IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
if the 'password does not expire' flag has been set, filling in the
PAC and netlogon reply correctly if so.
Andrew Bartlett
(This used to be commit c530ab5dc6)
This tries to ensure that when we are a client, we cope with mechs
(like GSSAPI) that only abort (unknown server) at first runtime.
Andrew Bartlett
(This used to be commit cb5d18c619)
credentials.
Consistantly rename these elements in the IDL to computer_name.
Fix the server-side code to always lookup by this name.
Add new, even nastier tests to RPC-SCHANNEL to prove this.
Andrew Bartlett
(This used to be commit 341a0abeb4)
We do need the gsskrb5_get_initiator_subkey() routine. But we should
ensure that we do always get a valid key, to prevent any segfaults.
Without this code, we get a different session key compared with
Win2k3, and so kerberised smb signing fails.
Andrew Bartlett
(This used to be commit cfd0df16b7)
isn't every parameter on NTLMSSP, but it is most of the important
ones.
This showed up that we had the '128bit && LM_KEY' case messed up.
This isn't supported, so we must look instead at the 56 bit flag.
Andrew Bartlett
(This used to be commit 990da31b5f)
We were causing mayhem by weakening the keys at the wrong point in time.
I think this is the correct place to do it. The session key for SMB
signing, and the 'smb session key' (used for encrypting password sets)
is never weakened.
The session key used for bulk data encryption/signing is weakened.
This also makes more sense, when we look at the NTLM2 code.
Andrew Bartlett
(This used to be commit 3fd32a1209)
The new RPC-SCHANNEL test shows that the full credentials state must
be kept in some shared memory, for some length of time. In
particular, clients will reconnect with SCHANNEL (after loosing all
connections) and expect that the credentials chain will remain in the
same place.
To achive this, we do the server-side crypto in a transaction,
including the fetch/store of the shared state.
Andrew Bartlett
(This used to be commit 982a6aa871)
responses...
Also trust OpenLDAP to be pedantic about it, breaking connections to AD.
In any case, we now get this 'right' (by nasty overloading hacks, but
hey), and we can now use system-supplied OpenLDAP libs and SASL/GSSAPI
to talk to Samba4.
Andrew Bartlett
(This used to be commit 0cbe18211a)
GSSAPI differs from GSS-SPNEGO in an additional 3 packets, negotiating
a buffer size and what integrity protection/privacy should be used.
I worked off draft-ietf-sasl-gssapi-03, and this works against Win2k3.
I'm doing this in the hope that Apple clients as well as SASL-based
LDAP tools may get a bit further.
I still can't get ldapsearch to work, it fails with the ever-helpful
'Local error'.
Andrew Bartlett
(This used to be commit 3e46289775)
In particular, I've used the --leak-report-full option to smbd to
track down memory that shouldn't be on a long-term context. This is
now talloc_free()ed much earlier.
Andrew Bartlett
(This used to be commit c6eb74f429)
From here we can add tests to Samba for kerberos, forcing it on and
off. In the process, I also remove the dependency of credentials on
GENSEC.
This also picks up on the idea of bringing 'set_boolean' into general
code from jpeach's cifsdd patch.
Andrew Bartlett
(This used to be commit 1ac7976ea6)
case) as the keytab.
This avoids issues in replicated setups, as we will replicate the
kpasswd key correctly (including from windows, which is why I care at
the moment).
Andrew Bartlett
(This used to be commit 849500d1aa)
length of the (possibly null) pointer.
In reality this should come to us either 16 or 0 bytes in length, but
this is the safest test.
This is bug 3401 in Samba3, thanks to Yau Lam Yiu <yiuext at cs.ust.hk>
Andrew Bartlett
(This used to be commit f3aa702944)
gsskrb5_get_initiator_subkey() routine is bougs. We can indeed use
gss_krb5_get_subkey().
This is fortunate, as there was a segfault bug in 'initiator' version.
Andrew Bartlett
(This used to be commit ec11870ca1)
we are going to try and have a 'real' NT token for these users, it is
going to get messy fast. I want to go down the idmap road, but we
don't have the infrustucure for that yet.
Andrew Bartlett
(This used to be commit c90d5e82ff)
structure that is more generic than just 'IP/port'.
It now passes make test, and has been reviewed and updated by
metze. (Thankyou *very* much).
This passes 'make test' as well as kerberos use (not currently in the
testsuite).
The original purpose of this patch was to have Samba able to pass a
socket address stucture from the BSD layer into the kerberos routines
and back again. It also removes nbt_peer_addr, which was being used
for a similar purpose.
It is a large change, but worthwhile I feel.
Andrew Bartlett
(This used to be commit 88198c4881)
with clients compiled against the MIT Kerberos implementation. (Which
checks for address in KRB-PRIV packets, hence my comments on socket
functions earlier today).
It also fixes the 'set password' operation to behave correctly (it was
previously a no-op).
This allows Samba3 to join Samba4. Some winbindd operations even work,
which I think is a good step forward. There is naturally a lot of work
to do, but I wanted at least the very basics of Samba3 domain membership
to be available for the tech preview.
Andrew Bartlett
(This used to be commit 4e80a557f9)
using pre-calculated passwords for all kerberos key types.
(Previously we could only use these for the NT# type).
The module handles all of the hash/string2key tasks for all parts of
Samba, which was previously in the rpc_server/samr/samr_password.c
code. We also update the msDS-KeyVersionNumber, and the password
history. This new module can be called at provision time, which
ensures we start with a database that is consistent in this respect.
By ensuring that the krb5key attribute is the only one we need to
retrieve, this also simplifies the run-time KDC logic. (Each value of
the multi-valued attribute is encoded as a 'Key' in ASN.1, using the
definition from Heimdal's HDB. This simplfies the KDC code.).
It is hoped that this will speed up the KDC enough that it can again
operate under valgrind.
(This used to be commit e902274321)
the difference between these at all, and in the future the
fact that INIT_OBJ_FILES include smb_build.h will be sufficient to
have recompiles at the right time.
(This used to be commit b24f2583ed)
We now use a different system for initializing the modules for a subsystem.
Most subsystems now have an init function that looks something like this:
init_module_fn static_init[] = STATIC_AUTH_MODULES;
init_module_fn *shared_init = load_samba_modules(NULL, "auth");
run_init_functions(static_init);
run_init_functions(shared_init);
talloc_free(shared_init);
I hope to eliminate the other init functions later on (the
init_programname_subsystems; defines).
(This used to be commit b6d2ad4ce0)
We still have Win2000 issues, but now we correctly handle the case
where NTLMSSP is chosen as an authentication mech, but the OID list
still contains Kerberos as a later option.
Andrew Bartlett
(This used to be commit dc2b2c33f8)
This extracts a remote windows domain into a keytab, suitable for use
in ethereal for kerberos decryption.
For the moment, like net samdump and net samsync, the 'password
server' smb.conf option must be set to the binding string for the
server. eg:
password server = ncacn_np:mypdc
Andrew Bartlett
(This used to be commit 272013438f)
- the objectClass needs to be added to the list of attributes to make
the check for objectClass=computer work
- the short version of the name needs to be used for the 'cn' in
cracknames
(This used to be commit 53f0fb77c3)
This is for use on user-supplied arguments to printf style format
strings which will become ldb filters. I have used it on LSA, SAMR
and the auth/ code so far.
Also add comments to cracknames code.
Andrew Bartlett
(This used to be commit 8308cf6e04)
command line processing system.
This is a little ugly at the moment, but works. What I cannot manage
to get to work is the extraction and propogation of command line
credentials into the js interface to ldb.
Andrew Bartlett
(This used to be commit f34ede763e)
backend.
The idea is that every time we open an LDB, we can provide a
session_info and/or credentials. This would allow any ldb to be remote
to LDAP. We should also support provisioning to a authenticated ldap
server.
(They are separate so we can say authenticate as foo for remote, but
here we just want a token of SYSTEM).
Andrew Bartlett
(This used to be commit ae2f3a64ee)
To avoid a circular depenency, it is not allowed to use Krb5 as an
authentication mechanism, so this must be removed from the list. An
extension to the credentials system allows this function.
Also remove proto.h use for any of the KDC, and use NTSTATUS returns
in more places.
Andrew Bartlett
(This used to be commit 5f9dddd02c)
GENSEC mechansims. This will allow a machine join to an NT4 domain to
avoid even trying kerberos, or a sensitive operation to require it.
Andrew Bartlett
(This used to be commit 11c7a89e52)
This merges Samba4 up to current lorikeet-heimdal, which includes a
replacement for some Samba-specific hacks.
In particular, the credentials system now supplies GSS client and
server credentials. These are imported into GSS with
gss_krb5_import_creds(). Unfortunetly this can't take an MEMORY
keytab, so we now create a FILE based keytab as provision and join
time.
Because the keytab is now created in advance, we don't spend .4s at
negprot doing sha1 s2k calls. Also, because the keytab is read in
real time, any change in the server key will be correctly picked up by
the the krb5 code.
To mark entries in the secrets which should be exported to a keytab,
there is a new kerberosSecret objectClass. The new routine
cli_credentials_update_all_keytabs() searches for these, and updates
the keytabs.
This is called in the provision.js via the ejs wrapper
credentials_update_all_keytabs().
We can now (in theory) use a system-provided /etc/krb5.keytab, if
krb5Keytab: FILE:/etc/krb5.keytab
is added to the secrets.ldb record. By default the attribute
privateKeytab: secrets.keytab
is set, pointing to allow the whole private directory to be moved
without breaking the internal links.
(This used to be commit 6b75573df4)
This allows a new password to be written in, and old entries removed
(we keep kvno and kvno-1).
Clean up the code a lot, and add comments on what it is doing...
Andrew Bartlett
(This used to be commit 0a911baaba)
messages. As discussed with Andrew, this will soon be replaced with a
system that marks the credentials to use the machine accout from the
database rather than pre-loading the machine account details here.
The reason we got the annoying messages is this was being called
before smb.conf is loaded, so the code doesn't yet know the location
of the private directory
(This used to be commit 6aeb4bf3fe)
This patch changes the way lsb_search is called and the meaning of the returned integer.
The last argument of ldb_search is changed from struct ldb_message to struct ldb_result
which contains a pointer to a struct ldb_message list and a count of the number of messages.
The return is not the count of messages anymore but instead it is an ldb error value.
I tryed to keep the patch as tiny as possible bu as you can guess I had to change a good
amount of places. I also tried to double check all my changes being sure that the calling
functions would still behave as before. But this patch is big enough that I fear some bug
may have been introduced anyway even if it passes the test suite. So if you are currently
working on any file being touched please give it a deep look and blame me for any error.
Simo.
(This used to be commit 22c8c97e6f)
We now put the PAC in the AS-REP, so that the client has it in the
TGT. We then validate it (and re-sign it) on a TGS-REQ, ie when the
client wants a ticket.
This should also allow us to interop with windows KDCs.
If we get an invalid PAC at the TGS stage, we just drop it.
I'm slowly trying to move the application logic out of hdb-ldb.c, and
back in with the rest of Samba's auth system, for consistancy. This
continues that trend.
Andrew Bartlett
(This used to be commit 36973b1eef)
kdc/hdb-ldb.c to share the routines used for auth/
This will require keeping the attribute list in sync, but I think it
is worth it for the next steps (sharing the server_info generation).
Andrew Bartlett
(This used to be commit da38bcefa7)
determining a mechanism to use.
Currently it doesn't to fallbacks like SPNEGO does, but this could be
added (to GENSEC, not to here).
This also adds a new function to GENSEC, which returns a list of SASL
names in our preference order (currently determined by the build
system of all things...).
Also make the similar function used for OIDs in SPNEGO do the same.
This is all a very long-winded way of moving from a hard-coded NTLM to
GSS-SPNEGO in our SASL client...
Andrew Bartlett
(This used to be commit 130eb9bb9a)
DCE_STYLE modified version, and add parametric options to control
delegation.
It turns out the only remaining issue is sending delegated credentials
to a windows server, probably due to the bug lha mentions in his blog
(using the wrong key).
If I turn delgation on in smbclient, but off in smbd, I can proxy a
cifs session.
I can't wait till Heimdal 0.8, so I'll see if I can figure out the fix
myself :-)
Andrew Bartlett
(This used to be commit fd5fd03570)
of the gsskrb5_acquire_cred hack.
Add support for delegated credentials into the auth and credentials
subsystem, and specifically into gensec_gssapi.
Add the CIFS NTVFS handler as a consumer of delegated credentials,
when no user/domain/password is specified.
Andrew Bartlett
(This used to be commit 55b89899ad)
authentication for user@realm logins and machine account logins.
This should avoid various protocol downgrade attacks.
Andrew Bartlett
(This used to be commit 76c2d204d0)
- Adds -rpath bin/ so you don't have to install Samba in order to use compiled binaries.
- Writes out pkg-config files when building shared libs
- Supports automatic fallback to MERGEDOBJ (which is the default) or
OBJ_LIST (if ld -r is not supported)
Building with shared libs reduces the size of the Samba binaries from
197 Mb to 60 Mb (including libraries) on my system (GCC4, with debugging).
To build with shared libraries support enabled, run:
LIBRARY_OUTPUT_TYPE=SHARED_LIBRARY ./config.status
init functions don't get called correctly yet when using shared libs, so
you won't be able to actually run anything with success :-)
Once init functions are done, I'll look at support for loading shared
modules once again.
Based on a patch by Peter Novodvorsky (nidd on IRC).
(This used to be commit 0b54405685)
This avoids the nasty user@DOMAIN test for now, as it has very odd
semantics with NTLMv2.
Allow only user accounts to do an interactive login.
Andrew Bartlett
(This used to be commit 690cad8083)
(Make it easy to see what was put into the keytab, so we can tell when
gssapi screams that it can't pull it out).
Andrew Bartlett
(This used to be commit c56142c4ac)
than doing ASN.1 parsing in Samba.
Also use the API function for getting a client from a ticket, rather
than just digging in the structure.
Andrew Bartlett
(This used to be commit 25d5ea6d72)
we can round-trip all the way back to a server_info structure, not
just a filled in PAC_DATA. (I was worried about generated fields being
incorrect, or some other logical flaw).
Andrew Bartlett
(This used to be commit 11b1d78cc5)
I'm sure this will not be the final resting place, but it will do for
now.
Use the cracknames code in auth/ for creating a server_info given a
principal name only (should avoid assumtions about spliting a
user@realm principal).
Andrew Bartlett
(This used to be commit c9d5d8e45d)
Add the kpasswd server to our KDC, implementing the 'original' and
Microsoft versions of the protocol.
This works with the Heimdal kpasswd client, but not with MIT, I think
due to ordering issues. It may not be worth the pain to have this
code go via GENSEC, as it is very, very tied to krb5.
This gets us one step closer to joins from Apple, Samba3 and other
similar implementations.
Andrew Bartlett
(This used to be commit ab5dbbe10a)
credentials. This works with the setup/secrets.ldif change from the
previous patch, and pretty much just re-invents the keytab.
Needed for kpasswdd work.
Andrew Bartlett
(This used to be commit cc9d167bab)
I'm also worried this might cause loops, if we get a 'force password
change', and the prompter tries to 'deal with it'.
Andrew Bartlett
(This used to be commit 5bc10c4e47)
authentication. This pulls the creating of the keytab back to the
credentials code, and removes the special case of 'use keberos keytab
= yes' for now.
This allows (and requires) the callers to specify the credentials for
the server credentails to GENSEC. This allows kpasswdd (soon to be
added) to use a different set of kerberos credentials.
The 'use kerberos keytab' code will be moved into the credentials
layer, as the layers below now expect a keytab.
We also now allow for the old secret to be stored into the
credentials, allowing service password changes.
Andrew Bartlett
(This used to be commit 205f77c579)
the client doesn't guess correctly on the mech to use. It must back
off and try the mech the server selected from the list.
I'm not particularly attached to our SPNEGO parser, so while I can't
easily use the SPNEGO application logic in Heimdal, I'm going to look
closely at using the asn1 routines to avoid some pain here.
Andrew Bartlett
(This used to be commit 9292173874)
authentication out of the various callers and into the kitchen
sink.. err, credentials subsystem.
This should ensure consistant logic, as well as get us one step closer
to security=server operation in future.
Andrew Bartlett
(This used to be commit 09c9576330)
most of the changes are fixes to make all the ldb code compile without
warnings on gcc4. Unfortunately That required a lot of casts :-(
I have also added the start of an 'operational' module, which will
replace the timestamp module, plus add support for some other
operational attributes
In ldb_msg_*() I added some new utility functions to make the
operational module sane, and remove the 'ldb' argument from the
ldb_msg_add_*() functions. That argument was only needed back in the
early days of ldb when we didn't use the hierarchical talloc and thus
needed a place to get the allocation function from. Now its just a
pain to pass around everywhere.
Also added a ldb_debug_set() function that calls ldb_debug() plus sets
the result using ldb_set_errstring(). That saves on some awkward
coding in a few places.
(This used to be commit f6818daecc)
ldap. Also ensure we put a objectclass on our private ldb's, so they
have some chance of being stored in ldap if you want to
(This used to be commit 1af2cc067f)
authenticated session down into LDB. This associates a session info
structure with the open LDB, allowing a future ldb_ntacl module to
allow/deny operations on that basis.
Along the way, I cleaned up a few things, and added new helper functions
to assist. In particular the LSA pipe uses simpler queries for some of
the setup.
In ldap_server, I have removed the 'ldasrv:hacked' module, which hasn't
been worked on (other than making it continue to compile) since January,
and I think the features of this module are being put into ldb anyway.
I have also changed the partitions in ldap_server to be initialised
after the connection, with the private pointer used to associate the ldb
with the incoming session.
Andrew Bartlett
(This used to be commit fd7203789a)
I also think the method of getting rid of pstring isn't the right
one. I certainly do want to get rid of pstring/fstring, but the reason
for removing them is the use of arbitrary sized fixed length strings
on the stack and in structures. Changing to another fixed length stack
string format isn't really a win, and moving to use strncpy() is
actually worse than pstrcpy() as strncpy() has the absolutely awful
semantics of always zeroing all remaining bytes, so it ends up taking
a lot of cpu doing pointless memory writes.
I'd rather move to more use of asprintf()/talloc_asprintf() and
similar functions for dynamic string allocation.
You also have to be very careful about some of these system defined
string limits. One some systems PATH_MAX could be 64k or even larger,
which can quickly blow the stack out when you allocate a few of them.
(This used to be commit 194efd26e4)
at the Samba4 socket layer.
The intention here is to ensure that other events may be processed while
heimdal is waiting on the KDC. The interface is designed to be
sufficiently flexible, so that the plugin may choose how to time
communication with the KDC (ie multiple outstanding requests, looking
for a functional KDC).
I've hacked the socket layer out of cldap.c to handle this very
specific case of one udp packet and reply. Likewise I also handle
TCP, stolen from the winbind code.
This same plugin system might also be useful for a self-contained
testing mode in Heimdal, in conjunction with libkdc. I would suggest
using socket-wrapper instead however.
Andrew Bartlett
(This used to be commit 3b09f9e8f9)
I still have issues with Win2k3 SP1, and Samba4 doesn't pass it's own
test for the moment, but I'm working on these issues :-)
This required a change to the credentials API, so that the special
case for NTLM logins using a principal was indeed handled as a
special, not general case.
Also don't set the realm from a ccache, as then it overrides --option=realm=.
Andrew Bartlett
(This used to be commit 194e8f07c0)
use raw krb5, not GSSAPI. I still keep the 'fake GSSAPI' code, but
under the module name 'fake_gssapi_krb5'.
Andrew Bartlett
(This used to be commit 99efec2758)
at every single build. Run 'scons configure=1' or delete sconf.cache
to force checks to be re-run.
Jelmer, I think this stuff is cached in the .sconf_cache directory but
the message is still displayed and it looks like it caches the compiled
test object file not the actual result of the test.
(This used to be commit 9d001dc083)
-Udomain\\user.
This will probably break in a few configurations, so please let me
know. I'll also work to have a way to inhibit kerberos/ntlmssp, as
this removes -k.
Andrew Bartlett
(This used to be commit 3c0dc570b8)
on the kerberos mailing lists a couple of weeks ago: Don't use DNS at
all for expanding short names into long names.
Using the 'override krb5_init_context' code already in the tree, this
removes the DNS lag on a kerberos session setup/connection.
Andrew Bartlett
(This used to be commit de3ceab3d0)
Samba to use the target principal name supplied in the mechTokenMIC of
an SPNEGO negTokenInit.
This isn't a great idea for security reasons, but is how Samba3 behaves,
and allows kerberos to function more often in some environments. It is
only available for CIFS session setups, due to the ordering of the
exchange.
Andrew Bartlett
(This used to be commit f6a6456441)
ideas I have had.
When I get a full list of things I want to do to a krb5_context I'll
either add gsskrb5_ wrappers, or a way of speicfying the krb5 context
per gssapi context.
(I want to ensure that the only krb5_context variables created while
executing Samba4 are via our wrapper).
Andrew Bartlett
(This used to be commit 8a22d46e70)
about the size of the signature. In particular, this works with AES,
which was previously broken Samba4/Samba4.
Reviewed by metze (and thanks for help with the previous IDL commit).
(This used to be commit 3c8be196cc)
data to be signed/sealed. We can use this to split the data from the
signature portion of the resultant wrapped packet.
This required merging the gsskrb5_wrap_size patch from
lorikeet-heimdal, and fixes AES encrption issues on DCE/RPC (we no
longer use a static 45 byte value).
This fixes one of the krb5 issues in my list.
Andrew Bartlett
(This used to be commit e4f2afc343)
with an aim to make the code simpiler and more correct.
Gone is the old (since the very early Samba 3.0 krb5 days) 'iterate over
all keytypes)' code in gensec_krb5, we now follow the approach used in
gensec_gssapi, and use a keytab.
I have also done a lot of work in the GSSAPI code, to try and reduce
the diff between us and upstream heimdal. It was becoming hard to
track patches in this code, and I also want this patch (the DCE_STYLE
support) to be in a 'manageable' state for when lha considers it for
merging. (metze assures me it still has memory leak problems, but
I've started to address some of that).
This patch also includes a simple update of other code to current
heimdal, as well as changes we need for better PAC verification.
On the PAC side of things we now match windows member servers by
checking the name and authtime on an incoming PAC. Not generating these
right was the cause of the PAC pain, and so now both the main code and
torture test validate this behaviour.
One thing doesn't work with this patch:
- the sealing of RPC pipes with kerberos, Samba -> Samba seems
broken. I'm pretty sure this is related to AES, and the need to break
apart the gss_wrap interface.
Andrew Bartlett
(This used to be commit a3aba57c00)
'MEMORY_WILDCARD' keytab type. (part of this checking is in effect a
merge from lorikeet-heimdal, where I removed this)
This is achieved by correctly using the GSSAPI gsskrb5_acquire_cred()
function, as this allows us to specify the target principal, regardless
of which alias the client may use.
This patch also tries to simplify some principal handling and fixes some
error cases.
Posted to samba-technical, reviewed by metze, and looked over by lha on IRC.
Andrew Bartlett
(This used to be commit 506a7b67ae)
Kerberos CCACHE into the system.
This again allows the use of the system ccache when no username is
specified, and brings more code in common between gensec_krb5 and
gensec_gssapi.
It also has a side-effect that may (or may not) be expected: If there
is a ccache, even if it is not used (perhaps the remote server didn't
want kerberos), it will change the default username.
Andrew Bartlett
(This used to be commit 6202267f6e)
