tracing: Reset ring buffer when changing trace_clocks
authorDavid Sharp <dhsharp@google.com>
Thu, 11 Oct 2012 23:27:52 +0000 (16:27 -0700)
committerSteven Rostedt <rostedt@goodmis.org>
Fri, 2 Nov 2012 14:21:47 +0000 (10:21 -0400)
Because the "tsc" clock isn't in nanoseconds, the ring buffer must be
reset when changing clocks so that incomparable timestamps don't end up
in the same trace.

Tested: Confirmed switching clocks resets the trace buffer.

Google-Bug-Id: 6980623
Link: http://lkml.kernel.org/r/1349998076-15495-3-git-send-email-dhsharp@google.com
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Signed-off-by: David Sharp <dhsharp@google.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
kernel/trace/trace.c

index 88111b08b2c1d6885c703d3322b274ba425b4db3..6ed6013dff2b45052068f44c92496dcc62ec32a9 100644 (file)
@@ -4073,6 +4073,14 @@ static ssize_t tracing_clock_write(struct file *filp, const char __user *ubuf,
        if (max_tr.buffer)
                ring_buffer_set_clock(max_tr.buffer, trace_clocks[i].func);
 
+       /*
+        * New clock may not be consistent with the previous clock.
+        * Reset the buffer so that it doesn't have incomparable timestamps.
+        */
+       tracing_reset_online_cpus(&global_trace);
+       if (max_tr.buffer)
+               tracing_reset_online_cpus(&max_tr);
+
        mutex_unlock(&trace_types_lock);
 
        *fpos += cnt;