lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20220817102722.wny7x5iwf62edpkd@wittgenstein>
Date:   Wed, 17 Aug 2022 12:27:22 +0200
From:   Christian Brauner <brauner@...nel.org>
To:     Amir Goldstein <amir73il@...il.com>
Cc:     天赐张 <zhangtianci.1997@...edance.com>,
        Miklos Szeredi <miklos@...redi.hu>,
        overlayfs <linux-unionfs@...r.kernel.org>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        Jiachen Zhang <zhangjiachen.jaycee@...edance.com>
Subject: Re: [External] Re: [PATCH] ovl: Do not override fsuid and fsgid in
 ovl_link()

On Wed, Aug 17, 2022 at 12:55:22PM +0300, Amir Goldstein wrote:
> On Wed, Aug 17, 2022 at 12:53 PM Amir Goldstein <amir73il@...il.com> wrote:
> >
> > On Wed, Aug 17, 2022 at 12:11 PM 天赐张 <zhangtianci.1997@...edance.com> wrote:
> > >
> > > On Wed, Aug 17, 2022 at 3:36 PM Amir Goldstein <amir73il@...il.com> wrote:
> > > >
> > > > On Wed, Aug 17, 2022 at 6:49 AM Zhang Tianci
> > > > <zhangtianci.1997@...edance.com> wrote:
> > > > >
> > > > > ovl_link() did not create a new inode after commit
> > > > > 51f7e52dc943 ("ovl: share inode for hard link"), so
> > > > > in ovl_create_or_link() we should not override cred's
> > > > > fsuid and fsgid when called by ovl_link().
> > > > >
> > > > > Signed-off-by: Zhang Tianci <zhangtianci.1997@...edance.com>
> > > > > Signed-off-by: Jiachen Zhang <zhangjiachen.jaycee@...edance.com>
> > > > > ---
> > > > >  fs/overlayfs/dir.c | 4 ++--
> > > > >  1 file changed, 2 insertions(+), 2 deletions(-)
> > > > >
> > > > > diff --git a/fs/overlayfs/dir.c b/fs/overlayfs/dir.c
> > > > > index 6b03457f72bb..568d338032db 100644
> > > > > --- a/fs/overlayfs/dir.c
> > > > > +++ b/fs/overlayfs/dir.c
> > > > > @@ -595,9 +595,9 @@ static int ovl_create_or_link(struct dentry *dentry, struct inode *inode,
> > > > >         err = -ENOMEM;
> > > > >         override_cred = prepare_creds();
> > > > >         if (override_cred) {
> > > > > -               override_cred->fsuid = inode->i_uid;
> > > > > -               override_cred->fsgid = inode->i_gid;
> > > > >                 if (!attr->hardlink) {
> > > > > +                       override_cred->fsuid = inode->i_uid;
> > > > > +                       override_cred->fsgid = inode->i_gid;
> > > > >                         err = security_dentry_create_files_as(dentry,
> > > > >                                         attr->mode, &dentry->d_name, old_cred,
> > > > >                                         override_cred);
> > > > > --
> > > >
> > > > This change looks incorrect.
> > > > Unless I am missing something, fsuid/fsgid still need to
> > > > be overridden for calling link() on underlying fs.
> > > > What made you do this change?
> > > >
> > > > Thanks,
> > > > Amir.
> > >
> > > Hi Amir,
> > >
> > > I ran into an error when I tested overlay on fuse:
> > >   $ mkdir /lower /fuse /merge
> > >   $ mount -t fuse /fuse
> > >   $ mkdir /fuse/upper /fuse/work
> > >   $ mount -t overlay /merge -o lowerdir=/lower,upperdir=/fuse/upper,workdir=work
> > >   $ touch /merge/file
> > >   $ chown bin.bin /merge/file // the file's caller becomes "bin"
> > >   $ ln /merge/file /merge/lnkfile
> > >
> > > Then I got an error(EACCES) because fuse daemon checks the link()'s
> > > caller is "bin", it denied this request.
> > > I browsed the changing history of ovl_link(). There are two key commits:
> > > The first is commit bb0d2b8ad296 ("ovl: fix sgid on directory") which
> > > overrides the cred's fsuid/fsgid using the new inode. The new inode's
> > > owner is initialized by inode_init_owner(), and inode->fsuid is
> > > assigned to the current user. So the override fsuid becomes the
> > > current user. We know link() is actually modifying the directory, so
> > > the caller must have the MAY_WRITE permission on the directory. The
> > > current caller may should have this permission. I think this is
> > > acceptable to use the caller's fsuid(But I still feel a little
> > > conflicted with the overlay's design).
> > > The second is commit 51f7e52dc943 ("ovl: share inode for hard link")
> > > which removed the inode creation in ovl_link(). This commit move
> > > inode_init_owner() into ovl_create_object(), so the ovl_link() just
> > > give the old inode to ovl_create_or_link(). Then the override fsuid
> > > becomes the old inode's fsuid, neither the caller nor the overlay's
> > > creator! So I think this is incorrect.
> > > I think the link() should be like unlink(), overlay fs should just use
> > > the creator cred to do underlying fs's operations.
> > >
> >
> > I see. The reproducer and explanation belong in the commit message.
> >
> > Your argument makes sense to me, but CC Christian to make
> > sure I am not missing anything related to ACLs and what not.
> 
> Once again with correct email address...

So we have:

ovl_create_object()
-> ovl_override_creds(ovl_sb)
-> ovl_new_inode()
   -> inode_init_owner()
      {
              inode->i_uid = current_fsuid();
              inode->i_gid = current_fsgid();
      }
-> ovl_create_or_link(inode, ...)
-> prepare_creds() // Copy of caller's creds
{
        override_creds->fsuid = inode->i_uid;
        override_creds->fsgid = inode->i_gid;
}
-> revert_creds()

which afaict means that the mounter's credentials are used apart from
the fs{g,u}id which is taken from inode->i_{g,u}id which should
correspond to current_fs{g,u}id().

The commit that is pointed out in the patch
51f7e52dc943 ("ovl: share inode for hard link")
seems to have broken that assumption.

Given that the intention was to use the creator's creds _with the
caller's fs{g,u}id_ wouldn't it make more sense to simply ensure that
the caller's fs{g,u}id are always used instead of using the full
creator's creds just for the link operation? So something like this
(untested):

diff --git a/fs/overlayfs/dir.c b/fs/overlayfs/dir.c
index 6b03457f72bb..4a3ee16a6d70 100644
--- a/fs/overlayfs/dir.c
+++ b/fs/overlayfs/dir.c
@@ -575,6 +575,9 @@ static int ovl_create_or_link(struct dentry *dentry, struct inode *inode,
        const struct cred *old_cred;
        struct cred *override_cred;
        struct dentry *parent = dentry->d_parent;
+       /* Retrieve caller's fs{g,u}id before we override creds below. */
+       kuid_t caller_fsuid = current_fsuid();
+       kgid_t caller_fsgid = current_fsgid();

        err = ovl_copy_up(parent);
        if (err)
@@ -595,8 +598,8 @@ static int ovl_create_or_link(struct dentry *dentry, struct inode *inode,
        err = -ENOMEM;
        override_cred = prepare_creds();
        if (override_cred) {
-               override_cred->fsuid = inode->i_uid;
-               override_cred->fsgid = inode->i_gid;
+               override_cred->fsuid = caller_fsuid;
+               override_cred->fsgid = caller_fsgid;
                if (!attr->hardlink) {
                        err = security_dentry_create_files_as(dentry,
                                        attr->mode, &dentry->d_name, old_cred,

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ