On Sat, May 19, 2018 at 10:55:04PM +0900, Tetsuo Handa wrote: > Ernesto A. Fernandez wrote: > > On Tue, May 15, 2018 at 07:08:24PM +0900, Tetsuo Handa wrote: > > > From f78a5fe168290cb9e009f4d907d04b5bfe277831 Mon Sep 17 00:00:00 2001 > > > From: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx> > > > Date: Tue, 15 May 2018 11:38:38 +0900 > > > Subject: [PATCH] hfsplus: don't return 0 when fill_super() failed > > > > > > syzbot is reporting NULL pointer dereference at mount_fs() [1]. > > > This is because hfsplus_fill_super() is by error returning 0 when > > > hfsplus_fill_super() detected invalid filesystem image, and mount_bdev() > > > is returning NULL because dget(s->s_root) == NULL if s->s_root == NULL, > > > and mount_fs() is accessing root->d_sb because IS_ERR(root) == false > > > if root == NULL. Fix this by returning -EINVAL when hfsplus_fill_super() > > > detected invalid filesystem image. > > > > > > [1] https://syzkaller.appspot.com/bug?id=21acb6850cecbc960c927229e597158cf35f33d0 > > > > > > Signed-off-by: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx> > > > Reported-by: syzbot <syzbot+01ffaf5d9568dd1609f7@xxxxxxxxxxxxxxxxxxxxxxxxx> > > > Cc: Al Viro <viro@xxxxxxxxxxxxxxxxxx> > > > --- > > > fs/hfsplus/super.c | 4 +++- > > > 1 file changed, 3 insertions(+), 1 deletion(-) > > > > > > diff --git a/fs/hfsplus/super.c b/fs/hfsplus/super.c > > > index 513c357..9e690ae 100644 > > > --- a/fs/hfsplus/super.c > > > +++ b/fs/hfsplus/super.c > > > @@ -524,8 +524,10 @@ static int hfsplus_fill_super(struct super_block *sb, void *data, int silent) > > > goto out_put_root; > > > if (!hfs_brec_read(&fd, &entry, sizeof(entry))) { > > > hfs_find_exit(&fd); > > > - if (entry.type != cpu_to_be16(HFSPLUS_FOLDER)) > > > + if (entry.type != cpu_to_be16(HFSPLUS_FOLDER)) { > > > + err = -EINVAL; > > > goto out_put_root; > > > + } > > > inode = hfsplus_iget(sb, be32_to_cpu(entry.folder.id)); > > > if (IS_ERR(inode)) { > > > err = PTR_ERR(inode); > > > -- > > > 1.8.3.1 > > > > > > > > > > I sent this same patch some time ago: > > > > https://www.spinics.net/lists/linux-fsdevel/msg125241.html > > > > This syzbot report did not reach me, though. I don't know why. > > > Oh, I didn't notice that you already wrote that patch. > Anyway, whom to ask for merge? Al Viro? Andrew Morton? I think Andrew Morton, but only one of my patches for hfsplus has been picked up so far, so I can't tell you for sure. Al Viro wasn't happy with this patchset and thought it was better to let ->put_super() handle the cleanup. I made a second version but I don't think he had the time to look at it yet.