2002-11-24 11:32:03 +03:00
/*
Unix SMB / CIFS implementation .
Winbind status program .
2004-04-06 20:44:24 +04:00
Copyright ( C ) Tim Potter 2000 - 2003
Copyright ( C ) Andrew Bartlett < abartlet @ samba . org > 2003 - 2004
2002-11-24 11:32:03 +03:00
Copyright ( C ) Francesco Chemolli < kinkie @ kame . usr . dsi . unimi . it > 2000
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"
2004-10-07 08:01:18 +04:00
# include "utils/ntlm_auth.h"
2002-11-24 11:32:03 +03:00
# undef DBGC_CLASS
# define DBGC_CLASS DBGC_WINBIND
# define SQUID_BUFFER_SIZE 2010
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
enum stdio_helper_mode {
2003-01-16 06:29:54 +03:00
SQUID_2_4_BASIC ,
2002-11-25 00:08:36 +03:00
SQUID_2_5_BASIC ,
2003-07-29 19:00:38 +04:00
SQUID_2_5_NTLMSSP ,
2003-12-30 11:52:46 +03:00
NTLMSSP_CLIENT_1 ,
2003-08-01 11:59:23 +04:00
GSS_SPNEGO ,
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
GSS_SPNEGO_CLIENT ,
2004-04-13 00:46:13 +04:00
NTLM_SERVER_1 ,
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
NUM_HELPER_MODES
} ;
typedef void ( * stdio_helper_function ) ( enum stdio_helper_mode stdio_helper_mode ,
char * buf , int length ) ;
static void manage_squid_basic_request ( enum stdio_helper_mode stdio_helper_mode ,
char * buf , int length ) ;
static void manage_squid_ntlmssp_request ( enum stdio_helper_mode stdio_helper_mode ,
char * buf , int length ) ;
2003-12-30 10:33:58 +03:00
static void manage_client_ntlmssp_request ( enum stdio_helper_mode stdio_helper_mode ,
char * buf , int length ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static void manage_gss_spnego_request ( enum stdio_helper_mode stdio_helper_mode ,
char * buf , int length ) ;
static void manage_gss_spnego_client_request ( enum stdio_helper_mode stdio_helper_mode ,
char * buf , int length ) ;
2004-04-13 00:46:13 +04:00
static void manage_ntlm_server_1_request ( enum stdio_helper_mode stdio_helper_mode ,
char * buf , int length ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static const struct {
enum stdio_helper_mode mode ;
const char * name ;
stdio_helper_function fn ;
} stdio_helper_protocols [ ] = {
{ SQUID_2_4_BASIC , " squid-2.4-basic " , manage_squid_basic_request } ,
{ SQUID_2_5_BASIC , " squid-2.5-basic " , manage_squid_basic_request } ,
{ SQUID_2_5_NTLMSSP , " squid-2.5-ntlmssp " , manage_squid_ntlmssp_request } ,
2003-12-30 11:52:46 +03:00
{ NTLMSSP_CLIENT_1 , " ntlmssp-client-1 " , manage_client_ntlmssp_request } ,
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
{ GSS_SPNEGO , " gss-spnego " , manage_gss_spnego_request } ,
{ GSS_SPNEGO_CLIENT , " gss-spnego-client " , manage_gss_spnego_client_request } ,
2004-04-13 00:46:13 +04:00
{ NTLM_SERVER_1 , " ntlm-server-1 " , manage_ntlm_server_1_request } ,
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
{ NUM_HELPER_MODES , NULL , NULL }
2002-11-25 00:08:36 +03:00
} ;
2002-11-24 11:32:03 +03:00
extern int winbindd_fd ;
2004-04-12 18:19:48 +04:00
const char * opt_username ;
const char * opt_domain ;
const char * opt_workstation ;
const char * opt_password ;
2003-04-07 11:10:53 +04:00
static DATA_BLOB opt_challenge ;
static DATA_BLOB opt_lm_response ;
static DATA_BLOB opt_nt_response ;
2003-03-24 12:54:13 +03:00
static int request_lm_key ;
2004-04-12 14:17:48 +04:00
static int request_user_session_key ;
2002-11-24 11:32:03 +03:00
2004-04-06 20:44:24 +04:00
static const char * require_membership_of ;
2004-10-01 07:28:39 +04:00
static const char * require_membership_of_sid ;
2002-11-24 11:32:03 +03:00
static char winbind_separator ( void )
{
struct winbindd_response response ;
static BOOL got_sep ;
static char sep ;
if ( got_sep )
return sep ;
ZERO_STRUCT ( response ) ;
/* Send off request */
2005-06-25 00:25:18 +04:00
if ( winbindd_request_response ( WINBINDD_INFO , NULL , & response ) ! =
2002-11-24 11:32:03 +03:00
NSS_STATUS_SUCCESS ) {
d_printf ( " could not obtain winbind separator! \n " ) ;
2004-04-12 14:17:48 +04:00
return * lp_winbind_separator ( ) ;
2002-11-24 11:32:03 +03:00
}
sep = response . data . info . winbind_separator ;
got_sep = True ;
if ( ! sep ) {
d_printf ( " winbind separator was NULL! \n " ) ;
2004-04-12 14:17:48 +04:00
return * lp_winbind_separator ( ) ;
2002-11-24 11:32:03 +03:00
}
return sep ;
}
2004-04-12 18:19:48 +04:00
const char * get_winbind_domain ( void )
2002-11-24 11:32:03 +03:00
{
struct winbindd_response response ;
static fstring winbind_domain ;
2003-01-16 06:29:54 +03:00
if ( * winbind_domain ) {
return winbind_domain ;
}
2002-11-24 11:32:03 +03:00
ZERO_STRUCT ( response ) ;
/* Send off request */
2005-06-25 00:25:18 +04:00
if ( winbindd_request_response ( WINBINDD_DOMAIN_NAME , NULL , & response ) ! =
2002-11-24 11:32:03 +03:00
NSS_STATUS_SUCCESS ) {
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
DEBUG ( 0 , ( " could not obtain winbind domain name! \n " ) ) ;
2004-04-12 14:17:48 +04:00
return lp_workgroup ( ) ;
2002-11-24 11:32:03 +03:00
}
fstrcpy ( winbind_domain , response . data . domain_name ) ;
return winbind_domain ;
}
2004-04-12 18:19:48 +04:00
const char * get_winbind_netbios_name ( void )
2003-01-16 06:29:54 +03:00
{
struct winbindd_response response ;
static fstring winbind_netbios_name ;
if ( * winbind_netbios_name ) {
return winbind_netbios_name ;
}
ZERO_STRUCT ( response ) ;
/* Send off request */
2005-06-25 00:25:18 +04:00
if ( winbindd_request_response ( WINBINDD_NETBIOS_NAME , NULL , & response ) ! =
2003-01-16 06:29:54 +03:00
NSS_STATUS_SUCCESS ) {
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
DEBUG ( 0 , ( " could not obtain winbind netbios name! \n " ) ) ;
2004-04-12 14:17:48 +04:00
return global_myname ( ) ;
2003-01-16 06:29:54 +03:00
}
fstrcpy ( winbind_netbios_name , response . data . netbios_name ) ;
return winbind_netbios_name ;
}
2004-04-12 18:19:48 +04:00
DATA_BLOB get_challenge ( void )
{
static DATA_BLOB chal ;
if ( opt_challenge . length )
return opt_challenge ;
chal = data_blob ( NULL , 8 ) ;
2004-07-14 08:36:01 +04:00
generate_random_buffer ( chal . data , chal . length ) ;
2004-04-12 18:19:48 +04:00
return chal ;
}
2004-04-06 20:44:24 +04:00
/* Copy of parse_domain_user from winbindd_util.c. Parse a string of the
form DOMAIN / user into a domain and a user */
static BOOL parse_ntlm_auth_domain_user ( const char * domuser , fstring domain ,
fstring user )
{
char * p = strchr ( domuser , winbind_separator ( ) ) ;
if ( ! p ) {
return False ;
}
fstrcpy ( user , p + 1 ) ;
fstrcpy ( domain , domuser ) ;
domain [ PTR_DIFF ( p , domuser ) ] = 0 ;
strupper_m ( domain ) ;
return True ;
}
static BOOL get_require_membership_sid ( void ) {
struct winbindd_request request ;
struct winbindd_response response ;
if ( ! require_membership_of ) {
return True ;
}
2004-10-01 07:28:39 +04:00
if ( require_membership_of_sid ) {
2004-04-06 20:44:24 +04:00
return True ;
}
/* Otherwise, ask winbindd for the name->sid request */
ZERO_STRUCT ( request ) ;
ZERO_STRUCT ( response ) ;
if ( ! parse_ntlm_auth_domain_user ( require_membership_of ,
request . data . name . dom_name ,
request . data . name . name ) ) {
DEBUG ( 0 , ( " Could not parse %s into seperate domain/name parts! \n " ,
require_membership_of ) ) ;
return False ;
}
2005-06-25 00:25:18 +04:00
if ( winbindd_request_response ( WINBINDD_LOOKUPNAME , & request , & response ) ! =
2004-04-06 20:44:24 +04:00
NSS_STATUS_SUCCESS ) {
DEBUG ( 0 , ( " Winbindd lookupname failed to resolve %s into a SID! \n " ,
require_membership_of ) ) ;
return False ;
}
2004-12-07 21:25:53 +03:00
require_membership_of_sid = SMB_STRDUP ( response . data . sid . sid ) ;
2004-04-06 20:44:24 +04:00
2004-10-01 07:28:39 +04:00
if ( require_membership_of_sid )
2004-04-06 20:44:24 +04:00
return True ;
return False ;
}
2002-11-24 11:32:03 +03:00
/* Authenticate a user with a plaintext password */
2004-04-06 20:44:24 +04:00
static BOOL check_plaintext_auth ( const char * user , const char * pass ,
BOOL stdout_diagnostics )
2002-11-24 11:32:03 +03:00
{
struct winbindd_request request ;
struct winbindd_response response ;
NSS_STATUS result ;
2004-04-06 20:44:24 +04:00
if ( ! get_require_membership_sid ( ) ) {
return False ;
}
2002-11-24 11:32:03 +03:00
/* Send off request */
ZERO_STRUCT ( request ) ;
ZERO_STRUCT ( response ) ;
fstrcpy ( request . data . auth . user , user ) ;
fstrcpy ( request . data . auth . pass , pass ) ;
2004-10-01 07:28:39 +04:00
if ( require_membership_of_sid )
fstrcpy ( request . data . auth . require_membership_of_sid , require_membership_of_sid ) ;
2002-11-24 11:32:03 +03:00
2005-06-25 00:25:18 +04:00
result = winbindd_request_response ( WINBINDD_PAM_AUTH , & request , & response ) ;
2002-11-24 11:32:03 +03:00
/* Display response */
if ( stdout_diagnostics ) {
if ( ( result ! = NSS_STATUS_SUCCESS ) & & ( response . data . auth . nt_status = = 0 ) ) {
d_printf ( " Reading winbind reply failed! (0x01) \n " ) ;
}
2003-04-07 11:10:53 +04:00
d_printf ( " %s: %s (0x%x) \n " ,
2002-11-24 11:32:03 +03:00
response . data . auth . nt_status_string ,
2003-04-07 11:10:53 +04:00
response . data . auth . error_string ,
2002-11-24 11:32:03 +03:00
response . data . auth . nt_status ) ;
} else {
if ( ( result ! = NSS_STATUS_SUCCESS ) & & ( response . data . auth . nt_status = = 0 ) ) {
DEBUG ( 1 , ( " Reading winbind reply failed! (0x01) \n " ) ) ;
}
2003-04-07 11:10:53 +04:00
DEBUG ( 3 , ( " %s: %s (0x%x) \n " ,
response . data . auth . nt_status_string ,
response . data . auth . error_string ,
response . data . auth . nt_status ) ) ;
2002-11-24 11:32:03 +03:00
}
return ( result = = NSS_STATUS_SUCCESS ) ;
}
2003-04-07 11:10:53 +04:00
/* authenticate a user with an encrypted username/password */
2004-04-12 18:19:48 +04:00
NTSTATUS contact_winbind_auth_crap ( const char * username ,
const char * domain ,
const char * workstation ,
const DATA_BLOB * challenge ,
const DATA_BLOB * lm_response ,
const DATA_BLOB * nt_response ,
uint32 flags ,
uint8 lm_key [ 8 ] ,
uint8 user_session_key [ 16 ] ,
char * * error_string ,
char * * unix_name )
2003-01-16 06:29:54 +03:00
{
2003-04-07 11:10:53 +04:00
NTSTATUS nt_status ;
NSS_STATUS result ;
2003-01-16 06:29:54 +03:00
struct winbindd_request request ;
struct winbindd_response response ;
2003-04-07 11:10:53 +04:00
2004-04-06 20:44:24 +04:00
if ( ! get_require_membership_sid ( ) ) {
return NT_STATUS_INVALID_PARAMETER ;
}
2003-01-16 06:29:54 +03:00
ZERO_STRUCT ( request ) ;
ZERO_STRUCT ( response ) ;
2003-07-07 09:11:10 +04:00
request . flags = flags ;
2003-01-16 06:29:54 +03:00
2005-11-08 09:19:34 +03:00
request . data . auth_crap . logon_parameters = MSV1_0_ALLOW_WORKSTATION_TRUST_ACCOUNT | MSV1_0_ALLOW_SERVER_TRUST_ACCOUNT ;
2004-10-01 07:28:39 +04:00
if ( require_membership_of_sid )
fstrcpy ( request . data . auth_crap . require_membership_of_sid , require_membership_of_sid ) ;
2004-04-06 20:44:24 +04:00
2004-10-01 07:28:39 +04:00
fstrcpy ( request . data . auth_crap . user , username ) ;
fstrcpy ( request . data . auth_crap . domain , domain ) ;
2003-04-07 11:10:53 +04:00
2004-10-01 07:28:39 +04:00
fstrcpy ( request . data . auth_crap . workstation ,
workstation ) ;
2003-04-07 11:10:53 +04:00
memcpy ( request . data . auth_crap . chal , challenge - > data , MIN ( challenge - > length , 8 ) ) ;
2003-01-16 06:29:54 +03:00
2003-04-07 11:10:53 +04:00
if ( lm_response & & lm_response - > length ) {
2004-04-06 20:44:24 +04:00
memcpy ( request . data . auth_crap . lm_resp ,
lm_response - > data ,
MIN ( lm_response - > length , sizeof ( request . data . auth_crap . lm_resp ) ) ) ;
2003-04-07 11:10:53 +04:00
request . data . auth_crap . lm_resp_len = lm_response - > length ;
}
if ( nt_response & & nt_response - > length ) {
2004-04-06 20:44:24 +04:00
memcpy ( request . data . auth_crap . nt_resp ,
nt_response - > data ,
MIN ( nt_response - > length , sizeof ( request . data . auth_crap . nt_resp ) ) ) ;
2003-04-07 11:10:53 +04:00
request . data . auth_crap . nt_resp_len = nt_response - > length ;
}
2005-06-25 00:25:18 +04:00
result = winbindd_request_response ( WINBINDD_PAM_AUTH_CRAP , & request , & response ) ;
2003-01-16 06:29:54 +03:00
/* Display response */
if ( ( result ! = NSS_STATUS_SUCCESS ) & & ( response . data . auth . nt_status = = 0 ) ) {
2003-04-07 11:10:53 +04:00
nt_status = NT_STATUS_UNSUCCESSFUL ;
if ( error_string )
* error_string = smb_xstrdup ( " Reading winbind reply failed! " ) ;
2004-06-14 06:56:48 +04:00
free_response ( & response ) ;
2003-04-07 11:10:53 +04:00
return nt_status ;
}
nt_status = ( NT_STATUS ( response . data . auth . nt_status ) ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
if ( error_string )
* error_string = smb_xstrdup ( response . data . auth . error_string ) ;
2004-06-14 06:56:48 +04:00
free_response ( & response ) ;
2003-04-07 11:10:53 +04:00
return nt_status ;
2003-01-16 06:29:54 +03:00
}
2004-04-12 14:17:48 +04:00
if ( ( flags & WBFLAG_PAM_LMKEY ) & & lm_key ) {
2003-04-07 11:10:53 +04:00
memcpy ( lm_key , response . data . auth . first_8_lm_hash ,
2004-04-12 14:17:48 +04:00
sizeof ( response . data . auth . first_8_lm_hash ) ) ;
2003-04-07 11:10:53 +04:00
}
2004-04-12 14:17:48 +04:00
if ( ( flags & WBFLAG_PAM_USER_SESSION_KEY ) & & user_session_key ) {
memcpy ( user_session_key , response . data . auth . user_session_key ,
2004-04-06 12:11:16 +04:00
sizeof ( response . data . auth . user_session_key ) ) ;
2003-04-07 11:10:53 +04:00
}
2003-12-30 16:20:39 +03:00
if ( flags & WBFLAG_PAM_UNIX_NAME ) {
2004-12-07 21:25:53 +03:00
* unix_name = SMB_STRDUP ( ( char * ) response . extra_data ) ;
2004-10-01 07:28:39 +04:00
if ( ! * unix_name ) {
2004-06-14 06:56:48 +04:00
free_response ( & response ) ;
2003-12-31 01:27:33 +03:00
return NT_STATUS_NO_MEMORY ;
}
2003-12-30 16:20:39 +03:00
}
2004-06-14 06:56:48 +04:00
free_response ( & response ) ;
2003-04-07 11:10:53 +04:00
return nt_status ;
}
2004-04-06 12:11:16 +04:00
static NTSTATUS winbind_pw_check ( struct ntlmssp_state * ntlmssp_state , DATA_BLOB * user_session_key , DATA_BLOB * lm_session_key )
2003-04-07 11:10:53 +04:00
{
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static const char zeros [ 16 ] ;
NTSTATUS nt_status ;
char * error_string ;
uint8 lm_key [ 8 ] ;
2004-04-12 14:17:48 +04:00
uint8 user_sess_key [ 16 ] ;
2003-12-30 16:20:39 +03:00
char * unix_name ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
nt_status = contact_winbind_auth_crap ( ntlmssp_state - > user , ntlmssp_state - > domain ,
ntlmssp_state - > workstation ,
& ntlmssp_state - > chal ,
& ntlmssp_state - > lm_resp ,
& ntlmssp_state - > nt_resp ,
2004-04-06 12:11:16 +04:00
WBFLAG_PAM_LMKEY | WBFLAG_PAM_USER_SESSION_KEY | WBFLAG_PAM_UNIX_NAME ,
2004-04-12 14:17:48 +04:00
lm_key , user_sess_key ,
2003-12-30 16:20:39 +03:00
& error_string , & unix_name ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
if ( NT_STATUS_IS_OK ( nt_status ) ) {
if ( memcmp ( lm_key , zeros , 8 ) ! = 0 ) {
* lm_session_key = data_blob ( NULL , 16 ) ;
memcpy ( lm_session_key - > data , lm_key , 8 ) ;
memset ( lm_session_key - > data + 8 , ' \0 ' , 8 ) ;
}
2004-04-12 14:17:48 +04:00
if ( memcmp ( user_sess_key , zeros , 16 ) ! = 0 ) {
* user_session_key = data_blob ( user_sess_key , 16 ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
}
2003-12-30 16:20:39 +03:00
ntlmssp_state - > auth_context = talloc_strdup ( ntlmssp_state - > mem_ctx , unix_name ) ;
SAFE_FREE ( unix_name ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
} else {
DEBUG ( NT_STATUS_EQUAL ( nt_status , NT_STATUS_ACCESS_DENIED ) ? 0 : 3 ,
( " Login for user [%s] \\ [%s]@[%s] failed due to [%s] \n " ,
2004-04-06 20:44:24 +04:00
ntlmssp_state - > domain , ntlmssp_state - > user ,
ntlmssp_state - > workstation ,
error_string ? error_string : " unknown error (NULL) " ) ) ;
2003-12-30 16:20:39 +03:00
ntlmssp_state - > auth_context = NULL ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
}
return nt_status ;
2003-01-16 06:29:54 +03:00
}
2004-04-06 12:11:16 +04:00
static NTSTATUS local_pw_check ( struct ntlmssp_state * ntlmssp_state , DATA_BLOB * user_session_key , DATA_BLOB * lm_session_key )
2003-12-30 10:33:58 +03:00
{
NTSTATUS nt_status ;
uint8 lm_pw [ 16 ] , nt_pw [ 16 ] ;
nt_lm_owf_gen ( opt_password , nt_pw , lm_pw ) ;
nt_status = ntlm_password_check ( ntlmssp_state - > mem_ctx ,
& ntlmssp_state - > chal ,
& ntlmssp_state - > lm_resp ,
& ntlmssp_state - > nt_resp ,
2004-04-03 19:41:32 +04:00
NULL , NULL ,
2003-12-30 10:33:58 +03:00
ntlmssp_state - > user ,
ntlmssp_state - > user ,
ntlmssp_state - > domain ,
2004-04-06 12:11:16 +04:00
lm_pw , nt_pw , user_session_key , lm_session_key ) ;
2003-12-30 10:33:58 +03:00
if ( NT_STATUS_IS_OK ( nt_status ) ) {
2004-04-06 20:44:24 +04:00
ntlmssp_state - > auth_context = talloc_asprintf ( ntlmssp_state - > mem_ctx ,
" %s%c%s " , ntlmssp_state - > domain ,
* lp_winbind_separator ( ) ,
ntlmssp_state - > user ) ;
2003-12-30 10:33:58 +03:00
} else {
DEBUG ( 3 , ( " Login for user [%s] \\ [%s]@[%s] failed due to [%s] \n " ,
ntlmssp_state - > domain , ntlmssp_state - > user , ntlmssp_state - > workstation ,
nt_errstr ( nt_status ) ) ) ;
2003-12-30 16:20:39 +03:00
ntlmssp_state - > auth_context = NULL ;
2003-12-30 10:33:58 +03:00
}
return nt_status ;
}
static NTSTATUS ntlm_auth_start_ntlmssp_client ( NTLMSSP_STATE * * client_ntlmssp_state )
{
NTSTATUS status ;
if ( ( opt_username = = NULL ) | | ( opt_domain = = NULL ) ) {
2005-09-30 21:13:37 +04:00
status = NT_STATUS_UNSUCCESSFUL ;
2003-12-30 10:33:58 +03:00
DEBUG ( 1 , ( " Need username and domain for NTLMSSP \n " ) ) ;
2004-10-01 07:28:39 +04:00
return NT_STATUS_INVALID_PARAMETER ;
2003-12-30 10:33:58 +03:00
}
status = ntlmssp_client_start ( client_ntlmssp_state ) ;
if ( ! NT_STATUS_IS_OK ( status ) ) {
DEBUG ( 1 , ( " Could not start NTLMSSP client: %s \n " ,
nt_errstr ( status ) ) ) ;
ntlmssp_end ( client_ntlmssp_state ) ;
return status ;
}
status = ntlmssp_set_username ( * client_ntlmssp_state , opt_username ) ;
if ( ! NT_STATUS_IS_OK ( status ) ) {
DEBUG ( 1 , ( " Could not set username: %s \n " ,
nt_errstr ( status ) ) ) ;
ntlmssp_end ( client_ntlmssp_state ) ;
return status ;
}
status = ntlmssp_set_domain ( * client_ntlmssp_state , opt_domain ) ;
if ( ! NT_STATUS_IS_OK ( status ) ) {
DEBUG ( 1 , ( " Could not set domain: %s \n " ,
nt_errstr ( status ) ) ) ;
ntlmssp_end ( client_ntlmssp_state ) ;
return status ;
}
status = ntlmssp_set_password ( * client_ntlmssp_state , opt_password ) ;
if ( ! NT_STATUS_IS_OK ( status ) ) {
DEBUG ( 1 , ( " Could not set password: %s \n " ,
nt_errstr ( status ) ) ) ;
ntlmssp_end ( client_ntlmssp_state ) ;
return status ;
}
return NT_STATUS_OK ;
}
static NTSTATUS ntlm_auth_start_ntlmssp_server ( NTLMSSP_STATE * * ntlmssp_state )
{
NTSTATUS status = ntlmssp_server_start ( ntlmssp_state ) ;
if ( ! NT_STATUS_IS_OK ( status ) ) {
DEBUG ( 1 , ( " Could not start NTLMSSP client: %s \n " ,
nt_errstr ( status ) ) ) ;
return status ;
}
/* Have we been given a local password, or should we ask winbind? */
if ( opt_password ) {
( * ntlmssp_state ) - > check_password = local_pw_check ;
( * ntlmssp_state ) - > get_domain = lp_workgroup ;
( * ntlmssp_state ) - > get_global_myname = global_myname ;
} else {
( * ntlmssp_state ) - > check_password = winbind_pw_check ;
( * ntlmssp_state ) - > get_domain = get_winbind_domain ;
( * ntlmssp_state ) - > get_global_myname = get_winbind_netbios_name ;
}
return NT_STATUS_OK ;
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static void manage_squid_ntlmssp_request ( enum stdio_helper_mode stdio_helper_mode ,
2003-01-16 06:29:54 +03:00
char * buf , int length )
{
2003-03-24 12:54:13 +03:00
static NTLMSSP_STATE * ntlmssp_state = NULL ;
2003-01-16 06:29:54 +03:00
DATA_BLOB request , reply ;
NTSTATUS nt_status ;
2003-03-24 12:54:13 +03:00
if ( strlen ( buf ) < 2 ) {
DEBUG ( 1 , ( " NTLMSSP query [%s] invalid " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
2003-12-30 10:33:58 +03:00
if ( strlen ( buf ) > 3 ) {
request = base64_decode_data_blob ( buf + 3 ) ;
} else {
request = data_blob ( NULL , 0 ) ;
}
if ( ( strncmp ( buf , " PW " , 3 ) = = 0 ) ) {
/* The calling application wants us to use a local password (rather than winbindd) */
2004-12-07 21:25:53 +03:00
opt_password = SMB_STRNDUP ( ( const char * ) request . data , request . length ) ;
2003-12-30 10:33:58 +03:00
if ( opt_password = = NULL ) {
DEBUG ( 1 , ( " Out of memory \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
data_blob_free ( & request ) ;
return ;
}
x_fprintf ( x_stdout , " OK \n " ) ;
data_blob_free ( & request ) ;
return ;
}
2003-12-24 12:56:51 +03:00
if ( strncmp ( buf , " YR " , 2 ) = = 0 ) {
2003-03-24 12:54:13 +03:00
if ( ntlmssp_state )
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
ntlmssp_end ( & ntlmssp_state ) ;
2003-12-24 12:56:51 +03:00
} else if ( strncmp ( buf , " KK " , 2 ) = = 0 ) {
2003-03-24 12:54:13 +03:00
} else {
DEBUG ( 1 , ( " NTLMSSP query [%s] invalid " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
2003-01-16 06:29:54 +03:00
if ( ! ntlmssp_state ) {
2003-12-30 10:33:58 +03:00
if ( ! NT_STATUS_IS_OK ( nt_status = ntlm_auth_start_ntlmssp_server ( & ntlmssp_state ) ) ) {
x_fprintf ( x_stdout , " BH %s \n " , nt_errstr ( nt_status ) ) ;
return ;
}
2003-01-16 06:29:54 +03:00
}
2003-03-24 12:54:13 +03:00
DEBUG ( 10 , ( " got NTLMSSP packet: \n " ) ) ;
2003-08-15 08:42:05 +04:00
dump_data ( 10 , ( const char * ) request . data , request . length ) ;
2003-01-16 06:29:54 +03:00
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
nt_status = ntlmssp_update ( ntlmssp_state , request , & reply ) ;
2003-01-16 06:29:54 +03:00
if ( NT_STATUS_EQUAL ( nt_status , NT_STATUS_MORE_PROCESSING_REQUIRED ) ) {
char * reply_base64 = base64_encode_data_blob ( reply ) ;
x_fprintf ( x_stdout , " TT %s \n " , reply_base64 ) ;
SAFE_FREE ( reply_base64 ) ;
data_blob_free ( & reply ) ;
2003-03-24 12:54:13 +03:00
DEBUG ( 10 , ( " NTLMSSP challenge \n " ) ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
} else if ( NT_STATUS_EQUAL ( nt_status , NT_STATUS_ACCESS_DENIED ) ) {
x_fprintf ( x_stdout , " BH %s \n " , nt_errstr ( nt_status ) ) ;
DEBUG ( 0 , ( " NTLMSSP BH: %s \n " , nt_errstr ( nt_status ) ) ) ;
2003-12-30 10:33:58 +03:00
ntlmssp_end ( & ntlmssp_state ) ;
2003-01-16 06:29:54 +03:00
} else if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
x_fprintf ( x_stdout , " NA %s \n " , nt_errstr ( nt_status ) ) ;
2003-03-24 12:54:13 +03:00
DEBUG ( 10 , ( " NTLMSSP %s \n " , nt_errstr ( nt_status ) ) ) ;
2003-01-16 06:29:54 +03:00
} else {
2003-12-30 16:20:39 +03:00
x_fprintf ( x_stdout , " AF %s \n " , ( char * ) ntlmssp_state - > auth_context ) ;
2003-03-24 12:54:13 +03:00
DEBUG ( 10 , ( " NTLMSSP OK! \n " ) ) ;
2003-01-16 06:29:54 +03:00
}
data_blob_free ( & request ) ;
}
2003-12-30 10:33:58 +03:00
static void manage_client_ntlmssp_request ( enum stdio_helper_mode stdio_helper_mode ,
char * buf , int length )
{
static NTLMSSP_STATE * ntlmssp_state = NULL ;
DATA_BLOB request , reply ;
NTSTATUS nt_status ;
BOOL first = False ;
if ( strlen ( buf ) < 2 ) {
DEBUG ( 1 , ( " NTLMSSP query [%s] invalid " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
if ( strlen ( buf ) > 3 ) {
request = base64_decode_data_blob ( buf + 3 ) ;
} else {
request = data_blob ( NULL , 0 ) ;
}
if ( strncmp ( buf , " PW " , 3 ) = = 0 ) {
/* We asked for a password and obviously got it :-) */
2004-12-07 21:25:53 +03:00
opt_password = SMB_STRNDUP ( ( const char * ) request . data , request . length ) ;
2003-12-30 10:33:58 +03:00
if ( opt_password = = NULL ) {
DEBUG ( 1 , ( " Out of memory \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
data_blob_free ( & request ) ;
return ;
}
x_fprintf ( x_stdout , " OK \n " ) ;
data_blob_free ( & request ) ;
return ;
}
if ( opt_password = = NULL ) {
/* Request a password from the calling process. After
sending it , the calling process should retry asking for the negotiate . */
DEBUG ( 10 , ( " Requesting password \n " ) ) ;
x_fprintf ( x_stdout , " PW \n " ) ;
return ;
}
if ( strncmp ( buf , " YR " , 2 ) = = 0 ) {
if ( ntlmssp_state )
ntlmssp_end ( & ntlmssp_state ) ;
} else if ( strncmp ( buf , " TT " , 2 ) = = 0 ) {
} else {
DEBUG ( 1 , ( " NTLMSSP query [%s] invalid " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
if ( ! ntlmssp_state ) {
if ( ! NT_STATUS_IS_OK ( nt_status = ntlm_auth_start_ntlmssp_client ( & ntlmssp_state ) ) ) {
x_fprintf ( x_stdout , " BH %s \n " , nt_errstr ( nt_status ) ) ;
return ;
}
first = True ;
}
DEBUG ( 10 , ( " got NTLMSSP packet: \n " ) ) ;
dump_data ( 10 , ( const char * ) request . data , request . length ) ;
nt_status = ntlmssp_update ( ntlmssp_state , request , & reply ) ;
if ( NT_STATUS_EQUAL ( nt_status , NT_STATUS_MORE_PROCESSING_REQUIRED ) ) {
char * reply_base64 = base64_encode_data_blob ( reply ) ;
if ( first ) {
x_fprintf ( x_stdout , " YR %s \n " , reply_base64 ) ;
} else {
x_fprintf ( x_stdout , " KK %s \n " , reply_base64 ) ;
}
SAFE_FREE ( reply_base64 ) ;
data_blob_free ( & reply ) ;
DEBUG ( 10 , ( " NTLMSSP challenge \n " ) ) ;
} else if ( NT_STATUS_IS_OK ( nt_status ) ) {
2005-09-30 21:13:37 +04:00
char * reply_base64 = base64_encode_data_blob ( reply ) ;
x_fprintf ( x_stdout , " AF %s \n " , reply_base64 ) ;
2006-03-08 11:22:42 +03:00
SAFE_FREE ( reply_base64 ) ;
2003-12-30 10:33:58 +03:00
DEBUG ( 10 , ( " NTLMSSP OK! \n " ) ) ;
if ( ntlmssp_state )
ntlmssp_end ( & ntlmssp_state ) ;
} else {
x_fprintf ( x_stdout , " BH %s \n " , nt_errstr ( nt_status ) ) ;
DEBUG ( 0 , ( " NTLMSSP BH: %s \n " , nt_errstr ( nt_status ) ) ) ;
if ( ntlmssp_state )
ntlmssp_end ( & ntlmssp_state ) ;
}
data_blob_free ( & request ) ;
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static void manage_squid_basic_request ( enum stdio_helper_mode stdio_helper_mode ,
2003-01-16 06:29:54 +03:00
char * buf , int length )
{
char * user , * pass ;
user = buf ;
pass = memchr ( buf , ' ' , length ) ;
if ( ! pass ) {
DEBUG ( 2 , ( " Password not found. Denying access \n " ) ) ;
2004-04-13 00:46:13 +04:00
x_fprintf ( x_stdout , " ERR \n " ) ;
2003-01-16 06:29:54 +03:00
return ;
}
* pass = ' \0 ' ;
pass + + ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
if ( stdio_helper_mode = = SQUID_2_5_BASIC ) {
2003-01-16 06:29:54 +03:00
rfc1738_unescape ( user ) ;
rfc1738_unescape ( pass ) ;
}
if ( check_plaintext_auth ( user , pass , False ) ) {
x_fprintf ( x_stdout , " OK \n " ) ;
} else {
x_fprintf ( x_stdout , " ERR \n " ) ;
}
}
2003-07-29 19:00:38 +04:00
static void offer_gss_spnego_mechs ( void ) {
DATA_BLOB token ;
SPNEGO_DATA spnego ;
ssize_t len ;
char * reply_base64 ;
2003-08-15 06:57:59 +04:00
pstring principal ;
pstring myname_lower ;
2003-07-29 19:00:38 +04:00
ZERO_STRUCT ( spnego ) ;
2003-07-31 14:24:10 +04:00
2003-08-15 06:57:59 +04:00
pstrcpy ( myname_lower , global_myname ( ) ) ;
strlower_m ( myname_lower ) ;
pstr_sprintf ( principal , " %s$@%s " , myname_lower , lp_realm ( ) ) ;
2003-07-31 14:24:10 +04:00
/* Server negTokenInit (mech offerings) */
2003-07-29 19:00:38 +04:00
spnego . type = SPNEGO_NEG_TOKEN_INIT ;
2005-10-18 07:24:00 +04:00
spnego . negTokenInit . mechTypes = SMB_XMALLOC_ARRAY ( const char * , 2 ) ;
2003-08-15 06:57:59 +04:00
# ifdef HAVE_KRB5
spnego . negTokenInit . mechTypes [ 0 ] = smb_xstrdup ( OID_KERBEROS5_OLD ) ;
spnego . negTokenInit . mechTypes [ 1 ] = smb_xstrdup ( OID_NTLMSSP ) ;
spnego . negTokenInit . mechTypes [ 2 ] = NULL ;
# else
2003-07-31 14:24:10 +04:00
spnego . negTokenInit . mechTypes [ 0 ] = smb_xstrdup ( OID_NTLMSSP ) ;
spnego . negTokenInit . mechTypes [ 1 ] = NULL ;
2003-08-15 06:57:59 +04:00
# endif
2003-07-29 19:00:38 +04:00
2003-08-15 06:57:59 +04:00
spnego . negTokenInit . mechListMIC = data_blob ( principal ,
strlen ( principal ) ) ;
2003-07-29 19:00:38 +04:00
len = write_spnego_data ( & token , & spnego ) ;
2003-07-31 14:24:10 +04:00
free_spnego_data ( & spnego ) ;
2003-07-29 19:00:38 +04:00
if ( len = = - 1 ) {
DEBUG ( 1 , ( " Could not write SPNEGO data blob \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
reply_base64 = base64_encode_data_blob ( token ) ;
2003-08-01 11:59:23 +04:00
x_fprintf ( x_stdout , " TT %s * \n " , reply_base64 ) ;
2003-07-29 19:00:38 +04:00
SAFE_FREE ( reply_base64 ) ;
data_blob_free ( & token ) ;
DEBUG ( 10 , ( " sent SPNEGO negTokenInit \n " ) ) ;
return ;
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static void manage_gss_spnego_request ( enum stdio_helper_mode stdio_helper_mode ,
2003-07-29 19:00:38 +04:00
char * buf , int length )
{
static NTLMSSP_STATE * ntlmssp_state = NULL ;
2003-08-12 05:54:26 +04:00
SPNEGO_DATA request , response ;
DATA_BLOB token ;
2003-07-29 19:00:38 +04:00
NTSTATUS status ;
ssize_t len ;
2003-08-12 23:00:08 +04:00
char * user = NULL ;
char * domain = NULL ;
2003-08-01 11:59:23 +04:00
const char * reply_code ;
char * reply_base64 ;
pstring reply_argument ;
2003-07-29 19:00:38 +04:00
if ( strlen ( buf ) < 2 ) {
2004-06-14 07:22:52 +04:00
DEBUG ( 1 , ( " SPENGO query [%s] invalid " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
2003-07-29 19:00:38 +04:00
2004-06-14 07:22:52 +04:00
if ( strncmp ( buf , " YR " , 2 ) = = 0 ) {
if ( ntlmssp_state )
ntlmssp_end ( & ntlmssp_state ) ;
} else if ( strncmp ( buf , " KK " , 2 ) = = 0 ) {
} else {
DEBUG ( 1 , ( " SPENGO query [%s] invalid " , buf ) ) ;
2003-07-29 19:00:38 +04:00
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
2004-06-14 07:22:52 +04:00
if ( ( strlen ( buf ) = = 2 ) ) {
2003-07-29 19:00:38 +04:00
2004-06-14 07:22:52 +04:00
/* no client data, get the negTokenInit offering
2003-07-29 19:00:38 +04:00
mechanisms */
offer_gss_spnego_mechs ( ) ;
return ;
}
2004-06-14 07:22:52 +04:00
/* All subsequent requests have a blob. This might be negTokenInit or negTokenTarg */
2003-07-29 19:00:38 +04:00
2004-06-14 07:22:52 +04:00
if ( strlen ( buf ) < = 3 ) {
2003-07-29 19:00:38 +04:00
DEBUG ( 1 , ( " GSS-SPNEGO query [%s] invalid \n " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
2003-08-12 05:54:26 +04:00
token = base64_decode_data_blob ( buf + 3 ) ;
len = read_spnego_data ( token , & request ) ;
data_blob_free ( & token ) ;
2003-07-29 19:00:38 +04:00
if ( len = = - 1 ) {
DEBUG ( 1 , ( " GSS-SPNEGO query [%s] invalid " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
2003-08-12 05:54:26 +04:00
if ( request . type = = SPNEGO_NEG_TOKEN_INIT ) {
2003-07-29 19:00:38 +04:00
/* Second request from Client. This is where the
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
client offers its mechanism to use . */
2003-07-29 19:00:38 +04:00
2003-08-12 05:54:26 +04:00
if ( ( request . negTokenInit . mechTypes = = NULL ) | |
( request . negTokenInit . mechTypes [ 0 ] = = NULL ) ) {
2003-07-29 19:00:38 +04:00
DEBUG ( 1 , ( " Client did not offer any mechanism " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
2005-09-30 21:13:37 +04:00
status = NT_STATUS_UNSUCCESSFUL ;
2003-08-12 23:00:08 +04:00
if ( strcmp ( request . negTokenInit . mechTypes [ 0 ] , OID_NTLMSSP ) = = 0 ) {
if ( request . negTokenInit . mechToken . data = = NULL ) {
DEBUG ( 1 , ( " Client did not provide NTLMSSP data \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
if ( ntlmssp_state ! = NULL ) {
DEBUG ( 1 , ( " Client wants a new NTLMSSP challenge, but "
" already got one \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
ntlmssp_end ( & ntlmssp_state ) ;
2003-08-12 23:00:08 +04:00
return ;
}
2003-12-30 10:33:58 +03:00
if ( ! NT_STATUS_IS_OK ( status = ntlm_auth_start_ntlmssp_server ( & ntlmssp_state ) ) ) {
x_fprintf ( x_stdout , " BH %s \n " , nt_errstr ( status ) ) ;
return ;
}
2003-08-12 23:00:08 +04:00
DEBUG ( 10 , ( " got NTLMSSP packet: \n " ) ) ;
2003-08-15 08:42:05 +04:00
dump_data ( 10 , ( const char * ) request . negTokenInit . mechToken . data ,
2003-08-12 23:00:08 +04:00
request . negTokenInit . mechToken . length ) ;
2003-08-13 00:50:56 +04:00
response . type = SPNEGO_NEG_TOKEN_TARG ;
2004-12-07 21:25:53 +03:00
response . negTokenTarg . supportedMech = SMB_STRDUP ( OID_NTLMSSP ) ;
2003-08-13 00:50:56 +04:00
response . negTokenTarg . mechListMIC = data_blob ( NULL , 0 ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
status = ntlmssp_update ( ntlmssp_state ,
2003-08-12 23:00:08 +04:00
request . negTokenInit . mechToken ,
& response . negTokenTarg . responseToken ) ;
2003-07-29 19:00:38 +04:00
}
2003-08-15 06:57:59 +04:00
# ifdef HAVE_KRB5
if ( strcmp ( request . negTokenInit . mechTypes [ 0 ] , OID_KERBEROS5_OLD ) = = 0 ) {
2005-10-20 21:33:17 +04:00
TALLOC_CTX * mem_ctx = talloc_init ( " manage_gss_spnego_request " ) ;
2003-08-15 06:57:59 +04:00
char * principal ;
DATA_BLOB ap_rep ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
DATA_BLOB session_key ;
2003-08-15 06:57:59 +04:00
if ( request . negTokenInit . mechToken . data = = NULL ) {
DEBUG ( 1 , ( " Client did not provide Kerberos data \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
response . type = SPNEGO_NEG_TOKEN_TARG ;
2004-12-07 21:25:53 +03:00
response . negTokenTarg . supportedMech = SMB_STRDUP ( OID_KERBEROS5_OLD ) ;
2003-08-15 06:57:59 +04:00
response . negTokenTarg . mechListMIC = data_blob ( NULL , 0 ) ;
response . negTokenTarg . responseToken = data_blob ( NULL , 0 ) ;
2005-09-30 21:13:37 +04:00
status = ads_verify_ticket ( mem_ctx , lp_realm ( ) ,
2003-08-15 06:57:59 +04:00
& request . negTokenInit . mechToken ,
2005-09-30 21:13:37 +04:00
& principal , NULL , & ap_rep ,
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
& session_key ) ;
2003-08-15 06:57:59 +04:00
2005-09-30 21:13:37 +04:00
talloc_destroy ( mem_ctx ) ;
2003-08-15 06:57:59 +04:00
/* Now in "principal" we have the name we are
authenticated as . */
if ( NT_STATUS_IS_OK ( status ) ) {
2004-10-27 04:41:41 +04:00
domain = strchr_m ( principal , ' @ ' ) ;
2003-08-15 06:57:59 +04:00
if ( domain = = NULL ) {
DEBUG ( 1 , ( " Did not get a valid principal "
" from ads_verify_ticket \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
* domain + + = ' \0 ' ;
2004-12-07 21:25:53 +03:00
domain = SMB_STRDUP ( domain ) ;
user = SMB_STRDUP ( principal ) ;
2003-08-15 06:57:59 +04:00
data_blob_free ( & ap_rep ) ;
SAFE_FREE ( principal ) ;
}
}
# endif
2003-07-29 19:00:38 +04:00
} else {
2003-08-13 00:50:56 +04:00
if ( ( request . negTokenTarg . supportedMech = = NULL ) | |
( strcmp ( request . negTokenTarg . supportedMech , OID_NTLMSSP ) ! = 0 ) ) {
/* Kerberos should never send a negTokenTarg, OID_NTLMSSP
is the only one we support that sends this stuff */
DEBUG ( 1 , ( " Got a negTokenTarg for something non-NTLMSSP: %s \n " ,
request . negTokenTarg . supportedMech ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
2003-07-29 19:00:38 +04:00
2003-08-12 05:54:26 +04:00
if ( request . negTokenTarg . responseToken . data = = NULL ) {
2003-08-13 00:50:56 +04:00
DEBUG ( 1 , ( " Got a negTokenTarg without a responseToken! \n " ) ) ;
2003-07-29 19:00:38 +04:00
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
status = ntlmssp_update ( ntlmssp_state ,
2003-08-12 05:54:26 +04:00
request . negTokenTarg . responseToken ,
& response . negTokenTarg . responseToken ) ;
2003-08-12 23:00:08 +04:00
2003-08-13 00:50:56 +04:00
response . type = SPNEGO_NEG_TOKEN_TARG ;
2004-12-07 21:25:53 +03:00
response . negTokenTarg . supportedMech = SMB_STRDUP ( OID_NTLMSSP ) ;
2003-08-13 00:50:56 +04:00
response . negTokenTarg . mechListMIC = data_blob ( NULL , 0 ) ;
2003-08-12 23:00:08 +04:00
if ( NT_STATUS_IS_OK ( status ) ) {
2004-12-07 21:25:53 +03:00
user = SMB_STRDUP ( ntlmssp_state - > user ) ;
domain = SMB_STRDUP ( ntlmssp_state - > domain ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
ntlmssp_end ( & ntlmssp_state ) ;
2003-08-12 23:00:08 +04:00
}
2003-08-12 05:54:26 +04:00
}
2003-07-29 19:00:38 +04:00
2003-08-12 05:54:26 +04:00
free_spnego_data ( & request ) ;
2003-07-29 19:00:38 +04:00
if ( NT_STATUS_IS_OK ( status ) ) {
2003-08-12 05:54:26 +04:00
response . negTokenTarg . negResult = SPNEGO_ACCEPT_COMPLETED ;
2003-08-01 11:59:23 +04:00
reply_code = " AF " ;
2003-08-12 23:00:08 +04:00
pstr_sprintf ( reply_argument , " %s \\ %s " , domain , user ) ;
2003-08-01 11:59:23 +04:00
} else if ( NT_STATUS_EQUAL ( status ,
NT_STATUS_MORE_PROCESSING_REQUIRED ) ) {
2003-08-12 05:54:26 +04:00
response . negTokenTarg . negResult = SPNEGO_ACCEPT_INCOMPLETE ;
2003-08-01 11:59:23 +04:00
reply_code = " TT " ;
pstr_sprintf ( reply_argument , " * " ) ;
} else {
2003-08-12 05:54:26 +04:00
response . negTokenTarg . negResult = SPNEGO_REJECT ;
2003-08-01 11:59:23 +04:00
reply_code = " NA " ;
pstrcpy ( reply_argument , nt_errstr ( status ) ) ;
2003-07-29 19:00:38 +04:00
}
2003-08-12 23:00:08 +04:00
SAFE_FREE ( user ) ;
SAFE_FREE ( domain ) ;
2003-08-12 05:54:26 +04:00
len = write_spnego_data ( & token , & response ) ;
free_spnego_data ( & response ) ;
2003-07-29 19:00:38 +04:00
if ( len = = - 1 ) {
DEBUG ( 1 , ( " Could not write SPNEGO data blob \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
reply_base64 = base64_encode_data_blob ( token ) ;
2003-08-01 11:59:23 +04:00
x_fprintf ( x_stdout , " %s %s %s \n " ,
reply_code , reply_base64 , reply_argument ) ;
2003-07-29 19:00:38 +04:00
SAFE_FREE ( reply_base64 ) ;
data_blob_free ( & token ) ;
return ;
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static NTLMSSP_STATE * client_ntlmssp_state = NULL ;
2003-08-01 11:59:23 +04:00
2003-08-04 17:10:43 +04:00
static BOOL manage_client_ntlmssp_init ( SPNEGO_DATA spnego )
2003-08-01 11:59:23 +04:00
{
NTSTATUS status ;
DATA_BLOB null_blob = data_blob ( NULL , 0 ) ;
DATA_BLOB to_server ;
char * to_server_base64 ;
const char * my_mechs [ ] = { OID_NTLMSSP , NULL } ;
DEBUG ( 10 , ( " Got spnego negTokenInit with NTLMSSP \n " ) ) ;
if ( client_ntlmssp_state ! = NULL ) {
DEBUG ( 1 , ( " Request for initial SPNEGO request where "
" we already have a state \n " ) ) ;
2003-08-04 17:10:43 +04:00
return False ;
2003-08-01 11:59:23 +04:00
}
2003-12-30 10:33:58 +03:00
if ( ! client_ntlmssp_state ) {
if ( ! NT_STATUS_IS_OK ( status = ntlm_auth_start_ntlmssp_client ( & client_ntlmssp_state ) ) ) {
x_fprintf ( x_stdout , " BH %s \n " , nt_errstr ( status ) ) ;
return False ;
}
2003-08-01 11:59:23 +04:00
}
2003-12-30 10:33:58 +03:00
2003-08-01 11:59:23 +04:00
if ( opt_password = = NULL ) {
/* Request a password from the calling process. After
sending it , the calling process should retry with
the negTokenInit . */
DEBUG ( 10 , ( " Requesting password \n " ) ) ;
x_fprintf ( x_stdout , " PW \n " ) ;
2003-08-04 17:10:43 +04:00
return True ;
2003-08-01 11:59:23 +04:00
}
spnego . type = SPNEGO_NEG_TOKEN_INIT ;
2005-10-18 07:24:00 +04:00
spnego . negTokenInit . mechTypes = my_mechs ;
2003-08-01 11:59:23 +04:00
spnego . negTokenInit . reqFlags = 0 ;
spnego . negTokenInit . mechListMIC = null_blob ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
status = ntlmssp_update ( client_ntlmssp_state , null_blob ,
2003-08-01 11:59:23 +04:00
& spnego . negTokenInit . mechToken ) ;
2005-09-30 21:13:37 +04:00
if ( ! ( NT_STATUS_EQUAL ( status , NT_STATUS_MORE_PROCESSING_REQUIRED ) | |
NT_STATUS_IS_OK ( status ) ) ) {
DEBUG ( 1 , ( " Expected OK or MORE_PROCESSING_REQUIRED, got: %s \n " ,
2003-08-01 11:59:23 +04:00
nt_errstr ( status ) ) ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
ntlmssp_end ( & client_ntlmssp_state ) ;
2003-08-04 17:10:43 +04:00
return False ;
2003-08-01 11:59:23 +04:00
}
write_spnego_data ( & to_server , & spnego ) ;
data_blob_free ( & spnego . negTokenInit . mechToken ) ;
to_server_base64 = base64_encode_data_blob ( to_server ) ;
data_blob_free ( & to_server ) ;
x_fprintf ( x_stdout , " KK %s \n " , to_server_base64 ) ;
SAFE_FREE ( to_server_base64 ) ;
2003-08-04 17:10:43 +04:00
return True ;
2003-08-01 11:59:23 +04:00
}
static void manage_client_ntlmssp_targ ( SPNEGO_DATA spnego )
{
NTSTATUS status ;
DATA_BLOB null_blob = data_blob ( NULL , 0 ) ;
DATA_BLOB request ;
DATA_BLOB to_server ;
char * to_server_base64 ;
DEBUG ( 10 , ( " Got spnego negTokenTarg with NTLMSSP \n " ) ) ;
if ( client_ntlmssp_state = = NULL ) {
DEBUG ( 1 , ( " Got NTLMSSP tArg without a client state \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
ntlmssp_end ( & client_ntlmssp_state ) ;
2003-08-01 11:59:23 +04:00
return ;
}
if ( spnego . negTokenTarg . negResult = = SPNEGO_REJECT ) {
x_fprintf ( x_stdout , " NA \n " ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
ntlmssp_end ( & client_ntlmssp_state ) ;
2003-08-01 11:59:23 +04:00
return ;
}
if ( spnego . negTokenTarg . negResult = = SPNEGO_ACCEPT_COMPLETED ) {
x_fprintf ( x_stdout , " AF \n " ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
ntlmssp_end ( & client_ntlmssp_state ) ;
2003-08-01 11:59:23 +04:00
return ;
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
status = ntlmssp_update ( client_ntlmssp_state ,
2003-08-01 11:59:23 +04:00
spnego . negTokenTarg . responseToken ,
& request ) ;
if ( ! NT_STATUS_EQUAL ( status , NT_STATUS_MORE_PROCESSING_REQUIRED ) ) {
DEBUG ( 1 , ( " Expected MORE_PROCESSING_REQUIRED from "
" ntlmssp_client_update, got: %s \n " ,
nt_errstr ( status ) ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
data_blob_free ( & request ) ;
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
ntlmssp_end ( & client_ntlmssp_state ) ;
2003-08-01 11:59:23 +04:00
return ;
}
spnego . type = SPNEGO_NEG_TOKEN_TARG ;
spnego . negTokenTarg . negResult = SPNEGO_ACCEPT_INCOMPLETE ;
2005-09-30 21:13:37 +04:00
spnego . negTokenTarg . supportedMech = ( char * ) OID_NTLMSSP ;
2003-08-01 11:59:23 +04:00
spnego . negTokenTarg . responseToken = request ;
spnego . negTokenTarg . mechListMIC = null_blob ;
write_spnego_data ( & to_server , & spnego ) ;
data_blob_free ( & request ) ;
to_server_base64 = base64_encode_data_blob ( to_server ) ;
data_blob_free ( & to_server ) ;
x_fprintf ( x_stdout , " KK %s \n " , to_server_base64 ) ;
SAFE_FREE ( to_server_base64 ) ;
return ;
}
2003-08-14 21:21:22 +04:00
# ifdef HAVE_KRB5
2003-08-04 17:10:43 +04:00
static BOOL manage_client_krb5_init ( SPNEGO_DATA spnego )
2003-08-01 11:59:23 +04:00
{
2003-08-13 00:50:56 +04:00
char * principal ;
DATA_BLOB tkt , to_server ;
2004-06-14 07:22:52 +04:00
DATA_BLOB session_key_krb5 = data_blob ( NULL , 0 ) ;
2003-08-13 00:50:56 +04:00
SPNEGO_DATA reply ;
char * reply_base64 ;
2004-01-08 11:19:18 +03:00
int retval ;
2003-08-13 00:50:56 +04:00
const char * my_mechs [ ] = { OID_KERBEROS5_OLD , NULL } ;
ssize_t len ;
if ( ( spnego . negTokenInit . mechListMIC . data = = NULL ) | |
( spnego . negTokenInit . mechListMIC . length = = 0 ) ) {
DEBUG ( 1 , ( " Did not get a principal for krb5 \n " ) ) ;
return False ;
}
2004-12-07 21:25:53 +03:00
principal = SMB_MALLOC ( spnego . negTokenInit . mechListMIC . length + 1 ) ;
2003-08-13 00:50:56 +04:00
if ( principal = = NULL ) {
DEBUG ( 1 , ( " Could not malloc principal \n " ) ) ;
return False ;
}
memcpy ( principal , spnego . negTokenInit . mechListMIC . data ,
spnego . negTokenInit . mechListMIC . length ) ;
principal [ spnego . negTokenInit . mechListMIC . length ] = ' \0 ' ;
2006-02-04 01:19:41 +03:00
retval = cli_krb5_get_ticket ( principal , 0 , & tkt , & session_key_krb5 , 0 , NULL ) ;
2003-08-13 00:50:56 +04:00
2004-01-08 11:19:18 +03:00
if ( retval ) {
2003-08-13 00:50:56 +04:00
pstring user ;
/* Let's try to first get the TGT, for that we need a
password . */
if ( opt_password = = NULL ) {
DEBUG ( 10 , ( " Requesting password \n " ) ) ;
x_fprintf ( x_stdout , " PW \n " ) ;
return True ;
}
pstr_sprintf ( user , " %s@%s " , opt_username , opt_domain ) ;
2006-03-20 22:05:44 +03:00
if ( ( retval = kerberos_kinit_password ( user , opt_password , 0 , NULL ) ) ) {
2004-01-08 11:19:18 +03:00
DEBUG ( 10 , ( " Requesting TGT failed: %s \n " , error_message ( retval ) ) ) ;
2004-06-14 07:22:52 +04:00
return False ;
2003-08-13 00:50:56 +04:00
}
2006-02-04 01:19:41 +03:00
retval = cli_krb5_get_ticket ( principal , 0 , & tkt , & session_key_krb5 , 0 , NULL ) ;
2004-01-08 11:19:18 +03:00
if ( retval ) {
DEBUG ( 10 , ( " Kinit suceeded, but getting a ticket failed: %s \n " , error_message ( retval ) ) ) ;
2004-06-14 07:22:52 +04:00
return False ;
2004-01-08 11:19:18 +03:00
}
2003-08-13 00:50:56 +04:00
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
data_blob_free ( & session_key_krb5 ) ;
2003-08-13 00:50:56 +04:00
ZERO_STRUCT ( reply ) ;
reply . type = SPNEGO_NEG_TOKEN_INIT ;
reply . negTokenInit . mechTypes = my_mechs ;
reply . negTokenInit . reqFlags = 0 ;
reply . negTokenInit . mechToken = tkt ;
reply . negTokenInit . mechListMIC = data_blob ( NULL , 0 ) ;
len = write_spnego_data ( & to_server , & reply ) ;
data_blob_free ( & tkt ) ;
if ( len = = - 1 ) {
DEBUG ( 1 , ( " Could not write SPNEGO data blob \n " ) ) ;
return False ;
}
reply_base64 = base64_encode_data_blob ( to_server ) ;
x_fprintf ( x_stdout , " KK %s * \n " , reply_base64 ) ;
SAFE_FREE ( reply_base64 ) ;
data_blob_free ( & to_server ) ;
DEBUG ( 10 , ( " sent GSS-SPNEGO KERBEROS5 negTokenInit \n " ) ) ;
return True ;
2003-08-01 11:59:23 +04:00
}
static void manage_client_krb5_targ ( SPNEGO_DATA spnego )
{
2003-08-13 00:50:56 +04:00
switch ( spnego . negTokenTarg . negResult ) {
case SPNEGO_ACCEPT_INCOMPLETE :
DEBUG ( 1 , ( " Got a Kerberos negTokenTarg with ACCEPT_INCOMPLETE \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
break ;
case SPNEGO_ACCEPT_COMPLETED :
DEBUG ( 10 , ( " Accept completed \n " ) ) ;
x_fprintf ( x_stdout , " AF \n " ) ;
break ;
case SPNEGO_REJECT :
DEBUG ( 10 , ( " Rejected \n " ) ) ;
x_fprintf ( x_stdout , " NA \n " ) ;
break ;
default :
DEBUG ( 1 , ( " Got an invalid negTokenTarg \n " ) ) ;
x_fprintf ( x_stdout , " AF \n " ) ;
}
2003-08-01 11:59:23 +04:00
}
2003-08-14 21:21:22 +04:00
# endif
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static void manage_gss_spnego_client_request ( enum stdio_helper_mode stdio_helper_mode ,
2003-08-01 11:59:23 +04:00
char * buf , int length )
{
DATA_BLOB request ;
SPNEGO_DATA spnego ;
ssize_t len ;
if ( strlen ( buf ) < = 3 ) {
DEBUG ( 1 , ( " SPNEGO query [%s] too short \n " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
request = base64_decode_data_blob ( buf + 3 ) ;
if ( strncmp ( buf , " PW " , 3 ) = = 0 ) {
/* We asked for a password and obviously got it :-) */
2004-12-07 21:25:53 +03:00
opt_password = SMB_STRNDUP ( ( const char * ) request . data , request . length ) ;
2004-04-13 00:46:13 +04:00
2003-08-01 11:59:23 +04:00
if ( opt_password = = NULL ) {
DEBUG ( 1 , ( " Out of memory \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
data_blob_free ( & request ) ;
return ;
}
x_fprintf ( x_stdout , " OK \n " ) ;
data_blob_free ( & request ) ;
return ;
}
if ( ( strncmp ( buf , " TT " , 3 ) ! = 0 ) & &
( strncmp ( buf , " AF " , 3 ) ! = 0 ) & &
( strncmp ( buf , " NA " , 3 ) ! = 0 ) ) {
DEBUG ( 1 , ( " SPNEGO request [%s] invalid \n " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
data_blob_free ( & request ) ;
return ;
}
/* So we got a server challenge to generate a SPNEGO
client - to - server request . . . */
len = read_spnego_data ( request , & spnego ) ;
data_blob_free ( & request ) ;
if ( len = = - 1 ) {
DEBUG ( 1 , ( " Could not read SPNEGO data for [%s] \n " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
if ( spnego . type = = SPNEGO_NEG_TOKEN_INIT ) {
/* The server offers a list of mechanisms */
2005-09-30 21:13:37 +04:00
const char * * mechType = ( const char * * ) spnego . negTokenInit . mechTypes ;
2003-08-01 11:59:23 +04:00
while ( * mechType ! = NULL ) {
2003-08-14 21:21:22 +04:00
# ifdef HAVE_KRB5
2003-08-04 17:10:43 +04:00
if ( ( strcmp ( * mechType , OID_KERBEROS5_OLD ) = = 0 ) | |
( strcmp ( * mechType , OID_KERBEROS5 ) = = 0 ) ) {
if ( manage_client_krb5_init ( spnego ) )
goto out ;
2003-08-01 11:59:23 +04:00
}
2003-08-14 21:21:22 +04:00
# endif
2003-08-01 11:59:23 +04:00
2003-08-04 17:10:43 +04:00
if ( strcmp ( * mechType , OID_NTLMSSP ) = = 0 ) {
if ( manage_client_ntlmssp_init ( spnego ) )
goto out ;
2003-08-01 11:59:23 +04:00
}
mechType + + ;
}
DEBUG ( 1 , ( " Server offered no compatible mechanism \n " ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
}
if ( spnego . type = = SPNEGO_NEG_TOKEN_TARG ) {
2003-08-04 17:10:43 +04:00
if ( spnego . negTokenTarg . supportedMech = = NULL ) {
/* On accept/reject Windows does not send the
mechanism anymore . Handle that here and
shut down the mechanisms . */
switch ( spnego . negTokenTarg . negResult ) {
case SPNEGO_ACCEPT_COMPLETED :
x_fprintf ( x_stdout , " AF \n " ) ;
break ;
case SPNEGO_REJECT :
x_fprintf ( x_stdout , " NA \n " ) ;
break ;
default :
DEBUG ( 1 , ( " Got a negTokenTarg with no mech and an "
" unknown negResult: %d \n " ,
spnego . negTokenTarg . negResult ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
ntlmssp_end ( & client_ntlmssp_state ) ;
2003-08-04 17:10:43 +04:00
goto out ;
}
2003-08-01 11:59:23 +04:00
if ( strcmp ( spnego . negTokenTarg . supportedMech ,
OID_NTLMSSP ) = = 0 ) {
manage_client_ntlmssp_targ ( spnego ) ;
goto out ;
}
2003-08-14 21:21:22 +04:00
# if HAVE_KRB5
2003-08-01 11:59:23 +04:00
if ( strcmp ( spnego . negTokenTarg . supportedMech ,
OID_KERBEROS5_OLD ) = = 0 ) {
manage_client_krb5_targ ( spnego ) ;
goto out ;
}
2003-08-14 21:21:22 +04:00
# endif
2003-08-01 11:59:23 +04:00
}
DEBUG ( 1 , ( " Got an SPNEGO token I could not handle [%s]! \n " , buf ) ) ;
x_fprintf ( x_stdout , " BH \n " ) ;
return ;
out :
free_spnego_data ( & spnego ) ;
return ;
}
2004-04-13 00:46:13 +04:00
static void manage_ntlm_server_1_request ( enum stdio_helper_mode stdio_helper_mode ,
char * buf , int length )
{
char * request , * parameter ;
static DATA_BLOB challenge ;
static DATA_BLOB lm_response ;
static DATA_BLOB nt_response ;
static char * full_username ;
static char * username ;
static char * domain ;
static char * plaintext_password ;
static BOOL ntlm_server_1_user_session_key ;
static BOOL ntlm_server_1_lm_session_key ;
if ( strequal ( buf , " . " ) ) {
if ( ! full_username & & ! username ) {
x_fprintf ( x_stdout , " Error: No username supplied! \n " ) ;
} else if ( plaintext_password ) {
/* handle this request as plaintext */
if ( ! full_username ) {
if ( asprintf ( & full_username , " %s%c%s " , domain , winbind_separator ( ) , username ) = = - 1 ) {
x_fprintf ( x_stdout , " Error: Out of memory in asprintf! \n . \n " ) ;
return ;
}
}
if ( check_plaintext_auth ( full_username , plaintext_password , False ) ) {
x_fprintf ( x_stdout , " Authenticated: Yes \n " ) ;
} else {
x_fprintf ( x_stdout , " Authenticated: No \n " ) ;
}
2004-04-14 00:57:26 +04:00
} else if ( ! lm_response . data & & ! nt_response . data ) {
2004-04-13 00:46:13 +04:00
x_fprintf ( x_stdout , " Error: No password supplied! \n " ) ;
} else if ( ! challenge . data ) {
x_fprintf ( x_stdout , " Error: No lanman-challenge supplied! \n " ) ;
} else {
char * error_string = NULL ;
uchar lm_key [ 8 ] ;
uchar user_session_key [ 16 ] ;
uint32 flags = 0 ;
if ( full_username & & ! username ) {
fstring fstr_user ;
fstring fstr_domain ;
if ( ! parse_ntlm_auth_domain_user ( full_username , fstr_user , fstr_domain ) ) {
/* username might be 'tainted', don't print into our new-line deleimianted stream */
x_fprintf ( x_stdout , " Error: Could not parse into domain and username \n " ) ;
}
SAFE_FREE ( username ) ;
SAFE_FREE ( domain ) ;
username = smb_xstrdup ( fstr_user ) ;
domain = smb_xstrdup ( fstr_domain ) ;
}
if ( ! domain ) {
domain = smb_xstrdup ( get_winbind_domain ( ) ) ;
}
if ( ntlm_server_1_lm_session_key )
flags | = WBFLAG_PAM_LMKEY ;
if ( ntlm_server_1_user_session_key )
flags | = WBFLAG_PAM_USER_SESSION_KEY ;
if ( ! NT_STATUS_IS_OK (
contact_winbind_auth_crap ( username ,
domain ,
global_myname ( ) ,
& challenge ,
& lm_response ,
& nt_response ,
flags ,
lm_key ,
user_session_key ,
& error_string ,
NULL ) ) ) {
x_fprintf ( x_stdout , " Authenticated: No \n " ) ;
x_fprintf ( x_stdout , " Authentication-Error: %s \n . \n " , error_string ) ;
SAFE_FREE ( error_string ) ;
} else {
static char zeros [ 16 ] ;
char * hex_lm_key ;
char * hex_user_session_key ;
x_fprintf ( x_stdout , " Authenticated: Yes \n " ) ;
if ( ntlm_server_1_lm_session_key
& & ( memcmp ( zeros , lm_key ,
sizeof ( lm_key ) ) ! = 0 ) ) {
2005-08-08 00:59:28 +04:00
hex_lm_key = hex_encode ( NULL ,
( const unsigned char * ) lm_key ,
sizeof ( lm_key ) ) ;
2004-04-13 00:46:13 +04:00
x_fprintf ( x_stdout , " LANMAN-Session-Key: %s \n " , hex_lm_key ) ;
2006-02-20 20:59:58 +03:00
TALLOC_FREE ( hex_lm_key ) ;
2004-04-13 00:46:13 +04:00
}
if ( ntlm_server_1_user_session_key
& & ( memcmp ( zeros , user_session_key ,
sizeof ( user_session_key ) ) ! = 0 ) ) {
2005-08-08 00:59:28 +04:00
hex_user_session_key = hex_encode ( NULL ,
( const unsigned char * ) user_session_key ,
sizeof ( user_session_key ) ) ;
2004-04-13 00:46:13 +04:00
x_fprintf ( x_stdout , " User-Session-Key: %s \n " , hex_user_session_key ) ;
2006-02-20 20:59:58 +03:00
TALLOC_FREE ( hex_user_session_key ) ;
2004-04-13 00:46:13 +04:00
}
}
}
/* clear out the state */
challenge = data_blob ( NULL , 0 ) ;
nt_response = data_blob ( NULL , 0 ) ;
lm_response = data_blob ( NULL , 0 ) ;
SAFE_FREE ( full_username ) ;
SAFE_FREE ( username ) ;
SAFE_FREE ( domain ) ;
SAFE_FREE ( plaintext_password ) ;
ntlm_server_1_user_session_key = False ;
ntlm_server_1_lm_session_key = False ;
x_fprintf ( x_stdout , " . \n " ) ;
return ;
}
request = buf ;
2004-04-14 00:57:26 +04:00
/* Indicates a base64 encoded structure */
parameter = strstr_m ( request , " :: " ) ;
2004-04-13 00:46:13 +04:00
if ( ! parameter ) {
2004-04-14 00:57:26 +04:00
parameter = strstr_m ( request , " : " ) ;
2004-04-13 00:46:13 +04:00
if ( ! parameter ) {
DEBUG ( 0 , ( " Parameter not found! \n " ) ) ;
x_fprintf ( x_stdout , " Error: Parameter not found! \n . \n " ) ;
return ;
}
2004-04-14 00:57:26 +04:00
2004-04-13 00:46:13 +04:00
parameter [ 0 ] = ' \0 ' ;
parameter + + ;
parameter [ 0 ] = ' \0 ' ;
parameter + + ;
} else {
parameter [ 0 ] = ' \0 ' ;
parameter + + ;
parameter [ 0 ] = ' \0 ' ;
parameter + + ;
2004-04-14 00:57:26 +04:00
parameter [ 0 ] = ' \0 ' ;
parameter + + ;
base64_decode_inplace ( parameter ) ;
2004-04-13 00:46:13 +04:00
}
if ( strequal ( request , " LANMAN-Challenge " ) ) {
2005-08-08 00:59:28 +04:00
challenge = strhex_to_data_blob ( NULL , parameter ) ;
2004-04-13 00:46:13 +04:00
if ( challenge . length ! = 8 ) {
x_fprintf ( x_stdout , " Error: hex decode of %s failed! (got %d bytes, expected 8) \n . \n " ,
parameter ,
( int ) challenge . length ) ;
challenge = data_blob ( NULL , 0 ) ;
}
} else if ( strequal ( request , " NT-Response " ) ) {
2005-08-08 00:59:28 +04:00
nt_response = strhex_to_data_blob ( NULL , parameter ) ;
2004-04-13 00:46:13 +04:00
if ( nt_response . length < 24 ) {
x_fprintf ( x_stdout , " Error: hex decode of %s failed! (only got %d bytes, needed at least 24) \n . \n " ,
parameter ,
2004-07-25 17:11:39 +04:00
( int ) nt_response . length ) ;
2004-04-13 00:46:13 +04:00
nt_response = data_blob ( NULL , 0 ) ;
}
} else if ( strequal ( request , " LANMAN-Response " ) ) {
2005-08-08 00:59:28 +04:00
lm_response = strhex_to_data_blob ( NULL , parameter ) ;
2004-04-13 00:46:13 +04:00
if ( lm_response . length ! = 24 ) {
x_fprintf ( x_stdout , " Error: hex decode of %s failed! (got %d bytes, expected 24) \n . \n " ,
parameter ,
( int ) lm_response . length ) ;
lm_response = data_blob ( NULL , 0 ) ;
}
} else if ( strequal ( request , " Password " ) ) {
plaintext_password = smb_xstrdup ( parameter ) ;
} else if ( strequal ( request , " NT-Domain " ) ) {
domain = smb_xstrdup ( parameter ) ;
} else if ( strequal ( request , " Username " ) ) {
username = smb_xstrdup ( parameter ) ;
} else if ( strequal ( request , " Full-Username " ) ) {
full_username = smb_xstrdup ( parameter ) ;
} else if ( strequal ( request , " Request-User-Session-Key " ) ) {
ntlm_server_1_user_session_key = strequal ( parameter , " Yes " ) ;
} else if ( strequal ( request , " Request-LanMan-Session-Key " ) ) {
ntlm_server_1_lm_session_key = strequal ( parameter , " Yes " ) ;
} else {
2004-04-13 12:58:02 +04:00
x_fprintf ( x_stdout , " Error: Unknown request %s \n . \n " , request ) ;
2004-04-13 00:46:13 +04:00
}
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static void manage_squid_request ( enum stdio_helper_mode helper_mode , stdio_helper_function fn )
2002-11-24 11:32:03 +03:00
{
char buf [ SQUID_BUFFER_SIZE + 1 ] ;
int length ;
2003-01-16 06:29:54 +03:00
char * c ;
2002-11-24 11:32:03 +03:00
static BOOL err ;
2003-03-24 12:54:13 +03:00
/* this is not a typo - x_fgets doesn't work too well under squid */
if ( fgets ( buf , sizeof ( buf ) - 1 , stdin ) = = NULL ) {
2004-06-14 08:37:29 +04:00
if ( ferror ( stdin ) ) {
DEBUG ( 1 , ( " fgets() failed! dying..... errno=%d (%s) \n " , ferror ( stdin ) ,
strerror ( ferror ( stdin ) ) ) ) ;
exit ( 1 ) ; /* BIIG buffer */
}
exit ( 0 ) ;
2002-11-24 11:32:03 +03:00
}
c = memchr ( buf , ' \n ' , sizeof ( buf ) - 1 ) ;
if ( c ) {
* c = ' \0 ' ;
length = c - buf ;
} else {
err = 1 ;
return ;
}
if ( err ) {
DEBUG ( 2 , ( " Oversized message \n " ) ) ;
x_fprintf ( x_stderr , " ERR \n " ) ;
err = 0 ;
return ;
}
DEBUG ( 10 , ( " Got '%s' from squid (length: %d). \n " , buf , length ) ) ;
if ( buf [ 0 ] = = ' \0 ' ) {
DEBUG ( 2 , ( " Invalid Request \n " ) ) ;
x_fprintf ( x_stderr , " ERR \n " ) ;
return ;
}
2002-11-25 00:08:36 +03:00
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
fn ( helper_mode , buf , length ) ;
2002-11-24 11:32:03 +03:00
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
static void squid_stream ( enum stdio_helper_mode stdio_mode , stdio_helper_function fn ) {
2002-11-24 11:32:03 +03:00
/* initialize FDescs */
x_setbuf ( x_stdout , NULL ) ;
x_setbuf ( x_stderr , NULL ) ;
while ( 1 ) {
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
manage_squid_request ( stdio_mode , fn ) ;
2002-11-24 11:32:03 +03:00
}
}
/* Authenticate a user with a challenge/response */
static BOOL check_auth_crap ( void )
{
2003-04-07 11:10:53 +04:00
NTSTATUS nt_status ;
uint32 flags = 0 ;
char lm_key [ 8 ] ;
2004-04-12 14:17:48 +04:00
char user_session_key [ 16 ] ;
2003-04-07 11:10:53 +04:00
char * hex_lm_key ;
2004-04-12 14:17:48 +04:00
char * hex_user_session_key ;
2003-04-07 11:10:53 +04:00
char * error_string ;
2003-03-24 12:54:13 +03:00
static uint8 zeros [ 16 ] ;
2003-05-12 05:49:03 +04:00
x_setbuf ( x_stdout , NULL ) ;
2003-03-24 12:54:13 +03:00
if ( request_lm_key )
2003-07-07 09:11:10 +04:00
flags | = WBFLAG_PAM_LMKEY ;
2003-03-24 12:54:13 +03:00
2004-04-12 14:17:48 +04:00
if ( request_user_session_key )
2004-04-06 12:11:16 +04:00
flags | = WBFLAG_PAM_USER_SESSION_KEY ;
2003-04-07 11:10:53 +04:00
2004-04-06 20:44:24 +04:00
flags | = WBFLAG_PAM_NT_STATUS_SQUASH ;
2003-04-07 11:10:53 +04:00
nt_status = contact_winbind_auth_crap ( opt_username , opt_domain ,
opt_workstation ,
& opt_challenge ,
& opt_lm_response ,
& opt_nt_response ,
flags ,
2003-08-15 08:42:05 +04:00
( unsigned char * ) lm_key ,
2004-04-12 14:17:48 +04:00
( unsigned char * ) user_session_key ,
2003-12-30 16:20:39 +03:00
& error_string , NULL ) ;
2003-04-07 11:10:53 +04:00
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
2003-05-12 04:18:45 +04:00
x_fprintf ( x_stdout , " %s (0x%x) \n " ,
error_string ,
NT_STATUS_V ( nt_status ) ) ;
2003-04-07 11:10:53 +04:00
SAFE_FREE ( error_string ) ;
return False ;
}
2003-03-24 12:54:13 +03:00
2003-04-07 11:10:53 +04:00
if ( request_lm_key
& & ( memcmp ( zeros , lm_key ,
sizeof ( lm_key ) ) ! = 0 ) ) {
2005-08-08 00:59:28 +04:00
hex_lm_key = hex_encode ( NULL , ( const unsigned char * ) lm_key ,
sizeof ( lm_key ) ) ;
2003-05-12 04:18:45 +04:00
x_fprintf ( x_stdout , " LM_KEY: %s \n " , hex_lm_key ) ;
2006-02-20 20:59:58 +03:00
TALLOC_FREE ( hex_lm_key ) ;
2003-04-07 11:10:53 +04:00
}
2004-04-12 14:17:48 +04:00
if ( request_user_session_key
& & ( memcmp ( zeros , user_session_key ,
sizeof ( user_session_key ) ) ! = 0 ) ) {
2005-08-08 00:59:28 +04:00
hex_user_session_key = hex_encode ( NULL , ( const unsigned char * ) user_session_key ,
sizeof ( user_session_key ) ) ;
2004-04-12 14:17:48 +04:00
x_fprintf ( x_stdout , " NT_KEY: %s \n " , hex_user_session_key ) ;
2006-02-20 20:59:58 +03:00
TALLOC_FREE ( hex_user_session_key ) ;
2003-04-07 11:10:53 +04:00
}
2002-11-24 11:32:03 +03:00
2003-04-07 11:10:53 +04:00
return True ;
}
2002-11-24 11:32:03 +03:00
/* Main program */
2003-01-17 07:12:12 +03:00
enum {
2002-11-24 11:32:03 +03:00
OPT_USERNAME = 1000 ,
OPT_DOMAIN ,
OPT_WORKSTATION ,
OPT_CHALLENGE ,
OPT_RESPONSE ,
OPT_LM ,
OPT_NT ,
2003-03-24 12:54:13 +03:00
OPT_PASSWORD ,
OPT_LM_KEY ,
2004-04-12 14:17:48 +04:00
OPT_USER_SESSION_KEY ,
2004-04-06 20:44:24 +04:00
OPT_DIAGNOSTICS ,
OPT_REQUIRE_MEMBERSHIP
2002-11-24 11:32:03 +03:00
} ;
2003-04-07 11:10:53 +04:00
int main ( int argc , const char * * argv )
2002-11-24 11:32:03 +03:00
{
int opt ;
2003-05-05 09:01:59 +04:00
static const char * helper_protocol ;
static int diagnostics ;
2002-11-24 11:32:03 +03:00
2003-04-07 11:10:53 +04:00
static const char * hex_challenge ;
static const char * hex_lm_response ;
static const char * hex_nt_response ;
2002-11-24 11:32:03 +03:00
poptContext pc ;
2003-04-07 11:10:53 +04:00
/* NOTE: DO NOT change this interface without considering the implications!
This is an external interface , which other programs will use to interact
with this helper .
*/
/* We do not use single-letter command abbreviations, because they harm future
interface stability . */
2002-11-24 11:32:03 +03:00
struct poptOption long_options [ ] = {
POPT_AUTOHELP
{ " helper-protocol " , 0 , POPT_ARG_STRING , & helper_protocol , OPT_DOMAIN , " operate as a stdio-based helper " , " helper protocol to use " } ,
2003-04-07 11:10:53 +04:00
{ " username " , 0 , POPT_ARG_STRING , & opt_username , OPT_USERNAME , " username " } ,
{ " domain " , 0 , POPT_ARG_STRING , & opt_domain , OPT_DOMAIN , " domain name " } ,
{ " workstation " , 0 , POPT_ARG_STRING , & opt_workstation , OPT_WORKSTATION , " workstation " } ,
2002-11-24 11:32:03 +03:00
{ " challenge " , 0 , POPT_ARG_STRING , & hex_challenge , OPT_CHALLENGE , " challenge (HEX encoded) " } ,
{ " lm-response " , 0 , POPT_ARG_STRING , & hex_lm_response , OPT_LM , " LM Response to the challenge (HEX encoded) " } ,
{ " nt-response " , 0 , POPT_ARG_STRING , & hex_nt_response , OPT_NT , " NT or NTLMv2 Response to the challenge (HEX encoded) " } ,
2003-04-07 11:10:53 +04:00
{ " password " , 0 , POPT_ARG_STRING , & opt_password , OPT_PASSWORD , " User's plaintext password " } ,
2005-04-25 06:30:12 +04:00
{ " request-lm-key " , 0 , POPT_ARG_NONE , & request_lm_key , OPT_LM_KEY , " Retrieve LM session key " } ,
{ " request-nt-key " , 0 , POPT_ARG_NONE , & request_user_session_key , OPT_USER_SESSION_KEY , " Retrieve User (NT) session key " } ,
2003-04-07 11:10:53 +04:00
{ " diagnostics " , 0 , POPT_ARG_NONE , & diagnostics , OPT_DIAGNOSTICS , " Perform diagnostics on the authentictaion chain " } ,
2004-04-06 20:44:24 +04:00
{ " require-membership-of " , 0 , POPT_ARG_STRING , & require_membership_of , OPT_REQUIRE_MEMBERSHIP , " Require that a user be a member of this group (either name or SID) for authentication to succeed " } ,
2003-04-14 07:59:04 +04:00
POPT_COMMON_SAMBA
POPT_TABLEEND
2002-11-24 11:32:03 +03:00
} ;
/* Samba client initialisation */
2005-12-29 01:48:54 +03:00
load_case_tables ( ) ;
2002-11-24 11:32:03 +03:00
dbf = x_stderr ;
2003-05-12 04:18:45 +04:00
/* Samba client initialisation */
2006-01-29 01:53:04 +03:00
if ( ! lp_load ( dyn_CONFIGFILE , True , False , False , True ) ) {
2004-09-01 00:36:55 +04:00
d_fprintf ( stderr , " ntlm_auth: error opening config file %s. Error was %s \n " ,
2003-05-12 04:18:45 +04:00
dyn_CONFIGFILE , strerror ( errno ) ) ;
exit ( 1 ) ;
}
2002-11-24 11:32:03 +03:00
/* Parse options */
pc = poptGetContext ( " ntlm_auth " , argc , argv , long_options , 0 ) ;
/* Parse command line options */
if ( argc = = 1 ) {
poptPrintHelp ( pc , stderr , 0 ) ;
return 1 ;
}
pc = poptGetContext ( NULL , argc , ( const char * * ) argv , long_options ,
POPT_CONTEXT_KEEP_FIRST ) ;
while ( ( opt = poptGetNextOpt ( pc ) ) ! = - 1 ) {
switch ( opt ) {
case OPT_CHALLENGE :
2005-08-08 00:59:28 +04:00
opt_challenge = strhex_to_data_blob ( NULL , hex_challenge ) ;
2004-04-06 20:44:24 +04:00
if ( opt_challenge . length ! = 8 ) {
2004-04-12 14:17:48 +04:00
x_fprintf ( x_stderr , " hex decode of %s failed! (only got %d bytes) \n " ,
hex_challenge ,
( int ) opt_challenge . length ) ;
2002-11-24 11:32:03 +03:00
exit ( 1 ) ;
}
break ;
case OPT_LM :
2005-08-08 00:59:28 +04:00
opt_lm_response = strhex_to_data_blob ( NULL , hex_lm_response ) ;
2004-04-06 20:44:24 +04:00
if ( opt_lm_response . length ! = 24 ) {
2004-04-12 14:17:48 +04:00
x_fprintf ( x_stderr , " hex decode of %s failed! (only got %d bytes) \n " ,
hex_lm_response ,
( int ) opt_lm_response . length ) ;
2002-11-24 11:32:03 +03:00
exit ( 1 ) ;
}
break ;
2004-04-06 20:44:24 +04:00
2003-03-24 12:54:13 +03:00
case OPT_NT :
2005-08-08 00:59:28 +04:00
opt_nt_response = strhex_to_data_blob ( NULL , hex_nt_response ) ;
2004-04-06 20:44:24 +04:00
if ( opt_nt_response . length < 24 ) {
2004-04-12 14:17:48 +04:00
x_fprintf ( x_stderr , " hex decode of %s failed! (only got %d bytes) \n " ,
hex_nt_response ,
( int ) opt_nt_response . length ) ;
2002-11-24 11:32:03 +03:00
exit ( 1 ) ;
}
2004-04-06 20:44:24 +04:00
break ;
case OPT_REQUIRE_MEMBERSHIP :
if ( StrnCaseCmp ( " S- " , require_membership_of , 2 ) = = 0 ) {
2004-10-01 07:28:39 +04:00
require_membership_of_sid = require_membership_of ;
2004-04-06 20:44:24 +04:00
}
2002-11-24 11:32:03 +03:00
break ;
}
}
if ( helper_protocol ) {
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
int i ;
for ( i = 0 ; i < NUM_HELPER_MODES ; i + + ) {
if ( strcmp ( helper_protocol , stdio_helper_protocols [ i ] . name ) = = 0 ) {
squid_stream ( stdio_helper_protocols [ i ] . mode , stdio_helper_protocols [ i ] . fn ) ;
exit ( 0 ) ;
}
2002-11-24 11:32:03 +03:00
}
2003-12-30 10:33:58 +03:00
x_fprintf ( x_stderr , " unknown helper protocol [%s] \n \n Valid helper protools: \n \n " , helper_protocol ) ;
for ( i = 0 ; i < NUM_HELPER_MODES ; i + + ) {
x_fprintf ( x_stderr , " %s \n " , stdio_helper_protocols [ i ] . name ) ;
}
Changes all over the shop, but all towards:
- NTLM2 support in the server
- KEY_EXCH support in the server
- variable length session keys.
In detail:
- NTLM2 is an extension of NTLMv1, that is compatible with existing
domain controllers (unlike NTLMv2, which requires a DC upgrade).
* This is known as 'NTLMv2 session security' *
(This is not yet implemented on the RPC pipes however, so there may
well still be issues for PDC setups, particuarly around password
changes. We do not fully understand the sign/seal implications of
NTLM2 on RPC pipes.)
This requires modifications to our authentication subsystem, as we
must handle the 'challege' input into the challenge-response algorithm
being changed. This also needs to be turned off for
'security=server', which does not support this.
- KEY_EXCH is another 'security' mechanism, whereby the session key
actually used by the server is sent by the client, rather than being
the shared-secret directly or indirectly.
- As both these methods change the session key, the auth subsystem
needed to be changed, to 'override' session keys provided by the
backend.
- There has also been a major overhaul of the NTLMSSP subsystem, to merge the 'client' and 'server' functions, so they both operate on a single structure. This should help the SPNEGO implementation.
- The 'names blob' in NTLMSSP is always in unicode - never in ascii.
Don't make an ascii version ever.
- The other big change is to allow variable length session keys. We
have always assumed that session keys are 16 bytes long - and padded
to this length if shorter. However, Kerberos session keys are 8 bytes
long, when the krb5 login uses DES.
* This fix allows SMB signging on machines not yet running MIT KRB5 1.3.1. *
- Add better DEBUG() messages to ntlm_auth, warning administrators of
misconfigurations that prevent access to the privileged pipe. This
should help reduce some of the 'it just doesn't work' issues.
- Fix data_blob_talloc() to behave the same way data_blob() does when
passed a NULL data pointer. (just allocate)
REMEMBER to make clean after this commit - I have changed plenty of data structures...
(This used to be commit f3bbc87b0dac63426cda6fac7a295d3aad810ecc)
2003-11-22 16:19:38 +03:00
exit ( 1 ) ;
2002-11-24 11:32:03 +03:00
}
2003-04-07 11:10:53 +04:00
if ( ! opt_username ) {
x_fprintf ( x_stderr , " username must be specified! \n \n " ) ;
poptPrintHelp ( pc , stderr , 0 ) ;
exit ( 1 ) ;
}
if ( opt_domain = = NULL ) {
opt_domain = get_winbind_domain ( ) ;
2002-11-24 11:32:03 +03:00
}
2003-04-07 11:10:53 +04:00
if ( opt_workstation = = NULL ) {
opt_workstation = " " ;
2002-11-24 11:32:03 +03:00
}
2003-04-07 11:10:53 +04:00
if ( opt_challenge . length ) {
2002-11-24 11:32:03 +03:00
if ( ! check_auth_crap ( ) ) {
exit ( 1 ) ;
}
2003-04-07 11:10:53 +04:00
exit ( 0 ) ;
}
if ( ! opt_password ) {
opt_password = getpass ( " password: " ) ;
}
if ( diagnostics ) {
if ( ! diagnose_ntlm_auth ( ) ) {
2004-04-12 14:17:48 +04:00
return 1 ;
2003-04-07 11:10:53 +04:00
}
} else {
2002-11-24 11:32:03 +03:00
fstring user ;
2003-04-07 11:10:53 +04:00
2003-07-23 16:33:59 +04:00
fstr_sprintf ( user , " %s%c%s " , opt_domain , winbind_separator ( ) , opt_username ) ;
2003-04-07 11:10:53 +04:00
if ( ! check_plaintext_auth ( user , opt_password , True ) ) {
2004-04-12 14:17:48 +04:00
return 1 ;
2002-11-24 11:32:03 +03:00
}
}
/* Exit code */
poptFreeContext ( pc ) ;
return 0 ;
}