[PATCH 10/10] trace task_io

[Wu Fengguang]

The "write_bytes" curve over time is very handy for viewing the
smoothness and fairness of page dirties for each task.

XXX: It looks not a good fit for writeback traces, shall we create a
include/trace/events/vfs.h for this?

XXX: the other fields of struct task_io_accounting are not dumped.

Signed-off-by: Wu Fengguang <fengguang.wu@xxxxxxxxx>
---
 include/trace/events/writeback.h |   32 +++++++++++++++++++++++++++++
 mm/page-writeback.c              |    1 
 2 files changed, 33 insertions(+)

--- linux-next.orig/include/trace/events/writeback.h	2011-08-22 11:58:21.000000000 +0800
+++ linux-next/include/trace/events/writeback.h	2011-08-22 12:22:38.000000000 +0800
@@ -180,6 +180,38 @@ TRACE_EVENT(writeback_queue_io,
 		__entry->moved)
 );
 
+TRACE_EVENT(task_io,
+	TP_PROTO(struct task_struct *task),
+	TP_ARGS(task),
+
+	TP_STRUCT__entry(
+		__field(unsigned long long,	read_bytes)
+		__field(unsigned long long,	write_bytes)
+		__field(unsigned long long,	cancelled_write_bytes)
+	),
+
+	TP_fast_assign(
+		struct task_io_accounting *ioac = &task->ioac;
+
+#ifdef CONFIG_TASK_IO_ACCOUNTING
+		__entry->read_bytes		= ioac->read_bytes;
+		__entry->write_bytes		= ioac->write_bytes;
+		__entry->cancelled_write_bytes	= ioac->cancelled_write_bytes;
+#else
+		__entry->read_bytes		= 0;
+		__entry->write_bytes		= 0;
+		__entry->cancelled_write_bytes	= 0;
+#endif
+	),
+
+	TP_printk("read=%llu write=%llu cancelled_write=%llu",
+		  __entry->read_bytes,
+		  __entry->write_bytes,
+		  __entry->cancelled_write_bytes
+	)
+);
+
+
 TRACE_EVENT(global_dirty_state,
 
 	TP_PROTO(unsigned long background_thresh,
--- linux-next.orig/mm/page-writeback.c	2011-08-22 12:10:33.000000000 +0800
+++ linux-next/mm/page-writeback.c	2011-08-22 12:20:33.000000000 +0800
@@ -407,6 +407,7 @@ void global_dirty_limits(unsigned long *
 	*pbackground = background;
 	*pdirty = dirty;
 	trace_global_dirty_state(background, dirty);
+	trace_task_io(current);
 }
 
 /**


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/