diff options
85 files changed, 2005 insertions, 690 deletions
diff --git a/.clang-format b/.clang-format index 48405c54ef27..d5c05db1a0d9 100644 --- a/.clang-format +++ b/.clang-format @@ -167,7 +167,7 @@ ForEachMacros: - 'drm_connector_for_each_possible_encoder' - 'drm_exec_for_each_locked_object' - 'drm_exec_for_each_locked_object_reverse' - - 'drm_for_each_bridge_in_chain' + - 'drm_for_each_bridge_in_chain_scoped' - 'drm_for_each_connector_iter' - 'drm_for_each_crtc' - 'drm_for_each_crtc_reverse' diff --git a/drivers/accel/amdxdna/aie2_pci.c b/drivers/accel/amdxdna/aie2_pci.c index 87c425e3d2b9..6e39c769bb6d 100644 --- a/drivers/accel/amdxdna/aie2_pci.c +++ b/drivers/accel/amdxdna/aie2_pci.c @@ -898,6 +898,12 @@ static int aie2_query_ctx_status_array(struct amdxdna_client *client, drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock)); + if (args->element_size > SZ_4K || args->num_element > SZ_1K) { + XDNA_DBG(xdna, "Invalid element size %d or number of element %d", + args->element_size, args->num_element); + return -EINVAL; + } + array_args.element_size = min(args->element_size, sizeof(struct amdxdna_drm_hwctx_entry)); array_args.buffer = args->buffer; diff --git a/drivers/accel/amdxdna/amdxdna_gem.c b/drivers/accel/amdxdna/amdxdna_gem.c index d407a36eb412..7f91863c3f24 100644 --- a/drivers/accel/amdxdna/amdxdna_gem.c +++ b/drivers/accel/amdxdna/amdxdna_gem.c @@ -392,35 +392,33 @@ static const struct dma_buf_ops amdxdna_dmabuf_ops = { .vunmap = drm_gem_dmabuf_vunmap, }; -static int amdxdna_gem_obj_vmap(struct drm_gem_object *obj, struct iosys_map *map) +static int amdxdna_gem_obj_vmap(struct amdxdna_gem_obj *abo, void **vaddr) { - struct amdxdna_gem_obj *abo = to_xdna_obj(obj); - - iosys_map_clear(map); - - dma_resv_assert_held(obj->resv); + struct iosys_map map = IOSYS_MAP_INIT_VADDR(NULL); + int ret; if (is_import_bo(abo)) - dma_buf_vmap(abo->dma_buf, map); + ret = dma_buf_vmap_unlocked(abo->dma_buf, &map); else - drm_gem_shmem_object_vmap(obj, map); - - if (!map->vaddr) - return -ENOMEM; + ret = drm_gem_vmap(to_gobj(abo), &map); - return 0; + *vaddr = map.vaddr; + return ret; } -static void amdxdna_gem_obj_vunmap(struct drm_gem_object *obj, struct iosys_map *map) +static void amdxdna_gem_obj_vunmap(struct amdxdna_gem_obj *abo) { - struct amdxdna_gem_obj *abo = to_xdna_obj(obj); + struct iosys_map map; + + if (!abo->mem.kva) + return; - dma_resv_assert_held(obj->resv); + iosys_map_set_vaddr(&map, abo->mem.kva); if (is_import_bo(abo)) - dma_buf_vunmap(abo->dma_buf, map); + dma_buf_vunmap_unlocked(abo->dma_buf, &map); else - drm_gem_shmem_object_vunmap(obj, map); + drm_gem_vunmap(to_gobj(abo), &map); } static struct dma_buf *amdxdna_gem_prime_export(struct drm_gem_object *gobj, int flags) @@ -455,7 +453,6 @@ static void amdxdna_gem_obj_free(struct drm_gem_object *gobj) { struct amdxdna_dev *xdna = to_xdna_dev(gobj->dev); struct amdxdna_gem_obj *abo = to_xdna_obj(gobj); - struct iosys_map map = IOSYS_MAP_INIT_VADDR(abo->mem.kva); XDNA_DBG(xdna, "BO type %d xdna_addr 0x%llx", abo->type, abo->mem.dev_addr); @@ -468,7 +465,7 @@ static void amdxdna_gem_obj_free(struct drm_gem_object *gobj) if (abo->type == AMDXDNA_BO_DEV_HEAP) drm_mm_takedown(&abo->mm); - drm_gem_vunmap(gobj, &map); + amdxdna_gem_obj_vunmap(abo); mutex_destroy(&abo->lock); if (is_import_bo(abo)) { @@ -489,8 +486,8 @@ static const struct drm_gem_object_funcs amdxdna_gem_shmem_funcs = { .pin = drm_gem_shmem_object_pin, .unpin = drm_gem_shmem_object_unpin, .get_sg_table = drm_gem_shmem_object_get_sg_table, - .vmap = amdxdna_gem_obj_vmap, - .vunmap = amdxdna_gem_obj_vunmap, + .vmap = drm_gem_shmem_object_vmap, + .vunmap = drm_gem_shmem_object_vunmap, .mmap = amdxdna_gem_obj_mmap, .vm_ops = &drm_gem_shmem_vm_ops, .export = amdxdna_gem_prime_export, @@ -663,7 +660,6 @@ amdxdna_drm_create_dev_heap(struct drm_device *dev, struct drm_file *filp) { struct amdxdna_client *client = filp->driver_priv; - struct iosys_map map = IOSYS_MAP_INIT_VADDR(NULL); struct amdxdna_dev *xdna = to_xdna_dev(dev); struct amdxdna_gem_obj *abo; int ret; @@ -692,12 +688,11 @@ amdxdna_drm_create_dev_heap(struct drm_device *dev, abo->mem.dev_addr = client->xdna->dev_info->dev_mem_base; drm_mm_init(&abo->mm, abo->mem.dev_addr, abo->mem.size); - ret = drm_gem_vmap(to_gobj(abo), &map); + ret = amdxdna_gem_obj_vmap(abo, &abo->mem.kva); if (ret) { XDNA_ERR(xdna, "Vmap heap bo failed, ret %d", ret); goto release_obj; } - abo->mem.kva = map.vaddr; client->dev_heap = abo; drm_gem_object_get(to_gobj(abo)); @@ -748,7 +743,6 @@ amdxdna_drm_create_cmd_bo(struct drm_device *dev, struct amdxdna_drm_create_bo *args, struct drm_file *filp) { - struct iosys_map map = IOSYS_MAP_INIT_VADDR(NULL); struct amdxdna_dev *xdna = to_xdna_dev(dev); struct amdxdna_gem_obj *abo; int ret; @@ -770,12 +764,11 @@ amdxdna_drm_create_cmd_bo(struct drm_device *dev, abo->type = AMDXDNA_BO_CMD; abo->client = filp->driver_priv; - ret = drm_gem_vmap(to_gobj(abo), &map); + ret = amdxdna_gem_obj_vmap(abo, &abo->mem.kva); if (ret) { XDNA_ERR(xdna, "Vmap cmd bo failed, ret %d", ret); goto release_obj; } - abo->mem.kva = map.vaddr; return abo; diff --git a/drivers/accel/ivpu/ivpu_debugfs.c b/drivers/accel/ivpu/ivpu_debugfs.c index cd24ccd20ba6..3bd85ee6c26b 100644 --- a/drivers/accel/ivpu/ivpu_debugfs.c +++ b/drivers/accel/ivpu/ivpu_debugfs.c @@ -398,35 +398,25 @@ static int dct_active_set(void *data, u64 active_percent) DEFINE_DEBUGFS_ATTRIBUTE(ivpu_dct_fops, dct_active_get, dct_active_set, "%llu\n"); +static void print_priority_band(struct seq_file *s, struct ivpu_hw_info *hw, + int band, const char *name) +{ + seq_printf(s, "%-9s: grace_period %9u process_grace_period %9u process_quantum %9u\n", + name, + hw->hws.grace_period[band], + hw->hws.process_grace_period[band], + hw->hws.process_quantum[band]); +} + static int priority_bands_show(struct seq_file *s, void *v) { struct ivpu_device *vdev = s->private; struct ivpu_hw_info *hw = vdev->hw; - for (int band = VPU_JOB_SCHEDULING_PRIORITY_BAND_IDLE; - band < VPU_JOB_SCHEDULING_PRIORITY_BAND_COUNT; band++) { - switch (band) { - case VPU_JOB_SCHEDULING_PRIORITY_BAND_IDLE: - seq_puts(s, "Idle: "); - break; - - case VPU_JOB_SCHEDULING_PRIORITY_BAND_NORMAL: - seq_puts(s, "Normal: "); - break; - - case VPU_JOB_SCHEDULING_PRIORITY_BAND_FOCUS: - seq_puts(s, "Focus: "); - break; - - case VPU_JOB_SCHEDULING_PRIORITY_BAND_REALTIME: - seq_puts(s, "Realtime: "); - break; - } - - seq_printf(s, "grace_period %9u process_grace_period %9u process_quantum %9u\n", - hw->hws.grace_period[band], hw->hws.process_grace_period[band], - hw->hws.process_quantum[band]); - } + print_priority_band(s, hw, VPU_JOB_SCHEDULING_PRIORITY_BAND_IDLE, "Idle"); + print_priority_band(s, hw, VPU_JOB_SCHEDULING_PRIORITY_BAND_NORMAL, "Normal"); + print_priority_band(s, hw, VPU_JOB_SCHEDULING_PRIORITY_BAND_FOCUS, "Focus"); + print_priority_band(s, hw, VPU_JOB_SCHEDULING_PRIORITY_BAND_REALTIME, "Realtime"); return 0; } diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c index 3289751b4757..a08f99c3ba4a 100644 --- a/drivers/accel/ivpu/ivpu_drv.c +++ b/drivers/accel/ivpu/ivpu_drv.c @@ -200,6 +200,9 @@ static int ivpu_get_param_ioctl(struct drm_device *dev, void *data, struct drm_f case DRM_IVPU_PARAM_CAPABILITIES: args->value = ivpu_is_capable(vdev, args->index); break; + case DRM_IVPU_PARAM_PREEMPT_BUFFER_SIZE: + args->value = ivpu_fw_preempt_buf_size(vdev); + break; default: ret = -EINVAL; break; diff --git a/drivers/accel/ivpu/ivpu_fw.c b/drivers/accel/ivpu/ivpu_fw.c index 9db741695401..32f513499829 100644 --- a/drivers/accel/ivpu/ivpu_fw.c +++ b/drivers/accel/ivpu/ivpu_fw.c @@ -26,6 +26,8 @@ #define FW_RUNTIME_MIN_ADDR (FW_GLOBAL_MEM_START) #define FW_RUNTIME_MAX_ADDR (FW_GLOBAL_MEM_END - FW_SHARED_MEM_SIZE) #define FW_FILE_IMAGE_OFFSET (VPU_FW_HEADER_SIZE + FW_VERSION_HEADER_SIZE) +#define FW_PREEMPT_BUF_MIN_SIZE SZ_4K +#define FW_PREEMPT_BUF_MAX_SIZE SZ_32M #define WATCHDOG_MSS_REDIRECT 32 #define WATCHDOG_NCE_REDIRECT 33 @@ -151,6 +153,47 @@ ivpu_fw_sched_mode_select(struct ivpu_device *vdev, const struct vpu_firmware_he return VPU_SCHEDULING_MODE_HW; } +static void +ivpu_preemption_config_parse(struct ivpu_device *vdev, const struct vpu_firmware_header *fw_hdr) +{ + struct ivpu_fw_info *fw = vdev->fw; + u32 primary_preempt_buf_size, secondary_preempt_buf_size; + + if (fw_hdr->preemption_buffer_1_max_size) + primary_preempt_buf_size = fw_hdr->preemption_buffer_1_max_size; + else + primary_preempt_buf_size = fw_hdr->preemption_buffer_1_size; + + if (fw_hdr->preemption_buffer_2_max_size) + secondary_preempt_buf_size = fw_hdr->preemption_buffer_2_max_size; + else + secondary_preempt_buf_size = fw_hdr->preemption_buffer_2_size; + + ivpu_dbg(vdev, FW_BOOT, "Preemption buffer size, primary: %u, secondary: %u\n", + primary_preempt_buf_size, secondary_preempt_buf_size); + + if (primary_preempt_buf_size < FW_PREEMPT_BUF_MIN_SIZE || + secondary_preempt_buf_size < FW_PREEMPT_BUF_MIN_SIZE) { + ivpu_warn(vdev, "Preemption buffers size too small\n"); + return; + } + + if (primary_preempt_buf_size > FW_PREEMPT_BUF_MAX_SIZE || + secondary_preempt_buf_size > FW_PREEMPT_BUF_MAX_SIZE) { + ivpu_warn(vdev, "Preemption buffers size too big\n"); + return; + } + + if (fw->sched_mode != VPU_SCHEDULING_MODE_HW) + return; + + if (ivpu_test_mode & IVPU_TEST_MODE_MIP_DISABLE) + return; + + vdev->fw->primary_preempt_buf_size = ALIGN(primary_preempt_buf_size, PAGE_SIZE); + vdev->fw->secondary_preempt_buf_size = ALIGN(secondary_preempt_buf_size, PAGE_SIZE); +} + static int ivpu_fw_parse(struct ivpu_device *vdev) { struct ivpu_fw_info *fw = vdev->fw; @@ -235,17 +278,9 @@ static int ivpu_fw_parse(struct ivpu_device *vdev) fw->sched_mode = ivpu_fw_sched_mode_select(vdev, fw_hdr); ivpu_info(vdev, "Scheduler mode: %s\n", fw->sched_mode ? "HW" : "OS"); - if (fw_hdr->preemption_buffer_1_max_size) - fw->primary_preempt_buf_size = fw_hdr->preemption_buffer_1_max_size; - else - fw->primary_preempt_buf_size = fw_hdr->preemption_buffer_1_size; - - if (fw_hdr->preemption_buffer_2_max_size) - fw->secondary_preempt_buf_size = fw_hdr->preemption_buffer_2_max_size; - else - fw->secondary_preempt_buf_size = fw_hdr->preemption_buffer_2_size; - ivpu_dbg(vdev, FW_BOOT, "Preemption buffer sizes: primary %u, secondary %u\n", - fw->primary_preempt_buf_size, fw->secondary_preempt_buf_size); + ivpu_preemption_config_parse(vdev, fw_hdr); + ivpu_dbg(vdev, FW_BOOT, "Mid-inference preemption %s supported\n", + ivpu_fw_preempt_buf_size(vdev) ? "is" : "is not"); if (fw_hdr->ro_section_start_address && !is_within_range(fw_hdr->ro_section_start_address, fw_hdr->ro_section_size, @@ -483,11 +518,6 @@ static void ivpu_fw_boot_params_print(struct ivpu_device *vdev, struct vpu_boot_ ivpu_dbg(vdev, FW_BOOT, "boot_params.cache_defaults[VPU_BOOT_L2_CACHE_CFG_NN].cfg = 0x%x\n", boot_params->cache_defaults[VPU_BOOT_L2_CACHE_CFG_NN].cfg); - ivpu_dbg(vdev, FW_BOOT, "boot_params.global_memory_allocator_base = 0x%llx\n", - boot_params->global_memory_allocator_base); - ivpu_dbg(vdev, FW_BOOT, "boot_params.global_memory_allocator_size = 0x%x\n", - boot_params->global_memory_allocator_size); - ivpu_dbg(vdev, FW_BOOT, "boot_params.shave_nn_fw_base = 0x%llx\n", boot_params->shave_nn_fw_base); @@ -495,10 +525,6 @@ static void ivpu_fw_boot_params_print(struct ivpu_device *vdev, struct vpu_boot_ boot_params->watchdog_irq_mss); ivpu_dbg(vdev, FW_BOOT, "boot_params.watchdog_irq_nce = 0x%x\n", boot_params->watchdog_irq_nce); - ivpu_dbg(vdev, FW_BOOT, "boot_params.host_to_vpu_irq = 0x%x\n", - boot_params->host_to_vpu_irq); - ivpu_dbg(vdev, FW_BOOT, "boot_params.job_done_irq = 0x%x\n", - boot_params->job_done_irq); ivpu_dbg(vdev, FW_BOOT, "boot_params.host_version_id = 0x%x\n", boot_params->host_version_id); diff --git a/drivers/accel/ivpu/ivpu_fw.h b/drivers/accel/ivpu/ivpu_fw.h index 7081913fb0dd..6fe2917abda6 100644 --- a/drivers/accel/ivpu/ivpu_fw.h +++ b/drivers/accel/ivpu/ivpu_fw.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0-only */ /* - * Copyright (C) 2020-2024 Intel Corporation + * Copyright (C) 2020-2025 Intel Corporation */ #ifndef __IVPU_FW_H__ @@ -52,4 +52,9 @@ static inline bool ivpu_fw_is_cold_boot(struct ivpu_device *vdev) return vdev->fw->entry_point == vdev->fw->cold_boot_entry_point; } +static inline u32 ivpu_fw_preempt_buf_size(struct ivpu_device *vdev) +{ + return vdev->fw->primary_preempt_buf_size + vdev->fw->secondary_preempt_buf_size; +} + #endif /* __IVPU_FW_H__ */ diff --git a/drivers/accel/ivpu/ivpu_gem.h b/drivers/accel/ivpu/ivpu_gem.h index aa8ff14f7aae..3ee996d503b2 100644 --- a/drivers/accel/ivpu/ivpu_gem.h +++ b/drivers/accel/ivpu/ivpu_gem.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0-only */ /* - * Copyright (C) 2020-2023 Intel Corporation + * Copyright (C) 2020-2025 Intel Corporation */ #ifndef __IVPU_GEM_H__ #define __IVPU_GEM_H__ @@ -96,4 +96,9 @@ static inline u32 cpu_to_vpu_addr(struct ivpu_bo *bo, void *cpu_addr) return bo->vpu_addr + (cpu_addr - ivpu_bo_vaddr(bo)); } +static inline bool ivpu_bo_is_mappable(struct ivpu_bo *bo) +{ + return bo->flags & DRM_IVPU_BO_MAPPABLE; +} + #endif /* __IVPU_GEM_H__ */ diff --git a/drivers/accel/ivpu/ivpu_job.c b/drivers/accel/ivpu/ivpu_job.c index 060f1fc031d3..044268d0fc87 100644 --- a/drivers/accel/ivpu/ivpu_job.c +++ b/drivers/accel/ivpu/ivpu_job.c @@ -34,22 +34,20 @@ static void ivpu_cmdq_ring_db(struct ivpu_device *vdev, struct ivpu_cmdq *cmdq) static int ivpu_preemption_buffers_create(struct ivpu_device *vdev, struct ivpu_file_priv *file_priv, struct ivpu_cmdq *cmdq) { - u64 primary_size = ALIGN(vdev->fw->primary_preempt_buf_size, PAGE_SIZE); - u64 secondary_size = ALIGN(vdev->fw->secondary_preempt_buf_size, PAGE_SIZE); - - if (vdev->fw->sched_mode != VPU_SCHEDULING_MODE_HW || - ivpu_test_mode & IVPU_TEST_MODE_MIP_DISABLE) + if (ivpu_fw_preempt_buf_size(vdev) == 0) return 0; cmdq->primary_preempt_buf = ivpu_bo_create(vdev, &file_priv->ctx, &vdev->hw->ranges.user, - primary_size, DRM_IVPU_BO_WC); + vdev->fw->primary_preempt_buf_size, + DRM_IVPU_BO_WC); if (!cmdq->primary_preempt_buf) { ivpu_err(vdev, "Failed to create primary preemption buffer\n"); return -ENOMEM; } cmdq->secondary_preempt_buf = ivpu_bo_create(vdev, &file_priv->ctx, &vdev->hw->ranges.dma, - secondary_size, DRM_IVPU_BO_WC); + vdev->fw->secondary_preempt_buf_size, + DRM_IVPU_BO_WC); if (!cmdq->secondary_preempt_buf) { ivpu_err(vdev, "Failed to create secondary preemption buffer\n"); goto err_free_primary; @@ -66,20 +64,39 @@ err_free_primary: static void ivpu_preemption_buffers_free(struct ivpu_device *vdev, struct ivpu_file_priv *file_priv, struct ivpu_cmdq *cmdq) { - if (vdev->fw->sched_mode != VPU_SCHEDULING_MODE_HW) - return; - if (cmdq->primary_preempt_buf) ivpu_bo_free(cmdq->primary_preempt_buf); if (cmdq->secondary_preempt_buf) ivpu_bo_free(cmdq->secondary_preempt_buf); } +static int ivpu_preemption_job_init(struct ivpu_device *vdev, struct ivpu_file_priv *file_priv, + struct ivpu_cmdq *cmdq, struct ivpu_job *job) +{ + int ret; + + /* Use preemption buffer provided by the user space */ + if (job->primary_preempt_buf) + return 0; + + if (!cmdq->primary_preempt_buf) { + /* Allocate per command queue preemption buffers */ + ret = ivpu_preemption_buffers_create(vdev, file_priv, cmdq); + if (ret) + return ret; + } + + /* Use preemption buffers allocated by the kernel */ + job->primary_preempt_buf = cmdq->primary_preempt_buf; + job->secondary_preempt_buf = cmdq->secondary_preempt_buf; + + return 0; +} + static struct ivpu_cmdq *ivpu_cmdq_alloc(struct ivpu_file_priv *file_priv) { struct ivpu_device *vdev = file_priv->vdev; struct ivpu_cmdq *cmdq; - int ret; cmdq = kzalloc(sizeof(*cmdq), GFP_KERNEL); if (!cmdq) @@ -89,10 +106,6 @@ static struct ivpu_cmdq *ivpu_cmdq_alloc(struct ivpu_file_priv *file_priv) if (!cmdq->mem) goto err_free_cmdq; - ret = ivpu_preemption_buffers_create(vdev, file_priv, cmdq); - if (ret) - ivpu_warn(vdev, "Failed to allocate preemption buffers, preemption limited\n"); - return cmdq; err_free_cmdq: @@ -219,11 +232,13 @@ static int ivpu_register_db(struct ivpu_file_priv *file_priv, struct ivpu_cmdq * ret = ivpu_jsm_register_db(vdev, file_priv->ctx.id, cmdq->db_id, cmdq->mem->vpu_addr, ivpu_bo_size(cmdq->mem)); - if (!ret) + if (!ret) { ivpu_dbg(vdev, JOB, "DB %d registered to cmdq %d ctx %d priority %d\n", cmdq->db_id, cmdq->id, file_priv->ctx.id, cmdq->priority); - else + } else { xa_erase(&vdev->db_xa, cmdq->db_id); + cmdq->db_id = 0; + } return ret; } @@ -427,17 +442,14 @@ static int ivpu_cmdq_push_job(struct ivpu_cmdq *cmdq, struct ivpu_job *job) if (unlikely(ivpu_test_mode & IVPU_TEST_MODE_NULL_SUBMISSION)) entry->flags = VPU_JOB_FLAGS_NULL_SUBMISSION_MASK; - if (vdev->fw->sched_mode == VPU_SCHEDULING_MODE_HW) { - if (cmdq->primary_preempt_buf) { - entry->primary_preempt_buf_addr = cmdq->primary_preempt_buf->vpu_addr; - entry->primary_preempt_buf_size = ivpu_bo_size(cmdq->primary_preempt_buf); - } + if (job->primary_preempt_buf) { + entry->primary_preempt_buf_addr = job->primary_preempt_buf->vpu_addr; + entry->primary_preempt_buf_size = ivpu_bo_size(job->primary_preempt_buf); + } - if (cmdq->secondary_preempt_buf) { - entry->secondary_preempt_buf_addr = cmdq->secondary_preempt_buf->vpu_addr; - entry->secondary_preempt_buf_size = - ivpu_bo_size(cmdq->secondary_preempt_buf); - } + if (job->secondary_preempt_buf) { + entry->secondary_preempt_buf_addr = job->secondary_preempt_buf->vpu_addr; + entry->secondary_preempt_buf_size = ivpu_bo_size(job->secondary_preempt_buf); } wmb(); /* Ensure that tail is updated after filling entry */ @@ -661,6 +673,13 @@ static int ivpu_job_submit(struct ivpu_job *job, u8 priority, u32 cmdq_id) goto err_unlock; } + ret = ivpu_preemption_job_init(vdev, file_priv, cmdq, job); + if (ret) { + ivpu_err(vdev, "Failed to initialize preemption buffers for job %d: %d\n", + job->job_id, ret); + goto err_unlock; + } + job->cmdq_id = cmdq->id; is_first_job = xa_empty(&vdev->submitted_jobs_xa); @@ -714,7 +733,7 @@ err_unlock: static int ivpu_job_prepare_bos_for_submit(struct drm_file *file, struct ivpu_job *job, u32 *buf_handles, - u32 buf_count, u32 commands_offset) + u32 buf_count, u32 commands_offset, u32 preempt_buffer_index) { struct ivpu_file_priv *file_priv = job->file_priv; struct ivpu_device *vdev = file_priv->vdev; @@ -750,6 +769,20 @@ ivpu_job_prepare_bos_for_submit(struct drm_file *file, struct ivpu_job *job, u32 job->cmd_buf_vpu_addr = bo->vpu_addr + commands_offset; + if (preempt_buffer_index) { + struct ivpu_bo *preempt_bo = job->bos[preempt_buffer_index]; + + if (ivpu_bo_size(preempt_bo) < ivpu_fw_preempt_buf_size(vdev)) { + ivpu_warn(vdev, "Preemption buffer is too small\n"); + return -EINVAL; + } + if (ivpu_bo_is_mappable(preempt_bo)) { + ivpu_warn(vdev, "Preemption buffer cannot be mappable\n"); + return -EINVAL; + } + job->primary_preempt_buf = preempt_bo; + } + ret = drm_gem_lock_reservations((struct drm_gem_object **)job->bos, buf_count, &acquire_ctx); if (ret) { @@ -780,7 +813,7 @@ unlock_reservations: static int ivpu_submit(struct drm_file *file, struct ivpu_file_priv *file_priv, u32 cmdq_id, u32 buffer_count, u32 engine, void __user *buffers_ptr, u32 cmds_offset, - u8 priority) + u32 preempt_buffer_index, u8 priority) { struct ivpu_device *vdev = file_priv->vdev; struct ivpu_job *job; @@ -812,7 +845,8 @@ static int ivpu_submit(struct drm_file *file, struct ivpu_file_priv *file_priv, goto err_exit_dev; } - ret = ivpu_job_prepare_bos_for_submit(file, job, buf_handles, buffer_count, cmds_offset); + ret = ivpu_job_prepare_bos_for_submit(file, job, buf_handles, buffer_count, cmds_offset, + preempt_buffer_index); if (ret) { ivpu_err(vdev, "Failed to prepare job: %d\n", ret); goto err_destroy_job; @@ -866,7 +900,7 @@ int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file) priority = ivpu_job_to_jsm_priority(args->priority); return ivpu_submit(file, file_priv, 0, args->buffer_count, args->engine, - (void __user *)args->buffers_ptr, args->commands_offset, priority); + (void __user *)args->buffers_ptr, args->commands_offset, 0, priority); } int ivpu_cmdq_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file) @@ -883,6 +917,9 @@ int ivpu_cmdq_submit_ioctl(struct drm_device *dev, void *data, struct drm_file * if (args->buffer_count == 0 || args->buffer_count > JOB_MAX_BUFFER_COUNT) return -EINVAL; + if (args->preempt_buffer_index >= args->buffer_count) + return -EINVAL; + if (!IS_ALIGNED(args->commands_offset, 8)) return -EINVAL; @@ -893,7 +930,8 @@ int ivpu_cmdq_submit_ioctl(struct drm_device *dev, void *data, struct drm_file * return -EBADFD; return ivpu_submit(file, file_priv, args->cmdq_id, args->buffer_count, VPU_ENGINE_COMPUTE, - (void __user *)args->buffers_ptr, args->commands_offset, 0); + (void __user *)args->buffers_ptr, args->commands_offset, + args->preempt_buffer_index, 0); } int ivpu_cmdq_create_ioctl(struct drm_device *dev, void *data, struct drm_file *file) @@ -1012,7 +1050,7 @@ void ivpu_context_abort_work_fn(struct work_struct *work) if (vdev->fw->sched_mode == VPU_SCHEDULING_MODE_HW) if (ivpu_jsm_reset_engine(vdev, 0)) - return; + goto runtime_put; mutex_lock(&vdev->context_list_lock); xa_for_each(&vdev->context_xa, ctx_id, file_priv) { @@ -1036,7 +1074,7 @@ void ivpu_context_abort_work_fn(struct work_struct *work) goto runtime_put; if (ivpu_jsm_hws_resume_engine(vdev, 0)) - return; + goto runtime_put; /* * In hardware scheduling mode NPU already has stopped processing jobs * and won't send us any further notifications, thus we have to free job related resources diff --git a/drivers/accel/ivpu/ivpu_job.h b/drivers/accel/ivpu/ivpu_job.h index 2e301c2eea7b..6c8b9c739b51 100644 --- a/drivers/accel/ivpu/ivpu_job.h +++ b/drivers/accel/ivpu/ivpu_job.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0-only */ /* - * Copyright (C) 2020-2024 Intel Corporation + * Copyright (C) 2020-2025 Intel Corporation */ #ifndef __IVPU_JOB_H__ @@ -55,6 +55,8 @@ struct ivpu_job { u32 job_id; u32 engine_idx; size_t bo_count; + struct ivpu_bo *primary_preempt_buf; + struct ivpu_bo *secondary_preempt_buf; struct ivpu_bo *bos[] __counted_by(bo_count); }; diff --git a/drivers/accel/ivpu/vpu_jsm_api.h b/drivers/accel/ivpu/vpu_jsm_api.h index 4b6b2b3d2583..de1b37ea1251 100644 --- a/drivers/accel/ivpu/vpu_jsm_api.h +++ b/drivers/accel/ivpu/vpu_jsm_api.h @@ -1,15 +1,16 @@ /* SPDX-License-Identifier: MIT */ /* - * Copyright (c) 2020-2024, Intel Corporation. + * Copyright (c) 2020-2025, Intel Corporation. + */ + +/** + * @addtogroup Jsm + * @{ */ /** * @file * @brief JSM shared definitions - * - * @ingroup Jsm - * @brief JSM shared definitions - * @{ */ #ifndef VPU_JSM_API_H #define VPU_JSM_API_H @@ -22,12 +23,12 @@ /* * Minor version changes when API backward compatibility is preserved. */ -#define VPU_JSM_API_VER_MINOR 29 +#define VPU_JSM_API_VER_MINOR 32 /* * API header changed (field names, documentation, formatting) but API itself has not been changed */ -#define VPU_JSM_API_VER_PATCH 0 +#define VPU_JSM_API_VER_PATCH 5 /* * Index in the API version table @@ -71,9 +72,12 @@ #define VPU_JSM_STATUS_MVNCI_OUT_OF_RESOURCES 0xAU #define VPU_JSM_STATUS_MVNCI_NOT_IMPLEMENTED 0xBU #define VPU_JSM_STATUS_MVNCI_INTERNAL_ERROR 0xCU -/* Job status returned when the job was preempted mid-inference */ +/* @deprecated (use VPU_JSM_STATUS_PREEMPTED_MID_COMMAND instead) */ #define VPU_JSM_STATUS_PREEMPTED_MID_INFERENCE 0xDU +/* Job status returned when the job was preempted mid-command */ +#define VPU_JSM_STATUS_PREEMPTED_MID_COMMAND 0xDU #define VPU_JSM_STATUS_MVNCI_CONTEXT_VIOLATION_HW 0xEU +#define VPU_JSM_STATUS_MVNCI_PREEMPTION_TIMED_OUT 0xFU /* * Host <-> VPU IPC channels. @@ -134,11 +138,21 @@ enum { * 2. Native fence queues are only supported on VPU 40xx onwards. */ VPU_JOB_QUEUE_FLAGS_USE_NATIVE_FENCE_MASK = (1 << 1U), - /* * Enable turbo mode for testing NPU performance; not recommended for regular usage. */ - VPU_JOB_QUEUE_FLAGS_TURBO_MODE = (1 << 2U) + VPU_JOB_QUEUE_FLAGS_TURBO_MODE = (1 << 2U), + /* + * Queue error detection mode flag + * For 'interactive' queues (this bit not set), the FW will identify queues that have not + * completed a job inside the TDR timeout as in error as part of engine reset sequence. + * For 'non-interactive' queues (this bit set), the FW will identify queues that have not + * progressed the heartbeat inside the non-interactive no-progress timeout as in error as + * part of engine reset sequence. Additionally, there is an upper limit applied to these + * queues: even if they progress the heartbeat, if they run longer than non-interactive + * timeout, then the FW will also identify them as in error. + */ + VPU_JOB_QUEUE_FLAGS_NON_INTERACTIVE = (1 << 3U) }; /* @@ -209,7 +223,7 @@ enum { */ #define VPU_INLINE_CMD_TYPE_FENCE_SIGNAL 0x2 -/* +/** * Job scheduling priority bands for both hardware scheduling and OS scheduling. */ enum vpu_job_scheduling_priority_band { @@ -220,16 +234,16 @@ enum vpu_job_scheduling_priority_band { VPU_JOB_SCHEDULING_PRIORITY_BAND_COUNT = 4, }; -/* +/** * Job format. * Jobs defines the actual workloads to be executed by a given engine. */ struct vpu_job_queue_entry { - /**< Address of VPU commands batch buffer */ + /** Address of VPU commands batch buffer */ u64 batch_buf_addr; - /**< Job ID */ + /** Job ID */ u32 job_id; - /**< Flags bit field, see VPU_JOB_FLAGS_* above */ + /** Flags bit field, see VPU_JOB_FLAGS_* above */ u32 flags; /** * Doorbell ring timestamp taken by KMD from SoC's global system clock, in @@ -237,20 +251,20 @@ struct vpu_job_queue_entry { * to match other profiling timestamps. */ u64 doorbell_timestamp; - /**< Extra id for job tracking, used only in the firmware perf traces */ + /** Extra id for job tracking, used only in the firmware perf traces */ u64 host_tracking_id; - /**< Address of the primary preemption buffer to use for this job */ + /** Address of the primary preemption buffer to use for this job */ u64 primary_preempt_buf_addr; - /**< Size of the primary preemption buffer to use for this job */ + /** Size of the primary preemption buffer to use for this job */ u32 primary_preempt_buf_size; - /**< Size of secondary preemption buffer to use for this job */ + /** Size of secondary preemption buffer to use for this job */ u32 secondary_preempt_buf_size; - /**< Address of secondary preemption buffer to use for this job */ + /** Address of secondary preemption buffer to use for this job */ u64 secondary_preempt_buf_addr; u64 reserved_0; }; -/* +/** * Inline command format. * Inline commands are the commands executed at scheduler level (typically, * synchronization directives). Inline command and job objects must be of @@ -258,34 +272,36 @@ struct vpu_job_queue_entry { */ struct vpu_inline_cmd { u64 reserved_0; - /* Inline command type, see VPU_INLINE_CMD_TYPE_* defines. */ + /** Inline command type, see VPU_INLINE_CMD_TYPE_* defines. */ u32 type; - /* Flags bit field, see VPU_JOB_FLAGS_* above. */ + /** Flags bit field, see VPU_JOB_FLAGS_* above. */ u32 flags; - /* Inline command payload. Depends on inline command type. */ - union { - /* Fence (wait and signal) commands' payload. */ - struct { - /* Fence object handle. */ + /** Inline command payload. Depends on inline command type. */ + union payload { + /** Fence (wait and signal) commands' payload. */ + struct fence { + /** Fence object handle. */ u64 fence_handle; - /* User VA of the current fence value. */ + /** User VA of the current fence value. */ u64 current_value_va; - /* User VA of the monitored fence value (read-only). */ + /** User VA of the monitored fence value (read-only). */ u64 monitored_value_va; - /* Value to wait for or write in fence location. */ + /** Value to wait for or write in fence location. */ u64 value; - /* User VA of the log buffer in which to add log entry on completion. */ + /** User VA of the log buffer in which to add log entry on completion. */ u64 log_buffer_va; - /* NPU private data. */ + /** NPU private data. */ u64 npu_private_data; } fence; - /* Other commands do not have a payload. */ - /* Payload definition for future inline commands can be inserted here. */ + /** + * Other commands do not have a payload: + * Payload definition for future inline commands can be inserted here. + */ u64 reserved_1[6]; } payload; }; -/* +/** * Job queue slots can be populated either with job objects or inline command objects. */ union vpu_jobq_slot { @@ -293,7 +309,7 @@ union vpu_jobq_slot { struct vpu_inline_cmd inline_cmd; }; -/* +/** * Job queue control registers. */ struct vpu_job_queue_header { @@ -301,18 +317,18 @@ struct vpu_job_queue_header { u32 head; u32 tail; u32 flags; - /* Set to 1 to indicate priority_band field is valid */ + /** Set to 1 to indicate priority_band field is valid */ u32 priority_band_valid; - /* + /** * Priority for the work of this job queue, valid only if the HWS is NOT used - * and the `priority_band_valid` is set to 1. It is applied only during - * the VPU_JSM_MSG_REGISTER_DB message processing. - * The device firmware might use the `priority_band` to optimize the power + * and the @ref priority_band_valid is set to 1. It is applied only during + * the @ref VPU_JSM_MSG_REGISTER_DB message processing. + * The device firmware might use the priority_band to optimize the power * management logic, but it will not affect the order of jobs. * Available priority bands: @see enum vpu_job_scheduling_priority_band */ u32 priority_band; - /* Inside realtime band assigns a further priority, limited to 0..31 range */ + /** Inside realtime band assigns a further priority, limited to 0..31 range */ u32 realtime_priority_level; u32 reserved_0[9]; }; @@ -337,16 +353,16 @@ enum vpu_trace_entity_type { VPU_TRACE_ENTITY_TYPE_HW_COMPONENT = 2, }; -/* +/** * HWS specific log buffer header details. * Total size is 32 bytes. */ struct vpu_hws_log_buffer_header { - /* Written by VPU after adding a log entry. Initialised by host to 0. */ + /** Written by VPU after adding a log entry. Initialised by host to 0. */ u32 first_free_entry_index; - /* Incremented by VPU every time the VPU writes the 0th entry; initialised by host to 0. */ + /** Incremented by VPU every time the VPU writes the 0th entry; initialised by host to 0. */ u32 wraparound_count; - /* + /** * This is the number of buffers that can be stored in the log buffer provided by the host. * It is written by host before passing buffer to VPU. VPU should consider it read-only. */ @@ -354,14 +370,14 @@ struct vpu_hws_log_buffer_header { u64 reserved[2]; }; -/* +/** * HWS specific log buffer entry details. * Total size is 32 bytes. */ struct vpu_hws_log_buffer_entry { - /* VPU timestamp must be an invariant timer tick (not impacted by DVFS) */ + /** VPU timestamp must be an invariant timer tick (not impacted by DVFS) */ u64 vpu_timestamp; - /* + /** * Operation type: * 0 - context state change * 1 - queue new work @@ -371,7 +387,7 @@ struct vpu_hws_log_buffer_entry { */ u32 operation_type; u32 reserved; - /* Operation data depends on operation type */ + /** Operation data depends on operation type */ u64 operation_data[2]; }; @@ -381,51 +397,54 @@ enum vpu_hws_native_fence_log_type { VPU_HWS_NATIVE_FENCE_LOG_TYPE_SIGNALS = 2 }; -/* HWS native fence log buffer header. */ +/** HWS native fence log buffer header. */ struct vpu_hws_native_fence_log_header { union { struct { - /* Index of the first free entry in buffer. */ + /** Index of the first free entry in buffer. */ u32 first_free_entry_idx; - /* Incremented each time NPU wraps around the buffer to write next entry. */ + /** + * Incremented each time NPU wraps around + * the buffer to write next entry. + */ u32 wraparound_count; }; - /* Field allowing atomic update of both fields above. */ + /** Field allowing atomic update of both fields above. */ u64 atomic_wraparound_and_entry_idx; }; - /* Log buffer type, see enum vpu_hws_native_fence_log_type. */ + /** Log buffer type, see enum vpu_hws_native_fence_log_type. */ u64 type; - /* Allocated number of entries in the log buffer. */ + /** Allocated number of entries in the log buffer. */ u64 entry_nb; u64 reserved[2]; }; -/* Native fence log operation types. */ +/** Native fence log operation types. */ enum vpu_hws_native_fence_log_op { VPU_HWS_NATIVE_FENCE_LOG_OP_SIGNAL_EXECUTED = 0, VPU_HWS_NATIVE_FENCE_LOG_OP_WAIT_UNBLOCKED = 1 }; -/* HWS native fence log entry. */ +/** HWS native fence log entry. */ struct vpu_hws_native_fence_log_entry { - /* Newly signaled/unblocked fence value. */ + /** Newly signaled/unblocked fence value. */ u64 fence_value; - /* Native fence object handle to which this operation belongs. */ + /** Native fence object handle to which this operation belongs. */ u64 fence_handle; - /* Operation type, see enum vpu_hws_native_fence_log_op. */ + /** Operation type, see enum vpu_hws_native_fence_log_op. */ u64 op_type; u64 reserved_0; - /* + /** * VPU_HWS_NATIVE_FENCE_LOG_OP_WAIT_UNBLOCKED only: Timestamp at which fence * wait was started (in NPU SysTime). */ u64 fence_wait_start_ts; u64 reserved_1; - /* Timestamp at which fence operation was completed (in NPU SysTime). */ + /** Timestamp at which fence operation was completed (in NPU SysTime). */ u64 fence_end_ts; }; -/* Native fence log buffer. */ +/** Native fence log buffer. */ struct vpu_hws_native_fence_log_buffer { struct vpu_hws_native_fence_log_header header; struct vpu_hws_native_fence_log_entry entry[]; @@ -450,8 +469,21 @@ enum vpu_ipc_msg_type { * after preemption or when resubmitting jobs to the queue. */ VPU_JSM_MSG_ENGINE_PREEMPT = 0x1101, + /** + * OS scheduling doorbell register command + * @see vpu_ipc_msg_payload_register_db + */ VPU_JSM_MSG_REGISTER_DB = 0x1102, + /** + * OS scheduling doorbell unregister command + * @see vpu_ipc_msg_payload_unregister_db + */ VPU_JSM_MSG_UNREGISTER_DB = 0x1103, + /** + * Query engine heartbeat. Heartbeat is expected to increase monotonically + * and increase while work is being progressed by NPU. + * @see vpu_ipc_msg_payload_query_engine_hb + */ VPU_JSM_MSG_QUERY_ENGINE_HB = 0x1104, VPU_JSM_MSG_GET_POWER_LEVEL_COUNT = 0x1105, VPU_JSM_MSG_GET_POWER_LEVEL = 0x1106, @@ -477,6 +509,7 @@ enum vpu_ipc_msg_type { * aborted and removed from internal scheduling queues. All doorbells assigned * to the host_ssid are unregistered and any internal FW resources belonging to * the host_ssid are released. + * @see vpu_ipc_msg_payload_ssid_release */ VPU_JSM_MSG_SSID_RELEASE = 0x110e, /** @@ -504,26 +537,51 @@ enum vpu_ipc_msg_type { * @see vpu_jsm_metric_streamer_start */ VPU_JSM_MSG_METRIC_STREAMER_INFO = 0x1112, - /** Control command: Priority band setup */ + /** + * Control command: Priority band setup + * @see vpu_ipc_msg_payload_hws_priority_band_setup + */ VPU_JSM_MSG_SET_PRIORITY_BAND_SETUP = 0x1113, - /** Control command: Create command queue */ + /** + * Control command: Create command queue + * @see vpu_ipc_msg_payload_hws_create_cmdq + */ VPU_JSM_MSG_CREATE_CMD_QUEUE = 0x1114, - /** Control command: Destroy command queue */ + /** + * Control command: Destroy command queue + * @see vpu_ipc_msg_payload_hws_destroy_cmdq + */ VPU_JSM_MSG_DESTROY_CMD_QUEUE = 0x1115, - /** Control command: Set context scheduling properties */ + /** + * Control command: Set context scheduling properties + * @see vpu_ipc_msg_payload_hws_set_context_sched_properties + */ VPU_JSM_MSG_SET_CONTEXT_SCHED_PROPERTIES = 0x1116, - /* + /** * Register a doorbell to notify VPU of new work. The doorbell may later be * deallocated or reassigned to another context. + * @see vpu_jsm_hws_register_db */ VPU_JSM_MSG_HWS_REGISTER_DB = 0x1117, - /** Control command: Log buffer setting */ + /** + * Control command: Log buffer setting + * @see vpu_ipc_msg_payload_hws_set_scheduling_log + */ VPU_JSM_MSG_HWS_SET_SCHEDULING_LOG = 0x1118, - /* Control command: Suspend command queue. */ + /** + * Control command: Suspend command queue. + * @see vpu_ipc_msg_payload_hws_suspend_cmdq + */ VPU_JSM_MSG_HWS_SUSPEND_CMDQ = 0x1119, - /* Control command: Resume command queue */ + /** + * Control command: Resume command queue + * @see vpu_ipc_msg_payload_hws_resume_cmdq + */ VPU_JSM_MSG_HWS_RESUME_CMDQ = 0x111a, - /* Control command: Resume engine after reset */ + /** + * Control command: Resume engine after reset + * @see vpu_ipc_msg_payload_hws_resume_engine + */ VPU_JSM_MSG_HWS_ENGINE_RESUME = 0x111b, /* Control command: Enable survivability/DCT mode */ VPU_JSM_MSG_DCT_ENABLE = 0x111c, @@ -540,7 +598,8 @@ enum vpu_ipc_msg_type { VPU_JSM_MSG_BLOB_DEINIT_DEPRECATED = VPU_JSM_MSG_GENERAL_CMD, /** * Control dyndbg behavior by executing a dyndbg command; equivalent to - * Linux command: `echo '<dyndbg_cmd>' > <debugfs>/dynamic_debug/control`. + * Linux command: + * @verbatim echo '<dyndbg_cmd>' > <debugfs>/dynamic_debug/control @endverbatim */ VPU_JSM_MSG_DYNDBG_CONTROL = 0x1201, /** @@ -550,15 +609,26 @@ enum vpu_ipc_msg_type { /* IPC Device -> Host, Job completion */ VPU_JSM_MSG_JOB_DONE = 0x2100, - /* IPC Device -> Host, Fence signalled */ + /** + * IPC Device -> Host, Fence signalled + * @see vpu_ipc_msg_payload_native_fence_signalled + */ VPU_JSM_MSG_NATIVE_FENCE_SIGNALLED = 0x2101, /* IPC Device -> Host, Async command completion */ VPU_JSM_MSG_ASYNC_CMD_DONE = 0x2200, + /** + * IPC Device -> Host, engine reset complete + * @see vpu_ipc_msg_payload_engine_reset_done + */ VPU_JSM_MSG_ENGINE_RESET_DONE = VPU_JSM_MSG_ASYNC_CMD_DONE, VPU_JSM_MSG_ENGINE_PREEMPT_DONE = 0x2201, VPU_JSM_MSG_REGISTER_DB_DONE = 0x2202, VPU_JSM_MSG_UNREGISTER_DB_DONE = 0x2203, + /** + * Response to query engine heartbeat. + * @see vpu_ipc_msg_payload_query_engine_hb_done + */ VPU_JSM_MSG_QUERY_ENGINE_HB_DONE = 0x2204, VPU_JSM_MSG_GET_POWER_LEVEL_COUNT_DONE = 0x2205, VPU_JSM_MSG_GET_POWER_LEVEL_DONE = 0x2206, @@ -575,7 +645,10 @@ enum vpu_ipc_msg_type { VPU_JSM_MSG_TRACE_GET_CAPABILITY_RSP = 0x220c, /** Response to VPU_JSM_MSG_TRACE_GET_NAME. */ VPU_JSM_MSG_TRACE_GET_NAME_RSP = 0x220d, - /** Response to VPU_JSM_MSG_SSID_RELEASE. */ + /** + * Response to VPU_JSM_MSG_SSID_RELEASE. + * @see vpu_ipc_msg_payload_ssid_release + */ VPU_JSM_MSG_SSID_RELEASE_DONE = 0x220e, /** * Response to VPU_JSM_MSG_METRIC_STREAMER_START. @@ -605,29 +678,56 @@ enum vpu_ipc_msg_type { /** * Asynchronous event sent from the VPU to the host either when the current * metric buffer is full or when the VPU has collected a multiple of - * @notify_sample_count samples as indicated through the start command - * (VPU_JSM_MSG_METRIC_STREAMER_START). Returns information about collected - * metric data. + * @ref vpu_jsm_metric_streamer_start::notify_sample_count samples as indicated + * through the start command (VPU_JSM_MSG_METRIC_STREAMER_START). Returns + * information about collected metric data. * @see vpu_jsm_metric_streamer_done */ VPU_JSM_MSG_METRIC_STREAMER_NOTIFICATION = 0x2213, - /** Response to control command: Priority band setup */ + /** + * Response to control command: Priority band setup + * @see vpu_ipc_msg_payload_hws_priority_band_setup + */ VPU_JSM_MSG_SET_PRIORITY_BAND_SETUP_RSP = 0x2214, - /** Response to control command: Create command queue */ + /** + * Response to control command: Create command queue + * @see vpu_ipc_msg_payload_hws_create_cmdq_rsp + */ VPU_JSM_MSG_CREATE_CMD_QUEUE_RSP = 0x2215, - /** Response to control command: Destroy command queue */ + /** + * Response to control command: Destroy command queue + * @see vpu_ipc_msg_payload_hws_destroy_cmdq + */ VPU_JSM_MSG_DESTROY_CMD_QUEUE_RSP = 0x2216, - /** Response to control command: Set context scheduling properties */ + /** + * Response to control command: Set context scheduling properties + * @see vpu_ipc_msg_payload_hws_set_context_sched_properties + */ VPU_JSM_MSG_SET_CONTEXT_SCHED_PROPERTIES_RSP = 0x2217, - /** Response to control command: Log buffer setting */ + /** + * Response to control command: Log buffer setting + * @see vpu_ipc_msg_payload_hws_set_scheduling_log + */ VPU_JSM_MSG_HWS_SET_SCHEDULING_LOG_RSP = 0x2218, - /* IPC Device -> Host, HWS notify index entry of log buffer written */ + /** + * IPC Device -> Host, HWS notify index entry of log buffer written + * @see vpu_ipc_msg_payload_hws_scheduling_log_notification + */ VPU_JSM_MSG_HWS_SCHEDULING_LOG_NOTIFICATION = 0x2219, - /* IPC Device -> Host, HWS completion of a context suspend request */ + /** + * IPC Device -> Host, HWS completion of a context suspend request + * @see vpu_ipc_msg_payload_hws_suspend_cmdq + */ VPU_JSM_MSG_HWS_SUSPEND_CMDQ_DONE = 0x221a, - /* Response to control command: Resume command queue */ + /** + * Response to control command: Resume command queue + * @see vpu_ipc_msg_payload_hws_resume_cmdq + */ VPU_JSM_MSG_HWS_RESUME_CMDQ_RSP = 0x221b, - /* Response to control command: Resume engine command response */ + /** + * Response to control command: Resume engine command response + * @see vpu_ipc_msg_payload_hws_resume_engine + */ VPU_JSM_MSG_HWS_RESUME_ENGINE_DONE = 0x221c, /* Response to control command: Enable survivability/DCT mode */ VPU_JSM_MSG_DCT_ENABLE_DONE = 0x221d, @@ -670,40 +770,44 @@ struct vpu_ipc_msg_payload_engine_preempt { u32 preempt_id; }; -/* - * @brief Register doorbell command structure. +/** + * Register doorbell command structure. * This structure supports doorbell registration for only OS scheduling. * @see VPU_JSM_MSG_REGISTER_DB */ struct vpu_ipc_msg_payload_register_db { - /* Index of the doorbell to register. */ + /** Index of the doorbell to register. */ u32 db_idx; - /* Reserved */ + /** Reserved */ u32 reserved_0; - /* Virtual address in Global GTT pointing to the start of job queue. */ + /** Virtual address in Global GTT pointing to the start of job queue. */ u64 jobq_base; - /* Size of the job queue in bytes. */ + /** Size of the job queue in bytes. */ u32 jobq_size; - /* Host sub-stream ID for the context assigned to the doorbell. */ + /** Host sub-stream ID for the context assigned to the doorbell. */ u32 host_ssid; }; /** - * @brief Unregister doorbell command structure. + * Unregister doorbell command structure. * Request structure to unregister a doorbell for both HW and OS scheduling. * @see VPU_JSM_MSG_UNREGISTER_DB */ struct vpu_ipc_msg_payload_unregister_db { - /* Index of the doorbell to unregister. */ + /** Index of the doorbell to unregister. */ u32 db_idx; - /* Reserved */ + /** Reserved */ u32 reserved_0; }; +/** + * Heartbeat request structure + * @see VPU_JSM_MSG_QUERY_ENGINE_HB + */ struct vpu_ipc_msg_payload_query_engine_hb { - /* Engine to return heartbeat value. */ + /** Engine to return heartbeat value. */ u32 engine_idx; - /* Reserved */ + /** Reserved */ u32 reserved_0; }; @@ -723,10 +827,14 @@ struct vpu_ipc_msg_payload_power_level { u32 reserved_0; }; +/** + * Structure for requesting ssid release + * @see VPU_JSM_MSG_SSID_RELEASE + */ struct vpu_ipc_msg_payload_ssid_release { - /* Host sub-stream ID for the context to be released. */ + /** Host sub-stream ID for the context to be released. */ u32 host_ssid; - /* Reserved */ + /** Reserved */ u32 reserved_0; }; @@ -752,7 +860,7 @@ struct vpu_jsm_metric_streamer_start { u64 sampling_rate; /** * If > 0 the VPU will send a VPU_JSM_MSG_METRIC_STREAMER_NOTIFICATION message - * after every @notify_sample_count samples is collected or dropped by the VPU. + * after every @ref notify_sample_count samples is collected or dropped by the VPU. * If set to UINT_MAX the VPU will only generate a notification when the metric * buffer is full. If set to 0 the VPU will never generate a notification. */ @@ -762,9 +870,9 @@ struct vpu_jsm_metric_streamer_start { * Address and size of the buffer where the VPU will write metric data. The * VPU writes all counters from enabled metric groups one after another. If * there is no space left to write data at the next sample period the VPU - * will switch to the next buffer (@see next_buffer_addr) and will optionally - * send a notification to the host driver if @notify_sample_count is non-zero. - * If @next_buffer_addr is NULL the VPU will stop collecting metric data. + * will switch to the next buffer (@ref next_buffer_addr) and will optionally + * send a notification to the host driver if @ref notify_sample_count is non-zero. + * If @ref next_buffer_addr is NULL the VPU will stop collecting metric data. */ u64 buffer_addr; u64 buffer_size; @@ -844,38 +952,47 @@ struct vpu_ipc_msg_payload_job_done { u64 cmdq_id; }; -/* +/** * Notification message upon native fence signalling. * @see VPU_JSM_MSG_NATIVE_FENCE_SIGNALLED */ struct vpu_ipc_msg_payload_native_fence_signalled { - /* Engine ID. */ + /** Engine ID. */ u32 engine_idx; - /* Host SSID. */ + /** Host SSID. */ u32 host_ssid; - /* CMDQ ID */ + /** CMDQ ID */ u64 cmdq_id; - /* Fence object handle. */ + /** Fence object handle. */ u64 fence_handle; }; +/** + * vpu_ipc_msg_payload_engine_reset_done will contain an array of this structure + * which contains which queues caused reset if FW was able to detect any error. + * @see vpu_ipc_msg_payload_engine_reset_done + */ struct vpu_jsm_engine_reset_context { - /* Host SSID */ + /** Host SSID */ u32 host_ssid; - /* Zero Padding */ + /** Zero Padding */ u32 reserved_0; - /* Command queue id */ + /** Command queue id */ u64 cmdq_id; - /* See VPU_ENGINE_RESET_CONTEXT_* defines */ + /** See VPU_ENGINE_RESET_CONTEXT_* defines */ u64 flags; }; +/** + * Engine reset response. + * @see VPU_JSM_MSG_ENGINE_RESET_DONE + */ struct vpu_ipc_msg_payload_engine_reset_done { - /* Engine ordinal */ + /** Engine ordinal */ u32 engine_idx; - /* Number of impacted contexts */ + /** Number of impacted contexts */ u32 num_impacted_contexts; - /* Array of impacted command queue ids and their flags */ + /** Array of impacted command queue ids and their flags */ struct vpu_jsm_engine_reset_context impacted_contexts[VPU_MAX_ENGINE_RESET_IMPACTED_CONTEXTS]; }; @@ -912,12 +1029,16 @@ struct vpu_ipc_msg_payload_unregister_db_done { u32 reserved_0; }; +/** + * Structure for heartbeat response + * @see VPU_JSM_MSG_QUERY_ENGINE_HB_DONE + */ struct vpu_ipc_msg_payload_query_engine_hb_done { - /* Engine returning heartbeat value. */ + /** Engine returning heartbeat value. */ u32 engine_idx; - /* Reserved */ + /** Reserved */ u32 reserved_0; - /* Heartbeat value. */ + /** Heartbeat value. */ u64 heartbeat; }; @@ -937,7 +1058,10 @@ struct vpu_ipc_msg_payload_get_power_level_count_done { u8 power_limit[16]; }; -/* HWS priority band setup request / response */ +/** + * HWS priority band setup request / response + * @see VPU_JSM_MSG_SET_PRIORITY_BAND_SETUP + */ struct vpu_ipc_msg_payload_hws_priority_band_setup { /* * Grace period in 100ns units when preempting another priority band for @@ -964,15 +1088,23 @@ struct vpu_ipc_msg_payload_hws_priority_band_setup { * TDR timeout value in milliseconds. Default value of 0 meaning no timeout. */ u32 tdr_timeout; + /* Non-interactive queue timeout for no progress of heartbeat in milliseconds. + * Default value of 0 meaning no timeout. + */ + u32 non_interactive_no_progress_timeout; + /* + * Non-interactive queue upper limit timeout value in milliseconds. Default + * value of 0 meaning no timeout. + */ + u32 non_interactive_timeout; }; -/* +/** * @brief HWS create command queue request. * Host will create a command queue via this command. * Note: Cmdq group is a handle of an object which * may contain one or more command queues. * @see VPU_JSM_MSG_CREATE_CMD_QUEUE - * @see VPU_JSM_MSG_CREATE_CMD_QUEUE_RSP */ struct vpu_ipc_msg_payload_hws_create_cmdq { /* Process id */ @@ -993,66 +1125,73 @@ struct vpu_ipc_msg_payload_hws_create_cmdq { u32 reserved_0; }; -/* - * @brief HWS create command queue response. - * @see VPU_JSM_MSG_CREATE_CMD_QUEUE +/** + * HWS create command queue response. * @see VPU_JSM_MSG_CREATE_CMD_QUEUE_RSP */ struct vpu_ipc_msg_payload_hws_create_cmdq_rsp { - /* Process id */ + /** Process id */ u64 process_id; - /* Host SSID */ + /** Host SSID */ u32 host_ssid; - /* Engine for which queue is being created */ + /** Engine for which queue is being created */ u32 engine_idx; - /* Command queue group */ + /** Command queue group */ u64 cmdq_group; - /* Command queue id */ + /** Command queue id */ u64 cmdq_id; }; -/* HWS destroy command queue request / response */ +/** + * HWS destroy command queue request / response + * @see VPU_JSM_MSG_DESTROY_CMD_QUEUE + * @see VPU_JSM_MSG_DESTROY_CMD_QUEUE_RSP + */ struct vpu_ipc_msg_payload_hws_destroy_cmdq { - /* Host SSID */ + /** Host SSID */ u32 host_ssid; - /* Zero Padding */ + /** Zero Padding */ u32 reserved; - /* Command queue id */ + /** Command queue id */ u64 cmdq_id; }; -/* HWS set context scheduling properties request / response */ +/** + * HWS set context scheduling properties request / response + * @see VPU_JSM_MSG_SET_CONTEXT_SCHED_PROPERTIES + * @see VPU_JSM_MSG_SET_CONTEXT_SCHED_PROPERTIES_RSP + */ struct vpu_ipc_msg_payload_hws_set_context_sched_properties { - /* Host SSID */ + /** Host SSID */ u32 host_ssid; - /* Zero Padding */ + /** Zero Padding */ u32 reserved_0; - /* Command queue id */ + /** Command queue id */ u64 cmdq_id; - /* + /** * Priority band to assign to work of this context. * Available priority bands: @see enum vpu_job_scheduling_priority_band */ u32 priority_band; - /* Inside realtime band assigns a further priority */ + /** Inside realtime band assigns a further priority */ u32 realtime_priority_level; - /* Priority relative to other contexts in the same process */ + /** Priority relative to other contexts in the same process */ s32 in_process_priority; - /* Zero padding / Reserved */ + /** Zero padding / Reserved */ u32 reserved_1; - /* + /** * Context quantum relative to other contexts of same priority in the same process * Minimum value supported by NPU is 1ms (10000 in 100ns units). */ u64 context_quantum; - /* Grace period when preempting context of the same priority within the same process */ + /** Grace period when preempting context of the same priority within the same process */ u64 grace_period_same_priority; - /* Grace period when preempting context of a lower priority within the same process */ + /** Grace period when preempting context of a lower priority within the same process */ u64 grace_period_lower_priority; }; -/* - * @brief Register doorbell command structure. +/** + * Register doorbell command structure. * This structure supports doorbell registration for both HW and OS scheduling. * Note: Queue base and size are added here so that the same structure can be used for * OS scheduling and HW scheduling. For OS scheduling, cmdq_id will be ignored @@ -1061,27 +1200,27 @@ struct vpu_ipc_msg_payload_hws_set_context_sched_properties { * @see VPU_JSM_MSG_HWS_REGISTER_DB */ struct vpu_jsm_hws_register_db { - /* Index of the doorbell to register. */ + /** Index of the doorbell to register. */ u32 db_id; - /* Host sub-stream ID for the context assigned to the doorbell. */ + /** Host sub-stream ID for the context assigned to the doorbell. */ u32 host_ssid; - /* ID of the command queue associated with the doorbell. */ + /** ID of the command queue associated with the doorbell. */ u64 cmdq_id; - /* Virtual address pointing to the start of command queue. */ + /** Virtual address pointing to the start of command queue. */ u64 cmdq_base; - /* Size of the command queue in bytes. */ + /** Size of the command queue in bytes. */ u64 cmdq_size; }; -/* - * @brief Structure to set another buffer to be used for scheduling-related logging. +/** + * Structure to set another buffer to be used for scheduling-related logging. * The size of the logging buffer and the number of entries is defined as part of the * buffer itself as described next. * The log buffer received from the host is made up of; - * - header: 32 bytes in size, as shown in 'struct vpu_hws_log_buffer_header'. + * - header: 32 bytes in size, as shown in @ref vpu_hws_log_buffer_header. * The header contains the number of log entries in the buffer. * - log entry: 0 to n-1, each log entry is 32 bytes in size, as shown in - * 'struct vpu_hws_log_buffer_entry'. + * @ref vpu_hws_log_buffer_entry. * The entry contains the VPU timestamp, operation type and data. * The host should provide the notify index value of log buffer to VPU. This is a * value defined within the log buffer and when written to will generate the @@ -1095,30 +1234,30 @@ struct vpu_jsm_hws_register_db { * @see VPU_JSM_MSG_HWS_SCHEDULING_LOG_NOTIFICATION */ struct vpu_ipc_msg_payload_hws_set_scheduling_log { - /* Engine ordinal */ + /** Engine ordinal */ u32 engine_idx; - /* Host SSID */ + /** Host SSID */ u32 host_ssid; - /* + /** * VPU log buffer virtual address. * Set to 0 to disable logging for this engine. */ u64 vpu_log_buffer_va; - /* + /** * Notify index of log buffer. VPU_JSM_MSG_HWS_SCHEDULING_LOG_NOTIFICATION * is generated when an event log is written to this index. */ u64 notify_index; - /* + /** * Field is now deprecated, will be removed when KMD is updated to support removal */ u32 enable_extra_events; - /* Zero Padding */ + /** Zero Padding */ u32 reserved_0; }; -/* - * @brief The scheduling log notification is generated by VPU when it writes +/** + * The scheduling log notification is generated by VPU when it writes * an event into the log buffer at the notify_index. VPU notifies host with * VPU_JSM_MSG_HWS_SCHEDULING_LOG_NOTIFICATION. This is an asynchronous * message from VPU to host. @@ -1126,14 +1265,14 @@ struct vpu_ipc_msg_payload_hws_set_scheduling_log { * @see VPU_JSM_MSG_HWS_SET_SCHEDULING_LOG */ struct vpu_ipc_msg_payload_hws_scheduling_log_notification { - /* Engine ordinal */ + /** Engine ordinal */ u32 engine_idx; - /* Zero Padding */ + /** Zero Padding */ u32 reserved_0; }; -/* - * @brief HWS suspend command queue request and done structure. +/** + * HWS suspend command queue request and done structure. * Host will request the suspend of contexts and VPU will; * - Suspend all work on this context * - Preempt any running work @@ -1152,21 +1291,21 @@ struct vpu_ipc_msg_payload_hws_scheduling_log_notification { * @see VPU_JSM_MSG_HWS_SUSPEND_CMDQ_DONE */ struct vpu_ipc_msg_payload_hws_suspend_cmdq { - /* Host SSID */ + /** Host SSID */ u32 host_ssid; - /* Zero Padding */ + /** Zero Padding */ u32 reserved_0; - /* Command queue id */ + /** Command queue id */ u64 cmdq_id; - /* + /** * Suspend fence value - reported by the VPU suspend context * completed once suspend is complete. */ u64 suspend_fence_value; }; -/* - * @brief HWS Resume command queue request / response structure. +/** + * HWS Resume command queue request / response structure. * Host will request the resume of a context; * - VPU will resume all work on this context * - Scheduler will allow this context to be scheduled @@ -1174,25 +1313,25 @@ struct vpu_ipc_msg_payload_hws_suspend_cmdq { * @see VPU_JSM_MSG_HWS_RESUME_CMDQ_RSP */ struct vpu_ipc_msg_payload_hws_resume_cmdq { - /* Host SSID */ + /** Host SSID */ u32 host_ssid; - /* Zero Padding */ + /** Zero Padding */ u32 reserved_0; - /* Command queue id */ + /** Command queue id */ u64 cmdq_id; }; -/* - * @brief HWS Resume engine request / response structure. - * After a HWS engine reset, all scheduling is stopped on VPU until a engine resume. +/** + * HWS Resume engine request / response structure. + * After a HWS engine reset, all scheduling is stopped on VPU until an engine resume. * Host shall send this command to resume scheduling of any valid queue. - * @see VPU_JSM_MSG_HWS_RESUME_ENGINE + * @see VPU_JSM_MSG_HWS_ENGINE_RESUME * @see VPU_JSM_MSG_HWS_RESUME_ENGINE_DONE */ struct vpu_ipc_msg_payload_hws_resume_engine { - /* Engine to be resumed */ + /** Engine to be resumed */ u32 engine_idx; - /* Reserved */ + /** Reserved */ u32 reserved_0; }; @@ -1326,7 +1465,7 @@ struct vpu_jsm_metric_streamer_done { /** * Metric group description placed in the metric buffer after successful completion * of the VPU_JSM_MSG_METRIC_STREAMER_INFO command. This is followed by one or more - * @vpu_jsm_metric_counter_descriptor records. + * @ref vpu_jsm_metric_counter_descriptor records. * @see VPU_JSM_MSG_METRIC_STREAMER_INFO */ struct vpu_jsm_metric_group_descriptor { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index c049848a56b2..1c9b5009304e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -198,7 +198,7 @@ static void amdgpu_gem_object_free(struct drm_gem_object *gobj) struct amdgpu_bo *aobj = gem_to_amdgpu_bo(gobj); amdgpu_hmm_unregister(aobj); - ttm_bo_put(&aobj->tbo); + ttm_bo_fini(&aobj->tbo); } int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size, diff --git a/drivers/gpu/drm/bridge/imx/imx8qxp-ldb.c b/drivers/gpu/drm/bridge/imx/imx8qxp-ldb.c index 5d272916e200..122502968927 100644 --- a/drivers/gpu/drm/bridge/imx/imx8qxp-ldb.c +++ b/drivers/gpu/drm/bridge/imx/imx8qxp-ldb.c @@ -683,11 +683,6 @@ static void imx8qxp_ldb_remove(struct platform_device *pdev) pm_runtime_disable(&pdev->dev); } -static int imx8qxp_ldb_runtime_suspend(struct device *dev) -{ - return 0; -} - static int imx8qxp_ldb_runtime_resume(struct device *dev) { struct imx8qxp_ldb *imx8qxp_ldb = dev_get_drvdata(dev); @@ -700,7 +695,7 @@ static int imx8qxp_ldb_runtime_resume(struct device *dev) } static const struct dev_pm_ops imx8qxp_ldb_pm_ops = { - RUNTIME_PM_OPS(imx8qxp_ldb_runtime_suspend, imx8qxp_ldb_runtime_resume, NULL) + RUNTIME_PM_OPS(NULL, imx8qxp_ldb_runtime_resume, NULL) }; static const struct of_device_id imx8qxp_ldb_dt_ids[] = { diff --git a/drivers/gpu/drm/display/drm_bridge_connector.c b/drivers/gpu/drm/display/drm_bridge_connector.c index baacd21e7341..a5bdd6c10643 100644 --- a/drivers/gpu/drm/display/drm_bridge_connector.c +++ b/drivers/gpu/drm/display/drm_bridge_connector.c @@ -137,10 +137,9 @@ static void drm_bridge_connector_hpd_notify(struct drm_connector *connector, { struct drm_bridge_connector *bridge_connector = to_drm_bridge_connector(connector); - struct drm_bridge *bridge; /* Notify all bridges in the pipeline of hotplug events. */ - drm_for_each_bridge_in_chain(bridge_connector->encoder, bridge) { + drm_for_each_bridge_in_chain_scoped(bridge_connector->encoder, bridge) { if (bridge->funcs->hpd_notify) bridge->funcs->hpd_notify(bridge, status); } @@ -639,7 +638,7 @@ struct drm_connector *drm_bridge_connector_init(struct drm_device *drm, struct drm_bridge_connector *bridge_connector; struct drm_connector *connector; struct i2c_adapter *ddc = NULL; - struct drm_bridge *bridge, *panel_bridge = NULL; + struct drm_bridge *panel_bridge = NULL; unsigned int supported_formats = BIT(HDMI_COLORSPACE_RGB); unsigned int max_bpc = 8; bool support_hdcp = false; @@ -667,7 +666,7 @@ struct drm_connector *drm_bridge_connector_init(struct drm_device *drm, * detection are available, we don't support hotplug detection at all. */ connector_type = DRM_MODE_CONNECTOR_Unknown; - drm_for_each_bridge_in_chain(encoder, bridge) { + drm_for_each_bridge_in_chain_scoped(encoder, bridge) { if (!bridge->interlace_allowed) connector->interlace_allowed = false; if (!bridge->ycbcr_420_allowed) @@ -818,7 +817,7 @@ struct drm_connector *drm_bridge_connector_init(struct drm_device *drm, if (bridge_connector->bridge_hdmi_cec && bridge_connector->bridge_hdmi_cec->ops & DRM_BRIDGE_OP_HDMI_CEC_NOTIFIER) { - bridge = bridge_connector->bridge_hdmi_cec; + struct drm_bridge *bridge = bridge_connector->bridge_hdmi_cec; ret = drmm_connector_hdmi_cec_notifier_register(connector, NULL, @@ -829,7 +828,7 @@ struct drm_connector *drm_bridge_connector_init(struct drm_device *drm, if (bridge_connector->bridge_hdmi_cec && bridge_connector->bridge_hdmi_cec->ops & DRM_BRIDGE_OP_HDMI_CEC_ADAPTER) { - bridge = bridge_connector->bridge_hdmi_cec; + struct drm_bridge *bridge = bridge_connector->bridge_hdmi_cec; ret = drmm_connector_hdmi_cec_register(connector, &drm_bridge_connector_hdmi_cec_funcs, diff --git a/drivers/gpu/drm/drm_atomic.c b/drivers/gpu/drm/drm_atomic.c index cd15cf52f0c9..ed5359a71f7e 100644 --- a/drivers/gpu/drm/drm_atomic.c +++ b/drivers/gpu/drm/drm_atomic.c @@ -1308,7 +1308,6 @@ drm_atomic_add_encoder_bridges(struct drm_atomic_state *state, struct drm_encoder *encoder) { struct drm_bridge_state *bridge_state; - struct drm_bridge *bridge; if (!encoder) return 0; @@ -1317,7 +1316,7 @@ drm_atomic_add_encoder_bridges(struct drm_atomic_state *state, "Adding all bridges for [encoder:%d:%s] to %p\n", encoder->base.id, encoder->name, state); - drm_for_each_bridge_in_chain(encoder, bridge) { + drm_for_each_bridge_in_chain_scoped(encoder, bridge) { /* Skip bridges that don't implement the atomic state hooks. */ if (!bridge->funcs->atomic_duplicate_state) continue; diff --git a/drivers/gpu/drm/drm_bridge.c b/drivers/gpu/drm/drm_bridge.c index d031447eebc9..5f265e8b4d49 100644 --- a/drivers/gpu/drm/drm_bridge.c +++ b/drivers/gpu/drm/drm_bridge.c @@ -1121,7 +1121,6 @@ drm_atomic_bridge_propagate_bus_flags(struct drm_bridge *bridge, struct drm_atomic_state *state) { struct drm_bridge_state *bridge_state, *next_bridge_state; - struct drm_bridge *next_bridge; u32 output_flags = 0; bridge_state = drm_atomic_get_new_bridge_state(state, bridge); @@ -1130,7 +1129,7 @@ drm_atomic_bridge_propagate_bus_flags(struct drm_bridge *bridge, if (!bridge_state) return; - next_bridge = drm_bridge_get_next_bridge(bridge); + struct drm_bridge *next_bridge __free(drm_bridge_put) = drm_bridge_get_next_bridge(bridge); /* * Let's try to apply the most common case here, that is, propagate @@ -1480,10 +1479,9 @@ static int encoder_bridges_show(struct seq_file *m, void *data) { struct drm_encoder *encoder = m->private; struct drm_printer p = drm_seq_file_printer(m); - struct drm_bridge *bridge; unsigned int idx = 0; - drm_for_each_bridge_in_chain(encoder, bridge) + drm_for_each_bridge_in_chain_scoped(encoder, bridge) drm_bridge_debugfs_show_bridge(&p, bridge, idx++); return 0; diff --git a/drivers/gpu/drm/drm_client_modeset.c b/drivers/gpu/drm/drm_client_modeset.c index 9c2c3b0c8c47..fc4caf7da5fc 100644 --- a/drivers/gpu/drm/drm_client_modeset.c +++ b/drivers/gpu/drm/drm_client_modeset.c @@ -1293,6 +1293,50 @@ int drm_client_modeset_dpms(struct drm_client_dev *client, int mode) } EXPORT_SYMBOL(drm_client_modeset_dpms); +/** + * drm_client_modeset_wait_for_vblank() - Wait for the next VBLANK to occur + * @client: DRM client + * @crtc_index: The ndex of the CRTC to wait on + * + * Block the caller until the given CRTC has seen a VBLANK. Do nothing + * if the CRTC is disabled. If there's another DRM master present, fail + * with -EBUSY. + * + * Returns: + * 0 on success, or negative error code otherwise. + */ +int drm_client_modeset_wait_for_vblank(struct drm_client_dev *client, unsigned int crtc_index) +{ + struct drm_device *dev = client->dev; + struct drm_crtc *crtc; + int ret; + + /* + * Rate-limit update frequency to vblank. If there's a DRM master + * present, it could interfere while we're waiting for the vblank + * event. Don't wait in this case. + */ + if (!drm_master_internal_acquire(dev)) + return -EBUSY; + + crtc = client->modesets[crtc_index].crtc; + + /* + * Only wait for a vblank event if the CRTC is enabled, otherwise + * just don't do anything, not even report an error. + */ + ret = drm_crtc_vblank_get(crtc); + if (!ret) { + drm_crtc_wait_one_vblank(crtc); + drm_crtc_vblank_put(crtc); + } + + drm_master_internal_release(dev); + + return 0; +} +EXPORT_SYMBOL(drm_client_modeset_wait_for_vblank); + #ifdef CONFIG_DRM_KUNIT_TEST #include "tests/drm_client_modeset_test.c" #endif diff --git a/drivers/gpu/drm/drm_fb_helper.c b/drivers/gpu/drm/drm_fb_helper.c index 11a5b60cb9ce..53e9dc0543de 100644 --- a/drivers/gpu/drm/drm_fb_helper.c +++ b/drivers/gpu/drm/drm_fb_helper.c @@ -368,6 +368,10 @@ static void drm_fb_helper_fb_dirty(struct drm_fb_helper *helper) unsigned long flags; int ret; + mutex_lock(&helper->lock); + drm_client_modeset_wait_for_vblank(&helper->client, 0); + mutex_unlock(&helper->lock); + if (drm_WARN_ON_ONCE(dev, !helper->funcs->fb_dirty)) return; @@ -1068,15 +1072,9 @@ int drm_fb_helper_ioctl(struct fb_info *info, unsigned int cmd, unsigned long arg) { struct drm_fb_helper *fb_helper = info->par; - struct drm_device *dev = fb_helper->dev; - struct drm_crtc *crtc; int ret = 0; - mutex_lock(&fb_helper->lock); - if (!drm_master_internal_acquire(dev)) { - ret = -EBUSY; - goto unlock; - } + guard(mutex)(&fb_helper->lock); switch (cmd) { case FBIO_WAITFORVSYNC: @@ -1096,28 +1094,12 @@ int drm_fb_helper_ioctl(struct fb_info *info, unsigned int cmd, * make. If we're not smart enough here, one should * just consider switch the userspace to KMS. */ - crtc = fb_helper->client.modesets[0].crtc; - - /* - * Only wait for a vblank event if the CRTC is - * enabled, otherwise just don't do anythintg, - * not even report an error. - */ - ret = drm_crtc_vblank_get(crtc); - if (!ret) { - drm_crtc_wait_one_vblank(crtc); - drm_crtc_vblank_put(crtc); - } - - ret = 0; + ret = drm_client_modeset_wait_for_vblank(&fb_helper->client, 0); break; default: ret = -ENOTTY; } - drm_master_internal_release(dev); -unlock: - mutex_unlock(&fb_helper->lock); return ret; } EXPORT_SYMBOL(drm_fb_helper_ioctl); diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c index 8d25cc65707d..cbeb76b2124f 100644 --- a/drivers/gpu/drm/drm_gem.c +++ b/drivers/gpu/drm/drm_gem.c @@ -101,10 +101,8 @@ drm_gem_init(struct drm_device *dev) vma_offset_manager = drmm_kzalloc(dev, sizeof(*vma_offset_manager), GFP_KERNEL); - if (!vma_offset_manager) { - DRM_ERROR("out of memory\n"); + if (!vma_offset_manager) return -ENOMEM; - } dev->vma_offset_manager = vma_offset_manager; drm_vma_offset_manager_init(vma_offset_manager, @@ -785,9 +783,10 @@ static int objects_lookup(struct drm_file *filp, u32 *handle, int count, int drm_gem_objects_lookup(struct drm_file *filp, void __user *bo_handles, int count, struct drm_gem_object ***objs_out) { - int ret; - u32 *handles; + struct drm_device *dev = filp->minor->dev; struct drm_gem_object **objs; + u32 *handles; + int ret; if (!count) return 0; @@ -807,7 +806,7 @@ int drm_gem_objects_lookup(struct drm_file *filp, void __user *bo_handles, if (copy_from_user(handles, bo_handles, count * sizeof(u32))) { ret = -EFAULT; - DRM_DEBUG("Failed to copy in GEM handles\n"); + drm_dbg_core(dev, "Failed to copy in GEM handles\n"); goto out; } @@ -855,12 +854,13 @@ EXPORT_SYMBOL(drm_gem_object_lookup); long drm_gem_dma_resv_wait(struct drm_file *filep, u32 handle, bool wait_all, unsigned long timeout) { - long ret; + struct drm_device *dev = filep->minor->dev; struct drm_gem_object *obj; + long ret; obj = drm_gem_object_lookup(filep, handle); if (!obj) { - DRM_DEBUG("Failed to look up GEM BO %d\n", handle); + drm_dbg_core(dev, "Failed to look up GEM BO %d\n", handle); return -EINVAL; } diff --git a/drivers/gpu/drm/drm_gem_dma_helper.c b/drivers/gpu/drm/drm_gem_dma_helper.c index 4f0320df858f..a507cf517015 100644 --- a/drivers/gpu/drm/drm_gem_dma_helper.c +++ b/drivers/gpu/drm/drm_gem_dma_helper.c @@ -582,7 +582,7 @@ drm_gem_dma_prime_import_sg_table_vmap(struct drm_device *dev, ret = dma_buf_vmap_unlocked(attach->dmabuf, &map); if (ret) { - DRM_ERROR("Failed to vmap PRIME buffer\n"); + drm_err(dev, "Failed to vmap PRIME buffer\n"); return ERR_PTR(ret); } diff --git a/drivers/gpu/drm/drm_gem_shmem_helper.c b/drivers/gpu/drm/drm_gem_shmem_helper.c index 5d1349c34afd..50594cf8e17c 100644 --- a/drivers/gpu/drm/drm_gem_shmem_helper.c +++ b/drivers/gpu/drm/drm_gem_shmem_helper.c @@ -48,28 +48,12 @@ static const struct drm_gem_object_funcs drm_gem_shmem_funcs = { .vm_ops = &drm_gem_shmem_vm_ops, }; -static struct drm_gem_shmem_object * -__drm_gem_shmem_create(struct drm_device *dev, size_t size, bool private, - struct vfsmount *gemfs) +static int __drm_gem_shmem_init(struct drm_device *dev, struct drm_gem_shmem_object *shmem, + size_t size, bool private, struct vfsmount *gemfs) { - struct drm_gem_shmem_object *shmem; - struct drm_gem_object *obj; + struct drm_gem_object *obj = &shmem->base; int ret = 0; - size = PAGE_ALIGN(size); - - if (dev->driver->gem_create_object) { - obj = dev->driver->gem_create_object(dev, size); - if (IS_ERR(obj)) - return ERR_CAST(obj); - shmem = to_drm_gem_shmem_obj(obj); - } else { - shmem = kzalloc(sizeof(*shmem), GFP_KERNEL); - if (!shmem) - return ERR_PTR(-ENOMEM); - obj = &shmem->base; - } - if (!obj->funcs) obj->funcs = &drm_gem_shmem_funcs; @@ -81,7 +65,7 @@ __drm_gem_shmem_create(struct drm_device *dev, size_t size, bool private, } if (ret) { drm_gem_private_object_fini(obj); - goto err_free; + return ret; } ret = drm_gem_create_mmap_offset(obj); @@ -102,14 +86,55 @@ __drm_gem_shmem_create(struct drm_device *dev, size_t size, bool private, __GFP_RETRY_MAYFAIL | __GFP_NOWARN); } - return shmem; - + return 0; err_release: drm_gem_object_release(obj); -err_free: - kfree(obj); + return ret; +} - return ERR_PTR(ret); +/** + * drm_gem_shmem_init - Initialize an allocated object. + * @dev: DRM device + * @obj: The allocated shmem GEM object. + * + * Returns: + * 0 on success, or a negative error code on failure. + */ +int drm_gem_shmem_init(struct drm_device *dev, struct drm_gem_shmem_object *shmem, size_t size) +{ + return __drm_gem_shmem_init(dev, shmem, size, false, NULL); +} +EXPORT_SYMBOL_GPL(drm_gem_shmem_init); + +static struct drm_gem_shmem_object * +__drm_gem_shmem_create(struct drm_device *dev, size_t size, bool private, + struct vfsmount *gemfs) +{ + struct drm_gem_shmem_object *shmem; + struct drm_gem_object *obj; + int ret = 0; + + size = PAGE_ALIGN(size); + + if (dev->driver->gem_create_object) { + obj = dev->driver->gem_create_object(dev, size); + if (IS_ERR(obj)) + return ERR_CAST(obj); + shmem = to_drm_gem_shmem_obj(obj); + } else { + shmem = kzalloc(sizeof(*shmem), GFP_KERNEL); + if (!shmem) + return ERR_PTR(-ENOMEM); + obj = &shmem->base; + } + + ret = __drm_gem_shmem_init(dev, shmem, size, private, gemfs); + if (ret) { + kfree(obj); + return ERR_PTR(ret); + } + + return shmem; } /** * drm_gem_shmem_create - Allocate an object with the given size @@ -150,13 +175,13 @@ struct drm_gem_shmem_object *drm_gem_shmem_create_with_mnt(struct drm_device *de EXPORT_SYMBOL_GPL(drm_gem_shmem_create_with_mnt); /** - * drm_gem_shmem_free - Free resources associated with a shmem GEM object - * @shmem: shmem GEM object to free + * drm_gem_shmem_release - Release resources associated with a shmem GEM object. + * @shmem: shmem GEM object * - * This function cleans up the GEM object state and frees the memory used to - * store the object itself. + * This function cleans up the GEM object state, but does not free the memory used to store the + * object itself. This function is meant to be a dedicated helper for the Rust GEM bindings. */ -void drm_gem_shmem_free(struct drm_gem_shmem_object *shmem) +void drm_gem_shmem_release(struct drm_gem_shmem_object *shmem) { struct drm_gem_object *obj = &shmem->base; @@ -183,6 +208,19 @@ void drm_gem_shmem_free(struct drm_gem_shmem_object *shmem) } drm_gem_object_release(obj); +} +EXPORT_SYMBOL_GPL(drm_gem_shmem_release); + +/** + * drm_gem_shmem_free - Free resources associated with a shmem GEM object + * @shmem: shmem GEM object to free + * + * This function cleans up the GEM object state and frees the memory used to + * store the object itself. + */ +void drm_gem_shmem_free(struct drm_gem_shmem_object *shmem) +{ + drm_gem_shmem_release(shmem); kfree(shmem); } EXPORT_SYMBOL_GPL(drm_gem_shmem_free); diff --git a/drivers/gpu/drm/drm_gem_vram_helper.c b/drivers/gpu/drm/drm_gem_vram_helper.c index b04cde4a60e7..90760d0ca071 100644 --- a/drivers/gpu/drm/drm_gem_vram_helper.c +++ b/drivers/gpu/drm/drm_gem_vram_helper.c @@ -107,7 +107,7 @@ static const struct drm_gem_object_funcs drm_gem_vram_object_funcs; static void drm_gem_vram_cleanup(struct drm_gem_vram_object *gbo) { - /* We got here via ttm_bo_put(), which means that the + /* We got here via ttm_bo_fini(), which means that the * TTM buffer object in 'bo' has already been cleaned * up; only release the GEM object. */ @@ -234,11 +234,11 @@ EXPORT_SYMBOL(drm_gem_vram_create); * drm_gem_vram_put() - Releases a reference to a VRAM-backed GEM object * @gbo: the GEM VRAM object * - * See ttm_bo_put() for more information. + * See ttm_bo_fini() for more information. */ void drm_gem_vram_put(struct drm_gem_vram_object *gbo) { - ttm_bo_put(&gbo->bo); + ttm_bo_fini(&gbo->bo); } EXPORT_SYMBOL(drm_gem_vram_put); diff --git a/drivers/gpu/drm/gud/gud_connector.c b/drivers/gpu/drm/gud/gud_connector.c index 4a15695fa933..62e349b06dbe 100644 --- a/drivers/gpu/drm/gud/gud_connector.c +++ b/drivers/gpu/drm/gud/gud_connector.c @@ -561,11 +561,11 @@ static int gud_connector_add_properties(struct gud_device *gdrm, struct gud_conn continue; /* not a DRM property */ property = gud_connector_property_lookup(connector, prop); - if (WARN_ON(IS_ERR(property))) + if (drm_WARN_ON(drm, IS_ERR(property))) continue; state_val = gud_connector_tv_state_val(prop, &gconn->initial_tv_state); - if (WARN_ON(IS_ERR(state_val))) + if (drm_WARN_ON(drm, IS_ERR(state_val))) continue; *state_val = val; @@ -593,7 +593,7 @@ int gud_connector_fill_properties(struct drm_connector_state *connector_state, unsigned int *state_val; state_val = gud_connector_tv_state_val(prop, &connector_state->tv); - if (WARN_ON_ONCE(IS_ERR(state_val))) + if (drm_WARN_ON_ONCE(connector_state->connector->dev, state_val)) return PTR_ERR(state_val); val = *state_val; @@ -667,7 +667,7 @@ static int gud_connector_create(struct gud_device *gdrm, unsigned int index, return ret; } - if (WARN_ON(connector->index != index)) + if (drm_WARN_ON(drm, connector->index != index)) return -EINVAL; if (flags & GUD_CONNECTOR_FLAGS_POLL_STATUS) diff --git a/drivers/gpu/drm/gud/gud_pipe.c b/drivers/gpu/drm/gud/gud_pipe.c index 54d9aa9998e5..3a208e956dff 100644 --- a/drivers/gpu/drm/gud/gud_pipe.c +++ b/drivers/gpu/drm/gud/gud_pipe.c @@ -61,7 +61,7 @@ static size_t gud_xrgb8888_to_r124(u8 *dst, const struct drm_format_info *format size_t len; void *buf; - WARN_ON_ONCE(format->char_per_block[0] != 1); + drm_WARN_ON_ONCE(fb->dev, format->char_per_block[0] != 1); /* Start on a byte boundary */ rect->x1 = ALIGN_DOWN(rect->x1, block_width); @@ -138,7 +138,7 @@ static size_t gud_xrgb8888_to_color(u8 *dst, const struct drm_format_info *forma pix = ((r >> 7) << 2) | ((g >> 7) << 1) | (b >> 7); break; default: - WARN_ON_ONCE(1); + drm_WARN_ON_ONCE(fb->dev, 1); return len; } @@ -527,7 +527,7 @@ int gud_plane_atomic_check(struct drm_plane *plane, drm_connector_list_iter_end(&conn_iter); } - if (WARN_ON_ONCE(!connector_state)) + if (drm_WARN_ON_ONCE(plane->dev, !connector_state)) return -ENOENT; len = struct_size(req, properties, @@ -539,7 +539,7 @@ int gud_plane_atomic_check(struct drm_plane *plane, gud_from_display_mode(&req->mode, mode); req->format = gud_from_fourcc(format->format); - if (WARN_ON_ONCE(!req->format)) { + if (drm_WARN_ON_ONCE(plane->dev, !req->format)) { ret = -EINVAL; goto out; } @@ -561,7 +561,7 @@ int gud_plane_atomic_check(struct drm_plane *plane, val = new_plane_state->rotation; break; default: - WARN_ON_ONCE(1); + drm_WARN_ON_ONCE(plane->dev, 1); ret = -EINVAL; goto out; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c index 1f4814968868..57bb111d65da 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c @@ -1029,7 +1029,7 @@ static void i915_ttm_delayed_free(struct drm_i915_gem_object *obj) { GEM_BUG_ON(!obj->ttm.created); - ttm_bo_put(i915_gem_to_ttm(obj)); + ttm_bo_fini(i915_gem_to_ttm(obj)); } static vm_fault_t vm_fault_ttm(struct vm_fault *vmf) @@ -1325,7 +1325,7 @@ int __i915_gem_ttm_object_init(struct intel_memory_region *mem, * If this function fails, it will call the destructor, but * our caller still owns the object. So no freeing in the * destructor until obj->ttm.created is true. - * Similarly, in delayed_destroy, we can't call ttm_bo_put() + * Similarly, in delayed_destroy, we can't call ttm_bo_fini() * until successful initialization. */ ret = ttm_bo_init_reserved(&i915->bdev, i915_gem_to_ttm(obj), bo_type, diff --git a/drivers/gpu/drm/imx/ipuv3/imx-tve.c b/drivers/gpu/drm/imx/ipuv3/imx-tve.c index c5629e155d25..63f23b821b0b 100644 --- a/drivers/gpu/drm/imx/ipuv3/imx-tve.c +++ b/drivers/gpu/drm/imx/ipuv3/imx-tve.c @@ -368,17 +368,20 @@ static unsigned long clk_tve_di_recalc_rate(struct clk_hw *hw, return 0; } -static long clk_tve_di_round_rate(struct clk_hw *hw, unsigned long rate, - unsigned long *prate) +static int clk_tve_di_determine_rate(struct clk_hw *hw, + struct clk_rate_request *req) { unsigned long div; - div = *prate / rate; + div = req->best_parent_rate / req->rate; if (div >= 4) - return *prate / 4; + req->rate = req->best_parent_rate / 4; else if (div >= 2) - return *prate / 2; - return *prate; + req->rate = req->best_parent_rate / 2; + else + req->rate = req->best_parent_rate; + + return 0; } static int clk_tve_di_set_rate(struct clk_hw *hw, unsigned long rate, @@ -409,7 +412,7 @@ static int clk_tve_di_set_rate(struct clk_hw *hw, unsigned long rate, } static const struct clk_ops clk_tve_di_ops = { - .round_rate = clk_tve_di_round_rate, + .determine_rate = clk_tve_di_determine_rate, .set_rate = clk_tve_di_set_rate, .recalc_rate = clk_tve_di_recalc_rate, }; diff --git a/drivers/gpu/drm/imx/ipuv3/parallel-display.c b/drivers/gpu/drm/imx/ipuv3/parallel-display.c index 6d8325c76697..dfdeb926fe9c 100644 --- a/drivers/gpu/drm/imx/ipuv3/parallel-display.c +++ b/drivers/gpu/drm/imx/ipuv3/parallel-display.c @@ -134,10 +134,10 @@ static int imx_pd_bridge_atomic_check(struct drm_bridge *bridge, struct imx_crtc_state *imx_crtc_state = to_imx_crtc_state(crtc_state); struct drm_display_info *di = &conn_state->connector->display_info; struct drm_bridge_state *next_bridge_state = NULL; - struct drm_bridge *next_bridge; u32 bus_flags, bus_fmt; - next_bridge = drm_bridge_get_next_bridge(bridge); + struct drm_bridge *next_bridge __free(drm_bridge_put) = drm_bridge_get_next_bridge(bridge); + if (next_bridge) next_bridge_state = drm_atomic_get_new_bridge_state(crtc_state->state, next_bridge); diff --git a/drivers/gpu/drm/loongson/lsdc_gem.c b/drivers/gpu/drm/loongson/lsdc_gem.c index a720d8f53209..22d0eced95da 100644 --- a/drivers/gpu/drm/loongson/lsdc_gem.c +++ b/drivers/gpu/drm/loongson/lsdc_gem.c @@ -57,7 +57,7 @@ static void lsdc_gem_object_free(struct drm_gem_object *obj) struct ttm_buffer_object *tbo = to_ttm_bo(obj); if (tbo) - ttm_bo_put(tbo); + ttm_bo_fini(tbo); } static int lsdc_gem_object_vmap(struct drm_gem_object *obj, struct iosys_map *map) diff --git a/drivers/gpu/drm/mcde/mcde_clk_div.c b/drivers/gpu/drm/mcde/mcde_clk_div.c index 3056ac566473..8c5af2677357 100644 --- a/drivers/gpu/drm/mcde/mcde_clk_div.c +++ b/drivers/gpu/drm/mcde/mcde_clk_div.c @@ -71,12 +71,15 @@ static int mcde_clk_div_choose_div(struct clk_hw *hw, unsigned long rate, return best_div; } -static long mcde_clk_div_round_rate(struct clk_hw *hw, unsigned long rate, - unsigned long *prate) +static int mcde_clk_div_determine_rate(struct clk_hw *hw, + struct clk_rate_request *req) { - int div = mcde_clk_div_choose_div(hw, rate, prate, true); + int div = mcde_clk_div_choose_div(hw, req->rate, + &req->best_parent_rate, true); - return DIV_ROUND_UP_ULL(*prate, div); + req->rate = DIV_ROUND_UP_ULL(req->best_parent_rate, div); + + return 0; } static unsigned long mcde_clk_div_recalc_rate(struct clk_hw *hw, @@ -132,7 +135,7 @@ static int mcde_clk_div_set_rate(struct clk_hw *hw, unsigned long rate, static const struct clk_ops mcde_clk_div_ops = { .enable = mcde_clk_div_enable, .recalc_rate = mcde_clk_div_recalc_rate, - .round_rate = mcde_clk_div_round_rate, + .determine_rate = mcde_clk_div_determine_rate, .set_rate = mcde_clk_div_set_rate, }; diff --git a/drivers/gpu/drm/nouveau/Kconfig b/drivers/gpu/drm/nouveau/Kconfig index c88776d1e784..3b5757aed9c8 100644 --- a/drivers/gpu/drm/nouveau/Kconfig +++ b/drivers/gpu/drm/nouveau/Kconfig @@ -28,6 +28,7 @@ config DRM_NOUVEAU select THERMAL if ACPI && X86 select ACPI_VIDEO if ACPI && X86 select SND_HDA_COMPONENT if SND_HDA_CORE + select PM_DEVFREQ if ARCH_TEGRA help Choose this option for open-source NVIDIA support. diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/tegra.h b/drivers/gpu/drm/nouveau/include/nvkm/core/tegra.h index 22f74fc88cd7..57bc542780bb 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/tegra.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/tegra.h @@ -9,6 +9,8 @@ struct nvkm_device_tegra { struct nvkm_device device; struct platform_device *pdev; + void __iomem *regs; + struct reset_control *rst; struct clk *clk; struct clk *clk_ref; diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/clk.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/clk.h index d5d8877064a7..6a09d397c651 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/clk.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/clk.h @@ -134,4 +134,5 @@ int gf100_clk_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct int gk104_clk_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_clk **); int gk20a_clk_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_clk **); int gm20b_clk_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_clk **); +int gp10b_clk_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_clk **); #endif diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.h b/drivers/gpu/drm/nouveau/nouveau_bo.h index d59fd12268b9..6c26beeb427f 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.h +++ b/drivers/gpu/drm/nouveau/nouveau_bo.h @@ -57,7 +57,7 @@ nouveau_bo(struct ttm_buffer_object *bo) static inline void nouveau_bo_fini(struct nouveau_bo *bo) { - ttm_bo_put(&bo->bo); + ttm_bo_fini(&bo->bo); } extern struct ttm_device_funcs nouveau_bo_driver; diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c index 690e10fbf0bd..395d92ab6271 100644 --- a/drivers/gpu/drm/nouveau/nouveau_gem.c +++ b/drivers/gpu/drm/nouveau/nouveau_gem.c @@ -87,7 +87,7 @@ nouveau_gem_object_del(struct drm_gem_object *gem) return; } - ttm_bo_put(&nvbo->bo); + ttm_bo_fini(&nvbo->bo); pm_runtime_mark_last_busy(dev); pm_runtime_put_autosuspend(dev); diff --git a/drivers/gpu/drm/nouveau/nouveau_platform.c b/drivers/gpu/drm/nouveau/nouveau_platform.c index 8d5853deeee4..9fd351273236 100644 --- a/drivers/gpu/drm/nouveau/nouveau_platform.c +++ b/drivers/gpu/drm/nouveau/nouveau_platform.c @@ -21,6 +21,8 @@ */ #include "nouveau_platform.h" +#include <nvkm/subdev/clk/gk20a_devfreq.h> + static int nouveau_platform_probe(struct platform_device *pdev) { const struct nvkm_device_tegra_func *func; @@ -40,6 +42,21 @@ static void nouveau_platform_remove(struct platform_device *pdev) nouveau_drm_device_remove(drm); } +#ifdef CONFIG_PM_SLEEP +static int nouveau_platform_suspend(struct device *dev) +{ + return gk20a_devfreq_suspend(dev); +} + +static int nouveau_platform_resume(struct device *dev) +{ + return gk20a_devfreq_resume(dev); +} + +static SIMPLE_DEV_PM_OPS(nouveau_pm_ops, nouveau_platform_suspend, + nouveau_platform_resume); +#endif + #if IS_ENABLED(CONFIG_OF) static const struct nvkm_device_tegra_func gk20a_platform_data = { .iommu_bit = 34, @@ -81,6 +98,9 @@ struct platform_driver nouveau_platform_driver = { .driver = { .name = "nouveau", .of_match_table = of_match_ptr(nouveau_platform_match), +#ifdef CONFIG_PM_SLEEP + .pm = &nouveau_pm_ops, +#endif }, .probe = nouveau_platform_probe, .remove = nouveau_platform_remove, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c index 3375a59ebf1a..2517b65d8faa 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c @@ -2280,6 +2280,7 @@ nv13b_chipset = { .acr = { 0x00000001, gp10b_acr_new }, .bar = { 0x00000001, gm20b_bar_new }, .bus = { 0x00000001, gf100_bus_new }, + .clk = { 0x00000001, gp10b_clk_new }, .fault = { 0x00000001, gp10b_fault_new }, .fb = { 0x00000001, gp10b_fb_new }, .fuse = { 0x00000001, gm107_fuse_new }, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c index 114e50ca1827..03aa6f09ec89 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c @@ -259,6 +259,10 @@ nvkm_device_tegra_new(const struct nvkm_device_tegra_func *func, tdev->func = func; tdev->pdev = pdev; + tdev->regs = devm_platform_ioremap_resource(pdev, 0); + if (IS_ERR(tdev->regs)) + return PTR_ERR(tdev->regs); + if (func->require_vdd) { tdev->vdd = devm_regulator_get(&pdev->dev, "vdd"); if (IS_ERR(tdev->vdd)) { diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/Kbuild b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/Kbuild index dcecd499d8df..be8f3283ee16 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/Kbuild @@ -10,6 +10,8 @@ nvkm-y += nvkm/subdev/clk/gf100.o nvkm-y += nvkm/subdev/clk/gk104.o nvkm-y += nvkm/subdev/clk/gk20a.o nvkm-y += nvkm/subdev/clk/gm20b.o +nvkm-y += nvkm/subdev/clk/gp10b.o +nvkm-$(CONFIG_PM_DEVFREQ) += nvkm/subdev/clk/gk20a_devfreq.o nvkm-y += nvkm/subdev/clk/pllnv04.o nvkm-y += nvkm/subdev/clk/pllgt215.o diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a.c b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a.c index d573fb0917fc..65f5d0f1f3bf 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a.c @@ -23,6 +23,7 @@ * */ #include "priv.h" +#include "gk20a_devfreq.h" #include "gk20a.h" #include <core/tegra.h> @@ -589,6 +590,10 @@ gk20a_clk_init(struct nvkm_clk *base) return ret; } + ret = gk20a_devfreq_init(base, &clk->devfreq); + if (ret) + return ret; + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a.h b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a.h index 286413ff4a9e..ea5b0bab4cce 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a.h @@ -118,6 +118,7 @@ struct gk20a_clk { const struct gk20a_clk_pllg_params *params; struct gk20a_pll pll; u32 parent_rate; + struct gk20a_devfreq *devfreq; u32 (*div_to_pl)(u32); u32 (*pl_to_div)(u32); diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a_devfreq.c b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a_devfreq.c new file mode 100644 index 000000000000..41003cbcdbfa --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a_devfreq.c @@ -0,0 +1,320 @@ +// SPDX-License-Identifier: MIT +#include <linux/clk.h> +#include <linux/math64.h> +#include <linux/platform_device.h> +#include <linux/pm_opp.h> + +#include <drm/drm_managed.h> + +#include <subdev/clk.h> + +#include "nouveau_drv.h" +#include "nouveau_chan.h" +#include "priv.h" +#include "gk20a_devfreq.h" +#include "gk20a.h" +#include "gp10b.h" + +#define PMU_BUSY_CYCLES_NORM_MAX 1000U + +#define PWR_PMU_IDLE_COUNTER_TOTAL 0U +#define PWR_PMU_IDLE_COUNTER_BUSY 4U + +#define PWR_PMU_IDLE_COUNT_REG_OFFSET 0x0010A508U +#define PWR_PMU_IDLE_COUNT_REG_SIZE 16U +#define PWR_PMU_IDLE_COUNT_MASK 0x7FFFFFFFU +#define PWR_PMU_IDLE_COUNT_RESET_VALUE (0x1U << 31U) + +#define PWR_PMU_IDLE_INTR_REG_OFFSET 0x0010A9E8U +#define PWR_PMU_IDLE_INTR_ENABLE_VALUE 0U + +#define PWR_PMU_IDLE_INTR_STATUS_REG_OFFSET 0x0010A9ECU +#define PWR_PMU_IDLE_INTR_STATUS_MASK 0x00000001U +#define PWR_PMU_IDLE_INTR_STATUS_RESET_VALUE 0x1U + +#define PWR_PMU_IDLE_THRESHOLD_REG_OFFSET 0x0010A8A0U +#define PWR_PMU_IDLE_THRESHOLD_REG_SIZE 4U +#define PWR_PMU_IDLE_THRESHOLD_MAX_VALUE 0x7FFFFFFFU + +#define PWR_PMU_IDLE_CTRL_REG_OFFSET 0x0010A50CU +#define PWR_PMU_IDLE_CTRL_REG_SIZE 16U +#define PWR_PMU_IDLE_CTRL_VALUE_MASK 0x3U +#define PWR_PMU_IDLE_CTRL_VALUE_BUSY 0x2U +#define PWR_PMU_IDLE_CTRL_VALUE_ALWAYS 0x3U +#define PWR_PMU_IDLE_CTRL_FILTER_MASK (0x1U << 2) +#define PWR_PMU_IDLE_CTRL_FILTER_DISABLED 0x0U + +#define PWR_PMU_IDLE_MASK_REG_OFFSET 0x0010A504U +#define PWR_PMU_IDLE_MASK_REG_SIZE 16U +#define PWM_PMU_IDLE_MASK_GR_ENABLED 0x1U +#define PWM_PMU_IDLE_MASK_CE_2_ENABLED 0x200000U + +/** + * struct gk20a_devfreq - Device frequency management + */ +struct gk20a_devfreq { + /** @devfreq: devfreq device. */ + struct devfreq *devfreq; + + /** @regs: Device registers. */ + void __iomem *regs; + + /** @gov_data: Governor data. */ + struct devfreq_simple_ondemand_data gov_data; + + /** @busy_time: Busy time. */ + ktime_t busy_time; + + /** @total_time: Total time. */ + ktime_t total_time; + + /** @time_last_update: Last update time. */ + ktime_t time_last_update; +}; + +static struct gk20a_devfreq *dev_to_gk20a_devfreq(struct device *dev) +{ + struct nouveau_drm *drm = dev_get_drvdata(dev); + struct nvkm_subdev *subdev = nvkm_device_subdev(drm->nvkm, NVKM_SUBDEV_CLK, 0); + struct nvkm_clk *base = nvkm_clk(subdev); + + switch (drm->nvkm->chipset) { + case 0x13b: return gp10b_clk(base)->devfreq; break; + default: return gk20a_clk(base)->devfreq; break; + } +} + +static void gk20a_pmu_init_perfmon_counter(struct gk20a_devfreq *gdevfreq) +{ + u32 data; + + // Set pmu idle intr status bit on total counter overflow + writel(PWR_PMU_IDLE_INTR_ENABLE_VALUE, + gdevfreq->regs + PWR_PMU_IDLE_INTR_REG_OFFSET); + + writel(PWR_PMU_IDLE_THRESHOLD_MAX_VALUE, + gdevfreq->regs + PWR_PMU_IDLE_THRESHOLD_REG_OFFSET + + (PWR_PMU_IDLE_COUNTER_TOTAL * PWR_PMU_IDLE_THRESHOLD_REG_SIZE)); + + // Setup counter for total cycles + data = readl(gdevfreq->regs + PWR_PMU_IDLE_CTRL_REG_OFFSET + + (PWR_PMU_IDLE_COUNTER_TOTAL * PWR_PMU_IDLE_CTRL_REG_SIZE)); + data &= ~(PWR_PMU_IDLE_CTRL_VALUE_MASK | PWR_PMU_IDLE_CTRL_FILTER_MASK); + data |= PWR_PMU_IDLE_CTRL_VALUE_ALWAYS | PWR_PMU_IDLE_CTRL_FILTER_DISABLED; + writel(data, gdevfreq->regs + PWR_PMU_IDLE_CTRL_REG_OFFSET + + (PWR_PMU_IDLE_COUNTER_TOTAL * PWR_PMU_IDLE_CTRL_REG_SIZE)); + + // Setup counter for busy cycles + writel(PWM_PMU_IDLE_MASK_GR_ENABLED | PWM_PMU_IDLE_MASK_CE_2_ENABLED, + gdevfreq->regs + PWR_PMU_IDLE_MASK_REG_OFFSET + + (PWR_PMU_IDLE_COUNTER_BUSY * PWR_PMU_IDLE_MASK_REG_SIZE)); + + data = readl(gdevfreq->regs + PWR_PMU_IDLE_CTRL_REG_OFFSET + + (PWR_PMU_IDLE_COUNTER_BUSY * PWR_PMU_IDLE_CTRL_REG_SIZE)); + data &= ~(PWR_PMU_IDLE_CTRL_VALUE_MASK | PWR_PMU_IDLE_CTRL_FILTER_MASK); + data |= PWR_PMU_IDLE_CTRL_VALUE_BUSY | PWR_PMU_IDLE_CTRL_FILTER_DISABLED; + writel(data, gdevfreq->regs + PWR_PMU_IDLE_CTRL_REG_OFFSET + + (PWR_PMU_IDLE_COUNTER_BUSY * PWR_PMU_IDLE_CTRL_REG_SIZE)); +} + +static u32 gk20a_pmu_read_idle_counter(struct gk20a_devfreq *gdevfreq, u32 counter_id) +{ + u32 ret; + + ret = readl(gdevfreq->regs + PWR_PMU_IDLE_COUNT_REG_OFFSET + + (counter_id * PWR_PMU_IDLE_COUNT_REG_SIZE)); + + return ret & PWR_PMU_IDLE_COUNT_MASK; +} + +static void gk20a_pmu_reset_idle_counter(struct gk20a_devfreq *gdevfreq, u32 counter_id) +{ + writel(PWR_PMU_IDLE_COUNT_RESET_VALUE, gdevfreq->regs + PWR_PMU_IDLE_COUNT_REG_OFFSET + + (counter_id * PWR_PMU_IDLE_COUNT_REG_SIZE)); +} + +static u32 gk20a_pmu_read_idle_intr_status(struct gk20a_devfreq *gdevfreq) +{ + u32 ret; + + ret = readl(gdevfreq->regs + PWR_PMU_IDLE_INTR_STATUS_REG_OFFSET); + + return ret & PWR_PMU_IDLE_INTR_STATUS_MASK; +} + +static void gk20a_pmu_clear_idle_intr_status(struct gk20a_devfreq *gdevfreq) +{ + writel(PWR_PMU_IDLE_INTR_STATUS_RESET_VALUE, + gdevfreq->regs + PWR_PMU_IDLE_INTR_STATUS_REG_OFFSET); +} + +static void gk20a_devfreq_update_utilization(struct gk20a_devfreq *gdevfreq) +{ + ktime_t now, last; + u64 busy_cycles, total_cycles; + u32 norm, intr_status; + + now = ktime_get(); + last = gdevfreq->time_last_update; + gdevfreq->total_time = ktime_us_delta(now, last); + + busy_cycles = gk20a_pmu_read_idle_counter(gdevfreq, PWR_PMU_IDLE_COUNTER_BUSY); + total_cycles = gk20a_pmu_read_idle_counter(gdevfreq, PWR_PMU_IDLE_COUNTER_TOTAL); + intr_status = gk20a_pmu_read_idle_intr_status(gdevfreq); + + gk20a_pmu_reset_idle_counter(gdevfreq, PWR_PMU_IDLE_COUNTER_BUSY); + gk20a_pmu_reset_idle_counter(gdevfreq, PWR_PMU_IDLE_COUNTER_TOTAL); + + if (intr_status != 0UL) { + norm = PMU_BUSY_CYCLES_NORM_MAX; + gk20a_pmu_clear_idle_intr_status(gdevfreq); + } else if (total_cycles == 0ULL || busy_cycles > total_cycles) { + norm = PMU_BUSY_CYCLES_NORM_MAX; + } else { + norm = (u32)div64_u64(busy_cycles * PMU_BUSY_CYCLES_NORM_MAX, + total_cycles); + } + + gdevfreq->busy_time = div_u64(gdevfreq->total_time * norm, PMU_BUSY_CYCLES_NORM_MAX); + gdevfreq->time_last_update = now; +} + +static int gk20a_devfreq_target(struct device *dev, unsigned long *freq, + u32 flags) +{ + struct nouveau_drm *drm = dev_get_drvdata(dev); + struct nvkm_subdev *subdev = nvkm_device_subdev(drm->nvkm, NVKM_SUBDEV_CLK, 0); + struct nvkm_clk *base = nvkm_clk(subdev); + struct nvkm_pstate *pstates = base->func->pstates; + int nr_pstates = base->func->nr_pstates; + int i, ret; + + for (i = 0; i < nr_pstates - 1; i++) + if (pstates[i].base.domain[nv_clk_src_gpc] * GK20A_CLK_GPC_MDIV >= *freq) + break; + + ret = nvkm_clk_ustate(base, pstates[i].pstate, 0); + ret |= nvkm_clk_ustate(base, pstates[i].pstate, 1); + if (ret) { + nvkm_error(subdev, "cannot update clock\n"); + return ret; + } + + *freq = pstates[i].base.domain[nv_clk_src_gpc] * GK20A_CLK_GPC_MDIV; + + return 0; +} + +static int gk20a_devfreq_get_cur_freq(struct device *dev, unsigned long *freq) +{ + struct nouveau_drm *drm = dev_get_drvdata(dev); + struct nvkm_subdev *subdev = nvkm_device_subdev(drm->nvkm, NVKM_SUBDEV_CLK, 0); + struct nvkm_clk *base = nvkm_clk(subdev); + + *freq = nvkm_clk_read(base, nv_clk_src_gpc) * GK20A_CLK_GPC_MDIV; + + return 0; +} + +static void gk20a_devfreq_reset(struct gk20a_devfreq *gdevfreq) +{ + gk20a_pmu_reset_idle_counter(gdevfreq, PWR_PMU_IDLE_COUNTER_BUSY); + gk20a_pmu_reset_idle_counter(gdevfreq, PWR_PMU_IDLE_COUNTER_TOTAL); + gk20a_pmu_clear_idle_intr_status(gdevfreq); + + gdevfreq->busy_time = 0; + gdevfreq->total_time = 0; + gdevfreq->time_last_update = ktime_get(); +} + +static int gk20a_devfreq_get_dev_status(struct device *dev, + struct devfreq_dev_status *status) +{ + struct nouveau_drm *drm = dev_get_drvdata(dev); + struct gk20a_devfreq *gdevfreq = dev_to_gk20a_devfreq(dev); + + gk20a_devfreq_get_cur_freq(dev, &status->current_frequency); + + gk20a_devfreq_update_utilization(gdevfreq); + + status->busy_time = ktime_to_ns(gdevfreq->busy_time); + status->total_time = ktime_to_ns(gdevfreq->total_time); + + gk20a_devfreq_reset(gdevfreq); + + NV_DEBUG(drm, "busy %lu total %lu %lu %% freq %lu MHz\n", + status->busy_time, status->total_time, + status->busy_time / (status->total_time / 100), + status->current_frequency / 1000 / 1000); + + return 0; +} + +static struct devfreq_dev_profile gk20a_devfreq_profile = { + .timer = DEVFREQ_TIMER_DELAYED, + .polling_ms = 50, + .target = gk20a_devfreq_target, + .get_cur_freq = gk20a_devfreq_get_cur_freq, + .get_dev_status = gk20a_devfreq_get_dev_status, +}; + +int gk20a_devfreq_init(struct nvkm_clk *base, struct gk20a_devfreq **gdevfreq) +{ + struct nvkm_device *device = base->subdev.device; + struct nouveau_drm *drm = dev_get_drvdata(device->dev); + struct nvkm_device_tegra *tdev = device->func->tegra(device); + struct nvkm_pstate *pstates = base->func->pstates; + int nr_pstates = base->func->nr_pstates; + struct gk20a_devfreq *new_gdevfreq; + int i; + + new_gdevfreq = drmm_kzalloc(drm->dev, sizeof(struct gk20a_devfreq), GFP_KERNEL); + if (!new_gdevfreq) + return -ENOMEM; + + new_gdevfreq->regs = tdev->regs; + + for (i = 0; i < nr_pstates; i++) + dev_pm_opp_add(base->subdev.device->dev, + pstates[i].base.domain[nv_clk_src_gpc] * GK20A_CLK_GPC_MDIV, 0); + + gk20a_pmu_init_perfmon_counter(new_gdevfreq); + gk20a_devfreq_reset(new_gdevfreq); + + gk20a_devfreq_profile.initial_freq = + nvkm_clk_read(base, nv_clk_src_gpc) * GK20A_CLK_GPC_MDIV; + + new_gdevfreq->gov_data.upthreshold = 45; + new_gdevfreq->gov_data.downdifferential = 5; + + new_gdevfreq->devfreq = devm_devfreq_add_device(device->dev, + &gk20a_devfreq_profile, + DEVFREQ_GOV_SIMPLE_ONDEMAND, + &new_gdevfreq->gov_data); + if (IS_ERR(new_gdevfreq->devfreq)) + return PTR_ERR(new_gdevfreq->devfreq); + + *gdevfreq = new_gdevfreq; + + return 0; +} + +int gk20a_devfreq_resume(struct device *dev) +{ + struct gk20a_devfreq *gdevfreq = dev_to_gk20a_devfreq(dev); + + if (!gdevfreq || !gdevfreq->devfreq) + return 0; + + return devfreq_resume_device(gdevfreq->devfreq); +} + +int gk20a_devfreq_suspend(struct device *dev) +{ + struct gk20a_devfreq *gdevfreq = dev_to_gk20a_devfreq(dev); + + if (!gdevfreq || !gdevfreq->devfreq) + return 0; + + return devfreq_suspend_device(gdevfreq->devfreq); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a_devfreq.h b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a_devfreq.h new file mode 100644 index 000000000000..5b7ca8a7a5cd --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gk20a_devfreq.h @@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __GK20A_DEVFREQ_H__ +#define __GK20A_DEVFREQ_H__ + +#include <linux/devfreq.h> + +struct gk20a_devfreq; + +#if defined(CONFIG_PM_DEVFREQ) +int gk20a_devfreq_init(struct nvkm_clk *base, struct gk20a_devfreq **devfreq); + +int gk20a_devfreq_resume(struct device *dev); +int gk20a_devfreq_suspend(struct device *dev); +#else +static inline int gk20a_devfreq_init(struct nvkm_clk *base, struct gk20a_devfreq **devfreq) +{ + return 0; +} + +static inline int gk20a_devfreq_resume(struct device dev) { return 0; } +static inline int gk20a_devfreq_suspend(struct device *dev) { return 0; } +#endif /* CONFIG_PM_DEVFREQ */ + +#endif /* __GK20A_DEVFREQ_H__ */ diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gm20b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gm20b.c index 7c33542f651b..fa8ca53acbd1 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gm20b.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gm20b.c @@ -27,6 +27,7 @@ #include <core/tegra.h> #include "priv.h" +#include "gk20a_devfreq.h" #include "gk20a.h" #define GPCPLL_CFG_SYNC_MODE BIT(2) @@ -869,6 +870,10 @@ gm20b_clk_init(struct nvkm_clk *base) return ret; } + ret = gk20a_devfreq_init(base, &clk->devfreq); + if (ret) + return ret; + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gp10b.c new file mode 100644 index 000000000000..492b62c0ee96 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gp10b.c @@ -0,0 +1,185 @@ +// SPDX-License-Identifier: MIT +#include <subdev/clk.h> +#include <subdev/timer.h> +#include <core/device.h> +#include <core/tegra.h> + +#include "priv.h" +#include "gk20a_devfreq.h" +#include "gk20a.h" +#include "gp10b.h" + +static int +gp10b_clk_init(struct nvkm_clk *base) +{ + struct gp10b_clk *clk = gp10b_clk(base); + struct nvkm_subdev *subdev = &clk->base.subdev; + int ret; + + /* Start with the highest frequency, matching the BPMP default */ + base->func->calc(base, &base->func->pstates[base->func->nr_pstates - 1].base); + ret = base->func->prog(base); + if (ret) { + nvkm_error(subdev, "cannot initialize clock\n"); + return ret; + } + + ret = gk20a_devfreq_init(base, &clk->devfreq); + if (ret) + return ret; + + return 0; +} + +static int +gp10b_clk_read(struct nvkm_clk *base, enum nv_clk_src src) +{ + struct gp10b_clk *clk = gp10b_clk(base); + struct nvkm_subdev *subdev = &clk->base.subdev; + + switch (src) { + case nv_clk_src_gpc: + return clk_get_rate(clk->clk) / GK20A_CLK_GPC_MDIV; + default: + nvkm_error(subdev, "invalid clock source %d\n", src); + return -EINVAL; + } +} + +static int +gp10b_clk_calc(struct nvkm_clk *base, struct nvkm_cstate *cstate) +{ + struct gp10b_clk *clk = gp10b_clk(base); + u32 target_rate = cstate->domain[nv_clk_src_gpc] * GK20A_CLK_GPC_MDIV; + + clk->new_rate = clk_round_rate(clk->clk, target_rate) / GK20A_CLK_GPC_MDIV; + + return 0; +} + +static int +gp10b_clk_prog(struct nvkm_clk *base) +{ + struct gp10b_clk *clk = gp10b_clk(base); + int ret; + + ret = clk_set_rate(clk->clk, clk->new_rate * GK20A_CLK_GPC_MDIV); + if (ret < 0) + return ret; + + clk->rate = clk_get_rate(clk->clk) / GK20A_CLK_GPC_MDIV; + + return 0; +} + +static struct nvkm_pstate +gp10b_pstates[] = { + { + .base = { + .domain[nv_clk_src_gpc] = 114750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 216750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 318750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 420750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 522750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 624750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 726750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 828750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 930750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 1032750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 1134750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 1236750, + }, + }, + { + .base = { + .domain[nv_clk_src_gpc] = 1300500, + }, + }, +}; + +static const struct nvkm_clk_func +gp10b_clk = { + .init = gp10b_clk_init, + .read = gp10b_clk_read, + .calc = gp10b_clk_calc, + .prog = gp10b_clk_prog, + .tidy = gk20a_clk_tidy, + .pstates = gp10b_pstates, + .nr_pstates = ARRAY_SIZE(gp10b_pstates), + .domains = { + { nv_clk_src_gpc, 0xff, 0, "core", GK20A_CLK_GPC_MDIV }, + { nv_clk_src_max } + } +}; + +int +gp10b_clk_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_clk **pclk) +{ + struct nvkm_device_tegra *tdev = device->func->tegra(device); + const struct nvkm_clk_func *func = &gp10b_clk; + struct gp10b_clk *clk; + int ret, i; + + clk = kzalloc(sizeof(*clk), GFP_KERNEL); + if (!clk) + return -ENOMEM; + *pclk = &clk->base; + clk->clk = tdev->clk; + + /* Finish initializing the pstates */ + for (i = 0; i < func->nr_pstates; i++) { + INIT_LIST_HEAD(&func->pstates[i].list); + func->pstates[i].pstate = i + 1; + } + + ret = nvkm_clk_ctor(func, device, type, inst, true, &clk->base); + if (ret) + return ret; + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gp10b.h b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gp10b.h new file mode 100644 index 000000000000..178e3bcdbbf7 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/gp10b.h @@ -0,0 +1,18 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVKM_CLK_GP10B_H__ +#define __NVKM_CLK_GP10B_H__ + +struct gp10b_clk { + /* currently applied parameters */ + struct nvkm_clk base; + struct gk20a_devfreq *devfreq; + struct clk *clk; + u32 rate; + + /* new parameters to apply */ + u32 new_rate; +}; + +#define gp10b_clk(p) container_of((p), struct gp10b_clk, base) + +#endif diff --git a/drivers/gpu/drm/omapdrm/omap_encoder.c b/drivers/gpu/drm/omapdrm/omap_encoder.c index 4dd05bc732da..195715b162e3 100644 --- a/drivers/gpu/drm/omapdrm/omap_encoder.c +++ b/drivers/gpu/drm/omapdrm/omap_encoder.c @@ -77,7 +77,6 @@ static void omap_encoder_mode_set(struct drm_encoder *encoder, struct omap_dss_device *output = omap_encoder->output; struct drm_device *dev = encoder->dev; struct drm_connector *connector; - struct drm_bridge *bridge; struct videomode vm = { 0 }; u32 bus_flags; @@ -97,8 +96,7 @@ static void omap_encoder_mode_set(struct drm_encoder *encoder, * * A better solution is to use DRM's bus-flags through the whole driver. */ - for (bridge = output->bridge; bridge; - bridge = drm_bridge_get_next_bridge(bridge)) { + drm_for_each_bridge_in_chain_from(output->bridge, bridge) { if (!bridge->timings) continue; diff --git a/drivers/gpu/drm/panel/panel-edp.c b/drivers/gpu/drm/panel/panel-edp.c index 62435e3cd9f4..d54db7a6a312 100644 --- a/drivers/gpu/drm/panel/panel-edp.c +++ b/drivers/gpu/drm/panel/panel-edp.c @@ -1909,6 +1909,7 @@ static const struct edp_panel_entry edp_panels[] = { EDP_PANEL_ENTRY('A', 'U', 'O', 0x8bba, &delay_200_500_e50, "B140UAN08.5"), EDP_PANEL_ENTRY('A', 'U', 'O', 0xa199, &delay_200_500_e50, "B116XAN06.1"), EDP_PANEL_ENTRY('A', 'U', 'O', 0xa7b3, &delay_200_500_e50, "B140UAN04.4"), + EDP_PANEL_ENTRY('A', 'U', 'O', 0xb7a9, &delay_200_500_e50, "B140HAK03.3"), EDP_PANEL_ENTRY('A', 'U', 'O', 0xc4b4, &delay_200_500_e50, "B116XAT04.1"), EDP_PANEL_ENTRY('A', 'U', 'O', 0xc9a8, &delay_200_500_e50, "B140QAN08.H"), EDP_PANEL_ENTRY('A', 'U', 'O', 0xcdba, &delay_200_500_e50, "B140UAX01.2"), @@ -1974,6 +1975,7 @@ static const struct edp_panel_entry edp_panels[] = { EDP_PANEL_ENTRY('B', 'O', 'E', 0x0c20, &delay_200_500_e80, "NT140FHM-N47"), EDP_PANEL_ENTRY('B', 'O', 'E', 0x0c93, &delay_200_500_e200, "Unknown"), EDP_PANEL_ENTRY('B', 'O', 'E', 0x0cb6, &delay_200_500_e200, "NT116WHM-N44"), + EDP_PANEL_ENTRY('B', 'O', 'E', 0x0cf2, &delay_200_500_e200, "NV156FHM-N4S"), EDP_PANEL_ENTRY('B', 'O', 'E', 0x0cf6, &delay_200_500_e200, "NV140WUM-N64"), EDP_PANEL_ENTRY('B', 'O', 'E', 0x0cfa, &delay_200_500_e50, "NV116WHM-A4D"), EDP_PANEL_ENTRY('B', 'O', 'E', 0x0d45, &delay_200_500_e80, "NV116WHM-N4B"), @@ -2007,10 +2009,12 @@ static const struct edp_panel_entry edp_panels[] = { EDP_PANEL_ENTRY('C', 'M', 'N', 0x1441, &delay_200_500_e80_d50, "N140JCA-ELK"), EDP_PANEL_ENTRY('C', 'M', 'N', 0x144f, &delay_200_500_e80_d50, "N140HGA-EA1"), EDP_PANEL_ENTRY('C', 'M', 'N', 0x1468, &delay_200_500_e80, "N140HGA-EA1"), + EDP_PANEL_ENTRY('C', 'M', 'N', 0x148f, &delay_200_500_e80, "N140HCA-EAC"), EDP_PANEL_ENTRY('C', 'M', 'N', 0x14a8, &delay_200_500_e80, "N140JCA-ELP"), EDP_PANEL_ENTRY('C', 'M', 'N', 0x14d4, &delay_200_500_e80_d50, "N140HCA-EAC"), EDP_PANEL_ENTRY('C', 'M', 'N', 0x14d6, &delay_200_500_e80_d50, "N140BGA-EA4"), EDP_PANEL_ENTRY('C', 'M', 'N', 0x14e5, &delay_200_500_e80_d50, "N140HGA-EA1"), + EDP_PANEL_ENTRY('C', 'M', 'N', 0x1565, &delay_200_500_e80, "N156HCA-EAB"), EDP_PANEL_ENTRY('C', 'M', 'N', 0x162b, &delay_200_500_e80_d50, "N160JCE-ELL"), EDP_PANEL_ENTRY('C', 'M', 'N', 0x7402, &delay_200_500_e200_d50, "N116BCA-EAK"), @@ -2022,10 +2026,12 @@ static const struct edp_panel_entry edp_panels[] = { EDP_PANEL_ENTRY('C', 'S', 'W', 0x1104, &delay_200_500_e50_d100, "MNB601LS1-4"), EDP_PANEL_ENTRY('C', 'S', 'W', 0x143f, &delay_200_500_e50, "MNE007QS3-6"), EDP_PANEL_ENTRY('C', 'S', 'W', 0x1448, &delay_200_500_e50, "MNE007QS3-7"), + EDP_PANEL_ENTRY('C', 'S', 'W', 0x144b, &delay_200_500_e80, "MNE001BS1-4"), EDP_PANEL_ENTRY('C', 'S', 'W', 0x1457, &delay_80_500_e80_p2e200, "MNE007QS3-8"), EDP_PANEL_ENTRY('C', 'S', 'W', 0x1462, &delay_200_500_e50, "MNE007QS5-2"), EDP_PANEL_ENTRY('C', 'S', 'W', 0x1468, &delay_200_500_e50, "MNE007QB2-2"), EDP_PANEL_ENTRY('C', 'S', 'W', 0x146e, &delay_80_500_e50_d50, "MNE007QB3-1"), + EDP_PANEL_ENTRY('C', 'S', 'W', 0x1519, &delay_200_500_e80_d50, "MNF601BS1-3"), EDP_PANEL_ENTRY('E', 'T', 'C', 0x0000, &delay_50_500_e200_d200_po2e335, "LP079QX1-SP0V"), @@ -2046,6 +2052,7 @@ static const struct edp_panel_entry edp_panels[] = { EDP_PANEL_ENTRY('K', 'D', 'B', 0x1212, &delay_200_500_e50, "KD116N0930A16"), EDP_PANEL_ENTRY('K', 'D', 'B', 0x1707, &delay_200_150_e50, "KD116N2130B12"), + EDP_PANEL_ENTRY('K', 'D', 'C', 0x0110, &delay_200_500_e50, "KD116N3730A07"), EDP_PANEL_ENTRY('K', 'D', 'C', 0x044f, &delay_200_500_e50, "KD116N9-30NH-F3"), EDP_PANEL_ENTRY('K', 'D', 'C', 0x05f1, &delay_200_500_e80_d50, "KD116N5-30NV-G7"), EDP_PANEL_ENTRY('K', 'D', 'C', 0x0809, &delay_200_500_e50, "KD116N2930A15"), diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h index 077525a3ad68..1e73efad02a8 100644 --- a/drivers/gpu/drm/panfrost/panfrost_device.h +++ b/drivers/gpu/drm/panfrost/panfrost_device.h @@ -10,11 +10,13 @@ #include <linux/pm.h> #include <linux/regulator/consumer.h> #include <linux/spinlock.h> +#include <drm/drm_auth.h> #include <drm/drm_device.h> #include <drm/drm_mm.h> #include <drm/gpu_scheduler.h> #include "panfrost_devfreq.h" +#include "panfrost_job.h" struct panfrost_device; struct panfrost_mmu; @@ -22,7 +24,6 @@ struct panfrost_job_slot; struct panfrost_job; struct panfrost_perfcnt; -#define NUM_JOB_SLOTS 3 #define MAX_PM_DOMAINS 5 enum panfrost_drv_comp_bits { @@ -206,13 +207,19 @@ struct panfrost_engine_usage { struct panfrost_file_priv { struct panfrost_device *pfdev; - struct drm_sched_entity sched_entity[NUM_JOB_SLOTS]; + struct xarray jm_ctxs; struct panfrost_mmu *mmu; struct panfrost_engine_usage engine_usage; }; +static inline bool panfrost_high_prio_allowed(struct drm_file *file) +{ + /* Higher priorities require CAP_SYS_NICE or DRM_MASTER */ + return (capable(CAP_SYS_NICE) || drm_is_current_master(file)); +} + static inline struct panfrost_device *to_panfrost_device(struct drm_device *ddev) { return ddev->dev_private; diff --git a/drivers/gpu/drm/panfrost/panfrost_drv.c b/drivers/gpu/drm/panfrost/panfrost_drv.c index 1ea6c509a5d5..3af4b4753ca4 100644 --- a/drivers/gpu/drm/panfrost/panfrost_drv.c +++ b/drivers/gpu/drm/panfrost/panfrost_drv.c @@ -109,6 +109,14 @@ static int panfrost_ioctl_get_param(struct drm_device *ddev, void *data, struct #endif break; + case DRM_PANFROST_PARAM_ALLOWED_JM_CTX_PRIORITIES: + param->value = BIT(PANFROST_JM_CTX_PRIORITY_LOW) | + BIT(PANFROST_JM_CTX_PRIORITY_MEDIUM); + + if (panfrost_high_prio_allowed(file)) + param->value |= BIT(PANFROST_JM_CTX_PRIORITY_HIGH); + break; + default: return -EINVAL; } @@ -279,9 +287,13 @@ static int panfrost_ioctl_submit(struct drm_device *dev, void *data, struct panfrost_file_priv *file_priv = file->driver_priv; struct drm_panfrost_submit *args = data; struct drm_syncobj *sync_out = NULL; + struct panfrost_jm_ctx *jm_ctx; struct panfrost_job *job; int ret = 0, slot; + if (args->pad) + return -EINVAL; + if (!args->jc) return -EINVAL; @@ -294,10 +306,16 @@ static int panfrost_ioctl_submit(struct drm_device *dev, void *data, return -ENODEV; } + jm_ctx = panfrost_jm_ctx_from_handle(file, args->jm_ctx_handle); + if (!jm_ctx) { + ret = -EINVAL; + goto out_put_syncout; + } + job = kzalloc(sizeof(*job), GFP_KERNEL); if (!job) { ret = -ENOMEM; - goto out_put_syncout; + goto out_put_jm_ctx; } kref_init(&job->refcount); @@ -307,12 +325,13 @@ static int panfrost_ioctl_submit(struct drm_device *dev, void *data, job->requirements = args->requirements; job->flush_id = panfrost_gpu_get_latest_flush_id(pfdev); job->mmu = file_priv->mmu; + job->ctx = panfrost_jm_ctx_get(jm_ctx); job->engine_usage = &file_priv->engine_usage; slot = panfrost_job_get_slot(job); ret = drm_sched_job_init(&job->base, - &file_priv->sched_entity[slot], + &jm_ctx->slot_entity[slot], 1, NULL, file->client_id); if (ret) goto out_put_job; @@ -338,6 +357,8 @@ out_cleanup_job: drm_sched_job_cleanup(&job->base); out_put_job: panfrost_job_put(job); +out_put_jm_ctx: + panfrost_jm_ctx_put(jm_ctx); out_put_syncout: if (sync_out) drm_syncobj_put(sync_out); @@ -536,6 +557,27 @@ err_put_obj: return ret; } +static int panfrost_ioctl_jm_ctx_create(struct drm_device *dev, void *data, + struct drm_file *file) +{ + return panfrost_jm_ctx_create(file, data); +} + +static int panfrost_ioctl_jm_ctx_destroy(struct drm_device *dev, void *data, + struct drm_file *file) +{ + const struct drm_panfrost_jm_ctx_destroy *args = data; + + if (args->pad) + return -EINVAL; + + /* We can't destroy the default context created when the file is opened. */ + if (!args->handle) + return -EINVAL; + + return panfrost_jm_ctx_destroy(file, args->handle); +} + int panfrost_unstable_ioctl_check(void) { if (!unstable_ioctls) @@ -564,7 +606,7 @@ panfrost_open(struct drm_device *dev, struct drm_file *file) goto err_free; } - ret = panfrost_job_open(panfrost_priv); + ret = panfrost_job_open(file); if (ret) goto err_job; @@ -583,7 +625,7 @@ panfrost_postclose(struct drm_device *dev, struct drm_file *file) struct panfrost_file_priv *panfrost_priv = file->driver_priv; panfrost_perfcnt_close(file); - panfrost_job_close(panfrost_priv); + panfrost_job_close(file); panfrost_mmu_ctx_put(panfrost_priv->mmu); kfree(panfrost_priv); @@ -603,6 +645,8 @@ static const struct drm_ioctl_desc panfrost_drm_driver_ioctls[] = { PANFROST_IOCTL(PERFCNT_DUMP, perfcnt_dump, DRM_RENDER_ALLOW), PANFROST_IOCTL(MADVISE, madvise, DRM_RENDER_ALLOW), PANFROST_IOCTL(SET_LABEL_BO, set_label_bo, DRM_RENDER_ALLOW), + PANFROST_IOCTL(JM_CTX_CREATE, jm_ctx_create, DRM_RENDER_ALLOW), + PANFROST_IOCTL(JM_CTX_DESTROY, jm_ctx_destroy, DRM_RENDER_ALLOW), }; static void panfrost_gpu_show_fdinfo(struct panfrost_device *pfdev, @@ -672,6 +716,47 @@ static int panthor_gems_show(struct seq_file *m, void *data) return 0; } +static void show_panfrost_jm_ctx(struct panfrost_jm_ctx *jm_ctx, u32 handle, + struct seq_file *m) +{ + struct drm_device *ddev = ((struct drm_info_node *)m->private)->minor->dev; + const char *prio = "UNKNOWN"; + + static const char * const prios[] = { + [DRM_SCHED_PRIORITY_HIGH] = "HIGH", + [DRM_SCHED_PRIORITY_NORMAL] = "NORMAL", + [DRM_SCHED_PRIORITY_LOW] = "LOW", + }; + + if (jm_ctx->slot_entity[0].priority != + jm_ctx->slot_entity[1].priority) + drm_warn(ddev, "Slot priorities should be the same in a single context"); + + if (jm_ctx->slot_entity[0].priority < ARRAY_SIZE(prios)) + prio = prios[jm_ctx->slot_entity[0].priority]; + + seq_printf(m, " JM context %u: priority %s\n", handle, prio); +} + +static int show_file_jm_ctxs(struct panfrost_file_priv *pfile, + struct seq_file *m) +{ + struct panfrost_jm_ctx *jm_ctx; + unsigned long i; + + xa_lock(&pfile->jm_ctxs); + xa_for_each(&pfile->jm_ctxs, i, jm_ctx) { + jm_ctx = panfrost_jm_ctx_get(jm_ctx); + xa_unlock(&pfile->jm_ctxs); + show_panfrost_jm_ctx(jm_ctx, i, m); + panfrost_jm_ctx_put(jm_ctx); + xa_lock(&pfile->jm_ctxs); + } + xa_unlock(&pfile->jm_ctxs); + + return 0; +} + static struct drm_info_list panthor_debugfs_list[] = { {"gems", panthor_gems_show, 0, NULL}, }; @@ -685,9 +770,64 @@ static int panthor_gems_debugfs_init(struct drm_minor *minor) return 0; } +static int show_each_file(struct seq_file *m, void *arg) +{ + struct drm_info_node *node = (struct drm_info_node *)m->private; + struct drm_device *ddev = node->minor->dev; + int (*show)(struct panfrost_file_priv *, struct seq_file *) = + node->info_ent->data; + struct drm_file *file; + int ret; + + ret = mutex_lock_interruptible(&ddev->filelist_mutex); + if (ret) + return ret; + + list_for_each_entry(file, &ddev->filelist, lhead) { + struct task_struct *task; + struct panfrost_file_priv *pfile = file->driver_priv; + struct pid *pid; + + /* + * Although we have a valid reference on file->pid, that does + * not guarantee that the task_struct who called get_pid() is + * still alive (e.g. get_pid(current) => fork() => exit()). + * Therefore, we need to protect this ->comm access using RCU. + */ + rcu_read_lock(); + pid = rcu_dereference(file->pid); + task = pid_task(pid, PIDTYPE_TGID); + seq_printf(m, "client_id %8llu pid %8d command %s:\n", + file->client_id, pid_nr(pid), + task ? task->comm : "<unknown>"); + rcu_read_unlock(); + + ret = show(pfile, m); + if (ret < 0) + break; + + seq_puts(m, "\n"); + } + + mutex_unlock(&ddev->filelist_mutex); + return ret; +} + +static struct drm_info_list panfrost_sched_debugfs_list[] = { + { "sched_ctxs", show_each_file, 0, show_file_jm_ctxs }, +}; + +static void panfrost_sched_debugfs_init(struct drm_minor *minor) +{ + drm_debugfs_create_files(panfrost_sched_debugfs_list, + ARRAY_SIZE(panfrost_sched_debugfs_list), + minor->debugfs_root, minor); +} + static void panfrost_debugfs_init(struct drm_minor *minor) { panthor_gems_debugfs_init(minor); + panfrost_sched_debugfs_init(minor); } #endif @@ -699,6 +839,8 @@ static void panfrost_debugfs_init(struct drm_minor *minor) * - 1.3 - adds JD_REQ_CYCLE_COUNT job requirement for SUBMIT * - adds SYSTEM_TIMESTAMP and SYSTEM_TIMESTAMP_FREQUENCY queries * - 1.4 - adds SET_LABEL_BO + * - 1.5 - adds JM_CTX_{CREATE,DESTROY} ioctls and extend SUBMIT to allow + * context creation with configurable priorities/affinity */ static const struct drm_driver panfrost_drm_driver = { .driver_features = DRIVER_RENDER | DRIVER_GEM | DRIVER_SYNCOBJ, diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c index 82acabb21b27..c47d14eabbae 100644 --- a/drivers/gpu/drm/panfrost/panfrost_job.c +++ b/drivers/gpu/drm/panfrost/panfrost_job.c @@ -22,6 +22,7 @@ #include "panfrost_mmu.h" #include "panfrost_dump.h" +#define MAX_JM_CTX_PER_FILE 64 #define JOB_TIMEOUT_MS 500 #define job_write(dev, reg, data) writel(data, dev->iomem + (reg)) @@ -359,6 +360,7 @@ static void panfrost_job_cleanup(struct kref *ref) kvfree(job->bos); } + panfrost_jm_ctx_put(job->ctx); kfree(job); } @@ -383,6 +385,9 @@ static struct dma_fence *panfrost_job_run(struct drm_sched_job *sched_job) int slot = panfrost_job_get_slot(job); struct dma_fence *fence = NULL; + if (job->ctx->destroyed) + return ERR_PTR(-ECANCELED); + if (unlikely(job->base.s_fence->finished.error)) return NULL; @@ -917,39 +922,176 @@ void panfrost_job_fini(struct panfrost_device *pfdev) destroy_workqueue(pfdev->reset.wq); } -int panfrost_job_open(struct panfrost_file_priv *panfrost_priv) +int panfrost_job_open(struct drm_file *file) +{ + struct panfrost_file_priv *panfrost_priv = file->driver_priv; + int ret; + + struct drm_panfrost_jm_ctx_create default_jm_ctx = { + .priority = PANFROST_JM_CTX_PRIORITY_MEDIUM, + }; + + xa_init_flags(&panfrost_priv->jm_ctxs, XA_FLAGS_ALLOC); + + ret = panfrost_jm_ctx_create(file, &default_jm_ctx); + if (ret) + return ret; + + /* We expect the default context to be assigned handle 0. */ + if (WARN_ON(default_jm_ctx.handle)) + return -EINVAL; + + return 0; +} + +void panfrost_job_close(struct drm_file *file) +{ + struct panfrost_file_priv *panfrost_priv = file->driver_priv; + struct panfrost_jm_ctx *jm_ctx; + unsigned long i; + + xa_for_each(&panfrost_priv->jm_ctxs, i, jm_ctx) + panfrost_jm_ctx_destroy(file, i); + + xa_destroy(&panfrost_priv->jm_ctxs); +} + +int panfrost_job_is_idle(struct panfrost_device *pfdev) { - struct panfrost_device *pfdev = panfrost_priv->pfdev; struct panfrost_job_slot *js = pfdev->js; - struct drm_gpu_scheduler *sched; - int ret, i; + int i; for (i = 0; i < NUM_JOB_SLOTS; i++) { - sched = &js->queue[i].sched; - ret = drm_sched_entity_init(&panfrost_priv->sched_entity[i], - DRM_SCHED_PRIORITY_NORMAL, &sched, - 1, NULL); - if (WARN_ON(ret)) - return ret; + /* If there are any jobs in the HW queue, we're not idle */ + if (atomic_read(&js->queue[i].sched.credit_count)) + return false; + } + + return true; +} + +static void panfrost_jm_ctx_release(struct kref *kref) +{ + struct panfrost_jm_ctx *jm_ctx = container_of(kref, struct panfrost_jm_ctx, refcnt); + + WARN_ON(!jm_ctx->destroyed); + + for (u32 i = 0; i < ARRAY_SIZE(jm_ctx->slot_entity); i++) + drm_sched_entity_destroy(&jm_ctx->slot_entity[i]); + + kfree(jm_ctx); +} + +void +panfrost_jm_ctx_put(struct panfrost_jm_ctx *jm_ctx) +{ + if (jm_ctx) + kref_put(&jm_ctx->refcnt, panfrost_jm_ctx_release); +} + +struct panfrost_jm_ctx * +panfrost_jm_ctx_get(struct panfrost_jm_ctx *jm_ctx) +{ + if (jm_ctx) + kref_get(&jm_ctx->refcnt); + + return jm_ctx; +} + +struct panfrost_jm_ctx * +panfrost_jm_ctx_from_handle(struct drm_file *file, u32 handle) +{ + struct panfrost_file_priv *priv = file->driver_priv; + struct panfrost_jm_ctx *jm_ctx; + + xa_lock(&priv->jm_ctxs); + jm_ctx = panfrost_jm_ctx_get(xa_load(&priv->jm_ctxs, handle)); + xa_unlock(&priv->jm_ctxs); + + return jm_ctx; +} + +static int jm_ctx_prio_to_drm_sched_prio(struct drm_file *file, + enum drm_panfrost_jm_ctx_priority in, + enum drm_sched_priority *out) +{ + switch (in) { + case PANFROST_JM_CTX_PRIORITY_LOW: + *out = DRM_SCHED_PRIORITY_LOW; + return 0; + case PANFROST_JM_CTX_PRIORITY_MEDIUM: + *out = DRM_SCHED_PRIORITY_NORMAL; + return 0; + case PANFROST_JM_CTX_PRIORITY_HIGH: + if (!panfrost_high_prio_allowed(file)) + return -EACCES; + + *out = DRM_SCHED_PRIORITY_HIGH; + return 0; + default: + return -EINVAL; } +} + +int panfrost_jm_ctx_create(struct drm_file *file, + struct drm_panfrost_jm_ctx_create *args) +{ + struct panfrost_file_priv *priv = file->driver_priv; + struct panfrost_device *pfdev = priv->pfdev; + enum drm_sched_priority sched_prio; + struct panfrost_jm_ctx *jm_ctx; + int ret; + + jm_ctx = kzalloc(sizeof(*jm_ctx), GFP_KERNEL); + if (!jm_ctx) + return -ENOMEM; + + kref_init(&jm_ctx->refcnt); + + ret = jm_ctx_prio_to_drm_sched_prio(file, args->priority, &sched_prio); + if (ret) + goto err_put_jm_ctx; + + for (u32 i = 0; i < NUM_JOB_SLOTS; i++) { + struct drm_gpu_scheduler *sched = &pfdev->js->queue[i].sched; + + ret = drm_sched_entity_init(&jm_ctx->slot_entity[i], sched_prio, + &sched, 1, NULL); + if (ret) + goto err_put_jm_ctx; + } + + ret = xa_alloc(&priv->jm_ctxs, &args->handle, jm_ctx, + XA_LIMIT(0, MAX_JM_CTX_PER_FILE), GFP_KERNEL); + if (ret) + goto err_put_jm_ctx; + return 0; + +err_put_jm_ctx: + jm_ctx->destroyed = true; + panfrost_jm_ctx_put(jm_ctx); + return ret; } -void panfrost_job_close(struct panfrost_file_priv *panfrost_priv) +int panfrost_jm_ctx_destroy(struct drm_file *file, u32 handle) { - struct panfrost_device *pfdev = panfrost_priv->pfdev; - int i; + struct panfrost_file_priv *priv = file->driver_priv; + struct panfrost_device *pfdev = priv->pfdev; + struct panfrost_jm_ctx *jm_ctx; - for (i = 0; i < NUM_JOB_SLOTS; i++) - drm_sched_entity_destroy(&panfrost_priv->sched_entity[i]); + jm_ctx = xa_erase(&priv->jm_ctxs, handle); + if (!jm_ctx) + return -EINVAL; + + jm_ctx->destroyed = true; /* Kill in-flight jobs */ spin_lock(&pfdev->js->job_lock); - for (i = 0; i < NUM_JOB_SLOTS; i++) { - struct drm_sched_entity *entity = &panfrost_priv->sched_entity[i]; - int j; + for (u32 i = 0; i < ARRAY_SIZE(jm_ctx->slot_entity); i++) { + struct drm_sched_entity *entity = &jm_ctx->slot_entity[i]; - for (j = ARRAY_SIZE(pfdev->jobs[0]) - 1; j >= 0; j--) { + for (int j = ARRAY_SIZE(pfdev->jobs[0]) - 1; j >= 0; j--) { struct panfrost_job *job = pfdev->jobs[i][j]; u32 cmd; @@ -980,18 +1122,7 @@ void panfrost_job_close(struct panfrost_file_priv *panfrost_priv) } } spin_unlock(&pfdev->js->job_lock); -} - -int panfrost_job_is_idle(struct panfrost_device *pfdev) -{ - struct panfrost_job_slot *js = pfdev->js; - int i; - for (i = 0; i < NUM_JOB_SLOTS; i++) { - /* If there are any jobs in the HW queue, we're not idle */ - if (atomic_read(&js->queue[i].sched.credit_count)) - return false; - } - - return true; + panfrost_jm_ctx_put(jm_ctx); + return 0; } diff --git a/drivers/gpu/drm/panfrost/panfrost_job.h b/drivers/gpu/drm/panfrost/panfrost_job.h index ec581b97852b..5a30ff1503c6 100644 --- a/drivers/gpu/drm/panfrost/panfrost_job.h +++ b/drivers/gpu/drm/panfrost/panfrost_job.h @@ -18,6 +18,7 @@ struct panfrost_job { struct panfrost_device *pfdev; struct panfrost_mmu *mmu; + struct panfrost_jm_ctx *ctx; /* Fence to be signaled by IRQ handler when the job is complete. */ struct dma_fence *done_fence; @@ -39,10 +40,30 @@ struct panfrost_job { u64 start_cycles; }; +struct panfrost_js_ctx { + struct drm_sched_entity sched_entity; + bool enabled; +}; + +#define NUM_JOB_SLOTS 3 + +struct panfrost_jm_ctx { + struct kref refcnt; + bool destroyed; + struct drm_sched_entity slot_entity[NUM_JOB_SLOTS]; +}; + +int panfrost_jm_ctx_create(struct drm_file *file, + struct drm_panfrost_jm_ctx_create *args); +int panfrost_jm_ctx_destroy(struct drm_file *file, u32 handle); +void panfrost_jm_ctx_put(struct panfrost_jm_ctx *jm_ctx); +struct panfrost_jm_ctx *panfrost_jm_ctx_get(struct panfrost_jm_ctx *jm_ctx); +struct panfrost_jm_ctx *panfrost_jm_ctx_from_handle(struct drm_file *file, u32 handle); + int panfrost_job_init(struct panfrost_device *pfdev); void panfrost_job_fini(struct panfrost_device *pfdev); -int panfrost_job_open(struct panfrost_file_priv *panfrost_priv); -void panfrost_job_close(struct panfrost_file_priv *panfrost_priv); +int panfrost_job_open(struct drm_file *file); +void panfrost_job_close(struct drm_file *file); int panfrost_job_get_slot(struct panfrost_job *job); int panfrost_job_push(struct panfrost_job *job); void panfrost_job_put(struct panfrost_job *job); diff --git a/drivers/gpu/drm/panthor/panthor_drv.c b/drivers/gpu/drm/panthor/panthor_drv.c index 4c202fc5ce05..fdbe89ef7f43 100644 --- a/drivers/gpu/drm/panthor/panthor_drv.c +++ b/drivers/gpu/drm/panthor/panthor_drv.c @@ -1105,7 +1105,7 @@ static int panthor_ioctl_group_create(struct drm_device *ddev, void *data, if (ret) goto out; - ret = panthor_group_create(pfile, args, queue_args); + ret = panthor_group_create(pfile, args, queue_args, file->client_id); if (ret < 0) goto out; args->group_handle = ret; diff --git a/drivers/gpu/drm/panthor/panthor_sched.c b/drivers/gpu/drm/panthor/panthor_sched.c index ba5dc3e443d9..0cc9055f4ee5 100644 --- a/drivers/gpu/drm/panthor/panthor_sched.c +++ b/drivers/gpu/drm/panthor/panthor_sched.c @@ -360,6 +360,9 @@ struct panthor_queue { /** @entity: DRM scheduling entity used for this queue. */ struct drm_sched_entity entity; + /** @name: DRM scheduler name for this queue. */ + char *name; + /** * @remaining_time: Time remaining before the job timeout expires. * @@ -901,6 +904,8 @@ static void group_free_queue(struct panthor_group *group, struct panthor_queue * if (queue->scheduler.ops) drm_sched_fini(&queue->scheduler); + kfree(queue->name); + panthor_queue_put_syncwait_obj(queue); panthor_kernel_bo_destroy(queue->ringbuf); @@ -1412,7 +1417,7 @@ cs_slot_process_fault_event_locked(struct panthor_device *ptdev, fault = cs_iface->output->fault; info = cs_iface->output->fault_info; - if (queue && CS_EXCEPTION_TYPE(fault) == DRM_PANTHOR_EXCEPTION_CS_INHERIT_FAULT) { + if (queue) { u64 cs_extract = queue->iface.output->extract; struct panthor_job *job; @@ -3308,9 +3313,10 @@ static u32 calc_profiling_ringbuf_num_slots(struct panthor_device *ptdev, static struct panthor_queue * group_create_queue(struct panthor_group *group, - const struct drm_panthor_queue_create *args) + const struct drm_panthor_queue_create *args, + u64 drm_client_id, u32 gid, u32 qid) { - const struct drm_sched_init_args sched_args = { + struct drm_sched_init_args sched_args = { .ops = &panthor_queue_sched_ops, .submit_wq = group->ptdev->scheduler->wq, .num_rqs = 1, @@ -3323,7 +3329,6 @@ group_create_queue(struct panthor_group *group, .credit_limit = args->ringbuf_size / sizeof(u64), .timeout = msecs_to_jiffies(JOB_TIMEOUT_MS), .timeout_wq = group->ptdev->reset.wq, - .name = "panthor-queue", .dev = group->ptdev->base.dev, }; struct drm_gpu_scheduler *drm_sched; @@ -3398,6 +3403,15 @@ group_create_queue(struct panthor_group *group, if (ret) goto err_free_queue; + /* assign a unique name */ + queue->name = kasprintf(GFP_KERNEL, "panthor-queue-%llu-%u-%u", drm_client_id, gid, qid); + if (!queue->name) { + ret = -ENOMEM; + goto err_free_queue; + } + + sched_args.name = queue->name; + ret = drm_sched_init(&queue->scheduler, &sched_args); if (ret) goto err_free_queue; @@ -3447,7 +3461,8 @@ static void add_group_kbo_sizes(struct panthor_device *ptdev, int panthor_group_create(struct panthor_file *pfile, const struct drm_panthor_group_create *group_args, - const struct drm_panthor_queue_create *queue_args) + const struct drm_panthor_queue_create *queue_args, + u64 drm_client_id) { struct panthor_device *ptdev = pfile->ptdev; struct panthor_group_pool *gpool = pfile->groups; @@ -3540,12 +3555,16 @@ int panthor_group_create(struct panthor_file *pfile, memset(group->syncobjs->kmap, 0, group_args->queues.count * sizeof(struct panthor_syncobj_64b)); + ret = xa_alloc(&gpool->xa, &gid, group, XA_LIMIT(1, MAX_GROUPS_PER_POOL), GFP_KERNEL); + if (ret) + goto err_put_group; + for (i = 0; i < group_args->queues.count; i++) { - group->queues[i] = group_create_queue(group, &queue_args[i]); + group->queues[i] = group_create_queue(group, &queue_args[i], drm_client_id, gid, i); if (IS_ERR(group->queues[i])) { ret = PTR_ERR(group->queues[i]); group->queues[i] = NULL; - goto err_put_group; + goto err_erase_gid; } group->queue_count++; @@ -3553,10 +3572,6 @@ int panthor_group_create(struct panthor_file *pfile, group->idle_queues = GENMASK(group->queue_count - 1, 0); - ret = xa_alloc(&gpool->xa, &gid, group, XA_LIMIT(1, MAX_GROUPS_PER_POOL), GFP_KERNEL); - if (ret) - goto err_put_group; - mutex_lock(&sched->reset.lock); if (atomic_read(&sched->reset.in_progress)) { panthor_group_stop(group); @@ -3575,6 +3590,9 @@ int panthor_group_create(struct panthor_file *pfile, return gid; +err_erase_gid: + xa_erase(&gpool->xa, gid); + err_put_group: group_put(group); return ret; diff --git a/drivers/gpu/drm/panthor/panthor_sched.h b/drivers/gpu/drm/panthor/panthor_sched.h index 742b0b4ff3a3..f4a475aa34c0 100644 --- a/drivers/gpu/drm/panthor/panthor_sched.h +++ b/drivers/gpu/drm/panthor/panthor_sched.h @@ -21,7 +21,8 @@ struct panthor_job; int panthor_group_create(struct panthor_file *pfile, const struct drm_panthor_group_create *group_args, - const struct drm_panthor_queue_create *queue_args); + const struct drm_panthor_queue_create *queue_args, + u64 drm_client_id); int panthor_group_destroy(struct panthor_file *pfile, u32 group_handle); int panthor_group_get_state(struct panthor_file *pfile, struct drm_panthor_group_get_state *get_state); diff --git a/drivers/gpu/drm/pl111/pl111_display.c b/drivers/gpu/drm/pl111/pl111_display.c index b9fe926a49e8..6d567e5c7c6f 100644 --- a/drivers/gpu/drm/pl111/pl111_display.c +++ b/drivers/gpu/drm/pl111/pl111_display.c @@ -473,12 +473,15 @@ static int pl111_clk_div_choose_div(struct clk_hw *hw, unsigned long rate, return best_div; } -static long pl111_clk_div_round_rate(struct clk_hw *hw, unsigned long rate, - unsigned long *prate) +static int pl111_clk_div_determine_rate(struct clk_hw *hw, + struct clk_rate_request *req) { - int div = pl111_clk_div_choose_div(hw, rate, prate, true); + int div = pl111_clk_div_choose_div(hw, req->rate, + &req->best_parent_rate, true); - return DIV_ROUND_UP_ULL(*prate, div); + req->rate = DIV_ROUND_UP_ULL(req->best_parent_rate, div); + + return 0; } static unsigned long pl111_clk_div_recalc_rate(struct clk_hw *hw, @@ -528,7 +531,7 @@ static int pl111_clk_div_set_rate(struct clk_hw *hw, unsigned long rate, static const struct clk_ops pl111_clk_div_ops = { .recalc_rate = pl111_clk_div_recalc_rate, - .round_rate = pl111_clk_div_round_rate, + .determine_rate = pl111_clk_div_determine_rate, .set_rate = pl111_clk_div_set_rate, }; diff --git a/drivers/gpu/drm/qxl/qxl_gem.c b/drivers/gpu/drm/qxl/qxl_gem.c index fc5e3763c359..d26043424e95 100644 --- a/drivers/gpu/drm/qxl/qxl_gem.c +++ b/drivers/gpu/drm/qxl/qxl_gem.c @@ -39,7 +39,7 @@ void qxl_gem_object_free(struct drm_gem_object *gobj) qxl_surface_evict(qdev, qobj, false); tbo = &qobj->tbo; - ttm_bo_put(tbo); + ttm_bo_fini(tbo); } int qxl_gem_object_create(struct qxl_device *qdev, int size, diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c index f86773f3db20..18ca1bcfd2f9 100644 --- a/drivers/gpu/drm/radeon/radeon_gem.c +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -86,7 +86,7 @@ static void radeon_gem_object_free(struct drm_gem_object *gobj) if (robj) { radeon_mn_unregister(robj); - ttm_bo_put(&robj->tbo); + ttm_bo_fini(&robj->tbo); } } diff --git a/drivers/gpu/drm/solomon/ssd130x.c b/drivers/gpu/drm/solomon/ssd130x.c index eec43d1a5595..8368f0ffbe1e 100644 --- a/drivers/gpu/drm/solomon/ssd130x.c +++ b/drivers/gpu/drm/solomon/ssd130x.c @@ -1498,7 +1498,7 @@ static int ssd130x_crtc_atomic_check(struct drm_crtc *crtc, if (ret) return ret; - ssd130x_state->data_array = kmalloc(ssd130x->width * pages, GFP_KERNEL); + ssd130x_state->data_array = kmalloc_array(ssd130x->width, pages, GFP_KERNEL); if (!ssd130x_state->data_array) return -ENOMEM; @@ -1519,7 +1519,7 @@ static int ssd132x_crtc_atomic_check(struct drm_crtc *crtc, if (ret) return ret; - ssd130x_state->data_array = kmalloc(columns * ssd130x->height, GFP_KERNEL); + ssd130x_state->data_array = kmalloc_array(columns, ssd130x->height, GFP_KERNEL); if (!ssd130x_state->data_array) return -ENOMEM; @@ -1546,7 +1546,7 @@ static int ssd133x_crtc_atomic_check(struct drm_crtc *crtc, pitch = drm_format_info_min_pitch(fi, 0, ssd130x->width); - ssd130x_state->data_array = kmalloc(pitch * ssd130x->height, GFP_KERNEL); + ssd130x_state->data_array = kmalloc_array(pitch, ssd130x->height, GFP_KERNEL); if (!ssd130x_state->data_array) return -ENOMEM; diff --git a/drivers/gpu/drm/stm/dw_mipi_dsi-stm.c b/drivers/gpu/drm/stm/dw_mipi_dsi-stm.c index 2c7bc064bc66..58eae6804cc8 100644 --- a/drivers/gpu/drm/stm/dw_mipi_dsi-stm.c +++ b/drivers/gpu/drm/stm/dw_mipi_dsi-stm.c @@ -274,8 +274,8 @@ static unsigned long dw_mipi_dsi_clk_recalc_rate(struct clk_hw *hw, return (unsigned long)pll_out_khz * 1000; } -static long dw_mipi_dsi_clk_round_rate(struct clk_hw *hw, unsigned long rate, - unsigned long *parent_rate) +static int dw_mipi_dsi_clk_determine_rate(struct clk_hw *hw, + struct clk_rate_request *req) { struct dw_mipi_dsi_stm *dsi = clk_to_dw_mipi_dsi_stm(hw); unsigned int idf, ndiv, odf, pll_in_khz, pll_out_khz; @@ -283,14 +283,14 @@ static long dw_mipi_dsi_clk_round_rate(struct clk_hw *hw, unsigned long rate, DRM_DEBUG_DRIVER("\n"); - pll_in_khz = (unsigned int)(*parent_rate / 1000); + pll_in_khz = (unsigned int)(req->best_parent_rate / 1000); /* Compute best pll parameters */ idf = 0; ndiv = 0; odf = 0; - ret = dsi_pll_get_params(dsi, pll_in_khz, rate / 1000, + ret = dsi_pll_get_params(dsi, pll_in_khz, req->rate / 1000, &idf, &ndiv, &odf); if (ret) DRM_WARN("Warning dsi_pll_get_params(): bad params\n"); @@ -298,7 +298,9 @@ static long dw_mipi_dsi_clk_round_rate(struct clk_hw *hw, unsigned long rate, /* Get the adjusted pll out value */ pll_out_khz = dsi_pll_get_clkout_khz(pll_in_khz, idf, ndiv, odf); - return pll_out_khz * 1000; + req->rate = pll_out_khz * 1000; + + return 0; } static int dw_mipi_dsi_clk_set_rate(struct clk_hw *hw, unsigned long rate, @@ -351,7 +353,7 @@ static const struct clk_ops dw_mipi_dsi_stm_clk_ops = { .disable = dw_mipi_dsi_clk_disable, .is_enabled = dw_mipi_dsi_clk_is_enabled, .recalc_rate = dw_mipi_dsi_clk_recalc_rate, - .round_rate = dw_mipi_dsi_clk_round_rate, + .determine_rate = dw_mipi_dsi_clk_determine_rate, .set_rate = dw_mipi_dsi_clk_set_rate, }; diff --git a/drivers/gpu/drm/stm/lvds.c b/drivers/gpu/drm/stm/lvds.c index 07788e8d3d83..fe38c0984b2b 100644 --- a/drivers/gpu/drm/stm/lvds.c +++ b/drivers/gpu/drm/stm/lvds.c @@ -682,8 +682,8 @@ static unsigned long lvds_pixel_clk_recalc_rate(struct clk_hw *hw, return (unsigned long)lvds->pixel_clock_rate; } -static long lvds_pixel_clk_round_rate(struct clk_hw *hw, unsigned long rate, - unsigned long *parent_rate) +static int lvds_pixel_clk_determine_rate(struct clk_hw *hw, + struct clk_rate_request *req) { struct stm_lvds *lvds = container_of(hw, struct stm_lvds, lvds_ck_px); unsigned int pll_in_khz, bdiv = 0, mdiv = 0, ndiv = 0; @@ -703,7 +703,7 @@ static long lvds_pixel_clk_round_rate(struct clk_hw *hw, unsigned long rate, mode = list_first_entry(&connector->modes, struct drm_display_mode, head); - pll_in_khz = (unsigned int)(*parent_rate / 1000); + pll_in_khz = (unsigned int)(req->best_parent_rate / 1000); if (lvds_is_dual_link(lvds->link_type)) multiplier = 2; @@ -719,14 +719,16 @@ static long lvds_pixel_clk_round_rate(struct clk_hw *hw, unsigned long rate, lvds->pixel_clock_rate = (unsigned long)pll_get_clkout_khz(pll_in_khz, bdiv, mdiv, ndiv) * 1000 * multiplier / 7; - return lvds->pixel_clock_rate; + req->rate = lvds->pixel_clock_rate; + + return 0; } static const struct clk_ops lvds_pixel_clk_ops = { .enable = lvds_pixel_clk_enable, .disable = lvds_pixel_clk_disable, .recalc_rate = lvds_pixel_clk_recalc_rate, - .round_rate = lvds_pixel_clk_round_rate, + .determine_rate = lvds_pixel_clk_determine_rate, }; static const struct clk_init_data clk_data = { diff --git a/drivers/gpu/drm/sun4i/sun4i_hdmi_ddc_clk.c b/drivers/gpu/drm/sun4i/sun4i_hdmi_ddc_clk.c index 12430b9d4e93..b1beadb9bb59 100644 --- a/drivers/gpu/drm/sun4i/sun4i_hdmi_ddc_clk.c +++ b/drivers/gpu/drm/sun4i/sun4i_hdmi_ddc_clk.c @@ -59,13 +59,15 @@ static unsigned long sun4i_ddc_calc_divider(unsigned long rate, return best_rate; } -static long sun4i_ddc_round_rate(struct clk_hw *hw, unsigned long rate, - unsigned long *prate) +static int sun4i_ddc_determine_rate(struct clk_hw *hw, + struct clk_rate_request *req) { struct sun4i_ddc *ddc = hw_to_ddc(hw); - return sun4i_ddc_calc_divider(rate, *prate, ddc->pre_div, - ddc->m_offset, NULL, NULL); + req->rate = sun4i_ddc_calc_divider(req->rate, req->best_parent_rate, + ddc->pre_div, ddc->m_offset, NULL, NULL); + + return 0; } static unsigned long sun4i_ddc_recalc_rate(struct clk_hw *hw, @@ -101,7 +103,7 @@ static int sun4i_ddc_set_rate(struct clk_hw *hw, unsigned long rate, static const struct clk_ops sun4i_ddc_ops = { .recalc_rate = sun4i_ddc_recalc_rate, - .round_rate = sun4i_ddc_round_rate, + .determine_rate = sun4i_ddc_determine_rate, .set_rate = sun4i_ddc_set_rate, }; diff --git a/drivers/gpu/drm/sun4i/sun4i_tcon_dclk.c b/drivers/gpu/drm/sun4i/sun4i_tcon_dclk.c index 03d7de1911cd..4afb12bd5281 100644 --- a/drivers/gpu/drm/sun4i/sun4i_tcon_dclk.c +++ b/drivers/gpu/drm/sun4i/sun4i_tcon_dclk.c @@ -67,8 +67,8 @@ static unsigned long sun4i_dclk_recalc_rate(struct clk_hw *hw, return parent_rate / val; } -static long sun4i_dclk_round_rate(struct clk_hw *hw, unsigned long rate, - unsigned long *parent_rate) +static int sun4i_dclk_determine_rate(struct clk_hw *hw, + struct clk_rate_request *req) { struct sun4i_dclk *dclk = hw_to_dclk(hw); struct sun4i_tcon *tcon = dclk->tcon; @@ -77,7 +77,7 @@ static long sun4i_dclk_round_rate(struct clk_hw *hw, unsigned long rate, int i; for (i = tcon->dclk_min_div; i <= tcon->dclk_max_div; i++) { - u64 ideal = (u64)rate * i; + u64 ideal = (u64)req->rate * i; unsigned long rounded; /* @@ -99,17 +99,19 @@ static long sun4i_dclk_round_rate(struct clk_hw *hw, unsigned long rate, goto out; } - if (abs(rate - rounded / i) < - abs(rate - best_parent / best_div)) { + if (abs(req->rate - rounded / i) < + abs(req->rate - best_parent / best_div)) { best_parent = rounded; best_div = i; } } out: - *parent_rate = best_parent; + req->best_parent_rate = best_parent; - return best_parent / best_div; + req->rate = best_parent / best_div; + + return 0; } static int sun4i_dclk_set_rate(struct clk_hw *hw, unsigned long rate, @@ -155,7 +157,7 @@ static const struct clk_ops sun4i_dclk_ops = { .is_enabled = sun4i_dclk_is_enabled, .recalc_rate = sun4i_dclk_recalc_rate, - .round_rate = sun4i_dclk_round_rate, + .determine_rate = sun4i_dclk_determine_rate, .set_rate = sun4i_dclk_set_rate, .get_phase = sun4i_dclk_get_phase, diff --git a/drivers/gpu/drm/tegra/hdmi.c b/drivers/gpu/drm/tegra/hdmi.c index 8cd2969e7d4b..c4820f5e7658 100644 --- a/drivers/gpu/drm/tegra/hdmi.c +++ b/drivers/gpu/drm/tegra/hdmi.c @@ -658,7 +658,7 @@ static void tegra_hdmi_write_infopack(struct tegra_hdmi *hdmi, const void *data, { const u8 *ptr = data; unsigned long offset; - size_t i, j; + size_t i; u32 value; switch (ptr[0]) { @@ -691,7 +691,7 @@ static void tegra_hdmi_write_infopack(struct tegra_hdmi *hdmi, const void *data, * - subpack_low: bytes 0 - 3 * - subpack_high: bytes 4 - 6 (with byte 7 padded to 0x00) */ - for (i = 3, j = 0; i < size; i += 7, j += 8) { + for (i = 3; i < size; i += 7) { size_t rem = size - i, num = min_t(size_t, rem, 4); value = tegra_hdmi_subpack(&ptr[i], num); diff --git a/drivers/gpu/drm/tegra/sor.c b/drivers/gpu/drm/tegra/sor.c index 21f3dfdcc5c9..bc7dd562cf6b 100644 --- a/drivers/gpu/drm/tegra/sor.c +++ b/drivers/gpu/drm/tegra/sor.c @@ -1864,7 +1864,7 @@ static void tegra_sor_hdmi_write_infopack(struct tegra_sor *sor, { const u8 *ptr = data; unsigned long offset; - size_t i, j; + size_t i; u32 value; switch (ptr[0]) { @@ -1897,7 +1897,7 @@ static void tegra_sor_hdmi_write_infopack(struct tegra_sor *sor, * - subpack_low: bytes 0 - 3 * - subpack_high: bytes 4 - 6 (with byte 7 padded to 0x00) */ - for (i = 3, j = 0; i < size; i += 7, j += 8) { + for (i = 3; i < size; i += 7) { size_t rem = size - i, num = min_t(size_t, rem, 4); value = tegra_sor_hdmi_subpack(&ptr[i], num); diff --git a/drivers/gpu/drm/tests/.kunitconfig b/drivers/gpu/drm/tests/.kunitconfig index 6ec04b4c979d..5be8e71f45d5 100644 --- a/drivers/gpu/drm/tests/.kunitconfig +++ b/drivers/gpu/drm/tests/.kunitconfig @@ -1,3 +1,5 @@ CONFIG_KUNIT=y CONFIG_DRM=y +CONFIG_DRM_VKMS=y +CONFIG_DRM_FBDEV_EMULATION=y CONFIG_DRM_KUNIT_TEST=y diff --git a/drivers/gpu/drm/ttm/tests/ttm_bo_test.c b/drivers/gpu/drm/ttm/tests/ttm_bo_test.c index 6c77550c51af..5426b435f702 100644 --- a/drivers/gpu/drm/ttm/tests/ttm_bo_test.c +++ b/drivers/gpu/drm/ttm/tests/ttm_bo_test.c @@ -379,7 +379,7 @@ static void ttm_bo_unreserve_bulk(struct kunit *test) dma_resv_fini(resv); } -static void ttm_bo_put_basic(struct kunit *test) +static void ttm_bo_fini_basic(struct kunit *test) { struct ttm_test_devices *priv = test->priv; struct ttm_buffer_object *bo; @@ -410,7 +410,7 @@ static void ttm_bo_put_basic(struct kunit *test) dma_resv_unlock(bo->base.resv); KUNIT_EXPECT_EQ(test, err, 0); - ttm_bo_put(bo); + ttm_bo_fini(bo); } static const char *mock_name(struct dma_fence *f) @@ -423,7 +423,7 @@ static const struct dma_fence_ops mock_fence_ops = { .get_timeline_name = mock_name, }; -static void ttm_bo_put_shared_resv(struct kunit *test) +static void ttm_bo_fini_shared_resv(struct kunit *test) { struct ttm_test_devices *priv = test->priv; struct ttm_buffer_object *bo; @@ -463,7 +463,7 @@ static void ttm_bo_put_shared_resv(struct kunit *test) bo->type = ttm_bo_type_device; bo->base.resv = external_resv; - ttm_bo_put(bo); + ttm_bo_fini(bo); } static void ttm_bo_pin_basic(struct kunit *test) @@ -616,8 +616,8 @@ static struct kunit_case ttm_bo_test_cases[] = { KUNIT_CASE(ttm_bo_unreserve_basic), KUNIT_CASE(ttm_bo_unreserve_pinned), KUNIT_CASE(ttm_bo_unreserve_bulk), - KUNIT_CASE(ttm_bo_put_basic), - KUNIT_CASE(ttm_bo_put_shared_resv), + KUNIT_CASE(ttm_bo_fini_basic), + KUNIT_CASE(ttm_bo_fini_shared_resv), KUNIT_CASE(ttm_bo_pin_basic), KUNIT_CASE(ttm_bo_pin_unpin_resource), KUNIT_CASE(ttm_bo_multiple_pin_one_unpin), diff --git a/drivers/gpu/drm/ttm/tests/ttm_bo_validate_test.c b/drivers/gpu/drm/ttm/tests/ttm_bo_validate_test.c index 1bcc67977f48..3a1eef83190c 100644 --- a/drivers/gpu/drm/ttm/tests/ttm_bo_validate_test.c +++ b/drivers/gpu/drm/ttm/tests/ttm_bo_validate_test.c @@ -144,7 +144,7 @@ static void ttm_bo_init_reserved_sys_man(struct kunit *test) drm_mm_node_allocated(&bo->base.vma_node.vm_node)); ttm_resource_free(bo, &bo->resource); - ttm_bo_put(bo); + ttm_bo_fini(bo); } static void ttm_bo_init_reserved_mock_man(struct kunit *test) @@ -186,7 +186,7 @@ static void ttm_bo_init_reserved_mock_man(struct kunit *test) drm_mm_node_allocated(&bo->base.vma_node.vm_node)); ttm_resource_free(bo, &bo->resource); - ttm_bo_put(bo); + ttm_bo_fini(bo); ttm_mock_manager_fini(priv->ttm_dev, mem_type); } @@ -221,7 +221,7 @@ static void ttm_bo_init_reserved_resv(struct kunit *test) KUNIT_EXPECT_PTR_EQ(test, bo->base.resv, &resv); ttm_resource_free(bo, &bo->resource); - ttm_bo_put(bo); + ttm_bo_fini(bo); } static void ttm_bo_validate_basic(struct kunit *test) @@ -265,7 +265,7 @@ static void ttm_bo_validate_basic(struct kunit *test) KUNIT_EXPECT_EQ(test, bo->resource->placement, DRM_BUDDY_TOPDOWN_ALLOCATION); - ttm_bo_put(bo); + ttm_bo_fini(bo); ttm_mock_manager_fini(priv->ttm_dev, snd_mem); } @@ -292,7 +292,7 @@ static void ttm_bo_validate_invalid_placement(struct kunit *test) KUNIT_EXPECT_EQ(test, err, -ENOMEM); - ttm_bo_put(bo); + ttm_bo_fini(bo); } static void ttm_bo_validate_failed_alloc(struct kunit *test) @@ -321,7 +321,7 @@ static void ttm_bo_validate_failed_alloc(struct kunit *test) KUNIT_EXPECT_EQ(test, err, -ENOMEM); - ttm_bo_put(bo); + ttm_bo_fini(bo); ttm_bad_manager_fini(priv->ttm_dev, mem_type); } @@ -353,7 +353,7 @@ static void ttm_bo_validate_pinned(struct kunit *test) ttm_bo_unpin(bo); dma_resv_unlock(bo->base.resv); - ttm_bo_put(bo); + ttm_bo_fini(bo); } static const struct ttm_bo_validate_test_case ttm_mem_type_cases[] = { @@ -403,7 +403,7 @@ static void ttm_bo_validate_same_placement(struct kunit *test) KUNIT_EXPECT_EQ(test, err, 0); KUNIT_EXPECT_EQ(test, ctx_val.bytes_moved, 0); - ttm_bo_put(bo); + ttm_bo_fini(bo); if (params->mem_type != TTM_PL_SYSTEM) ttm_mock_manager_fini(priv->ttm_dev, params->mem_type); @@ -452,7 +452,7 @@ static void ttm_bo_validate_busy_placement(struct kunit *test) KUNIT_EXPECT_EQ(test, bo->resource->mem_type, snd_mem); KUNIT_ASSERT_TRUE(test, list_is_singular(&man->lru[bo->priority])); - ttm_bo_put(bo); + ttm_bo_fini(bo); ttm_bad_manager_fini(priv->ttm_dev, fst_mem); ttm_mock_manager_fini(priv->ttm_dev, snd_mem); } @@ -495,7 +495,7 @@ static void ttm_bo_validate_multihop(struct kunit *test) KUNIT_EXPECT_EQ(test, ctx_val.bytes_moved, size * 2); KUNIT_EXPECT_EQ(test, bo->resource->mem_type, final_mem); - ttm_bo_put(bo); + ttm_bo_fini(bo); ttm_mock_manager_fini(priv->ttm_dev, fst_mem); ttm_mock_manager_fini(priv->ttm_dev, tmp_mem); @@ -567,7 +567,7 @@ static void ttm_bo_validate_no_placement_signaled(struct kunit *test) KUNIT_ASSERT_TRUE(test, flags & TTM_TT_FLAG_ZERO_ALLOC); } - ttm_bo_put(bo); + ttm_bo_fini(bo); } static int threaded_dma_resv_signal(void *arg) @@ -635,7 +635,7 @@ static void ttm_bo_validate_no_placement_not_signaled(struct kunit *test) /* Make sure we have an idle object at this point */ dma_resv_wait_timeout(bo->base.resv, usage, false, MAX_SCHEDULE_TIMEOUT); - ttm_bo_put(bo); + ttm_bo_fini(bo); } static void ttm_bo_validate_move_fence_signaled(struct kunit *test) @@ -668,7 +668,7 @@ static void ttm_bo_validate_move_fence_signaled(struct kunit *test) KUNIT_EXPECT_EQ(test, bo->resource->mem_type, mem_type); KUNIT_EXPECT_EQ(test, ctx.bytes_moved, size); - ttm_bo_put(bo); + ttm_bo_fini(bo); dma_fence_put(man->move); } @@ -753,7 +753,7 @@ static void ttm_bo_validate_move_fence_not_signaled(struct kunit *test) else KUNIT_EXPECT_EQ(test, bo->resource->mem_type, fst_mem); - ttm_bo_put(bo); + ttm_bo_fini(bo); ttm_mock_manager_fini(priv->ttm_dev, fst_mem); ttm_mock_manager_fini(priv->ttm_dev, snd_mem); } @@ -807,8 +807,8 @@ static void ttm_bo_validate_happy_evict(struct kunit *test) KUNIT_EXPECT_EQ(test, bos[1].resource->mem_type, mem_type); for (i = 0; i < bo_no; i++) - ttm_bo_put(&bos[i]); - ttm_bo_put(bo_val); + ttm_bo_fini(&bos[i]); + ttm_bo_fini(bo_val); ttm_mock_manager_fini(priv->ttm_dev, mem_type); ttm_mock_manager_fini(priv->ttm_dev, mem_multihop); @@ -852,12 +852,12 @@ static void ttm_bo_validate_all_pinned_evict(struct kunit *test) KUNIT_EXPECT_EQ(test, err, -ENOMEM); - ttm_bo_put(bo_small); + ttm_bo_fini(bo_small); ttm_bo_reserve(bo_big, false, false, NULL); ttm_bo_unpin(bo_big); dma_resv_unlock(bo_big->base.resv); - ttm_bo_put(bo_big); + ttm_bo_fini(bo_big); ttm_mock_manager_fini(priv->ttm_dev, mem_type); ttm_mock_manager_fini(priv->ttm_dev, mem_multihop); @@ -916,13 +916,13 @@ static void ttm_bo_validate_allowed_only_evict(struct kunit *test) KUNIT_EXPECT_EQ(test, bo_evictable->resource->mem_type, mem_type_evict); KUNIT_EXPECT_EQ(test, ctx_val.bytes_moved, size * 2 + BO_SIZE); - ttm_bo_put(bo); - ttm_bo_put(bo_evictable); + ttm_bo_fini(bo); + ttm_bo_fini(bo_evictable); ttm_bo_reserve(bo_pinned, false, false, NULL); ttm_bo_unpin(bo_pinned); dma_resv_unlock(bo_pinned->base.resv); - ttm_bo_put(bo_pinned); + ttm_bo_fini(bo_pinned); ttm_mock_manager_fini(priv->ttm_dev, mem_type); ttm_mock_manager_fini(priv->ttm_dev, mem_multihop); @@ -973,8 +973,8 @@ static void ttm_bo_validate_deleted_evict(struct kunit *test) KUNIT_EXPECT_NULL(test, bo_big->ttm); KUNIT_EXPECT_NULL(test, bo_big->resource); - ttm_bo_put(bo_small); - ttm_bo_put(bo_big); + ttm_bo_fini(bo_small); + ttm_bo_fini(bo_big); ttm_mock_manager_fini(priv->ttm_dev, mem_type); } @@ -1025,8 +1025,8 @@ static void ttm_bo_validate_busy_domain_evict(struct kunit *test) KUNIT_EXPECT_EQ(test, bo_init->resource->mem_type, mem_type); KUNIT_EXPECT_NULL(test, bo_val->resource); - ttm_bo_put(bo_init); - ttm_bo_put(bo_val); + ttm_bo_fini(bo_init); + ttm_bo_fini(bo_val); ttm_mock_manager_fini(priv->ttm_dev, mem_type); ttm_bad_manager_fini(priv->ttm_dev, mem_type_evict); @@ -1070,8 +1070,8 @@ static void ttm_bo_validate_evict_gutting(struct kunit *test) KUNIT_ASSERT_NULL(test, bo_evict->resource); KUNIT_ASSERT_TRUE(test, bo_evict->ttm->page_flags & TTM_TT_FLAG_ZERO_ALLOC); - ttm_bo_put(bo_evict); - ttm_bo_put(bo); + ttm_bo_fini(bo_evict); + ttm_bo_fini(bo); ttm_mock_manager_fini(priv->ttm_dev, mem_type); } @@ -1128,9 +1128,9 @@ static void ttm_bo_validate_recrusive_evict(struct kunit *test) ttm_mock_manager_fini(priv->ttm_dev, mem_type); ttm_mock_manager_fini(priv->ttm_dev, mem_type_evict); - ttm_bo_put(bo_val); - ttm_bo_put(bo_tt); - ttm_bo_put(bo_mock); + ttm_bo_fini(bo_val); + ttm_bo_fini(bo_tt); + ttm_bo_fini(bo_mock); } static struct kunit_case ttm_bo_validate_test_cases[] = { diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 29423ceeec5c..fba2a68a556e 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -318,18 +318,17 @@ static void ttm_bo_release(struct kref *kref) bo->destroy(bo); } -/** - * ttm_bo_put - * - * @bo: The buffer object. - * - * Unreference a buffer object. - */ +/* TODO: remove! */ void ttm_bo_put(struct ttm_buffer_object *bo) { kref_put(&bo->kref, ttm_bo_release); } -EXPORT_SYMBOL(ttm_bo_put); + +void ttm_bo_fini(struct ttm_buffer_object *bo) +{ + ttm_bo_put(bo); +} +EXPORT_SYMBOL(ttm_bo_fini); static int ttm_bo_bounce_temp_buffer(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, diff --git a/drivers/gpu/drm/ttm/ttm_bo_internal.h b/drivers/gpu/drm/ttm/ttm_bo_internal.h index 9d8b747a34db..e0d48eac74b0 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_internal.h +++ b/drivers/gpu/drm/ttm/ttm_bo_internal.h @@ -55,4 +55,6 @@ ttm_bo_get_unless_zero(struct ttm_buffer_object *bo) return bo; } +void ttm_bo_put(struct ttm_buffer_object *bo); + #endif diff --git a/drivers/gpu/drm/vc4/Kconfig b/drivers/gpu/drm/vc4/Kconfig index 123ab0ce1781..bb8c40be3250 100644 --- a/drivers/gpu/drm/vc4/Kconfig +++ b/drivers/gpu/drm/vc4/Kconfig @@ -35,6 +35,7 @@ config DRM_VC4_HDMI_CEC bool "Broadcom VC4 HDMI CEC Support" depends on DRM_VC4 select CEC_CORE + select DRM_DISPLAY_HDMI_CEC_HELPER help Choose this option if you have a Broadcom VC4 GPU and want to use CEC. diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.c b/drivers/gpu/drm/vc4/vc4_hdmi.c index 07c91b450f93..049c92dd5d27 100644 --- a/drivers/gpu/drm/vc4/vc4_hdmi.c +++ b/drivers/gpu/drm/vc4/vc4_hdmi.c @@ -32,6 +32,7 @@ */ #include <drm/display/drm_hdmi_audio_helper.h> +#include <drm/display/drm_hdmi_cec_helper.h> #include <drm/display/drm_hdmi_helper.h> #include <drm/display/drm_hdmi_state_helper.h> #include <drm/display/drm_scdc_helper.h> @@ -375,14 +376,6 @@ static void vc4_hdmi_handle_hotplug(struct vc4_hdmi *vc4_hdmi, drm_atomic_helper_connector_hdmi_hotplug(connector, status); - if (status == connector_status_disconnected) { - cec_phys_addr_invalidate(vc4_hdmi->cec_adap); - return; - } - - cec_s_phys_addr(vc4_hdmi->cec_adap, - connector->display_info.source_physical_address, false); - if (status != connector_status_connected) return; @@ -2384,8 +2377,8 @@ static irqreturn_t vc4_cec_irq_handler_rx_thread(int irq, void *priv) struct vc4_hdmi *vc4_hdmi = priv; if (vc4_hdmi->cec_rx_msg.len) - cec_received_msg(vc4_hdmi->cec_adap, - &vc4_hdmi->cec_rx_msg); + drm_connector_hdmi_cec_received_msg(&vc4_hdmi->connector, + &vc4_hdmi->cec_rx_msg); return IRQ_HANDLED; } @@ -2395,15 +2388,17 @@ static irqreturn_t vc4_cec_irq_handler_tx_thread(int irq, void *priv) struct vc4_hdmi *vc4_hdmi = priv; if (vc4_hdmi->cec_tx_ok) { - cec_transmit_done(vc4_hdmi->cec_adap, CEC_TX_STATUS_OK, - 0, 0, 0, 0); + drm_connector_hdmi_cec_transmit_done(&vc4_hdmi->connector, + CEC_TX_STATUS_OK, + 0, 0, 0, 0); } else { /* * This CEC implementation makes 1 retry, so if we * get a NACK, then that means it made 2 attempts. */ - cec_transmit_done(vc4_hdmi->cec_adap, CEC_TX_STATUS_NACK, - 0, 2, 0, 0); + drm_connector_hdmi_cec_transmit_done(&vc4_hdmi->connector, + CEC_TX_STATUS_NACK, + 0, 2, 0, 0); } return IRQ_HANDLED; } @@ -2560,9 +2555,9 @@ static irqreturn_t vc4_cec_irq_handler(int irq, void *priv) return ret; } -static int vc4_hdmi_cec_enable(struct cec_adapter *adap) +static int vc4_hdmi_cec_enable(struct drm_connector *connector) { - struct vc4_hdmi *vc4_hdmi = cec_get_drvdata(adap); + struct vc4_hdmi *vc4_hdmi = connector_to_vc4_hdmi(connector); struct drm_device *drm = vc4_hdmi->connector.dev; /* clock period in microseconds */ const u32 usecs = 1000000 / CEC_CLOCK_FREQ; @@ -2627,9 +2622,9 @@ static int vc4_hdmi_cec_enable(struct cec_adapter *adap) return 0; } -static int vc4_hdmi_cec_disable(struct cec_adapter *adap) +static int vc4_hdmi_cec_disable(struct drm_connector *connector) { - struct vc4_hdmi *vc4_hdmi = cec_get_drvdata(adap); + struct vc4_hdmi *vc4_hdmi = connector_to_vc4_hdmi(connector); struct drm_device *drm = vc4_hdmi->connector.dev; unsigned long flags; int idx; @@ -2663,17 +2658,17 @@ static int vc4_hdmi_cec_disable(struct cec_adapter *adap) return 0; } -static int vc4_hdmi_cec_adap_enable(struct cec_adapter *adap, bool enable) +static int vc4_hdmi_cec_adap_enable(struct drm_connector *connector, bool enable) { if (enable) - return vc4_hdmi_cec_enable(adap); + return vc4_hdmi_cec_enable(connector); else - return vc4_hdmi_cec_disable(adap); + return vc4_hdmi_cec_disable(connector); } -static int vc4_hdmi_cec_adap_log_addr(struct cec_adapter *adap, u8 log_addr) +static int vc4_hdmi_cec_adap_log_addr(struct drm_connector *connector, u8 log_addr) { - struct vc4_hdmi *vc4_hdmi = cec_get_drvdata(adap); + struct vc4_hdmi *vc4_hdmi = connector_to_vc4_hdmi(connector); struct drm_device *drm = vc4_hdmi->connector.dev; unsigned long flags; int idx; @@ -2699,10 +2694,10 @@ static int vc4_hdmi_cec_adap_log_addr(struct cec_adapter *adap, u8 log_addr) return 0; } -static int vc4_hdmi_cec_adap_transmit(struct cec_adapter *adap, u8 attempts, +static int vc4_hdmi_cec_adap_transmit(struct drm_connector *connector, u8 attempts, u32 signal_free_time, struct cec_msg *msg) { - struct vc4_hdmi *vc4_hdmi = cec_get_drvdata(adap); + struct vc4_hdmi *vc4_hdmi = connector_to_vc4_hdmi(connector); struct drm_device *dev = vc4_hdmi->connector.dev; unsigned long flags; u32 val; @@ -2745,84 +2740,65 @@ static int vc4_hdmi_cec_adap_transmit(struct cec_adapter *adap, u8 attempts, return 0; } -static const struct cec_adap_ops vc4_hdmi_cec_adap_ops = { - .adap_enable = vc4_hdmi_cec_adap_enable, - .adap_log_addr = vc4_hdmi_cec_adap_log_addr, - .adap_transmit = vc4_hdmi_cec_adap_transmit, -}; - -static void vc4_hdmi_cec_release(void *ptr) -{ - struct vc4_hdmi *vc4_hdmi = ptr; - - cec_unregister_adapter(vc4_hdmi->cec_adap); - vc4_hdmi->cec_adap = NULL; -} - -static int vc4_hdmi_cec_init(struct vc4_hdmi *vc4_hdmi) +static int vc4_hdmi_cec_init(struct drm_connector *connector) { - struct cec_connector_info conn_info; + struct vc4_hdmi *vc4_hdmi = connector_to_vc4_hdmi(connector); struct platform_device *pdev = vc4_hdmi->pdev; struct device *dev = &pdev->dev; int ret; - if (!of_property_present(dev->of_node, "interrupts")) { - dev_warn(dev, "'interrupts' DT property is missing, no CEC\n"); - return 0; - } - - vc4_hdmi->cec_adap = cec_allocate_adapter(&vc4_hdmi_cec_adap_ops, - vc4_hdmi, - vc4_hdmi->variant->card_name, - CEC_CAP_DEFAULTS | - CEC_CAP_CONNECTOR_INFO, 1); - ret = PTR_ERR_OR_ZERO(vc4_hdmi->cec_adap); - if (ret < 0) - return ret; - - cec_fill_conn_info_from_drm(&conn_info, &vc4_hdmi->connector); - cec_s_conn_info(vc4_hdmi->cec_adap, &conn_info); - if (vc4_hdmi->variant->external_irq_controller) { ret = devm_request_threaded_irq(dev, platform_get_irq_byname(pdev, "cec-rx"), vc4_cec_irq_handler_rx_bare, vc4_cec_irq_handler_rx_thread, 0, "vc4 hdmi cec rx", vc4_hdmi); if (ret) - goto err_delete_cec_adap; + return ret; ret = devm_request_threaded_irq(dev, platform_get_irq_byname(pdev, "cec-tx"), vc4_cec_irq_handler_tx_bare, vc4_cec_irq_handler_tx_thread, 0, "vc4 hdmi cec tx", vc4_hdmi); if (ret) - goto err_delete_cec_adap; + return ret; } else { ret = devm_request_threaded_irq(dev, platform_get_irq(pdev, 0), vc4_cec_irq_handler, vc4_cec_irq_handler_thread, 0, "vc4 hdmi cec", vc4_hdmi); if (ret) - goto err_delete_cec_adap; + return ret; } - ret = cec_register_adapter(vc4_hdmi->cec_adap, &pdev->dev); - if (ret < 0) - goto err_delete_cec_adap; + return 0; +} + +static const struct drm_connector_hdmi_cec_funcs vc4_hdmi_cec_funcs = { + .init = vc4_hdmi_cec_init, + .enable = vc4_hdmi_cec_adap_enable, + .log_addr = vc4_hdmi_cec_adap_log_addr, + .transmit = vc4_hdmi_cec_adap_transmit, +}; + +static int vc4_hdmi_cec_register(struct vc4_hdmi *vc4_hdmi) +{ + struct platform_device *pdev = vc4_hdmi->pdev; + struct device *dev = &pdev->dev; + + if (!of_property_present(dev->of_node, "interrupts")) { + dev_warn(dev, "'interrupts' DT property is missing, no CEC\n"); + return 0; + } /* - * NOTE: Strictly speaking, we should probably use a DRM-managed - * registration there to avoid removing the CEC adapter by the - * time the DRM driver doesn't have any user anymore. + * NOTE: the CEC adapter will be unregistered by drmm cleanup from + * drm_managed_release(), which is called from drm_dev_release() + * during device unbind. * * However, the CEC framework already cleans up the CEC adapter * only when the last user has closed its file descriptor, so we * don't need to handle it in DRM. * - * By the time the device-managed hook is executed, we will give - * up our reference to the CEC adapter and therefore don't - * really care when it's actually freed. - * * There's still a problematic sequence: if we unregister our * CEC adapter, but the userspace keeps a handle on the CEC * adapter but not the DRM device for some reason. In such a @@ -2833,19 +2809,14 @@ static int vc4_hdmi_cec_init(struct vc4_hdmi *vc4_hdmi) * the CEC framework already handles this too, by calling * cec_is_registered() in cec_ioctl() and cec_poll(). */ - ret = devm_add_action_or_reset(dev, vc4_hdmi_cec_release, vc4_hdmi); - if (ret) - return ret; - - return 0; - -err_delete_cec_adap: - cec_delete_adapter(vc4_hdmi->cec_adap); - - return ret; + return drmm_connector_hdmi_cec_register(&vc4_hdmi->connector, + &vc4_hdmi_cec_funcs, + vc4_hdmi->variant->card_name, + 1, + &pdev->dev); } #else -static int vc4_hdmi_cec_init(struct vc4_hdmi *vc4_hdmi) +static int vc4_hdmi_cec_register(struct vc4_hdmi *vc4_hdmi) { return 0; } @@ -3250,7 +3221,7 @@ static int vc4_hdmi_bind(struct device *dev, struct device *master, void *data) if (ret) goto err_put_runtime_pm; - ret = vc4_hdmi_cec_init(vc4_hdmi); + ret = vc4_hdmi_cec_register(vc4_hdmi); if (ret) goto err_put_runtime_pm; diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.h b/drivers/gpu/drm/vc4/vc4_hdmi.h index a31157c99bee..8d069718df00 100644 --- a/drivers/gpu/drm/vc4/vc4_hdmi.h +++ b/drivers/gpu/drm/vc4/vc4_hdmi.h @@ -147,7 +147,6 @@ struct vc4_hdmi { */ bool disable_wifi_frequencies; - struct cec_adapter *cec_adap; struct cec_msg cec_rx_msg; bool cec_tx_ok; bool cec_irq_was_rx; diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_gem.c b/drivers/gpu/drm/vmwgfx/vmwgfx_gem.c index eedf1fe60be7..39f8c46550c2 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_gem.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_gem.c @@ -37,7 +37,7 @@ static void vmw_gem_object_free(struct drm_gem_object *gobj) { struct ttm_buffer_object *bo = drm_gem_ttm_of_gem(gobj); if (bo) - ttm_bo_put(bo); + ttm_bo_fini(bo); } static int vmw_gem_object_open(struct drm_gem_object *obj, diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c index 870f43347281..ebcd191034df 100644 --- a/drivers/gpu/drm/xe/xe_bo.c +++ b/drivers/gpu/drm/xe/xe_bo.c @@ -1696,7 +1696,7 @@ static void xe_gem_object_free(struct drm_gem_object *obj) * refcount directly if needed. */ __xe_bo_vunmap(gem_to_xe_bo(obj)); - ttm_bo_put(container_of(obj, struct ttm_buffer_object, base)); + ttm_bo_fini(container_of(obj, struct ttm_buffer_object, base)); } static void xe_gem_object_close(struct drm_gem_object *obj, diff --git a/drivers/gpu/host1x/bus.c b/drivers/gpu/host1x/bus.c index 344cc9e741c1..723a80895cd4 100644 --- a/drivers/gpu/host1x/bus.c +++ b/drivers/gpu/host1x/bus.c @@ -471,6 +471,18 @@ static int host1x_device_add(struct host1x *host1x, mutex_unlock(&clients_lock); + /* + * Add device even if there are no subdevs to ensure syncpoint functionality + * is available regardless of whether any engine subdevices are present + */ + if (list_empty(&device->subdevs)) { + err = device_add(&device->dev); + if (err < 0) + dev_err(&device->dev, "failed to add device: %d\n", err); + else + device->registered = true; + } + return 0; } diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c index 1f93e5e276c0..e365df6af353 100644 --- a/drivers/gpu/host1x/dev.c +++ b/drivers/gpu/host1x/dev.c @@ -585,14 +585,8 @@ static int host1x_probe(struct platform_device *pdev) } host->clk = devm_clk_get(&pdev->dev, NULL); - if (IS_ERR(host->clk)) { - err = PTR_ERR(host->clk); - - if (err != -EPROBE_DEFER) - dev_err(&pdev->dev, "failed to get clock: %d\n", err); - - return err; - } + if (IS_ERR(host->clk)) + return dev_err_probe(&pdev->dev, PTR_ERR(host->clk), "failed to get clock\n"); err = host1x_get_resets(host); if (err) @@ -821,6 +815,7 @@ u64 host1x_get_dma_mask(struct host1x *host1x) } EXPORT_SYMBOL(host1x_get_dma_mask); +MODULE_SOFTDEP("post: tegra-drm"); MODULE_AUTHOR("Thierry Reding <thierry.reding@avionic-design.de>"); MODULE_AUTHOR("Terje Bergstrom <tbergstrom@nvidia.com>"); MODULE_DESCRIPTION("Host1x driver for Tegra products"); diff --git a/drivers/gpu/host1x/hw/channel_hw.c b/drivers/gpu/host1x/hw/channel_hw.c index d44b8de890be..2df6a16d484e 100644 --- a/drivers/gpu/host1x/hw/channel_hw.c +++ b/drivers/gpu/host1x/hw/channel_hw.c @@ -47,24 +47,11 @@ static void trace_write_gather(struct host1x_cdma *cdma, struct host1x_bo *bo, } } -static void submit_wait(struct host1x_job *job, u32 id, u32 threshold, - u32 next_class) +static void submit_wait(struct host1x_job *job, u32 id, u32 threshold) { struct host1x_cdma *cdma = &job->channel->cdma; -#if HOST1X_HW >= 6 - u32 stream_id; - - /* - * If a memory context has been set, use it. Otherwise - * (if context isolation is disabled) use the engine's - * firmware stream ID. - */ - if (job->memory_context) - stream_id = job->memory_context->stream_id; - else - stream_id = job->engine_fallback_streamid; - +#if HOST1X_HW >= 2 host1x_cdma_push_wide(cdma, host1x_opcode_setclass( HOST1X_CLASS_HOST1X, @@ -76,23 +63,6 @@ static void submit_wait(struct host1x_job *job, u32 id, u32 threshold, id, HOST1X_OPCODE_NOP ); - host1x_cdma_push_wide(&job->channel->cdma, - host1x_opcode_setclass(job->class, 0, 0), - host1x_opcode_setpayload(stream_id), - host1x_opcode_setstreamid(job->engine_streamid_offset / 4), - HOST1X_OPCODE_NOP); -#elif HOST1X_HW >= 2 - host1x_cdma_push_wide(cdma, - host1x_opcode_setclass( - HOST1X_CLASS_HOST1X, - HOST1X_UCLASS_LOAD_SYNCPT_PAYLOAD_32, - /* WAIT_SYNCPT_32 is at SYNCPT_PAYLOAD_32+2 */ - BIT(0) | BIT(2) - ), - threshold, - id, - host1x_opcode_setclass(next_class, 0, 0) - ); #else /* TODO add waitchk or use waitbases or other mitigation */ host1x_cdma_push(cdma, @@ -103,6 +73,32 @@ static void submit_wait(struct host1x_job *job, u32 id, u32 threshold, ), host1x_class_host_wait_syncpt(id, threshold) ); +#endif +} + +static void submit_setclass(struct host1x_job *job, u32 next_class) +{ + struct host1x_cdma *cdma = &job->channel->cdma; + +#if HOST1X_HW >= 6 + u32 stream_id; + + /* + * If a memory context has been set, use it. Otherwise + * (if context isolation is disabled) use the engine's + * firmware stream ID. + */ + if (job->memory_context) + stream_id = job->memory_context->stream_id; + else + stream_id = job->engine_fallback_streamid; + + host1x_cdma_push_wide(cdma, + host1x_opcode_setclass(next_class, 0, 0), + host1x_opcode_setpayload(stream_id), + host1x_opcode_setstreamid(job->engine_streamid_offset / 4), + HOST1X_OPCODE_NOP); +#else host1x_cdma_push(cdma, host1x_opcode_setclass(next_class, 0, 0), HOST1X_OPCODE_NOP @@ -110,7 +106,8 @@ static void submit_wait(struct host1x_job *job, u32 id, u32 threshold, #endif } -static void submit_gathers(struct host1x_job *job, u32 job_syncpt_base) +static void submit_gathers(struct host1x_job *job, struct host1x_job_cmd *cmds, u32 num_cmds, + u32 job_syncpt_base) { struct host1x_cdma *cdma = &job->channel->cdma; #if HOST1X_HW < 6 @@ -119,8 +116,8 @@ static void submit_gathers(struct host1x_job *job, u32 job_syncpt_base) unsigned int i; u32 threshold; - for (i = 0; i < job->num_cmds; i++) { - struct host1x_job_cmd *cmd = &job->cmds[i]; + for (i = 0; i < num_cmds; i++) { + struct host1x_job_cmd *cmd = &cmds[i]; if (cmd->is_wait) { if (cmd->wait.relative) @@ -128,7 +125,8 @@ static void submit_gathers(struct host1x_job *job, u32 job_syncpt_base) else threshold = cmd->wait.threshold; - submit_wait(job, cmd->wait.id, threshold, cmd->wait.next_class); + submit_wait(job, cmd->wait.id, threshold); + submit_setclass(job, cmd->wait.next_class); } else { struct host1x_job_gather *g = &cmd->gather; @@ -216,7 +214,34 @@ static void channel_program_cdma(struct host1x_job *job) #if HOST1X_HW >= 6 u32 fence; + int i = 0; + + if (job->num_cmds == 0) + goto prefences_done; + if (!job->cmds[0].is_wait || job->cmds[0].wait.relative) + goto prefences_done; + + /* Enter host1x class with invalid stream ID for prefence waits. */ + host1x_cdma_push_wide(cdma, + host1x_opcode_acquire_mlock(1), + host1x_opcode_setclass(1, 0, 0), + host1x_opcode_setpayload(0), + host1x_opcode_setstreamid(0x1fffff)); + + for (i = 0; i < job->num_cmds; i++) { + struct host1x_job_cmd *cmd = &job->cmds[i]; + + if (!cmd->is_wait || cmd->wait.relative) + break; + + submit_wait(job, cmd->wait.id, cmd->wait.threshold); + } + + host1x_cdma_push(cdma, + HOST1X_OPCODE_NOP, + host1x_opcode_release_mlock(1)); +prefences_done: /* Enter engine class with invalid stream ID. */ host1x_cdma_push_wide(cdma, host1x_opcode_acquire_mlock(job->class), @@ -230,11 +255,12 @@ static void channel_program_cdma(struct host1x_job *job) host1x_opcode_nonincr(HOST1X_UCLASS_INCR_SYNCPT, 1), HOST1X_UCLASS_INCR_SYNCPT_INDX_F(job->syncpt->id) | HOST1X_UCLASS_INCR_SYNCPT_COND_F(4)); - submit_wait(job, job->syncpt->id, fence, job->class); + submit_wait(job, job->syncpt->id, fence); + submit_setclass(job, job->class); /* Submit work. */ job->syncpt_end = host1x_syncpt_incr_max(sp, job->syncpt_incrs); - submit_gathers(job, job->syncpt_end - job->syncpt_incrs); + submit_gathers(job, job->cmds + i, job->num_cmds - i, job->syncpt_end - job->syncpt_incrs); /* Before releasing MLOCK, ensure engine is idle again. */ fence = host1x_syncpt_incr_max(sp, 1); @@ -242,7 +268,7 @@ static void channel_program_cdma(struct host1x_job *job) host1x_opcode_nonincr(HOST1X_UCLASS_INCR_SYNCPT, 1), HOST1X_UCLASS_INCR_SYNCPT_INDX_F(job->syncpt->id) | HOST1X_UCLASS_INCR_SYNCPT_COND_F(4)); - submit_wait(job, job->syncpt->id, fence, job->class); + submit_wait(job, job->syncpt->id, fence); /* Release MLOCK. */ host1x_cdma_push(cdma, @@ -272,7 +298,7 @@ static void channel_program_cdma(struct host1x_job *job) job->syncpt_end = host1x_syncpt_incr_max(sp, job->syncpt_incrs); - submit_gathers(job, job->syncpt_end - job->syncpt_incrs); + submit_gathers(job, job->cmds, job->num_cmds, job->syncpt_end - job->syncpt_incrs); #endif } diff --git a/drivers/gpu/host1x/syncpt.c b/drivers/gpu/host1x/syncpt.c index f63d14a57a1d..acc7d82e0585 100644 --- a/drivers/gpu/host1x/syncpt.c +++ b/drivers/gpu/host1x/syncpt.c @@ -345,8 +345,6 @@ static void syncpt_release(struct kref *ref) sp->locked = false; - mutex_lock(&sp->host->syncpt_mutex); - host1x_syncpt_base_free(sp->base); kfree(sp->name); sp->base = NULL; @@ -369,7 +367,7 @@ void host1x_syncpt_put(struct host1x_syncpt *sp) if (!sp) return; - kref_put(&sp->ref, syncpt_release); + kref_put_mutex(&sp->ref, syncpt_release, &sp->host->syncpt_mutex); } EXPORT_SYMBOL(host1x_syncpt_put); diff --git a/include/drm/drm_bridge.h b/include/drm/drm_bridge.h index 76e05930f50e..0ff7ab4aa868 100644 --- a/include/drm/drm_bridge.h +++ b/include/drm/drm_bridge.h @@ -1362,6 +1362,13 @@ drm_bridge_get_current_state(struct drm_bridge *bridge) * drm_bridge_get_next_bridge() - Get the next bridge in the chain * @bridge: bridge object * + * The caller is responsible of having a reference to @bridge via + * drm_bridge_get() or equivalent. This function leaves the refcount of + * @bridge unmodified. + * + * The refcount of the returned bridge is incremented. Use drm_bridge_put() + * when done with it. + * * RETURNS: * the next bridge in the chain after @bridge, or NULL if @bridge is the last. */ @@ -1371,7 +1378,7 @@ drm_bridge_get_next_bridge(struct drm_bridge *bridge) if (list_is_last(&bridge->chain_node, &bridge->encoder->bridge_chain)) return NULL; - return list_next_entry(bridge, chain_node); + return drm_bridge_get(list_next_entry(bridge, chain_node)); } /** @@ -1434,15 +1441,61 @@ drm_bridge_chain_get_last_bridge(struct drm_encoder *encoder) } /** - * drm_for_each_bridge_in_chain() - Iterate over all bridges present in a chain + * drm_bridge_get_next_bridge_and_put - Get the next bridge in the chain + * and put the previous + * @bridge: bridge object + * + * Same as drm_bridge_get_next_bridge() but additionally puts the @bridge. + * + * RETURNS: + * the next bridge in the chain after @bridge, or NULL if @bridge is the last. + */ +static inline struct drm_bridge * +drm_bridge_get_next_bridge_and_put(struct drm_bridge *bridge) +{ + struct drm_bridge *next = drm_bridge_get_next_bridge(bridge); + + drm_bridge_put(bridge); + + return next; +} + +/** + * drm_for_each_bridge_in_chain_scoped - iterate over all bridges attached + * to an encoder * @encoder: the encoder to iterate bridges on * @bridge: a bridge pointer updated to point to the current bridge at each * iteration * * Iterate over all bridges present in the bridge chain attached to @encoder. + * + * Automatically gets/puts the bridge reference while iterating, and puts + * the reference even if returning or breaking in the middle of the loop. + */ +#define drm_for_each_bridge_in_chain_scoped(encoder, bridge) \ + for (struct drm_bridge *bridge __free(drm_bridge_put) = \ + drm_bridge_chain_get_first_bridge(encoder); \ + bridge; \ + bridge = drm_bridge_get_next_bridge_and_put(bridge)) + +/** + * drm_for_each_bridge_in_chain_from - iterate over all bridges starting + * from the given bridge + * @first_bridge: the bridge to start from + * @bridge: a bridge pointer updated to point to the current bridge at each + * iteration + * + * Iterate over all bridges in the encoder chain starting from + * @first_bridge, included. + * + * Automatically gets/puts the bridge reference while iterating, and puts + * the reference even if returning or breaking in the middle of the loop. */ -#define drm_for_each_bridge_in_chain(encoder, bridge) \ - list_for_each_entry(bridge, &(encoder)->bridge_chain, chain_node) +#define drm_for_each_bridge_in_chain_from(first_bridge, bridge) \ + for (struct drm_bridge *bridge __free(drm_bridge_put) = \ + drm_bridge_get(first_bridge); \ + bridge; \ + bridge = drm_bridge_get_next_bridge_and_put(bridge)) enum drm_mode_status drm_bridge_chain_mode_valid(struct drm_bridge *bridge, diff --git a/include/drm/drm_client.h b/include/drm/drm_client.h index 146ca80e35db..bdd845e383ef 100644 --- a/include/drm/drm_client.h +++ b/include/drm/drm_client.h @@ -220,6 +220,7 @@ int drm_client_modeset_check(struct drm_client_dev *client); int drm_client_modeset_commit_locked(struct drm_client_dev *client); int drm_client_modeset_commit(struct drm_client_dev *client); int drm_client_modeset_dpms(struct drm_client_dev *client, int mode); +int drm_client_modeset_wait_for_vblank(struct drm_client_dev *client, unsigned int crtc_index); /** * drm_client_for_each_modeset() - Iterate over client modesets diff --git a/include/drm/drm_gem_shmem_helper.h b/include/drm/drm_gem_shmem_helper.h index 92f5db84b9c2..589f7bfe7506 100644 --- a/include/drm/drm_gem_shmem_helper.h +++ b/include/drm/drm_gem_shmem_helper.h @@ -107,10 +107,12 @@ struct drm_gem_shmem_object { #define to_drm_gem_shmem_obj(obj) \ container_of(obj, struct drm_gem_shmem_object, base) +int drm_gem_shmem_init(struct drm_device *dev, struct drm_gem_shmem_object *shmem, size_t size); struct drm_gem_shmem_object *drm_gem_shmem_create(struct drm_device *dev, size_t size); struct drm_gem_shmem_object *drm_gem_shmem_create_with_mnt(struct drm_device *dev, size_t size, struct vfsmount *gemfs); +void drm_gem_shmem_release(struct drm_gem_shmem_object *shmem); void drm_gem_shmem_free(struct drm_gem_shmem_object *shmem); void drm_gem_shmem_put_pages_locked(struct drm_gem_shmem_object *shmem); diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h index 323a505e6e6a..fb88301b3c45 100644 --- a/include/drm/gpu_scheduler.h +++ b/include/drm/gpu_scheduler.h @@ -546,7 +546,7 @@ struct drm_sched_backend_ops { * @num_rqs: Number of run-queues. This is at most DRM_SCHED_PRIORITY_COUNT, * as there's usually one run-queue per priority, but could be less. * @sched_rq: An allocated array of run-queues of size @num_rqs; - * @job_scheduled: once @drm_sched_entity_do_release is called the scheduler + * @job_scheduled: once drm_sched_entity_flush() is called the scheduler * waits on this wait queue until all the scheduled jobs are * finished. * @job_id_count: used to assign unique id to the each job. diff --git a/include/drm/ttm/ttm_bo.h b/include/drm/ttm/ttm_bo.h index e664a96540eb..bca3a8849d47 100644 --- a/include/drm/ttm/ttm_bo.h +++ b/include/drm/ttm/ttm_bo.h @@ -391,7 +391,7 @@ int ttm_bo_wait_ctx(struct ttm_buffer_object *bo, int ttm_bo_validate(struct ttm_buffer_object *bo, struct ttm_placement *placement, struct ttm_operation_ctx *ctx); -void ttm_bo_put(struct ttm_buffer_object *bo); +void ttm_bo_fini(struct ttm_buffer_object *bo); void ttm_bo_set_bulk_move(struct ttm_buffer_object *bo, struct ttm_lru_bulk_move *bulk); bool ttm_bo_eviction_valuable(struct ttm_buffer_object *bo, diff --git a/include/uapi/drm/ivpu_accel.h b/include/uapi/drm/ivpu_accel.h index 160ee1411d4a..e470b0221e02 100644 --- a/include/uapi/drm/ivpu_accel.h +++ b/include/uapi/drm/ivpu_accel.h @@ -90,6 +90,7 @@ extern "C" { #define DRM_IVPU_PARAM_TILE_CONFIG 11 #define DRM_IVPU_PARAM_SKU 12 #define DRM_IVPU_PARAM_CAPABILITIES 13 +#define DRM_IVPU_PARAM_PREEMPT_BUFFER_SIZE 14 #define DRM_IVPU_PLATFORM_TYPE_SILICON 0 @@ -176,6 +177,9 @@ struct drm_ivpu_param { * * %DRM_IVPU_PARAM_CAPABILITIES: * Supported capabilities (read-only) + * + * %DRM_IVPU_PARAM_PREEMPT_BUFFER_SIZE: + * Size of the preemption buffer (read-only) */ __u32 param; @@ -371,6 +375,13 @@ struct drm_ivpu_cmdq_submit { * to be executed. The offset has to be 8-byte aligned. */ __u32 commands_offset; + /** + * @preempt_buffer_index: + * + * Index of the preemption buffer in the buffers_ptr array. + */ + __u32 preempt_buffer_index; + __u32 reserved; }; /* drm_ivpu_bo_wait job status codes */ diff --git a/include/uapi/drm/panfrost_drm.h b/include/uapi/drm/panfrost_drm.h index ed67510395bd..e8b47c9f6976 100644 --- a/include/uapi/drm/panfrost_drm.h +++ b/include/uapi/drm/panfrost_drm.h @@ -22,6 +22,8 @@ extern "C" { #define DRM_PANFROST_PERFCNT_DUMP 0x07 #define DRM_PANFROST_MADVISE 0x08 #define DRM_PANFROST_SET_LABEL_BO 0x09 +#define DRM_PANFROST_JM_CTX_CREATE 0x0a +#define DRM_PANFROST_JM_CTX_DESTROY 0x0b #define DRM_IOCTL_PANFROST_SUBMIT DRM_IOW(DRM_COMMAND_BASE + DRM_PANFROST_SUBMIT, struct drm_panfrost_submit) #define DRM_IOCTL_PANFROST_WAIT_BO DRM_IOW(DRM_COMMAND_BASE + DRM_PANFROST_WAIT_BO, struct drm_panfrost_wait_bo) @@ -31,6 +33,8 @@ extern "C" { #define DRM_IOCTL_PANFROST_GET_BO_OFFSET DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_GET_BO_OFFSET, struct drm_panfrost_get_bo_offset) #define DRM_IOCTL_PANFROST_MADVISE DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_MADVISE, struct drm_panfrost_madvise) #define DRM_IOCTL_PANFROST_SET_LABEL_BO DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_SET_LABEL_BO, struct drm_panfrost_set_label_bo) +#define DRM_IOCTL_PANFROST_JM_CTX_CREATE DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_JM_CTX_CREATE, struct drm_panfrost_jm_ctx_create) +#define DRM_IOCTL_PANFROST_JM_CTX_DESTROY DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_JM_CTX_DESTROY, struct drm_panfrost_jm_ctx_destroy) /* * Unstable ioctl(s): only exposed when the unsafe unstable_ioctls module @@ -71,6 +75,12 @@ struct drm_panfrost_submit { /** A combination of PANFROST_JD_REQ_* */ __u32 requirements; + + /** JM context handle. Zero if you want to use the default context. */ + __u32 jm_ctx_handle; + + /** Padding field. MBZ. */ + __u32 pad; }; /** @@ -177,6 +187,7 @@ enum drm_panfrost_param { DRM_PANFROST_PARAM_AFBC_FEATURES, DRM_PANFROST_PARAM_SYSTEM_TIMESTAMP, DRM_PANFROST_PARAM_SYSTEM_TIMESTAMP_FREQUENCY, + DRM_PANFROST_PARAM_ALLOWED_JM_CTX_PRIORITIES, }; struct drm_panfrost_get_param { @@ -299,6 +310,45 @@ struct panfrost_dump_registers { __u32 value; }; +enum drm_panfrost_jm_ctx_priority { + /** + * @PANFROST_JM_CTX_PRIORITY_LOW: Low priority context. + */ + PANFROST_JM_CTX_PRIORITY_LOW = 0, + + /** + * @PANFROST_JM_CTX_PRIORITY_MEDIUM: Medium priority context. + */ + PANFROST_JM_CTX_PRIORITY_MEDIUM, + + /** + * @PANFROST_JM_CTX_PRIORITY_HIGH: High priority context. + * + * Requires CAP_SYS_NICE or DRM_MASTER. + */ + PANFROST_JM_CTX_PRIORITY_HIGH, +}; + +struct drm_panfrost_jm_ctx_create { + /** @handle: Handle of the created JM context */ + __u32 handle; + + /** @priority: Context priority (see enum drm_panfrost_jm_ctx_priority). */ + __u32 priority; +}; + +struct drm_panfrost_jm_ctx_destroy { + /** + * @handle: Handle of the JM context to destroy. + * + * Must be a valid context handle returned by DRM_IOCTL_PANTHOR_JM_CTX_CREATE. + */ + __u32 handle; + + /** @pad: Padding field, MBZ. */ + __u32 pad; +}; + #if defined(__cplusplus) } #endif |