selinux: check sidtab limit before adding a new entry
diff mbox series

Message ID 20190722132111.25743-1-omosnace@redhat.com
State New
Headers show
Series
  • selinux: check sidtab limit before adding a new entry
Related show

Commit Message

Ondrej Mosnacek July 22, 2019, 1:21 p.m. UTC
We need to error out when trying to add an entry above SIDTAB_MAX in
sidtab_reverse_lookup() to avoid overflow on the odd chance that this
happens.

Fixes: ee1a84fdfeed ("selinux: overhaul sidtab to fix bug and improve performance")
Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>
---
 security/selinux/ss/sidtab.c | 5 +++++
 1 file changed, 5 insertions(+)

Comments

William Roberts July 22, 2019, 2:17 p.m. UTC | #1
On Mon, Jul 22, 2019 at 8:34 AM Ondrej Mosnacek <omosnace@redhat.com> wrote:
>
> We need to error out when trying to add an entry above SIDTAB_MAX in
> sidtab_reverse_lookup() to avoid overflow on the odd chance that this
> happens.
>
> Fixes: ee1a84fdfeed ("selinux: overhaul sidtab to fix bug and improve performance")
> Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>
> ---
>  security/selinux/ss/sidtab.c | 5 +++++
>  1 file changed, 5 insertions(+)
>
> diff --git a/security/selinux/ss/sidtab.c b/security/selinux/ss/sidtab.c
> index e63a90ff2728..54c1ba1e79ab 100644
> --- a/security/selinux/ss/sidtab.c
> +++ b/security/selinux/ss/sidtab.c
> @@ -286,6 +286,11 @@ static int sidtab_reverse_lookup(struct sidtab *s, struct context *context,
>                 ++count;
>         }
>
> +       /* bail out if we already reached max entries */
> +       rc = -ENOMEM;

Wouldn't -EOVERFLOW be better?

> +       if (count == SIDTAB_MAX)
> +               goto out_unlock;
> +
>         /* insert context into new entry */
>         rc = -ENOMEM;
>         dst = sidtab_do_lookup(s, count, 1);
> --
> 2.21.0
>
Kees Cook July 22, 2019, 4:50 p.m. UTC | #2
On Mon, Jul 22, 2019 at 03:21:11PM +0200, Ondrej Mosnacek wrote:
> We need to error out when trying to add an entry above SIDTAB_MAX in
> sidtab_reverse_lookup() to avoid overflow on the odd chance that this
> happens.
> 
> Fixes: ee1a84fdfeed ("selinux: overhaul sidtab to fix bug and improve performance")
> Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>

Is this reachable from unprivileged userspace?

> ---
>  security/selinux/ss/sidtab.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/security/selinux/ss/sidtab.c b/security/selinux/ss/sidtab.c
> index e63a90ff2728..54c1ba1e79ab 100644
> --- a/security/selinux/ss/sidtab.c
> +++ b/security/selinux/ss/sidtab.c
> @@ -286,6 +286,11 @@ static int sidtab_reverse_lookup(struct sidtab *s, struct context *context,
>  		++count;
>  	}
>  
> +	/* bail out if we already reached max entries */
> +	rc = -ENOMEM;
> +	if (count == SIDTAB_MAX)

Do you want to use >= here instead?

-Kees

> +		goto out_unlock;
> +
>  	/* insert context into new entry */
>  	rc = -ENOMEM;
>  	dst = sidtab_do_lookup(s, count, 1);
> -- 
> 2.21.0
>
Paul Moore July 23, 2019, 12:36 a.m. UTC | #3
On Mon, Jul 22, 2019 at 12:50 PM Kees Cook <keescook@chromium.org> wrote:
> On Mon, Jul 22, 2019 at 03:21:11PM +0200, Ondrej Mosnacek wrote:
> > We need to error out when trying to add an entry above SIDTAB_MAX in
> > sidtab_reverse_lookup() to avoid overflow on the odd chance that this
> > happens.
> >
> > Fixes: ee1a84fdfeed ("selinux: overhaul sidtab to fix bug and improve performance")
> > Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>
>
> Is this reachable from unprivileged userspace?

I believe it's reachable via selinuxfs under /sys/fs/selinux/context,
and the DAC permissions are for the relevant files are 0666, but the
SELinux policy might restrict that.

> > ---
> >  security/selinux/ss/sidtab.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> >
> > diff --git a/security/selinux/ss/sidtab.c b/security/selinux/ss/sidtab.c
> > index e63a90ff2728..54c1ba1e79ab 100644
> > --- a/security/selinux/ss/sidtab.c
> > +++ b/security/selinux/ss/sidtab.c
> > @@ -286,6 +286,11 @@ static int sidtab_reverse_lookup(struct sidtab *s, struct context *context,
> >               ++count;
> >       }
> >
> > +     /* bail out if we already reached max entries */
> > +     rc = -ENOMEM;
> > +     if (count == SIDTAB_MAX)
>
> Do you want to use >= here instead?

Yes, definitely.
Ondrej Mosnacek July 23, 2019, 6:48 a.m. UTC | #4
On Mon, Jul 22, 2019 at 4:17 PM William Roberts
<bill.c.roberts@gmail.com> wrote:
> On Mon, Jul 22, 2019 at 8:34 AM Ondrej Mosnacek <omosnace@redhat.com> wrote:
> >
> > We need to error out when trying to add an entry above SIDTAB_MAX in
> > sidtab_reverse_lookup() to avoid overflow on the odd chance that this
> > happens.
> >
> > Fixes: ee1a84fdfeed ("selinux: overhaul sidtab to fix bug and improve performance")
> > Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>
> > ---
> >  security/selinux/ss/sidtab.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> >
> > diff --git a/security/selinux/ss/sidtab.c b/security/selinux/ss/sidtab.c
> > index e63a90ff2728..54c1ba1e79ab 100644
> > --- a/security/selinux/ss/sidtab.c
> > +++ b/security/selinux/ss/sidtab.c
> > @@ -286,6 +286,11 @@ static int sidtab_reverse_lookup(struct sidtab *s, struct context *context,
> >                 ++count;
> >         }
> >
> > +       /* bail out if we already reached max entries */
> > +       rc = -ENOMEM;
>
> Wouldn't -EOVERFLOW be better?

Good point. Will change it in v2.

>
> > +       if (count == SIDTAB_MAX)
> > +               goto out_unlock;
> > +
> >         /* insert context into new entry */
> >         rc = -ENOMEM;
> >         dst = sidtab_do_lookup(s, count, 1);
> > --
> > 2.21.0
> >

Thanks,
Ondrej Mosnacek July 23, 2019, 6:48 a.m. UTC | #5
On Mon, Jul 22, 2019 at 6:50 PM Kees Cook <keescook@chromium.org> wrote:
> On Mon, Jul 22, 2019 at 03:21:11PM +0200, Ondrej Mosnacek wrote:
> > We need to error out when trying to add an entry above SIDTAB_MAX in
> > sidtab_reverse_lookup() to avoid overflow on the odd chance that this
> > happens.
> >
> > Fixes: ee1a84fdfeed ("selinux: overhaul sidtab to fix bug and improve performance")
> > Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>
>
> Is this reachable from unprivileged userspace?
>
> > ---
> >  security/selinux/ss/sidtab.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> >
> > diff --git a/security/selinux/ss/sidtab.c b/security/selinux/ss/sidtab.c
> > index e63a90ff2728..54c1ba1e79ab 100644
> > --- a/security/selinux/ss/sidtab.c
> > +++ b/security/selinux/ss/sidtab.c
> > @@ -286,6 +286,11 @@ static int sidtab_reverse_lookup(struct sidtab *s, struct context *context,
> >               ++count;
> >       }
> >
> > +     /* bail out if we already reached max entries */
> > +     rc = -ENOMEM;
> > +     if (count == SIDTAB_MAX)
>
> Do you want to use >= here instead?

Makes sense. Also staged for v2.

>
> -Kees
>
> > +             goto out_unlock;
> > +
> >       /* insert context into new entry */
> >       rc = -ENOMEM;
> >       dst = sidtab_do_lookup(s, count, 1);
> > --
> > 2.21.0
> >
>
> --
> Kees Cook

Thanks,

Patch
diff mbox series

diff --git a/security/selinux/ss/sidtab.c b/security/selinux/ss/sidtab.c
index e63a90ff2728..54c1ba1e79ab 100644
--- a/security/selinux/ss/sidtab.c
+++ b/security/selinux/ss/sidtab.c
@@ -286,6 +286,11 @@  static int sidtab_reverse_lookup(struct sidtab *s, struct context *context,
 		++count;
 	}
 
+	/* bail out if we already reached max entries */
+	rc = -ENOMEM;
+	if (count == SIDTAB_MAX)
+		goto out_unlock;
+
 	/* insert context into new entry */
 	rc = -ENOMEM;
 	dst = sidtab_do_lookup(s, count, 1);