This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH, i386]: Revert "Align DImode to 32 bits" patch.
- From: Eric Botcazou <ebotcazou at adacore dot com>
- To: Uros Bizjak <ubizjak at gmail dot com>
- Cc: gcc-patches at gcc dot gnu dot org, "H.J. Lu" <hjl dot tools at gmail dot com>
- Date: Tue, 25 Mar 2008 19:00:03 +0100
- Subject: Re: [PATCH, i386]: Revert "Align DImode to 32 bits" patch.
- References: <47E6480A.4010406@gmail.com>
> Due to above problems, I have revert DImode alignment patch, so DImode
> is now aligned back to its natural alignment (8bytes).
Supporting kludge in the Ada compiler reverted as well.
Tested on i586-suse-linux, applied on the mainline.
2008-03-25 Eric Botcazou <ebotcazou@adacore.com>
Revert
2008-03-05 Eric Botcazou <ebotcazou@adacore.com>
PR ada/35186
* decl.c (maybe_pad_type): Avoid padding an integral type when
bumping its alignment is sufficient.
--
Eric Botcazou
Index: decl.c
===================================================================
--- decl.c (revision 133506)
+++ decl.c (working copy)
@@ -5529,18 +5529,6 @@ maybe_pad_type (tree type, tree size, un
if (align == 0 && !size)
return type;
- /* If no size is specified and we have an integral type, and changing
- the alignment won't change its size, return a copy of the type
- with the specified alignment. */
- if (!size
- && INTEGRAL_TYPE_P (type)
- && value_factor_p (orig_size, align))
- {
- type = copy_type (type);
- TYPE_ALIGN (type) = align;
- return type;
- }
-
/* We used to modify the record in place in some cases, but that could
generate incorrect debugging information. So make a new record
type and name. */