canonical-ubuntu-qa team mailing list archive

[Jürg Häfliger <1869134@xxxxxxxxxxxxxxxxxx>]

** Tags added: kernel-daily-bug

-- 
You received this bug notification because you are a member of Canonical
Platform QA Team, which is subscribed to ubuntu-kernel-tests.
https://bugs.launchpad.net/bugs/1869134

Title:
  tcp_fastopen_backup_key.sh from net in ubuntu_kernel_selftests failed
  on Eoan LPAR

Status in ubuntu-kernel-tests:
  Fix Released
Status in linux package in Ubuntu:
  Fix Released
Status in linux source package in Focal:
  Fix Released
Status in linux source package in Groovy:
  Fix Released

Bug description:
  == SRU Justification Focal, Groovy ==

  Running the tcp_fastopen_backup_key.sh from net in
  ubuntu_kernel_selftests fails on big endian systems. This is a
  regression that occurred because of commit 438ac88009bc ("net:
  fastopen: robustness and endianness fixes for SipHash").

  The test rotates TFO keys for ipv4/ipv6. It demonstrate how the
  primary and backup TFO keys can be rotated while minimizing the number
  of client cookies that are rejected.  Due to an endianess issue, big
  endian systems were getting endian reversed values, so the values
  being from /proc/sys/net/ipv4/tcp_fastopen_key were reversed.

  /proc/sys/net/ipv4/tcp_fastopen_key

  == Fix ==

  Upstream (linux-next) fix:

  commit f19008e676366c44e9241af57f331b6c6edf9552
  Author: Jason Baron <jbaron@xxxxxxxxxx>
  Date:   Mon Aug 10 13:38:39 2020 -0400

      tcp: correct read of TFO keys on big endian systems

  == Test case ==

  Run the kernel self tests tcp_fastopen_backup_key.sh.  Currently this
  generally fails on the last few test cases.  With the fix the test
  passes.

  == Regression Potential ==

  This fix touches the reading of the tcp fast open key proc interface
  /proc/sys/net/ipv4/tcp_fastopen_key and the getsockopt
  TCP_FASTOPEN_KEY interface so there is a potential for a regression
  when accessing data from these read-only interfaces.

  There is some data copying to keys using unaligned puts and memcpy()
  so there is a potential that we may have data bounds errors causing
  memory corruption.

  ----

  Issue found on Eoan (5.3.0-42.34, 5.3.0-43.36) with all s390x instances:
    - LPAR s2lp4
    - zVM kernel03
    - zKVM s2lp6g003

   # selftests: net: tcp_fastopen_backup_key.sh
   # PASS
   # PASS
   # PASS
   # PASS
   # PASS
   # PASS
   # PASS
   # PASS
   # PASS
   # FAIL: TcpExtTCPFastOpenPassiveFail non-zero
   not ok 30 selftests: net: tcp_fastopen_backup_key.sh # exit=1

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-kernel-tests/+bug/1869134/+subscriptions