diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdkfd/kfd_topology.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 53 |
1 files changed, 40 insertions, 13 deletions
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c index e3843c5929ed..09da91644f9f 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c @@ -36,6 +36,7 @@ #include "kfd_topology.h" #include "kfd_device_queue_manager.h" #include "kfd_iommu.h" +#include "amdgpu_amdkfd.h" /* topology_device_list - Master list of all topology devices */ static struct list_head topology_device_list; @@ -100,7 +101,25 @@ struct kfd_dev *kfd_device_by_pci_dev(const struct pci_dev *pdev) down_read(&topology_lock); list_for_each_entry(top_dev, &topology_device_list, list) - if (top_dev->gpu->pdev == pdev) { + if (top_dev->gpu && top_dev->gpu->pdev == pdev) { + device = top_dev->gpu; + break; + } + + up_read(&topology_lock); + + return device; +} + +struct kfd_dev *kfd_device_by_kgd(const struct kgd_dev *kgd) +{ + struct kfd_topology_device *top_dev; + struct kfd_dev *device = NULL; + + down_read(&topology_lock); + + list_for_each_entry(top_dev, &topology_device_list, list) + if (top_dev->gpu && top_dev->gpu->kgd == kgd) { device = top_dev->gpu; break; } @@ -1052,7 +1071,7 @@ static uint32_t kfd_generate_gpu_id(struct kfd_dev *gpu) if (!gpu) return 0; - gpu->kfd2kgd->get_local_mem_info(gpu->kgd, &local_mem_info); + amdgpu_amdkfd_get_local_mem_info(gpu->kgd, &local_mem_info); local_mem_size = local_mem_info.local_mem_size_private + local_mem_info.local_mem_size_public; @@ -1074,8 +1093,6 @@ static uint32_t kfd_generate_gpu_id(struct kfd_dev *gpu) * the GPU device is not already present in the topology device * list then return NULL. This means a new topology device has to * be created for this GPU. - * TODO: Rather than assiging @gpu to first topology device withtout - * gpu attached, it will better to have more stringent check. */ static struct kfd_topology_device *kfd_assign_gpu(struct kfd_dev *gpu) { @@ -1083,12 +1100,20 @@ static struct kfd_topology_device *kfd_assign_gpu(struct kfd_dev *gpu) struct kfd_topology_device *out_dev = NULL; down_write(&topology_lock); - list_for_each_entry(dev, &topology_device_list, list) + list_for_each_entry(dev, &topology_device_list, list) { + /* Discrete GPUs need their own topology device list + * entries. Don't assign them to CPU/APU nodes. + */ + if (!gpu->device_info->needs_iommu_device && + dev->node_props.cpu_cores_count) + continue; + if (!dev->gpu && (dev->node_props.simd_count > 0)) { dev->gpu = gpu; out_dev = dev; break; } + } up_write(&topology_lock); return out_dev; } @@ -1118,8 +1143,7 @@ static void kfd_fill_mem_clk_max_info(struct kfd_topology_device *dev) * for APUs - If CRAT from ACPI reports more than one bank, then * all the banks will report the same mem_clk_max information */ - dev->gpu->kfd2kgd->get_local_mem_info(dev->gpu->kgd, - &local_mem_info); + amdgpu_amdkfd_get_local_mem_info(dev->gpu->kgd, &local_mem_info); list_for_each_entry(mem, &dev->mem_props, list) mem->mem_clk_max = local_mem_info.mem_clk_max; @@ -1240,7 +1264,7 @@ int kfd_topology_add_device(struct kfd_dev *gpu) * needed for the topology */ - dev->gpu->kfd2kgd->get_cu_info(dev->gpu->kgd, &cu_info); + amdgpu_amdkfd_get_cu_info(dev->gpu->kgd, &cu_info); dev->node_props.simd_arrays_per_engine = cu_info.num_shader_arrays_per_engine; @@ -1249,7 +1273,7 @@ int kfd_topology_add_device(struct kfd_dev *gpu) dev->node_props.location_id = PCI_DEVID(gpu->pdev->bus->number, gpu->pdev->devfn); dev->node_props.max_engine_clk_fcompute = - dev->gpu->kfd2kgd->get_max_engine_clock_in_mhz(dev->gpu->kgd); + amdgpu_amdkfd_get_max_engine_clock_in_mhz(dev->gpu->kgd); dev->node_props.max_engine_clk_ccompute = cpufreq_quick_get_max(0) / 1000; dev->node_props.drm_render_minor = @@ -1272,12 +1296,14 @@ int kfd_topology_add_device(struct kfd_dev *gpu) case CHIP_FIJI: case CHIP_POLARIS10: case CHIP_POLARIS11: + case CHIP_POLARIS12: pr_debug("Adding doorbell packet type capability\n"); dev->node_props.capability |= ((HSA_CAP_DOORBELL_TYPE_1_0 << HSA_CAP_DOORBELL_TYPE_TOTALBITS_SHIFT) & HSA_CAP_DOORBELL_TYPE_TOTALBITS_MASK); break; case CHIP_VEGA10: + case CHIP_VEGA12: case CHIP_VEGA20: case CHIP_RAVEN: dev->node_props.capability |= ((HSA_CAP_DOORBELL_TYPE_2_0 << @@ -1372,7 +1398,6 @@ int kfd_topology_enum_kfd_devices(uint8_t idx, struct kfd_dev **kdev) static int kfd_cpumask_to_apic_id(const struct cpumask *cpumask) { - const struct cpuinfo_x86 *cpuinfo; int first_cpu_of_numa_node; if (!cpumask || cpumask == cpu_none_mask) @@ -1380,9 +1405,11 @@ static int kfd_cpumask_to_apic_id(const struct cpumask *cpumask) first_cpu_of_numa_node = cpumask_first(cpumask); if (first_cpu_of_numa_node >= nr_cpu_ids) return -1; - cpuinfo = &cpu_data(first_cpu_of_numa_node); - - return cpuinfo->apicid; +#ifdef CONFIG_X86_64 + return cpu_data(first_cpu_of_numa_node).apicid; +#else + return first_cpu_of_numa_node; +#endif } /* kfd_numa_node_to_apic_id - Returns the APIC ID of the first logical processor |