diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
88 files changed, 2134 insertions, 983 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig b/drivers/gpu/drm/amd/amdgpu/Kconfig index 22d88f8ef527..b0365cc1374e 100644 --- a/drivers/gpu/drm/amd/amdgpu/Kconfig +++ b/drivers/gpu/drm/amd/amdgpu/Kconfig @@ -2,13 +2,15 @@ config DRM_AMDGPU tristate "AMD GPU" - depends on DRM && PCI && MMU + depends on DRM + depends on DRM_DISPLAY_DP_HELPER + depends on DRM_DISPLAY_HDCP_HELPER + depends on DRM_DISPLAY_HDMI_HELPER + depends on DRM_DISPLAY_HELPER + depends on MMU + depends on PCI depends on !UML select FW_LOADER - select DRM_DISPLAY_DP_HELPER - select DRM_DISPLAY_HDMI_HELPER - select DRM_DISPLAY_HDCP_HELPER - select DRM_DISPLAY_HELPER select DRM_KMS_HELPER select DRM_SCHED select DRM_TTM diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 4536c8ad0e11..1f6b56ec99f6 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -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 9c62552bec34..e0d7f4ee7e16 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -210,6 +210,7 @@ extern int amdgpu_async_gfx_ring; extern int amdgpu_mcbp; extern int amdgpu_discovery; extern int amdgpu_mes; +extern int amdgpu_mes_log_enable; extern int amdgpu_mes_kiq; extern int amdgpu_noretry; extern int amdgpu_force_asic_type; @@ -605,7 +606,7 @@ 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 */ @@ -1407,7 +1408,8 @@ 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 493982f94649..c50202215f6b 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 2da50e095883..5ef6b745f222 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_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index 35dd6effa9a3..3b4591f554f1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -748,7 +748,7 @@ bool amdgpu_amdkfd_is_fed(struct amdgpu_device *adev) } 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); } @@ -769,12 +769,20 @@ 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 0ef223c2affb..c51954c9052e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -336,12 +336,13 @@ 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); 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_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index df58a6a1a67e..2131de36e3da 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -1854,6 +1854,7 @@ err_node_allow: err_bo_create: amdgpu_amdkfd_unreserve_mem_limit(adev, aligned_size, flags, xcp_id); err_reserve_limit: + amdgpu_sync_free(&(*mem)->sync); mutex_destroy(&(*mem)->lock); if (gobj) drm_gem_object_put(gobj); @@ -2900,13 +2901,12 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu * amdgpu_sync_create(&sync_obj); - /* Validate BOs and map them to GPUVM (update VM page tables). */ + /* Validate BOs managed by KFD */ list_for_each_entry(mem, &process_info->kfd_bo_list, validate_list) { struct amdgpu_bo *bo = mem->bo; uint32_t domain = mem->domain; - struct kfd_mem_attachment *attachment; struct dma_resv_iter cursor; struct dma_fence *fence; @@ -2931,6 +2931,25 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu * goto validate_map_fail; } } + } + + if (failed_size) + pr_debug("0x%lx/0x%lx in system\n", failed_size, total_size); + + /* Validate PDs, PTs and evicted DMABuf imports last. Otherwise BO + * validations above would invalidate DMABuf imports again. + */ + ret = process_validate_vms(process_info, &exec.ticket); + if (ret) { + pr_debug("Validating VMs failed, ret: %d\n", ret); + goto validate_map_fail; + } + + /* Update mappings managed by KFD. */ + list_for_each_entry(mem, &process_info->kfd_bo_list, + validate_list) { + struct kfd_mem_attachment *attachment; + list_for_each_entry(attachment, &mem->attachments, list) { if (!attachment->is_mapped) continue; @@ -2947,18 +2966,6 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu * } } - if (failed_size) - pr_debug("0x%lx/0x%lx in system\n", failed_size, total_size); - - /* Validate PDs, PTs and evicted DMABuf imports last. Otherwise BO - * validations above would invalidate DMABuf imports again. - */ - ret = process_validate_vms(process_info, &exec.ticket); - if (ret) { - pr_debug("Validating VMs failed, ret: %d\n", ret); - goto validate_map_fail; - } - /* Update mappings not managed by KFD */ list_for_each_entry(peer_vm, &process_info->vm_list_head, vm_list_node) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c index 6857c586ded7..a6d64bdbbb14 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; }; /* @@ -872,6 +873,10 @@ 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_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 0a4b09709cfb..ec888fc6ead8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -819,7 +819,7 @@ retry: p->bytes_moved += ctx.bytes_moved; if (!amdgpu_gmc_vram_full_visible(&adev->gmc) && - amdgpu_bo_in_cpu_visible_vram(bo)) + amdgpu_res_cpu_visible(adev, bo->tbo.resource)) p->bytes_moved_vis += ctx.bytes_moved; if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) { 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 000000000000..64fe564b8036 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c @@ -0,0 +1,346 @@ +// 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); + + /* 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 000000000000..52459512cb2b --- /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 5dc24c971b41..f3b7cb18fd46 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 @@ -1455,7 +1535,7 @@ 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 && @@ -4069,6 +4149,13 @@ 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) { @@ -4135,18 +4222,22 @@ int amdgpu_device_init(struct amdgpu_device *adev, adev->ip_blocks[i].status.hw = true; } } + } else if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 10) && + !amdgpu_device_has_display_hardware(adev)) { + r = psp_gpu_reset(adev); } else { - tmp = amdgpu_reset_method; - /* It should do a default reset when loading or reloading the driver, - * regardless of the module parameter reset_method. - */ - amdgpu_reset_method = AMD_RESET_METHOD_NONE; - r = amdgpu_asic_reset(adev); - amdgpu_reset_method = tmp; - if (r) { - dev_err(adev->dev, "asic reset on init failed\n"); - goto failed; - } + tmp = amdgpu_reset_method; + /* It should do a default reset when loading or reloading the driver, + * regardless of the module parameter reset_method. + */ + amdgpu_reset_method = AMD_RESET_METHOD_NONE; + r = amdgpu_asic_reset(adev); + amdgpu_reset_method = tmp; + } + + if (r) { + dev_err(adev->dev, "asic reset on init failed\n"); + goto failed; } } @@ -4539,6 +4630,8 @@ int amdgpu_device_prepare(struct drm_device *dev) if (r) goto unprepare; + flush_delayed_work(&adev->gfx.gfx_off_delay_work); + for (i = 0; i < adev->num_ip_blocks; i++) { if (!adev->ip_blocks[i].status.valid) continue; @@ -4968,12 +5061,15 @@ 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 */ @@ -5261,7 +5357,9 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle, /* Try reset handler method first */ tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device, reset_list); - amdgpu_reset_reg_dumps(tmp_adev); + + if (!test_bit(AMDGPU_SKIP_COREDUMP, &reset_context->flags)) + amdgpu_reset_reg_dumps(tmp_adev); reset_context->reset_device_list = device_list_handle; r = amdgpu_reset_perform_reset(tmp_adev, reset_context); @@ -5334,7 +5432,8 @@ 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"); @@ -5532,6 +5631,23 @@ 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 * @@ -5603,6 +5719,12 @@ 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); @@ -5685,11 +5807,12 @@ 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.*/ @@ -5768,6 +5891,7 @@ 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); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index a07e4b87d4ca..0e31bdb4b7cb 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) { @@ -1896,6 +1900,9 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &smu_v13_0_ip_block); 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: @@ -2237,6 +2244,7 @@ static int amdgpu_discovery_set_umsch_mm_ip_blocks(struct amdgpu_device *adev) { switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) { case IP_VERSION(4, 0, 5): + case IP_VERSION(4, 0, 6): if (amdgpu_umsch_mm & 0x1) { amdgpu_device_ip_block_add(adev, &umsch_mm_v4_0_ip_block); adev->enable_umsch_mm = true; @@ -2676,6 +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_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 80b9642f2bc4..c512f70b8272 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -195,6 +195,7 @@ int amdgpu_async_gfx_ring = 1; int amdgpu_mcbp = -1; int amdgpu_discovery = -1; int amdgpu_mes; +int amdgpu_mes_log_enable = 0; int amdgpu_mes_kiq; int amdgpu_noretry = -1; int amdgpu_force_asic_type = -1; @@ -668,6 +669,15 @@ MODULE_PARM_DESC(mes, module_param_named(mes, amdgpu_mes, int, 0444); /** + * DOC: mes_log_enable (int) + * Enable Micro Engine Scheduler log. This is used to enable/disable MES internal log. + * (0 = disabled (default), 1 = enabled) + */ +MODULE_PARM_DESC(mes_log_enable, + "Enable Micro Engine Scheduler log (0 = disabled (default), 1 = enabled)"); +module_param_named(mes_log_enable, amdgpu_mes_log_enable, int, 0444); + +/** * DOC: mes_kiq (int) * Enable Micro Engine Scheduler KIQ. This is a new engine pipe for kiq. * (0 = disabled (default), 1 = enabled) @@ -2471,6 +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) { @@ -2734,7 +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); } @@ -2774,7 +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_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index 8fcf889ddce9..04a86dff71e6 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); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfxhub.h index c7b44aeb671b..103a837ccc71 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_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index 4b3000c21ef2..e4742b65032d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -304,12 +304,15 @@ static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job) dma_fence_set_error(finished, -ECANCELED); if (finished->error < 0) { - DRM_INFO("Skip scheduling IBs!\n"); + dev_dbg(adev->dev, "Skip scheduling IBs in ring(%s)", + ring->name); } else { r = amdgpu_ib_schedule(ring, job->num_ibs, job->ibs, job, &fence); if (r) - DRM_ERROR("Error scheduling IBs (%d)\n", r); + dev_err(adev->dev, + "Error scheduling IBs (%d) in ring(%s)", r, + ring->name); } job->job_run_counter++; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index a2df3025a754..a0ea6fe8d060 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 24ad4b97177b..0734490347db 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 b964110ed1e0..e5bf07ce3451 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 a98e03e0a51f..7d215f280aab 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c @@ -40,7 +40,6 @@ 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 +64,6 @@ 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; @@ -102,7 +100,10 @@ static int amdgpu_mes_event_log_init(struct amdgpu_device *adev) { int r; - r = amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE, + if (!amdgpu_mes_log_enable) + return 0; + + r = amdgpu_bo_create_kernel(adev, AMDGPU_MES_LOG_BUFFER_SIZE, PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT, &adev->mes.event_log_gpu_obj, &adev->mes.event_log_gpu_addr, @@ -653,7 +654,7 @@ 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) @@ -711,8 +712,7 @@ 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); @@ -766,8 +766,7 @@ 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); @@ -1471,7 +1470,7 @@ 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; @@ -1549,12 +1548,11 @@ static int amdgpu_debugfs_mes_event_log_show(struct seq_file *m, void *unused) uint32_t *mem = (uint32_t *)(adev->mes.event_log_cpu_addr); seq_hex_dump(m, "", DUMP_PREFIX_OFFSET, 32, 4, - mem, PAGE_SIZE, false); + mem, AMDGPU_MES_LOG_BUFFER_SIZE, false); return 0; } - DEFINE_SHOW_ATTRIBUTE(amdgpu_debugfs_mes_event_log); #endif @@ -1565,7 +1563,7 @@ void amdgpu_debugfs_mes_event_log_init(struct amdgpu_device *adev) #if defined(CONFIG_DEBUG_FS) struct drm_minor *minor = adev_to_drm(adev)->primary; struct dentry *root = minor->debugfs_root; - if (adev->enable_mes) + if (adev->enable_mes && amdgpu_mes_log_enable) debugfs_create_file("amdgpu_mes_event_log", 0444, root, adev, &amdgpu_debugfs_mes_event_log_fops); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h index 7d4f93fea937..6b3e1844eac5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h @@ -52,6 +52,7 @@ enum amdgpu_mes_priority_level { #define AMDGPU_MES_PROC_CTX_SIZE 0x1000 /* one page area */ #define AMDGPU_MES_GANG_CTX_SIZE 0x1000 /* one page area */ +#define AMDGPU_MES_LOG_BUFFER_SIZE 0x4000 /* Maximu log buffer size for MES */ struct amdgpu_mes_funcs; @@ -140,6 +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 { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h index 1ca9d4ed8063..95d676ee207f 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 010b0cb7693c..492aebc44e51 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 @@ -173,6 +174,12 @@ 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++; } @@ -595,8 +602,7 @@ 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->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 | @@ -617,8 +623,7 @@ int amdgpu_bo_create(struct amdgpu_device *adev, return r; if (!amdgpu_gmc_vram_full_visible(&adev->gmc) && - bo->tbo.resource->mem_type == TTM_PL_VRAM && - amdgpu_bo_in_cpu_visible_vram(bo)) + amdgpu_res_cpu_visible(adev, bo->tbo.resource)) amdgpu_cs_report_moved_bytes(adev, ctx.bytes_moved, ctx.bytes_moved); else @@ -628,7 +633,7 @@ 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; @@ -1272,23 +1277,25 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo, bool evict) void amdgpu_bo_get_memory(struct amdgpu_bo *bo, struct amdgpu_mem_stats *stats) { + struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); + struct ttm_resource *res = bo->tbo.resource; uint64_t size = amdgpu_bo_size(bo); struct drm_gem_object *obj; unsigned int domain; bool shared; /* Abort if the BO doesn't currently have a backing store */ - if (!bo->tbo.resource) + if (!res) return; obj = &bo->tbo.base; shared = drm_gem_object_is_shared_for_memory_stats(obj); - domain = amdgpu_mem_type_to_domain(bo->tbo.resource->mem_type); + domain = amdgpu_mem_type_to_domain(res->mem_type); switch (domain) { case AMDGPU_GEM_DOMAIN_VRAM: stats->vram += size; - if (amdgpu_bo_in_cpu_visible_vram(bo)) + if (amdgpu_res_cpu_visible(adev, bo->tbo.resource)) stats->visible_vram += size; if (shared) stats->vram_shared += size; @@ -1359,8 +1366,9 @@ 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); } @@ -1389,10 +1397,7 @@ vm_fault_t amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo) /* Remember that this BO was accessed by the CPU */ abo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED; - if (bo->resource->mem_type != TTM_PL_VRAM) - return 0; - - if (amdgpu_bo_in_cpu_visible_vram(abo)) + if (amdgpu_res_cpu_visible(adev, bo->resource)) return 0; /* Can't move a pinned BO to visible VRAM */ @@ -1415,7 +1420,7 @@ vm_fault_t amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo) /* this should never happen */ if (bo->resource->mem_type == TTM_PL_VRAM && - !amdgpu_bo_in_cpu_visible_vram(abo)) + !amdgpu_res_cpu_visible(adev, bo->resource)) return VM_FAULT_SIGBUS; ttm_bo_move_to_lru_tail_unlocked(bo); @@ -1579,6 +1584,7 @@ uint32_t amdgpu_bo_get_preferred_domain(struct amdgpu_device *adev, */ u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m) { + struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); struct dma_buf_attachment *attachment; struct dma_buf *dma_buf; const char *placement; @@ -1587,10 +1593,11 @@ u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m) if (dma_resv_trylock(bo->tbo.base.resv)) { unsigned int domain; + domain = amdgpu_mem_type_to_domain(bo->tbo.resource->mem_type); switch (domain) { case AMDGPU_GEM_DOMAIN_VRAM: - if (amdgpu_bo_in_cpu_visible_vram(bo)) + if (amdgpu_res_cpu_visible(adev, bo->tbo.resource)) placement = "VRAM VISIBLE"; else placement = "VRAM"; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index be679c42b0b8..fa03d9e4874c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -251,28 +251,6 @@ static inline u64 amdgpu_bo_mmap_offset(struct amdgpu_bo *bo) } /** - * amdgpu_bo_in_cpu_visible_vram - check if BO is (partly) in visible VRAM - */ -static inline bool amdgpu_bo_in_cpu_visible_vram(struct amdgpu_bo *bo) -{ - struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); - struct amdgpu_res_cursor cursor; - - if (!bo->tbo.resource || bo->tbo.resource->mem_type != TTM_PL_VRAM) - return false; - - amdgpu_res_first(bo->tbo.resource, 0, amdgpu_bo_size(bo), &cursor); - while (cursor.remaining) { - if (cursor.start < adev->gmc.visible_vram_size) - return true; - - amdgpu_res_next(&cursor, cursor.size); - } - - return false; -} - -/** * amdgpu_bo_explicit_sync - return whether the bo is explicitly synced */ static inline bool amdgpu_bo_explicit_sync(struct amdgpu_bo *bo) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index 94b310fdb719..4bd4602d11b1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -1053,6 +1053,11 @@ 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 +2265,15 @@ 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 +2631,8 @@ 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 +2662,8 @@ 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 +3289,12 @@ 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 ee16f134ae92..3635303e6548 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 8ebab6f22e5a..352ce16a0963 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1045,6 +1045,7 @@ 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 +1053,28 @@ 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 +1083,44 @@ 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 +1133,79 @@ 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); } } } @@ -1265,7 +1269,8 @@ 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 +1278,7 @@ 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 +1292,14 @@ 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 +1335,21 @@ 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); + 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_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 +1361,7 @@ 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 +1375,12 @@ 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 +1391,7 @@ 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 +2056,7 @@ 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); @@ -2384,6 +2399,19 @@ 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 +2421,21 @@ 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 +2480,6 @@ 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); } } @@ -2694,7 +2723,7 @@ static int amdgpu_ras_page_retirement_thread(void *param) atomic_dec(&con->page_retirement_req_cnt); amdgpu_umc_bad_page_polling_timeout(adev, - false, MAX_UMC_POISON_POLLING_TIME_ASYNC); + 0, MAX_UMC_POISON_POLLING_TIME_ASYNC); } return 0; @@ -3036,6 +3065,35 @@ 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 +3414,8 @@ 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 +3532,39 @@ 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); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index e0f8ce9d8440..8d26989c75c8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -64,6 +64,14 @@ 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 +427,21 @@ 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; +}; + struct amdgpu_ras { /* ras infrastructure */ /* for ras itself. */ @@ -479,6 +502,11 @@ struct amdgpu_ras { atomic_t page_retirement_req_cnt; /* 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 { @@ -879,4 +907,6 @@ 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); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index b12808c0c331..06a62a8a992e 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 381101d2bf05..50fcd86e1033 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 147100c27c2d..ea4873f6ccd1 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 60522963aaca..b11d190ece53 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_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c index 5505d646f43a..06f0a6534a94 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c @@ -524,46 +524,58 @@ static ssize_t amdgpu_debugfs_mqd_read(struct file *f, char __user *buf, { struct amdgpu_ring *ring = file_inode(f)->i_private; volatile u32 *mqd; - int r; + u32 *kbuf; + int r, i; uint32_t value, result; if (*pos & 3 || size & 3) return -EINVAL; - result = 0; + kbuf = kmalloc(ring->mqd_size, GFP_KERNEL); + if (!kbuf) + return -ENOMEM; r = amdgpu_bo_reserve(ring->mqd_obj, false); if (unlikely(r != 0)) - return r; + goto err_free; r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&mqd); - if (r) { - amdgpu_bo_unreserve(ring->mqd_obj); - return r; - } + if (r) + goto err_unreserve; + /* + * Copy to local buffer to avoid put_user(), which might fault + * and acquire mmap_sem, under reservation_ww_class_mutex. + */ + for (i = 0; i < ring->mqd_size/sizeof(u32); i++) + kbuf[i] = mqd[i]; + + amdgpu_bo_kunmap(ring->mqd_obj); + amdgpu_bo_unreserve(ring->mqd_obj); + + result = 0; while (size) { if (*pos >= ring->mqd_size) - goto done; + break; - value = mqd[*pos/4]; + value = kbuf[*pos/4]; r = put_user(value, (uint32_t *)buf); if (r) - goto done; + goto err_free; buf += 4; result += 4; size -= 4; *pos += 4; } -done: - amdgpu_bo_kunmap(ring->mqd_obj); - mqd = NULL; - amdgpu_bo_unreserve(ring->mqd_obj); - if (r) - return r; - + kfree(kbuf); return result; + +err_unreserve: + amdgpu_bo_unreserve(ring->mqd_obj); +err_free: + kfree(kbuf); + return r; } static const struct file_operations amdgpu_debugfs_mqd_fops = { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h index ff4435181055..ec9d12f85f39 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_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index fc418e670fda..6b48bcf53ce9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -133,7 +133,7 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo, } else if (!amdgpu_gmc_vram_full_visible(&adev->gmc) && !(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) && - amdgpu_bo_in_cpu_visible_vram(abo)) { + amdgpu_res_cpu_visible(adev, bo->resource)) { /* Try evicting to the CPU inaccessible part of VRAM * first, but only set GTT as busy placement, so this @@ -378,11 +378,12 @@ 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; } @@ -403,40 +404,55 @@ error: return r; } -/* - * amdgpu_mem_visible - Check that memory can be accessed by ttm_bo_move_memcpy +/** + * amdgpu_res_cpu_visible - Check that resource can be accessed by CPU + * @adev: amdgpu device + * @res: the resource to check * - * Called by amdgpu_bo_move() + * Returns: true if the full resource is CPU visible, false otherwise. */ -static bool amdgpu_mem_visible(struct amdgpu_device *adev, - struct ttm_resource *mem) +bool amdgpu_res_cpu_visible(struct amdgpu_device *adev, + struct ttm_resource *res) { - u64 mem_size = (u64)mem->size; struct amdgpu_res_cursor cursor; - u64 end; - if (mem->mem_type == TTM_PL_SYSTEM || - mem->mem_type == TTM_PL_TT) + if (!res) + return false; + + if (res->mem_type == TTM_PL_SYSTEM || res->mem_type == TTM_PL_TT || + res->mem_type == AMDGPU_PL_PREEMPT) return true; - if (mem->mem_type != TTM_PL_VRAM) + + if (res->mem_type != TTM_PL_VRAM) return false; - amdgpu_res_first(mem, 0, mem_size, &cursor); - end = cursor.start + cursor.size; + amdgpu_res_first(res, 0, res->size, &cursor); while (cursor.remaining) { + if ((cursor.start + cursor.size) >= adev->gmc.visible_vram_size) + return false; amdgpu_res_next(&cursor, cursor.size); + } - if (!cursor.remaining) - break; + return true; +} - /* ttm_resource_ioremap only supports contiguous memory */ - if (end != cursor.start) - return false; +/* + * amdgpu_res_copyable - Check that memory can be accessed by ttm_bo_move_memcpy + * + * Called by amdgpu_bo_move() + */ +static bool amdgpu_res_copyable(struct amdgpu_device *adev, + struct ttm_resource *mem) +{ + if (!amdgpu_res_cpu_visible(adev, mem)) + return false; - end = cursor.start + cursor.size; - } + /* ttm_resource_ioremap only supports contiguous memory */ + if (mem->mem_type == TTM_PL_VRAM && + !(mem->placement & TTM_PL_FLAG_CONTIGUOUS)) + return false; - return end <= adev->gmc.visible_vram_size; + return true; } /* @@ -529,8 +545,8 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict, if (r) { /* Check that all memory is CPU accessible */ - if (!amdgpu_mem_visible(adev, old_mem) || - !amdgpu_mem_visible(adev, new_mem)) { + if (!amdgpu_res_copyable(adev, old_mem) || + !amdgpu_res_copyable(adev, new_mem)) { pr_err("Move buffer fallback to memcpy unavailable\n"); return r; } @@ -557,7 +573,6 @@ static int amdgpu_ttm_io_mem_reserve(struct ttm_device *bdev, struct ttm_resource *mem) { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); - size_t bus_size = (size_t)mem->size; switch (mem->mem_type) { case TTM_PL_SYSTEM: @@ -568,9 +583,6 @@ static int amdgpu_ttm_io_mem_reserve(struct ttm_device *bdev, break; case TTM_PL_VRAM: mem->bus.offset = mem->start << PAGE_SHIFT; - /* check if it's visible */ - if ((mem->bus.offset + bus_size) > adev->gmc.visible_vram_size) - return -EINVAL; if (adev->mman.aper_base_kaddr && mem->placement & TTM_PL_FLAG_CONTIGUOUS) @@ -2215,6 +2227,71 @@ 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 65ec82141a8e..4f5e70ee9ad0 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; @@ -139,6 +137,9 @@ int amdgpu_vram_mgr_reserve_range(struct amdgpu_vram_mgr *mgr, int amdgpu_vram_mgr_query_page_status(struct amdgpu_vram_mgr *mgr, uint64_t start); +bool amdgpu_res_cpu_visible(struct amdgpu_device *adev, + struct ttm_resource *res); + int amdgpu_ttm_init(struct amdgpu_device *adev); void amdgpu_ttm_fini(struct amdgpu_device *adev); void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, @@ -155,6 +156,9 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, 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 619445760037..105d4de0613a 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 20436f81856a..f486510fc94c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -177,7 +177,7 @@ 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); @@ -186,9 +186,7 @@ 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); } @@ -196,7 +194,7 @@ 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 = { @@ -238,8 +236,7 @@ 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); } @@ -247,7 +244,7 @@ int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev, } 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 ret = AMDGPU_RAS_SUCCESS; @@ -311,7 +308,8 @@ 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) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index 26d2ae498daf..563b0249247e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -66,8 +66,6 @@ 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); }; struct amdgpu_umc_funcs { @@ -103,7 +101,7 @@ 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_process_ecc_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry); @@ -123,5 +121,5 @@ 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); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c index ab820cf52668..0df97c3e3a70 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c @@ -189,10 +189,13 @@ static void setup_vpe_queue(struct amdgpu_device *adev, mqd->rptr_val = 0; mqd->unmapped = 1; + if (adev->vpe.collaborate_mode) + memcpy(++mqd, test->mqd_data_cpu_addr, sizeof(struct MQD_INFO)); + qinfo->mqd_addr = test->mqd_data_gpu_addr; qinfo->csa_addr = test->ctx_data_gpu_addr + offsetof(struct umsch_mm_test_ctx_data, vpe_ctx_csa); - qinfo->doorbell_offset_0 = (adev->doorbell_index.vpe_ring + 1) << 1; + qinfo->doorbell_offset_0 = 0; qinfo->doorbell_offset_1 = 0; } @@ -287,7 +290,10 @@ static int submit_vpe_queue(struct amdgpu_device *adev, struct umsch_mm_test *te ring[5] = 0; mqd->wptr_val = (6 << 2); - // WDOORBELL32(adev->umsch_mm.agdb_index[CONTEXT_PRIORITY_LEVEL_NORMAL], mqd->wptr_val); + if (adev->vpe.collaborate_mode) + (++mqd)->wptr_val = (6 << 2); + + WDOORBELL32(adev->umsch_mm.agdb_index[CONTEXT_PRIORITY_LEVEL_NORMAL], mqd->wptr_val); for (i = 0; i < adev->usec_timeout; i++) { if (*fence == test_pattern) @@ -571,6 +577,7 @@ int amdgpu_umsch_mm_init_microcode(struct amdgpu_umsch_mm *umsch) switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) { case IP_VERSION(4, 0, 5): + case IP_VERSION(4, 0, 6): fw_name = "amdgpu/umsch_mm_4_0_0.bin"; break; default: @@ -750,6 +757,7 @@ static int umsch_mm_early_init(void *handle) switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) { case IP_VERSION(4, 0, 5): + case IP_VERSION(4, 0, 6): umsch_mm_v4_0_set_funcs(&adev->umsch_mm); break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.h index 8258a43a6236..5014b5af95fd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.h @@ -33,13 +33,6 @@ enum UMSCH_SWIP_ENGINE_TYPE { UMSCH_SWIP_ENGINE_TYPE_MAX }; -enum UMSCH_SWIP_AFFINITY_TYPE { - UMSCH_SWIP_AFFINITY_TYPE_ANY = 0, - UMSCH_SWIP_AFFINITY_TYPE_VCN0 = 1, - UMSCH_SWIP_AFFINITY_TYPE_VCN1 = 2, - UMSCH_SWIP_AFFINITY_TYPE_MAX -}; - enum UMSCH_CONTEXT_PRIORITY_LEVEL { CONTEXT_PRIORITY_LEVEL_IDLE = 0, CONTEXT_PRIORITY_LEVEL_NORMAL = 1, @@ -51,13 +44,15 @@ enum UMSCH_CONTEXT_PRIORITY_LEVEL { struct umsch_mm_set_resource_input { uint32_t vmid_mask_mm_vcn; uint32_t vmid_mask_mm_vpe; + uint32_t collaboration_mask_vpe; uint32_t logging_vmid; uint32_t engine_mask; union { struct { uint32_t disable_reset : 1; uint32_t disable_umsch_mm_log : 1; - uint32_t reserved : 30; + uint32_t use_rs64mem_for_proc_ctx_csa : 1; + uint32_t reserved : 29; }; uint32_t uint32_all; }; @@ -78,15 +73,18 @@ struct umsch_mm_add_queue_input { uint32_t doorbell_offset_1; enum UMSCH_SWIP_ENGINE_TYPE engine_type; uint32_t affinity; - enum UMSCH_SWIP_AFFINITY_TYPE affinity_type; uint64_t mqd_addr; uint64_t h_context; uint64_t h_queue; uint32_t vm_context_cntl; + uint32_t process_csa_array_index; + uint32_t context_csa_array_index; + struct { uint32_t is_context_suspended : 1; - uint32_t reserved : 31; + uint32_t collaboration_mode : 1; + uint32_t reserved : 30; }; }; @@ -94,6 +92,7 @@ struct umsch_mm_remove_queue_input { uint32_t doorbell_offset_0; uint32_t doorbell_offset_1; uint64_t context_csa_addr; + uint32_t context_csa_array_index; }; struct MQD_INFO { @@ -103,6 +102,7 @@ struct MQD_INFO { uint32_t wptr_val; uint32_t rptr_val; uint32_t unmapped; + uint32_t vmid; }; struct amdgpu_umsch_mm; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index 9c514a606a2f..bb85772b1374 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; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 7a4eae36778a..54ab51a4ada7 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 3f59b7b5523f..642f1fd287d8 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_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 4299ce386322..4e2391c83d7c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -886,6 +886,44 @@ 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 +954,8 @@ 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 +965,8 @@ 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 +986,9 @@ 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 +1071,18 @@ 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; @@ -1613,6 +1647,37 @@ static void amdgpu_vm_bo_insert_map(struct amdgpu_device *adev, trace_amdgpu_vm_bo_map(bo_va, mapping); } +/* Validate operation parameters to prevent potential abuse */ +static int amdgpu_vm_verify_parameters(struct amdgpu_device *adev, + struct amdgpu_bo *bo, + uint64_t saddr, + uint64_t offset, + uint64_t size) +{ + uint64_t tmp, lpfn; + + if (saddr & AMDGPU_GPU_PAGE_MASK + || offset & AMDGPU_GPU_PAGE_MASK + || size & AMDGPU_GPU_PAGE_MASK) + return -EINVAL; + + if (check_add_overflow(saddr, size, &tmp) + || check_add_overflow(offset, size, &tmp) + || size == 0 /* which also leads to end < begin */) + return -EINVAL; + + /* make sure object fit at this offset */ + if (bo && offset + size > amdgpu_bo_size(bo)) + return -EINVAL; + + /* Ensure last pfn not exceed max_pfn */ + lpfn = (saddr + size - 1) >> AMDGPU_GPU_PAGE_SHIFT; + if (lpfn >= adev->vm_manager.max_pfn) + return -EINVAL; + + return 0; +} + /** * amdgpu_vm_bo_map - map bo inside a vm * @@ -1639,21 +1704,14 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev, struct amdgpu_bo *bo = bo_va->base.bo; struct amdgpu_vm *vm = bo_va->base.vm; uint64_t eaddr; + int r; - /* validate the parameters */ - if (saddr & ~PAGE_MASK || offset & ~PAGE_MASK || size & ~PAGE_MASK) - return -EINVAL; - if (saddr + size <= saddr || offset + size <= offset) - return -EINVAL; - - /* make sure object fit at this offset */ - eaddr = saddr + size - 1; - if ((bo && offset + size > amdgpu_bo_size(bo)) || - (eaddr >= adev->vm_manager.max_pfn << AMDGPU_GPU_PAGE_SHIFT)) - return -EINVAL; + r = amdgpu_vm_verify_parameters(adev, bo, saddr, offset, size); + if (r) + return r; saddr /= AMDGPU_GPU_PAGE_SIZE; - eaddr /= AMDGPU_GPU_PAGE_SIZE; + eaddr = saddr + (size - 1) / AMDGPU_GPU_PAGE_SIZE; tmp = amdgpu_vm_it_iter_first(&vm->va, saddr, eaddr); if (tmp) { @@ -1706,17 +1764,9 @@ int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev, uint64_t eaddr; int r; - /* validate the parameters */ - if (saddr & ~PAGE_MASK || offset & ~PAGE_MASK || size & ~PAGE_MASK) - return -EINVAL; - if (saddr + size <= saddr || offset + size <= offset) - return -EINVAL; - - /* make sure object fit at this offset */ - eaddr = saddr + size - 1; - if ((bo && offset + size > amdgpu_bo_size(bo)) || - (eaddr >= adev->vm_manager.max_pfn << AMDGPU_GPU_PAGE_SHIFT)) - return -EINVAL; + r = amdgpu_vm_verify_parameters(adev, bo, saddr, offset, size); + if (r) + return r; /* Allocate all the needed memory */ mapping = kmalloc(sizeof(*mapping), GFP_KERNEL); @@ -1730,7 +1780,7 @@ int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev, } saddr /= AMDGPU_GPU_PAGE_SIZE; - eaddr /= AMDGPU_GPU_PAGE_SIZE; + eaddr = saddr + (size - 1) / AMDGPU_GPU_PAGE_SIZE; mapping->start = saddr; mapping->last = eaddr; @@ -1817,10 +1867,14 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev, struct amdgpu_bo_va_mapping *before, *after, *tmp, *next; LIST_HEAD(removed); uint64_t eaddr; + int r; + + r = amdgpu_vm_verify_parameters(adev, NULL, saddr, 0, size); + if (r) + return r; - eaddr = saddr + size - 1; saddr /= AMDGPU_GPU_PAGE_SIZE; - eaddr /= AMDGPU_GPU_PAGE_SIZE; + eaddr = saddr + (size - 1) / AMDGPU_GPU_PAGE_SIZE; /* Allocate all the needed memory */ before = kzalloc(sizeof(*before), GFP_KERNEL); @@ -2391,6 +2445,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, 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); @@ -2924,6 +2979,14 @@ 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 047ec1930d12..54d7da396de0 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 6e31621452de..3895bd7d176a 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 124389a6bf48..7fdd306a48a0 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); + } } /** @@ -972,10 +996,8 @@ 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 349416e176a1..66e8a016126b 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 000000000000..51cddfa3f1e8 --- /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_vpe.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c index 7a65a2b128ec..6695481f870f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c @@ -396,6 +396,12 @@ static int vpe_hw_init(void *handle) struct amdgpu_vpe *vpe = &adev->vpe; int ret; + /* Power on VPE */ + ret = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, + AMD_PG_STATE_UNGATE); + if (ret) + return ret; + ret = vpe_load_microcode(vpe); if (ret) return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index 8db880244324..e494f5bf136a 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; @@ -501,6 +502,9 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, if (place->flags & TTM_PL_FLAG_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; @@ -571,7 +575,7 @@ 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 +608,7 @@ 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 +916,7 @@ 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 0e04e42cf809..b256cbc2bc27 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 20d51f6c9bb8..dd2ec48cf5c2 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 1592c63b3099..a3bfc16de6d4 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 51a14f6d93bd..fb2b394bb9c5 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 d6f808acfb17..414ea3f560a7 100644 --- a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c +++ b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c @@ -62,6 +62,11 @@ void aqua_vanjaram_doorbell_index_init(struct amdgpu_device *adev) adev->doorbell_index.max_assignment = AMDGPU_DOORBELL_LAYOUT1_MAX_ASSIGNMENT << 1; } +static bool aqua_vanjaram_xcp_vcn_shared(struct amdgpu_device *adev) +{ + return (adev->xcp_mgr->num_xcps > adev->vcn.num_vcn_inst); +} + static void aqua_vanjaram_set_xcp_id(struct amdgpu_device *adev, uint32_t inst_idx, struct amdgpu_ring *ring) { @@ -87,7 +92,7 @@ static void aqua_vanjaram_set_xcp_id(struct amdgpu_device *adev, case AMDGPU_RING_TYPE_VCN_ENC: case AMDGPU_RING_TYPE_VCN_JPEG: ip_blk = AMDGPU_XCP_VCN; - if (adev->xcp_mgr->mode == AMDGPU_CPX_PARTITION_MODE) + if (aqua_vanjaram_xcp_vcn_shared(adev)) inst_mask = 1 << (inst_idx * 2); break; default: @@ -140,10 +145,12 @@ static int aqua_vanjaram_xcp_sched_list_update( aqua_vanjaram_xcp_gpu_sched_update(adev, ring, ring->xcp_id); - /* VCN is shared by two partitions under CPX MODE */ + /* VCN may be shared by two partitions under CPX MODE in certain + * configs. + */ if ((ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC || - ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG) && - adev->xcp_mgr->mode == AMDGPU_CPX_PARTITION_MODE) + ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG) && + aqua_vanjaram_xcp_vcn_shared(adev)) aqua_vanjaram_xcp_gpu_sched_update(adev, ring, ring->xcp_id + 1); } @@ -623,7 +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 */ @@ -635,7 +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/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index a3a643254d7a..fdbc26346b54 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; } } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index f90905ef32c7..d5b924222903 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -3964,7 +3964,7 @@ 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; @@ -4518,7 +4518,7 @@ 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; @@ -8317,7 +8317,7 @@ 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 */ } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index 1770e496c1b7..5dbfef49dd5d 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; @@ -1635,7 +1635,7 @@ static void gfx_v11_0_setup_rb(struct amdgpu_device *adev) active_rb_bitmap |= (0x3 << (i * rb_bitmap_width_per_sa)); } - active_rb_bitmap |= global_active_rb_bitmap; + active_rb_bitmap &= global_active_rb_bitmap; adev->gfx.config.backend_enable_mask = active_rb_bitmap; adev->gfx.config.num_rbs = hweight32(active_rb_bitmap); } @@ -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); @@ -5465,6 +5460,7 @@ static void gfx_v11_0_ring_emit_vm_flush(struct amdgpu_ring *ring, /* Make sure that we can't skip the SET_Q_MODE packets when the VM * changed in any way. */ + ring->set_q_mode_offs = 0; ring->set_q_mode_ptr = NULL; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 6f97a6d0e6d0..71b555993b7a 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)) 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 065b2bd5f5a6..3f4fd2f08163 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 b53c8fd4e8cf..fc33354f1d3d 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, }; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c index 22175da0e16a..d200310d1731 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 49aecdcee006..77df8c9cbad2 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_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 47b63a4ce68b..c4ec1358f3aa 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; diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c b/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c index ad4ad39f128f..c757ef99e3c5 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); diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c index b8da0fc29378..29ed78798070 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); diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c index 072c478665ad..81833395324a 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c @@ -111,7 +111,7 @@ static int mes_v11_0_submit_pkt_and_poll_completion(struct amdgpu_mes *mes, struct amdgpu_device *adev = mes->adev; struct amdgpu_ring *ring = &mes->ring; unsigned long flags; - signed long timeout = adev->usec_timeout; + signed long timeout = 3000000; /* 3000 ms */ if (amdgpu_emu_mode) { timeout *= 100; @@ -411,14 +411,47 @@ static int mes_v11_0_set_hw_resources(struct amdgpu_mes *mes) mes_set_hw_res_pkt.enable_reg_active_poll = 1; mes_set_hw_res_pkt.enable_level_process_quantum_check = 1; mes_set_hw_res_pkt.oversubscription_timer = 50; - mes_set_hw_res_pkt.enable_mes_event_int_logging = 1; - mes_set_hw_res_pkt.event_intr_history_gpu_mc_ptr = mes->event_log_gpu_addr; + if (amdgpu_mes_log_enable) { + mes_set_hw_res_pkt.enable_mes_event_int_logging = 1; + mes_set_hw_res_pkt.event_intr_history_gpu_mc_ptr = + mes->event_log_gpu_addr; + } 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, 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, @@ -1200,6 +1233,14 @@ 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"); @@ -1223,6 +1264,11 @@ 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; } diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c index c0fc44cdd658..7a1ff298417a 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 a2bd2c3b1ef9..0c7275bca8f7 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 77f5b55decf9..aba00d961627 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/psp_v14_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c index 78a95f8f370b..f08a32c18694 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c @@ -169,7 +169,8 @@ 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) @@ -177,6 +178,10 @@ 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) { @@ -653,6 +658,7 @@ 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_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c index 34237a1b1f2e..7bb0652159ac 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c @@ -1602,19 +1602,9 @@ static int sdma_v4_4_2_set_ecc_irq_state(struct amdgpu_device *adev, u32 sdma_cntl; sdma_cntl = RREG32_SDMA(type, regSDMA_CNTL); - switch (state) { - case AMDGPU_IRQ_STATE_DISABLE: - sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA_CNTL, - DRAM_ECC_INT_ENABLE, 0); - WREG32_SDMA(type, regSDMA_CNTL, sdma_cntl); - break; - /* sdma ecc interrupt is enabled by default - * driver doesn't need to do anything to - * enable the interrupt */ - case AMDGPU_IRQ_STATE_ENABLE: - default: - break; - } + sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA_CNTL, DRAM_ECC_INT_ENABLE, + state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0); + WREG32_SDMA(type, regSDMA_CNTL, sdma_cntl); return 0; } @@ -2189,35 +2179,39 @@ 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, &report->info); - if (ret) - return ret; + ret = aca_bank_info_decode(bank, &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; @@ -2236,7 +2230,7 @@ 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_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c index 361835a61f2e..67a4d8b1512b 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); diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c index 23e4ef4fff7c..67e179c7e347 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 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 000000000000..2a51a70d4846 --- /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 000000000000..6e617f832d90 --- /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 dec81ccf6240..c8abbf5da736 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -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; } } diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.c b/drivers/gpu/drm/amd/amdgpu/soc21.c index 581a3bd11481..43ca63fe85ac 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc21.c +++ b/drivers/gpu/drm/amd/amdgpu/soc21.c @@ -457,10 +457,8 @@ static bool soc21_need_full_reset(struct amdgpu_device *adev) { switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): - return amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC); case IP_VERSION(11, 0, 2): case IP_VERSION(11, 0, 3): - return false; default: return true; } @@ -722,7 +720,10 @@ static int soc21_common_early_init(void *handle) AMD_PG_SUPPORT_VCN | AMD_PG_SUPPORT_JPEG | AMD_PG_SUPPORT_GFX_PG; - adev->external_rev_id = adev->rev_id + 0x1; + if (adev->rev_id == 0) + adev->external_rev_id = 0x1; + else + adev->external_rev_id = adev->rev_id + 0x10; break; case IP_VERSION(11, 5, 1): adev->cg_flags = @@ -869,10 +870,35 @@ static int soc21_common_suspend(void *handle) return soc21_common_hw_fini(adev); } +static bool soc21_need_reset_on_resume(struct amdgpu_device *adev) +{ + u32 sol_reg1, sol_reg2; + + /* Will reset for the following suspend abort cases. + * 1) Only reset dGPU side. + * 2) S3 suspend got aborted and TOS is active. + */ + if (!(adev->flags & AMD_IS_APU) && adev->in_s3 && + !adev->suspend_complete) { + sol_reg1 = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81); + msleep(100); + sol_reg2 = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_81); + + return (sol_reg1 != sol_reg2); + } + + return false; +} + static int soc21_common_resume(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + if (soc21_need_reset_on_resume(adev)) { + dev_info(adev->dev, "S3 suspend aborted, resetting..."); + soc21_asic_reset(adev); + } + return soc21_common_hw_init(adev); } diff --git a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h index 056d4df8fa1f..3ac56a9645eb 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/umc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c index 77af4e25ff46..a0122b22eda4 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c @@ -28,28 +28,6 @@ #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, -}; - static inline uint64_t get_umc_v12_0_reg_offset(struct amdgpu_device *adev, uint32_t node_inst, uint32_t umc_inst, @@ -192,99 +170,22 @@ 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) -{ - bool result = 0; - int i; - - for (i = 0; i < 32; i++) - result = result ^ ((val >> i) & 0x1); - - return result; -} - -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; - - /* 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; -} - 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) + struct ras_err_data *err_data, + struct ta_ras_query_address_input *addr_in) { 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; - 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; + 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); - 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); + return; + } soc_pa = addr_out.pa.pa; bank = addr_out.pa.bank; @@ -309,7 +210,7 @@ static void umc_v12_0_convert_error_address(struct amdgpu_device *adev, "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); + retired_page, channel_index, addr_in->ma.umc_inst); /* shift R13 bit */ retired_page ^= (0x1ULL << UMC_V12_0_PA_R13_BIT); @@ -317,7 +218,7 @@ static void umc_v12_0_convert_error_address(struct amdgpu_device *adev, "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); + retired_page, channel_index, addr_in->ma.umc_inst); } } @@ -325,10 +226,11 @@ 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 +259,19 @@ 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 */ @@ -404,10 +317,16 @@ static int umc_v12_0_err_cnt_init_per_channel(struct amdgpu_device *adev, 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 +337,16 @@ 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 +362,10 @@ 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 +374,7 @@ 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 */ @@ -498,43 +422,44 @@ 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; - } + 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 0; + 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, }; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h index 5973bfb14fce..1d5f44dcffdd 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h @@ -55,67 +55,12 @@ #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_IPID_LO_2_UMC_CH(_ipid_lo) (((((_ipid_lo) >> 20) & 0x1) * 4) + \ (((_ipid_lo) >> 12) & 0xF)) @@ -127,11 +72,6 @@ bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_ 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 c4c77257710c..a32f87992f20 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/umsch_mm_v4_0.c b/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c index 8e7b763cfdb7..bd57896ab85d 100644 --- a/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c @@ -60,7 +60,7 @@ static int umsch_mm_v4_0_load_microcode(struct amdgpu_umsch_mm *umsch) umsch->cmd_buf_curr_ptr = umsch->cmd_buf_ptr; - if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 5)) { + if (amdgpu_ip_version(adev, VCN_HWIP, 0) >= IP_VERSION(4, 0, 5)) { WREG32_SOC15(VCN, 0, regUVD_IPX_DLDO_CONFIG, 1 << UVD_IPX_DLDO_CONFIG__ONO0_PWR_CONFIG__SHIFT); SOC15_WAIT_ON_RREG(VCN, 0, regUVD_IPX_DLDO_STATUS, @@ -225,6 +225,8 @@ static int umsch_mm_v4_0_ring_start(struct amdgpu_umsch_mm *umsch) WREG32_SOC15(VCN, 0, regVCN_UMSCH_RB_SIZE, ring->ring_size); + ring->wptr = 0; + data = RREG32_SOC15(VCN, 0, regVCN_RB_ENABLE); data &= ~(VCN_RB_ENABLE__AUDIO_RB_EN_MASK); WREG32_SOC15(VCN, 0, regVCN_RB_ENABLE, data); @@ -248,7 +250,7 @@ static int umsch_mm_v4_0_ring_stop(struct amdgpu_umsch_mm *umsch) data = REG_SET_FIELD(data, VCN_UMSCH_RB_DB_CTRL, EN, 0); WREG32_SOC15(VCN, 0, regVCN_UMSCH_RB_DB_CTRL, data); - if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 5)) { + if (amdgpu_ip_version(adev, VCN_HWIP, 0) >= IP_VERSION(4, 0, 5)) { WREG32_SOC15(VCN, 0, regUVD_IPX_DLDO_CONFIG, 2 << UVD_IPX_DLDO_CONFIG__ONO0_PWR_CONFIG__SHIFT); SOC15_WAIT_ON_RREG(VCN, 0, regUVD_IPX_DLDO_STATUS, @@ -271,6 +273,8 @@ static int umsch_mm_v4_0_set_hw_resources(struct amdgpu_umsch_mm *umsch) set_hw_resources.vmid_mask_mm_vcn = umsch->vmid_mask_mm_vcn; set_hw_resources.vmid_mask_mm_vpe = umsch->vmid_mask_mm_vpe; + set_hw_resources.collaboration_mask_vpe = + adev->vpe.collaborate_mode ? 0x3 : 0x0; set_hw_resources.engine_mask = umsch->engine_mask; set_hw_resources.vcn0_hqd_mask[0] = umsch->vcn0_hqd_mask; @@ -346,6 +350,7 @@ static int umsch_mm_v4_0_add_queue(struct amdgpu_umsch_mm *umsch, add_queue.h_queue = input_ptr->h_queue; add_queue.vm_context_cntl = input_ptr->vm_context_cntl; add_queue.is_context_suspended = input_ptr->is_context_suspended; + add_queue.collaboration_mode = adev->vpe.collaborate_mode ? 1 : 0; add_queue.api_status.api_completion_fence_addr = umsch->ring.fence_drv.gpu_addr; add_queue.api_status.api_completion_fence_value = ++umsch->ring.fence_drv.sync_seq; diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 1a98812981f4..2415355b037c 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; } |