[PATCH v5 2/3] overlayfs: check CAP_MKNOD before issuing vfs_whiteout

[PATCH v5 2/3] overlayfs: check CAP_MKNOD before issuing vfs_whiteout

[Mark Salyzyn]

Assumption never checked, should fail if the mounter creds are not
sufficient.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Vivek Goyal <vgoyal@redhat.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: Amir Goldstein <amir73il@gmail.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Stephen Smalley <sds@tycho.nsa.gov>
Cc: linux-unionfs@vger.kernel.org
Cc: linux-doc@vger.kernel.org
Cc: linux-kernel@vger.kernel.org

v5
- dependency of "overlayfs: override_creds=off option bypass creator_cred"
---
 fs/overlayfs/overlayfs.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h
index 7538b9b56237..bf3a80157d42 100644
--- a/fs/overlayfs/overlayfs.h
+++ b/fs/overlayfs/overlayfs.h
@@ -176,7 +176,7 @@ static inline int ovl_do_rename(struct inode *olddir, struct dentry *olddentry,
 
 static inline int ovl_do_whiteout(struct inode *dir, struct dentry *dentry)
 {
-	int err = vfs_whiteout(dir, dentry);
+	int err = capable(CAP_MKNOD) ? vfs_whiteout(dir, dentry) : -EPERM;
 	pr_debug("whiteout(%pd2) = %i\n", dentry, err);
 	return err;
 }
-- 
2.19.0.rc0.228.g281dcd1b4d0-goog

Re: [PATCH v5 2/3] overlayfs: check CAP_MKNOD before issuing vfs_whiteout

[Amir Goldstein]

On Tue, Aug 28, 2018 at 7:53 PM Mark Salyzyn <salyzyn@android.com> wrote:
>
> Assumption never checked, should fail if the mounter creds are not
> sufficient.
>
> Signed-off-by: Mark Salyzyn <salyzyn@android.com>
> Cc: Miklos Szeredi <miklos@szeredi.hu>
> Cc: Jonathan Corbet <corbet@lwn.net>
> Cc: Vivek Goyal <vgoyal@redhat.com>
> Cc: Eric W. Biederman <ebiederm@xmission.com>
> Cc: Amir Goldstein <amir73il@gmail.com>
> Cc: Randy Dunlap <rdunlap@infradead.org>
> Cc: Stephen Smalley <sds@tycho.nsa.gov>
> Cc: linux-unionfs@vger.kernel.org
> Cc: linux-doc@vger.kernel.org
> Cc: linux-kernel@vger.kernel.org
>
> v5
> - dependency of "overlayfs: override_creds=off option bypass creator_cred"
> ---
>  fs/overlayfs/overlayfs.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h
> index 7538b9b56237..bf3a80157d42 100644
> --- a/fs/overlayfs/overlayfs.h
> +++ b/fs/overlayfs/overlayfs.h
> @@ -176,7 +176,7 @@ static inline int ovl_do_rename(struct inode *olddir, struct dentry *olddentry,
>
>  static inline int ovl_do_whiteout(struct inode *dir, struct dentry *dentry)
>  {
> -       int err = vfs_whiteout(dir, dentry);
> +       int err = capable(CAP_MKNOD) ? vfs_whiteout(dir, dentry) : -EPERM;

Should that be ns_capable()? Should the test go into vfs_whiteout()?
I feel there is no convention at all.

Thanks,
Amir.

Re: [PATCH v5 2/3] overlayfs: check CAP_MKNOD before issuing vfs_whiteout

[Amir Goldstein]

On Tue, Aug 28, 2018 at 8:43 PM Amir Goldstein <amir73il@gmail.com> wrote:
>
> On Tue, Aug 28, 2018 at 7:53 PM Mark Salyzyn <salyzyn@android.com> wrote:
> >
> > Assumption never checked, should fail if the mounter creds are not
> > sufficient.
> >
> > Signed-off-by: Mark Salyzyn <salyzyn@android.com>
> > Cc: Miklos Szeredi <miklos@szeredi.hu>
> > Cc: Jonathan Corbet <corbet@lwn.net>
> > Cc: Vivek Goyal <vgoyal@redhat.com>
> > Cc: Eric W. Biederman <ebiederm@xmission.com>
> > Cc: Amir Goldstein <amir73il@gmail.com>
> > Cc: Randy Dunlap <rdunlap@infradead.org>
> > Cc: Stephen Smalley <sds@tycho.nsa.gov>
> > Cc: linux-unionfs@vger.kernel.org
> > Cc: linux-doc@vger.kernel.org
> > Cc: linux-kernel@vger.kernel.org
> >
> > v5
> > - dependency of "overlayfs: override_creds=off option bypass creator_cred"
> > ---
> >  fs/overlayfs/overlayfs.h | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h
> > index 7538b9b56237..bf3a80157d42 100644
> > --- a/fs/overlayfs/overlayfs.h
> > +++ b/fs/overlayfs/overlayfs.h
> > @@ -176,7 +176,7 @@ static inline int ovl_do_rename(struct inode *olddir, struct dentry *olddentry,
> >
> >  static inline int ovl_do_whiteout(struct inode *dir, struct dentry *dentry)
> >  {
> > -       int err = vfs_whiteout(dir, dentry);
> > +       int err = capable(CAP_MKNOD) ? vfs_whiteout(dir, dentry) : -EPERM;
>
> Should that be ns_capable()? Should the test go into vfs_whiteout()?
> I feel there is no convention at all.
>

Nevermind, I don't think creating a whiteout poses any risk, so don't think
we need to worry about CAP_MKNOD.

Thanks,
Amir.

Re: [PATCH v5 2/3] overlayfs: check CAP_MKNOD before issuing vfs_whiteout

[Mark Salyzyn]

On 08/28/2018 11:42 AM, Amir Goldstein wrote:
> On Tue, Aug 28, 2018 at 8:43 PM Amir Goldstein <amir73il@gmail.com> wrote:
>> On Tue, Aug 28, 2018 at 7:53 PM Mark Salyzyn <salyzyn@android.com> wrote:
>>> Assumption never checked, should fail if the mounter creds are not
>>> sufficient.
>>>
>>> Signed-off-by: Mark Salyzyn <salyzyn@android.com>
>>> Cc: Miklos Szeredi <miklos@szeredi.hu>
>>> Cc: Jonathan Corbet <corbet@lwn.net>
>>> Cc: Vivek Goyal <vgoyal@redhat.com>
>>> Cc: Eric W. Biederman <ebiederm@xmission.com>
>>> Cc: Amir Goldstein <amir73il@gmail.com>
>>> Cc: Randy Dunlap <rdunlap@infradead.org>
>>> Cc: Stephen Smalley <sds@tycho.nsa.gov>
>>> Cc: linux-unionfs@vger.kernel.org
>>> Cc: linux-doc@vger.kernel.org
>>> Cc: linux-kernel@vger.kernel.org
>>>
>>> v5
>>> - dependency of "overlayfs: override_creds=off option bypass creator_cred"
>>> ---
>>>   fs/overlayfs/overlayfs.h | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h
>>> index 7538b9b56237..bf3a80157d42 100644
>>> --- a/fs/overlayfs/overlayfs.h
>>> +++ b/fs/overlayfs/overlayfs.h
>>> @@ -176,7 +176,7 @@ static inline int ovl_do_rename(struct inode *olddir, struct dentry *olddentry,
>>>
>>>   static inline int ovl_do_whiteout(struct inode *dir, struct dentry *dentry)
>>>   {
>>> -       int err = vfs_whiteout(dir, dentry);
>>> +       int err = capable(CAP_MKNOD) ? vfs_whiteout(dir, dentry) : -EPERM;
>> Should that be ns_capable()? Should the test go into vfs_whiteout()?
>> I feel there is no convention at all.
>>
> Nevermind, I don't think creating a whiteout poses any risk, so don't think
> we need to worry about CAP_MKNOD.
>
> Thanks,
> Amir.

Ok, will discard from the set, we can address this later if it creates 
concern (as in, not a dependency to my proposed feature flag). So we 
feel that whiteout node in the writeable playground of workdir/upperdir 
is not in itself a security concern. Other (more dangerous) mknod will 
be checked against the caller's credentials coming in.

-- Mark