Re: [PATCH v3] fs/sysv: Null check to prevent null-ptr-deref bug

[Christian Brauner]

On Tue, May 30, 2023 at 07:59:16AM -0700, Prince Kumar Maurya wrote:
> On Tue, May 30, 2023 at 1:26 AM Christian Brauner <brauner@xxxxxxxxxx> wrote:
> >
> > On Sun, May 28, 2023 at 11:44:22AM -0700, Prince Kumar Maurya wrote:
> > > sb_getblk(inode->i_sb, parent) return a null ptr and taking lock on
> > > that leads to the null-ptr-deref bug.
> > >
> > > Reported-by: syzbot+aad58150cbc64ba41bdc@xxxxxxxxxxxxxxxxxxxxxxxxx
> > > Closes: https://syzkaller.appspot.com/bug?extid=aad58150cbc64ba41bdc
> > > Signed-off-by: Prince Kumar Maurya <princekumarmaurya06@xxxxxxxxx>
> > > ---
> > > Change since v2: Updated subject and added Reported-by and closes tags.
> > >
> > >  fs/sysv/itree.c | 2 ++
> > >  1 file changed, 2 insertions(+)
> > >
> > > diff --git a/fs/sysv/itree.c b/fs/sysv/itree.c
> > > index b22764fe669c..3a6b66e719fd 100644
> > > --- a/fs/sysv/itree.c
> > > +++ b/fs/sysv/itree.c
> > > @@ -145,6 +145,8 @@ static int alloc_branch(struct inode *inode,
> > >                */
> > >               parent = block_to_cpu(SYSV_SB(inode->i_sb), branch[n-1].key);
> > >               bh = sb_getblk(inode->i_sb, parent);
> > > +             if (!bh)
> > > +                     break;
> >
> > When you break here you'll hit:
> >
> > /* Allocation failed, free what we already allocated */
> > for (i = 1; i < n; i++)
> >         bforget(branch[i].bh);
> > for (i = 0; i < n; i++)
> >         sysv_free_block(inode->i_sb, branch[i].key);
> >
> > below. The cleanup paths were coded in the assumption that sb_getblk()
> > can't fail. So bforget() can assume that branch[i].bh has been allocated
> > and set up. So that bforget(branch[i].bh) is your next pending NULL
> > deref afaict.
> 
> 
> I doubt that would happen. There is a break above as well, before we do
> sb_getblk().
> 
> /* Allocate the next block */
> branch[n].key = sysv_new_block(inode->i_sb);
> if (!branch[n].key)
>    break;
> 
> The clean up code path runs till i is less than n not equal to n which
> would have caused the problem.

But then aren't you leaking branch[n].key if you break after failed sb_getblk()
after sysv_new_block() succeeded?