[PATCH] Trace: initialize variable to clear warning

[PATCH] Trace: initialize variable to clear warning

[Zhengwang Ruan]

GCC warns that 'page2' is used without being initialized previously before
being used, this is to clear it.

Signed-off-by: Zhengwang Ruan <ruan.zhengwang@gmail.com>
---
 kernel/trace/trace.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index a3f1bc5..47376ba 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -3694,7 +3694,7 @@ tracing_mark_write(struct file *filp, const char __user *ubuf,
 	int nr_pages = 1;
 	ssize_t written;
 	void *page1;
-	void *page2;
+	void *uninitialized_var(page2);
 	int offset;
 	int size;
 	int len;
-- 
1.6.0.4

Re: [PATCH] Trace: initialize variable to clear warning

[Borislav Petkov]

On Tue, Feb 28, 2012 at 05:09:34PM +0800, Zhengwang Ruan wrote:
> GCC warns that 'page2' is used without being initialized previously before
> being used, this is to clear it.
> 
> Signed-off-by: Zhengwang Ruan <ruan.zhengwang@gmail.com>
> ---
>  kernel/trace/trace.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> index a3f1bc5..47376ba 100644
> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -3694,7 +3694,7 @@ tracing_mark_write(struct file *filp, const char __user *ubuf,
>  	int nr_pages = 1;
>  	ssize_t written;
>  	void *page1;
> -	void *page2;
> +	void *uninitialized_var(page2);
>  	int offset;
>  	int size;
>  	int len;

We had this already - gcc 4.5 warns incorrectly about this, you should
upgrade to 4.6: http://marc.info/?l=linux-kernel&m=132083844427859

-- 
Regards/Gruss,
Boris.

Re: [PATCH] Trace: initialize variable to clear warning

[Steven Rostedt]

On Tue, 2012-02-28 at 17:09 +0800, Zhengwang Ruan wrote:
> GCC warns that 'page2' is used without being initialized previously before
> being used, this is to clear it.
> 

This is a bug in gcc that's fixed in 4.6 and beyond. I don't want to
remove this warning because it will hide a real bug if page2 really does
become uninitialized. If 4.6 gives a warning here, then I'll fix it.

Note, we can't just initialize it here either, because it is initialized
in the code, and if it gets used later without that initialization then
it will break. No default will work.

Thanks,

-- Steve

> Signed-off-by: Zhengwang Ruan <ruan.zhengwang@gmail.com>
> ---
>  kernel/trace/trace.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> index a3f1bc5..47376ba 100644
> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -3694,7 +3694,7 @@ tracing_mark_write(struct file *filp, const char __user *ubuf,
>  	int nr_pages = 1;
>  	ssize_t written;
>  	void *page1;
> -	void *page2;
> +	void *uninitialized_var(page2);
>  	int offset;
>  	int size;
>  	int len;

Re: [PATCH] Trace: initialize variable to clear warning

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 477 bytes --]

On Tue, 28 Feb 2012 09:16:31 EST, Steven Rostedt said:

> This is a bug in gcc that's fixed in 4.6 and beyond. I don't want to
> remove this warning because it will hide a real bug if page2 really does
> become uninitialized. If 4.6 gives a warning here, then I'll fix it.

> > -	void *page2;
> > +	void *uninitialized_var(page2);

How would you feel about this?

-	void *page2;
+	void *page2; /* gcc 4.5 bug causes incorrect "uninitialized" warning */

or something similar?


[-- Attachment #2: Type: application/pgp-signature, Size: 865 bytes --]

Re: [PATCH] Trace: initialize variable to clear warning

[Steven Rostedt]

On Wed, 2012-02-29 at 17:40 -0500, Valdis.Kletnieks@vt.edu wrote:
> On Tue, 28 Feb 2012 09:16:31 EST, Steven Rostedt said:
> 
> > This is a bug in gcc that's fixed in 4.6 and beyond. I don't want to
> > remove this warning because it will hide a real bug if page2 really does
> > become uninitialized. If 4.6 gives a warning here, then I'll fix it.
> 
> > > -	void *page2;
> > > +	void *uninitialized_var(page2);
> 
> How would you feel about this?
> 
> -	void *page2;
> +	void *page2; /* gcc 4.5 bug causes incorrect "uninitialized" warning */
> 
> or something similar?

/* if you get a warning here, upgrade your gcc to 4.6 */


-- Steve

Re: [PATCH] Trace: initialize variable to clear warning

[Zhengwang Ruan]

-------- Original Message --------
From: Steven Rostedt
Sent: 2012年02月29日 星期三 20时09分18秒
To: Valdis.Kletnieks
Subject: Re: [PATCH] Trace: initialize variable to clear warning
> On Wed, 2012-02-29 at 17:40 -0500, Valdis.Kletnieks@vt.edu wrote:
>> On Tue, 28 Feb 2012 09:16:31 EST, Steven Rostedt said:
>>
>>> This is a bug in gcc that's fixed in 4.6 and beyond. I don't want to
>>> remove this warning because it will hide a real bug if page2 really does
>>> become uninitialized. If 4.6 gives a warning here, then I'll fix it.
>>>> -	void *page2;
>>>> +	void *uninitialized_var(page2);
>> How would you feel about this?
>>
>> -	void *page2;
>> +	void *page2; /* gcc 4.5 bug causes incorrect "uninitialized" warning */
>>
>> or something similar?
> /* if you get a warning here, upgrade your gcc to 4.6 */
>

OK, I will resend this patch.

Thanks,
-Zhengwang

> -- Steve
>
>
>
>

Re: [PATCH] Trace: initialize variable to clear warning

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 272 bytes --]

On Thu, 01 Mar 2012 10:08:56 +0800, Zhengwang Ruan said:

> > /* if you get a warning here, upgrade your gcc to 4.6 */
> >
> 
> OK, I will resend this patch.

For the version with Steven's suggested comment:

Acked-By: Valdis Kletnieks <valdis.kletnieks@vt.edu>

[-- Attachment #2: Type: application/pgp-signature, Size: 865 bytes --]