Re: [PATCH v11 12/21] vfs: Remove get_xip_mem

[Matthew Wilcox <willy@linux.intel.com>]

On Thu, Oct 16, 2014 at 02:14:46PM +0200, Mathieu Desnoyers wrote:
> > +++ b/fs/open.c
> > @@ -655,11 +655,8 @@ int open_check_o_direct(struct file *f)
> >  {
> >  	/* NB: we're sure to have correct a_ops only after f_op->open */
> >  	if (f->f_flags & O_DIRECT) {
> > -		if (!f->f_mapping->a_ops ||
> > -		    ((!f->f_mapping->a_ops->direct_IO) &&
> > -		    (!f->f_mapping->a_ops->get_xip_mem))) {
> > +		if (!f->f_mapping->a_ops || !f->f_mapping->a_ops->direct_IO)
> 
> Why is it OK to remove the check for get_xip_mem callback here, rather
> than replacing it with a IS_DAX check like the rest of this patch does ?
> I'm probably missing something.

XIP used to intercept I/Os by having the filesystem's ->read & ->write
methods call xip_file_read (/write).  That would do the I/O, and so there
was no need to have a ->direct_IO element in a_ops.  For DAX, we use the
generic VFS code to call back into the filesystem's ->direct_IO entry
point, so the check above for ->direct_IO now checks for both regular
and DAX support.

Or to put it another way, DAX now requires that the filesystem support
O_DIRECT.  Which is pretty much the way it has to be anyway, since DAX
is direct!