Re: FC4 compilation issue

["Duilio Protti" <dprotti () fceia unr edu ar>]

> That raises the question of why HAVE___ATTRIBUTE__ is not being
> defined on your FC4 box (it is defined on my FC3). The definition is
> made as part of this rather simple AC_C___ATTRIBUTE__ macro in
> nbase/aclocal.m4:
>
> dnl
> dnl Test for __attribute__
> dnl
>
> AC_DEFUN(AC_C___ATTRIBUTE__, [
> AC_MSG_CHECKING(for __attribute__)
> AC_CACHE_VAL(ac_cv___attribute__, [
> AC_TRY_COMPILE([
> #include <stdlib.h>
> ],
> [
> static void foo(void) __attribute__ ((noreturn));
>
> static void
> foo(void)
> {
>  exit(1);
> }
> ],
> ac_cv___attribute__=yes,
> ac_cv___attribute__=no)])
> if test "$ac_cv___attribute__" = "yes"; then
>  AC_DEFINE(HAVE___ATTRIBUTE__, 1, [define if your compiler has
>  __attribute__])
> fi
> AC_MSG_RESULT($ac_cv___attribute__)
> ])
>
> If you, or someone else with a system that can reproduce the problem,
> can figure out what is wrong, I'd be happy to apply the patch. Maybe
> you could check your config.log to determine why this test seems to be
> failing.
>
> Cheers,
> Fyodor

This is the relevant part of nbase/config.log file on FC4:

configure:3528: checking for __attribute__
configure:3560: gcc -c -g -O2 -Wall   conftest.c >&5
conftest.c: In function 'main':
conftest.c:52: error: invalid storage class for function 'foo'
conftest.c:56: error: invalid storage class for function 'foo'
configure:3566: $? = 1
configure: failed program was:
| /* confdefs.h.  */
|
| #define PACKAGE_NAME ""
| #define PACKAGE_TARNAME ""
| #define PACKAGE_VERSION ""
| #define PACKAGE_STRING ""
| #define PACKAGE_BUGREPORT ""
| #define LINUX 1
| #define PCAP_TIMEOUT_IGNORED 1
| #define STDC_HEADERS 1
| #define HAVE_SYS_TYPES_H 1
| #define HAVE_SYS_STAT_H 1
| #define HAVE_STDLIB_H 1
| #define HAVE_STRING_H 1
| #define HAVE_MEMORY_H 1
| #define HAVE_STRINGS_H 1
| #define HAVE_INTTYPES_H 1
| #define HAVE_STDINT_H 1
| #define HAVE_UNISTD_H 1
| #define HAVE_STRING_H 1
| #define HAVE_GETOPT_H 1
| #define HAVE_STRINGS_H 1
| #define HAVE_MEMORY_H 1
| #define HAVE_SYS_PARAM_H 1
| #define HAVE_SYS_TIME_H 1
| #define HAVE_UNISTD_H 1
| #define HAVE_ERRNO_H 1
| #define HAVE_SYS_TYPES_H 1
| #define HAVE_SYS_SOCKET_H 1
| #define HAVE_NETINET_IN_H 1
| #define HAVE_RPC_TYPES_H 1
| #define HAVE_NETDB_H 1
| #define HAVE_PWD_H 1
| #define HAVE_ARPA_INET_H 1
| #define HAVE_SYS_RESOURCE_H 1
| #define HAVE_SYS_STAT_H 1
| #define HAVE_SYS_WAIT_H 1
| #define HAVE_NET_IF_H 1
| #define HAVE_LIBIBERTY_H 1
| #define HAVE_FCNTL_H 1
| #define HAVE_NET_IF_H 1
| #define HAVE_NETINET_IF_ETHER_H 1
| #define TIME_WITH_SYS_TIME 1
| /* end confdefs.h.  */
|
| #include <stdlib.h>
|
| int
| main ()
| {
|
| static void foo(void) __attribute__ ((noreturn));
|
| static void
| foo(void)
| {
|   exit(1);
| }
|
|   ;
|   return 0;
| }


Trying to compile the above C program with FC4's default gcc (4.0.2):

$ gcc -c -O2 -Wall test2.c
test2.c: In function 'main':
test2.c:7: error: invalid storage class for function 'foo'
test2.c:11: error: invalid storage class for function 'foo'

But with the old gcc on the same FC4 it works good:

$ gcc32 -c -O2 -Wall test2.c
$ gcc32 --version
gcc32 (GCC) 3.2.3 20030502 (Red Hat Linux 3.2.3-47.fc4)

and the attribute test pass Ok.

I think the fastest way to fix the problem, is doing the
AC_C___ATTRIBUTE__ test by ourself, modifying nbase/configure.ac to move
the foo() declaration outside the main() body. A patch for this is
attached. I have tested successfully on FC4 with both gcc 4.x and gcc 3.x.


Cheers,
Duilio Protti.

Attachment: attribute-test.patch
Description:

_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev