fcs 147 shark preliminary patches.

[Chris Phillips]

Hi Damien,

Thanks!

Yes I know that Xerxes has submitted the 2
changes you point out, I stand on the shoulders of giants!
(I am an amateur in both zero and shark and a total
newby in LLVM... )

My Intent is to get a build of icedtea zero/shark with jdk7 147 fcs  and
begin
the work of debugging zero and shark. The 2 patches Xerxes provided were
necessary but not sufficient for b147. 

I now have both product and debug icedtea with zero and shark building
albeit with a very hacked Makefile (remove some patches)  and only with
llvm 2.9, (2.8 can't find getLShr etal?) but shark is dying very early :
#  Internal Error
(/home/local/icedtea/latest/icedtea7/20110708/icedtea7/openjdk/hotspot/src/share/vm/interpreter/linkResolver.cpp:807),
pid=6904, tid=47420332263168
# 
assert(instanceKlass::cast(resolved_method->method_holder())->is_linked())
failed: must be linked

[linkResolver.cpp:807 looks funny... assert is there twice? ]

and  zero is in need of further jsr292 work, which is why I am leery of
integrating the patches yet.  :-(

I am attaching yet another patch set, (except for one, what it takes to
get debug to build
and that one owes its origin to Xerxes, too. It may be a possible
candidate for the
failure, but more work is needed.)

Chris

On 14/07/11 06:58 PM, Damien Raude-Morvan wrote:
> Hi Chris,
>
> Le mercredi 13 juillet 2011 18:42:12, Andrew John Hughes a écrit :
>> On Wed, Jul 13, 2011 at 11:42:00AM -0400, Chris Phillips wrote:
>>> Hi Andrew
>>>
>>> On 13/07/11 11:26 AM, Andrew John Hughes wrote:
>>>> What version of IcedTea is this for?
>>> Its originally for openjdk jdk7 fcs (b147)
>>> for icedtea7, now matches
>> Ok, probably easiest to just work against the IcedTea7 &
>> IcedTea8 forests (http://icedtea.classpath.org/hg/icedtea8-forest/hotspot
>> being the latter).
>> You should have push access (if not, ask Mark Wielaard).
>>
>> I don't know who can actually review this; to my knowledge, only
>> Gary and Xerxes will be familiar enough with Shark.  If it's broken
>> without this patch, and it works with it, I'm happy enough for you
>> to commit it, but please check with Xerxes first.
> Parts of your patch [1] have already been reported/fixed by others :
>
> - src/cpu/zero/vm/stack_zero.cpp issue has been reported as #753 [2] and 
> Xerxes already provided a patch. It has been reviewed by John Rose [3]. I 
> think it should be pushed to icedtea7 + icedtea (until upstream merge it)
>
> - src/share/vm/runtime/vmStructs.cpp issue has been reported as #757 [4] and 
> seems ok for me (built and jtreg results). Should be pushed too.
>
> Regarding you changes to sharedRuntime_zero.cpp and methodHandles_zero.hpp it 
> seems that first reference to methodHandles_zero.hpp comes from 7045514 [5], 
> but final commit into hotspot-comp repository doesn't contains this file... a 
> bit weird.
>
> [1] http://lgonqn.org/temp/ChrisPhi/webrev/
> [2] http://icedtea.classpath.org/bugzilla/show_bug.cgi?id=753
> [3] http://mail.openjdk.java.net/pipermail/zero-dev/2011-July/000387.html
> [4] http://icedtea.classpath.org/bugzilla/show_bug.cgi?id=757
>
> [5] http://cr.openjdk.java.net/~twisti/7045514/
> [6] http://hg.openjdk.java.net/hsx/hotspot-comp/hotspot/rev/cba7b5c2d53f
>
> Cheers,

-------------- next part --------------
A non-text attachment was scrubbed...
Name: hotspot.patch
Type: text/x-patch
Size: 7798 bytes
Desc: not available
Url : http://mail.openjdk.java.net/pipermail/distro-pkg-dev/attachments/20110715/29383738/hotspot.patch 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: webrev.zip
Type: application/zip
Size: 337556 bytes
Desc: not available
Url : http://mail.openjdk.java.net/pipermail/distro-pkg-dev/attachments/20110715/29383738/webrev.zip