Re: PATCH: Add --enable-__cxa_atexit

["H . J . Lu" <hjl at lucon dot org>]

On Fri, May 31, 2002 at 11:21:18AM -0700, Zack Weinberg wrote:
> On Fri, May 31, 2002 at 10:43:11AM -0700, H . J . Lu wrote:
> > +# Enable __cxa_atexit for C++.
> > +AC_ARG_ENABLE(__cxa_atexit,
> > +[  --enable-__cxa_atexit   enable __cxa_atexit for C++],
> > +[], [enable___cxa_atexit=no])
> > +if test x$enable___cxa_atexit != xno; then
> > +  AC_DEFINE(ENABLE_CXA_ATEXIT, 1,
> > +  [Define if you want to use __cxa_atexit, rather than, atexit to
> > +   register C++ destructors for local statics and global objects.])
> > +fi
> 
> 1) Why is this an --enable switch rather than a tm.h #define?  This is
>    the major concern with the patch.

There is no way to reliably check if target's __cxa_atexit is available
and correct. Think about cross compiler. I couldn't even figure out
a clean way to check if it is available for cross compiler. I don't
want to enable it for Linux by default since it is a relatively new
feature in glibc.

> 
> 2) No comma after "than".
> 3) Call the macro DEFAULT_USE_CXA_ATEXIT.
> 

Here is an update.

Thanks.


H.J.
----
2002-05-31  H.J. Lu  (hjl@gnu.org)

	* configure.in: Add --enable-__cxa_atexit.
	* configure: Regenerated.
	* config.in: Likewise.

	* defaults.h (DEFAULT_USE_CXA_ATEXIT): New. Defined to be 0 if
	not defined.

2002-05-31  H.J. Lu  (hjl@gnu.org)

	* decl2.c (flag_use_cxa_atexit): Set to DEFAULT_USE_CXA_ATEXIT.

--- gcc/config.in.atexit	Thu May 23 16:28:28 2002
+++ gcc/config.in	Fri May 31 11:36:31 2002
@@ -296,6 +296,10 @@
    every opportunity.  This is extremely expensive. */
 #undef ENABLE_GC_ALWAYS_COLLECT
 
+/* Define if you want to use __cxa_atexit, rather than atexit to
+   register C++ destructors for local statics and global objects. */
+#undef DEFAULT_USE_CXA_ATEXIT
+
 /* Define if you want the C and C++ compilers to support multibyte
    character sets for source code. */
 #undef MULTIBYTE_CHARS
--- gcc/configure.in.atexit	Fri May 31 11:33:23 2002
+++ gcc/configure.in	Fri May 31 11:34:07 2002
@@ -259,6 +259,16 @@ elif test x$withval != xno; then
   cpp_install_dir=$withval
 fi])
 
+# Enable __cxa_atexit for C++.
+AC_ARG_ENABLE(__cxa_atexit,
+[  --enable-__cxa_atexit   enable __cxa_atexit for C++],
+[], [enable___cxa_atexit=no])
+if test x$enable___cxa_atexit != xno; then
+  AC_DEFINE(DEFAULT_USE_CXA_ATEXIT, 1,
+  [Define if you want to use __cxa_atexit, rather than atexit to
+   register C++ destructors for local statics and global objects.])
+fi
+  
 # Enable Multibyte Characters for C/C++
 AC_ARG_ENABLE(c-mbchar,
 [  --enable-c-mbchar       enable multibyte characters for C and C++],
--- gcc/cp/decl2.c.atexit	Tue Apr 30 10:11:37 2002
+++ gcc/cp/decl2.c	Fri May 31 11:36:07 2002
@@ -376,7 +376,7 @@ int flag_weak = 1;
 /* Nonzero to use __cxa_atexit, rather than atexit, to register
    destructors for local statics and global objects.  */
 
-int flag_use_cxa_atexit;
+int flag_use_cxa_atexit = DEFAULT_USE_CXA_ATEXIT;
 
 /* Maximum template instantiation depth.  This limit is rather
    arbitrary, but it exists to limit the time it takes to notice
--- gcc/defaults.h.atexit	Sun May 19 21:43:24 2002
+++ gcc/defaults.h	Fri May 31 11:35:49 2002
@@ -462,4 +462,10 @@ You Lose!  You must define PREFERRED_DEB
 #define MODE_BASE_REG_CLASS(MODE) BASE_REG_CLASS
 #endif
 
+/* Determine if __cxa_atexit, rather than atexit is used to register
+   C++ destructors for local statics and global objects. */
+#ifndef DEFAULT_USE_CXA_ATEXIT
+#define DEFAULT_USE_CXA_ATEXIT 0
+#endif
+
 #endif  /* ! GCC_DEFAULTS_H */