[PATCHES] GPO support for the AD DC (vers 5)

Fri Nov 17 16:44:29 UTC 2017

Have you looked at this Garming?

On 11/09/2017 02:11 PM, Garming Sam wrote:
> Hi,
>
> It seems to pass the tests now. The tests also complain about the
> executable header in gpclass.py being unnecessary, so that probably
> needs to just be removed.
>
> I will start looking over the additional patches so that I can put my
> review on them.
>
>
> Cheers,
>
> Garming
>
> On 08/11/17 07:08, David Mulder wrote:
>> Here is the fix.
>> We also need to ensure we have a talloc stackframe in ads_connect().
>> Also, samba_gpoupdate needs to be executable or the samba4.gpo.apply
>> test fails to execute the script.
>> Again, I've also got the patches posted on my github repo:
>>
>> https://github.com/dmulder/samba/tree/garming-gpo-david-rebased
>>
>> Please re-test and hopefully let's get these pushed into master.
>>
>> On 11/06/2017 07:45 PM, Garming Sam wrote:
>>> Another failure here:
>>>
>>> [1246(6605)/2209 at 1h56m16s] samba4.gpo.apply(ad_dc:local)
>>> smbtorture 4.8.0pre1-DEVELOPERBUILD
>>> Using seed 1510022136
>>> no talloc stackframe at ../source3/libads/ldap.c:88, leaking memory
>>> PANIC (pid 15202): No talloc stackframe
>>> BACKTRACE: 17 stack frames:
>>>   #0 bin/shared/libsmbconf.so.0(log_stack_trace+0x1f) [0x2aafce37106e]
>>>   #1 bin/shared/libsmbconf.so.0(smb_panic_s3+0x6f) [0x2aafce370eb9]
>>>   #2 bin/shared/libsamba-util.so.0(smb_panic+0x28) [0x2aafcb72edf8]
>>>   #3 bin/shared/libsamba-util.so.0(_talloc_tos+0x10a) [0x2aafcb7323bc]
>>>   #4 bin/shared/private/libads-samba4.so(ldap_open_with_timeout+0xf6)
>>> [0x2aafce30539e]
>>>   #5 bin/shared/private/libads-samba4.so(ads_connect+0x5eb)
>>> [0x2aafce306ed8]
>>>   #6 bin/shared/private/libads-samba4.so(ads_connect_user_creds+0x2b)
>>> [0x2aafce30733f]
>>>   #7 bin/python/samba/gpo.so(+0x290a) [0x2aafce27e90a]
>>>   #8 python(PyEval_EvalFrameEx+0x118c) [0x524cdc]
>>>   #9 python(PyEval_EvalFrameEx+0xc9a) [0x5247ea]
>>>   #10 python(PyEval_EvalFrameEx+0xc9a) [0x5247ea]
>>>   #11 python() [0x567d14]
>>>   #12 python(PyRun_FileExFlags+0x92) [0x465bf4]
>>>   #13 python(PyRun_SimpleFileExFlags+0x2ee) [0x46612d]
>>>   #14 python(Py_Main+0xb5e) [0x466d92]
>>>   #15 /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf5)
>>> [0x2aafca044f45]
>>>   #16 python() [0x577c2e]
>>> smb_panic(): calling panic action
>>> [/home/ubuntu/autobuild/b4818/samba/selftest/gdb_backtrace 15202]
>>> [Thread debugging using libthread_db enabled]
>>>
>>> Garming
>>>
>>>
>>> On 03/11/17 03:12, David Mulder wrote:
>>>> I've rebased Andrew's "Use talloc_stackframe() not talloc_tos() in
>>>> namequery.c" commit with this patch and pushed it to github:
>>>> https://github.com/dmulder/samba/tree/garming-gpo-david-rebased
>>>> I've also attached the rebased commit here.
>>>>
>>>> Garming, could you re-run your tests please?
>>>>
>>>> On 11/02/2017 07:49 AM, David Mulder via samba-technical wrote:
>>>>> This patch fixes the crash.
>>>>> We should free the frame from all return cases in resolve_name()
>>>>>
>>>>> On 11/01/2017 04:23 PM, Garming Sam wrote:
>>>>>> Hi,
>>>>>>
>>>>>> Seen a failure in our test suites with the patches:
>>>>>>
>>>>>> [57(928)/2209 at 47m30s] samba.tests.tdb_util
>>>>>> [58(929)/2209 at 47m31s] samba.tests.tdb_util.python3
>>>>>> [59(930)/2209 at 47m31s] samba.tests.pam_winbind(local)(ad_member)
>>>>>> ERROR: Testsuite[samba.tests.pam_winbind(local)(ad_member)]
>>>>>> REASON: unable to set up environment ad_member - exiting
>>>>>>
>>>>>>
>>>>>> Freed frame ../source3/utils/net.c:917, expected
>>>>>> ../source3/libsmb/namequery.c:2871.
>>>>>> PANIC (pid 30133): Frame not freed in order.
>>>>>> BACKTRACE: 10 stack frames:
>>>>>>    #0 bin/shared/libsmbconf.so.0(log_stack_trace+0x1f)
>>>>>> [0x2b386416106e]
>>>>>>    #1 bin/shared/libsmbconf.so.0(smb_panic_s3+0x6f) [0x2b3864160eb9]
>>>>>>    #2 bin/shared/libsamba-util.so.0(smb_panic+0x28) [0x2b3863f60df8]
>>>>>>    #3 bin/shared/libsamba-util.so.0(+0x25fe3) [0x2b3863f63fe3]
>>>>>>    #4 bin/shared/private/libtalloc.so.2(+0x36a6) [0x2b38641bd6a6]
>>>>>>    #5 bin/shared/private/libtalloc.so.2(+0x3a2c) [0x2b38641bda2c]
>>>>>>    #6 bin/shared/private/libtalloc.so.2(_talloc_free+0x102)
>>>>>> [0x2b38641bece7]
>>>>>>    #7 ./bin/net(main+0xf24) [0x2b38637dc214]
>>>>>>    #8 /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf5)
>>>>>> [0x2b38658a7f45]
>>>>>>    #9 ./bin/net(+0x246b9) [0x2b38637c26b9]
>>>>>>
>>>>>>
>>>>>> It looks like the talloc stackframe work hasn't quite been
>>>>>> completed.
>>>>>> There seems be another location which has not yet been included.
>>>>>>
>>>>>>
>>>>>> Cheers,
>>>>>>
>>>>>>
>>>>>> Garming
>>>>>>
>>>>>>
>>>>>> On 02/11/17 10:01, David Mulder via samba-technical wrote:
>>>>>>> Here is a patch to improve the formatting.
>>>>>>>
>>>>>>> On 11/01/2017 01:08 PM, Andrew Bartlett wrote:
>>>>>>>> On Thu, 2017-11-02 at 06:10 +1300, Andrew Bartlett via
>>>>>>>> samba-technical
>>>>>>>> wrote:
>>>>>>>>> On Wed, 2017-11-01 at 11:01 -0600, David Mulder wrote:
>>>>>>>>>> Oh, sorry. It actually all came through in a single email
>>>>>>>>>> (labeled
>>>>>>>>>> one
>>>>>>>>>> of thirty).
>>>>>>>>> Sorry, didn't scroll down ;-)
>>>>>>>>>
>>>>>>>>>> But it looks as though I missed a commit anyhow, since there
>>>>>>>>>> should be 31 commits, not 30.
>>>>>>>>>> Here is an attachment, and is posted to my github repo:
>>>>>>>>>> https://github.com/dmulder/samba/tree/garming-gpo-david-rebased
>>>>>>>>> Thanks!
>>>>>>>> I've pushed it back to garming-gpo-david-rebased after some
>>>>>>>> squashing.
>>>>>>>>
>>>>>>>> The diff is:
>>>>>>>>
>>>>>>>> diff --git a/libgpo/pygpo.c b/libgpo/pygpo.c
>>>>>>>> index 38844e5a43e..4f53981ec83 100644
>>>>>>>> --- a/libgpo/pygpo.c
>>>>>>>> +++ b/libgpo/pygpo.c
>>>>>>>> @@ -70,7 +70,7 @@ static PyObject *py_gpo_get_unix_path(PyObject
>>>>>>>> *self, PyObject *args,
>>>>>>>>            TALLOC_CTX *frame = NULL;
>>>>>>>>            static const char *kwlist[] = {"cache_dir", NULL};
>>>>>>>>            struct GROUP_POLICY_OBJECT *gpo_ptr \
>>>>>>>> -               = pytalloc_get_type(self, struct
>>>>>>>> GROUP_POLICY_OBJECT);
>>>>>>>> +               = (struct GROUP_POLICY_OBJECT
>>>>>>>> *)pytalloc_get_ptr(self);
>>>>>>>>              if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s",
>>>>>>>> discard_const_p(char *, kwlist), &cache_dir)) {
>>>>>>>>                    PyErr_SetString(PyExc_SystemError, "Failed to
>>>>>>>> parse
>>>>>>>> arguments to gpo_get_unix_path()");
>>>>>>>>
>>>>>>>>
>>>>>>>> I'll try to get this reviewed into master during the next little
>>>>>>>> while.
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>>
>>>>>>>> Andrew Bartlett
>>>>>>>>
>>>
>
>

-- 
David Mulder
SUSE Labs Software Engineer - Samba
dmulder at suse.com
SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg)