proc: make proc_fd_permission() thread-friendly
Oleg Nesterov [Wed, 11 Sep 2013 21:24:37 +0000 (14:24 -0700)]
proc_fd_permission() says "process can still access /proc/self/fd after it
has executed a setuid()", but the "task_pid() = proc_pid() check only
helps if the task is group leader, /proc/self points to
/proc/<leader-pid>.

Change this check to use task_tgid() so that the whole thread group can
access its /proc/self/fd or /proc/<tid-of-sub-thread>/fd.

Notes:
- CLONE_THREAD does not require CLONE_FILES so task->files
  can differ, but I don't think this can lead to any security
  problem. And this matches same_thread_group() in
  __ptrace_may_access().

- /proc/self should probably point to /proc/<thread-tid>, but
  it is too late to change the rules. Perhaps it makes sense
  to add /proc/thread though.

Test-case:

void *tfunc(void *arg)
{
assert(opendir("/proc/self/fd"));
return NULL;
}

int main(void)
{
pthread_t t;
pthread_create(&t, NULL, tfunc, NULL);
pthread_join(t, NULL);
return 0;
}

fails if, say, this executable is not readable and suid_dumpable = 0.

Bug 200226999

Change-Id: I4a0f2648e257b75c8b03516a63f655a1adbe942c
(Cherry-picked from 96d0df79f2644fc823f26c06491e182d87a90c2a)
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Vaibhav Shinde <vashinde@nvidia.com>
Reviewed-on: http://git-master/r/1212863
Reviewed-by: Sri Krishna Chowdary <schowdary@nvidia.com>
Reviewed-by: Dhiren Parmar <dparmar@nvidia.com>
Reviewed-by: Gagan Grover <ggrover@nvidia.com>

fs/proc/fd.c

index d7a4a28..50ac26b 100644 (file)
@@ -305,7 +305,7 @@ int proc_fd_permission(struct inode *inode, int mask)
        int rv = generic_permission(inode, mask);
        if (rv == 0)
                return 0;
-       if (task_pid(current) == proc_pid(inode))
+       if (task_tgid(current) == proc_pid(inode))
                rv = 0;
        return rv;
 }