path: root/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c

/*
 * Copyright 2019 Advanced Micro Devices, Inc.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
 * OTHER DEALINGS IN THE SOFTWARE.
 *
 */

#include <linux/firmware.h>
#include <drm/drm_exec.h>

#include "amdgpu_mes.h"
#include "amdgpu.h"
#include "soc15_common.h"
#include "amdgpu_mes_ctx.h"

#define AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS 1024
#define AMDGPU_ONE_DOORBELL_SIZE 8

int amdgpu_mes_doorbell_process_slice(struct amdgpu_device *adev)
{
	return roundup(AMDGPU_ONE_DOORBELL_SIZE *
		       AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS,
		       PAGE_SIZE);
}

static int amdgpu_mes_kernel_doorbell_get(struct amdgpu_device *adev,
					 struct amdgpu_mes_process *process,
					 int ip_type, uint64_t *doorbell_index)
{
	unsigned int offset, found;
	struct amdgpu_mes *mes = &adev->mes;

	if (ip_type == AMDGPU_RING_TYPE_SDMA)
		offset = adev->doorbell_index.sdma_engine[0];
	else
		offset = 0;

	found = find_next_zero_bit(mes->doorbell_bitmap, mes->num_mes_dbs, offset);
	if (found >= mes->num_mes_dbs) {
		DRM_WARN("No doorbell available\n");
		return -ENOSPC;
	}

	set_bit(found, mes->doorbell_bitmap);

	/* Get the absolute doorbell index on BAR */
	*doorbell_index = mes->db_start_dw_offset + found * 2;
	return 0;
}

static void amdgpu_mes_kernel_doorbell_free(struct amdgpu_device *adev,
					   struct amdgpu_mes_process *process,
					   uint32_t doorbell_index)
{
	unsigned int old, rel_index;
	struct amdgpu_mes *mes = &adev->mes;

	/* Find the relative index of the doorbell in this object */
	rel_index = (doorbell_index - mes->db_start_dw_offset) / 2;
	old = test_and_clear_bit(rel_index, mes->doorbell_bitmap);
	WARN_ON(!old);
}

static int amdgpu_mes_doorbell_init(struct amdgpu_device *adev)
{
	int i;
	struct amdgpu_mes *mes = &adev->mes;

	/* Bitmap for dynamic allocation of kernel doorbells */
	mes->doorbell_bitmap = bitmap_zalloc(PAGE_SIZE / sizeof(u32), GFP_KERNEL);
	if (!mes->doorbell_bitmap) {
		DRM_ERROR("Failed to allocate MES doorbell bitmap\n");
		return -ENOMEM;
	}

	mes->num_mes_dbs = PAGE_SIZE / AMDGPU_ONE_DOORBELL_SIZE;
	for (i = 0; i < AMDGPU_MES_PRIORITY_NUM_LEVELS; i++) {
		adev->mes.aggregated_doorbells[i] = mes->db_start_dw_offset + i * 2;
		set_bit(i, mes->doorbell_bitmap);
	}

	return 0;
}

static void amdgpu_mes_doorbell_free(struct amdgpu_device *adev)
{
	bitmap_free(adev->mes.doorbell_bitmap);
}

int amdgpu_mes_init(struct amdgpu_device *adev)
{
	int i, r;

	adev->mes.adev = adev;

	idr_init(&adev->mes.pasid_idr);
	idr_init(&adev->mes.gang_id_idr);
	idr_init(&adev->mes.queue_id_idr);
	ida_init(&adev->mes.doorbell_ida);
	spin_lock_init(&adev->mes.queue_id_lock);
	spin_lock_init(&adev->mes.ring_lock);
	mutex_init(&adev->mes.mutex_hidden);

	adev->mes.total_max_queue = AMDGPU_FENCE_MES_QUEUE_ID_MASK;
	adev->mes.vmid_mask_mmhub = 0xffffff00;
	adev->mes.vmid_mask_gfxhub = 0xffffff00;

	for (i = 0; i < AMDGPU_MES_MAX_COMPUTE_PIPES; i++) {
		/* use only 1st MEC pipes */
		if (i >= 4)
			continue;
		adev->mes.compute_hqd_mask[i] = 0xc;
	}

	for (i = 0; i < AMDGPU_MES_MAX_GFX_PIPES; i++)
		adev->mes.gfx_hqd_mask[i] = i ? 0 : 0xfffffffe;

	for (i = 0; i < AMDGPU_MES_MAX_SDMA_PIPES; i++) {
		if (adev->ip_versions[SDMA0_HWIP][0] < IP_VERSION(6, 0, 0))
			adev->mes.sdma_hqd_mask[i] = i ? 0 : 0x3fc;
		/* zero sdma_hqd_mask for non-existent engine */
		else if (adev->sdma.num_instances == 1)
			adev->mes.sdma_hqd_mask[i] = i ? 0 : 0xfc;
		else
			adev->mes.sdma_hqd_mask[i] = 0xfc;
	}

	r = amdgpu_device_wb_get(adev, &adev->mes.sch_ctx_offs);
	if (r) {
		dev_err(adev->dev,
			"(%d) ring trail_fence_offs wb alloc failed\n", r);
		goto error_ids;
	}
	adev->mes.sch_ctx_gpu_addr =
		adev->wb.gpu_addr + (adev->mes.sch_ctx_offs * 4);
	adev->mes.sch_ctx_ptr =
		(uint64_t *)&adev->wb.wb[adev->mes.sch_ctx_offs];

	r = amdgpu_device_wb_get(adev, &adev->mes.query_status_fence_offs);
	if (r) {
		amdgpu_device_wb_free(adev, adev->mes.sch_ctx_offs);
		dev_err(adev->dev,
			"(%d) query_status_fence_offs wb alloc failed\n", r);
		goto error_ids;
	}
	adev->mes.query_status_fence_gpu_addr =
		adev->wb.gpu_addr + (adev->mes.query_status_fence_offs * 4);
	adev->mes.query_status_fence_ptr =
		(uint64_t *)&adev->wb.wb[adev->mes.query_status_fence_offs];

	r = amdgpu_device_wb_get(adev, &adev->mes.read_val_offs);
	if (r) {
		amdgpu_device_wb_free(adev, adev->mes.sch_ctx_offs);
		amdgpu_device_wb_free(adev, adev->mes.query_status_fence_offs);
		dev_err(adev->dev,
			"(%d) read_val_offs alloc failed\n", r);
		goto error_ids;
	}
	adev->mes.read_val_gpu_addr =
		adev->wb.gpu_addr + (adev->mes.read_val_offs * 4);
	adev->mes.read_val_ptr =
		(uint32_t *)&adev->wb.wb[adev->mes.read_val_offs];

	r = amdgpu_mes_doorbell_init(adev);
	if (r)
		goto error;

	return 0;

error:
	amdgpu_device_wb_free(adev, adev->mes.sch_ctx_offs);
	amdgpu_device_wb_free(adev, adev->mes.query_status_fence_offs);
	amdgpu_device_wb_free(adev, adev->mes.read_val_offs);
error_ids:
	idr_destroy(&adev->mes.pasid_idr);
	idr_destroy(&adev->mes.gang_id_idr);
	idr_destroy(&adev->mes.queue_id_idr);
	ida_destroy(&adev->mes.doorbell_ida);
	mutex_destroy(&adev->mes.mutex_hidden);
	return r;
}

void amdgpu_mes_fini(struct amdgpu_device *adev)
{
	amdgpu_device_wb_free(adev, adev->mes.sch_ctx_offs);
	amdgpu_device_wb_free(adev, adev->mes.query_status_fence_offs);
	amdgpu_device_wb_free(adev, adev->mes.read_val_offs);
	amdgpu_mes_doorbell_free(adev);

	idr_destroy(&adev->mes.pasid_idr);
	idr_destroy(&adev->mes.gang_id_idr);
	idr_destroy(&adev->mes.queue_id_idr);
	ida_destroy(&adev->mes.doorbell_ida);
	mutex_destroy(&adev->mes.mutex_hidden);
}

static void amdgpu_mes_queue_free_mqd(struct amdgpu_mes_queue *q)
{
	amdgpu_bo_free_kernel(&q->mqd_obj,
			      &q->mqd_gpu_addr,
			      &q->mqd_cpu_ptr);
}

int amdgpu_mes_create_process(struct amdgpu_device *adev, int pasid,
			      struct amdgpu_vm *vm)
{
	struct amdgpu_mes_process *process;
	int r;

	/* allocate the mes process buffer */
	process = kzalloc(sizeof(struct amdgpu_mes_process), GFP_KERNEL);
	if (!process) {
		DRM_ERROR("no more memory to create mes process\n");
		return -ENOMEM;
	}

	/* allocate the process context bo and map it */
	r = amdgpu_bo_create_kernel(adev, AMDGPU_MES_PROC_CTX_SIZE, PAGE_SIZE,
				    AMDGPU_GEM_DOMAIN_GTT,
				    &process->proc_ctx_bo,
				    &process->proc_ctx_gpu_addr,
				    &process->proc_ctx_cpu_ptr);
	if (r) {
		DRM_ERROR("failed to allocate process context bo\n");
		goto clean_up_memory;
	}
	memset(process->proc_ctx_cpu_ptr, 0, AMDGPU_MES_PROC_CTX_SIZE);

	/*
	 * Avoid taking any other locks under MES lock to avoid circular
	 * lock dependencies.
	 */
	amdgpu_mes_lock(&adev->mes);

	/* add the mes process to idr list */
	r = idr_alloc(&adev->mes.pasid_idr, process, pasid, pasid + 1,
		      GFP_KERNEL);
	if (r <