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 <asm/mce.h>
37
38 #include "mce-internal.h"
39
apei_mce_report_mem_error(int corrected,struct cper_sec_mem_err * mem_err)40 void apei_mce_report_mem_error(int corrected, struct cper_sec_mem_err *mem_err)
41 {
42 struct mce m;
43
44 /* Only corrected MC is reported */
45 if (!corrected)
46 return;
47
48 mce_setup(&m);
49 m.bank = 1;
50 /* Fake a memory read corrected error with unknown channel */
51 m.status = MCI_STATUS_VAL | MCI_STATUS_EN | MCI_STATUS_ADDRV | 0x9f;
52 m.addr = mem_err->physical_addr;
53 mce_log(&m);
54 mce_notify_irq();
55 }
56 EXPORT_SYMBOL_GPL(apei_mce_report_mem_error);
57
58 #define CPER_CREATOR_MCE \
59 UUID_LE(0x75a574e3, 0x5052, 0x4b29, 0x8a, 0x8e, 0xbe, 0x2c, \
60 0x64, 0x90, 0xb8, 0x9d)
61 #define CPER_SECTION_TYPE_MCE \
62 UUID_LE(0xfe08ffbe, 0x95e4, 0x4be7, 0xbc, 0x73, 0x40, 0x96, \
63 0x04, 0x4a, 0x38, 0xfc)
64
65 /*
66 * CPER specification (in UEFI specification 2.3 appendix N) requires
67 * byte-packed.
68 */
69 struct cper_mce_record {
70 struct cper_record_header hdr;
71 struct cper_section_descriptor sec_hdr;
72 struct mce mce;
73 } __packed;
74
apei_write_mce(struct mce * m)75 int apei_write_mce(struct mce *m)
76 {
77 struct cper_mce_record rcd;
78
79 memset(&rcd, 0, sizeof(rcd));
80 memcpy(rcd.hdr.signature, CPER_SIG_RECORD, CPER_SIG_SIZE);
81 rcd.hdr.revision = CPER_RECORD_REV;
82 rcd.hdr.signature_end = CPER_SIG_END;
83 rcd.hdr.section_count = 1;
84 rcd.hdr.error_severity = CPER_SEV_FATAL;
85 /* timestamp, platform_id, partition_id are all invalid */
86 rcd.hdr.validation_bits = 0;
87 rcd.hdr.record_length = sizeof(rcd);
88 rcd.hdr.creator_id = CPER_CREATOR_MCE;
89 rcd.hdr.notification_type = CPER_NOTIFY_MCE;
90 rcd.hdr.record_id = cper_next_record_id();
91 rcd.hdr.flags = CPER_HW_ERROR_FLAGS_PREVERR;
92
93 rcd.sec_hdr.section_offset = (void *)&rcd.mce - (void *)&rcd;
94 rcd.sec_hdr.section_length = sizeof(rcd.mce);
95 rcd.sec_hdr.revision = CPER_SEC_REV;
96 /* fru_id and fru_text is invalid */
97 rcd.sec_hdr.validation_bits = 0;
98 rcd.sec_hdr.flags = CPER_SEC_PRIMARY;
99 rcd.sec_hdr.section_type = CPER_SECTION_TYPE_MCE;
100 rcd.sec_hdr.section_severity = CPER_SEV_FATAL;
101
102 memcpy(&rcd.mce, m, sizeof(*m));
103
104 return erst_write(&rcd.hdr);
105 }
106
apei_read_mce(struct mce * m,u64 * record_id)107 ssize_t apei_read_mce(struct mce *m, u64 *record_id)
108 {
109 struct cper_mce_record rcd;
110 int rc, pos;
111
112 rc = erst_get_record_id_begin(&pos);
113 if (rc)
114 return rc;
115 retry:
116 rc = erst_get_record_id_next(&pos, record_id);
117 if (rc)
118 goto out;
119 /* no more record */
120 if (*record_id == APEI_ERST_INVALID_RECORD_ID)
121 goto out;
122 rc = erst_read(*record_id, &rcd.hdr, sizeof(rcd));
123 /* someone else has cleared the record, try next one */
124 if (rc == -ENOENT)
125 goto retry;
126 else if (rc < 0)
127 goto out;
128 /* try to skip other type records in storage */
129 else if (rc != sizeof(rcd) ||
130 uuid_le_cmp(rcd.hdr.creator_id, CPER_CREATOR_MCE))
131 goto retry;
132 memcpy(m, &rcd.mce, sizeof(*m));
133 rc = sizeof(*m);
134 out:
135 erst_get_record_id_end();
136
137 return rc;
138 }
139
140 /* Check whether there is record in ERST */
apei_check_mce(void)141 int apei_check_mce(void)
142 {
143 return erst_get_record_count();
144 }
145
apei_clear_mce(u64 record_id)146 int apei_clear_mce(u64 record_id)
147 {
148 return erst_clear(record_id);
149 }
150