linux 4.14.0-12.14 ADT test failure with linux 4.14.0-12.14

  ======================================================================
  FAIL: test_02_stack_limit (__main__.KernelASLRCollisionsTest)
  Process stack is normal size
  ----------------------------------------------------------------------
  Traceback (most recent call last):
    File "./test-kernel-aslr-collisions.py", line 69, in test_02_stack_limit
      self.assertEqual(-1, stack[1])
  AssertionError: -1 != 8388608

  ======================================================================
  FAIL: test_stack_collision (__main__.KernelASLRCollisionsTest)
  Check if stack crashes into mmap in 100,000 execs (amd64 only?) (LP: #504164)
  ----------------------------------------------------------------------
  Traceback (most recent call last):
    File "./test-kernel-aslr-collisions.py", line 88, in test_stack_collision
      self.assertShellExitEquals(0, ['./aslr-crash'])
    File "/tmp/autopkgtest.PPgnXF/build.Dmb/src/autotest/client/tmp/ubuntu_qrt_kernel_aslr_collisions/src/qa-regression-testing/scripts/testlib.py", line 1134, in assertShellExitEquals
      self.assertEqual(expected, rc, msg + result + report)
  AssertionError: Got exit code 1, expected 0
  Command: './aslr-crash'
  Output:
  Failed to set stack rlimit:: Invalid argument

This looks to be due to a change in behavior regarding the hard RLIMIT_STACK value on setuid. Previously the hard limit of RLIM_INFINITY would have been preserved across the setuid exec, now it has been limited so that values greater than 8MB (including RLIM_INFINITY) are clamped down to that value. This is no doubt due to 04e35f4495dd "exec: avoid RLIMIT_STACK races with prlimit()." The test expects the hard limit to always be RLIM_INIFINITY.

The patch was reverted upstream, so I'll do the same in bionic.