Re: [PATCH 2/2] selftests/exec: add a test for execveat()'s comm

[Tycho Andersen <tycho@tycho.pizza>]

On Wed, Nov 27, 2024 at 02:25:29PM +0000, Mark Brown wrote:
> On Wed, Oct 30, 2024 at 02:37:32PM -0600, Tycho Andersen wrote:
> > From: Tycho Andersen <tandersen@netflix.com>
> > 
> > In the previous patch we've defined a couple behaviors:
> > 
> > 1. execveat(fd, AT_EMPTY_PATH, {"foo"}, ...) should render argv[0] as
> >    /proc/pid/comm
> > 2. execveat(fd, AT_EMPTY_PATH, {NULL}, ...) should keep the old behavior of
> >    rendering the fd as /proc/pid/comm
> > 
> > and just to be sure keeps working with symlinks, which was a concern in
> > [1], I've added a test for that as well.
> > 
> > The test itself is a bit ugly, because the existing check_execveat_fail()
> > helpers use a hardcoded envp and argv, and we want to "pass" things via the
> > environment to test various argument values, but it seemed cleaner than
> > passing one in everywhere in all the existing tests.
> 
> This test doesn't pass in my CI, running on an i.MX8MP Verdin board.
> This is an arm64 system and I'm running the tests on NFS.
> 
> > Output looks like:
> 
> >     ok 51 Check success of execveat(6, 'home/tycho/packages/...yyyyyyyyyyyyyyyyyyyy', 0)...
> >     # Check execveat(AT_EMPTY_PATH)'s comm is sentinel
> >     ok 52 Check success of execveat(9, '', 4096)...
> >     # Check execveat(AT_EMPTY_PATH)'s comm is sentinel
> >     ok 53 Check success of execveat(11, '', 4096)...
> >     # Check execveat(AT_EMPTY_PATH)'s comm is 9
> >     [   25.579272] process 'execveat' launched '/dev/fd/9' with NULL argv: empty string added
> >     ok 54 Check success of execveat(9, '', 4096)...
> 
> The output when things fail is:
> 
> # # Check execveat(AT_EMPTY_PATH)'s comm is sentinel
> # # bad comm, got: 11 expected: sentinel# child 8257 exited with 1 neither 0 nor 0
> # not ok 52 Check success of execveat(11, '', 4096)... 
> # # Check execveat(AT_EMPTY_PATH)'s comm is sentinel
> # # bad comm, got: 13 expected: sentinel# child 8258 exited with 1 neither 0 nor 0
> # not ok 53 Check success of execveat(13, '', 4096)... 
> 
> Full log from a failing job at:
> 
>    https://lava.sirena.org.uk/scheduler/job/993508
> 
> I didn't do any investigation beyond this.

Strange... but this series has been rejected by Linus anyway, so
probably not worth investigating further.

Tycho