Volker Lendecke [Thu, 30 Dec 2021 15:58:58 +0000 (16:58 +0100)]
smbd: Remove a few vfs_stat() calls
openat_pathref_fsp() does not need them anymore
Signed-off-by: Volker Lendecke <vl@samba.org> Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Fri Mar 11 19:19:21 UTC 2022 on sn-devel-184
Volker Lendecke [Sat, 8 Jan 2022 09:08:16 +0000 (10:08 +0100)]
smbd: Return ISLNK from non_widelink_open() in smb_fname
Soon we want to not require stat() calls before entering
openat_pathref_fsp() anymore but rely on the fstat on the O_PATH file
handle (alternatively the call to fstatat(AT_SYMLINK_NOFOLLOW)) done
properly from within fd_openat(). The callers of non_widelink_open()
expect the stat information to be correct in "smb_fname". Copy it in
case of not opening a symlink in the posix case.
Signed-off-by: Volker Lendecke <vl@samba.org> Reviewed-by: Ralph Boehme <slow@samba.org>
Volker Lendecke [Thu, 30 Dec 2021 15:49:45 +0000 (16:49 +0100)]
smbd: Don't require a valid stat for openat_pathref_fsp()
With the simplifications in non_widelink_open() (don't depend on the
is_directory fsp flag) the main reason for requiring a valid stat
struct in openat_pathref_fsp() is gone. With this change
openat_pathref_fsp() is now capable of being the very first (and
authoritative) name-referencing operation with openat(O_PATH) for a
name.
Without having the stat information around before calling
openat_pathref_fsp(), the call to check_same_dev_ino() becomes
obsolete here.
Signed-off-by: Volker Lendecke <vl@samba.org> Reviewed-by: Ralph Boehme <slow@samba.org>
Volker Lendecke [Tue, 8 Mar 2022 13:31:32 +0000 (14:31 +0100)]
smbd: No need to set O_DIRECTORY in openat_pathref_fsp()
If I read Linux' man 2 open right (and susv4 agrees), O_DIRECTORY is
around to make sure opendir() is not raced against non-directory
files. opendir() needs to make sure the underlying object is actually
a directory. O_DIRECTORY is not required for opening directories in
RDONLY mode, regardless of having O_PATH or not.
At this point in openat_pathref_fsp() we don't care about the type of
the underlying object, we do fstat() and distinguish between files and
directories later according to the mode returned from fstat().
Signed-off-by: Volker Lendecke <vl@samba.org> Reviewed-by: Ralph Boehme <slow@samba.org>
Volker Lendecke [Tue, 8 Mar 2022 11:44:33 +0000 (12:44 +0100)]
smbd: Always use O_NONBLOCK in openat_pathref_fsp()
There's no reason why we would ever want to block on open(O_PATH). The
only cases that to me right now seem relevant is oplock breaks and
FIFOs, which can block forever. Oplock breaks don't happen for
O_PATH (hopefully...) but for the non-O_PATH case we don't want to
block either but we do handle this higher up.
We're handling EWOULDBLOCK for the oplock case correctly in
open_file_ntcreate() by setting up polling. So far we haven't done
this for the implicit openat_pathref_fsp() from filename_convert()
yet. But as our kernel oplock implementation lacks in functionality
big time anyway I would rather fail an open with NETWORK_BUSY than to
sit waiting for an oplock break for 30 seconds.
Signed-off-by: Volker Lendecke <vl@samba.org> Reviewed-by: Ralph Boehme <slow@samba.org>
Volker Lendecke [Thu, 10 Mar 2022 18:18:44 +0000 (19:18 +0100)]
smbd: Simplify dos_mode_check_compressed()
btrfs_fget_compression() is the only real implementation of
VFS_GET_COMPRESSION. It does not use the mem_ctx argument, so it seems
unnecessary to do a full malloc()/free() cycle here. Moreover, if this
was actually required, talloc_stackframe() would be more appropriate
these days as deep within the smbd even loop it does not go through
the libc malloc, but just increments a pointer.
Signed-off-by: Volker Lendecke <vl@samba.org> Reviewed-by: Ralph Boehme <slow@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Joseph Sutton <josephsutton@catalyst.net.nz>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Fri Mar 11 18:06:47 UTC 2022 on sn-devel-184
Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Fri Mar 11 11:05:55 UTC 2022 on sn-devel-184
Volker Lendecke [Fri, 4 Mar 2022 13:56:24 +0000 (14:56 +0100)]
smbd: Simplify non_widelink_open()
Don't depend on fsp->fsp_flags.is_directory: We can always take the
parent directory fname, chdir into it and openat(O_PATH|O_NOFOLLOW)
the relative file name. To properly handle the symlink case without
having O_PATH, upon failure we need the call to
fstatat(AT_SYMLINK_NOFOLLOW) as a replacement for the fstat-call that
we can do when we successfully opened the relative file name with
O_NOFOLLOW.
Signed-off-by: Volker Lendecke <vl@samba.org> Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu Mar 10 19:19:06 UTC 2022 on sn-devel-184
Volker Lendecke [Mon, 7 Mar 2022 17:00:20 +0000 (18:00 +0100)]
vfs: Convert get_real_filename() to NTSTATUS
This makes it possible to more easily handle STOPPED_ON_SYMLINK vs
OBJECT_PATH_NOT_FOUND vs OBJECT_NAME_NOT_FOUND and so on. The next
patch needs this to properly handle symlinks.
Signed-off-by: Volker Lendecke <vl@samba.org> Reviewed-by: Jeremy Allison <jra@samba.org>
Jeremy Allison [Tue, 8 Mar 2022 01:47:15 +0000 (17:47 -0800)]
s3: smbd: Plumb in and use smbd_smb2_server_connection_read_handler() when server min protocol > NT1 (i.e. SMB2-only).
This will allow us to remove the SMB1 server specific code
when we disable SMB1, and still retain the ability to negotiate
up from SMB1 -> SMB2 for old clients.
Signed-off-by: Jeremy Allison <jra@samba.org> Reviewed-by: David Mulder <dmulder@samba.org>
Autobuild-User(master): David Mulder <dmulder@samba.org>
Autobuild-Date(master): Thu Mar 10 17:53:26 UTC 2022 on sn-devel-184
Restricts negotiation to SMB2-only. This will make it easier
to remove the SMB1-only parts of the server later.
The only allowed pre-SMB2 requests are a NBSSrequest
(to set the client NetBIOS name) and a 'normal' NBSSmessage
containing an SMB1 negprot. This allows smbd_smb2_server_connection_read_handler()
to work with older clients that use an initial SMB1negprot to
bootstrap into SMB2.
Eventually all other parts of the SMB1 server will
be removed.
Not yet used.
Signed-off-by: Jeremy Allison <jra@samba.org> Reviewed-by: David Mulder <dmulder@samba.org>
s4:auth: let authenticate_ldap_simple_bind() pass down the mapped nt4names
authenticate_ldap_simple_bind*() needs to pass the
result of the cracknames operation into the auth stack
as user_info->client.{account,domain}_name, because
user_info->client.{account,domain}_name is also used
when forwarding the request via netrLogonSamLogon*
to a remote server, for exactly that the values are
also used in order to map a AUTH_PASSWORD_PLAIN into
AUTH_PASSWORD_RESPONSE, where the NTLMv2 response
contains the account and domain names passed in the
netr_IdentityInfo value.
Otherwise it would not be possible to forward the
LDAP simple bind authentication request to a remote
DC.
Currently this only applies to an RODC that forwards
the request to an RWDC.
But note that LDAP simple binds (as on Windows) only
work for users in the DCs forest, as the DsCrackNames
need to work and it can't work for users of remote
forests. I tested that in a DC of a forest root domain,
if rejected the LDAP simple bind against a different forest,
but allowed it for a users of a child domain in the
same forest. The NTLMSSP bind worked in both cases.
Signed-off-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Thu Mar 10 04:10:54 UTC 2022 on sn-devel-184
auth: let auth logging prefer user_info->orig_client.{account,domain}_name if available
The optional user_info->orig_client.{account,domain}_name are
the once really used by the client and should be used in
audit logging. But we still fallback to
user_info->client.{account,domain}_name.
s4:auth: rename user_info->mapped_state to user_info->cracknames_called
This makes it much clearer what it is used for and
it is a special hack for authenticate_ldap_simple_bind_send()
in order to avoid some additional work in
authsam_check_password_internals().
winbindd: don't set mapped_state in winbindd_dual_auth_passdb()
mapped_state is a special hack for authenticate_ldap_simple_bind_send()
in order to avoid some additional work in authsam_check_password_internals()
This doesn't apply here. We should also handle wbinfo -a
authentication UPN names, e.g. administrator@DOMAIN,
even if the account belongs to the local sam.
With this change the behavior is consistent also locally on DCs and
also an RODC can handle these requests locally for cached accounts.
s3:auth: make_user_info_map() should not set mapped_state
mapped_state is only evaluated in authsam_check_password_internals()
of auth_sam.c in source4, so setting it in the auth3 code
doesn't make any difference. I've proved that with
an SMB_ASSERT() and a full pipeline not triggering it.
s4:auth: check for user_info->mapped.account_name if it needs to be filled
mapped_state is a special hack for authenticate_ldap_simple_bind_send()
in order to avoid some additional work in authsam_check_password_internals().
But that code will be changed in the next commits, so we can simplify
the logic and only check for user_info->mapped.account_name being NULL.
As it's the important factor that user_info->mapped.account_name is
non-NULL down in the auth stack.
s4:auth: encrypt_user_info() should set password_state instead of mapped_state
user_info->mapped_state has nothing to do with enum auth_password_state,
user_info->password_state is the one that holds the auth_password_state value.
Luckily user_info->password_state was never referenced in the
encrypt_user_info() callers.
s4:auth: a simple bind uses the DCs name as workstation
I've seen that in LogonSamLogonEx request triggered
by a simple bind with a user of a trusted domain
within the same forest. Note simple binds don't
work with users for another forest/external domain,
as the DsCrackNames call on the bind_dn fails.
Signed-off-by: Garming Sam <garming@catalyst.net.nz> Reviewed-by: Stefan Metzmacher <metze@samba.org> Reviewed-by: Andrew Bartlett <abartlet@samba.org>
s4:auth_sam: use USER_INFO_INTERACTIVE_LOGON as inducation for an interactive logon
Using != AUTH_PASSWORD_RESPONSE is not the correct indication
due to the local mappings from AUTH_PASSWORD_PLAIN via
AUTH_PASSWORD_HASH to AUTH_PASSWORD_RESPONSE.
It means an LDAP simble bind will now honour
'old password allowed period'.
Signed-off-by: Jeremy Allison <jra@samba.org> Signed-off-by: David Mulder <dmulder@samba.org>
Autobuild-User(master): David Mulder <dmulder@samba.org>
Autobuild-Date(master): Tue Mar 8 23:05:19 UTC 2022 on sn-devel-184
Jeremy Allison [Tue, 8 Mar 2022 03:04:23 +0000 (19:04 -0800)]
s3: smbd: Split srv_init_signing() into 2 static functions smb1_srv_init_signing() and smb2_srv_init_signing().
Correctly initialize and look at xconn->smb2.signing_mandatory
for the SMB2 signing state (this gets set correctly for the AD-DC
case etc. inside smb2_srv_init_signing()).
Signed-off-by: Jeremy Allison <jra@samba.org> Signed-off-by: David Mulder <dmulder@samba.org>
Jule Anger [Fri, 4 Mar 2022 08:02:28 +0000 (09:02 +0100)]
s3:utils: assign ids to struct to list shares correctly
The commit "99d1f1fa10d smbd: Remove unused "struct connections_key"" removes
also the assignment of information to connections_data, which are needed to list
shares.
Pavel Filipenský [Thu, 17 Feb 2022 18:20:46 +0000 (19:20 +0100)]
s3:rpcclient: Fix crash in rpcclient
rpcclient SERVER -c 'dfsenum 5' dumps core
Signed-off-by: Pavel Filipenský <pfilipen@redhat.com> Reviewed-by: Andrew Bartlett <abartlet@samba.org> Reviewed-by: Joseph Sutton <josephsutton@catalyst.net.nz>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Mon Mar 7 00:00:32 UTC 2022 on sn-devel-184
Pavel Filipenský [Thu, 17 Feb 2022 18:20:46 +0000 (19:20 +0100)]
s3:rpcclient: Fix trailing whitespace in cmd_dfs.c
Signed-off-by: Pavel Filipenský <pfilipen@redhat.com> Reviewed-by: Andrew Bartlett <abartlet@samba.org> Reviewed-by: Joseph Sutton <josephsutton@catalyst.net.nz>
Pavel Filipenský [Wed, 23 Feb 2022 16:39:46 +0000 (17:39 +0100)]
s3:script: Blackbox tests for the rpcclient DFS commands
Signed-off-by: Pavel Filipenský <pfilipen@redhat.com> Reviewed-by: Andrew Bartlett <abartlet@samba.org> Reviewed-by: Joseph Sutton <josephsutton@catalyst.net.nz>
s4:kdc: redirect pre-authentication failures to an RWDC
The most important case is that we still have a previous
password cached at the RODC and the inbound replication
hasn't wiped the cache yet and we also haven't triggered
a new replication yet.
projects / thirdparty / samba.git / log
