[PATCH] ring_buffer: reset write when reserve buffer fail

[PATCH] ring_buffer: reset write when reserve buffer fail

[Lai Jiangshan]

Impact: reset struct buffer_page.write when interrupt storm

if struct buffer_page.write is not reset, any succedent committing
will corrupted ring_buffer:

static inline void
rb_set_commit_to_write(struct ring_buffer_per_cpu *cpu_buffer)
{
	......
		cpu_buffer->commit_page->commit =
			cpu_buffer->commit_page->write;
	......
}


when "if (RB_WARN_ON(cpu_buffer, next_page == reader_page))", ring_buffer
is disabled, but some reserved buffers may haven't been committed.
we need reset struct buffer_page.write.

when "if (unlikely(next_page == cpu_buffer->commit_page))", ring_buffer
is still available, we should not corrupt it.

Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
---
diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 8b0daf0..18bfd76 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -1025,12 +1025,8 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
 		}
 
 		if (next_page == head_page) {
-			if (!(buffer->flags & RB_FL_OVERWRITE)) {
-				/* reset write */
-				if (tail <= BUF_PAGE_SIZE)
-					local_set(&tail_page->write, tail);
+			if (!(buffer->flags & RB_FL_OVERWRITE))
 				goto out_unlock;
-			}
 
 			/* tail_page has not moved yet? */
 			if (tail_page == cpu_buffer->tail_page) {
@@ -1105,6 +1101,10 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
 	return event;
 
  out_unlock:
+	/* reset write */
+	if (tail <= BUF_PAGE_SIZE)
+		local_set(&tail_page->write, tail);
+
 	__raw_spin_unlock(&cpu_buffer->lock);
 	local_irq_restore(flags);
 	return NULL;

Re: [PATCH] ring_buffer: reset write when reserve buffer fail

[Lai Jiangshan]

Hi, Steven Rostedt,

Have you seen these two patches?
(other patch is "[PATCH -tip] ftrace: fix trace_output")

Thanks, Lai

Lai Jiangshan wrote:
> Impact: reset struct buffer_page.write when interrupt storm
> 
> if struct buffer_page.write is not reset, any succedent committing
> will corrupted ring_buffer:
> 
> static inline void
> rb_set_commit_to_write(struct ring_buffer_per_cpu *cpu_buffer)
> {
> 	......
> 		cpu_buffer->commit_page->commit =
> 			cpu_buffer->commit_page->write;
> 	......
> }
> 
> 
> when "if (RB_WARN_ON(cpu_buffer, next_page == reader_page))", ring_buffer
> is disabled, but some reserved buffers may haven't been committed.
> we need reset struct buffer_page.write.
> 
> when "if (unlikely(next_page == cpu_buffer->commit_page))", ring_buffer
> is still available, we should not corrupt it.
> 
> Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
> ---
> diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
> index 8b0daf0..18bfd76 100644
> --- a/kernel/trace/ring_buffer.c
> +++ b/kernel/trace/ring_buffer.c
> @@ -1025,12 +1025,8 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
>  		}
>  
>  		if (next_page == head_page) {
> -			if (!(buffer->flags & RB_FL_OVERWRITE)) {
> -				/* reset write */
> -				if (tail <= BUF_PAGE_SIZE)
> -					local_set(&tail_page->write, tail);
> +			if (!(buffer->flags & RB_FL_OVERWRITE))
>  				goto out_unlock;
> -			}
>  
>  			/* tail_page has not moved yet? */
>  			if (tail_page == cpu_buffer->tail_page) {
> @@ -1105,6 +1101,10 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
>  	return event;
>  
>   out_unlock:
> +	/* reset write */
> +	if (tail <= BUF_PAGE_SIZE)
> +		local_set(&tail_page->write, tail);
> +
>  	__raw_spin_unlock(&cpu_buffer->lock);
>  	local_irq_restore(flags);
>  	return NULL;
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 
> 
> 

Re: [PATCH] ring_buffer: reset write when reserve buffer fail

[Steven Rostedt]

Hi Lai,

Sorry for the late response, I've got a million things going on at once, 
and this needed more thought than a quick review.

On Mon, 12 Jan 2009, Lai Jiangshan wrote:

> 
> Impact: reset struct buffer_page.write when interrupt storm
> 
> if struct buffer_page.write is not reset, any succedent committing
> will corrupted ring_buffer:
> 
> static inline void
> rb_set_commit_to_write(struct ring_buffer_per_cpu *cpu_buffer)
> {
> 	......
> 		cpu_buffer->commit_page->commit =
> 			cpu_buffer->commit_page->write;
> 	......
> }
> 
> 
> when "if (RB_WARN_ON(cpu_buffer, next_page == reader_page))", ring_buffer
> is disabled, but some reserved buffers may haven't been committed.
> we need reset struct buffer_page.write.

So what you are saying is, when we hit this anomaly, another cpu might be 
writing to the ring buffer, and will soon do a commit.

> 
> when "if (unlikely(next_page == cpu_buffer->commit_page))", ring_buffer
> is still available, we should not corrupt it.

OK, makes sense. I'll queue it up.

Thanks!

-- Steve