[PATCH 0/2] ring-buffer updates for tip

[PATCH 0/2] ring-buffer updates for tip

[Steven Rostedt]

Ingo,

The following patches are in:

  git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git

    branch: tip/devel


Steven Rostedt (2):
      ring-buffer: fix deadlock from reader_lock in read_start
      ring-buffer: no preempt for sched_clock

----
 kernel/trace/ring_buffer.c |   37 ++++++++++++++++++++++++-------------
 1 files changed, 24 insertions(+), 13 deletions(-)

[PATCH 1/2] ring-buffer: fix deadlock from reader_lock in read_start

[Steven Rostedt]

[-- Attachment #1: 0001-ring-buffer-fix-deadlock-from-reader_lock-in-read_s.patch --]
[-- Type: text/plain, Size: 2276 bytes --]

Impact: deadlock fix in ring_buffer_read_start

The ring_buffer_iter_reset was called from ring_buffer_read_start
where both grabbed the reader_lock.

This patch separates out the internals of ring_buffer_iter_reset
to its own function so that both APIs may grab the reader_lock.

Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
 kernel/trace/ring_buffer.c |   29 +++++++++++++++++------------
 1 files changed, 17 insertions(+), 12 deletions(-)

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index c04c433..86dc353 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -1475,19 +1475,9 @@ unsigned long ring_buffer_overruns(struct ring_buffer *buffer)
 	return overruns;
 }
 
-/**
- * ring_buffer_iter_reset - reset an iterator
- * @iter: The iterator to reset
- *
- * Resets the iterator, so that it will start from the beginning
- * again.
- */
-void ring_buffer_iter_reset(struct ring_buffer_iter *iter)
+static void rb_iter_reset(struct ring_buffer_iter *iter)
 {
 	struct ring_buffer_per_cpu *cpu_buffer = iter->cpu_buffer;
-	unsigned long flags;
-
-	spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
 
 	/* Iterator usage is expected to have record disabled */
 	if (list_empty(&cpu_buffer->reader_page->list)) {
@@ -1501,7 +1491,22 @@ void ring_buffer_iter_reset(struct ring_buffer_iter *iter)
 		iter->read_stamp = cpu_buffer->read_stamp;
 	else
 		iter->read_stamp = iter->head_page->time_stamp;
+}
 
+/**
+ * ring_buffer_iter_reset - reset an iterator
+ * @iter: The iterator to reset
+ *
+ * Resets the iterator, so that it will start from the beginning
+ * again.
+ */
+void ring_buffer_iter_reset(struct ring_buffer_iter *iter)
+{
+	struct ring_buffer_per_cpu *cpu_buffer = iter->cpu_buffer;
+	unsigned long flags;
+
+	spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
+	rb_iter_reset(iter);
 	spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
 }
 
@@ -1957,7 +1962,7 @@ ring_buffer_read_start(struct ring_buffer *buffer, int cpu)
 
 	spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
 	__raw_spin_lock(&cpu_buffer->lock);
-	ring_buffer_iter_reset(iter);
+	rb_iter_reset(iter);
 	__raw_spin_unlock(&cpu_buffer->lock);
 	spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
 
-- 
1.5.6.5

-- 

[PATCH 2/2] ring-buffer: no preempt for sched_clock

[Steven Rostedt]

[-- Attachment #1: 0002-ring-buffer-no-preempt-for-sched_clock.patch --]
[-- Type: text/plain, Size: 1049 bytes --]

Impact: disable preemption no calling sched_clock

The ring_buffer_time_stamp still uses sched_clock as its counter.
But it is a bug to call it with preemption enabled. This requirement
should not be pushed to the ring_buffer_time_stamp callers, so
the ring_buffer_time_stamp needs to disable preemption when calling
sched_clock.

Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
 kernel/trace/ring_buffer.c |    8 +++++++-
 1 files changed, 7 insertions(+), 1 deletions(-)

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 86dc353..2d6c2cf 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -53,8 +53,14 @@ void tracing_off(void)
 /* FIXME!!! */
 u64 ring_buffer_time_stamp(int cpu)
 {
+	u64 time;
+
+	preempt_disable_notrace();
 	/* shift to debug/test normalization and TIME_EXTENTS */
-	return sched_clock() << DEBUG_SHIFT;
+	time = sched_clock() << DEBUG_SHIFT;
+	preempt_enable_notrace();
+
+	return time;
 }
 
 void ring_buffer_normalize_time_stamp(int cpu, u64 *ts)
-- 
1.5.6.5

-- 

Re: [PATCH 2/2] ring-buffer: no preempt for sched_clock

[Ingo Molnar]

* Steven Rostedt <rostedt@goodmis.org> wrote:

> Impact: disable preemption no calling sched_clock
> 
> The ring_buffer_time_stamp still uses sched_clock as its counter.
> But it is a bug to call it with preemption enabled. This requirement
> should not be pushed to the ring_buffer_time_stamp callers, so
> the ring_buffer_time_stamp needs to disable preemption when calling
> sched_clock.
> 
> Signed-off-by: Steven Rostedt <srostedt@redhat.com>
> ---
>  kernel/trace/ring_buffer.c |    8 +++++++-
>  1 files changed, 7 insertions(+), 1 deletions(-)

applied to tip/tracing/urgent, thanks Steve!

	Ingo

Re: [PATCH 1/2] ring-buffer: fix deadlock from reader_lock in read_start

[Ingo Molnar]

* Steven Rostedt <rostedt@goodmis.org> wrote:

> Impact: deadlock fix in ring_buffer_read_start
> 
> The ring_buffer_iter_reset was called from ring_buffer_read_start
> where both grabbed the reader_lock.
> 
> This patch separates out the internals of ring_buffer_iter_reset
> to its own function so that both APIs may grab the reader_lock.
> 
> Signed-off-by: Steven Rostedt <srostedt@redhat.com>
> ---
>  kernel/trace/ring_buffer.c |   29 +++++++++++++++++------------
>  1 files changed, 17 insertions(+), 12 deletions(-)

applied to tip/tracing/ring-buffer, thanks Steve!

	Ingo