RFR(XS): 8050228: Rename 'rem_size' in compactibleFreeListSpace.cpp because of name clashes on AIX" -f 8050228_rename_rem_size.patch

Volker Simonis volker.simonis at gmail.com
Thu Jul 17 09:44:19 UTC 2014


Hi John,

thanks a lot for reviewing and sponsoring his change.
You can find the new, adapted webrev here:

http://cr.openjdk.java.net/~simonis/webrevs/8050228.v2/

Regards,
Volker


On Wed, Jul 16, 2014 at 10:45 PM, Jon Masamitsu
<jon.masamitsu at oracle.com> wrote:
>
> On 7/15/2014 8:02 PM, Vitaly Davidovich wrote:
>>
>> How about rem_sz? :)
>
>
> Yes, rem_sz, please.  There are example of such abbreviations
> in the file.  I'll sponsor and push.
>
> Sending an hg patch with summary and reviewed-by
> info that I can import would be the easiest I think.
>
> Jon
>
>
>
>>
>> Sent from my phone
>> On Jul 15, 2014 10:14 PM, "David Holmes" <david.holmes at oracle.com> wrote:
>>
>>> On 16/07/2014 2:54 AM, Volker Simonis wrote:
>>>
>>>> Hi,
>>>>
>>>> could somebody please review and sponsor this little change:
>>>>
>>> GC code should be taken by GC folk. (aka buck passing :) )
>>>
>>> FWIW change looks fine to me. The name remain_size grates a little but
>>> spelling out remaining_size grates even more. I don't have a better
>>> suggestion but perhaps GC folk will.
>>>
>>> Cheers,
>>> David
>>>
>>>   http://cr.openjdk.java.net/~simonis/webrevs/8050228/
>>>>
>>>> https://bugs.openjdk.java.net/browse/JDK-8050228
>>>>
>>>> Background:
>>>>
>>>> I know this sounds crazy but it's true: there's an AIX header which
>>>> unconditionally defines rem_size:
>>>>
>>>> /usr/include/sys/xmem.h
>>>> struct xmem {
>>>> ...
>>>> #define rem_size u2._subspace_id2
>>>> };
>>>>
>>>> This breaks the compilation of
>>>> CompactibleFreeListSpace::splitChunkAndReturnRemainder() which uses a
>>>> local variable of the same name.
>>>>
>>>> Until now, we've worked around this problem by simply undefining
>>>> 'rem_size' in the platform specific file os_aix.inline.hpp but after
>>>> "8042195: Introduce umbrella header orderAccess.inline.hpp" this
>>>> doesn't seems to be enough any more.
>>>>
>>>> So before introducing yet another ugly platform dependent hack in
>>>> shared code or depending on a certain include order of otherwise
>>>> unrelated platform headers in shared code I suggest so simply give up
>>>> and rename the local variable.
>>>>
>>>> In this change I've renamed 'rem_size' to 'remain_size' because "rem"
>>>> is used as abbreviation of "remainder" in the code. But actually I'd
>>>> be happy with any other name which differs from "rem_size".
>>>>
>>>> Thank you and best regards,
>>>> Volker
>>>>
>>>>
>


More information about the hotspot-dev mailing list