1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
|
<html>
<head>
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
<meta http-equiv="Content-Language" content="en" />
<title>s6-dns: the s6dns_engine library interface</title>
<meta name="Description" content="s6-dns: the s6dns_engine library interface" />
<meta name="Keywords" content="s6-dns dns s6dns_engine library libs6dns" />
<!-- <link rel="stylesheet" type="text/css" href="//skarnet.org/default.css" /> -->
</head>
<body>
<p>
<a href="index.html">libs6dns</a><br />
<a href="../">s6-dns</a><br />
<a href="//skarnet.org/software/">Software</a><br />
<a href="//skarnet.org/">skarnet.org</a>
</p>
<h1> The <tt>s6dns_engine</tt> library interface </h1>
<p>
The following functions are declared in the <tt>s6-dns/s6dns-engine.h</tt> header,
and implemented in the <tt>libs6dns.a</tt> or <tt>libs6dns.so</tt> library.
</p>
<h2> General information </h2>
<p>
<tt>s6dns_engine</tt> is the nitty-gritty of DNS query management. These
are the low-level asynchronous primitives sending DNS queries over the
network, and getting answers.
</p>
<p>
<tt>s6dns_engine</tt> has been inspired by Dan J. Bernstein's
<a href="https://cr.yp.to/djbdns/dns_transmit.html">dns_transmit</a>
library, but does not borrow any code from it. Unlike
<tt>dns_transmit</tt>, <tt>s6dns_engine</tt> does not assume that
network send operations are instant successes; <tt>s6dns_engine</tt>
descriptors can be selected for writing as well as for reading.
Also, if the underlying <a href="//skarnet.org/software/skalibs">
skalibs</a> has been compiled with IPv6 support, <tt>s6dns_engine</tt>
supports native IPv6 transport.
</p>
<p>
The <tt>s6dns_engine</tt> functions are used in the implementation of the
<tt>s6dns_resolven_loop</tt> function - which is nothing more than a
simple event loop around the <tt>s6dns_engine</tt> primitives - and the
<a href="../skadns/skadnsd.html">skadnsd</a> daemon. Both pieces of code are
good examples of how to use <tt>s6dns_engine</tt>.
</p>
<p>
However, unless you're implementing a DNS cache, you probably should
not call the
<tt>s6dns_engine</tt> functions directly: they are very low-level. If you
need synchronous resolution, use the
<a href="s6dns-resolve.html">s6dns_resolve</a> functions. If you need
asynchronous DNS operations, use the
<a href="../skadns/index.html">skadns</a> functions, which are
designed to provide a higher level interface to multiple asynchronous
DNS queries.
</p>
<h2> Data structures </h2>
<p>
A <tt>s6dns_engine_t</tt> structure holds all the data necessary to
manage a DNS query (and response). It must be initialized to S6DNS_ENGINE_ZERO
when first declared, and recycled with <tt>s6dns_engine_recycle()</tt>
after each use. It contains a stralloc, so it must be freed with
<tt>s6dns_engine_free()</tt> before being discarded, to avoid memory leaks.
</p>
<h2> Functions </h2>
<h3> <tt>s6dns_engine_t</tt> life cycle </h3>
<p>
<code> int s6dns_engine_init (s6dns_engine_t *dt, s6dns_ip46list_t const *servers, uint32_t options, char const *q, unsigned int qlen, uint16_t qtype, tain_t const *deadline, tain_t const *stamp) </code>
</p>
<p>
Initializes <em>dt</em> with query <em>q</em> of length <em>qlen</em>
and type <em>qtype</em>. If <tt>d</tt> is an
encoded <tt>s6dns_domain_t</tt>, then <tt>d.s</tt> and <tt>d.len</tt>
are appropriate candidates for arguments <em>q</em> and <em>qlen</em>
respectively.
</p>
<p>
<em>options</em> can be 0 or an OR'ed
combination of the following, defined in <tt>s6-dns/s6dns-constants.h</tt>:
</p>
<ul>
<li> S6DNS_O_RECURSIVE: the query will be recursive and assuming it is
sent to a DNS cache, instead of iterative and assuming it is sent to a
DNS server. </li>
<li> S6DNS_O_STRICT: the library will only accept authoritative answers
to iterative queries. This is normally the sane behaviour, but badly
configured DNS software around the world - notably, <a href="../bind.html">
BIND</a> when it's configured to be both a cache and a server - often
serve <em>non-</em>authoritative data even when they could, so it
breaks things, hence why the option isn't set by default. </li>
</ul>
<p>
<em>servers</em> must point to a list of IP addresses as defined in
<a href="s6dns-ip46.html">s6-dns/s6dns-ip46.h</a>. Such a list can be
obtained from the <tt>/etc/resolv.conf</tt> file via the
<a href="s6dns-rci.html">s6dns_rci_fill()</a> call when performing a
recursive query, or it must be constructed from a list of relevant
NS addresses when performing an iterative query.
</p>
<p>
<em>stamp</em> must be an accurate enough timestamp. <em>deadline</em>
sets up a deadline for the query: if the query hasn't been
satisfactorily answered by <em>deadline</em> - no matter how many
round-trips to network servers the library performs internally - then
it will be considered failed, and a timeout will be reported.
</p>
<p>
The function returns 1 if all went well, and 0 if an error occurred.
It returns instantly; it <em>does not</em> perform any network operation,
it just prepares <em>dt</em> to send a query. The actual data sending
will take place on the next <tt>s6dns_engine_event()</tt> call.
</p>
<p>
<code> void s6dns_engine_recycle (s6dns_engine_t *dt) </code>
</p>
<p>
Recycles <tt>dt</tt>, making it ready for another use. This function
does not deallocate the heap memory used by dt, so it's faster than
<tt>s6dns_engine_free()</tt> and does not cause heap fragmentation.
</p>
<p>
<code> void s6dns_engine_free (s6dns_engine_t *dt) </code>
</p>
<p>
Frees the heap memory used by <tt>dt</tt>. Also makes <tt>dt</tt>
ready for another use. It's advised to only use this function when
certain that <em>dt</em> will not be reused.
</p>
<h3> Before the <tt>iopause()</tt> </h3>
<p>
The descriptor to select on is available as the <tt>fd</tt> field in
the <tt>s6dns_engine_t</tt> structure.
<em>dt</em>→fd should be read every iteration, because it can
change between iterations even if no event or timeout is reported
for <em>dt</em>.
</p>
<p>
<code> void s6dns_engine_nextdeadline (s6dns_engine_t const *dt, tain_t *a) </code>
</p>
<p>
If <em>dt</em> needs handling before the absolute date *<em>a</em>,
then *<em>a</em> is updated
so it contains the earlier date. This is useful to compute the next
deadline in an <tt>iopause()</tt> loop.
</p>
<p>
<code> int s6dns_engine_isreadable (s6dns_engine_t const *dt) </code>
</p>
<p>
Returns nonzero iff <em>dt</em>→fd is to be selected for reading.
Should be called in every iteration.
</p>
<p>
<code> int s6dns_engine_iswritable (s6dns_engine_t const *dt) </code>
</p>
<p>
Returns nonzero iff <em>dt</em>→fd is to be selected for writing.
Should be called in every iteration.
</p>
<h3> After the <tt>iopause()</tt> </h3>
<p>
<code> int s6dns_engine_timeout (s6dns_engine_t *dt, tain_t const *stamp) </code>
</p>
<p>
This function should be called if your selecting function returned 0, which
means that an event timed out.
<em>stamp</em> should contain the current time. The function returns -1 if
an error occurred, 1 if <em>dt</em> actually timed out, and 0 if nothing
special happened to <em>dt</em> (and your iopause timeout was caused by
something else). If the return value is not 0, <em>dt</em> is automatically
recycled.
</p>
<p>
<code> int s6dns_engine_event (s6dns_engine_t *dt, tain_t const *stamp) </code>
</p>
<p>
This function should be called if your selecting function returned a positive
number, which means that some event got triggered.
<em>stamp</em> should contain the current time. The function returns
-1 if an error occurred (and <em>dt</em> is automatically recycled). It
returns 0 if no answer has arrived yet, and 1 if an answer is available.
</p>
<p>
The <tt>s6dns_engine_timeout()</tt> and <tt>s6dns_engine_event()</tt> functions,
when returning -1, make use of the following error codes:
</p>
<ul>
<li> EINVAL: Invalid <em>dt</em>.
<li> ENETUNREACH: All the servers in the <em>servers</em> list have been
unsuccessfully tried. </li>
<li> EPROTO: An answer arrived, but it didn't follow the DNS protocol. </li>
<li> Other error codes reporting socket or system failures. </li>
</ul>
<p>
<code> char *s6dns_engine_packet (s6dns_engine_t const *dt) </code>
</p>
<p>
Points to the response packet received from the network,
if <tt>s6dns_engine_event()</tt> returned 1.
</p>
<p>
<code> unsigned int s6dns_engine_packetlen (s6dns_engine_t const *dt) </code>
</p>
<p>
Gives the length of the response packet,
if <tt>s6dns_engine_event()</tt> returned 1.
</p>
<p>
You should recycle or free <em>dt</em> after reading the response packet.
</p>
<p>
<code> void s6dns_engine_query (s6dns_engine_t const *dt, char **name, uint16_t *len, uint16_t *qtype) </code>
</p>
<p>
Recalls the query encoded in <em>*dt</em>: stores a pointer to the encoded name in
<em>*name</em>, its length in <em>*len</em> and the query type in <em>*qtype</em>.
You can call this function
even if <em>dt</em> has been recycled - but, of course, not if it has been freed.
</p>
</body>
</html>
|
