[Bug 2081210] Re: pv FTBFS with valgrind 3.23

[Dan Bungert]

** Changed in: pv (Ubuntu)
       Status: New => Fix Committed

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to valgrind in Ubuntu.
https://bugs.launchpad.net/bugs/2081210

Title:
  pv FTBFS with valgrind 3.23

Status in pv package in Ubuntu:
  Fix Committed
Status in valgrind package in Ubuntu:
  New

Bug description:
  pv employs valgrind as part of the test suite which runs at build time.
  As of valgrind 3.23, 3 tests are failing:

  https://launchpadlibrarian.net/749245201/buildlog_ubuntu-oracular-
  amd64.pv_1.8.10-1_BUILDING.txt.gz

  FAIL: tests/Memory_safety_-_Basic
  =================================

  memory check failed - see file `valgrind.out'.
  FAIL tests/Memory_safety_-_Basic.test (exit status: 1)

  FAIL: tests/Memory_safety_-_Remote_control_receiver
  ===================================================

  memory check failed - see file `valgrind.out'.
  FAIL tests/Memory_safety_-_Remote_control_receiver.test (exit status: 1)

  FAIL: tests/Memory_safety_-_Remote_control_sender
  =================================================

  memory check failed - see file `valgrind.out'.
  FAIL tests/Memory_safety_-_Remote_control_sender.test (exit status: 1)

  However these tests aren't failing for memory safety reasons, they are
  failing for errors on open file descriptors that have been inherited.

  valgrind --quiet makes this easy to see:

  ==1907696== FILE DESCRIPTORS: 4 open (2 std) at exit.
  ==1907696== Open file descriptor 9: /dev/pts/6
  ==1907696==    <inherited from parent>
  ==1907696==
  ==1907696== Open file descriptor 3: /tmp/tmp.KBPoCHdMro
  ==1907696==    <inherited from parent>
  ==1907696==
  ==1907696==
  ==1907696== ERROR SUMMARY: 2 errors from 2 contexts (suppressed: 0 from 0)

  I'm not sure where fd9 came from, fd3 is a logging desciptor used with
  valgrind's --log-fd feature.

  Reverting to valgrind 3.22 means the above are no longer errors.

  Valgrind 3.23 has relevant text in the relase notes:
  https://valgrind.org/docs/manual/dist.news.html

  * --track-fds=yes also produces "real" errors now which can be
    suppressed and work with --error-exitcode. When combined with
    --xml the xml-output now also includes FdBadClose and FdNotClosed
    error kinds (see docs/internals/xml-output-protocol5.txt).

  This raises some questions:
  * should an open log-fd be considered an error (likely not, IMO)
  * should arbitrary other inherited fds be considered an error (less certain)

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/pv/+bug/2081210/+subscriptions