[PATCH FOR APPROVAL]: Support hs16

[Andrew John Hughes]

2009/10/22 Matthias Klose <doko at ubuntu.com>:
> On 22.10.2009 01:46, Andrew John Hughes wrote:
>>
>> 2009/10/21 Matthias Klose<doko at ubuntu.com>:
>>>
>>> On 21.10.2009 18:04, Andrew John Hughes wrote:
>>>>
>>>> Sun's proprietary JDK has moved to hs16 as its base: see
>>>> http://hg.openjdk.java.net/hsx/hsx16/master. We should do the same.
>>>>
>>>> The following changesets:
>>>>
>>>> http://icedtea.classpath.org/people/andrew/icedtea6/rev/71a19c78147c
>>>> http://icedtea.classpath.org/people/andrew/icedtea6/rev/093116b4e715
>>>> http://icedtea.classpath.org/people/andrew/icedtea6/rev/86a05a0cec75
>>>>
>>>> add support for building with hs16 as we did before for hs14 (though
>>>> not yet by default) i.e. --with-hotspot-build=default
>>>>
>>>> Using hs16 makes four IcedTea patches redundant, but requires a new
>>>> one of mine backported from OpenJDK7 (to ignore -XX:DeoptimizeALot on
>>>> some new tests)
>>>> (http://hg.openjdk.java.net/jdk7/jdk7/hotspot/rev/d6b9fd78f389).
>>>> With that patch applied, all HotSpot JTReg tests pass on hs16.
>>>>
>>>> I also used this opportunity to remove the redundant gcc-4.3 patch,
>>>> which broke when applied to hs16.  It doesn't appear to do anything
>>>> but attempt (and fail) to turn -Werror off. Both hs14 and hs16 build
>>>> fine with gcc 4.3 and 4.4, so there seems to be no need for this patch
>>>> any more.
>>>
>>> let's see if it is necessary when building with -fstack-protector
>>> -D_FORTIFY_SOURCE. I'll see that when building on Ubuntu.
>>>
>>
>> I'd be interested to know if it does.  I think your own patches
>> already cover this.  The gcc-4.3 patch doesn't seem to have been doing
>> anything as on my builds it certainly hasn't been stopping -Werror
>> failures in the past, and, if the current patch set is anything to go
>> by, it hasn't for anyone else either.  Part of the patch turns off
>> -Werror if an option is set to turn it on which seems simply bizarre.
>>
>>>> I did get one warning/error failure with the hs16 build,
>>>> which is fixed by another of my patches from JDK7
>>>> (http://hg.openjdk.java.net/jdk7/jdk7/hotspot/rev/5fdbe2cdf565).  I'll
>>>> document both on the wiki once this patch is approved.
>>>>
>>>> Ok for HEAD?
>>>
>>> please go for it, and thanks for work!
>>>
>>
>> Pushed! Thanks.
>
> hmm, did write this before the shark build (didn't) finish:
>
> In file included from ../generated/incls/_precompiled.incl:77:
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/asm/assembler.hpp:
> In constructor 'RegisterOrConstant::RegisterOrConstant()':
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/asm/assembler.hpp:152:
> error: 'noreg' was not declared in this scope
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/asm/assembler.hpp:
> In constructor 'RegisterOrConstant::RegisterOrConstant(intptr_t)':
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/asm/assembler.hpp:154:
> error: 'noreg' was not declared in this scope
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/asm/assembler.hpp:
> In member function 'bool RegisterOrConstant::is_register() const':
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/asm/assembler.hpp:162:
> error: 'noreg' was not declared in this scope
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/asm/assembler.hpp:
> In member function 'bool RegisterOrConstant::is_constant() const':
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/asm/assembler.hpp:163:
> error: 'noreg' was not declared in this scope
> In file included from ../generated/incls/_precompiled.incl:253:
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/interpreter/interpreter.hpp:
> In constructor 'CodeletMark::CodeletMark(InterpreterMacroAssembler*&, const
> char*, Bytecodes::Code)':
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/interpreter/interpreter.hpp:107:
> error: cannot allocate an object of abstract type
> 'InterpreterMacroAssembler'
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/cpu/zero/vm/interp_masm_zero.hpp:28:
> note:   because the following virtual functions are pure within
> 'InterpreterMacroAssembler':
> /scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/hotspot/src/share/vm/asm/assembler.hpp:319:
> note:        virtual RegisterOrConstant
> AbstractAssembler::delayed_value_impl(intptr_t*, RegisterImpl*, int)
> make[8]: *** [incls/_precompiled.incl.gch] Error 1
> make[8]: Leaving directory
> `/scratch/packages/openjdk/b17/openjdk-6-6b17~pre1/build/zerovm/openjdk/build/linux-i586/hotspot/outputdir/linux_zero_shark/product'
>


Aaah, sorry I forgot to do Zero and Shark runs... :(

This is fixed in IcedTea7, but we probably need to make this
conditional on hs16; I seem to remember it doesn't work with hs14.
-- 
Andrew :-)

Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

Support Free Java!
Contribute to GNU Classpath and the OpenJDK
http://www.gnu.org/software/classpath
http://openjdk.java.net

PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
Fingerprint: F8EF F1EA 401E 2E60 15FA  7927 142C 2591 94EF D9D8