PATCH: Fix demangler in symbol versioning

[H. J. Lu]

On Thu, Dec 04, 2003 at 06:11:32PM -0500, Ian Lance Taylor wrote:
> "H. J. Lu" <hjl@lucon.org> writes:
> 
> > > DMGL_TYPES tells the demangler to see if the string is the mangled
> > > version of a type encoding.  I don't think the compiler will ever use
> > > a type encoding as an external symbol name--that doesn't even really
> > > make sense--so I don't think there is any need for DMGL_TYPES here.
> > > 
> > 
> > Will it lead to incorrect output? If not, I'd like to add it.
> 
> It could conceivably lead to incorrect output.  I don't know.  For
> example, if you call cplus_demangle ("i", DMGL_TYPES), you will get
> "int".  So if there is an external variable named `i', it seems
> conceivable that the wrong thing would happen.  I haven't thought
> about whether this could really happen, though.

In that case, we have a regression with cxxfilt. The old one:

# cxxfilt i
i

The new one:

# ./cxxfilt i
int

Here is the new patch.


H.J.
----
2003-12-04  H.J. Lu  <hongjiu.lu@intel.com>

	* ldlang.c (lang_vers_match): Updated calls to cplus_demangle.

--- ld/ldlang.c.dem	2003-12-04 11:36:55.000000000 -0800
+++ ld/ldlang.c	2003-12-04 15:19:19.000000000 -0800
@@ -4996,13 +4996,13 @@ lang_vers_match (struct bfd_elf_version_
 
   if (head->mask & BFD_ELF_VERSION_CXX_TYPE)
     {
-      cxx_sym = cplus_demangle (sym, /* DMGL_NO_TPARAMS */ 0);
+      cxx_sym = cplus_demangle (sym, DMGL_PARAMS | DMGL_ANSI);
       if (!cxx_sym)
 	cxx_sym = sym;
     }
   if (head->mask & BFD_ELF_VERSION_JAVA_TYPE)
     {
-      java_sym = cplus_demangle (sym, DMGL_JAVA);
+      java_sym = java_demangle_v3 (sym);
       if (!java_sym)
 	java_sym = sym;
     }