LXR linux/arch/x86/kernel/cpu/mcheck/mce-apei.c

   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/kernel.h>
  32#include <linux/acpi.h>
  33#include <linux/cper.h>
  34#include <acpi/apei.h>
  35#include <asm/mce.h>
  36
  37#include "mce-internal.h"
  38
  39void apei_mce_report_mem_error(int corrected, struct cper_sec_mem_err *mem_err)
  40{
  41        struct mce m;
  42
  43        /* Only corrected MC is reported */
  44        if (!corrected)
  45                return;
  46
  47        mce_setup(&m);
  48        m.bank = 1;
  49        /* Fake a memory read corrected error with unknown channel */
  50        m.status = MCI_STATUS_VAL | MCI_STATUS_EN | MCI_STATUS_ADDRV | 0x9f;
  51        m.addr = mem_err->physical_addr;
  52        mce_log(&m);
  53        mce_notify_irq();
  54}
  55EXPORT_SYMBOL_GPL(apei_mce_report_mem_error);
  56
  57#define CPER_CREATOR_MCE                                                \
  58        UUID_LE(0x75a574e3, 0x5052, 0x4b29, 0x8a, 0x8e, 0xbe, 0x2c,     \
  59                0x64, 0x90, 0xb8, 0x9d)
  60#define CPER_SECTION_TYPE_MCE                                           \
  61        UUID_LE(0xfe08ffbe, 0x95e4, 0x4be7, 0xbc, 0x73, 0x40, 0x96,     \
  62                0x04, 0x4a, 0x38, 0xfc)
  63
  64/*
  65 * CPER specification (in UEFI specification 2.3 appendix N) requires
  66 * byte-packed.
  67 */
  68struct cper_mce_record {
  69        struct cper_record_header hdr;
  70        struct cper_section_descriptor sec_hdr;
  71        struct mce mce;
  72} __packed;
  73
  74int apei_write_mce(struct mce *m)
  75{
  76        struct cper_mce_record rcd;
  77
  78        memset(&rcd, 0, sizeof(rcd));
  79        memcpy(rcd.hdr.signature, CPER_SIG_RECORD, CPER_SIG_SIZE);
  80        rcd.hdr.revision = CPER_RECORD_REV;
  81        rcd.hdr.signature_end = CPER_SIG_END;
  82        rcd.hdr.section_count = 1;
  83        rcd.hdr.error_severity = CPER_SEV_FATAL;
  84        /* timestamp, platform_id, partition_id are all invalid */
  85        rcd.hdr.validation_bits = 0;
  86        rcd.hdr.record_length = sizeof(rcd);
  87        rcd.hdr.creator_id = CPER_CREATOR_MCE;
  88        rcd.hdr.notification_type = CPER_NOTIFY_MCE;
  89        rcd.hdr.record_id = cper_next_record_id();
  90        rcd.hdr.flags = CPER_HW_ERROR_FLAGS_PREVERR;
  91
  92        rcd.sec_hdr.section_offset = (void *)&rcd.mce - (void *)&rcd;
  93        rcd.sec_hdr.section_length = sizeof(rcd.mce);
  94        rcd.sec_hdr.revision = CPER_SEC_REV;
  95        /* fru_id and fru_text is invalid */
  96        rcd.sec_hdr.validation_bits = 0;
  97        rcd.sec_hdr.flags = CPER_SEC_PRIMARY;
  98        rcd.sec_hdr.section_type = CPER_SECTION_TYPE_MCE;
  99        rcd.sec_hdr.section_severity = CPER_SEV_FATAL;
 100
 101        memcpy(&rcd.mce, m, sizeof(*m));
 102
 103        return erst_write(&rcd.hdr);
 104}
 105
 106ssize_t apei_read_mce(struct mce *m, u64 *record_id)
 107{
 108        struct cper_mce_record rcd;
 109        int rc, pos;
 110
 111        rc = erst_get_record_id_begin(&pos);
 112        if (rc)
 113                return rc;
 114retry:
 115        rc = erst_get_record_id_next(&pos, record_id);
 116        if (rc)
 117                goto out;
 118        /* no more record */
 119        if (*record_id == APEI_ERST_INVALID_RECORD_ID)
 120                goto out;
 121        rc = erst_read(*record_id, &rcd.hdr, sizeof(rcd));
 122        /* someone else has cleared the record, try next one */
 123        if (rc == -ENOENT)
 124                goto retry;
 125        else if (rc < 0)
 126                goto out;
 127        /* try to skip other type records in storage */
 128        else if (rc != sizeof(rcd) ||
 129                 uuid_le_cmp(rcd.hdr.creator_id, CPER_CREATOR_MCE))
 130                goto retry;
 131        memcpy(m, &rcd.mce, sizeof(*m));
 132        rc = sizeof(*m);
 133out:
 134        erst_get_record_id_end();
 135
 136        return rc;
 137}
 138
 139/* Check whether there is record in ERST */
 140int apei_check_mce(void)
 141{
 142        return erst_get_record_count();
 143}
 144
 145int apei_clear_mce(u64 record_id)
 146{
 147        return erst_clear(record_id);
 148}
 149