GM_USE_MG_CRED allows a mechanism glue credential to be used by a mechanism
without additional wrapping. Although no extant mechanisms use this flag, the
flag had sporadic support in the mechanism glue already. In the interest of
consistency, add missing GM_USE_MG_CRED checks. If this functionality is not
desired, then all checks should be removed.
Add helper APIs to the mechanism glue from the Apple implementation.
gss_mg_export_name() can be used by a mechanism to create an exported name
(this is used by the future SAnon implementation). _gss_mech_import_name()
allows the mechanism to pass in a list of parsing functions for specific name
tiypes. This commit only adds support to the mechanism glue itself.
The GSS mechglue uses singly linked lists for mechanisms and mechanism objects,
to which new entries are inserted at the head. This breaks ordering of
mechanisms specified in OID sets and in /etc/gss/mech, as they will be back to
front. Use a tail queue instead so that new entries are inserted at the end.
Make mechanism interfaces that are optional in usage, optional in
implementation, i.e. do not require them to be present when dynamically loading
a mechanism. In order to aid minimal mechanism implementation, more of these
should likely be made optional (this will require an additional check in the
mechglue to return GSS_S_UNAVAILABLE if the function pointer is NULL). This is
left as a future exercise.
krb5_cc_cache_match() searches all ccache collections for a ccache that
has credentials for a given principal name. This includes MEMORY
ccaches, which means it can find the same ccache as is referenced by a
GSS cred handle given to gss_store_cred(), which means that
gss_store_cred() can fail.
For now we work around this by including a private variant of
krb5_cc_cache_match() that only searches the default ccache, not all
collections. Eventually we should ensure that krb5_cc_default() also
searches all collection-type (other than MEMORY) ccaches for a default
credential, then we can go back to using krb5_cc_cache_match() (though
we'll need to make sure that MEMORY is searched last or not at all).
gss_acquire_cred_from() is an exported function and therefore
must be tagged with GSSAPI_LIB_FUNCTION and GSSAPI_LIB_CALL.
Change-Id: I80918cb8083eaeac2d0eba5347f7b428e997cfaa
Implementation of gss_set_neg_mechs() and gss_get_neg_mechs() as defined in RFC
4178. New gss_release_cred_by_mech() API for dropping a credential from a
mechanism glue credential.
Apple's Heimdal impelmentation uses a number of utility functions for
allocating names and credentials, to avoid calling malloc or calloc directly.
Import them.
Implement the GSS-API credential store API extensions defined by MIT here:
https://k5wiki.kerberos.org/wiki/Projects/Credential_Store_extensions
Note: we kill off gss_acquire_cred_ext() here. This was never a public API,
although mechanisms could have implemented it and I briefly used it in my
BrowserID prototype mechanism. gss_acquire_cred_ext_from() occupies the place
in the dispatch table where gss_acquire_cred_ext() used to, but this structure
was never visible outside Heimdal (i.e. it is only used by internal
mechanisms);
(Mechanisms that need to accept arbitrary key/value dictionaries from
applications should now implement gss_acquire_cred_from().)
gss_krb5_set_allowable_enctypes() was declared with an array of int32_t types
representing the enctype list, but the definition had an array of krb5_enctype.
Whilst these are likely the same size, they may not be. On the receiving end,
allocate an array of krb5_enctype.
* check `ctx->gc_ctx` in `gss_inquire_context()`
* check `gm_inquire_cred != NULL` in `gss_inquire_context()`
* check `min_lifetime` in `gss_inquire_cred()`
* check `gm_inquire_cred_by_mech != NULL` in `gss_inquire_cred_by_mech()`
* set mech error in `gss_inquire_cred_by_oid()`
* don't clobber error in `gss_inquire_cred_by_oid()`
* don't pass NULL minor_status to `gss_krb5_free_lucid_sec_context()`
* allow NULL ccache in `gss_krb5_ccache_name()`
* NULL names OK in `_gss_find_mn()`
* allow empty names in `gss_import_name()` (removes `input_name_buffer` length
check). to support ANONYMOUS. in `gss_import_name()`, ignore mech name
import failure as long as it's possible to import the name in some other
mechanism
* better argument validation in `gss_export_sec_context()`
* in `gss_compare_name()`, check `mn2 != NULL`
* check `gss_add_oid_set_member()` return code in `gss_indicate_mechs()`
* in `gss_destroy_cred()`, set output cred handle to `GSS_C_NO_CREDENTIAL`
* cast size_t to OM_uint32 where required
In the case that one name has no MNs and the second name only has an MN, then
the order in which the two names are passed in matters, but it shouldn't. Run
the comparison both ways.
It turns out gss_add_cred() really needed a complete rewrite. It's much
better to first have a gss_duplicate_cred() (which has been needed for
other reasons anyways), and use that when the input_cred_handle is not
GSS_C_NO_CREDENTIAL and output_cred_handle is not NULL, then mutate that
duplicate credential handle (or the input_cred_handle if
output_cred_handle is NULL).
Make memcmp() compare the name1 and name2 value instead of comparing
name1 with itself.
The memcmp() is only executed if the left-hand side of the || is false
i.e. when both length are equal so the length argument is correct (no out-of-bounds reads).
restore correct OID for GSS_C_PEER_HAS_UPDATED_SPNEGO, this should have no
ABI implications, it's for internal use only. The current OID was incorrectly
copied in commit dbeeb18a, it should belong to 1.3.6.1.4.1.5322.19 which is
... enterprise(1) padl(5322) gssKrb5Extensions(19). The OID we were camping
on belongs to another party.
This is generated from lib/gssapi/oid.txt using lib/gssapi/gen-oid.pl,
which sorts the entries to ensure minimal diff churn when an oid is
added or changed.
The lack of effective changes can be seen by sorting both versions, a
bit like this:
$ git show HEAD~~:lib/gssapi/mech/gss_oid.c | sort > /tmp/gss_oid.c-OLD
$ cat lib/gssapi/mech/gss_oid.c | sort > /tmp/gss_oid.c-NEW
$ diff -u /tmp/gss_oid.c*
$ #Nothing to see!
This is of course not a reliable check in general, but works for this
simple file in concert with ordinary inspection.
Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
These missed out on the rk_UNCONST()ification by virtue of being added
in a parallel branch. In the diagram below, they got added in 02cf28e,
while the rk_UNCONSTs were added in f5f9014.
* cc47c8f Turn on -Wextra -Wno-sign-compare -Wno-unused-paramter and fix issues.
* 3069d80 Merge branch 'master' into lukeh/acquire-cred-ex
|\
| * f5f9014 Warning fixes from Christos Zoulas
* | 02cf28e implement gss_acquire_cred_ex with password support
|/
* 2170219 add more oids
rk_UNCONST amounts to a cast to (void *), removing const.
Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
gss_init_sec_context() with input_cred_handle != GSS_C_NO_CREDENTIAL
should NOT proceed if there is no element in the given credential for
the requested mechanism.
Solaris'/Illumos' gss_acquire_cred_with_password() does not have
side-effects. MIT and Heimdal have differed, but it's now agreed that
the Solaris/Illumos behavior is correct.
To make a credential obained with gss_acquire_cred_with_password()
available to other processes, use gss_store_cred().