Commit 5b87245f authored by Amber Lin's avatar Amber Lin Committed by Alex Deucher
Browse files

drm/amdkfd: Simplify kfd2kgd interface 



After amdkfd module is merged into amdgpu, KFD can call amdgpu directly
and no longer needs to use the function pointer. Replace those function
pointers with functions if they are not ASIC dependent.

Signed-off-by: default avatarAmber Lin <Amber.Lin@amd.com>
Acked-by: default avatarChristian König <christian.koenig@amd.com>
Reviewed-by: default avatarAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: default avatarFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: default avatarAlex Deucher <alexander.deucher@amd.com>
parent 17288014

drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c

+2 −1
Original line number Diff line number Diff line
@@ -23,6 +23,7 @@ 
#include "kfd_priv.h"

#include "kfd_events.h"

#include "cik_int.h"

#include "amdgpu_amdkfd.h"



static bool cik_event_interrupt_isr(struct kfd_dev *dev,

					const uint32_t *ih_ring_entry,
@@ -107,7 +108,7 @@ static void cik_event_interrupt_wq(struct kfd_dev *dev, 
		kfd_process_vm_fault(dev->dqm, pasid);



		memset(&info, 0, sizeof(info));

		dev->kfd2kgd->get_vm_fault_info(dev->kgd, &info);

		amdgpu_amdkfd_gpuvm_get_vm_fault_info(dev->kgd, &info);

		if (!info.page_addr && !info.status)

			return;

drivers/gpu/drm/amd/amdkfd/kfd_chardev.c

+11 −10
Original line number Diff line number Diff line
@@ -37,6 +37,7 @@ 
#include "kfd_priv.h"

#include "kfd_device_queue_manager.h"

#include "kfd_dbgmgr.h"

#include "amdgpu_amdkfd.h"



static long kfd_ioctl(struct file *, unsigned int, unsigned long);

static int kfd_open(struct inode *, struct file *);
@@ -834,8 +835,7 @@ static int kfd_ioctl_get_clock_counters(struct file *filep, 
	dev = kfd_device_by_id(args->gpu_id);

	if (dev)

		/* Reading GPU clock counter from KGD */

		args->gpu_clock_counter =

			dev->kfd2kgd->get_gpu_clock_counter(dev->kgd);

		args->gpu_clock_counter = get_gpu_clock_counter(dev->kgd);

	else

		/* Node without GPU resource */

		args->gpu_clock_counter = 0;
@@ -1042,7 +1042,7 @@ static int kfd_ioctl_create_event(struct file *filp, struct kfd_process *p, 
		}

		mutex_unlock(&p->mutex);



		err = kfd->kfd2kgd->map_gtt_bo_to_kernel(kfd->kgd,

		err = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(kfd->kgd,

						mem, &kern_addr, &size);

		if (err) {

			pr_err("Failed to map event page to kernel\n");
@@ -1240,7 +1240,7 @@ bool kfd_dev_is_large_bar(struct kfd_dev *dev) 
	if (dev->device_info->needs_iommu_device)

		return false;



	dev->kfd2kgd->get_local_mem_info(dev->kgd, &mem_info);

	get_local_mem_info(dev->kgd, &mem_info);

	if (mem_info.local_mem_size_private == 0 &&

			mem_info.local_mem_size_public > 0)

		return true;
@@ -1281,7 +1281,7 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep, 
		goto err_unlock;

	}



	err = dev->kfd2kgd->alloc_memory_of_gpu(

	err = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(

		dev->kgd, args->va_addr, args->size,

		pdd->vm, (struct kgd_mem **) &mem, &offset,

		flags);
@@ -1303,7 +1303,7 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep, 
	return 0;



err_free:

	dev->kfd2kgd->free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem);

	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem);

err_unlock:

	mutex_unlock(&p->mutex);

	return err;
@@ -1338,7 +1338,8 @@ static int kfd_ioctl_free_memory_of_gpu(struct file *filep, 
		goto err_unlock;

	}



	ret = dev->kfd2kgd->free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem);

	ret = amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd,

						(struct kgd_mem *)mem);



	/* If freeing the buffer failed, leave the handle in place for

	 * clean-up during process tear-down.
@@ -1418,7 +1419,7 @@ static int kfd_ioctl_map_memory_to_gpu(struct file *filep, 
			err = PTR_ERR(peer_pdd);

			goto get_mem_obj_from_handle_failed;

		}

		err = peer->kfd2kgd->map_memory_to_gpu(

		err = amdgpu_amdkfd_gpuvm_map_memory_to_gpu(

			peer->kgd, (struct kgd_mem *)mem, peer_pdd->vm);

		if (err) {

			pr_err("Failed to map to gpu %d/%d\n",
@@ -1430,7 +1431,7 @@ static int kfd_ioctl_map_memory_to_gpu(struct file *filep, 


	mutex_unlock(&p->mutex);



	err = dev->kfd2kgd->sync_memory(dev->kgd, (struct kgd_mem *) mem, true);

	err = amdgpu_amdkfd_gpuvm_sync_memory(dev->kgd, (struct kgd_mem *) mem, true);

	if (err) {

		pr_debug("Sync memory failed, wait interrupted by user signal\n");

		goto sync_memory_failed;
@@ -1525,7 +1526,7 @@ static int kfd_ioctl_unmap_memory_from_gpu(struct file *filep, 
			err = -ENODEV;

			goto get_mem_obj_from_handle_failed;

		}

		err = dev->kfd2kgd->unmap_memory_to_gpu(

		err = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(

			peer->kgd, (struct kgd_mem *)mem, peer_pdd->vm);

		if (err) {

			pr_err("Failed to unmap from gpu %d/%d\n",

drivers/gpu/drm/amd/amdkfd/kfd_crat.c

+3 −2
Original line number Diff line number Diff line
@@ -26,6 +26,7 @@ 
#include "kfd_priv.h"

#include "kfd_topology.h"

#include "kfd_iommu.h"

#include "amdgpu_amdkfd.h"



/* GPU Processor ID base for dGPUs for which VCRAT needs to be created.

 * GPU processor ID are expressed with Bit[31]=1.
@@ -1159,7 +1160,7 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image, 
	cu->flags |= CRAT_CU_FLAGS_GPU_PRESENT;

	cu->proximity_domain = proximity_domain;



	kdev->kfd2kgd->get_cu_info(kdev->kgd, &cu_info);

	get_cu_info(kdev->kgd, &cu_info);

	cu->num_simd_per_cu = cu_info.simd_per_cu;

	cu->num_simd_cores = cu_info.simd_per_cu * cu_info.cu_active_number;

	cu->max_waves_simd = cu_info.max_waves_per_simd;
@@ -1190,7 +1191,7 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image, 
	 * report the total FB size (public+private) as a single

	 * private heap.

	 */

	kdev->kfd2kgd->get_local_mem_info(kdev->kgd, &local_mem_info);

	get_local_mem_info(kdev->kgd, &local_mem_info);

	sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr +

			sub_type_hdr->length);

drivers/gpu/drm/amd/amdkfd/kfd_device.c

+4 −3
Original line number Diff line number Diff line
@@ -28,6 +28,7 @@ 
#include "kfd_pm4_headers_vi.h"

#include "cwsr_trap_handler.h"

#include "kfd_iommu.h"

#include "amdgpu_amdkfd.h"



#define MQD_SIZE_ALIGNED 768
@@ -478,7 +479,7 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd, 
	/* add another 512KB for all other allocations on gart (HPD, fences) */

	size += 512 * 1024;



	if (kfd->kfd2kgd->init_gtt_mem_allocation(

	if (alloc_gtt_mem(

			kfd->kgd, size, &kfd->gtt_mem,

			&kfd->gtt_start_gpu_addr, &kfd->gtt_start_cpu_ptr,

			false)) {
@@ -552,7 +553,7 @@ kfd_topology_add_device_error: 
kfd_doorbell_error:

	kfd_gtt_sa_fini(kfd);

kfd_gtt_sa_init_error:

	kfd->kfd2kgd->free_gtt_mem(kfd->kgd, kfd->gtt_mem);

	free_gtt_mem(kfd->kgd, kfd->gtt_mem);

	dev_err(kfd_device,

		"device %x:%x NOT added due to errors\n",

		kfd->pdev->vendor, kfd->pdev->device);
@@ -569,7 +570,7 @@ void kgd2kfd_device_exit(struct kfd_dev *kfd) 
		kfd_topology_remove_device(kfd);

		kfd_doorbell_fini(kfd);

		kfd_gtt_sa_fini(kfd);

		kfd->kfd2kgd->free_gtt_mem(kfd->kgd, kfd->gtt_mem);

		free_gtt_mem(kfd->kgd, kfd->gtt_mem);

	}



	kfree(kfd);

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c

+8 −7
Original line number Diff line number Diff line
@@ -33,6 +33,7 @@ 
#include "kfd_mqd_manager.h"

#include "cik_regs.h"

#include "kfd_kernel_queue.h"

#include "amdgpu_amdkfd.h"



/* Size of the per-pipe EOP queue */

#define CIK_HPD_EOP_BYTES_LOG2 11
@@ -219,7 +220,7 @@ static int flush_texture_cache_nocpsch(struct kfd_dev *kdev, 
	if (ret)

		return ret;



	return kdev->kfd2kgd->submit_ib(kdev->kgd, KGD_ENGINE_MEC1, qpd->vmid,

	return amdgpu_amdkfd_submit_ib(kdev->kgd, KGD_ENGINE_MEC1, qpd->vmid,

				qpd->ib_base, (uint32_t *)qpd->ib_kaddr,

				pmf->release_mem_size / sizeof(uint32_t));

}
@@ -672,7 +673,7 @@ static int restore_process_queues_nocpsch(struct device_queue_manager *dqm, 


	pdd = qpd_to_pdd(qpd);

	/* Retrieve PD base */

	pd_base = dqm->dev->kfd2kgd->get_process_page_dir(pdd->vm);

	pd_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->vm);



	dqm_lock(dqm);

	if (WARN_ON_ONCE(!qpd->evicted)) /* already restored, do nothing */
@@ -743,7 +744,7 @@ static int restore_process_queues_cpsch(struct device_queue_manager *dqm, 


	pdd = qpd_to_pdd(qpd);

	/* Retrieve PD base */

	pd_base = dqm->dev->kfd2kgd->get_process_page_dir(pdd->vm);

	pd_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->vm);



	dqm_lock(dqm);

	if (WARN_ON_ONCE(!qpd->evicted)) /* already restored, do nothing */
@@ -793,7 +794,7 @@ static int register_process(struct device_queue_manager *dqm, 


	pdd = qpd_to_pdd(qpd);

	/* Retrieve PD base */

	pd_base = dqm->dev->kfd2kgd->get_process_page_dir(pdd->vm);

	pd_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->vm);



	dqm_lock(dqm);

	list_add(&n->list, &dqm->queues);
@@ -805,7 +806,7 @@ static int register_process(struct device_queue_manager *dqm, 
	retval = dqm->asic_ops.update_qpd(dqm, qpd);



	if (dqm->processes_count++ == 0)

		dqm->dev->kfd2kgd->set_compute_idle(dqm->dev->kgd, false);

		amdgpu_amdkfd_set_compute_idle(dqm->dev->kgd, false);



	dqm_unlock(dqm);
@@ -829,7 +830,7 @@ static int unregister_process(struct device_queue_manager *dqm, 
			list_del(&cur->list);

			kfree(cur);

			if (--dqm->processes_count == 0)

				dqm->dev->kfd2kgd->set_compute_idle(

				amdgpu_amdkfd_set_compute_idle(

					dqm->dev->kgd, true);

			goto out;

		}
@@ -1796,7 +1797,7 @@ static void kfd_process_hw_exception(struct work_struct *work) 
{

	struct device_queue_manager *dqm = container_of(work,

			struct device_queue_manager, hw_exception_work);

	dqm->dev->kfd2kgd->gpu_recover(dqm->dev->kgd);

	amdgpu_amdkfd_gpu_reset(dqm->dev->kgd);

}



#if defined(CONFIG_DEBUG_FS)

CRA Git | Maintained and supported by SUSTech CRA and CCSE