iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
linux/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c

1020 lines
42 KiB
C
Raw Normal View History

drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
/*
* Copyright 2020 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/kernel.h>
#include "amdgpu.h"
#include "amdgpu_gfx.h"
#include "soc15.h"
#include "soc15d.h"
#include "amdgpu_atomfirmware.h"
#include "amdgpu_pm.h"
#include "gc/gc_9_4_1_offset.h"
#include "gc/gc_9_4_1_sh_mask.h"
#include "soc15_common.h"
#include "gfx_v9_4.h"
#include "amdgpu_ras.h"
static const struct soc15_reg_entry gfx_v9_4_edc_counter_regs[] = {
/* CPC */
{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT), 0, 1, 1 },
{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_UCODE_CNT), 0, 1, 1 },
/* DC */
{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_STATE_CNT), 0, 1, 1 },
{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT), 0, 1, 1 },
{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT), 0, 1, 1 },
/* CPF */
{ SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT), 0, 1, 1 },
{ SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_TAG_CNT), 0, 1, 1 },
/* GDS */
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_CNT), 0, 1, 1 },
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_GRBM_CNT), 0, 1, 1 },
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_DED), 0, 1, 1 },
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT), 0, 1, 1 },
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT), 0, 1, 1 },
/* SPI */
{ SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT), 0, 4, 1 },
/* SQ */
drm/amdgpu: correct SE number for arcturus gfx ras When resetting EDC related register, all CUs needs to be visited, otherwise, garbage data from EDC register of missed SEs would present. Signed-off-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-08-26 15:43:42 +08:00
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT), 0, 8, 16 },
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_DED_CNT), 0, 8, 16 },
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_INFO), 0, 8, 16 },
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_SEC_CNT), 0, 8, 16 },
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
/* SQC */
{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT), 0, 4, 6 },
{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2), 0, 4, 6 },
{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3), 0, 4, 6 },
{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3), 0, 4, 6 },
/* TA */
{ SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT), 0, 4, 16 },
/* TCA */
{ SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT), 0, 1, 2 },
/* TCC */
{ SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT), 0, 1, 16 },
{ SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2), 0, 1, 16 },
/* TCI */
{ SOC15_REG_ENTRY(GC, 0, mmTCI_EDC_CNT), 0, 1, 72 },
/* TCP */
{ SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW), 0, 4, 16 },
{ SOC15_REG_ENTRY(GC, 0, mmTCP_ATC_EDC_GATCL1_CNT), 0, 4, 16 },
{ SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT), 0, 4, 16 },
/* TD */
{ SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT), 0, 4, 16 },
/* GCEA */
{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT), 0, 1, 32 },
{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2), 0, 1, 32 },
{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 1, 32 },
/* RLC */
{ SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT), 0, 1, 1 },
{ SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2), 0, 1, 1 },
};
static void gfx_v9_4_select_se_sh(struct amdgpu_device *adev, u32 se_num,
u32 sh_num, u32 instance)
{
u32 data;
if (instance == 0xffffffff)
data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
INSTANCE_BROADCAST_WRITES, 1);
else
data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX,
instance);
if (se_num == 0xffffffff)
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES,
1);
else
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
if (sh_num == 0xffffffff)
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES,
1);
else
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
WREG32_SOC15_RLC_SHADOW(GC, 0, mmGRBM_GFX_INDEX, data);
}
static const struct soc15_ras_field_entry gfx_v9_4_ras_fields[] = {
/* CPC */
{ "CPC_SCRATCH", SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT),
SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, SEC_COUNT),
SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, DED_COUNT) },
{ "CPC_UCODE", SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_UCODE_CNT),
SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, SEC_COUNT),
SOC15_REG_FIELD(CPC_EDC_UCODE_CNT, DED_COUNT) },
{ "CPC_DC_STATE_RAM_ME1", SOC15_REG_ENTRY(GC, 0, mmDC_EDC_STATE_CNT),
SOC15_REG_FIELD(DC_EDC_STATE_CNT, SEC_COUNT_ME1),
SOC15_REG_FIELD(DC_EDC_STATE_CNT, DED_COUNT_ME1) },
{ "CPC_DC_CSINVOC_RAM_ME1",
SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT),
SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, SEC_COUNT_ME1),
SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, DED_COUNT_ME1) },
{ "CPC_DC_RESTORE_RAM_ME1",
SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT),
SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, SEC_COUNT_ME1),
SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, DED_COUNT_ME1) },
{ "CPC_DC_CSINVOC_RAM1_ME1",
SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT),
SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, SEC_COUNT1_ME1),
SOC15_REG_FIELD(DC_EDC_CSINVOC_CNT, DED_COUNT1_ME1) },
{ "CPC_DC_RESTORE_RAM1_ME1",
SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT),
SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, SEC_COUNT1_ME1),
SOC15_REG_FIELD(DC_EDC_RESTORE_CNT, DED_COUNT1_ME1) },
/* CPF */
{ "CPF_ROQ_ME2", SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT),
SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, SEC_COUNT_ME2),
SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, DED_COUNT_ME2) },
{ "CPF_ROQ_ME1", SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT),
SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, SEC_COUNT_ME1),
SOC15_REG_FIELD(CPF_EDC_ROQ_CNT, DED_COUNT_ME1) },
{ "CPF_TCIU_TAG", SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_TAG_CNT),
SOC15_REG_FIELD(CPF_EDC_TAG_CNT, SEC_COUNT),
SOC15_REG_FIELD(CPF_EDC_TAG_CNT, DED_COUNT) },
/* GDS */
{ "GDS_GRBM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_GRBM_CNT),
SOC15_REG_FIELD(GDS_EDC_GRBM_CNT, SEC),
SOC15_REG_FIELD(GDS_EDC_GRBM_CNT, DED) },
{ "GDS_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_CNT),
SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_SEC),
SOC15_REG_FIELD(GDS_EDC_CNT, GDS_MEM_DED) },
{ "GDS_PHY_CMD_RAM_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT),
SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_SEC),
SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_CMD_RAM_MEM_DED) },
{ "GDS_PHY_DATA_RAM_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT),
SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_DATA_RAM_MEM_SEC),
SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, PHY_DATA_RAM_MEM_DED) },
{ "GDS_ME0_CS_PIPE_MEM", SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT),
SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_SEC),
SOC15_REG_FIELD(GDS_EDC_OA_PHY_CNT, ME0_CS_PIPE_MEM_DED) },
{ "GDS_ME1_PIPE0_PIPE_MEM",
SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_SEC),
SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE0_PIPE_MEM_DED) },
{ "GDS_ME1_PIPE1_PIPE_MEM",
SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_SEC),
SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE1_PIPE_MEM_DED) },
{ "GDS_ME1_PIPE2_PIPE_MEM",
SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_SEC),
SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE2_PIPE_MEM_DED) },
{ "GDS_ME1_PIPE3_PIPE_MEM",
SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT),
SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_SEC),
SOC15_REG_FIELD(GDS_EDC_OA_PIPE_CNT, ME1_PIPE3_PIPE_MEM_DED) },
/* SPI */
{ "SPI_SR_MEM", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_SR_MEM_SEC_COUNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_SR_MEM_DED_COUNT) },
{ "SPI_GDS_EXPREQ", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_GDS_EXPREQ_SEC_COUNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_GDS_EXPREQ_DED_COUNT) },
{ "SPI_WB_GRANT_30", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_30_SEC_COUNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_30_DED_COUNT) },
{ "SPI_WB_GRANT_61", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_61_SEC_COUNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_WB_GRANT_61_DED_COUNT) },
{ "SPI_LIFE_CNT", SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_LIFE_CNT_SEC_COUNT),
SOC15_REG_FIELD(SPI_EDC_CNT, SPI_LIFE_CNT_DED_COUNT) },
/* SQ */
{ "SQ_SGPR", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_SEC_COUNT),
SOC15_REG_FIELD(SQ_EDC_CNT, SGPR_DED_COUNT) },
{ "SQ_LDS_D", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_SEC_COUNT),
SOC15_REG_FIELD(SQ_EDC_CNT, LDS_D_DED_COUNT) },
{ "SQ_LDS_I", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_SEC_COUNT),
SOC15_REG_FIELD(SQ_EDC_CNT, LDS_I_DED_COUNT) },
{ "SQ_VGPR0", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_SEC_COUNT),
SOC15_REG_FIELD(SQ_EDC_CNT, VGPR0_DED_COUNT) },
{ "SQ_VGPR1", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_SEC_COUNT),
SOC15_REG_FIELD(SQ_EDC_CNT, VGPR1_DED_COUNT) },
{ "SQ_VGPR2", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_SEC_COUNT),
SOC15_REG_FIELD(SQ_EDC_CNT, VGPR2_DED_COUNT) },
{ "SQ_VGPR3", SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT),
SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_SEC_COUNT),
SOC15_REG_FIELD(SQ_EDC_CNT, VGPR3_DED_COUNT) },
/* SQC */
{ "SQC_INST_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT2, INST_UTCL1_LFIFO_DED_COUNT) },
{ "SQC_DATA_CU0_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_WRITE_DATA_BUF_DED_COUNT) },
{ "SQC_DATA_CU0_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU0_UTCL1_LFIFO_DED_COUNT) },
{ "SQC_DATA_CU1_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_WRITE_DATA_BUF_DED_COUNT) },
{ "SQC_DATA_CU1_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU1_UTCL1_LFIFO_DED_COUNT) },
{ "SQC_DATA_CU2_WRITE_DATA_BUF", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_WRITE_DATA_BUF_DED_COUNT) },
{ "SQC_DATA_CU2_UTCL1_LFIFO", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT, DATA_CU2_UTCL1_LFIFO_DED_COUNT) },
{ "SQC_INST_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_TAG_RAM_DED_COUNT) },
{ "SQC_INST_BANKA_UTCL1_MISS_FIFO",
SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
INST_BANKA_UTCL1_MISS_FIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
INST_BANKA_UTCL1_MISS_FIFO_DED_COUNT) },
{ "SQC_INST_BANKA_MISS_FIFO",
SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKA_MISS_FIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
INST_BANKA_MISS_FIFO_DED_COUNT) },
{ "SQC_INST_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT2, INST_BANKA_BANK_RAM_DED_COUNT) },
{ "SQC_DATA_BANKA_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_TAG_RAM_DED_COUNT) },
{ "SQC_DATA_BANKA_HIT_FIFO",
SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_HIT_FIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_HIT_FIFO_DED_COUNT) },
{ "SQC_DATA_BANKA_MISS_FIFO",
SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKA_MISS_FIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
DATA_BANKA_MISS_FIFO_DED_COUNT) },
{ "SQC_DATA_BANKA_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2),
SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT2, DATA_BANKA_BANK_RAM_DED_COUNT) },
{ "SQC_INST_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_TAG_RAM_DED_COUNT) },
{ "SQC_INST_BANKB_UTCL1_MISS_FIFO",
SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
INST_BANKB_UTCL1_MISS_FIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
INST_BANKB_UTCL1_MISS_FIFO_DED_COUNT) },
{ "SQC_INST_BANKB_MISS_FIFO",
SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, INST_BANKB_MISS_FIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
INST_BANKB_MISS_FIFO_DED_COUNT) },
{ "SQC_INST_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT3, INST_BANKB_BANK_RAM_DED_COUNT) },
{ "SQC_DATA_BANKB_TAG_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_TAG_RAM_DED_COUNT) },
{ "SQC_DATA_BANKB_HIT_FIFO",
SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_HIT_FIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_HIT_FIFO_DED_COUNT) },
{ "SQC_DATA_BANKB_MISS_FIFO",
SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_PARITY_CNT3),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3, DATA_BANKB_MISS_FIFO_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_PARITY_CNT3,
DATA_BANKB_MISS_FIFO_DED_COUNT) },
{ "SQC_DATA_BANKB_BANK_RAM", SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3),
SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_SEC_COUNT),
SOC15_REG_FIELD(SQC_EDC_CNT3, DATA_BANKB_BANK_RAM_DED_COUNT) },
/* TA */
{ "TA_FS_DFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_SEC_COUNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_DFIFO_DED_COUNT) },
{ "TA_FS_AFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_SEC_COUNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_AFIFO_DED_COUNT) },
{ "TA_FL_LFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FL_LFIFO_SEC_COUNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FL_LFIFO_DED_COUNT) },
{ "TA_FX_LFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FX_LFIFO_SEC_COUNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FX_LFIFO_DED_COUNT) },
{ "TA_FS_CFIFO", SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_CFIFO_SEC_COUNT),
SOC15_REG_FIELD(TA_EDC_CNT, TA_FS_CFIFO_DED_COUNT) },
/* TCA */
{ "TCA_HOLE_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT),
SOC15_REG_FIELD(TCA_EDC_CNT, HOLE_FIFO_SEC_COUNT),
SOC15_REG_FIELD(TCA_EDC_CNT, HOLE_FIFO_DED_COUNT) },
{ "TCA_REQ_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT),
SOC15_REG_FIELD(TCA_EDC_CNT, REQ_FIFO_SEC_COUNT),
SOC15_REG_FIELD(TCA_EDC_CNT, REQ_FIFO_DED_COUNT) },
/* TCC */
{ "TCC_CACHE_DATA", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DATA_DED_COUNT) },
{ "TCC_CACHE_DIRTY", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT, CACHE_DIRTY_DED_COUNT) },
{ "TCC_HIGH_RATE_TAG", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT, HIGH_RATE_TAG_DED_COUNT) },
{ "TCC_LOW_RATE_TAG", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT, LOW_RATE_TAG_DED_COUNT) },
{ "TCC_IN_USE_DEC", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_DEC_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_DEC_DED_COUNT) },
{ "TCC_IN_USE_TRANSFER", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_TRANSFER_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT2, IN_USE_TRANSFER_DED_COUNT) },
{ "TCC_RETURN_DATA", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_DATA_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_DATA_DED_COUNT) },
{ "TCC_RETURN_CONTROL", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_CONTROL_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT2, RETURN_CONTROL_DED_COUNT) },
{ "TCC_UC_ATOMIC_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
SOC15_REG_FIELD(TCC_EDC_CNT2, UC_ATOMIC_FIFO_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT2, UC_ATOMIC_FIFO_DED_COUNT) },
{ "TCC_WRITE_RETURN", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_RETURN_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_RETURN_DED_COUNT) },
{ "TCC_WRITE_CACHE_READ", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_CACHE_READ_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT2, WRITE_CACHE_READ_DED_COUNT) },
{ "TCC_SRC_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT, SRC_FIFO_DED_COUNT) },
{ "TCC_CACHE_TAG_PROBE_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2),
SOC15_REG_FIELD(TCC_EDC_CNT2, CACHE_TAG_PROBE_FIFO_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT2, CACHE_TAG_PROBE_FIFO_DED_COUNT) },
{ "TCC_LATENCY_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_DED_COUNT) },
{ "TCC_LATENCY_FIFO_NEXT_RAM", SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT),
SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_NEXT_RAM_SEC_COUNT),
SOC15_REG_FIELD(TCC_EDC_CNT, LATENCY_FIFO_NEXT_RAM_DED_COUNT) },
/* TCI */
{ "TCI_WRITE_RAM", SOC15_REG_ENTRY(GC, 0, mmTCI_EDC_CNT),
SOC15_REG_FIELD(TCI_EDC_CNT, WRITE_RAM_SEC_COUNT),
SOC15_REG_FIELD(TCI_EDC_CNT, WRITE_RAM_DED_COUNT) },
/* TCP */
{ "TCP_CACHE_RAM", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_SEC_COUNT),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CACHE_RAM_DED_COUNT) },
{ "TCP_LFIFO_RAM", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_SEC_COUNT),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, LFIFO_RAM_DED_COUNT) },
{ "TCP_CMD_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CMD_FIFO_SEC_COUNT),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, CMD_FIFO_DED_COUNT) },
{ "TCP_VM_FIFO", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, VM_FIFO_SEC_COUNT),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, VM_FIFO_DED_COUNT) },
{ "TCP_DB_RAM", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, DB_RAM_SED_COUNT), 0, 0 },
{ "TCP_UTCL1_LFIFO0", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_SEC_COUNT),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO0_DED_COUNT) },
{ "TCP_UTCL1_LFIFO1", SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT_NEW),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_SEC_COUNT),
SOC15_REG_FIELD(TCP_EDC_CNT_NEW, UTCL1_LFIFO1_DED_COUNT) },
/* TD */
{ "TD_SS_FIFO_LO", SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT),
SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_SEC_COUNT),
SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_LO_DED_COUNT) },
{ "TD_SS_FIFO_HI", SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT),
SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_SEC_COUNT),
SOC15_REG_FIELD(TD_EDC_CNT, SS_FIFO_HI_DED_COUNT) },
{ "TD_CS_FIFO", SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT),
SOC15_REG_FIELD(TD_EDC_CNT, CS_FIFO_SEC_COUNT),
SOC15_REG_FIELD(TD_EDC_CNT, CS_FIFO_DED_COUNT) },
/* EA */
{ "EA_DRAMRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_CMDMEM_DED_COUNT) },
{ "EA_DRAMWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_CMDMEM_DED_COUNT) },
{ "EA_DRAMWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_DATAMEM_DED_COUNT) },
{ "EA_RRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, RRET_TAGMEM_DED_COUNT) },
{ "EA_WRET_TAGMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, WRET_TAGMEM_DED_COUNT) },
{ "EA_GMIRD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_CMDMEM_DED_COUNT) },
{ "EA_GMIWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_CMDMEM_DED_COUNT) },
{ "EA_GMIWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_DATAMEM_DED_COUNT) },
{ "EA_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMRD_PAGEMEM_SED_COUNT), 0, 0 },
{ "EA_DRAMRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
SOC15_REG_FIELD(GCEA_EDC_CNT3, DRAMRD_PAGEMEM_DED_COUNT) },
{ "EA_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, DRAMWR_PAGEMEM_SED_COUNT), 0, 0 },
{ "EA_DRAMWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
SOC15_REG_FIELD(GCEA_EDC_CNT3, DRAMWR_PAGEMEM_DED_COUNT) },
{ "EA_IORD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, IORD_CMDMEM_SED_COUNT), 0, 0 },
{ "EA_IORD_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
SOC15_REG_FIELD(GCEA_EDC_CNT3, IORD_CMDMEM_DED_COUNT) },
{ "EA_IOWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_CMDMEM_SED_COUNT), 0, 0 },
{ "EA_IOWR_CMDMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
SOC15_REG_FIELD(GCEA_EDC_CNT3, IOWR_CMDMEM_DED_COUNT) },
{ "EA_IOWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, IOWR_DATAMEM_SED_COUNT), 0, 0 },
{ "EA_IOWR_DATAMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
SOC15_REG_FIELD(GCEA_EDC_CNT3, IOWR_DATAMEM_DED_COUNT) },
{ "EA_GMIRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIRD_PAGEMEM_SED_COUNT), 0, 0 },
{ "EA_GMIRD_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
SOC15_REG_FIELD(GCEA_EDC_CNT3, GMIRD_PAGEMEM_DED_COUNT) },
{ "EA_GMIWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
SOC15_REG_FIELD(GCEA_EDC_CNT2, GMIWR_PAGEMEM_SED_COUNT), 0, 0 },
{ "EA_GMIWR_PAGEMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
SOC15_REG_FIELD(GCEA_EDC_CNT3, GMIWR_PAGEMEM_DED_COUNT) },
{ "EA_MAM_D0MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D0MEM_SED_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D0MEM_DED_COUNT) },
{ "EA_MAM_D1MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D1MEM_SED_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D1MEM_DED_COUNT) },
{ "EA_MAM_D2MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D2MEM_SED_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D2MEM_DED_COUNT) },
{ "EA_MAM_D3MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2),
SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D3MEM_SED_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT2, MAM_D3MEM_DED_COUNT) },
{ "EA_MAM_A0MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3),
SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A0MEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A0MEM_DED_COUNT) },
{ "EA_MAM_A1MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3),
SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A1MEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A1MEM_DED_COUNT) },
{ "EA_MAM_A2MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3),
SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A2MEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A2MEM_DED_COUNT) },
{ "EA_MAM_A3MEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3),
SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A3MEM_SEC_COUNT),
SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_A3MEM_DED_COUNT) },
{ "EA_MAM_AFMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT),
SOC15_REG_FIELD(GCEA_EDC_CNT, MAM_AFMEM_SEC_COUNT), 0, 0 },
{ "EA_MAM_AFMEM", SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT3), 0, 0,
SOC15_REG_FIELD(GCEA_EDC_CNT3, MAM_AFMEM_DED_COUNT) },
/* RLC */
{ "RLCG_INSTR_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_INSTR_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_INSTR_RAM_DED_COUNT) },
{ "RLCG_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLCG_SCRATCH_RAM_DED_COUNT) },
{ "RLCV_INSTR_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_INSTR_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_INSTR_RAM_DED_COUNT) },
{ "RLCV_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLCV_SCRATCH_RAM_DED_COUNT) },
{ "RLC_TCTAG_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLC_TCTAG_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLC_TCTAG_RAM_DED_COUNT) },
{ "RLC_SPM_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SPM_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SPM_SCRATCH_RAM_DED_COUNT) },
{ "RLC_SRM_DATA_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_DATA_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_DATA_RAM_DED_COUNT) },
{ "RLC_SRM_ADDR_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_ADDR_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT, RLC_SRM_ADDR_RAM_DED_COUNT) },
{ "RLC_SPM_SE0_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE0_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE0_SCRATCH_RAM_DED_COUNT) },
{ "RLC_SPM_SE1_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE1_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE1_SCRATCH_RAM_DED_COUNT) },
{ "RLC_SPM_SE2_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE2_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE2_SCRATCH_RAM_DED_COUNT) },
{ "RLC_SPM_SE3_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE3_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE3_SCRATCH_RAM_DED_COUNT) },
{ "RLC_SPM_SE4_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE4_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE4_SCRATCH_RAM_DED_COUNT) },
{ "RLC_SPM_SE5_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE5_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE5_SCRATCH_RAM_DED_COUNT) },
{ "RLC_SPM_SE6_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE6_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE6_SCRATCH_RAM_DED_COUNT) },
{ "RLC_SPM_SE7_SCRATCH_RAM", SOC15_REG_ENTRY(GC, 0, mmRLC_EDC_CNT2),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE7_SCRATCH_RAM_SEC_COUNT),
SOC15_REG_FIELD(RLC_EDC_CNT2, RLC_SPM_SE7_SCRATCH_RAM_DED_COUNT) },
};
static const char * const vml2_mems[] = {
"UTC_VML2_BANK_CACHE_0_BIGK_MEM0",
"UTC_VML2_BANK_CACHE_0_BIGK_MEM1",
"UTC_VML2_BANK_CACHE_0_4K_MEM0",
"UTC_VML2_BANK_CACHE_0_4K_MEM1",
"UTC_VML2_BANK_CACHE_1_BIGK_MEM0",
"UTC_VML2_BANK_CACHE_1_BIGK_MEM1",
"UTC_VML2_BANK_CACHE_1_4K_MEM0",
"UTC_VML2_BANK_CACHE_1_4K_MEM1",
"UTC_VML2_BANK_CACHE_2_BIGK_MEM0",
"UTC_VML2_BANK_CACHE_2_BIGK_MEM1",
"UTC_VML2_BANK_CACHE_2_4K_MEM0",
"UTC_VML2_BANK_CACHE_2_4K_MEM1",
"UTC_VML2_BANK_CACHE_3_BIGK_MEM0",
"UTC_VML2_BANK_CACHE_3_BIGK_MEM1",
"UTC_VML2_BANK_CACHE_3_4K_MEM0",
"UTC_VML2_BANK_CACHE_3_4K_MEM1",
"UTC_VML2_IFIFO_GROUP0",
"UTC_VML2_IFIFO_GROUP1",
"UTC_VML2_IFIFO_GROUP2",
"UTC_VML2_IFIFO_GROUP3",
"UTC_VML2_IFIFO_GROUP4",
"UTC_VML2_IFIFO_GROUP5",
"UTC_VML2_IFIFO_GROUP6",
"UTC_VML2_IFIFO_GROUP7",
"UTC_VML2_IFIFO_GROUP8",
"UTC_VML2_IFIFO_GROUP9",
"UTC_VML2_IFIFO_GROUP10",
"UTC_VML2_IFIFO_GROUP11",
"UTC_VML2_IFIFO_GROUP12",
"UTC_VML2_IFIFO_GROUP13",
"UTC_VML2_IFIFO_GROUP14",
"UTC_VML2_IFIFO_GROUP15",
"UTC_VML2_IFIFO_GROUP16",
"UTC_VML2_IFIFO_GROUP17",
"UTC_VML2_IFIFO_GROUP18",
"UTC_VML2_IFIFO_GROUP19",
"UTC_VML2_IFIFO_GROUP20",
"UTC_VML2_IFIFO_GROUP21",
"UTC_VML2_IFIFO_GROUP22",
"UTC_VML2_IFIFO_GROUP23",
"UTC_VML2_IFIFO_GROUP24",
};
static const char * const vml2_walker_mems[] = {
"UTC_VML2_CACHE_PDE0_MEM0",
"UTC_VML2_CACHE_PDE0_MEM1",
"UTC_VML2_CACHE_PDE1_MEM0",
"UTC_VML2_CACHE_PDE1_MEM1",
"UTC_VML2_CACHE_PDE2_MEM0",
"UTC_VML2_CACHE_PDE2_MEM1",
"UTC_VML2_RDIF_ARADDRS",
"UTC_VML2_RDIF_LOG_FIFO",
"UTC_VML2_QUEUE_REQ",
"UTC_VML2_QUEUE_RET",
};
static const char * const utcl2_router_mems[] = {
"UTCL2_ROUTER_GROUP0_VML2_REQ_FIFO0",
"UTCL2_ROUTER_GROUP1_VML2_REQ_FIFO1",
"UTCL2_ROUTER_GROUP2_VML2_REQ_FIFO2",
"UTCL2_ROUTER_GROUP3_VML2_REQ_FIFO3",
"UTCL2_ROUTER_GROUP4_VML2_REQ_FIFO4",
"UTCL2_ROUTER_GROUP5_VML2_REQ_FIFO5",
"UTCL2_ROUTER_GROUP6_VML2_REQ_FIFO6",
"UTCL2_ROUTER_GROUP7_VML2_REQ_FIFO7",
"UTCL2_ROUTER_GROUP8_VML2_REQ_FIFO8",
"UTCL2_ROUTER_GROUP9_VML2_REQ_FIFO9",
"UTCL2_ROUTER_GROUP10_VML2_REQ_FIFO10",
"UTCL2_ROUTER_GROUP11_VML2_REQ_FIFO11",
"UTCL2_ROUTER_GROUP12_VML2_REQ_FIFO12",
"UTCL2_ROUTER_GROUP13_VML2_REQ_FIFO13",
"UTCL2_ROUTER_GROUP14_VML2_REQ_FIFO14",
"UTCL2_ROUTER_GROUP15_VML2_REQ_FIFO15",
"UTCL2_ROUTER_GROUP16_VML2_REQ_FIFO16",
"UTCL2_ROUTER_GROUP17_VML2_REQ_FIFO17",
"UTCL2_ROUTER_GROUP18_VML2_REQ_FIFO18",
"UTCL2_ROUTER_GROUP19_VML2_REQ_FIFO19",
"UTCL2_ROUTER_GROUP20_VML2_REQ_FIFO20",
"UTCL2_ROUTER_GROUP21_VML2_REQ_FIFO21",
"UTCL2_ROUTER_GROUP22_VML2_REQ_FIFO22",
"UTCL2_ROUTER_GROUP23_VML2_REQ_FIFO23",
"UTCL2_ROUTER_GROUP24_VML2_REQ_FIFO24",
};
static const char * const atc_l2_cache_2m_mems[] = {
"UTC_ATCL2_CACHE_2M_BANK0_WAY0_MEM",
"UTC_ATCL2_CACHE_2M_BANK0_WAY1_MEM",
"UTC_ATCL2_CACHE_2M_BANK1_WAY0_MEM",
"UTC_ATCL2_CACHE_2M_BANK1_WAY1_MEM",
};
static const char * const atc_l2_cache_4k_mems[] = {
"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM0",
"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM1",
"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM2",
"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM3",
"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM4",
"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM5",
"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM6",
"UTC_ATCL2_CACHE_4K_BANK0_WAY0_MEM7",
"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM0",
"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM1",
"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM2",
"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM3",
"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM4",
"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM5",
"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM6",
"UTC_ATCL2_CACHE_4K_BANK0_WAY1_MEM7",
"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM0",
"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM1",
"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM2",
"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM3",
"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM4",
"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM5",
"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM6",
"UTC_ATCL2_CACHE_4K_BANK1_WAY0_MEM7",
"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM0",
"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM1",
"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM2",
"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM3",
"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM4",
"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM5",
"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM6",
"UTC_ATCL2_CACHE_4K_BANK1_WAY1_MEM7",
};
static int gfx_v9_4_query_utc_edc_status(struct amdgpu_device *adev,
struct ras_err_data *err_data)
{
uint32_t i, data;
uint32_t sec_count, ded_count;
WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_CNTL, 0);
WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_CNTL, 0);
WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_CNTL, 0);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, 255);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL, 0);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL, 0);
for (i = 0; i < ARRAY_SIZE(vml2_mems); i++) {
WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, i);
data = RREG32_SOC15(GC, 0, mmVML2_MEM_ECC_CNTL);
sec_count = REG_GET_FIELD(data, VML2_MEM_ECC_CNTL, SEC_COUNT);
if (sec_count) {
drm/amdgpu: update RAS related dmesg print prefix RAS error related dmesg print with pci device info Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: John Clements <john.clements@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-07 16:54:33 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, SEC %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
vml2_mems[i], sec_count);
err_data->ce_count += sec_count;
}
ded_count = REG_GET_FIELD(data, VML2_MEM_ECC_CNTL, DED_COUNT);
if (ded_count) {
drm/amdgpu: update RAS related dmesg print prefix RAS error related dmesg print with pci device info Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: John Clements <john.clements@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-07 16:54:33 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, DED %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
vml2_mems[i], ded_count);
err_data->ue_count += ded_count;
}
}
for (i = 0; i < ARRAY_SIZE(vml2_walker_mems); i++) {
WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, i);
data = RREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_CNTL);
sec_count = REG_GET_FIELD(data, VML2_WALKER_MEM_ECC_CNTL,
SEC_COUNT);
if (sec_count) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, SEC %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
vml2_walker_mems[i], sec_count);
err_data->ce_count += sec_count;
}
ded_count = REG_GET_FIELD(data, VML2_WALKER_MEM_ECC_CNTL,
DED_COUNT);
if (ded_count) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, DED %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
vml2_walker_mems[i], ded_count);
err_data->ue_count += ded_count;
}
}
for (i = 0; i < ARRAY_SIZE(utcl2_router_mems); i++) {
WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, i);
data = RREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_CNTL);
sec_count = REG_GET_FIELD(data, UTCL2_MEM_ECC_CNTL, SEC_COUNT);
if (sec_count) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, SEC %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
utcl2_router_mems[i], sec_count);
err_data->ce_count += sec_count;
}
ded_count = REG_GET_FIELD(data, UTCL2_MEM_ECC_CNTL, DED_COUNT);
if (ded_count) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, DED %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
utcl2_router_mems[i], ded_count);
err_data->ue_count += ded_count;
}
}
for (i = 0; i < ARRAY_SIZE(atc_l2_cache_2m_mems); i++) {
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, i);
data = RREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL);
sec_count = REG_GET_FIELD(data, ATC_L2_CACHE_2M_DSM_CNTL,
SEC_COUNT);
if (sec_count) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, SEC %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
atc_l2_cache_2m_mems[i], sec_count);
err_data->ce_count += sec_count;
}
ded_count = REG_GET_FIELD(data, ATC_L2_CACHE_2M_DSM_CNTL,
DED_COUNT);
if (ded_count) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, DED %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
atc_l2_cache_2m_mems[i], ded_count);
err_data->ue_count += ded_count;
}
}
for (i = 0; i < ARRAY_SIZE(atc_l2_cache_4k_mems); i++) {
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, i);
data = RREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_CNTL);
sec_count = REG_GET_FIELD(data, ATC_L2_CACHE_4K_DSM_CNTL,
SEC_COUNT);
if (sec_count) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, SEC %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
atc_l2_cache_4k_mems[i], sec_count);
err_data->ce_count += sec_count;
}
ded_count = REG_GET_FIELD(data, ATC_L2_CACHE_4K_DSM_CNTL,
DED_COUNT);
if (ded_count) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"Instance[%d]: SubBlock %s, DED %d\n", i,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
atc_l2_cache_4k_mems[i], ded_count);
err_data->ue_count += ded_count;
}
}
WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, 255);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
return 0;
}
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
static int gfx_v9_4_ras_error_count(struct amdgpu_device *adev,
const struct soc15_reg_entry *reg,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
uint32_t se_id, uint32_t inst_id,
uint32_t value, uint32_t *sec_count,
uint32_t *ded_count)
{
uint32_t i;
uint32_t sec_cnt, ded_cnt;
for (i = 0; i < ARRAY_SIZE(gfx_v9_4_ras_fields); i++) {
if (gfx_v9_4_ras_fields[i].reg_offset != reg->reg_offset ||
gfx_v9_4_ras_fields[i].seg != reg->seg ||
gfx_v9_4_ras_fields[i].inst != reg->inst)
continue;
sec_cnt = (value & gfx_v9_4_ras_fields[i].sec_count_mask) >>
gfx_v9_4_ras_fields[i].sec_count_shift;
if (sec_cnt) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"GFX SubBlock %s, Instance[%d][%d], SEC %d\n",
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
gfx_v9_4_ras_fields[i].name, se_id, inst_id,
sec_cnt);
*sec_count += sec_cnt;
}
ded_cnt = (value & gfx_v9_4_ras_fields[i].ded_count_mask) >>
gfx_v9_4_ras_fields[i].ded_count_shift;
if (ded_cnt) {
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
dev_info(adev->dev,
"GFX SubBlock %s, Instance[%d][%d], DED %d\n",
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
gfx_v9_4_ras_fields[i].name, se_id, inst_id,
ded_cnt);
*ded_count += ded_cnt;
}
}
return 0;
}
drm/amdgpu: Modify gfx block to fit for the unified ras block data and ops 1.Modify gfx block to fit for the unified ras block data and ops. 2.Change amdgpu_gfx_ras_funcs to amdgpu_gfx_ras, and the corresponding variable name remove _funcs suffix. 3.Remove the const flag of gfx ras variable so that gfx ras block can be able to be inserted into amdgpu device ras block link list. 4.Invoke amdgpu_ras_register_ras_block function to register gfx ras block into amdgpu device ras block link list. 5.Remove the redundant code about gfx in amdgpu_ras.c after using the unified ras block. 6.Fill unified ras block .name .block .ras_late_init and .ras_fini for all of gfx versions. If .ras_late_init and .ras_fini had been defined by the selected gfx version, the defined functions will take effect; if not defined, default fill with amdgpu_gfx_ras_late_init and amdgpu_gfx_ras_fini. Signed-off-by: yipechai <YiPeng.Chai@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: John Clements <john.clements@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2022-01-04 13:52:46 +08:00
static void gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev,
drm/amdgpu: split gfx callbacks into ras and non-ras ones gfx ras is only available in cerntain ip generations. Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: John Clements <John.Clements@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2021-03-19 16:59:09 +08:00
void *ras_error_status)
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
{
struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
uint32_t sec_count = 0, ded_count = 0;
uint32_t i, j, k;
uint32_t reg_value;
if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
drm/amdgpu: Modify gfx block to fit for the unified ras block data and ops 1.Modify gfx block to fit for the unified ras block data and ops. 2.Change amdgpu_gfx_ras_funcs to amdgpu_gfx_ras, and the corresponding variable name remove _funcs suffix. 3.Remove the const flag of gfx ras variable so that gfx ras block can be able to be inserted into amdgpu device ras block link list. 4.Invoke amdgpu_ras_register_ras_block function to register gfx ras block into amdgpu device ras block link list. 5.Remove the redundant code about gfx in amdgpu_ras.c after using the unified ras block. 6.Fill unified ras block .name .block .ras_late_init and .ras_fini for all of gfx versions. If .ras_late_init and .ras_fini had been defined by the selected gfx version, the defined functions will take effect; if not defined, default fill with amdgpu_gfx_ras_late_init and amdgpu_gfx_ras_fini. Signed-off-by: yipechai <YiPeng.Chai@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: John Clements <john.clements@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2022-01-04 13:52:46 +08:00
return;
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
err_data->ue_count = 0;
err_data->ce_count = 0;
mutex_lock(&adev->grbm_idx_mutex);
for (i = 0; i < ARRAY_SIZE(gfx_v9_4_edc_counter_regs); i++) {
for (j = 0; j < gfx_v9_4_edc_counter_regs[i].se_num; j++) {
for (k = 0; k < gfx_v9_4_edc_counter_regs[i].instance;
k++) {
gfx_v9_4_select_se_sh(adev, j, 0, k);
reg_value = RREG32(SOC15_REG_ENTRY_OFFSET(
gfx_v9_4_edc_counter_regs[i]));
if (reg_value)
drm/amdgpu: replace DRM prefix with PCI device info for gfx/mmhub Prefix RAS message printing in gfx/mmhub with PCI device info, which assists the debug in multiple GPU case. Reviewed-by: Guchun Chen <guchun.chen@amd.com> Signed-off-by: Dennis Li <Dennis.Li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-04-18 12:08:20 +08:00
gfx_v9_4_ras_error_count(adev,
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
&gfx_v9_4_edc_counter_regs[i],
j, k, reg_value, &sec_count,
&ded_count);
}
}
}
err_data->ce_count += sec_count;
err_data->ue_count += ded_count;
gfx_v9_4_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
mutex_unlock(&adev->grbm_idx_mutex);
gfx_v9_4_query_utc_edc_status(adev, err_data);
}
drm/amdgpu: split gfx callbacks into ras and non-ras ones gfx ras is only available in cerntain ip generations. Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: John Clements <John.Clements@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2021-03-19 16:59:09 +08:00
static void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev)
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
{
int i, j, k;
drm/amdgpu: check GFX RAS capability before reset counters disallow the logical to be enabled on platforms that don't support gfx ras at this stage, like sriov skus, dgpu with legacy ras.etc Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Monk Liu <monk.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-03-09 16:34:37 +08:00
if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
return;
drm/amdgpu: add RAS support for the gfx block of Arcturus Implement functions to do the RAS error injection and query EDC counter. Signed-off-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-01-16 13:40:52 +08:00
mutex_lock(&adev->grbm_idx_mutex);
for (i = 0; i < ARRAY_SIZE(gfx_v9_4_edc_counter_regs); i++) {
for (j = 0; j < gfx_v9_4_edc_counter_regs[i].se_num; j++) {
for (k = 0; k < gfx_v9_4_edc_counter_regs[i].instance;
k++) {
gfx_v9_4_select_se_sh(adev, j, 0x0, k);
RREG32(SOC15_REG_ENTRY_OFFSET(
gfx_v9_4_edc_counter_regs[i]));
}
}
}
WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, 0xe0000000);
mutex_unlock(&adev->grbm_idx_mutex);
WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_CNTL, 0);
WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_CNTL, 0);
WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_CNTL, 0);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, 255);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL, 0);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL, 0);
for (i = 0; i < ARRAY_SIZE(vml2_mems); i++) {
WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, i);
RREG32_SOC15(GC, 0, mmVML2_MEM_ECC_CNTL);
}
for (i = 0; i < ARRAY_SIZE(vml2_walker_mems); i++) {
WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, i);
RREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_CNTL);
}
for (i = 0; i < ARRAY_SIZE(utcl2_router_mems); i++) {
WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, i);
RREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_CNTL);
}
for (i = 0; i < ARRAY_SIZE(atc_l2_cache_2m_mems); i++) {
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, i);
RREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_CNTL);
}
for (i = 0; i < ARRAY_SIZE(atc_l2_cache_4k_mems); i++) {
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, i);
RREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_CNTL);
}
WREG32_SOC15(GC, 0, mmVML2_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmVML2_WALKER_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmUTCL2_MEM_ECC_INDEX, 255);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_DSM_INDEX, 255);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
}
drm/amdgpu: only harvest gcea/mmea error status in arcturus SDP RdRspStatus/WrRspStatus or first parity error on RdRsp data can cause system fatal error in arcturus. GPU will be freezed in such case. Driver needs to harvest these error information before reset the GPU. Check error type to avoid harvest normal gcea/mmea information. Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Stanley Yang <Stanley.Yang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2021-04-16 17:34:13 +08:00
static const struct soc15_reg_entry gfx_v9_4_ea_err_status_regs =
drm/amdgpu: update athub interrupt harvesting handle GCEA/MMHUB EA error should not result to DF freeze, this is fixed in next generation, but for some reasons the GCEA/MMHUB EA error will result to DF freeze in previous generation, diver should avoid to indicate GCEA/MMHUB EA error as hw fatal error in kernel message by read GCEA/MMHUB err status registers. Changed from V1: make query_ras_error_status function more general make read mmhub er status register more friendly Changed from V2: move ras error status query function into do_recovery workqueue Changed from V3: remove useless code from V2, print GCEA error status instance number Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-09-15 16:15:05 +08:00
{ SOC15_REG_ENTRY(GC, 0, mmGCEA_ERR_STATUS), 0, 1, 32 };
drm/amdgpu: split gfx callbacks into ras and non-ras ones gfx ras is only available in cerntain ip generations. Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: John Clements <John.Clements@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2021-03-19 16:59:09 +08:00
static void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev)
drm/amdgpu: update athub interrupt harvesting handle GCEA/MMHUB EA error should not result to DF freeze, this is fixed in next generation, but for some reasons the GCEA/MMHUB EA error will result to DF freeze in previous generation, diver should avoid to indicate GCEA/MMHUB EA error as hw fatal error in kernel message by read GCEA/MMHUB err status registers. Changed from V1: make query_ras_error_status function more general make read mmhub er status register more friendly Changed from V2: move ras error status query function into do_recovery workqueue Changed from V3: remove useless code from V2, print GCEA error status instance number Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-09-15 16:15:05 +08:00
{
uint32_t i, j;
uint32_t reg_value;
if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
return;
mutex_lock(&adev->grbm_idx_mutex);
drm/amdgpu: only harvest gcea/mmea error status in arcturus SDP RdRspStatus/WrRspStatus or first parity error on RdRsp data can cause system fatal error in arcturus. GPU will be freezed in such case. Driver needs to harvest these error information before reset the GPU. Check error type to avoid harvest normal gcea/mmea information. Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Stanley Yang <Stanley.Yang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2021-04-16 17:34:13 +08:00
for (i = 0; i < gfx_v9_4_ea_err_status_regs.se_num; i++) {
for (j = 0; j < gfx_v9_4_ea_err_status_regs.instance;
drm/amdgpu: update athub interrupt harvesting handle GCEA/MMHUB EA error should not result to DF freeze, this is fixed in next generation, but for some reasons the GCEA/MMHUB EA error will result to DF freeze in previous generation, diver should avoid to indicate GCEA/MMHUB EA error as hw fatal error in kernel message by read GCEA/MMHUB err status registers. Changed from V1: make query_ras_error_status function more general make read mmhub er status register more friendly Changed from V2: move ras error status query function into do_recovery workqueue Changed from V3: remove useless code from V2, print GCEA error status instance number Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-09-15 16:15:05 +08:00
j++) {
gfx_v9_4_select_se_sh(adev, i, 0, j);
reg_value = RREG32(SOC15_REG_ENTRY_OFFSET(
drm/amdgpu: only harvest gcea/mmea error status in arcturus SDP RdRspStatus/WrRspStatus or first parity error on RdRsp data can cause system fatal error in arcturus. GPU will be freezed in such case. Driver needs to harvest these error information before reset the GPU. Check error type to avoid harvest normal gcea/mmea information. Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Stanley Yang <Stanley.Yang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2021-04-16 17:34:13 +08:00
gfx_v9_4_ea_err_status_regs));
if (REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_STATUS) ||
REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_WRRSP_STATUS) ||
REG_GET_FIELD(reg_value, GCEA_ERR_STATUS, SDP_RDRSP_DATAPARITY_ERROR)) {
/* SDP read/write error/parity error in FUE_IS_FATAL mode
* can cause system fatal error in arcturas. Harvest the error
* status before GPU reset */
drm/amdgpu: update athub interrupt harvesting handle GCEA/MMHUB EA error should not result to DF freeze, this is fixed in next generation, but for some reasons the GCEA/MMHUB EA error will result to DF freeze in previous generation, diver should avoid to indicate GCEA/MMHUB EA error as hw fatal error in kernel message by read GCEA/MMHUB err status registers. Changed from V1: make query_ras_error_status function more general make read mmhub er status register more friendly Changed from V2: move ras error status query function into do_recovery workqueue Changed from V3: remove useless code from V2, print GCEA error status instance number Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-09-15 16:15:05 +08:00
dev_warn(adev->dev, "GCEA err detected at instance: %d, status: 0x%x!\n",
j, reg_value);
drm/amdgpu: only harvest gcea/mmea error status in arcturus SDP RdRspStatus/WrRspStatus or first parity error on RdRsp data can cause system fatal error in arcturus. GPU will be freezed in such case. Driver needs to harvest these error information before reset the GPU. Check error type to avoid harvest normal gcea/mmea information. Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Stanley Yang <Stanley.Yang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2021-04-16 17:34:13 +08:00
}
drm/amdgpu: update athub interrupt harvesting handle GCEA/MMHUB EA error should not result to DF freeze, this is fixed in next generation, but for some reasons the GCEA/MMHUB EA error will result to DF freeze in previous generation, diver should avoid to indicate GCEA/MMHUB EA error as hw fatal error in kernel message by read GCEA/MMHUB err status registers. Changed from V1: make query_ras_error_status function more general make read mmhub er status register more friendly Changed from V2: move ras error status query function into do_recovery workqueue Changed from V3: remove useless code from V2, print GCEA error status instance number Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2020-09-15 16:15:05 +08:00
}
}
gfx_v9_4_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
mutex_unlock(&adev->grbm_idx_mutex);
}
drm/amdgpu: split gfx callbacks into ras and non-ras ones gfx ras is only available in cerntain ip generations. Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: John Clements <John.Clements@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2021-03-19 16:59:09 +08:00
drm/amdgpu: Modify gfx block to fit for the unified ras block data and ops 1.Modify gfx block to fit for the unified ras block data and ops. 2.Change amdgpu_gfx_ras_funcs to amdgpu_gfx_ras, and the corresponding variable name remove _funcs suffix. 3.Remove the const flag of gfx ras variable so that gfx ras block can be able to be inserted into amdgpu device ras block link list. 4.Invoke amdgpu_ras_register_ras_block function to register gfx ras block into amdgpu device ras block link list. 5.Remove the redundant code about gfx in amdgpu_ras.c after using the unified ras block. 6.Fill unified ras block .name .block .ras_late_init and .ras_fini for all of gfx versions. If .ras_late_init and .ras_fini had been defined by the selected gfx version, the defined functions will take effect; if not defined, default fill with amdgpu_gfx_ras_late_init and amdgpu_gfx_ras_fini. Signed-off-by: yipechai <YiPeng.Chai@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: John Clements <john.clements@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2022-01-04 13:52:46 +08:00
const struct amdgpu_ras_block_hw_ops gfx_v9_4_ras_ops = {
.query_ras_error_count = &gfx_v9_4_query_ras_error_count,
.reset_ras_error_count = &gfx_v9_4_reset_ras_error_count,
.query_ras_error_status = &gfx_v9_4_query_ras_error_status,
};
struct amdgpu_gfx_ras gfx_v9_4_ras = {
.ras_block = {
.hw_ops = &gfx_v9_4_ras_ops,
},
drm/amdgpu: split gfx callbacks into ras and non-ras ones gfx ras is only available in cerntain ip generations. Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Dennis Li <Dennis.Li@amd.com> Reviewed-by: John Clements <John.Clements@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2021-03-19 16:59:09 +08:00
};
Reference in New Issue Copy Permalink