RFR [8u] (XS) JDK-8047382: hotspot build failed with gcc version Red Hat 4.4.6-4

[Lois Foltan]

On 4/15/2015 4:44 AM, Stefan Karlsson wrote:
> Hi Lois,
>
> On 2015-04-15 01:40, Lois Foltan wrote:
>> Hello,
>>
>> Please review the following fix:
>>
>> Webrev:
>>     http://cr.openjdk.java.net/~lfoltan/bug_jdk8047382/
>>
>
> Looks good to me.
>
> Is there a reason why cast_from_oop<void *>(...) isn't used here?

Thanks Stefan for the review!  You are right, cast_from_oop could be 
used here.  I went back to my original work to get CHECK_UNHANDLED_OOPS 
working on all platforms,

http://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/2013-September/009380.html

What I do recall is that the cast_to_oop/cast_from_oop inline template 
functions were introduced primarily for the cases where an oop was being 
cast to and from an integral type.  The original change had many cases 
that directly cast to (void *) without use of the cast_*_oop functions.  
If I remember correctly the reason was we felt that having the direct 
cast in the (void *) case would be easier to read than changing all 
those cases to cast_*_oop().

Thanks,
Lois




>
> Thanks,
> StefanK
>
>> Bug: hotspot build failed with gcc version Red Hat 4.4.6-4
>>     https://bugs.openjdk.java.net/browse/JDK-8047382
>>
>> Summary of Fix:
>> Removed the Solaris specific conditionalization for casting to void * 
>> within calls to HS_DTRACE_PROBE* from DTRACE_CLASSINIT_PROBE* and 
>> DTRACE_CLASSLOAD_PROBE macro definitions.  gcc 4.4.6-4 needs this 
>> cast as well.  This only affects JDK 8u.  Thank you to Mario Torre 
>> for providing the webrev fix for this issue.
>>
>> Tests:
>>     jprt build submission to test compilation with all compilers
>>     vm.quick.testlist, hotspot/test/*
>>     Built and testing as well with a gcc 4.4.6-4 compiler (in progress)
>