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.
The main difference in this new version is the extra data structure generated
between the IDL data structure and the NDR parser:
IDL -> NDR -> { ndr_parser, ndr_header, eparser, etc }
This makes the ndr_parser.pm internals much more sane.
Other changes include:
- Remove unnecessary calls with NDR_BUFFERS (for example, GUID doesn't have any buffers, just scalars) as well as some (unnecessary) nested setting of flags.
- Parse array loops in the C code rather then calling ndr_pull_array(). This allows us to have, for example, arrays of pointers or arrays of pointers to arrays, etc..
- Use if() {} rather then if () goto foo; everywhere
- NDR_IN no longer implies LIBNDR_FLAG_REF_ALLOC
- By default, top level pointers are now "ref" (as is the default in
most other IDL compilers). This can be overridden using the
default_pointer_top() property.
- initial work on new ethereal parser generators by Alan DeKok and me
- pidl now writes errors in the standard format used by compilers, which
is parsable by most editors
- ability to warn about the fact that pidl extension(s) have been used,
useful for making sure IDL files work with other IDL compilers.
oh, and there's probably some other things I can't think of right now..
(This used to be commit 13cf227615)
I wanted to add a simple 'workstation' argument to the DCERPC
authenticated binding calls, but this patch kind of grew from there.
With SCHANNEL, the 'workstation' name (the netbios name of the client)
matters, as this is what ties the session between the NETLOGON ops and
the SCHANNEL bind. This changes a lot of files, and these will again
be changed when jelmer does the credentials work.
I also correct some schannel IDL to distinguish between workstation
names and account names. The distinction matters for domain trust
accounts.
Issues in handling this (issues with lifetime of talloc pointers)
caused me to change the 'creds_CredentialsState' and 'struct
dcerpc_binding' pointers to always be talloc()ed pointers.
In the schannel DB, we now store both the domain and computername, and
query on both. This should ensure we fault correctly when the domain
is specified incorrectly in the SCHANNEL bind.
In the RPC-SCHANNEL test, I finally fixed a bug that vl pointed out,
where the comment claimed we re-used a connection, but in fact we made
a new connection.
This was achived by breaking apart some of the
dcerpc_secondary_connection() logic.
The addition of workstation handling was also propogated to NTLMSSP
and GENSEC, for completeness.
The RPC-SAMSYNC test has been cleaned up a little, using a loop over
usernames/passwords rather than manually expanded tests. This will be
expanded further (the code in #if 0 in this patch) to use a newly
created user account for testing.
In making this test pass test_rpc.sh, I found a bug in the RPC-ECHO
server, caused by the removal of [ref] and the assoicated pointer from
the IDL. This has been re-added, until the underlying pidl issues are
solved.
(This used to be commit 824289dcc2)
less likely that anyone will use pstring for new code
- got rid of winbind_client.h from includes.h. This one triggered a
huge change, as winbind_client.h was including system/filesys.h and
defining the old uint32 and uint16 types, as well as its own
pstring and fstring.
(This used to be commit 9db6c79e90)
make it possible to add optimisations to the events code such as
keeping the next timed event in a sorted list, and using epoll for
file descriptor events.
I also removed the loop events code, as it wasn't being used anywhere,
and changed timed events to always be one-shot (as adding a new timed
event in the event handler is so easy to do if needed)
(This used to be commit d7b4b6de51)
servers in smbd. The old code still contained a fairly bit of legacy
from the time when smbd was only handling SMB connection. The new code
gets rid of all of the smb_server specific code in smbd/, and creates
a much simpler infrastructures for new server code.
Major changes include:
- simplified the process model code a lot.
- got rid of the top level server and service structures
completely. The top level context is now the event_context. This
got rid of service.h and server.h completely (they were the most
confusing parts of the old code)
- added service_stream.[ch] for the helper functions that are
specific to stream type services (services that handle streams, and
use a logically separate process per connection)
- got rid of the builtin idle_handler code in the service logic, as
none of the servers were using it, and it can easily be handled by
a server in future by adding its own timed_event to the event
context.
- fixed some major memory leaks in the rpc server code.
- added registration of servers, rather than hard coding our list of
possible servers. This allows for servers as modules in the future.
- temporarily disabled the winbind code until I add the helper
functions for that type of server
- added error checking on service startup. If a configured server
fails to startup then smbd doesn't startup.
- cleaned up the command line handling in smbd, removing unused options
(This used to be commit cf6a46c3cb)
- added async support to the negprot client code
- removed two unused parameters from smbcli_full_connection() code
- converted smbclient to use smbcli_full_connection() rather than
reinventing everything itself
(This used to be commit 71cbe28734)
- stream_socket services
the smb, ldap and rpc service which sets up a srtam socket end then
waits for connections
and
- task services
which this you can create a seperate task that do something
(this is also going through the process_model subsystem
so with -M standard a new process for this created
with -M thread a new thread ...
I'll add datagram services later when we whave support for datagram sockets in lib/socket/
see the next commit as an example for service_task's
metze
(This used to be commit d5fa02746c)
the backend should check for
(dce_call->state_flags & DCESRV_CALL_STATE_FLAG_MAY_ASYNC)
then it's allowed to reply async
then the backend should mark that call as async with
dce_call->state_flags |= DCESRV_CALL_STATE_FLAG_ASYNC;
later it has to manualy set r->out.result
and then send the reply by calling
status = dcesrv_reply(p->dce_call);
NOTE: that ncacn_np doesn't support async replies yet
- implement an async version of echo_TestSleep
- reenable the echo_TestSleep torture test
(this need to be more strict when we have support for async ncacn_np)
metze
(This used to be commit f0a0dbeb25)
this stage does the following:
- simplifies the dcerpc_handle handling, and all the callers of it
- split out the context_id depenent state into a linked list of established contexts
- fixed some talloc handling in several rpc servers that i noticed while doing the above
(This used to be commit fde042b3fc)
use function pointers anymore
- make the module init much easier
- a lot of cleanups
don't try to read the diff in auth/ better read the new files
it passes test_echo.sh and test_rpc.sh
abartlet: please fix spelling fixes
metze
(This used to be commit 3c0d16b823)
- there is no alter_nak or alter_ack packet, its all done in an
alter_response
- auto-allocated the contex_ids
- tried to fix up the dcom code to work again with
alter_context. Jelmer, please take a look :)
(This used to be commit dd1c54add8)
- added #if TALLOC_DEPRECATED around the _p functions
- fixes the code that broke from the above
while doing this I fixed quite a number of places that were
incorrectly using the non type-safe talloc functions to use the type
safe ones. Some were even doing multiplies for array allocation, which
is potentially unsafe.
(This used to be commit 6e7754abd0)
main rpc server code. let the backends specify a ndr_push/ndr_pull function
like we already do with the dispatch() function.
this allows an interface implmentation to work as real proxy
without needing to know the idl for an interface that means
just the plain decrypted payload can be forwarded
If someone want to write such a backend, patches are wellcome
metze
(This used to be commit a150bdf140)
- nt4 doesn't setup the pfc flags correctly for rpc packet types
other than normal requests, so don't check for fragmented packets
unless they are of type request
- ensure we give STATUS_BUFFER_OVERFLOW when we return a partial
fragment in SMBtrans requests on ncacn_np
(This used to be commit 83ebffec32)
rather then a large table in librpc/gen_ndr/tables.c. This will allow us
to only link in only the required gen_ndr files (speeds up linking quite a
bit, makes binaries smaller).
Each gen_ndr_* file now has a init function that calls the init functions
of the interfaces it contains. I did it this way to keep pidl's code simple,
though it might hurt startup time a bit. I'd be happy to change it if
people like one function better.
(This used to be commit 3c436590ae)
Both subsystems and modules can now have init functions, which can be
specified in .mk files (INIT_FUNCTION = ...)
The build system will define :
- SUBSYSTEM_init_static_modules that calls the init functions of all statically compiled modules. Failing to load will generate an error which is not fatal
- BINARY_init_subsystems that calls the init functions (if defined) for the subsystems the binary depends on
This removes the hack with the "static bool Initialised = " and the
"lazy_init" functions
(This used to be commit 7a8244761b)
deferred reply is short-circuited immediately when the file is
closed by another user, allowing it to be opened by the waiting user.
- added a sane set of timeval manipulation routines
- converted all the events code and code that uses it to use struct
timeval instead of time_t, which allows for microsecond resolution
instead of 1 second resolution. This was needed for doing the pvfs
deferred open code, and is why the patch is so big.
(This used to be commit 0d51511d40)
- tidied up some of the system includes
- moved a few more structures back from misc.idl to netlogon.idl and samr.idl now that pidl
knows about inter-IDL dependencies
(This used to be commit 7b7477ac42)
The thing that finally convinced me that minimal includes was worth
pursuing for rpc was a compiler (tcc) that failed to build Samba due
to reaching internal limits of the size of include files. Also the
fact that includes.h.gch was 16MB, which really seems excessive. This
patch brings it back to 12M, which is still too large, but
better. Note that this patch speeds up compile times for both the pch
and non-pch case.
This change also includes the addition iof a "depends()" option in our
IDL files, allowing you to specify that one IDL file depends on
another. This capability was needed for the auto-includes generation.
(This used to be commit b8f5fa8ac8)
Examples of binding strings are :
ncalrpc:[EPMAPPER]
ncacn_unix_stream:[/tmp/epmapper]
N.B. The unix socket support in lib/socket/ appears to close and remove the
socket it is listening on after the first client disconnects so until
that has been fixed, it is only possible to do one ncalrpc: or ncacn_unix_stream: request per instance of smbd :-)
Support for looking up NCALRPC names via the endpoint mapper will be added later.
(This used to be commit 426f3e63ca)
specifying a endpoint is now also 'endpoint' instead of 'endpoints'. The
default endpoint (if none is specified) is still "ncacn_np:[\\pipe\\ifacename]",
where ifacename is the name of the interface.
Examples:
[
uuid(60a15ec5-4de8-11d7-a637-005056a20182),
endpoint("ncacn_np:[\\pipe\\rpcecho]", "ncacn_ip_tcp:")
]
interface rpcecho
{
void dummy();
}
dcerpc_binding is now converted to ep_description in the server, but I hope to
completely eliminate ep_description later on.
The eventual goal of all these changes is to make it easier to add
transports as I'm going to add support for
ncalrpc (local RPC over named pipes) and ncacn_unix_stream (Unix sockets).
(This used to be commit f3da7c8b44)
existing call rather than creating a new one. This prevents call
structures hanging around on the rpc connection context until it is
closed
(This used to be commit c51ca7c0e7)
taking a context (so when you pass a NULL pointer you end up with
memory in a top level context). Fixed it by changing the API to take a
context. The context is only used if the pointer you are reallocing is
NULL.
(This used to be commit 8dc23821c9)
rather than manual reference counts
- properly support SMBexit in the cifs and posix backends
- added a logoff method to all backends
With these changes the RAW-CONTEXT test now passes against the posix backend
(This used to be commit c315d6ac1c)
server code. This fixes a number of memory leaks I found when testing
with valgrind and smbtorture, as the cascading effect of a
talloc_free() ensures that anything derived from the top level object
is destroyed on disconnect.
(This used to be commit 76d0b8206c)
'authenticated' connections.
Fix kerberos session key issues - we need to call the
routine for extracting the session key, not just read the cache.
Andrew Bartlett
(This used to be commit b80d849b6b)
Fix indent, and add a few more useful debug messages.
Send a fault, if the bind is not accepted - don't just leave the client hanging.
Andrew Bartlett
(This used to be commit 486215edc1)
This version does the following:
1) talloc_free(), talloc_realloc() and talloc_steal() lose their
(redundent) first arguments
2) you can use _any_ talloc pointer as a talloc context to allocate
more memory. This allows you to create complex data structures
where the top level structure is the logical parent of the next
level down, and those are the parents of the level below
that. Then destroy either the lot with a single talloc_free() or
destroy any sub-part with a talloc_free() of that part
3) you can name any pointer. Use talloc_named() which is just like
talloc() but takes the printf style name argument as well as the
parent context and the size.
The whole thing ends up being a very simple piece of code, although
some of the pointer walking gets hairy.
So far, I'm just using the new talloc() like the old one. The next
step is to actually take advantage of the new interface
properly. Expect some new commits soon that simplify some common
coding styles in samba4 by using the new talloc().
(This used to be commit e35bb094c5)
in pidl. This mechanism should be much easier to extend to the
"retrospective subcontexts" that jelmer needs.
also produced more standards complient full-pointer offsets. This
keeps ethereal happy with decoding our epmapper frames.
(This used to be commit ecb7378bbc)
the idea is to have services as modules (smb, dcerpc, swat, ...)
the process_model don't know about the service it self anymore.
TODO:
- the smbsrv should use the smbsrv_send function
- the service subsystem init should be done like for other modules
- we need to have a generic socket subsystem, which handle stream, datagram,
and virtuell other sockets( e.g. for the ntvfs_ipc module to connect to the dcerpc server
, or for smb or dcerpc or whatever to connect to a server wide auth service)
- and other fixes...
NOTE: process model pthread seems to be broken( but also before this patch!)
metze
(This used to be commit bbe5e00715)
This implements gensec for Samba's server side, and brings gensec up
to the standards of a full subsystem.
This means that use of the subsystem is by gensec_* functions, not
function pointers in structures (this is internal). This causes
changes in all the existing gensec users.
Our RPC server no longer contains it's own generalised security
scheme, and now calls gensec directly.
Gensec has also taken over the role of auth/auth_ntlmssp.c
An important part of gensec, is the output of the 'session_info'
struct. This is now reference counted, so that we can correctly free
it when a pipe is closed, no matter if it was inherited, or created by
per-pipe authentication.
The schannel code is reworked, to be in the same file for client and
server.
ntlm_auth is reworked to use gensec.
The major problem with this code is the way it relies on subsystem
auto-initialisation. The primary reason for this commit now.is to
allow these problems to be looked at, and fixed.
There are problems with the new code:
- I've tested it with smbtorture, but currently don't have VMware and
valgrind working (this I'll fix soon).
- The SPNEGO code is client-only at this point.
- We still do not do kerberos.
Andrew Bartlett
(This used to be commit 07fd885fd4)
- added workstation to auth_session_info in rpc servers
- added session key fetch hook in crypto backends in dcesrv
- store and fetch seed as well as a session key in schannel ldb
- when a client uses schannel to setup a netlogon pipe connection we
also need to setup the credentials from the schannel negotiation so
credentials chaining works
- added server side netr_LogonGetDomainInfo() call
(This used to be commit a35459387d)
the dce_conn->auth_state.session_info
( the ntlmssp one works fine, but the schannel one isn't implemented yet)
this is also set by the ntvfs_ipc backend on the endpoint connect.
metze
(This used to be commit ad3dd1789e)
makes it much easier to develop the IDL for new requests, especially
for sealed pipes, where ethereal cannot easily extract the data.
(This used to be commit 0cde043592)
I made it much more generic, and we should be able to add a
module interface to this code, so that other DCERPC_AUTH types can be added
via modules...
metze
(This used to be commit d09abeb686)
names rather than our crazy naming scheme. So DES is now called
des_crypt() rather than smbhash()
- added the code from the solution of the ADS crypto challenge that
allows Samba to correctly handle a 128 bit session key in all of the
netr_ServerAuthenticateX() varients. A huge thanks to Luke Howard
from PADL for solving this one!
- restructured the server side rpc authentication to allow for other
than NTLMSSP sign and seal. This commit just adds the structure, the
next commit will add schannel server side support.
- added 128 bit session key support to our client side code, and
testing against w2k3 with smbtorture. Works well.
(This used to be commit 729b2f41c9)
servers. Previously the server pipe code needed to return the RPC
level status (nearly always "OK") and separately set the function call
return using r->out.result. All the programmers writing servers
(metze, jelmer and me) were often getting this wrong, by doing things
like "return NT_STATUS_NO_MEMORY" which was really quite meaningless
as there is no code like that at the dcerpc level.
I have now modified pidl to generate the necessary boilerplate so that
just returning the status you want from the function will work. So for
a NTSTATUS function you return NT_STATUS_XXX and from a WERROR
function you return WERR_XXX. If you really want to generate a DCERPC
level fault rather than just a return value in your function then you
should use the DCESRV_FAULT() macro which will correctly generate a
fault for you.
As a side effect, this also adds automatic type checking of all of our
server side rpc functions, which was impossible with the old API. When
I changed the API I found and fixed quite a few functions with the
wrong type information, so this is definately useful.
I have also changed the server side template generation to generate a
DCERPC "operation range error" by default when you have not yet filled
in a server side function. This allows us to correctly implement
functions in any order in our rpc pipe servers and give the client the
right information about the fault.
(This used to be commit a4df5c7cf8)
common case of there being no pending partial data and a full dcerpc
packet being received.
We should use this same model for the smb server. It gives us
efficient memory usage while allowing for completely async socket
operations.
(This used to be commit 9aab321fb6)
1.) We now register endpoint servers add startup via register_backend()
and later use the smb.conf 'dcerpc endpoint servers' parameter to setup the dcesrv_context
2.) each endpoint server can register at context creation time as much interfaces as it wants
(multiple interfaces on one endpoint are supported!)
(NOTE: there's a difference between 'endpoint server' and 'endpoint'!
for details look at rpc_server/dcesrv_server.h)
3.) one endpoint can have a security descriptor registered to it self
this will be checked in the future when a client wants to connect
to an smb pipe endpoint.
4.) we now have a 'remote' endpoint server, which works like the ntvfs_cifs module
it takes this options in the [globals] section:
dcerpc remote:interfaces = srvsvc, winreg, w32time, epmapper
dcerpc remote:binding = ...
dcerpc remote:user = ...
dcerpc remote:password = ...
5.) we currently have tree endpoint servers: epmapper, rpcecho and remote
the default for the 'dcerpc endpiont servers = epmapper, rpcecho'
for testing you can also do
dcerpc endpoint servers = rpcecho, remote, epmapper
dcerpc remote:interfaces = srvsvc, samr, netlogon
6,) please notice the the epmapper now only returns NO_ENTRIES
(but I think we'll find a solution for this too:-)
7.) also there're some other stuff left, but step by step :-)
This patch also includes updates for the
register_subsystem() , ntvfs_init(), and some other funtions
to check for duplicate subsystem registration
metze
(hmmm, my first large commit...I hope it works as supposed :-)
(This used to be commit 917e45dafd)
uint32 followed by a GUID. I needed to fix this to support running in
mixed-mode rpc (where smbtorture is bigendian and w2k3 is
little-endian). Otherwise when you send back a policy handle the
server doesn't recognise it.
(This used to be commit 9b1c76a8e9)
This adds support for bigendian rpc in the client. I have installed
SUN pcnetlink locally and am using it to test the samba4 rpc
code. This allows us to easily find places where we have stuffed up
the types (such as 2 uint16 versus a uint32), as testing both
big-endian and little-endian easily shows which is correct. I have now
used this to fix several bugs like that in the samba4 IDL.
In order to make this work I also had to redefine a GUID as a true
structure, not a blob. From the pcnetlink wire it is clear that it is
indeed defined as a structure (the byte order changes). This required
changing lots of Samba code to use a GUID as a structure.
I also had to fix the if_version code in dcerpc syntax IDs, as it
turns out they are a single uint32 not two uint16s.
The big-endian support is a bit ugly at the moment, and breaks the
layering in some places. More work is needed, especially on the server
side.
(This used to be commit bb1af644a5)
I needed to hack the ntlmssp code a little, as the auth code in samba4
is out of date relative to the samba3 auth code. I need to do a merge :)
(This used to be commit 6ee0935afe)
server endpoints. We can now successfully setup listening endpoints on
high ports, then use our endpoint mapper redirect incoming clients to
the right port.
also greatly cleanup the rpc over tcp session handling.
(This used to be commit 593bc29bbe)
needs quite a bit more work to get it finished. The biggest missing
feature is the lack of NTLMSSP which is needed for basic
authentication over tcp
(This used to be commit 9fb0f03693)
for some reason the epm_Lookup replies can't be parsed by ethereal,
although w2k parses then fine as does the Samba4 NDR code.
(This used to be commit 097e7ca99d)
implements the epm_Lookup() call, I'll add the other important calls
soon. I was rather pleased to find that epm_Lookup() worked first
time, which is particularly surprising given its complexity.
This required quite a bit of new infrastructure:
* a generic way of handling dcerpc policy handles in the rpc server
* added type checked varients of talloc. These are much less error
prone. I'd like to move to using these for nearly all uses of
talloc.
* added more dcerpc fault handling code, and translation from
NTSTATUS to a dcerpc fault code
* added data_blob_talloc_zero() for allocating an initially zero
blob
* added a endpoint enumeration hook in the dcerpc endpoint server
operations
(This used to be commit 3f85f9b782)
don't cause fragmented pdus (I'll add fragments shortly)
* change data_blob_talloc() to not zero memory when the 2nd argument
is NULL. The zeroing just masks bugs, and can't even allow a DOS
attack
* modified pidl to ensure that [ref] arguments to the out side of
functions are allocated when parsing the in side. This allows rpc
backends to assume that [ref] variables are all setup. Doesn't work
correctly for [ref] arrays yet
* changed DLIST_ADD_END() to take the type instead of a tmp
variable. This means you don't need to declare a silly tmp variable in
the caller
(This used to be commit 46e0a35819)
