Documentation/security/Yama.txt

   1 Yama is a Linux Security Module that collects system-wide DAC security
   2 protections that are not handled by the core kernel itself. This is
   3 selectable at build-time with CONFIG_SECURITY_YAMA, and can be controlled
   4 at run-time through sysctls in /proc/sys/kernel/yama:
   5
   6 - ptrace_scope
   7
   8 ==============================================================
   9
  10 ptrace_scope:
  11
  12 As Linux grows in popularity, it will become a larger target for
  13 malware. One particularly troubling weakness of the Linux process
  14 interfaces is that a single user is able to examine the memory and
  15 running state of any of their processes. For example, if one application
  16 (e.g. Pidgin) was compromised, it would be possible for an attacker to
  17 attach to other running processes (e.g. Firefox, SSH sessions, GPG agent,
  18 etc) to extract additional credentials and continue to expand the scope
  19 of their attack without resorting to user-assisted phishing.
  20
  21 This is not a theoretical problem. SSH session hijacking
  22 (http://www.storm.net.nz/projects/7) and arbitrary code injection
  23 (http://c-skills.blogspot.com/2007/05/injectso.html) attacks already
  24 exist and remain possible if ptrace is allowed to operate as before.
  25 Since ptrace is not commonly used by non-developers and non-admins, system
  26 builders should be allowed the option to disable this debugging system.
  27
  28 For a solution, some applications use prctl(PR_SET_DUMPABLE, ...) to
  29 specifically disallow such ptrace attachment (e.g. ssh-agent), but many
  30 do not. A more general solution is to only allow ptrace directly from a
  31 parent to a child process (i.e. direct "gdb EXE" and "strace EXE" still
  32 work), or with CAP_SYS_PTRACE (i.e. "gdb --pid=PID", and "strace -p PID"
  33 still work as root).
  34
  35 In mode 1, software that has defined application-specific relationships
  36 between a debugging process and its inferior (crash handlers, etc),
  37 prctl(PR_SET_PTRACER, pid, ...) can be used. An inferior can declare which
  38 other process (and its descendants) are allowed to call PTRACE_ATTACH
  39 against it. Only one such declared debugging process can exists for
  40 each inferior at a time. For example, this is used by KDE, Chromium, and
  41 Firefox's crash handlers, and by Wine for allowing only Wine processes
  42 to ptrace each other. If a process wishes to entirely disable these ptrace
  43 restrictions, it can call prctl(PR_SET_PTRACER, PR_SET_PTRACER_ANY, ...)
  44 so that any otherwise allowed process (even those in external pid namespaces)
  45 may attach.
  46
  47 The sysctl settings (writable only with CAP_SYS_PTRACE) are:
  48
  49 0 - classic ptrace permissions: a process can PTRACE_ATTACH to any other
  50     process running under the same uid, as long as it is dumpable (i.e.
  51     did not transition uids, start privileged, or have called
  52     prctl(PR_SET_DUMPABLE...) already). Similarly, PTRACE_TRACEME is
  53     unchanged.
  54
  55 1 - restricted ptrace: a process must have a predefined relationship
  56     with the inferior it wants to call PTRACE_ATTACH on. By default,
  57     this relationship is that of only its descendants when the above
  58     classic criteria is also met. To change the relationship, an
  59     inferior can call prctl(PR_SET_PTRACER, debugger, ...) to declare
  60     an allowed debugger PID to call PTRACE_ATTACH on the inferior.
  61     Using PTRACE_TRACEME is unchanged.
  62
  63 2 - admin-only attach: only processes with CAP_SYS_PTRACE may use ptrace
  64     with PTRACE_ATTACH, or through children calling PTRACE_TRACEME.
  65
  66 3 - no attach: no processes may use ptrace with PTRACE_ATTACH nor via
  67     PTRACE_TRACEME. Once set, this sysctl value cannot be changed.
  68
  69 The original children-only logic was based on the restrictions in grsecurity.
  70
  71 ==============================================================