November 22, 2006

The Linux Test project ltp-20061121 Released

Michael Reed writes "The Linux Test Project test suite has been released. The latest version of the testsuite contains 2900+ tests for the Linux OS. Our web site also contains other information such as:
  • A Linux test tools matrix
  • Technical papers
  • How To's on Linux testing
  • Code coverage analysis tool.

We encourage the community to post results to,
and patches, new tests, or comments/questions to

Release Highlights:
Code Cleanups by Darrick Wong, Mike Frysinger, Jacky Malcles, Jane Lv
Michael Reed

See ChangeLog Below

  • Removed an invalid errno value check.
  • Fixed the setting of LTPROOT for the bin directory, not just the
  • descend into timers subdir when cleaning
  • This patch amends pounder to build kernel 2.6.18 instead of 2.6.15.
  • dynamically calculate TST_TOTAL rather than hardcoding it
    . shmctl01.c, shmctl02.c, shmctl03.c
  • Fix from Jane Lv to move large arrays off the stack and into global
  • This patch makes the pounder control program handle SIGHUP by killing children
  • The readme file has been updated for nfs
  • Disable this test case for kernels below 2.6.17 and not 2.6.16.
  • This is a patch that allows pounder to record the contents of sysfs when it
    starts, thus enabling us to take a better snapshot of the system state when
    pounder starts. It also adds "lspci" into the PCI Device Tree heading per
    user request.
  • Since 2.6.17, _syscall* isn't exported into the user space any more, so all
    the test cases using _syscall* will fail to be compiled This patch
    remove _syscall5 from llseek.c and convert to syscall provided by libc.It
    also includes some cleanups and makes the build cleaner and removes the
    warning for
    scsimain.c, llseek.c, scsimain.c
  • Backed out the "fix" in Bug 1592647, b/c I don't think the testcase is
    actually broken.
  • touchup code a little to make the expected behavior of test4 more obvious
  • The random syscall generator has a bug on x86-64 where it only generates
    random bits for the upper 32 bits of the syscall argument, and then truncates
    the upper 32 bits anyway! This patch replaces the rand() weirdness with a
    function that generates N bytes of random data instead. It also adds a -z
    flag in case one *wants* the zero-arguments behavior; Max Asbock and Russ
    Weight requested it because zero is a quick way to find kernel code paths
    that don't check userland pointers.
    CHANGELOG, randasys.c, random_syscall
  • Applied fix for Bugs item #1592647
    . diotest4.c
  • A patch by Jacky Malcles (The IA-64 architecture sets SHMLBA to 1MB and) The
    value of the pointer address passed to shmat being set in such a way that
    rounding down to the nearest multiple of SHMLBA results above the current
    heap. shmt09.c
  • Made a change to a comment above the change for 1.11 mc_cmds
  • Disabling the ping test for now. The test can pass without this additional
    check and can cause failures on some distros.

  • A patch by Suzuki Kp, that removed the definition of HZ from user space for
    Bug # 28023.

  • Comment out set -x to remove debug output




  • Linux
Click Here!