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.
effective gid which could mean that the user gets group 0 in their
group list for acl interpretation
this is a replacement fix for the one richard did in 2.2 (which didn't
cope wiith variable behaviour depending on which nss module was in
use)
(This used to be commit cfc5ca3416)
security descriptors. We need to calculate the maximum offset and set the offset
back after reading/writing every field in the SEC_DESC.
This was *nasty* to find....
Jeremy.
(This used to be commit 175d43980e)
a separate tdb per printer, but only keeps (currently one) tdb open at
a time (although this is easily changed by changing a #define). Needs
scalability testing with large numbers of printers now....
Jeremy.
(This used to be commit b0909cfa14)
server. The real problem is all the special cases we had for when we
are a wins server as opposed to when we are using a 'real' wins
server.
This patch removes the special cases. We now accept non-broadcast
packets from ourselves and we use ourselves as a wins server when we
are one. This gets rid of the special cases and simplifies things
quite a bit.
It all seems to work, but there are bound to be problems found later.
(This used to be commit 3e843d3015)
Add a function to display 'sid types' as strings - makes rpcclient outptut
and DEBUG() logs much eaiser to understand.
Move the enum for SID types to smb.h, becouse is really isn't LSA specific any
more.
Andrew Bartlett
(This used to be commit fc97398611)
We now cope wiith multiple WINS groups and multiple failover servers
for release and refresh as well as registration. We also do the regitrations
in the same fashion as W2K does, where we don't try to register the next
IP in the list for a name until the WINS server has acked the previos IP.
This prevents us flooding the WINS server and also seems to make for much
more reliable multi-homed registration.
I also changed the dead WINS server code to mark pairs of IPs dead,
not individual IPs. The idea is that a WINS server might be dead from
the point of view of one of our interfaces, but not another, so we
need to keep talking to it on one while moving onto a failover WINS
server on the other interface. This copes much better with partial
LAN outages and weird routing tables.
(This used to be commit 313f2c9ff7)
accept an extended syntax for 'wins server' like this:
wins server = group1:192.168.2.10 group2:192.168.3.99 group1:192.168.0.1
The tags before the IPs don't mean anything, they are just a way of
grouping IPs together. If you use the old syntax (ie. no ':') then
an implicit group name of '*' is used. In general I'd recommend people
use interface names for the group names, but it doesn't matter much.
When we register in nmbd we try to register all our IPs with each group
of WINS servers. We keep trying until all of them are registered with
every group, falling back to the failover WINS servers for each group
as we go.
When we do a WINS lookup we try each of the WINS servers for each group.
If a WINS server for a group gives a negative answer then we give up
on that group and move to the next group. If it times out then
we move to the next failover wins server in the group.
In either case, if a WINS server doesn't respond then we mark it dead
for 10 minutes, to prevent lengthy waits for dead servers.
(This used to be commit e125f06058)
gives us a good grounding to properly support multiple wins servers
for different interfaces (which will be coming soon ...)
- fixed our wins registration failover code to actually do failover!
We were not trying to register with a secondary wins server at all
when the primary was down. We now fallback correctly.
- fixed the multi-homed name registration packets so that they work
even in a non-connected network (ie. when one of our interfaces is not
routable from the wins server. Yes, this really happens in the real
world).
(This used to be commit a049360d5b)
The aim of this execise is to give the 'security>=user' code a straight paper
path. Security=share will sill call authorise_login(), but otherwise we avoid
that mess.
This allow *much* more accurate error code reporting, beocuse we don't start
pretending that we can use the (nonexistant) password etc.
Also in this patch is code to create the 'homes' share at session setup time
(as we have done in the past - been broken recently) and to record this on
the user's vuser struct for later reference. The changes here should also
allow for much better use of %H (some more changes to come here).
The service.c changes move a lot of code around, but are not as drastric
as they look...
(Also included is a fix to srv_srvsvc_nt.c where 'total_entries' not
'*total_entries' was compared).
This code is needs testing, but passes my basic tests.
I expect we have lost some functionality, but the stuff I had expected
to loose was already broken before I started. In particular, we don't 'fall
back' to guest if the user cannot access a share (for security=user). If you
want this kind of stuff then you really want security=share anyway.
Andrew Bartlett
(This used to be commit 4c0cbcaed9)
It extends the 'server mutex' to conver security=server, becouse the connection
race condition exists here too, and while people *should* use security=domain,
some sites don't....
(This probably should be done in 2.2 as well).
Also, start to actually extract and use the information that the remote
server returns in the info3 struct.
The server mutex code is now in a new file.
Andrew Bartlett
(This used to be commit 9b0dabdf4e)
rather than a string when configuring mulitple backends.
Also adjust some of the users of get_global_sam_sid() to cope with the fact
that it just might not exist (uninitialised, can't access secrets.tdb).
More places need conversion.
Add some const and remove silly casts.
Andrew Bartlett
(This used to be commit c264bf2ec9)
to using SIDs instead of RIDs.
The new funciton sid_peek_check_rid() takes an 'expected domain sid' argument.
The idea here is to prevent mistakes where the SID is implict, but isn't
the same one that we have in the struct.
Andrew Bartlett
(This used to be commit 04f9a8ff4c)
the (now static) global_sam_sid.
The only place it was being used was to return global_sid_NULL to some
uid->sid functions - and I'm not convinced this is correct in any case.
Andrew Bartlett
(This used to be commit e2a76a7fc9)
initialising function. This patch thanks to the work of
"Stefan (metze) Metzmacher" <metze@metzemix.de>
This is partly to enable the transition to SIDs in the the passdb.
Andrew Bartlett
(This used to be commit 96afea638e)
a file that is linked with the passdb.
This is to avoid linking insanity when this global becomes a self-initing
function.
(This used to be commit 743afd96cb)
to correctly allow password changes on expired passwords. (No security
implications, as its just a 'will I let you talk to the server' check).
pam_winbind checks the password prior to changing it, so that users don't
have to make up and type their new password when they havn't even got the
old one right. This also helps with stacking etc.
Andrew Bartlett
(This used to be commit 2b78d49300)
Importantly:
The removal of the silly 'delete user script' behaviour when secuity=domain.
I have left the name the same - as it still does the (previously documented,
but not in smb.conf(5)) sane behaviour of deleting users on request.
When we decide what to do with the 'add user' functionality, we might
rename it.
Andrew Bartlett
(This used to be commit cdcfe3671e)
These might be reimplmented as simple pass-through functions, but all users
really should be doing 'getpwnam_alloc' or 'getpwuid_alloc' to ensure that
there are not shared static buffers.
I don't beleive we actually need a getpw*() cache inside samba - if we do
then I think we should look at our code design first.
(some of these changes are for platforms I don't have access to, but
they look sane)
Andrew Bartlett
(This used to be commit 9d8431b04f)
Make some code static, add some const to the PAM code, and make the plaintext
password code actually function - particulary without the requirement to
modify the 'struct passwd' (which it assumed was made up of fstrings)
This kills some particularly ugly code in lib/util_pw.c
Andrew Bartlett
(This used to be commit 302dad4990)
the DATA_BLOB code into its own file.
It would be nice to go over some of the other util.c functions, and check
that we still use them all, and that we use them in more than one place.
Andrew Bartlett
(This used to be commit d0ea70fce5)
this is a first step only passdb stuff has beein "classized".
- so what can you do?
set debug level to: 1 poasdb:10
that will make all the code run at debug level 1 except the code in
passdb/* files that will run at level 10
TODO: fix the man page
- also smbcontrol has this nice feature so smbcontrol smbd debug 3 passdb:5
will set every smbd to have a default log level of 3 while passdb stuff
will be at level 5
and so no..
minor cosmetic fix to pdbedit is there too
(This used to be commit be5c3b3f57)
(only function that used it was unused, and this helps bring TNG and HEAD
closer)
Its also cleaner.
Andrew Bartlett
(This used to be commit 78f47c8333)
cleanup some of the code in net_rpc_join re const warnings and
fstrings.
Passdb:
Make the %u and %U substituions in passdb work.
This is done by declaring these paramters to be 'const' and doing
the substitution manually. I'm told this is us going full circle,
but I can't really see a better way.
Finally these things actually seem to work properly...
Make the lanman code use the pdb's recorded values for homedir etc
rather than the values from lp_*()
Add code to set the plaintext password in the passdb, where it can
decide how to store/set it. For use with a future 'ldap password
change' option, or somthing like that...
Add pdb_unix, so as to remove the 'not in passdb' special cases from the
local_lookup_*() code. Quite small, as it uses the new 'struct passwd ->
SAM_ACCOUNT' code that is now in just one place. (also used by pdb_smbpasswd)
Other:
Fix up the adding of [homes] at session setup time to actually pass
the right string, that is the unix homedir, not the UNC path.
Fix up [homes] so that for winbind users is picks the correct name.
(bad interactions with the default domain code previously)
Change the rpc_server/srv_lsa_nt.c code to match NT when for the
SATUS_NONE_MAPPED reply: This was only being triggered on
no queries, now it is on the 'no mappings' (ie all mappings failed).
Checked against Win2k.
Policy Question: Should SID -> unix_user.234/unix_group.364 be
considered a mapping or not? Currently it isn't.
Andrew Bartlett
(This used to be commit c28668068b)
This option was badly maintained, useless and confused our users and
distirbutors. (its SSL, therfore it must be good...)
No windows client uses this protocol without help from an SSL tunnel.
I can't see any reason why setting up a unix-side SSL wrapper would
be any more difficult than the > 10 config options this mess added
to samba in any case.
On the Samba client end, I think the LIBSMB_PROG hack should be
sufficient to start stunnel on the unix side. We might extend this
to take %i and %p (IP and port) if there is demand.
Andrew Bartlett
(This used to be commit b04561d3fd)
(ie. ENOTDIR) to the NT status code NT_STATUS_NOT_A_DIRECTORY. NT seems
to use NT_STATUS_OBJECT_PATH_NOT_FOUND. I'm hoping this will fix the
access binaries served from a Samba share bug...
Jeremy.
(This used to be commit 6f2b76c239)
this was a very nasty bug with filename corruption and NT4 clients. The
exact termination conditions are quite critical ...
(This used to be commit a538efe7d0)
Thanks to Ollie Oldham <ollie.oldham@metro-optix.com> for spotting it.
few mods to make it easier to compile the tests.
addedd the "Ollie" test to the floating point ones.
(This used to be commit 415f9d92bc)
<mimir@diament.ists.pwr.wroc.pl>) this patch allows samba to correctly
enumerate its trusted domains - by exaimining the keys in the secrets.tdb file.
This patch has been tested with both NT4 and rpcclient/wbinfo, and adds
some extra functionality to talloc and rpc_parse to allow it to deal with
already unicode strings.
Finally, this cleans up some const warnings that were in net_rpc.c by pushing
another dash of const into the rpc client code.
Andrew Bartlett
(This used to be commit 0bdd94cb99)
All uids and gids must create valid RIDs, becouse other code expects this, and
can't handle the failure case. (ACL code in particular)
Allow admins to adjust the base of the RID algorithm, so avoid clashes with
users brought in from NT (for example).
Put all the algorithm code back in one place, so that this change is global.
Better coping with NULL sid pointers - but it still breaks a lot of stuff.
BONUS: manpage entry for new paramater :-)
counter based rids for normal users in tdbsam is disabled for the timebeing,
idra and I will work out some things here soon I hope.
Andrew Bartlett
(This used to be commit 5275c94cdf)
the hash for this scheme is *much* larger (approximately 31 bits) and
the code is written to be very fast, correctly handling multibyte
while not doing any actual multi-byte conversions in the vast majority
of cases
you can select this scheme using "mangling method = hash2", although I
may make it the default if it works out well.
(This used to be commit bb173c1a7e)
fixed tdbsam memory corruption (and segfault)
reducing calls to pdb_uid_to_user_rid and countrary to 0 to move to a non alghoritmic rid allocation with some passdb modules.
(This used to be commit 9836af7cd6)
in the reverse).
* add in new printer change notify code from SAMBA_2_2
* add in se_map_standard() from 2.2 in _spoolss_open_printer_ex()
* sync up the _print_queue_struct in smb.h (why did someone change the
user/file names in fs_user/fs_file (or vice-versa) ? )
* sync up some cli_spoolss_XXX functions
(This used to be commit 5760315c1d)
this is needed because W2K will send a TCP reset to any open
connections that have not done a negprot when a second connection is
made. This meant that under heavy netlogon load a Samba domain member
would fail authentications.
Jeremy, you may wish to port this to 2.2.x
(This used to be commit eb196070e6)
receive_smb: You might think that we ought to set smb_read_error here,
but apparently that breaks the recursive main loop in oplock.c.
Global variables suck. :-/
(This used to be commit b6d5d02aa1)
This adds code to do generic PAM -> NTSTATUS and NTSTATUS -> PAM error
conversions, and uses them to make the error handling in pam_winbind sane.
In particular, pam_winbind now uses PAM error codes, not silly '-1, -2 ...'
stuff, and logs the NTSTATUS error that winbind now sends over the pipe.
Added code to wbinfo to display these - makes a big difference in debugging
winbindd.
The main change here is the code to allow pam_winbind password changing to
correctly stack - This code ripped from pam_unix, and the copyright attached.
(Same as for all pam modules, including pam_winbind)
Andrew Bartlett
(This used to be commit dc1a72f896)
processing work correctly in winbindd. This is a really good patch
that gives full select semantics to the Samba modified select.
Jeremy.
(This used to be commit 3af16ade17)
non-domain Samba server from a NT4 client.
Note that this exactly reverses a change by Jeremy on the 18th of
December 2001, reverting the code back to what JF originally wrote. I
have looked carefully with a sniffer and JFs original NULL sid is
correct (ie. it matches what NT4 does) and also fixes the problem.
Sending a blank sid (which is what jeremy's patch did) causes NT4 to
give a classic "parameter is incorrect error" and prevents the
addition of new ACLs.
(This used to be commit 9930cf9733)
<a.bokovoy@sam-solutions.net>.
This patch is designed to remove the 'special cases' required for this support.
In particular this now kills off winbind_initgroups, as it appears no longer to
be required.
Andrew Bartlett
(This used to be commit f1d8d50976)
this might need a bit more work - or at least documentation.
This is certainly a worthwile little hack, as it avoids the need to invert the
group database. I don't think we should allow unqualified domains here - as
that allows us to distinguish between (at least some) usernames and these
'special' groups.
(This used to be commit 151dd7bc6c)
correctly) be no longer needed. This is in aid of the 'winbind default domain'
code - which works much better when smbd always goes via the standard unix
interfaces.
Andrew Bartlett
(This used to be commit a41fe2f6c8)
routines can work. The code was copied into both convert_string and
convert_string_allocate -- I split it into a little static function,
and removed an apparently duplicate call to init_valid_table().
(This used to be commit 7f7d22880d)
degree of seperation betwen reading/writing the raw NamedPipe SMB packets
and the matching operations inside smbd's RPC components.
This patch is designed for no change in behaviour, and my tests hold that to be
true. This patch does however allow for the future loadable modules interface
to specify function pointers in replacement of the fixed state.
The pipes_struct has been split into two peices, with smb_np_struct taking the
information that should be generic to where the data ends up.
Some other minor changes are made: we get another small helper function in
util_sock.c and some of the original code has better failure debugs and
variable use. (As per on-list comments).
Andrew Bartlett
(This used to be commit 8ef13cabdd)
smbd, and also makes it much cleaner inside winbindd.
It is mostly my code, with a few changes and testing performed by Alexander
Bokovoy <a.bokovoy@sam-solutions.net>. ab has tested it in security=domain and
security=ads, but more testing is always appricatiated.
The idea is that we no longer cart around a 'domain\user' string, we keep them
seperate until the last moment - when we push that string into a pwent on onto
the socket.
This removes the need to be constantly parsing that string - the domain prefix
is almost always already provided, (only a couple of functions actually changed
arguments in all this).
Some consequential changes to the RPC client code, to stop it concatonating the
two strings (it now passes them both back as params).
I havn't changed the cache code, however the usernames will no longer have a
double domain prefix in the key string. The actual structures are unchanged
- but the meaning of 'username' in the 'rid' will have changed. (The cache is
invalidated at startup, so on-disk formats are not an issue here).
Andrew Bartlett
(This used to be commit e870f0e727)
<a.bokovoy@sam-solutions.net>.
The idea is the domain\username is rather harsh for unix systems - people don't
expect to have to FTP, SSH and (in particular) e-mail with a username like
that.
This 'corrects' that - but is not without its own problems.
As you can see from the changes to files like username.c and wb_client.c (smbd's
winbind client code) a lot of assumptions are made in a lot of places about
lp_winbind_seperator determining a users's status as a domain or local user.
The main change I will shortly be making is to investigate and kill off
winbind_initgroups() - as far as I know it was a workaround for an old bug in
winbind itself (and a bug in RH 5.2) and should no longer be relevent.
I am also going to move to using the 'winbind uid' and 'winbind gid' paramaters
to determine a user/groups's 'local' status, rather than the presence of the
seperator.
As such, this functionality is recommended for servers providing unix services,
but is currently less than optimal for windows clients.
(TODO: remove all references to lp_winbind_seperator() and
lp_winbind_use_default_domain() from smbd)
Andrew Bartlett
(This used to be commit 07a21fcd23)
Samba (ab)uses the returns from getpwnam() a lot - in particular it keeps
them around for a long time - often past the next call...
This adds a getpwnam_alloc and a getpwuid_alloc to the collection.
These function as expected, returning a malloced structure that can be
free()ed with passwd_free(&passwd).
This patch also cuts down on the number of calls to getpwnam - mostly by
taking advantage of the fact that the passdb interface is already
case-insensiteve.
With this patch most of the recursive cases have been removed (that I know
of) and the problems are reduced further by not using the sys_ interface
in the new code. This means that pointers to the cache won't be affected.
(This is a tempoary HACK, I intend to kill the password cache entirly).
The only change I'm a little worried about is the change to
rpc_server/srv_samr_nt.c for private groups. In this case we are getting
groups from the new group mapping DB. Do we still need to check for private
groups? I've toned down the check to a case sensitve match with the new code,
but we might be able to kill it entirly.
I've also added a make_modifyable_passwd() function, that copies a passwd
struct into the form that the old sys_getpw* code provided. As far as I can
tell this is only actually used in the pass_check.c crazies, where I moved
the final 'special case' for shadow passwords (out of _Get_Pwnam()).
The matching case for getpwent() is dealt with already, in lib/util_getent.c
Also included in here is a small change to register the [homes] share at vuid
creation rather than just in one varient of the session setup. (This picks
up the SPNEGO cases). The home directory is now stored on the vuid, and I
am hoping this might provide a saner way to do %H substitions.
TODO: Kill off remaining Get_Pwnam_Modify calls (they are not needed), change
the remaining sys_getpwnam() callers to use getpwnam_alloc() and move
Get_Pwnam to return an allocated struct.
Andrew Bartlett
(This used to be commit 1d86c7f942)
to move this from being a static to matching its mate in lib/util_sock.c.
In any case, this should discorage anybody from using the 'wrong' version of
this function. (ie the one from TNG, which needs a bit more error checking
depending on use).
Andrew Bartlett
(This used to be commit e6a3a01f79)
Also change the structure so it has its own (optional) 'free' pointer - so we
don't free() a talloc'ed version.
also split out the data_blob_clear() functionaility.
Andrew Bartlett
(This used to be commit 207ee8aac4)
