Re: [PATCH v1 0/2] fuse: fix missing barriers in io-uring init

On Tue, Jul 14, 2026 at 1:51 AM Joanne Koong <[email protected]> wrote:
>
> On Mon, Jul 13, 2026 at 3:11 PM Joanne Koong <[email protected]> wrote:
> >
> > On Mon, Jul 13, 2026 at 10:54 AM Joanne Koong <[email protected]> wrote:
> > >
> > > These are two pre-existing issues Sashiko reported [1] on the fuse zerocopy
> > > series.
> > >
> > > [1] https://sashiko.dev/#/patchset/20260630211436.2062816-1-joannelkoong%40gmail.com
> >
> > Sashiko noted some other places that are also missing barriers [1].
> > Will send v2 to add these places as well.
>
> These additional ones aren't real bugs. They're not reachable on a
> well-behaved server and on a malicious/buggy server, the failure is
> benign / self-correcting.
>
> I think we need some sort of way to help Sashiko understand so it
> doesn't keep flagging this on every future patch anyone submits. I
> think there's a few other places where some of Sashiko's
> assumptions/analysis are wrong (eg uring-cmd sqe stability semantics
> [1] in the zero-copy series). I'll look into this unless anyone
> already knows how to do this.

You may want to talk with your colleague Chris Mason about adding review
prompts for FUSE, because he is also running AI analysis on existing subsystems
regardless of mailing list patches.
I didn't try, but you can probably figure out how to add FUSE specific
review prompts
to Sashiko yourself...

Thanks,
Amir.