Re: [PATCH RFC 38/41] fuse: refuse files on pseudo filesystems as passthrough backing files

[Amir Goldstein]

On Fri, Nov 13, 2026 at 11:25:57AM +0100, Alexander Mikhalitsyn wrote:
> Hi Christoph,
> 
> On Wed, 2026-11-11 at 16:12 +0100, Christoph Hellwig wrote:
> > I've seen setups where people use fuse to mediate access to sysfs
> > sysfs in odd ways.  No idea if that's still around, but if so, this
> > patch will break it.
> 
> 
> Oh, that's painful. Do they use backing files for that?
> It's a relatively new feature in FUSE...
> 

Very unlikely that they use FUSE passthrough and if they did
very unlikely that it would break their filesystem.

The API for fuse passthrough is that the server requests to setup a
backing file before open via ioctl and if ioctl fails (as this patch
does) then server falls back to non-passthrough open.

The ioctl can already fail for many reasons such as MAX_STACK_DEPTH on
procfs as the commit message says, so server needs to cope with the
fall back.

Thanks,
Amir.