tracing: Support to dump instance traces by ftrace_dump_on_oops

Currently ftrace only dumps the global trace buffer on an OOPs. For
debugging a production usecase, instance trace will be helpful to
check specific problems since global trace buffer may be used for
other purposes.

This patch extend the ftrace_dump_on_oops parameter to dump a specific
or multiple trace instances:

  - ftrace_dump_on_oops=0: as before -- don't dump
  - ftrace_dump_on_oops[=1]: as before -- dump the global trace buffer
  on all CPUs
  - ftrace_dump_on_oops=2 or =orig_cpu: as before -- dump the global
  trace buffer on CPU that triggered the oops
  - ftrace_dump_on_oops=<instance_name>: new behavior -- dump the
  tracing instance matching <instance_name>
  - ftrace_dump_on_oops[=2/orig_cpu],<instance1_name>[=2/orig_cpu],
  <instrance2_name>[=2/orig_cpu]: new behavior -- dump the global trace
  buffer and multiple instance buffer on all CPUs, or only dump on CPU
  that triggered the oops if =2 or =orig_cpu is given

Also, the sysctl node can handle the input accordingly.

Link: https://lore.kernel.org/linux-trace-kernel/20240223083126.1817731-1-quic_hyiwei@quicinc.com

Cc: Ross Zwisler <zwisler@google.com>
Cc: <mhiramat@kernel.org>
Cc: <mark.rutland@arm.com>
Cc: <mcgrof@kernel.org>
Cc: <keescook@chromium.org>
Cc: <j.granados@samsung.com>
Cc: <mathieu.desnoyers@efficios.com>
Cc: <corbet@lwn.net>
Signed-off-by: Huang Yiwei <quic_hyiwei@quicinc.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>

3 files changed, 119 insertions, 43 deletions

diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 157f7ce2942d..81cc974913bb 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -1710,9 +1710,9 @@ static struct ctl_table kern_table[] = {
 	{
 		.procname	= "ftrace_dump_on_oops",
 		.data		= &ftrace_dump_on_oops,
-		.maxlen		= sizeof(int),
+		.maxlen		= MAX_TRACER_SIZE,
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dostring,
 	},
 	{
 		.procname	= "traceoff_on_warning",
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 1e1fd377a1cf..233d1af39fff 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -130,9 +130,12 @@ cpumask_var_t __read_mostly	tracing_buffer_mask;
  * /proc/sys/kernel/ftrace_dump_on_oops
  * Set 1 if you want to dump buffers of all CPUs
  * Set 2 if you want to dump the buffer of the CPU that triggered oops
+ * Set instance name if you want to dump the specific trace instance
+ * Multiple instance dump is also supported, and instances are seperated
+ * by commas.
  */
-
-enum ftrace_dump_mode ftrace_dump_on_oops;
+/* Set to string format zero to disable by default */
+char ftrace_dump_on_oops[MAX_TRACER_SIZE] = "0";
 
 /* When set, tracing will stop when a WARN*() is hit */
 int __disable_trace_on_warning;
@@ -178,7 +181,6 @@ static void ftrace_trace_userstack(struct trace_array *tr,
 				   struct trace_buffer *buffer,
 				   unsigned int trace_ctx);
 
-#define MAX_TRACER_SIZE		100
 static char bootup_tracer_buf[MAX_TRACER_SIZE] __initdata;
 static char *default_bootup_tracer;
 
@@ -201,19 +203,33 @@ static int __init set_cmdline_ftrace(char *str)
 }
 __setup("ftrace=", set_cmdline_ftrace);
 
+int ftrace_dump_on_oops_enabled(void)
+{
+	if (!strcmp("0", ftrace_dump_on_oops))
+		return 0;
+	else
+		return 1;
+}
+
 static int __init set_ftrace_dump_on_oops(char *str)
 {
-	if (*str++ != '=' || !*str || !strcmp("1", str)) {
-		ftrace_dump_on_oops = DUMP_ALL;
+	if (!*str) {
+		strscpy(ftrace_dump_on_oops, "1", MAX_TRACER_SIZE);
 		return 1;
 	}
 
-	if (!strcmp("orig_cpu", str) || !strcmp("2", str)) {
-		ftrace_dump_on_oops = DUMP_ORIG;
-                return 1;
-        }
+	if (*str == ',') {
+		strscpy(ftrace_dump_on_oops, "1", MAX_TRACER_SIZE);
+		strscpy(ftrace_dump_on_oops + 1, str, MAX_TRACER_SIZE - 1);
+		return 1;
+	}
+
+	if (*str++ == '=') {
+		strscpy(ftrace_dump_on_oops, str, MAX_TRACER_SIZE);
+		return 1;
+	}
 
-        return 0;
+	return 0;
 }
 __setup("ftrace_dump_on_oops", set_ftrace_dump_on_oops);
 
@@ -9832,14 +9848,14 @@ static struct notifier_block trace_die_notifier = {
 static int trace_die_panic_handler(struct notifier_block *self,
 				unsigned long ev, void *unused)
 {
-	if (!ftrace_dump_on_oops)
+	if (!ftrace_dump_on_oops_enabled())
 		return NOTIFY_DONE;
 
 	/* The die notifier requires DIE_OOPS to trigger */
 	if (self == &trace_die_notifier && ev != DIE_OOPS)
 		return NOTIFY_DONE;
 
-	ftrace_dump(ftrace_dump_on_oops);
+	ftrace_dump(DUMP_PARAM);
 
 	return NOTIFY_DONE;
 }
@@ -9880,12 +9896,12 @@ trace_printk_seq(struct trace_seq *s)
 	trace_seq_init(s);
 }
 
-void trace_init_global_iter(struct trace_iterator *iter)
+static void trace_init_iter(struct trace_iterator *iter, struct trace_array *tr)
 {
-	iter->tr = &global_trace;
+	iter->tr = tr;
 	iter->trace = iter->tr->current_trace;
 	iter->cpu_file = RING_BUFFER_ALL_CPUS;
-	iter->array_buffer = &global_trace.array_buffer;
+	iter->array_buffer = &tr->array_buffer;
 
 	if (iter->trace && iter->trace->open)
 		iter->trace->open(iter);
@@ -9905,22 +9921,19 @@ void trace_init_global_iter(struct trace_iterator *iter)
 	iter->fmt_size = STATIC_FMT_BUF_SIZE;
 }
 
-void ftrace_dump(enum ftrace_dump_mode oops_dump_mode)
+void trace_init_global_iter(struct trace_iterator *iter)
+{
+	trace_init_iter(iter, &global_trace);
+}
+
+static void ftrace_dump_one(struct trace_array *tr, enum ftrace_dump_mode dump_mode)
 {
 	/* use static because iter can be a bit big for the stack */
 	static struct trace_iterator iter;
-	static atomic_t dump_running;
-	struct trace_array *tr = &global_trace;
 	unsigned int old_userobj;
 	unsigned long flags;
 	int cnt = 0, cpu;
 
-	/* Only allow one dump user at a time. */
-	if (atomic_inc_return(&dump_running) != 1) {
-		atomic_dec(&dump_running);
-		return;
-	}
-
 	/*
 	 * Always turn off tracing when we dump.
 	 * We don't need to show trace output of what happens
@@ -9929,12 +9942,12 @@ void ftrace_dump(enum ftrace_dump_mode oops_dump_mode)
 	 * If the user does a sysrq-z, then they can re-enable
 	 * tracing with echo 1 > tracing_on.
 	 */
-	tracing_off();
+	tracer_tracing_off(tr);
 
 	local_irq_save(flags);
 
 	/* Simulate the iterator */
-	trace_init_global_iter(&iter);
+	trace_init_iter(&iter, tr);
 
 	for_each_tracing_cpu(cpu) {
 		atomic_inc(&per_cpu_ptr(iter.array_buffer->data, cpu)->disabled);
@@ -9945,21 +9958,15 @@ void ftrace_dump(enum ftrace_dump_mode oops_dump_mode)
 	/* don't look at user memory in panic mode */
 	tr->trace_flags &= ~TRACE_ITER_SYM_USEROBJ;
 
-	switch (oops_dump_mode) {
-	case DUMP_ALL:
-		iter.cpu_file = RING_BUFFER_ALL_CPUS;
-		break;
-	case DUMP_ORIG:
+	if (dump_mode == DUMP_ORIG)
 		iter.cpu_file = raw_smp_processor_id();
-		break;
-	case DUMP_NONE:
-		goto out_enable;
-	default:
-		printk(KERN_TRACE "Bad dumping mode, switching to all CPUs dump\n");
+	else
 		iter.cpu_file = RING_BUFFER_ALL_CPUS;
-	}
 
-	printk(KERN_TRACE "Dumping ftrace buffer:\n");
+	if (tr == &global_trace)
+		printk(KERN_TRACE "Dumping ftrace buffer:\n");
+	else
+		printk(KERN_TRACE "Dumping ftrace instance %s buffer:\n", tr->name);
 
 	/* Did function tracer already get disabled? */
 	if (ftrace_is_dead()) {
@@ -10001,15 +10008,84 @@ void ftrace_dump(enum ftrace_dump_mode oops_dump_mode)
 	else
 		printk(KERN_TRACE "---------------------------------\n");
 
- out_enable:
 	tr->trace_flags |= old_userobj;
 
 	for_each_tracing_cpu(cpu) {
 		atomic_dec(&per_cpu_ptr(iter.array_buffer->data, cpu)->disabled);
 	}
-	atomic_dec(&dump_running);
 	local_irq_restore(flags);
 }
+
+static void ftrace_dump_by_param(void)
+{
+	bool first_param = true;
+	char dump_param[MAX_TRACER_SIZE];
+	char *buf, *token, *inst_name;
+	struct trace_array *tr;
+
+	strscpy(dump_param, ftrace_dump_on_oops, MAX_TRACER_SIZE);
+	buf = dump_param;
+
+	while ((token = strsep(&buf, ",")) != NULL) {
+		if (first_param) {
+			first_param = false;
+			if (!strcmp("0", token))
+				continue;
+			else if (!strcmp("1", token)) {
+				ftrace_dump_one(&global_trace, DUMP_ALL);
+				continue;
+			}
+			else if (!strcmp("2", token) ||
+			  !strcmp("orig_cpu", token)) {
+				ftrace_dump_one(&global_trace, DUMP_ORIG);
+				continue;
+			}
+		}
+
+		inst_name = strsep(&token, "=");
+		tr = trace_array_find(inst_name);
+		if (!tr) {
+			printk(KERN_TRACE "Instance %s not found\n", inst_name);
+			continue;
+		}
+
+		if (token && (!strcmp("2", token) ||
+			  !strcmp("orig_cpu", token)))
+			ftrace_dump_one(tr, DUMP_ORIG);
+		else
+			ftrace_dump_one(tr, DUMP_ALL);
+	}
+}
+
+void ftrace_dump(enum ftrace_dump_mode oops_dump_mode)
+{
+	static atomic_t dump_running;
+
+	/* Only allow one dump user at a time. */
+	if (atomic_inc_return(&dump_running) != 1) {
+		atomic_dec(&dump_running);
+		return;
+	}
+
+	switch (oops_dump_mode) {
+	case DUMP_ALL:
+		ftrace_dump_one(&global_trace, DUMP_ALL);
+		break;
+	case DUMP_ORIG:
+		ftrace_dump_one(&global_trace, DUMP_ORIG);
+		break;
+	case DUMP_PARAM:
+		ftrace_dump_by_param();
+		break;
+	case DUMP_NONE:
+		break;
+	default:
+		printk(KERN_TRACE "Bad dumping mode, switching to all CPUs dump\n");
+		ftrace_dump_one(&global_trace, DUMP_ALL);
+	}
+
+	atomic_dec(&dump_running);
+}
 EXPORT_SYMBOL_GPL(ftrace_dump);
 
 #define WRITE_BUFSIZE  4096
diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c
index 529590499b1f..e9c5058a8efd 100644
--- a/kernel/trace/trace_selftest.c
+++ b/kernel/trace/trace_selftest.c
@@ -768,7 +768,7 @@ static int trace_graph_entry_watchdog(struct ftrace_graph_ent *trace)
 	if (unlikely(++graph_hang_thresh > GRAPH_MAX_FUNC_TEST)) {
 		ftrace_graph_stop();
 		printk(KERN_WARNING "BUG: Function graph tracer hang!\n");
-		if (ftrace_dump_on_oops) {
+		if (ftrace_dump_on_oops_enabled()) {
 			ftrace_dump(DUMP_ALL);
 			/* ftrace_dump() disables tracing */
 			tracing_on();