Re: [PATCH 4/8] fuse: signal that a fuse filesystem should exhibit local fs behaviors

["Darrick J. Wong" <djwong@kernel.org>]

On Wed, Sep 17, 2025 at 10:18:40AM -0700, Joanne Koong wrote:
> On Mon, Sep 15, 2025 at 5:25 PM Darrick J. Wong <djwong@kernel.org> wrote:
> >
> > From: Darrick J. Wong <djwong@kernel.org>
> >
> > Create a new fuse context flag that indicates that the kernel should
> > implement various local filesystem behaviors instead of passing vfs
> > commands straight through to the fuse server and expecting the server to
> > do all the work.  For example, this means that we'll use the kernel to
> > transform some ACL updates into mode changes, and later to do
> > enforcement of the immutable and append iflags.
> >
> > Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
> > ---
> >  fs/fuse/fuse_i.h |    4 ++++
> >  fs/fuse/inode.c  |    2 ++
> >  2 files changed, 6 insertions(+)
> >
> > diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
> > index e93a3c3f11d901..e13e8270f4f58d 100644
> > --- a/fs/fuse/fuse_i.h
> > +++ b/fs/fuse/fuse_i.h
> > @@ -603,6 +603,7 @@ struct fuse_fs_context {
> >         bool no_control:1;
> >         bool no_force_umount:1;
> >         bool legacy_opts_show:1;
> > +       bool local_fs:1;
> >         enum fuse_dax_mode dax_mode;
> >         unsigned int max_read;
> >         unsigned int blksize;
> > @@ -901,6 +902,9 @@ struct fuse_conn {
> >         /* Is link not implemented by fs? */
> >         unsigned int no_link:1;
> >
> > +       /* Should this filesystem behave like a local filesystem? */
> > +       unsigned int local_fs:1;
> > +
> >         /* Use io_uring for communication */
> >         unsigned int io_uring;
> >
> > diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
> > index c94aba627a6f11..c8dd0bcb7e6f9f 100644
> > --- a/fs/fuse/inode.c
> > +++ b/fs/fuse/inode.c
> > @@ -1862,6 +1862,7 @@ int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx)
> >         fc->destroy = ctx->destroy;
> >         fc->no_control = ctx->no_control;
> >         fc->no_force_umount = ctx->no_force_umount;
> > +       fc->local_fs = ctx->local_fs;
> >
> 
> If I'm understanding it correctly, fc->local_fs is set to true if it's
> a fuseblk device? Why do we need a new "ctx->local_fs" instead of
> reusing ctx->is_bdev?

Eventually, enabling iomap will also set local_fs=1, as Miklos and I
sort of touched on a couple weeks ago:

https://lore.kernel.org/linux-fsdevel/CAJfpegvmXnZc=nC4UGw5Gya2cAr-kR0s=WNecnMhdTM_mGyuUg@mail.gmail.com/

--D

> Thanks,
> Joanne
> 
> >         err = -ENOMEM;
> >         root = fuse_get_root_inode(sb, ctx->rootmode);
> > @@ -2029,6 +2030,7 @@ static int fuse_init_fs_context(struct fs_context *fsc)
> >         if (fsc->fs_type == &fuseblk_fs_type) {
> >                 ctx->is_bdev = true;
> >                 ctx->destroy = true;
> > +               ctx->local_fs = true;
> >         }
> >  #endif
> >
> >