77 lines
2.1 KiB
Plaintext
77 lines
2.1 KiB
Plaintext
==============================================================
|
|
Test Suite for the Control Group Freezer and Signal Subsystems
|
|
==============================================================
|
|
Author: Matt Helsley
|
|
Date: June, 2008
|
|
Last update: January 19th, 2009
|
|
|
|
About:
|
|
------
|
|
These tests exercise the freezer and signal subsystems (aka controllers)
|
|
available since kernel 2.6.28 or later.
|
|
|
|
These tests exercise the freezer control group controller. This controller is
|
|
a useful for implementing container checkpointing or if you wish to gradually
|
|
freeze tasks in cgroups before doing a suspend.
|
|
|
|
The common case is embodied in the freeze_thaw.sh test file. However,
|
|
the scripts are also designed to test a few obscure cases such as:
|
|
freeze self -- freeze the control group that the current process is
|
|
in
|
|
freeze cancellation -- freeze but then backout of the freeze processing
|
|
and issue a "thaw" command before reaching the frozen state
|
|
signalling freezing/frozen tasks -- this should delay delivery of the
|
|
signal until the task is unfrozen
|
|
vfork -- The vfork system call introduces an obscure special case for
|
|
the freezer kernel code. This test case attempts to trigger any
|
|
unanticipated corner cases involving vfork.
|
|
|
|
|
|
NOTES:
|
|
|
|
Using the cgroup freezer prevents the tasks from being automatically
|
|
woken up upon a "Power Management Resume". You will need to
|
|
manually unfreeze cgroups.
|
|
|
|
These tests assume that most of the library functions in
|
|
libcgroup_subsys take much less than $sample_sleep time to execute. If
|
|
this assumption is incorrect then many of these tests may fail in
|
|
unusual ways. So make sure $sample_sleep is a "reasonable" amount of
|
|
time.
|
|
|
|
Building:
|
|
---------
|
|
|
|
Run the command line:
|
|
make all
|
|
|
|
|
|
Installing:
|
|
-----------
|
|
|
|
Run the command line:
|
|
|
|
|
|
Cleaning:
|
|
---------
|
|
|
|
Run the command line:
|
|
|
|
make clean
|
|
|
|
|
|
Running:
|
|
--------
|
|
|
|
Run the command line:
|
|
|
|
./run.sh
|
|
|
|
Results:
|
|
--------
|
|
|
|
The results should show up as PASS/FAIL in the LTP logs for each script. The
|
|
output frequently will be more specific (beyond which test script failed)
|
|
about where and what failure occurred.
|
|
|