kernel-wide: fix missing validations on __get/__put/__copy_to/__copy_from_user()
authorMathieu Desnoyers <mathieu.desnoyers@efficios.com>
Wed, 11 Sep 2013 21:23:18 +0000 (14:23 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 11 Sep 2013 22:58:18 +0000 (15:58 -0700)
commit3ddc5b46a8e90f3c9251338b60191d0a804b0d92
tree5c76cd730cb94e75f30953d6cd1aed9386fcee37
parent20d0e57017b69e7e4ae7166c43f3a3f023ab9702
kernel-wide: fix missing validations on __get/__put/__copy_to/__copy_from_user()

I found the following pattern that leads in to interesting findings:

  grep -r "ret.*|=.*__put_user" *
  grep -r "ret.*|=.*__get_user" *
  grep -r "ret.*|=.*__copy" *

The __put_user() calls in compat_ioctl.c, ptrace compat, signal compat,
since those appear in compat code, we could probably expect the kernel
addresses not to be reachable in the lower 32-bit range, so I think they
might not be exploitable.

For the "__get_user" cases, I don't think those are exploitable: the worse
that can happen is that the kernel will copy kernel memory into in-kernel
buffers, and will fail immediately afterward.

The alpha csum_partial_copy_from_user() seems to be missing the
access_ok() check entirely.  The fix is inspired from x86.  This could
lead to information leak on alpha.  I also noticed that many architectures
map csum_partial_copy_from_user() to csum_partial_copy_generic(), but I
wonder if the latter is performing the access checks on every
architectures.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
arch/alpha/lib/csum_partial_copy.c
arch/sparc/kernel/sys_sparc32.c
block/compat_ioctl.c
kernel/signal.c
net/socket.c