[LLVMdev] Type inconsistency in LLVM 3.1: CGDebugInfo.cpp

[Gaster, Benedict]

I tracked this issue down and it turns out that Visual Studio is correct and this is a bug in the LLVM code with respect to C++11. (Note: it is valid C++ 2003, it is just that VS 2012 RC is closer to C++11 now, at least in this regard.)

The C++11 specification has change from C++ 2003 and is described in the ISO spec:

"template<class U, class V> pair(const pair<U, V>& p);
Remark: This constructor shall not participate in overload resolution unless const U& is implicitly convertible to first_type and const V& is implicitly convertible to second_type."

This is does not hold for the code example and so an explicit cast is required:

if (T.Verify() && T.isForwardDecl())
    ReplaceMap.push_back(std::make_pair(Ty.getAsOpaquePtr(),static_cast<llvm::Value*>(T)));

With this small change the code is both valid C++ 2003 and valid C++11.

I checked out the top of tree Clang and discovered that this change has already been made, which is great. 

Regards,

Ben

----------------------------
Benedict R. Gaster
Advanced Micro Devices

Fancy a trip to Berlin? Consider submitting a paper to MULTIPROG 2013: http://multiprog.ac.upc.edu/

-----Original Message-----
From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On Behalf Of Gaster, Benedict
Sent: Thursday, August 09, 2012 11:12 AM
To: Benjamin Kramer
Cc: llvmdev at cs.uiuc.edu
Subject: Re: [LLVMdev] Type inconsistency in LLVM 3.1: CGDebugInfo.cpp

Hi Ben,

Thanks that helped a lot.

The problem seems to be that with the move to C++11 we now have:

void std::vector<_Ty>::push_back(std::pair<_Ty1,_Ty2> &&)' 

and there no conversion operator that can be applied to convert:

'std::pair<_Ty1,_Ty2>' to 'std::pair<_Ty3,_Ty4> &&

Where:

      [
              _Ty1=void *,
             _Ty2=llvm::DIType
       ]
       and
       [
              _Ty3=void *,
              _Ty4=llvm::WeakVH
        ]

The reason it gives is std::pair<_Ty1,_Ty2>' to 'std::pair<_Ty3,_Ty4>, but as you point out there is indeed a conversion operator for DITYPE to WeakVH. Putting the explicit cast to give:

if (T.Verify() && T.isForwardDecl())
    ReplaceMap.push_back(std::make_pair(Ty.getAsOpaquePtr(), (llvm::MDNode *)T));

and all it well again. This does indeed seem to be a bug in VS, unless there is some C++11 changes that stops implicit conversion under std::pair.

Regards,

Ben

- Benedict R. Gaster
Enjoy Berlin and submit a paper to Multiprog 2013: http://multiprog.ac.upc.edu/

> -----Original Message-----
> From: Benjamin Kramer [mailto:benny.kra at gmail.com]
> Sent: Thursday, August 09, 2012 10:55 AM
> To: Gaster, Benedict
> Cc: llvmdev at cs.uiuc.edu
> Subject: Re: [LLVMdev] Type inconsistency in LLVM 3.1: CGDebugInfo.cpp
> 
> 
> On 09.08.2012, at 19:43, "Gaster, Benedict" <Benedict.Gaster at amd.com>
> wrote:
> 
> > I'm probably missing something simple here but in:
> >
> > CGDebugInfo.h:
> >
> >    std::vector<std::pair<void *, llvm::WeakVH> >ReplaceMap;
> >
> > but then in
> >
> > CGDebugInfo.cpp:
> >
> >   llvm::DIType TC = getTypeOrNull(Ty);
> >
> >   void * v = Ty.getAsOpaquePtr();
> >   std::pair<void *, llvm::WeakVH> tmp = std::make_pair(v, TC);
> >
> >   if (TC.Verify() && TC.isForwardDecl())
> >     ReplaceMap.push_back(Ty.getTypeOrNull(), TC);
> >
> > Note that TC is of type llvm:DIType and not WeakVH.
> >
> > What am I missing, as this does not compile under Visual Studio 2012 RC?
> 
> Tricky code, DIType (or rather its superclass DIDescriptor) implicitly 
> converts to MDNode* via a operator overload. MDNode is a subclass of 
> Value, and WeakVH is constructible from Value*.
> 
> Wouldn't surprise me if MSVC has problems eating that, lots of magic 
> involved here.
> 
> - Ben
> >
> > Regards,
> >
> > Ben
> >
> > - Benedict R. Gaster
> > Enjoy Berlin and submit a paper to Multiprog 2013:
> http://multiprog.ac.upc.edu/
> >
> > _______________________________________________
> > LLVM Developers mailing list
> > LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
> 



_______________________________________________
LLVM Developers mailing list
LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev