On Mon, Jul 13, 2026 at 1:20 AM Baokun Li <[email protected]> wrote:
>
> Hi Xiang,
>
> On 2026/7/10 05:11, Xiang Mei wrote:
> > The fuse-io-uring transport copies req->in.h out to the ring in
> > fuse_uring_copy_to_ring() and req->out.h back in fuse_uring_commit().
> > Both headers live inside the fuse_request slab object, whose cache
> > (fuse_req_cachep) is created without a usercopy whitelist,
>
>
> Then why not allocate "fuse_request" with kmem_cache_create_usercopy()
> to add a usercopy whitelist instead?
>
> That would avoid the extra stack usage for the bounce headers and
> the 56 bytes of copying they incur.
>
>
> Thanks,
> Baokun
You are right. I have dropped previous reviewers' tags and sent a v4
since it's a hot path:
https://lore.kernel.org/fuse-devel/[email protected]/T/#t
Xiang
>
> > so copying
> > them directly to/from userspace trips CONFIG_HARDENED_USERCOPY and
> > panics:
> >
> > usercopy: Kernel memory exposure attempt detected from SLUB object
> > 'fuse_request' (offset 56, size 40)!
> > kernel BUG at mm/usercopy.c:102!
> > RIP: 0010:usercopy_abort+0x6c/0x80
> > Call Trace:
> > __check_heap_object
> > __check_object_size
> > copy_header_to_ring fs/fuse/dev_uring.c:618
> > fuse_uring_prepare_send
> > fuse_uring_send_in_task
> > ...
> > __do_sys_io_uring_enter
> > entry_SYSCALL_64_after_hwframe
> >
> > Bounce both headers through an on-stack copy so the usercopy touches
> > stack memory, not the slab object.
> >
> > Fixes: c090c8abae4b ("fuse: Add io-uring sqe commit and fetch support")
> > Cc: [email protected]
> > Reported-by: Weiming Shi <[email protected]>
> > Assisted-by: Claude:claude-opus-4-8
> > Signed-off-by: Xiang Mei <[email protected]>
> > Reviewed-by: Bernd Schubert <[email protected]>
> > Reviewed-by: Joanne Koong <[email protected]>
> > ---
> > v3: no context change; add Bernd's Reviewed-by
> >
> > fs/fuse/dev_uring.c | 12 ++++++++----
> > 1 file changed, 8 insertions(+), 4 deletions(-)
> >
> > diff --git a/fs/fuse/dev_uring.c b/fs/fuse/dev_uring.c
> > index 77c8cec43d9c..0814681eb04b 100644
> > --- a/fs/fuse/dev_uring.c
> > +++ b/fs/fuse/dev_uring.c
> > @@ -744,6 +744,7 @@ static int fuse_uring_copy_to_ring(struct fuse_ring_ent
> > *ent,
> > {
> > struct fuse_ring_queue *queue = ent->queue;
> > struct fuse_ring *ring = queue->ring;
> > + struct fuse_in_header in_header;
> > int err;
> >
> > err = -EIO;
> > @@ -765,8 +766,9 @@ static int fuse_uring_copy_to_ring(struct fuse_ring_ent
> > *ent,
> > }
> >
> > /* copy fuse_in_header */
> > - return copy_header_to_ring(ent, FUSE_URING_HEADER_IN_OUT, &req->in.h,
> > - sizeof(req->in.h));
> > + in_header = req->in.h;
> > + return copy_header_to_ring(ent, FUSE_URING_HEADER_IN_OUT, &in_header,
> > + sizeof(in_header));
> > }
> >
> > static int fuse_uring_prepare_send(struct fuse_ring_ent *ent,
> > @@ -871,11 +873,13 @@ static void fuse_uring_commit(struct fuse_ring_ent
> > *ent, struct fuse_req *req,
> > unsigned int issue_flags)
> > {
> > struct fuse_ring *ring = ent->queue->ring;
> > + struct fuse_out_header out_header;
> > ssize_t err = -EFAULT;
> >
> > - if (copy_header_from_ring(ent, FUSE_URING_HEADER_IN_OUT, &req->out.h,
> > - sizeof(req->out.h)))
> > + if (copy_header_from_ring(ent, FUSE_URING_HEADER_IN_OUT, &out_header,
> > + sizeof(out_header)))
> > goto out;
> > + req->out.h = out_header;
> >
> > err = fuse_uring_out_header_has_err(&req->out.h, req);
> > if (err) {
>
>
