diff options
| author | Maxime Ripard <maxime@cerno.tech> | 2022-04-05 12:06:58 +0300 |
|---|---|---|
| committer | Maxime Ripard <maxime@cerno.tech> | 2022-04-05 12:06:58 +0300 |
| commit | 9cbbd694a58bdf24def2462276514c90cab7cf80 (patch) | |
| tree | 98a504890134d34631a6a0ecbce94d3f1ecc21fc /drivers/gpu/drm/amd/amdgpu | |
| parent | 71d637823cac7748079a912e0373476c7cf6f985 (diff) | |
| parent | 3123109284176b1532874591f7c81f3837bbdc17 (diff) | |
| download | linux-9cbbd694a58bdf24def2462276514c90cab7cf80.tar.xz | |
Merge drm/drm-next into drm-misc-next
Let's start the 5.19 development cycle.
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
154 files changed, 4601 insertions, 3234 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 7fedbb725e17..40e2c6e2df79 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -46,18 +46,18 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \ atom.o amdgpu_fence.o amdgpu_ttm.o amdgpu_object.o amdgpu_gart.o \ amdgpu_encoders.o amdgpu_display.o amdgpu_i2c.o \ amdgpu_gem.o amdgpu_ring.o \ - amdgpu_cs.o amdgpu_bios.o amdgpu_benchmark.o amdgpu_test.o \ + amdgpu_cs.o amdgpu_bios.o amdgpu_benchmark.o \ atombios_dp.o amdgpu_afmt.o amdgpu_trace_points.o \ atombios_encoders.o amdgpu_sa.o atombios_i2c.o \ amdgpu_dma_buf.o amdgpu_vm.o amdgpu_ib.o amdgpu_pll.o \ amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o \ amdgpu_gtt_mgr.o amdgpu_preempt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o \ amdgpu_atomfirmware.o amdgpu_vf_error.o amdgpu_sched.o \ - amdgpu_debugfs.o amdgpu_ids.o amdgpu_gmc.o amdgpu_mmhub.o \ + amdgpu_debugfs.o amdgpu_ids.o amdgpu_gmc.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_fw_attestation.o amdgpu_securedisplay.o amdgpu_hdp.o \ + amdgpu_fw_attestation.o amdgpu_securedisplay.o \ amdgpu_eeprom.o amdgpu_mca.o amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran.c b/drivers/gpu/drm/amd/amdgpu/aldebaran.c index bcfdb63b1d42..c6cc493a5486 100644 --- a/drivers/gpu/drm/amd/amdgpu/aldebaran.c +++ b/drivers/gpu/drm/amd/amdgpu/aldebaran.c @@ -31,6 +31,17 @@ #include "amdgpu_psp.h" #include "amdgpu_xgmi.h" +static bool aldebaran_is_mode2_default(struct amdgpu_reset_control *reset_ctl) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle; + + if ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2) && + adev->gmc.xgmi.connected_to_cpu)) + return true; + + return false; +} + static struct amdgpu_reset_handler * aldebaran_get_reset_handler(struct amdgpu_reset_control *reset_ctl, struct amdgpu_reset_context *reset_context) @@ -48,7 +59,7 @@ aldebaran_get_reset_handler(struct amdgpu_reset_control *reset_ctl, } } - if (adev->gmc.xgmi.connected_to_cpu) { + if (aldebaran_is_mode2_default(reset_ctl)) { list_for_each_entry(handler, &reset_ctl->reset_handlers, handler_list) { if (handler->reset_method == AMD_RESET_METHOD_MODE2) { @@ -136,18 +147,31 @@ static int aldebaran_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl, struct amdgpu_reset_context *reset_context) { - struct amdgpu_device *tmp_adev = NULL; struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle; + struct amdgpu_device *tmp_adev = NULL; + struct list_head reset_device_list; int r = 0; dev_dbg(adev->dev, "aldebaran perform hw reset\n"); - if (reset_context->hive == NULL) { + if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2) && + reset_context->hive == NULL) { /* Wrong context, return error */ return -EINVAL; } - list_for_each_entry(tmp_adev, &reset_context->hive->device_list, - gmc.xgmi.head) { + INIT_LIST_HEAD(&reset_device_list); + if (reset_context->hive) { + list_for_each_entry (tmp_adev, + &reset_context->hive->device_list, + gmc.xgmi.head) + list_add_tail(&tmp_adev->reset_list, + &reset_device_list); + } else { + list_add_tail(&reset_context->reset_req_dev->reset_list, + &reset_device_list); + } + + list_for_each_entry (tmp_adev, &reset_device_list, reset_list) { mutex_lock(&tmp_adev->reset_cntl->reset_lock); tmp_adev->reset_cntl->active_reset = AMD_RESET_METHOD_MODE2; } @@ -155,8 +179,7 @@ aldebaran_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl, * Mode2 reset doesn't need any sync between nodes in XGMI hive, instead launch * them together so that they can be completed asynchronously on multiple nodes */ - list_for_each_entry(tmp_adev, &reset_context->hive->device_list, - gmc.xgmi.head) { + list_for_each_entry (tmp_adev, &reset_device_list, reset_list) { /* For XGMI run all resets in parallel to speed up the process */ if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { if (!queue_work(system_unbound_wq, @@ -174,9 +197,7 @@ aldebaran_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl, /* For XGMI wait for all resets to complete before proceed */ if (!r) { - list_for_each_entry(tmp_adev, - &reset_context->hive->device_list, - gmc.xgmi.head) { + list_for_each_entry (tmp_adev, &reset_device_list, reset_list) { if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { flush_work(&tmp_adev->reset_cntl->reset_work); r = tmp_adev->asic_reset_res; @@ -186,8 +207,7 @@ aldebaran_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl, } } - list_for_each_entry(tmp_adev, &reset_context->hive->device_list, - gmc.xgmi.head) { + list_for_each_entry (tmp_adev, &reset_device_list, reset_list) { mutex_unlock(&tmp_adev->reset_cntl->reset_lock); tmp_adev->reset_cntl->active_reset = AMD_RESET_METHOD_NONE; } @@ -260,7 +280,7 @@ static int aldebaran_mode2_restore_ip(struct amdgpu_device *adev) adev->gfx.rlc.funcs->resume(adev); /* Wait for FW reset event complete */ - r = smu_wait_for_event(adev, SMU_EVENT_RESET_COMPLETE, 0); + r = amdgpu_dpm_wait_for_event(adev, SMU_EVENT_RESET_COMPLETE, 0); if (r) { dev_err(adev->dev, "Failed to get response from firmware after reset\n"); @@ -319,16 +339,30 @@ static int aldebaran_mode2_restore_hwcontext(struct amdgpu_reset_control *reset_ctl, struct amdgpu_reset_context *reset_context) { - int r; struct amdgpu_device *tmp_adev = NULL; + struct list_head reset_device_list; + int r; - if (reset_context->hive == NULL) { + if (reset_context->reset_req_dev->ip_versions[MP1_HWIP][0] == + IP_VERSION(13, 0, 2) && + reset_context->hive == NULL) { /* Wrong context, return error */ return -EINVAL; } - list_for_each_entry(tmp_adev, &reset_context->hive->device_list, - gmc.xgmi.head) { + INIT_LIST_HEAD(&reset_device_list); + if (reset_context->hive) { + list_for_each_entry (tmp_adev, + &reset_context->hive->device_list, + gmc.xgmi.head) + list_add_tail(&tmp_adev->reset_list, + &reset_device_list); + } else { + list_add_tail(&reset_context->reset_req_dev->reset_list, + &reset_device_list); + } + + list_for_each_entry (tmp_adev, &reset_device_list, reset_list) { dev_info(tmp_adev->dev, "GPU reset succeeded, trying to resume\n"); r = aldebaran_mode2_restore_ip(tmp_adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index b89406b01694..cdf0818088b3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -60,7 +60,6 @@ #include <drm/amdgpu_drm.h> #include <drm/drm_gem.h> #include <drm/drm_ioctl.h> -#include <drm/gpu_scheduler.h> #include <kgd_kfd_interface.h> #include "dm_pp_interface.h" @@ -99,7 +98,6 @@ #include "amdgpu_gem.h" #include "amdgpu_doorbell.h" #include "amdgpu_amdkfd.h" -#include "amdgpu_smu.h" #include "amdgpu_discovery.h" #include "amdgpu_mes.h" #include "amdgpu_umc.h" @@ -109,6 +107,7 @@ #include "amdgpu_smuio.h" #include "amdgpu_fdinfo.h" #include "amdgpu_mca.h" +#include "amdgpu_ras.h" #define MAX_GPU_INSTANCE 16 @@ -155,8 +154,6 @@ extern int amdgpu_vis_vram_limit; extern int amdgpu_gart_size; extern int amdgpu_gtt_size; extern int amdgpu_moverate; -extern int amdgpu_benchmarking; -extern int amdgpu_testing; extern int amdgpu_audio; extern int amdgpu_disp_priority; extern int amdgpu_hw_i2c; @@ -197,7 +194,6 @@ extern int amdgpu_emu_mode; extern uint amdgpu_smu_memory_pool_size; extern int amdgpu_smu_pptable_id; extern uint amdgpu_dc_feature_mask; -extern uint amdgpu_freesync_vid_mode; extern uint amdgpu_dc_debug_mask; extern uint amdgpu_dm_abm_level; extern int amdgpu_backlight; @@ -214,6 +210,7 @@ extern int amdgpu_mes; extern int amdgpu_noretry; extern int amdgpu_force_asic_type; extern int amdgpu_smartshift_bias; +extern int amdgpu_use_xgmi_p2p; #ifdef CONFIG_HSA_AMD extern int sched_policy; extern bool debug_evictions; @@ -235,6 +232,9 @@ extern int amdgpu_cik_support; #endif extern int amdgpu_num_kcq; +#define AMDGPU_VCNFW_LOG_SIZE (32 * 1024) +extern int amdgpu_vcnfw_log; + #define AMDGPU_VM_MAX_NUM_CTX 4096 #define AMDGPU_SG_THRESHOLD (256*1024*1024) #define AMDGPU_DEFAULT_GTT_SIZE_MB 3072ULL /* 3GB by default */ @@ -276,9 +276,6 @@ extern int amdgpu_num_kcq; #define AMDGPU_SMARTSHIFT_MIN_BIAS (-100) struct amdgpu_device; -struct amdgpu_ib; -struct amdgpu_cs_parser; -struct amdgpu_job; struct amdgpu_irq_src; struct amdgpu_fpriv; struct amdgpu_bo_va_mapping; @@ -373,7 +370,8 @@ int amdgpu_device_ip_block_add(struct amdgpu_device *adev, */ bool amdgpu_get_bios(struct amdgpu_device *adev); bool amdgpu_read_bios(struct amdgpu_device *adev); - +bool amdgpu_soc15_read_bios_from_rom(struct amdgpu_device *adev, + u8 *bios, u32 length_bytes); /* * Clocks */ @@ -466,20 +464,6 @@ struct amdgpu_flip_work { /* - * CP & rings. - */ - -struct amdgpu_ib { - struct amdgpu_sa_bo *sa_bo; - uint32_t length_dw; - uint64_t gpu_addr; - uint32_t *ptr; - uint32_t flags; -}; - -extern const struct drm_sched_backend_ops amdgpu_sched_ops; - -/* * file private structure */ @@ -494,79 +478,6 @@ struct amdgpu_fpriv { int amdgpu_file_to_fpriv(struct file *filp, struct amdgpu_fpriv **fpriv); -int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm, - unsigned size, - enum amdgpu_ib_pool_type pool, - struct amdgpu_ib *ib); -void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib, - struct dma_fence *f); -int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, - struct amdgpu_ib *ibs, struct amdgpu_job *job, - struct dma_fence **f); -int amdgpu_ib_pool_init(struct amdgpu_device *adev); -void amdgpu_ib_pool_fini(struct amdgpu_device *adev); -int amdgpu_ib_ring_tests(struct amdgpu_device *adev); - -/* - * CS. - */ -struct amdgpu_cs_chunk { - uint32_t chunk_id; - uint32_t length_dw; - void *kdata; -}; - -struct amdgpu_cs_post_dep { - struct drm_syncobj *syncobj; - struct dma_fence_chain *chain; - u64 point; -}; - -struct amdgpu_cs_parser { - struct amdgpu_device *adev; - struct drm_file *filp; - struct amdgpu_ctx *ctx; - - /* chunks */ - unsigned nchunks; - struct amdgpu_cs_chunk *chunks; - - /* scheduler job object */ - struct amdgpu_job *job; - struct drm_sched_entity *entity; - - /* buffer objects */ - struct ww_acquire_ctx ticket; - struct amdgpu_bo_list *bo_list; - struct amdgpu_mn *mn; - struct amdgpu_bo_list_entry vm_pd; - struct list_head validated; - struct dma_fence *fence; - uint64_t bytes_moved_threshold; - uint64_t bytes_moved_vis_threshold; - uint64_t bytes_moved; - uint64_t bytes_moved_vis; - - /* user fence */ - struct amdgpu_bo_list_entry uf_entry; - - unsigned num_post_deps; - struct amdgpu_cs_post_dep *post_deps; -}; - -static inline u32 amdgpu_get_ib_value(struct amdgpu_cs_parser *p, - uint32_t ib_idx, int idx) -{ - return p->job->ibs[ib_idx].ptr[idx]; -} - -static inline void amdgpu_set_ib_value(struct amdgpu_cs_parser *p, - uint32_t ib_idx, int idx, - uint32_t value) -{ - p->job->ibs[ib_idx].ptr[idx] = value; -} - /* * Writeback */ @@ -586,13 +497,7 @@ void amdgpu_device_wb_free(struct amdgpu_device *adev, u32 wb); /* * Benchmarking */ -void amdgpu_benchmark(struct amdgpu_device *adev, int test_number); - - -/* - * Testing - */ -void amdgpu_test_moves(struct amdgpu_device *adev); +int amdgpu_benchmark(struct amdgpu_device *adev, int test_number); /* * ASIC specific register table accessible by UMD @@ -771,6 +676,8 @@ struct amd_powerplay { const struct amd_pm_funcs *pp_funcs; }; +struct ip_discovery_top; + /* polaris10 kickers */ #define ASICID_IS_P20(did, rid) (((did == 0x67DF) && \ ((rid == 0xE3) || \ @@ -952,12 +859,6 @@ struct amdgpu_device { /* powerplay */ struct amd_powerplay powerplay; - bool pp_force_state_enabled; - - /* smu */ - struct smu_context smu; - - /* dpm */ struct amdgpu_pm pm; u32 cg_flags; u32 pg_flags; @@ -1100,7 +1001,18 @@ struct amdgpu_device { uint32_t ip_versions[MAX_HWIP][HWIP_MAX_INSTANCE]; bool ram_is_direct_mapped; + + struct list_head ras_list; + + struct ip_discovery_top *ip_top; + struct amdgpu_reset_domain *reset_domain; + + struct mutex benchmark_mutex; + + /* reset dump register */ + uint32_t *reset_dump_reg_list; + int num_regs; }; static inline struct amdgpu_device *drm_to_adev(struct drm_device *ddev) @@ -1299,6 +1211,7 @@ int amdgpu_device_gpu_recover_imp(struct amdgpu_device *adev, void amdgpu_device_pci_config_reset(struct amdgpu_device *adev); int amdgpu_device_pci_reset(struct amdgpu_device *adev); bool amdgpu_device_need_post(struct amdgpu_device *adev); +bool amdgpu_device_should_use_aspm(struct amdgpu_device *adev); void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes, u64 num_vis_bytes); @@ -1324,6 +1237,10 @@ void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring); void amdgpu_device_halt(struct amdgpu_device *adev); +u32 amdgpu_device_pcie_port_rreg(struct amdgpu_device *adev, + u32 reg); +void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev, + u32 reg, u32 v); /* atpx handler */ #if defined(CONFIG_VGA_SWITCHEROO) @@ -1411,12 +1328,10 @@ int amdgpu_acpi_smart_shift_update(struct drm_device *dev, enum amdgpu_ss ss_sta int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev); void amdgpu_acpi_get_backlight_caps(struct amdgpu_dm_backlight_caps *caps); -bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev); void amdgpu_acpi_detect(void); #else static inline int amdgpu_acpi_init(struct amdgpu_device *adev) { return 0; } static inline void amdgpu_acpi_fini(struct amdgpu_device *adev) { } -static inline bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) { return false; } static inline void amdgpu_acpi_detect(void) { } static inline bool amdgpu_acpi_is_power_shift_control_supported(void) { return false; } static inline int amdgpu_acpi_power_shift_control(struct amdgpu_device *adev, @@ -1425,9 +1340,13 @@ static inline int amdgpu_acpi_smart_shift_update(struct drm_device *dev, enum amdgpu_ss ss_state) { return 0; } #endif -int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser, - uint64_t addr, struct amdgpu_bo **bo, - struct amdgpu_bo_va_mapping **mapping); +#if defined(CONFIG_ACPI) && defined(CONFIG_SUSPEND) +bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev); +bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev); +#else +static inline bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) { return false; } +static inline bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev) { return false; } +#endif #if defined(CONFIG_DRM_AMD_DC) int amdgpu_dm_display_resume(struct amdgpu_device *adev ); @@ -1455,6 +1374,15 @@ int amdgpu_device_set_cg_state(struct amdgpu_device *adev, int amdgpu_device_set_pg_state(struct amdgpu_device *adev, enum amd_powergating_state state); +static inline bool amdgpu_device_has_timeouts_enabled(struct amdgpu_device *adev) +{ + return amdgpu_gpu_recovery != 0 && + adev->gfx_timeout != MAX_SCHEDULE_TIMEOUT && + adev->compute_timeout != MAX_SCHEDULE_TIMEOUT && + adev->sdma_timeout != MAX_SCHEDULE_TIMEOUT && + adev->video_timeout != MAX_SCHEDULE_TIMEOUT; +} + #include "amdgpu_object.h" static inline bool amdgpu_is_tmz(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index 4811b0faafd9..0e12315fa0cb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -1031,6 +1031,20 @@ void amdgpu_acpi_detect(void) } } +#if IS_ENABLED(CONFIG_SUSPEND) +/** + * amdgpu_acpi_is_s3_active + * + * @adev: amdgpu_device_pointer + * + * returns true if supported, false if not. + */ +bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev) +{ + return !(adev->flags & AMD_IS_APU) || + (pm_suspend_target_state == PM_SUSPEND_MEM); +} + /** * amdgpu_acpi_is_s0ix_active * @@ -1040,11 +1054,24 @@ void amdgpu_acpi_detect(void) */ bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) { -#if IS_ENABLED(CONFIG_AMD_PMC) && IS_ENABLED(CONFIG_SUSPEND) - if (acpi_gbl_FADT.flags & ACPI_FADT_LOW_POWER_S0) { - if (adev->flags & AMD_IS_APU) - return pm_suspend_target_state == PM_SUSPEND_TO_IDLE; + if (!(adev->flags & AMD_IS_APU) || + (pm_suspend_target_state != PM_SUSPEND_TO_IDLE)) + return false; + + if (!(acpi_gbl_FADT.flags & ACPI_FADT_LOW_POWER_S0)) { + dev_warn_once(adev->dev, + "Power consumption will be higher as BIOS has not been configured for suspend-to-idle.\n" + "To use suspend-to-idle change the sleep mode in BIOS setup.\n"); + return false; } -#endif + +#if !IS_ENABLED(CONFIG_AMD_PMC) + dev_warn_once(adev->dev, + "Power consumption will be higher as the kernel has not been compiled with CONFIG_AMD_PMC.\n"); return false; +#else + return true; +#endif /* CONFIG_AMD_PMC */ } + +#endif /* CONFIG_SUSPEND */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index ac841ae8f5cc..4cb14c2fe53f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -131,6 +131,7 @@ struct amdkfd_process_info { atomic_t evicted_bos; struct delayed_work restore_userptr_work; struct pid *pid; + bool block_mmu_notifications; }; int amdgpu_amdkfd_init(void); @@ -268,7 +269,7 @@ uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *drm_priv); int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu( struct amdgpu_device *adev, uint64_t va, uint64_t size, void *drm_priv, struct kgd_mem **mem, - uint64_t *offset, uint32_t flags); + uint64_t *offset, uint32_t flags, bool criu_resume); int amdgpu_amdkfd_gpuvm_free_memory_of_gpu( struct amdgpu_device *adev, struct kgd_mem *mem, void *drm_priv, uint64_t *size); @@ -297,6 +298,10 @@ int amdgpu_amdkfd_get_tile_config(struct amdgpu_device *adev, struct tile_config *config); void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev, bool reset); +bool amdgpu_amdkfd_bo_mapped_to_dev(struct amdgpu_device *adev, struct kgd_mem *mem); +void amdgpu_amdkfd_block_mmu_notifications(void *p); +int amdgpu_amdkfd_criu_resume(void *p); + #if IS_ENABLED(CONFIG_HSA_AMD) void amdgpu_amdkfd_gpuvm_init_mem_limits(void); void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c index 46cd4ee6bafb..c8935d718207 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c @@ -37,10 +37,7 @@ const struct kfd2kgd_calls aldebaran_kfd2kgd = { .hqd_sdma_is_occupied = kgd_arcturus_hqd_sdma_is_occupied, .hqd_destroy = kgd_gfx_v9_hqd_destroy, .hqd_sdma_destroy = kgd_arcturus_hqd_sdma_destroy, - .address_watch_disable = kgd_gfx_v9_address_watch_disable, - .address_watch_execute = kgd_gfx_v9_address_watch_execute, .wave_control_execute = kgd_gfx_v9_wave_control_execute, - .address_watch_get_offset = kgd_gfx_v9_address_watch_get_offset, .get_atc_vmid_pasid_mapping_info = kgd_gfx_v9_get_atc_vmid_pasid_mapping_info, .set_vm_context_page_table_base = kgd_gfx_v9_set_vm_context_page_table_base, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c index abe93b3ff765..4191af5a3f13 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c @@ -289,10 +289,7 @@ const struct kfd2kgd_calls arcturus_kfd2kgd = { .hqd_sdma_is_occupied = kgd_arcturus_hqd_sdma_is_occupied, .hqd_destroy = kgd_gfx_v9_hqd_destroy, .hqd_sdma_destroy = kgd_arcturus_hqd_sdma_destroy, - .address_watch_disable = kgd_gfx_v9_address_watch_disable, - .address_watch_execute = kgd_gfx_v9_address_watch_execute, .wave_control_execute = kgd_gfx_v9_wave_control_execute, - .address_watch_get_offset = kgd_gfx_v9_address_watch_get_offset, .get_atc_vmid_pasid_mapping_info = kgd_gfx_v9_get_atc_vmid_pasid_mapping_info, .set_vm_context_page_table_base = diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c index 7b7f4b2764c1..9378fc79e9ea 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c @@ -671,20 +671,6 @@ static bool get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev, return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK); } -static int kgd_address_watch_disable(struct amdgpu_device *adev) -{ - return 0; -} - -static int kgd_address_watch_execute(struct amdgpu_device *adev, - unsigned int watch_point_id, - uint32_t cntl_val, - uint32_t addr_hi, - uint32_t addr_lo) -{ - return 0; -} - static int kgd_wave_control_execute(struct amdgpu_device *adev, uint32_t gfx_index_val, uint32_t sq_cmd) @@ -709,13 +695,6 @@ static int kgd_wave_control_execute(struct amdgpu_device *adev, return 0; } -static uint32_t kgd_address_watch_get_offset(struct amdgpu_device *adev, - unsigned int watch_point_id, - unsigned int reg_offset) -{ - return 0; -} - static void set_vm_context_page_table_base(struct amdgpu_device *adev, uint32_t vmid, uint64_t page_table_base) { @@ -767,10 +746,7 @@ const struct kfd2kgd_calls gfx_v10_kfd2kgd = { .hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied, .hqd_destroy = kgd_hqd_destroy, .hqd_sdma_destroy = kgd_hqd_sdma_destroy, - .address_watch_disable = kgd_address_watch_disable, - .address_watch_execute = kgd_address_watch_execute, .wave_control_execute = kgd_wave_control_execute, - .address_watch_get_offset = kgd_address_watch_get_offset, .get_atc_vmid_pasid_mapping_info = get_atc_vmid_pasid_mapping_info, .set_vm_context_page_table_base = set_vm_context_page_table_base, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c index 1f37d3574001..ba21ec6b35e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c @@ -26,6 +26,8 @@ #include "gc/gc_10_3_0_sh_mask.h" #include "oss/osssys_5_0_0_offset.h" #include "oss/osssys_5_0_0_sh_mask.h" +#include "athub/athub_2_1_0_offset.h" +#include "athub/athub_2_1_0_sh_mask.h" #include "soc15_common.h" #include "v10_structs.h" #include "nv.h" @@ -582,21 +584,6 @@ static int hqd_sdma_destroy_v10_3(struct amdgpu_device *adev, void *mqd, return 0; } - -static int address_watch_disable_v10_3(struct amdgpu_device *adev) -{ - return 0; -} - -static int address_watch_execute_v10_3(struct amdgpu_device *adev, - unsigned int watch_point_id, - uint32_t cntl_val, - uint32_t addr_hi, - uint32_t addr_lo) -{ - return 0; -} - static int wave_control_execute_v10_3(struct amdgpu_device *adev, uint32_t gfx_index_val, uint32_t sq_cmd) @@ -621,11 +608,16 @@ static int wave_control_execute_v10_3(struct amdgpu_device *adev, return 0; } -static uint32_t address_watch_get_offset_v10_3(struct amdgpu_device *adev, - unsigned int watch_point_id, - unsigned int reg_offset) +static bool get_atc_vmid_pasid_mapping_info_v10_3(struct amdgpu_device *adev, + uint8_t vmid, uint16_t *p_pasid) { - return 0; + uint32_t value; + + value = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING) + + vmid); + *p_pasid = value & ATC_VMID0_PASID_MAPPING__PASID_MASK; + + return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK); } static void set_vm_context_page_table_base_v10_3(struct amdgpu_device *adev, @@ -809,11 +801,8 @@ const struct kfd2kgd_calls gfx_v10_3_kfd2kgd = { .hqd_sdma_is_occupied = hqd_sdma_is_occupied_v10_3, .hqd_destroy = hqd_destroy_v10_3, .hqd_sdma_destroy = hqd_sdma_destroy_v10_3, - .address_watch_disable = address_watch_disable_v10_3, - .address_watch_execute = address_watch_execute_v10_3, .wave_control_execute = wave_control_execute_v10_3, - .address_watch_get_offset = address_watch_get_offset_v10_3, - .get_atc_vmid_pasid_mapping_info = NULL, + .get_atc_vmid_pasid_mapping_info = get_atc_vmid_pasid_mapping_info_v10_3, .set_vm_context_page_table_base = set_vm_context_page_table_base_v10_3, .program_trap_handler_settings = program_trap_handler_settings_v10_3, #if 0 diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c index 36528dad7684..65552bb7d2f2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c @@ -45,43 +45,6 @@ enum { MAX_WATCH_ADDRESSES = 4 }; -enum { - ADDRESS_WATCH_REG_ADDR_HI = 0, - ADDRESS_WATCH_REG_ADDR_LO, - ADDRESS_WATCH_REG_CNTL, - ADDRESS_WATCH_REG_MAX -}; - -/* not defined in the CI/KV reg file */ -enum { - ADDRESS_WATCH_REG_CNTL_ATC_BIT = 0x10000000UL, - ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK = 0x00FFFFFF, - ADDRESS_WATCH_REG_ADDLOW_MASK_EXTENSION = 0x03000000, - /* extend the mask to 26 bits to match the low address field */ - ADDRESS_WATCH_REG_ADDLOW_SHIFT = 6, - ADDRESS_WATCH_REG_ADDHIGH_MASK = 0xFFFF -}; - -static const uint32_t watchRegs[MAX_WATCH_ADDRESSES * ADDRESS_WATCH_REG_MAX] = { - mmTCP_WATCH0_ADDR_H, mmTCP_WATCH0_ADDR_L, mmTCP_WATCH0_CNTL, - mmTCP_WATCH1_ADDR_H, mmTCP_WATCH1_ADDR_L, mmTCP_WATCH1_CNTL, - mmTCP_WATCH2_ADDR_H, mmTCP_WATCH2_ADDR_L, mmTCP_WATCH2_CNTL, - mmTCP_WATCH3_ADDR_H, mmTCP_WATCH3_ADDR_L, mmTCP_WATCH3_CNTL -}; - -union TCP_WATCH_CNTL_BITS { - struct { - uint32_t mask:24; - uint32_t vmid:4; - uint32_t atc:1; - uint32_t mode:2; - uint32_t valid:1; - } bitfields, bits; - uint32_t u32All; - signed int i32All; - float f32All; -}; - static void lock_srbm(struct amdgpu_device *adev, uint32_t mec, uint32_t pipe, uint32_t queue, uint32_t vmid) { @@ -529,55 +492,6 @@ static int kgd_hqd_sdma_destroy(struct amdgpu_device *adev, void *mqd, return 0; } -static int kgd_address_watch_disable(struct amdgpu_device *adev) -{ - union TCP_WATCH_CNTL_BITS cntl; - unsigned int i; - - cntl.u32All = 0; - - cntl.bitfields.valid = 0; - cntl.bitfields.mask = ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK; - cntl.bitfields.atc = 1; - - /* Turning off this address until we set all the registers */ - for (i = 0; i < MAX_WATCH_ADDRESSES; i++) - WREG32(watchRegs[i * ADDRESS_WATCH_REG_MAX + - ADDRESS_WATCH_REG_CNTL], cntl.u32All); - - return 0; -} - -static int kgd_address_watch_execute(struct amdgpu_device *adev, - unsigned int watch_point_id, - uint32_t cntl_val, - uint32_t addr_hi, - uint32_t addr_lo) -{ - union TCP_WATCH_CNTL_BITS cntl; - - cntl.u32All = cntl_val; - - /* Turning off this watch point until we set all the registers */ - cntl.bitfields.valid = 0; - WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX + - ADDRESS_WATCH_REG_CNTL], cntl.u32All); - - WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX + - ADDRESS_WATCH_REG_ADDR_HI], addr_hi); - - WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX + - ADDRESS_WATCH_REG_ADDR_LO], addr_lo); - - /* Enable the watch point */ - cntl.bitfields.valid = 1; - - WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX + - ADDRESS_WATCH_REG_CNTL], cntl.u32All); - - return 0; -} - static int kgd_wave_control_execute(struct amdgpu_device *adev, uint32_t gfx_index_val, uint32_t sq_cmd) @@ -602,13 +516,6 @@ static int kgd_wave_control_execute(struct amdgpu_device *adev, return 0; } -static uint32_t kgd_address_watch_get_offset(struct amdgpu_device *adev, - unsigned int watch_point_id, - unsigned int reg_offset) -{ - return watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX + reg_offset]; -} - static bool get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev, uint8_t vmid, uint16_t *p_pasid) { @@ -665,10 +572,7 @@ const struct kfd2kgd_calls gfx_v7_kfd2kgd = { .hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied, .hqd_destroy = kgd_hqd_destroy, .hqd_sdma_destroy = kgd_hqd_sdma_destroy, - .address_watch_disable = kgd_address_watch_disable, - .address_watch_execute = kgd_address_watch_execute, .wave_control_execute = kgd_wave_control_execute, - .address_watch_get_offset = kgd_address_watch_get_offset, .get_atc_vmid_pasid_mapping_info = get_atc_vmid_pasid_mapping_info, .set_scratch_backing_va = set_scratch_backing_va, .set_vm_context_page_table_base = set_vm_context_page_table_base, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c index 52832cd69a93..9dc5f2a0cc07 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c @@ -538,20 +538,6 @@ static bool get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev, return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK); } -static int kgd_address_watch_disable(struct amdgpu_device *adev) -{ - return 0; -} - -static int kgd_address_watch_execute(struct amdgpu_device *adev, - unsigned int watch_point_id, - uint32_t cntl_val, - uint32_t addr_hi, - uint32_t addr_lo) -{ - return 0; -} - static int kgd_wave_control_execute(struct amdgpu_device *adev, uint32_t gfx_index_val, uint32_t sq_cmd) @@ -576,13 +562,6 @@ static int kgd_wave_control_execute(struct amdgpu_device *adev, return 0; } -static uint32_t kgd_address_watch_get_offset(struct amdgpu_device *adev, - unsigned int watch_point_id, - unsigned int reg_offset) -{ - return 0; -} - static void set_scratch_backing_va(struct amdgpu_device *adev, uint64_t va, uint32_t vmid) { @@ -614,10 +593,7 @@ const struct kfd2kgd_calls gfx_v8_kfd2kgd = { .hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied, .hqd_destroy = kgd_hqd_destroy, .hqd_sdma_destroy = kgd_hqd_sdma_destroy, - .address_watch_disable = kgd_address_watch_disable, - .address_watch_execute = kgd_address_watch_execute, .wave_control_execute = kgd_wave_control_execute, - .address_watch_get_offset = kgd_address_watch_get_offset, .get_atc_vmid_pasid_mapping_info = get_atc_vmid_pasid_mapping_info, .set_scratch_backing_va = set_scratch_backing_va, 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 1abf662a0e91..81e3b528bbc9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c @@ -622,20 +622,6 @@ bool kgd_gfx_v9_get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev, return !!(value & ATC_VMID0_PASID_MAPPING__VALID_MASK); } -int kgd_gfx_v9_address_watch_disable(struct amdgpu_device *adev) -{ - return 0; -} - -int kgd_gfx_v9_address_watch_execute(struct amdgpu_device *adev, - unsigned int watch_point_id, - uint32_t cntl_val, - uint32_t addr_hi, - uint32_t addr_lo) -{ - return 0; -} - int kgd_gfx_v9_wave_control_execute(struct amdgpu_device *adev, uint32_t gfx_index_val, uint32_t sq_cmd) @@ -660,13 +646,6 @@ int kgd_gfx_v9_wave_control_execute(struct amdgpu_device *adev, return 0; } -uint32_t kgd_gfx_v9_address_watch_get_offset(struct amdgpu_device *adev, - unsigned int watch_point_id, - unsigned int reg_offset) -{ - return 0; -} - void kgd_gfx_v9_set_vm_context_page_table_base(struct amdgpu_device *adev, uint32_t vmid, uint64_t page_table_base) { @@ -736,7 +715,7 @@ static void get_wave_count(struct amdgpu_device *adev, int queue_idx, * process whose pasid is provided as a parameter. The process could have ZERO * or more queues running and submitting waves to compute units. * - * @kgd: Handle of device from which to get number of waves in flight + * @adev: Handle of device from which to get number of waves in flight * @pasid: Identifies the process for which this query call is invoked * @pasid_wave_cnt: Output parameter updated with number of waves in flight that * belong to process with given pasid @@ -745,7 +724,7 @@ static void get_wave_count(struct amdgpu_device *adev, int queue_idx, * * Note: It's possible that the device has too many queues (oversubscription) * in which case a VMID could be remapped to a different PASID. This could lead - * to an iaccurate wave count. Following is a high-level sequence: + * to an inaccurate wave count. Following is a high-level sequence: * Time T1: vmid = getVmid(); vmid is associated with Pasid P1 * Time T2: passId = getPasId(vmid); vmid is associated with Pasid P2 * In the sequence above wave count obtained from time T1 will be incorrectly @@ -888,10 +867,7 @@ const struct kfd2kgd_calls gfx_v9_kfd2kgd = { .hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied, .hqd_destroy = kgd_gfx_v9_hqd_destroy, .hqd_sdma_destroy = kgd_hqd_sdma_destroy, - .address_watch_disable = kgd_gfx_v9_address_watch_disable, - .address_watch_execute = kgd_gfx_v9_address_watch_execute, .wave_control_execute = kgd_gfx_v9_wave_control_execute, - .address_watch_get_offset = kgd_gfx_v9_address_watch_get_offset, .get_atc_vmid_pasid_mapping_info = kgd_gfx_v9_get_atc_vmid_pasid_mapping_info, .set_vm_context_page_table_base = kgd_gfx_v9_set_vm_context_page_table_base, 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 24be49df26fd..c7ed3bc9053c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h @@ -46,19 +46,9 @@ int kgd_gfx_v9_hqd_destroy(struct amdgpu_device *adev, void *mqd, enum kfd_preempt_type reset_type, unsigned int utimeout, uint32_t pipe_id, uint32_t queue_id); -int kgd_gfx_v9_address_watch_disable(struct amdgpu_device *adev); -int kgd_gfx_v9_address_watch_execute(struct amdgpu_device *adev, - unsigned int watch_point_id, - uint32_t cntl_val, - uint32_t addr_hi, - uint32_t addr_lo); int kgd_gfx_v9_wave_control_execute(struct amdgpu_device *adev, uint32_t gfx_index_val, uint32_t sq_cmd); -uint32_t kgd_gfx_v9_address_watch_get_offset(struct amdgpu_device *adev, - unsigned int watch_point_id, - unsigned int reg_offset); - bool kgd_gfx_v9_get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev, uint8_t vmid, uint16_t *p_pasid); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index b6f266f612ea..900ed2a7483b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -121,7 +121,7 @@ static size_t amdgpu_amdkfd_acc_size(uint64_t size) } /** - * @amdgpu_amdkfd_reserve_mem_limit() - Decrease available memory by size + * amdgpu_amdkfd_reserve_mem_limit() - Decrease available memory by size * of buffer including any reserved for control structures * * @adev: Device to which allocated BO belongs to @@ -743,7 +743,7 @@ unwind: continue; if (attachment[i]->bo_va) { amdgpu_bo_reserve(bo[i], true); - amdgpu_vm_bo_rmv(adev, attachment[i]->bo_va); + amdgpu_vm_bo_del(adev, attachment[i]->bo_va); amdgpu_bo_unreserve(bo[i]); list_del(&attachment[i]->list); } @@ -760,7 +760,7 @@ static void kfd_mem_detach(struct kfd_mem_attachment *attachment) pr_debug("\t remove VA 0x%llx in entry %p\n", attachment->va, attachment); - amdgpu_vm_bo_rmv(attachment->adev, attachment->bo_va); + amdgpu_vm_bo_del(attachment->adev, attachment->bo_va); drm_gem_object_put(&bo->tbo.base); list_del(&attachment->list); kfree(attachment); @@ -807,7 +807,8 @@ static void remove_kgd_mem_from_kfd_bo_list(struct kgd_mem *mem, * * Returns 0 for success, negative errno for errors. */ -static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr) +static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr, + bool criu_resume) { struct amdkfd_process_info *process_info = mem->process_info; struct amdgpu_bo *bo = mem->bo; @@ -829,6 +830,18 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr) goto out; } + if (criu_resume) { + /* + * During a CRIU restore operation, the userptr buffer objects + * will be validated in the restore_userptr_work worker at a + * later stage when it is scheduled by another ioctl called by + * CRIU master process for the target pid for restore. + */ + atomic_inc(&mem->invalid); + mutex_unlock(&process_info->lock); + return 0; + } + ret = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages); if (ret) { pr_err("%s: Failed to get user pages: %d\n", __func__, ret); @@ -1417,10 +1430,39 @@ uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *drm_priv) return avm->pd_phys_addr; } +void amdgpu_amdkfd_block_mmu_notifications(void *p) +{ + struct amdkfd_process_info *pinfo = (struct amdkfd_process_info *)p; + + mutex_lock(&pinfo->lock); + WRITE_ONCE(pinfo->block_mmu_notifications, true); + mutex_unlock(&pinfo->lock); +} + +int amdgpu_amdkfd_criu_resume(void *p) +{ + int ret = 0; + struct amdkfd_process_info *pinfo = (struct amdkfd_process_info *)p; + + mutex_lock(&pinfo->lock); + pr_debug("scheduling work\n"); + atomic_inc(&pinfo->evicted_bos); + if (!READ_ONCE(pinfo->block_mmu_notifications)) { + ret = -EINVAL; + goto out_unlock; + } + WRITE_ONCE(pinfo->block_mmu_notifications, false); + schedule_delayed_work(&pinfo->restore_userptr_work, 0); + +out_unlock: + mutex_unlock(&pinfo->lock); + return ret; +} + int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu( struct amdgpu_device *adev, uint64_t va, uint64_t size, void *drm_priv, struct kgd_mem **mem, - uint64_t *offset, uint32_t flags) + uint64_t *offset, uint32_t flags, bool criu_resume) { struct amdgpu_vm *avm = drm_priv_to_vm(drm_priv); enum ttm_bo_type bo_type = ttm_bo_type_device; @@ -1523,7 +1565,8 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu( add_kgd_mem_to_kfd_bo_list(*mem, avm->process_info, user_addr); if (user_addr) { - ret = init_user_pages(*mem, user_addr); + pr_debug("creating userptr BO for user_addr = %llu\n", user_addr); + ret = init_user_pages(*mem, user_addr, criu_resume); if (ret) goto allocate_init_user_pages_failed; } else if (flags & (KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL | @@ -1778,12 +1821,6 @@ int amdgpu_amdkfd_gpuvm_map_memory_to_gpu( true); ret = unreserve_bo_and_vms(&ctx, false, false); - /* Only apply no TLB flush on Aldebaran to - * workaround regressions on other Asics. - */ - if (table_freed && (adev->asic_type != CHIP_ALDEBARAN)) - *table_freed = true; - goto out; out_unreserve: @@ -2033,6 +2070,10 @@ int amdgpu_amdkfd_evict_userptr(struct kgd_mem *mem, int evicted_bos; int r = 0; + /* Do not process MMU notifications until stage-4 IOCTL is received */ + if (READ_ONCE(process_info->block_mmu_notifications)) + return 0; + atomic_inc(&mem->invalid); evicted_bos = atomic_inc_return(&process_info->evicted_bos); if (evicted_bos == 1) { @@ -2600,3 +2641,14 @@ int amdgpu_amdkfd_get_tile_config(struct amdgpu_device *adev, return 0; } + +bool amdgpu_amdkfd_bo_mapped_to_dev(struct amdgpu_device *adev, struct kgd_mem *mem) +{ + struct kfd_mem_attachment *entry; + + list_for_each_entry(entry, &mem->attachments, list) { + if (entry->is_mapped && entry->adev == adev) + return true; + } + return false; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c index 12a6b1c99c93..9ba4817a9148 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c @@ -1083,6 +1083,7 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev, return 0; } +#ifdef CONFIG_DRM_AMDGPU_SI int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev, u32 clock, bool strobe_mode, @@ -1503,6 +1504,7 @@ int amdgpu_atombios_init_mc_reg_table(struct amdgpu_device *adev, } return -EINVAL; } +#endif bool amdgpu_atombios_has_gpu_virtualization_table(struct amdgpu_device *adev) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h index 27e74b1fc260..4153d520e2a3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h @@ -160,6 +160,7 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev, bool strobe_mode, struct atom_clock_dividers *dividers); +#ifdef CONFIG_DRM_AMDGPU_SI int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev, u32 clock, bool strobe_mode, @@ -179,6 +180,17 @@ int amdgpu_atombios_get_voltage_table(struct amdgpu_device *adev, int amdgpu_atombios_init_mc_reg_table(struct amdgpu_device *adev, u8 module_index, struct atom_mc_reg_table *reg_table); +int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type, + u16 voltage_id, u16 *voltage); +int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device *adev, + u16 *voltage, + u16 leakage_idx); +void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev, + u16 *vddc, u16 *vddci, u16 *mvdd); +int amdgpu_atombios_get_svi2_info(struct amdgpu_device *adev, + u8 voltage_type, + u8 *svd_gpio_id, u8 *svc_gpio_id); +#endif bool amdgpu_atombios_has_gpu_virtualization_table(struct amdgpu_device *adev); @@ -190,21 +202,11 @@ void amdgpu_atombios_scratch_regs_set_backlight_level(struct amdgpu_device *adev bool amdgpu_atombios_scratch_need_asic_init(struct amdgpu_device *adev); void amdgpu_atombios_copy_swap(u8 *dst, u8 *src, u8 num_bytes, bool to_le); -int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type, - u16 voltage_id, u16 *voltage); -int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device *adev, - u16 *voltage, - u16 leakage_idx); -void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev, - u16 *vddc, u16 *vddci, u16 *mvdd); int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev, u8 clock_type, u32 clock, bool strobe_mode, struct atom_clock_dividers *dividers); -int amdgpu_atombios_get_svi2_info(struct amdgpu_device *adev, - u8 voltage_type, - u8 *svd_gpio_id, u8 *svc_gpio_id); int amdgpu_atombios_get_data_table(struct amdgpu_device *adev, uint32_t table, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c index 313517f7cf10..edc6377ec5ff 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c @@ -29,14 +29,13 @@ #define AMDGPU_BENCHMARK_COMMON_MODES_N 17 static int amdgpu_benchmark_do_move(struct amdgpu_device *adev, unsigned size, - uint64_t saddr, uint64_t daddr, int n) + uint64_t saddr, uint64_t daddr, int n, s64 *time_ms) { - unsigned long start_jiffies; - unsigned long end_jiffies; + ktime_t stime, etime; struct dma_fence *fence; int i, r; - start_jiffies = jiffies; + stime = ktime_get(); for (i = 0; i < n; i++) { struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; r = amdgpu_copy_buffer(ring, saddr, daddr, size, NULL, &fence, @@ -48,120 +47,81 @@ static int amdgpu_benchmark_do_move(struct amdgpu_device *adev, unsigned size, if (r) goto exit_do_move; } - end_jiffies = jiffies; - r = jiffies_to_msecs(end_jiffies - start_jiffies); exit_do_move: + etime = ktime_get(); + *time_ms = ktime_ms_delta(etime, stime); + return r; } -static void amdgpu_benchmark_log_results(int n, unsigned size, - unsigned int time, +static void amdgpu_benchmark_log_results(struct amdgpu_device *adev, + int n, unsigned size, + s64 time_ms, unsigned sdomain, unsigned ddomain, char *kind) { - unsigned int throughput = (n * (size >> 10)) / time; - DRM_INFO("amdgpu: %s %u bo moves of %u kB from" - " %d to %d in %u ms, throughput: %u Mb/s or %u MB/s\n", - kind, n, size >> 10, sdomain, ddomain, time, + s64 throughput = (n * (size >> 10)); + + throughput = div64_s64(throughput, time_ms); + + dev_info(adev->dev, "amdgpu: %s %u bo moves of %u kB from" + " %d to %d in %lld ms, throughput: %lld Mb/s or %lld MB/s\n", + kind, n, size >> 10, sdomain, ddomain, time_ms, throughput * 8, throughput); } -static void amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size, - unsigned sdomain, unsigned ddomain) +static int amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size, + unsigned sdomain, unsigned ddomain) { struct amdgpu_bo *dobj = NULL; struct amdgpu_bo *sobj = NULL; - struct amdgpu_bo_param bp; uint64_t saddr, daddr; + s64 time_ms; int r, n; - int time; - - memset(&bp, 0, sizeof(bp)); - bp.size = size; - bp.byte_align = PAGE_SIZE; - bp.domain = sdomain; - bp.flags = 0; - bp.type = ttm_bo_type_kernel; - bp.resv = NULL; - bp.bo_ptr_size = sizeof(struct amdgpu_bo); n = AMDGPU_BENCHMARK_ITERATIONS; - r = amdgpu_bo_create(adev, &bp, &sobj); - if (r) { - goto out_cleanup; - } - r = amdgpu_bo_reserve(sobj, false); - if (unlikely(r != 0)) - goto out_cleanup; - r = amdgpu_bo_pin(sobj, sdomain); - if (r) { - amdgpu_bo_unreserve(sobj); - goto out_cleanup; - } - r = amdgpu_ttm_alloc_gart(&sobj->tbo); - amdgpu_bo_unreserve(sobj); - if (r) { - goto out_cleanup; - } - saddr = amdgpu_bo_gpu_offset(sobj); - bp.domain = ddomain; - r = amdgpu_bo_create(adev, &bp, &dobj); - if (r) { - goto out_cleanup; - } - r = amdgpu_bo_reserve(dobj, false); - if (unlikely(r != 0)) - goto out_cleanup; - r = amdgpu_bo_pin(dobj, ddomain); - if (r) { - amdgpu_bo_unreserve(sobj); + + r = amdgpu_bo_create_kernel(adev, size, + PAGE_SIZE, sdomain, + &sobj, + &saddr, + NULL); + if (r) goto out_cleanup; - } - r = amdgpu_ttm_alloc_gart(&dobj->tbo); - amdgpu_bo_unreserve(dobj); - if (r) { + r = amdgpu_bo_create_kernel(adev, size, + PAGE_SIZE, ddomain, + &dobj, + &daddr, + NULL); + if (r) goto out_cleanup; - } - daddr = amdgpu_bo_gpu_offset(dobj); if (adev->mman.buffer_funcs) { - time = amdgpu_benchmark_do_move(adev, size, saddr, daddr, n); - if (time < 0) + r = amdgpu_benchmark_do_move(adev, size, saddr, daddr, n, &time_ms); + if (r) goto out_cleanup; - if (time > 0) - amdgpu_benchmark_log_results(n, size, time, + else + amdgpu_benchmark_log_results(adev, n, size, time_ms, sdomain, ddomain, "dma"); } out_cleanup: /* Check error value now. The value can be overwritten when clean up.*/ - if (r) { - DRM_ERROR("Error while benchmarking BO move.\n"); - } + if (r < 0) + dev_info(adev->dev, "Error while benchmarking BO move.\n"); - if (sobj) { - r = amdgpu_bo_reserve(sobj, true); - if (likely(r == 0)) { - amdgpu_bo_unpin(sobj); - amdgpu_bo_unreserve(sobj); - } - amdgpu_bo_unref(&sobj); - } - if (dobj) { - r = amdgpu_bo_reserve(dobj, true); - if (likely(r == 0)) { - amdgpu_bo_unpin(dobj); - amdgpu_bo_unreserve(dobj); - } - amdgpu_bo_unref(&dobj); - } + if (sobj) + amdgpu_bo_free_kernel(&sobj, &saddr, NULL); + if (dobj) + amdgpu_bo_free_kernel(&dobj, &daddr, NULL); + return r; } -void amdgpu_benchmark(struct amdgpu_device *adev, int test_number) +int amdgpu_benchmark(struct amdgpu_device *adev, int test_number) { - int i; + int i, r; static const int common_modes[AMDGPU_BENCHMARK_COMMON_MODES_N] = { 640 * 480 * 4, 720 * 480 * 4, @@ -182,63 +142,119 @@ void amdgpu_benchmark(struct amdgpu_device *adev, int test_number) 1920 * 1200 * 4 }; + mutex_lock(&adev->benchmark_mutex); switch (test_number) { case 1: + dev_info(adev->dev, + "benchmark test: %d (simple test, VRAM to GTT and GTT to VRAM)\n", + test_number); /* simple test, VRAM to GTT and GTT to VRAM */ - amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_GTT, - AMDGPU_GEM_DOMAIN_VRAM); - amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_DOMAIN_GTT); + r = amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_GTT, + AMDGPU_GEM_DOMAIN_VRAM); + if (r) + goto done; + r = amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM, + AMDGPU_GEM_DOMAIN_GTT); + if (r) + goto done; break; case 2: + dev_info(adev->dev, + "benchmark test: %d (simple test, VRAM to VRAM)\n", + test_number); /* simple test, VRAM to VRAM */ - amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_DOMAIN_VRAM); + r = amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM, + AMDGPU_GEM_DOMAIN_VRAM); + if (r) + goto done; break; case 3: + dev_info(adev->dev, + "benchmark test: %d (GTT to VRAM, buffer size sweep, powers of 2)\n", + test_number); /* GTT to VRAM, buffer size sweep, powers of 2 */ - for (i = 1; i <= 16384; i <<= 1) - amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE, - AMDGPU_GEM_DOMAIN_GTT, - AMDGPU_GEM_DOMAIN_VRAM); + for (i = 1; i <= 16384; i <<= 1) { + r = amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE, + AMDGPU_GEM_DOMAIN_GTT, + AMDGPU_GEM_DOMAIN_VRAM); + if (r) + goto done; + } break; case 4: + dev_info(adev->dev, + "benchmark test: %d (VRAM to GTT, buffer size sweep, powers of 2)\n", + test_number); /* VRAM to GTT, buffer size sweep, powers of 2 */ - for (i = 1; i <= 16384; i <<= 1) - amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_DOMAIN_GTT); + for (i = 1; i <= 16384; i <<= 1) { + r = amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, + AMDGPU_GEM_DOMAIN_GTT); + if (r) + goto done; + } break; case 5: + dev_info(adev->dev, + "benchmark test: %d (VRAM to VRAM, buffer size sweep, powers of 2)\n", + test_number); /* VRAM to VRAM, buffer size sweep, powers of 2 */ - for (i = 1; i <= 16384; i <<= 1) - amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_DOMAIN_VRAM); + for (i = 1; i <= 16384; i <<= 1) { + r = amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, + AMDGPU_GEM_DOMAIN_VRAM); + if (r) + goto done; + } break; case 6: + dev_info(adev->dev, + "benchmark test: %d (GTT to VRAM, buffer size sweep, common modes)\n", + test_number); /* GTT to VRAM, buffer size sweep, common modes */ - for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++) - amdgpu_benchmark_move(adev, common_modes[i], - AMDGPU_GEM_DOMAIN_GTT, - AMDGPU_GEM_DOMAIN_VRAM); + for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++) { + r = amdgpu_benchmark_move(adev, common_modes[i], + AMDGPU_GEM_DOMAIN_GTT, + AMDGPU_GEM_DOMAIN_VRAM); + if (r) + goto done; + } break; case 7: + dev_info(adev->dev, + "benchmark test: %d (VRAM to GTT, buffer size sweep, common modes)\n", + test_number); /* VRAM to GTT, buffer size sweep, common modes */ - for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++) - amdgpu_benchmark_move(adev, common_modes[i], - AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_DOMAIN_GTT); + for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++) { + r = amdgpu_benchmark_move(adev, common_modes[i], + AMDGPU_GEM_DOMAIN_VRAM, + AMDGPU_GEM_DOMAIN_GTT); + if (r) + goto done; + } break; case 8: + dev_info(adev->dev, + "benchmark test: %d (VRAM to VRAM, buffer size sweep, common modes)\n", + test_number); /* VRAM to VRAM, buffer size sweep, common modes */ - for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++) - amdgpu_benchmark_move(adev, common_modes[i], + for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++) { + r = amdgpu_benchmark_move(adev, common_modes[i], AMDGPU_GEM_DOMAIN_VRAM, AMDGPU_GEM_DOMAIN_VRAM); + if (r) + goto done; + } break; default: - DRM_ERROR("Unknown benchmark\n"); + dev_info(adev->dev, "Unknown benchmark %d\n", test_number); + r = -EINVAL; + break; } + +done: + mutex_unlock(&adev->benchmark_mutex); + + return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c index 27b19503773b..0eddca795e96 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c @@ -464,3 +464,41 @@ success: adev->is_atom_fw = (adev->asic_type >= CHIP_VEGA10) ? true : false; return true; } + +/* helper function for soc15 and onwards to read bios from rom */ +bool amdgpu_soc15_read_bios_from_rom(struct amdgpu_device *adev, + u8 *bios, u32 length_bytes) +{ + u32 *dw_ptr; + u32 i, length_dw; + u32 rom_index_offset; + u32 rom_data_offset; + + if (bios == NULL) + return false; + if (length_bytes == 0) + return false; + /* APU vbios image is part of sbios image */ + if (adev->flags & AMD_IS_APU) + return false; + if (!adev->smuio.funcs || + !adev->smuio.funcs->get_rom_index_offset || + !adev->smuio.funcs->get_rom_data_offset) + return false; + + dw_ptr = (u32 *)bios; + length_dw = ALIGN(length_bytes, 4) / 4; + + rom_index_offset = + adev->smuio.funcs->get_rom_index_offset(adev); + rom_data_offset = + adev->smuio.funcs->get_rom_data_offset(adev); + + /* set rom index to 0 */ + WREG32(rom_index_offset, 0); + /* read out the rom data */ + for (i = 0; i < length_dw; i++) + dw_ptr[i] = RREG32(rom_data_offset); + + return true; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c index fa20261aa928..673078faa27a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c @@ -626,7 +626,7 @@ amdgpu_connector_fixup_lcd_native_mode(struct drm_encoder *encoder, if (mode->type & DRM_MODE_TYPE_PREFERRED) { if (mode->hdisplay != native_mode->hdisplay || mode->vdisplay != native_mode->vdisplay) - memcpy(native_mode, mode, sizeof(*mode)); + drm_mode_copy(native_mode, mode); } } @@ -635,7 +635,7 @@ amdgpu_connector_fixup_lcd_native_mode(struct drm_encoder *encoder, list_for_each_entry_safe(mode, t, &connector->probed_modes, head) { if (mode->hdisplay == native_mode->hdisplay && mode->vdisplay == native_mode->vdisplay) { - *native_mode = *mode; + drm_mode_copy(native_mode, mode); drm_mode_set_crtcinfo(native_mode, CRTC_INTERLACE_HALVE_V); DRM_DEBUG_KMS("Determined LVDS native mode details from EDID\n"); break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index ee59b6928b11..e85e347eb670 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -32,6 +32,7 @@ #include <drm/amdgpu_drm.h> #include <drm/drm_syncobj.h> +#include "amdgpu_cs.h" #include "amdgpu.h" #include "amdgpu_trace.h" #include "amdgpu_gmc.h" @@ -127,8 +128,6 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs goto free_chunk; } - mutex_lock(&p->ctx->lock); - /* skip guilty context job */ if (atomic_read(&p->ctx->guilty) == 1) { ret = -ECANCELED; @@ -341,7 +340,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev, if (free_vram >= 128 * 1024 * 1024 || free_vram >= total_vram / 8) { s64 min_us; - /* Be more aggresive on dGPUs. Try to fill a portion of free + /* Be more aggressive on dGPUs. Try to fill a portion of free * VRAM now. */ if (!(adev->flags & AMD_IS_APU)) @@ -585,6 +584,16 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, } } + /* Move fence waiting after getting reservation lock of + * PD root. Then there is no need on a ctx mutex lock. + */ + r = amdgpu_ctx_wait_prev_fence(p->ctx, p->entity); + if (unlikely(r != 0)) { + if (r != -ERESTARTSYS) + DRM_ERROR("amdgpu_ctx_wait_prev_fence failed.\n"); + goto error_validate; + } + amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold, &p->bytes_moved_vis_threshold); p->bytes_moved = 0; @@ -700,7 +709,6 @@ static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, dma_fence_put(parser->fence); if (parser->ctx) { - mutex_unlock(&parser->ctx->lock); amdgpu_ctx_put(parser->ctx); } if (parser->bo_list) @@ -775,12 +783,12 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p) memcpy(ib->ptr, kptr, chunk_ib->ib_bytes); amdgpu_bo_kunmap(aobj); - r = amdgpu_ring_parse_cs(ring, p, j); + r = amdgpu_ring_parse_cs(ring, p, p->job, ib); if (r) return r; } else { ib->ptr = (uint32_t *)kptr; - r = amdgpu_ring_patch_cs_in_place(ring, p, j); + r = amdgpu_ring_patch_cs_in_place(ring, p, p->job, ib); amdgpu_bo_kunmap(aobj); if (r) return r; @@ -944,7 +952,7 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev, if (parser->job->uf_addr && ring->funcs->no_user_fence) return -EINVAL; - return amdgpu_ctx_wait_prev_fence(parser->ctx, parser->entity); + return 0; } static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p, @@ -1365,7 +1373,6 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) goto out; r = amdgpu_cs_submit(&parser, cs); - out: amdgpu_cs_parser_fini(&parser, r, reserved_buffers); @@ -1511,6 +1518,7 @@ int amdgpu_cs_fence_to_handle_ioctl(struct drm_device *dev, void *data, return 0; default: + dma_fence_put(fence); return -EINVAL; } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h new file mode 100644 index 000000000000..30ecc4917f81 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h @@ -0,0 +1,80 @@ +/* + * Copyright 2022 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_CS_H__ +#define __AMDGPU_CS_H__ + +#include "amdgpu_job.h" +#include "amdgpu_bo_list.h" +#include "amdgpu_ring.h" + +struct amdgpu_bo_va_mapping; + +struct amdgpu_cs_chunk { + uint32_t chunk_id; + uint32_t length_dw; + void *kdata; +}; + +struct amdgpu_cs_post_dep { + struct drm_syncobj *syncobj; + struct dma_fence_chain *chain; + u64 point; +}; + +struct amdgpu_cs_parser { + struct amdgpu_device *adev; + struct drm_file *filp; + struct amdgpu_ctx *ctx; + + /* chunks */ + unsigned nchunks; + struct amdgpu_cs_chunk *chunks; + + /* scheduler job object */ + struct amdgpu_job *job; + struct drm_sched_entity *entity; + + /* buffer objects */ + struct ww_acquire_ctx ticket; + struct amdgpu_bo_list *bo_list; + struct amdgpu_mn *mn; + struct amdgpu_bo_list_entry vm_pd; + struct list_head validated; + struct dma_fence *fence; + uint64_t bytes_moved_threshold; + uint64_t bytes_moved_vis_threshold; + uint64_t bytes_moved; + uint64_t bytes_moved_vis; + + /* user fence */ + struct amdgpu_bo_list_entry uf_entry; + + unsigned num_post_deps; + struct amdgpu_cs_post_dep *post_deps; +}; + +int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser, + uint64_t addr, struct amdgpu_bo **bo, + struct amdgpu_bo_va_mapping **mapping); + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c index da21e60bb827..c6d4d41c4393 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c @@ -98,7 +98,7 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, if (r) { DRM_ERROR("failed to do bo_map on static CSA, err=%d\n", r); - amdgpu_vm_bo_rmv(adev, *bo_va); + amdgpu_vm_bo_del(adev, *bo_va); ttm_eu_backoff_reservation(&ticket, &list); return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index 468003583b2a..5981c7d9bd48 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -23,6 +23,7 @@ */ #include <drm/drm_auth.h> +#include <drm/drm_drv.h> #include "amdgpu.h" #include "amdgpu_sched.h" #include "amdgpu_ras.h" @@ -204,9 +205,15 @@ static int amdgpu_ctx_init_entity(struct amdgpu_ctx *ctx, u32 hw_ip, if (r) goto error_free_entity; - ctx->entities[hw_ip][ring] = entity; + /* It's not an error if we fail to install the new entity */ + if (cmpxchg(&ctx->entities[hw_ip][ring], NULL, entity)) + goto cleanup_entity; + return 0; +cleanup_entity: + drm_sched_entity_fini(&entity->entity); + error_free_entity: kfree(entity); @@ -230,13 +237,13 @@ static int amdgpu_ctx_init(struct amdgpu_device *adev, kref_init(&ctx->refcount); spin_lock_init(&ctx->ring_lock); - mutex_init(&ctx->lock); ctx->reset_counter = atomic_read(&adev->gpu_reset_counter); ctx->reset_counter_query = ctx->reset_counter; ctx->vram_lost_counter = atomic_read(&adev->vram_lost_counter); ctx->init_priority = priority; ctx->override_priority = AMDGPU_CTX_PRIORITY_UNSET; + ctx->stable_pstate = AMDGPU_CTX_STABLE_PSTATE_NONE; return 0; } @@ -255,11 +262,85 @@ static void amdgpu_ctx_fini_entity(struct amdgpu_ctx_entity *entity) kfree(entity); } +static int amdgpu_ctx_get_stable_pstate(struct amdgpu_ctx *ctx, + u32 *stable_pstate) +{ + struct amdgpu_device *adev = ctx->adev; + enum amd_dpm_forced_level current_level; + + current_level = amdgpu_dpm_get_performance_level(adev); + + switch (current_level) { + case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD: + *stable_pstate = AMDGPU_CTX_STABLE_PSTATE_STANDARD; + break; + case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK: + *stable_pstate = AMDGPU_CTX_STABLE_PSTATE_MIN_SCLK; + break; + case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK: + *stable_pstate = AMDGPU_CTX_STABLE_PSTATE_MIN_MCLK; + break; + case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK: + *stable_pstate = AMDGPU_CTX_STABLE_PSTATE_PEAK; + break; + default: + *stable_pstate = AMDGPU_CTX_STABLE_PSTATE_NONE; + break; + } + return 0; +} + +static int amdgpu_ctx_set_stable_pstate(struct amdgpu_ctx *ctx, + u32 stable_pstate) +{ + struct amdgpu_device *adev = ctx->adev; + enum amd_dpm_forced_level level; + int r; + + mutex_lock(&adev->pm.stable_pstate_ctx_lock); + if (adev->pm.stable_pstate_ctx && adev->pm.stable_pstate_ctx != ctx) { + r = -EBUSY; + goto done; + } + + switch (stable_pstate) { + case AMDGPU_CTX_STABLE_PSTATE_NONE: + level = AMD_DPM_FORCED_LEVEL_AUTO; + break; + case AMDGPU_CTX_STABLE_PSTATE_STANDARD: + level = AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD; + break; + case AMDGPU_CTX_STABLE_PSTATE_MIN_SCLK: + level = AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK; + break; + case AMDGPU_CTX_STABLE_PSTATE_MIN_MCLK: + level = AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK; + break; + case AMDGPU_CTX_STABLE_PSTATE_PEAK: + level = AMD_DPM_FORCED_LEVEL_PROFILE_PEAK; + break; + default: + r = -EINVAL; + goto done; + } + + r = amdgpu_dpm_force_performance_level(adev, level); + + if (level == AMD_DPM_FORCED_LEVEL_AUTO) + adev->pm.stable_pstate_ctx = NULL; + else + adev->pm.stable_pstate_ctx = ctx; +done: + mutex_unlock(&adev->pm.stable_pstate_ctx_lock); + + return r; +} + static void amdgpu_ctx_fini(struct kref *ref) { struct amdgpu_ctx *ctx = container_of(ref, struct amdgpu_ctx, refcount); struct amdgpu_device *adev = ctx->adev; - unsigned i, j; + unsigned i, j, idx; if (!adev) return; @@ -271,7 +352,11 @@ static void amdgpu_ctx_fini(struct kref *ref) } } - mutex_destroy(&ctx->lock); + if (drm_dev_enter(&adev->ddev, &idx)) { + amdgpu_ctx_set_stable_pstate(ctx, AMDGPU_CTX_STABLE_PSTATE_NONE); + drm_dev_exit(idx); + } + kfree(ctx); } @@ -467,11 +552,41 @@ static int amdgpu_ctx_query2(struct amdgpu_device *adev, return 0; } + + +static int amdgpu_ctx_stable_pstate(struct amdgpu_device *adev, + struct amdgpu_fpriv *fpriv, uint32_t id, + bool set, u32 *stable_pstate) +{ + struct amdgpu_ctx *ctx; + struct amdgpu_ctx_mgr *mgr; + int r; + + if (!fpriv) + return -EINVAL; + + mgr = &fpriv->ctx_mgr; + mutex_lock(&mgr->lock); + ctx = idr_find(&mgr->ctx_handles, id); + if (!ctx) { + mutex_unlock(&mgr->lock); + return -EINVAL; + } + + if (set) + r = amdgpu_ctx_set_stable_pstate(ctx, *stable_pstate); + else + r = amdgpu_ctx_get_stable_pstate(ctx, stable_pstate); + + mutex_unlock(&mgr->lock); + return r; +} + int amdgpu_ctx_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) { int r; - uint32_t id; + uint32_t id, stable_pstate; int32_t priority; union drm_amdgpu_ctx *args = data; @@ -500,6 +615,21 @@ int amdgpu_ctx_ioctl(struct drm_device *dev, void *data, case AMDGPU_CTX_OP_QUERY_STATE2: r = amdgpu_ctx_query2(adev, fpriv, id, &args->out); break; + case AMDGPU_CTX_OP_GET_STABLE_PSTATE: + if (args->in.flags) + return -EINVAL; + r = amdgpu_ctx_stable_pstate(adev, fpriv, id, false, &stable_pstate); + if (!r) + args->out.pstate.flags = stable_pstate; + break; + case AMDGPU_CTX_OP_SET_STABLE_PSTATE: + if (args->in.flags & ~AMDGPU_CTX_STABLE_PSTATE_FLAGS_MASK) + return -EINVAL; + stable_pstate = args->in.flags & AMDGPU_CTX_STABLE_PSTATE_FLAGS_MASK; + if (stable_pstate > AMDGPU_CTX_STABLE_PSTATE_PEAK) + return -EINVAL; + r = amdgpu_ctx_stable_pstate(adev, fpriv, id, true, &stable_pstate); + break; default: return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h index a44b8b8ed39c..d0cbfcea90f7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h @@ -49,10 +49,10 @@ struct amdgpu_ctx { bool preamble_presented; int32_t init_priority; int32_t override_priority; - struct mutex lock; atomic_t guilty; unsigned long ras_counter_ce; unsigned long ras_counter_ue; + uint32_t stable_pstate; }; struct amdgpu_ctx_mgr { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c index c3728061d65a..5d04d24a0d5f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c @@ -730,7 +730,7 @@ static ssize_t amdgpu_debugfs_gca_config_read(struct file *f, char __user *buf, return -ENOMEM; /* version, increment each time something is added */ - config[no_regs++] = 3; + config[no_regs++] = 4; config[no_regs++] = adev->gfx.config.max_shader_engines; config[no_regs++] = adev->gfx.config.max_tile_pipes; config[no_regs++] = adev->gfx.config.max_cu_per_sh; @@ -770,6 +770,9 @@ static ssize_t amdgpu_debugfs_gca_config_read(struct file *f, char __user *buf, config[no_regs++] = adev->pdev->subsystem_device; config[no_regs++] = adev->pdev->subsystem_vendor; + /* rev==4 APU flag */ + config[no_regs++] = adev->flags & AMD_IS_APU ? 1 : 0; + while (size && (*pos < no_regs * 4)) { uint32_t value; @@ -1122,8 +1125,10 @@ static ssize_t amdgpu_debugfs_gfxoff_read(struct file *f, char __user *buf, return -EINVAL; r = pm_runtime_get_sync(adev_to_drm(adev)->dev); - if (r < 0) + if (r < 0) { + pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; + } while (size) { uint32_t value; @@ -1359,6 +1364,25 @@ static int amdgpu_debugfs_evict_gtt(void *data, u64 *val) return 0; } +static int amdgpu_debugfs_benchmark(void *data, u64 val) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)data; + struct drm_device *dev = adev_to_drm(adev); + int r; + + r = pm_runtime_get_sync(dev->dev); + if (r < 0) { + pm_runtime_put_autosuspend(dev->dev); + return r; + } + + r = amdgpu_benchmark(adev, val); + + pm_runtime_mark_last_busy(dev->dev); + pm_runtime_put_autosuspend(dev->dev); + + return r; +} static int amdgpu_debugfs_vm_info_show(struct seq_file *m, void *unused) { @@ -1395,6 +1419,8 @@ DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_evict_vram_fops, amdgpu_debugfs_evict_vram, NULL, "%lld\n"); DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_evict_gtt_fops, amdgpu_debugfs_evict_gtt, NULL, "%lld\n"); +DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_benchmark_fops, NULL, amdgpu_debugfs_benchmark, + "%lld\n"); static void amdgpu_ib_preempt_fences_swap(struct amdgpu_ring *ring, struct dma_fence **fences) @@ -1587,22 +1613,25 @@ static int amdgpu_debugfs_sclk_set(void *data, u64 val) return ret; } - if (is_support_sw_smu(adev)) { - ret = smu_get_dpm_freq_range(&adev->smu, SMU_SCLK, &min_freq, &max_freq); - if (ret || val > max_freq || val < min_freq) - return -EINVAL; - ret = smu_set_soft_freq_range(&adev->smu, SMU_SCLK, (uint32_t)val, (uint32_t)val); - } else { - return 0; + ret = amdgpu_dpm_get_dpm_freq_range(adev, PP_SCLK, &min_freq, &max_freq); + if (ret == -EOPNOTSUPP) { + ret = 0; + goto out; } + if (ret || val > max_freq || val < min_freq) { + ret = -EINVAL; + goto out; + } + + ret = amdgpu_dpm_set_soft_freq_range(adev, PP_SCLK, (uint32_t)val, (uint32_t)val); + if (ret) + ret = -EINVAL; +out: pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); - if (ret) - return -EINVAL; - - return 0; + return ret; } DEFINE_DEBUGFS_ATTRIBUTE(fops_ib_preempt, NULL, @@ -1611,6 +1640,91 @@ DEFINE_DEBUGFS_ATTRIBUTE(fops_ib_preempt, NULL, DEFINE_DEBUGFS_ATTRIBUTE(fops_sclk_set, NULL, amdgpu_debugfs_sclk_set, "%llu\n"); +static ssize_t amdgpu_reset_dump_register_list_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; + char reg_offset[12]; + int i, ret, len = 0; + + if (*pos) + return 0; + + memset(reg_offset, 0, 12); + ret = down_read_killable(&adev->reset_domain->sem); + if (ret) + return ret; + + for (i = 0; i < adev->num_regs; i++) { + sprintf(reg_offset, "0x%x\n", adev->reset_dump_reg_list[i]); + up_read(&adev->reset_domain->sem); + if (copy_to_user(buf + len, reg_offset, strlen(reg_offset))) + return -EFAULT; + + len += strlen(reg_offset); + ret = down_read_killable(&adev->reset_domain->sem); + if (ret) + return ret; + } + + up_read(&adev->reset_domain->sem); + *pos += len; + + return len; +} + +static ssize_t amdgpu_reset_dump_register_list_write(struct file *f, + const char __user *buf, size_t size, loff_t *pos) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private; + char reg_offset[11]; + uint32_t *new, *tmp = NULL; + int ret, i = 0, len = 0; + + do { + memset(reg_offset, 0, 11); + if (copy_from_user(reg_offset, buf + len, + min(10, ((int)size-len)))) { + ret = -EFAULT; + goto error_free; + } + + new = krealloc_array(tmp, i + 1, sizeof(uint32_t), GFP_KERNEL); + if (!new) { + ret = -ENOMEM; + goto error_free; + } + tmp = new; + if (sscanf(reg_offset, "%X %n", &tmp[i], &ret) != 1) { + ret = -EINVAL; + goto error_free; + } + + len += ret; + i++; + } while (len < size); + + ret = down_write_killable(&adev->reset_domain->sem); + if (ret) + goto error_free; + + swap(adev->reset_dump_reg_list, tmp); + adev->num_regs = i; + up_write(&adev->reset_domain->sem); + ret = size; + +error_free: + kfree(tmp); + return ret; +} + +static const struct file_operations amdgpu_reset_dump_register_list = { + .owner = THIS_MODULE, + .read = amdgpu_reset_dump_register_list_read, + .write = amdgpu_reset_dump_register_list_write, + .llseek = default_llseek +}; + int amdgpu_debugfs_init(struct amdgpu_device *adev) { struct dentry *root = adev_to_drm(adev)->primary->debugfs_root; @@ -1664,6 +1778,16 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev) amdgpu_debugfs_ring_init(adev, ring); } + for ( i = 0; i < adev->vcn.num_vcn_inst; i++) { + if (!amdgpu_vcnfw_log) + break; + + if (adev->vcn.harvest_config & (1 << i)) + continue; + + amdgpu_debugfs_vcn_fwlog_init(adev, i, &adev->vcn.inst[i]); + } + amdgpu_ras_debugfs_create_all(adev); amdgpu_rap_debugfs_init(adev); amdgpu_securedisplay_debugfs_init(adev); @@ -1677,6 +1801,10 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev) &amdgpu_debugfs_test_ib_fops); debugfs_create_file("amdgpu_vm_info", 0444, root, adev, &amdgpu_debugfs_vm_info_fops); + debugfs_create_file("amdgpu_benchmark", 0200, root, adev, + &amdgpu_benchmark_fops); + debugfs_create_file("amdgpu_reset_dump_register_list", 0644, root, adev, + &amdgpu_reset_dump_register_list); adev->debugfs_vbios_blob.data = adev->bios; adev->debugfs_vbios_blob.size = adev->bios_size; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index cd1f9140a878..3987ecb24ef4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -31,6 +31,7 @@ #include <linux/console.h> #include <linux/slab.h> #include <linux/iommu.h> +#include <linux/pci.h> #include <drm/drm_atomic_helper.h> #include <drm/drm_probe_helper.h> @@ -55,7 +56,6 @@ #include "soc15.h" #include "nv.h" #include "bif/bif_4_1_d.h" -#include <linux/pci.h> #include <linux/firmware.h> #include "amdgpu_vf_error.h" @@ -80,14 +80,11 @@ MODULE_FIRMWARE("amdgpu/raven_gpu_info.bin"); MODULE_FIRMWARE("amdgpu/picasso_gpu_info.bin"); MODULE_FIRMWARE("amdgpu/raven2_gpu_info.bin"); MODULE_FIRMWARE("amdgpu/arcturus_gpu_info.bin"); -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/yellow_carp_gpu_info.bin"); #define AMDGPU_RESUME_MS 2000 +#define AMDGPU_MAX_RETRY_LIMIT 2 +#define AMDGPU_RETRY_SRIOV_RESET(r) ((r) == -EBUSY || (r) == -ETIMEDOUT || (r) == -EINVAL) const char *amdgpu_asic_name[] = { "TAHITI", @@ -554,7 +551,11 @@ void amdgpu_device_wreg(struct amdgpu_device *adev, /** * amdgpu_mm_wreg_mmio_rlc - write register either with direct/indirect mmio or with RLC path if in range * - * this function is invoked only the debugfs register access + * @adev: amdgpu_device pointer + * @reg: mmio/rlc register + * @v: value to write + * + * this function is invoked only for the debugfs register access */ void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, uint32_t reg, uint32_t v) @@ -566,7 +567,7 @@ void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, adev->gfx.rlc.funcs && adev->gfx.rlc.funcs->is_rlcg_access_range) { if (adev->gfx.rlc.funcs->is_rlcg_access_range(adev, reg)) - return adev->gfx.rlc.funcs->sriov_wreg(adev, reg, v, 0, 0); + return amdgpu_sriov_wreg(adev, reg, v, 0, 0); } else if ((reg * 4) >= adev->rmmio_size) { adev->pcie_wreg(adev, reg * 4, v); } else { @@ -1312,6 +1313,31 @@ bool amdgpu_device_need_post(struct amdgpu_device *adev) return true; } +/** + * amdgpu_device_should_use_aspm - check if the device should program ASPM + * + * @adev: amdgpu_device pointer + * + * Confirm whether the module parameter and pcie bridge agree that ASPM should + * be set for this device. + * + * Returns true if it should be used or false if not. + */ +bool amdgpu_device_should_use_aspm(struct amdgpu_device *adev) +{ + switch (amdgpu_aspm) { + case -1: + break; + case 0: + return false; + case 1: + return true; + default: + return false; + } + return pcie_aspm_enabled(adev->pdev); +} + /* if we get transitioned to only one device, take VGA back */ /** * amdgpu_device_vga_set_decode - enable/disable vga decode @@ -1446,7 +1472,8 @@ static int amdgpu_device_init_apu_flags(struct amdgpu_device *adev) case CHIP_YELLOW_CARP: break; case CHIP_CYAN_SKILLFISH: - if (adev->pdev->device == 0x13FE) + if ((adev->pdev->device == 0x13FE) || + (adev->pdev->device == 0x143F)) adev->apu_flags |= AMD_APU_IS_CYAN_SKILLFISH2; break; default: @@ -1507,6 +1534,11 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev) amdgpu_sched_hw_submission = roundup_pow_of_two(amdgpu_sched_hw_submission); } + if (amdgpu_reset_method < -1 || amdgpu_reset_method > 4) { + dev_warn(adev->dev, "invalid option for reset method, reverting to default\n"); + amdgpu_reset_method = -1; + } + amdgpu_device_check_smu_prv_buffer_size(adev); amdgpu_device_check_vm_size(adev); @@ -1517,7 +1549,6 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev) amdgpu_gmc_tmz_set(adev); - amdgpu_gmc_noretry_set(adev); return 0; } @@ -1955,27 +1986,9 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev) case CHIP_ARCTURUS: chip_name = "arcturus"; break; - case CHIP_RENOIR: - if (adev->apu_flags & AMD_APU_IS_RENOIR) - chip_name = "renoir"; - else - chip_name = "green_sardine"; - break; - case CHIP_NAVI10: - chip_name = "navi10"; - break; - case CHIP_NAVI14: - chip_name = "navi14"; - break; case CHIP_NAVI12: chip_name = "navi12"; break; - case CHIP_VANGOGH: - chip_name = "vangogh"; - break; - case CHIP_YELLOW_CARP: - chip_name = "yellow_carp"; - break; } snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_gpu_info.bin", chip_name); @@ -2073,6 +2086,8 @@ out: */ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) { + struct drm_device *dev = adev_to_drm(adev); + struct pci_dev *parent; int i, r; amdgpu_device_enable_virtual_display(adev); @@ -2137,6 +2152,18 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) break; } + if (amdgpu_has_atpx() && + (amdgpu_is_atpx_hybrid() || + amdgpu_has_atpx_dgpu_power_cntl()) && + ((adev->flags & AMD_IS_APU) == 0) && + !pci_is_thunderbolt_attached(to_pci_dev(dev->dev))) + adev->flags |= AMD_IS_PX; + + if (!(adev->flags & AMD_IS_APU)) { + parent = pci_upstream_bridge(adev->pdev); + adev->has_pr3 = parent ? pci_pr3_present(parent) : false; + } + amdgpu_amdkfd_device_probe(adev); adev->pm.pp_feature = amdgpu_pp_feature_mask; @@ -2673,6 +2700,12 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) adev->ip_blocks[i].status.late_initialized = true; } + r = amdgpu_ras_late_init(adev); + if (r) { + DRM_ERROR("amdgpu_ras_late_init failed %d", r); + return r; + } + amdgpu_ras_set_error_query_ready(adev, true); amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE); @@ -2687,7 +2720,7 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) /* For passthrough configuration on arcturus and aldebaran, enable special handling SBR */ if (amdgpu_passthrough(adev) && ((adev->asic_type == CHIP_ARCTURUS && adev->gmc.xgmi.num_physical_nodes > 1)|| adev->asic_type == CHIP_ALDEBARAN )) - smu_handle_passthrough_sbr(&adev->smu, true); + amdgpu_dpm_handle_passthrough_sbr(adev, true); if (adev->gmc.xgmi.num_physical_nodes > 1) { mutex_lock(&mgpu_info.mutex); @@ -2771,11 +2804,11 @@ static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev) } } - amdgpu_amdkfd_suspend(adev, false); - amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE); amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE); + amdgpu_amdkfd_suspend(adev, false); + /* Workaroud for ASICs need to disable SMC first */ amdgpu_device_smu_fini_early(adev); @@ -2944,7 +2977,7 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) int i, r; if (adev->in_s0ix) - amdgpu_gfx_state_change_set(adev, sGpuChangeState_D3Entry); + amdgpu_dpm_gfx_state_change(adev, sGpuChangeState_D3Entry); for (i = adev->num_ip_blocks - 1; i >= 0; i--) { if (!adev->ip_blocks[i].status.valid) @@ -3370,9 +3403,9 @@ static void amdgpu_device_xgmi_reset_func(struct work_struct *__work) if (adev->asic_reset_res) goto fail; - if (adev->mmhub.ras_funcs && - adev->mmhub.ras_funcs->reset_ras_error_count) - adev->mmhub.ras_funcs->reset_ras_error_count(adev); + if (adev->mmhub.ras && adev->mmhub.ras->ras_block.hw_ops && + adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count) + adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count(adev); } else { task_barrier_full(&hive->tb); @@ -3558,8 +3591,10 @@ int amdgpu_device_init(struct amdgpu_device *adev, hash_init(adev->mn_hash); mutex_init(&adev->psp.mutex); mutex_init(&adev->notifier_lock); + mutex_init(&adev->pm.stable_pstate_ctx_lock); + mutex_init(&adev->benchmark_mutex); - amdgpu_device_init_apu_flags(adev); + amdgpu_device_init_apu_flags(adev); r = amdgpu_device_check_arguments(adev); if (r) @@ -3580,6 +3615,8 @@ int amdgpu_device_init(struct amdgpu_device *adev, INIT_LIST_HEAD(&adev->reset_list); + INIT_LIST_HEAD(&adev->ras_list); + INIT_DELAYED_WORK(&adev->delayed_init_work, amdgpu_device_delayed_init_work_handler); INIT_DELAYED_WORK(&adev->gfx.gfx_off_delay_work, @@ -3629,6 +3666,15 @@ int amdgpu_device_init(struct amdgpu_device *adev, if (amdgpu_mes && adev->asic_type >= CHIP_NAVI10) adev->enable_mes = true; + /* + * Reset domain needs to be present early, before XGMI hive discovered + * (if any) and intitialized to use reset sem and in_gpu reset flag + * early on during init and before calling to RREG32. + */ + adev->reset_domain = amdgpu_reset_create_reset_domain(SINGLE_DEVICE, "amdgpu-reset-dev"); + if (!adev->reset_domain) + return -ENOMEM; + /* detect hw virtualization here */ amdgpu_detect_virtualization(adev); @@ -3638,20 +3684,12 @@ int amdgpu_device_init(struct amdgpu_device *adev, return r; } - /* - * Reset domain needs to be present early, before XGMI hive discovered - * (if any) and intitialized to use reset sem and in_gpu reset flag - * early on during init. - */ - adev->reset_domain = amdgpu_reset_create_reset_domain(SINGLE_DEVICE ,"amdgpu-reset-dev"); - if (!adev->reset_domain) - return -ENOMEM; - /* early init functions */ r = amdgpu_device_ip_early_init(adev); if (r) return r; + amdgpu_gmc_noretry_set(adev); /* Need to get xgmi info early to decide the reset behavior*/ if (adev->gmc.xgmi.supported) { r = adev->gfxhub.funcs->get_xgmi_info(adev); @@ -3819,19 +3857,6 @@ fence_driver_init: } else adev->ucode_sysfs_en = true; - if ((amdgpu_testing & 1)) { - if (adev->accel_working) - amdgpu_test_moves(adev); - else - DRM_INFO("amdgpu: acceleration disabled, skipping move tests\n"); - } - if (amdgpu_benchmarking) { - if (adev->accel_working) - amdgpu_benchmark(adev, amdgpu_benchmarking); - else - DRM_INFO("amdgpu: acceleration disabled, skipping benchmarks\n"); - } - /* * Register gpu instance before amdgpu_device_enable_mgpu_fan_boost. * Otherwise the mgpu fan boost feature will be skipped due to the @@ -4117,7 +4142,7 @@ int amdgpu_device_resume(struct drm_device *dev, bool fbcon) return 0; if (adev->in_s0ix) - amdgpu_gfx_state_change_set(adev, sGpuChangeState_D0Entry); + amdgpu_dpm_gfx_state_change(adev, sGpuChangeState_D0Entry); /* post card */ if (amdgpu_device_need_post(adev)) { @@ -4420,7 +4445,9 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev, { int r; struct amdgpu_hive_info *hive = NULL; + int retry_limit = 0; +retry: amdgpu_amdkfd_pre_reset(adev); amdgpu_amdkfd_pre_reset(adev); @@ -4469,6 +4496,14 @@ error: } amdgpu_virt_release_full_gpu(adev, true); + if (AMDGPU_RETRY_SRIOV_RESET(r)) { + if (retry_limit < AMDGPU_MAX_RETRY_LIMIT) { + retry_limit++; + goto retry; + } else + DRM_ERROR("GPU reset retry is beyond the retry limit\n"); + } + return r; } @@ -4660,6 +4695,22 @@ int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, return r; } +static int amdgpu_reset_reg_dumps(struct amdgpu_device *adev) +{ + uint32_t reg_value; + int i; + + lockdep_assert_held(&adev->reset_domain->sem); + dump_stack(); + + for (i = 0; i < adev->num_regs; i++) { + reg_value = RREG32(adev->reset_dump_reg_list[i]); + trace_amdgpu_reset_reg_dumps(adev->reset_dump_reg_list[i], reg_value); + } + + return 0; +} + int amdgpu_do_asic_reset(struct list_head *device_list_handle, struct amdgpu_reset_context *reset_context) { @@ -4670,6 +4721,7 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle, /* Try reset handler method first */ tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device, reset_list); + amdgpu_reset_reg_dumps(tmp_adev); r = amdgpu_reset_perform_reset(tmp_adev, reset_context); /* If reset handler not implemented, continue; otherwise return */ if (r == -ENOSYS) @@ -4718,9 +4770,9 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle, if (!r && amdgpu_ras_intr_triggered()) { list_for_each_entry(tmp_adev, device_list_handle, reset_list) { - if (tmp_adev->mmhub.ras_funcs && - tmp_adev->mmhub.ras_funcs->reset_ras_error_count) - tmp_adev->mmhub.ras_funcs->reset_ras_error_count(tmp_adev); + if (tmp_adev->mmhub.ras && tmp_adev->mmhub.ras->ras_block.hw_ops && + tmp_adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count) + tmp_adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count(tmp_adev); } amdgpu_ras_intr_cleared(); @@ -5183,6 +5235,9 @@ skip_hw_reset: drm_helper_resume_force_mode(adev_to_drm(tmp_adev)); } + if (tmp_adev->asic_reset_res) + r = tmp_adev->asic_reset_res; + tmp_adev->asic_reset_res = 0; if (r) { @@ -5746,3 +5801,36 @@ void amdgpu_device_halt(struct amdgpu_device *adev) pci_disable_device(pdev); pci_wait_for_pending_transaction(pdev); } + +u32 amdgpu_device_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; +} + +void amdgpu_device_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); +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index 81bfee978b74..e4fcbb385a62 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -360,8 +360,11 @@ out: return r; } +static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev); + void amdgpu_discovery_fini(struct amdgpu_device *adev) { + amdgpu_discovery_sysfs_fini(adev); kfree(adev->mman.discovery_bin); adev->mman.discovery_bin = NULL; } @@ -382,6 +385,578 @@ static int amdgpu_discovery_validate_ip(const struct ip *ip) return 0; } +static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, + uint32_t *vcn_harvest_count) +{ + struct binary_header *bhdr; + struct ip_discovery_header *ihdr; + struct die_header *dhdr; + struct ip *ip; + uint16_t die_offset, ip_offset, num_dies, num_ips; + int i, j; + + bhdr = (struct binary_header *)adev->mman.discovery_bin; + ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin + + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); + num_dies = le16_to_cpu(ihdr->num_dies); + + /* scan harvest bit of all IP data structures */ + for (i = 0; i < num_dies; i++) { + die_offset = le16_to_cpu(ihdr->die_info[i].die_offset); + dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset); + num_ips = le16_to_cpu(dhdr->num_ips); + ip_offset = die_offset + sizeof(*dhdr); + + for (j = 0; j < num_ips; j++) { + ip = (struct ip *)(adev->mman.discovery_bin + ip_offset); + + if (amdgpu_discovery_validate_ip(ip)) + goto next_ip; + + if (le16_to_cpu(ip->harvest) == 1) { + switch (le16_to_cpu(ip->hw_id)) { + case VCN_HWID: + (*vcn_harvest_count)++; + if (ip->number_instance == 0) + adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN0; + else + adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN1; + break; + case DMU_HWID: + adev->harvest_ip_mask |= AMD_HARVEST_IP_DMU_MASK; + break; + default: + break; + } + } +next_ip: + ip_offset += sizeof(*ip) + 4 * (ip->num_base_address - 1); + } + } +} + +static void amdgpu_discovery_read_from_harvest_table(struct amdgpu_device *adev, + uint32_t *vcn_harvest_count) +{ + struct binary_header *bhdr; + struct harvest_table *harvest_info; + int i; + + bhdr = (struct binary_header *)adev->mman.discovery_bin; + harvest_info = (struct harvest_table *)(adev->mman.discovery_bin + + le16_to_cpu(bhdr->table_list[HARVEST_INFO].offset)); + for (i = 0; i < 32; i++) { + if (le16_to_cpu(harvest_info->list[i].hw_id) == 0) + break; + + switch (le16_to_cpu(harvest_info->list[i].hw_id)) { + case VCN_HWID: + (*vcn_harvest_count)++; + if (harvest_info->list[i].number_instance == 0) + adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN0; + else + adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN1; + break; + case DMU_HWID: + adev->harvest_ip_mask |= AMD_HARVEST_IP_DMU_MASK; + break; + default: + break; + } + } +} + +/* ================================================== */ + +struct ip_hw_instance { + struct kobject kobj; /* ip_discovery/die/#die/#hw_id/#instance/<attrs...> */ + + int hw_id; + u8 num_instance; + u8 major, minor, revision; + u8 harvest; + + int num_base_addresses; + u32 base_addr[]; +}; + +struct ip_hw_id { + struct kset hw_id_kset; /* ip_discovery/die/#die/#hw_id/, contains ip_hw_instance */ + int hw_id; +}; + +struct ip_die_entry { + struct kset ip_kset; /* ip_discovery/die/#die/, contains ip_hw_id */ + u16 num_ips; +}; + +/* -------------------------------------------------- */ + +struct ip_hw_instance_attr { + struct attribute attr; + ssize_t (*show)(struct ip_hw_instance *ip_hw_instance, char *buf); +}; + +static ssize_t hw_id_show(struct ip_hw_instance *ip_hw_instance, char *buf) +{ + return sysfs_emit(buf, "%d\n", ip_hw_instance->hw_id); +} + +static ssize_t num_instance_show(struct ip_hw_instance *ip_hw_instance, char *buf) +{ + return sysfs_emit(buf, "%d\n", ip_hw_instance->num_instance); +} + +static ssize_t major_show(struct ip_hw_instance *ip_hw_instance, char *buf) +{ + return sysfs_emit(buf, "%d\n", ip_hw_instance->major); +} + +static ssize_t minor_show(struct ip_hw_instance *ip_hw_instance, char *buf) +{ + return sysfs_emit(buf, "%d\n", ip_hw_instance->minor); +} + +static ssize_t revision_show(struct ip_hw_instance *ip_hw_instance, char *buf) +{ + return sysfs_emit(buf, "%d\n", ip_hw_instance->revision); +} + +static ssize_t harvest_show(struct ip_hw_instance *ip_hw_instance, char *buf) +{ + return sysfs_emit(buf, "0x%01X\n", ip_hw_instance->harvest); +} + +static ssize_t num_base_addresses_show(struct ip_hw_instance *ip_hw_instance, char *buf) +{ + return sysfs_emit(buf, "%d\n", ip_hw_instance->num_base_addresses); +} + +static ssize_t base_addr_show(struct ip_hw_instance *ip_hw_instance, char *buf) +{ + ssize_t res, at; + int ii; + + for (res = at = ii = 0; ii < ip_hw_instance->num_base_addresses; ii++) { + /* Here we satisfy the condition that, at + size <= PAGE_SIZE. + */ + if (at + 12 > PAGE_SIZE) + break; + res = sysfs_emit_at(buf, at, "0x%08X\n", + ip_hw_instance->base_addr[ii]); + if (res <= 0) + break; + at += res; + } + + return res < 0 ? res : at; +} + +static struct ip_hw_instance_attr ip_hw_attr[] = { + __ATTR_RO(hw_id), + __ATTR_RO(num_instance), + __ATTR_RO(major), + __ATTR_RO(minor), + __ATTR_RO(revision), + __ATTR_RO(harvest), + __ATTR_RO(num_base_addresses), + __ATTR_RO(base_addr), +}; + +static struct attribute *ip_hw_instance_attrs[ARRAY_SIZE(ip_hw_attr) + 1]; +ATTRIBUTE_GROUPS(ip_hw_instance); + +#define to_ip_hw_instance(x) container_of(x, struct ip_hw_instance, kobj) +#define to_ip_hw_instance_attr(x) container_of(x, struct ip_hw_instance_attr, attr) + +static ssize_t ip_hw_instance_attr_show(struct kobject *kobj, + struct attribute *attr, + char *buf) +{ + struct ip_hw_instance *ip_hw_instance = to_ip_hw_instance(kobj); + struct ip_hw_instance_attr *ip_hw_attr = to_ip_hw_instance_attr(attr); + + if (!ip_hw_attr->show) + return -EIO; + + return ip_hw_attr->show(ip_hw_instance, buf); +} + +static const struct sysfs_ops ip_hw_instance_sysfs_ops = { + .show = ip_hw_instance_attr_show, +}; + +static void ip_hw_instance_release(struct kobject *kobj) +{ + struct ip_hw_instance *ip_hw_instance = to_ip_hw_instance(kobj); + + kfree(ip_hw_instance); +} + +static struct kobj_type ip_hw_instance_ktype = { + .release = ip_hw_instance_release, + .sysfs_ops = &ip_hw_instance_sysfs_ops, + .default_groups = ip_hw_instance_groups, +}; + +/* -------------------------------------------------- */ + +#define to_ip_hw_id(x) container_of(to_kset(x), struct ip_hw_id, hw_id_kset) + +static void ip_hw_id_release(struct kobject *kobj) +{ + struct ip_hw_id *ip_hw_id = to_ip_hw_id(kobj); + + if (!list_empty(&ip_hw_id->hw_id_kset.list)) + DRM_ERROR("ip_hw_id->hw_id_kset is not empty"); + kfree(ip_hw_id); +} + +static struct kobj_type ip_hw_id_ktype = { + .release = ip_hw_id_release, + .sysfs_ops = &kobj_sysfs_ops, +}; + +/* -------------------------------------------------- */ + +static void die_kobj_release(struct kobject *kobj); +static void ip_disc_release(struct kobject *kobj); + +struct ip_die_entry_attribute { + struct attribute attr; + ssize_t (*show)(struct ip_die_entry *ip_die_entry, char *buf); +}; + +#define to_ip_die_entry_attr(x) container_of(x, struct ip_die_entry_attribute, attr) + +static ssize_t num_ips_show(struct ip_die_entry *ip_die_entry, char *buf) +{ + return sysfs_emit(buf, "%d\n", ip_die_entry->num_ips); +} + +/* If there are more ip_die_entry attrs, other than the number of IPs, + * we can make this intro an array of attrs, and then initialize + * ip_die_entry_attrs in a loop. + */ +static struct ip_die_entry_attribute num_ips_attr = + __ATTR_RO(num_ips); + +static struct attribute *ip_die_entry_attrs[] = { + &num_ips_attr.attr, + NULL, +}; +ATTRIBUTE_GROUPS(ip_die_entry); /* ip_die_entry_groups */ + +#define to_ip_die_entry(x) container_of(to_kset(x), struct ip_die_entry, ip_kset) + +static ssize_t ip_die_entry_attr_show(struct kobject *kobj, + struct attribute *attr, + char *buf) +{ + struct ip_die_entry_attribute *ip_die_entry_attr = to_ip_die_entry_attr(attr); + struct ip_die_entry *ip_die_entry = to_ip_die_entry(kobj); + + if (!ip_die_entry_attr->show) + return -EIO; + + return ip_die_entry_attr->show(ip_die_entry, buf); +} + +static void ip_die_entry_release(struct kobject *kobj) +{ + struct ip_die_entry *ip_die_entry = to_ip_die_entry(kobj); + + if (!list_empty(&ip_die_entry->ip_kset.list)) + DRM_ERROR("ip_die_entry->ip_kset is not empty"); + kfree(ip_die_entry); +} + +static const struct sysfs_ops ip_die_entry_sysfs_ops = { + .show = ip_die_entry_attr_show, +}; + +static struct kobj_type ip_die_entry_ktype = { + .release = ip_die_entry_release, + .sysfs_ops = &ip_die_entry_sysfs_ops, + .default_groups = ip_die_entry_groups, +}; + +static struct kobj_type die_kobj_ktype = { + .release = die_kobj_release, + .sysfs_ops = &kobj_sysfs_ops, +}; + +static struct kobj_type ip_discovery_ktype = { + .release = ip_disc_release, + .sysfs_ops = &kobj_sysfs_ops, +}; + +struct ip_discovery_top { + struct kobject kobj; /* ip_discovery/ */ + struct kset die_kset; /* ip_discovery/die/, contains ip_die_entry */ + struct amdgpu_device *adev; +}; + +static void die_kobj_release(struct kobject *kobj) +{ + struct ip_discovery_top *ip_top = container_of(to_kset(kobj), + struct ip_discovery_top, + die_kset); + if (!list_empty(&ip_top->die_kset.list)) + DRM_ERROR("ip_top->die_kset is not empty"); +} + +static void ip_disc_release(struct kobject *kobj) +{ + struct ip_discovery_top *ip_top = container_of(kobj, struct ip_discovery_top, + kobj); + struct amdgpu_device *adev = ip_top->adev; + + adev->ip_top = NULL; + kfree(ip_top); +} + +static int amdgpu_discovery_sysfs_ips(struct amdgpu_device *adev, + struct ip_die_entry *ip_die_entry, + const size_t _ip_offset, const int num_ips) +{ + int ii, jj, kk, res; + + DRM_DEBUG("num_ips:%d", num_ips); + + /* Find all IPs of a given HW ID, and add their instance to + * #die/#hw_id/#instance/<attributes> + */ + for (ii = 0; ii < HW_ID_MAX; ii++) { + struct ip_hw_id *ip_hw_id = NULL; + size_t ip_offset = _ip_offset; + + for (jj = 0; jj < num_ips; jj++) { + struct ip *ip; + struct ip_hw_instance *ip_hw_instance; + + ip = (struct ip *)(adev->mman.discovery_bin + ip_offset); + if (amdgpu_discovery_validate_ip(ip) || + le16_to_cpu(ip->hw_id) != ii) + goto next_ip; + + DRM_DEBUG("match:%d @ ip_offset:%zu", ii, ip_offset); + + /* We have a hw_id match; register the hw + * block if not yet registered. + */ + if (!ip_hw_id) { + ip_hw_id = kzalloc(sizeof(*ip_hw_id), GFP_KERNEL); + if (!ip_hw_id) + return -ENOMEM; + ip_hw_id->hw_id = ii; + + kobject_set_name(&ip_hw_id->hw_id_kset.kobj, "%d", ii); + ip_hw_id->hw_id_kset.kobj.kset = &ip_die_entry->ip_kset; + ip_hw_id->hw_id_kset.kobj.ktype = &ip_hw_id_ktype; + res = kset_register(&ip_hw_id->hw_id_kset); + if (res) { + DRM_ERROR("Couldn't register ip_hw_id kset"); + kfree(ip_hw_id); + return res; + } + if (hw_id_names[ii]) { + res = sysfs_create_link(&ip_die_entry->ip_kset.kobj, + &ip_hw_id->hw_id_kset.kobj, + hw_id_names[ii]); + if (res) { + DRM_ERROR("Couldn't create IP link %s in IP Die:%s\n", + hw_id_names[ii], + kobject_name(&ip_die_entry->ip_kset.kobj)); + } + } + } + + /* Now register its instance. + */ + ip_hw_instance = kzalloc(struct_size(ip_hw_instance, + base_addr, + ip->num_base_address), + GFP_KERNEL); + if (!ip_hw_instance) { + DRM_ERROR("no memory for ip_hw_instance"); + return -ENOMEM; + } + ip_hw_instance->hw_id = le16_to_cpu(ip->hw_id); /* == ii */ + ip_hw_instance->num_instance = ip->number_instance; + ip_hw_instance->major = ip->major; + ip_hw_instance->minor = ip->minor; + ip_hw_instance->revision = ip->revision; + ip_hw_instance->harvest = ip->harvest; + ip_hw_instance->num_base_addresses = ip->num_base_address; + + for (kk = 0; kk < ip_hw_instance->num_base_addresses; kk++) + ip_hw_instance->base_addr[kk] = ip->base_address[kk]; + + kobject_init(&ip_hw_instance->kobj, &ip_hw_instance_ktype); + ip_hw_instance->kobj.kset = &ip_hw_id->hw_id_kset; + res = kobject_add(&ip_hw_instance->kobj, NULL, + "%d", ip_hw_instance->num_instance); +next_ip: + ip_offset += sizeof(*ip) + 4 * (ip->num_base_address - 1); + } + } + + return 0; +} + +static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev) +{ + struct binary_header *bhdr; + struct ip_discovery_header *ihdr; + struct die_header *dhdr; + struct kset *die_kset = &adev->ip_top->die_kset; + u16 num_dies, die_offset, num_ips; + size_t ip_offset; + int ii, res; + + bhdr = (struct binary_header *)adev->mman.discovery_bin; + ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin + + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); + num_dies = le16_to_cpu(ihdr->num_dies); + + DRM_DEBUG("number of dies: %d\n", num_dies); + + for (ii = 0; ii < num_dies; ii++) { + struct ip_die_entry *ip_die_entry; + + die_offset = le16_to_cpu(ihdr->die_info[ii].die_offset); + dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset); + num_ips = le16_to_cpu(dhdr->num_ips); + ip_offset = die_offset + sizeof(*dhdr); + + /* Add the die to the kset. + * + * dhdr->die_id == ii, which was checked in + * amdgpu_discovery_reg_base_init(). + */ + + ip_die_entry = kzalloc(sizeof(*ip_die_entry), GFP_KERNEL); + if (!ip_die_entry) + return -ENOMEM; + + ip_die_entry->num_ips = num_ips; + + kobject_set_name(&ip_die_entry->ip_kset.kobj, "%d", le16_to_cpu(dhdr->die_id)); + ip_die_entry->ip_kset.kobj.kset = die_kset; + ip_die_entry->ip_kset.kobj.ktype = &ip_die_entry_ktype; + res = kset_register(&ip_die_entry->ip_kset); + if (res) { + DRM_ERROR("Couldn't register ip_die_entry kset"); + kfree(ip_die_entry); + return res; + } + + amdgpu_discovery_sysfs_ips(adev, ip_die_entry, ip_offset, num_ips); + } + + return 0; +} + +static int amdgpu_discovery_sysfs_init(struct amdgpu_device *adev) +{ + struct kset *die_kset; + int res, ii; + + adev->ip_top = kzalloc(sizeof(*adev->ip_top), GFP_KERNEL); + if (!adev->ip_top) + return -ENOMEM; + + adev->ip_top->adev = adev; + + res = kobject_init_and_add(&adev->ip_top->kobj, &ip_discovery_ktype, + &adev->dev->kobj, "ip_discovery"); + if (res) { + DRM_ERROR("Couldn't init and add ip_discovery/"); + goto Err; + } + + die_kset = &adev->ip_top->die_kset; + kobject_set_name(&die_kset->kobj, "%s", "die"); + die_kset->kobj.parent = &adev->ip_top->kobj; + die_kset->kobj.ktype = &die_kobj_ktype; + res = kset_register(&adev->ip_top->die_kset); + if (res) { + DRM_ERROR("Couldn't register die_kset"); + goto Err; + } + + for (ii = 0; ii < ARRAY_SIZE(ip_hw_attr); ii++) + ip_hw_instance_attrs[ii] = &ip_hw_attr[ii].attr; + ip_hw_instance_attrs[ii] = NULL; + + res = amdgpu_discovery_sysfs_recurse(adev); + + return res; +Err: + kobject_put(&adev->ip_top->kobj); + return res; +} + +/* -------------------------------------------------- */ + +#define list_to_kobj(el) container_of(el, struct kobject, entry) + +static void amdgpu_discovery_sysfs_ip_hw_free(struct ip_hw_id *ip_hw_id) +{ + struct list_head *el, *tmp; + struct kset *hw_id_kset; + + hw_id_kset = &ip_hw_id->hw_id_kset; + spin_lock(&hw_id_kset->list_lock); + list_for_each_prev_safe(el, tmp, &hw_id_kset->list) { + list_del_init(el); + spin_unlock(&hw_id_kset->list_lock); + /* kobject is embedded in ip_hw_instance */ + kobject_put(list_to_kobj(el)); + spin_lock(&hw_id_kset->list_lock); + } + spin_unlock(&hw_id_kset->list_lock); + kobject_put(&ip_hw_id->hw_id_kset.kobj); +} + +static void amdgpu_discovery_sysfs_die_free(struct ip_die_entry *ip_die_entry) +{ + struct list_head *el, *tmp; + struct kset *ip_kset; + + ip_kset = &ip_die_entry->ip_kset; + spin_lock(&ip_kset->list_lock); + list_for_each_prev_safe(el, tmp, &ip_kset->list) { + list_del_init(el); + spin_unlock(&ip_kset->list_lock); + amdgpu_discovery_sysfs_ip_hw_free(to_ip_hw_id(list_to_kobj(el))); + spin_lock(&ip_kset->list_lock); + } + spin_unlock(&ip_kset->list_lock); + kobject_put(&ip_die_entry->ip_kset.kobj); +} + +static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev) +{ + struct list_head *el, *tmp; + struct kset *die_kset; + + die_kset = &adev->ip_top->die_kset; + spin_lock(&die_kset->list_lock); + list_for_each_prev_safe(el, tmp, &die_kset->list) { + list_del_init(el); + spin_unlock(&die_kset->list_lock); + amdgpu_discovery_sysfs_die_free(to_ip_die_entry(list_to_kobj(el))); + spin_lock(&die_kset->list_lock); + } + spin_unlock(&die_kset->list_lock); + kobject_put(&adev->ip_top->die_kset.kobj); + kobject_put(&adev->ip_top->kobj); +} + +/* ================================================== */ + int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev) { struct binary_header *bhdr; @@ -492,6 +1067,8 @@ next_ip: } } + amdgpu_discovery_sysfs_init(adev); + return 0; } @@ -545,32 +1122,26 @@ int amdgpu_discovery_get_ip_version(struct amdgpu_device *adev, int hw_id, int n void amdgpu_discovery_harvest_ip(struct amdgpu_device *adev) { - struct binary_header *bhdr; - struct harvest_table *harvest_info; - int i, vcn_harvest_count = 0; - - bhdr = (struct binary_header *)adev->mman.discovery_bin; - harvest_info = (struct harvest_table *)(adev->mman.discovery_bin + - le16_to_cpu(bhdr->table_list[HARVEST_INFO].offset)); - - for (i = 0; i < 32; i++) { - if (le16_to_cpu(harvest_info->list[i].hw_id) == 0) - break; + int vcn_harvest_count = 0; - switch (le16_to_cpu(harvest_info->list[i].hw_id)) { - case VCN_HWID: - vcn_harvest_count++; - if (harvest_info->list[i].number_instance == 0) - adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN0; - else - adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN1; - break; - case DMU_HWID: - adev->harvest_ip_mask |= AMD_HARVEST_IP_DMU_MASK; - break; - default: - break; - } + /* + * Harvest table does not fit Navi1x and legacy GPUs, + * so read harvest bit per IP data structure to set + * harvest configuration. + */ + if (adev->ip_versions[GC_HWIP][0] < IP_VERSION(10, 2, 0)) { + if ((adev->pdev->device == 0x731E && + (adev->pdev->revision == 0xC6 || + adev->pdev->revision == 0xC7)) || + (adev->pdev->device == 0x7340 && + adev->pdev->revision == 0xC9) || + (adev->pdev->device == 0x7360 && + adev->pdev->revision == 0xC7)) + amdgpu_discovery_read_harvest_bit_per_ip(adev, + &vcn_harvest_count); + } else { + amdgpu_discovery_read_from_harvest_table(adev, + &vcn_harvest_count); } amdgpu_discovery_harvest_config_quirk(adev); @@ -674,12 +1245,15 @@ static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 3): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): + case IP_VERSION(10, 3, 7): amdgpu_device_ip_block_add(adev, &nv_common_ip_block); break; default: @@ -709,12 +1283,15 @@ static int amdgpu_discovery_set_gmc_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 3): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): + case IP_VERSION(10, 3, 7): amdgpu_device_ip_block_add(adev, &gmc_v10_0_ip_block); break; default: @@ -790,6 +1367,8 @@ static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(13, 0, 1): case IP_VERSION(13, 0, 2): case IP_VERSION(13, 0, 3): + case IP_VERSION(13, 0, 5): + case IP_VERSION(13, 0, 8): amdgpu_device_ip_block_add(adev, &psp_v13_0_ip_block); break; default: @@ -831,6 +1410,8 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(13, 0, 1): case IP_VERSION(13, 0, 2): case IP_VERSION(13, 0, 3): + case IP_VERSION(13, 0, 5): + case IP_VERSION(13, 0, 8): amdgpu_device_ip_block_add(adev, &smu_v13_0_ip_block); break; default: @@ -846,8 +1427,14 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) { if (adev->enable_virtual_display || amdgpu_sriov_vf(adev)) { amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block); + return 0; + } + + if (!amdgpu_device_has_dc_support(adev)) + return 0; + #if defined(CONFIG_DRM_AMD_DC) - } else if (adev->ip_versions[DCE_HWIP][0]) { + if (adev->ip_versions[DCE_HWIP][0]) { switch (adev->ip_versions[DCE_HWIP][0]) { case IP_VERSION(1, 0, 0): case IP_VERSION(1, 0, 1): @@ -861,6 +1448,8 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(3, 0, 1): case IP_VERSION(3, 1, 2): case IP_VERSION(3, 1, 3): + case IP_VERSION(3, 1, 5): + case IP_VERSION(3, 1, 6): amdgpu_device_ip_block_add(adev, &dm_ip_block); break; default: @@ -882,8 +1471,8 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) adev->ip_versions[DCI_HWIP][0]); return -EINVAL; } -#endif } +#endif return 0; } @@ -904,12 +1493,15 @@ static int amdgpu_discovery_set_gc_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): amdgpu_device_ip_block_add(adev, &gfx_v10_0_ip_block); break; default: @@ -944,8 +1536,10 @@ static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(5, 2, 2): case IP_VERSION(5, 2, 4): case IP_VERSION(5, 2, 5): + case IP_VERSION(5, 2, 6): case IP_VERSION(5, 2, 3): case IP_VERSION(5, 2, 1): + case IP_VERSION(5, 2, 7): amdgpu_device_ip_block_add(adev, &sdma_v5_2_ip_block); break; default: @@ -1012,6 +1606,7 @@ static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(3, 0, 0): case IP_VERSION(3, 0, 16): case IP_VERSION(3, 1, 1): + case IP_VERSION(3, 1, 2): case IP_VERSION(3, 0, 2): case IP_VERSION(3, 0, 192): amdgpu_device_ip_block_add(adev, &vcn_v3_0_ip_block); @@ -1038,12 +1633,14 @@ static int amdgpu_discovery_set_mes_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 3): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): amdgpu_device_ip_block_add(adev, &mes_v10_1_ip_block); break; default: @@ -1217,11 +1814,6 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) return -EINVAL; amdgpu_discovery_harvest_ip(adev); - - if (!adev->mman.discovery_bin) { - DRM_ERROR("ip discovery uninitialized\n"); - return -EINVAL; - } break; } @@ -1242,6 +1834,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 4): @@ -1254,10 +1847,32 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(10, 3, 3): adev->family = AMDGPU_FAMILY_YC; break; + case IP_VERSION(10, 3, 6): + adev->family = AMDGPU_FAMILY_GC_10_3_6; + break; + case IP_VERSION(10, 3, 7): + adev->family = AMDGPU_FAMILY_GC_10_3_7; + break; default: return -EINVAL; } + switch (adev->ip_versions[GC_HWIP][0]) { + case IP_VERSION(9, 1, 0): + case IP_VERSION(9, 2, 2): + case IP_VERSION(9, 3, 0): + case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): + case IP_VERSION(10, 3, 1): + case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 6): + case IP_VERSION(10, 3, 7): + adev->flags |= AMD_IS_APU; + break; + default: + break; + } + if (adev->ip_versions[XGMI_HWIP][0] == IP_VERSION(4, 8, 0)) adev->gmc.xgmi.supported = true; @@ -1285,7 +1900,9 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) break; case IP_VERSION(7, 2, 0): case IP_VERSION(7, 2, 1): + case IP_VERSION(7, 3, 0): case IP_VERSION(7, 5, 0): + case IP_VERSION(7, 5, 1): adev->nbio.funcs = &nbio_v7_2_funcs; adev->nbio.hdp_flush_reg = &nbio_v7_2_hdp_flush_reg; break; @@ -1368,6 +1985,8 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(11, 0, 11): case IP_VERSION(11, 5, 0): case IP_VERSION(13, 0, 1): + case IP_VERSION(13, 0, 9): + case IP_VERSION(13, 0, 10): adev->smuio.funcs = &smuio_v11_0_6_funcs; break; case IP_VERSION(13, 0, 2): diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index e76b96d55551..fae5c1debfad 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -506,28 +506,9 @@ uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev, */ if ((bo_flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) && amdgpu_bo_support_uswc(bo_flags) && - amdgpu_device_asic_has_dc_support(adev->asic_type)) { - switch (adev->asic_type) { - case CHIP_CARRIZO: - case CHIP_STONEY: - domain |= AMDGPU_GEM_DOMAIN_GTT; - break; - case CHIP_RAVEN: - /* enable S/G on PCO and RV2 */ - if ((adev->apu_flags & AMD_APU_IS_RAVEN2) || - (adev->apu_flags & AMD_APU_IS_PICASSO)) - domain |= AMDGPU_GEM_DOMAIN_GTT; - break; - case CHIP_RENOIR: - case CHIP_VANGOGH: - case CHIP_YELLOW_CARP: - domain |= AMDGPU_GEM_DOMAIN_GTT; - break; - - default: - break; - } - } + amdgpu_device_asic_has_dc_support(adev->asic_type) && + adev->mode_info.gpu_vm_support) + domain |= AMDGPU_GEM_DOMAIN_GTT; #endif return domain; @@ -710,9 +691,9 @@ static int convert_tiling_flags_to_modifier(struct amdgpu_framebuffer *afb) return -EINVAL; } - if (adev->asic_type >= CHIP_SIENNA_CICHLID) + if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) version = AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS; - else if (adev->family == AMDGPU_FAMILY_NV) + else if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 0, 0)) version = AMD_FMT_MOD_TILE_VER_GFX10; else version = AMD_FMT_MOD_TILE_VER_GFX9; @@ -806,7 +787,7 @@ static int convert_tiling_flags_to_modifier(struct amdgpu_framebuffer *afb) if (adev->family >= AMDGPU_FAMILY_NV) { int extra_pipe = 0; - if (adev->asic_type >= CHIP_SIENNA_CICHLID && + if ((adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) && pipes == packers && pipes > 1) extra_pipe = 1; @@ -1143,7 +1124,7 @@ int amdgpu_display_framebuffer_init(struct drm_device *dev, if (ret) return ret; - if (dev->mode_config.fb_modifiers_not_supported) { + if (dev->mode_config.fb_modifiers_not_supported && !adev->enable_virtual_display) { drm_WARN_ONCE(dev, adev->family >= AMDGPU_FAMILY_AI, "GFX9+ requires FB check based on format modifier\n"); ret = check_tiling_flags_gfx6(rfb); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 4c83f1db8a24..bb1c025d9001 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -99,9 +99,11 @@ * - 3.42.0 - Add 16bpc fixed point display support * - 3.43.0 - Add device hot plug/unplug support * - 3.44.0 - DCN3 supports DCC independent block settings: !64B && 128B, 64B && 128B + * - 3.45.0 - Add context ioctl stable pstate interface + * * 3.46.0 - To enable hot plug amdgpu tests in libdrm */ #define KMS_DRIVER_MAJOR 3 -#define KMS_DRIVER_MINOR 44 +#define KMS_DRIVER_MINOR 46 #define KMS_DRIVER_PATCHLEVEL 0 int amdgpu_vram_limit; @@ -109,8 +111,6 @@ 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; -int amdgpu_testing; int amdgpu_audio = -1; int amdgpu_disp_priority; int amdgpu_hw_i2c; @@ -174,10 +174,11 @@ int amdgpu_mes; int amdgpu_noretry = -1; int amdgpu_force_asic_type = -1; int amdgpu_tmz = -1; /* auto */ -uint amdgpu_freesync_vid_mode; int amdgpu_reset_method = -1; /* auto */ int amdgpu_num_kcq = -1; int amdgpu_smartshift_bias; +int amdgpu_use_xgmi_p2p = 1; +int amdgpu_vcnfw_log; static void amdgpu_drv_delayed_reset_work_handler(struct work_struct *work); @@ -232,20 +233,6 @@ MODULE_PARM_DESC(moverate, "Maximum buffer migration rate in MB/s. (32, 64, etc. module_param_named(moverate, amdgpu_moverate, int, 0600); /** - * DOC: benchmark (int) - * Run benchmarks. The default is 0 (Skip benchmarks). - */ -MODULE_PARM_DESC(benchmark, "Run benchmark"); -module_param_named(benchmark, amdgpu_benchmarking, int, 0444); - -/** - * DOC: test (int) - * Test BO GTT->VRAM and VRAM->GTT GPU copies. The default is 0 (Skip test, only set 1 to run test). - */ -MODULE_PARM_DESC(test, "Run tests"); -module_param_named(test, amdgpu_testing, int, 0444); - -/** * DOC: audio (int) * Set HDMI/DPAudio. Only affects non-DC display handling. The default is -1 (Enabled), set 0 to disabled it. */ @@ -667,6 +654,13 @@ MODULE_PARM_DESC(force_asic_type, "A non negative value used to specify the asic type for all supported GPUs"); module_param_named(force_asic_type, amdgpu_force_asic_type, int, 0444); +/** + * DOC: use_xgmi_p2p (int) + * Enables/disables XGMI P2P interface (0 = disable, 1 = enable). + */ +MODULE_PARM_DESC(use_xgmi_p2p, + "Enable XGMI P2P interface (0 = disable; 1 = enable (default))"); +module_param_named(use_xgmi_p2p, amdgpu_use_xgmi_p2p, int, 0444); #ifdef CONFIG_HSA_AMD @@ -740,7 +734,7 @@ MODULE_PARM_DESC(debug_largebar, * systems with a broken CRAT table. * * Default is auto (according to asic type, iommu_v2, and crat table, to decide - * whehter use CRAT) + * whether use CRAT) */ int ignore_crat; module_param(ignore_crat, int, 0444); @@ -844,36 +838,10 @@ MODULE_PARM_DESC(tmz, "Enable TMZ feature (-1 = auto (default), 0 = off, 1 = on) module_param_named(tmz, amdgpu_tmz, int, 0444); /** - * DOC: freesync_video (uint) - * Enable the optimization to adjust front porch timing to achieve seamless - * mode change experience when setting a freesync supported mode for which full - * modeset is not needed. - * - * The Display Core will add a set of modes derived from the base FreeSync - * video mode into the corresponding connector's mode list based on commonly - * used refresh rates and VRR range of the connected display, when users enable - * this feature. From the userspace perspective, they can see a seamless mode - * change experience when the change between different refresh rates under the - * same resolution. Additionally, userspace applications such as Video playback - * can read this modeset list and change the refresh rate based on the video - * frame rate. Finally, the userspace can also derive an appropriate mode for a - * particular refresh rate based on the FreeSync Mode and add it to the - * connector's mode list. - * - * Note: This is an experimental feature. - * - * The default value: 0 (off). - */ -MODULE_PARM_DESC( - freesync_video, - "Enable freesync modesetting optimization feature (0 = off (default), 1 = on)"); -module_param_named(freesync_video, amdgpu_freesync_vid_mode, uint, 0444); - -/** * DOC: reset_method (int) - * GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco, 5 = pci) + * GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco) */ -MODULE_PARM_DESC(reset_method, "GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco/bamaco, 5 = pci)"); +MODULE_PARM_DESC(reset_method, "GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco/bamaco)"); module_param_named(reset_method, amdgpu_reset_method, int, 0444); /** @@ -889,6 +857,13 @@ MODULE_PARM_DESC(num_kcq, "number of kernel compute queue user want to setup (8 module_param_named(num_kcq, amdgpu_num_kcq, int, 0444); /** + * DOC: vcnfw_log (int) + * Enable vcnfw log output for debugging, the default is disabled. + */ +MODULE_PARM_DESC(vcnfw_log, "Enable vcnfw log(0 = disable (default value), 1 = enable)"); +module_param_named(vcnfw_log, amdgpu_vcnfw_log, int, 0444); + +/** * DOC: smu_pptable_id (int) * Used to override pptable id. id = 0 use VBIOS pptable. * id > 0 use the soft pptable with specicfied id. @@ -1942,13 +1917,14 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x73FF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_DIMGREY_CAVEFISH}, /* Aldebaran */ - {0x1002, 0x7408, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN|AMD_EXP_HW_SUPPORT}, - {0x1002, 0x740C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN|AMD_EXP_HW_SUPPORT}, - {0x1002, 0x740F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN|AMD_EXP_HW_SUPPORT}, - {0x1002, 0x7410, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN|AMD_EXP_HW_SUPPORT}, + {0x1002, 0x7408, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN}, + {0x1002, 0x740C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN}, + {0x1002, 0x740F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN}, + {0x1002, 0x7410, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN}, /* CYAN_SKILLFISH */ {0x1002, 0x13FE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU}, + {0x1002, 0x143F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU}, /* BEIGE_GOBY */ {0x1002, 0x7420, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BEIGE_GOBY}, @@ -1994,6 +1970,28 @@ static bool amdgpu_is_fw_framebuffer(resource_size_t base, return found; } +static void amdgpu_get_secondary_funcs(struct amdgpu_device *adev) +{ + struct pci_dev *p = NULL; + int i; + + /* 0 - GPU + * 1 - audio + * 2 - USB + * 3 - UCSI + */ + for (i = 1; i < 4; i++) { + p = pci_get_domain_bus_and_slot(pci_domain_nr(adev->pdev->bus), + adev->pdev->bus->number, i); + if (p) { + pm_runtime_get_sync(&p->dev); + pm_runtime_mark_last_busy(&p->dev); + pm_runtime_put_autosuspend(&p->dev); + pci_dev_put(p); + } + } +} + static int amdgpu_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) { @@ -2011,6 +2009,9 @@ static int amdgpu_pci_probe(struct pci_dev *pdev, return -ENODEV; } + if (amdgpu_aspm == -1 && !pcie_aspm_enabled(pdev)) + amdgpu_aspm = 0; + if (amdgpu_virtual_display || amdgpu_device_asic_has_dc_support(flags & AMD_ASIC_MASK)) supports_atomic = true; @@ -2126,6 +2127,48 @@ retry_init: if (ret) DRM_ERROR("Creating debugfs files failed (%d).\n", ret); + if (adev->runpm) { + /* only need to skip on ATPX */ + if (amdgpu_device_supports_px(ddev)) + dev_pm_set_driver_flags(ddev->dev, DPM_FLAG_NO_DIRECT_COMPLETE); + /* we want direct complete for BOCO */ + if (amdgpu_device_supports_boco(ddev)) + dev_pm_set_driver_flags(ddev->dev, DPM_FLAG_SMART_PREPARE | + DPM_FLAG_SMART_SUSPEND | + DPM_FLAG_MAY_SKIP_RESUME); + pm_runtime_use_autosuspend(ddev->dev); + pm_runtime_set_autosuspend_delay(ddev->dev, 5000); + + pm_runtime_allow(ddev->dev); + + pm_runtime_mark_last_busy(ddev->dev); + pm_runtime_put_autosuspend(ddev->dev); + + /* + * For runpm implemented via BACO, PMFW will handle the + * timing for BACO in and out: + * - put ASIC into BACO state only when both video and + * audio functions are in D3 state. + * - pull ASIC out of BACO state when either video or + * audio function is in D0 state. + * Also, at startup, PMFW assumes both functions are in + * D0 state. + * + * So if snd driver was loaded prior to amdgpu driver + * and audio function was put into D3 state, there will + * be no PMFW-aware D-state transition(D0->D3) on runpm + * suspend. Thus the BACO will be not correctly kicked in. + * + * Via amdgpu_get_secondary_funcs(), the audio dev is put + * into D0 state. Then there will be a PMFW-aware D-state + * transition(D0->D3) on runpm suspend. + */ + if (amdgpu_device_supports_baco(ddev) && + !(adev->flags & AMD_IS_APU) && + (adev->asic_type >= CHIP_NAVI10)) + amdgpu_get_secondary_funcs(adev); + } + return 0; err_pci: @@ -2137,8 +2180,15 @@ static void amdgpu_pci_remove(struct pci_dev *pdev) { struct drm_device *dev = pci_get_drvdata(pdev); + struct amdgpu_device *adev = drm_to_adev(dev); drm_dev_unplug(dev); + + if (adev->runpm) { + pm_runtime_get_sync(dev->dev); + pm_runtime_forbid(dev->dev); + } + amdgpu_driver_unload_kms(dev); /* @@ -2246,13 +2296,20 @@ static void amdgpu_drv_delayed_reset_work_handler(struct work_struct *work) static int amdgpu_pmops_prepare(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); + struct amdgpu_device *adev = drm_to_adev(drm_dev); /* Return a positive number here so * DPM_FLAG_SMART_SUSPEND works properly */ if (amdgpu_device_supports_boco(drm_dev)) - return pm_runtime_suspended(dev) && - pm_suspend_via_firmware(); + return pm_runtime_suspended(dev); + + /* if we will not support s3 or s2i for the device + * then skip suspend + */ + if (!amdgpu_acpi_is_s0ix_active(adev) && + !amdgpu_acpi_is_s3_active(adev)) + return 1; return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c index 2a786e788627..ecada5eadfe3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c @@ -30,7 +30,6 @@ #include "amdgpu_eeprom.h" #define FRU_EEPROM_MADDR 0x60000 -#define I2C_PRODUCT_INFO_OFFSET 0xC0 static bool is_fru_eeprom_supported(struct amdgpu_device *adev) { @@ -40,7 +39,13 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev) */ struct atom_context *atom_ctx = adev->mode_info.atom_context; - /* VBIOS is of the format ###-DXXXYY-##. For SKU identification, + /* The i2c access is blocked on VF + * TODO: Need other way to get the info + */ + if (amdgpu_sriov_vf(adev)) + return false; + + /* VBIOS is of the format ###-DXXXYYYY-##. For SKU identification, * we can use just the "DXXX" portion. If there were more models, we * could convert the 3 characters to a hex integer and use a switch * for ease/speed/readability. For now, 2 string comparisons are @@ -59,17 +64,24 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev) case CHIP_ALDEBARAN: /* All Aldebaran SKUs have the FRU */ return true; + case CHIP_SIENNA_CICHLID: + if (strnstr(atom_ctx->vbios_version, "D603", + sizeof(atom_ctx->vbios_version))) + return true; + else + return false; default: return false; } } static int amdgpu_fru_read_eeprom(struct amdgpu_device *adev, uint32_t addrptr, - unsigned char *buff) + unsigned char *buf, size_t buf_size) { - int ret, size; + int ret; + u8 size; - ret = amdgpu_eeprom_read(&adev->pm.smu_i2c, addrptr, buff, 1); + ret = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addrptr, buf, 1); if (ret < 1) { DRM_WARN("FRU: Failed to get size field"); return ret; @@ -78,9 +90,11 @@ static int amdgpu_fru_read_eeprom(struct amdgpu_device *adev, uint32_t addrptr, /* The size returned by the i2c requires subtraction of 0xC0 since the * size apparently always reports as 0xC0+actual size. */ - size = buff[0] - I2C_PRODUCT_INFO_OFFSET; + size = buf[0] & 0x3F; + size = min_t(size_t, size, buf_size); - ret = amdgpu_eeprom_read(&adev->pm.smu_i2c, addrptr + 1, buff, size); + ret = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addrptr + 1, + buf, size); if (ret < 1) { DRM_WARN("FRU: Failed to get data field"); return ret; @@ -91,19 +105,15 @@ static int amdgpu_fru_read_eeprom(struct amdgpu_device *adev, uint32_t addrptr, int amdgpu_fru_get_product_info(struct amdgpu_device *adev) { - unsigned char buff[AMDGPU_PRODUCT_NAME_LEN+2]; + unsigned char buf[AMDGPU_PRODUCT_NAME_LEN]; u32 addrptr; int size, len; - int offset = 2; if (!is_fru_eeprom_supported(adev)) return 0; - if (adev->asic_type == CHIP_ALDEBARAN) - offset = 0; - /* If algo exists, it means that the i2c_adapter's initialized */ - if (!adev->pm.smu_i2c.algo) { + if (!adev->pm.fru_eeprom_i2c_bus || !adev->pm.fru_eeprom_i2c_bus->algo) { DRM_WARN("Cannot access FRU, EEPROM accessor not initialized"); return -ENODEV; } @@ -121,7 +131,7 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) * and the language field, so just start from 0xb, manufacturer size */ addrptr = FRU_EEPROM_MADDR + 0xb; - size = amdgpu_fru_read_eeprom(adev, addrptr, buff); + size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); if (size < 1) { DRM_ERROR("Failed to read FRU Manufacturer, ret:%d", size); return -EINVAL; @@ -131,7 +141,7 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) * size field being 1 byte. This pattern continues below. */ addrptr += size + 1; - size = amdgpu_fru_read_eeprom(adev, addrptr, buff); + size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); if (size < 1) { DRM_ERROR("Failed to read FRU product name, ret:%d", size); return -EINVAL; @@ -143,12 +153,11 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) AMDGPU_PRODUCT_NAME_LEN); len = AMDGPU_PRODUCT_NAME_LEN - 1; } - /* Start at 2 due to buff using fields 0 and 1 for the address */ - memcpy(adev->product_name, &buff[offset], len); + memcpy(adev->product_name, buf, len); adev->product_name[len] = '\0'; addrptr += size + 1; - size = amdgpu_fru_read_eeprom(adev, addrptr, buff); + size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); if (size < 1) { DRM_ERROR("Failed to read FRU product number, ret:%d", size); return -EINVAL; @@ -162,11 +171,11 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) DRM_WARN("FRU Product Number is larger than 16 characters. This is likely a mistake"); len = sizeof(adev->product_number) - 1; } - memcpy(adev->product_number, &buff[offset], len); + memcpy(adev->product_number, buf, len); adev->product_number[len] = '\0'; addrptr += size + 1; - size = amdgpu_fru_read_eeprom(adev, addrptr, buff); + size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); if (size < 1) { DRM_ERROR("Failed to read FRU product version, ret:%d", size); @@ -174,7 +183,7 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) } addrptr += size + 1; - size = amdgpu_fru_read_eeprom(adev, addrptr, buff); + size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); if (size < 1) { DRM_ERROR("Failed to read FRU serial number, ret:%d", size); @@ -189,7 +198,7 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) DRM_WARN("FRU Serial Number is larger than 16 characters. This is likely a mistake"); len = sizeof(adev->serial) - 1; } - memcpy(adev->serial, &buff[offset], len); + memcpy(adev->serial, buf, len); adev->serial[len] = '\0'; return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c index 645950a653a0..01cb89ffbd56 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c @@ -150,7 +150,7 @@ void amdgpu_gart_table_vram_free(struct amdgpu_device *adev) * replaces them with the dummy page (all asics). * Returns 0 for success, -EINVAL for failure. */ -int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, +void amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, int pages) { unsigned t; @@ -161,13 +161,11 @@ int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, uint64_t flags = 0; int idx; - if (!adev->gart.ready) { - WARN(1, "trying to unbind memory from uninitialized GART !\n"); - return -EINVAL; - } + if (!adev->gart.ptr) + return; if (!drm_dev_enter(adev_to_drm(adev), &idx)) - return 0; + return; t = offset / AMDGPU_GPU_PAGE_SIZE; p = t / AMDGPU_GPU_PAGES_IN_CPU_PAGE; @@ -188,7 +186,6 @@ int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, amdgpu_gmc_flush_gpu_tlb(adev, 0, i, 0); drm_dev_exit(idx); - return 0; } /** @@ -204,7 +201,7 @@ int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, * Map the dma_addresses into GART entries (all asics). * Returns 0 for success, -EINVAL for failure. */ -int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset, +void amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset, int pages, dma_addr_t *dma_addr, uint64_t flags, void *dst) { @@ -212,13 +209,8 @@ int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset, unsigned i, j, t; int idx; - if (!adev->gart.ready) { - WARN(1, "trying to bind memory to uninitialized GART !\n"); - return -EINVAL; - } - if (!drm_dev_enter(adev_to_drm(adev), &idx)) - return 0; + return; t = offset / AMDGPU_GPU_PAGE_SIZE; @@ -230,7 +222,6 @@ int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset, } } drm_dev_exit(idx); - return 0; } /** @@ -246,20 +237,14 @@ int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset, * (all asics). * Returns 0 for success, -EINVAL for failure. */ -int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset, +void amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset, int pages, dma_addr_t *dma_addr, uint64_t flags) { - if (!adev->gart.ready) { - WARN(1, "trying to bind memory to uninitialized GART !\n"); - return -EINVAL; - } - if (!adev->gart.ptr) - return 0; + return; - return amdgpu_gart_map(adev, offset, pages, dma_addr, flags, - adev->gart.ptr); + amdgpu_gart_map(adev, offset, pages, dma_addr, flags, adev->gart.ptr); } /** @@ -274,6 +259,9 @@ void amdgpu_gart_invalidate_tlb(struct amdgpu_device *adev) { int i; + if (!adev->gart.ptr) + return; + mb(); amdgpu_device_flush_hdp(adev, NULL); for (i = 0; i < adev->num_vmhubs; i++) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h index 78895413cf9f..8fea3e04e411 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h @@ -46,7 +46,6 @@ struct amdgpu_gart { unsigned num_gpu_pages; unsigned num_cpu_pages; unsigned table_size; - bool ready; /* Asic default pte flags */ uint64_t gart_pte_flags; @@ -58,12 +57,12 @@ int amdgpu_gart_table_vram_pin(struct amdgpu_device *adev); void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev); int amdgpu_gart_init(struct amdgpu_device *adev); void amdgpu_gart_dummy_page_fini(struct amdgpu_device *adev); -int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, - int pages); -int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset, - int pages, dma_addr_t *dma_addr, uint64_t flags, - void *dst); -int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset, - int pages, dma_addr_t *dma_addr, uint64_t flags); +void amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, + int pages); +void amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset, + int pages, dma_addr_t *dma_addr, uint64_t flags, + void *dst); +void amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset, + int pages, dma_addr_t *dma_addr, uint64_t flags); void amdgpu_gart_invalidate_tlb(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 9b12cab5e606..57b74d35052f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -222,7 +222,7 @@ static void amdgpu_gem_object_close(struct drm_gem_object *obj, if (!bo_va || --bo_va->ref_count) goto out_unlock; - amdgpu_vm_bo_rmv(adev, bo_va); + amdgpu_vm_bo_del(adev, bo_va); if (!amdgpu_vm_ready(vm)) goto out_unlock; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 1916ec84dd71..8fe939976224 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -615,74 +615,35 @@ int amdgpu_get_gfx_off_status(struct amdgpu_device *adev, uint32_t *value) mutex_lock(&adev->gfx.gfx_off_mutex); - r = smu_get_status_gfxoff(adev, value); + r = amdgpu_dpm_get_status_gfxoff(adev, value); mutex_unlock(&adev->gfx.gfx_off_mutex); return r; } -int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev) +int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block) { int r; - struct ras_fs_if fs_info = { - .sysfs_name = "gfx_err_count", - }; - struct ras_ih_if ih_info = { - .cb = amdgpu_gfx_process_ras_data_cb, - }; - - if (!adev->gfx.ras_if) { - adev->gfx.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); - if (!adev->gfx.ras_if) - return -ENOMEM; - adev->gfx.ras_if->block = AMDGPU_RAS_BLOCK__GFX; - adev->gfx.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; - adev->gfx.ras_if->sub_block_index = 0; - } - fs_info.head = ih_info.head = *adev->gfx.ras_if; - r = amdgpu_ras_late_init(adev, adev->gfx.ras_if, - &fs_info, &ih_info); + r = amdgpu_ras_block_late_init(adev, ras_block); if (r) - goto free; + return r; - if (amdgpu_ras_is_supported(adev, adev->gfx.ras_if->block)) { + if (amdgpu_ras_is_supported(adev, ras_block->block)) { if (!amdgpu_persistent_edc_harvesting_supported(adev)) amdgpu_ras_reset_error_status(adev, AMDGPU_RAS_BLOCK__GFX); r = amdgpu_irq_get(adev, &adev->gfx.cp_ecc_error_irq, 0); if (r) goto late_fini; - } else { - /* free gfx ras_if if ras is not supported */ - r = 0; - goto free; } return 0; late_fini: - amdgpu_ras_late_fini(adev, adev->gfx.ras_if, &ih_info); -free: - kfree(adev->gfx.ras_if); - adev->gfx.ras_if = NULL; + amdgpu_ras_block_late_fini(adev, ras_block); return r; } -void amdgpu_gfx_ras_fini(struct amdgpu_device *adev) -{ - if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX) && - adev->gfx.ras_if) { - struct ras_common_if *ras_if = adev->gfx.ras_if; - struct ras_ih_if ih_info = { - .head = *ras_if, - .cb = amdgpu_gfx_process_ras_data_cb, - }; - - amdgpu_ras_late_fini(adev, ras_if, &ih_info); - kfree(ras_if); - } -} - int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev, void *err_data, struct amdgpu_iv_entry *entry) @@ -695,9 +656,9 @@ int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev, */ if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) { kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->query_ras_error_count) - adev->gfx.ras_funcs->query_ras_error_count(adev, err_data); + if (adev->gfx.ras && adev->gfx.ras->ras_block.hw_ops && + adev->gfx.ras->ras_block.hw_ops->query_ras_error_count) + adev->gfx.ras->ras_block.hw_ops->query_ras_error_count(adev, err_data); amdgpu_ras_reset_gpu(adev); } return AMDGPU_RAS_SUCCESS; @@ -852,19 +813,3 @@ int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev) } return amdgpu_num_kcq; } - -/* amdgpu_gfx_state_change_set - Handle gfx power state change set - * @adev: amdgpu_device pointer - * @state: gfx power state(1 -sGpuChangeState_D0Entry and 2 -sGpuChangeState_D3Entry) - * - */ - -void amdgpu_gfx_state_change_set(struct amdgpu_device *adev, enum gfx_change_state state) -{ - mutex_lock(&adev->pm.mutex); - if (adev->powerplay.pp_funcs && - adev->powerplay.pp_funcs->gfx_state_change_set) - ((adev)->powerplay.pp_funcs->gfx_state_change_set( - (adev)->powerplay.pp_handle, state)); - mutex_unlock(&adev->pm.mutex); -} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index f851196c83a5..dcb3c7871c73 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h @@ -31,6 +31,7 @@ #include "amdgpu_ring.h" #include "amdgpu_rlc.h" #include "soc15.h" +#include "amdgpu_ras.h" /* GFX current status */ #define AMDGPU_GFX_NORMAL_MODE 0x00000000L @@ -47,12 +48,6 @@ enum amdgpu_gfx_pipe_priority { AMDGPU_GFX_PIPE_PRIO_HIGH = AMDGPU_RING_PRIO_2 }; -/* Argument for PPSMC_MSG_GpuChangeState */ -enum gfx_change_state { - sGpuChangeState_D0Entry = 1, - sGpuChangeState_D3Entry, -}; - #define AMDGPU_GFX_QUEUE_PRIORITY_MINIMUM 0 #define AMDGPU_GFX_QUEUE_PRIORITY_MAXIMUM 15 @@ -204,16 +199,8 @@ struct amdgpu_cu_info { uint32_t bitmap[4][4]; }; -struct amdgpu_gfx_ras_funcs { - int (*ras_late_init)(struct amdgpu_device *adev); - void (*ras_fini)(struct amdgpu_device *adev); - int (*ras_error_inject)(struct amdgpu_device *adev, - void *inject_if); - int (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); - void (*reset_ras_error_count)(struct amdgpu_device *adev); - void (*query_ras_error_status)(struct amdgpu_device *adev); - void (*reset_ras_error_status)(struct amdgpu_device *adev); +struct amdgpu_gfx_ras { + struct amdgpu_ras_block_object ras_block; void (*enable_watchdog_timer)(struct amdgpu_device *adev); }; @@ -337,7 +324,7 @@ struct amdgpu_gfx { /*ras */ struct ras_common_if *ras_if; - const struct amdgpu_gfx_ras_funcs *ras_funcs; + struct amdgpu_gfx_ras *ras; }; #define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev)) @@ -399,8 +386,7 @@ bool amdgpu_gfx_is_me_queue_enabled(struct amdgpu_device *adev, int me, int pipe, int queue); void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable); int amdgpu_get_gfx_off_status(struct amdgpu_device *adev, uint32_t *value); -int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev); -void amdgpu_gfx_ras_fini(struct amdgpu_device *adev); +int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block); int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev, void *err_data, struct amdgpu_iv_entry *entry); @@ -410,5 +396,4 @@ int amdgpu_gfx_cp_ecc_error_irq(struct amdgpu_device *adev, 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); -void amdgpu_gfx_state_change_set(struct amdgpu_device *adev, enum gfx_change_state state); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 2430d6223c2d..ca2cfb65f976 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -436,82 +436,25 @@ void amdgpu_gmc_filter_faults_remove(struct amdgpu_device *adev, uint64_t addr, } while (fault->timestamp < tmp); } -int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev) +int amdgpu_gmc_ras_early_init(struct amdgpu_device *adev) { - int r; - - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->ras_late_init) { - r = adev->umc.ras_funcs->ras_late_init(adev); - if (r) - return r; - } - - if (adev->mmhub.ras_funcs && - adev->mmhub.ras_funcs->ras_late_init) { - r = adev->mmhub.ras_funcs->ras_late_init(adev); - if (r) - return r; - } - - if (!adev->gmc.xgmi.connected_to_cpu) - adev->gmc.xgmi.ras_funcs = &xgmi_ras_funcs; - - if (adev->gmc.xgmi.ras_funcs && - adev->gmc.xgmi.ras_funcs->ras_late_init) { - r = adev->gmc.xgmi.ras_funcs->ras_late_init(adev); - if (r) - return r; + if (!adev->gmc.xgmi.connected_to_cpu) { + adev->gmc.xgmi.ras = &xgmi_ras; + amdgpu_ras_register_ras_block(adev, &adev->gmc.xgmi.ras->ras_block); + adev->gmc.xgmi.ras_if = &adev->gmc.xgmi.ras->ras_block.ras_comm; } - if (adev->hdp.ras_funcs && - adev->hdp.ras_funcs->ras_late_init) { - r = adev->hdp.ras_funcs->ras_late_init(adev); - if (r) - return r; - } - - if (adev->mca.mp0.ras_funcs && - adev->mca.mp0.ras_funcs->ras_late_init) { - r = adev->mca.mp0.ras_funcs->ras_late_init(adev); - if (r) - return r; - } - - if (adev->mca.mp1.ras_funcs && - adev->mca.mp1.ras_funcs->ras_late_init) { - r = adev->mca.mp1.ras_funcs->ras_late_init(adev); - if (r) - return r; - } - - if (adev->mca.mpio.ras_funcs && - adev->mca.mpio.ras_funcs->ras_late_init) { - r = adev->mca.mpio.ras_funcs->ras_late_init(adev); - if (r) - return r; - } + return 0; +} +int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev) +{ return 0; } void amdgpu_gmc_ras_fini(struct amdgpu_device *adev) { - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->ras_fini) - adev->umc.ras_funcs->ras_fini(adev); - if (adev->mmhub.ras_funcs && - adev->mmhub.ras_funcs->ras_fini) - adev->mmhub.ras_funcs->ras_fini(adev); - - if (adev->gmc.xgmi.ras_funcs && - adev->gmc.xgmi.ras_funcs->ras_fini) - adev->gmc.xgmi.ras_funcs->ras_fini(adev); - - if (adev->hdp.ras_funcs && - adev->hdp.ras_funcs->ras_fini) - adev->hdp.ras_funcs->ras_fini(adev); } /* @@ -584,6 +527,7 @@ void amdgpu_gmc_tmz_set(struct amdgpu_device *adev) case CHIP_NAVI12: case CHIP_VANGOGH: case CHIP_YELLOW_CARP: + case CHIP_IP_DISCOVERY: /* Don't enable it by default yet. */ if (amdgpu_tmz < 1) { @@ -615,11 +559,11 @@ void amdgpu_gmc_noretry_set(struct amdgpu_device *adev) { struct amdgpu_gmc *gmc = &adev->gmc; - switch (adev->asic_type) { - case CHIP_VEGA10: - case CHIP_VEGA20: - case CHIP_ARCTURUS: - case CHIP_ALDEBARAN: + switch (adev->ip_versions[GC_HWIP][0]) { + case IP_VERSION(9, 0, 1): + case IP_VERSION(9, 4, 0): + case IP_VERSION(9, 4, 1): + case IP_VERSION(9, 4, 2): /* * noretry = 0 will cause kfd page fault tests fail * for some ASICs, so set default to 1 for these ASICs. @@ -629,7 +573,6 @@ void amdgpu_gmc_noretry_set(struct amdgpu_device *adev) else gmc->noretry = amdgpu_noretry; break; - case CHIP_RAVEN: default: /* Raven currently has issues with noretry * regardless of what we decide for other @@ -680,6 +623,13 @@ void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev) unsigned size; /* + * Some ASICs need to reserve a region of video memory to avoid access + * from driver + */ + adev->mman.stolen_reserved_offset = 0; + adev->mman.stolen_reserved_size = 0; + + /* * TODO: * Currently there is a bug where some memory client outside * of the driver writes to first 8M of VRAM on S3 resume, @@ -689,10 +639,25 @@ void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev) */ switch (adev->asic_type) { case CHIP_VEGA10: + adev->mman.keep_stolen_vga_memory = true; + /* + * VEGA10 SRIOV VF needs some firmware reserved area. + */ + if (amdgpu_sriov_vf(adev)) { + adev->mman.stolen_reserved_offset = 0x100000; + adev->mman.stolen_reserved_size = 0x600000; + } + break; case CHIP_RAVEN: case CHIP_RENOIR: adev->mman.keep_stolen_vga_memory = true; break; + case CHIP_YELLOW_CARP: + if (amdgpu_discovery == 0) { + adev->mman.stolen_reserved_offset = 0x1ffb0000; + adev->mman.stolen_reserved_size = 64 * PAGE_SIZE; + } + break; default: adev->mman.keep_stolen_vga_memory = false; break; @@ -813,21 +778,48 @@ uint64_t amdgpu_gmc_vram_cpu_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo return amdgpu_bo_gpu_offset(bo) - adev->gmc.vram_start + adev->gmc.aper_base; } -void amdgpu_gmc_get_reserved_allocation(struct amdgpu_device *adev) +int amdgpu_gmc_vram_checking(struct amdgpu_device *adev) { - /* Some ASICs need to reserve a region of video memory to avoid access - * from driver */ - adev->mman.stolen_reserved_offset = 0; - adev->mman.stolen_reserved_size = 0; + struct amdgpu_bo *vram_bo = NULL; + uint64_t vram_gpu = 0; + void *vram_ptr = NULL; + + int ret, size = 0x100000; + uint8_t cptr[10]; + + ret = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, + &vram_bo, + &vram_gpu, + &vram_ptr); + if (ret) + return ret; + + memset(vram_ptr, 0x86, size); + memset(cptr, 0x86, 10); + + /** + * Check the start, the mid, and the end of the memory if the content of + * each byte is the pattern "0x86". If yes, we suppose the vram bo is + * workable. + * + * Note: If check the each byte of whole 1M bo, it will cost too many + * seconds, so here, we just pick up three parts for emulation. + */ + ret = memcmp(vram_ptr, cptr, 10); + if (ret) + return ret; - switch (adev->asic_type) { - case CHIP_YELLOW_CARP: - if (amdgpu_discovery == 0) { - adev->mman.stolen_reserved_offset = 0x1ffb0000; - adev->mman.stolen_reserved_size = 64 * PAGE_SIZE; - } - break; - default: - break; - } + ret = memcmp(vram_ptr + (size / 2), cptr, 10); + if (ret) + return ret; + + ret = memcmp(vram_ptr + size - 10, cptr, 10); + if (ret) + return ret; + + amdgpu_bo_free_kernel(&vram_bo, &vram_gpu, + &vram_ptr); + + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h index 8458cebc6d5b..032b0313f277 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h @@ -29,6 +29,7 @@ #include <linux/types.h> #include "amdgpu_irq.h" +#include "amdgpu_ras.h" /* VA hole for 48bit addresses on Vega10 */ #define AMDGPU_GMC_HOLE_START 0x0000800000000000ULL @@ -135,12 +136,8 @@ struct amdgpu_gmc_funcs { unsigned int (*get_vbios_fb_size)(struct amdgpu_device *adev); }; -struct amdgpu_xgmi_ras_funcs { - int (*ras_late_init)(struct amdgpu_device *adev); - void (*ras_fini)(struct amdgpu_device *adev); - int (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); - void (*reset_ras_error_count)(struct amdgpu_device *adev); +struct amdgpu_xgmi_ras { + struct amdgpu_ras_block_object ras_block; }; struct amdgpu_xgmi { @@ -159,7 +156,7 @@ struct amdgpu_xgmi { struct ras_common_if *ras_if; bool connected_to_cpu; bool pending_reset; - const struct amdgpu_xgmi_ras_funcs *ras_funcs; + struct amdgpu_xgmi_ras *ras; }; struct amdgpu_gmc { @@ -321,6 +318,7 @@ bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev, uint16_t pasid, uint64_t timestamp); void amdgpu_gmc_filter_faults_remove(struct amdgpu_device *adev, uint64_t addr, uint16_t pasid); +int amdgpu_gmc_ras_early_init(struct amdgpu_device *adev); int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev); void amdgpu_gmc_ras_fini(struct amdgpu_device *adev); int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev); @@ -333,10 +331,10 @@ amdgpu_gmc_set_vm_fault_masks(struct amdgpu_device *adev, int hub_type, bool enable); void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev); -void amdgpu_gmc_get_reserved_allocation(struct amdgpu_device *adev); void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev); uint64_t amdgpu_gmc_vram_mc2pa(struct amdgpu_device *adev, uint64_t mc_addr); uint64_t amdgpu_gmc_vram_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo); uint64_t amdgpu_gmc_vram_cpu_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo); +int amdgpu_gmc_vram_checking(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c index 68494b959116..8c6b2284cf56 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c @@ -187,26 +187,21 @@ static void amdgpu_gtt_mgr_del(struct ttm_resource_manager *man, * * Re-init the gart for each known BO in the GTT. */ -int amdgpu_gtt_mgr_recover(struct amdgpu_gtt_mgr *mgr) +void amdgpu_gtt_mgr_recover(struct amdgpu_gtt_mgr *mgr) { struct ttm_range_mgr_node *node; struct drm_mm_node *mm_node; struct amdgpu_device *adev; - int r = 0; adev = container_of(mgr, typeof(*adev), mman.gtt_mgr); spin_lock(&mgr->lock); drm_mm_for_each_node(mm_node, &mgr->mm) { node = container_of(mm_node, typeof(*node), mm_nodes[0]); - r = amdgpu_ttm_recover_gart(node->base.bo); - if (r) - break; + amdgpu_ttm_recover_gart(node->base.bo); } spin_unlock(&mgr->lock); amdgpu_gart_invalidate_tlb(adev); - - return r; } /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c deleted file mode 100644 index a766e1aad2b9..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c +++ /dev/null @@ -1,68 +0,0 @@ -/* - * Copyright 2021 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_ras.h" - -int amdgpu_hdp_ras_late_init(struct amdgpu_device *adev) -{ - int r; - struct ras_ih_if ih_info = { - .cb = NULL, - }; - struct ras_fs_if fs_info = { - .sysfs_name = "hdp_err_count", - }; - - if (!adev->hdp.ras_if) { - adev->hdp.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); - if (!adev->hdp.ras_if) - return -ENOMEM; - adev->hdp.ras_if->block = AMDGPU_RAS_BLOCK__HDP; - adev->hdp.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; - adev->hdp.ras_if->sub_block_index = 0; - } - ih_info.head = fs_info.head = *adev->hdp.ras_if; - r = amdgpu_ras_late_init(adev, adev->hdp.ras_if, - &fs_info, &ih_info); - if (r || !amdgpu_ras_is_supported(adev, adev->hdp.ras_if->block)) { - kfree(adev->hdp.ras_if); - adev->hdp.ras_if = NULL; - } - - return r; -} - -void amdgpu_hdp_ras_fini(struct amdgpu_device *adev) -{ - if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__HDP) && - adev->hdp.ras_if) { - struct ras_common_if *ras_if = adev->hdp.ras_if; - struct ras_ih_if ih_info = { - .cb = NULL, - }; - - amdgpu_ras_late_fini(adev, ras_if, &ih_info); - kfree(ras_if); - } -} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h index 7ec99d591584..9181c7bef7c6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h @@ -22,13 +22,10 @@ */ #ifndef __AMDGPU_HDP_H__ #define __AMDGPU_HDP_H__ +#include "amdgpu_ras.h" -struct amdgpu_hdp_ras_funcs { - int (*ras_late_init)(struct amdgpu_device *adev); - void (*ras_fini)(struct amdgpu_device *adev); - void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); - void (*reset_ras_error_count)(struct amdgpu_device *adev); +struct amdgpu_hdp_ras { + struct amdgpu_ras_block_object ras_block; }; struct amdgpu_hdp_funcs { @@ -43,9 +40,8 @@ struct amdgpu_hdp_funcs { struct amdgpu_hdp { struct ras_common_if *ras_if; const struct amdgpu_hdp_funcs *funcs; - const struct amdgpu_hdp_ras_funcs *ras_funcs; + struct amdgpu_hdp_ras *ras; }; -int amdgpu_hdp_ras_late_init(struct amdgpu_device *adev); -void amdgpu_hdp_ras_fini(struct amdgpu_device *adev); +int amdgpu_hdp_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block); #endif /* __AMDGPU_HDP_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index bc1297dcdf97..d583766ea392 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c @@ -166,8 +166,8 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, } if ((ib->flags & AMDGPU_IB_FLAGS_SECURE) && - (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)) { - dev_err(adev->dev, "secure submissions not supported on compute rings\n"); + (!ring->funcs->secure_submission_supported)) { + dev_err(adev->dev, "secure submissions not supported on ring <%s>\n", ring->name); return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c index 888d97143177..81207737c716 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c @@ -187,7 +187,7 @@ static int amdgpu_vmid_grab_idle(struct amdgpu_vm *vm, unsigned i; int r; - if (ring->vmid_wait && !dma_fence_is_signaled(ring->vmid_wait)) + if (!dma_fence_is_signaled(ring->vmid_wait)) return amdgpu_sync_fence(sync, ring->vmid_wait); fences = kmalloc_array(id_mgr->num_ids, sizeof(void *), GFP_KERNEL); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index f5cbc2747ac6..ea3e8c66211f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -199,13 +199,13 @@ static irqreturn_t amdgpu_irq_handler(int irq, void *arg) * ack the interrupt if it is there */ if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__PCIE_BIF)) { - if (adev->nbio.ras_funcs && - adev->nbio.ras_funcs->handle_ras_controller_intr_no_bifring) - adev->nbio.ras_funcs->handle_ras_controller_intr_no_bifring(adev); + if (adev->nbio.ras && + adev->nbio.ras->handle_ras_controller_intr_no_bifring) + adev->nbio.ras->handle_ras_controller_intr_no_bifring(adev); - if (adev->nbio.ras_funcs && - adev->nbio.ras_funcs->handle_ras_err_event_athub_intr_no_bifring) - adev->nbio.ras_funcs->handle_ras_err_event_athub_intr_no_bifring(adev); + if (adev->nbio.ras && + adev->nbio.ras->handle_ras_err_event_athub_intr_no_bifring) + adev->nbio.ras->handle_ras_err_event_athub_intr_no_bifring(adev); } return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index 38c9fd7b7ad4..67f66f2f1809 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -37,6 +37,7 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job) struct amdgpu_task_info ti; struct amdgpu_device *adev = ring->adev; int idx; + int r; if (!drm_dev_enter(adev_to_drm(adev), &idx)) { DRM_INFO("%s - device unplugged skipping recovery on scheduler:%s", @@ -63,7 +64,9 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job) ti.process_name, ti.tgid, ti.task_name, ti.pid); if (amdgpu_device_should_recover_gpu(ring->adev)) { - amdgpu_device_gpu_recover_imp(ring->adev, job); + r = amdgpu_device_gpu_recover_imp(ring->adev, job); + if (r) + DRM_ERROR("GPU Recovery Failed: %d\n", r); } else { drm_sched_suspend_timeout(&ring->sched); if (amdgpu_sriov_vf(adev)) @@ -78,14 +81,10 @@ exit: int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs, struct amdgpu_job **job, struct amdgpu_vm *vm) { - size_t size = sizeof(struct amdgpu_job); - if (num_ibs == 0) return -EINVAL; - size += sizeof(struct amdgpu_ib) * num_ibs; - - *job = kzalloc(size, GFP_KERNEL); + *job = kzalloc(struct_size(*job, ibs, num_ibs), GFP_KERNEL); if (!*job) return -ENOMEM; @@ -95,7 +94,6 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs, */ (*job)->base.sched = &adev->rings[0]->sched; (*job)->vm = vm; - (*job)->ibs = (void *)&(*job)[1]; (*job)->num_ibs = num_ibs; amdgpu_sync_create(&(*job)->sync); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h index 9e65730193b8..d599c0540b46 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h @@ -23,6 +23,10 @@ #ifndef __AMDGPU_JOB_H__ #define __AMDGPU_JOB_H__ +#include <drm/gpu_scheduler.h> +#include "amdgpu_sync.h" +#include "amdgpu_ring.h" + /* bit set means command submit involves a preamble IB */ #define AMDGPU_PREAMBLE_IB_PRESENT (1 << 0) /* bit set means preamble IB is first presented in belonging context */ @@ -45,12 +49,10 @@ struct amdgpu_job { struct amdgpu_vm *vm; struct amdgpu_sync sync; struct amdgpu_sync sched_sync; - struct amdgpu_ib *ibs; struct dma_fence hw_fence; struct dma_fence *external_hw_fence; uint32_t preamble_status; uint32_t preemption_status; - uint32_t num_ibs; bool vm_needs_flush; uint64_t vm_pd_addr; unsigned vmid; @@ -66,6 +68,9 @@ struct amdgpu_job { /* job_run_counter >= 1 means a resubmit job */ uint32_t job_run_counter; + + uint32_t num_ibs; + struct amdgpu_ib ibs[]; }; int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index 0beab961b18b..6b626c293e72 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -87,11 +87,6 @@ void amdgpu_driver_unload_kms(struct drm_device *dev) if (adev->rmmio == NULL) return; - if (adev->runpm) { - pm_runtime_get_sync(dev->dev); - pm_runtime_forbid(dev->dev); - } - if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DRV_UNLOAD)) DRM_WARN("smart shift update failed\n"); @@ -124,22 +119,6 @@ void amdgpu_register_gpu_instance(struct amdgpu_device *adev) mutex_unlock(&mgpu_info.mutex); } -static void amdgpu_get_audio_func(struct amdgpu_device *adev) -{ - struct pci_dev *p = NULL; - - p = pci_get_domain_bus_and_slot(pci_domain_nr(adev->pdev->bus), - adev->pdev->bus->number, 1); - if (p) { - pm_runtime_get_sync(&p->dev); - - pm_runtime_mark_last_busy(&p->dev); - pm_runtime_put_autosuspend(&p->dev); - - pci_dev_put(p); - } -} - /** * amdgpu_driver_load_kms - Main load function for KMS. * @@ -152,21 +131,10 @@ static void amdgpu_get_audio_func(struct amdgpu_device *adev) int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags) { struct drm_device *dev; - struct pci_dev *parent; int r, acpi_status; dev = adev_to_drm(adev); - if (amdgpu_has_atpx() && - (amdgpu_is_atpx_hybrid() || - amdgpu_has_atpx_dgpu_power_cntl()) && - ((flags & AMD_IS_APU) == 0) && - !pci_is_thunderbolt_attached(to_pci_dev(dev->dev))) - flags |= AMD_IS_PX; - - parent = pci_upstream_bridge(adev->pdev); - adev->has_pr3 = parent ? pci_pr3_present(parent) : false; - /* amdgpu_device_init should report only fatal error * like memory allocation failure or iomapping failure, * or memory manager initialization failure, it must @@ -224,58 +192,12 @@ int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags) if (acpi_status) dev_dbg(dev->dev, "Error during ACPI methods call\n"); - if (adev->runpm) { - /* only need to skip on ATPX */ - if (amdgpu_device_supports_px(dev)) - dev_pm_set_driver_flags(dev->dev, DPM_FLAG_NO_DIRECT_COMPLETE); - /* we want direct complete for BOCO */ - if (amdgpu_device_supports_boco(dev)) - dev_pm_set_driver_flags(dev->dev, DPM_FLAG_SMART_PREPARE | - DPM_FLAG_SMART_SUSPEND | - DPM_FLAG_MAY_SKIP_RESUME); - pm_runtime_use_autosuspend(dev->dev); - pm_runtime_set_autosuspend_delay(dev->dev, 5000); - - pm_runtime_allow(dev->dev); - - pm_runtime_mark_last_busy(dev->dev); - pm_runtime_put_autosuspend(dev->dev); - - /* - * For runpm implemented via BACO, PMFW will handle the - * timing for BACO in and out: - * - put ASIC into BACO state only when both video and - * audio functions are in D3 state. - * - pull ASIC out of BACO state when either video or - * audio function is in D0 state. - * Also, at startup, PMFW assumes both functions are in - * D0 state. - * - * So if snd driver was loaded prior to amdgpu driver - * and audio function was put into D3 state, there will - * be no PMFW-aware D-state transition(D0->D3) on runpm - * suspend. Thus the BACO will be not correctly kicked in. - * - * Via amdgpu_get_audio_func(), the audio dev is put - * into D0 state. Then there will be a PMFW-aware D-state - * transition(D0->D3) on runpm suspend. - */ - if (amdgpu_device_supports_baco(dev) && - !(adev->flags & AMD_IS_APU) && - (adev->asic_type >= CHIP_NAVI10)) - amdgpu_get_audio_func(adev); - } - if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DRV_LOAD)) DRM_WARN("smart shift update failed\n"); out: - if (r) { - /* balance pm_runtime_get_sync in amdgpu_driver_unload_kms */ - if (adev->rmmio && adev->runpm) - pm_runtime_put_noidle(dev->dev); + if (r) amdgpu_driver_unload_kms(dev); - } return r; } @@ -406,6 +328,10 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info, fw_info->ver = adev->psp.toc.fw_version; fw_info->feature = adev->psp.toc.feature_version; break; + case AMDGPU_INFO_FW_CAP: + fw_info->ver = adev->psp.cap_fw_version; + fw_info->feature = adev->psp.cap_feature_version; + break; default: return -EINVAL; } @@ -1270,18 +1196,20 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev, if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_VCE) != NULL) amdgpu_vce_free_handles(adev, file_priv); - amdgpu_vm_bo_rmv(adev, fpriv->prt_va); - if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) { /* TODO: how to handle reserve failure */ BUG_ON(amdgpu_bo_reserve(adev->virt.csa_obj, true)); - amdgpu_vm_bo_rmv(adev, fpriv->csa_va); + amdgpu_vm_bo_del(adev, fpriv->csa_va); fpriv->csa_va = NULL; amdgpu_bo_unreserve(adev->virt.csa_obj); } pasid = fpriv->vm.pasid; pd = amdgpu_bo_ref(fpriv->vm.root.bo); + if (!WARN_ON(amdgpu_bo_reserve(pd, true))) { + amdgpu_vm_bo_del(adev, fpriv->prt_va); + amdgpu_bo_unreserve(pd); + } amdgpu_ctx_mgr_fini(&fpriv->ctx_mgr); amdgpu_vm_fini(adev, &fpriv->vm); @@ -1429,8 +1357,7 @@ static int amdgpu_debugfs_firmware_info_show(struct seq_file *m, void *unused) struct drm_amdgpu_info_firmware fw_info; struct drm_amdgpu_query_fw query_fw; struct atom_context *ctx = adev->mode_info.atom_context; - uint8_t smu_minor, smu_debug; - uint16_t smu_major; + uint8_t smu_program, smu_major, smu_minor, smu_debug; int ret, i; static const char *ta_fw_name[TA_FW_TYPE_MAX_INDEX] = { @@ -1576,11 +1503,12 @@ static int amdgpu_debugfs_firmware_info_show(struct seq_file *m, void *unused) ret = amdgpu_firmware_info(&fw_info, &query_fw, adev); if (ret) return ret; - smu_major = (fw_info.ver >> 16) & 0xffff; + smu_program = (fw_info.ver >> 24) & 0xff; + smu_major = (fw_info.ver >> 16) & 0xff; smu_minor = (fw_info.ver >> 8) & 0xff; smu_debug = (fw_info.ver >> 0) & 0xff; - seq_printf(m, "SMC feature version: %u, firmware version: 0x%08x (%d.%d.%d)\n", - fw_info.feature, fw_info.ver, smu_major, smu_minor, smu_debug); + seq_printf(m, "SMC feature version: %u, program: %d, firmware version: 0x%08x (%d.%d.%d)\n", + fw_info.feature, smu_program, fw_info.ver, smu_major, smu_minor, smu_debug); /* SDMA */ query_fw.fw_type = AMDGPU_INFO_FW_SDMA; @@ -1625,6 +1553,16 @@ static int amdgpu_debugfs_firmware_info_show(struct seq_file *m, void *unused) seq_printf(m, "TOC feature version: %u, firmware version: 0x%08x\n", fw_info.feature, fw_info.ver); + /* CAP */ + if (adev->psp.cap_fw) { + query_fw.fw_type = AMDGPU_INFO_FW_CAP; + ret = amdgpu_firmware_info(&fw_info, &query_fw, adev); + if (ret) + return ret; + seq_printf(m, "CAP feature version: %u, firmware version: 0x%08x\n", + fw_info.feature, fw_info.ver); + } + seq_printf(m, "VBIOS version: %s\n", ctx->vbios_version); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c index ce538f4819f9..51c2a82e2fa4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c @@ -70,48 +70,3 @@ void amdgpu_mca_query_ras_error_count(struct amdgpu_device *adev, amdgpu_mca_reset_error_count(adev, mc_status_addr); } - -int amdgpu_mca_ras_late_init(struct amdgpu_device *adev, - struct amdgpu_mca_ras *mca_dev) -{ - int r; - struct ras_ih_if ih_info = { - .cb = NULL, - }; - struct ras_fs_if fs_info = { - .sysfs_name = mca_dev->ras_funcs->sysfs_name, - }; - - if (!mca_dev->ras_if) { - mca_dev->ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); - if (!mca_dev->ras_if) - return -ENOMEM; - mca_dev->ras_if->block = mca_dev->ras_funcs->ras_block; - mca_dev->ras_if->sub_block_index = mca_dev->ras_funcs->ras_sub_block; - mca_dev->ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; - } - ih_info.head = fs_info.head = *mca_dev->ras_if; - r = amdgpu_ras_late_init(adev, mca_dev->ras_if, - &fs_info, &ih_info); - if (r || !amdgpu_ras_is_supported(adev, mca_dev->ras_if->block)) { - kfree(mca_dev->ras_if); - mca_dev->ras_if = NULL; - } - - return r; -} - -void amdgpu_mca_ras_fini(struct amdgpu_device *adev, - struct amdgpu_mca_ras *mca_dev) -{ - struct ras_ih_if ih_info = { - .cb = NULL, - }; - - if (!mca_dev->ras_if) - return; - - amdgpu_ras_late_fini(adev, mca_dev->ras_if, &ih_info); - kfree(mca_dev->ras_if); - mca_dev->ras_if = NULL; -}
\ No newline at end of file diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h index c74bc7177066..7ce16d16e34b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h @@ -21,21 +21,13 @@ #ifndef __AMDGPU_MCA_H__ #define __AMDGPU_MCA_H__ -struct amdgpu_mca_ras_funcs { - int (*ras_late_init)(struct amdgpu_device *adev); - void (*ras_fini)(struct amdgpu_device *adev); - void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); - void (*query_ras_error_address)(struct amdgpu_device *adev, - void *ras_error_status); - uint32_t ras_block; - uint32_t ras_sub_block; - const char* sysfs_name; +struct amdgpu_mca_ras_block { + struct amdgpu_ras_block_object ras_block; }; struct amdgpu_mca_ras { struct ras_common_if *ras_if; - const struct amdgpu_mca_ras_funcs *ras_funcs; + struct amdgpu_mca_ras_block *ras; }; struct amdgpu_mca_funcs { @@ -64,10 +56,4 @@ void amdgpu_mca_query_ras_error_count(struct amdgpu_device *adev, uint64_t mc_status_addr, void *ras_error_status); -int amdgpu_mca_ras_late_init(struct amdgpu_device *adev, - struct amdgpu_mca_ras *mca_dev); - -void amdgpu_mca_ras_fini(struct amdgpu_device *adev, - struct amdgpu_mca_ras *mca_dev); - #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c deleted file mode 100644 index 24297dc51434..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c +++ /dev/null @@ -1,68 +0,0 @@ -/* - * 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 "amdgpu_ras.h" - -int amdgpu_mmhub_ras_late_init(struct amdgpu_device *adev) -{ - int r; - struct ras_ih_if ih_info = { - .cb = NULL, - }; - struct ras_fs_if fs_info = { - .sysfs_name = "mmhub_err_count", - }; - - if (!adev->mmhub.ras_if) { - adev->mmhub.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); - if (!adev->mmhub.ras_if) - return -ENOMEM; - adev->mmhub.ras_if->block = AMDGPU_RAS_BLOCK__MMHUB; - adev->mmhub.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; - adev->mmhub.ras_if->sub_block_index = 0; - } - ih_info.head = fs_info.head = *adev->mmhub.ras_if; - r = amdgpu_ras_late_init(adev, adev->mmhub.ras_if, - &fs_info, &ih_info); - if (r || !amdgpu_ras_is_supported(adev, adev->mmhub.ras_if->block)) { - kfree(adev->mmhub.ras_if); - adev->mmhub.ras_if = NULL; - } - - return r; -} - -void amdgpu_mmhub_ras_fini(struct amdgpu_device *adev) -{ - if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB) && - adev->mmhub.ras_if) { - struct ras_common_if *ras_if = adev->mmhub.ras_if; - struct ras_ih_if ih_info = { - .cb = NULL, - }; - - amdgpu_ras_late_fini(adev, ras_if, &ih_info); - kfree(ras_if); - } -} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h index b27fcbccce2b..9f1540f0ebf9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h @@ -21,14 +21,8 @@ #ifndef __AMDGPU_MMHUB_H__ #define __AMDGPU_MMHUB_H__ -struct amdgpu_mmhub_ras_funcs { - int (*ras_late_init)(struct amdgpu_device *adev); - void (*ras_fini)(struct amdgpu_device *adev); - void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); - void (*query_ras_error_status)(struct amdgpu_device *adev); - void (*reset_ras_error_count)(struct amdgpu_device *adev); - void (*reset_ras_error_status)(struct amdgpu_device *adev); +struct amdgpu_mmhub_ras { + struct amdgpu_ras_block_object ras_block; }; struct amdgpu_mmhub_funcs { @@ -50,10 +44,8 @@ struct amdgpu_mmhub_funcs { struct amdgpu_mmhub { struct ras_common_if *ras_if; const struct amdgpu_mmhub_funcs *funcs; - const struct amdgpu_mmhub_ras_funcs *ras_funcs; + struct amdgpu_mmhub_ras *ras; }; -int amdgpu_mmhub_ras_late_init(struct amdgpu_device *adev); -void amdgpu_mmhub_ras_fini(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index 902235fae4cd..a546cb3cfa18 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -341,6 +341,7 @@ struct amdgpu_mode_info { int num_crtc; /* number of crtcs */ int num_hpd; /* number of hpd pins */ int num_dig; /* number of dig blocks */ + bool gpu_vm_support; /* supports display from GTT */ int disp_priority; const struct amdgpu_display_funcs *funcs; const enum drm_plane_type *plane_type; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c index 6afb02fef8cf..37d779b8e4a6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c @@ -22,61 +22,24 @@ #include "amdgpu.h" #include "amdgpu_ras.h" -int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev) +int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block) { int r; - struct ras_ih_if ih_info = { - .cb = NULL, - }; - struct ras_fs_if fs_info = { - .sysfs_name = "pcie_bif_err_count", - }; - - if (!adev->nbio.ras_if) { - adev->nbio.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); - if (!adev->nbio.ras_if) - return -ENOMEM; - adev->nbio.ras_if->block = AMDGPU_RAS_BLOCK__PCIE_BIF; - adev->nbio.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; - adev->nbio.ras_if->sub_block_index = 0; - } - ih_info.head = fs_info.head = *adev->nbio.ras_if; - r = amdgpu_ras_late_init(adev, adev->nbio.ras_if, - &fs_info, &ih_info); + r = amdgpu_ras_block_late_init(adev, ras_block); if (r) - goto free; + return r; - if (amdgpu_ras_is_supported(adev, adev->nbio.ras_if->block)) { + if (amdgpu_ras_is_supported(adev, ras_block->block)) { r = amdgpu_irq_get(adev, &adev->nbio.ras_controller_irq, 0); if (r) goto late_fini; r = amdgpu_irq_get(adev, &adev->nbio.ras_err_event_athub_irq, 0); if (r) goto late_fini; - } else { - r = 0; - goto free; } return 0; late_fini: - amdgpu_ras_late_fini(adev, adev->nbio.ras_if, &ih_info); -free: - kfree(adev->nbio.ras_if); - adev->nbio.ras_if = NULL; + amdgpu_ras_block_late_fini(adev, ras_block); return r; } - -void amdgpu_nbio_ras_fini(struct amdgpu_device *adev) -{ - if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__PCIE_BIF) && - adev->nbio.ras_if) { - struct ras_common_if *ras_if = adev->nbio.ras_if; - struct ras_ih_if ih_info = { - .cb = NULL, - }; - - amdgpu_ras_late_fini(adev, ras_if, &ih_info); - kfree(ras_if); - } -} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h index 843052205bd5..3d13e601fc35 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h @@ -47,15 +47,12 @@ struct nbio_hdp_flush_reg { u32 ref_and_mask_sdma7; }; -struct amdgpu_nbio_ras_funcs { +struct amdgpu_nbio_ras { + struct amdgpu_ras_block_object ras_block; void (*handle_ras_controller_intr_no_bifring)(struct amdgpu_device *adev); void (*handle_ras_err_event_athub_intr_no_bifring)(struct amdgpu_device *adev); int (*init_ras_controller_interrupt)(struct amdgpu_device *adev); int (*init_ras_err_event_athub_interrupt)(struct amdgpu_device *adev); - void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); - int (*ras_late_init)(struct amdgpu_device *adev); - void (*ras_fini)(struct amdgpu_device *adev); }; struct amdgpu_nbio_funcs { @@ -104,9 +101,8 @@ struct amdgpu_nbio { struct amdgpu_irq_src ras_err_event_athub_irq; struct ras_common_if *ras_if; const struct amdgpu_nbio_funcs *funcs; - const struct amdgpu_nbio_ras_funcs *ras_funcs; + struct amdgpu_nbio_ras *ras; }; -int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev); -void amdgpu_nbio_ras_fini(struct amdgpu_device *adev); +int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c index ea0cde4904f0..25731719c627 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c @@ -575,6 +575,9 @@ int amdgpu_bo_create(struct amdgpu_device *adev, if (!amdgpu_bo_support_uswc(bo->flags)) bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC; + if (adev->ras_enabled) + bo->flags |= AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE; + bo->tbo.bdev = &adev->mman.bdev; if (bp->domain & (AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA | AMDGPU_GEM_DOMAIN_GDS)) @@ -1303,7 +1306,8 @@ void amdgpu_bo_release_notify(struct ttm_buffer_object *bo) !(abo->flags & AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE)) return; - dma_resv_lock(bo->base.resv, NULL); + if (WARN_ON_ONCE(!dma_resv_trylock(bo->base.resv))) + return; r = amdgpu_fill_buffer(abo, AMDGPU_POISON, bo->base.resv, &fence); if (!WARN_ON(r)) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index dee17a0e1187..a6acec1a6155 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -133,6 +133,8 @@ static int psp_early_init(void *handle) break; case IP_VERSION(13, 0, 1): case IP_VERSION(13, 0, 3): + case IP_VERSION(13, 0, 5): + case IP_VERSION(13, 0, 8): psp_v13_0_set_psp_funcs(psp); psp->autoload_supported = true; break; @@ -259,6 +261,32 @@ static bool psp_get_runtime_db_entry(struct amdgpu_device *adev, return ret; } +static int psp_init_sriov_microcode(struct psp_context *psp) +{ + struct amdgpu_device *adev = psp->adev; + int ret = 0; + + switch (adev->ip_versions[MP0_HWIP][0]) { + case IP_VERSION(9, 0, 0): + ret = psp_init_cap_microcode(psp, "vega10"); + break; + case IP_VERSION(11, 0, 9): + ret = psp_init_cap_microcode(psp, "navi12"); + break; + case IP_VERSION(11, 0, 7): + ret = psp_init_cap_microcode(psp, "sienna_cichlid"); + break; + case IP_VERSION(13, 0, 2): + ret = psp_init_cap_microcode(psp, "aldebaran"); + break; + default: + BUG(); + break; + } + + return ret; +} + static int psp_sw_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -273,21 +301,19 @@ static int psp_sw_init(void *handle) ret = -ENOMEM; } - if (!amdgpu_sriov_vf(adev)) { + if (amdgpu_sriov_vf(adev)) + ret = psp_init_sriov_microcode(psp); + else ret = psp_init_microcode(psp); - if (ret) { - DRM_ERROR("Failed to load psp firmware!\n"); - return ret; - } - } else if (amdgpu_sriov_vf(adev) && - adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 2)) { - ret = psp_init_ta_microcode(psp, "aldebaran"); - if (ret) { - DRM_ERROR("Failed to initialize ta microcode!\n"); - return ret; - } + if (ret) { + DRM_ERROR("Failed to load psp firmware!\n"); + return ret; } + adev->psp.xgmi_context.supports_extended_data = + !adev->gmc.xgmi.connected_to_cpu && + adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 2); + memset(&boot_cfg_entry, 0, sizeof(boot_cfg_entry)); if (psp_get_runtime_db_entry(adev, PSP_RUNTIME_ENTRY_TYPE_BOOT_CONFIG, @@ -353,6 +379,10 @@ static int psp_sw_fini(void *handle) release_firmware(psp->ta_fw); psp->ta_fw = NULL; } + if (adev->psp.cap_fw) { + release_firmware(psp->cap_fw); + psp->cap_fw = NULL; + } if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(11, 0, 0) || adev->ip_versions[MP0_HWIP][0] == IP_VERSION(11, 0, 7)) @@ -491,7 +521,10 @@ psp_cmd_submit_buf(struct psp_context *psp, DRM_WARN("psp gfx command %s(0x%X) failed and response status is (0x%X)\n", psp_gfx_cmd_name(psp->cmd_buf_mem->cmd_id), psp->cmd_buf_mem->cmd_id, psp->cmd_buf_mem->resp.status); - if (!timeout) { + /* If we load CAP FW, PSP must return 0 under SRIOV + * also return failure in case of timeout + */ + if ((ucode && (ucode->ucode_id == AMDGPU_UCODE_ID_CAP)) || !timeout) { ret = -EINVAL; goto exit; } @@ -914,19 +947,15 @@ static void psp_prep_ta_load_cmd_buf(struct psp_gfx_cmd_resp *cmd, static int psp_ta_init_shared_buf(struct psp_context *psp, struct ta_mem_context *mem_ctx) { - int ret; - /* * Allocate 16k memory aligned to 4k from Frame Buffer (local * physical) for ta to host memory */ - ret = amdgpu_bo_create_kernel(psp->adev, mem_ctx->shared_mem_size, + return amdgpu_bo_create_kernel(psp->adev, mem_ctx->shared_mem_size, PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM, &mem_ctx->shared_bo, &mem_ctx->shared_mc_addr, &mem_ctx->shared_buf); - - return ret; } static void psp_ta_free_shared_buf(struct ta_mem_context *mem_ctx) @@ -1308,6 +1337,11 @@ static void psp_ras_ta_check_status(struct psp_context *psp) break; case TA_RAS_STATUS__SUCCESS: break; + case TA_RAS_STATUS__TEE_ERROR_ACCESS_DENIED: + if (ras_cmd->cmd_id == TA_RAS_COMMAND__TRIGGER_ERROR) + dev_warn(psp->adev->dev, + "RAS WARNING: Inject error to critical region is not allowed\n"); + break; default: dev_warn(psp->adev->dev, "RAS WARNING: ras status = 0x%X\n", ras_cmd->ras_status); @@ -1520,7 +1554,9 @@ int psp_ras_trigger_error(struct psp_context *psp, if (amdgpu_ras_intr_triggered()) return 0; - if (ras_cmd->ras_status) + if (ras_cmd->ras_status == TA_RAS_STATUS__TEE_ERROR_ACCESS_DENIED) + return -EACCES; + else if (ras_cmd->ras_status) return -EINVAL; return 0; @@ -2051,6 +2087,9 @@ static int psp_get_fw_type(struct amdgpu_firmware_info *ucode, enum psp_gfx_fw_type *type) { switch (ucode->ucode_id) { + case AMDGPU_UCODE_ID_CAP: + *type = GFX_FW_TYPE_CAP; + break; case AMDGPU_UCODE_ID_SDMA0: *type = GFX_FW_TYPE_SDMA0; break; @@ -2973,7 +3012,6 @@ static int psp_init_sos_base_fw(struct amdgpu_device *adev) adev->psp.sos.size_bytes = le32_to_cpu(sos_hdr->sos.size_bytes); adev->psp.sos.start_addr = ucode_array_start_addr + le32_to_cpu(sos_hdr->sos.offset_bytes); - adev->psp.xgmi_context.supports_extended_data = false; } else { /* Load alternate PSP SOS FW */ sos_hdr_v1_3 = (const struct psp_firmware_header_v1_3 *)adev->psp.sos_fw->data; @@ -2988,7 +3026,6 @@ static int psp_init_sos_base_fw(struct amdgpu_device *adev) adev->psp.sos.size_bytes = le32_to_cpu(sos_hdr_v1_3->sos_aux.size_bytes); adev->psp.sos.start_addr = ucode_array_start_addr + le32_to_cpu(sos_hdr_v1_3->sos_aux.offset_bytes); - adev->psp.xgmi_context.supports_extended_data = true; } if ((adev->psp.sys.size_bytes == 0) || (adev->psp.sos.size_bytes == 0)) { @@ -3217,6 +3254,58 @@ out: return err; } +int psp_init_cap_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 *cap_hdr_v1_0; + struct amdgpu_firmware_info *info = NULL; + int err = 0; + + if (!chip_name) { + dev_err(adev->dev, "invalid chip name for cap microcode\n"); + return -EINVAL; + } + + if (!amdgpu_sriov_vf(adev)) { + dev_err(adev->dev, "cap microcode should only be loaded under SRIOV\n"); + return -EINVAL; + } + + snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_cap.bin", chip_name); + err = request_firmware(&adev->psp.cap_fw, fw_name, adev->dev); + if (err) { + dev_warn(adev->dev, "cap microcode does not exist, skip\n"); + err = 0; + goto out; + } + + err = amdgpu_ucode_validate(adev->psp.cap_fw); + if (err) { + dev_err(adev->dev, "fail to initialize cap microcode\n"); + goto out; + } + + info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CAP]; + info->ucode_id = AMDGPU_UCODE_ID_CAP; + info->fw = adev->psp.cap_fw; + cap_hdr_v1_0 = (const struct psp_firmware_header_v1_0 *) + adev->psp.cap_fw->data; + adev->firmware.fw_size += ALIGN( + le32_to_cpu(cap_hdr_v1_0->header.ucode_size_bytes), PAGE_SIZE); + adev->psp.cap_fw_version = le32_to_cpu(cap_hdr_v1_0->header.ucode_version); + adev->psp.cap_feature_version = le32_to_cpu(cap_hdr_v1_0->sos.fw_version); + adev->psp.cap_ucode_size = le32_to_cpu(cap_hdr_v1_0->header.ucode_size_bytes); + + return 0; + +out: + release_firmware(adev->psp.cap_fw); + adev->psp.cap_fw = NULL; + return err; +} + static int psp_set_clockgating_state(void *handle, enum amd_clockgating_state state) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index f29afabbff1f..ff7d533eb746 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -306,6 +306,9 @@ struct psp_context /* toc firmware */ const struct firmware *toc_fw; + /* cap firmware */ + const struct firmware *cap_fw; + /* fence buffer */ struct amdgpu_bo *fence_buf_bo; uint64_t fence_buf_mc_addr; @@ -327,6 +330,10 @@ struct psp_context const struct firmware *ta_fw; uint32_t ta_fw_version; + uint32_t cap_fw_version; + uint32_t cap_feature_version; + uint32_t cap_ucode_size; + struct ta_context asd_context; struct psp_xgmi_context xgmi_context; struct psp_ras_context ras_context; @@ -440,6 +447,8 @@ 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_init_cap_microcode(struct psp_context *psp, + const char *chip_name); int psp_get_fw_attestation_records_addr(struct psp_context *psp, uint64_t *output_ptr); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 8f47c14ecbc7..424c22a841f4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -75,6 +75,13 @@ const char *ras_mca_block_string[] = { "mca_iohc", }; +struct amdgpu_ras_block_list { + /* ras block link */ + struct list_head node; + + struct amdgpu_ras_block_object *ras_obj; +}; + const char *get_ras_block_str(struct ras_common_if *ras_block) { if (!ras_block) @@ -89,6 +96,9 @@ const char *get_ras_block_str(struct ras_common_if *ras_block) return ras_block_string[ras_block->block]; } +#define ras_block_str(_BLOCK_) \ + (((_BLOCK_) < ARRAY_SIZE(ras_block_string)) ? ras_block_string[_BLOCK_] : "Out Of Range") + #define ras_err_str(i) (ras_error_string[ffs(i)]) #define RAS_DEFAULT_FLAGS (AMDGPU_RAS_FLAG_INIT_BY_VBIOS) @@ -155,14 +165,9 @@ static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t addre } memset(&err_rec, 0x0, sizeof(struct eeprom_table_record)); - - err_rec.address = address; - err_rec.retired_page = address >> AMDGPU_GPU_PAGE_SHIFT; - err_rec.ts = (uint64_t)ktime_get_real_seconds(); - err_rec.err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; - err_data.err_addr = &err_rec; - err_data.err_addr_cnt = 1; + amdgpu_umc_fill_error_record(&err_data, address, + (address >> AMDGPU_GPU_PAGE_SHIFT), 0, 0); if (amdgpu_bad_page_threshold != 0) { amdgpu_ras_add_bad_pages(adev, err_data.err_addr, @@ -452,7 +457,7 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, } if (ret) - return -EINVAL; + return ret; return size; } @@ -866,30 +871,47 @@ static int amdgpu_ras_enable_all_features(struct amdgpu_device *adev, } /* feature ctl end */ +static int amdgpu_ras_block_match_default(struct amdgpu_ras_block_object *block_obj, + enum amdgpu_ras_block block) +{ + if (!block_obj) + return -EINVAL; -static void amdgpu_ras_mca_query_error_status(struct amdgpu_device *adev, - struct ras_common_if *ras_block, - struct ras_err_data *err_data) + if (block_obj->ras_comm.block == block) + return 0; + + return -EINVAL; +} + +static struct amdgpu_ras_block_object *amdgpu_ras_get_ras_block(struct amdgpu_device *adev, + enum amdgpu_ras_block block, uint32_t sub_block_index) { - switch (ras_block->sub_block_index) { - case AMDGPU_RAS_MCA_BLOCK__MP0: - if (adev->mca.mp0.ras_funcs && - adev->mca.mp0.ras_funcs->query_ras_error_count) - adev->mca.mp0.ras_funcs->query_ras_error_count(adev, &err_data); - break; - case AMDGPU_RAS_MCA_BLOCK__MP1: - if (adev->mca.mp1.ras_funcs && - adev->mca.mp1.ras_funcs->query_ras_error_count) - adev->mca.mp1.ras_funcs->query_ras_error_count(adev, &err_data); - break; - case AMDGPU_RAS_MCA_BLOCK__MPIO: - if (adev->mca.mpio.ras_funcs && - adev->mca.mpio.ras_funcs->query_ras_error_count) - adev->mca.mpio.ras_funcs->query_ras_error_count(adev, &err_data); - break; - default: - break; + struct amdgpu_ras_block_list *node, *tmp; + struct amdgpu_ras_block_object *obj; + + if (block >= AMDGPU_RAS_BLOCK__LAST) + return NULL; + + if (!amdgpu_ras_is_supported(adev, block)) + return NULL; + + list_for_each_entry_safe(node, tmp, &adev->ras_list, node) { + if (!node->ras_obj) { + dev_warn(adev->dev, "Warning: abnormal ras list node.\n"); + continue; + } + + obj = node->ras_obj; + if (obj->ras_block_match) { + if (obj->ras_block_match(obj, block, sub_block_index) == 0) + return obj; + } else { + if (amdgpu_ras_block_match_default(obj, block) == 0) + return obj; + } } + + return NULL; } static void amdgpu_ras_get_ecc_info(struct amdgpu_device *adev, struct ras_err_data *err_data) @@ -901,26 +923,26 @@ static void amdgpu_ras_get_ecc_info(struct amdgpu_device *adev, struct ras_err_d * choosing right query method according to * whether smu support query error information */ - ret = smu_get_ecc_info(&adev->smu, (void *)&(ras->umc_ecc)); + ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(ras->umc_ecc)); if (ret == -EOPNOTSUPP) { - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->query_ras_error_count) - adev->umc.ras_funcs->query_ras_error_count(adev, err_data); + if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops && + adev->umc.ras->ras_block.hw_ops->query_ras_error_count) + adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev, err_data); /* umc query_ras_error_address is also responsible for clearing * error status */ - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->query_ras_error_address) - adev->umc.ras_funcs->query_ras_error_address(adev, err_data); + if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops && + adev->umc.ras->ras_block.hw_ops->query_ras_error_address) + adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev, err_data); } else if (!ret) { - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->ecc_info_query_ras_error_count) - adev->umc.ras_funcs->ecc_info_query_ras_error_count(adev, err_data); + if (adev->umc.ras && + adev->umc.ras->ecc_info_query_ras_error_count) + adev->umc.ras->ecc_info_query_ras_error_count(adev, err_data); - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->ecc_info_query_ras_error_address) - adev->umc.ras_funcs->ecc_info_query_ras_error_address(adev, err_data); + if (adev->umc.ras && + adev->umc.ras->ecc_info_query_ras_error_address) + adev->umc.ras->ecc_info_query_ras_error_address(adev, err_data); } } @@ -928,62 +950,32 @@ static void amdgpu_ras_get_ecc_info(struct amdgpu_device *adev, struct ras_err_d int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info) { + struct amdgpu_ras_block_object *block_obj = NULL; struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head); struct ras_err_data err_data = {0, 0, 0, NULL}; - int i; if (!obj) return -EINVAL; - switch (info->head.block) { - case AMDGPU_RAS_BLOCK__UMC: + if (info->head.block == AMDGPU_RAS_BLOCK__UMC) { amdgpu_ras_get_ecc_info(adev, &err_data); - break; - case AMDGPU_RAS_BLOCK__SDMA: - if (adev->sdma.funcs->query_ras_error_count) { - for (i = 0; i < adev->sdma.num_instances; i++) - adev->sdma.funcs->query_ras_error_count(adev, i, - &err_data); + } else { + block_obj = amdgpu_ras_get_ras_block(adev, info->head.block, 0); + if (!block_obj || !block_obj->hw_ops) { + dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", + get_ras_block_str(&info->head)); + return -EINVAL; } - break; - case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->query_ras_error_count) - adev->gfx.ras_funcs->query_ras_error_count(adev, &err_data); - - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->query_ras_error_status) - adev->gfx.ras_funcs->query_ras_error_status(adev); - break; - case AMDGPU_RAS_BLOCK__MMHUB: - if (adev->mmhub.ras_funcs && - adev->mmhub.ras_funcs->query_ras_error_count) - adev->mmhub.ras_funcs->query_ras_error_count(adev, &err_data); - - if (adev->mmhub.ras_funcs && - adev->mmhub.ras_funcs->query_ras_error_status) - adev->mmhub.ras_funcs->query_ras_error_status(adev); - break; - case AMDGPU_RAS_BLOCK__PCIE_BIF: - if (adev->nbio.ras_funcs && - adev->nbio.ras_funcs->query_ras_error_count) - adev->nbio.ras_funcs->query_ras_error_count(adev, &err_data); - break; - case AMDGPU_RAS_BLOCK__XGMI_WAFL: - if (adev->gmc.xgmi.ras_funcs && - adev->gmc.xgmi.ras_funcs->query_ras_error_count) - adev->gmc.xgmi.ras_funcs->query_ras_error_count(adev, &err_data); - break; - case AMDGPU_RAS_BLOCK__HDP: - if (adev->hdp.ras_funcs && - adev->hdp.ras_funcs->query_ras_error_count) - adev->hdp.ras_funcs->query_ras_error_count(adev, &err_data); - break; - case AMDGPU_RAS_BLOCK__MCA: - amdgpu_ras_mca_query_error_status(adev, &info->head, &err_data); - break; - default: - break; + + if (block_obj->hw_ops->query_ras_error_count) + block_obj->hw_ops->query_ras_error_count(adev, &err_data); + + if ((info->head.block == AMDGPU_RAS_BLOCK__SDMA) || + (info->head.block == AMDGPU_RAS_BLOCK__GFX) || + (info->head.block == AMDGPU_RAS_BLOCK__MMHUB)) { + if (block_obj->hw_ops->query_ras_error_status) + block_obj->hw_ops->query_ras_error_status(adev); + } } obj->err_data.ue_count += err_data.ue_count; @@ -1040,68 +1032,27 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, enum amdgpu_ras_block block) { + struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0); + if (!amdgpu_ras_is_supported(adev, block)) return -EINVAL; - switch (block) { - case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->reset_ras_error_count) - adev->gfx.ras_funcs->reset_ras_error_count(adev); - - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->reset_ras_error_status) - adev->gfx.ras_funcs->reset_ras_error_status(adev); - break; - case AMDGPU_RAS_BLOCK__MMHUB: - if (adev->mmhub.ras_funcs && - adev->mmhub.ras_funcs->reset_ras_error_count) - adev->mmhub.ras_funcs->reset_ras_error_count(adev); - - if (adev->mmhub.ras_funcs && - adev->mmhub.ras_funcs->reset_ras_error_status) - adev->mmhub.ras_funcs->reset_ras_error_status(adev); - break; - case AMDGPU_RAS_BLOCK__SDMA: - if (adev->sdma.funcs->reset_ras_error_count) - adev->sdma.funcs->reset_ras_error_count(adev); - break; - case AMDGPU_RAS_BLOCK__HDP: - if (adev->hdp.ras_funcs && - adev->hdp.ras_funcs->reset_ras_error_count) - adev->hdp.ras_funcs->reset_ras_error_count(adev); - break; - default: - break; + if (!block_obj || !block_obj->hw_ops) { + dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", + ras_block_str(block)); + return -EINVAL; } - return 0; -} - -/* Trigger XGMI/WAFL error */ -static int amdgpu_ras_error_inject_xgmi(struct amdgpu_device *adev, - struct ta_ras_trigger_error_input *block_info) -{ - int ret; - - if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_DISALLOW)) - dev_warn(adev->dev, "Failed to disallow df cstate"); - - if (amdgpu_dpm_allow_xgmi_power_down(adev, false)) - dev_warn(adev->dev, "Failed to disallow XGMI power down"); - - ret = psp_ras_trigger_error(&adev->psp, block_info); + if (block_obj->hw_ops->reset_ras_error_count) + block_obj->hw_ops->reset_ras_error_count(adev); - if (amdgpu_ras_intr_triggered()) - return ret; - - if (amdgpu_dpm_allow_xgmi_power_down(adev, true)) - dev_warn(adev->dev, "Failed to allow XGMI power down"); - - if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_ALLOW)) - dev_warn(adev->dev, "Failed to allow df cstate"); + if ((block == AMDGPU_RAS_BLOCK__GFX) || + (block == AMDGPU_RAS_BLOCK__MMHUB)) { + if (block_obj->hw_ops->reset_ras_error_status) + block_obj->hw_ops->reset_ras_error_status(adev); + } - return ret; + return 0; } /* wrapper of psp_ras_trigger_error */ @@ -1116,11 +1067,20 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, .address = info->address, .value = info->value, }; - int ret = 0; + int ret = -EINVAL; + struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, + info->head.block, + info->head.sub_block_index); if (!obj) return -EINVAL; + if (!block_obj || !block_obj->hw_ops) { + dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", + get_ras_block_str(&info->head)); + return -EINVAL; + } + /* Calculate XGMI relative offset */ if (adev->gmc.xgmi.num_physical_nodes > 1) { block_info.address = @@ -1128,28 +1088,15 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, block_info.address); } - switch (info->head.block) { - case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->ras_error_inject) - ret = adev->gfx.ras_funcs->ras_error_inject(adev, info); - else - ret = -EINVAL; - break; - case AMDGPU_RAS_BLOCK__UMC: - case AMDGPU_RAS_BLOCK__SDMA: - case AMDGPU_RAS_BLOCK__MMHUB: - case AMDGPU_RAS_BLOCK__PCIE_BIF: - case AMDGPU_RAS_BLOCK__MCA: - ret = psp_ras_trigger_error(&adev->psp, &block_info); - break; - case AMDGPU_RAS_BLOCK__XGMI_WAFL: - ret = amdgpu_ras_error_inject_xgmi(adev, &block_info); - break; - default: - dev_info(adev->dev, "%s error injection is not supported yet\n", - get_ras_block_str(&info->head)); - ret = -EINVAL; + if (info->head.block == AMDGPU_RAS_BLOCK__GFX) { + if (block_obj->hw_ops->ras_error_inject) + ret = block_obj->hw_ops->ras_error_inject(adev, info); + } else { + /* If defined special ras_error_inject(e.g: xgmi), implement special ras_error_inject */ + if (block_obj->hw_ops->ras_error_inject) + ret = block_obj->hw_ops->ras_error_inject(adev, &block_info); + else /*If not defined .ras_error_inject, use default ras_error_inject*/ + ret = psp_ras_trigger_error(&adev->psp, &block_info); } if (ret) @@ -1329,18 +1276,17 @@ static int amdgpu_ras_sysfs_remove_feature_node(struct amdgpu_device *adev) } int amdgpu_ras_sysfs_create(struct amdgpu_device *adev, - struct ras_fs_if *head) + struct ras_common_if *head) { - struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head->head); + struct ras_manager *obj = amdgpu_ras_find_obj(adev, head); if (!obj || obj->attr_inuse) return -EINVAL; get_obj(obj); - memcpy(obj->fs_data.sysfs_name, - head->sysfs_name, - sizeof(obj->fs_data.sysfs_name)); + snprintf(obj->fs_data.sysfs_name, sizeof(obj->fs_data.sysfs_name), + "%s_err_count", head->name); obj->sysfs_attr = (struct device_attribute){ .attr = { @@ -1647,9 +1593,9 @@ int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev, } int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev, - struct ras_ih_if *info) + struct ras_common_if *head) { - struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head); + struct ras_manager *obj = amdgpu_ras_find_obj(adev, head); struct ras_ih_data *data; if (!obj) @@ -1669,24 +1615,27 @@ int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev, } int amdgpu_ras_interrupt_add_handler(struct amdgpu_device *adev, - struct ras_ih_if *info) + struct ras_common_if *head) { - struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head); + struct ras_manager *obj = amdgpu_ras_find_obj(adev, head); struct ras_ih_data *data; + struct amdgpu_ras_block_object *ras_obj; if (!obj) { /* in case we registe the IH before enable ras feature */ - obj = amdgpu_ras_create_obj(adev, &info->head); + obj = amdgpu_ras_create_obj(adev, head); if (!obj) return -EINVAL; } else get_obj(obj); + ras_obj = container_of(head, struct amdgpu_ras_block_object, ras_comm); + data = &obj->ih_data; /* add the callback.etc */ *data = (struct ras_ih_data) { .inuse = 0, - .cb = info->cb, + .cb = ras_obj->ras_cb, .element_size = sizeof(struct amdgpu_iv_entry), .rptr = 0, .wptr = 0, @@ -1715,10 +1664,7 @@ static int amdgpu_ras_interrupt_remove_all(struct amdgpu_device *adev) struct ras_manager *obj, *tmp; list_for_each_entry_safe(obj, tmp, &con->head, node) { - struct ras_ih_if info = { - .head = obj->head, - }; - amdgpu_ras_interrupt_remove_handler(adev, &info); + amdgpu_ras_interrupt_remove_handler(adev, &obj->head); } return 0; @@ -1766,24 +1712,28 @@ static void amdgpu_ras_log_on_err_counter(struct amdgpu_device *adev) static void amdgpu_ras_error_status_query(struct amdgpu_device *adev, struct ras_query_if *info) { + struct amdgpu_ras_block_object *block_obj; /* * Only two block need to query read/write * RspStatus at current state */ - switch (info->head.block) { - case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->query_ras_error_status) - adev->gfx.ras_funcs->query_ras_error_status(adev); - break; - case AMDGPU_RAS_BLOCK__MMHUB: - if (adev->mmhub.ras_funcs && - adev->mmhub.ras_funcs->query_ras_error_status) - adev->mmhub.ras_funcs->query_ras_error_status(adev); - break; - default: - break; + if ((info->head.block != AMDGPU_RAS_BLOCK__GFX) && + (info->head.block != AMDGPU_RAS_BLOCK__MMHUB)) + return; + + block_obj = amdgpu_ras_get_ras_block(adev, + info->head.block, + info->head.sub_block_index); + + if (!block_obj || !block_obj->hw_ops) { + dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", + get_ras_block_str(&info->head)); + return; } + + if (block_obj->hw_ops->query_ras_error_status) + block_obj->hw_ops->query_ras_error_status(adev); + } static void amdgpu_ras_query_err_status(struct amdgpu_device *adev) @@ -2118,6 +2068,7 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev) mutex_init(&con->recovery_lock); INIT_WORK(&con->recovery_work, amdgpu_ras_do_recovery); atomic_set(&con->in_recovery, 0); + con->eeprom_control.bad_channel_bitmap = 0; max_eeprom_records_count = amdgpu_ras_eeprom_max_record_count(); amdgpu_ras_validate_threshold(adev, max_eeprom_records_count); @@ -2141,8 +2092,12 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev) if (ret) goto free; - if (adev->smu.ppt_funcs && adev->smu.ppt_funcs->send_hbm_bad_pages_num) - adev->smu.ppt_funcs->send_hbm_bad_pages_num(&adev->smu, con->eeprom_control.ras_num_recs); + amdgpu_dpm_send_hbm_bad_pages_num(adev, con->eeprom_control.ras_num_recs); + + if (con->update_channel_flag == true) { + amdgpu_dpm_send_hbm_bad_channel_flag(adev, con->eeprom_control.bad_channel_bitmap); + con->update_channel_flag = false; + } } #ifdef CONFIG_X86_MCE_AMD @@ -2336,6 +2291,7 @@ int amdgpu_ras_init(struct amdgpu_device *adev) goto release_con; } + con->update_channel_flag = false; con->features = 0; INIT_LIST_HEAD(&con->head); /* Might need get this flag from vbios. */ @@ -2348,24 +2304,27 @@ int amdgpu_ras_init(struct amdgpu_device *adev) case CHIP_VEGA20: case CHIP_ARCTURUS: case CHIP_ALDEBARAN: - if (!adev->gmc.xgmi.connected_to_cpu) - adev->nbio.ras_funcs = &nbio_v7_4_ras_funcs; + if (!adev->gmc.xgmi.connected_to_cpu) { + adev->nbio.ras = &nbio_v7_4_ras; + amdgpu_ras_register_ras_block(adev, &adev->nbio.ras->ras_block); + adev->nbio.ras_if = &adev->nbio.ras->ras_block.ras_comm; + } break; default: /* nbio ras is not available */ break; } - if (adev->nbio.ras_funcs && - adev->nbio.ras_funcs->init_ras_controller_interrupt) { - r = adev->nbio.ras_funcs->init_ras_controller_interrupt(adev); + if (adev->nbio.ras && + adev->nbio.ras->init_ras_controller_interrupt) { + r = adev->nbio.ras->init_ras_controller_interrupt(adev); if (r) goto release_con; } - if (adev->nbio.ras_funcs && - adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt) { - r = adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt(adev); + if (adev->nbio.ras && + adev->nbio.ras->init_ras_err_event_athub_interrupt) { + r = adev->nbio.ras->init_ras_err_event_athub_interrupt(adev); if (r) goto release_con; } @@ -2377,12 +2336,12 @@ int amdgpu_ras_init(struct amdgpu_device *adev) } else if (adev->df.funcs && adev->df.funcs->query_ras_poison_mode && - adev->umc.ras_funcs && - adev->umc.ras_funcs->query_ras_poison_mode) { + adev->umc.ras && + adev->umc.ras->query_ras_poison_mode) { df_poison = adev->df.funcs->query_ras_poison_mode(adev); umc_poison = - adev->umc.ras_funcs->query_ras_poison_mode(adev); + adev->umc.ras->query_ras_poison_mode(adev); /* Only poison is set in both DF and UMC, we can support it */ if (df_poison && umc_poison) con->poison_supported = true; @@ -2445,11 +2404,10 @@ bool amdgpu_ras_is_poison_mode_supported(struct amdgpu_device *adev) } /* helper function to handle common stuff in ip late init phase */ -int amdgpu_ras_late_init(struct amdgpu_device *adev, - struct ras_common_if *ras_block, - struct ras_fs_if *fs_info, - struct ras_ih_if *ih_info) +int amdgpu_ras_block_late_init(struct amdgpu_device *adev, + struct ras_common_if *ras_block) { + struct amdgpu_ras_block_object *ras_obj = NULL; struct amdgpu_ras *con = amdgpu_ras_get_context(adev); unsigned long ue_count, ce_count; int r; @@ -2477,15 +2435,16 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev, if (adev->in_suspend || amdgpu_in_reset(adev)) return 0; - if (ih_info->cb) { - r = amdgpu_ras_interrupt_add_handler(adev, ih_info); + ras_obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm); + if (ras_obj->ras_cb) { + r = amdgpu_ras_interrupt_add_handler(adev, ras_block); if (r) - goto interrupt; + goto cleanup; } - r = amdgpu_ras_sysfs_create(adev, fs_info); + r = amdgpu_ras_sysfs_create(adev, ras_block); if (r) - goto sysfs; + goto interrupt; /* Those are the cached values at init. */ @@ -2495,27 +2454,40 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev, } return 0; -cleanup: - amdgpu_ras_sysfs_remove(adev, ras_block); -sysfs: - if (ih_info->cb) - amdgpu_ras_interrupt_remove_handler(adev, ih_info); + interrupt: + if (ras_obj->ras_cb) + amdgpu_ras_interrupt_remove_handler(adev, ras_block); +cleanup: amdgpu_ras_feature_enable(adev, ras_block, 0); return r; } +static int amdgpu_ras_block_late_init_default(struct amdgpu_device *adev, + struct ras_common_if *ras_block) +{ + return amdgpu_ras_block_late_init(adev, ras_block); +} + /* helper function to remove ras fs node and interrupt handler */ -void amdgpu_ras_late_fini(struct amdgpu_device *adev, - struct ras_common_if *ras_block, - struct ras_ih_if *ih_info) +void amdgpu_ras_block_late_fini(struct amdgpu_device *adev, + struct ras_common_if *ras_block) { - if (!ras_block || !ih_info) + struct amdgpu_ras_block_object *ras_obj; + if (!ras_block) return; amdgpu_ras_sysfs_remove(adev, ras_block); - if (ih_info->cb) - amdgpu_ras_interrupt_remove_handler(adev, ih_info); + + ras_obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm); + if (ras_obj->ras_cb) + amdgpu_ras_interrupt_remove_handler(adev, ras_block); +} + +static void amdgpu_ras_block_late_fini_default(struct amdgpu_device *adev, + struct ras_common_if *ras_block) +{ + return amdgpu_ras_block_late_fini(adev, ras_block); } /* do some init work after IP late init as dependence. @@ -2568,6 +2540,33 @@ void amdgpu_ras_suspend(struct amdgpu_device *adev) amdgpu_ras_disable_all_features(adev, 1); } +int amdgpu_ras_late_init(struct amdgpu_device *adev) +{ + struct amdgpu_ras_block_list *node, *tmp; + struct amdgpu_ras_block_object *obj; + int r; + + list_for_each_entry_safe(node, tmp, &adev->ras_list, node) { + if (!node->ras_obj) { + dev_warn(adev->dev, "Warning: abnormal ras list node.\n"); + continue; + } + + obj = node->ras_obj; + if (obj->ras_late_init) { + r = obj->ras_late_init(adev, &obj->ras_comm); + if (r) { + dev_err(adev->dev, "%s failed to execute ras_late_init! ret:%d\n", + obj->ras_comm.name, r); + return r; + } + } else + amdgpu_ras_block_late_init_default(adev, &obj->ras_comm); + } + + return 0; +} + /* do some fini work before IP fini as dependence */ int amdgpu_ras_pre_fini(struct amdgpu_device *adev) { @@ -2585,11 +2584,28 @@ int amdgpu_ras_pre_fini(struct amdgpu_device *adev) int amdgpu_ras_fini(struct amdgpu_device *adev) { + struct amdgpu_ras_block_list *ras_node, *tmp; + struct amdgpu_ras_block_object *obj = NULL; struct amdgpu_ras *con = amdgpu_ras_get_context(adev); if (!adev->ras_enabled || !con) return 0; + list_for_each_entry_safe(ras_node, tmp, &adev->ras_list, node) { + if (ras_node->ras_obj) { + obj = ras_node->ras_obj; + if (amdgpu_ras_is_supported(adev, obj->ras_comm.block) && + obj->ras_fini) + obj->ras_fini(adev, &obj->ras_comm); + else + amdgpu_ras_block_late_fini_default(adev, &obj->ras_comm); + } + + /* Clear ras blocks from ras_list and free ras block list node */ + list_del(&ras_node->node); + kfree(ras_node); + } + amdgpu_ras_fs_fini(adev); amdgpu_ras_interrupt_remove_all(adev); @@ -2717,8 +2733,6 @@ static int amdgpu_bad_page_notifier(struct notifier_block *nb, dev_info(adev->dev, "Uncorrectable error detected in UMC inst: %d, chan_idx: %d", umc_inst, ch_inst); - memset(&err_rec, 0x0, sizeof(struct eeprom_table_record)); - /* * Translate UMC channel address to Physical address */ @@ -2730,16 +2744,10 @@ static int amdgpu_bad_page_notifier(struct notifier_block *nb, ADDR_OF_256B_BLOCK(channel_index) | OFFSET_IN_256B_BLOCK(m->addr); - err_rec.address = m->addr; - err_rec.retired_page = retired_page >> AMDGPU_GPU_PAGE_SHIFT; - err_rec.ts = (uint64_t)ktime_get_real_seconds(); - err_rec.err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; - err_rec.cu = 0; - err_rec.mem_channel = channel_index; - err_rec.mcumc_id = umc_inst; - + memset(&err_rec, 0x0, sizeof(struct eeprom_table_record)); err_data.err_addr = &err_rec; - err_data.err_addr_cnt = 1; + amdgpu_umc_fill_error_record(&err_data, m->addr, + retired_page, channel_index, umc_inst); if (amdgpu_bad_page_threshold != 0) { amdgpu_ras_add_bad_pages(adev, err_data.err_addr, @@ -2777,3 +2785,63 @@ static void amdgpu_register_bad_pages_mca_notifier(struct amdgpu_device *adev) } } #endif + +struct amdgpu_ras *amdgpu_ras_get_context(struct amdgpu_device *adev) +{ + if (!adev) + return NULL; + + return adev->psp.ras_context.ras; +} + +int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con) +{ + if (!adev) + return -EINVAL; + + adev->psp.ras_context.ras = ras_con; + return 0; +} + +/* check if ras is supported on block, say, sdma, gfx */ +int amdgpu_ras_is_supported(struct amdgpu_device *adev, + unsigned int block) +{ + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + + if (block >= AMDGPU_RAS_BLOCK_COUNT) + return 0; + return ras && (adev->ras_enabled & (1 << block)); +} + +int amdgpu_ras_reset_gpu(struct amdgpu_device *adev) +{ + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + + if (atomic_cmpxchg(&ras->in_recovery, 0, 1) == 0) + schedule_work(&ras->recovery_work); + return 0; +} + + +/* Register each ip ras block into amdgpu ras */ +int amdgpu_ras_register_ras_block(struct amdgpu_device *adev, + struct amdgpu_ras_block_object *ras_block_obj) +{ + struct amdgpu_ras_block_list *ras_node; + if (!adev || !ras_block_obj) + return -EINVAL; + + if (!amdgpu_ras_asic_supported(adev)) + return 0; + + ras_node = kzalloc(sizeof(*ras_node), GFP_KERNEL); + if (!ras_node) + return -ENOMEM; + + INIT_LIST_HEAD(&ras_node->node); + ras_node->ras_obj = ras_block_obj; + list_add_tail(&ras_node->node, &adev->ras_list); + + return 0; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 1c708122d492..9314fde81e68 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -26,11 +26,11 @@ #include <linux/debugfs.h> #include <linux/list.h> -#include "amdgpu.h" -#include "amdgpu_psp.h" #include "ta_ras_if.h" #include "amdgpu_ras_eeprom.h" +struct amdgpu_iv_entry; + #define AMDGPU_RAS_FLAG_INIT_BY_VBIOS (0x1 << 0) enum amdgpu_ras_block { @@ -374,6 +374,9 @@ struct amdgpu_ras { /* record umc error info queried from smu */ struct umc_ecc_info umc_ecc; + + /* Indicates smu whether need update bad channel info */ + bool update_channel_flag; }; struct ras_fs_data { @@ -484,6 +487,27 @@ struct ras_debug_if { }; int op; }; + +struct amdgpu_ras_block_object { + struct ras_common_if ras_comm; + + int (*ras_block_match)(struct amdgpu_ras_block_object *block_obj, + enum amdgpu_ras_block block, uint32_t sub_block_index); + int (*ras_late_init)(struct amdgpu_device *adev, struct ras_common_if *ras_block); + void (*ras_fini)(struct amdgpu_device *adev, struct ras_common_if *ras_block); + ras_ih_cb ras_cb; + const struct amdgpu_ras_block_hw_ops *hw_ops; +}; + +struct amdgpu_ras_block_hw_ops { + int (*ras_error_inject)(struct amdgpu_device *adev, void *inject_if); + void (*query_ras_error_count)(struct amdgpu_device *adev, void *ras_error_status); + void (*query_ras_error_status)(struct amdgpu_device *adev); + void (*query_ras_error_address)(struct amdgpu_device *adev, void *ras_error_status); + void (*reset_ras_error_count)(struct amdgpu_device *adev); + void (*reset_ras_error_status)(struct amdgpu_device *adev); +}; + /* work flow * vbios * 1: ras feature enable (enabled by default) @@ -498,19 +522,6 @@ struct ras_debug_if { * 8: feature disable */ -#define amdgpu_ras_get_context(adev) ((adev)->psp.ras_context.ras) -#define amdgpu_ras_set_context(adev, ras_con) ((adev)->psp.ras_context.ras = (ras_con)) - -/* check if ras is supported on block, say, sdma, gfx */ -static inline int amdgpu_ras_is_supported(struct amdgpu_device *adev, - unsigned int block) -{ - struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); - - if (block >= AMDGPU_RAS_BLOCK_COUNT) - return 0; - return ras && (adev->ras_enabled & (1 << block)); -} int amdgpu_ras_recovery_init(struct amdgpu_device *adev); @@ -527,15 +538,6 @@ int amdgpu_ras_add_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); - - if (atomic_cmpxchg(&ras->in_recovery, 0, 1) == 0) - schedule_work(&ras->recovery_work); - return 0; -} - static inline enum ta_ras_block amdgpu_ras_block_to_ta(enum amdgpu_ras_block block) { switch (block) { @@ -596,15 +598,15 @@ amdgpu_ras_error_to_ta(enum amdgpu_ras_error_type error) { /* called in ip_init and ip_fini */ int amdgpu_ras_init(struct amdgpu_device *adev); +int amdgpu_ras_late_init(struct amdgpu_device *adev); int amdgpu_ras_fini(struct amdgpu_device *adev); int amdgpu_ras_pre_fini(struct amdgpu_device *adev); -int amdgpu_ras_late_init(struct amdgpu_device *adev, - struct ras_common_if *ras_block, - struct ras_fs_if *fs_info, - struct ras_ih_if *ih_info); -void amdgpu_ras_late_fini(struct amdgpu_device *adev, - struct ras_common_if *ras_block, - struct ras_ih_if *ih_info); + +int amdgpu_ras_block_late_init(struct amdgpu_device *adev, + struct ras_common_if *ras_block); + +void amdgpu_ras_block_late_fini(struct amdgpu_device *adev, + struct ras_common_if *ras_block); int amdgpu_ras_feature_enable(struct amdgpu_device *adev, struct ras_common_if *head, bool enable); @@ -613,7 +615,7 @@ int amdgpu_ras_feature_enable_on_boot(struct amdgpu_device *adev, struct ras_common_if *head, bool enable); int amdgpu_ras_sysfs_create(struct amdgpu_device *adev, - struct ras_fs_if *head); + struct ras_common_if *head); int amdgpu_ras_sysfs_remove(struct amdgpu_device *adev, struct ras_common_if *head); @@ -630,10 +632,10 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, struct ras_inject_if *info); int amdgpu_ras_interrupt_add_handler(struct amdgpu_device *adev, - struct ras_ih_if *info); + struct ras_common_if *head); int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev, - struct ras_ih_if *info); + struct ras_common_if *head); int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev, struct ras_dispatch_if *info); @@ -667,4 +669,14 @@ const char *get_ras_block_str(struct ras_common_if *ras_block); bool amdgpu_ras_is_poison_mode_supported(struct amdgpu_device *adev); +int amdgpu_ras_is_supported(struct amdgpu_device *adev, unsigned int block); + +int amdgpu_ras_reset_gpu(struct amdgpu_device *adev); + +struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev); + +int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con); + +int amdgpu_ras_register_ras_block(struct amdgpu_device *adev, + struct amdgpu_ras_block_object *ras_block_obj); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index d3e055314804..c4283987bb1e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -196,7 +196,7 @@ static int __write_table_header(struct amdgpu_ras_eeprom_control *control) /* i2c may be unstable in gpu reset */ down_read(&adev->reset_domain->sem); - res = amdgpu_eeprom_write(&adev->pm.smu_i2c, + res = amdgpu_eeprom_write(adev->pm.ras_eeprom_i2c_bus, control->i2c_address + control->ras_header_offset, buf, RAS_TABLE_HEADER_SIZE); @@ -265,7 +265,9 @@ static int amdgpu_ras_eeprom_correct_header_tag( */ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control) { + struct amdgpu_device *adev = to_amdgpu_device(control); struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr; + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); u8 csum; int res; @@ -284,6 +286,12 @@ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control) control->ras_num_recs = 0; control->ras_fri = 0; + amdgpu_dpm_send_hbm_bad_pages_num(adev, control->ras_num_recs); + + control->bad_channel_bitmap = 0; + amdgpu_dpm_send_hbm_bad_channel_flag(adev, control->bad_channel_bitmap); + con->update_channel_flag = false; + amdgpu_ras_debugfs_set_ret_size(control); mutex_unlock(&control->ras_tbl_mutex); @@ -391,7 +399,7 @@ static int __amdgpu_ras_eeprom_write(struct amdgpu_ras_eeprom_control *control, /* i2c may be unstable in gpu reset */ down_read(&adev->reset_domain->sem); buf_size = num * RAS_TABLE_RECORD_SIZE; - res = amdgpu_eeprom_write(&adev->pm.smu_i2c, + res = amdgpu_eeprom_write(adev->pm.ras_eeprom_i2c_bus, control->i2c_address + RAS_INDEX_TO_OFFSET(control, fri), buf, buf_size); @@ -417,6 +425,7 @@ amdgpu_ras_eeprom_append_table(struct amdgpu_ras_eeprom_control *control, struct eeprom_table_record *record, const u32 num) { + struct amdgpu_ras *con = amdgpu_ras_get_context(to_amdgpu_device(control)); u32 a, b, i; u8 *buf, *pp; int res; @@ -428,9 +437,16 @@ amdgpu_ras_eeprom_append_table(struct amdgpu_ras_eeprom_control *control, /* Encode all of them in one go. */ pp = buf; - for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) + for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) { __encode_table_record_to_buf(control, &record[i], pp); + /* update bad channel bitmap */ + if (!(control->bad_channel_bitmap & (1 << record[i].mem_channel))) { + control->bad_channel_bitmap |= 1 << record[i].mem_channel; + con->update_channel_flag = true; + } + } + /* a, first record index to write into. * b, last record index to write into. * a = first index to read (fri) + number of records in the table, @@ -550,7 +566,7 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control) } down_read(&adev->reset_domain->sem); - res = amdgpu_eeprom_read(&adev->pm.smu_i2c, + res = amdgpu_eeprom_read(adev->pm.ras_eeprom_i2c_bus, control->i2c_address + control->ras_record_offset, buf, buf_size); @@ -646,7 +662,7 @@ static int __amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control, /* i2c may be unstable in gpu reset */ down_read(&adev->reset_domain->sem); buf_size = num * RAS_TABLE_RECORD_SIZE; - res = amdgpu_eeprom_read(&adev->pm.smu_i2c, + res = amdgpu_eeprom_read(adev->pm.ras_eeprom_i2c_bus, control->i2c_address + RAS_INDEX_TO_OFFSET(control, fri), buf, buf_size); @@ -683,6 +699,7 @@ int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control, const u32 num) { struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); int i, res; u8 *buf, *pp; u32 g0, g1; @@ -750,8 +767,15 @@ int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control, /* Read up everything? Then transform. */ pp = buf; - for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) + for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) { __decode_table_record_from_buf(control, &record[i], pp); + + /* update bad channel bitmap */ + if (!(control->bad_channel_bitmap & (1 << record[i].mem_channel))) { + control->bad_channel_bitmap |= 1 << record[i].mem_channel; + con->update_channel_flag = true; + } + } Out: kfree(buf); mutex_unlock(&control->ras_tbl_mutex); @@ -1011,7 +1035,7 @@ static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control return -ENOMEM; } - res = amdgpu_eeprom_read(&adev->pm.smu_i2c, + res = amdgpu_eeprom_read(adev->pm.ras_eeprom_i2c_bus, control->i2c_address + control->ras_header_offset, buf, buf_size); @@ -1047,7 +1071,7 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control, return 0; /* Verify i2c adapter is initialized */ - if (!adev->pm.smu_i2c.algo) + if (!adev->pm.ras_eeprom_i2c_bus || !adev->pm.ras_eeprom_i2c_bus->algo) return -ENOENT; if (!__get_eeprom_i2c_addr(adev, control)) @@ -1059,7 +1083,7 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control, mutex_init(&control->ras_tbl_mutex); /* Read the table header from EEPROM address */ - res = amdgpu_eeprom_read(&adev->pm.smu_i2c, + res = amdgpu_eeprom_read(adev->pm.ras_eeprom_i2c_bus, control->i2c_address + control->ras_header_offset, buf, RAS_TABLE_HEADER_SIZE); if (res < RAS_TABLE_HEADER_SIZE) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h index 6bb00578bfbb..54d9bfe0881d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h @@ -80,6 +80,10 @@ struct amdgpu_ras_eeprom_control { /* Protect table access via this mutex. */ struct mutex ras_tbl_mutex; + + /* Record channel info which occurred bad pages + */ + u32 bad_channel_bitmap; }; /* diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c index 248d64158721..c80af0889773 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c @@ -36,8 +36,8 @@ int amdgpu_reset_init(struct amdgpu_device *adev) { int ret = 0; - switch (adev->asic_type) { - case CHIP_ALDEBARAN: + switch (adev->ip_versions[MP1_HWIP][0]) { + case IP_VERSION(13, 0, 2): ret = aldebaran_reset_init(adev); break; default: @@ -51,8 +51,8 @@ int amdgpu_reset_fini(struct amdgpu_device *adev) { int ret = 0; - switch (adev->asic_type) { - case CHIP_ALDEBARAN: + switch (adev->ip_versions[MP1_HWIP][0]) { + case IP_VERSION(13, 0, 2): ret = aldebaran_reset_fini(adev); break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c index 35bcb6dc1816..7f33ae87cb41 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c @@ -193,6 +193,7 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring, adev->rings[ring->idx] = ring; ring->num_hw_submission = sched_hw_submission; ring->sched_score = sched_score; + ring->vmid_wait = dma_fence_get_stub(); r = amdgpu_fence_driver_init_ring(ring); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h index 48365da213dc..5320bb0883d8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h @@ -28,6 +28,13 @@ #include <drm/gpu_scheduler.h> #include <drm/drm_print.h> +struct amdgpu_device; +struct amdgpu_ring; +struct amdgpu_ib; +struct amdgpu_cs_parser; +struct amdgpu_job; +struct amdgpu_vm; + /* max number of rings */ #define AMDGPU_MAX_RINGS 28 #define AMDGPU_MAX_HWIP_RINGS 8 @@ -82,11 +89,13 @@ enum amdgpu_ib_pool_type { AMDGPU_IB_POOL_MAX }; -struct amdgpu_device; -struct amdgpu_ring; -struct amdgpu_ib; -struct amdgpu_cs_parser; -struct amdgpu_job; +struct amdgpu_ib { + struct amdgpu_sa_bo *sa_bo; + uint32_t length_dw; + uint64_t gpu_addr; + uint32_t *ptr; + uint32_t flags; +}; struct amdgpu_sched { u32 num_scheds; @@ -111,6 +120,8 @@ struct amdgpu_fence_driver { struct dma_fence **fences; }; +extern const struct drm_sched_backend_ops amdgpu_sched_ops; + void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring); void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring); @@ -144,6 +155,7 @@ struct amdgpu_ring_funcs { u32 nop; bool support_64bit_ptrs; bool no_user_fence; + bool secure_submission_supported; unsigned vmhub; unsigned extra_dw; @@ -152,8 +164,12 @@ struct amdgpu_ring_funcs { u64 (*get_wptr)(struct amdgpu_ring *ring); void (*set_wptr)(struct amdgpu_ring *ring); /* validating and patching of IBs */ - int (*parse_cs)(struct amdgpu_cs_parser *p, uint32_t ib_idx); - int (*patch_cs_in_place)(struct amdgpu_cs_parser *p, uint32_t ib_idx); + int (*parse_cs)(struct amdgpu_cs_parser *p, + struct amdgpu_job *job, + struct amdgpu_ib *ib); + int (*patch_cs_in_place)(struct amdgpu_cs_parser *p, + struct amdgpu_job *job, + struct amdgpu_ib *ib); /* constants to calculate how many DW are needed for an emit */ unsigned emit_frame_size; unsigned emit_ib_size; @@ -253,8 +269,8 @@ struct amdgpu_ring { atomic_t *sched_score; }; -#define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib))) -#define amdgpu_ring_patch_cs_in_place(r, p, ib) ((r)->funcs->patch_cs_in_place((p), (ib))) +#define amdgpu_ring_parse_cs(r, p, job, ib) ((r)->funcs->parse_cs((p), (job), (ib))) +#define amdgpu_ring_patch_cs_in_place(r, p, job, ib) ((r)->funcs->patch_cs_in_place((p), (job), (ib))) #define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r)) #define amdgpu_ring_test_ib(r, t) (r)->funcs->test_ib((r), (t)) #define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r)) @@ -352,4 +368,29 @@ int amdgpu_ring_test_helper(struct amdgpu_ring *ring); void amdgpu_debugfs_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring); + +static inline u32 amdgpu_ib_get_value(struct amdgpu_ib *ib, int idx) +{ + return ib->ptr[idx]; +} + +static inline void amdgpu_ib_set_value(struct amdgpu_ib *ib, int idx, + uint32_t value) +{ + ib->ptr[idx] = value; +} + +int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm, + unsigned size, + enum amdgpu_ib_pool_type pool, + struct amdgpu_ib *ib); +void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib, + struct dma_fence *f); +int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, + struct amdgpu_ib *ibs, struct amdgpu_job *job, + struct dma_fence **f); +int amdgpu_ib_pool_init(struct amdgpu_device *adev); +void amdgpu_ib_pool_fini(struct amdgpu_device *adev); +int amdgpu_ib_ring_tests(struct amdgpu_device *adev); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h index 00afd0dcae86..3f671a62b009 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h @@ -127,11 +127,19 @@ struct amdgpu_rlc_funcs { void (*reset)(struct amdgpu_device *adev); void (*start)(struct amdgpu_device *adev); void (*update_spm_vmid)(struct amdgpu_device *adev, unsigned vmid); - void (*sriov_wreg)(struct amdgpu_device *adev, u32 offset, u32 v, u32 acc_flags, u32 hwip); - u32 (*sriov_rreg)(struct amdgpu_device *adev, u32 offset, u32 acc_flags, u32 hwip); bool (*is_rlcg_access_range)(struct amdgpu_device *adev, uint32_t reg); }; +struct amdgpu_rlcg_reg_access_ctrl { + uint32_t scratch_reg0; + uint32_t scratch_reg1; + uint32_t scratch_reg2; + uint32_t scratch_reg3; + uint32_t grbm_cntl; + uint32_t grbm_idx; + uint32_t spare_int; +}; + struct amdgpu_rlc { /* for power gating */ struct amdgpu_bo *save_restore_obj; @@ -191,6 +199,10 @@ struct amdgpu_rlc { struct amdgpu_bo *rlc_toc_bo; uint64_t rlc_toc_gpu_addr; void *rlc_toc_buf; + + bool rlcg_reg_access_supported; + /* registers for rlcg indirect reg access */ + struct amdgpu_rlcg_reg_access_ctrl reg_access_ctrl; }; void amdgpu_gfx_rlc_enter_safe_mode(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c index 65debb65a5df..e1835fd4b237 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c @@ -87,73 +87,30 @@ uint64_t amdgpu_sdma_get_csa_mc_addr(struct amdgpu_ring *ring, } int amdgpu_sdma_ras_late_init(struct amdgpu_device *adev, - void *ras_ih_info) + struct ras_common_if *ras_block) { int r, i; - struct ras_ih_if *ih_info = (struct ras_ih_if *)ras_ih_info; - struct ras_fs_if fs_info = { - .sysfs_name = "sdma_err_count", - }; - - if (!ih_info) - return -EINVAL; - - if (!adev->sdma.ras_if) { - adev->sdma.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); - if (!adev->sdma.ras_if) - return -ENOMEM; - adev->sdma.ras_if->block = AMDGPU_RAS_BLOCK__SDMA; - adev->sdma.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; - adev->sdma.ras_if->sub_block_index = 0; - } - fs_info.head = ih_info->head = *adev->sdma.ras_if; - r = amdgpu_ras_late_init(adev, adev->sdma.ras_if, - &fs_info, ih_info); + r = amdgpu_ras_block_late_init(adev, ras_block); if (r) - goto free; + return r; - if (amdgpu_ras_is_supported(adev, adev->sdma.ras_if->block)) { + if (amdgpu_ras_is_supported(adev, ras_block->block)) { for (i = 0; i < adev->sdma.num_instances; i++) { r = amdgpu_irq_get(adev, &adev->sdma.ecc_irq, AMDGPU_SDMA_IRQ_INSTANCE0 + i); if (r) goto late_fini; } - } else { - r = 0; - goto free; } return 0; late_fini: - amdgpu_ras_late_fini(adev, adev->sdma.ras_if, ih_info); -free: - kfree(adev->sdma.ras_if); - adev->sdma.ras_if = NULL; + amdgpu_ras_block_late_fini(adev, ras_block); return r; } -void amdgpu_sdma_ras_fini(struct amdgpu_device *adev) -{ - if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA) && - adev->sdma.ras_if) { - struct ras_common_if *ras_if = adev->sdma.ras_if; - struct ras_ih_if ih_info = { - .head = *ras_if, - /* the cb member will not be used by - * amdgpu_ras_interrupt_remove_handler, init it only - * to cheat the check in ras_late_fini - */ - .cb = amdgpu_sdma_process_ras_data_cb, - }; - - amdgpu_ras_late_fini(adev, ras_if, &ih_info); - kfree(ras_if); - } -} - int amdgpu_sdma_process_ras_data_cb(struct amdgpu_device *adev, void *err_data, struct amdgpu_iv_entry *entry) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h index f8fb755e3aa6..53ac3ebae8d6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h @@ -23,6 +23,7 @@ #ifndef __AMDGPU_SDMA_H__ #define __AMDGPU_SDMA_H__ +#include "amdgpu_ras.h" /* max number of IP instances */ #define AMDGPU_MAX_SDMA_INSTANCES 8 @@ -50,13 +51,8 @@ struct amdgpu_sdma_instance { bool burst_nop; }; -struct amdgpu_sdma_ras_funcs { - int (*ras_late_init)(struct amdgpu_device *adev, - void *ras_ih_info); - void (*ras_fini)(struct amdgpu_device *adev); - int (*query_ras_error_count)(struct amdgpu_device *adev, - uint32_t instance, void *ras_error_status); - void (*reset_ras_error_count)(struct amdgpu_device *adev); +struct amdgpu_sdma_ras { + struct amdgpu_ras_block_object ras_block; }; struct amdgpu_sdma { @@ -73,7 +69,7 @@ struct amdgpu_sdma { uint32_t srbm_soft_reset; bool has_page_queue; struct ras_common_if *ras_if; - const struct amdgpu_sdma_ras_funcs *funcs; + struct amdgpu_sdma_ras *ras; }; /* @@ -121,8 +117,7 @@ amdgpu_sdma_get_instance_from_ring(struct amdgpu_ring *ring); int amdgpu_sdma_get_index_from_ring(struct amdgpu_ring *ring, uint32_t *index); uint64_t amdgpu_sdma_get_csa_mc_addr(struct amdgpu_ring *ring, unsigned vmid); int amdgpu_sdma_ras_late_init(struct amdgpu_device *adev, - void *ras_ih_info); -void amdgpu_sdma_ras_fini(struct amdgpu_device *adev); + struct ras_common_if *ras_block); int amdgpu_sdma_process_ras_data_cb(struct amdgpu_device *adev, void *err_data, struct amdgpu_iv_entry *entry); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c deleted file mode 100644 index 909d830b513e..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c +++ /dev/null @@ -1,250 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0 OR MIT -/* - * Copyright 2009 VMware, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Michel Dänzer - */ - -#include <drm/amdgpu_drm.h> -#include "amdgpu.h" -#include "amdgpu_uvd.h" -#include "amdgpu_vce.h" - -/* Test BO GTT->VRAM and VRAM->GTT GPU copies across the whole GTT aperture */ -static void amdgpu_do_test_moves(struct amdgpu_device *adev) -{ - struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; - struct amdgpu_bo *vram_obj = NULL; - struct amdgpu_bo **gtt_obj = NULL; - struct amdgpu_bo_param bp; - uint64_t gart_addr, vram_addr; - unsigned n, size; - int i, r; - - size = 1024 * 1024; - - /* Number of tests = - * (Total GTT - gart_pin_size - (2 transfer windows for buffer moves)) / test size - */ - n = adev->gmc.gart_size - atomic64_read(&adev->gart_pin_size); - n -= AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GTT_NUM_TRANSFER_WINDOWS * - AMDGPU_GPU_PAGE_SIZE; - n /= size; - - gtt_obj = kcalloc(n, sizeof(*gtt_obj), GFP_KERNEL); - if (!gtt_obj) { - DRM_ERROR("Failed to allocate %d pointers\n", n); - r = 1; - goto out_cleanup; - } - memset(&bp, 0, sizeof(bp)); - bp.size = size; - bp.byte_align = PAGE_SIZE; - bp.domain = AMDGPU_GEM_DOMAIN_VRAM; - bp.flags = 0; - bp.type = ttm_bo_type_kernel; - bp.resv = NULL; - bp.bo_ptr_size = sizeof(struct amdgpu_bo); - - r = amdgpu_bo_create(adev, &bp, &vram_obj); - if (r) { - DRM_ERROR("Failed to create VRAM object\n"); - goto out_cleanup; - } - r = amdgpu_bo_reserve(vram_obj, false); - if (unlikely(r != 0)) - goto out_unref; - r = amdgpu_bo_pin(vram_obj, AMDGPU_GEM_DOMAIN_VRAM); - if (r) { - DRM_ERROR("Failed to pin VRAM object\n"); - goto out_unres; - } - vram_addr = amdgpu_bo_gpu_offset(vram_obj); - for (i = 0; i < n; i++) { - void *gtt_map, *vram_map; - void **gart_start, **gart_end; - void **vram_start, **vram_end; - struct dma_fence *fence = NULL; - - bp.domain = AMDGPU_GEM_DOMAIN_GTT; - r = amdgpu_bo_create(adev, &bp, gtt_obj + i); - if (r) { - DRM_ERROR("Failed to create GTT object %d\n", i); - goto out_lclean; - } - - r = amdgpu_bo_reserve(gtt_obj[i], false); - if (unlikely(r != 0)) - goto out_lclean_unref; - r = amdgpu_bo_pin(gtt_obj[i], AMDGPU_GEM_DOMAIN_GTT); - if (r) { - DRM_ERROR("Failed to pin GTT object %d\n", i); - goto out_lclean_unres; - } - r = amdgpu_ttm_alloc_gart(>t_obj[i]->tbo); - if (r) { - DRM_ERROR("%p bind failed\n", gtt_obj[i]); - goto out_lclean_unpin; - } - gart_addr = amdgpu_bo_gpu_offset(gtt_obj[i]); - - r = amdgpu_bo_kmap(gtt_obj[i], >t_map); - if (r) { - DRM_ERROR("Failed to map GTT object %d\n", i); - goto out_lclean_unpin; - } - - for (gart_start = gtt_map, gart_end = gtt_map + size; - gart_start < gart_end; - gart_start++) - *gart_start = gart_start; - - amdgpu_bo_kunmap(gtt_obj[i]); - - r = amdgpu_copy_buffer(ring, gart_addr, vram_addr, - size, NULL, &fence, false, false, false); - - if (r) { - DRM_ERROR("Failed GTT->VRAM copy %d\n", i); - goto out_lclean_unpin; - } - - r = dma_fence_wait(fence, false); - if (r) { - DRM_ERROR("Failed to wait for GTT->VRAM fence %d\n", i); - goto out_lclean_unpin; - } - - dma_fence_put(fence); - fence = NULL; - - r = amdgpu_bo_kmap(vram_obj, &vram_map); - if (r) { - DRM_ERROR("Failed to map VRAM object after copy %d\n", i); - goto out_lclean_unpin; - } - - for (gart_start = gtt_map, gart_end = gtt_map + size, - vram_start = vram_map, vram_end = vram_map + size; - vram_start < vram_end; - gart_start++, vram_start++) { - if (*vram_start != gart_start) { - DRM_ERROR("Incorrect GTT->VRAM copy %d: Got 0x%p, " - "expected 0x%p (GTT/VRAM offset " - "0x%16llx/0x%16llx)\n", - i, *vram_start, gart_start, - (unsigned long long) - (gart_addr - adev->gmc.gart_start + - (void *)gart_start - gtt_map), - (unsigned long long) - (vram_addr - adev->gmc.vram_start + - (void *)gart_start - gtt_map)); - amdgpu_bo_kunmap(vram_obj); - goto out_lclean_unpin; - } - *vram_start = vram_start; - } - - amdgpu_bo_kunmap(vram_obj); - - r = amdgpu_copy_buffer(ring, vram_addr, gart_addr, - size, NULL, &fence, false, false, false); - - if (r) { - DRM_ERROR("Failed VRAM->GTT copy %d\n", i); - goto out_lclean_unpin; - } - - r = dma_fence_wait(fence, false); - if (r) { - DRM_ERROR("Failed to wait for VRAM->GTT fence %d\n", i); - goto out_lclean_unpin; - } - - dma_fence_put(fence); - fence = NULL; - - r = amdgpu_bo_kmap(gtt_obj[i], >t_map); - if (r) { - DRM_ERROR("Failed to map GTT object after copy %d\n", i); - goto out_lclean_unpin; - } - - for (gart_start = gtt_map, gart_end = gtt_map + size, - vram_start = vram_map, vram_end = vram_map + size; - gart_start < gart_end; - gart_start++, vram_start++) { - if (*gart_start != vram_start) { - DRM_ERROR("Incorrect VRAM->GTT copy %d: Got 0x%p, " - "expected 0x%p (VRAM/GTT offset " - "0x%16llx/0x%16llx)\n", - i, *gart_start, vram_start, - (unsigned long long) - (vram_addr - adev->gmc.vram_start + - (void *)vram_start - vram_map), - (unsigned long long) - (gart_addr - adev->gmc.gart_start + - (void *)vram_start - vram_map)); - amdgpu_bo_kunmap(gtt_obj[i]); - goto out_lclean_unpin; - } - } - - amdgpu_bo_kunmap(gtt_obj[i]); - - DRM_INFO("Tested GTT->VRAM and VRAM->GTT copy for GTT offset 0x%llx\n", - gart_addr - adev->gmc.gart_start); - continue; - -out_lclean_unpin: - amdgpu_bo_unpin(gtt_obj[i]); -out_lclean_unres: - amdgpu_bo_unreserve(gtt_obj[i]); -out_lclean_unref: - amdgpu_bo_unref(>t_obj[i]); -out_lclean: - for (--i; i >= 0; --i) { - amdgpu_bo_unpin(gtt_obj[i]); - amdgpu_bo_unreserve(gtt_obj[i]); - amdgpu_bo_unref(>t_obj[i]); - } - if (fence) - dma_fence_put(fence); - break; - } - - amdgpu_bo_unpin(vram_obj); -out_unres: - amdgpu_bo_unreserve(vram_obj); -out_unref: - amdgpu_bo_unref(&vram_obj); -out_cleanup: - kfree(gtt_obj); - if (r) { - pr_warn("Error while testing BO move\n"); - } -} - -void amdgpu_test_moves(struct amdgpu_device *adev) -{ - if (adev->mman.buffer_funcs) - amdgpu_do_test_moves(adev); -} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h index d855cb53c7e0..06dfcf297a8d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h @@ -358,11 +358,10 @@ TRACE_EVENT(amdgpu_vm_update_ptes, } ), TP_printk("pid:%u vm_ctx:0x%llx start:0x%010llx end:0x%010llx," - " flags:0x%llx, incr:%llu, dst:\n%s%s", __entry->pid, + " flags:0x%llx, incr:%llu, dst:\n%s", __entry->pid, __entry->vm_ctx, __entry->start, __entry->end, __entry->flags, __entry->incr, __print_array( - __get_dynamic_array(dst), min(__entry->nptes, 32u), 8), - __entry->nptes > 32 ? "..." : "") + __get_dynamic_array(dst), __entry->nptes, 8)) ); TRACE_EVENT(amdgpu_vm_set_ptes, @@ -537,6 +536,22 @@ TRACE_EVENT(amdgpu_ib_pipe_sync, __entry->seqno) ); +TRACE_EVENT(amdgpu_reset_reg_dumps, + TP_PROTO(uint32_t address, uint32_t value), + TP_ARGS(address, value), + TP_STRUCT__entry( + __field(uint32_t, address) + __field(uint32_t, value) + ), + TP_fast_assign( + __entry->address = address; + __entry->value = value; + ), + TP_printk("amdgpu register dump 0x%x: 0x%x", + __entry->address, + __entry->value) +); + #undef AMDGPU_JOB_GET_TIMELINE_NAME #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c index 57c6c39ba064..b96d885f6e33 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c @@ -23,6 +23,7 @@ */ #include <drm/amdgpu_drm.h> +#include "amdgpu_cs.h" #include "amdgpu.h" #define CREATE_TRACE_POINTS diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 57ac118fc266..f7f149588432 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -50,6 +50,7 @@ #include <drm/ttm/ttm_range_manager.h> #include <drm/amdgpu_drm.h> +#include <drm/drm_drv.h> #include "amdgpu.h" #include "amdgpu_object.h" @@ -170,10 +171,10 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo, * @bo: buffer object to map * @mem: memory object to map * @mm_cur: range to map - * @num_pages: number of pages to map * @window: which GART window to use * @ring: DMA ring to use for the copy * @tmz: if we should setup a TMZ enabled mapping + * @size: in number of bytes to map, out number of bytes mapped * @addr: resulting address inside the MC address space * * Setup one of the GART windows to access a specific piece of memory or return @@ -182,15 +183,14 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo, static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, struct ttm_resource *mem, struct amdgpu_res_cursor *mm_cur, - unsigned num_pages, unsigned window, - struct amdgpu_ring *ring, bool tmz, - uint64_t *addr) + unsigned window, struct amdgpu_ring *ring, + bool tmz, uint64_t *size, uint64_t *addr) { struct amdgpu_device *adev = ring->adev; - struct amdgpu_job *job; - unsigned num_dw, num_bytes; - struct dma_fence *fence; + unsigned offset, num_pages, num_dw, num_bytes; uint64_t src_addr, dst_addr; + struct dma_fence *fence; + struct amdgpu_job *job; void *cpu_addr; uint64_t flags; unsigned int i; @@ -198,7 +198,9 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, BUG_ON(adev->mman.buffer_funcs->copy_max_bytes < AMDGPU_GTT_MAX_TRANSFER_SIZE * 8); - BUG_ON(mem->mem_type == AMDGPU_PL_PREEMPT); + + if (WARN_ON(mem->mem_type == AMDGPU_PL_PREEMPT)) + return -EINVAL; /* Map only what can't be accessed directly */ if (!tmz && mem->start != AMDGPU_BO_INVALID_OFFSET) { @@ -207,10 +209,22 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, return 0; } + + /* + * If start begins at an offset inside the page, then adjust the size + * and addr accordingly + */ + offset = mm_cur->start & ~PAGE_MASK; + + num_pages = PFN_UP(*size + offset); + num_pages = min_t(uint32_t, num_pages, AMDGPU_GTT_MAX_TRANSFER_SIZE); + + *size = min(*size, (uint64_t)num_pages * PAGE_SIZE - offset); + *addr = adev->gmc.gart_start; *addr += (u64)window * AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GPU_PAGE_SIZE; - *addr += mm_cur->start & ~PAGE_MASK; + *addr += offset; num_dw = ALIGN(adev->mman.buffer_funcs->copy_num_dw, 8); num_bytes = num_pages * 8 * AMDGPU_GPU_PAGES_IN_CPU_PAGE; @@ -241,10 +255,7 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, dma_addr_t *dma_addr; dma_addr = &bo->ttm->dma_address[mm_cur->start >> PAGE_SHIFT]; - r = amdgpu_gart_map(adev, 0, num_pages, dma_addr, flags, - cpu_addr); - if (r) - goto error_free; + amdgpu_gart_map(adev, 0, num_pages, dma_addr, flags, cpu_addr); } else { dma_addr_t dma_address; @@ -252,11 +263,8 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, dma_address += adev->vm_manager.vram_base_offset; for (i = 0; i < num_pages; ++i) { - r = amdgpu_gart_map(adev, i << PAGE_SHIFT, 1, - &dma_address, flags, cpu_addr); - if (r) - goto error_free; - + amdgpu_gart_map(adev, i << PAGE_SHIFT, 1, &dma_address, + flags, cpu_addr); dma_address += PAGE_SIZE; } } @@ -297,9 +305,6 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, struct dma_resv *resv, struct dma_fence **f) { - const uint32_t GTT_MAX_BYTES = (AMDGPU_GTT_MAX_TRANSFER_SIZE * - AMDGPU_GPU_PAGE_SIZE); - struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; struct amdgpu_res_cursor src_mm, dst_mm; struct dma_fence *fence = NULL; @@ -315,29 +320,20 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, mutex_lock(&adev->mman.gtt_window_lock); while (src_mm.remaining) { - uint32_t src_page_offset = src_mm.start & ~PAGE_MASK; - uint32_t dst_page_offset = dst_mm.start & ~PAGE_MASK; + uint64_t from, to, cur_size; struct dma_fence *next; - uint32_t cur_size; - uint64_t from, to; - /* Copy size cannot exceed GTT_MAX_BYTES. So if src or dst - * begins at an offset, then adjust the size accordingly - */ - cur_size = max(src_page_offset, dst_page_offset); - cur_size = min(min3(src_mm.size, dst_mm.size, size), - (uint64_t)(GTT_MAX_BYTES - cur_size)); + /* Never copy more than 256MiB at once to avoid a timeout */ + cur_size = min3(src_mm.size, dst_mm.size, 256ULL << 20); /* Map src to window 0 and dst to window 1. */ r = amdgpu_ttm_map_buffer(src->bo, src->mem, &src_mm, - PFN_UP(cur_size + src_page_offset), - 0, ring, tmz, &from); + 0, ring, tmz, &cur_size, &from); if (r) goto error; r = amdgpu_ttm_map_buffer(dst->bo, dst->mem, &dst_mm, - PFN_UP(cur_size + dst_page_offset), - 1, ring, tmz, &to); + 1, ring, tmz, &cur_size, &to); if (r) goto error; @@ -396,8 +392,7 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo, (abo->flags & AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE)) { struct dma_fence *wipe_fence = NULL; - r = amdgpu_fill_buffer(ttm_to_amdgpu_bo(bo), AMDGPU_POISON, - NULL, &wipe_fence); + r = amdgpu_fill_buffer(abo, AMDGPU_POISON, NULL, &wipe_fence); if (r) { goto error; } else if (wipe_fence) { @@ -821,14 +816,13 @@ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_device *bdev, #endif } -static int amdgpu_ttm_gart_bind(struct amdgpu_device *adev, - struct ttm_buffer_object *tbo, - uint64_t flags) +static void amdgpu_ttm_gart_bind(struct amdgpu_device *adev, + struct ttm_buffer_object *tbo, + uint64_t flags) { struct amdgpu_bo *abo = ttm_to_amdgpu_bo(tbo); struct ttm_tt *ttm = tbo->ttm; struct amdgpu_ttm_tt *gtt = (void *)ttm; - int r; if (amdgpu_bo_encrypted(abo)) flags |= AMDGPU_PTE_TMZ; @@ -836,10 +830,8 @@ static int amdgpu_ttm_gart_bind(struct amdgpu_device *adev, if (abo->flags & AMDGPU_GEM_CREATE_CP_MQD_GFX9) { uint64_t page_idx = 1; - r = amdgpu_gart_bind(adev, gtt->offset, page_idx, - gtt->ttm.dma_address, flags); - if (r) - goto gart_bind_fail; + amdgpu_gart_bind(adev, gtt->offset, page_idx, + gtt->ttm.dma_address, flags); /* The memory type of the first page defaults to UC. Now * modify the memory type to NC from the second page of @@ -848,21 +840,13 @@ static int amdgpu_ttm_gart_bind(struct amdgpu_device *adev, flags &= ~AMDGPU_PTE_MTYPE_VG10_MASK; flags |= AMDGPU_PTE_MTYPE_VG10(AMDGPU_MTYPE_NC); - r = amdgpu_gart_bind(adev, - gtt->offset + (page_idx << PAGE_SHIFT), - ttm->num_pages - page_idx, - &(gtt->ttm.dma_address[page_idx]), flags); + amdgpu_gart_bind(adev, gtt->offset + (page_idx << PAGE_SHIFT), + ttm->num_pages - page_idx, + &(gtt->ttm.dma_address[page_idx]), flags); } else { - r = amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages, - gtt->ttm.dma_address, flags); + amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages, + gtt->ttm.dma_address, flags); } - -gart_bind_fail: - if (r) - DRM_ERROR("failed to bind %u pages at 0x%08llX\n", - ttm->num_pages, gtt->offset); - - return r; } /* @@ -878,7 +862,7 @@ static int amdgpu_ttm_backend_bind(struct ttm_device *bdev, struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); struct amdgpu_ttm_tt *gtt = (void*)ttm; uint64_t flags; - int r = 0; + int r; if (!bo_mem) return -EINVAL; @@ -925,14 +909,10 @@ static int amdgpu_ttm_backend_bind(struct ttm_device *bdev, /* bind pages into GART page tables */ gtt->offset = (u64)bo_mem->start << PAGE_SHIFT; - r = amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages, - gtt->ttm.dma_address, flags); - - if (r) - DRM_ERROR("failed to bind %u pages at 0x%08llX\n", - ttm->num_pages, gtt->offset); + amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages, + gtt->ttm.dma_address, flags); gtt->bound = true; - return r; + return 0; } /* @@ -982,12 +962,7 @@ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo) /* Bind pages */ gtt->offset = (u64)tmp->start << PAGE_SHIFT; - r = amdgpu_ttm_gart_bind(adev, bo, flags); - if (unlikely(r)) { - ttm_resource_free(bo, &tmp); - return r; - } - + amdgpu_ttm_gart_bind(adev, bo, flags); amdgpu_gart_invalidate_tlb(adev); ttm_resource_free(bo, &bo->resource); ttm_bo_assign_mem(bo, tmp); @@ -1001,19 +976,16 @@ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo) * Called by amdgpu_gtt_mgr_recover() from amdgpu_device_reset() to * rebind GTT pages during a GPU reset. */ -int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo) +void amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo) { struct amdgpu_device *adev = amdgpu_ttm_adev(tbo->bdev); uint64_t flags; - int r; if (!tbo->ttm) - return 0; + return; flags = amdgpu_ttm_tt_pte_flags(adev, tbo->ttm, tbo->resource); - r = amdgpu_ttm_gart_bind(adev, tbo, flags); - - return r; + amdgpu_ttm_gart_bind(adev, tbo, flags); } /* @@ -1027,7 +999,6 @@ static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev, { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); struct amdgpu_ttm_tt *gtt = (void *)ttm; - int r; /* if the pages have userptr pinning then clear that first */ if (gtt->userptr) { @@ -1047,10 +1018,7 @@ static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev, return; /* unbind shouldn't be done for GDS/GWS/OA in ttm_bo_clean_mm */ - r = amdgpu_gart_unbind(adev, gtt->offset, ttm->num_pages); - if (r) - DRM_ERROR("failed to unbind %u pages at 0x%08llX\n", - gtt->ttm.num_pages, gtt->offset); + amdgpu_gart_unbind(adev, gtt->offset, ttm->num_pages); gtt->bound = false; } @@ -1169,6 +1137,26 @@ static void amdgpu_ttm_tt_unpopulate(struct ttm_device *bdev, } /** + * amdgpu_ttm_tt_get_userptr - Return the userptr GTT ttm_tt for the current + * task + * + * @tbo: The ttm_buffer_object that contains the userptr + * @user_addr: The returned value + */ +int amdgpu_ttm_tt_get_userptr(const struct ttm_buffer_object *tbo, + uint64_t *user_addr) +{ + struct amdgpu_ttm_tt *gtt; + + if (!tbo->ttm) + return -EINVAL; + + gtt = (void *)tbo->ttm; + *user_addr = gtt->userptr; + return 0; +} + +/** * amdgpu_ttm_tt_set_userptr - Initialize userptr GTT ttm_tt for the current * task * @@ -1433,6 +1421,63 @@ static void amdgpu_ttm_vram_mm_access(struct amdgpu_device *adev, loff_t pos, } } +static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo, + unsigned long offset, void *buf, int len, int write) +{ + struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo); + struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev); + struct amdgpu_res_cursor src_mm; + struct amdgpu_job *job; + struct dma_fence *fence; + uint64_t src_addr, dst_addr; + unsigned int num_dw; + int r, idx; + + if (len != PAGE_SIZE) + return -EINVAL; + + if (!adev->mman.sdma_access_ptr) + return -EACCES; + + if (!drm_dev_enter(adev_to_drm(adev), &idx)) + return -ENODEV; + + if (write) + memcpy(adev->mman.sdma_access_ptr, buf, len); + + num_dw = ALIGN(adev->mman.buffer_funcs->copy_num_dw, 8); + r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, AMDGPU_IB_POOL_DELAYED, &job); + if (r) + goto out; + + amdgpu_res_first(abo->tbo.resource, offset, len, &src_mm); + src_addr = amdgpu_ttm_domain_start(adev, bo->resource->mem_type) + src_mm.start; + dst_addr = amdgpu_bo_gpu_offset(adev->mman.sdma_access_bo); + if (write) + swap(src_addr, dst_addr); + + amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr, dst_addr, PAGE_SIZE, false); + + amdgpu_ring_pad_ib(adev->mman.buffer_funcs_ring, &job->ibs[0]); + WARN_ON(job->ibs[0].length_dw > num_dw); + + r = amdgpu_job_submit(job, &adev->mman.entity, AMDGPU_FENCE_OWNER_UNDEFINED, &fence); + if (r) { + amdgpu_job_free(job); + goto out; + } + + if (!dma_fence_wait_timeout(fence, false, adev->sdma_timeout)) + r = -ETIMEDOUT; + dma_fence_put(fence); + + if (!(r || write)) + memcpy(buf, adev->mman.sdma_access_ptr, len); +out: + drm_dev_exit(idx); + return r; +} + /** * amdgpu_ttm_access_memory - Read or Write memory that backs a buffer object. * @@ -1457,6 +1502,10 @@ static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo, if (bo->resource->mem_type != TTM_PL_VRAM) return -EIO; + if (amdgpu_device_has_timeouts_enabled(adev) && + !amdgpu_ttm_access_memory_sdma(bo, offset, buf, len, write)) + return len; + amdgpu_res_first(bo->resource, offset, len, &cursor); while (cursor.remaining) { size_t count, size = cursor.size; @@ -1796,6 +1845,12 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) return r; } + if (amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_GTT, + &adev->mman.sdma_access_bo, NULL, + &adev->mman.sdma_access_ptr)) + DRM_WARN("Debug VRAM access will use slowpath MM access\n"); + return 0; } @@ -1817,6 +1872,8 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev) if (adev->mman.stolen_reserved_size) amdgpu_bo_free_kernel(&adev->mman.stolen_reserved_memory, NULL, NULL); + amdgpu_bo_free_kernel(&adev->mman.sdma_access_bo, NULL, + &adev->mman.sdma_access_ptr); amdgpu_ttm_fw_reserve_vram_fini(adev); if (drm_dev_enter(adev_to_drm(adev), &idx)) { @@ -1887,23 +1944,55 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable) adev->mman.buffer_funcs_enabled = enable; } +static int amdgpu_ttm_prepare_job(struct amdgpu_device *adev, + bool direct_submit, + unsigned int num_dw, + struct dma_resv *resv, + bool vm_needs_flush, + struct amdgpu_job **job) +{ + enum amdgpu_ib_pool_type pool = direct_submit ? + AMDGPU_IB_POOL_DIRECT : + AMDGPU_IB_POOL_DELAYED; + int r; + + r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, pool, job); + if (r) + return r; + + if (vm_needs_flush) { + (*job)->vm_pd_addr = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo ? + adev->gmc.pdb0_bo : + adev->gart.bo); + (*job)->vm_needs_flush = true; + } + if (resv) { + r = amdgpu_sync_resv(adev, &(*job)->sync, resv, + AMDGPU_SYNC_ALWAYS, + AMDGPU_FENCE_OWNER_UNDEFINED); + if (r) { + DRM_ERROR("sync failed (%d).\n", r); + amdgpu_job_free(*job); + return r; + } + } + return 0; +} + int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, struct dma_resv *resv, struct dma_fence **fence, bool direct_submit, bool vm_needs_flush, bool tmz) { - enum amdgpu_ib_pool_type pool = direct_submit ? AMDGPU_IB_POOL_DIRECT : - AMDGPU_IB_POOL_DELAYED; struct amdgpu_device *adev = ring->adev; + unsigned num_loops, num_dw; struct amdgpu_job *job; - uint32_t max_bytes; - unsigned num_loops, num_dw; unsigned i; int r; - if (direct_submit && !ring->sched.ready) { + if (!direct_submit && !ring->sched.ready) { DRM_ERROR("Trying to move memory with ring turned off.\n"); return -EINVAL; } @@ -1911,26 +2000,11 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, max_bytes = adev->mman.buffer_funcs->copy_max_bytes; num_loops = DIV_ROUND_UP(byte_count, max_bytes); num_dw = ALIGN(num_loops * adev->mman.buffer_funcs->copy_num_dw, 8); - - r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, pool, &job); + r = amdgpu_ttm_prepare_job(adev, direct_submit, num_dw, + resv, vm_needs_flush, &job); if (r) return r; - if (vm_needs_flush) { - job->vm_pd_addr = amdgpu_gmc_pd_addr(adev->gmc.pdb0_bo ? - adev->gmc.pdb0_bo : adev->gart.bo); - job->vm_needs_flush = true; - } - if (resv) { - r = amdgpu_sync_resv(adev, &job->sync, resv, - AMDGPU_SYNC_ALWAYS, - AMDGPU_FENCE_OWNER_UNDEFINED); - if (r) { - DRM_ERROR("sync failed (%d).\n", r); - goto error_free; - } - } - for (i = 0; i < num_loops; i++) { uint32_t cur_size_in_bytes = min(byte_count, max_bytes); @@ -1960,77 +2034,35 @@ error_free: return r; } -int amdgpu_fill_buffer(struct amdgpu_bo *bo, - uint32_t src_data, - struct dma_resv *resv, - struct dma_fence **fence) +static int amdgpu_ttm_fill_mem(struct amdgpu_ring *ring, uint32_t src_data, + uint64_t dst_addr, uint32_t byte_count, + struct dma_resv *resv, + struct dma_fence **fence, + bool vm_needs_flush) { - struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); - uint32_t max_bytes = adev->mman.buffer_funcs->fill_max_bytes; - struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; - - struct amdgpu_res_cursor cursor; + struct amdgpu_device *adev = ring->adev; unsigned int num_loops, num_dw; - uint64_t num_bytes; - struct amdgpu_job *job; + uint32_t max_bytes; + unsigned int i; int r; - if (!adev->mman.buffer_funcs_enabled) { - DRM_ERROR("Trying to clear memory with ring turned off.\n"); - return -EINVAL; - } - - if (bo->tbo.resource->mem_type == AMDGPU_PL_PREEMPT) { - DRM_ERROR("Trying to clear preemptible memory.\n"); - return -EINVAL; - } - - if (bo->tbo.resource->mem_type == TTM_PL_TT) { - r = amdgpu_ttm_alloc_gart(&bo->tbo); - if (r) - return r; - } - - num_bytes = bo->tbo.resource->num_pages << PAGE_SHIFT; - num_loops = 0; - - amdgpu_res_first(bo->tbo.resource, 0, num_bytes, &cursor); - while (cursor.remaining) { - num_loops += DIV_ROUND_UP_ULL(cursor.size, max_bytes); - amdgpu_res_next(&cursor, cursor.size); - } - num_dw = num_loops * adev->mman.buffer_funcs->fill_num_dw; - - /* for IB padding */ - num_dw += 64; - - r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, AMDGPU_IB_POOL_DELAYED, - &job); + max_bytes = adev->mman.buffer_funcs->fill_max_bytes; + num_loops = DIV_ROUND_UP_ULL(byte_count, max_bytes); + num_dw = ALIGN(num_loops * adev->mman.buffer_funcs->fill_num_dw, 8); + r = amdgpu_ttm_prepare_job(adev, false, num_dw, resv, vm_needs_flush, + &job); if (r) return r; - if (resv) { - r = amdgpu_sync_resv(adev, &job->sync, resv, - AMDGPU_SYNC_ALWAYS, - AMDGPU_FENCE_OWNER_UNDEFINED); - if (r) { - DRM_ERROR("sync failed (%d).\n", r); - goto error_free; - } - } - - amdgpu_res_first(bo->tbo.resource, 0, num_bytes, &cursor); - while (cursor.remaining) { - uint32_t cur_size = min_t(uint64_t, cursor.size, max_bytes); - uint64_t dst_addr = cursor.start; + for (i = 0; i < num_loops; i++) { + uint32_t cur_size = min(byte_count, max_bytes); - dst_addr += amdgpu_ttm_domain_start(adev, - bo->tbo.resource->mem_type); amdgpu_emit_fill_buffer(adev, &job->ibs[0], src_data, dst_addr, cur_size); - amdgpu_res_next(&cursor, cur_size); + dst_addr += cur_size; + byte_count -= cur_size; } amdgpu_ring_pad_ib(ring, &job->ibs[0]); @@ -2047,6 +2079,55 @@ error_free: return r; } +int amdgpu_fill_buffer(struct amdgpu_bo *bo, + uint32_t src_data, + struct dma_resv *resv, + struct dma_fence **f) +{ + struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); + struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; + struct dma_fence *fence = NULL; + struct amdgpu_res_cursor dst; + int r; + + if (!adev->mman.buffer_funcs_enabled) { + DRM_ERROR("Trying to clear memory with ring turned off.\n"); + return -EINVAL; + } + + amdgpu_res_first(bo->tbo.resource, 0, amdgpu_bo_size(bo), &dst); + + mutex_lock(&adev->mman.gtt_window_lock); + while (dst.remaining) { + struct dma_fence *next; + uint64_t cur_size, to; + + /* Never fill more than 256MiB at once to avoid timeouts */ + cur_size = min(dst.size, 256ULL << 20); + + r = amdgpu_ttm_map_buffer(&bo->tbo, bo->tbo.resource, &dst, + 1, ring, false, &cur_size, &to); + if (r) + goto error; + + r = amdgpu_ttm_fill_mem(ring, src_data, to, cur_size, resv, + &next, true); + if (r) + goto error; + + dma_fence_put(fence); + fence = next; + + amdgpu_res_next(&dst, cur_size); + } +error: + mutex_unlock(&adev->mman.gtt_window_lock); + if (f) + *f = dma_fence_get(fence); + dma_fence_put(fence); + return r; +} + /** * amdgpu_ttm_evict_resources - evict memory buffers * @adev: amdgpu device object diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index 58c64871c94a..9120ae80ef52 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -91,6 +91,10 @@ struct amdgpu_mman { u64 fw_vram_usage_size; struct amdgpu_bo *fw_vram_usage_reserved_bo; void *fw_vram_usage_va; + + /* PAGE_SIZE'd BO for process memory r/w over SDMA. */ + struct amdgpu_bo *sdma_access_bo; + void *sdma_access_ptr; }; struct amdgpu_copy_mem { @@ -107,7 +111,7 @@ int amdgpu_vram_mgr_init(struct amdgpu_device *adev); void amdgpu_vram_mgr_fini(struct amdgpu_device *adev); bool amdgpu_gtt_mgr_has_gart_addr(struct ttm_resource *mem); -int amdgpu_gtt_mgr_recover(struct amdgpu_gtt_mgr *mgr); +void amdgpu_gtt_mgr_recover(struct amdgpu_gtt_mgr *mgr); uint64_t amdgpu_preempt_mgr_usage(struct ttm_resource_manager *man); @@ -149,7 +153,7 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo, struct dma_fence **fence); int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo); -int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo); +void amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo); uint64_t amdgpu_ttm_domain_start(struct amdgpu_device *adev, uint32_t type); #if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR) @@ -168,6 +172,8 @@ static inline bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm) #endif void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages); +int amdgpu_ttm_tt_get_userptr(const struct ttm_buffer_object *tbo, + uint64_t *user_addr); int amdgpu_ttm_tt_set_userptr(struct ttm_buffer_object *bo, uint64_t addr, uint32_t flags); bool amdgpu_ttm_tt_has_userptr(struct ttm_tt *ttm); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h index 7c2538db3cd5..40dffbac85a0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h @@ -343,7 +343,8 @@ union amdgpu_firmware_header { * fw loading support */ enum AMDGPU_UCODE_ID { - AMDGPU_UCODE_ID_SDMA0 = 0, + AMDGPU_UCODE_ID_CAP = 0, + AMDGPU_UCODE_ID_SDMA0, AMDGPU_UCODE_ID_SDMA1, AMDGPU_UCODE_ID_SDMA2, AMDGPU_UCODE_ID_SDMA3, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index 46264a4002f7..aad3c8b4c810 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -21,7 +21,7 @@ * */ -#include "amdgpu_ras.h" +#include "amdgpu.h" static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, void *ras_error_status, @@ -33,14 +33,14 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, int ret = 0; kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); - ret = smu_get_ecc_info(&adev->smu, (void *)&(con->umc_ecc)); + ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc)); if (ret == -EOPNOTSUPP) { - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->query_ras_error_count) - adev->umc.ras_funcs->query_ras_error_count(adev, ras_error_status); + if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops && + adev->umc.ras->ras_block.hw_ops->query_ras_error_count) + adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev, ras_error_status); - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->query_ras_error_address && + if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops && + adev->umc.ras->ras_block.hw_ops->query_ras_error_address && adev->umc.max_ras_err_cnt_per_query) { err_data->err_addr = kcalloc(adev->umc.max_ras_err_cnt_per_query, @@ -56,15 +56,15 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, /* umc query_ras_error_address is also responsible for clearing * error status */ - adev->umc.ras_funcs->query_ras_error_address(adev, ras_error_status); + adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev, ras_error_status); } } else if (!ret) { - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->ecc_info_query_ras_error_count) - adev->umc.ras_funcs->ecc_info_query_ras_error_count(adev, ras_error_status); + if (adev->umc.ras && + adev->umc.ras->ecc_info_query_ras_error_count) + adev->umc.ras->ecc_info_query_ras_error_count(adev, ras_error_status); - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->ecc_info_query_ras_error_address && + if (adev->umc.ras && + adev->umc.ras->ecc_info_query_ras_error_address && adev->umc.max_ras_err_cnt_per_query) { err_data->err_addr = kcalloc(adev->umc.max_ras_err_cnt_per_query, @@ -80,7 +80,7 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, /* umc query_ras_error_address is also responsible for clearing * error status */ - adev->umc.ras_funcs->ecc_info_query_ras_error_address(adev, ras_error_status); + adev->umc.ras->ecc_info_query_ras_error_address(adev, ras_error_status); } } @@ -96,8 +96,12 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, err_data->err_addr_cnt); amdgpu_ras_save_bad_pages(adev); - if (adev->smu.ppt_funcs && adev->smu.ppt_funcs->send_hbm_bad_pages_num) - adev->smu.ppt_funcs->send_hbm_bad_pages_num(&adev->smu, con->eeprom_control.ras_num_recs); + amdgpu_dpm_send_hbm_bad_pages_num(adev, con->eeprom_control.ras_num_recs); + + if (con->update_channel_flag == true) { + amdgpu_dpm_send_hbm_bad_channel_flag(adev, con->eeprom_control.bad_channel_bitmap); + con->update_channel_flag = false; + } } if (reset) @@ -130,78 +134,39 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev, return ret; } -static int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, +int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, void *ras_error_status, struct amdgpu_iv_entry *entry) { return amdgpu_umc_do_page_retirement(adev, ras_error_status, entry, true); } -int amdgpu_umc_ras_late_init(struct amdgpu_device *adev) +int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block) { int r; - struct ras_fs_if fs_info = { - .sysfs_name = "umc_err_count", - }; - struct ras_ih_if ih_info = { - .cb = amdgpu_umc_process_ras_data_cb, - }; - if (!adev->umc.ras_if) { - adev->umc.ras_if = - kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); - if (!adev->umc.ras_if) - return -ENOMEM; - adev->umc.ras_if->block = AMDGPU_RAS_BLOCK__UMC; - adev->umc.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; - adev->umc.ras_if->sub_block_index = 0; - } - ih_info.head = fs_info.head = *adev->umc.ras_if; - - r = amdgpu_ras_late_init(adev, adev->umc.ras_if, - &fs_info, &ih_info); + r = amdgpu_ras_block_late_init(adev, ras_block); if (r) - goto free; + return r; - if (amdgpu_ras_is_supported(adev, adev->umc.ras_if->block)) { + if (amdgpu_ras_is_supported(adev, ras_block->block)) { r = amdgpu_irq_get(adev, &adev->gmc.ecc_irq, 0); if (r) goto late_fini; - } else { - r = 0; - goto free; } /* ras init of specific umc version */ - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->err_cnt_init) - adev->umc.ras_funcs->err_cnt_init(adev); + if (adev->umc.ras && + adev->umc.ras->err_cnt_init) + adev->umc.ras->err_cnt_init(adev); return 0; late_fini: - amdgpu_ras_late_fini(adev, adev->umc.ras_if, &ih_info); -free: - kfree(adev->umc.ras_if); - adev->umc.ras_if = NULL; + amdgpu_ras_block_late_fini(adev, ras_block); return r; } -void amdgpu_umc_ras_fini(struct amdgpu_device *adev) -{ - if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC) && - adev->umc.ras_if) { - struct ras_common_if *ras_if = adev->umc.ras_if; - struct ras_ih_if ih_info = { - .head = *ras_if, - .cb = amdgpu_umc_process_ras_data_cb, - }; - - amdgpu_ras_late_fini(adev, ras_if, &ih_info); - kfree(ras_if); - } -} - int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) @@ -219,3 +184,24 @@ int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev, amdgpu_ras_interrupt_dispatch(adev, &ih_data); return 0; } + +void amdgpu_umc_fill_error_record(struct ras_err_data *err_data, + uint64_t err_addr, + uint64_t retired_page, + uint32_t channel_index, + uint32_t umc_inst) +{ + struct eeprom_table_record *err_rec = + &err_data->err_addr[err_data->err_addr_cnt]; + + err_rec->address = err_addr; + /* page frame address is saved */ + err_rec->retired_page = retired_page >> AMDGPU_GPU_PAGE_SHIFT; + err_rec->ts = (uint64_t)ktime_get_real_seconds(); + err_rec->err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; + err_rec->cu = 0; + err_rec->mem_channel = channel_index; + err_rec->mcumc_id = umc_inst; + + err_data->err_addr_cnt++; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index b72194e8bfe5..2ec6698aa1fe 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -20,6 +20,7 @@ */ #ifndef __AMDGPU_UMC_H__ #define __AMDGPU_UMC_H__ +#include "amdgpu_ras.h" /* * (addr / 256) * 4096, the higher 26 bits in ErrorAddr @@ -40,14 +41,9 @@ #define LOOP_UMC_CH_INST(ch_inst) for ((ch_inst) = 0; (ch_inst) < adev->umc.channel_inst_num; (ch_inst)++) #define LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) LOOP_UMC_INST((umc_inst)) LOOP_UMC_CH_INST((ch_inst)) -struct amdgpu_umc_ras_funcs { +struct amdgpu_umc_ras { + struct amdgpu_ras_block_object ras_block; void (*err_cnt_init)(struct amdgpu_device *adev); - int (*ras_late_init)(struct amdgpu_device *adev); - void (*ras_fini)(struct amdgpu_device *adev); - void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); - void (*query_ras_error_address)(struct amdgpu_device *adev, - void *ras_error_status); bool (*query_ras_poison_mode)(struct amdgpu_device *adev); void (*ecc_info_query_ras_error_count)(struct amdgpu_device *adev, void *ras_error_status); @@ -73,15 +69,23 @@ struct amdgpu_umc { struct ras_common_if *ras_if; const struct amdgpu_umc_funcs *funcs; - const struct amdgpu_umc_ras_funcs *ras_funcs; + struct amdgpu_umc_ras *ras; }; -int amdgpu_umc_ras_late_init(struct amdgpu_device *adev); -void amdgpu_umc_ras_fini(struct amdgpu_device *adev); +int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block); int amdgpu_umc_poison_handler(struct amdgpu_device *adev, void *ras_error_status, bool reset); int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry); +void amdgpu_umc_fill_error_record(struct ras_err_data *err_data, + uint64_t err_addr, + uint64_t retired_page, + uint32_t channel_index, + uint32_t umc_inst); + +int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, + void *ras_error_status, + struct amdgpu_iv_entry *entry); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c index 6f8de11a17f1..39c74d9fa7cc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c @@ -37,6 +37,7 @@ #include "amdgpu.h" #include "amdgpu_pm.h" #include "amdgpu_uvd.h" +#include "amdgpu_cs.h" #include "cikd.h" #include "uvd/uvd_4_2_d.h" @@ -98,7 +99,7 @@ struct amdgpu_uvd_cs_ctx { unsigned reg, count; unsigned data0, data1; unsigned idx; - unsigned ib_idx; + struct amdgpu_ib *ib; /* does the IB has a msg command */ bool has_msg_cmd; @@ -557,8 +558,8 @@ static u64 amdgpu_uvd_get_addr_from_ctx(struct amdgpu_uvd_cs_ctx *ctx) uint32_t lo, hi; uint64_t addr; - lo = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data0); - hi = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data1); + lo = amdgpu_ib_get_value(ctx->ib, ctx->data0); + hi = amdgpu_ib_get_value(ctx->ib, ctx->data1); addr = ((uint64_t)lo) | (((uint64_t)hi) << 32); return addr; @@ -589,7 +590,7 @@ static int amdgpu_uvd_cs_pass1(struct amdgpu_uvd_cs_ctx *ctx) if (!ctx->parser->adev->uvd.address_64_bit) { /* check if it's a message or feedback command */ - cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx) >> 1; + cmd = amdgpu_ib_get_value(ctx->ib, ctx->idx) >> 1; if (cmd == 0x0 || cmd == 0x3) { /* yes, force it into VRAM */ uint32_t domain = AMDGPU_GEM_DOMAIN_VRAM; @@ -834,6 +835,7 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx, handle = msg[2]; if (handle == 0) { + amdgpu_bo_kunmap(bo); DRM_ERROR("Invalid UVD handle!\n"); return -EINVAL; } @@ -892,6 +894,7 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx, DRM_ERROR("Illegal UVD message type (%d)!\n", msg_type); } + amdgpu_bo_kunmap(bo); return -EINVAL; } @@ -925,12 +928,10 @@ static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx) addr -= mapping->start * AMDGPU_GPU_PAGE_SIZE; start += addr; - amdgpu_set_ib_value(ctx->parser, ctx->ib_idx, ctx->data0, - lower_32_bits(start)); - amdgpu_set_ib_value(ctx->parser, ctx->ib_idx, ctx->data1, - upper_32_bits(start)); + amdgpu_ib_set_value(ctx->ib, ctx->data0, lower_32_bits(start)); + amdgpu_ib_set_value(ctx->ib, ctx->data1, upper_32_bits(start)); - cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx) >> 1; + cmd = amdgpu_ib_get_value(ctx->ib, ctx->idx) >> 1; if (cmd < 0x4) { if ((end - start) < ctx->buf_sizes[cmd]) { DRM_ERROR("buffer (%d) to small (%d / %d)!\n", cmd, @@ -990,14 +991,13 @@ static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx) static int amdgpu_uvd_cs_reg(struct amdgpu_uvd_cs_ctx *ctx, int (*cb)(struct amdgpu_uvd_cs_ctx *ctx)) { - struct amdgpu_ib *ib = &ctx->parser->job->ibs[ctx->ib_idx]; int i, r; ctx->idx++; for (i = 0; i <= ctx->count; ++i) { unsigned reg = ctx->reg + i; - if (ctx->idx >= ib->length_dw) { + if (ctx->idx >= ctx->ib->length_dw) { DRM_ERROR("Register command after end of CS!\n"); return -EINVAL; } @@ -1037,11 +1037,10 @@ static int amdgpu_uvd_cs_reg(struct amdgpu_uvd_cs_ctx *ctx, static int amdgpu_uvd_cs_packets(struct amdgpu_uvd_cs_ctx *ctx, int (*cb)(struct amdgpu_uvd_cs_ctx *ctx)) { - struct amdgpu_ib *ib = &ctx->parser->job->ibs[ctx->ib_idx]; int r; - for (ctx->idx = 0 ; ctx->idx < ib->length_dw; ) { - uint32_t cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx); + for (ctx->idx = 0 ; ctx->idx < ctx->ib->length_dw; ) { + uint32_t cmd = amdgpu_ib_get_value(ctx->ib, ctx->idx); unsigned type = CP_PACKET_GET_TYPE(cmd); switch (type) { case PACKET_TYPE0: @@ -1066,11 +1065,14 @@ static int amdgpu_uvd_cs_packets(struct amdgpu_uvd_cs_ctx *ctx, * amdgpu_uvd_ring_parse_cs - UVD command submission parser * * @parser: Command submission parser context - * @ib_idx: Which indirect buffer to use + * @job: the job to parse + * @ib: the IB to patch * * Parse the command stream, patch in addresses as necessary. */ -int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx) +int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, + struct amdgpu_job *job, + struct amdgpu_ib *ib) { struct amdgpu_uvd_cs_ctx ctx = {}; unsigned buf_sizes[] = { @@ -1080,10 +1082,9 @@ int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx) [0x00000003] = 2048, [0x00000004] = 0xFFFFFFFF, }; - struct amdgpu_ib *ib = &parser->job->ibs[ib_idx]; int r; - parser->job->vm = NULL; + job->vm = NULL; ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo); if (ib->length_dw % 16) { @@ -1094,7 +1095,7 @@ int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx) ctx.parser = parser; ctx.buf_sizes = buf_sizes; - ctx.ib_idx = ib_idx; + ctx.ib = ib; /* first round only required on chips without UVD 64 bit address support */ if (!parser->adev->uvd.address_64_bit) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h index 76ac9699885d..9f89bb7cd60b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h @@ -82,7 +82,9 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle, bool direct, struct dma_fence **fence); void amdgpu_uvd_free_handles(struct amdgpu_device *adev, struct drm_file *filp); -int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx); +int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, + struct amdgpu_job *job, + struct amdgpu_ib *ib); void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring); void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring); int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index 344f711ad144..02cb3a12dd76 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -34,6 +34,7 @@ #include "amdgpu.h" #include "amdgpu_pm.h" #include "amdgpu_vce.h" +#include "amdgpu_cs.h" #include "cikd.h" /* 1 second timeout */ @@ -587,8 +588,7 @@ err: /** * amdgpu_vce_validate_bo - make sure not to cross 4GB boundary * - * @p: parser context - * @ib_idx: indirect buffer to use + * @ib: indirect buffer to use * @lo: address of lower dword * @hi: address of higher dword * @size: minimum size @@ -596,8 +596,9 @@ err: * * Make sure that no BO cross a 4GB boundary. */ -static int amdgpu_vce_validate_bo(struct amdgpu_cs_parser *p, uint32_t ib_idx, - int lo, int hi, unsigned size, int32_t index) +static int amdgpu_vce_validate_bo(struct amdgpu_cs_parser *p, + struct amdgpu_ib *ib, int lo, int hi, + unsigned size, int32_t index) { int64_t offset = ((uint64_t)size) * ((int64_t)index); struct ttm_operation_ctx ctx = { false, false }; @@ -607,8 +608,8 @@ static int amdgpu_vce_validate_bo(struct amdgpu_cs_parser *p, uint32_t ib_idx, uint64_t addr; int r; - addr = ((uint64_t)amdgpu_get_ib_value(p, ib_idx, lo)) | - ((uint64_t)amdgpu_get_ib_value(p, ib_idx, hi)) << 32; + addr = ((uint64_t)amdgpu_ib_get_value(ib, lo)) | + ((uint64_t)amdgpu_ib_get_value(ib, hi)) << 32; if (index >= 0) { addr += offset; fpfn = PAGE_ALIGN(offset) >> PAGE_SHIFT; @@ -638,7 +639,7 @@ static int amdgpu_vce_validate_bo(struct amdgpu_cs_parser *p, uint32_t ib_idx, * amdgpu_vce_cs_reloc - command submission relocation * * @p: parser context - * @ib_idx: indirect buffer to use + * @ib: indirect buffer to use * @lo: address of lower dword * @hi: address of higher dword * @size: minimum size @@ -646,7 +647,7 @@ static int amdgpu_vce_validate_bo(struct amdgpu_cs_parser *p, uint32_t ib_idx, * * Patch relocation inside command stream with real buffer address */ -static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, +static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, struct amdgpu_ib *ib, int lo, int hi, unsigned size, uint32_t index) { struct amdgpu_bo_va_mapping *mapping; @@ -657,8 +658,8 @@ static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, if (index == 0xffffffff) index = 0; - addr = ((uint64_t)amdgpu_get_ib_value(p, ib_idx, lo)) | - ((uint64_t)amdgpu_get_ib_value(p, ib_idx, hi)) << 32; + addr = ((uint64_t)amdgpu_ib_get_value(ib, lo)) | + ((uint64_t)amdgpu_ib_get_value(ib, hi)) << 32; addr += ((uint64_t)size) * ((uint64_t)index); r = amdgpu_cs_find_mapping(p, addr, &bo, &mapping); @@ -679,8 +680,8 @@ static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, addr += amdgpu_bo_gpu_offset(bo); addr -= ((uint64_t)size) * ((uint64_t)index); - amdgpu_set_ib_value(p, ib_idx, lo, lower_32_bits(addr)); - amdgpu_set_ib_value(p, ib_idx, hi, upper_32_bits(addr)); + amdgpu_ib_set_value(ib, lo, lower_32_bits(addr)); + amdgpu_ib_set_value(ib, hi, upper_32_bits(addr)); return 0; } @@ -729,11 +730,13 @@ static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p, * amdgpu_vce_ring_parse_cs - parse and validate the command stream * * @p: parser context - * @ib_idx: indirect buffer to use + * @job: the job to parse + * @ib: the IB to patch */ -int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) +int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, + struct amdgpu_job *job, + struct amdgpu_ib *ib) { - struct amdgpu_ib *ib = &p->job->ibs[ib_idx]; unsigned fb_idx = 0, bs_idx = 0; int session_idx = -1; uint32_t destroyed = 0; @@ -744,12 +747,12 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) unsigned idx; int i, r = 0; - p->job->vm = NULL; + job->vm = NULL; ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo); for (idx = 0; idx < ib->length_dw;) { - uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx); - uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1); + uint32_t len = amdgpu_ib_get_value(ib, idx); + uint32_t cmd = amdgpu_ib_get_value(ib, idx + 1); if ((len < 8) || (len & 3)) { DRM_ERROR("invalid VCE command length (%d)!\n", len); @@ -759,52 +762,52 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) switch (cmd) { case 0x00000002: /* task info */ - fb_idx = amdgpu_get_ib_value(p, ib_idx, idx + 6); - bs_idx = amdgpu_get_ib_value(p, ib_idx, idx + 7); + fb_idx = amdgpu_ib_get_value(ib, idx + 6); + bs_idx = amdgpu_ib_get_value(ib, idx + 7); break; case 0x03000001: /* encode */ - r = amdgpu_vce_validate_bo(p, ib_idx, idx + 10, - idx + 9, 0, 0); + r = amdgpu_vce_validate_bo(p, ib, idx + 10, idx + 9, + 0, 0); if (r) goto out; - r = amdgpu_vce_validate_bo(p, ib_idx, idx + 12, - idx + 11, 0, 0); + r = amdgpu_vce_validate_bo(p, ib, idx + 12, idx + 11, + 0, 0); if (r) goto out; break; case 0x05000001: /* context buffer */ - r = amdgpu_vce_validate_bo(p, ib_idx, idx + 3, - idx + 2, 0, 0); + r = amdgpu_vce_validate_bo(p, ib, idx + 3, idx + 2, + 0, 0); if (r) goto out; break; case 0x05000004: /* video bitstream buffer */ - tmp = amdgpu_get_ib_value(p, ib_idx, idx + 4); - r = amdgpu_vce_validate_bo(p, ib_idx, idx + 3, idx + 2, + tmp = amdgpu_ib_get_value(ib, idx + 4); + r = amdgpu_vce_validate_bo(p, ib, idx + 3, idx + 2, tmp, bs_idx); if (r) goto out; break; case 0x05000005: /* feedback buffer */ - r = amdgpu_vce_validate_bo(p, ib_idx, idx + 3, idx + 2, + r = amdgpu_vce_validate_bo(p, ib, idx + 3, idx + 2, 4096, fb_idx); if (r) goto out; break; case 0x0500000d: /* MV buffer */ - r = amdgpu_vce_validate_bo(p, ib_idx, idx + 3, - idx + 2, 0, 0); + r = amdgpu_vce_validate_bo(p, ib, idx + 3, idx + 2, + 0, 0); if (r) goto out; - r = amdgpu_vce_validate_bo(p, ib_idx, idx + 8, - idx + 7, 0, 0); + r = amdgpu_vce_validate_bo(p, ib, idx + 8, idx + 7, + 0, 0); if (r) goto out; break; @@ -814,12 +817,12 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) } for (idx = 0; idx < ib->length_dw;) { - uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx); - uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1); + uint32_t len = amdgpu_ib_get_value(ib, idx); + uint32_t cmd = amdgpu_ib_get_value(ib, idx + 1); switch (cmd) { case 0x00000001: /* session */ - handle = amdgpu_get_ib_value(p, ib_idx, idx + 2); + handle = amdgpu_ib_get_value(ib, idx + 2); session_idx = amdgpu_vce_validate_handle(p, handle, &allocated); if (session_idx < 0) { @@ -830,8 +833,8 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) break; case 0x00000002: /* task info */ - fb_idx = amdgpu_get_ib_value(p, ib_idx, idx + 6); - bs_idx = amdgpu_get_ib_value(p, ib_idx, idx + 7); + fb_idx = amdgpu_ib_get_value(ib, idx + 6); + bs_idx = amdgpu_ib_get_value(ib, idx + 7); break; case 0x01000001: /* create */ @@ -846,8 +849,8 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) goto out; } - *size = amdgpu_get_ib_value(p, ib_idx, idx + 8) * - amdgpu_get_ib_value(p, ib_idx, idx + 10) * + *size = amdgpu_ib_get_value(ib, idx + 8) * + amdgpu_ib_get_value(ib, idx + 10) * 8 * 3 / 2; break; @@ -876,12 +879,12 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) break; case 0x03000001: /* encode */ - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 10, idx + 9, + r = amdgpu_vce_cs_reloc(p, ib, idx + 10, idx + 9, *size, 0); if (r) goto out; - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 12, idx + 11, + r = amdgpu_vce_cs_reloc(p, ib, idx + 12, idx + 11, *size / 3, 0); if (r) goto out; @@ -892,35 +895,35 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) break; case 0x05000001: /* context buffer */ - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, + r = amdgpu_vce_cs_reloc(p, ib, idx + 3, idx + 2, *size * 2, 0); if (r) goto out; break; case 0x05000004: /* video bitstream buffer */ - tmp = amdgpu_get_ib_value(p, ib_idx, idx + 4); - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, + tmp = amdgpu_ib_get_value(ib, idx + 4); + r = amdgpu_vce_cs_reloc(p, ib, idx + 3, idx + 2, tmp, bs_idx); if (r) goto out; break; case 0x05000005: /* feedback buffer */ - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, + r = amdgpu_vce_cs_reloc(p, ib, idx + 3, idx + 2, 4096, fb_idx); if (r) goto out; break; case 0x0500000d: /* MV buffer */ - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, - idx + 2, *size, 0); + r = amdgpu_vce_cs_reloc(p, ib, idx + 3, + idx + 2, *size, 0); if (r) goto out; - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 8, - idx + 7, *size / 12, 0); + r = amdgpu_vce_cs_reloc(p, ib, idx + 8, + idx + 7, *size / 12, 0); if (r) goto out; break; @@ -965,11 +968,13 @@ out: * amdgpu_vce_ring_parse_cs_vm - parse the command stream in VM mode * * @p: parser context - * @ib_idx: indirect buffer to use + * @job: the job to parse + * @ib: the IB to patch */ -int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx) +int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, + struct amdgpu_job *job, + struct amdgpu_ib *ib) { - struct amdgpu_ib *ib = &p->job->ibs[ib_idx]; int session_idx = -1; uint32_t destroyed = 0; uint32_t created = 0; @@ -978,8 +983,8 @@ int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx) int i, r = 0, idx = 0; while (idx < ib->length_dw) { - uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx); - uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1); + uint32_t len = amdgpu_ib_get_value(ib, idx); + uint32_t cmd = amdgpu_ib_get_value(ib, idx + 1); if ((len < 8) || (len & 3)) { DRM_ERROR("invalid VCE command length (%d)!\n", len); @@ -989,7 +994,7 @@ int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx) switch (cmd) { case 0x00000001: /* session */ - handle = amdgpu_get_ib_value(p, ib_idx, idx + 2); + handle = amdgpu_ib_get_value(ib, idx + 2); session_idx = amdgpu_vce_validate_handle(p, handle, &allocated); if (session_idx < 0) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h index be4a6e773c5b..ea680fc9a6c3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h @@ -59,8 +59,11 @@ int amdgpu_vce_entity_init(struct amdgpu_device *adev); int amdgpu_vce_suspend(struct amdgpu_device *adev); int amdgpu_vce_resume(struct amdgpu_device *adev); void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp); -int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx); -int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx); +int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, struct amdgpu_job *job, + struct amdgpu_ib *ib); +int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, + struct amdgpu_job *job, + struct amdgpu_ib *ib); void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job, struct amdgpu_ib *ib, uint32_t flags); void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index 9a19a6a57b23..f99093f2ebc7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -27,6 +27,7 @@ #include <linux/firmware.h> #include <linux/module.h> #include <linux/pci.h> +#include <linux/debugfs.h> #include <drm/drm_drv.h> #include "amdgpu.h" @@ -51,6 +52,7 @@ #define FIRMWARE_ALDEBARAN "amdgpu/aldebaran_vcn.bin" #define FIRMWARE_BEIGE_GOBY "amdgpu/beige_goby_vcn.bin" #define FIRMWARE_YELLOW_CARP "amdgpu/yellow_carp_vcn.bin" +#define FIRMWARE_VCN_3_1_2 "amdgpu/vcn_3_1_2_vcn.bin" MODULE_FIRMWARE(FIRMWARE_RAVEN); MODULE_FIRMWARE(FIRMWARE_PICASSO); @@ -68,6 +70,7 @@ MODULE_FIRMWARE(FIRMWARE_VANGOGH); MODULE_FIRMWARE(FIRMWARE_DIMGREY_CAVEFISH); MODULE_FIRMWARE(FIRMWARE_BEIGE_GOBY); MODULE_FIRMWARE(FIRMWARE_YELLOW_CARP); +MODULE_FIRMWARE(FIRMWARE_VCN_3_1_2); static void amdgpu_vcn_idle_work_handler(struct work_struct *work); @@ -77,6 +80,7 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) const char *fw_name; const struct common_firmware_header *hdr; unsigned char fw_check; + unsigned int fw_shared_size, log_offset; int i, r; INIT_DELAYED_WORK(&adev->vcn.idle_work, amdgpu_vcn_idle_work_handler); @@ -165,6 +169,12 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)) adev->vcn.indirect_sram = true; break; + case IP_VERSION(3, 1, 2): + fw_name = FIRMWARE_VCN_3_1_2; + 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; } @@ -218,7 +228,12 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) bo_size = AMDGPU_VCN_STACK_SIZE + AMDGPU_VCN_CONTEXT_SIZE; if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8); - bo_size += AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)); + fw_shared_size = AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)); + log_offset = offsetof(struct amdgpu_fw_shared, fw_log); + bo_size += fw_shared_size; + + if (amdgpu_vcnfw_log) + bo_size += AMDGPU_VCNFW_LOG_SIZE; for (i = 0; i < adev->vcn.num_vcn_inst; i++) { if (adev->vcn.harvest_config & (1 << i)) @@ -232,10 +247,18 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) return r; } - adev->vcn.inst[i].fw_shared_cpu_addr = adev->vcn.inst[i].cpu_addr + - bo_size - AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)); - adev->vcn.inst[i].fw_shared_gpu_addr = adev->vcn.inst[i].gpu_addr + - bo_size - AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)); + adev->vcn.inst[i].fw_shared.cpu_addr = adev->vcn.inst[i].cpu_addr + + bo_size - fw_shared_size; + adev->vcn.inst[i].fw_shared.gpu_addr = adev->vcn.inst[i].gpu_addr + + bo_size - fw_shared_size; + + adev->vcn.inst[i].fw_shared.mem_size = fw_shared_size; + + if (amdgpu_vcnfw_log) { + adev->vcn.inst[i].fw_shared.cpu_addr -= AMDGPU_VCNFW_LOG_SIZE; + adev->vcn.inst[i].fw_shared.gpu_addr -= AMDGPU_VCNFW_LOG_SIZE; + adev->vcn.inst[i].fw_shared.log_offset = log_offset; + } if (adev->vcn.indirect_sram) { r = amdgpu_bo_create_kernel(adev, 64 * 2 * 4, PAGE_SIZE, @@ -971,3 +994,112 @@ void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev) dev_info(adev->dev, "Will use PSP to load VCN firmware\n"); } } + +/* + * debugfs for mapping vcn firmware log buffer. + */ +#if defined(CONFIG_DEBUG_FS) +static ssize_t amdgpu_debugfs_vcn_fwlog_read(struct file *f, char __user *buf, + size_t size, loff_t *pos) +{ + struct amdgpu_vcn_inst *vcn; + void *log_buf; + volatile struct amdgpu_vcn_fwlog *plog; + unsigned int read_pos, write_pos, available, i, read_bytes = 0; + unsigned int read_num[2] = {0}; + + vcn = file_inode(f)->i_private; + if (!vcn) + return -ENODEV; + + if (!vcn->fw_shared.cpu_addr || !amdgpu_vcnfw_log) + return -EFAULT; + + log_buf = vcn->fw_shared.cpu_addr + vcn->fw_shared.mem_size; + + plog = (volatile struct amdgpu_vcn_fwlog *)log_buf; + read_pos = plog->rptr; + write_pos = plog->wptr; + + if (read_pos > AMDGPU_VCNFW_LOG_SIZE || write_pos > AMDGPU_VCNFW_LOG_SIZE) + return -EFAULT; + + if (!size || (read_pos == write_pos)) + return 0; + + if (write_pos > read_pos) { + available = write_pos - read_pos; + read_num[0] = min(size, (size_t)available); + } else { + read_num[0] = AMDGPU_VCNFW_LOG_SIZE - read_pos; + available = read_num[0] + write_pos - plog->header_size; + if (size > available) + read_num[1] = write_pos - plog->header_size; + else if (size > read_num[0]) + read_num[1] = size - read_num[0]; + else + read_num[0] = size; + } + + for (i = 0; i < 2; i++) { + if (read_num[i]) { + if (read_pos == AMDGPU_VCNFW_LOG_SIZE) + read_pos = plog->header_size; + if (read_num[i] == copy_to_user((buf + read_bytes), + (log_buf + read_pos), read_num[i])) + return -EFAULT; + + read_bytes += read_num[i]; + read_pos += read_num[i]; + } + } + + plog->rptr = read_pos; + *pos += read_bytes; + return read_bytes; +} + +static const struct file_operations amdgpu_debugfs_vcnfwlog_fops = { + .owner = THIS_MODULE, + .read = amdgpu_debugfs_vcn_fwlog_read, + .llseek = default_llseek +}; +#endif + +void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev, uint8_t i, + struct amdgpu_vcn_inst *vcn) +{ +#if defined(CONFIG_DEBUG_FS) + struct drm_minor *minor = adev_to_drm(adev)->primary; + struct dentry *root = minor->debugfs_root; + char name[32]; + + sprintf(name, "amdgpu_vcn_%d_fwlog", i); + debugfs_create_file_size(name, S_IFREG | S_IRUGO, root, vcn, + &amdgpu_debugfs_vcnfwlog_fops, + AMDGPU_VCNFW_LOG_SIZE); +#endif +} + +void amdgpu_vcn_fwlog_init(struct amdgpu_vcn_inst *vcn) +{ +#if defined(CONFIG_DEBUG_FS) + volatile uint32_t *flag = vcn->fw_shared.cpu_addr; + void *fw_log_cpu_addr = vcn->fw_shared.cpu_addr + vcn->fw_shared.mem_size; + uint64_t fw_log_gpu_addr = vcn->fw_shared.gpu_addr + vcn->fw_shared.mem_size; + volatile struct amdgpu_vcn_fwlog *log_buf = fw_log_cpu_addr; + volatile struct amdgpu_fw_shared_fw_logging *fw_log = vcn->fw_shared.cpu_addr + + vcn->fw_shared.log_offset; + *flag |= cpu_to_le32(AMDGPU_VCN_FW_LOGGING_FLAG); + fw_log->is_enabled = 1; + fw_log->addr_lo = cpu_to_le32(fw_log_gpu_addr & 0xFFFFFFFF); + fw_log->addr_hi = cpu_to_le32(fw_log_gpu_addr >> 32); + fw_log->size = cpu_to_le32(AMDGPU_VCNFW_LOG_SIZE); + + log_buf->header_size = sizeof(struct amdgpu_vcn_fwlog); + log_buf->buffer_size = AMDGPU_VCNFW_LOG_SIZE; + log_buf->rptr = log_buf->header_size; + log_buf->wptr = log_buf->header_size; + log_buf->wrapped = 0; +#endif +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h index 5d3728b027d3..e2fde88aaf5e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h @@ -158,6 +158,7 @@ #define AMDGPU_VCN_FW_SHARED_FLAG_0_RB (1 << 6) #define AMDGPU_VCN_MULTI_QUEUE_FLAG (1 << 8) #define AMDGPU_VCN_SW_RING_FLAG (1 << 9) +#define AMDGPU_VCN_FW_LOGGING_FLAG (1 << 10) #define AMDGPU_VCN_IB_FLAG_DECODE_BUFFER 0x00000001 #define AMDGPU_VCN_CMD_FLAG_MSG_BUFFER 0x00000001 @@ -205,6 +206,13 @@ struct amdgpu_vcn_reg{ unsigned scratch9; }; +struct amdgpu_vcn_fw_shared { + void *cpu_addr; + uint64_t gpu_addr; + uint32_t mem_size; + uint32_t log_offset; +}; + struct amdgpu_vcn_inst { struct amdgpu_bo *vcpu_bo; void *cpu_addr; @@ -221,8 +229,7 @@ struct amdgpu_vcn_inst { uint64_t dpg_sram_gpu_addr; uint32_t *dpg_sram_curr_addr; atomic_t dpg_enc_submission_cnt; - void *fw_shared_cpu_addr; - uint64_t fw_shared_gpu_addr; + struct amdgpu_vcn_fw_shared fw_shared; }; struct amdgpu_vcn { @@ -265,6 +272,13 @@ struct amdgpu_fw_shared_sw_ring { uint8_t padding[3]; }; +struct amdgpu_fw_shared_fw_logging { + uint8_t is_enabled; + uint32_t addr_lo; + uint32_t addr_hi; + uint32_t size; +}; + struct amdgpu_fw_shared { uint32_t present_flag_0; uint8_t pad[44]; @@ -272,6 +286,15 @@ struct amdgpu_fw_shared { uint8_t pad1[1]; struct amdgpu_fw_shared_multi_queue multi_queue; struct amdgpu_fw_shared_sw_ring sw_ring; + struct amdgpu_fw_shared_fw_logging fw_log; +}; + +struct amdgpu_vcn_fwlog { + uint32_t rptr; + uint32_t wptr; + uint32_t buffer_size; + uint32_t header_size; + uint8_t wrapped; }; struct amdgpu_vcn_decode_buffer { @@ -313,4 +336,7 @@ enum amdgpu_ring_priority_level amdgpu_vcn_get_enc_ring_prio(int ring); void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev); +void amdgpu_vcn_fwlog_init(struct amdgpu_vcn_inst *vcn); +void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev, + uint8_t i, struct amdgpu_vcn_inst *vcn); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 3a25dd220786..a025f080aa6a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -822,3 +822,148 @@ void amdgpu_virt_update_sriov_video_codec(struct amdgpu_device *adev, } } } + +static bool amdgpu_virt_get_rlcg_reg_access_flag(struct amdgpu_device *adev, + u32 acc_flags, u32 hwip, + bool write, u32 *rlcg_flag) +{ + bool ret = false; + + switch (hwip) { + case GC_HWIP: + if (amdgpu_sriov_reg_indirect_gc(adev)) { + *rlcg_flag = + write ? AMDGPU_RLCG_GC_WRITE : AMDGPU_RLCG_GC_READ; + ret = true; + /* only in new version, AMDGPU_REGS_NO_KIQ and + * AMDGPU_REGS_RLC are enabled simultaneously */ + } else if ((acc_flags & AMDGPU_REGS_RLC) && + !(acc_flags & AMDGPU_REGS_NO_KIQ) && write) { + *rlcg_flag = AMDGPU_RLCG_GC_WRITE_LEGACY; + ret = true; + } + break; + case MMHUB_HWIP: + if (amdgpu_sriov_reg_indirect_mmhub(adev) && + (acc_flags & AMDGPU_REGS_RLC) && write) { + *rlcg_flag = AMDGPU_RLCG_MMHUB_WRITE; + ret = true; + } + break; + default: + break; + } + return ret; +} + +static u32 amdgpu_virt_rlcg_reg_rw(struct amdgpu_device *adev, u32 offset, u32 v, u32 flag) +{ + struct amdgpu_rlcg_reg_access_ctrl *reg_access_ctrl; + uint32_t timeout = 50000; + uint32_t i, tmp; + uint32_t ret = 0; + static void *scratch_reg0; + static void *scratch_reg1; + static void *scratch_reg2; + static void *scratch_reg3; + static void *spare_int; + + if (!adev->gfx.rlc.rlcg_reg_access_supported) { + dev_err(adev->dev, + "indirect registers access through rlcg is not available\n"); + return 0; + } + + reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl; + scratch_reg0 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg0; + scratch_reg1 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg1; + scratch_reg2 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg2; + scratch_reg3 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg3; + if (reg_access_ctrl->spare_int) + spare_int = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->spare_int; + + if (offset == reg_access_ctrl->grbm_cntl) { + /* if the target reg offset is grbm_cntl, write to scratch_reg2 */ + writel(v, scratch_reg2); + writel(v, ((void __iomem *)adev->rmmio) + (offset * 4)); + } else if (offset == reg_access_ctrl->grbm_idx) { + /* if the target reg offset is grbm_idx, write to scratch_reg3 */ + writel(v, scratch_reg3); + writel(v, ((void __iomem *)adev->rmmio) + (offset * 4)); + } else { + /* + * SCRATCH_REG0 = read/write value + * SCRATCH_REG1[30:28] = command + * SCRATCH_REG1[19:0] = address in dword + * SCRATCH_REG1[26:24] = Error reporting + */ + writel(v, scratch_reg0); + writel((offset | flag), scratch_reg1); + if (reg_access_ctrl->spare_int) + writel(1, spare_int); + + for (i = 0; i < timeout; i++) { + tmp = readl(scratch_reg1); + if (!(tmp & AMDGPU_RLCG_SCRATCH1_ADDRESS_MASK)) + break; + udelay(10); + } + + if (i >= timeout) { + if (amdgpu_sriov_rlcg_error_report_enabled(adev)) { + if (tmp & AMDGPU_RLCG_VFGATE_DISABLED) { + dev_err(adev->dev, + "vfgate is disabled, rlcg failed to program reg: 0x%05x\n", offset); + } else if (tmp & AMDGPU_RLCG_WRONG_OPERATION_TYPE) { + dev_err(adev->dev, + "wrong operation type, rlcg failed to program reg: 0x%05x\n", offset); + } else if (tmp & AMDGPU_RLCG_REG_NOT_IN_RANGE) { + dev_err(adev->dev, + "regiser is not in range, rlcg failed to program reg: 0x%05x\n", offset); + } else { + dev_err(adev->dev, + "unknown error type, rlcg failed to program reg: 0x%05x\n", offset); + } + } else { + dev_err(adev->dev, + "timeout: rlcg faled to program reg: 0x%05x\n", offset); + } + } + } + + ret = readl(scratch_reg0); + return ret; +} + +void amdgpu_sriov_wreg(struct amdgpu_device *adev, + u32 offset, u32 value, + u32 acc_flags, u32 hwip) +{ + u32 rlcg_flag; + + if (!amdgpu_sriov_runtime(adev) && + amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags, hwip, true, &rlcg_flag)) { + amdgpu_virt_rlcg_reg_rw(adev, offset, value, rlcg_flag); + return; + } + + if (acc_flags & AMDGPU_REGS_NO_KIQ) + WREG32_NO_KIQ(offset, value); + else + WREG32(offset, value); +} + +u32 amdgpu_sriov_rreg(struct amdgpu_device *adev, + u32 offset, u32 acc_flags, u32 hwip) +{ + u32 rlcg_flag; + + if (!amdgpu_sriov_runtime(adev) && + amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags, hwip, false, &rlcg_flag)) + return amdgpu_virt_rlcg_reg_rw(adev, offset, 0, rlcg_flag); + + if (acc_flags & AMDGPU_REGS_NO_KIQ) + return RREG32_NO_KIQ(offset); + else + return RREG32(offset); +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index 9adfb8d63280..239f232f9c02 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -32,6 +32,19 @@ #define AMDGPU_PASSTHROUGH_MODE (1 << 3) /* thw whole GPU is pass through for VM */ #define AMDGPU_SRIOV_CAPS_RUNTIME (1 << 4) /* is out of full access mode */ +/* flags for indirect register access path supported by rlcg for sriov */ +#define AMDGPU_RLCG_GC_WRITE_LEGACY (0x8 << 28) +#define AMDGPU_RLCG_GC_WRITE (0x0 << 28) +#define AMDGPU_RLCG_GC_READ (0x1 << 28) +#define AMDGPU_RLCG_MMHUB_WRITE (0x2 << 28) + +/* error code for indirect register access path supported by rlcg for sriov */ +#define AMDGPU_RLCG_VFGATE_DISABLED 0x4000000 +#define AMDGPU_RLCG_WRONG_OPERATION_TYPE 0x2000000 +#define AMDGPU_RLCG_REG_NOT_IN_RANGE 0x1000000 + +#define AMDGPU_RLCG_SCRATCH1_ADDRESS_MASK 0xFFFFF + /* all asic after AI use this offset */ #define mmRCC_IOV_FUNC_IDENTIFIER 0xDE5 /* tonga/fiji use this offset */ @@ -275,13 +288,18 @@ struct amdgpu_video_codec_info; (amdgpu_sriov_vf((adev)) && \ ((adev)->virt.reg_access & (AMDGIM_FEATURE_GC_REG_RLC_EN))) +#define amdgpu_sriov_rlcg_error_report_enabled(adev) \ + (amdgpu_sriov_reg_indirect_mmhub(adev) || amdgpu_sriov_reg_indirect_gc(adev)) + #define amdgpu_passthrough(adev) \ ((adev)->virt.caps & AMDGPU_PASSTHROUGH_MODE) static inline bool is_virtual_machine(void) { -#ifdef CONFIG_X86 +#if defined(CONFIG_X86) return boot_cpu_has(X86_FEATURE_HYPERVISOR); +#elif defined(CONFIG_ARM64) + return !is_kernel_in_hyp_mode(); #else return false; #endif @@ -293,7 +311,6 @@ static inline bool is_virtual_machine(void) ((!amdgpu_in_reset(adev)) && adev->virt.tdr_debug) #define amdgpu_sriov_is_normal(adev) \ ((!amdgpu_in_reset(adev)) && (!adev->virt.tdr_debug)) - bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev); void amdgpu_virt_init_setting(struct amdgpu_device *adev); void amdgpu_virt_kiq_reg_write_reg_wait(struct amdgpu_device *adev, @@ -321,4 +338,9 @@ enum amdgpu_sriov_vf_mode amdgpu_virt_get_sriov_vf_mode(struct amdgpu_device *ad void amdgpu_virt_update_sriov_video_codec(struct amdgpu_device *adev, struct amdgpu_video_codec_info *encode, uint32_t encode_array_size, struct amdgpu_video_codec_info *decode, uint32_t decode_array_size); +void amdgpu_sriov_wreg(struct amdgpu_device *adev, + u32 offset, u32 value, + u32 acc_flags, u32 hwip); +u32 amdgpu_sriov_rreg(struct amdgpu_device *adev, + u32 offset, u32 acc_flags, u32 hwip); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c index d99c8779b51e..5224d9a39737 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c @@ -391,7 +391,6 @@ static struct drm_plane *amdgpu_vkms_plane_init(struct drm_device *dev, int index) { struct drm_plane *plane; - uint64_t modifiers[] = {DRM_FORMAT_MOD_LINEAR, DRM_FORMAT_MOD_INVALID}; int ret; plane = kzalloc(sizeof(*plane), GFP_KERNEL); @@ -402,7 +401,7 @@ static struct drm_plane *amdgpu_vkms_plane_init(struct drm_device *dev, &amdgpu_vkms_plane_funcs, amdgpu_vkms_formats, ARRAY_SIZE(amdgpu_vkms_formats), - modifiers, type, NULL); + NULL, type, NULL); if (ret) { kfree(plane); return ERR_PTR(ret); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 28f5e8b21a99..5d11978c162e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -375,6 +375,8 @@ static void amdgpu_vm_bo_base_init(struct amdgpu_vm_bo_base *base, if (bo->tbo.base.resv != vm->root.bo->tbo.base.resv) return; + dma_resv_assert_held(vm->root.bo->tbo.base.resv); + ttm_bo_set_bulk_move(&bo->tbo, &vm->lru_bulk_move); if (bo->tbo.type == ttm_bo_type_kernel && bo->parent) amdgpu_vm_bo_relocated(base); @@ -710,11 +712,17 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm, * Check if all VM PDs/PTs are ready for updates * * Returns: - * True if eviction list is empty. + * True if VM is not evicting. */ bool amdgpu_vm_ready(struct amdgpu_vm *vm) { - return list_empty(&vm->evicted); + bool ret; + + amdgpu_vm_eviction_lock(vm); + ret = !vm->evicting; + amdgpu_vm_eviction_unlock(vm); + + return ret && list_empty(&vm->evicted); } /** @@ -1580,7 +1588,7 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params, nptes = max(nptes, 1u); trace_amdgpu_vm_update_ptes(params, frag_start, upd_end, - nptes, dst, incr, upd_flags, + min(nptes, 32u), dst, incr, upd_flags, vm->task_info.pid, vm->immediate.fence_context); amdgpu_vm_update_flags(params, to_amdgpu_bo_vm(pt), @@ -2203,6 +2211,7 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev, if (!bo) return bo_va; + dma_resv_assert_held(bo->tbo.base.resv); if (amdgpu_dmabuf_is_xgmi_accessible(adev, bo)) { bo_va->is_xgmi = true; /* Power up XGMI if it can be potentially used */ @@ -2580,7 +2589,7 @@ void amdgpu_vm_bo_trace_cs(struct amdgpu_vm *vm, struct ww_acquire_ctx *ticket) } /** - * amdgpu_vm_bo_rmv - remove a bo to a specific vm + * amdgpu_vm_bo_del - remove a bo from a specific vm * * @adev: amdgpu_device pointer * @bo_va: requested bo_va @@ -2589,7 +2598,7 @@ void amdgpu_vm_bo_trace_cs(struct amdgpu_vm *vm, struct ww_acquire_ctx *ticket) * * Object have to be reserved! */ -void amdgpu_vm_bo_rmv(struct amdgpu_device *adev, +void amdgpu_vm_bo_del(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va) { struct amdgpu_bo_va_mapping *mapping, *next; @@ -2597,7 +2606,10 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev, struct amdgpu_vm *vm = bo_va->base.vm; struct amdgpu_vm_bo_base **base; + dma_resv_assert_held(vm->root.bo->tbo.base.resv); + if (bo) { + dma_resv_assert_held(bo->tbo.base.resv); if (bo->tbo.base.resv == vm->root.bo->tbo.base.resv) ttm_bo_set_bulk_move(&bo->tbo, NULL); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 4d236682a118..bd7892482bbf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -433,7 +433,7 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev, struct amdgpu_bo_va_mapping *amdgpu_vm_bo_lookup_mapping(struct amdgpu_vm *vm, uint64_t addr); void amdgpu_vm_bo_trace_cs(struct amdgpu_vm *vm, struct ww_acquire_ctx *ticket); -void amdgpu_vm_bo_rmv(struct amdgpu_device *adev, +void amdgpu_vm_bo_del(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va); void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size, uint32_t fragment_size_default, unsigned max_level, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index e50fe25fbcb8..0a7611648573 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -283,7 +283,7 @@ int amdgpu_vram_mgr_reserve_range(struct amdgpu_vram_mgr *mgr, rsv->mm_node.size = size >> PAGE_SHIFT; spin_lock(&mgr->lock); - list_add_tail(&mgr->reservations_pending, &rsv->node); + list_add_tail(&rsv->node, &mgr->reservations_pending); amdgpu_vram_mgr_do_reserve(&mgr->manager); spin_unlock(&mgr->lock); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index eb06322d2972..1b108d03e785 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -34,7 +34,6 @@ #include "amdgpu_reset.h" -#define smnPCS_XGMI23_PCS_ERROR_STATUS 0x11a01210 #define smnPCS_XGMI3X16_PCS_ERROR_STATUS 0x11a0020c #define smnPCS_GOPX1_PCS_ERROR_STATUS 0x12200210 @@ -69,17 +68,6 @@ static const int wafl_pcs_err_status_reg_arct[] = { smnPCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS + 0x100000, }; -static const int xgmi23_pcs_err_status_reg_aldebaran[] = { - smnPCS_XGMI23_PCS_ERROR_STATUS, - smnPCS_XGMI23_PCS_ERROR_STATUS + 0x100000, - smnPCS_XGMI23_PCS_ERROR_STATUS + 0x200000, - smnPCS_XGMI23_PCS_ERROR_STATUS + 0x300000, - smnPCS_XGMI23_PCS_ERROR_STATUS + 0x400000, - smnPCS_XGMI23_PCS_ERROR_STATUS + 0x500000, - smnPCS_XGMI23_PCS_ERROR_STATUS + 0x600000, - smnPCS_XGMI23_PCS_ERROR_STATUS + 0x700000 -}; - static const int xgmi3x16_pcs_err_status_reg_aldebaran[] = { smnPCS_XGMI3X16_PCS_ERROR_STATUS, smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x100000, @@ -757,53 +745,15 @@ int amdgpu_xgmi_remove_device(struct amdgpu_device *adev) return psp_xgmi_terminate(&adev->psp); } -static int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) +static int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block) { - int r; - struct ras_ih_if ih_info = { - .cb = NULL, - }; - struct ras_fs_if fs_info = { - .sysfs_name = "xgmi_wafl_err_count", - }; - if (!adev->gmc.xgmi.supported || adev->gmc.xgmi.num_physical_nodes == 0) return 0; - adev->gmc.xgmi.ras_funcs->reset_ras_error_count(adev); - - if (!adev->gmc.xgmi.ras_if) { - adev->gmc.xgmi.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); - if (!adev->gmc.xgmi.ras_if) - return -ENOMEM; - adev->gmc.xgmi.ras_if->block = AMDGPU_RAS_BLOCK__XGMI_WAFL; - adev->gmc.xgmi.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; - adev->gmc.xgmi.ras_if->sub_block_index = 0; - } - ih_info.head = fs_info.head = *adev->gmc.xgmi.ras_if; - r = amdgpu_ras_late_init(adev, adev->gmc.xgmi.ras_if, - &fs_info, &ih_info); - if (r || !amdgpu_ras_is_supported(adev, adev->gmc.xgmi.ras_if->block)) { - kfree(adev->gmc.xgmi.ras_if); - adev->gmc.xgmi.ras_if = NULL; - } - - return r; -} + adev->gmc.xgmi.ras->ras_block.hw_ops->reset_ras_error_count(adev); -static void amdgpu_xgmi_ras_fini(struct amdgpu_device *adev) -{ - if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL) && - adev->gmc.xgmi.ras_if) { - struct ras_common_if *ras_if = adev->gmc.xgmi.ras_if; - struct ras_ih_if ih_info = { - .cb = NULL, - }; - - amdgpu_ras_late_fini(adev, ras_if, &ih_info); - kfree(ras_if); - } + return amdgpu_ras_block_late_init(adev, ras_block); } uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev, @@ -835,9 +785,6 @@ static void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev) xgmi_pcs_err_status_reg_vg20[i]); break; case CHIP_ALDEBARAN: - for (i = 0; i < ARRAY_SIZE(xgmi23_pcs_err_status_reg_aldebaran); i++) - pcs_clear_status(adev, - xgmi23_pcs_err_status_reg_aldebaran[i]); for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_aldebaran); i++) pcs_clear_status(adev, xgmi3x16_pcs_err_status_reg_aldebaran[i]); @@ -890,7 +837,7 @@ static int amdgpu_xgmi_query_pcs_error_status(struct amdgpu_device *adev, return 0; } -static int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, +static void amdgpu_xgmi_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; @@ -899,7 +846,7 @@ static int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, uint32_t ue_cnt = 0, ce_cnt = 0; if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL)) - return -EINVAL; + return ; err_data->ue_count = 0; err_data->ce_count = 0; @@ -938,13 +885,6 @@ static int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, } break; case CHIP_ALDEBARAN: - /* check xgmi23 pcs error */ - for (i = 0; i < ARRAY_SIZE(xgmi23_pcs_err_status_reg_aldebaran); i++) { - data = RREG32_PCIE(xgmi23_pcs_err_status_reg_aldebaran[i]); - if (data) - amdgpu_xgmi_query_pcs_error_status(adev, - data, &ue_cnt, &ce_cnt, true); - } /* check xgmi3x16 pcs error */ for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_aldebaran); i++) { data = RREG32_PCIE(xgmi3x16_pcs_err_status_reg_aldebaran[i]); @@ -965,17 +905,53 @@ static int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, break; } - adev->gmc.xgmi.ras_funcs->reset_ras_error_count(adev); + adev->gmc.xgmi.ras->ras_block.hw_ops->reset_ras_error_count(adev); err_data->ue_count += ue_cnt; err_data->ce_count += ce_cnt; +} - return 0; +/* Trigger XGMI/WAFL error */ +static int amdgpu_ras_error_inject_xgmi(struct amdgpu_device *adev, void *inject_if) +{ + int ret = 0; + struct ta_ras_trigger_error_input *block_info = + (struct ta_ras_trigger_error_input *)inject_if; + + if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_DISALLOW)) + dev_warn(adev->dev, "Failed to disallow df cstate"); + + if (amdgpu_dpm_allow_xgmi_power_down(adev, false)) + dev_warn(adev->dev, "Failed to disallow XGMI power down"); + + ret = psp_ras_trigger_error(&adev->psp, block_info); + + if (amdgpu_ras_intr_triggered()) + return ret; + + if (amdgpu_dpm_allow_xgmi_power_down(adev, true)) + dev_warn(adev->dev, "Failed to allow XGMI power down"); + + if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_ALLOW)) + dev_warn(adev->dev, "Failed to allow df cstate"); + + return ret; } -const struct amdgpu_xgmi_ras_funcs xgmi_ras_funcs = { - .ras_late_init = amdgpu_xgmi_ras_late_init, - .ras_fini = amdgpu_xgmi_ras_fini, +struct amdgpu_ras_block_hw_ops xgmi_ras_hw_ops = { .query_ras_error_count = amdgpu_xgmi_query_ras_error_count, .reset_ras_error_count = amdgpu_xgmi_reset_ras_error_count, + .ras_error_inject = amdgpu_ras_error_inject_xgmi, +}; + +struct amdgpu_xgmi_ras xgmi_ras = { + .ras_block = { + .ras_comm = { + .name = "xgmi_wafl", + .block = AMDGPU_RAS_BLOCK__XGMI_WAFL, + .type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE, + }, + .hw_ops = &xgmi_ras_hw_ops, + .ras_late_init = amdgpu_xgmi_ras_late_init, + }, }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h index dcaad22be492..552e6fb55aa8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h @@ -24,7 +24,7 @@ #include <drm/task_barrier.h> #include "amdgpu_psp.h" - +#include "amdgpu_ras.h" struct amdgpu_hive_info { struct kobject kobj; @@ -51,7 +51,7 @@ struct amdgpu_pcs_ras_field { uint32_t pcs_err_shift; }; -extern const struct amdgpu_xgmi_ras_funcs xgmi_ras_funcs; +extern struct amdgpu_xgmi_ras xgmi_ras; struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev); void amdgpu_put_xgmi_hive(struct amdgpu_hive_info *hive); int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev); @@ -67,7 +67,8 @@ uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev, static inline bool amdgpu_xgmi_same_hive(struct amdgpu_device *adev, struct amdgpu_device *bo_adev) { - return (adev != bo_adev && + return (amdgpu_use_xgmi_p2p && + adev != bo_adev && adev->gmc.xgmi.hive_id && adev->gmc.xgmi.hive_id == bo_adev->gmc.xgmi.hive_id); } diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c index 3ea557864320..88642e7ecdf4 100644 --- a/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c @@ -68,12 +68,13 @@ int athub_v1_0_set_clockgating(struct amdgpu_device *adev, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->asic_type) { - case CHIP_VEGA10: - case CHIP_VEGA12: - case CHIP_VEGA20: - case CHIP_RAVEN: - case CHIP_RENOIR: + switch (adev->ip_versions[ATHUB_HWIP][0]) { + case IP_VERSION(9, 0, 0): + case IP_VERSION(9, 1, 0): + case IP_VERSION(9, 2, 0): + case IP_VERSION(9, 3, 0): + case IP_VERSION(9, 4, 0): + case IP_VERSION(1, 5, 0): athub_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE); athub_update_medium_grain_light_sleep(adev, diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c index ab6a07e5e8c4..a720436857b4 100644 --- a/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c @@ -78,6 +78,7 @@ int athub_v2_0_set_clockgating(struct amdgpu_device *adev, return 0; switch (adev->ip_versions[ATHUB_HWIP][0]) { + case IP_VERSION(1, 3, 1): case IP_VERSION(2, 0, 0): case IP_VERSION(2, 0, 2): athub_v2_0_update_medium_grain_clock_gating(adev, diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c index 2edefd10e56c..ad8e87d3d2cb 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, case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): + case IP_VERSION(2, 4, 0): athub_v2_1_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE); athub_v2_1_update_medium_grain_light_sleep(adev, state == AMD_CG_STATE_GATE); break; diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index f10ce740a29c..de6d10390ab2 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik.c +++ b/drivers/gpu/drm/amd/amdgpu/cik.c @@ -1719,7 +1719,7 @@ static void cik_program_aspm(struct amdgpu_device *adev) bool disable_l0s = false, disable_l1 = false, disable_plloff_in_l1 = false; bool disable_clkreq = false; - if (amdgpu_aspm == 0) + if (!amdgpu_device_should_use_aspm(adev)) return; if (pci_is_root_bus(adev->pdev->bus)) diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c index fb61c0814115..288fce7dc0ed 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c @@ -2532,7 +2532,7 @@ static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode) break; } /* adjust pm to dpms */ - amdgpu_pm_compute_clocks(adev); + amdgpu_dpm_compute_clocks(adev); } static void dce_v10_0_crtc_prepare(struct drm_crtc *crtc) diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c index 17942a11366d..cbe5250b31cb 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c @@ -2608,7 +2608,7 @@ static void dce_v11_0_crtc_dpms(struct drm_crtc *crtc, int mode) break; } /* adjust pm to dpms */ - amdgpu_pm_compute_clocks(adev); + amdgpu_dpm_compute_clocks(adev); } static void dce_v11_0_crtc_prepare(struct drm_crtc *crtc) diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c index 2ec99ec8e1a3..982855e6cf52 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c @@ -2424,7 +2424,7 @@ static void dce_v6_0_crtc_dpms(struct drm_crtc *crtc, int mode) break; } /* adjust pm to dpms */ - amdgpu_pm_compute_clocks(adev); + amdgpu_dpm_compute_clocks(adev); } static void dce_v6_0_crtc_prepare(struct drm_crtc *crtc) diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c index de11fbe5aba2..84440741c60b 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c @@ -2433,7 +2433,7 @@ static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode) break; } /* adjust pm to dpms */ - amdgpu_pm_compute_clocks(adev); + amdgpu_dpm_compute_clocks(adev); } static void dce_v8_0_crtc_prepare(struct drm_crtc *crtc) diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c index 43c5e3ec9a39..f4dfca013ec5 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c @@ -458,7 +458,7 @@ 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, - int counter_idx, uint64_t config, + uint64_t config, int counter_idx, bool is_deferred) { @@ -476,8 +476,8 @@ static int df_v3_6_pmc_set_deferred(struct amdgpu_device *adev, } static bool df_v3_6_pmc_is_deferred(struct amdgpu_device *adev, - int counter_idx, - uint64_t config) + uint64_t config, + int counter_idx) { return (df_v3_6_pmc_has_counter(adev, config, counter_idx) && (adev->df_perfmon_config_assign_mask[counter_idx] diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index dbe7442fb25c..f4c6accd3226 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -56,10 +56,6 @@ #define GFX10_NUM_GFX_RINGS_Sienna_Cichlid 1 #define GFX10_MEC_HPD_SIZE 2048 -#define RLCG_VFGATE_DISABLED 0x4000000 -#define RLCG_WRONG_OPERATION_TYPE 0x2000000 -#define RLCG_NOT_IN_RANGE 0x1000000 - #define F32_CE_PROGRAM_RAM_SIZE 65536 #define RLCG_UCODE_LOADING_START_ADDRESS 0x00002000L @@ -110,6 +106,12 @@ #define mmGOLDEN_TSC_COUNT_UPPER_Vangogh_BASE_IDX 1 #define mmGOLDEN_TSC_COUNT_LOWER_Vangogh 0x0026 #define mmGOLDEN_TSC_COUNT_LOWER_Vangogh_BASE_IDX 1 + +#define mmGOLDEN_TSC_COUNT_UPPER_GC_10_3_6 0x002d +#define mmGOLDEN_TSC_COUNT_UPPER_GC_10_3_6_BASE_IDX 1 +#define mmGOLDEN_TSC_COUNT_LOWER_GC_10_3_6 0x002e +#define mmGOLDEN_TSC_COUNT_LOWER_GC_10_3_6_BASE_IDX 1 + #define mmSPI_CONFIG_CNTL_1_Vangogh 0x2441 #define mmSPI_CONFIG_CNTL_1_Vangogh_BASE_IDX 1 #define mmVGT_TF_MEMORY_BASE_HI_Vangogh 0x2261 @@ -180,14 +182,6 @@ #define mmRLC_SPARE_INT_0_Sienna_Cichlid 0x4ca5 #define mmRLC_SPARE_INT_0_Sienna_Cichlid_BASE_IDX 1 -#define GFX_RLCG_GC_WRITE_OLD (0x8 << 28) -#define GFX_RLCG_GC_WRITE (0x0 << 28) -#define GFX_RLCG_GC_READ (0x1 << 28) -#define GFX_RLCG_MMHUB_WRITE (0x2 << 28) - -#define RLCG_ERROR_REPORT_ENABLED(adev) \ - (amdgpu_sriov_reg_indirect_mmhub(adev) || amdgpu_sriov_reg_indirect_gc(adev)) - MODULE_FIRMWARE("amdgpu/navi10_ce.bin"); MODULE_FIRMWARE("amdgpu/navi10_pfp.bin"); MODULE_FIRMWARE("amdgpu/navi10_me.bin"); @@ -256,13 +250,6 @@ MODULE_FIRMWARE("amdgpu/yellow_carp_mec.bin"); MODULE_FIRMWARE("amdgpu/yellow_carp_mec2.bin"); MODULE_FIRMWARE("amdgpu/yellow_carp_rlc.bin"); -MODULE_FIRMWARE("amdgpu/cyan_skillfish_ce.bin"); -MODULE_FIRMWARE("amdgpu/cyan_skillfish_pfp.bin"); -MODULE_FIRMWARE("amdgpu/cyan_skillfish_me.bin"); -MODULE_FIRMWARE("amdgpu/cyan_skillfish_mec.bin"); -MODULE_FIRMWARE("amdgpu/cyan_skillfish_mec2.bin"); -MODULE_FIRMWARE("amdgpu/cyan_skillfish_rlc.bin"); - MODULE_FIRMWARE("amdgpu/cyan_skillfish2_ce.bin"); MODULE_FIRMWARE("amdgpu/cyan_skillfish2_pfp.bin"); MODULE_FIRMWARE("amdgpu/cyan_skillfish2_me.bin"); @@ -270,6 +257,20 @@ MODULE_FIRMWARE("amdgpu/cyan_skillfish2_mec.bin"); MODULE_FIRMWARE("amdgpu/cyan_skillfish2_mec2.bin"); MODULE_FIRMWARE("amdgpu/cyan_skillfish2_rlc.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_6_ce.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_6_pfp.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_6_me.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_6_mec.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_6_mec2.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_6_rlc.bin"); + +MODULE_FIRMWARE("amdgpu/gc_10_3_7_ce.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_7_pfp.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_7_me.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_7_mec.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_7_mec2.bin"); +MODULE_FIRMWARE("amdgpu/gc_10_3_7_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), @@ -1463,143 +1464,6 @@ static const struct soc15_reg_golden golden_settings_gc_10_1_2[] = SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00c00000) }; -static bool gfx_v10_get_rlcg_flag(struct amdgpu_device *adev, u32 acc_flags, u32 hwip, - int write, u32 *rlcg_flag) -{ - switch (hwip) { - case GC_HWIP: - if (amdgpu_sriov_reg_indirect_gc(adev)) { - *rlcg_flag = write ? GFX_RLCG_GC_WRITE : GFX_RLCG_GC_READ; - - return true; - /* only in new version, AMDGPU_REGS_NO_KIQ and AMDGPU_REGS_RLC enabled simultaneously */ - } else if ((acc_flags & AMDGPU_REGS_RLC) && !(acc_flags & AMDGPU_REGS_NO_KIQ)) { - *rlcg_flag = GFX_RLCG_GC_WRITE_OLD; - - return true; - } - - break; - case MMHUB_HWIP: - if (amdgpu_sriov_reg_indirect_mmhub(adev) && - (acc_flags & AMDGPU_REGS_RLC) && write) { - *rlcg_flag = GFX_RLCG_MMHUB_WRITE; - return true; - } - - break; - default: - DRM_DEBUG("Not program register by RLCG\n"); - } - - return false; -} - -static u32 gfx_v10_rlcg_rw(struct amdgpu_device *adev, u32 offset, u32 v, uint32_t flag) -{ - static void *scratch_reg0; - static void *scratch_reg1; - static void *scratch_reg2; - static void *scratch_reg3; - static void *spare_int; - static uint32_t grbm_cntl; - static uint32_t grbm_idx; - uint32_t i = 0; - uint32_t retries = 50000; - u32 ret = 0; - u32 tmp; - - scratch_reg0 = adev->rmmio + - (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG0) * 4; - scratch_reg1 = adev->rmmio + - (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG1) * 4; - scratch_reg2 = adev->rmmio + - (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG2) * 4; - scratch_reg3 = adev->rmmio + - (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG3) * 4; - - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) { - spare_int = adev->rmmio + - (adev->reg_offset[GC_HWIP][0][mmRLC_SPARE_INT_0_Sienna_Cichlid_BASE_IDX] - + mmRLC_SPARE_INT_0_Sienna_Cichlid) * 4; - } else { - spare_int = adev->rmmio + - (adev->reg_offset[GC_HWIP][0][mmRLC_SPARE_INT_BASE_IDX] + mmRLC_SPARE_INT) * 4; - } - - grbm_cntl = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_CNTL_BASE_IDX] + mmGRBM_GFX_CNTL; - grbm_idx = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_INDEX_BASE_IDX] + mmGRBM_GFX_INDEX; - - if (offset == grbm_cntl || offset == grbm_idx) { - if (offset == grbm_cntl) - writel(v, scratch_reg2); - else if (offset == grbm_idx) - writel(v, scratch_reg3); - - writel(v, ((void __iomem *)adev->rmmio) + (offset * 4)); - } else { - writel(v, scratch_reg0); - writel(offset | flag, scratch_reg1); - writel(1, spare_int); - - for (i = 0; i < retries; i++) { - tmp = readl(scratch_reg1); - if (!(tmp & flag)) - break; - - udelay(10); - } - - if (i >= retries) { - if (RLCG_ERROR_REPORT_ENABLED(adev)) { - if (tmp & RLCG_VFGATE_DISABLED) - pr_err("The vfgate is disabled, program reg:0x%05x failed!\n", offset); - else if (tmp & RLCG_WRONG_OPERATION_TYPE) - pr_err("Wrong operation type, program reg:0x%05x failed!\n", offset); - else if (tmp & RLCG_NOT_IN_RANGE) - pr_err("The register is not in range, program reg:0x%05x failed!\n", offset); - else - pr_err("Unknown error type, program reg:0x%05x failed!\n", offset); - } else - pr_err("timeout: rlcg program reg:0x%05x failed!\n", offset); - } - } - - ret = readl(scratch_reg0); - - return ret; -} - -static void gfx_v10_sriov_wreg(struct amdgpu_device *adev, u32 offset, u32 value, u32 acc_flags, u32 hwip) -{ - u32 rlcg_flag; - - if (!amdgpu_sriov_runtime(adev) && - gfx_v10_get_rlcg_flag(adev, acc_flags, hwip, 1, &rlcg_flag)) { - gfx_v10_rlcg_rw(adev, offset, value, rlcg_flag); - return; - } - - if (acc_flags & AMDGPU_REGS_NO_KIQ) - WREG32_NO_KIQ(offset, value); - else - WREG32(offset, value); -} - -static u32 gfx_v10_sriov_rreg(struct amdgpu_device *adev, u32 offset, u32 acc_flags, u32 hwip) -{ - u32 rlcg_flag; - - if (!amdgpu_sriov_runtime(adev) && - gfx_v10_get_rlcg_flag(adev, acc_flags, hwip, 0, &rlcg_flag)) - return gfx_v10_rlcg_rw(adev, offset, 0, rlcg_flag); - - if (acc_flags & AMDGPU_REGS_NO_KIQ) - return RREG32_NO_KIQ(offset); - else - return RREG32(offset); -} - static const struct soc15_reg_golden golden_settings_gc_10_1_nv14[] = { /* Pending on emulation bring up */ @@ -3557,6 +3421,57 @@ static const struct soc15_reg_golden golden_settings_gc_10_0_cyan_skillfish[] = SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0x00800000, 0x00800000) }; +static const struct soc15_reg_golden golden_settings_gc_10_3_6[] = +{ + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0xff7f0fff, 0x78000100), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0x000000ff, 0x00000044), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c200), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000280), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1807ff, 0x00000042), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x1ff1ffff, 0x00000500), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0x000000ff, 0x00000044), + 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, mmLDS_CONFIG, 0x000001ff, 0x00000020), + 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, 0x00000820), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQG_CONFIG, 0x000017ff, 0x00001000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSX_DEBUG_1, 0xffffff7f, 0x00010020), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00100000) +}; + +static const struct soc15_reg_golden golden_settings_gc_10_3_7[] = { + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0xff7f0fff, 0x78000100), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0x000000ff, 0x000000e4), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c200), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000280), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1807ff, 0x00000041), + 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, 0xffffffff), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffff), + 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, mmLDS_CONFIG, 0x000001ff, 0x00000020), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf000003f, 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, 0x00000820), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQG_CONFIG, 0x000017ff, 0x00001000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSX_DEBUG_1, 0xffffff7f, 0x00010020), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00100000) +}; + #define DEFAULT_SH_MEM_CONFIG \ ((SH_MEM_ADDRESS_MODE_64 << SH_MEM_CONFIG__ADDRESS_MODE__SHIFT) | \ (SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | \ @@ -3790,10 +3705,21 @@ static void gfx_v10_0_init_golden_registers(struct amdgpu_device *adev) (const u32)ARRAY_SIZE(golden_settings_gc_10_3_5)); break; case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): soc15_program_register_sequence(adev, golden_settings_gc_10_0_cyan_skillfish, (const u32)ARRAY_SIZE(golden_settings_gc_10_0_cyan_skillfish)); break; + case IP_VERSION(10, 3, 6): + soc15_program_register_sequence(adev, + golden_settings_gc_10_3_6, + (const u32)ARRAY_SIZE(golden_settings_gc_10_3_6)); + break; + case IP_VERSION(10, 3, 7): + soc15_program_register_sequence(adev, + golden_settings_gc_10_3_7, + (const u32)ARRAY_SIZE(golden_settings_gc_10_3_7)); + break; default: break; } @@ -3968,6 +3894,7 @@ static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev) case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): if ((adev->gfx.me_fw_version >= 0x00000046) && (adev->gfx.me_feature_version >= 27) && (adev->gfx.pfp_fw_version >= 0x00000068) && @@ -3981,7 +3908,9 @@ static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): adev->gfx.cp_fw_write_wait = true; break; default: @@ -4102,11 +4031,15 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev) case IP_VERSION(10, 3, 3): chip_name = "yellow_carp"; break; + case IP_VERSION(10, 3, 6): + chip_name = "gc_10_3_6"; + break; case IP_VERSION(10, 1, 3): - if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2) - chip_name = "cyan_skillfish2"; - else - chip_name = "cyan_skillfish"; + case IP_VERSION(10, 1, 4): + chip_name = "cyan_skillfish2"; + break; + case IP_VERSION(10, 3, 7): + chip_name = "gc_10_3_7"; break; default: BUG(); @@ -4448,6 +4381,30 @@ static void gfx_v10_0_rlc_fini(struct amdgpu_device *adev) (void **)&adev->gfx.rlc.cp_table_ptr); } +static void gfx_v10_0_init_rlcg_reg_access_ctrl(struct amdgpu_device *adev) +{ + struct amdgpu_rlcg_reg_access_ctrl *reg_access_ctrl; + + reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl; + reg_access_ctrl->scratch_reg0 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG0); + reg_access_ctrl->scratch_reg1 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG1); + reg_access_ctrl->scratch_reg2 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG2); + reg_access_ctrl->scratch_reg3 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG3); + reg_access_ctrl->grbm_cntl = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_CNTL); + reg_access_ctrl->grbm_idx = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX); + switch (adev->ip_versions[GC_HWIP][0]) { + case IP_VERSION(10, 3, 0): + reg_access_ctrl->spare_int = + SOC15_REG_OFFSET(GC, 0, mmRLC_SPARE_INT_0_Sienna_Cichlid); + break; + default: + reg_access_ctrl->spare_int = + SOC15_REG_OFFSET(GC, 0, mmRLC_SPARE_INT); + break; + } + adev->gfx.rlc.rlcg_reg_access_supported = true; +} + static int gfx_v10_0_rlc_init(struct amdgpu_device *adev) { const struct cs_section_def *cs_data; @@ -4468,6 +4425,7 @@ static int gfx_v10_0_rlc_init(struct amdgpu_device *adev) if (adev->gfx.rlc.funcs->update_spm_vmid) adev->gfx.rlc.funcs->update_spm_vmid(adev, 0xf); + return 0; } @@ -4678,7 +4636,9 @@ static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): 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; @@ -4689,6 +4649,7 @@ static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev) 1 << REG_GET_FIELD(gb_addr_config, GB_ADDR_CONFIG, NUM_PKRS); break; case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): 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; @@ -4801,6 +4762,7 @@ static int gfx_v10_0_sw_init(void *handle) case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): adev->gfx.me.num_me = 1; adev->gfx.me.num_pipe_per_me = 1; adev->gfx.me.num_queue_per_pipe = 1; @@ -4813,7 +4775,9 @@ static int gfx_v10_0_sw_init(void *handle) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): adev->gfx.me.num_me = 1; adev->gfx.me.num_pipe_per_me = 1; adev->gfx.me.num_queue_per_pipe = 1; @@ -4865,10 +4829,14 @@ static int gfx_v10_0_sw_init(void *handle) if (r) return r; - r = gfx_v10_0_rlc_init(adev); - if (r) { - DRM_ERROR("Failed to init rlc BOs!\n"); - return r; + if (adev->gfx.rlc.funcs) { + if (adev->gfx.rlc.funcs->init) { + r = adev->gfx.rlc.funcs->init(adev); + if (r) { + dev_err(adev->dev, "Failed to init rlc BOs!\n"); + return r; + } + } } r = gfx_v10_0_mec_init(adev); @@ -5047,7 +5015,8 @@ static void gfx_v10_0_setup_rb(struct amdgpu_device *adev) for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { bitmap = i * adev->gfx.config.max_sh_per_se + j; if (((adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0)) || - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 3))) && + (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 3)) || + (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 6))) && ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1)) continue; gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff); @@ -6321,7 +6290,9 @@ static void gfx_v10_0_cp_gfx_set_doorbell(struct amdgpu_device *adev, case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): 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); @@ -6458,7 +6429,9 @@ static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): WREG32_SOC15(GC, 0, mmCP_MEC_CNTL_Sienna_Cichlid, 0); break; default: @@ -6472,7 +6445,9 @@ static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): WREG32_SOC15(GC, 0, mmCP_MEC_CNTL_Sienna_Cichlid, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK)); @@ -6570,7 +6545,9 @@ static void gfx_v10_0_kiq_setting(struct amdgpu_ring *ring) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): tmp = RREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS_Sienna_Cichlid); tmp &= 0xffffff00; tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue); @@ -7300,6 +7277,8 @@ static bool gfx_v10_0_check_grbm_cam_remapping(struct amdgpu_device *adev) break; case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 6): + case IP_VERSION(10, 3, 7): return true; default: data = RREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE); @@ -7334,7 +7313,9 @@ static void gfx_v10_0_setup_grbm_cam_remapping(struct amdgpu_device *adev) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): /* 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) | @@ -7654,6 +7635,7 @@ static int gfx_v10_0_soft_reset(void *handle) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): if (REG_GET_FIELD(tmp, GRBM_STATUS2, RLC_BUSY_Sienna_Cichlid)) grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, @@ -7721,6 +7703,21 @@ static uint64_t gfx_v10_0_get_gpu_clock_counter(struct amdgpu_device *adev) preempt_enable(); clock = clock_lo | (clock_hi << 32ULL); break; + case IP_VERSION(10, 3, 6): + preempt_disable(); + clock_hi = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_GC_10_3_6); + clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_GC_10_3_6); + hi_check = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_GC_10_3_6); + /* The SMUIO TSC clock frequency is 100MHz, which sets 32-bit carry over + * roughly every 42 seconds. + */ + if (hi_check != clock_hi) { + clock_lo = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_GC_10_3_6); + clock_hi = hi_check; + } + preempt_enable(); + clock = clock_lo | (clock_hi << 32ULL); + break; default: preempt_disable(); clock_hi = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER); @@ -7778,6 +7775,7 @@ static int gfx_v10_0_early_init(void *handle) case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): adev->gfx.num_gfx_rings = GFX10_NUM_GFX_RINGS_NV1X; break; case IP_VERSION(10, 3, 0): @@ -7785,7 +7783,9 @@ static int gfx_v10_0_early_init(void *handle) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): adev->gfx.num_gfx_rings = GFX10_NUM_GFX_RINGS_Sienna_Cichlid; break; default: @@ -7801,6 +7801,9 @@ static int gfx_v10_0_early_init(void *handle) gfx_v10_0_set_gds_init(adev); gfx_v10_0_set_rlc_funcs(adev); + /* init rlcg reg access ctrl */ + gfx_v10_0_init_rlcg_reg_access_ctrl(adev); + return 0; } @@ -7843,7 +7846,9 @@ static void gfx_v10_0_set_safe_mode(struct amdgpu_device *adev) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE_Sienna_Cichlid, data); /* wait for RLC_SAFE_MODE */ @@ -7879,7 +7884,9 @@ static void gfx_v10_0_unset_safe_mode(struct amdgpu_device *adev) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE_Sienna_Cichlid, data); break; default: @@ -8333,6 +8340,8 @@ static void gfx_v10_cntl_power_gating(struct amdgpu_device *adev, bool enable) switch (adev->ip_versions[GC_HWIP][0]) { case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 6): + case IP_VERSION(10, 3, 7): data = 0x4E20 & RLC_PG_DELAY_3__CGCG_ACTIVE_BEFORE_CGPG_MASK_Vangogh; WREG32_SOC15(GC, 0, mmRLC_PG_DELAY_3, data); break; @@ -8377,8 +8386,6 @@ static const struct amdgpu_rlc_funcs gfx_v10_0_rlc_funcs_sriov = { .reset = gfx_v10_0_rlc_reset, .start = gfx_v10_0_rlc_start, .update_spm_vmid = gfx_v10_0_update_spm_vmid, - .sriov_wreg = gfx_v10_sriov_wreg, - .sriov_rreg = gfx_v10_sriov_rreg, .is_rlcg_access_range = gfx_v10_0_is_rlcg_access_range, }; @@ -8403,6 +8410,8 @@ static int gfx_v10_0_set_powergating_state(void *handle, break; case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 6): + case IP_VERSION(10, 3, 7): gfx_v10_cntl_pg(adev, enable); amdgpu_gfx_off_ctrl(adev, enable); break; @@ -8429,7 +8438,9 @@ static int gfx_v10_0_set_clockgating_state(void *handle, case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): gfx_v10_0_update_gfx_clock_gating(adev, state == AMD_CG_STATE_GATE); break; @@ -9366,6 +9377,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_gfx = { .align_mask = 0xff, .nop = PACKET3(PACKET3_NOP, 0x3FFF), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_GFXHUB_0, .get_rptr = gfx_v10_0_ring_get_rptr_gfx, .get_wptr = gfx_v10_0_ring_get_wptr_gfx, @@ -9537,11 +9549,14 @@ static void gfx_v10_0_set_rlc_funcs(struct amdgpu_device *adev) case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): adev->gfx.rlc.funcs = &gfx_v10_0_rlc_funcs; break; case IP_VERSION(10, 1, 2): @@ -9634,7 +9649,9 @@ static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev, for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { bitmap = i * adev->gfx.config.max_sh_per_se + j; if (((adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0)) || - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 3))) && + (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 3)) || + (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 6)) || + (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 7))) && ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1)) continue; mask = 1; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 9189fb85a4dd..46d4bf27ebbb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -63,13 +63,6 @@ #define mmGCEA_PROBE_MAP 0x070c #define mmGCEA_PROBE_MAP_BASE_IDX 0 -#define GFX9_RLCG_GC_WRITE_OLD (0x8 << 28) -#define GFX9_RLCG_GC_WRITE (0x0 << 28) -#define GFX9_RLCG_GC_READ (0x1 << 28) -#define GFX9_RLCG_VFGATE_DISABLED 0x4000000 -#define GFX9_RLCG_WRONG_OPERATION_TYPE 0x2000000 -#define GFX9_RLCG_NOT_IN_RANGE 0x1000000 - MODULE_FIRMWARE("amdgpu/vega10_ce.bin"); MODULE_FIRMWARE("amdgpu/vega10_pfp.bin"); MODULE_FIRMWARE("amdgpu/vega10_me.bin"); @@ -746,128 +739,6 @@ static const u32 GFX_RLC_SRM_INDEX_CNTL_DATA_OFFSETS[] = mmRLC_SRM_INDEX_CNTL_DATA_7 - mmRLC_SRM_INDEX_CNTL_DATA_0, }; -static u32 gfx_v9_0_rlcg_rw(struct amdgpu_device *adev, u32 offset, u32 v, uint32_t flag) -{ - static void *scratch_reg0; - static void *scratch_reg1; - static void *scratch_reg2; - static void *scratch_reg3; - static void *spare_int; - static uint32_t grbm_cntl; - static uint32_t grbm_idx; - uint32_t i = 0; - uint32_t retries = 50000; - u32 ret = 0; - u32 tmp; - - scratch_reg0 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG0)*4; - scratch_reg1 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG1)*4; - scratch_reg2 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG2_BASE_IDX] + mmSCRATCH_REG2)*4; - scratch_reg3 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG3_BASE_IDX] + mmSCRATCH_REG3)*4; - spare_int = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmRLC_SPARE_INT_BASE_IDX] + mmRLC_SPARE_INT)*4; - - grbm_cntl = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_CNTL_BASE_IDX] + mmGRBM_GFX_CNTL; - grbm_idx = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_INDEX_BASE_IDX] + mmGRBM_GFX_INDEX; - - if (offset == grbm_cntl || offset == grbm_idx) { - if (offset == grbm_cntl) - writel(v, scratch_reg2); - else if (offset == grbm_idx) - writel(v, scratch_reg3); - - writel(v, ((void __iomem *)adev->rmmio) + (offset * 4)); - } else { - /* - * SCRATCH_REG0 = read/write value - * SCRATCH_REG1[30:28] = command - * SCRATCH_REG1[19:0] = address in dword - * SCRATCH_REG1[26:24] = Error reporting - */ - writel(v, scratch_reg0); - writel(offset | flag, scratch_reg1); - writel(1, spare_int); - - for (i = 0; i < retries; i++) { - tmp = readl(scratch_reg1); - if (!(tmp & flag)) - break; - - udelay(10); - } - - if (i >= retries) { - if (amdgpu_sriov_reg_indirect_gc(adev)) { - if (tmp & GFX9_RLCG_VFGATE_DISABLED) - pr_err("The vfgate is disabled, program reg:0x%05x failed!\n", offset); - else if (tmp & GFX9_RLCG_WRONG_OPERATION_TYPE) - pr_err("Wrong operation type, program reg:0x%05x failed!\n", offset); - else if (tmp & GFX9_RLCG_NOT_IN_RANGE) - pr_err("The register is not in range, program reg:0x%05x failed!\n", offset); - else - pr_err("Unknown error type, program reg:0x%05x failed!\n", offset); - } else - pr_err("timeout: rlcg program reg:0x%05x failed!\n", offset); - } - } - - ret = readl(scratch_reg0); - - return ret; -} - -static bool gfx_v9_0_get_rlcg_flag(struct amdgpu_device *adev, u32 acc_flags, u32 hwip, - int write, u32 *rlcg_flag) -{ - - switch (hwip) { - case GC_HWIP: - if (amdgpu_sriov_reg_indirect_gc(adev)) { - *rlcg_flag = write ? GFX9_RLCG_GC_WRITE : GFX9_RLCG_GC_READ; - - return true; - /* only in new version, AMDGPU_REGS_NO_KIQ and AMDGPU_REGS_RLC enabled simultaneously */ - } else if ((acc_flags & AMDGPU_REGS_RLC) && !(acc_flags & AMDGPU_REGS_NO_KIQ) && write) { - *rlcg_flag = GFX9_RLCG_GC_WRITE_OLD; - return true; - } - - break; - default: - return false; - } - - return false; -} - -static u32 gfx_v9_0_sriov_rreg(struct amdgpu_device *adev, u32 offset, u32 acc_flags, u32 hwip) -{ - u32 rlcg_flag; - - if (!amdgpu_sriov_runtime(adev) && gfx_v9_0_get_rlcg_flag(adev, acc_flags, hwip, 0, &rlcg_flag)) - return gfx_v9_0_rlcg_rw(adev, offset, 0, rlcg_flag); - - if (acc_flags & AMDGPU_REGS_NO_KIQ) - return RREG32_NO_KIQ(offset); - else - return RREG32(offset); -} - -static void gfx_v9_0_sriov_wreg(struct amdgpu_device *adev, u32 offset, - u32 value, u32 acc_flags, u32 hwip) -{ - u32 rlcg_flag; - - if (!amdgpu_sriov_runtime(adev) && gfx_v9_0_get_rlcg_flag(adev, acc_flags, hwip, 1, &rlcg_flag)) { - gfx_v9_0_rlcg_rw(adev, offset, value, rlcg_flag); - return; - } - - if (acc_flags & AMDGPU_REGS_NO_KIQ) - WREG32_NO_KIQ(offset, value); - else - WREG32(offset, value); -} - #define VEGA10_GB_ADDR_CONFIG_GOLDEN 0x2a114042 #define VEGA12_GB_ADDR_CONFIG_GOLDEN 0x24104041 #define RAVEN_GB_ADDR_CONFIG_GOLDEN 0x24000042 @@ -882,7 +753,7 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev, 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); -static int gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev, +static void gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev, void *ras_error_status); static int gfx_v9_0_ras_error_inject(struct amdgpu_device *adev, void *inject_if); @@ -2008,6 +1879,21 @@ static int gfx_v9_0_cp_jump_table_num(struct amdgpu_device *adev) return 4; } +static void gfx_v9_0_init_rlcg_reg_access_ctrl(struct amdgpu_device *adev) +{ + struct amdgpu_rlcg_reg_access_ctrl *reg_access_ctrl; + + reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl; + reg_access_ctrl->scratch_reg0 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG0); + reg_access_ctrl->scratch_reg1 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG1); + reg_access_ctrl->scratch_reg2 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG2); + reg_access_ctrl->scratch_reg3 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG3); + reg_access_ctrl->grbm_cntl = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_CNTL); + reg_access_ctrl->grbm_idx = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX); + reg_access_ctrl->spare_int = SOC15_REG_OFFSET(GC, 0, mmRLC_SPARE_INT); + adev->gfx.rlc.rlcg_reg_access_supported = true; +} + static int gfx_v9_0_rlc_init(struct amdgpu_device *adev) { const struct cs_section_def *cs_data; @@ -2197,12 +2083,16 @@ static const struct amdgpu_gfx_funcs gfx_v9_0_gfx_funcs = { .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q, }; -static const struct amdgpu_gfx_ras_funcs gfx_v9_0_ras_funcs = { - .ras_late_init = amdgpu_gfx_ras_late_init, - .ras_fini = amdgpu_gfx_ras_fini, - .ras_error_inject = &gfx_v9_0_ras_error_inject, - .query_ras_error_count = &gfx_v9_0_query_ras_error_count, - .reset_ras_error_count = &gfx_v9_0_reset_ras_error_count, +const struct amdgpu_ras_block_hw_ops gfx_v9_0_ras_ops = { + .ras_error_inject = &gfx_v9_0_ras_error_inject, + .query_ras_error_count = &gfx_v9_0_query_ras_error_count, + .reset_ras_error_count = &gfx_v9_0_reset_ras_error_count, +}; + +static struct amdgpu_gfx_ras gfx_v9_0_ras = { + .ras_block = { + .hw_ops = &gfx_v9_0_ras_ops, + }, }; static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) @@ -2231,7 +2121,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) DRM_INFO("fix gfx.config for vega12\n"); break; case IP_VERSION(9, 4, 0): - adev->gfx.ras_funcs = &gfx_v9_0_ras_funcs; + adev->gfx.ras = &gfx_v9_0_ras; 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; @@ -2258,7 +2148,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) gb_addr_config = RAVEN_GB_ADDR_CONFIG_GOLDEN; break; case IP_VERSION(9, 4, 1): - adev->gfx.ras_funcs = &gfx_v9_4_ras_funcs; + adev->gfx.ras = &gfx_v9_4_ras; 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; @@ -2279,7 +2169,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) gb_addr_config |= 0x22010042; break; case IP_VERSION(9, 4, 2): - adev->gfx.ras_funcs = &gfx_v9_4_2_ras_funcs; + adev->gfx.ras = &gfx_v9_4_2_ras; 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; @@ -2298,6 +2188,27 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) break; } + if (adev->gfx.ras) { + err = amdgpu_ras_register_ras_block(adev, &adev->gfx.ras->ras_block); + if (err) { + DRM_ERROR("Failed to register gfx ras block!\n"); + return err; + } + + strcpy(adev->gfx.ras->ras_block.ras_comm.name, "gfx"); + adev->gfx.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__GFX; + adev->gfx.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; + adev->gfx.ras_if = &adev->gfx.ras->ras_block.ras_comm; + + /* If not define special ras_late_init function, use gfx default ras_late_init */ + if (!adev->gfx.ras->ras_block.ras_late_init) + adev->gfx.ras->ras_block.ras_late_init = amdgpu_gfx_ras_late_init; + + /* If not defined special ras_cb function, use default ras_cb */ + if (!adev->gfx.ras->ras_block.ras_cb) + adev->gfx.ras->ras_block.ras_cb = amdgpu_gfx_process_ras_data_cb; + } + adev->gfx.config.gb_addr_config = gb_addr_config; adev->gfx.config.gb_addr_config_fields.num_pipes = 1 << @@ -2434,10 +2345,14 @@ static int gfx_v9_0_sw_init(void *handle) return r; } - r = adev->gfx.rlc.funcs->init(adev); - if (r) { - DRM_ERROR("Failed to init rlc BOs!\n"); - return r; + if (adev->gfx.rlc.funcs) { + if (adev->gfx.rlc.funcs->init) { + r = adev->gfx.rlc.funcs->init(adev); + if (r) { + dev_err(adev->dev, "Failed to init rlc BOs!\n"); + return r; + } + } } r = gfx_v9_0_mec_init(adev); @@ -2513,10 +2428,6 @@ static int gfx_v9_0_sw_fini(void *handle) int i; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->ras_fini) - adev->gfx.ras_funcs->ras_fini(adev); - for (i = 0; i < adev->gfx.num_gfx_rings; i++) amdgpu_ring_fini(&adev->gfx.gfx_ring[i]); for (i = 0; i < adev->gfx.num_compute_rings; i++) @@ -4840,6 +4751,9 @@ static int gfx_v9_0_early_init(void *handle) gfx_v9_0_set_gds_init(adev); gfx_v9_0_set_rlc_funcs(adev); + /* init rlcg reg access ctrl */ + gfx_v9_0_init_rlcg_reg_access_ctrl(adev); + return 0; } @@ -4870,16 +4784,9 @@ static int gfx_v9_0_ecc_late_init(void *handle) if (r) return r; - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->ras_late_init) { - r = adev->gfx.ras_funcs->ras_late_init(adev); - if (r) - return r; - } - - if (adev->gfx.ras_funcs && - adev->gfx.ras_funcs->enable_watchdog_timer) - adev->gfx.ras_funcs->enable_watchdog_timer(adev); + if (adev->gfx.ras && + adev->gfx.ras->enable_watchdog_timer) + adev->gfx.ras->enable_watchdog_timer(adev); return 0; } @@ -5250,8 +5157,6 @@ static const struct amdgpu_rlc_funcs gfx_v9_0_rlc_funcs = { .reset = gfx_v9_0_rlc_reset, .start = gfx_v9_0_rlc_start, .update_spm_vmid = gfx_v9_0_update_spm_vmid, - .sriov_wreg = gfx_v9_0_sriov_wreg, - .sriov_rreg = gfx_v9_0_sriov_rreg, .is_rlcg_access_range = gfx_v9_0_is_rlcg_access_range, }; @@ -6819,7 +6724,7 @@ static void gfx_v9_0_reset_ras_error_count(struct amdgpu_device *adev) WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, 255); } -static int gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev, +static void gfx_v9_0_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; @@ -6828,7 +6733,7 @@ static int gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev, uint32_t reg_value; if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) - return -EINVAL; + return; err_data->ue_count = 0; err_data->ce_count = 0; @@ -6857,8 +6762,6 @@ static int gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev, mutex_unlock(&adev->grbm_idx_mutex); gfx_v9_0_query_utc_edc_status(adev, err_data); - - return 0; } static void gfx_v9_0_emit_mem_sync(struct amdgpu_ring *ring) @@ -6962,6 +6865,7 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = { .align_mask = 0xff, .nop = PACKET3(PACKET3_NOP, 0x3FFF), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_GFXHUB_0, .get_rptr = gfx_v9_0_ring_get_rptr_gfx, .get_wptr = gfx_v9_0_ring_get_wptr_gfx, diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c index b4789dfc2bb9..c67e387a97f5 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c @@ -863,7 +863,7 @@ static int gfx_v9_4_ras_error_count(struct amdgpu_device *adev, return 0; } -static int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, +static void gfx_v9_4_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; @@ -872,7 +872,7 @@ static int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, uint32_t reg_value; if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) - return -EINVAL; + return; err_data->ue_count = 0; err_data->ce_count = 0; @@ -903,7 +903,6 @@ static int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, gfx_v9_4_query_utc_edc_status(adev, err_data); - return 0; } static void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev) @@ -1029,11 +1028,16 @@ static void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev) mutex_unlock(&adev->grbm_idx_mutex); } -const struct amdgpu_gfx_ras_funcs gfx_v9_4_ras_funcs = { - .ras_late_init = amdgpu_gfx_ras_late_init, - .ras_fini = amdgpu_gfx_ras_fini, - .ras_error_inject = &gfx_v9_4_ras_error_inject, - .query_ras_error_count = &gfx_v9_4_query_ras_error_count, - .reset_ras_error_count = &gfx_v9_4_reset_ras_error_count, - .query_ras_error_status = &gfx_v9_4_query_ras_error_status, + +const struct amdgpu_ras_block_hw_ops gfx_v9_4_ras_ops = { + .ras_error_inject = &gfx_v9_4_ras_error_inject, + .query_ras_error_count = &gfx_v9_4_query_ras_error_count, + .reset_ras_error_count = &gfx_v9_4_reset_ras_error_count, + .query_ras_error_status = &gfx_v9_4_query_ras_error_status, +}; + +struct amdgpu_gfx_ras gfx_v9_4_ras = { + .ras_block = { + .hw_ops = &gfx_v9_4_ras_ops, + }, }; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h index bdd16b568021..ca520a767267 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h @@ -24,6 +24,6 @@ #ifndef __GFX_V9_4_H__ #define __GFX_V9_4_H__ -extern const struct amdgpu_gfx_ras_funcs gfx_v9_4_ras_funcs; +extern struct amdgpu_gfx_ras gfx_v9_4_ras; #endif /* __GFX_V9_4_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c index c4f37a161875..7653ebd0e67b 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c @@ -1641,14 +1641,14 @@ static int gfx_v9_4_2_query_utc_edc_count(struct amdgpu_device *adev, return 0; } -static int gfx_v9_4_2_query_ras_error_count(struct amdgpu_device *adev, +static void gfx_v9_4_2_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; uint32_t sec_count = 0, ded_count = 0; if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) - return -EINVAL; + return; err_data->ue_count = 0; err_data->ce_count = 0; @@ -1661,7 +1661,6 @@ static int gfx_v9_4_2_query_ras_error_count(struct amdgpu_device *adev, err_data->ce_count += sec_count; err_data->ue_count += ded_count; - return 0; } static void gfx_v9_4_2_reset_utc_err_status(struct amdgpu_device *adev) @@ -1931,13 +1930,17 @@ static void gfx_v9_4_2_reset_sq_timeout_status(struct amdgpu_device *adev) mutex_unlock(&adev->grbm_idx_mutex); } -const struct amdgpu_gfx_ras_funcs gfx_v9_4_2_ras_funcs = { - .ras_late_init = amdgpu_gfx_ras_late_init, - .ras_fini = amdgpu_gfx_ras_fini, - .ras_error_inject = &gfx_v9_4_2_ras_error_inject, - .query_ras_error_count = &gfx_v9_4_2_query_ras_error_count, - .reset_ras_error_count = &gfx_v9_4_2_reset_ras_error_count, - .query_ras_error_status = &gfx_v9_4_2_query_ras_error_status, - .reset_ras_error_status = &gfx_v9_4_2_reset_ras_error_status, +struct amdgpu_ras_block_hw_ops gfx_v9_4_2_ras_ops = { + .ras_error_inject = &gfx_v9_4_2_ras_error_inject, + .query_ras_error_count = &gfx_v9_4_2_query_ras_error_count, + .reset_ras_error_count = &gfx_v9_4_2_reset_ras_error_count, + .query_ras_error_status = &gfx_v9_4_2_query_ras_error_status, + .reset_ras_error_status = &gfx_v9_4_2_reset_ras_error_status, +}; + +struct amdgpu_gfx_ras gfx_v9_4_2_ras = { + .ras_block = { + .hw_ops = &gfx_v9_4_2_ras_ops, + }, .enable_watchdog_timer = &gfx_v9_4_2_enable_watchdog_timer, }; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h index 6db1f88509af..7584624b641c 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h @@ -31,6 +31,6 @@ void gfx_v9_4_2_init_golden_registers(struct amdgpu_device *adev, void gfx_v9_4_2_set_power_brake_sequence(struct amdgpu_device *adev); int gfx_v9_4_2_do_edc_gpr_workarounds(struct amdgpu_device *adev); -extern const struct amdgpu_gfx_ras_funcs gfx_v9_4_2_ras_funcs; +extern struct amdgpu_gfx_ras gfx_v9_4_2_ras; #endif /* __GFX_V9_4_2_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c index b4eddf6e98a6..ff738e9725ee 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c @@ -543,7 +543,9 @@ static void gfxhub_v2_1_utcl2_harvest(struct amdgpu_device *adev) adev->gfx.config.max_sh_per_se * adev->gfx.config.max_shader_engines); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 3)) { + switch (adev->ip_versions[GC_HWIP][0]) { + case IP_VERSION(10, 3, 1): + case IP_VERSION(10, 3, 3): /* Get SA disabled bitmap from eFuse setting */ efuse_setting = RREG32_SOC15(GC, 0, mmCC_GC_SA_UNIT_DISABLE); efuse_setting &= CC_GC_SA_UNIT_DISABLE__SA_DISABLE_MASK; @@ -566,6 +568,9 @@ static void gfxhub_v2_1_utcl2_harvest(struct amdgpu_device *adev) disabled_sa = tmp; WREG32_SOC15(GC, 0, mmGCUTCL2_HARVEST_BYPASS_GROUPS_YELLOW_CARP, disabled_sa); + break; + default: + break; } } diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index 222b1da9d601..3c1d440824a7 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -666,11 +666,27 @@ static void gmc_v10_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V8_7_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V8_7_PER_CHANNEL_OFFSET_SIENNA; adev->umc.channel_idx_tbl = &umc_v8_7_channel_idx_tbl[0][0]; - adev->umc.ras_funcs = &umc_v8_7_ras_funcs; + adev->umc.ras = &umc_v8_7_ras; break; default: break; } + if (adev->umc.ras) { + amdgpu_ras_register_ras_block(adev, &adev->umc.ras->ras_block); + + strcpy(adev->umc.ras->ras_block.ras_comm.name, "umc"); + adev->umc.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__UMC; + adev->umc.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; + adev->umc.ras_if = &adev->umc.ras->ras_block.ras_comm; + + /* If don't define special ras_late_init function, use default ras_late_init */ + if (!adev->umc.ras->ras_block.ras_late_init) + adev->umc.ras->ras_block.ras_late_init = amdgpu_umc_ras_late_init; + + /* If not defined special ras_cb function, use default ras_cb */ + if (!adev->umc.ras->ras_block.ras_cb) + adev->umc.ras->ras_block.ras_cb = amdgpu_umc_process_ras_data_cb; + } } @@ -679,6 +695,7 @@ static void gmc_v10_0_set_mmhub_funcs(struct amdgpu_device *adev) switch (adev->ip_versions[MMHUB_HWIP][0]) { case IP_VERSION(2, 3, 0): case IP_VERSION(2, 4, 0): + case IP_VERSION(2, 4, 1): adev->mmhub.funcs = &mmhub_v2_3_funcs; break; default: @@ -695,7 +712,9 @@ static void gmc_v10_0_set_gfxhub_funcs(struct amdgpu_device *adev) case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): adev->gfxhub.funcs = &gfxhub_v2_1_funcs; break; default: @@ -707,6 +726,7 @@ static void gmc_v10_0_set_gfxhub_funcs(struct amdgpu_device *adev) static int gmc_v10_0_early_init(void *handle) { + int r; struct amdgpu_device *adev = (struct amdgpu_device *)handle; gmc_v10_0_set_mmhub_funcs(adev); @@ -722,6 +742,10 @@ static int gmc_v10_0_early_init(void *handle) adev->gmc.private_aperture_end = adev->gmc.private_aperture_start + (4ULL << 30) - 1; + r = amdgpu_gmc_ras_early_init(adev); + if (r) + return r; + return 0; } @@ -864,12 +888,15 @@ static int gmc_v10_0_sw_init(void *handle) case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 4): case IP_VERSION(10, 3, 5): + case IP_VERSION(10, 3, 6): case IP_VERSION(10, 3, 3): + case IP_VERSION(10, 3, 7): adev->num_vmhubs = 2; /* * To fulfill 4-level page support, @@ -921,7 +948,6 @@ static int gmc_v10_0_sw_init(void *handle) return r; amdgpu_gmc_get_vbios_allocations(adev); - amdgpu_gmc_get_reserved_allocation(adev); /* Memory manager */ r = amdgpu_bo_init(adev); @@ -988,14 +1014,7 @@ static int gmc_v10_0_gart_enable(struct amdgpu_device *adev) return -EINVAL; } - if (amdgpu_sriov_vf(adev) && amdgpu_in_reset(adev)) - goto skip_pin_bo; - - r = amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); - if (r) - return r; - -skip_pin_bo: + amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); r = adev->gfxhub.funcs->gart_enable(adev); if (r) return r; @@ -1021,8 +1040,6 @@ skip_pin_bo: (unsigned)(adev->gmc.gart_size >> 20), (unsigned long long)amdgpu_bo_gpu_offset(adev->gart.bo)); - adev->gart.ready = true; - return 0; } @@ -1045,6 +1062,12 @@ static int gmc_v10_0_hw_init(void *handle) if (r) return r; + if (amdgpu_emu_mode == 1) { + r = amdgpu_gmc_vram_checking(adev); + if (r) + return r; + } + if (adev->umc.funcs && adev->umc.funcs->init_registers) adev->umc.funcs->init_registers(adev); @@ -1142,6 +1165,10 @@ static void gmc_v10_0_get_clockgating_state(void *handle, u32 *flags) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 3) || + adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 4)) + return; + adev->mmhub.funcs->get_clockgating(adev, flags); if (adev->ip_versions[ATHUB_HWIP][0] >= IP_VERSION(2, 1, 0)) diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c index cd6c38e083d0..ec291d28edff 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c @@ -469,16 +469,14 @@ static void gmc_v6_0_set_prt(struct amdgpu_device *adev, bool enable) static int gmc_v6_0_gart_enable(struct amdgpu_device *adev) { uint64_t table_addr; - int r, i; u32 field; + int i; if (adev->gart.bo == NULL) { dev_err(adev->dev, "No VRAM object for PCIE GART.\n"); return -EINVAL; } - r = amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); - if (r) - return r; + amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); table_addr = amdgpu_bo_gpu_offset(adev->gart.bo); @@ -558,7 +556,6 @@ static int gmc_v6_0_gart_enable(struct amdgpu_device *adev) dev_info(adev->dev, "PCIE GART of %uM enabled (table at 0x%016llX).\n", (unsigned)(adev->gmc.gart_size >> 20), (unsigned long long)table_addr); - adev->gart.ready = true; return 0; } @@ -922,7 +919,10 @@ static int gmc_v6_0_hw_init(void *handle) if (r) return r; - return r; + if (amdgpu_emu_mode == 1) + return amdgpu_gmc_vram_checking(adev); + else + return r; } static int gmc_v6_0_hw_fini(void *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c index ab8adbff9e2d..344d819b4c1b 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c @@ -613,17 +613,14 @@ static void gmc_v7_0_set_prt(struct amdgpu_device *adev, bool enable) static int gmc_v7_0_gart_enable(struct amdgpu_device *adev) { uint64_t table_addr; - int r, i; u32 tmp, field; + int i; if (adev->gart.bo == NULL) { dev_err(adev->dev, "No VRAM object for PCIE GART.\n"); return -EINVAL; } - r = amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); - if (r) - return r; - + amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); table_addr = amdgpu_bo_gpu_offset(adev->gart.bo); /* Setup TLB control */ @@ -712,7 +709,6 @@ static int gmc_v7_0_gart_enable(struct amdgpu_device *adev) DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n", (unsigned)(adev->gmc.gart_size >> 20), (unsigned long long)table_addr); - adev->gart.ready = true; return 0; } @@ -1111,7 +1107,10 @@ static int gmc_v7_0_hw_init(void *handle) if (r) return r; - return r; + if (amdgpu_emu_mode == 1) + return amdgpu_gmc_vram_checking(adev); + else + return r; } static int gmc_v7_0_hw_fini(void *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index 054733838292..ca9841d5669f 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -837,17 +837,14 @@ static void gmc_v8_0_set_prt(struct amdgpu_device *adev, bool enable) static int gmc_v8_0_gart_enable(struct amdgpu_device *adev) { uint64_t table_addr; - int r, i; u32 tmp, field; + int i; if (adev->gart.bo == NULL) { dev_err(adev->dev, "No VRAM object for PCIE GART.\n"); return -EINVAL; } - r = amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); - if (r) - return r; - + amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); table_addr = amdgpu_bo_gpu_offset(adev->gart.bo); /* Setup TLB control */ @@ -953,7 +950,6 @@ static int gmc_v8_0_gart_enable(struct amdgpu_device *adev) DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n", (unsigned)(adev->gmc.gart_size >> 20), (unsigned long long)table_addr); - adev->gart.ready = true; return 0; } @@ -1242,7 +1238,10 @@ static int gmc_v8_0_hw_init(void *handle) if (r) return r; - return r; + if (amdgpu_emu_mode == 1) + return amdgpu_gmc_vram_checking(adev); + else + return r; } static int gmc_v8_0_hw_fini(void *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 3a5efe969735..431742eb7811 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -1204,7 +1204,7 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V6_1_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V6_1_PER_CHANNEL_OFFSET_VG20; adev->umc.channel_idx_tbl = &umc_v6_1_channel_idx_tbl[0][0]; - adev->umc.ras_funcs = &umc_v6_1_ras_funcs; + adev->umc.ras = &umc_v6_1_ras; break; case IP_VERSION(6, 1, 2): adev->umc.max_ras_err_cnt_per_query = UMC_V6_1_TOTAL_CHANNEL_NUM; @@ -1212,15 +1212,16 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V6_1_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V6_1_PER_CHANNEL_OFFSET_ARCT; adev->umc.channel_idx_tbl = &umc_v6_1_channel_idx_tbl[0][0]; - adev->umc.ras_funcs = &umc_v6_1_ras_funcs; + adev->umc.ras = &umc_v6_1_ras; break; case IP_VERSION(6, 7, 0): - adev->umc.max_ras_err_cnt_per_query = UMC_V6_7_TOTAL_CHANNEL_NUM; + adev->umc.max_ras_err_cnt_per_query = + UMC_V6_7_TOTAL_CHANNEL_NUM * UMC_V6_7_BAD_PAGE_NUM_PER_CHANNEL; adev->umc.channel_inst_num = UMC_V6_7_CHANNEL_INSTANCE_NUM; adev->umc.umc_inst_num = UMC_V6_7_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V6_7_PER_CHANNEL_OFFSET; if (!adev->gmc.xgmi.connected_to_cpu) - adev->umc.ras_funcs = &umc_v6_7_ras_funcs; + adev->umc.ras = &umc_v6_7_ras; if (1 & adev->smuio.funcs->get_die_id(adev)) adev->umc.channel_idx_tbl = &umc_v6_7_channel_idx_tbl_first[0][0]; else @@ -1229,6 +1230,23 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) default: break; } + + if (adev->umc.ras) { + amdgpu_ras_register_ras_block(adev, &adev->umc.ras->ras_block); + + strcpy(adev->umc.ras->ras_block.ras_comm.name, "umc"); + adev->umc.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__UMC; + adev->umc.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; + adev->umc.ras_if = &adev->umc.ras->ras_block.ras_comm; + + /* If don't define special ras_late_init function, use default ras_late_init */ + if (!adev->umc.ras->ras_block.ras_late_init) + adev->umc.ras->ras_block.ras_late_init = amdgpu_umc_ras_late_init; + + /* If not defined special ras_cb function, use default ras_cb */ + if (!adev->umc.ras->ras_block.ras_cb) + adev->umc.ras->ras_block.ras_cb = amdgpu_umc_process_ras_data_cb; + } } static void gmc_v9_0_set_mmhub_funcs(struct amdgpu_device *adev) @@ -1250,18 +1268,27 @@ static void gmc_v9_0_set_mmhub_ras_funcs(struct amdgpu_device *adev) { switch (adev->ip_versions[MMHUB_HWIP][0]) { case IP_VERSION(9, 4, 0): - adev->mmhub.ras_funcs = &mmhub_v1_0_ras_funcs; + adev->mmhub.ras = &mmhub_v1_0_ras; break; case IP_VERSION(9, 4, 1): - adev->mmhub.ras_funcs = &mmhub_v9_4_ras_funcs; + adev->mmhub.ras = &mmhub_v9_4_ras; break; case IP_VERSION(9, 4, 2): - adev->mmhub.ras_funcs = &mmhub_v1_7_ras_funcs; + adev->mmhub.ras = &mmhub_v1_7_ras; break; default: /* mmhub ras is not available */ break; } + + if (adev->mmhub.ras) { + amdgpu_ras_register_ras_block(adev, &adev->mmhub.ras->ras_block); + + strcpy(adev->mmhub.ras->ras_block.ras_comm.name, "mmhub"); + adev->mmhub.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__MMHUB; + adev->mmhub.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; + adev->mmhub.ras_if = &adev->mmhub.ras->ras_block.ras_comm; + } } static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev) @@ -1271,7 +1298,9 @@ static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev) static void gmc_v9_0_set_hdp_ras_funcs(struct amdgpu_device *adev) { - adev->hdp.ras_funcs = &hdp_v4_0_ras_funcs; + adev->hdp.ras = &hdp_v4_0_ras; + amdgpu_ras_register_ras_block(adev, &adev->hdp.ras->ras_block); + adev->hdp.ras_if = &adev->hdp.ras->ras_block.ras_comm; } static void gmc_v9_0_set_mca_funcs(struct amdgpu_device *adev) @@ -1289,6 +1318,7 @@ static void gmc_v9_0_set_mca_funcs(struct amdgpu_device *adev) static int gmc_v9_0_early_init(void *handle) { + int r; struct amdgpu_device *adev = (struct amdgpu_device *)handle; /* ARCT and VEGA20 don't have XGMI defined in their IP discovery tables */ @@ -1318,6 +1348,10 @@ static int gmc_v9_0_early_init(void *handle) adev->gmc.private_aperture_end = adev->gmc.private_aperture_start + (4ULL << 30) - 1; + r = amdgpu_gmc_ras_early_init(adev); + if (r) + return r; + return 0; } @@ -1344,13 +1378,13 @@ static int gmc_v9_0_late_init(void *handle) } if (!amdgpu_persistent_edc_harvesting_supported(adev)) { - if (adev->mmhub.ras_funcs && - adev->mmhub.ras_funcs->reset_ras_error_count) - adev->mmhub.ras_funcs->reset_ras_error_count(adev); + if (adev->mmhub.ras && adev->mmhub.ras->ras_block.hw_ops && + adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count) + adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count(adev); - if (adev->hdp.ras_funcs && - adev->hdp.ras_funcs->reset_ras_error_count) - adev->hdp.ras_funcs->reset_ras_error_count(adev); + if (adev->hdp.ras && adev->hdp.ras->ras_block.hw_ops && + adev->hdp.ras->ras_block.hw_ops->reset_ras_error_count) + adev->hdp.ras->ras_block.hw_ops->reset_ras_error_count(adev); } r = amdgpu_gmc_ras_late_init(adev); @@ -1519,7 +1553,7 @@ static void gmc_v9_0_save_registers(struct amdgpu_device *adev) static int gmc_v9_0_sw_init(void *handle) { - int r, vram_width = 0, vram_type = 0, vram_vendor = 0; + int r, vram_width = 0, vram_type = 0, vram_vendor = 0, dma_addr_bits; struct amdgpu_device *adev = (struct amdgpu_device *)handle; adev->gfxhub.funcs->init(adev); @@ -1635,12 +1669,13 @@ static int gmc_v9_0_sw_init(void *handle) */ adev->gmc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */ - r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(44)); + dma_addr_bits = adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2) ? 48:44; + r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(dma_addr_bits)); if (r) { printk(KERN_WARNING "amdgpu: No suitable DMA available.\n"); return r; } - adev->need_swiotlb = drm_need_swiotlb(44); + adev->need_swiotlb = drm_need_swiotlb(dma_addr_bits); r = gmc_v9_0_mc_init(adev); if (r) @@ -1754,14 +1789,7 @@ static int gmc_v9_0_gart_enable(struct amdgpu_device *adev) return -EINVAL; } - if (amdgpu_sriov_vf(adev) && amdgpu_in_reset(adev)) - goto skip_pin_bo; - - r = amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); - if (r) - return r; - -skip_pin_bo: + amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr); r = adev->gfxhub.funcs->gart_enable(adev); if (r) return r; @@ -1778,7 +1806,6 @@ skip_pin_bo: DRM_INFO("PTB located at 0x%016llX\n", (unsigned long long)amdgpu_bo_gpu_offset(adev->gart.bo)); - adev->gart.ready = true; return 0; } @@ -1786,7 +1813,7 @@ static int gmc_v9_0_hw_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; bool value; - int i; + int i, r; /* The sequence of these two function calls matters.*/ gmc_v9_0_init_golden_registers(adev); @@ -1821,7 +1848,14 @@ static int gmc_v9_0_hw_init(void *handle) if (adev->umc.funcs && adev->umc.funcs->init_registers) adev->umc.funcs->init_registers(adev); - return gmc_v9_0_gart_enable(adev); + r = gmc_v9_0_gart_enable(adev); + if (r) + return r; + + if (amdgpu_emu_mode == 1) + return amdgpu_gmc_vram_checking(adev); + else + return r; } /** diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c index eecfb1545c1e..046216635262 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c @@ -146,17 +146,29 @@ static void hdp_v4_0_init_registers(struct amdgpu_device *adev) WREG32_FIELD15(HDP, 0, HDP_MISC_CNTL, FLUSH_INVALIDATE_CACHE, 1); + if (adev->ip_versions[HDP_HWIP][0] == IP_VERSION(4, 4, 0)) + WREG32_FIELD15(HDP, 0, HDP_MISC_CNTL, READ_BUFFER_WATERMARK, 2); + WREG32_SOC15(HDP, 0, mmHDP_NONSURFACE_BASE, (adev->gmc.vram_start >> 8)); WREG32_SOC15(HDP, 0, mmHDP_NONSURFACE_BASE_HI, (adev->gmc.vram_start >> 40)); } -const struct amdgpu_hdp_ras_funcs hdp_v4_0_ras_funcs = { - .ras_late_init = amdgpu_hdp_ras_late_init, - .ras_fini = amdgpu_hdp_ras_fini, +struct amdgpu_ras_block_hw_ops hdp_v4_0_ras_hw_ops = { .query_ras_error_count = hdp_v4_0_query_ras_error_count, .reset_ras_error_count = hdp_v4_0_reset_ras_error_count, }; +struct amdgpu_hdp_ras hdp_v4_0_ras = { + .ras_block = { + .ras_comm = { + .name = "hdp", + .block = AMDGPU_RAS_BLOCK__HDP, + .type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE, + }, + .hw_ops = &hdp_v4_0_ras_hw_ops, + }, +}; + const struct amdgpu_hdp_funcs hdp_v4_0_funcs = { .flush_hdp = hdp_v4_0_flush_hdp, .invalidate_hdp = hdp_v4_0_invalidate_hdp, diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.h b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.h index dc3a1b81dd62..c44eee9282ab 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.h +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.h @@ -27,6 +27,6 @@ #include "soc15_common.h" extern const struct amdgpu_hdp_funcs hdp_v4_0_funcs; -extern const struct amdgpu_hdp_ras_funcs hdp_v4_0_ras_funcs; +extern struct amdgpu_hdp_ras hdp_v4_0_ras; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c index 01c242c5abc3..41a00851b6c5 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c @@ -50,11 +50,16 @@ static int jpeg_v3_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (adev->asic_type != CHIP_YELLOW_CARP) { - u32 harvest = RREG32_SOC15(JPEG, 0, mmCC_UVD_HARVESTING); + u32 harvest; + switch (adev->ip_versions[UVD_HWIP][0]) { + case IP_VERSION(3, 1, 1): + break; + default: + harvest = RREG32_SOC15(JPEG, 0, mmCC_UVD_HARVESTING); if (harvest & CC_UVD_HARVESTING__UVD_DISABLE_MASK) return -ENOENT; + break; } adev->jpeg.num_jpeg_inst = 1; diff --git a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c index 8f7107d392af..d4bd7d1d2649 100644 --- a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c @@ -37,24 +37,36 @@ static void mca_v3_0_mp0_query_ras_error_count(struct amdgpu_device *adev, ras_error_status); } -static int mca_v3_0_mp0_ras_late_init(struct amdgpu_device *adev) +static int mca_v3_0_ras_block_match(struct amdgpu_ras_block_object *block_obj, + enum amdgpu_ras_block block, uint32_t sub_block_index) { - return amdgpu_mca_ras_late_init(adev, &adev->mca.mp0); -} + if (!block_obj) + return -EINVAL; -static void mca_v3_0_mp0_ras_fini(struct amdgpu_device *adev) -{ - amdgpu_mca_ras_fini(adev, &adev->mca.mp0); + if ((block_obj->ras_comm.block == block) && + (block_obj->ras_comm.sub_block_index == sub_block_index)) { + return 0; + } + + return -EINVAL; } -const struct amdgpu_mca_ras_funcs mca_v3_0_mp0_ras_funcs = { - .ras_late_init = mca_v3_0_mp0_ras_late_init, - .ras_fini = mca_v3_0_mp0_ras_fini, +const struct amdgpu_ras_block_hw_ops mca_v3_0_mp0_hw_ops = { .query_ras_error_count = mca_v3_0_mp0_query_ras_error_count, .query_ras_error_address = NULL, - .ras_block = AMDGPU_RAS_BLOCK__MCA, - .ras_sub_block = AMDGPU_RAS_MCA_BLOCK__MP0, - .sysfs_name = "mp0_err_count", +}; + +struct amdgpu_mca_ras_block mca_v3_0_mp0_ras = { + .ras_block = { + .ras_comm = { + .block = AMDGPU_RAS_BLOCK__MCA, + .sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP0, + .type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE, + .name = "mp0", + }, + .hw_ops = &mca_v3_0_mp0_hw_ops, + .ras_block_match = mca_v3_0_ras_block_match, + }, }; static void mca_v3_0_mp1_query_ras_error_count(struct amdgpu_device *adev, @@ -65,24 +77,22 @@ static void mca_v3_0_mp1_query_ras_error_count(struct amdgpu_device *adev, ras_error_status); } -static int mca_v3_0_mp1_ras_late_init(struct amdgpu_device *adev) -{ - return amdgpu_mca_ras_late_init(adev, &adev->mca.mp1); -} - -static void mca_v3_0_mp1_ras_fini(struct amdgpu_device *adev) -{ - amdgpu_mca_ras_fini(adev, &adev->mca.mp1); -} - -const struct amdgpu_mca_ras_funcs mca_v3_0_mp1_ras_funcs = { - .ras_late_init = mca_v3_0_mp1_ras_late_init, - .ras_fini = mca_v3_0_mp1_ras_fini, +const struct amdgpu_ras_block_hw_ops mca_v3_0_mp1_hw_ops = { .query_ras_error_count = mca_v3_0_mp1_query_ras_error_count, .query_ras_error_address = NULL, - .ras_block = AMDGPU_RAS_BLOCK__MCA, - .ras_sub_block = AMDGPU_RAS_MCA_BLOCK__MP1, - .sysfs_name = "mp1_err_count", +}; + +struct amdgpu_mca_ras_block mca_v3_0_mp1_ras = { + .ras_block = { + .ras_comm = { + .block = AMDGPU_RAS_BLOCK__MCA, + .sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP1, + .type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE, + .name = "mp1", + }, + .hw_ops = &mca_v3_0_mp1_hw_ops, + .ras_block_match = mca_v3_0_ras_block_match, + }, }; static void mca_v3_0_mpio_query_ras_error_count(struct amdgpu_device *adev, @@ -93,24 +103,22 @@ static void mca_v3_0_mpio_query_ras_error_count(struct amdgpu_device *adev, ras_error_status); } -static int mca_v3_0_mpio_ras_late_init(struct amdgpu_device *adev) -{ - return amdgpu_mca_ras_late_init(adev, &adev->mca.mpio); -} - -static void mca_v3_0_mpio_ras_fini(struct amdgpu_device *adev) -{ - amdgpu_mca_ras_fini(adev, &adev->mca.mpio); -} - -const struct amdgpu_mca_ras_funcs mca_v3_0_mpio_ras_funcs = { - .ras_late_init = mca_v3_0_mpio_ras_late_init, - .ras_fini = mca_v3_0_mpio_ras_fini, +const struct amdgpu_ras_block_hw_ops mca_v3_0_mpio_hw_ops = { .query_ras_error_count = mca_v3_0_mpio_query_ras_error_count, .query_ras_error_address = NULL, - .ras_block = AMDGPU_RAS_BLOCK__MCA, - .ras_sub_block = AMDGPU_RAS_MCA_BLOCK__MPIO, - .sysfs_name = "mpio_err_count", +}; + +struct amdgpu_mca_ras_block mca_v3_0_mpio_ras = { + .ras_block = { + .ras_comm = { + .block = AMDGPU_RAS_BLOCK__MCA, + .sub_block_index = AMDGPU_RAS_MCA_BLOCK__MPIO, + .type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE, + .name = "mpio", + }, + .hw_ops = &mca_v3_0_mpio_hw_ops, + .ras_block_match = mca_v3_0_ras_block_match, + }, }; @@ -118,9 +126,15 @@ static void mca_v3_0_init(struct amdgpu_device *adev) { struct amdgpu_mca *mca = &adev->mca; - mca->mp0.ras_funcs = &mca_v3_0_mp0_ras_funcs; - mca->mp1.ras_funcs = &mca_v3_0_mp1_ras_funcs; - mca->mpio.ras_funcs = &mca_v3_0_mpio_ras_funcs; + mca->mp0.ras = &mca_v3_0_mp0_ras; + mca->mp1.ras = &mca_v3_0_mp1_ras; + mca->mpio.ras = &mca_v3_0_mpio_ras; + amdgpu_ras_register_ras_block(adev, &mca->mp0.ras->ras_block); + amdgpu_ras_register_ras_block(adev, &mca->mp1.ras->ras_block); + amdgpu_ras_register_ras_block(adev, &mca->mpio.ras->ras_block); + mca->mp0.ras_if = &mca->mp0.ras->ras_block.ras_comm; + mca->mp1.ras_if = &mca->mp1.ras->ras_block.ras_comm; + mca->mpio.ras_if = &mca->mpio.ras->ras_block.ras_comm; } const struct amdgpu_mca_funcs mca_v3_0_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c index 1da2ec692057..4c9f0c0f3116 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c @@ -774,13 +774,17 @@ static void mmhub_v1_0_reset_ras_error_count(struct amdgpu_device *adev) } } -const struct amdgpu_mmhub_ras_funcs mmhub_v1_0_ras_funcs = { - .ras_late_init = amdgpu_mmhub_ras_late_init, - .ras_fini = amdgpu_mmhub_ras_fini, +struct amdgpu_ras_block_hw_ops mmhub_v1_0_ras_hw_ops = { .query_ras_error_count = mmhub_v1_0_query_ras_error_count, .reset_ras_error_count = mmhub_v1_0_reset_ras_error_count, }; +struct amdgpu_mmhub_ras mmhub_v1_0_ras = { + .ras_block = { + .hw_ops = &mmhub_v1_0_ras_hw_ops, + }, +}; + const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs = { .get_fb_location = mmhub_v1_0_get_fb_location, .init = mmhub_v1_0_init, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h index 4661b094e007..dae7ca48bd8b 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h @@ -24,6 +24,6 @@ #define __MMHUB_V1_0_H__ extern const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs; -extern const struct amdgpu_mmhub_ras_funcs mmhub_v1_0_ras_funcs; +extern struct amdgpu_mmhub_ras mmhub_v1_0_ras; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c index f5f7181f9af5..3b901f941627 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c @@ -1321,15 +1321,19 @@ static void mmhub_v1_7_reset_ras_error_status(struct amdgpu_device *adev) } } -const struct amdgpu_mmhub_ras_funcs mmhub_v1_7_ras_funcs = { - .ras_late_init = amdgpu_mmhub_ras_late_init, - .ras_fini = amdgpu_mmhub_ras_fini, +struct amdgpu_ras_block_hw_ops mmhub_v1_7_ras_hw_ops = { .query_ras_error_count = mmhub_v1_7_query_ras_error_count, .reset_ras_error_count = mmhub_v1_7_reset_ras_error_count, .query_ras_error_status = mmhub_v1_7_query_ras_error_status, .reset_ras_error_status = mmhub_v1_7_reset_ras_error_status, }; +struct amdgpu_mmhub_ras mmhub_v1_7_ras = { + .ras_block = { + .hw_ops = &mmhub_v1_7_ras_hw_ops, + }, +}; + const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs = { .get_fb_location = mmhub_v1_7_get_fb_location, .init = mmhub_v1_7_init, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h index a7f9dfc24697..629f49052137 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h @@ -24,6 +24,6 @@ #define __MMHUB_V1_7_H__ extern const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs; -extern const struct amdgpu_mmhub_ras_funcs mmhub_v1_7_ras_funcs; +extern struct amdgpu_mmhub_ras mmhub_v1_7_ras; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c index 9e16da28505a..1957fb098c4d 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c @@ -93,6 +93,7 @@ mmhub_v2_3_print_l2_protection_fault_status(struct amdgpu_device *adev, switch (adev->ip_versions[MMHUB_HWIP][0]) { case IP_VERSION(2, 3, 0): case IP_VERSION(2, 4, 0): + case IP_VERSION(2, 4, 1): mmhub_cid = mmhub_client_ids_vangogh[cid][rw]; break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c index ff49eeaf7882..619106f7d23d 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c @@ -1655,14 +1655,18 @@ static void mmhub_v9_4_query_ras_error_status(struct amdgpu_device *adev) } } -const struct amdgpu_mmhub_ras_funcs mmhub_v9_4_ras_funcs = { - .ras_late_init = amdgpu_mmhub_ras_late_init, - .ras_fini = amdgpu_mmhub_ras_fini, +const struct amdgpu_ras_block_hw_ops mmhub_v9_4_ras_hw_ops = { .query_ras_error_count = mmhub_v9_4_query_ras_error_count, .reset_ras_error_count = mmhub_v9_4_reset_ras_error_count, .query_ras_error_status = mmhub_v9_4_query_ras_error_status, }; +struct amdgpu_mmhub_ras mmhub_v9_4_ras = { + .ras_block = { + .hw_ops = &mmhub_v9_4_ras_hw_ops, + }, +}; + const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = { .get_fb_location = mmhub_v9_4_get_fb_location, .init = mmhub_v9_4_init, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h index 90436efa92ef..a48329d95f71 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h @@ -24,6 +24,6 @@ #define __MMHUB_V9_4_H__ extern const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs; -extern const struct amdgpu_mmhub_ras_funcs mmhub_v9_4_ras_funcs; +extern struct amdgpu_mmhub_ras mmhub_v9_4_ras; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c index 531cfba759dd..7b63d30b9b79 100644 --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c @@ -546,7 +546,7 @@ static int xgpu_vi_mailbox_rcv_irq(struct amdgpu_device *adev, { int r; - /* trigger gpu-reset by hypervisor only if TDR disbaled */ + /* trigger gpu-reset by hypervisor only if TDR disabled */ if (!amdgpu_gpu_recovery) { /* see what event we get */ r = xgpu_vi_mailbox_rcv_msg(adev, IDH_FLR_NOTIFICATION); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c index 3444332ea110..6f81de6f3cc4 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c @@ -59,10 +59,16 @@ static u32 nbio_v7_2_get_rev_id(struct amdgpu_device *adev) { u32 tmp; - if (adev->asic_type == CHIP_YELLOW_CARP) + switch (adev->ip_versions[NBIO_HWIP][0]) { + case IP_VERSION(7, 2, 1): + case IP_VERSION(7, 3, 0): + case IP_VERSION(7, 5, 0): tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0_YC); - else + break; + default: tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0); + break; + } 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; @@ -72,20 +78,26 @@ static u32 nbio_v7_2_get_rev_id(struct amdgpu_device *adev) static void nbio_v7_2_mc_access_enable(struct amdgpu_device *adev, bool enable) { - if (enable) - if (adev->asic_type == CHIP_YELLOW_CARP) + switch (adev->ip_versions[NBIO_HWIP][0]) { + case IP_VERSION(7, 2, 1): + case IP_VERSION(7, 3, 0): + case IP_VERSION(7, 5, 0): + if (enable) WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN_YC, 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_YC, 0); + break; + default: + 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 - if (adev->asic_type == CHIP_YELLOW_CARP) - WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN_YC, 0); else WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_FB_EN, 0); + break; + } } static u32 nbio_v7_2_get_memsize(struct amdgpu_device *adev) @@ -250,7 +262,10 @@ static void nbio_v7_2_update_medium_grain_light_sleep(struct amdgpu_device *adev { uint32_t def, data; - if (adev->asic_type == CHIP_YELLOW_CARP) { + switch (adev->ip_versions[NBIO_HWIP][0]) { + case IP_VERSION(7, 2, 1): + case IP_VERSION(7, 3, 0): + case IP_VERSION(7, 5, 0): 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; @@ -260,8 +275,8 @@ static void nbio_v7_2_update_medium_grain_light_sleep(struct amdgpu_device *adev if (def != data) WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2), data); - data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regBIF1_PCIE_TX_POWER_CTRL_1)); - def = data; + def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, + regBIF1_PCIE_TX_POWER_CTRL_1)); if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) data |= (BIF1_PCIE_TX_POWER_CTRL_1__MST_MEM_LS_EN_MASK | BIF1_PCIE_TX_POWER_CTRL_1__REPLAY_MEM_LS_EN_MASK); @@ -272,7 +287,8 @@ static void nbio_v7_2_update_medium_grain_light_sleep(struct amdgpu_device *adev if (def != data) WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regBIF1_PCIE_TX_POWER_CTRL_1), data); - } else { + break; + default: 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 | @@ -285,6 +301,7 @@ static void nbio_v7_2_update_medium_grain_light_sleep(struct amdgpu_device *adev if (def != data) WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CNTL2), data); + break; } } @@ -352,7 +369,10 @@ const struct nbio_hdp_flush_reg nbio_v7_2_hdp_flush_reg = { static void nbio_v7_2_init_registers(struct amdgpu_device *adev) { uint32_t def, data; - if (adev->asic_type == CHIP_YELLOW_CARP) { + switch (adev->ip_versions[NBIO_HWIP][0]) { + case IP_VERSION(7, 2, 1): + case IP_VERSION(7, 3, 0): + case IP_VERSION(7, 5, 0): def = data = RREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regBIF1_PCIE_MST_CTRL_3)); data = REG_SET_FIELD(data, BIF1_PCIE_MST_CTRL_3, CI_SWUS_MAX_READ_REQUEST_SIZE_MODE, 1); @@ -361,7 +381,8 @@ static void nbio_v7_2_init_registers(struct amdgpu_device *adev) if (def != data) WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regBIF1_PCIE_MST_CTRL_3), data); - } else { + break; + default: 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); @@ -370,6 +391,7 @@ static void nbio_v7_2_init_registers(struct amdgpu_device *adev) if (def != data) WREG32_PCIE_PORT(SOC15_REG_OFFSET(NBIO, 0, regPCIE_CONFIG_CNTL), data); + break; } if (amdgpu_sriov_vf(adev)) diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c index dc5e93756fea..c2357e83a8c4 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c @@ -362,9 +362,24 @@ const struct nbio_hdp_flush_reg nbio_v7_4_hdp_flush_reg_ald = { static void nbio_v7_4_init_registers(struct amdgpu_device *adev) { + uint32_t baco_cntl; + if (amdgpu_sriov_vf(adev)) adev->rmmio_remap.reg_offset = SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_DEV0_EPF0_VF0_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2; + + if (adev->ip_versions[NBIO_HWIP][0] == IP_VERSION(7, 4, 4) && + !amdgpu_sriov_vf(adev)) { + baco_cntl = RREG32_SOC15(NBIO, 0, mmBACO_CNTL); + if (baco_cntl & + (BACO_CNTL__BACO_DUMMY_EN_MASK | BACO_CNTL__BACO_EN_MASK)) { + baco_cntl &= ~(BACO_CNTL__BACO_DUMMY_EN_MASK | + BACO_CNTL__BACO_EN_MASK); + dev_dbg(adev->dev, "Unsetting baco dummy mode %x", + baco_cntl); + WREG32_SOC15(NBIO, 0, mmBACO_CNTL, baco_cntl); + } + } } static void nbio_v7_4_handle_ras_controller_intr_no_bifring(struct amdgpu_device *adev) @@ -658,16 +673,27 @@ static void nbio_v7_4_enable_doorbell_interrupt(struct amdgpu_device *adev, DOORBELL_INTERRUPT_DISABLE, enable ? 0 : 1); } -const struct amdgpu_nbio_ras_funcs nbio_v7_4_ras_funcs = { +const struct amdgpu_ras_block_hw_ops nbio_v7_4_ras_hw_ops = { + .query_ras_error_count = nbio_v7_4_query_ras_error_count, +}; + +struct amdgpu_nbio_ras nbio_v7_4_ras = { + .ras_block = { + .ras_comm = { + .name = "pcie_bif", + .block = AMDGPU_RAS_BLOCK__PCIE_BIF, + .type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE, + }, + .hw_ops = &nbio_v7_4_ras_hw_ops, + .ras_late_init = amdgpu_nbio_ras_late_init, + }, .handle_ras_controller_intr_no_bifring = nbio_v7_4_handle_ras_controller_intr_no_bifring, .handle_ras_err_event_athub_intr_no_bifring = nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring, .init_ras_controller_interrupt = nbio_v7_4_init_ras_controller_interrupt, .init_ras_err_event_athub_interrupt = nbio_v7_4_init_ras_err_event_athub_interrupt, - .query_ras_error_count = nbio_v7_4_query_ras_error_count, - .ras_late_init = amdgpu_nbio_ras_late_init, - .ras_fini = amdgpu_nbio_ras_fini, }; + static void nbio_v7_4_program_ltr(struct amdgpu_device *adev) { uint32_t def, data; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h index cc5692db6f98..7490022d79d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h @@ -29,6 +29,6 @@ extern const struct nbio_hdp_flush_reg nbio_v7_4_hdp_flush_reg; extern const struct nbio_hdp_flush_reg nbio_v7_4_hdp_flush_reg_ald; extern const struct amdgpu_nbio_funcs nbio_v7_4_funcs; -extern const struct amdgpu_nbio_ras_funcs nbio_v7_4_ras_funcs; +extern struct amdgpu_nbio_ras nbio_v7_4_ras; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c index 2ec1ffb36b1f..e19f14c3ef59 100644 --- a/drivers/gpu/drm/amd/amdgpu/nv.c +++ b/drivers/gpu/drm/amd/amdgpu/nv.c @@ -204,6 +204,7 @@ static int nv_query_video_codecs(struct amdgpu_device *adev, bool encode, *codecs = &sc_video_codecs_decode; return 0; case IP_VERSION(3, 1, 1): + case IP_VERSION(3, 1, 2): if (encode) *codecs = &nv_video_codecs_encode; else @@ -258,21 +259,6 @@ 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; @@ -283,21 +269,6 @@ 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; @@ -360,38 +331,6 @@ static bool nv_read_disabled_bios(struct amdgpu_device *adev) return false; } -static bool nv_read_bios_from_rom(struct amdgpu_device *adev, - u8 *bios, u32 length_bytes) -{ - u32 *dw_ptr; - u32 i, length_dw; - u32 rom_index_offset, rom_data_offset; - - if (bios == NULL) - return false; - if (length_bytes == 0) - return false; - /* APU vbios image is part of sbios image */ - if (adev->flags & AMD_IS_APU) - return false; - - dw_ptr = (u32 *)bios; - length_dw = ALIGN(length_bytes, 4) / 4; - - rom_index_offset = - adev->smuio.funcs->get_rom_index_offset(adev); - rom_data_offset = - adev->smuio.funcs->get_rom_data_offset(adev); - - /* set rom index to 0 */ - WREG32(rom_index_offset, 0); - /* read out the rom data */ - for (i = 0; i < length_dw; i++) - dw_ptr[i] = RREG32(rom_data_offset); - - return true; -} - static struct soc15_allowed_register_entry nv_allowed_read_registers[] = { { SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS)}, { SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS2)}, @@ -515,6 +454,8 @@ nv_asic_reset_method(struct amdgpu_device *adev) case IP_VERSION(11, 5, 0): case IP_VERSION(13, 0, 1): case IP_VERSION(13, 0, 3): + case IP_VERSION(13, 0, 5): + case IP_VERSION(13, 0, 8): return AMD_RESET_METHOD_MODE2; case IP_VERSION(11, 0, 7): case IP_VERSION(11, 0, 11): @@ -584,7 +525,7 @@ static void nv_pcie_gen3_enable(struct amdgpu_device *adev) static void nv_program_aspm(struct amdgpu_device *adev) { - if (!amdgpu_aspm) + if (!amdgpu_device_should_use_aspm(adev)) return; if (!(adev->flags & AMD_IS_APU) && @@ -699,7 +640,8 @@ static int nv_update_umd_stable_pstate(struct amdgpu_device *adev, adev->gfx.funcs->update_perfmon_mgcg(adev, !enter); if (!(adev->flags & AMD_IS_APU) && - (adev->nbio.funcs->enable_aspm)) + (adev->nbio.funcs->enable_aspm) && + amdgpu_device_should_use_aspm(adev)) adev->nbio.funcs->enable_aspm(adev, !enter); return 0; @@ -708,7 +650,7 @@ static int nv_update_umd_stable_pstate(struct amdgpu_device *adev, static const struct amdgpu_asic_funcs nv_asic_funcs = { .read_disabled_bios = &nv_read_disabled_bios, - .read_bios_from_rom = &nv_read_bios_from_rom, + .read_bios_from_rom = &amdgpu_soc15_read_bios_from_rom, .read_register = &nv_read_register, .reset = &nv_asic_reset, .reset_method = &nv_asic_reset_method, @@ -742,8 +684,8 @@ 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; + adev->pciep_rreg = amdgpu_device_pcie_port_rreg; + adev->pciep_wreg = amdgpu_device_pcie_port_wreg; /* TODO: will add them during VCN v2 implementation */ adev->uvd_ctx_rreg = NULL; @@ -964,10 +906,63 @@ static int nv_common_early_init(void *handle) adev->external_rev_id = adev->rev_id + 0x01; break; case IP_VERSION(10, 1, 3): + case IP_VERSION(10, 1, 4): adev->cg_flags = 0; adev->pg_flags = 0; adev->external_rev_id = adev->rev_id + 0x82; break; + case IP_VERSION(10, 3, 6): + adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | + AMD_CG_SUPPORT_GFX_MGLS | + 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_GFX_RLC_LS | + AMD_CG_SUPPORT_GFX_CP_LS | + AMD_CG_SUPPORT_GFX_FGCG | + AMD_CG_SUPPORT_MC_MGCG | + AMD_CG_SUPPORT_MC_LS | + AMD_CG_SUPPORT_SDMA_LS | + AMD_CG_SUPPORT_HDP_MGCG | + AMD_CG_SUPPORT_HDP_LS | + AMD_CG_SUPPORT_ATHUB_MGCG | + AMD_CG_SUPPORT_ATHUB_LS | + AMD_CG_SUPPORT_IH_CG | + 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; + adev->external_rev_id = adev->rev_id + 0x01; + break; + case IP_VERSION(10, 3, 7): + adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | + AMD_CG_SUPPORT_GFX_MGLS | + 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_GFX_RLC_LS | + AMD_CG_SUPPORT_GFX_CP_LS | + AMD_CG_SUPPORT_GFX_FGCG | + AMD_CG_SUPPORT_MC_MGCG | + AMD_CG_SUPPORT_MC_LS | + AMD_CG_SUPPORT_SDMA_LS | + AMD_CG_SUPPORT_HDP_MGCG | + AMD_CG_SUPPORT_HDP_LS | + AMD_CG_SUPPORT_ATHUB_MGCG | + AMD_CG_SUPPORT_ATHUB_LS | + AMD_CG_SUPPORT_IH_CG | + AMD_CG_SUPPORT_VCN_MGCG | + AMD_CG_SUPPORT_JPEG_MGCG; + adev->pg_flags = AMD_PG_SUPPORT_VCN | + AMD_PG_SUPPORT_VCN_DPG | + AMD_PG_SUPPORT_JPEG | + AMD_PG_SUPPORT_GFX_PG; + adev->external_rev_id = adev->rev_id + 0x01; + break; default: /* FIXME: not supported yet */ return -EINVAL; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h index dd0dce254901..1f276ddd26e9 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h +++ b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h @@ -258,6 +258,7 @@ enum psp_gfx_fw_type { GFX_FW_TYPE_SDMA6 = 56, /* SDMA6 MI */ GFX_FW_TYPE_SDMA7 = 57, /* SDMA7 MI */ GFX_FW_TYPE_VCN1 = 58, /* VCN1 MI */ + GFX_FW_TYPE_CAP = 62, /* CAP_FW */ GFX_FW_TYPE_REG_LIST = 67, /* REG_LIST MI */ GFX_FW_TYPE_MAX }; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index d0e76b36d4ab..9518b4394a6e 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -53,11 +53,13 @@ MODULE_FIRMWARE("amdgpu/navi14_ta.bin"); MODULE_FIRMWARE("amdgpu/navi12_sos.bin"); MODULE_FIRMWARE("amdgpu/navi12_asd.bin"); MODULE_FIRMWARE("amdgpu/navi12_ta.bin"); +MODULE_FIRMWARE("amdgpu/navi12_cap.bin"); MODULE_FIRMWARE("amdgpu/arcturus_sos.bin"); MODULE_FIRMWARE("amdgpu/arcturus_asd.bin"); MODULE_FIRMWARE("amdgpu/arcturus_ta.bin"); MODULE_FIRMWARE("amdgpu/sienna_cichlid_sos.bin"); MODULE_FIRMWARE("amdgpu/sienna_cichlid_ta.bin"); +MODULE_FIRMWARE("amdgpu/sienna_cichlid_cap.bin"); MODULE_FIRMWARE("amdgpu/navy_flounder_sos.bin"); MODULE_FIRMWARE("amdgpu/navy_flounder_ta.bin"); MODULE_FIRMWARE("amdgpu/vangogh_asd.bin"); @@ -177,8 +179,6 @@ static int psp_v11_0_init_microcode(struct psp_context *psp) 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); err = request_firmware(&adev->psp.ta_fw, fw_name, adev->dev); if (err) { diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c index 17655bc6d2f1..024f60631faf 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c @@ -31,9 +31,16 @@ MODULE_FIRMWARE("amdgpu/aldebaran_sos.bin"); MODULE_FIRMWARE("amdgpu/aldebaran_ta.bin"); +MODULE_FIRMWARE("amdgpu/aldebaran_cap.bin"); MODULE_FIRMWARE("amdgpu/yellow_carp_asd.bin"); MODULE_FIRMWARE("amdgpu/yellow_carp_toc.bin"); MODULE_FIRMWARE("amdgpu/yellow_carp_ta.bin"); +MODULE_FIRMWARE("amdgpu/psp_13_0_5_asd.bin"); +MODULE_FIRMWARE("amdgpu/psp_13_0_5_toc.bin"); +MODULE_FIRMWARE("amdgpu/psp_13_0_5_ta.bin"); +MODULE_FIRMWARE("amdgpu/psp_13_0_8_asd.bin"); +MODULE_FIRMWARE("amdgpu/psp_13_0_8_toc.bin"); +MODULE_FIRMWARE("amdgpu/psp_13_0_8_ta.bin"); /* For large FW files the time to complete can be very long */ #define USBC_PD_POLLING_LIMIT_S 240 @@ -55,6 +62,12 @@ static int psp_v13_0_init_microcode(struct psp_context *psp) case IP_VERSION(13, 0, 3): chip_name = "yellow_carp"; break; + case IP_VERSION(13, 0, 5): + chip_name = "psp_13_0_5"; + break; + case IP_VERSION(13, 0, 8): + chip_name = "psp_13_0_8"; + break; default: BUG(); } @@ -69,6 +82,8 @@ static int psp_v13_0_init_microcode(struct psp_context *psp) break; case IP_VERSION(13, 0, 1): case IP_VERSION(13, 0, 3): + case IP_VERSION(13, 0, 5): + case IP_VERSION(13, 0, 8): err = psp_init_asd_microcode(psp, chip_name); if (err) return err; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c index 1ed357cb0f49..01f3bcc62a6c 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c @@ -44,6 +44,7 @@ MODULE_FIRMWARE("amdgpu/vega10_sos.bin"); MODULE_FIRMWARE("amdgpu/vega10_asd.bin"); +MODULE_FIRMWARE("amdgpu/vega10_cap.bin"); MODULE_FIRMWARE("amdgpu/vega12_sos.bin"); MODULE_FIRMWARE("amdgpu/vega12_asd.bin"); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c index 4509bd4cce2d..1d8bbcbd7a37 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c @@ -1142,6 +1142,7 @@ static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = { .align_mask = 0xf, .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), .support_64bit_ptrs = false, + .secure_submission_supported = true, .get_rptr = sdma_v2_4_ring_get_rptr, .get_wptr = sdma_v2_4_ring_get_wptr, .set_wptr = sdma_v2_4_ring_set_wptr, diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index 135727b59c41..4ef4feff5649 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -1580,6 +1580,7 @@ static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = { .align_mask = 0xf, .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), .support_64bit_ptrs = false, + .secure_submission_supported = true, .get_rptr = sdma_v3_0_ring_get_rptr, .get_wptr = sdma_v3_0_ring_get_wptr, .set_wptr = sdma_v3_0_ring_set_wptr, diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index e8e4749e9c79..d7e8f7232364 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -1885,22 +1885,16 @@ static int sdma_v4_0_process_ras_data_cb(struct amdgpu_device *adev, static int sdma_v4_0_late_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct ras_ih_if ih_info = { - .cb = sdma_v4_0_process_ras_data_cb, - }; sdma_v4_0_setup_ulv(adev); if (!amdgpu_persistent_edc_harvesting_supported(adev)) { - if (adev->sdma.funcs && - adev->sdma.funcs->reset_ras_error_count) - adev->sdma.funcs->reset_ras_error_count(adev); + if (adev->sdma.ras && adev->sdma.ras->ras_block.hw_ops && + adev->sdma.ras->ras_block.hw_ops->reset_ras_error_count) + adev->sdma.ras->ras_block.hw_ops->reset_ras_error_count(adev); } - if (adev->sdma.funcs && adev->sdma.funcs->ras_late_init) - return adev->sdma.funcs->ras_late_init(adev, &ih_info); - else - return 0; + return 0; } static int sdma_v4_0_sw_init(void *handle) @@ -2001,9 +1995,6 @@ static int sdma_v4_0_sw_fini(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; int i; - if (adev->sdma.funcs && adev->sdma.funcs->ras_fini) - adev->sdma.funcs->ras_fini(adev); - for (i = 0; i < adev->sdma.num_instances; i++) { amdgpu_ring_fini(&adev->sdma.instance[i].ring); if (adev->sdma.has_page_queue) @@ -2057,6 +2048,10 @@ static int sdma_v4_0_suspend(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + /* SMU saves SDMA state for us */ + if (adev->in_s0ix) + return 0; + return sdma_v4_0_hw_fini(adev); } @@ -2064,6 +2059,10 @@ static int sdma_v4_0_resume(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + /* SMU restores SDMA state for us */ + if (adev->in_s0ix) + return 0; + return sdma_v4_0_hw_init(adev); } @@ -2415,6 +2414,7 @@ static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = { .align_mask = 0xf, .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_0, .get_rptr = sdma_v4_0_ring_get_rptr, .get_wptr = sdma_v4_0_ring_get_wptr, @@ -2451,6 +2451,7 @@ static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs_2nd_mmhub = { .align_mask = 0xf, .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_1, .get_rptr = sdma_v4_0_ring_get_rptr, .get_wptr = sdma_v4_0_ring_get_wptr, @@ -2483,6 +2484,7 @@ static const struct amdgpu_ring_funcs sdma_v4_0_page_ring_funcs = { .align_mask = 0xf, .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_0, .get_rptr = sdma_v4_0_ring_get_rptr, .get_wptr = sdma_v4_0_page_ring_get_wptr, @@ -2515,6 +2517,7 @@ static const struct amdgpu_ring_funcs sdma_v4_0_page_ring_funcs_2nd_mmhub = { .align_mask = 0xf, .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_1, .get_rptr = sdma_v4_0_ring_get_rptr, .get_wptr = sdma_v4_0_page_ring_get_wptr, @@ -2740,7 +2743,7 @@ static void sdma_v4_0_get_ras_error_count(uint32_t value, } } -static int sdma_v4_0_query_ras_error_count(struct amdgpu_device *adev, +static int sdma_v4_0_query_ras_error_count_by_instance(struct amdgpu_device *adev, uint32_t instance, void *ras_error_status) { struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; @@ -2762,6 +2765,18 @@ static int sdma_v4_0_query_ras_error_count(struct amdgpu_device *adev, return 0; }; +static void sdma_v4_0_query_ras_error_count(struct amdgpu_device *adev, void *ras_error_status) +{ + int i = 0; + + for (i = 0; i < adev->sdma.num_instances; i++) { + if (sdma_v4_0_query_ras_error_count_by_instance(adev, i, ras_error_status)) { + dev_err(adev->dev, "Query ras error count failed in SDMA%d\n", i); + return; + } + } +} + static void sdma_v4_0_reset_ras_error_count(struct amdgpu_device *adev) { int i; @@ -2773,26 +2788,48 @@ static void sdma_v4_0_reset_ras_error_count(struct amdgpu_device *adev) } } -static const struct amdgpu_sdma_ras_funcs sdma_v4_0_ras_funcs = { - .ras_late_init = amdgpu_sdma_ras_late_init, - .ras_fini = amdgpu_sdma_ras_fini, +const struct amdgpu_ras_block_hw_ops sdma_v4_0_ras_hw_ops = { .query_ras_error_count = sdma_v4_0_query_ras_error_count, .reset_ras_error_count = sdma_v4_0_reset_ras_error_count, }; +static struct amdgpu_sdma_ras sdma_v4_0_ras = { + .ras_block = { + .hw_ops = &sdma_v4_0_ras_hw_ops, + .ras_cb = sdma_v4_0_process_ras_data_cb, + }, +}; + static void sdma_v4_0_set_ras_funcs(struct amdgpu_device *adev) { switch (adev->ip_versions[SDMA0_HWIP][0]) { case IP_VERSION(4, 2, 0): case IP_VERSION(4, 2, 2): - adev->sdma.funcs = &sdma_v4_0_ras_funcs; + adev->sdma.ras = &sdma_v4_0_ras; break; case IP_VERSION(4, 4, 0): - adev->sdma.funcs = &sdma_v4_4_ras_funcs; + adev->sdma.ras = &sdma_v4_4_ras; break; default: break; } + + if (adev->sdma.ras) { + amdgpu_ras_register_ras_block(adev, &adev->sdma.ras->ras_block); + + strcpy(adev->sdma.ras->ras_block.ras_comm.name, "sdma"); + adev->sdma.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__SDMA; + adev->sdma.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; + adev->sdma.ras_if = &adev->sdma.ras->ras_block.ras_comm; + + /* If don't define special ras_late_init function, use default ras_late_init */ + if (!adev->sdma.ras->ras_block.ras_late_init) + adev->sdma.ras->ras_block.ras_late_init = amdgpu_sdma_ras_late_init; + + /* If not defined special ras_cb function, use default ras_cb */ + if (!adev->sdma.ras->ras_block.ras_cb) + adev->sdma.ras->ras_block.ras_cb = amdgpu_sdma_process_ras_data_cb; + } } const struct amdgpu_ip_block_version sdma_v4_0_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c index bf95007f0843..6f9895cdddb1 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.c @@ -188,7 +188,7 @@ static void sdma_v4_4_get_ras_error_count(struct amdgpu_device *adev, } } -static int sdma_v4_4_query_ras_error_count(struct amdgpu_device *adev, +static int sdma_v4_4_query_ras_error_count_by_instance(struct amdgpu_device *adev, uint32_t instance, void *ras_error_status) { @@ -245,9 +245,26 @@ static void sdma_v4_4_reset_ras_error_count(struct amdgpu_device *adev) } } -const struct amdgpu_sdma_ras_funcs sdma_v4_4_ras_funcs = { - .ras_late_init = amdgpu_sdma_ras_late_init, - .ras_fini = amdgpu_sdma_ras_fini, +static void sdma_v4_4_query_ras_error_count(struct amdgpu_device *adev, void *ras_error_status) +{ + int i = 0; + + for (i = 0; i < adev->sdma.num_instances; i++) { + if (sdma_v4_4_query_ras_error_count_by_instance(adev, i, ras_error_status)) { + dev_err(adev->dev, "Query ras error count failed in SDMA%d\n", i); + return; + } + } + +} + +const struct amdgpu_ras_block_hw_ops sdma_v4_4_ras_hw_ops = { .query_ras_error_count = sdma_v4_4_query_ras_error_count, .reset_ras_error_count = sdma_v4_4_reset_ras_error_count, }; + +struct amdgpu_sdma_ras sdma_v4_4_ras = { + .ras_block = { + .hw_ops = &sdma_v4_4_ras_hw_ops, + }, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h index 74a6e5b5e949..a9f0c68359e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4.h @@ -23,6 +23,6 @@ #ifndef __SDMA_V4_4_H__ #define __SDMA_V4_4_H__ -extern const struct amdgpu_sdma_ras_funcs sdma_v4_4_ras_funcs; +extern struct amdgpu_sdma_ras sdma_v4_4_ras; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c index 81e033549dda..a8d49c005f73 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c @@ -51,9 +51,6 @@ MODULE_FIRMWARE("amdgpu/navi14_sdma1.bin"); MODULE_FIRMWARE("amdgpu/navi12_sdma.bin"); MODULE_FIRMWARE("amdgpu/navi12_sdma1.bin"); -MODULE_FIRMWARE("amdgpu/cyan_skillfish_sdma.bin"); -MODULE_FIRMWARE("amdgpu/cyan_skillfish_sdma1.bin"); - MODULE_FIRMWARE("amdgpu/cyan_skillfish2_sdma.bin"); MODULE_FIRMWARE("amdgpu/cyan_skillfish2_sdma1.bin"); @@ -264,10 +261,7 @@ static int sdma_v5_0_init_microcode(struct amdgpu_device *adev) chip_name = "navi12"; break; case IP_VERSION(5, 0, 1): - if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2) - chip_name = "cyan_skillfish2"; - else - chip_name = "cyan_skillfish"; + chip_name = "cyan_skillfish2"; break; default: BUG(); @@ -1696,6 +1690,7 @@ static const struct amdgpu_ring_funcs sdma_v5_0_ring_funcs = { .align_mask = 0xf, .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_GFXHUB_0, .get_rptr = sdma_v5_0_ring_get_rptr, .get_wptr = sdma_v5_0_ring_get_wptr, diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c index d3d6d5b045b8..824eace69884 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c @@ -51,6 +51,8 @@ MODULE_FIRMWARE("amdgpu/beige_goby_sdma.bin"); MODULE_FIRMWARE("amdgpu/vangogh_sdma.bin"); MODULE_FIRMWARE("amdgpu/yellow_carp_sdma.bin"); +MODULE_FIRMWARE("amdgpu/sdma_5_2_6.bin"); +MODULE_FIRMWARE("amdgpu/sdma_5_2_7.bin"); #define SDMA1_REG_OFFSET 0x600 #define SDMA3_REG_OFFSET 0x400 @@ -138,28 +140,34 @@ static int sdma_v5_2_init_microcode(struct amdgpu_device *adev) switch (adev->ip_versions[SDMA0_HWIP][0]) { case IP_VERSION(5, 2, 0): - chip_name = "sienna_cichlid"; + chip_name = "sienna_cichlid_sdma"; break; case IP_VERSION(5, 2, 2): - chip_name = "navy_flounder"; + chip_name = "navy_flounder_sdma"; break; case IP_VERSION(5, 2, 1): - chip_name = "vangogh"; + chip_name = "vangogh_sdma"; break; case IP_VERSION(5, 2, 4): - chip_name = "dimgrey_cavefish"; + chip_name = "dimgrey_cavefish_sdma"; break; case IP_VERSION(5, 2, 5): - chip_name = "beige_goby"; + chip_name = "beige_goby_sdma"; break; case IP_VERSION(5, 2, 3): - chip_name = "yellow_carp"; + chip_name = "yellow_carp_sdma"; + break; + case IP_VERSION(5, 2, 6): + chip_name = "sdma_5_2_6"; + break; + case IP_VERSION(5, 2, 7): + chip_name = "sdma_5_2_7"; break; default: BUG(); } - snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name); + snprintf(fw_name, sizeof(fw_name), "amdgpu/%s.bin", chip_name); err = request_firmware(&adev->sdma.instance[0].fw, fw_name, adev->dev); if (err) @@ -1617,6 +1625,7 @@ static int sdma_v5_2_set_clockgating_state(void *handle, case IP_VERSION(5, 2, 1): case IP_VERSION(5, 2, 4): case IP_VERSION(5, 2, 5): + case IP_VERSION(5, 2, 6): case IP_VERSION(5, 2, 3): sdma_v5_2_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE); @@ -1644,6 +1653,11 @@ static void sdma_v5_2_get_clockgating_state(void *handle, u32 *flags) if (amdgpu_sriov_vf(adev)) *flags = 0; + /* AMD_CG_SUPPORT_SDMA_MGCG */ + data = RREG32(sdma_v5_2_get_reg_offset(adev, 0, mmSDMA0_CLK_CTRL)); + if (!(data & SDMA0_CLK_CTRL__CGCG_EN_OVERRIDE_MASK)) + *flags |= AMD_CG_SUPPORT_SDMA_MGCG; + /* AMD_CG_SUPPORT_SDMA_LS */ data = RREG32_KIQ(sdma_v5_2_get_reg_offset(adev, 0, mmSDMA0_POWER_CNTL)); if (data & SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK) @@ -1673,6 +1687,7 @@ static const struct amdgpu_ring_funcs sdma_v5_2_ring_funcs = { .align_mask = 0xf, .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_GFXHUB_0, .get_rptr = sdma_v5_2_ring_get_rptr, .get_wptr = sdma_v5_2_ring_get_wptr, diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c index e6d2f74a7976..7f99e130acd0 100644 --- a/drivers/gpu/drm/amd/amdgpu/si.c +++ b/drivers/gpu/drm/amd/amdgpu/si.c @@ -2453,7 +2453,7 @@ static void si_program_aspm(struct amdgpu_device *adev) bool disable_l0s = false, disable_l1 = false, disable_plloff_in_l1 = false; bool disable_clkreq = false; - if (amdgpu_aspm == 0) + if (!amdgpu_device_should_use_aspm(adev)) return; if (adev->flags & AMD_IS_APU) diff --git a/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.c b/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.c index 73ffa8fde3df..dd2d66090d23 100644 --- a/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.c +++ b/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.c @@ -26,6 +26,7 @@ #include "smu_v11_0_i2c.h" #include "amdgpu.h" +#include "amdgpu_dpm.h" #include "soc15_common.h" #include <drm/drm_fixed.h> #include <drm/drm_drv.h> @@ -43,11 +44,10 @@ #define I2C_X_RESTART BIT(31) -#define to_amdgpu_device(x) (container_of(x, struct amdgpu_device, pm.smu_i2c)) - static void smu_v11_0_i2c_set_clock_gating(struct i2c_adapter *control, bool en) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; uint32_t reg = RREG32_SOC15(SMUIO, 0, mmSMUIO_PWRMGT); reg = REG_SET_FIELD(reg, SMUIO_PWRMGT, i2c_clk_gate_en, en ? 1 : 0); @@ -75,7 +75,8 @@ static void smu_v11_0_i2c_set_clock_gating(struct i2c_adapter *control, bool en) static int smu_v11_0_i2c_enable(struct i2c_adapter *control, bool enable) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; WREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_ENABLE, enable ? 1 : 0); @@ -100,7 +101,8 @@ static int smu_v11_0_i2c_enable(struct i2c_adapter *control, bool enable) static void smu_v11_0_i2c_clear_status(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; /* do */ { RREG32_SOC15(SMUIO, 0, mmCKSVII2C_IC_CLR_INTR); @@ -110,7 +112,8 @@ static void smu_v11_0_i2c_clear_status(struct i2c_adapter *control) static void smu_v11_0_i2c_configure(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; uint32_t reg = 0; reg = REG_SET_FIELD(reg, CKSVII2C_IC_CON, IC_SLAVE_DISABLE, 1); @@ -131,7 +134,8 @@ static void smu_v11_0_i2c_configure(struct i2c_adapter *control) static void smu_v11_0_i2c_set_clock(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; /* * Standard mode speed, These values are taken from SMUIO MAS, @@ -154,7 +158,8 @@ static void smu_v11_0_i2c_set_clock(struct i2c_adapter *control) static void smu_v11_0_i2c_set_address(struct i2c_adapter *control, u16 address) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; /* The IC_TAR::IC_TAR field is 10-bits wide. * It takes a 7-bit or 10-bit addresses as an address, @@ -165,7 +170,8 @@ static void smu_v11_0_i2c_set_address(struct i2c_adapter *control, u16 address) static uint32_t smu_v11_0_i2c_poll_tx_status(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; uint32_t ret = I2C_OK; uint32_t reg, reg_c_tx_abrt_source; @@ -216,7 +222,8 @@ static uint32_t smu_v11_0_i2c_poll_tx_status(struct i2c_adapter *control) static uint32_t smu_v11_0_i2c_poll_rx_status(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; uint32_t ret = I2C_OK; uint32_t reg_ic_status, reg_c_tx_abrt_source; @@ -262,7 +269,8 @@ static uint32_t smu_v11_0_i2c_transmit(struct i2c_adapter *control, u16 address, u8 *data, u32 numbytes, u32 i2c_flag) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; u32 bytes_sent, reg, ret = I2C_OK; unsigned long timeout_counter; @@ -360,7 +368,8 @@ static uint32_t smu_v11_0_i2c_receive(struct i2c_adapter *control, u16 address, u8 *data, u32 numbytes, u32 i2c_flag) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; uint32_t bytes_received, ret = I2C_OK; bytes_received = 0; @@ -431,7 +440,8 @@ static uint32_t smu_v11_0_i2c_receive(struct i2c_adapter *control, static void smu_v11_0_i2c_abort(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; uint32_t reg = 0; /* Enable I2C engine; */ @@ -447,7 +457,8 @@ static void smu_v11_0_i2c_abort(struct i2c_adapter *control) static bool smu_v11_0_i2c_activity_done(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; const uint32_t IDLE_TIMEOUT = 1024; uint32_t timeout_count = 0; @@ -508,7 +519,8 @@ static void smu_v11_0_i2c_init(struct i2c_adapter *control) static void smu_v11_0_i2c_fini(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; u32 status, enable, en_stat; int res; @@ -543,7 +555,8 @@ static void smu_v11_0_i2c_fini(struct i2c_adapter *control) static bool smu_v11_0_i2c_bus_lock(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; /* Send PPSMC_MSG_RequestI2CBus */ if (!amdgpu_dpm_smu_i2c_bus_access(adev, true)) @@ -554,7 +567,8 @@ static bool smu_v11_0_i2c_bus_lock(struct i2c_adapter *control) static bool smu_v11_0_i2c_bus_unlock(struct i2c_adapter *control) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(control); + struct amdgpu_device *adev = smu_i2c->adev; /* Send PPSMC_MSG_ReleaseI2CBus */ if (!amdgpu_dpm_smu_i2c_bus_access(adev, false)) @@ -587,16 +601,17 @@ static uint32_t smu_v11_0_i2c_write_data(struct i2c_adapter *control, if (ret != I2C_OK) DRM_ERROR("WriteI2CData() - I2C error occurred :%x", ret); - + return ret; } static void lock_bus(struct i2c_adapter *i2c, unsigned int flags) { - struct amdgpu_device *adev = to_amdgpu_device(i2c); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(i2c); + struct amdgpu_device *adev = smu_i2c->adev; - mutex_lock(&adev->pm.smu_i2c_mutex); + mutex_lock(&smu_i2c->mutex); if (!smu_v11_0_i2c_bus_lock(i2c)) DRM_ERROR("Failed to lock the bus from SMU"); else @@ -611,13 +626,14 @@ static int trylock_bus(struct i2c_adapter *i2c, unsigned int flags) static void unlock_bus(struct i2c_adapter *i2c, unsigned int flags) { - struct amdgpu_device *adev = to_amdgpu_device(i2c); + struct amdgpu_smu_i2c_bus *smu_i2c = i2c_get_adapdata(i2c); + struct amdgpu_device *adev = smu_i2c->adev; if (!smu_v11_0_i2c_bus_unlock(i2c)) DRM_ERROR("Failed to unlock the bus from SMU"); else adev->pm.bus_locked = false; - mutex_unlock(&adev->pm.smu_i2c_mutex); + mutex_unlock(&smu_i2c->mutex); } static const struct i2c_lock_operations smu_v11_0_i2c_i2c_lock_ops = { @@ -706,19 +722,26 @@ static const struct i2c_adapter_quirks smu_v11_0_i2c_control_quirks = { .flags = I2C_AQ_NO_ZERO_LEN, }; -int smu_v11_0_i2c_control_init(struct i2c_adapter *control) +int smu_v11_0_i2c_control_init(struct amdgpu_device *adev) { - struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_smu_i2c_bus *smu_i2c = &adev->pm.smu_i2c[0]; + struct i2c_adapter *control = &smu_i2c->adapter; int res; - mutex_init(&adev->pm.smu_i2c_mutex); + smu_i2c->adev = adev; + smu_i2c->port = 0; + mutex_init(&smu_i2c->mutex); control->owner = THIS_MODULE; control->class = I2C_CLASS_HWMON; control->dev.parent = &adev->pdev->dev; control->algo = &smu_v11_0_i2c_algo; - snprintf(control->name, sizeof(control->name), "AMDGPU SMU"); + snprintf(control->name, sizeof(control->name), "AMDGPU SMU 0"); control->lock_ops = &smu_v11_0_i2c_i2c_lock_ops; control->quirks = &smu_v11_0_i2c_control_quirks; + i2c_set_adapdata(control, smu_i2c); + + adev->pm.ras_eeprom_i2c_bus = &adev->pm.smu_i2c[0].adapter; + adev->pm.fru_eeprom_i2c_bus = &adev->pm.smu_i2c[0].adapter; res = i2c_add_adapter(control); if (res) @@ -727,9 +750,13 @@ int smu_v11_0_i2c_control_init(struct i2c_adapter *control) return res; } -void smu_v11_0_i2c_control_fini(struct i2c_adapter *control) +void smu_v11_0_i2c_control_fini(struct amdgpu_device *adev) { + struct i2c_adapter *control = adev->pm.ras_eeprom_i2c_bus; + i2c_del_adapter(control); + adev->pm.ras_eeprom_i2c_bus = NULL; + adev->pm.fru_eeprom_i2c_bus = NULL; } /* diff --git a/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.h b/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.h index 44467c05f642..96ad14288a0c 100644 --- a/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.h +++ b/drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.h @@ -26,9 +26,9 @@ #include <linux/types.h> -struct i2c_adapter; +struct amdgpu_device; -int smu_v11_0_i2c_control_init(struct i2c_adapter *control); -void smu_v11_0_i2c_control_fini(struct i2c_adapter *control); +int smu_v11_0_i2c_control_init(struct amdgpu_device *adev); +void smu_v11_0_i2c_control_fini(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index 0fc1747e4a70..3d0251ef8d79 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -375,39 +375,6 @@ static bool soc15_read_disabled_bios(struct amdgpu_device *adev) return false; } -static bool soc15_read_bios_from_rom(struct amdgpu_device *adev, - u8 *bios, u32 length_bytes) -{ - u32 *dw_ptr; - u32 i, length_dw; - uint32_t rom_index_offset; - uint32_t rom_data_offset; - - if (bios == NULL) - return false; - if (length_bytes == 0) - return false; - /* APU vbios image is part of sbios image */ - if (adev->flags & AMD_IS_APU) - return false; - - dw_ptr = (u32 *)bios; - length_dw = ALIGN(length_bytes, 4) / 4; - - rom_index_offset = - adev->smuio.funcs->get_rom_index_offset(adev); - rom_data_offset = - adev->smuio.funcs->get_rom_data_offset(adev); - - /* set rom index to 0 */ - WREG32(rom_index_offset, 0); - /* read out the rom data */ - for (i = 0; i < length_dw; i++) - dw_ptr[i] = RREG32(rom_data_offset); - - return true; -} - static struct soc15_allowed_register_entry soc15_allowed_read_registers[] = { { SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS)}, { SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS2)}, @@ -619,8 +586,8 @@ soc15_asic_reset_method(struct amdgpu_device *adev) static int soc15_asic_reset(struct amdgpu_device *adev) { /* original raven doesn't have full asic reset */ - if ((adev->apu_flags & AMD_APU_IS_RAVEN) && - !(adev->apu_flags & AMD_APU_IS_RAVEN2)) + if ((adev->apu_flags & AMD_APU_IS_RAVEN) || + (adev->apu_flags & AMD_APU_IS_RAVEN2)) return 0; switch (soc15_asic_reset_method(adev)) { @@ -703,7 +670,7 @@ static void soc15_pcie_gen3_enable(struct amdgpu_device *adev) static void soc15_program_aspm(struct amdgpu_device *adev) { - if (!amdgpu_aspm) + if (!amdgpu_device_should_use_aspm(adev)) return; if (!(adev->flags & AMD_IS_APU) && @@ -886,6 +853,10 @@ static bool soc15_need_reset_on_init(struct amdgpu_device *adev) { u32 sol_reg; + /* CP hangs in IGT reloading test on RN, reset to WA */ + if (adev->asic_type == CHIP_RENOIR) + return true; + /* Just return false for soc15 GPUs. Reset does not seem to * be necessary. */ @@ -925,7 +896,7 @@ static void soc15_pre_asic_init(struct amdgpu_device *adev) static const struct amdgpu_asic_funcs soc15_asic_funcs = { .read_disabled_bios = &soc15_read_disabled_bios, - .read_bios_from_rom = &soc15_read_bios_from_rom, + .read_bios_from_rom = &amdgpu_soc15_read_bios_from_rom, .read_register = &soc15_read_register, .reset = &soc15_asic_reset, .reset_method = &soc15_asic_reset_method, @@ -947,7 +918,7 @@ static const struct amdgpu_asic_funcs soc15_asic_funcs = static const struct amdgpu_asic_funcs vega20_asic_funcs = { .read_disabled_bios = &soc15_read_disabled_bios, - .read_bios_from_rom = &soc15_read_bios_from_rom, + .read_bios_from_rom = &amdgpu_soc15_read_bios_from_rom, .read_register = &soc15_read_register, .reset = &soc15_asic_reset, .reset_method = &soc15_asic_reset_method, @@ -1114,8 +1085,11 @@ static int soc15_common_early_init(void *handle) AMD_CG_SUPPORT_SDMA_LS | AMD_CG_SUPPORT_VCN_MGCG; + /* + * MMHUB PG needs to be disabled for Picasso for + * stability reasons. + */ adev->pg_flags = AMD_PG_SUPPORT_SDMA | - AMD_PG_SUPPORT_MMHUB | AMD_PG_SUPPORT_VCN; } else { adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | @@ -1219,16 +1193,11 @@ static int soc15_common_early_init(void *handle) static int soc15_common_late_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - int r = 0; if (amdgpu_sriov_vf(adev)) xgpu_ai_mailbox_get_irq(adev); - if (adev->nbio.ras_funcs && - adev->nbio.ras_funcs->ras_late_init) - r = adev->nbio.ras_funcs->ras_late_init(adev); - - return r; + return 0; } static int soc15_common_sw_init(void *handle) @@ -1249,10 +1218,6 @@ static int soc15_common_sw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (adev->nbio.ras_funcs && - adev->nbio.ras_funcs->ras_fini) - adev->nbio.ras_funcs->ras_fini(adev); - if (adev->df.funcs && adev->df.funcs->sw_fini) adev->df.funcs->sw_fini(adev); @@ -1318,11 +1283,11 @@ static int soc15_common_hw_fini(void *handle) if (adev->nbio.ras_if && amdgpu_ras_is_supported(adev, adev->nbio.ras_if->block)) { - if (adev->nbio.ras_funcs && - adev->nbio.ras_funcs->init_ras_controller_interrupt) + if (adev->nbio.ras && + adev->nbio.ras->init_ras_controller_interrupt) amdgpu_irq_put(adev, &adev->nbio.ras_controller_irq, 0); - if (adev->nbio.ras_funcs && - adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt) + if (adev->nbio.ras && + adev->nbio.ras->init_ras_err_event_athub_interrupt) amdgpu_irq_put(adev, &adev->nbio.ras_err_event_athub_irq, 0); } diff --git a/drivers/gpu/drm/amd/amdgpu/soc15_common.h b/drivers/gpu/drm/amd/amdgpu/soc15_common.h index 473767e03676..acce8c2e0328 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15_common.h +++ b/drivers/gpu/drm/amd/amdgpu/soc15_common.h @@ -28,13 +28,13 @@ #define SOC15_REG_OFFSET(ip, inst, reg) (adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) #define __WREG32_SOC15_RLC__(reg, value, flag, hwip) \ - ((amdgpu_sriov_vf(adev) && adev->gfx.rlc.funcs && adev->gfx.rlc.funcs->sriov_wreg) ? \ - adev->gfx.rlc.funcs->sriov_wreg(adev, reg, value, flag, hwip) : \ + ((amdgpu_sriov_vf(adev) && adev->gfx.rlc.funcs && adev->gfx.rlc.rlcg_reg_access_supported) ? \ + amdgpu_sriov_wreg(adev, reg, value, flag, hwip) : \ WREG32(reg, value)) #define __RREG32_SOC15_RLC__(reg, flag, hwip) \ - ((amdgpu_sriov_vf(adev) && adev->gfx.rlc.funcs && adev->gfx.rlc.funcs->sriov_rreg) ? \ - adev->gfx.rlc.funcs->sriov_rreg(adev, reg, flag, hwip) : \ + ((amdgpu_sriov_vf(adev) && adev->gfx.rlc.funcs && adev->gfx.rlc.rlcg_reg_access_supported) ? \ + amdgpu_sriov_rreg(adev, reg, flag, hwip) : \ RREG32(reg)) #define WREG32_FIELD15(ip, idx, reg, field, val) \ diff --git a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h index 5093826a43d1..509d8a1945eb 100644 --- a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h +++ b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h @@ -64,7 +64,8 @@ enum ta_ras_status { TA_RAS_STATUS__ERROR_PCS_STATE_ERROR = 0xA016, TA_RAS_STATUS__ERROR_PCS_STATE_HANG = 0xA017, TA_RAS_STATUS__ERROR_PCS_STATE_UNKNOWN = 0xA018, - TA_RAS_STATUS__ERROR_UNSUPPORTED_ERROR_INJ = 0xA019 + TA_RAS_STATUS__ERROR_UNSUPPORTED_ERROR_INJ = 0xA019, + TA_RAS_STATUS__TEE_ERROR_ACCESS_DENIED = 0xA01A }; enum ta_ras_block { diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c index 20b44983ac94..939cb203f7ad 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c @@ -300,7 +300,6 @@ static void umc_v6_1_query_error_address(struct amdgpu_device *adev, { uint32_t lsb, mc_umc_status_addr; uint64_t mc_umc_status, err_addr, retired_page, mc_umc_addrt0; - struct eeprom_table_record *err_rec; uint32_t channel_index = adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst]; if (adev->asic_type == CHIP_ARCTURUS) { @@ -328,8 +327,6 @@ static void umc_v6_1_query_error_address(struct amdgpu_device *adev, return; } - err_rec = &err_data->err_addr[err_data->err_addr_cnt]; - /* calculate error address if ue/ce error is detected */ if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 && (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 || @@ -348,18 +345,9 @@ static void umc_v6_1_query_error_address(struct amdgpu_device *adev, /* we only save ue error information currently, ce is skipped */ if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) - == 1) { - err_rec->address = err_addr; - /* page frame address is saved */ - err_rec->retired_page = retired_page >> AMDGPU_GPU_PAGE_SHIFT; - err_rec->ts = (uint64_t)ktime_get_real_seconds(); - err_rec->err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; - err_rec->cu = 0; - err_rec->mem_channel = channel_index; - err_rec->mcumc_id = umc_inst; - - err_data->err_addr_cnt++; - } + == 1) + amdgpu_umc_fill_error_record(err_data, err_addr, + retired_page, channel_index, umc_inst); } /* clear umc status */ @@ -465,10 +453,14 @@ static void umc_v6_1_err_cnt_init(struct amdgpu_device *adev) umc_v6_1_enable_umc_index_mode(adev); } -const struct amdgpu_umc_ras_funcs umc_v6_1_ras_funcs = { - .err_cnt_init = umc_v6_1_err_cnt_init, - .ras_late_init = amdgpu_umc_ras_late_init, - .ras_fini = amdgpu_umc_ras_fini, +const struct amdgpu_ras_block_hw_ops umc_v6_1_ras_hw_ops = { .query_ras_error_count = umc_v6_1_query_ras_error_count, .query_ras_error_address = umc_v6_1_query_ras_error_address, }; + +struct amdgpu_umc_ras umc_v6_1_ras = { + .ras_block = { + .hw_ops = &umc_v6_1_ras_hw_ops, + }, + .err_cnt_init = umc_v6_1_err_cnt_init, +};
\ No newline at end of file diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h index 5dc36c730bb2..50c632eb4cc6 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h @@ -45,7 +45,7 @@ /* umc ce count initial value */ #define UMC_V6_1_CE_CNT_INIT (UMC_V6_1_CE_CNT_MAX - UMC_V6_1_CE_INT_THRESHOLD) -extern const struct amdgpu_umc_ras_funcs umc_v6_1_ras_funcs; +extern struct amdgpu_umc_ras umc_v6_1_ras; extern const uint32_t umc_v6_1_channel_idx_tbl[UMC_V6_1_UMC_INSTANCE_NUM][UMC_V6_1_CHANNEL_INSTANCE_NUM]; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c index 6dd1e19e8d43..c45d9c14ecbc 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c @@ -47,6 +47,13 @@ static inline uint32_t get_umc_v6_7_reg_offset(struct amdgpu_device *adev, uint32_t umc_inst, uint32_t ch_inst) { + uint32_t index = umc_inst * adev->umc.channel_inst_num + ch_inst; + + /* adjust umc and channel index offset, + * the register address is not linear on each umc instace */ + umc_inst = index / 4; + ch_inst = index % 4; + return adev->umc.channel_offs * ch_inst + UMC_V6_7_INST_DIST * umc_inst; } @@ -58,49 +65,74 @@ static inline uint32_t get_umc_v6_7_channel_index(struct amdgpu_device *adev, } static void umc_v6_7_ecc_info_query_correctable_error_count(struct amdgpu_device *adev, - uint32_t channel_index, + uint32_t umc_inst, uint32_t ch_inst, unsigned long *error_count) { - uint32_t ecc_err_cnt; uint64_t mc_umc_status; + uint32_t eccinfo_table_idx; struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); - /* - * select the lower chip and check the error count - * skip add error count, calc error counter only from mca_umc_status - */ - ecc_err_cnt = ras->umc_ecc.ecc[channel_index].ce_count_lo_chip; - - /* - * select the higher chip and check the err counter - * skip add error count, calc error counter only from mca_umc_status - */ - ecc_err_cnt = ras->umc_ecc.ecc[channel_index].ce_count_hi_chip; - + eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst; /* check for SRAM correctable error MCUMC_STATUS is a 64 bit register */ - mc_umc_status = ras->umc_ecc.ecc[channel_index].mca_umc_status; + mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status; if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 && REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1) *error_count += 1; } static void umc_v6_7_ecc_info_querry_uncorrectable_error_count(struct amdgpu_device *adev, - uint32_t channel_index, + uint32_t umc_inst, uint32_t ch_inst, unsigned long *error_count) { uint64_t mc_umc_status; + uint32_t eccinfo_table_idx; + uint32_t umc_reg_offset; + uint32_t mc_umc_addr; + uint64_t reg_value; struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + umc_reg_offset = get_umc_v6_7_reg_offset(adev, + umc_inst, ch_inst); + + eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst; /* check the MCUMC_STATUS */ - mc_umc_status = ras->umc_ecc.ecc[channel_index].mca_umc_status; + mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status; if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) && (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 || REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 || REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 || REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 || - REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1)) + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1)) { *error_count += 1; + + if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1) + dev_info(adev->dev, "Deferred error, no user action is needed.\n"); + + if (mc_umc_status) + dev_info(adev->dev, "MCA STATUS 0x%llx, umc_reg_offset 0x%x\n", mc_umc_status, umc_reg_offset); + + /* print IPID registers value */ + mc_umc_addr = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_IPIDT0); + reg_value = RREG64_PCIE((mc_umc_addr + umc_reg_offset) * 4); + if (reg_value) + dev_info(adev->dev, "MCA IPID 0x%llx, umc_reg_offset 0x%x\n", reg_value, umc_reg_offset); + + /* print SYND registers value */ + mc_umc_addr = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_SYNDT0); + reg_value = RREG64_PCIE((mc_umc_addr + umc_reg_offset) * 4); + if (reg_value) + dev_info(adev->dev, "MCA SYND 0x%llx, umc_reg_offset 0x%x\n", reg_value, umc_reg_offset); + + /* print MISC0 registers value */ + mc_umc_addr = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_MISC0T0); + reg_value = RREG64_PCIE((mc_umc_addr + umc_reg_offset) * 4); + if (reg_value) + dev_info(adev->dev, "MCA MISC0 0x%llx, umc_reg_offset 0x%x\n", reg_value, umc_reg_offset); + } } static void umc_v6_7_ecc_info_query_ras_error_count(struct amdgpu_device *adev, @@ -110,42 +142,34 @@ static void umc_v6_7_ecc_info_query_ras_error_count(struct amdgpu_device *adev, uint32_t umc_inst = 0; uint32_t ch_inst = 0; - uint32_t umc_reg_offset = 0; - uint32_t channel_index = 0; /*TODO: driver needs to toggle DF Cstate to ensure * safe access of UMC registers. Will add the protection */ LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { - umc_reg_offset = get_umc_v6_7_reg_offset(adev, - umc_inst, - ch_inst); - channel_index = get_umc_v6_7_channel_index(adev, - umc_inst, - ch_inst); umc_v6_7_ecc_info_query_correctable_error_count(adev, - channel_index, + umc_inst, ch_inst, &(err_data->ce_count)); umc_v6_7_ecc_info_querry_uncorrectable_error_count(adev, - channel_index, + umc_inst, ch_inst, &(err_data->ue_count)); } } static void umc_v6_7_ecc_info_query_error_address(struct amdgpu_device *adev, struct ras_err_data *err_data, - uint32_t umc_reg_offset, uint32_t ch_inst, uint32_t umc_inst) { - uint64_t mc_umc_status, err_addr, retired_page; - struct eeprom_table_record *err_rec; + uint64_t mc_umc_status, err_addr, soc_pa, retired_page, column; uint32_t channel_index; + uint32_t eccinfo_table_idx; struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst; channel_index = adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst]; - mc_umc_status = ras->umc_ecc.ecc[channel_index].mca_umc_status; + mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status; if (mc_umc_status == 0) return; @@ -153,34 +177,41 @@ static void umc_v6_7_ecc_info_query_error_address(struct amdgpu_device *adev, if (!err_data->err_addr) return; - err_rec = &err_data->err_addr[err_data->err_addr_cnt]; - /* calculate error address if ue/ce error is detected */ if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 && (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 || REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)) { - err_addr = ras->umc_ecc.ecc[channel_index].mca_umc_addr; + err_addr = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_addr; err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr); /* translate umc channel address to soc pa, 3 parts are included */ - retired_page = ADDR_OF_8KB_BLOCK(err_addr) | + soc_pa = ADDR_OF_8KB_BLOCK(err_addr) | ADDR_OF_256B_BLOCK(channel_index) | OFFSET_IN_256B_BLOCK(err_addr); + /* The umc channel bits are not original values, they are hashed */ + SET_CHANNEL_HASH(channel_index, soc_pa); + + /* clear [C4 C3 C2] in soc physical address */ + soc_pa &= ~(0x7ULL << UMC_V6_7_PA_C2_BIT); + /* we only save ue error information currently, ce is skipped */ if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1) { - err_rec->address = err_addr; - /* page frame address is saved */ - err_rec->retired_page = retired_page >> AMDGPU_GPU_PAGE_SHIFT; - err_rec->ts = (uint64_t)ktime_get_real_seconds(); - err_rec->err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; - err_rec->cu = 0; - err_rec->mem_channel = channel_index; - err_rec->mcumc_id = umc_inst; - - err_data->err_addr_cnt++; + /* loop for all possibilities of [C4 C3 C2] */ + for (column = 0; column < UMC_V6_7_NA_MAP_PA_NUM; column++) { + retired_page = soc_pa | (column << UMC_V6_7_PA_C2_BIT); + dev_info(adev->dev, "Error Address(PA): 0x%llx\n", retired_page); + amdgpu_umc_fill_error_record(err_data, err_addr, + retired_page, channel_index, umc_inst); + + /* shift R14 bit */ + retired_page ^= (0x1ULL << UMC_V6_7_PA_R14_BIT); + dev_info(adev->dev, "Error Address(PA): 0x%llx\n", retired_page); + amdgpu_umc_fill_error_record(err_data, err_addr, + retired_page, channel_index, umc_inst); + } } } } @@ -192,18 +223,13 @@ static void umc_v6_7_ecc_info_query_ras_error_address(struct amdgpu_device *adev uint32_t umc_inst = 0; uint32_t ch_inst = 0; - uint32_t umc_reg_offset = 0; /*TODO: driver needs to toggle DF Cstate to ensure * safe access of UMC resgisters. Will add the protection * when firmware interface is ready */ LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { - umc_reg_offset = get_umc_v6_7_reg_offset(adev, - umc_inst, - ch_inst); umc_v6_7_ecc_info_query_error_address(adev, err_data, - umc_reg_offset, ch_inst, umc_inst); } @@ -261,6 +287,8 @@ static void umc_v6_7_querry_uncorrectable_error_count(struct amdgpu_device *adev { uint64_t mc_umc_status; uint32_t mc_umc_status_addr; + uint32_t mc_umc_addr; + uint64_t reg_value; mc_umc_status_addr = SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0); @@ -272,8 +300,36 @@ static void umc_v6_7_querry_uncorrectable_error_count(struct amdgpu_device *adev REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 || REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 || REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 || - REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1)) + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1)) { *error_count += 1; + + if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1) + dev_info(adev->dev, "Deferred error, no user action is needed.\n"); + + if (mc_umc_status) + dev_info(adev->dev, "MCA STATUS 0x%llx, umc_reg_offset 0x%x\n", mc_umc_status, umc_reg_offset); + + /* print IPID registers value */ + mc_umc_addr = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_IPIDT0); + reg_value = RREG64_PCIE((mc_umc_addr + umc_reg_offset) * 4); + if (reg_value) + dev_info(adev->dev, "MCA IPID 0x%llx, umc_reg_offset 0x%x\n", reg_value, umc_reg_offset); + + /* print SYND registers value */ + mc_umc_addr = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_SYNDT0); + reg_value = RREG64_PCIE((mc_umc_addr + umc_reg_offset) * 4); + if (reg_value) + dev_info(adev->dev, "MCA SYND 0x%llx, umc_reg_offset 0x%x\n", reg_value, umc_reg_offset); + + /* print MISC0 registers value */ + mc_umc_addr = + SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_MISC0T0); + reg_value = RREG64_PCIE((mc_umc_addr + umc_reg_offset) * 4); + if (reg_value) + dev_info(adev->dev, "MCA MISC0 0x%llx, umc_reg_offset 0x%x\n", reg_value, umc_reg_offset); + } } static void umc_v6_7_reset_error_count_per_channel(struct amdgpu_device *adev, @@ -365,9 +421,9 @@ static void umc_v6_7_query_error_address(struct amdgpu_device *adev, uint32_t umc_inst) { uint32_t mc_umc_status_addr; - uint64_t mc_umc_status, err_addr, retired_page, mc_umc_addrt0; - struct eeprom_table_record *err_rec; uint32_t channel_index; + uint64_t mc_umc_status, mc_umc_addrt0; + uint64_t err_addr, soc_pa, retired_page, column; mc_umc_status_addr = SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0); @@ -385,8 +441,6 @@ static void umc_v6_7_query_error_address(struct amdgpu_device *adev, return; } - err_rec = &err_data->err_addr[err_data->err_addr_cnt]; - channel_index = adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst]; @@ -399,23 +453,32 @@ static void umc_v6_7_query_error_address(struct amdgpu_device *adev, err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr); /* translate umc channel address to soc pa, 3 parts are included */ - retired_page = ADDR_OF_8KB_BLOCK(err_addr) | + soc_pa = ADDR_OF_8KB_BLOCK(err_addr) | ADDR_OF_256B_BLOCK(channel_index) | OFFSET_IN_256B_BLOCK(err_addr); + /* The umc channel bits are not original values, they are hashed */ + SET_CHANNEL_HASH(channel_index, soc_pa); + + /* clear [C4 C3 C2] in soc physical address */ + soc_pa &= ~(0x7ULL << UMC_V6_7_PA_C2_BIT); + /* we only save ue error information currently, ce is skipped */ if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1) { - err_rec->address = err_addr; - /* page frame address is saved */ - err_rec->retired_page = retired_page >> AMDGPU_GPU_PAGE_SHIFT; - err_rec->ts = (uint64_t)ktime_get_real_seconds(); - err_rec->err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; - err_rec->cu = 0; - err_rec->mem_channel = channel_index; - err_rec->mcumc_id = umc_inst; - - err_data->err_addr_cnt++; + /* loop for all possibilities of [C4 C3 C2] */ + for (column = 0; column < UMC_V6_7_NA_MAP_PA_NUM; column++) { + retired_page = soc_pa | (column << UMC_V6_7_PA_C2_BIT); + dev_info(adev->dev, "Error Address(PA): 0x%llx\n", retired_page); + amdgpu_umc_fill_error_record(err_data, err_addr, + retired_page, channel_index, umc_inst); + + /* shift R14 bit */ + retired_page ^= (0x1ULL << UMC_V6_7_PA_R14_BIT); + dev_info(adev->dev, "Error Address(PA): 0x%llx\n", retired_page); + amdgpu_umc_fill_error_record(err_data, err_addr, + retired_page, channel_index, umc_inst); + } } } @@ -463,28 +526,24 @@ static uint32_t umc_v6_7_query_ras_poison_mode_per_channel( static bool umc_v6_7_query_ras_poison_mode(struct amdgpu_device *adev) { - uint32_t umc_inst = 0; - uint32_t ch_inst = 0; uint32_t umc_reg_offset = 0; - LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { - umc_reg_offset = get_umc_v6_7_reg_offset(adev, - umc_inst, - ch_inst); - /* Enabling fatal error in one channel will be considered - as fatal error mode */ - if (umc_v6_7_query_ras_poison_mode_per_channel(adev, umc_reg_offset)) - return false; - } - - return true; + /* Enabling fatal error in umc instance0 channel0 will be + * considered as fatal error mode + */ + umc_reg_offset = get_umc_v6_7_reg_offset(adev, 0, 0); + return !umc_v6_7_query_ras_poison_mode_per_channel(adev, umc_reg_offset); } -const struct amdgpu_umc_ras_funcs umc_v6_7_ras_funcs = { - .ras_late_init = amdgpu_umc_ras_late_init, - .ras_fini = amdgpu_umc_ras_fini, +const struct amdgpu_ras_block_hw_ops umc_v6_7_ras_hw_ops = { .query_ras_error_count = umc_v6_7_query_ras_error_count, .query_ras_error_address = umc_v6_7_query_ras_error_address, +}; + +struct amdgpu_umc_ras umc_v6_7_ras = { + .ras_block = { + .hw_ops = &umc_v6_7_ras_hw_ops, + }, .query_ras_poison_mode = umc_v6_7_query_ras_poison_mode, .ecc_info_query_ras_error_count = umc_v6_7_ecc_info_query_ras_error_count, .ecc_info_query_ras_error_address = umc_v6_7_ecc_info_query_ras_error_address, diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h index 57f2557e7aca..fe41ed2f5945 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h @@ -41,9 +41,32 @@ #define UMC_V6_7_CHANNEL_INSTANCE_NUM 8 /* total channel instances in one umc block */ #define UMC_V6_7_TOTAL_CHANNEL_NUM (UMC_V6_7_CHANNEL_INSTANCE_NUM * UMC_V6_7_UMC_INSTANCE_NUM) +/* one piece of normalizing address is mapped to 8 pieces of physical address */ +#define UMC_V6_7_NA_MAP_PA_NUM 8 +/* R14 bit shift should be considered, double the number */ +#define UMC_V6_7_BAD_PAGE_NUM_PER_CHANNEL (UMC_V6_7_NA_MAP_PA_NUM * 2) +/* The CH4 bit in SOC physical address */ +#define UMC_V6_7_PA_CH4_BIT 12 +/* The C2 bit in SOC physical address */ +#define UMC_V6_7_PA_C2_BIT 17 +/* The R14 bit in SOC physical address */ +#define UMC_V6_7_PA_R14_BIT 34 /* UMC regiser per channel offset */ #define UMC_V6_7_PER_CHANNEL_OFFSET 0x400 -extern const struct amdgpu_umc_ras_funcs umc_v6_7_ras_funcs; + +/* XOR bit 20, 25, 34 of PA into CH4 bit (bit 12 of PA), + * hash bit is only effective when related setting is enabled + */ +#define CHANNEL_HASH(channel_idx, pa) (((channel_idx) >> 4) ^ \ + (((pa) >> 20) & 0x1ULL & adev->df.hash_status.hash_64k) ^ \ + (((pa) >> 25) & 0x1ULL & adev->df.hash_status.hash_2m) ^ \ + (((pa) >> 34) & 0x1ULL & adev->df.hash_status.hash_1g)) +#define SET_CHANNEL_HASH(channel_idx, pa) do { \ + (pa) &= ~(0x1ULL << UMC_V6_7_PA_CH4_BIT); \ + (pa) |= (CHANNEL_HASH(channel_idx, pa) << UMC_V6_7_PA_CH4_BIT); \ + } while (0) + +extern struct amdgpu_umc_ras umc_v6_7_ras; extern const uint32_t umc_v6_7_channel_idx_tbl_second[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM]; extern const uint32_t diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c index af59a35788e3..f35253e0eaa6 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c @@ -40,13 +40,137 @@ const uint32_t {9, 0}, {15, 6} }; -static inline uint32_t get_umc_8_reg_offset(struct amdgpu_device *adev, +static inline uint32_t get_umc_v8_7_reg_offset(struct amdgpu_device *adev, uint32_t umc_inst, uint32_t ch_inst) { return adev->umc.channel_offs*ch_inst + UMC_8_INST_DIST*umc_inst; } +static void umc_v8_7_ecc_info_query_correctable_error_count(struct amdgpu_device *adev, + uint32_t umc_inst, uint32_t ch_inst, + unsigned long *error_count) +{ + uint64_t mc_umc_status; + uint32_t eccinfo_table_idx; + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + + eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst; + + /* check for SRAM correctable error + * MCUMC_STATUS is a 64 bit register + */ + mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status; + if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 && + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1) + *error_count += 1; +} + +static void umc_v8_7_ecc_info_querry_uncorrectable_error_count(struct amdgpu_device *adev, + uint32_t umc_inst, uint32_t ch_inst, + unsigned long *error_count) +{ + uint64_t mc_umc_status; + uint32_t eccinfo_table_idx; + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + + eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst; + + /* check the MCUMC_STATUS */ + mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status; + if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) && + (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC) == 1)) + *error_count += 1; +} + +static void umc_v8_7_ecc_info_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; + + uint32_t umc_inst = 0; + uint32_t ch_inst = 0; + + /* TODO: driver needs to toggle DF Cstate to ensure + * safe access of UMC registers. Will add the protection + */ + LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { + umc_v8_7_ecc_info_query_correctable_error_count(adev, + umc_inst, ch_inst, + &(err_data->ce_count)); + umc_v8_7_ecc_info_querry_uncorrectable_error_count(adev, + umc_inst, ch_inst, + &(err_data->ue_count)); + } +} + +static void umc_v8_7_ecc_info_query_error_address(struct amdgpu_device *adev, + struct ras_err_data *err_data, + uint32_t ch_inst, + uint32_t umc_inst) +{ + uint64_t mc_umc_status, err_addr, retired_page; + uint32_t channel_index; + uint32_t eccinfo_table_idx; + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + + eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst; + channel_index = + adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst]; + + mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status; + + if (mc_umc_status == 0) + return; + + if (!err_data->err_addr) + return; + + /* calculate error address if ue/ce error is detected */ + if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 && + (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 || + REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)) { + + err_addr = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_addr; + err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr); + + /* translate umc channel address to soc pa, 3 parts are included */ + retired_page = ADDR_OF_4KB_BLOCK(err_addr) | + ADDR_OF_256B_BLOCK(channel_index) | + OFFSET_IN_256B_BLOCK(err_addr); + + /* we only save ue error information currently, ce is skipped */ + if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) + == 1) + amdgpu_umc_fill_error_record(err_data, err_addr, + retired_page, channel_index, umc_inst); + } +} + +static void umc_v8_7_ecc_info_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; + + uint32_t umc_inst = 0; + uint32_t ch_inst = 0; + + /* TODO: driver needs to toggle DF Cstate to ensure + * safe access of UMC resgisters. Will add the protection + * when firmware interface is ready + */ + LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { + umc_v8_7_ecc_info_query_error_address(adev, + err_data, + ch_inst, + umc_inst); + } +} + static void umc_v8_7_clear_error_count_per_channel(struct amdgpu_device *adev, uint32_t umc_reg_offset) { @@ -92,7 +216,7 @@ static void umc_v8_7_clear_error_count(struct amdgpu_device *adev) uint32_t umc_reg_offset = 0; LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { - umc_reg_offset = get_umc_8_reg_offset(adev, + umc_reg_offset = get_umc_v8_7_reg_offset(adev, umc_inst, ch_inst); @@ -178,7 +302,7 @@ static void umc_v8_7_query_ras_error_count(struct amdgpu_device *adev, uint32_t umc_reg_offset = 0; LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { - umc_reg_offset = get_umc_8_reg_offset(adev, + umc_reg_offset = get_umc_v8_7_reg_offset(adev, umc_inst, ch_inst); @@ -201,7 +325,6 @@ static void umc_v8_7_query_error_address(struct amdgpu_device *adev, { uint32_t lsb, mc_umc_status_addr; uint64_t mc_umc_status, err_addr, retired_page, mc_umc_addrt0; - struct eeprom_table_record *err_rec; uint32_t channel_index = adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst]; mc_umc_status_addr = @@ -220,8 +343,6 @@ static void umc_v8_7_query_error_address(struct amdgpu_device *adev, return; } - err_rec = &err_data->err_addr[err_data->err_addr_cnt]; - /* calculate error address if ue/ce error is detected */ if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 && (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 || @@ -240,18 +361,9 @@ static void umc_v8_7_query_error_address(struct amdgpu_device *adev, /* we only save ue error information currently, ce is skipped */ if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) - == 1) { - err_rec->address = err_addr; - /* page frame address is saved */ - err_rec->retired_page = retired_page >> AMDGPU_GPU_PAGE_SHIFT; - err_rec->ts = (uint64_t)ktime_get_real_seconds(); - err_rec->err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; - err_rec->cu = 0; - err_rec->mem_channel = channel_index; - err_rec->mcumc_id = umc_inst; - - err_data->err_addr_cnt++; - } + == 1) + amdgpu_umc_fill_error_record(err_data, err_addr, + retired_page, channel_index, umc_inst); } /* clear umc status */ @@ -268,7 +380,7 @@ static void umc_v8_7_query_ras_error_address(struct amdgpu_device *adev, uint32_t umc_reg_offset = 0; LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { - umc_reg_offset = get_umc_8_reg_offset(adev, + umc_reg_offset = get_umc_v8_7_reg_offset(adev, umc_inst, ch_inst); @@ -316,7 +428,7 @@ static void umc_v8_7_err_cnt_init(struct amdgpu_device *adev) uint32_t umc_reg_offset = 0; LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) { - umc_reg_offset = get_umc_8_reg_offset(adev, + umc_reg_offset = get_umc_v8_7_reg_offset(adev, umc_inst, ch_inst); @@ -324,10 +436,16 @@ static void umc_v8_7_err_cnt_init(struct amdgpu_device *adev) } } -const struct amdgpu_umc_ras_funcs umc_v8_7_ras_funcs = { - .err_cnt_init = umc_v8_7_err_cnt_init, - .ras_late_init = amdgpu_umc_ras_late_init, - .ras_fini = amdgpu_umc_ras_fini, +const struct amdgpu_ras_block_hw_ops umc_v8_7_ras_hw_ops = { .query_ras_error_count = umc_v8_7_query_ras_error_count, .query_ras_error_address = umc_v8_7_query_ras_error_address, }; + +struct amdgpu_umc_ras umc_v8_7_ras = { + .ras_block = { + .hw_ops = &umc_v8_7_ras_hw_ops, + }, + .err_cnt_init = umc_v8_7_err_cnt_init, + .ecc_info_query_ras_error_count = umc_v8_7_ecc_info_query_ras_error_count, + .ecc_info_query_ras_error_address = umc_v8_7_ecc_info_query_ras_error_address, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h index 37e6dc7c28e0..dd4993f5f78f 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h @@ -44,7 +44,7 @@ /* umc ce count initial value */ #define UMC_V8_7_CE_CNT_INIT (UMC_V8_7_CE_CNT_MAX - UMC_V8_7_CE_INT_THRESHOLD) -extern const struct amdgpu_umc_ras_funcs umc_v8_7_ras_funcs; +extern struct amdgpu_umc_ras umc_v8_7_ras; extern const uint32_t umc_v8_7_channel_idx_tbl[UMC_V8_7_UMC_INSTANCE_NUM][UMC_V8_7_CHANNEL_INSTANCE_NUM]; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c index b483f03b4591..2f15b8e0f7d7 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c @@ -25,6 +25,7 @@ #include "amdgpu.h" #include "amdgpu_uvd.h" +#include "amdgpu_cs.h" #include "soc15.h" #include "soc15d.h" #include "soc15_common.h" @@ -1275,14 +1276,15 @@ static int uvd_v7_0_ring_test_ring(struct amdgpu_ring *ring) * uvd_v7_0_ring_patch_cs_in_place - Patch the IB for command submission. * * @p: the CS parser with the IBs - * @ib_idx: which IB to patch + * @job: which job this ib is in + * @ib: which IB to patch * */ static int uvd_v7_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p, - uint32_t ib_idx) + struct amdgpu_job *job, + struct amdgpu_ib *ib) { - struct amdgpu_ring *ring = to_amdgpu_ring(p->entity->rq->sched); - struct amdgpu_ib *ib = &p->job->ibs[ib_idx]; + struct amdgpu_ring *ring = to_amdgpu_ring(job->base.sched); unsigned i; /* No patching necessary for the first instance */ @@ -1290,12 +1292,12 @@ static int uvd_v7_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p, return 0; for (i = 0; i < ib->length_dw; i += 2) { - uint32_t reg = amdgpu_get_ib_value(p, ib_idx, i); + uint32_t reg = amdgpu_ib_get_value(ib, i); reg -= p->adev->reg_offset[UVD_HWIP][0][1]; reg += p->adev->reg_offset[UVD_HWIP][1][1]; - amdgpu_set_ib_value(p, ib_idx, i, reg); + amdgpu_ib_set_value(ib, i, reg); } return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c index 3799226defc0..dff54190b96c 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c @@ -148,6 +148,13 @@ static int vcn_v1_0_sw_init(void *handle) adev->vcn.pause_dpg_mode = vcn_v1_0_pause_dpg_mode; + if (amdgpu_vcnfw_log) { + volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr; + + fw_shared->present_flag_0 = 0; + amdgpu_vcn_fwlog_init(adev->vcn.inst); + } + r = jpeg_v1_0_sw_init(handle); return r; @@ -1095,13 +1102,8 @@ static int vcn_v1_0_start_dpg_mode(struct amdgpu_device *adev) static int vcn_v1_0_start(struct amdgpu_device *adev) { - int r; - - if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) - r = vcn_v1_0_start_dpg_mode(adev); - else - r = vcn_v1_0_start_spg_mode(adev); - return r; + return (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ? + vcn_v1_0_start_dpg_mode(adev) : vcn_v1_0_start_spg_mode(adev); } /** @@ -1903,6 +1905,7 @@ static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = { .align_mask = 0xf, .support_64bit_ptrs = false, .no_user_fence = true, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_0, .get_rptr = vcn_v1_0_dec_ring_get_rptr, .get_wptr = vcn_v1_0_dec_ring_get_wptr, diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c index 313fc1b53999..7a7f35e83dd5 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c @@ -172,8 +172,12 @@ static int vcn_v2_0_sw_init(void *handle) if (r) return r; - fw_shared = adev->vcn.inst->fw_shared_cpu_addr; + fw_shared = adev->vcn.inst->fw_shared.cpu_addr; fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG); + + if (amdgpu_vcnfw_log) + amdgpu_vcn_fwlog_init(adev->vcn.inst); + return 0; } @@ -188,7 +192,7 @@ static int vcn_v2_0_sw_fini(void *handle) { int r, idx; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr; + volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr; if (drm_dev_enter(adev_to_drm(adev), &idx)) { fw_shared->present_flag_0 = 0; @@ -364,9 +368,9 @@ static void vcn_v2_0_mc_resume(struct amdgpu_device *adev) /* non-cache window */ WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW, - lower_32_bits(adev->vcn.inst->fw_shared_gpu_addr)); + lower_32_bits(adev->vcn.inst->fw_shared.gpu_addr)); WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH, - upper_32_bits(adev->vcn.inst->fw_shared_gpu_addr)); + upper_32_bits(adev->vcn.inst->fw_shared.gpu_addr)); WREG32_SOC15(UVD, 0, mmUVD_VCPU_NONCACHE_OFFSET0, 0); WREG32_SOC15(UVD, 0, mmUVD_VCPU_NONCACHE_SIZE0, AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared))); @@ -455,10 +459,10 @@ static void vcn_v2_0_mc_resume_dpg_mode(struct amdgpu_device *adev, bool indirec /* non-cache window */ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET( UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW), - lower_32_bits(adev->vcn.inst->fw_shared_gpu_addr), 0, indirect); + lower_32_bits(adev->vcn.inst->fw_shared.gpu_addr), 0, indirect); WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET( UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH), - upper_32_bits(adev->vcn.inst->fw_shared_gpu_addr), 0, indirect); + upper_32_bits(adev->vcn.inst->fw_shared.gpu_addr), 0, indirect); WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET( UVD, 0, mmUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect); WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET( @@ -784,7 +788,7 @@ static void vcn_v2_0_enable_static_power_gating(struct amdgpu_device *adev) static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect) { - volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr; + volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr; struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec; uint32_t rb_bufsz, tmp; @@ -921,7 +925,7 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect) static int vcn_v2_0_start(struct amdgpu_device *adev) { - volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr; + volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr; struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec; uint32_t rb_bufsz, tmp; uint32_t lmi_swap_cntl; @@ -1094,8 +1098,10 @@ static int vcn_v2_0_start(struct amdgpu_device *adev) static int vcn_v2_0_stop_dpg_mode(struct amdgpu_device *adev) { + struct dpg_pause_state state = {.fw_based = VCN_DPG_STATE__UNPAUSE}; uint32_t tmp; + vcn_v2_0_pause_dpg_mode(adev, 0, &state); /* Wait for power status to be 1 */ SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS, 1, UVD_POWER_STATUS__UVD_POWER_STATUS_MASK); @@ -1207,7 +1213,7 @@ static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev, UVD_POWER_STATUS__UVD_POWER_STATUS_MASK); if (!ret_code) { - volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr; + volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr; /* pause DPG */ reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK; WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data); @@ -2003,6 +2009,7 @@ static const struct amd_ip_funcs vcn_v2_0_ip_funcs = { static const struct amdgpu_ring_funcs vcn_v2_0_dec_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_VCN_DEC, .align_mask = 0xf, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_0, .get_rptr = vcn_v2_0_dec_ring_get_rptr, .get_wptr = vcn_v2_0_dec_ring_get_wptr, diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index 44fc4c218433..1bf672966a62 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -196,8 +196,11 @@ static int vcn_v2_5_sw_init(void *handle) return r; } - fw_shared = adev->vcn.inst[j].fw_shared_cpu_addr; + fw_shared = adev->vcn.inst[j].fw_shared.cpu_addr; fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG); + + if (amdgpu_vcnfw_log) + amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]); } if (amdgpu_sriov_vf(adev)) { @@ -229,7 +232,7 @@ static int vcn_v2_5_sw_fini(void *handle) for (i = 0; i < adev->vcn.num_vcn_inst; i++) { if (adev->vcn.harvest_config & (1 << i)) continue; - fw_shared = adev->vcn.inst[i].fw_shared_cpu_addr; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; fw_shared->present_flag_0 = 0; } drm_dev_exit(idx); @@ -423,9 +426,9 @@ static void vcn_v2_5_mc_resume(struct amdgpu_device *adev) /* non-cache window */ WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW, - lower_32_bits(adev->vcn.inst[i].fw_shared_gpu_addr)); + lower_32_bits(adev->vcn.inst[i].fw_shared.gpu_addr)); WREG32_SOC15(VCN, i, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH, - upper_32_bits(adev->vcn.inst[i].fw_shared_gpu_addr)); + upper_32_bits(adev->vcn.inst[i].fw_shared.gpu_addr)); WREG32_SOC15(VCN, i, mmUVD_VCPU_NONCACHE_OFFSET0, 0); WREG32_SOC15(VCN, i, mmUVD_VCPU_NONCACHE_SIZE0, AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared))); @@ -513,10 +516,10 @@ static void vcn_v2_5_mc_resume_dpg_mode(struct amdgpu_device *adev, int inst_idx /* non-cache window */ WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( VCN, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW), - lower_32_bits(adev->vcn.inst[inst_idx].fw_shared_gpu_addr), 0, indirect); + lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect); WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( VCN, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH), - upper_32_bits(adev->vcn.inst[inst_idx].fw_shared_gpu_addr), 0, indirect); + upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect); WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( VCN, 0, mmUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect); WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( @@ -757,7 +760,7 @@ static void vcn_v2_5_enable_clock_gating(struct amdgpu_device *adev) static int vcn_v2_5_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, bool indirect) { - volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared_cpu_addr; + volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr; struct amdgpu_ring *ring; uint32_t rb_bufsz, tmp; @@ -981,7 +984,7 @@ static int vcn_v2_5_start(struct amdgpu_device *adev) vcn_v2_5_mc_resume(adev); for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { - volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[i].fw_shared_cpu_addr; + volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; if (adev->vcn.harvest_config & (1 << i)) continue; /* VCN global tiling registers */ @@ -1403,7 +1406,7 @@ static int vcn_v2_5_pause_dpg_mode(struct amdgpu_device *adev, UVD_POWER_STATUS__UVD_POWER_STATUS_MASK); if (!ret_code) { - volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared_cpu_addr; + volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr; /* pause DPG */ reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK; @@ -1512,6 +1515,7 @@ static void vcn_v2_5_dec_ring_set_wptr(struct amdgpu_ring *ring) static const struct amdgpu_ring_funcs vcn_v2_5_dec_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_VCN_DEC, .align_mask = 0xf, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_1, .get_rptr = vcn_v2_5_dec_ring_get_rptr, .get_wptr = vcn_v2_5_dec_ring_get_wptr, @@ -1542,6 +1546,7 @@ static const struct amdgpu_ring_funcs vcn_v2_5_dec_ring_vm_funcs = { static const struct amdgpu_ring_funcs vcn_v2_6_dec_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_VCN_DEC, .align_mask = 0xf, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_0, .get_rptr = vcn_v2_5_dec_ring_get_rptr, .get_wptr = vcn_v2_5_dec_ring_get_wptr, diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c index da11ceba0698..c87263ed20ec 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c @@ -25,6 +25,7 @@ #include "amdgpu.h" #include "amdgpu_vcn.h" #include "amdgpu_pm.h" +#include "amdgpu_cs.h" #include "soc15.h" #include "soc15d.h" #include "vcn_v2_0.h" @@ -213,11 +214,14 @@ static int vcn_v3_0_sw_init(void *handle) return r; } - fw_shared = adev->vcn.inst[i].fw_shared_cpu_addr; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_SW_RING_FLAG) | cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG) | cpu_to_le32(AMDGPU_VCN_FW_SHARED_FLAG_0_RB); fw_shared->sw_ring.is_enabled = cpu_to_le32(DEC_SW_RING_ENABLED); + + if (amdgpu_vcnfw_log) + amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]); } if (amdgpu_sriov_vf(adev)) { @@ -249,7 +253,7 @@ static int vcn_v3_0_sw_fini(void *handle) if (adev->vcn.harvest_config & (1 << i)) continue; - fw_shared = adev->vcn.inst[i].fw_shared_cpu_addr; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; fw_shared->present_flag_0 = 0; fw_shared->sw_ring.is_enabled = false; } @@ -295,6 +299,7 @@ static int vcn_v3_0_hw_init(void *handle) ring = &adev->vcn.inst[i].ring_dec; if (amdgpu_vcn_is_disabled_vcn(adev, VCN_DECODE_RING, i)) { ring->sched.ready = false; + ring->no_scheduler = true; dev_info(adev->dev, "ring %s is disabled by hypervisor\n", ring->name); } else { ring->wptr = 0; @@ -307,6 +312,7 @@ static int vcn_v3_0_hw_init(void *handle) ring = &adev->vcn.inst[i].ring_enc[j]; if (amdgpu_vcn_is_disabled_vcn(adev, VCN_ENCODE_RING, i)) { ring->sched.ready = false; + ring->no_scheduler = true; dev_info(adev->dev, "ring %s is disabled by hypervisor\n", ring->name); } else { ring->wptr = 0; @@ -469,9 +475,9 @@ static void vcn_v3_0_mc_resume(struct amdgpu_device *adev, int inst) /* non-cache window */ WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW, - lower_32_bits(adev->vcn.inst[inst].fw_shared_gpu_addr)); + lower_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr)); WREG32_SOC15(VCN, inst, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH, - upper_32_bits(adev->vcn.inst[inst].fw_shared_gpu_addr)); + upper_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr)); WREG32_SOC15(VCN, inst, mmUVD_VCPU_NONCACHE_OFFSET0, 0); WREG32_SOC15(VCN, inst, mmUVD_VCPU_NONCACHE_SIZE0, AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared))); @@ -558,10 +564,10 @@ static void vcn_v3_0_mc_resume_dpg_mode(struct amdgpu_device *adev, int inst_idx /* non-cache window */ WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( VCN, inst_idx, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW), - lower_32_bits(adev->vcn.inst[inst_idx].fw_shared_gpu_addr), 0, indirect); + lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect); WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( VCN, inst_idx, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH), - upper_32_bits(adev->vcn.inst[inst_idx].fw_shared_gpu_addr), 0, indirect); + upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect); WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( VCN, inst_idx, mmUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect); WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( @@ -923,7 +929,7 @@ static void vcn_v3_0_enable_clock_gating(struct amdgpu_device *adev, int inst) static int vcn_v3_0_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, bool indirect) { - volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared_cpu_addr; + volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr; struct amdgpu_ring *ring; uint32_t rb_bufsz, tmp; @@ -1220,7 +1226,7 @@ static int vcn_v3_0_start(struct amdgpu_device *adev) tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1); WREG32_SOC15(VCN, i, mmUVD_RBC_RB_CNTL, tmp); - fw_shared = adev->vcn.inst[i].fw_shared_cpu_addr; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; fw_shared->multi_queue.decode_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET); /* programm the RB_BASE for ring buffer */ @@ -1611,7 +1617,7 @@ static int vcn_v3_0_pause_dpg_mode(struct amdgpu_device *adev, if (adev->ip_versions[UVD_HWIP][0] != IP_VERSION(3, 0, 33)) { /* Restore */ - fw_shared = adev->vcn.inst[inst_idx].fw_shared_cpu_addr; + fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr; fw_shared->multi_queue.encode_generalpurpose_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET); ring = &adev->vcn.inst[inst_idx].ring_enc[0]; ring->wptr = 0; @@ -1700,7 +1706,7 @@ static void vcn_v3_0_dec_ring_set_wptr(struct amdgpu_ring *ring) if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) { /*whenever update RBC_RB_WPTR, we save the wptr in shared rb.wptr and scratch2 */ - fw_shared = adev->vcn.inst[ring->me].fw_shared_cpu_addr; + fw_shared = adev->vcn.inst[ring->me].fw_shared.cpu_addr; fw_shared->rb.wptr = lower_32_bits(ring->wptr); WREG32_SOC15(VCN, ring->me, mmUVD_SCRATCH2, lower_32_bits(ring->wptr)); @@ -1780,6 +1786,7 @@ static const struct amdgpu_ring_funcs vcn_v3_0_dec_sw_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_VCN_DEC, .align_mask = 0x3f, .nop = VCN_DEC_SW_CMD_NO_OP, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_0, .get_rptr = vcn_v3_0_dec_ring_get_rptr, .get_wptr = vcn_v3_0_dec_ring_get_wptr, @@ -1806,21 +1813,23 @@ static const struct amdgpu_ring_funcs vcn_v3_0_dec_sw_ring_vm_funcs = { .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, }; -static int vcn_v3_0_limit_sched(struct amdgpu_cs_parser *p) +static int vcn_v3_0_limit_sched(struct amdgpu_cs_parser *p, + struct amdgpu_job *job) { struct drm_gpu_scheduler **scheds; /* The create msg must be in the first IB submitted */ - if (atomic_read(&p->entity->fence_seq)) + if (atomic_read(&job->base.entity->fence_seq)) return -EINVAL; scheds = p->adev->gpu_sched[AMDGPU_HW_IP_VCN_DEC] [AMDGPU_RING_PRIO_DEFAULT].sched; - drm_sched_entity_modify_sched(p->entity, scheds, 1); + drm_sched_entity_modify_sched(job->base.entity, scheds, 1); return 0; } -static int vcn_v3_0_dec_msg(struct amdgpu_cs_parser *p, uint64_t addr) +static int vcn_v3_0_dec_msg(struct amdgpu_cs_parser *p, struct amdgpu_job *job, + uint64_t addr) { struct ttm_operation_ctx ctx = { false, false }; struct amdgpu_bo_va_mapping *map; @@ -1891,7 +1900,7 @@ static int vcn_v3_0_dec_msg(struct amdgpu_cs_parser *p, uint64_t addr) if (create[0] == 0x7 || create[0] == 0x10 || create[0] == 0x11) continue; - r = vcn_v3_0_limit_sched(p); + r = vcn_v3_0_limit_sched(p, job); if (r) goto out; } @@ -1902,10 +1911,10 @@ out: } static int vcn_v3_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p, - uint32_t ib_idx) + struct amdgpu_job *job, + struct amdgpu_ib *ib) { - struct amdgpu_ring *ring = to_amdgpu_ring(p->entity->rq->sched); - struct amdgpu_ib *ib = &p->job->ibs[ib_idx]; + struct amdgpu_ring *ring = to_amdgpu_ring(job->base.sched); uint32_t msg_lo = 0, msg_hi = 0; unsigned i; int r; @@ -1915,8 +1924,8 @@ static int vcn_v3_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p, return 0; for (i = 0; i < ib->length_dw; i += 2) { - uint32_t reg = amdgpu_get_ib_value(p, ib_idx, i); - uint32_t val = amdgpu_get_ib_value(p, ib_idx, i + 1); + uint32_t reg = amdgpu_ib_get_value(ib, i); + uint32_t val = amdgpu_ib_get_value(ib, i + 1); if (reg == PACKET0(p->adev->vcn.internal.data0, 0)) { msg_lo = val; @@ -1924,7 +1933,8 @@ static int vcn_v3_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p, msg_hi = val; } else if (reg == PACKET0(p->adev->vcn.internal.cmd, 0) && val == 0) { - r = vcn_v3_0_dec_msg(p, ((u64)msg_hi) << 32 | msg_lo); + r = vcn_v3_0_dec_msg(p, job, + ((u64)msg_hi) << 32 | msg_lo); if (r) return r; } @@ -1935,6 +1945,7 @@ static int vcn_v3_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p, static const struct amdgpu_ring_funcs vcn_v3_0_dec_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_VCN_DEC, .align_mask = 0xf, + .secure_submission_supported = true, .vmhub = AMDGPU_MMHUB_0, .get_rptr = vcn_v3_0_dec_ring_get_rptr, .get_wptr = vcn_v3_0_dec_ring_get_wptr, diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 6645ebbd2696..039b90cdc3bc 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -1140,7 +1140,7 @@ static void vi_program_aspm(struct amdgpu_device *adev) bool bL1SS = false; bool bClkReqSupport = true; - if (!amdgpu_aspm) + if (!amdgpu_device_should_use_aspm(adev)) return; if (adev->flags & AMD_IS_APU || |