2005-01-01 03:19:08 +03:00
/*
Unix SMB / CIFS implementation .
Kerberos backend for GENSEC
2005-06-28 04:55:44 +04:00
Copyright ( C ) Andrew Bartlett < abartlet @ samba . org > 2004 - 2005
Copyright ( C ) Stefan Metzmacher < metze @ samba . org > 2004 - 2005
2005-01-01 03:19:08 +03:00
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"
# include "system/kerberos.h"
2005-05-11 16:03:48 +04:00
# include "auth/kerberos/kerberos.h"
# include "librpc/gen_ndr/ndr_krb5pac.h"
2005-01-01 03:19:08 +03:00
# include "auth/auth.h"
struct gensec_gssapi_state {
gss_ctx_id_t gssapi_context ;
struct gss_channel_bindings_struct * input_chan_bindings ;
gss_name_t server_name ;
gss_name_t client_name ;
2005-01-24 17:30:22 +03:00
OM_uint32 want_flags , got_flags ;
2005-01-03 02:53:14 +03:00
const gss_OID_desc * gss_oid ;
2005-05-11 16:03:48 +04:00
DATA_BLOB session_key ;
DATA_BLOB pac ;
2005-05-16 03:42:11 +04:00
2005-06-04 15:17:05 +04:00
struct smb_krb5_context * smb_krb5_context ;
2005-11-02 03:31:22 +03:00
struct gssapi_creds_container * client_cred ;
r11995: A big kerberos-related update.
This merges Samba4 up to current lorikeet-heimdal, which includes a
replacement for some Samba-specific hacks.
In particular, the credentials system now supplies GSS client and
server credentials. These are imported into GSS with
gss_krb5_import_creds(). Unfortunetly this can't take an MEMORY
keytab, so we now create a FILE based keytab as provision and join
time.
Because the keytab is now created in advance, we don't spend .4s at
negprot doing sha1 s2k calls. Also, because the keytab is read in
real time, any change in the server key will be correctly picked up by
the the krb5 code.
To mark entries in the secrets which should be exported to a keytab,
there is a new kerberosSecret objectClass. The new routine
cli_credentials_update_all_keytabs() searches for these, and updates
the keytabs.
This is called in the provision.js via the ejs wrapper
credentials_update_all_keytabs().
We can now (in theory) use a system-provided /etc/krb5.keytab, if
krb5Keytab: FILE:/etc/krb5.keytab
is added to the secrets.ldb record. By default the attribute
privateKeytab: secrets.keytab
is set, pointing to allow the whole private directory to be moved
without breaking the internal links.
(This used to be commit 6b75573df49c6210e1b9d71e108a9490976bd41d)
2005-12-01 08:20:39 +03:00
struct gssapi_creds_container * server_cred ;
2005-05-16 03:42:11 +04:00
2005-11-02 03:31:22 +03:00
gss_cred_id_t delegated_cred_handle ;
2005-01-01 03:19:08 +03:00
} ;
2005-05-16 03:42:11 +04:00
static char * gssapi_error_string ( TALLOC_CTX * mem_ctx ,
OM_uint32 maj_stat , OM_uint32 min_stat )
{
OM_uint32 disp_min_stat , disp_maj_stat ;
gss_buffer_desc maj_error_message ;
gss_buffer_desc min_error_message ;
OM_uint32 msg_ctx = 0 ;
char * ret ;
maj_error_message . value = NULL ;
min_error_message . value = NULL ;
disp_maj_stat = gss_display_status ( & disp_min_stat , maj_stat , GSS_C_GSS_CODE ,
GSS_C_NULL_OID , & msg_ctx , & maj_error_message ) ;
disp_maj_stat = gss_display_status ( & disp_min_stat , min_stat , GSS_C_MECH_CODE ,
GSS_C_NULL_OID , & msg_ctx , & min_error_message ) ;
ret = talloc_asprintf ( mem_ctx , " %s: %s " , ( char * ) maj_error_message . value , ( char * ) min_error_message . value ) ;
gss_release_buffer ( & disp_min_stat , & maj_error_message ) ;
gss_release_buffer ( & disp_min_stat , & min_error_message ) ;
return ret ;
}
2005-01-01 03:19:08 +03:00
static int gensec_gssapi_destory ( void * ptr )
{
struct gensec_gssapi_state * gensec_gssapi_state = ptr ;
OM_uint32 maj_stat , min_stat ;
2005-05-16 03:42:11 +04:00
2005-11-02 03:31:22 +03:00
if ( gensec_gssapi_state - > delegated_cred_handle ! = GSS_C_NO_CREDENTIAL ) {
maj_stat = gss_release_cred ( & min_stat ,
& gensec_gssapi_state - > delegated_cred_handle ) ;
}
2005-01-01 03:19:08 +03:00
if ( gensec_gssapi_state - > gssapi_context ! = GSS_C_NO_CONTEXT ) {
maj_stat = gss_delete_sec_context ( & min_stat ,
& gensec_gssapi_state - > gssapi_context ,
GSS_C_NO_BUFFER ) ;
}
if ( gensec_gssapi_state - > server_name ! = GSS_C_NO_NAME ) {
maj_stat = gss_release_name ( & min_stat , & gensec_gssapi_state - > server_name ) ;
}
if ( gensec_gssapi_state - > client_name ! = GSS_C_NO_NAME ) {
maj_stat = gss_release_name ( & min_stat , & gensec_gssapi_state - > client_name ) ;
}
return 0 ;
}
static NTSTATUS gensec_gssapi_start ( struct gensec_security * gensec_security )
{
struct gensec_gssapi_state * gensec_gssapi_state ;
2005-05-16 03:42:11 +04:00
krb5_error_code ret ;
2005-01-27 10:08:20 +03:00
gensec_gssapi_state = talloc ( gensec_security , struct gensec_gssapi_state ) ;
2005-01-01 03:19:08 +03:00
if ( ! gensec_gssapi_state ) {
return NT_STATUS_NO_MEMORY ;
}
gensec_security - > private_data = gensec_gssapi_state ;
gensec_gssapi_state - > gssapi_context = GSS_C_NO_CONTEXT ;
gensec_gssapi_state - > server_name = GSS_C_NO_NAME ;
gensec_gssapi_state - > client_name = GSS_C_NO_NAME ;
/* TODO: Fill in channel bindings */
gensec_gssapi_state - > input_chan_bindings = GSS_C_NO_CHANNEL_BINDINGS ;
2005-11-02 06:48:49 +03:00
gensec_gssapi_state - > want_flags = 0 ;
if ( lp_parm_bool ( - 1 , " gensec_gssapi " , " mutual " , True ) ) {
gensec_gssapi_state - > want_flags | = GSS_C_MUTUAL_FLAG ;
}
2005-11-02 07:12:47 +03:00
if ( lp_parm_bool ( - 1 , " gensec_gssapi " , " delegation " , True ) ) {
2005-11-02 06:48:49 +03:00
gensec_gssapi_state - > want_flags | = GSS_C_DELEG_FLAG ;
}
2005-01-01 03:19:08 +03:00
gensec_gssapi_state - > got_flags = 0 ;
2005-05-11 16:03:48 +04:00
gensec_gssapi_state - > session_key = data_blob ( NULL , 0 ) ;
gensec_gssapi_state - > pac = data_blob ( NULL , 0 ) ;
2005-11-02 03:31:22 +03:00
gensec_gssapi_state - > delegated_cred_handle = GSS_C_NO_CREDENTIAL ;
2005-05-16 03:42:11 +04:00
talloc_set_destructor ( gensec_gssapi_state , gensec_gssapi_destory ) ;
2005-01-01 03:19:08 +03:00
if ( gensec_security - > want_features & GENSEC_FEATURE_SIGN ) {
gensec_gssapi_state - > want_flags | = GSS_C_INTEG_FLAG ;
}
if ( gensec_security - > want_features & GENSEC_FEATURE_SEAL ) {
gensec_gssapi_state - > want_flags | = GSS_C_CONF_FLAG ;
}
2005-05-11 16:03:48 +04:00
if ( gensec_security - > want_features & GENSEC_FEATURE_DCE_STYLE ) {
gensec_gssapi_state - > want_flags | = GSS_C_DCE_STYLE ;
}
2005-01-01 03:19:08 +03:00
2005-05-16 05:31:22 +04:00
gensec_gssapi_state - > gss_oid = gss_mech_krb5 ;
2005-05-16 03:42:11 +04:00
2005-06-04 15:17:05 +04:00
ret = smb_krb5_init_context ( gensec_gssapi_state ,
& gensec_gssapi_state - > smb_krb5_context ) ;
2005-05-16 03:42:11 +04:00
if ( ret ) {
DEBUG ( 1 , ( " gensec_krb5_start: krb5_init_context failed (%s) \n " ,
2005-06-04 15:17:05 +04:00
error_message ( ret ) ) ) ;
2005-05-16 03:42:11 +04:00
return NT_STATUS_INTERNAL_ERROR ;
}
2005-01-01 03:19:08 +03:00
return NT_STATUS_OK ;
}
static NTSTATUS gensec_gssapi_server_start ( struct gensec_security * gensec_security )
{
NTSTATUS nt_status ;
2005-10-20 07:47:55 +04:00
int ret ;
2005-01-01 03:19:08 +03:00
struct gensec_gssapi_state * gensec_gssapi_state ;
2005-06-22 06:12:26 +04:00
struct cli_credentials * machine_account ;
r11995: A big kerberos-related update.
This merges Samba4 up to current lorikeet-heimdal, which includes a
replacement for some Samba-specific hacks.
In particular, the credentials system now supplies GSS client and
server credentials. These are imported into GSS with
gss_krb5_import_creds(). Unfortunetly this can't take an MEMORY
keytab, so we now create a FILE based keytab as provision and join
time.
Because the keytab is now created in advance, we don't spend .4s at
negprot doing sha1 s2k calls. Also, because the keytab is read in
real time, any change in the server key will be correctly picked up by
the the krb5 code.
To mark entries in the secrets which should be exported to a keytab,
there is a new kerberosSecret objectClass. The new routine
cli_credentials_update_all_keytabs() searches for these, and updates
the keytabs.
This is called in the provision.js via the ejs wrapper
credentials_update_all_keytabs().
We can now (in theory) use a system-provided /etc/krb5.keytab, if
krb5Keytab: FILE:/etc/krb5.keytab
is added to the secrets.ldb record. By default the attribute
privateKeytab: secrets.keytab
is set, pointing to allow the whole private directory to be moved
without breaking the internal links.
(This used to be commit 6b75573df49c6210e1b9d71e108a9490976bd41d)
2005-12-01 08:20:39 +03:00
struct gssapi_creds_container * gcc ;
2005-01-01 03:19:08 +03:00
nt_status = gensec_gssapi_start ( gensec_security ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
return nt_status ;
}
gensec_gssapi_state = gensec_security - > private_data ;
2005-10-20 07:47:55 +04:00
machine_account = gensec_get_credentials ( gensec_security ) ;
2005-06-22 06:12:26 +04:00
2005-10-20 07:47:55 +04:00
if ( ! machine_account ) {
DEBUG ( 3 , ( " No machine account credentials specified \n " ) ) ;
return NT_STATUS_INVALID_PARAMETER ;
2005-06-22 06:12:26 +04:00
} else {
r11995: A big kerberos-related update.
This merges Samba4 up to current lorikeet-heimdal, which includes a
replacement for some Samba-specific hacks.
In particular, the credentials system now supplies GSS client and
server credentials. These are imported into GSS with
gss_krb5_import_creds(). Unfortunetly this can't take an MEMORY
keytab, so we now create a FILE based keytab as provision and join
time.
Because the keytab is now created in advance, we don't spend .4s at
negprot doing sha1 s2k calls. Also, because the keytab is read in
real time, any change in the server key will be correctly picked up by
the the krb5 code.
To mark entries in the secrets which should be exported to a keytab,
there is a new kerberosSecret objectClass. The new routine
cli_credentials_update_all_keytabs() searches for these, and updates
the keytabs.
This is called in the provision.js via the ejs wrapper
credentials_update_all_keytabs().
We can now (in theory) use a system-provided /etc/krb5.keytab, if
krb5Keytab: FILE:/etc/krb5.keytab
is added to the secrets.ldb record. By default the attribute
privateKeytab: secrets.keytab
is set, pointing to allow the whole private directory to be moved
without breaking the internal links.
(This used to be commit 6b75573df49c6210e1b9d71e108a9490976bd41d)
2005-12-01 08:20:39 +03:00
ret = cli_credentials_get_server_gss_creds ( machine_account , & gcc ) ;
2005-10-20 07:47:55 +04:00
if ( ret ) {
r11995: A big kerberos-related update.
This merges Samba4 up to current lorikeet-heimdal, which includes a
replacement for some Samba-specific hacks.
In particular, the credentials system now supplies GSS client and
server credentials. These are imported into GSS with
gss_krb5_import_creds(). Unfortunetly this can't take an MEMORY
keytab, so we now create a FILE based keytab as provision and join
time.
Because the keytab is now created in advance, we don't spend .4s at
negprot doing sha1 s2k calls. Also, because the keytab is read in
real time, any change in the server key will be correctly picked up by
the the krb5 code.
To mark entries in the secrets which should be exported to a keytab,
there is a new kerberosSecret objectClass. The new routine
cli_credentials_update_all_keytabs() searches for these, and updates
the keytabs.
This is called in the provision.js via the ejs wrapper
credentials_update_all_keytabs().
We can now (in theory) use a system-provided /etc/krb5.keytab, if
krb5Keytab: FILE:/etc/krb5.keytab
is added to the secrets.ldb record. By default the attribute
privateKeytab: secrets.keytab
is set, pointing to allow the whole private directory to be moved
without breaking the internal links.
(This used to be commit 6b75573df49c6210e1b9d71e108a9490976bd41d)
2005-12-01 08:20:39 +03:00
DEBUG ( 1 , ( " Aquiring acceptor credentials failed: %s \n " ,
error_message ( ret ) ) ) ;
2005-10-20 07:47:55 +04:00
return NT_STATUS_CANT_ACCESS_DOMAIN_INFO ;
2005-06-22 06:12:26 +04:00
}
}
r11995: A big kerberos-related update.
This merges Samba4 up to current lorikeet-heimdal, which includes a
replacement for some Samba-specific hacks.
In particular, the credentials system now supplies GSS client and
server credentials. These are imported into GSS with
gss_krb5_import_creds(). Unfortunetly this can't take an MEMORY
keytab, so we now create a FILE based keytab as provision and join
time.
Because the keytab is now created in advance, we don't spend .4s at
negprot doing sha1 s2k calls. Also, because the keytab is read in
real time, any change in the server key will be correctly picked up by
the the krb5 code.
To mark entries in the secrets which should be exported to a keytab,
there is a new kerberosSecret objectClass. The new routine
cli_credentials_update_all_keytabs() searches for these, and updates
the keytabs.
This is called in the provision.js via the ejs wrapper
credentials_update_all_keytabs().
We can now (in theory) use a system-provided /etc/krb5.keytab, if
krb5Keytab: FILE:/etc/krb5.keytab
is added to the secrets.ldb record. By default the attribute
privateKeytab: secrets.keytab
is set, pointing to allow the whole private directory to be moved
without breaking the internal links.
(This used to be commit 6b75573df49c6210e1b9d71e108a9490976bd41d)
2005-12-01 08:20:39 +03:00
gensec_gssapi_state - > server_cred = gcc ;
2005-01-01 03:19:08 +03:00
return NT_STATUS_OK ;
2005-06-22 06:12:26 +04:00
2005-01-01 03:19:08 +03:00
}
static NTSTATUS gensec_gssapi_client_start ( struct gensec_security * gensec_security )
{
struct gensec_gssapi_state * gensec_gssapi_state ;
2005-06-28 04:55:44 +04:00
struct cli_credentials * creds = gensec_get_credentials ( gensec_security ) ;
2005-08-29 08:30:22 +04:00
krb5_error_code ret ;
2005-01-01 03:19:08 +03:00
NTSTATUS nt_status ;
gss_buffer_desc name_token ;
2005-09-17 13:46:20 +04:00
gss_OID name_type ;
2005-01-01 03:19:08 +03:00
OM_uint32 maj_stat , min_stat ;
2005-07-09 05:58:38 +04:00
const char * hostname = gensec_get_target_hostname ( gensec_security ) ;
2005-09-17 13:46:20 +04:00
const char * principal ;
2005-11-02 03:31:22 +03:00
struct gssapi_creds_container * gcc ;
2005-07-09 05:58:38 +04:00
if ( ! hostname ) {
DEBUG ( 1 , ( " Could not determine hostname for target computer, cannot use kerberos \n " ) ) ;
return NT_STATUS_INVALID_PARAMETER ;
}
if ( is_ipaddress ( hostname ) ) {
2005-09-28 08:50:02 +04:00
DEBUG ( 2 , ( " Cannot do GSSAPI to an IP address \n " ) ) ;
return NT_STATUS_INVALID_PARAMETER ;
}
if ( strequal ( hostname , " localhost " ) ) {
DEBUG ( 2 , ( " GSSAPI to 'localhost' does not make sense \n " ) ) ;
2005-07-09 05:58:38 +04:00
return NT_STATUS_INVALID_PARAMETER ;
}
2005-01-05 13:21:08 +03:00
2005-01-01 03:19:08 +03:00
nt_status = gensec_gssapi_start ( gensec_security ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
return nt_status ;
}
gensec_gssapi_state = gensec_security - > private_data ;
2005-09-17 13:46:20 +04:00
principal = gensec_get_target_principal ( gensec_security ) ;
if ( principal & & lp_client_use_spnego_principal ( ) ) {
2005-10-24 19:38:07 +04:00
name_token . value = discard_const_p ( uint8_t , principal ) ;
name_token . length = strlen ( principal ) ;
2005-09-17 13:46:20 +04:00
name_type = GSS_C_NULL_OID ;
} else {
2005-10-24 19:38:07 +04:00
principal = talloc_asprintf ( gensec_gssapi_state , " %s@%s " ,
gensec_get_target_service ( gensec_security ) ,
hostname ) ;
name_token . value = discard_const_p ( uint8_t , principal ) ;
name_token . length = strlen ( principal ) ;
2005-09-17 13:46:20 +04:00
name_type = GSS_C_NT_HOSTBASED_SERVICE ;
}
2005-01-01 03:19:08 +03:00
maj_stat = gss_import_name ( & min_stat ,
& name_token ,
2005-09-17 13:46:20 +04:00
name_type ,
2005-01-01 03:19:08 +03:00
& gensec_gssapi_state - > server_name ) ;
2005-05-16 03:42:11 +04:00
if ( maj_stat ) {
2005-07-26 09:07:15 +04:00
DEBUG ( 2 , ( " GSS Import name of %s failed: %s \n " ,
2005-05-16 03:42:11 +04:00
( char * ) name_token . value ,
gssapi_error_string ( gensec_gssapi_state , maj_stat , min_stat ) ) ) ;
2005-07-20 14:54:50 +04:00
return NT_STATUS_INVALID_PARAMETER ;
2005-05-16 03:42:11 +04:00
}
2005-11-02 03:31:22 +03:00
ret = cli_credentials_get_client_gss_creds ( creds , & gcc ) ;
switch ( ret ) {
case 0 :
break ;
case KRB5_KDC_UNREACH :
DEBUG ( 3 , ( " Cannot reach a KDC we require \n " ) ) ;
return NT_STATUS_INVALID_PARAMETER ; /* Make SPNEGO ignore us, we can't go any further here */
default :
DEBUG ( 1 , ( " Aquiring initiator credentails failed \n " ) ) ;
return NT_STATUS_UNSUCCESSFUL ;
2005-01-01 03:19:08 +03:00
}
2005-05-16 03:42:11 +04:00
2005-11-02 03:31:22 +03:00
gensec_gssapi_state - > client_cred = gcc ;
2005-01-01 03:19:08 +03:00
return NT_STATUS_OK ;
}
2005-06-22 06:12:26 +04:00
/**
* Check if the packet is one for this mechansim
*
* @ param gensec_security GENSEC state
* @ param in The request , as a DATA_BLOB
* @ return Error , INVALID_PARAMETER if it ' s not a packet for us
* or NT_STATUS_OK if the packet is ok .
*/
static NTSTATUS gensec_gssapi_magic ( struct gensec_security * gensec_security ,
const DATA_BLOB * in )
{
if ( gensec_gssapi_check_oid ( in , GENSEC_OID_KERBEROS5 ) ) {
return NT_STATUS_OK ;
} else {
return NT_STATUS_INVALID_PARAMETER ;
}
}
2005-01-01 03:19:08 +03:00
/**
* Next state function for the GSSAPI GENSEC mechanism
*
* @ param gensec_gssapi_state GSSAPI State
* @ param out_mem_ctx The TALLOC_CTX for * out to be allocated on
* @ param in The request , as a DATA_BLOB
* @ param out The reply , as an talloc ( ) ed DATA_BLOB , on * out_mem_ctx
* @ return Error , MORE_PROCESSING_REQUIRED if a reply is sent ,
* or NT_STATUS_OK if the user is authenticated .
*/
static NTSTATUS gensec_gssapi_update ( struct gensec_security * gensec_security ,
TALLOC_CTX * out_mem_ctx ,
const DATA_BLOB in , DATA_BLOB * out )
{
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
2005-11-02 03:31:22 +03:00
NTSTATUS nt_status = NT_STATUS_LOGON_FAILURE ;
2005-01-01 03:19:08 +03:00
OM_uint32 maj_stat , min_stat ;
OM_uint32 min_stat2 ;
gss_buffer_desc input_token , output_token ;
2005-01-03 02:53:14 +03:00
gss_OID gss_oid_p ;
2005-01-01 03:19:08 +03:00
input_token . length = in . length ;
input_token . value = in . data ;
switch ( gensec_security - > gensec_role ) {
case GENSEC_CLIENT :
{
maj_stat = gss_init_sec_context ( & min_stat ,
2005-11-02 03:31:22 +03:00
gensec_gssapi_state - > client_cred - > creds ,
2005-01-01 03:19:08 +03:00
& gensec_gssapi_state - > gssapi_context ,
gensec_gssapi_state - > server_name ,
2005-01-03 02:53:14 +03:00
discard_const_p ( gss_OID_desc , gensec_gssapi_state - > gss_oid ) ,
2005-01-01 03:19:08 +03:00
gensec_gssapi_state - > want_flags ,
0 ,
gensec_gssapi_state - > input_chan_bindings ,
& input_token ,
NULL ,
& output_token ,
& gensec_gssapi_state - > got_flags , /* ret flags */
NULL ) ;
break ;
}
case GENSEC_SERVER :
{
maj_stat = gss_accept_sec_context ( & min_stat ,
2005-01-24 17:30:22 +03:00
& gensec_gssapi_state - > gssapi_context ,
r11995: A big kerberos-related update.
This merges Samba4 up to current lorikeet-heimdal, which includes a
replacement for some Samba-specific hacks.
In particular, the credentials system now supplies GSS client and
server credentials. These are imported into GSS with
gss_krb5_import_creds(). Unfortunetly this can't take an MEMORY
keytab, so we now create a FILE based keytab as provision and join
time.
Because the keytab is now created in advance, we don't spend .4s at
negprot doing sha1 s2k calls. Also, because the keytab is read in
real time, any change in the server key will be correctly picked up by
the the krb5 code.
To mark entries in the secrets which should be exported to a keytab,
there is a new kerberosSecret objectClass. The new routine
cli_credentials_update_all_keytabs() searches for these, and updates
the keytabs.
This is called in the provision.js via the ejs wrapper
credentials_update_all_keytabs().
We can now (in theory) use a system-provided /etc/krb5.keytab, if
krb5Keytab: FILE:/etc/krb5.keytab
is added to the secrets.ldb record. By default the attribute
privateKeytab: secrets.keytab
is set, pointing to allow the whole private directory to be moved
without breaking the internal links.
(This used to be commit 6b75573df49c6210e1b9d71e108a9490976bd41d)
2005-12-01 08:20:39 +03:00
gensec_gssapi_state - > server_cred - > creds ,
2005-01-01 03:19:08 +03:00
& input_token ,
gensec_gssapi_state - > input_chan_bindings ,
& gensec_gssapi_state - > client_name ,
2005-01-03 02:53:14 +03:00
& gss_oid_p ,
2005-01-01 03:19:08 +03:00
& output_token ,
& gensec_gssapi_state - > got_flags ,
NULL ,
2005-11-02 03:31:22 +03:00
& gensec_gssapi_state - > delegated_cred_handle ) ;
2005-01-03 02:53:14 +03:00
gensec_gssapi_state - > gss_oid = gss_oid_p ;
2005-01-01 03:19:08 +03:00
break ;
}
default :
return NT_STATUS_INVALID_PARAMETER ;
}
if ( maj_stat = = GSS_S_COMPLETE ) {
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
* out = data_blob_talloc ( out_mem_ctx , output_token . value , output_token . length ) ;
gss_release_buffer ( & min_stat2 , & output_token ) ;
2005-11-02 03:31:22 +03:00
if ( gensec_gssapi_state - > got_flags & GSS_C_DELEG_FLAG ) {
DEBUG ( 5 , ( " gensec_gssapi: credentials were delegated \n " ) ) ;
} else {
DEBUG ( 5 , ( " gensec_gssapi: NO credentials were delegated \n " ) ) ;
}
2005-01-01 03:19:08 +03:00
return NT_STATUS_OK ;
} else if ( maj_stat = = GSS_S_CONTINUE_NEEDED ) {
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
* out = data_blob_talloc ( out_mem_ctx , output_token . value , output_token . length ) ;
gss_release_buffer ( & min_stat2 , & output_token ) ;
2005-01-01 03:19:08 +03:00
return NT_STATUS_MORE_PROCESSING_REQUIRED ;
2005-09-28 08:50:02 +04:00
} else if ( ( gensec_gssapi_state - > gss_oid - > length = = gss_mech_krb5 - > length )
& & ( memcmp ( gensec_gssapi_state - > gss_oid - > elements , gss_mech_krb5 - > elements ,
gensec_gssapi_state - > gss_oid - > length ) = = 0 ) ) {
switch ( min_stat ) {
2005-10-29 17:13:52 +04:00
case KRB5_KDC_UNREACH :
DEBUG ( 3 , ( " Cannot reach a KDC we require: %s \n " ,
gssapi_error_string ( gensec_gssapi_state , maj_stat , min_stat ) ) ) ;
return NT_STATUS_INVALID_PARAMETER ; /* Make SPNEGO ignore us, we can't go any further here */
2005-09-28 08:50:02 +04:00
case KRB5KDC_ERR_S_PRINCIPAL_UNKNOWN :
DEBUG ( 3 , ( " Server is not registered with our KDC: %s \n " ,
gssapi_error_string ( gensec_gssapi_state , maj_stat , min_stat ) ) ) ;
return NT_STATUS_INVALID_PARAMETER ; /* Make SPNEGO ignore us, we can't go any further here */
case KRB5KRB_AP_ERR_MSG_TYPE :
2005-05-16 05:31:22 +04:00
/* garbage input, possibly from the auto-mech detection */
return NT_STATUS_INVALID_PARAMETER ;
2005-09-28 08:50:02 +04:00
default :
DEBUG ( 1 , ( " GSS(krb5) Update failed: %s \n " ,
gssapi_error_string ( out_mem_ctx , maj_stat , min_stat ) ) ) ;
return nt_status ;
2005-05-16 05:31:22 +04:00
}
2005-09-28 08:50:02 +04:00
} else {
2005-05-16 03:42:11 +04:00
DEBUG ( 1 , ( " GSS Update failed: %s \n " ,
gssapi_error_string ( out_mem_ctx , maj_stat , min_stat ) ) ) ;
2005-01-01 03:19:08 +03:00
return nt_status ;
}
}
static NTSTATUS gensec_gssapi_wrap ( struct gensec_security * gensec_security ,
TALLOC_CTX * mem_ctx ,
const DATA_BLOB * in ,
DATA_BLOB * out )
{
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
OM_uint32 maj_stat , min_stat ;
gss_buffer_desc input_token , output_token ;
int conf_state ;
input_token . length = in - > length ;
input_token . value = in - > data ;
maj_stat = gss_wrap ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
gensec_have_feature ( gensec_security , GENSEC_FEATURE_SEAL ) ,
GSS_C_QOP_DEFAULT ,
& input_token ,
& conf_state ,
& output_token ) ;
if ( GSS_ERROR ( maj_stat ) ) {
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
DEBUG ( 1 , ( " gensec_gssapi_wrap: GSS Wrap failed: %s \n " ,
2005-05-16 03:42:11 +04:00
gssapi_error_string ( mem_ctx , maj_stat , min_stat ) ) ) ;
2005-01-01 03:19:08 +03:00
return NT_STATUS_ACCESS_DENIED ;
}
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
* out = data_blob_talloc ( mem_ctx , output_token . value , output_token . length ) ;
2005-01-01 03:19:08 +03:00
gss_release_buffer ( & min_stat , & output_token ) ;
if ( gensec_have_feature ( gensec_security , GENSEC_FEATURE_SEAL )
& & ! conf_state ) {
return NT_STATUS_ACCESS_DENIED ;
}
return NT_STATUS_OK ;
}
static NTSTATUS gensec_gssapi_unwrap ( struct gensec_security * gensec_security ,
TALLOC_CTX * mem_ctx ,
const DATA_BLOB * in ,
DATA_BLOB * out )
{
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
OM_uint32 maj_stat , min_stat ;
gss_buffer_desc input_token , output_token ;
int conf_state ;
gss_qop_t qop_state ;
input_token . length = in - > length ;
input_token . value = in - > data ;
maj_stat = gss_unwrap ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
& input_token ,
& output_token ,
& conf_state ,
& qop_state ) ;
if ( GSS_ERROR ( maj_stat ) ) {
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
DEBUG ( 1 , ( " gensec_gssapi_unwrap: GSS UnWrap failed: %s \n " ,
2005-05-16 03:42:11 +04:00
gssapi_error_string ( mem_ctx , maj_stat , min_stat ) ) ) ;
2005-01-01 03:19:08 +03:00
return NT_STATUS_ACCESS_DENIED ;
}
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
* out = data_blob_talloc ( mem_ctx , output_token . value , output_token . length ) ;
2005-01-01 03:19:08 +03:00
gss_release_buffer ( & min_stat , & output_token ) ;
if ( gensec_have_feature ( gensec_security , GENSEC_FEATURE_SEAL )
& & ! conf_state ) {
return NT_STATUS_ACCESS_DENIED ;
}
return NT_STATUS_OK ;
}
2005-09-11 15:19:02 +04:00
static size_t gensec_gssapi_sig_size ( struct gensec_security * gensec_security , size_t data_size )
2005-05-11 16:03:48 +04:00
{
2005-09-11 15:19:02 +04:00
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
OM_uint32 maj_stat , min_stat ;
OM_uint32 output_size ;
if ( ( gensec_gssapi_state - > gss_oid - > length ! = gss_mech_krb5 - > length )
| | ( memcmp ( gensec_gssapi_state - > gss_oid - > elements , gss_mech_krb5 - > elements ,
gensec_gssapi_state - > gss_oid - > length ) ! = 0 ) ) {
DEBUG ( 1 , ( " NO sig size available for this mech \n " ) ) ;
return 0 ;
}
maj_stat = gsskrb5_wrap_size ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
gensec_have_feature ( gensec_security , GENSEC_FEATURE_SEAL ) ,
GSS_C_QOP_DEFAULT ,
data_size ,
& output_size ) ;
if ( GSS_ERROR ( maj_stat ) ) {
TALLOC_CTX * mem_ctx = talloc_new ( NULL ) ;
DEBUG ( 1 , ( " gensec_gssapi_seal_packet: determinaing signature size with gss_wrap_size_limit failed: %s \n " ,
gssapi_error_string ( mem_ctx , maj_stat , min_stat ) ) ) ;
talloc_free ( mem_ctx ) ;
return 0 ;
}
if ( output_size < data_size ) {
return 0 ;
}
/* The difference between the max output and the max input must be the signature */
return output_size - data_size ;
2005-05-11 16:03:48 +04:00
}
static NTSTATUS gensec_gssapi_seal_packet ( struct gensec_security * gensec_security ,
TALLOC_CTX * mem_ctx ,
uint8_t * data , size_t length ,
const uint8_t * whole_pdu , size_t pdu_length ,
DATA_BLOB * sig )
{
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
OM_uint32 maj_stat , min_stat ;
gss_buffer_desc input_token , output_token ;
int conf_state ;
2005-09-11 15:19:02 +04:00
ssize_t sig_length ;
2005-05-11 16:03:48 +04:00
input_token . length = length ;
input_token . value = data ;
maj_stat = gss_wrap ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
gensec_have_feature ( gensec_security , GENSEC_FEATURE_SEAL ) ,
GSS_C_QOP_DEFAULT ,
& input_token ,
& conf_state ,
& output_token ) ;
if ( GSS_ERROR ( maj_stat ) ) {
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
DEBUG ( 1 , ( " gensec_gssapi_seal_packet: GSS Wrap failed: %s \n " ,
2005-05-16 03:42:11 +04:00
gssapi_error_string ( mem_ctx , maj_stat , min_stat ) ) ) ;
2005-05-11 16:03:48 +04:00
return NT_STATUS_ACCESS_DENIED ;
}
2005-09-11 15:19:02 +04:00
sig_length = gensec_gssapi_sig_size ( gensec_security , length ) ;
/* Caller must pad to right boundary */
if ( output_token . length ! = ( length + sig_length ) ) {
2005-11-30 05:08:15 +03:00
DEBUG ( 1 , ( " gensec_gssapi_seal_packet: GSS Wrap length [%ld] does not match caller length [%ld] plus sig size [%ld] = [%ld] \n " ,
( long ) output_token . length , ( long ) length , ( long ) sig_length , ( long ) ( length + sig_length ) ) ) ;
2005-05-11 16:03:48 +04:00
return NT_STATUS_INTERNAL_ERROR ;
}
memcpy ( data , ( ( uint8_t * ) output_token . value ) + sig_length , length ) ;
* sig = data_blob_talloc ( mem_ctx , ( uint8_t * ) output_token . value , sig_length ) ;
2005-05-16 03:42:11 +04:00
dump_data_pw ( " gensec_gssapi_seal_packet: sig \n " , sig - > data , sig - > length ) ;
dump_data_pw ( " gensec_gssapi_seal_packet: clear \n " , data , length ) ;
dump_data_pw ( " gensec_gssapi_seal_packet: sealed \n " , ( ( uint8_t * ) output_token . value ) + sig_length , output_token . length - sig_length ) ;
2005-05-11 16:03:48 +04:00
gss_release_buffer ( & min_stat , & output_token ) ;
if ( gensec_have_feature ( gensec_security , GENSEC_FEATURE_SEAL )
& & ! conf_state ) {
return NT_STATUS_ACCESS_DENIED ;
}
return NT_STATUS_OK ;
}
static NTSTATUS gensec_gssapi_unseal_packet ( struct gensec_security * gensec_security ,
TALLOC_CTX * mem_ctx ,
uint8_t * data , size_t length ,
const uint8_t * whole_pdu , size_t pdu_length ,
const DATA_BLOB * sig )
{
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
OM_uint32 maj_stat , min_stat ;
gss_buffer_desc input_token , output_token ;
int conf_state ;
gss_qop_t qop_state ;
DATA_BLOB in ;
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
dump_data_pw ( " gensec_gssapi_unseal_packet: sig \n " , sig - > data , sig - > length ) ;
2005-05-11 16:03:48 +04:00
in = data_blob_talloc ( mem_ctx , NULL , sig - > length + length ) ;
memcpy ( in . data , sig - > data , sig - > length ) ;
memcpy ( in . data + sig - > length , data , length ) ;
input_token . length = in . length ;
input_token . value = in . data ;
maj_stat = gss_unwrap ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
& input_token ,
& output_token ,
& conf_state ,
& qop_state ) ;
if ( GSS_ERROR ( maj_stat ) ) {
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
DEBUG ( 1 , ( " gensec_gssapi_unseal_packet: GSS UnWrap failed: %s \n " ,
2005-05-16 03:42:11 +04:00
gssapi_error_string ( mem_ctx , maj_stat , min_stat ) ) ) ;
2005-05-11 16:03:48 +04:00
return NT_STATUS_ACCESS_DENIED ;
}
if ( output_token . length ! = length ) {
return NT_STATUS_INTERNAL_ERROR ;
}
memcpy ( data , output_token . value , length ) ;
gss_release_buffer ( & min_stat , & output_token ) ;
if ( gensec_have_feature ( gensec_security , GENSEC_FEATURE_SEAL )
& & ! conf_state ) {
return NT_STATUS_ACCESS_DENIED ;
}
return NT_STATUS_OK ;
}
static NTSTATUS gensec_gssapi_sign_packet ( struct gensec_security * gensec_security ,
TALLOC_CTX * mem_ctx ,
const uint8_t * data , size_t length ,
const uint8_t * whole_pdu , size_t pdu_length ,
DATA_BLOB * sig )
{
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
OM_uint32 maj_stat , min_stat ;
gss_buffer_desc input_token , output_token ;
int conf_state ;
ssize_t sig_length = 0 ;
input_token . length = length ;
input_token . value = discard_const_p ( uint8_t * , data ) ;
maj_stat = gss_wrap ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
0 ,
GSS_C_QOP_DEFAULT ,
& input_token ,
& conf_state ,
& output_token ) ;
if ( GSS_ERROR ( maj_stat ) ) {
2005-05-16 03:42:11 +04:00
DEBUG ( 1 , ( " GSS Wrap failed: %s \n " ,
gssapi_error_string ( mem_ctx , maj_stat , min_stat ) ) ) ;
2005-05-11 16:03:48 +04:00
return NT_STATUS_ACCESS_DENIED ;
}
if ( output_token . length < length ) {
return NT_STATUS_INTERNAL_ERROR ;
}
2005-09-11 15:19:02 +04:00
sig_length = gensec_gssapi_sig_size ( gensec_security , length ) ;
/* Caller must pad to right boundary */
if ( output_token . length ! = ( length + sig_length ) ) {
2005-11-30 05:08:15 +03:00
DEBUG ( 1 , ( " gensec_gssapi_sign_packet: GSS Wrap length [%ld] does not match caller length [%ld] plus sig size [%ld] = [%ld] \n " ,
( long ) output_token . length , ( long ) length , ( long ) sig_length , ( long ) ( length + sig_length ) ) ) ;
2005-09-11 15:19:02 +04:00
return NT_STATUS_INTERNAL_ERROR ;
}
2005-05-11 16:03:48 +04:00
* sig = data_blob_talloc ( mem_ctx , ( uint8_t * ) output_token . value , sig_length ) ;
2005-05-16 03:42:11 +04:00
dump_data_pw ( " gensec_gssapi_seal_packet: sig \n " , sig - > data , sig - > length ) ;
2005-05-11 16:03:48 +04:00
gss_release_buffer ( & min_stat , & output_token ) ;
return NT_STATUS_OK ;
}
static NTSTATUS gensec_gssapi_check_packet ( struct gensec_security * gensec_security ,
TALLOC_CTX * mem_ctx ,
const uint8_t * data , size_t length ,
const uint8_t * whole_pdu , size_t pdu_length ,
const DATA_BLOB * sig )
{
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
OM_uint32 maj_stat , min_stat ;
gss_buffer_desc input_token , output_token ;
int conf_state ;
gss_qop_t qop_state ;
DATA_BLOB in ;
2005-05-16 03:42:11 +04:00
dump_data_pw ( " gensec_gssapi_seal_packet: sig \n " , sig - > data , sig - > length ) ;
2005-05-11 16:03:48 +04:00
in = data_blob_talloc ( mem_ctx , NULL , sig - > length + length ) ;
memcpy ( in . data , sig - > data , sig - > length ) ;
memcpy ( in . data + sig - > length , data , length ) ;
input_token . length = in . length ;
input_token . value = in . data ;
maj_stat = gss_unwrap ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
& input_token ,
& output_token ,
& conf_state ,
& qop_state ) ;
if ( GSS_ERROR ( maj_stat ) ) {
2005-05-16 03:42:11 +04:00
DEBUG ( 1 , ( " GSS UnWrap failed: %s \n " ,
gssapi_error_string ( mem_ctx , maj_stat , min_stat ) ) ) ;
2005-05-11 16:03:48 +04:00
return NT_STATUS_ACCESS_DENIED ;
}
if ( output_token . length ! = length ) {
return NT_STATUS_INTERNAL_ERROR ;
}
gss_release_buffer ( & min_stat , & output_token ) ;
return NT_STATUS_OK ;
}
2005-01-01 03:19:08 +03:00
static BOOL gensec_gssapi_have_feature ( struct gensec_security * gensec_security ,
2005-02-10 08:22:53 +03:00
uint32_t feature )
2005-01-01 03:19:08 +03:00
{
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
if ( feature & GENSEC_FEATURE_SIGN ) {
return gensec_gssapi_state - > got_flags & GSS_C_INTEG_FLAG ;
}
if ( feature & GENSEC_FEATURE_SEAL ) {
return gensec_gssapi_state - > got_flags & GSS_C_CONF_FLAG ;
}
2005-05-11 16:03:48 +04:00
if ( feature & GENSEC_FEATURE_SESSION_KEY ) {
2005-05-16 05:31:22 +04:00
if ( ( gensec_gssapi_state - > gss_oid - > length = = gss_mech_krb5 - > length )
& & ( memcmp ( gensec_gssapi_state - > gss_oid - > elements , gss_mech_krb5 - > elements , gensec_gssapi_state - > gss_oid - > length ) = = 0 ) ) {
2005-05-11 16:03:48 +04:00
return True ;
}
2005-08-20 10:08:52 +04:00
}
if ( feature & GENSEC_FEATURE_DCE_STYLE ) {
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
return gensec_gssapi_state - > got_flags & GSS_C_DCE_STYLE ;
2005-08-20 10:08:52 +04:00
}
if ( feature & GENSEC_FEATURE_ASYNC_REPLIES ) {
return True ;
2005-05-12 06:07:53 +04:00
}
2005-01-01 03:19:08 +03:00
return False ;
}
2005-05-11 16:03:48 +04:00
static NTSTATUS gensec_gssapi_session_key ( struct gensec_security * gensec_security ,
DATA_BLOB * session_key )
{
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
if ( gensec_gssapi_state - > session_key . data ) {
* session_key = gensec_gssapi_state - > session_key ;
return NT_STATUS_OK ;
}
2005-05-16 03:42:11 +04:00
/* Ensure we only call this for GSSAPI/krb5, otherwise things could get very ugly */
2005-05-16 05:31:22 +04:00
if ( ( gensec_gssapi_state - > gss_oid - > length = = gss_mech_krb5 - > length )
& & ( memcmp ( gensec_gssapi_state - > gss_oid - > elements , gss_mech_krb5 - > elements ,
2005-05-16 03:42:11 +04:00
gensec_gssapi_state - > gss_oid - > length ) = = 0 ) ) {
2006-01-12 10:13:36 +03:00
OM_uint32 maj_stat ;
krb5_keyblock * skey ;
2005-05-11 16:03:48 +04:00
2006-01-12 10:13:36 +03:00
maj_stat = gss_krb5_get_subkey ( gensec_gssapi_state - > gssapi_context ,
& skey ) ;
2005-05-11 16:03:48 +04:00
if ( maj_stat = = 0 ) {
2005-07-17 13:20:52 +04:00
DEBUG ( 10 , ( " Got KRB5 session key of length %d \n " ,
2006-01-12 10:13:36 +03:00
( int ) KRB5_KEY_LENGTH ( skey ) ) ) ;
2005-05-11 16:03:48 +04:00
gensec_gssapi_state - > session_key = data_blob_talloc ( gensec_gssapi_state ,
2006-01-12 10:13:36 +03:00
KRB5_KEY_DATA ( skey ) , KRB5_KEY_LENGTH ( skey ) ) ;
2005-05-11 16:03:48 +04:00
* session_key = gensec_gssapi_state - > session_key ;
dump_data_pw ( " KRB5 Session Key: \n " , session_key - > data , session_key - > length ) ;
2006-01-12 10:13:36 +03:00
krb5_free_keyblock ( gensec_gssapi_state - > smb_krb5_context - > krb5_context , skey ) ;
2005-05-11 16:03:48 +04:00
return NT_STATUS_OK ;
}
return NT_STATUS_NO_USER_SESSION_KEY ;
}
DEBUG ( 1 , ( " NO session key for this mech \n " ) ) ;
return NT_STATUS_NO_USER_SESSION_KEY ;
}
static NTSTATUS gensec_gssapi_session_info ( struct gensec_security * gensec_security ,
struct auth_session_info * * _session_info )
{
NTSTATUS nt_status ;
2005-06-28 04:55:44 +04:00
TALLOC_CTX * mem_ctx ;
2005-05-11 16:03:48 +04:00
struct gensec_gssapi_state * gensec_gssapi_state = gensec_security - > private_data ;
struct auth_serversupplied_info * server_info = NULL ;
struct auth_session_info * session_info = NULL ;
2005-06-28 04:55:44 +04:00
struct PAC_LOGON_INFO * logon_info ;
2005-05-11 16:03:48 +04:00
OM_uint32 maj_stat , min_stat ;
gss_buffer_desc name_token ;
2005-06-28 04:55:44 +04:00
gss_buffer_desc pac ;
2005-06-28 12:27:50 +04:00
krb5_keyblock * keyblock ;
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
time_t authtime ;
krb5_principal principal ;
char * principal_string ;
2005-10-20 14:15:31 +04:00
DATA_BLOB pac_blob ;
2005-05-11 16:03:48 +04:00
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
if ( ( gensec_gssapi_state - > gss_oid - > length ! = gss_mech_krb5 - > length )
| | ( memcmp ( gensec_gssapi_state - > gss_oid - > elements , gss_mech_krb5 - > elements ,
gensec_gssapi_state - > gss_oid - > length ) ! = 0 ) ) {
DEBUG ( 1 , ( " NO session info available for this mech \n " ) ) ;
return NT_STATUS_INVALID_PARAMETER ;
}
2005-06-28 04:55:44 +04:00
mem_ctx = talloc_named ( gensec_gssapi_state , 0 , " gensec_gssapi_session_info context " ) ;
NT_STATUS_HAVE_NO_MEMORY ( mem_ctx ) ;
2005-05-11 16:03:48 +04:00
maj_stat = gss_display_name ( & min_stat ,
gensec_gssapi_state - > client_name ,
& name_token ,
NULL ) ;
if ( maj_stat ) {
return NT_STATUS_FOOBAR ;
}
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
principal_string = talloc_strndup ( mem_ctx , name_token . value , name_token . length ) ;
2005-05-11 16:03:48 +04:00
gss_release_buffer ( & min_stat , & name_token ) ;
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
if ( ! principal_string ) {
2005-06-28 04:55:44 +04:00
talloc_free ( mem_ctx ) ;
return NT_STATUS_NO_MEMORY ;
}
2005-05-11 16:03:48 +04:00
2005-06-28 12:27:50 +04:00
maj_stat = gss_krb5_copy_service_keyblock ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
& keyblock ) ;
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
if ( maj_stat = = 0 ) {
maj_stat = gsskrb5_extract_authtime_from_sec_context ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
& authtime ) ;
}
if ( maj_stat = = 0 ) {
maj_stat = gsskrb5_extract_authz_data_from_sec_context ( & min_stat ,
gensec_gssapi_state - > gssapi_context ,
2005-10-27 03:41:01 +04:00
KRB5_AUTHDATA_WIN2K_PAC ,
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
& pac ) ;
}
2005-10-20 14:15:31 +04:00
if ( maj_stat = = 0 ) {
pac_blob = data_blob_talloc ( mem_ctx , pac . value , pac . length ) ;
gss_release_buffer ( & min_stat , & pac ) ;
}
2005-06-28 04:55:44 +04:00
2005-10-24 02:20:42 +04:00
/* IF we have the PAC - otherwise we need to get this
* data from elsewere - local ldb , or ( TODO ) lookup of some
* kind . . .
*/
2005-06-28 04:55:44 +04:00
if ( maj_stat = = 0 ) {
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
krb5_error_code ret ;
ret = krb5_parse_name ( gensec_gssapi_state - > smb_krb5_context - > krb5_context ,
principal_string , & principal ) ;
if ( ret ) {
talloc_free ( mem_ctx ) ;
return NT_STATUS_INVALID_PARAMETER ;
}
2005-06-28 04:55:44 +04:00
/* decode and verify the pac */
2005-10-27 03:41:01 +04:00
nt_status = kerberos_pac_logon_info ( mem_ctx , & logon_info , pac_blob ,
2005-08-27 16:23:37 +04:00
gensec_gssapi_state - > smb_krb5_context - > krb5_context ,
2005-11-07 05:29:37 +03:00
NULL , keyblock , principal , authtime , NULL ) ;
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
krb5_free_principal ( gensec_gssapi_state - > smb_krb5_context - > krb5_context , principal ) ;
2005-06-28 04:55:44 +04:00
if ( NT_STATUS_IS_OK ( nt_status ) ) {
union netr_Validation validation ;
validation . sam3 = & logon_info - > info3 ;
nt_status = make_server_info_netlogon_validation ( gensec_gssapi_state ,
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
NULL ,
2005-06-28 04:55:44 +04:00
3 , & validation ,
& server_info ) ;
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
talloc_free ( mem_ctx ) ;
return nt_status ;
}
} else {
maj_stat = 1 ;
}
}
if ( maj_stat ) {
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
DEBUG ( 1 , ( " Unable to use PAC, resorting to local user lookup! \n " ) ) ;
2005-10-24 02:20:42 +04:00
nt_status = sam_get_server_info_principal ( mem_ctx , principal_string ,
& server_info ) ;
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
2005-06-28 04:55:44 +04:00
if ( ! NT_STATUS_IS_OK ( nt_status ) ) {
talloc_free ( mem_ctx ) ;
return nt_status ;
}
2005-05-11 16:03:48 +04:00
}
/* references the server_info into the session_info */
nt_status = auth_generate_session_info ( gensec_gssapi_state , server_info , & session_info ) ;
r10066: This is the second in my patches to work on Samba4's kerberos support,
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00a9c5318f4706db55d03f64e8bea60c)
2005-09-08 01:52:50 +04:00
talloc_free ( mem_ctx ) ;
2005-05-11 16:03:48 +04:00
talloc_free ( server_info ) ;
NT_STATUS_NOT_OK_RETURN ( nt_status ) ;
nt_status = gensec_gssapi_session_key ( gensec_security , & session_info - > session_key ) ;
NT_STATUS_NOT_OK_RETURN ( nt_status ) ;
2005-11-02 03:31:22 +03:00
if ( ! ( gensec_gssapi_state - > got_flags & GSS_C_DELEG_FLAG ) ) {
2005-11-05 09:38:47 +03:00
DEBUG ( 10 , ( " gensec_gssapi: NO delegated credentials supplied by client \n " ) ) ;
2005-11-02 03:31:22 +03:00
} else {
krb5_error_code ret ;
DEBUG ( 10 , ( " gensec_gssapi: delegated credentials supplied by client \n " ) ) ;
session_info - > credentials = cli_credentials_init ( session_info ) ;
if ( ! session_info - > credentials ) {
return NT_STATUS_NO_MEMORY ;
}
cli_credentials_set_conf ( session_info - > credentials ) ;
ret = cli_credentials_set_client_gss_creds ( session_info - > credentials ,
gensec_gssapi_state - > delegated_cred_handle ,
CRED_SPECIFIED ) ;
if ( ret ) {
return NT_STATUS_NO_MEMORY ;
}
/* It has been taken from this place... */
gensec_gssapi_state - > delegated_cred_handle = GSS_C_NO_CREDENTIAL ;
}
2005-05-11 16:03:48 +04:00
* _session_info = session_info ;
return NT_STATUS_OK ;
}
2005-09-24 07:31:01 +04:00
static const char * gensec_gssapi_krb5_oids [ ] = {
2005-05-16 03:42:11 +04:00
GENSEC_OID_KERBEROS5 ,
GENSEC_OID_KERBEROS5_OLD ,
NULL
2005-01-01 03:19:08 +03:00
} ;
2005-05-11 16:11:35 +04:00
/* As a server, this could in theory accept any GSSAPI mech */
2005-05-16 03:42:11 +04:00
static const struct gensec_security_ops gensec_gssapi_krb5_security_ops = {
. name = " gssapi_krb5 " ,
2005-11-05 14:02:37 +03:00
. sasl_name = " GSSAPI " ,
2005-08-23 20:13:39 +04:00
. auth_type = DCERPC_AUTH_TYPE_KRB5 ,
2005-09-24 07:31:01 +04:00
. oid = gensec_gssapi_krb5_oids ,
2005-05-11 16:11:35 +04:00
. client_start = gensec_gssapi_client_start ,
. server_start = gensec_gssapi_server_start ,
2005-06-22 06:12:26 +04:00
. magic = gensec_gssapi_magic ,
2005-05-11 16:11:35 +04:00
. update = gensec_gssapi_update ,
. session_key = gensec_gssapi_session_key ,
. session_info = gensec_gssapi_session_info ,
. sig_size = gensec_gssapi_sig_size ,
. sign_packet = gensec_gssapi_sign_packet ,
. check_packet = gensec_gssapi_check_packet ,
. seal_packet = gensec_gssapi_seal_packet ,
. unseal_packet = gensec_gssapi_unseal_packet ,
. wrap = gensec_gssapi_wrap ,
. unwrap = gensec_gssapi_unwrap ,
. have_feature = gensec_gssapi_have_feature ,
2005-09-21 01:29:29 +04:00
. enabled = True
2005-01-03 02:53:14 +03:00
} ;
2005-01-01 03:19:08 +03:00
NTSTATUS gensec_gssapi_init ( void )
{
NTSTATUS ret ;
2005-01-03 02:53:14 +03:00
ret = gensec_register ( & gensec_gssapi_krb5_security_ops ) ;
if ( ! NT_STATUS_IS_OK ( ret ) ) {
DEBUG ( 0 , ( " Failed to register '%s' gensec backend! \n " ,
gensec_gssapi_krb5_security_ops . name ) ) ;
return ret ;
}
2005-01-01 03:19:08 +03:00
return ret ;
}