From patchwork Wed Sep 5 13:38:08 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Stefan Liebler X-Patchwork-Id: 29205 Received: (qmail 77001 invoked by alias); 5 Sep 2018 13:38:17 -0000 Mailing-List: contact libc-alpha-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-alpha-owner@sourceware.org Delivered-To: mailing list libc-alpha@sourceware.org Received: (qmail 76992 invoked by uid 89); 5 Sep 2018 13:38:17 -0000 Authentication-Results: sourceware.org; auth=none X-Spam-SWARE-Status: No, score=-25.8 required=5.0 tests=AWL, BAYES_00, GIT_PATCH_0, GIT_PATCH_1, GIT_PATCH_2, GIT_PATCH_3, RCVD_IN_DNSWL_LOW, SPF_PASS autolearn=ham version=3.3.2 spammy=inclusive, termination X-HELO: mx0a-001b2d01.pphosted.com To: GNU C Library Cc: Adhemerval Zanella From: Stefan Liebler Subject: [PATCH] Fix segfault in maybe_script_execute. Date: Wed, 5 Sep 2018 15:38:08 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.9.1 MIME-Version: 1.0 x-cbid: 18090513-0020-0000-0000-000002C1A308 X-IBM-AV-DETECTION: SAVI=unused REMOTE=unused XFE=unused x-cbparentid: 18090513-0021-0000-0000-0000210ED133 Message-Id: <6fab031c-9748-42e4-7137-55bf9ae59545@linux.ibm.com> Hi, If glibc is built with gcc 8 and -march=z900, the testcase posix/tst-spawn4-compat crashes with a segfault. In function maybe_script_execute, the new_argv array is dynamically initialized on stack with (argc + 1) elements. The function wants to add _PATH_BSHELL as the first argument. But if argc == 1, it writes three instead of two elements: new_argv[0] = (char *) _PATH_BSHELL; new_argv[1] = (char *) args->file; new_argv[2] = NULL; The latter write access writes to the same location where the pointer args is stored on stack. Then it segfaults while accessing args: args->exec (new_argv[0], new_argv, args->envp); If argc > 1, the arguments inclusive the NULL termination of new_argv is copied from argv with memcpy. The implementation assumes that argc == 0 will never happen! Okay to commit? Bye Stefan --- ChangeLog: * sysdeps/unix/sysv/linux/spawni.c (maybe_script_execute): Increment size of new_argv by one. commit d099212b378f23fef38ee3f51168bb247d5f719d Author: Stefan Liebler Date: Wed Sep 5 13:34:44 2018 +0200 Fix segfault in maybe_script_execute. If glibc is built with gcc 8 and -march=z900, the testcase posix/tst-spawn4-compat crashes with a segfault. In function maybe_script_execute, the new_argv array is dynamically initialized on stack with (argc + 1) elements. The function wants to add _PATH_BSHELL as the first argument. But if argc == 1, it writes three instead of two elements: new_argv[0] = (char *) _PATH_BSHELL; new_argv[1] = (char *) args->file; new_argv[2] = NULL; The latter write access writes to the same location where the pointer args is stored on stack. Then it segfaults while accessing args: args->exec (new_argv[0], new_argv, args->envp); If argc > 1, the arguments inclusive the NULL termination of new_argv is copied from argv with memcpy. The implementation assumes that argc == 0 will never happen! ChangeLog: * sysdeps/unix/sysv/linux/spawni.c (maybe_script_execute): Increment size of new_argv by one. diff --git a/sysdeps/unix/sysv/linux/spawni.c b/sysdeps/unix/sysv/linux/spawni.c index cf0213ece5..85239cedbf 100644 --- a/sysdeps/unix/sysv/linux/spawni.c +++ b/sysdeps/unix/sysv/linux/spawni.c @@ -101,7 +101,7 @@ maybe_script_execute (struct posix_spawn_args *args) ptrdiff_t argc = args->argc; /* Construct an argument list for the shell. */ - char *new_argv[argc + 1]; + char *new_argv[argc + 2]; new_argv[0] = (char *) _PATH_BSHELL; new_argv[1] = (char *) args->file; if (argc > 1)