1 /*
2 * Bridge between MCE and APEI
3 *
4 * On some machine, corrected memory errors are reported via APEI
5 * generic hardware error source (GHES) instead of corrected Machine
6 * Check. These corrected memory errors can be reported to user space
7 * through /dev/mcelog via faking a corrected Machine Check, so that
8 * the error memory page can be offlined by /sbin/mcelog if the error
9 * count for one page is beyond the threshold.
10 *
11 * For fatal MCE, save MCE record into persistent storage via ERST, so
12 * that the MCE record can be logged after reboot via ERST.
13 *
14 * Copyright 2010 Intel Corp.
15 * Author: Huang Ying <ying.huang@intel.com>
16 *
17 * This program is free software; you can redistribute it and/or
18 * modify it under the terms of the GNU General Public License version
19 * 2 as published by the Free Software Foundation.
20 *
21 * This program is distributed in the hope that it will be useful,
22 * but WITHOUT ANY WARRANTY; without even the implied warranty of
23 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
24 * GNU General Public License for more details.
25 *
26 * You should have received a copy of the GNU General Public License
27 * along with this program; if not, write to the Free Software
28 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
29 */
30
31 #include <linux/export.h>
32 #include <linux/kernel.h>
33 #include <linux/acpi.h>
34 #include <linux/cper.h>
35 #include <acpi/apei.h>
36 #include <acpi/ghes.h>
37 #include <asm/mce.h>
38
39 #include "mce-internal.h"
40
apei_mce_report_mem_error(int severity,struct cper_sec_mem_err * mem_err)41 void apei_mce_report_mem_error(int severity, struct cper_sec_mem_err *mem_err)
42 {
43 struct mce m;
44
45 if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
46 return;
47
48 mce_setup(&m);
49 m.bank = 1;
50 /* Fake a memory read error with unknown channel */
51 m.status = MCI_STATUS_VAL | MCI_STATUS_EN | MCI_STATUS_ADDRV | 0x9f;
52
53 if (severity >= GHES_SEV_RECOVERABLE)
54 m.status |= MCI_STATUS_UC;
55 if (severity >= GHES_SEV_PANIC)
56 m.status |= MCI_STATUS_PCC;
57
58 m.addr = mem_err->physical_addr;
59 mce_log(&m);
60 }
61 EXPORT_SYMBOL_GPL(apei_mce_report_mem_error);
62
63 #define CPER_CREATOR_MCE \
64 UUID_LE(0x75a574e3, 0x5052, 0x4b29, 0x8a, 0x8e, 0xbe, 0x2c, \
65 0x64, 0x90, 0xb8, 0x9d)
66 #define CPER_SECTION_TYPE_MCE \
67 UUID_LE(0xfe08ffbe, 0x95e4, 0x4be7, 0xbc, 0x73, 0x40, 0x96, \
68 0x04, 0x4a, 0x38, 0xfc)
69
70 /*
71 * CPER specification (in UEFI specification 2.3 appendix N) requires
72 * byte-packed.
73 */
74 struct cper_mce_record {
75 struct cper_record_header hdr;
76 struct cper_section_descriptor sec_hdr;
77 struct mce mce;
78 } __packed;
79
apei_write_mce(struct mce * m)80 int apei_write_mce(struct mce *m)
81 {
82 struct cper_mce_record rcd;
83
84 memset(&rcd, 0, sizeof(rcd));
85 memcpy(rcd.hdr.signature, CPER_SIG_RECORD, CPER_SIG_SIZE);
86 rcd.hdr.revision = CPER_RECORD_REV;
87 rcd.hdr.signature_end = CPER_SIG_END;
88 rcd.hdr.section_count = 1;
89 rcd.hdr.error_severity = CPER_SEV_FATAL;
90 /* timestamp, platform_id, partition_id are all invalid */
91 rcd.hdr.validation_bits = 0;
92 rcd.hdr.record_length = sizeof(rcd);
93 rcd.hdr.creator_id = CPER_CREATOR_MCE;
94 rcd.hdr.notification_type = CPER_NOTIFY_MCE;
95 rcd.hdr.record_id = cper_next_record_id();
96 rcd.hdr.flags = CPER_HW_ERROR_FLAGS_PREVERR;
97
98 rcd.sec_hdr.section_offset = (void *)&rcd.mce - (void *)&rcd;
99 rcd.sec_hdr.section_length = sizeof(rcd.mce);
100 rcd.sec_hdr.revision = CPER_SEC_REV;
101 /* fru_id and fru_text is invalid */
102 rcd.sec_hdr.validation_bits = 0;
103 rcd.sec_hdr.flags = CPER_SEC_PRIMARY;
104 rcd.sec_hdr.section_type = CPER_SECTION_TYPE_MCE;
105 rcd.sec_hdr.section_severity = CPER_SEV_FATAL;
106
107 memcpy(&rcd.mce, m, sizeof(*m));
108
109 return erst_write(&rcd.hdr);
110 }
111
apei_read_mce(struct mce * m,u64 * record_id)112 ssize_t apei_read_mce(struct mce *m, u64 *record_id)
113 {
114 struct cper_mce_record rcd;
115 int rc, pos;
116
117 rc = erst_get_record_id_begin(&pos);
118 if (rc)
119 return rc;
120 retry:
121 rc = erst_get_record_id_next(&pos, record_id);
122 if (rc)
123 goto out;
124 /* no more record */
125 if (*record_id == APEI_ERST_INVALID_RECORD_ID)
126 goto out;
127 rc = erst_read(*record_id, &rcd.hdr, sizeof(rcd));
128 /* someone else has cleared the record, try next one */
129 if (rc == -ENOENT)
130 goto retry;
131 else if (rc < 0)
132 goto out;
133 /* try to skip other type records in storage */
134 else if (rc != sizeof(rcd) ||
135 uuid_le_cmp(rcd.hdr.creator_id, CPER_CREATOR_MCE))
136 goto retry;
137 memcpy(m, &rcd.mce, sizeof(*m));
138 rc = sizeof(*m);
139 out:
140 erst_get_record_id_end();
141
142 return rc;
143 }
144
145 /* Check whether there is record in ERST */
apei_check_mce(void)146 int apei_check_mce(void)
147 {
148 return erst_get_record_count();
149 }
150
apei_clear_mce(u64 record_id)151 int apei_clear_mce(u64 record_id)
152 {
153 return erst_clear(record_id);
154 }
155