lib: Process kernel taints

A small library routine to read '/proc/sys/kernel/taints' and check for
a fatal condition. This is currently used by the runner, but is also
useful for some tests.

v2,3: function docs

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Petri Latvala <petri.latvala@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>

5 files changed, 140 insertions, 59 deletions

diff --git a/lib/Makefile.sources b/lib/Makefile.sources
index 67b38645..7102f95e 100644
--- a/lib/Makefile.sources
+++ b/lib/Makefile.sources
@@ -64,6 +64,8 @@ lib_source_list =	 	\
 	igt_sysfs.h		\
 	igt_sysrq.c		\
 	igt_sysrq.h		\
+	igt_taints.c		\
+	igt_taints.h		\
 	igt_thread.c		\
 	igt_thread.h		\
 	igt_x86.h		\
diff --git a/lib/igt_taints.c b/lib/igt_taints.c
new file mode 100644
index 00000000..6f5b827c
--- /dev/null
+++ b/lib/igt_taints.c
@@ -0,0 +1,104 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright © 2021 Intel Corporation
+ */
+
+#include <stdio.h>
+
+#include "igt_taints.h"
+
+/* see Linux's include/linux/kernel.h */
+static const struct {
+	int bit;
+	int bad;
+	const char *explanation;
+} abort_taints[] = {
+  { 5, 1, "TAINT_BAD_PAGE: Bad page reference or an unexpected page flags." },
+  { 7, 1, "TAINT_DIE: Kernel has died - BUG/OOPS." },
+  { 9, 1, "TAINT_WARN: WARN_ON has happened." },
+  { -1 }
+};
+
+/**
+ * igt_explain_taints:
+ * @taints: mask of taints requiring an explanation [inout]
+ *
+ * Inspects the mask and looks up the first reason corresponding to a set
+ * bit in the mask. It returns the reason as a string constant, and removes
+ * the bit from the mask. If the mask is empty, or we have no known reason
+ * matching the mask, NULL is returned.
+ *
+ * This may be used in a loop to extract all known reasons for why the
+ * kernel is tainted:
+ *
+ * while (reason = igt_explain_taints(&taints))
+ * 	igt_info("%s", reason);
+ *
+ * Returns the first reason corresponding to a taint bit.
+ */
+const char *igt_explain_taints(unsigned long *taints)
+{
+	for (typeof(*abort_taints) *taint = abort_taints;
+	     taint->bit >= 0;
+	     taint++) {
+		if (*taints & (1ul << taint->bit)) {
+			*taints &= ~(1ul << taint->bit);
+			return taint->explanation;
+		}
+	}
+
+	return NULL;
+}
+
+/**
+ * igt_bad_taints:
+ *
+ * Returns the mask of kernel taints that IGT considers fatal.
+ * Such as TAINT_WARN set when the kernel oopses.
+ */
+unsigned long igt_bad_taints(void)
+{
+	static unsigned long bad_taints;
+
+	if (!bad_taints) {
+		for (typeof(*abort_taints) *taint = abort_taints;
+		     taint->bit >= 0;
+		     taint++) {
+			if (taint->bad)
+				bad_taints |= 1ul << taint->bit;
+		}
+	}
+
+	return bad_taints;
+}
+
+/**
+ * igt_kernel_tainted:
+ * @taints: bitmask of kernel taints [out]
+ *
+ * Reads the bitmask of kernel taints from "/proc/sys/kernel/tainted",
+ * see linux/kernel.h for the full set of flags. These are set whenever
+ * the kernel encounters an exceptional condition that may impair functionality.
+ * The kernel only sets the taint once, and so once a "fatal" condition has
+ * been encountered, it is generally not advisable to continue testing, as at
+ * least all future taint reporting will be lost.
+ *
+ * igt_kernel_tainted() returns the set of _all_ taints reported via @taints,
+ * and also the set of _fatal_ taints as its return value.
+ *
+ * Returns a mask of fatal taints; 0 if untainted.
+ */
+unsigned long igt_kernel_tainted(unsigned long *taints)
+{
+	FILE *f;
+
+	*taints = 0;
+
+	f = fopen("/proc/sys/kernel/tainted", "r");
+	if (f) {
+		fscanf(f, "%lu", taints);
+		fclose(f);
+	}
+
+	return is_tainted(*taints);
+}
diff --git a/lib/igt_taints.h b/lib/igt_taints.h
new file mode 100644
index 00000000..be4195c5
--- /dev/null
+++ b/lib/igt_taints.h
@@ -0,0 +1,19 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright © 2021 Intel Corporation
+ */
+
+#ifndef __IGT_TAINTS_H__
+#define __IGT_TAINTS_H__
+
+unsigned long igt_kernel_tainted(unsigned long *taints);
+const char *igt_explain_taints(unsigned long *taints);
+
+unsigned long igt_bad_taints(void);
+
+static inline unsigned long is_tainted(unsigned long taints)
+{
+	return taints & igt_bad_taints();
+}
+
+#endif /* __IGT_TAINTS_H__ */
diff --git a/lib/meson.build b/lib/meson.build
index 540facb2..3abc42cb 100644
--- a/lib/meson.build
+++ b/lib/meson.build
@@ -27,6 +27,7 @@ lib_sources = [
 	'igt_syncobj.c',
 	'igt_sysfs.c',
 	'igt_sysrq.c',
+	'igt_taints.c',
 	'igt_thread.c',
 	'igt_vec.c',
 	'igt_vgem.c',
diff --git a/runner/executor.c b/runner/executor.c
index faf272d8..93db8bb3 100644
--- a/runner/executor.c
+++ b/runner/executor.c
@@ -24,6 +24,7 @@
 
 #include "igt_aux.h"
 #include "igt_core.h"
+#include "igt_taints.h"
 #include "executor.h"
 #include "output_strings.h"
 
@@ -307,70 +308,23 @@ static char *handle_lockdep(void)
 	return NULL;
 }
 
-/* see Linux's include/linux/kernel.h */
-static const struct {
-	unsigned long bit;
-	const char *explanation;
-} abort_taints[] = {
-  {(1 << 5), "TAINT_BAD_PAGE: Bad page reference or an unexpected page flags."},
-  {(1 << 7), "TAINT_DIE: Kernel has died - BUG/OOPS."},
-  {(1 << 9), "TAINT_WARN: WARN_ON has happened."},
-  {0, 0}};
-
-static unsigned long bad_taints(void)
-{
-	static unsigned long __bad_taints;
-
-	if (!__bad_taints) {
-		for (typeof(*abort_taints) *taint = abort_taints;
-		     taint->bit;
-		     taint++)
-			__bad_taints |= taint->bit;
-	}
-
-	return __bad_taints;
-}
-
-static unsigned long is_tainted(unsigned long taints)
-{
-	return taints & bad_taints();
-}
-
-static unsigned long tainted(unsigned long *taints)
-{
-	FILE *f;
-
-	*taints = 0;
-
-	f = fopen("/proc/sys/kernel/tainted", "r");
-	if (f) {
-		fscanf(f, "%lu", taints);
-		fclose(f);
-	}
-
-	return is_tainted(*taints);
-}
-
 static char *handle_taint(void)
 {
-	unsigned long taints;
+	unsigned long taints, bad;
+	char *explain;
 	char *reason;
 
-	if (!tainted(&taints))
+	bad = igt_kernel_tainted(&taints);
+	if (!bad)
 		return NULL;
 
-	asprintf(&reason, "Kernel badly tainted (%#lx) (check dmesg for details):\n",
-		 taints);
-
-	for (typeof(*abort_taints) *taint = abort_taints; taint->bit; taint++) {
-		if (taint->bit & taints) {
-			char *old_reason = reason;
-			asprintf(&reason, "%s\t(%#lx) %s\n",
-					old_reason,
-					taint->bit,
-					taint->explanation);
-			free(old_reason);
-		}
+	asprintf(&reason, "Kernel badly tainted (%#lx, %#lx) (check dmesg for details):\n",
+		 taints, bad);
+
+	while ((explain = igt_explain_taints(&bad))) {
+		char *old_reason = reason;
+		asprintf(&reason, "%s\t%s\n", old_reason, explain);
+		free(old_reason);
 	}
 
 	return reason;
@@ -1142,7 +1096,8 @@ static int monitor_output(pid_t child,
 			sigfd = -1; /* we are dying, no signal handling for now */
 		}
 
-		timeout_reason = need_to_timeout(settings, killed, tainted(&taints),
+		timeout_reason = need_to_timeout(settings, killed,
+						 igt_kernel_tainted(&taints),
 						 igt_time_elapsed(&time_last_activity, &time_now),
 						 igt_time_elapsed(&time_last_subtest, &time_now),
 						 igt_time_elapsed(&time_killed, &time_now),