undefined ia64_double_zero in hs19

[David Holmes]

I've initiated the steps to getting this fixed.

David Holmes

Dr Andrew John Hughes said the following on 11/29/10 23:12:
> On 11:05 Mon 29 Nov     , Gary Benson wrote:
>> Matthias Klose wrote:
>>> On 29.11.2010 10:18, David Holmes wrote:
>>>> Matthias Klose said the following on 11/29/10 18:06:
>>>>> The hotspot build fails on ia64, ia64_double_zero is not defined
>>>>> anymore. The code references a "IA64 gcc bug", but I cannot find
>>>>> any open GCC report for this issue. Is this fixed upstream, or
>>>>> is the workaround still necessary? If it's the latter, please
>>>>> restore the ia64_double_zero definition.
>>>> That was done a year ago:
>>>>
>>>> http://hg.openjdk.java.net/jdk7/hotspot-comp/hotspot/rev/8e7adf982378
>>>>
>>>> I suspect it was seen as some opportune "clean up".
>>> apparently it's still in OpenJDK6.
>> That surprises me... doesn't OpenJDK6 regularly get it's HotSpot
>> replaced by the latest and greatest?  
> 
> I regularly merge in the stable branches when they become available.
> We never merge from OpenJDK7 directly.
> 
>> This fix has been in HotSpot
>> for almost as long as Zero, so it *should* be there...
>>
> 
> It was.  Someone decided to revert it... :-S
> Clearly they didn't actually build on ia64.
> 
>> IcedTea always used to have a patch for this.   Could it be that
>> a now-unrequired IcedTea patch is being reversed?
>>
> 
> We removed the patch in IcedTea when it went upstream, first through
> our own backport and then through the import of the hs17 stable branch.
> 
> Importing hs19 reverted the fix by bringing in 6953477.
> 
> So can someone please fix this again?
> 
>> Cheers,
>> Gary
>>
>> -- 
>> http://gbenson.net/
>