/*
* Unix SMB / CIFS implementation .
* RPC Pipe client / server routines
* Copyright ( C ) Andrew Tridgell 1992 - 1998 ,
* Copyright ( C ) Luke Kenneth Casson Leighton 1996 - 1998 ,
* Copyright ( C ) Paul Ashton 1998.
* Copyright ( C ) Jeremy Allison 1999.
* Copyright ( C ) Andrew Bartlett 2003.
*
* This program is free software ; you can redistribute it and / or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation ; either version 2 of the License , or
* ( at your option ) any later version .
*
* This program is distributed in the hope that it will be useful ,
* but WITHOUT ANY WARRANTY ; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE . See the
* GNU General Public License for more details .
*
* You should have received a copy of the GNU General Public License
* along with this program ; if not , write to the Free Software
* Foundation , Inc . , 675 Mass Ave , Cambridge , MA 0213 9 , USA .
*/
# include "includes.h"
# undef DBGC_CLASS
# define DBGC_CLASS DBGC_RPC_CLI
extern struct pipe_id_info pipe_names [ ] ;
static void get_auth_type_level ( int pipe_auth_flags , int * auth_type , int * auth_level )
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
{
* auth_type = 0 ;
* auth_level = 0 ;
if ( pipe_auth_flags & AUTH_PIPE_SEAL ) {
* auth_level = RPC_PIPE_AUTH_SEAL_LEVEL ;
} else if ( pipe_auth_flags & AUTH_PIPE_SIGN ) {
* auth_level = RPC_PIPE_AUTH_SIGN_LEVEL ;
}
if ( pipe_auth_flags & AUTH_PIPE_NETSEC ) {
* auth_type = NETSEC_AUTH_TYPE ;
} else if ( pipe_auth_flags & AUTH_PIPE_NTLMSSP ) {
* auth_type = NTLMSSP_AUTH_TYPE ;
}
}
/********************************************************************
Rpc pipe call id .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
static uint32 get_rpc_call_id ( void )
{
static uint32 call_id = 0 ;
return + + call_id ;
}
/*******************************************************************
Use SMBreadX to get rest of one fragment ' s worth of rpc data .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
static BOOL rpc_read ( struct cli_state * cli , prs_struct * rdata , uint32 data_to_read , uint32 * rdata_offset )
{
size_t size = ( size_t ) cli - > max_recv_frag ;
int stream_offset = 0 ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
int num_read ;
char * pdata ;
int extra_data_size = ( ( int ) * rdata_offset ) + ( ( int ) data_to_read ) - ( int ) prs_data_size ( rdata ) ;
DEBUG ( 5 , ( " rpc_read: data_to_read: %u rdata offset: %u extra_data_size: %d \n " ,
( int ) data_to_read , ( unsigned int ) * rdata_offset , extra_data_size ) ) ;
/*
* Grow the buffer if needed to accommodate the data to be read .
*/
if ( extra_data_size > 0 ) {
if ( ! prs_force_grow ( rdata , ( uint32 ) extra_data_size ) ) {
DEBUG ( 0 , ( " rpc_read: Failed to grow parse struct by %d bytes. \n " , extra_data_size ) ) ;
return False ;
}
DEBUG ( 5 , ( " rpc_read: grew buffer by %d bytes to %u \n " , extra_data_size , prs_data_size ( rdata ) ) ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
}
pdata = prs_data_p ( rdata ) + * rdata_offset ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
do /* read data using SMBreadX */
{
uint32 ecode ;
uint8 eclass ;
if ( size > ( size_t ) data_to_read )
size = ( size_t ) data_to_read ;
num_read = ( int ) cli_read ( cli , cli - > nt_pipe_fnum , pdata , ( off_t ) stream_offset , size ) ;
DEBUG ( 5 , ( " rpc_read: num_read = %d, read offset: %d, to read: %d \n " ,
num_read , stream_offset , data_to_read ) ) ;
if ( cli_is_dos_error ( cli ) ) {
cli_dos_error ( cli , & eclass , & ecode ) ;
if ( eclass ! = ERRDOS & & ecode ! = ERRmoredata ) {
DEBUG ( 0 , ( " rpc_read: Error %d/%u in cli_read \n " ,
eclass , ( unsigned int ) ecode ) ) ;
return False ;
}
}
data_to_read - = num_read ;
stream_offset + = num_read ;
pdata + = num_read ;
} while ( num_read > 0 & & data_to_read > 0 ) ;
/* && err == (0x80000000 | STATUS_BUFFER_OVERFLOW)); */
/*
* Update the current offset into rdata by the amount read .
*/
* rdata_offset + = stream_offset ;
return True ;
}
/****************************************************************************
Checks the header . This will set the endian bit in the rdata prs_struct . JRA .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
static BOOL rpc_check_hdr ( prs_struct * rdata , RPC_HDR * rhdr ,
BOOL * first , BOOL * last , uint32 * len )
{
DEBUG ( 5 , ( " rpc_check_hdr: rdata->data_size = %u \n " , ( uint32 ) prs_data_size ( rdata ) ) ) ;
/* Next call sets endian bit. */
if ( ! smb_io_rpc_hdr ( " rpc_hdr " , rhdr , rdata , 0 ) ) {
DEBUG ( 0 , ( " rpc_check_hdr: Failed to unmarshall RPC_HDR. \n " ) ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
return False ;
}
if ( prs_offset ( rdata ) ! = RPC_HEADER_LEN ) {
DEBUG ( 0 , ( " rpc_check_hdr: offset was %x, should be %x. \n " , prs_offset ( rdata ) , RPC_HEADER_LEN ) ) ;
return False ;
}
( * first ) = ( ( rhdr - > flags & RPC_FLG_FIRST ) ! = 0 ) ;
( * last ) = ( ( rhdr - > flags & RPC_FLG_LAST ) ! = 0 ) ;
( * len ) = ( uint32 ) rhdr - > frag_len - prs_data_size ( rdata ) ;
return ( rhdr - > pkt_type ! = RPC_FAULT ) ;
}
/****************************************************************************
Verify data on an rpc pipe .
The VERIFY & SEAL code is only executed on packets that look like this :
Request / Response PDU ' s look like the following . . .
| < - - - - - - - - - - - - - - - - - - PDU len - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - > |
| < - HDR_LEN - - > | < - - REQ LEN - - - - - - > | . . . . . . . . . . . . . | < - AUTH_HDRLEN - > | < - AUTH_LEN - - > |
+ - - - - - - - - - - - - + - - - - - - - - - - - - - - - - - + - - - - - - - - - - - - - + - - - - - - - - - - - - - - - + - - - - - - - - - - - - - +
| RPC HEADER | REQ / RESP HEADER | DATA . . . . . . | AUTH_HDR | AUTH DATA |
+ - - - - - - - - - - - - + - - - - - - - - - - - - - - - - - + - - - - - - - - - - - - - + - - - - - - - - - - - - - - - + - - - - - - - - - - - - - +
Never on bind requests / responses .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
static BOOL rpc_auth_pipe ( struct cli_state * cli , prs_struct * rdata ,
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
uint32 fragment_start , int len , int auth_len , uint8 pkt_type ,
int * pauth_padding_len )
{
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
/*
* The following is that length of the data we must sign or seal .
* This doesn ' t include the RPC headers or the auth_len or the RPC_HDR_AUTH_LEN
* preceeding the auth_data .
*/
int data_len = len - RPC_HEADER_LEN - RPC_HDR_RESP_LEN - RPC_HDR_AUTH_LEN - auth_len ;
/*
* The start of the data to sign / seal is just after the RPC headers .
*/
char * reply_data = prs_data_p ( rdata ) + fragment_start + RPC_HEADER_LEN + RPC_HDR_REQ_LEN ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
RPC_HDR_AUTH rhdr_auth ;
char * dp = prs_data_p ( rdata ) + fragment_start + len -
RPC_HDR_AUTH_LEN - auth_len ;
prs_struct auth_verf ;
* pauth_padding_len = 0 ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( auth_len = = 0 ) {
if ( cli - > pipe_auth_flags = = 0 ) {
/* move along, nothing to see here */
return True ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
DEBUG ( 2 , ( " No authenticaton header recienved on reply, but this pipe is authenticated \n " ) ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
DEBUG ( 5 , ( " rpc_auth_pipe: pkt_type: %d len: %d auth_len: %d NTLMSSP %s schannel %s sign %s seal %s \n " ,
pkt_type , len , auth_len ,
BOOLSTR ( cli - > pipe_auth_flags & AUTH_PIPE_NTLMSSP ) ,
BOOLSTR ( cli - > pipe_auth_flags & AUTH_PIPE_NETSEC ) ,
BOOLSTR ( cli - > pipe_auth_flags & AUTH_PIPE_SIGN ) ,
BOOLSTR ( cli - > pipe_auth_flags & AUTH_PIPE_SEAL ) ) ) ;
if ( dp - prs_data_p ( rdata ) > prs_data_size ( rdata ) ) {
DEBUG ( 0 , ( " rpc_auth_pipe: schannel auth data > data size ! \n " ) ) ;
return False ;
}
DEBUG ( 10 , ( " rpc_auth_pipe: packet: \n " ) ) ;
dump_data ( 100 , dp , auth_len ) ;
prs_init ( & auth_verf , 0 , cli - > mem_ctx , UNMARSHALL ) ;
/* The endinness must be preserved. JRA. */
prs_set_endian_data ( & auth_verf , rdata - > bigendian_data ) ;
/* Point this new parse struct at the auth section of the main
parse struct - rather than copying it . Avoids needing to
free it on every error
*/
prs_give_memory ( & auth_verf , dp , RPC_HDR_AUTH_LEN + auth_len , False /* not dynamic */ ) ;
prs_set_offset ( & auth_verf , 0 ) ;
{
int auth_type ;
int auth_level ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! smb_io_rpc_hdr_auth ( " auth_hdr " , & rhdr_auth , & auth_verf , 0 ) ) {
DEBUG ( 0 , ( " rpc_auth_pipe: Could not parse auth header \n " ) ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
/* Let the caller know how much padding at the end of the data */
* pauth_padding_len = rhdr_auth . padding ;
/* Check it's the type of reply we were expecting to decode */
get_auth_type_level ( cli - > pipe_auth_flags , & auth_type , & auth_level ) ;
if ( rhdr_auth . auth_type ! = auth_type ) {
DEBUG ( 0 , ( " BAD auth type %d (should be %d) \n " ,
rhdr_auth . auth_type , auth_type ) ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( rhdr_auth . auth_level ! = auth_level ) {
DEBUG ( 0 , ( " BAD auth level %d (should be %d) \n " ,
rhdr_auth . auth_level , auth_level ) ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( pkt_type = = RPC_BINDACK ) {
if ( cli - > pipe_auth_flags & AUTH_PIPE_NTLMSSP ) {
/* copy the next auth_len bytes into a buffer for
later use */
DATA_BLOB ntlmssp_verf = data_blob ( NULL , auth_len ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
/* save the reply away, for use a little later */
prs_copy_data_out ( ntlmssp_verf . data , & auth_verf , auth_len ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
return ( NT_STATUS_IS_OK ( ntlmssp_client_store_response ( cli - > ntlmssp_pipe_state ,
ntlmssp_verf ) ) ) ;
}
else if ( cli - > pipe_auth_flags & AUTH_PIPE_NETSEC ) {
/* nothing to do here - we don't seem to be able to
validate the bindack based on VL ' s comments */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
return True ;
}
}
if ( cli - > pipe_auth_flags & AUTH_PIPE_NTLMSSP ) {
NTSTATUS nt_status ;
DATA_BLOB sig ;
if ( ( cli - > pipe_auth_flags & AUTH_PIPE_SIGN ) | |
( cli - > pipe_auth_flags & AUTH_PIPE_SEAL ) ) {
if ( auth_len ! = RPC_AUTH_NTLMSSP_CHK_LEN ) {
DEBUG ( 0 , ( " rpc_auth_pipe: wrong ntlmssp auth len %d \n " , auth_len ) ) ;
return False ;
}
sig = data_blob ( NULL , auth_len ) ;
prs_copy_data_out ( sig . data , & auth_verf , auth_len ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
}
/*
* Unseal any sealed data in the PDU , not including the
* 8 byte auth_header or the auth_data .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
/*
* Now unseal and check the auth verifier in the auth_data at
* the end of the packet .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( cli - > pipe_auth_flags & AUTH_PIPE_SEAL ) {
if ( data_len < 0 ) {
DEBUG ( 1 , ( " Can't unseal - data_len < 0!! \n " ) ) ;
return False ;
}
nt_status = ntlmssp_client_unseal_packet ( cli - > ntlmssp_pipe_state ,
reply_data , data_len ,
& sig ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
}
else if ( cli - > pipe_auth_flags & AUTH_PIPE_SIGN ) {
nt_status = ntlmssp_client_check_packet ( cli - > ntlmssp_pipe_state ,
reply_data , data_len ,
& sig ) ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
data_blob_free ( & sig ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
DEBUG ( 0 , ( " rpc_auth_pipe: could not validate "
" incoming NTLMSSP packet! \n " ) ) ;
return False ;
}
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( cli - > pipe_auth_flags & AUTH_PIPE_NETSEC ) {
RPC_AUTH_NETSEC_CHK chk ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( auth_len ! = RPC_AUTH_NETSEC_CHK_LEN ) {
DEBUG ( 0 , ( " rpc_auth_pipe: wrong schannel auth len %d \n " , auth_len ) ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! smb_io_rpc_auth_netsec_chk ( " schannel_auth_sign " ,
& chk , & auth_verf , 0 ) ) {
DEBUG ( 0 , ( " rpc_auth_pipe: schannel unmarshalling "
" RPC_AUTH_NETSECK_CHK failed \n " ) ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! netsec_decode ( & cli - > auth_info ,
cli - > pipe_auth_flags ,
SENDER_IS_ACCEPTOR ,
& chk , reply_data , data_len ) ) {
DEBUG ( 0 , ( " rpc_auth_pipe: Could not decode schannel \n " ) ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
cli - > auth_info . seq_num + + ;
}
return True ;
}
/****************************************************************************
Send data on an rpc pipe via trans , which * must * be the last fragment .
receive response data from an rpc pipe , which may be large . . .
Read the first fragment : unfortunately have to use SMBtrans for the first
bit , then SMBreadX for subsequent bits .
If first fragment received also wasn ' t the last fragment , continue
getting fragments until we _do_ receive the last fragment .
Request / Response PDU ' s look like the following . . .
| < - - - - - - - - - - - - - - - - - - PDU len - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - > |
| < - HDR_LEN - - > | < - - REQ LEN - - - - - - > | . . . . . . . . . . . . . | < - AUTH_HDRLEN - > | < - AUTH_LEN - - > |
+ - - - - - - - - - - - - + - - - - - - - - - - - - - - - - - + - - - - - - - - - - - - - + - - - - - - - - - - - - - - - + - - - - - - - - - - - - - +
| RPC HEADER | REQ / RESP HEADER | DATA . . . . . . | AUTH_HDR | AUTH DATA |
+ - - - - - - - - - - - - + - - - - - - - - - - - - - - - - - + - - - - - - - - - - - - - + - - - - - - - - - - - - - - - + - - - - - - - - - - - - - +
Where the presence of the AUTH_HDR and AUTH are dependent on the
signing & sealing being negotiated .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
static BOOL rpc_api_pipe ( struct cli_state * cli , prs_struct * data , prs_struct * rdata ,
uint8 expected_pkt_type )
{
uint32 len ;
char * rparam = NULL ;
uint32 rparam_len = 0 ;
uint16 setup [ 2 ] ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
BOOL first = True ;
BOOL last = True ;
RPC_HDR rhdr ;
char * pdata = data ? prs_data_p ( data ) : NULL ;
uint32 data_len = data ? prs_offset ( data ) : 0 ;
char * prdata = NULL ;
uint32 rdata_len = 0 ;
uint32 current_offset = 0 ;
uint32 fragment_start = 0 ;
uint32 max_data = cli - > max_xmit_frag ? cli - > max_xmit_frag : 1024 ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
int auth_padding_len = 0 ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* Create setup parameters - must be in native byte order. */
setup [ 0 ] = TRANSACT_DCERPCCMD ;
setup [ 1 ] = cli - > nt_pipe_fnum ; /* Pipe file handle. */
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
DEBUG ( 5 , ( " rpc_api_pipe: fnum:%x \n " , ( int ) cli - > nt_pipe_fnum ) ) ;
/* Send the RPC request and receive a response. For short RPC
calls ( about 1024 bytes or so ) the RPC request and response
appears in a SMBtrans request and response . Larger RPC
responses are received further on . */
if ( ! cli_api_pipe ( cli , " \\ PIPE \\ " ,
setup , 2 , 0 , /* Setup, length, max */
NULL , 0 , 0 , /* Params, length, max */
pdata , data_len , max_data , /* data, length, max */
& rparam , & rparam_len , /* return params, len */
& prdata , & rdata_len ) ) /* return data, len */
{
DEBUG ( 0 , ( " cli_pipe: return critical error. Error was %s \n " , cli_errstr ( cli ) ) ) ;
return False ;
}
/* Throw away returned params - we know we won't use them. */
SAFE_FREE ( rparam ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
if ( prdata = = NULL ) {
DEBUG ( 0 , ( " rpc_api_pipe: pipe %x failed to return data. \n " ,
( int ) cli - > nt_pipe_fnum ) ) ;
return False ;
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/*
* Give this memory as dynamically allocated to the return parse
* struct .
*/
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
prs_give_memory ( rdata , prdata , rdata_len , True ) ;
current_offset = rdata_len ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* This next call sets the endian bit correctly in rdata. */
if ( ! rpc_check_hdr ( rdata , & rhdr , & first , & last , & len ) ) {
prs_mem_free ( rdata ) ;
return False ;
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
if ( rhdr . pkt_type = = RPC_BINDACK ) {
if ( ! last & & ! first ) {
DEBUG ( 5 , ( " rpc_api_pipe: bug in server (AS/U?), setting fragment first/last ON. \n " ) ) ;
first = True ;
last = True ;
}
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( rhdr . pkt_type = = RPC_BINDNACK ) {
DEBUG ( 3 , ( " Bind NACK received on pipe %x! \n " , ( int ) cli - > nt_pipe_fnum ) ) ;
prs_mem_free ( rdata ) ;
return False ;
}
if ( rhdr . pkt_type = = RPC_RESPONSE ) {
RPC_HDR_RESP rhdr_resp ;
if ( ! smb_io_rpc_hdr_resp ( " rpc_hdr_resp " , & rhdr_resp , rdata , 0 ) ) {
DEBUG ( 5 , ( " rpc_api_pipe: failed to unmarshal RPC_HDR_RESP. \n " ) ) ;
prs_mem_free ( rdata ) ;
return False ;
}
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( rhdr . pkt_type ! = expected_pkt_type ) {
DEBUG ( 3 , ( " Connection to pipe %x got an unexpected RPC packet type - %d, not %d \n " , ( int ) cli - > nt_pipe_fnum , rhdr . pkt_type , expected_pkt_type ) ) ;
prs_mem_free ( rdata ) ;
return False ;
}
DEBUG ( 5 , ( " rpc_api_pipe: len left: %u smbtrans read: %u \n " ,
( unsigned int ) len , ( unsigned int ) rdata_len ) ) ;
/* check if data to be sent back was too large for one SMBtrans */
/* err status is only informational: the _real_ check is on the
length */
if ( len > 0 ) {
/* || err == (0x80000000 | STATUS_BUFFER_OVERFLOW)) */
/* Read the remaining part of the first response fragment */
if ( ! rpc_read ( cli , rdata , len , & current_offset ) ) {
prs_mem_free ( rdata ) ;
return False ;
}
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/*
* Now we have a complete PDU , check the auth struct if any was sent .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! rpc_auth_pipe ( cli , rdata , fragment_start , rhdr . frag_len ,
rhdr . auth_len , rhdr . pkt_type , & auth_padding_len ) ) {
prs_mem_free ( rdata ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( rhdr . auth_len ! = 0 ) {
/*
* Drop the auth footers from the current offset .
* We need this if there are more fragments .
* The auth footers consist of the auth_data and the
* preceeding 8 byte auth_header .
*/
current_offset - = ( auth_padding_len + RPC_HDR_AUTH_LEN + rhdr . auth_len ) ;
}
/*
* Only one rpc fragment , and it has been read .
*/
if ( first & & last ) {
DEBUG ( 6 , ( " rpc_api_pipe: fragment first and last both set \n " ) ) ;
return True ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
}
/*
* Read more fragments using SMBreadX until we get one with the
* last bit set .
*/
while ( ! last ) {
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
RPC_HDR_RESP rhdr_resp ;
int num_read ;
char hdr_data [ RPC_HEADER_LEN + RPC_HDR_RESP_LEN ] ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
prs_struct hps ;
uint8 eclass ;
uint32 ecode ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
/*
* First read the header of the next PDU .
*/
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
prs_init ( & hps , 0 , cli - > mem_ctx , UNMARSHALL ) ;
prs_give_memory ( & hps , hdr_data , sizeof ( hdr_data ) , False ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
num_read = cli_read ( cli , cli - > nt_pipe_fnum , hdr_data , 0 , RPC_HEADER_LEN + RPC_HDR_RESP_LEN ) ;
if ( cli_is_dos_error ( cli ) ) {
cli_dos_error ( cli , & eclass , & ecode ) ;
if ( eclass ! = ERRDOS & & ecode ! = ERRmoredata ) {
DEBUG ( 0 , ( " rpc_api_pipe: cli_read error : %d/%d \n " , eclass , ecode ) ) ;
return False ;
}
}
DEBUG ( 5 , ( " rpc_api_pipe: read header (size:%d) \n " , num_read ) ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
if ( num_read ! = RPC_HEADER_LEN + RPC_HDR_RESP_LEN ) {
DEBUG ( 0 , ( " rpc_api_pipe: Error : requested %d bytes, got %d. \n " ,
RPC_HEADER_LEN + RPC_HDR_RESP_LEN , num_read ) ) ;
return False ;
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* This call sets the endianness in hps. */
if ( ! rpc_check_hdr ( & hps , & rhdr , & first , & last , & len ) )
return False ;
/* Ensure the endianness in rdata is set correctly - must be same as hps. */
if ( hps . bigendian_data ! = rdata - > bigendian_data ) {
DEBUG ( 0 , ( " rpc_api_pipe: Error : Endianness changed from %s to %s \n " ,
rdata - > bigendian_data ? " big " : " little " ,
hps . bigendian_data ? " big " : " little " ) ) ;
return False ;
}
if ( ! smb_io_rpc_hdr_resp ( " rpc_hdr_resp " , & rhdr_resp , & hps , 0 ) ) {
DEBUG ( 0 , ( " rpc_api_pipe: Error in unmarshalling RPC_HDR_RESP. \n " ) ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
return False ;
}
if ( first ) {
DEBUG ( 0 , ( " rpc_api_pipe: secondary PDU rpc header has 'first' set ! \n " ) ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
return False ;
}
/*
* Now read the rest of the PDU .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! rpc_read ( cli , rdata , len , & current_offset ) ) {
prs_mem_free ( rdata ) ;
return False ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
}
fragment_start = current_offset - len - RPC_HEADER_LEN - RPC_HDR_RESP_LEN ;
/*
* Verify any authentication footer .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! rpc_auth_pipe ( cli , rdata , fragment_start , rhdr . frag_len ,
rhdr . auth_len , rhdr . pkt_type , & auth_padding_len ) ) {
prs_mem_free ( rdata ) ;
return False ;
}
if ( rhdr . auth_len ! = 0 ) {
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
/*
* Drop the auth footers from the current offset .
* The auth footers consist of the auth_data and the
* preceeding 8 byte auth_header .
* We need this if there are more fragments .
*/
current_offset - = ( auth_padding_len + RPC_HDR_AUTH_LEN + rhdr . auth_len ) ;
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
}
return True ;
}
/*******************************************************************
creates a DCE / RPC bind request
- initialises the parse structure .
- dynamically allocates the header data structure
- caller is expected to free the header data structure once used .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
static NTSTATUS create_rpc_bind_req ( struct cli_state * cli , prs_struct * rpc_out ,
uint32 rpc_call_id ,
RPC_IFACE * abstract , RPC_IFACE * transfer ,
const char * my_name , const char * domain )
{
RPC_HDR hdr ;
RPC_HDR_RB hdr_rb ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
RPC_HDR_AUTH hdr_auth ;
int auth_len = 0 ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
int auth_type , auth_level ;
size_t saved_hdr_offset ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_struct auth_info ;
prs_init ( & auth_info , RPC_HDR_AUTH_LEN , /* we will need at least this much */
prs_get_mem_context ( rpc_out ) , MARSHALL ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( cli - > pipe_auth_flags ) {
get_auth_type_level ( cli - > pipe_auth_flags , & auth_type , & auth_level ) ;
/*
* Create the auth structs we will marshall .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
init_rpc_hdr_auth ( & hdr_auth , auth_type , auth_level , 0x00 , 1 ) ;
/*
* Now marshall the data into the temporary parse_struct .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! smb_io_rpc_hdr_auth ( " hdr_auth " , & hdr_auth , & auth_info , 0 ) ) {
DEBUG ( 0 , ( " create_rpc_bind_req: failed to marshall RPC_HDR_AUTH. \n " ) ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_mem_free ( & auth_info ) ;
return NT_STATUS_NO_MEMORY ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
saved_hdr_offset = prs_offset ( & auth_info ) ;
}
if ( cli - > pipe_auth_flags & AUTH_PIPE_NTLMSSP ) {
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
NTSTATUS nt_status ;
DATA_BLOB null_blob = data_blob ( NULL , 0 ) ;
DATA_BLOB request ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
DEBUG ( 5 , ( " Processing NTLMSSP Negotiate \n " ) ) ;
nt_status = ntlmssp_client_update ( cli - > ntlmssp_pipe_state ,
null_blob ,
& request ) ;
if ( ! NT_STATUS_EQUAL ( nt_status ,
NT_STATUS_MORE_PROCESSING_REQUIRED ) ) {
prs_mem_free ( & auth_info ) ;
return nt_status ;
}
/* Auth len in the rpc header doesn't include auth_header. */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
auth_len = request . length ;
prs_copy_data_in ( & auth_info , request . data , request . length ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
DEBUG ( 5 , ( " NTLMSSP Negotiate: \n " ) ) ;
dump_data ( 5 , request . data , request . length ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
data_blob_free ( & request ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
}
else if ( cli - > pipe_auth_flags & AUTH_PIPE_NETSEC ) {
RPC_AUTH_NETSEC_NEG netsec_neg ;
/* Use lp_workgroup() if domain not specified */
if ( ! domain | | ! domain [ 0 ] )
domain = lp_workgroup ( ) ;
init_rpc_auth_netsec_neg ( & netsec_neg , domain , my_name ) ;
/*
* Now marshall the data into the temporary parse_struct .
*/
if ( ! smb_io_rpc_auth_netsec_neg ( " netsec_neg " ,
& netsec_neg , & auth_info , 0 ) ) {
DEBUG ( 0 , ( " Failed to marshall RPC_AUTH_NETSEC_NEG. \n " ) ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_mem_free ( & auth_info ) ;
return NT_STATUS_NO_MEMORY ;
}
/* Auth len in the rpc header doesn't include auth_header. */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
auth_len = prs_offset ( & auth_info ) - saved_hdr_offset ;
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* create the request RPC_HDR */
init_rpc_hdr ( & hdr , RPC_BIND , 0x3 , rpc_call_id ,
RPC_HEADER_LEN + RPC_HDR_RB_LEN + prs_offset ( & auth_info ) ,
auth_len ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
if ( ! smb_io_rpc_hdr ( " hdr " , & hdr , rpc_out , 0 ) ) {
DEBUG ( 0 , ( " create_rpc_bind_req: failed to marshall RPC_HDR. \n " ) ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_mem_free ( & auth_info ) ;
return NT_STATUS_NO_MEMORY ;
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* create the bind request RPC_HDR_RB */
init_rpc_hdr_rb ( & hdr_rb , MAX_PDU_FRAG_LEN , MAX_PDU_FRAG_LEN , 0x0 ,
0x1 , 0x0 , 0x1 , abstract , transfer ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* Marshall the bind request data */
if ( ! smb_io_rpc_hdr_rb ( " " , & hdr_rb , rpc_out , 0 ) ) {
DEBUG ( 0 , ( " create_rpc_bind_req: failed to marshall RPC_HDR_RB. \n " ) ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_mem_free ( & auth_info ) ;
return NT_STATUS_NO_MEMORY ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
}
/*
* Grow the outgoing buffer to store any auth info .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( auth_len ! = 0 ) {
if ( ! prs_append_prs_data ( rpc_out , & auth_info ) ) {
DEBUG ( 0 , ( " create_rpc_bind_req: failed to grow parse struct to add auth. \n " ) ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_mem_free ( & auth_info ) ;
return NT_STATUS_NO_MEMORY ;
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
return NT_STATUS_OK ;
}
/*******************************************************************
Creates a DCE / RPC bind authentication response .
This is the packet that is sent back to the server once we
have received a BIND - ACK , to finish the third leg of
the authentication handshake .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
static NTSTATUS create_rpc_bind_resp ( struct cli_state * cli ,
uint32 rpc_call_id ,
prs_struct * rpc_out )
{
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
NTSTATUS nt_status ;
RPC_HDR hdr ;
RPC_HDR_AUTHA hdr_autha ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
DATA_BLOB ntlmssp_null_response = data_blob ( NULL , 0 ) ;
DATA_BLOB ntlmssp_reply ;
int auth_type , auth_level ;
/* The response is picked up from the internal cache,
where it was placed by the rpc_auth_pipe ( ) code */
nt_status = ntlmssp_client_update ( cli - > ntlmssp_pipe_state ,
ntlmssp_null_response ,
& ntlmssp_reply ) ;
if ( ! NT_STATUS_EQUAL ( nt_status , NT_STATUS_MORE_PROCESSING_REQUIRED ) ) {
return nt_status ;
}
/* Create the request RPC_HDR */
init_rpc_hdr ( & hdr , RPC_BINDRESP , 0x0 , rpc_call_id ,
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
RPC_HEADER_LEN + RPC_HDR_AUTHA_LEN + ntlmssp_reply . length ,
ntlmssp_reply . length ) ;
/* Marshall it. */
if ( ! smb_io_rpc_hdr ( " hdr " , & hdr , rpc_out , 0 ) ) {
DEBUG ( 0 , ( " create_rpc_bind_resp: failed to marshall RPC_HDR. \n " ) ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
data_blob_free ( & ntlmssp_reply ) ;
return NT_STATUS_NO_MEMORY ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
get_auth_type_level ( cli - > pipe_auth_flags , & auth_type , & auth_level ) ;
/* Create the request RPC_HDR_AUTHA */
init_rpc_hdr_autha ( & hdr_autha , MAX_PDU_FRAG_LEN , MAX_PDU_FRAG_LEN ,
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
auth_type , auth_level , 0x00 ) ;
if ( ! smb_io_rpc_hdr_autha ( " hdr_autha " , & hdr_autha , rpc_out , 0 ) ) {
DEBUG ( 0 , ( " create_rpc_bind_resp: failed to marshall RPC_HDR_AUTHA. \n " ) ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
data_blob_free ( & ntlmssp_reply ) ;
return NT_STATUS_NO_MEMORY ;
}
/*
* Append the auth data to the outgoing buffer .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! prs_copy_data_in ( rpc_out , ntlmssp_reply . data , ntlmssp_reply . length ) ) {
DEBUG ( 0 , ( " create_rpc_bind_req: failed to grow parse struct to add auth. \n " ) ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
data_blob_free ( & ntlmssp_reply ) ;
return NT_STATUS_NO_MEMORY ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( cli - > pipe_auth_flags & AUTH_PIPE_SIGN ) {
nt_status = ntlmssp_client_sign_init ( cli - > ntlmssp_pipe_state ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
return nt_status ;
}
}
data_blob_free ( & ntlmssp_reply ) ;
return NT_STATUS_OK ;
}
/*******************************************************************
Creates a DCE / RPC request .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
static uint32 create_rpc_request ( prs_struct * rpc_out , uint8 op_num , int data_len , int auth_len , uint8 flags , uint32 oldid , uint32 data_left )
{
uint32 alloc_hint ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
RPC_HDR hdr ;
RPC_HDR_REQ hdr_req ;
uint32 callid = oldid ? oldid : get_rpc_call_id ( ) ;
DEBUG ( 5 , ( " create_rpc_request: opnum: 0x%x data_len: 0x%x \n " , op_num , data_len ) ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* create the rpc header RPC_HDR */
init_rpc_hdr ( & hdr , RPC_REQUEST , flags ,
callid , data_len , auth_len ) ;
/*
* The alloc hint should be the amount of data , not including
* RPC headers & footers .
*/
if ( auth_len ! = 0 )
alloc_hint = data_len - RPC_HEADER_LEN - RPC_HDR_AUTH_LEN - auth_len ;
else
alloc_hint = data_len - RPC_HEADER_LEN ;
DEBUG ( 10 , ( " create_rpc_request: data_len: %x auth_len: %x alloc_hint: %x \n " ,
data_len , auth_len , alloc_hint ) ) ;
/* Create the rpc request RPC_HDR_REQ */
init_rpc_hdr_req ( & hdr_req , alloc_hint , op_num ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* stream-time... */
if ( ! smb_io_rpc_hdr ( " hdr " , & hdr , rpc_out , 0 ) )
return 0 ;
if ( ! smb_io_rpc_hdr_req ( " hdr_req " , & hdr_req , rpc_out , 0 ) )
return 0 ;
if ( prs_offset ( rpc_out ) ! = RPC_HEADER_LEN + RPC_HDR_REQ_LEN )
return 0 ;
return callid ;
}
/*******************************************************************
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
Puts an auth header into an rpc request .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
static BOOL create_auth_hdr ( prs_struct * outgoing_packet ,
int auth_type ,
int auth_level , int padding )
{
RPC_HDR_AUTH hdr_auth ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
init_rpc_hdr_auth ( & hdr_auth , auth_type , auth_level ,
padding , 1 ) ;
if ( ! smb_io_rpc_hdr_auth ( " hdr_auth " , & hdr_auth ,
outgoing_packet , 0 ) ) {
DEBUG ( 0 , ( " create_auth_hdr:Failed to marshal RPC_HDR_AUTH. \n " ) ) ;
return False ;
}
return True ;
}
/**
* Send a request on an RPC pipe and get a response .
*
* @ param data NDR contents of the request to be sent .
* @ param rdata Unparsed NDR response data .
* */
BOOL rpc_api_pipe_req ( struct cli_state * cli , uint8 op_num ,
prs_struct * data , prs_struct * rdata )
{
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
uint32 auth_len , real_auth_len , auth_hdr_len , max_data , data_left , data_sent ;
NTSTATUS nt_status ;
BOOL ret = False ;
uint32 callid = 0 ;
fstring dump_name ;
auth_len = 0 ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
real_auth_len = 0 ;
auth_hdr_len = 0 ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( cli - > pipe_auth_flags & AUTH_PIPE_SIGN ) {
if ( cli - > pipe_auth_flags & AUTH_PIPE_NTLMSSP ) {
auth_len = RPC_AUTH_NTLMSSP_CHK_LEN ;
}
if ( cli - > pipe_auth_flags & AUTH_PIPE_NETSEC ) {
auth_len = RPC_AUTH_NETSEC_CHK_LEN ;
}
auth_hdr_len = RPC_HDR_AUTH_LEN ;
}
/*
* calc how much actual data we can send in a PDU fragment
*/
max_data = cli - > max_xmit_frag - RPC_HEADER_LEN - RPC_HDR_REQ_LEN -
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
auth_hdr_len - auth_len - 8 ;
for ( data_left = prs_offset ( data ) , data_sent = 0 ; data_left > 0 ; ) {
prs_struct outgoing_packet ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_struct sec_blob ;
uint32 data_len , send_size ;
uint8 flags = 0 ;
uint32 auth_padding = 0 ;
RPC_AUTH_NETSEC_CHK verf ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
DATA_BLOB sign_blob ;
/*
* how much will we send this time
*/
send_size = MIN ( data_left , max_data ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! prs_init ( & sec_blob , send_size , /* will need at least this much */
cli - > mem_ctx , MARSHALL ) ) {
DEBUG ( 0 , ( " Could not malloc %u bytes " ,
send_size + auth_padding ) ) ;
return False ;
}
if ( ! prs_append_some_prs_data ( & sec_blob , data ,
data_sent , send_size ) ) {
DEBUG ( 0 , ( " Failed to append data to netsec blob \n " ) ) ;
prs_mem_free ( & sec_blob ) ;
return False ;
}
/*
* NT expects the data that is sealed to be 8 - byte
* aligned . The padding must be encrypted as well and
* taken into account when generating the
* authentication verifier . The amount of padding must
* be stored in the auth header .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( cli - > pipe_auth_flags ) {
size_t data_and_padding_size ;
int auth_type ;
int auth_level ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_align_uint64 ( & sec_blob ) ;
get_auth_type_level ( cli - > pipe_auth_flags , & auth_type , & auth_level ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
data_and_padding_size = prs_offset ( & sec_blob ) ;
auth_padding = data_and_padding_size - send_size ;
/* insert the auth header */
if ( ! create_auth_hdr ( & sec_blob , auth_type , auth_level , auth_padding ) ) {
prs_mem_free ( & sec_blob ) ;
return False ;
}
/* create an NTLMSSP signature */
if ( cli - > pipe_auth_flags & AUTH_PIPE_NTLMSSP ) {
/*
* Seal the outgoing data if requested .
*/
if ( cli - > pipe_auth_flags & AUTH_PIPE_SEAL ) {
nt_status = ntlmssp_client_seal_packet ( cli - > ntlmssp_pipe_state ,
( unsigned char * ) prs_data_p ( & sec_blob ) ,
data_and_padding_size ,
& sign_blob ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
prs_mem_free ( & sec_blob ) ;
return False ;
}
}
else if ( cli - > pipe_auth_flags & AUTH_PIPE_SIGN ) {
nt_status = ntlmssp_client_sign_packet ( cli - > ntlmssp_pipe_state ,
( unsigned char * ) prs_data_p ( & sec_blob ) ,
data_and_padding_size , & sign_blob ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
prs_mem_free ( & sec_blob ) ;
return False ;
}
}
/* write auth footer onto the packet */
real_auth_len = sign_blob . length ;
prs_copy_data_in ( & sec_blob , sign_blob . data , sign_blob . length ) ;
data_blob_free ( & sign_blob ) ;
}
else if ( cli - > pipe_auth_flags & AUTH_PIPE_NETSEC ) {
static const uchar netsec_sig [ 8 ] = NETSEC_SIGNATURE ;
static const uchar nullbytes [ 8 ] = { 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 } ;
size_t parse_offset_marker ;
DEBUG ( 10 , ( " SCHANNEL seq_num=%d \n " , cli - > auth_info . seq_num ) ) ;
init_rpc_auth_netsec_chk ( & verf , netsec_sig , nullbytes ,
nullbytes , nullbytes ) ;
netsec_encode ( & cli - > auth_info ,
cli - > pipe_auth_flags ,
SENDER_IS_INITIATOR ,
& verf ,
prs_data_p ( & sec_blob ) ,
data_and_padding_size ) ;
cli - > auth_info . seq_num + + ;
/* write auth footer onto the packet */
parse_offset_marker = prs_offset ( & sec_blob ) ;
if ( ! smb_io_rpc_auth_netsec_chk ( " " , & verf ,
& sec_blob , 0 ) ) {
prs_mem_free ( & sec_blob ) ;
return False ;
}
real_auth_len = prs_offset ( & sec_blob ) - parse_offset_marker ;
}
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
data_len = RPC_HEADER_LEN + RPC_HDR_REQ_LEN + prs_offset ( & sec_blob ) ;
/*
* Malloc parse struct to hold it ( and enough for alignments ) .
*/
if ( ! prs_init ( & outgoing_packet , data_len + 8 ,
cli - > mem_ctx , MARSHALL ) ) {
DEBUG ( 0 , ( " rpc_api_pipe_req: Failed to malloc %u bytes. \n " , ( unsigned int ) data_len ) ) ;
return False ;
}
if ( data_left = = prs_offset ( data ) )
flags | = RPC_FLG_FIRST ;
if ( data_left < = max_data )
flags | = RPC_FLG_LAST ;
/*
* Write out the RPC header and the request header .
*/
if ( ! ( callid = create_rpc_request ( & outgoing_packet , op_num ,
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
data_len , real_auth_len , flags ,
callid , data_left ) ) ) {
DEBUG ( 0 , ( " rpc_api_pipe_req: Failed to create RPC request. \n " ) ) ;
prs_mem_free ( & outgoing_packet ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_mem_free ( & sec_blob ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_append_prs_data ( & outgoing_packet , & sec_blob ) ;
prs_mem_free ( & sec_blob ) ;
DEBUG ( 100 , ( " data_len: %x data_calc_len: %x \n " , data_len ,
prs_offset ( & outgoing_packet ) ) ) ;
if ( flags & RPC_FLG_LAST )
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
ret = rpc_api_pipe ( cli , & outgoing_packet ,
rdata , RPC_RESPONSE ) ;
else {
cli_write ( cli , cli - > nt_pipe_fnum , 0x0008 ,
prs_data_p ( & outgoing_packet ) ,
data_sent , data_len ) ;
}
prs_mem_free ( & outgoing_packet ) ;
data_sent + = send_size ;
data_left - = send_size ;
}
/* Also capture received data */
slprintf ( dump_name , sizeof ( dump_name ) - 1 , " reply_%s " ,
cli_pipe_get_name ( cli ) ) ;
prs_dump ( dump_name , op_num , rdata ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
return ret ;
}
/****************************************************************************
Set the handle state .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
static BOOL rpc_pipe_set_hnd_state ( struct cli_state * cli , const char * pipe_name , uint16 device_state )
{
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
BOOL state_set = False ;
char param [ 2 ] ;
uint16 setup [ 2 ] ; /* only need 2 uint16 setup parameters */
char * rparam = NULL ;
char * rdata = NULL ;
uint32 rparam_len , rdata_len ;
if ( pipe_name = = NULL )
return False ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
DEBUG ( 5 , ( " Set Handle state Pipe[%x]: %s - device state:%x \n " ,
cli - > nt_pipe_fnum , pipe_name , device_state ) ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* create parameters: device state */
SSVAL ( param , 0 , device_state ) ;
/* create setup parameters. */
setup [ 0 ] = 0x0001 ;
setup [ 1 ] = cli - > nt_pipe_fnum ; /* pipe file handle. got this from an SMBOpenX. */
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* send the data on \PIPE\ */
if ( cli_api_pipe ( cli , " \\ PIPE \\ " ,
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
setup , 2 , 0 , /* setup, length, max */
param , 2 , 0 , /* param, length, max */
NULL , 0 , 1024 , /* data, length, max */
& rparam , & rparam_len , /* return param, length */
& rdata , & rdata_len ) ) /* return data, length */
{
DEBUG ( 5 , ( " Set Handle state: return OK \n " ) ) ;
state_set = True ;
}
SAFE_FREE ( rparam ) ;
SAFE_FREE ( rdata ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
return state_set ;
}
/****************************************************************************
check the rpc bind acknowledge response
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
int get_pipe_index ( const char * pipe_name )
{
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
int pipe_idx = 0 ;
while ( pipe_names [ pipe_idx ] . client_pipe ! = NULL ) {
if ( strequal ( pipe_name , pipe_names [ pipe_idx ] . client_pipe ) )
return pipe_idx ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
pipe_idx + + ;
} ;
return - 1 ;
}
/****************************************************************************
check the rpc bind acknowledge response
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
const char * get_pipe_name_from_index ( const int pipe_index )
{
if ( ( pipe_index < 0 ) | | ( pipe_index > = PI_MAX_PIPES ) )
return NULL ;
return pipe_names [ pipe_index ] . client_pipe ;
}
/****************************************************************************
Check to see if this pipe index points to one of
the pipes only supported by Win2k
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
BOOL is_win2k_pipe ( const int pipe_idx )
{
switch ( pipe_idx )
{
case PI_LSARPC_DS :
return True ;
}
return False ;
}
/****************************************************************************
check the rpc bind acknowledge response
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
static BOOL valid_pipe_name ( const int pipe_idx , RPC_IFACE * abstract , RPC_IFACE * transfer )
{
if ( pipe_idx > = PI_MAX_PIPES ) {
DEBUG ( 0 , ( " valid_pipe_name: Programmer error! Invalid pipe index [%d] \n " ,
pipe_idx ) ) ;
return False ;
}
DEBUG ( 5 , ( " Bind Abstract Syntax: " ) ) ;
dump_data ( 5 , ( char * ) & ( pipe_names [ pipe_idx ] . abstr_syntax ) ,
sizeof ( pipe_names [ pipe_idx ] . abstr_syntax ) ) ;
DEBUG ( 5 , ( " Bind Transfer Syntax: " ) ) ;
dump_data ( 5 , ( char * ) & ( pipe_names [ pipe_idx ] . trans_syntax ) ,
sizeof ( pipe_names [ pipe_idx ] . trans_syntax ) ) ;
/* copy the required syntaxes out so we can do the right bind */
* transfer = pipe_names [ pipe_idx ] . trans_syntax ;
* abstract = pipe_names [ pipe_idx ] . abstr_syntax ;
return True ;
}
/****************************************************************************
check the rpc bind acknowledge response
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
static BOOL check_bind_response ( RPC_HDR_BA * hdr_ba , const int pipe_idx , RPC_IFACE * transfer )
{
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
int i = 0 ;
if ( hdr_ba - > addr . len < = 0 )
return False ;
if ( ! strequal ( hdr_ba - > addr . str , pipe_names [ pipe_idx ] . server_pipe ) )
{
DEBUG ( 4 , ( " bind_rpc_pipe: pipe_name %s != expected pipe %s. oh well! \n " ,
pipe_names [ i ] . server_pipe , hdr_ba - > addr . str ) ) ;
return False ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
}
DEBUG ( 5 , ( " bind_rpc_pipe: server pipe_name found: %s \n " , pipe_names [ i ] . server_pipe ) ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
if ( pipe_names [ pipe_idx ] . server_pipe = = NULL ) {
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
DEBUG ( 2 , ( " bind_rpc_pipe: pipe name %s unsupported \n " , hdr_ba - > addr . str ) ) ;
return False ;
}
/* check the transfer syntax */
if ( ( hdr_ba - > transfer . version ! = transfer - > version ) | |
( memcmp ( & hdr_ba - > transfer . uuid , & transfer - > uuid , sizeof ( transfer - > uuid ) ) ! = 0 ) ) {
DEBUG ( 2 , ( " bind_rpc_pipe: transfer syntax differs \n " ) ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
return False ;
}
/* lkclXXXX only accept one result: check the result(s) */
if ( hdr_ba - > res . num_results ! = 0x1 | | hdr_ba - > res . result ! = 0 ) {
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
DEBUG ( 2 , ( " bind_rpc_pipe: bind denied results: %d reason: %x \n " ,
hdr_ba - > res . num_results , hdr_ba - > res . reason ) ) ;
}
DEBUG ( 5 , ( " bind_rpc_pipe: accepted! \n " ) ) ;
return True ;
}
/****************************************************************************
Create and send the third packet in an RPC auth .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
static BOOL rpc_send_auth_reply ( struct cli_state * cli , prs_struct * rdata , uint32 rpc_call_id )
{
prs_struct rpc_out ;
ssize_t ret ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_init ( & rpc_out , RPC_HEADER_LEN + RPC_HDR_AUTHA_LEN , /* need at least this much */
cli - > mem_ctx , MARSHALL ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
create_rpc_bind_resp ( cli , rpc_call_id ,
& rpc_out ) ;
if ( ( ret = cli_write ( cli , cli - > nt_pipe_fnum , 0x8 , prs_data_p ( & rpc_out ) ,
0 , ( size_t ) prs_offset ( & rpc_out ) ) ) ! = ( ssize_t ) prs_offset ( & rpc_out ) ) {
DEBUG ( 0 , ( " rpc_send_auth_reply: cli_write failed. Return was %d \n " , ( int ) ret ) ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_mem_free ( & rpc_out ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_mem_free ( & rpc_out ) ;
return True ;
}
/****************************************************************************
Do an rpc bind .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
static BOOL rpc_pipe_bind ( struct cli_state * cli , int pipe_idx , const char * my_name )
{
RPC_IFACE abstract ;
RPC_IFACE transfer ;
prs_struct rpc_out ;
prs_struct rdata ;
uint32 rpc_call_id ;
char buffer [ MAX_PDU_FRAG_LEN ] ;
if ( ( pipe_idx < 0 ) | | ( pipe_idx > = PI_MAX_PIPES ) )
return False ;
DEBUG ( 5 , ( " Bind RPC Pipe[%x]: %s \n " , cli - > nt_pipe_fnum , pipe_names [ pipe_idx ] . client_pipe ) ) ;
if ( ! valid_pipe_name ( pipe_idx , & abstract , & transfer ) )
return False ;
prs_init ( & rpc_out , 0 , cli - > mem_ctx , MARSHALL ) ;
/*
* Use the MAX_PDU_FRAG_LEN buffer to store the bind request .
*/
prs_give_memory ( & rpc_out , buffer , sizeof ( buffer ) , False ) ;
rpc_call_id = get_rpc_call_id ( ) ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( cli - > pipe_auth_flags & AUTH_PIPE_NTLMSSP ) {
NTSTATUS nt_status ;
fstring password ;
DEBUG ( 5 , ( " NTLMSSP authenticated pipe selected \n " ) ) ;
nt_status = ntlmssp_client_start ( & cli - > ntlmssp_pipe_state ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) )
return False ;
nt_status = ntlmssp_set_username ( cli - > ntlmssp_pipe_state ,
cli - > user_name ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) )
return False ;
nt_status = ntlmssp_set_domain ( cli - > ntlmssp_pipe_state ,
cli - > domain ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) )
return False ;
pwd_get_cleartext ( & cli - > pwd , password ) ;
nt_status = ntlmssp_set_password ( cli - > ntlmssp_pipe_state ,
password ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) )
return False ;
if ( cli - > pipe_auth_flags & AUTH_PIPE_SIGN ) {
cli - > ntlmssp_pipe_state - > neg_flags | = NTLMSSP_NEGOTIATE_SIGN ;
}
if ( cli - > pipe_auth_flags & AUTH_PIPE_SEAL ) {
cli - > ntlmssp_pipe_state - > neg_flags | = NTLMSSP_NEGOTIATE_SEAL ;
}
} else if ( cli - > pipe_auth_flags & AUTH_PIPE_NETSEC ) {
cli - > auth_info . seq_num = 0 ;
}
/* Marshall the outgoing data. */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
create_rpc_bind_req ( cli , & rpc_out , rpc_call_id ,
& abstract , & transfer ,
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
global_myname ( ) , cli - > domain ) ;
/* Initialize the incoming data struct. */
prs_init ( & rdata , 0 , cli - > mem_ctx , UNMARSHALL ) ;
/* send data on \PIPE\. receive a response */
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( rpc_api_pipe ( cli , & rpc_out , & rdata , RPC_BINDACK ) ) {
RPC_HDR_BA hdr_ba ;
DEBUG ( 5 , ( " rpc_pipe_bind: rpc_api_pipe returned OK. \n " ) ) ;
if ( ! smb_io_rpc_hdr_ba ( " " , & hdr_ba , & rdata , 0 ) ) {
DEBUG ( 0 , ( " rpc_pipe_bind: Failed to unmarshall RPC_HDR_BA. \n " ) ) ;
prs_mem_free ( & rdata ) ;
return False ;
}
if ( ! check_bind_response ( & hdr_ba , pipe_idx , & transfer ) ) {
DEBUG ( 2 , ( " rpc_pipe_bind: check_bind_response failed. \n " ) ) ;
prs_mem_free ( & rdata ) ;
return False ;
}
cli - > max_xmit_frag = hdr_ba . bba . max_tsize ;
cli - > max_recv_frag = hdr_ba . bba . max_rsize ;
/*
* If we ' re doing NTLMSSP auth we need to send a reply to
* the bind - ack to complete the 3 - way challenge response
* handshake .
*/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ( cli - > pipe_auth_flags & AUTH_PIPE_NTLMSSP )
& & ! rpc_send_auth_reply ( cli , & rdata , rpc_call_id ) ) {
DEBUG ( 0 , ( " rpc_pipe_bind: rpc_send_auth_reply failed. \n " ) ) ;
prs_mem_free ( & rdata ) ;
return False ;
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
prs_mem_free ( & rdata ) ;
return True ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
return False ;
}
/****************************************************************************
Open a session .
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
BOOL cli_nt_session_open ( struct cli_state * cli , const int pipe_idx )
{
int fnum ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
/* At the moment we can't have more than one pipe open over
a cli connection . ) - : */
SMB_ASSERT ( cli - > nt_pipe_fnum = = 0 ) ;
/* The pipe index must fall within our array */
SMB_ASSERT ( ( pipe_idx > = 0 ) & & ( pipe_idx < PI_MAX_PIPES ) ) ;
if ( cli - > capabilities & CAP_NT_SMBS ) {
if ( ( fnum = cli_nt_create ( cli , & pipe_names [ pipe_idx ] . client_pipe [ 5 ] , DESIRED_ACCESS_PIPE ) ) = = - 1 ) {
DEBUG ( 0 , ( " cli_nt_session_open: cli_nt_create failed on pipe %s to machine %s. Error was %s \n " ,
& pipe_names [ pipe_idx ] . client_pipe [ 5 ] , cli - > desthost , cli_errstr ( cli ) ) ) ;
return False ;
}
cli - > nt_pipe_fnum = ( uint16 ) fnum ;
} else {
if ( ( fnum = cli_open ( cli , pipe_names [ pipe_idx ] . client_pipe , O_CREAT | O_RDWR , DENY_NONE ) ) = = - 1 ) {
DEBUG ( 0 , ( " cli_nt_session_open: cli_open failed on pipe %s to machine %s. Error was %s \n " ,
pipe_names [ pipe_idx ] . client_pipe , cli - > desthost , cli_errstr ( cli ) ) ) ;
return False ;
}
cli - > nt_pipe_fnum = ( uint16 ) fnum ;
/**************** Set Named Pipe State ***************/
if ( ! rpc_pipe_set_hnd_state ( cli , pipe_names [ pipe_idx ] . client_pipe , 0x4300 ) ) {
DEBUG ( 0 , ( " cli_nt_session_open: pipe hnd state failed. Error was %s \n " ,
cli_errstr ( cli ) ) ) ;
cli_close ( cli , cli - > nt_pipe_fnum ) ;
return False ;
}
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
}
/******************* bind request on pipe *****************/
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! rpc_pipe_bind ( cli , pipe_idx , global_myname ( ) ) ) {
DEBUG ( 2 , ( " cli_nt_session_open: rpc bind to %s failed \n " ,
get_pipe_name_from_index ( pipe_idx ) ) ) ;
cli_close ( cli , cli - > nt_pipe_fnum ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
return False ;
}
/*
* Setup the remote server name prefixed by \ and the machine account name .
*/
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
fstrcpy ( cli - > srv_name_slash , " \\ \\ " ) ;
fstrcat ( cli - > srv_name_slash , cli - > desthost ) ;
strupper_m ( cli - > srv_name_slash ) ;
fstrcpy ( cli - > clnt_name_slash , " \\ \\ " ) ;
fstrcat ( cli - > clnt_name_slash , global_myname ( ) ) ;
strupper_m ( cli - > clnt_name_slash ) ;
fstrcpy ( cli - > mach_acct , global_myname ( ) ) ;
fstrcat ( cli - > mach_acct , " $ " ) ;
strupper_m ( cli - > mach_acct ) ;
/* Remember which pipe we're talking to */
fstrcpy ( cli - > pipe_name , pipe_names [ pipe_idx ] . client_pipe ) ;
Makefile.in :
- added srvsvc client files
clientgen.c :
- replaced cli_error(cli, int *cls, int *err) with
cli_error(cli, uint8 cls, uint32 *err). this version detects
32 bit status messages. the DOS error "MORE_DATA", the
equivalent of the 32 bit *warning* 0x8000 0005
(STATUS_BUFFER_OVERFLOW), was being processed as an error,
terminating the cli_receive_trans() call.
cli_pipe.c :
- replaced calls that had been incorrectly modified from
32 bit warnings (0x8000 0005 - STATUS_BUFFER_OVERFLOW)
to 8 bit DOS errors (0x01 0xEA - MORE_DATA).
the use of the old version of cli_error (DOS only)
instead of the new one (DOS and 32 bit) caused the
dce/rpc client code to fail.
- replaced 2 space indentation with tab indentation in all functions.
cli_srvsvc.c :
cmd_srvsvc.c :
- added these files back in, fixing them up to use jeremy's
modified versions of the dce/rpc client functions.
parse_srv.c :
- added back in some "unused" functions required by dce/rpc
client-side code. it would be helpful if all such "unused"
functions could be added back in.
rpcclient.c :
- added "session", "file", "share", "connection" enumeration
functions back in. these are equivalent to nt's "NetXXXXXEnum"
Win32 (MSDN) functions.
- added "srvinfo" function back in. this is equivalent to
nt's NetServerGetInfo Win32 (MSDN) function.
-
return True ;
}
/****************************************************************************
Open a session to the NETLOGON pipe using schannel .
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
( Assumes that the netlogon pipe is already open )
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
NTSTATUS cli_nt_establish_netlogon ( struct cli_state * cli , int sec_chan ,
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
const uchar trust_password [ 16 ] )
{
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
NTSTATUS result ;
uint32 neg_flags = 0x000001ff ;
int fnum ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
cli_nt_netlogon_netsec_session_close ( cli ) ;
if ( lp_client_schannel ( ) ! = False )
neg_flags | = NETLOGON_NEG_SCHANNEL ;
result = cli_nt_setup_creds ( cli , sec_chan , trust_password ,
& neg_flags , 2 ) ;
if ( ! NT_STATUS_IS_OK ( result ) ) {
cli_nt_session_close ( cli ) ;
return result ;
}
if ( ( lp_client_schannel ( ) = = True ) & &
( ( neg_flags & NETLOGON_NEG_SCHANNEL ) = = 0 ) ) {
DEBUG ( 3 , ( " Server did not offer schannel \n " ) ) ;
cli_nt_session_close ( cli ) ;
return NT_STATUS_UNSUCCESSFUL ;
}
if ( ( lp_client_schannel ( ) = = False ) | |
( ( neg_flags & NETLOGON_NEG_SCHANNEL ) = = 0 ) ) {
return NT_STATUS_OK ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
/* keep the existing connection to NETLOGON open */
}
/* Server offered schannel, so try it. */
memcpy ( cli - > auth_info . sess_key , cli - > sess_key ,
sizeof ( cli - > auth_info . sess_key ) ) ;
cli - > saved_netlogon_pipe_fnum = cli - > nt_pipe_fnum ;
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
cli - > pipe_auth_flags = AUTH_PIPE_NETSEC ;
cli - > pipe_auth_flags | = AUTH_PIPE_SIGN ;
cli - > pipe_auth_flags | = AUTH_PIPE_SEAL ;
if ( cli - > capabilities & CAP_NT_SMBS ) {
/* The secure channel connection must be opened on the same
session ( TCP connection ) as the one the challenge was
requested from . */
if ( ( fnum = cli_nt_create ( cli , PIPE_NETLOGON_PLAIN ,
DESIRED_ACCESS_PIPE ) ) = = - 1 ) {
DEBUG ( 0 , ( " cli_nt_create failed to %s machine %s. "
" Error was %s \n " ,
PIPE_NETLOGON , cli - > desthost ,
cli_errstr ( cli ) ) ) ;
return NT_STATUS_UNSUCCESSFUL ;
}
cli - > nt_pipe_fnum = ( uint16 ) fnum ;
} else {
if ( ( fnum = cli_open ( cli , PIPE_NETLOGON ,
O_CREAT | O_RDWR , DENY_NONE ) ) = = - 1 ) {
DEBUG ( 0 , ( " cli_open failed on pipe %s to machine %s. "
" Error was %s \n " ,
PIPE_NETLOGON , cli - > desthost ,
cli_errstr ( cli ) ) ) ;
return NT_STATUS_UNSUCCESSFUL ;
}
cli - > nt_pipe_fnum = ( uint16 ) fnum ;
/**************** Set Named Pipe State ***************/
if ( ! rpc_pipe_set_hnd_state ( cli , PIPE_NETLOGON , 0x4300 ) ) {
DEBUG ( 0 , ( " Pipe hnd state failed. Error was %s \n " ,
cli_errstr ( cli ) ) ) ;
cli_close ( cli , cli - > nt_pipe_fnum ) ;
return NT_STATUS_UNSUCCESSFUL ;
}
}
Jeremy requested that I get my NTLMSSP patch into CVS. He didn't request
the schannel code, but I've included that anyway. :-)
This patch revives the client-side NTLMSSP support for RPC named pipes
in Samba, and cleans up the client and server schannel code. The use of the
new code is enabled by the 'sign', 'seal' and 'schannel' commands in
rpcclient.
The aim was to prove that our separate NTLMSSP client library actually
implements NTLMSSP signing and sealing as per Microsoft's NTLMv1 implementation,
in the hope that knowing this will assist us in correctly implementing
NTLMSSP signing for SMB packets. (Still not yet functional)
This patch replaces the NTLMSSP implementation in rpc_client/cli_pipe.c with
calls to libsmb/ntlmssp.c. In the process, we have gained the ability to
use the more secure NT password, and the ability to sign-only, instead of
having to seal the pipe connection. (Previously we were limited to sealing,
and could only use the LM-password derived key).
Our new client-side NTLMSSP code also needed alteration to cope with our
comparatively simple server-side implementation. A future step is to replace
it with calls to the same NTLMSSP library.
Also included in this patch is the schannel 'sign only' patch I submitted to
the team earlier. While not enabled (and not functional, at this stage) the
work in this patch makes the code paths *much* easier to follow. I have also
included similar hooks in rpccleint to allow the use of schannel on *any* pipe.
rpcclient now defaults to not using schannel (or any other extra per-pipe
authenticiation) for any connection. The 'schannel' command enables schannel
for all pipes until disabled.
This code is also much more secure than the previous code, as changes to our
cli_pipe routines ensure that the authentication footer cannot be removed
by an attacker, and more error states are correctly handled.
(The same needs to be done to our server)
Andrew Bartlett
-
if ( ! rpc_pipe_bind ( cli , PI_NETLOGON , global_myname ( ) ) ) {
DEBUG ( 2 , ( " rpc bind to %s failed \n " , PIPE_NETLOGON ) ) ;
cli_close ( cli , cli - > nt_pipe_fnum ) ;
return NT_STATUS_UNSUCCESSFUL ;
}
return NT_STATUS_OK ;
}
NTSTATUS cli_nt_setup_netsec ( struct cli_state * cli , int sec_chan ,
const uchar trust_password [ 16 ] )
{
NTSTATUS result ;
uint32 neg_flags = 0x000001ff ;
cli - > pipe_auth_flags = 0 ;
if ( lp_client_schannel ( ) = = False ) {
return NT_STATUS_OK ;
}
if ( ! cli_nt_session_open ( cli , PI_NETLOGON ) ) {
DEBUG ( 0 , ( " Could not initialise %s \n " ,
get_pipe_name_from_index ( PI_NETLOGON ) ) ) ;
return NT_STATUS_UNSUCCESSFUL ;
}
if ( lp_client_schannel ( ) ! = False )
neg_flags | = NETLOGON_NEG_SCHANNEL ;
neg_flags | = NETLOGON_NEG_SCHANNEL ;
result = cli_nt_setup_creds ( cli , sec_chan , trust_password ,
& neg_flags , 2 ) ;
if ( ! ( neg_flags & NETLOGON_NEG_SCHANNEL )
& & lp_client_schannel ( ) = = True ) {
DEBUG ( 1 , ( " Could not negotiate SCHANNEL with the DC! \n " ) ) ;
result = NT_STATUS_UNSUCCESSFUL ;
}
if ( ! NT_STATUS_IS_OK ( result ) ) {
ZERO_STRUCT ( cli - > auth_info . sess_key ) ;
ZERO_STRUCT ( cli - > sess_key ) ;
cli - > pipe_auth_flags = 0 ;
cli_nt_session_close ( cli ) ;
return result ;
}
memcpy ( cli - > auth_info . sess_key , cli - > sess_key ,
sizeof ( cli - > auth_info . sess_key ) ) ;
cli - > saved_netlogon_pipe_fnum = cli - > nt_pipe_fnum ;
cli - > nt_pipe_fnum = 0 ;
/* doing schannel, not per-user auth */
cli - > pipe_auth_flags = AUTH_PIPE_NETSEC | AUTH_PIPE_SIGN | AUTH_PIPE_SEAL ;
return NT_STATUS_OK ;
}
const char * cli_pipe_get_name ( struct cli_state * cli )
{
return cli - > pipe_name ;
}