diff options
| author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-08-07 13:18:06 +0000 |
|---|---|---|
| committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-08-07 13:18:06 +0000 |
| commit | 638a9e433ecd61e64761352dbec1fa4f5874c941 (patch) | |
| tree | fdbff74a238d7a5a7d1cef071b7230bc064b9f25 /drivers/gpu/drm/amd/amdgpu | |
| parent | Releasing progress-linux version 6.9.12-1~progress7.99u1. (diff) | |
| download | linux-638a9e433ecd61e64761352dbec1fa4f5874c941.tar.xz linux-638a9e433ecd61e64761352dbec1fa4f5874c941.zip | |
Merging upstream version 6.10.3.
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
145 files changed, 3259 insertions, 1025 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 4536c8ad0e..078f588e99 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -23,7 +23,7 @@ # Makefile for the drm device driver. This driver provides support for the # Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. -FULL_AMD_PATH=$(srctree)/$(src)/.. +FULL_AMD_PATH=$(src)/.. DISPLAY_FOLDER_NAME=display FULL_AMD_DISPLAY_PATH = $(FULL_AMD_PATH)/$(DISPLAY_FOLDER_NAME) @@ -70,7 +70,8 @@ amdgpu-y += amdgpu_device.o amdgpu_doorbell_mgr.o amdgpu_kms.o \ amdgpu_cs.o amdgpu_bios.o amdgpu_benchmark.o \ atombios_dp.o amdgpu_afmt.o amdgpu_trace_points.o \ atombios_encoders.o amdgpu_sa.o atombios_i2c.o \ - amdgpu_dma_buf.o amdgpu_vm.o amdgpu_vm_pt.o amdgpu_ib.o amdgpu_pll.o \ + amdgpu_dma_buf.o amdgpu_vm.o amdgpu_vm_pt.o amdgpu_vm_tlb_fence.o \ + amdgpu_ib.o amdgpu_pll.o \ amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o \ amdgpu_gtt_mgr.o amdgpu_preempt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o \ amdgpu_atomfirmware.o amdgpu_vf_error.o amdgpu_sched.o \ @@ -80,7 +81,7 @@ amdgpu-y += amdgpu_device.o amdgpu_doorbell_mgr.o amdgpu_kms.o \ amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o \ amdgpu_fw_attestation.o amdgpu_securedisplay.o \ amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o \ - amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o + amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o amdgpu_dev_coredump.o amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o @@ -247,7 +248,8 @@ amdgpu-y += \ smuio_v11_0_6.o \ smuio_v13_0.o \ smuio_v13_0_3.o \ - smuio_v13_0_6.o + smuio_v13_0_6.o \ + smuio_v14_0_2.o # add reset block amdgpu-y += \ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index b3b8464720..f87d53e183 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -139,6 +139,14 @@ enum amdgpu_ss { AMDGPU_SS_DRV_UNLOAD }; +struct amdgpu_hwip_reg_entry { + u32 hwip; + u32 inst; + u32 seg; + u32 reg_offset; + const char *reg_name; +}; + struct amdgpu_watchdog_timer { bool timeout_fatal_disable; uint32_t period; /* maxCycles = (1 << period), the number of cycles before a timeout */ @@ -494,6 +502,7 @@ struct amdgpu_wb { uint64_t gpu_addr; u32 num_wb; /* Number of wb slots actually reserved for amdgpu. */ unsigned long used[DIV_ROUND_UP(AMDGPU_MAX_WB, BITS_PER_LONG)]; + spinlock_t lock; }; int amdgpu_device_wb_get(struct amdgpu_device *adev, u32 *wb); @@ -606,7 +615,7 @@ struct amdgpu_asic_funcs { /* PCIe replay counter */ uint64_t (*get_pcie_replay_count)(struct amdgpu_device *adev); /* device supports BACO */ - bool (*supports_baco)(struct amdgpu_device *adev); + int (*supports_baco)(struct amdgpu_device *adev); /* pre asic_init quirks */ void (*pre_asic_init)(struct amdgpu_device *adev); /* enter/exit umd stable pstate */ @@ -1408,7 +1417,8 @@ bool amdgpu_device_supports_atpx(struct drm_device *dev); bool amdgpu_device_supports_px(struct drm_device *dev); bool amdgpu_device_supports_boco(struct drm_device *dev); bool amdgpu_device_supports_smart_shift(struct drm_device *dev); -bool amdgpu_device_supports_baco(struct drm_device *dev); +int amdgpu_device_supports_baco(struct drm_device *dev); +void amdgpu_device_detect_runtime_pm_mode(struct amdgpu_device *adev); bool amdgpu_device_is_peer_accessible(struct amdgpu_device *adev, struct amdgpu_device *peer_adev); int amdgpu_device_baco_enter(struct drm_device *dev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c index 493982f946..c50202215f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c @@ -28,7 +28,7 @@ #define ACA_BANK_HWID(type, hwid, mcatype) [ACA_HWIP_TYPE_##type] = {hwid, mcatype} -typedef int bank_handler_t(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type, void *data); +typedef int bank_handler_t(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type, void *data); struct aca_banks { int nr_banks; @@ -86,7 +86,7 @@ static void aca_banks_release(struct aca_banks *banks) } } -static int aca_smu_get_valid_aca_count(struct amdgpu_device *adev, enum aca_error_type type, u32 *count) +static int aca_smu_get_valid_aca_count(struct amdgpu_device *adev, enum aca_smu_type type, u32 *count) { struct amdgpu_aca *aca = &adev->aca; const struct aca_smu_funcs *smu_funcs = aca->smu_funcs; @@ -116,20 +116,22 @@ static struct aca_regs_dump { {"CONTROL_MASK", ACA_REG_IDX_CTL_MASK}, }; -static void aca_smu_bank_dump(struct amdgpu_device *adev, int idx, int total, struct aca_bank *bank) +static void aca_smu_bank_dump(struct amdgpu_device *adev, int idx, int total, struct aca_bank *bank, + struct ras_query_context *qctx) { + u64 event_id = qctx ? qctx->event_id : 0ULL; int i; - dev_info(adev->dev, HW_ERR "Accelerator Check Architecture events logged\n"); + RAS_EVENT_LOG(adev, event_id, HW_ERR "Accelerator Check Architecture events logged\n"); /* plus 1 for output format, e.g: ACA[08/08]: xxxx */ for (i = 0; i < ARRAY_SIZE(aca_regs); i++) - dev_info(adev->dev, HW_ERR "ACA[%02d/%02d].%s=0x%016llx\n", - idx + 1, total, aca_regs[i].name, bank->regs[aca_regs[i].reg_idx]); + RAS_EVENT_LOG(adev, event_id, HW_ERR "ACA[%02d/%02d].%s=0x%016llx\n", + idx + 1, total, aca_regs[i].name, bank->regs[aca_regs[i].reg_idx]); } -static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_error_type type, +static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_smu_type type, int start, int count, - struct aca_banks *banks) + struct aca_banks *banks, struct ras_query_context *qctx) { struct amdgpu_aca *aca = &adev->aca; const struct aca_smu_funcs *smu_funcs = aca->smu_funcs; @@ -143,13 +145,12 @@ static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_erro return -EOPNOTSUPP; switch (type) { - case ACA_ERROR_TYPE_UE: + case ACA_SMU_TYPE_UE: max_count = smu_funcs->max_ue_bank_count; break; - case ACA_ERROR_TYPE_CE: + case ACA_SMU_TYPE_CE: max_count = smu_funcs->max_ce_bank_count; break; - case ACA_ERROR_TYPE_DEFERRED: default: return -EINVAL; } @@ -164,7 +165,9 @@ static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_erro if (ret) return ret; - aca_smu_bank_dump(adev, i, count, &bank); + bank.type = type; + + aca_smu_bank_dump(adev, i, count, &bank, qctx); ret = aca_banks_add_bank(banks, &bank); if (ret) @@ -195,7 +198,7 @@ static bool aca_bank_hwip_is_matched(struct aca_bank *bank, enum aca_hwip_type t return hwip->hwid == hwid && hwip->mcatype == mcatype; } -static bool aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type) +static bool aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type) { const struct aca_bank_ops *bank_ops = handle->bank_ops; @@ -273,59 +276,49 @@ static struct aca_bank_error *get_bank_error(struct aca_error *aerr, struct aca_ return new_bank_error(aerr, info); } -static int aca_log_errors(struct aca_handle *handle, enum aca_error_type type, - struct aca_bank_report *report) +int aca_error_cache_log_bank_error(struct aca_handle *handle, struct aca_bank_info *info, + enum aca_error_type type, u64 count) { struct aca_error_cache *error_cache = &handle->error_cache; struct aca_bank_error *bank_error; struct aca_error *aerr; - if (!handle || !report) + if (!handle || !info || type >= ACA_ERROR_TYPE_COUNT) return -EINVAL; - if (!report->count[type]) + if (!count) return 0; aerr = &error_cache->errors[type]; - bank_error = get_bank_error(aerr, &report->info); + bank_error = get_bank_error(aerr, info); if (!bank_error) return -ENOMEM; - bank_error->count[type] += report->count[type]; + bank_error->count += count; return 0; } -static int aca_generate_bank_report(struct aca_handle *handle, struct aca_bank *bank, - enum aca_error_type type, struct aca_bank_report *report) +static int aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type) { const struct aca_bank_ops *bank_ops = handle->bank_ops; - if (!bank || !report) + if (!bank) return -EINVAL; - if (!bank_ops->aca_bank_generate_report) + if (!bank_ops->aca_bank_parser) return -EOPNOTSUPP; - memset(report, 0, sizeof(*report)); - return bank_ops->aca_bank_generate_report(handle, bank, type, - report, handle->data); + return bank_ops->aca_bank_parser(handle, bank, type, + handle->data); } static int handler_aca_log_bank_error(struct aca_handle *handle, struct aca_bank *bank, - enum aca_error_type type, void *data) + enum aca_smu_type type, void *data) { - struct aca_bank_report report; int ret; - ret = aca_generate_bank_report(handle, bank, type, &report); - if (ret) - return ret; - - if (!report.count[type]) - return 0; - - ret = aca_log_errors(handle, type, &report); + ret = aca_bank_parser(handle, bank, type); if (ret) return ret; @@ -333,7 +326,7 @@ static int handler_aca_log_bank_error(struct aca_handle *handle, struct aca_bank } static int aca_dispatch_bank(struct aca_handle_manager *mgr, struct aca_bank *bank, - enum aca_error_type type, bank_handler_t handler, void *data) + enum aca_smu_type type, bank_handler_t handler, void *data) { struct aca_handle *handle; int ret; @@ -354,7 +347,7 @@ static int aca_dispatch_bank(struct aca_handle_manager *mgr, struct aca_bank *ba } static int aca_dispatch_banks(struct aca_handle_manager *mgr, struct aca_banks *banks, - enum aca_error_type type, bank_handler_t handler, void *data) + enum aca_smu_type type, bank_handler_t handler, void *data) { struct aca_bank_node *node; struct aca_bank *bank; @@ -378,8 +371,28 @@ static int aca_dispatch_banks(struct aca_handle_manager *mgr, struct aca_banks * return 0; } -static int aca_banks_update(struct amdgpu_device *adev, enum aca_error_type type, - bank_handler_t handler, void *data) +static bool aca_bank_should_update(struct amdgpu_device *adev, enum aca_smu_type type) +{ + struct amdgpu_aca *aca = &adev->aca; + bool ret = true; + + /* + * Because the UE Valid MCA count will only be cleared after reset, + * in order to avoid repeated counting of the error count, + * the aca bank is only updated once during the gpu recovery stage. + */ + if (type == ACA_SMU_TYPE_UE) { + if (amdgpu_ras_intr_triggered()) + ret = atomic_cmpxchg(&aca->ue_update_flag, 0, 1) == 0; + else + atomic_set(&aca->ue_update_flag, 0); + } + + return ret; +} + +static int aca_banks_update(struct amdgpu_device *adev, enum aca_smu_type type, + bank_handler_t handler, struct ras_query_context *qctx, void *data) { struct amdgpu_aca *aca = &adev->aca; struct aca_banks banks; @@ -389,9 +402,8 @@ static int aca_banks_update(struct amdgpu_device *adev, enum aca_error_type type if (list_empty(&aca->mgr.list)) return 0; - /* NOTE: pmfw is only support UE and CE */ - if (type == ACA_ERROR_TYPE_DEFERRED) - type = ACA_ERROR_TYPE_CE; + if (!aca_bank_should_update(adev, type)) + return 0; ret = aca_smu_get_valid_aca_count(adev, type, &count); if (ret) @@ -402,7 +414,7 @@ static int aca_banks_update(struct amdgpu_device *adev, enum aca_error_type type aca_banks_init(&banks); - ret = aca_smu_get_valid_aca_banks(adev, type, 0, count, &banks); + ret = aca_smu_get_valid_aca_banks(adev, type, 0, count, &banks, qctx); if (ret) goto err_release_banks; @@ -431,7 +443,7 @@ static int aca_log_aca_error_data(struct aca_bank_error *bank_error, enum aca_er if (type >= ACA_ERROR_TYPE_COUNT) return -EINVAL; - count = bank_error->count[type]; + count = bank_error->count; if (!count) return 0; @@ -447,6 +459,8 @@ static int aca_log_aca_error_data(struct aca_bank_error *bank_error, enum aca_er amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, NULL, count); break; case ACA_ERROR_TYPE_DEFERRED: + amdgpu_ras_error_statistic_de_count(err_data, &mcm_info, NULL, count); + break; default: break; } @@ -477,12 +491,25 @@ out_unlock: } static int __aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle, enum aca_error_type type, - struct ras_err_data *err_data) + struct ras_err_data *err_data, struct ras_query_context *qctx) { + enum aca_smu_type smu_type; int ret; + switch (type) { + case ACA_ERROR_TYPE_UE: + smu_type = ACA_SMU_TYPE_UE; + break; + case ACA_ERROR_TYPE_CE: + case ACA_ERROR_TYPE_DEFERRED: + smu_type = ACA_SMU_TYPE_CE; + break; + default: + return -EINVAL; + } + /* udpate aca bank to aca source error_cache first */ - ret = aca_banks_update(adev, type, handler_aca_log_bank_error, NULL); + ret = aca_banks_update(adev, smu_type, handler_aca_log_bank_error, qctx, NULL); if (ret) return ret; @@ -498,10 +525,9 @@ static bool aca_handle_is_valid(struct aca_handle *handle) } int amdgpu_aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle, - enum aca_error_type type, void *data) + enum aca_error_type type, struct ras_err_data *err_data, + struct ras_query_context *qctx) { - struct ras_err_data *err_data = (struct ras_err_data *)data; - if (!handle || !err_data) return -EINVAL; @@ -511,7 +537,7 @@ int amdgpu_aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *han if (!(BIT(type) & handle->mask)) return 0; - return __aca_get_error_data(adev, handle, type, err_data); + return __aca_get_error_data(adev, handle, type, err_data, qctx); } static void aca_error_init(struct aca_error *aerr, enum aca_error_type type) @@ -668,6 +694,8 @@ int amdgpu_aca_init(struct amdgpu_device *adev) struct amdgpu_aca *aca = &adev->aca; int ret; + atomic_set(&aca->ue_update_flag, 0); + ret = aca_manager_init(&aca->mgr); if (ret) return ret; @@ -680,6 +708,8 @@ void amdgpu_aca_fini(struct amdgpu_device *adev) struct amdgpu_aca *aca = &adev->aca; aca_manager_fini(&aca->mgr); + + atomic_set(&aca->ue_update_flag, 0); } int amdgpu_aca_reset(struct amdgpu_device *adev) @@ -723,23 +753,13 @@ int aca_bank_info_decode(struct aca_bank *bank, struct aca_bank_info *info) static int aca_bank_get_error_code(struct amdgpu_device *adev, struct aca_bank *bank) { - int error_code; - - switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { - case IP_VERSION(13, 0, 6): - if (!(adev->flags & AMD_IS_APU) && adev->pm.fw_version >= 0x00555600) { - error_code = ACA_REG__SYND__ERRORINFORMATION(bank->regs[ACA_REG_IDX_SYND]); - return error_code & 0xff; - } - break; - default: - break; - } + struct amdgpu_aca *aca = &adev->aca; + const struct aca_smu_funcs *smu_funcs = aca->smu_funcs; - /* NOTE: the true error code is encoded in status.errorcode[0:7] */ - error_code = ACA_REG__STATUS__ERRORCODE(bank->regs[ACA_REG_IDX_STATUS]); + if (!smu_funcs || !smu_funcs->parse_error_code) + return -EOPNOTSUPP; - return error_code & 0xff; + return smu_funcs->parse_error_code(adev, bank); } int aca_bank_check_error_codes(struct amdgpu_device *adev, struct aca_bank *bank, int *err_codes, int size) @@ -750,6 +770,9 @@ int aca_bank_check_error_codes(struct amdgpu_device *adev, struct aca_bank *bank return -EINVAL; error_code = aca_bank_get_error_code(adev, bank); + if (error_code < 0) + return error_code; + for (i = 0; i < size; i++) { if (err_codes[i] == error_code) return 0; @@ -784,7 +807,7 @@ static int amdgpu_aca_smu_debug_mode_set(void *data, u64 val) return 0; } -static void aca_dump_entry(struct seq_file *m, struct aca_bank *bank, enum aca_error_type type, int idx) +static void aca_dump_entry(struct seq_file *m, struct aca_bank *bank, enum aca_smu_type type, int idx) { struct aca_bank_info info; int i, ret; @@ -793,7 +816,7 @@ static void aca_dump_entry(struct seq_file *m, struct aca_bank *bank, enum aca_e if (ret) return; - seq_printf(m, "aca entry[%d].type: %s\n", idx, type == ACA_ERROR_TYPE_UE ? "UE" : "CE"); + seq_printf(m, "aca entry[%d].type: %s\n", idx, type == ACA_SMU_TYPE_UE ? "UE" : "CE"); seq_printf(m, "aca entry[%d].info: socketid:%d aid:%d hwid:0x%03x mcatype:0x%04x\n", idx, info.socket_id, info.die_id, info.hwid, info.mcatype); @@ -807,7 +830,7 @@ struct aca_dump_context { }; static int handler_aca_bank_dump(struct aca_handle *handle, struct aca_bank *bank, - enum aca_error_type type, void *data) + enum aca_smu_type type, void *data) { struct aca_dump_context *ctx = (struct aca_dump_context *)data; @@ -816,7 +839,7 @@ static int handler_aca_bank_dump(struct aca_handle *handle, struct aca_bank *ban return handler_aca_log_bank_error(handle, bank, type, NULL); } -static int aca_dump_show(struct seq_file *m, enum aca_error_type type) +static int aca_dump_show(struct seq_file *m, enum aca_smu_type type) { struct amdgpu_device *adev = (struct amdgpu_device *)m->private; struct aca_dump_context context = { @@ -824,12 +847,12 @@ static int aca_dump_show(struct seq_file *m, enum aca_error_type type) .idx = 0, }; - return aca_banks_update(adev, type, handler_aca_bank_dump, (void *)&context); + return aca_banks_update(adev, type, handler_aca_bank_dump, NULL, (void *)&context); } static int aca_dump_ce_show(struct seq_file *m, void *unused) { - return aca_dump_show(m, ACA_ERROR_TYPE_CE); + return aca_dump_show(m, ACA_SMU_TYPE_CE); } static int aca_dump_ce_open(struct inode *inode, struct file *file) @@ -847,7 +870,7 @@ static const struct file_operations aca_ce_dump_debug_fops = { static int aca_dump_ue_show(struct seq_file *m, void *unused) { - return aca_dump_show(m, ACA_ERROR_TYPE_UE); + return aca_dump_show(m, ACA_SMU_TYPE_UE); } static int aca_dump_ue_open(struct inode *inode, struct file *file) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h index 2da50e0958..5ef6b745f2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h @@ -26,6 +26,9 @@ #include <linux/list.h> +struct ras_err_data; +struct ras_query_context; + #define ACA_MAX_REGS_COUNT (16) #define ACA_REG_FIELD(x, h, l) (((x) & GENMASK_ULL(h, l)) >> l) @@ -99,7 +102,14 @@ enum aca_error_type { ACA_ERROR_TYPE_COUNT }; +enum aca_smu_type { + ACA_SMU_TYPE_UE = 0, + ACA_SMU_TYPE_CE, + ACA_SMU_TYPE_COUNT, +}; + struct aca_bank { + enum aca_smu_type type; u64 regs[ACA_MAX_REGS_COUNT]; }; @@ -115,15 +125,10 @@ struct aca_bank_info { int mcatype; }; -struct aca_bank_report { - struct aca_bank_info info; - u64 count[ACA_ERROR_TYPE_COUNT]; -}; - struct aca_bank_error { struct list_head node; struct aca_bank_info info; - u64 count[ACA_ERROR_TYPE_COUNT]; + u64 count; }; struct aca_error { @@ -157,9 +162,8 @@ struct aca_handle { }; struct aca_bank_ops { - int (*aca_bank_generate_report)(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type, - struct aca_bank_report *report, void *data); - bool (*aca_bank_is_valid)(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type, + int (*aca_bank_parser)(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type, void *data); + bool (*aca_bank_is_valid)(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type, void *data); }; @@ -167,13 +171,15 @@ struct aca_smu_funcs { int max_ue_bank_count; int max_ce_bank_count; int (*set_debug_mode)(struct amdgpu_device *adev, bool enable); - int (*get_valid_aca_count)(struct amdgpu_device *adev, enum aca_error_type type, u32 *count); - int (*get_valid_aca_bank)(struct amdgpu_device *adev, enum aca_error_type type, int idx, struct aca_bank *bank); + int (*get_valid_aca_count)(struct amdgpu_device *adev, enum aca_smu_type type, u32 *count); + int (*get_valid_aca_bank)(struct amdgpu_device *adev, enum aca_smu_type type, int idx, struct aca_bank *bank); + int (*parse_error_code)(struct amdgpu_device *adev, struct aca_bank *bank); }; struct amdgpu_aca { struct aca_handle_manager mgr; const struct aca_smu_funcs *smu_funcs; + atomic_t ue_update_flag; bool is_enabled; }; @@ -196,7 +202,10 @@ int amdgpu_aca_add_handle(struct amdgpu_device *adev, struct aca_handle *handle, const char *name, const struct aca_info *aca_info, void *data); void amdgpu_aca_remove_handle(struct aca_handle *handle); int amdgpu_aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle, - enum aca_error_type type, void *data); + enum aca_error_type type, struct ras_err_data *err_data, + struct ras_query_context *qctx); int amdgpu_aca_smu_set_debug_mode(struct amdgpu_device *adev, bool en); void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root); +int aca_error_cache_log_bank_error(struct aca_handle *handle, struct aca_bank_info *info, + enum aca_error_type type, u64 count); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c index 6d72355ac4..bf6c4a0d05 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c @@ -637,6 +637,8 @@ static const struct amd_ip_funcs acp_ip_funcs = { .soft_reset = acp_soft_reset, .set_clockgating_state = acp_set_clockgating_state, .set_powergating_state = acp_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version acp_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index 7291c3fd8c..e3738d4172 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -750,10 +750,17 @@ bool amdgpu_amdkfd_is_fed(struct amdgpu_device *adev) return amdgpu_ras_get_fed_status(adev); } +void amdgpu_amdkfd_ras_pasid_poison_consumption_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block, uint16_t pasid, + pasid_notify pasid_fn, void *data, uint32_t reset) +{ + amdgpu_umc_pasid_poison_handler(adev, block, pasid, pasid_fn, data, reset); +} + void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev, - enum amdgpu_ras_block block, bool reset) + enum amdgpu_ras_block block, uint32_t reset) { - amdgpu_umc_poison_handler(adev, block, reset); + amdgpu_umc_pasid_poison_handler(adev, block, 0, NULL, NULL, reset); } int amdgpu_amdkfd_send_close_event_drain_irq(struct amdgpu_device *adev, @@ -772,12 +779,20 @@ int amdgpu_amdkfd_send_close_event_drain_irq(struct amdgpu_device *adev, return 0; } -bool amdgpu_amdkfd_ras_query_utcl2_poison_status(struct amdgpu_device *adev) +bool amdgpu_amdkfd_ras_query_utcl2_poison_status(struct amdgpu_device *adev, + int hub_inst, int hub_type) { - if (adev->gfx.ras && adev->gfx.ras->query_utcl2_poison_status) - return adev->gfx.ras->query_utcl2_poison_status(adev); - else - return false; + if (!hub_type) { + if (adev->gfxhub.funcs->query_utcl2_poison_status) + return adev->gfxhub.funcs->query_utcl2_poison_status(adev, hub_inst); + else + return false; + } else { + if (adev->mmhub.funcs->query_utcl2_poison_status) + return adev->mmhub.funcs->query_utcl2_poison_status(adev, hub_inst); + else + return false; + } } int amdgpu_amdkfd_check_and_lock_kfd(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index 0ef223c2af..1de021ebdd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -336,12 +336,18 @@ void amdgpu_amdkfd_debug_mem_fence(struct amdgpu_device *adev); int amdgpu_amdkfd_get_tile_config(struct amdgpu_device *adev, struct tile_config *config); void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev, - enum amdgpu_ras_block block, bool reset); + enum amdgpu_ras_block block, uint32_t reset); + +void amdgpu_amdkfd_ras_pasid_poison_consumption_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block, uint16_t pasid, + pasid_notify pasid_fn, void *data, uint32_t reset); + bool amdgpu_amdkfd_is_fed(struct amdgpu_device *adev); bool amdgpu_amdkfd_bo_mapped_to_dev(struct amdgpu_device *adev, struct kgd_mem *mem); void amdgpu_amdkfd_block_mmu_notifications(void *p); int amdgpu_amdkfd_criu_resume(void *p); -bool amdgpu_amdkfd_ras_query_utcl2_poison_status(struct amdgpu_device *adev); +bool amdgpu_amdkfd_ras_query_utcl2_poison_status(struct amdgpu_device *adev, + int hub_inst, int hub_type); int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev, uint64_t size, u32 alloc_flag, int8_t xcp_id); void amdgpu_amdkfd_unreserve_mem_limit(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 69810b3f1c..3ab6c3aa0a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c @@ -881,6 +881,7 @@ uint32_t kgd_gfx_v10_set_wave_launch_mode(struct amdgpu_device *adev, } #define TCP_WATCH_STRIDE (mmTCP_WATCH1_ADDR_H - mmTCP_WATCH0_ADDR_H) +#define SQ_WATCH_STRIDE (mmSQ_WATCH1_ADDR_H - mmSQ_WATCH0_ADDR_H) uint32_t kgd_gfx_v10_set_address_watch(struct amdgpu_device *adev, uint64_t watch_address, uint32_t watch_address_mask, @@ -889,55 +890,93 @@ uint32_t kgd_gfx_v10_set_address_watch(struct amdgpu_device *adev, uint32_t debug_vmid, uint32_t inst) { + /* SQ_WATCH?_ADDR_* and TCP_WATCH?_ADDR_* are programmed with the + * same values. + */ uint32_t watch_address_high; uint32_t watch_address_low; - uint32_t watch_address_cntl; - - watch_address_cntl = 0; + uint32_t tcp_watch_address_cntl; + uint32_t sq_watch_address_cntl; watch_address_low = lower_32_bits(watch_address); watch_address_high = upper_32_bits(watch_address) & 0xffff; - watch_address_cntl = REG_SET_FIELD(watch_address_cntl, + tcp_watch_address_cntl = 0; + tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl, TCP_WATCH0_CNTL, VMID, debug_vmid); - watch_address_cntl = REG_SET_FIELD(watch_address_cntl, + tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl, TCP_WATCH0_CNTL, MODE, watch_mode); - watch_address_cntl = REG_SET_FIELD(watch_address_cntl, + tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl, TCP_WATCH0_CNTL, MASK, watch_address_mask >> 7); + sq_watch_address_cntl = 0; + sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl, + SQ_WATCH0_CNTL, + VMID, + debug_vmid); + sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl, + SQ_WATCH0_CNTL, + MODE, + watch_mode); + sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl, + SQ_WATCH0_CNTL, + MASK, + watch_address_mask >> 6); + /* Turning off this watch point until we set all the registers */ - watch_address_cntl = REG_SET_FIELD(watch_address_cntl, + tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl, TCP_WATCH0_CNTL, VALID, 0); - WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) + (watch_id * TCP_WATCH_STRIDE)), - watch_address_cntl); + tcp_watch_address_cntl); + + sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl, + SQ_WATCH0_CNTL, + VALID, + 0); + WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_CNTL) + + (watch_id * SQ_WATCH_STRIDE)), + sq_watch_address_cntl); + /* Program {TCP,SQ}_WATCH?_ADDR* */ WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_H) + (watch_id * TCP_WATCH_STRIDE)), watch_address_high); - WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_L) + (watch_id * TCP_WATCH_STRIDE)), watch_address_low); + WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_ADDR_H) + + (watch_id * SQ_WATCH_STRIDE)), + watch_address_high); + WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_ADDR_L) + + (watch_id * SQ_WATCH_STRIDE)), + watch_address_low); + /* Enable the watch point */ - watch_address_cntl = REG_SET_FIELD(watch_address_cntl, + tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl, TCP_WATCH0_CNTL, VALID, 1); - WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) + (watch_id * TCP_WATCH_STRIDE)), - watch_address_cntl); + tcp_watch_address_cntl); + + sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl, + SQ_WATCH0_CNTL, + VALID, + 1); + WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_CNTL) + + (watch_id * SQ_WATCH_STRIDE)), + sq_watch_address_cntl); return 0; } @@ -953,8 +992,14 @@ uint32_t kgd_gfx_v10_clear_address_watch(struct amdgpu_device *adev, (watch_id * TCP_WATCH_STRIDE)), watch_address_cntl); + WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_CNTL) + + (watch_id * SQ_WATCH_STRIDE)), + watch_address_cntl); + return 0; } +#undef TCP_WATCH_STRIDE +#undef SQ_WATCH_STRIDE /* kgd_gfx_v10_get_iq_wait_times: Returns the mmCP_IQ_WAIT_TIME1/2 values diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 8975cf41a9..48ad0c04aa 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -196,7 +196,7 @@ int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev, return -EINVAL; vram_size = KFD_XCP_MEMORY_SIZE(adev, xcp_id); - if (adev->gmc.is_app_apu || adev->flags & AMD_IS_APU) { + if (adev->flags & AMD_IS_APU) { system_mem_needed = size; ttm_mem_needed = size; } @@ -233,7 +233,7 @@ int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev, if (adev && xcp_id >= 0) { adev->kfd.vram_used[xcp_id] += vram_needed; adev->kfd.vram_used_aligned[xcp_id] += - (adev->gmc.is_app_apu || adev->flags & AMD_IS_APU) ? + (adev->flags & AMD_IS_APU) ? vram_needed : ALIGN(vram_needed, VRAM_AVAILABLITY_ALIGN); } @@ -261,7 +261,7 @@ void amdgpu_amdkfd_unreserve_mem_limit(struct amdgpu_device *adev, if (adev) { adev->kfd.vram_used[xcp_id] -= size; - if (adev->gmc.is_app_apu || adev->flags & AMD_IS_APU) { + if (adev->flags & AMD_IS_APU) { adev->kfd.vram_used_aligned[xcp_id] -= size; kfd_mem_limit.system_mem_used -= size; kfd_mem_limit.ttm_mem_used -= size; @@ -890,7 +890,7 @@ static int kfd_mem_attach(struct amdgpu_device *adev, struct kgd_mem *mem, * if peer device has large BAR. In contrast, access over xGMI is * allowed for both small and large BAR configurations of peer device */ - if ((adev != bo_adev && !(adev->gmc.is_app_apu || adev->flags & AMD_IS_APU)) && + if ((adev != bo_adev && !(adev->flags & AMD_IS_APU)) && ((mem->domain == AMDGPU_GEM_DOMAIN_VRAM) || (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL) || (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP))) { @@ -1658,7 +1658,7 @@ size_t amdgpu_amdkfd_get_available_memory(struct amdgpu_device *adev, - atomic64_read(&adev->vram_pin_size) - reserved_for_pt; - if (adev->gmc.is_app_apu || adev->flags & AMD_IS_APU) { + if (adev->flags & AMD_IS_APU) { system_mem_available = no_system_mem_limit ? kfd_mem_limit.max_system_mem_limit : kfd_mem_limit.max_system_mem_limit - @@ -1706,7 +1706,7 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu( if (flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) { domain = alloc_domain = AMDGPU_GEM_DOMAIN_VRAM; - if (adev->gmc.is_app_apu || adev->flags & AMD_IS_APU) { + if (adev->flags & AMD_IS_APU) { domain = AMDGPU_GEM_DOMAIN_GTT; alloc_domain = AMDGPU_GEM_DOMAIN_GTT; alloc_flags = 0; @@ -1953,7 +1953,7 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu( if (size) { if (!is_imported && (mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_VRAM || - ((adev->gmc.is_app_apu || adev->flags & AMD_IS_APU) && + ((adev->flags & AMD_IS_APU) && mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_GTT))) *size = bo_size; else @@ -2376,7 +2376,7 @@ static int import_obj_create(struct amdgpu_device *adev, (*mem)->bo = bo; (*mem)->va = va; (*mem)->domain = (bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) && - !(adev->gmc.is_app_apu || adev->flags & AMD_IS_APU) ? + !(adev->flags & AMD_IS_APU) ? AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT; (*mem)->mapped_to_gpu_memory = 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c index 12b4885182..2e13c7c4b2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c @@ -34,6 +34,7 @@ union firmware_info { struct atom_firmware_info_v3_2 v32; struct atom_firmware_info_v3_3 v33; struct atom_firmware_info_v3_4 v34; + struct atom_firmware_info_v3_5 v35; }; /* @@ -887,6 +888,10 @@ int amdgpu_atomfirmware_get_fw_reserved_fb_size(struct amdgpu_device *adev) fw_reserved_fb_size = (firmware_info->v34.fw_reserved_size_in_kb << 10); break; + case 5: + fw_reserved_fb_size = + (firmware_info->v35.fw_reserved_size_in_kb << 10); + break; default: fw_reserved_fb_size = 0; break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c index edc6377ec5..199693369c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c @@ -39,7 +39,7 @@ static int amdgpu_benchmark_do_move(struct amdgpu_device *adev, unsigned size, for (i = 0; i < n; i++) { struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; r = amdgpu_copy_buffer(ring, saddr, daddr, size, NULL, &fence, - false, false, false); + false, false, 0); if (r) goto exit_do_move; r = dma_fence_wait(fence, false); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c new file mode 100644 index 0000000000..c1cb626836 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c @@ -0,0 +1,360 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2024 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include <generated/utsrelease.h> +#include <linux/devcoredump.h> +#include "amdgpu_dev_coredump.h" +#include "atom.h" + +#ifndef CONFIG_DEV_COREDUMP +void amdgpu_coredump(struct amdgpu_device *adev, bool vram_lost, + struct amdgpu_reset_context *reset_context) +{ +} +#else + +const char *hw_ip_names[MAX_HWIP] = { + [GC_HWIP] = "GC", + [HDP_HWIP] = "HDP", + [SDMA0_HWIP] = "SDMA0", + [SDMA1_HWIP] = "SDMA1", + [SDMA2_HWIP] = "SDMA2", + [SDMA3_HWIP] = "SDMA3", + [SDMA4_HWIP] = "SDMA4", + [SDMA5_HWIP] = "SDMA5", + [SDMA6_HWIP] = "SDMA6", + [SDMA7_HWIP] = "SDMA7", + [LSDMA_HWIP] = "LSDMA", + [MMHUB_HWIP] = "MMHUB", + [ATHUB_HWIP] = "ATHUB", + [NBIO_HWIP] = "NBIO", + [MP0_HWIP] = "MP0", + [MP1_HWIP] = "MP1", + [UVD_HWIP] = "UVD/JPEG/VCN", + [VCN1_HWIP] = "VCN1", + [VCE_HWIP] = "VCE", + [VPE_HWIP] = "VPE", + [DF_HWIP] = "DF", + [DCE_HWIP] = "DCE", + [OSSSYS_HWIP] = "OSSSYS", + [SMUIO_HWIP] = "SMUIO", + [PWR_HWIP] = "PWR", + [NBIF_HWIP] = "NBIF", + [THM_HWIP] = "THM", + [CLK_HWIP] = "CLK", + [UMC_HWIP] = "UMC", + [RSMU_HWIP] = "RSMU", + [XGMI_HWIP] = "XGMI", + [DCI_HWIP] = "DCI", + [PCIE_HWIP] = "PCIE", +}; + +static void amdgpu_devcoredump_fw_info(struct amdgpu_device *adev, + struct drm_printer *p) +{ + uint32_t version; + uint32_t feature; + uint8_t smu_program, smu_major, smu_minor, smu_debug; + struct atom_context *ctx = adev->mode_info.atom_context; + + drm_printf(p, "VCE feature version: %u, fw version: 0x%08x\n", + adev->vce.fb_version, adev->vce.fw_version); + drm_printf(p, "UVD feature version: %u, fw version: 0x%08x\n", 0, + adev->uvd.fw_version); + drm_printf(p, "GMC feature version: %u, fw version: 0x%08x\n", 0, + adev->gmc.fw_version); + drm_printf(p, "ME feature version: %u, fw version: 0x%08x\n", + adev->gfx.me_feature_version, adev->gfx.me_fw_version); + drm_printf(p, "PFP feature version: %u, fw version: 0x%08x\n", + adev->gfx.pfp_feature_version, adev->gfx.pfp_fw_version); + drm_printf(p, "CE feature version: %u, fw version: 0x%08x\n", + adev->gfx.ce_feature_version, adev->gfx.ce_fw_version); + drm_printf(p, "RLC feature version: %u, fw version: 0x%08x\n", + adev->gfx.rlc_feature_version, adev->gfx.rlc_fw_version); + + drm_printf(p, "RLC SRLC feature version: %u, fw version: 0x%08x\n", + adev->gfx.rlc_srlc_feature_version, + adev->gfx.rlc_srlc_fw_version); + drm_printf(p, "RLC SRLG feature version: %u, fw version: 0x%08x\n", + adev->gfx.rlc_srlg_feature_version, + adev->gfx.rlc_srlg_fw_version); + drm_printf(p, "RLC SRLS feature version: %u, fw version: 0x%08x\n", + adev->gfx.rlc_srls_feature_version, + adev->gfx.rlc_srls_fw_version); + drm_printf(p, "RLCP feature version: %u, fw version: 0x%08x\n", + adev->gfx.rlcp_ucode_feature_version, + adev->gfx.rlcp_ucode_version); + drm_printf(p, "RLCV feature version: %u, fw version: 0x%08x\n", + adev->gfx.rlcv_ucode_feature_version, + adev->gfx.rlcv_ucode_version); + drm_printf(p, "MEC feature version: %u, fw version: 0x%08x\n", + adev->gfx.mec_feature_version, adev->gfx.mec_fw_version); + + if (adev->gfx.mec2_fw) + drm_printf(p, "MEC2 feature version: %u, fw version: 0x%08x\n", + adev->gfx.mec2_feature_version, + adev->gfx.mec2_fw_version); + + drm_printf(p, "IMU feature version: %u, fw version: 0x%08x\n", 0, + adev->gfx.imu_fw_version); + drm_printf(p, "PSP SOS feature version: %u, fw version: 0x%08x\n", + adev->psp.sos.feature_version, adev->psp.sos.fw_version); + drm_printf(p, "PSP ASD feature version: %u, fw version: 0x%08x\n", + adev->psp.asd_context.bin_desc.feature_version, + adev->psp.asd_context.bin_desc.fw_version); + + drm_printf(p, "TA XGMI feature version: 0x%08x, fw version: 0x%08x\n", + adev->psp.xgmi_context.context.bin_desc.feature_version, + adev->psp.xgmi_context.context.bin_desc.fw_version); + drm_printf(p, "TA RAS feature version: 0x%08x, fw version: 0x%08x\n", + adev->psp.ras_context.context.bin_desc.feature_version, + adev->psp.ras_context.context.bin_desc.fw_version); + drm_printf(p, "TA HDCP feature version: 0x%08x, fw version: 0x%08x\n", + adev->psp.hdcp_context.context.bin_desc.feature_version, + adev->psp.hdcp_context.context.bin_desc.fw_version); + drm_printf(p, "TA DTM feature version: 0x%08x, fw version: 0x%08x\n", + adev->psp.dtm_context.context.bin_desc.feature_version, + adev->psp.dtm_context.context.bin_desc.fw_version); + drm_printf(p, "TA RAP feature version: 0x%08x, fw version: 0x%08x\n", + adev->psp.rap_context.context.bin_desc.feature_version, + adev->psp.rap_context.context.bin_desc.fw_version); + drm_printf(p, + "TA SECURE DISPLAY feature version: 0x%08x, fw version: 0x%08x\n", + adev->psp.securedisplay_context.context.bin_desc.feature_version, + adev->psp.securedisplay_context.context.bin_desc.fw_version); + + /* SMC firmware */ + version = adev->pm.fw_version; + + smu_program = (version >> 24) & 0xff; + smu_major = (version >> 16) & 0xff; + smu_minor = (version >> 8) & 0xff; + smu_debug = (version >> 0) & 0xff; + drm_printf(p, + "SMC feature version: %u, program: %d, fw version: 0x%08x (%d.%d.%d)\n", + 0, smu_program, version, smu_major, smu_minor, smu_debug); + + /* SDMA firmware */ + for (int i = 0; i < adev->sdma.num_instances; i++) { + drm_printf(p, + "SDMA%d feature version: %u, firmware version: 0x%08x\n", + i, adev->sdma.instance[i].feature_version, + adev->sdma.instance[i].fw_version); + } + + drm_printf(p, "VCN feature version: %u, fw version: 0x%08x\n", 0, + adev->vcn.fw_version); + drm_printf(p, "DMCU feature version: %u, fw version: 0x%08x\n", 0, + adev->dm.dmcu_fw_version); + drm_printf(p, "DMCUB feature version: %u, fw version: 0x%08x\n", 0, + adev->dm.dmcub_fw_version); + drm_printf(p, "PSP TOC feature version: %u, fw version: 0x%08x\n", + adev->psp.toc.feature_version, adev->psp.toc.fw_version); + + version = adev->mes.kiq_version & AMDGPU_MES_VERSION_MASK; + feature = (adev->mes.kiq_version & AMDGPU_MES_FEAT_VERSION_MASK) >> + AMDGPU_MES_FEAT_VERSION_SHIFT; + drm_printf(p, "MES_KIQ feature version: %u, fw version: 0x%08x\n", + feature, version); + + version = adev->mes.sched_version & AMDGPU_MES_VERSION_MASK; + feature = (adev->mes.sched_version & AMDGPU_MES_FEAT_VERSION_MASK) >> + AMDGPU_MES_FEAT_VERSION_SHIFT; + drm_printf(p, "MES feature version: %u, fw version: 0x%08x\n", feature, + version); + + drm_printf(p, "VPE feature version: %u, fw version: 0x%08x\n", + adev->vpe.feature_version, adev->vpe.fw_version); + + drm_printf(p, "\nVBIOS Information\n"); + drm_printf(p, "vbios name : %s\n", ctx->name); + drm_printf(p, "vbios pn : %s\n", ctx->vbios_pn); + drm_printf(p, "vbios version : %d\n", ctx->version); + drm_printf(p, "vbios ver_str : %s\n", ctx->vbios_ver_str); + drm_printf(p, "vbios date : %s\n", ctx->date); +} + +static ssize_t +amdgpu_devcoredump_read(char *buffer, loff_t offset, size_t count, + void *data, size_t datalen) +{ + struct drm_printer p; + struct amdgpu_coredump_info *coredump = data; + struct drm_print_iterator iter; + struct amdgpu_vm_fault_info *fault_info; + int i, ver; + + iter.data = buffer; + iter.offset = 0; + iter.start = offset; + iter.remain = count; + + p = drm_coredump_printer(&iter); + + drm_printf(&p, "**** AMDGPU Device Coredump ****\n"); + drm_printf(&p, "version: " AMDGPU_COREDUMP_VERSION "\n"); + drm_printf(&p, "kernel: " UTS_RELEASE "\n"); + drm_printf(&p, "module: " KBUILD_MODNAME "\n"); + drm_printf(&p, "time: %lld.%09ld\n", coredump->reset_time.tv_sec, + coredump->reset_time.tv_nsec); + + if (coredump->reset_task_info.pid) + drm_printf(&p, "process_name: %s PID: %d\n", + coredump->reset_task_info.process_name, + coredump->reset_task_info.pid); + + /* GPU IP's information of the SOC */ + drm_printf(&p, "\nIP Information\n"); + drm_printf(&p, "SOC Family: %d\n", coredump->adev->family); + drm_printf(&p, "SOC Revision id: %d\n", coredump->adev->rev_id); + drm_printf(&p, "SOC External Revision id: %d\n", coredump->adev->external_rev_id); + + for (int i = 1; i < MAX_HWIP; i++) { + for (int j = 0; j < HWIP_MAX_INSTANCE; j++) { + ver = coredump->adev->ip_versions[i][j]; + if (ver) + drm_printf(&p, "HWIP: %s[%d][%d]: v%d.%d.%d.%d.%d\n", + hw_ip_names[i], i, j, + IP_VERSION_MAJ(ver), + IP_VERSION_MIN(ver), + IP_VERSION_REV(ver), + IP_VERSION_VARIANT(ver), + IP_VERSION_SUBREV(ver)); + } + } + + /* IP firmware information */ + drm_printf(&p, "\nIP Firmwares\n"); + amdgpu_devcoredump_fw_info(coredump->adev, &p); + + if (coredump->ring) { + drm_printf(&p, "\nRing timed out details\n"); + drm_printf(&p, "IP Type: %d Ring Name: %s\n", + coredump->ring->funcs->type, + coredump->ring->name); + } + + /* Add page fault information */ + fault_info = &coredump->adev->vm_manager.fault_info; + drm_printf(&p, "\n[%s] Page fault observed\n", + fault_info->vmhub ? "mmhub" : "gfxhub"); + drm_printf(&p, "Faulty page starting at address: 0x%016llx\n", fault_info->addr); + drm_printf(&p, "Protection fault status register: 0x%x\n\n", fault_info->status); + + /* dump the ip state for each ip */ + drm_printf(&p, "IP Dump\n"); + for (int i = 0; i < coredump->adev->num_ip_blocks; i++) { + if (coredump->adev->ip_blocks[i].version->funcs->print_ip_state) { + drm_printf(&p, "IP: %s\n", + coredump->adev->ip_blocks[i] + .version->funcs->name); + coredump->adev->ip_blocks[i] + .version->funcs->print_ip_state( + (void *)coredump->adev, &p); + drm_printf(&p, "\n"); + } + } + + /* Add ring buffer information */ + drm_printf(&p, "Ring buffer information\n"); + for (int i = 0; i < coredump->adev->num_rings; i++) { + int j = 0; + struct amdgpu_ring *ring = coredump->adev->rings[i]; + + drm_printf(&p, "ring name: %s\n", ring->name); + drm_printf(&p, "Rptr: 0x%llx Wptr: 0x%llx RB mask: %x\n", + amdgpu_ring_get_rptr(ring), + amdgpu_ring_get_wptr(ring), + ring->buf_mask); + drm_printf(&p, "Ring size in dwords: %d\n", + ring->ring_size / 4); + drm_printf(&p, "Ring contents\n"); + drm_printf(&p, "Offset \t Value\n"); + + while (j < ring->ring_size) { + drm_printf(&p, "0x%x \t 0x%x\n", j, ring->ring[j / 4]); + j += 4; + } + } + + if (coredump->reset_vram_lost) + drm_printf(&p, "VRAM is lost due to GPU reset!\n"); + if (coredump->adev->reset_info.num_regs) { + drm_printf(&p, "AMDGPU register dumps:\nOffset: Value:\n"); + + for (i = 0; i < coredump->adev->reset_info.num_regs; i++) + drm_printf(&p, "0x%08x: 0x%08x\n", + coredump->adev->reset_info.reset_dump_reg_list[i], + coredump->adev->reset_info.reset_dump_reg_value[i]); + } + + return count - iter.remain; +} + +static void amdgpu_devcoredump_free(void *data) +{ + kfree(data); +} + +void amdgpu_coredump(struct amdgpu_device *adev, bool vram_lost, + struct amdgpu_reset_context *reset_context) +{ + struct amdgpu_coredump_info *coredump; + struct drm_device *dev = adev_to_drm(adev); + struct amdgpu_job *job = reset_context->job; + struct drm_sched_job *s_job; + + coredump = kzalloc(sizeof(*coredump), GFP_NOWAIT); + + if (!coredump) { + DRM_ERROR("%s: failed to allocate memory for coredump\n", __func__); + return; + } + + coredump->reset_vram_lost = vram_lost; + + if (reset_context->job && reset_context->job->vm) { + struct amdgpu_task_info *ti; + struct amdgpu_vm *vm = reset_context->job->vm; + + ti = amdgpu_vm_get_task_info_vm(vm); + if (ti) { + coredump->reset_task_info = *ti; + amdgpu_vm_put_task_info(ti); + } + } + + if (job) { + s_job = &job->base; + coredump->ring = to_amdgpu_ring(s_job->sched); + } + + coredump->adev = adev; + + ktime_get_ts64(&coredump->reset_time); + + dev_coredumpm(dev->dev, THIS_MODULE, coredump, 0, GFP_NOWAIT, + amdgpu_devcoredump_read, amdgpu_devcoredump_free); +} +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.h new file mode 100644 index 0000000000..52459512cb --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.h @@ -0,0 +1,47 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2024 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_DEV_COREDUMP_H__ +#define __AMDGPU_DEV_COREDUMP_H__ + +#include "amdgpu.h" +#include "amdgpu_reset.h" + +#ifdef CONFIG_DEV_COREDUMP + +#define AMDGPU_COREDUMP_VERSION "1" + +struct amdgpu_coredump_info { + struct amdgpu_device *adev; + struct amdgpu_task_info reset_task_info; + struct timespec64 reset_time; + bool reset_vram_lost; + struct amdgpu_ring *ring; +}; +#endif + +void amdgpu_coredump(struct amdgpu_device *adev, bool vram_lost, + struct amdgpu_reset_context *reset_context); + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index eb8af02332..ee7df1d84e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -74,6 +74,7 @@ #include "amdgpu_fru_eeprom.h" #include "amdgpu_reset.h" #include "amdgpu_virt.h" +#include "amdgpu_dev_coredump.h" #include <linux/suspend.h> #include <drm/task_barrier.h> @@ -143,6 +144,8 @@ const char *amdgpu_asic_name[] = { "LAST", }; +static inline void amdgpu_device_stop_pending_resets(struct amdgpu_device *adev); + /** * DOC: pcie_replay_count * @@ -335,16 +338,93 @@ bool amdgpu_device_supports_boco(struct drm_device *dev) * * @dev: drm_device pointer * - * Returns true if the device supporte BACO, - * otherwise return false. + * Return: + * 1 if the device supporte BACO; + * 3 if the device support MACO (only works if BACO is supported) + * otherwise return 0. */ -bool amdgpu_device_supports_baco(struct drm_device *dev) +int amdgpu_device_supports_baco(struct drm_device *dev) { struct amdgpu_device *adev = drm_to_adev(dev); return amdgpu_asic_supports_baco(adev); } +void amdgpu_device_detect_runtime_pm_mode(struct amdgpu_device *adev) +{ + struct drm_device *dev; + int bamaco_support; + + dev = adev_to_drm(adev); + + adev->pm.rpm_mode = AMDGPU_RUNPM_NONE; + bamaco_support = amdgpu_device_supports_baco(dev); + + switch (amdgpu_runtime_pm) { + case 2: + if (bamaco_support & MACO_SUPPORT) { + adev->pm.rpm_mode = AMDGPU_RUNPM_BAMACO; + dev_info(adev->dev, "Forcing BAMACO for runtime pm\n"); + } else if (bamaco_support == BACO_SUPPORT) { + adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; + dev_info(adev->dev, "Requested mode BAMACO not available,fallback to use BACO\n"); + } + break; + case 1: + if (bamaco_support & BACO_SUPPORT) { + adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; + dev_info(adev->dev, "Forcing BACO for runtime pm\n"); + } + break; + case -1: + case -2: + if (amdgpu_device_supports_px(dev)) { /* enable PX as runtime mode */ + adev->pm.rpm_mode = AMDGPU_RUNPM_PX; + dev_info(adev->dev, "Using ATPX for runtime pm\n"); + } else if (amdgpu_device_supports_boco(dev)) { /* enable boco as runtime mode */ + adev->pm.rpm_mode = AMDGPU_RUNPM_BOCO; + dev_info(adev->dev, "Using BOCO for runtime pm\n"); + } else { + if (!bamaco_support) + goto no_runtime_pm; + + switch (adev->asic_type) { + case CHIP_VEGA20: + case CHIP_ARCTURUS: + /* BACO are not supported on vega20 and arctrus */ + break; + case CHIP_VEGA10: + /* enable BACO as runpm mode if noretry=0 */ + if (!adev->gmc.noretry) + adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; + break; + default: + /* enable BACO as runpm mode on CI+ */ + adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; + break; + } + + if (adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) { + if (bamaco_support & MACO_SUPPORT) { + adev->pm.rpm_mode = AMDGPU_RUNPM_BAMACO; + dev_info(adev->dev, "Using BAMACO for runtime pm\n"); + } else { + dev_info(adev->dev, "Using BACO for runtime pm\n"); + } + } + } + break; + case 0: + dev_info(adev->dev, "runtime pm is manually disabled\n"); + break; + default: + break; + } + +no_runtime_pm: + if (adev->pm.rpm_mode == AMDGPU_RUNPM_NONE) + dev_info(adev->dev, "Runtime PM not available\n"); +} /** * amdgpu_device_supports_smart_shift - Is the device dGPU with * smart shift support @@ -1402,13 +1482,17 @@ static int amdgpu_device_wb_init(struct amdgpu_device *adev) */ int amdgpu_device_wb_get(struct amdgpu_device *adev, u32 *wb) { - unsigned long offset = find_first_zero_bit(adev->wb.used, adev->wb.num_wb); + unsigned long flags, offset; + spin_lock_irqsave(&adev->wb.lock, flags); + offset = find_first_zero_bit(adev->wb.used, adev->wb.num_wb); if (offset < adev->wb.num_wb) { __set_bit(offset, adev->wb.used); + spin_unlock_irqrestore(&adev->wb.lock, flags); *wb = offset << 3; /* convert to dw offset */ return 0; } else { + spin_unlock_irqrestore(&adev->wb.lock, flags); return -EINVAL; } } @@ -1423,9 +1507,13 @@ int amdgpu_device_wb_get(struct amdgpu_device *adev, u32 *wb) */ void amdgpu_device_wb_free(struct amdgpu_device *adev, u32 wb) { + unsigned long flags; + wb >>= 3; + spin_lock_irqsave(&adev->wb.lock, flags); if (wb < adev->wb.num_wb) __clear_bit(wb, adev->wb.used); + spin_unlock_irqrestore(&adev->wb.lock, flags); } /** @@ -1455,7 +1543,7 @@ int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev) /* PCI_EXT_CAP_ID_VNDR extended capability is located at 0x100 */ if (!pci_find_ext_capability(adev->pdev, PCI_EXT_CAP_ID_VNDR)) - DRM_WARN("System can't access extended configuration space,please check!!\n"); + DRM_WARN("System can't access extended configuration space, please check!!\n"); /* skip if the bios has already enabled large BAR */ if (adev->gmc.real_vram_size && @@ -3981,6 +4069,7 @@ int amdgpu_device_init(struct amdgpu_device *adev, spin_lock_init(&adev->se_cac_idx_lock); spin_lock_init(&adev->audio_endpt_idx_lock); spin_lock_init(&adev->mm_stats.lock); + spin_lock_init(&adev->wb.lock); INIT_LIST_HEAD(&adev->shadow_list); mutex_init(&adev->shadow_list_lock); @@ -4069,6 +4158,13 @@ int amdgpu_device_init(struct amdgpu_device *adev, /* Enable TMZ based on IP_VERSION */ amdgpu_gmc_tmz_set(adev); + if (amdgpu_sriov_vf(adev) && + amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0)) + /* VF MMIO access (except mailbox range) from CPU + * will be blocked during sriov runtime + */ + adev->virt.caps |= AMDGPU_VF_MMIO_ACCESS_PROTECT; + amdgpu_gmc_noretry_set(adev); /* Need to get xgmi info early to decide the reset behavior*/ if (adev->gmc.xgmi.supported) { @@ -4974,12 +5070,15 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev, retry: amdgpu_amdkfd_pre_reset(adev); + amdgpu_device_stop_pending_resets(adev); + if (from_hypervisor) r = amdgpu_virt_request_full_gpu(adev, true); else r = amdgpu_virt_reset_gpu(adev); if (r) return r; + amdgpu_ras_set_fed(adev, false); amdgpu_irq_gpu_reset_resume_helper(adev); /* some sw clean up VF needs to do before recover */ @@ -5266,11 +5365,21 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle, struct amdgpu_device *tmp_adev = NULL; bool need_full_reset, skip_hw_reset, vram_lost = false; int r = 0; + uint32_t i; /* Try reset handler method first */ tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device, reset_list); - amdgpu_reset_reg_dumps(tmp_adev); + + if (!test_bit(AMDGPU_SKIP_COREDUMP, &reset_context->flags)) { + amdgpu_reset_reg_dumps(tmp_adev); + + /* Trigger ip dump before we reset the asic */ + for (i = 0; i < tmp_adev->num_ip_blocks; i++) + if (tmp_adev->ip_blocks[i].version->funcs->dump_ip_state) + tmp_adev->ip_blocks[i].version->funcs + ->dump_ip_state((void *)tmp_adev); + } reset_context->reset_device_list = device_list_handle; r = amdgpu_reset_perform_reset(tmp_adev, reset_context); @@ -5343,7 +5452,8 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle, vram_lost = amdgpu_device_check_vram_lost(tmp_adev); - amdgpu_coredump(tmp_adev, vram_lost, reset_context); + if (!test_bit(AMDGPU_SKIP_COREDUMP, &reset_context->flags)) + amdgpu_coredump(tmp_adev, vram_lost, reset_context); if (vram_lost) { DRM_INFO("VRAM is lost due to GPU reset!\n"); @@ -5541,6 +5651,23 @@ static inline void amdgpu_device_stop_pending_resets(struct amdgpu_device *adev) } +static int amdgpu_device_health_check(struct list_head *device_list_handle) +{ + struct amdgpu_device *tmp_adev; + int ret = 0; + u32 status; + + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { + pci_read_config_dword(tmp_adev->pdev, PCI_COMMAND, &status); + if (PCI_POSSIBLE_ERROR(status)) { + dev_err(tmp_adev->dev, "device lost from bus!"); + ret = -ENODEV; + } + } + + return ret; +} + /** * amdgpu_device_gpu_recover - reset the asic and recover scheduler * @@ -5612,6 +5739,12 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, device_list_handle = &device_list; } + if (!amdgpu_sriov_vf(adev)) { + r = amdgpu_device_health_check(device_list_handle); + if (r) + goto end_reset; + } + /* We need to lock reset domain only once both for XGMI and single device */ tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device, reset_list); @@ -5694,11 +5827,12 @@ retry: /* Rest of adevs pre asic reset from XGMI hive. */ tmp_adev->asic_reset_res = r; } - /* - * Drop all pending non scheduler resets. Scheduler resets - * were already dropped during drm_sched_stop - */ - amdgpu_device_stop_pending_resets(tmp_adev); + if (!amdgpu_sriov_vf(tmp_adev)) + /* + * Drop all pending non scheduler resets. Scheduler resets + * were already dropped during drm_sched_stop + */ + amdgpu_device_stop_pending_resets(tmp_adev); } /* Actual ASIC resets if needed.*/ @@ -5777,6 +5911,7 @@ skip_sched_resume: reset_list); amdgpu_device_unlock_reset_domain(tmp_adev->reset_domain); +end_reset: if (hive) { mutex_unlock(&hive->hive_lock); amdgpu_put_xgmi_hive(hive); @@ -6038,7 +6173,7 @@ int amdgpu_device_baco_exit(struct drm_device *dev) adev->nbio.funcs->enable_doorbell_interrupt) adev->nbio.funcs->enable_doorbell_interrupt(adev, true); - if (amdgpu_passthrough(adev) && + if (amdgpu_passthrough(adev) && adev->nbio.funcs && adev->nbio.funcs->clear_doorbell_interrupt) adev->nbio.funcs->clear_doorbell_interrupt(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index ac5bf01fe8..0e31bdb4b7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -97,6 +97,7 @@ #include "smuio_v13_0.h" #include "smuio_v13_0_3.h" #include "smuio_v13_0_6.h" +#include "smuio_v14_0_2.h" #include "vcn_v5_0_0.h" #include "jpeg_v5_0_0.h" @@ -245,6 +246,9 @@ static int amdgpu_discovery_read_binary_from_sysmem(struct amdgpu_device *adev, return -ENOENT; } +#define IP_DISCOVERY_V2 2 +#define IP_DISCOVERY_V4 4 + static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, uint8_t *binary) { @@ -259,14 +263,14 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, * wait for this to complete. Once the C2PMSG is updated, we can * continue. */ - if (dev_is_removable(&adev->pdev->dev)) { - for (i = 0; i < 1000; i++) { - msg = RREG32(mmMP0_SMN_C2PMSG_33); - if (msg & 0x80000000) - break; - msleep(1); - } + + for (i = 0; i < 1000; i++) { + msg = RREG32(mmMP0_SMN_C2PMSG_33); + if (msg & 0x80000000) + break; + usleep_range(1000, 1100); } + vram_size = (uint64_t)RREG32(mmRCC_CONFIG_MEMSIZE) << 20; if (vram_size) { @@ -1897,6 +1901,8 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev) break; case IP_VERSION(14, 0, 0): case IP_VERSION(14, 0, 1): + case IP_VERSION(14, 0, 2): + case IP_VERSION(14, 0, 3): amdgpu_device_ip_block_add(adev, &smu_v14_0_ip_block); break; default: @@ -2678,6 +2684,9 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(14, 0, 1): adev->smuio.funcs = &smuio_v13_0_6_funcs; break; + case IP_VERSION(14, 0, 2): + adev->smuio.funcs = &smuio_v14_0_2_funcs; + break; default: break; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c index 055ba2ea4c..662d0f28f3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c @@ -41,8 +41,6 @@ #include <linux/dma-buf.h> #include <linux/dma-fence-array.h> #include <linux/pci-p2pdma.h> -#include <linux/pm_runtime.h> -#include "amdgpu_trace.h" /** * amdgpu_dma_buf_attach - &dma_buf_ops.attach implementation @@ -58,42 +56,11 @@ static int amdgpu_dma_buf_attach(struct dma_buf *dmabuf, struct drm_gem_object *obj = dmabuf->priv; struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); - int r; if (pci_p2pdma_distance(adev->pdev, attach->dev, false) < 0) attach->peer2peer = false; - r = pm_runtime_get_sync(adev_to_drm(adev)->dev); - trace_amdgpu_runpm_reference_dumps(1, __func__); - if (r < 0) - goto out; - return 0; - -out: - pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); - trace_amdgpu_runpm_reference_dumps(0, __func__); - return r; -} - -/** - * amdgpu_dma_buf_detach - &dma_buf_ops.detach implementation - * - * @dmabuf: DMA-buf where we remove the attachment from - * @attach: the attachment to remove - * - * Called when an attachment is removed from the DMA-buf. - */ -static void amdgpu_dma_buf_detach(struct dma_buf *dmabuf, - struct dma_buf_attachment *attach) -{ - struct drm_gem_object *obj = dmabuf->priv; - struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); - struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); - - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); - pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); - trace_amdgpu_runpm_reference_dumps(0, __func__); } /** @@ -267,7 +234,6 @@ static int amdgpu_dma_buf_begin_cpu_access(struct dma_buf *dma_buf, const struct dma_buf_ops amdgpu_dmabuf_ops = { .attach = amdgpu_dma_buf_attach, - .detach = amdgpu_dma_buf_detach, .pin = amdgpu_dma_buf_pin, .unpin = amdgpu_dma_buf_unpin, .map_dma_buf = amdgpu_dma_buf_map, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index e4277298cf..ea14f1c8f4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -925,7 +925,7 @@ module_param_named(freesync_video, amdgpu_freesync_vid_mode, uint, 0444); * GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco) */ MODULE_PARM_DESC(reset_method, "GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco/bamaco)"); -module_param_named(reset_method, amdgpu_reset_method, int, 0444); +module_param_named(reset_method, amdgpu_reset_method, int, 0644); /** * DOC: bad_page_threshold (int) Bad page threshold is specifies the @@ -2481,6 +2481,7 @@ static void amdgpu_drv_delayed_reset_work_handler(struct work_struct *work) /* Use a common context, just need to make sure full reset is done */ set_bit(AMDGPU_SKIP_HW_RESET, &reset_context.flags); + set_bit(AMDGPU_SKIP_COREDUMP, &reset_context.flags); r = amdgpu_do_asic_reset(&device_list, &reset_context); if (r) { @@ -2744,7 +2745,8 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev) drm_dev->switch_power_state = DRM_SWITCH_POWER_DYNAMIC_OFF; } else if (adev->pm.rpm_mode == AMDGPU_RUNPM_BOCO) { /* nothing to do */ - } else if (adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) { + } else if ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) || + (adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)) { amdgpu_device_baco_enter(drm_dev); } @@ -2784,7 +2786,8 @@ static int amdgpu_pmops_runtime_resume(struct device *dev) * PCI core handles it for _PR3. */ pci_set_master(pdev); - } else if (adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) { + } else if ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) || + (adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)) { amdgpu_device_baco_exit(drm_dev); } ret = amdgpu_device_resume(drm_dev, false); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index 10832b4704..bc3ac73b6b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -181,7 +181,6 @@ int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, struct amd amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr, seq, flags | AMDGPU_FENCE_FLAG_INT); pm_runtime_get_noresume(adev_to_drm(adev)->dev); - trace_amdgpu_runpm_reference_dumps(1, __func__); ptr = &ring->fence_drv.fences[seq & ring->fence_drv.num_fences_mask]; if (unlikely(rcu_dereference_protected(*ptr, 1))) { struct dma_fence *old; @@ -309,7 +308,6 @@ bool amdgpu_fence_process(struct amdgpu_ring *ring) dma_fence_put(fence); pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); - trace_amdgpu_runpm_reference_dumps(0, __func__); } while (last_seq != seq); return true; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index 67c234bcf8..3adaa46701 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -108,6 +108,7 @@ int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size, memset(&bp, 0, sizeof(bp)); *obj = NULL; + flags |= AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE; bp.size = size; bp.byte_align = alignment; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 1d955652f3..e92bdc9a39 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -329,8 +329,9 @@ int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev, int xcc_id) ring->eop_gpu_addr = kiq->eop_gpu_addr; ring->no_scheduler = true; - snprintf(ring->name, sizeof(ring->name), "kiq_%d.%d.%d.%d", - xcc_id, ring->me, ring->pipe, ring->queue); + snprintf(ring->name, sizeof(ring->name), "kiq_%hhu.%hhu.%hhu.%hhu", + (unsigned char)xcc_id, (unsigned char)ring->me, + (unsigned char)ring->pipe, (unsigned char)ring->queue); r = amdgpu_ring_init(adev, ring, 1024, irq, AMDGPU_CP_KIQ_IRQ_DRIVER0, AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index 8fcf889ddc..64f197bbc8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h @@ -259,7 +259,6 @@ struct amdgpu_cu_info { struct amdgpu_gfx_ras { struct amdgpu_ras_block_object ras_block; void (*enable_watchdog_timer)(struct amdgpu_device *adev); - bool (*query_utcl2_poison_status)(struct amdgpu_device *adev); int (*rlc_gc_fed_irq)(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry); @@ -434,6 +433,10 @@ struct amdgpu_gfx { uint32_t num_xcc_per_xcp; struct mutex partition_mutex; bool mcbp; /* mid command buffer preemption */ + + /* IP reg dump */ + uint32_t *ip_dump; + uint32_t reg_count; }; struct amdgpu_gfx_ras_reg_entry { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h index c7b44aeb67..103a837ccc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h @@ -38,6 +38,8 @@ struct amdgpu_gfxhub_funcs { void (*mode2_save_regs)(struct amdgpu_device *adev); void (*mode2_restore_regs)(struct amdgpu_device *adev); void (*halt)(struct amdgpu_device *adev); + bool (*query_utcl2_poison_status)(struct amdgpu_device *adev, + int xcc_id); }; struct amdgpu_gfxhub { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 08b9dfb653..86b096ad03 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -878,7 +878,6 @@ void amdgpu_gmc_noretry_set(struct amdgpu_device *adev) struct amdgpu_gmc *gmc = &adev->gmc; uint32_t gc_ver = amdgpu_ip_version(adev, GC_HWIP, 0); bool noretry_default = (gc_ver == IP_VERSION(9, 0, 1) || - gc_ver == IP_VERSION(9, 3, 0) || gc_ver == IP_VERSION(9, 4, 0) || gc_ver == IP_VERSION(9, 4, 1) || gc_ver == IP_VERSION(9, 4, 2) || diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c index d79cb13e1a..00d6211e0f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c @@ -279,7 +279,7 @@ amdgpu_i2c_lookup(struct amdgpu_device *adev, return NULL; } -static void amdgpu_i2c_get_byte(struct amdgpu_i2c_chan *i2c_bus, +static int amdgpu_i2c_get_byte(struct amdgpu_i2c_chan *i2c_bus, u8 slave_addr, u8 addr, u8 *val) @@ -304,16 +304,18 @@ static void amdgpu_i2c_get_byte(struct amdgpu_i2c_chan *i2c_bus, out_buf[0] = addr; out_buf[1] = 0; - if (i2c_transfer(&i2c_bus->adapter, msgs, 2) == 2) { - *val = in_buf[0]; - DRM_DEBUG("val = 0x%02x\n", *val); - } else { - DRM_DEBUG("i2c 0x%02x 0x%02x read failed\n", - addr, *val); + if (i2c_transfer(&i2c_bus->adapter, msgs, 2) != 2) { + DRM_DEBUG("i2c 0x%02x read failed\n", addr); + return -EIO; } + + *val = in_buf[0]; + DRM_DEBUG("val = 0x%02x\n", *val); + + return 0; } -static void amdgpu_i2c_put_byte(struct amdgpu_i2c_chan *i2c_bus, +static int amdgpu_i2c_put_byte(struct amdgpu_i2c_chan *i2c_bus, u8 slave_addr, u8 addr, u8 val) @@ -329,9 +331,12 @@ static void amdgpu_i2c_put_byte(struct amdgpu_i2c_chan *i2c_bus, out_buf[0] = addr; out_buf[1] = val; - if (i2c_transfer(&i2c_bus->adapter, &msg, 1) != 1) - DRM_DEBUG("i2c 0x%02x 0x%02x write failed\n", - addr, val); + if (i2c_transfer(&i2c_bus->adapter, &msg, 1) != 1) { + DRM_DEBUG("i2c 0x%02x 0x%02x write failed\n", addr, val); + return -EIO; + } + + return 0; } /* ddc router switching */ @@ -346,16 +351,18 @@ amdgpu_i2c_router_select_ddc_port(const struct amdgpu_connector *amdgpu_connecto if (!amdgpu_connector->router_bus) return; - amdgpu_i2c_get_byte(amdgpu_connector->router_bus, + if (amdgpu_i2c_get_byte(amdgpu_connector->router_bus, amdgpu_connector->router.i2c_addr, - 0x3, &val); + 0x3, &val)) + return; val &= ~amdgpu_connector->router.ddc_mux_control_pin; amdgpu_i2c_put_byte(amdgpu_connector->router_bus, amdgpu_connector->router.i2c_addr, 0x3, val); - amdgpu_i2c_get_byte(amdgpu_connector->router_bus, + if (amdgpu_i2c_get_byte(amdgpu_connector->router_bus, amdgpu_connector->router.i2c_addr, - 0x1, &val); + 0x1, &val)) + return; val &= ~amdgpu_connector->router.ddc_mux_control_pin; val |= amdgpu_connector->router.ddc_mux_state; amdgpu_i2c_put_byte(amdgpu_connector->router_bus, @@ -375,16 +382,18 @@ amdgpu_i2c_router_select_cd_port(const struct amdgpu_connector *amdgpu_connector if (!amdgpu_connector->router_bus) return; - amdgpu_i2c_get_byte(amdgpu_connector->router_bus, + if (amdgpu_i2c_get_byte(amdgpu_connector->router_bus, amdgpu_connector->router.i2c_addr, - 0x3, &val); + 0x3, &val)) + return; val &= ~amdgpu_connector->router.cd_mux_control_pin; amdgpu_i2c_put_byte(amdgpu_connector->router_bus, amdgpu_connector->router.i2c_addr, 0x3, val); - amdgpu_i2c_get_byte(amdgpu_connector->router_bus, + if (amdgpu_i2c_get_byte(amdgpu_connector->router_bus, amdgpu_connector->router.i2c_addr, - 0x1, &val); + 0x1, &val)) + return; val &= ~amdgpu_connector->router.cd_mux_control_pin; val |= amdgpu_connector->router.cd_mux_state; amdgpu_i2c_put_byte(amdgpu_connector->router_bus, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index 665c63f552..013ff373e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -279,7 +279,7 @@ int amdgpu_irq_init(struct amdgpu_device *adev) adev->irq.msi_enabled = false; if (!amdgpu_msi_ok(adev)) - flags = PCI_IRQ_LEGACY; + flags = PCI_IRQ_INTX; else flags = PCI_IRQ_ALL_TYPES; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index a2df3025a7..a0ea6fe8d0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -149,38 +149,7 @@ int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags) goto out; } - adev->pm.rpm_mode = AMDGPU_RUNPM_NONE; - if (amdgpu_device_supports_px(dev) && - (amdgpu_runtime_pm != 0)) { /* enable PX as runtime mode */ - adev->pm.rpm_mode = AMDGPU_RUNPM_PX; - dev_info(adev->dev, "Using ATPX for runtime pm\n"); - } else if (amdgpu_device_supports_boco(dev) && - (amdgpu_runtime_pm != 0)) { /* enable boco as runtime mode */ - adev->pm.rpm_mode = AMDGPU_RUNPM_BOCO; - dev_info(adev->dev, "Using BOCO for runtime pm\n"); - } else if (amdgpu_device_supports_baco(dev) && - (amdgpu_runtime_pm != 0)) { - switch (adev->asic_type) { - case CHIP_VEGA20: - case CHIP_ARCTURUS: - /* enable BACO as runpm mode if runpm=1 */ - if (amdgpu_runtime_pm > 0) - adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; - break; - case CHIP_VEGA10: - /* enable BACO as runpm mode if noretry=0 */ - if (!adev->gmc.noretry) - adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; - break; - default: - /* enable BACO as runpm mode on CI+ */ - adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; - break; - } - - if (adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) - dev_info(adev->dev, "Using BACO for runtime pm\n"); - } + amdgpu_device_detect_runtime_pm_mode(adev); /* Call ACPI methods: require modeset init * but failure is not fatal diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c index 24ad4b9717..0734490347 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c @@ -210,22 +210,26 @@ int amdgpu_mca_smu_set_debug_mode(struct amdgpu_device *adev, bool enable) return -EOPNOTSUPP; } -static void amdgpu_mca_smu_mca_bank_dump(struct amdgpu_device *adev, int idx, struct mca_bank_entry *entry) +static void amdgpu_mca_smu_mca_bank_dump(struct amdgpu_device *adev, int idx, struct mca_bank_entry *entry, + struct ras_query_context *qctx) { - dev_info(adev->dev, HW_ERR "Accelerator Check Architecture events logged\n"); - dev_info(adev->dev, HW_ERR "aca entry[%02d].STATUS=0x%016llx\n", - idx, entry->regs[MCA_REG_IDX_STATUS]); - dev_info(adev->dev, HW_ERR "aca entry[%02d].ADDR=0x%016llx\n", - idx, entry->regs[MCA_REG_IDX_ADDR]); - dev_info(adev->dev, HW_ERR "aca entry[%02d].MISC0=0x%016llx\n", - idx, entry->regs[MCA_REG_IDX_MISC0]); - dev_info(adev->dev, HW_ERR "aca entry[%02d].IPID=0x%016llx\n", - idx, entry->regs[MCA_REG_IDX_IPID]); - dev_info(adev->dev, HW_ERR "aca entry[%02d].SYND=0x%016llx\n", - idx, entry->regs[MCA_REG_IDX_SYND]); + u64 event_id = qctx->event_id; + + RAS_EVENT_LOG(adev, event_id, HW_ERR "Accelerator Check Architecture events logged\n"); + RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].STATUS=0x%016llx\n", + idx, entry->regs[MCA_REG_IDX_STATUS]); + RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].ADDR=0x%016llx\n", + idx, entry->regs[MCA_REG_IDX_ADDR]); + RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].MISC0=0x%016llx\n", + idx, entry->regs[MCA_REG_IDX_MISC0]); + RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].IPID=0x%016llx\n", + idx, entry->regs[MCA_REG_IDX_IPID]); + RAS_EVENT_LOG(adev, event_id, HW_ERR "aca entry[%02d].SYND=0x%016llx\n", + idx, entry->regs[MCA_REG_IDX_SYND]); } -int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type, struct ras_err_data *err_data) +int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type, + struct ras_err_data *err_data, struct ras_query_context *qctx) { struct amdgpu_smuio_mcm_config_info mcm_info; struct ras_err_addr err_addr = {0}; @@ -244,7 +248,7 @@ int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_blo list_for_each_entry(node, &mca_set.list, node) { entry = &node->entry; - amdgpu_mca_smu_mca_bank_dump(adev, i++, entry); + amdgpu_mca_smu_mca_bank_dump(adev, i++, entry, qctx); count = 0; ret = amdgpu_mca_smu_parse_mca_error_count(adev, blk, type, entry, &count); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h index b964110ed1..e5bf07ce34 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h @@ -169,6 +169,7 @@ void amdgpu_mca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root void amdgpu_mca_bank_set_init(struct mca_bank_set *mca_set); int amdgpu_mca_bank_set_add_entry(struct mca_bank_set *mca_set, struct mca_bank_entry *entry); void amdgpu_mca_bank_set_release(struct mca_bank_set *mca_set); -int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type, struct ras_err_data *err_data); +int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type, + struct ras_err_data *err_data, struct ras_query_context *qctx); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c index 1569bef030..5ca5c47ab5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c @@ -32,6 +32,18 @@ #define AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS 1024 #define AMDGPU_ONE_DOORBELL_SIZE 8 +signed long amdgpu_mes_fence_wait_polling(u64 *fence, + u64 wait_seq, + signed long timeout) +{ + + while ((s64)(wait_seq - *fence) > 0 && timeout > 0) { + udelay(2); + timeout -= 2; + } + return timeout > 0 ? timeout : 0; +} + int amdgpu_mes_doorbell_process_slice(struct amdgpu_device *adev) { return roundup(AMDGPU_ONE_DOORBELL_SIZE * @@ -40,7 +52,6 @@ int amdgpu_mes_doorbell_process_slice(struct amdgpu_device *adev) } static int amdgpu_mes_kernel_doorbell_get(struct amdgpu_device *adev, - struct amdgpu_mes_process *process, int ip_type, uint64_t *doorbell_index) { unsigned int offset, found; @@ -65,7 +76,6 @@ static int amdgpu_mes_kernel_doorbell_get(struct amdgpu_device *adev, } static void amdgpu_mes_kernel_doorbell_free(struct amdgpu_device *adev, - struct amdgpu_mes_process *process, uint32_t doorbell_index) { unsigned int old, rel_index; @@ -656,7 +666,7 @@ int amdgpu_mes_add_hw_queue(struct amdgpu_device *adev, int gang_id, *queue_id = queue->queue_id = r; /* allocate a doorbell index for the queue */ - r = amdgpu_mes_kernel_doorbell_get(adev, gang->process, + r = amdgpu_mes_kernel_doorbell_get(adev, qprops->queue_type, &qprops->doorbell_off); if (r) @@ -714,8 +724,7 @@ int amdgpu_mes_add_hw_queue(struct amdgpu_device *adev, int gang_id, return 0; clean_up_doorbell: - amdgpu_mes_kernel_doorbell_free(adev, gang->process, - qprops->doorbell_off); + amdgpu_mes_kernel_doorbell_free(adev, qprops->doorbell_off); clean_up_queue_id: spin_lock_irqsave(&adev->mes.queue_id_lock, flags); idr_remove(&adev->mes.queue_id_idr, queue->queue_id); @@ -769,8 +778,7 @@ int amdgpu_mes_remove_hw_queue(struct amdgpu_device *adev, int queue_id) queue_id); list_del(&queue->list); - amdgpu_mes_kernel_doorbell_free(adev, gang->process, - queue->doorbell_off); + amdgpu_mes_kernel_doorbell_free(adev, queue->doorbell_off); amdgpu_mes_unlock(&adev->mes); amdgpu_mes_queue_free_mqd(queue); @@ -778,6 +786,28 @@ int amdgpu_mes_remove_hw_queue(struct amdgpu_device *adev, int queue_id) return 0; } +int amdgpu_mes_map_legacy_queue(struct amdgpu_device *adev, + struct amdgpu_ring *ring) +{ + struct mes_map_legacy_queue_input queue_input; + int r; + + memset(&queue_input, 0, sizeof(queue_input)); + + queue_input.queue_type = ring->funcs->type; + queue_input.doorbell_offset = ring->doorbell_index; + queue_input.pipe_id = ring->pipe; + queue_input.queue_id = ring->queue; + queue_input.mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj); + queue_input.wptr_addr = ring->wptr_gpu_addr; + + r = adev->mes.funcs->map_legacy_queue(&adev->mes, &queue_input); + if (r) + DRM_ERROR("failed to map legacy queue\n"); + + return r; +} + int amdgpu_mes_unmap_legacy_queue(struct amdgpu_device *adev, struct amdgpu_ring *ring, enum amdgpu_unmap_queues_action action, @@ -1475,7 +1505,7 @@ int amdgpu_mes_init_microcode(struct amdgpu_device *adev, int pipe) const struct mes_firmware_header_v1_0 *mes_hdr; struct amdgpu_firmware_info *info; char ucode_prefix[30]; - char fw_name[40]; + char fw_name[50]; bool need_retry = false; int r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h index 4c8fc3117e..df9f0404d8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h @@ -141,6 +141,12 @@ struct amdgpu_mes { /* ip specific functions */ const struct amdgpu_mes_funcs *funcs; + + /* mes resource_1 bo*/ + struct amdgpu_bo *resource_1; + uint64_t resource_1_gpu_addr; + void *resource_1_addr; + }; struct amdgpu_mes_process { @@ -242,6 +248,15 @@ struct mes_remove_queue_input { uint64_t gang_context_addr; }; +struct mes_map_legacy_queue_input { + uint32_t queue_type; + uint32_t doorbell_offset; + uint32_t pipe_id; + uint32_t queue_id; + uint64_t mqd_addr; + uint64_t wptr_addr; +}; + struct mes_unmap_legacy_queue_input { enum amdgpu_unmap_queues_action action; uint32_t queue_type; @@ -318,6 +333,9 @@ struct amdgpu_mes_funcs { int (*remove_hw_queue)(struct amdgpu_mes *mes, struct mes_remove_queue_input *input); + int (*map_legacy_queue)(struct amdgpu_mes *mes, + struct mes_map_legacy_queue_input *input); + int (*unmap_legacy_queue)(struct amdgpu_mes *mes, struct mes_unmap_legacy_queue_input *input); @@ -334,6 +352,10 @@ struct amdgpu_mes_funcs { #define amdgpu_mes_kiq_hw_init(adev) (adev)->mes.kiq_hw_init((adev)) #define amdgpu_mes_kiq_hw_fini(adev) (adev)->mes.kiq_hw_fini((adev)) +signed long amdgpu_mes_fence_wait_polling(u64 *fence, + u64 wait_seq, + signed long timeout); + int amdgpu_mes_ctx_get_offs(struct amdgpu_ring *ring, unsigned int id_offs); int amdgpu_mes_init_microcode(struct amdgpu_device *adev, int pipe); @@ -357,6 +379,8 @@ int amdgpu_mes_add_hw_queue(struct amdgpu_device *adev, int gang_id, int *queue_id); int amdgpu_mes_remove_hw_queue(struct amdgpu_device *adev, int queue_id); +int amdgpu_mes_map_legacy_queue(struct amdgpu_device *adev, + struct amdgpu_ring *ring); int amdgpu_mes_unmap_legacy_queue(struct amdgpu_device *adev, struct amdgpu_ring *ring, enum amdgpu_unmap_queues_action action, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h index 1ca9d4ed80..95d676ee20 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h @@ -63,6 +63,8 @@ struct amdgpu_mmhub_funcs { uint64_t page_table_base); void (*update_power_gating)(struct amdgpu_device *adev, bool enable); + bool (*query_utcl2_poison_status)(struct amdgpu_device *adev, + int hub_inst); }; struct amdgpu_mmhub { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c index f6d503432a..c556c8b653 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c @@ -39,6 +39,7 @@ #include "amdgpu.h" #include "amdgpu_trace.h" #include "amdgpu_amdkfd.h" +#include "amdgpu_vram_mgr.h" /** * DOC: amdgpu_object @@ -153,8 +154,10 @@ void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain) else places[c].flags |= TTM_PL_FLAG_TOPDOWN; - if (flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS) + if (abo->tbo.type == ttm_bo_type_kernel && + flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS) places[c].flags |= TTM_PL_FLAG_CONTIGUOUS; + c++; } @@ -173,6 +176,12 @@ void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain) abo->flags & AMDGPU_GEM_CREATE_PREEMPTIBLE ? AMDGPU_PL_PREEMPT : TTM_PL_TT; places[c].flags = 0; + /* + * When GTT is just an alternative to VRAM make sure that we + * only use it as fallback and still try to fill up VRAM first. + */ + if (domain & abo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) + places[c].flags |= TTM_PL_FLAG_FALLBACK; c++; } @@ -595,9 +604,6 @@ int amdgpu_bo_create(struct amdgpu_device *adev, if (!amdgpu_bo_support_uswc(bo->flags)) bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC; - if (adev->ras_enabled) - bo->flags |= AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE; - bo->tbo.bdev = &adev->mman.bdev; if (bp->domain & (AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA | AMDGPU_GEM_DOMAIN_GDS)) @@ -629,7 +635,7 @@ int amdgpu_bo_create(struct amdgpu_device *adev, bo->tbo.resource->mem_type == TTM_PL_VRAM) { struct dma_fence *fence; - r = amdgpu_fill_buffer(bo, 0, bo->tbo.base.resv, &fence, true); + r = amdgpu_ttm_clear_buffer(bo, bo->tbo.base.resv, &fence); if (unlikely(r)) goto fail_unreserve; @@ -759,7 +765,7 @@ int amdgpu_bo_restore_shadow(struct amdgpu_bo *shadow, struct dma_fence **fence) return amdgpu_copy_buffer(ring, shadow_addr, parent_addr, amdgpu_bo_size(shadow), NULL, fence, - true, false, false); + true, false, 0); } /** @@ -961,6 +967,10 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain, if (!bo->placements[i].lpfn || (lpfn && lpfn < bo->placements[i].lpfn)) bo->placements[i].lpfn = lpfn; + + if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS && + bo->placements[i].mem_type == TTM_PL_VRAM) + bo->placements[i].flags |= TTM_PL_FLAG_CONTIGUOUS; } r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx); @@ -1366,8 +1376,9 @@ void amdgpu_bo_release_notify(struct ttm_buffer_object *bo) if (WARN_ON_ONCE(!dma_resv_trylock(bo->base.resv))) return; - r = amdgpu_fill_buffer(abo, AMDGPU_POISON, bo->base.resv, &fence, true); + r = amdgpu_fill_buffer(abo, 0, bo->base.resv, &fence, true); if (!WARN_ON(r)) { + amdgpu_vram_mgr_set_cleared(bo->resource); amdgpu_bo_fence(abo, fence, false); dma_fence_put(fence); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index 94b310fdb7..cef9dd0a01 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -640,6 +640,20 @@ static const char *psp_gfx_cmd_name(enum psp_gfx_cmd_id cmd_id) } } +static bool psp_err_warn(struct psp_context *psp) +{ + struct psp_gfx_cmd_resp *cmd = psp->cmd_buf_mem; + + /* This response indicates reg list is already loaded */ + if (amdgpu_ip_version(psp->adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 2) && + cmd->cmd_id == GFX_CMD_ID_LOAD_IP_FW && + cmd->cmd.cmd_load_ip_fw.fw_type == GFX_FW_TYPE_REG_LIST && + cmd->resp.status == TEE_ERROR_CANCEL) + return false; + + return true; +} + static int psp_cmd_submit_buf(struct psp_context *psp, struct amdgpu_firmware_info *ucode, @@ -699,10 +713,13 @@ psp_cmd_submit_buf(struct psp_context *psp, dev_warn(psp->adev->dev, "failed to load ucode %s(0x%X) ", amdgpu_ucode_name(ucode->ucode_id), ucode->ucode_id); - dev_warn(psp->adev->dev, - "psp gfx command %s(0x%X) failed and response status is (0x%X)\n", - psp_gfx_cmd_name(psp->cmd_buf_mem->cmd_id), psp->cmd_buf_mem->cmd_id, - psp->cmd_buf_mem->resp.status); + if (psp_err_warn(psp)) + dev_warn( + psp->adev->dev, + "psp gfx command %s(0x%X) failed and response status is (0x%X)\n", + psp_gfx_cmd_name(psp->cmd_buf_mem->cmd_id), + psp->cmd_buf_mem->cmd_id, + psp->cmd_buf_mem->resp.status); /* If any firmware (including CAP) load fails under SRIOV, it should * return failure to stop the VF from initializing. * Also return failure in case of timeout @@ -1053,6 +1070,11 @@ static int psp_asd_initialize(struct psp_context *psp) if (amdgpu_sriov_vf(psp->adev) || !psp->asd_context.bin_desc.size_bytes) return 0; + /* bypass asd if display hardware is not available */ + if (!amdgpu_device_has_display_hardware(psp->adev) && + amdgpu_ip_version(psp->adev, MP0_HWIP, 0) >= IP_VERSION(13, 0, 10)) + return 0; + psp->asd_context.mem_context.shared_mc_addr = 0; psp->asd_context.mem_context.shared_mem_size = PSP_ASD_SHARED_MEM_SIZE; psp->asd_context.ta_load_type = GFX_CMD_ID_LOAD_ASD; @@ -2260,6 +2282,15 @@ static int psp_hw_start(struct psp_context *psp) } } + if ((is_psp_fw_valid(psp->ipkeymgr_drv)) && + (psp->funcs->bootloader_load_ipkeymgr_drv != NULL)) { + ret = psp_bootloader_load_ipkeymgr_drv(psp); + if (ret) { + dev_err(adev->dev, "PSP load ipkeymgr_drv failed!\n"); + return ret; + } + } + if ((is_psp_fw_valid(psp->sos)) && (psp->funcs->bootloader_load_sos != NULL)) { ret = psp_bootloader_load_sos(psp); @@ -2617,7 +2648,8 @@ static int psp_load_p2s_table(struct psp_context *psp) struct amdgpu_firmware_info *ucode = &adev->firmware.ucode[AMDGPU_UCODE_ID_P2S_TABLE]; - if (adev->in_runpm && (adev->pm.rpm_mode == AMDGPU_RUNPM_BACO)) + if (adev->in_runpm && ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) || + (adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO))) return 0; if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6)) { @@ -2647,7 +2679,8 @@ static int psp_load_smu_fw(struct psp_context *psp) * Skip SMU FW reloading in case of using BACO for runpm only, * as SMU is always alive. */ - if (adev->in_runpm && (adev->pm.rpm_mode == AMDGPU_RUNPM_BACO)) + if (adev->in_runpm && ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) || + (adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO))) return 0; if (!ucode->fw || amdgpu_sriov_vf(psp->adev)) @@ -3273,6 +3306,12 @@ static int parse_sos_bin_descriptor(struct psp_context *psp, psp->ras_drv.size_bytes = le32_to_cpu(desc->size_bytes); psp->ras_drv.start_addr = ucode_start_addr; break; + case PSP_FW_TYPE_PSP_IPKEYMGR_DRV: + psp->ipkeymgr_drv.fw_version = le32_to_cpu(desc->fw_version); + psp->ipkeymgr_drv.feature_version = le32_to_cpu(desc->fw_version); + psp->ipkeymgr_drv.size_bytes = le32_to_cpu(desc->size_bytes); + psp->ipkeymgr_drv.start_addr = ucode_start_addr; + break; default: dev_warn(psp->adev->dev, "Unsupported PSP FW type: %d\n", desc->fw_type); break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index ee16f134ae..3635303e65 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -73,8 +73,10 @@ enum psp_bootloader_cmd { PSP_BL__LOAD_KEY_DATABASE = 0x80000, PSP_BL__LOAD_SOCDRV = 0xB0000, PSP_BL__LOAD_DBGDRV = 0xC0000, + PSP_BL__LOAD_HADDRV = PSP_BL__LOAD_DBGDRV, PSP_BL__LOAD_INTFDRV = 0xD0000, - PSP_BL__LOAD_RASDRV = 0xE0000, + PSP_BL__LOAD_RASDRV = 0xE0000, + PSP_BL__LOAD_IPKEYMGRDRV = 0xF0000, PSP_BL__DRAM_LONG_TRAIN = 0x100000, PSP_BL__DRAM_SHORT_TRAIN = 0x200000, PSP_BL__LOAD_TOS_SPL_TABLE = 0x10000000, @@ -117,6 +119,7 @@ struct psp_funcs { int (*bootloader_load_intf_drv)(struct psp_context *psp); int (*bootloader_load_dbg_drv)(struct psp_context *psp); int (*bootloader_load_ras_drv)(struct psp_context *psp); + int (*bootloader_load_ipkeymgr_drv)(struct psp_context *psp); int (*bootloader_load_sos)(struct psp_context *psp); int (*ring_create)(struct psp_context *psp, enum psp_ring_type ring_type); @@ -336,6 +339,7 @@ struct psp_context { struct psp_bin_desc intf_drv; struct psp_bin_desc dbg_drv; struct psp_bin_desc ras_drv; + struct psp_bin_desc ipkeymgr_drv; /* tmr buffer */ struct amdgpu_bo *tmr_bo; @@ -424,6 +428,9 @@ struct amdgpu_psp_funcs { #define psp_bootloader_load_ras_drv(psp) \ ((psp)->funcs->bootloader_load_ras_drv ? \ (psp)->funcs->bootloader_load_ras_drv((psp)) : 0) +#define psp_bootloader_load_ipkeymgr_drv(psp) \ + ((psp)->funcs->bootloader_load_ipkeymgr_drv ? \ + (psp)->funcs->bootloader_load_ipkeymgr_drv((psp)) : 0) #define psp_bootloader_load_sos(psp) \ ((psp)->funcs->bootloader_load_sos ? (psp)->funcs->bootloader_load_sos((psp)) : 0) #define psp_smu_reload_quirk(psp) \ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 8ebab6f22e..1adc81a557 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -122,6 +122,8 @@ const char *get_ras_block_str(struct ras_common_if *ras_block) #define MAX_UMC_POISON_POLLING_TIME_ASYNC 100 //ms +#define AMDGPU_RAS_RETIRE_PAGE_INTERVAL 100 //ms + enum amdgpu_ras_retire_page_reservation { AMDGPU_RAS_RETIRE_PAGE_RESERVED, AMDGPU_RAS_RETIRE_PAGE_PENDING, @@ -1045,6 +1047,7 @@ static void amdgpu_ras_get_ecc_info(struct amdgpu_device *adev, struct ras_err_d static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, struct ras_manager *ras_mgr, struct ras_err_data *err_data, + struct ras_query_context *qctx, const char *blk_name, bool is_ue, bool is_de) @@ -1052,27 +1055,28 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, struct amdgpu_smuio_mcm_config_info *mcm_info; struct ras_err_node *err_node; struct ras_err_info *err_info; + u64 event_id = qctx->event_id; if (is_ue) { for_each_ras_error(err_node, err_data) { err_info = &err_node->err_info; mcm_info = &err_info->mcm_info; if (err_info->ue_count) { - dev_info(adev->dev, "socket: %d, die: %d, " - "%lld new uncorrectable hardware errors detected in %s block\n", - mcm_info->socket_id, - mcm_info->die_id, - err_info->ue_count, - blk_name); + RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, " + "%lld new uncorrectable hardware errors detected in %s block\n", + mcm_info->socket_id, + mcm_info->die_id, + err_info->ue_count, + blk_name); } } for_each_ras_error(err_node, &ras_mgr->err_data) { err_info = &err_node->err_info; mcm_info = &err_info->mcm_info; - dev_info(adev->dev, "socket: %d, die: %d, " - "%lld uncorrectable hardware errors detected in total in %s block\n", - mcm_info->socket_id, mcm_info->die_id, err_info->ue_count, blk_name); + RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, " + "%lld uncorrectable hardware errors detected in total in %s block\n", + mcm_info->socket_id, mcm_info->die_id, err_info->ue_count, blk_name); } } else { @@ -1081,44 +1085,44 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, err_info = &err_node->err_info; mcm_info = &err_info->mcm_info; if (err_info->de_count) { - dev_info(adev->dev, "socket: %d, die: %d, " - "%lld new deferred hardware errors detected in %s block\n", - mcm_info->socket_id, - mcm_info->die_id, - err_info->de_count, - blk_name); + RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, " + "%lld new deferred hardware errors detected in %s block\n", + mcm_info->socket_id, + mcm_info->die_id, + err_info->de_count, + blk_name); } } for_each_ras_error(err_node, &ras_mgr->err_data) { err_info = &err_node->err_info; mcm_info = &err_info->mcm_info; - dev_info(adev->dev, "socket: %d, die: %d, " - "%lld deferred hardware errors detected in total in %s block\n", - mcm_info->socket_id, mcm_info->die_id, - err_info->de_count, blk_name); + RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, " + "%lld deferred hardware errors detected in total in %s block\n", + mcm_info->socket_id, mcm_info->die_id, + err_info->de_count, blk_name); } } else { for_each_ras_error(err_node, err_data) { err_info = &err_node->err_info; mcm_info = &err_info->mcm_info; if (err_info->ce_count) { - dev_info(adev->dev, "socket: %d, die: %d, " - "%lld new correctable hardware errors detected in %s block\n", - mcm_info->socket_id, - mcm_info->die_id, - err_info->ce_count, - blk_name); + RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, " + "%lld new correctable hardware errors detected in %s block\n", + mcm_info->socket_id, + mcm_info->die_id, + err_info->ce_count, + blk_name); } } for_each_ras_error(err_node, &ras_mgr->err_data) { err_info = &err_node->err_info; mcm_info = &err_info->mcm_info; - dev_info(adev->dev, "socket: %d, die: %d, " - "%lld correctable hardware errors detected in total in %s block\n", - mcm_info->socket_id, mcm_info->die_id, - err_info->ce_count, blk_name); + RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d, " + "%lld correctable hardware errors detected in total in %s block\n", + mcm_info->socket_id, mcm_info->die_id, + err_info->ce_count, blk_name); } } } @@ -1131,77 +1135,79 @@ static inline bool err_data_has_source_info(struct ras_err_data *data) static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, struct ras_query_if *query_if, - struct ras_err_data *err_data) + struct ras_err_data *err_data, + struct ras_query_context *qctx) { struct ras_manager *ras_mgr = amdgpu_ras_find_obj(adev, &query_if->head); const char *blk_name = get_ras_block_str(&query_if->head); + u64 event_id = qctx->event_id; if (err_data->ce_count) { if (err_data_has_source_info(err_data)) { - amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, + amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, qctx, blk_name, false, false); } else if (!adev->aid_mask && adev->smuio.funcs && adev->smuio.funcs->get_socket_id && adev->smuio.funcs->get_die_id) { - dev_info(adev->dev, "socket: %d, die: %d " - "%ld correctable hardware errors " - "detected in %s block\n", - adev->smuio.funcs->get_socket_id(adev), - adev->smuio.funcs->get_die_id(adev), - ras_mgr->err_data.ce_count, - blk_name); + RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d " + "%ld correctable hardware errors " + "detected in %s block\n", + adev->smuio.funcs->get_socket_id(adev), + adev->smuio.funcs->get_die_id(adev), + ras_mgr->err_data.ce_count, + blk_name); } else { - dev_info(adev->dev, "%ld correctable hardware errors " - "detected in %s block\n", - ras_mgr->err_data.ce_count, - blk_name); + RAS_EVENT_LOG(adev, event_id, "%ld correctable hardware errors " + "detected in %s block\n", + ras_mgr->err_data.ce_count, + blk_name); } } if (err_data->ue_count) { if (err_data_has_source_info(err_data)) { - amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, + amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, qctx, blk_name, true, false); } else if (!adev->aid_mask && adev->smuio.funcs && adev->smuio.funcs->get_socket_id && adev->smuio.funcs->get_die_id) { - dev_info(adev->dev, "socket: %d, die: %d " - "%ld uncorrectable hardware errors " - "detected in %s block\n", - adev->smuio.funcs->get_socket_id(adev), - adev->smuio.funcs->get_die_id(adev), - ras_mgr->err_data.ue_count, - blk_name); + RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d " + "%ld uncorrectable hardware errors " + "detected in %s block\n", + adev->smuio.funcs->get_socket_id(adev), + adev->smuio.funcs->get_die_id(adev), + ras_mgr->err_data.ue_count, + blk_name); } else { - dev_info(adev->dev, "%ld uncorrectable hardware errors " - "detected in %s block\n", - ras_mgr->err_data.ue_count, - blk_name); + RAS_EVENT_LOG(adev, event_id, "%ld uncorrectable hardware errors " + "detected in %s block\n", + ras_mgr->err_data.ue_count, + blk_name); } } if (err_data->de_count) { if (err_data_has_source_info(err_data)) { - amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, + amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, qctx, blk_name, false, true); } else if (!adev->aid_mask && adev->smuio.funcs && adev->smuio.funcs->get_socket_id && adev->smuio.funcs->get_die_id) { - dev_info(adev->dev, "socket: %d, die: %d " - "%ld deferred hardware errors " - "detected in %s block\n", - adev->smuio.funcs->get_socket_id(adev), - adev->smuio.funcs->get_die_id(adev), - ras_mgr->err_data.de_count, - blk_name); + RAS_EVENT_LOG(adev, event_id, "socket: %d, die: %d " + "%ld deferred hardware errors " + "detected in %s block\n", + adev->smuio.funcs->get_socket_id(adev), + adev->smuio.funcs->get_die_id(adev), + ras_mgr->err_data.de_count, + blk_name); } else { - dev_info(adev->dev, "%ld deferred hardware errors " - "detected in %s block\n", - ras_mgr->err_data.de_count, - blk_name); + RAS_EVENT_LOG(adev, event_id, "%ld deferred hardware errors " + "detected in %s block\n", + ras_mgr->err_data.de_count, + blk_name); } } } @@ -1244,6 +1250,10 @@ int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk, { struct ras_manager *obj; + /* in resume phase, no need to create aca fs node */ + if (adev->in_suspend || amdgpu_in_reset(adev)) + return 0; + obj = get_ras_manager(adev, blk); if (!obj) return -EINVAL; @@ -1265,7 +1275,8 @@ int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk) } static int amdgpu_aca_log_ras_error_data(struct amdgpu_device *adev, enum amdgpu_ras_block blk, - enum aca_error_type type, struct ras_err_data *err_data) + enum aca_error_type type, struct ras_err_data *err_data, + struct ras_query_context *qctx) { struct ras_manager *obj; @@ -1273,7 +1284,7 @@ static int amdgpu_aca_log_ras_error_data(struct amdgpu_device *adev, enum amdgpu if (!obj) return -EINVAL; - return amdgpu_aca_get_error_data(adev, &obj->aca_handle, type, err_data); + return amdgpu_aca_get_error_data(adev, &obj->aca_handle, type, err_data, qctx); } ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr, @@ -1287,13 +1298,14 @@ ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *a if (amdgpu_ras_query_error_status(obj->adev, &info)) return -EINVAL; - return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count, - "ce", info.ce_count); + return sysfs_emit(buf, "%s: %lu\n%s: %lu\n%s: %lu\n", "ue", info.ue_count, + "ce", info.ce_count, "de", info.ue_count); } static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev, struct ras_query_if *info, struct ras_err_data *err_data, + struct ras_query_context *qctx, unsigned int error_query_mode) { enum amdgpu_ras_block blk = info ? info->head.block : AMDGPU_RAS_BLOCK_COUNT; @@ -1329,17 +1341,21 @@ static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev, } } else { if (amdgpu_aca_is_enabled(adev)) { - ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_UE, err_data); + ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_UE, err_data, qctx); + if (ret) + return ret; + + ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_CE, err_data, qctx); if (ret) return ret; - ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_CE, err_data); + ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_DEFERRED, err_data, qctx); if (ret) return ret; } else { /* FIXME: add code to check return value later */ - amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_UE, err_data); - amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_CE, err_data); + amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_UE, err_data, qctx); + amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_CE, err_data, qctx); } } @@ -1351,6 +1367,7 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_i { struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head); struct ras_err_data err_data; + struct ras_query_context qctx; unsigned int error_query_mode; int ret; @@ -1364,8 +1381,12 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_i if (!amdgpu_ras_get_error_query_mode(adev, &error_query_mode)) return -EINVAL; + memset(&qctx, 0, sizeof(qctx)); + qctx.event_id = amdgpu_ras_acquire_event_id(adev, amdgpu_ras_intr_triggered() ? + RAS_EVENT_TYPE_ISR : RAS_EVENT_TYPE_INVALID); ret = amdgpu_ras_query_error_status_helper(adev, info, &err_data, + &qctx, error_query_mode); if (ret) goto out_fini_err_data; @@ -1376,7 +1397,7 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_i info->ce_count = obj->err_data.ce_count; info->de_count = obj->err_data.de_count; - amdgpu_ras_error_generate_report(adev, info, &err_data); + amdgpu_ras_error_generate_report(adev, info, &err_data, &qctx); out_fini_err_data: amdgpu_ras_error_data_fini(&err_data); @@ -2041,7 +2062,7 @@ static void amdgpu_ras_interrupt_poison_consumption_handler(struct ras_manager * } } - amdgpu_umc_poison_handler(adev, obj->head.block, false); + amdgpu_umc_poison_handler(adev, obj->head.block, 0); if (block_obj->hw_ops && block_obj->hw_ops->handle_poison_consumption) poison_stat = block_obj->hw_ops->handle_poison_consumption(adev); @@ -2061,6 +2082,17 @@ static void amdgpu_ras_interrupt_poison_creation_handler(struct ras_manager *obj { dev_info(obj->adev->dev, "Poison is created\n"); + + if (amdgpu_ip_version(obj->adev, UMC_HWIP, 0) >= IP_VERSION(12, 0, 0)) { + struct amdgpu_ras *con = amdgpu_ras_get_context(obj->adev); + + amdgpu_ras_put_poison_req(obj->adev, + AMDGPU_RAS_BLOCK__UMC, 0, NULL, NULL, false); + + atomic_inc(&con->page_retirement_req_cnt); + + wake_up(&con->page_retirement_wq); + } } static void amdgpu_ras_interrupt_umc_handler(struct ras_manager *obj, @@ -2371,7 +2403,7 @@ static int amdgpu_ras_badpages_read(struct amdgpu_device *adev, .flags = AMDGPU_RAS_RETIRE_PAGE_RESERVED, }; status = amdgpu_vram_mgr_query_page_status(&adev->mman.vram_mgr, - data->bps[i].retired_page); + data->bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT); if (status == -EBUSY) (*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_PENDING; else if (status == -ENOENT) @@ -2384,6 +2416,19 @@ out: return ret; } +static void amdgpu_ras_set_fed_all(struct amdgpu_device *adev, + struct amdgpu_hive_info *hive, bool status) +{ + struct amdgpu_device *tmp_adev; + + if (hive) { + list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) + amdgpu_ras_set_fed(tmp_adev, status); + } else { + amdgpu_ras_set_fed(adev, status); + } +} + static void amdgpu_ras_do_recovery(struct work_struct *work) { struct amdgpu_ras *ras = @@ -2393,8 +2438,21 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) struct list_head device_list, *device_list_handle = NULL; struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev); - if (hive) + if (hive) { atomic_set(&hive->ras_recovery, 1); + + /* If any device which is part of the hive received RAS fatal + * error interrupt, set fatal error status on all. This + * condition will need a recovery, and flag will be cleared + * as part of recovery. + */ + list_for_each_entry(remote_adev, &hive->device_list, + gmc.xgmi.head) + if (amdgpu_ras_get_fed_status(remote_adev)) { + amdgpu_ras_set_fed_all(adev, hive, true); + break; + } + } if (!ras->disable_ras_err_cnt_harvest) { /* Build list of devices to query RAS related errors */ @@ -2439,18 +2497,6 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) ras->gpu_reset_flags &= ~AMDGPU_RAS_GPU_RESET_MODE1_RESET; set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); - /* For any RAS error that needs a full reset to - * recover, set the fatal error status - */ - if (hive) { - list_for_each_entry(remote_adev, - &hive->device_list, - gmc.xgmi.head) - amdgpu_ras_set_fed(remote_adev, - true); - } else { - amdgpu_ras_set_fed(adev, true); - } psp_fatal_error_recovery_quirk(&adev->psp); } } @@ -2516,9 +2562,7 @@ int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev, goto out; } - amdgpu_vram_mgr_reserve_range(&adev->mman.vram_mgr, - bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT, - AMDGPU_GPU_PAGE_SIZE); + amdgpu_ras_reserve_page(adev, bps[i].retired_page); memcpy(&data->bps[data->count], &bps[i], sizeof(*data->bps)); data->count++; @@ -2674,10 +2718,167 @@ static void amdgpu_ras_validate_threshold(struct amdgpu_device *adev, } } +int amdgpu_ras_put_poison_req(struct amdgpu_device *adev, + enum amdgpu_ras_block block, uint16_t pasid, + pasid_notify pasid_fn, void *data, uint32_t reset) +{ + int ret = 0; + struct ras_poison_msg poison_msg; + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + + memset(&poison_msg, 0, sizeof(poison_msg)); + poison_msg.block = block; + poison_msg.pasid = pasid; + poison_msg.reset = reset; + poison_msg.pasid_fn = pasid_fn; + poison_msg.data = data; + + ret = kfifo_put(&con->poison_fifo, poison_msg); + if (!ret) { + dev_err(adev->dev, "Poison message fifo is full!\n"); + return -ENOSPC; + } + + return 0; +} + +static int amdgpu_ras_get_poison_req(struct amdgpu_device *adev, + struct ras_poison_msg *poison_msg) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + + return kfifo_get(&con->poison_fifo, poison_msg); +} + +static void amdgpu_ras_ecc_log_init(struct ras_ecc_log_info *ecc_log) +{ + mutex_init(&ecc_log->lock); + + /* Set any value as siphash key */ + memset(&ecc_log->ecc_key, 0xad, sizeof(ecc_log->ecc_key)); + + INIT_RADIX_TREE(&ecc_log->de_page_tree, GFP_KERNEL); + ecc_log->de_updated = false; +} + +static void amdgpu_ras_ecc_log_fini(struct ras_ecc_log_info *ecc_log) +{ + struct radix_tree_iter iter; + void __rcu **slot; + struct ras_ecc_err *ecc_err; + + mutex_lock(&ecc_log->lock); + radix_tree_for_each_slot(slot, &ecc_log->de_page_tree, &iter, 0) { + ecc_err = radix_tree_deref_slot(slot); + kfree(ecc_err->err_pages.pfn); + kfree(ecc_err); + radix_tree_iter_delete(&ecc_log->de_page_tree, &iter, slot); + } + mutex_unlock(&ecc_log->lock); + + mutex_destroy(&ecc_log->lock); + ecc_log->de_updated = false; +} + +static void amdgpu_ras_do_page_retirement(struct work_struct *work) +{ + struct amdgpu_ras *con = container_of(work, struct amdgpu_ras, + page_retirement_dwork.work); + struct amdgpu_device *adev = con->adev; + struct ras_err_data err_data; + + if (amdgpu_in_reset(adev) || atomic_read(&con->in_recovery)) + return; + + amdgpu_ras_error_data_init(&err_data); + + amdgpu_umc_handle_bad_pages(adev, &err_data); + + amdgpu_ras_error_data_fini(&err_data); + + mutex_lock(&con->umc_ecc_log.lock); + if (radix_tree_tagged(&con->umc_ecc_log.de_page_tree, + UMC_ECC_NEW_DETECTED_TAG)) + schedule_delayed_work(&con->page_retirement_dwork, + msecs_to_jiffies(AMDGPU_RAS_RETIRE_PAGE_INTERVAL)); + mutex_unlock(&con->umc_ecc_log.lock); +} + +static int amdgpu_ras_query_ecc_status(struct amdgpu_device *adev, + enum amdgpu_ras_block ras_block, uint32_t timeout_ms) +{ + int ret = 0; + struct ras_ecc_log_info *ecc_log; + struct ras_query_if info; + uint32_t timeout = timeout_ms; + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + + memset(&info, 0, sizeof(info)); + info.head.block = ras_block; + + ecc_log = &ras->umc_ecc_log; + ecc_log->de_updated = false; + do { + ret = amdgpu_ras_query_error_status(adev, &info); + if (ret) { + dev_err(adev->dev, "Failed to query ras error! ret:%d\n", ret); + return ret; + } + + if (timeout && !ecc_log->de_updated) { + msleep(1); + timeout--; + } + } while (timeout && !ecc_log->de_updated); + + if (timeout_ms && !timeout) { + dev_warn(adev->dev, "Can't find deferred error\n"); + return -ETIMEDOUT; + } + + return 0; +} + +static void amdgpu_ras_poison_creation_handler(struct amdgpu_device *adev, + uint32_t timeout) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + int ret; + + ret = amdgpu_ras_query_ecc_status(adev, AMDGPU_RAS_BLOCK__UMC, timeout); + if (!ret) + schedule_delayed_work(&con->page_retirement_dwork, 0); +} + +static int amdgpu_ras_poison_consumption_handler(struct amdgpu_device *adev, + struct ras_poison_msg *poison_msg) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + uint32_t reset = poison_msg->reset; + uint16_t pasid = poison_msg->pasid; + + kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); + + if (poison_msg->pasid_fn) + poison_msg->pasid_fn(adev, pasid, poison_msg->data); + + if (reset) { + flush_delayed_work(&con->page_retirement_dwork); + + con->gpu_reset_flags |= reset; + amdgpu_ras_reset_gpu(adev); + } + + return 0; +} + static int amdgpu_ras_page_retirement_thread(void *param) { struct amdgpu_device *adev = (struct amdgpu_device *)param; struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct ras_poison_msg poison_msg; + enum amdgpu_ras_block ras_block; + bool poison_creation_is_handled = false; while (!kthread_should_stop()) { @@ -2688,13 +2889,34 @@ static int amdgpu_ras_page_retirement_thread(void *param) if (kthread_should_stop()) break; - dev_info(adev->dev, "Start processing page retirement. request:%d\n", - atomic_read(&con->page_retirement_req_cnt)); - atomic_dec(&con->page_retirement_req_cnt); - amdgpu_umc_bad_page_polling_timeout(adev, - false, MAX_UMC_POISON_POLLING_TIME_ASYNC); + if (!amdgpu_ras_get_poison_req(adev, &poison_msg)) + continue; + + ras_block = poison_msg.block; + + dev_info(adev->dev, "Start processing ras block %s(%d)\n", + ras_block_str(ras_block), ras_block); + + if (ras_block == AMDGPU_RAS_BLOCK__UMC) { + amdgpu_ras_poison_creation_handler(adev, + MAX_UMC_POISON_POLLING_TIME_ASYNC); + poison_creation_is_handled = true; + } else { + /* poison_creation_is_handled: + * false: no poison creation interrupt, but it has poison + * consumption interrupt. + * true: It has poison creation interrupt at the beginning, + * but it has no poison creation interrupt later. + */ + amdgpu_ras_poison_creation_handler(adev, + poison_creation_is_handled ? + 0 : MAX_UMC_POISON_POLLING_TIME_ASYNC); + + amdgpu_ras_poison_consumption_handler(adev, &poison_msg); + poison_creation_is_handled = false; + } } return 0; @@ -2763,6 +2985,8 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev) } } + mutex_init(&con->page_rsv_lock); + INIT_KFIFO(con->poison_fifo); mutex_init(&con->page_retirement_lock); init_waitqueue_head(&con->page_retirement_wq); atomic_set(&con->page_retirement_req_cnt, 0); @@ -2773,6 +2997,8 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev) dev_warn(adev->dev, "Failed to create umc_page_retirement thread!!!\n"); } + INIT_DELAYED_WORK(&con->page_retirement_dwork, amdgpu_ras_do_page_retirement); + amdgpu_ras_ecc_log_init(&con->umc_ecc_log); #ifdef CONFIG_X86_MCE_AMD if ((adev->asic_type == CHIP_ALDEBARAN) && (adev->gmc.xgmi.connected_to_cpu)) @@ -2813,8 +3039,14 @@ static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev) atomic_set(&con->page_retirement_req_cnt, 0); + mutex_destroy(&con->page_rsv_lock); + cancel_work_sync(&con->recovery_work); + cancel_delayed_work_sync(&con->page_retirement_dwork); + + amdgpu_ras_ecc_log_fini(&con->umc_ecc_log); + mutex_lock(&con->recovery_lock); con->eh_data = NULL; kfree(data->bps); @@ -3036,6 +3268,35 @@ static int amdgpu_get_ras_schema(struct amdgpu_device *adev) AMDGPU_RAS_ERROR__PARITY; } +static void ras_event_mgr_init(struct ras_event_manager *mgr) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(mgr->seqnos); i++) + atomic64_set(&mgr->seqnos[i], 0); +} + +static void amdgpu_ras_event_mgr_init(struct amdgpu_device *adev) +{ + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + struct amdgpu_hive_info *hive; + + if (!ras) + return; + + hive = amdgpu_get_xgmi_hive(adev); + ras->event_mgr = hive ? &hive->event_mgr : &ras->__event_mgr; + + /* init event manager with node 0 on xgmi system */ + if (!amdgpu_in_reset(adev)) { + if (!hive || adev->gmc.xgmi.node_id == 0) + ras_event_mgr_init(ras->event_mgr); + } + + if (hive) + amdgpu_put_xgmi_hive(hive); +} + int amdgpu_ras_init(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); @@ -3356,6 +3617,8 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) return 0; + amdgpu_ras_event_mgr_init(adev); + if (amdgpu_aca_is_enabled(adev)) { if (amdgpu_in_reset(adev)) r = amdgpu_aca_reset(adev); @@ -3472,14 +3735,39 @@ void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status) atomic_set(&ras->fed, !!status); } +bool amdgpu_ras_event_id_is_valid(struct amdgpu_device *adev, u64 id) +{ + return !(id & BIT_ULL(63)); +} + +u64 amdgpu_ras_acquire_event_id(struct amdgpu_device *adev, enum ras_event_type type) +{ + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + u64 id; + + switch (type) { + case RAS_EVENT_TYPE_ISR: + id = (u64)atomic64_read(&ras->event_mgr->seqnos[type]); + break; + case RAS_EVENT_TYPE_INVALID: + default: + id = BIT_ULL(63) | 0ULL; + break; + } + + return id; +} + void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev) { if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) { struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + u64 event_id = (u64)atomic64_inc_return(&ras->event_mgr->seqnos[RAS_EVENT_TYPE_ISR]); - dev_info(adev->dev, "uncorrectable hardware error" - "(ERREVENT_ATHUB_INTERRUPT) detected!\n"); + RAS_EVENT_LOG(adev, event_id, "uncorrectable hardware error" + "(ERREVENT_ATHUB_INTERRUPT) detected!\n"); + amdgpu_ras_set_fed(adev, true); ras->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE1_RESET; amdgpu_ras_reset_gpu(adev); } @@ -3998,6 +4286,8 @@ void amdgpu_ras_add_mca_err_addr(struct ras_err_info *err_info, struct ras_err_a { struct ras_err_addr *mca_err_addr; + /* This function will be retired. */ + return; mca_err_addr = kzalloc(sizeof(*mca_err_addr), GFP_KERNEL); if (!mca_err_addr) return; @@ -4195,3 +4485,19 @@ void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances) amdgpu_ras_boot_time_error_reporting(adev, i, boot_error); } } + +int amdgpu_ras_reserve_page(struct amdgpu_device *adev, uint64_t pfn) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr; + uint64_t start = pfn << AMDGPU_GPU_PAGE_SHIFT; + int ret = 0; + + mutex_lock(&con->page_rsv_lock); + ret = amdgpu_vram_mgr_query_page_status(mgr, start); + if (ret == -ENOENT) + ret = amdgpu_vram_mgr_reserve_range(mgr, start, AMDGPU_GPU_PAGE_SIZE); + mutex_unlock(&con->page_rsv_lock); + + return ret; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index db9cb2b4e9..7021c4a66f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -26,6 +26,9 @@ #include <linux/debugfs.h> #include <linux/list.h> +#include <linux/kfifo.h> +#include <linux/radix-tree.h> +#include <linux/siphash.h> #include "ta_ras_if.h" #include "amdgpu_ras_eeprom.h" #include "amdgpu_smuio.h" @@ -64,6 +67,14 @@ struct amdgpu_iv_entry; /* The high three bits indicates socketid */ #define AMDGPU_RAS_GET_FEATURES(val) ((val) & ~AMDGPU_RAS_FEATURES_SOCKETID_MASK) +#define RAS_EVENT_LOG(_adev, _id, _fmt, ...) \ +do { \ + if (amdgpu_ras_event_id_is_valid((_adev), (_id))) \ + dev_info((_adev)->dev, "{%llu}" _fmt, (_id), ##__VA_ARGS__); \ + else \ + dev_info((_adev)->dev, _fmt, ##__VA_ARGS__); \ +} while (0) + enum amdgpu_ras_block { AMDGPU_RAS_BLOCK__UMC = 0, AMDGPU_RAS_BLOCK__SDMA, @@ -419,6 +430,52 @@ struct umc_ecc_info { int record_ce_addr_supported; }; +enum ras_event_type { + RAS_EVENT_TYPE_INVALID = -1, + RAS_EVENT_TYPE_ISR = 0, + RAS_EVENT_TYPE_COUNT, +}; + +struct ras_event_manager { + atomic64_t seqnos[RAS_EVENT_TYPE_COUNT]; +}; + +struct ras_query_context { + enum ras_event_type type; + u64 event_id; +}; + +typedef int (*pasid_notify)(struct amdgpu_device *adev, + uint16_t pasid, void *data); + +struct ras_poison_msg { + enum amdgpu_ras_block block; + uint16_t pasid; + uint32_t reset; + pasid_notify pasid_fn; + void *data; +}; + +struct ras_err_pages { + uint32_t count; + uint64_t *pfn; +}; + +struct ras_ecc_err { + u64 hash_index; + uint64_t status; + uint64_t ipid; + uint64_t addr; + struct ras_err_pages err_pages; +}; + +struct ras_ecc_log_info { + struct mutex lock; + siphash_key_t ecc_key; + struct radix_tree_root de_page_tree; + bool de_updated; +}; + struct amdgpu_ras { /* ras infrastructure */ /* for ras itself. */ @@ -477,8 +534,18 @@ struct amdgpu_ras { wait_queue_head_t page_retirement_wq; struct mutex page_retirement_lock; atomic_t page_retirement_req_cnt; + struct mutex page_rsv_lock; + DECLARE_KFIFO(poison_fifo, struct ras_poison_msg, 128); + struct ras_ecc_log_info umc_ecc_log; + struct delayed_work page_retirement_dwork; + /* Fatal error detected flag */ atomic_t fed; + + /* RAS event manager */ + struct ras_event_manager __event_mgr; + struct ras_event_manager *event_mgr; + }; struct ras_fs_data { @@ -512,6 +579,7 @@ struct ras_err_data { unsigned long de_count; unsigned long err_addr_cnt; struct eeprom_table_record *err_addr; + unsigned long err_addr_len; u32 err_list_count; struct list_head err_node_list; }; @@ -879,4 +947,13 @@ void amdgpu_ras_del_mca_err_addr(struct ras_err_info *err_info, void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status); bool amdgpu_ras_get_fed_status(struct amdgpu_device *adev); +bool amdgpu_ras_event_id_is_valid(struct amdgpu_device *adev, u64 id); +u64 amdgpu_ras_acquire_event_id(struct amdgpu_device *adev, enum ras_event_type type); + +int amdgpu_ras_reserve_page(struct amdgpu_device *adev, uint64_t pfn); + +int amdgpu_ras_put_poison_req(struct amdgpu_device *adev, + enum amdgpu_ras_block block, uint16_t pasid, + pasid_notify pasid_fn, void *data, uint32_t reset); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index b12808c0c3..06a62a8a99 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -404,6 +404,22 @@ static int amdgpu_ras_eeprom_correct_header_tag( return res; } +static void amdgpu_ras_set_eeprom_table_version(struct amdgpu_ras_eeprom_control *control) +{ + struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr; + + switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) { + case IP_VERSION(8, 10, 0): + case IP_VERSION(12, 0, 0): + hdr->version = RAS_TABLE_VER_V2_1; + return; + default: + hdr->version = RAS_TABLE_VER_V1; + return; + } +} + /** * amdgpu_ras_eeprom_reset_table -- Reset the RAS EEPROM table * @control: pointer to control structure @@ -423,11 +439,7 @@ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control) mutex_lock(&control->ras_tbl_mutex); hdr->header = RAS_TABLE_HDR_VAL; - if (adev->umc.ras && - adev->umc.ras->set_eeprom_table_version) - adev->umc.ras->set_eeprom_table_version(hdr); - else - hdr->version = RAS_TABLE_VER_V1; + amdgpu_ras_set_eeprom_table_version(control); if (hdr->version == RAS_TABLE_VER_V2_1) { hdr->first_rec_offset = RAS_RECORD_START_V2_1; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h index 381101d2bf..50fcd86e10 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h @@ -164,4 +164,29 @@ static inline void amdgpu_res_next(struct amdgpu_res_cursor *cur, uint64_t size) } } +/** + * amdgpu_res_cleared - check if blocks are cleared + * + * @cur: the cursor to extract the block + * + * Check if the @cur block is cleared + */ +static inline bool amdgpu_res_cleared(struct amdgpu_res_cursor *cur) +{ + struct drm_buddy_block *block; + + switch (cur->mem_type) { + case TTM_PL_VRAM: + block = cur->node; + + if (!amdgpu_vram_mgr_is_cleared(block)) + return false; + break; + default: + return false; + } + + return true; +} + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c index 147100c27c..ea4873f6cc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c @@ -21,9 +21,6 @@ * */ -#include <linux/devcoredump.h> -#include <generated/utsrelease.h> - #include "amdgpu_reset.h" #include "aldebaran.h" #include "sienna_cichlid.h" @@ -161,105 +158,3 @@ void amdgpu_device_unlock_reset_domain(struct amdgpu_reset_domain *reset_domain) atomic_set(&reset_domain->in_gpu_reset, 0); up_write(&reset_domain->sem); } - -#ifndef CONFIG_DEV_COREDUMP -void amdgpu_coredump(struct amdgpu_device *adev, bool vram_lost, - struct amdgpu_reset_context *reset_context) -{ -} -#else -static ssize_t -amdgpu_devcoredump_read(char *buffer, loff_t offset, size_t count, - void *data, size_t datalen) -{ - struct drm_printer p; - struct amdgpu_coredump_info *coredump = data; - struct drm_print_iterator iter; - int i; - - iter.data = buffer; - iter.offset = 0; - iter.start = offset; - iter.remain = count; - - p = drm_coredump_printer(&iter); - - drm_printf(&p, "**** AMDGPU Device Coredump ****\n"); - drm_printf(&p, "version: " AMDGPU_COREDUMP_VERSION "\n"); - drm_printf(&p, "kernel: " UTS_RELEASE "\n"); - drm_printf(&p, "module: " KBUILD_MODNAME "\n"); - drm_printf(&p, "time: %lld.%09ld\n", coredump->reset_time.tv_sec, - coredump->reset_time.tv_nsec); - - if (coredump->reset_task_info.pid) - drm_printf(&p, "process_name: %s PID: %d\n", - coredump->reset_task_info.process_name, - coredump->reset_task_info.pid); - - if (coredump->ring) { - drm_printf(&p, "\nRing timed out details\n"); - drm_printf(&p, "IP Type: %d Ring Name: %s\n", - coredump->ring->funcs->type, - coredump->ring->name); - } - - if (coredump->reset_vram_lost) - drm_printf(&p, "VRAM is lost due to GPU reset!\n"); - if (coredump->adev->reset_info.num_regs) { - drm_printf(&p, "AMDGPU register dumps:\nOffset: Value:\n"); - - for (i = 0; i < coredump->adev->reset_info.num_regs; i++) - drm_printf(&p, "0x%08x: 0x%08x\n", - coredump->adev->reset_info.reset_dump_reg_list[i], - coredump->adev->reset_info.reset_dump_reg_value[i]); - } - - return count - iter.remain; -} - -static void amdgpu_devcoredump_free(void *data) -{ - kfree(data); -} - -void amdgpu_coredump(struct amdgpu_device *adev, bool vram_lost, - struct amdgpu_reset_context *reset_context) -{ - struct amdgpu_coredump_info *coredump; - struct drm_device *dev = adev_to_drm(adev); - struct amdgpu_job *job = reset_context->job; - struct drm_sched_job *s_job; - - coredump = kzalloc(sizeof(*coredump), GFP_NOWAIT); - - if (!coredump) { - DRM_ERROR("%s: failed to allocate memory for coredump\n", __func__); - return; - } - - coredump->reset_vram_lost = vram_lost; - - if (reset_context->job && reset_context->job->vm) { - struct amdgpu_task_info *ti; - struct amdgpu_vm *vm = reset_context->job->vm; - - ti = amdgpu_vm_get_task_info_vm(vm); - if (ti) { - coredump->reset_task_info = *ti; - amdgpu_vm_put_task_info(ti); - } - } - - if (job) { - s_job = &job->base; - coredump->ring = to_amdgpu_ring(s_job->sched); - } - - coredump->adev = adev; - - ktime_get_ts64(&coredump->reset_time); - - dev_coredumpm(dev->dev, THIS_MODULE, coredump, 0, GFP_NOWAIT, - amdgpu_devcoredump_read, amdgpu_devcoredump_free); -} -#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h index 60522963aa..b11d190ece 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h @@ -32,6 +32,7 @@ enum AMDGPU_RESET_FLAGS { AMDGPU_NEED_FULL_RESET = 0, AMDGPU_SKIP_HW_RESET = 1, + AMDGPU_SKIP_COREDUMP = 2, }; struct amdgpu_reset_context { @@ -88,19 +89,6 @@ struct amdgpu_reset_domain { atomic_t reset_res; }; -#ifdef CONFIG_DEV_COREDUMP - -#define AMDGPU_COREDUMP_VERSION "1" - -struct amdgpu_coredump_info { - struct amdgpu_device *adev; - struct amdgpu_task_info reset_task_info; - struct timespec64 reset_time; - bool reset_vram_lost; - struct amdgpu_ring *ring; -}; -#endif - int amdgpu_reset_init(struct amdgpu_device *adev); int amdgpu_reset_fini(struct amdgpu_device *adev); @@ -141,9 +129,6 @@ void amdgpu_device_lock_reset_domain(struct amdgpu_reset_domain *reset_domain); void amdgpu_device_unlock_reset_domain(struct amdgpu_reset_domain *reset_domain); -void amdgpu_coredump(struct amdgpu_device *adev, bool vram_lost, - struct amdgpu_reset_context *reset_context); - #define for_each_handler(i, handler, reset_ctl) \ for (i = 0; (i < AMDGPU_RESET_MAX_HANDLERS) && \ (handler = (*reset_ctl->reset_handlers)[i]); \ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h index 173a2a3080..b51a82e711 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h @@ -132,7 +132,7 @@ struct amdgpu_buffer_funcs { uint64_t dst_offset, /* number of byte to transfer */ uint32_t byte_count, - bool tmz); + uint32_t copy_flags); /* maximum bytes in a single operation */ uint32_t fill_max_bytes; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h index ff44351810..ec9d12f85f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h @@ -44,6 +44,7 @@ struct amdgpu_smuio_funcs { u32 (*get_socket_id)(struct amdgpu_device *adev); enum amdgpu_pkg_type (*get_pkg_type)(struct amdgpu_device *adev); bool (*is_host_gpu_xgmi_supported)(struct amdgpu_device *adev); + u64 (*get_gpu_clock_counter)(struct amdgpu_device *adev); }; struct amdgpu_smuio { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h index f539b1d002..383fce40d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h @@ -178,10 +178,10 @@ TRACE_EVENT(amdgpu_cs_ioctl, TP_fast_assign( __entry->sched_job_id = job->base.id; - __assign_str(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job)); + __assign_str(timeline); __entry->context = job->base.s_fence->finished.context; __entry->seqno = job->base.s_fence->finished.seqno; - __assign_str(ring, to_amdgpu_ring(job->base.sched)->name); + __assign_str(ring); __entry->num_ibs = job->num_ibs; ), TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u", @@ -203,10 +203,10 @@ TRACE_EVENT(amdgpu_sched_run_job, TP_fast_assign( __entry->sched_job_id = job->base.id; - __assign_str(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job)); + __assign_str(timeline); __entry->context = job->base.s_fence->finished.context; __entry->seqno = job->base.s_fence->finished.seqno; - __assign_str(ring, to_amdgpu_ring(job->base.sched)->name); + __assign_str(ring); __entry->num_ibs = job->num_ibs; ), TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u", @@ -231,7 +231,7 @@ TRACE_EVENT(amdgpu_vm_grab_id, TP_fast_assign( __entry->pasid = vm->pasid; - __assign_str(ring, ring->name); + __assign_str(ring); __entry->vmid = job->vmid; __entry->vm_hub = ring->vm_hub, __entry->pd_addr = job->vm_pd_addr; @@ -425,7 +425,7 @@ TRACE_EVENT(amdgpu_vm_flush, ), TP_fast_assign( - __assign_str(ring, ring->name); + __assign_str(ring); __entry->vmid = vmid; __entry->vm_hub = ring->vm_hub; __entry->pd_addr = pd_addr; @@ -526,7 +526,7 @@ TRACE_EVENT(amdgpu_ib_pipe_sync, ), TP_fast_assign( - __assign_str(ring, sched_job->base.sched->name); + __assign_str(ring); __entry->id = sched_job->base.id; __entry->fence = fence; __entry->ctx = fence->context; @@ -554,21 +554,6 @@ TRACE_EVENT(amdgpu_reset_reg_dumps, __entry->value) ); -TRACE_EVENT(amdgpu_runpm_reference_dumps, - TP_PROTO(uint32_t index, const char *func), - TP_ARGS(index, func), - TP_STRUCT__entry( - __field(uint32_t, index) - __string(func, func) - ), - TP_fast_assign( - __entry->index = index; - __assign_str(func, func); - ), - TP_printk("amdgpu runpm reference dump 0x%x: 0x%s\n", - __entry->index, - __get_str(func)) -); #undef AMDGPU_JOB_GET_TIMELINE_NAME #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 29c197c000..e785f12841 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -236,7 +236,7 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, dst_addr = amdgpu_bo_gpu_offset(adev->gart.bo); dst_addr += window * AMDGPU_GTT_MAX_TRANSFER_SIZE * 8; amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr, - dst_addr, num_bytes, false); + dst_addr, num_bytes, 0); amdgpu_ring_pad_ib(ring, &job->ibs[0]); WARN_ON(job->ibs[0].length_dw > num_dw); @@ -296,6 +296,8 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, struct dma_fence *fence = NULL; int r = 0; + uint32_t copy_flags = 0; + if (!adev->mman.buffer_funcs_enabled) { DRM_ERROR("Trying to move memory with ring turned off.\n"); return -EINVAL; @@ -323,8 +325,11 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, if (r) goto error; - r = amdgpu_copy_buffer(ring, from, to, cur_size, - resv, &next, false, true, tmz); + if (tmz) + copy_flags |= AMDGPU_COPY_FLAGS_TMZ; + + r = amdgpu_copy_buffer(ring, from, to, cur_size, resv, + &next, false, true, copy_flags); if (r) goto error; @@ -378,11 +383,12 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo, (abo->flags & AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE)) { struct dma_fence *wipe_fence = NULL; - r = amdgpu_fill_buffer(abo, AMDGPU_POISON, NULL, &wipe_fence, - false); + r = amdgpu_fill_buffer(abo, 0, NULL, &wipe_fence, + false); if (r) { goto error; } else if (wipe_fence) { + amdgpu_vram_mgr_set_cleared(bo->resource); dma_fence_put(fence); fence = wipe_fence; } @@ -1492,7 +1498,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo, swap(src_addr, dst_addr); amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr, dst_addr, - PAGE_SIZE, false); + PAGE_SIZE, 0); amdgpu_ring_pad_ib(adev->mman.buffer_funcs_ring, &job->ibs[0]); WARN_ON(job->ibs[0].length_dw > num_dw); @@ -2143,7 +2149,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, struct dma_resv *resv, struct dma_fence **fence, bool direct_submit, - bool vm_needs_flush, bool tmz) + bool vm_needs_flush, uint32_t copy_flags) { struct amdgpu_device *adev = ring->adev; unsigned int num_loops, num_dw; @@ -2169,8 +2175,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, uint32_t cur_size_in_bytes = min(byte_count, max_bytes); amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_offset, - dst_offset, cur_size_in_bytes, tmz); - + dst_offset, cur_size_in_bytes, copy_flags); src_offset += cur_size_in_bytes; dst_offset += cur_size_in_bytes; byte_count -= cur_size_in_bytes; @@ -2230,6 +2235,71 @@ static int amdgpu_ttm_fill_mem(struct amdgpu_ring *ring, uint32_t src_data, return 0; } +/** + * amdgpu_ttm_clear_buffer - clear memory buffers + * @bo: amdgpu buffer object + * @resv: reservation object + * @fence: dma_fence associated with the operation + * + * Clear the memory buffer resource. + * + * Returns: + * 0 for success or a negative error code on failure. + */ +int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo, + struct dma_resv *resv, + struct dma_fence **fence) +{ + struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); + struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; + struct amdgpu_res_cursor cursor; + u64 addr; + int r; + + if (!adev->mman.buffer_funcs_enabled) + return -EINVAL; + + if (!fence) + return -EINVAL; + + *fence = dma_fence_get_stub(); + + amdgpu_res_first(bo->tbo.resource, 0, amdgpu_bo_size(bo), &cursor); + + mutex_lock(&adev->mman.gtt_window_lock); + while (cursor.remaining) { + struct dma_fence *next = NULL; + u64 size; + + if (amdgpu_res_cleared(&cursor)) { + amdgpu_res_next(&cursor, cursor.size); + continue; + } + + /* Never clear more than 256MiB at once to avoid timeouts */ + size = min(cursor.size, 256ULL << 20); + + r = amdgpu_ttm_map_buffer(&bo->tbo, bo->tbo.resource, &cursor, + 1, ring, false, &size, &addr); + if (r) + goto err; + + r = amdgpu_ttm_fill_mem(ring, 0, addr, size, resv, + &next, true, true); + if (r) + goto err; + + dma_fence_put(*fence); + *fence = next; + + amdgpu_res_next(&cursor, size); + } +err: + mutex_unlock(&adev->mman.gtt_window_lock); + + return r; +} + int amdgpu_fill_buffer(struct amdgpu_bo *bo, uint32_t src_data, struct dma_resv *resv, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index 32cf6b6f6e..b6f53129de 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -38,8 +38,6 @@ #define AMDGPU_GTT_MAX_TRANSFER_SIZE 512 #define AMDGPU_GTT_NUM_TRANSFER_WINDOWS 2 -#define AMDGPU_POISON 0xd0bed0be - extern const struct attribute_group amdgpu_vram_mgr_attr_group; extern const struct attribute_group amdgpu_gtt_mgr_attr_group; @@ -111,6 +109,8 @@ struct amdgpu_copy_mem { unsigned long offset; }; +#define AMDGPU_COPY_FLAGS_TMZ (1 << 0) + int amdgpu_gtt_mgr_init(struct amdgpu_device *adev, uint64_t gtt_size); void amdgpu_gtt_mgr_fini(struct amdgpu_device *adev); int amdgpu_preempt_mgr_init(struct amdgpu_device *adev); @@ -151,13 +151,16 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, struct dma_resv *resv, struct dma_fence **fence, bool direct_submit, - bool vm_needs_flush, bool tmz); + bool vm_needs_flush, uint32_t copy_flags); int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, const struct amdgpu_copy_mem *src, const struct amdgpu_copy_mem *dst, uint64_t size, bool tmz, struct dma_resv *resv, struct dma_fence **f); +int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo, + struct dma_resv *resv, + struct dma_fence **fence); int amdgpu_fill_buffer(struct amdgpu_bo *bo, uint32_t src_data, struct dma_resv *resv, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h index 6194457600..105d4de061 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h @@ -125,6 +125,7 @@ enum psp_fw_type { PSP_FW_TYPE_PSP_INTF_DRV, PSP_FW_TYPE_PSP_DBG_DRV, PSP_FW_TYPE_PSP_RAS_DRV, + PSP_FW_TYPE_PSP_IPKEYMGR_DRV, PSP_FW_TYPE_MAX_INDEX, }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index 6f7451e3ee..540e0f066b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -21,10 +21,13 @@ * */ +#include <linux/sort.h> #include "amdgpu.h" #include "umc_v6_7.h" #define MAX_UMC_POISON_POLLING_TIME_SYNC 20 //ms +#define MAX_UMC_HASH_STRING_SIZE 256 + static int amdgpu_umc_convert_error_address(struct amdgpu_device *adev, struct ras_err_data *err_data, uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst) @@ -63,6 +66,8 @@ int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev, goto out_fini_err_data; } + err_data.err_addr_len = adev->umc.max_ras_err_cnt_per_query; + /* * Translate UMC channel address to Physical address */ @@ -86,7 +91,7 @@ out_fini_err_data: return ret; } -static void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev, +void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev, void *ras_error_status) { struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; @@ -118,6 +123,8 @@ static void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev, if(!err_data->err_addr) dev_warn(adev->dev, "Failed to alloc memory for " "umc error address record!\n"); + else + err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query; /* umc query_ras_error_address is also responsible for clearing * error status @@ -143,6 +150,8 @@ static void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev, if(!err_data->err_addr) dev_warn(adev->dev, "Failed to alloc memory for " "umc error address record!\n"); + else + err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query; /* umc query_ras_error_address is also responsible for clearing * error status @@ -178,7 +187,7 @@ static void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev, static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, void *ras_error_status, struct amdgpu_iv_entry *entry, - bool reset) + uint32_t reset) { struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; struct amdgpu_ras *con = amdgpu_ras_get_context(adev); @@ -187,9 +196,7 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, amdgpu_umc_handle_bad_pages(adev, ras_error_status); if (err_data->ue_count && reset) { - /* use mode-2 reset for poison consumption */ - if (!entry) - con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET; + con->gpu_reset_flags |= reset; amdgpu_ras_reset_gpu(adev); } @@ -197,7 +204,7 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, } int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev, - bool reset, uint32_t timeout_ms) + uint32_t reset, uint32_t timeout_ms) { struct ras_err_data err_data; struct ras_common_if head = { @@ -239,16 +246,16 @@ int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev, if (reset) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - /* use mode-2 reset for poison consumption */ - con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET; + con->gpu_reset_flags |= reset; amdgpu_ras_reset_gpu(adev); } return 0; } -int amdgpu_umc_poison_handler(struct amdgpu_device *adev, - enum amdgpu_ras_block block, bool reset) +int amdgpu_umc_pasid_poison_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block, uint16_t pasid, + pasid_notify pasid_fn, void *data, uint32_t reset) { int ret = AMDGPU_RAS_SUCCESS; @@ -286,16 +293,14 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev, amdgpu_ras_error_data_fini(&err_data); } else { - if (reset) { - amdgpu_umc_bad_page_polling_timeout(adev, - reset, MAX_UMC_POISON_POLLING_TIME_SYNC); - } else { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + amdgpu_ras_put_poison_req(adev, + block, pasid, pasid_fn, data, reset); + atomic_inc(&con->page_retirement_req_cnt); wake_up(&con->page_retirement_wq); - } } } else { if (adev->virt.ops && adev->virt.ops->ras_poison_handler) @@ -308,11 +313,19 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev, return ret; } +int amdgpu_umc_poison_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block, uint32_t reset) +{ + return amdgpu_umc_pasid_poison_handler(adev, + block, 0, NULL, NULL, reset); +} + int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, void *ras_error_status, struct amdgpu_iv_entry *entry) { - return amdgpu_umc_do_page_retirement(adev, ras_error_status, entry, true); + return amdgpu_umc_do_page_retirement(adev, ras_error_status, entry, + AMDGPU_RAS_GPU_RESET_MODE1_RESET); } int amdgpu_umc_ras_sw_init(struct amdgpu_device *adev) @@ -389,14 +402,20 @@ int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev, return 0; } -void amdgpu_umc_fill_error_record(struct ras_err_data *err_data, +int amdgpu_umc_fill_error_record(struct ras_err_data *err_data, uint64_t err_addr, uint64_t retired_page, uint32_t channel_index, uint32_t umc_inst) { - struct eeprom_table_record *err_rec = - &err_data->err_addr[err_data->err_addr_cnt]; + struct eeprom_table_record *err_rec; + + if (!err_data || + !err_data->err_addr || + (err_data->err_addr_cnt >= err_data->err_addr_len)) + return -EINVAL; + + err_rec = &err_data->err_addr[err_data->err_addr_cnt]; err_rec->address = err_addr; /* page frame address is saved */ @@ -408,6 +427,8 @@ void amdgpu_umc_fill_error_record(struct ras_err_data *err_data, err_rec->mcumc_id = umc_inst; err_data->err_addr_cnt++; + + return 0; } int amdgpu_umc_loop_channels(struct amdgpu_device *adev, @@ -440,3 +461,76 @@ int amdgpu_umc_loop_channels(struct amdgpu_device *adev, return 0; } + +int amdgpu_umc_update_ecc_status(struct amdgpu_device *adev, + uint64_t status, uint64_t ipid, uint64_t addr) +{ + if (adev->umc.ras->update_ecc_status) + return adev->umc.ras->update_ecc_status(adev, + status, ipid, addr); + return 0; +} + +static int amdgpu_umc_uint64_cmp(const void *a, const void *b) +{ + uint64_t *addr_a = (uint64_t *)a; + uint64_t *addr_b = (uint64_t *)b; + + if (*addr_a > *addr_b) + return 1; + else if (*addr_a < *addr_b) + return -1; + else + return 0; +} + +/* Use string hash to avoid logging the same bad pages repeatedly */ +int amdgpu_umc_build_pages_hash(struct amdgpu_device *adev, + uint64_t *pfns, int len, uint64_t *val) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + char buf[MAX_UMC_HASH_STRING_SIZE] = {0}; + int offset = 0, i = 0; + uint64_t hash_val; + + if (!pfns || !len) + return -EINVAL; + + sort(pfns, len, sizeof(uint64_t), amdgpu_umc_uint64_cmp, NULL); + + for (i = 0; i < len; i++) + offset += snprintf(&buf[offset], sizeof(buf) - offset, "%llx", pfns[i]); + + hash_val = siphash(buf, offset, &con->umc_ecc_log.ecc_key); + + *val = hash_val; + + return 0; +} + +int amdgpu_umc_logs_ecc_err(struct amdgpu_device *adev, + struct radix_tree_root *ecc_tree, struct ras_ecc_err *ecc_err) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct ras_ecc_log_info *ecc_log; + int ret; + + ecc_log = &con->umc_ecc_log; + + mutex_lock(&ecc_log->lock); + ret = radix_tree_insert(ecc_tree, ecc_err->hash_index, ecc_err); + if (!ret) { + struct ras_err_pages *err_pages = &ecc_err->err_pages; + int i; + + /* Reserve memory */ + for (i = 0; i < err_pages->count; i++) + amdgpu_ras_reserve_page(adev, err_pages->pfn[i]); + + radix_tree_tag_set(ecc_tree, + ecc_err->hash_index, UMC_ECC_NEW_DETECTED_TAG); + } + mutex_unlock(&ecc_log->lock); + + return ret; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index 26d2ae498d..5f50c69c3c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -52,6 +52,8 @@ #define LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) \ LOOP_UMC_NODE_INST((node_inst)) LOOP_UMC_INST_AND_CH((umc_inst), (ch_inst)) +/* Page retirement tag */ +#define UMC_ECC_NEW_DETECTED_TAG 0x1 typedef int (*umc_func)(struct amdgpu_device *adev, uint32_t node_inst, uint32_t umc_inst, uint32_t ch_inst, void *data); @@ -66,8 +68,8 @@ struct amdgpu_umc_ras { void *ras_error_status); bool (*check_ecc_err_status)(struct amdgpu_device *adev, enum amdgpu_mca_error_type type, void *ras_error_status); - /* support different eeprom table version for different asic */ - void (*set_eeprom_table_version)(struct amdgpu_ras_eeprom_table_header *hdr); + int (*update_ecc_status)(struct amdgpu_device *adev, + uint64_t status, uint64_t ipid, uint64_t addr); }; struct amdgpu_umc_funcs { @@ -103,11 +105,14 @@ struct amdgpu_umc { int amdgpu_umc_ras_sw_init(struct amdgpu_device *adev); int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block); int amdgpu_umc_poison_handler(struct amdgpu_device *adev, - enum amdgpu_ras_block block, bool reset); + enum amdgpu_ras_block block, uint32_t reset); +int amdgpu_umc_pasid_poison_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block, uint16_t pasid, + pasid_notify pasid_fn, void *data, uint32_t reset); int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry); -void amdgpu_umc_fill_error_record(struct ras_err_data *err_data, +int amdgpu_umc_fill_error_record(struct ras_err_data *err_data, uint64_t err_addr, uint64_t retired_page, uint32_t channel_index, @@ -123,5 +128,15 @@ int amdgpu_umc_loop_channels(struct amdgpu_device *adev, umc_func func, void *data); int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev, - bool reset, uint32_t timeout_ms); + uint32_t reset, uint32_t timeout_ms); + +int amdgpu_umc_update_ecc_status(struct amdgpu_device *adev, + uint64_t status, uint64_t ipid, uint64_t addr); +int amdgpu_umc_build_pages_hash(struct amdgpu_device *adev, + uint64_t *pfns, int len, uint64_t *val); +int amdgpu_umc_logs_ecc_err(struct amdgpu_device *adev, + struct radix_tree_root *ecc_tree, struct ras_ecc_err *ecc_err); + +void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev, + void *ras_error_status); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c index f7c73533e3..e01c1c8e64 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c @@ -878,6 +878,8 @@ static const struct amd_ip_funcs umsch_mm_v4_0_ip_funcs = { .hw_fini = umsch_mm_hw_fini, .suspend = umsch_mm_suspend, .resume = umsch_mm_resume, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version umsch_mm_v4_0_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index 9c514a606a..677eb14155 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -93,7 +93,7 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work); int amdgpu_vcn_early_init(struct amdgpu_device *adev) { - char ucode_prefix[30]; + char ucode_prefix[25]; char fw_name[40]; int r, i; @@ -185,7 +185,10 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8); - if (amdgpu_ip_version(adev, UVD_HWIP, 0) >= IP_VERSION(4, 0, 0)) { + if (amdgpu_ip_version(adev, UVD_HWIP, 0) >= IP_VERSION(5, 0, 0)) { + fw_shared_size = AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn5_fw_shared)); + log_offset = offsetof(struct amdgpu_vcn5_fw_shared, fw_log); + } else if (amdgpu_ip_version(adev, UVD_HWIP, 0) >= IP_VERSION(4, 0, 0)) { fw_shared_size = AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)); log_offset = offsetof(struct amdgpu_vcn4_fw_shared, fw_log); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h index a418393d89..9f06def236 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h @@ -454,6 +454,16 @@ struct amdgpu_vcn_rb_metadata { uint8_t pad[26]; }; +struct amdgpu_vcn5_fw_shared { + uint32_t present_flag_0; + uint8_t pad[12]; + struct amdgpu_fw_shared_unified_queue_struct sq; + uint8_t pad1[8]; + struct amdgpu_fw_shared_fw_logging fw_log; + struct amdgpu_fw_shared_rb_setup rb_setup; + uint8_t pad2[4]; +}; + #define VCN_BLOCK_ENCODE_DISABLE_MASK 0x80 #define VCN_BLOCK_DECODE_DISABLE_MASK 0x40 #define VCN_BLOCK_QUEUE_DISABLE_MASK 0xC0 diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 7a4eae3677..54ab51a4ad 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -32,6 +32,7 @@ #include "amdgpu.h" #include "amdgpu_ras.h" +#include "amdgpu_reset.h" #include "vi.h" #include "soc15.h" #include "nv.h" @@ -424,7 +425,7 @@ static int amdgpu_virt_read_pf2vf_data(struct amdgpu_device *adev) return -EINVAL; if (pf2vf_info->size > 1024) { - DRM_ERROR("invalid pf2vf message size\n"); + dev_err(adev->dev, "invalid pf2vf message size: 0x%x\n", pf2vf_info->size); return -EINVAL; } @@ -435,7 +436,9 @@ static int amdgpu_virt_read_pf2vf_data(struct amdgpu_device *adev) 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"); + dev_err(adev->dev, + "invalid pf2vf message: header checksum=0x%x calculated checksum=0x%x\n", + checksum, checkval); return -EINVAL; } @@ -449,7 +452,9 @@ static int amdgpu_virt_read_pf2vf_data(struct amdgpu_device *adev) adev->virt.fw_reserve.p_pf2vf, pf2vf_info->size, 0, checksum); if (checksum != checkval) { - DRM_ERROR("invalid pf2vf message\n"); + dev_err(adev->dev, + "invalid pf2vf message: header checksum=0x%x calculated checksum=0x%x\n", + checksum, checkval); return -EINVAL; } @@ -485,7 +490,7 @@ static int amdgpu_virt_read_pf2vf_data(struct amdgpu_device *adev) ((struct amd_sriov_msg_pf2vf_info *)pf2vf_info)->uuid; break; default: - DRM_ERROR("invalid pf2vf version\n"); + dev_err(adev->dev, "invalid pf2vf version: 0x%x\n", pf2vf_info->version); return -EINVAL; } @@ -571,6 +576,11 @@ static int amdgpu_virt_write_vf2pf_data(struct amdgpu_device *adev) vf2pf_info->decode_usage = 0; vf2pf_info->dummy_page_addr = (uint64_t)adev->dummy_page_addr; + vf2pf_info->mes_info_addr = (uint64_t)adev->mes.resource_1_gpu_addr; + + if (adev->mes.resource_1) { + vf2pf_info->mes_info_size = adev->mes.resource_1->tbo.base.size; + } vf2pf_info->checksum = amd_sriov_msg_checksum( vf2pf_info, vf2pf_info->header.size, 0, 0); @@ -584,8 +594,22 @@ static void amdgpu_virt_update_vf2pf_work_item(struct work_struct *work) int ret; ret = amdgpu_virt_read_pf2vf_data(adev); - if (ret) + if (ret) { + adev->virt.vf2pf_update_retry_cnt++; + if ((adev->virt.vf2pf_update_retry_cnt >= AMDGPU_VF2PF_UPDATE_MAX_RETRY_LIMIT) && + amdgpu_sriov_runtime(adev) && !amdgpu_in_reset(adev)) { + amdgpu_ras_set_fed(adev, true); + if (amdgpu_reset_domain_schedule(adev->reset_domain, + &adev->virt.flr_work)) + return; + else + dev_err(adev->dev, "Failed to queue work! at %s", __func__); + } + goto out; + } + + adev->virt.vf2pf_update_retry_cnt = 0; amdgpu_virt_write_vf2pf_data(adev); out: @@ -606,6 +630,7 @@ void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev) adev->virt.fw_reserve.p_pf2vf = NULL; adev->virt.fw_reserve.p_vf2pf = NULL; adev->virt.vf2pf_update_interval_ms = 0; + adev->virt.vf2pf_update_retry_cnt = 0; if (adev->mman.fw_vram_usage_va && adev->mman.drv_vram_usage_va) { DRM_WARN("Currently fw_vram and drv_vram should not have values at the same time!"); @@ -705,12 +730,6 @@ void amdgpu_detect_virtualization(struct amdgpu_device *adev) adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE; } - if (amdgpu_sriov_vf(adev) && adev->asic_type == CHIP_SIENNA_CICHLID) - /* VF MMIO access (except mailbox range) from CPU - * will be blocked during sriov runtime - */ - adev->virt.caps |= AMDGPU_VF_MMIO_ACCESS_PROTECT; - /* we have the ability to check now */ if (amdgpu_sriov_vf(adev)) { switch (adev->asic_type) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index 3f59b7b552..642f1fd287 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -52,6 +52,8 @@ /* tonga/fiji use this offset */ #define mmBIF_IOV_FUNC_IDENTIFIER 0x1503 +#define AMDGPU_VF2PF_UPDATE_MAX_RETRY_LIMIT 5 + enum amdgpu_sriov_vf_mode { SRIOV_VF_MODE_BARE_METAL = 0, SRIOV_VF_MODE_ONE_VF, @@ -130,6 +132,8 @@ enum AMDGIM_FEATURE_FLAG { AMDGIM_FEATURE_AV1_SUPPORT = (1 << 6), /* VCN RB decouple */ AMDGIM_FEATURE_VCN_RB_DECOUPLE = (1 << 7), + /* MES info */ + AMDGIM_FEATURE_MES_INFO_ENABLE = (1 << 8), }; enum AMDGIM_REG_ACCESS_FLAG { @@ -257,6 +261,7 @@ struct amdgpu_virt { /* vf2pf message */ struct delayed_work vf2pf_work; uint32_t vf2pf_update_interval_ms; + int vf2pf_update_retry_cnt; /* multimedia bandwidth config */ bool is_mm_bw_enabled; @@ -332,6 +337,8 @@ static inline bool is_virtual_machine(void) ((adev)->virt.gim_feature & AMDGIM_FEATURE_AV1_SUPPORT) #define amdgpu_sriov_is_vcn_rb_decouple(adev) \ ((adev)->virt.gim_feature & AMDGIM_FEATURE_VCN_RB_DECOUPLE) +#define amdgpu_sriov_is_mes_info_enable(adev) \ + ((adev)->virt.gim_feature & AMDGIM_FEATURE_MES_INFO_ENABLE) bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev); void amdgpu_virt_init_setting(struct amdgpu_device *adev); int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c index ba6d1876ce..fde66225c4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c @@ -672,6 +672,8 @@ static const struct amd_ip_funcs amdgpu_vkms_ip_funcs = { .soft_reset = amdgpu_vkms_soft_reset, .set_clockgating_state = amdgpu_vkms_set_clockgating_state, .set_powergating_state = amdgpu_vkms_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version amdgpu_vkms_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 94089069c9..0f71060664 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -434,7 +434,7 @@ uint64_t amdgpu_vm_generation(struct amdgpu_device *adev, struct amdgpu_vm *vm) if (!vm) return result; - result += vm->generation; + result += lower_32_bits(vm->generation); /* Add one if the page tables will be re-generated on next CS */ if (drm_sched_entity_error(&vm->delayed)) ++result; @@ -463,13 +463,14 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm, int (*validate)(void *p, struct amdgpu_bo *bo), void *param) { + uint64_t new_vm_generation = amdgpu_vm_generation(adev, vm); struct amdgpu_vm_bo_base *bo_base; struct amdgpu_bo *shadow; struct amdgpu_bo *bo; int r; - if (drm_sched_entity_error(&vm->delayed)) { - ++vm->generation; + if (vm->generation != new_vm_generation) { + vm->generation = new_vm_generation; amdgpu_vm_bo_reset_state_machine(vm); amdgpu_vm_fini_entities(vm); r = amdgpu_vm_init_entities(adev, vm); @@ -886,6 +887,44 @@ static void amdgpu_vm_tlb_seq_cb(struct dma_fence *fence, } /** + * amdgpu_vm_tlb_flush - prepare TLB flush + * + * @params: parameters for update + * @fence: input fence to sync TLB flush with + * @tlb_cb: the callback structure + * + * Increments the tlb sequence to make sure that future CS execute a VM flush. + */ +static void +amdgpu_vm_tlb_flush(struct amdgpu_vm_update_params *params, + struct dma_fence **fence, + struct amdgpu_vm_tlb_seq_struct *tlb_cb) +{ + struct amdgpu_vm *vm = params->vm; + + if (!fence || !*fence) + return; + + tlb_cb->vm = vm; + if (!dma_fence_add_callback(*fence, &tlb_cb->cb, + amdgpu_vm_tlb_seq_cb)) { + dma_fence_put(vm->last_tlb_flush); + vm->last_tlb_flush = dma_fence_get(*fence); + } else { + amdgpu_vm_tlb_seq_cb(NULL, &tlb_cb->cb); + } + + /* Prepare a TLB flush fence to be attached to PTs */ + if (!params->unlocked && vm->is_compute_context) { + amdgpu_vm_tlb_fence_create(params->adev, vm, fence); + + /* Makes sure no PD/PT is freed before the flush */ + dma_resv_add_fence(vm->root.bo->tbo.base.resv, *fence, + DMA_RESV_USAGE_BOOKKEEP); + } +} + +/** * amdgpu_vm_update_range - update a range in the vm page table * * @adev: amdgpu_device pointer to use for commands @@ -916,8 +955,8 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm, struct ttm_resource *res, dma_addr_t *pages_addr, struct dma_fence **fence) { - struct amdgpu_vm_update_params params; struct amdgpu_vm_tlb_seq_struct *tlb_cb; + struct amdgpu_vm_update_params params; struct amdgpu_res_cursor cursor; enum amdgpu_sync_mode sync_mode; int r, idx; @@ -927,8 +966,8 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm, tlb_cb = kmalloc(sizeof(*tlb_cb), GFP_KERNEL); if (!tlb_cb) { - r = -ENOMEM; - goto error_unlock; + drm_dev_exit(idx); + return -ENOMEM; } /* Vega20+XGMI where PTEs get inadvertently cached in L2 texture cache, @@ -948,7 +987,9 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm, params.immediate = immediate; params.pages_addr = pages_addr; params.unlocked = unlocked; + params.needs_flush = flush_tlb; params.allow_override = allow_override; + INIT_LIST_HEAD(¶ms.tlb_flush_waitlist); /* Implicitly sync to command submissions in the same VM before * unmapping. Sync to moving fences before mapping. @@ -1031,24 +1072,18 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm, } r = vm->update_funcs->commit(¶ms, fence); + if (r) + goto error_free; - if (flush_tlb || params.table_freed) { - tlb_cb->vm = vm; - if (fence && *fence && - !dma_fence_add_callback(*fence, &tlb_cb->cb, - amdgpu_vm_tlb_seq_cb)) { - dma_fence_put(vm->last_tlb_flush); - vm->last_tlb_flush = dma_fence_get(*fence); - } else { - amdgpu_vm_tlb_seq_cb(NULL, &tlb_cb->cb); - } + if (params.needs_flush) { + amdgpu_vm_tlb_flush(¶ms, fence, tlb_cb); tlb_cb = NULL; } + amdgpu_vm_pt_free_list(adev, ¶ms); + error_free: kfree(tlb_cb); - -error_unlock: amdgpu_vm_eviction_unlock(vm); drm_dev_exit(idx); return r; @@ -2407,10 +2442,11 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, vm->last_update = dma_fence_get_stub(); vm->last_unlocked = dma_fence_get_stub(); vm->last_tlb_flush = dma_fence_get_stub(); - vm->generation = 0; + vm->generation = amdgpu_vm_generation(adev, NULL); mutex_init(&vm->eviction_lock); vm->evicting = false; + vm->tlb_fence_context = dma_fence_context_alloc(1); r = amdgpu_vm_pt_create(adev, vm, adev->vm_manager.root_level, false, &root, xcp_id); @@ -2944,6 +2980,14 @@ void amdgpu_vm_update_fault_cache(struct amdgpu_device *adev, if (vm && status) { vm->fault_info.addr = addr; vm->fault_info.status = status; + /* + * Update the fault information globally for later usage + * when vm could be stale or freed. + */ + adev->vm_manager.fault_info.addr = addr; + adev->vm_manager.fault_info.vmhub = vmhub; + adev->vm_manager.fault_info.status = status; + if (AMDGPU_IS_GFXHUB(vmhub)) { vm->fault_info.vmhub = AMDGPU_VMHUB_TYPE_GFX; vm->fault_info.vmhub |= diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 047ec1930d..54d7da396d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -257,15 +257,20 @@ struct amdgpu_vm_update_params { unsigned int num_dw_left; /** - * @table_freed: return true if page table is freed when updating + * @needs_flush: true whenever we need to invalidate the TLB */ - bool table_freed; + bool needs_flush; /** * @allow_override: true for memory that is not uncached: allows MTYPE * to be overridden for NUMA local memory. */ bool allow_override; + + /** + * @tlb_flush_waitlist: temporary storage for BOs until tlb_flush + */ + struct list_head tlb_flush_waitlist; }; struct amdgpu_vm_update_funcs { @@ -342,6 +347,7 @@ struct amdgpu_vm { atomic64_t tlb_seq; struct dma_fence *last_tlb_flush; atomic64_t kfd_last_flushed_seq; + uint64_t tlb_fence_context; /* How many times we had to re-generate the page tables */ uint64_t generation; @@ -422,6 +428,8 @@ struct amdgpu_vm_manager { * look up VM of a page fault */ struct xarray pasids; + /* Global registration of recent page fault information */ + struct amdgpu_vm_fault_info fault_info; }; struct amdgpu_bo_va_mapping; @@ -544,6 +552,8 @@ int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params, uint64_t start, uint64_t end, uint64_t dst, uint64_t flags); void amdgpu_vm_pt_free_work(struct work_struct *work); +void amdgpu_vm_pt_free_list(struct amdgpu_device *adev, + struct amdgpu_vm_update_params *params); #if defined(CONFIG_DEBUG_FS) void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m); @@ -609,5 +619,8 @@ void amdgpu_vm_update_fault_cache(struct amdgpu_device *adev, uint64_t addr, uint32_t status, unsigned int vmhub); +void amdgpu_vm_tlb_fence_create(struct amdgpu_device *adev, + struct amdgpu_vm *vm, + struct dma_fence **fence); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c index 6e31621452..3895bd7d17 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c @@ -108,7 +108,9 @@ static int amdgpu_vm_cpu_update(struct amdgpu_vm_update_params *p, static int amdgpu_vm_cpu_commit(struct amdgpu_vm_update_params *p, struct dma_fence **fence) { - /* Flush HDP */ + if (p->needs_flush) + atomic64_inc(&p->vm->tlb_seq); + mb(); amdgpu_device_flush_hdp(p->adev, NULL); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c index 124389a6bf..f07647a9a9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c @@ -622,40 +622,58 @@ void amdgpu_vm_pt_free_work(struct work_struct *work) } /** - * amdgpu_vm_pt_free_dfs - free PD/PT levels + * amdgpu_vm_pt_free_list - free PD/PT levels * * @adev: amdgpu device structure - * @vm: amdgpu vm structure - * @start: optional cursor where to start freeing PDs/PTs - * @unlocked: vm resv unlock status + * @params: see amdgpu_vm_update_params definition * - * Free the page directory or page table level and all sub levels. + * Free the page directory objects saved in the flush list */ -static void amdgpu_vm_pt_free_dfs(struct amdgpu_device *adev, - struct amdgpu_vm *vm, - struct amdgpu_vm_pt_cursor *start, - bool unlocked) +void amdgpu_vm_pt_free_list(struct amdgpu_device *adev, + struct amdgpu_vm_update_params *params) { - struct amdgpu_vm_pt_cursor cursor; - struct amdgpu_vm_bo_base *entry; + struct amdgpu_vm_bo_base *entry, *next; + struct amdgpu_vm *vm = params->vm; + bool unlocked = params->unlocked; + + if (list_empty(¶ms->tlb_flush_waitlist)) + return; if (unlocked) { spin_lock(&vm->status_lock); - for_each_amdgpu_vm_pt_dfs_safe(adev, vm, start, cursor, entry) - list_move(&entry->vm_status, &vm->pt_freed); - - if (start) - list_move(&start->entry->vm_status, &vm->pt_freed); + list_splice_init(¶ms->tlb_flush_waitlist, &vm->pt_freed); spin_unlock(&vm->status_lock); schedule_work(&vm->pt_free_work); return; } - for_each_amdgpu_vm_pt_dfs_safe(adev, vm, start, cursor, entry) + list_for_each_entry_safe(entry, next, ¶ms->tlb_flush_waitlist, vm_status) amdgpu_vm_pt_free(entry); +} - if (start) - amdgpu_vm_pt_free(start->entry); +/** + * amdgpu_vm_pt_add_list - add PD/PT level to the flush list + * + * @params: parameters for the update + * @cursor: first PT entry to start DF search from, non NULL + * + * This list will be freed after TLB flush. + */ +static void amdgpu_vm_pt_add_list(struct amdgpu_vm_update_params *params, + struct amdgpu_vm_pt_cursor *cursor) +{ + struct amdgpu_vm_pt_cursor seek; + struct amdgpu_vm_bo_base *entry; + + spin_lock(¶ms->vm->status_lock); + for_each_amdgpu_vm_pt_dfs_safe(params->adev, params->vm, cursor, seek, entry) { + if (entry && entry->bo) + list_move(&entry->vm_status, ¶ms->tlb_flush_waitlist); + } + + /* enter start node now */ + list_move(&cursor->entry->vm_status, ¶ms->tlb_flush_waitlist); + spin_unlock(¶ms->vm->status_lock); } /** @@ -667,7 +685,13 @@ static void amdgpu_vm_pt_free_dfs(struct amdgpu_device *adev, */ void amdgpu_vm_pt_free_root(struct amdgpu_device *adev, struct amdgpu_vm *vm) { - amdgpu_vm_pt_free_dfs(adev, vm, NULL, false); + struct amdgpu_vm_pt_cursor cursor; + struct amdgpu_vm_bo_base *entry; + + for_each_amdgpu_vm_pt_dfs_safe(adev, vm, NULL, cursor, entry) { + if (entry) + amdgpu_vm_pt_free(entry); + } } /** @@ -682,11 +706,15 @@ int amdgpu_vm_pde_update(struct amdgpu_vm_update_params *params, struct amdgpu_vm_bo_base *entry) { struct amdgpu_vm_bo_base *parent = amdgpu_vm_pt_parent(entry); - struct amdgpu_bo *bo = parent->bo, *pbo; + struct amdgpu_bo *bo, *pbo; struct amdgpu_vm *vm = params->vm; uint64_t pde, pt, flags; unsigned int level; + if (WARN_ON(!parent)) + return -EINVAL; + + bo = parent->bo; for (level = 0, pbo = bo->parent; pbo; ++level) pbo = pbo->parent; @@ -972,10 +1000,8 @@ int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params, while (cursor.pfn < frag_start) { /* Make sure previous mapping is freed */ if (cursor.entry->bo) { - params->table_freed = true; - amdgpu_vm_pt_free_dfs(adev, params->vm, - &cursor, - params->unlocked); + params->needs_flush = true; + amdgpu_vm_pt_add_list(params, &cursor); } amdgpu_vm_pt_next(adev, &cursor); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c index 349416e176..66e8a01612 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c @@ -126,6 +126,10 @@ static int amdgpu_vm_sdma_commit(struct amdgpu_vm_update_params *p, WARN_ON(ib->length_dw == 0); amdgpu_ring_pad_ib(ring, ib); + + if (p->needs_flush) + atomic64_inc(&p->vm->tlb_seq); + WARN_ON(ib->length_dw > p->num_dw_left); f = amdgpu_job_submit(p->job); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c new file mode 100644 index 0000000000..51cddfa3f1 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c @@ -0,0 +1,112 @@ +// SPDX-License-Identifier: GPL-2.0 OR MIT +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include <linux/dma-fence.h> +#include <linux/workqueue.h> + +#include "amdgpu.h" +#include "amdgpu_vm.h" +#include "amdgpu_gmc.h" + +struct amdgpu_tlb_fence { + struct dma_fence base; + struct amdgpu_device *adev; + struct dma_fence *dependency; + struct work_struct work; + spinlock_t lock; + uint16_t pasid; + +}; + +static const char *amdgpu_tlb_fence_get_driver_name(struct dma_fence *fence) +{ + return "amdgpu tlb fence"; +} + +static const char *amdgpu_tlb_fence_get_timeline_name(struct dma_fence *f) +{ + return "amdgpu tlb timeline"; +} + +static void amdgpu_tlb_fence_work(struct work_struct *work) +{ + struct amdgpu_tlb_fence *f = container_of(work, typeof(*f), work); + int r; + + if (f->dependency) { + dma_fence_wait(f->dependency, false); + dma_fence_put(f->dependency); + f->dependency = NULL; + } + + r = amdgpu_gmc_flush_gpu_tlb_pasid(f->adev, f->pasid, 2, true, 0); + if (r) { + dev_err(f->adev->dev, "TLB flush failed for PASID %d.\n", + f->pasid); + dma_fence_set_error(&f->base, r); + } + + dma_fence_signal(&f->base); + dma_fence_put(&f->base); +} + +static const struct dma_fence_ops amdgpu_tlb_fence_ops = { + .use_64bit_seqno = true, + .get_driver_name = amdgpu_tlb_fence_get_driver_name, + .get_timeline_name = amdgpu_tlb_fence_get_timeline_name +}; + +void amdgpu_vm_tlb_fence_create(struct amdgpu_device *adev, struct amdgpu_vm *vm, + struct dma_fence **fence) +{ + struct amdgpu_tlb_fence *f; + + f = kmalloc(sizeof(*f), GFP_KERNEL); + if (!f) { + /* + * We can't fail since the PDEs and PTEs are already updated, so + * just block for the dependency and execute the TLB flush + */ + if (*fence) + dma_fence_wait(*fence, false); + + amdgpu_gmc_flush_gpu_tlb_pasid(adev, vm->pasid, 2, true, 0); + *fence = dma_fence_get_stub(); + return; + } + + f->adev = adev; + f->dependency = *fence; + f->pasid = vm->pasid; + INIT_WORK(&f->work, amdgpu_tlb_fence_work); + spin_lock_init(&f->lock); + + dma_fence_init(&f->base, &amdgpu_tlb_fence_ops, &f->lock, + vm->tlb_fence_context, atomic64_read(&vm->tlb_seq)); + + /* TODO: We probably need a separate wq here */ + dma_fence_get(&f->base); + schedule_work(&f->work); + + *fence = &f->base; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index 8db8802443..6c30eceec8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -450,6 +450,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, { struct amdgpu_vram_mgr *mgr = to_vram_mgr(man); struct amdgpu_device *adev = to_amdgpu_device(mgr); + struct amdgpu_bo *bo = ttm_to_amdgpu_bo(tbo); u64 vis_usage = 0, max_bytes, min_block_size; struct amdgpu_vram_mgr_resource *vres; u64 size, remaining_size, lpfn, fpfn; @@ -468,7 +469,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, if (tbo->type != ttm_bo_type_kernel) max_bytes -= AMDGPU_VM_RESERVED_VRAM; - if (place->flags & TTM_PL_FLAG_CONTIGUOUS) { + if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS) { pages_per_block = ~0ul; } else { #ifdef CONFIG_TRANSPARENT_HUGEPAGE @@ -477,7 +478,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, /* default to 2MB */ pages_per_block = 2UL << (20UL - PAGE_SHIFT); #endif - pages_per_block = max_t(uint32_t, pages_per_block, + pages_per_block = max_t(u32, pages_per_block, tbo->page_alignment); } @@ -498,9 +499,12 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, if (place->flags & TTM_PL_FLAG_TOPDOWN) vres->flags |= DRM_BUDDY_TOPDOWN_ALLOCATION; - if (place->flags & TTM_PL_FLAG_CONTIGUOUS) + if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS) vres->flags |= DRM_BUDDY_CONTIGUOUS_ALLOCATION; + if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CLEARED) + vres->flags |= DRM_BUDDY_CLEAR_ALLOCATION; + if (fpfn || lpfn != mgr->mm.size) /* Allocate blocks in desired range */ vres->flags |= DRM_BUDDY_RANGE_ALLOCATION; @@ -514,21 +518,31 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, else min_block_size = mgr->default_page_size; - BUG_ON(min_block_size < mm->chunk_size); - /* Limit maximum size to 2GiB due to SG table limitations */ size = min(remaining_size, 2ULL << 30); if ((size >= (u64)pages_per_block << PAGE_SHIFT) && - !(size & (((u64)pages_per_block << PAGE_SHIFT) - 1))) + !(size & (((u64)pages_per_block << PAGE_SHIFT) - 1))) min_block_size = (u64)pages_per_block << PAGE_SHIFT; + BUG_ON(min_block_size < mm->chunk_size); + r = drm_buddy_alloc_blocks(mm, fpfn, lpfn, size, min_block_size, &vres->blocks, vres->flags); + + if (unlikely(r == -ENOSPC) && pages_per_block == ~0ul && + !(place->flags & TTM_PL_FLAG_CONTIGUOUS)) { + vres->flags &= ~DRM_BUDDY_CONTIGUOUS_ALLOCATION; + pages_per_block = max_t(u32, 2UL << (20UL - PAGE_SHIFT), + tbo->page_alignment); + + continue; + } + if (unlikely(r)) goto error_free_blocks; @@ -571,7 +585,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, return 0; error_free_blocks: - drm_buddy_free_list(mm, &vres->blocks); + drm_buddy_free_list(mm, &vres->blocks, 0); mutex_unlock(&mgr->lock); error_fini: ttm_resource_fini(man, &vres->base); @@ -604,7 +618,7 @@ static void amdgpu_vram_mgr_del(struct ttm_resource_manager *man, amdgpu_vram_mgr_do_reserve(man); - drm_buddy_free_list(mm, &vres->blocks); + drm_buddy_free_list(mm, &vres->blocks, vres->flags); mutex_unlock(&mgr->lock); atomic64_sub(vis_usage, &mgr->vis_usage); @@ -912,7 +926,7 @@ void amdgpu_vram_mgr_fini(struct amdgpu_device *adev) kfree(rsv); list_for_each_entry_safe(rsv, temp, &mgr->reserved_pages, blocks) { - drm_buddy_free_list(&mgr->mm, &rsv->allocated); + drm_buddy_free_list(&mgr->mm, &rsv->allocated, 0); kfree(rsv); } if (!adev->gmc.is_app_apu) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h index 0e04e42cf8..b256cbc2bc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h @@ -53,10 +53,20 @@ static inline u64 amdgpu_vram_mgr_block_size(struct drm_buddy_block *block) return (u64)PAGE_SIZE << drm_buddy_block_order(block); } +static inline bool amdgpu_vram_mgr_is_cleared(struct drm_buddy_block *block) +{ + return drm_buddy_block_is_clear(block); +} + static inline struct amdgpu_vram_mgr_resource * to_amdgpu_vram_mgr_resource(struct ttm_resource *res) { return container_of(res, struct amdgpu_vram_mgr_resource, base); } +static inline void amdgpu_vram_mgr_set_cleared(struct ttm_resource *res) +{ + to_amdgpu_vram_mgr_resource(res)->flags |= DRM_BUDDY_CLEARED; +} + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 20d51f6c9b..dd2ec48cf5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -1035,15 +1035,16 @@ int amdgpu_xgmi_remove_device(struct amdgpu_device *adev) return 0; } -static int xgmi_v6_4_0_aca_bank_generate_report(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type, - struct aca_bank_report *report, void *data) +static int xgmi_v6_4_0_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) { struct amdgpu_device *adev = handle->adev; + struct aca_bank_info info; const char *error_str; - u64 status; + u64 status, count; int ret, ext_error_code; - ret = aca_bank_info_decode(bank, &report->info); + ret = aca_bank_info_decode(bank, &info); if (ret) return ret; @@ -1055,15 +1056,28 @@ static int xgmi_v6_4_0_aca_bank_generate_report(struct aca_handle *handle, struc if (error_str) dev_info(adev->dev, "%s detected\n", error_str); - if ((type == ACA_ERROR_TYPE_UE && ext_error_code == 0) || - (type == ACA_ERROR_TYPE_CE && ext_error_code == 6)) - report->count[type] = ACA_REG__MISC0__ERRCNT(bank->regs[ACA_REG_IDX_MISC0]); + count = ACA_REG__MISC0__ERRCNT(bank->regs[ACA_REG_IDX_MISC0]); - return 0; + switch (type) { + case ACA_SMU_TYPE_UE: + if (ext_error_code != 0 && ext_error_code != 9) + count = 0ULL; + + ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE, count); + break; + case ACA_SMU_TYPE_CE: + count = ext_error_code == 6 ? count : 0ULL; + ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_CE, count); + break; + default: + return -EINVAL; + } + + return ret; } static const struct aca_bank_ops xgmi_v6_4_0_aca_bank_ops = { - .aca_bank_generate_report = xgmi_v6_4_0_aca_bank_generate_report, + .aca_bank_parser = xgmi_v6_4_0_aca_bank_parser, }; static const struct aca_info xgmi_v6_4_0_aca_info = { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h index 1592c63b30..a3bfc16de6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h @@ -44,6 +44,7 @@ struct amdgpu_hive_info { struct amdgpu_reset_domain *reset_domain; atomic_t ras_recovery; + struct ras_event_manager event_mgr; }; struct amdgpu_pcs_ras_field { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h index 51a14f6d93..fb2b394bb9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h @@ -94,7 +94,8 @@ union amd_sriov_msg_feature_flags { uint32_t reg_indirect_acc : 1; uint32_t av1_support : 1; uint32_t vcn_rb_decouple : 1; - uint32_t reserved : 24; + uint32_t mes_info_enable : 1; + uint32_t reserved : 23; } flags; uint32_t all; }; @@ -157,7 +158,7 @@ struct amd_sriov_msg_pf2vf_info_header { uint32_t reserved[2]; }; -#define AMD_SRIOV_MSG_PF2VF_INFO_FILLED_SIZE (48) +#define AMD_SRIOV_MSG_PF2VF_INFO_FILLED_SIZE (49) struct amd_sriov_msg_pf2vf_info { /* header contains size and version */ struct amd_sriov_msg_pf2vf_info_header header; @@ -208,6 +209,8 @@ struct amd_sriov_msg_pf2vf_info { struct amd_sriov_msg_uuid_info uuid_info; /* PCIE atomic ops support flag */ uint32_t pcie_atomic_ops_support_flags; + /* Portion of GPU memory occupied by VF. MAX value is 65535, but set to uint32_t to maintain alignment with reserved size */ + uint32_t gpu_capacity; /* reserved */ uint32_t reserved[256 - AMD_SRIOV_MSG_PF2VF_INFO_FILLED_SIZE]; }; @@ -221,7 +224,7 @@ struct amd_sriov_msg_vf2pf_info_header { uint32_t reserved[2]; }; -#define AMD_SRIOV_MSG_VF2PF_INFO_FILLED_SIZE (70) +#define AMD_SRIOV_MSG_VF2PF_INFO_FILLED_SIZE (73) struct amd_sriov_msg_vf2pf_info { /* header contains size and version */ struct amd_sriov_msg_vf2pf_info_header header; @@ -265,7 +268,9 @@ struct amd_sriov_msg_vf2pf_info { uint32_t version; } ucode_info[AMD_SRIOV_MSG_RESERVE_UCODE]; uint64_t dummy_page_addr; - + /* FB allocated for guest MES to record UQ info */ + uint64_t mes_info_addr; + uint32_t mes_info_size; /* reserved */ uint32_t reserved[256 - AMD_SRIOV_MSG_VF2PF_INFO_FILLED_SIZE]; }; diff --git a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c index fbb43ae762..d4e2aed2ef 100644 --- a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c +++ b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c @@ -422,7 +422,7 @@ __aqua_vanjaram_get_auto_mode(struct amdgpu_xcp_mgr *xcp_mgr) if (adev->gmc.num_mem_partitions == num_xcc / 2) return (adev->flags & AMD_IS_APU) ? AMDGPU_TPX_PARTITION_MODE : - AMDGPU_QPX_PARTITION_MODE; + AMDGPU_CPX_PARTITION_MODE; if (adev->gmc.num_mem_partitions == 2 && !(adev->flags & AMD_IS_APU)) return AMDGPU_DPX_PARTITION_MODE; @@ -630,7 +630,7 @@ static int aqua_vanjaram_xcp_mgr_init(struct amdgpu_device *adev) int aqua_vanjaram_init_soc_config(struct amdgpu_device *adev) { - u32 mask, inst_mask = adev->sdma.sdma_mask; + u32 mask, avail_inst, inst_mask = adev->sdma.sdma_mask; int ret, i; /* generally 1 AID supports 4 instances */ @@ -642,7 +642,9 @@ int aqua_vanjaram_init_soc_config(struct amdgpu_device *adev) for (mask = (1 << adev->sdma.num_inst_per_aid) - 1; inst_mask; inst_mask >>= adev->sdma.num_inst_per_aid, ++i) { - if ((inst_mask & mask) == mask) + avail_inst = inst_mask & mask; + if (avail_inst == mask || avail_inst == 0x3 || + avail_inst == 0xc) adev->aid_mask |= (1 << i); } diff --git a/drivers/gpu/drm/amd/amdgpu/atom.c b/drivers/gpu/drm/amd/amdgpu/atom.c index 72362df352..d552e01335 100644 --- a/drivers/gpu/drm/amd/amdgpu/atom.c +++ b/drivers/gpu/drm/amd/amdgpu/atom.c @@ -1243,6 +1243,7 @@ static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, ectx.ps_size = params_size; ectx.abort = false; ectx.last_jump = 0; + ectx.last_jump_jiffies = 0; if (ws) { ectx.ws = kcalloc(4, ws, GFP_KERNEL); ectx.ws_size = ws; diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index a3a643254d..cf1d5d462b 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik.c +++ b/drivers/gpu/drm/amd/amdgpu/cik.c @@ -1375,14 +1375,14 @@ static int cik_asic_pci_config_reset(struct amdgpu_device *adev) return r; } -static bool cik_asic_supports_baco(struct amdgpu_device *adev) +static int cik_asic_supports_baco(struct amdgpu_device *adev) { switch (adev->asic_type) { case CHIP_BONAIRE: case CHIP_HAWAII: return amdgpu_dpm_is_baco_supported(adev); default: - return false; + return 0; } } @@ -2210,6 +2210,8 @@ static const struct amd_ip_funcs cik_common_ip_funcs = { .soft_reset = cik_common_soft_reset, .set_clockgating_state = cik_common_set_clockgating_state, .set_powergating_state = cik_common_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ip_block_version cik_common_ip_block = diff --git a/drivers/gpu/drm/amd/amdgpu/cik_ih.c b/drivers/gpu/drm/amd/amdgpu/cik_ih.c index f24e34dc33..576baa9dbb 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/cik_ih.c @@ -435,6 +435,8 @@ static const struct amd_ip_funcs cik_ih_ip_funcs = { .soft_reset = cik_ih_soft_reset, .set_clockgating_state = cik_ih_set_clockgating_state, .set_powergating_state = cik_ih_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ih_funcs cik_ih_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c index a3fccc4c1f..6948ebda0f 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c @@ -1228,6 +1228,8 @@ static const struct amd_ip_funcs cik_sdma_ip_funcs = { .soft_reset = cik_sdma_soft_reset, .set_clockgating_state = cik_sdma_set_clockgating_state, .set_powergating_state = cik_sdma_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = { @@ -1290,7 +1292,7 @@ static void cik_sdma_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: is this a secure operation + * @copy_flags: unused * * Copy GPU buffers using the DMA engine (CIK). * Used by the amdgpu ttm implementation to move pages if @@ -1300,7 +1302,7 @@ static void cik_sdma_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY, SDMA_COPY_SUB_OPCODE_LINEAR, 0); ib->ptr[ib->length_dw++] = byte_count; diff --git a/drivers/gpu/drm/amd/amdgpu/cz_ih.c b/drivers/gpu/drm/amd/amdgpu/cz_ih.c index c19681492e..0726437873 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/cz_ih.c @@ -433,6 +433,8 @@ static const struct amd_ip_funcs cz_ih_ip_funcs = { .soft_reset = cz_ih_soft_reset, .set_clockgating_state = cz_ih_set_clockgating_state, .set_powergating_state = cz_ih_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ih_funcs cz_ih_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c index 221af054d8..b44fce44c0 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c @@ -3333,6 +3333,8 @@ static const struct amd_ip_funcs dce_v10_0_ip_funcs = { .soft_reset = dce_v10_0_soft_reset, .set_clockgating_state = dce_v10_0_set_clockgating_state, .set_powergating_state = dce_v10_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static void diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c index 69e8b0db6c..80b2e7f79a 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c @@ -3464,6 +3464,8 @@ static const struct amd_ip_funcs dce_v11_0_ip_funcs = { .soft_reset = dce_v11_0_soft_reset, .set_clockgating_state = dce_v11_0_set_clockgating_state, .set_powergating_state = dce_v11_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static void diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c index 60d40201fd..db20012600 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c @@ -3154,6 +3154,8 @@ static const struct amd_ip_funcs dce_v6_0_ip_funcs = { .soft_reset = dce_v6_0_soft_reset, .set_clockgating_state = dce_v6_0_set_clockgating_state, .set_powergating_state = dce_v6_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static void diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c index 5a5fcc45e4..5b56100ec9 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c @@ -3242,6 +3242,8 @@ static const struct amd_ip_funcs dce_v8_0_ip_funcs = { .soft_reset = dce_v8_0_soft_reset, .set_clockgating_state = dce_v8_0_set_clockgating_state, .set_powergating_state = dce_v8_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static void diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 701146d649..536287ddd2 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -276,6 +276,99 @@ MODULE_FIRMWARE("amdgpu/gc_10_3_7_mec.bin"); MODULE_FIRMWARE("amdgpu/gc_10_3_7_mec2.bin"); MODULE_FIRMWARE("amdgpu/gc_10_3_7_rlc.bin"); +static const struct amdgpu_hwip_reg_entry gc_reg_list_10_1[] = { + SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS2), + SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS3), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_STALLED_STAT1), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_STALLED_STAT2), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_STALLED_STAT1), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_STALLED_STAT1), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_BUSY_STAT), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_BUSY_STAT), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_BUSY_STAT), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_BUSY_STAT2), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_BUSY_STAT2), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPF_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_ERROR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_HPD_STATUS0), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_BASE), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_RPTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_WPTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB0_BASE), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB0_RPTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB0_WPTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB1_BASE), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB1_RPTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB1_WPTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB2_BASE), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB2_WPTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB2_WPTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_CMD_BUFSZ), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_CMD_BUFSZ), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_CMD_BUFSZ), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_CMD_BUFSZ), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_BASE_LO), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_BASE_HI), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB1_BUFSZ), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_BASE_LO), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_BASE_HI), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_IB2_BUFSZ), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_BASE_LO), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_BASE_HI), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB1_BUFSZ), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_BASE_LO), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_BASE_HI), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_IB2_BUFSZ), + SOC15_REG_ENTRY_STR(GC, 0, mmCPF_UTCL1_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmCPC_UTCL1_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmCPG_UTCL1_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmGDS_PROTECTION_FAULT), + SOC15_REG_ENTRY_STR(GC, 0, mmGDS_VM_PROTECTION_FAULT), + SOC15_REG_ENTRY_STR(GC, 0, mmIA_UTCL1_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmIA_UTCL1_STATUS_2), + SOC15_REG_ENTRY_STR(GC, 0, mmPA_CL_CNTL_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_UTCL1_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmRMI_UTCL1_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmSQC_DCACHE_UTCL0_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmSQC_ICACHE_UTCL0_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmSQG_UTCL0_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmTCP_UTCL0_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmWD_UTCL1_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL), + SOC15_REG_ENTRY_STR(GC, 0, mmGCVM_L2_PROTECTION_FAULT_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_DEBUG), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_CNTL), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_MES_CNTL), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_INSTR_PNTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC1_INSTR_PNTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC2_INSTR_PNTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_MES_DEBUG_INTERRUPT_INSTR_PNTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_MES_INSTR_PNTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_INSTR_PNTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_INSTR_PNTR), + SOC15_REG_ENTRY_STR(GC, 0, mmCP_CPC_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_STAT), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_COMMAND), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_MESSAGE), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_1), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_2), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_3), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_ARGUMENT_4), + SOC15_REG_ENTRY_STR(GC, 0, mmSMU_RLC_RESPONSE), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SAFE_MODE), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_SAFE_MODE), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_RLCS_GPM_STAT_2), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SPP_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_RLCS_BOOTLOAD_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_INT_STAT), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_GENERAL_6), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_DEBUG_INST_A), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_DEBUG_INST_B), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_DEBUG_INST_ADDR), + SOC15_REG_ENTRY_STR(GC, 0, mmRLC_LX6_CORE_PDEBUG_INST) +}; + static const struct soc15_reg_golden golden_settings_gc_10_1[] = { SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_4, 0xffffffff, 0x00400014), SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_CPF_CLK_CTRL, 0xfcff8fff, 0xf8000100), @@ -3964,7 +4057,7 @@ static void gfx_v10_0_check_gfxoff_flag(struct amdgpu_device *adev) static int gfx_v10_0_init_microcode(struct amdgpu_device *adev) { - char fw_name[40]; + char fw_name[53]; char ucode_prefix[30]; const char *wks = ""; int err; @@ -4490,6 +4583,22 @@ static int gfx_v10_0_compute_ring_init(struct amdgpu_device *adev, int ring_id, hw_prio, NULL); } +static void gfx_v10_0_alloc_dump_mem(struct amdgpu_device *adev) +{ + uint32_t reg_count = ARRAY_SIZE(gc_reg_list_10_1); + uint32_t *ptr; + + ptr = kcalloc(reg_count, sizeof(uint32_t), GFP_KERNEL); + if (ptr == NULL) { + DRM_ERROR("Failed to allocate memory for IP Dump\n"); + adev->gfx.ip_dump = NULL; + adev->gfx.reg_count = 0; + } else { + adev->gfx.ip_dump = ptr; + adev->gfx.reg_count = reg_count; + } +} + static int gfx_v10_0_sw_init(void *handle) { int i, j, k, r, ring_id = 0; @@ -4518,7 +4627,7 @@ static int gfx_v10_0_sw_init(void *handle) case IP_VERSION(10, 3, 3): case IP_VERSION(10, 3, 7): adev->gfx.me.num_me = 1; - adev->gfx.me.num_pipe_per_me = 1; + adev->gfx.me.num_pipe_per_me = 2; adev->gfx.me.num_queue_per_pipe = 1; adev->gfx.mec.num_mec = 2; adev->gfx.mec.num_pipe_per_mec = 4; @@ -4642,6 +4751,8 @@ static int gfx_v10_0_sw_init(void *handle) gfx_v10_0_gpu_early_init(adev); + gfx_v10_0_alloc_dump_mem(adev); + return 0; } @@ -4694,6 +4805,8 @@ static int gfx_v10_0_sw_fini(void *handle) gfx_v10_0_free_microcode(adev); + kfree(adev->gfx.ip_dump); + return 0; } @@ -8317,7 +8430,7 @@ static void gfx_v10_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) } reg_mem_engine = 0; } else { - ref_and_mask = nbio_hf_reg->ref_and_mask_cp0; + ref_and_mask = nbio_hf_reg->ref_and_mask_cp0 << ring->pipe; reg_mem_engine = 1; /* pfp */ } @@ -9154,6 +9267,36 @@ static void gfx_v10_0_emit_mem_sync(struct amdgpu_ring *ring) amdgpu_ring_write(ring, gcr_cntl); /* GCR_CNTL */ } +static void gfx_v10_ip_print(void *handle, struct drm_printer *p) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + uint32_t i; + uint32_t reg_count = ARRAY_SIZE(gc_reg_list_10_1); + + if (!adev->gfx.ip_dump) + return; + + for (i = 0; i < reg_count; i++) + drm_printf(p, "%-50s \t 0x%08x\n", + gc_reg_list_10_1[i].reg_name, + adev->gfx.ip_dump[i]); +} + +static void gfx_v10_ip_dump(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + uint32_t i; + uint32_t reg_count = ARRAY_SIZE(gc_reg_list_10_1); + + if (!adev->gfx.ip_dump) + return; + + amdgpu_gfx_off_ctrl(adev, false); + for (i = 0; i < reg_count; i++) + adev->gfx.ip_dump[i] = RREG32(SOC15_REG_ENTRY_OFFSET(gc_reg_list_10_1[i])); + amdgpu_gfx_off_ctrl(adev, true); +} + static const struct amd_ip_funcs gfx_v10_0_ip_funcs = { .name = "gfx_v10_0", .early_init = gfx_v10_0_early_init, @@ -9170,6 +9313,8 @@ static const struct amd_ip_funcs gfx_v10_0_ip_funcs = { .set_clockgating_state = gfx_v10_0_set_clockgating_state, .set_powergating_state = gfx_v10_0_set_powergating_state, .get_clockgating_state = gfx_v10_0_get_clockgating_state, + .dump_ip_state = gfx_v10_ip_dump, + .print_ip_state = gfx_v10_ip_print, }; static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_gfx = { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index f00e05aba4..ad6431013c 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c @@ -510,7 +510,7 @@ static void gfx_v11_0_check_fw_cp_gfx_shadow(struct amdgpu_device *adev) static int gfx_v11_0_init_microcode(struct amdgpu_device *adev) { char fw_name[40]; - char ucode_prefix[30]; + char ucode_prefix[25]; int err; const struct rlc_firmware_header_v2_0 *rlc_hdr; uint16_t version_major; @@ -4506,14 +4506,11 @@ static int gfx_v11_0_soft_reset(void *handle) gfx_v11_0_set_safe_mode(adev, 0); + mutex_lock(&adev->srbm_mutex); for (i = 0; i < adev->gfx.mec.num_mec; ++i) { for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) { for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) { - tmp = RREG32_SOC15(GC, 0, regGRBM_GFX_CNTL); - tmp = REG_SET_FIELD(tmp, GRBM_GFX_CNTL, MEID, i); - tmp = REG_SET_FIELD(tmp, GRBM_GFX_CNTL, QUEUEID, j); - tmp = REG_SET_FIELD(tmp, GRBM_GFX_CNTL, PIPEID, k); - WREG32_SOC15(GC, 0, regGRBM_GFX_CNTL, tmp); + soc21_grbm_select(adev, i, k, j, 0); WREG32_SOC15(GC, 0, regCP_HQD_DEQUEUE_REQUEST, 0x2); WREG32_SOC15(GC, 0, regSPI_COMPUTE_QUEUE_RESET, 0x1); @@ -4523,16 +4520,14 @@ static int gfx_v11_0_soft_reset(void *handle) for (i = 0; i < adev->gfx.me.num_me; ++i) { for (j = 0; j < adev->gfx.me.num_queue_per_pipe; j++) { for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) { - tmp = RREG32_SOC15(GC, 0, regGRBM_GFX_CNTL); - tmp = REG_SET_FIELD(tmp, GRBM_GFX_CNTL, MEID, i); - tmp = REG_SET_FIELD(tmp, GRBM_GFX_CNTL, QUEUEID, j); - tmp = REG_SET_FIELD(tmp, GRBM_GFX_CNTL, PIPEID, k); - WREG32_SOC15(GC, 0, regGRBM_GFX_CNTL, tmp); + soc21_grbm_select(adev, i, k, j, 0); WREG32_SOC15(GC, 0, regCP_GFX_HQD_DEQUEUE_REQUEST, 0x1); } } } + soc21_grbm_select(adev, 0, 0, 0, 0); + mutex_unlock(&adev->srbm_mutex); /* Try to acquire the gfx mutex before access to CP_VMID_RESET */ r = gfx_v11_0_request_gfx_index_mutex(adev, 1); @@ -6174,6 +6169,8 @@ static const struct amd_ip_funcs gfx_v11_0_ip_funcs = { .set_clockgating_state = gfx_v11_0_set_clockgating_state, .set_powergating_state = gfx_v11_0_set_powergating_state, .get_clockgating_state = gfx_v11_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_gfx = { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c index 34f9211b26..d0992ce9fb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c @@ -3457,6 +3457,8 @@ static const struct amd_ip_funcs gfx_v6_0_ip_funcs = { .soft_reset = gfx_v6_0_soft_reset, .set_clockgating_state = gfx_v6_0_set_clockgating_state, .set_powergating_state = gfx_v6_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_gfx = { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index 86a4865b1a..541dbd70d8 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -4977,6 +4977,8 @@ static const struct amd_ip_funcs gfx_v7_0_ip_funcs = { .soft_reset = gfx_v7_0_soft_reset, .set_clockgating_state = gfx_v7_0_set_clockgating_state, .set_powergating_state = gfx_v7_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 202ddda57f..2f0e72caee 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -6878,6 +6878,8 @@ static const struct amd_ip_funcs gfx_v8_0_ip_funcs = { .set_clockgating_state = gfx_v8_0_set_clockgating_state, .set_powergating_state = gfx_v8_0_set_powergating_state, .get_clockgating_state = gfx_v8_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 99dbd23411..3c8c5abf35 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -1249,7 +1249,7 @@ static void gfx_v9_0_check_if_need_gfxoff(struct amdgpu_device *adev) static int gfx_v9_0_init_cp_gfx_microcode(struct amdgpu_device *adev, char *chip_name) { - char fw_name[30]; + char fw_name[50]; int err; snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name); @@ -1282,7 +1282,7 @@ out: static int gfx_v9_0_init_rlc_microcode(struct amdgpu_device *adev, char *chip_name) { - char fw_name[30]; + char fw_name[53]; int err; const struct rlc_firmware_header_v2_0 *rlc_hdr; uint16_t version_major; @@ -1337,7 +1337,7 @@ static bool gfx_v9_0_load_mec2_fw_bin_support(struct amdgpu_device *adev) static int gfx_v9_0_init_cp_compute_microcode(struct amdgpu_device *adev, char *chip_name) { - char fw_name[30]; + char fw_name[50]; int err; if (amdgpu_sriov_vf(adev) && (adev->asic_type == CHIP_ALDEBARAN)) @@ -6856,6 +6856,8 @@ static const struct amd_ip_funcs gfx_v9_0_ip_funcs = { .set_clockgating_state = gfx_v9_0_set_clockgating_state, .set_powergating_state = gfx_v9_0_set_powergating_state, .get_clockgating_state = gfx_v9_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c index 065b2bd5f5..3f4fd2f081 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c @@ -1909,18 +1909,7 @@ static void gfx_v9_4_2_reset_sq_timeout_status(struct amdgpu_device *adev) mutex_unlock(&adev->grbm_idx_mutex); } -static bool gfx_v9_4_2_query_uctl2_poison_status(struct amdgpu_device *adev) -{ - u32 status = 0; - struct amdgpu_vmhub *hub; - - hub = &adev->vmhub[AMDGPU_GFXHUB(0)]; - status = RREG32(hub->vm_l2_pro_fault_status); - /* reset page fault status */ - WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1); - return REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, FED); -} struct amdgpu_ras_block_hw_ops gfx_v9_4_2_ras_ops = { .query_ras_error_count = &gfx_v9_4_2_query_ras_error_count, @@ -1934,5 +1923,4 @@ struct amdgpu_gfx_ras gfx_v9_4_2_ras = { .hw_ops = &gfx_v9_4_2_ras_ops, }, .enable_watchdog_timer = &gfx_v9_4_2_enable_watchdog_timer, - .query_utcl2_poison_status = gfx_v9_4_2_query_uctl2_poison_status, }; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c index b10fdd8b54..f5b9f443cf 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c @@ -680,38 +680,44 @@ static const struct amdgpu_gfx_funcs gfx_v9_4_3_gfx_funcs = { .ih_node_to_logical_xcc = &gfx_v9_4_3_ih_to_xcc_inst, }; -static int gfx_v9_4_3_aca_bank_generate_report(struct aca_handle *handle, - struct aca_bank *bank, enum aca_error_type type, - struct aca_bank_report *report, void *data) +static int gfx_v9_4_3_aca_bank_parser(struct aca_handle *handle, + struct aca_bank *bank, enum aca_smu_type type, + void *data) { - u64 status, misc0; + struct aca_bank_info info; + u64 misc0; u32 instlo; int ret; - status = bank->regs[ACA_REG_IDX_STATUS]; - if ((type == ACA_ERROR_TYPE_UE && - ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_FAULT) || - (type == ACA_ERROR_TYPE_CE && - ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_CE)) { + ret = aca_bank_info_decode(bank, &info); + if (ret) + return ret; - ret = aca_bank_info_decode(bank, &report->info); - if (ret) - return ret; + /* NOTE: overwrite info.die_id with xcd id for gfx */ + instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]); + instlo &= GENMASK(31, 1); + info.die_id = instlo == mmSMNAID_XCD0_MCA_SMU ? 0 : 1; - /* NOTE: overwrite info.die_id with xcd id for gfx */ - instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]); - instlo &= GENMASK(31, 1); - report->info.die_id = instlo == mmSMNAID_XCD0_MCA_SMU ? 0 : 1; + misc0 = bank->regs[ACA_REG_IDX_MISC0]; - misc0 = bank->regs[ACA_REG_IDX_MISC0]; - report->count[type] = ACA_REG__MISC0__ERRCNT(misc0); + switch (type) { + case ACA_SMU_TYPE_UE: + ret = aca_error_cache_log_bank_error(handle, &info, + ACA_ERROR_TYPE_UE, 1ULL); + break; + case ACA_SMU_TYPE_CE: + ret = aca_error_cache_log_bank_error(handle, &info, + ACA_ERROR_TYPE_CE, ACA_REG__MISC0__ERRCNT(misc0)); + break; + default: + return -EINVAL; } - return 0; + return ret; } static bool gfx_v9_4_3_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, - enum aca_error_type type, void *data) + enum aca_smu_type type, void *data) { u32 instlo; @@ -730,7 +736,7 @@ static bool gfx_v9_4_3_aca_bank_is_valid(struct aca_handle *handle, struct aca_b } static const struct aca_bank_ops gfx_v9_4_3_aca_bank_ops = { - .aca_bank_generate_report = gfx_v9_4_3_aca_bank_generate_report, + .aca_bank_parser = gfx_v9_4_3_aca_bank_parser, .aca_bank_is_valid = gfx_v9_4_3_aca_bank_is_valid, }; @@ -2398,10 +2404,10 @@ gfx_v9_4_3_xcc_update_coarse_grain_clock_gating(struct amdgpu_device *adev, if (def != data) WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGTT_MGCG_OVERRIDE, data); - /* enable cgcg FSM(0x0000363F) */ + /* CGCG Hysteresis: 400us */ def = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGCG_CGLS_CTRL); - data = (0x36 + data = (0x2710 << RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) | RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK; if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS) @@ -2410,10 +2416,10 @@ gfx_v9_4_3_xcc_update_coarse_grain_clock_gating(struct amdgpu_device *adev, if (def != data) WREG32_SOC15(GC, GET_INST(GC, xcc_id), regRLC_CGCG_CGLS_CTRL, data); - /* set IDLE_POLL_COUNT(0x00900100) */ + /* set IDLE_POLL_COUNT(0x33450100)*/ def = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_RB_WPTR_POLL_CNTL); data = (0x0100 << CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY__SHIFT) | - (0x0090 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT); + (0x3345 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT); if (def != data) WREG32_SOC15(GC, GET_INST(GC, xcc_id), regCP_RB_WPTR_POLL_CNTL, data); } else { @@ -4010,6 +4016,8 @@ static const struct amd_ip_funcs gfx_v9_4_3_ip_funcs = { .set_clockgating_state = gfx_v9_4_3_set_clockgating_state, .set_powergating_state = gfx_v9_4_3_set_powergating_state, .get_clockgating_state = gfx_v9_4_3_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs gfx_v9_4_3_ring_funcs_compute = { diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c index 22175da0e1..d200310d17 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c @@ -443,6 +443,22 @@ static void gfxhub_v1_0_init(struct amdgpu_device *adev) mmVM_INVALIDATE_ENG0_ADDR_RANGE_LO32; } +static bool gfxhub_v1_0_query_utcl2_poison_status(struct amdgpu_device *adev, + int xcc_id) +{ + u32 status = 0; + struct amdgpu_vmhub *hub; + + if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 2)) + return false; + + hub = &adev->vmhub[AMDGPU_GFXHUB(0)]; + status = RREG32(hub->vm_l2_pro_fault_status); + /* reset page fault status */ + WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1); + + return REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, FED); +} const struct amdgpu_gfxhub_funcs gfxhub_v1_0_funcs = { .get_mc_fb_offset = gfxhub_v1_0_get_mc_fb_offset, @@ -452,4 +468,5 @@ const struct amdgpu_gfxhub_funcs gfxhub_v1_0_funcs = { .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, + .query_utcl2_poison_status = gfxhub_v1_0_query_utcl2_poison_status, }; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c index 49aecdcee0..77df8c9cba 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c @@ -620,6 +620,20 @@ static int gfxhub_v1_2_get_xgmi_info(struct amdgpu_device *adev) return 0; } +static bool gfxhub_v1_2_query_utcl2_poison_status(struct amdgpu_device *adev, + int xcc_id) +{ + u32 fed, status; + + status = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regVM_L2_PROTECTION_FAULT_STATUS); + fed = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, FED); + /* reset page fault status */ + WREG32_P(SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id), + regVM_L2_PROTECTION_FAULT_STATUS), 1, ~1); + + return fed; +} + const struct amdgpu_gfxhub_funcs gfxhub_v1_2_funcs = { .get_mc_fb_offset = gfxhub_v1_2_get_mc_fb_offset, .setup_vm_pt_regs = gfxhub_v1_2_setup_vm_pt_regs, @@ -628,6 +642,7 @@ const struct amdgpu_gfxhub_funcs gfxhub_v1_2_funcs = { .set_fault_enable_default = gfxhub_v1_2_set_fault_enable_default, .init = gfxhub_v1_2_init, .get_xgmi_info = gfxhub_v1_2_get_xgmi_info, + .query_utcl2_poison_status = gfxhub_v1_2_query_utcl2_poison_status, }; static int gfxhub_v1_2_xcp_resume(void *handle, uint32_t inst_mask) diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c index 23b4786399..3e38d8bfcb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c @@ -1115,6 +1115,8 @@ static const struct amd_ip_funcs gmc_v6_0_ip_funcs = { .soft_reset = gmc_v6_0_soft_reset, .set_clockgating_state = gmc_v6_0_set_clockgating_state, .set_powergating_state = gmc_v6_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_gmc_funcs gmc_v6_0_gmc_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c index 3da7b6a2b0..85df8fc810 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c @@ -1354,6 +1354,8 @@ static const struct amd_ip_funcs gmc_v7_0_ip_funcs = { .soft_reset = gmc_v7_0_soft_reset, .set_clockgating_state = gmc_v7_0_set_clockgating_state, .set_powergating_state = gmc_v7_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_gmc_funcs gmc_v7_0_gmc_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index d20e5f20ee..fc97757e33 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -1717,6 +1717,8 @@ static const struct amd_ip_funcs gmc_v8_0_ip_funcs = { .set_clockgating_state = gmc_v8_0_set_clockgating_state, .set_powergating_state = gmc_v8_0_set_powergating_state, .get_clockgating_state = gmc_v8_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_gmc_funcs gmc_v8_0_gmc_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 47b63a4ce6..f7f4924751 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -548,7 +548,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, { bool retry_fault = !!(entry->src_data[1] & 0x80); bool write_fault = !!(entry->src_data[1] & 0x20); - uint32_t status = 0, cid = 0, rw = 0; + uint32_t status = 0, cid = 0, rw = 0, fed = 0; struct amdgpu_task_info *task_info; struct amdgpu_vmhub *hub; const char *mmhub_cid; @@ -664,6 +664,13 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, status = RREG32(hub->vm_l2_pro_fault_status); cid = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, CID); rw = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, RW); + fed = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, FED); + + /* for fed error, kfd will handle it, return directly */ + if (fed && amdgpu_ras_is_poison_mode_supported(adev) && + (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(9, 4, 2))) + return 0; + WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1); amdgpu_vm_update_fault_cache(adev, entry->pasid, addr, status, vmhub); @@ -1450,7 +1457,6 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.channel_offs = UMC_V12_0_PER_CHANNEL_OFFSET; adev->umc.active_mask = adev->aid_mask; adev->umc.retire_unit = UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL; - adev->umc.channel_idx_tbl = &umc_v12_0_channel_idx_tbl[0][0][0]; if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) adev->umc.ras = &umc_v12_0_ras; break; @@ -1904,7 +1910,7 @@ gmc_v9_0_init_sw_mem_ranges(struct amdgpu_device *adev, break; } - size = adev->gmc.real_vram_size >> AMDGPU_GPU_PAGE_SHIFT; + size = (adev->gmc.real_vram_size + SZ_16M) >> AMDGPU_GPU_PAGE_SHIFT; size /= adev->gmc.num_mem_partitions; for (i = 0; i < adev->gmc.num_mem_partitions; ++i) { diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c index 2c02ae6988..07984f7c3a 100644 --- a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c @@ -425,6 +425,8 @@ static const struct amd_ip_funcs iceland_ih_ip_funcs = { .soft_reset = iceland_ih_soft_reset, .set_clockgating_state = iceland_ih_set_clockgating_state, .set_powergating_state = iceland_ih_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ih_funcs iceland_ih_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c b/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c index ad4ad39f12..3cb64c8f71 100644 --- a/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c @@ -346,6 +346,21 @@ static int ih_v6_0_irq_init(struct amdgpu_device *adev) DELAY, 3); WREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL, tmp); + /* Redirect the interrupts to IH RB1 for dGPU */ + if (adev->irq.ih1.ring_size) { + tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX); + tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_INDEX, INDEX, 0); + WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX, tmp); + + tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA); + tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, CLIENT_ID, 0xa); + tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, SOURCE_ID, 0x0); + tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, + SOURCE_ID_MATCH_ENABLE, 0x1); + + WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA, tmp); + } + pci_set_master(adev->pdev); /* enable interrupts */ @@ -549,8 +564,15 @@ static int ih_v6_0_sw_init(void *handle) adev->irq.ih.use_doorbell = true; adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1; - adev->irq.ih1.ring_size = 0; - adev->irq.ih2.ring_size = 0; + if (!(adev->flags & AMD_IS_APU)) { + r = amdgpu_ih_ring_init(adev, &adev->irq.ih1, IH_RING_SIZE, + use_bus_addr); + if (r) + return r; + + adev->irq.ih1.use_doorbell = true; + adev->irq.ih1.doorbell_index = (adev->doorbell_index.ih + 1) << 1; + } /* initialize ih control register offset */ ih_v6_0_init_register_offset(adev); @@ -748,6 +770,8 @@ static const struct amd_ip_funcs ih_v6_0_ip_funcs = { .set_clockgating_state = ih_v6_0_set_clockgating_state, .set_powergating_state = ih_v6_0_set_powergating_state, .get_clockgating_state = ih_v6_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ih_funcs ih_v6_0_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c index b8da0fc293..0fbf5fa7b0 100644 --- a/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c +++ b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c @@ -346,6 +346,21 @@ static int ih_v6_1_irq_init(struct amdgpu_device *adev) DELAY, 3); WREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL, tmp); + /* Redirect the interrupts to IH RB1 for dGPU */ + if (adev->irq.ih1.ring_size) { + tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX); + tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_INDEX, INDEX, 0); + WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_INDEX, tmp); + + tmp = RREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA); + tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, CLIENT_ID, 0xa); + tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, SOURCE_ID, 0x0); + tmp = REG_SET_FIELD(tmp, IH_RING1_CLIENT_CFG_DATA, + SOURCE_ID_MATCH_ENABLE, 0x1); + + WREG32_SOC15(OSSSYS, 0, regIH_RING1_CLIENT_CFG_DATA, tmp); + } + pci_set_master(adev->pdev); /* enable interrupts */ @@ -550,8 +565,15 @@ static int ih_v6_1_sw_init(void *handle) adev->irq.ih.use_doorbell = true; adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1; - adev->irq.ih1.ring_size = 0; - adev->irq.ih2.ring_size = 0; + if (!(adev->flags & AMD_IS_APU)) { + r = amdgpu_ih_ring_init(adev, &adev->irq.ih1, IH_RING_SIZE, + use_bus_addr); + if (r) + return r; + + adev->irq.ih1.use_doorbell = true; + adev->irq.ih1.doorbell_index = (adev->doorbell_index.ih + 1) << 1; + } /* initialize ih control register offset */ ih_v6_1_init_register_offset(adev); @@ -753,6 +775,8 @@ static const struct amd_ip_funcs ih_v6_1_ip_funcs = { .set_clockgating_state = ih_v6_1_set_clockgating_state, .set_powergating_state = ih_v6_1_set_powergating_state, .get_clockgating_state = ih_v6_1_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ih_funcs ih_v6_1_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c index 7aed96fa10..aa6235dd4f 100644 --- a/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c @@ -749,6 +749,8 @@ static const struct amd_ip_funcs ih_v7_0_ip_funcs = { .set_clockgating_state = ih_v7_0_set_clockgating_state, .set_powergating_state = ih_v7_0_set_powergating_state, .get_clockgating_state = ih_v7_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ih_funcs ih_v7_0_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c index 1c8116d75f..ef3e42f6b8 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c @@ -759,6 +759,8 @@ static const struct amd_ip_funcs jpeg_v2_0_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = jpeg_v2_0_set_clockgating_state, .set_powergating_state = jpeg_v2_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs jpeg_v2_0_dec_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c index 99cd49ee8e..afeaf3c64e 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c @@ -632,6 +632,8 @@ static const struct amd_ip_funcs jpeg_v2_5_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = jpeg_v2_5_set_clockgating_state, .set_powergating_state = jpeg_v2_5_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amd_ip_funcs jpeg_v2_6_ip_funcs = { @@ -652,6 +654,8 @@ static const struct amd_ip_funcs jpeg_v2_6_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = jpeg_v2_5_set_clockgating_state, .set_powergating_state = jpeg_v2_5_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs jpeg_v2_5_dec_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c index a92481da60..1c7cf4800b 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c @@ -557,6 +557,8 @@ static const struct amd_ip_funcs jpeg_v3_0_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = jpeg_v3_0_set_clockgating_state, .set_powergating_state = jpeg_v3_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs jpeg_v3_0_dec_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c index 88ea58d5c4..237fe5df5a 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c @@ -719,6 +719,8 @@ static const struct amd_ip_funcs jpeg_v4_0_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = jpeg_v4_0_set_clockgating_state, .set_powergating_state = jpeg_v4_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs jpeg_v4_0_dec_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c index 32caeb37ce..d66af11aa6 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c @@ -1053,6 +1053,8 @@ static const struct amd_ip_funcs jpeg_v4_0_3_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = jpeg_v4_0_3_set_clockgating_state, .set_powergating_state = jpeg_v4_0_3_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs jpeg_v4_0_3_dec_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c index edf5bcdd2b..da6bb9022b 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c @@ -762,6 +762,8 @@ static const struct amd_ip_funcs jpeg_v4_0_5_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = jpeg_v4_0_5_set_clockgating_state, .set_powergating_state = jpeg_v4_0_5_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs jpeg_v4_0_5_dec_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c index e70200f975..64c856bfe0 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c @@ -513,6 +513,8 @@ static const struct amd_ip_funcs jpeg_v5_0_0_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = jpeg_v5_0_0_set_clockgating_state, .set_powergating_state = jpeg_v5_0_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs jpeg_v5_0_0_dec_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c index 1e5ad1e08d..a626bf9049 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c @@ -1176,6 +1176,8 @@ static const struct amd_ip_funcs mes_v10_1_ip_funcs = { .hw_fini = mes_v10_1_hw_fini, .suspend = mes_v10_1_suspend, .resume = mes_v10_1_resume, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version mes_v10_1_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c index 63f281a998..32d4519541 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c @@ -100,18 +100,76 @@ static const struct amdgpu_ring_funcs mes_v11_0_ring_funcs = { .insert_nop = amdgpu_ring_insert_nop, }; +static const char *mes_v11_0_opcodes[] = { + "SET_HW_RSRC", + "SET_SCHEDULING_CONFIG", + "ADD_QUEUE", + "REMOVE_QUEUE", + "PERFORM_YIELD", + "SET_GANG_PRIORITY_LEVEL", + "SUSPEND", + "RESUME", + "RESET", + "SET_LOG_BUFFER", + "CHANGE_GANG_PRORITY", + "QUERY_SCHEDULER_STATUS", + "PROGRAM_GDS", + "SET_DEBUG_VMID", + "MISC", + "UPDATE_ROOT_PAGE_TABLE", + "AMD_LOG", +}; + +static const char *mes_v11_0_misc_opcodes[] = { + "WRITE_REG", + "INV_GART", + "QUERY_STATUS", + "READ_REG", + "WAIT_REG_MEM", + "SET_SHADER_DEBUGGER", +}; + +static const char *mes_v11_0_get_op_string(union MESAPI__MISC *x_pkt) +{ + const char *op_str = NULL; + + if (x_pkt->header.opcode < ARRAY_SIZE(mes_v11_0_opcodes)) + op_str = mes_v11_0_opcodes[x_pkt->header.opcode]; + + return op_str; +} + +static const char *mes_v11_0_get_misc_op_string(union MESAPI__MISC *x_pkt) +{ + const char *op_str = NULL; + + if ((x_pkt->header.opcode == MES_SCH_API_MISC) && + (x_pkt->opcode < ARRAY_SIZE(mes_v11_0_misc_opcodes))) + op_str = mes_v11_0_misc_opcodes[x_pkt->opcode]; + + return op_str; +} + static int mes_v11_0_submit_pkt_and_poll_completion(struct amdgpu_mes *mes, void *pkt, int size, int api_status_off) { - int ndw = size / 4; - signed long r; - union MESAPI__ADD_QUEUE *x_pkt = pkt; - struct MES_API_STATUS *api_status; + union MESAPI__QUERY_MES_STATUS mes_status_pkt; + signed long timeout = 3000000; /* 3000 ms */ struct amdgpu_device *adev = mes->adev; struct amdgpu_ring *ring = &mes->ring; + struct MES_API_STATUS *api_status; + union MESAPI__MISC *x_pkt = pkt; + const char *op_str, *misc_op_str; unsigned long flags; - signed long timeout = adev->usec_timeout; + u64 status_gpu_addr; + u32 status_offset; + u64 *status_ptr; + signed long r; + int ret; + + if (x_pkt->header.opcode >= MES_SCH_API_MAX) + return -EINVAL; if (amdgpu_emu_mode) { timeout *= 100; @@ -119,37 +177,82 @@ static int mes_v11_0_submit_pkt_and_poll_completion(struct amdgpu_mes *mes, /* Worst case in sriov where all other 15 VF timeout, each VF needs about 600ms */ timeout = 15 * 600 * 1000; } - BUG_ON(size % 4 != 0); + + ret = amdgpu_device_wb_get(adev, &status_offset); + if (ret) + return ret; + + status_gpu_addr = adev->wb.gpu_addr + (status_offset * 4); + status_ptr = (u64 *)&adev->wb.wb[status_offset]; + *status_ptr = 0; spin_lock_irqsave(&mes->ring_lock, flags); - if (amdgpu_ring_alloc(ring, ndw)) { - spin_unlock_irqrestore(&mes->ring_lock, flags); - return -ENOMEM; - } + r = amdgpu_ring_alloc(ring, (size + sizeof(mes_status_pkt)) / 4); + if (r) + goto error_unlock_free; api_status = (struct MES_API_STATUS *)((char *)pkt + api_status_off); - api_status->api_completion_fence_addr = mes->ring.fence_drv.gpu_addr; - api_status->api_completion_fence_value = ++mes->ring.fence_drv.sync_seq; + api_status->api_completion_fence_addr = status_gpu_addr; + api_status->api_completion_fence_value = 1; + + amdgpu_ring_write_multiple(ring, pkt, size / 4); + + memset(&mes_status_pkt, 0, sizeof(mes_status_pkt)); + mes_status_pkt.header.type = MES_API_TYPE_SCHEDULER; + mes_status_pkt.header.opcode = MES_SCH_API_QUERY_SCHEDULER_STATUS; + mes_status_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS; + mes_status_pkt.api_status.api_completion_fence_addr = + ring->fence_drv.gpu_addr; + mes_status_pkt.api_status.api_completion_fence_value = + ++ring->fence_drv.sync_seq; + + amdgpu_ring_write_multiple(ring, &mes_status_pkt, + sizeof(mes_status_pkt) / 4); - amdgpu_ring_write_multiple(ring, pkt, ndw); amdgpu_ring_commit(ring); spin_unlock_irqrestore(&mes->ring_lock, flags); - DRM_DEBUG("MES msg=%d was emitted\n", x_pkt->header.opcode); + op_str = mes_v11_0_get_op_string(x_pkt); + misc_op_str = mes_v11_0_get_misc_op_string(x_pkt); - r = amdgpu_fence_wait_polling(ring, ring->fence_drv.sync_seq, - timeout); - if (r < 1) { - DRM_ERROR("MES failed to response msg=%d\n", - x_pkt->header.opcode); + if (misc_op_str) + dev_dbg(adev->dev, "MES msg=%s (%s) was emitted\n", op_str, + misc_op_str); + else if (op_str) + dev_dbg(adev->dev, "MES msg=%s was emitted\n", op_str); + else + dev_dbg(adev->dev, "MES msg=%d was emitted\n", + x_pkt->header.opcode); + + r = amdgpu_fence_wait_polling(ring, ring->fence_drv.sync_seq, timeout); + if (r < 1 || !*status_ptr) { + + if (misc_op_str) + dev_err(adev->dev, "MES failed to respond to msg=%s (%s)\n", + op_str, misc_op_str); + else if (op_str) + dev_err(adev->dev, "MES failed to respond to msg=%s\n", + op_str); + else + dev_err(adev->dev, "MES failed to respond to msg=%d\n", + x_pkt->header.opcode); while (halt_if_hws_hang) schedule(); - return -ETIMEDOUT; + r = -ETIMEDOUT; + goto error_wb_free; } + amdgpu_device_wb_free(adev, status_offset); return 0; + +error_unlock_free: + spin_unlock_irqrestore(&mes->ring_lock, flags); + +error_wb_free: + amdgpu_device_wb_free(adev, status_offset); + return r; } static int convert_to_mes_queue_type(int queue_type) @@ -422,6 +525,36 @@ static int mes_v11_0_set_hw_resources(struct amdgpu_mes *mes) offsetof(union MESAPI_SET_HW_RESOURCES, api_status)); } +static int mes_v11_0_set_hw_resources_1(struct amdgpu_mes *mes) +{ + int size = 128 * PAGE_SIZE; + int ret = 0; + struct amdgpu_device *adev = mes->adev; + union MESAPI_SET_HW_RESOURCES_1 mes_set_hw_res_pkt; + memset(&mes_set_hw_res_pkt, 0, sizeof(mes_set_hw_res_pkt)); + + mes_set_hw_res_pkt.header.type = MES_API_TYPE_SCHEDULER; + mes_set_hw_res_pkt.header.opcode = MES_SCH_API_SET_HW_RSRC_1; + mes_set_hw_res_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS; + mes_set_hw_res_pkt.enable_mes_info_ctx = 1; + + ret = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, + &mes->resource_1, + &mes->resource_1_gpu_addr, + &mes->resource_1_addr); + if (ret) { + dev_err(adev->dev, "(%d) failed to create mes resource_1 bo\n", ret); + return ret; + } + + mes_set_hw_res_pkt.mes_info_ctx_mc_addr = mes->resource_1_gpu_addr; + mes_set_hw_res_pkt.mes_info_ctx_size = mes->resource_1->tbo.base.size; + return mes_v11_0_submit_pkt_and_poll_completion(mes, + &mes_set_hw_res_pkt, sizeof(mes_set_hw_res_pkt), + offsetof(union MESAPI_SET_HW_RESOURCES_1, api_status)); +} + static const struct amdgpu_mes_funcs mes_v11_0_funcs = { .add_hw_queue = mes_v11_0_add_hw_queue, .remove_hw_queue = mes_v11_0_remove_hw_queue, @@ -1203,6 +1336,14 @@ static int mes_v11_0_hw_init(void *handle) if (r) goto failure; + if (amdgpu_sriov_is_mes_info_enable(adev)) { + r = mes_v11_0_set_hw_resources_1(&adev->mes); + if (r) { + DRM_ERROR("failed mes_v11_0_set_hw_resources_1, r=%d\n", r); + goto failure; + } + } + r = mes_v11_0_query_sched_status(&adev->mes); if (r) { DRM_ERROR("MES is busy\n"); @@ -1226,6 +1367,11 @@ failure: static int mes_v11_0_hw_fini(void *handle) { + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + if (amdgpu_sriov_is_mes_info_enable(adev)) { + amdgpu_bo_free_kernel(&adev->mes.resource_1, &adev->mes.resource_1_gpu_addr, + &adev->mes.resource_1_addr); + } return 0; } @@ -1291,6 +1437,8 @@ static const struct amd_ip_funcs mes_v11_0_ip_funcs = { .hw_fini = mes_v11_0_hw_fini, .suspend = mes_v11_0_suspend, .resume = mes_v11_0_resume, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version mes_v11_0_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c index c0fc44cdd6..7a1ff29841 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c @@ -559,6 +559,20 @@ static void mmhub_v1_8_get_clockgating(struct amdgpu_device *adev, u64 *flags) } +static bool mmhub_v1_8_query_utcl2_poison_status(struct amdgpu_device *adev, + int hub_inst) +{ + u32 fed, status; + + status = RREG32_SOC15(MMHUB, hub_inst, regVM_L2_PROTECTION_FAULT_STATUS); + fed = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, FED); + /* reset page fault status */ + WREG32_P(SOC15_REG_OFFSET(MMHUB, hub_inst, + regVM_L2_PROTECTION_FAULT_STATUS), 1, ~1); + + return fed; +} + const struct amdgpu_mmhub_funcs mmhub_v1_8_funcs = { .get_fb_location = mmhub_v1_8_get_fb_location, .init = mmhub_v1_8_init, @@ -568,6 +582,7 @@ const struct amdgpu_mmhub_funcs mmhub_v1_8_funcs = { .setup_vm_pt_regs = mmhub_v1_8_setup_vm_pt_regs, .set_clockgating = mmhub_v1_8_set_clockgating, .get_clockgating = mmhub_v1_8_get_clockgating, + .query_utcl2_poison_status = mmhub_v1_8_query_utcl2_poison_status, }; static const struct amdgpu_ras_err_status_reg_entry mmhub_v1_8_ce_reg_list[] = { @@ -706,28 +721,32 @@ static const struct amdgpu_ras_block_hw_ops mmhub_v1_8_ras_hw_ops = { .reset_ras_error_count = mmhub_v1_8_reset_ras_error_count, }; -static int mmhub_v1_8_aca_bank_generate_report(struct aca_handle *handle, - struct aca_bank *bank, enum aca_error_type type, - struct aca_bank_report *report, void *data) +static int mmhub_v1_8_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) { - u64 status, misc0; + struct aca_bank_info info; + u64 misc0; int ret; - status = bank->regs[ACA_REG_IDX_STATUS]; - if ((type == ACA_ERROR_TYPE_UE && - ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_FAULT) || - (type == ACA_ERROR_TYPE_CE && - ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_CE)) { - - ret = aca_bank_info_decode(bank, &report->info); - if (ret) - return ret; - - misc0 = bank->regs[ACA_REG_IDX_MISC0]; - report->count[type] = ACA_REG__MISC0__ERRCNT(misc0); + ret = aca_bank_info_decode(bank, &info); + if (ret) + return ret; + + misc0 = bank->regs[ACA_REG_IDX_MISC0]; + switch (type) { + case ACA_SMU_TYPE_UE: + ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE, + 1ULL); + break; + case ACA_SMU_TYPE_CE: + ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_CE, + ACA_REG__MISC0__ERRCNT(misc0)); + break; + default: + return -EINVAL; } - return 0; + return ret; } /* reference to smu driver if header file */ @@ -741,7 +760,7 @@ static int mmhub_v1_8_err_codes[] = { }; static bool mmhub_v1_8_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, - enum aca_error_type type, void *data) + enum aca_smu_type type, void *data) { u32 instlo; @@ -760,7 +779,7 @@ static bool mmhub_v1_8_aca_bank_is_valid(struct aca_handle *handle, struct aca_b } static const struct aca_bank_ops mmhub_v1_8_aca_bank_ops = { - .aca_bank_generate_report = mmhub_v1_8_aca_bank_generate_report, + .aca_bank_parser = mmhub_v1_8_aca_bank_parser, .aca_bank_is_valid = mmhub_v1_8_aca_bank_is_valid, }; diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c index a2bd2c3b1e..0c7275bca8 100644 --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c @@ -276,6 +276,8 @@ static void xgpu_ai_mailbox_flr_work(struct work_struct *work) timeout -= 10; } while (timeout > 1); + dev_warn(adev->dev, "waiting IDH_FLR_NOTIFICATION_CMPL timeout\n"); + flr_done: atomic_set(&adev->reset_domain->in_gpu_reset, 0); up_write(&adev->reset_domain->sem); diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c index 77f5b55dec..aba00d9616 100644 --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c @@ -309,6 +309,8 @@ static void xgpu_nv_mailbox_flr_work(struct work_struct *work) timeout -= 10; } while (timeout > 1); + dev_warn(adev->dev, "waiting IDH_FLR_NOTIFICATION_CMPL timeout\n"); + flr_done: atomic_set(&adev->reset_domain->in_gpu_reset, 0); up_write(&adev->reset_domain->sem); @@ -444,7 +446,6 @@ static void xgpu_nv_ras_poison_handler(struct amdgpu_device *adev, amdgpu_virt_fini_data_exchange(adev); xgpu_nv_send_access_requests_with_param(adev, IDH_RAS_POISON, block, 0, 0); - amdgpu_virt_init_data_exchange(adev); } } diff --git a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c index 4178f4e5da..b281462093 100644 --- a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c @@ -713,6 +713,8 @@ static const struct amd_ip_funcs navi10_ih_ip_funcs = { .set_clockgating_state = navi10_ih_set_clockgating_state, .set_powergating_state = navi10_ih_set_powergating_state, .get_clockgating_state = navi10_ih_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ih_funcs navi10_ih_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c index 4d7976b777..12e54047bf 100644 --- a/drivers/gpu/drm/amd/amdgpu/nv.c +++ b/drivers/gpu/drm/amd/amdgpu/nv.c @@ -110,7 +110,7 @@ static const struct amdgpu_video_codec_info sc_video_codecs_decode_array_vcn0[] {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 4096, 4096, 4)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)}, }; @@ -121,7 +121,7 @@ static const struct amdgpu_video_codec_info sc_video_codecs_decode_array_vcn1[] {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 4096, 4096, 4)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, }; @@ -199,7 +199,7 @@ static const struct amdgpu_video_codec_info yc_video_codecs_decode_array[] = { {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)}, }; @@ -1131,4 +1131,6 @@ static const struct amd_ip_funcs nv_common_ip_funcs = { .set_clockgating_state = nv_common_set_clockgating_state, .set_powergating_state = nv_common_set_powergating_state, .get_clockgating_state = nv_common_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h index 7566973ed8..37b5ddd6f1 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h +++ b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h @@ -464,8 +464,9 @@ struct psp_gfx_rb_frame #define PSP_ERR_UNKNOWN_COMMAND 0x00000100 enum tee_error_code { - TEE_SUCCESS = 0x00000000, - TEE_ERROR_NOT_SUPPORTED = 0xFFFF000A, + TEE_SUCCESS = 0x00000000, + TEE_ERROR_CANCEL = 0xFFFF0002, + TEE_ERROR_NOT_SUPPORTED = 0xFFFF000A, }; #endif /* _PSP_TEE_GFX_IF_H_ */ diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c index 238abd9807..40b28298af 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c @@ -174,7 +174,8 @@ static int psp_v14_0_bootloader_load_intf_drv(struct psp_context *psp) static int psp_v14_0_bootloader_load_dbg_drv(struct psp_context *psp) { - return psp_v14_0_bootloader_load_component(psp, &psp->dbg_drv, PSP_BL__LOAD_DBGDRV); + /* dbg_drv was renamed to had_drv in psp v14 */ + return psp_v14_0_bootloader_load_component(psp, &psp->dbg_drv, PSP_BL__LOAD_HADDRV); } static int psp_v14_0_bootloader_load_ras_drv(struct psp_context *psp) @@ -182,6 +183,10 @@ static int psp_v14_0_bootloader_load_ras_drv(struct psp_context *psp) return psp_v14_0_bootloader_load_component(psp, &psp->ras_drv, PSP_BL__LOAD_RASDRV); } +static int psp_v14_0_bootloader_load_ipkeymgr_drv(struct psp_context *psp) +{ + return psp_v14_0_bootloader_load_component(psp, &psp->ipkeymgr_drv, PSP_BL__LOAD_IPKEYMGRDRV); +} static int psp_v14_0_bootloader_load_sos(struct psp_context *psp) { @@ -658,6 +663,7 @@ static const struct psp_funcs psp_v14_0_funcs = { .bootloader_load_intf_drv = psp_v14_0_bootloader_load_intf_drv, .bootloader_load_dbg_drv = psp_v14_0_bootloader_load_dbg_drv, .bootloader_load_ras_drv = psp_v14_0_bootloader_load_ras_drv, + .bootloader_load_ipkeymgr_drv = psp_v14_0_bootloader_load_ipkeymgr_drv, .bootloader_load_sos = psp_v14_0_bootloader_load_sos, .ring_create = psp_v14_0_ring_create, .ring_stop = psp_v14_0_ring_stop, diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c index 07e19caf2b..ac8a9b9b3e 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c @@ -1113,6 +1113,8 @@ static const struct amd_ip_funcs sdma_v2_4_ip_funcs = { .soft_reset = sdma_v2_4_soft_reset, .set_clockgating_state = sdma_v2_4_set_clockgating_state, .set_powergating_state = sdma_v2_4_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = { @@ -1176,7 +1178,7 @@ static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: unused + * @copy_flags: unused * * Copy GPU buffers using the DMA engine (VI). * Used by the amdgpu ttm implementation to move pages if @@ -1186,7 +1188,7 @@ static void sdma_v2_4_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index 2ad615be4b..b8ebdc4ae6 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -1553,6 +1553,8 @@ static const struct amd_ip_funcs sdma_v3_0_ip_funcs = { .set_clockgating_state = sdma_v3_0_set_clockgating_state, .set_powergating_state = sdma_v3_0_set_powergating_state, .get_clockgating_state = sdma_v3_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = { @@ -1616,7 +1618,7 @@ static void sdma_v3_0_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: unused + * @copy_flags: unused * * Copy GPU buffers using the DMA engine (VI). * Used by the amdgpu ttm implementation to move pages if @@ -1626,7 +1628,7 @@ static void sdma_v3_0_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index 3fac292334..772604feb6 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -2451,7 +2451,7 @@ static void sdma_v4_0_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: if a secure copy should be used + * @copy_flags: copy flags for the buffers * * Copy GPU buffers using the DMA engine (VEGA10/12). * Used by the amdgpu ttm implementation to move pages if @@ -2461,11 +2461,11 @@ static void sdma_v4_0_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) | - SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0); + SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0); ib->ptr[ib->length_dw++] = byte_count - 1; ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ ib->ptr[ib->length_dw++] = lower_32_bits(src_offset); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c index e708468ac5..341b24d832 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c @@ -1945,7 +1945,7 @@ static void sdma_v4_4_2_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: if a secure copy should be used + * @copy_flags: copy flags for the buffers * * Copy GPU buffers using the DMA engine. * Used by the amdgpu ttm implementation to move pages if @@ -1955,11 +1955,11 @@ static void sdma_v4_4_2_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) | - SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0); + SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0); ib->ptr[ib->length_dw++] = byte_count - 1; ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ ib->ptr[ib->length_dw++] = lower_32_bits(src_offset); @@ -2180,35 +2180,39 @@ static const struct amdgpu_ras_block_hw_ops sdma_v4_4_2_ras_hw_ops = { .reset_ras_error_count = sdma_v4_4_2_reset_ras_error_count, }; -static int sdma_v4_4_2_aca_bank_generate_report(struct aca_handle *handle, - struct aca_bank *bank, enum aca_error_type type, - struct aca_bank_report *report, void *data) +static int sdma_v4_4_2_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) { - u64 status, misc0; + struct aca_bank_info info; + u64 misc0; int ret; - status = bank->regs[ACA_REG_IDX_STATUS]; - if ((type == ACA_ERROR_TYPE_UE && - ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_FAULT) || - (type == ACA_ERROR_TYPE_CE && - ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_CE)) { + ret = aca_bank_info_decode(bank, &info); + if (ret) + return ret; - ret = aca_bank_info_decode(bank, &report->info); - if (ret) - return ret; - - misc0 = bank->regs[ACA_REG_IDX_MISC0]; - report->count[type] = ACA_REG__MISC0__ERRCNT(misc0); + misc0 = bank->regs[ACA_REG_IDX_MISC0]; + switch (type) { + case ACA_SMU_TYPE_UE: + ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE, + 1ULL); + break; + case ACA_SMU_TYPE_CE: + ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_CE, + ACA_REG__MISC0__ERRCNT(misc0)); + break; + default: + return -EINVAL; } - return 0; + return ret; } /* CODE_SDMA0 - CODE_SDMA4, reference to smu driver if header file */ static int sdma_v4_4_2_err_codes[] = { 33, 34, 35, 36 }; static bool sdma_v4_4_2_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, - enum aca_error_type type, void *data) + enum aca_smu_type type, void *data) { u32 instlo; @@ -2227,7 +2231,7 @@ static bool sdma_v4_4_2_aca_bank_is_valid(struct aca_handle *handle, struct aca_ } static const struct aca_bank_ops sdma_v4_4_2_aca_bank_ops = { - .aca_bank_generate_report = sdma_v4_4_2_aca_bank_generate_report, + .aca_bank_parser = sdma_v4_4_2_aca_bank_parser, .aca_bank_is_valid = sdma_v4_4_2_aca_bank_is_valid, }; diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c index 883e8a1b8a..b7d33d78bc 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c @@ -999,7 +999,8 @@ static int sdma_v5_0_ring_test_ring(struct amdgpu_ring *ring) r = amdgpu_ring_alloc(ring, 20); if (r) { DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_device_wb_free(adev, index); + if (!ring->is_mes_queue) + amdgpu_device_wb_free(adev, index); return r; } @@ -1805,7 +1806,7 @@ static void sdma_v5_0_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: if a secure copy should be used + * @copy_flags: copy flags for the buffers * * Copy GPU buffers using the DMA engine (NAVI10). * Used by the amdgpu ttm implementation to move pages if @@ -1815,11 +1816,11 @@ static void sdma_v5_0_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) | - SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0); + SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0); ib->ptr[ib->length_dw++] = byte_count - 1; ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ ib->ptr[ib->length_dw++] = lower_32_bits(src_offset); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c index da01b524b9..af1e90159c 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c @@ -176,6 +176,14 @@ static void sdma_v5_2_ring_set_wptr(struct amdgpu_ring *ring) DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n", ring->doorbell_index, ring->wptr << 2); WDOORBELL64(ring->doorbell_index, ring->wptr << 2); + /* SDMA seems to miss doorbells sometimes when powergating kicks in. + * Updating the wptr directly will wake it. This is only safe because + * we disallow gfxoff in begin_use() and then allow it again in end_use(). + */ + WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR), + lower_32_bits(ring->wptr << 2)); + WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI), + upper_32_bits(ring->wptr << 2)); } else { DRM_DEBUG("Not using doorbell -- " "mmSDMA%i_GFX_RB_WPTR == 0x%08x " @@ -839,7 +847,8 @@ static int sdma_v5_2_ring_test_ring(struct amdgpu_ring *ring) r = amdgpu_ring_alloc(ring, 20); if (r) { DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_device_wb_free(adev, index); + if (!ring->is_mes_queue) + amdgpu_device_wb_free(adev, index); return r; } @@ -1646,6 +1655,10 @@ static void sdma_v5_2_ring_begin_use(struct amdgpu_ring *ring) * but it shouldn't hurt for other parts since * this GFXOFF will be disallowed anyway when SDMA is * active, this just makes it explicit. + * sdma_v5_2_ring_set_wptr() takes advantage of this + * to update the wptr because sometimes SDMA seems to miss + * doorbells when entering PG. If you remove this, update + * sdma_v5_2_ring_set_wptr() as well! */ amdgpu_gfx_off_ctrl(adev, false); } @@ -1751,7 +1764,7 @@ static void sdma_v5_2_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: if a secure copy should be used + * @copy_flags: copy flags for the buffers * * Copy GPU buffers using the DMA engine. * Used by the amdgpu ttm implementation to move pages if @@ -1761,11 +1774,11 @@ static void sdma_v5_2_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) | - SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0); + SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0); ib->ptr[ib->length_dw++] = byte_count - 1; ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ ib->ptr[ib->length_dw++] = lower_32_bits(src_offset); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c index 361835a61f..c833b6b837 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c @@ -507,6 +507,13 @@ static int sdma_v6_0_gfx_resume(struct amdgpu_device *adev) /* set minor_ptr_update to 0 after wptr programed */ WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_MINOR_PTR_UPDATE), 0); + /* Set up sdma hang watchdog */ + temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_WATCHDOG_CNTL)); + /* 100ms per unit */ + temp = REG_SET_FIELD(temp, SDMA0_WATCHDOG_CNTL, QUEUE_HANG_COUNT, + max(adev->usec_timeout/100000, 1)); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_WATCHDOG_CNTL), temp); + /* Set up RESP_MODE to non-copy addresses */ temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_CNTL)); temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3); @@ -854,7 +861,8 @@ static int sdma_v6_0_ring_test_ring(struct amdgpu_ring *ring) r = amdgpu_ring_alloc(ring, 5); if (r) { DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_device_wb_free(adev, index); + if (!ring->is_mes_queue) + amdgpu_device_wb_free(adev, index); return r; } @@ -1567,7 +1575,7 @@ static void sdma_v6_0_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: if a secure copy should be used + * @copy_flags: copy flags for the buffers * * Copy GPU buffers using the DMA engine. * Used by the amdgpu ttm implementation to move pages if @@ -1577,11 +1585,11 @@ static void sdma_v6_0_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COPY) | SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) | - SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0); + SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0); ib->ptr[ib->length_dw++] = byte_count - 1; ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ ib->ptr[ib->length_dw++] = lower_32_bits(src_offset); diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c index 23e4ef4fff..85235470e8 100644 --- a/drivers/gpu/drm/amd/amdgpu/si.c +++ b/drivers/gpu/drm/amd/amdgpu/si.c @@ -1409,9 +1409,9 @@ static int si_gpu_pci_config_reset(struct amdgpu_device *adev) return r; } -static bool si_asic_supports_baco(struct amdgpu_device *adev) +static int si_asic_supports_baco(struct amdgpu_device *adev) { - return false; + return 0; } static enum amd_reset_method @@ -2706,6 +2706,8 @@ static const struct amd_ip_funcs si_common_ip_funcs = { .soft_reset = si_common_soft_reset, .set_clockgating_state = si_common_set_clockgating_state, .set_powergating_state = si_common_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ip_block_version si_common_ip_block = diff --git a/drivers/gpu/drm/amd/amdgpu/si_dma.c b/drivers/gpu/drm/amd/amdgpu/si_dma.c index 9aa0e11ee6..11db5b7558 100644 --- a/drivers/gpu/drm/amd/amdgpu/si_dma.c +++ b/drivers/gpu/drm/amd/amdgpu/si_dma.c @@ -708,6 +708,8 @@ static const struct amd_ip_funcs si_dma_ip_funcs = { .soft_reset = si_dma_soft_reset, .set_clockgating_state = si_dma_set_clockgating_state, .set_powergating_state = si_dma_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs si_dma_ring_funcs = { @@ -761,7 +763,7 @@ static void si_dma_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: is this a secure operation + * @copy_flags: unused * * Copy GPU buffers using the DMA engine (VI). * Used by the amdgpu ttm implementation to move pages if @@ -771,7 +773,7 @@ static void si_dma_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_COPY, 1, 0, 0, byte_count); diff --git a/drivers/gpu/drm/amd/amdgpu/si_ih.c b/drivers/gpu/drm/amd/amdgpu/si_ih.c index cada9f300a..5237395e4f 100644 --- a/drivers/gpu/drm/amd/amdgpu/si_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/si_ih.c @@ -296,6 +296,8 @@ static const struct amd_ip_funcs si_ih_ip_funcs = { .soft_reset = si_ih_soft_reset, .set_clockgating_state = si_ih_set_clockgating_state, .set_powergating_state = si_ih_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ih_funcs si_ih_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.c b/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.c index 04c797d545..0af648931d 100644 --- a/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.c +++ b/drivers/gpu/drm/amd/amdgpu/smu_v13_0_10.c @@ -91,7 +91,7 @@ static int smu_v13_0_10_mode2_suspend_ip(struct amdgpu_device *adev) adev->ip_blocks[i].status.hw = false; } - return r; + return 0; } static int diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.c b/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.c new file mode 100644 index 0000000000..2a51a70d48 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.c @@ -0,0 +1,62 @@ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "smuio_v14_0_2.h" +#include "smuio/smuio_14_0_2_offset.h" +#include "smuio/smuio_14_0_2_sh_mask.h" +#include <linux/preempt.h> + +static u32 smuio_v14_0_2_get_rom_index_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(SMUIO, 0, regROM_INDEX); +} + +static u32 smuio_v14_0_2_get_rom_data_offset(struct amdgpu_device *adev) +{ + return SOC15_REG_OFFSET(SMUIO, 0, regROM_DATA); +} + +static u64 smuio_v14_0_2_get_gpu_clock_counter(struct amdgpu_device *adev) +{ + u64 clock; + u64 clock_counter_lo, clock_counter_hi_pre, clock_counter_hi_after; + + preempt_disable(); + clock_counter_hi_pre = (u64)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_UPPER); + clock_counter_lo = (u64)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_LOWER); + /* the clock counter may be udpated during polling the counters */ + clock_counter_hi_after = (u64)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_UPPER); + if (clock_counter_hi_pre != clock_counter_hi_after) + clock_counter_lo = (u64)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_LOWER); + preempt_enable(); + + clock = clock_counter_lo | (clock_counter_hi_after << 32ULL); + + return clock; +} + +const struct amdgpu_smuio_funcs smuio_v14_0_2_funcs = { + .get_rom_index_offset = smuio_v14_0_2_get_rom_index_offset, + .get_rom_data_offset = smuio_v14_0_2_get_rom_data_offset, + .get_gpu_clock_counter = smuio_v14_0_2_get_gpu_clock_counter, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.h b/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.h new file mode 100644 index 0000000000..6e617f832d --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/smuio_v14_0_2.h @@ -0,0 +1,30 @@ +/* + * Copyright 2023 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 __SMUIO_V14_0_2_H__ +#define __SMUIO_V14_0_2_H__ + +#include "soc15_common.h" + +extern const struct amdgpu_smuio_funcs smuio_v14_0_2_funcs; + +#endif /* __SMUIO_V14_0_2_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index dec81ccf62..170f02e967 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -143,7 +143,7 @@ static const struct amdgpu_video_codec_info rn_video_codecs_decode_array[] = {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 4096, 4096, 4)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, }; @@ -156,7 +156,7 @@ static const struct amdgpu_video_codecs rn_video_codecs_decode = static const struct amdgpu_video_codec_info vcn_4_0_3_video_codecs_decode_array[] = { {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)}, }; @@ -502,7 +502,7 @@ static int soc15_asic_baco_reset(struct amdgpu_device *adev) static enum amd_reset_method soc15_asic_reset_method(struct amdgpu_device *adev) { - bool baco_reset = false; + int baco_reset = 0; bool connected_to_cpu = false; struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); @@ -540,7 +540,7 @@ soc15_asic_reset_method(struct amdgpu_device *adev) */ if (ras && adev->ras_enabled && adev->pm.fw_version <= 0x283400) - baco_reset = false; + baco_reset = 0; } else { baco_reset = amdgpu_dpm_is_baco_supported(adev); } @@ -620,7 +620,7 @@ static int soc15_asic_reset(struct amdgpu_device *adev) } } -static bool soc15_supports_baco(struct amdgpu_device *adev) +static int soc15_supports_baco(struct amdgpu_device *adev) { switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(9, 0, 0): @@ -628,13 +628,13 @@ static bool soc15_supports_baco(struct amdgpu_device *adev) if (adev->asic_type == CHIP_VEGA20) { if (adev->psp.sos.fw_version >= 0x80067) return amdgpu_dpm_is_baco_supported(adev); - return false; + return 0; } else { return amdgpu_dpm_is_baco_supported(adev); } break; default: - return false; + return 0; } } @@ -1501,4 +1501,6 @@ static const struct amd_ip_funcs soc15_common_ip_funcs = { .set_clockgating_state = soc15_common_set_clockgating_state, .set_powergating_state = soc15_common_set_powergating_state, .get_clockgating_state= soc15_common_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.h b/drivers/gpu/drm/amd/amdgpu/soc15.h index 1444b7765e..282584a48b 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.h +++ b/drivers/gpu/drm/amd/amdgpu/soc15.h @@ -88,6 +88,8 @@ struct soc15_ras_field_entry { }; #define SOC15_REG_ENTRY(ip, inst, reg) ip##_HWIP, inst, reg##_BASE_IDX, reg +#define SOC15_REG_ENTRY_STR(ip, inst, reg) \ + { ip##_HWIP, inst, reg##_BASE_IDX, reg, #reg } #define SOC15_REG_ENTRY_OFFSET(entry) (adev->reg_offset[entry.hwip][entry.inst][entry.seg] + entry.reg_offset) diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.c b/drivers/gpu/drm/amd/amdgpu/soc21.c index 43ca63fe85..fb67974675 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc21.c +++ b/drivers/gpu/drm/amd/amdgpu/soc21.c @@ -72,7 +72,7 @@ static const struct amdgpu_video_codecs vcn_4_0_0_video_codecs_encode_vcn1 = { static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_decode_array_vcn0[] = { {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)}, }; @@ -80,7 +80,7 @@ static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_decode_array_ static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_decode_array_vcn1[] = { {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, }; @@ -985,4 +985,6 @@ static const struct amd_ip_funcs soc21_common_ip_funcs = { .set_clockgating_state = soc21_common_set_clockgating_state, .set_powergating_state = soc21_common_set_powergating_state, .get_clockgating_state = soc21_common_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; diff --git a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h index 056d4df8fa..3ac56a9645 100644 --- a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h +++ b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h @@ -146,6 +146,7 @@ struct ta_ras_mca_addr { uint32_t ch_inst; uint32_t umc_inst; uint32_t node_inst; + uint32_t socket_id; }; struct ta_ras_phy_addr { diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c index 450b6e8315..24d49d8136 100644 --- a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c @@ -486,6 +486,8 @@ static const struct amd_ip_funcs tonga_ih_ip_funcs = { .post_soft_reset = tonga_ih_post_soft_reset, .set_clockgating_state = tonga_ih_set_clockgating_state, .set_powergating_state = tonga_ih_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ih_funcs tonga_ih_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c index 77af4e25ff..bfe61d86ee 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c @@ -28,27 +28,7 @@ #include "umc/umc_12_0_0_sh_mask.h" #include "mp/mp_13_0_6_sh_mask.h" -const uint32_t - umc_v12_0_channel_idx_tbl[] - [UMC_V12_0_UMC_INSTANCE_NUM] - [UMC_V12_0_CHANNEL_INSTANCE_NUM] = { - {{3, 7, 11, 15, 2, 6, 10, 14}, {1, 5, 9, 13, 0, 4, 8, 12}, - {19, 23, 27, 31, 18, 22, 26, 30}, {17, 21, 25, 29, 16, 20, 24, 28}}, - {{47, 43, 39, 35, 46, 42, 38, 34}, {45, 41, 37, 33, 44, 40, 36, 32}, - {63, 59, 55, 51, 62, 58, 54, 50}, {61, 57, 53, 49, 60, 56, 52, 48}}, - {{79, 75, 71, 67, 78, 74, 70, 66}, {77, 73, 69, 65, 76, 72, 68, 64}, - {95, 91, 87, 83, 94, 90, 86, 82}, {93, 89, 85, 81, 92, 88, 84, 80}}, - {{99, 103, 107, 111, 98, 102, 106, 110}, {97, 101, 105, 109, 96, 100, 104, 108}, - {115, 119, 123, 127, 114, 118, 122, 126}, {113, 117, 121, 125, 112, 116, 120, 124}} - }; - -/* mapping of MCA error address to normalized address */ -static const uint32_t umc_v12_0_ma2na_mapping[] = { - 0, 5, 6, 8, 9, 14, 12, 13, - 10, 11, 15, 16, 17, 18, 19, 20, - 21, 22, 23, 24, 25, 26, 27, 28, - 24, 7, 29, 30, -}; +#define MAX_ECC_NUM_PER_RETIREMENT 32 static inline uint64_t get_umc_v12_0_reg_offset(struct amdgpu_device *adev, uint32_t node_inst, @@ -192,99 +172,74 @@ static void umc_v12_0_query_ras_error_count(struct amdgpu_device *adev, umc_v12_0_reset_error_count(adev); } -static bool umc_v12_0_bit_wise_xor(uint32_t val) +static void umc_v12_0_convert_error_address(struct amdgpu_device *adev, + struct ras_err_data *err_data, + struct ta_ras_query_address_input *addr_in) { - bool result = 0; - int i; + uint32_t col, row, row_xor, bank, channel_index; + uint64_t soc_pa, retired_page, column, err_addr; + struct ta_ras_query_address_output addr_out; - for (i = 0; i < 32; i++) - result = result ^ ((val >> i) & 0x1); + err_addr = addr_in->ma.err_addr; + addr_in->addr_type = TA_RAS_MCA_TO_PA; + if (psp_ras_query_address(&adev->psp, addr_in, &addr_out)) { + dev_warn(adev->dev, "Failed to query RAS physical address for 0x%llx", + err_addr); - return result; -} + return; + } + + soc_pa = addr_out.pa.pa; + bank = addr_out.pa.bank; + channel_index = addr_out.pa.channel_idx; -static void umc_v12_0_mca_addr_to_pa(struct amdgpu_device *adev, - uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst, - uint32_t node_inst, - struct ta_ras_query_address_output *addr_out) -{ - uint32_t channel_index, i; - uint64_t na, soc_pa; - uint32_t bank_hash0, bank_hash1, bank_hash2, bank_hash3, col, row; - uint32_t bank0, bank1, bank2, bank3, bank; - - bank_hash0 = (err_addr >> UMC_V12_0_MCA_B0_BIT) & 0x1ULL; - bank_hash1 = (err_addr >> UMC_V12_0_MCA_B1_BIT) & 0x1ULL; - bank_hash2 = (err_addr >> UMC_V12_0_MCA_B2_BIT) & 0x1ULL; - bank_hash3 = (err_addr >> UMC_V12_0_MCA_B3_BIT) & 0x1ULL; col = (err_addr >> 1) & 0x1fULL; row = (err_addr >> 10) & 0x3fffULL; + row_xor = row ^ (0x1ULL << 13); + /* clear [C3 C2] in soc physical address */ + soc_pa &= ~(0x3ULL << UMC_V12_0_PA_C2_BIT); + /* clear [C4] in soc physical address */ + soc_pa &= ~(0x1ULL << UMC_V12_0_PA_C4_BIT); + + /* loop for all possibilities of [C4 C3 C2] */ + for (column = 0; column < UMC_V12_0_NA_MAP_PA_NUM; column++) { + retired_page = soc_pa | ((column & 0x3) << UMC_V12_0_PA_C2_BIT); + retired_page |= (((column & 0x4) >> 2) << UMC_V12_0_PA_C4_BIT); + /* include column bit 0 and 1 */ + col &= 0x3; + col |= (column << 2); + dev_info(adev->dev, + "Error Address(PA):0x%-10llx Row:0x%-4x Col:0x%-2x Bank:0x%x Channel:0x%x\n", + retired_page, row, col, bank, channel_index); + amdgpu_umc_fill_error_record(err_data, err_addr, + retired_page, channel_index, addr_in->ma.umc_inst); - /* apply bank hash algorithm */ - bank0 = - bank_hash0 ^ (UMC_V12_0_XOR_EN0 & - (umc_v12_0_bit_wise_xor(col & UMC_V12_0_COL_XOR0) ^ - (umc_v12_0_bit_wise_xor(row & UMC_V12_0_ROW_XOR0)))); - bank1 = - bank_hash1 ^ (UMC_V12_0_XOR_EN1 & - (umc_v12_0_bit_wise_xor(col & UMC_V12_0_COL_XOR1) ^ - (umc_v12_0_bit_wise_xor(row & UMC_V12_0_ROW_XOR1)))); - bank2 = - bank_hash2 ^ (UMC_V12_0_XOR_EN2 & - (umc_v12_0_bit_wise_xor(col & UMC_V12_0_COL_XOR2) ^ - (umc_v12_0_bit_wise_xor(row & UMC_V12_0_ROW_XOR2)))); - bank3 = - bank_hash3 ^ (UMC_V12_0_XOR_EN3 & - (umc_v12_0_bit_wise_xor(col & UMC_V12_0_COL_XOR3) ^ - (umc_v12_0_bit_wise_xor(row & UMC_V12_0_ROW_XOR3)))); - - bank = bank0 | (bank1 << 1) | (bank2 << 2) | (bank3 << 3); - err_addr &= ~0x3c0ULL; - err_addr |= (bank << UMC_V12_0_MCA_B0_BIT); - - na = 0x0; - /* convert mca error address to normalized address */ - for (i = 1; i < ARRAY_SIZE(umc_v12_0_ma2na_mapping); i++) - na |= ((err_addr >> i) & 0x1ULL) << umc_v12_0_ma2na_mapping[i]; - - channel_index = - adev->umc.channel_idx_tbl[node_inst * adev->umc.umc_inst_num * - adev->umc.channel_inst_num + - umc_inst * adev->umc.channel_inst_num + - ch_inst]; - /* translate umc channel address to soc pa, 3 parts are included */ - soc_pa = ADDR_OF_32KB_BLOCK(na) | - ADDR_OF_256B_BLOCK(channel_index) | - OFFSET_IN_256B_BLOCK(na); - - /* the umc channel bits are not original values, they are hashed */ - UMC_V12_0_SET_CHANNEL_HASH(channel_index, soc_pa); - - addr_out->pa.pa = soc_pa; - addr_out->pa.bank = bank; - addr_out->pa.channel_idx = channel_index; + /* shift R13 bit */ + retired_page ^= (0x1ULL << UMC_V12_0_PA_R13_BIT); + dev_info(adev->dev, + "Error Address(PA):0x%-10llx Row:0x%-4x Col:0x%-2x Bank:0x%x Channel:0x%x\n", + retired_page, row_xor, col, bank, channel_index); + amdgpu_umc_fill_error_record(err_data, err_addr, + retired_page, channel_index, addr_in->ma.umc_inst); + } } -static void umc_v12_0_convert_error_address(struct amdgpu_device *adev, - struct ras_err_data *err_data, uint64_t err_addr, - uint32_t ch_inst, uint32_t umc_inst, - uint32_t node_inst) +static int umc_v12_0_convert_err_addr(struct amdgpu_device *adev, + struct ta_ras_query_address_input *addr_in, + uint64_t *pfns, int len) { uint32_t col, row, row_xor, bank, channel_index; - uint64_t soc_pa, retired_page, column; - struct ta_ras_query_address_input addr_in; + uint64_t soc_pa, retired_page, column, err_addr; struct ta_ras_query_address_output addr_out; + uint32_t pos = 0; - addr_in.addr_type = TA_RAS_MCA_TO_PA; - addr_in.ma.err_addr = err_addr; - addr_in.ma.ch_inst = ch_inst; - addr_in.ma.umc_inst = umc_inst; - addr_in.ma.node_inst = node_inst; - - if (psp_ras_query_address(&adev->psp, &addr_in, &addr_out)) - /* fallback to old path if fail to get pa from psp */ - umc_v12_0_mca_addr_to_pa(adev, err_addr, ch_inst, umc_inst, - node_inst, &addr_out); + err_addr = addr_in->ma.err_addr; + addr_in->addr_type = TA_RAS_MCA_TO_PA; + if (psp_ras_query_address(&adev->psp, addr_in, &addr_out)) { + dev_warn(adev->dev, "Failed to query RAS physical address for 0x%llx", + err_addr); + return 0; + } soc_pa = addr_out.pa.pa; bank = addr_out.pa.bank; @@ -302,33 +257,42 @@ static void umc_v12_0_convert_error_address(struct amdgpu_device *adev, for (column = 0; column < UMC_V12_0_NA_MAP_PA_NUM; column++) { retired_page = soc_pa | ((column & 0x3) << UMC_V12_0_PA_C2_BIT); retired_page |= (((column & 0x4) >> 2) << UMC_V12_0_PA_C4_BIT); + + if (pos >= len) + return 0; + pfns[pos++] = retired_page >> AMDGPU_GPU_PAGE_SHIFT; + /* include column bit 0 and 1 */ col &= 0x3; col |= (column << 2); dev_info(adev->dev, "Error Address(PA):0x%-10llx Row:0x%-4x Col:0x%-2x Bank:0x%x Channel:0x%x\n", retired_page, row, col, bank, channel_index); - amdgpu_umc_fill_error_record(err_data, err_addr, - retired_page, channel_index, umc_inst); /* shift R13 bit */ retired_page ^= (0x1ULL << UMC_V12_0_PA_R13_BIT); + + if (pos >= len) + return 0; + pfns[pos++] = retired_page >> AMDGPU_GPU_PAGE_SHIFT; + dev_info(adev->dev, "Error Address(PA):0x%-10llx Row:0x%-4x Col:0x%-2x Bank:0x%x Channel:0x%x\n", retired_page, row_xor, col, bank, channel_index); - amdgpu_umc_fill_error_record(err_data, err_addr, - retired_page, channel_index, umc_inst); } + + return pos; } static int umc_v12_0_query_error_address(struct amdgpu_device *adev, uint32_t node_inst, uint32_t umc_inst, uint32_t ch_inst, void *data) { + struct ras_err_data *err_data = (struct ras_err_data *)data; + struct ta_ras_query_address_input addr_in; uint64_t mc_umc_status_addr; uint64_t mc_umc_status, err_addr; uint64_t mc_umc_addrt0; - struct ras_err_data *err_data = (struct ras_err_data *)data; uint64_t umc_reg_offset = get_umc_v12_0_reg_offset(adev, node_inst, umc_inst, ch_inst); @@ -357,8 +321,19 @@ static int umc_v12_0_query_error_address(struct amdgpu_device *adev, err_addr = REG_GET_FIELD(err_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr); - umc_v12_0_convert_error_address(adev, err_data, err_addr, - ch_inst, umc_inst, node_inst); + if (!adev->aid_mask && + adev->smuio.funcs && + adev->smuio.funcs->get_socket_id) + addr_in.ma.socket_id = adev->smuio.funcs->get_socket_id(adev); + else + addr_in.ma.socket_id = 0; + + addr_in.ma.err_addr = err_addr; + addr_in.ma.ch_inst = ch_inst; + addr_in.ma.umc_inst = umc_inst; + addr_in.ma.node_inst = node_inst; + + umc_v12_0_convert_error_address(adev, err_data, &addr_in); } /* clear umc status */ @@ -401,13 +376,20 @@ static int umc_v12_0_err_cnt_init_per_channel(struct amdgpu_device *adev, return 0; } +#ifdef TO_BE_REMOVED static void umc_v12_0_ecc_info_query_ras_error_count(struct amdgpu_device *adev, void *ras_error_status) { + struct ras_query_context qctx; + + memset(&qctx, 0, sizeof(qctx)); + qctx.event_id = amdgpu_ras_acquire_event_id(adev, amdgpu_ras_intr_triggered() ? + RAS_EVENT_TYPE_ISR : RAS_EVENT_TYPE_INVALID); + amdgpu_mca_smu_log_ras_error(adev, - AMDGPU_RAS_BLOCK__UMC, AMDGPU_MCA_ERROR_TYPE_CE, ras_error_status); + AMDGPU_RAS_BLOCK__UMC, AMDGPU_MCA_ERROR_TYPE_CE, ras_error_status, &qctx); amdgpu_mca_smu_log_ras_error(adev, - AMDGPU_RAS_BLOCK__UMC, AMDGPU_MCA_ERROR_TYPE_UE, ras_error_status); + AMDGPU_RAS_BLOCK__UMC, AMDGPU_MCA_ERROR_TYPE_UE, ras_error_status, &qctx); } static void umc_v12_0_ecc_info_query_ras_error_address(struct amdgpu_device *adev, @@ -418,12 +400,16 @@ static void umc_v12_0_ecc_info_query_ras_error_address(struct amdgpu_device *ade struct ras_err_info *err_info; struct ras_err_addr *mca_err_addr, *tmp; struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; + struct ta_ras_query_address_input addr_in; for_each_ras_error(err_node, err_data) { err_info = &err_node->err_info; if (list_empty(&err_info->err_addr_list)) continue; + addr_in.ma.node_inst = err_info->mcm_info.die_id; + addr_in.ma.socket_id = err_info->mcm_info.socket_id; + list_for_each_entry_safe(mca_err_addr, tmp, &err_info->err_addr_list, node) { mc_umc_status = mca_err_addr->err_status; if (mc_umc_status && @@ -439,6 +425,10 @@ static void umc_v12_0_ecc_info_query_ras_error_address(struct amdgpu_device *ade MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr); InstanceIdLo = REG_GET_FIELD(mca_ipid, MCMP1_IPIDT0, InstanceIdLo); + addr_in.ma.err_addr = err_addr; + addr_in.ma.ch_inst = MCA_IPID_LO_2_UMC_CH(InstanceIdLo); + addr_in.ma.umc_inst = MCA_IPID_LO_2_UMC_INST(InstanceIdLo); + dev_info(adev->dev, "UMC:IPID:0x%llx, aid:%d, inst:%d, ch:%d, err_addr:0x%llx\n", mca_ipid, err_info->mcm_info.die_id, @@ -447,10 +437,7 @@ static void umc_v12_0_ecc_info_query_ras_error_address(struct amdgpu_device *ade err_addr); umc_v12_0_convert_error_address(adev, - err_data, err_addr, - MCA_IPID_LO_2_UMC_CH(InstanceIdLo), - MCA_IPID_LO_2_UMC_INST(InstanceIdLo), - err_info->mcm_info.die_id); + err_data, &addr_in); } /* Delete error address node from list and free memory */ @@ -458,6 +445,7 @@ static void umc_v12_0_ecc_info_query_ras_error_address(struct amdgpu_device *ade } } } +#endif static bool umc_v12_0_check_ecc_err_status(struct amdgpu_device *adev, enum amdgpu_mca_error_type type, void *ras_error_status) @@ -498,43 +486,49 @@ const struct amdgpu_ras_block_hw_ops umc_v12_0_ras_hw_ops = { .query_ras_error_address = umc_v12_0_query_ras_error_address, }; -static int umc_v12_0_aca_bank_generate_report(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type, - struct aca_bank_report *report, void *data) +static int umc_v12_0_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) { struct amdgpu_device *adev = handle->adev; - u64 status; + struct aca_bank_info info; + enum aca_error_type err_type; + u64 status, count; + u32 ext_error_code; int ret; - ret = aca_bank_info_decode(bank, &report->info); + status = bank->regs[ACA_REG_IDX_STATUS]; + if (umc_v12_0_is_deferred_error(adev, status)) + err_type = ACA_ERROR_TYPE_DEFERRED; + else if (umc_v12_0_is_uncorrectable_error(adev, status)) + err_type = ACA_ERROR_TYPE_UE; + else if (umc_v12_0_is_correctable_error(adev, status)) + err_type = ACA_ERROR_TYPE_CE; + else + return 0; + + ret = aca_bank_info_decode(bank, &info); if (ret) return ret; - status = bank->regs[ACA_REG_IDX_STATUS]; - switch (type) { - case ACA_ERROR_TYPE_UE: - if (umc_v12_0_is_uncorrectable_error(adev, status)) { - report->count[type] = 1; - } - break; - case ACA_ERROR_TYPE_CE: - if (umc_v12_0_is_correctable_error(adev, status)) { - report->count[type] = 1; - } - break; - default: - return -EINVAL; - } + amdgpu_umc_update_ecc_status(adev, + bank->regs[ACA_REG_IDX_STATUS], + bank->regs[ACA_REG_IDX_IPID], + bank->regs[ACA_REG_IDX_ADDR]); - return 0; + ext_error_code = ACA_REG__STATUS__ERRORCODEEXT(status); + count = ext_error_code == 0 ? + ACA_REG__MISC0__ERRCNT(bank->regs[ACA_REG_IDX_MISC0]) : 1ULL; + + return aca_error_cache_log_bank_error(handle, &info, err_type, count); } static const struct aca_bank_ops umc_v12_0_aca_bank_ops = { - .aca_bank_generate_report = umc_v12_0_aca_bank_generate_report, + .aca_bank_parser = umc_v12_0_aca_bank_parser, }; const struct aca_info umc_v12_0_aca_info = { .hwip = ACA_HWIP_TYPE_UMC, - .mask = ACA_ERROR_UE_MASK | ACA_ERROR_CE_MASK, + .mask = ACA_ERROR_UE_MASK | ACA_ERROR_CE_MASK | ACA_ERROR_DEFERRED_MASK, .bank_ops = &umc_v12_0_aca_bank_ops, }; @@ -554,6 +548,152 @@ static int umc_v12_0_ras_late_init(struct amdgpu_device *adev, struct ras_common return 0; } +static int umc_v12_0_update_ecc_status(struct amdgpu_device *adev, + uint64_t status, uint64_t ipid, uint64_t addr) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + uint16_t hwid, mcatype; + struct ta_ras_query_address_input addr_in; + uint64_t page_pfn[UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL]; + uint64_t err_addr, hash_val = 0; + struct ras_ecc_err *ecc_err; + int count; + int ret; + + hwid = REG_GET_FIELD(ipid, MCMP1_IPIDT0, HardwareID); + mcatype = REG_GET_FIELD(ipid, MCMP1_IPIDT0, McaType); + + if ((hwid != MCA_UMC_HWID_V12_0) || (mcatype != MCA_UMC_MCATYPE_V12_0)) + return 0; + + if (!status) + return 0; + + if (!umc_v12_0_is_deferred_error(adev, status)) + return 0; + + err_addr = REG_GET_FIELD(addr, + MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr); + + dev_info(adev->dev, + "UMC:IPID:0x%llx, socket:%llu, aid:%llu, inst:%llu, ch:%llu, err_addr:0x%llx\n", + ipid, + MCA_IPID_2_SOCKET_ID(ipid), + MCA_IPID_2_DIE_ID(ipid), + MCA_IPID_2_UMC_INST(ipid), + MCA_IPID_2_UMC_CH(ipid), + err_addr); + + memset(page_pfn, 0, sizeof(page_pfn)); + + memset(&addr_in, 0, sizeof(addr_in)); + addr_in.ma.err_addr = err_addr; + addr_in.ma.ch_inst = MCA_IPID_2_UMC_CH(ipid); + addr_in.ma.umc_inst = MCA_IPID_2_UMC_INST(ipid); + addr_in.ma.node_inst = MCA_IPID_2_DIE_ID(ipid); + addr_in.ma.socket_id = MCA_IPID_2_SOCKET_ID(ipid); + + count = umc_v12_0_convert_err_addr(adev, + &addr_in, page_pfn, ARRAY_SIZE(page_pfn)); + if (count <= 0) { + dev_warn(adev->dev, "Fail to convert error address! count:%d\n", count); + return 0; + } + + ret = amdgpu_umc_build_pages_hash(adev, + page_pfn, count, &hash_val); + if (ret) { + dev_err(adev->dev, "Fail to build error pages hash\n"); + return ret; + } + + ecc_err = kzalloc(sizeof(*ecc_err), GFP_KERNEL); + if (!ecc_err) + return -ENOMEM; + + ecc_err->err_pages.pfn = kcalloc(count, sizeof(*ecc_err->err_pages.pfn), GFP_KERNEL); + if (!ecc_err->err_pages.pfn) { + kfree(ecc_err); + return -ENOMEM; + } + + memcpy(ecc_err->err_pages.pfn, page_pfn, count * sizeof(*ecc_err->err_pages.pfn)); + ecc_err->err_pages.count = count; + + ecc_err->hash_index = hash_val; + ecc_err->status = status; + ecc_err->ipid = ipid; + ecc_err->addr = addr; + + ret = amdgpu_umc_logs_ecc_err(adev, &con->umc_ecc_log.de_page_tree, ecc_err); + if (ret) { + if (ret == -EEXIST) + con->umc_ecc_log.de_updated = true; + else + dev_err(adev->dev, "Fail to log ecc error! ret:%d\n", ret); + + kfree(ecc_err->err_pages.pfn); + kfree(ecc_err); + return ret; + } + + con->umc_ecc_log.de_updated = true; + + return 0; +} + +static int umc_v12_0_fill_error_record(struct amdgpu_device *adev, + struct ras_ecc_err *ecc_err, void *ras_error_status) +{ + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; + uint32_t i = 0; + int ret = 0; + + if (!err_data || !ecc_err) + return -EINVAL; + + for (i = 0; i < ecc_err->err_pages.count; i++) { + ret = amdgpu_umc_fill_error_record(err_data, + ecc_err->addr, + ecc_err->err_pages.pfn[i] << AMDGPU_GPU_PAGE_SHIFT, + MCA_IPID_2_UMC_CH(ecc_err->ipid), + MCA_IPID_2_UMC_INST(ecc_err->ipid)); + if (ret) + break; + } + + err_data->de_count++; + + return ret; +} + +static void umc_v12_0_query_ras_ecc_err_addr(struct amdgpu_device *adev, + void *ras_error_status) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct ras_ecc_err *entries[MAX_ECC_NUM_PER_RETIREMENT]; + struct radix_tree_root *ecc_tree; + int new_detected, ret, i; + + ecc_tree = &con->umc_ecc_log.de_page_tree; + + mutex_lock(&con->umc_ecc_log.lock); + new_detected = radix_tree_gang_lookup_tag(ecc_tree, (void **)entries, + 0, ARRAY_SIZE(entries), UMC_ECC_NEW_DETECTED_TAG); + for (i = 0; i < new_detected; i++) { + if (!entries[i]) + continue; + + ret = umc_v12_0_fill_error_record(adev, entries[i], ras_error_status); + if (ret) { + dev_err(adev->dev, "Fail to fill umc error record, ret:%d\n", ret); + break; + } + radix_tree_tag_clear(ecc_tree, entries[i]->hash_index, UMC_ECC_NEW_DETECTED_TAG); + } + mutex_unlock(&con->umc_ecc_log.lock); +} + struct amdgpu_umc_ras umc_v12_0_ras = { .ras_block = { .hw_ops = &umc_v12_0_ras_hw_ops, @@ -561,8 +701,8 @@ struct amdgpu_umc_ras umc_v12_0_ras = { }, .err_cnt_init = umc_v12_0_err_cnt_init, .query_ras_poison_mode = umc_v12_0_query_ras_poison_mode, - .ecc_info_query_ras_error_count = umc_v12_0_ecc_info_query_ras_error_count, - .ecc_info_query_ras_error_address = umc_v12_0_ecc_info_query_ras_error_address, + .ecc_info_query_ras_error_address = umc_v12_0_query_ras_ecc_err_addr, .check_ecc_err_status = umc_v12_0_check_ecc_err_status, + .update_ecc_status = umc_v12_0_update_ecc_status, }; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h index 5973bfb14f..b497479385 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h @@ -55,83 +55,38 @@ #define UMC_V12_0_NA_MAP_PA_NUM 8 /* R13 bit shift should be considered, double the number */ #define UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL (UMC_V12_0_NA_MAP_PA_NUM * 2) -/* bank bits in MCA error address */ -#define UMC_V12_0_MCA_B0_BIT 6 -#define UMC_V12_0_MCA_B1_BIT 7 -#define UMC_V12_0_MCA_B2_BIT 8 -#define UMC_V12_0_MCA_B3_BIT 9 + /* column bits in SOC physical address */ #define UMC_V12_0_PA_C2_BIT 15 #define UMC_V12_0_PA_C4_BIT 21 /* row bits in SOC physical address */ #define UMC_V12_0_PA_R13_BIT 35 -/* channel index bits in SOC physical address */ -#define UMC_V12_0_PA_CH4_BIT 12 -#define UMC_V12_0_PA_CH5_BIT 13 -#define UMC_V12_0_PA_CH6_BIT 14 - -/* bank hash settings */ -#define UMC_V12_0_XOR_EN0 1 -#define UMC_V12_0_XOR_EN1 1 -#define UMC_V12_0_XOR_EN2 1 -#define UMC_V12_0_XOR_EN3 1 -#define UMC_V12_0_COL_XOR0 0x0 -#define UMC_V12_0_COL_XOR1 0x0 -#define UMC_V12_0_COL_XOR2 0x800 -#define UMC_V12_0_COL_XOR3 0x1000 -#define UMC_V12_0_ROW_XOR0 0x11111 -#define UMC_V12_0_ROW_XOR1 0x22222 -#define UMC_V12_0_ROW_XOR2 0x4444 -#define UMC_V12_0_ROW_XOR3 0x8888 - -/* channel hash settings */ -#define UMC_V12_0_HASH_4K 0 -#define UMC_V12_0_HASH_64K 1 -#define UMC_V12_0_HASH_2M 1 -#define UMC_V12_0_HASH_1G 1 -#define UMC_V12_0_HASH_1T 1 - -/* XOR some bits of PA into CH4~CH6 bits (bits 12~14 of PA), - * hash bit is only effective when related setting is enabled - */ -#define UMC_V12_0_CHANNEL_HASH_CH4(channel_idx, pa) ((((channel_idx) >> 5) & 0x1) ^ \ - (((pa) >> 20) & 0x1ULL & UMC_V12_0_HASH_64K) ^ \ - (((pa) >> 27) & 0x1ULL & UMC_V12_0_HASH_2M) ^ \ - (((pa) >> 34) & 0x1ULL & UMC_V12_0_HASH_1G) ^ \ - (((pa) >> 41) & 0x1ULL & UMC_V12_0_HASH_1T)) -#define UMC_V12_0_CHANNEL_HASH_CH5(channel_idx, pa) ((((channel_idx) >> 6) & 0x1) ^ \ - (((pa) >> 21) & 0x1ULL & UMC_V12_0_HASH_64K) ^ \ - (((pa) >> 28) & 0x1ULL & UMC_V12_0_HASH_2M) ^ \ - (((pa) >> 35) & 0x1ULL & UMC_V12_0_HASH_1G) ^ \ - (((pa) >> 42) & 0x1ULL & UMC_V12_0_HASH_1T)) -#define UMC_V12_0_CHANNEL_HASH_CH6(channel_idx, pa) ((((channel_idx) >> 4) & 0x1) ^ \ - (((pa) >> 19) & 0x1ULL & UMC_V12_0_HASH_64K) ^ \ - (((pa) >> 26) & 0x1ULL & UMC_V12_0_HASH_2M) ^ \ - (((pa) >> 33) & 0x1ULL & UMC_V12_0_HASH_1G) ^ \ - (((pa) >> 40) & 0x1ULL & UMC_V12_0_HASH_1T) ^ \ - (((pa) >> 47) & 0x1ULL & UMC_V12_0_HASH_4K)) -#define UMC_V12_0_SET_CHANNEL_HASH(channel_idx, pa) do { \ - (pa) &= ~(0x7ULL << UMC_V12_0_PA_CH4_BIT); \ - (pa) |= (UMC_V12_0_CHANNEL_HASH_CH4(channel_idx, pa) << UMC_V12_0_PA_CH4_BIT); \ - (pa) |= (UMC_V12_0_CHANNEL_HASH_CH5(channel_idx, pa) << UMC_V12_0_PA_CH5_BIT); \ - (pa) |= (UMC_V12_0_CHANNEL_HASH_CH6(channel_idx, pa) << UMC_V12_0_PA_CH6_BIT); \ - } while (0) + +#define MCA_UMC_HWID_V12_0 0x96 +#define MCA_UMC_MCATYPE_V12_0 0x0 #define MCA_IPID_LO_2_UMC_CH(_ipid_lo) (((((_ipid_lo) >> 20) & 0x1) * 4) + \ (((_ipid_lo) >> 12) & 0xF)) #define MCA_IPID_LO_2_UMC_INST(_ipid_lo) (((_ipid_lo) >> 21) & 0x7) +#define MCA_IPID_2_DIE_ID(ipid) ((REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdHi) >> 2) & 0x03) + +#define MCA_IPID_2_UMC_CH(ipid) \ + (MCA_IPID_LO_2_UMC_CH(REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdLo))) + +#define MCA_IPID_2_UMC_INST(ipid) \ + (MCA_IPID_LO_2_UMC_INST(REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdLo))) + +#define MCA_IPID_2_SOCKET_ID(ipid) \ + (((REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdLo) & 0x1) << 2) | \ + (REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdHi) & 0x03)) + bool umc_v12_0_is_deferred_error(struct amdgpu_device *adev, uint64_t mc_umc_status); bool umc_v12_0_is_uncorrectable_error(struct amdgpu_device *adev, uint64_t mc_umc_status); bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_status); typedef bool (*check_error_type_func)(struct amdgpu_device *adev, uint64_t mc_umc_status); -extern const uint32_t - umc_v12_0_channel_idx_tbl[] - [UMC_V12_0_UMC_INSTANCE_NUM] - [UMC_V12_0_CHANNEL_INSTANCE_NUM]; - extern struct amdgpu_umc_ras umc_v12_0_ras; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c index c4c7725771..a32f87992f 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c @@ -442,11 +442,6 @@ static void umc_v8_10_ecc_info_query_ras_error_address(struct amdgpu_device *ade umc_v8_10_ecc_info_query_error_address, ras_error_status); } -static void umc_v8_10_set_eeprom_table_version(struct amdgpu_ras_eeprom_table_header *hdr) -{ - hdr->version = RAS_TABLE_VER_V2_1; -} - const struct amdgpu_ras_block_hw_ops umc_v8_10_ras_hw_ops = { .query_ras_error_count = umc_v8_10_query_ras_error_count, .query_ras_error_address = umc_v8_10_query_ras_error_address, @@ -460,5 +455,4 @@ struct amdgpu_umc_ras umc_v8_10_ras = { .query_ras_poison_mode = umc_v8_10_query_ras_poison_mode, .ecc_info_query_ras_error_count = umc_v8_10_ecc_info_query_ras_error_count, .ecc_info_query_ras_error_address = umc_v8_10_ecc_info_query_ras_error_address, - .set_eeprom_table_version = umc_v8_10_set_eeprom_table_version, }; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c b/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c index a6006f231c..805d6662c8 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v3_1.c @@ -819,6 +819,8 @@ static const struct amd_ip_funcs uvd_v3_1_ip_funcs = { .soft_reset = uvd_v3_1_soft_reset, .set_clockgating_state = uvd_v3_1_set_clockgating_state, .set_powergating_state = uvd_v3_1_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version uvd_v3_1_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c index 1aa09ad7bb..3f19c606f4 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c @@ -769,6 +769,8 @@ static const struct amd_ip_funcs uvd_v4_2_ip_funcs = { .soft_reset = uvd_v4_2_soft_reset, .set_clockgating_state = uvd_v4_2_set_clockgating_state, .set_powergating_state = uvd_v4_2_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c index f8b229b754..efd903c21d 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c @@ -877,6 +877,8 @@ static const struct amd_ip_funcs uvd_v5_0_ip_funcs = { .set_clockgating_state = uvd_v5_0_set_clockgating_state, .set_powergating_state = uvd_v5_0_set_powergating_state, .get_clockgating_state = uvd_v5_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c index a9a6880f44..495de50684 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c @@ -1545,6 +1545,8 @@ static const struct amd_ip_funcs uvd_v6_0_ip_funcs = { .set_clockgating_state = uvd_v6_0_set_clockgating_state, .set_powergating_state = uvd_v6_0_set_powergating_state, .get_clockgating_state = uvd_v6_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs uvd_v6_0_ring_phys_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c index a08e7abca4..66fada199b 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c @@ -626,6 +626,8 @@ static const struct amd_ip_funcs vce_v2_0_ip_funcs = { .soft_reset = vce_v2_0_soft_reset, .set_clockgating_state = vce_v2_0_set_clockgating_state, .set_powergating_state = vce_v2_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs vce_v2_0_ring_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index f4760748d3..32517c364c 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -913,6 +913,8 @@ static const struct amd_ip_funcs vce_v3_0_ip_funcs = { .set_clockgating_state = vce_v3_0_set_clockgating_state, .set_powergating_state = vce_v3_0_set_powergating_state, .get_clockgating_state = vce_v3_0_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs vce_v3_0_ring_phys_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c index aaceecd558..cb253bd3a2 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c @@ -1902,6 +1902,8 @@ static const struct amd_ip_funcs vcn_v1_0_ip_funcs = { .post_soft_reset = NULL /* vcn_v1_0_post_soft_reset */, .set_clockgating_state = vcn_v1_0_set_clockgating_state, .set_powergating_state = vcn_v1_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; /* diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c index e357d8cf0c..f18fd61c43 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c @@ -2008,6 +2008,8 @@ static const struct amd_ip_funcs vcn_v2_0_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = vcn_v2_0_set_clockgating_state, .set_powergating_state = vcn_v2_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ring_funcs vcn_v2_0_dec_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index 1cd8a94b0f..baec14bde2 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -1901,6 +1901,8 @@ static const struct amd_ip_funcs vcn_v2_5_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = vcn_v2_5_set_clockgating_state, .set_powergating_state = vcn_v2_5_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amd_ip_funcs vcn_v2_6_ip_funcs = { @@ -1921,6 +1923,8 @@ static const struct amd_ip_funcs vcn_v2_6_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = vcn_v2_5_set_clockgating_state, .set_powergating_state = vcn_v2_5_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version vcn_v2_5_ip_block = diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c index 8f82fb887e..6b31cf4b8a 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c @@ -359,6 +359,7 @@ static int vcn_v3_0_hw_init(void *handle) } } + return 0; done: if (!r) DRM_INFO("VCN decode and encode initialized successfully(under %s).\n", @@ -2230,6 +2231,8 @@ static const struct amd_ip_funcs vcn_v3_0_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = vcn_v3_0_set_clockgating_state, .set_powergating_state = vcn_v3_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version vcn_v3_0_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c index 832d15f7b5..9a33d3d000 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c @@ -288,6 +288,7 @@ static int vcn_v4_0_hw_init(void *handle) } } + return 0; done: if (!r) DRM_INFO("VCN decode and encode initialized successfully(under %s).\n", @@ -1052,6 +1053,9 @@ static int vcn_v4_0_start(struct amdgpu_device *adev) amdgpu_dpm_enable_uvd(adev, true); for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + if (adev->vcn.harvest_config & (1 << i)) + continue; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) { @@ -1505,6 +1509,9 @@ static int vcn_v4_0_stop(struct amdgpu_device *adev) int i, r = 0; for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + if (adev->vcn.harvest_config & (1 << i)) + continue; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF; @@ -2130,6 +2137,8 @@ static const struct amd_ip_funcs vcn_v4_0_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = vcn_v4_0_set_clockgating_state, .set_powergating_state = vcn_v4_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version vcn_v4_0_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c index 203fa98832..2279d8fce0 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c @@ -1660,6 +1660,8 @@ static const struct amd_ip_funcs vcn_v4_0_3_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = vcn_v4_0_3_set_clockgating_state, .set_powergating_state = vcn_v4_0_3_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version vcn_v4_0_3_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c index 501e53e69f..30e80c6f11 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c @@ -237,6 +237,7 @@ static int vcn_v4_0_5_hw_init(void *handle) goto done; } + return 0; done: if (!r) DRM_INFO("VCN decode and encode initialized successfully(under %s).\n", @@ -963,6 +964,9 @@ static int vcn_v4_0_5_start(struct amdgpu_device *adev) amdgpu_dpm_enable_uvd(adev, true); for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + if (adev->vcn.harvest_config & (1 << i)) + continue; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) { @@ -1167,6 +1171,9 @@ static int vcn_v4_0_5_stop(struct amdgpu_device *adev) int i, r = 0; for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + if (adev->vcn.harvest_config & (1 << i)) + continue; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF; @@ -1752,6 +1759,8 @@ static const struct amd_ip_funcs vcn_v4_0_5_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = vcn_v4_0_5_set_clockgating_state, .set_powergating_state = vcn_v4_0_5_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version vcn_v4_0_5_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c index bc60c554eb..fbd3f7a582 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c @@ -95,7 +95,7 @@ static int vcn_v5_0_0_sw_init(void *handle) return r; for (i = 0; i < adev->vcn.num_vcn_inst; i++) { - volatile struct amdgpu_vcn4_fw_shared *fw_shared; + volatile struct amdgpu_vcn5_fw_shared *fw_shared; if (adev->vcn.harvest_config & (1 << i)) continue; @@ -154,7 +154,7 @@ static int vcn_v5_0_0_sw_fini(void *handle) if (drm_dev_enter(adev_to_drm(adev), &idx)) { for (i = 0; i < adev->vcn.num_vcn_inst; i++) { - volatile struct amdgpu_vcn4_fw_shared *fw_shared; + volatile struct amdgpu_vcn5_fw_shared *fw_shared; if (adev->vcn.harvest_config & (1 << i)) continue; @@ -203,6 +203,7 @@ static int vcn_v5_0_0_hw_init(void *handle) goto done; } + return 0; done: if (!r) DRM_INFO("VCN decode and encode initialized successfully(under %s).\n", @@ -334,7 +335,7 @@ static void vcn_v5_0_0_mc_resume(struct amdgpu_device *adev, int inst) upper_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr)); WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_OFFSET0, 0); WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_SIZE0, - AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared))); + AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn5_fw_shared))); } /** @@ -438,7 +439,7 @@ static void vcn_v5_0_0_mc_resume_dpg_mode(struct amdgpu_device *adev, int inst_i VCN, inst_idx, regUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect); WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET( VCN, inst_idx, regUVD_VCPU_NONCACHE_SIZE0), - AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)), 0, indirect); + AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn5_fw_shared)), 0, indirect); /* VCN global tiling registers */ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET( @@ -615,7 +616,7 @@ static void vcn_v5_0_0_enable_clock_gating(struct amdgpu_device *adev, int inst) */ static int vcn_v5_0_0_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, bool indirect) { - volatile struct amdgpu_vcn4_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr; + volatile struct amdgpu_vcn5_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr; struct amdgpu_ring *ring; uint32_t tmp; @@ -712,7 +713,7 @@ static int vcn_v5_0_0_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, b */ static int vcn_v5_0_0_start(struct amdgpu_device *adev) { - volatile struct amdgpu_vcn4_fw_shared *fw_shared; + volatile struct amdgpu_vcn5_fw_shared *fw_shared; struct amdgpu_ring *ring; uint32_t tmp; int i, j, k, r; @@ -721,6 +722,9 @@ static int vcn_v5_0_0_start(struct amdgpu_device *adev) amdgpu_dpm_enable_uvd(adev, true); for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + if (adev->vcn.harvest_config & (1 << i)) + continue; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) { @@ -893,11 +897,14 @@ static void vcn_v5_0_0_stop_dpg_mode(struct amdgpu_device *adev, int inst_idx) */ static int vcn_v5_0_0_stop(struct amdgpu_device *adev) { - volatile struct amdgpu_vcn4_fw_shared *fw_shared; + volatile struct amdgpu_vcn5_fw_shared *fw_shared; uint32_t tmp; int i, r = 0; for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + if (adev->vcn.harvest_config & (1 << i)) + continue; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF; @@ -1328,6 +1335,8 @@ static const struct amd_ip_funcs vcn_v5_0_0_ip_funcs = { .post_soft_reset = NULL, .set_clockgating_state = vcn_v5_0_0_set_clockgating_state, .set_powergating_state = vcn_v5_0_0_set_powergating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; const struct amdgpu_ip_block_version vcn_v5_0_0_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 1a98812981..d39c670f62 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -897,7 +897,7 @@ static int vi_asic_pci_config_reset(struct amdgpu_device *adev) return r; } -static bool vi_asic_supports_baco(struct amdgpu_device *adev) +static int vi_asic_supports_baco(struct amdgpu_device *adev) { switch (adev->asic_type) { case CHIP_FIJI: @@ -908,14 +908,14 @@ static bool vi_asic_supports_baco(struct amdgpu_device *adev) case CHIP_TOPAZ: return amdgpu_dpm_is_baco_supported(adev); default: - return false; + return 0; } } static enum amd_reset_method vi_asic_reset_method(struct amdgpu_device *adev) { - bool baco_reset; + int baco_reset; if (amdgpu_reset_method == AMD_RESET_METHOD_LEGACY || amdgpu_reset_method == AMD_RESET_METHOD_BACO) @@ -935,7 +935,7 @@ vi_asic_reset_method(struct amdgpu_device *adev) baco_reset = amdgpu_dpm_is_baco_supported(adev); break; default: - baco_reset = false; + baco_reset = 0; break; } @@ -2058,6 +2058,8 @@ static const struct amd_ip_funcs vi_common_ip_funcs = { .set_clockgating_state = vi_common_set_clockgating_state, .set_powergating_state = vi_common_set_powergating_state, .get_clockgating_state = vi_common_get_clockgating_state, + .dump_ip_state = NULL, + .print_ip_state = NULL, }; static const struct amdgpu_ip_block_version vi_common_ip_block = |