Stefan Metzmacher [Mon, 11 Sep 2006 15:42:31 +0000 (15:42 +0000)]
r18384: this is a ref pointer like in samba3 and for the Gid version
metze
Stefan Metzmacher [Mon, 11 Sep 2006 14:08:51 +0000 (14:08 +0000)]
r18383: ok we need _XOPEN_SOURCE 500 and include standards.h on Tru64
to get MAP_FAILED
metze
Stefan Metzmacher [Mon, 11 Sep 2006 13:47:29 +0000 (13:47 +0000)]
r18382: define _XOPEN_SOURCE to hopefully bring in MAP_FAILED in
sys/mman.h on Tru64
metze
Stefan Metzmacher [Mon, 11 Sep 2006 12:47:40 +0000 (12:47 +0000)]
r18381: make sure autoconf doesn't add '-O2' to CFLAGS
metze
Andrew Tridgell [Mon, 11 Sep 2006 10:32:39 +0000 (10:32 +0000)]
r18378: try enabling _OSF_SOURCE to see if it fixes the tru64 build problems -
thanks to volker for the suggestion
Andrew Tridgell [Mon, 11 Sep 2006 09:21:58 +0000 (09:21 +0000)]
r18377: disable native iconv with valgrind
Andrew Tridgell [Mon, 11 Sep 2006 09:19:58 +0000 (09:19 +0000)]
r18376: added iconv:native=false option to turn off native iconv. Needed under
valgrind as native iconv is so full of overflows
Andrew Tridgell [Mon, 11 Sep 2006 09:04:51 +0000 (09:04 +0000)]
r18375: Volker noticed that this is in fact an 8bit number - well spotted!
Andrew Tridgell [Mon, 11 Sep 2006 08:28:25 +0000 (08:28 +0000)]
r18374: don't go past the end of this array
should fix the intermittend failure of the HMACSHA1 test
Andrew Tridgell [Mon, 11 Sep 2006 07:59:02 +0000 (07:59 +0000)]
r18372: don't use the system inotify until in the build farm until we work out
how to work around a problem with it giving unpredictable orderings in
renames (sometimes target is first event, sometimes source)
Andrew Tridgell [Mon, 11 Sep 2006 07:58:08 +0000 (07:58 +0000)]
r18371: made the directory depth notify test independent of timing
Andrew Tridgell [Mon, 11 Sep 2006 07:56:23 +0000 (07:56 +0000)]
r18370: allow system inotify to be disabled
Andrew Bartlett [Mon, 11 Sep 2006 07:51:30 +0000 (07:51 +0000)]
r18368: Don't list GENSEC mechs that only have client implementations in our
supportedSASLMechanism list.
Andrew Bartlett
Andrew Bartlett [Mon, 11 Sep 2006 07:50:49 +0000 (07:50 +0000)]
r18367: When converting to entryUUID, ensure we don't double-convert a
string-format GUID.
Andrew Bartlett
Stefan Metzmacher [Mon, 11 Sep 2006 07:15:49 +0000 (07:15 +0000)]
r18366: check if we can link using the selected flags
metze
Andrew Bartlett [Mon, 11 Sep 2006 06:49:24 +0000 (06:49 +0000)]
r18365: When adding a new structure member, always remember to fragment it.
Andrew Bartlett
Andrew Bartlett [Mon, 11 Sep 2006 06:29:58 +0000 (06:29 +0000)]
r18364: Get us closer to schema compliance. The corrent names for "secret"
and "priorSecret" are "currentValue" and "priorValue".
With this, we pass RPC-LSA against OpenLDAP.
Andrew Bartlett
Andrew Bartlett [Mon, 11 Sep 2006 06:17:12 +0000 (06:17 +0000)]
r18363: Found a rather nasty bug in our fragment handling.
We were adding packet fragments onto the *reply* queue, not the
recieve queue. This worked, as long as we got a whole packet before
we did any reply work, but failed once the backend called a remote
LDAP server (and I presume something invoked the event loop).
Andrew Bartlett
Andrew Bartlett [Mon, 11 Sep 2006 06:15:39 +0000 (06:15 +0000)]
r18362: Make LookupSids map onto LookupSids2, as they both take a policy
handle. Avoids a lookup for the basic domain information for every
request.
Andrew Bartlett
Andrew Bartlett [Mon, 11 Sep 2006 05:11:10 +0000 (05:11 +0000)]
r18361: Invert the way we handle LookupSids2/LookupSids3 and LookupNames3/LookupNames4
The latter calls don't supply a policy handle
The latter calls now acquire a policy handle, then call the earlier
calls. This means we still share the codepaths, but don't need to
fetch policy state when it is already provided.
Andrew Bartlett
Andrew Tridgell [Mon, 11 Sep 2006 04:47:56 +0000 (04:47 +0000)]
r18360: better handling of child process killing in standard mode
Andrew Tridgell [Mon, 11 Sep 2006 04:47:48 +0000 (04:47 +0000)]
r18359: better handling of child process killing in standard mode
Andrew Tridgell [Mon, 11 Sep 2006 04:28:34 +0000 (04:28 +0000)]
r18358: handle errors in the RPC-LSA async test
Andrew Bartlett [Mon, 11 Sep 2006 04:18:16 +0000 (04:18 +0000)]
r18357: Convert more crypto tests from using function results as initialisers.
(Fails on older Unix C compilers)
Andrew Bartlett
Andrew Tridgell [Mon, 11 Sep 2006 02:58:45 +0000 (02:58 +0000)]
r18356: doing 1000 of these requests can take over an hour under
valgrind. Reduce to 50, which is plenty for testing the code
Andrew Tridgell [Mon, 11 Sep 2006 02:09:03 +0000 (02:09 +0000)]
r18355: increase the delays waiting for all events to propogate in the
RAW-NOTIFY test.
Andrew Bartlett [Mon, 11 Sep 2006 01:42:13 +0000 (01:42 +0000)]
r18354: It seems safe to enable the DIGEST-MD5 module now.
Andrew Bartlett
Andrew Tridgell [Mon, 11 Sep 2006 01:38:50 +0000 (01:38 +0000)]
r18353: try to fix the assumption of NULL being handled in printf()
permanently by replacing printf() on systems that don't have a C99
printf lib
Andrew Tridgell [Mon, 11 Sep 2006 01:22:28 +0000 (01:22 +0000)]
r18352: printf("%s") is not safe for NULL strings on all platforms, but
d_printf() is. This fixes a crash on solaris
Andrew Tridgell [Mon, 11 Sep 2006 00:59:31 +0000 (00:59 +0000)]
r18351: functions as initialisers in structures doesn't work on some compilers
Andrew Tridgell [Mon, 11 Sep 2006 00:50:57 +0000 (00:50 +0000)]
r18350: we have to check for lstat() to keep roken happen on hpux
Andrew Tridgell [Mon, 11 Sep 2006 00:43:48 +0000 (00:43 +0000)]
r18349: don't do validation when using valgrind. It's way too slow (can be
hours for a single test)
Andrew Tridgell [Mon, 11 Sep 2006 00:24:21 +0000 (00:24 +0000)]
r18348: fixed a valgrind error in RPC-SRVSVC
Andrew Tridgell [Sun, 10 Sep 2006 22:53:21 +0000 (22:53 +0000)]
r18347: run slapd in the foreground so timelimit can kill it
Andrew Tridgell [Sun, 10 Sep 2006 22:45:45 +0000 (22:45 +0000)]
r18346: make sure we kill off the slapd process
Simo Sorce [Sun, 10 Sep 2006 22:06:49 +0000 (22:06 +0000)]
r18345: test also flags
Simo Sorce [Sun, 10 Sep 2006 21:51:46 +0000 (21:51 +0000)]
r18344: add a test to show hoe NetNameValidate behaves.
fix the misleading idl names
Andrew Tridgell [Sun, 10 Sep 2006 14:19:38 +0000 (14:19 +0000)]
r18343: fixed setlinebuf() prototype, added test for it, and use it in two
places to avoid a #ifdef
Andrew Tridgell [Sun, 10 Sep 2006 13:51:57 +0000 (13:51 +0000)]
r18342: a bit more explanation of these strange values
Andrew Tridgell [Sun, 10 Sep 2006 13:48:42 +0000 (13:48 +0000)]
r18341: ooh this gets subtle - this needed for xfs on linux
Andrew Tridgell [Sun, 10 Sep 2006 13:16:27 +0000 (13:16 +0000)]
r18340: some HPUX boxes don't have ptrdiff_t
Andrew Tridgell [Sun, 10 Sep 2006 13:02:54 +0000 (13:02 +0000)]
r18339: need these checks for roken.h on hpux
Stefan Metzmacher [Sun, 10 Sep 2006 12:51:30 +0000 (12:51 +0000)]
r18338: this isn't needed anymore
metze
Andrew Tridgell [Sun, 10 Sep 2006 12:45:37 +0000 (12:45 +0000)]
r18337: more -Ae tests in check_cc.m4
honor the MMAP_BLACKLIST
Andrew Tridgell [Sun, 10 Sep 2006 12:38:32 +0000 (12:38 +0000)]
r18336: autoconf tries to force on C89 mode on HP-UX, using the -Ae
flag. Unfortunately that flag conflicts with the -AC99 flag, and we
get lots of breakage.
This is a trick to force off the -Ae option
Andrew Tridgell [Sun, 10 Sep 2006 12:23:47 +0000 (12:23 +0000)]
r18335: more portable bool tests
Andrew Tridgell [Sun, 10 Sep 2006 12:12:24 +0000 (12:12 +0000)]
r18334: AIX 5.1 doesn't have any of 'users', 'guest', 'other' or 'unknown'
it does have 'usr'
Stefan Metzmacher [Sun, 10 Sep 2006 11:53:44 +0000 (11:53 +0000)]
r18333: $< is broken in older BSD version:
when $@ is foo/bar.o, $< could be torture/foo/bar.c
if it also exists. So better use $* which is foo/bar
to get foo/bar.c
metze
Andrew Tridgell [Sun, 10 Sep 2006 11:47:21 +0000 (11:47 +0000)]
r18332: added back in our shared mmap test code
Andrew Tridgell [Sun, 10 Sep 2006 11:32:42 +0000 (11:32 +0000)]
r18331: fixed a warning
Andrew Tridgell [Sun, 10 Sep 2006 11:32:31 +0000 (11:32 +0000)]
r18330: don't mix pointer types in RPC replies
Stefan Metzmacher [Sun, 10 Sep 2006 11:23:29 +0000 (11:23 +0000)]
r18329: print out special make variables when failing, to track down the
build failures on older BSD systems with broken make
metze
Andrew Tridgell [Sun, 10 Sep 2006 11:16:26 +0000 (11:16 +0000)]
r18328: when we bail out early, set *offset to 0
Andrew Tridgell [Sun, 10 Sep 2006 11:16:09 +0000 (11:16 +0000)]
r18327: 'struct token' is defined on some hosts
Stefan Metzmacher [Sun, 10 Sep 2006 10:58:15 +0000 (10:58 +0000)]
r18326: these are not used anymore, I found the bug on *BSD
it's the broken BSD make again where we endup with this:
... -c torture/libnet/libnet_rpc.c -o libnet/libnet_rpc.o
metze
Andrew Tridgell [Sun, 10 Sep 2006 10:43:31 +0000 (10:43 +0000)]
r18325: more warnings and one compile error on aix fixed
Andrew Tridgell [Sun, 10 Sep 2006 10:30:33 +0000 (10:30 +0000)]
r18324: fixed a uninitialised variable
Andrew Tridgell [Sun, 10 Sep 2006 10:30:10 +0000 (10:30 +0000)]
r18323: this function returns a pointer, not a bool
Andrew Tridgell [Sun, 10 Sep 2006 10:02:10 +0000 (10:02 +0000)]
r18322: fixed a compilation problem on AIX caused by lex not putting config.h
first. That leads to a conflicting define for lseek() due to
_LARGE_FILES being defined after standards headers are included
Andrew Tridgell [Sun, 10 Sep 2006 10:00:42 +0000 (10:00 +0000)]
r18321: fixed some warnings on AIX
Andrew Tridgell [Sun, 10 Sep 2006 08:08:20 +0000 (08:08 +0000)]
r18320: the raw write size limit test hurts too many hosts. Limit it to 2^33
bytes if --dangerous is not set.
Andrew Tridgell [Sun, 10 Sep 2006 07:24:41 +0000 (07:24 +0000)]
r18319: fixed the directory search resume code on IRIX
The problem was twofold:
1) irix returns 64 bit numbers in telldir(). The protocol uses a 32
bit resume key. We now cope with this properly using the code in
pvfs_list_seek_ofs().
2) irix returns 0xFFFFFFFF from telldir() for the last entry in the
directory. When added to DIR_OFFSET_BASE this became
DIR_OFFSET_DOTDOT which meant an infinite loop!
Simo Sorce [Sun, 10 Sep 2006 03:58:00 +0000 (03:58 +0000)]
r18318: Implement DiskEnum
Fix spelling and consistencies issues in idl
Simo Sorce [Sun, 10 Sep 2006 03:11:03 +0000 (03:11 +0000)]
r18317: Make sure we actually have a valid reply or fail
Simo Sorce [Sun, 10 Sep 2006 02:46:17 +0000 (02:46 +0000)]
r18316: Build the parent too
Simo Sorce [Sun, 10 Sep 2006 00:41:15 +0000 (00:41 +0000)]
r18315: Implement netSrvGetInfo level 102
fix names to be consistent between level 101 and 102 and typo in idl
Simo Sorce [Sat, 9 Sep 2006 23:50:44 +0000 (23:50 +0000)]
r18314: Handle the case where a dir has the sticky bit set and the OS gives back
EPERM when trying to access user xattrs. Just pretend no attributes are set.
Simo.
Andrew Tridgell [Sat, 9 Sep 2006 12:57:45 +0000 (12:57 +0000)]
r18309: FreeBSD 6.1 has a symbol ldap_new_connection() in the system ldap
library. Even though we don't like to that library, it gets loaded via
nss-ldap, which means nss-ldap calls into the samba ldap lib with the
wrong parameters, and crashes.
We really need to use a completely different namespace in libcli/ldap/
Andrew Tridgell [Sat, 9 Sep 2006 12:51:32 +0000 (12:51 +0000)]
r18308: get this right ....
Andrew Tridgell [Sat, 9 Sep 2006 11:46:38 +0000 (11:46 +0000)]
r18307: fixed a warning
Stefan Metzmacher [Sat, 9 Sep 2006 11:07:39 +0000 (11:07 +0000)]
r18306: I hope that works better...
metze
Stefan Metzmacher [Sat, 9 Sep 2006 11:03:12 +0000 (11:03 +0000)]
r18305: get some more debug output to fix the build on sambajail
metze
Andrew Tridgell [Sat, 9 Sep 2006 10:29:13 +0000 (10:29 +0000)]
r18304: fixed misuse of size_t in dopr()
Stefan Metzmacher [Sat, 9 Sep 2006 10:24:42 +0000 (10:24 +0000)]
r18303: we should just use -Ilib/popt as in all other places
metze
Andrew Tridgell [Sat, 9 Sep 2006 10:22:47 +0000 (10:22 +0000)]
r18302: fixed test_cifsdd.sh - the unnecessary sum operations were breaking on
solaris
Andrew Tridgell [Sat, 9 Sep 2006 10:05:58 +0000 (10:05 +0000)]
r18301: I discovered how to load the warnings from a build farm build into
emacs compile mode (hint, paste to a file, and compile as "cat
filename").
This allowed me to fix nearly all the warnings for a IA_64 SuSE build
very quickly.
Andrew Tridgell [Sat, 9 Sep 2006 10:01:17 +0000 (10:01 +0000)]
r18300: fixed a type bug in heimdal - lha, you happy with this upstream? It
showed up on ia_64 systems
Andrew Tridgell [Sat, 9 Sep 2006 10:00:35 +0000 (10:00 +0000)]
r18299: POPT_CREDENTIALS should depend on LIBPOPT
Andrew Tridgell [Sat, 9 Sep 2006 09:20:26 +0000 (09:20 +0000)]
r18298: fixed mmap failure test
Stefan Metzmacher [Sat, 9 Sep 2006 09:16:53 +0000 (09:16 +0000)]
r18297: - remove leading ./
- add CFLAGS to the dynconfig-devel.o rule
metze
Stefan Metzmacher [Sat, 9 Sep 2006 09:08:29 +0000 (09:08 +0000)]
r18296: this isn't needed anymore
metze
Andrew Tridgell [Sat, 9 Sep 2006 09:06:20 +0000 (09:06 +0000)]
r18295: pass write type for packet_size
this fixes a crash on IA_64 systems
Stefan Metzmacher [Sat, 9 Sep 2006 09:05:39 +0000 (09:05 +0000)]
r18294: remove leading ././ from .mk files
metze
Stefan Metzmacher [Sat, 9 Sep 2006 08:46:25 +0000 (08:46 +0000)]
r18293: use the correct name it's not always under $srcdir...
metze
Stefan Metzmacher [Sat, 9 Sep 2006 08:34:48 +0000 (08:34 +0000)]
r18292: import some autoconf 2.60 specific macros,
so that older autoconf versions still work
metze
Stefan Metzmacher [Sat, 9 Sep 2006 08:17:48 +0000 (08:17 +0000)]
r18291: don't add ./ for files in the top dir '.'
metze
Andrew Tridgell [Sat, 9 Sep 2006 07:51:30 +0000 (07:51 +0000)]
r18290: finally worked out why we were failing tests on solaris. It has
kerberos on port 750 in /etc/services! Work around this by using
127.0.0.1:88 in the krb5.conf
Andrew Tridgell [Sat, 9 Sep 2006 06:24:13 +0000 (06:24 +0000)]
r18289: don't check for inline till we've worked out the main compiler flags
Andrew Tridgell [Sat, 9 Sep 2006 06:22:57 +0000 (06:22 +0000)]
r18288: autoconf already has a C99 test builtin!
If this works well, we can remove our own test
Andrew Tridgell [Sat, 9 Sep 2006 03:45:04 +0000 (03:45 +0000)]
r18287: add support for the -qlanglvl=extc99 and -qlanglvl=stdc99 flags,
needed on AIX 5.2 for C99 structures
Andrew Tridgell [Sat, 9 Sep 2006 03:29:40 +0000 (03:29 +0000)]
r18286: ldb now needs install-sh too
Andrew Tridgell [Sat, 9 Sep 2006 03:28:49 +0000 (03:28 +0000)]
r18285: tdb now needs install-sh
Andrew Tridgell [Sat, 9 Sep 2006 03:16:05 +0000 (03:16 +0000)]
r18284: enable _XOPEN_SOURCE_EXTENDED to fix a HP-UX bug with the definition
of sendfile()
Andrew Tridgell [Sat, 9 Sep 2006 02:52:41 +0000 (02:52 +0000)]
r18283: libreplace.m4 needs to be early in configure.ac in other packages too
Andrew Tridgell [Sat, 9 Sep 2006 02:35:38 +0000 (02:35 +0000)]
r18282: dynconfig.o build rule needs cflags.pl too
Andrew Tridgell [Sat, 9 Sep 2006 02:32:12 +0000 (02:32 +0000)]
r18281: a workaround for an infinite dependency loop in the dependency
generation for smbreadline.d. The problem was caused by
extra_flags.txt having ./lib/smbreadline/smbreadline.o but the build
using lib/smbreadline/smbreadline.o, which means cflags.pl didn't
match the target.
Hopefully the workaround can be removed when metze or jelmer wake up
and give me a hint on how to make SMB_SUBSYSTEM() not add the ./
prefix :-)
Andrew Tridgell [Sat, 9 Sep 2006 02:12:09 +0000 (02:12 +0000)]
r18280: more portability tidyups, ensuring we use libreplace everywhere
Andrew Tridgell [Sat, 9 Sep 2006 02:06:01 +0000 (02:06 +0000)]
r18279: libreplace.m4 needs to come first or autoconf complains, as libreplace
checks for global system flags
Andrew Tridgell [Sat, 9 Sep 2006 01:49:38 +0000 (01:49 +0000)]
r18278: move more header checks and _GNU_SOURCE into libreplace
Andrew Tridgell [Sat, 9 Sep 2006 01:21:10 +0000 (01:21 +0000)]
r18277: on some of our slower hosts we need more than 2700 seconds for smbd to
complete all tests