Jeremy Allison [Wed, 9 Mar 2005 00:06:27 +0000 (00:06 +0000)]
r5702: Fix bug #2271. Correctly pull out and use resume names in a
directory listing (we were incorrectly understanding what was
returned in the "last name" entry).
Jeremy.
Jeremy Allison [Tue, 8 Mar 2005 23:03:38 +0000 (23:03 +0000)]
r5700: Fix problems where we're not pointing to the start of entry for a
couple of info levels - W2K3 always points to the start, not the name.
Jeremy.
Gerald Carter [Tue, 8 Mar 2005 17:22:39 +0000 (17:22 +0000)]
r5691: wrapping the pause/resume/purge printer commands in
{become,unbecome}_root() blocks. We've already done
a print_access_check() to ensure the user is admin.
The means that non-root users can pause and manage printers.
I really don't see how this worked before without setuid
binaries on the server.
Also update print_queue_update() interface to allow an smbd
to update the print queue cache locally rather than going through
the bg lpq daemon. This is needed for things like pjob_delete()
to ensure the cache is current for the specific client.
Jeremy Allison [Sat, 5 Mar 2005 01:22:53 +0000 (01:22 +0000)]
r5655: Added support for Novell NDS universal password. Code donated by
Vince Brimhall <vbrimhall@novell.com> - slight tidyup by me to
use Samba conventions.
Vince - thanks a *lot* for this code - please test to make sure
I haven't messed anything up.
Jeremy.
Jeremy Allison [Thu, 3 Mar 2005 03:43:57 +0000 (03:43 +0000)]
r5636: Re-add the allocation size - parameterized by share as
"allocation roundup size", by default set as 1Mb. From
advice by BlueArc about Windows client behaviour. VC++
people can set this to zero to turn it off.
Jeremy.
Jeremy Allison [Thu, 3 Mar 2005 02:07:00 +0000 (02:07 +0000)]
r5634: Fix 64-bit overflow problems found by BlueArc torture tester.
We still have a few strange bugs with 64-bit locking values. I will
get traces.
Jeremy.
Jeremy Allison [Thu, 3 Mar 2005 02:04:36 +0000 (02:04 +0000)]
r5632: Fix infinite looping bug found by nasty BlueArc test :-).
When finding a singleton directory remember that we're
at the end and don't continuously return the same name.
Jeremy.
Jeremy Allison [Wed, 2 Mar 2005 18:19:32 +0000 (18:19 +0000)]
r5623: Horrible band-aid patch to fix Blue-Arc torture tester.
I know this isn't right but will work until I can refactor
the deny mode code with the Samba4 algorithm.
Jeremy.
Jeremy Allison [Wed, 2 Mar 2005 03:41:44 +0000 (03:41 +0000)]
r5616: Forgot about the sticky bit on directories (commonly set on /tmp). If this is set
then only the owner or root can delete a file. We now use
the same algorithm to check file delete.
Jeremy.
Jeremy Allison [Mon, 28 Feb 2005 07:01:52 +0000 (07:01 +0000)]
r5589: Ensure the torture code that uses cli_list and cli_list_old
matches the new MS-DFS function prototypes. Should fix the
build farm tests.
Jeremy.
Jim McDonough [Sat, 26 Feb 2005 15:26:55 +0000 (15:26 +0000)]
r5580: Fix "net rpc trustdom add". Much closer to what windows does. Also stop
referencing unknown_6 from sam, because it's just fixed at 1260, the max
len of LOGON_HRS. Need to go in and mark it as "remove me" from passdb.
Gerald Carter [Sat, 26 Feb 2005 15:03:16 +0000 (15:03 +0000)]
r5578: get 'recurse; dir' working across multiple levels of dfs referrals
note that this does not handle the situation where the same \\server\share
is mounted mutliple times in the dfs tree since I store a single mount
path per struct cli_state *
Gerald Carter [Fri, 25 Feb 2005 15:56:13 +0000 (15:56 +0000)]
r5562: * bump version to 3.0.12pre2
* change special character in gd's valid workstation
check to a '+' to be more in line with the characters
used by valid users
Jeremy Allison [Fri, 25 Feb 2005 02:22:44 +0000 (02:22 +0000)]
r5548: Stop lying about allocation sizes to Windows clients. It was a nice
idea, and aparently improved performance in some circumstances, but it
breaks the VC++ compiler :-(. Not cool. Fix bug #2146.
Jeremy.
Gerald Carter [Thu, 24 Feb 2005 21:54:52 +0000 (21:54 +0000)]
r5545: move cli_cm_XXX() connection handling code to clidfs and out of client.c; client.c still maintains a pointer to the first connection so the change is fairly reansparent to other smbclient functions such as -L and -M
Jeremy Allison [Thu, 24 Feb 2005 21:06:11 +0000 (21:06 +0000)]
r5543: Fix for bug #962 - using MB sharenames containing a ']' character.
Processing a share name is now MB safe so long as the correct
unix charset is in scope.
Jeremy.
Günther Deschner [Thu, 24 Feb 2005 00:26:24 +0000 (00:26 +0000)]
r5528: Expand the invalid-workstation-scheme. Workstation-Names with leading
'@'-sign are expanded on-the-fly as posix-groups of workstations. This
allows optional, more flexible login-control in larger networks.
Gerald Carter [Wed, 23 Feb 2005 19:26:32 +0000 (19:26 +0000)]
r5520: fix last remaining dfs issues with smbclient.
* all the unix extension commands should work
* send the correct TRANS2_FINDFIRST format to 2k to
get a listing from a msdfs root share (tested against
smbd as well).
* mkdir, rmdir, etc... all seem ok.
I'm sure bugs will pop up so keep testing.
Last thing I plan on doing is to clean up the horrible
mess with connection management in smbclient and global
variables (so i can move the cli_cm_xx() routines to a
separate file).
Gerald Carter [Wed, 23 Feb 2005 17:29:28 +0000 (17:29 +0000)]
r5518: Add initial msdfs support to smbclient. Currently I can only
cd up and down the tree and get directory listings.
Still have to figure out how to get a directory listing on a
2k dfs root. Also have to work out some issues with relative paths
that cross dfs mount points.
We're protected from the new code paths when connecting to
a non-dfs root share ( the flag from the tcon&X is stored
in the struct cli_state* )
Gerald Carter [Wed, 23 Feb 2005 16:36:44 +0000 (16:36 +0000)]
r5517: code cleanup; rename the sorted_tree to pathtree (used by registry code)
I was going to use this for tracking dfs mounts in smbclient
but found another way. Still the cleanup is valid so commiting it.
should be minimally disruptive since it is not widely used.
Jeremy Allison [Tue, 22 Feb 2005 20:55:48 +0000 (20:55 +0000)]
r5510: Optimisation to only do can_delete check if client asks for FILE_SHARE_DELETE.
Not completely correct but will catch the XP SP2 problem.
Jeremy.
Jeremy Allison [Tue, 22 Feb 2005 05:11:37 +0000 (05:11 +0000)]
r5497: Fix for DIR1 failures in build farm. It struck me that we
only care about failing with ACCESS_DENIED if we can't delete
with DELETE access requested. All other errors will be processed
as normal.
Jeremy.
Gerald Carter [Tue, 22 Feb 2005 03:31:22 +0000 (03:31 +0000)]
r5495: * add in some code from Mike Nix <mnix@wanm.com.au> for the SMBsplopen
and SMBsplclose commands (BUG 2010)
* clarify some debug messages in smbspool (also from Mike)
my changes:
* start adding msdfs client routines
* enable smbclient to maintain multiple connections
* set the CAP_DFS flag for our internal clienht routines.
I actualy have a dfs referral working in do_cd() but that code
is too ugly to live so I'm not checking it in just yet.
Further work is to merge with vl's changes in trunk to support multiple
TIDs per cli_state *.
Volker Lendecke [Sun, 20 Feb 2005 15:15:33 +0000 (15:15 +0000)]
r5471: In cli_samr_lookup_rids, flags is not a flags but an array size. W2k3 rejects
everything but 1000 here, so there's no point in exposing that to the caller.
Volker Lendecke [Sun, 20 Feb 2005 13:47:16 +0000 (13:47 +0000)]
r5467: Optimize _samr_query_groupmem with LDAP backend for large domains.
Could someone else please look at this patch, verifying that I did not break
the ldapsam:trusted = False fallback to the old behaviour? It works fine for
me, but you never know. You're certainly free to review the new code as well :-)
Jim McDonough [Sat, 19 Feb 2005 11:09:52 +0000 (11:09 +0000)]
r5460: Fix "restrict anonymous = 1". If we have schannel connection, we must be
validated with a user, so allow it even if pipe itself had an anonymous
connection.
Jim McDonough [Sat, 19 Feb 2005 10:56:40 +0000 (10:56 +0000)]
r5458: Generate a sane response to exceeding lookupsids limit. Truncate list to zero
and return NT_STATUS_NONE_MAPPED. This does not crash windows and maintains
the benefit of not overallocating memory. The previous response of
truncating to the MAX limit was not useful because it crashed lsass.exe on
windows (bug opened with MS), and it was also misleading the client to
believe that a complete answer was received.
Jim McDonough [Sat, 19 Feb 2005 10:41:51 +0000 (10:41 +0000)]
r5456: Increase limit of mapped SIDS to 0x5000, which is what 2k and later do.
NT has no limit. We still don't respond the way 2k would to requests
larger, which is to actually allocate the memory and send back the entire
response, plus a return status of NT_STATUS_NONE_MAPPED. Still looking
into ways of doing this without crashing windows.
and *any* change in your entire DIT will show up immediately (after a
second change of an object even as object diff). It's very interesting
to see the interaction of GPOs, the various steps of account
modification with dsa.msc, etc.
Gracefully ignore some parts of adssearch.pl that are rather immature...
Volker Lendecke [Thu, 17 Feb 2005 14:27:34 +0000 (14:27 +0000)]
r5428: Apply some const. LDAP attribs should now be declared const char *attr[]. This
gives some new warnings in smbldap.c, but a the callers are cleaned up.
projects / thirdparty / samba.git / log
