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1); + tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3); + tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, + ENABLE_ADVANCED_DRIVER_MODEL, 1); + tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, + SYSTEM_APERTURE_UNMAPPED_ACCESS, 0); + tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0); + tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, + MTYPE, MTYPE_UC); /* UC, uncached */ + + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL, tmp); +} + +static void mmhub_v2_3_init_cache_regs(struct amdgpu_device *adev) +{ + uint32_t tmp; + + /* Setup L2 cache */ + tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 1); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, + ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1); + /* XXX for emulation, Refer to closed source code.*/ + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE, + 0); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 0); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0); + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL, tmp); + + tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL2); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1); + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL2, tmp); + + tmp = mmMMVM_L2_CNTL3_DEFAULT; + if (adev->gmc.translate_further) { + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 12); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, + L2_CACHE_BIGK_FRAGMENT_SIZE, 9); + } else { + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, BANK_SELECT, 9); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL3, + L2_CACHE_BIGK_FRAGMENT_SIZE, 6); + } + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL3, tmp); + + tmp = mmMMVM_L2_CNTL4_DEFAULT; + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0); + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL4, tmp); + + tmp = mmMMVM_L2_CNTL5_DEFAULT; + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL5, L2_CACHE_SMALLK_FRAGMENT_SIZE, 0); + WREG32_SOC15(GC, 0, mmMMVM_L2_CNTL5, tmp); +} + +static void mmhub_v2_3_enable_system_domain(struct amdgpu_device *adev) +{ + uint32_t tmp; + + tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_CNTL); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, + RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0); + WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_CNTL, tmp); +} + +static void mmhub_v2_3_disable_identity_aperture(struct amdgpu_device *adev) +{ + WREG32_SOC15(MMHUB, 0, + mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32, + 0xFFFFFFFF); + WREG32_SOC15(MMHUB, 0, + mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32, + 0x0000000F); + + WREG32_SOC15(MMHUB, 0, + mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0); + WREG32_SOC15(MMHUB, 0, + mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0); + + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, + 0); + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, + 0); +} + +static void mmhub_v2_3_setup_vmid_config(struct amdgpu_device *adev) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + int i; + uint32_t tmp; + + for (i = 0; i <= 14; i++) { + tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_CNTL, i); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH, + adev->vm_manager.num_level); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, + RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, + DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, + 1); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, + PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, + VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, + READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, + WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, + EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1); + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, + PAGE_TABLE_BLOCK_SIZE, + adev->vm_manager.block_size - 9); + /* Send no-retry XNACK on fault to suppress VM fault storm. */ + tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, + RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, + !amdgpu_noretry); + WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_CNTL, + i * hub->ctx_distance, tmp); + WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, + i * hub->ctx_addr_distance, 0); + WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, + i * hub->ctx_addr_distance, 0); + WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32, + i * hub->ctx_addr_distance, + lower_32_bits(adev->vm_manager.max_pfn - 1)); + WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32, + i * hub->ctx_addr_distance, + upper_32_bits(adev->vm_manager.max_pfn - 1)); + } +} + +static void mmhub_v2_3_program_invalidation(struct amdgpu_device *adev) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + unsigned i; + + for (i = 0; i < 18; ++i) { + WREG32_SOC15_OFFSET(MMHUB, 0, + mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32, + i * hub->eng_addr_distance, 0xffffffff); + WREG32_SOC15_OFFSET(MMHUB, 0, + mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_HI32, + i * hub->eng_addr_distance, 0x1f); + } +} + +static int mmhub_v2_3_gart_enable(struct amdgpu_device *adev) +{ + if (amdgpu_sriov_vf(adev)) { + /* + * MMMC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are + * VF copy registers so vbios post doesn't program them, for + * SRIOV driver need to program them + */ + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_FB_LOCATION_BASE, + adev->gmc.vram_start >> 24); + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_FB_LOCATION_TOP, + adev->gmc.vram_end >> 24); + } + + /* GART Enable. */ + mmhub_v2_3_init_gart_aperture_regs(adev); + mmhub_v2_3_init_system_aperture_regs(adev); + mmhub_v2_3_init_tlb_regs(adev); + mmhub_v2_3_init_cache_regs(adev); + + mmhub_v2_3_enable_system_domain(adev); + mmhub_v2_3_disable_identity_aperture(adev); + mmhub_v2_3_setup_vmid_config(adev); + mmhub_v2_3_program_invalidation(adev); + + return 0; +} + +static void mmhub_v2_3_gart_disable(struct amdgpu_device *adev) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + u32 tmp; + u32 i; + + /* Disable all tables */ + for (i = 0; i < 16; i++) + WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_CNTL, + i * hub->ctx_distance, 0); + + /* Setup TLB control */ + tmp = RREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL); + tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0); + tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, + ENABLE_ADVANCED_DRIVER_MODEL, 0); + WREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL, tmp); + + /* Setup L2 cache */ + tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL); + tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 0); + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL, tmp); + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL3, 0); +} + +/** + * mmhub_v2_3_set_fault_enable_default - update GART/VM fault handling + * + * @adev: amdgpu_device pointer + * @value: true redirects VM faults to the default page + */ +static void mmhub_v2_3_set_fault_enable_default(struct amdgpu_device *adev, + bool value) +{ + u32 tmp; + tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT, + value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + READ_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value); + if (!value) { + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + CRASH_ON_NO_RETRY_FAULT, 1); + tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL, + CRASH_ON_RETRY_FAULT, 1); + } + WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL, tmp); +} + +static const struct amdgpu_vmhub_funcs mmhub_v2_3_vmhub_funcs = { + .print_l2_protection_fault_status = mmhub_v2_3_print_l2_protection_fault_status, + .get_invalidate_req = mmhub_v2_3_get_invalidate_req, +}; + +static void mmhub_v2_3_init(struct amdgpu_device *adev) +{ + struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB_0]; + + hub->ctx0_ptb_addr_lo32 = + SOC15_REG_OFFSET(MMHUB, 0, + mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32); + hub->ctx0_ptb_addr_hi32 = + SOC15_REG_OFFSET(MMHUB, 0, + mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32); + hub->vm_inv_eng0_sem = + SOC15_REG_OFFSET(MMHUB, 0, + mmMMVM_INVALIDATE_ENG0_SEM); + hub->vm_inv_eng0_req = + SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_REQ); + hub->vm_inv_eng0_ack = + SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_ACK); + hub->vm_context0_cntl = + SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_CNTL); + hub->vm_l2_pro_fault_status = + SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_STATUS); + hub->vm_l2_pro_fault_cntl = + SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL); + + hub->ctx_distance = mmMMVM_CONTEXT1_CNTL - mmMMVM_CONTEXT0_CNTL; + hub->ctx_addr_distance = mmMMVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 - + mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32; + hub->eng_distance = mmMMVM_INVALIDATE_ENG1_REQ - + mmMMVM_INVALIDATE_ENG0_REQ; + hub->eng_addr_distance = mmMMVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 - + mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32; + + hub->vm_cntx_cntl_vm_fault = MMVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK | + MMVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK | + MMVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK | + MMVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK | + MMVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK | + MMVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK | + MMVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK; + + hub->vmhub_funcs = &mmhub_v2_3_vmhub_funcs; +} + +static void +mmhub_v2_3_update_medium_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + uint32_t def, data, def1, data1; + + def = data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG); + def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2); + + if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)) { + data |= MM_ATC_L2_MISC_CG__ENABLE_MASK; + + data1 &= ~(DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK); + + } else { + data &= ~MM_ATC_L2_MISC_CG__ENABLE_MASK; + + data1 |= (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK); + } + + if (def != data) + WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG, data); + if (def1 != data1) + WREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2, data1); +} + +static void +mmhub_v2_3_update_medium_grain_light_sleep(struct amdgpu_device *adev, + bool enable) +{ + uint32_t def, data; + + def = data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG); + + if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS)) + data |= MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK; + else + data &= ~MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK; + + if (def != data) + WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG, data); +} + +static int mmhub_v2_3_set_clockgating(struct amdgpu_device *adev, + enum amd_clockgating_state state) +{ + if (amdgpu_sriov_vf(adev)) + return 0; + + mmhub_v2_3_update_medium_grain_clock_gating(adev, + state == AMD_CG_STATE_GATE ? true : false); + mmhub_v2_3_update_medium_grain_light_sleep(adev, + state == AMD_CG_STATE_GATE ? true : false); + + return 0; +} + +static void mmhub_v2_3_get_clockgating(struct amdgpu_device *adev, u32 *flags) +{ + int data, data1; + + if (amdgpu_sriov_vf(adev)) + *flags = 0; + + data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG); + data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2); + + /* AMD_CG_SUPPORT_MC_MGCG */ + if ((data & MM_ATC_L2_MISC_CG__ENABLE_MASK) && + !(data1 & (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK | + DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK))) + *flags |= AMD_CG_SUPPORT_MC_MGCG; + + /* AMD_CG_SUPPORT_MC_LS */ + if (data & MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK) + *flags |= AMD_CG_SUPPORT_MC_LS; +} + +const struct amdgpu_mmhub_funcs mmhub_v2_3_funcs = { + .ras_late_init = amdgpu_mmhub_ras_late_init, + .init = mmhub_v2_3_init, + .gart_enable = mmhub_v2_3_gart_enable, + .set_fault_enable_default = mmhub_v2_3_set_fault_enable_default, + .gart_disable = mmhub_v2_3_gart_disable, + .set_clockgating = mmhub_v2_3_set_clockgating, + .get_clockgating = mmhub_v2_3_get_clockgating, + .setup_vm_pt_regs = mmhub_v2_3_setup_vm_pt_regs, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.h new file mode 100644 index 000000000000..2926d21dea8b --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.h @@ -0,0 +1,28 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __MMHUB_V2_3_H__ +#define __MMHUB_V2_3_H__ + +extern const struct amdgpu_mmhub_funcs mmhub_v2_3_funcs; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c index 74b1e7dc49a9..837769fcb35b 100644 --- a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c @@ -314,6 +314,8 @@ static int navi10_ih_irq_init(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN_Sienna_Cichlid); ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1); @@ -660,8 +662,11 @@ static int navi10_ih_sw_init(void *handle) /* use gpu virtual address for ih ring * until ih_checken is programmed to allow * use bus address for ih ring by psp bl */ - use_bus_addr = - (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) ? false : true; + if ((adev->flags & AMD_IS_APU) || + (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) + use_bus_addr = false; + else + use_bus_addr = true; r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, use_bus_addr); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c index 7429f30398b9..e0048806afaa 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c @@ -28,10 +28,12 @@ #include "nbio/nbio_2_3_offset.h" #include "nbio/nbio_2_3_sh_mask.h" #include <uapi/linux/kfd_ioctl.h> +#include <linux/pci.h> #define smnPCIE_CONFIG_CNTL 0x11180044 #define smnCPM_CONTROL 0x11180460 #define smnPCIE_CNTL2 0x11180070 +#define smnPCIE_LC_CNTL 0x11140280 #define mmBIF_SDMA2_DOORBELL_RANGE 0x01d6 #define mmBIF_SDMA2_DOORBELL_RANGE_BASE_IDX 2 @@ -312,6 +314,42 @@ static void nbio_v2_3_init_registers(struct amdgpu_device *adev) WREG32_PCIE(smnPCIE_CONFIG_CNTL, data); } +#define NAVI10_PCIE__LC_L0S_INACTIVITY_DEFAULT 0x00000000 // off by default, no gains over L1 +#define NAVI10_PCIE__LC_L1_INACTIVITY_DEFAULT 0x00000009 // 1=1us, 9=1ms +#define NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT 0x0000000E // 4ms + +static void nbio_v2_3_enable_aspm(struct amdgpu_device *adev, + bool enable) +{ + uint32_t def, data; + + def = data = RREG32_PCIE(smnPCIE_LC_CNTL); + + if (enable) { + /* Disable ASPM L0s/L1 first */ + data &= ~(PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK | PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK); + + data |= NAVI10_PCIE__LC_L0S_INACTIVITY_DEFAULT << PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT; + + if (pci_is_thunderbolt_attached(adev->pdev)) + data |= NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT; + else + data |= NAVI10_PCIE__LC_L1_INACTIVITY_DEFAULT << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT; + + data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK; + } else { + /* Disbale ASPM L1 */ + data &= ~PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK; + /* Disable ASPM TxL0s */ + data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK; + /* Disable ACPI L1 */ + data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK; + } + + if (def != data) + WREG32_PCIE(smnPCIE_LC_CNTL, data); +} + const struct amdgpu_nbio_funcs nbio_v2_3_funcs = { .get_hdp_flush_req_offset = nbio_v2_3_get_hdp_flush_req_offset, .get_hdp_flush_done_offset = nbio_v2_3_get_hdp_flush_done_offset, @@ -332,4 +370,5 @@ const struct amdgpu_nbio_funcs nbio_v2_3_funcs = { .ih_control = nbio_v2_3_ih_control, .init_registers = nbio_v2_3_init_registers, .remap_hdp_registers = nbio_v2_3_remap_hdp_registers, + .enable_aspm = nbio_v2_3_enable_aspm, }; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c b/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c index 7b2fb050407d..d2f1fe55d388 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c @@ -32,7 +32,7 @@ static u32 nbio_v6_1_get_rev_id(struct amdgpu_device *adev) { - u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0); + u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0); tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK; tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT; @@ -114,7 +114,7 @@ static void nbio_v6_1_enable_doorbell_selfring_aperture(struct amdgpu_device *ad static void nbio_v6_1_ih_doorbell_range(struct amdgpu_device *adev, bool use_doorbell, int doorbell_index) { - u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0 , mmBIF_IH_DOORBELL_RANGE); + u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE); if (use_doorbell) { ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, OFFSET, doorbell_index); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c index d34628e113fc..ae685813c419 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c @@ -43,7 +43,7 @@ static void nbio_v7_0_remap_hdp_registers(struct amdgpu_device *adev) static u32 nbio_v7_0_get_rev_id(struct amdgpu_device *adev) { - u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0); + u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0); tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK; tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT; @@ -126,7 +126,7 @@ static void nbio_v7_0_enable_doorbell_selfring_aperture(struct amdgpu_device *ad static void nbio_v7_0_ih_doorbell_range(struct amdgpu_device *adev, bool use_doorbell, int doorbell_index) { - u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0 , mmBIF_IH_DOORBELL_RANGE); + u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE); if (use_doorbell) { ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, OFFSET, doorbell_index); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c new file mode 100644 index 000000000000..aa36022670f9 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c @@ -0,0 +1,341 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "amdgpu_atombios.h" +#include "nbio_v7_2.h" + +#include "nbio/nbio_7_2_0_offset.h" +#include "nbio/nbio_7_2_0_sh_mask.h" +#include <uapi/linux/kfd_ioctl.h> + +static void nbio_v7_2_remap_hdp_registers(struct amdgpu_device *adev) +{ + WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_MEM_FLUSH_CNTL, + adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL); + WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_REG_FLUSH_CNTL, + adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_REG_FLUSH_CNTL); +} + +static u32 nbio_v7_2_get_rev_id(struct amdgpu_device *adev) +{ + u32 tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0); + + tmp &= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK; + tmp >>= RCC_STRAP0_RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT; + + return tmp; +} + +static void nbio_v7_2_mc_access_enable(struct amdgpu_device *adev, bool enable) +{ + if (enable) + WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN, + BIF_BX0_BIF_FB_EN__FB_READ_EN_MASK | + BIF_BX0_BIF_FB_EN__FB_WRITE_EN_MASK); + else + WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN, 0); +} + +static void nbio_v7_2_hdp_flush(struct amdgpu_device *adev, + struct amdgpu_ring *ring) +{ + if (!ring || !ring->funcs->emit_wreg) + WREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); + else + amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); +} + +static u32 nbio_v7_2_get_memsize(struct amdgpu_device *adev) +{ + return RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_CONFIG_MEMSIZE); +} + +static void nbio_v7_2_sdma_doorbell_range(struct amdgpu_device *adev, int instance, + bool use_doorbell, int doorbell_index, + int doorbell_size) +{ + u32 reg = SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_SDMA0_DOORBELL_RANGE); + u32 doorbell_range = RREG32_PCIE_PORT(reg); + + if (use_doorbell) { + doorbell_range = REG_SET_FIELD(doorbell_range, + GDC0_BIF_SDMA0_DOORBELL_RANGE, + OFFSET, doorbell_index); + doorbell_range = REG_SET_FIELD(doorbell_range, + GDC0_BIF_SDMA0_DOORBELL_RANGE, + SIZE, doorbell_size); + } else { + doorbell_range = REG_SET_FIELD(doorbell_range, + GDC0_BIF_SDMA0_DOORBELL_RANGE, + SIZE, 0); + } + + WREG32_PCIE_PORT(reg, doorbell_range); +} + +static void nbio_v7_2_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell, + int doorbell_index, int instance) +{ + u32 reg = SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VCN0_DOORBELL_RANGE); + u32 doorbell_range = RREG32_PCIE_PORT(reg); + + if (use_doorbell) { + doorbell_range = REG_SET_FIELD(doorbell_range, + GDC0_BIF_VCN0_DOORBELL_RANGE, OFFSET, + doorbell_index); + doorbell_range = REG_SET_FIELD(doorbell_range, + GDC0_BIF_VCN0_DOORBELL_RANGE, SIZE, 8); + } else { + doorbell_range = REG_SET_FIELD(doorbell_range, + GDC0_BIF_VCN0_DOORBELL_RANGE, SIZE, 0); + } + + WREG32_PCIE_PORT(reg, doorbell_range); +} + +static void nbio_v7_2_enable_doorbell_aperture(struct amdgpu_device *adev, + bool enable) +{ + u32 reg; + + reg = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN); + reg = REG_SET_FIELD(reg, RCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN, + BIF_DOORBELL_APER_EN, enable ? 1 : 0); + + WREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF0_0_RCC_DOORBELL_APER_EN, reg); +} + +static void nbio_v7_2_enable_doorbell_selfring_aperture(struct amdgpu_device *adev, + bool enable) +{ + u32 tmp = 0; + + if (enable) { + tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, + DOORBELL_SELFRING_GPA_APER_EN, 1) | + REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, + DOORBELL_SELFRING_GPA_APER_MODE, 1) | + REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, + DOORBELL_SELFRING_GPA_APER_SIZE, 0); + + WREG32_SOC15(NBIO, 0, + regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_LOW, + lower_32_bits(adev->doorbell.base)); + WREG32_SOC15(NBIO, 0, + regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_HIGH, + upper_32_bits(adev->doorbell.base)); + } + + WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, + tmp); +} + + +static void nbio_v7_2_ih_doorbell_range(struct amdgpu_device *adev, + bool use_doorbell, int doorbell_index) +{ + u32 ih_doorbell_range = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_IH_DOORBELL_RANGE)); + + if (use_doorbell) { + ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, + GDC0_BIF_IH_DOORBELL_RANGE, OFFSET, + doorbell_index); + ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, + GDC0_BIF_IH_DOORBELL_RANGE, SIZE, + 2); + } else { + ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, + GDC0_BIF_IH_DOORBELL_RANGE, SIZE, + 0); + } + + WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_IH_DOORBELL_RANGE), + ih_doorbell_range); +} + +static void nbio_v7_2_ih_control(struct amdgpu_device *adev) +{ + u32 interrupt_cntl; + + /* setup interrupt control */ + WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL2, + adev->dummy_page_addr >> 8); + + interrupt_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL); + /* + * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi + * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN + */ + interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL, + IH_DUMMY_RD_OVERRIDE, 0); + + /* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */ + interrupt_cntl = REG_SET_FIELD(interrupt_cntl, BIF_BX0_INTERRUPT_CNTL, + IH_REQ_NONSNOOP_EN, 0); + + WREG32_SOC15(NBIO, 0, regBIF_BX0_INTERRUPT_CNTL, interrupt_cntl); +} + +static void nbio_v7_2_update_medium_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + uint32_t def, data; + + def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regCPM_CONTROL)); + if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG)) { + data |= (CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK | + CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK | + CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK | + CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK | + CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK | + CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK); + } else { + data &= ~(CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK | + CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK | + CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK | + CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK | + CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK | + CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK); + } + + if (def != data) + WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regCPM_CONTROL), data); +} + +static void nbio_v7_2_update_medium_grain_light_sleep(struct amdgpu_device *adev, + bool enable) +{ + uint32_t def, data; + + def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2)); + if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) { + data |= (PCIE_CNTL2__SLV_MEM_LS_EN_MASK | + PCIE_CNTL2__MST_MEM_LS_EN_MASK | + PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK); + } else { + data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK | + PCIE_CNTL2__MST_MEM_LS_EN_MASK | + PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK); + } + + if (def != data) + WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2), data); +} + +static void nbio_v7_2_get_clockgating_state(struct amdgpu_device *adev, + u32 *flags) +{ + int data; + + /* AMD_CG_SUPPORT_BIF_MGCG */ + data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regCPM_CONTROL)); + if (data & CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK) + *flags |= AMD_CG_SUPPORT_BIF_MGCG; + + /* AMD_CG_SUPPORT_BIF_LS */ + data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2)); + if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK) + *flags |= AMD_CG_SUPPORT_BIF_LS; +} + +static u32 nbio_v7_2_get_hdp_flush_req_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_REQ); +} + +static u32 nbio_v7_2_get_hdp_flush_done_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_GPU_HDP_FLUSH_DONE); +} + +static u32 nbio_v7_2_get_pcie_index_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX0_PCIE_INDEX2); +} + +static u32 nbio_v7_2_get_pcie_data_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX0_PCIE_DATA2); +} + +static u32 nbio_v7_2_get_pcie_port_index_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_INDEX); +} + +static u32 nbio_v7_2_get_pcie_port_data_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(NBIO, 0, regBIF_BX_PF0_RSMU_DATA); +} + +const struct nbio_hdp_flush_reg nbio_v7_2_hdp_flush_reg = { + .ref_and_mask_cp0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP0_MASK, + .ref_and_mask_cp1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP1_MASK, + .ref_and_mask_cp2 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP2_MASK, + .ref_and_mask_cp3 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP3_MASK, + .ref_and_mask_cp4 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP4_MASK, + .ref_and_mask_cp5 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP5_MASK, + .ref_and_mask_cp6 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP6_MASK, + .ref_and_mask_cp7 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP7_MASK, + .ref_and_mask_cp8 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP8_MASK, + .ref_and_mask_cp9 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP9_MASK, + .ref_and_mask_sdma0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA0_MASK, + .ref_and_mask_sdma1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA1_MASK, +}; + +static void nbio_v7_2_init_registers(struct amdgpu_device *adev) +{ + uint32_t def, data; + + def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CONFIG_CNTL)); + data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL, CI_SWUS_MAX_READ_REQUEST_SIZE_MODE, 1); + data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL, CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV, 1); + + if (def != data) + WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CONFIG_CNTL), + data); +} + +const struct amdgpu_nbio_funcs nbio_v7_2_funcs = { + .get_hdp_flush_req_offset = nbio_v7_2_get_hdp_flush_req_offset, + .get_hdp_flush_done_offset = nbio_v7_2_get_hdp_flush_done_offset, + .get_pcie_index_offset = nbio_v7_2_get_pcie_index_offset, + .get_pcie_data_offset = nbio_v7_2_get_pcie_data_offset, + .get_pcie_port_index_offset = nbio_v7_2_get_pcie_port_index_offset, + .get_pcie_port_data_offset = nbio_v7_2_get_pcie_port_data_offset, + .get_rev_id = nbio_v7_2_get_rev_id, + .mc_access_enable = nbio_v7_2_mc_access_enable, + .hdp_flush = nbio_v7_2_hdp_flush, + .get_memsize = nbio_v7_2_get_memsize, + .sdma_doorbell_range = nbio_v7_2_sdma_doorbell_range, + .vcn_doorbell_range = nbio_v7_2_vcn_doorbell_range, + .enable_doorbell_aperture = nbio_v7_2_enable_doorbell_aperture, + .enable_doorbell_selfring_aperture = nbio_v7_2_enable_doorbell_selfring_aperture, + .ih_doorbell_range = nbio_v7_2_ih_doorbell_range, + .update_medium_grain_clock_gating = nbio_v7_2_update_medium_grain_clock_gating, + .update_medium_grain_light_sleep = nbio_v7_2_update_medium_grain_light_sleep, + .get_clockgating_state = nbio_v7_2_get_clockgating_state, + .ih_control = nbio_v7_2_ih_control, + .init_registers = nbio_v7_2_init_registers, + .remap_hdp_registers = nbio_v7_2_remap_hdp_registers, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.h new file mode 100644 index 000000000000..a8e8e65648a0 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.h @@ -0,0 +1,32 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NBIO_V7_2_H__ +#define __NBIO_V7_2_H__ + +#include "soc15_common.h" + +extern const struct nbio_hdp_flush_reg nbio_v7_2_hdp_flush_reg; +extern const struct amdgpu_nbio_funcs nbio_v7_2_funcs; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c index 03462c857498..e33d8022cc32 100644 --- a/drivers/gpu/drm/amd/amdgpu/nv.c +++ b/drivers/gpu/drm/amd/amdgpu/nv.c @@ -49,6 +49,7 @@ #include "gfxhub_v2_0.h" #include "mmhub_v2_0.h" #include "nbio_v2_3.h" +#include "nbio_v7_2.h" #include "nv.h" #include "navi10_ih.h" #include "gfx_v10_0.h" @@ -95,6 +96,21 @@ static u64 nv_pcie_rreg64(struct amdgpu_device *adev, u32 reg) return amdgpu_device_indirect_rreg64(adev, address, data, reg); } +static u32 nv_pcie_port_rreg(struct amdgpu_device *adev, u32 reg) +{ + unsigned long flags, address, data; + u32 r; + address = adev->nbio.funcs->get_pcie_port_index_offset(adev); + data = adev->nbio.funcs->get_pcie_port_data_offset(adev); + + spin_lock_irqsave(&adev->pcie_idx_lock, flags); + WREG32(address, reg * 4); + (void)RREG32(address); + r = RREG32(data); + spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); + return r; +} + static void nv_pcie_wreg64(struct amdgpu_device *adev, u32 reg, u64 v) { unsigned long address, data; @@ -105,6 +121,21 @@ static void nv_pcie_wreg64(struct amdgpu_device *adev, u32 reg, u64 v) amdgpu_device_indirect_wreg64(adev, address, data, reg, v); } +static void nv_pcie_port_wreg(struct amdgpu_device *adev, u32 reg, u32 v) +{ + unsigned long flags, address, data; + + address = adev->nbio.funcs->get_pcie_port_index_offset(adev); + data = adev->nbio.funcs->get_pcie_port_data_offset(adev); + + spin_lock_irqsave(&adev->pcie_idx_lock, flags); + WREG32(address, reg * 4); + (void)RREG32(address); + WREG32(data, v); + (void)RREG32(data); + spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); +} + static u32 nv_didt_rreg(struct amdgpu_device *adev, u32 reg) { unsigned long flags, address, data; @@ -254,7 +285,8 @@ static int nv_read_register(struct amdgpu_device *adev, u32 se_num, *value = 0; for (i = 0; i < ARRAY_SIZE(nv_allowed_read_registers); i++) { en = &nv_allowed_read_registers[i]; - if (reg_offset != + if ((i == 7 && (adev->sdma.num_instances == 1)) || /* some asics don't have SDMA1 */ + reg_offset != (adev->reg_offset[en->hwip][en->inst][en->seg] + en->reg_offset)) continue; @@ -443,6 +475,12 @@ legacy_init: case CHIP_NAVY_FLOUNDER: sienna_cichlid_reg_base_init(adev); break; + case CHIP_VANGOGH: + vangogh_reg_base_init(adev); + break; + case CHIP_DIMGREY_CAVEFISH: + dimgrey_cavefish_reg_base_init(adev); + break; default: return -EINVAL; } @@ -455,10 +493,11 @@ void nv_set_virt_ops(struct amdgpu_device *adev) adev->virt.ops = &xgpu_nv_virt_ops; } -static bool nv_is_blockchain_sku(struct pci_dev *pdev) +static bool nv_is_headless_sku(struct pci_dev *pdev) { - if (pdev->device == 0x731E && - (pdev->revision == 0xC6 || pdev->revision == 0xC7)) + if ((pdev->device == 0x731E && + (pdev->revision == 0xC6 || pdev->revision == 0xC7)) || + (pdev->device == 0x7340 && pdev->revision == 0xC9)) return true; return false; } @@ -467,8 +506,13 @@ int nv_set_ip_blocks(struct amdgpu_device *adev) { int r; - adev->nbio.funcs = &nbio_v2_3_funcs; - adev->nbio.hdp_flush_reg = &nbio_v2_3_hdp_flush_reg; + if (adev->flags & AMD_IS_APU) { + adev->nbio.funcs = &nbio_v7_2_funcs; + adev->nbio.hdp_flush_reg = &nbio_v7_2_hdp_flush_reg; + } else { + adev->nbio.funcs = &nbio_v2_3_funcs; + adev->nbio.hdp_flush_reg = &nbio_v2_3_hdp_flush_reg; + } if (adev->asic_type == CHIP_SIENNA_CICHLID) adev->gmc.xgmi.supported = true; @@ -492,7 +536,7 @@ int nv_set_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block); #if defined(CONFIG_DRM_AMD_DC) else if (amdgpu_device_has_dc_support(adev) && - !nv_is_blockchain_sku(adev->pdev)) + !nv_is_headless_sku(adev->pdev)) amdgpu_device_ip_block_add(adev, &dm_ip_block); #endif amdgpu_device_ip_block_add(adev, &gfx_v10_0_ip_block); @@ -500,7 +544,7 @@ int nv_set_ip_blocks(struct amdgpu_device *adev) if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT && !amdgpu_sriov_vf(adev)) amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block); - if (!nv_is_blockchain_sku(adev->pdev)) + if (!nv_is_headless_sku(adev->pdev)) amdgpu_device_ip_block_add(adev, &vcn_v2_0_ip_block); amdgpu_device_ip_block_add(adev, &jpeg_v2_0_ip_block); if (adev->enable_mes) @@ -575,6 +619,44 @@ int nv_set_ip_blocks(struct amdgpu_device *adev) is_support_sw_smu(adev)) amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block); break; + case CHIP_VANGOGH: + amdgpu_device_ip_block_add(adev, &nv_common_ip_block); + amdgpu_device_ip_block_add(adev, &gmc_v10_0_ip_block); + amdgpu_device_ip_block_add(adev, &navi10_ih_ip_block); + if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) + amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block); + amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block); + if (adev->enable_virtual_display || amdgpu_sriov_vf(adev)) + amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block); +#if defined(CONFIG_DRM_AMD_DC) + else if (amdgpu_device_has_dc_support(adev)) + amdgpu_device_ip_block_add(adev, &dm_ip_block); +#endif + amdgpu_device_ip_block_add(adev, &gfx_v10_0_ip_block); + amdgpu_device_ip_block_add(adev, &sdma_v5_2_ip_block); + amdgpu_device_ip_block_add(adev, &vcn_v3_0_ip_block); + amdgpu_device_ip_block_add(adev, &jpeg_v3_0_ip_block); + break; + case CHIP_DIMGREY_CAVEFISH: + amdgpu_device_ip_block_add(adev, &nv_common_ip_block); + amdgpu_device_ip_block_add(adev, &gmc_v10_0_ip_block); + amdgpu_device_ip_block_add(adev, &navi10_ih_ip_block); + if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) + amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block); + if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP && + is_support_sw_smu(adev)) + amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block); + if (adev->enable_virtual_display || amdgpu_sriov_vf(adev)) + amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block); +#if defined(CONFIG_DRM_AMD_DC) + else if (amdgpu_device_has_dc_support(adev)) + amdgpu_device_ip_block_add(adev, &dm_ip_block); +#endif + amdgpu_device_ip_block_add(adev, &gfx_v10_0_ip_block); + amdgpu_device_ip_block_add(adev, &sdma_v5_2_ip_block); + amdgpu_device_ip_block_add(adev, &vcn_v3_0_ip_block); + amdgpu_device_ip_block_add(adev, &jpeg_v3_0_ip_block); + break; default: return -EINVAL; } @@ -671,6 +753,29 @@ static void nv_pre_asic_init(struct amdgpu_device *adev) { } +static int nv_update_umd_stable_pstate(struct amdgpu_device *adev, + bool enter) +{ + if (enter) + amdgpu_gfx_rlc_enter_safe_mode(adev); + else + amdgpu_gfx_rlc_exit_safe_mode(adev); + + if (adev->gfx.funcs->update_perfmon_mgcg) + adev->gfx.funcs->update_perfmon_mgcg(adev, !enter); + + /* + * The ASPM function is not fully enabled and verified on + * Navi yet. Temporarily skip this until ASPM enabled. + */ +#if 0 + if (adev->nbio.funcs->enable_aspm) + adev->nbio.funcs->enable_aspm(adev, !enter); +#endif + + return 0; +} + static const struct amdgpu_asic_funcs nv_asic_funcs = { .read_disabled_bios = &nv_read_disabled_bios, @@ -691,6 +796,7 @@ static const struct amdgpu_asic_funcs nv_asic_funcs = .get_pcie_replay_count = &nv_get_pcie_replay_count, .supports_baco = &nv_asic_supports_baco, .pre_asic_init = &nv_pre_asic_init, + .update_umd_stable_pstate = &nv_update_umd_stable_pstate, }; static int nv_common_early_init(void *handle) @@ -706,6 +812,8 @@ static int nv_common_early_init(void *handle) adev->pcie_wreg = &nv_pcie_wreg; adev->pcie_rreg64 = &nv_pcie_rreg64; adev->pcie_wreg64 = &nv_pcie_wreg64; + adev->pciep_rreg = &nv_pcie_port_rreg; + adev->pciep_wreg = &nv_pcie_port_wreg; /* TODO: will add them during VCN v2 implementation */ adev->uvd_ctx_rreg = NULL; @@ -833,6 +941,46 @@ static int nv_common_early_init(void *handle) adev->external_rev_id = adev->rev_id + 0x32; break; + case CHIP_VANGOGH: + adev->apu_flags |= AMD_APU_IS_VANGOGH; + adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | + AMD_CG_SUPPORT_GFX_MGLS | + AMD_CG_SUPPORT_GFX_CP_LS | + AMD_CG_SUPPORT_GFX_RLC_LS | + AMD_CG_SUPPORT_GFX_CGCG | + AMD_CG_SUPPORT_GFX_CGLS | + AMD_CG_SUPPORT_GFX_3D_CGCG | + AMD_CG_SUPPORT_GFX_3D_CGLS | + AMD_CG_SUPPORT_MC_MGCG | + AMD_CG_SUPPORT_MC_LS | + AMD_CG_SUPPORT_GFX_FGCG | + AMD_CG_SUPPORT_VCN_MGCG | + AMD_CG_SUPPORT_JPEG_MGCG; + adev->pg_flags = AMD_PG_SUPPORT_GFX_PG | + AMD_PG_SUPPORT_VCN | + AMD_PG_SUPPORT_VCN_DPG | + AMD_PG_SUPPORT_JPEG; + if (adev->apu_flags & AMD_APU_IS_VANGOGH) + adev->external_rev_id = adev->rev_id + 0x01; + break; + case CHIP_DIMGREY_CAVEFISH: + adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | + AMD_CG_SUPPORT_GFX_CGCG | + AMD_CG_SUPPORT_GFX_3D_CGCG | + AMD_CG_SUPPORT_VCN_MGCG | + AMD_CG_SUPPORT_JPEG_MGCG | + AMD_CG_SUPPORT_MC_MGCG | + AMD_CG_SUPPORT_MC_LS | + AMD_CG_SUPPORT_HDP_MGCG | + AMD_CG_SUPPORT_HDP_LS | + AMD_CG_SUPPORT_IH_CG; + adev->pg_flags = AMD_PG_SUPPORT_VCN | + AMD_PG_SUPPORT_VCN_DPG | + AMD_PG_SUPPORT_JPEG | + AMD_PG_SUPPORT_ATHUB | + AMD_PG_SUPPORT_MMHUB; + adev->external_rev_id = adev->rev_id + 0x3c; + break; default: /* FIXME: not supported yet */ return -EINVAL; @@ -1060,6 +1208,7 @@ static int nv_common_set_clockgating_state(void *handle, case CHIP_NAVI12: case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: adev->nbio.funcs->update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE); adev->nbio.funcs->update_medium_grain_light_sleep(adev, diff --git a/drivers/gpu/drm/amd/amdgpu/nv.h b/drivers/gpu/drm/amd/amdgpu/nv.h index aeef50a6a54b..515d67bf249f 100644 --- a/drivers/gpu/drm/amd/amdgpu/nv.h +++ b/drivers/gpu/drm/amd/amdgpu/nv.h @@ -34,4 +34,6 @@ int navi10_reg_base_init(struct amdgpu_device *adev); int navi14_reg_base_init(struct amdgpu_device *adev); int navi12_reg_base_init(struct amdgpu_device *adev); int sienna_cichlid_reg_base_init(struct amdgpu_device *adev); +void vangogh_reg_base_init(struct amdgpu_device *adev); +int dimgrey_cavefish_reg_base_init(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h index 4137dc710aaf..d65a5339d354 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h +++ b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h @@ -86,21 +86,22 @@ struct psp_gfx_ctrl /* TEE Gfx Command IDs for the ring buffer interface. */ enum psp_gfx_cmd_id { - GFX_CMD_ID_LOAD_TA = 0x00000001, /* load TA */ - GFX_CMD_ID_UNLOAD_TA = 0x00000002, /* unload TA */ - GFX_CMD_ID_INVOKE_CMD = 0x00000003, /* send command to TA */ - GFX_CMD_ID_LOAD_ASD = 0x00000004, /* load ASD Driver */ - GFX_CMD_ID_SETUP_TMR = 0x00000005, /* setup TMR region */ - GFX_CMD_ID_LOAD_IP_FW = 0x00000006, /* load HW IP FW */ - GFX_CMD_ID_DESTROY_TMR = 0x00000007, /* destroy TMR region */ - GFX_CMD_ID_SAVE_RESTORE = 0x00000008, /* save/restore HW IP FW */ - GFX_CMD_ID_SETUP_VMR = 0x00000009, /* setup VMR region */ - GFX_CMD_ID_DESTROY_VMR = 0x0000000A, /* destroy VMR region */ - GFX_CMD_ID_PROG_REG = 0x0000000B, /* program regs */ - GFX_CMD_ID_CLEAR_VF_FW = 0x0000000D, /* Clear VF FW, to be used on VF shutdown. */ + GFX_CMD_ID_LOAD_TA = 0x00000001, /* load TA */ + GFX_CMD_ID_UNLOAD_TA = 0x00000002, /* unload TA */ + GFX_CMD_ID_INVOKE_CMD = 0x00000003, /* send command to TA */ + GFX_CMD_ID_LOAD_ASD = 0x00000004, /* load ASD Driver */ + GFX_CMD_ID_SETUP_TMR = 0x00000005, /* setup TMR region */ + GFX_CMD_ID_LOAD_IP_FW = 0x00000006, /* load HW IP FW */ + GFX_CMD_ID_DESTROY_TMR = 0x00000007, /* destroy TMR region */ + GFX_CMD_ID_SAVE_RESTORE = 0x00000008, /* save/restore HW IP FW */ + GFX_CMD_ID_SETUP_VMR = 0x00000009, /* setup VMR region */ + GFX_CMD_ID_DESTROY_VMR = 0x0000000A, /* destroy VMR region */ + GFX_CMD_ID_PROG_REG = 0x0000000B, /* program regs */ + GFX_CMD_ID_CLEAR_VF_FW = 0x0000000D, /* Clear VF FW, to be used on VF shutdown. */ + GFX_CMD_ID_GET_FW_ATTESTATION = 0x0000000F, /* Query GPUVA of the Fw Attestation DB */ /* IDs upto 0x1F are reserved for older programs (Raven, Vega 10/12/20) */ - GFX_CMD_ID_LOAD_TOC = 0x00000020, /* Load TOC and obtain TMR size */ - GFX_CMD_ID_AUTOLOAD_RLC = 0x00000021, /* Indicates all graphics fw loaded, start RLC autoload */ + GFX_CMD_ID_LOAD_TOC = 0x00000020, /* Load TOC and obtain TMR size */ + GFX_CMD_ID_AUTOLOAD_RLC = 0x00000021, /* Indicates all graphics fw loaded, start RLC autoload */ }; /* Command to load Trusted Application binary into PSP OS. */ @@ -285,6 +286,25 @@ union psp_gfx_commands struct psp_gfx_cmd_load_toc cmd_load_toc; }; +struct psp_gfx_uresp_reserved +{ + uint32_t reserved[8]; +}; + +/* Command-specific response for Fw Attestation Db */ +struct psp_gfx_uresp_fwar_db_info +{ + uint32_t fwar_db_addr_lo; + uint32_t fwar_db_addr_hi; +}; + +/* Union of command-specific responses for GPCOM ring. */ +union psp_gfx_uresp +{ + struct psp_gfx_uresp_reserved reserved; + struct psp_gfx_uresp_fwar_db_info fwar_db_info; +}; + /* Structure of GFX Response buffer. * For GPCOM I/F it is part of GFX_CMD_RESP buffer, for RBI * it is separate buffer. @@ -297,9 +317,11 @@ struct psp_gfx_resp uint32_t fw_addr_hi; /* +12 bits [63:32] of FW address within TMR (in response to cmd_load_ip_fw command) */ uint32_t tmr_size; /* +16 size of the TMR to be reserved including MM fw and Gfx fw in response to cmd_load_toc command */ - uint32_t reserved[3]; + uint32_t reserved[11]; + + union psp_gfx_uresp uresp; /* +64 response union containing command-specific responses */ - /* total 32 bytes */ + /* total 96 bytes */ }; /* Structure of Command buffer pointed by psp_gfx_rb_frame.cmd_buf_addr_hi diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index 6c5d9612abcb..03e88dbf92be 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -59,6 +59,10 @@ MODULE_FIRMWARE("amdgpu/sienna_cichlid_sos.bin"); MODULE_FIRMWARE("amdgpu/sienna_cichlid_ta.bin"); MODULE_FIRMWARE("amdgpu/navy_flounder_sos.bin"); MODULE_FIRMWARE("amdgpu/navy_flounder_ta.bin"); +MODULE_FIRMWARE("amdgpu/vangogh_asd.bin"); +MODULE_FIRMWARE("amdgpu/vangogh_toc.bin"); +MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_sos.bin"); +MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_asd.bin"); /* address block */ #define smnMP1_FIRMWARE_FLAGS 0x3010024 @@ -77,7 +81,7 @@ static int psp_v11_0_init_microcode(struct psp_context *psp) { struct amdgpu_device *adev = psp->adev; const char *chip_name; - char fw_name[30]; + char fw_name[PSP_FW_NAME_LEN]; int err = 0; const struct ta_firmware_header_v1_0 *ta_hdr; @@ -105,24 +109,26 @@ static int psp_v11_0_init_microcode(struct psp_context *psp) case CHIP_NAVY_FLOUNDER: chip_name = "navy_flounder"; break; + case CHIP_VANGOGH: + chip_name = "vangogh"; + break; + case CHIP_DIMGREY_CAVEFISH: + chip_name = "dimgrey_cavefish"; + break; default: BUG(); } - err = psp_init_sos_microcode(psp, chip_name); - if (err) - return err; - - if (adev->asic_type != CHIP_SIENNA_CICHLID && - adev->asic_type != CHIP_NAVY_FLOUNDER) { - err = psp_init_asd_microcode(psp, chip_name); - if (err) - return err; - } switch (adev->asic_type) { case CHIP_VEGA20: case CHIP_ARCTURUS: + err = psp_init_sos_microcode(psp, chip_name); + if (err) + return err; + err = psp_init_asd_microcode(psp, chip_name); + if (err) + return err; snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ta.bin", chip_name); err = request_firmware(&adev->psp.ta_fw, fw_name, adev->dev); if (err) { @@ -150,6 +156,12 @@ static int psp_v11_0_init_microcode(struct psp_context *psp) case CHIP_NAVI10: case CHIP_NAVI14: case CHIP_NAVI12: + err = psp_init_sos_microcode(psp, chip_name); + if (err) + return err; + err = psp_init_asd_microcode(psp, chip_name); + if (err) + return err; if (amdgpu_sriov_vf(adev)) break; snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ta.bin", chip_name); @@ -180,10 +192,26 @@ static int psp_v11_0_init_microcode(struct psp_context *psp) break; case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + err = psp_init_sos_microcode(psp, chip_name); + if (err) + return err; err = psp_init_ta_microcode(&adev->psp, chip_name); if (err) return err; break; + case CHIP_DIMGREY_CAVEFISH: + err = psp_init_sos_microcode(psp, chip_name); + if (err) + return err; + break; + case CHIP_VANGOGH: + err = psp_init_asd_microcode(psp, chip_name); + if (err) + return err; + err = psp_init_toc_microcode(psp, chip_name); + if (err) + return err; + break; default: BUG(); } @@ -407,8 +435,8 @@ static int psp_v11_0_ring_init(struct psp_context *psp, struct amdgpu_device *adev = psp->adev; if ((!amdgpu_sriov_vf(adev)) && - (adev->asic_type != CHIP_SIENNA_CICHLID) && - (adev->asic_type != CHIP_NAVY_FLOUNDER)) + !(adev->asic_type >= CHIP_SIENNA_CICHLID && + adev->asic_type <= CHIP_DIMGREY_CAVEFISH)) psp_v11_0_reroute_ih(psp); ring = &psp->km_ring; @@ -615,7 +643,7 @@ static int psp_v11_0_memory_training_send_msg(struct psp_context *psp, int msg) static int psp_v11_0_memory_training(struct psp_context *psp, uint32_t ops) { struct psp_memory_training_context *ctx = &psp->mem_train_ctx; - uint32_t *pcache = (uint32_t*)ctx->sys_cache; + uint32_t *pcache = (uint32_t *)ctx->sys_cache; struct amdgpu_device *adev = psp->adev; uint32_t p2c_header[4]; uint32_t sz; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c index 75489313dbad..dff5c15b4858 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c @@ -39,6 +39,7 @@ MODULE_FIRMWARE("amdgpu/renoir_asd.bin"); MODULE_FIRMWARE("amdgpu/renoir_ta.bin"); +MODULE_FIRMWARE("amdgpu/green_sardine_asd.bin"); /* address block */ #define smnMP1_FIRMWARE_FLAGS 0x3010024 @@ -54,7 +55,10 @@ static int psp_v12_0_init_microcode(struct psp_context *psp) switch (adev->asic_type) { case CHIP_RENOIR: - chip_name = "renoir"; + if (adev->apu_flags & AMD_APU_IS_RENOIR) + chip_name = "renoir"; + else + chip_name = "green_sardine"; break; default: BUG(); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index 86fb1eddf5a6..dc74ca2aa892 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -69,6 +69,7 @@ MODULE_FIRMWARE("amdgpu/picasso_sdma.bin"); MODULE_FIRMWARE("amdgpu/raven2_sdma.bin"); MODULE_FIRMWARE("amdgpu/arcturus_sdma.bin"); MODULE_FIRMWARE("amdgpu/renoir_sdma.bin"); +MODULE_FIRMWARE("amdgpu/green_sardine_sdma.bin"); #define SDMA0_POWER_CNTL__ON_OFF_CONDITION_HOLD_TIME_MASK 0x000000F8L #define SDMA0_POWER_CNTL__ON_OFF_STATUS_DURATION_TIME_MASK 0xFC000000L @@ -568,7 +569,7 @@ static void sdma_v4_0_destroy_inst_ctx(struct amdgpu_device *adev) break; } - memset((void*)adev->sdma.instance, 0, + memset((void *)adev->sdma.instance, 0, sizeof(struct amdgpu_sdma_instance) * AMDGPU_MAX_SDMA_INSTANCES); } @@ -619,7 +620,10 @@ static int sdma_v4_0_init_microcode(struct amdgpu_device *adev) chip_name = "arcturus"; break; case CHIP_RENOIR: - chip_name = "renoir"; + if (adev->apu_flags & AMD_APU_IS_RENOIR) + chip_name = "renoir"; + else + chip_name = "green_sardine"; break; default: BUG(); @@ -639,8 +643,8 @@ static int sdma_v4_0_init_microcode(struct amdgpu_device *adev) if (adev->asic_type == CHIP_ARCTURUS) { /* Acturus will leverage the same FW memory for every SDMA instance */ - memcpy((void*)&adev->sdma.instance[i], - (void*)&adev->sdma.instance[0], + memcpy((void *)&adev->sdma.instance[i], + (void *)&adev->sdma.instance[0], sizeof(struct amdgpu_sdma_instance)); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c index 9f3952723c63..cb5a6f1437f8 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c @@ -46,6 +46,9 @@ MODULE_FIRMWARE("amdgpu/sienna_cichlid_sdma.bin"); MODULE_FIRMWARE("amdgpu/navy_flounder_sdma.bin"); +MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_sdma.bin"); + +MODULE_FIRMWARE("amdgpu/vangogh_sdma.bin"); #define SDMA1_REG_OFFSET 0x600 #define SDMA3_REG_OFFSET 0x400 @@ -87,6 +90,8 @@ static void sdma_v5_2_init_golden_registers(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: break; default: break; @@ -124,7 +129,7 @@ static void sdma_v5_2_destroy_inst_ctx(struct amdgpu_device *adev) break; } - memset((void*)adev->sdma.instance, 0, + memset((void *)adev->sdma.instance, 0, sizeof(struct amdgpu_sdma_instance) * AMDGPU_MAX_SDMA_INSTANCES); } @@ -160,6 +165,12 @@ static int sdma_v5_2_init_microcode(struct amdgpu_device *adev) case CHIP_NAVY_FLOUNDER: chip_name = "navy_flounder"; break; + case CHIP_VANGOGH: + chip_name = "vangogh"; + break; + case CHIP_DIMGREY_CAVEFISH: + chip_name = "dimgrey_cavefish"; + break; default: BUG(); } @@ -175,10 +186,10 @@ static int sdma_v5_2_init_microcode(struct amdgpu_device *adev) goto out; for (i = 1; i < adev->sdma.num_instances; i++) { - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) { - memcpy((void*)&adev->sdma.instance[i], - (void*)&adev->sdma.instance[0], + if (adev->asic_type >= CHIP_SIENNA_CICHLID && + adev->asic_type <= CHIP_DIMGREY_CAVEFISH) { + memcpy((void *)&adev->sdma.instance[i], + (void *)&adev->sdma.instance[0], sizeof(struct amdgpu_sdma_instance)); } else { snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma%d.bin", chip_name, i); @@ -696,7 +707,7 @@ static int sdma_v5_2_gfx_resume(struct amdgpu_device *adev) temp &= 0xFF0FFF; temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) | (CACHE_WRITE_POLICY_L2__DEFAULT << 14) | - 0x01000000); + SDMA0_UTCL1_PAGE__LLC_NOALLOC_MASK); WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE), temp); if (!amdgpu_sriov_vf(adev)) { @@ -1169,8 +1180,12 @@ static int sdma_v5_2_early_init(void *handle) adev->sdma.num_instances = 4; break; case CHIP_NAVY_FLOUNDER: + case CHIP_DIMGREY_CAVEFISH: adev->sdma.num_instances = 2; break; + case CHIP_VANGOGH: + adev->sdma.num_instances = 1; + break; default: break; } @@ -1567,6 +1582,8 @@ static int sdma_v5_2_set_clockgating_state(void *handle, switch (adev->asic_type) { case CHIP_SIENNA_CICHLID: case CHIP_NAVY_FLOUNDER: + case CHIP_VANGOGH: + case CHIP_DIMGREY_CAVEFISH: sdma_v5_2_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE ? true : false); sdma_v5_2_update_medium_grain_light_sleep(adev, diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c index e5e336fd9e94..3cf0589bfea5 100644 --- a/drivers/gpu/drm/amd/amdgpu/si.c +++ b/drivers/gpu/drm/amd/amdgpu/si.c @@ -1350,7 +1350,7 @@ static void si_vga_set_state(struct amdgpu_device *adev, bool state) static u32 si_get_xclk(struct amdgpu_device *adev) { - u32 reference_clock = adev->clock.spll.reference_freq; + u32 reference_clock = adev->clock.spll.reference_freq; u32 tmp; tmp = RREG32(CG_CLKPIN_CNTL_2); diff --git a/drivers/gpu/drm/amd/amdgpu/si_ih.c b/drivers/gpu/drm/amd/amdgpu/si_ih.c index 621727d7fd18..51880f6ef634 100644 --- a/drivers/gpu/drm/amd/amdgpu/si_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/si_ih.c @@ -43,7 +43,7 @@ static void si_ih_enable_interrupts(struct amdgpu_device *adev) WREG32(IH_RB_CNTL, ih_rb_cntl); adev->irq.ih.enabled = true; } - + static void si_ih_disable_interrupts(struct amdgpu_device *adev) { u32 ih_rb_cntl = RREG32(IH_RB_CNTL); diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index afcccc6c0fc6..5b79ce9e0699 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -62,6 +62,7 @@ #include "nbio_v7_0.h" #include "nbio_v7_4.h" #include "vega10_ih.h" +#include "navi10_ih.h" #include "sdma_v4_0.h" #include "uvd_v7_0.h" #include "vce_v4_0.h" @@ -734,9 +735,15 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev) else amdgpu_device_ip_block_add(adev, &psp_v3_1_ip_block); } - amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block); + if (adev->asic_type == CHIP_VEGA20) + amdgpu_device_ip_block_add(adev, &navi10_ih_ip_block); + else + amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block); } else { - amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block); + if (adev->asic_type == CHIP_VEGA20) + amdgpu_device_ip_block_add(adev, &navi10_ih_ip_block); + else + amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block); if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) { if (adev->asic_type == CHIP_VEGA20) amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block); @@ -787,9 +794,9 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) { if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block); - amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block); + amdgpu_device_ip_block_add(adev, &navi10_ih_ip_block); } else { - amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block); + amdgpu_device_ip_block_add(adev, &navi10_ih_ip_block); if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block); } @@ -822,7 +829,7 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block); #if defined(CONFIG_DRM_AMD_DC) else if (amdgpu_device_has_dc_support(adev)) - amdgpu_device_ip_block_add(adev, &dm_ip_block); + amdgpu_device_ip_block_add(adev, &dm_ip_block); #endif amdgpu_device_ip_block_add(adev, &vcn_v2_0_ip_block); amdgpu_device_ip_block_add(adev, &jpeg_v2_0_ip_block); @@ -1195,8 +1202,7 @@ static int soc15_common_early_init(void *handle) adev->pg_flags = AMD_PG_SUPPORT_SDMA | AMD_PG_SUPPORT_MMHUB | - AMD_PG_SUPPORT_VCN | - AMD_PG_SUPPORT_VCN_DPG; + AMD_PG_SUPPORT_VCN; } else { adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS | @@ -1243,7 +1249,15 @@ static int soc15_common_early_init(void *handle) break; case CHIP_RENOIR: adev->asic_funcs = &soc15_asic_funcs; - adev->apu_flags |= AMD_APU_IS_RENOIR; + if (adev->pdev->device == 0x1636) + adev->apu_flags |= AMD_APU_IS_RENOIR; + else + adev->apu_flags |= AMD_APU_IS_GREEN_SARDINE; + + if (adev->apu_flags & AMD_APU_IS_RENOIR) + adev->external_rev_id = adev->rev_id + 0x91; + else + adev->external_rev_id = adev->rev_id + 0xa1; adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS | AMD_CG_SUPPORT_GFX_3D_CGCG | @@ -1268,7 +1282,6 @@ static int soc15_common_early_init(void *handle) AMD_PG_SUPPORT_VCN | AMD_PG_SUPPORT_JPEG | AMD_PG_SUPPORT_VCN_DPG; - adev->external_rev_id = adev->rev_id + 0x91; break; default: /* FIXME: not supported yet */ diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c index 5288617ca552..96d7769609f4 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c @@ -253,7 +253,7 @@ static void umc_v6_1_querry_uncorrectable_error_count(struct amdgpu_device *adev static void umc_v6_1_query_ras_error_count(struct amdgpu_device *adev, void *ras_error_status) { - struct ras_err_data* err_data = (struct ras_err_data*)ras_error_status; + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; uint32_t umc_inst = 0; uint32_t ch_inst = 0; @@ -368,7 +368,7 @@ static void umc_v6_1_query_error_address(struct amdgpu_device *adev, static void umc_v6_1_query_ras_error_address(struct amdgpu_device *adev, void *ras_error_status) { - struct ras_err_data* err_data = (struct ras_err_data*)ras_error_status; + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; uint32_t umc_inst = 0; uint32_t ch_inst = 0; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c index 5665c77a9d58..a064c097690c 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c @@ -170,7 +170,7 @@ static void umc_v8_7_querry_uncorrectable_error_count(struct amdgpu_device *adev static void umc_v8_7_query_ras_error_count(struct amdgpu_device *adev, void *ras_error_status) { - struct ras_err_data* err_data = (struct ras_err_data*)ras_error_status; + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; uint32_t umc_inst = 0; uint32_t ch_inst = 0; @@ -260,7 +260,7 @@ static void umc_v8_7_query_error_address(struct amdgpu_device *adev, static void umc_v8_7_query_ras_error_address(struct amdgpu_device *adev, void *ras_error_status) { - struct ras_err_data* err_data = (struct ras_err_data*)ras_error_status; + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; uint32_t umc_inst = 0; uint32_t ch_inst = 0; diff --git a/drivers/gpu/drm/amd/amdgpu/vangogh_reg_init.c b/drivers/gpu/drm/amd/amdgpu/vangogh_reg_init.c new file mode 100644 index 000000000000..d64d681a05dc --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/vangogh_reg_init.c @@ -0,0 +1,50 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "nv.h" + +#include "soc15_common.h" +#include "soc15_hw_ip.h" +#include "vangogh_ip_offset.h" + +void vangogh_reg_base_init(struct amdgpu_device *adev) +{ + /* HW has more IP blocks, only initialized the blocke needed by driver */ + uint32_t i; + for (i = 0 ; i < MAX_INSTANCE ; ++i) { + adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i])); + adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i])); + adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i])); + adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i])); + adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i])); + adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i])); + adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i])); + adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(VCN_BASE.instance[i])); + adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i])); + adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DCN_BASE.instance[i])); + adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i])); + adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i])); + adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i])); + adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i])); + } +} diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c index e5d29dee0c88..e285f9c9d460 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c @@ -45,7 +45,7 @@ #define mmUVD_SCRATCH9_INTERNAL_OFFSET 0xc01d #define mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET 0x1e1 -#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x5a6 +#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x5a6 #define mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET 0x5a7 #define mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET 0x1e2 @@ -800,7 +800,7 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect) WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, tmp); if (indirect) - adev->vcn.inst->dpg_sram_curr_addr = (uint32_t*)adev->vcn.inst->dpg_sram_cpu_addr; + adev->vcn.inst->dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst->dpg_sram_cpu_addr; /* enable clock gating */ vcn_v2_0_clock_gating_dpg_mode(adev, 0, indirect); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index 0f1d3ef8baa7..b6e0f4ba6272 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -45,7 +45,7 @@ #define mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET 0x431 #define mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET 0x3b4 -#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x3b5 +#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x3b5 #define mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET 0x25c #define VCN25_MAX_HW_INSTANCES_ARCTURUS 2 @@ -777,7 +777,7 @@ static int vcn_v2_5_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, boo WREG32_SOC15(VCN, inst_idx, mmUVD_POWER_STATUS, tmp); if (indirect) - adev->vcn.inst[inst_idx].dpg_sram_curr_addr = (uint32_t*)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr; + adev->vcn.inst[inst_idx].dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr; /* enable clock gating */ vcn_v2_5_clock_gating_dpg_mode(adev, 0, inst_idx, indirect); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c index e074f7ed388c..269002816109 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c @@ -44,10 +44,10 @@ #define mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET 0x431 #define mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET 0x3b4 -#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x3b5 +#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x3b5 #define mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET 0x25c -#define VCN_INSTANCES_SIENNA_CICHLID 2 +#define VCN_INSTANCES_SIENNA_CICHLID 2 static int amdgpu_ih_clientid_vcns[] = { SOC15_IH_CLIENTID_VCN, @@ -55,8 +55,8 @@ static int amdgpu_ih_clientid_vcns[] = { }; static int amdgpu_ucode_id_vcns[] = { - AMDGPU_UCODE_ID_VCN, - AMDGPU_UCODE_ID_VCN1 + AMDGPU_UCODE_ID_VCN, + AMDGPU_UCODE_ID_VCN1 }; static int vcn_v3_0_start_sriov(struct amdgpu_device *adev); @@ -915,7 +915,7 @@ static int vcn_v3_0_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, boo WREG32_SOC15(VCN, inst_idx, mmUVD_POWER_STATUS, tmp); if (indirect) - adev->vcn.inst[inst_idx].dpg_sram_curr_addr = (uint32_t*)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr; + adev->vcn.inst[inst_idx].dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr; /* enable clock gating */ vcn_v3_0_clock_gating_dpg_mode(adev, 0, inst_idx, indirect); @@ -1371,7 +1371,7 @@ static int vcn_v3_0_start_sriov(struct amdgpu_device *adev) } /* Update init table header in memory */ - size = sizeof(struct mmsch_v3_0_init_header); + size = sizeof(struct mmsch_v3_0_init_header); table_loc = (uint32_t *)table->cpu_addr; memcpy((void *)table_loc, &header, size); |