diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdkfd')
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_chardev.c | 27 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_device.c | 11 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 7 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c | 27 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c | 4 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_events.c | 13 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_migrate.c | 3 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c | 15 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 2 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_process.c | 18 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 10 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 2 |
12 files changed, 87 insertions, 52 deletions
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index 6d291aa6386b..a0e30f21e12e 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -1065,6 +1065,20 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep, mutex_unlock(&p->svms.lock); return -EADDRINUSE; } + + /* When register user buffer check if it has been registered by svm by + * buffer cpu virtual address. + */ + if ((flags & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR) && + interval_tree_iter_first(&p->svms.objects, + args->mmap_offset >> PAGE_SHIFT, + (args->mmap_offset + args->size - 1) >> PAGE_SHIFT)) { + pr_err("User Buffer Address: 0x%llx already allocated by SVM\n", + args->mmap_offset); + mutex_unlock(&p->svms.lock); + return -EADDRINUSE; + } + mutex_unlock(&p->svms.lock); #endif mutex_lock(&p->mutex); @@ -1127,8 +1141,13 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep, } /* Update the VRAM usage count */ - if (flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) - WRITE_ONCE(pdd->vram_usage, pdd->vram_usage + args->size); + if (flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) { + uint64_t size = args->size; + + if (flags & KFD_IOC_ALLOC_MEM_FLAGS_AQL_QUEUE_MEM) + size >>= 1; + WRITE_ONCE(pdd->vram_usage, pdd->vram_usage + PAGE_ALIGN(size)); + } mutex_unlock(&p->mutex); @@ -2879,8 +2898,8 @@ static int kfd_mmio_mmap(struct kfd_dev *dev, struct kfd_process *process, address = dev->adev->rmmio_remap.bus_addr; - vma->vm_flags |= VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE | - VM_DONTDUMP | VM_PFNMAP; + vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE | + VM_DONTDUMP | VM_PFNMAP); vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c index b8936340742b..3de7f616a001 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c @@ -262,23 +262,12 @@ struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf) f2g = &gfx_v8_kfd2kgd; break; case CHIP_FIJI: - gfx_target_version = 80003; - f2g = &gfx_v8_kfd2kgd; - break; case CHIP_POLARIS10: gfx_target_version = 80003; f2g = &gfx_v8_kfd2kgd; break; case CHIP_POLARIS11: - gfx_target_version = 80003; - if (!vf) - f2g = &gfx_v8_kfd2kgd; - break; case CHIP_POLARIS12: - gfx_target_version = 80003; - if (!vf) - f2g = &gfx_v8_kfd2kgd; - break; case CHIP_VEGAM: gfx_target_version = 80003; if (!vf) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c index ecb4c3abc629..7a95698d83f7 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c @@ -200,7 +200,7 @@ static int add_queue_mes(struct device_queue_manager *dqm, struct queue *q, queue_input.wptr_addr = (uint64_t)q->properties.write_ptr; if (q->wptr_bo) { - wptr_addr_off = (uint64_t)q->properties.write_ptr - (uint64_t)q->wptr_bo->kfd_bo->va; + wptr_addr_off = (uint64_t)q->properties.write_ptr & (PAGE_SIZE - 1); queue_input.wptr_mc_addr = ((uint64_t)q->wptr_bo->tbo.resource->start << PAGE_SHIFT) + wptr_addr_off; } @@ -2373,7 +2373,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev) if (init_mqd_managers(dqm)) goto out_free; - if (allocate_hiq_sdma_mqd(dqm)) { + if (!dev->shared_resources.enable_mes && allocate_hiq_sdma_mqd(dqm)) { pr_err("Failed to allocate hiq sdma mqd trunk buffer\n"); goto out_free; } @@ -2397,7 +2397,8 @@ static void deallocate_hiq_sdma_mqd(struct kfd_dev *dev, void device_queue_manager_uninit(struct device_queue_manager *dqm) { dqm->ops.uninitialize(dqm); - deallocate_hiq_sdma_mqd(dqm->dev, &dqm->hiq_sdma_mqd); + if (!dqm->dev->shared_resources.enable_mes) + deallocate_hiq_sdma_mqd(dqm->dev, &dqm->hiq_sdma_mqd); kfree(dqm); } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c index d119070956fb..8b2dd2670ab7 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c @@ -59,30 +59,27 @@ static int update_qpd_v9(struct device_queue_manager *dqm, /* check if sh_mem_config register already configured */ if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - SH_MEM_ALIGNMENT_MODE_UNALIGNED << + qpd->sh_mem_config = SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT; - if (KFD_GC_VERSION(dqm->dev) == IP_VERSION(9, 4, 2)) { - /* Aldebaran can safely support different XNACK modes - * per process - */ - if (!pdd->process->xnack_enabled) - qpd->sh_mem_config |= - 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT; - } else if (dqm->dev->noretry && - !dqm->dev->use_iommu_v2) { - qpd->sh_mem_config |= - 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT; - } + if (dqm->dev->noretry && !dqm->dev->use_iommu_v2) + qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT; qpd->sh_mem_ape1_limit = 0; qpd->sh_mem_ape1_base = 0; } + if (KFD_SUPPORT_XNACK_PER_PROCESS(dqm->dev)) { + if (!pdd->process->xnack_enabled) + qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT; + else + qpd->sh_mem_config &= ~(1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT); + } + qpd->sh_mem_bases = compute_sh_mem_bases_64bit(pdd); - pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases); + pr_debug("sh_mem_bases 0x%X sh_mem_config 0x%X\n", qpd->sh_mem_bases, + qpd->sh_mem_config); return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c b/drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c index cd4e61bf0493..cbef2e147da5 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c @@ -159,8 +159,8 @@ int kfd_doorbell_mmap(struct kfd_dev *dev, struct kfd_process *process, address = kfd_get_process_doorbells(pdd); if (!address) return -ENOMEM; - vma->vm_flags |= VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE | - VM_DONTDUMP | VM_PFNMAP; + vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE | + VM_DONTDUMP | VM_PFNMAP); vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c index 729d26d648af..c894cf8f7c50 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c @@ -778,16 +778,13 @@ static struct kfd_event_waiter *alloc_event_waiters(uint32_t num_events) struct kfd_event_waiter *event_waiters; uint32_t i; - event_waiters = kmalloc_array(num_events, - sizeof(struct kfd_event_waiter), - GFP_KERNEL); + event_waiters = kcalloc(num_events, sizeof(struct kfd_event_waiter), + GFP_KERNEL); if (!event_waiters) return NULL; - for (i = 0; (event_waiters) && (i < num_events) ; i++) { + for (i = 0; i < num_events; i++) init_wait(&event_waiters[i].wait); - event_waiters[i].activated = false; - } return event_waiters; } @@ -1052,8 +1049,8 @@ int kfd_event_mmap(struct kfd_process *p, struct vm_area_struct *vma) pfn = __pa(page->kernel_address); pfn >>= PAGE_SHIFT; - vma->vm_flags |= VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE - | VM_DONTDUMP | VM_PFNMAP; + vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE + | VM_DONTDUMP | VM_PFNMAP); pr_debug("Mapping signal page\n"); pr_debug(" start user address == 0x%08lx\n", vma->vm_start); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c index 10048ce16aea..de8ce72344fc 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c @@ -1027,8 +1027,7 @@ int svm_migrate_init(struct amdgpu_device *adev) /* Disable SVM support capability */ pgmap->type = 0; if (pgmap->type == MEMORY_DEVICE_PRIVATE) - devm_release_mem_region(adev->dev, res->start, - res->end - res->start + 1); + devm_release_mem_region(adev->dev, res->start, resource_size(res)); return PTR_ERR(r); } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c index 4f6390f3236e..4a9af800b1f1 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c @@ -308,11 +308,16 @@ static void init_mqd_sdma(struct mqd_manager *mm, void **mqd, struct queue_properties *q) { struct v11_sdma_mqd *m; + int size; m = (struct v11_sdma_mqd *) mqd_mem_obj->cpu_ptr; - memset(m, 0, sizeof(struct v11_sdma_mqd)); + if (mm->dev->shared_resources.enable_mes) + size = PAGE_SIZE; + else + size = sizeof(struct v11_sdma_mqd); + memset(m, 0, size); *mqd = m; if (gart_addr) *gart_addr = mqd_mem_obj->gpu_addr; @@ -443,6 +448,14 @@ struct mqd_manager *mqd_manager_init_v11(enum KFD_MQD_TYPE type, #if defined(CONFIG_DEBUG_FS) mqd->debugfs_show_mqd = debugfs_show_mqd_sdma; #endif + /* + * To allocate SDMA MQDs by generic functions + * when MES is enabled. + */ + if (dev->shared_resources.enable_mes) { + mqd->allocate_mqd = allocate_mqd; + mqd->free_mqd = kfd_free_mqd_cp; + } pr_debug("%s@%i\n", __func__, __LINE__); break; default: diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index 552c3ac85a13..bfa30d12406b 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -206,6 +206,8 @@ enum cache_policy { #define KFD_GC_VERSION(dev) ((dev)->adev->ip_versions[GC_HWIP][0]) #define KFD_IS_SOC15(dev) ((KFD_GC_VERSION(dev)) >= (IP_VERSION(9, 0, 1))) +#define KFD_SUPPORT_XNACK_PER_PROCESS(dev)\ + (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 2)) struct kfd_event_interrupt_class { bool (*interrupt_isr)(struct kfd_dev *dev, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index 51b1683ac5c1..7acd55a814b2 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -1330,7 +1330,7 @@ bool kfd_process_xnack_mode(struct kfd_process *p, bool supported) * per-process XNACK mode selection. But let the dev->noretry * setting still influence the default XNACK mode. */ - if (supported && KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 2)) + if (supported && KFD_SUPPORT_XNACK_PER_PROCESS(dev)) continue; /* GFXv10 and later GPUs do not support shader preemption @@ -1563,6 +1563,8 @@ err_free_pdd: int kfd_process_device_init_vm(struct kfd_process_device *pdd, struct file *drm_file) { + struct amdgpu_fpriv *drv_priv; + struct amdgpu_vm *avm; struct kfd_process *p; struct kfd_dev *dev; int ret; @@ -1573,10 +1575,15 @@ int kfd_process_device_init_vm(struct kfd_process_device *pdd, if (pdd->drm_priv) return -EBUSY; + ret = amdgpu_file_to_fpriv(drm_file, &drv_priv); + if (ret) + return ret; + avm = &drv_priv->vm; + p = pdd->process; dev = pdd->dev; - ret = amdgpu_amdkfd_gpuvm_acquire_process_vm(dev->adev, drm_file, + ret = amdgpu_amdkfd_gpuvm_acquire_process_vm(dev->adev, avm, &p->kgd_process_info, &p->ef); if (ret) { @@ -1593,7 +1600,7 @@ int kfd_process_device_init_vm(struct kfd_process_device *pdd, if (ret) goto err_init_cwsr; - ret = amdgpu_amdkfd_gpuvm_set_vm_pasid(dev->adev, drm_file, p->pasid); + ret = amdgpu_amdkfd_gpuvm_set_vm_pasid(dev->adev, avm, p->pasid); if (ret) goto err_set_pasid; @@ -1607,6 +1614,7 @@ err_init_cwsr: kfd_process_device_destroy_ib_mem(pdd); err_reserve_ib_mem: pdd->drm_priv = NULL; + amdgpu_amdkfd_gpuvm_destroy_cb(dev->adev, avm); return ret; } @@ -1978,8 +1986,8 @@ int kfd_reserved_mem_mmap(struct kfd_dev *dev, struct kfd_process *process, return -ENOMEM; } - vma->vm_flags |= VM_IO | VM_DONTCOPY | VM_DONTEXPAND - | VM_NORESERVE | VM_DONTDUMP | VM_PFNMAP; + vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND + | VM_NORESERVE | VM_DONTDUMP | VM_PFNMAP); /* Mapping pages to user process */ return remap_pfn_range(vma, vma->vm_start, PFN_DOWN(__pa(qpd->cwsr_kaddr)), diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 814f99888ab1..dc6fd6967050 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -23,6 +23,7 @@ #include <linux/types.h> #include <linux/sched/task.h> +#include <drm/ttm/ttm_tt.h> #include "amdgpu_sync.h" #include "amdgpu_object.h" #include "amdgpu_vm.h" @@ -570,6 +571,15 @@ svm_range_vram_node_new(struct amdgpu_device *adev, struct svm_range *prange, goto reserve_bo_failed; } + if (clear) { + r = amdgpu_bo_sync_wait(bo, AMDGPU_FENCE_OWNER_KFD, false); + if (r) { + pr_debug("failed %d to sync bo\n", r); + amdgpu_bo_unreserve(bo); + goto reserve_bo_failed; + } + } + r = dma_resv_reserve_fences(bo->tbo.base.resv, 1); if (r) { pr_debug("failed %d to reserve bo\n", r); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c index bceb1a5b2518..3fdaba56be6f 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c @@ -801,7 +801,7 @@ static int kfd_build_sysfs_node_entry(struct kfd_topology_device *dev, p2plink->attr.name = "properties"; p2plink->attr.mode = KFD_SYSFS_FILE_MODE; - sysfs_attr_init(&iolink->attr); + sysfs_attr_init(&p2plink->attr); ret = sysfs_create_file(p2plink->kobj, &p2plink->attr); if (ret < 0) return ret; |