* [PATCH v2 01/10] rasdaemon: Support cpu fault isolation for corrected errors
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
2022-10-03 16:17 ` [PATCH v2 02/10] rasdaemon: Support cpu fault isolation for recoverable errors shiju.jose
` (8 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shengwei Luo <luoshengwei@huawei.com>
When the corrected errors exceed the set limit in cycle, try to
offline the related cpu core.
Signed-off-by: Shengwei Luo <luoshengwei@huawei.com>
Signed-off-by: Junchong Pan <panjunchong@hisilicon.com>
Signed-off-by: Lei Feng <fenglei47@h-partners.com>
Signed-off-by: Xiaofei Tan <tanxiaofei@huawei.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
Makefile.am | 6 +-
configure.ac | 11 ++
misc/rasdaemon.env | 17 ++
queue.c | 119 ++++++++++++++
queue.h | 39 +++++
ras-arm-handler.c | 97 +++++++++++
ras-arm-handler.h | 18 ++
ras-cpu-isolation.c | 388 ++++++++++++++++++++++++++++++++++++++++++++
ras-cpu-isolation.h | 67 ++++++++
ras-events.c | 9 +-
10 files changed, 769 insertions(+), 2 deletions(-)
create mode 100644 queue.c
create mode 100644 queue.h
create mode 100644 ras-cpu-isolation.c
create mode 100644 ras-cpu-isolation.h
diff --git a/Makefile.am b/Makefile.am
index a322b9a..36e7d4e 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -69,13 +69,17 @@ endif
if WITH_AMP_NS_DECODE
rasdaemon_SOURCES += non-standard-ampere.c
endif
+if WITH_CPU_FAULT_ISOLATION
+ rasdaemon_SOURCES += ras-cpu-isolation.c queue.c
+endif
rasdaemon_LDADD = -lpthread $(SQLITE3_LIBS) libtrace/libtrace.a
include_HEADERS = config.h ras-events.h ras-logger.h ras-mc-handler.h \
ras-aer-handler.h ras-mce-handler.h ras-record.h bitfield.h ras-report.h \
ras-extlog-handler.h ras-arm-handler.h ras-non-standard-handler.h \
ras-devlink-handler.h ras-diskerror-handler.h rbtree.h ras-page-isolation.h \
- non-standard-hisilicon.h non-standard-ampere.h ras-memory-failure-handler.h
+ non-standard-hisilicon.h non-standard-ampere.h ras-memory-failure-handler.h \
+ ras-cpu-isolation.h queue.h
# This rule can't be called with more than one Makefile job (like make -j8)
# I can't figure out a way to fix that
diff --git a/configure.ac b/configure.ac
index a77991f..e0ed751 100644
--- a/configure.ac
+++ b/configure.ac
@@ -161,6 +161,16 @@ AS_IF([test "x$enable_amp_ns_decode" = "xyes" || test "x$enable_all" == "xyes"],
AM_CONDITIONAL([WITH_AMP_NS_DECODE], [test x$enable_amp_ns_decode = xyes || test x$enable_all == xyes])
AM_COND_IF([WITH_AMP_NS_DECODE], [USE_AMP_NS_DECODE="yes"], [USE_AMP_NS_DECODE="no"])
+AC_ARG_ENABLE([cpu_fault_isolation],
+ AS_HELP_STRING([--enable-cpu-fault-isolation], [enable cpu online fault isolation]))
+
+AS_IF([test "x$enable_cpu_fault_isolation" = "xyes" || test "x$enable_all" == "xyes"], [
+ AC_DEFINE(HAVE_CPU_FAULT_ISOLATION,1,"have cpu online fault isolation")
+ AC_SUBST([WITH_CPU_FAULT_ISOLATION])
+])
+AM_CONDITIONAL([WITH_CPU_FAULT_ISOLATION], [test x$enable_cpu_fault_isolation = xyes || test x$enable_all == xyes])
+AM_COND_IF([WITH_CPU_FAULT_ISOLATION], [USE_CPU_FAULT_ISOLATION="yes"], [USE_CPU_FAULT_ISOLATION="no"])
+
test "$sysconfdir" = '${prefix}/etc' && sysconfdir=/etc
CFLAGS="$CFLAGS -Wall -Wmissing-prototypes -Wstrict-prototypes"
@@ -201,4 +211,5 @@ compile time options summary
Memory Failure : $USE_MEMORY_FAILURE
Memory CE PFA : $USE_MEMORY_CE_PFA
AMP RAS errors : $USE_AMP_NS_DECODE
+ CPU fault isolation : $USE_CPU_FAULT_ISOLATION
EOF
diff --git a/misc/rasdaemon.env b/misc/rasdaemon.env
index 12fd766..7cb18e8 100644
--- a/misc/rasdaemon.env
+++ b/misc/rasdaemon.env
@@ -27,3 +27,20 @@ PAGE_CE_THRESHOLD="50"
# soft-then-hard First try to soft offline, then try hard offlining.
# Note: default offline choice is "soft".
PAGE_CE_ACTION="soft"
+
+# CPU Online Fault Isolation
+# Whether to enable cpu online fault isolation (yes|no).
+CPU_ISOLATION_ENABLE="no"
+# Specify the threshold of CE numbers.
+#
+# Format:
+# [0-9]+[unit]
+#
+# Supported units:
+# CPU_CE_THRESHOLD: no unit
+# CPU_ISOLATION_CYCLE: D|d (day), H|h (hour), M|m (minute), S|s (second), default is in second
+CPU_CE_THRESHOLD="18"
+CPU_ISOLATION_CYCLE="24h"
+
+# Prevent excessive isolation from causing an avalanche effect
+CPU_ISOLATION_LIMIT="10"
\ No newline at end of file
diff --git a/queue.c b/queue.c
new file mode 100644
index 0000000..65b6fb8
--- /dev/null
+++ b/queue.c
@@ -0,0 +1,119 @@
+/*
+ * Copyright (c) Huawei Technologies Co., Ltd. 2021-2021. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include "queue.h"
+#include "ras-logger.h"
+
+int is_empty(struct link_queue *queue)
+{
+ if (queue)
+ return queue->size == 0;
+
+ return 1;
+}
+
+struct link_queue *init_queue(void)
+{
+ struct link_queue *queue = NULL;
+
+ queue = (struct link_queue *)malloc(sizeof(struct link_queue));
+ if (queue == NULL) {
+ log(TERM, LOG_ERR, "Failed to allocate memory for queue.\n");
+ return NULL;
+ }
+
+ queue->size = 0;
+ queue->head = NULL;
+ queue->tail = NULL;
+
+ return queue;
+}
+
+void clear_queue(struct link_queue *queue)
+{
+ if (queue == NULL)
+ return;
+
+ struct queue_node *node = queue->head;
+ struct queue_node *tmp = NULL;
+
+ while (node != NULL) {
+ tmp = node;
+ node = node->next;
+ free(tmp);
+ }
+
+ queue->head = NULL;
+ queue->tail = NULL;
+ queue->size = 0;
+}
+
+void free_queue(struct link_queue *queue)
+{
+ clear_queue(queue);
+
+ if (queue)
+ free(queue);
+}
+
+/* It should be guranteed that the param is not NULL */
+void push(struct link_queue *queue, struct queue_node *node)
+{
+ /* there is no element in the queue */
+ if (queue->head == NULL)
+ queue->head = node;
+ else
+ queue->tail->next = node;
+
+ queue->tail = node;
+ (queue->size)++;
+}
+
+int pop(struct link_queue *queue)
+{
+ struct queue_node *tmp = NULL;
+
+ if (queue == NULL || is_empty(queue))
+ return -1;
+
+ tmp = queue->head;
+ queue->head = queue->head->next;
+ free(tmp);
+ (queue->size)--;
+
+ return 0;
+}
+
+struct queue_node *front(struct link_queue *queue)
+{
+ if (queue == NULL)
+ return NULL;
+
+ return queue->head;
+}
+
+struct queue_node *node_create(time_t time, unsigned int value)
+{
+ struct queue_node *node = NULL;
+
+ node = (struct queue_node *)malloc(sizeof(struct queue_node));
+ if (node != NULL) {
+ node->time = time;
+ node->value = value;
+ node->next = NULL;
+ }
+
+ return node;
+}
diff --git a/queue.h b/queue.h
new file mode 100644
index 0000000..5459f40
--- /dev/null
+++ b/queue.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright (c) Huawei Technologies Co., Ltd. 2021-2021. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#ifndef __RAS_QUEUE_H
+#define __RAS_QUEUE_H
+
+struct queue_node {
+ time_t time;
+ unsigned int value;
+ struct queue_node *next;
+};
+
+struct link_queue {
+ struct queue_node *head;
+ struct queue_node *tail;
+ int size;
+};
+
+int is_empty(struct link_queue *queue);
+struct link_queue *init_queue(void);
+void clear_queue(struct link_queue *queue);
+void free_queue(struct link_queue *queue);
+void push(struct link_queue *queue, struct queue_node *node);
+int pop(struct link_queue *queue);
+struct queue_node *front(struct link_queue *queue);
+struct queue_node *node_create(time_t time, unsigned int value);
+
+#endif
diff --git a/ras-arm-handler.c b/ras-arm-handler.c
index 1149dc6..9c7a3c3 100644
--- a/ras-arm-handler.c
+++ b/ras-arm-handler.c
@@ -22,6 +22,10 @@
#include "ras-report.h"
#include "ras-non-standard-handler.h"
#include "non-standard-ampere.h"
+#include "ras-cpu-isolation.h"
+
+#define ARM_ERR_VALID_ERROR_COUNT BIT(0)
+#define ARM_ERR_VALID_FLAGS BIT(1)
void display_raw_data(struct trace_seq *s,
const uint8_t *buf,
@@ -42,6 +46,93 @@ void display_raw_data(struct trace_seq *s,
}
}
+#ifdef HAVE_CPU_FAULT_ISOLATION
+static int count_errors(struct ras_arm_event *ev)
+{
+ struct ras_arm_err_info *err_info;
+ int num_pei;
+ int err_info_size = sizeof(struct ras_arm_err_info);
+ int num = 0;
+ int i;
+ int error_count;
+
+ if (ev->pei_len % err_info_size != 0) {
+ log(TERM, LOG_ERR,
+ "The event data does not match to the ARM Processor Error Information Structure\n");
+ return num;
+ }
+ num_pei = ev->pei_len / err_info_size;
+ err_info = (struct ras_arm_err_info *)(ev->pei_error);
+
+ for (i = 0; i < num_pei; ++i) {
+ error_count = 1;
+ if (err_info->validation_bits & ARM_ERR_VALID_ERROR_COUNT) {
+ /*
+ * The value of this field is defined as follows:
+ * 0: Single Error
+ * 1: Multiple Errors
+ * 2-65535: Error Count
+ */
+ error_count = err_info->multiple_error + 1;
+ }
+
+ num += error_count;
+ err_info += 1;
+ }
+ log(TERM, LOG_INFO, "%d error in cpu core catched\n", num);
+ return num;
+}
+
+static int ras_handle_cpu_error(struct trace_seq *s,
+ struct pevent_record *record,
+ struct event_format *event,
+ struct ras_arm_event *ev, time_t now)
+{
+ unsigned long long val;
+ int cpu;
+ char *severity;
+ struct error_info err_info;
+
+ if (pevent_get_field_val(s, event, "cpu", record, &val, 1) < 0)
+ return -1;
+ cpu = val;
+ trace_seq_printf(s, "\n cpu: %d", cpu);
+
+ /* record cpu error */
+ if (pevent_get_field_val(s, event, "sev", record, &val, 1) < 0)
+ return -1;
+ /* refer to UEFI_2_9 specification chapter N2.2 Table N-5 */
+ switch (val) {
+ case GHES_SEV_NO:
+ severity = "Informational";
+ break;
+ case GHES_SEV_CORRECTED:
+ severity = "Corrected";
+ break;
+ case GHES_SEV_RECOVERABLE:
+ severity = "Recoverable";
+ break;
+ default:
+ case GHES_SEV_PANIC:
+ severity = "Fatal";
+ }
+ trace_seq_printf(s, "\n severity: %s", severity);
+
+ if (val == GHES_SEV_CORRECTED) {
+ int nums = count_errors(ev);
+
+ if (nums > 0) {
+ err_info.nums = nums;
+ err_info.time = now;
+ err_info.err_type = val;
+ ras_record_cpu_error(&err_info, cpu);
+ }
+ }
+
+ return 0;
+}
+#endif
+
int ras_arm_event_handler(struct trace_seq *s,
struct pevent_record *record,
struct event_format *event, void *context)
@@ -52,6 +143,7 @@ int ras_arm_event_handler(struct trace_seq *s,
struct tm *tm;
struct ras_arm_event ev;
int len = 0;
+
memset(&ev, 0, sizeof(ev));
/*
@@ -139,6 +231,11 @@ int ras_arm_event_handler(struct trace_seq *s,
display_raw_data(s, ev.vsei_error, ev.oem_len);
#endif
+#ifdef HAVE_CPU_FAULT_ISOLATION
+ if (ras_handle_cpu_error(s, record, event, &ev, now) < 0)
+ return -1;
+#endif
+
/* Insert data into the SGBD */
#ifdef HAVE_SQLITE3
ras_store_arm_record(ras, &ev);
diff --git a/ras-arm-handler.h b/ras-arm-handler.h
index 563a2d3..52813e7 100644
--- a/ras-arm-handler.h
+++ b/ras-arm-handler.h
@@ -17,6 +17,24 @@
#include "ras-events.h"
#include "libtrace/event-parse.h"
+/*
+ * ARM Processor Error Information Structure, According to
+ * UEFI_2_9 specification chapter N2.4.4.
+ */
+#pragma pack(1)
+struct ras_arm_err_info {
+ uint8_t version;
+ uint8_t length;
+ uint16_t validation_bits;
+ uint8_t type;
+ uint16_t multiple_error;
+ uint8_t flags;
+ uint64_t error_info;
+ uint64_t virt_fault_addr;
+ uint64_t physical_fault_addr;
+};
+#pragma pack()
+
int ras_arm_event_handler(struct trace_seq *s,
struct pevent_record *record,
struct event_format *event, void *context);
diff --git a/ras-cpu-isolation.c b/ras-cpu-isolation.c
new file mode 100644
index 0000000..1694a08
--- /dev/null
+++ b/ras-cpu-isolation.c
@@ -0,0 +1,388 @@
+/*
+ * Copyright (c) Huawei Technologies Co., Ltd. 2021-2021. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <fcntl.h>
+#include <errno.h>
+#include <unistd.h>
+#include <limits.h>
+#include <ctype.h>
+#include "ras-logger.h"
+#include "ras-cpu-isolation.h"
+
+#define SECOND_OF_MON (30 * 24 * 60 * 60)
+#define SECOND_OF_DAY (24 * 60 * 60)
+#define SECOND_OF_HOU (60 * 60)
+#define SECOND_OF_MIN (60)
+
+#define LIMIT_OF_CPU_THRESHOLD 10000
+#define INIT_OF_CPU_THRESHOLD 18
+#define DEC_CHECK 10
+#define LAST_BIT_OF_UL 5
+
+static struct cpu_info *cpu_infos;
+static unsigned int ncores;
+static unsigned int enabled = 1;
+static const char *cpu_path_format = "/sys/devices/system/cpu/cpu%d/online";
+
+static const struct param normal_units[] = {
+ {"", 1},
+ {}
+};
+
+static const struct param cycle_units[] = {
+ {"d", SECOND_OF_DAY},
+ {"h", SECOND_OF_HOU},
+ {"m", SECOND_OF_MIN},
+ {"s", 1},
+ {}
+};
+
+static struct isolation_param threshold = {
+ .name = "CPU_CE_THRESHOLD",
+ .units = normal_units,
+ .value = INIT_OF_CPU_THRESHOLD,
+ .limit = LIMIT_OF_CPU_THRESHOLD
+};
+
+static struct isolation_param cpu_limit = {
+ .name = "CPU_ISOLATION_LIMIT",
+ .units = normal_units
+};
+
+static struct isolation_param cycle = {
+ .name = "CPU_ISOLATION_CYCLE",
+ .units = cycle_units,
+ .value = SECOND_OF_DAY,
+ .limit = SECOND_OF_MON
+};
+
+static const char * const cpu_state[] = {
+ [CPU_OFFLINE] = "offline",
+ [CPU_ONLINE] = "online",
+ [CPU_OFFLINE_FAILED] = "offline-failed",
+ [CPU_UNKNOWN] = "unknown"
+};
+
+static int open_sys_file(unsigned int cpu, int __oflag, const char *format)
+{
+ int fd;
+ char path[PATH_MAX] = "";
+ char real_path[PATH_MAX] = "";
+
+ snprintf(path, sizeof(path), format, cpu);
+ if (strlen(path) > PATH_MAX || realpath(path, real_path) == NULL) {
+ log(TERM, LOG_ERR, "[%s]:open file: %s failed\n", __func__, path);
+ return -1;
+ }
+ fd = open(real_path, __oflag);
+ if (fd == -1) {
+ log(TERM, LOG_ERR, "[%s]:open file: %s failed\n", __func__, real_path);
+ return -1;
+ }
+
+ return fd;
+}
+
+static int get_cpu_status(unsigned int cpu)
+{
+ int fd, num;
+ char buf[2] = "";
+
+ fd = open_sys_file(cpu, O_RDONLY, cpu_path_format);
+ if (fd == -1)
+ return CPU_UNKNOWN;
+
+ if (read(fd, buf, 1) <= 0 || sscanf(buf, "%d", &num) != 1)
+ num = CPU_UNKNOWN;
+
+ close(fd);
+
+ return (num < 0 || num > CPU_UNKNOWN) ? CPU_UNKNOWN : num;
+}
+
+static int init_cpu_info(unsigned int cpus)
+{
+ ncores = cpus;
+ cpu_infos = (struct cpu_info *)malloc(sizeof(*cpu_infos) * cpus);
+ if (!cpu_infos) {
+ log(TERM, LOG_ERR,
+ "Failed to allocate memory for cpu infos in %s.\n", __func__);
+ return -1;
+ }
+
+ for (unsigned int i = 0; i < cpus; ++i) {
+ cpu_infos[i].ce_nums = 0;
+ cpu_infos[i].state = get_cpu_status(i);
+ cpu_infos[i].ce_queue = init_queue();
+
+ if (cpu_infos[i].ce_queue == NULL) {
+ log(TERM, LOG_ERR,
+ "Failed to allocate memory for cpu ce queue in %s.\n", __func__);
+ return -1;
+ }
+ }
+ /* set limit of offlined cpu limit according to number of cpu */
+ cpu_limit.limit = cpus - 1;
+ cpu_limit.value = 0;
+
+ return 0;
+}
+
+static void check_config(struct isolation_param *config)
+{
+ if (config->value > config->limit) {
+ log(TERM, LOG_WARNING, "Value: %lu exceed limit: %lu, set to limit\n",
+ config->value, config->limit);
+ config->value = config->limit;
+ }
+}
+
+static int parse_ul_config(struct isolation_param *config, char *env, unsigned long *value)
+{
+ char *unit = NULL;
+ int env_size, has_unit = 0;
+
+ if (!env || strlen(env) == 0)
+ return -1;
+
+ env_size = strlen(env);
+ unit = env + env_size - 1;
+
+ if (isalpha(*unit)) {
+ has_unit = 1;
+ env_size--;
+ if (env_size <= 0)
+ return -1;
+ }
+
+ for (int i = 0; i < env_size; ++i) {
+ if (isdigit(env[i])) {
+ if (*value > ULONG_MAX / DEC_CHECK ||
+ (*value == ULONG_MAX / DEC_CHECK && env[i] - '0' > LAST_BIT_OF_UL)) {
+ log(TERM, LOG_ERR, "%s is out of range: %lu\n", env, ULONG_MAX);
+ return -1;
+ }
+ *value = DEC_CHECK * (*value) + (env[i] - '0');
+ } else
+ return -1;
+ }
+
+ if (!has_unit)
+ return 0;
+
+ for (const struct param *units = config->units; units->name; units++) {
+ /* value character and unit character are both valid */
+ if (!strcasecmp(unit, units->name)) {
+ if (*value > (ULONG_MAX / units->value)) {
+ log(TERM, LOG_ERR,
+ "%s is out of range: %lu\n", env, ULONG_MAX);
+ return -1;
+ }
+ *value = (*value) * units->value;
+ return 0;
+ }
+ }
+ log(TERM, LOG_ERR, "Invalid unit %s\n", unit);
+ return -1;
+}
+
+static void init_config(struct isolation_param *config)
+{
+ char *env = getenv(config->name);
+ unsigned long value = 0;
+
+ if (parse_ul_config(config, env, &value) < 0) {
+ log(TERM, LOG_ERR, "Invalid %s: %s! Use default value %lu.\n",
+ config->name, env, config->value);
+ return;
+ }
+
+ config->value = value;
+ check_config(config);
+}
+
+static int check_config_status(void)
+{
+ char *env = getenv("CPU_ISOLATION_ENABLE");
+
+ if (env == NULL || strcasecmp(env, "yes"))
+ return -1;
+
+ return 0;
+}
+
+void ras_cpu_isolation_init(unsigned int cpus)
+{
+ if (init_cpu_info(cpus) < 0 || check_config_status() < 0) {
+ enabled = 0;
+ log(TERM, LOG_WARNING, "Cpu fault isolation is disabled\n");
+ return;
+ }
+
+ log(TERM, LOG_INFO, "Cpu fault isolation is enabled\n");
+ init_config(&threshold);
+ init_config(&cpu_limit);
+ init_config(&cycle);
+}
+
+void cpu_infos_free(void)
+{
+ if (cpu_infos) {
+ for (int i = 0; i < ncores; ++i)
+ free_queue(cpu_infos[i].ce_queue);
+
+ free(cpu_infos);
+ }
+}
+
+static int do_cpu_offline(unsigned int cpu)
+{
+ int fd, rc;
+ char buf[2] = "";
+
+ cpu_infos[cpu].state = CPU_OFFLINE_FAILED;
+ fd = open_sys_file(cpu, O_RDWR, cpu_path_format);
+ if (fd == -1)
+ return HANDLE_FAILED;
+
+ strcpy(buf, "0");
+ rc = write(fd, buf, strlen(buf));
+ if (rc < 0) {
+ log(TERM, LOG_ERR, "cpu%u offline failed, errno:%d\n", cpu, errno);
+ close(fd);
+ return HANDLE_FAILED;
+ }
+
+ close(fd);
+ /* check wthether the cpu is isolated successfully */
+ cpu_infos[cpu].state = get_cpu_status(cpu);
+
+ if (cpu_infos[cpu].state == CPU_OFFLINE)
+ return HANDLE_SUCCEED;
+
+ return HANDLE_FAILED;
+}
+
+static int do_ce_handler(unsigned int cpu)
+{
+ struct link_queue *queue = cpu_infos[cpu].ce_queue;
+ unsigned int tmp;
+ /*
+ * Since we just count all error numbers in setted cycle, we store the time
+ * and error numbers from current event to the queue, then everytime we
+ * calculate the period from beginning time to ending time, if the period
+ * exceeds setted cycle, we pop the beginning time and error until the period
+ * from new beginning time to ending time is less than cycle.
+ */
+ while (queue->head && queue->tail && queue->tail->time - queue->head->time > cycle.value) {
+ tmp = queue->head->value;
+ if (pop(queue) == 0)
+ cpu_infos[cpu].ce_nums -= tmp;
+ }
+ log(TERM, LOG_INFO,
+ "Current number of Corrected Errors in cpu%d in the cycle is %lu\n",
+ cpu, cpu_infos[cpu].ce_nums);
+
+ if (cpu_infos[cpu].ce_nums >= threshold.value) {
+ log(TERM, LOG_INFO,
+ "Corrected Errors exceeded threshold %lu, try to offline cpu%u\n",
+ threshold.value, cpu);
+ return do_cpu_offline(cpu);
+ }
+ return HANDLE_NOTHING;
+}
+
+static int error_handler(unsigned int cpu, struct error_info *err_info)
+{
+ int ret = HANDLE_NOTHING;
+
+ switch (err_info->err_type) {
+ case CE:
+ ret = do_ce_handler(cpu);
+ break;
+ default:
+ break;
+ }
+
+ return ret;
+}
+
+static void record_error_info(unsigned int cpu, struct error_info *err_info)
+{
+ switch (err_info->err_type) {
+ case CE:
+ {
+ struct queue_node *node = node_create(err_info->time, err_info->nums);
+
+ if (node == NULL) {
+ log(TERM, LOG_ERR, "Fail to allocate memory for queue node\n");
+ return;
+ }
+ push(cpu_infos[cpu].ce_queue, node);
+ cpu_infos[cpu].ce_nums += err_info->nums;
+ break;
+ }
+ default:
+ break;
+ }
+}
+
+void ras_record_cpu_error(struct error_info *err_info, int cpu)
+{
+ int ret;
+
+ if (enabled == 0)
+ return;
+
+ if (cpu >= ncores || cpu < 0) {
+ log(TERM, LOG_ERR,
+ "The current cpu %d has exceed the total number of cpu:%u\n", cpu, ncores);
+ return;
+ }
+
+ log(TERM, LOG_INFO, "Handling error on cpu%d\n", cpu);
+ cpu_infos[cpu].state = get_cpu_status(cpu);
+
+ if (cpu_infos[cpu].state != CPU_ONLINE) {
+ log(TERM, LOG_INFO, "Cpu%d is not online or unknown, ignore\n", cpu);
+ return;
+ }
+
+ record_error_info(cpu, err_info);
+ /*
+ * Since user may change cpu state, we get current offlined
+ * cpu numbers every recording time.
+ */
+ if (ncores - sysconf(_SC_NPROCESSORS_ONLN) >= cpu_limit.value) {
+ log(TERM, LOG_WARNING,
+ "Offlined cpus have exceeded limit: %lu, choose to do nothing\n",
+ cpu_limit.value);
+ return;
+ }
+
+ ret = error_handler(cpu, err_info);
+ if (ret == HANDLE_NOTHING)
+ log(TERM, LOG_WARNING, "Doing nothing in the cpu%d\n", cpu);
+ else if (ret == HANDLE_SUCCEED) {
+ log(TERM, LOG_INFO, "Offline cpu%d succeed, the state is %s\n",
+ cpu, cpu_state[cpu_infos[cpu].state]);
+ clear_queue(cpu_infos[cpu].ce_queue);
+ cpu_infos[cpu].ce_nums = 0;
+ } else
+ log(TERM, LOG_WARNING, "Offline cpu%d fail, the state is %s\n",
+ cpu, cpu_state[cpu_infos[cpu].state]);
+}
diff --git a/ras-cpu-isolation.h b/ras-cpu-isolation.h
new file mode 100644
index 0000000..35b5225
--- /dev/null
+++ b/ras-cpu-isolation.h
@@ -0,0 +1,67 @@
+/*
+ * Copyright (c) Huawei Technologies Co., Ltd. 2021-2021. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#ifndef __RAS_CPU_ISOLATION_H
+#define __RAS_CPU_ISOLATION_H
+
+#include "queue.h"
+
+#define MAX_BUF_LEN 1024
+
+struct param {
+ char *name;
+ unsigned long value;
+};
+
+struct isolation_param {
+ char *name;
+ const struct param *units;
+ unsigned long value;
+ unsigned long limit;
+};
+
+enum cpu_state {
+ CPU_OFFLINE,
+ CPU_ONLINE,
+ CPU_OFFLINE_FAILED,
+ CPU_UNKNOWN,
+};
+
+enum error_handle_result {
+ HANDLE_FAILED = -1,
+ HANDLE_SUCCEED,
+ HANDLE_NOTHING,
+};
+
+enum error_type {
+ CE = 1
+};
+
+struct cpu_info {
+ unsigned long ce_nums;
+ struct link_queue *ce_queue;
+ enum cpu_state state;
+};
+
+struct error_info {
+ unsigned long nums;
+ time_t time;
+ enum error_type err_type;
+};
+
+void ras_cpu_isolation_init(unsigned int cpus);
+void ras_record_cpu_error(struct error_info *err_info, int cpu);
+void cpu_infos_free(void);
+
+#endif
diff --git a/ras-events.c b/ras-events.c
index 39cab20..beda655 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -42,6 +42,7 @@
#include "ras-record.h"
#include "ras-logger.h"
#include "ras-page-isolation.h"
+#include "ras-cpu-isolation.h"
/*
* Polling time, if read() doesn't block. Currently, trace_pipe_raw never
@@ -856,6 +857,10 @@ int handle_ras_events(int record_events)
cpus = get_num_cpus(ras);
+#ifdef HAVE_CPU_FAULT_ISOLATION
+ ras_cpu_isolation_init(cpus);
+#endif
+
#ifdef HAVE_MCE
rc = register_mce_handler(ras, cpus);
if (rc)
@@ -982,6 +987,8 @@ err:
}
free(ras);
}
-
+#ifdef HAVE_CPU_FAULT_ISOLATION
+ cpu_infos_free();
+#endif
return rc;
}
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH v2 02/10] rasdaemon: Support cpu fault isolation for recoverable errors
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
2022-10-03 16:17 ` [PATCH v2 01/10] rasdaemon: Support cpu fault isolation for corrected errors shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
2022-10-03 16:17 ` [PATCH v2 03/10] rasdaemon: Modify recording Hisilicon common error data shiju.jose
` (7 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shengwei Luo <luoshengwei@huawei.com>
When the recoverable errors in cpu core occurred, try to offline
the related cpu core.
Signed-off-by: Shengwei Luo <luoshengwei@huawei.com>
Signed-off-by: Junchong Pan <panjunchong@hisilicon.com>
Signed-off-by: Lei Feng <fenglei47@h-partners.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
ras-arm-handler.c | 22 +++++++++++++++++++---
ras-cpu-isolation.c | 17 +++++++++++++++++
ras-cpu-isolation.h | 4 +++-
3 files changed, 39 insertions(+), 4 deletions(-)
diff --git a/ras-arm-handler.c b/ras-arm-handler.c
index 9c7a3c3..a0dfc51 100644
--- a/ras-arm-handler.c
+++ b/ras-arm-handler.c
@@ -26,6 +26,7 @@
#define ARM_ERR_VALID_ERROR_COUNT BIT(0)
#define ARM_ERR_VALID_FLAGS BIT(1)
+#define BIT2 2
void display_raw_data(struct trace_seq *s,
const uint8_t *buf,
@@ -47,7 +48,20 @@ void display_raw_data(struct trace_seq *s,
}
#ifdef HAVE_CPU_FAULT_ISOLATION
-static int count_errors(struct ras_arm_event *ev)
+static int is_core_failure(struct ras_arm_err_info *err_info)
+{
+ if (err_info->validation_bits & ARM_ERR_VALID_FLAGS) {
+ /*
+ * core failure:
+ * Bit 0\1\3: (at lease 1)
+ * Bit 2: 0
+ */
+ return (err_info->flags & 0xf) && !(err_info->flags & (0x1 << BIT2));
+ }
+ return 0;
+}
+
+static int count_errors(struct ras_arm_event *ev, int sev)
{
struct ras_arm_err_info *err_info;
int num_pei;
@@ -75,6 +89,8 @@ static int count_errors(struct ras_arm_event *ev)
*/
error_count = err_info->multiple_error + 1;
}
+ if (sev == GHES_SEV_RECOVERABLE && !is_core_failure(err_info))
+ error_count = 0;
num += error_count;
err_info += 1;
@@ -118,8 +134,8 @@ static int ras_handle_cpu_error(struct trace_seq *s,
}
trace_seq_printf(s, "\n severity: %s", severity);
- if (val == GHES_SEV_CORRECTED) {
- int nums = count_errors(ev);
+ if (val == GHES_SEV_CORRECTED || val == GHES_SEV_RECOVERABLE) {
+ int nums = count_errors(ev, val);
if (nums > 0) {
err_info.nums = nums;
diff --git a/ras-cpu-isolation.c b/ras-cpu-isolation.c
index 1694a08..90633fd 100644
--- a/ras-cpu-isolation.c
+++ b/ras-cpu-isolation.c
@@ -126,6 +126,7 @@ static int init_cpu_info(unsigned int cpus)
for (unsigned int i = 0; i < cpus; ++i) {
cpu_infos[i].ce_nums = 0;
+ cpu_infos[i].uce_nums = 0;
cpu_infos[i].state = get_cpu_status(i);
cpu_infos[i].ce_queue = init_queue();
@@ -306,6 +307,15 @@ static int do_ce_handler(unsigned int cpu)
return HANDLE_NOTHING;
}
+static int do_uce_handler(unsigned int cpu)
+{
+ if (cpu_infos[cpu].uce_nums > 0) {
+ log(TERM, LOG_INFO, "Uncorrected Errors occurred, try to offline cpu%u\n", cpu);
+ return do_cpu_offline(cpu);
+ }
+ return HANDLE_NOTHING;
+}
+
static int error_handler(unsigned int cpu, struct error_info *err_info)
{
int ret = HANDLE_NOTHING;
@@ -314,6 +324,9 @@ static int error_handler(unsigned int cpu, struct error_info *err_info)
case CE:
ret = do_ce_handler(cpu);
break;
+ case UCE:
+ ret = do_uce_handler(cpu);
+ break;
default:
break;
}
@@ -336,6 +349,9 @@ static void record_error_info(unsigned int cpu, struct error_info *err_info)
cpu_infos[cpu].ce_nums += err_info->nums;
break;
}
+ case UCE:
+ cpu_infos[cpu].uce_nums++;
+ break;
default:
break;
}
@@ -382,6 +398,7 @@ void ras_record_cpu_error(struct error_info *err_info, int cpu)
cpu, cpu_state[cpu_infos[cpu].state]);
clear_queue(cpu_infos[cpu].ce_queue);
cpu_infos[cpu].ce_nums = 0;
+ cpu_infos[cpu].uce_nums = 0;
} else
log(TERM, LOG_WARNING, "Offline cpu%d fail, the state is %s\n",
cpu, cpu_state[cpu_infos[cpu].state]);
diff --git a/ras-cpu-isolation.h b/ras-cpu-isolation.h
index 35b5225..5682106 100644
--- a/ras-cpu-isolation.h
+++ b/ras-cpu-isolation.h
@@ -45,10 +45,12 @@ enum error_handle_result {
};
enum error_type {
- CE = 1
+ CE = 1,
+ UCE
};
struct cpu_info {
+ unsigned long uce_nums;
unsigned long ce_nums;
struct link_queue *ce_queue;
enum cpu_state state;
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH v2 03/10] rasdaemon: Modify recording Hisilicon common error data
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
2022-10-03 16:17 ` [PATCH v2 01/10] rasdaemon: Support cpu fault isolation for corrected errors shiju.jose
2022-10-03 16:17 ` [PATCH v2 02/10] rasdaemon: Support cpu fault isolation for recoverable errors shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
2022-10-03 16:17 ` [PATCH v2 04/10] rasdaemon: ras-mc-ctl: Modify error statistics for HiSilicon KunPeng9xx common errors shiju.jose
` (6 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
The error statistics for the Hisilicon common
error need to do based on module, error severity etc.
Modify recording Hisilicon common error data as separate fields
in the sql db table instead of the combined single field.
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
non-standard-hisilicon.c | 126 ++++++++++++++++++++++++++++++++-------
1 file changed, 104 insertions(+), 22 deletions(-)
diff --git a/non-standard-hisilicon.c b/non-standard-hisilicon.c
index 1432163..d1e1774 100644
--- a/non-standard-hisilicon.c
+++ b/non-standard-hisilicon.c
@@ -17,6 +17,7 @@
#include "non-standard-hisilicon.h"
#define HISI_BUF_LEN 2048
+#define HISI_PCIE_INFO_BUF_LEN 256
struct hisi_common_error_section {
uint32_t val_bits;
@@ -63,12 +64,25 @@ enum {
enum {
HISI_COMMON_FIELD_ID,
HISI_COMMON_FIELD_TIMESTAMP,
- HISI_COMMON_FIELD_ERR_INFO,
+ HISI_COMMON_FIELD_VERSION,
+ HISI_COMMON_FIELD_SOC_ID,
+ HISI_COMMON_FIELD_SOCKET_ID,
+ HISI_COMMON_FIELD_TOTEM_ID,
+ HISI_COMMON_FIELD_NIMBUS_ID,
+ HISI_COMMON_FIELD_SUB_SYSTEM_ID,
+ HISI_COMMON_FIELD_MODULE_ID,
+ HISI_COMMON_FIELD_SUB_MODULE_ID,
+ HISI_COMMON_FIELD_CORE_ID,
+ HISI_COMMON_FIELD_PORT_ID,
+ HISI_COMMON_FIELD_ERR_TYPE,
+ HISI_COMMON_FIELD_PCIE_INFO,
+ HISI_COMMON_FIELD_ERR_SEVERITY,
HISI_COMMON_FIELD_REGS_DUMP,
};
struct hisi_event {
char error_msg[HISI_BUF_LEN];
+ char pcie_info[HISI_PCIE_INFO_BUF_LEN];
char reg_msg[HISI_BUF_LEN];
};
@@ -132,14 +146,26 @@ int step_vendor_data_tab(struct ras_ns_ev_decoder *ev_decoder, const char *name)
#ifdef HAVE_SQLITE3
static const struct db_fields hisi_common_section_fields[] = {
- { .name = "id", .type = "INTEGER PRIMARY KEY" },
- { .name = "timestamp", .type = "TEXT" },
- { .name = "err_info", .type = "TEXT" },
+ { .name = "id", .type = "INTEGER PRIMARY KEY" },
+ { .name = "timestamp", .type = "TEXT" },
+ { .name = "version", .type = "INTEGER" },
+ { .name = "soc_id", .type = "INTEGER" },
+ { .name = "socket_id", .type = "INTEGER" },
+ { .name = "totem_id", .type = "INTEGER" },
+ { .name = "nimbus_id", .type = "INTEGER" },
+ { .name = "sub_system_id", .type = "INTEGER" },
+ { .name = "module_id", .type = "TEXT" },
+ { .name = "sub_module_id", .type = "INTEGER" },
+ { .name = "core_id", .type = "INTEGER" },
+ { .name = "port_id", .type = "INTEGER" },
+ { .name = "err_type", .type = "INTEGER" },
+ { .name = "pcie_info", .type = "TEXT" },
+ { .name = "err_severity", .type = "TEXT" },
{ .name = "regs_dump", .type = "TEXT" },
};
static const struct db_table_descriptor hisi_common_section_tab = {
- .name = "hisi_common_section",
+ .name = "hisi_common_section_v2",
.fields = hisi_common_section_fields,
.num_fields = ARRAY_SIZE(hisi_common_section_fields),
};
@@ -199,12 +225,20 @@ static const char* get_soc_desc(uint8_t soc_id)
return soc_desc[soc_id];
}
-static void decode_module(struct hisi_event *event, uint8_t module_id)
+static void decode_module(struct ras_ns_ev_decoder *ev_decoder,
+ struct hisi_event *event, uint8_t module_id)
{
- if (module_id >= sizeof(module_name)/sizeof(char *))
+ if (module_id >= sizeof(module_name)/sizeof(char *)) {
HISI_SNPRINTF(event->error_msg, "module=unknown(id=%hhu) ", module_id);
- else
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_TEXT,
+ HISI_COMMON_FIELD_MODULE_ID,
+ 0, "unknown");
+ } else {
HISI_SNPRINTF(event->error_msg, "module=%s ", module_name[module_id]);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_TEXT,
+ HISI_COMMON_FIELD_MODULE_ID,
+ 0, module_name[module_id]);
+ }
}
static void decode_hisi_common_section_hdr(struct ras_ns_ev_decoder *ev_decoder,
@@ -212,43 +246,93 @@ static void decode_hisi_common_section_hdr(struct ras_ns_ev_decoder *ev_decoder,
struct hisi_event *event)
{
HISI_SNPRINTF(event->error_msg, "[ table_version=%hhu", err->version);
- if (err->val_bits & BIT(HISI_COMMON_VALID_SOC_ID))
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_VERSION,
+ err->version, NULL);
+ if (err->val_bits & BIT(HISI_COMMON_VALID_SOC_ID)) {
HISI_SNPRINTF(event->error_msg, "soc=%s", get_soc_desc(err->soc_id));
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_SOC_ID,
+ err->soc_id, NULL);
+ }
- if (err->val_bits & BIT(HISI_COMMON_VALID_SOCKET_ID))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_SOCKET_ID)) {
HISI_SNPRINTF(event->error_msg, "socket_id=%hhu", err->socket_id);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_SOCKET_ID,
+ err->socket_id, NULL);
+ }
- if (err->val_bits & BIT(HISI_COMMON_VALID_TOTEM_ID))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_TOTEM_ID)) {
HISI_SNPRINTF(event->error_msg, "totem_id=%hhu", err->totem_id);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_TOTEM_ID,
+ err->totem_id, NULL);
+ }
- if (err->val_bits & BIT(HISI_COMMON_VALID_NIMBUS_ID))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_NIMBUS_ID)) {
HISI_SNPRINTF(event->error_msg, "nimbus_id=%hhu", err->nimbus_id);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_NIMBUS_ID,
+ err->nimbus_id, NULL);
+ }
- if (err->val_bits & BIT(HISI_COMMON_VALID_SUBSYSTEM_ID))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_SUBSYSTEM_ID)) {
HISI_SNPRINTF(event->error_msg, "subsystem_id=%hhu", err->subsystem_id);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_SUB_SYSTEM_ID,
+ err->subsystem_id, NULL);
+ }
if (err->val_bits & BIT(HISI_COMMON_VALID_MODULE_ID))
- decode_module(event, err->module_id);
+ decode_module(ev_decoder, event, err->module_id);
- if (err->val_bits & BIT(HISI_COMMON_VALID_SUBMODULE_ID))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_SUBMODULE_ID)) {
HISI_SNPRINTF(event->error_msg, "submodule_id=%hhu", err->submodule_id);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_SUB_MODULE_ID,
+ err->submodule_id, NULL);
+ }
- if (err->val_bits & BIT(HISI_COMMON_VALID_CORE_ID))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_CORE_ID)) {
HISI_SNPRINTF(event->error_msg, "core_id=%hhu", err->core_id);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_CORE_ID,
+ err->core_id, NULL);
+ }
- if (err->val_bits & BIT(HISI_COMMON_VALID_PORT_ID))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_PORT_ID)) {
HISI_SNPRINTF(event->error_msg, "port_id=%hhu", err->port_id);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_PORT_ID,
+ err->port_id, NULL);
+ }
- if (err->val_bits & BIT(HISI_COMMON_VALID_ERR_TYPE))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_ERR_TYPE)) {
HISI_SNPRINTF(event->error_msg, "err_type=%hu", err->err_type);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_INT,
+ HISI_COMMON_FIELD_ERR_TYPE,
+ err->err_type, NULL);
+ }
- if (err->val_bits & BIT(HISI_COMMON_VALID_PCIE_INFO))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_PCIE_INFO)) {
HISI_SNPRINTF(event->error_msg, "pcie_device_id=%04x:%02x:%02x.%x",
err->pcie_info.segment, err->pcie_info.bus,
err->pcie_info.device, err->pcie_info.function);
+ HISI_SNPRINTF(event->pcie_info, "%04x:%02x:%02x.%x",
+ err->pcie_info.segment, err->pcie_info.bus,
+ err->pcie_info.device, err->pcie_info.function);
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_TEXT,
+ HISI_COMMON_FIELD_PCIE_INFO,
+ 0, event->pcie_info);
+ }
- if (err->val_bits & BIT(HISI_COMMON_VALID_ERR_SEVERITY))
+ if (err->val_bits & BIT(HISI_COMMON_VALID_ERR_SEVERITY)) {
HISI_SNPRINTF(event->error_msg, "err_severity=%s", err_severity(err->err_severity));
+ record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_TEXT,
+ HISI_COMMON_FIELD_ERR_SEVERITY,
+ 0, err_severity(err->err_severity));
+ }
HISI_SNPRINTF(event->error_msg, "]");
}
@@ -293,8 +377,6 @@ static int decode_hisi_common_section(struct ras_events *ras,
record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_TEXT,
HISI_COMMON_FIELD_TIMESTAMP,
0, event->timestamp);
- record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_TEXT,
- HISI_COMMON_FIELD_ERR_INFO, 0, hevent.error_msg);
record_vendor_data(ev_decoder, HISI_OEM_DATA_TYPE_TEXT,
HISI_COMMON_FIELD_REGS_DUMP, 0, hevent.reg_msg);
step_vendor_data_tab(ev_decoder, "hisi_common_section_tab");
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH v2 04/10] rasdaemon: ras-mc-ctl: Modify error statistics for HiSilicon KunPeng9xx common errors
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
` (2 preceding siblings ...)
2022-10-03 16:17 ` [PATCH v2 03/10] rasdaemon: Modify recording Hisilicon common error data shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
2022-10-03 16:17 ` [PATCH v2 05/10] rasdaemon: ras-mc-ctl: Reformat error info of the HiSilicon Kunpeng920 shiju.jose
` (5 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Modify the error statistics for the HiSilicon KunPeng9xx platforms common errors
to display the statistics and error info based on the module and the error severity.
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
util/ras-mc-ctl.in | 40 +++++++++++++++++++++++++++++-----------
1 file changed, 29 insertions(+), 11 deletions(-)
diff --git a/util/ras-mc-ctl.in b/util/ras-mc-ctl.in
index b22dd60..08eb287 100755
--- a/util/ras-mc-ctl.in
+++ b/util/ras-mc-ctl.in
@@ -1537,7 +1537,7 @@ sub vendor_errors_summary
require DBI;
my ($num_args, $platform_id);
my ($query, $query_handle, $count, $out);
- my ($module_id, $sub_module_id, $err_severity, $err_sev, $err_info);
+ my ($module_id, $sub_module_id, $err_severity, $err_sev);
$num_args = $#ARGV + 1;
$platform_id = 0;
@@ -1614,13 +1614,18 @@ sub vendor_errors_summary
# HiSilicon Kunpeng9xx common errors
if ($platform_id eq HISILICON_KUNPENG_9XX) {
- $query = "select err_info, count(*) from hisi_common_section";
+ $query = "select err_severity, module_id, count(*) from hisi_common_section_v2 group by err_severity, module_id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
- $query_handle->bind_columns(\($err_info, $count));
+ $query_handle->bind_columns(\($err_severity, $module_id, $count));
$out = "";
+ $err_sev = "";
while($query_handle->fetch()) {
- $out .= "\terrors: $count\n";
+ if ($err_severity ne $err_sev) {
+ $out .= "$err_severity errors:\n";
+ $err_sev = $err_severity;
+ }
+ $out .= "\t$module_id: $count\n";
}
if ($out ne "") {
print "HiSilicon Kunpeng9xx common error events summary:\n$out\n";
@@ -1638,8 +1643,8 @@ sub vendor_errors
require DBI;
my ($num_args, $platform_id);
my ($query, $query_handle, $id, $timestamp, $out);
- my ($version, $soc_id, $socket_id, $nimbus_id, $core_id, $port_id);
- my ($module_id, $sub_module_id, $err_severity, $err_type, $err_info, $regs);
+ my ($version, $soc_id, $socket_id, $totem_id, $nimbus_id, $sub_system_id, $core_id, $port_id);
+ my ($module_id, $sub_module_id, $err_severity, $err_type, $pcie_info, $regs);
$num_args = $#ARGV + 1;
$platform_id = 0;
@@ -1727,15 +1732,28 @@ sub vendor_errors
# HiSilicon Kunpeng9xx common errors
if ($platform_id eq HISILICON_KUNPENG_9XX) {
- $query = "select id, timestamp, err_info, regs_dump from hisi_common_section order by id";
+ $query = "select id, timestamp, version, soc_id, socket_id, totem_id, nimbus_id, sub_system_id, module_id, sub_module_id, core_id, port_id, err_type, pcie_info, err_severity, regs_dump from hisi_common_section_v2 order by id, module_id, err_severity";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
- $query_handle->bind_columns(\($id, $timestamp, $err_info, $regs));
+ $query_handle->bind_columns(\($id, $timestamp, $version, $soc_id, $socket_id, $totem_id, $nimbus_id, $sub_system_id, $module_id, $sub_module_id, $core_id, $port_id, $err_type, $pcie_info, $err_severity, $regs));
$out = "";
while($query_handle->fetch()) {
- $out .= "$id. $timestamp ";
- $out .= "Error Info:$err_info \n" if ($err_info);
- $out .= "Error Registers: $regs\n\n" if ($regs);
+ $out .= "$id. $timestamp Error Info: ";
+ $out .= "version=$version, ";
+ $out .= "soc_id=$soc_id, " if ($soc_id);
+ $out .= "socket_id=$socket_id, " if ($socket_id);
+ $out .= "totem_id=$totem_id, " if ($totem_id);
+ $out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
+ $out .= "sub_system_id=$sub_system_id, " if ($sub_system_id);
+ $out .= "module_id=$module_id, " if ($module_id);
+ $out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
+ $out .= "core_id=$core_id, " if ($core_id);
+ $out .= "port_id=$port_id, " if ($port_id);
+ $out .= "err_type=$err_type, " if ($err_type);
+ $out .= "pcie_info=$pcie_info, " if ($pcie_info);
+ $out .= "err_severity=$err_severity, " if ($err_severity);
+ $out .= "Error Registers: $regs" if ($regs);
+ $out .= "\n\n";
}
if ($out ne "") {
print "HiSilicon Kunpeng9xx common error events:\n$out\n";
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH v2 05/10] rasdaemon: ras-mc-ctl: Reformat error info of the HiSilicon Kunpeng920
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
` (3 preceding siblings ...)
2022-10-03 16:17 ` [PATCH v2 04/10] rasdaemon: ras-mc-ctl: Modify error statistics for HiSilicon KunPeng9xx common errors shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
2022-10-03 16:17 ` [PATCH v2 06/10] rasdaemon: ras-mc-ctl: Add printing usage if necessary parameters are not passed for the vendor-error options shiju.jose
` (4 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Reformat the code to display the error info of HiSilicon Kunpeng920.
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
util/ras-mc-ctl.in | 15 +++++++++------
1 file changed, 9 insertions(+), 6 deletions(-)
diff --git a/util/ras-mc-ctl.in b/util/ras-mc-ctl.in
index 08eb287..8755b6f 100755
--- a/util/ras-mc-ctl.in
+++ b/util/ras-mc-ctl.in
@@ -1671,8 +1671,9 @@ sub vendor_errors
$out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
$out .= "module_id=$module_id, " if ($module_id);
$out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
- $out .= "err_severity=$err_severity, \n" if ($err_severity);
- $out .= "Error Registers: $regs\n\n" if ($regs);
+ $out .= "err_severity=$err_severity, " if ($err_severity);
+ $out .= "Error Registers: $regs " if ($regs);
+ $out .= "\n\n";
}
if ($out ne "") {
print "HiSilicon Kunpeng920 OEM type1 error events:\n$out\n";
@@ -1694,8 +1695,9 @@ sub vendor_errors
$out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
$out .= "module_id=$module_id, " if ($module_id);
$out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
- $out .= "err_severity=$err_severity, \n" if ($err_severity);
- $out .= "Error Registers: $regs\n\n" if ($regs);
+ $out .= "err_severity=$err_severity, " if ($err_severity);
+ $out .= "Error Registers: $regs " if ($regs);
+ $out .= "\n\n";
}
if ($out ne "") {
print "HiSilicon Kunpeng920 OEM type2 error events:\n$out\n";
@@ -1719,8 +1721,9 @@ sub vendor_errors
$out .= "core_id=$core_id, " if ($core_id);
$out .= "port_id=$port_id, " if ($port_id);
$out .= "err_severity=$err_severity, " if ($err_severity);
- $out .= "err_type=$err_type, \n" if ($err_type);
- $out .= "Error Registers: $regs\n\n" if ($regs);
+ $out .= "err_type=$err_type, " if ($err_type);
+ $out .= "Error Registers: $regs " if ($regs);
+ $out .= "\n\n";
}
if ($out ne "") {
print "HiSilicon Kunpeng920 PCIe controller error events:\n$out\n";
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH v2 06/10] rasdaemon: ras-mc-ctl: Add printing usage if necessary parameters are not passed for the vendor-error options
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
` (4 preceding siblings ...)
2022-10-03 16:17 ` [PATCH v2 05/10] rasdaemon: ras-mc-ctl: Reformat error info of the HiSilicon Kunpeng920 shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
2022-10-03 16:17 ` [PATCH v2 07/10] rasdaemon: ras-mc-ctl: Add support to display the HiSilicon vendor errors for a specified module shiju.jose
` (3 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Add printing usage if necessary parameters are not passed
for the vendor-errors options.
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
util/ras-mc-ctl.in | 2 ++
1 file changed, 2 insertions(+)
diff --git a/util/ras-mc-ctl.in b/util/ras-mc-ctl.in
index 8755b6f..959ea6b 100755
--- a/util/ras-mc-ctl.in
+++ b/util/ras-mc-ctl.in
@@ -1544,6 +1544,7 @@ sub vendor_errors_summary
if ($num_args ne 0) {
$platform_id = $ARGV[0];
} else {
+ usage(1);
return;
}
@@ -1651,6 +1652,7 @@ sub vendor_errors
if ($num_args ne 0) {
$platform_id = $ARGV[0];
} else {
+ usage(1);
return;
}
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH v2 07/10] rasdaemon: ras-mc-ctl: Add support to display the HiSilicon vendor errors for a specified module
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
` (5 preceding siblings ...)
2022-10-03 16:17 ` [PATCH v2 06/10] rasdaemon: ras-mc-ctl: Add printing usage if necessary parameters are not passed for the vendor-error options shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
2022-10-03 16:17 ` [PATCH v2 08/10] rasdaemon: ras-mc-ctl: Relocate reading and display Kunpeng920 errors to under Kunpeng9xx shiju.jose
` (2 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Add support to display the HiSilicon vendor errors for a specified module.
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
util/ras-mc-ctl.in | 145 +++++++++++++++++++++++++++------------------
1 file changed, 87 insertions(+), 58 deletions(-)
diff --git a/util/ras-mc-ctl.in b/util/ras-mc-ctl.in
index 959ea6b..296eb87 100755
--- a/util/ras-mc-ctl.in
+++ b/util/ras-mc-ctl.in
@@ -96,8 +96,9 @@ Usage: $prog [OPTIONS...]
--errors Shows the errors stored at the error database.
--error-count Shows the corrected and uncorrected error counts using sysfs.
--vendor-errors-summary <platform-id> Presents a summary of the vendor-specific logged errors.
- --vendor-errors <platform-id> Shows the vendor-specific errors stored in the error database.
- --vendor-platforms Shows the supported platforms with platform-ids for the vendor-specific errors.
+ --vendor-errors <platform-id> Shows the vendor-specific errors stored in the error database.
+ --vendor-errors <platform-id> <module-name> Shows the vendor-specific errors for a specific module stored in the error database.
+ --vendor-platforms List the supported platforms with platform-ids for the vendor-specific errors.
--help This help message.
EOF
@@ -1535,12 +1536,14 @@ use constant {
sub vendor_errors_summary
{
require DBI;
- my ($num_args, $platform_id);
+ my ($num_args, $platform_id, $found_platform);
my ($query, $query_handle, $count, $out);
my ($module_id, $sub_module_id, $err_severity, $err_sev);
$num_args = $#ARGV + 1;
$platform_id = 0;
+ $found_platform = 0;
+
if ($num_args ne 0) {
$platform_id = $ARGV[0];
} else {
@@ -1552,6 +1555,7 @@ sub vendor_errors_summary
# HiSilicon Kunpeng920 errors
if ($platform_id eq HISILICON_KUNPENG_920) {
+ $found_platform = 1;
$query = "select err_severity, module_id, count(*) from hip08_oem_type1_event_v2 group by err_severity, module_id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
@@ -1615,6 +1619,7 @@ sub vendor_errors_summary
# HiSilicon Kunpeng9xx common errors
if ($platform_id eq HISILICON_KUNPENG_9XX) {
+ $found_platform = 1;
$query = "select err_severity, module_id, count(*) from hisi_common_section_v2 group by err_severity, module_id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
@@ -1636,21 +1641,31 @@ sub vendor_errors_summary
$query_handle->finish;
}
+ if ($platform_id && !($found_platform)) {
+ print "Platform ID $platform_id is not valid\n";
+ }
+
undef($dbh);
}
sub vendor_errors
{
require DBI;
- my ($num_args, $platform_id);
+ my ($num_args, $platform_id, $found_platform, $module, $found_module);
my ($query, $query_handle, $id, $timestamp, $out);
my ($version, $soc_id, $socket_id, $totem_id, $nimbus_id, $sub_system_id, $core_id, $port_id);
my ($module_id, $sub_module_id, $err_severity, $err_type, $pcie_info, $regs);
$num_args = $#ARGV + 1;
$platform_id = 0;
+ $found_platform = 0;
+ $module = 0;
+ $found_module = 0;
if ($num_args ne 0) {
$platform_id = $ARGV[0];
+ if ($num_args gt 1) {
+ $module = $ARGV[1];
+ }
} else {
usage(1);
return;
@@ -1660,27 +1675,29 @@ sub vendor_errors
# HiSilicon Kunpeng920 errors
if ($platform_id eq HISILICON_KUNPENG_920) {
+ $found_platform = 1;
$query = "select id, timestamp, version, soc_id, socket_id, nimbus_id, module_id, sub_module_id, err_severity, regs_dump from hip08_oem_type1_event_v2 order by id, module_id, err_severity";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $version, $soc_id, $socket_id, $nimbus_id, $module_id, $sub_module_id, $err_severity, $regs));
$out = "";
while($query_handle->fetch()) {
- $out .= "$id. $timestamp Error Info: ";
- $out .= "version=$version, ";
- $out .= "soc_id=$soc_id, " if ($soc_id);
- $out .= "socket_id=$socket_id, " if ($socket_id);
- $out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
- $out .= "module_id=$module_id, " if ($module_id);
- $out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
- $out .= "err_severity=$err_severity, " if ($err_severity);
- $out .= "Error Registers: $regs " if ($regs);
- $out .= "\n\n";
+ if ($module eq 0 || ($module_id && uc($module) eq uc($module_id))) {
+ $out .= "$id. $timestamp Error Info: ";
+ $out .= "version=$version, ";
+ $out .= "soc_id=$soc_id, " if ($soc_id);
+ $out .= "socket_id=$socket_id, " if ($socket_id);
+ $out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
+ $out .= "module_id=$module_id, " if ($module_id);
+ $out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
+ $out .= "err_severity=$err_severity, " if ($err_severity);
+ $out .= "Error Registers: $regs " if ($regs);
+ $out .= "\n\n";
+ $found_module = 1;
+ }
}
if ($out ne "") {
print "HiSilicon Kunpeng920 OEM type1 error events:\n$out\n";
- } else {
- print "No HiSilicon Kunpeng920 OEM type1 errors.\n";
}
$query_handle->finish;
@@ -1690,21 +1707,22 @@ sub vendor_errors
$query_handle->bind_columns(\($id, $timestamp, $version, $soc_id, $socket_id, $nimbus_id, $module_id, $sub_module_id, $err_severity, $regs));
$out = "";
while($query_handle->fetch()) {
- $out .= "$id. $timestamp Error Info: ";
- $out .= "version=$version, ";
- $out .= "soc_id=$soc_id, " if ($soc_id);
- $out .= "socket_id=$socket_id, " if ($socket_id);
- $out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
- $out .= "module_id=$module_id, " if ($module_id);
- $out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
- $out .= "err_severity=$err_severity, " if ($err_severity);
- $out .= "Error Registers: $regs " if ($regs);
- $out .= "\n\n";
+ if ($module eq 0 || ($module_id && uc($module) eq uc($module_id))) {
+ $out .= "$id. $timestamp Error Info: ";
+ $out .= "version=$version, ";
+ $out .= "soc_id=$soc_id, " if ($soc_id);
+ $out .= "socket_id=$socket_id, " if ($socket_id);
+ $out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
+ $out .= "module_id=$module_id, " if ($module_id);
+ $out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
+ $out .= "err_severity=$err_severity, " if ($err_severity);
+ $out .= "Error Registers: $regs " if ($regs);
+ $out .= "\n\n";
+ $found_module = 1;
+ }
}
if ($out ne "") {
print "HiSilicon Kunpeng920 OEM type2 error events:\n$out\n";
- } else {
- print "No HiSilicon Kunpeng920 OEM type2 errors.\n";
}
$query_handle->finish;
@@ -1714,51 +1732,56 @@ sub vendor_errors
$query_handle->bind_columns(\($id, $timestamp, $version, $soc_id, $socket_id, $nimbus_id, $sub_module_id, $core_id, $port_id, $err_severity, $err_type, $regs));
$out = "";
while($query_handle->fetch()) {
- $out .= "$id. $timestamp Error Info: ";
- $out .= "version=$version, ";
- $out .= "soc_id=$soc_id, " if ($soc_id);
- $out .= "socket_id=$socket_id, " if ($socket_id);
- $out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
- $out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
- $out .= "core_id=$core_id, " if ($core_id);
- $out .= "port_id=$port_id, " if ($port_id);
- $out .= "err_severity=$err_severity, " if ($err_severity);
- $out .= "err_type=$err_type, " if ($err_type);
- $out .= "Error Registers: $regs " if ($regs);
- $out .= "\n\n";
+ if ($module eq 0 || ($sub_module_id && uc($module) eq uc($sub_module_id))) {
+ $out .= "$id. $timestamp Error Info: ";
+ $out .= "version=$version, ";
+ $out .= "soc_id=$soc_id, " if ($soc_id);
+ $out .= "socket_id=$socket_id, " if ($socket_id);
+ $out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
+ $out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
+ $out .= "core_id=$core_id, " if ($core_id);
+ $out .= "port_id=$port_id, " if ($port_id);
+ $out .= "err_severity=$err_severity, " if ($err_severity);
+ $out .= "err_type=$err_type, " if ($err_type);
+ $out .= "Error Registers: $regs " if ($regs);
+ $out .= "\n\n";
+ $found_module = 1;
+ }
}
if ($out ne "") {
print "HiSilicon Kunpeng920 PCIe controller error events:\n$out\n";
- } else {
- print "No HiSilicon Kunpeng920 PCIe controller errors.\n";
}
$query_handle->finish;
}
# HiSilicon Kunpeng9xx common errors
if ($platform_id eq HISILICON_KUNPENG_9XX) {
+ $found_platform = 1;
$query = "select id, timestamp, version, soc_id, socket_id, totem_id, nimbus_id, sub_system_id, module_id, sub_module_id, core_id, port_id, err_type, pcie_info, err_severity, regs_dump from hisi_common_section_v2 order by id, module_id, err_severity";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $version, $soc_id, $socket_id, $totem_id, $nimbus_id, $sub_system_id, $module_id, $sub_module_id, $core_id, $port_id, $err_type, $pcie_info, $err_severity, $regs));
$out = "";
while($query_handle->fetch()) {
- $out .= "$id. $timestamp Error Info: ";
- $out .= "version=$version, ";
- $out .= "soc_id=$soc_id, " if ($soc_id);
- $out .= "socket_id=$socket_id, " if ($socket_id);
- $out .= "totem_id=$totem_id, " if ($totem_id);
- $out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
- $out .= "sub_system_id=$sub_system_id, " if ($sub_system_id);
- $out .= "module_id=$module_id, " if ($module_id);
- $out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
- $out .= "core_id=$core_id, " if ($core_id);
- $out .= "port_id=$port_id, " if ($port_id);
- $out .= "err_type=$err_type, " if ($err_type);
- $out .= "pcie_info=$pcie_info, " if ($pcie_info);
- $out .= "err_severity=$err_severity, " if ($err_severity);
- $out .= "Error Registers: $regs" if ($regs);
- $out .= "\n\n";
+ if ($module eq 0 || ($module_id && uc($module) eq uc($module_id))) {
+ $out .= "$id. $timestamp Error Info: ";
+ $out .= "version=$version, ";
+ $out .= "soc_id=$soc_id, " if ($soc_id);
+ $out .= "socket_id=$socket_id, " if ($socket_id);
+ $out .= "totem_id=$totem_id, " if ($totem_id);
+ $out .= "nimbus_id=$nimbus_id, " if ($nimbus_id);
+ $out .= "sub_system_id=$sub_system_id, " if ($sub_system_id);
+ $out .= "module_id=$module_id, " if ($module_id);
+ $out .= "sub_module_id=$sub_module_id, " if ($sub_module_id);
+ $out .= "core_id=$core_id, " if ($core_id);
+ $out .= "port_id=$port_id, " if ($port_id);
+ $out .= "err_type=$err_type, " if ($err_type);
+ $out .= "pcie_info=$pcie_info, " if ($pcie_info);
+ $out .= "err_severity=$err_severity, " if ($err_severity);
+ $out .= "Error Registers: $regs" if ($regs);
+ $out .= "\n\n";
+ $found_module = 1;
+ }
}
if ($out ne "") {
print "HiSilicon Kunpeng9xx common error events:\n$out\n";
@@ -1768,6 +1791,12 @@ sub vendor_errors
$query_handle->finish;
}
+ if ($platform_id && !($found_platform)) {
+ print "Platform ID $platform_id is not valid\n";
+ } elsif ($module && !($found_module)) {
+ print "No error record for the module $module\n";
+ }
+
undef($dbh);
}
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH v2 08/10] rasdaemon: ras-mc-ctl: Relocate reading and display Kunpeng920 errors to under Kunpeng9xx
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
` (6 preceding siblings ...)
2022-10-03 16:17 ` [PATCH v2 07/10] rasdaemon: ras-mc-ctl: Add support to display the HiSilicon vendor errors for a specified module shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
2022-10-03 16:17 ` [PATCH v2 09/10] rasdaemon: ras-mc-ctl: Updated HiSilicon platform name shiju.jose
2022-10-03 16:17 ` [PATCH v2 10/10] rasdaemon: Fix for a memory out-of-bounds issue and optimized code to remove duplicate function shiju.jose
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Relocate reading and display Kunpeng920 errors to under Kunpeng9xx.
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
util/ras-mc-ctl.in | 40 ++++++++++------------------------------
1 file changed, 10 insertions(+), 30 deletions(-)
diff --git a/util/ras-mc-ctl.in b/util/ras-mc-ctl.in
index 296eb87..75981a0 100755
--- a/util/ras-mc-ctl.in
+++ b/util/ras-mc-ctl.in
@@ -1529,7 +1529,6 @@ sub errors
# Definitions of the vendor platform IDs.
use constant {
- HISILICON_KUNPENG_920 => "Kunpeng920",
HISILICON_KUNPENG_9XX => "Kunpeng9xx",
};
@@ -1553,8 +1552,8 @@ sub vendor_errors_summary
my $dbh = DBI->connect("dbi:SQLite:dbname=$dbname", "", "", {});
- # HiSilicon Kunpeng920 errors
- if ($platform_id eq HISILICON_KUNPENG_920) {
+ # HiSilicon Kunpeng9xx errors
+ if ($platform_id eq HISILICON_KUNPENG_9XX) {
$found_platform = 1;
$query = "select err_severity, module_id, count(*) from hip08_oem_type1_event_v2 group by err_severity, module_id";
$query_handle = $dbh->prepare($query);
@@ -1570,9 +1569,7 @@ sub vendor_errors_summary
$out .= "\t$module_id: $count\n";
}
if ($out ne "") {
- print "HiSilicon Kunpeng920 OEM type1 error events summary:\n$out\n";
- } else {
- print "No HiSilicon Kunpeng920 OEM type1 errors.\n\n";
+ print "HiSilicon Kunpeng9xx OEM type1 error events summary:\n$out\n";
}
$query_handle->finish;
@@ -1590,9 +1587,7 @@ sub vendor_errors_summary
$out .= "\t$module_id: $count\n";
}
if ($out ne "") {
- print "HiSilicon Kunpeng920 OEM type2 error events summary:\n$out\n";
- } else {
- print "No HiSilicon Kunpeng920 OEM type2 errors.\n\n";
+ print "HiSilicon Kunpeng9xx OEM type2 error events summary:\n$out\n";
}
$query_handle->finish;
@@ -1610,16 +1605,10 @@ sub vendor_errors_summary
$out .= "\t$sub_module_id: $count\n";
}
if ($out ne "") {
- print "HiSilicon Kunpeng920 PCIe controller error events summary:\n$out\n";
- } else {
- print "No HiSilicon Kunpeng920 PCIe controller errors.\n\n";
+ print "HiSilicon Kunpeng9xx PCIe controller error events summary:\n$out\n";
}
$query_handle->finish;
- }
- # HiSilicon Kunpeng9xx common errors
- if ($platform_id eq HISILICON_KUNPENG_9XX) {
- $found_platform = 1;
$query = "select err_severity, module_id, count(*) from hisi_common_section_v2 group by err_severity, module_id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
@@ -1635,8 +1624,6 @@ sub vendor_errors_summary
}
if ($out ne "") {
print "HiSilicon Kunpeng9xx common error events summary:\n$out\n";
- } else {
- print "No HiSilicon Kunpeng9xx common errors.\n\n";
}
$query_handle->finish;
}
@@ -1673,8 +1660,8 @@ sub vendor_errors
my $dbh = DBI->connect("dbi:SQLite:dbname=$dbname", "", "", {});
- # HiSilicon Kunpeng920 errors
- if ($platform_id eq HISILICON_KUNPENG_920) {
+ # HiSilicon Kunpeng9xx errors
+ if ($platform_id eq HISILICON_KUNPENG_9XX) {
$found_platform = 1;
$query = "select id, timestamp, version, soc_id, socket_id, nimbus_id, module_id, sub_module_id, err_severity, regs_dump from hip08_oem_type1_event_v2 order by id, module_id, err_severity";
$query_handle = $dbh->prepare($query);
@@ -1697,7 +1684,7 @@ sub vendor_errors
}
}
if ($out ne "") {
- print "HiSilicon Kunpeng920 OEM type1 error events:\n$out\n";
+ print "HiSilicon Kunpeng9xx OEM type1 error events:\n$out\n";
}
$query_handle->finish;
@@ -1722,7 +1709,7 @@ sub vendor_errors
}
}
if ($out ne "") {
- print "HiSilicon Kunpeng920 OEM type2 error events:\n$out\n";
+ print "HiSilicon Kunpeng9xx OEM type2 error events:\n$out\n";
}
$query_handle->finish;
@@ -1749,14 +1736,10 @@ sub vendor_errors
}
}
if ($out ne "") {
- print "HiSilicon Kunpeng920 PCIe controller error events:\n$out\n";
+ print "HiSilicon Kunpeng9xx PCIe controller error events:\n$out\n";
}
$query_handle->finish;
- }
- # HiSilicon Kunpeng9xx common errors
- if ($platform_id eq HISILICON_KUNPENG_9XX) {
- $found_platform = 1;
$query = "select id, timestamp, version, soc_id, socket_id, totem_id, nimbus_id, sub_system_id, module_id, sub_module_id, core_id, port_id, err_type, pcie_info, err_severity, regs_dump from hisi_common_section_v2 order by id, module_id, err_severity";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
@@ -1785,8 +1768,6 @@ sub vendor_errors
}
if ($out ne "") {
print "HiSilicon Kunpeng9xx common error events:\n$out\n";
- } else {
- print "No HiSilicon Kunpeng9xx common errors.\n";
}
$query_handle->finish;
}
@@ -1803,7 +1784,6 @@ sub vendor_errors
sub vendor_platforms
{
print "\nSupported platforms for the vendor-specific errors:\n";
- print "\tHiSilicon Kunpeng920, platform-id=\"", HISILICON_KUNPENG_920, "\"\n";
print "\tHiSilicon Kunpeng9xx, platform-id=\"", HISILICON_KUNPENG_9XX, "\"\n";
print "\n";
}
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH v2 09/10] rasdaemon: ras-mc-ctl: Updated HiSilicon platform name
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
` (7 preceding siblings ...)
2022-10-03 16:17 ` [PATCH v2 08/10] rasdaemon: ras-mc-ctl: Relocate reading and display Kunpeng920 errors to under Kunpeng9xx shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
2022-10-03 16:17 ` [PATCH v2 10/10] rasdaemon: Fix for a memory out-of-bounds issue and optimized code to remove duplicate function shiju.jose
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Updated the HiSilicon platform name as KunPeng9xx.
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
util/ras-mc-ctl.in | 24 ++++++++++++------------
1 file changed, 12 insertions(+), 12 deletions(-)
diff --git a/util/ras-mc-ctl.in b/util/ras-mc-ctl.in
index 75981a0..1cc19b3 100755
--- a/util/ras-mc-ctl.in
+++ b/util/ras-mc-ctl.in
@@ -1529,7 +1529,7 @@ sub errors
# Definitions of the vendor platform IDs.
use constant {
- HISILICON_KUNPENG_9XX => "Kunpeng9xx",
+ HISILICON_KUNPENG_9XX => "KunPeng9xx",
};
sub vendor_errors_summary
@@ -1552,7 +1552,7 @@ sub vendor_errors_summary
my $dbh = DBI->connect("dbi:SQLite:dbname=$dbname", "", "", {});
- # HiSilicon Kunpeng9xx errors
+ # HiSilicon KunPeng9xx errors
if ($platform_id eq HISILICON_KUNPENG_9XX) {
$found_platform = 1;
$query = "select err_severity, module_id, count(*) from hip08_oem_type1_event_v2 group by err_severity, module_id";
@@ -1569,7 +1569,7 @@ sub vendor_errors_summary
$out .= "\t$module_id: $count\n";
}
if ($out ne "") {
- print "HiSilicon Kunpeng9xx OEM type1 error events summary:\n$out\n";
+ print "HiSilicon KunPeng9xx OEM type1 error events summary:\n$out\n";
}
$query_handle->finish;
@@ -1587,7 +1587,7 @@ sub vendor_errors_summary
$out .= "\t$module_id: $count\n";
}
if ($out ne "") {
- print "HiSilicon Kunpeng9xx OEM type2 error events summary:\n$out\n";
+ print "HiSilicon KunPeng9xx OEM type2 error events summary:\n$out\n";
}
$query_handle->finish;
@@ -1605,7 +1605,7 @@ sub vendor_errors_summary
$out .= "\t$sub_module_id: $count\n";
}
if ($out ne "") {
- print "HiSilicon Kunpeng9xx PCIe controller error events summary:\n$out\n";
+ print "HiSilicon KunPeng9xx PCIe controller error events summary:\n$out\n";
}
$query_handle->finish;
@@ -1623,7 +1623,7 @@ sub vendor_errors_summary
$out .= "\t$module_id: $count\n";
}
if ($out ne "") {
- print "HiSilicon Kunpeng9xx common error events summary:\n$out\n";
+ print "HiSilicon KunPeng9xx common error events summary:\n$out\n";
}
$query_handle->finish;
}
@@ -1660,7 +1660,7 @@ sub vendor_errors
my $dbh = DBI->connect("dbi:SQLite:dbname=$dbname", "", "", {});
- # HiSilicon Kunpeng9xx errors
+ # HiSilicon KunPeng9xx errors
if ($platform_id eq HISILICON_KUNPENG_9XX) {
$found_platform = 1;
$query = "select id, timestamp, version, soc_id, socket_id, nimbus_id, module_id, sub_module_id, err_severity, regs_dump from hip08_oem_type1_event_v2 order by id, module_id, err_severity";
@@ -1684,7 +1684,7 @@ sub vendor_errors
}
}
if ($out ne "") {
- print "HiSilicon Kunpeng9xx OEM type1 error events:\n$out\n";
+ print "HiSilicon KunPeng9xx OEM type1 error events:\n$out\n";
}
$query_handle->finish;
@@ -1709,7 +1709,7 @@ sub vendor_errors
}
}
if ($out ne "") {
- print "HiSilicon Kunpeng9xx OEM type2 error events:\n$out\n";
+ print "HiSilicon KunPeng9xx OEM type2 error events:\n$out\n";
}
$query_handle->finish;
@@ -1736,7 +1736,7 @@ sub vendor_errors
}
}
if ($out ne "") {
- print "HiSilicon Kunpeng9xx PCIe controller error events:\n$out\n";
+ print "HiSilicon KunPeng9xx PCIe controller error events:\n$out\n";
}
$query_handle->finish;
@@ -1767,7 +1767,7 @@ sub vendor_errors
}
}
if ($out ne "") {
- print "HiSilicon Kunpeng9xx common error events:\n$out\n";
+ print "HiSilicon KunPeng9xx common error events:\n$out\n";
}
$query_handle->finish;
}
@@ -1784,7 +1784,7 @@ sub vendor_errors
sub vendor_platforms
{
print "\nSupported platforms for the vendor-specific errors:\n";
- print "\tHiSilicon Kunpeng9xx, platform-id=\"", HISILICON_KUNPENG_9XX, "\"\n";
+ print "\tHiSilicon KunPeng9xx, platform-id=\"", HISILICON_KUNPENG_9XX, "\"\n";
print "\n";
}
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread* [PATCH v2 10/10] rasdaemon: Fix for a memory out-of-bounds issue and optimized code to remove duplicate function.
2022-10-03 16:17 [PATCH v2 00/10] rasdaemon: Add cpu fault isolation support and improvements to the HiSilicon vendor specific code shiju.jose
` (8 preceding siblings ...)
2022-10-03 16:17 ` [PATCH v2 09/10] rasdaemon: ras-mc-ctl: Updated HiSilicon platform name shiju.jose
@ 2022-10-03 16:17 ` shiju.jose
9 siblings, 0 replies; 11+ messages in thread
From: shiju.jose @ 2022-10-03 16:17 UTC (permalink / raw)
To: linux-edac, mchehab
Cc: linuxarm, tanxiaofei, jonathan.cameron, prime.zeng, luoshengwei,
panjunchong, fenglei47, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Fixed a memory out-of-bounds issue with string pointers and
optimized code structure to remove duplicate function.
Signed-off-by: Lei Feng <fenglei47@h-partners.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
non-standard-hisi_hip08.c | 6 +++---
non-standard-hisilicon.c | 2 +-
ras-non-standard-handler.c | 16 +---------------
3 files changed, 5 insertions(+), 19 deletions(-)
diff --git a/non-standard-hisi_hip08.c b/non-standard-hisi_hip08.c
index 9092183..4ef47ea 100644
--- a/non-standard-hisi_hip08.c
+++ b/non-standard-hisi_hip08.c
@@ -1014,15 +1014,15 @@ static int decode_hip08_pcie_local_error(struct ras_events *ras,
static struct ras_ns_ev_decoder hip08_ns_ev_decoder[] = {
{
- .sec_type = "1f8161e155d641e6bd107afd1dc5f7c5",
+ .sec_type = "1f8161e1-55d6-41e6-bd10-7afd1dc5f7c5",
.decode = decode_hip08_oem_type1_error,
},
{
- .sec_type = "45534ea6ce2341158535e07ab3aef91d",
+ .sec_type = "45534ea6-ce23-4115-8535-e07ab3aef91d",
.decode = decode_hip08_oem_type2_error,
},
{
- .sec_type = "b2889fc9e7d74f9da867af42e98be772",
+ .sec_type = "b2889fc9-e7d7-4f9d-a867-af42e98be772",
.decode = decode_hip08_pcie_local_error,
},
};
diff --git a/non-standard-hisilicon.c b/non-standard-hisilicon.c
index d1e1774..6ee9271 100644
--- a/non-standard-hisilicon.c
+++ b/non-standard-hisilicon.c
@@ -387,7 +387,7 @@ static int decode_hisi_common_section(struct ras_events *ras,
static struct ras_ns_ev_decoder hisi_section_ns_ev_decoder[] = {
{
- .sec_type = "c8b328a899174af69a132e08ab2e7586",
+ .sec_type = "c8b328a8-9917-4af6-9a13-2e08ab2e7586",
.decode = decode_hisi_common_section,
},
};
diff --git a/ras-non-standard-handler.c b/ras-non-standard-handler.c
index 6d5a6f8..6932e58 100644
--- a/ras-non-standard-handler.c
+++ b/ras-non-standard-handler.c
@@ -52,20 +52,6 @@ static char *uuid_le(const char *uu)
return uuid;
}
-static int uuid_le_cmp(const char *sec_type, const char *uuid2)
-{
- static char uuid1[32];
- char *p = uuid1;
- int i;
- static const unsigned char le[16] = {
- 3, 2, 1, 0, 5, 4, 7, 6, 8, 9, 10, 11, 12, 13, 14, 15};
-
- for (i = 0; i < 16; i++)
- p += sprintf(p, "%.2x", (unsigned char) sec_type[le[i]]);
- *p = 0;
- return strncmp(uuid1, uuid2, 32);
-}
-
int register_ns_ev_decoder(struct ras_ns_ev_decoder *ns_ev_decoder)
{
struct ras_ns_ev_decoder *list;
@@ -96,7 +82,7 @@ static int find_ns_ev_decoder(const char *sec_type, struct ras_ns_ev_decoder **p
ns_ev_decoder = ras_ns_ev_dec_list;
while (ns_ev_decoder) {
- if (uuid_le_cmp(sec_type, ns_ev_decoder->sec_type) == 0) {
+ if (strcmp(uuid_le(sec_type), ns_ev_decoder->sec_type) == 0) {
*p_ns_ev_dec = ns_ev_decoder;
match = 1;
break;
--
2.25.1
^ permalink raw reply related [flat|nested] 11+ messages in thread