Re: [AVFS] avfs-fuse compilation bug with amd64

"Peter" <[email protected]> Fri, 2 Jun 2006 15:30:30 +0000 (UTC)
Newsgroups gmane.comp.file-systems.avfs.user
Message-ID <[email protected]>
On Mon, 22 May 2006 13:47:47 +0200, Ralf Hoffmann wrote:

> Hi,
> 
> On 2006-05-21 22:23, Ralf Hoffmann wrote:
>> Looks like the configure test LARGEFILE doesn't add this compile flag. I
>> have access to an AMD64 machine and I will try to find the problem. I
>> didn't check the URL from above but is the config.log for the failed
>> compilation available?
> 
> The LARGEFILE test says FILE_OFFSET_BITS is not required for large
> file support which makes sense on a 64 bit architecture. But the fuse
> include file "fuse_common.h" consists on this definition.
> 
> Miklos, what's your opinion about this? All we can do in avfs is to
> define this flag whether it is required or not (at least for fuse
> support) but I'm unsure about possible side effects.
> 
> Best Regards,
> 
> Ralf Hoffmann

Ralf, I took a peek at configure.in from cvs, line 130

if test "x$ac_cv_sys_file_offset_bits" != x -a "$ac_cv_sys_file_offset_bits" != no; then
  CPPFLAGS="$CPPFLAGS -D_FILE_OFFSET_BITS=$ac_cv_sys_file_offset_bits"
fi
if test "x$ac_cv_sys_large_files" != x -a "$ac_cv_sys_large_files" != no; then
  CPPFLAGS="$CPPFLAGS -D_LARGE_FILES=$ac_cv_sys_large_files"
fi


Seems to me if we rework the first test to accommodate an AMD64 which may
have the bits set already, we can force _FILE_OFFSET_BITS=64 even though
it is redundant. The alternative is to have packagers (me) hack the build
script. I would rather not :)


-- 
Peter
+++++
Do not reply to this email, it is a spam trap and not monitored.
Please reply to pete4abw at comcast dot net