]> git.ipfire.org Git - thirdparty/libcgroup.git/log
thirdparty/libcgroup.git
5 years ago tests: Add unit tests for cgroup_compare_wildcard_procname()
Tom Hromatka [Wed, 20 Nov 2019 17:09:33 +0000 (17:09 +0000)] 
 tests: Add unit tests for cgroup_compare_wildcard_procname()

This commit adds unit tests for cgroup_compare_wildcard_procname().

The following tests are included in this commit:

WildcardProcnameSimpleMatch - Compares an ignore rule with a
    wildcarded procname with a matching procname.  Expects true.

WildcardProcnameNoMatch - Compares an ignore rule with a wildcarded
    procname with a process that has a non-matching name.  Expects
    false.

ProcnameWildcard_AsteriskNoMatch - Compares a wildcard rule with
    a process name that doesn't match.  Expects false.

ProcnameWildcard_AsteriskMatch - Compares a wildcard rule with
    a process name that does match the rule.  Expects true.

ProcnameWildcard_AsteriskNoMatch2 - Compares a wildcard rule
    with a process name that is shorter than the rule and thus
    does not match.  Expects false.

ProcnameWildcard_AsteriskMatchExactly - Compares a wildcard
    rule with a process name that matches the rule exactly.
    Expects true.

ProcnameWildcard_NoAsteriskMatchExactly - Compares a rule
    with no wildcards with a process name that exactly matches
    the rule.  Expects false as the function exits early due
    to no asterisk in the rule.

The results from googletest are reported below:

[----------] 13 tests from CgroupCompareIgnoreRuleTest
<snip>
[ RUN      ] CgroupCompareIgnoreRuleTest.WildcardProcnameSimpleMatch
[       OK ] CgroupCompareIgnoreRuleTest.WildcardProcnameSimpleMatch (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.WildcardProcnameNoMatch
[       OK ] CgroupCompareIgnoreRuleTest.WildcardProcnameNoMatch (0 ms)
[----------] 13 tests from CgroupCompareIgnoreRuleTest (1 ms total)

[----------] 7 tests from ProcnameWildcardTest
[ RUN      ] ProcnameWildcardTest.ProcnameWildcard_NoAsterisk
[       OK ] ProcnameWildcardTest.ProcnameWildcard_NoAsterisk (0 ms)
[ RUN      ] ProcnameWildcardTest.ProcnameWildcard_AsteriskNoMatch
[       OK ] ProcnameWildcardTest.ProcnameWildcard_AsteriskNoMatch (0 ms)
[ RUN      ] ProcnameWildcardTest.ProcnameWildcard_AsteriskMatch
[       OK ] ProcnameWildcardTest.ProcnameWildcard_AsteriskMatch (0 ms)
[ RUN      ] ProcnameWildcardTest.ProcnameWildcard_AsteriskNoMatch2
[       OK ] ProcnameWildcardTest.ProcnameWildcard_AsteriskNoMatch2 (0 ms)
[ RUN      ] ProcnameWildcardTest.ProcnameWildcard_AsteriskMatchExactly
[       OK ] ProcnameWildcardTest.ProcnameWildcard_AsteriskMatchExactly (0 ms)
[ RUN      ] ProcnameWildcardTest.ProcnameWildcard_NoAsteriskMatchExactly
[       OK ] ProcnameWildcardTest.ProcnameWildcard_NoAsteriskMatchExactly (0 ms)
[ RUN      ] ProcnameWildcardTest.ProcnameWildcard_AsteriskFirstChar
[       OK ] ProcnameWildcardTest.ProcnameWildcard_AsteriskFirstChar (0 ms)
[----------] 7 tests from ProcnameWildcardTest (0 ms total)

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agocgrulesengd: Add wildcard matching for process names
Tom Hromatka [Mon, 4 Nov 2019 23:23:22 +0000 (23:23 +0000)] 
cgrulesengd: Add wildcard matching for process names

This commit adds wildcard matching to process name
parsing in cgrulesengd.  Note that wildcard matching
works for standard rules and ignore rules.

For example, given the following rule in cgrules.conf

\#<user> <controllers> <destination>
tom:foo* memory FooCG/

Processes named 'foo', foo1', 'foo2', etc. would be moved
to the FooCG/ cgroup.  Processes named 'bar', 'fo', etc.
would not match this rule.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agotests: Add unit tests for cgroup_compare_ignore_rule()
Tom Hromatka [Mon, 4 Nov 2019 23:20:01 +0000 (23:20 +0000)] 
tests: Add unit tests for cgroup_compare_ignore_rule()

The following tests are in this commit:

NotAnIgnore() - Test that non-ignore rules are not processed

SimpleMatch() - Test that an ignore rule with a matching cgroup
    and controller match

CgroupsDontMatch() - Test that cgroups with similar but not matching
    names do not match

ControllersDontMatch() - Test that controllers with non-matching
    names do not match

CombinedControllers() - Test that controllers that have been mounted
    on the same path, e.g. cpu,cpuacct, can be matched by a rule

MatchChildFolder() - Compares a rule with a single-level hierarchy
    with a rule that matches that hierarchy

MatchGrandchildFolder() - Compares a rule with a multi-level hierarchy
    with a process that matches that hierachy

MatchSimilarChildFolder() - Compares a child folder that is similar
    to the rule.  This mimics a misconfiguration that a user could
    make

RealWorldMatch() - Test that a realistic cgroup proc file matches
    a new process

RealWorldNoMatch() - Test that a realistic cgroup proc file will
    not match a new process

SimilarFolderNoMatch() - Compares a rule with a folder name that
   is similar to the process' cgroup, but not a match

RootDestinationMatch() - Compares a rule with the root as the
   destination and a process also running in the root cgroup

RootDestinationNoMatch() - Compares a rule with the root as the
   destination and a process in another cgroup folder

The results from googletest are reported below:

[----------] 13 tests from CgroupCompareIgnoreRuleTest
[ RUN      ] CgroupCompareIgnoreRuleTest.NotAnIgnore
[       OK ] CgroupCompareIgnoreRuleTest.NotAnIgnore (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.SimpleMatch
[       OK ] CgroupCompareIgnoreRuleTest.SimpleMatch (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.CgroupsDontMatch
[       OK ] CgroupCompareIgnoreRuleTest.CgroupsDontMatch (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.ControllersDontMatch
[       OK ] CgroupCompareIgnoreRuleTest.ControllersDontMatch (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.CombinedControllers
[       OK ] CgroupCompareIgnoreRuleTest.CombinedControllers (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.MatchChildFolder
[       OK ] CgroupCompareIgnoreRuleTest.MatchChildFolder (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.MatchGrandchildFolder
[       OK ] CgroupCompareIgnoreRuleTest.MatchGrandchildFolder (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.MatchSimilarChildFolder
[       OK ] CgroupCompareIgnoreRuleTest.MatchSimilarChildFolder (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.RealWorldMatch
[       OK ] CgroupCompareIgnoreRuleTest.RealWorldMatch (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.RealWorldNoMatch
[       OK ] CgroupCompareIgnoreRuleTest.RealWorldNoMatch (1 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.SimilarFolderNoMatch
[       OK ] CgroupCompareIgnoreRuleTest.SimilarFolderNoMatch (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.RootDestinationMatch
[       OK ] CgroupCompareIgnoreRuleTest.RootDestinationMatch (0 ms)
[ RUN      ] CgroupCompareIgnoreRuleTest.RootDestinationNoMatch
[       OK ] CgroupCompareIgnoreRuleTest.RootDestinationNoMatch (0 ms)
[----------] 13 tests from CgroupCompareIgnoreRuleTest (3 ms total)

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agoapi.c: Add logic to process ignore rules
Tom Hromatka [Mon, 4 Nov 2019 23:18:48 +0000 (23:18 +0000)] 
api.c: Add logic to process ignore rules

This commit updates the cgrulesengd logic that matches the new pid
to its corresponding rule by adding support for ignore rules.  If
a user has created an ignore rule and the pid matches that rule
completely, then cgrulesengd will perform no further processing
(e.g. moving the pid to a different cgroup) on that pid.  It is
assumed that a separate process - outside of libcgroup - will
manage that process and its cgroups.

A few examples:

<user>      <controller>        <destination>        <options>
*           cpuset              IgnoreCgroup         ignore
*           cpuset              DefaultCgroup

For the above example:
    * A new process named "foo" spawned in the IgnoreCgroup will be
      ignored by cgrulesengd and will not be moved to DefaultCgroup
    * A new process named "bar" spawned in the root cgroup will not
      match the ignore rule and will be moved to the DefaultCgroup
      by the second rule
    * Any processes not in IgnoreCgroup will be automatically moved
      to the default cgroup

<user>      <controller>        <destination>        <options>
*           memory              IgnoreCgroup2/       ignore
*           memory              DefaultCgroup

For the above example:
    * A new process named "bar" spawned in IgnoreCgroup2/childcg
      will match the first rule and thus be ignored by cgrulesengd
      and will not be moved
    * In fact, any process in IgnoreCgroup2 or its children cgroup(s)
      will be ignored by cgrulesengd
    * cgrulesengd will attempt to move all other processes that don't
      match the ignore rule to the default cgroup

<user>      <controller>        <destination>        <options>
jdoe:foo    cpu                 IgnoreCgroup         ignore
*           cpu                 DefaultCgroup

For the above example:
    * If a process named "foo" is spawned by user jdoe and is in the
      IgnoreCgroup cgroup, then cgrulesengd will ignore it and it
      will not be moved
    * If a process named "bar" is spawned in IgnoreCgroup, it will
      not match the first rule and will fall into the default
      rule.  Thus it will be moved to the DefaultCgroup

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agotests: Add unit tests for cg_get_cgroups_from_proc_cgroups()
Tom Hromatka [Mon, 4 Nov 2019 23:16:15 +0000 (23:16 +0000)] 
tests: Add unit tests for cg_get_cgroups_from_proc_cgroups()

The following tests are in this commit:

ReadSingleLine() - Read a single valid line with a non-root controller

ReadSingleLine2() - Read a single valid line with a root controller

ReadEmptyController() - Some systems list a controller with no name.
    This test simulates that scenario

ReadExampleFile() - Read several lines that are similar to a typical
    /proc/{pid}/cgroups file

The results from googletest are reported below:

[----------] 4 tests from GetCgroupsFromProcCgroupsTest
[ RUN      ] GetCgroupsFromProcCgroupsTest.ReadSingleLine
[       OK ] GetCgroupsFromProcCgroupsTest.ReadSingleLine (0 ms)
[ RUN      ] GetCgroupsFromProcCgroupsTest.ReadSingleLine2
[       OK ] GetCgroupsFromProcCgroupsTest.ReadSingleLine2 (1 ms)
[ RUN      ] GetCgroupsFromProcCgroupsTest.ReadEmptyController
[       OK ] GetCgroupsFromProcCgroupsTest.ReadEmptyController (0 ms)
[ RUN      ] GetCgroupsFromProcCgroupsTest.ReadExampleFile
[       OK ] GetCgroupsFromProcCgroupsTest.ReadExampleFile (0 ms)
[----------] 4 tests from GetCgroupsFromProcCgroupsTest (1 ms total)

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agoapi.c: Add function to read /proc/{pid}/cgroup
Tom Hromatka [Mon, 4 Nov 2019 23:16:00 +0000 (23:16 +0000)] 
api.c: Add function to read /proc/{pid}/cgroup

This commit adds a function that, given a pid, can read in
the controllers and cgroups listed in /proc/{pid}/cgroup.
The caller is expected to allocate the controller_list[]
and cgroup_list[] arrays as well as null each entry in the
arrays.  cg_get_cgroups_from_proc_cgroups() will allocate
the necessary memory for each string within the arrays.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agotests: Add unit tests for cgroup_parse_rules_options()
Tom Hromatka [Mon, 4 Nov 2019 23:15:32 +0000 (23:15 +0000)] 
tests: Add unit tests for cgroup_parse_rules_options()

The following tests are in this commit:

RulesOptions_Ignore() - The valid option "ignore" is tested

RulesOptions_IgnoreWithComma() - The valid (but syntactically
    strange) option "ignore," is tested

RulesOptions_InvalidOption() - An invalid option is tested

RulesOptions_InvalidOption2() - An invalid option along with
    a valid option is tested

RulesOptions_EmptyOptions() - An empty string is tested

RulesOptions_NullOptions() - A null-pointer option string is
    tested

The results from googletest are reported below:

[----------] 6 tests from ParseRulesOptionsTest
[ RUN      ] ParseRulesOptionsTest.RulesOptions_Ignore
[       OK ] ParseRulesOptionsTest.RulesOptions_Ignore (0 ms)
[ RUN      ] ParseRulesOptionsTest.RulesOptions_IgnoreWithComma
[       OK ] ParseRulesOptionsTest.RulesOptions_IgnoreWithComma (0 ms)
[ RUN      ] ParseRulesOptionsTest.RulesOptions_InvalidOption
[       OK ] ParseRulesOptionsTest.RulesOptions_InvalidOption (0 ms)
[ RUN      ] ParseRulesOptionsTest.RulesOptions_InvalidOption2
[       OK ] ParseRulesOptionsTest.RulesOptions_InvalidOption2 (0 ms)
[ RUN      ] ParseRulesOptionsTest.RulesOptions_EmptyOptions
[       OK ] ParseRulesOptionsTest.RulesOptions_EmptyOptions (0 ms)
[ RUN      ] ParseRulesOptionsTest.RulesOptions_NullOptions
[       OK ] ParseRulesOptionsTest.RulesOptions_NullOptions (0 ms)
[----------] 6 tests from ParseRulesOptionsTest (0 ms total)

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agoapi.c: Add options field for rules defined in cgrules conf file
Tom Hromatka [Mon, 4 Nov 2019 23:15:11 +0000 (23:15 +0000)] 
api.c: Add options field for rules defined in cgrules conf file

This commit adds a fourth field called options to the rules
entry in cgrules configuration files.  Note that the field is
optional and existing rules will be parsed exactly as before.

Also, this commit only adds the parsing of the options field.
It doesn't change the rule behavior logic; that will come in
a subsequent commit.

An example cgrules.conf using this feature:

<user>    <controller>    <destination>           <options>
*         cpu             MyCgroup                ignore
*         cpu             DefaultCgroup

In the above example (and once the subsequent processing is
added), any process currently in the cpu controller and
MyCgroup will be ignored by cgrulesengd.  In other words,
cgrules will not try to move these processes to another cgroup
or manage them in any fashion.  It is anticipated that a
separate user process (outside of the scope of libcgroup) will
manage pids in this scenario.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agoapi: Increase MAX_MNT_ELEMENTS to 16
Tom Hromatka [Mon, 18 Nov 2019 16:32:08 +0000 (16:32 +0000)] 
api: Increase MAX_MNT_ELEMENTS to 16

Newer kernels now support many cgroup controllers; thus we
need to increase the number of elements we can safely read
from /proc/{pid}/cgroup.  This commit increases the number
of elements from 8 to 16.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agotests: Add STATIC and UNIT_TEST definitions to automake files
Tom Hromatka [Mon, 4 Nov 2019 23:14:16 +0000 (23:14 +0000)] 
tests: Add STATIC and UNIT_TEST definitions to automake files

This commit adds two new definitions to the automake Makefile.am
files.  This change has no effect on the deliverable binaries.

STATIC - For a standard libcgroup library build, STATIC is equal
         to the "static" keyword and behaves exactly the same.
         For a unit test build, STATIC is mapped to an empty
         string.  This allows the unit tests to invoke static
         functions that would otherwise be unavailable to the
         test suite.

UNIT_TEST - This define is used in libcgroup-internal.h to
         wrap the function prototypes of STATIC functions.
         In a standard libcgroup library build, UNIT_TEST
         is _not_ defined, and thus the prototypes are not
         available.  In a unit test build, UNIT_TEST is
         defined and the function prototypes are available
         for the unit tests to use.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
5 years agoREADME: Add README.md
Tom Hromatka [Thu, 10 Oct 2019 15:52:03 +0000 (09:52 -0600)] 
README: Add README.md

This commit adds README.md.  It contains graphics that show the
status of the continuous integration and code coverage results.
It also contains a link to the original README.  Note that github
doesn't allow the inlining of files in markdown due to security
concerns, so a link is the best we can do.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Gather code coverage data after a successful CI run
Tom Hromatka [Mon, 23 Sep 2019 22:14:50 +0000 (22:14 +0000)] 
tests: Gather code coverage data after a successful CI run

Upon successful completion of a continuous integration (CI) build
through Travis CI, this commit enables coveralls.io to generate
code coverage results by utilizing the gcov data generated
during both the unit and functional test runs.

Note that the m4/ax_code_coverage.m4 file is provided by the
autoconf team, and it has been used verbatim.  The original
file is available here:
https://git.savannah.gnu.org/gitweb/?p=autoconf-archive.git;a=blob_plain;f=m4/ax_code_coverage.m4

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Switch to using LXD for functional tests
Tom Hromatka [Mon, 23 Sep 2019 22:09:48 +0000 (22:09 +0000)] 
tests: Switch to using LXD for functional tests

Prior to this commit, the functional tests used LXC.  LXD is
better for configuring a shared device between the host and
container that can be written to by the container.  This
feature is critical for tracking code coverage in the
functional tests.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Add Travis CI yaml configuration file
Tom Hromatka [Thu, 19 Sep 2019 20:57:29 +0000 (14:57 -0600)] 
tests: Add Travis CI yaml configuration file

This commit adds the Travis CI yaml configuration file that
will automatically invoke continuous integration for the
libcgroup project.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Disable the legacy tests
Tom Hromatka [Thu, 19 Sep 2019 20:55:12 +0000 (14:55 -0600)] 
tests: Disable the legacy tests

The legacy tests do not successfully run in Travis CI.  This
commit disables them from running but does leave the test
code in place.  The future of these tests and whether they will
be removed from the repo or not will be revisited at a later
date.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Use python3 in the functional tests
Tom Hromatka [Thu, 19 Sep 2019 20:51:55 +0000 (14:51 -0600)] 
tests: Use python3 in the functional tests

Python 2.7 will not be maintained beyond January 1, 2020.  To
avoid potential problems when python 2 is retired, this commit
explicitly instructs the functional tests to use python3.

Suggested-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Fix inconsistent return value in functional test's Container() class
Tom Hromatka [Mon, 7 Oct 2019 20:44:46 +0000 (14:44 -0600)] 
tests: Fix inconsistent return value in functional test's Container() class

This commit fixes the Container() class' inconsistent usage
of decode().  In the rootfs() method, it returned a binary
array in one case and a string in another.  Now the rootfs()
method will always return a string.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Fix errors in functional test error path handling
Tom Hromatka [Mon, 7 Oct 2019 20:42:21 +0000 (14:42 -0600)] 
tests: Fix errors in functional test error path handling

The RunError() exception class had two bugs that prevented
it from properly formatting a run exception:
    1) It wasn't being created properly due to a misplaced
       parenthesis
    2) It had a syntax error in its __str__() method where
       it was using self.message rather than self.command

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Improve the unit test framework integration with automake
Tom Hromatka [Thu, 19 Sep 2019 20:40:37 +0000 (14:40 -0600)] 
tests: Improve the unit test framework integration with automake

Prior to this commit, the unit test framework behaved slightly
differently depending upon how the libcgroup git repo was
cloned.  This commit standardizes the unit test framework's
behavior by initializing the googletest submodule when
bootstrap.sh is invoked.

This commit also cleans up a potential issue on some versions
of automake that don't properly expand $(top_builddir) early
in the automake process.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Add functional test for basic cgget operations
Tom Hromatka [Thu, 27 Jun 2019 17:34:00 +0000 (11:34 -0600)] 
tests: Add functional test for basic cgget operations

This commit adds a functional test for a basic cgget call.  The
test utilizes the new functional text framework and lxc
containers to establish a non-destructive cgroup test
environment.

Tests added:
    001-cgget-basic_cgget.py - Test simple cgget functionality

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Add functional test suite
Tom Hromatka [Thu, 27 Jun 2019 17:28:53 +0000 (11:28 -0600)] 
tests: Add functional test suite

This commit adds a functional test suite that utilizes lxc
containers to guarantee a non-destructive test environment.

The tests can be invoked individually, as a group of related
tests, or from automake via the standard 'make check'
command.

No tests are included as part of this commit.

Example test invocations:

    Run a single test (first cd to tests/ftests):
        ./001-cgget-basic_cgget.py
        or
        ./ftests.py -N 15      # Run test #015

    Run a suite of tests (first cd to tests/ftests):
        ./ftests.py -s cgget   # Run all cgget tests

    Run all the tests by hand
        ./ftests.py
        # This may be advantageous over running make check
        # because it will try to re-use the same lxc
        # container for all of the tests.  This should
        # provide a significant performance increase

    Run the tests from automake
        make check
        # Then examine the *.trs and *.log files for
        # specifics regarding each test result

Example output from a test run:

Test Results:
        Run Date:                     Jun 03 13:41:35
        Passed:                               1  test
        Skipped:                              0 tests
        Failed:                               0 tests
-----------------------------------------------------------------
Timing Results:
        Test                               Time (sec)
        ---------------------------------------------------------
        setup                                    6.95
        001-cgget-basic_cgget.py                 0.07
        teardown                                 0.00
        ---------------------------------------------------------
        Total Run Time                           7.02

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Add functional test LXC container class
Tom Hromatka [Fri, 21 Jun 2019 15:40:23 +0000 (09:40 -0600)] 
tests: Add functional test LXC container class

This commit adds a Container() class for managing LXC containers.
This class provides methods to abstract the management of LXC
containers and simplify their usage in the functional test suite.

Example usages:
  # instantiate the class
  mycontainer = Container('TheNameOfMyContainer')

  # create and start the container
  mycontainer.create()
  mycontainer.start()

Container() can raise ValueError on invalid parameters,
ContainerError when a container operation unexpectedly failes,
or RunError when running a command fails.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Add functional test cgroup class
Tom Hromatka [Fri, 21 Jun 2019 15:37:20 +0000 (09:37 -0600)] 
tests: Add functional test cgroup class

This commit adds a Cgroup() class for managing cgroups.  This
class provides static methods for interacting with libcgroup's
interfaces including cgset, cgget, cgcreate, etc.

Example usages:
  # create a cgroup in the cpuset controller named foo
  Cgroup.create(config, 'cpuset', 'foo')

  # set cpu.shares for foobar to 500
  Cgroup.set(config, 'foobar', 'cpu.shares', '500')

  # get the limit_in_bytes for AnotherCgroup.  Have libcgroup
  # strip off all of the decorations so that only the value is
  # returned
  limit_in_bytes = Cgroup.get(config, controller=None,
      cgname='AnotherCgroup', setting='memory.limit_in_bytes',
      print_headers=False, values_only=True)

Providing invalid parameters to a Cgroup method will result in
a ValueError while a failure to execute a command will result
in a RunError.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Add functional test run class
Tom Hromatka [Fri, 21 Jun 2019 15:32:44 +0000 (09:32 -0600)] 
tests: Add functional test run class

This commit adds a Run() class that can invoke Python's Popen
command.  This class provides a static method, run(), that will
execute the command via Python's subprocess module.

Example usages:
  # delete foo.conf
  Run.run(['rm', '-f', '/tmp/foo.conf'])

  # Use cgset to set SomeCgroup's cpu.shares to 500
  cmd = ['cgset', '-r', 'cpu.shares=500', 'SomeCgroup']
  Run.run(cmd)

  # get info on current user
  Run.run('id')

  # write to a file.  Note that this must be run in a shell
  Run.run(['echo', 'some data', '>>', 'some_file'], shell_bool=True)

Example output:

  Jun 27 12:34:18: DEBUG: run:
          command = sudo lxc-info --config=lxc.rootfs -n test_libcg
          ret = 0
          stdout = lxc.rootfs = /container/test_libcg/rootfs
          stderr =

  Jun 27 12:34:18: DEBUG: run:
          command = sudo lxc-attach -n test_libcg -- /libcg/src/tools/.libs/cgget -n -v -r cpu.shares 001cgget
          ret = 0
          stdout = 512
          stderr =

If the command fails, Run.run() throws a RunError exception which
contains the return code, stdout, and stderr from the command.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Reviewed-by: Dhaval Giani <dhaval.giani@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Add functional test logging class
Tom Hromatka [Fri, 21 Jun 2019 15:34:28 +0000 (09:34 -0600)] 
tests: Add functional test logging class

This commit adds a Log() class for the functional test
suite.  This class provides several static methods for
logging information during a test run.  The data logged
is often useful for debugging failed tests.

Example usages:
  Log.log_debug('Running test %s' % test name)

  except Exception as e:
    Log.log_error(e)

Both the log level and log file are configurable.  By
default, the functional tests will log all messages of
criticality LOG_WARNING or higher to libcgroup-ftests.log.
These settings can be overridden by the following command
line options:
  -l LOGLEVEL, --loglevel LOGLEVEL
  -L LOGFILE, --logfile LOGFILE

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Reviewed-by: Dhaval Giani <dhaval.giani@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agotests: Add functional test folder
Tom Hromatka [Fri, 21 Jun 2019 15:29:41 +0000 (09:29 -0600)] 
tests: Add functional test folder

This commit adds a folder that will hold the functional tests
and their framework.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Reviewed-by: Dhaval Giani <dhaval.giani@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
5 years agoAdd googletest/ to .gitignore
Dhaval Giani [Tue, 17 Sep 2019 22:11:26 +0000 (15:11 -0700)] 
Add googletest/ to .gitignore

Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
6 years agocgrulesengd: Do not ignore changes of short-lived processes
Michal Koutný [Tue, 11 Jun 2019 07:05:59 +0000 (09:05 +0200)] 
cgrulesengd: Do not ignore changes of short-lived processes

When a double-forking daemon spawns the shortlived forking process and
we fail to classify it in time, the child does not inherit (the
intended) cgroup membership.

We could process all children after receiving PROC_EVENT_FORK to remedy
this. But since we already have the timestamp logic introduced in

    8953fc07c049 ("Changelog v2:  * Use clock_gettime(2) for getting
    timestamp since a system boot.  * Change parent_info's memory to
    dynamic allocation.")

and it may be too much work for all fork(2) calls, we extend the usage
of parent_info by assuming the parent would have changed its cgroup
membership by our actions even if it terminated quickly.

v2: Handle non-existent /proc/$PID/tasks as short-lived process too
    Use cgroup_get_last_errno() helper

Signed-off-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Dhaval Giani <dhaval.giani@gmail.com>
6 years agoChange googletest to the https URL
Dhaval Giani [Thu, 30 May 2019 17:59:08 +0000 (10:59 -0700)] 
Change googletest to the https URL

Change it from the ssh URL, since we don't need ssh access.

Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
6 years agotests: Add unit tests for cg_build_path()
Tom Hromatka [Thu, 30 May 2019 15:22:28 +0000 (09:22 -0600)] 
tests: Add unit tests for cg_build_path()

This commit adds unit tests for the internal function
cg_build_path().  While code coverage is not (yet) enabled,
I believe these tests provide full code coverage for the
aforementioned function and its sister function
cg_build_path_locked().

The following tests are in this commit:

BuildPathV1_ControllerMismatch() - Calls cg_build_path() with a
    controller that isn't in cg_mount_table[].  Expects a return
    value of NULL

BuildPathV1_ControllerMatch() - Calls cg_build_path() with a valid
    controller in cg_mount_table[]

BuildPathV1_ControllerMatchWithName() - Calls cg_build_path() with
    a valid controller and a cgroup name

BuildPathV1_ControllerMatchWithNs() - Calls cg_build_path() with a
    valid controller that has a namespace

BuildPathV1_ControllerMatchWithNameAndNs() - Calls cg_build_path()
    with a valid controller, a cgroup name, and the controller has
    a namespace

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
6 years agotests: Add googletest unit test framework
Tom Hromatka [Thu, 30 May 2019 15:22:27 +0000 (09:22 -0600)] 
tests: Add googletest unit test framework

This commit adds support for the googletest unit test
framework.  To enable the testing of non-exported
functions, a more permissive map file was added.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Signed-off-by: Dhaval Giani <dhaval.giani@oracle.com>
6 years agocgset: fix misleading error message
Tom Hromatka [Tue, 15 Jan 2019 23:33:22 +0000 (16:33 -0700)] 
cgset: fix misleading error message

When cgset fails to modify a cgroup, it output a
hardcoded error message which can cause user confusion.
This commit updates the error message to behave like
other error messages throughout libcgroup and produce
a more user-friendly output.

Signed-off-by: Tom Hromatka <tom.hromatka@oracle.com>
Acked-by: Nikola Forró <nforro@redhat.com>
6 years agoFix issues discovered by Coverity scan
Nikola Forró [Fri, 23 Nov 2018 15:50:43 +0000 (16:50 +0100)] 
Fix issues discovered by Coverity scan

Signed-off-by: Nikola Forró <nforro@redhat.com>
6 years agoapi.c: always move all tasks of a process to a cgroup
Nikola Forró [Mon, 23 Jul 2018 15:38:26 +0000 (17:38 +0200)] 
api.c: always move all tasks of a process to a cgroup

Move the thread enumeration introduced in commit 2186c97
from cgroup_change_all_cgroups() to cgroup_change_cgroup_path()
to ensure it works in every case.

Signed-off-by: Nikola Forró <nforro@redhat.com>
6 years agoAdd support for nosuid, noexec and nodev
Balbir Singh [Fri, 12 Oct 2018 08:57:24 +0000 (10:57 +0200)] 
Add support for nosuid, noexec and nodev

Inspired by the contents of https://lwn.net/Articles/647757/, there
is no easy way of passing these mount options with cgroups. For existing
users, it makes sense to support these

Signed-off-by: Balbir Singh <bsingharora@gmail.com>
Acked-by: Nikola Forró <nforro@redhat.com>
7 years agocgrulesengd: remove umask(0)
Michal Hocko [Wed, 18 Jul 2018 09:24:29 +0000 (11:24 +0200)] 
cgrulesengd: remove umask(0)

One of our partners has noticed that cgred daemon is creating a log file
(/var/log/cgred) with too wide permissions (0666) and that is seen as
a security bug because an untrusted user can write to otherwise
restricted area. CVE-2018-14348 has been assigned to this issue.

Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Balbir Singh <bsingharora@gmail.com>
9 years agoapi.c: change cgroup of every thread of a process
Nikola Forró [Tue, 8 Dec 2015 16:09:08 +0000 (17:09 +0100)] 
api.c: change cgroup of every thread of a process

When changing cgroup of multi-threaded process, only the main threads
cgroup actually changed. Now all threads of a process are enumerated
and cgroup is changed for each of them.

9 years agoapi.c: fix infinite loop
Nikola Forró [Tue, 8 Dec 2015 15:53:41 +0000 (16:53 +0100)] 
api.c: fix infinite loop

If getgrnam or getpwuid functions failed, the program entered
an infinite loop, because the rule pointer was never advanced.
This is now fixed by updating the pointer before continuing
to the next iteration.

10 years agoapi.c: support for setting multiline values in control files
Jan Chaloupka [Mon, 15 Sep 2014 11:29:39 +0000 (13:29 +0200)] 
api.c: support for setting multiline values in control files

As of now, libcgroup does not support multiline values setting from configuration files. i.e. values in a form:

net_prio.ifpriomap="lo 7
eth0 66
eth1 5
eth2 4
eth3 3";

Thus, setting of more network interfaces can not be done from configuration file. Or

devices.allow="a *:* w
c 8:* r";

thus setting list of allow devices can not be set as well. The only way is to set it from userspace, e.g.:
# echo "lo 7" > /sys/fs/cgroup/net_prio/testGroup/net_prio.ifpriomap
# echo "eth 0" > /sys/fs/cgroup/net_prio/testGroup/net_prio.ifpriomap
# echo "eth 1" > /sys/fs/cgroup/net_prio/testGroup/net_prio.ifpriomap
# echo "eth 2" > /sys/fs/cgroup/net_prio/testGroup/net_prio.ifpriomap
# echo "eth 3" > /sys/fs/cgroup/net_prio/testGroup/net_prio.ifpriomap

This patch allows setting of multiline variables.

How this support works:
Multiline value is broken in lines and each line is set by write (man 2 write) syscall (without bufferring).
This implies change of fopen with open, fclose with close.
There is no control on multiline value, thus "eth0\n  \t\n" can be set. However, setting
of "  \t" will fail as write command returns -1. Thus administrator has to set correct
multiline values.

Tested on virtual machine with fedora and rhel with network interface lo, eth0-eth3. Configuration file:

# cat /etc/cgconfig.conf
group testGroup {
        net_prio {
                net_prio.ifpriomap="lo 7
eth0    66
eth1 5
eth2 4
eth3 3";
        }
}

net_prio has to be created before:
# modprobe netprio_cgroup
# mkdir /sys/fs/cgroup/net_prio
# mount -t cgroup -onet_prio none /sys/fs/cgroup/net_prio

Changelog:
test of success of strdup call
free str_val before return (str_val is changing in while cycle,
thus str_start_val points to the start of str_val before while)

Signed-off-by: Jan Chaloupka <jchaloup@redhat.com>
10 years agoparse.y: extending cgroup names with "default"
Jan Chaloupka [Mon, 8 Sep 2014 11:28:18 +0000 (13:28 +0200)] 
parse.y: extending cgroup names with "default"

default is used as a keyword for defining default group permissions. Having "default" as a group name (with double quotes "), parsing is successfull. Howerver, using default without double quotes, parsing fails. This patch modifies parsing rule and lex for default group name without double quotes.

Tested on:

default {
perm {
task {
uid = root;
gid = rgel6610;
fperm = 770;
}
admin {
uid = root;
gid = root;
dperm = 775;
fperm = 744;
}
}
}

group default {
        cpuset {
                cpuset.mems = 0;
                cpuset.cpus = 0;
        }
}

Signed-off-by: Jan Chaloupka <jchaloup@redhat.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
10 years agodocumentation: add examples to man pages
Ivana Hutarova Varekova [Fri, 15 Aug 2014 14:11:45 +0000 (16:11 +0200)] 
documentation: add examples to man pages

This patch add examples and missing info in file section libcgroup man pages.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Jan Chaloupka<jchaloup@redhat.com>
11 years agocgroup_config_create_template_group: fix inverted error check.
Anton Khirnov [Thu, 31 Jul 2014 09:50:34 +0000 (11:50 +0200)] 
cgroup_config_create_template_group: fix inverted error check.

The aux_cgroup pointer will be NULL when an error occurs, but the check
is for it being non-NULL.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
11 years agocgcreate: use "*" character as a meta character for all mounted controllers
Ivana Hutarova Varekova [Thu, 31 Jul 2014 09:40:00 +0000 (11:40 +0200)] 
cgcreate: use "*" character as a meta character for all mounted controllers

This patch adds the possibility to use meta character "*" as a shortcut for all mounted controllers. This meta character can be used in "-g" option.

For example:
    $ cgcreate -g *:first  -g cpu:second
    $ lssubsys
    cpuset:/
    cpuset:/first
    cpu,cpuacct:/
    cpu,cpuacct:/first
    cpu,cpuacct:/second
    memory:/
    memory:/first
    devices:/
    devices:/first
    freezer:/
    freezer:/first
    net_cls,net_prio:/
    net_cls,net_prio:/first
    blkio:/
    blkio:/first
    perf_event:/
    perf_event:/first
    hugetlb:/
    hugetlb:/first

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Reviewed-by: Jan Chaloupka <jchaloup@redhat.com>
11 years agoapi.c: add cgroup_add_all_controllers function
Ivana Hutarova Varekova [Thu, 31 Jul 2014 09:37:26 +0000 (11:37 +0200)] 
api.c: add cgroup_add_all_controllers function

cgroup_add_all_controllers function attach all mounted controllers to a given cgroup. This function just modifies internal libcgroup structure, not the kernel control group.
 input parameter: cgroup
 output parameter: zero or error number

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Reviewed-by: Jan Chaloupka <jchaloup@redhat.com>
11 years agocgrules.d: new man page
Jan Chaloupka [Thu, 31 Jul 2014 09:23:08 +0000 (11:23 +0200)] 
cgrules.d: new man page

This patch contains new man page for cgrules.d directory and corresponding
update of existing man pages. It describes the way additional configuration
files should be created and potential conflicts resulting from an arbitrary
order or reading files in the directory.

Changelog:
* line wrapping (max to 80 characters)
* hiearchy typo corrected
* diffrent typo corrected
* succesfully typo corrected
* /etc/cgconfig.conf added into FILES section
* formulation of advice for creating rules updated. Is it enough or
  still confusing? If so, can you be more specific?
* reformulation of the first and the second paragraph

Signed-off-by: Jan Chaloupka <jchaloup@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agoapi.c: Adding support for loading configuration files from cgrules.d directory
Jan Chaloupka [Thu, 31 Jul 2014 09:21:19 +0000 (11:21 +0200)] 
api.c: Adding support for loading configuration files from cgrules.d directory

Implementation of loading rules from /etc/cgrules.d/. Explanation is in the
cover letter. New function cgroup_parse_rules_file created,
calling cgroup_parse_rules. cgroup_parse_rules is invoked only in
cgroup_change_cgroup_flags, cgroup_init_rules_cache and
cgroup_reload_cached_rules functions. For them the change in cgroup_parse_rules
implementation is transpart.

Tested with two configuration files in /etc/cgrules.d/ and all rules
in /etc/cgrules.conf commented out:

$ cat /etc/cgrules.conf
#*:ls       cpu             strom/%u
#*:sleep    memory          les/%g

$ cat /etc/cgrules.d/cgrules1.conf
*:ls       cpu             strom/%u

$ cat /etc/cgrules.d/cgrules2.conf
*:sleep    memory          les/%g

plus cgconfig.[c|d/*] files:
$ cat /etc/cgconfig.conf
mount {
        cpuset  = /cgroup/cpuset;
        cpu     = /cgroup/cpu;
        cpuacct = /cgroup/cpuacct;
        memory  = /cgroup/memory;
        devices = /cgroup/devices;
        freezer = /cgroup/freezer;
        net_cls = /cgroup/net_cls;
        blkio   = /cgroup/blkio;
}

$ cat /etc/cgconfig.d/small.conf
template strom/%u {
        cpu {}
}

group vetev {
        cpu {}
}

$ cat /etc/cgconfig.d/medium.conf
template les/%g {
        memory {}
}

group drevo/listi {
        memory {}
}

Tested to cache reload as well by sending SIGUSR2 signal to running
cgrulesengd process. After first reload I commented out all rules => no
rules match after invoking ls command. After second reload I uncommented
out ls rule => rule match after invoking ls command. After third reload
I uncommented out ls and sleep rule => debug output of cgrulesengd shows
both rules loaded in the cache

Changelog:
* CGROUP_PARSE_STATE_UNLOCK removed
* reformulation of comment to "Cannot read directory. However,
  CGRULES_CONF_FILE is succesfully parsed. Thus return as a success for
  back compatibility."
* errno = 0 removed, once it is not zero, function returns, so need to set
  it to 0 in every iteration
* fprintf replaced by cgroup_err
* added missing unlocks + new label unlock_list for all returns

This
- * finds a rule matching the given UID or GID.  It will store this rule in rl,
+ * finds a rule matching the given UID or GID.  It will store this rule in trl,
is valid correction. "It will store this rule in rl" talks about cache being
disabled. In this case, it is  stored into trl.

Signed-off-by: Jan Chaloupka <jchaloup@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgdelete: fix the display problem
Weng Meiling [Thu, 31 Jul 2014 08:30:51 +0000 (10:30 +0200)] 
cgdelete: fix the display problem

When multi subsystems mounted on one place like this:

$ lssubsys -m
cpu,cpuacct /cgroup/cpu

$ lscgroup
cpu,cpuacct:/
cpu,cpuacct:/test

if we delete the cgroup with the cgdelete -g, and specifying multi
controllers like this:

$ cgdelete -g cpu,cpuacct:test

or

$ cgdelete -g cpu:test -g cpuacct:test

it will report error:
cgdelete: cannot remove group 'test': No such file or directory

this patch fix the problem.

v1 -> v2
 - make cgdelete -g cpu:/test -g cpu:test failed.
v2 -> v3
 - make cgdelete -g cpu:test -g cpu:test1 ok.
v3 -> v4
 - make cgdelete -g cpuacct:test -g cpu:test -g cpuacct:test failed.
 - add some comments
 - fix the uninitialized warning
v4 -> v5 (created by Ivana Hutarova Varekova, acked by Weng Meiling)
 - make cgdelete -g cpuacct:test -g cpu:test -g cpu:test failed
 - join extended cgroup list variables to one record, do the list dynamic (static version can exceed the given limit and the safe static limit is quite high)

Signed-off-by: Weng Meiling <wengmeiling.weng@huawei.com>
Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Weng Meiling <wengmeiling.weng@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agomanpages: update of manpages for loading configuration
Jan Chaloupka [Thu, 19 Jun 2014 09:42:14 +0000 (11:42 +0200)] 
manpages: update of manpages for loading configuration

Manual pages for new functionality. Because loading files from directory is already implemented in cgconfig tool, there is nothing new. cgrulesengd reads only templates from given files. Rules for writing templates are already specified.

Signed-off-by: Jan Chaloupka <jchaloup@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agoMakefile: update for cgruleseng.d Makefiles
Jan Chaloupka [Thu, 19 Jun 2014 09:39:37 +0000 (11:39 +0200)] 
Makefile: update for cgruleseng.d Makefiles

Modification of makefiles. cgrulesengd.c uses new functions for loading files from /etc/cgconfig.d/. Thus there is a dependency on tools-common.[c|h] files. Unfortunatelly they are in ../tools directory, adding dependency on files in a diffrent directory but on the same level. Makefile.in is regenerated from Makefile.am using autoreconf -ivf.

Tested on my local machine, all files are loaded properly, cgrulesengd properly creates groups based on templates from files from /etc/cgconfig.d/.

Signed-off-by: Jan Chaloupka <jchaloup@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgrulesengd.c: loading of multiples files during init/reload
Jan Chaloupka [Thu, 19 Jun 2014 09:04:26 +0000 (11:04 +0200)] 
cgrulesengd.c: loading of multiples files during init/reload

This patch use new implementation. In cgrulesengd.c, static variable template_files is used to save all config files, i.e. /etc/cgconfig.conf and /etc/cgconfig.d/*. This list of files is then passed to cgroup_templates_cache_set_source_files, which saves files to template_files in config.c module. This list is accessed through invocation of cgroup_init_templates_cache_from_files and cgroup_reload_templates_cache_from_files, which can access variables only from config.c, thus presenting cgroup_templates_cache_set_source_files function. At the end, this list is set free.

Signed-off-by: Jan Chaloupka <jchaloup@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agoconfig.c: loading templates from multiple files into one template table
Jan Chaloupka [Thu, 19 Jun 2014 08:34:32 +0000 (10:34 +0200)] 
config.c: loading templates from multiple files into one template table

This patch implements loading files from /etc/cgconfig.d/. It uses already implemented functions. To hold list of files to be read for template reloading, new static variable for this list (list of template files) has been added to config.c. It is the same list as file list read with cgconfigparser (cgconfig.c file), i.e. the same functions are called. Because parser reads only one file, it is called for each file in template file list. Thus initialization, resp. template duplication has to be called before, resp. after each file parsing. Thus new functions has been presented, thus modifying config.[h|c] api and libcgroup.map. Again parsing more files for templates is analogy to parsing files for groups in cgconfig.c.

cgroup_load_cached_templates_from_file is a modification of cgroup_reload_cached_templates and cgroup_init_templates_cache. Only for loop and auxiliary variables added.

Changelog:
        Description of config.h updated. [init|reload] replaced by load
        TEMPLATE_ACTION_[INIT|RELOAD] removed
        Instead of returning error on uninitialized list of template files, CGCONFIG_CONF_FILE with old cgroup_[init|reload]_templates_cache functions is called.
        cgroup_config_create_template_group reports which template file caused unsuccesfull loading of template files

Signed-off-by: Jan Chaloupka <jchaloup@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agoThere is a situation when moving a task from child cgroup
Weng Meiling [Fri, 4 Apr 2014 07:58:20 +0000 (09:58 +0200)] 
There is a situation when moving a task from child cgroup
to parent cgroup, the task exited. This will lead to service
cgconfig start failed. So ignore the error ESRCH.

Signed-off-by: Weng Meiling <wengmeiling.weng@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agolex: add \ character into regexp for ID token
Jan Chaloupka [Mon, 10 Mar 2014 17:50:57 +0000 (18:50 +0100)] 
lex: add \ character into regexp for ID token

Having virtual machine up, group name contains \ character after
cgsnapshot output. lex.l does not consider this case. For example:
group machine.slice/machine-qemu\x2drhel6.4.scope {...}
Adding \ character into regexp for ID token resolves this problem.

Signed-off-by: Jan Chaloupka <jchaloup@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
11 years agoRelease v0.41 v0.41
Ivana Hutarova Varekova [Mon, 13 Jan 2014 14:48:09 +0000 (15:48 +0100)] 
Release v0.41

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agov0.40.rc1
Ivana Hutarova Varekova [Thu, 21 Nov 2013 17:37:40 +0000 (18:37 +0100)] 
v0.40.rc1

releasing v0.40.rc1

This is bugfixing release.

List of all patches:
Ivana Hutarova Varekova (37):
      fix permissions of logger.sh
      cgroup_get_cgroup: remove bogus comment
      fix template function to copy template groups properly
      documentation: fix cgroup_copy cgroup description
      config.c: better detect whether template group already is created
      parse: fix template parsing
      lex: suppress useless warning message during compilation
      create_cgroup_from_name_value_pairs can't deal n/v pairs from the same controller
      cgsnapshot: don't display mountpoints which are not wanted
      cgclassify: polish usage function
      cgclassify: unify help output
      cgclear: polish usage function
      cgconfig: polish usage function
      cgclear: unify help output
      cgconfig: unify the help output
      cgcreate: polish usage function
      cgdelete: polish usage function
      cgdelete: unification of usage function output
      cgexec: polish usage function
      cgexec: unify the help output
      cgget: polish usage function
      cgset: polish usage function
      cgsnapshot: polish usage function
      cgsnapshot: unify help output
      lscgroup: polish usage function
      lscgroup: unify help output
      lssubsys: polish usage function
      lssubsys: unify help output
      cgroup_init: stop mntopt parsing if there is a bug in named hierarchy
      cgroup_copy_with_slash: two typos
      cgroup_get_current_controller_path: test whether strtok_r returns NULL
      cgroup_config_create_template_group: fix strncpy limits
      cgroup_config_create_template_group: test the right variable
      cgroup_{reload|init}_cached_templates:memory leaks
      cgroup_create_cgroup: write error on fail
      cgconfig.conf: enable sign "@" in cgroup name
      bindings: remove -nostdinc compiler option

Libo Chen (8):
      log.h: add head file <stdarg.h>
      cgcreate: add err check
      cgconfig: add err check
      cgconfig: fix the excursive processing
      cgconfig: add parameter check
      cgconfig: correct freeing default_cgroup
      cgconfig: add check and nonzero
      cgconfig: change function usage behavior

Peter Schiffer (8):
      Use cache when reading /etc/passwd file
      Cgred service won't start if /etc/cgrules.conf is missing
      Typo in debug message in cgexec.c file
      Incorrect cast
      Invalid reallocation of config_table in config_insert_cgroup()
      Templates letter case is switched
      Use cache when calling cgroup_change_cgroup_flags()
      Variable procname leaks

Weng Meiling (6):
      cgsnapshot: fix the print message when use cgsnapshot with -f
      when we use cgsnapshot with the given controller, the result always     displays     the last mounted controller which is not specified, so fix it.
      It's more readable and friendly to use the enum value instead of number.
      cgclassify: use the error message instead of error number
      cgred:cgred should start when cgroup is mounted
      cgcreate: fix the wrong uid/gid set.

11 years agoRevert "'make dist' command compiles "bindings" directory content even if bindings...
Ivana Hutarova Varekova [Thu, 21 Nov 2013 13:50:25 +0000 (14:50 +0100)] 
Revert "'make dist' command compiles "bindings" directory content even if bindings are not enabled in configuration file."

This reverts commit e9b2de289163327c229f095b53b44c0767d60e38.
It is necessary becouse of binding tools can be used in some versions.

11 years ago'make dist' command compiles "bindings" directory content even if bindings are not...
Ivana Hutarova Varekova [Mon, 18 Nov 2013 10:29:27 +0000 (11:29 +0100)] 
'make dist' command compiles "bindings" directory content even if bindings are not enabled in configuration file.
This patch fix it.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agobindings: remove -nostdinc compiler option
Ivana Hutarova Varekova [Mon, 18 Nov 2013 10:25:02 +0000 (11:25 +0100)] 
bindings: remove -nostdinc compiler option

stdarg.h header file is now (libstdc++-devel-4.6.3) in libctdc++ package, thus compilation of bindings failed:
../../include/libcgroup/log.h:12:20: fatal error: stdarg.h: No such file or directory
Thus -nostdinc compiler option have to be removed.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgconfig.conf: enable sign "@" in cgroup name
Ivana Hutarova Varekova [Wed, 30 Oct 2013 13:26:42 +0000 (14:26 +0100)] 
cgconfig.conf: enable sign "@" in cgroup name

This patch enables to have "@" sign in cgroup name in cgconfig.conf as
well.
It is because of systemd, which uses sign "@" to separate the template name
from the instance identifier in its cgroup names.
For example: systemd-fsck@.service.
Without this patch, it's not possible to use these names in cgconfig.conf file.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Peter Schiffer <pschiffe@redhat.com>
11 years agocgroup_create_cgroup: write error on fail
Ivana Hutarova Varekova [Wed, 30 Oct 2013 13:21:33 +0000 (14:21 +0100)] 
cgroup_create_cgroup: write error on fail

In the current version cgroup_create_cgroup returns warning if it fails and
can't set some variable. Thus on default configuration there is no
information about the problem. This situation happens if the cgconfigparser
reads cgconfig.conf. The error message where is the problem is very useful
in this case. The patch do the message visible in standard configuration.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Peter Schiffer <pschiffe@redhat.com>
11 years agocgroup_{reload|init}_cached_templates:memory leaks
Ivana Hutarova Varekova [Wed, 30 Oct 2013 13:17:15 +0000 (14:17 +0100)] 
cgroup_{reload|init}_cached_templates:memory leaks

cgroup_{reload|init}_cached_templates does not free memory if no template
is set. This patch fix the problem.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Peter Schiffer <pschiffe@redhat.com>
11 years agoVariable procname leaks
Peter Schiffer [Fri, 25 Oct 2013 15:07:43 +0000 (17:07 +0200)] 
Variable procname leaks

Variable procname in cgroup_change_all_cgroups() function is leaking. This patch fixes the problem.

Signed-off-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agoUse cache when calling cgroup_change_cgroup_flags()
Peter Schiffer [Mon, 14 Oct 2013 06:49:34 +0000 (08:49 +0200)] 
Use cache when calling cgroup_change_cgroup_flags()

When system contains cgrules.conf file with few thousand rules,
cgroup_change_cgroup_flags() function without CGFLAG_USECACHE flag is very
slow.

As Ivana pointed out, cgexec and cgclassify tools might have problems with
cache
reloading, so I've removed them from the patch for now.

Currently, the patch affects only cgrulesengd daemon. Testing was done with
approximately 4000 users, rules in cgrules.conf file and same amount of
entries
in cgconfig.conf. Without the CGFLAG_USECACHE the cgrulesengd is starting
for
about 15 - 20 minutes (this is because of cgroup_change_all_cgroups()
function,
which scans all running processes on cgrulesengd start).

Signed-off-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agoTemplates letter case is switched
Peter Schiffer [Mon, 14 Oct 2013 06:43:24 +0000 (08:43 +0200)] 
Templates letter case is switched

Man page cgrules.conf(5) says:
     %u     username, uid if name resolving fails
     %U     uid
     %g     group name, gid if name resolving fails
     %G     gid
     %p     process name, pid if name not available
     %P     pid

However, in cgroup_change_cgroup_flags() function, the letter case is switched.
This patch fixes the problem.

Signed-off-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agoInvalid reallocation of config_table in config_insert_cgroup()
Peter Schiffer [Mon, 14 Oct 2013 04:48:06 +0000 (06:48 +0200)] 
Invalid reallocation of config_table in config_insert_cgroup()

There's invalid reallocation of config_table variable in config_insert_cgroup() function.

Signed-off-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agoIncorrect cast
Peter Schiffer [Mon, 14 Oct 2013 04:45:10 +0000 (06:45 +0200)] 
Incorrect cast

This patch fixes incorrect cast in cgclassify.c file.

Signed-off-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agoTypo in debug message in cgexec.c file
Peter Schiffer [Mon, 14 Oct 2013 04:41:38 +0000 (06:41 +0200)] 
Typo in debug message in cgexec.c file

This patch fixes typo in debug message.

Signed-off-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgroup_config_create_template_group: test the right variable
Ivana Hutarova Varekova [Wed, 9 Oct 2013 09:43:38 +0000 (11:43 +0200)] 
cgroup_config_create_template_group: test the right variable

cgroup_config_create_template_group tests the return value of bogus
variable. This patch fix it.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
11 years agocgroup_config_create_template_group: fix strncpy limits
Ivana Hutarova Varekova [Wed, 9 Oct 2013 09:43:25 +0000 (11:43 +0200)] 
cgroup_config_create_template_group: fix strncpy limits

fix strncpy limits in cgroup_config_create_template_group function

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
11 years agocgroup_get_current_controller_path: test whether strtok_r returns NULL
Ivana Hutarova Varekova [Wed, 9 Oct 2013 09:41:22 +0000 (11:41 +0200)] 
cgroup_get_current_controller_path: test whether strtok_r returns NULL

cgroup_get_current_controller_path  should test whether strtok_r returns NULL before it continues.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
11 years agocgroup_copy_with_slash: two typos
Ivana Hutarova Varekova [Wed, 9 Oct 2013 09:39:47 +0000 (11:39 +0200)] 
cgroup_copy_with_slash: two typos

Fix a typo when the '/' character is tested on the last but one position instead of the last.
There have to be more allocated more memory in the same function.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
11 years agocgroup_init: stop mntopt parsing if there is a bug in named hierarchy
Ivana Hutarova Varekova [Wed, 9 Oct 2013 09:14:36 +0000 (11:14 +0200)] 
cgroup_init: stop mntopt parsing if there is a bug in named hierarchy

cgroup_init should stop parsing of named hierarchy if there is a bug.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Peter Schiffer <pschiffe@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
11 years agocgcreate: fix the wrong uid/gid set.
Weng Meiling [Fri, 27 Sep 2013 05:27:30 +0000 (07:27 +0200)] 
cgcreate: fix the wrong uid/gid set.

When setting the cgconfig.conf as follow:

mount {
memory = /cgroup/memory;
}

group . {
perm {
task {
gid = w00227741;
uid = w00227741;
fperm =755;
}
admin {
gid = w00227741;
uid = w00227741;
fperm =755;
dperm = 755;
}
}
memory {
}
}

and restart the cgconfig service. Then switch to the user w00227741
and cgcreate cgroup like this:

# lscgroup
memory:/

# cgcreate -g memory:test
cgcreate: can't create cgroup test: Operation not permitted

the command report the error, but the cgroup test has created:

# lscgroup
memory:/
memory:/test

this patch fix the problem.

Signed-off-by: Weng Meiling <wengmeiling.weng@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agolssubsys: unify help output
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:18:53 +0000 (07:18 +0200)] 
lssubsys: unify help output

lssubsys: unify help output
* add missing options to syntax row
* change the order of options (now in alphabetical order)

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agolssubsys: polish usage function
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:18:43 +0000 (07:18 +0200)] 
lssubsys: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of lssubsys tool. Remove a level of indentation.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agolscgroup: unify help output
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:18:23 +0000 (07:18 +0200)] 
lscgroup: unify help output

lscgroup: unify help output
* use tabulators instead of spaces
* at first describe the functionality and then describe options
* use printf instead fprintf(stdout,..)

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agolscgroup: polish usage function
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:18:03 +0000 (07:18 +0200)] 
lscgroup: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of lscgroup tool. Remove a level of indentation.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgsnapshot: unify help output
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:17:52 +0000 (07:17 +0200)] 
cgsnapshot: unify help output

cgsnapshot: unify help output
* use tabulators instead of spaces
* change formulation
* change fprintf(stdout,...) to printf()

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgsnapshot: polish usage function
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:17:19 +0000 (07:17 +0200)] 
cgsnapshot: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of cgsnapshot tool. Remove a level of indentation.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgset: polish usage function
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:17:07 +0000 (07:17 +0200)] 
cgset: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of cgset tool. Remove a level of indentation.
Use tabulators instead of spaces.
Based on Jan's comment change typo display->set in -g description

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgget: polish usage function
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:11:19 +0000 (07:11 +0200)] 
cgget: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of cgget tool. Remove a level of indentation.
Use printf instead of fprintf(stdout,...)

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgexec: unify the help output
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:11:02 +0000 (07:11 +0200)] 
cgexec: unify the help output

cgexec: unify the help output - add -h option info

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgexec: polish usage function
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:10:21 +0000 (07:10 +0200)] 
cgexec: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of cgexec tool. Remove a level of
indentation.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgdelete: unification of usage function output
Ivana Hutarova Varekova [Fri, 27 Sep 2013 05:09:56 +0000 (07:09 +0200)] 
cgdelete: unification of usage function output

cgdelete use spaces instead of tabulator
use printf instead of fprintf(stdout,...)

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgdelete: polish usage function
Ivana Hutarova Varekova [Thu, 26 Sep 2013 17:30:04 +0000 (19:30 +0200)] 
cgdelete: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of cgdelete tool. Remove a level of indentation.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgcreate: polish usage function
Ivana Hutarova Varekova [Thu, 26 Sep 2013 10:00:16 +0000 (12:00 +0200)] 
cgcreate: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of cgcreate tool. Remove a level of indentation.
use printf instead of fprintf(stderr,...)

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgconfig: unify the help output
Ivana Hutarova Varekova [Thu, 26 Sep 2013 09:59:32 +0000 (11:59 +0200)] 
cgconfig: unify the help output

Remove cgconfig input free row and change the output program name to standardize
the output with the other libcgroup tools.
sort options based on alphabetical order

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgclear: unify help output
Ivana Hutarova Varekova [Thu, 26 Sep 2013 09:58:39 +0000 (11:58 +0200)] 
cgclear: unify help output

cgclear: unify help output
put the options to chronological order

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgconfig: polish usage function
Ivana Hutarova Varekova [Thu, 26 Sep 2013 09:57:26 +0000 (11:57 +0200)] 
cgconfig: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of cgconfig tool. Remove a level of indentation.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgclear: polish usage function
Ivana Hutarova Varekova [Thu, 26 Sep 2013 09:57:06 +0000 (11:57 +0200)] 
cgclear: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage" function of cgclear tool. Remove a level of indentation.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgclassify: unify help output
Ivana Hutarova Varekova [Thu, 26 Sep 2013 09:56:33 +0000 (11:56 +0200)] 
cgclassify: unify help output

cgclassify: unify help output
* -g option is optional
* use the alphabetical order
* add -h option description

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
11 years agocgclassify: polish usage function
Ivana Hutarova Varekova [Thu, 26 Sep 2013 09:53:33 +0000 (11:53 +0200)] 
cgclassify: polish usage function

This change is suggested by Dhaval Giani <dhaval.giani@gmail.com>.
Polish "usage function and remove a level of indentation.

Signed-off-by: Ivana Hutarova Varekova <varekova@redhat.com>
Acked-by: Jan Safranek <jsafrane@redhat.com>
Acked-by: Dhaval Giani <dhaval.giani@gmail.com>
11 years agocgconfig: change function usage behavior
Libo Chen [Thu, 12 Sep 2013 06:38:34 +0000 (08:38 +0200)] 
cgconfig: change function usage behavior

remove exit from function usage and add a structure like usage function
in the other tools.
This is suggested by Ivana Hutarova Varekova <varekova@redhat.com>

now usage willn't exit, so we should help it in err case.

v4:
 - fix some code style reported by checkpatch.pl

Signed-off-by: Libo Chen <clbchenlibo.chen@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgconfig: add check and nonzero
Libo Chen [Tue, 10 Sep 2013 06:29:17 +0000 (08:29 +0200)] 
cgconfig: add check and nonzero

1. check return value of cgroup_string_list_init.
2. add nonzero for -a and -t

v4:
 - add nonzero for cgroup_new_cgroup

Signed-off-by: Libo Chen <clbchenlibo.chen@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgconfig: correct freeing default_cgroup
Libo Chen [Tue, 10 Sep 2013 06:27:05 +0000 (08:27 +0200)] 
cgconfig: correct freeing default_cgroup

We are freeing up default_cgroup even it has not be allocated,
there is no need to do that.

Yes, it is just a cleanup suggested
by Dhaval Giani <dhaval.giani@gmail.com>

Signed-off-by: Libo Chen <clbchenlibo.chen@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgconfig: add parameter check
Libo Chen [Tue, 10 Sep 2013 06:22:04 +0000 (08:22 +0200)] 
cgconfig: add parameter check

we should check invalite parameter, and then print a warning

v3:
 - usage has changed, so no need of fprintf output
v2:
 - change print info format. This is suggested
   by Dhaval Giani <dhaval.giani@gmail.com>

Signed-off-by: Libo Chen <clbchenlibo.chen@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgconfig: fix the excursive processing
Libo Chen [Tue, 10 Sep 2013 06:18:32 +0000 (08:18 +0200)] 
cgconfig: fix the excursive processing

error and ret should be unified

v4:
 - use error for cgroup_string_list_add_item
   goto err lable instead of exit

Signed-off-by: Libo Chen <clbchenlibo.chen@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgconfig: add err check
Libo Chen [Tue, 10 Sep 2013 06:13:56 +0000 (08:13 +0200)] 
cgconfig: add err check

cgroup still can be created even if giving invalid -d,-f,-s parameter.
e.g.
        cgconfigparser  -d 888 -l /etc/cgconfig.conf
        "888" is invalid, but cgroup will be created.

        So we should exit when -d,-f or -s is invalid.

v3:
 - add nonzero return value
 - use error instead of ret

Signed-off-by: Libo Chen <clbchenlibo.chen@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgcreate: add err check
Libo Chen [Tue, 10 Sep 2013 06:11:56 +0000 (08:11 +0200)] 
cgcreate: add err check

cgroup still can be created even if giving invalid -d,-f,-s parameter
e.g.
        cgcreate -g cpu:/test -d 888
        "888" is invalid, but test will be cgreated.

        So we should exit if -d,-f or -s is invalid

v2:
 - add nonzero return value

Signed-off-by: Libo Chen <clbchenlibo.chen@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>
11 years agocgred:cgred should start when cgroup is mounted
Weng Meiling [Tue, 10 Sep 2013 06:06:20 +0000 (08:06 +0200)] 
cgred:cgred should start when cgroup is mounted

If we mount the cgroup without the name cgroup like the following:
cpu on /cgroup/cpu type cgroup (rw,cpu)

the cgred service will start failed, because it just check the name.
So changing cgred to check the mounted system's type instead of the name.

v1 -> v2
Using the right method to check cgroup filesystem. The first version
patch's method is wrong, so resend the second patch.

Signed-off-by: Weng Meiling <wengmeiling.weng@huawei.com>
Acked-by: Ivana Hutarova Varekova <varekova@redhat.com>