If you look in the code, you can find the problem very quickly:
//FIXME: Should be sizeof(ModuleTy)
Elements.push_back(llvm::ConstantInt::get(LongTy, 16));
I've fixed this to pull the size from the target data. Hopefully it
should be a bit closer to working now.
David
On 27 Jan 2009, at 15:41, Niels Grewe wrote:
> On Tue, Jan 27, 2009 at 10:31:01AM +0000, David Chisnall wrote:
>> If anyone feels like doing this on an x86-64 system, I'd be
>> interested
>> to hear if it works.
>
> Okay, I just did that and MsgSendSmallInt.bc was generated by clang
> without errors. Unfortunately Smalltalk still doesn't work for me.
> All the tests from the testsuite fail because edlc comes back with the
> following error:
>
> Module (null) version 8 doesn't match runtime 8
>
> So apparently the runtime can't be convinced to go on here. I've got
> this problem both with libobjc from gcc 4.3.2 and with the one that
> shipped with gcc 4.3.3 (on Debian, at least). The error message is a
> little misleading, because judging from the code that prints the above
> message (the init_check_module_version function in libobjc's init.c)
> there is some size mismatch between the size field in the module
> struct
> (which is 16) and the value reported by sizeof(module), which is
> 32. Since I don't know what to make of that I have attached a
> backtrace
> from the IntegerAdditions test for the perusal of the more
> knowledgable.
>
> best regards,
>
>
> Niels
> <TestIntegerAddition-
> bt.txt>_______________________________________________
> Etoile-dev mailing list
> [email protected]
> https://mail.gna.org/listinfo/etoile-dev
_______________________________________________
Etoile-dev mailing list
[email protected]
https://mail.gna.org/listinfo/etoile-dev
