Re: Re: [PATCH 5.10 00/19] 5.10.59-rc1 review

Re: Re: [PATCH 5.10 00/19] 5.10.59-rc1 review

[Tim Lewis]

Is there a reason for skipping the selftests/netfilter tests?

eg;
https://qa-reports.linaro.org/lkft/linux-stable-rc-linux-5.10.y/build/v5.10.57-155-ged2493daa915/testrun/5442487/suite/kselftest-net/test/netfilter:nft_trans_stress.sh/details/

I have not double checked (n=1) but it looks like this patch is the
first for that test to hang on for 13 hours. The state just before I
killed the test;

ps ax --forest
...
   2013 pts/2    Ss     0:00  \_ -bash
   2025 pts/2    S+     0:00      \_ bash scripts/kernel_test_diff.sh
   2064 pts/2    S+     0:00          \_ make SKIP_TARGETS=ptrace rtc
breakpoints kselftest
   2285 pts/2    S+     0:00              \_ make -C
./tools/testing/selftests run_tests
   2953 pts/2    S+     0:00                  \_ /bin/sh -c for TARGET
in android arm64 bpf capabilities cgroup clone3 core cpufreq
cpu-hotplug drivers/dma-buf efivarfs exec filesystems
filesystems/binde
  13616 pts/2    S+     0:00                      \_ make
OUTPUT=/media/btrfs/linux/tools/testing/selftests/netfilter -C
netfilter run_tests
  13617 pts/2    S+     0:00                          \_ /bin/sh -c
BASE_DIR="/media/btrfs/linux/tools/testing/selftests"; .
/media/btrfs/linux/tools/testing/selftests/kselftest/runner.sh; if [
"X" != "X
  13624 pts/2    S+     0:00                              \_ /bin/sh
-c BASE_DIR="/media/btrfs/linux/tools/testing/selftests"; .
/media/btrfs/linux/tools/testing/selftests/kselftest/runner.sh; if [
"X" !
  13625 pts/2    S+     0:00                                  \_
/bin/sh -c BASE_DIR="/media/btrfs/linux/tools/testing/selftests"; .
/media/btrfs/linux/tools/testing/selftests/kselftest/runner.sh; if [ "
  13626 pts/2    S+     0:00                                      \_
/bin/sh -c BASE_DIR="/media/btrfs/linux/tools/testing/selftests"; .
/media/btrfs/linux/tools/testing/selftests/kselftest/runner.sh; if
  13629 pts/2    S+     0:00
\_ /bin/sh -c BASE_DIR="/media/btrfs/linux/tools/testing/selftests"; .
/media/btrfs/linux/tools/testing/selftests/kselftest/runner.sh
  13631 pts/2    S+     0:00
   \_ perl /media/btrfs/linux/tools/testing/selftests/kselftest/prefix.pl
  16853 pts/2    R+   188:32 ping -4 127.0.0.1 -fq
  16854 pts/2    R+   188:32 ping -6 ::1 -fq
  16856 pts/2    R+   189:36 ping -4 127.0.0.1 -fq
  16857 pts/2    R+   189:36 ping -6 ::1 -fq
  16859 pts/2    R+   189:24 ping -4 127.0.0.1 -fq
  16860 pts/2    R+   189:24 ping -6 ::1 -fq
  16862 pts/2    R+   189:22 ping -4 127.0.0.1 -fq
  16863 pts/2    R+   189:22 ping -6 ::1 -fq




dmesg
...
[  248.609199] BUG: scheduling while atomic: cpuhp/1/15/0x00000002
...
[  248.609658] Call trace:
[  248.609670]  dump_backtrace+0x0/0x1a8
[  248.609675]  show_stack+0x18/0x68
[  248.609681]  dump_stack+0xd0/0x12c
[  248.609687]  __schedule_bug+0x60/0x78
[  248.609692]  __schedule+0x5d8/0x650
[  248.609696]  schedule+0x70/0x108
[  248.609702]  schedule_timeout+0x228/0x288
[  248.609707]  wait_for_completion_killable+0x88/0x170
[  248.609712]  __kthread_create_on_node+0xbc/0x1a8
[  248.609716]  kthread_create_on_node+0x58/0x80
[  248.609723]  spl_kthread_create+0x9c/0x110
[  248.609727]  taskq_thread_create+0x64/0x108
[  248.609732]  spl_taskq_expand+0xc4/0xf0
[  248.609738]  cpuhp_invoke_callback+0x180/0x210
[  248.609743]  cpuhp_thread_fun+0xd8/0x170
[  248.609747]  smpboot_thread_fn+0x1c4/0x280
[  248.609752]  kthread+0x140/0x160
[  248.609756]  ret_from_fork+0x10/0x34
...
[  385.330256] IPv6: ADDRCONF(NETDEV_CHANGE): ns2eth3: link becomes ready

Re: Re: [PATCH 5.10 00/19] 5.10.59-rc1 review

[Naresh Kamboju]

Hi Tim,

On Sun, 15 Aug 2021 at 08:41, Tim Lewis <elatllat@gmail.com> wrote:
>
> Is there a reason for skipping the selftests/netfilter tests?
>
> eg;
> https://qa-reports.linaro.org/lkft/linux-stable-rc-linux-5.10.y/build/v5.10.57-155-ged2493daa915/testrun/5442487/suite/kselftest-net/test/netfilter:nft_trans_stress.sh/details/
>
> I have not double checked (n=1) but it looks like this patch is the
> first for that test to hang on for 13 hours. The state just before I
> killed the test;

I have seen a set of test cases hang so I have skipped them.

x86_64 is using NFS root file system the net and netfilter tests performing
network interfaces up and down causing network failure and system hang.

The reasons for failure are listed in this document.
https://github.com/Linaro/test-definitions/blob/master/automated/linux/kselftest/skipfile-lkft.yaml

- Naresh