[LLVMdev] Fwd: [windows) how to use weak references with llvm 3.4 and windows?

[Carl]

Thanks.
You've been very helpful.
Le 12 févr. 2014 21:37, "Reid Kleckner" <rnk at google.com> a écrit :

> Yep, llvm::GlobalValue::ExternalWeakLinkage.
>
>
> On Wed, Feb 12, 2014 at 2:31 AM, Carl <name.is.carl at gmail.com> wrote:
>
>> So instead of using linkagetypes.weak I should use externalweaklinkage ?
>> Le 11 févr. 2014 18:29, "Reid Kleckner" <rnk at google.com> a écrit :
>>
>> You'd have to use extern_weak linkage.  Clang compiles the foo
>>> declaration to:
>>> $ clang -cc1 -emit-llvm -o - t.c | grep declare.*@foo
>>> declare extern_weak i32 @foo(...) #1
>>>
>>>
>>>
>>> On Tue, Feb 11, 2014 at 4:16 AM, Carl <name.is.carl at gmail.com> wrote:
>>>
>>>> Thanks for your clear answer. Do you know what modifier should I use to
>>>> declare such weak symbols in my llvm intermediate code?
>>>>
>>>> So that it can be compiled to the .o file with the weak attribute ?
>>>> Le 10 févr. 2014 19:44, "Reid Kleckner" <rnk at google.com> a écrit :
>>>>
>>>> COFF doesn't support the same kind of concept of 'weak' that ELF does.
>>>>>  This is the issue that users brought up with mingw:
>>>>> https://sourceware.org/bugzilla/show_bug.cgi?id=9687
>>>>>
>>>>> Instead, COFF supports weak externals, which you can use like:
>>>>>
>>>>> $ cat t.c
>>>>> int foo() __attribute__((weak));
>>>>> int main() {
>>>>>   if (foo)
>>>>>     return foo();
>>>>>   return 13;
>>>>> }
>>>>>
>>>>> $ cat t2.c
>>>>> int foo() {
>>>>>   return 42;
>>>>> }
>>>>>
>>>>> $ clang t.c t2.c -o t && ./t ; echo $?
>>>>> 42
>>>>>
>>>>> So, we got the definition of foo from t2.c.  If t2.c hadn't been
>>>>> linked, foo would be null.
>>>>>
>>>>>
>>>>> On Mon, Feb 10, 2014 at 6:28 AM, Carl <name.is.carl at gmail.com> wrote:
>>>>>
>>>>>>
>>>>>> Hello,
>>>>>>
>>>>>> I'm generating C code (and the resulting obj files) using llvm 3.4
>>>>>> for both unix and windows.
>>>>>> And I use the dreaded weak references, that, for windows, are not too
>>>>>> widely supported.
>>>>>>
>>>>>> When I link my application on linux, I have no issue.
>>>>>> But when I'm doing the same on windows using mingw I got a duplicate
>>>>>> symbol error :
>>>>>>
>>>>>> ..\robovm-0.0.8\lib\robovm-rt.jar\dalvik\system\BlockGuard$BlockGuardPolicyException.class.o:(.text$dalvik_system_BlockGuard$24BlockGuardPolicyException_getPolicy__I_lookup[_dalvik_system_BlockGuard$24BlockGuardPolicyException_getPolicy__I_lookup]+0x0):
>>>>>> multiple definition of
>>>>>> `dalvik_system_BlockGuard$24BlockGuardPolicyException_getPolicy__I_lookup'
>>>>>> ..\Temp\robovm3774596063679264132.tmp\linker.o:(.text+0x340): first
>>>>>> defined here
>>>>>> ..\robovm-0.0.8\lib\robovm-rt.jar\dalvik\system\BlockGuard$BlockGuardPolicyException.class.o:(.text$dalvik_system_BlockGuard$24BlockGuardPolicyException_getPolicyViolation__I_lookup[_dalvik_system_BlockGuard$24BlockGuardPolicyException_getPolicyViolation__I_lookup]+0x0):
>>>>>> multiple definition of
>>>>>> `dalvik_system_BlockGuard$24BlockGuardPolicyException_getPolicyViolation__I_lookup'
>>>>>> ..\Temp\robovm3774596063679264132.tmp\linker.o:(.text+0x350): first
>>>>>> defined here
>>>>>> ..\robovm-0.0.8\lib\robovm-rt.jar\dalvik\system\BlockGuard$BlockGuardPolicyException.class.o:(.text$dalvik_system_BlockGuard$24BlockGuardPolicyException_getMessage__Ljava_lang_String$3B_lookup[_dalvik_system_BlockGuard$24BlockGuardPolicyException_getMessage__Ljava_lang_String$3B_lookup]+0x0):
>>>>>> multiple definition of
>>>>>> `dalvik_system_BlockGuard$24BlockGuardPolicyException_getMessage__Ljava_lang_String$3B_lookup'
>>>>>>
>>>>>> Because those symbols are declared as weak.
>>>>>>
>>>>>> What is the status about the support of weak symbols on windows?
>>>>>> Are they supposed to work? Are they supposed to *never *work ?
>>>>>>
>>>>>> Do you know any way to fix this (or work around it). I use weak
>>>>>> symbols everywhere in my code generation, so I would prefer a fix that
>>>>>> doesn't involve to rewrite my code.
>>>>>>
>>>>>> I'm considering rewriting the obj files using objcopy.
>>>>>> Has this any chance to work ?
>>>>>> Here are the o.
>>>>>>
>>>>>> Cheers,
>>>>>> Carl.
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> LLVM Developers mailing list
>>>>>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>>>>>>
>>>>>>
>>>>>
>>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20140213/a4bec561/attachment.html>