blob: 313c09b48e63321b90bb4be8eafc9011ae5b213e [file]
/*
* Copyright 2025 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 "amdgpu.h"
#include "amdgpu_imu.h"
#include "amdgpu_dpm.h"
#include "imu_v12_1.h"
#include "gc/gc_12_1_0_offset.h"
#include "gc/gc_12_1_0_sh_mask.h"
#include "mmhub/mmhub_4_2_0_offset.h"
#define TRANSFER_RAM_MASK 0x001c0000
static int imu_v12_1_init_microcode(struct amdgpu_device *adev)
{
char ucode_prefix[15];
int err;
const struct imu_firmware_header_v1_0 *imu_hdr;
struct amdgpu_firmware_info *info = NULL;
DRM_DEBUG("\n");
amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
"amdgpu/%s_imu.bin", ucode_prefix);
if (err)
goto out;
imu_hdr = (const struct imu_firmware_header_v1_0 *)adev->gfx.imu_fw->data;
adev->gfx.imu_fw_version = le32_to_cpu(imu_hdr->header.ucode_version);
if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
info = &adev->firmware.ucode[AMDGPU_UCODE_ID_IMU_I];
info->ucode_id = AMDGPU_UCODE_ID_IMU_I;
info->fw = adev->gfx.imu_fw;
adev->firmware.fw_size +=
ALIGN(le32_to_cpu(imu_hdr->imu_iram_ucode_size_bytes), PAGE_SIZE);
info = &adev->firmware.ucode[AMDGPU_UCODE_ID_IMU_D];
info->ucode_id = AMDGPU_UCODE_ID_IMU_D;
info->fw = adev->gfx.imu_fw;
adev->firmware.fw_size +=
ALIGN(le32_to_cpu(imu_hdr->imu_dram_ucode_size_bytes), PAGE_SIZE);
}
out:
if (err) {
dev_err(adev->dev,
"gfx12: Failed to load firmware \"%s_imu.bin\"\n",
ucode_prefix);
amdgpu_ucode_release(&adev->gfx.imu_fw);
}
return err;
}
static void imu_v12_1_xcc_load_microcode(struct amdgpu_device *adev,
int xcc_id)
{
const struct imu_firmware_header_v1_0 *hdr;
const __le32 *fw_data;
unsigned i, fw_size;
hdr = (const struct imu_firmware_header_v1_0 *)adev->gfx.imu_fw->data;
fw_data = (const __le32 *)(adev->gfx.imu_fw->data +
le32_to_cpu(hdr->header.ucode_array_offset_bytes));
fw_size = le32_to_cpu(hdr->imu_iram_ucode_size_bytes) / 4;
WREG32_SOC15(GC, GET_INST(GC, xcc_id), regGFX_IMU_I_RAM_ADDR, 0);
for (i = 0; i < fw_size; i++)
WREG32_SOC15(GC, GET_INST(GC, xcc_id),
regGFX_IMU_I_RAM_DATA,
le32_to_cpup(fw_data++));
WREG32_SOC15(GC, GET_INST(GC, xcc_id),
regGFX_IMU_I_RAM_ADDR,
adev->gfx.imu_fw_version);
fw_data = (const __le32 *)(adev->gfx.imu_fw->data +
le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
le32_to_cpu(hdr->imu_iram_ucode_size_bytes));
fw_size = le32_to_cpu(hdr->imu_dram_ucode_size_bytes) / 4;
WREG32_SOC15(GC, GET_INST(GC, xcc_id), regGFX_IMU_D_RAM_ADDR, 0);
for (i = 0; i < fw_size; i++)
WREG32_SOC15(GC, GET_INST(GC, xcc_id),
regGFX_IMU_D_RAM_DATA,
le32_to_cpup(fw_data++));
WREG32_SOC15(GC, GET_INST(GC, xcc_id),
regGFX_IMU_D_RAM_ADDR,
adev->gfx.imu_fw_version);
}
static int imu_v12_1_load_microcode(struct amdgpu_device *adev)
{
int i, num_xcc;
if (!adev->gfx.imu_fw)
return -EINVAL;
num_xcc = NUM_XCC(adev->gfx.xcc_mask);
for (i = 0; i < num_xcc; i++) {
imu_v12_1_xcc_load_microcode(adev, i);
}
return 0;
}
#define regGFX_IMU_PARTITION_SWITCH 0x5f8c
#define regGFX_IMU_PARTITION_SWITCH_BASE_IDX 1
#define GFX_IMU_PARTITION_SWITCH__PARTITION_MODE__SHIFT 0x0
#define GFX_IMU_PARTITION_SWITCH__TOTAL_XCCS_IN_XCP__SHIFT 0x2
#define GFX_IMU_PARTITION_SWITCH__VIRTUAL_XCC_ID__SHIFT 0x9
#define GFX_IMU_PARTITION_SWITCH__PHYSICAL_XCC_MASK__SHIFT 0x10
#define GFX_IMU_PARTITION_SWITCH__PARTITION_MODE_MASK 0x00000003L
#define GFX_IMU_PARTITION_SWITCH__TOTAL_XCCS_IN_XCP_MASK 0x0000003CL
#define GFX_IMU_PARTITION_SWITCH__VIRTUAL_XCC_ID_MASK 0x00000E00L
#define GFX_IMU_PARTITION_SWITCH__PHYSICAL_XCC_MASK_MASK 0x00FF0000L
static int imu_v12_1_switch_compute_partition(struct amdgpu_device *adev,
int num_xccs_per_xcp,
int compute_partition_mode)
{
int ret, i, num_xcc;
u32 tmp = 0, physical_xcc_mask = 0;
if (adev->psp.funcs) {
ret = psp_spatial_partition(&adev->psp, compute_partition_mode);
if (ret)
return ret;
} else {
num_xcc = NUM_XCC(adev->gfx.xcc_mask);
/* 00 - SPX; 01 - DPX; 10 - QPX; 11 - CPX */
switch (compute_partition_mode) {
case AMDGPU_SPX_PARTITION_MODE:
compute_partition_mode = 0;
break;
case AMDGPU_DPX_PARTITION_MODE:
compute_partition_mode = 1;
break;
case AMDGPU_QPX_PARTITION_MODE:
compute_partition_mode = 2;
break;
case AMDGPU_CPX_PARTITION_MODE:
compute_partition_mode = 3;
break;
default:
dev_err(adev->dev, "Invalid compute partition mode\n");
return -EINVAL;
}
num_xcc = NUM_XCC(adev->gfx.xcc_mask);
for(i = 0; i < num_xcc; i++) {
if (i % num_xccs_per_xcp == 0) {
physical_xcc_mask =
adev->gfx.xcc_mask &
(GENMASK(GET_INST(GC, i + num_xccs_per_xcp - 1),
GET_INST(GC, i)));
}
tmp = REG_SET_FIELD(tmp, GFX_IMU_PARTITION_SWITCH,
PARTITION_MODE, compute_partition_mode);
tmp = REG_SET_FIELD(tmp, GFX_IMU_PARTITION_SWITCH,
TOTAL_XCCS_IN_XCP, num_xccs_per_xcp);
tmp = REG_SET_FIELD(tmp, GFX_IMU_PARTITION_SWITCH,
VIRTUAL_XCC_ID, i % num_xccs_per_xcp);
tmp = REG_SET_FIELD(tmp, GFX_IMU_PARTITION_SWITCH,
PHYSICAL_XCC_MASK, physical_xcc_mask);
WREG32_SOC15(GC, GET_INST(GC, i), regGFX_IMU_PARTITION_SWITCH, tmp);
}
ret = 0;
}
adev->gfx.num_xcc_per_xcp = num_xccs_per_xcp;
return 0;
}
#define regGFX_IMU_MCM_ADDR_LUT_PF 0x5f8d
#define regGFX_IMU_MCM_ADDR_LUT_PF_BASE_IDX 1
static void imu_v12_1_init_mcm_addr_lut(struct amdgpu_device *adev)
{
/* Encoding mcm_addr_lut in SPX mode for now
* I've noticed variations in encoding
* Need to revisit the configuration later
* XCD0:: Bits-[3:0 ] AID-01 XCD-00
* XCD1:: Bits-[7:4 ] AID-01 XCD-01
* XCD2:: Bits-[11:8 ] AID-01 XCD-10
* XCD3:: Bits-[15:12] AID-01 XCD-11
* XCD4:: Bits-[19:16] AID-10 XCD-00
* XCD5:: Bits-[23:20] AID-10 XCD-01
* XCD6:: Bits-[27:24] AID-10 XCD-10
* XCD7:: Bits-[31:28] AID-10 XCD-11
*/
int data, encode_data;
int i, j, num_xcc;
num_xcc = NUM_XCC(adev->gfx.xcc_mask);
for (i = 0; i < (num_xcc / adev->gfx.num_xcc_per_xcp); i++) {
encode_data = 0;
for (j = 0; j < adev->gfx.num_xcc_per_xcp; j++) {
data = ((GET_INST(GC, i * adev->gfx.num_xcc_per_xcp + j) / 4 + 1) << 2) |
(GET_INST(GC, i * adev->gfx.num_xcc_per_xcp + j) % 4);
encode_data |= data << (j * 4);
}
for (j = 0; j < adev->gfx.num_xcc_per_xcp; j++)
WREG32_SOC15(GC, GET_INST(GC, i * adev->gfx.num_xcc_per_xcp + j),
regGFX_IMU_MCM_ADDR_LUT_PF, encode_data);
}
}
const struct amdgpu_imu_funcs gfx_v12_1_imu_funcs = {
.init_microcode = imu_v12_1_init_microcode,
.load_microcode = imu_v12_1_load_microcode,
.switch_compute_partition = imu_v12_1_switch_compute_partition,
.init_mcm_addr_lut = imu_v12_1_init_mcm_addr_lut,
};