On Tue, Mar 30, 2021 at 5:09 PM ronnie sahlberg
<ronniesahlberg@xxxxxxxxx> wrote:
>
> On Wed, Mar 31, 2021 at 12:22 AM Aurélien Aptel <aaptel@xxxxxxxx> wrote:
> >
> > Ronnie Sahlberg <lsahlber@xxxxxxxxxx> writes:
> > > +static long smb3_collapse_range(struct file *file, struct cifs_tcon *tcon,
> > > + loff_t off, loff_t len)
> > > +{
> > > + int rc;
> > > + unsigned int xid;
> > > + struct cifsFileInfo *cfile = file->private_data;
> > > + __le64 eof;
> > > +
> > > + xid = get_xid();
> > > +
> > > + if (off + len < off) {
> > > + rc = -EFBIG;
> > > + goto out;
> > > + }
> >
> > loff_t is defined as 'long long' for me which is signed, and signed
> > overflow is Undefined Behaviour, unless we compile with -fwrapv which
> > I'm not sure it is something we can assume.
> >
> > Also, vfs_fallocate() in fs/open.c already does an overflow check before
> > calling f_op->falloc(), this is probably not needed. (It's also relying
> > on signed overflow so I guess it is ok...?)
>
> Thanks.
> Steve, can you drop this check from the patch?
Yes. I was trying to run some xfstests on it vs. Windows with REFS
(and Samba with BTRFS after that). Any observations on these two
patches and xfstests?
> > Rest of the patch looks good otherwise.
> >
> > Cheers,
> > --
> > Aurélien Aptel / SUSE Labs Samba Team
--
Thanks,
Steve
