1
0
mirror of https://github.com/samba-team/samba.git synced 2025-10-23 11:33:16 +03:00
Commit Graph

5551 Commits

Author SHA1 Message Date
Andrew Tridgell
f48abaaaca r8753: fixed directory handling on systems that do not return . and .. as the
first two entries in a directory. This is what caused the FC3 system
shelob in the build farm to fail the RAW-UNLINK and RAW-SEARCH tests.
2007-10-10 13:30:02 -05:00
Andrew Bartlett
1e0f7792bb r8752: With all the infrustructure done, details like a SamSync migration
into LDB are actually quite easy.

This brings us the users, and sets basic domain information.

You are expected to have provisioned with the settings for the target
domain, and have joined the domain as a BDC.  Then simply 'net
samsync'.

Now we just need to flesh out the delta types.

Andrew Bartlett
2007-10-10 13:30:02 -05:00
Andrew Tridgell
d974bf3589 r8750: drat, on some systems dirfd() is a macro 2007-10-10 13:30:02 -05:00
Andrew Tridgell
ce022e4090 r8749: for completeness, add rewinddir() and dirfd() 2007-10-10 13:30:02 -05:00
Andrew Tridgell
9411bd4e5e r8748: fixed build. Andrew, please check. 2007-10-10 13:30:01 -05:00
Andrew Tridgell
60a07da59d r8747: remove unused code 2007-10-10 13:30:01 -05:00
Andrew Tridgell
e1bf7c4279 r8746: replace opendir/readdir/telldir/seekdir/closedir on systems where they
are broken (apparently all BSD systems). This breakage leads to unlink
on files in an open directory causing a later seekdir to miss
files. The bug happens due to a block boundary bug in the BSD libc
implementation of these calls.

This replacement code also fixes a severe memory usage problem with
telldir that can cause closedir() to take an arbitrary amount of time.

I have reported the bug in readdir to Greg Lehey (a FreeBSD maintainer)
2007-10-10 13:30:01 -05:00
Andrew Bartlett
2eadcf4669 r8744: Split 'net samdump' out into a separate file
Work on the talloc memory tree, as I think talloc_reference and other
things were biting me.

Crush unions in the name of code reform. ;-)

Andrew Bartlett
2007-10-10 13:30:01 -05:00
Andrew Tridgell
8086c37df6 r8742: fixed handling of zero length names in mprObject() 2007-10-10 13:30:00 -05:00
Andrew Bartlett
258e5e302e r8741: Kill warnings about enums not fully enumerated, as we will never use
all the branches.

Andrew Bartlett
2007-10-10 13:30:00 -05:00
Andrew Bartlett
0cc58f5c3c r8740: Extend the rdn_name module to handle adding the rdn as an attribute. ie:
dn: cn=foo,ou=bar
objectClass: person

implies

dn: cn=foo,ou=bar
objectClass: person
cn: foo
(as well as a pile more default attributes)

We also correct the case in the attirbute to match that in the DN
(win2k3 behaviour) and I have a testsuite (in ejs) to prove it.

This module also found a bug in our provision.ldif, so and reduces
code complexity in the samdb module.

Andrew Bartlett
2007-10-10 13:30:00 -05:00
Andrew Tridgell
7b37923b72 r8736: this fixes the ldb speed (raises BENCH-WINS from 15 ops/sec to over 4000)
simo, we need to be careful to cope with this sort of direct dn query efficiently
2007-10-10 13:30:00 -05:00
Andrew Tridgell
519812a9e8 r8735: added NBT and WINS testing to 'make test' so we will know if it breaks again 2007-10-10 13:30:00 -05:00
Andrew Tridgell
83727bf72c r8734: fixed the wins server for the new ldb DN restrictions.
This gets it working, but I'm rather worried about the speed. We used
to get more than 5000 ops/sec, but now we are down to around 15
ops/sec. I suspect a bug in ldb.
2007-10-10 13:29:59 -05:00
Andrew Tridgell
8c4d0cd7b7 r8729: make the RAW-SEARCH test more robust to servers with hash based directory ordering 2007-10-10 13:29:59 -05:00
Andrew Tridgell
d40e590674 r8726: increase default session timeout to 15 minutes to be less painful for developers 2007-10-10 13:29:59 -05:00
Rafal Szczesniak
1a779f8643 r8721: Further work on libnet_rpc_usermod function. Now it can change
both account name and full name.

rafal
2007-10-10 13:29:58 -05:00
Rafal Szczesniak
4b3aa69e70 r8717: Add monitor message emiting in userdel and usermod calls.
rafal
2007-10-10 13:29:58 -05:00
Andrew Tridgell
c6bae3e87e r8715: - revert the %PRIi64 stuff. Tim, we explicitly check for %llu support
in configure, and replace snprintf if the system doesn't support
  it. Our replacement code does not handle the "%PRIi64" stuff, so
  using it would break us on lots of platforms

- fixed constant array initialisers to work on HPUX.
2007-10-10 13:29:58 -05:00
Tim Potter
1d2b708da6 r8714: Oops - get it right this time! 2007-10-10 13:29:58 -05:00
Tim Potter
7b316f119b r8713: Experiment to try and fix warnings on 64-bit machines without breaking
32-bit ones.  Yes, this weird looking macros are part of C99.
2007-10-10 13:29:57 -05:00
Andrew Tridgell
1d830bcd47 r8712: cleanup old search test code to use talloc 2007-10-10 13:29:57 -05:00
Andrew Tridgell
4d63bd27d4 r8711: add m4 for finding return type for signal handlers
this patch brought to you by the continuing saga of HPUX portability
2007-10-10 13:29:57 -05:00
Andrew Tridgell
eb3b3c8b40 r8710: another attempt at fixing HPUX 2007-10-10 13:29:56 -05:00
Andrew Tridgell
92afff5fe7 r8709: fixed the assumption in RAW-SEARCH that directories are returned in sorted order 2007-10-10 13:29:56 -05:00
Andrew Tridgell
b9b8714297 r8707: this typedef isn't used, and breaks the build on HPUX, so I've removed it 2007-10-10 13:29:56 -05:00
Andrew Bartlett
c3cc14542e r8706: My previous patch oversimplied the previous change to session setup -
we didn't cope with the 'anonymous NTLM under SPNEGO' login.

Andrew Bartlett
2007-10-10 13:29:56 -05:00
Andrew Tridgell
ff156f7aae r8703: cope with null string in interpret_addr() 2007-10-10 13:29:56 -05:00
Andrew Tridgell
3c12b46e32 r8702: fixed ntlm_auth build. Andrew, can you check I got this right? 2007-10-10 13:29:55 -05:00
Andrew Bartlett
1bc5a11907 r8701: Fix up auth_developer for recent changes.
Andrew Bartlett
2007-10-10 13:29:55 -05:00
Andrew Bartlett
ecbd2235a3 r8700: Propmted by tridge's need to do plaintext auth in ejs, rework the
user_info strcture in auth/

This moves it to a pattern much like that found in ntvfs, with
functions to migrate between PAIN, HASH and RESPONSE passwords.

Instead of make_user_info*() functions, we simply fill in the control
block in the callers, per recent dicussions on the lists.  This
removed a lot of data copies as well as error paths, as we can grab
much of it with talloc.

Andrew Bartlett
2007-10-10 13:29:55 -05:00
Andrew Bartlett
f5910ceef5 r8699: removed invalid comment 2007-10-10 13:29:55 -05:00
Andrew Tridgell
c84c516b17 r8698: attempt to cope with lack of strtoull() on HPUX 2007-10-10 13:29:55 -05:00
Andrew Tridgell
80177b29f4 r8696: fixed ejs to more strictly follow the va_list rules. Might fix that
pesky panic on x86_64
2007-10-10 13:29:55 -05:00
Rafal Szczesniak
19f5ea9c9c r8693: Test code for usermod routine.
rafal
2007-10-10 13:29:54 -05:00
Rafal Szczesniak
bee1c9ec2d r8692: Starting parts of code to provide user modify functionality.
It's more like a placeholder now, than a working code. Just don't
want to hang it around my laptop only.

rafal
2007-10-10 13:29:54 -05:00
Jelmer Vernooij
7384b3c4a1 r8690: Code cleanups, improvements necessary for dependencies (not added yet) 2007-10-10 13:29:54 -05:00
Andrew Tridgell
802724cf69 r8685: an alpha in the build farm is dying with a floating point
exception. Catch these in out fault handler, so we get a backtrace
2007-10-10 13:29:54 -05:00
Andrew Tridgell
c2f5884e2f r8684: this should fix the panic on x86_64, and possibly alpha 2007-10-10 13:29:54 -05:00
Andrew Tridgell
e6d9cde482 r8681: if SOCKET_WRAPPER_DIR starts with ./ then strip it internally. This saves us 2 more chars
in the name, which is enough to get IRIX 6.4 working
2007-10-10 13:29:54 -05:00
Andrew Tridgell
e84871dd32 r8680: try harder to find the binary for gdb in the backtrace 2007-10-10 13:29:53 -05:00
Andrew Tridgell
f8a2b032a7 r8679: only call fault setup once (thanks to andrew for pointing this out) 2007-10-10 13:29:53 -05:00
Andrew Tridgell
acf8c8fd49 r8678: setup for gdb backtrace in 'make test' 2007-10-10 13:29:53 -05:00
Andrew Bartlett
2303e24be7 r8677: The first part of the domain name may not be equal to the netbios domain name.
Remove the use of flatname from the main domain object, we no longer
reference it.

Andrew Bartlett
2007-10-10 13:29:53 -05:00
Andrew Tridgell
3a49626ae1 r8676: attribute lists in ldb searches must be NULL terminated
this is what was causing the panic on the s390 box
2007-10-10 13:29:53 -05:00
Andrew Bartlett
da48e77e7c r8674: With the rdn_name module, we don't need this duplication in the samdb
module any more.

Andrew Bartlett
2007-10-10 13:29:53 -05:00
Andrew Tridgell
f9693e313d r8671: use much shorter names for the selftest directory and socket wrapper
code. I'm afraid this is needed by irix 6.4 which silently truncates
names in unix domain sockets in recvfrom() to 16 chars. My apologies
for having to move to such short names :-(
2007-10-10 13:29:52 -05:00
Andrew Bartlett
968bcc4fe8 r8670: Remove GUID code from SAMR, it is handled lower down now. I notice
this code also does string SIDs, but I'm not quite sure where that
fits in.

Andrew Bartlett
2007-10-10 13:29:52 -05:00
Andrew Bartlett
8054abc76e r8669: The objectguid module belongs in Samba's ldb module collection, not in
ldb, as it can't build without the NDR and GUID code.

Also make it properly use the NDR encoding for the GUID (I forgot last
time, and used a string), as well as set the dependencies on the
module correctly.

Andrew Bartlett
2007-10-10 13:29:52 -05:00
Andrew Tridgell
756796ad2e r8668: fixed a segv during upgrade of a very old ldb.
Thanks to volker for finding this one too. Keep them coming!
2007-10-10 13:29:52 -05:00