Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755999Ab0F3Ais (ORCPT ); Tue, 29 Jun 2010 20:38:48 -0400 Received: from smtp.outflux.net ([198.145.64.163]:54710 "EHLO smtp.outflux.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754328Ab0F3Aip (ORCPT ); Tue, 29 Jun 2010 20:38:45 -0400 Date: Tue, 29 Jun 2010 17:38:44 -0700 From: Kees Cook To: linux-security-module@vger.kernel.org Cc: linux-kernel@vger.kernel.org Subject: [PATCH 0/2] Yama: add PTRACE exception tracking Message-ID: <20100630003844.GE4837@outflux.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Organization: Canonical X-HELO: www.outflux.net Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1949 Lines: 47 The primary exception to Yama's descendant-based PTRACE restrictions is when an application has a predefined crash handler that is spawned in parallel with the crashed application (e.g. KDE, Chromium). These applications want to bypass the common RLIMIT_CORE=0, and gather state information from the process for upstream problem reporting. When the main application crashes, it generally has access to the PID of what will debug it (e.g. when a KDE app crashes, it knows the parent PID of the debugger that will be spawned). So, since this programmatic method of PTRACEing is useful, there should be a way for processes to actively declare who can PTRACE them. This patch adds a prctl hook for Yama so that processes can exempt themselves from the PTRACE restrictions in the case of a crash when they know their debugger's PID. As a matter of demonstration, here is what the patch to KDE4 would look like to support Yama, or other PTRACE-restricting LSMs that wanted to grant a similar exception: --- kde4libs-4.4.90.orig/kdeui/util/kcrash.cpp 2010-06-28 17:07:28.667869954 -0700 +++ kde4libs-4.4.90/kdeui/util/kcrash.cpp 2010-06-28 17:09:32.089958401 -0700 @@ -41,6 +41,7 @@ #include #include #include +#include #include #include @@ -437,6 +438,7 @@ //if the process was started directly, use waitpid(), as it's a child... while(waitpid(-1, NULL, 0) != pid) {} } else { + prctl(PR_SET_PTRACER, pid, 0, 0, 0); //...else poll its status using kill() while(kill(pid, 0) >= 0) { sleep(1); -- Kees Cook Ubuntu Security Team -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/