2009-03-22 23:15:39

by Frederic Weisbecker

Subject: [PATCH] tracing/events: don't discard an event after commit

Note: depends on "tracing/events: don't use wake up for events"

When we want to filter an event, the filter test is done after
the event is commited to the ring-buffer to be discarded later if
needed.

But a reader could be reading this event while we are trying to discard
it. Other kind of racy events can even happen because the event is
commited and can be read and/or consumed.

What we want is to discard the event before committing it.

Signed-off-by: Frederic Weisbecker <[email protected]>
---
kernel/trace/trace_events_stage_3.h | 5 +++--
1 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/kernel/trace/trace_events_stage_3.h b/kernel/trace/trace_events_stage_3.h
index 9a3bd49..9d2fa78 100644
--- a/kernel/trace/trace_events_stage_3.h
+++ b/kernel/trace/trace_events_stage_3.h
@@ -222,10 +222,11 @@ static void ftrace_raw_event_##call(proto) \
\
assign; \
\
- trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
- \
if (call->preds && !filter_match_preds(call, entry)) \
ring_buffer_event_discard(event); \
+ \
+ trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
+ \
} \
\
static int ftrace_raw_reg_event_##call(void) \
--
1.6.1

2009-03-22 23:18:54

by Frederic Weisbecker

Subject: [PATCH v2] tracing/events: don't discard an event after commit

Note: depends on "tracing/events: don't use wake up for events"

When we want to filter an event, the filter test is done after
the event is commited to the ring-buffer to be discarded later if
needed.

But a reader could be reading this event while we are trying to discard
it. Other kind of racy events can even happen because the event is
commited and can be read and/or consumed.

What we want is to discard the event before committing it.

v2: add a reported-by Steven Rostedt

Reported-by: Steven Rostedt <[email protected]>
Signed-off-by: Frederic Weisbecker <[email protected]>
---
kernel/trace/trace_events_stage_3.h | 5 +++--
1 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/kernel/trace/trace_events_stage_3.h b/kernel/trace/trace_events_stage_3.h
index 9a3bd49..9d2fa78 100644
--- a/kernel/trace/trace_events_stage_3.h
+++ b/kernel/trace/trace_events_stage_3.h
@@ -222,10 +222,11 @@ static void ftrace_raw_event_##call(proto) \
\
assign; \
\
- trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
- \
if (call->preds && !filter_match_preds(call, entry)) \
ring_buffer_event_discard(event); \
+ \
+ trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
+ \
} \
\
static int ftrace_raw_reg_event_##call(void) \
--
1.6.1

2009-03-23 08:32:39

by Frederic Weisbecker

Subject: [tip:tracing/filters] tracing/events: don't discard an event after commit

Commit-ID: b118415bfad6d75792a85ac999e25149db8e6919
Gitweb: http://git.kernel.org/tip/b118415bfad6d75792a85ac999e25149db8e6919
Author: Frederic Weisbecker <[email protected]>
AuthorDate: Mon, 23 Mar 2009 00:18:39 +0100
Committer: Ingo Molnar <[email protected]>
CommitDate: Mon, 23 Mar 2009 09:22:15 +0100

tracing/events: don't discard an event after commit

When we want to filter an event, the filter test is done after
the event is commited to the ring-buffer to be discarded later if
needed.

But a reader could be reading this event while we are trying to discard
it. Other kind of racy events can even happen because the event is
commited and can be read and/or consumed.

What we want is to discard the event before committing it.

Reported-by: Steven Rostedt <[email protected]>
Signed-off-by: Frederic Weisbecker <[email protected]>
Cc: Tom Zanussi <[email protected]>
LKML-Reference: <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>


---
kernel/trace/trace_events_stage_3.h | 5 +++--
1 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/kernel/trace/trace_events_stage_3.h b/kernel/trace/trace_events_stage_3.h
index 9a3bd49..9d2fa78 100644
--- a/kernel/trace/trace_events_stage_3.h
+++ b/kernel/trace/trace_events_stage_3.h
@@ -222,10 +222,11 @@ static void ftrace_raw_event_##call(proto) \
\
assign; \
\
- trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
- \
if (call->preds && !filter_match_preds(call, entry)) \
ring_buffer_event_discard(event); \
+ \
+ trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
+ \
} \
\
static int ftrace_raw_reg_event_##call(void) \

2009-03-23 08:34:01

by Tom Zanussi

Subject: Re: [PATCH v2] tracing/events: don't discard an event after commit

Hi Frederic,

On Mon, 2009-03-23 at 00:18 +0100, Frederic Weisbecker wrote:
> Note: depends on "tracing/events: don't use wake up for events"
>
> When we want to filter an event, the filter test is done after
> the event is commited to the ring-buffer to be discarded later if
> needed.
>
> But a reader could be reading this event while we are trying to discard
> it. Other kind of racy events can even happen because the event is
> commited and can be read and/or consumed.
>
> What we want is to discard the event before committing it.
>

Thanks for fixing this - I had intend on doing this, but it completely
slipped my mind.

Thanks for all your other comments and fixes too...

Tom

> v2: add a reported-by Steven Rostedt
>
> Reported-by: Steven Rostedt <[email protected]>
> Signed-off-by: Frederic Weisbecker <[email protected]>
> ---
> kernel/trace/trace_events_stage_3.h | 5 +++--
> 1 files changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/trace/trace_events_stage_3.h b/kernel/trace/trace_events_stage_3.h
> index 9a3bd49..9d2fa78 100644
> --- a/kernel/trace/trace_events_stage_3.h
> +++ b/kernel/trace/trace_events_stage_3.h
> @@ -222,10 +222,11 @@ static void ftrace_raw_event_##call(proto) \
> \
> assign; \
> \
> - trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
> - \
> if (call->preds && !filter_match_preds(call, entry)) \
> ring_buffer_event_discard(event); \
> + \
> + trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
> + \
> } \
> \
> static int ftrace_raw_reg_event_##call(void) \