Re: [PATCH] bfs: add sanity check at bfs_fill_super().

[Tigran Aivazian <aivazian.tigran@xxxxxxxxx>]

Thank you. Please see comments below:

On 14 June 2018 at 11:45, Tetsuo Handa
<penguin-kernel@xxxxxxxxxxxxxxxxxxx> wrote:
> (Which patch did you mean? Hmm, I paste both patches instead of waiting for your answer.)

I only meant the BFS patch.

> commit 0c54bbdb89992eeff50866b64366a1a0cbd0e6fb
> Author: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
> Date:   Sat May 26 14:15:41 2018 +1000
>
>     bfs: add sanity check at bfs_fill_super()
>
>     syzbot is reporting too large memory allocation at bfs_fill_super() [1].
>     Since file system image is corrupted such that bfs_sb->s_start == 0,
>     bfs_fill_super() is trying to allocate 8MB of continuous memory.  Fix this
>     by adding a sanity check on bfs_sb->s_start, __GFP_NOWARN and printf().
>
>     [1] https://syzkaller.appspot.com/bug?id=16a87c236b951351374a84c8a32f40edbc034e96
>
>     Link: http://lkml.kernel.org/r/1525862104-3407-1-git-send-email-penguin-kernel@xxxxxxxxxxxxxxxxxxx
>     Signed-off-by: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
>     Reported-by: syzbot <syzbot+71c6b5d68e91149fc8a4@xxxxxxxxxxxxxxxxxxxxxxxxx>
>     Reviewed-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
>     Cc: Tigran Aivazian <aivazian.tigran@xxxxxxxxx>
>     Cc: Matthew Wilcox <willy@xxxxxxxxxxxxx>
>     Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
>     Signed-off-by: Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx>
>
> diff --git a/fs/bfs/inode.c b/fs/bfs/inode.c
> index 9a69392..d81c148 100644
> --- a/fs/bfs/inode.c
> +++ b/fs/bfs/inode.c
> @@ -350,7 +350,8 @@ static int bfs_fill_super(struct super_block *s, void *data, int silent)
>
>         s->s_magic = BFS_MAGIC;
>
> -       if (le32_to_cpu(bfs_sb->s_start) > le32_to_cpu(bfs_sb->s_end)) {
> +       if (le32_to_cpu(bfs_sb->s_start) > le32_to_cpu(bfs_sb->s_end) ||
> +           le32_to_cpu(bfs_sb->s_start) < BFS_BSIZE) {
>                 printf("Superblock is corrupted\n");
>                 goto out1;
>         }

Yes, this is acceptable.

> @@ -359,9 +360,11 @@ static int bfs_fill_super(struct super_block *s, void *data, int silent)
>                                         sizeof(struct bfs_inode)
>                                         + BFS_ROOT_INO - 1;
>         imap_len = (info->si_lasti / 8) + 1;
> -       info->si_imap = kzalloc(imap_len, GFP_KERNEL);
> -       if (!info->si_imap)
> +       info->si_imap = kzalloc(imap_len, GFP_KERNEL | __GFP_NOWARN);
> +       if (!info->si_imap) {
> +               printf("Cannot allocate %u bytes\n", imap_len);
>                 goto out1;
> +       }

This is unnecessary. As Andrew Morton pointed out, if people set
panic_on_warn then they do really want a panic on allocation failure
warnings. Please update the patch with just the above sanity check for
s_start < BFS_BSIZE.

Kind regards,
Tigran