1
0
mirror of https://github.com/samba-team/samba.git synced 2024-12-29 11:21:54 +03:00
samba-mirror/source3/sam/idmap_ldap.c

791 lines
22 KiB
C
Raw Normal View History

/*
Unix SMB/CIFS implementation.
idmap LDAP backend
Copyright (C) Tim Potter 2000
Copyright (C) Jim McDonough <jmcd@us.ibm.com> 2003
Copyright (C) Simo Sorce 2003
Copyright (C) Gerald Carter 2003
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
#include "includes.h"
#undef DBGC_CLASS
#define DBGC_CLASS DBGC_IDMAP
#include <lber.h>
#include <ldap.h>
#include "smbldap.h"
struct ldap_idmap_state {
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
struct smbldap_state *smbldap_state;
TALLOC_CTX *mem_ctx;
};
static struct ldap_idmap_state ldap_state;
/* number tries while allocating new id */
#define LDAP_MAX_ALLOC_ID 128
/***********************************************************************
This function cannot be called to modify a mapping, only set a new one
***********************************************************************/
static NTSTATUS ldap_set_mapping(const DOM_SID *sid, unid_t id, int id_type)
{
pstring dn;
pstring id_str;
fstring type;
LDAPMod **mods = NULL;
int rc = -1;
int ldap_op;
fstring sid_string;
LDAPMessage *entry = NULL;
sid_to_string( sid_string, sid );
ldap_op = LDAP_MOD_ADD;
pstr_sprintf(dn, "%s=%s,%s", get_attr_key2string( sidmap_attr_list, LDAP_ATTR_SID),
sid_string, lp_ldap_idmap_suffix());
if ( id_type & ID_USERID )
fstrcpy( type, get_attr_key2string( sidmap_attr_list, LDAP_ATTR_UIDNUMBER ) );
else
fstrcpy( type, get_attr_key2string( sidmap_attr_list, LDAP_ATTR_GIDNUMBER ) );
pstr_sprintf(id_str, "%lu", ((id_type & ID_USERID) ? (unsigned long)id.uid :
(unsigned long)id.gid));
smbldap_set_mod( &mods, LDAP_MOD_ADD, "objectClass", LDAP_OBJ_IDMAP_ENTRY );
smbldap_make_mod( ldap_state.smbldap_state->ldap_struct,
entry, &mods, type, id_str );
smbldap_make_mod( ldap_state.smbldap_state->ldap_struct,
entry, &mods,
get_attr_key2string(sidmap_attr_list, LDAP_ATTR_SID),
sid_string );
/* There may well be nothing at all to do */
if (mods) {
smbldap_set_mod( &mods, LDAP_MOD_ADD, "objectClass", LDAP_OBJ_SID_ENTRY );
rc = smbldap_add(ldap_state.smbldap_state, dn, mods);
ldap_mods_free( mods, True );
} else {
rc = LDAP_SUCCESS;
}
if (rc != LDAP_SUCCESS) {
char *ld_error = NULL;
ldap_get_option(ldap_state.smbldap_state->ldap_struct, LDAP_OPT_ERROR_STRING,
&ld_error);
DEBUG(0,("ldap_set_mapping_internals: Failed to %s mapping from %s to %lu [%s]\n",
(ldap_op == LDAP_MOD_ADD) ? "add" : "replace",
sid_string, (unsigned long)((id_type & ID_USERID) ? id.uid : id.gid), type));
DEBUG(0, ("ldap_set_mapping_internals: Error was: %s (%s)\n",
ld_error ? ld_error : "(NULL)", ldap_err2string (rc)));
return NT_STATUS_UNSUCCESSFUL;
}
DEBUG(10,("ldap_set_mapping: Successfully created mapping from %s to %lu [%s]\n",
sid_string, ((id_type & ID_USERID) ? (unsigned long)id.uid :
(unsigned long)id.gid), type));
return NT_STATUS_OK;
}
/**********************************************************************
Even if the sambaDomain attribute in LDAP tells us that this RID is
safe to use, always check before use.
*********************************************************************/
static BOOL sid_in_use(struct ldap_idmap_state *state,
const DOM_SID *sid, int *error)
{
fstring filter;
fstring sid_string;
LDAPMessage *result = NULL;
int count;
int rc;
char *sid_attr[] = {LDAP_ATTRIBUTE_SID, NULL};
slprintf(filter, sizeof(filter)-1, "(%s=%s)", LDAP_ATTRIBUTE_SID, sid_to_string(sid_string, sid));
rc = smbldap_search_suffix(state->smbldap_state,
filter, sid_attr, &result);
if (rc != LDAP_SUCCESS) {
char *ld_error = NULL;
ldap_get_option(state->smbldap_state->ldap_struct, LDAP_OPT_ERROR_STRING, &ld_error);
DEBUG(2, ("Failed to check if sid %s is alredy in use: %s\n",
sid_string, ld_error));
SAFE_FREE(ld_error);
*error = rc;
return True;
}
if ((count = ldap_count_entries(state->smbldap_state->ldap_struct, result)) > 0) {
DEBUG(3, ("Sid %s already in use - trying next RID\n",
sid_string));
ldap_msgfree(result);
return True;
}
ldap_msgfree(result);
/* good, sid is not in use */
return False;
}
/**********************************************************************
Set the new nextRid attribute, and return one we can use.
This also checks that this RID is actually free - in case the admin
manually stole it :-).
*********************************************************************/
static NTSTATUS ldap_next_rid(struct ldap_idmap_state *state, uint32 *rid,
int rid_type)
{
NTSTATUS ret = NT_STATUS_UNSUCCESSFUL;
int rc;
LDAPMessage *domain_result = NULL;
LDAPMessage *entry = NULL;
char *dn;
LDAPMod **mods = NULL;
fstring old_rid_string;
fstring next_rid_string;
fstring algorithmic_rid_base_string;
uint32 next_rid;
uint32 alg_rid_base;
int attempts = 0;
char *ld_error = NULL;
while (attempts < 10) {
if (!NT_STATUS_IS_OK(ret = smbldap_search_domain_info(state->smbldap_state,
&domain_result, get_global_sam_name(), True))) {
return ret;
}
entry = ldap_first_entry(state->smbldap_state->ldap_struct, domain_result);
if (!entry) {
DEBUG(0, ("Could not get domain info entry\n"));
ldap_msgfree(domain_result);
return ret;
}
if ((dn = smbldap_get_dn(state->smbldap_state->ldap_struct, entry)) == NULL) {
DEBUG(0, ("Could not get domain info DN\n"));
ldap_msgfree(domain_result);
return ret;
}
/* yes, we keep 3 seperate counters, one for rids between 1000 (BASE_RID) and
algorithmic_rid_base. The other two are to avoid stomping on the
different sets of algorithmic RIDs */
if (smbldap_get_single_attribute(state->smbldap_state->ldap_struct, entry,
get_attr_key2string(dominfo_attr_list, LDAP_ATTR_ALGORITHMIC_RID_BASE),
algorithmic_rid_base_string)) {
alg_rid_base = (uint32)atol(algorithmic_rid_base_string);
} else {
alg_rid_base = algorithmic_rid_base();
/* Try to make the modification atomically by enforcing the
old value in the delete mod. */
slprintf(algorithmic_rid_base_string, sizeof(algorithmic_rid_base_string)-1, "%d", alg_rid_base);
smbldap_make_mod(state->smbldap_state->ldap_struct, entry, &mods,
get_attr_key2string(dominfo_attr_list, LDAP_ATTR_ALGORITHMIC_RID_BASE),
algorithmic_rid_base_string);
}
next_rid = 0;
if (alg_rid_base > BASE_RID) {
/* we have a non-default 'algorithmic rid base', so we have 'low' rids that we
can allocate to new users */
if (smbldap_get_single_attribute(state->smbldap_state->ldap_struct, entry,
get_attr_key2string(dominfo_attr_list, LDAP_ATTR_NEXT_RID),
old_rid_string)) {
*rid = (uint32)atol(old_rid_string);
} else {
*rid = BASE_RID;
}
next_rid = *rid+1;
if (next_rid >= alg_rid_base) {
ldap_msgfree(domain_result);
return NT_STATUS_UNSUCCESSFUL;
}
slprintf(next_rid_string, sizeof(next_rid_string)-1, "%d", next_rid);
/* Try to make the modification atomically by enforcing the
old value in the delete mod. */
smbldap_make_mod(state->smbldap_state->ldap_struct, entry, &mods,
get_attr_key2string(dominfo_attr_list, LDAP_ATTR_NEXT_RID),
next_rid_string);
}
if (!next_rid) { /* not got one already */
switch (rid_type) {
case USER_RID_TYPE:
if (smbldap_get_single_attribute(state->smbldap_state->ldap_struct, entry,
get_attr_key2string(dominfo_attr_list, LDAP_ATTR_NEXT_USERRID),
old_rid_string)) {
*rid = (uint32)atol(old_rid_string);
}
break;
case GROUP_RID_TYPE:
if (smbldap_get_single_attribute(state->smbldap_state->ldap_struct, entry,
get_attr_key2string(dominfo_attr_list, LDAP_ATTR_NEXT_GROUPRID),
old_rid_string)) {
*rid = (uint32)atol(old_rid_string);
}
break;
}
/* This is the core of the whole routine. If we had
scheme-style closures, there would be a *lot* less code
duplication... */
next_rid = *rid+RID_MULTIPLIER;
slprintf(next_rid_string, sizeof(next_rid_string)-1, "%d", next_rid);
switch (rid_type) {
case USER_RID_TYPE:
/* Try to make the modification atomically by enforcing the
old value in the delete mod. */
smbldap_make_mod(state->smbldap_state->ldap_struct, entry, &mods,
get_attr_key2string(dominfo_attr_list, LDAP_ATTR_NEXT_USERRID),
next_rid_string);
break;
case GROUP_RID_TYPE:
/* Try to make the modification atomically by enforcing the
old value in the delete mod. */
smbldap_make_mod(state->smbldap_state->ldap_struct, entry, &mods,
get_attr_key2string(dominfo_attr_list, LDAP_ATTR_NEXT_GROUPRID),
next_rid_string);
break;
}
}
if ((rc = smbldap_modify(state->smbldap_state, dn, mods)) == LDAP_SUCCESS) {
DOM_SID dom_sid;
DOM_SID sid;
pstring domain_sid_string;
int error = 0;
if (!smbldap_get_single_attribute(state->smbldap_state->ldap_struct, domain_result,
get_attr_key2string(dominfo_attr_list, LDAP_ATTR_DOM_SID),
domain_sid_string)) {
ldap_mods_free(mods, True);
SAFE_FREE(dn);
ldap_msgfree(domain_result);
return ret;
}
if (!string_to_sid(&dom_sid, domain_sid_string)) {
ldap_mods_free(mods, True);
SAFE_FREE(dn);
ldap_msgfree(domain_result);
return ret;
}
ldap_mods_free(mods, True);
mods = NULL;
SAFE_FREE(dn);
ldap_msgfree(domain_result);
sid_copy(&sid, &dom_sid);
sid_append_rid(&sid, *rid);
/* check RID is not in use */
if (sid_in_use(state, &sid, &error)) {
if (error) {
return ret;
}
continue;
}
return NT_STATUS_OK;
}
ld_error = NULL;
ldap_get_option(state->smbldap_state->ldap_struct, LDAP_OPT_ERROR_STRING, &ld_error);
DEBUG(2, ("Failed to modify rid: %s\n", ld_error ? ld_error : "(NULL"));
SAFE_FREE(ld_error);
ldap_mods_free(mods, True);
mods = NULL;
SAFE_FREE(dn);
ldap_msgfree(domain_result);
domain_result = NULL;
{
/* Sleep for a random timeout */
unsigned sleeptime = (sys_random()*sys_getpid()*attempts);
attempts += 1;
sleeptime %= 100;
smb_msleep(sleeptime);
}
}
DEBUG(0, ("Failed to set new RID\n"));
return ret;
}
/*****************************************************************************
Allocate a new RID
*****************************************************************************/
static NTSTATUS ldap_allocate_rid(uint32 *rid, int rid_type)
{
return ldap_next_rid( &ldap_state, rid, rid_type );
}
/*****************************************************************************
Allocate a new uid or gid
*****************************************************************************/
static NTSTATUS ldap_allocate_id(unid_t *id, int id_type)
{
NTSTATUS ret = NT_STATUS_UNSUCCESSFUL;
int rc = LDAP_SERVER_DOWN;
int count = 0;
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
LDAPMessage *result = NULL;
LDAPMessage *entry = NULL;
pstring id_str, new_id_str;
LDAPMod **mods = NULL;
const char *type;
char *dn = NULL;
char **attr_list;
pstring filter;
uid_t luid, huid;
gid_t lgid, hgid;
type = (id_type & ID_USERID) ?
get_attr_key2string( idpool_attr_list, LDAP_ATTR_UIDNUMBER ) :
get_attr_key2string( idpool_attr_list, LDAP_ATTR_GIDNUMBER );
pstr_sprintf(filter, "(objectClass=%s)", LDAP_OBJ_IDPOOL);
attr_list = get_attr_list( idpool_attr_list );
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
rc = smbldap_search(ldap_state.smbldap_state, lp_ldap_idmap_suffix(),
LDAP_SCOPE_SUBTREE, filter,
attr_list, 0, &result);
free_attr_list( attr_list );
if (rc != LDAP_SUCCESS) {
DEBUG(0,("ldap_allocate_id: %s object not found\n", LDAP_OBJ_IDPOOL));
goto out;
}
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
count = ldap_count_entries(ldap_state.smbldap_state->ldap_struct, result);
if (count != 1) {
DEBUG(0,("ldap_allocate_id: single %s object not found\n", LDAP_OBJ_IDPOOL));
goto out;
}
dn = smbldap_get_dn(ldap_state.smbldap_state->ldap_struct, result);
if (!dn) {
goto out;
}
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
entry = ldap_first_entry(ldap_state.smbldap_state->ldap_struct, result);
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
if (!smbldap_get_single_attribute(ldap_state.smbldap_state->ldap_struct, entry, type, id_str)) {
DEBUG(0,("ldap_allocate_id: %s attribute not found\n",
type));
goto out;
}
/* this must succeed or else we wouldn't have initialized */
lp_idmap_uid( &luid, &huid);
lp_idmap_gid( &lgid, &hgid);
/* make sure we still have room to grow */
if (id_type & ID_USERID) {
id->uid = strtoul(id_str, NULL, 10);
if (id->uid > huid ) {
DEBUG(0,("ldap_allocate_id: Cannot allocate uid above %lu!\n",
(unsigned long)huid));
goto out;
}
}
else {
id->gid = strtoul(id_str, NULL, 10);
if (id->gid > hgid ) {
DEBUG(0,("ldap_allocate_id: Cannot allocate gid above %lu!\n",
(unsigned long)hgid));
goto out;
}
}
pstr_sprintf(new_id_str, "%lu",
((id_type & ID_USERID) ? (unsigned long)id->uid :
(unsigned long)id->gid) + 1);
smbldap_set_mod( &mods, LDAP_MOD_DELETE, type, id_str );
smbldap_set_mod( &mods, LDAP_MOD_ADD, type, new_id_str );
if (mods == NULL) {
DEBUG(0,("ldap_allocate_id: smbldap_set_mod() failed.\n"));
goto out;
}
rc = smbldap_modify(ldap_state.smbldap_state, dn, mods);
ldap_mods_free( mods, True );
if (rc != LDAP_SUCCESS) {
DEBUG(0,("ldap_allocate_id: Failed to allocate new %s. ldap_modify() failed.\n",
type));
goto out;
}
ret = NT_STATUS_OK;
out:
SAFE_FREE(dn);
if (result != NULL)
ldap_msgfree(result);
return ret;
}
/*****************************************************************************
get a sid from an id
*****************************************************************************/
static NTSTATUS ldap_get_sid_from_id(DOM_SID *sid, unid_t id, int id_type)
{
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
LDAPMessage *result = NULL;
LDAPMessage *entry = NULL;
pstring sid_str;
pstring filter;
pstring suffix;
const char *type;
int rc;
int count;
NTSTATUS ret = NT_STATUS_UNSUCCESSFUL;
char **attr_list;
if ( id_type & ID_USERID )
type = get_attr_key2string( idpool_attr_list, LDAP_ATTR_UIDNUMBER );
else
type = get_attr_key2string( idpool_attr_list, LDAP_ATTR_GIDNUMBER );
pstrcpy( suffix, lp_ldap_idmap_suffix() );
pstr_sprintf(filter, "(&(objectClass=%s)(%s=%lu))",
LDAP_OBJ_IDMAP_ENTRY, type,
((id_type & ID_USERID) ? (unsigned long)id.uid : (unsigned long)id.gid));
attr_list = get_attr_list( sidmap_attr_list );
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
rc = smbldap_search(ldap_state.smbldap_state, suffix, LDAP_SCOPE_SUBTREE,
filter, attr_list, 0, &result);
if (rc != LDAP_SUCCESS) {
DEBUG(3,("ldap_get_isd_from_id: Failure looking up entry (%s)\n",
ldap_err2string(rc) ));
goto out;
}
count = ldap_count_entries(ldap_state.smbldap_state->ldap_struct, result);
if (count != 1) {
DEBUG(0,("ldap_get_sid_from_id: mapping not found for %s: %lu\n",
type, ((id_type & ID_USERID) ? (unsigned long)id.uid :
(unsigned long)id.gid)));
goto out;
}
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
entry = ldap_first_entry(ldap_state.smbldap_state->ldap_struct, result);
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
if ( !smbldap_get_single_attribute(ldap_state.smbldap_state->ldap_struct, entry, LDAP_ATTRIBUTE_SID, sid_str) )
goto out;
if (!string_to_sid(sid, sid_str))
goto out;
ret = NT_STATUS_OK;
out:
free_attr_list( attr_list );
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
if (result)
ldap_msgfree(result);
return ret;
}
/***********************************************************************
Get an id from a sid
***********************************************************************/
static NTSTATUS ldap_get_id_from_sid(unid_t *id, int *id_type, const DOM_SID *sid)
{
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
LDAPMessage *result = NULL;
LDAPMessage *entry = NULL;
pstring sid_str;
pstring filter;
pstring id_str;
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
const char *suffix;
const char *type;
int rc;
int count;
char **attr_list;
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
char *dn = NULL;
NTSTATUS ret = NT_STATUS_UNSUCCESSFUL;
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
sid_to_string(sid_str, sid);
DEBUG(8,("ldap_get_id_from_sid: %s (%s)\n", sid_str,
(*id_type & ID_GROUPID ? "group" : "user") ));
suffix = lp_ldap_idmap_suffix();
pstr_sprintf(filter, "(&(objectClass=%s)(%s=%s))",
LDAP_OBJ_IDMAP_ENTRY, LDAP_ATTRIBUTE_SID, sid_str);
if ( *id_type & ID_GROUPID )
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
type = get_attr_key2string( sidmap_attr_list, LDAP_ATTR_GIDNUMBER );
else
type = get_attr_key2string( sidmap_attr_list, LDAP_ATTR_UIDNUMBER );
/* do the search and check for errors */
attr_list = get_attr_list( sidmap_attr_list );
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
rc = smbldap_search(ldap_state.smbldap_state, suffix, LDAP_SCOPE_SUBTREE,
filter, attr_list, 0, &result);
if (rc != LDAP_SUCCESS) {
DEBUG(3,("ldap_get_id_from_sid: Failure looking up idmap entry (%s)\n",
ldap_err2string(rc) ));
goto out;
}
/* check for the number of entries returned */
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
count = ldap_count_entries(ldap_state.smbldap_state->ldap_struct, result);
if ( count > 1 ) {
DEBUG(0, ("ldap_get_id_from_sid: (2nd) search %s returned [%d] entries!\n",
filter, count));
goto out;
}
/* try to allocate a new id if we still haven't found one */
if ( !count ) {
int i;
if (*id_type & ID_QUERY_ONLY) {
DEBUG(5,("ldap_get_id_from_sid: No matching entry found and QUERY_ONLY flag set\n"));
goto out;
}
DEBUG(8,("ldap_get_id_from_sid: Allocating new id\n"));
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
for (i = 0; i < LDAP_MAX_ALLOC_ID; i++) {
ret = ldap_allocate_id(id, *id_type);
if ( NT_STATUS_IS_OK(ret) )
break;
}
if ( !NT_STATUS_IS_OK(ret) ) {
DEBUG(0,("ldap_allocate_id: cannot acquire id lock!\n"));
goto out;
}
DEBUG(10,("ldap_get_id_from_sid: Allocated new %cid [%ul]\n",
(*id_type & ID_GROUPID ? 'g' : 'u'), (uint32)id->uid ));
ret = ldap_set_mapping(sid, *id, *id_type);
/* all done */
goto out;
}
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
DEBUG(10,("ldap_get_id_from_sid: success\n"));
entry = ldap_first_entry(ldap_state.smbldap_state->ldap_struct, result);
dn = smbldap_get_dn(ldap_state.smbldap_state->ldap_struct, result);
if (!dn)
goto out;
DEBUG(10, ("Found mapping entry at dn=%s, looking for %s\n", dn, type));
if ( smbldap_get_single_attribute(ldap_state.smbldap_state->ldap_struct, entry, type, id_str) ) {
if ( (*id_type & ID_USERID) )
id->uid = strtoul(id_str, NULL, 10);
else
id->gid = strtoul(id_str, NULL, 10);
ret = NT_STATUS_OK;
goto out;
}
out:
free_attr_list( attr_list );
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
if (result)
ldap_msgfree(result);
SAFE_FREE(dn);
return ret;
}
/**********************************************************************
Verify the sambaUnixIdPool entry in the directiry.
**********************************************************************/
static NTSTATUS verify_idpool( void )
{
fstring filter;
int rc;
char **attr_list;
LDAPMessage *result = NULL;
LDAPMod **mods = NULL;
int count;
fstr_sprintf( filter, "(objectclass=%s)", LDAP_OBJ_IDPOOL );
attr_list = get_attr_list( idpool_attr_list );
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
rc = smbldap_search(ldap_state.smbldap_state, lp_ldap_idmap_suffix(),
LDAP_SCOPE_SUBTREE, filter, attr_list, 0, &result);
free_attr_list ( attr_list );
if (rc != LDAP_SUCCESS)
return NT_STATUS_UNSUCCESSFUL;
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
count = ldap_count_entries(ldap_state.smbldap_state->ldap_struct, result);
ldap_msgfree(result);
if ( count > 1 ) {
DEBUG(0,("ldap_idmap_init: multiple entries returned from %s (base == %s)\n",
filter, lp_ldap_idmap_suffix() ));
return NT_STATUS_UNSUCCESSFUL;
}
else if (count == 0) {
uid_t luid, huid;
gid_t lgid, hgid;
fstring uid_str, gid_str;
if ( !lp_idmap_uid(&luid, &huid) || !lp_idmap_gid( &lgid, &hgid ) ) {
DEBUG(0,("ldap_idmap_init: idmap uid/gid parameters not specified\n"));
return NT_STATUS_UNSUCCESSFUL;
}
fstr_sprintf( uid_str, "%lu", (unsigned long)luid );
fstr_sprintf( gid_str, "%lu", (unsigned long)lgid );
smbldap_set_mod( &mods, LDAP_MOD_ADD, "objectClass", LDAP_OBJ_IDPOOL );
smbldap_set_mod( &mods, LDAP_MOD_ADD,
get_attr_key2string(idpool_attr_list, LDAP_ATTR_UIDNUMBER), uid_str );
smbldap_set_mod( &mods, LDAP_MOD_ADD,
get_attr_key2string(idpool_attr_list, LDAP_ATTR_GIDNUMBER), gid_str );
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
rc = smbldap_modify(ldap_state.smbldap_state, lp_ldap_idmap_suffix(), mods);
}
return ( rc==LDAP_SUCCESS ? NT_STATUS_OK : NT_STATUS_UNSUCCESSFUL );
}
/*****************************************************************************
Initialise idmap database.
*****************************************************************************/
static NTSTATUS ldap_idmap_init( char *params )
{
NTSTATUS nt_status;
ldap_state.mem_ctx = talloc_init("idmap_ldap");
if (!ldap_state.mem_ctx) {
return NT_STATUS_NO_MEMORY;
}
/* assume location is the only parameter */
if (!NT_STATUS_IS_OK(nt_status =
smbldap_init(ldap_state.mem_ctx, params,
&ldap_state.smbldap_state))) {
talloc_destroy(ldap_state.mem_ctx);
return nt_status;
}
/* see if the idmap suffix and sub entries exists */
nt_status = verify_idpool();
if ( !NT_STATUS_IS_OK(nt_status) )
return nt_status;
return NT_STATUS_OK;
}
/*****************************************************************************
End the LDAP session
*****************************************************************************/
static NTSTATUS ldap_idmap_close(void)
{
This patch cleans up some of our ldap code, for better behaviour: We now always read the Domain SID out of LDAP. If the local secrets.tdb is ever different to LDAP, it is overwritten out of LDAP. We also store the 'algorithmic rid base' into LDAP, and assert if it changes. (This ensures cross-host synchronisation, and allows for possible integration with idmap). If we fail to read/add the domain entry, we just fallback to the old behaviour. We always use an existing DN when adding IDMAP entries to LDAP, unless no suitable entry is available. This means that a user's posixAccount will have a SID added to it, or a user's sambaSamAccount will have a UID added. Where we cannot us an existing DN, we use 'sambaSid=S-x-y-z,....' as the DN. The code now allows modifications to the ID mapping in many cases. Likewise, we now check more carefully when adding new user entires to LDAP, to not duplicate SIDs (for users, at this stage), and to add the sambaSamAccount onto the idmap entry for that user, if it is already established (ensuring we do not duplicate sambaSid entries in the directory). The allocated UID code has been expanded to take into account the space between '1000 - algorithmic rid base'. This much better fits into what an NT4 does - allocating in the bottom part of the RID range. On the code cleanup side of things, we now share as much code as possible between idmap_ldap and pdb_ldap. We also no longer use the race-prone 'enumerate all users' method for finding the next RID to allocate. Instead, we just start at the bottom of the range, and increment again if the user already exists. The first time this is run, it may well take a long time, but next time will just be able to use the next Rid. Thanks to metze and AB for double-checking parts of this. Andrew Bartlett (This used to be commit 9c595c8c2327b92a86901d84c3f2c284dabd597e)
2003-07-04 17:29:42 +04:00
smbldap_free_struct(&(ldap_state).smbldap_state);
talloc_destroy(ldap_state.mem_ctx);
DEBUG(5,("The connection to the LDAP server was closed\n"));
/* maybe free the results here --metze */
return NT_STATUS_OK;
}
/* This function doesn't make as much sense in an LDAP world since the calling
node doesn't really control the ID ranges */
static void ldap_idmap_status(void)
{
DEBUG(0, ("LDAP IDMAP Status not available\n"));
}
static struct idmap_methods ldap_methods = {
ldap_idmap_init,
ldap_allocate_rid,
ldap_allocate_id,
ldap_get_sid_from_id,
ldap_get_id_from_sid,
ldap_set_mapping,
ldap_idmap_close,
ldap_idmap_status
};
NTSTATUS idmap_ldap_init(void)
{
return smb_register_idmap(SMB_IDMAP_INTERFACE_VERSION, "ldap", &ldap_methods);
}