]>
Commit | Line | Data |
---|---|---|
7a68c94a | 1 | @node Name Service Switch, Users and Groups, Job Control, Top |
706074a5 | 2 | @chapter System Databases and Name Service Switch |
7a68c94a | 3 | @c %MENU% Accessing system databases |
2c6fe0bd UD |
4 | @cindex Name Service Switch |
5 | @cindex NSS | |
6d52618b | 6 | @cindex databases |
7a68c94a | 7 | |
706074a5 UD |
8 | Various functions in the C Library need to be configured to work |
9 | correctly in the local environment. Traditionally, this was done by | |
6952e59e | 10 | using files (e.g., @file{/etc/passwd}), but other nameservices (like the |
706074a5 UD |
11 | Network Information Service (NIS) and the Domain Name Service (DNS)) |
12 | became popular, and were hacked into the C library, usually with a fixed | |
41f49342 | 13 | search order. |
706074a5 | 14 | |
5455692a | 15 | @Theglibc{} contains a cleaner solution to this problem. It is |
706074a5 | 16 | designed after a method used by Sun Microsystems in the C library of |
1f77f049 | 17 | @w{Solaris 2}. @Theglibc{} follows their name and calls this |
706074a5 UD |
18 | scheme @dfn{Name Service Switch} (NSS). |
19 | ||
20 | Though the interface might be similar to Sun's version there is no | |
21 | common code. We never saw any source code of Sun's implementation and | |
6952e59e | 22 | so the internal interface is incompatible. This also manifests in the |
706074a5 UD |
23 | file names we use as we will see later. |
24 | ||
25 | ||
26 | @menu | |
27 | * NSS Basics:: What is this NSS good for. | |
28 | * NSS Configuration File:: Configuring NSS. | |
29 | * NSS Module Internals:: How does it work internally. | |
30 | * Extending NSS:: What to do to add services or databases. | |
31 | @end menu | |
32 | ||
33 | @node NSS Basics, NSS Configuration File, Name Service Switch, Name Service Switch | |
34 | @section NSS Basics | |
35 | ||
36 | The basic idea is to put the implementation of the different services | |
37 | offered to access the databases in separate modules. This has some | |
38 | advantages: | |
39 | ||
40 | @enumerate | |
41 | @item | |
1f77f049 | 42 | Contributors can add new services without adding them to @theglibc{}. |
706074a5 UD |
43 | @item |
44 | The modules can be updated separately. | |
45 | @item | |
46 | The C library image is smaller. | |
47 | @end enumerate | |
48 | ||
5455692a | 49 | To fulfill the first goal above, the ABI of the modules will be described |
706074a5 UD |
50 | below. For getting the implementation of a new service right it is |
51 | important to understand how the functions in the modules get called. | |
52 | They are in no way designed to be used by the programmer directly. | |
53 | Instead the programmer should only use the documented and standardized | |
54 | functions to access the databases. | |
55 | ||
56 | @noindent | |
57 | The databases available in the NSS are | |
58 | ||
59 | @cindex ethers | |
60 | @cindex group | |
61 | @cindex hosts | |
ba1ffaa1 | 62 | @cindex netgroup |
b3a86ae1 | 63 | @cindex networks |
706074a5 UD |
64 | @cindex protocols |
65 | @cindex passwd | |
66 | @cindex rpc | |
67 | @cindex services | |
68 | @cindex shadow | |
2fe82ca6 | 69 | @table @code |
26dee9c4 UD |
70 | @item aliases |
71 | Mail aliases | |
72 | @comment @pxref{Mail Aliases}. | |
706074a5 UD |
73 | @item ethers |
74 | Ethernet numbers, | |
75 | @comment @pxref{Ethernet Numbers}. | |
76 | @item group | |
77 | Groups of users, @pxref{Group Database}. | |
78 | @item hosts | |
79 | Host names and numbers, @pxref{Host Names}. | |
ba1ffaa1 UD |
80 | @item netgroup |
81 | Network wide list of host and users, @pxref{Netgroup Database}. | |
b3a86ae1 | 82 | @item networks |
706074a5 UD |
83 | Network names and numbers, @pxref{Networks Database}. |
84 | @item protocols | |
85 | Network protocols, @pxref{Protocols Database}. | |
86 | @item passwd | |
841785ba | 87 | User identities, @pxref{User Database}. |
706074a5 | 88 | @item rpc |
841785ba | 89 | Remote procedure call names and numbers. |
706074a5 UD |
90 | @comment @pxref{RPC Database}. |
91 | @item services | |
92 | Network services, @pxref{Services Database}. | |
93 | @item shadow | |
841785ba ZW |
94 | User passphrase hashes and related information. |
95 | @comment @pxref{Shadow Passphrase Database}. | |
2fe82ca6 | 96 | @end table |
706074a5 UD |
97 | |
98 | @noindent | |
26dee9c4 UD |
99 | There will be some more added later (@code{automount}, @code{bootparams}, |
100 | @code{netmasks}, and @code{publickey}). | |
706074a5 UD |
101 | |
102 | @node NSS Configuration File, NSS Module Internals, NSS Basics, Name Service Switch | |
103 | @section The NSS Configuration File | |
104 | ||
105 | @cindex @file{/etc/nsswitch.conf} | |
106 | @cindex @file{nsswitch.conf} | |
107 | Somehow the NSS code must be told about the wishes of the user. For | |
108 | this reason there is the file @file{/etc/nsswitch.conf}. For each | |
5455692a | 109 | database, this file contains a specification of how the lookup process should |
706074a5 UD |
110 | work. The file could look like this: |
111 | ||
112 | @example | |
113 | @include nsswitch.texi | |
114 | @end example | |
115 | ||
116 | The first column is the database as you can guess from the table above. | |
117 | The rest of the line specifies how the lookup process works. Please | |
118 | note that you specify the way it works for each database individually. | |
119 | This cannot be done with the old way of a monolithic implementation. | |
120 | ||
121 | The configuration specification for each database can contain two | |
122 | different items: | |
123 | ||
124 | @itemize @bullet | |
125 | @item | |
126 | the service specification like @code{files}, @code{db}, or @code{nis}. | |
127 | @item | |
26dee9c4 | 128 | the reaction on lookup result like @code{[NOTFOUND=return]}. |
706074a5 UD |
129 | @end itemize |
130 | ||
131 | @menu | |
ba1ffaa1 | 132 | * Services in the NSS configuration:: Service names in the NSS configuration. |
6d52618b | 133 | * Actions in the NSS configuration:: React appropriately to the lookup result. |
706074a5 UD |
134 | * Notes on NSS Configuration File:: Things to take care about while |
135 | configuring NSS. | |
136 | @end menu | |
137 | ||
138 | @node Services in the NSS configuration, Actions in the NSS configuration, NSS Configuration File, NSS Configuration File | |
139 | @subsection Services in the NSS configuration File | |
140 | ||
16d2c631 JM |
141 | The above example file mentions five different services: @code{files}, |
142 | @code{db}, @code{dns}, @code{nis}, and @code{nisplus}. This does not | |
143 | mean these | |
5455692a | 144 | services are available on all sites and neither does it mean these are |
706074a5 UD |
145 | all the services which will ever be available. |
146 | ||
147 | In fact, these names are simply strings which the NSS code uses to find | |
148 | the implicitly addressed functions. The internal interface will be | |
149 | described later. Visible to the user are the modules which implement an | |
150 | individual service. | |
151 | ||
152 | Assume the service @var{name} shall be used for a lookup. The code for | |
153 | this service is implemented in a module called @file{libnss_@var{name}}. | |
154 | On a system supporting shared libraries this is in fact a shared library | |
c66dbe00 | 155 | with the name (for example) @file{libnss_@var{name}.so.2}. The number |
706074a5 UD |
156 | at the end is the currently used version of the interface which will not |
157 | change frequently. Normally the user should not have to be cognizant of | |
158 | these files since they should be placed in a directory where they are | |
159 | found automatically. Only the names of all available services are | |
160 | important. | |
161 | ||
162 | @node Actions in the NSS configuration, Notes on NSS Configuration File, Services in the NSS configuration, NSS Configuration File | |
163 | @subsection Actions in the NSS configuration | |
164 | ||
165 | The second item in the specification gives the user much finer control | |
166 | on the lookup process. Action items are placed between two service | |
167 | names and are written within brackets. The general form is | |
168 | ||
2de99474 | 169 | @display |
57ba7bb4 | 170 | @code{[} ( @code{!}? @var{status} @code{=} @var{action} )+ @code{]} |
2de99474 | 171 | @end display |
706074a5 UD |
172 | |
173 | @noindent | |
174 | where | |
175 | ||
176 | @smallexample | |
177 | @var{status} @result{} success | notfound | unavail | tryagain | |
178 | @var{action} @result{} return | continue | |
179 | @end smallexample | |
180 | ||
181 | The case of the keywords is insignificant. The @var{status} | |
182 | values are the results of a call to a lookup function of a specific | |
ced8f893 | 183 | service. They mean: |
706074a5 UD |
184 | |
185 | @ftable @samp | |
186 | @item success | |
6d52618b | 187 | No error occurred and the wanted entry is returned. The default action |
706074a5 UD |
188 | for this is @code{return}. |
189 | ||
190 | @item notfound | |
191 | The lookup process works ok but the needed value was not found. The | |
192 | default action is @code{continue}. | |
193 | ||
194 | @item unavail | |
195 | @cindex DNS server unavailable | |
196 | The service is permanently unavailable. This can either mean the needed | |
197 | file is not available, or, for DNS, the server is not available or does | |
198 | not allow queries. The default action is @code{continue}. | |
199 | ||
200 | @item tryagain | |
201 | The service is temporarily unavailable. This could mean a file is | |
202 | locked or a server currently cannot accept more connections. The | |
203 | default action is @code{continue}. | |
204 | @end ftable | |
205 | ||
ced8f893 SG |
206 | @noindent |
207 | The @var{action} values mean: | |
208 | ||
209 | @ftable @samp | |
210 | @item return | |
211 | ||
212 | If the status matches, stop the lookup process at this service | |
213 | specification. If an entry is available, provide it to the application. | |
214 | If an error occurred, report it to the application. In case of a prior | |
215 | @samp{merge} action, the data is combined with previous lookup results, | |
216 | as explained below. | |
217 | ||
218 | @item continue | |
219 | ||
220 | If the status matches, proceed with the lookup process at the next | |
221 | entry, discarding the result of the current lookup (and any merged | |
222 | data). An exception is the @samp{initgroups} database and the | |
223 | @samp{success} status, where @samp{continue} acts like @code{merge} | |
224 | below. | |
225 | ||
226 | @item merge | |
227 | ||
228 | Proceed with the lookup process, retaining the current lookup result. | |
229 | This action is useful only with the @samp{success} status. If a | |
230 | subsequent service lookup succeeds and has a matching @samp{return} | |
231 | specification, the results are merged, the lookup process ends, and the | |
232 | merged results are returned to the application. If the following service | |
233 | has a matching @samp{merge} action, the lookup process continues, | |
234 | retaining the combined data from this and any previous lookups. | |
235 | ||
236 | After a @code{merge} action, errors from subsequent lookups are ignored, | |
237 | and the data gathered so far will be returned. | |
238 | ||
239 | The @samp{merge} only applies to the @samp{success} status. It is | |
240 | currently implemented for the @samp{group} database and its group | |
241 | members field, @samp{gr_mem}. If specified for other databases, it | |
242 | causes the lookup to fail (if the @var{status} matches). | |
243 | ||
244 | When processing @samp{merge} for @samp{group} membership, the group GID | |
245 | and name must be identical for both entries. If only one or the other is | |
246 | a match, the behavior is undefined. | |
247 | ||
248 | @end ftable | |
249 | ||
706074a5 UD |
250 | @noindent |
251 | If we have a line like | |
252 | ||
253 | @smallexample | |
254 | ethers: nisplus [NOTFOUND=return] db files | |
255 | @end smallexample | |
256 | ||
257 | @noindent | |
258 | this is equivalent to | |
259 | ||
260 | @smallexample | |
261 | ethers: nisplus [SUCCESS=return NOTFOUND=return UNAVAIL=continue | |
262 | TRYAGAIN=continue] | |
263 | db [SUCCESS=return NOTFOUND=continue UNAVAIL=continue | |
264 | TRYAGAIN=continue] | |
265 | files | |
266 | @end smallexample | |
267 | ||
268 | @noindent | |
269 | (except that it would have to be written on one line). The default | |
270 | value for the actions are normally what you want, and only need to be | |
271 | changed in exceptional cases. | |
272 | ||
273 | If the optional @code{!} is placed before the @var{status} this means | |
4f8dbcb1 | 274 | the following action is used for all statuses but @var{status} itself. |
706074a5 UD |
275 | I.e., @code{!} is negation as in the C language (and others). |
276 | ||
277 | Before we explain the exception which makes this action item necessary | |
278 | one more remark: obviously it makes no sense to add another action | |
279 | item after the @code{files} service. Since there is no other service | |
280 | following the action @emph{always} is @code{return}. | |
281 | ||
282 | @cindex nisplus, and completeness | |
283 | Now, why is this @code{[NOTFOUND=return]} action useful? To understand | |
284 | this we should know that the @code{nisplus} service is often | |
285 | complete; i.e., if an entry is not available in the NIS+ tables it is | |
286 | not available anywhere else. This is what is expressed by this action | |
287 | item: it is useless to examine further services since they will not give | |
288 | us a result. | |
289 | ||
290 | @cindex nisplus, and booting | |
291 | @cindex bootstrapping, and services | |
292 | The situation would be different if the NIS+ service is not available | |
293 | because the machine is booting. In this case the return value of the | |
294 | lookup function is not @code{notfound} but instead @code{unavail}. And | |
295 | as you can see in the complete form above: in this situation the | |
296 | @code{db} and @code{files} services are used. Neat, isn't it? The | |
297 | system administrator need not pay special care for the time the system | |
298 | is not completely ready to work (while booting or shutdown or | |
299 | network problems). | |
300 | ||
301 | ||
302 | @node Notes on NSS Configuration File, , Actions in the NSS configuration, NSS Configuration File | |
303 | @subsection Notes on the NSS Configuration File | |
304 | ||
305 | Finally a few more hints. The NSS implementation is not completely | |
306 | helpless if @file{/etc/nsswitch.conf} does not exist. For | |
307 | all supported databases there is a default value so it should normally | |
308 | be possible to get the system running even if the file is corrupted or | |
309 | missing. | |
310 | ||
ba1ffaa1 | 311 | @cindex default value, and NSS |
b3a86ae1 | 312 | For the @code{hosts} and @code{networks} databases the default value is |
ba1ffaa1 UD |
313 | @code{dns [!UNAVAIL=return] files}. I.e., the system is prepared for |
314 | the DNS service not to be available but if it is available the answer it | |
04b9968b | 315 | returns is definitive. |
ba1ffaa1 | 316 | |
10dc2a90 UD |
317 | The @code{passwd}, @code{group}, and @code{shadow} databases are |
318 | traditionally handled in a special way. The appropriate files in the | |
319 | @file{/etc} directory are read but if an entry with a name starting | |
320 | with a @code{+} character is found NIS is used. This kind of lookup | |
a1a78204 SE |
321 | remains possible if @theglibc{} was configured with the |
322 | @code{--enable-obsolete-nsl} option and the special lookup service | |
323 | @code{compat} is used. If @theglibc{} was configured with the | |
324 | @code{--enable-obsolete-nsl} option the default value for the three | |
325 | databases above is @code{compat [NOTFOUND=return] files}. If the | |
326 | @code{--enable-obsolete-nsl} option was not used the default value | |
327 | for the services is @code{files}. | |
328 | ||
329 | For all other databases the default value is @code{files} unless | |
330 | @theglibc{} was configured with @code{--enable-obsolete-rpc} option, in | |
331 | which case it the default value is @code{nis [NOTFOUND=return] files}. | |
ba1ffaa1 UD |
332 | |
333 | @cindex optimizing NSS | |
706074a5 | 334 | A second point is that the user should try to optimize the lookup |
10dc2a90 UD |
335 | process. The different service have different response times. |
336 | A simple file look up on a local file could be fast, but if the file | |
337 | is long and the needed entry is near the end of the file this may take | |
338 | quite some time. In this case it might be better to use the @code{db} | |
339 | service which allows fast local access to large data sets. | |
706074a5 UD |
340 | |
341 | Often the situation is that some global information like NIS must be | |
342 | used. So it is unavoidable to use service entries like @code{nis} etc. | |
343 | But one should avoid slow services like this if possible. | |
344 | ||
345 | ||
346 | @node NSS Module Internals, Extending NSS, NSS Configuration File, Name Service Switch | |
347 | @section NSS Module Internals | |
348 | ||
04b9968b | 349 | Now it is time to describe what the modules look like. The functions |
706074a5 UD |
350 | contained in a module are identified by their names. I.e., there is no |
351 | jump table or the like. How this is done is of no interest here; those | |
352 | interested in this topic should read about Dynamic Linking. | |
353 | @comment @ref{Dynamic Linking}. | |
354 | ||
355 | ||
356 | @menu | |
357 | * NSS Module Names:: Construction of the interface function of | |
358 | the NSS modules. | |
359 | * NSS Modules Interface:: Programming interface in the NSS module | |
360 | functions. | |
361 | @end menu | |
362 | ||
363 | @node NSS Module Names, NSS Modules Interface, NSS Module Internals, NSS Module Internals | |
364 | @subsection The Naming Scheme of the NSS Modules | |
365 | ||
366 | @noindent | |
5455692a | 367 | The name of each function consists of various parts: |
706074a5 UD |
368 | |
369 | @quotation | |
370 | _nss_@var{service}_@var{function} | |
371 | @end quotation | |
372 | ||
373 | @var{service} of course corresponds to the name of the module this | |
04b9968b UD |
374 | function is found in.@footnote{Now you might ask why this information is |
375 | duplicated. The answer is that we want to make it possible to link | |
706074a5 UD |
376 | directly with these shared objects.} The @var{function} part is derived |
377 | from the interface function in the C library itself. If the user calls | |
378 | the function @code{gethostbyname} and the service used is @code{files} | |
379 | the function | |
380 | ||
381 | @smallexample | |
382 | _nss_files_gethostbyname_r | |
383 | @end smallexample | |
384 | ||
385 | @noindent | |
386 | in the module | |
387 | ||
388 | @smallexample | |
c66dbe00 | 389 | libnss_files.so.2 |
706074a5 UD |
390 | @end smallexample |
391 | ||
392 | @noindent | |
393 | @cindex reentrant NSS functions | |
394 | is used. You see, what is explained above in not the whole truth. In | |
395 | fact the NSS modules only contain reentrant versions of the lookup | |
396 | functions. I.e., if the user would call the @code{gethostbyname_r} | |
397 | function this also would end in the above function. For all user | |
398 | interface functions the C library maps this call to a call to the | |
399 | reentrant function. For reentrant functions this is trivial since the | |
5455692a | 400 | interface is (nearly) the same. For the non-reentrant version the |
26dee9c4 UD |
401 | library keeps internal buffers which are used to replace the user |
402 | supplied buffer. | |
706074a5 UD |
403 | |
404 | I.e., the reentrant functions @emph{can} have counterparts. No service | |
405 | module is forced to have functions for all databases and all kinds to | |
406 | access them. If a function is not available it is simply treated as if | |
407 | the function would return @code{unavail} | |
408 | (@pxref{Actions in the NSS configuration}). | |
409 | ||
c66dbe00 UD |
410 | The file name @file{libnss_files.so.2} would be on a @w{Solaris 2} |
411 | system @file{nss_files.so.2}. This is the difference mentioned above. | |
2de99474 UD |
412 | Sun's NSS modules are usable as modules which get indirectly loaded |
413 | only. | |
414 | ||
1f77f049 | 415 | The NSS modules in @theglibc{} are prepared to be used as normal |
b642f101 | 416 | libraries themselves. This is @emph{not} true at the moment, though. |
04b9968b | 417 | However, the organization of the name space in the modules does not make it |
2de99474 UD |
418 | impossible like it is for Solaris. Now you can see why the modules are |
419 | still libraries.@footnote{There is a second explanation: we were too | |
420 | lazy to change the Makefiles to allow the generation of shared objects | |
04b9968b | 421 | not starting with @file{lib} but don't tell this to anybody.} |
2de99474 | 422 | |
706074a5 UD |
423 | |
424 | @node NSS Modules Interface, , NSS Module Names, NSS Module Internals | |
425 | @subsection The Interface of the Function in NSS Modules | |
426 | ||
427 | Now we know about the functions contained in the modules. It is now | |
428 | time to describe the types. When we mentioned the reentrant versions of | |
429 | the functions above, this means there are some additional arguments | |
5455692a | 430 | (compared with the standard, non-reentrant versions). The prototypes for |
706074a5 UD |
431 | the non-reentrant and reentrant versions of our function above are: |
432 | ||
433 | @smallexample | |
434 | struct hostent *gethostbyname (const char *name) | |
435 | ||
a18f587d UD |
436 | int gethostbyname_r (const char *name, struct hostent *result_buf, |
437 | char *buf, size_t buflen, struct hostent **result, | |
438 | int *h_errnop) | |
706074a5 UD |
439 | @end smallexample |
440 | ||
441 | @noindent | |
ba1ffaa1 | 442 | The actual prototype of the function in the NSS modules in this case is |
706074a5 UD |
443 | |
444 | @smallexample | |
ba1ffaa1 UD |
445 | enum nss_status _nss_files_gethostbyname_r (const char *name, |
446 | struct hostent *result_buf, | |
a18f587d | 447 | char *buf, size_t buflen, |
c66dbe00 | 448 | int *errnop, int *h_errnop) |
706074a5 UD |
449 | @end smallexample |
450 | ||
ba1ffaa1 | 451 | I.e., the interface function is in fact the reentrant function with the |
3a35923e FW |
452 | change of the return value, the omission of the @var{result} parameter, |
453 | and the addition of the @var{errnop} parameter. While the user-level | |
454 | function returns a pointer to the result the reentrant function return | |
455 | an @code{enum nss_status} value: | |
706074a5 | 456 | |
b642f101 | 457 | @vtable @code |
706074a5 UD |
458 | @item NSS_STATUS_TRYAGAIN |
459 | numeric value @code{-2} | |
460 | ||
461 | @item NSS_STATUS_UNAVAIL | |
462 | numeric value @code{-1} | |
463 | ||
464 | @item NSS_STATUS_NOTFOUND | |
465 | numeric value @code{0} | |
466 | ||
467 | @item NSS_STATUS_SUCCESS | |
468 | numeric value @code{1} | |
b642f101 | 469 | @end vtable |
706074a5 UD |
470 | |
471 | @noindent | |
472 | Now you see where the action items of the @file{/etc/nsswitch.conf} file | |
473 | are used. | |
474 | ||
ba1ffaa1 UD |
475 | If you study the source code you will find there is a fifth value: |
476 | @code{NSS_STATUS_RETURN}. This is an internal use only value, used by a | |
477 | few functions in places where none of the above value can be used. If | |
478 | necessary the source code should be examined to learn about the details. | |
479 | ||
7be8096f UD |
480 | In case the interface function has to return an error it is important |
481 | that the correct error code is stored in @code{*@var{errnop}}. Some | |
5455692a | 482 | return status values have only one associated error code, others have |
7be8096f UD |
483 | more. |
484 | ||
485 | @multitable @columnfractions .3 .2 .50 | |
486 | @item | |
487 | @code{NSS_STATUS_TRYAGAIN} @tab | |
04b9968b | 488 | @code{EAGAIN} @tab One of the functions used ran temporarily out of |
7be8096f UD |
489 | resources or a service is currently not available. |
490 | @item | |
491 | @tab | |
492 | @code{ERANGE} @tab The provided buffer is not large enough. | |
493 | The function should be called again with a larger buffer. | |
494 | @item | |
495 | @code{NSS_STATUS_UNAVAIL} @tab | |
496 | @code{ENOENT} @tab A necessary input file cannot be found. | |
497 | @item | |
498 | @code{NSS_STATUS_NOTFOUND} @tab | |
499 | @code{ENOENT} @tab The requested entry is not available. | |
d4e301c5 CD |
500 | |
501 | @item | |
502 | @code{NSS_STATUS_NOTFOUND} @tab | |
503 | @code{SUCCESS} @tab There are no entries. | |
504 | Use this to avoid returning errors for inactive services which may | |
505 | be enabled at a later time. This is not the same as the service | |
506 | being temporarily unavailable. | |
7be8096f UD |
507 | @end multitable |
508 | ||
509 | These are proposed values. There can be other error codes and the | |
510 | described error codes can have different meaning. @strong{With one | |
511 | exception:} when returning @code{NSS_STATUS_TRYAGAIN} the error code | |
512 | @code{ERANGE} @emph{must} mean that the user provided buffer is too | |
5455692a | 513 | small. Everything else is non-critical. |
7be8096f | 514 | |
3a35923e FW |
515 | In statically linked programs, the main application and NSS modules do |
516 | not share the same thread-local variable @code{errno}, which is the | |
517 | reason why there is an explicit @var{errnop} function argument. | |
518 | ||
ba1ffaa1 | 519 | The above function has something special which is missing for almost all |
706074a5 UD |
520 | the other module functions. There is an argument @var{h_errnop}. This |
521 | points to a variable which will be filled with the error code in case | |
3a35923e FW |
522 | the execution of the function fails for some reason. (In statically |
523 | linked programs, the thread-local variable @code{h_errno} is not shared | |
524 | with the main application.) | |
706074a5 UD |
525 | |
526 | The @code{get@var{XXX}by@var{YYY}} functions are the most important | |
527 | functions in the NSS modules. But there are others which implement | |
528 | the other ways to access system databases (say for the | |
841785ba | 529 | user database, there are @code{setpwent}, @code{getpwent}, and |
706074a5 UD |
530 | @code{endpwent}). These will be described in more detail later. |
531 | Here we give a general way to determine the | |
532 | signature of the module function: | |
533 | ||
534 | @itemize @bullet | |
535 | @item | |
3a35923e | 536 | the return value is @code{enum nss_status}; |
706074a5 | 537 | @item |
3a35923e | 538 | the name (@pxref{NSS Module Names}); |
706074a5 UD |
539 | @item |
540 | the first arguments are identical to the arguments of the non-reentrant | |
541 | function; | |
542 | @item | |
3a35923e | 543 | the next four arguments are: |
706074a5 UD |
544 | |
545 | @table @code | |
26dee9c4 | 546 | @item STRUCT_TYPE *result_buf |
706074a5 UD |
547 | pointer to buffer where the result is stored. @code{STRUCT_TYPE} is |
548 | normally a struct which corresponds to the database. | |
549 | @item char *buffer | |
0bc93a2f | 550 | pointer to a buffer where the function can store additional data for |
706074a5 | 551 | the result etc. |
26dee9c4 | 552 | @item size_t buflen |
706074a5 | 553 | length of the buffer pointed to by @var{buffer}. |
3a35923e FW |
554 | @item int *errnop |
555 | the low-level error code to return to the application. If the return | |
556 | value is not @code{NSS_STATUS_SUCCESS}, @code{*@var{errnop}} needs to be | |
557 | set to a non-zero value. An NSS module should never set | |
558 | @code{*@var{errnop}} to zero. The value @code{ERANGE} is special, as | |
559 | described above. | |
706074a5 UD |
560 | @end table |
561 | ||
562 | @item | |
563 | possibly a last argument @var{h_errnop}, for the host name and network | |
3a35923e FW |
564 | name lookup functions. If the return value is not |
565 | @code{NSS_STATUS_SUCCESS}, @code{*@var{h_errnop}} needs to be set to a | |
566 | non-zero value. A generic error code is @code{NETDB_INTERNAL}, which | |
567 | instructs the caller to examine @code{*@var{errnop}} for further | |
568 | details. (This includes the @code{ERANGE} special case.) | |
706074a5 UD |
569 | @end itemize |
570 | ||
571 | @noindent | |
572 | This table is correct for all functions but the @code{set@dots{}ent} | |
573 | and @code{end@dots{}ent} functions. | |
574 | ||
575 | ||
576 | @node Extending NSS, , NSS Module Internals, Name Service Switch | |
577 | @section Extending NSS | |
578 | ||
579 | One of the advantages of NSS mentioned above is that it can be extended | |
580 | quite easily. There are two ways in which the extension can happen: | |
581 | adding another database or adding another service. The former is | |
582 | normally done only by the C library developers. It is | |
583 | here only important to remember that adding another database is | |
584 | independent from adding another service because a service need not | |
585 | support all databases or lookup functions. | |
586 | ||
ef48b196 | 587 | A designer/implementer of a new service is therefore free to choose the |
706074a5 UD |
588 | databases s/he is interested in and leave the rest for later (or |
589 | completely aside). | |
590 | ||
591 | @menu | |
592 | * Adding another Service to NSS:: What is to do to add a new service. | |
593 | * NSS Module Function Internals:: Guidelines for writing new NSS | |
594 | service functions. | |
595 | @end menu | |
596 | ||
597 | @node Adding another Service to NSS, NSS Module Function Internals, Extending NSS, Extending NSS | |
598 | @subsection Adding another Service to NSS | |
599 | ||
1f77f049 JM |
600 | The sources for a new service need not (and should not) be part of @theglibc{} |
601 | itself. The developer retains complete control over the | |
706074a5 UD |
602 | sources and its development. The links between the C library and the |
603 | new service module consists solely of the interface functions. | |
604 | ||
605 | Each module is designed following a specific interface specification. | |
c66dbe00 UD |
606 | For now the version is 2 (the interface in version 1 was not adequate) |
607 | and this manifests in the version number of the shared library object of | |
608 | the NSS modules: they have the extension @code{.2}. If the interface | |
609 | changes again in an incompatible way, this number will be increased. | |
706074a5 UD |
610 | Modules using the old interface will still be usable. |
611 | ||
612 | Developers of a new service will have to make sure that their module is | |
613 | created using the correct interface number. This means the file itself | |
0bc93a2f | 614 | must have the correct name and on ELF systems the @dfn{soname} (Shared |
706074a5 UD |
615 | Object Name) must also have this number. Building a module from a bunch |
616 | of object files on an ELF system using GNU CC could be done like this: | |
617 | ||
618 | @smallexample | |
c66dbe00 | 619 | gcc -shared -o libnss_NAME.so.2 -Wl,-soname,libnss_NAME.so.2 OBJECTS |
706074a5 UD |
620 | @end smallexample |
621 | ||
622 | @noindent | |
623 | @ref{Link Options, Options for Linking, , gcc, GNU CC}, to learn | |
624 | more about this command line. | |
625 | ||
626 | To use the new module the library must be able to find it. This can be | |
627 | achieved by using options for the dynamic linker so that it will search | |
04b9968b | 628 | the directory where the binary is placed. For an ELF system this could be |
706074a5 UD |
629 | done by adding the wanted directory to the value of |
630 | @code{LD_LIBRARY_PATH}. | |
631 | ||
04b9968b | 632 | But this is not always possible since some programs (those which run |
706074a5 UD |
633 | under IDs which do not belong to the user) ignore this variable. |
634 | Therefore the stable version of the module should be placed into a | |
635 | directory which is searched by the dynamic linker. Normally this should | |
636 | be the directory @file{$prefix/lib}, where @file{$prefix} corresponds to | |
637 | the value given to configure using the @code{--prefix} option. But be | |
638 | careful: this should only be done if it is clear the module does not | |
639 | cause any harm. System administrators should be careful. | |
640 | ||
641 | ||
642 | @node NSS Module Function Internals, , Adding another Service to NSS, Extending NSS | |
643 | @subsection Internals of the NSS Module Functions | |
644 | ||
645 | Until now we only provided the syntactic interface for the functions in | |
04b9968b | 646 | the NSS module. In fact there is not much more we can say since the |
706074a5 UD |
647 | implementation obviously is different for each function. But a few |
648 | general rules must be followed by all functions. | |
649 | ||
650 | In fact there are four kinds of different functions which may appear in | |
651 | the interface. All derive from the traditional ones for system databases. | |
652 | @var{db} in the following table is normally an abbreviation for the | |
841785ba | 653 | database (e.g., it is @code{pw} for the user database). |
706074a5 UD |
654 | |
655 | @table @code | |
a18f587d | 656 | @item enum nss_status _nss_@var{database}_set@var{db}ent (void) |
706074a5 UD |
657 | This function prepares the service for following operations. For a |
658 | simple file based lookup this means files could be opened, for other | |
659 | services this function simply is a noop. | |
660 | ||
661 | One special case for this function is that it takes an additional | |
662 | argument for some @var{database}s (i.e., the interface is | |
663 | @code{int set@var{db}ent (int)}). @ref{Host Names}, which describes the | |
664 | @code{sethostent} function. | |
665 | ||
666 | The return value should be @var{NSS_STATUS_SUCCESS} or according to the | |
667 | table above in case of an error (@pxref{NSS Modules Interface}). | |
668 | ||
a18f587d | 669 | @item enum nss_status _nss_@var{database}_end@var{db}ent (void) |
706074a5 UD |
670 | This function simply closes all files which are still open or removes |
671 | buffer caches. If there are no files or buffers to remove this is again | |
672 | a simple noop. | |
673 | ||
5455692a | 674 | There normally is no return value other than @var{NSS_STATUS_SUCCESS}. |
706074a5 | 675 | |
c66dbe00 | 676 | @item enum nss_status _nss_@var{database}_get@var{db}ent_r (@var{STRUCTURE} *result, char *buffer, size_t buflen, int *errnop) |
706074a5 UD |
677 | Since this function will be called several times in a row to retrieve |
678 | one entry after the other it must keep some kind of state. But this | |
679 | also means the functions are not really reentrant. They are reentrant | |
680 | only in that simultaneous calls to this function will not try to | |
681 | write the retrieved data in the same place (as it would be the case for | |
682 | the non-reentrant functions); instead, it writes to the structure | |
683 | pointed to by the @var{result} parameter. But the calls share a common | |
684 | state and in the case of a file access this means they return neighboring | |
685 | entries in the file. | |
686 | ||
687 | The buffer of length @var{buflen} pointed to by @var{buffer} can be used | |
688 | for storing some additional data for the result. It is @emph{not} | |
689 | guaranteed that the same buffer will be passed for the next call of this | |
690 | function. Therefore one must not misuse this buffer to save some state | |
691 | information from one call to another. | |
692 | ||
3a35923e | 693 | Before the function returns with a failure code, the implementation |
010fe231 | 694 | should store the value of the local @code{errno} variable in the variable |
3a35923e FW |
695 | pointed to be @var{errnop}. This is important to guarantee the module |
696 | working in statically linked programs. The stored value must not be | |
697 | zero. | |
c66dbe00 | 698 | |
706074a5 UD |
699 | As explained above this function could also have an additional last |
700 | argument. This depends on the database used; it happens only for | |
b3a86ae1 | 701 | @code{host} and @code{networks}. |
706074a5 | 702 | |
04b9968b | 703 | The function shall return @code{NSS_STATUS_SUCCESS} as long as there are |
706074a5 UD |
704 | more entries. When the last entry was read it should return |
705 | @code{NSS_STATUS_NOTFOUND}. When the buffer given as an argument is too | |
706 | small for the data to be returned @code{NSS_STATUS_TRYAGAIN} should be | |
707 | returned. When the service was not formerly initialized by a call to | |
5455692a | 708 | @code{_nss_@var{DATABASE}_set@var{db}ent} all return values allowed for |
706074a5 UD |
709 | this function can also be returned here. |
710 | ||
c66dbe00 | 711 | @item enum nss_status _nss_@var{DATABASE}_get@var{db}by@var{XX}_r (@var{PARAMS}, @var{STRUCTURE} *result, char *buffer, size_t buflen, int *errnop) |
706074a5 UD |
712 | This function shall return the entry from the database which is |
713 | addressed by the @var{PARAMS}. The type and number of these arguments | |
714 | vary. It must be individually determined by looking to the user-level | |
715 | interface functions. All arguments given to the non-reentrant version | |
716 | are here described by @var{PARAMS}. | |
717 | ||
718 | The result must be stored in the structure pointed to by @var{result}. | |
5455692a | 719 | If there are additional data to return (say strings, where the |
706074a5 | 720 | @var{result} structure only contains pointers) the function must use the |
5455692a | 721 | @var{buffer} of length @var{buflen}. There must not be any references |
706074a5 UD |
722 | to non-constant global data. |
723 | ||
0bc93a2f | 724 | The implementation of this function should honor the @var{stayopen} |
706074a5 UD |
725 | flag set by the @code{set@var{DB}ent} function whenever this makes sense. |
726 | ||
5455692a | 727 | Before the function returns, the implementation should store the value of |
010fe231 | 728 | the local @code{errno} variable in the variable pointed to by |
5455692a | 729 | @var{errnop}. This is important to guarantee the module works in |
c66dbe00 UD |
730 | statically linked programs. |
731 | ||
706074a5 | 732 | Again, this function takes an additional last argument for the |
b3a86ae1 | 733 | @code{host} and @code{networks} database. |
706074a5 UD |
734 | |
735 | The return value should as always follow the rules given above | |
736 | (@pxref{NSS Modules Interface}). | |
737 | ||
738 | @end table |