diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
89 files changed, 2117 insertions, 702 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 4009d2e30727..87f095dc385c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -49,6 +49,8 @@ #include <linux/rbtree.h> #include <linux/hashtable.h> #include <linux/dma-fence.h> +#include <linux/pci.h> +#include <linux/aer.h> #include <drm/ttm/ttm_bo_api.h> #include <drm/ttm/ttm_bo_driver.h> @@ -102,6 +104,7 @@ #include "amdgpu_mes.h" #include "amdgpu_umc.h" #include "amdgpu_mmhub.h" +#include "amdgpu_gfxhub.h" #include "amdgpu_df.h" #define MAX_GPU_INSTANCE 16 @@ -879,6 +882,9 @@ struct amdgpu_device { /* mmhub */ struct amdgpu_mmhub mmhub; + /* gfxhub */ + struct amdgpu_gfxhub gfxhub; + /* gfx */ struct amdgpu_gfx gfx; @@ -987,6 +993,9 @@ struct amdgpu_device { atomic_t throttling_logging_enabled; struct ratelimit_state throttling_logging_rs; uint32_t ras_features; + + bool in_pci_err_recovery; + struct pci_saved_state *pci_state; }; static inline struct amdgpu_device *drm_to_adev(struct drm_device *ddev) @@ -1011,18 +1020,32 @@ int amdgpu_gpu_wait_for_idle(struct amdgpu_device *adev); void amdgpu_device_vram_access(struct amdgpu_device *adev, loff_t pos, uint32_t *buf, size_t size, bool write); -uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg, +uint32_t amdgpu_device_rreg(struct amdgpu_device *adev, + uint32_t reg, uint32_t acc_flags); +void amdgpu_device_wreg(struct amdgpu_device *adev, + uint32_t reg, uint32_t v, uint32_t acc_flags); -void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, - uint32_t acc_flags); -void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, uint32_t reg, uint32_t v, - uint32_t acc_flags); +void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, + uint32_t reg, uint32_t v); void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value); uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset); u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg); void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v); +u32 amdgpu_device_indirect_rreg(struct amdgpu_device *adev, + u32 pcie_index, u32 pcie_data, + u32 reg_addr); +u64 amdgpu_device_indirect_rreg64(struct amdgpu_device *adev, + u32 pcie_index, u32 pcie_data, + u32 reg_addr); +void amdgpu_device_indirect_wreg(struct amdgpu_device *adev, + u32 pcie_index, u32 pcie_data, + u32 reg_addr, u32 reg_data); +void amdgpu_device_indirect_wreg64(struct amdgpu_device *adev, + u32 pcie_index, u32 pcie_data, + u32 reg_addr, u64 reg_data); + bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type); bool amdgpu_device_has_dc_support(struct amdgpu_device *adev); @@ -1033,8 +1056,8 @@ int emu_soc_asic_init(struct amdgpu_device *adev); */ #define AMDGPU_REGS_NO_KIQ (1<<1) -#define RREG32_NO_KIQ(reg) amdgpu_mm_rreg(adev, (reg), AMDGPU_REGS_NO_KIQ) -#define WREG32_NO_KIQ(reg, v) amdgpu_mm_wreg(adev, (reg), (v), AMDGPU_REGS_NO_KIQ) +#define RREG32_NO_KIQ(reg) amdgpu_device_rreg(adev, (reg), AMDGPU_REGS_NO_KIQ) +#define WREG32_NO_KIQ(reg, v) amdgpu_device_wreg(adev, (reg), (v), AMDGPU_REGS_NO_KIQ) #define RREG32_KIQ(reg) amdgpu_kiq_rreg(adev, (reg)) #define WREG32_KIQ(reg, v) amdgpu_kiq_wreg(adev, (reg), (v)) @@ -1042,9 +1065,9 @@ int emu_soc_asic_init(struct amdgpu_device *adev); #define RREG8(reg) amdgpu_mm_rreg8(adev, (reg)) #define WREG8(reg, v) amdgpu_mm_wreg8(adev, (reg), (v)) -#define RREG32(reg) amdgpu_mm_rreg(adev, (reg), 0) -#define DREG32(reg) printk(KERN_INFO "REGISTER: " #reg " : 0x%08X\n", amdgpu_mm_rreg(adev, (reg), 0)) -#define WREG32(reg, v) amdgpu_mm_wreg(adev, (reg), (v), 0) +#define RREG32(reg) amdgpu_device_rreg(adev, (reg), 0) +#define DREG32(reg) printk(KERN_INFO "REGISTER: " #reg " : 0x%08X\n", amdgpu_device_rreg(adev, (reg), 0)) +#define WREG32(reg, v) amdgpu_device_wreg(adev, (reg), (v), 0) #define REG_SET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK) #define REG_GET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK) #define RREG32_PCIE(reg) adev->pcie_rreg(adev, (reg)) @@ -1090,7 +1113,7 @@ int emu_soc_asic_init(struct amdgpu_device *adev); WREG32_SMC(_Reg, tmp); \ } while (0) -#define DREG32_SYS(sqf, adev, reg) seq_printf((sqf), #reg " : 0x%08X\n", amdgpu_mm_rreg((adev), (reg), false)) +#define DREG32_SYS(sqf, adev, reg) seq_printf((sqf), #reg " : 0x%08X\n", amdgpu_device_rreg((adev), (reg), false)) #define RREG32_IO(reg) amdgpu_io_rreg(adev, (reg)) #define WREG32_IO(reg, v) amdgpu_io_wreg(adev, (reg), (v)) @@ -1260,6 +1283,15 @@ static inline int amdgpu_dm_display_resume(struct amdgpu_device *adev) { return void amdgpu_register_gpu_instance(struct amdgpu_device *adev); void amdgpu_unregister_gpu_instance(struct amdgpu_device *adev); +pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, + pci_channel_state_t state); +pci_ers_result_t amdgpu_pci_mmio_enabled(struct pci_dev *pdev); +pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev); +void amdgpu_pci_resume(struct pci_dev *pdev); + +bool amdgpu_device_cache_pci_state(struct pci_dev *pdev); +bool amdgpu_device_load_pci_state(struct pci_dev *pdev); + #include "amdgpu_object.h" /* used by df_v3_6.c and amdgpu_pmu.c */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index 4a93b880c6bf..165b02e267b0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -806,8 +806,8 @@ int amdgpu_acpi_init(struct amdgpu_device *adev) } adev->atif = atif; - if (atif->notifications.brightness_change) { #if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE) + if (atif->notifications.brightness_change) { if (amdgpu_device_has_dc_support(adev)) { #if defined(CONFIG_DRM_AMD_DC) struct amdgpu_display_manager *dm = &adev->dm; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index edff1b7f282a..0544460653b9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -36,6 +36,8 @@ */ uint64_t amdgpu_amdkfd_total_mem_size; +static bool kfd_initialized; + int amdgpu_amdkfd_init(void) { struct sysinfo si; @@ -51,19 +53,26 @@ int amdgpu_amdkfd_init(void) #else ret = -ENOENT; #endif + kfd_initialized = !ret; return ret; } void amdgpu_amdkfd_fini(void) { - kgd2kfd_exit(); + if (kfd_initialized) { + kgd2kfd_exit(); + kfd_initialized = false; + } } void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev) { bool vf = amdgpu_sriov_vf(adev); + if (!kfd_initialized) + return; + adev->kfd.dev = kgd2kfd_probe((struct kgd_dev *)adev, adev->pdev, adev->asic_type, vf); @@ -572,6 +581,13 @@ uint32_t amdgpu_amdkfd_get_asic_rev_id(struct kgd_dev *kgd) return adev->rev_id; } +int amdgpu_amdkfd_get_noretry(struct kgd_dev *kgd) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)kgd; + + return adev->gmc.noretry; +} + int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine, uint32_t vmid, uint64_t gpu_addr, uint32_t *ib_cmd, uint32_t ib_len) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index a10507ecb750..ea391ca7f2f1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -181,6 +181,7 @@ uint64_t amdgpu_amdkfd_get_unique_id(struct kgd_dev *kgd); uint64_t amdgpu_amdkfd_get_mmio_remap_phys_addr(struct kgd_dev *kgd); uint32_t amdgpu_amdkfd_get_num_gws(struct kgd_dev *kgd); uint32_t amdgpu_amdkfd_get_asic_rev_id(struct kgd_dev *kgd); +int amdgpu_amdkfd_get_noretry(struct kgd_dev *kgd); uint8_t amdgpu_amdkfd_get_xgmi_hops_count(struct kgd_dev *dst, struct kgd_dev *src); /* Read user wptr from a specified user address space with page fault @@ -207,11 +208,11 @@ uint8_t amdgpu_amdkfd_get_xgmi_hops_count(struct kgd_dev *dst, struct kgd_dev *s }) /* GPUVM API */ -int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, unsigned int pasid, +int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, u32 pasid, void **vm, void **process_info, struct dma_fence **ef); int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct kgd_dev *kgd, - struct file *filp, unsigned int pasid, + struct file *filp, u32 pasid, void **vm, void **process_info, struct dma_fence **ef); void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev, 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 df0aab0fc67e..4763bab7a4d0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c @@ -32,7 +32,6 @@ #include "v10_structs.h" #include "nv.h" #include "nvd.h" -#include "gfxhub_v2_0.h" enum hqd_dequeue_request_type { NO_ACTION = 0, @@ -105,7 +104,7 @@ static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid, unlock_srbm(kgd); } -static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid, +static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, u32 pasid, unsigned int vmid) { struct amdgpu_device *adev = get_amdgpu_device(kgd); @@ -753,7 +752,7 @@ static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid, } /* SDMA is on gfxhub as well for Navi1* series */ - gfxhub_v2_0_setup_vm_pt_regs(adev, vmid, page_table_base); + adev->gfxhub.funcs->setup_vm_pt_regs(adev, vmid, page_table_base); } const struct kfd2kgd_calls gfx_v10_kfd2kgd = { 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 5b38f848b772..50016bf9c427 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 @@ -31,7 +31,6 @@ #include "v10_structs.h" #include "nv.h" #include "nvd.h" -#include "gfxhub_v2_1.h" enum hqd_dequeue_request_type { NO_ACTION = 0, @@ -657,7 +656,7 @@ static void set_vm_context_page_table_base_v10_3(struct kgd_dev *kgd, uint32_t v struct amdgpu_device *adev = get_amdgpu_device(kgd); /* SDMA is on gfxhub as well for Navi1* series */ - gfxhub_v2_1_setup_vm_pt_regs(adev, vmid, page_table_base); + adev->gfxhub.funcs->setup_vm_pt_regs(adev, vmid, page_table_base); } #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 275f20399373..b91d27e39bad 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c @@ -139,7 +139,7 @@ static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid, unlock_srbm(kgd); } -static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid, +static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, u32 pasid, unsigned int vmid) { struct amdgpu_device *adev = get_amdgpu_device(kgd); 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 4997189d8b36..5ce0ce704a21 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c @@ -96,7 +96,7 @@ static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid, unlock_srbm(kgd); } -static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid, +static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, u32 pasid, unsigned int vmid) { struct amdgpu_device *adev = get_amdgpu_device(kgd); 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 e6aede725197..43b18863a8b8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c @@ -36,9 +36,7 @@ #include "v9_structs.h" #include "soc15.h" #include "soc15d.h" -#include "mmhub_v1_0.h" -#include "gfxhub_v1_0.h" - +#include "gfx_v9_0.h" enum hqd_dequeue_request_type { NO_ACTION = 0, @@ -110,7 +108,7 @@ void kgd_gfx_v9_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid, unlock_srbm(kgd); } -int kgd_gfx_v9_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid, +int kgd_gfx_v9_set_pasid_vmid_mapping(struct kgd_dev *kgd, u32 pasid, unsigned int vmid) { struct amdgpu_device *adev = get_amdgpu_device(kgd); @@ -703,7 +701,180 @@ void kgd_gfx_v9_set_vm_context_page_table_base(struct kgd_dev *kgd, adev->mmhub.funcs->setup_vm_pt_regs(adev, vmid, page_table_base); - gfxhub_v1_0_setup_vm_pt_regs(adev, vmid, page_table_base); + adev->gfxhub.funcs->setup_vm_pt_regs(adev, vmid, page_table_base); +} + +static void lock_spi_csq_mutexes(struct amdgpu_device *adev) +{ + mutex_lock(&adev->srbm_mutex); + mutex_lock(&adev->grbm_idx_mutex); + +} + +static void unlock_spi_csq_mutexes(struct amdgpu_device *adev) +{ + mutex_unlock(&adev->grbm_idx_mutex); + mutex_unlock(&adev->srbm_mutex); +} + +/** + * @get_wave_count: Read device registers to get number of waves in flight for + * a particular queue. The method also returns the VMID associated with the + * queue. + * + * @adev: Handle of device whose registers are to be read + * @queue_idx: Index of queue in the queue-map bit-field + * @wave_cnt: Output parameter updated with number of waves in flight + * @vmid: Output parameter updated with VMID of queue whose wave count + * is being collected + */ +static void get_wave_count(struct amdgpu_device *adev, int queue_idx, + int *wave_cnt, int *vmid) +{ + int pipe_idx; + int queue_slot; + unsigned int reg_val; + + /* + * Program GRBM with appropriate MEID, PIPEID, QUEUEID and VMID + * parameters to read out waves in flight. Get VMID if there are + * non-zero waves in flight. + */ + *vmid = 0xFF; + *wave_cnt = 0; + pipe_idx = queue_idx / adev->gfx.mec.num_queue_per_pipe; + queue_slot = queue_idx % adev->gfx.mec.num_queue_per_pipe; + soc15_grbm_select(adev, 1, pipe_idx, queue_slot, 0); + reg_val = RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_CSQ_WF_ACTIVE_COUNT_0) + + queue_slot); + *wave_cnt = reg_val & SPI_CSQ_WF_ACTIVE_COUNT_0__COUNT_MASK; + if (*wave_cnt != 0) + *vmid = (RREG32_SOC15(GC, 0, mmCP_HQD_VMID) & + CP_HQD_VMID__VMID_MASK) >> CP_HQD_VMID__VMID__SHIFT; +} + +/** + * @kgd_gfx_v9_get_cu_occupancy: Reads relevant registers associated with each + * shader engine and aggregates the number of waves that are in flight for the + * 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 + * @pasid: Identifies the process for which this query call is invoked + * @wave_cnt: Output parameter updated with number of waves in flight that + * belong to process with given pasid + * @max_waves_per_cu: Output parameter updated with maximum number of waves + * possible per Compute Unit + * + * @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: + * 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 + * lost or added to total wave count. + * + * The registers that provide the waves in flight are: + * + * SPI_CSQ_WF_ACTIVE_STATUS - bit-map of queues per pipe. The bit is ON if a + * queue is slotted, OFF if there is no queue. A process could have ZERO or + * more queues slotted and submitting waves to be run on compute units. Even + * when there is a queue it is possible there could be zero wave fronts, this + * can happen when queue is waiting on top-of-pipe events - e.g. waitRegMem + * command + * + * For each bit that is ON from above: + * + * Read (SPI_CSQ_WF_ACTIVE_COUNT_0 + queue_idx) register. It provides the + * number of waves that are in flight for the queue at specified index. The + * index ranges from 0 to 7. + * + * If non-zero waves are in flight, read CP_HQD_VMID register to obtain VMID + * of the wave(s). + * + * Determine if VMID from above step maps to pasid provided as parameter. If + * it matches agrregate the wave count. That the VMID will not match pasid is + * a normal condition i.e. a device is expected to support multiple queues + * from multiple proceses. + * + * Reading registers referenced above involves programming GRBM appropriately + */ +static void kgd_gfx_v9_get_cu_occupancy(struct kgd_dev *kgd, int pasid, + int *pasid_wave_cnt, int *max_waves_per_cu) +{ + int qidx; + int vmid; + int se_idx; + int sh_idx; + int se_cnt; + int sh_cnt; + int wave_cnt; + int queue_map; + int pasid_tmp; + int max_queue_cnt; + int vmid_wave_cnt = 0; + struct amdgpu_device *adev; + DECLARE_BITMAP(cp_queue_bitmap, KGD_MAX_QUEUES); + + adev = get_amdgpu_device(kgd); + lock_spi_csq_mutexes(adev); + soc15_grbm_select(adev, 1, 0, 0, 0); + + /* + * Iterate through the shader engines and arrays of the device + * to get number of waves in flight + */ + bitmap_complement(cp_queue_bitmap, adev->gfx.mec.queue_bitmap, + KGD_MAX_QUEUES); + max_queue_cnt = adev->gfx.mec.num_pipe_per_mec * + adev->gfx.mec.num_queue_per_pipe; + sh_cnt = adev->gfx.config.max_sh_per_se; + se_cnt = adev->gfx.config.max_shader_engines; + for (se_idx = 0; se_idx < se_cnt; se_idx++) { + for (sh_idx = 0; sh_idx < sh_cnt; sh_idx++) { + + gfx_v9_0_select_se_sh(adev, se_idx, sh_idx, 0xffffffff); + queue_map = RREG32(SOC15_REG_OFFSET(GC, 0, + mmSPI_CSQ_WF_ACTIVE_STATUS)); + + /* + * Assumption: queue map encodes following schema: four + * pipes per each micro-engine, with each pipe mapping + * eight queues. This schema is true for GFX9 devices + * and must be verified for newer device families + */ + for (qidx = 0; qidx < max_queue_cnt; qidx++) { + + /* Skip qeueus that are not associated with + * compute functions + */ + if (!test_bit(qidx, cp_queue_bitmap)) + continue; + + if (!(queue_map & (1 << qidx))) + continue; + + /* Get number of waves in flight and aggregate them */ + get_wave_count(adev, qidx, &wave_cnt, &vmid); + if (wave_cnt != 0) { + pasid_tmp = + RREG32(SOC15_REG_OFFSET(OSSSYS, 0, + mmIH_VMID_0_LUT) + vmid); + if (pasid_tmp == pasid) + vmid_wave_cnt += wave_cnt; + } + } + } + } + + gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + soc15_grbm_select(adev, 0, 0, 0, 0); + unlock_spi_csq_mutexes(adev); + + /* Update the output parameters and return */ + *pasid_wave_cnt = vmid_wave_cnt; + *max_waves_per_cu = adev->gfx.cu_info.simd_per_cu * + adev->gfx.cu_info.max_waves_per_simd; } const struct kfd2kgd_calls gfx_v9_kfd2kgd = { @@ -726,4 +897,5 @@ const struct kfd2kgd_calls gfx_v9_kfd2kgd = { .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, + .get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy, }; 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 f098e88d3a0d..fc8934b86d93 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h @@ -26,7 +26,7 @@ void kgd_gfx_v9_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid, uint32_t sh_mem_config, uint32_t sh_mem_ape1_base, uint32_t sh_mem_ape1_limit, uint32_t sh_mem_bases); -int kgd_gfx_v9_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid, +int kgd_gfx_v9_set_pasid_vmid_mapping(struct kgd_dev *kgd, u32 pasid, unsigned int vmid); int kgd_gfx_v9_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id); int kgd_gfx_v9_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 64d4b5ff95d6..054a1c2d5054 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -996,7 +996,7 @@ create_evict_fence_fail: return ret; } -int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, unsigned int pasid, +int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, u32 pasid, void **vm, void **process_info, struct dma_fence **ef) { @@ -1032,7 +1032,7 @@ amdgpu_vm_init_fail: } int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct kgd_dev *kgd, - struct file *filp, unsigned int pasid, + struct file *filp, u32 pasid, void **vm, void **process_info, struct dma_fence **ef) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c index 17c010d0431f..b4df6460e45a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c @@ -543,6 +543,7 @@ int amdgpu_mem_train_support(struct amdgpu_device *adev) case HW_REV(11, 0, 0): case HW_REV(11, 0, 5): case HW_REV(11, 0, 7): + case HW_REV(11, 0, 11): ret = 1; break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c index 3e35a8f2c5e5..7abe9500c0c6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c @@ -616,7 +616,7 @@ static bool amdgpu_atpx_detect(void) while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) { vga_count++; - has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true); + has_atpx |= amdgpu_atpx_pci_probe_handle(pdev); parent_pdev = pci_upstream_bridge(pdev); d3_supported |= parent_pdev && parent_pdev->bridge_d3; @@ -626,7 +626,7 @@ static bool amdgpu_atpx_detect(void) while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_OTHER << 8, pdev)) != NULL) { vga_count++; - has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true); + has_atpx |= amdgpu_atpx_pci_probe_handle(pdev); parent_pdev = pci_upstream_bridge(pdev); d3_supported |= parent_pdev && parent_pdev->bridge_d3; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c index a5e08dc54e4a..065937482239 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c @@ -267,7 +267,7 @@ static int amdgpu_debugfs_process_reg_op(bool read, struct file *f, } else { r = get_user(value, (uint32_t *)buf); if (!r) - amdgpu_mm_wreg_mmio_rlc(adev, *pos >> 2, value, 0); + amdgpu_mm_wreg_mmio_rlc(adev, *pos >> 2, value); } if (r) { result = r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index f7307af76452..37da3537ba2e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -80,8 +80,6 @@ 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/sienna_cichlid_gpu_info.bin"); -MODULE_FIRMWARE("amdgpu/navy_flounder_gpu_info.bin"); #define AMDGPU_RESUME_MS 2000 @@ -303,10 +301,10 @@ void amdgpu_device_vram_access(struct amdgpu_device *adev, loff_t pos, } /* - * MMIO register access helper functions. + * register access helper functions. */ /** - * amdgpu_mm_rreg - read a memory mapped IO register + * amdgpu_device_rreg - read a memory mapped IO or indirect register * * @adev: amdgpu_device pointer * @reg: dword aligned register offset @@ -314,30 +312,29 @@ void amdgpu_device_vram_access(struct amdgpu_device *adev, loff_t pos, * * Returns the 32 bit value from the offset specified. */ -uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg, - uint32_t acc_flags) +uint32_t amdgpu_device_rreg(struct amdgpu_device *adev, + uint32_t reg, uint32_t acc_flags) { uint32_t ret; - if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev) && - down_read_trylock(&adev->reset_sem)) { - ret = amdgpu_kiq_rreg(adev, reg); - up_read(&adev->reset_sem); - return ret; - } - - if ((reg * 4) < adev->rmmio_size) - ret = readl(((void __iomem *)adev->rmmio) + (reg * 4)); - else { - unsigned long flags; + if (adev->in_pci_err_recovery) + return 0; - spin_lock_irqsave(&adev->mmio_idx_lock, flags); - writel((reg * 4), ((void __iomem *)adev->rmmio) + (mmMM_INDEX * 4)); - ret = readl(((void __iomem *)adev->rmmio) + (mmMM_DATA * 4)); - spin_unlock_irqrestore(&adev->mmio_idx_lock, flags); + if ((reg * 4) < adev->rmmio_size) { + if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && + amdgpu_sriov_runtime(adev) && + down_read_trylock(&adev->reset_sem)) { + ret = amdgpu_kiq_rreg(adev, reg); + up_read(&adev->reset_sem); + } else { + ret = readl(((void __iomem *)adev->rmmio) + (reg * 4)); + } + } else { + ret = adev->pcie_rreg(adev, reg * 4); } - trace_amdgpu_mm_rreg(adev->pdev->device, reg, ret); + trace_amdgpu_device_rreg(adev->pdev->device, reg, ret); + return ret; } @@ -355,7 +352,11 @@ uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg, * * Returns the 8 bit value from the offset specified. */ -uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset) { +uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset) +{ + if (adev->in_pci_err_recovery) + return 0; + if (offset < adev->rmmio_size) return (readb(adev->rmmio + offset)); BUG(); @@ -376,33 +377,19 @@ uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset) { * * Writes the value specified to the offset specified. */ -void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value) { +void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value) +{ + if (adev->in_pci_err_recovery) + return; + if (offset < adev->rmmio_size) writeb(value, adev->rmmio + offset); else BUG(); } -static inline void amdgpu_mm_wreg_mmio(struct amdgpu_device *adev, - uint32_t reg, uint32_t v, - uint32_t acc_flags) -{ - trace_amdgpu_mm_wreg(adev->pdev->device, reg, v); - - if ((reg * 4) < adev->rmmio_size) - writel(v, ((void __iomem *)adev->rmmio) + (reg * 4)); - else { - unsigned long flags; - - spin_lock_irqsave(&adev->mmio_idx_lock, flags); - writel((reg * 4), ((void __iomem *)adev->rmmio) + (mmMM_INDEX * 4)); - writel(v, ((void __iomem *)adev->rmmio) + (mmMM_DATA * 4)); - spin_unlock_irqrestore(&adev->mmio_idx_lock, flags); - } -} - /** - * amdgpu_mm_wreg - write to a memory mapped IO register + * amdgpu_device_wreg - write to a memory mapped IO or indirect register * * @adev: amdgpu_device pointer * @reg: dword aligned register offset @@ -411,17 +398,27 @@ static inline void amdgpu_mm_wreg_mmio(struct amdgpu_device *adev, * * Writes the value specified to the offset specified. */ -void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, - uint32_t acc_flags) +void amdgpu_device_wreg(struct amdgpu_device *adev, + uint32_t reg, uint32_t v, + uint32_t acc_flags) { - if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev) && - down_read_trylock(&adev->reset_sem)) { - amdgpu_kiq_wreg(adev, reg, v); - up_read(&adev->reset_sem); + if (adev->in_pci_err_recovery) return; + + if ((reg * 4) < adev->rmmio_size) { + if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && + amdgpu_sriov_runtime(adev) && + down_read_trylock(&adev->reset_sem)) { + amdgpu_kiq_wreg(adev, reg, v); + up_read(&adev->reset_sem); + } else { + writel(v, ((void __iomem *)adev->rmmio) + (reg * 4)); + } + } else { + adev->pcie_wreg(adev, reg * 4, v); } - amdgpu_mm_wreg_mmio(adev, reg, v, acc_flags); + trace_amdgpu_device_wreg(adev->pdev->device, reg, v); } /* @@ -429,18 +426,20 @@ void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, * * this function is invoked only the debugfs register access * */ -void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, uint32_t reg, uint32_t v, - uint32_t acc_flags) +void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, + uint32_t reg, uint32_t v) { - if (amdgpu_sriov_fullaccess(adev) && - adev->gfx.rlc.funcs && - adev->gfx.rlc.funcs->is_rlcg_access_range) { + if (adev->in_pci_err_recovery) + return; + if (amdgpu_sriov_fullaccess(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->rlcg_wreg(adev, reg, v); + } else { + writel(v, ((void __iomem *)adev->rmmio) + (reg * 4)); } - - amdgpu_mm_wreg_mmio(adev, reg, v, acc_flags); } /** @@ -453,6 +452,9 @@ void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, uint32_t reg, uint32_t */ u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg) { + if (adev->in_pci_err_recovery) + return 0; + if ((reg * 4) < adev->rio_mem_size) return ioread32(adev->rio_mem + (reg * 4)); else { @@ -472,6 +474,9 @@ u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg) */ void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v) { + if (adev->in_pci_err_recovery) + return; + if ((reg * 4) < adev->rio_mem_size) iowrite32(v, adev->rio_mem + (reg * 4)); else { @@ -491,6 +496,9 @@ void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v) */ u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index) { + if (adev->in_pci_err_recovery) + return 0; + if (index < adev->doorbell.num_doorbells) { return readl(adev->doorbell.ptr + index); } else { @@ -511,6 +519,9 @@ u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index) */ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v) { + if (adev->in_pci_err_recovery) + return; + if (index < adev->doorbell.num_doorbells) { writel(v, adev->doorbell.ptr + index); } else { @@ -529,6 +540,9 @@ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v) */ u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index) { + if (adev->in_pci_err_recovery) + return 0; + if (index < adev->doorbell.num_doorbells) { return atomic64_read((atomic64_t *)(adev->doorbell.ptr + index)); } else { @@ -549,6 +563,9 @@ u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index) */ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v) { + if (adev->in_pci_err_recovery) + return; + if (index < adev->doorbell.num_doorbells) { atomic64_set((atomic64_t *)(adev->doorbell.ptr + index), v); } else { @@ -557,6 +574,135 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v) } /** + * amdgpu_device_indirect_rreg - read an indirect register + * + * @adev: amdgpu_device pointer + * @pcie_index: mmio register offset + * @pcie_data: mmio register offset + * + * Returns the value of indirect register @reg_addr + */ +u32 amdgpu_device_indirect_rreg(struct amdgpu_device *adev, + u32 pcie_index, u32 pcie_data, + u32 reg_addr) +{ + unsigned long flags; + u32 r; + void __iomem *pcie_index_offset; + void __iomem *pcie_data_offset; + + spin_lock_irqsave(&adev->pcie_idx_lock, flags); + pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4; + pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4; + + writel(reg_addr, pcie_index_offset); + readl(pcie_index_offset); + r = readl(pcie_data_offset); + spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); + + return r; +} + +/** + * amdgpu_device_indirect_rreg64 - read a 64bits indirect register + * + * @adev: amdgpu_device pointer + * @pcie_index: mmio register offset + * @pcie_data: mmio register offset + * + * Returns the value of indirect register @reg_addr + */ +u64 amdgpu_device_indirect_rreg64(struct amdgpu_device *adev, + u32 pcie_index, u32 pcie_data, + u32 reg_addr) +{ + unsigned long flags; + u64 r; + void __iomem *pcie_index_offset; + void __iomem *pcie_data_offset; + + spin_lock_irqsave(&adev->pcie_idx_lock, flags); + pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4; + pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4; + + /* read low 32 bits */ + writel(reg_addr, pcie_index_offset); + readl(pcie_index_offset); + r = readl(pcie_data_offset); + /* read high 32 bits */ + writel(reg_addr + 4, pcie_index_offset); + readl(pcie_index_offset); + r |= ((u64)readl(pcie_data_offset) << 32); + spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); + + return r; +} + +/** + * amdgpu_device_indirect_wreg - write an indirect register address + * + * @adev: amdgpu_device pointer + * @pcie_index: mmio register offset + * @pcie_data: mmio register offset + * @reg_addr: indirect register offset + * @reg_data: indirect register data + * + */ +void amdgpu_device_indirect_wreg(struct amdgpu_device *adev, + u32 pcie_index, u32 pcie_data, + u32 reg_addr, u32 reg_data) +{ + unsigned long flags; + void __iomem *pcie_index_offset; + void __iomem *pcie_data_offset; + + spin_lock_irqsave(&adev->pcie_idx_lock, flags); + pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4; + pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4; + + writel(reg_addr, pcie_index_offset); + readl(pcie_index_offset); + writel(reg_data, pcie_data_offset); + readl(pcie_data_offset); + spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); +} + +/** + * amdgpu_device_indirect_wreg64 - write a 64bits indirect register address + * + * @adev: amdgpu_device pointer + * @pcie_index: mmio register offset + * @pcie_data: mmio register offset + * @reg_addr: indirect register offset + * @reg_data: indirect register data + * + */ +void amdgpu_device_indirect_wreg64(struct amdgpu_device *adev, + u32 pcie_index, u32 pcie_data, + u32 reg_addr, u64 reg_data) +{ + unsigned long flags; + void __iomem *pcie_index_offset; + void __iomem *pcie_data_offset; + + spin_lock_irqsave(&adev->pcie_idx_lock, flags); + pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4; + pcie_data_offset = (void __iomem *)adev->rmmio + pcie_data * 4; + + /* write low 32 bits */ + writel(reg_addr, pcie_index_offset); + readl(pcie_index_offset); + writel((u32)(reg_data & 0xffffffffULL), pcie_data_offset); + readl(pcie_data_offset); + /* write high 32 bits */ + writel(reg_addr + 4, pcie_index_offset); + readl(pcie_index_offset); + writel((u32)(reg_data >> 32), pcie_data_offset); + readl(pcie_data_offset); + spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); +} + +/** * amdgpu_invalid_rreg - dummy reg read function * * @adev: amdgpu device pointer @@ -1224,11 +1370,15 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev) amdgpu_gmc_tmz_set(adev); - if (amdgpu_num_kcq > 8 || amdgpu_num_kcq < 0) { + if (amdgpu_num_kcq == -1) { + amdgpu_num_kcq = 8; + } else if (amdgpu_num_kcq > 8 || amdgpu_num_kcq < 0) { amdgpu_num_kcq = 8; dev_warn(adev->dev, "set kernel compute queue number to 8 due to invalid parameter provided by user\n"); } + amdgpu_gmc_noretry_set(adev); + return 0; } @@ -1256,7 +1406,7 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; pci_set_power_state(dev->pdev, PCI_D0); - pci_restore_state(dev->pdev); + amdgpu_device_load_pci_state(dev->pdev); r = pci_enable_device(dev->pdev); if (r) DRM_WARN("pci_enable_device failed (%d)\n", r); @@ -1269,7 +1419,7 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, drm_kms_helper_poll_disable(dev); dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; amdgpu_device_suspend(dev, true); - pci_save_state(dev->pdev); + amdgpu_device_cache_pci_state(dev->pdev); /* Shut down the device */ pci_disable_device(dev->pdev); pci_set_power_state(dev->pdev, PCI_D3cold); @@ -1631,6 +1781,8 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev) case CHIP_CARRIZO: case CHIP_STONEY: case CHIP_VEGA20: + case CHIP_SIENNA_CICHLID: + case CHIP_NAVY_FLOUNDER: default: return 0; case CHIP_VEGA10: @@ -1662,12 +1814,6 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev) case CHIP_NAVI12: chip_name = "navi12"; break; - case CHIP_SIENNA_CICHLID: - chip_name = "sienna_cichlid"; - break; - case CHIP_NAVY_FLOUNDER: - chip_name = "navy_flounder"; - break; } snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_gpu_info.bin", chip_name); @@ -2999,6 +3145,7 @@ static const struct attribute *amdgpu_dev_attributes[] = { NULL }; + /** * amdgpu_device_init - initialize the driver * @@ -3170,13 +3317,13 @@ int amdgpu_device_init(struct amdgpu_device *adev, r = amdgpu_device_get_job_timeout_settings(adev); if (r) { dev_err(adev->dev, "invalid lockup_timeout parameter syntax\n"); - return r; + goto failed_unmap; } /* early init functions */ r = amdgpu_device_ip_early_init(adev); if (r) - return r; + goto failed_unmap; /* doorbell bar mapping and doorbell index init*/ amdgpu_device_doorbell_init(adev); @@ -3217,6 +3364,8 @@ int amdgpu_device_init(struct amdgpu_device *adev, } } + pci_enable_pcie_error_reporting(adev->ddev.pdev); + /* Post card if necessary */ if (amdgpu_device_need_post(adev)) { if (!adev->bios) { @@ -3359,16 +3508,18 @@ fence_driver_init: flush_delayed_work(&adev->delayed_init_work); r = sysfs_create_files(&adev->dev->kobj, amdgpu_dev_attributes); - if (r) { + if (r) dev_err(adev->dev, "Could not create amdgpu device attr\n"); - return r; - } if (IS_ENABLED(CONFIG_PERF_EVENTS)) r = amdgpu_pmu_init(adev); if (r) dev_err(adev->dev, "amdgpu_pmu_init failed\n"); + /* Have stored pci confspace at hand for restore in sudden PCI error */ + if (amdgpu_device_cache_pci_state(adev->pdev)) + pci_restore_state(pdev); + return 0; failed: @@ -3376,6 +3527,10 @@ failed: if (boco) vga_switcheroo_fini_domain_pm_ops(adev->dev); +failed_unmap: + iounmap(adev->rmmio); + adev->rmmio = NULL; + return r; } @@ -3393,11 +3548,15 @@ void amdgpu_device_fini(struct amdgpu_device *adev) flush_delayed_work(&adev->delayed_init_work); adev->shutdown = true; + kfree(adev->pci_state); + /* make sure IB test finished before entering exclusive mode * to avoid preemption on IB test * */ - if (amdgpu_sriov_vf(adev)) + if (amdgpu_sriov_vf(adev)) { amdgpu_virt_request_full_gpu(adev, false); + amdgpu_virt_fini_data_exchange(adev); + } /* disable all interrupts */ amdgpu_irq_disable_all(adev); @@ -4031,6 +4190,11 @@ static int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, amdgpu_debugfs_wait_dump(adev); + if (amdgpu_sriov_vf(adev)) { + /* stop the data exchange thread */ + amdgpu_virt_fini_data_exchange(adev); + } + /* block all schedulers and reset given job's ring */ for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { struct amdgpu_ring *ring = adev->rings[i]; @@ -4072,7 +4236,8 @@ static int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive, struct list_head *device_list_handle, - bool *need_full_reset_arg) + bool *need_full_reset_arg, + bool skip_hw_reset) { struct amdgpu_device *tmp_adev = NULL; bool need_full_reset = *need_full_reset_arg, vram_lost = false; @@ -4082,7 +4247,7 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive, * ASIC reset has to be done on all HGMI hive nodes ASAP * to allow proper links negotiation in FW (within 1 sec) */ - if (need_full_reset) { + if (!skip_hw_reset && need_full_reset) { list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { /* For XGMI run all resets in parallel to speed up the process */ if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { @@ -4460,7 +4625,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, retry: /* Rest of adevs pre asic reset from XGMI hive. */ list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { r = amdgpu_device_pre_asic_reset(tmp_adev, - NULL, + (tmp_adev == adev) ? job : NULL, &need_full_reset); /*TODO Should we stop ?*/ if (r) { @@ -4477,7 +4642,7 @@ retry: /* Rest of adevs pre asic reset from XGMI hive. */ if (r) adev->asic_reset_res = r; } else { - r = amdgpu_do_asic_reset(hive, device_list_handle, &need_full_reset); + r = amdgpu_do_asic_reset(hive, device_list_handle, &need_full_reset, false); if (r && r == -EAGAIN) goto retry; } @@ -4705,3 +4870,235 @@ int amdgpu_device_baco_exit(struct drm_device *dev) return 0; } + +static void amdgpu_cancel_all_tdr(struct amdgpu_device *adev) +{ + int i; + + for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { + struct amdgpu_ring *ring = adev->rings[i]; + + if (!ring || !ring->sched.thread) + continue; + + cancel_delayed_work_sync(&ring->sched.work_tdr); + } +} + +/** + * amdgpu_pci_error_detected - Called when a PCI error is detected. + * @pdev: PCI device struct + * @state: PCI channel state + * + * Description: Called when a PCI error is detected. + * + * Return: PCI_ERS_RESULT_NEED_RESET or PCI_ERS_RESULT_DISCONNECT. + */ +pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_state_t state) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + struct amdgpu_device *adev = drm_to_adev(dev); + int i; + + DRM_INFO("PCI error: detected callback, state(%d)!!\n", state); + + if (adev->gmc.xgmi.num_physical_nodes > 1) { + DRM_WARN("No support for XGMI hive yet..."); + return PCI_ERS_RESULT_DISCONNECT; + } + + switch (state) { + case pci_channel_io_normal: + return PCI_ERS_RESULT_CAN_RECOVER; + /* Fatal error, prepare for slot reset */ + case pci_channel_io_frozen: + /* + * Cancel and wait for all TDRs in progress if failing to + * set adev->in_gpu_reset in amdgpu_device_lock_adev + * + * Locking adev->reset_sem will prevent any external access + * to GPU during PCI error recovery + */ + while (!amdgpu_device_lock_adev(adev, NULL)) + amdgpu_cancel_all_tdr(adev); + + /* + * Block any work scheduling as we do for regular GPU reset + * for the duration of the recovery + */ + for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { + struct amdgpu_ring *ring = adev->rings[i]; + + if (!ring || !ring->sched.thread) + continue; + + drm_sched_stop(&ring->sched, NULL); + } + return PCI_ERS_RESULT_NEED_RESET; + case pci_channel_io_perm_failure: + /* Permanent error, prepare for device removal */ + return PCI_ERS_RESULT_DISCONNECT; + } + + return PCI_ERS_RESULT_NEED_RESET; +} + +/** + * amdgpu_pci_mmio_enabled - Enable MMIO and dump debug registers + * @pdev: pointer to PCI device + */ +pci_ers_result_t amdgpu_pci_mmio_enabled(struct pci_dev *pdev) +{ + + DRM_INFO("PCI error: mmio enabled callback!!\n"); + + /* TODO - dump whatever for debugging purposes */ + + /* This called only if amdgpu_pci_error_detected returns + * PCI_ERS_RESULT_CAN_RECOVER. Read/write to the device still + * works, no need to reset slot. + */ + + return PCI_ERS_RESULT_RECOVERED; +} + +/** + * amdgpu_pci_slot_reset - Called when PCI slot has been reset. + * @pdev: PCI device struct + * + * Description: This routine is called by the pci error recovery + * code after the PCI slot has been reset, just before we + * should resume normal operations. + */ +pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + struct amdgpu_device *adev = drm_to_adev(dev); + int r, i; + bool need_full_reset = true; + u32 memsize; + struct list_head device_list; + + DRM_INFO("PCI error: slot reset callback!!\n"); + + INIT_LIST_HEAD(&device_list); + list_add_tail(&adev->gmc.xgmi.head, &device_list); + + /* wait for asic to come out of reset */ + msleep(500); + + /* Restore PCI confspace */ + amdgpu_device_load_pci_state(pdev); + + /* confirm ASIC came out of reset */ + for (i = 0; i < adev->usec_timeout; i++) { + memsize = amdgpu_asic_get_config_memsize(adev); + + if (memsize != 0xffffffff) + break; + udelay(1); + } + if (memsize == 0xffffffff) { + r = -ETIME; + goto out; + } + + adev->in_pci_err_recovery = true; + r = amdgpu_device_pre_asic_reset(adev, NULL, &need_full_reset); + adev->in_pci_err_recovery = false; + if (r) + goto out; + + r = amdgpu_do_asic_reset(NULL, &device_list, &need_full_reset, true); + +out: + if (!r) { + if (amdgpu_device_cache_pci_state(adev->pdev)) + pci_restore_state(adev->pdev); + + DRM_INFO("PCIe error recovery succeeded\n"); + } else { + DRM_ERROR("PCIe error recovery failed, err:%d", r); + amdgpu_device_unlock_adev(adev); + } + + return r ? PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_RECOVERED; +} + +/** + * amdgpu_pci_resume() - resume normal ops after PCI reset + * @pdev: pointer to PCI device + * + * Called when the error recovery driver tells us that its + * OK to resume normal operation. Use completion to allow + * halted scsi ops to resume. + */ +void amdgpu_pci_resume(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + struct amdgpu_device *adev = drm_to_adev(dev); + int i; + + + DRM_INFO("PCI error: resume callback!!\n"); + + for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { + struct amdgpu_ring *ring = adev->rings[i]; + + if (!ring || !ring->sched.thread) + continue; + + + drm_sched_resubmit_jobs(&ring->sched); + drm_sched_start(&ring->sched, true); + } + + amdgpu_device_unlock_adev(adev); +} + +bool amdgpu_device_cache_pci_state(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + struct amdgpu_device *adev = drm_to_adev(dev); + int r; + + r = pci_save_state(pdev); + if (!r) { + kfree(adev->pci_state); + + adev->pci_state = pci_store_saved_state(pdev); + + if (!adev->pci_state) { + DRM_ERROR("Failed to store PCI saved state"); + return false; + } + } else { + DRM_WARN("Failed to save PCI state, err:%d\n", r); + return false; + } + + return true; +} + +bool amdgpu_device_load_pci_state(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + struct amdgpu_device *adev = drm_to_adev(dev); + int r; + + if (!adev->pci_state) + return false; + + r = pci_load_saved_state(pdev, adev->pci_state); + + if (!r) { + pci_restore_state(pdev); + } else { + DRM_WARN("Failed to load PCI state, err:%d\n", r); + return false; + } + + return true; +} + + diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h index 61a26c15c8dd..373cdebe0e2f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_df.h @@ -44,9 +44,9 @@ struct amdgpu_df_funcs { void (*enable_ecc_force_par_wr_rmw)(struct amdgpu_device *adev, bool enable); int (*pmc_start)(struct amdgpu_device *adev, uint64_t config, - int is_enable); + int is_add); int (*pmc_stop)(struct amdgpu_device *adev, uint64_t config, - int is_disable); + int is_remove); void (*pmc_get_count)(struct amdgpu_device *adev, uint64_t config, uint64_t *count); uint64_t (*get_fica)(struct amdgpu_device *adev, uint32_t ficaa_val); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index 4cba095b6c44..b25faaee6f0e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -293,7 +293,7 @@ int amdgpu_display_crtc_set_config(struct drm_mode_set *set, take the current one */ if (active && !adev->have_disp_power_ref) { adev->have_disp_power_ref = true; - goto out; + return ret; } /* if we have no active crtcs, then drop the power ref we got before */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index f359ffb77aa5..8b30915aa972 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -32,7 +32,6 @@ #include <drm/drm_pciids.h> #include <linux/console.h> #include <linux/module.h> -#include <linux/pci.h> #include <linux/pm_runtime.h> #include <linux/vga_switcheroo.h> #include <drm/drm_probe_helper.h> @@ -148,7 +147,7 @@ int amdgpu_async_gfx_ring = 1; int amdgpu_mcbp = 0; int amdgpu_discovery = -1; int amdgpu_mes = 0; -int amdgpu_noretry; +int amdgpu_noretry = -1; int amdgpu_force_asic_type = -1; int amdgpu_tmz = 0; int amdgpu_reset_method = -1; /* auto */ @@ -597,8 +596,13 @@ MODULE_PARM_DESC(mes, "Enable Micro Engine Scheduler (0 = disabled (default), 1 = enabled)"); module_param_named(mes, amdgpu_mes, int, 0444); +/** + * DOC: noretry (int) + * Disable retry faults in the GPU memory controller. + * (0 = retry enabled, 1 = retry disabled, -1 auto (default)) + */ MODULE_PARM_DESC(noretry, - "Disable retry faults (0 = retry enabled (default), 1 = retry disabled)"); + "Disable retry faults (0 = retry enabled, 1 = retry disabled, -1 auto (default))"); module_param_named(noretry, amdgpu_noretry, int, 0644); /** @@ -1062,6 +1066,7 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x7319, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10}, {0x1002, 0x731A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10}, {0x1002, 0x731B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10}, + {0x1002, 0x731E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10}, {0x1002, 0x731F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10}, /* Navi14 */ {0x1002, 0x7340, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI14}, @@ -1073,8 +1078,16 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x1636, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RENOIR|AMD_IS_APU}, /* Navi12 */ - {0x1002, 0x7360, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12|AMD_EXP_HW_SUPPORT}, - {0x1002, 0x7362, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12|AMD_EXP_HW_SUPPORT}, + {0x1002, 0x7360, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12}, + {0x1002, 0x7362, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12}, + + /* Sienna_Cichlid */ + {0x1002, 0x73A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, + {0x1002, 0x73A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, + {0x1002, 0x73A3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, + {0x1002, 0x73AB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, + {0x1002, 0x73AE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, + {0x1002, 0x73BF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, {0, 0, 0} }; @@ -1102,6 +1115,16 @@ static int amdgpu_pci_probe(struct pci_dev *pdev, return -ENODEV; } + /* Due to hardware bugs, S/G Display on raven requires a 1:1 IOMMU mapping, + * however, SME requires an indirect IOMMU mapping because the encryption + * bit is beyond the DMA mask of the chip. + */ + if (mem_encrypt_active() && ((flags & AMD_ASIC_MASK) == CHIP_RAVEN)) { + dev_info(&pdev->dev, + "SME is not compatible with RAVEN\n"); + return -ENOTSUPP; + } + #ifdef CONFIG_DRM_AMDGPU_SI if (!amdgpu_si_support) { switch (flags & AMD_ASIC_MASK) { @@ -1308,7 +1331,7 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev) if (amdgpu_is_atpx_hybrid()) { pci_ignore_hotplug(pdev); } else { - pci_save_state(pdev); + amdgpu_device_cache_pci_state(pdev); pci_disable_device(pdev); pci_ignore_hotplug(pdev); pci_set_power_state(pdev, PCI_D3cold); @@ -1341,7 +1364,7 @@ static int amdgpu_pmops_runtime_resume(struct device *dev) pci_set_master(pdev); } else { pci_set_power_state(pdev, PCI_D0); - pci_restore_state(pdev); + amdgpu_device_load_pci_state(pdev); ret = pci_enable_device(pdev); if (ret) return ret; @@ -1514,6 +1537,13 @@ static struct drm_driver kms_driver = { .patchlevel = KMS_DRIVER_PATCHLEVEL, }; +static struct pci_error_handlers amdgpu_pci_err_handler = { + .error_detected = amdgpu_pci_error_detected, + .mmio_enabled = amdgpu_pci_mmio_enabled, + .slot_reset = amdgpu_pci_slot_reset, + .resume = amdgpu_pci_resume, +}; + static struct pci_driver amdgpu_kms_pci_driver = { .name = DRIVER_NAME, .id_table = pciidlist, @@ -1521,6 +1551,7 @@ static struct pci_driver amdgpu_kms_pci_driver = { .remove = amdgpu_pci_remove, .shutdown = amdgpu_pci_shutdown, .driver.pm = &amdgpu_pm_ops, + .err_handler = &amdgpu_pci_err_handler, }; static int __init amdgpu_init(void) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c index e811fecc540f..8f4a8f8d8146 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c @@ -34,18 +34,31 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev) { - /* TODO: Gaming SKUs don't have the FRU EEPROM. - * Use this hack to address hangs on modprobe on gaming SKUs - * until a proper solution can be implemented by only supporting - * the explicit chip IDs for VG20 Server cards - * - * TODO: Add list of supported Arcturus DIDs once confirmed + /* Only server cards have the FRU EEPROM + * TODO: See if we can figure this out dynamically instead of + * having to parse VBIOS versions. */ - if ((adev->asic_type == CHIP_VEGA20 && adev->pdev->device == 0x66a0) || - (adev->asic_type == CHIP_VEGA20 && adev->pdev->device == 0x66a1) || - (adev->asic_type == CHIP_VEGA20 && adev->pdev->device == 0x66a4)) - return true; - return false; + struct atom_context *atom_ctx = adev->mode_info.atom_context; + + /* VBIOS is of the format ###-DXXXYY-##. 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 + * reasonable and not too expensive + */ + switch (adev->asic_type) { + case CHIP_VEGA20: + /* D161 and D163 are the VG20 server SKUs */ + if (strnstr(atom_ctx->vbios_version, "D161", + sizeof(atom_ctx->vbios_version)) || + strnstr(atom_ctx->vbios_version, "D163", + 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, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.h index f29a8611d69b..1308d976d60e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.h @@ -26,4 +26,4 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev); -#endif // __AMDGPU_PRODINFO_H__ +#endif // __AMDGPU_FRU_EEPROM_H__ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index be08a63ef58c..8ea6fc745769 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -609,6 +609,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data, struct ww_acquire_ctx ticket; struct list_head list, duplicates; uint64_t va_flags; + uint64_t vm_size; int r = 0; if (args->va_address < AMDGPU_VA_RESERVED_SIZE) { @@ -629,6 +630,15 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data, args->va_address &= AMDGPU_GMC_HOLE_MASK; + vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE; + vm_size -= AMDGPU_VA_RESERVED_SIZE; + if (args->va_address + args->map_size > vm_size) { + dev_dbg(&dev->pdev->dev, + "va_address 0x%llx is in top reserved area 0x%llx\n", + args->va_address + args->map_size, vm_size); + return -EINVAL; + } + if ((args->flags & ~valid_flags) && (args->flags & ~prt_flags)) { dev_dbg(&dev->pdev->dev, "invalid flags combination 0x%08X\n", args->flags); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index d6981425ec51..8c9bacfdbc30 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -693,6 +693,9 @@ uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg) struct amdgpu_kiq *kiq = &adev->gfx.kiq; struct amdgpu_ring *ring = &kiq->ring; + if (adev->in_pci_err_recovery) + return 0; + BUG_ON(!ring->funcs->emit_rreg); spin_lock_irqsave(&kiq->ring_lock, flags); @@ -757,6 +760,9 @@ void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v) BUG_ON(!ring->funcs->emit_wreg); + if (adev->in_pci_err_recovery) + return; + spin_lock_irqsave(&kiq->ring_lock, flags); amdgpu_ring_alloc(ring, 32); amdgpu_ring_emit_wreg(ring, reg, v); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index a611e78dd4ba..258498cbf1eb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h @@ -217,6 +217,7 @@ struct amdgpu_gfx_funcs { int (*query_ras_error_count) (struct amdgpu_device *adev, void *ras_error_status); void (*reset_ras_error_count) (struct amdgpu_device *adev); void (*init_spm_golden)(struct amdgpu_device *adev); + void (*query_ras_error_status) (struct amdgpu_device *adev); }; struct sq_work { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h new file mode 100644 index 000000000000..66ebc2e3b2ad --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h @@ -0,0 +1,43 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __AMDGPU_GFXHUB_H__ +#define __AMDGPU_GFXHUB_H__ + +struct amdgpu_gfxhub_funcs { + u64 (*get_fb_location)(struct amdgpu_device *adev); + u64 (*get_mc_fb_offset)(struct amdgpu_device *adev); + void (*setup_vm_pt_regs)(struct amdgpu_device *adev, uint32_t vmid, + uint64_t page_table_base); + int (*gart_enable)(struct amdgpu_device *adev); + + void (*gart_disable)(struct amdgpu_device *adev); + void (*set_fault_enable_default)(struct amdgpu_device *adev, bool value); + void (*init)(struct amdgpu_device *adev); + int (*get_xgmi_info)(struct amdgpu_device *adev); +}; + +struct amdgpu_gfxhub { + const struct amdgpu_gfxhub_funcs *funcs; +}; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 0e35023b5703..cc86f431a3d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -409,6 +409,44 @@ void amdgpu_gmc_tmz_set(struct amdgpu_device *adev) } } +/** + * amdgpu_noretry_set -- set per asic noretry defaults + * @adev: amdgpu_device pointer + * + * Set a per asic default for the no-retry parameter. + * + */ +void amdgpu_gmc_noretry_set(struct amdgpu_device *adev) +{ + struct amdgpu_gmc *gmc = &adev->gmc; + + switch (adev->asic_type) { + case CHIP_RAVEN: + /* Raven currently has issues with noretry + * regardless of what we decide for other + * asics, we should leave raven with + * noretry = 0 until we root cause the + * issues. + */ + if (amdgpu_noretry == -1) + gmc->noretry = 0; + else + gmc->noretry = amdgpu_noretry; + break; + default: + /* default this to 0 for now, but we may want + * to change this in the future for certain + * GPUs as it can increase performance in + * certain cases. + */ + if (amdgpu_noretry == -1) + gmc->noretry = 0; + else + gmc->noretry = amdgpu_noretry; + break; + } +} + void amdgpu_gmc_set_vm_fault_masks(struct amdgpu_device *adev, int hub_type, bool enable) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h index d61bbde4c7d2..aa0c83776ce0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h @@ -239,6 +239,7 @@ struct amdgpu_gmc { struct amdgpu_xgmi xgmi; struct amdgpu_irq_src ecc_irq; + int noretry; }; #define amdgpu_gmc_flush_gpu_tlb(adev, vmid, vmhub, type) ((adev)->gmc.gmc_funcs->flush_gpu_tlb((adev), (vmid), (vmhub), (type))) @@ -300,6 +301,7 @@ void amdgpu_gmc_ras_fini(struct amdgpu_device *adev); int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev); extern void amdgpu_gmc_tmz_set(struct amdgpu_device *adev); +extern void amdgpu_gmc_noretry_set(struct amdgpu_device *adev); extern void amdgpu_gmc_set_vm_fault_masks(struct amdgpu_device *adev, int hub_type, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c index 7521f4ab55de..6e9a9e5dbea0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c @@ -43,7 +43,7 @@ static DEFINE_IDA(amdgpu_pasid_ida); /* Helper to free pasid from a fence callback */ struct amdgpu_pasid_cb { struct dma_fence_cb cb; - unsigned int pasid; + u32 pasid; }; /** @@ -79,7 +79,7 @@ int amdgpu_pasid_alloc(unsigned int bits) * amdgpu_pasid_free - Free a PASID * @pasid: PASID to free */ -void amdgpu_pasid_free(unsigned int pasid) +void amdgpu_pasid_free(u32 pasid) { trace_amdgpu_pasid_freed(pasid); ida_simple_remove(&amdgpu_pasid_ida, pasid); @@ -105,7 +105,7 @@ static void amdgpu_pasid_free_cb(struct dma_fence *fence, * Free the pasid only after all the fences in resv are signaled. */ void amdgpu_pasid_free_delayed(struct dma_resv *resv, - unsigned int pasid) + u32 pasid) { struct dma_fence *fence, **fences; struct amdgpu_pasid_cb *cb; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h index 8e58325bbca2..0c3b4fa1f936 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h @@ -71,9 +71,9 @@ struct amdgpu_vmid_mgr { }; int amdgpu_pasid_alloc(unsigned int bits); -void amdgpu_pasid_free(unsigned int pasid); +void amdgpu_pasid_free(u32 pasid); void amdgpu_pasid_free_delayed(struct dma_resv *resv, - unsigned int pasid); + u32 pasid); bool amdgpu_vmid_had_gpu_reset(struct amdgpu_device *adev, struct amdgpu_vmid *id); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index 456a4a93b337..efda38349a03 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -177,7 +177,7 @@ int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags) break; case CHIP_VEGA10: /* turn runpm on if noretry=0 */ - if (!amdgpu_noretry) + if (!adev->gmc.noretry) adev->runpm = true; break; default: @@ -282,14 +282,25 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info, fw_info->feature = 0; break; case AMDGPU_INFO_FW_TA: - if (query_fw->index > 1) - return -EINVAL; - if (query_fw->index == 0) { + switch (query_fw->index) { + case 0: fw_info->ver = adev->psp.ta_fw_version; fw_info->feature = adev->psp.ta_xgmi_ucode_version; - } else { + break; + case 1: fw_info->ver = adev->psp.ta_fw_version; fw_info->feature = adev->psp.ta_ras_ucode_version; + break; + case 2: + fw_info->ver = adev->psp.ta_fw_version; + fw_info->feature = adev->psp.ta_hdcp_ucode_version; + break; + case 3: + fw_info->ver = adev->psp.ta_fw_version; + fw_info->feature = adev->psp.ta_dtm_ucode_version; + break; + default: + return -EINVAL; } break; case AMDGPU_INFO_FW_SDMA: @@ -1080,7 +1091,7 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev, struct amdgpu_fpriv *fpriv = file_priv->driver_priv; struct amdgpu_bo_list *list; struct amdgpu_bo *pd; - unsigned int pasid; + u32 pasid; int handle; if (!fpriv) @@ -1385,13 +1396,31 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data) fw_info.feature, fw_info.ver); query_fw.fw_type = AMDGPU_INFO_FW_TA; - for (i = 0; i < 2; i++) { + for (i = 0; i < 4; i++) { query_fw.index = i; ret = amdgpu_firmware_info(&fw_info, &query_fw, adev); if (ret) continue; - seq_printf(m, "TA %s feature version: %u, firmware version: 0x%08x\n", - i ? "RAS" : "XGMI", fw_info.feature, fw_info.ver); + switch (query_fw.index) { + case 0: + seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n", + "RAS", fw_info.feature, fw_info.ver); + break; + case 1: + seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n", + "XGMI", fw_info.feature, fw_info.ver); + break; + case 2: + seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n", + "HDCP", fw_info.feature, fw_info.ver); + break; + case 3: + seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n", + "DTM", fw_info.feature, fw_info.ver); + break; + default: + return -EINVAL; + } } /* SMC */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h index 0c43d7fe893c..1ae9bdae7311 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h @@ -40,6 +40,7 @@ struct amdgpu_mmhub_funcs { uint64_t page_table_base); void (*update_power_gating)(struct amdgpu_device *adev, bool enable); + void (*query_ras_error_status)(struct amdgpu_device *adev); }; struct amdgpu_mmhub { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index 04a430e0e2e1..a04decb934b0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -46,6 +46,7 @@ #include <drm/drm_dp_mst_helper.h> #include "modules/inc/mod_freesync.h" +#include "amdgpu_dm_irq_params.h" struct amdgpu_bo; struct amdgpu_device; @@ -404,7 +405,8 @@ struct amdgpu_crtc { struct amdgpu_flip_work *pflip_works; enum amdgpu_flip_status pflip_status; int deferred_flip_completion; - u32 last_flip_vblank; + /* parameters access from DM IRQ handler */ + struct dm_irq_params dm_irq_params; /* pll sharing */ struct amdgpu_atom_ss ss; bool ss_enabled; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index d6c38e24f130..96a9699f87ba 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -161,10 +161,12 @@ static int psp_sw_init(void *handle) struct psp_context *psp = &adev->psp; int ret; - ret = psp_init_microcode(psp); - if (ret) { - DRM_ERROR("Failed to load psp firmware!\n"); - return ret; + if (!amdgpu_sriov_vf(adev)) { + ret = psp_init_microcode(psp); + if (ret) { + DRM_ERROR("Failed to load psp firmware!\n"); + return ret; + } } ret = psp_memory_training_init(psp); @@ -178,7 +180,7 @@ static int psp_sw_init(void *handle) return ret; } - if (adev->asic_type == CHIP_NAVI10) { + if (adev->asic_type == CHIP_NAVI10 || adev->asic_type == CHIP_SIENNA_CICHLID) { ret= psp_sysfs_init(adev); if (ret) { return ret; @@ -206,7 +208,8 @@ static int psp_sw_fini(void *handle) adev->psp.ta_fw = NULL; } - if (adev->asic_type == CHIP_NAVI10) + if (adev->asic_type == CHIP_NAVI10 || + adev->asic_type == CHIP_SIENNA_CICHLID) psp_sysfs_fini(adev); return 0; @@ -219,6 +222,9 @@ int psp_wait_for(struct psp_context *psp, uint32_t reg_index, int i; struct amdgpu_device *adev = psp->adev; + if (psp->adev->in_pci_err_recovery) + return 0; + for (i = 0; i < adev->usec_timeout; i++) { val = RREG32(reg_index); if (check_changed) { @@ -245,6 +251,9 @@ psp_cmd_submit_buf(struct psp_context *psp, bool ras_intr = false; bool skip_unsupport = false; + if (psp->adev->in_pci_err_recovery) + return 0; + mutex_lock(&psp->mutex); memset(psp->cmd_buf_mem, 0, PSP_CMD_BUFFER_SIZE); @@ -929,6 +938,7 @@ static int psp_ras_load(struct psp_context *psp) { int ret; struct psp_gfx_cmd_resp *cmd; + struct ta_ras_shared_memory *ras_cmd; /* * TODO: bypass the loading in sriov for now @@ -952,11 +962,20 @@ static int psp_ras_load(struct psp_context *psp) ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr); + ras_cmd = (struct ta_ras_shared_memory*)psp->ras.ras_shared_buf; + if (!ret) { - psp->ras.ras_initialized = true; psp->ras.session_id = cmd->resp.session_id; + + if (!ras_cmd->ras_status) + psp->ras.ras_initialized = true; + else + dev_warn(psp->adev->dev, "RAS Init Status: 0x%X\n", ras_cmd->ras_status); } + if (ret || ras_cmd->ras_status) + amdgpu_ras_fini(psp->adev); + kfree(cmd); return ret; @@ -1732,6 +1751,12 @@ static int psp_get_fw_type(struct amdgpu_firmware_info *ucode, case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM: *type = GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM; break; + case AMDGPU_UCODE_ID_RLC_IRAM: + *type = GFX_FW_TYPE_RLC_IRAM; + break; + case AMDGPU_UCODE_ID_RLC_DRAM: + *type = GFX_FW_TYPE_RLC_DRAM_BOOT; + break; case AMDGPU_UCODE_ID_SMC: *type = GFX_FW_TYPE_SMU; break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index e5ea14774c0c..4e36551ab50b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1027,58 +1027,6 @@ static ssize_t amdgpu_ras_sysfs_features_read(struct device *dev, return scnprintf(buf, PAGE_SIZE, "feature mask: 0x%x\n", con->features); } -static void amdgpu_ras_sysfs_add_bad_page_node(struct amdgpu_device *adev) -{ - struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - struct attribute_group group; - struct bin_attribute *bin_attrs[] = { - &con->badpages_attr, - NULL, - }; - - con->badpages_attr = (struct bin_attribute) { - .attr = { - .name = "gpu_vram_bad_pages", - .mode = S_IRUGO, - }, - .size = 0, - .private = NULL, - .read = amdgpu_ras_sysfs_badpages_read, - }; - - group.name = RAS_FS_NAME; - group.bin_attrs = bin_attrs; - - sysfs_bin_attr_init(bin_attrs[0]); - - sysfs_update_group(&adev->dev->kobj, &group); -} - -static int amdgpu_ras_sysfs_create_feature_node(struct amdgpu_device *adev) -{ - struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - struct attribute *attrs[] = { - &con->features_attr.attr, - NULL - }; - struct attribute_group group = { - .name = RAS_FS_NAME, - .attrs = attrs, - }; - - con->features_attr = (struct device_attribute) { - .attr = { - .name = "features", - .mode = S_IRUGO, - }, - .show = amdgpu_ras_sysfs_features_read, - }; - - sysfs_attr_init(attrs[0]); - - return sysfs_create_group(&adev->dev->kobj, &group); -} - static void amdgpu_ras_sysfs_remove_bad_page_node(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); @@ -1300,13 +1248,43 @@ static void amdgpu_ras_debugfs_remove_all(struct amdgpu_device *adev) /* debugfs end */ /* ras fs */ - +static BIN_ATTR(gpu_vram_bad_pages, S_IRUGO, + amdgpu_ras_sysfs_badpages_read, NULL, 0); +static DEVICE_ATTR(features, S_IRUGO, + amdgpu_ras_sysfs_features_read, NULL); static int amdgpu_ras_fs_init(struct amdgpu_device *adev) { - amdgpu_ras_sysfs_create_feature_node(adev); + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct attribute_group group = { + .name = RAS_FS_NAME, + }; + struct attribute *attrs[] = { + &con->features_attr.attr, + NULL + }; + struct bin_attribute *bin_attrs[] = { + NULL, + NULL, + }; + int r; - if (amdgpu_bad_page_threshold != 0) - amdgpu_ras_sysfs_add_bad_page_node(adev); + /* add features entry */ + con->features_attr = dev_attr_features; + group.attrs = attrs; + sysfs_attr_init(attrs[0]); + + if (amdgpu_bad_page_threshold != 0) { + /* add bad_page_features entry */ + bin_attr_gpu_vram_bad_pages.private = NULL; + con->badpages_attr = bin_attr_gpu_vram_bad_pages; + bin_attrs[0] = &con->badpages_attr; + group.bin_attrs = bin_attrs; + sysfs_bin_attr_init(bin_attrs[0]); + } + + r = sysfs_create_group(&adev->dev->kobj, &group); + if (r) + dev_err(adev->dev, "Failed to create RAS sysfs group!"); return 0; } @@ -1498,6 +1476,45 @@ static void amdgpu_ras_log_on_err_counter(struct amdgpu_device *adev) } } +/* Parse RdRspStatus and WrRspStatus */ +void amdgpu_ras_error_status_query(struct amdgpu_device *adev, + struct ras_query_if *info) +{ + /* + * Only two block need to query read/write + * RspStatus at current state + */ + switch (info->head.block) { + case AMDGPU_RAS_BLOCK__GFX: + if (adev->gfx.funcs->query_ras_error_status) + adev->gfx.funcs->query_ras_error_status(adev); + break; + case AMDGPU_RAS_BLOCK__MMHUB: + if (adev->mmhub.funcs->query_ras_error_status) + adev->mmhub.funcs->query_ras_error_status(adev); + break; + default: + break; + } +} + +static void amdgpu_ras_query_err_status(struct amdgpu_device *adev) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct ras_manager *obj; + + if (!con) + return; + + list_for_each_entry(obj, &con->head, node) { + struct ras_query_if info = { + .head = obj->head, + }; + + amdgpu_ras_error_status_query(adev, &info); + } +} + /* recovery begin */ /* return 0 on success. @@ -1568,8 +1585,10 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) } list_for_each_entry(remote_adev, - device_list_handle, gmc.xgmi.head) + device_list_handle, gmc.xgmi.head) { + amdgpu_ras_query_err_status(remote_adev); amdgpu_ras_log_on_err_counter(remote_adev); + } amdgpu_put_xgmi_hive(hive); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h index 60bb3e8b3118..aeaaae713c59 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h @@ -168,12 +168,16 @@ struct amdgpu_rlc { u32 save_restore_list_cntl_size_bytes; u32 save_restore_list_gpm_size_bytes; u32 save_restore_list_srm_size_bytes; + u32 rlc_iram_ucode_size_bytes; + u32 rlc_dram_ucode_size_bytes; u32 *register_list_format; u32 *register_restore; u8 *save_restore_list_cntl; u8 *save_restore_list_gpm; u8 *save_restore_list_srm; + u8 *rlc_iram_ucode; + u8 *rlc_dram_ucode; bool is_rlc_v2_1; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h index 63e734a125fb..ee9480d14cbc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h @@ -35,7 +35,7 @@ #define AMDGPU_JOB_GET_TIMELINE_NAME(job) \ job->base.s_fence->finished.ops->get_timeline_name(&job->base.s_fence->finished) -TRACE_EVENT(amdgpu_mm_rreg, +TRACE_EVENT(amdgpu_device_rreg, TP_PROTO(unsigned did, uint32_t reg, uint32_t value), TP_ARGS(did, reg, value), TP_STRUCT__entry( @@ -54,7 +54,7 @@ TRACE_EVENT(amdgpu_mm_rreg, (unsigned long)__entry->value) ); -TRACE_EVENT(amdgpu_mm_wreg, +TRACE_EVENT(amdgpu_device_wreg, TP_PROTO(unsigned did, uint32_t reg, uint32_t value), TP_ARGS(did, reg, value), TP_STRUCT__entry( @@ -321,6 +321,49 @@ DEFINE_EVENT(amdgpu_vm_mapping, amdgpu_vm_bo_cs, TP_ARGS(mapping) ); +TRACE_EVENT(amdgpu_vm_update_ptes, + TP_PROTO(struct amdgpu_vm_update_params *p, + uint64_t start, uint64_t end, + unsigned int nptes, uint64_t dst, + uint64_t incr, uint64_t flags, + pid_t pid, uint64_t vm_ctx), + TP_ARGS(p, start, end, nptes, dst, incr, flags, pid, vm_ctx), + TP_STRUCT__entry( + __field(u64, start) + __field(u64, end) + __field(u64, flags) + __field(unsigned int, nptes) + __field(u64, incr) + __field(pid_t, pid) + __field(u64, vm_ctx) + __dynamic_array(u64, dst, nptes) + ), + + TP_fast_assign( + unsigned int i; + + __entry->start = start; + __entry->end = end; + __entry->flags = flags; + __entry->incr = incr; + __entry->nptes = nptes; + __entry->pid = pid; + __entry->vm_ctx = vm_ctx; + for (i = 0; i < nptes; ++i) { + u64 addr = p->pages_addr ? amdgpu_vm_map_gart( + p->pages_addr, dst) : dst; + + ((u64 *)__get_dynamic_array(dst))[i] = addr; + dst += incr; + } + ), + TP_printk("pid:%u vm_ctx:0x%llx start:0x%010llx end:0x%010llx," + " 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), __entry->nptes, 8)) +); + TRACE_EVENT(amdgpu_vm_set_ptes, TP_PROTO(uint64_t pe, uint64_t addr, unsigned count, uint32_t incr, uint64_t flags, bool direct), diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index beacd00221d8..ddb1c8e9eea4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -1044,6 +1044,7 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_bo_device *bdev, release_sg: kfree(ttm->sg); + ttm->sg = NULL; return r; } @@ -1187,7 +1188,12 @@ static int amdgpu_ttm_backend_bind(struct ttm_bo_device *bdev, } /** - * amdgpu_ttm_alloc_gart - Allocate GART memory for buffer object + * amdgpu_ttm_alloc_gart - Make sure buffer object is accessible either + * through AGP or GART aperture. + * + * If bo is accessible through AGP aperture, then use AGP aperture + * to access bo; otherwise allocate logical space in GART aperture + * and map bo to GART aperture. */ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c index 55fe19a2f332..b313ce4c3e97 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c @@ -500,6 +500,8 @@ static int amdgpu_ucode_init_single_fw(struct amdgpu_device *adev, ucode->ucode_id != AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL && ucode->ucode_id != AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM && ucode->ucode_id != AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM && + ucode->ucode_id != AMDGPU_UCODE_ID_RLC_IRAM && + ucode->ucode_id != AMDGPU_UCODE_ID_RLC_DRAM && ucode->ucode_id != AMDGPU_UCODE_ID_DMCU_ERAM && ucode->ucode_id != AMDGPU_UCODE_ID_DMCU_INTV && ucode->ucode_id != AMDGPU_UCODE_ID_DMCUB)) { @@ -556,6 +558,14 @@ static int amdgpu_ucode_init_single_fw(struct amdgpu_device *adev, ucode->ucode_size = adev->gfx.rlc.save_restore_list_srm_size_bytes; memcpy(ucode->kaddr, adev->gfx.rlc.save_restore_list_srm, ucode->ucode_size); + } else if (ucode->ucode_id == AMDGPU_UCODE_ID_RLC_IRAM) { + ucode->ucode_size = adev->gfx.rlc.rlc_iram_ucode_size_bytes; + memcpy(ucode->kaddr, adev->gfx.rlc.rlc_iram_ucode, + ucode->ucode_size); + } else if (ucode->ucode_id == AMDGPU_UCODE_ID_RLC_DRAM) { + ucode->ucode_size = adev->gfx.rlc.rlc_dram_ucode_size_bytes; + memcpy(ucode->kaddr, adev->gfx.rlc.rlc_dram_ucode, + ucode->ucode_size); } else if (ucode->ucode_id == AMDGPU_UCODE_ID_CP_MES) { ucode->ucode_size = le32_to_cpu(mes_hdr->mes_ucode_size_bytes); memcpy(ucode->kaddr, (void *)((uint8_t *)adev->mes.fw->data + diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h index 3c23c6293ff9..0e43b46d3ab5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h @@ -222,6 +222,15 @@ struct rlc_firmware_header_v2_1 { uint32_t save_restore_list_srm_offset_bytes; }; +/* version_major=2, version_minor=1 */ +struct rlc_firmware_header_v2_2 { + struct rlc_firmware_header_v2_1 v2_1; + uint32_t rlc_iram_ucode_size_bytes; + uint32_t rlc_iram_ucode_offset_bytes; + uint32_t rlc_dram_ucode_size_bytes; + uint32_t rlc_dram_ucode_offset_bytes; +}; + /* version_major=1, version_minor=0 */ struct sdma_firmware_header_v1_0 { struct common_firmware_header header; @@ -339,6 +348,8 @@ enum AMDGPU_UCODE_ID { AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL, AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM, AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM, + AMDGPU_UCODE_ID_RLC_IRAM, + AMDGPU_UCODE_ID_RLC_DRAM, AMDGPU_UCODE_ID_RLC_G, AMDGPU_UCODE_ID_STORAGE, AMDGPU_UCODE_ID_SMC, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index 495c3d7bb2b2..f3b7287e84c4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -68,6 +68,7 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) INIT_DELAYED_WORK(&adev->vcn.idle_work, amdgpu_vcn_idle_work_handler); mutex_init(&adev->vcn.vcn_pg_lock); + mutex_init(&adev->vcn.vcn1_jpeg1_workaround); atomic_set(&adev->vcn.total_submission_cnt, 0); for (i = 0; i < adev->vcn.num_vcn_inst; i++) atomic_set(&adev->vcn.inst[i].dpg_enc_submission_cnt, 0); @@ -237,6 +238,7 @@ int amdgpu_vcn_sw_fini(struct amdgpu_device *adev) } release_firmware(adev->vcn.fw); + mutex_destroy(&adev->vcn.vcn1_jpeg1_workaround); mutex_destroy(&adev->vcn.vcn_pg_lock); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h index 7a9b804bc988..17691158f783 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h @@ -220,6 +220,7 @@ struct amdgpu_vcn { struct amdgpu_vcn_inst inst[AMDGPU_MAX_VCN_INSTANCES]; struct amdgpu_vcn_reg internal; struct mutex vcn_pg_lock; + struct mutex vcn1_jpeg1_workaround; atomic_t total_submission_cnt; unsigned harvest_config; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index f76961d17246..d0aea5e39531 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -31,6 +31,12 @@ #include "soc15.h" #include "nv.h" +#define POPULATE_UCODE_INFO(vf2pf_info, ucode, ver) \ + do { \ + vf2pf_info->ucode_info[ucode].id = ucode; \ + vf2pf_info->ucode_info[ucode].version = ver; \ + } while (0) + bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev) { /* By now all MMIO pages except mailbox are blocked */ @@ -239,10 +245,10 @@ void amdgpu_virt_free_mm_table(struct amdgpu_device *adev) } -int amdgpu_virt_fw_reserve_get_checksum(void *obj, - unsigned long obj_size, - unsigned int key, - unsigned int chksum) +unsigned int amd_sriov_msg_checksum(void *obj, + unsigned long obj_size, + unsigned int key, + unsigned int checksum) { unsigned int ret = key; unsigned long i = 0; @@ -252,9 +258,9 @@ int amdgpu_virt_fw_reserve_get_checksum(void *obj, /* calculate checksum */ for (i = 0; i < obj_size; ++i) ret += *(pos + i); - /* minus the chksum itself */ - pos = (char *)&chksum; - for (i = 0; i < sizeof(chksum); ++i) + /* minus the checksum itself */ + pos = (char *)&checksum; + for (i = 0; i < sizeof(checksum); ++i) ret -= *(pos + i); return ret; } @@ -415,33 +421,188 @@ static void amdgpu_virt_add_bad_page(struct amdgpu_device *adev, } } -void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev) +static int amdgpu_virt_read_pf2vf_data(struct amdgpu_device *adev) { - uint32_t pf2vf_size = 0; - uint32_t checksum = 0; + struct amd_sriov_msg_pf2vf_info_header *pf2vf_info = adev->virt.fw_reserve.p_pf2vf; + uint32_t checksum; uint32_t checkval; - char *str; + + if (adev->virt.fw_reserve.p_pf2vf == NULL) + return -EINVAL; + + if (pf2vf_info->size > 1024) { + DRM_ERROR("invalid pf2vf message size\n"); + return -EINVAL; + } + + switch (pf2vf_info->version) { + case 1: + checksum = ((struct amdgim_pf2vf_info_v1 *)pf2vf_info)->checksum; + checkval = amd_sriov_msg_checksum( + adev->virt.fw_reserve.p_pf2vf, pf2vf_info->size, + adev->virt.fw_reserve.checksum_key, checksum); + if (checksum != checkval) { + DRM_ERROR("invalid pf2vf message\n"); + return -EINVAL; + } + + adev->virt.gim_feature = + ((struct amdgim_pf2vf_info_v1 *)pf2vf_info)->feature_flags; + break; + case 2: + /* TODO: missing key, need to add it later */ + checksum = ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->checksum; + checkval = amd_sriov_msg_checksum( + adev->virt.fw_reserve.p_pf2vf, pf2vf_info->size, + 0, checksum); + if (checksum != checkval) { + DRM_ERROR("invalid pf2vf message\n"); + return -EINVAL; + } + + adev->virt.vf2pf_update_interval_ms = + ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->vf2pf_update_interval_ms; + adev->virt.gim_feature = + ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->feature_flags.all; + + break; + default: + DRM_ERROR("invalid pf2vf version\n"); + return -EINVAL; + } + + /* correct too large or too little interval value */ + if (adev->virt.vf2pf_update_interval_ms < 200 || adev->virt.vf2pf_update_interval_ms > 10000) + adev->virt.vf2pf_update_interval_ms = 2000; + + return 0; +} + +static void amdgpu_virt_populate_vf2pf_ucode_info(struct amdgpu_device *adev) +{ + struct amd_sriov_msg_vf2pf_info *vf2pf_info; + vf2pf_info = (struct amd_sriov_msg_vf2pf_info *) adev->virt.fw_reserve.p_vf2pf; + + if (adev->virt.fw_reserve.p_vf2pf == NULL) + return; + + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_VCE, adev->vce.fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_UVD, adev->uvd.fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_MC, adev->gmc.fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_ME, adev->gfx.me_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_PFP, adev->gfx.pfp_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_CE, adev->gfx.ce_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_RLC, adev->gfx.rlc_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_RLC_SRLC, adev->gfx.rlc_srlc_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_RLC_SRLG, adev->gfx.rlc_srlg_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_RLC_SRLS, adev->gfx.rlc_srls_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_MEC, adev->gfx.mec_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_MEC2, adev->gfx.mec2_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_SOS, adev->psp.sos_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_ASD, adev->psp.asd_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_TA_RAS, adev->psp.ta_ras_ucode_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_TA_XGMI, adev->psp.ta_xgmi_ucode_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_SMC, adev->pm.fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_SDMA, adev->sdma.instance[0].fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_SDMA2, adev->sdma.instance[1].fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_VCN, adev->vcn.fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_DMCU, adev->dm.dmcu_fw_version); +} + +static int amdgpu_virt_write_vf2pf_data(struct amdgpu_device *adev) +{ + struct amd_sriov_msg_vf2pf_info *vf2pf_info; + struct ttm_resource_manager *vram_man = ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM); + + vf2pf_info = (struct amd_sriov_msg_vf2pf_info *) adev->virt.fw_reserve.p_vf2pf; + + if (adev->virt.fw_reserve.p_vf2pf == NULL) + return -EINVAL; + + memset(vf2pf_info, 0, sizeof(struct amd_sriov_msg_vf2pf_info)); + + vf2pf_info->header.size = sizeof(struct amd_sriov_msg_vf2pf_info); + vf2pf_info->header.version = AMD_SRIOV_MSG_FW_VRAM_VF2PF_VER; + +#ifdef MODULE + if (THIS_MODULE->version != NULL) + strcpy(vf2pf_info->driver_version, THIS_MODULE->version); + else +#endif + strcpy(vf2pf_info->driver_version, "N/A"); + + vf2pf_info->pf2vf_version_required = 0; // no requirement, guest understands all + vf2pf_info->driver_cert = 0; + vf2pf_info->os_info.all = 0; + + vf2pf_info->fb_usage = amdgpu_vram_mgr_usage(vram_man) >> 20; + vf2pf_info->fb_vis_usage = amdgpu_vram_mgr_vis_usage(vram_man) >> 20; + vf2pf_info->fb_size = adev->gmc.real_vram_size >> 20; + vf2pf_info->fb_vis_size = adev->gmc.visible_vram_size >> 20; + + amdgpu_virt_populate_vf2pf_ucode_info(adev); + + /* TODO: read dynamic info */ + vf2pf_info->gfx_usage = 0; + vf2pf_info->compute_usage = 0; + vf2pf_info->encode_usage = 0; + vf2pf_info->decode_usage = 0; + + vf2pf_info->checksum = + amd_sriov_msg_checksum( + vf2pf_info, vf2pf_info->header.size, 0, 0); + + return 0; +} + +void amdgpu_virt_update_vf2pf_work_item(struct work_struct *work) +{ + struct amdgpu_device *adev = container_of(work, struct amdgpu_device, virt.vf2pf_work.work); + + amdgpu_virt_read_pf2vf_data(adev); + amdgpu_virt_write_vf2pf_data(adev); + + schedule_delayed_work(&(adev->virt.vf2pf_work), adev->virt.vf2pf_update_interval_ms); +} + +void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev) +{ + if (adev->virt.vf2pf_update_interval_ms != 0) { + DRM_INFO("clean up the vf2pf work item\n"); + flush_delayed_work(&adev->virt.vf2pf_work); + cancel_delayed_work_sync(&adev->virt.vf2pf_work); + } +} + +void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev) +{ uint64_t bp_block_offset = 0; uint32_t bp_block_size = 0; - struct amdgim_pf2vf_info_v2 *pf2vf_v2 = NULL; + struct amd_sriov_msg_pf2vf_info *pf2vf_v2 = NULL; adev->virt.fw_reserve.p_pf2vf = NULL; adev->virt.fw_reserve.p_vf2pf = NULL; + adev->virt.vf2pf_update_interval_ms = 0; if (adev->mman.fw_vram_usage_va != NULL) { + adev->virt.vf2pf_update_interval_ms = 2000; + adev->virt.fw_reserve.p_pf2vf = - (struct amd_sriov_msg_pf2vf_info_header *)( - adev->mman.fw_vram_usage_va + AMDGIM_DATAEXCHANGE_OFFSET); - AMDGPU_FW_VRAM_PF2VF_READ(adev, header.size, &pf2vf_size); - AMDGPU_FW_VRAM_PF2VF_READ(adev, checksum, &checksum); - AMDGPU_FW_VRAM_PF2VF_READ(adev, feature_flags, &adev->virt.gim_feature); - - /* pf2vf message must be in 4K */ - if (pf2vf_size > 0 && pf2vf_size < 4096) { - if (adev->virt.fw_reserve.p_pf2vf->version == 2) { - pf2vf_v2 = (struct amdgim_pf2vf_info_v2 *)adev->virt.fw_reserve.p_pf2vf; - bp_block_offset = ((uint64_t)pf2vf_v2->bp_block_offset_L & 0xFFFFFFFF) | - ((((uint64_t)pf2vf_v2->bp_block_offset_H) << 32) & 0xFFFFFFFF00000000); + (struct amd_sriov_msg_pf2vf_info_header *) + (adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10)); + adev->virt.fw_reserve.p_vf2pf = + (struct amd_sriov_msg_vf2pf_info_header *) + (adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB << 10)); + + amdgpu_virt_read_pf2vf_data(adev); + amdgpu_virt_write_vf2pf_data(adev); + + /* bad page handling for version 2 */ + if (adev->virt.fw_reserve.p_pf2vf->version == 2) { + pf2vf_v2 = (struct amd_sriov_msg_pf2vf_info *)adev->virt.fw_reserve.p_pf2vf; + + bp_block_offset = ((uint64_t)pf2vf_v2->bp_block_offset_low & 0xFFFFFFFF) | + ((((uint64_t)pf2vf_v2->bp_block_offset_high) << 32) & 0xFFFFFFFF00000000); bp_block_size = pf2vf_v2->bp_block_size; if (bp_block_size && !adev->virt.ras_init_done) @@ -450,37 +611,11 @@ void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev) if (adev->virt.ras_init_done) amdgpu_virt_add_bad_page(adev, bp_block_offset, bp_block_size); } + } - checkval = amdgpu_virt_fw_reserve_get_checksum( - adev->virt.fw_reserve.p_pf2vf, pf2vf_size, - adev->virt.fw_reserve.checksum_key, checksum); - if (checkval == checksum) { - adev->virt.fw_reserve.p_vf2pf = - ((void *)adev->virt.fw_reserve.p_pf2vf + - pf2vf_size); - memset((void *)adev->virt.fw_reserve.p_vf2pf, 0, - sizeof(amdgim_vf2pf_info)); - AMDGPU_FW_VRAM_VF2PF_WRITE(adev, header.version, - AMDGPU_FW_VRAM_VF2PF_VER); - AMDGPU_FW_VRAM_VF2PF_WRITE(adev, header.size, - sizeof(amdgim_vf2pf_info)); - AMDGPU_FW_VRAM_VF2PF_READ(adev, driver_version, - &str); -#ifdef MODULE - if (THIS_MODULE->version != NULL) - strcpy(str, THIS_MODULE->version); - else -#endif - strcpy(str, "N/A"); - AMDGPU_FW_VRAM_VF2PF_WRITE(adev, driver_cert, - 0); - AMDGPU_FW_VRAM_VF2PF_WRITE(adev, checksum, - amdgpu_virt_fw_reserve_get_checksum( - adev->virt.fw_reserve.p_vf2pf, - pf2vf_size, - adev->virt.fw_reserve.checksum_key, 0)); - } - } + if (adev->virt.vf2pf_update_interval_ms != 0) { + INIT_DELAYED_WORK(&adev->virt.vf2pf_work, amdgpu_virt_update_vf2pf_work_item); + schedule_delayed_work(&(adev->virt.vf2pf_work), adev->virt.vf2pf_update_interval_ms); } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index b2046c3a404d..8dd624c20f89 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -24,6 +24,8 @@ #ifndef AMDGPU_VIRT_H #define AMDGPU_VIRT_H +#include "amdgv_sriovmsg.h" + #define AMDGPU_SRIOV_CAPS_SRIOV_VBIOS (1 << 0) /* vBIOS is sr-iov ready */ #define AMDGPU_SRIOV_CAPS_ENABLE_IOV (1 << 1) /* sr-iov is enabled on this GPU */ #define AMDGPU_SRIOV_CAPS_IS_VF (1 << 2) /* this GPU is a virtual function */ @@ -79,7 +81,10 @@ struct amdgpu_virt_fw_reserve { struct amd_sriov_msg_vf2pf_info_header *p_vf2pf; unsigned int checksum_key; }; + /* + * Legacy GIM header + * * Defination between PF and VF * Structures forcibly aligned to 4 to keep the same style as PF. */ @@ -101,15 +106,7 @@ enum AMDGIM_FEATURE_FLAG { AMDGIM_FEATURE_PP_ONE_VF = (1 << 4), }; -struct amd_sriov_msg_pf2vf_info_header { - /* the total structure size in byte. */ - uint32_t size; - /* version of this structure, written by the GIM */ - uint32_t version; - /* reserved */ - uint32_t reserved[2]; -} __aligned(4); -struct amdgim_pf2vf_info_v1 { +struct amdgim_pf2vf_info_v1 { /* header contains size and version */ struct amd_sriov_msg_pf2vf_info_header header; /* max_width * max_height */ @@ -128,54 +125,6 @@ struct amdgim_pf2vf_info_v1 { unsigned int checksum; } __aligned(4); -struct amdgim_pf2vf_info_v2 { - /* header contains size and version */ - struct amd_sriov_msg_pf2vf_info_header header; - /* use private key from mailbox 2 to create chueksum */ - uint32_t checksum; - /* The features flags of the GIM driver supports. */ - uint32_t feature_flags; - /* max_width * max_height */ - uint32_t uvd_enc_max_pixels_count; - /* 16x16 pixels/sec, codec independent */ - uint32_t uvd_enc_max_bandwidth; - /* max_width * max_height */ - uint32_t vce_enc_max_pixels_count; - /* 16x16 pixels/sec, codec independent */ - uint32_t vce_enc_max_bandwidth; - /* Bad pages block position in BYTE */ - uint32_t bp_block_offset_L; - uint32_t bp_block_offset_H; - /* Bad pages block size in BYTE */ - uint32_t bp_block_size; - /* MEC FW position in kb from the start of VF visible frame buffer */ - uint32_t mecfw_kboffset_L; - uint32_t mecfw_kboffset_H; - /* MEC FW size in KB */ - uint32_t mecfw_ksize; - /* UVD FW position in kb from the start of VF visible frame buffer */ - uint32_t uvdfw_kboffset_L; - uint32_t uvdfw_kboffset_H; - /* UVD FW size in KB */ - uint32_t uvdfw_ksize; - /* VCE FW position in kb from the start of VF visible frame buffer */ - uint32_t vcefw_kboffset_L; - uint32_t vcefw_kboffset_H; - /* VCE FW size in KB */ - uint32_t vcefw_ksize; - uint32_t reserved[AMDGIM_GET_STRUCTURE_RESERVED_SIZE(256, 0, 0, (18 + sizeof(struct amd_sriov_msg_pf2vf_info_header)/sizeof(uint32_t)), 0)]; -} __aligned(4); - - -struct amd_sriov_msg_vf2pf_info_header { - /* the total structure size in byte. */ - uint32_t size; - /*version of this structure, written by the guest */ - uint32_t version; - /* reserved */ - uint32_t reserved[2]; -} __aligned(4); - struct amdgim_vf2pf_info_v1 { /* header contains size and version */ struct amd_sriov_msg_vf2pf_info_header header; @@ -237,31 +186,6 @@ struct amdgim_vf2pf_info_v2 { uint32_t reserved[AMDGIM_GET_STRUCTURE_RESERVED_SIZE(256, 64, 0, (12 + sizeof(struct amd_sriov_msg_vf2pf_info_header)/sizeof(uint32_t)), 0)]; } __aligned(4); -#define AMDGPU_FW_VRAM_VF2PF_VER 2 -typedef struct amdgim_vf2pf_info_v2 amdgim_vf2pf_info ; - -#define AMDGPU_FW_VRAM_VF2PF_WRITE(adev, field, val) \ - do { \ - ((amdgim_vf2pf_info *)adev->virt.fw_reserve.p_vf2pf)->field = (val); \ - } while (0) - -#define AMDGPU_FW_VRAM_VF2PF_READ(adev, field, val) \ - do { \ - (*val) = ((amdgim_vf2pf_info *)adev->virt.fw_reserve.p_vf2pf)->field; \ - } while (0) - -#define AMDGPU_FW_VRAM_PF2VF_READ(adev, field, val) \ - do { \ - if (!adev->virt.fw_reserve.p_pf2vf) \ - *(val) = 0; \ - else { \ - if (adev->virt.fw_reserve.p_pf2vf->version == 1) \ - *(val) = ((struct amdgim_pf2vf_info_v1 *)adev->virt.fw_reserve.p_pf2vf)->field; \ - if (adev->virt.fw_reserve.p_pf2vf->version == 2) \ - *(val) = ((struct amdgim_pf2vf_info_v2 *)adev->virt.fw_reserve.p_pf2vf)->field; \ - } \ - } while (0) - struct amdgpu_virt_ras_err_handler_data { /* point to bad page records array */ struct eeprom_table_record *bps; @@ -285,7 +209,7 @@ struct amdgpu_virt { struct work_struct flr_work; struct amdgpu_mm_table mm_table; const struct amdgpu_virt_ops *ops; - struct amdgpu_vf_error_buffer vf_errors; + struct amdgpu_vf_error_buffer vf_errors; struct amdgpu_virt_fw_reserve fw_reserve; uint32_t gim_feature; uint32_t reg_access_mode; @@ -293,6 +217,10 @@ struct amdgpu_virt { bool tdr_debug; struct amdgpu_virt_ras_err_handler_data *virt_eh_data; bool ras_init_done; + + /* vf2pf message */ + struct delayed_work vf2pf_work; + uint32_t vf2pf_update_interval_ms; }; #define amdgpu_sriov_enabled(adev) \ @@ -341,11 +269,9 @@ void amdgpu_virt_request_init_data(struct amdgpu_device *adev); int amdgpu_virt_wait_reset(struct amdgpu_device *adev); int amdgpu_virt_alloc_mm_table(struct amdgpu_device *adev); void amdgpu_virt_free_mm_table(struct amdgpu_device *adev); -int amdgpu_virt_fw_reserve_get_checksum(void *obj, unsigned long obj_size, - unsigned int key, - unsigned int chksum); void amdgpu_virt_release_ras_err_handler_data(struct amdgpu_device *adev); void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev); +void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev); void amdgpu_detect_virtualization(struct amdgpu_device *adev); bool amdgpu_virt_can_access_debugfs(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index f0e6fafd0938..38b59a4fc04c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -1502,6 +1502,8 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params, pt = cursor.entry->base.bo; shift = parent_shift; + frag_end = max(frag_end, ALIGN(frag_start + 1, + 1ULL << shift)); } /* Looks good so far, calculate parameters for the update */ @@ -1513,19 +1515,26 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params, entry_end = min(entry_end, end); do { + struct amdgpu_vm *vm = params->vm; uint64_t upd_end = min(entry_end, frag_end); unsigned nptes = (upd_end - frag_start) >> shift; + uint64_t upd_flags = flags | AMDGPU_PTE_FRAG(frag); /* This can happen when we set higher level PDs to * silent to stop fault floods. */ nptes = max(nptes, 1u); + + trace_amdgpu_vm_update_ptes(params, frag_start, upd_end, + nptes, dst, incr, upd_flags, + vm->task_info.pid, + vm->immediate.fence_context); amdgpu_vm_update_flags(params, pt, cursor.level, pe_start, dst, nptes, incr, - flags | AMDGPU_PTE_FRAG(frag)); + upd_flags); pe_start += nptes * 8; - dst += (uint64_t)nptes * AMDGPU_GPU_PAGE_SIZE << shift; + dst += nptes * incr; frag_start = upd_end; if (frag_start >= frag_end) { @@ -2795,7 +2804,7 @@ long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout) * 0 for success, error for failure. */ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, - int vm_context, unsigned int pasid) + int vm_context, u32 pasid) { struct amdgpu_bo_param bp; struct amdgpu_bo *root; @@ -2966,7 +2975,7 @@ static int amdgpu_vm_check_clean_reserved(struct amdgpu_device *adev, * 0 for success, -errno for errors. */ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm, - unsigned int pasid) + u32 pasid) { bool pte_support_ats = (adev->asic_type == CHIP_RAVEN); int r; @@ -3264,7 +3273,7 @@ int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) * @pasid: PASID identifier for VM * @task_info: task_info to fill. */ -void amdgpu_vm_get_task_info(struct amdgpu_device *adev, unsigned int pasid, +void amdgpu_vm_get_task_info(struct amdgpu_device *adev, u32 pasid, struct amdgpu_task_info *task_info) { struct amdgpu_vm *vm; @@ -3308,7 +3317,7 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm) * Try to gracefully handle a VM fault. Return true if the fault was handled and * shouldn't be reported any more. */ -bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, unsigned int pasid, +bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, uint64_t addr) { struct amdgpu_bo *root; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 770025a5e500..58c83a7ad0fd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -98,7 +98,7 @@ struct amdgpu_bo_list_entry; #define AMDGPU_PTE_MTYPE_NV10(a) ((uint64_t)(a) << 48) #define AMDGPU_PTE_MTYPE_NV10_MASK AMDGPU_PTE_MTYPE_NV10(7ULL) -/* How to programm VM fault handling */ +/* How to program VM fault handling */ #define AMDGPU_VM_FAULT_STOP_NEVER 0 #define AMDGPU_VM_FAULT_STOP_FIRST 1 #define AMDGPU_VM_FAULT_STOP_ALWAYS 2 @@ -112,8 +112,8 @@ struct amdgpu_bo_list_entry; #define AMDGPU_MMHUB_0 1 #define AMDGPU_MMHUB_1 2 -/* hardcode that limit for now */ -#define AMDGPU_VA_RESERVED_SIZE (1ULL << 20) +/* Reserve 2MB at top/bottom of address space for kernel use */ +#define AMDGPU_VA_RESERVED_SIZE (2ULL << 20) /* max vmids dedicated for process */ #define AMDGPU_VM_MAX_RESERVED_VMID 1 @@ -372,8 +372,8 @@ void amdgpu_vm_manager_fini(struct amdgpu_device *adev); long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout); int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, - int vm_context, unsigned int pasid); -int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm, unsigned int pasid); + int vm_context, u32 pasid); +int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm, u32 pasid); void amdgpu_vm_release_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm); void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm); void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm, @@ -430,9 +430,9 @@ bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring, struct amdgpu_job *job); void amdgpu_vm_check_compute_bug(struct amdgpu_device *adev); -void amdgpu_vm_get_task_info(struct amdgpu_device *adev, unsigned int pasid, +void amdgpu_vm_get_task_info(struct amdgpu_device *adev, u32 pasid, struct amdgpu_task_info *task_info); -bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, unsigned int pasid, +bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, uint64_t addr); void amdgpu_vm_set_task_info(struct amdgpu_vm *vm); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c index 39c704a1fb0e..0786e7555554 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c @@ -59,7 +59,7 @@ static int amdgpu_vm_cpu_prepare(struct amdgpu_vm_update_params *p, * * @p: see amdgpu_vm_update_params definition * @bo: PD/PT to update - * @pe: kmap addr of the page entry + * @pe: byte offset of the PDE/PTE, relative to start of PDB/PTB * @addr: dst addr to write into pe * @count: number of page entries to update * @incr: increase next addr by incr bytes diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c index 189d46ea603b..db790574dc2e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c @@ -155,7 +155,7 @@ static void amdgpu_vm_sdma_copy_ptes(struct amdgpu_vm_update_params *p, * * @p: see amdgpu_vm_update_params definition * @bo: PD/PT to update - * @pe: addr of the page entry + * @pe: byte offset of the PDE/PTE, relative to start of PDB/PTB * @addr: dst addr to write into pe * @count: number of page entries to update * @incr: increase next addr by incr bytes @@ -187,7 +187,7 @@ static void amdgpu_vm_sdma_set_ptes(struct amdgpu_vm_update_params *p, * * @p: see amdgpu_vm_update_params definition * @bo: PD/PT to update - * @pe: addr of the page entry + * @pe: byte offset of the PDE/PTE, relative to start of PDB/PTB * @addr: dst addr to write into pe * @count: number of page entries to update * @incr: increase next addr by incr bytes diff --git a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h new file mode 100644 index 000000000000..5355827ed0ae --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h @@ -0,0 +1,276 @@ +/* + * Copyright 2018-2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef AMDGV_SRIOV_MSG__H_ +#define AMDGV_SRIOV_MSG__H_ + +/* unit in kilobytes */ +#define AMD_SRIOV_MSG_VBIOS_OFFSET 0 +#define AMD_SRIOV_MSG_VBIOS_SIZE_KB 64 +#define AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB AMD_SRIOV_MSG_VBIOS_SIZE_KB +#define AMD_SRIOV_MSG_DATAEXCHANGE_SIZE_KB 4 + +/* + * layout + * 0 64KB 65KB 66KB + * | VBIOS | PF2VF | VF2PF | Bad Page | ... + * | 64KB | 1KB | 1KB | + */ +#define AMD_SRIOV_MSG_SIZE_KB 1 +#define AMD_SRIOV_MSG_PF2VF_OFFSET_KB AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB +#define AMD_SRIOV_MSG_VF2PF_OFFSET_KB (AMD_SRIOV_MSG_PF2VF_OFFSET_KB + AMD_SRIOV_MSG_SIZE_KB) +#define AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB (AMD_SRIOV_MSG_VF2PF_OFFSET_KB + AMD_SRIOV_MSG_SIZE_KB) + +/* + * PF2VF history log: + * v1 defined in amdgim + * v2 current + * + * VF2PF history log: + * v1 defined in amdgim + * v2 defined in amdgim + * v3 current + */ +#define AMD_SRIOV_MSG_FW_VRAM_PF2VF_VER 2 +#define AMD_SRIOV_MSG_FW_VRAM_VF2PF_VER 3 + +#define AMD_SRIOV_MSG_RESERVE_UCODE 24 + +enum amd_sriov_ucode_engine_id { + AMD_SRIOV_UCODE_ID_VCE = 0, + AMD_SRIOV_UCODE_ID_UVD, + AMD_SRIOV_UCODE_ID_MC, + AMD_SRIOV_UCODE_ID_ME, + AMD_SRIOV_UCODE_ID_PFP, + AMD_SRIOV_UCODE_ID_CE, + AMD_SRIOV_UCODE_ID_RLC, + AMD_SRIOV_UCODE_ID_RLC_SRLC, + AMD_SRIOV_UCODE_ID_RLC_SRLG, + AMD_SRIOV_UCODE_ID_RLC_SRLS, + AMD_SRIOV_UCODE_ID_MEC, + AMD_SRIOV_UCODE_ID_MEC2, + AMD_SRIOV_UCODE_ID_SOS, + AMD_SRIOV_UCODE_ID_ASD, + AMD_SRIOV_UCODE_ID_TA_RAS, + AMD_SRIOV_UCODE_ID_TA_XGMI, + AMD_SRIOV_UCODE_ID_SMC, + AMD_SRIOV_UCODE_ID_SDMA, + AMD_SRIOV_UCODE_ID_SDMA2, + AMD_SRIOV_UCODE_ID_VCN, + AMD_SRIOV_UCODE_ID_DMCU, + AMD_SRIOV_UCODE_ID__MAX +}; + +#pragma pack(push, 1) // PF2VF / VF2PF data areas are byte packed + +union amd_sriov_msg_feature_flags { + struct { + uint32_t error_log_collect : 1; + uint32_t host_load_ucodes : 1; + uint32_t host_flr_vramlost : 1; + uint32_t mm_bw_management : 1; + uint32_t pp_one_vf_mode : 1; + uint32_t reserved : 27; + } flags; + uint32_t all; +}; + +union amd_sriov_msg_os_info { + struct { + uint32_t windows : 1; + uint32_t reserved : 31; + } info; + uint32_t all; +}; + +struct amd_sriov_msg_pf2vf_info_header { + /* the total structure size in byte */ + uint32_t size; + /* version of this structure, written by the HOST */ + uint32_t version; + /* reserved */ + uint32_t reserved[2]; +}; + +struct amd_sriov_msg_pf2vf_info { + /* header contains size and version */ + struct amd_sriov_msg_pf2vf_info_header header; + /* use private key from mailbox 2 to create checksum */ + uint32_t checksum; + /* The features flags of the HOST driver supports */ + union amd_sriov_msg_feature_flags feature_flags; + /* (max_width * max_height * fps) / (16 * 16) */ + uint32_t hevc_enc_max_mb_per_second; + /* (max_width * max_height) / (16 * 16) */ + uint32_t hevc_enc_max_mb_per_frame; + /* (max_width * max_height * fps) / (16 * 16) */ + uint32_t avc_enc_max_mb_per_second; + /* (max_width * max_height) / (16 * 16) */ + uint32_t avc_enc_max_mb_per_frame; + /* MEC FW position in BYTE from the start of VF visible frame buffer */ + uint64_t mecfw_offset; + /* MEC FW size in BYTE */ + uint32_t mecfw_size; + /* UVD FW position in BYTE from the start of VF visible frame buffer */ + uint64_t uvdfw_offset; + /* UVD FW size in BYTE */ + uint32_t uvdfw_size; + /* VCE FW position in BYTE from the start of VF visible frame buffer */ + uint64_t vcefw_offset; + /* VCE FW size in BYTE */ + uint32_t vcefw_size; + /* Bad pages block position in BYTE */ + uint32_t bp_block_offset_low; + uint32_t bp_block_offset_high; + /* Bad pages block size in BYTE */ + uint32_t bp_block_size; + /* frequency for VF to update the VF2PF area in msec, 0 = manual */ + uint32_t vf2pf_update_interval_ms; + /* identification in ROCm SMI */ + uint64_t uuid; + uint32_t fcn_idx; + /* reserved */ + uint32_t reserved[256-26]; +}; + +struct amd_sriov_msg_vf2pf_info_header { + /* the total structure size in byte */ + uint32_t size; + /* version of this structure, written by the guest */ + uint32_t version; + /* reserved */ + uint32_t reserved[2]; +}; + +struct amd_sriov_msg_vf2pf_info { + /* header contains size and version */ + struct amd_sriov_msg_vf2pf_info_header header; + uint32_t checksum; + /* driver version */ + uint8_t driver_version[64]; + /* driver certification, 1=WHQL, 0=None */ + uint32_t driver_cert; + /* guest OS type and version */ + union amd_sriov_msg_os_info os_info; + /* guest fb information in the unit of MB */ + uint32_t fb_usage; + /* guest gfx engine usage percentage */ + uint32_t gfx_usage; + /* guest gfx engine health percentage */ + uint32_t gfx_health; + /* guest compute engine usage percentage */ + uint32_t compute_usage; + /* guest compute engine health percentage */ + uint32_t compute_health; + /* guest avc engine usage percentage. 0xffff means N/A */ + uint32_t avc_enc_usage; + /* guest avc engine health percentage. 0xffff means N/A */ + uint32_t avc_enc_health; + /* guest hevc engine usage percentage. 0xffff means N/A */ + uint32_t hevc_enc_usage; + /* guest hevc engine usage percentage. 0xffff means N/A */ + uint32_t hevc_enc_health; + /* combined encode/decode usage */ + uint32_t encode_usage; + uint32_t decode_usage; + /* Version of PF2VF that VF understands */ + uint32_t pf2vf_version_required; + /* additional FB usage */ + uint32_t fb_vis_usage; + uint32_t fb_vis_size; + uint32_t fb_size; + /* guest ucode data, each one is 1.25 Dword */ + struct { + uint8_t id; + uint32_t version; + } ucode_info[AMD_SRIOV_MSG_RESERVE_UCODE]; + + /* reserved */ + uint32_t reserved[256-68]; +}; + +/* mailbox message send from guest to host */ +enum amd_sriov_mailbox_request_message { + MB_REQ_MSG_REQ_GPU_INIT_ACCESS = 1, + MB_REQ_MSG_REL_GPU_INIT_ACCESS, + MB_REQ_MSG_REQ_GPU_FINI_ACCESS, + MB_REQ_MSG_REL_GPU_FINI_ACCESS, + MB_REQ_MSG_REQ_GPU_RESET_ACCESS, + MB_REQ_MSG_REQ_GPU_INIT_DATA, + + MB_REQ_MSG_LOG_VF_ERROR = 200, +}; + +/* mailbox message send from host to guest */ +enum amd_sriov_mailbox_response_message { + MB_RES_MSG_CLR_MSG_BUF = 0, + MB_RES_MSG_READY_TO_ACCESS_GPU = 1, + MB_RES_MSG_FLR_NOTIFICATION, + MB_RES_MSG_FLR_NOTIFICATION_COMPLETION, + MB_RES_MSG_SUCCESS, + MB_RES_MSG_FAIL, + MB_RES_MSG_QUERY_ALIVE, + MB_RES_MSG_GPU_INIT_DATA_READY, + + MB_RES_MSG_TEXT_MESSAGE = 255 +}; + +/* version data stored in MAILBOX_MSGBUF_RCV_DW1 for future expansion */ +enum amd_sriov_gpu_init_data_version { + GPU_INIT_DATA_READY_V1 = 1, +}; + +#pragma pack(pop) // Restore previous packing option + +/* checksum function between host and guest */ +unsigned int amd_sriov_msg_checksum(void *obj, + unsigned long obj_size, + unsigned int key, + unsigned int checksum); + +/* assertion at compile time */ +#ifdef __linux__ +#define stringification(s) _stringification(s) +#define _stringification(s) #s + +_Static_assert( + sizeof(struct amd_sriov_msg_vf2pf_info) == AMD_SRIOV_MSG_SIZE_KB << 10, + "amd_sriov_msg_vf2pf_info must be " stringification(AMD_SRIOV_MSG_SIZE_KB) " KB"); + +_Static_assert( + sizeof(struct amd_sriov_msg_pf2vf_info) == AMD_SRIOV_MSG_SIZE_KB << 10, + "amd_sriov_msg_pf2vf_info must be " stringification(AMD_SRIOV_MSG_SIZE_KB) " KB"); + +_Static_assert( + AMD_SRIOV_MSG_RESERVE_UCODE % 4 == 0, + "AMD_SRIOV_MSG_RESERVE_UCODE must be multiple of 4"); + +_Static_assert( + AMD_SRIOV_MSG_RESERVE_UCODE > AMD_SRIOV_UCODE_ID__MAX, + "AMD_SRIOV_MSG_RESERVE_UCODE must be bigger than AMD_SRIOV_UCODE_ID__MAX"); + +#undef _stringification +#undef stringification +#endif + +#endif /* AMDGV_SRIOV_MSG__H_ */ diff --git a/drivers/gpu/drm/amd/amdgpu/cik_ih.c b/drivers/gpu/drm/amd/amdgpu/cik_ih.c index 401c99f0b2d0..db953e95f3d2 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/cik_ih.c @@ -316,14 +316,9 @@ static int cik_ih_sw_fini(void *handle) static int cik_ih_hw_init(void *handle) { - int r; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - r = cik_ih_irq_init(adev); - if (r) - return r; - - return 0; + return cik_ih_irq_init(adev); } static int cik_ih_hw_fini(void *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c index cc93577dee03..b4d4b76538d2 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c @@ -47,6 +47,9 @@ static void dce_virtual_set_display_funcs(struct amdgpu_device *adev); static void dce_virtual_set_irq_funcs(struct amdgpu_device *adev); static int dce_virtual_connector_encoder_init(struct amdgpu_device *adev, int index); +static int dce_virtual_pageflip(struct amdgpu_device *adev, + unsigned crtc_id); +static enum hrtimer_restart dce_virtual_vblank_timer_handle(struct hrtimer *vblank_timer); static void dce_virtual_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev, int crtc, enum amdgpu_interrupt_state state); @@ -171,8 +174,10 @@ static void dce_virtual_crtc_commit(struct drm_crtc *crtc) static void dce_virtual_crtc_disable(struct drm_crtc *crtc) { struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc); + struct drm_device *dev = crtc->dev; - drm_crtc_vblank_off(crtc); + if (dev->num_crtcs) + drm_crtc_vblank_off(crtc); amdgpu_crtc->enabled = false; amdgpu_crtc->pll_id = ATOM_PPLL_INVALID; @@ -247,6 +252,11 @@ static int dce_virtual_crtc_init(struct amdgpu_device *adev, int index) amdgpu_crtc->vsync_timer_enabled = AMDGPU_IRQ_STATE_DISABLE; drm_crtc_helper_add(&amdgpu_crtc->base, &dce_virtual_crtc_helper_funcs); + hrtimer_init(&amdgpu_crtc->vblank_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + hrtimer_set_expires(&amdgpu_crtc->vblank_timer, DCE_VIRTUAL_VBLANK_PERIOD); + amdgpu_crtc->vblank_timer.function = dce_virtual_vblank_timer_handle; + hrtimer_start(&amdgpu_crtc->vblank_timer, + DCE_VIRTUAL_VBLANK_PERIOD, HRTIMER_MODE_REL); return 0; } @@ -476,7 +486,7 @@ static int dce_virtual_hw_fini(void *handle) for (i = 0; i<adev->mode_info.num_crtc; i++) if (adev->mode_info.crtcs[i]) - dce_virtual_set_crtc_vblank_interrupt_state(adev, i, AMDGPU_IRQ_STATE_DISABLE); + hrtimer_cancel(&adev->mode_info.crtcs[i]->vblank_timer); return 0; } @@ -698,9 +708,15 @@ static enum hrtimer_restart dce_virtual_vblank_timer_handle(struct hrtimer *vbla struct amdgpu_crtc, vblank_timer); struct drm_device *ddev = amdgpu_crtc->base.dev; struct amdgpu_device *adev = drm_to_adev(ddev); + struct amdgpu_irq_src *source = adev->irq.client[AMDGPU_IRQ_CLIENTID_LEGACY].sources + [VISLANDS30_IV_SRCID_SMU_DISP_TIMER2_TRIGGER]; + int irq_type = amdgpu_display_crtc_idx_to_irq_type(adev, + amdgpu_crtc->crtc_id); - drm_handle_vblank(ddev, amdgpu_crtc->crtc_id); - dce_virtual_pageflip(adev, amdgpu_crtc->crtc_id); + if (amdgpu_irq_enabled(adev, source, irq_type)) { + drm_handle_vblank(ddev, amdgpu_crtc->crtc_id); + dce_virtual_pageflip(adev, amdgpu_crtc->crtc_id); + } hrtimer_start(vblank_timer, DCE_VIRTUAL_VBLANK_PERIOD, HRTIMER_MODE_REL); @@ -716,21 +732,6 @@ static void dce_virtual_set_crtc_vblank_interrupt_state(struct amdgpu_device *ad return; } - if (state && !adev->mode_info.crtcs[crtc]->vsync_timer_enabled) { - DRM_DEBUG("Enable software vsync timer\n"); - hrtimer_init(&adev->mode_info.crtcs[crtc]->vblank_timer, - CLOCK_MONOTONIC, HRTIMER_MODE_REL); - hrtimer_set_expires(&adev->mode_info.crtcs[crtc]->vblank_timer, - DCE_VIRTUAL_VBLANK_PERIOD); - adev->mode_info.crtcs[crtc]->vblank_timer.function = - dce_virtual_vblank_timer_handle; - hrtimer_start(&adev->mode_info.crtcs[crtc]->vblank_timer, - DCE_VIRTUAL_VBLANK_PERIOD, HRTIMER_MODE_REL); - } else if (!state && adev->mode_info.crtcs[crtc]->vsync_timer_enabled) { - DRM_DEBUG("Disable software vsync timer\n"); - hrtimer_cancel(&adev->mode_info.crtcs[crtc]->vblank_timer); - } - adev->mode_info.crtcs[crtc]->vsync_timer_enabled = state; DRM_DEBUG("[FM]set crtc %d vblank interrupt state %d\n", crtc, state); } diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c index 2eab808fffeb..7b89fd2aa44a 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c @@ -455,7 +455,8 @@ static int df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev, uint32_t *lo_base_addr, uint32_t *hi_base_addr, uint32_t *lo_val, - uint32_t *hi_val) + uint32_t *hi_val, + bool is_enable) { uint32_t eventsel, instance, unitmask; @@ -477,7 +478,8 @@ static int df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev, instance_5432 = (instance >> 2) & 0xf; instance_76 = (instance >> 6) & 0x3; - *lo_val = (unitmask << 8) | (instance_10 << 6) | eventsel | (1 << 22); + *lo_val = (unitmask << 8) | (instance_10 << 6) | eventsel; + *lo_val = is_enable ? *lo_val | (1 << 22) : *lo_val & ~(1 << 22); *hi_val = (instance_76 << 29) | instance_5432; DRM_DEBUG_DRIVER("config=%llx addr=%08x:%08x val=%08x:%08x", @@ -572,14 +574,14 @@ static void df_v3_6_reset_perfmon_cntr(struct amdgpu_device *adev, } static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config, - int is_enable) + int is_add) { uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val; int err = 0, ret = 0; switch (adev->asic_type) { case CHIP_VEGA20: - if (is_enable) + if (is_add) return df_v3_6_pmc_add_cntr(adev, config); df_v3_6_reset_perfmon_cntr(adev, config); @@ -589,7 +591,8 @@ static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config, &lo_base_addr, &hi_base_addr, &lo_val, - &hi_val); + &hi_val, + true); if (ret) return ret; @@ -612,7 +615,7 @@ static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config, } static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config, - int is_disable) + int is_remove) { uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val; int ret = 0; @@ -624,15 +627,17 @@ static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config, &lo_base_addr, &hi_base_addr, &lo_val, - &hi_val); + &hi_val, + false); if (ret) return ret; - df_v3_6_reset_perfmon_cntr(adev, config); - if (is_disable) + if (is_remove) { + df_v3_6_reset_perfmon_cntr(adev, config); df_v3_6_pmc_release_cntr(adev, config); + } break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index d502e30f67d9..56fdbe626d30 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -112,6 +112,22 @@ #define mmCP_HYP_ME_UCODE_DATA 0x5817 #define mmCP_HYP_ME_UCODE_DATA_BASE_IDX 1 +//CC_GC_SA_UNIT_DISABLE +#define mmCC_GC_SA_UNIT_DISABLE 0x0fe9 +#define mmCC_GC_SA_UNIT_DISABLE_BASE_IDX 0 +#define CC_GC_SA_UNIT_DISABLE__SA_DISABLE__SHIFT 0x8 +#define CC_GC_SA_UNIT_DISABLE__SA_DISABLE_MASK 0x0000FF00L +//GC_USER_SA_UNIT_DISABLE +#define mmGC_USER_SA_UNIT_DISABLE 0x0fea +#define mmGC_USER_SA_UNIT_DISABLE_BASE_IDX 0 +#define GC_USER_SA_UNIT_DISABLE__SA_DISABLE__SHIFT 0x8 +#define GC_USER_SA_UNIT_DISABLE__SA_DISABLE_MASK 0x0000FF00L +//PA_SC_ENHANCE_3 +#define mmPA_SC_ENHANCE_3 0x1085 +#define mmPA_SC_ENHANCE_3_BASE_IDX 0 +#define PA_SC_ENHANCE_3__FORCE_PBB_WORKLOAD_MODE_TO_ZERO__SHIFT 0x3 +#define PA_SC_ENHANCE_3__FORCE_PBB_WORKLOAD_MODE_TO_ZERO_MASK 0x00000008L + MODULE_FIRMWARE("amdgpu/navi10_ce.bin"); MODULE_FIRMWARE("amdgpu/navi10_pfp.bin"); MODULE_FIRMWARE("amdgpu/navi10_me.bin"); @@ -3091,6 +3107,7 @@ static const struct soc15_reg_golden golden_settings_gc_10_3[] = SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf), SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008), SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x10f80988), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG, 0x00000020, 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), @@ -3188,6 +3205,8 @@ static int gfx_v10_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev); static void gfx_v10_0_ring_emit_ce_meta(struct amdgpu_ring *ring, bool resume); static void gfx_v10_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume); static void gfx_v10_0_ring_emit_frame_cntl(struct amdgpu_ring *ring, bool start, bool secure); +static u32 gfx_v10_3_get_disabled_sa(struct amdgpu_device *adev); +static void gfx_v10_3_program_pbb_mode(struct amdgpu_device *adev); static void gfx10_kiq_set_resources(struct amdgpu_ring *kiq_ring, uint64_t queue_mask) { @@ -3560,7 +3579,7 @@ static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev) break; } - if (adev->gfx.cp_fw_write_wait == false) + if (!adev->gfx.cp_fw_write_wait) DRM_WARN_ONCE("CP firmware version too old, please update!"); } @@ -3586,6 +3605,17 @@ static void gfx_v10_0_init_rlc_ext_microcode(struct amdgpu_device *adev) le32_to_cpu(rlc_hdr->reg_list_format_direct_reg_list_length); } +static void gfx_v10_0_init_rlc_iram_dram_microcode(struct amdgpu_device *adev) +{ + const struct rlc_firmware_header_v2_2 *rlc_hdr; + + rlc_hdr = (const struct rlc_firmware_header_v2_2 *)adev->gfx.rlc_fw->data; + adev->gfx.rlc.rlc_iram_ucode_size_bytes = le32_to_cpu(rlc_hdr->rlc_iram_ucode_size_bytes); + adev->gfx.rlc.rlc_iram_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->rlc_iram_ucode_offset_bytes); + adev->gfx.rlc.rlc_dram_ucode_size_bytes = le32_to_cpu(rlc_hdr->rlc_dram_ucode_size_bytes); + adev->gfx.rlc.rlc_dram_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->rlc_dram_ucode_offset_bytes); +} + static bool gfx_v10_0_navi10_gfxoff_should_enable(struct amdgpu_device *adev) { bool ret = false; @@ -3610,6 +3640,9 @@ static void gfx_v10_0_check_gfxoff_flag(struct amdgpu_device *adev) if (!gfx_v10_0_navi10_gfxoff_should_enable(adev)) adev->pm.pp_feature &= ~PP_GFXOFF_MASK; break; + case CHIP_NAVY_FLOUNDER: + adev->pm.pp_feature &= ~PP_GFXOFF_MASK; + break; default: break; } @@ -3698,8 +3731,6 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev) rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data; version_major = le16_to_cpu(rlc_hdr->header.header_version_major); version_minor = le16_to_cpu(rlc_hdr->header.header_version_minor); - if (version_major == 2 && version_minor == 1) - adev->gfx.rlc.is_rlc_v2_1 = true; adev->gfx.rlc_fw_version = le32_to_cpu(rlc_hdr->header.ucode_version); adev->gfx.rlc_feature_version = le32_to_cpu(rlc_hdr->ucode_feature_version); @@ -3741,8 +3772,12 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev) for (i = 0 ; i < (rlc_hdr->reg_list_size_bytes >> 2); i++) adev->gfx.rlc.register_restore[i] = le32_to_cpu(tmp[i]); - if (adev->gfx.rlc.is_rlc_v2_1) - gfx_v10_0_init_rlc_ext_microcode(adev); + if (version_major == 2) { + if (version_minor >= 1) + gfx_v10_0_init_rlc_ext_microcode(adev); + if (version_minor == 2) + gfx_v10_0_init_rlc_iram_dram_microcode(adev); + } } snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec%s.bin", chip_name, wks); @@ -3803,8 +3838,7 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev) adev->firmware.fw_size += ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE); } - if (adev->gfx.rlc.is_rlc_v2_1 && - adev->gfx.rlc.save_restore_list_cntl_size_bytes && + if (adev->gfx.rlc.save_restore_list_cntl_size_bytes && adev->gfx.rlc.save_restore_list_gpm_size_bytes && adev->gfx.rlc.save_restore_list_srm_size_bytes) { info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL]; @@ -3824,6 +3858,21 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev) info->fw = adev->gfx.rlc_fw; adev->firmware.fw_size += ALIGN(adev->gfx.rlc.save_restore_list_srm_size_bytes, PAGE_SIZE); + + if (adev->gfx.rlc.rlc_iram_ucode_size_bytes && + adev->gfx.rlc.rlc_dram_ucode_size_bytes) { + info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_IRAM]; + info->ucode_id = AMDGPU_UCODE_ID_RLC_IRAM; + info->fw = adev->gfx.rlc_fw; + adev->firmware.fw_size += + ALIGN(adev->gfx.rlc.rlc_iram_ucode_size_bytes, PAGE_SIZE); + + info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_DRAM]; + info->ucode_id = AMDGPU_UCODE_ID_RLC_DRAM; + info->fw = adev->gfx.rlc_fw; + adev->firmware.fw_size += + ALIGN(adev->gfx.rlc.rlc_dram_ucode_size_bytes, PAGE_SIZE); + } } info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1]; @@ -4533,12 +4582,17 @@ static void gfx_v10_0_setup_rb(struct amdgpu_device *adev) int i, j; u32 data; u32 active_rbs = 0; + u32 bitmap; u32 rb_bitmap_width_per_sh = adev->gfx.config.max_backends_per_se / adev->gfx.config.max_sh_per_se; mutex_lock(&adev->grbm_idx_mutex); for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { + bitmap = i * adev->gfx.config.max_sh_per_se + j; + if ((adev->asic_type == CHIP_SIENNA_CICHLID) && + ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1)) + continue; gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff); data = gfx_v10_0_get_rb_active_bitmap(adev); active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) * @@ -6947,6 +7001,9 @@ static int gfx_v10_0_hw_init(void *handle) if (r) return r; + if (adev->asic_type == CHIP_SIENNA_CICHLID) + gfx_v10_3_program_pbb_mode(adev); + return r; } @@ -6980,15 +7037,19 @@ static int gfx_v10_0_hw_fini(void *handle) amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0); amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0); + + if (!adev->in_pci_err_recovery) { #ifndef BRING_UP_DEBUG - if (amdgpu_async_gfx_ring) { - r = gfx_v10_0_kiq_disable_kgq(adev); - if (r) - DRM_ERROR("KGQ disable failed\n"); - } + if (amdgpu_async_gfx_ring) { + r = gfx_v10_0_kiq_disable_kgq(adev); + if (r) + DRM_ERROR("KGQ disable failed\n"); + } #endif - if (amdgpu_gfx_disable_kcq(adev)) - DRM_ERROR("KCQ disable failed\n"); + if (amdgpu_gfx_disable_kcq(adev)) + DRM_ERROR("KCQ disable failed\n"); + } + if (amdgpu_sriov_vf(adev)) { gfx_v10_0_cp_gfx_enable(adev, false); /* Program KIQ position of RLC_CP_SCHEDULERS during destroy */ @@ -8756,6 +8817,10 @@ static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev, mutex_lock(&adev->grbm_idx_mutex); for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { + bitmap = i * adev->gfx.config.max_sh_per_se + j; + if ((adev->asic_type == CHIP_SIENNA_CICHLID) && + ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1)) + continue; mask = 1; ao_bitmap = 0; counter = 0; @@ -8790,6 +8855,47 @@ static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev, return 0; } +static u32 gfx_v10_3_get_disabled_sa(struct amdgpu_device *adev) +{ + uint32_t efuse_setting, vbios_setting, disabled_sa, max_sa_mask; + + efuse_setting = RREG32_SOC15(GC, 0, mmCC_GC_SA_UNIT_DISABLE); + efuse_setting &= CC_GC_SA_UNIT_DISABLE__SA_DISABLE_MASK; + efuse_setting >>= CC_GC_SA_UNIT_DISABLE__SA_DISABLE__SHIFT; + + vbios_setting = RREG32_SOC15(GC, 0, mmGC_USER_SA_UNIT_DISABLE); + vbios_setting &= GC_USER_SA_UNIT_DISABLE__SA_DISABLE_MASK; + vbios_setting >>= GC_USER_SA_UNIT_DISABLE__SA_DISABLE__SHIFT; + + max_sa_mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_sh_per_se * + adev->gfx.config.max_shader_engines); + disabled_sa = efuse_setting | vbios_setting; + disabled_sa &= max_sa_mask; + + return disabled_sa; +} + +static void gfx_v10_3_program_pbb_mode(struct amdgpu_device *adev) +{ + uint32_t max_sa_per_se, max_sa_per_se_mask, max_shader_engines; + uint32_t disabled_sa_mask, se_index, disabled_sa_per_se; + + disabled_sa_mask = gfx_v10_3_get_disabled_sa(adev); + + max_sa_per_se = adev->gfx.config.max_sh_per_se; + max_sa_per_se_mask = (1 << max_sa_per_se) - 1; + max_shader_engines = adev->gfx.config.max_shader_engines; + + for (se_index = 0; max_shader_engines > se_index; se_index++) { + disabled_sa_per_se = disabled_sa_mask >> (se_index * max_sa_per_se); + disabled_sa_per_se &= max_sa_per_se_mask; + if (disabled_sa_per_se == max_sa_per_se_mask) { + WREG32_FIELD15(GC, 0, PA_SC_ENHANCE_3, FORCE_PBB_WORKLOAD_MODE_TO_ZERO, 1); + break; + } + } +} + const struct amdgpu_ip_block_version gfx_v10_0_ip_block = { .type = AMD_IP_BLOCK_TYPE_GFX, diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 93c63ff3b35e..6959aebae6d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -49,6 +49,7 @@ #include "amdgpu_ras.h" #include "gfx_v9_4.h" +#include "gfx_v9_0.h" #include "asic_reg/pwr/pwr_10_0_offset.h" #include "asic_reg/pwr/pwr_10_0_sh_mask.h" @@ -788,7 +789,6 @@ static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev); static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev, struct amdgpu_cu_info *cu_info); static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev); -static void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance); 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, @@ -2075,6 +2075,7 @@ static const struct amdgpu_gfx_funcs gfx_v9_4_gfx_funcs = { .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, }; static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) @@ -2196,7 +2197,6 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) static int gfx_v9_0_compute_ring_init(struct amdgpu_device *adev, int ring_id, int mec, int pipe, int queue) { - int r; unsigned irq_type; struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id]; unsigned int hw_prio; @@ -2221,13 +2221,8 @@ static int gfx_v9_0_compute_ring_init(struct amdgpu_device *adev, int ring_id, hw_prio = amdgpu_gfx_is_high_priority_compute_queue(adev, ring->queue) ? AMDGPU_GFX_PIPE_PRIO_HIGH : AMDGPU_GFX_PIPE_PRIO_NORMAL; /* type-2 packets are deprecated on MEC, use type-3 instead */ - r = amdgpu_ring_init(adev, ring, 1024, - &adev->gfx.eop_irq, irq_type, hw_prio); - if (r) - return r; - - - return 0; + return amdgpu_ring_init(adev, ring, 1024, + &adev->gfx.eop_irq, irq_type, hw_prio); } static int gfx_v9_0_sw_init(void *handle) @@ -2402,7 +2397,8 @@ static void gfx_v9_0_tiling_mode_table_init(struct amdgpu_device *adev) /* TODO */ } -static void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance) +void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, + u32 instance) { u32 data; @@ -2560,14 +2556,14 @@ static void gfx_v9_0_constants_init(struct amdgpu_device *adev) tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE, SH_MEM_ALIGNMENT_MODE_UNALIGNED); tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_DISABLE, - !!amdgpu_noretry); + !!adev->gmc.noretry); WREG32_SOC15_RLC(GC, 0, mmSH_MEM_CONFIG, tmp); WREG32_SOC15_RLC(GC, 0, mmSH_MEM_BASES, 0); } else { tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE, SH_MEM_ALIGNMENT_MODE_UNALIGNED); tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_DISABLE, - !!amdgpu_noretry); + !!adev->gmc.noretry); WREG32_SOC15_RLC(GC, 0, mmSH_MEM_CONFIG, tmp); tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE, (adev->gmc.private_aperture_start >> 48)); @@ -2800,7 +2796,7 @@ static void pwr_10_0_gfxip_control_over_cgpg(struct amdgpu_device *adev, uint32_t default_data = 0; default_data = data = RREG32(SOC15_REG_OFFSET(PWR, 0, mmPWR_MISC_CNTL_STATUS)); - if (enable == true) { + if (enable) { /* enable GFXIP control over CGPG */ data |= PWR_MISC_CNTL_STATUS__PWR_GFX_RLC_CGPG_EN_MASK; if(default_data != data) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h index fa5a3fbaf6ab..dfe8d4841f58 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h @@ -26,9 +26,7 @@ extern const struct amdgpu_ip_block_version gfx_v9_0_ip_block; -void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num); - -uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev); -int gfx_v9_0_get_cu_info(struct amdgpu_device *adev, struct amdgpu_cu_info *cu_info); +void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, + u32 instance); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c index bd85aed3523a..bc699d680ce8 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c @@ -992,3 +992,32 @@ int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, void *inject_if) return ret; } + +static const struct soc15_reg_entry gfx_v9_4_rdrsp_status_regs = + { SOC15_REG_ENTRY(GC, 0, mmGCEA_ERR_STATUS), 0, 1, 32 }; + +void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev) +{ + uint32_t i, j; + uint32_t reg_value; + + if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) + return; + + mutex_lock(&adev->grbm_idx_mutex); + + for (i = 0; i < gfx_v9_4_rdrsp_status_regs.se_num; i++) { + for (j = 0; j < gfx_v9_4_rdrsp_status_regs.instance; + j++) { + gfx_v9_4_select_se_sh(adev, i, 0, j); + reg_value = RREG32(SOC15_REG_ENTRY_OFFSET( + gfx_v9_4_rdrsp_status_regs)); + if (reg_value) + dev_warn(adev->dev, "GCEA err detected at instance: %d, status: 0x%x!\n", + j, reg_value); + } + } + + gfx_v9_4_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + mutex_unlock(&adev->grbm_idx_mutex); +} diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h index 1ffecc5c0f0a..875f18473a98 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h @@ -34,4 +34,6 @@ int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev); +void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev); + #endif /* __GFX_V9_4_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c index 529e46386a50..fad887a66886 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c @@ -245,7 +245,7 @@ static void gfxhub_v1_0_setup_vmid_config(struct amdgpu_device *adev) /* Send no-retry XNACK on fault to suppress VM fault storm. */ tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, - !amdgpu_noretry); + !adev->gmc.noretry); WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_CNTL, i * hub->ctx_distance, tmp); WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, @@ -403,3 +403,13 @@ void gfxhub_v1_0_init(struct amdgpu_device *adev) hub->eng_addr_distance = mmVM_INVALIDATE_ENG1_ADDR_RANGE_LO32 - mmVM_INVALIDATE_ENG0_ADDR_RANGE_LO32; } + + +const struct amdgpu_gfxhub_funcs gfxhub_v1_0_funcs = { + .get_mc_fb_offset = gfxhub_v1_0_get_mc_fb_offset, + .setup_vm_pt_regs = gfxhub_v1_0_setup_vm_pt_regs, + .gart_enable = gfxhub_v1_0_gart_enable, + .gart_disable = gfxhub_v1_0_gart_disable, + .set_fault_enable_default = gfxhub_v1_0_set_fault_enable_default, + .init = gfxhub_v1_0_init, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h index 92d3a70cd9b1..0c46672bbf49 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h @@ -33,4 +33,5 @@ u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev); void gfxhub_v1_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, uint64_t page_table_base); +extern const struct amdgpu_gfxhub_funcs gfxhub_v1_0_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c index c0ab71df0d90..1e24b6d51e41 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.c @@ -21,6 +21,7 @@ * */ #include "amdgpu.h" +#include "gfxhub_v1_0.h" #include "gfxhub_v1_1.h" #include "gc/gc_9_2_1_offset.h" @@ -28,7 +29,7 @@ #include "soc15_common.h" -int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev) +static int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev) { u32 xgmi_lfb_cntl = RREG32_SOC15(GC, 0, mmMC_VM_XGMI_LFB_CNTL); u32 max_region = @@ -66,3 +67,13 @@ int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev) return 0; } + +const struct amdgpu_gfxhub_funcs gfxhub_v1_1_funcs = { + .get_mc_fb_offset = gfxhub_v1_0_get_mc_fb_offset, + .setup_vm_pt_regs = gfxhub_v1_0_setup_vm_pt_regs, + .gart_enable = gfxhub_v1_0_gart_enable, + .gart_disable = gfxhub_v1_0_gart_disable, + .set_fault_enable_default = gfxhub_v1_0_set_fault_enable_default, + .init = gfxhub_v1_0_init, + .get_xgmi_info = gfxhub_v1_1_get_xgmi_info, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h index d753cf28a0a6..ae5759ffbee3 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_1.h @@ -24,6 +24,6 @@ #ifndef __GFXHUB_V1_1_H__ #define __GFXHUB_V1_1_H__ -int gfxhub_v1_1_get_xgmi_info(struct amdgpu_device *adev); +extern const struct amdgpu_gfxhub_funcs gfxhub_v1_1_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c index b882ac59879a..456360bf58fa 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c @@ -102,7 +102,7 @@ gfxhub_v2_0_print_l2_protection_fault_status(struct amdgpu_device *adev, GCVM_L2_PROTECTION_FAULT_STATUS, RW)); } -u64 gfxhub_v2_0_get_fb_location(struct amdgpu_device *adev) +static u64 gfxhub_v2_0_get_fb_location(struct amdgpu_device *adev) { u64 base = RREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_BASE); @@ -112,12 +112,12 @@ u64 gfxhub_v2_0_get_fb_location(struct amdgpu_device *adev) return base; } -u64 gfxhub_v2_0_get_mc_fb_offset(struct amdgpu_device *adev) +static u64 gfxhub_v2_0_get_mc_fb_offset(struct amdgpu_device *adev) { return (u64)RREG32_SOC15(GC, 0, mmGCMC_VM_FB_OFFSET) << 24; } -void gfxhub_v2_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, +static void gfxhub_v2_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, uint64_t page_table_base) { struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB_0]; @@ -153,11 +153,6 @@ static void gfxhub_v2_0_init_system_aperture_regs(struct amdgpu_device *adev) uint64_t value; if (!amdgpu_sriov_vf(adev)) { - /* - * the new L1 policy will block SRIOV guest from writing - * these regs, and they will be programed at host. - * so skip programing these regs. - */ /* Disable AGP. */ WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_BASE, 0); WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_TOP, 0); @@ -318,7 +313,7 @@ static void gfxhub_v2_0_setup_vmid_config(struct amdgpu_device *adev) /* Send no-retry XNACK on fault to suppress VM fault storm. */ tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, - !amdgpu_noretry); + !adev->gmc.noretry); WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL, i * hub->ctx_distance, tmp); WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, @@ -347,7 +342,7 @@ static void gfxhub_v2_0_program_invalidation(struct amdgpu_device *adev) } } -int gfxhub_v2_0_gart_enable(struct amdgpu_device *adev) +static int gfxhub_v2_0_gart_enable(struct amdgpu_device *adev) { /* GART Enable. */ gfxhub_v2_0_init_gart_aperture_regs(adev); @@ -363,7 +358,7 @@ int gfxhub_v2_0_gart_enable(struct amdgpu_device *adev) return 0; } -void gfxhub_v2_0_gart_disable(struct amdgpu_device *adev) +static void gfxhub_v2_0_gart_disable(struct amdgpu_device *adev) { struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB_0]; u32 tmp; @@ -394,7 +389,7 @@ void gfxhub_v2_0_gart_disable(struct amdgpu_device *adev) * @adev: amdgpu_device pointer * @value: true redirects VM faults to the default page */ -void gfxhub_v2_0_set_fault_enable_default(struct amdgpu_device *adev, +static void gfxhub_v2_0_set_fault_enable_default(struct amdgpu_device *adev, bool value) { u32 tmp; @@ -436,7 +431,7 @@ static const struct amdgpu_vmhub_funcs gfxhub_v2_0_vmhub_funcs = { .get_invalidate_req = gfxhub_v2_0_get_invalidate_req, }; -void gfxhub_v2_0_init(struct amdgpu_device *adev) +static void gfxhub_v2_0_init(struct amdgpu_device *adev) { struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB_0]; @@ -477,3 +472,13 @@ void gfxhub_v2_0_init(struct amdgpu_device *adev) hub->vmhub_funcs = &gfxhub_v2_0_vmhub_funcs; } + +const struct amdgpu_gfxhub_funcs gfxhub_v2_0_funcs = { + .get_fb_location = gfxhub_v2_0_get_fb_location, + .get_mc_fb_offset = gfxhub_v2_0_get_mc_fb_offset, + .setup_vm_pt_regs = gfxhub_v2_0_setup_vm_pt_regs, + .gart_enable = gfxhub_v2_0_gart_enable, + .gart_disable = gfxhub_v2_0_gart_disable, + .set_fault_enable_default = gfxhub_v2_0_set_fault_enable_default, + .init = gfxhub_v2_0_init, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.h index 392b8cd94fc0..9ddc35cd53d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.h +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.h @@ -24,14 +24,6 @@ #ifndef __GFXHUB_V2_0_H__ #define __GFXHUB_V2_0_H__ -u64 gfxhub_v2_0_get_fb_location(struct amdgpu_device *adev); -int gfxhub_v2_0_gart_enable(struct amdgpu_device *adev); -void gfxhub_v2_0_gart_disable(struct amdgpu_device *adev); -void gfxhub_v2_0_set_fault_enable_default(struct amdgpu_device *adev, - bool value); -void gfxhub_v2_0_init(struct amdgpu_device *adev); -u64 gfxhub_v2_0_get_mc_fb_offset(struct amdgpu_device *adev); -void gfxhub_v2_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, - uint64_t page_table_base); +extern const struct amdgpu_gfxhub_funcs gfxhub_v2_0_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c index 237a9ff5afa0..724bb29e9bb4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c @@ -102,7 +102,7 @@ gfxhub_v2_1_print_l2_protection_fault_status(struct amdgpu_device *adev, GCVM_L2_PROTECTION_FAULT_STATUS, RW)); } -u64 gfxhub_v2_1_get_fb_location(struct amdgpu_device *adev) +static u64 gfxhub_v2_1_get_fb_location(struct amdgpu_device *adev) { u64 base = RREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_BASE); @@ -112,12 +112,12 @@ u64 gfxhub_v2_1_get_fb_location(struct amdgpu_device *adev) return base; } -u64 gfxhub_v2_1_get_mc_fb_offset(struct amdgpu_device *adev) +static u64 gfxhub_v2_1_get_mc_fb_offset(struct amdgpu_device *adev) { return (u64)RREG32_SOC15(GC, 0, mmGCMC_VM_FB_OFFSET) << 24; } -void gfxhub_v2_1_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, +static void gfxhub_v2_1_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, uint64_t page_table_base) { struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB_0]; @@ -319,7 +319,7 @@ static void gfxhub_v2_1_setup_vmid_config(struct amdgpu_device *adev) /* Send no-retry XNACK on fault to suppress VM fault storm. */ tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, - !amdgpu_noretry); + !adev->gmc.noretry); WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL, i * hub->ctx_distance, tmp); WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, @@ -348,7 +348,7 @@ static void gfxhub_v2_1_program_invalidation(struct amdgpu_device *adev) } } -int gfxhub_v2_1_gart_enable(struct amdgpu_device *adev) +static int gfxhub_v2_1_gart_enable(struct amdgpu_device *adev) { if (amdgpu_sriov_vf(adev)) { /* @@ -376,7 +376,7 @@ int gfxhub_v2_1_gart_enable(struct amdgpu_device *adev) return 0; } -void gfxhub_v2_1_gart_disable(struct amdgpu_device *adev) +static void gfxhub_v2_1_gart_disable(struct amdgpu_device *adev) { struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB_0]; u32 tmp; @@ -405,7 +405,7 @@ void gfxhub_v2_1_gart_disable(struct amdgpu_device *adev) * @adev: amdgpu_device pointer * @value: true redirects VM faults to the default page */ -void gfxhub_v2_1_set_fault_enable_default(struct amdgpu_device *adev, +static void gfxhub_v2_1_set_fault_enable_default(struct amdgpu_device *adev, bool value) { u32 tmp; @@ -454,7 +454,7 @@ static const struct amdgpu_vmhub_funcs gfxhub_v2_1_vmhub_funcs = { .get_invalidate_req = gfxhub_v2_1_get_invalidate_req, }; -void gfxhub_v2_1_init(struct amdgpu_device *adev) +static void gfxhub_v2_1_init(struct amdgpu_device *adev) { struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB_0]; @@ -496,7 +496,7 @@ void gfxhub_v2_1_init(struct amdgpu_device *adev) hub->vmhub_funcs = &gfxhub_v2_1_vmhub_funcs; } -int gfxhub_v2_1_get_xgmi_info(struct amdgpu_device *adev) +static int gfxhub_v2_1_get_xgmi_info(struct amdgpu_device *adev) { u32 xgmi_lfb_cntl = RREG32_SOC15(GC, 0, mmGCMC_VM_XGMI_LFB_CNTL); u32 max_region = @@ -531,3 +531,14 @@ int gfxhub_v2_1_get_xgmi_info(struct amdgpu_device *adev) return 0; } + +const struct amdgpu_gfxhub_funcs gfxhub_v2_1_funcs = { + .get_fb_location = gfxhub_v2_1_get_fb_location, + .get_mc_fb_offset = gfxhub_v2_1_get_mc_fb_offset, + .setup_vm_pt_regs = gfxhub_v2_1_setup_vm_pt_regs, + .gart_enable = gfxhub_v2_1_gart_enable, + .gart_disable = gfxhub_v2_1_gart_disable, + .set_fault_enable_default = gfxhub_v2_1_set_fault_enable_default, + .init = gfxhub_v2_1_init, + .get_xgmi_info = gfxhub_v2_1_get_xgmi_info, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.h index 3452a4e9a3da..f75c2eccfad9 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.h +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.h @@ -24,16 +24,6 @@ #ifndef __GFXHUB_V2_1_H__ #define __GFXHUB_V2_1_H__ -u64 gfxhub_v2_1_get_fb_location(struct amdgpu_device *adev); -int gfxhub_v2_1_gart_enable(struct amdgpu_device *adev); -void gfxhub_v2_1_gart_disable(struct amdgpu_device *adev); -void gfxhub_v2_1_set_fault_enable_default(struct amdgpu_device *adev, - bool value); -void gfxhub_v2_1_init(struct amdgpu_device *adev); -u64 gfxhub_v2_1_get_mc_fb_offset(struct amdgpu_device *adev); -void gfxhub_v2_1_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, - uint64_t page_table_base); - -int gfxhub_v2_1_get_xgmi_info(struct amdgpu_device *adev); +extern const struct amdgpu_gfxhub_funcs gfxhub_v2_1_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index 31359e519d69..dbc8b76b9b78 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -634,11 +634,26 @@ static void gmc_v10_0_set_mmhub_funcs(struct amdgpu_device *adev) adev->mmhub.funcs = &mmhub_v2_0_funcs; } +static void gmc_v10_0_set_gfxhub_funcs(struct amdgpu_device *adev) +{ + switch (adev->asic_type) { + case CHIP_SIENNA_CICHLID: + case CHIP_NAVY_FLOUNDER: + adev->gfxhub.funcs = &gfxhub_v2_1_funcs; + break; + default: + adev->gfxhub.funcs = &gfxhub_v2_0_funcs; + break; + } +} + + static int gmc_v10_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; gmc_v10_0_set_mmhub_funcs(adev); + gmc_v10_0_set_gfxhub_funcs(adev); gmc_v10_0_set_gmc_funcs(adev); gmc_v10_0_set_irq_funcs(adev); gmc_v10_0_set_umc_funcs(adev); @@ -676,11 +691,7 @@ static void gmc_v10_0_vram_gtt_location(struct amdgpu_device *adev, { u64 base = 0; - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) - base = gfxhub_v2_1_get_fb_location(adev); - else - base = gfxhub_v2_0_get_fb_location(adev); + base = adev->gfxhub.funcs->get_fb_location(adev); /* add the xgmi offset of the physical node */ base += adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size; @@ -689,11 +700,7 @@ static void gmc_v10_0_vram_gtt_location(struct amdgpu_device *adev, amdgpu_gmc_gart_location(adev, mc); /* base offset of vram pages */ - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) - adev->vm_manager.vram_base_offset = gfxhub_v2_1_get_mc_fb_offset(adev); - else - adev->vm_manager.vram_base_offset = gfxhub_v2_0_get_mc_fb_offset(adev); + adev->vm_manager.vram_base_offset = adev->gfxhub.funcs->get_mc_fb_offset(adev); /* add the xgmi offset of the physical node */ adev->vm_manager.vram_base_offset += @@ -777,11 +784,7 @@ static int gmc_v10_0_sw_init(void *handle) int r, vram_width = 0, vram_type = 0, vram_vendor = 0; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) - gfxhub_v2_1_init(adev); - else - gfxhub_v2_0_init(adev); + adev->gfxhub.funcs->init(adev); adev->mmhub.funcs->init(adev); @@ -852,7 +855,7 @@ static int gmc_v10_0_sw_init(void *handle) } if (adev->gmc.xgmi.supported) { - r = gfxhub_v2_1_get_xgmi_info(adev); + r = adev->gfxhub.funcs->get_xgmi_info(adev); if (r) return r; } @@ -944,11 +947,7 @@ static int gmc_v10_0_gart_enable(struct amdgpu_device *adev) if (r) return r; - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) - r = gfxhub_v2_1_gart_enable(adev); - else - r = gfxhub_v2_0_gart_enable(adev); + r = adev->gfxhub.funcs->gart_enable(adev); if (r) return r; @@ -969,11 +968,7 @@ static int gmc_v10_0_gart_enable(struct amdgpu_device *adev) value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ? false : true; - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) - gfxhub_v2_1_set_fault_enable_default(adev, value); - else - gfxhub_v2_0_set_fault_enable_default(adev, value); + adev->gfxhub.funcs->set_fault_enable_default(adev, value); adev->mmhub.funcs->set_fault_enable_default(adev, value); gmc_v10_0_flush_gpu_tlb(adev, 0, AMDGPU_MMHUB_0, 0); gmc_v10_0_flush_gpu_tlb(adev, 0, AMDGPU_GFXHUB_0, 0); @@ -1014,11 +1009,7 @@ static int gmc_v10_0_hw_init(void *handle) */ static void gmc_v10_0_gart_disable(struct amdgpu_device *adev) { - if (adev->asic_type == CHIP_SIENNA_CICHLID || - adev->asic_type == CHIP_NAVY_FLOUNDER) - gfxhub_v2_1_gart_disable(adev); - else - gfxhub_v2_0_gart_disable(adev); + adev->gfxhub.funcs->gart_disable(adev); adev->mmhub.funcs->gart_disable(adev); amdgpu_gart_table_vram_unpin(adev); } diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 91629c2b1d5c..3ebbddb63705 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -269,7 +269,6 @@ static const char *mmhub_client_ids_arcturus[][2] = { [14][1] = "HDP", [15][1] = "SDMA0", [32+15][1] = "SDMA1", - [32+15][1] = "SDMA1", [64+15][1] = "SDMA2", [96+15][1] = "SDMA3", [128+15][1] = "SDMA4", @@ -1165,6 +1164,19 @@ static void gmc_v9_0_set_mmhub_funcs(struct amdgpu_device *adev) } } +static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev) +{ + switch (adev->asic_type) { + case CHIP_ARCTURUS: + case CHIP_VEGA20: + adev->gfxhub.funcs = &gfxhub_v1_1_funcs; + break; + default: + adev->gfxhub.funcs = &gfxhub_v1_0_funcs; + break; + } +} + static int gmc_v9_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -1173,6 +1185,7 @@ static int gmc_v9_0_early_init(void *handle) gmc_v9_0_set_irq_funcs(adev); gmc_v9_0_set_umc_funcs(adev); gmc_v9_0_set_mmhub_funcs(adev); + gmc_v9_0_set_gfxhub_funcs(adev); adev->gmc.shared_aperture_start = 0x2000000000000000ULL; adev->gmc.shared_aperture_end = @@ -1194,21 +1207,16 @@ static int gmc_v9_0_late_init(void *handle) r = amdgpu_gmc_allocate_vm_inv_eng(adev); if (r) return r; - /* Check if ecc is available */ + + /* + * Workaround performance drop issue with VBIOS enables partial + * writes, while disables HBM ECC for vega10. + */ if (!amdgpu_sriov_vf(adev) && (adev->asic_type == CHIP_VEGA10)) { - r = amdgpu_atomfirmware_mem_ecc_supported(adev); - if (!r) { - DRM_INFO("ECC is not present.\n"); + if (!(adev->ras_features & (1 << AMDGPU_RAS_BLOCK__UMC))) { if (adev->df.funcs->enable_ecc_force_par_wr_rmw) adev->df.funcs->enable_ecc_force_par_wr_rmw(adev, false); - } else - DRM_INFO("ECC is active.\n"); - - r = amdgpu_atomfirmware_sram_ecc_supported(adev); - if (!r) - DRM_INFO("SRAM ECC is not present.\n"); - else - DRM_INFO("SRAM ECC is active.\n"); + } } if (adev->mmhub.funcs && adev->mmhub.funcs->reset_ras_error_count) @@ -1235,7 +1243,7 @@ static void gmc_v9_0_vram_gtt_location(struct amdgpu_device *adev, amdgpu_gmc_gart_location(adev, mc); amdgpu_gmc_agp_location(adev, mc); /* base offset of vram pages */ - adev->vm_manager.vram_base_offset = gfxhub_v1_0_get_mc_fb_offset(adev); + adev->vm_manager.vram_base_offset = adev->gfxhub.funcs->get_mc_fb_offset(adev); /* XXX: add the xgmi offset of the physical node? */ adev->vm_manager.vram_base_offset += @@ -1270,7 +1278,7 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev) #ifdef CONFIG_X86_64 if (adev->flags & AMD_IS_APU) { - adev->gmc.aper_base = gfxhub_v1_0_get_mc_fb_offset(adev); + adev->gmc.aper_base = adev->gfxhub.funcs->get_mc_fb_offset(adev); adev->gmc.aper_size = adev->gmc.real_vram_size; } #endif @@ -1340,7 +1348,7 @@ static int gmc_v9_0_sw_init(void *handle) int r, vram_width = 0, vram_type = 0, vram_vendor = 0; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - gfxhub_v1_0_init(adev); + adev->gfxhub.funcs->init(adev); adev->mmhub.funcs->init(adev); @@ -1454,7 +1462,7 @@ static int gmc_v9_0_sw_init(void *handle) adev->need_swiotlb = drm_need_swiotlb(44); if (adev->gmc.xgmi.supported) { - r = gfxhub_v1_1_get_xgmi_info(adev); + r = adev->gfxhub.funcs->get_xgmi_info(adev); if (r) return r; } @@ -1546,8 +1554,11 @@ static void gmc_v9_0_init_golden_registers(struct amdgpu_device *adev) */ void gmc_v9_0_restore_registers(struct amdgpu_device *adev) { - if (adev->asic_type == CHIP_RAVEN) + if (adev->asic_type == CHIP_RAVEN) { WREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0, adev->gmc.sdpif_register); + WARN_ON(adev->gmc.sdpif_register != + RREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0)); + } } /** @@ -1567,7 +1578,7 @@ static int gmc_v9_0_gart_enable(struct amdgpu_device *adev) if (r) return r; - r = gfxhub_v1_0_gart_enable(adev); + r = adev->gfxhub.funcs->gart_enable(adev); if (r) return r; @@ -1634,7 +1645,7 @@ static int gmc_v9_0_hw_init(void *handle) value = true; if (!amdgpu_sriov_vf(adev)) { - gfxhub_v1_0_set_fault_enable_default(adev, value); + adev->gfxhub.funcs->set_fault_enable_default(adev, value); adev->mmhub.funcs->set_fault_enable_default(adev, value); } for (i = 0; i < adev->num_vmhubs; ++i) @@ -1657,7 +1668,7 @@ static int gmc_v9_0_hw_init(void *handle) */ static void gmc_v9_0_gart_disable(struct amdgpu_device *adev) { - gfxhub_v1_0_gart_disable(adev); + adev->gfxhub.funcs->gart_disable(adev); adev->mmhub.funcs->gart_disable(adev); amdgpu_gart_table_vram_unpin(adev); } @@ -1681,14 +1692,9 @@ static int gmc_v9_0_hw_fini(void *handle) static int gmc_v9_0_suspend(void *handle) { - int r; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - r = gmc_v9_0_hw_fini(adev); - if (r) - return r; - - return 0; + return gmc_v9_0_hw_fini(adev); } static int gmc_v9_0_resume(void *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c index bc300283b6ab..c600b61b5f45 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c @@ -33,6 +33,7 @@ static void jpeg_v1_0_set_dec_ring_funcs(struct amdgpu_device *adev); static void jpeg_v1_0_set_irq_funcs(struct amdgpu_device *adev); +static void jpeg_v1_0_ring_begin_use(struct amdgpu_ring *ring); static void jpeg_v1_0_decode_ring_patch_wreg(struct amdgpu_ring *ring, uint32_t *ptr, uint32_t reg_offset, uint32_t val) { @@ -564,8 +565,8 @@ static const struct amdgpu_ring_funcs jpeg_v1_0_decode_ring_vm_funcs = { .insert_start = jpeg_v1_0_decode_ring_insert_start, .insert_end = jpeg_v1_0_decode_ring_insert_end, .pad_ib = amdgpu_ring_generic_pad_ib, - .begin_use = vcn_v1_0_ring_begin_use, - .end_use = amdgpu_vcn_ring_end_use, + .begin_use = jpeg_v1_0_ring_begin_use, + .end_use = vcn_v1_0_ring_end_use, .emit_wreg = jpeg_v1_0_decode_ring_emit_wreg, .emit_reg_wait = jpeg_v1_0_decode_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, @@ -586,3 +587,22 @@ static void jpeg_v1_0_set_irq_funcs(struct amdgpu_device *adev) { adev->jpeg.inst->irq.funcs = &jpeg_v1_0_irq_funcs; } + +static void jpeg_v1_0_ring_begin_use(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + bool set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work); + int cnt = 0; + + mutex_lock(&adev->vcn.vcn1_jpeg1_workaround); + + if (amdgpu_fence_wait_empty(&adev->vcn.inst->ring_dec)) + DRM_ERROR("JPEG dec: vcn dec ring may not be empty\n"); + + for (cnt = 0; cnt < adev->vcn.num_enc_rings; cnt++) { + if (amdgpu_fence_wait_empty(&adev->vcn.inst->ring_enc[cnt])) + DRM_ERROR("JPEG dec: vcn enc ring[%d] may not be empty\n", cnt); + } + + vcn_v1_0_set_pg_for_begin_use(ring, set_clocks); +} diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c index 4b746584a797..1c22d8393b21 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c @@ -832,7 +832,6 @@ static int mes_v10_1_queue_init(struct amdgpu_device *adev) static int mes_v10_1_ring_init(struct amdgpu_device *adev) { struct amdgpu_ring *ring; - int r; ring = &adev->mes.ring; @@ -849,11 +848,7 @@ static int mes_v10_1_ring_init(struct amdgpu_device *adev) ring->no_scheduler = true; sprintf(ring->name, "mes_%d.%d.%d", ring->me, ring->pipe, ring->queue); - r = amdgpu_ring_init(adev, ring, 1024, NULL, 0, AMDGPU_RING_PRIO_DEFAULT); - if (r) - return r; - - return 0; + return amdgpu_ring_init(adev, ring, 1024, NULL, 0, AMDGPU_RING_PRIO_DEFAULT); } static int mes_v10_1_mqd_sw_init(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c index 45a902b1acb7..f84701c562bf 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c @@ -268,7 +268,7 @@ static void mmhub_v1_0_setup_vmid_config(struct amdgpu_device *adev) /* Send no-retry XNACK on fault to suppress VM fault storm. */ tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, - !amdgpu_noretry); + !adev->gmc.noretry); WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_CNTL, i * hub->ctx_distance, tmp); WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c index 2d88278c50bf..2063700f0bc6 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c @@ -201,11 +201,6 @@ static void mmhub_v2_0_init_system_aperture_regs(struct amdgpu_device *adev) WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_BOT, 0x00FFFFFF); if (!amdgpu_sriov_vf(adev)) { - /* - * the new L1 policy will block SRIOV guest from writing - * these regs, and they will be programed at host. - * so skip programing these regs. - */ /* Program the system aperture low logical page number. */ WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_LOW_ADDR, adev->gmc.vram_start >> 18); @@ -374,7 +369,7 @@ static void mmhub_v2_0_setup_vmid_config(struct amdgpu_device *adev) /* Send no-retry XNACK on fault to suppress VM fault storm. */ tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, - !amdgpu_noretry); + !adev->gmc.noretry); WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_CNTL, i * hub->ctx_distance, tmp); WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c index 6c6ad529c65c..66748bb01b52 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c @@ -330,7 +330,7 @@ static void mmhub_v9_4_setup_vmid_config(struct amdgpu_device *adev, int hubid) /* Send no-retry XNACK on fault to suppress VM fault storm. */ tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL, RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, - !amdgpu_noretry); + !adev->gmc.noretry); WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2VC0_VM_CONTEXT1_CNTL, hubid * MMHUB_INSTANCE_REGISTER_OFFSET + i * hub->ctx_distance, tmp); @@ -1624,6 +1624,34 @@ static void mmhub_v9_4_reset_ras_error_count(struct amdgpu_device *adev) } } +static const struct soc15_reg_entry mmhub_v9_4_err_status_regs[] = { + { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA0_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA1_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA2_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA3_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA4_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA5_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA6_ERR_STATUS), 0, 0, 0 }, + { SOC15_REG_ENTRY(MMHUB, 0, mmMMEA7_ERR_STATUS), 0, 0, 0 }, +}; + +static void mmhub_v9_4_query_ras_error_status(struct amdgpu_device *adev) +{ + int i; + uint32_t reg_value; + + if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB)) + return; + + for (i = 0; i < ARRAY_SIZE(mmhub_v9_4_err_status_regs); i++) { + reg_value = + RREG32(SOC15_REG_ENTRY_OFFSET(mmhub_v9_4_err_status_regs[i])); + if (reg_value) + dev_warn(adev->dev, "MMHUB EA err detected at instance: %d, status: 0x%x!\n", + i, reg_value); + } +} + const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = { .ras_late_init = amdgpu_mmhub_ras_late_init, .query_ras_error_count = mmhub_v9_4_query_ras_error_count, @@ -1636,4 +1664,5 @@ const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = { .set_clockgating = mmhub_v9_4_set_clockgating, .get_clockgating = mmhub_v9_4_get_clockgating, .setup_vm_pt_regs = mmhub_v9_4_setup_vm_pt_regs, + .query_ras_error_status = mmhub_v9_4_query_ras_error_status, }; diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c index 9c07014d9bd6..f5ce9a9f4cf5 100644 --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c @@ -262,7 +262,8 @@ flr_done: /* Trigger recovery for world switch failure if no TDR */ if (amdgpu_device_should_recover_gpu(adev) - && (amdgpu_device_has_job_running(adev) || adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT)) + && (!amdgpu_device_has_job_running(adev) || + adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT)) amdgpu_device_gpu_recover(adev, NULL); } diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c index 9c23abf9b140..666ed99cc14b 100644 --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c @@ -283,7 +283,7 @@ flr_done: /* Trigger recovery for world switch failure if no TDR */ if (amdgpu_device_should_recover_gpu(adev) - && (amdgpu_device_has_job_running(adev) || + && (!amdgpu_device_has_job_running(adev) || adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT || adev->gfx_timeout == MAX_SCHEDULE_TIMEOUT || adev->compute_timeout == MAX_SCHEDULE_TIMEOUT || diff --git a/drivers/gpu/drm/amd/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c index 4d1402356262..03462c857498 100644 --- a/drivers/gpu/drm/amd/amdgpu/nv.c +++ b/drivers/gpu/drm/amd/amdgpu/nv.c @@ -69,75 +69,40 @@ static const struct amd_ip_funcs nv_common_ip_funcs; */ static u32 nv_pcie_rreg(struct amdgpu_device *adev, u32 reg) { - unsigned long flags, address, data; - u32 r; + unsigned long address, data; address = adev->nbio.funcs->get_pcie_index_offset(adev); data = adev->nbio.funcs->get_pcie_data_offset(adev); - spin_lock_irqsave(&adev->pcie_idx_lock, flags); - WREG32(address, reg); - (void)RREG32(address); - r = RREG32(data); - spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); - return r; + return amdgpu_device_indirect_rreg(adev, address, data, reg); } static void nv_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v) { - unsigned long flags, address, data; + unsigned long address, data; address = adev->nbio.funcs->get_pcie_index_offset(adev); data = adev->nbio.funcs->get_pcie_data_offset(adev); - spin_lock_irqsave(&adev->pcie_idx_lock, flags); - WREG32(address, reg); - (void)RREG32(address); - WREG32(data, v); - (void)RREG32(data); - spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); + amdgpu_device_indirect_wreg(adev, address, data, reg, v); } static u64 nv_pcie_rreg64(struct amdgpu_device *adev, u32 reg) { - unsigned long flags, address, data; - u64 r; + unsigned long address, data; address = adev->nbio.funcs->get_pcie_index_offset(adev); data = adev->nbio.funcs->get_pcie_data_offset(adev); - spin_lock_irqsave(&adev->pcie_idx_lock, flags); - /* read low 32 bit */ - WREG32(address, reg); - (void)RREG32(address); - r = RREG32(data); - - /* read high 32 bit*/ - WREG32(address, reg + 4); - (void)RREG32(address); - r |= ((u64)RREG32(data) << 32); - spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); - return r; + return amdgpu_device_indirect_rreg64(adev, address, data, reg); } static void nv_pcie_wreg64(struct amdgpu_device *adev, u32 reg, u64 v) { - unsigned long flags, address, data; + unsigned long address, data; address = adev->nbio.funcs->get_pcie_index_offset(adev); data = adev->nbio.funcs->get_pcie_data_offset(adev); - spin_lock_irqsave(&adev->pcie_idx_lock, flags); - /* write low 32 bit */ - WREG32(address, reg); - (void)RREG32(address); - WREG32(data, (u32)(v & 0xffffffffULL)); - (void)RREG32(data); - - /* write high 32 bit */ - WREG32(address, reg + 4); - (void)RREG32(address); - WREG32(data, (u32)(v >> 32)); - (void)RREG32(data); - spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); + amdgpu_device_indirect_wreg64(adev, address, data, reg, v); } static u32 nv_didt_rreg(struct amdgpu_device *adev, u32 reg) @@ -311,7 +276,7 @@ static int nv_asic_mode1_reset(struct amdgpu_device *adev) /* disable BM */ pci_clear_master(adev->pdev); - pci_save_state(adev->pdev); + amdgpu_device_cache_pci_state(adev->pdev); if (amdgpu_dpm_is_mode1_reset_supported(adev)) { dev_info(adev->dev, "GPU smu mode1 reset\n"); @@ -323,7 +288,7 @@ static int nv_asic_mode1_reset(struct amdgpu_device *adev) if (ret) dev_err(adev->dev, "GPU mode1 reset failed\n"); - pci_restore_state(adev->pdev); + amdgpu_device_load_pci_state(adev->pdev); /* wait for asic to come out of reset */ for (i = 0; i < adev->usec_timeout; i++) { @@ -490,6 +455,14 @@ void nv_set_virt_ops(struct amdgpu_device *adev) adev->virt.ops = &xgpu_nv_virt_ops; } +static bool nv_is_blockchain_sku(struct pci_dev *pdev) +{ + if (pdev->device == 0x731E && + (pdev->revision == 0xC6 || pdev->revision == 0xC7)) + return true; + return false; +} + int nv_set_ip_blocks(struct amdgpu_device *adev) { int r; @@ -518,7 +491,8 @@ int nv_set_ip_blocks(struct amdgpu_device *adev) if (adev->enable_virtual_display || amdgpu_sriov_vf(adev)) amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block); #if defined(CONFIG_DRM_AMD_DC) - else if (amdgpu_device_has_dc_support(adev)) + else if (amdgpu_device_has_dc_support(adev) && + !nv_is_blockchain_sku(adev->pdev)) amdgpu_device_ip_block_add(adev, &dm_ip_block); #endif amdgpu_device_ip_block_add(adev, &gfx_v10_0_ip_block); @@ -526,7 +500,8 @@ int nv_set_ip_blocks(struct amdgpu_device *adev) if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT && !amdgpu_sriov_vf(adev)) amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block); - amdgpu_device_ip_block_add(adev, &vcn_v2_0_ip_block); + if (!nv_is_blockchain_sku(adev->pdev)) + amdgpu_device_ip_block_add(adev, &vcn_v2_0_ip_block); amdgpu_device_ip_block_add(adev, &jpeg_v2_0_ip_block); if (adev->enable_mes) amdgpu_device_ip_block_add(adev, &mes_v10_1_ip_block); @@ -621,7 +596,7 @@ static void nv_invalidate_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring) { if (!ring || !ring->funcs->emit_wreg) { - WREG32_SOC15_NO_KIQ(NBIO, 0, mmHDP_READ_CACHE_INVALIDATE, 1); + WREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE, 1); } else { amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET( HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 1); diff --git a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h index cbc04a5c0fe1..4137dc710aaf 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h +++ b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h @@ -83,19 +83,6 @@ struct psp_gfx_ctrl */ #define GFX_FLAG_RESPONSE 0x80000000 -/* Gbr IH registers ID */ -enum ih_reg_id { - IH_RB = 0, // IH_RB_CNTL - IH_RB_RNG1 = 1, // IH_RB_CNTL_RING1 - IH_RB_RNG2 = 2, // IH_RB_CNTL_RING2 -}; - -/* Command to setup Gibraltar IH register */ -struct psp_gfx_cmd_gbr_ih_reg { - uint32_t reg_value; /* Value to be set to the IH_RB_CNTL... register*/ - enum ih_reg_id reg_id; /* ID of the register */ -}; - /* TEE Gfx Command IDs for the ring buffer interface. */ enum psp_gfx_cmd_id { @@ -214,7 +201,7 @@ enum psp_gfx_fw_type { GFX_FW_TYPE_UVD1 = 23, /* UVD1 VG-20 */ GFX_FW_TYPE_TOC = 24, /* TOC NV-10 */ GFX_FW_TYPE_RLC_P = 25, /* RLC P NV */ - GFX_FW_TYPE_RLX6 = 26, /* RLX6 NV */ + GFX_FW_TYPE_RLC_IRAM = 26, /* RLC_IRAM NV */ GFX_FW_TYPE_GLOBAL_TAP_DELAYS = 27, /* GLOBAL TAP DELAYS NV */ GFX_FW_TYPE_SE0_TAP_DELAYS = 28, /* SE0 TAP DELAYS NV */ GFX_FW_TYPE_SE1_TAP_DELAYS = 29, /* SE1 TAP DELAYS NV */ @@ -236,7 +223,7 @@ enum psp_gfx_fw_type { GFX_FW_TYPE_ACCUM_CTRL_RAM = 45, /* ACCUM CTRL RAM NV */ GFX_FW_TYPE_RLCP_CAM = 46, /* RLCP CAM NV */ GFX_FW_TYPE_RLC_SPP_CAM_EXT = 47, /* RLC SPP CAM EXT NV */ - GFX_FW_TYPE_RLX6_DRAM_BOOT = 48, /* RLX6 DRAM BOOT NV */ + GFX_FW_TYPE_RLC_DRAM_BOOT = 48, /* RLC DRAM BOOT NV */ GFX_FW_TYPE_VCN0_RAM = 49, /* VCN_RAM NV + RN */ GFX_FW_TYPE_VCN1_RAM = 50, /* VCN_RAM NV + RN */ GFX_FW_TYPE_DMUB = 51, /* DMUB RN */ diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index e16874f30d5d..6c5d9612abcb 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -58,7 +58,7 @@ MODULE_FIRMWARE("amdgpu/arcturus_ta.bin"); MODULE_FIRMWARE("amdgpu/sienna_cichlid_sos.bin"); MODULE_FIRMWARE("amdgpu/sienna_cichlid_ta.bin"); MODULE_FIRMWARE("amdgpu/navy_flounder_sos.bin"); -MODULE_FIRMWARE("amdgpu/navy_flounder_asd.bin"); +MODULE_FIRMWARE("amdgpu/navy_flounder_ta.bin"); /* address block */ #define smnMP1_FIRMWARE_FLAGS 0x3010024 diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index 856c50386c86..86fb1eddf5a6 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -592,6 +592,9 @@ static int sdma_v4_0_init_microcode(struct amdgpu_device *adev) struct amdgpu_firmware_info *info = NULL; const struct common_firmware_header *header = NULL; + if (amdgpu_sriov_vf(adev)) + return 0; + DRM_DEBUG("\n"); switch (adev->asic_type) { @@ -1000,7 +1003,7 @@ static void sdma_v4_0_page_stop(struct amdgpu_device *adev) sdma[i] = &adev->sdma.instance[i].page; if ((adev->mman.buffer_funcs_ring == sdma[i]) && - (unset == false)) { + (!unset)) { amdgpu_ttm_set_buffer_funcs_status(adev, false); unset = true; } @@ -1063,6 +1066,15 @@ static void sdma_v4_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable) WREG32_SDMA(i, mmSDMA0_PHASE2_QUANTUM, phase_quantum); } WREG32_SDMA(i, mmSDMA0_CNTL, f32_cntl); + + /* + * Enable SDMA utilization. Its only supported on + * Arcturus for the moment and firmware version 14 + * and above. + */ + if (adev->asic_type == CHIP_ARCTURUS && + adev->sdma.instance[i].fw_version >= 14) + WREG32_SDMA(i, mmSDMA0_PUB_DUMMY_REG2, enable); } } @@ -1080,7 +1092,7 @@ static void sdma_v4_0_enable(struct amdgpu_device *adev, bool enable) u32 f32_cntl; int i; - if (enable == false) { + if (!enable) { sdma_v4_0_gfx_stop(adev); sdma_v4_0_rlc_stop(adev); if (adev->sdma.has_page_queue) diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c index e2232dd12d8e..9c72b95b7463 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c @@ -203,6 +203,9 @@ static int sdma_v5_0_init_microcode(struct amdgpu_device *adev) const struct common_firmware_header *header = NULL; const struct sdma_firmware_header_v1_0 *hdr; + if (amdgpu_sriov_vf(adev)) + return 0; + DRM_DEBUG("\n"); switch (adev->asic_type) { @@ -616,7 +619,7 @@ static void sdma_v5_0_enable(struct amdgpu_device *adev, bool enable) u32 f32_cntl; int i; - if (enable == false) { + if (!enable) { sdma_v5_0_gfx_stop(adev); sdma_v5_0_rlc_stop(adev); } diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c index 46a9617fee5f..9f3952723c63 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c @@ -148,6 +148,9 @@ static int sdma_v5_2_init_microcode(struct amdgpu_device *adev) struct amdgpu_firmware_info *info = NULL; const struct common_firmware_header *header = NULL; + if (amdgpu_sriov_vf(adev)) + return 0; + DRM_DEBUG("\n"); switch (adev->asic_type) { @@ -559,7 +562,7 @@ static void sdma_v5_2_enable(struct amdgpu_device *adev, bool enable) u32 f32_cntl; int i; - if (enable == false) { + if (!enable) { sdma_v5_2_gfx_stop(adev); sdma_v5_2_rlc_stop(adev); } diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c index 455d5e366c69..e5e336fd9e94 100644 --- a/drivers/gpu/drm/amd/amdgpu/si.c +++ b/drivers/gpu/drm/amd/amdgpu/si.c @@ -1339,7 +1339,7 @@ static void si_vga_set_state(struct amdgpu_device *adev, bool state) uint32_t temp; temp = RREG32(CONFIG_CNTL); - if (state == false) { + if (!state) { temp &= ~(1<<0); temp |= (1<<1); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index 2f93c475d6d8..afcccc6c0fc6 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -101,75 +101,40 @@ */ static u32 soc15_pcie_rreg(struct amdgpu_device *adev, u32 reg) { - unsigned long flags, address, data; - u32 r; + unsigned long address, data; address = adev->nbio.funcs->get_pcie_index_offset(adev); data = adev->nbio.funcs->get_pcie_data_offset(adev); - spin_lock_irqsave(&adev->pcie_idx_lock, flags); - WREG32(address, reg); - (void)RREG32(address); - r = RREG32(data); - spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); - return r; + return amdgpu_device_indirect_rreg(adev, address, data, reg); } static void soc15_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v) { - unsigned long flags, address, data; + unsigned long address, data; address = adev->nbio.funcs->get_pcie_index_offset(adev); data = adev->nbio.funcs->get_pcie_data_offset(adev); - spin_lock_irqsave(&adev->pcie_idx_lock, flags); - WREG32(address, reg); - (void)RREG32(address); - WREG32(data, v); - (void)RREG32(data); - spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); + amdgpu_device_indirect_wreg(adev, address, data, reg, v); } static u64 soc15_pcie_rreg64(struct amdgpu_device *adev, u32 reg) { - unsigned long flags, address, data; - u64 r; + unsigned long address, data; address = adev->nbio.funcs->get_pcie_index_offset(adev); data = adev->nbio.funcs->get_pcie_data_offset(adev); - spin_lock_irqsave(&adev->pcie_idx_lock, flags); - /* read low 32 bit */ - WREG32(address, reg); - (void)RREG32(address); - r = RREG32(data); - - /* read high 32 bit*/ - WREG32(address, reg + 4); - (void)RREG32(address); - r |= ((u64)RREG32(data) << 32); - spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); - return r; + return amdgpu_device_indirect_rreg64(adev, address, data, reg); } static void soc15_pcie_wreg64(struct amdgpu_device *adev, u32 reg, u64 v) { - unsigned long flags, address, data; + unsigned long address, data; address = adev->nbio.funcs->get_pcie_index_offset(adev); data = adev->nbio.funcs->get_pcie_data_offset(adev); - spin_lock_irqsave(&adev->pcie_idx_lock, flags); - /* write low 32 bit */ - WREG32(address, reg); - (void)RREG32(address); - WREG32(data, (u32)(v & 0xffffffffULL)); - (void)RREG32(data); - - /* write high 32 bit */ - WREG32(address, reg + 4); - (void)RREG32(address); - WREG32(data, (u32)(v >> 32)); - (void)RREG32(data); - spin_unlock_irqrestore(&adev->pcie_idx_lock, flags); + amdgpu_device_indirect_wreg64(adev, address, data, reg, v); } static u32 soc15_uvd_ctx_rreg(struct amdgpu_device *adev, u32 reg) @@ -484,13 +449,13 @@ static int soc15_asic_mode1_reset(struct amdgpu_device *adev) /* disable BM */ pci_clear_master(adev->pdev); - pci_save_state(adev->pdev); + amdgpu_device_cache_pci_state(adev->pdev); ret = psp_gpu_reset(adev); if (ret) dev_err(adev->dev, "GPU mode1 reset failed\n"); - pci_restore_state(adev->pdev); + amdgpu_device_load_pci_state(adev->pdev); /* wait for asic to come out of reset */ for (i = 0; i < adev->usec_timeout; i++) { @@ -697,12 +662,12 @@ static void soc15_reg_base_init(struct amdgpu_device *adev) * it doesn't support SRIOV. */ if (amdgpu_discovery) { r = amdgpu_discovery_reg_base_init(adev); - if (r) { - DRM_WARN("failed to init reg base from ip discovery table, " - "fallback to legacy init method\n"); - vega10_reg_base_init(adev); - } + if (r == 0) + break; + DRM_WARN("failed to init reg base from ip discovery table, " + "fallback to legacy init method\n"); } + vega10_reg_base_init(adev); break; case CHIP_VEGA20: vega20_reg_base_init(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c index 3cafba726587..b0c0c438fc93 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c @@ -348,7 +348,7 @@ static int uvd_v4_2_start(struct amdgpu_device *adev) /* Set the write pointer delay */ WREG32(mmUVD_RBC_RB_WPTR_CNTL, 0); - /* programm the 4GB memory segment for rptr and ring buffer */ + /* program the 4GB memory segment for rptr and ring buffer */ WREG32(mmUVD_LMI_EXT40_ADDR, upper_32_bits(ring->gpu_addr) | (0x7 << 16) | (0x1 << 31)); @@ -541,7 +541,7 @@ static void uvd_v4_2_mc_resume(struct amdgpu_device *adev) uint64_t addr; uint32_t size; - /* programm the VCPU memory controller bits 0-27 */ + /* program the VCPU memory controller bits 0-27 */ addr = (adev->uvd.inst->gpu_addr + AMDGPU_UVD_FIRMWARE_OFFSET) >> 3; size = AMDGPU_UVD_FIRMWARE_SIZE(adev) >> 3; WREG32(mmUVD_VCPU_CACHE_OFFSET0, addr); diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c index a566ff926e90..6e57001f6d0a 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c @@ -253,7 +253,7 @@ static void uvd_v5_0_mc_resume(struct amdgpu_device *adev) uint64_t offset; uint32_t size; - /* programm memory controller bits 0-27 */ + /* program memory controller bits 0-27 */ WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW, lower_32_bits(adev->uvd.inst->gpu_addr)); WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH, @@ -404,7 +404,7 @@ static int uvd_v5_0_start(struct amdgpu_device *adev) /* set the wb address */ WREG32(mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2)); - /* programm the RB_BASE for ring buffer */ + /* program the RB_BASE for ring buffer */ WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr)); WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c index 0a880bc101b8..666bfa4a0b8e 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c @@ -583,7 +583,7 @@ static void uvd_v6_0_mc_resume(struct amdgpu_device *adev) uint64_t offset; uint32_t size; - /* programm memory controller bits 0-27 */ + /* program memory controller bits 0-27 */ WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW, lower_32_bits(adev->uvd.inst->gpu_addr)); WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH, @@ -825,7 +825,7 @@ static int uvd_v6_0_start(struct amdgpu_device *adev) /* set the wb address */ WREG32(mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2)); - /* programm the RB_BASE for ring buffer */ + /* program the RB_BASE for ring buffer */ WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr)); WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, @@ -1240,8 +1240,8 @@ static int uvd_v6_0_process_interrupt(struct amdgpu_device *adev, break; } - if (false == int_handled) - DRM_ERROR("Unhandled interrupt: %d %d\n", + if (!int_handled) + DRM_ERROR("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c index e07e3fae99b5..b44c8677ce8d 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c @@ -1073,7 +1073,7 @@ static int uvd_v7_0_start(struct amdgpu_device *adev) WREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2)); - /* programm the RB_BASE for ring buffer */ + /* program the RB_BASE for ring buffer */ WREG32_SOC15(UVD, k, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr)); WREG32_SOC15(UVD, k, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c index 927c330fad21..86e1ef732ebe 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c @@ -54,6 +54,7 @@ static int vcn_v1_0_pause_dpg_mode(struct amdgpu_device *adev, int inst_idx, struct dpg_pause_state *new_state); static void vcn_v1_0_idle_work_handler(struct work_struct *work); +static void vcn_v1_0_ring_begin_use(struct amdgpu_ring *ring); /** * vcn_v1_0_early_init - set function pointers @@ -910,7 +911,7 @@ static int vcn_v1_0_start_spg_mode(struct amdgpu_device *adev) WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2)); - /* programm the RB_BASE for ring buffer */ + /* program the RB_BASE for ring buffer */ WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr)); WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, @@ -1068,7 +1069,7 @@ static int vcn_v1_0_start_dpg_mode(struct amdgpu_device *adev) WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2)); - /* programm the RB_BASE for ring buffer */ + /* program the RB_BASE for ring buffer */ WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr)); WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, @@ -1804,11 +1805,24 @@ static void vcn_v1_0_idle_work_handler(struct work_struct *work) } } -void vcn_v1_0_ring_begin_use(struct amdgpu_ring *ring) +static void vcn_v1_0_ring_begin_use(struct amdgpu_ring *ring) { - struct amdgpu_device *adev = ring->adev; + struct amdgpu_device *adev = ring->adev; bool set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work); + mutex_lock(&adev->vcn.vcn1_jpeg1_workaround); + + if (amdgpu_fence_wait_empty(&ring->adev->jpeg.inst->ring_dec)) + DRM_ERROR("VCN dec: jpeg dec ring may not be empty\n"); + + vcn_v1_0_set_pg_for_begin_use(ring, set_clocks); + +} + +void vcn_v1_0_set_pg_for_begin_use(struct amdgpu_ring *ring, bool set_clocks) +{ + struct amdgpu_device *adev = ring->adev; + if (set_clocks) { amdgpu_gfx_off_ctrl(adev, false); if (adev->pm.dpm_enabled) @@ -1844,6 +1858,12 @@ void vcn_v1_0_ring_begin_use(struct amdgpu_ring *ring) } } +void vcn_v1_0_ring_end_use(struct amdgpu_ring *ring) +{ + schedule_delayed_work(&ring->adev->vcn.idle_work, VCN_IDLE_TIMEOUT); + mutex_unlock(&ring->adev->vcn.vcn1_jpeg1_workaround); +} + static const struct amd_ip_funcs vcn_v1_0_ip_funcs = { .name = "vcn_v1_0", .early_init = vcn_v1_0_early_init, @@ -1891,7 +1911,7 @@ static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = { .insert_end = vcn_v1_0_dec_ring_insert_end, .pad_ib = amdgpu_ring_generic_pad_ib, .begin_use = vcn_v1_0_ring_begin_use, - .end_use = amdgpu_vcn_ring_end_use, + .end_use = vcn_v1_0_ring_end_use, .emit_wreg = vcn_v1_0_dec_ring_emit_wreg, .emit_reg_wait = vcn_v1_0_dec_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, @@ -1923,7 +1943,7 @@ static const struct amdgpu_ring_funcs vcn_v1_0_enc_ring_vm_funcs = { .insert_end = vcn_v1_0_enc_ring_insert_end, .pad_ib = amdgpu_ring_generic_pad_ib, .begin_use = vcn_v1_0_ring_begin_use, - .end_use = amdgpu_vcn_ring_end_use, + .end_use = vcn_v1_0_ring_end_use, .emit_wreg = vcn_v1_0_enc_ring_emit_wreg, .emit_reg_wait = vcn_v1_0_enc_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.h b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.h index f67d7391fc21..1f1cc7f0ece7 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.h +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.h @@ -24,7 +24,8 @@ #ifndef __VCN_V1_0_H__ #define __VCN_V1_0_H__ -void vcn_v1_0_ring_begin_use(struct amdgpu_ring *ring); +void vcn_v1_0_ring_end_use(struct amdgpu_ring *ring); +void vcn_v1_0_set_pg_for_begin_use(struct amdgpu_ring *ring, bool set_clocks); extern const struct amdgpu_ip_block_version vcn_v1_0_ip_block; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c index 23a9eb5b2c8a..e5d29dee0c88 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c @@ -900,7 +900,7 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect) WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2)); - /* programm the RB_BASE for ring buffer */ + /* program the RB_BASE for ring buffer */ WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr)); WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, @@ -1060,7 +1060,7 @@ static int vcn_v2_0_start(struct amdgpu_device *adev) WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp); fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET; - /* programm the RB_BASE for ring buffer */ + /* program the RB_BASE for ring buffer */ WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr)); WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index 139fac0d8e76..0f1d3ef8baa7 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -882,7 +882,7 @@ static int vcn_v2_5_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, boo WREG32_SOC15(VCN, inst_idx, mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2)); - /* programm the RB_BASE for ring buffer */ + /* program the RB_BASE for ring buffer */ WREG32_SOC15(VCN, inst_idx, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr)); WREG32_SOC15(VCN, inst_idx, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, @@ -1062,7 +1062,7 @@ static int vcn_v2_5_start(struct amdgpu_device *adev) WREG32_SOC15(VCN, i, mmUVD_RBC_RB_CNTL, tmp); fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET; - /* programm the RB_BASE for ring buffer */ + /* program the RB_BASE for ring buffer */ WREG32_SOC15(VCN, i, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr)); WREG32_SOC15(VCN, i, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c index 589d6cd8adec..e074f7ed388c 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c @@ -746,18 +746,18 @@ static void vcn_v3_0_disable_clock_gating(struct amdgpu_device *adev, int inst) | UVD_SUVD_CGC_GATE__IME_HEVC_MASK | UVD_SUVD_CGC_GATE__EFC_MASK | UVD_SUVD_CGC_GATE__SAOE_MASK - | 0x08000000 + | UVD_SUVD_CGC_GATE__SRE_AV1_MASK | UVD_SUVD_CGC_GATE__FBC_PCLK_MASK | UVD_SUVD_CGC_GATE__FBC_CCLK_MASK - | 0x40000000 + | UVD_SUVD_CGC_GATE__SCM_AV1_MASK | UVD_SUVD_CGC_GATE__SMPA_MASK); WREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_GATE, data); data = RREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_GATE2); data |= (UVD_SUVD_CGC_GATE2__MPBE0_MASK | UVD_SUVD_CGC_GATE2__MPBE1_MASK - | 0x00000004 - | 0x00000008 + | UVD_SUVD_CGC_GATE2__SIT_AV1_MASK + | UVD_SUVD_CGC_GATE2__SDB_AV1_MASK | UVD_SUVD_CGC_GATE2__MPC1_MASK); WREG32_SOC15(VCN, inst, mmUVD_SUVD_CGC_GATE2, data); @@ -776,8 +776,8 @@ static void vcn_v3_0_disable_clock_gating(struct amdgpu_device *adev, int inst) | UVD_SUVD_CGC_CTRL__SMPA_MODE_MASK | UVD_SUVD_CGC_CTRL__MPBE0_MODE_MASK | UVD_SUVD_CGC_CTRL__MPBE1_MODE_MASK - | 0x00008000 - | 0x00010000 + | UVD_SUVD_CGC_CTRL__SIT_AV1_MODE_MASK + | UVD_SUVD_CGC_CTRL__SDB_AV1_MODE_MASK | UVD_SUVD_CGC_CTRL__MPC1_MODE_MASK | UVD_SUVD_CGC_CTRL__FBC_PCLK_MASK | UVD_SUVD_CGC_CTRL__FBC_CCLK_MASK); @@ -892,8 +892,8 @@ static void vcn_v3_0_enable_clock_gating(struct amdgpu_device *adev, int inst) | UVD_SUVD_CGC_CTRL__SMPA_MODE_MASK | UVD_SUVD_CGC_CTRL__MPBE0_MODE_MASK | UVD_SUVD_CGC_CTRL__MPBE1_MODE_MASK - | 0x00008000 - | 0x00010000 + | UVD_SUVD_CGC_CTRL__SIT_AV1_MODE_MASK + | UVD_SUVD_CGC_CTRL__SDB_AV1_MODE_MASK | UVD_SUVD_CGC_CTRL__MPC1_MODE_MASK | UVD_SUVD_CGC_CTRL__FBC_PCLK_MASK | UVD_SUVD_CGC_CTRL__FBC_CCLK_MASK); |