Re: [PATCH] fs/exec: require argv[0] presence in do_execveat_common()

["Eric W. Biederman" <ebiederm@xmission.com>]

Kees Cook <keescook@chromium.org> writes:

> On Wed, Jan 26, 2022 at 12:33:39PM +0000, Heikki Kallasjoki wrote:
>> On Wed, Jan 26, 2022 at 05:18:58AM -0600, Ariadne Conill wrote:
>> > On Tue, 25 Jan 2022, Kees Cook wrote:
>> > > Lots of stuff likes to do:
>> > > execve(path, NULL, NULL);
>> > 
>> > I looked at these, and these seem to basically be lazily-written test cases
>> > which should be fixed.  I didn't see any example of real-world applications
>> > doing this.  As noted in some of the test cases, there are comments like
>> > "Solaris doesn't support this," etc.
>> 
>> See also the (small) handful of instances of `execlp(cmd, NULL);` out
>> there, which I imagine would start to fail:
>> https://codesearch.debian.net/search?q=execlp%3F%5Cs*%5C%28%5B%5E%2C%5D%2B%2C%5Cs*NULL&literal=0
>> 
>> Two of the hits (ispell, nauty) would seem to be non-test use cases.
>
> Ah yeah, I've added this to the Issue tracker:
> https://github.com/KSPP/linux/issues/176

I just took a slightly deeper look at these.

There are two patterns found by that search.

-  execlp("/proc/self/exec", NULL)

    Which in both both the proot and care packages is a testt that
    deliberately loops and checks to see if it can generate "argc == 0".

    That is the case where changing argc == 0 into { "", NULL }
    will loop forever.

    For that test failing to exec the "argc == 0" will cause a test
    failure but for a security issue that seems a reasonable thing
    to do to a test.

- execlp(MACRO, NULL)

    The macro happens to contain commas so what looks via inspection
    will generate an application run with "argc == 0" actually
    does not.

Eric