linux-edac.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH V4 0/4] rasdaemon: Add support for the CXL error events
@ 2023-02-14 11:21 shiju.jose
  2023-02-14 11:21 ` [PATCH V4 1/4] rasdaemon: Move definition for BIT and BIT_ULL to a common file shiju.jose
                   ` (3 more replies)
  0 siblings, 4 replies; 7+ messages in thread
From: shiju.jose @ 2023-02-14 11:21 UTC (permalink / raw)
  To: mchehab, linux-edac, linux-cxl; +Cc: jonathan.cameron, linuxarm, shiju.jose

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="yes", Size: 2340 bytes --]

From: Shiju Jose <shiju.jose@huawei.com>

Log and record the following CXL errors reported through the kernel
trace events. CXL poison errors, CXL AER uncorrectable errors and CXL AER
correctable errors.

Note1: The default poll method in the rasdaemon to receive
the trace events does not work due to a commit in the kernel trace system.
Solution 1:
3e46d910d8acf94e5360126593b68bf4fee4c4a1("tracing: Fix poll() and select()
 do not work on per_cpu trace_pipe an……d trace_pipe_raw") in Linux 6.2-rc7
 or later. and
https://lore.kernel.org/lkml/20230204193345.842-1-shiju.jose@huawei.com/T/

Solution 2: 
Thus instead used the pthread way for testing the CXL error events.
To do so, please make following change in the ras-events.c
<change start ...>
/* rc = read_ras_event_all_cpus(data, cpus); */
rc = -255;
< ...change end >
/* Poll doesn't work on this kernel. Fallback to pthread way */
if (rc == -255) {
...

Shiju Jose (4):
  rasdaemon: Move definition for BIT and BIT_ULL to a common file
  rasdaemon: Add support for the CXL poison events
  rasdaemon: Add support for the CXL AER uncorrectable errors
  rasdaemon: Add support for the CXL AER correctable errors

Changes:
V3 -> V4
1. Modifications for the changes in the kernel patches
   a) https://lore.kernel.org/lkml/cover.1675983077.git.alison.schofield@intel.com/
   b) https://lore.kernel.org/linux-cxl/63e5ed38d77d9_138fbc2947a@iweiny-mobl.notmuch/T/#t

V2 -> V3
1. Fix for the comments from Dave Jiang.

RFC V1 -> V2
1. Rename uuid to region_uuid in the log and SQLite DB.
2. Rebase to the latest rasdaemon code.
3. Modify to match the name changes of interface structures and
   functions in the latest libtraceevent-dev, use in the rasdaemon. 


 Makefile.am                |   7 +-
 configure.ac               |  11 +
 ras-cxl-handler.c          | 398 +++++++++++++++++++++++++++++++++++++
 ras-cxl-handler.h          |  32 +++
 ras-events.c               |  33 +++
 ras-events.h               |   3 +
 ras-non-standard-handler.h |   3 -
 ras-record.c               | 209 +++++++++++++++++++
 ras-record.h               |  52 +++++
 ras-report.c               | 225 +++++++++++++++++++++
 ras-report.h               |   6 +
 11 files changed, 975 insertions(+), 4 deletions(-)
 create mode 100644 ras-cxl-handler.c
 create mode 100644 ras-cxl-handler.h

-- 
2.25.1


^ permalink raw reply	[flat|nested] 7+ messages in thread

* [PATCH V4 1/4] rasdaemon: Move definition for BIT and BIT_ULL to a common file
  2023-02-14 11:21 [PATCH V4 0/4] rasdaemon: Add support for the CXL error events shiju.jose
@ 2023-02-14 11:21 ` shiju.jose
  2023-02-14 11:21 ` [PATCH V4 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 @ 2023-02-14 11:21 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

* [PATCH V4 2/4] rasdaemon: Add support for the CXL poison events
  2023-02-14 11:21 [PATCH V4 0/4] rasdaemon: Add support for the CXL error events shiju.jose
  2023-02-14 11:21 ` [PATCH V4 1/4] rasdaemon: Move definition for BIT and BIT_ULL to a common file shiju.jose
@ 2023-02-14 11:21 ` shiju.jose
  2023-02-14 11:21 ` [PATCH V4 3/4] rasdaemon: Add support for the CXL AER uncorrectable errors shiju.jose
  2023-02-14 11:21 ` [PATCH V4 4/4] rasdaemon: Add support for the CXL AER correctable errors shiju.jose
  3 siblings, 0 replies; 7+ messages in thread
From: shiju.jose @ 2023-02-14 11:21 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/lkml/cover.1675983077.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 | 178 ++++++++++++++++++++++++++++++++++++++++++++++
 ras-cxl-handler.h |  24 +++++++
 ras-events.c      |  15 ++++
 ras-events.h      |   1 +
 ras-record.c      |  83 +++++++++++++++++++++
 ras-record.h      |  21 ++++++
 ras-report.c      |  85 ++++++++++++++++++++++
 ras-report.h      |   2 +
 10 files changed, 426 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..b6175bf
--- /dev/null
+++ b/ras-cxl-handler.c
@@ -0,0 +1,178 @@
+/*
+ * 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.host = tep_get_field_raw(s, event, "host",
+				    record, &len, 1);
+	if (!ev.host)
+		return -1;
+	if (trace_seq_printf(s, "host:%s ", ev.host) <= 0)
+		return -1;
+
+	if (tep_get_field_val(s, event, "serial", record, &val, 1) < 0)
+		return -1;
+	ev.serial = val;
+	if (trace_seq_printf(s, "serial:0x%llx ", (unsigned long long)ev.serial) <= 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..9db5d93 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -559,6 +559,69 @@ 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 = "host",                 .type = "TEXT" },
+	{ .name = "serial",               .type = "INTEGER" },
+	{ .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->host, -1, NULL);
+	sqlite3_bind_int64(priv->stmt_cxl_poison_event, 4, ev->serial);
+	sqlite3_bind_text(priv->stmt_cxl_poison_event, 5, ev->region, -1, NULL);
+	sqlite3_bind_text(priv->stmt_cxl_poison_event, 6, ev->uuid, -1, NULL);
+	sqlite3_bind_int64(priv->stmt_cxl_poison_event, 7, ev->hpa);
+	sqlite3_bind_int64(priv->stmt_cxl_poison_event, 8, ev->dpa);
+	sqlite3_bind_int(priv->stmt_cxl_poison_event, 9, ev->length);
+	sqlite3_bind_text(priv->stmt_cxl_poison_event, 10, ev->source, -1, NULL);
+	sqlite3_bind_int(priv->stmt_cxl_poison_event, 11, ev->flags);
+	sqlite3_bind_text(priv->stmt_cxl_poison_event, 12, 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 +959,16 @@ 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 +1081,16 @@ 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..9ede444 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -114,6 +114,21 @@ struct ras_mf_event {
 	const char *action_result;
 };
 
+struct ras_cxl_poison_event {
+	char timestamp[64];
+	const char *memdev;
+	const char *host;
+	uint64_t serial;
+	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 +138,7 @@ struct mce_event;
 struct devlink_event;
 struct diskerror_event;
 struct ras_mf_event;
+struct ras_cxl_poison_event;
 
 #ifdef HAVE_SQLITE3
 
@@ -155,6 +171,9 @@ 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 +201,7 @@ 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 +215,7 @@ 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..b1b7614 100644
--- a/ras-report.c
+++ b/ras-report.c
@@ -331,6 +331,44 @@ 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"		\
+						"host=%s\n"		\
+						"serial=0x%lx\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->host,		\
+						ev->serial,		\
+						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 +406,9 @@ 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 +817,47 @@ 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

* [PATCH V4 3/4] rasdaemon: Add support for the CXL AER uncorrectable errors
  2023-02-14 11:21 [PATCH V4 0/4] rasdaemon: Add support for the CXL error events shiju.jose
  2023-02-14 11:21 ` [PATCH V4 1/4] rasdaemon: Move definition for BIT and BIT_ULL to a common file shiju.jose
  2023-02-14 11:21 ` [PATCH V4 2/4] rasdaemon: Add support for the CXL poison events shiju.jose
@ 2023-02-14 11:21 ` shiju.jose
  2023-02-17 16:58   ` Dave Jiang
  2023-02-14 11:21 ` [PATCH V4 4/4] rasdaemon: Add support for the CXL AER correctable errors shiju.jose
  3 siblings, 1 reply; 7+ messages in thread
From: shiju.jose @ 2023-02-14 11:21 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 patches are here:
https://lore.kernel.org/linux-cxl/166974401763.1608150.5424589924034481387.stgit@djiang5-desk3.ch.intel.com/T/#t
https://lore.kernel.org/linux-cxl/63e5ed38d77d9_138fbc2947a@iweiny-mobl.notmuch/T/#t

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 | 145 ++++++++++++++++++++++++++++++++++++++++++++++
 ras-cxl-handler.h |   5 ++
 ras-events.c      |   9 +++
 ras-events.h      |   1 +
 ras-record.c      |  67 +++++++++++++++++++++
 ras-record.h      |  17 ++++++
 ras-report.c      |  71 +++++++++++++++++++++++
 ras-report.h      |   2 +
 8 files changed, 317 insertions(+)

diff --git a/ras-cxl-handler.c b/ras-cxl-handler.c
index b6175bf..3bcefd6 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)
@@ -176,3 +177,147 @@ 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.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.host = tep_get_field_raw(s, event, "host",
+				    record, &len, 1);
+	if (!ev.host)
+		return -1;
+	if (trace_seq_printf(s, "host:%s ", ev.host) <= 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 9db5d93..bd297ad 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -620,6 +620,56 @@ 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 = "memdev",               .type = "TEXT" },
+	{ .name = "host",                 .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->memdev, -1, NULL);
+	sqlite3_bind_text(priv->stmt_cxl_aer_ue_event, 3, ev->host, -1, NULL);
+	sqlite3_bind_int(priv->stmt_cxl_aer_ue_event, 4, ev->error_status);
+	sqlite3_bind_int(priv->stmt_cxl_aer_ue_event, 5, ev->first_error);
+	sqlite3_bind_blob(priv->stmt_cxl_aer_ue_event, 6, 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
 
 /*
@@ -967,6 +1017,15 @@ 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;
@@ -1089,6 +1148,14 @@ 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 9ede444..e2dcd19 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -129,6 +129,19 @@ 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 *memdev;
+	const char *host;
+	uint32_t error_status;
+	uint32_t first_error;
+	uint32_t *header_log;
+};
+
 struct ras_mc_event;
 struct ras_aer_event;
 struct ras_extlog_event;
@@ -139,6 +152,7 @@ struct devlink_event;
 struct diskerror_event;
 struct ras_mf_event;
 struct ras_cxl_poison_event;
+struct ras_cxl_aer_ue_event;
 
 #ifdef HAVE_SQLITE3
 
@@ -173,6 +187,7 @@ struct sqlite3_priv {
 #endif
 #ifdef HAVE_CXL
 	sqlite3_stmt	*stmt_cxl_poison_event;
+	sqlite3_stmt	*stmt_cxl_aer_ue_event;
 #endif
 };
 
@@ -202,6 +217,7 @@ 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; };
@@ -216,6 +232,7 @@ 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 b1b7614..b3cc7dc 100644
--- a/ras-report.c
+++ b/ras-report.c
@@ -369,6 +369,30 @@ 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"	\
+						"memdev=%s\n"		\
+						"host=%s\n"		\
+						"error_status=%u\n"	\
+						"first_error=%u\n"	\
+						ev->timestamp,		\
+						ev->memdev,		\
+						ev->host,		\
+						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;
@@ -409,6 +433,9 @@ 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;
 	}
@@ -861,3 +888,47 @@ 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

* [PATCH V4 4/4] rasdaemon: Add support for the CXL AER correctable errors
  2023-02-14 11:21 [PATCH V4 0/4] rasdaemon: Add support for the CXL error events shiju.jose
                   ` (2 preceding siblings ...)
  2023-02-14 11:21 ` [PATCH V4 3/4] rasdaemon: Add support for the CXL AER uncorrectable errors shiju.jose
@ 2023-02-14 11:21 ` shiju.jose
  2023-02-17 16:59   ` Dave Jiang
  3 siblings, 1 reply; 7+ messages in thread
From: shiju.jose @ 2023-02-14 11:21 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 patches are here:
https://lore.kernel.org/linux-cxl/166974401763.1608150.5424589924034481387.stgit@djiang5-desk3.ch.intel.com/T/#t
https://lore.kernel.org/linux-cxl/63e5ed38d77d9_138fbc2947a@iweiny-mobl.notmuch/T/#t

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
 ras-cxl-handler.c | 75 +++++++++++++++++++++++++++++++++++++++++++++++
 ras-cxl-handler.h |  3 ++
 ras-events.c      |  9 ++++++
 ras-events.h      |  1 +
 ras-record.c      | 59 +++++++++++++++++++++++++++++++++++++
 ras-record.h      | 11 +++++++
 ras-report.c      | 69 +++++++++++++++++++++++++++++++++++++++++++
 ras-report.h      |  2 ++
 8 files changed, 229 insertions(+)

diff --git a/ras-cxl-handler.c b/ras-cxl-handler.c
index 3bcefd6..b344b58 100644
--- a/ras-cxl-handler.c
+++ b/ras-cxl-handler.c
@@ -196,6 +196,14 @@ 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;
@@ -219,6 +227,16 @@ 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)
@@ -321,3 +339,60 @@ 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.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.host = tep_get_field_raw(s, event, "host",
+				    record, &len, 1);
+	if (!ev.host)
+		return -1;
+	if (trace_seq_printf(s, "host:%s ", ev.host) <= 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 bd297ad..7ff7298 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -670,6 +670,50 @@ 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 = "memdev",               .type = "TEXT" },
+	{ .name = "host",                 .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->memdev, -1, NULL);
+	sqlite3_bind_text(priv->stmt_cxl_aer_ce_event, 3, ev->host, -1, NULL);
+	sqlite3_bind_int(priv->stmt_cxl_aer_ce_event, 4, 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
 
 /*
@@ -1026,6 +1070,13 @@ 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;
@@ -1156,6 +1207,14 @@ 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 e2dcd19..2d56a48 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -142,6 +142,13 @@ struct ras_cxl_aer_ue_event {
 	uint32_t *header_log;
 };
 
+struct ras_cxl_aer_ce_event {
+	char timestamp[64];
+	const char *memdev;
+	const char *host;
+	uint32_t error_status;
+};
+
 struct ras_mc_event;
 struct ras_aer_event;
 struct ras_extlog_event;
@@ -153,6 +160,7 @@ 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
 
@@ -188,6 +196,7 @@ 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
 };
 
@@ -218,6 +227,7 @@ 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; };
@@ -233,6 +243,7 @@ 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 b3cc7dc..17fa66b 100644
--- a/ras-report.c
+++ b/ras-report.c
@@ -393,6 +393,28 @@ 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"	\
+						"memdev=%s\n"		\
+						"host=%s\n"		\
+						"error_status=%u\n"	\
+						ev->timestamp,		\
+						ev->memdev,		\
+						ev->host,		\
+						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;
@@ -436,6 +458,9 @@ 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;
 	}
@@ -932,3 +957,47 @@ 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

* Re: [PATCH V4 3/4] rasdaemon: Add support for the CXL AER uncorrectable errors
  2023-02-14 11:21 ` [PATCH V4 3/4] rasdaemon: Add support for the CXL AER uncorrectable errors shiju.jose
@ 2023-02-17 16:58   ` Dave Jiang
  0 siblings, 0 replies; 7+ messages in thread
From: Dave Jiang @ 2023-02-17 16:58 UTC (permalink / raw)
  To: shiju.jose, mchehab, linux-edac, linux-cxl; +Cc: jonathan.cameron, linuxarm



On 2/14/23 4:21 AM, shiju.jose@huawei.com wrote:
> From: Shiju Jose <shiju.jose@huawei.com>
> 
> Add support to log and record the CXL AER uncorrectable errors.
> 
> The corresponding Kernel patches are here:
> https://lore.kernel.org/linux-cxl/166974401763.1608150.5424589924034481387.stgit@djiang5-desk3.ch.intel.com/T/#t
> https://lore.kernel.org/linux-cxl/63e5ed38d77d9_138fbc2947a@iweiny-mobl.notmuch/T/#t
> 
> 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>

Reviewed-by: Dave Jiang <dave.jiang@intel.com>

> ---
>   ras-cxl-handler.c | 145 ++++++++++++++++++++++++++++++++++++++++++++++
>   ras-cxl-handler.h |   5 ++
>   ras-events.c      |   9 +++
>   ras-events.h      |   1 +
>   ras-record.c      |  67 +++++++++++++++++++++
>   ras-record.h      |  17 ++++++
>   ras-report.c      |  71 +++++++++++++++++++++++
>   ras-report.h      |   2 +
>   8 files changed, 317 insertions(+)
> 
> diff --git a/ras-cxl-handler.c b/ras-cxl-handler.c
> index b6175bf..3bcefd6 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)
> @@ -176,3 +177,147 @@ 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.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.host = tep_get_field_raw(s, event, "host",
> +				    record, &len, 1);
> +	if (!ev.host)
> +		return -1;
> +	if (trace_seq_printf(s, "host:%s ", ev.host) <= 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 9db5d93..bd297ad 100644
> --- a/ras-record.c
> +++ b/ras-record.c
> @@ -620,6 +620,56 @@ 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 = "memdev",               .type = "TEXT" },
> +	{ .name = "host",                 .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->memdev, -1, NULL);
> +	sqlite3_bind_text(priv->stmt_cxl_aer_ue_event, 3, ev->host, -1, NULL);
> +	sqlite3_bind_int(priv->stmt_cxl_aer_ue_event, 4, ev->error_status);
> +	sqlite3_bind_int(priv->stmt_cxl_aer_ue_event, 5, ev->first_error);
> +	sqlite3_bind_blob(priv->stmt_cxl_aer_ue_event, 6, 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
>   
>   /*
> @@ -967,6 +1017,15 @@ 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;
> @@ -1089,6 +1148,14 @@ 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 9ede444..e2dcd19 100644
> --- a/ras-record.h
> +++ b/ras-record.h
> @@ -129,6 +129,19 @@ 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 *memdev;
> +	const char *host;
> +	uint32_t error_status;
> +	uint32_t first_error;
> +	uint32_t *header_log;
> +};
> +
>   struct ras_mc_event;
>   struct ras_aer_event;
>   struct ras_extlog_event;
> @@ -139,6 +152,7 @@ struct devlink_event;
>   struct diskerror_event;
>   struct ras_mf_event;
>   struct ras_cxl_poison_event;
> +struct ras_cxl_aer_ue_event;
>   
>   #ifdef HAVE_SQLITE3
>   
> @@ -173,6 +187,7 @@ struct sqlite3_priv {
>   #endif
>   #ifdef HAVE_CXL
>   	sqlite3_stmt	*stmt_cxl_poison_event;
> +	sqlite3_stmt	*stmt_cxl_aer_ue_event;
>   #endif
>   };
>   
> @@ -202,6 +217,7 @@ 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; };
> @@ -216,6 +232,7 @@ 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 b1b7614..b3cc7dc 100644
> --- a/ras-report.c
> +++ b/ras-report.c
> @@ -369,6 +369,30 @@ 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"	\
> +						"memdev=%s\n"		\
> +						"host=%s\n"		\
> +						"error_status=%u\n"	\
> +						"first_error=%u\n"	\
> +						ev->timestamp,		\
> +						ev->memdev,		\
> +						ev->host,		\
> +						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;
> @@ -409,6 +433,9 @@ 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;
>   	}
> @@ -861,3 +888,47 @@ 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
>   

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH V4 4/4] rasdaemon: Add support for the CXL AER correctable errors
  2023-02-14 11:21 ` [PATCH V4 4/4] rasdaemon: Add support for the CXL AER correctable errors shiju.jose
@ 2023-02-17 16:59   ` Dave Jiang
  0 siblings, 0 replies; 7+ messages in thread
From: Dave Jiang @ 2023-02-17 16:59 UTC (permalink / raw)
  To: shiju.jose, mchehab, linux-edac, linux-cxl; +Cc: jonathan.cameron, linuxarm



On 2/14/23 4:21 AM, shiju.jose@huawei.com wrote:
> From: Shiju Jose <shiju.jose@huawei.com>
> 
> Add support to log and record the CXL AER correctable errors.
> 
> The corresponding Kernel patches are here:
> https://lore.kernel.org/linux-cxl/166974401763.1608150.5424589924034481387.stgit@djiang5-desk3.ch.intel.com/T/#t
> https://lore.kernel.org/linux-cxl/63e5ed38d77d9_138fbc2947a@iweiny-mobl.notmuch/T/#t
> 
> 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-cxl-handler.c | 75 +++++++++++++++++++++++++++++++++++++++++++++++
>   ras-cxl-handler.h |  3 ++
>   ras-events.c      |  9 ++++++
>   ras-events.h      |  1 +
>   ras-record.c      | 59 +++++++++++++++++++++++++++++++++++++
>   ras-record.h      | 11 +++++++
>   ras-report.c      | 69 +++++++++++++++++++++++++++++++++++++++++++
>   ras-report.h      |  2 ++
>   8 files changed, 229 insertions(+)
> 
> diff --git a/ras-cxl-handler.c b/ras-cxl-handler.c
> index 3bcefd6..b344b58 100644
> --- a/ras-cxl-handler.c
> +++ b/ras-cxl-handler.c
> @@ -196,6 +196,14 @@ 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;
> @@ -219,6 +227,16 @@ 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)
> @@ -321,3 +339,60 @@ 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.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.host = tep_get_field_raw(s, event, "host",
> +				    record, &len, 1);
> +	if (!ev.host)
> +		return -1;
> +	if (trace_seq_printf(s, "host:%s ", ev.host) <= 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 bd297ad..7ff7298 100644
> --- a/ras-record.c
> +++ b/ras-record.c
> @@ -670,6 +670,50 @@ 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 = "memdev",               .type = "TEXT" },
> +	{ .name = "host",                 .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->memdev, -1, NULL);
> +	sqlite3_bind_text(priv->stmt_cxl_aer_ce_event, 3, ev->host, -1, NULL);
> +	sqlite3_bind_int(priv->stmt_cxl_aer_ce_event, 4, 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
>   
>   /*
> @@ -1026,6 +1070,13 @@ 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;
> @@ -1156,6 +1207,14 @@ 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 e2dcd19..2d56a48 100644
> --- a/ras-record.h
> +++ b/ras-record.h
> @@ -142,6 +142,13 @@ struct ras_cxl_aer_ue_event {
>   	uint32_t *header_log;
>   };
>   
> +struct ras_cxl_aer_ce_event {
> +	char timestamp[64];
> +	const char *memdev;
> +	const char *host;
> +	uint32_t error_status;
> +};
> +
>   struct ras_mc_event;
>   struct ras_aer_event;
>   struct ras_extlog_event;
> @@ -153,6 +160,7 @@ 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
>   
> @@ -188,6 +196,7 @@ 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
>   };
>   
> @@ -218,6 +227,7 @@ 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; };
> @@ -233,6 +243,7 @@ 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 b3cc7dc..17fa66b 100644
> --- a/ras-report.c
> +++ b/ras-report.c
> @@ -393,6 +393,28 @@ 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"	\
> +						"memdev=%s\n"		\
> +						"host=%s\n"		\
> +						"error_status=%u\n"	\
> +						ev->timestamp,		\
> +						ev->memdev,		\
> +						ev->host,		\
> +						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;
> @@ -436,6 +458,9 @@ 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;
>   	}
> @@ -932,3 +957,47 @@ 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
>   

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2023-02-17 16:59 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-02-14 11:21 [PATCH V4 0/4] rasdaemon: Add support for the CXL error events shiju.jose
2023-02-14 11:21 ` [PATCH V4 1/4] rasdaemon: Move definition for BIT and BIT_ULL to a common file shiju.jose
2023-02-14 11:21 ` [PATCH V4 2/4] rasdaemon: Add support for the CXL poison events shiju.jose
2023-02-14 11:21 ` [PATCH V4 3/4] rasdaemon: Add support for the CXL AER uncorrectable errors shiju.jose
2023-02-17 16:58   ` Dave Jiang
2023-02-14 11:21 ` [PATCH V4 4/4] rasdaemon: Add support for the CXL AER correctable errors shiju.jose
2023-02-17 16:59   ` Dave Jiang

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).