proxy_certificates.txt

开源的ssl算法openssl,版本0.9.8H

<DRAFT!>
			HOWTO proxy certificates

0. WARNING

NONE OF THE CODE PRESENTED HERE HAVE BEEN CHECKED!  They are just an
example to show you how things can be done.  There may be typos or
type conflicts, and you will have to resolve them.

1. Introduction

Proxy certificates are defined in RFC 3820.  They are really usual
certificates with the mandatory extension proxyCertInfo.

Proxy certificates are issued by an End Entity (typically a user),
either directly with the EE certificate as issuing certificate, or by
extension through an already issued proxy certificate..  They are used
to extend rights to some other entity (a computer process, typically,
or sometimes to the user itself), so it can perform operations in the
name of the owner of the EE certificate.

See http://www.ietf.org/rfc/rfc3820.txt for more information.


2. A warning about proxy certificates

Noone seems to have tested proxy certificates with security in mind.
Basically, to this date, it seems that proxy certificates have only
been used in a world that's highly aware of them.  What would happen
if an unsuspecting application is to validate a chain of certificates
that contains proxy certificates?  It would usually consider the leaf
to be the certificate to check for authorisation data, and since proxy
certificates are controlled by the EE certificate owner alone, it's
would be normal to consider what the EE certificate owner could do
with them.

subjectAltName and issuerAltName are forbidden in proxy certificates,
and this is enforced in OpenSSL.  The subject must be the same as the
issuer, with one commonName added on.

Possible threats are, as far as has been imagined so far:

 - impersonation through commonName (think server certificates).
 - use of additional extensions, possibly non-standard ones used in
   certain environments, that would grant extra or different
   authorisation rights.

For this reason, OpenSSL requires that the use of proxy certificates
be explicitely allowed.  Currently, this can be done using the
following methods:

 - if the application calls X509_verify_cert() itself, it can do the
   following prior to that call (ctx is the pointer passed in the call
   to X509_verify_cert()):

	X509_STORE_CTX_set_flags(ctx, X509_V_FLAG_ALLOW_PROXY_CERTS);

 - in all other cases, proxy certificate validation can be enabled
   before starting the application by setting the envirnoment variable
   OPENSSL_ALLOW_PROXY with some non-empty value.

There are thoughts to allow proxy certificates with a line in the
default openssl.cnf, but that's still in the future.


3. How to create proxy cerificates

It's quite easy to create proxy certificates, by taking advantage of
the lack of checks of the 'openssl x509' application (*ahem*).  But
first, you need to create a configuration section that contains a
definition of the proxyCertInfo extension, a little like this:

  [ v3_proxy ]
  # A proxy certificate MUST NEVER be a CA certificate.
  basicConstraints=CA:FALSE

  # Usual authority key ID
  authorityKeyIdentifier=keyid,issuer:always

  # Now, for the extension that marks this certificate as a proxy one
  proxyCertInfo=critical,language:id-ppl-anyLanguage,pathlen:1,policy:text:AB

It's also possible to give the proxy extension in a separate section:

  proxyCertInfo=critical,@proxy_ext

  [ proxy_ext ]
  language=id-ppl-anyLanguage
  pathlen=0
  policy=text:BC

The policy value has a specific syntax, {syntag}:{string}, where the
syntag determines what will be done with the string.  The recognised
syntags are as follows:

  text	indicates that the string is simply the bytes, not
	encoded in any kind of way:

		policy=text:r鋕sm鰎g錽

	Previous versions of this design had a specific tag
	for UTF-8 text.  However, since the bytes are copied
	as-is anyway, there's no need for it.  Instead, use
	the text: tag, like this:

		policy=text:r盲ksm枚rg氓s

  hex	indicates the string is encoded in hex, with colons
	between each byte (every second hex digit):

		policy=hex:72:E4:6B:73:6D:F6:72:67:E5:73

	Previous versions of this design had a tag to insert a
	complete DER blob.  However, the only legal use for
	this would be to surround the bytes that would go with
	the hex: tag with what's needed to construct a correct
	OCTET STRING.  Since hex: does that, the DER tag felt
	superfluous, and was therefore removed.

  file	indicates that the text of the policy should really be
	taken from a file.  The string is then really a file
	name.  This is useful for policies that are large
	(more than a few of lines) XML documents, for example.

The 'policy' setting can be split up in multiple lines like this:

  0.policy=This is
  1.polisy= a multi-
  2.policy=line policy.

NOTE: the proxy policy value is the part that determines the rights
granted to the process using the proxy certificate.  The value is
completely dependent on the application reading and interpretting it!

Now that you have created an extension section for your proxy
certificate, you can now easily create a proxy certificate like this:

  openssl req -new -config openssl.cnf \
	  -out proxy.req -keyout proxy.key
  openssl x509 -req -CAcreateserial -in proxy.req -days 7 \
	  -out proxy.crt -CA user.crt -CAkey user.key \
	  -extfile openssl.cnf -extensions v3_proxy

It's just as easy to create a proxy certificate using another proxy
certificate as issuer (note that I'm using a different configuration
section for it):

  openssl req -new -config openssl.cnf \
	  -out proxy2.req -keyout proxy2.key
  openssl x509 -req -CAcreateserial -in proxy2.req -days 7 \
	  -out proxy2.crt -CA proxy.crt -CAkey proxy.key \
	  -extfile openssl.cnf -extensions v3_proxy2


4. How to have your application interpret the policy?

The basic way to interpret proxy policies is to prepare some default
rights, then do a check of the proxy certificate against the a chain
of proxy certificates, user certificate and CA certificates, and see
what rights came out by the end.  Sounds easy, huh?  It almost is.

The slightly complicated part is how to pass data between your
application and the certificate validation procedure.

You need the following ingredients:

 - a callback routing that will be called for every certificate that's
   validated.  It will be called several times for each certificates,
   so you must be attentive to when it's a good time to do the proxy
   policy interpretation and check, as well as to fill in the defaults
   when the EE certificate is checked.

 - a structure of data that's shared between your application code and
   the callback.

 - a wrapper function that sets it all up.

 - an ex_data index function that creates an index into the generic
   ex_data store that's attached to an X509 validation context.

This is some cookbook code for you to fill in:

  /* In this example, I will use a view of granted rights as a bit
     array, one bit for each possible right.  */
  typedef struct your_rights {
    unsigned char rights[total_rights / 8];
  } YOUR_RIGHTS;

  /* The following procedure will create an index for the ex_data
     store in the X509 validation context the first time it's called.
     Subsequent calls will return the same index.  */
  static int get_proxy_auth_ex_data_idx(void)
  {
    static volatile int idx = -1;
    if (idx < 0)
      {
        CRYPTO_w_lock(CRYPTO_LOCK_X509_STORE);
        if (idx < 0)
          {
            idx = X509_STORE_CTX_get_ex_new_index(0,
                                                  "for verify callback",
                                                  NULL,NULL,NULL);
          }
        CRYPTO_w_unlock(CRYPTO_LOCK_X509_STORE);
      }
    return idx;
  }

  /* Callback to be given to the X509 validation procedure.  */
  static int verify_callback(int ok, X509_STORE_CTX *ctx)
  {
    if (ok == 1) /* It's REALLY important you keep the proxy policy
                    check within this secion.  It's important to know
                    that when ok is 1, the certificates are checked
                    from top to bottom.  You get the CA root first,
                    followed by the possible chain of intermediate
                    CAs, followed by the EE certificate, followed by
                    the possible proxy certificates.  */
      {
        X509 *xs = ctx->current_cert;

        if (xs->ex_flags & EXFLAG_PROXY)
          {
	    YOUR_RIGHTS *rights =
              (YOUR_RIGHTS *)X509_STORE_CTX_get_ex_data(ctx,
                get_proxy_auth_ex_data_idx());
            PROXY_CERT_INFO_EXTENSION *pci =
              X509_get_ext_d2i(xs, NID_proxyCertInfo, NULL, NULL);

            switch (OBJ_obj2nid(pci->proxyPolicy->policyLanguage))
              {
              case NID_Independent:
                /* Do whatever you need to grant explicit rights to
                   this particular proxy certificate, usually by
                   pulling them from some database.  If there are none
                   to be found, clear all rights (making this and any
                   subsequent proxy certificate void of any rights).
                */
                memset(rights->rights, 0, sizeof(rights->rights));
                break;
              case NID_id_ppl_inheritAll:
                /* This is basically a NOP, we simply let the current
                   rights stand as they are. */
                break;
              default:
                /* This is usually the most complex section of code.
                   You really do whatever you want as long as you
                   follow RFC 3820.  In the example we use here, the
                   simplest thing to do is to build another, temporary
                   bit array and fill it with the rights granted by
                   the current proxy certificate, then use it as a
                   mask on the accumulated rights bit array, and
                   voil