Re: [PATCH v7 01/31] iov_iter: Add ITER_XARRAY

[Al Viro <viro@zeniv.linux.org.uk>]

On Sun, Apr 25, 2021 at 02:58:02PM +0100, David Howells wrote:

> But for the moment, I guess I should just add:
> 
> 	i->iov_offset += bytes;
> 
> to all three (kvec, bvec and xarray)?

No.  First of all, you'd need ->count updated as well; for kvec and bvec you
*REALLY* don't have to end up with ->iov_offset exceeding the size of current
kvec or bvec resp.; Bad Shit(tm) happens that way.

> 
> > > @@ -1246,7 +1349,8 @@ unsigned long iov_iter_alignment(const struct iov_iter *i)
> > >  	iterate_all_kinds(i, size, v,
> > >  		(res |= (unsigned long)v.iov_base | v.iov_len, 0),
> > >  		res |= v.bv_offset | v.bv_len,
> > > -		res |= (unsigned long)v.iov_base | v.iov_len
> > > +		res |= (unsigned long)v.iov_base | v.iov_len,
> > > +		res |= v.bv_offset | v.bv_len
> > >  	)
> > >  	return res;
> > >  }
> > 
> > Hmm...  That looks like a really bad overkill - do you need anything beyond
> > count and iov_offset in that case + perhaps "do we have the very last page"?
> > IOW, do you need to iterate anything at all here?  What am I missing here?
> 
> Good point.  I wonder, even, if the alignment could just be set to 1.  There's
> no kdoc description on the function that says what the result is meant to
> represent.

Huh?  It's the worst alignment of all segment boundaries, what else?  As in
if (iov_iter_alignment(i) & 1023)
	// we have something in there that isn't 1K-aligned.