Re: [Mingw-w64-public] Help building a win64 native compiler from Linux

[Kai T. <kti...@go...>]

2009/8/21 Ozkan Sezer <se...@gm...>:
> On Fri, Aug 21, 2009 at 10:57 AM, Kai Tietz<kti...@go...> wrote:
>> 2009/8/21 Ozkan Sezer <se...@gm...>:
>>> On Fri, Aug 21, 2009 at 10:24 AM, Kai Tietz<kti...@go...> wrote:
>>>> 2009/8/21 t6...@gm... <t6...@gm...>:
>>>>> Hello,
>>>>> I filed a bug report at gcc bug tracker
>>>>> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41136
>>>>> Is anyone encountering this bug and is there any possible workaround ?
>>>>>
>>>>> ------------------------------------------------------------------------------
>>>>> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
>>>>> trial. Simplify your report design, integration and deployment - and focus on
>>>>> what you do best, core application coding. Discover what's new with
>>>>> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
>>>>> _______________________________________________
>>>>> Mingw-w64-public mailing list
>>>>> Min...@li...
>>>>> https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
>>>>>
>>>>
>>>> Hello,
>>>>
>>>> thank you for reporting this. Please report such kind of failures to
>>>> mingw-w64 project's bug tracker first. This looks not as a gcc bug
>>>> itself, it seems to be more reasoned by other issues possibly reasoned
>>>> by our runtime.
>>>>
>>>> Please describe also the environment you are using to build the native
>>>> compiler, the date of runtime headers and crt you are using, and the
>>>> binutils version you are using.
>>>>
>>>> Regards,
>>>> Kai
>>>
>>> This happens because the autotools idiocy decides
>>> that LFS is not available and _GLIBCXX_USE_LFS is
>>> left undefined which is the first issue, it falls back to
>>> fstat instead of fstat64 and fstat wraps onto fstat64i32.
>>> This is actually a GCC bug, IMO.
>>> The second issue is that fstat64i32 is in _fstat64i32.c
>>> but fstat() being defined as an inline wrapping around
>>> it in sys/stat.h is marked __MINGW_ATTRIB_NO_OPTIMIZE
>>> The no-optimize maybe causing an issue here ?.
>>>
>>> --
>>> Ozkan
>>>
>>
>> Hmm, well the issue in libstdc++ could be a bug in it. But the issue
>
> Yes. The problem is in acinclude.m4 of libstdc++ with its
> GLIBCXX_CHECK_LFS procedure.
>
>> about double present symbols is an issue related to the fact that it
>> uses the inlined version. Therefore the __NO_INLINE__ is in our
>> headers. Is this function not protected by an __NO_INLINE__?
>
> Yes, it is protected.
>
>> This bug is reasoned IMHO by the fact, that within the file we
>> implement fstat64i32 and the fstat functions are within one .c file.
>> We should split this file, so that each function is within one
>> separate file here. So we avoid this.
>
> Can you explain? (I'm newly awoken..)
>

in crt stdio/_fstat64i32.c we implement the following functions:
"_fstat64i32", "_stat64i32", "fstat", and "stat". So if one of those
functions is already in a library (e.g _fstat64i32"), but fstat is
referenced by this library, too, then linker want to take the function
"fstat" from object stdio/_fstat64i32.o. But this object also includes
_fstat64i32 function, so it leads to double defined linker errors. We
should split this file into its four different functions.

Cheers,
Kai

-- 
|  (\_/) This is Bunny. Copy and paste
| (='.'=) Bunny into your signature to help
| (")_(") him gain world domination