2f1a370cd3
hack for gss-wrap-iov to it work
...
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-28 13:31:12 -07:00
f030b4e59a
free context
2009-08-27 18:30:29 -07:00
9a4e91b1de
don't reset handle twice
2009-08-27 18:30:28 -07:00
74538fc2af
Plug memory leak in prf function
2009-08-27 18:30:28 -07:00
6c3f3fafa3
Don't leak kerberos credentials when trying dns canon
2009-08-27 18:30:28 -07:00
45cfe3f971
Fix server context client context order to match callee
2009-08-27 18:30:28 -07:00
1999c85670
Make mech glue layer aware of composite mechs that uses mech glue layer credentials
...
This make it possible to use krb5/ntlm credentials with SPNEGO.
Needs some more work to avoid double fetching credentials.
2009-08-27 12:12:44 -07:00
32ee735d73
drop RCSID
2009-08-26 23:15:35 -07:00
a2820df666
spelling
2009-08-26 22:53:38 -07:00
d18cdee577
don't reset EC
2009-08-26 22:52:26 -07:00
ebb2e72c61
make error message more unique
2009-08-26 22:43:25 -07:00
022e7d4319
Return unwrapped delegated credentials if the actual mech is not the called mech
...
Assumes that pseudo mechs are are of how mechglue credentails look like and
return credentials like that.
Pointed out on krbdev by Nicolas Williams
2009-08-26 22:32:50 -07:00
559103b218
if not trailer set, init EC to 0
2009-08-26 21:40:07 -07:00
03998aeccb
gsskrb5: fix test_context. after gss_wrap_iov changes
...
metze
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-25 23:34:38 -07:00
40a6abd116
gsskrb5: make the check for dcestyle and conf_req_flag == 0 more explicit
...
metze
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-25 23:34:38 -07:00
560cb0c132
gsskrb5: fix ec and padding handling in _gssapi_unwrap_cfx_iov()
...
metze
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-25 23:34:38 -07:00
76f0fb9170
gsskrb5: fix ec and padding handling in _gssapi_wrap_cfx_iov()
...
metze
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-25 23:34:38 -07:00
f286dd5d64
gsskrb5: fix _gssapi_wrap_iov_length_cfx() - there's more than just krb5 overhead...
...
metze
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-25 23:34:38 -07:00
1a0423fd3d
gsskrb5: make _gk_allocate_buffer() non static
...
metze
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-25 23:34:38 -07:00
60725fd2f5
gsskrb5: add _gk_verify_buffers()
...
metze
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-25 23:34:37 -07:00
9ccc79c5b6
Don't leak context if nsi_probe failes
...
Deduced from valgrind log produced by Markus Moeller
2009-08-22 10:52:22 -07:00
6618ca5ffc
switch to EVP_MD_CTX_create() and thus make smaller
2009-08-21 07:22:49 -07:00
56f90c5b19
switch to EVP_MD_CTX_create() and thus make smaller
2009-08-21 07:16:28 -07:00
f465930be7
switch to EVP_MD_CTX_create() and thus make smaller
2009-08-21 07:16:19 -07:00
dfd40e4403
switch to EVP_MD_CTX_create() and thus make smaller
2009-08-21 07:16:09 -07:00
03cb3aa56b
use EVP_MD_CTX_create
2009-08-20 17:13:09 -07:00
88d55a1d06
Make compile for weak crypto global (HEIM_WEAK_CRYPTO) and use it for GSSAPI too
2009-08-17 18:06:42 +02:00
fc702a97f5
switch to use EVP interface instead of old crypto interface
2009-08-17 17:30:59 +02:00
62433c844c
switch to use EVP interface instead of old crypto interface
2009-08-17 16:02:45 +02:00
fcfa32b0b9
Use constant time memcmp
2009-08-17 12:04:51 +02:00
42cec58cb4
switch to use EVP interface instead of old crypto interface
2009-08-17 11:43:24 +02:00
dfd1edb54d
switch to use EVP interface instead of old MDX_ style interface
2009-08-17 10:20:01 +02:00
ddb54ca483
switch to use EVP interface instead of old MDX_ style interface
2009-08-17 10:16:13 +02:00
13c3b9b1c6
switch to use EVP interface instead of old MDX_ style interface
2009-08-17 10:15:31 +02:00
639e93d436
switch to use EVP interface instead of old MDX_ style interface
2009-08-17 10:14:24 +02:00
3ef05891ee
switch to use EVP interface instead of old MDX_ style interface
2009-08-17 10:13:04 +02:00
ddb8230917
switch to use EVP interface instead of old MDX_ style interface
2009-08-17 10:10:42 +02:00
6ac304d156
Use min() instead of MIN()
2009-08-14 20:05:36 +02:00
07f0c8be5d
Clean better
2009-08-06 10:18:52 +02:00
95993f222c
Fix order of flags, passes regression test now
2009-08-05 13:42:34 +02:00
0ede7ac561
Pass down the use-dce-style flag instead of the while gssapi krb5 context
2009-08-05 12:00:07 +02:00
ab9e5d13ec
gsskrb5: try to be compatible with windows for gss_wrap* and cfx
...
The good thing is that windows and heimdal both use EC=0
in the non DCE_STYLE case, so we need the windows compat hack
only in DCE_STYLE mode.
metze
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-04 20:22:05 +02:00
0297d047a4
gsskrb5: add support for DCE_STYLE and des and des3 keys
...
Only the des keys are tested as windows doesn't support des3
metze
Signed-off-by: Love Hornquist Astrand <lha@h5l.org >
2009-08-04 20:21:20 +02:00
3cebc3767f
add more test to test_acquire_cred that removes the need of test_init_creds.c
2009-08-03 13:05:36 +02:00
3608b815b4
Don't bother checking usage of minor_status [CID-23]
2009-07-30 14:01:57 +02:00
901bac07e8
Don't need to look check *input_name twice [CID-27].
2009-07-30 14:00:48 +02:00
08256017e4
Don't dereference input_name [CID-27].
2009-07-30 13:59:42 +02:00
896391a56b
Double free of sp on empty list of creds [CID-183]
2009-07-30 07:46:37 +02:00
9581e59bde
FORWARD_NULL fixes [CID-163] and friends
2009-07-30 07:36:03 +02:00
b9644d7060
Test on wrong variable
2009-07-30 07:30:27 +02:00
Stefan Metzmacher