Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 20 Sep 2022 15:21:11 -0700
From:   Kees Cook <keescook@chromium.org>
To:     "Eric W. Biederman" <ebiederm@xmission.com>
Cc:     Ren Zhijie <renzhijie2@huawei.com>, viro@zeniv.linux.org.uk,
        linux-mm@kvack.org, linux-fsdevel@vger.kernel.org,
        linux-kernel@vger.kernel.org, tanghui20@huawei.com
Subject: Re: [PATCH] exec: Force binary name when argv is empty
Message-ID: <202209201516.966D6EF@keescook>
References: <20220920120812.231417-1-renzhijie2@huawei.com>
 <87sfkmyumv.fsf@email.froward.int.ebiederm.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <87sfkmyumv.fsf@email.froward.int.ebiederm.org>
Precedence: bulk

On Tue, Sep 20, 2022 at 09:42:48AM -0500, Eric W. Biederman wrote:
> Ren Zhijie <renzhijie2@huawei.com> writes:
> > From: Hui Tang <tanghui20@huawei.com>
> >
> > First run './execv-main execv-child', there is empty in 'COMMAND' column
> > when run 'ps -u'.
> >
> >  USER       PID %CPU %MEM    VSZ   RSS TTY    [...] TIME COMMAND
> >  root       368  0.3  0.0   4388   764 ttyS0        0:00 ./execv-main
> >  root       369  0.6  0.0   4520   812 ttyS0        0:00
> >
> > The program 'execv-main' as follows:
> >
> >  int main(int argc, char **argv)
> >  {
> >    char *execv_argv[] = {NULL};
> >    pid_t pid = fork();
> >
> >    if (pid == 0) {
> >      execv(argv[1], execv_argv);
> >    } else if (pid > 0) {
> >      wait(NULL);
> >    }
> >    return 0;
> >  }

The correct fix is to userspace here:

  int main(int argc, char **argv)
  {
-   char *execv_argv[] = {NULL};
+   char *execv_argv[] = { argv[1], NULL };
    pid_t pid = fork();

    if (pid == 0) {

> [...]
> For a rare case that should essentially never happen why make it
> friendlier to use?  Why not fix userspace to add the friendly name
> instead of the kernel?
> 
> Unless there is a good reason for it, it would be my hope that in
> a couple of years all of the userspace programs that trigger
> the warning when they start up could be fixed, and we could have
> execve start failing in those cases.

Agreed -- the goal is to help userspace fix how execve(2) is called.

Speaking to the proposed patch, this idea was considered during the
development of the ""-adding patch, with the basic outcome being
that creating a _new_ behavior was not a good idea, and might cause more
confusion. You can see the thread here:

https://lore.kernel.org/all/202202021229.9681AD39B0@keescook/

-Kees

-- 
Kees Cook