bug on mainline

bug on mainline

[Andreas Jaeger]

With current GCC 3.1 and current binutils I get this error when
compiling glibc CVS on i686-linux:

/tmp/ccDWlLRk.s: Assembler messages:
/tmp/ccDWlLRk.s:7722: Error: symbol `daylight' can not be both weak and common
/tmp/ccDWlLRk.s:7722: Error: symbol `timezone' can not be both weak and common
make[2]: *** [/builds/glibc/gcc-3.1/time/tzset.o] Error 1

Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj

Re: bug on mainline

[Ulrich Drepper]

[-- Attachment #1: Type: text/plain, Size: 920 bytes --]

On Sat, 2002-03-02 at 22:41, Andreas Jaeger wrote:
> 
> With current GCC 3.1 and current binutils I get this error when
> compiling glibc CVS on i686-linux:
> 
> /tmp/ccDWlLRk.s: Assembler messages:
> /tmp/ccDWlLRk.s:7722: Error: symbol `daylight' can not be both weak and common
> /tmp/ccDWlLRk.s:7722: Error: symbol `timezone' can not be both weak and common
> make[2]: *** [/builds/glibc/gcc-3.1/time/tzset.o] Error 1

I've seen in the gcc ChangeLogs this horrible mistake called
-fno-zero-initialized-in-bss.  Are you using a compiler with this?

Unless the default gets set back to something reasonable we have to add
a configure test to detect a compiler with this option.

-- 
---------------.                          ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Red Hat          `--' drepper at redhat.com   `------------------------

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 232 bytes --]

Re: bug on mainline

[Richard Henderson]

On Sat, Mar 02, 2002 at 10:54:39PM -0800, Ulrich Drepper wrote:
> > With current GCC 3.1 and current binutils I get this error when
> > compiling glibc CVS on i686-linux:
> > 
> > /tmp/ccDWlLRk.s: Assembler messages:
> > /tmp/ccDWlLRk.s:7722: Error: symbol `daylight' can not be both
> > weak and common
[...]
> I've seen in the gcc ChangeLogs this horrible mistake called
> -fno-zero-initialized-in-bss.  Are you using a compiler with this?

I don't think it's a mistake in theory.

But the patch was supposed to generate proper bss symbols,
not common symbols.  Poke Kaveh for a start; I'll look at
it if he doesn't take care of it.


r~

Re: bug on mainline

[Richard Henderson]

On Sun, Mar 03, 2002 at 12:33:23AM -0800, Richard Henderson wrote:
> But the patch was supposed to generate proper bss symbols,
> not common symbols.  Poke Kaveh for a start; I'll look at
> it if he doesn't take care of it.

Nevermind, I found it anyway.


r~



Index: c-decl.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/c-decl.c,v
retrieving revision 1.303
diff -u -p -r1.303 c-decl.c
--- c-decl.c	2002/02/28 13:29:47	1.303
+++ c-decl.c	2002/03/03 08:54:38
@@ -3476,7 +3476,7 @@ start_decl (declarator, declspecs, initi
      initializer equal to zero.  (Section 3.7.2)
      -fno-common gives strict ANSI behavior.  Usually you don't want it.
      This matters only for variables with external linkage.  */
-  if (! flag_no_common || ! TREE_PUBLIC (decl))
+  if (!initialized && (! flag_no_common || ! TREE_PUBLIC (decl)))
     DECL_COMMON (decl) = 1;
 
   /* Set attributes here so if duplicate decl, will have proper attributes.  */

Re: bug on mainline

[Andreas Jaeger]

Ulrich Drepper <drepper@redhat.com> writes:

> On Sat, 2002-03-02 at 22:41, Andreas Jaeger wrote:
>> 
>> With current GCC 3.1 and current binutils I get this error when
>> compiling glibc CVS on i686-linux:
>> 
>> /tmp/ccDWlLRk.s: Assembler messages:
>> /tmp/ccDWlLRk.s:7722: Error: symbol `daylight' can not be both weak and common
>> /tmp/ccDWlLRk.s:7722: Error: symbol `timezone' can not be both weak and common
>> make[2]: *** [/builds/glibc/gcc-3.1/time/tzset.o] Error 1
>
> I've seen in the gcc ChangeLogs this horrible mistake called
> -fno-zero-initialized-in-bss.  Are you using a compiler with this?

Yes, I do.  Ok, that might explain it.

Thanks,
Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj