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.
instead of either sysv or mmap shared memory or lock files.
this means we can now completely remove
locking_shm.c
locking_slow.c
shmem.c
shmem_sysv.c
and lots of other things also got simpler
locking.c got a bit larger, but is much better compartmentalised now
suite and a very simple tool for manuipulating the databases.
the main code is in tdb/tdb.c and includes both mmap and file based
IO. All databases auto-expand and allow multiple simultaneous writers.
the next step is using this new capability in lots of places in Samba
where we have existing ad-hoc databases
in a /etc/nsswitch.conf hosts line.
Only tested on RH6.1, but should work on a broad range of Linux
distributions. It could probably be made to work with Solaris pretty
easily.
It does not build by default. Build it with "make nsswitch"
done a minimal amout of clean-up in the Makefile, removing unnecessary
modules from the link stage. this is not complete, yet, and will
involve some changes, for example to smbd, to remove dependencies on
the password database API that shouldn't be there. for example,
smbd should not ever call getsmbpwXXX() it should call the Samr or Lsa
API.
this first implementation has minor problems with not reinstantiating
the same services as the caller. the "homes" service is a good example.
pdus, and then feeds them over either a "local" function call or a "remote"
function call to an msrpc service. the "remote" msrpc daemon, on the
other side of a unix socket, then calls the same "local" function that
smbd would, if the msrpc service were being run from inside smbd.
this allows a transition from local msrpc services (inside the same smbd
process) to remote (over a unix socket).
removed reference to pipes_struct in msrpc services. all msrpc processing
functions take rpcsrv_struct which is a structure containing state info
for the msrpc functions to decode and create pdus.
created become_vuser() which does everything not related to connection_struct
that become_user() does.
removed, as best i could, connection_struct dependencies from the nt spoolss
printing code.
todo: remove dcinfo from rpcsrv_struct because this stores NETLOGON-specific
info on a per-connection basis, and if the connection dies then so does
the info, and that's a fairly serious problem.
had to put pretty much everything that is in user_struct into parse_creds.c
to feed unix user info over to the msrpc daemons. why? because it's
expensive to do unix password/group database lookups, and it's definitely
expensive to do nt user profile lookups, not to mention pretty difficult
and if you did either of these it would introduce a complication /
unnecessary interdependency. so, send uid/gid/num_groups/gid_t* +
SID+num_rids+domain_group_rids* + unix username + nt username + nt domain
+ user session key etc. this is the MINIMUM info identified so far that's
actually implemented. missing bits include the called and calling
netbios names etc. (basically, anything that can be loaded into
standard_sub() and standard_sub_basic()...)
time out of sending the session setup on Solaris 2.6. No idea.
I'll work on it some tomorrow. This is to fix the "Unable to
setup password vectors" thingy.
Also changed an inet_aton() to inet_addr() as the former is
not very portable :-)
Luke, I set the redir flag to false because the connection to
the smb-agent was failing and smbpasswd bombed. Double check me
on this one.
-jc
at present, a unix password is missing from the unix credentials, but is
not _actually_ expected to be needed. weeelll... maybe :-)
this is used to transfer credentials between smbd and msrpc daemons, down
a unix socket, so that the unix and nt credentials can be inherited by
an msrpc daemon called up from smbd.
damn, this one is bad.
started, at least two days ago, to add an authentication mechanism to
the smbd<->msrpc redirector/relay, such that sufficient unix / nt
information could be transferred across the unix socket to do a
become_user() on the other side of the socket.
it is necessary that the msrpc daemon inherit the same unix and nt
credentials as the smbd process from which it was spawned, until
such time as the msrpc daemon receives an authentication request
of its own, whereupon the msrpc daemon is responsible for authenticating
the new credentials and doing yet another become_user() etc sequence.
one horrible cut / paste job from smbd, plus a code split of shared
components between the two.
the job is not _yet_ complete, as i need to be able to do a become_user()
call for security reasons. i picked lsarpcd first because you don't
_need_ security on it (microsoft botched so badly on this one, it's not
real. at least they fixed this in nt5 with restrictanonymous=0x2).
fixing this involves sending the current smb and unix credentials down
the unix pipe so that the daemon it eventually goes to can pick them
up at the other end.
i can't believe this all worked!!!
received properly when a UDP "retry" occurs. it's because reads and
writes must be interleaved / matched.
scenario:
nmblookup connects to agent, sends request.
agent receives request, broadcasts it on 137.
agent RECEIVES 137 broadcast, sends it to nmblookup
agent receives RESPONSE to 137 broadcast, sends it to nmblookup.
if reads are not equally interspersed with writes, then second send
will fail.
if you think this is odd behaviour and that the agent should be filtering
its own UDP traffic, think again.
agent will be, potentially, redirecting nmbd traffic (including WINS
server) not just client programs.
created an "nmb-agent" utility that, yes: it connects to the 137 socket
and accepts unix socket connections which it redirects onto port 137.
it uses the name_trn_id field to filter requests to the correct
location.
name_query() and name_status() are the first victims to use this
feature (by specifying a file descriptor of -1).
moved smb-agent over to a single-process model instead of fork()
in order to reuse client connections. except, of course, you
can't do a select() on the same socket connections! argh!
ideas from ssh-agent.
the intent is to be able to share smb sessions using cli_net_use_add()
across multiple processes, where one process knows the target server
name, user name and domain, but not the smb password.
lp_trusted_domains() parameter, so trusted domain logins should work,
right, if you put user = TRUSTED_DOMAIN\NTuser in "domain name map", right?
right - as _long_ as you're not using NTLMv2, because the damn NT username
gets mapped to the damn unix name too early, and NTLMv2 challenge-responses
are based on the client's user name, client's domain name, client's host name
etc damn etc.
so it becomes necessary to stop using char* username because this allows
for massive amounts of confusion as to which username is being referred to.
the underlying unix username on the local unix system that is associated with
the smbd process that represents the NT username? or the NT username itself?
