[patch 1/2] audit: let userspace fully control TTY input auditing

Remove the code that automatically disables TTY input auditing in processes
that open TTYs when they have no other TTY open; this heuristic was
intended to automatically handle daemons, but it has false positives (e.g.
with sshd) that make it impossible to control TTY input auditing from a PAM
module.  With this patch, TTY input auditing is controlled from user-space
only.

On the other hand, not even for daemons does it make sense to audit "input"
from PTY masters; this data was produced by a program writing to the PTY
slave, and does not represent data entered by the user.

Signed-off-by: Miloslav Trmac <mitr@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

1 files changed, 1 insertions, 4 deletions

diff --git a/drivers/char/tty_io.c b/drivers/char/tty_io.c
index 4d3c7018f0c3..afddccf1bb3d 100644
--- a/drivers/char/tty_io.c
+++ b/drivers/char/tty_io.c
@@ -2755,7 +2755,6 @@ got_driver:
                 __proc_set_tty(current, tty);
         spin_unlock_irq(&current->sighand->siglock);
         mutex_unlock(&tty_mutex);
-        tty_audit_opening();
         return 0;
 }
 
@@ -2818,10 +2817,8 @@ static int ptmx_open(struct inode *inode, struct file *filp)
 
         check_tty_count(tty, "tty_open");
         retval = ptm_driver->open(tty, filp);
-        if (!retval) {
-                tty_audit_opening();
+        if (!retval)
                 return 0;
-        }
 out1:
         release_dev(filp);
         return retval;