Add an internal-use setter accessor for use by mssfu.c when principal names are
replaced. This also fixes a leak where r->client_princ was not freed before
being replaced with the impersonated client name.
As we move more and more state into astgs_request_t we need to be
careful not to leave locals and function arguments aliasing objects from
the astgs_request_t.
We weren't checking for errors, and we weren't using the convience
SEQUENCE OF add_METHOD_DATA() and add_ETYPE_INFO() functions generated
by the ASN.1 compiler. Using those utilities made get_pa_etype_info()
and make_etype_info_entry() sufficiently simpler that merging the two
makes for simpler, more readable code.
Build the entry to write to the log file in memory, the write it with
O_APPEND. This should make writes to the request log file safer in
multi-process KDC configurations.
Also, check results of krb5_store_*() functions (coverity alerted to
this).
We will use the kdc_ rather than krb5_kdc_ prefix for new public APIs exported
from libkdc. Amend the recently introduced
request_{get,set,copy,delete}_attribute APIs to conform.
If use_strongest_session_key and no keys are found the return
code will be set to KRB5KDC_ERR_ETYPE_NOSUPP. However, before
exiting if the principal is a weak exception then the return
code will be set to success and a NULL key pointer will be
dereferenced. Skip the weak exception check if the return
code is non-zero.
Change-Id: I36d0e1b179b3392171ad6767c75111d59c3f2f7d
Samba clients are often machine accounts with non-default salts that
will fail if they can't use the AES encryption type they know the KDC
supports. The problem is that arcfour-hmac-md5 has no salt so was
being used in preference.
Samba started to fail when
kdc_config->preauth_use_strongest_session_key = true;
was forced into the KDC configuration.
The history here is an attempt to avoid Kerberos v4 salts in des-cbc-crc
keys, but this instead broke Samba clients with AES-keys on machine accounts
as these have a non-default salt by default. These accounts were incorrectly
restricted to arcfour-hmac-md5 and they didn't like that.
A broader fix than Samba commit 8e1efd8bd3bf698dc0b6ed2081919f49b1412b53
REF: https://lists.samba.org/archive/samba/2021-October/237844.html
Samba BUG: https://bugzilla.samba.org/show_bug.cgi?id=14864
Change-Id: Ia8908a5a2eef107e6b133d7f0e4343c1988c18bb
Tighten up all of the call sites of hx509_request_get_san()
to free the output string returned upon failure.
Use frees(&s) instead of free(s); s = NULL;.
Change-Id: I71035d7c1d2330a1a3a1b3b730cdd6ba1e6b7da3
Change-Id: I4e872126f5c1deb48d36f5bdac22da7c0df3c8a1
warning: missing braces around initializer [-Wmissing-braces]
warning: (near initialization for 'fastreq.fast_options') [-Wmissing-braces]
warning: missing initializer for field 'padata' of 'struct KrbFastReq' [-Wmissing-field-initializers]
Now that ETYPE_xxx names are macros to the KRB5_ENCTYPE_xxx
symbols there is no longer a need to cast to (krb5_enctype).
Change-Id: Ib561f6e45346abac7a53eb1db0bfef60ee3bcb74
Remove hdb_entry_ex and revert to the original design of hdb_entry (except with
an additional context member in hdb_entry which is managed by the free_entry
method in HDB).
Decorate HDB_entry with context and move free_entry callback into HDB structure
itself. Requires updating hdb_free_entry() signature to include HDB parameter.
A follow-up commit will consolidate hdb_entry_ex (which has a single hdb_entry
member) into hdb_entry.
This is not a very good fix, though the warnings remain. Such errors
can in principle be a problem because in C there is no standard enum
sizing.
In this case we have two enums with the same elements and so the same
size, so it's clearly not a problem.
Recent auditing changes (b1dcc1a4) introduced a leak into pa_enc_ts_validate()
where the encryption type name and error message could be leaked.
This patch fixes a compile error in the original commit.
