xref: /linux/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h (revision 2c1ed907520c50326b8f604907a8478b27881a2e)
1 /*
2  * Copyright 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 in
12  * all copies or substantial portions of the Software.
13  *
14  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15  * 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) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
18  * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
19  * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
20  * OTHER DEALINGS IN THE SOFTWARE.
21  *
22  */
23 
24 #ifndef _AMDGPU_RAS_EEPROM_H
25 #define _AMDGPU_RAS_EEPROM_H
26 
27 #include <linux/i2c.h>
28 
29 #define RAS_TABLE_VER_V1           0x00010000
30 #define RAS_TABLE_VER_V2_1         0x00021000
31 
32 struct amdgpu_device;
33 
34 enum amdgpu_ras_gpu_health_status {
35 	GPU_HEALTH_USABLE = 0,
36 	GPU_RETIRED__ECC_REACH_THRESHOLD = 2,
37 };
38 
39 enum amdgpu_ras_eeprom_err_type {
40 	AMDGPU_RAS_EEPROM_ERR_NA,
41 	AMDGPU_RAS_EEPROM_ERR_RECOVERABLE,
42 	AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE,
43 	AMDGPU_RAS_EEPROM_ERR_COUNT,
44 };
45 
46 /*
47  * one UMC MCA address could map to multiply physical address (PA),
48  * such as 1:16, we use eeprom_table_record.address to store MCA
49  * address and use eeprom_table_record.retired_page to save PA.
50  *
51  * AMDGPU_RAS_EEPROM_REC_PA: one record store one PA
52  * AMDGPU_RAS_EEPROM_REC_MCA: one record store one MCA address
53  */
54 enum amdgpu_ras_eeprom_rec_type {
55 	AMDGPU_RAS_EEPROM_REC_PA,
56 	AMDGPU_RAS_EEPROM_REC_MCA,
57 };
58 
59 struct amdgpu_ras_eeprom_table_header {
60 	uint32_t header;
61 	uint32_t version;
62 	uint32_t first_rec_offset;
63 	uint32_t tbl_size;
64 	uint32_t checksum;
65 } __packed;
66 
67 struct amdgpu_ras_eeprom_table_ras_info {
68 	u8  rma_status;
69 	u8  health_percent;
70 	u16 ecc_page_threshold;
71 	u32 padding[64 - 1];
72 } __packed;
73 
74 struct amdgpu_ras_eeprom_control {
75 	struct amdgpu_ras_eeprom_table_header tbl_hdr;
76 
77 	struct amdgpu_ras_eeprom_table_ras_info tbl_rai;
78 
79 	/* Base I2C EEPPROM 19-bit memory address,
80 	 * where the table is located. For more information,
81 	 * see top of amdgpu_eeprom.c.
82 	 */
83 	u32 i2c_address;
84 
85 	/* The byte offset off of @i2c_address
86 	 * where the table header is found,
87 	 * and where the records start--always
88 	 * right after the header.
89 	 */
90 	u32 ras_header_offset;
91 	u32 ras_info_offset;
92 	u32 ras_record_offset;
93 
94 	/* Number of records in the table.
95 	 */
96 	u32 ras_num_recs;
97 
98 	/* the bad page number is ras_num_recs or
99 	 * ras_num_recs * umc.retire_unit
100 	 */
101 	u32 ras_num_bad_pages;
102 
103 	/* First record index to read, 0-based.
104 	 * Range is [0, num_recs-1]. This is
105 	 * an absolute index, starting right after
106 	 * the table header.
107 	 */
108 	u32 ras_fri;
109 
110 	/* Maximum possible number of records
111 	 * we could store, i.e. the maximum capacity
112 	 * of the table.
113 	 */
114 	u32 ras_max_record_count;
115 
116 	/* Protect table access via this mutex.
117 	 */
118 	struct mutex ras_tbl_mutex;
119 
120 	/* Record channel info which occurred bad pages
121 	 */
122 	u32 bad_channel_bitmap;
123 	enum amdgpu_ras_eeprom_rec_type rec_type;
124 };
125 
126 /*
127  * Represents single table record. Packed to be easily serialized into byte
128  * stream.
129  */
130 struct eeprom_table_record {
131 
132 	union {
133 		uint64_t address;
134 		uint64_t offset;
135 	};
136 
137 	uint64_t retired_page;
138 	uint64_t ts;
139 
140 	enum amdgpu_ras_eeprom_err_type err_type;
141 
142 	union {
143 		unsigned char bank;
144 		unsigned char cu;
145 	};
146 
147 	unsigned char mem_channel;
148 	unsigned char mcumc_id;
149 } __packed;
150 
151 int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control);
152 
153 int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control);
154 
155 bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev);
156 
157 int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
158 			   struct eeprom_table_record *records, const u32 num);
159 
160 int amdgpu_ras_eeprom_append(struct amdgpu_ras_eeprom_control *control,
161 			     struct eeprom_table_record *records, const u32 num);
162 
163 uint32_t amdgpu_ras_eeprom_max_record_count(struct amdgpu_ras_eeprom_control *control);
164 
165 void amdgpu_ras_debugfs_set_ret_size(struct amdgpu_ras_eeprom_control *control);
166 
167 int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control);
168 
169 extern const struct file_operations amdgpu_ras_debugfs_eeprom_size_ops;
170 extern const struct file_operations amdgpu_ras_debugfs_eeprom_table_ops;
171 
172 #endif // _AMDGPU_RAS_EEPROM_H
173