* [RESEND PATCH V3 1/4] rasdaemon: Move definition for BIT and BIT_ULL to a common file
2023-02-02 18:18 [RESEND PATCH V3 0/4] rasdaemon: Add support for the CXL error events shiju.jose
@ 2023-02-02 18:18 ` shiju.jose
2023-02-02 18:18 ` [RESEND PATCH V3 2/4] rasdaemon: Add support for the CXL poison events shiju.jose
` (2 subsequent siblings)
3 siblings, 0 replies; 7+ messages in thread
From: shiju.jose @ 2023-02-02 18:18 UTC (permalink / raw)
To: mchehab, linux-edac, linux-cxl; +Cc: jonathan.cameron, linuxarm, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Move definition for BIT() and BIT_ULL() to the
common file ras-record.h
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
---
ras-non-standard-handler.h | 3 ---
ras-record.h | 3 +++
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/ras-non-standard-handler.h b/ras-non-standard-handler.h
index 4d9f938..c360eaf 100644
--- a/ras-non-standard-handler.h
+++ b/ras-non-standard-handler.h
@@ -17,9 +17,6 @@
#include "ras-events.h"
#include <traceevent/event-parse.h>
-#define BIT(nr) (1UL << (nr))
-#define BIT_ULL(nr) (1ULL << (nr))
-
struct ras_ns_ev_decoder {
struct ras_ns_ev_decoder *next;
const char *sec_type;
diff --git a/ras-record.h b/ras-record.h
index d9f7733..219f10b 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -25,6 +25,9 @@
#define ARRAY_SIZE(x) (sizeof(x)/sizeof(*(x)))
+#define BIT(nr) (1UL << (nr))
+#define BIT_ULL(nr) (1ULL << (nr))
+
extern long user_hz;
struct ras_events;
--
2.25.1
^ permalink raw reply related [flat|nested] 7+ messages in thread* [RESEND PATCH V3 2/4] rasdaemon: Add support for the CXL poison events
2023-02-02 18:18 [RESEND PATCH V3 0/4] rasdaemon: Add support for the CXL error events shiju.jose
2023-02-02 18:18 ` [RESEND PATCH V3 1/4] rasdaemon: Move definition for BIT and BIT_ULL to a common file shiju.jose
@ 2023-02-02 18:18 ` shiju.jose
2023-02-02 19:41 ` Alison Schofield
2023-02-02 18:18 ` [RESEND PATCH V3 3/4] rasdaemon: Add support for the CXL AER uncorrectable errors shiju.jose
2023-02-02 18:18 ` [RESEND PATCH V3 4/4] rasdaemon: Add support for the CXL AER correctable errors shiju.jose
3 siblings, 1 reply; 7+ messages in thread
From: shiju.jose @ 2023-02-02 18:18 UTC (permalink / raw)
To: mchehab, linux-edac, linux-cxl; +Cc: jonathan.cameron, linuxarm, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Add support to log and record the CXL poison events.
The corresponding Kernel patches here:
https://lore.kernel.org/linux-cxl/de11785ff05844299b40b100f8e0f56c7eef7f08.1674070170.git.alison.schofield@intel.com/
Presently RFC draft version for logging, could be extended for the policy
based recovery action for the frequent poison events depending on the above
kernel patches.
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
Makefile.am | 7 +-
configure.ac | 11 +++
ras-cxl-handler.c | 172 ++++++++++++++++++++++++++++++++++++++++++++++
ras-cxl-handler.h | 24 +++++++
ras-events.c | 15 ++++
ras-events.h | 1 +
ras-record.c | 81 ++++++++++++++++++++++
ras-record.h | 20 ++++++
ras-report.c | 83 ++++++++++++++++++++++
ras-report.h | 2 +
10 files changed, 415 insertions(+), 1 deletion(-)
create mode 100644 ras-cxl-handler.c
create mode 100644 ras-cxl-handler.h
diff --git a/Makefile.am b/Makefile.am
index a9832cc..bd7b2ae 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -73,6 +73,11 @@ endif
if WITH_CPU_FAULT_ISOLATION
rasdaemon_SOURCES += ras-cpu-isolation.c queue.c
endif
+
+if WITH_CXL
+ rasdaemon_SOURCES += ras-cxl-handler.c
+endif
+
rasdaemon_LDADD = -lpthread $(SQLITE3_LIBS) $(LIBTRACEEVENT_LIBS)
rasdaemon_CFLAGS = $(SQLITE3_CFLAGS) $(LIBTRACEEVENT_CFLAGS)
@@ -81,7 +86,7 @@ include_HEADERS = config.h ras-events.h ras-logger.h ras-mc-handler.h \
ras-extlog-handler.h ras-arm-handler.h ras-non-standard-handler.h \
ras-devlink-handler.h ras-diskerror-handler.h rbtree.h ras-page-isolation.h \
non-standard-hisilicon.h non-standard-ampere.h ras-memory-failure-handler.h \
- ras-cpu-isolation.h queue.h
+ ras-cxl-handler.h ras-cpu-isolation.h queue.h
# This rule can't be called with more than one Makefile job (like make -j8)
# I can't figure out a way to fix that
diff --git a/configure.ac b/configure.ac
index c973aaf..028b9b3 100644
--- a/configure.ac
+++ b/configure.ac
@@ -127,6 +127,16 @@ AS_IF([test "x$enable_memory_failure" = "xyes" || test "x$enable_all" = "xyes"],
AM_CONDITIONAL([WITH_MEMORY_FAILURE], [test x$enable_memory_failure = xyes || test x$enable_all = xyes])
AM_COND_IF([WITH_MEMORY_FAILURE], [USE_MEMORY_FAILURE="yes"], [USE_MEMORY_FAILURE="no"])
+AC_ARG_ENABLE([cxl],
+ AS_HELP_STRING([--enable-cxl], [enable CXL events (currently experimental)]))
+
+AS_IF([test "x$enable_cxl" = "xyes" || test "x$enable_all" == "xyes"], [
+ AC_DEFINE(HAVE_CXL,1,"have CXL events collect")
+ AC_SUBST([WITH_CXL])
+])
+AM_CONDITIONAL([WITH_CXL], [test x$enable_cxl = xyes || test x$enable_all == xyes])
+AM_COND_IF([WITH_CXL], [USE_CXL="yes"], [USE_CXL="no"])
+
AC_ARG_ENABLE([abrt_report],
AS_HELP_STRING([--enable-abrt-report], [enable report event to ABRT (currently experimental)]))
@@ -215,6 +225,7 @@ compile time options summary
DEVLINK : $USE_DEVLINK
Disk I/O errors : $USE_DISKERROR
Memory Failure : $USE_MEMORY_FAILURE
+ CXL events : $USE_CXL
Memory CE PFA : $USE_MEMORY_CE_PFA
AMP RAS errors : $USE_AMP_NS_DECODE
CPU fault isolation : $USE_CPU_FAULT_ISOLATION
diff --git a/ras-cxl-handler.c b/ras-cxl-handler.c
new file mode 100644
index 0000000..0ba2519
--- /dev/null
+++ b/ras-cxl-handler.c
@@ -0,0 +1,172 @@
+/*
+ * Copyright (c) Huawei Technologies Co., Ltd. 2023. 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 <unistd.h>
+#include <traceevent/kbuffer.h>
+#include "ras-cxl-handler.h"
+#include "ras-record.h"
+#include "ras-logger.h"
+#include "ras-report.h"
+
+/* Poison List: Payload out flags */
+#define CXL_POISON_FLAG_MORE BIT(0)
+#define CXL_POISON_FLAG_OVERFLOW BIT(1)
+#define CXL_POISON_FLAG_SCANNING BIT(2)
+
+/* CXL poison - source types */
+enum cxl_poison_source {
+ CXL_POISON_SOURCE_UNKNOWN = 0,
+ CXL_POISON_SOURCE_EXTERNAL = 1,
+ CXL_POISON_SOURCE_INTERNAL = 2,
+ CXL_POISON_SOURCE_INJECTED = 3,
+ CXL_POISON_SOURCE_VENDOR = 7,
+};
+
+int ras_cxl_poison_event_handler(struct trace_seq *s,
+ struct tep_record *record,
+ struct tep_event *event, void *context)
+{
+ int len;
+ unsigned long long val;
+ struct ras_events *ras = context;
+ time_t now;
+ struct tm *tm;
+ struct ras_cxl_poison_event ev;
+
+ now = record->ts / user_hz + ras->uptime_diff;
+ tm = localtime(&now);
+ if (tm)
+ strftime(ev.timestamp, sizeof(ev.timestamp),
+ "%Y-%m-%d %H:%M:%S %z", tm);
+ else
+ strncpy(ev.timestamp, "1970-01-01 00:00:00 +0000", sizeof(ev.timestamp));
+ if (trace_seq_printf(s, "%s ", ev.timestamp) <= 0)
+ return -1;
+
+ ev.memdev = tep_get_field_raw(s, event, "memdev",
+ record, &len, 1);
+ if (!ev.memdev)
+ return -1;
+ if (trace_seq_printf(s, "memdev:%s ", ev.memdev) <= 0)
+ return -1;
+
+ ev.pcidev = tep_get_field_raw(s, event, "pcidev",
+ record, &len, 1);
+ if (!ev.pcidev)
+ return -1;
+ if (trace_seq_printf(s, "pcidev:%s ", ev.pcidev) <= 0)
+ return -1;
+
+ ev.region = tep_get_field_raw(s, event, "region",
+ record, &len, 1);
+ if (!ev.region)
+ return -1;
+ if (trace_seq_printf(s, "region:%s ", ev.region) <= 0)
+ return -1;
+
+ ev.uuid = tep_get_field_raw(s, event, "uuid",
+ record, &len, 1);
+ if (!ev.uuid)
+ return -1;
+ if (trace_seq_printf(s, "region_uuid:%s ", ev.uuid) <= 0)
+ return -1;
+
+ if (tep_get_field_val(s, event, "hpa", record, &val, 1) < 0)
+ return -1;
+ ev.hpa = val;
+ if (trace_seq_printf(s, "poison list: hpa:0x%llx ", (unsigned long long)ev.hpa) <= 0)
+ return -1;
+
+ if (tep_get_field_val(s, event, "dpa", record, &val, 1) < 0)
+ return -1;
+ ev.dpa = val;
+ if (trace_seq_printf(s, "dpa:0x%llx ", (unsigned long long)ev.dpa) <= 0)
+ return -1;
+
+ if (tep_get_field_val(s, event, "length", record, &val, 1) < 0)
+ return -1;
+ ev.length = val;
+ if (trace_seq_printf(s, "length:%d ", ev.length) <= 0)
+ return -1;
+
+ if (tep_get_field_val(s, event, "source", record, &val, 1) < 0)
+ return -1;
+
+ switch (val) {
+ case CXL_POISON_SOURCE_UNKNOWN:
+ ev.source = "Unknown";
+ break;
+ case CXL_POISON_SOURCE_EXTERNAL:
+ ev.source = "External";
+ break;
+ case CXL_POISON_SOURCE_INTERNAL:
+ ev.source = "Internal";
+ break;
+ case CXL_POISON_SOURCE_INJECTED:
+ ev.source = "Injected";
+ break;
+ case CXL_POISON_SOURCE_VENDOR:
+ ev.source = "Vendor";
+ break;
+ default:
+ ev.source = "Invalid";
+ }
+ if (trace_seq_printf(s, "source:%s ", ev.source) <= 0)
+ return -1;
+
+ if (tep_get_field_val(s, event, "flags", record, &val, 1) < 0)
+ return -1;
+ ev.flags = val;
+ if (trace_seq_printf(s, "flags:%d ", ev.flags) <= 0)
+ return -1;
+
+ if (ev.flags & CXL_POISON_FLAG_OVERFLOW) {
+ if (tep_get_field_val(s, event, "overflow_t", record, &val, 1) < 0)
+ return -1;
+ if (val) {
+ /* CXL Specification 3.0
+ * Overflow timestamp - The number of unsigned nanoseconds
+ * that have elapsed since midnight, 01-Jan-1970 UTC
+ */
+ time_t ovf_ts_secs = val / 1000000000ULL;
+
+ tm = localtime(&ovf_ts_secs);
+ if (tm) {
+ strftime(ev.overflow_ts, sizeof(ev.overflow_ts),
+ "%Y-%m-%d %H:%M:%S %z", tm);
+ }
+ }
+ if (!val || !tm)
+ strncpy(ev.overflow_ts, "1970-01-01 00:00:00 +0000",
+ sizeof(ev.overflow_ts));
+ } else
+ strncpy(ev.overflow_ts, "1970-01-01 00:00:00 +0000", sizeof(ev.overflow_ts));
+ if (trace_seq_printf(s, "overflow timestamp:%s\n", ev.overflow_ts) <= 0)
+ return -1;
+
+ /* Insert data into the SGBD */
+#ifdef HAVE_SQLITE3
+ ras_store_cxl_poison_event(ras, &ev);
+#endif
+
+#ifdef HAVE_ABRT_REPORT
+ /* Report event to ABRT */
+ ras_report_cxl_poison_event(ras, &ev);
+#endif
+
+ return 0;
+}
diff --git a/ras-cxl-handler.h b/ras-cxl-handler.h
new file mode 100644
index 0000000..84d5cc6
--- /dev/null
+++ b/ras-cxl-handler.h
@@ -0,0 +1,24 @@
+/*
+ * Copyright (c) Huawei Technologies Co., Ltd. 2023. 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_CXL_HANDLER_H
+#define __RAS_CXL_HANDLER_H
+
+#include "ras-events.h"
+#include <traceevent/event-parse.h>
+
+int ras_cxl_poison_event_handler(struct trace_seq *s,
+ struct tep_record *record,
+ struct tep_event *event, void *context);
+#endif
diff --git a/ras-events.c b/ras-events.c
index 39f9ce2..6555125 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -40,6 +40,7 @@
#include "ras-devlink-handler.h"
#include "ras-diskerror-handler.h"
#include "ras-memory-failure-handler.h"
+#include "ras-cxl-handler.h"
#include "ras-record.h"
#include "ras-logger.h"
#include "ras-page-isolation.h"
@@ -243,6 +244,10 @@ int toggle_ras_mc_event(int enable)
rc |= __toggle_ras_mc_event(ras, "ras", "memory_failure_event", enable);
#endif
+#ifdef HAVE_CXL
+ rc |= __toggle_ras_mc_event(ras, "cxl", "cxl_poison", enable);
+#endif
+
free_ras:
free(ras);
return rc;
@@ -951,6 +956,16 @@ int handle_ras_events(int record_events)
"ras", "memory_failure_event");
#endif
+#ifdef HAVE_CXL
+ rc = add_event_handler(ras, pevent, page_size, "cxl", "cxl_poison",
+ ras_cxl_poison_event_handler, NULL, CXL_POISON_EVENT);
+ if (!rc)
+ num_events++;
+ else
+ log(ALL, LOG_ERR, "Can't get traces from %s:%s\n",
+ "cxl", "cxl_poison");
+#endif
+
if (!num_events) {
log(ALL, LOG_INFO,
"Failed to trace all supported RAS events. Aborting.\n");
diff --git a/ras-events.h b/ras-events.h
index 6c9f507..fc51070 100644
--- a/ras-events.h
+++ b/ras-events.h
@@ -39,6 +39,7 @@ enum {
DEVLINK_EVENT,
DISKERROR_EVENT,
MF_EVENT,
+ CXL_POISON_EVENT,
NR_EVENTS
};
diff --git a/ras-record.c b/ras-record.c
index a367939..f54fb41 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -559,6 +559,67 @@ int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev)
}
#endif
+#ifdef HAVE_CXL
+/*
+ * Table and functions to handle cxl:cxl_poison
+ */
+static const struct db_fields cxl_poison_event_fields[] = {
+ { .name = "id", .type = "INTEGER PRIMARY KEY" },
+ { .name = "timestamp", .type = "TEXT" },
+ { .name = "memdev", .type = "TEXT" },
+ { .name = "pcidev", .type = "TEXT" },
+ { .name = "region", .type = "TEXT" },
+ { .name = "region_uuid", .type = "TEXT" },
+ { .name = "hpa", .type = "INTEGER" },
+ { .name = "dpa", .type = "INTEGER" },
+ { .name = "length", .type = "INTEGER" },
+ { .name = "source", .type = "TEXT" },
+ { .name = "flags", .type = "INTEGER" },
+ { .name = "overflow_ts", .type = "TEXT" },
+};
+
+static const struct db_table_descriptor cxl_poison_event_tab = {
+ .name = "cxl_poison_event",
+ .fields = cxl_poison_event_fields,
+ .num_fields = ARRAY_SIZE(cxl_poison_event_fields),
+};
+
+int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev)
+{
+ int rc;
+ struct sqlite3_priv *priv = ras->db_priv;
+
+ if (!priv || !priv->stmt_cxl_poison_event)
+ return 0;
+ log(TERM, LOG_INFO, "cxl_poison_event store: %p\n", priv->stmt_cxl_poison_event);
+
+ sqlite3_bind_text(priv->stmt_cxl_poison_event, 1, ev->timestamp, -1, NULL);
+ sqlite3_bind_text(priv->stmt_cxl_poison_event, 2, ev->memdev, -1, NULL);
+ sqlite3_bind_text(priv->stmt_cxl_poison_event, 3, ev->pcidev, -1, NULL);
+ sqlite3_bind_text(priv->stmt_cxl_poison_event, 4, ev->region, -1, NULL);
+ sqlite3_bind_text(priv->stmt_cxl_poison_event, 5, ev->uuid, -1, NULL);
+ sqlite3_bind_int64(priv->stmt_cxl_poison_event, 6, ev->hpa);
+ sqlite3_bind_int64(priv->stmt_cxl_poison_event, 7, ev->dpa);
+ sqlite3_bind_int(priv->stmt_cxl_poison_event, 8, ev->length);
+ sqlite3_bind_text(priv->stmt_cxl_poison_event, 9, ev->source, -1, NULL);
+ sqlite3_bind_int(priv->stmt_cxl_poison_event, 10, ev->flags);
+ sqlite3_bind_text(priv->stmt_cxl_poison_event, 11, ev->overflow_ts, -1, NULL);
+
+ rc = sqlite3_step(priv->stmt_cxl_poison_event);
+ if (rc != SQLITE_OK && rc != SQLITE_DONE)
+ log(TERM, LOG_ERR,
+ "Failed to do cxl_poison_event step on sqlite: error = %d\n", rc);
+ rc = sqlite3_reset(priv->stmt_cxl_poison_event);
+ if (rc != SQLITE_OK && rc != SQLITE_DONE)
+ log(TERM, LOG_ERR,
+ "Failed reset cxl_poison_event on sqlite: error = %d\n",
+ rc);
+ log(TERM, LOG_INFO, "register inserted at db\n");
+
+ return rc;
+}
+#endif
+
/*
* Generic code
*/
@@ -896,6 +957,16 @@ int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras)
}
#endif
+#ifdef HAVE_CXL
+ rc = ras_mc_create_table(priv, &cxl_poison_event_tab);
+ if (rc == SQLITE_OK) {
+ rc = ras_mc_prepare_stmt(priv, &priv->stmt_cxl_poison_event,
+ &cxl_poison_event_tab);
+ if (rc != SQLITE_OK)
+ goto error;
+ }
+#endif
+
ras->db_priv = priv;
return 0;
@@ -1008,6 +1079,16 @@ int ras_mc_event_closedb(unsigned int cpu, struct ras_events *ras)
}
#endif
+#ifdef HAVE_CXL
+ if (priv->stmt_cxl_poison_event) {
+ rc = sqlite3_finalize(priv->stmt_cxl_poison_event);
+ if (rc != SQLITE_OK)
+ log(TERM, LOG_ERR,
+ "cpu %u: Failed to finalize cxl_poison_event sqlite: error = %d\n",
+ cpu, rc);
+ }
+#endif
+
rc = sqlite3_close_v2(db);
if (rc != SQLITE_OK)
log(TERM, LOG_ERR,
diff --git a/ras-record.h b/ras-record.h
index 219f10b..e5bf483 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -114,6 +114,20 @@ struct ras_mf_event {
const char *action_result;
};
+struct ras_cxl_poison_event {
+ char timestamp[64];
+ const char *memdev;
+ const char *pcidev;
+ const char *region;
+ const char *uuid;
+ uint64_t hpa;
+ uint64_t dpa;
+ uint32_t length;
+ const char *source;
+ uint8_t flags;
+ char overflow_ts[64];
+};
+
struct ras_mc_event;
struct ras_aer_event;
struct ras_extlog_event;
@@ -123,6 +137,7 @@ struct mce_event;
struct devlink_event;
struct diskerror_event;
struct ras_mf_event;
+struct ras_cxl_poison_event;
#ifdef HAVE_SQLITE3
@@ -155,6 +170,9 @@ struct sqlite3_priv {
#ifdef HAVE_MEMORY_FAILURE
sqlite3_stmt *stmt_mf_event;
#endif
+#ifdef HAVE_CXL
+ sqlite3_stmt *stmt_cxl_poison_event;
+#endif
};
struct db_fields {
@@ -182,6 +200,7 @@ int ras_store_arm_record(struct ras_events *ras, struct ras_arm_event *ev);
int ras_store_devlink_event(struct ras_events *ras, struct devlink_event *ev);
int ras_store_diskerror_event(struct ras_events *ras, struct diskerror_event *ev);
int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev);
+int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev);
#else
static inline int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras) { return 0; };
@@ -195,6 +214,7 @@ static inline int ras_store_arm_record(struct ras_events *ras, struct ras_arm_ev
static inline int ras_store_devlink_event(struct ras_events *ras, struct devlink_event *ev) { return 0; };
static inline int ras_store_diskerror_event(struct ras_events *ras, struct diskerror_event *ev) { return 0; };
static inline int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev) { return 0; };
+static inline int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev) { return 0; };
#endif
diff --git a/ras-report.c b/ras-report.c
index 62d5eb7..589e640 100644
--- a/ras-report.c
+++ b/ras-report.c
@@ -331,6 +331,42 @@ static int set_mf_event_backtrace(char *buf, struct ras_mf_event *ev)
return 0;
}
+static int set_cxl_poison_event_backtrace(char *buf, struct ras_cxl_poison_event *ev)
+{
+ char bt_buf[MAX_BACKTRACE_SIZE];
+
+ if (!buf || !ev)
+ return -1;
+
+ sprintf(bt_buf, "BACKTRACE=" \
+ "timestamp=%s\n" \
+ "memdev=%s\n" \
+ "pcidev=%s\n" \
+ "region=%s\n" \
+ "uuid=%s\n" \
+ "hpa=0x%lx\n" \
+ "dpa=0x%lx\n" \
+ "length=%d\n" \
+ "source=%s\n" \
+ "flags=%d\n" \
+ "overflow_timestamp=%s\n" \
+ ev->timestamp, \
+ ev->memdev, \
+ ev->pcidev, \
+ ev->region, \
+ ev->uuid, \
+ ev->hpa, \
+ ev->dpa, \
+ ev->length, \
+ ev->source, \
+ ev->flags, \
+ ev->overflow_ts);
+
+ strcat(buf, bt_buf);
+
+ return 0;
+}
+
static int commit_report_backtrace(int sockfd, int type, void *ev){
char buf[MAX_BACKTRACE_SIZE];
char *pbuf = buf;
@@ -368,6 +404,9 @@ static int commit_report_backtrace(int sockfd, int type, void *ev){
case MF_EVENT:
rc = set_mf_event_backtrace(buf, (struct ras_mf_event *)ev);
break;
+ case CXL_POISON_EVENT:
+ rc = set_cxl_poison_event_backtrace(buf, (struct ras_cxl_poison_event *)ev);
+ break;
default:
return -1;
}
@@ -776,3 +815,47 @@ mf_fail:
else
return -1;
}
+
+int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev)
+{
+ char buf[MAX_MESSAGE_SIZE];
+ int sockfd = 0;
+ int done = 0;
+ int rc = -1;
+
+ memset(buf, 0, sizeof(buf));
+
+ sockfd = setup_report_socket();
+ if (sockfd < 0)
+ return -1;
+
+ rc = commit_report_basic(sockfd);
+ if (rc < 0)
+ goto cxl_poison_fail;
+
+ rc = commit_report_backtrace(sockfd, CXL_POISON_EVENT, ev);
+ if (rc < 0)
+ goto cxl_poison_fail;
+
+ sprintf(buf, "ANALYZER=%s", "rasdaemon-cxl-poison");
+ rc = write(sockfd, buf, strlen(buf) + 1);
+ if (rc < strlen(buf) + 1)
+ goto cxl_poison_fail;
+
+ sprintf(buf, "REASON=%s", "CXL poison");
+ rc = write(sockfd, buf, strlen(buf) + 1);
+ if (rc < strlen(buf) + 1)
+ goto cxl_poison_fail;
+
+ done = 1;
+
+cxl_poison_fail:
+
+ if (sockfd >= 0)
+ close(sockfd);
+
+ if (done)
+ return 0;
+ else
+ return -1;
+}
diff --git a/ras-report.h b/ras-report.h
index e605eb1..d1591ce 100644
--- a/ras-report.h
+++ b/ras-report.h
@@ -39,6 +39,7 @@ int ras_report_arm_event(struct ras_events *ras, struct ras_arm_event *ev);
int ras_report_devlink_event(struct ras_events *ras, struct devlink_event *ev);
int ras_report_diskerror_event(struct ras_events *ras, struct diskerror_event *ev);
int ras_report_mf_event(struct ras_events *ras, struct ras_mf_event *ev);
+int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev);
#else
@@ -50,6 +51,7 @@ static inline int ras_report_arm_event(struct ras_events *ras, struct ras_arm_ev
static inline int ras_report_devlink_event(struct ras_events *ras, struct devlink_event *ev) { return 0; };
static inline int ras_report_diskerror_event(struct ras_events *ras, struct diskerror_event *ev) { return 0; };
static inline int ras_report_mf_event(struct ras_events *ras, struct ras_mf_event *ev) { return 0; };
+static inline int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev) { return 0; };
#endif
--
2.25.1
^ permalink raw reply related [flat|nested] 7+ messages in thread* Re: [RESEND PATCH V3 2/4] rasdaemon: Add support for the CXL poison events
2023-02-02 18:18 ` [RESEND PATCH V3 2/4] rasdaemon: Add support for the CXL poison events shiju.jose
@ 2023-02-02 19:41 ` Alison Schofield
2023-02-03 9:31 ` Shiju Jose
0 siblings, 1 reply; 7+ messages in thread
From: Alison Schofield @ 2023-02-02 19:41 UTC (permalink / raw)
To: shiju.jose; +Cc: mchehab, linux-edac, linux-cxl, jonathan.cameron, linuxarm
On Thu, Feb 02, 2023 at 06:18:44PM +0000, shiju.jose@huawei.com wrote:
> From: Shiju Jose <shiju.jose@huawei.com>
>
> Add support to log and record the CXL poison events.
>
> The corresponding Kernel patches here:
> https://lore.kernel.org/linux-cxl/de11785ff05844299b40b100f8e0f56c7eef7f08.1674070170.git.alison.schofield@intel.com/
>
> Presently RFC draft version for logging, could be extended for the policy
> based recovery action for the frequent poison events depending on the above
> kernel patches.
Hi Shiju,
Looks good to me based on the kernel patches you reference above.
I want to let you know that a v6 is in the works, and will lead to
some naming changes below - but minor stuff.
Alison
>
> Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
> Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> ---
> Makefile.am | 7 +-
> configure.ac | 11 +++
> ras-cxl-handler.c | 172 ++++++++++++++++++++++++++++++++++++++++++++++
> ras-cxl-handler.h | 24 +++++++
> ras-events.c | 15 ++++
> ras-events.h | 1 +
> ras-record.c | 81 ++++++++++++++++++++++
> ras-record.h | 20 ++++++
> ras-report.c | 83 ++++++++++++++++++++++
> ras-report.h | 2 +
> 10 files changed, 415 insertions(+), 1 deletion(-)
> create mode 100644 ras-cxl-handler.c
> create mode 100644 ras-cxl-handler.h
>
> diff --git a/Makefile.am b/Makefile.am
> index a9832cc..bd7b2ae 100644
> --- a/Makefile.am
> +++ b/Makefile.am
> @@ -73,6 +73,11 @@ endif
> if WITH_CPU_FAULT_ISOLATION
> rasdaemon_SOURCES += ras-cpu-isolation.c queue.c
> endif
> +
> +if WITH_CXL
> + rasdaemon_SOURCES += ras-cxl-handler.c
> +endif
> +
> rasdaemon_LDADD = -lpthread $(SQLITE3_LIBS) $(LIBTRACEEVENT_LIBS)
> rasdaemon_CFLAGS = $(SQLITE3_CFLAGS) $(LIBTRACEEVENT_CFLAGS)
>
> @@ -81,7 +86,7 @@ include_HEADERS = config.h ras-events.h ras-logger.h ras-mc-handler.h \
> ras-extlog-handler.h ras-arm-handler.h ras-non-standard-handler.h \
> ras-devlink-handler.h ras-diskerror-handler.h rbtree.h ras-page-isolation.h \
> non-standard-hisilicon.h non-standard-ampere.h ras-memory-failure-handler.h \
> - ras-cpu-isolation.h queue.h
> + ras-cxl-handler.h ras-cpu-isolation.h queue.h
>
> # This rule can't be called with more than one Makefile job (like make -j8)
> # I can't figure out a way to fix that
> diff --git a/configure.ac b/configure.ac
> index c973aaf..028b9b3 100644
> --- a/configure.ac
> +++ b/configure.ac
> @@ -127,6 +127,16 @@ AS_IF([test "x$enable_memory_failure" = "xyes" || test "x$enable_all" = "xyes"],
> AM_CONDITIONAL([WITH_MEMORY_FAILURE], [test x$enable_memory_failure = xyes || test x$enable_all = xyes])
> AM_COND_IF([WITH_MEMORY_FAILURE], [USE_MEMORY_FAILURE="yes"], [USE_MEMORY_FAILURE="no"])
>
> +AC_ARG_ENABLE([cxl],
> + AS_HELP_STRING([--enable-cxl], [enable CXL events (currently experimental)]))
> +
> +AS_IF([test "x$enable_cxl" = "xyes" || test "x$enable_all" == "xyes"], [
> + AC_DEFINE(HAVE_CXL,1,"have CXL events collect")
> + AC_SUBST([WITH_CXL])
> +])
> +AM_CONDITIONAL([WITH_CXL], [test x$enable_cxl = xyes || test x$enable_all == xyes])
> +AM_COND_IF([WITH_CXL], [USE_CXL="yes"], [USE_CXL="no"])
> +
> AC_ARG_ENABLE([abrt_report],
> AS_HELP_STRING([--enable-abrt-report], [enable report event to ABRT (currently experimental)]))
>
> @@ -215,6 +225,7 @@ compile time options summary
> DEVLINK : $USE_DEVLINK
> Disk I/O errors : $USE_DISKERROR
> Memory Failure : $USE_MEMORY_FAILURE
> + CXL events : $USE_CXL
> Memory CE PFA : $USE_MEMORY_CE_PFA
> AMP RAS errors : $USE_AMP_NS_DECODE
> CPU fault isolation : $USE_CPU_FAULT_ISOLATION
> diff --git a/ras-cxl-handler.c b/ras-cxl-handler.c
> new file mode 100644
> index 0000000..0ba2519
> --- /dev/null
> +++ b/ras-cxl-handler.c
> @@ -0,0 +1,172 @@
> +/*
> + * Copyright (c) Huawei Technologies Co., Ltd. 2023. 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 <unistd.h>
> +#include <traceevent/kbuffer.h>
> +#include "ras-cxl-handler.h"
> +#include "ras-record.h"
> +#include "ras-logger.h"
> +#include "ras-report.h"
> +
> +/* Poison List: Payload out flags */
> +#define CXL_POISON_FLAG_MORE BIT(0)
> +#define CXL_POISON_FLAG_OVERFLOW BIT(1)
> +#define CXL_POISON_FLAG_SCANNING BIT(2)
> +
> +/* CXL poison - source types */
> +enum cxl_poison_source {
> + CXL_POISON_SOURCE_UNKNOWN = 0,
> + CXL_POISON_SOURCE_EXTERNAL = 1,
> + CXL_POISON_SOURCE_INTERNAL = 2,
> + CXL_POISON_SOURCE_INJECTED = 3,
> + CXL_POISON_SOURCE_VENDOR = 7,
> +};
> +
> +int ras_cxl_poison_event_handler(struct trace_seq *s,
> + struct tep_record *record,
> + struct tep_event *event, void *context)
> +{
> + int len;
> + unsigned long long val;
> + struct ras_events *ras = context;
> + time_t now;
> + struct tm *tm;
> + struct ras_cxl_poison_event ev;
> +
> + now = record->ts / user_hz + ras->uptime_diff;
> + tm = localtime(&now);
> + if (tm)
> + strftime(ev.timestamp, sizeof(ev.timestamp),
> + "%Y-%m-%d %H:%M:%S %z", tm);
> + else
> + strncpy(ev.timestamp, "1970-01-01 00:00:00 +0000", sizeof(ev.timestamp));
> + if (trace_seq_printf(s, "%s ", ev.timestamp) <= 0)
> + return -1;
> +
> + ev.memdev = tep_get_field_raw(s, event, "memdev",
> + record, &len, 1);
> + if (!ev.memdev)
> + return -1;
> + if (trace_seq_printf(s, "memdev:%s ", ev.memdev) <= 0)
> + return -1;
> +
> + ev.pcidev = tep_get_field_raw(s, event, "pcidev",
> + record, &len, 1);
> + if (!ev.pcidev)
> + return -1;
> + if (trace_seq_printf(s, "pcidev:%s ", ev.pcidev) <= 0)
> + return -1;
> +
> + ev.region = tep_get_field_raw(s, event, "region",
> + record, &len, 1);
> + if (!ev.region)
> + return -1;
> + if (trace_seq_printf(s, "region:%s ", ev.region) <= 0)
> + return -1;
> +
> + ev.uuid = tep_get_field_raw(s, event, "uuid",
> + record, &len, 1);
> + if (!ev.uuid)
> + return -1;
> + if (trace_seq_printf(s, "region_uuid:%s ", ev.uuid) <= 0)
> + return -1;
> +
> + if (tep_get_field_val(s, event, "hpa", record, &val, 1) < 0)
> + return -1;
> + ev.hpa = val;
> + if (trace_seq_printf(s, "poison list: hpa:0x%llx ", (unsigned long long)ev.hpa) <= 0)
> + return -1;
> +
> + if (tep_get_field_val(s, event, "dpa", record, &val, 1) < 0)
> + return -1;
> + ev.dpa = val;
> + if (trace_seq_printf(s, "dpa:0x%llx ", (unsigned long long)ev.dpa) <= 0)
> + return -1;
> +
> + if (tep_get_field_val(s, event, "length", record, &val, 1) < 0)
> + return -1;
> + ev.length = val;
> + if (trace_seq_printf(s, "length:%d ", ev.length) <= 0)
> + return -1;
> +
> + if (tep_get_field_val(s, event, "source", record, &val, 1) < 0)
> + return -1;
> +
> + switch (val) {
> + case CXL_POISON_SOURCE_UNKNOWN:
> + ev.source = "Unknown";
> + break;
> + case CXL_POISON_SOURCE_EXTERNAL:
> + ev.source = "External";
> + break;
> + case CXL_POISON_SOURCE_INTERNAL:
> + ev.source = "Internal";
> + break;
> + case CXL_POISON_SOURCE_INJECTED:
> + ev.source = "Injected";
> + break;
> + case CXL_POISON_SOURCE_VENDOR:
> + ev.source = "Vendor";
> + break;
> + default:
> + ev.source = "Invalid";
> + }
> + if (trace_seq_printf(s, "source:%s ", ev.source) <= 0)
> + return -1;
> +
> + if (tep_get_field_val(s, event, "flags", record, &val, 1) < 0)
> + return -1;
> + ev.flags = val;
> + if (trace_seq_printf(s, "flags:%d ", ev.flags) <= 0)
> + return -1;
> +
> + if (ev.flags & CXL_POISON_FLAG_OVERFLOW) {
> + if (tep_get_field_val(s, event, "overflow_t", record, &val, 1) < 0)
> + return -1;
> + if (val) {
> + /* CXL Specification 3.0
> + * Overflow timestamp - The number of unsigned nanoseconds
> + * that have elapsed since midnight, 01-Jan-1970 UTC
> + */
> + time_t ovf_ts_secs = val / 1000000000ULL;
> +
> + tm = localtime(&ovf_ts_secs);
> + if (tm) {
> + strftime(ev.overflow_ts, sizeof(ev.overflow_ts),
> + "%Y-%m-%d %H:%M:%S %z", tm);
> + }
> + }
> + if (!val || !tm)
> + strncpy(ev.overflow_ts, "1970-01-01 00:00:00 +0000",
> + sizeof(ev.overflow_ts));
> + } else
> + strncpy(ev.overflow_ts, "1970-01-01 00:00:00 +0000", sizeof(ev.overflow_ts));
> + if (trace_seq_printf(s, "overflow timestamp:%s\n", ev.overflow_ts) <= 0)
> + return -1;
> +
> + /* Insert data into the SGBD */
> +#ifdef HAVE_SQLITE3
> + ras_store_cxl_poison_event(ras, &ev);
> +#endif
> +
> +#ifdef HAVE_ABRT_REPORT
> + /* Report event to ABRT */
> + ras_report_cxl_poison_event(ras, &ev);
> +#endif
> +
> + return 0;
> +}
> diff --git a/ras-cxl-handler.h b/ras-cxl-handler.h
> new file mode 100644
> index 0000000..84d5cc6
> --- /dev/null
> +++ b/ras-cxl-handler.h
> @@ -0,0 +1,24 @@
> +/*
> + * Copyright (c) Huawei Technologies Co., Ltd. 2023. 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_CXL_HANDLER_H
> +#define __RAS_CXL_HANDLER_H
> +
> +#include "ras-events.h"
> +#include <traceevent/event-parse.h>
> +
> +int ras_cxl_poison_event_handler(struct trace_seq *s,
> + struct tep_record *record,
> + struct tep_event *event, void *context);
> +#endif
> diff --git a/ras-events.c b/ras-events.c
> index 39f9ce2..6555125 100644
> --- a/ras-events.c
> +++ b/ras-events.c
> @@ -40,6 +40,7 @@
> #include "ras-devlink-handler.h"
> #include "ras-diskerror-handler.h"
> #include "ras-memory-failure-handler.h"
> +#include "ras-cxl-handler.h"
> #include "ras-record.h"
> #include "ras-logger.h"
> #include "ras-page-isolation.h"
> @@ -243,6 +244,10 @@ int toggle_ras_mc_event(int enable)
> rc |= __toggle_ras_mc_event(ras, "ras", "memory_failure_event", enable);
> #endif
>
> +#ifdef HAVE_CXL
> + rc |= __toggle_ras_mc_event(ras, "cxl", "cxl_poison", enable);
> +#endif
> +
> free_ras:
> free(ras);
> return rc;
> @@ -951,6 +956,16 @@ int handle_ras_events(int record_events)
> "ras", "memory_failure_event");
> #endif
>
> +#ifdef HAVE_CXL
> + rc = add_event_handler(ras, pevent, page_size, "cxl", "cxl_poison",
> + ras_cxl_poison_event_handler, NULL, CXL_POISON_EVENT);
> + if (!rc)
> + num_events++;
> + else
> + log(ALL, LOG_ERR, "Can't get traces from %s:%s\n",
> + "cxl", "cxl_poison");
> +#endif
> +
> if (!num_events) {
> log(ALL, LOG_INFO,
> "Failed to trace all supported RAS events. Aborting.\n");
> diff --git a/ras-events.h b/ras-events.h
> index 6c9f507..fc51070 100644
> --- a/ras-events.h
> +++ b/ras-events.h
> @@ -39,6 +39,7 @@ enum {
> DEVLINK_EVENT,
> DISKERROR_EVENT,
> MF_EVENT,
> + CXL_POISON_EVENT,
> NR_EVENTS
> };
>
> diff --git a/ras-record.c b/ras-record.c
> index a367939..f54fb41 100644
> --- a/ras-record.c
> +++ b/ras-record.c
> @@ -559,6 +559,67 @@ int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev)
> }
> #endif
>
> +#ifdef HAVE_CXL
> +/*
> + * Table and functions to handle cxl:cxl_poison
> + */
> +static const struct db_fields cxl_poison_event_fields[] = {
> + { .name = "id", .type = "INTEGER PRIMARY KEY" },
> + { .name = "timestamp", .type = "TEXT" },
> + { .name = "memdev", .type = "TEXT" },
> + { .name = "pcidev", .type = "TEXT" },
> + { .name = "region", .type = "TEXT" },
> + { .name = "region_uuid", .type = "TEXT" },
> + { .name = "hpa", .type = "INTEGER" },
> + { .name = "dpa", .type = "INTEGER" },
> + { .name = "length", .type = "INTEGER" },
> + { .name = "source", .type = "TEXT" },
> + { .name = "flags", .type = "INTEGER" },
> + { .name = "overflow_ts", .type = "TEXT" },
> +};
> +
> +static const struct db_table_descriptor cxl_poison_event_tab = {
> + .name = "cxl_poison_event",
> + .fields = cxl_poison_event_fields,
> + .num_fields = ARRAY_SIZE(cxl_poison_event_fields),
> +};
> +
> +int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev)
> +{
> + int rc;
> + struct sqlite3_priv *priv = ras->db_priv;
> +
> + if (!priv || !priv->stmt_cxl_poison_event)
> + return 0;
> + log(TERM, LOG_INFO, "cxl_poison_event store: %p\n", priv->stmt_cxl_poison_event);
> +
> + sqlite3_bind_text(priv->stmt_cxl_poison_event, 1, ev->timestamp, -1, NULL);
> + sqlite3_bind_text(priv->stmt_cxl_poison_event, 2, ev->memdev, -1, NULL);
> + sqlite3_bind_text(priv->stmt_cxl_poison_event, 3, ev->pcidev, -1, NULL);
> + sqlite3_bind_text(priv->stmt_cxl_poison_event, 4, ev->region, -1, NULL);
> + sqlite3_bind_text(priv->stmt_cxl_poison_event, 5, ev->uuid, -1, NULL);
> + sqlite3_bind_int64(priv->stmt_cxl_poison_event, 6, ev->hpa);
> + sqlite3_bind_int64(priv->stmt_cxl_poison_event, 7, ev->dpa);
> + sqlite3_bind_int(priv->stmt_cxl_poison_event, 8, ev->length);
> + sqlite3_bind_text(priv->stmt_cxl_poison_event, 9, ev->source, -1, NULL);
> + sqlite3_bind_int(priv->stmt_cxl_poison_event, 10, ev->flags);
> + sqlite3_bind_text(priv->stmt_cxl_poison_event, 11, ev->overflow_ts, -1, NULL);
> +
> + rc = sqlite3_step(priv->stmt_cxl_poison_event);
> + if (rc != SQLITE_OK && rc != SQLITE_DONE)
> + log(TERM, LOG_ERR,
> + "Failed to do cxl_poison_event step on sqlite: error = %d\n", rc);
> + rc = sqlite3_reset(priv->stmt_cxl_poison_event);
> + if (rc != SQLITE_OK && rc != SQLITE_DONE)
> + log(TERM, LOG_ERR,
> + "Failed reset cxl_poison_event on sqlite: error = %d\n",
> + rc);
> + log(TERM, LOG_INFO, "register inserted at db\n");
> +
> + return rc;
> +}
> +#endif
> +
> /*
> * Generic code
> */
> @@ -896,6 +957,16 @@ int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras)
> }
> #endif
>
> +#ifdef HAVE_CXL
> + rc = ras_mc_create_table(priv, &cxl_poison_event_tab);
> + if (rc == SQLITE_OK) {
> + rc = ras_mc_prepare_stmt(priv, &priv->stmt_cxl_poison_event,
> + &cxl_poison_event_tab);
> + if (rc != SQLITE_OK)
> + goto error;
> + }
> +#endif
> +
> ras->db_priv = priv;
> return 0;
>
> @@ -1008,6 +1079,16 @@ int ras_mc_event_closedb(unsigned int cpu, struct ras_events *ras)
> }
> #endif
>
> +#ifdef HAVE_CXL
> + if (priv->stmt_cxl_poison_event) {
> + rc = sqlite3_finalize(priv->stmt_cxl_poison_event);
> + if (rc != SQLITE_OK)
> + log(TERM, LOG_ERR,
> + "cpu %u: Failed to finalize cxl_poison_event sqlite: error = %d\n",
> + cpu, rc);
> + }
> +#endif
> +
> rc = sqlite3_close_v2(db);
> if (rc != SQLITE_OK)
> log(TERM, LOG_ERR,
> diff --git a/ras-record.h b/ras-record.h
> index 219f10b..e5bf483 100644
> --- a/ras-record.h
> +++ b/ras-record.h
> @@ -114,6 +114,20 @@ struct ras_mf_event {
> const char *action_result;
> };
>
> +struct ras_cxl_poison_event {
> + char timestamp[64];
> + const char *memdev;
> + const char *pcidev;
> + const char *region;
> + const char *uuid;
> + uint64_t hpa;
> + uint64_t dpa;
> + uint32_t length;
> + const char *source;
> + uint8_t flags;
> + char overflow_ts[64];
> +};
> +
> struct ras_mc_event;
> struct ras_aer_event;
> struct ras_extlog_event;
> @@ -123,6 +137,7 @@ struct mce_event;
> struct devlink_event;
> struct diskerror_event;
> struct ras_mf_event;
> +struct ras_cxl_poison_event;
>
> #ifdef HAVE_SQLITE3
>
> @@ -155,6 +170,9 @@ struct sqlite3_priv {
> #ifdef HAVE_MEMORY_FAILURE
> sqlite3_stmt *stmt_mf_event;
> #endif
> +#ifdef HAVE_CXL
> + sqlite3_stmt *stmt_cxl_poison_event;
> +#endif
> };
>
> struct db_fields {
> @@ -182,6 +200,7 @@ int ras_store_arm_record(struct ras_events *ras, struct ras_arm_event *ev);
> int ras_store_devlink_event(struct ras_events *ras, struct devlink_event *ev);
> int ras_store_diskerror_event(struct ras_events *ras, struct diskerror_event *ev);
> int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev);
> +int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev);
>
> #else
> static inline int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras) { return 0; };
> @@ -195,6 +214,7 @@ static inline int ras_store_arm_record(struct ras_events *ras, struct ras_arm_ev
> static inline int ras_store_devlink_event(struct ras_events *ras, struct devlink_event *ev) { return 0; };
> static inline int ras_store_diskerror_event(struct ras_events *ras, struct diskerror_event *ev) { return 0; };
> static inline int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev) { return 0; };
> +static inline int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev) { return 0; };
>
> #endif
>
> diff --git a/ras-report.c b/ras-report.c
> index 62d5eb7..589e640 100644
> --- a/ras-report.c
> +++ b/ras-report.c
> @@ -331,6 +331,42 @@ static int set_mf_event_backtrace(char *buf, struct ras_mf_event *ev)
> return 0;
> }
>
> +static int set_cxl_poison_event_backtrace(char *buf, struct ras_cxl_poison_event *ev)
> +{
> + char bt_buf[MAX_BACKTRACE_SIZE];
> +
> + if (!buf || !ev)
> + return -1;
> +
> + sprintf(bt_buf, "BACKTRACE=" \
> + "timestamp=%s\n" \
> + "memdev=%s\n" \
> + "pcidev=%s\n" \
> + "region=%s\n" \
> + "uuid=%s\n" \
> + "hpa=0x%lx\n" \
> + "dpa=0x%lx\n" \
> + "length=%d\n" \
> + "source=%s\n" \
> + "flags=%d\n" \
> + "overflow_timestamp=%s\n" \
> + ev->timestamp, \
> + ev->memdev, \
> + ev->pcidev, \
> + ev->region, \
> + ev->uuid, \
> + ev->hpa, \
> + ev->dpa, \
> + ev->length, \
> + ev->source, \
> + ev->flags, \
> + ev->overflow_ts);
> +
> + strcat(buf, bt_buf);
> +
> + return 0;
> +}
> +
> static int commit_report_backtrace(int sockfd, int type, void *ev){
> char buf[MAX_BACKTRACE_SIZE];
> char *pbuf = buf;
> @@ -368,6 +404,9 @@ static int commit_report_backtrace(int sockfd, int type, void *ev){
> case MF_EVENT:
> rc = set_mf_event_backtrace(buf, (struct ras_mf_event *)ev);
> break;
> + case CXL_POISON_EVENT:
> + rc = set_cxl_poison_event_backtrace(buf, (struct ras_cxl_poison_event *)ev);
> + break;
> default:
> return -1;
> }
> @@ -776,3 +815,47 @@ mf_fail:
> else
> return -1;
> }
> +
> +int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev)
> +{
> + char buf[MAX_MESSAGE_SIZE];
> + int sockfd = 0;
> + int done = 0;
> + int rc = -1;
> +
> + memset(buf, 0, sizeof(buf));
> +
> + sockfd = setup_report_socket();
> + if (sockfd < 0)
> + return -1;
> +
> + rc = commit_report_basic(sockfd);
> + if (rc < 0)
> + goto cxl_poison_fail;
> +
> + rc = commit_report_backtrace(sockfd, CXL_POISON_EVENT, ev);
> + if (rc < 0)
> + goto cxl_poison_fail;
> +
> + sprintf(buf, "ANALYZER=%s", "rasdaemon-cxl-poison");
> + rc = write(sockfd, buf, strlen(buf) + 1);
> + if (rc < strlen(buf) + 1)
> + goto cxl_poison_fail;
> +
> + sprintf(buf, "REASON=%s", "CXL poison");
> + rc = write(sockfd, buf, strlen(buf) + 1);
> + if (rc < strlen(buf) + 1)
> + goto cxl_poison_fail;
> +
> + done = 1;
> +
> +cxl_poison_fail:
> +
> + if (sockfd >= 0)
> + close(sockfd);
> +
> + if (done)
> + return 0;
> + else
> + return -1;
> +}
> diff --git a/ras-report.h b/ras-report.h
> index e605eb1..d1591ce 100644
> --- a/ras-report.h
> +++ b/ras-report.h
> @@ -39,6 +39,7 @@ int ras_report_arm_event(struct ras_events *ras, struct ras_arm_event *ev);
> int ras_report_devlink_event(struct ras_events *ras, struct devlink_event *ev);
> int ras_report_diskerror_event(struct ras_events *ras, struct diskerror_event *ev);
> int ras_report_mf_event(struct ras_events *ras, struct ras_mf_event *ev);
> +int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev);
>
> #else
>
> @@ -50,6 +51,7 @@ static inline int ras_report_arm_event(struct ras_events *ras, struct ras_arm_ev
> static inline int ras_report_devlink_event(struct ras_events *ras, struct devlink_event *ev) { return 0; };
> static inline int ras_report_diskerror_event(struct ras_events *ras, struct diskerror_event *ev) { return 0; };
> static inline int ras_report_mf_event(struct ras_events *ras, struct ras_mf_event *ev) { return 0; };
> +static inline int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev) { return 0; };
>
> #endif
>
> --
> 2.25.1
>
^ permalink raw reply [flat|nested] 7+ messages in thread* RE: [RESEND PATCH V3 2/4] rasdaemon: Add support for the CXL poison events
2023-02-02 19:41 ` Alison Schofield
@ 2023-02-03 9:31 ` Shiju Jose
0 siblings, 0 replies; 7+ messages in thread
From: Shiju Jose @ 2023-02-03 9:31 UTC (permalink / raw)
To: Alison Schofield
Cc: mchehab, linux-edac, linux-cxl, Jonathan Cameron, Linuxarm
>-----Original Message-----
>From: Alison Schofield <alison.schofield@intel.com>
>Sent: 02 February 2023 19:42
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: mchehab@kernel.org; linux-edac@vger.kernel.org; linux-
>cxl@vger.kernel.org; Jonathan Cameron <jonathan.cameron@huawei.com>;
>Linuxarm <linuxarm@huawei.com>
>Subject: Re: [RESEND PATCH V3 2/4] rasdaemon: Add support for the CXL poison
>events
>
>On Thu, Feb 02, 2023 at 06:18:44PM +0000, shiju.jose@huawei.com wrote:
>> From: Shiju Jose <shiju.jose@huawei.com>
>>
>> Add support to log and record the CXL poison events.
>>
>> The corresponding Kernel patches here:
>> https://lore.kernel.org/linux-cxl/de11785ff05844299b40b100f8e0f56c7eef
>> 7f08.1674070170.git.alison.schofield@intel.com/
>>
>> Presently RFC draft version for logging, could be extended for the
>> policy based recovery action for the frequent poison events depending
>> on the above kernel patches.
>
>Hi Shiju,
>
>Looks good to me based on the kernel patches you reference above.
>I want to let you know that a v6 is in the works, and will lead to some naming
>changes below - but minor stuff.
Hi Alison,
Thanks for the information. I will make changes according to the v6.
>
>Alison
>
Thanks,
Shiju
^ permalink raw reply [flat|nested] 7+ messages in thread
* [RESEND PATCH V3 3/4] rasdaemon: Add support for the CXL AER uncorrectable errors
2023-02-02 18:18 [RESEND PATCH V3 0/4] rasdaemon: Add support for the CXL error events shiju.jose
2023-02-02 18:18 ` [RESEND PATCH V3 1/4] rasdaemon: Move definition for BIT and BIT_ULL to a common file shiju.jose
2023-02-02 18:18 ` [RESEND PATCH V3 2/4] rasdaemon: Add support for the CXL poison events shiju.jose
@ 2023-02-02 18:18 ` shiju.jose
2023-02-02 18:18 ` [RESEND PATCH V3 4/4] rasdaemon: Add support for the CXL AER correctable errors shiju.jose
3 siblings, 0 replies; 7+ messages in thread
From: shiju.jose @ 2023-02-02 18:18 UTC (permalink / raw)
To: mchehab, linux-edac, linux-cxl; +Cc: jonathan.cameron, linuxarm, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Add support to log and record the CXL AER uncorrectable errors.
The corresponding Kernel patch here:
https://patchwork.kernel.org/project/cxl/patch/166974413388.1608150.5875712482260436188.stgit@djiang5-desk3.ch.intel.com/
It was found that the header log data to be converted to the
big-endian format to correctly store in the SQLite database likely
because the SQLite database seems uses the big-endian storage.
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
ras-cxl-handler.c | 138 ++++++++++++++++++++++++++++++++++++++++++++++
ras-cxl-handler.h | 5 ++
ras-events.c | 9 +++
ras-events.h | 1 +
ras-record.c | 65 ++++++++++++++++++++++
ras-record.h | 16 ++++++
ras-report.c | 69 +++++++++++++++++++++++
ras-report.h | 2 +
8 files changed, 305 insertions(+)
diff --git a/ras-cxl-handler.c b/ras-cxl-handler.c
index 0ba2519..50bbdb0 100644
--- a/ras-cxl-handler.c
+++ b/ras-cxl-handler.c
@@ -21,6 +21,7 @@
#include "ras-record.h"
#include "ras-logger.h"
#include "ras-report.h"
+#include <endian.h>
/* Poison List: Payload out flags */
#define CXL_POISON_FLAG_MORE BIT(0)
@@ -170,3 +171,140 @@ int ras_cxl_poison_event_handler(struct trace_seq *s,
return 0;
}
+
+/* CXL AER Errors */
+
+#define CXL_AER_UE_CACHE_DATA_PARITY BIT(0)
+#define CXL_AER_UE_CACHE_ADDR_PARITY BIT(1)
+#define CXL_AER_UE_CACHE_BE_PARITY BIT(2)
+#define CXL_AER_UE_CACHE_DATA_ECC BIT(3)
+#define CXL_AER_UE_MEM_DATA_PARITY BIT(4)
+#define CXL_AER_UE_MEM_ADDR_PARITY BIT(5)
+#define CXL_AER_UE_MEM_BE_PARITY BIT(6)
+#define CXL_AER_UE_MEM_DATA_ECC BIT(7)
+#define CXL_AER_UE_REINIT_THRESH BIT(8)
+#define CXL_AER_UE_RSVD_ENCODE BIT(9)
+#define CXL_AER_UE_POISON BIT(10)
+#define CXL_AER_UE_RECV_OVERFLOW BIT(11)
+#define CXL_AER_UE_INTERNAL_ERR BIT(14)
+#define CXL_AER_UE_IDE_TX_ERR BIT(15)
+#define CXL_AER_UE_IDE_RX_ERR BIT(16)
+
+struct cxl_error_list {
+ uint32_t bit;
+ const char *error;
+};
+
+static const struct cxl_error_list cxl_aer_ue[] = {
+ { .bit = CXL_AER_UE_CACHE_DATA_PARITY, .error = "Cache Data Parity Error" },
+ { .bit = CXL_AER_UE_CACHE_ADDR_PARITY, .error = "Cache Address Parity Error" },
+ { .bit = CXL_AER_UE_CACHE_BE_PARITY, .error = "Cache Byte Enable Parity Error" },
+ { .bit = CXL_AER_UE_CACHE_DATA_ECC, .error = "Cache Data ECC Error" },
+ { .bit = CXL_AER_UE_MEM_DATA_PARITY, .error = "Memory Data Parity Error" },
+ { .bit = CXL_AER_UE_MEM_ADDR_PARITY, .error = "Memory Address Parity Error" },
+ { .bit = CXL_AER_UE_MEM_BE_PARITY, .error = "Memory Byte Enable Parity Error" },
+ { .bit = CXL_AER_UE_MEM_DATA_ECC, .error = "Memory Data ECC Error" },
+ { .bit = CXL_AER_UE_REINIT_THRESH, .error = "REINIT Threshold Hit" },
+ { .bit = CXL_AER_UE_RSVD_ENCODE, .error = "Received Unrecognized Encoding" },
+ { .bit = CXL_AER_UE_POISON, .error = "Received Poison From Peer" },
+ { .bit = CXL_AER_UE_RECV_OVERFLOW, .error = "Receiver Overflow" },
+ { .bit = CXL_AER_UE_INTERNAL_ERR, .error = "Component Specific Error" },
+ { .bit = CXL_AER_UE_IDE_TX_ERR, .error = "IDE Tx Error" },
+ { .bit = CXL_AER_UE_IDE_RX_ERR, .error = "IDE Rx Error" },
+};
+
+static int decode_cxl_error_status(struct trace_seq *s, uint32_t status,
+ const struct cxl_error_list *cxl_error_list,
+ uint8_t num_elems)
+{
+ int i;
+
+ for (i = 0; i < num_elems; i++) {
+ if (status & cxl_error_list[i].bit)
+ if (trace_seq_printf(s, "\'%s\' ", cxl_error_list[i].error) <= 0)
+ return -1;
+ }
+ return 0;
+}
+
+int ras_cxl_aer_ue_event_handler(struct trace_seq *s,
+ struct tep_record *record,
+ struct tep_event *event, void *context)
+{
+ int len, i;
+ unsigned long long val;
+ time_t now;
+ struct tm *tm;
+ struct ras_events *ras = context;
+ struct ras_cxl_aer_ue_event ev;
+
+ memset(&ev, 0, sizeof(ev));
+ now = record->ts / user_hz + ras->uptime_diff;
+ tm = localtime(&now);
+ if (tm)
+ strftime(ev.timestamp, sizeof(ev.timestamp),
+ "%Y-%m-%d %H:%M:%S %z", tm);
+ else
+ strncpy(ev.timestamp, "1970-01-01 00:00:00 +0000", sizeof(ev.timestamp));
+ if (trace_seq_printf(s, "%s ", ev.timestamp) <= 0)
+ return -1;
+
+ ev.dev_name = tep_get_field_raw(s, event, "dev_name",
+ record, &len, 1);
+ if (!ev.dev_name)
+ return -1;
+ if (trace_seq_printf(s, "dev_name:%s ", ev.dev_name) <= 0)
+ return -1;
+
+ if (tep_get_field_val(s, event, "status", record, &val, 1) < 0)
+ return -1;
+ ev.error_status = val;
+
+ if (trace_seq_printf(s, "error status:") <= 0)
+ return -1;
+ if (decode_cxl_error_status(s, ev.error_status,
+ cxl_aer_ue, ARRAY_SIZE(cxl_aer_ue)) < 0)
+ return -1;
+
+ if (tep_get_field_val(s, event, "first_error", record, &val, 1) < 0)
+ return -1;
+ ev.first_error = val;
+
+ if (trace_seq_printf(s, "first error:") <= 0)
+ return -1;
+ if (decode_cxl_error_status(s, ev.first_error,
+ cxl_aer_ue, ARRAY_SIZE(cxl_aer_ue)) < 0)
+ return -1;
+
+ ev.header_log = tep_get_field_raw(s, event, "header_log",
+ record, &len, 1);
+ if (!ev.header_log)
+ return -1;
+ if (trace_seq_printf(s, "header log:\n") <= 0)
+ return -1;
+ for (i = 0; i < CXL_HEADERLOG_SIZE_U32; i++) {
+ if (trace_seq_printf(s, "%08x ", ev.header_log[i]) <= 0)
+ break;
+ if ((i > 0) && ((i % 20) == 0))
+ if (trace_seq_printf(s, "\n") <= 0)
+ break;
+ /* Convert header log data to the big-endian format because
+ * the SQLite database seems uses the big-endian storage.
+ */
+ ev.header_log[i] = htobe32(ev.header_log[i]);
+ }
+ if (i < CXL_HEADERLOG_SIZE_U32)
+ return -1;
+
+ /* Insert data into the SGBD */
+#ifdef HAVE_SQLITE3
+ ras_store_cxl_aer_ue_event(ras, &ev);
+#endif
+
+#ifdef HAVE_ABRT_REPORT
+ /* Report event to ABRT */
+ ras_report_cxl_aer_ue_event(ras, &ev);
+#endif
+
+ return 0;
+}
diff --git a/ras-cxl-handler.h b/ras-cxl-handler.h
index 84d5cc6..18b3120 100644
--- a/ras-cxl-handler.h
+++ b/ras-cxl-handler.h
@@ -21,4 +21,9 @@
int ras_cxl_poison_event_handler(struct trace_seq *s,
struct tep_record *record,
struct tep_event *event, void *context);
+
+int ras_cxl_aer_ue_event_handler(struct trace_seq *s,
+ struct tep_record *record,
+ struct tep_event *event, void *context);
+
#endif
diff --git a/ras-events.c b/ras-events.c
index 6555125..ead792b 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -246,6 +246,7 @@ int toggle_ras_mc_event(int enable)
#ifdef HAVE_CXL
rc |= __toggle_ras_mc_event(ras, "cxl", "cxl_poison", enable);
+ rc |= __toggle_ras_mc_event(ras, "cxl", "cxl_aer_uncorrectable_error", enable);
#endif
free_ras:
@@ -964,6 +965,14 @@ int handle_ras_events(int record_events)
else
log(ALL, LOG_ERR, "Can't get traces from %s:%s\n",
"cxl", "cxl_poison");
+
+ rc = add_event_handler(ras, pevent, page_size, "cxl", "cxl_aer_uncorrectable_error",
+ ras_cxl_aer_ue_event_handler, NULL, CXL_AER_UE_EVENT);
+ if (!rc)
+ num_events++;
+ else
+ log(ALL, LOG_ERR, "Can't get traces from %s:%s\n",
+ "cxl", "cxl_aer_uncorrectable_error");
#endif
if (!num_events) {
diff --git a/ras-events.h b/ras-events.h
index fc51070..65f9d9a 100644
--- a/ras-events.h
+++ b/ras-events.h
@@ -40,6 +40,7 @@ enum {
DISKERROR_EVENT,
MF_EVENT,
CXL_POISON_EVENT,
+ CXL_AER_UE_EVENT,
NR_EVENTS
};
diff --git a/ras-record.c b/ras-record.c
index f54fb41..4703790 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -618,6 +618,54 @@ int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_eve
return rc;
}
+
+/*
+ * Table and functions to handle cxl:cxl_aer_uncorrectable_error
+ */
+static const struct db_fields cxl_aer_ue_event_fields[] = {
+ { .name = "id", .type = "INTEGER PRIMARY KEY" },
+ { .name = "timestamp", .type = "TEXT" },
+ { .name = "dev_name", .type = "TEXT" },
+ { .name = "error_status", .type = "INTEGER" },
+ { .name = "first_error", .type = "INTEGER" },
+ { .name = "header_log", .type = "BLOB" },
+};
+
+static const struct db_table_descriptor cxl_aer_ue_event_tab = {
+ .name = "cxl_aer_ue_event",
+ .fields = cxl_aer_ue_event_fields,
+ .num_fields = ARRAY_SIZE(cxl_aer_ue_event_fields),
+};
+
+int ras_store_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev)
+{
+ int rc;
+ struct sqlite3_priv *priv = ras->db_priv;
+
+ if (!priv || !priv->stmt_cxl_aer_ue_event)
+ return 0;
+ log(TERM, LOG_INFO, "cxl_aer_ue_event store: %p\n", priv->stmt_cxl_aer_ue_event);
+
+ sqlite3_bind_text(priv->stmt_cxl_aer_ue_event, 1, ev->timestamp, -1, NULL);
+ sqlite3_bind_text(priv->stmt_cxl_aer_ue_event, 2, ev->dev_name, -1, NULL);
+ sqlite3_bind_int(priv->stmt_cxl_aer_ue_event, 3, ev->error_status);
+ sqlite3_bind_int(priv->stmt_cxl_aer_ue_event, 4, ev->first_error);
+ sqlite3_bind_blob(priv->stmt_cxl_aer_ue_event, 5, ev->header_log, CXL_HEADERLOG_SIZE, NULL);
+
+ rc = sqlite3_step(priv->stmt_cxl_aer_ue_event);
+ if (rc != SQLITE_OK && rc != SQLITE_DONE)
+ log(TERM, LOG_ERR,
+ "Failed to do cxl_aer_ue_event step on sqlite: error = %d\n", rc);
+ rc = sqlite3_reset(priv->stmt_cxl_aer_ue_event);
+ if (rc != SQLITE_OK && rc != SQLITE_DONE)
+ log(TERM, LOG_ERR,
+ "Failed reset cxl_aer_ue_event on sqlite: error = %d\n",
+ rc);
+ log(TERM, LOG_INFO, "register inserted at db\n");
+
+ return rc;
+}
+
#endif
/*
@@ -965,6 +1013,15 @@ int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras)
if (rc != SQLITE_OK)
goto error;
}
+
+ rc = ras_mc_create_table(priv, &cxl_aer_ue_event_tab);
+ if (rc == SQLITE_OK) {
+ rc = ras_mc_prepare_stmt(priv, &priv->stmt_cxl_aer_ue_event,
+ &cxl_aer_ue_event_tab);
+ if (rc != SQLITE_OK)
+ goto error;
+ }
+
#endif
ras->db_priv = priv;
@@ -1087,6 +1144,14 @@ int ras_mc_event_closedb(unsigned int cpu, struct ras_events *ras)
"cpu %u: Failed to finalize cxl_poison_event sqlite: error = %d\n",
cpu, rc);
}
+
+ if (priv->stmt_cxl_aer_ue_event) {
+ rc = sqlite3_finalize(priv->stmt_cxl_aer_ue_event);
+ if (rc != SQLITE_OK)
+ log(TERM, LOG_ERR,
+ "cpu %u: Failed to finalize cxl_aer_ue_event sqlite: error = %d\n",
+ cpu, rc);
+ }
#endif
rc = sqlite3_close_v2(db);
diff --git a/ras-record.h b/ras-record.h
index e5bf483..0e2c178 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -128,6 +128,18 @@ struct ras_cxl_poison_event {
char overflow_ts[64];
};
+#define SZ_512 0x200
+#define CXL_HEADERLOG_SIZE SZ_512
+#define CXL_HEADERLOG_SIZE_U32 (SZ_512 / sizeof(uint32_t))
+
+struct ras_cxl_aer_ue_event {
+ char timestamp[64];
+ const char *dev_name;
+ uint32_t error_status;
+ uint32_t first_error;
+ uint32_t *header_log;
+};
+
struct ras_mc_event;
struct ras_aer_event;
struct ras_extlog_event;
@@ -138,6 +150,7 @@ struct devlink_event;
struct diskerror_event;
struct ras_mf_event;
struct ras_cxl_poison_event;
+struct ras_cxl_aer_ue_event;
#ifdef HAVE_SQLITE3
@@ -172,6 +185,7 @@ struct sqlite3_priv {
#endif
#ifdef HAVE_CXL
sqlite3_stmt *stmt_cxl_poison_event;
+ sqlite3_stmt *stmt_cxl_aer_ue_event;
#endif
};
@@ -201,6 +215,7 @@ int ras_store_devlink_event(struct ras_events *ras, struct devlink_event *ev);
int ras_store_diskerror_event(struct ras_events *ras, struct diskerror_event *ev);
int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev);
int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev);
+int ras_store_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev);
#else
static inline int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras) { return 0; };
@@ -215,6 +230,7 @@ static inline int ras_store_devlink_event(struct ras_events *ras, struct devlink
static inline int ras_store_diskerror_event(struct ras_events *ras, struct diskerror_event *ev) { return 0; };
static inline int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev) { return 0; };
static inline int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev) { return 0; };
+static inline int ras_store_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev) { return 0; };
#endif
diff --git a/ras-report.c b/ras-report.c
index 589e640..4c09061 100644
--- a/ras-report.c
+++ b/ras-report.c
@@ -367,6 +367,28 @@ static int set_cxl_poison_event_backtrace(char *buf, struct ras_cxl_poison_event
return 0;
}
+static int set_cxl_aer_ue_event_backtrace(char *buf, struct ras_cxl_aer_ue_event *ev)
+{
+ char bt_buf[MAX_BACKTRACE_SIZE];
+
+ if (!buf || !ev)
+ return -1;
+
+ sprintf(bt_buf, "BACKTRACE=" \
+ "timestamp=%s\n" \
+ "dev_name=%s\n" \
+ "error_status=%u\n" \
+ "first_error=%u\n" \
+ ev->timestamp, \
+ ev->dev_name, \
+ ev->error_status, \
+ ev->first_error);
+
+ strcat(buf, bt_buf);
+
+ return 0;
+}
+
static int commit_report_backtrace(int sockfd, int type, void *ev){
char buf[MAX_BACKTRACE_SIZE];
char *pbuf = buf;
@@ -407,6 +429,9 @@ static int commit_report_backtrace(int sockfd, int type, void *ev){
case CXL_POISON_EVENT:
rc = set_cxl_poison_event_backtrace(buf, (struct ras_cxl_poison_event *)ev);
break;
+ case CXL_AER_UE_EVENT:
+ rc = set_cxl_aer_ue_event_backtrace(buf, (struct ras_cxl_aer_ue_event *)ev);
+ break;
default:
return -1;
}
@@ -859,3 +884,47 @@ cxl_poison_fail:
else
return -1;
}
+
+int ras_report_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev)
+{
+ char buf[MAX_MESSAGE_SIZE];
+ int sockfd = 0;
+ int done = 0;
+ int rc = -1;
+
+ memset(buf, 0, sizeof(buf));
+
+ sockfd = setup_report_socket();
+ if (sockfd < 0)
+ return -1;
+
+ rc = commit_report_basic(sockfd);
+ if (rc < 0)
+ goto cxl_aer_ue_fail;
+
+ rc = commit_report_backtrace(sockfd, CXL_AER_UE_EVENT, ev);
+ if (rc < 0)
+ goto cxl_aer_ue_fail;
+
+ sprintf(buf, "ANALYZER=%s", "rasdaemon-cxl-aer-uncorrectable-error");
+ rc = write(sockfd, buf, strlen(buf) + 1);
+ if (rc < strlen(buf) + 1)
+ goto cxl_aer_ue_fail;
+
+ sprintf(buf, "REASON=%s", "CXL AER uncorrectable error");
+ rc = write(sockfd, buf, strlen(buf) + 1);
+ if (rc < strlen(buf) + 1)
+ goto cxl_aer_ue_fail;
+
+ done = 1;
+
+cxl_aer_ue_fail:
+
+ if (sockfd >= 0)
+ close(sockfd);
+
+ if (done)
+ return 0;
+ else
+ return -1;
+}
diff --git a/ras-report.h b/ras-report.h
index d1591ce..dfe89d1 100644
--- a/ras-report.h
+++ b/ras-report.h
@@ -40,6 +40,7 @@ int ras_report_devlink_event(struct ras_events *ras, struct devlink_event *ev);
int ras_report_diskerror_event(struct ras_events *ras, struct diskerror_event *ev);
int ras_report_mf_event(struct ras_events *ras, struct ras_mf_event *ev);
int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev);
+int ras_report_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev);
#else
@@ -52,6 +53,7 @@ static inline int ras_report_devlink_event(struct ras_events *ras, struct devlin
static inline int ras_report_diskerror_event(struct ras_events *ras, struct diskerror_event *ev) { return 0; };
static inline int ras_report_mf_event(struct ras_events *ras, struct ras_mf_event *ev) { return 0; };
static inline int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev) { return 0; };
+static inline int ras_report_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev) { return 0; };
#endif
--
2.25.1
^ permalink raw reply related [flat|nested] 7+ messages in thread* [RESEND PATCH V3 4/4] rasdaemon: Add support for the CXL AER correctable errors
2023-02-02 18:18 [RESEND PATCH V3 0/4] rasdaemon: Add support for the CXL error events shiju.jose
` (2 preceding siblings ...)
2023-02-02 18:18 ` [RESEND PATCH V3 3/4] rasdaemon: Add support for the CXL AER uncorrectable errors shiju.jose
@ 2023-02-02 18:18 ` shiju.jose
3 siblings, 0 replies; 7+ messages in thread
From: shiju.jose @ 2023-02-02 18:18 UTC (permalink / raw)
To: mchehab, linux-edac, linux-cxl; +Cc: jonathan.cameron, linuxarm, shiju.jose
From: Shiju Jose <shiju.jose@huawei.com>
Add support to log and record the CXL AER correctable errors.
The corresponding Kernel patch here:
https://patchwork.kernel.org/project/cxl/patch/166974413388.1608150.5875712482260436188.stgit@djiang5-desk3.ch.intel.com/
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
ras-cxl-handler.c | 68 +++++++++++++++++++++++++++++++++++++++++++++++
ras-cxl-handler.h | 3 +++
ras-events.c | 9 +++++++
ras-events.h | 1 +
ras-record.c | 57 +++++++++++++++++++++++++++++++++++++++
ras-record.h | 10 +++++++
ras-report.c | 67 ++++++++++++++++++++++++++++++++++++++++++++++
ras-report.h | 2 ++
8 files changed, 217 insertions(+)
diff --git a/ras-cxl-handler.c b/ras-cxl-handler.c
index 50bbdb0..5ba350a 100644
--- a/ras-cxl-handler.c
+++ b/ras-cxl-handler.c
@@ -190,6 +190,14 @@ int ras_cxl_poison_event_handler(struct trace_seq *s,
#define CXL_AER_UE_IDE_TX_ERR BIT(15)
#define CXL_AER_UE_IDE_RX_ERR BIT(16)
+#define CXL_AER_CE_CACHE_DATA_ECC BIT(0)
+#define CXL_AER_CE_MEM_DATA_ECC BIT(1)
+#define CXL_AER_CE_CRC_THRESH BIT(2)
+#define CXL_AER_CE_RETRY_THRESH BIT(3)
+#define CXL_AER_CE_CACHE_POISON BIT(4)
+#define CXL_AER_CE_MEM_POISON BIT(5)
+#define CXL_AER_CE_PHYS_LAYER_ERR BIT(6)
+
struct cxl_error_list {
uint32_t bit;
const char *error;
@@ -213,6 +221,16 @@ static const struct cxl_error_list cxl_aer_ue[] = {
{ .bit = CXL_AER_UE_IDE_RX_ERR, .error = "IDE Rx Error" },
};
+static const struct cxl_error_list cxl_aer_ce[] = {
+ { .bit = CXL_AER_CE_CACHE_DATA_ECC, .error = "Cache Data ECC Error" },
+ { .bit = CXL_AER_CE_MEM_DATA_ECC, .error = "Memory Data ECC Error" },
+ { .bit = CXL_AER_CE_CRC_THRESH, .error = "CRC Threshold Hit" },
+ { .bit = CXL_AER_CE_RETRY_THRESH, .error = "Retry Threshold" },
+ { .bit = CXL_AER_CE_CACHE_POISON, .error = "Received Cache Poison From Peer" },
+ { .bit = CXL_AER_CE_MEM_POISON, .error = "Received Memory Poison From Peer" },
+ { .bit = CXL_AER_CE_PHYS_LAYER_ERR, .error = "Received Error From Physical Layer" },
+};
+
static int decode_cxl_error_status(struct trace_seq *s, uint32_t status,
const struct cxl_error_list *cxl_error_list,
uint8_t num_elems)
@@ -308,3 +326,53 @@ int ras_cxl_aer_ue_event_handler(struct trace_seq *s,
return 0;
}
+
+int ras_cxl_aer_ce_event_handler(struct trace_seq *s,
+ struct tep_record *record,
+ struct tep_event *event, void *context)
+{
+ int len;
+ unsigned long long val;
+ time_t now;
+ struct tm *tm;
+ struct ras_events *ras = context;
+ struct ras_cxl_aer_ce_event ev;
+
+ now = record->ts / user_hz + ras->uptime_diff;
+ tm = localtime(&now);
+ if (tm)
+ strftime(ev.timestamp, sizeof(ev.timestamp),
+ "%Y-%m-%d %H:%M:%S %z", tm);
+ else
+ strncpy(ev.timestamp, "1970-01-01 00:00:00 +0000", sizeof(ev.timestamp));
+ if (trace_seq_printf(s, "%s ", ev.timestamp) <= 0)
+ return -1;
+
+ ev.dev_name = tep_get_field_raw(s, event, "dev_name",
+ record, &len, 1);
+ if (!ev.dev_name)
+ return -1;
+ if (trace_seq_printf(s, "dev_name:%s ", ev.dev_name) <= 0)
+ return -1;
+
+ if (tep_get_field_val(s, event, "status", record, &val, 1) < 0)
+ return -1;
+ ev.error_status = val;
+ if (trace_seq_printf(s, "error status:") <= 0)
+ return -1;
+ if (decode_cxl_error_status(s, ev.error_status,
+ cxl_aer_ce, ARRAY_SIZE(cxl_aer_ce)) < 0)
+ return -1;
+
+ /* Insert data into the SGBD */
+#ifdef HAVE_SQLITE3
+ ras_store_cxl_aer_ce_event(ras, &ev);
+#endif
+
+#ifdef HAVE_ABRT_REPORT
+ /* Report event to ABRT */
+ ras_report_cxl_aer_ce_event(ras, &ev);
+#endif
+
+ return 0;
+}
diff --git a/ras-cxl-handler.h b/ras-cxl-handler.h
index 18b3120..711daf4 100644
--- a/ras-cxl-handler.h
+++ b/ras-cxl-handler.h
@@ -26,4 +26,7 @@ int ras_cxl_aer_ue_event_handler(struct trace_seq *s,
struct tep_record *record,
struct tep_event *event, void *context);
+int ras_cxl_aer_ce_event_handler(struct trace_seq *s,
+ struct tep_record *record,
+ struct tep_event *event, void *context);
#endif
diff --git a/ras-events.c b/ras-events.c
index ead792b..3691311 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -247,6 +247,7 @@ int toggle_ras_mc_event(int enable)
#ifdef HAVE_CXL
rc |= __toggle_ras_mc_event(ras, "cxl", "cxl_poison", enable);
rc |= __toggle_ras_mc_event(ras, "cxl", "cxl_aer_uncorrectable_error", enable);
+ rc |= __toggle_ras_mc_event(ras, "cxl", "cxl_aer_correctable_error", enable);
#endif
free_ras:
@@ -973,6 +974,14 @@ int handle_ras_events(int record_events)
else
log(ALL, LOG_ERR, "Can't get traces from %s:%s\n",
"cxl", "cxl_aer_uncorrectable_error");
+
+ rc = add_event_handler(ras, pevent, page_size, "cxl", "cxl_aer_correctable_error",
+ ras_cxl_aer_ce_event_handler, NULL, CXL_AER_CE_EVENT);
+ if (!rc)
+ num_events++;
+ else
+ log(ALL, LOG_ERR, "Can't get traces from %s:%s\n",
+ "cxl", "cxl_aer_correctable_error");
#endif
if (!num_events) {
diff --git a/ras-events.h b/ras-events.h
index 65f9d9a..dc7bdfb 100644
--- a/ras-events.h
+++ b/ras-events.h
@@ -41,6 +41,7 @@ enum {
MF_EVENT,
CXL_POISON_EVENT,
CXL_AER_UE_EVENT,
+ CXL_AER_CE_EVENT,
NR_EVENTS
};
diff --git a/ras-record.c b/ras-record.c
index 4703790..c318a18 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -666,6 +666,48 @@ int ras_store_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_eve
return rc;
}
+/*
+ * Table and functions to handle cxl:cxl_aer_correctable_error
+ */
+static const struct db_fields cxl_aer_ce_event_fields[] = {
+ { .name = "id", .type = "INTEGER PRIMARY KEY" },
+ { .name = "timestamp", .type = "TEXT" },
+ { .name = "dev_name", .type = "TEXT" },
+ { .name = "error_status", .type = "INTEGER" },
+};
+
+static const struct db_table_descriptor cxl_aer_ce_event_tab = {
+ .name = "cxl_aer_ce_event",
+ .fields = cxl_aer_ce_event_fields,
+ .num_fields = ARRAY_SIZE(cxl_aer_ce_event_fields),
+};
+
+int ras_store_cxl_aer_ce_event(struct ras_events *ras, struct ras_cxl_aer_ce_event *ev)
+{
+ int rc;
+ struct sqlite3_priv *priv = ras->db_priv;
+
+ if (!priv || !priv->stmt_cxl_aer_ce_event)
+ return 0;
+ log(TERM, LOG_INFO, "cxl_aer_ce_event store: %p\n", priv->stmt_cxl_aer_ce_event);
+
+ sqlite3_bind_text(priv->stmt_cxl_aer_ce_event, 1, ev->timestamp, -1, NULL);
+ sqlite3_bind_text(priv->stmt_cxl_aer_ce_event, 2, ev->dev_name, -1, NULL);
+ sqlite3_bind_int(priv->stmt_cxl_aer_ce_event, 3, ev->error_status);
+
+ rc = sqlite3_step(priv->stmt_cxl_aer_ce_event);
+ if (rc != SQLITE_OK && rc != SQLITE_DONE)
+ log(TERM, LOG_ERR,
+ "Failed to do cxl_aer_ce_event step on sqlite: error = %d\n", rc);
+ rc = sqlite3_reset(priv->stmt_cxl_aer_ce_event);
+ if (rc != SQLITE_OK && rc != SQLITE_DONE)
+ log(TERM, LOG_ERR,
+ "Failed reset cxl_aer_ce_event on sqlite: error = %d\n",
+ rc);
+ log(TERM, LOG_INFO, "register inserted at db\n");
+
+ return rc;
+}
#endif
/*
@@ -1022,6 +1064,13 @@ int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras)
goto error;
}
+ rc = ras_mc_create_table(priv, &cxl_aer_ce_event_tab);
+ if (rc == SQLITE_OK) {
+ rc = ras_mc_prepare_stmt(priv, &priv->stmt_cxl_aer_ce_event,
+ &cxl_aer_ce_event_tab);
+ if (rc != SQLITE_OK)
+ goto error;
+ }
#endif
ras->db_priv = priv;
@@ -1152,6 +1201,14 @@ int ras_mc_event_closedb(unsigned int cpu, struct ras_events *ras)
"cpu %u: Failed to finalize cxl_aer_ue_event sqlite: error = %d\n",
cpu, rc);
}
+
+ if (priv->stmt_cxl_aer_ce_event) {
+ rc = sqlite3_finalize(priv->stmt_cxl_aer_ce_event);
+ if (rc != SQLITE_OK)
+ log(TERM, LOG_ERR,
+ "cpu %u: Failed to finalize cxl_aer_ce_event sqlite: error = %d\n",
+ cpu, rc);
+ }
#endif
rc = sqlite3_close_v2(db);
diff --git a/ras-record.h b/ras-record.h
index 0e2c178..1f28cc1 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -140,6 +140,12 @@ struct ras_cxl_aer_ue_event {
uint32_t *header_log;
};
+struct ras_cxl_aer_ce_event {
+ char timestamp[64];
+ const char *dev_name;
+ uint32_t error_status;
+};
+
struct ras_mc_event;
struct ras_aer_event;
struct ras_extlog_event;
@@ -151,6 +157,7 @@ struct diskerror_event;
struct ras_mf_event;
struct ras_cxl_poison_event;
struct ras_cxl_aer_ue_event;
+struct ras_cxl_aer_ce_event;
#ifdef HAVE_SQLITE3
@@ -186,6 +193,7 @@ struct sqlite3_priv {
#ifdef HAVE_CXL
sqlite3_stmt *stmt_cxl_poison_event;
sqlite3_stmt *stmt_cxl_aer_ue_event;
+ sqlite3_stmt *stmt_cxl_aer_ce_event;
#endif
};
@@ -216,6 +224,7 @@ int ras_store_diskerror_event(struct ras_events *ras, struct diskerror_event *ev
int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev);
int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev);
int ras_store_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev);
+int ras_store_cxl_aer_ce_event(struct ras_events *ras, struct ras_cxl_aer_ce_event *ev);
#else
static inline int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras) { return 0; };
@@ -231,6 +240,7 @@ static inline int ras_store_diskerror_event(struct ras_events *ras, struct diske
static inline int ras_store_mf_event(struct ras_events *ras, struct ras_mf_event *ev) { return 0; };
static inline int ras_store_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev) { return 0; };
static inline int ras_store_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev) { return 0; };
+static inline int ras_store_cxl_aer_ce_event(struct ras_events *ras, struct ras_cxl_aer_ce_event *ev) { return 0; };
#endif
diff --git a/ras-report.c b/ras-report.c
index 4c09061..796abab 100644
--- a/ras-report.c
+++ b/ras-report.c
@@ -389,6 +389,26 @@ static int set_cxl_aer_ue_event_backtrace(char *buf, struct ras_cxl_aer_ue_event
return 0;
}
+static int set_cxl_aer_ce_event_backtrace(char *buf, struct ras_cxl_aer_ce_event *ev)
+{
+ char bt_buf[MAX_BACKTRACE_SIZE];
+
+ if (!buf || !ev)
+ return -1;
+
+ sprintf(bt_buf, "BACKTRACE=" \
+ "timestamp=%s\n" \
+ "dev_name=%s\n" \
+ "error_status=%u\n" \
+ ev->timestamp, \
+ ev->dev_name, \
+ ev->error_status);
+
+ strcat(buf, bt_buf);
+
+ return 0;
+}
+
static int commit_report_backtrace(int sockfd, int type, void *ev){
char buf[MAX_BACKTRACE_SIZE];
char *pbuf = buf;
@@ -432,6 +452,9 @@ static int commit_report_backtrace(int sockfd, int type, void *ev){
case CXL_AER_UE_EVENT:
rc = set_cxl_aer_ue_event_backtrace(buf, (struct ras_cxl_aer_ue_event *)ev);
break;
+ case CXL_AER_CE_EVENT:
+ rc = set_cxl_aer_ce_event_backtrace(buf, (struct ras_cxl_aer_ce_event *)ev);
+ break;
default:
return -1;
}
@@ -928,3 +951,47 @@ cxl_aer_ue_fail:
else
return -1;
}
+
+int ras_report_cxl_aer_ce_event(struct ras_events *ras, struct ras_cxl_aer_ce_event *ev)
+{
+ char buf[MAX_MESSAGE_SIZE];
+ int sockfd = 0;
+ int done = 0;
+ int rc = -1;
+
+ memset(buf, 0, sizeof(buf));
+
+ sockfd = setup_report_socket();
+ if (sockfd < 0)
+ return -1;
+
+ rc = commit_report_basic(sockfd);
+ if (rc < 0)
+ goto cxl_aer_ce_fail;
+
+ rc = commit_report_backtrace(sockfd, CXL_AER_CE_EVENT, ev);
+ if (rc < 0)
+ goto cxl_aer_ce_fail;
+
+ sprintf(buf, "ANALYZER=%s", "rasdaemon-cxl-aer-correctable-error");
+ rc = write(sockfd, buf, strlen(buf) + 1);
+ if (rc < strlen(buf) + 1)
+ goto cxl_aer_ce_fail;
+
+ sprintf(buf, "REASON=%s", "CXL AER correctable error");
+ rc = write(sockfd, buf, strlen(buf) + 1);
+ if (rc < strlen(buf) + 1)
+ goto cxl_aer_ce_fail;
+
+ done = 1;
+
+cxl_aer_ce_fail:
+
+ if (sockfd >= 0)
+ close(sockfd);
+
+ if (done)
+ return 0;
+ else
+ return -1;
+}
diff --git a/ras-report.h b/ras-report.h
index dfe89d1..46155ee 100644
--- a/ras-report.h
+++ b/ras-report.h
@@ -41,6 +41,7 @@ int ras_report_diskerror_event(struct ras_events *ras, struct diskerror_event *e
int ras_report_mf_event(struct ras_events *ras, struct ras_mf_event *ev);
int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev);
int ras_report_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev);
+int ras_report_cxl_aer_ce_event(struct ras_events *ras, struct ras_cxl_aer_ce_event *ev);
#else
@@ -54,6 +55,7 @@ static inline int ras_report_diskerror_event(struct ras_events *ras, struct disk
static inline int ras_report_mf_event(struct ras_events *ras, struct ras_mf_event *ev) { return 0; };
static inline int ras_report_cxl_poison_event(struct ras_events *ras, struct ras_cxl_poison_event *ev) { return 0; };
static inline int ras_report_cxl_aer_ue_event(struct ras_events *ras, struct ras_cxl_aer_ue_event *ev) { return 0; };
+static inline int ras_report_cxl_aer_ce_event(struct ras_events *ras, struct ras_cxl_aer_ce_event *ev) { return 0; };
#endif
--
2.25.1
^ permalink raw reply related [flat|nested] 7+ messages in thread