[PATCH] tracing: Do PTR_ERR() after IS_ERR()

Li Qiong posted 1 patch 3 years, 8 months ago
kernel/trace/ring_buffer_benchmark.c | 10 ++++++----
1 file changed, 6 insertions(+), 4 deletions(-)
[PATCH] tracing: Do PTR_ERR() after IS_ERR()
Posted by Li Qiong 3 years, 8 months ago
Check IS_ERR() firstly, then do PTR_ERR().

Signed-off-by: Li Qiong <liqiong@nfschina.com>
---
 kernel/trace/ring_buffer_benchmark.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/kernel/trace/ring_buffer_benchmark.c b/kernel/trace/ring_buffer_benchmark.c
index 78e576575b79..a8f6b0725c45 100644
--- a/kernel/trace/ring_buffer_benchmark.c
+++ b/kernel/trace/ring_buffer_benchmark.c
@@ -439,17 +439,19 @@ static int __init ring_buffer_benchmark_init(void)
 	if (!disable_reader) {
 		consumer = kthread_create(ring_buffer_consumer_thread,
 					  NULL, "rb_consumer");
-		ret = PTR_ERR(consumer);
-		if (IS_ERR(consumer))
+		if (IS_ERR(consumer)) {
+			ret = PTR_ERR(consumer);
 			goto out_fail;
+		}
 	}
 
 	producer = kthread_run(ring_buffer_producer_thread,
 			       NULL, "rb_producer");
-	ret = PTR_ERR(producer);
 
-	if (IS_ERR(producer))
+	if (IS_ERR(producer)) {
+		ret = PTR_ERR(producer);
 		goto out_kill;
+	}
 
 	/*
 	 * Run them as low-prio background tasks by default:
-- 
2.11.0
Re: [PATCH] tracing: Do PTR_ERR() after IS_ERR()
Posted by Steven Rostedt 3 years, 8 months ago
On Wed, 27 Jul 2022 23:35:19 +0800
Li Qiong <liqiong@nfschina.com> wrote:

> Check IS_ERR() firstly, then do PTR_ERR().

Why?

The code is fine as is.

-- Steve
Re: [PATCH] tracing: Do PTR_ERR() after IS_ERR()
Posted by liqiong 3 years, 8 months ago
在 2022年07月28日 00:28, Steven Rostedt 写道:
> On Wed, 27 Jul 2022 23:35:19 +0800
> Li Qiong <liqiong@nfschina.com> wrote:
>
>> Check IS_ERR() firstly, then do PTR_ERR().
> Why?
>
> The code is fine as is.
>
> -- Steve
>

Hi, 

It's all right, assign  PTR_ERR()  to 'ret'  anyway.
But this assignment is valid only at the "IS_ERR()" path.
Maybe it is better put "PTR_ERR()" at error path, keep the code clear.

-- 
李力琼 <13524287433>
上海市浦东新区海科路99号中科院上海高等研究院3号楼3楼

Re: [PATCH] tracing: Do PTR_ERR() after IS_ERR()
Posted by Steven Rostedt 3 years, 8 months ago
On Thu, 28 Jul 2022 08:28:08 +0800
liqiong <liqiong@nfschina.com> wrote:

> It's all right, assign  PTR_ERR()  to 'ret'  anyway.
> But this assignment is valid only at the "IS_ERR()" path.
> Maybe it is better put "PTR_ERR()" at error path, keep the code clear.

No it does not. It adds unnecessary brackets.

It is common in the kernel to have:

	ret = ERROR;
	if (some_condition())
		goto out;

	ret = ERROR1;
	if (some_other_condition())
		goto out;

	ret = ERROR2;
	if (some_new_condition())
		goto out;

	ret = 0;
out:
	return ret;

And your change breaks this.

-- Steve
Re: [PATCH] tracing: Do PTR_ERR() after IS_ERR()
Posted by 李力琼 3 years, 8 months ago
在 2022/7/30 00:56, Steven Rostedt 写道:
> On Thu, 28 Jul 2022 08:28:08 +0800
> liqiong <liqiong@nfschina.com> wrote:
>
>> It's all right, assign  PTR_ERR()  to 'ret'  anyway.
>> But this assignment is valid only at the "IS_ERR()" path.
>> Maybe it is better put "PTR_ERR()" at error path, keep the code clear.
> No it does not. It adds unnecessary brackets.
>
> It is common in the kernel to have:
>
> 	ret = ERROR;
> 	if (some_condition())
> 		goto out;
>
> 	ret = ERROR1;
> 	if (some_other_condition())
> 		goto out;
>
> 	ret = ERROR2;
> 	if (some_new_condition())
> 		goto out;
>
> 	ret = 0;
> out:
> 	return ret;
>
> And your change breaks this.
>
> -- Steve

I got it, Thanks.