Jim Li wrote:
> Evan,
>
> You are right. It's g_free(). The fact is that g_free() on an 
> uninitialized  pointer  cause  a  crash  on  SPARC, not NULL  pointer.
> Sorry for not accurate description!
>   
Then I think the right fix is to add proper initialization for the
variable, but not to remove the g_free() call, right?

Best,
-Evan

> Thanks
> jim
>   
>> Jim,
>>
>> I'm confused. In your patch, it's g_free(), not free().
>>
>> And in free()'s man page, it says  "If ptr is a null pointer, no action
>> occurs."
>> I wrote a simple test program, free(0) doesn't crash on my box.
>>
>> -Evan
>>
>> Jim Li wrote:
>>   
>>     
>>> Evan ,
>>>   
>>>     
>>>       
>>>> g_free() on a NULL pointer should be ok with no crash.
>>>> If it crash at g_free (desc), it looks more like a memory corruption to
>>>> me, i.e. desc pointed to some illegal memory.
>>>>   
>>>>     
>>>>       
>>>>         
>>> It's free() not g_free(). :-)
>>>
>>> free() on a NULL pointer works on X86, but not on SPARC.
>>>
>>> Jim
>>>   
>>>     
>>>       
>>>> -Evan
>>>>
>>>> Jim Li wrote:
>>>>   
>>>>     
>>>>       
>>>>         
>>>>> All,
>>>>>
>>>>> This patch just remove free() for a null pointer which cause the crash.
>>>>>
>>>>> Thanks
>>>>> Jim
>>>>>   
>>>>>     
>>>>>       
>>>>>         
>>>>>           
>>>   
>>>     
>>>       
>
>   

Reply via email to