This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] mips: Allow larger values for __aligned__
On Fri, Oct 13, 2006 at 09:41:53PM +0200, Eric Botcazou wrote:
> > You are compiling with -Werror, is that normal? I don't compile with
> > -Werror for testing.
> -Werror is automatically enabled during a 3-stage bootstrap on mainline.
> > The elfos.h header should always be included first.
> Where is that written?
Sorry, I wasn't being clear enough. The order should be that elfos.h is
included before the target headers redefine the ELF related defaults, so
the guard was superfluous.
What do you think of the following convservative approach to correct the
(650) 331-3385 x716
--- gcc/ChangeLog (revision 117705)
+++ gcc/ChangeLog (working copy)
@@ -1,3 +1,8 @@
+2006-10-13 Carlos O'Donell <email@example.com>
+ * config/elfos.h (MAX_OFILE_ALIGNMENT): Default to largest
+ alignment representable by a 32-bit system.
2006-10-13 Richard Guenther <firstname.lastname@example.org>
--- gcc/config/elfos.h (revision 117705)
+++ gcc/config/elfos.h (working copy)
@@ -44,14 +44,10 @@ Boston, MA 02110-1301, USA. */
section alignment up to (0x80000000 * 8), while 64-bit ELF
supports (0x8000000000000000 * 8). If this macro is not
defined, the default is the largest alignment supported by
- 32-bit ELF. Use this macro to limit the alignment which can
- be specified using the `__attribute__ ((aligned (N)))'
- construct. */
-#if (HOST_BITS_PER_WIDEST_INT >= 64)
-#define MAX_OFILE_ALIGNMENT (((unsigned HOST_WIDEST_INT) 1 << 31) * 8)
+ 32-bit ELF on a 32-bit target. Use this macro to limit the
+ alignment which can be specified using the
+ `__attribute__ ((aligned (N)))' construct. */
#define MAX_OFILE_ALIGNMENT (((unsigned HOST_WIDEST_INT) 1 << 28) * 8)
/* Use periods rather than dollar signs in special g++ assembler names. */