tprouty/samba.git
16 years agor20402: Fix spelling: samba bug #4292 debian #402392
Andrew Bartlett [Fri, 29 Dec 2006 08:25:51 +0000 (08:25 +0000)]
r20402: Fix spelling: samba bug #4292 debian #402392

16 years agor20394: This is a *VERY* early start of my work on notify.
Volker Lendecke [Thu, 28 Dec 2006 21:50:31 +0000 (21:50 +0000)]
r20394: This is a *VERY* early start of my work on notify.

Checking in because Jeremy was bugging me. Potentially this becomes quite
intrusive, I'm not sure if I should open a temporary branch for this.

Jeremy, Jerry, do you think 3_0 is the right place for this?

Volker

16 years agor20393: Debug level 1 is a bit excessive here
Volker Lendecke [Thu, 28 Dec 2006 21:34:31 +0000 (21:34 +0000)]
r20393: Debug level 1 is a bit excessive here

16 years agor20389: this needs to be after AC_LIBREPLACE_BROKEN_CHECKS :-)
Stefan Metzmacher [Thu, 28 Dec 2006 19:04:32 +0000 (19:04 +0000)]
r20389: this needs to be after AC_LIBREPLACE_BROKEN_CHECKS :-)

metze

16 years agor20388: add the library that contains dlopen to the global LIBS as it was before
Stefan Metzmacher [Thu, 28 Dec 2006 18:57:21 +0000 (18:57 +0000)]
r20388: add the library that contains dlopen to the global LIBS as it was before

metze

16 years agor20386: remove unneeded dlopen related configure checks and includes
Stefan Metzmacher [Thu, 28 Dec 2006 18:09:32 +0000 (18:09 +0000)]
r20386: remove unneeded dlopen related configure checks and includes
this is done by libreplace...

this hopefully fixes the build on HPUX 11.00

metze

16 years agor20385: merge from samba4:
Stefan Metzmacher [Thu, 28 Dec 2006 17:17:16 +0000 (17:17 +0000)]
r20385: merge from samba4:

only try to find dlfcn.h if the dlopen symbol was found,
it hopefully fixes systems where dlfcn.h but no library with dlopen

metze

16 years agor20372: Make NTrename the same as SMBmv w.r.t. wildcards.
Jeremy Allison [Thu, 28 Dec 2006 00:24:34 +0000 (00:24 +0000)]
r20372: Make NTrename the same as SMBmv w.r.t. wildcards.
Jeremy.

16 years agor20371: In renames (via SMBmv) both src and dest can contain
Jeremy Allison [Thu, 28 Dec 2006 00:10:55 +0000 (00:10 +0000)]
r20371: In renames (via SMBmv) both src and dest can contain
wcards. I had forgotten this and had refused to allow
dest to contain wcards. We now pass all the normal
Samba4 smbtorture RAW-RENAME tests.
Jeremy

16 years agor20370: Now we've gone to the trouble of getting an NT status
Jeremy Allison [Thu, 28 Dec 2006 00:01:12 +0000 (00:01 +0000)]
r20370: Now we've gone to the trouble of getting an NT status
from open_file_ntcreate make sure we return it on rename
error.
Jeremy.

16 years agor20368: Fix valgrind report of uninitialized memory
Jeremy Allison [Wed, 27 Dec 2006 23:52:47 +0000 (23:52 +0000)]
r20368: Fix valgrind report of uninitialized memory
being returned (false positive). Up debug
level on "open_directory" message.
Jeremy.

16 years agor20363: Fix any possible valgrind errors in srvstr_get_XX or srvstr_pull_XX
Jeremy Allison [Wed, 27 Dec 2006 20:45:12 +0000 (20:45 +0000)]
r20363: Fix any possible valgrind errors in srvstr_get_XX or srvstr_pull_XX
by ensuring we pass in a valid src_len (or zero when appropriate).
Volker is correct in that this is a *horrible* interface and he is
now free to generally clean it up everywhere :-). Go for it Volker !
Jeremy.

16 years agor20362: Fix valgrind issues where we may read params
Jeremy Allison [Wed, 27 Dec 2006 18:51:09 +0000 (18:51 +0000)]
r20362: Fix valgrind issues where we may read params
before checking length. Found by Volker.
Jeremy.

16 years agor20361: Prevent strnlen_w reading beyond max. Valgrind
Jeremy Allison [Wed, 27 Dec 2006 18:36:00 +0000 (18:36 +0000)]
r20361: Prevent strnlen_w reading beyond max. Valgrind
found by Volker.
Jeremy

16 years agor20356: Consolidate the calls to parent_dirname() per open to one.
Volker Lendecke [Wed, 27 Dec 2006 10:57:59 +0000 (10:57 +0000)]
r20356: Consolidate the calls to parent_dirname() per open to one.

This involved passing the dirname as argument to a few routines instead of
calling parent_dirname() deep down.

Volker

16 years agor20355: Fix some C++ warnings
Volker Lendecke [Wed, 27 Dec 2006 10:26:32 +0000 (10:26 +0000)]
r20355: Fix some C++ warnings

16 years agor20345: Simplify call_nt_transact_create by using tmp_talloc_ctx()
Volker Lendecke [Sun, 24 Dec 2006 17:09:08 +0000 (17:09 +0000)]
r20345: Simplify call_nt_transact_create by using tmp_talloc_ctx()

16 years agor20344: Move the calls to inherit_access_acls and change_owner_to_parent together.
Volker Lendecke [Sun, 24 Dec 2006 16:37:22 +0000 (16:37 +0000)]
r20344: Move the calls to inherit_access_acls and change_owner_to_parent together.

Jeremy, I'm 100% sure you watch me closely here, I count on you :-)

Volker

16 years agor20343: Split change_owner_to_parent fd-based (for file opens) and a name-based (for
Volker Lendecke [Sun, 24 Dec 2006 15:29:09 +0000 (15:29 +0000)]
r20343: Split change_owner_to_parent fd-based (for file opens) and a name-based (for
directory opens) routines.

Volker

16 years agor20342: Move the lstat into mkdir_internal, for the FILE_OPEN case in open_directory
Volker Lendecke [Sun, 24 Dec 2006 15:11:03 +0000 (15:11 +0000)]
r20342: Move the lstat into mkdir_internal, for the FILE_OPEN case in open_directory
we want to follow symlinks.

Volker

16 years agor20340: Join vfs_MkDir to its only caller
Volker Lendecke [Sun, 24 Dec 2006 14:29:19 +0000 (14:29 +0000)]
r20340: Join vfs_MkDir to its only caller

16 years agor20339: Make mkdir_internal static to open.c
Volker Lendecke [Sun, 24 Dec 2006 11:45:55 +0000 (11:45 +0000)]
r20339: Make mkdir_internal static to open.c

16 years agor20338: Restructure open_directory a bit. This gets rid of a race condition regarding
Volker Lendecke [Sun, 24 Dec 2006 11:13:32 +0000 (11:13 +0000)]
r20338: Restructure open_directory a bit. This gets rid of a race condition regarding
error messages: We relied upon a stat that a directory did not exist to later
on then do the mkdir or not. This does the mkdir directly and copes with a
potential error.

The second one is more important: It's possible with Samba 3 to do a
ntcreate&x with NTCREATEX_OPTIONS_DIRECTORY and we happily do a NT_STATUS_OK.

Also move up the use_nt_status() logic a bit. I think this does not belong
into the core routines, the smb server as such should take care of it.

Jeremy, do you think this should go to 3.0.24?

I'll update samba4torture when the build farm has picked up this checkin.

Volker

16 years agor20336: Fix some memleaks in trans2open and simplify trans2mkdir a bit: We now have
Volker Lendecke [Sat, 23 Dec 2006 19:35:36 +0000 (19:35 +0000)]
r20336: Fix some memleaks in trans2open and simplify trans2mkdir a bit: We now have
tmp_talloc_ctx().

Volker

16 years agor20335: Port oplock return logic from ntcreate&x to NT_TRANSACT_CREATE. Fixes parts of
Volker Lendecke [Sat, 23 Dec 2006 18:55:08 +0000 (18:55 +0000)]
r20335: Port oplock return logic from ntcreate&x to NT_TRANSACT_CREATE. Fixes parts of
raw-open.

Volker

16 years agor20334: Activate RAW-MKDIR
Volker Lendecke [Sat, 23 Dec 2006 18:01:34 +0000 (18:01 +0000)]
r20334: Activate RAW-MKDIR

16 years agor20333: Fix a couple of Coverity errors in one run, this was a potential NULL dereference
Volker Lendecke [Sat, 23 Dec 2006 16:47:50 +0000 (16:47 +0000)]
r20333: Fix a couple of Coverity errors in one run, this was a potential NULL dereference

16 years agor20330: And here's the fix for the parent winbindd crashing
Jeremy Allison [Sat, 23 Dec 2006 02:26:18 +0000 (02:26 +0000)]
r20330: And here's the fix for the parent winbindd crashing
after it's child died unexpectedly whilst the parent
was waiting for a reply. We need to clean up the request
we're not going to service, plus we still need to call
the continuation function with a "False" flag so it
can clean things up. Still testing this, but I think
I'm right.
Jeremy

16 years agor20329: Fix a winbindd crash bug. If someone pulls
Jeremy Allison [Sat, 23 Dec 2006 00:17:15 +0000 (00:17 +0000)]
r20329: Fix a winbindd crash bug. If someone pulls
the network cable out of the machine *exactly*
after the init_dc_connect() call in cm_connect_sam()
or cm_connect_lsa() call succeeded but before any
of the other calls fail, and they have debug level
10 set in the log, then we'd crash due to dereferencing
a now NULL pointer (conn->cli gets set to NULL when
the init_dc_connect() call called from cm_get_schannel_dcinfo()
fails). Yes, before you ask this *did* happen on a
customer site :-).
Jeremy.

16 years agor20313: Make trans2_mkdir also go through open_directory for all the fancy inherit_acl
Volker Lendecke [Thu, 21 Dec 2006 22:35:10 +0000 (22:35 +0000)]
r20313: Make trans2_mkdir also go through open_directory for all the fancy inherit_acl
stuff & friends.

Survives raw-mkdir. I'll activate that tomorrow when all build farm machines
have picked up the changed samba4 torture test.

Volker

16 years agor20312: Fix a memleak
Volker Lendecke [Thu, 21 Dec 2006 22:30:21 +0000 (22:30 +0000)]
r20312: Fix a memleak

16 years agor20310: We've checked for EA's and support for them further up already.
Volker Lendecke [Thu, 21 Dec 2006 21:30:11 +0000 (21:30 +0000)]
r20310: We've checked for EA's and support for them further up already.

Volker

16 years agor20304: Smaller fixes for pam_winbind:
Günther Deschner [Thu, 21 Dec 2006 13:31:23 +0000 (13:31 +0000)]
r20304: Smaller fixes for pam_winbind:

* fail on invalid credential flags in pam_sm_setcred
* parse config file for pam_sm_acct_mgmt and pam_sm_open_session

Guenther

16 years agor20296: If we're going to overwrite krb5.conf
Jeremy Allison [Thu, 21 Dec 2006 00:43:21 +0000 (00:43 +0000)]
r20296: If we're going to overwrite krb5.conf
only do it for our primary domain.
Jeremy.

16 years agor20290: Remove unused call
Simo Sorce [Wed, 20 Dec 2006 19:05:07 +0000 (19:05 +0000)]
r20290: Remove unused call

16 years agor20289: IDMAP is part of winbind but not the main process.
Simo Sorce [Wed, 20 Dec 2006 17:56:26 +0000 (17:56 +0000)]
r20289: IDMAP is part of winbind but not the main process.
Make sure we route all request to remote DCs via the main process
so that IDMAP can correctly reuse DC connections and use the
async interface.

This fixes also idmap_nss so that it is able to resolve local
group names (requires patch on the samba dc earlier committed
to SAMBA_3_0 to make it resolve both the mapped and the unmapped
name).

Simo.

16 years agor20287: Allow a NULL sid when setting up the list of trusted domains.
Gerald Carter [Wed, 20 Dec 2006 16:57:10 +0000 (16:57 +0000)]
r20287: Allow a NULL sid when setting up the list of trusted domains.

16 years agor20279: Fix winbind segfault in winbindd_getsidaliases.
Günther Deschner [Wed, 20 Dec 2006 14:23:41 +0000 (14:23 +0000)]
r20279: Fix winbind segfault in winbindd_getsidaliases.

Jeremy: sidstr formerly could be NULL (when num_aliases was 0), since we
strdup here it needs to exist.

Guenther

16 years agor20273: Map KRB5_KDCREP_SKEW to NT_STATUS_TIME_DIFFERENCE_AT_DC.
Günther Deschner [Wed, 20 Dec 2006 10:54:09 +0000 (10:54 +0000)]
r20273: Map KRB5_KDCREP_SKEW to NT_STATUS_TIME_DIFFERENCE_AT_DC.

This gives much nicer error messages when failing to join due to clock
skew.

Guenther

16 years agor20271: Fix the build
Volker Lendecke [Wed, 20 Dec 2006 07:38:58 +0000 (07:38 +0000)]
r20271: Fix the build

16 years agor20270: Even with the dual daemon mode the parent winbindd
Jeremy Allison [Wed, 20 Dec 2006 04:50:46 +0000 (04:50 +0000)]
r20270: Even with the dual daemon mode the parent winbindd
still needs to contact the DC's for non async requests
like enumerate users/groups etc. Now that online
DC detection is tied to async events we must enable
the processing of events in the main loop of winbindd.
Finally got rid of the last hard coded domain->initialized = 1
code in init_child_recv() - now all domain->initialized = True
gets done only in the connection manager code when either
we're online and have spoken to the DC or are offline and
we know we can't talk to the DC.
Jeremy.

16 years agor20269: merge -r20264:20267 from SAMBA_3_0_24
Herb Lewis [Wed, 20 Dec 2006 01:10:04 +0000 (01:10 +0000)]
r20269: merge -r20264:20267 from SAMBA_3_0_24
more no previous prototype warnings

16 years agor20268: merge -r 20261:20263 from samba_3_0_24
Herb Lewis [Wed, 20 Dec 2006 01:07:21 +0000 (01:07 +0000)]
r20268: merge -r 20261:20263 from samba_3_0_24
get rid of previous prototype warnings

16 years agor20264: Two lines above this we just checked if (fsp->is_directory) ...
Volker Lendecke [Tue, 19 Dec 2006 22:13:10 +0000 (22:13 +0000)]
r20264: Two lines above this we just checked if (fsp->is_directory) ...

16 years agor20261: merge 20260 from samba_3_0_24
Herb Lewis [Tue, 19 Dec 2006 20:16:52 +0000 (20:16 +0000)]
r20261: merge 20260 from samba_3_0_24
clean up a bunch of no previous prototype warnings

16 years agor20257: Hey, change_owner_to_parent is now static :-)
Volker Lendecke [Tue, 19 Dec 2006 18:08:51 +0000 (18:08 +0000)]
r20257: Hey, change_owner_to_parent is now static :-)

16 years agor20255: Activate BASE-SAMBA3ERROR in the build farm
Volker Lendecke [Tue, 19 Dec 2006 18:02:24 +0000 (18:02 +0000)]
r20255: Activate BASE-SAMBA3ERROR in the build farm

16 years agor20254: The pam_chauthtok needs to go through the async interface as well.
Günther Deschner [Tue, 19 Dec 2006 17:35:47 +0000 (17:35 +0000)]
r20254: The pam_chauthtok needs to go through the async interface as well.
This fixes pam password changes in the online case.

Guenther

16 years agor20253: Reduce some code duplication, make reply_mkdir go through the same code paths
Volker Lendecke [Tue, 19 Dec 2006 16:36:54 +0000 (16:36 +0000)]
r20253: Reduce some code duplication, make reply_mkdir go through the same code paths
ncreate does. This is a bit slower (about 10-20%), because it goes touches the
share mode db, but I think not having to call change_owner_to_parent and
friends in fewer places outweighs this. And, mkdir is not the way current
Windows boxes create directories, they do it via the ncreate call.

Volker

16 years agor20252: Revert back to const, but I have a fleble feeling we should go the ther way.
Simo Sorce [Tue, 19 Dec 2006 14:01:05 +0000 (14:01 +0000)]
r20252: Revert back to const, but I have a fleble feeling we should go the ther way.

16 years agor20250: If we've come from being globally offline we
Jeremy Allison [Tue, 19 Dec 2006 00:48:39 +0000 (00:48 +0000)]
r20250: If we've come from being globally offline we
don't have a check online event handler set.
We need to add one once we're been asked to
go back online as this is the only way to actually
go into the online state. Doh ! :-).
Jeremy.

16 years agor20249: Fail when parsing invalid options in _pam_parse.
Günther Deschner [Mon, 18 Dec 2006 22:10:09 +0000 (22:10 +0000)]
r20249: Fail when parsing invalid options in _pam_parse.

Guenther

16 years agor20248: merge 20247 from samba_3_0_24
Herb Lewis [Mon, 18 Dec 2006 22:01:39 +0000 (22:01 +0000)]
r20248: merge 20247 from samba_3_0_24
get rid of warning: value computed is not used

16 years agor20245: merge 20244 from samba_3_0_24
Herb Lewis [Mon, 18 Dec 2006 20:37:26 +0000 (20:37 +0000)]
r20245: merge 20244 from samba_3_0_24
get rid of more nested extern declarations warnings

16 years agor20243: Make lookup_name resolve both the mapped and the real unix group name
Simo Sorce [Mon, 18 Dec 2006 20:05:50 +0000 (20:05 +0000)]
r20243: Make lookup_name resolve both the mapped and the real unix group name

16 years agor20242: these are not really const as we set them in the function
Simo Sorce [Mon, 18 Dec 2006 20:04:54 +0000 (20:04 +0000)]
r20242: these are not really const as we set them in the function

16 years agor20241: Slightly improve readability of the pam_vsyslog replacement function.
Günther Deschner [Mon, 18 Dec 2006 15:35:52 +0000 (15:35 +0000)]
r20241: Slightly improve readability of the pam_vsyslog replacement function.

Guenther

16 years agor20240: Be a little more verbose about the credential flags when the debug flag
Günther Deschner [Mon, 18 Dec 2006 15:33:57 +0000 (15:33 +0000)]
r20240: Be a little more verbose about the credential flags when the debug flag
is set.

Guenther

16 years agor20239: Parse the configfile for pam_sm_setcred as well
Günther Deschner [Mon, 18 Dec 2006 15:29:05 +0000 (15:29 +0000)]
r20239: Parse the configfile for pam_sm_setcred as well
(e.g. to get the debug flag)

Guenther

16 years agor20237: Replace exit_server with exit_server_cleanly where appropriate. All
James Peach [Mon, 18 Dec 2006 04:25:21 +0000 (04:25 +0000)]
r20237: Replace exit_server with exit_server_cleanly where appropriate. All
send_smb failures should be clean exits. All times when we exit as
a matter of policy should also be clean exits.

16 years agor20235: Don't use realloc directly.
James Peach [Mon, 18 Dec 2006 01:59:42 +0000 (01:59 +0000)]
r20235: Don't use realloc directly.

16 years agor20229: Avoid a silly function call in mkdir_internal. None of the callers look at
Volker Lendecke [Sun, 17 Dec 2006 18:41:44 +0000 (18:41 +0000)]
r20229: Avoid a silly function call in mkdir_internal. None of the callers look at
errno, all go straight to ERROR_NT(status).

Volker

16 years agor20228: Bring the calling conventions of inherit_access_acl and change_owner_to_parent
Volker Lendecke [Sun, 17 Dec 2006 18:30:44 +0000 (18:30 +0000)]
r20228: Bring the calling conventions of inherit_access_acl and change_owner_to_parent
a bit closer together: Move the lp_inherit_perms() check into the callers.

Volker

16 years agor20227: Simplify logic in vfs_MkDir() a bit
Volker Lendecke [Sun, 17 Dec 2006 18:10:15 +0000 (18:10 +0000)]
r20227: Simplify logic in vfs_MkDir() a bit

16 years agor20218: Same fix as r20217 -- can we consolidate that duplicated code a bit?
Volker Lendecke [Sat, 16 Dec 2006 18:41:22 +0000 (18:41 +0000)]
r20218: Same fix as r20217 -- can we consolidate that duplicated code a bit?

16 years agor20217: uid_t and gid_t are not necessarily 32 bit. In assignments we get an automatic
Volker Lendecke [Sat, 16 Dec 2006 18:37:33 +0000 (18:37 +0000)]
r20217: uid_t and gid_t are not necessarily 32 bit. In assignments we get an automatic
conversion, but not when we pass pointers down to other functions.

Simo, please check.

Volker

16 years agor20216: Fix fallback code.
Simo Sorce [Sat, 16 Dec 2006 18:13:12 +0000 (18:13 +0000)]
r20216: Fix fallback code.
A reversed check made it impossile to fallback to the Unix Domain mapping code.
Also fix a potential use of a freed array.

Jerry,
my tests shows that this code now correctly handle the fallback to Unix Domain
when our Domain member is asked for a mapped group that has a unix name different
from the Windows name against a Samba DC and we do not use winbindd but share
users/groups by other means (ldap / sync of passwd and group files)

Immediate Fix would be to discuss if we should answer back when DOMAIN\unixgroup -> SID
is asked for, in the case the unixgroup name is mapped to a different name.
IE: DOMAIN\Domain Admins -> ntadmins

Currently if we are asked for "DOMAIN\Domain Admins" we return the dom admins SID
If we are asked for "DOMAIN\ntadmins we return "not found", but we may consider to
return the Domain admins SID in this case too.

Comments are welcome on this point!

Long term fix I think is the unixinfo pipe and of course an idmap_unixinfo moudle.

Simo.

16 years agor20215: Next step trying to fix the Solaris build.
Volker Lendecke [Sat, 16 Dec 2006 17:54:16 +0000 (17:54 +0000)]
r20215: Next step trying to fix the Solaris build.

I think "anonimous" is correctly spelled "anonymous". The Solaris compile is
referring to this as "anonymous" in line 814 of smbldap.c. Simo, please check.

Thanks,

Volker

16 years agor20214: Fix a type-punned warning. Simo, Jerry, please check
Volker Lendecke [Sat, 16 Dec 2006 17:47:22 +0000 (17:47 +0000)]
r20214: Fix a type-punned warning. Simo, Jerry, please check

16 years agor20212: Attempt to fix the Solaris build
Volker Lendecke [Sat, 16 Dec 2006 11:15:03 +0000 (11:15 +0000)]
r20212: Attempt to fix the Solaris build

16 years agor20211: send_nt_replies does not need the inbuf arg
Volker Lendecke [Sat, 16 Dec 2006 10:36:25 +0000 (10:36 +0000)]
r20211: send_nt_replies does not need the inbuf arg

16 years agor20210: registry_fetch_values is not needed anymore, fix typo
Volker Lendecke [Sat, 16 Dec 2006 09:36:10 +0000 (09:36 +0000)]
r20210: registry_fetch_values is not needed anymore, fix typo

16 years agor20209: Fix two memleaks
Volker Lendecke [Sat, 16 Dec 2006 09:33:17 +0000 (09:33 +0000)]
r20209: Fix two memleaks

16 years agor20208: Change sprintf_append() never to use malloc,
Jeremy Allison [Sat, 16 Dec 2006 05:02:21 +0000 (05:02 +0000)]
r20208: Change sprintf_append() never to use malloc,
but always use a talloc context.
Thanks to simo for pointing this out.
Jeremy.

16 years agor20207: Fix a couple more places where extra_data was
Jeremy Allison [Sat, 16 Dec 2006 01:52:06 +0000 (01:52 +0000)]
r20207: Fix a couple more places where extra_data was
being talloc'ed off the NULL context instead
of being malloced.
Jeremy.

16 years agor20206: Start cleaning up the talloc_ctx mess.
Jeremy Allison [Sat, 16 Dec 2006 01:32:57 +0000 (01:32 +0000)]
r20206: Start cleaning up the talloc_ctx mess.
child->mem_ctx isn't actually used for
anything, so remove it.
Jeremy.

16 years agor20202: fix make clean
Herb Lewis [Sat, 16 Dec 2006 00:51:15 +0000 (00:51 +0000)]
r20202: fix make clean

16 years agor20201: merge makefile changes from 3_0_24
Herb Lewis [Sat, 16 Dec 2006 00:25:34 +0000 (00:25 +0000)]
r20201: merge makefile changes from 3_0_24

16 years agor20197: merge talloc fixes from samba4:
Stefan Metzmacher [Fri, 15 Dec 2006 22:58:06 +0000 (22:58 +0000)]
r20197: merge talloc fixes from samba4:

- make most static functions inline
- handle NULL pointers in talloc_parent_chunk()
- use talloc_parent_chunk() in talloc_parent_name()
  to fix a bug found by the IBM checker

metze

16 years agor20186: Fix winbind crash bug in WINBIND_GETGROUPS.
Günther Deschner [Fri, 15 Dec 2006 16:45:39 +0000 (16:45 +0000)]
r20186: Fix winbind crash bug in WINBIND_GETGROUPS.

response_extra_sent() expects to free a malloced
extra_data.data while the add_XX_to_array functions all return talloced
memory now. Jeremy, please check.

Guenther

16 years agor20183: Fix the build.
Günther Deschner [Fri, 15 Dec 2006 12:35:20 +0000 (12:35 +0000)]
r20183: Fix the build.

Guenther

16 years agor20181: Improve NT status to pam error mapping.
Jeremy Allison [Fri, 15 Dec 2006 06:44:16 +0000 (06:44 +0000)]
r20181: Improve NT status to pam error mapping.
Jeremy.

16 years agor20180: Ensure that pam returns the correct error messages
Jeremy Allison [Fri, 15 Dec 2006 06:06:15 +0000 (06:06 +0000)]
r20180: Ensure that pam returns the correct error messages
when offline and or doing password changes.
Jeremy.

16 years agor20179: Sync up with Samba4 - remove blank lines at the
Jeremy Allison [Fri, 15 Dec 2006 01:50:04 +0000 (01:50 +0000)]
r20179: Sync up with Samba4 - remove blank lines at the
end parsing a file.
Jeremy.

16 years agor20178: Ensure we allocate the intermediate trans structs
Jeremy Allison [Fri, 15 Dec 2006 00:49:12 +0000 (00:49 +0000)]
r20178: Ensure we allocate the intermediate trans structs
off conn->mem_ctx, not the null context so we can
safefy free everything on conn close. Should fix
possible memleak.
Jeremy.

16 years agor20174: If we're only going to call one handler per message
Jeremy Allison [Thu, 14 Dec 2006 22:11:17 +0000 (22:11 +0000)]
r20174: If we're only going to call one handler per message
then terminate the traversal once we've done that.
Jeremy.

16 years agor20173: DNS update fixes:
Gerald Carter [Thu, 14 Dec 2006 17:00:10 +0000 (17:00 +0000)]
r20173: DNS update fixes:

* Fix DNS updates for multi-homed hosts
* Child domains often don't have an NS record in
  DNS so we have to fall back to looking up the the NS
  records for the forest root.
* Fix compile warning caused by mismatched 'struct in_addr'
  and 'in_addr_t' parameters called to DoDNSUpdate()

16 years agor20171: Don't delete the krb5 credential if others still reference to it.
Günther Deschner [Thu, 14 Dec 2006 16:34:24 +0000 (16:34 +0000)]
r20171: Don't delete the krb5 credential if others still reference to it.

Guenther

16 years agor20170: Fix secure DNS updates to work against
Gerald Carter [Thu, 14 Dec 2006 16:27:45 +0000 (16:27 +0000)]
r20170: Fix secure DNS updates to work against
Wnidows 2000 DNS which expects the TKEY payload to
be in the answer section and not in the additional
set of records (like Windows 2003 and the RFC).

16 years agor20169: Support for fallback to legacy mapping code was not completely tested.
Simo Sorce [Thu, 14 Dec 2006 15:30:54 +0000 (15:30 +0000)]
r20169: Support for fallback to legacy mapping code was not completely tested.
Add necessary fixes.

16 years agor20165: Change messaging subsystem to only allow one message
Jeremy Allison [Thu, 14 Dec 2006 01:00:16 +0000 (01:00 +0000)]
r20165: Change messaging subsystem to only allow one message
per type - this is all we use right now and makes
re-entrancy problems with deleting handlers with
a message dispatch loop go away.
Jeremy.

16 years agor20160: Fix long-standing (ie. from initial code I think) bug
Jeremy Allison [Wed, 13 Dec 2006 22:19:10 +0000 (22:19 +0000)]
r20160: Fix long-standing (ie. from initial code I think) bug
in tdb message processing. If we're inside a dispatch
function and we delete our own handler we'd walk onto
the next pointer from a deleted memory block. Fixes
crash bug in winbindd (and goodness knows where else).
Jeremy.

16 years agor20155: revert, I misread
Simo Sorce [Wed, 13 Dec 2006 21:05:47 +0000 (21:05 +0000)]
r20155: revert, I misread

16 years agor20154: Fix nasty typo, hunting another bug
Simo Sorce [Wed, 13 Dec 2006 21:03:38 +0000 (21:03 +0000)]
r20154: Fix nasty typo, hunting another bug

16 years agor20151: remove meaningless checks
Simo Sorce [Wed, 13 Dec 2006 18:29:00 +0000 (18:29 +0000)]
r20151: remove meaningless checks

16 years agor20150: better memory handling for some functions, make sure we don't
Simo Sorce [Wed, 13 Dec 2006 16:39:50 +0000 (16:39 +0000)]
r20150: better memory handling for some functions, make sure we don't
leak memory by using the wrong(long lived)  mem context

16 years agor20148: Forgot to merge. Ensure open_any_socket_out()
Jeremy Allison [Wed, 13 Dec 2006 05:54:39 +0000 (05:54 +0000)]
r20148: Forgot to merge. Ensure open_any_socket_out()
is safe from signals.
Jeremy.

16 years agor20146: Now online checks are fully async we can do them
Jeremy Allison [Wed, 13 Dec 2006 01:11:29 +0000 (01:11 +0000)]
r20146: Now online checks are fully async we can do them
every cache timeout times.
Jeremy.

16 years agor20140: Make online/offline detection completely asynchronous.
Jeremy Allison [Tue, 12 Dec 2006 22:41:42 +0000 (22:41 +0000)]
r20140: Make online/offline detection completely asynchronous.
Now I've done this I might be able to reduce the probe
timeout and reduce the backoff algorithm, going back
to checking every cache time seconds (5 mins by default),
as the parent or forked domain child will never block.
Jeremy.

16 years agor20136: Fix #4290. Properly compute time to password expiration in message from
Jim McDonough [Tue, 12 Dec 2006 22:05:48 +0000 (22:05 +0000)]
r20136: Fix #4290.  Properly compute time to password expiration in message from
pam_winbind.  Thanks to Andrew Benham <andrew.benham@thus.net>