tracing: fix comments about trace buffer resizing
authorSteven Rostedt <srostedt@redhat.com>
Thu, 12 Mar 2009 15:21:08 +0000 (11:21 -0400)
committerSteven Rostedt <srostedt@redhat.com>
Fri, 13 Mar 2009 01:14:58 +0000 (21:14 -0400)
Impact: cleanup

Some of the comments about the trace buffer resizing is gobbledygook.
And I wonder why people question if I'm a native English speaker.

This patch makes the comments make a bit more sense.

Reported-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
kernel/trace/trace.c

index c3946a6df34e8be4081e04ee58130cfc2ba45e6f..c61ee85c50bbea3fac9bd63f240f3faeb0550b9e 100644 (file)
@@ -2336,7 +2336,8 @@ static int tracing_resize_ring_buffer(unsigned long size)
 
        /*
         * If kernel or user changes the size of the ring buffer
-        * it get completed.
+        * we use the size that was given, and we can forget about
+        * expanding it later.
         */
        ring_buffer_expanded = 1;
 
@@ -2351,8 +2352,20 @@ static int tracing_resize_ring_buffer(unsigned long size)
                r = ring_buffer_resize(global_trace.buffer,
                                       global_trace.entries);
                if (r < 0) {
-                       /* AARGH! We are left with different
-                        * size max buffer!!!! */
+                       /*
+                        * AARGH! We are left with different
+                        * size max buffer!!!!
+                        * The max buffer is our "snapshot" buffer.
+                        * When a tracer needs a snapshot (one of the
+                        * latency tracers), it swaps the max buffer
+                        * with the saved snap shot. We succeeded to
+                        * update the size of the main buffer, but failed to
+                        * update the size of the max buffer. But when we tried
+                        * to reset the main buffer to the original size, we
+                        * failed there too. This is very unlikely to
+                        * happen, but if it does, warn and kill all
+                        * tracing.
+                        */
                        WARN_ON(1);
                        tracing_disabled = 1;
                }