Re: [build] Move ENABLE_EXECUTE_STACK to toplevel libgcc

[Paolo Bonzini <bonzini@gnu.org>]

On 05/30/2011 05:59 PM, Rainer Orth wrote:
> This is my hopefully last patch for toplevel libgcc moves: it moves
> ENABLE_EXECUTE_STACK to $target-lib.h headers in libgcc/config.  Since
> gcc/config/sol2.h is only used on Solaris targets anymore and Solaris 8
> is the minimal supported version, I've removed both a Solaris 2.6
> workaround and include<sys/mman.h>  directly.  Same thing in osf5-lib.h.
>
> Since the existance of the macro is used in alpha/alpha.c, i386/i386.c,
> and sparc/sparc.c to enable calls to __enable_execute_stack, I had to
> define HAVE_ENABLE_EXECUTE_STACK in the gcc/config headers to convey the
> necessary information.
>
> The new libgcc/config/$target-lib.h headers are added to libgcc_tm_file
> in gcc/config.gcc.  I'd rather add them to libgcc/config.host instead so
> the information is kept local to libgcc.

Did you have any problems doing so?

Long term, it would be even better to do something like this in 
libgcc/config.host:

enable_execute_stack=enable-execute-stack-empty.c
case $host in
   ...) enable_execute_stack=config/enable-execute-stack-mprotect.c ;;
   ...
esac

in libgcc/configure.ac:

AC_CONFIG_LINKS(enable-execute-stack.c:$enable_execute_stack)

in libgcc/Makefile.in:

LIB2ADD += enable-execute-stack.c

and drop this hunk altogether from gcc/libgcc2.c:

#ifdef L_enable_execute_stack
/* Attempt to turn on execute permission for the stack.  */

#ifdef ENABLE_EXECUTE_STACK
   ENABLE_EXECUTE_STACK
#else
void
__enable_execute_stack (void *addr __attribute__((__unused__)))
{}
#endif /* ENABLE_EXECUTE_STACK */

#endif /* L_enable_execute_stack */


> Bootstrapped without regressions on i386-pc-solaris2.11 and
> sparc-sun-solaris2.11.
>
> I've Cc'ed the OS port maintainers of the other affected targets and
> would appreciate testing/review.  An OpenBSD maintainer isn't listed,
> unfortunately.  Also included are the CPU port maintainers for the
> modified backends.
>
> Ok for mainline after a week if no problems occur in testing on the
> other targets?

It's a good start, but at least you need changes to the documentation; 
if you can make the above work, that would be great as an example of how 
to move stuff to the libgcc toplevel directory.

Paolo