XS RFR: 8006103: [parfait] Possible null pointer dereference at hotspot/src/os/linux/vm/os_linux.cpp; os_windows.cpp; os_solaris.cpp; os_bsd.cpp

[Daniel D. Daugherty]

The four files showed up as merge conflicts in my preliminary sync from
Main_Baseline -> RT_Baseline. I can easily back out this fix as part of
this week's sync.

Dan


On 4/5/13 5:43 AM, Coleen Phillimore wrote:
> On 4/4/2013 9:09 PM, David Holmes wrote:
>> Hi Calvin,
>>
>> On 5/04/2013 11:06 AM, Calvin Cheung wrote:
>>> I've filed https://jbs.oracle.com/bugs/browse/JDK-8011588 to backout 
>>> the
>>> fix and will contact parfait team about this false positive.
>>
>> I don't want to see even more time and resources wasted on this. 
>> Discuss with Karen whether to back this out or just leave it.
>
> I agree, this is not worth spending more time on because it was a 
> small change.  There should be a parfait bug filed for the false 
> positive so we don't have to fix any others.
>
> Coleen
>
>>
>> Thanks,
>> David
>>
>>> Calvin
>>>
>>> On 4/4/2013 4:58 PM, serguei.spitsyn at oracle.com wrote:
>>>> On 4/4/13 2:16 PM, David Holmes wrote:
>>>>> On 5/04/2013 4:50 AM, Calvin Cheung wrote:
>>>>>> On 4/3/2013 9:56 PM, David Holmes wrote:
>>>>>>> On 2/04/2013 3:10 AM, Calvin Cheung wrote:
>>>>>>>> Simple fix is to add null check on pelements.
>>>>>>>
>>>>>>> This warning is bogus. You will only get NULL if n==0 and if 
>>>>>>> n==0 then
>>>>>>> you don't dereference pelements.
>>>>>> After reading the os::split_path(), n will always be set to 0 if 
>>>>>> a NULL
>>>>>> is returned.
>>>>>> Yes, I agreed with the above.
>>>>>> But adding a null check should do no harm and make the code 
>>>>>> easier to
>>>>>> understand.
>>>>>
>>>>> I disagree. Adding the null check is redundant overhead that is
>>>>> completely unnecessary.
>>>>>
>>>>>> I'll check with the parfait dev team and see if they can make the 
>>>>>> tool
>>>>>> "smarter".
>>>>>
>>>>> This CR should be marked as a false positive and no code change
>>>>> should be made in my opinion.
>>>>
>>>> Adding extra code just to make the parfait happy does not look right.
>>>> This particular case is a false positive and needs to be fixed in the
>>>> tool.
>>>>
>>>> Thanks,
>>>> Serguei
>>>>
>>>>>
>>>>> David
>>>>>
>>>>>> thanks,
>>>>>> Calvin
>>>>>>
>>>>>>>
>>>>>>> David
>>>>>>> -----
>>>>>>>
>>>>>>>> webrev: http://cr.openjdk.java.net/~ccheung/8006103/webrev/
>>>>>>>> bug: https://jbs.oracle.com/bugs/browse/JDK-8006103
>>>>>>>>
>>>>>>>> Test: jprt run was successful.
>>>>>>>>
>>>>>>>> thanks,
>>>>>>>> Calvin
>>>>>>
>>>>
>>>
>
>
>