RFR: 8247236: fieldDescriptor::print_on_for prints extra newline after NULL

Stefan Karlsson stefan.karlsson at oracle.com
Tue Jun 9 15:30:15 UTC 2020


Thanks for reviewing!

StefanK

On 2020-06-09 17:27, Thomas Schatzl wrote:
> Hi Stefan,
>
> On 09.06.20 17:05, Stefan Karlsson wrote:
>> Hi all,
>>
>> Please review this small patch to remove a newline that is 
>> incorrectly printed after NULL.
>>
>> https://cr.openjdk.java.net/~stefank/8247236/webrev.01/
>> https://bugs.openjdk.java.net/browse/JDK-8247236
>>
>> Ioi reported that this bug could be seen by running:
>>
>> ---
>> (gdb) call pp(m)
>> "Executing pp"
>> java.lang.Class
>> {0x00000007bf0008b0} - klass: 'java/lang/Class'
>>   - ---- fields (total size 14 words):
>>   - private volatile transient 'classRedefinedCount' 'I' @12 0
>>   - private volatile transient strict 'cachedConstructor' 
>> 'Ljava/lang/reflect/Constructor;' @40 NULL
>>   (0)
>>   - private transient 'name' 'Ljava/lang/String;' @44 NULL
>>   (0)
>>
>> There are newlines between NULL and (0).
>> ---
>>
>> This can be seen whenever an InstanceKlass is printed and one of the 
>> oops are NULL. There's no newline if the object is non-null or if 
>> printing primitives. Note, that this is only printed in non-product 
>> builds, making it less likely that someone depends on the output format.
>>
>> After the patch, the lines look like this:
>>   - private volatile transient strict 'cachedConstructor' 
>> 'Ljava/lang/reflect/Constructor;' @40 NULL (0)
>>   - private transient 'name' 'Ljava/lang/String;' @44 NULL (0)
>
> thanks a lot for fixing up the hs_err log output. Looks good.
>
> Thomas



More information about the hotspot-dev mailing list