Re: [PATCH v1 2/2] fs/writeback: skip inodes with potential writeback hang in wait_sb_inodes()

[Joanne Koong <joannelkoong@gmail.com>]

On Wed, Dec 3, 2025 at 1:28 AM Miklos Szeredi <miklos@szeredi.hu> wrote:
>
> On Wed, 26 Nov 2025 at 18:58, Joanne Koong <joannelkoong@gmail.com> wrote:
> >
> > On Wed, Nov 26, 2025 at 2:55 AM David Hildenbrand (Red Hat)
> > <david@kernel.org> wrote:
> > > >
> > > >> having a flag that states something like that that
> > > >> "AS_NO_WRITEBACK_WAIT_ON_DATA_SYNC" would probable be what we would want
> > > >> to add to avoid waiting for writeback with clear semantics why it is ok
> > > >> in that specific scenario.
> > > >
> > > > Having a separate AS_NO_WRITEBACK_WAIT_ON_DATA_SYNC mapping flag
> > > > sounds reasonable to me and I agree is more clearer semantically.
> > >
> > > Good. Then it's clear that we are not waiting because writeback is
> > > shaky, but because even if it would be working, because we don't have to
> > > because there are no such guarantees.
> > >
> > > Maybe
> > >
> > > AS_NO_DATA_INTEGRITY
> > >
> > > or similar would be cleaner, I'll have to leave that to you and Miklos
> > > to decide what exactly the semantics are that fuse currently doesn't
> > > provide.
> >
> > After reading Miklos's reply, I must have misunderstood this then - my
> > understanding was that the reason we couldn't guarantee data integrity
> > in fuse was because of the temp pages design where checking the
> > writeback flag on the real folio doesn't reflect writeback state, but
> > that removing the temp pages and using the real folio now does
> > guarantee this. But it seems like it's not as simple as that and
> > there's no data integrity guarantees for other reasons.
> >
> > Changing this to AS_NO_DATA_INTEGRITY sounds good to me, if that
> > sounds good to Miklos as well. Or do you have another preference,
> > Miklos?
>
> Sure, sounds good.
>
> (Sorry about the delay, missed this.)

Is the reason we can't guarantee data integrity because the server
ultimately controls disk persistence whereby it can claim to have
written data even if it didn't? If so, it seems like NFS / the other
network-based filesystems would also have to set that
AS_NO_DATA_INTEGRITY mapping flag too to be consistent? Or is there
some other reason?

Thanks,
Joanne


>
> Thanks,
> Miklos