kselftest: devices: Add test to detect device error logs

Log errors are the most widely used mechanism for reporting issues in
the kernel. When an error is logged using the device helpers, eg
dev_err(), it gets metadata attached that identifies the subsystem and
device where the message is coming from. Introduce a new test that makes
use of that metadata to report which devices logged errors (or more
critical messages).

Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
Acked-by: Shuah Khan <skhan@linuxfoundation.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://lore.kernel.org/r/20240705-dev-err-log-selftest-v2-3-163b9cd7b3c1@collabora.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

3 files changed, 89 insertions, 0 deletions

diff --git a/tools/testing/selftests/Makefile b/tools/testing/selftests/Makefile
index 7bd78b9f5cdd..c4937c87df22 100644
--- a/tools/testing/selftests/Makefile
+++ b/tools/testing/selftests/Makefile
@@ -13,6 +13,7 @@ TARGETS += core
 TARGETS += cpufreq
 TARGETS += cpu-hotplug
 TARGETS += damon
+TARGETS += devices/error_logs
 TARGETS += devices/probe
 TARGETS += dmabuf-heaps
 TARGETS += drivers/dma-buf
diff --git a/tools/testing/selftests/devices/error_logs/Makefile b/tools/testing/selftests/devices/error_logs/Makefile
new file mode 100644
index 000000000000..d546c3fb0a7f
--- /dev/null
+++ b/tools/testing/selftests/devices/error_logs/Makefile
@@ -0,0 +1,3 @@
+TEST_PROGS := test_device_error_logs.py
+
+include ../../lib.mk
diff --git a/tools/testing/selftests/devices/error_logs/test_device_error_logs.py b/tools/testing/selftests/devices/error_logs/test_device_error_logs.py
new file mode 100755
index 000000000000..3dd56c8ec92c
--- /dev/null
+++ b/tools/testing/selftests/devices/error_logs/test_device_error_logs.py
@@ -0,0 +1,85 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: GPL-2.0
+#
+# Copyright (c) 2024 Collabora Ltd
+#
+# This test checks for the presence of error (or more critical) log messages
+# coming from devices in the kernel log.
+#
+# One failed test case is reported for each device that has outputted error
+# logs. Devices with no errors do not produce a passing test case to avoid
+# polluting the results, therefore a successful run will list 0 tests run.
+#
+
+import glob
+import os
+import re
+import sys
+
+# Allow ksft module to be imported from different directory
+this_dir = os.path.dirname(os.path.realpath(__file__))
+sys.path.append(os.path.join(this_dir, "../../kselftest/"))
+
+import ksft
+
+kmsg = "/dev/kmsg"
+
+RE_log = re.compile(
+    r"(?P<prefix>[0-9]+),(?P<sequence>[0-9]+),(?P<timestamp>[0-9]+),(?P<flag>[^;]*)(,[^;]*)*;(?P<message>.*)"
+)
+RE_tag = re.compile(r" (?P<key>[^=]+)=(?P<value>.*)")
+
+PREFIX_ERROR = 3
+
+logs = []
+error_log_per_device = {}
+
+
+def parse_kmsg():
+    current_log = {}
+
+    with open(kmsg) as f:
+        os.set_blocking(f.fileno(), False)
+
+        for line in f:
+            tag_line = RE_tag.match(line)
+            log_line = RE_log.match(line)
+
+            if log_line:
+                if current_log:
+                    logs.append(current_log)  # Save last log
+
+                current_log = {
+                    "prefix": int(log_line.group("prefix")),
+                    "sequence": int(log_line.group("sequence")),
+                    "timestamp": int(log_line.group("timestamp")),
+                    "flag": log_line.group("flag"),
+                    "message": log_line.group("message"),
+                }
+            elif tag_line:
+                current_log[tag_line.group("key")] = tag_line.group("value")
+
+
+def generate_per_device_error_log():
+    for log in logs:
+        if log.get("DEVICE") and log["prefix"] <= PREFIX_ERROR:
+            if not error_log_per_device.get(log["DEVICE"]):
+                error_log_per_device[log["DEVICE"]] = []
+            error_log_per_device[log["DEVICE"]].append(log)
+
+
+parse_kmsg()
+
+generate_per_device_error_log()
+num_tests = len(error_log_per_device)
+
+ksft.print_header()
+ksft.set_plan(num_tests)
+
+for device in error_log_per_device:
+    for log in error_log_per_device[device]:
+        ksft.print_msg(log["message"])
+    ksft.test_result_fail(device)
+if num_tests == 0:
+    ksft.print_msg("No device error logs found")
+ksft.finished()