diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
23 files changed, 616 insertions, 155 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 97cb56c791f3..dc2d53081e80 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -192,7 +192,6 @@ extern int amdgpu_emu_mode; extern uint amdgpu_smu_memory_pool_size; extern int amdgpu_smu_pptable_id; extern uint amdgpu_dc_feature_mask; -extern uint amdgpu_freesync_vid_mode; extern uint amdgpu_dc_debug_mask; extern uint amdgpu_dc_visual_confirm; extern uint amdgpu_dm_abm_level; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index a5a2b06c6588..2bca37044ad0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -706,7 +706,7 @@ int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev, atcs_input.size = sizeof(struct atcs_pref_req_input); /* client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */ - atcs_input.client_id = adev->pdev->devfn | (adev->pdev->bus->number << 8); + atcs_input.client_id = pci_dev_id(adev->pdev); atcs_input.valid_flags_mask = ATCS_VALID_FLAGS_MASK; atcs_input.flags = ATCS_WAIT_FOR_COMPLETION; if (advertise) @@ -776,7 +776,7 @@ int amdgpu_acpi_power_shift_control(struct amdgpu_device *adev, atcs_input.size = sizeof(struct atcs_pwr_shift_input); /* dGPU id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */ - atcs_input.dgpu_id = adev->pdev->devfn | (adev->pdev->bus->number << 8); + atcs_input.dgpu_id = pci_dev_id(adev->pdev); atcs_input.dev_acpi_state = dev_state; atcs_input.drv_state = drv_state; @@ -1141,7 +1141,7 @@ int amdgpu_acpi_get_tmr_info(struct amdgpu_device *adev, u64 *tmr_offset, if (!tmr_offset || !tmr_size) return -EINVAL; - bdf = (adev->pdev->bus->number << 8) | adev->pdev->devfn; + bdf = pci_dev_id(adev->pdev); dev_info = amdgpu_acpi_get_dev(bdf); if (!dev_info) return -ENOENT; @@ -1162,7 +1162,7 @@ int amdgpu_acpi_get_mem_info(struct amdgpu_device *adev, int xcc_id, if (!numa_info) return -EINVAL; - bdf = (adev->pdev->bus->number << 8) | adev->pdev->devfn; + bdf = pci_dev_id(adev->pdev); dev_info = amdgpu_acpi_get_dev(bdf); if (!dev_info) return -ENOENT; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c index e2fed6edbdd0..aff08321e976 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c @@ -163,12 +163,6 @@ static uint32_t kgd_gfx_aldebaran_set_address_watch( return watch_address_cntl; } -static uint32_t kgd_gfx_aldebaran_clear_address_watch(struct amdgpu_device *adev, - uint32_t watch_id) -{ - return 0; -} - const struct kfd2kgd_calls aldebaran_kfd2kgd = { .program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings, .set_pasid_vmid_mapping = kgd_gfx_v9_set_pasid_vmid_mapping, @@ -193,7 +187,7 @@ const struct kfd2kgd_calls aldebaran_kfd2kgd = { .set_wave_launch_trap_override = kgd_aldebaran_set_wave_launch_trap_override, .set_wave_launch_mode = kgd_aldebaran_set_wave_launch_mode, .set_address_watch = kgd_gfx_aldebaran_set_address_watch, - .clear_address_watch = kgd_gfx_aldebaran_clear_address_watch, + .clear_address_watch = kgd_gfx_v9_clear_address_watch, .get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times, .build_grace_period_packet_info = kgd_gfx_v9_build_grace_period_packet_info, .program_trap_handler_settings = kgd_gfx_v9_program_trap_handler_settings, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index a136fba9f29b..7d6daf8d2bfa 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -39,7 +39,6 @@ #include "amdgpu_xgmi.h" #include "kfd_priv.h" #include "kfd_smi_events.h" -#include <drm/ttm/ttm_tt.h> /* Userptr restore delay, just long enough to allow consecutive VM * changes to accumulate diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 91d7f9e4fefc..e77f048c99d8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -305,10 +305,16 @@ size_t amdgpu_device_aper_access(struct amdgpu_device *adev, loff_t pos, if (write) { memcpy_toio(addr, buf, count); + /* Make sure HDP write cache flush happens without any reordering + * after the system memory contents are sent over PCIe device + */ mb(); amdgpu_device_flush_hdp(adev, NULL); } else { amdgpu_device_invalidate_hdp(adev, NULL); + /* Make sure HDP read cache is invalidated before issuing a read + * to the PCIe device + */ mb(); memcpy_fromio(buf, addr, count); } @@ -3487,10 +3493,11 @@ static void amdgpu_device_set_mcbp(struct amdgpu_device *adev) { if (amdgpu_mcbp == 1) adev->gfx.mcbp = true; - - if ((adev->ip_versions[GC_HWIP][0] >= IP_VERSION(9, 0, 0)) && - (adev->ip_versions[GC_HWIP][0] < IP_VERSION(10, 0, 0)) && - adev->gfx.num_gfx_rings) + else if (amdgpu_mcbp == 0) + adev->gfx.mcbp = false; + else if ((adev->ip_versions[GC_HWIP][0] >= IP_VERSION(9, 0, 0)) && + (adev->ip_versions[GC_HWIP][0] < IP_VERSION(10, 0, 0)) && + adev->gfx.num_gfx_rings) adev->gfx.mcbp = true; if (amdgpu_sriov_vf(adev)) @@ -4153,6 +4160,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon) drm_fb_helper_set_suspend_unlocked(adev_to_drm(adev)->fb_helper, true); cancel_delayed_work_sync(&adev->delayed_init_work); + flush_delayed_work(&adev->gfx.gfx_off_delay_work); amdgpu_ras_suspend(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h index a4ecb802d2f9..09f6727e7c73 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h @@ -334,14 +334,14 @@ enum AMDGPU_DOORBELL_ASSIGNMENT_LAYOUT1 { AMDGPU_DOORBELL_LAYOUT1_sDMA_ENGINE_END = 0x19F, /* IH: 0x1A0 ~ 0x1AF */ AMDGPU_DOORBELL_LAYOUT1_IH = 0x1A0, - /* VCN: 0x1B0 ~ 0x1D4 */ + /* VCN: 0x1B0 ~ 0x1E8 */ AMDGPU_DOORBELL_LAYOUT1_VCN_START = 0x1B0, - AMDGPU_DOORBELL_LAYOUT1_VCN_END = 0x1D4, + AMDGPU_DOORBELL_LAYOUT1_VCN_END = 0x1E8, AMDGPU_DOORBELL_LAYOUT1_FIRST_NON_CP = AMDGPU_DOORBELL_LAYOUT1_sDMA_ENGINE_START, AMDGPU_DOORBELL_LAYOUT1_LAST_NON_CP = AMDGPU_DOORBELL_LAYOUT1_VCN_END, - AMDGPU_DOORBELL_LAYOUT1_MAX_ASSIGNMENT = 0x1D4, + AMDGPU_DOORBELL_LAYOUT1_MAX_ASSIGNMENT = 0x1E8, AMDGPU_DOORBELL_LAYOUT1_INVALID = 0xFFFF }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index d11fa7dec4c4..81edf66dbea8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -187,7 +187,6 @@ int amdgpu_mes_kiq; int amdgpu_noretry = -1; int amdgpu_force_asic_type = -1; int amdgpu_tmz = -1; /* auto */ -uint amdgpu_freesync_vid_mode; int amdgpu_reset_method = -1; /* auto */ int amdgpu_num_kcq = -1; int amdgpu_smartshift_bias; @@ -348,8 +347,9 @@ module_param_named(aspm, amdgpu_aspm, int, 0444); * Override for runtime power management control for dGPUs. The amdgpu driver can dynamically power down * the dGPUs when they are idle if supported. The default is -1 (auto enable). * Setting the value to 0 disables this functionality. + * Setting the value to -2 is auto enabled with power down when displays are attached. */ -MODULE_PARM_DESC(runpm, "PX runtime pm (2 = force enable with BAMACO, 1 = force enable with BACO, 0 = disable, -1 = auto)"); +MODULE_PARM_DESC(runpm, "PX runtime pm (2 = force enable with BAMACO, 1 = force enable with BACO, 0 = disable, -1 = auto, -2 = autowith displays)"); module_param_named(runpm, amdgpu_runtime_pm, int, 0444); /** @@ -872,32 +872,6 @@ MODULE_PARM_DESC(tmz, "Enable TMZ feature (-1 = auto (default), 0 = off, 1 = on) module_param_named(tmz, amdgpu_tmz, int, 0444); /** - * DOC: freesync_video (uint) - * Enable the optimization to adjust front porch timing to achieve seamless - * mode change experience when setting a freesync supported mode for which full - * modeset is not needed. - * - * The Display Core will add a set of modes derived from the base FreeSync - * video mode into the corresponding connector's mode list based on commonly - * used refresh rates and VRR range of the connected display, when users enable - * this feature. From the userspace perspective, they can see a seamless mode - * change experience when the change between different refresh rates under the - * same resolution. Additionally, userspace applications such as Video playback - * can read this modeset list and change the refresh rate based on the video - * frame rate. Finally, the userspace can also derive an appropriate mode for a - * particular refresh rate based on the FreeSync Mode and add it to the - * connector's mode list. - * - * Note: This is an experimental feature. - * - * The default value: 0 (off). - */ -MODULE_PARM_DESC( - freesync_video, - "Enable freesync modesetting optimization feature (0 = off (default), 1 = on)"); -module_param_named(freesync_video, amdgpu_freesync_vid_mode, uint, 0444); - -/** * DOC: reset_method (int) * GPU reset method (-1 = auto (default), 0 = legacy, 1 = mode0, 2 = mode1, 3 = mode2, 4 = baco) */ @@ -2523,24 +2497,26 @@ static int amdgpu_runtime_idle_check_display(struct device *dev) struct drm_connector_list_iter iter; int ret = 0; - /* XXX: Return busy if any displays are connected to avoid - * possible display wakeups after runtime resume due to - * hotplug events in case any displays were connected while - * the GPU was in suspend. Remove this once that is fixed. - */ - mutex_lock(&drm_dev->mode_config.mutex); - drm_connector_list_iter_begin(drm_dev, &iter); - drm_for_each_connector_iter(list_connector, &iter) { - if (list_connector->status == connector_status_connected) { - ret = -EBUSY; - break; + if (amdgpu_runtime_pm != -2) { + /* XXX: Return busy if any displays are connected to avoid + * possible display wakeups after runtime resume due to + * hotplug events in case any displays were connected while + * the GPU was in suspend. Remove this once that is fixed. + */ + mutex_lock(&drm_dev->mode_config.mutex); + drm_connector_list_iter_begin(drm_dev, &iter); + drm_for_each_connector_iter(list_connector, &iter) { + if (list_connector->status == connector_status_connected) { + ret = -EBUSY; + break; + } } - } - drm_connector_list_iter_end(&iter); - mutex_unlock(&drm_dev->mode_config.mutex); + drm_connector_list_iter_end(&iter); + mutex_unlock(&drm_dev->mode_config.mutex); - if (ret) - return ret; + if (ret) + return ret; + } if (adev->dc_enabled) { struct drm_crtc *crtc; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index c694b41f6461..7537f5aa76f0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -552,6 +552,41 @@ int amdgpu_fence_driver_sw_init(struct amdgpu_device *adev) } /** + * amdgpu_fence_need_ring_interrupt_restore - helper function to check whether + * fence driver interrupts need to be restored. + * + * @ring: ring that to be checked + * + * Interrupts for rings that belong to GFX IP don't need to be restored + * when the target power state is s0ix. + * + * Return true if need to restore interrupts, false otherwise. + */ +static bool amdgpu_fence_need_ring_interrupt_restore(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + bool is_gfx_power_domain = false; + + switch (ring->funcs->type) { + case AMDGPU_RING_TYPE_SDMA: + /* SDMA 5.x+ is part of GFX power domain so it's covered by GFXOFF */ + if (adev->ip_versions[SDMA0_HWIP][0] >= IP_VERSION(5, 0, 0)) + is_gfx_power_domain = true; + break; + case AMDGPU_RING_TYPE_GFX: + case AMDGPU_RING_TYPE_COMPUTE: + case AMDGPU_RING_TYPE_KIQ: + case AMDGPU_RING_TYPE_MES: + is_gfx_power_domain = true; + break; + default: + break; + } + + return !(adev->in_s0ix && is_gfx_power_domain); +} + +/** * amdgpu_fence_driver_hw_fini - tear down the fence driver * for all possible rings. * @@ -579,7 +614,8 @@ void amdgpu_fence_driver_hw_fini(struct amdgpu_device *adev) amdgpu_fence_driver_force_completion(ring); if (!drm_dev_is_unplugged(adev_to_drm(adev)) && - ring->fence_drv.irq_src) + ring->fence_drv.irq_src && + amdgpu_fence_need_ring_interrupt_restore(ring)) amdgpu_irq_put(adev, ring->fence_drv.irq_src, ring->fence_drv.irq_type); @@ -655,7 +691,8 @@ void amdgpu_fence_driver_hw_init(struct amdgpu_device *adev) continue; /* enable the interrupt */ - if (ring->fence_drv.irq_src) + if (ring->fence_drv.irq_src && + amdgpu_fence_need_ring_interrupt_restore(ring)) amdgpu_irq_get(adev, ring->fence_drv.irq_src, ring->fence_drv.irq_type); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index c76b6bfc4dab..2382921710ec 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -700,15 +700,8 @@ void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable) if (adev->gfx.gfx_off_req_count == 0 && !adev->gfx.gfx_off_state) { - /* If going to s2idle, no need to wait */ - if (adev->in_s0ix) { - if (!amdgpu_dpm_set_powergating_by_smu(adev, - AMD_IP_BLOCK_TYPE_GFX, true)) - adev->gfx.gfx_off_state = true; - } else { - schedule_delayed_work(&adev->gfx.gfx_off_delay_work, + schedule_delayed_work(&adev->gfx.gfx_off_delay_work, delay); - } } } else { if (adev->gfx.gfx_off_req_count == 0) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index 631c5ab3f7dc..99f4df133ed3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -1019,7 +1019,7 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) case AMDGPU_INFO_SENSOR_GPU_AVG_POWER: /* get average GPU power */ if (amdgpu_dpm_read_sensor(adev, - AMDGPU_PP_SENSOR_GPU_POWER, + AMDGPU_PP_SENSOR_GPU_AVG_POWER, (void *)&ui32, &ui32_size)) { return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index 3c7d9051b8c6..3384eb94fde0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -132,6 +132,7 @@ struct psp_funcs { int (*read_usbc_pd_fw)(struct psp_context *psp, uint32_t *fw_ver); int (*update_spirom)(struct psp_context *psp, uint64_t fw_pri_mc_addr); int (*vbflash_stat)(struct psp_context *psp); + int (*fatal_error_recovery_quirk)(struct psp_context *psp); }; struct ta_funcs { @@ -445,6 +446,10 @@ struct amdgpu_psp_funcs { ((psp)->funcs->vbflash_stat ? \ (psp)->funcs->vbflash_stat((psp)) : -EINVAL) +#define psp_fatal_error_recovery_quirk(psp) \ + ((psp)->funcs->fatal_error_recovery_quirk ? \ + (psp)->funcs->fatal_error_recovery_quirk((psp)) : 0) + extern const struct amd_ip_funcs psp_ip_funcs; extern const struct amdgpu_ip_block_version psp_v3_1_ip_block; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 886c7c6f1c3b..7689395e44fd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2064,6 +2064,8 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) if (ras->gpu_reset_flags & AMDGPU_RAS_GPU_RESET_MODE1_RESET) { ras->gpu_reset_flags &= ~AMDGPU_RAS_GPU_RESET_MODE1_RESET; set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); + + psp_fatal_error_recovery_quirk(&adev->psp); } } @@ -2970,10 +2972,6 @@ int amdgpu_ras_fini(struct amdgpu_device *adev) void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev) { - amdgpu_ras_check_supported(adev); - if (!adev->ras_hw_enabled) - return; - if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) { struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index 4287743e1212..4764d2171f92 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -158,6 +158,7 @@ static bool __is_ras_eeprom_supported(struct amdgpu_device *adev) case IP_VERSION(11, 0, 7): /* Sienna cichlid */ case IP_VERSION(13, 0, 0): case IP_VERSION(13, 0, 2): /* Aldebaran */ + case IP_VERSION(13, 0, 6): case IP_VERSION(13, 0, 10): return true; default: @@ -212,6 +213,7 @@ static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev, control->i2c_address = EEPROM_I2C_MADDR_0; return true; case IP_VERSION(13, 0, 0): + case IP_VERSION(13, 0, 6): case IP_VERSION(13, 0, 10): control->i2c_address = EEPROM_I2C_MADDR_4; return true; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c index b779ee4bbaa7..e1ee1c7117fb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c @@ -397,7 +397,7 @@ void amdgpu_sw_ring_ib_begin(struct amdgpu_ring *ring) struct amdgpu_ring_mux *mux = &adev->gfx.muxer; WARN_ON(!ring->is_sw_ring); - if (ring->hw_prio > AMDGPU_RING_PRIO_DEFAULT) { + if (adev->gfx.mcbp && ring->hw_prio > AMDGPU_RING_PRIO_DEFAULT) { if (amdgpu_mcbp_scan(mux) > 0) amdgpu_mcbp_trigger_preempt(mux); return; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index dea848bb55c1..4e51dce3aab5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -49,7 +49,6 @@ #include <drm/ttm/ttm_tt.h> #include <drm/amdgpu_drm.h> -#include <drm/drm_drv.h> #include "amdgpu.h" #include "amdgpu_object.h" @@ -2420,7 +2419,7 @@ static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf, struct page *p; void *ptr; - bytes = bytes < size ? bytes : size; + bytes = min(bytes, size); /* Translate the bus address to a physical address. If * the domain is NULL it means there is no IOMMU active @@ -2475,7 +2474,7 @@ static ssize_t amdgpu_iomem_write(struct file *f, const char __user *buf, struct page *p; void *ptr; - bytes = bytes < size ? bytes : size; + bytes = min(bytes, size); addr = dom ? iommu_iova_to_phys(dom, addr) : addr; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c index 9c9cca129498..565a1fa436d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c @@ -239,8 +239,13 @@ static int amdgpu_xcp_dev_alloc(struct amdgpu_device *adev) for (i = 1; i < MAX_XCP; i++) { ret = amdgpu_xcp_drm_dev_alloc(&p_ddev); - if (ret) + if (ret == -ENOSPC) { + dev_warn(adev->dev, + "Skip xcp node #%d when out of drm node resource.", i); + return 0; + } else if (ret) { return ret; + } /* Redirect all IOCTLs to the primary device */ adev->xcp_mgr->xcp[i].rdev = p_ddev->render->dev; @@ -328,6 +333,9 @@ int amdgpu_xcp_dev_register(struct amdgpu_device *adev, return 0; for (i = 1; i < MAX_XCP; i++) { + if (!adev->xcp_mgr->xcp[i].ddev) + break; + ret = drm_dev_register(adev->xcp_mgr->xcp[i].ddev, ent->driver_data); if (ret) return ret; @@ -345,6 +353,9 @@ void amdgpu_xcp_dev_unplug(struct amdgpu_device *adev) return; for (i = 1; i < MAX_XCP; i++) { + if (!adev->xcp_mgr->xcp[i].ddev) + break; + p_ddev = adev->xcp_mgr->xcp[i].ddev; drm_dev_unplug(p_ddev); p_ddev->render->dev = adev->xcp_mgr->xcp[i].rdev; diff --git a/drivers/gpu/drm/amd/amdgpu/atom.h b/drivers/gpu/drm/amd/amdgpu/atom.h index 55bf99d5288d..c11cf18a0f18 100644 --- a/drivers/gpu/drm/amd/amdgpu/atom.h +++ b/drivers/gpu/drm/amd/amdgpu/atom.h @@ -117,12 +117,15 @@ struct drm_device; struct card_info { struct drm_device *dev; - void (* reg_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ - uint32_t (* reg_read)(struct card_info *, uint32_t); /* filled by driver */ - void (* mc_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ - uint32_t (* mc_read)(struct card_info *, uint32_t); /* filled by driver */ - void (* pll_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ - uint32_t (* pll_read)(struct card_info *, uint32_t); /* filled by driver */ + void (*reg_write)(struct card_info *info, + u32 reg, uint32_t val); /* filled by driver */ + uint32_t (*reg_read)(struct card_info *info, uint32_t reg); /* filled by driver */ + void (*mc_write)(struct card_info *info, + u32 reg, uint32_t val); /* filled by driver */ + uint32_t (*mc_read)(struct card_info *info, uint32_t reg); /* filled by driver */ + void (*pll_write)(struct card_info *info, + u32 reg, uint32_t val); /* filled by driver */ + uint32_t (*pll_read)(struct card_info *info, uint32_t reg); /* filled by driver */ }; struct atom_context { @@ -152,10 +155,10 @@ struct atom_context { extern int amdgpu_atom_debug; -struct atom_context *amdgpu_atom_parse(struct card_info *, void *); -int amdgpu_atom_execute_table(struct atom_context *, int, uint32_t *); -int amdgpu_atom_asic_init(struct atom_context *); -void amdgpu_atom_destroy(struct atom_context *); +struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios); +int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params); +int amdgpu_atom_asic_init(struct atom_context *ctx); +void amdgpu_atom_destroy(struct atom_context *ctx); bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index, uint16_t *size, uint8_t *frev, uint8_t *crev, uint16_t *data_start); bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index, 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 d8d6807e7b96..57ed4e5c294c 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c @@ -337,13 +337,11 @@ static uint64_t gfx_v9_4_3_get_gpu_clock_counter(struct amdgpu_device *adev) { uint64_t clock; - amdgpu_gfx_off_ctrl(adev, false); mutex_lock(&adev->gfx.gpu_clock_mutex); WREG32_SOC15(GC, GET_INST(GC, 0), regRLC_CAPTURE_GPU_CLOCK_COUNT, 1); clock = (uint64_t)RREG32_SOC15(GC, GET_INST(GC, 0), regRLC_GPU_CLOCK_COUNT_LSB) | ((uint64_t)RREG32_SOC15(GC, GET_INST(GC, 0), regRLC_GPU_CLOCK_COUNT_MSB) << 32ULL); mutex_unlock(&adev->gfx.gpu_clock_mutex); - amdgpu_gfx_off_ctrl(adev, true); return clock; } diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 880460cd3239..f9a5a2c0573e 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -1998,6 +1998,19 @@ static int gmc_v9_0_init_mem_ranges(struct amdgpu_device *adev) return 0; } +static void gmc_v9_4_3_init_vram_info(struct amdgpu_device *adev) +{ + static const u32 regBIF_BIOS_SCRATCH_4 = 0x50; + u32 vram_info; + + if (!amdgpu_sriov_vf(adev)) { + vram_info = RREG32(regBIF_BIOS_SCRATCH_4); + adev->gmc.vram_vendor = vram_info & 0xF; + } + adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM; + adev->gmc.vram_width = 128 * 64; +} + static int gmc_v9_0_sw_init(void *handle) { int r, vram_width = 0, vram_type = 0, vram_vendor = 0, dma_addr_bits; @@ -2010,15 +2023,12 @@ static int gmc_v9_0_sw_init(void *handle) spin_lock_init(&adev->gmc.invalidate_lock); - if (!(adev->bios) || adev->gmc.is_app_apu) { + if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) { + gmc_v9_4_3_init_vram_info(adev); + } else if (!adev->bios) { if (adev->flags & AMD_IS_APU) { - if (adev->gmc.is_app_apu) { - adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM; - adev->gmc.vram_width = 128 * 64; - } else { - adev->gmc.vram_type = AMDGPU_VRAM_TYPE_DDR4; - adev->gmc.vram_width = 64 * 64; - } + adev->gmc.vram_type = AMDGPU_VRAM_TYPE_DDR4; + adev->gmc.vram_width = 64 * 64; } else { adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM; adev->gmc.vram_width = 128 * 64; diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c index 153731d6ce8b..15612915bb6c 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c @@ -26,6 +26,7 @@ #include "soc15.h" #include "soc15d.h" #include "jpeg_v4_0_3.h" +#include "mmsch_v4_0_3.h" #include "vcn/vcn_4_0_3_offset.h" #include "vcn/vcn_4_0_3_sh_mask.h" @@ -41,6 +42,7 @@ static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev); static int jpeg_v4_0_3_set_powergating_state(void *handle, enum amd_powergating_state state); static void jpeg_v4_0_3_set_ras_funcs(struct amdgpu_device *adev); +static void jpeg_v4_0_3_dec_ring_set_wptr(struct amdgpu_ring *ring); static int amdgpu_ih_srcid_jpeg[] = { VCN_4_0__SRCID__JPEG_DECODE, @@ -109,9 +111,20 @@ static int jpeg_v4_0_3_sw_init(void *handle) ring = &adev->jpeg.inst[i].ring_dec[j]; ring->use_doorbell = true; ring->vm_hub = AMDGPU_MMHUB0(adev->jpeg.inst[i].aid_id); - ring->doorbell_index = - (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + - 1 + j + 9 * jpeg_inst; + if (!amdgpu_sriov_vf(adev)) { + ring->doorbell_index = + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 1 + j + 9 * jpeg_inst; + } else { + if (j < 4) + ring->doorbell_index = + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 4 + j + 32 * jpeg_inst; + else + ring->doorbell_index = + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 8 + j + 32 * jpeg_inst; + } sprintf(ring->name, "jpeg_dec_%d.%d", adev->jpeg.inst[i].aid_id, j); r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0, AMDGPU_RING_PRIO_DEFAULT, NULL); @@ -160,6 +173,119 @@ static int jpeg_v4_0_3_sw_fini(void *handle) return r; } +static int jpeg_v4_0_3_start_sriov(struct amdgpu_device *adev) +{ + struct amdgpu_ring *ring; + uint64_t ctx_addr; + uint32_t param, resp, expected; + uint32_t tmp, timeout; + + struct amdgpu_mm_table *table = &adev->virt.mm_table; + uint32_t *table_loc; + uint32_t table_size; + uint32_t size, size_dw, item_offset; + uint32_t init_status; + int i, j, jpeg_inst; + + struct mmsch_v4_0_cmd_direct_write + direct_wt = { {0} }; + struct mmsch_v4_0_cmd_end end = { {0} }; + struct mmsch_v4_0_3_init_header header; + + direct_wt.cmd_header.command_type = + MMSCH_COMMAND__DIRECT_REG_WRITE; + end.cmd_header.command_type = + MMSCH_COMMAND__END; + + for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) { + jpeg_inst = GET_INST(JPEG, i); + + memset(&header, 0, sizeof(struct mmsch_v4_0_3_init_header)); + header.version = MMSCH_VERSION; + header.total_size = sizeof(struct mmsch_v4_0_3_init_header) >> 2; + + table_loc = (uint32_t *)table->cpu_addr; + table_loc += header.total_size; + + item_offset = header.total_size; + + for (j = 0; j < adev->jpeg.num_jpeg_rings; j++) { + ring = &adev->jpeg.inst[i].ring_dec[j]; + table_size = 0; + + tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_LOW); + MMSCH_V4_0_INSERT_DIRECT_WT(tmp, lower_32_bits(ring->gpu_addr)); + tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_HIGH); + MMSCH_V4_0_INSERT_DIRECT_WT(tmp, upper_32_bits(ring->gpu_addr)); + tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_SIZE); + MMSCH_V4_0_INSERT_DIRECT_WT(tmp, ring->ring_size / 4); + + if (j <= 3) { + header.mjpegdec0[j].table_offset = item_offset; + header.mjpegdec0[j].init_status = 0; + header.mjpegdec0[j].table_size = table_size; + } else { + header.mjpegdec1[j - 4].table_offset = item_offset; + header.mjpegdec1[j - 4].init_status = 0; + header.mjpegdec1[j - 4].table_size = table_size; + } + header.total_size += table_size; + item_offset += table_size; + } + + MMSCH_V4_0_INSERT_END(); + + /* send init table to MMSCH */ + size = sizeof(struct mmsch_v4_0_3_init_header); + table_loc = (uint32_t *)table->cpu_addr; + memcpy((void *)table_loc, &header, size); + + ctx_addr = table->gpu_addr; + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr)); + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr)); + + tmp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID); + tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK; + tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT); + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID, tmp); + + size = header.total_size; + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_SIZE, size); + + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP, 0); + + param = 0x00000001; + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_HOST, param); + tmp = 0; + timeout = 1000; + resp = 0; + expected = MMSCH_VF_MAILBOX_RESP__OK; + init_status = + ((struct mmsch_v4_0_3_init_header *)(table_loc))->mjpegdec0[i].init_status; + while (resp != expected) { + resp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP); + + if (resp != 0) + break; + udelay(10); + tmp = tmp + 10; + if (tmp >= timeout) { + DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\ + " waiting for regMMSCH_VF_MAILBOX_RESP "\ + "(expected=0x%08x, readback=0x%08x)\n", + tmp, expected, resp); + return -EBUSY; + } + } + if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE && + init_status != MMSCH_VF_ENGINE_STATUS__PASS) + DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init status for jpeg: %x\n", + resp, init_status); + + } + return 0; +} + /** * jpeg_v4_0_3_hw_init - start and test JPEG block * @@ -172,31 +298,47 @@ static int jpeg_v4_0_3_hw_init(void *handle) struct amdgpu_ring *ring; int i, j, r, jpeg_inst; - for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { - jpeg_inst = GET_INST(JPEG, i); + if (amdgpu_sriov_vf(adev)) { + r = jpeg_v4_0_3_start_sriov(adev); + if (r) + return r; - ring = adev->jpeg.inst[i].ring_dec; + for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { + ring = &adev->jpeg.inst[i].ring_dec[j]; + ring->wptr = 0; + ring->wptr_old = 0; + jpeg_v4_0_3_dec_ring_set_wptr(ring); + ring->sched.ready = true; + } + } + } else { + for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { + jpeg_inst = GET_INST(JPEG, i); - if (ring->use_doorbell) - adev->nbio.funcs->vcn_doorbell_range( - adev, ring->use_doorbell, - (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + - 9 * jpeg_inst, - adev->jpeg.inst[i].aid_id); + ring = adev->jpeg.inst[i].ring_dec; - for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { - ring = &adev->jpeg.inst[i].ring_dec[j]; if (ring->use_doorbell) - WREG32_SOC15_OFFSET( - VCN, GET_INST(VCN, i), - regVCN_JPEG_DB_CTRL, - (ring->pipe ? (ring->pipe - 0x15) : 0), - ring->doorbell_index + adev->nbio.funcs->vcn_doorbell_range( + adev, ring->use_doorbell, + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 9 * jpeg_inst, + adev->jpeg.inst[i].aid_id); + + for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { + ring = &adev->jpeg.inst[i].ring_dec[j]; + if (ring->use_doorbell) + WREG32_SOC15_OFFSET( + VCN, GET_INST(VCN, i), + regVCN_JPEG_DB_CTRL, + (ring->pipe ? (ring->pipe - 0x15) : 0), + ring->doorbell_index << VCN_JPEG_DB_CTRL__OFFSET__SHIFT | - VCN_JPEG_DB_CTRL__EN_MASK); - r = amdgpu_ring_test_helper(ring); - if (r) - return r; + VCN_JPEG_DB_CTRL__EN_MASK); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + } } } DRM_DEV_INFO(adev->dev, "JPEG decode initialized successfully.\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/mmsch_v4_0_3.h b/drivers/gpu/drm/amd/amdgpu/mmsch_v4_0_3.h new file mode 100644 index 000000000000..db7eb5260295 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/mmsch_v4_0_3.h @@ -0,0 +1,37 @@ +/* + * 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 __MMSCH_V4_0_3_H__ +#define __MMSCH_V4_0_3_H__ + +#include "amdgpu_vcn.h" +#include "mmsch_v4_0.h" + +struct mmsch_v4_0_3_init_header { + uint32_t version; + uint32_t total_size; + struct mmsch_v4_0_table_info vcn0; + struct mmsch_v4_0_table_info mjpegdec0[4]; + struct mmsch_v4_0_table_info mjpegdec1[4]; +}; +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c index 58db1ee631b3..10b17bd5aebe 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c @@ -691,6 +691,27 @@ static int psp_v13_0_vbflash_status(struct psp_context *psp) return RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_115); } +static int psp_v13_0_fatal_error_recovery_quirk(struct psp_context *psp) +{ + struct amdgpu_device *adev = psp->adev; + + if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 10)) { + uint32_t reg_data; + /* MP1 fatal error: trigger PSP dram read to unhalt PSP + * during MP1 triggered sync flood. + */ + reg_data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67); + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67, reg_data + 0x10); + + /* delay 1000ms for the mode1 reset for fatal error + * to be recovered back. + */ + msleep(1000); + } + + return 0; +} + static const struct psp_funcs psp_v13_0_funcs = { .init_microcode = psp_v13_0_init_microcode, .bootloader_load_kdb = psp_v13_0_bootloader_load_kdb, @@ -710,7 +731,8 @@ static const struct psp_funcs psp_v13_0_funcs = { .load_usbc_pd_fw = psp_v13_0_load_usbc_pd_fw, .read_usbc_pd_fw = psp_v13_0_read_usbc_pd_fw, .update_spirom = psp_v13_0_update_spirom, - .vbflash_stat = psp_v13_0_vbflash_status + .vbflash_stat = psp_v13_0_vbflash_status, + .fatal_error_recovery_quirk = psp_v13_0_fatal_error_recovery_quirk, }; void psp_v13_0_set_psp_funcs(struct psp_context *psp) diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c index e62472e6e7b3..f85d18cd74ec 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c @@ -31,6 +31,7 @@ #include "soc15d.h" #include "soc15_hw_ip.h" #include "vcn_v2_0.h" +#include "mmsch_v4_0_3.h" #include "vcn/vcn_4_0_3_offset.h" #include "vcn/vcn_4_0_3_sh_mask.h" @@ -44,6 +45,7 @@ #define VCN_VID_SOC_ADDRESS_2_0 0x1fb00 #define VCN1_VID_SOC_ADDRESS_3_0 0x48300 +static int vcn_v4_0_3_start_sriov(struct amdgpu_device *adev); static void vcn_v4_0_3_set_unified_ring_funcs(struct amdgpu_device *adev); static void vcn_v4_0_3_set_irq_funcs(struct amdgpu_device *adev); static int vcn_v4_0_3_set_powergating_state(void *handle, @@ -111,9 +113,16 @@ static int vcn_v4_0_3_sw_init(void *handle) ring = &adev->vcn.inst[i].ring_enc[0]; ring->use_doorbell = true; - ring->doorbell_index = - (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + - 9 * vcn_inst; + + if (!amdgpu_sriov_vf(adev)) + ring->doorbell_index = + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 9 * vcn_inst; + else + ring->doorbell_index = + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 32 * vcn_inst; + ring->vm_hub = AMDGPU_MMHUB0(adev->vcn.inst[i].aid_id); sprintf(ring->name, "vcn_unified_%d", adev->vcn.inst[i].aid_id); r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0, @@ -130,6 +139,12 @@ static int vcn_v4_0_3_sw_init(void *handle) amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]); } + if (amdgpu_sriov_vf(adev)) { + r = amdgpu_virt_alloc_mm_table(adev); + if (r) + return r; + } + if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) adev->vcn.pause_dpg_mode = vcn_v4_0_3_pause_dpg_mode; @@ -167,6 +182,9 @@ static int vcn_v4_0_3_sw_fini(void *handle) drm_dev_exit(idx); } + if (amdgpu_sriov_vf(adev)) + amdgpu_virt_free_mm_table(adev); + r = amdgpu_vcn_suspend(adev); if (r) return r; @@ -189,33 +207,47 @@ static int vcn_v4_0_3_hw_init(void *handle) struct amdgpu_ring *ring; int i, r, vcn_inst; - for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { - vcn_inst = GET_INST(VCN, i); - ring = &adev->vcn.inst[i].ring_enc[0]; + if (amdgpu_sriov_vf(adev)) { + r = vcn_v4_0_3_start_sriov(adev); + if (r) + goto done; - if (ring->use_doorbell) { - adev->nbio.funcs->vcn_doorbell_range( - adev, ring->use_doorbell, - (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + - 9 * vcn_inst, - adev->vcn.inst[i].aid_id); - - WREG32_SOC15( - VCN, GET_INST(VCN, ring->me), - regVCN_RB1_DB_CTRL, - ring->doorbell_index - << VCN_RB1_DB_CTRL__OFFSET__SHIFT | - VCN_RB1_DB_CTRL__EN_MASK); - - /* Read DB_CTRL to flush the write DB_CTRL command. */ - RREG32_SOC15( - VCN, GET_INST(VCN, ring->me), - regVCN_RB1_DB_CTRL); + for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + ring = &adev->vcn.inst[i].ring_enc[0]; + ring->wptr = 0; + ring->wptr_old = 0; + vcn_v4_0_3_unified_ring_set_wptr(ring); + ring->sched.ready = true; } + } else { + for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + vcn_inst = GET_INST(VCN, i); + ring = &adev->vcn.inst[i].ring_enc[0]; + + if (ring->use_doorbell) { + adev->nbio.funcs->vcn_doorbell_range( + adev, ring->use_doorbell, + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 9 * vcn_inst, + adev->vcn.inst[i].aid_id); + + WREG32_SOC15( + VCN, GET_INST(VCN, ring->me), + regVCN_RB1_DB_CTRL, + ring->doorbell_index + << VCN_RB1_DB_CTRL__OFFSET__SHIFT | + VCN_RB1_DB_CTRL__EN_MASK); + + /* Read DB_CTRL to flush the write DB_CTRL command. */ + RREG32_SOC15( + VCN, GET_INST(VCN, ring->me), + regVCN_RB1_DB_CTRL); + } - r = amdgpu_ring_test_helper(ring); - if (r) - goto done; + r = amdgpu_ring_test_helper(ring); + if (r) + goto done; + } } done: @@ -813,6 +845,193 @@ static int vcn_v4_0_3_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, b return 0; } +static int vcn_v4_0_3_start_sriov(struct amdgpu_device *adev) +{ + int i, vcn_inst; + struct amdgpu_ring *ring_enc; + uint64_t cache_addr; + uint64_t rb_enc_addr; + uint64_t ctx_addr; + uint32_t param, resp, expected; + uint32_t offset, cache_size; + uint32_t tmp, timeout; + + struct amdgpu_mm_table *table = &adev->virt.mm_table; + uint32_t *table_loc; + uint32_t table_size; + uint32_t size, size_dw; + uint32_t init_status; + uint32_t enabled_vcn; + + struct mmsch_v4_0_cmd_direct_write + direct_wt = { {0} }; + struct mmsch_v4_0_cmd_direct_read_modify_write + direct_rd_mod_wt = { {0} }; + struct mmsch_v4_0_cmd_end end = { {0} }; + struct mmsch_v4_0_3_init_header header; + + volatile struct amdgpu_vcn4_fw_shared *fw_shared; + volatile struct amdgpu_fw_shared_rb_setup *rb_setup; + + direct_wt.cmd_header.command_type = + MMSCH_COMMAND__DIRECT_REG_WRITE; + direct_rd_mod_wt.cmd_header.command_type = + MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE; + end.cmd_header.command_type = MMSCH_COMMAND__END; + + for (i = 0; i < adev->vcn.num_vcn_inst; i++) { + vcn_inst = GET_INST(VCN, i); + + memset(&header, 0, sizeof(struct mmsch_v4_0_3_init_header)); + header.version = MMSCH_VERSION; + header.total_size = sizeof(struct mmsch_v4_0_3_init_header) >> 2; + + table_loc = (uint32_t *)table->cpu_addr; + table_loc += header.total_size; + + table_size = 0; + + MMSCH_V4_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCN, 0, regUVD_STATUS), + ~UVD_STATUS__UVD_BUSY, UVD_STATUS__UVD_BUSY); + + cache_size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.fw->size + 4); + + if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) { + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), + adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_lo); + + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), + adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_hi); + + offset = 0; + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_VCPU_CACHE_OFFSET0), 0); + } else { + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), + lower_32_bits(adev->vcn.inst[i].gpu_addr)); + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), + upper_32_bits(adev->vcn.inst[i].gpu_addr)); + offset = cache_size; + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_VCPU_CACHE_OFFSET0), + AMDGPU_UVD_FIRMWARE_OFFSET >> 3); + } + + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_VCPU_CACHE_SIZE0), + cache_size); + + cache_addr = adev->vcn.inst[vcn_inst].gpu_addr + offset; + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), lower_32_bits(cache_addr)); + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), upper_32_bits(cache_addr)); + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_VCPU_CACHE_OFFSET1), 0); + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE); + + cache_addr = adev->vcn.inst[vcn_inst].gpu_addr + offset + + AMDGPU_VCN_STACK_SIZE; + + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW), lower_32_bits(cache_addr)); + + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH), upper_32_bits(cache_addr)); + + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_VCPU_CACHE_OFFSET2), 0); + + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE); + + fw_shared = adev->vcn.inst[vcn_inst].fw_shared.cpu_addr; + rb_setup = &fw_shared->rb_setup; + + ring_enc = &adev->vcn.inst[vcn_inst].ring_enc[0]; + ring_enc->wptr = 0; + rb_enc_addr = ring_enc->gpu_addr; + + rb_setup->is_rb_enabled_flags |= RB_ENABLED; + rb_setup->rb_addr_lo = lower_32_bits(rb_enc_addr); + rb_setup->rb_addr_hi = upper_32_bits(rb_enc_addr); + rb_setup->rb_size = ring_enc->ring_size / 4; + fw_shared->present_flag_0 |= cpu_to_le32(AMDGPU_VCN_VF_RB_SETUP_FLAG); + + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW), + lower_32_bits(adev->vcn.inst[vcn_inst].fw_shared.gpu_addr)); + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH), + upper_32_bits(adev->vcn.inst[vcn_inst].fw_shared.gpu_addr)); + MMSCH_V4_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCN, 0, + regUVD_VCPU_NONCACHE_SIZE0), + AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared))); + MMSCH_V4_0_INSERT_END(); + + header.vcn0.init_status = 0; + header.vcn0.table_offset = header.total_size; + header.vcn0.table_size = table_size; + header.total_size += table_size; + + /* Send init table to mmsch */ + size = sizeof(struct mmsch_v4_0_3_init_header); + table_loc = (uint32_t *)table->cpu_addr; + memcpy((void *)table_loc, &header, size); + + ctx_addr = table->gpu_addr; + WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr)); + WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr)); + + tmp = RREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_VMID); + tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK; + tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT); + WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_VMID, tmp); + + size = header.total_size; + WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_CTX_SIZE, size); + + WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_MAILBOX_RESP, 0); + + param = 0x00000001; + WREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_MAILBOX_HOST, param); + tmp = 0; + timeout = 1000; + resp = 0; + expected = MMSCH_VF_MAILBOX_RESP__OK; + while (resp != expected) { + resp = RREG32_SOC15(VCN, vcn_inst, regMMSCH_VF_MAILBOX_RESP); + if (resp != 0) + break; + + udelay(10); + tmp = tmp + 10; + if (tmp >= timeout) { + DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\ + " waiting for regMMSCH_VF_MAILBOX_RESP "\ + "(expected=0x%08x, readback=0x%08x)\n", + tmp, expected, resp); + return -EBUSY; + } + } + + enabled_vcn = amdgpu_vcn_is_disabled_vcn(adev, VCN_DECODE_RING, 0) ? 1 : 0; + init_status = ((struct mmsch_v4_0_3_init_header *)(table_loc))->vcn0.init_status; + if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE + && init_status != MMSCH_VF_ENGINE_STATUS__PASS) { + DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init "\ + "status for VCN%x: 0x%x\n", resp, enabled_vcn, init_status); + } + } + + return 0; +} + /** * vcn_v4_0_3_start - VCN start * @@ -1317,6 +1536,15 @@ static int vcn_v4_0_3_set_powergating_state(void *handle, struct amdgpu_device *adev = (struct amdgpu_device *)handle; int ret; + /* for SRIOV, guest should not control VCN Power-gating + * MMSCH FW should control Power-gating and clock-gating + * guest should avoid touching CGC and PG + */ + if (amdgpu_sriov_vf(adev)) { + adev->vcn.cur_state = AMD_PG_STATE_UNGATE; + return 0; + } + if (state == adev->vcn.cur_state) return 0; |