Re: [selftests/bpf] 69d96519db: kernel_selftests.bpf.test_socket_cookie.fail

[Stanislav Fomichev]

On 06/21, Andrii Nakryiko wrote:
> )
> 
> On Fri, Jun 21, 2019 at 9:11 AM Stanislav Fomichev <sdf@xxxxxxxxxxx> wrote:
> >
> > On 06/21, kernel test robot wrote:
> > > FYI, we noticed the following commit (built with gcc-7):
> > >
> > > commit: 69d96519dbf0bfa1868dc8597d4b9b2cdeb009d7 ("selftests/bpf: convert socket_cookie test to sk storage")
> > > https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master
> > >
> > > in testcase: kernel_selftests
> > > with following parameters:
> > >
> > >       group: kselftests-00
> > >
> > > test-description: The kernel contains a set of "self tests" under the tools/testing/selftests/ directory. These are intended to be small unit tests to exercise individual code paths in the kernel.
> > > test-url: https://www.kernel.org/doc/Documentation/kselftest.txt
> > >
> > >
> > > on test machine: qemu-system-x86_64 -enable-kvm -cpu SandyBridge -smp 2 -m 4G
> > >
> > > caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace):
> > >
> > >
> > > If you fix the issue, kindly add following tag
> > > Reported-by: kernel test robot <rong.a.chen@xxxxxxxxx>
> > >
> > > # selftests: bpf: test_socket_cookie
> > > # libbpf: failed to create map (name: 'socket_cookies'): Invalid
> > > # argument
> > Another case of old clang trying to create a map that depends on BTF?
> > Should we maybe switch those BTF checks in the kernel to return
> > EOPNOTSUPP to make it easy to diagnose?
> 
> For older compilers that don't generate DATASEC/VAR, you'll see a clear message:
> 
> libbpf: DATASEC '.maps' not found.
> 
> So this must be something else. I just confirmed with clang version
> 7.0.20180201 that for ./test_socket_cookie that's the first line
> that's emitted on failure.
Thanks for checking, I also took a look at the attached kernel_selftests.xz,
here is what it has:
2019-06-21 11:58:35 ln -sf /usr/bin/clang-6.0 /usr/bin/clang
2019-06-21 11:58:35 ln -sf /usr/bin/llc-6.0 /usr/bin/llc
...
# BTF libbpf test[1] (test_btf_haskv.o): SKIP. No ELF .BTF found
# BTF libbpf test[2] (test_btf_nokv.o): SKIP. No ELF .BTF found
...
# Test case #0 (btf_dump_test_case_syntax): test_btf_dump_case:71:FAIL
# failed to load test BTF: -2
# Test case #1 (btf_dump_test_case_ordering): test_btf_dump_case:71:FAIL
# failed to load test BTF: -2
...

And so on. So there is clearly an old clang that doesn't emit any
BTF. And I also don't see your recent abd29c931459 before 69d96519dbf0 in
linux-next, that's why it doesn't complain about missing/corrupt BTF.

We need to convince lkp people to upgrade clang, otherwise, I suppose,
we'll get more of these reportings after your recent df0b77925982 :-(

> > > # libbpf: failed to load object './socket_cookie_prog.o'
> > > # (test_socket_cookie.c:149: errno: Invalid argument) Failed to load
> > > # ./socket_cookie_prog.o
> > > # FAILED
> > > not ok 15 selftests: bpf: test_socket_cookie
> > >
> > >
> > >
> > >
> > > To reproduce:
> > >
> > >         # build kernel
> > >       cd linux
> > >       cp config-5.2.0-rc2-00598-g69d9651 .config
> > >       make HOSTCC=gcc-7 CC=gcc-7 ARCH=x86_64 olddefconfig
> > >       make HOSTCC=gcc-7 CC=gcc-7 ARCH=x86_64 prepare
> > >       make HOSTCC=gcc-7 CC=gcc-7 ARCH=x86_64 modules_prepare
> > >       make HOSTCC=gcc-7 CC=gcc-7 ARCH=x86_64 SHELL=/bin/bash
> > >       make HOSTCC=gcc-7 CC=gcc-7 ARCH=x86_64 bzImage
> > >
> > >
> > >         git clone https://github.com/intel/lkp-tests.git
> > >         cd lkp-tests
> > >         bin/lkp qemu -k <bzImage> job-script # job-script is attached in this email
> > >
> > >
> > >
> > > Thanks,
> > > Rong Chen
> > >
> >
> 
> <mega snip>