From: Masami Hiramatsu (Google) <mhiramat@kernel.org>

Record trace_clock information in the trace_scratch area and recover
the trace_clock when boot, so that reader can docode the timestamp
correctly.
Note that since most trace_clocks records the timestamp in nano-
seconds, this is not a bug. But some trace_clock, like counter and
tsc will record the counter value. Only for those trace_clock user
needs this information.

Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
---
 Changes in v2:
   - instead of exposing it via last_boot_info, set the current
     trace_clock as the same clock we used in the last boot.
---
 include/linux/trace_clock.h |    2 ++
 kernel/trace/trace.c        |   17 +++++++++++++++++
 2 files changed, 19 insertions(+)

diff --git a/include/linux/trace_clock.h b/include/linux/trace_clock.h
index 00e8f98c940f..2245848270d2 100644
--- a/include/linux/trace_clock.h
+++ b/include/linux/trace_clock.h
@@ -15,6 +15,8 @@
 
 #include <asm/trace_clock.h>
 
+#define TRACE_CLOCK_NAME_MAX 16
+
 extern u64 notrace trace_clock_local(void);
 extern u64 notrace trace_clock(void);
 extern u64 notrace trace_clock_jiffies(void);
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 8ddf6b17215c..6ce5d05ec64d 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -6004,6 +6004,7 @@ struct trace_mod_entry {
 };
 
 struct trace_scratch {
+	char			clock[TRACE_CLOCK_NAME_MAX];
 	unsigned long		text_addr;
 	unsigned long		nr_entries;
 	struct trace_mod_entry	entries[];
@@ -6114,6 +6115,8 @@ static void update_last_data(struct trace_array *tr)
 	if (tr->scratch) {
 		struct trace_scratch *tscratch = tr->scratch;
 
+		strscpy(tscratch->clock, trace_clocks[tr->clock_id].name,
+			TRACE_CLOCK_NAME_MAX);
 		memset(tscratch->entries, 0,
 		       flex_array_size(tscratch, entries, tscratch->nr_entries));
 		tscratch->nr_entries = 0;
@@ -7289,6 +7292,12 @@ int tracing_set_clock(struct trace_array *tr, const char *clockstr)
 	tracing_reset_online_cpus(&tr->max_buffer);
 #endif
 
+	if (tr->scratch && !(tr->flags & TRACE_ARRAY_FL_LAST_BOOT)) {
+		struct trace_scratch *tscratch = tr->scratch;
+
+		strscpy(tscratch->clock, trace_clocks[i].name, TRACE_CLOCK_NAME_MAX);
+	}
+
 	mutex_unlock(&trace_types_lock);
 
 	return 0;
@@ -9514,6 +9523,14 @@ static void setup_trace_scratch(struct trace_array *tr,
 
 	/* Scan modules to make text delta for modules. */
 	module_for_each_mod(make_mod_delta, tr);
+
+	/* Set trace_clock as the same of the previous boot. */
+	if (tscratch->clock[0]) {
+		if (tracing_set_clock(tr, tscratch->clock) < 0) {
+			pr_info("the previous trace_clock info is not valid.");
+			goto reset;
+		}
+	}
 	return;
  reset:
 	/* Invalid trace modules */

On Thu, 17 Apr 2025 14:17:29 +0900
"Masami Hiramatsu (Google)" <mhiramat@kernel.org> wrote:

> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -6004,6 +6004,7 @@ struct trace_mod_entry {
>  };
>  
>  struct trace_scratch {
> +	char			clock[TRACE_CLOCK_NAME_MAX];
>  	unsigned long		text_addr;
>  	unsigned long		nr_entries;
>  	struct trace_mod_entry	entries[];
> @@ -6114,6 +6115,8 @@ static void update_last_data(struct trace_array *tr)
>  	if (tr->scratch) {
>  		struct trace_scratch *tscratch = tr->scratch;
>  
> +		strscpy(tscratch->clock, trace_clocks[tr->clock_id].name,

Why copy the name and not the clock_id?

The clock ids should not change between kernels.

-- Steve

> +			TRACE_CLOCK_NAME_MAX);
>  		memset(tscratch->entries, 0,
>  		       flex_array_size(tscratch, entries, tscratch->nr_entries));
>  		tscratch->nr_entries = 0;

On Thu, 17 Apr 2025 10:18:04 -0400
Steven Rostedt <rostedt@goodmis.org> wrote:

> On Thu, 17 Apr 2025 14:17:29 +0900
> "Masami Hiramatsu (Google)" <mhiramat@kernel.org> wrote:
> 
> > --- a/kernel/trace/trace.c
> > +++ b/kernel/trace/trace.c
> > @@ -6004,6 +6004,7 @@ struct trace_mod_entry {
> >  };
> >  
> >  struct trace_scratch {
> > +	char			clock[TRACE_CLOCK_NAME_MAX];
> >  	unsigned long		text_addr;
> >  	unsigned long		nr_entries;
> >  	struct trace_mod_entry	entries[];
> > @@ -6114,6 +6115,8 @@ static void update_last_data(struct trace_array *tr)
> >  	if (tr->scratch) {
> >  		struct trace_scratch *tscratch = tr->scratch;
> >  
> > +		strscpy(tscratch->clock, trace_clocks[tr->clock_id].name,
> 
> Why copy the name and not the clock_id?

The name is more robust if we rebooted in the different kernel.

> 
> The clock ids should not change between kernels.

Hmm, I thought it could be changed because we didn't define it.

----
static struct {
	u64 (*func)(void);
	const char *name;
	int in_ns;		/* is this clock in nanoseconds? */
} trace_clocks[] = {
	{ trace_clock_local,		"local",	1 },
	{ trace_clock_global,		"global",	1 },
	{ trace_clock_counter,		"counter",	0 },
	{ trace_clock_jiffies,		"uptime",	0 },
	{ trace_clock,			"perf",		1 },
	{ ktime_get_mono_fast_ns,	"mono",		1 },
	{ ktime_get_raw_fast_ns,	"mono_raw",	1 },
	{ ktime_get_boot_fast_ns,	"boot",		1 },
	{ ktime_get_tai_fast_ns,	"tai",		1 },
	ARCH_TRACE_CLOCKS
};
----

So the clock id is just an index of this array. That can be changed
easily between the kernel version. If we discard the previous boot
data in that case, I agree with using clock ids.

(Another reason for the implementation is that tracing_set_clock()
 only accepts a name, but this is just due to laziness on my part. :-( )

Thank you,

> 
> -- Steve
> 
> > +			TRACE_CLOCK_NAME_MAX);
> >  		memset(tscratch->entries, 0,
> >  		       flex_array_size(tscratch, entries, tscratch->nr_entries));
> >  		tscratch->nr_entries = 0;


-- 
Masami Hiramatsu (Google) <mhiramat@kernel.org>