On Jun 29, 2021, at 8:36 AM, Ye Bin <yebin10@xxxxxxxxxx> wrote:
>
> Now sbi->s_mmp_tsk is created with kthread_run, then kmmpd maybe
> already running and even exit as exception. Even though we set
> sbi->s_mmp_tsk with NULL before kmmpd kthread exit, but
> "sbi->s_mmp_tsk=kthread_run(XX)" may set after set with NULL.
> mount kmmpd
> | |
> |-call kthread_run |
> | |-kmmpd runing
> | |-kmmpd exit sbi->s_mmp_tsk=NULL
> | |
> |-kthread_run return |
> | and set sbi->s_mmp_tsk |
> | |
> |-then we get wild ptr"sbi->s_mmp_tsk" and later trigger UAF
>
> This patch is base on previous "ext4: Fix use-after-free about sbi->s_mmp_tsk".
> Previous patch ensure kmmpd kthread exit by itself will set sbi->s_mmp_tsk with
> NULL. We can create kthread first, and then wakeup kmmpd kthread later.
>
> Signed-off-by: Ye Bin <yebin10@xxxxxxxxxx>
Please note minor typo in the patch subject, should be "use-after-free".
Reviewed-by: Andreas Dilger <adilger@xxxxxxxxx>
> ---
> fs/ext4/mmp.c | 10 ++++++----
> 1 file changed, 6 insertions(+), 4 deletions(-)
>
> diff --git a/fs/ext4/mmp.c b/fs/ext4/mmp.c
> index fc18a8c205c7..6ec1ea182cc0 100644
> --- a/fs/ext4/mmp.c
> +++ b/fs/ext4/mmp.c
> @@ -394,16 +394,18 @@ int ext4_multi_mount_protect(struct super_block *sb,
> /*
> * Start a kernel thread to update the MMP block periodically.
> */
> - EXT4_SB(sb)->s_mmp_tsk = kthread_run(kmmpd, sb, "kmmpd-%.*s",
> - (int)sizeof(mmp->mmp_bdevname),
> - bdevname(bh->b_bdev,
> - mmp->mmp_bdevname));
> + EXT4_SB(sb)->s_mmp_tsk = kthread_create(kmmpd, sb, "kmmpd-%.*s",
> + (int)sizeof(mmp->mmp_bdevname),
> + bdevname(bh->b_bdev,
> + mmp->mmp_bdevname));
> +
> if (IS_ERR(EXT4_SB(sb)->s_mmp_tsk)) {
> EXT4_SB(sb)->s_mmp_tsk = NULL;
> ext4_warning(sb, "Unable to create kmmpd thread for %s.",
> sb->s_id);
> goto failed;
> }
> + wake_up_process(EXT4_SB(sb)->s_mmp_tsk);
>
> return 0;
>
> --
> 2.31.1
>
Cheers, Andreas
Attachment:
signature.asc
Description: Message signed with OpenPGP
