diff options
author | Stephen Gallagher <sgallagh@redhat.com> | 2010-10-15 11:35:31 -0400 |
---|---|---|
committer | Stephen Gallagher <sgallagh@redhat.com> | 2010-10-18 13:12:04 -0400 |
commit | 55769ee01eac9ce8ce55b29222f14e1c4362fc3c (patch) | |
tree | bb98ca4f3863a5b1cebe60267f093abbc7695b1f | |
parent | d5d2af3a47753527e96cef47162eecedfb6b2b5b (diff) | |
download | sssd-55769ee01eac9ce8ce55b29222f14e1c4362fc3c.tar.gz sssd-55769ee01eac9ce8ce55b29222f14e1c4362fc3c.tar.bz2 sssd-55769ee01eac9ce8ce55b29222f14e1c4362fc3c.zip |
Handle nested groups in RFC2307bis
This first approach handles the non-optimized "pure" RFC2307bis
case. It recursively calls into nested groups until it it has
found them all or hits the pre-defined nesting limit.
It then saves all member users first, then all groups to the sysdb
-rw-r--r-- | src/providers/ldap/sdap_async_accounts.c | 777 |
1 files changed, 776 insertions, 1 deletions
diff --git a/src/providers/ldap/sdap_async_accounts.c b/src/providers/ldap/sdap_async_accounts.c index 1684af63..3ee8a4ea 100644 --- a/src/providers/ldap/sdap_async_accounts.c +++ b/src/providers/ldap/sdap_async_accounts.c @@ -1458,6 +1458,9 @@ struct sdap_get_groups_state { struct sysdb_attrs **groups; size_t count; size_t check_count; + + hash_table_t *user_hash; + hash_table_t *group_hash; }; static void sdap_get_groups_process(struct tevent_req *subreq); @@ -1504,6 +1507,15 @@ struct tevent_req *sdap_get_groups_send(TALLOC_CTX *memctx, return req; } +static struct tevent_req *sdap_nested_group_process_send( + TALLOC_CTX *mem_ctx, struct tevent_context *ev, + struct sss_domain_info *domain, + struct sysdb_ctx *sysdb, struct sysdb_attrs *group, + hash_table_t *users, hash_table_t *groups, + struct sdap_options *opts, struct sdap_handle *sh, + uint32_t nesting); +static void sdap_nested_done(struct tevent_req *req); +static errno_t sdap_nested_group_process_recv(struct tevent_req *req); static void sdap_get_groups_process(struct tevent_req *subreq) { struct tevent_req *req = @@ -1523,9 +1535,56 @@ static void sdap_get_groups_process(struct tevent_req *subreq) DEBUG(6, ("Search for groups, returned %d results.\n", state->count)); - if (state->count == 0) { + switch(state->count) { + case 0: tevent_req_error(req, ENOENT); return; + + case 1: + /* Single group search */ + if ((state->opts->schema_type == SDAP_SCHEMA_RFC2307) || + (dp_opt_get_int(state->opts->basic, SDAP_NESTING_LEVEL) == 0)) { + /* Either this is RFC2307 or we have disabled nested group + * support for RFC2307bis. Either way, we'll process the + * groups in single-level, multiple-request mode. + */ + break; + } + + /* Prepare hashes for nested user processing */ + ret = sss_hash_create(state, 32, &state->user_hash); + if (ret != EOK) { + tevent_req_error(req, ret); + return; + } + + ret = sss_hash_create(state, 32, &state->group_hash); + if (ret != EOK) { + tevent_req_error(req, ret); + return; + } + + subreq = sdap_nested_group_process_send(state, + state->ev, + state->dom, + state->sysdb, + state->groups[0], + state->user_hash, + state->group_hash, + state->opts, + state->sh, + 0); + if (!subreq) { + tevent_req_error(req, EIO); + return; + } + + tevent_req_set_callback(subreq, sdap_nested_done, req); + return; + + default: + /* Enumeration */ + break; } state->check_count = state->count; @@ -1594,6 +1653,89 @@ int sdap_get_groups_recv(struct tevent_req *req, return EOK; } +static void sdap_nested_done(struct tevent_req *subreq) +{ + errno_t ret; + int hret; + unsigned long i; + unsigned long count; + hash_value_t *values; + struct sysdb_attrs **users = NULL; + struct sysdb_attrs **groups = NULL; + struct tevent_req *req = tevent_req_callback_data(subreq, + struct tevent_req); + struct sdap_get_groups_state *state = tevent_req_data(req, + struct sdap_get_groups_state); + + ret = sdap_nested_group_process_recv(subreq); + talloc_zfree(subreq); + if (ret != EOK) { + DEBUG(1, ("Nested group processing failed: [%d][%s]\n", + ret, strerror(ret))); + tevent_req_error(req, ret); + return; + } + + hret = hash_values(state->user_hash, &count, &values); + if (hret != HASH_SUCCESS) { + tevent_req_error(req, EIO); + } + + if (count) { + users = talloc_array(state, struct sysdb_attrs *, count); + if (!users) { + talloc_free(values); + tevent_req_error(req, ENOMEM); + return; + } + + for (i = 0; i < count; i++) { + users[i] = talloc_get_type(values[i].ptr, struct sysdb_attrs); + } + talloc_zfree(values); + } + + /* Save all of the users first so that they are in + * place for the groups to add them. + */ + ret = sdap_save_users(state, state->sysdb, state->dom, state->opts, + users, count, &state->higher_timestamp); + if (ret != EOK) { + tevent_req_error(req, ret); + return; + } + + /* Users are all saved. Now save groups */ + hret = hash_values(state->group_hash, &count, &values); + if (hret != HASH_SUCCESS) { + tevent_req_error(req, EIO); + return; + } + + groups = talloc_array(state, struct sysdb_attrs *, count); + if (!groups) { + talloc_free(values); + tevent_req_error(req, ENOMEM); + return; + } + + for (i = 0; i < count; i++) { + groups[i] = talloc_get_type(values[i].ptr, struct sysdb_attrs); + } + talloc_zfree(values); + + ret = sdap_save_groups(state, state->sysdb, state->dom, state->opts, + groups, count, false, &state->higher_timestamp); + if (ret != EOK) { + tevent_req_error(req, ret); + return; + } + + /* Processing complete */ + tevent_req_done(req); +} + + /* ==Save-fake-group-list=====================================*/ static errno_t sdap_add_incomplete_groups(struct sysdb_ctx *sysdb, struct sss_domain_info *dom, @@ -2296,3 +2438,636 @@ int sdap_get_initgr_recv(struct tevent_req *req) return EOK; } +struct sdap_nested_group_ctx { + struct tevent_context *ev; + struct sysdb_ctx *sysdb; + struct sss_domain_info *domain; + + hash_table_t *users; + hash_table_t *groups; + + struct sdap_options *opts; + struct sdap_handle *sh; + + uint32_t nesting_level; + + struct ldb_message_element *members; + uint32_t member_index; + char *member_dn; +}; + +static errno_t sdap_nested_group_process_step(struct tevent_req *req); +static struct tevent_req *sdap_nested_group_process_send( + TALLOC_CTX *mem_ctx, struct tevent_context *ev, + struct sss_domain_info *domain, + struct sysdb_ctx *sysdb, struct sysdb_attrs *group, + hash_table_t *users, hash_table_t *groups, + struct sdap_options *opts, struct sdap_handle *sh, + uint32_t nesting) +{ + errno_t ret; + int hret; + struct tevent_req *req; + struct sdap_nested_group_ctx *state; + const char *groupname; + hash_key_t key; + hash_value_t value; + + req = tevent_req_create(mem_ctx, &state, struct sdap_nested_group_ctx); + if (!req) { + return NULL; + } + + state->ev = ev; + state->sysdb = sysdb; + state->domain = domain; + state->users = users; + state->groups = groups; + state->opts = opts; + state->sh = sh; + state->nesting_level = nesting; + + /* If this is too many levels deep, just return success */ + if (nesting > dp_opt_get_int(opts->basic, SDAP_NESTING_LEVEL)) { + ret = EOK; + goto immediate; + } + + /* Add the current group to the groups hash so we don't + * look it up more than once + */ + key.type = HASH_KEY_STRING; + + ret = sysdb_attrs_get_string( + group, + opts->group_map[SDAP_AT_GROUP_NAME].sys_name, + &groupname); + if (ret != EOK) goto immediate; + + key.str = talloc_strdup(state, groupname); + if (!key.str) { + ret = ENOMEM; + goto immediate; + } + + if (hash_has_key(groups, &key)) { + /* This group has already been processed + * (or is in progress) + * Skip it and just return success + */ + ret = EOK; + goto immediate; + } + + value.type = HASH_VALUE_PTR; + value.ptr = talloc_steal(groups, group); + + hret = hash_enter(groups, &key, &value); + if (hret != HASH_SUCCESS) { + ret = EIO; + goto immediate; + } + talloc_free(key.str); + + /* Process group memberships */ + + /* TODO: future enhancement, check for memberuid as well + * See https://fedorahosted.org/sssd/ticket/445 + */ + + ret = sysdb_attrs_get_el( + group, + opts->group_map[SDAP_AT_GROUP_MEMBER].sys_name, + &state->members); + if (ret != EOK) { + if (ret == ENOENT) { + /* No members to process */ + ret = EOK; + } + goto immediate; + } + + state->member_index = 0; + + ret = sdap_nested_group_process_step(req); + if (ret != EAGAIN) goto immediate; + + return req; + +immediate: + if (ret == EOK) { + tevent_req_done(req); + } else { + tevent_req_error(req, ret); + } + tevent_req_post(req, ev); + return req; +} + + +static void sdap_nested_group_process_ldap_user(struct tevent_req *subreq); +static void sdap_nested_group_process_user(struct tevent_req *subreq); +static errno_t sdap_nested_group_lookup_user(struct tevent_req *req, + tevent_req_fn fn); +static errno_t sdap_nested_group_lookup_group(struct tevent_req *req); +static errno_t sdap_nested_group_process_step(struct tevent_req *req) +{ + errno_t ret; + struct sdap_nested_group_ctx *state = + tevent_req_data(req, struct sdap_nested_group_ctx); + char *filter; + static const char *attrs[] = SYSDB_PW_ATTRS; + size_t count; + struct ldb_message **msgs; + uint64_t expiration; + bool has_key = false; + hash_key_t key; + uint8_t *data; + time_t now = time(NULL); + + while (true) { + /* Continue to loop through until all entries have been + * processed. + */ + do { + if (state->member_index >= state->members->num_values) { + /* No more entries to check. Return success */ + return EOK; + } + + /* First check whether this origDN is present (and not expired) + * in the sysdb + */ + data = state->members->values[state->member_index].data; + state->member_dn = talloc_strdup(state, (const char *)data); + if (!state->member_dn) { + ret = ENOMEM; + goto error; + } + + /* Check the user hash + * If it's there, we can save ourselves a trip to the + * sysdb and possibly LDAP as well + */ + key.type = HASH_KEY_STRING; + key.str = state->member_dn; + has_key = hash_has_key(state->users, &key); + if (has_key) { + talloc_zfree(state->member_dn); + state->member_index++; + continue; + } + + + } while (has_key); + + /* Check for the specified origDN in the sysdb */ + filter = talloc_asprintf(NULL, "(%s=%s)", + SYSDB_ORIG_DN, + state->member_dn); + if (!filter) { + ret = ENOMEM; + goto error; + } + + /* Try users first */ + ret = sysdb_search_users(state, state->sysdb, state->domain, filter, + attrs, &count, &msgs); + talloc_zfree(filter); + if (ret != EOK && ret != ENOENT) { + goto error; + } if (ret == ENOENT || count == 0) { + /* It wasn't a user. Check whether it's a group */ + if (ret == EOK) talloc_zfree(msgs); + + filter = talloc_asprintf(NULL, "(%s=%s)", + SYSDB_ORIG_DN, + state->member_dn); + if (!filter) { + ret = ENOMEM; + goto error; + } + ret = sysdb_search_groups(state, state->sysdb, state->domain, + filter, attrs, &count, &msgs); + talloc_zfree(filter); + if (ret != EOK && ret != ENOENT) { + ret = EIO; + goto error; + } else if (ret == ENOENT || count == 9) { + if (ret == EOK) talloc_zfree(msgs); + + /* It wasn't found in the groups either + * We'll have to do a blind lookup for both + */ + + /* Try users first */ + ret = sdap_nested_group_lookup_user( + req, sdap_nested_group_process_ldap_user); + if (ret != EOK) { + tevent_req_error(req, ret); + } + return EAGAIN; + } + + /* We found a group with this origDN in the sysdb */ + + /* Check whether the entry is valid */ + if (count != 1) { + DEBUG(1, ("More than one entry with this origDN? Skipping\n")); + state->member_index++; + talloc_zfree(state->member_dn); + continue; + } + + expiration = ldb_msg_find_attr_as_uint64(msgs[0], + SYSDB_CACHE_EXPIRE, + 0); + if (expiration && expiration > now) { + DEBUG(6, ("Cached values are still valid. Skipping\n")); + state->member_index++; + talloc_zfree(state->member_dn); + continue; + } + + /* Refresh the group from LDAP */ + ret = sdap_nested_group_lookup_group(req); + if (ret != EOK) goto error; + + return EAGAIN; + } + + /* We found a user with this origDN in the sysdb */ + + /* Check whether the entry is valid */ + if (count != 1) { + DEBUG(1, ("More than one entry with this origDN? Skipping\n")); + state->member_index++; + talloc_zfree(state->member_dn); + continue; + } + + expiration = ldb_msg_find_attr_as_uint64(msgs[0], + SYSDB_CACHE_EXPIRE, + 0); + if (expiration && expiration > now) { + DEBUG(6, ("Cached values are still valid. Skipping\n")); + state->member_index++; + talloc_zfree(state->member_dn); + continue; + } + + /* Refresh the user from LDAP */ + ret = sdap_nested_group_lookup_user( + req, sdap_nested_group_process_user); + if (ret != EOK) goto error; + + return EAGAIN; + } /* while (true) */ + +error: + talloc_zfree(state->member_dn); + return ret; +} + +static errno_t sdap_nested_group_lookup_user(struct tevent_req *req, + tevent_req_fn fn) +{ + errno_t ret; + const char **sdap_attrs; + char *filter; + struct tevent_req *subreq; + struct sdap_nested_group_ctx *state = + tevent_req_data(req, struct sdap_nested_group_ctx); + + ret = build_attrs_from_map(state, state->opts->user_map, + SDAP_OPTS_USER, &sdap_attrs); + if (ret != EOK) { + return ret; + } + + filter = talloc_asprintf( + sdap_attrs, "(objectclass=%s)", + state->opts->user_map[SDAP_OC_USER].name); + if (!filter) { + talloc_free(sdap_attrs); + return ENOMEM; + } + + subreq = sdap_get_generic_send(state, state->ev, state->opts, + state->sh, state->member_dn, + LDAP_SCOPE_BASE, + filter, sdap_attrs, + state->opts->user_map, + SDAP_OPTS_USER); + if (!subreq) { + talloc_free(sdap_attrs); + return EIO; + } + talloc_steal(subreq, sdap_attrs); + + tevent_req_set_callback(subreq, fn, req); + return EOK; +} + +static void sdap_nested_group_process_group(struct tevent_req *subreq); +static errno_t sdap_nested_group_lookup_group(struct tevent_req *req) +{ + errno_t ret; + const char **sdap_attrs; + char *filter; + struct tevent_req *subreq; + struct sdap_nested_group_ctx *state = + tevent_req_data(req, struct sdap_nested_group_ctx); + + ret = build_attrs_from_map(state, state->opts->group_map, + SDAP_OPTS_GROUP, &sdap_attrs); + if (ret != EOK) { + return ret; + } + + filter = talloc_asprintf( + sdap_attrs, "(objectclass=%s)", + state->opts->group_map[SDAP_OC_GROUP].name); + if (!filter) { + talloc_free(sdap_attrs); + return ENOMEM; + } + + subreq = sdap_get_generic_send(state, state->ev, state->opts, + state->sh, state->member_dn, + LDAP_SCOPE_BASE, + filter, sdap_attrs, + state->opts->group_map, + SDAP_OPTS_GROUP); + if (!subreq) { + talloc_free(sdap_attrs); + return EIO; + } + talloc_steal(subreq, sdap_attrs); + + tevent_req_set_callback(subreq, sdap_nested_group_process_group, req); + return EOK; +} + +static void sdap_nested_group_process_user(struct tevent_req *subreq) +{ + errno_t ret; + struct tevent_req *req = + tevent_req_callback_data(subreq, struct tevent_req); + struct sdap_nested_group_ctx *state = + tevent_req_data(req, struct sdap_nested_group_ctx); + TALLOC_CTX *tmp_ctx; + size_t count; + struct sysdb_attrs **replies; + int hret; + hash_key_t key; + hash_value_t value; + + tmp_ctx = talloc_new(NULL); + if (!tmp_ctx) { + tevent_req_error(req, ENOMEM); + return; + } + + ret = sdap_get_generic_recv(subreq, tmp_ctx, &count, &replies); + talloc_zfree(subreq); + if (ret != EOK && ret != ENOENT) { + tevent_req_error(req, ret); + goto done; + } else if (ret == ENOENT || count == 0) { + /* Nothing to do if the user doesn't exist */ + goto skip; + } + + if (count != 1) { + /* There should only ever be one reply for a + * BASE search. If otherwise, it's a serious + * error. + */ + DEBUG(1,("Received multiple replies for a BASE search!\n")); + tevent_req_error(req, EIO); + goto done; + } + + /* Save the user attributes to the user hash so we can store + * them all at once later. + */ + + key.type = HASH_KEY_STRING; + key.str = state->member_dn; + + value.type = HASH_VALUE_PTR; + value.ptr = replies[0]; + + hret = hash_enter(state->users, &key, &value); + if (hret != HASH_SUCCESS) { + tevent_req_error(req, EIO); + goto done; + } + talloc_steal(state->users, replies[0]); + +skip: + state->member_index++; + talloc_zfree(state->member_dn); + ret = sdap_nested_group_process_step(req); + if (ret == EOK) { + /* EOK means it's complete */ + tevent_req_done(req); + } else if (ret != EAGAIN) { + tevent_req_error(req, ret); + } + + /* EAGAIN means that we should re-enter + * the mainloop + */ + +done: + talloc_free(tmp_ctx); +} + +static void sdap_group_internal_nesting_done(struct tevent_req *subreq); +static void sdap_nested_group_process_group(struct tevent_req *subreq) +{ + errno_t ret; + struct tevent_req *req = + tevent_req_callback_data(subreq, struct tevent_req); + struct sdap_nested_group_ctx *state = + tevent_req_data(req, struct sdap_nested_group_ctx); + TALLOC_CTX *tmp_ctx; + size_t count; + struct sysdb_attrs **replies; + + tmp_ctx = talloc_new(NULL); + if (!tmp_ctx) { + tevent_req_error(req, ENOMEM); + return; + } + + ret = sdap_get_generic_recv(subreq, tmp_ctx, &count, &replies); + talloc_zfree(subreq); + if (ret != EOK && ret != ENOENT) { + tevent_req_error(req, ret); + goto done; + } else if (ret == ENOENT || count == 0) { + /* Nothing to do if the group doesn't exist */ + goto skip; + } + + if (count != 1) { + /* There should only ever be one reply for a + * BASE search. If otherwise, it's a serious + * error. + */ + DEBUG(1,("Received multiple replies for a BASE search!\n")); + tevent_req_error(req, EIO); + goto done; + } + + /* Recurse down into the member group */ + subreq = sdap_nested_group_process_send(state, state->ev, state->domain, + state->sysdb, replies[0], + state->users, state->groups, + state->opts, state->sh, + state->nesting_level + 1); + if (!subreq) { + tevent_req_error(req, EIO); + goto done; + } + tevent_req_set_callback(subreq, sdap_group_internal_nesting_done, req); + + talloc_free(tmp_ctx); + return; + +skip: + state->member_index++; + talloc_zfree(state->member_dn); + ret = sdap_nested_group_process_step(req); + if (ret == EOK) { + /* EOK means it's complete */ + tevent_req_done(req); + } else if (ret != EAGAIN) { + tevent_req_error(req, ret); + } + + /* EAGAIN means that we should re-enter + * the mainloop + */ + +done: + talloc_free(tmp_ctx); +} + +static void sdap_group_internal_nesting_done(struct tevent_req *subreq) +{ + errno_t ret; + struct tevent_req *req = + tevent_req_callback_data(subreq, struct tevent_req); + struct sdap_nested_group_ctx *state = + tevent_req_data(req, struct sdap_nested_group_ctx); + + ret = sdap_nested_group_process_recv(subreq); + talloc_zfree(subreq); + if (ret != EOK) { + tevent_req_error(req, ret); + } + + state->member_index++; + talloc_zfree(state->member_dn); + ret = sdap_nested_group_process_step(req); + if (ret == EOK) { + /* EOK means it's complete */ + tevent_req_done(req); + } else if (ret != EAGAIN) { + tevent_req_error(req, ret); + } + + /* EAGAIN means that we should re-enter + * the mainloop + */ +} + +static void sdap_nested_group_process_ldap_user(struct tevent_req *subreq) +{ + errno_t ret; + struct tevent_req *req = + tevent_req_callback_data(subreq, struct tevent_req); + struct sdap_nested_group_ctx *state = + tevent_req_data(req, struct sdap_nested_group_ctx); + TALLOC_CTX *tmp_ctx; + size_t count; + struct sysdb_attrs **replies; + int hret; + hash_key_t key; + hash_value_t value; + + tmp_ctx = talloc_new(NULL); + if (!tmp_ctx) { + tevent_req_error(req, ENOMEM); + return; + } + + ret = sdap_get_generic_recv(subreq, tmp_ctx, &count, &replies); + talloc_zfree(subreq); + if (ret != EOK && ret != ENOENT) { + tevent_req_error(req, ret); + goto done; + } else if (ret == ENOENT || count == 0) { + /* No user found. Assume it's a group */ + ret = sdap_nested_group_lookup_group(req); + if (ret != EOK) { + tevent_req_error(req, ret); + } + goto done; + } + + if (count != 1) { + /* There should only ever be one reply for a + * BASE search. If otherwise, it's a serious + * error. + */ + DEBUG(1,("Received multiple replies for a BASE search!\n")); + tevent_req_error(req, EIO); + goto done; + } + + /* Save the user attributes to the user hash so we can store + * them all at once later. + */ + key.type = HASH_KEY_STRING; + key.str = state->member_dn; + + value.type = HASH_VALUE_PTR; + value.ptr = replies[0]; + + hret = hash_enter(state->users, &key, &value); + if (hret != HASH_SUCCESS) { + tevent_req_error(req, EIO); + goto done; + } + talloc_steal(state->users, replies[0]); + + /* Move on to the next member */ + state->member_index++; + talloc_zfree(state->member_dn); + ret = sdap_nested_group_process_step(req); + if (ret == EOK) { + /* EOK means it's complete */ + tevent_req_done(req); + } else if (ret != EAGAIN) { + tevent_req_error(req, ret); + } + + /* EAGAIN means that we should re-enter + * the mainloop + */ + +done: + talloc_free(tmp_ctx); +} + +static errno_t sdap_nested_group_process_recv(struct tevent_req *req) +{ + TEVENT_REQ_RETURN_ON_ERROR(req); + + return EOK; +} |