xref: /freebsd/lib/libopenbsd/ohash_init.3 (revision d6b92ffa)
1.\"	$OpenBSD: ohash_init.3,v 1.2 2014/05/13 14:01:41 jmc Exp $
2.\" Copyright (c) 1999 Marc Espie <espie@openbsd.org>
3.\"
4.\" Permission to use, copy, modify, and distribute this software for any
5.\" purpose with or without fee is hereby granted, provided that the above
6.\" copyright notice and this permission notice appear in all copies.
7.\"
8.\" THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9.\" WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10.\" MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11.\" ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12.\" WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13.\" ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14.\" OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15.\"
16.\" $FreeBSD$
17.\"
18.Dd May 12, 2014
19.Dt OHASH_INIT 3
20.Os
21.Sh NAME
22.Nm ohash_init ,
23.Nm ohash_delete ,
24.Nm ohash_lookup_interval ,
25.Nm ohash_lookup_memory ,
26.Nm ohash_find ,
27.Nm ohash_remove ,
28.Nm ohash_insert ,
29.Nm ohash_first ,
30.Nm ohash_next ,
31.Nm ohash_entries
32.Nd light-weight open hashing
33.Sh SYNOPSIS
34.In stdint.h
35.In stddef.h
36.In ohash.h
37.Ft void
38.Fn ohash_init "struct ohash *h" "unsigned int size" "struct ohash_info *info"
39.Ft void
40.Fn ohash_delete "struct ohash *h"
41.Ft "unsigned int"
42.Fn ohash_lookup_interval "struct ohash *h" "const char *start" "const char *end" "uint32_t hv"
43.Ft "unsigned int"
44.Fn ohash_lookup_memory "struct ohash *h" "const char *k" "size_t s" "uint32_t hv"
45.Ft void *
46.Fn ohash_find "struct ohash *h" "unsigned int i"
47.Ft void *
48.Fn ohash_remove "struct ohash *h" "unsigned int i"
49.Ft void *
50.Fn ohash_insert "struct ohash *h" "unsigned int i" "void *p"
51.Ft void *
52.Fn ohash_first "struct ohash *h" "unsigned int *i"
53.Ft void *
54.Fn ohash_next "struct ohash *h" "unsigned int *i"
55.Ft "unsigned int"
56.Fn ohash_entries "struct ohash *h"
57.Sh DESCRIPTION
58These functions have been designed as a fast, extensible alternative to
59the usual hash table functions.
60They provide storage and retrieval of records indexed by keys,
61where a key is a contiguous sequence of bytes at a fixed position in
62each record.
63Keys can either be NUL-terminated strings or fixed-size memory areas.
64All functions take a pointer to an ohash structure as the
65.Fa h
66function argument.
67Storage for this structure should be provided by user code.
68.Pp
69.Fn ohash_init
70initializes the table to store roughly 2 to the power
71.Fa size
72elements.
73.Fa info
74is a pointer to a
75.Fa struct ohash_info .
76.Bd -literal -offset indent
77struct ohash_info {
78	ptrdiff_t key_offset;
79	void *data;	/* user data */
80	void *(*calloc)(size_t, size_t, void *);
81	void (*free)(void *, void *);
82	void *(*alloc)(size_t, void *);
83};
84.Ed
85.Pp
86The
87.Va offset
88field holds the position of the key in each record;
89the
90.Va calloc
91and
92.Va free
93fields are pointers to
94.Xr calloc 3
95and
96.Xr free 3 Ns -like
97functions, used for managing the table internal storage;
98the
99.Va alloc
100field is only used by the utility function
101.Xr ohash_create_entry 3 .
102.Pp
103Each of these functions are called similarly to their standard counterpart,
104but with an extra
105.Ft void *
106parameter corresponding to the content of the field
107.Fa data ,
108which can be used to communicate specific information to the functions.
109.Pp
110.Fn ohash_init
111stores a copy of those fields internally, so
112.Fa info
113can be reclaimed after initialization.
114.Pp
115.Fn ohash_delete
116frees storage internal to
117.Fa h .
118Elements themselves should be freed by the user first, using for instance
119.Fn ohash_first
120and
121.Fn ohash_next .
122.Pp
123.Fn ohash_lookup_interval
124and
125.Fn ohash_lookup_memory
126are the basic look-up element functions.
127The hashing function result is provided by the user as
128.Fa hv .
129These return a
130.Qq slot
131in the ohash table
132.Fa h ,
133to be used with
134.Fn ohash_find ,
135.Fn ohash_insert ,
136or
137.Fn ohash_remove .
138This slot is only valid up to the next call to
139.Fn ohash_insert
140or
141.Fn ohash_remove .
142.Pp
143.Fn ohash_lookup_interval
144handles string-like keys.
145.Fn ohash_lookup_interval
146assumes the key is the interval between
147.Fa start
148and
149.Fa end ,
150exclusive,
151though the actual elements stored in the table should only contain
152NUL-terminated keys.
153.Pp
154.Fn ohash_lookup_memory
155assumes the key is the memory area starting at
156.Fa k
157of size
158.Fa s .
159All bytes are significant in key comparison.
160.Pp
161.Fn ohash_find
162retrieves an element from a slot
163.Fa i
164returned by the
165.Fn ohash_lookup*
166functions.
167It returns
168.Dv NULL
169if the slot is empty.
170.Pp
171.Fn ohash_insert
172inserts a new element
173.Fa p
174at slot
175.Fa i .
176Slot
177.Fa i
178must be empty and element
179.Fa p
180must have a key corresponding to the
181.Fn ohash_lookup*
182call.
183.Pp
184.Fn ohash_remove
185removes the element at slot
186.Fa i .
187It returns the removed element, for user code to dispose of, or
188.Dv NULL
189if the slot was empty.
190.Pp
191.Fn ohash_first
192and
193.Fn ohash_next
194can be used to access all elements in an ohash table, like this:
195.Bd -literal -offset indent
196for (n = ohash_first(h, &i); n != NULL; n = ohash_next(h, &i))
197	do_something_with(n);
198.Ed
199.Pp
200.Fa i
201points to an auxiliary unsigned integer used to record the current position
202in the ohash table.
203Those functions are safe to use even while entries are added to/removed
204from the table, but in such a case they don't guarantee that new entries
205will be returned.
206As a special case, they can safely be used to free elements in the table.
207.Pp
208.Fn ohash_entries
209returns the number of elements in the hash table.
210.Sh STORAGE HANDLING
211Only
212.Fn ohash_init ,
213.Fn ohash_insert ,
214.Fn ohash_remove
215and
216.Fn ohash_delete
217may call the user-supplied memory functions:
218.Bd -literal -offset indent
219p = (*info->calloc)(n, sizeof_record, info->data);
220/* copy data from old to p */
221(*info->free)(old, info->data);
222.Ed
223.Pp
224It is the responsibility of the user memory allocation code to verify
225that those calls did not fail.
226.Pp
227If memory allocation fails,
228.Fn ohash_init
229returns a useless hash table.
230.Fn ohash_insert
231and
232.Fn ohash_remove
233still perform the requested operation, but the returned table should be
234considered read-only.
235It can still be accessed by
236.Fn ohash_lookup* ,
237.Fn ohash_find ,
238.Fn ohash_first
239and
240.Fn ohash_next
241to dump relevant information to disk before aborting.
242.Sh THREAD SAFETY
243The open hashing functions are not thread-safe by design.
244In particular, in a threaded environment, there is no guarantee that a
245.Qq slot
246will not move between a
247.Fn ohash_lookup*
248and a
249.Fn ohash_find ,
250.Fn ohash_insert
251or
252.Fn ohash_remove
253call.
254.Pp
255Multi-threaded applications should explicitly protect ohash table access.
256.Sh SEE ALSO
257.Xr hcreate 3 ,
258.Xr ohash_interval 3
259.Rs
260.%A Donald E. Knuth
261.%B The Art of Computer Programming
262.%V Vol. 3
263.%P pp 506-550
264.%D 1973
265.Re
266.Sh STANDARDS
267Those functions are completely non-standard and should be avoided in
268portable programs.
269.Sh HISTORY
270Those functions were designed and written for
271.Ox
272.Xr make 1
273by Marc Espie in 1999.
274