[F/OEM-5.6][SRU][PATCH 0/1] selftests/net: bump timeout to 5 minutes

Po-Hsu Lin po-hsu.lin at canonical.com
Thu May 13 08:39:36 UTC 2021


When first submitting this for Bionic [1], I forgot to check if
F/OEM-5.6 need this. (We have the SAUCE patch for F/OEM-5.10)

[1] https://lists.ubuntu.com/archives/kernel-team/2021-May/119865.html

[Impact]
These 3 tests will fail with timeout error when running the whole
"net" test in ubuntu_kernel_selftests:
  * not ok 12 selftests: net: xfrm_policy.sh # TIMEOUT
  * not ok 16 selftests: net: pmtu.sh # TIMEOUT
  * not ok 19 selftests: net: udpgso_bench.sh # TIMEOUT

They will pass if you run them manually. This is because of the
default 45 seconds timeout in kselftest framework.

A quick test shows these tests will take about:
  xfrm_policy.sh - 2m19.690s
  pmtu.sh - 3m6.832s
  udpgso_bench.sh - 0m57.985s

[Fix]
* b881d089c7c9c7 ("selftests/net: bump timeout to 5 minutes")

We have commit 852c8cbf34d3b3 ("selftests/kselftest/runner.sh: Add 45
second timeout per test") for default timeout added since Bionic.

However there is a SAUCE patch ("UBUNTU: SAUCE: selftests/net --
disable timeout") to disable timeout for the net test in newer
releases. I think we can leave it as-is for the moment unless some
test is hanging too long because of that.

Therefore Only Bionic needs this patch, the patch can be applied with
some context adjustment.

We have the SAUCE patch for F/OEM-5.10 as well, this patch can be
cherry-picked for F/OEM-5.6

[Test]
With this patch applied, these tests should have a chance to finish.

[Where problems could occur]
The fix is just for testing tool, no actual impact to real kernel
functions. If this 5 minutes timeout is not enough, we might still
seeing this kind of failures in the test report.

Po-Hsu Lin (1):
  selftests/net: bump timeout to 5 minutes

 tools/testing/selftests/net/Makefile | 2 ++
 tools/testing/selftests/net/settings | 1 +
 2 files changed, 3 insertions(+)
 create mode 100644 tools/testing/selftests/net/settings

-- 
2.7.4




More information about the kernel-team mailing list