On Wed, Aug 7, 2013 at 11:51 AM, Paul Pluzhnikov <ppluzhni...@google.com> wrote:
>
> The following source:
>
> template<typename T> static void f();
> void g() { f<int>(); }
>
> results in "_Z1fIiEvv" under g++, but in "_ZL1fIiEvv" under clang.
>
> Richard Smith says:
>
> The ABI doesn't cover manglings for local symbols ...
> ... and c++filt is not able to cope with the L prefix here.
>
> I'm having a hard time seeing how this isn't a g++ bug and a matching
> c++filt bug.
>
> It's hard for me to argue that this is a 'g++' bug (since there is no
> ABI violation here), but c++filt should be able to handle this, and does
> with attached patch.
OK, I finally took a look.
The demangler treates _ZL as a local-source-name. This is not part of
the C++ ABI (http://codesourcery.com/cxx-abi/). It was added by Geoff
Keating: http://gcc.gnu.org/PR31775,
http://gcc.gnu.org/ml/gcc-patches/2007-05/msg00309.html .
With Geoff's patch the demangling is
_Z <encoding>
<encoding> => <(data) name>
_Z <(data) name>
<name> => <unscoped-name>
_Z <unscoped-name>
<unscoped-name> => <unqualified-name>
_Z <unqualified-name>
<unqualified-name> => <local-source-name> (not in ABI)
_Z <local-source-name>
<local-source-name> => L <source-name> [<discriminator>] (not in ABI)
_ZL <source-name> [<discriminator>]
<source-name> => <number> <identifier>
_ZL1f [<discriminator>]
The discriminator is not present, and at this point we are in trouble
because IiE is extraneous.
But really the demangling should be
_Z <encoding>
<encoding> => <(data) name>
_Z <(data) name>
<name> => <unscoped-template-name> <template-args>
_Z <unscoped-template-name> <template-args>
<unscoped-template-name> => <unscoped-name>
_Z <unscoped-name> <template-args>
<unscoped-name> => <unqualified-name>
_Z <unqualified-name> <template-args>
<unqualified-name> => <local-source-name> (not in ABI)
_Z <local-source-name> <template-args>
<local-source-name> => L <source-name> [<discriminator>] (not in ABI)
_ZL <source-name> [<discriminator>] <template-args>
<source-name> => <number> <identifier>
_ZL1f [<discriminator>] <template-args>
[<discriminator>] =>
_ZL1f <template-args>
etc.
So it looks to me like Geoff's original patch was wrong, and your
patch is correct.
So this is OK.
Thanks.
Sorry for ignoring this for so long.
Ian
