On Fri, 2025-02-14 at 11:17 +0100, Miklos Szeredi wrote:
> On Tue, 4 Feb 2025 at 16:04, Jeff Layton <jlayton@xxxxxxxxxx> wrote:
> >
> > I see no reason to set the private_data on the file to this value. Just
> > grab the result of the atomic_read() and output it without setting
> > private_data.
> >
> > Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
> > ---
> > fs/fuse/control.c | 5 ++---
> > 1 file changed, 2 insertions(+), 3 deletions(-)
> >
> > diff --git a/fs/fuse/control.c b/fs/fuse/control.c
> > index 2a730d88cc3bdb50ea1f8a3185faad5f05fc6e74..17ef07cf0c38e44bd7eadb3450bd53a8acc5e885 100644
> > --- a/fs/fuse/control.c
> > +++ b/fs/fuse/control.c
> > @@ -49,18 +49,17 @@ static ssize_t fuse_conn_waiting_read(struct file *file, char __user *buf,
> > {
> > char tmp[32];
> > size_t size;
> > + int value;
> >
> > if (!*ppos) {
> > - long value;
> > struct fuse_conn *fc = fuse_ctl_file_conn_get(file);
> > if (!fc)
> > return 0;
> >
> > value = atomic_read(&fc->num_waiting);
> > - file->private_data = (void *)value;
> > fuse_conn_put(fc);
> > }
>
> "value" is uninitialized if *ppos is non-zero.
>
Good catch. Thanks. Do you want me to resend?
> I also wonder why this patch is an improvement (with the bug fixed)?
>
It's just an unnecessary assignment. Nothing will look at or use
private_data in this codepath, so there is no need to set it.
--
Jeff Layton <jlayton@xxxxxxxxxx>
