ira/wip.git
15 years agoAvoid valgrind errors
Volker Lendecke [Tue, 27 Jan 2009 18:45:51 +0000 (19:45 +0100)]
Avoid valgrind errors

In event handlers, we might destroy other events that are pending in the lists.
We can only run one event safely per select call.

Yes, I've seen these valgrind errors :-)

Jeremy, with ccdd921e61 you had checked in the change to run multiple events.
Do you remember why it was necessary and could not be solved in a different
way?

Volker

15 years agoFix a valgrind error when the socket dies
Volker Lendecke [Tue, 27 Jan 2009 18:41:34 +0000 (19:41 +0100)]
Fix a valgrind error when the socket dies

Don't reference anything that might have been deleted in the async_req_error
call.

15 years agos3:winbind: remove prototype for non-existent function from winbind_proto.h
Michael Adam [Wed, 28 Jan 2009 11:41:04 +0000 (12:41 +0100)]
s3:winbind: remove prototype for non-existent function from winbind_proto.h

Michael

15 years agos3:test: exit the net rpc registry tests when give_administrative_rights fails
Michael Adam [Wed, 28 Jan 2009 11:19:06 +0000 (12:19 +0100)]
s3:test: exit the net rpc registry tests when give_administrative_rights fails

Michael

15 years agos3:selftest: fix warnings with empty *_TEST_LOG variables
Stefan Metzmacher [Wed, 28 Jan 2009 11:03:59 +0000 (12:03 +0100)]
s3:selftest: fix warnings with empty *_TEST_LOG variables

metze

15 years agos3:registry: always provision the registry, not only in smbd.
Michael Adam [Wed, 28 Jan 2009 11:04:53 +0000 (12:04 +0100)]
s3:registry: always provision the registry, not only in smbd.

This will make tools like "net registry" also fill the registry
with the basic hives and keys.

Michael

15 years agoselftest: set 'lock dir' in client.conf
Stefan Metzmacher [Wed, 28 Jan 2009 10:59:26 +0000 (11:59 +0100)]
selftest: set 'lock dir' in client.conf

metze

15 years agoasync_sock: Move to top level
Kai Blin [Sat, 24 Jan 2009 09:00:13 +0000 (10:00 +0100)]
async_sock: Move to top level

15 years agos3:torture: support SMB_CONF_PATH envvar in smbtorture
Stefan Metzmacher [Wed, 28 Jan 2009 09:50:55 +0000 (10:50 +0100)]
s3:torture: support SMB_CONF_PATH envvar in smbtorture

metze

15 years agos3:util_tdb: remove duplicate prototypes (from ../lib/util/util_tdb.h)
Michael Adam [Wed, 28 Jan 2009 09:35:17 +0000 (10:35 +0100)]
s3:util_tdb: remove duplicate prototypes (from ../lib/util/util_tdb.h)

Michael

15 years agos3: try to make 'make selftest' work again
Stefan Metzmacher [Wed, 28 Jan 2009 08:35:44 +0000 (09:35 +0100)]
s3: try to make 'make selftest' work again

metze

15 years agos3:build: don't use $(FLAGS) in linking targets - it is useless
Michael Adam [Wed, 28 Jan 2009 08:42:42 +0000 (09:42 +0100)]
s3:build: don't use $(FLAGS) in linking targets - it is useless

Michael

15 years agos3:tdb_validate: clean up leading tabs/spaces and trailing spaces.
Michael Adam [Wed, 28 Jan 2009 07:56:17 +0000 (08:56 +0100)]
s3:tdb_validate: clean up leading tabs/spaces and trailing spaces.

Michael

15 years agos3: separate tdb validation code out into its own source file
Michael Adam [Wed, 28 Jan 2009 07:52:53 +0000 (08:52 +0100)]
s3: separate tdb validation code out into its own source file

So this gets now linked only into its single user: winbindd
(needed by winbindd_cache.c)

Michael

15 years agoFix detection of dns_sd libraries.
Jeremy Allison [Wed, 28 Jan 2009 00:19:09 +0000 (16:19 -0800)]
Fix detection of dns_sd libraries.
Jeremy.

15 years agoMove S3 specific parameters to S3 conf file in make test.
Steven Danneman [Tue, 27 Jan 2009 21:24:38 +0000 (13:24 -0800)]
Move S3 specific parameters to S3 conf file in make test.

15 years agos3: Fix shadowed declaration
Tim Prouty [Tue, 27 Jan 2009 18:37:51 +0000 (10:37 -0800)]
s3: Fix shadowed declaration

15 years agolibreplace: fix detection of netinet/ip.h on solaris 8
Michael Adam [Tue, 27 Jan 2009 14:46:25 +0000 (15:46 +0100)]
libreplace: fix detection of netinet/ip.h on solaris 8

(The test needs to additionally include <netinet/in_systm.h>.)

Michael

15 years agos3:rpc_server: use get_client_fd() instead of smbd_server_fd()
Stefan Metzmacher [Tue, 27 Jan 2009 08:48:48 +0000 (09:48 +0100)]
s3:rpc_server: use get_client_fd() instead of smbd_server_fd()

Both are absolutely the same, but smbd_server_fd() is otherwise
only used in smbd/*.c and would like to remove smbd_server_fd() soon.

metze

15 years agos3:smbd: wait for new connection via fd events
Stefan Metzmacher [Tue, 27 Jan 2009 09:02:43 +0000 (10:02 +0100)]
s3:smbd: wait for new connection via fd events

metze

15 years agos3:smbd: make mdns registration event driven.
Stefan Metzmacher [Mon, 26 Jan 2009 13:55:54 +0000 (14:55 +0100)]
s3:smbd: make mdns registration event driven.

metze

15 years agos3:smbd: make inetd code path more clear
Stefan Metzmacher [Fri, 23 Jan 2009 12:08:22 +0000 (13:08 +0100)]
s3:smbd: make inetd code path more clear

metze

15 years agos3:smbd: make kernel oplocks event driven
Stefan Metzmacher [Fri, 23 Jan 2009 09:08:44 +0000 (10:08 +0100)]
s3:smbd: make kernel oplocks event driven

And use signal events for Linux oplocks.

metze

15 years agos3:smbd: remove pointless respond_to_all_remaining_local_messages() function
Stefan Metzmacher [Fri, 23 Jan 2009 09:07:45 +0000 (10:07 +0100)]
s3:smbd: remove pointless respond_to_all_remaining_local_messages() function

This the process_kernel_oplock() function never response to messages,
it only generates messages to ourself.

metze

15 years agos3:smbd: convert aio to use tevent_signal
Stefan Metzmacher [Thu, 22 Jan 2009 17:04:17 +0000 (18:04 +0100)]
s3:smbd: convert aio to use tevent_signal

metze

15 years agos3:nmbd: handle SIG_TERM and SIGHUP via tevent
Stefan Metzmacher [Thu, 22 Jan 2009 15:17:23 +0000 (16:17 +0100)]
s3:nmbd: handle SIG_TERM and SIGHUP via tevent

metze

15 years agos3:nmbd: as the sig_term() handler only sets a flag we don't need to block SIGTERM
Stefan Metzmacher [Fri, 23 Jan 2009 09:15:00 +0000 (10:15 +0100)]
s3:nmbd: as the sig_term() handler only sets a flag we don't need to block SIGTERM

The arguments of commit d98bea900ee694cdba83149620c65bd7f8765f26
are no longer valid.

metze

15 years agos3:nmbd: install the SIG_DFL handler for SIGTERM while we're waiting for interfaces
Stefan Metzmacher [Fri, 23 Jan 2009 09:13:29 +0000 (10:13 +0100)]
s3:nmbd: install the SIG_DFL handler for SIGTERM while we're waiting for interfaces

We should handle all 3 cases where we actively wait for interfaces
in the same way.

metze

15 years agos3:winbindd: handle SIG_TERM, SIGHUP, SIGCHLD and SIGUSR2 via tevent
Stefan Metzmacher [Thu, 22 Jan 2009 13:54:21 +0000 (14:54 +0100)]
s3:winbindd: handle SIG_TERM, SIGHUP, SIGCHLD and SIGUSR2 via tevent

metze

15 years agos3:smbd: use signal events for SIGTERM, SIGHUP and SIGCHLD
Stefan Metzmacher [Wed, 21 Jan 2009 22:24:18 +0000 (23:24 +0100)]
s3:smbd: use signal events for SIGTERM, SIGHUP and SIGCHLD

metze

15 years agos3:smbd: move all code that handles a client connection into smbd_process()
Stefan Metzmacher [Thu, 22 Jan 2009 11:36:42 +0000 (12:36 +0100)]
s3:smbd: move all code that handles a client connection into smbd_process()

metze

15 years agos3:smbd: we should setup the client socket in the same way in all code pathes
Stefan Metzmacher [Thu, 22 Jan 2009 09:13:45 +0000 (10:13 +0100)]
s3:smbd: we should setup the client socket in the same way in all code pathes

We should behave the same in inetd, interactive and deamon modes.

metze

15 years agos3:namecache: remove unused namecache_shutdown() function
Stefan Metzmacher [Thu, 22 Jan 2009 09:19:14 +0000 (10:19 +0100)]
s3:namecache: remove unused namecache_shutdown() function

metze

15 years agotevent: change SA_INFO_QUEUE_COUNT from 10 to 100
Stefan Metzmacher [Fri, 23 Jan 2009 09:42:31 +0000 (10:42 +0100)]
tevent: change SA_INFO_QUEUE_COUNT from 10 to 100

The samba3 aio code requires lp_max_mux() with a default of 50
and the samba3 linux oplock code requires 100.

May we could make the size dynamic later.

metze

15 years agoasync_req: Fix the S4 build
Kai Blin [Tue, 27 Jan 2009 09:37:35 +0000 (10:37 +0100)]
async_req: Fix the S4 build

15 years agolibreplace: fix bug #6066 - netinet/ip.h present but cannot be compiled
Michael Adam [Tue, 27 Jan 2009 11:11:49 +0000 (12:11 +0100)]
libreplace: fix bug #6066 - netinet/ip.h present but cannot be compiled

under solaris

Michael

15 years agos3 build: Eliminate the gmake-specific Makefile syntax
Tim Prouty [Tue, 27 Jan 2009 00:50:31 +0000 (16:50 -0800)]
s3 build: Eliminate the gmake-specific Makefile syntax

15 years agos3 make test: Fix a bug in the previous commit
Tim Prouty [Mon, 26 Jan 2009 23:38:19 +0000 (15:38 -0800)]
s3 make test: Fix a bug in the previous commit

Luckily the build farm caught this

15 years agos3 make test: Add the ability to specify a custom smb.conf for make test
Tim Prouty [Sun, 25 Jan 2009 22:38:44 +0000 (14:38 -0800)]
s3 make test: Add the ability to specify a custom smb.conf for make test

- Adds new -c <custom conf> option to selftest.sh that when specified
  adds a line to make test's server.conf: "include <custom conf>"
- Adds getopts processing to selftest.sh
- Changes selftest.sh shrdir arg to use -s <shrdir>
- Changes selftest.sh smbtorture4_path arg to use -t <smbtortur4 path>
- Adds configure option --with-selftest-custom-conf=<custom conf>
- Updates Makefile.in to take advantage of the new/changed parameters

15 years agoMerge branch 'master' of ssh://git.samba.org/data/git/samba
Jelmer Vernooij [Mon, 26 Jan 2009 17:14:41 +0000 (18:14 +0100)]
Merge branch 'master' of ssh://git.samba.org/data/git/samba

15 years agopyrpc: Raise proper NDR error codes with matching strings rather than mapping
Jelmer Vernooij [Mon, 26 Jan 2009 16:31:50 +0000 (17:31 +0100)]
pyrpc: Raise proper NDR error codes with matching strings rather than mapping
to NTSTATUS first.

15 years agoAdd function for mapping NDR error codes to strings.
Jelmer Vernooij [Mon, 26 Jan 2009 16:31:22 +0000 (17:31 +0100)]
Add function for mapping NDR error codes to strings.

15 years agoDocument default of the printing config variable.
Andreas Schneider [Mon, 26 Jan 2009 13:38:34 +0000 (14:38 +0100)]
Document default of the printing config variable.

Signed-off-by: Andreas Schneider <anschneider@suse.de>
15 years agodocs: Add manpage for vfs_shadow_copy2.
Karolin Seeger [Mon, 26 Jan 2009 13:17:23 +0000 (14:17 +0100)]
docs: Add manpage for vfs_shadow_copy2.

Karolin

15 years agoDecouple clistr_pull from struct cli_state->inbuf
Volker Lendecke [Mon, 26 Jan 2009 02:38:05 +0000 (03:38 +0100)]
Decouple clistr_pull from struct cli_state->inbuf

15 years agoFix an error path memleak
Volker Lendecke [Sun, 25 Jan 2009 19:33:22 +0000 (20:33 +0100)]
Fix an error path memleak

15 years agoMake smb_bytes_push_str deal with a NULL buf returning NULL
Volker Lendecke [Sun, 25 Jan 2009 19:32:59 +0000 (20:32 +0100)]
Make smb_bytes_push_str deal with a NULL buf returning NULL

15 years agoFix chain_reply for pipe reads
Volker Lendecke [Sun, 25 Jan 2009 11:22:20 +0000 (12:22 +0100)]
Fix chain_reply for pipe reads

The caller might have over-allocated reply->outbuf. Deal with that.

Sorry, Günther, for giving you so much pain ...

Volker

15 years agoMerge branch 'master' of ssh://git.samba.org/data/git/samba
Jelmer Vernooij [Sun, 25 Jan 2009 04:35:21 +0000 (05:35 +0100)]
Merge branch 'master' of ssh://git.samba.org/data/git/samba

15 years agoAdd prototypes required by samba-gtk.
Jelmer Vernooij [Sun, 25 Jan 2009 04:32:30 +0000 (05:32 +0100)]
Add prototypes required by samba-gtk.

15 years agoFix a segfault in parse_logentry
Volker Lendecke [Sat, 24 Jan 2009 14:08:14 +0000 (15:08 +0100)]
Fix a segfault in parse_logentry

entry->data_record.strings is type smb_ucs2, so the pointer arithmetic
increments by two :-)

15 years agoModify OneFS module to support new NTIMES interfaces
todd stecher [Sat, 24 Jan 2009 00:55:18 +0000 (16:55 -0800)]
Modify OneFS module to support new NTIMES interfaces

15 years agoExtend NTIMES to allow setting create_time
todd stecher [Fri, 23 Jan 2009 22:40:19 +0000 (14:40 -0800)]
Extend NTIMES to allow setting create_time

1) Add in smb_file_time struct to clarify code and make room for createtime.
2) Get and set create time from SMB messages.
3) Fixup existing VFS modules + examples Some OS'es allow for the
setting of the birthtime through kernel interfaces. This value is
generically used for Windows createtime, but is not settable in the
code today.

15 years agoThis change allows for the autoconfigre detection of sub-second time resolution in...
todd stecher [Fri, 23 Jan 2009 22:38:40 +0000 (14:38 -0800)]
This change allows for the autoconfigre detection of sub-second time resolution in the FreeBSD stat structure

15 years agos4-smbtorture: add my copyright.
Günther Deschner [Tue, 20 Jan 2009 23:33:01 +0000 (00:33 +0100)]
s4-smbtorture: add my copyright.

Guenther

15 years agos4-smbtorture: add BackupLog handling in eventlog test.
Günther Deschner [Tue, 20 Jan 2009 23:25:50 +0000 (00:25 +0100)]
s4-smbtorture: add BackupLog handling in eventlog test.

Guenther

15 years agos3-rpcclient: add eventlog_loginfo command.
Günther Deschner [Tue, 20 Jan 2009 22:01:58 +0000 (23:01 +0100)]
s3-rpcclient: add eventlog_loginfo command.

Guenther

15 years agos3-rpcclient: add eventlog_backuplog command.
Günther Deschner [Tue, 20 Jan 2009 21:18:29 +0000 (22:18 +0100)]
s3-rpcclient: add eventlog_backuplog command.

Guenther

15 years agos3-rpcclient: add eventlog_registerevsource command.
Günther Deschner [Tue, 20 Jan 2009 19:10:16 +0000 (20:10 +0100)]
s3-rpcclient: add eventlog_registerevsource command.

Guenther

15 years agos3-rpcclient: add eventlog_reporteventsource command.
Günther Deschner [Tue, 20 Jan 2009 18:26:41 +0000 (19:26 +0100)]
s3-rpcclient: add eventlog_reporteventsource command.

Guenther

15 years agos4-smbtorture: add test_GetLogIntormation test.
Günther Deschner [Tue, 20 Jan 2009 18:01:43 +0000 (19:01 +0100)]
s4-smbtorture: add test_GetLogIntormation test.

Guenther

15 years agos3-rpcclient: add eventlog_reportevent command.
Günther Deschner [Tue, 20 Jan 2009 18:09:53 +0000 (19:09 +0100)]
s3-rpcclient: add eventlog_reportevent command.

Guenther

15 years agos4-smbtorture: add ReportEventW test.
Günther Deschner [Tue, 20 Jan 2009 17:42:47 +0000 (18:42 +0100)]
s4-smbtorture: add ReportEventW test.

Guenther

15 years agos3-eventlog: add missing _eventlog_ReportEventAndSourceW stub call.
Günther Deschner [Thu, 22 Jan 2009 18:35:33 +0000 (19:35 +0100)]
s3-eventlog: add missing _eventlog_ReportEventAndSourceW stub call.

Guenther

15 years agos3: re-run make samba3-idl.
Günther Deschner [Sat, 24 Jan 2009 02:30:48 +0000 (03:30 +0100)]
s3: re-run make samba3-idl.

Guenther

15 years agoeventlog: fill in IDL for eventlog_OpenBackupEventLogW.
Günther Deschner [Tue, 20 Jan 2009 21:18:53 +0000 (22:18 +0100)]
eventlog: fill in IDL for eventlog_OpenBackupEventLogW.

Guenther

15 years agoeventlog: fill in IDL for eventlog_BackupEventLogW.
Günther Deschner [Tue, 20 Jan 2009 21:06:01 +0000 (22:06 +0100)]
eventlog: fill in IDL for eventlog_BackupEventLogW.

Guenther

15 years agoeventlog: fill in IDL for eventlog_DeregisterEventSource.
Günther Deschner [Tue, 20 Jan 2009 19:00:02 +0000 (20:00 +0100)]
eventlog: fill in IDL for eventlog_DeregisterEventSource.

Guenther

15 years agoeventlog: fill in IDL for eventlog_RegisterEventSourceW.
Günther Deschner [Tue, 20 Jan 2009 18:58:17 +0000 (19:58 +0100)]
eventlog: fill in IDL for eventlog_RegisterEventSourceW.

Guenther

15 years agoeventlog: add missing IDL (opnum 24) for eventlog_ReportEventAndSourceW.
Günther Deschner [Tue, 20 Jan 2009 18:20:26 +0000 (19:20 +0100)]
eventlog: add missing IDL (opnum 24) for eventlog_ReportEventAndSourceW.

Guenther

15 years agoeventlog: fill in IDL for eventlog_GetLogIntormation.
Günther Deschner [Tue, 20 Jan 2009 17:56:34 +0000 (18:56 +0100)]
eventlog: fill in IDL for eventlog_GetLogIntormation.

Guenther

15 years agoeventlog: fill in IDL for eventlog_ReportEventW.
Günther Deschner [Tue, 20 Jan 2009 17:32:01 +0000 (18:32 +0100)]
eventlog: fill in IDL for eventlog_ReportEventW.

Guenther

15 years agoRevert "s3-eventlog: trying to avoid to pass around full ndr structures in rpc_server."
Günther Deschner [Sat, 24 Jan 2009 01:59:24 +0000 (02:59 +0100)]
Revert "s3-eventlog: trying to avoid to pass around full ndr structures in rpc_server."

This reverts commit c0f1309010b79be67557af5ffbd1c429a3f3d99a.

15 years agos3-eventlog: fix various invalid memcpy in read_package_entry().
Günther Deschner [Sat, 24 Jan 2009 01:10:47 +0000 (02:10 +0100)]
s3-eventlog: fix various invalid memcpy in read_package_entry().

Guenther

15 years agos3-eventlog: fix buffer allocation in eventlog read call.
Günther Deschner [Sat, 24 Jan 2009 01:07:40 +0000 (02:07 +0100)]
s3-eventlog: fix buffer allocation in eventlog read call.

This broke with f6fa3080fee1b20df9f1968500840a88cf0ee592 back in April 2007...

Guenther

15 years agotalloc: Change the two other definitions of TALLOC_FREE to match the primary version
Tim Prouty [Sat, 24 Jan 2009 01:29:21 +0000 (17:29 -0800)]
talloc: Change the two other definitions of TALLOC_FREE to match the primary version

Eventually these two other definitions should be removed and all of
the TALLOC macros should live in the same header.  Until then, this
patch eliminates some build warnings.

15 years agos3 ldb: Fix vasprintf segfault that has been the source of build machine failures...
Tim Prouty [Sat, 24 Jan 2009 00:09:13 +0000 (16:09 -0800)]
s3 ldb: Fix vasprintf segfault that has been the source of build machine failures this week

The bug was introduced when converting samba3's ldb_search interface
to match samba4:
47951fc5d0085e124666b7667715bba98076820e
f3ecb701335b1129947127d4c45eef970b66875c

15 years agotalloc_free() copes with a NULL pointer just fine
Volker Lendecke [Fri, 23 Jan 2009 22:15:05 +0000 (23:15 +0100)]
talloc_free() copes with a NULL pointer just fine

Saved around 10kb of code on my box :-)

15 years agos3: OneFS unmappable sids support.
Dan Sledz [Fri, 9 Jan 2009 01:09:28 +0000 (17:09 -0800)]
s3: OneFS unmappable sids support.

Updates the onefs vfs module to add configurable behavior to deal
with sids that are unknown to us.  The best examples are aces that
come from robocopy/xcopy.

Adds the following share level options (which are specific to the vfs_onefs
module):

onefs: ignore unmappable sids (Default = false)
If this option is set to true, sids which could not be resolved to
a uid/gid are ignored.  If an unmappable sid is encountered as the owner
or group, the owner/group is converted to BUILTIN\Administrators.

onefs: unmappable sids ignore list (Default = empty)
Only the sids in the list are ignored.

onefs: ignore sacls (Default = false)
SACLs are ignored

onefs: unmappable sids deny everyone (Default = false)
If an unmappable sid is found in a deny ACE, the ACE's identity is
changed to Everyone.

15 years agos4: add ../lib/async_req to the s4 build
Kai Blin [Fri, 23 Jan 2009 11:25:49 +0000 (12:25 +0100)]
s4: add ../lib/async_req to the s4 build

15 years agoasync_req: Move to top level dir
Kai Blin [Fri, 23 Jan 2009 10:54:30 +0000 (11:54 +0100)]
async_req: Move to top level dir

15 years agodocs: Fix formatting issue in man libsmbclient.
Karolin Seeger [Fri, 23 Jan 2009 12:02:46 +0000 (13:02 +0100)]
docs: Fix formatting issue in man libsmbclient.

Karolin

15 years agos4: we're on the way to alhpa7
Stefan Metzmacher [Fri, 23 Jan 2009 10:24:25 +0000 (11:24 +0100)]
s4: we're on the way to alhpa7

metze

15 years agoApply same logic fix for #4308 Excel save operation corrupts file ACLs
Jeremy Allison [Fri, 23 Jan 2009 00:21:02 +0000 (16:21 -0800)]
Apply same logic fix for #4308 Excel save operation corrupts file ACLs
to NFSv4 ACL code as this uses the same flawed logic as posix_acls.c.
Jeremy.

15 years agoFix logic error in try_chown - we shouldn't arbitrarily chown
Jeremy Allison [Thu, 22 Jan 2009 23:57:41 +0000 (15:57 -0800)]
Fix logic error in try_chown - we shouldn't arbitrarily chown
to ourselves unless that was passed in.
Jeremy.

15 years agoSecond part of the attemt to fix #4308 - Excel save operation corrupts file ACLs.
Jeremy Allison [Thu, 22 Jan 2009 22:38:57 +0000 (14:38 -0800)]
Second part of the attemt to fix #4308 - Excel save operation corrupts file ACLs.
If the chown succeeds then the ACL set should also. Ensure this is the case
(refactor some of this code to make it simpler to read also).
Jeremy.

15 years agoAbstract away the transport in cli_pipe.c
Volker Lendecke [Thu, 22 Jan 2009 17:52:15 +0000 (18:52 +0100)]
Abstract away the transport in cli_pipe.c

Sorry for the monster checkin, I could not really find a way to do this in
steps.

15 years agoAdd the socket rpc client transport
Volker Lendecke [Thu, 22 Jan 2009 17:39:29 +0000 (18:39 +0100)]
Add the socket rpc client transport

15 years agoAdd the named pipe rpc client transport
Volker Lendecke [Thu, 22 Jan 2009 17:34:06 +0000 (18:34 +0100)]
Add the named pipe rpc client transport

15 years agoRPC_CLIENT_OBJ is the right variable for cli_pipe.o
Volker Lendecke [Thu, 22 Jan 2009 21:04:55 +0000 (22:04 +0100)]
RPC_CLIENT_OBJ is the right variable for cli_pipe.o

15 years agoAdd struct rpc_cli_transport
Volker Lendecke [Thu, 22 Jan 2009 17:50:37 +0000 (18:50 +0100)]
Add struct rpc_cli_transport

15 years agoMerge branch 'master' of ssh://git.samba.org/data/git/samba
Jelmer Vernooij [Thu, 22 Jan 2009 21:49:30 +0000 (22:49 +0100)]
Merge branch 'master' of ssh://git.samba.org/data/git/samba

15 years agos3: Fix warning
Tim Prouty [Thu, 22 Jan 2009 21:14:04 +0000 (13:14 -0800)]
s3: Fix warning

15 years agoAnother attempt to fix bug #4308 - Excel save operation corrupts file ACLs.
Jeremy Allison [Thu, 22 Jan 2009 18:57:10 +0000 (10:57 -0800)]
Another attempt to fix bug #4308 - Excel save operation corrupts file ACLs.
Simo is completely correct. We should be doing the chown *first*, and fail the
ACL set if this fails. The long standing assumption I made when writing the
initial POSIX ACL code was that Windows didn't control who could chown a file
in the same was as POSIX. In POSIX only root can do this whereas I wasn't sure
who could do this in Windows at the time (I didn't understand the privilege
model). So the assumption was that setting the ACL was more important (early
tests showed many failed ACL set's due to inability to chown). But now we have
privileges in smbd, and we must always fail an ACL set when we can't chown
first. The key that Simo noticed is that the CREATOR_OWNER bits in the ACL
incoming are relative to the *new* owner, not the old one. This is why the old
user owner disappears on ACL set - their access was set via the USER_OBJ in the
creator POSIX ACL and when the ownership changes they lose their access.

Patch is simple - just ensure we do the chown first before evaluating the
incoming ACL re-read the owners. We already have code to do this it just wasn't
rigorously being applied.
Jeremy.

15 years agoDo not start a transaction this way.
Simo Sorce [Thu, 22 Jan 2009 16:51:37 +0000 (11:51 -0500)]
Do not start a transaction this way.
Because we still want to commit any changes that successfully apply we
never want to do a global cancel, and because of how transaction nesting
works that means we never cancel any "transaction" at the single modify
operation level.

Let each operation start its own transaction so that the transaction is
properly canceled if any error is returned and half committed operations
(though plugins failing) are properly reverted on a per operation basis.

15 years agoFix a segfault: rpccli_* expect the reply_pdu to always be initialized
Volker Lendecke [Thu, 22 Jan 2009 16:53:22 +0000 (17:53 +0100)]
Fix a segfault: rpccli_* expect the reply_pdu to always be initialized

15 years agoRemove obsolete samr Python module - use samba.dcerpc.samr instead.
Jelmer Vernooij [Thu, 22 Jan 2009 14:23:32 +0000 (15:23 +0100)]
Remove obsolete samr Python module - use samba.dcerpc.samr instead.

15 years agoImplement as_sddl.
Jelmer Vernooij [Thu, 22 Jan 2009 13:49:51 +0000 (14:49 +0100)]
Implement as_sddl.

15 years agoSupport parsing sddl for security descriptors.
Jelmer Vernooij [Thu, 22 Jan 2009 13:37:59 +0000 (14:37 +0100)]
Support parsing sddl for security descriptors.

15 years agolib/replace: add defines to let the callers find out if pwrite and pread are thread...
Stefan Metzmacher [Thu, 22 Jan 2009 10:52:54 +0000 (11:52 +0100)]
lib/replace: add defines to let the callers find out if pwrite and pread are thread/fork safe

metze