Re: ia32 userland and XFS

Re: ia32 userland and XFS

[Aníbal Monsalve Salazar]

[-- Attachment #1.1: Type: text/plain, Size: 482 bytes --]

Adding xfs@oss.sgi.com to the cc list so all the XFS folk see this.

On Mon, Dec 01, 2008 at 07:37:16PM -0600, Ron Johnson wrote:
>https://alioth.debian.org/docman/view.php/30192/21/debian-amd64-howto.html#id292806
>
>According to this (seemingly 2+ year old) web page, the XFS file system
>chokes on the combination of 32 bit userland and 64 bit kernel.
>
>Is this still true, and why should a low-level driver hidden under a
>virtual fs care what user apps access it via the vfs?

[-- Attachment #1.2: Digital signature --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

[-- Attachment #2: Type: text/plain, Size: 121 bytes --]

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: ia32 userland and XFS

[Christoph Hellwig]

On Tue, Dec 02, 2008 at 08:33:50PM +1100, An?bal Monsalve Salazar wrote:
> On Mon, Dec 01, 2008 at 07:37:16PM -0600, Ron Johnson wrote:
> >According to this (seemingly 2+ year old) web page, the XFS file system
> >chokes on the combination of 32 bit userland and 64 bit kernel.
> >
> >Is this still true, and why should a low-level driver hidden under a
> >virtual fs care what user apps access it via the vfs?

XFS as in the plain posix filesystem works perfectly fine with a 64 bit
kernel and 32 bit userspace.

But various advance capabilities or administration interfaces which are
used by tools from xfsprogs are implemented as ioctls, and unfortunately
most of them have been designed very badly and aren't wordsize clean.

There have been handlers for a few of them for a while, but only as of
today a full set of compat handlers has been commited.  That code will be
release with 2.6.29, but could also be backported.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs