Volker Lendecke [Sun, 12 Aug 2007 11:22:26 +0000 (11:22 +0000)]
r24347: fake_sendfile does not need Inbuf/Outbuf
In the future, we might put the new Linux splice(2) syscall here. This
should also work for reply_write, but getting that in is a bit trickier.
We need to decide very early before fetching the whole buffer that we
have a write call.
Volker Lendecke [Sun, 12 Aug 2007 09:16:54 +0000 (09:16 +0000)]
r24346: Push reply_prep_legacy into send_file_readX
Jelmer Vernooij [Sun, 12 Aug 2007 01:06:02 +0000 (01:06 +0000)]
r24341: Make libreplace provide offsetof.
Jelmer Vernooij [Sun, 12 Aug 2007 00:55:03 +0000 (00:55 +0000)]
r24340: Use standard data type uint32_t rather than tdb-specific u32.
Volker Lendecke [Sat, 11 Aug 2007 16:28:10 +0000 (16:28 +0000)]
r24332: schedule_aio_read_and_X does not need InBuf/OutBuf
Volker Lendecke [Sat, 11 Aug 2007 16:20:27 +0000 (16:20 +0000)]
r24331: Fix an error where gid_t != 32 bit
Volker Lendecke [Sat, 11 Aug 2007 16:18:11 +0000 (16:18 +0000)]
r24330: Fix a 64-bit error
Volker Lendecke [Sat, 11 Aug 2007 16:15:39 +0000 (16:15 +0000)]
r24329: Fix a 64-bit bug
enums are not necessarily represented as 32-bit uints. On assignment
(see line 1029) implicit conversion happens, but not when pointers are
taken.
Volker Lendecke [Sat, 11 Aug 2007 15:02:58 +0000 (15:02 +0000)]
r24328: Restore chain_reply error handling
Get the whole smb header from the second chained function, in particular
the error fields
Volker Lendecke [Sat, 11 Aug 2007 14:37:39 +0000 (14:37 +0000)]
r24327: First round of fixes to chain_reply
The argument to smb_setlen does not contain the nbt header of 4 bytes
The chained function might allocate outbuf itself (as now happens with
reply_read_and_X). This would erroneously overwrite the caller's outbuf.
Give it an outbuf pointer of it's own
Volker Lendecke [Sat, 11 Aug 2007 11:50:53 +0000 (11:50 +0000)]
r24326: Fix the build for Solaris CC
Volker Lendecke [Sat, 11 Aug 2007 11:20:21 +0000 (11:20 +0000)]
r24325: Push down reply_prep_legacy in reply_read_and_X
Volker Lendecke [Sat, 11 Aug 2007 10:53:36 +0000 (10:53 +0000)]
r24324: No reply_prep_legacy() in reply_pipe_read_and_X
Volker Lendecke [Sat, 11 Aug 2007 10:40:35 +0000 (10:40 +0000)]
r24323: Remove the reply_read_and_X wrapper function
Volker Lendecke [Sat, 11 Aug 2007 10:26:40 +0000 (10:26 +0000)]
r24322: Wrap reply_read_and_X in reply_prep_legacy
Volker Lendecke [Sat, 11 Aug 2007 09:53:42 +0000 (09:53 +0000)]
r24321: Attempt to fix the "hape" cc build
Volker Lendecke [Fri, 10 Aug 2007 21:34:55 +0000 (21:34 +0000)]
r24320: Fix the wct checks in reply_trans2 and the reply_trans calls
No idea what I've been smoking here.
Thanks to Stefan for closely looking
Volker Lendecke [Fri, 10 Aug 2007 21:33:58 +0000 (21:33 +0000)]
r24319: Check wct in reply_read_and_X
Volker Lendecke [Fri, 10 Aug 2007 20:53:27 +0000 (20:53 +0000)]
r24318: Print a prominent warning when reply_prep_legacy is called
This is a temporary function anyway
Volker Lendecke [Fri, 10 Aug 2007 20:03:31 +0000 (20:03 +0000)]
r24316: Fix an uninitialized read
Jerry, please check this!
Thanks,
Volker
Volker Lendecke [Fri, 10 Aug 2007 14:01:14 +0000 (14:01 +0000)]
r24314: Attempt to fix bug 4868
Stefan Metzmacher [Fri, 10 Aug 2007 10:31:28 +0000 (10:31 +0000)]
r24312: revision 24091 removed the 'FORCE'...
metze
Stefan Metzmacher [Fri, 10 Aug 2007 10:28:09 +0000 (10:28 +0000)]
r24311: add a reply_force_nterror() macro
metze
Stefan Metzmacher [Fri, 10 Aug 2007 10:15:25 +0000 (10:15 +0000)]
r24310: we should not call END_PROFILE(SMBntcreateX); twice...
metze
Stefan Metzmacher [Fri, 10 Aug 2007 10:11:40 +0000 (10:11 +0000)]
r24309: if no protocol matches, just return choise 0xffff
as we've done before revision 24002
metze
Stefan Metzmacher [Fri, 10 Aug 2007 09:57:30 +0000 (09:57 +0000)]
r24308: this needs a better fix that compiles...
metze
Stefan Metzmacher [Fri, 10 Aug 2007 09:56:13 +0000 (09:56 +0000)]
r24307: fill in choise with value -1 on error
as we've done before revision 24002
metze
Stefan Metzmacher [Fri, 10 Aug 2007 09:53:42 +0000 (09:53 +0000)]
r24306: the check_fsp() function has an explicit user argument
so use it...
metze
Stefan Metzmacher [Fri, 10 Aug 2007 09:44:13 +0000 (09:44 +0000)]
r24304: patch from Bjoern JAcke <bj@SerNet.DE>:
attached patches add EA support for Solaris. If no one disagrees, can
someone check this in please?
metze
Stefan Metzmacher [Fri, 10 Aug 2007 09:05:18 +0000 (09:05 +0000)]
r24302: add a module that overloads the file_id_create VFS function
and alters the device id depending on the configured algorithm.
The algorithm is configured via "fileid:algorithm":
- "fsname" (default) uses a uint64 hash over the mount point
- "fsid" uses the fsid returned from statfs()
This is needed for "clustering = yes" on some clusterfilesystems
metze
Volker Lendecke [Thu, 9 Aug 2007 20:54:58 +0000 (20:54 +0000)]
r24291: Fix Coverity ID 364
We've checked num_rids != 0 above.
Herb Lewis [Thu, 9 Aug 2007 19:03:23 +0000 (19:03 +0000)]
r24290: fix debug statments to match function name
Jeremy Allison [Wed, 8 Aug 2007 23:56:55 +0000 (23:56 +0000)]
r24281: Fix bug found by Herb. The vuid entry in the cli_state structure gets
left as nonzero as returned by the failed cli_session_setup_spnego. When we then try
to authenticate as the user in cli_session_setup this returns an
error "Bad userid" (as seen in wireshark).
"We should only leave cli->vuid != 0 on success. Looks like it's
getting set in the cli_session_setup_blob_receive() call and not
cleared again on error."
Jeremy.
Volker Lendecke [Wed, 8 Aug 2007 20:06:17 +0000 (20:06 +0000)]
r24280: Fix the build of vfs_afsacl.c
Volker Lendecke [Wed, 8 Aug 2007 19:05:30 +0000 (19:05 +0000)]
r24279: Remove reply_prep_legacy from reply_write_and_X
Volker Lendecke [Wed, 8 Aug 2007 18:40:26 +0000 (18:40 +0000)]
r24278: Push down reply_prep_legacy in reply_write_and_X
Remove the need for reply_prep_legacy for reply_pipe_write_and_X
Stefan Metzmacher [Tue, 7 Aug 2007 14:06:27 +0000 (14:06 +0000)]
r24272: try to fix the build on irix...
metze
Volker Lendecke [Tue, 7 Aug 2007 13:43:02 +0000 (13:43 +0000)]
r24271: Push reply_prep_legacy into reply_write_and_X
Volker Lendecke [Tue, 7 Aug 2007 13:14:49 +0000 (13:14 +0000)]
r24270: Fix some END_PROFILE(SMBwriteX)
Volker Lendecke [Tue, 7 Aug 2007 13:12:46 +0000 (13:12 +0000)]
r24269: Check wct in reply_write_and_X
Volker Lendecke [Tue, 7 Aug 2007 13:10:21 +0000 (13:10 +0000)]
r24268: Fix two crashes for spoolss
strlcpy does not like a NULL source. Fix a use-after-free.
Volker Lendecke [Tue, 7 Aug 2007 11:50:03 +0000 (11:50 +0000)]
r24267: Fix the build farm
I had only tested with "net getlocalsid". posix_locking_init() calls this
with a NULL name...
Volker Lendecke [Tue, 7 Aug 2007 11:28:04 +0000 (11:28 +0000)]
r24266: Remove the "open files database hash size" parameter
This has been superseded by the "tdb_hashsize:<tdbname>" parameter
Volker Lendecke [Tue, 7 Aug 2007 11:21:48 +0000 (11:21 +0000)]
r24265: Add the tdb_hashsize:<tdbname> parameter
This makes it possible to set the default hashsize for any tdb. I would
like to remove the "open files database hash size" in favor of this one.
I'll check that removal in with the next commit, please complain/revert
if it's not ok.
Volker
Volker Lendecke [Mon, 6 Aug 2007 21:03:03 +0000 (21:03 +0000)]
r24254: Fix a segfault in r24102
The really pure bugfix would have replaced pparam by *pparam, but we pulled
the fid anyway.
Metze, in line 2421 the FSP_BELONGS_CONN is commented out, is that intended?
The FSCTL_CREATE_OR_GET_OBJECT_ID ioctl did not have it.
Volker
Jeremy Allison [Mon, 6 Aug 2007 18:54:26 +0000 (18:54 +0000)]
r24253: From Jan Martin <Jan.Martin@rwedea.com>.
----------------------------------------------------------
In rare cases, Samba 3.0.25b shows directory contents at the wrong
position in the file tree when displaying a subdirectory of a DFS link.
The problem occurs whenever Windows XP asks for a DFS referral for a
subdirectory of a DFS link with a trailing backslash.
Windows does not do this very often, but we saw it several times per day
on our central DFS server.
smbd/msdfs.c, dfs_path_lookup() does the following with the requested
path:
- in line 390, the local copy 'localpath' is 'unix_convert'ed; the
trailing backslash is removed inside unix_convert
- in lines 417-20, 'dfspath' (another copy of the requested path) is
mangled another way without removing trailing backslashes
That's why the following loop (lines 435-461) that is meant to
synchronously cut off the last path component from both strings until it
comes to a DFS link, does not handle both strings the same. When the
original path ended with a backslash, 'canon_dfspath' has always one
component more than 'localpath', so that *consumedcntp gets too big in
line 446. This value is reported to the client.
----------------------------------------------------------
Bug #4860.
Jeremy.
Günther Deschner [Mon, 6 Aug 2007 14:03:11 +0000 (14:03 +0000)]
r24252: Dump guid of msExchMailboxGuid when returned.
Guenther
Günther Deschner [Mon, 6 Aug 2007 13:48:57 +0000 (13:48 +0000)]
r24251: Neverending fun:
Heimdal doesn't accept all OIDs and gss_import_name() fails with
GSS_S_BAD_NAMETYPE using this one. Use the GSS_KRB5_NT_PRINCIPAL_NAME OID
instead (which works with at least MIT 1.6.1 and Heimdal 1.0.1).
Guenther
Günther Deschner [Mon, 6 Aug 2007 10:39:27 +0000 (10:39 +0000)]
r24250: Merge W_ERROR_HAVE_NO_MEMORY macro from Samba 4.
Guenther
Volker Lendecke [Sun, 5 Aug 2007 11:34:57 +0000 (11:34 +0000)]
r24244: Complete the conversion of reply_trans[s]
Remove reply_prep/post_legacy from send_trans_replies
Volker Lendecke [Sun, 5 Aug 2007 11:19:24 +0000 (11:19 +0000)]
r24243: Remove reply_prep_legacy from api_reply
Volker Lendecke [Sun, 5 Aug 2007 10:57:04 +0000 (10:57 +0000)]
r24242: Remove reply_prep_legacy from api_rpc_trans_reply
Volker Lendecke [Sun, 5 Aug 2007 10:48:35 +0000 (10:48 +0000)]
r24241: Remove reply_prep_legacy from api_WNPHS and api_SNPHS
Volker Lendecke [Sun, 5 Aug 2007 10:37:08 +0000 (10:37 +0000)]
r24240: Push down reply_prep_legacy one level inside api_fd_reply
Volker Lendecke [Sun, 5 Aug 2007 10:28:12 +0000 (10:28 +0000)]
r24239: Push reply_prep_legacy into api_fd_reply
Volker Lendecke [Sun, 5 Aug 2007 10:17:58 +0000 (10:17 +0000)]
r24238: Push down reply_prep_legacy one level
Volker Lendecke [Sun, 5 Aug 2007 10:04:11 +0000 (10:04 +0000)]
r24237: Add send_trans_reply_new
Volker Lendecke [Sun, 5 Aug 2007 10:01:58 +0000 (10:01 +0000)]
r24236: Push reply_prep_legacy() into named_pipe()
Volker Lendecke [Sun, 5 Aug 2007 09:46:17 +0000 (09:46 +0000)]
r24235: Make handle_trans reply to errors itself
Volker Lendecke [Sun, 5 Aug 2007 09:19:21 +0000 (09:19 +0000)]
r24234: Reformatting -- SCNR
Volker Lendecke [Sun, 5 Aug 2007 09:18:14 +0000 (09:18 +0000)]
r24233: Push reply_prep_legacy into handle_trans
Volker Lendecke [Sun, 5 Aug 2007 09:01:15 +0000 (09:01 +0000)]
r24232: Push down reply_prep_legacy inside reply_transs
Volker Lendecke [Sun, 5 Aug 2007 08:47:09 +0000 (08:47 +0000)]
r24231: Push allocating InBuffer/OutBuffer into reply_transs
Volker Lendecke [Sun, 5 Aug 2007 08:31:41 +0000 (08:31 +0000)]
r24230: Push down reply_prep_legacy in reply_trans
The last checkin was for reply_trans, not reply_transs
Volker Lendecke [Sun, 5 Aug 2007 08:04:11 +0000 (08:04 +0000)]
r24229: Push allocating InBuffer/OutBuffer into reply_transs
Volker Lendecke [Sat, 4 Aug 2007 20:44:33 +0000 (20:44 +0000)]
r24225: Convert reply_flush to the new API
Volker Lendecke [Sat, 4 Aug 2007 20:28:46 +0000 (20:28 +0000)]
r24224: Check wct in reply_trans
Volker Lendecke [Sat, 4 Aug 2007 20:08:35 +0000 (20:08 +0000)]
r24223: Convert reply_echo to the new API
Volker Lendecke [Sat, 4 Aug 2007 10:32:54 +0000 (10:32 +0000)]
r24167: Fix Coverity ID 365
We checked argc>=3 some lines above
Volker Lendecke [Sat, 4 Aug 2007 10:25:27 +0000 (10:25 +0000)]
r24166: Fix Coverity ID 391
Volker Lendecke [Sat, 4 Aug 2007 10:18:33 +0000 (10:18 +0000)]
r24165: Fix Coverity ID 369.
This was not really a bug I think, but this change cleans up the code a
bit.
Jeremy Allison [Fri, 3 Aug 2007 16:51:43 +0000 (16:51 +0000)]
r24164: Fix for write cache corruption bug reported by Jean-Francois Panisset <panisset@A52.com>.
Awaiting confirmation from reporter.
Jeremy.
Volker Lendecke [Fri, 3 Aug 2007 16:06:44 +0000 (16:06 +0000)]
r24163: Remove the send_trans2_replies_new wrapper
This changes send_trans2_replies to not depend on large buffers anymore
and finishes the trans2 conversion.
Volker Lendecke [Fri, 3 Aug 2007 15:45:41 +0000 (15:45 +0000)]
r24162: send_trans2_replies is now static in trans2.c
Volker Lendecke [Fri, 3 Aug 2007 15:27:28 +0000 (15:27 +0000)]
r24161: handle_trans2 takes care of errors itself, no need to do it in reply_transs2
Volker Lendecke [Fri, 3 Aug 2007 14:33:38 +0000 (14:33 +0000)]
r24160: process_trans2 in smbd/blocking.c used send_trans2_replies.
Fake a struct smb_request here.
Volker
Volker Lendecke [Fri, 3 Aug 2007 14:02:07 +0000 (14:02 +0000)]
r24159: Convert reply_transs2 to the new API
Gerald Carter [Fri, 3 Aug 2007 13:53:25 +0000 (13:53 +0000)]
r24158: SE_GROUP_RESOURCE in the other_sids list apparently means a
domain local group.
Fix a typo in the PAC debugging routine
Volker Lendecke [Fri, 3 Aug 2007 13:34:02 +0000 (13:34 +0000)]
r24156: Convert reply_trans2 to the new API
Volker Lendecke [Fri, 3 Aug 2007 13:13:57 +0000 (13:13 +0000)]
r24155: Check wct in reply_trans2
Volker Lendecke [Fri, 3 Aug 2007 12:44:06 +0000 (12:44 +0000)]
r24154: handle_trans2 does not reference inbuf/outbuf anymore
Volker Lendecke [Fri, 3 Aug 2007 12:15:58 +0000 (12:15 +0000)]
r24153: Convert call_trans2setfsinfo to the new API.
This *might* break the transport encryption stuff. I need to check that.
Volker Lendecke [Fri, 3 Aug 2007 11:57:39 +0000 (11:57 +0000)]
r24152: Convert call_trans2ioctl to the new API
Volker Lendecke [Fri, 3 Aug 2007 10:51:59 +0000 (10:51 +0000)]
r24151: Convert call_trans2getdfsreferral to the new API
Volker Lendecke [Fri, 3 Aug 2007 10:32:41 +0000 (10:32 +0000)]
r24150: Convert call_trans2qfsinfo to the new API
Volker Lendecke [Fri, 3 Aug 2007 10:07:18 +0000 (10:07 +0000)]
r24149: Convert call_trans2setfilepathinfo to the new API
Volker Lendecke [Fri, 3 Aug 2007 09:45:42 +0000 (09:45 +0000)]
r24148: Remove some inbuf refs from subcommands of call_trans2setfilepathinfo
Volker Lendecke [Fri, 3 Aug 2007 09:09:00 +0000 (09:09 +0000)]
r24147: Remove some inbuf refs from call_trans2setfilepathinfo
Volker Lendecke [Fri, 3 Aug 2007 07:00:19 +0000 (07:00 +0000)]
r24145: Fix a comment
Volker Lendecke [Thu, 2 Aug 2007 20:56:40 +0000 (20:56 +0000)]
r24144: Fix the build
In a function returning void the "return;" needs to stand alone on some
compilers
Volker Lendecke [Thu, 2 Aug 2007 20:19:27 +0000 (20:19 +0000)]
r24143: Convert call_trans2qfilepathinfo to the new API
Volker Lendecke [Thu, 2 Aug 2007 20:06:15 +0000 (20:06 +0000)]
r24142: Remove some direct inbuf references from call_trans2qfilepathinfo
Volker Lendecke [Thu, 2 Aug 2007 19:50:56 +0000 (19:50 +0000)]
r24141: Add check_fsp as a replacement for CHECK_FSP
Volker Lendecke [Thu, 2 Aug 2007 19:48:44 +0000 (19:48 +0000)]
r24140: Convert call_trans2findnotifyfirst/next to the new API
Volker Lendecke [Thu, 2 Aug 2007 19:35:27 +0000 (19:35 +0000)]
r24139: Convert call_trans2mkdir to the new API
Volker Lendecke [Thu, 2 Aug 2007 19:22:14 +0000 (19:22 +0000)]
r24138: Convert call_trans2findnext to the new API
Volker Lendecke [Thu, 2 Aug 2007 19:08:21 +0000 (19:08 +0000)]
r24137: Convert call_trans2findfirst to the new API
Volker Lendecke [Thu, 2 Aug 2007 18:50:17 +0000 (18:50 +0000)]
r24136: get_lanman2_dir_entry() does not need access to inbuf/outbuf
It did not use inbuf anyway. Outbuf was used for alignment and to the
flags2. The alignment works ok if we align relative to base_data because this
and outbuf are both malloc'ed, and we put send_trans2_replies aligns out_data
properly relative to the start of the outbuf.
I don't know if the explicit align_string() calls in get_lanman2_dir_entry are
really necessary. I think that srvstr_push also does the same.
Please also note that this changes dst_len in srvstr_push from -1 to the real
length remaining.
Volker
Volker Lendecke [Thu, 2 Aug 2007 18:28:41 +0000 (18:28 +0000)]
r24135: Convert call_trans2open to the new API
This itself won't help much, because send_trans2_replies_new still allocates
the big buffers, but stay tuned :-)
Also add/update my copyright on stuff I recently touched.
Volker
Volker Lendecke [Thu, 2 Aug 2007 18:13:57 +0000 (18:13 +0000)]
r24134: talloc smb_request for handle_trans2
When starting to convert the individual trans2 subcalls, I need the new
API conventions to be present there. This means that those calls fill in
req->outbuf when there's something to ship