Re: [PATCH v1 1/2] ptrace: Set PF_SUPERPRIV when checking capability

From: Jann Horn
Date: Fri Oct 30 2020 - 14:00:46 EST


On Fri, Oct 30, 2020 at 5:06 PM Mickaël Salaün <mic@xxxxxxxxxxx> wrote:
> On 30/10/2020 16:47, Jann Horn wrote:
> > On Fri, Oct 30, 2020 at 1:39 PM Mickaël Salaün <mic@xxxxxxxxxxx> wrote:
> >> Commit 69f594a38967 ("ptrace: do not audit capability check when outputing
> >> /proc/pid/stat") replaced the use of ns_capable() with
> >> has_ns_capability{,_noaudit}() which doesn't set PF_SUPERPRIV.
> >>
> >> Commit 6b3ad6649a4c ("ptrace: reintroduce usage of subjective credentials in
> >> ptrace_has_cap()") replaced has_ns_capability{,_noaudit}() with
> >> security_capable(), which doesn't set PF_SUPERPRIV neither.
> >>
> >> Since commit 98f368e9e263 ("kernel: Add noaudit variant of ns_capable()"), a
> >> new ns_capable_noaudit() helper is available. Let's use it!
> >>
> >> As a result, the signature of ptrace_has_cap() is restored to its original one.
> >>
> >> Cc: Christian Brauner <christian.brauner@xxxxxxxxxx>
> >> Cc: Eric Paris <eparis@xxxxxxxxxx>
> >> Cc: Jann Horn <jannh@xxxxxxxxxx>
> >> Cc: Kees Cook <keescook@xxxxxxxxxxxx>
> >> Cc: Oleg Nesterov <oleg@xxxxxxxxxx>
> >> Cc: Serge E. Hallyn <serge@xxxxxxxxxx>
> >> Cc: Tyler Hicks <tyhicks@xxxxxxxxxxxxxxxxxxx>
> >> Cc: stable@xxxxxxxxxxxxxxx
> >> Fixes: 6b3ad6649a4c ("ptrace: reintroduce usage of subjective credentials in ptrace_has_cap()")
> >> Fixes: 69f594a38967 ("ptrace: do not audit capability check when outputing /proc/pid/stat")
> >> Signed-off-by: Mickaël Salaün <mic@xxxxxxxxxxxxxxxxxxx>
> >
> > Yeah... I guess this makes sense. (We'd have to undo or change it if
> > we ever end up needing to use a different set of credentials, e.g.
> > from ->f_cred, but I guess that's really something we should avoid
> > anyway.)
> >
> > Reviewed-by: Jann Horn <jannh@xxxxxxxxxx>
> >
> > with one nit:
> >
> >
> > [...]
> >> /* Returns 0 on success, -errno on denial. */
> >> static int __ptrace_may_access(struct task_struct *task, unsigned int mode)
> >> {
> >> - const struct cred *cred = current_cred(), *tcred;
> >> + const struct cred *const cred = current_cred(), *tcred;
> >
> > This is an unrelated change, and almost no kernel code marks local
> > pointer variables as "const". I would drop this change from the patch.
>
> This give guarantee that the cred variable will not be used for
> something else than current_cred(), which kinda prove that this patch
> doesn't change the behavior of __ptrace_may_access() by not using cred
> in ptrace_has_cap(). It doesn't hurt and I think it could be useful to
> spot issues when backporting.

And it might require an extra fixup while backporting because the next
line is different and that might cause the patch to not apply.