Work at SourceForge, help us to make it a better place! We have an immediate need for a Support Technician in our San Francisco or Denver office.

Close

Git Log


Commit Date  
[7bc2bb] (20110606) by Caspar Zhang Caspar Zhang , pushed by Cyril Hrubis Cyril Hrubis

mm: thp testcase come from CVE reproducer

This is a reproducer of CVE-2011-0999, which fixed by mainline commit
a7d6e4ecdb7648478ddec76d30d87d03d6e22b31:

"Transparent hugepages can only be created if rmap is fully functional.
So we must prevent hugepages to be created while
is_vma_temporary_stack() is true."

When running in a loop, it can trigger panic like this, if kernel
unpatched:

kernel BUG at mm/huge_memory.c:1260!
invalid opcode: 0000 [#1] SMP
last sysfs file:
/sys/devices/system/cpu/cpu23/cache/index2/shared_cpu_map
....

So I recommend to run it as 'thp01 -I xxx'.

Signed-off-by: Pingtian Han <phan@redhat.com>
Signed-off-by: Caspar Zhang <czhang@redhat.com>

2011-05-31 15:34:04 Tree
[a1cb1b] by Peter W Morreale Peter W Morreale , pushed by Cyril Hrubis Cyril Hrubis

Fix getschedparam 1-3 test.

This test fails when attempting to set the priority in an thread
attributes because of out-of-order operations on setting the attributes.

The sched policy must be set first in the attributes, then you can set a
prirotiy. This is because the default policy is SCHED_OTHER, which does
not have a priority associated with it. So move the setting of the
policy before attempting to set the priority.

Signed-off-by: Peter W. Morreale <pmorreale@novell.com>

2011-05-31 14:11:20 Tree
[2f879f] by Peter W Morreale Peter W Morreale , pushed by Cyril Hrubis Cyril Hrubis

Fix coding style.

Fix the coding style for the pthread_attr_setschedpolicy() 4-1 test.

Signed-off-by: Peter W. Morreale <pmorreale@novell.com>

2011-05-31 14:01:34 Tree
[1823a2] by Peter W Morreale Peter W Morreale , pushed by Cyril Hrubis Cyril Hrubis

Change 5-1 of pthread_attr_setscope().

Modify the 5-1 test of pthread_attr_setscope to test for unsupported
scoping policies. The standard says either or both of
PTHREAD_SCOPE_SYSTEM, and PTHREAD_SCOPE_PROCESS must be supported.

Change the test to reflect that...

Signed-off-by: Peter W. Morreale <pmorreale@novell.com>

2011-05-25 19:40:51 Tree
[d7453e] by Peter W Morreale Peter W Morreale , pushed by Cyril Hrubis Cyril Hrubis

Modify 1-2 of pthread_cond_broadcast()

This test will create 10,000 threads waiting on a single condition
varible. This seems grossly excessive and may well result in failures
on smaller memory machines (even through the code sets the thread stack
to the minimum)

Reset the children to 1,000.

Signed-off-by: Peter W. Morreale <pmorreale@novell.com>

2011-05-25 19:25:07 Tree
[41b098] by Peter W Morreale Peter W Morreale , pushed by Cyril Hrubis Cyril Hrubis

Remove the speculative directory.

The functionality of the code in the speculative directory is now
provided by the 4-1 and 5-1 tests.

Signed-off-by: Peter W. Morreale <pmorreale@novell.com>

2011-05-25 16:05:19 Tree
[12ea6d] by Peter W Morreale Peter W Morreale , pushed by Cyril Hrubis Cyril Hrubis

Rewrite the 5-1 test.

Since we are looking for unsupported policies (ENOTSUP) from
pthread_attr_setschedpolicy(), we should test for the *required* polices
to ensure that they are supported.

Rewrite the test to verify the supported policies.

Signed-off-by: Peter W. Morreale <pmorreale@novell.com>

2011-05-25 16:05:19 Tree
[789758] by Peter W Morreale Peter W Morreale , pushed by Cyril Hrubis Cyril Hrubis

Add mutex test

Assertion 2 states that threads at different scheduling priorities will
obtain a mutex in priority order. Test that.

Signed-off-by: Peter W. Morreale <pmorreale@novell.com>

2011-05-25 16:05:19 Tree
[3d3b44] by Peter W Morreale Peter W Morreale , pushed by Cyril Hrubis Cyril Hrubis

Rewrite the policy test.

The current policy test (1-1.c) is inadquate and returns a false positive.
The deficiencies of the test are:

o fails to set the inheritance policy within the attributes, so the
policy setting is completely ignored during thread creation.
o fails to set a priority in the attributes.
o fails to verify the created thread's policy and priority.
o only attempted to test the SCHED_FIFO policy.

Fix this test by creating 3 new test src files to test SCHED_FIFO,
SCHED_RR, and SCHED_OTHER. Since the bulk of the test code is identical,
place the bulk of the code into a common file that is included in the three
individual test cases.

The tests for SCHED_FIFO and SCHED_RR also set and verify a priority.

Also normalize the error messages as well as make them more informative.

Signed-off-by: Peter W. Morreale <pmorreale@novell.com>

2011-05-25 16:05:19 Tree
[800540] by Peng Haitao Peng Haitao , pushed by Cyril Hrubis Cyril Hrubis

ksm cases should call tst_kvercmp()

ksm is added to the kernel in 2.6.32, so the cases should call tst_kvercmp().

Signed-off-by: Peng Haitao <penght@cn.fujitsu.com>

2011-05-24 13:40:15 Tree
[ab2bc5] by Garrett Cooper Garrett Cooper

Remove the last vestige of the selinux testsuite.

Leave the selinux .m4 files hanging around for the proc01 testcase.

Signed-off-by: Garrett Cooper <yanegomi@gmail.com>

2011-05-23 19:29:23 Tree
[0ed440] by Garrett Cooper Garrett Cooper

Remove the selinux testsuite from LTP.

Please see
http://www.mail-archive.com/ltp-list@lists.sourceforge.net/msg12241.html
for more discussion on this topic.

Signed-off-by: Garrett Cooper <yanegomi@gmail.com>

2011-05-23 19:25:59 Tree
[640435] by Garrett Cooper Garrett Cooper

Some testcases are not subsystem specific, So there's no need
to mount all subsystems at "cgroup/". Furthre more, if some
subsystem is already mounted, it would fail if we're trying
to mount all subsystems at a mountpoint.
This patch creates a hierarchy with no bound subsystems.

Signed-off-by: Gui Jianfeng <guijianfeng@cn.fujitsu.com>

2011-05-23 19:22:51 Tree
[de9230] by Caspar Zhang Caspar Zhang , pushed by Cyril Hrubis Cyril Hrubis

chown03: fix EPERM issue

in setup phase of chown03 testcase, it set effective gid/uid before
create tmpdir, this might cause EPERM error under some circumstance:

chown03 1 TBROK : chown(/tmp/ltp-1FYYTiULQl/chos2UBh3, -1, 0) failed: errno=EPERM(1): Operation not permitted
chown03 2 TBROK : Remaining cases broken
chown03 0 TWARN : tst_brk: Invalid Type: 4. Using TBROK
chown03 3 TBROK : tmpdir_cleanup: no user cleanup function called before exiting

This patch creates tmpdir first, then set the effective gid/uid to test.

Signed-off-by: Caspar Zhang <czhang@redhat.com>

2011-05-20 12:39:18 Tree
[346aab] by Caspar Zhang Caspar Zhang , pushed by Cyril Hrubis Cyril Hrubis

ksm: fix pagesize

ksm test fails on ppc64 systems because of the pagesize on ppc64 is
larger than 4KB, so page_sharing will be a different value. This patch
replace the hard-coded number 256 to a variable: 1MB/pagesize.

Signed-off-by: Caspar Zhang <czhang@redhat.com>

2011-05-19 17:01:28 Tree
[e6bb54] by Caspar Zhang Caspar Zhang , pushed by Cyril Hrubis Cyril Hrubis

ksm02,04: fix typo

Signed-off-by: Caspar Zhang <czhang@redhat.com>

2011-05-19 15:20:32 Tree
[61ac0e] by Cyril Hrubis Cyril Hrubis

Add newline to include/cleanup.c to stop GCC from whining.

Signed-off-by: Cyril Hrubis <chrubis@suse.cz>

2011-05-11 15:08:32 Tree
[9ef186] by Cyril Hrubis Cyril Hrubis

Add attribute noreturn for tst_brkm() too.

Signed-off-by: Cyril Hrubis <chrubis@suse.cz>

2011-05-11 11:13:12 Tree
[22faba] by Caspar Zhang Caspar Zhang , pushed by Cyril Hrubis Cyril Hrubis

remove unused libmm

As discussed in previous thread, some tests under mem/ links against
libmm which seems not necessary. So I removed libmm and the links to
clean up.

Signed-off-by: Caspar Zhang <czhang@redhat.com>

2011-05-06 12:02:01 Tree
[6b0bf5] by Cyril Hrubis Cyril Hrubis

Fix mq_getattr/speculative/7-1.c

* Cleaned up the coding style

* Make the test return PTS_UNRESOLVED when mq_getattr() succedes
to be consistent with the rest of "may fail" testcases.

* Do a proper unlink for the message queue.

If the message queue is not cleaned up, it remains in the system and
causes other mq_open testcases fail (namely mq_open_speculative/26-1.c).

In this case even the testcase itself was failing after being run more
than maximal number of possible POSIX message queues in system.

Signed-off-by: Cyril Hrubis <chrubis@suse.cz>

2011-05-05 15:37:23 Tree
[afcdf0] by Cyril Hrubis Cyril Hrubis

Merge ../ltp-dev

2011-05-04 17:17:11 Tree
[52a1bb] by Cyril Hrubis Cyril Hrubis

Fix fork/19-1.c

* Cleaned up the coding style and get rid of the testfrmw

* Do a proper unlink for the message queue

If the message queue is not cleaned up, it remains in the system and
causes other mq_open testcases fail (namely mq_open_speculative/26-1.c).

Signed-off-by: Cyril Hrubis <chrubis@suse.cz>

2011-05-04 17:09:13 Tree
[e93579] by Garrett Cooper Garrett Cooper

Fix race condition in kill05.

There is a race condition when starting child. Master child can go
forward before child changes its reuid.
Fix this by waiting until child is ready.

Signed-off-by: Jan Stancek <jstancek@redhat.com>

2011-05-04 16:57:34 Tree
[203573] by Garrett Cooper Garrett Cooper

Fix typo.

a typo fix, forgot NULL in tst_brkm. And seems tst_resm+tst_exit could
be replaced by tst_brkm().

Signed-off-by: Caspar Zhang <czhang@redhat.com>

2011-05-04 16:55:59 Tree
[21e14b] by Garrett Cooper Garrett Cooper

Fix out-of-build tree srcdir.

For out of source tree build, realtime_config.h is created under the
top_builddir. Add it into the include path or you will get a missing
file issue.

Signed-off-by: Andrew Chen <ypchen@faraday-tech.com>

2011-05-04 16:52:35 Tree
Older >