This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[C++ PATCH] Add sorry() when mangling CALL_EXPR


this patch adds a sorry() when write_expression meets a CALL_EXPR. Since
this is a defect in the C++ ABI, we currently have no way to mangle
correctly such an expression. This patch lets the compilation aborts
gracefully instead of segfaulting.

This issue was discussed here:

This patch has been bootstrapped on i686-pc-cygwin (c, c++, java), and
regtested (together with my following patch) with "make check-g++" with no
new regressions. OK for mainline? I suggest to include this also in the

Giovanni Bajo

2003-06-25  Giovanni Bajo  <>

        * mangle.c (write_expression): Exit gracefully when trying to
        mangle a CALL_EXPR.

2003-06-25  Giovanni Bajo  <>

        * g++.dg/other/error9.C: New test.

Index: mangle.c
RCS file: /cvsroot/gcc/gcc/gcc/cp/mangle.c,v
retrieving revision 1.68
diff -c -w -p -r1.68 mangle.c
*** mangle.c    3 Apr 2003 03:45:48 -0000       1.68
--- mangle.c    24 Jun 2003 22:56:27 -0000
*************** write_expression (expr)
*** 1986,1991 ****
--- 1986,1995 ----

        switch (code)
+         case CALL_EXPR:
+         sorry ("call_expr cannot be mangled due to a defect in the C++
+         break;
          case CAST_EXPR:
          write_type (TREE_TYPE (expr));
          write_expression (TREE_VALUE (TREE_OPERAND (expr, 0)));

// { dg-do compile }
// Make sure we emit a decent error message when trying to mangle an
//  expression not supported by the C++ ABI due to a defect.

template <int N>
struct A {};

struct B
  static int foo(void);

template <class T>
A<sizeof(T::foo())> func(void);

int main()
  func<B>();  // { dg-error "sorry, unimplemented" }

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]