proc: maps protection
authorKees Cook <kees@outflux.net>
Tue, 8 May 2007 07:26:04 +0000 (00:26 -0700)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Tue, 8 May 2007 18:15:02 +0000 (11:15 -0700)
commit5096add84b9e96e2e0a9c72675c442fe5433388a
treef0444013cb7db32596d2b6febafc1ee4c2a4ea1f
parent4a1ccb5b1eff949a90ab830869cb23d6609c3d5f
proc: maps protection

The /proc/pid/ "maps", "smaps", and "numa_maps" files contain sensitive
information about the memory location and usage of processes.  Issues:

- maps should not be world-readable, especially if programs expect any
  kind of ASLR protection from local attackers.
- maps cannot just be 0400 because "-D_FORTIFY_SOURCE=2 -O2" makes glibc
  check the maps when %n is in a *printf call, and a setuid(getuid())
  process wouldn't be able to read its own maps file.  (For reference
  see http://lkml.org/lkml/2006/1/22/150)
- a system-wide toggle is needed to allow prior behavior in the case of
  non-root applications that depend on access to the maps contents.

This change implements a check using "ptrace_may_attach" before allowing
access to read the maps contents.  To control this protection, the new knob
/proc/sys/kernel/maps_protect has been added, with corresponding updates to
the procfs documentation.

[akpm@linux-foundation.org: build fixes]
[akpm@linux-foundation.org: New sysctl numbers are old hat]
Signed-off-by: Kees Cook <kees@outflux.net>
Cc: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
CREDITS
Documentation/filesystems/proc.txt
fs/proc/base.c
fs/proc/internal.h
fs/proc/task_mmu.c
fs/proc/task_nommu.c
kernel/sysctl.c