abartlet/samba.git/.git
16 years agor18584: found one of the fd leaks. The registry backend was using a
Andrew Tridgell [Sat, 16 Sep 2006 16:59:37 +0000 (16:59 +0000)]
r18584: found one of the fd leaks. The registry backend was using a
talloc(NULL, xxx) to allocate the registry context. That had two
consequences

1) it was a massive memory leak, as all winreg operations leaked their
   entire context (including an open ldb database) every time

2) event_context_find() never found the exsting event context, so we
   used a new event context each time, which called epoll_create()
   each time, which caused a fd to be allocated
(This used to be commit 1c0a3de39828b43149d8981fc7f10e7c8b59a392)

16 years agor18583: a nasty hack to allow me to monitor the number of open file
Andrew Tridgell [Sat, 16 Sep 2006 16:27:46 +0000 (16:27 +0000)]
r18583: a nasty hack to allow me to monitor the number of open file
descriptors in smbd on all build farm machines. I suspect we have a fd
leak somewhere, but its hard to track down.

I know part of the problem is the number of (fake) network interfaces
we define in the build farm tests, with each of them listening on a
whole bunch of different protocol ports. That chews up around 48 file
descriptors just to startup. I don't think thats the real problem
though, and I suspect something else is leaking file descriptors on
some hosts.
(This used to be commit 83bf458ec98d9f2a21b3748802f41abcbcc1e0b2)

16 years agor18582: disable synchronous ldb in the build farm. Some hosts have very slow
Andrew Tridgell [Sat, 16 Sep 2006 15:51:41 +0000 (15:51 +0000)]
r18582: disable synchronous ldb in the build farm. Some hosts have very slow
filesystems and the tests are taking more than 1.5 hours. We will
still be testing the sync code paths, as the ldb tests still run
synchronous, but doing all of our other tests synchronously just costs
too much time.
(This used to be commit ce2baf7385e019383ffb292bd30e77c31eecbbef)

16 years agor18581: also check for SEC_STD_DELETE, and split out the check into a separate
Andrew Tridgell [Sat, 16 Sep 2006 15:37:45 +0000 (15:37 +0000)]
r18581: also check for SEC_STD_DELETE, and split out the check into a separate
static function
(This used to be commit 024ca6a91cdf2c0f8999c220b4459a72c45bfd32)

16 years agor18580: map the PVFS_FLAG_READONLY bit in the posix backend onto
Andrew Tridgell [Sat, 16 Sep 2006 15:31:53 +0000 (15:31 +0000)]
r18580: map the PVFS_FLAG_READONLY bit in the posix backend onto
NT_STATUS_ACCESS_DENIED in the access mask checks
(This used to be commit ceffc34f3e9f47a8a44dad52054688f9855eeb37)

16 years agor18579: fixed boolean parameters on big endian hosts which have
Andrew Tridgell [Sat, 16 Sep 2006 14:58:51 +0000 (14:58 +0000)]
r18579: fixed boolean parameters on big endian hosts which have
sizeof(BOOL) != sizeof(int)

this broke with the conversion to a real BOOL type
(This used to be commit 75dab73ac603968ce49c605e07d43051dbfa7398)

16 years agor18578: steve, I think you may have forgotten to commit test_cifsposix.sh ?
Andrew Tridgell [Sat, 16 Sep 2006 05:07:36 +0000 (05:07 +0000)]
r18578: steve, I think you may have forgotten to commit test_cifsposix.sh ?
(This used to be commit fea75d2c9732f91cb79e178ef6c57c3945e4f52e)

16 years agor18577: reduce the number of tests run in 'make quicktest' again, so it
Andrew Tridgell [Sat, 16 Sep 2006 05:06:57 +0000 (05:06 +0000)]
r18577: reduce the number of tests run in 'make quicktest' again, so it
completes in a minute on my laptop (the benchmark I usually use).

Simo, I removed th ldb tests, as ldb is pretty well tested by lots of
the other tests, and its a pretty slow test. We could instead add a
"ldb-testquick.sh" which does only minimal testing. The full tests
will of course be run with 'make test' and in the build farm.
(This used to be commit 3cee6a7863bba61c92f94087df5c6442f02a2da7)

16 years agor18576: unfortunately our current build farm results are a bit too good!
Andrew Tridgell [Sat, 16 Sep 2006 04:54:15 +0000 (04:54 +0000)]
r18576: unfortunately our current build farm results are a bit too good!

when I changed the test output to be easier to read, I also broke it
so that it always gives a zero exit status. Fix now, and unfortunately
I expect to see some more red in the farm as a result :(
(This used to be commit c2fc3471023b613571089cb018638a40091d08ab)

16 years agor18575: - use the right variable to teststatus
Andrew Tridgell [Fri, 15 Sep 2006 23:00:32 +0000 (23:00 +0000)]
r18575: - use the right variable to teststatus

- don't display the "(NN tests failed so far)" message unless a test
  has failed
(This used to be commit da37e963cef065bb1c50543596e352c78e712ddc)

16 years agor18574: re-enable the echo pipe in smbscript. I suspect disabling this was an
Andrew Tridgell [Fri, 15 Sep 2006 22:59:18 +0000 (22:59 +0000)]
r18574: re-enable the echo pipe in smbscript. I suspect disabling this was an
accidental part of commit in the recent echo.idl changes (in r18565)
(This used to be commit 5e143267d7ae18536f1a1be8966dc5e783ea867e)

16 years agor18573: disable the echo.js testing of echo_TestCall() for now.
Andrew Tridgell [Fri, 15 Sep 2006 22:56:45 +0000 (22:56 +0000)]
r18573: disable the echo.js testing of echo_TestCall() for now.

Jelmer, we need to fix pidl to be able to handle the double pointers
in the ejs generated code
(This used to be commit 63760acbb7ef6bc32e82ac843adf4f0155f0cb0a)

16 years agor18571: try to make it a lot more obvious when 'make test' or 'make quicktest'
Andrew Tridgell [Fri, 15 Sep 2006 22:39:30 +0000 (22:39 +0000)]
r18571: try to make it a lot more obvious when 'make test' or 'make quicktest'
has failed. The output was too subtle for people who aren't used to
it.
(This used to be commit a57decb96aee05d3238b64d50a136c8b8001f811)

16 years agor18570: Fix up function names in cifs unix/posix extensions backend. Enable tiny...
Steve French [Fri, 15 Sep 2006 21:39:38 +0000 (21:39 +0000)]
r18570: Fix up function names in cifs unix/posix extensions backend. Enable tiny quick test for torture for them
(This used to be commit 64062d16a2cc215af320eb30827887eb2531e3b0)

16 years agor18569: add really simple testing of the 'simple' ntvfs backend, so we know
Andrew Tridgell [Fri, 15 Sep 2006 21:05:03 +0000 (21:05 +0000)]
r18569: add really simple testing of the 'simple' ntvfs backend, so we know
when it breaks. It isn't much good as a template for developers to use
unless it works :-)
(This used to be commit 262bee07dcb140b413d5ae5a8a754b3e57235323)

16 years agor18568: this warning is not needed now that it is the job of the unixuid ntvfs
Andrew Tridgell [Fri, 15 Sep 2006 20:57:29 +0000 (20:57 +0000)]
r18568: this warning is not needed now that it is the job of the unixuid ntvfs
module to handle euid
(This used to be commit 6784058923ef532da882830296f9bb5f58cddb43)

16 years agor18567: fixed the winreg js code for the new names of the fields in winreg.idl
Andrew Tridgell [Fri, 15 Sep 2006 20:55:43 +0000 (20:55 +0000)]
r18567: fixed the winreg js code for the new names of the fields in winreg.idl

When changing a field name in idl, please remember to check for use of
those functions in any js code as well.
(This used to be commit 7005806aa6842ffc3d5ed98682f2aefc59759580)

16 years agor18566: fixed the winreg pipe and winreg tests
Andrew Tridgell [Fri, 15 Sep 2006 20:36:38 +0000 (20:36 +0000)]
r18566: fixed the winreg pipe and winreg tests

Jerry, there is a big difference on the wire between these two:

  [out] uint32 x;
and
  [out] uint32 *x;

if you change from

  [out] uint32 x;

then you need to change to:

  [out,ref] uint32 *x;

otherwise it changes the format on the wire, which means we are no
longer compatible with MS servers.

but be aware that even if you change to a ref ptr, you also need to
change all the client code to set all the return variables in the out
part of the structure. That's why I don't like the MIDL restriction of
forcing the use of ref pointers for output variables - it makes life
much harder when writing client code, and makes the code much more
error prone (just look at all the extra code needed to make this work
again).

I know we could auto-allocate these variables in the generated client
side NDR code, but if we did that then we would have no way of doing a
_real_ ref out pointer, which we really wanted to set to some already
allocated variable.

So please hold off on changing our idl to use the MIDL convention for
output variables until Jelmer and I have had a good "chat" about this :-)
(This used to be commit 555aed43ba3c08360ca7fa921622b80732a7f657)

16 years agor18565: Fix echo.idl to be Samba3-, MIDL and midlc compatible
Jelmer Vernooij [Fri, 15 Sep 2006 20:07:55 +0000 (20:07 +0000)]
r18565: Fix echo.idl to be Samba3-, MIDL and midlc compatible
(This used to be commit ab0a798c57564901f0adcd8aedc1ef0928e79edd)

16 years agor18564: update for cifs unix/posix extensions stub version to build and beginnings...
Steve French [Fri, 15 Sep 2006 19:24:38 +0000 (19:24 +0000)]
r18564: update for cifs unix/posix extensions stub version to build and beginnings of smbtorture test for it
(This used to be commit 57f56957176ca04f3abb579b557aade71f8d361d)

16 years agor18563: - move more of the header checks into lib/replace/
Andrew Tridgell [Fri, 15 Sep 2006 19:14:36 +0000 (19:14 +0000)]
r18563: - move more of the header checks into lib/replace/

- change the test for net/if.h to do a full compile, not just an
  existance test. net/if.h is completely broken on hpux, and can never
  compile (it uses stuff before it defines it), so by using a
  AC_TRY_COMPILE() test we avoid using net/if.h on hpux, which should
  fix the build
(This used to be commit bde18f3d5ce837f600bae8d63f31d92a579fe1f2)

16 years agor18561: Fix [out] pointers in winreg IDL
Gerald Carter [Fri, 15 Sep 2006 18:34:03 +0000 (18:34 +0000)]
r18561: Fix [out] pointers in winreg IDL
(This used to be commit 694677dafefdd94fa0a9bed93efab70c528dcb26)

16 years agor18559: [string] always applies to the last pointer
Jelmer Vernooij [Fri, 15 Sep 2006 17:34:46 +0000 (17:34 +0000)]
r18559: [string] always applies to the last pointer
(This used to be commit 86b4624226d6e72645221cadb8669b0f1aba0903)

16 years agor18558: Fix ShareCheck which was assuming all paths are "C:\"
Simo Sorce [Fri, 15 Sep 2006 16:27:55 +0000 (16:27 +0000)]
r18558: Fix ShareCheck which was assuming all paths are "C:\"

Also cope with the fact that we define the FSTYPE as NTFS by default.
We never use this anywhere else, so we may just change it, but just
detect the fact and return DISK in share_classic for now.
(This used to be commit 4daf5f7764ce69c14066f7320961c90141f0863a)

16 years agor18555: use C:\ as default not C:
Stefan Metzmacher [Fri, 15 Sep 2006 14:42:24 +0000 (14:42 +0000)]
r18555: use C:\ as default not C:

metze
(This used to be commit d9137edbdb61f2847453357b94050e5afc2fe68b)

16 years agor18554: Fix warnings about [out] arguments.
Jelmer Vernooij [Fri, 15 Sep 2006 14:32:30 +0000 (14:32 +0000)]
r18554: Fix warnings about [out] arguments.
(This used to be commit caac6cf817b09c864554e992f89437f6205b948b)

16 years agor18553: - confdefs.h is always included in configure checks no need to include them...
Stefan Metzmacher [Fri, 15 Sep 2006 14:30:23 +0000 (14:30 +0000)]
r18553: - confdefs.h is always included in configure checks no need to include them explicit
- undefine _XOPEN_SOURCE_EXTENDED for the AIX interface detection test
  #define _XOPEN_SOURCE_EXTENDED 1 brings in sa_len to sockaddr on Tru64
  which means the AIX code compiles...

metze
(This used to be commit 04a0a5144a249e12a9dcb327c5b6055e6a6135db)

16 years agor18550: Return a path not just a disk drive
Simo Sorce [Fri, 15 Sep 2006 12:45:08 +0000 (12:45 +0000)]
r18550: Return a path not just a disk drive
(This used to be commit 60076fa87d6c68295ebedc28cdbbd4556cfd0ae7)

16 years agor18549: move gcc version check to libreplace and reorder the tests a bit
Stefan Metzmacher [Fri, 15 Sep 2006 10:54:18 +0000 (10:54 +0000)]
r18549: move gcc version check to libreplace and reorder the tests a bit
for nicer output

metze
(This used to be commit 888a769af557d050d99df703ce5f651688c837c5)

16 years agor18548: don't use #elif as we don't notice when 2 HAVE_IFACE_ versions are defined
Stefan Metzmacher [Fri, 15 Sep 2006 09:30:32 +0000 (09:30 +0000)]
r18548: don't use #elif as we don't notice when 2 HAVE_IFACE_ versions are defined
try to find the problem on Tru64...where configure says the AIX method finds 1 interface
but later can't compile netif.c. (revision 18486 was the last that detects ifconf with 2 interfaces)

metze
(This used to be commit 8fa9852160680054ddb5316c4ee03a03f8553a37)

16 years agor18545: if yapp isn't availabe touch the target file, as it is commited to svn,
Stefan Metzmacher [Fri, 15 Sep 2006 08:10:49 +0000 (08:10 +0000)]
r18545: if yapp isn't availabe touch the target file, as it is commited to svn,
to prevent rebuilding.

we do

make everything
make everything
make bin/smbtorture
make test

in the buildfarm and rebuilding parts isn't that nice

metze
(This used to be commit 212fa7977d514f81ce4301902a8038869ee02b4e)

16 years agor18544: - use AC_LIBREPLACE_LOCATION_CHECKS in samba4
Stefan Metzmacher [Fri, 15 Sep 2006 07:54:13 +0000 (07:54 +0000)]
r18544: - use AC_LIBREPLACE_LOCATION_CHECKS in samba4
- to get the ordering right we need to specify AC_CANONICAL_BUILD explicit
- add AC_CANONICAL_TARGET

metze
(This used to be commit 1ea52d75849f004752cdbe11a3dddd10b4afe47d)

16 years agor18542: Some late nite work.
Simo Sorce [Fri, 15 Sep 2006 05:18:53 +0000 (05:18 +0000)]
r18542: Some late nite work.
Now we can add and remove a share from the "Computer Management"
console (not yet modify!) usinf share backend = ldb
(This used to be commit ae2f6d4a5a372a37b9783a02bb8e7f16588b21f0)

16 years agor18541: with 100 old style searches we can run out of file descriptors on some
Andrew Tridgell [Fri, 15 Sep 2006 03:29:08 +0000 (03:29 +0000)]
r18541: with 100 old style searches we can run out of file descriptors on some
systems. drop to 50
(This used to be commit ce5d5a187264af2b87e2919bdcc5f478cce02df9)

16 years agor18540: show the additional smbd output from each test, rather than only
Andrew Tridgell [Fri, 15 Sep 2006 03:24:15 +0000 (03:24 +0000)]
r18540: show the additional smbd output from each test, rather than only
showing on test failure and showing all the smbd output each time.
(This used to be commit 1dfac4666423d7e0565795e3eae72da7e25a5b54)

16 years agor18539: 'make distclean' should delete config.cache
Andrew Tridgell [Thu, 14 Sep 2006 20:08:21 +0000 (20:08 +0000)]
r18539: 'make distclean' should delete config.cache
(This used to be commit 5a8becb1be1b41b4fecf9f9f47a60eed5f77c264)

16 years agor18538: we need a pipe here, not a logical OR
Andrew Tridgell [Thu, 14 Sep 2006 20:07:45 +0000 (20:07 +0000)]
r18538: we need a pipe here, not a logical OR

also fix cn name for this record, and ensure tests/tmp is created
(This used to be commit 588c91f06b90467279600fe6a1222273b5d53915)

16 years agor18537: after testing for poptGetContext, if the test fails, don't try to use
Andrew Tridgell [Thu, 14 Sep 2006 19:36:26 +0000 (19:36 +0000)]
r18537: after testing for poptGetContext, if the test fails, don't try to use
the external popt (this affects hpisgr8)
(This used to be commit 9afbd60f91bca84dd645372ed7139e42f712384d)

16 years agor18536: fixed the loading of external binary files from ldif into ldb
Andrew Tridgell [Thu, 14 Sep 2006 18:50:12 +0000 (18:50 +0000)]
r18536: fixed the loading of external binary files from ldif into ldb
(This used to be commit fbe13ed83e2f3508db6d77f4bd65a913ef12ff02)

16 years agor18535: move the AC_CANONICAL_HOST and host specific flag tests into
Andrew Tridgell [Thu, 14 Sep 2006 18:46:10 +0000 (18:46 +0000)]
r18535: move the AC_CANONICAL_HOST and host specific flag tests into
libreplace. This should fix the standalone build of tdb on HPUX, where
we need to blacklist mmap.

Unfortunately this requires that we have a copy of config.guess and
config.sub in each of our project subdirectories. I tried to find a
way to use something like AC_CONFIG_AUX_DIR($libreplacedir) and just
put config.{guess,sub} in the lib/replace/ directory, but I couldn't
figure out how to do that in a way that kept autoconf happy for each
of our separate builds. Any autoconf guru out there see a way to do
this?
(This used to be commit 823cd3ab35456769dcefee17bdaca21f01ba0f63)

16 years agor18528: work around what appears to be a compiler bug in gcc on irix. It
Andrew Tridgell [Thu, 14 Sep 2006 16:08:13 +0000 (16:08 +0000)]
r18528: work around what appears to be a compiler bug in gcc on irix. It
caused the RPC-SECRETS test to crash smbd in an inlined version of
this memcmp() call. This patch should have absolutely no effect at
all, but in fact it prevents the crash.

Disassembling at the point of the crash, it shows that gcc is inlining
the memcmp(). I don't know enough MIPS assembler to actually spot the
bug. In case anyone reading this does know MIPS assembler, here is the
gcc generated code that crashes:

0x105e0218 <gssapi_krb5_verify_header+168>:     lw      $t1,52($sp)
0x105e021c <gssapi_krb5_verify_header+172>:     lw      $t1,0($t1)
0x105e0220 <gssapi_krb5_verify_header+176>:     lhu     $t1,0($t1)
0x105e0224 <gssapi_krb5_verify_header+180>:     lw      $t2,68($sp)
0x105e0228 <gssapi_krb5_verify_header+184>:     lhu     $t2,0($t2)
0x105e022c <gssapi_krb5_verify_header+188>:     subu    $t1,$t1,$t2

it gets a segv at 0x105e0220.

lha, what do you think of this? The change should be innocuous on all
other platforms, apart from making the code harder to read :(
(This used to be commit 95455b57893c99d6d2dc20c4f75042ae4c1cfe85)

16 years agor18524: Pre-allocate out arguments.
Jelmer Vernooij [Thu, 14 Sep 2006 13:37:17 +0000 (13:37 +0000)]
r18524: Pre-allocate out arguments.
(This used to be commit 0ee42669d3a5ec062ec14ecda94342b0df519964)

16 years agor18523: Allow [out] on arrays as well as pointers, use in unixinfo.
Jelmer Vernooij [Thu, 14 Sep 2006 12:25:41 +0000 (12:25 +0000)]
r18523: Allow [out] on arrays as well as pointers, use in unixinfo.
(This used to be commit f67b4d58acd2c3746e5ffeffa804e888ca3c49a5)

16 years agor18522: It's probably not a good idea to use rpc_netdfs_init() as initialization
Jelmer Vernooij [Thu, 14 Sep 2006 12:23:02 +0000 (12:23 +0000)]
r18522: It's probably not a good idea to use rpc_netdfs_init() as initialization
function name for _all_ RPC servers...
(This used to be commit 234272e39d7dc79f5e7e0399aebf5681a92981b2)

16 years agor18521: implement volkers suggestion for avoiding the type punning warnings
Andrew Tridgell [Thu, 14 Sep 2006 12:18:44 +0000 (12:18 +0000)]
r18521: implement volkers suggestion for avoiding the type punning warnings
(This used to be commit 9b9f058edb033c999c4430add4f05459ac43c9e2)

16 years agor18520: Add note about warnings on [out] arrays.
Jelmer Vernooij [Thu, 14 Sep 2006 12:08:06 +0000 (12:08 +0000)]
r18520: Add note about warnings on [out] arrays.
(This used to be commit 7a2afc3e0c0e8c0e9360805e8b4326e257b1e9dc)

16 years agor18518: we replace snprintf() if its not C99, so we should also add the rep_
Andrew Tridgell [Thu, 14 Sep 2006 11:55:49 +0000 (11:55 +0000)]
r18518: we replace snprintf() if its not C99, so we should also add the rep_
prototype in that case
(This used to be commit 64b5ff171d04bd93bd41309edfbe532c01a46e4f)

16 years agor18517: I find it less confusing if tests are run in the order they are
Andrew Tridgell [Thu, 14 Sep 2006 11:47:13 +0000 (11:47 +0000)]
r18517: I find it less confusing if tests are run in the order they are
added.
(This used to be commit 3e1844b2b229d6e84436d6c455bef707d0efcafc)

16 years agor18516: I'm surprised that compilers allowed this at all
Andrew Tridgell [Thu, 14 Sep 2006 11:43:59 +0000 (11:43 +0000)]
r18516: I'm surprised that compilers allowed this at all
(This used to be commit 46aa8a5e8914184575fa6b9ec9edced8dd615eab)

16 years agor18515: Remove calls to "time" for now as some sh implementations don't have it,
Jelmer Vernooij [Thu, 14 Sep 2006 10:35:56 +0000 (10:35 +0000)]
r18515: Remove calls to "time" for now as some sh implementations don't have it,
causing this test to fail.
(This used to be commit fc3f41a4ab38fd32386c5bdb1979cc26892afcf8)

16 years agor18514: fix configure caching of AC_VERIFY_C_PROTOTYPE() macro
Stefan Metzmacher [Thu, 14 Sep 2006 10:34:03 +0000 (10:34 +0000)]
r18514: fix configure caching of AC_VERIFY_C_PROTOTYPE() macro

metze
(This used to be commit 8e49dc69e7d588c85c9e9e7b24b595c61c1da95a)

16 years agor18505: add configure checks for telldir() and seekdir()
Stefan Metzmacher [Thu, 14 Sep 2006 08:26:41 +0000 (08:26 +0000)]
r18505: add configure checks for telldir() and seekdir()

telldir() is different on older OpenBSD versions
seekdir() is different Tru64

tridge: OpenBSD seems to like this replacement:-)
        while MAC OS 10 gets 134 runtime error:-(
lets wait what Tru64 will give

metze
(This used to be commit 1f4e602ff239b7feabb2dd1d6938dedf91bde5cd)

16 years agor18504: Handle mappings for RENAME and KEEP attributes better. We don't need
Andrew Bartlett [Thu, 14 Sep 2006 07:57:49 +0000 (07:57 +0000)]
r18504: Handle mappings for RENAME and KEEP attributes better.  We don't need
to mess with the values in these cases.

Where we do convert the values, try and convert substrings.  This
isn't going to be perfect, but we should try rather than segfault.

This also avoids using the wrong arm of the union for the attribute
name

The change in the entryUUID module is to correct the case of
sAMAccountName, due to the case sensitive ldap.js test.

Andrew Bartlett
(This used to be commit 81d9a692c1e74ec9078bf718003eafdba85b4324)

16 years agor18503: - _GNU_SOURCE is defined by libreplace
Stefan Metzmacher [Thu, 14 Sep 2006 07:30:46 +0000 (07:30 +0000)]
r18503: - _GNU_SOURCE is defined by libreplace
- move AS_HELP_STRING replacement to libreplace

metze
(This used to be commit ace406a36ebbed230bf6c844e4639bac5e6c9882)

16 years agor18502: remove unused lines
Stefan Metzmacher [Thu, 14 Sep 2006 07:22:38 +0000 (07:22 +0000)]
r18502: remove unused lines

metze
(This used to be commit 37f55c08c304feddf2ca5d0d20deeb9a6317e2c3)

16 years agor18501: libreplace needs 'long long', bailout if not present
Stefan Metzmacher [Thu, 14 Sep 2006 06:39:48 +0000 (06:39 +0000)]
r18501: libreplace needs 'long long', bailout if not present
or the size is not 8 bytes or more.

samba4 doesn't need type checks anymore

metze
(This used to be commit d8fdd05482fc6b9bfb48d72db6b467e3e5c05e4d)

16 years agor18500: move make related checks into its own file
Stefan Metzmacher [Thu, 14 Sep 2006 06:22:02 +0000 (06:22 +0000)]
r18500: move make related checks into its own file

metze
(This used to be commit 6a6729fb7219ac7eb160b787827acdf1f7af7cc8)

16 years agor18499: this hopefully lets the code compile on OpenBSD
Stefan Metzmacher [Thu, 14 Sep 2006 05:33:54 +0000 (05:33 +0000)]
r18499: this hopefully lets the code compile on OpenBSD

the prototype was fixed in this revision
http://www.openbsd.org/cgi-bin/cvsweb/src/include/dirent.h.diff?r1=1.15&r2=1.16

so we'll need a configure test to find the prototype of
telldir later

metze
(This used to be commit c4da3b43640368aff98e501b6ca7801821fd2fbd)

16 years agor18498: While passing stack values into torture_tcase_add_test is bad, values
Andrew Bartlett [Thu, 14 Sep 2006 05:13:20 +0000 (05:13 +0000)]
r18498: While passing stack values into torture_tcase_add_test is bad, values
assigned into a static initialiser must be constant.

Wishing sweet dreams on the jet-lagged tridge :-)

Andrew Bartlett
(This used to be commit 174014d956cb1d246bd07218ea289418195a73fb)

16 years agor18497: fixed crash bug in irpc test - the torture_tcase_add_test() call
Andrew Tridgell [Thu, 14 Sep 2006 03:23:52 +0000 (03:23 +0000)]
r18497: fixed crash bug in irpc test - the torture_tcase_add_test() call
is passed a stack pointer from values[], which is not valid when the
torture suite is run.

Jelmer, need to be careful when converting test code not to use stack
values from the torture_XXX() call in the test suite
(This used to be commit 2744a0b385f58aaf0b45f33bec0d0a16c09709d7)

16 years agor18496: yet another dlopen valgrind suppression
Andrew Tridgell [Thu, 14 Sep 2006 03:17:11 +0000 (03:17 +0000)]
r18496: yet another dlopen valgrind suppression
(This used to be commit 041c2c1202b8e5bf2a6a8d90530751560dacd9f4)

16 years agor18495: More work on the LDAP backend (which now passes a lot of our tests!)
Andrew Bartlett [Thu, 14 Sep 2006 03:15:30 +0000 (03:15 +0000)]
r18495: More work on the LDAP backend (which now passes a lot of our tests!)

This adds a list of attributes that are in our wildcard seaches, but
the remote server requires to be explicitly listed.  This also cleans
up the handling of wildcards in ldb_map to be more consistant.

Also fix the partitions module to rebase the search, if on the GC
port, we do a subtree search.  (Otherwise backends can rightly
complain that the search is not in their scope).

Andrew Bartlett
(This used to be commit bc58792b7102f086b19353635d5d5ef9d40a0aae)

16 years agor18494: don't count 'DWARF2 CFI reader' messages as valgrind failures
Andrew Tridgell [Thu, 14 Sep 2006 03:13:02 +0000 (03:13 +0000)]
r18494: don't count 'DWARF2 CFI reader' messages as valgrind failures
(This used to be commit f7eaa03d6f6f28408d8e6a7e33c540e50cbed3a1)

16 years agor18493: another "blind coding" attempt at a getdirentries() based readdir()
Andrew Tridgell [Thu, 14 Sep 2006 02:37:00 +0000 (02:37 +0000)]
r18493: another "blind coding" attempt at a getdirentries() based readdir()
replacement
(This used to be commit 94b73d692bf85604c7be811bad0b8c6a08b18103)

16 years agor18492: an attempt at replacing readdir() with something based on
Andrew Tridgell [Thu, 14 Sep 2006 02:25:57 +0000 (02:25 +0000)]
r18492: an attempt at replacing readdir() with something based on
getdirentries()

This is untested on any of the platforms that matter, so we'll just
have to see if the build farm machines like it
(This used to be commit 343b0871b16c86b68badd4581420f475f94e79c9)

16 years agor18491: fixed a warning
Andrew Tridgell [Wed, 13 Sep 2006 23:13:44 +0000 (23:13 +0000)]
r18491: fixed a warning
(This used to be commit 1b17f0f08f85238717e88966cba311ff4b933a9a)

16 years agor18490: on irix _XOPEN_SOURCE_EXTENDED needs to be defined to 1, not just
Andrew Tridgell [Wed, 13 Sep 2006 23:13:32 +0000 (23:13 +0000)]
r18490: on irix _XOPEN_SOURCE_EXTENDED needs to be defined to 1, not just
defined
(This used to be commit 39d984738b1931e16cb0e6abfdb58108fc97da90)

16 years agor18489: dlopen is so amazingly broken on linux
Andrew Tridgell [Wed, 13 Sep 2006 22:42:23 +0000 (22:42 +0000)]
r18489: dlopen is so amazingly broken on linux

supress the valgrind warnings
(This used to be commit fb69f8cb2c73d123c2099011fcd423f514a93da3)

16 years agor18488: we have to make sure any extensions flags also make it into
Andrew Tridgell [Wed, 13 Sep 2006 22:31:25 +0000 (22:31 +0000)]
r18488: we have to make sure any extensions flags also make it into
confdefs.h, otherwise the real build and the configure tests will not
be significant

this change fixes the build of libreplace on hpux with gcc
(This used to be commit d37fc315325c93414ce0942d28f0f47b42873b95)

16 years agor18486: hopefully this fixes the panics on solaris
Stefan Metzmacher [Wed, 13 Sep 2006 17:38:49 +0000 (17:38 +0000)]
r18486: hopefully this fixes the panics on solaris

metze
(This used to be commit 22cfc3404ccb621dad194fe04c5f2885fb16f79c)

16 years agor18485: include libreplace.m4 in the aclocal.m4 files
Stefan Metzmacher [Wed, 13 Sep 2006 16:51:45 +0000 (16:51 +0000)]
r18485: include libreplace.m4 in the aclocal.m4 files
and use the macros in configure.ac

metze
(This used to be commit 95d33e4d71b4c97af8413bcd136f393aa3e380dd)

16 years agor18482: Fix a couple of small issues in the generated server code.
Jelmer Vernooij [Wed, 13 Sep 2006 16:29:14 +0000 (16:29 +0000)]
r18482: Fix a couple of small issues in the generated server code.
(This used to be commit ddfe3c54439eb76327608e04df9381621ab875b8)

16 years agor18479: start hiding libreplace configure tests behind macros
Stefan Metzmacher [Wed, 13 Sep 2006 16:15:38 +0000 (16:15 +0000)]
r18479: start hiding libreplace configure tests behind macros

AC_LIBREPLACE_CC_CHECKS
AC_LIBREPLACE_BROKEN_CHECKS
and
AC_LIBREPLACE_ALL_CHECKS
which calls the 2 others

I'll add some more, so that samba3/samba4 can later call
them in the wanted order and all standalone builds
use AC_LIBREPLACE_ALL_CHECKS.

metze
(This used to be commit e7a30456c76f4bf9a79cdcff6b15c894bc20c954)

16 years agor18478: Add missing declaration for status.
Jelmer Vernooij [Wed, 13 Sep 2006 15:54:08 +0000 (15:54 +0000)]
r18478: Add missing declaration for status.
(This used to be commit 64909a0d69a938e96f7a04b4288ca851fb8fe18a)

16 years agor18477: Finish server code generator for Samba 3.
Jelmer Vernooij [Wed, 13 Sep 2006 15:45:20 +0000 (15:45 +0000)]
r18477: Finish server code generator for Samba 3.
(This used to be commit 4e2f20042448721ba1df5bbbb77710e155f23953)

16 years agor18475: Start working on server code generator that uses libndr.
Jelmer Vernooij [Wed, 13 Sep 2006 14:13:48 +0000 (14:13 +0000)]
r18475: Start working on server code generator that uses libndr.
(This used to be commit aa1c550d371c5874668baf06be7168c85dc5d48b)

16 years agor18474: - we don't need to explicit check stdint.h and inttypes.h
Stefan Metzmacher [Wed, 13 Sep 2006 14:10:25 +0000 (14:10 +0000)]
r18474: - we don't need to explicit check stdint.h and inttypes.h
  alsready done by autoconf magic
- display the sizes of all standard C types
- check for int64_t and uint64_t

metze
(This used to be commit 371a33a871b67f12af177696bae6aa87d2c1d9f9)

16 years agor18472: - use STDC_HEADERS for stdlib.h and stddef.h as autoconf does
Stefan Metzmacher [Wed, 13 Sep 2006 13:50:51 +0000 (13:50 +0000)]
r18472: - use STDC_HEADERS for stdlib.h and stddef.h as autoconf does
- AC_HEADERS_STDC is not explicit needed

metze
(This used to be commit 8f20d2cfe564164ea049dbb9f52d885e11098c2a)

16 years agor18471: Remove other Samba3 parser generator support as well - it's no
Jelmer Vernooij [Wed, 13 Sep 2006 13:49:17 +0000 (13:49 +0000)]
r18471: Remove other Samba3 parser generator support as well - it's no
longer necessary as we can use libndr now.
(This used to be commit 22142a9f3d5e759742c79a591413e5e8af04b22f)

16 years agor18470: Remove obsolete client generator code for Samba3 (we're now using the
Jelmer Vernooij [Wed, 13 Sep 2006 13:45:36 +0000 (13:45 +0000)]
r18470: Remove obsolete client generator code for Samba3 (we're now using the
new code that uses libndr)
(This used to be commit ea0ef1542f78e3a58d86b5693ec17c145050526b)

16 years agor18468: Add IDL for dfs_EnumEx. Still lacks test(s)
Jelmer Vernooij [Wed, 13 Sep 2006 13:37:52 +0000 (13:37 +0000)]
r18468: Add IDL for dfs_EnumEx. Still lacks test(s)
(This used to be commit 79b6001f919d398e52ceec29f5ebe6f0018e5660)

16 years agor18465: much better fix for a roken specifiv problem
Stefan Metzmacher [Wed, 13 Sep 2006 12:32:27 +0000 (12:32 +0000)]
r18465: much better fix for a roken specifiv problem

metze
(This used to be commit 1ae1e68fe9e1004f1a11847714b2e71715a1ce03)

16 years agor18463: add some more warning flags, tested with gcc 2.95.3 on SuSE 7.3
Stefan Metzmacher [Wed, 13 Sep 2006 11:39:40 +0000 (11:39 +0000)]
r18463: add some more warning flags, tested with gcc 2.95.3 on SuSE 7.3

metze
(This used to be commit 553a90cbe200d2030e4caefe949b48d63abe382d)

16 years agor18462: make the output a bit nicer
Stefan Metzmacher [Wed, 13 Sep 2006 11:08:50 +0000 (11:08 +0000)]
r18462: make the output a bit nicer

metze
(This used to be commit 0b61ea3ed7b833dbb5a7ca8ef2bb29c1b68f5cad)

16 years agor18461: - extend make showflags
Stefan Metzmacher [Wed, 13 Sep 2006 10:58:40 +0000 (10:58 +0000)]
r18461: - extend make showflags
- let make test trigger showflags

metze
(This used to be commit 43c1d775b900b305345afe394383b78e83e4e5cb)

16 years agor18460: split out timegm test and only add timegm.o when needed
Stefan Metzmacher [Wed, 13 Sep 2006 10:51:26 +0000 (10:51 +0000)]
r18460: split out timegm test and only add timegm.o when needed

metze
(This used to be commit f9bff4dbdad8c7acc649d13a5666b58967bf5d92)

16 years agor18459: Set access to test LDAP server as system to anonymous, but also note
Andrew Bartlett [Wed, 13 Sep 2006 10:26:32 +0000 (10:26 +0000)]
r18459: Set access to test LDAP server as system to anonymous, but also note
the correct authz-regexp for future use.

Fail to test if slapd fails and TEST_LDAP=yes
(This used to be commit 1daf4e7bd47df4f127447ab9b5d583bbe7b2a10a)

16 years agor18458: move wins32 stuff to the top dir
Stefan Metzmacher [Wed, 13 Sep 2006 10:13:01 +0000 (10:13 +0000)]
r18458: move wins32 stuff to the top dir

metze
(This used to be commit 521e94f2693eab9053d9e3f1bbc56cdf2e0adbcd)

16 years agor18456: Tru64 doesn't have getdents() so we don't need this ifdef's...
Stefan Metzmacher [Wed, 13 Sep 2006 10:12:03 +0000 (10:12 +0000)]
r18456: Tru64 doesn't have getdents() so we don't need this ifdef's...

metze
(This used to be commit 5af86eb3f07e26aead67ab0dd46576e0aefe8eb4)

16 years agor18455: check if getdents() and getdirentries() are available.
Stefan Metzmacher [Wed, 13 Sep 2006 10:10:20 +0000 (10:10 +0000)]
r18455: check if getdents() and getdirentries() are available.
(only for the nicer configure output)

metze
(This used to be commit d53eb6bd10903d696c5ed9aad3bca3ddf35b2b51)

16 years agor18454: this was just for debugging...
Stefan Metzmacher [Wed, 13 Sep 2006 10:04:46 +0000 (10:04 +0000)]
r18454: this was just for debugging...

metze
(This used to be commit f9f14a94297d72af13b043db99fa892e5e3dcb46)

16 years agor18452: add configure test for a getdirentries() based replacement for broken readdir()
Stefan Metzmacher [Wed, 13 Sep 2006 10:00:45 +0000 (10:00 +0000)]
r18452: add configure test for a getdirentries() based replacement for broken readdir()

Tru64 need this, linux also has getdirentries() but the native readdir() works
but it means we can write the code on linux and when it works it may work on Tru64
too.

tridge: can you try to implement this?

metze
(This used to be commit dd791d255ca61159b38a59b89f954a61fc5e9cf6)

16 years agor18451: move repdir/ into the top dir
Stefan Metzmacher [Wed, 13 Sep 2006 09:56:21 +0000 (09:56 +0000)]
r18451: move repdir/ into the top dir

metze
(This used to be commit a564194817d9b78e353abb6bf0548b2dd9eb130b)

16 years agor18450: - autogenerate the OBJ_FILES for LIBREPLACE
Stefan Metzmacher [Wed, 13 Sep 2006 09:47:41 +0000 (09:47 +0000)]
r18450: - autogenerate the OBJ_FILES for LIBREPLACE
- remove samba specific stuff from libreplace
- and include the readdir replacement stuff in the standalone builds

metze
(This used to be commit 3cac61152ef9a32313d7f7e5d38651f03a31f251)

16 years agor18448: this gives nicer output...
Stefan Metzmacher [Wed, 13 Sep 2006 09:27:33 +0000 (09:27 +0000)]
r18448: this gives nicer output...

metze
(This used to be commit 6bdb329da5771b4be792d8889d3ece8f93a495b0)

16 years agor18447: make sure we bail out if the config.h is not generated
Stefan Metzmacher [Wed, 13 Sep 2006 09:23:05 +0000 (09:23 +0000)]
r18447: make sure we bail out if the config.h is not generated
by samba's configure.

this could happen when you have done a standalone build in
lib/replace/, lib/talloc, ...

metze
(This used to be commit d8e88386748d3952c6a82726c9d36cad870992fd)

16 years agor18445: splitout the dlfcn related tests
Stefan Metzmacher [Wed, 13 Sep 2006 08:14:10 +0000 (08:14 +0000)]
r18445: splitout the dlfcn related tests

metze
(This used to be commit 8662e1481504c50a45416ae09ec19b834164e77c)

16 years agor18444: return the correct value on error
Stefan Metzmacher [Wed, 13 Sep 2006 07:35:42 +0000 (07:35 +0000)]
r18444: return the correct value on error

metze
(This used to be commit d65e94fea910ff435b5cb6fe146f81b01df9abf6)

16 years agor18443: add object files only when needed
Stefan Metzmacher [Wed, 13 Sep 2006 06:28:14 +0000 (06:28 +0000)]
r18443: add object files only when needed

metze
(This used to be commit 5fddb66def8dd29a9f8d13b4b679df26aca6cfab)

16 years agor18442: run the os2_delete test in the testsuite too
Stefan Metzmacher [Wed, 13 Sep 2006 06:11:53 +0000 (06:11 +0000)]
r18442: run the os2_delete test in the testsuite too

that shows that Tru64 is broken...

It doesn't have getdirent() and the the replacement
code doesn't work there.

tridge: do you have any idea how to fix this?

metze
(This used to be commit a5ebfd0ab71f4f4288697f6fd3183a440ce7506b)