Björn Jacke [Tue, 18 Oct 2011 08:54:56 +0000 (10:54 +0200)]
s3/doc: add man page for aio_fork vfs module
thanks to Volker for the content
Autobuild-User: Björn Jacke <bj@sernet.de>
Autobuild-Date: Tue Oct 18 12:24:35 CEST 2011 on sn-devel-104
(cherry picked from commit 56328a4d61c8d0a52f6841097bf8fc4ffd46bfb6)
Fix uninitialized memory problem in group_sids_to_info3 (fixes bug #8455).
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Mon Oct 17 23:32:58 CEST 2011 on sn-devel-104
(cherry picked from commit c52b571506874987ba626c25e9692fbe2251b7e2)
s3: Fix bug 8455 -- Samba PDC is looking up only primary user group
group_sids_to_info3 does a sid_peek_check_rid on the domain sid before adding
the rids to the array. If the domain sid is 0x0, then the check will always
fail.
Autobuild-User: Volker Lendecke <vlendec@samba.org>
Autobuild-Date: Sat Sep 17 00:51:27 CEST 2011 on sn-devel-104
(cherry picked from commit 3dcec44f3edbc9c4f1946ead3480f6d01cd53e7a)
Jeremy Allison [Wed, 12 Oct 2011 21:00:32 +0000 (14:00 -0700)]
Another part of the fix for Bug 8473 - smb2_find uses a hard coded max reply size of 0x10000 instead of smb2_max_trans.
Change the defaults and the documentation to reflect the 64k limit.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User: Stefan Metzmacher <metze@samba.org>
Autobuild-Date: Thu Oct 13 12:22:23 CEST 2011 on sn-devel-104
(cherry picked from commit d2a9e8df35be1516f22f0ef78356de6d4b02d4d3)
s3:smb2_server: get/set info are limited by max_trans size (bug #8473)
metze
Autobuild-User: Stefan Metzmacher <metze@samba.org>
Autobuild-Date: Thu Oct 13 03:32:02 CEST 2011 on sn-devel-104
(cherry picked from commit 5e04231e961f10656384a6f16104d0d55b6f3e0e)
s3:smbd: disconnect the socket if we got an unexpected request
If we got a SMB2_OP_NEGPROT after the protocol is already negotiated
or if we got a non SMB2_OP_NEGPROT before the protocol was negotiated
we should close the connection (as windows does).
metze
Autobuild-User: Stefan Metzmacher <metze@samba.org>
Autobuild-Date: Fri Sep 23 12:30:50 CEST 2011 on sn-devel-104
(cherry picked from commit 5494856294384df4360d85e68abc03c43dc16e06)
Fix bug #8515 - Empty CIFS share can be blocked for other clients by deleting it via empty path (DELETE_PENDING until the last client)
Disallow "." in can_set_delete_on_close().
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Wed Oct 12 21:07:27 CEST 2011 on sn-devel-104
(cherry picked from commit bd260f03ab492d03c2890db47dc6fb4f1b824a1a)
s3:smb2_server: return OBJECT_NAME_INVALID if the path is terminated in SMB2_FIND/QUERY_DIRECTORY
metze
Autobuild-User: Stefan Metzmacher <metze@samba.org>
Autobuild-Date: Wed Sep 7 12:15:51 CEST 2011 on sn-devel-104
(cherry picked from commit 9bc4decc1cba701926fc8081c3903aac754a6f51)
Christian Ambach [Mon, 10 Oct 2011 15:48:18 +0000 (17:48 +0200)]
s3:smb2_create: fix allocation size return value when opening existing files
at least on GPFS, using the stat information gave wrong results
example: FileInfo gave 12582912000, Create Response gave 25769803776000
This makes the create part use the same method as fileinfo,
matching up the replies of both calls
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Mon Oct 10 21:49:53 CEST 2011 on sn-devel-104
(cherry picked from commit ec0c1f2c34ea520b4a7d636ba284a9c7d3ecd96f)
Jeremy Allison [Wed, 7 Sep 2011 21:57:11 +0000 (14:57 -0700)]
Tidy up previous patch for bug #8442 - NFSv4 DENY ACLs always include SYNCHRONIZE flag - blocking renames.
aceType is an enum field, not a bitmask.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Thu Sep 8 03:34:39 CEST 2011 on sn-devel-104
(cherry picked from commit 0aa6751e919d842cdae044be26f6da725063eac2)
Ira Cooper [Wed, 7 Sep 2011 19:24:22 +0000 (12:24 -0700)]
Fix bug #8442 - NFSv4 DENY ACLs always include SYNCHRONIZE flag - blocking renames.
Thanks to Youzhong Yang for discovering this issue.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Wed Sep 7 22:56:06 CEST 2011 on sn-devel-104
(cherry picked from commit 600451da33728cbd377b122c85cf146f5b319998)
Karolin Seeger [Sat, 8 Oct 2011 18:54:37 +0000 (20:54 +0200)]
s3-docs: Fix typo.
Karolin
Autobuild-User: Karolin Seeger <kseeger@samba.org>
Autobuild-Date: Sat Oct 8 22:27:07 CEST 2011 on sn-devel-104
(cherry picked from commit 851d6625d5e08ac9659a6d23d105f9e0c47f9692)
Bug 7551: Return error of cli_push when 'put - /some/file' is used
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Thu Sep 29 23:47:02 CEST 2011 on sn-devel-104
(cherry picked from commit d883cc664cac81633a60e5b04f99f23a3577ae65)
Jeremy Allison [Fri, 7 Oct 2011 18:09:44 +0000 (11:09 -0700)]
Fix bug #8509 - Read-only handles on SAMR allow SAMR_DOMAIN_ACCESS_CREATE_USER.
Not a security issue as we also check inside _samr_CreateUser2.
Thanks to Andreas Schneider <asn@samba.org> for finding and testing this.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Fri Oct 7 21:51:27 CEST 2011 on sn-devel-104
(cherry picked from commit c80ba57169cee2ec66e8afe3616956c17958a3ae)
Jeremy Allison [Tue, 4 Oct 2011 21:46:00 +0000 (14:46 -0700)]
Fix bug #8507 - smbd doesn't correctly honor the "force create mode" bits from a cifsfs create.
Don't manipulate the new_dos_attributes bits until we know it's not a POSIX open.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Wed Oct 5 01:19:17 CEST 2011 on sn-devel-104
(cherry picked from commit f93fd128eb623e324459806a34052bd40afe82e5)
Jeremy Allison [Thu, 8 Sep 2011 20:48:27 +0000 (13:48 -0700)]
Second part of fix for bug #8443 - Default user entry is set to minimal permissions on incoming ACL change with no user specified.
Be smarter about setting default permissions when a ACL_USER_OBJ isn't given. Use the principle of least surprises for the user.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Fri Sep 9 00:26:08 CEST 2011 on sn-devel-104
(cherry picked from commit e30b8c72def13e2abc14858ea64eb849ea665b80)
Jeremy Allison [Thu, 8 Sep 2011 19:51:18 +0000 (12:51 -0700)]
First part of fix for bug #8443 - Default user entry is set to minimal permissions on incoming ACL change with no user specified.
create_default_mode() is not needed - it's taken care of by code
inside ensure_canon_entry_valid().
(cherry picked from commit 793bd527fdd0b188aba8f3b4bffd8fa8f69a9cd1)
Jeremy Allison [Fri, 2 Sep 2011 19:22:34 +0000 (12:22 -0700)]
Part 3 of bugfix for bug #7509 - smb_acl_to_posix: ACL is invalid for set (Invalid argument)
Don't call check_owning_objs() to convert ACL_USER->ACL_USER_OBJ and
AC_GROUP->ACL_GROUP_OBJ for default (directory) ACLs, we do this separately
inside ensure_canon_entry_valid().
(cherry picked from commit c528fc5cacaae7e0e83041eb98150052b436071e)
Jeremy Allison [Fri, 2 Sep 2011 18:58:56 +0000 (11:58 -0700)]
Part 2 of bugfix for bug #7509 - smb_acl_to_posix: ACL is invalid for set (Invalid argument)
Only map CREATOR_OWNER/CREATOR_GROUP to ACL_USER_OBJ/ACL_GROUP_OBJ in
a default(directory) ACL set.
(cherry picked from commit a5038ace24559bb02eec8262d3af5b5e78634d16)
Jeremy Allison [Fri, 2 Sep 2011 18:21:08 +0000 (11:21 -0700)]
Part 1 of bugfix for bug #7509 - smb_acl_to_posix: ACL is invalid for set (Invalid argument)
Remove the code I added for bug "6878 - Cannot change ACL's inherit flag". It is incorrect
and causes the POSIX ACL ACL_USER_OBJ duplication.
(cherry picked from commit 2b935b49f3d975759eb1cbcf2b11bf7c9d982804)
s3:smb2_server: SMB2_OP_CANCEL requests don't have to be signed
metze
Autobuild-User: Stefan Metzmacher <metze@samba.org>
Autobuild-Date: Thu Sep 29 02:58:41 CEST 2011 on sn-devel-104
(cherry picked from commit ea00f0e452f64868fdc78cf0dc8e9ce3af19fe32)
Fix bug #8503 (The SMB2 server should ignore signing on CANCEL requests).
Jeremy Allison [Tue, 27 Sep 2011 17:23:08 +0000 (10:23 -0700)]
These modules are no longer experimental but production-ready (especially the acl_xattr code). Remove the "experimental" tag.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Tue Sep 27 21:00:12 CEST 2011 on sn-devel-104
(cherry picked from commit fa2236c066890b687483408ed0d82a4bb1f95157)
Fix bug #8494 (Remove "experimental" label on VFS ACL modules).
David Disseldorp [Fri, 23 Sep 2011 18:46:06 +0000 (11:46 -0700)]
Fix bug 8480 - acl_xattr can free an invalid pointer if no blob is loaded.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Fri Sep 23 22:20:55 CEST 2011 on sn-devel-104
(cherry picked from commit ca8c25abebb8715176f68be9916217db9a8400dd)
Jeremy Allison [Thu, 22 Sep 2011 16:06:48 +0000 (09:06 -0700)]
Fix bug #8476 - Samba asserts when SMB2 client breaks the crediting rules.
Just drop the connection, not SMB_ASSERT.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Thu Sep 22 19:41:31 CEST 2011 on sn-devel-104
(cherry picked from commit d50fa9c21e1c3b6f7126c33e24966099c7567225)
Jeremy Allison [Thu, 22 Sep 2011 20:42:56 +0000 (13:42 -0700)]
Fix bug #8477 - Map to guest can return uninitialized blob of data.
Found by Codenomicon at SNIA SDC.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Fri Sep 23 03:19:46 CEST 2011 on sn-devel-104
(cherry picked from commit f0f91d0117b2ccf778382e61a411b5e4f0ea2b14)
Jeremy Allison [Wed, 31 Aug 2011 00:37:19 +0000 (17:37 -0700)]
Fix bug 8429 - Compound SMB2 requests on an IPC connection can corrupt the reply stream.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Wed Aug 31 21:18:11 CEST 2011 on sn-devel-104
(cherry picked from commit 786fe9fab223723e4d2340f285592b2a44945d73)
because from the echo responder we always read using the normal smb1 protocol
handling routine. If that is a bit down the smb2 stream, we get a non-negprot
packet and panic.
BTW, the echo responder is not required for smb2 anyway, Microsoft confirmed
that it probes the server liveness using TCP keepalives and not smb2 echo
requests.
The last 2 patches address bug #8334 (smb2: smbd logs Invalid SMB packet: first
request: 0x0008 and crashes).
s3:smb2_server: make sure we prefer responses over requests on the client socket
metze
Autobuild-User: Stefan Metzmacher <metze@samba.org>
Autobuild-Date: Fri Aug 12 16:46:43 CEST 2011 on sn-devel-104
(cherry picked from commit 42cde0480bd6a5e2dddaa66917e1fa71e6a4edcd)
According to [MS-SMB2] 3.3.5.9.7
(http://msdn.microsoft.com/en-us/library/cc246784%28v=PROT.13%29.aspx),
smbd must reply with NT_STATUS_OBJECT_NAME_NOT_FOUND as it does not
support durable file-handles yet.
I have seen w2k8r2 running xcopy /C ending up in an endless loop
trying to get back the original file handle from smbd sending the same
requests over and over.
Jeremy Allison [Mon, 29 Aug 2011 23:47:16 +0000 (16:47 -0700)]
Fix bug 8412 - Microsoft Office 2007 (Microsoft Word) fails to save as on a Samba share with SMB2. (cherry picked from commit dec3b21cd1737b317749e7ebced5aa1c2115ebdf)
Volker Lendecke [Fri, 26 Aug 2011 09:38:21 +0000 (11:38 +0200)]
s3: Use sys_read in fork_domain_child
I've seen
[2011/08/26 01:44:10.872057, 1] winbindd/winbindd_dual.c:1336(fork_domain_child)
fork_domain_child: Could not read child status: nread=-1, error=Interrupted system call
Volker Lendecke [Fri, 26 Aug 2011 09:39:41 +0000 (11:39 +0200)]
s3: Use sys_write in fork_domain_child
Counterpart for last checkin. A lot less likely, but not impossible in a child.
Autobuild-User: Volker Lendecke <vlendec@samba.org>
Autobuild-Date: Fri Aug 26 13:14:27 CEST 2011 on sn-devel-104
(cherry picked from commit e0e3d215b1902ecbeb3d469af9a196856711921e)
Volker Lendecke [Fri, 26 Aug 2011 14:54:18 +0000 (16:54 +0200)]
s3: Fix a winbind race leading to 100% CPU
This fixes a race condition that leads to the winbindd_children list becoming
corrupted. It happens when on a busy winbind SIGCHLD is a bit late.
Imagine a winbind with multiple requests in the queue for a single child. Child
dies, and before the SIGCHLD handler is called we find the socket to be dead.
wb_child_request_done is called, receiving an error from wb_simple_trans_recv.
It closes the socket. Then immediately the wb_child_request_trigger will do
another fork_domain_child before the signal handler is called. This means that
we do another fork_domain_child, we have child->sock==-1 at this point.
fork_domain_child will do a DLIST_ADD(winbindd_children, child) a second time
where the child is already part of that list. This corrupts the list. Then the
signal handler kicks in, spinning in
forever. Not good. This patch makes sure that both conditions (sock==-1 and not
part of the list) for a winbindd_child struct match up.
Autobuild-User: Volker Lendecke <vlendec@samba.org>
Autobuild-Date: Fri Aug 26 18:51:24 CEST 2011 on sn-devel-104
(cherry picked from commit ed058f4dc3e059967f3c0538bc220866e9c63588)
SATOH Fumiyasu [Mon, 22 Aug 2011 04:48:40 +0000 (13:48 +0900)]
s3: Fix the build of vfs_aixacl2.c
Autobuild-User: Volker Lendecke <vlendec@samba.org>
Autobuild-Date: Thu Aug 25 23:06:12 CEST 2011 on sn-devel-104
(cherry picked from commit faa769ec19b6486675048734837a6e2237d6f0a4)
Volker Lendecke [Mon, 22 Aug 2011 12:16:26 +0000 (14:16 +0200)]
s3: Fix bug 8385
Poll and select behave differently regarding error handling. When doing the
connect(2), we can not rely on poll telling us both readability and writability
upon error. Just always try a second connect(2). At least on Linux it returns 0
when it succeeded.
Signed-off-by: Jeremy Allison <jra@samba.org>
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Thu Aug 25 19:39:12 CEST 2011 on sn-devel-104
(cherry picked from commit 50e30afa608dfdeae8a260730ead9761ed424dad)
Missing assignment means this loop will never terminate. Need to be applied
to 3.5.x and 3.6.1.
(cherry picked from commit 218cccfdf1820a9840a5cbc7df3915b6b58a762d)
Ira Cooper [Mon, 22 Aug 2011 19:46:41 +0000 (12:46 -0700)]
Fix bug #8395 - optimize serverid_exists() for Solaris.
jra asked:
Do you have any idea on how many calls it saves,
and what help the optimization gives ?
As far as calls: Not 100% sure, dtrace pointed the issue out to me, and when I
saw 10-50%(Or more), I looked for an easy fix.
The real expense here is the fnctl locks on the database. That's pretty
painful. 1000ns per call, and probably 3-4 calls in that chain. So it doesn't
take much to come out ahead was my math.
Andrew Tridgell [Fri, 29 Jul 2011 01:57:07 +0000 (11:57 +1000)]
talloc: added test suite for talloc_free_children()
this tests the fix from Simo
Autobuild-User: Andrew Tridgell <tridge@samba.org>
Autobuild-Date: Fri Jul 29 11:30:13 CEST 2011 on sn-devel-104
(cherry picked from commit d004fd0b53fb6f3ae64f0e24cf51f4471d434574)
Autobuild-User: Jelmer Vernooij <jelmer@samba.org>
Autobuild-Date: Sun Apr 24 03:27:54 CEST 2011 on sn-devel-104
(cherry picked from commit fb05e82c99f0779bd44371a2bdafdd7147448dd5)
Andrew Tridgell [Mon, 8 Aug 2011 08:24:32 +0000 (18:24 +1000)]
talloc: ensure the sibling linked list remains valid during a free
This ensures that the sibling list of a pointer doesn't become invalid
during a free operation. It is an alternative fix to the fix in 6f51a1f45bf4de062cce7a562477e8140630a53d, and avoids the problem of
trying to calculate the parent pointer early
This should fix the subtle spoolss talloc bug that Simo found
Autobuild-User: Andrew Tridgell <tridge@samba.org>
Autobuild-Date: Tue Aug 9 01:53:17 CEST 2011 on sn-devel-104
(cherry picked from commit cf986f200804ce873b43c1ecf2d5e1bd08eb8a25)