On Thu, Jan 27, 2022 at 12:07:24AM +0000, Ariadne Conill wrote:
In several other operating systems, it is a hard requirement that the second argument to execve(2) be the name of a program, thus prohibiting a scenario where argc < 1. POSIX 2017 also recommends this behaviour, but it is not an explicit requirement[0]:
The argument arg0 should point to a filename string that is associated with the process being started by one of the exec functions.To ensure that execve(2) with argc < 1 is not a useful tool for shellcode to use, we can validate this in do_execveat_common() and fail for this scenario, effectively blocking successful exploitation of CVE-2021-4034 and similar bugs which depend on execve(2) working with argc < 1.
We use -EINVAL for this case, mirroring recent changes to FreeBSD and OpenBSD. -EINVAL is also used by QNX for this, while Solaris uses -EFAULT.
In earlier versions of the patch, it was proposed that we create a fake argv for applications to use when argc < 1, but it was concluded that it would be better to just fail the execve(2) in these cases, as launching a process with an empty or NULL argv[0] was likely to just cause more problems.
Let's do it and see what breaks. :)
I do see at least tools/testing/selftests/exec/recursion-depth.c will need a fix. And maybe testcases/kernel/syscalls/execveat/execveat.h in LTP.
Acked-by: Kees Cook keescook@chromium.org
Interestingly, Michael Kerrisk opened an issue about this in 2008[1], but there was no consensus to support fixing this issue then. Hopefully now that CVE-2021-4034 shows practical exploitative use[2] of this bug in a shellcode, we can reconsider.
This issue is being tracked in the KSPP issue tracker[3].
There are a few[4][5] minor edge cases (primarily in test suites) that are caught by this, but we plan to work with the projects to fix those edge cases.
Changes from v2:
- Switch to using -EINVAL as the error code for this.
- Use pr_warn_once() to warn when an execve(2) is rejected due to NULL argv.
Changes from v1:
- Rework commit message significantly.
- Make the argv[0] check explicit rather than hijacking the error-check for count().
Reported-by: Michael Kerrisk mtk.manpages@gmail.com To: Andrew Morton akpm@linux-foundation.org Cc: Matthew Wilcox willy@infradead.org Cc: Christian Brauner brauner@kernel.org Cc: Rich Felker dalias@libc.org Cc: Eric Biederman ebiederm@xmission.com Cc: Alexander Viro viro@zeniv.linux.org.uk Cc: Kees Cook keescook@chromium.org Cc: linux-fsdevel@vger.kernel.org Cc: linux-mm@kvack.org Cc: stable@vger.kernel.org Signed-off-by: Ariadne Conill ariadne@dereferenced.org
fs/exec.c | 4 ++++ 1 file changed, 4 insertions(+)
diff --git a/fs/exec.c b/fs/exec.c index 79f2c9483302..982730cfe3b8 100644 --- a/fs/exec.c +++ b/fs/exec.c @@ -1897,6 +1897,10 @@ static int do_execveat_common(int fd, struct filename *filename, } retval = count(argv, MAX_ARG_STRINGS);
- if (retval == 0) {
pr_warn_once("Attempted to run process '%s' with NULL argv\n", bprm->filename);retval = -EINVAL;- } if (retval < 0) goto out_free; bprm->argc = retval;
-- 2.34.1