Remember back on Apr 4 when you checked in this:
revision 1.3
date: 2006/04/04 15:04:25; author: slang; state: Exp; lines: +46 -2
commit of murali's 64bit fixes.
It, in part, removes -Werror from these kernel configure tests:
titan$ cvs diff -r1.2 -r1.3 maint/config/kernel.m4
Index: maint/config/kernel.m4
[..]
AC_MSG_CHECKING(for second arg type int in address_space_operations
releasepage)
tmp_cflags=$CFLAGS
- CFLAGS="$CFLAGS -Werror"
+ CFLAGS="$CFLAGS"
AC_TRY_COMPILE([
#define __KERNEL__
#include <linux/fs.h>
@@ -290,7 +303,7 @@ AC_DEFUN([AX_KERNEL_FEATURES],
AC_MSG_CHECKING(for int return in inode_operations follow_link)
tmp_cflags=$CFLAGS
- CFLAGS="$CFLAGS -Werror"
+ CFLAGS="$CFLAGS"
AC_TRY_COMPILE([
#define __KERNEL__
#include <linux/fs.h>
@@ -305,6 +318,37 @@ AC_DEFUN([AX_KERNEL_FEATURES],
AC_MSG_RESULT(no)
)
CFLAGS=$tmp_cflags
I can't figure out how to get those tests to work with out the
use of -Werror. What we're trying to do here is to differentiate
between two possible function prototypes in struct
address_space_operations:
int (*releasepage)(struct page *, int);
and
int (*releasepage)(struct page *, gfp_t);
The current test looks like:
AC_TRY_COMPILE([
#define __KERNEL__
#include <linux/fs.h>
extern int rp(struct page *page, int foo);
], [
struct address_space_operations aso = {
.releasepage = rp
};
],
[..]
but I would be happy for anything that works.
Can you (or Murali) describe what was the problem that required
removing -Werror? I'll keep trying to think of ways to do the test
without -Werror. As things stand, though, configure generates the
wrong value for x86_64 2.6.16, and warnings pop up at compile time.
-- Pete
_______________________________________________
Pvfs2-developers mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers
