[PATCH 1/3] lib/kobject: simplify the kobject_init function

[PATCH 1/3] lib/kobject: simplify the kobject_init function

[Dong Hao]

From: Dong Hao <haodong@linux.vnet.ibm.com>

The printk() function at the end of function kobject_init() already had '\n',
so remove the duplicated one.

Signed-off-by: Dong Hao <haodong@linux.vnet.ibm.com>
---
 lib/kobject.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/kobject.c b/lib/kobject.c
index e07ee1f..279a172 100644
--- a/lib/kobject.c
+++ b/lib/kobject.c
@@ -274,7 +274,7 @@ void kobject_init(struct kobject *kobj, struct kobj_type *ktype)
 		goto error;
 	}
 	if (!ktype) {
-		err_str = "must have a ktype to be initialized properly!\n";
+		err_str = "must have a ktype to be initialized properly!";
 		goto error;
 	}
 	if (kobj->state_initialized) {
-- 
1.7.9.5

[PATCH 2/3] lib/kobject: Panic when kobj or ktype is not properly assigned

[Dong Hao]

From: Dong Hao <haodong@linux.vnet.ibm.com>

kobj and ktype are two important attributes which will be used after kobject_init(),
and (!kobj)|(!ktype) may cause FS corruption which could not be recovered.
Panic instead of dump_stack() when neither kobj nor ktype is properly assigned to
detect the bug early.

Signed-off-by: Dong Hao <haodong@linux.vnet.ibm.com>
---
 lib/kobject.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/kobject.c b/lib/kobject.c
index 279a172..ff9b3c3 100644
--- a/lib/kobject.c
+++ b/lib/kobject.c
@@ -290,7 +290,7 @@ void kobject_init(struct kobject *kobj, struct kobj_type *ktype)
 
 error:
 	printk(KERN_ERR "kobject (%p): %s\n", kobj, err_str);
-	dump_stack();
+	BUG_ON(1);
 }
 EXPORT_SYMBOL(kobject_init);
 
-- 
1.7.9.5

[PATCH 3/3] lib/kobject: WARN_ON corresponds to BUG_ON

[Dong Hao]

From: Dong Hao <haodong@linux.vnet.ibm.com>

WARN_ON corresponds to BUG_ON and also gets more info.

Signed-off-by: Dong Hao <haodong@linux.vnet.ibm.com>
---
 lib/kobject.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/kobject.c b/lib/kobject.c
index ff9b3c3..f7d9f31 100644
--- a/lib/kobject.c
+++ b/lib/kobject.c
@@ -281,7 +281,7 @@ void kobject_init(struct kobject *kobj, struct kobj_type *ktype)
 		/* do not error out as sometimes we can recover */
 		printk(KERN_ERR "kobject (%p): tried to init an initialized "
 		       "object, something is seriously wrong.\n", kobj);
-		dump_stack();
+		WARN_ON(1);
 	}
 
 	kobject_init_internal(kobj);
-- 
1.7.9.5

Re: [PATCH 2/3] lib/kobject: Panic when kobj or ktype is not properly assigned

[Xiao Guangrong]

On 03/19/2013 02:47 PM, Dong Hao wrote:
> From: Dong Hao <haodong@linux.vnet.ibm.com>
> 
> kobj and ktype are two important attributes which will be used after kobject_init(),

> and (!kobj)|(!ktype) may cause FS corruption which could not be recovered.

delete this line, i am not very convinced about corruption  ...

> Panic instead of dump_stack() when neither kobj nor ktype is properly assigned to
> detect the bug early.
> 
> Signed-off-by: Dong Hao <haodong@linux.vnet.ibm.com>
> ---
>  lib/kobject.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/lib/kobject.c b/lib/kobject.c
> index 279a172..ff9b3c3 100644
> --- a/lib/kobject.c
> +++ b/lib/kobject.c
> @@ -290,7 +290,7 @@ void kobject_init(struct kobject *kobj, struct kobj_type *ktype)
> 
>  error:
>  	printk(KERN_ERR "kobject (%p): %s\n", kobj, err_str);
> -	dump_stack();
> +	BUG_ON(1);
>  }
>  EXPORT_SYMBOL(kobject_init);
> 

Re: [PATCH 3/3] lib/kobject: WARN_ON corresponds to BUG_ON

[Xiao Guangrong]

On 03/19/2013 02:47 PM, Dong Hao wrote:
> From: Dong Hao <haodong@linux.vnet.ibm.com>
> 
> WARN_ON corresponds to BUG_ON and also gets more info.

Change the log like this:
like the comments said, double initialize a kobj is wrong
but can be recoverd, using WARN_ON to warn the developer also
it can get more info.

(Or you refine the code by yourself.)

> 
> Signed-off-by: Dong Hao <haodong@linux.vnet.ibm.com>
> ---
>  lib/kobject.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/lib/kobject.c b/lib/kobject.c
> index ff9b3c3..f7d9f31 100644
> --- a/lib/kobject.c
> +++ b/lib/kobject.c
> @@ -281,7 +281,7 @@ void kobject_init(struct kobject *kobj, struct kobj_type *ktype)
>  		/* do not error out as sometimes we can recover */
>  		printk(KERN_ERR "kobject (%p): tried to init an initialized "
>  		       "object, something is seriously wrong.\n", kobj);
> -		dump_stack();
> +		WARN_ON(1);
>  	}
> 
>  	kobject_init_internal(kobj);
> 

Re: [PATCH 3/3] lib/kobject: WARN_ON corresponds to BUG_ON

[Xiao Guangrong]

On 03/19/2013 03:03 PM, Xiao Guangrong wrote:
> On 03/19/2013 02:47 PM, Dong Hao wrote:
>> From: Dong Hao <haodong@linux.vnet.ibm.com>
>>
>> WARN_ON corresponds to BUG_ON and also gets more info.
> 
> Change the log like this:
> like the comments said, double initialize a kobj is wrong
> but can be recoverd, using WARN_ON to warn the developer also
> it can get more info.
> 
> (Or you refine the code by yourself.)

s/code/comment

> 
>>
>> Signed-off-by: Dong Hao <haodong@linux.vnet.ibm.com>
>> ---
>>  lib/kobject.c |    2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/lib/kobject.c b/lib/kobject.c
>> index ff9b3c3..f7d9f31 100644
>> --- a/lib/kobject.c
>> +++ b/lib/kobject.c
>> @@ -281,7 +281,7 @@ void kobject_init(struct kobject *kobj, struct kobj_type *ktype)
>>  		/* do not error out as sometimes we can recover */
>>  		printk(KERN_ERR "kobject (%p): tried to init an initialized "
>>  		       "object, something is seriously wrong.\n", kobj);
>> -		dump_stack();
>> +		WARN_ON(1);
>>  	}
>>
>>  	kobject_init_internal(kobj);
>>
> 

Re: [PATCH 2/3] lib/kobject: Panic when kobj or ktype is not properly assigned

[Greg KH]

On Tue, Mar 19, 2013 at 02:47:00PM +0800, Dong Hao wrote:
> From: Dong Hao <haodong@linux.vnet.ibm.com>
> 
> kobj and ktype are two important attributes which will be used after kobject_init(),
> and (!kobj)|(!ktype) may cause FS corruption which could not be recovered.
> Panic instead of dump_stack() when neither kobj nor ktype is properly assigned to
> detect the bug early.
> 
> Signed-off-by: Dong Hao <haodong@linux.vnet.ibm.com>
> ---
>  lib/kobject.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/lib/kobject.c b/lib/kobject.c
> index 279a172..ff9b3c3 100644
> --- a/lib/kobject.c
> +++ b/lib/kobject.c
> @@ -290,7 +290,7 @@ void kobject_init(struct kobject *kobj, struct kobj_type *ktype)
>  
>  error:
>  	printk(KERN_ERR "kobject (%p): %s\n", kobj, err_str);
> -	dump_stack();
> +	BUG_ON(1);

You just crashed the machine.  That's horrible, and unacceptable.

*plonk*

Re: [PATCH 1/3] lib/kobject: simplify the kobject_init function

[Greg KH]

On Tue, Mar 19, 2013 at 02:46:59PM +0800, Dong Hao wrote:
> From: Dong Hao <haodong@linux.vnet.ibm.com>
> 
> The printk() function at the end of function kobject_init() already had '\n',
> so remove the duplicated one.

Your subject doesn't match this description.

Re: [PATCH 3/3] lib/kobject: WARN_ON corresponds to BUG_ON

[Greg KH]

On Tue, Mar 19, 2013 at 02:47:01PM +0800, Dong Hao wrote:
> From: Dong Hao <haodong@linux.vnet.ibm.com>
> 
> WARN_ON corresponds to BUG_ON and also gets more info.

Huh?  No.

Please stop, don't resend these.

greg k-h