[v2] NFS: Avoid cross-structure casting
diff mbox

Message ID 20170405152914.GA57260@beast
State New
Headers show

Commit Message

Kees Cook April 5, 2017, 3:29 p.m. UTC
When the call to nfs_devname() fails, the error path attempts to retain
the error via the mnt variable, but this requires a cast across very
different types (char * to struct vfsmount *), which the upcoming
structure layout randomization plugin flags as being potentially
dangerous in the face of randomization. This is a false positive, but
what this code actually wants to do is retain the error value, so this
patch explicitly sets it, instead of using what seems to be an
unexpected cast.

Signed-off-by: Kees Cook <keescook@chromium.org>
---
v2: duh, use ERR_CAST. thanks neilb!
---
 fs/nfs/namespace.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

Comments

Kees Cook April 12, 2017, 6:59 p.m. UTC | #1
On Wed, Apr 5, 2017 at 8:29 AM, Kees Cook <keescook@chromium.org> wrote:
> When the call to nfs_devname() fails, the error path attempts to retain
> the error via the mnt variable, but this requires a cast across very
> different types (char * to struct vfsmount *), which the upcoming
> structure layout randomization plugin flags as being potentially
> dangerous in the face of randomization. This is a false positive, but
> what this code actually wants to do is retain the error value, so this
> patch explicitly sets it, instead of using what seems to be an
> unexpected cast.
>
> Signed-off-by: Kees Cook <keescook@chromium.org>

If I can get an Acked-by on this, I could push it via the gcc-plugin tree.

Thanks!

-Kees

> ---
> v2: duh, use ERR_CAST. thanks neilb!
> ---
>  fs/nfs/namespace.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c
> index 786f17580582..8ca5d147124d 100644
> --- a/fs/nfs/namespace.c
> +++ b/fs/nfs/namespace.c
> @@ -259,9 +259,10 @@ struct vfsmount *nfs_do_submount(struct dentry *dentry, struct nfs_fh *fh,
>         if (page == NULL)
>                 goto out;
>         devname = nfs_devname(dentry, page, PAGE_SIZE);
> -       mnt = (struct vfsmount *)devname;
> -       if (IS_ERR(devname))
> +       if (IS_ERR(devname)) {
> +               mnt = ERR_CAST(devname);
>                 goto free_page;
> +       }
>         mnt = nfs_do_clone_mount(NFS_SB(dentry->d_sb), devname, &mountdata);
>  free_page:
>         free_page((unsigned long)page);
> --
> 2.7.4
>
>
> --
> Kees Cook
> Pixel Security
Trond Myklebust April 20, 2017, 6:04 p.m. UTC | #2
On Wed, 2017-04-12 at 11:59 -0700, Kees Cook wrote:
> On Wed, Apr 5, 2017 at 8:29 AM, Kees Cook <keescook@chromium.org>

> wrote:

> > When the call to nfs_devname() fails, the error path attempts to

> > retain

> > the error via the mnt variable, but this requires a cast across

> > very

> > different types (char * to struct vfsmount *), which the upcoming

> > structure layout randomization plugin flags as being potentially

> > dangerous in the face of randomization. This is a false positive,

> > but

> > what this code actually wants to do is retain the error value, so

> > this

> > patch explicitly sets it, instead of using what seems to be an

> > unexpected cast.

> > 

> > Signed-off-by: Kees Cook <keescook@chromium.org>

> 

> If I can get an Acked-by on this, I could push it via the gcc-plugin

> tree.

> 

> Thanks!

> 

> -Kees

> 

> > ---

> > v2: duh, use ERR_CAST. thanks neilb!

> > ---

> >  fs/nfs/namespace.c | 5 +++--

> >  1 file changed, 3 insertions(+), 2 deletions(-)

> > 

> > diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c

> > index 786f17580582..8ca5d147124d 100644

> > --- a/fs/nfs/namespace.c

> > +++ b/fs/nfs/namespace.c

> > @@ -259,9 +259,10 @@ struct vfsmount *nfs_do_submount(struct dentry

> > *dentry, struct nfs_fh *fh,

> >         if (page == NULL)

> >                 goto out;

> >         devname = nfs_devname(dentry, page, PAGE_SIZE);

> > -       mnt = (struct vfsmount *)devname;

> > -       if (IS_ERR(devname))

> > +       if (IS_ERR(devname)) {

> > +               mnt = ERR_CAST(devname);

> >                 goto free_page;

> > +       }

> >         mnt = nfs_do_clone_mount(NFS_SB(dentry->d_sb), devname,

> > &mountdata);

> >  free_page:

> >         free_page((unsigned long)page);

> > --

> > 2.7.4

> > 

> > 

> > --

> > Kees Cook

> > Pixel Security

> 


Acked-by: Trond Myklebust <trond.myklebust@primarydata.com>


-- 
Trond Myklebust
Linux NFS client maintainer, PrimaryData
trond.myklebust@primarydata.com

Patch
diff mbox

diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c
index 786f17580582..8ca5d147124d 100644
--- a/fs/nfs/namespace.c
+++ b/fs/nfs/namespace.c
@@ -259,9 +259,10 @@  struct vfsmount *nfs_do_submount(struct dentry *dentry, struct nfs_fh *fh,
 	if (page == NULL)
 		goto out;
 	devname = nfs_devname(dentry, page, PAGE_SIZE);
-	mnt = (struct vfsmount *)devname;
-	if (IS_ERR(devname))
+	if (IS_ERR(devname)) {
+		mnt = ERR_CAST(devname);
 		goto free_page;
+	}
 	mnt = nfs_do_clone_mount(NFS_SB(dentry->d_sb), devname, &mountdata);
 free_page:
 	free_page((unsigned long)page);