Loading...
1/*
2 * Copyright (C) 2019 Advanced Micro Devices, Inc.
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice shall be included
12 * in all copies or substantial portions of the Software.
13 *
14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
15 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
17 * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
18 * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
19 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
20 */
21#ifndef __AMDGPU_UMC_H__
22#define __AMDGPU_UMC_H__
23#include "amdgpu_ras.h"
24
25/*
26 * (addr / 256) * 4096, the higher 26 bits in ErrorAddr
27 * is the index of 4KB block
28 */
29#define ADDR_OF_4KB_BLOCK(addr) (((addr) & ~0xffULL) << 4)
30/*
31 * (addr / 256) * 8192, the higher 26 bits in ErrorAddr
32 * is the index of 8KB block
33 */
34#define ADDR_OF_8KB_BLOCK(addr) (((addr) & ~0xffULL) << 5)
35/* channel index is the index of 256B block */
36#define ADDR_OF_256B_BLOCK(channel_index) ((channel_index) << 8)
37/* offset in 256B block */
38#define OFFSET_IN_256B_BLOCK(addr) ((addr) & 0xffULL)
39
40#define LOOP_UMC_INST(umc_inst) for ((umc_inst) = 0; (umc_inst) < adev->umc.umc_inst_num; (umc_inst)++)
41#define LOOP_UMC_CH_INST(ch_inst) for ((ch_inst) = 0; (ch_inst) < adev->umc.channel_inst_num; (ch_inst)++)
42#define LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) LOOP_UMC_INST((umc_inst)) LOOP_UMC_CH_INST((ch_inst))
43
44#define LOOP_UMC_NODE_INST(node_inst) \
45 for ((node_inst) = 0; (node_inst) < adev->umc.node_inst_num; (node_inst)++)
46
47#define LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) \
48 LOOP_UMC_NODE_INST((node_inst)) LOOP_UMC_INST_AND_CH((umc_inst), (ch_inst))
49
50struct amdgpu_umc_ras {
51 struct amdgpu_ras_block_object ras_block;
52 void (*err_cnt_init)(struct amdgpu_device *adev);
53 bool (*query_ras_poison_mode)(struct amdgpu_device *adev);
54 void (*ecc_info_query_ras_error_count)(struct amdgpu_device *adev,
55 void *ras_error_status);
56 void (*ecc_info_query_ras_error_address)(struct amdgpu_device *adev,
57 void *ras_error_status);
58};
59
60struct amdgpu_umc_funcs {
61 void (*init_registers)(struct amdgpu_device *adev);
62};
63
64struct amdgpu_umc {
65 /* max error count in one ras query call */
66 uint32_t max_ras_err_cnt_per_query;
67 /* number of umc channel instance with memory map register access */
68 uint32_t channel_inst_num;
69 /* number of umc instance with memory map register access */
70 uint32_t umc_inst_num;
71
72 /*number of umc node instance with memory map register access*/
73 uint32_t node_inst_num;
74
75 /* UMC regiser per channel offset */
76 uint32_t channel_offs;
77 /* channel index table of interleaved memory */
78 const uint32_t *channel_idx_tbl;
79 struct ras_common_if *ras_if;
80
81 const struct amdgpu_umc_funcs *funcs;
82 struct amdgpu_umc_ras *ras;
83};
84
85int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block);
86int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset);
87int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev,
88 struct amdgpu_irq_src *source,
89 struct amdgpu_iv_entry *entry);
90void amdgpu_umc_fill_error_record(struct ras_err_data *err_data,
91 uint64_t err_addr,
92 uint64_t retired_page,
93 uint32_t channel_index,
94 uint32_t umc_inst);
95
96int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
97 void *ras_error_status,
98 struct amdgpu_iv_entry *entry);
99int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev,
100 uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst);
101#endif
1/*
2 * Copyright (C) 2019 Advanced Micro Devices, Inc.
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice shall be included
12 * in all copies or substantial portions of the Software.
13 *
14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
15 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
17 * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
18 * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
19 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
20 */
21#ifndef __AMDGPU_UMC_H__
22#define __AMDGPU_UMC_H__
23
24/* implement 64 bits REG operations via 32 bits interface */
25#define RREG64_UMC(reg) (RREG32(reg) | \
26 ((uint64_t)RREG32((reg) + 1) << 32))
27#define WREG64_UMC(reg, v) \
28 do { \
29 WREG32((reg), lower_32_bits(v)); \
30 WREG32((reg) + 1, upper_32_bits(v)); \
31 } while (0)
32
33/*
34 * void (*func)(struct amdgpu_device *adev, struct ras_err_data *err_data,
35 * uint32_t umc_reg_offset, uint32_t channel_index)
36 */
37#define amdgpu_umc_for_each_channel(func) \
38 struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; \
39 uint32_t umc_inst, channel_inst, umc_reg_offset, channel_index; \
40 for (umc_inst = 0; umc_inst < adev->umc.umc_inst_num; umc_inst++) { \
41 /* enable the index mode to query eror count per channel */ \
42 adev->umc.funcs->enable_umc_index_mode(adev, umc_inst); \
43 for (channel_inst = 0; \
44 channel_inst < adev->umc.channel_inst_num; \
45 channel_inst++) { \
46 /* calc the register offset according to channel instance */ \
47 umc_reg_offset = adev->umc.channel_offs * channel_inst; \
48 /* get channel index of interleaved memory */ \
49 channel_index = adev->umc.channel_idx_tbl[ \
50 umc_inst * adev->umc.channel_inst_num + channel_inst]; \
51 (func)(adev, err_data, umc_reg_offset, channel_index); \
52 } \
53 } \
54 adev->umc.funcs->disable_umc_index_mode(adev);
55
56struct amdgpu_umc_funcs {
57 void (*ras_init)(struct amdgpu_device *adev);
58 void (*query_ras_error_count)(struct amdgpu_device *adev,
59 void *ras_error_status);
60 void (*query_ras_error_address)(struct amdgpu_device *adev,
61 void *ras_error_status);
62 void (*enable_umc_index_mode)(struct amdgpu_device *adev,
63 uint32_t umc_instance);
64 void (*disable_umc_index_mode)(struct amdgpu_device *adev);
65};
66
67struct amdgpu_umc {
68 /* max error count in one ras query call */
69 uint32_t max_ras_err_cnt_per_query;
70 /* number of umc channel instance with memory map register access */
71 uint32_t channel_inst_num;
72 /* number of umc instance with memory map register access */
73 uint32_t umc_inst_num;
74 /* UMC regiser per channel offset */
75 uint32_t channel_offs;
76 /* channel index table of interleaved memory */
77 const uint32_t *channel_idx_tbl;
78
79 const struct amdgpu_umc_funcs *funcs;
80};
81
82#endif