path: root/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c

/*
 * Copyright 2014 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.
 *
 * Authors: Alex Deucher
 */
#include <linux/firmware.h>
#include <drm/drmP.h>
#include "amdgpu.h"
#include "amdgpu_ucode.h"
#include "amdgpu_trace.h"
#include "vi.h"
#include "vid.h"

#include "oss/oss_3_0_d.h"
#include "oss/oss_3_0_sh_mask.h"

#include "gmc/gmc_8_1_d.h"
#include "gmc/gmc_8_1_sh_mask.h"

#include "gca/gfx_8_0_d.h"
#include "gca/gfx_8_0_enum.h"
#include "gca/gfx_8_0_sh_mask.h"

#include "bif/bif_5_0_d.h"
#include "bif/bif_5_0_sh_mask.h"

#include "tonga_sdma_pkt_open.h"

static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev);
static void sdma_v3_0_set_buffer_funcs(struct amdgpu_device *adev);
static void sdma_v3_0_set_vm_pte_funcs(struct amdgpu_device *adev);
static void sdma_v3_0_set_irq_funcs(struct amdgpu_device *adev);

MODULE_FIRMWARE("radeon/tonga_sdma.bin");
MODULE_FIRMWARE("radeon/tonga_sdma1.bin");
MODULE_FIRMWARE("radeon/carrizo_sdma.bin");
MODULE_FIRMWARE("radeon/carrizo_sdma1.bin");

static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
{
	SDMA0_REGISTER_OFFSET,
	SDMA1_REGISTER_OFFSET
};

static const u32 golden_settings_tonga_a11[] =
{
	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
	mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100,
	mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
	mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
	mmSDMA1_GFX_IB_CNTL, 0x800f0111, 0x00000100,
	mmSDMA1_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
	mmSDMA1_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
};

static const u32 tonga_mgcg_cgcg_init[] =
{
	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
};

static const u32 cz_golden_settings_a11[] =
{
	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
	mmSDMA0_GFX_IB_CNTL, 0x00000100, 0x00000100,
	mmSDMA0_POWER_CNTL, 0x00000800, 0x0003c800,
	mmSDMA0_RLC0_IB_CNTL, 0x00000100, 0x00000100,
	mmSDMA0_RLC1_IB_CNTL, 0x00000100, 0x00000100,
	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
	mmSDMA1_GFX_IB_CNTL, 0x00000100, 0x00000100,
	mmSDMA1_POWER_CNTL, 0x00000800, 0x0003c800,
	mmSDMA1_RLC0_IB_CNTL, 0x00000100, 0x00000100,
	mmSDMA1_RLC1_IB_CNTL, 0x00000100, 0x00000100,
};

static const u32 cz_mgcg_cgcg_init[] =
{
	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
};

/*
 * sDMA - System DMA
 * Starting with CIK, the GPU has new asynchronous
 * DMA engines.  These engines are used for compute
 * and gfx.  There are two DMA engines (SDMA0, SDMA1)
 * and each one supports 1 ring buffer used for gfx
 * and 2 queues used for compute.
 *
 * The programming model is very similar to the CP
 * (ring buffer, IBs, etc.), but sDMA has it's own
 * packet format that is different from the PM4 format
 * used by the CP. sDMA supports copying data, writing
 * embedded data, solid fills, and a number of other
 * things.  It also has support for tiling/detiling of
 * buffers.
 */

static void sdma_v3_0_init_golden_registers(struct amdgpu_device *adev)
{
	switch (adev->asic_type) {
	case CHIP_TONGA:
		amdgpu_program_register_sequence(adev,
						 tonga_mgcg_cgcg_init,
						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
		amdgpu_program_register_sequence(adev,
						 golden_settings_tonga_a11,
						 (const u32)ARRAY_SIZE(golden_settings_tonga_a11));
		break;
	case CHIP_CARRIZO:
		amdgpu_program_register_sequence(adev,
						 cz_mgcg_cgcg_init,
						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
		amdgpu_program_register_sequence(adev,
						 cz_golden_settings_a11,
						 (const u32)ARRAY_SIZE(cz_golden_settings_a11));
		break;
	default:
		break;
	}
}

/**
 * sdma_v3_0_init_microcode - load ucode images from disk
 *
 * @adev: amdgpu_device pointer
 *
 * Use the firmware interface to load the ucode images into
 * the driver (not loaded into hw).
 * Returns 0 on success, error on failure.
 */
static int sdma_v3_0_init_microcode(struct amdgpu_device *adev)
{
	const char *chip_name;
	char fw_name[30];
	int err, i;
	struct amdgpu_firmware_info *info = NULL;
	const struct common_firmware_header *header = NULL;

	DRM_DEBUG("\n");

	switch (adev->asic_type) {
	case CHIP_TONGA:
		chip_name = "tonga";
		break;
	case CHIP_CARRIZO:
		chip_name = "carrizo";
		break;
	default: BUG();
	}

	for (i = 0; i < SDMA_MAX_INSTANCE; i++) {
		if (i == 0)
			snprintf(fw_name, sizeof(fw_name), "radeon/%s_sdma.bin", chip_name);
		else
			snprintf(fw_name, sizeof(fw_name), "radeon/%s_sdma1.bin", chip_name);
		err = request_firmware(&adev->sdma[i].fw, fw_name, adev->dev);
		if (err)
			goto out;
		err = amdgpu_ucode_validate(adev->sdma[i].fw);
		if (err)
			goto out;

		if (adev->firmware.smu_load) {
			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
			info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
			info->fw = adev->sdma[i].fw;
			header = (const struct common_firmware_header *)info->fw->data;
			adev->firmware.fw_size +=
				ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
		}
	}
out:
	if (err) {
		printk(KERN_ERR
		       "sdma_v3_0: Failed to load firmware \"%s\"\n",
		       fw_name);
		for (i = 0; i < SDMA_MAX_INSTANCE; i++) {
			release_firmware(adev->sdma[i].fw);
			adev->sdma[i].fw = NULL;
		}
	}
	return err;
}

/**
 * sdma_v3_0_ring_get_rptr - get the current read pointer
 *
 * @ring: amdgpu ring pointer
 *
 * Get the current rptr from the hardware (VI+).
 */
static uint32_t sdma_v3_0_ring_get_rptr(struct amdgpu_ring *ring)
{
	u32 rptr;

	/* XXX check if swapping is necessary on BE */
	rptr = ring->adev->wb.wb[ring->rptr_offs] >> 2;

	return rptr;
}

/**
 * sdma_v3_0_ring_get_wptr - get the current write pointer
 *
 * @ring: amdgpu ring pointer
 *
 * Get the current wptr from the hardware (VI+).
 */
static uint32_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)
{
	struct amdgpu_device *adev = ring->adev;
	u32 wptr;

	if (ring->use_doorbell) {
		/* XXX check if swapping is necessary on BE */
		wptr = ring->adev->wb.wb[ring->wptr_offs] >> 2;
	} else {
		int me = (ring == &ring->adev->sdma[0].ring) ? 0 : 1;

		wptr = RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >> 2;
	}

	return wptr;
}

/**
 * sdma_v3_0_ring_set_wptr - commit the write pointer
 *
 * @ring: amdgpu ring pointer
 *
 * Write the wptr back to the hardware (VI+).
 */
static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *rin