Re: [PATCH 0/4] Use generic_file_read_iter() in hugetlbfs

[Matthew Wilcox <willy@infradead.org>]

On Wed, Jul 08, 2026 at 04:31:48PM -0700, jane.chu@oracle.com wrote:
> On 7/7/2026 11:17 AM, Matthew Wilcox (Oracle) wrote:
> > This set of patches are separated out from Jane's earlier series.
> > https://lore.kernel.org/linux-mm/20260617172534.1740152-1-jane.chu@oracle.com/
> > I feel that they represent a reasonable amount of change to send
> > separately, although I'm going to keep looking at the rest of the sseries
> > because I actually need it to do a further improvement that I'm working on.
> > 
> > I'm not 100% confident in all of this, so I'd appreciate feedback.
> > I've run the libhugetlbfs test suite against it and observe no new
> > failures, but I'm not sure it's thoroughly exercising the read() path,
> > and I'm absolutely certain it's not exercising the memory-failure paths.
> 
> I can help testing memory-failure once I finish the stuff at hand.

Thanks!  Sashiko had a lot to say, so I'm working on a v2 with a lot
of changes.  I'd hold off on testing for now.  One last big one to go,
which is that we need to support holes in hugetlb files reading as zeroes
rather than allocating folios.  I have some ideas, but nothing I like
yet.