On Fri, Feb 01, 2019 at 08:43:59AM +0100, Christoph Hellwig wrote:
> On Mon, Jan 28, 2019 at 12:18:05PM -0800, Matthew Wilcox wrote:
> > On Mon, Jan 28, 2019 at 08:31:19PM +0800, zhengbin wrote:
> > > The 'end_byte' parameter of filemap_range_has_page is required to be
> > > inclusive, so follow the rule.
> >
> > Reviewed-by: Matthew Wilcox <willy@xxxxxxxxxxxxx>
> > Fixes: 6be96d3ad34a ("fs: return if direct I/O will trigger writeback")
> >
> > Adding the people in the sign-off chain to the Cc.
>
> This looks correct to me:
>
> Acked-by: Christoph Hellwig <hch@xxxxxx>
>
> I wish we'd kill these stupid range calling conventions, though -
> offset + len is a lot more intuitive, and we already use it very
> widely all over the kernel.
It has its own problems though; you have to check that offset + len -
1 doesn't wrap past zero. Really, it's the transition from (offset,
len) to (min, max) that needs to be avoided as much as possible within
a subsystem.
