Re: [PATCH v2 01/11] ntfsplus: in-memory, on-disk structures and headers

[Namjae Jeon]

On Mon, Dec 1, 2025 at 5:47 PM Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote:
>
> On Sun, Nov 30, 2025 at 11:14:24PM -0800, Christoph Hellwig wrote:
> > > + * ntfs_read_mapping_folio - map a folio into accessible memory, reading it if necessary
> >
> > The very long comment for something that is just a trivial wrapper
> > around read_mapping_folio is odd.  Also why does ntrfs need the special
> > EINTR handling that other file systems don't?
>
> I would presume that this is because NTFS is using the page cache for
> metadata and they don't want the metadata read to be interrupted by a
> fatal signal.
Right.
> Of course, this turns into a spinning instead of sleeping
> wait, so very bad for CPU usage.
CPU intensive spinning only occurs if signals are delivered extremely
frequently...
Are there any ways to improve this EINTR handling?
Thanks!