are not returned. serverctrlsp This result parameter will be filled in with an allocated array of controls copied out of ref. The control array should be freed by calling ldap_controls_free(). If serverctrlsp is NULL, no controls are returned. freeit A boolean that determines whether or not the ref parameter is disposed of. Pass any non-zero value to have these functions free ref after extracting the requested information. This option is provided as a convenience; you can also use ldap_msgfree() to free the result later. The ldap_parse_reference() function returns an LDAP error code that indicates whether the reference could be successfully parsed (LDAP_SUCCESS if all goes well). 2 Encoded_ASN.1 LDAP contains functions that may be used to encode and decode BER-encoded ASN.1 values, which are often used inside of control and extension values. The following additional integral types are defined for use in manipulation of BER encoded ASN.1 values: typedef unsigned long ber_tag_t; /* for BER tags */ typedef long ber_int_t; /* for BER ints, enums, and Booleans */ With the exceptions of two new functions, ber_flatten() and ber_ init(), these functions are compatible with the University of Michigan LDAP 3.3 implementation of BER. typedef struct berval { ber_len_t bv_len; char *bv_val; } BerValue; A struct berval contains a sequence of bytes and an indication of its length. The bv_val is not null terminated. A bv_len must always be a nonnegative number. Applications may allocate their own berval structures. typedef struct berelement { /* opaque */ } BerElement; The BerElement structure contains not only a copy of the encoded value, but also state information used in encoding or decoding. Applications cannot allocate their own BerElement structures. The internal state is neither thread-specific nor locked, so two threads should not manipulate the same BerElement value simultaneously. A single BerElement value cannot be used for both encoding and decoding. void ber_bvfree( struct berval *bv ); The ber_bvfree() function frees a berval returned from this API. Both the bv->bv_val string and the berval itself are freed. Applications should not use ber_bvfree() with bervals which the application has allocated. void ber_bvecfree ( struct berval **bv ); The ber_bvecfree() function frees an array of bervals returned from this API. Each of the bervals in the array are freed using ber_bvfree(), then the array itself is freed. struct berval *ber_bvdup (struct berval *bv ); The ber_bvdup() function returns a copy of a berval. The bv_val field in the returned berval points to a different area of memory as the bv_val field in the argument berval. The null pointer is returned on error (for example, out of memory). void ber_free ( BerElement *ber, int fbuf ); The ber_free() function frees a BerElement which is returned from the API calls ber_alloc_t() or ber_init(). Each BerElement must be freed by the caller. The second argument fbuf should always be set to 1 to ensure that the internal buffer used by the BER functions is freed as well as the BerElement container itself. 3 Encoding The following is an example of encoding: BerElement *ber_alloc_t(int options); The ber_alloc_t() function constructs and returns BerElement. The null pointer is returned on error. The options field contains a bitwise-or of options which are to be used when generating the encoding of this BerElement. One option is defined and must always be supplied: #define LBER_USE_DER 0x01 When this option is present, lengths will always be encoded in the minimum number of octets. Note that this option does not cause values of sets and sequences to be rearranged in tag and byte order, so these functions are not sufficient for generating DER output as defined in X.509 and X.680. If the caller takes responsibility for ordering values of sets and sequences correctly, DER output as defined in X.509 and X.680 can be produced. Unrecognized option bits are ignored. The BerElement returned by ber_alloc_t() is initially empty. Calls to ber_printf() will append bytes to the end of the BerElement. int ber_printf(BerElement *ber, char *fmt, ... ) The ber_printf() function is used to encode a BER element in much the same way that sprintf() works. One important difference, though, is that state information is kept in the BER argument so that multiple calls can be made to ber_printf() to append to the end of the BER element. BER must be a pointer to a BerElement returned by ber_alloc_t(). The ber_printf() function interprets and formats its arguments according to the format string fmt. The ber_printf() function returns -1 if there is an error during encoding and a positive number if successful. As with sprintf(), each character in fmt refers to an argument to ber_printf(). The format string can contain the following format characters: t Tag. The next argument is a ber_tag_t specifying the tag to override the next element to be written to the ber. This works across calls. The value must contain the tag class, constructed bit, and tag value. The tag value must fit in a single octet (tag value is less than 32). For example, a tag of "[3]" for a constructed type is 0xA3. b Boolean. The next argument is a ber_int_t, containing either 0 for FALSE or 0xff for TRUE. A boolean element is output. If this format character is not preceded by the 't' format modifier, the tag 0x01 is used for the element. e Enumerated. The next argument is a ber_int_t, containing the enumerated value in the host's byte order. An enumerated element is output. If this format character is not preceded by the 't' format modifier, the tag 0x0A is used for the element. i Integer. The next argument is a ber_int_t, containing the integer in the host's byte order. An integer element is output. If this format character is not preceded by the 't' format modifier, the tag 0x02 is used for the element. B Bitstring. The next two arguments are a char * pointer to the start of the bitstring, followed by a ber_len_t containing the number of bits in the bitstring. A bitstring element is output, in primitive form. If this format character is not preceded by the 't' format modifier, the tag 0x03 is used for the element. n Null. No argument is required. An ASN.1 NULL element is output. If this format character is not preceded by the 't' format modifier, the tag 0x05 is used for the element. o Octet string. The next two arguments are a char *, followed by a ber_len_t with the length of the string. The string may contain null bytes and need not by zero-terminated. An octet string element is output, in primitive form. If this format character is not preceded by the 't' format modifier, the tag 0x04 is used for the element. s Octet string. The next argument is a char * pointing to a zero-terminated string. An octet string element in primitive form is output, which does not include the trailing '\0' byte. If this format character is not preceded by the 't' format modifier, the tag 0x04 is used for the element. v Several octet strings. The next argument is a char **, an array of char * pointers to zero-terminated strings. The last element in the array must be a null pointer. The octet strings do not include the leading SEQUENCE OF octet strings. The 't' format modifier cannot be used with this format character. V Several octet strings. A NULL-terminated array of struct berval *'s is supplied. Note that a construct like '{V}' is required to get an actual SEQUENCE OF octet strings. The 't' format modifier cannot be used with this format character. { Begin sequence. No argument is required. If this format character is not preceded by the 't' format modifier, the tag 0x30 is used. } End sequence. No argument is required. The 't' format modifier cannot be used with this format character. [ Begin set. No argument is required. If this format character is not preceded by the 't' format modifier, the tag 0x31 is used. ] End set. No argument is required. The 't' format modifier cannot be used with this format character. Each use of a '{' format character must be matched by a '}' character, either later in the format string, or in the format string of a subsequent call to ber_printf() for that BerElement. The same applies to the '[' and ']'. Sequences and sets nest, and implementations of this API must maintain internal state to be able to properly calculate the lengths. int ber_flatten (BerElement *ber, struct berval **bvPtr); The ber_flatten() function allocates a struct berval whose contents are a BER encoding taken from the ber argument. The bvPtr pointer points to the returned berval, which must be freed using ber_bvfree(). This function returns 0 on success and -1 on error. The ber_flatten() API call is not present in U-M LDAP 3.3. The use of ber_flatten() on a BerElement in which all '{' and '}' format modifiers have not been properly matched is an error (that is, -1 will be returned by ber_flatten() if this situation is exists). 3 Decoding The following two symbols are available to applications. #define LBER_ERROR 0xffffffffL #define LBER_DEFAULT 0xffffffffL BerElement *ber_init (struct berval *bv); The ber_init() function constructs a BerElement and returns a new BerElement containing a copy of the data in the bv argument. The ber_init() function returns the null pointer on error. ber_tag_t ber_scanf (BerElement *ber, char *fmt, ... ); The ber_scanf() function is used to decode a BER element in much the same way that sscanf() works. One important difference, though, is that some state information is kept with the ber argument so that multiple calls can be made to ber_scanf() to sequentially read from the BER element. The ber argument must be a pointer to a BerElement returned by ber_init(). The ber_ scanf() function interprets function the bytes according to the format string fmt, and stores the results in its additional arguments. The ber_scanf() function returns LBER_ERROR on error, and a different value on success. The format string contains conversion specifications which are used to direct the interpretation of the BER element. The format string can contain the following characters: a Octet string. A char ** argument should be supplied. Memory is allocated, filled with the contents of the octet string, null- terminated, and the pointer to the string is stored in the argument. The returned value must be freed using ldap_memfree(). The tag of the element must indicate the primitive form (constructed strings are not supported) but is otherwise ignored and discarded during the decoding. This format cannot be used with octet strings which could contain null bytes. O Octet string. A struct berval ** argument should be supplied, which upon return points to a allocated struct berval containing the octet string and its length. The ber_bvfree() function must be called to free the allocated memory. The tag of the element must indicate the primitive form (constructed strings are not supported) but is otherwise ignored during the decoding. b Boolean. A pointer to a ber_int_t should be supplied. The value stored will be 0 for FALSE or nonzero for TRUE. The tag of the element must indicate the primitive form but is otherwise ignored during the decoding. e Enumerated value stored will be in host byte order. The tag of the element must indicate the primitive form but is otherwise ignored during the decoding. The ber_scanf() function will return an error if the enumerated value cannot be stored in a ber_int_t. i Integer. A pointer to a ber_int_t should be supplied. The value stored will be in host byte order. The tag of the element must indicate the primitive form but is otherwise ignored during the decoding. The ber_scanf() function will return an error if the integer cannot be stored in a ber_ int_t. B Bitstring. A char ** argument should be supplied which will point to the allocated bits, followed by a ber_len_t * argument, which will point to the length (in bits) of the bit-string returned. The ldap_memfree() function must be called to free the bit-string. The tag of the element must indicate the primitive form (constructed bitstrings are not supported) but is otherwise ignored during the decoding. n Null. No argument is required. The element is simply skipped if it is recognized as a zero-length element. The tag is ignored. v Several octet strings. A char *** argument should be supplied, which upon return points to a allocated null- terminated array of char *'s containing the octet strings. NULL is stored if the sequence is empty. The ldap_ memfree() function must be called to free each element of the array and the array itself. The tag of the sequence and of the octet strings are ignored. V Several octet strings (which could contain null bytes). A struct berval *** should be supplied, which upon return points to a allocated null-terminated array of struct berval *'s containing the octet strings and their lengths. NULL is stored if the sequence is empty. The ber_ bvecfree() function can be called to free the allocated memory. The tag of the sequence and of the octet strings are ignored. x Skip element. The next element is skipped. No argument is required. { Begin sequence. No argument is required. The initial sequence tag and length are skipped. } End sequence. No argument is required. [ Begin set. No argument is required. The initial set tag and length are skipped. ] End set. No argument is required. ber_tag_t ber_peek_tag (BerElement *ber, ber_len_t *lenPtr); The ber_peek_tag() function returns the tag of the next element to be parsed in the BerElement argument. The length of this element is stored in the *lenPtr argument. LBER_DEFAULT is returned if there is no further data to be read. The ber argument is not modified. ber_tag_t ber_skip_tag (BerElement *ber, ber_len_t *lenPtr); The ber_skip_tag() function is similar to ber_peek_tag(), except that the state pointer in the BerElement argument is advanced past the first tag and length, and is pointed to the value part of the next element. This function should only be used with constructed types and situations when a BER encoding is used as the value of an OCTET STRING. The length of the value is stored in *lenPtr. ber_tag_t ber_first_element(BerElement *ber, ber_len_t *lenPtr, char **opaquePtr); ber_tag_t ber_next_element (BerElement *ber, ber_len_t *lenPtr, char *opaque); The ber_first_element() and ber_next_element() functions are used to traverse a SET, SET OF, SEQUENCE or SEQUENCE OF data value. The ber_first_element() function calls ber_skip_tag(), stores internal information in *lenPtr and *opaquePtr, and calls ber_peek_tag() for the first element inside the constructed value. LBER_DEFAULT is returned if the constructed value is empty. The ber_next_element() function positions the state at the start of the next element in the constructed type. LBER_DEFAULT is returned if there are no further values. The len and opaque values should not be used by applications other than as arguments to ber_next_element(). (Refer to the VSI OpenVMS Utility Routines Manual for an example of this usage.) 2 Using_SSL Secure Sockets Layer (SSL) is the open standard security protocol for the secure transfer of sensitive information over the Internet. You can establish SSL on an LDAP session if the server supports such sessions. SSL uses X.509 public key technology to provide the following security functions: o Integrity and confidentiality of the LDAP dialog This is the most common use of SSL. The bytes sent over the wire are encrypted. o Authentication of the client Some servers use SSL to authenticate the client and make access control decisions based on the client identity. In this case, the client must have access to its private key and its certificate. The client certificate subject is a DN. o Authentication of the server It might be important for the client to verify the identity of the server to which it is talking. In this case, the client must have access to the appropriate certification authority (CA) public keys. There are several versions of SSL: SSLv2 (2.0), SSLv3 (3.0), and TLSv1 (3.1). TLS is the latest Internet standard. It does not require the use of RSA algorithms. Usually the client specifies the highest version it supports, and the server negotiates downward, if necessary. The client library supports all the versions listed here. You can establish SSL over LDAP two different ways: o LDAPS This older, de facto standard uses a separate TCP/IP port (usually 636) specifically for SSL over LDAP. In this case, the second parameter to the ldap_tls_start() function must be set to zero. o StartTLS This proposed Internet standard uses a regular LDAP port (usually 389) and requires the client to request the use of SSL. In this case, the second parameter to the ldap_tls_ start() function must be set to 1. 3 SSL_Certificate_Options The following session-handle options are specific to SSL and can be set by the ldap_set_option() function: o LDAP_OPT_TLS_CERT_REQUIRED (0x7001) void * Set to LDAP_OPT_ON if the client library requires a server certificate to be present the next time the ldap_tls_start() function is called. The default value is LDAP_OPT_OFF; a server certificate is not required. o LDAP_OPT_TLS_VERIFY_REQUIRED (0x7002) void * Set to LDAP_OPT_ON if the client library requires that a server certificate path be validated the next time the ldap_ tls_start() function is called. The default value is LDAP_OPT_ OFF; the server certificate, if any, is not verified. o LDAP_OPT_TLS_CERT_FILE (0x7003) char * Set to the name of a file containing the client's certificate for use by the ldap_tls_start() function. o LDAP_OPT_TLS_PKEY_FILE (0x7004) char * Set to the name of a file containing the client's private key for use by the ldap_tls_start() function. o LDAP_OPT_TLS_CA_FILE (0x7005) char * Set to the name of a file containing CA public keys used for validation of the server by the ldap_tls_start() function. o LDAP_OPT_TLS_CA_PATH (0x7006) char * Set to the name of a directory on disk containing CA public key files used for validation of the server by the ldap_tls_ start() function. o LDAP_OPT_TLS_VERSION (0x7007) int * Set to the desired SSL protocol version. This option takes one of the following values: 1: TLSv1 only 20: SSLv2 only 23: SSLv2 or SSLv3 30: SSLv3 only (default) 31: TLSv1 only If LDAP_OPT_TLS_VERIFY_REQUIRED is set to ON, either the LDAP_ OPT_TLS_CA_FILE or the LDAP_OPT_TLS_CA_PATH option must be set. If client authentication is required, both LDAP_OPT_TLS_CERT_FILE and LDAP_OPT_TLS_PKEY_FILE must be set. 3 Obtaining_a_Key_Pair In order for TLS to authenticate a client, the client must have a private key and a certificate. Obtain these from either a Certification Authority or a self-sign program. A self-sign program is included in the Open Source Security for OpenVMS product.