| blob | c8c91c42d1ccc9efa8547c5e11be8e06ae4f3f27 |
3 THIS IS INCOMPLETE! I'M ONLY COMMITING IT IN ORDER TO SOLICIT COMMENTS
4 FROM A FEW PEOPLE. DON'T TAKE THIS AS THE FINAL VERSION YET.
7 Samba4 Programming Guide
8 ========================
10 .. contents::
12 The internals of Samba4 are quite different from previous versions of
13 Samba, so even if you are an experienced Samba developer please take
14 the time to read through this document.
16 This document will explain both the broad structure of Samba4, and
17 some of the common coding elements such as memory management and
18 dealing with macros.
21 Coding Style
22 ------------
24 In past versions of Samba we have basically let each programmer choose
25 their own programming style. Unfortunately the result has often been
26 that code that other members of the team find difficult to read. For
27 Samba version 4 I would like to standardise on a common coding style
28 to make the whole tree more readable. For those of you who are
29 horrified at the idea of having to learn a new style, I can assure you
30 that it isn't as painful as you might think. I was forced to adopt a
31 new style when I started working on the Linux kernel, and after some
32 initial pain found it quite easy.
34 That said, I don't want to invent a new style, instead I would like to
35 adopt the style used by the Linux kernel. It is a widely used style
36 with plenty of support tools available. See Documentation/CodingStyle
37 in the Linux source tree. This is the style that I have used to write
38 all of the core infrastructure for Samba4 and I think that we should
39 continue with that style.
41 I also think that we should most definately *not* adopt an automatic
42 reformatting system in cvs (or whatever other source code system we
43 end up using in the future). Such automatic formatters are, in my
44 experience, incredibly error prone and don't understand the necessary
45 exceptions. I don't mind if people use automated tools to reformat
46 their own code before they commit it, but please do not run such
47 automated tools on large slabs of existing code without being willing
48 to spend a *lot* of time hand checking the results.
50 Finally, I think that for code that is parsing or formatting protocol
51 packets the code layout should strongly reflect the packet
52 format. That means ordring the code so that it parses in the same
53 order as the packet is stored on the wire (where possible) and using
54 white space to align packet offsets so that a reader can immediately
55 map any line of the code to the corresponding place in the packet.
58 Static and Global Data
59 ----------------------
61 The basic rule is "avoid static and global data like the plague". What
62 do I mean by static data? The way to tell if you have static data in a
63 file is to use the "size" utility in Linux. For example if we run::
65 size libcli/raw/*.o
67 in Samba4 then you get the following::
69 text data bss dec hex filename
70 2015 0 0 2015 7df libcli/raw/clikrb5.o
71 202 0 0 202 ca libcli/raw/clioplock.o
72 35 0 0 35 23 libcli/raw/clirewrite.o
73 3891 0 0 3891 f33 libcli/raw/clisession.o
74 869 0 0 869 365 libcli/raw/clisocket.o
75 4962 0 0 4962 1362 libcli/raw/clispnego.o
76 1223 0 0 1223 4c7 libcli/raw/clitransport.o
77 2294 0 0 2294 8f6 libcli/raw/clitree.o
78 1081 0 0 1081 439 libcli/raw/raweas.o
79 6765 0 0 6765 1a6d libcli/raw/rawfile.o
80 6824 0 0 6824 1aa8 libcli/raw/rawfileinfo.o
81 2944 0 0 2944 b80 libcli/raw/rawfsinfo.o
82 541 0 0 541 21d libcli/raw/rawioctl.o
83 1728 0 0 1728 6c0 libcli/raw/rawnegotiate.o
84 723 0 0 723 2d3 libcli/raw/rawnotify.o
85 3779 0 0 3779 ec3 libcli/raw/rawreadwrite.o
86 6597 0 0 6597 19c5 libcli/raw/rawrequest.o
87 5580 0 0 5580 15cc libcli/raw/rawsearch.o
88 3034 0 0 3034 bda libcli/raw/rawsetfileinfo.o
89 5187 0 0 5187 1443 libcli/raw/rawtrans.o
90 2033 0 0 2033 7f1 libcli/raw/smb_signing.o
92 notice that the "data" and "bss" columns are all zero? That is
93 good. If there are any non-zero values in data or bss then that
94 indicates static data and is bad (as a rule of thumb).
96 Lets compare that result to the equivalent in Samba3::
98 text data bss dec hex filename
99 3978 0 0 3978 f8a libsmb/asn1.o
100 18963 0 288 19251 4b33 libsmb/cliconnect.o
101 2815 0 1024 3839 eff libsmb/clidgram.o
102 4038 0 0 4038 fc6 libsmb/clientgen.o
103 3337 664 256 4257 10a1 libsmb/clierror.o
104 10043 0 0 10043 273b libsmb/clifile.o
105 332 0 0 332 14c libsmb/clifsinfo.o
106 166 0 0 166 a6 libsmb/clikrb5.o
107 5212 0 0 5212 145c libsmb/clilist.o
108 1367 0 0 1367 557 libsmb/climessage.o
109 259 0 0 259 103 libsmb/clioplock.o
110 1584 0 0 1584 630 libsmb/cliprint.o
111 7565 0 256 7821 1e8d libsmb/cliquota.o
112 7694 0 0 7694 1e0e libsmb/clirap.o
113 27440 0 0 27440 6b30 libsmb/clirap2.o
114 2905 0 0 2905 b59 libsmb/clireadwrite.o
115 1698 0 0 1698 6a2 libsmb/clisecdesc.o
116 5517 0 0 5517 158d libsmb/clispnego.o
117 485 0 0 485 1e5 libsmb/clistr.o
118 8449 0 0 8449 2101 libsmb/clitrans.o
119 2053 0 4 2057 809 libsmb/conncache.o
120 3041 0 256 3297 ce1 libsmb/credentials.o
121 1261 0 1024 2285 8ed libsmb/doserr.o
122 14560 0 0 14560 38e0 libsmb/errormap.o
123 3645 0 0 3645 e3d libsmb/namecache.o
124 16815 0 8 16823 41b7 libsmb/namequery.o
125 1626 0 0 1626 65a libsmb/namequery_dc.o
126 14301 0 1076 15377 3c11 libsmb/nmblib.o
127 24516 0 2048 26564 67c4 libsmb/nterr.o
128 8661 0 8 8669 21dd libsmb/ntlmssp.o
129 3188 0 0 3188 c74 libsmb/ntlmssp_parse.o
130 4945 0 0 4945 1351 libsmb/ntlmssp_sign.o
131 1303 0 0 1303 517 libsmb/passchange.o
132 1221 0 0 1221 4c5 libsmb/pwd_cache.o
133 2475 0 4 2479 9af libsmb/samlogon_cache.o
134 10768 32 0 10800 2a30 libsmb/smb_signing.o
135 4524 0 16 4540 11bc libsmb/smbdes.o
136 5708 0 0 5708 164c libsmb/smbencrypt.o
137 7049 0 3072 10121 2789 libsmb/smberr.o
138 2995 0 0 2995 bb3 libsmb/spnego.o
139 3186 0 0 3186 c72 libsmb/trustdom_cache.o
140 1742 0 0 1742 6ce libsmb/trusts_util.o
141 918 0 28 946 3b2 libsmb/unexpected.o
143 notice all of the non-zero data and bss elements? Every bit of that
144 data is a bug waiting to happen.
146 Static data is evil as it has the following consequences:
147 - it makes code much less likely to be thread-safe
148 - it makes code much less likely to be recursion-safe
149 - it leads to subtle side effects when the same code is called from multiple places
150 - doesn't play well with shared libraries or plugins
152 Static data is particularly evil in library code (such as our internal
153 smb and rpc libraries). If you can get rid of all static data in
154 libraries then you can make some fairly strong guarantees about the
155 behaviour of functions in that library, which really helps.
157 Of course, it is possible to write code that uses static data and is
158 safe, it's just much harder to do that than just avoid static data in
159 the first place. We have been tripped up countless times by subtle
160 bugs in Samba due to the use of static data, so I think it is time to
161 start avoiding it in new code. Much of the core infrastructure of
162 Samba4 was specifically written to avoid static data, so I'm going to
163 be really annoyed if everyone starts adding lots of static data back
164 in.
166 So, how do we avoid static data? The basic method is to use context
167 pointers. When reading the Samba4 code you will notice that just about
168 every function takes a pointer to a context structure as its first
169 argument. Any data that the function needs that isn't an explicit
170 argument to the function can be found by traversing that context.
172 Note that this includes all of the little caches that we have lying
173 all over the code in Samba3. I'm referring to the ones that generally
174 have a "static int initialised" and then some static string or integer
175 that remembers the last return value of the function. Get rid of them!
176 If you are *REALLY* absolutely completely certain that your personal
177 favourite mini-cache is needed then you should do it properly by
178 putting it into the appropriate context rather than doing it the lazy
179 way by putting it inside the target function. I would suggest however
180 that the vast majority of those little caches are useless - don't
181 stick it in unless you have really firm benchmarking results that show
182 that it is needed and helps by a significant amount.
184 Note that Samba4 is not yet completely clean of static data like
185 this. I've gotten the smbd/ directory down to 24 bytes of static data,
186 and libcli/raw/ down to zero. I've also gotten the ntvfs layer and all
187 backends down to just 8 bytes in ntvfs_base.c. The rest still needs
188 some more work.
190 Also note that truly constant data is OK, and will not in fact show up
191 in the data and bss columns in "size" anyway (it will be included in
192 "text"). So you can have constant tables of protocol data.
195 How to use talloc
196 -----------------
198 Please see the separate document, lib/talloc/talloc_guide.txt
199 You _must_ read this if you want to program in Samba4.
202 Interface Structures
203 --------------------
205 One of the biggest changes in Samba4 is the universal use of interface
206 structures. Go take a look through libcli/raw/interfaces.h now to get
207 an idea of what I am talking about.
209 In Samba3 many of the core wire structures in the SMB protocol were
210 never explicitly defined in Samba. Instead, our parse and generation
211 functions just worked directly with wire buffers. The biggest problem
212 with this is that is tied our parse code with our "business logic"
213 much too closely, which meant the code got extremely confusing to
214 read.
216 In Samba4 we have explicitly defined interface structures for
217 everything in the protocol. When we receive a buffer we always parse
218 it completely into one of these structures, then we pass a pointer to
219 that structure to a backend handler. What we must *not* do is make any
220 decisions about the data inside the parse functions. That is critical
221 as different backends will need different portions of the data. This
222 leads to a golden rule for Samba4:
224 "don't design interfaces that lose information"
226 In Samba3 our backends often received "condensed" versions of the
227 information sent from clients, but this inevitably meant that some
228 backends could not get at the data they needed to do what they wanted,
229 so from now on we should expose the backends to all of the available
230 information and let them choose which bits they want.
232 Ok, so now some of you will be thinking "this sounds just like our
233 msrpc code from Samba3", and while to some extent this is true there
234 are extremely important differences in the approach that are worth
235 pointing out.
237 In the Samba3 msrpc code we used explicit parse structures for all
238 msrpc functions. The problem is that we didn't just put all of the
239 real variables in these structures, we also put in all the artifacts
240 as well. A good example is the security descriptor strucrure that
241 looks like this in Samba3::
243 typedef struct security_descriptor_info
244 {
245 uint16 revision;
246 uint16 type;
248 uint32 off_owner_sid;
249 uint32 off_grp_sid;
250 uint32 off_sacl;
251 uint32 off_dacl;
253 SEC_ACL *dacl;
254 SEC_ACL *sacl;
255 DOM_SID *owner_sid;
256 DOM_SID *grp_sid;
257 } SEC_DESC;
259 The problem with this structure is all the off_* variables. Those are
260 not part of the interface, and do not appear in any real descriptions
261 of Microsoft security descriptors. They are parsing artifacts
262 generated by the IDL compiler that Microsoft use. That doesn't mean
263 they aren't needed on the wire - indeed they are as they tell the
264 parser where to find the following four variables, but they should
265 *NOT* be in the interface structure.
267 In Samba3 there were unwritten rules about which variables in a
268 structure a high level caller has to fill in and which ones are filled
269 in by the marshalling code. In Samba4 those rules are gone, because
270 the redundent artifact variables are gone. The high level caller just
271 sets up the real variables and the marshalling code worries about
272 generating the right offsets.
274 The same rule applies to strings. In many places in the SMB and MSRPC
275 protocols complex strings are used on the wire, with complex rules
276 about padding, format, alighment, termination etc. None of that
277 information is useful to a high level calling routine or to a backend - its
278 all just so much wire fluff. So, in Samba4 these strings are
279 just "char \*" and are always in our internal multi-byte format (which
280 is usually UTF8). It is up to the parse functions to worry about
281 translating the format and getting the padding right.
283 The one exception to this is the use of the WIRE_STRING type, but that
284 has a very good justification in terms of regression testing. Go and
285 read the comment in smb_interfaces.h about that now.
287 So, here is another rule to code by. When writing an interface
288 structure think carefully about what variables in the structure can be
289 left out as they are redundent. If some length is effectively defined
290 twice on the wire then only put it once in the packet. If a length can
291 be inferred from a null termination then do that and leave the length
292 out of the structure completely. Don't put redundent stuff in
293 structures!
296 Async Design
297 ------------
299 Samba4 has an asynchronous design. That affects *lots* of the code,
300 and the implications of the asynchronous design needs to be considered
301 just about everywhere.
303 The first aspect of the async design to look at is the SMB client
304 library. Lets take a look at the following three functions in
305 libcli/raw/rawfile.c::
307 struct cli_request *smb_raw_seek_send(struct cli_tree *tree, struct smb_seek *parms);
308 NTSTATUS smb_raw_seek_recv(struct cli_request *req, struct smb_seek *parms);
309 NTSTATUS smb_raw_seek(struct cli_tree *tree, struct smb_seek *parms);
311 Go and read them now then come back.
313 Ok, first notice there there are 3 separate functions, whereas the
314 equivalent code in Samba3 had just one. Also note that the 3rd
315 function is extremely simple - its just a wrapper around calling the
316 first two in order.
318 The three separate functions are needed because we need to be able to
319 generate SMB calls asynchronously. The first call, which for smb calls
320 is always called smb_raw_XXXX_send(), constructs and sends a SMB
321 request and returns a "struct cli_request" which acts as a handle for
322 the request. The caller is then free to do lots of other calls if it
323 wants to, then when it is ready it can call the smb_raw_XXX_recv()
324 function to receive the reply.
326 If all you want is a synchronous call then call the 3rd interface, the
327 one called smb_raw_XXXX(). That just calls the first two in order, and
328 blocks waiting for the reply.
330 But what if you want to be called when the reply comes in? Yes, thats
331 possible. You can do things like this::
333 struct cli_request *req;
335 req = smb_raw_XXX_send(tree, params);
337 req->async.fn = my_callback;
338 req->async.private = my_private_data;
340 then in your callback function you can call the smb_raw_XXXX_recv()
341 function to receive the reply. Your callback will receive the "req"
342 pointer, which you can use to retrieve your private data from
343 req->async.private.
345 Then all you need to do is ensure that the main loop in the client
346 library gets called. You can either do that by polling the connection
347 using cli_transport_pending() and cli_request_receive_next() or you
348 can use transport->idle.func to setup an idle function handler to call
349 back to your main code. Either way, you can build a fully async
350 application.
352 In order to support all of this we have to make sure that when we
353 write a piece of library code (SMB, MSRPC etc) that we build the
354 separate _send() and _recv() functions. It really is worth the effort.
356 Now about async in smbd, a much more complex topic.
358 The SMB protocol is inherently async. Some functions (such as change
359 notify) often don't return for hours, while hundreds of other
360 functions pass through the socket. Take a look at the RAW-MUX test in
361 the Samba4 smbtorture to see some really extreme examples of the sort
362 of async operations that Windows supports. I particularly like the
363 open/open/close sequence where the 2nd open (which conflicts with the
364 first) succeeds because the subsequent close is answered out of order.
366 In Samba3 we handled this stuff very badly. We had awful "pending
367 request" queues that allocated full 128k packet buffers, and even with
368 all that crap we got the semantics wrong. In Samba4 I intend to make
369 sure we get this stuff right.
371 So, how do we do this? We now have an async interface between smbd and
372 the NTVFS backends. Whenever smbd calls into a backend the backend has
373 an option of answer the request in a synchronous fashion if it wants
374 to just like in Samba3, but it also has the option of answering the
375 request asynchronously. The only backend that currently does this is
376 the CIFS backend, but I hope the other backends will soon do this to.
378 To make this work you need to do things like this in the backend::
380 req->control_flags |= REQ_CONTROL_ASYNC;
382 that tells smbd that the backend has elected to reply later rather
383 than replying immediately. The backend must *only* do this if
384 req->async.send_fn is not NULL. If send_fn is NULL then it means that
385 the smbd front end cannot handle this function being replied to in an
386 async fashion.
388 If the backend does this then it is up to the backend to call
389 req->async.send_fn() when it is ready to reply. It the meantime smbd
390 puts the call on hold and goes back to answering other requests on the
391 socket.
393 Inside smbd you will find that there is code to support this. The most
394 obvious change is that smbd splits each SMB reply function into two
395 parts - just like the client library has a _send() and _recv()
396 function, so smbd has a _send() function and the parse function for
397 each SMB.
399 As an example go and have a look at reply_getatr_send() and
400 reply_getatr() in smb_server/smb/reply.c. Read them? Good.
402 Notice that reply_getatr() sets up the req->async structure to contain
403 the send function. Thats how the backend gets to do an async reply, it
404 calls this function when it is ready. Also notice that reply_getatr()
405 only does the parsing of the request, and does not do the reply
406 generation. That is done by the _send() function.
409 NTVFS
410 -----
412 One of the most noticeable changes in Samba4 is the introduction of
413 the NTVFS layer. This provided the initial motivation for the design
414 of Samba4 and in many ways lies at the heart of the design.
416 In Samba3 the main file serving process (smbd) combined the handling
417 of the SMB protocol with the mapping to POSIX semantics in the same
418 code. If you look in smbd/reply.c in Samba3 you see numerous places
419 where POSIX assumptions are mixed tightly with SMB parsing code. We
420 did have a VFS layer in Samba3, but it was a POSIX-like VFS layer, so
421 no matter how you wrote a plugin you could not bypass the POSIX
422 mapping decisions that had already been made before the VFS layer was
423 called.
425 In Samba4 things are quite different. All SMB parsing is performed in
426 the smbd front end, then fully parsed requests are passed to the NTVFS
427 backend. That backend makes any semantic mapping decisions and fills
428 in the 'out' portion of the request. The front end is then responsible
429 for putting those results into wire format and sending them to the
430 client.
432 Lets have a look at one of those request structures. Go and read the
433 definition of "union smb_write" and "enum write_level" in
434 libcli/raw/interfaces.h. (no, don't just skip reading it, really go
435 and read it. Yes, that means you!).
437 Notice the union? That's how Samba4 allows a single NTVFS backend
438 interface to handle the several different ways of doing a write
439 operation in the SMB protocol. Now lets look at one section of that
440 union::
442 /* SMBwriteX interface */
443 struct {
444 enum smb_write_level level;
445 struct {
446 union smb_handle file;
447 uint64_t offset;
448 uint16_t wmode;
449 uint16_t remaining;
450 uint32_t count;
451 const uint8_t *data;
452 } in;
453 struct {
454 uint32_t nwritten;
455 uint16_t remaining;
456 } out;
457 } writex, generic;
459 see the "in" and "out" sections? The "in" section is for parameters
460 that the SMB client sends on the wire as part of the request. The smbd
461 front end parse code parses the wire request and fills in all those
462 parameters. It then calls the NTVFS interface which looks like this::
464 NTSTATUS (*write)(struct request_context *req, union smb_write *io);
466 and the NTVFS backend does the write request. The backend then fills
467 in the "out" section of the writex structure and gives the union back
468 to the front end (either by returning, or if done in an async fashion
469 then by calling the async send function. See the async discussion
470 elsewhere in this document).
472 The NTVFS backend knows which particular function is being requested
473 by looking at io->generic.level. Notice that this enum is also
474 repeated inside each of the sub-structures in the union, so the
475 backend could just as easily look at io->writex.level and would get
476 the same variable.
478 Notice also that some levels (such as splwrite) don't have an "out"
479 section. This happens because there is no return value apart from a
480 status code from those SMB calls.
482 So what about status codes? The status code is returned directly by
483 the backend NTVFS interface when the call is performed
484 synchronously. When performed asynchronously then the status code is
485 put into req->async.status before the req->async.send_fn() callback is
486 called.
488 Currently the most complete NTVFS backend is the CIFS backend. I don't
489 expect this backend will be used much in production, but it does
490 provide the ideal test case for our NTVFS design. As it offers the
491 full capabilities that are possible with a CIFS server we can be sure
492 that we don't have any gaping holes in our APIs, and that the front
493 end code is flexible enough to handle any advances in the NT style
494 feature sets of Unix filesystems that make come along.
497 Process Models
498 --------------
500 In Samba3 we supported just one process model. It just so happens that
501 the process model that Samba3 supported is the "right" one for most
502 users, but there are situations where this model wasn't ideal.
504 In Samba4 you can choose the smbd process model on the smbd command
505 line.
508 DCERPC binding strings
509 ----------------------
511 When connecting to a dcerpc service you need to specify a binding
512 string.
514 The format is:
516 TRANSPORT:host[flags]
518 where TRANSPORT is either ncacn_np for SMB or ncacn_ip_tcp for RPC/TCP
520 "host" is an IP or hostname or netbios name. If the binding string
521 identifies the server side of an endpoint, "host" may be an empty
522 string.
524 "flags" can include a SMB pipe name if using the ncacn_np transport or
525 a TCP port number if using the ncacn_ip_tcp transport, otherwise they
526 will be auto-determined.
528 other recognised flags are:
530 sign : enable ntlmssp signing
531 seal : enable ntlmssp sealing
532 spnego : use SPNEGO instead of NTLMSSP authentication
533 krb5 : use KRB5 instead of NTLMSSP authentication
534 connect : enable rpc connect level auth (auth, but no sign or seal)
535 validate : enable the NDR validator
536 print : enable debugging of the packets
537 bigendian : use bigendian RPC
538 padcheck : check reply data for non-zero pad bytes
541 Here are some examples:
543 ncacn_np:myserver
544 ncacn_np:myserver[samr]
545 ncacn_np:myserver[\pipe\samr]
546 ncacn_np:myserver[/pipe/samr]
547 ncacn_np:myserver[samr,sign,print]
548 ncacn_np:myserver[sign,spnego]
549 ncacn_np:myserver[\pipe\samr,sign,seal,bigendian]
550 ncacn_np:myserver[/pipe/samr,seal,validate]
551 ncacn_np:
552 ncacn_np:[/pipe/samr]
553 ncacn_ip_tcp:myserver
554 ncacn_ip_tcp:myserver[1024]
555 ncacn_ip_tcp:myserver[sign,seal]
556 ncacn_ip_tcp:myserver[spnego,seal]
559 IDEA: Maybe extend UNC names like this?
561 smbclient //server/share
562 smbclient //server/share[sign,seal,spnego]
564 DCERPC Handles
565 --------------
566 The various handles that are used in the RPC servers should be created and
567 fetch using the dcesrv_handle_* functions.
569 Use dcesrv_handle_new(struct dcesrv_connection \*, uint8 handle_type) to obtain
570 a new handle of the specified type. Handle types are unique within each
571 pipe.
573 The handle can later be fetched again using::
575 struct dcesrv_handle *dcesrv_handle_fetch(struct dcesrv_connection *dce_conn, struct policy_handle *p, uint8 handle_type)
577 and destroyed by::
579 dcesrv_handle_destroy(struct dcesrv_handle *).
581 User data should be stored in the 'data' member of the dcesrv_handle struct.
584 MSRPC
585 -----
589 - ntvfs
590 - testing
591 - command line handling
592 - libcli structure
593 - posix reliance
594 - uid/gid handling
595 - process models
596 - static data
597 - msrpc
600 don't zero structures! avoid ZERO_STRUCT() and talloc_zero()
603 GMT vs TZ in printout of QFILEINFO timezones
605 put in full UNC path in tconx
607 test timezone handling by using a server in different zone from client
609 do {} while (0) system
611 NT_STATUS_IS_OK() is NOT the opposite of NT_STATUS_IS_ERR()
613 need to implement secondary parts of trans2 and nttrans in server and
614 client
616 document access_mask in openx reply
618 check all capabilities and flag1, flag2 fields (eg. EAs)
620 large files -> pass thru levels
622 setpathinfo is very fussy about null termination of the file name
624 the overwrite flag doesn't seem to work on setpathinfo RENAME_INFORMATION
626 END_OF_FILE_INFORMATION and ALLOCATION_INFORMATION don't seem to work
627 via setpathinfo
629 on w2k3 setpathinfo DISPOSITION_INFORMATION fails, but does have an
630 effect. It leaves the file with SHARING_VIOLATION.
632 on w2k3 trans2 setpathinfo with any invalid low numbered level causes
633 the file to get into a state where DELETE_PENDING is reported, and the
634 file cannot be deleted until you reboot
636 trans2 qpathinfo doesn't see the delete_pending flag correctly, but
637 qfileinfo does!
639 get rid of strtok
641 add programming documentation note about lp_set_cmdline()
643 need to add a wct checking function in all client parsing code,
644 similar to REQ_CHECK_WCT()
646 need to make sure that NTTIME is a round number of seconds when
647 converted from time_t
649 not using a zero next offset in SMB_FILE_STREAM_INFORMATION for last
650 entry causes explorer exception under win2000
653 if the server sets the session key the same for a second SMB socket as
654 an initial socket then the client will not re-authenticate, it will go
655 straight to a tconx, skipping session setup and will use all the
656 existing parameters! This allows two sockets with the same keys!?
659 removed blocking lock code, we now queue the whole request the same as
660 we queue any other pending request. This allows for things like a
661 close() while a pending blocking lock is being processed to operate
662 sanely.
664 disabled change notify code
666 disabled oplock code
670 MILESTONES
671 ==========
674 client library and test code
675 ----------------------------
677 convert client library to new structure
678 get smbtorture working
679 get smbclient working
680 expand client library for all requests
681 write per-request test suite
682 gentest randomised test suite
683 separate client code as a library for non-Samba use
685 server code
686 -----------
687 add remaining core SMB requests
688 add IPC layer
689 add nttrans layer
690 add rpc layer
691 fix auth models (share, server, rpc)
692 get net command working
693 connect CIFS backend to server level auth
694 get nmbd working
695 get winbindd working
696 reconnect printing code
697 restore removed smbd options
698 add smb.conf macro substitution code
699 add async backend notification
700 add generic timer event mechanism
702 clustering code
703 ---------------
705 write CIFS backend
706 new server models (break 1-1)
707 test clustered models
708 add fulcrum statistics gathering
710 docs
711 ----
713 conference paper
714 developer docs
716 svn instructions
718 Ideas
719 -----
721 - store all config in config.ldb
723 - load from smb.conf if modtime changes
725 - dump full system config with ldbsearch
727 - will need the ability to form a ldif difference file
729 - advanced web admin via a web ldb editor
731 - normal web admin via web forms -> ldif
733 - config.ldb will replace smb.conf, secrets.tdb, shares.tdb etc
735 - subsystems in smbd will load config parameters for a share
736 using ldbsearch at tconx time
738 - need a loadparm equivalent module that provides parameter defaults
740 - start smbd like this: "smbd -C tdb://etc/samba/config.ldb" or
741 "smbd -C ldapi://var/run/ldapi"
743 - write a tool that generates a template ldap schema from an existing
744 ldb+tdb file
746 - no need to HUP smbd to reload config
748 - how to handle configuration comments? same problem as SWAT
751 BUGS:
752 add a test case for last_entry_offset in trans2 find interfaces
753 conn refused
754 connect -> errno
755 no 137 resolution not possible
756 should not fallback to anon when pass supplied
757 should check pass-thu cap bit, and skip lots of tests
758 possibly allow the test suite to say "allow oversized replies" for trans2 and other calls
759 handle servers that don't have the setattre call in torture
760 add max file coponent length test and max path len test
761 check for alloc failure in all core reply.c and trans2.c code where allocation size depends on client parameter
763 case-insenstive idea:
764 all filenames on disk lowercase
765 real case in extended attribute
766 keep cache of what dirs are all lowercase
767 when searching for name, don't search if dir is definately all lowercase
768 when creating file, use dnotify to tell if someone else creates at
769 same time
771 solve del *.* idea:
772 make mangle cache dynamic size
773 fill during a dir scan
774 setup a timer
775 destroy cache after 30 sec
776 destroy if a 2nd dir scan happens on same dir