/src/zlib-ng/deflate_medium.c

Source
/* deflate_medium.c -- The deflate_medium deflate strategy
 *
 * Copyright (C) 2013 Intel Corporation. All rights reserved.
 * Authors:
 *  Arjan van de Ven    <arjan@linux.intel.com>
 *
 * For conditions of distribution and use, see copyright notice in zlib.h
 */
#ifndef NO_MEDIUM_STRATEGY
#include "zbuild.h"
#include "deflate.h"
#include "deflate_p.h"
#include "functable.h"
#include "insert_string_p.h"

struct match {
    uint16_t match_start;
    uint16_t match_length;
    uint16_t strstart;
    uint16_t orgstart;
};

static int emit_match(deflate_state *s, struct match match) {
    int bflush = 0;

    /* matches that are not long enough we need to emit as literals */
    if (match.match_length < WANT_MIN_MATCH) {
        while (match.match_length) {
            bflush += zng_tr_tally_lit(s, s->window[match.strstart]);
            s->lookahead--;
            match.strstart++;
            match.match_length--;
        }
        return bflush;
    }

    check_match(s, match.strstart, match.match_start, match.match_length);

    bflush += zng_tr_tally_dist(s, match.strstart - match.match_start, match.match_length - STD_MIN_MATCH);

    s->lookahead -= match.match_length;
    return bflush;
}

static void insert_match(deflate_state *s, struct match match) {
    if (UNLIKELY(s->lookahead <= (unsigned int)(match.match_length + WANT_MIN_MATCH)))
        return;

    /* matches that are not long enough we need to emit as literals */
    if (LIKELY(match.match_length < WANT_MIN_MATCH)) {
        match.strstart++;
        match.match_length--;
        if (UNLIKELY(match.match_length > 0)) {
            if (match.strstart >= match.orgstart) {
                if (match.strstart + match.match_length - 1 >= match.orgstart) {
                    insert_string(s, match.strstart, match.match_length);
                } else {
                    insert_string(s, match.strstart, match.orgstart - match.strstart + 1);
                }
                match.strstart += match.match_length;
                match.match_length = 0;
            }
        }
        return;
    }

    /* Insert new strings in the hash table only if the match length
     * is not too large. This saves time but degrades compression.
     */
    if (match.match_length <= 16 * s->max_insert_length && s->lookahead >= WANT_MIN_MATCH) {
        match.match_length--; /* string at strstart already in table */
        match.strstart++;

        if (LIKELY(match.strstart >= match.orgstart)) {
            if (LIKELY(match.strstart + match.match_length - 1 >= match.orgstart)) {
                insert_string(s, match.strstart, match.match_length);
            } else {
                insert_string(s, match.strstart, match.orgstart - match.strstart + 1);
            }
        } else if (match.orgstart < match.strstart + match.match_length) {
            insert_string(s, match.orgstart, match.strstart + match.match_length - match.orgstart);
        }
        match.strstart += match.match_length;
        match.match_length = 0;
    } else {
        match.strstart += match.match_length;
        match.match_length = 0;

        if (match.strstart >= (STD_MIN_MATCH - 2))
            quick_insert_string(s, match.strstart + 2 - STD_MIN_MATCH);

        /* If lookahead < WANT_MIN_MATCH, ins_h is garbage, but it does not
         * matter since it will be recomputed at next deflate call.
         */
    }
}

static void fizzle_matches(deflate_state *s, struct match *current, struct match *next) {
    unsigned char *window;
    unsigned char *match, *orig;
    struct match c, n;
    int changed = 0;
    Pos limit;
    /* step zero: sanity checks */

    if (current->match_length <= 1)
        return;

    if (UNLIKELY(current->match_length > 1 + next->match_start))
        return;

    if (UNLIKELY(current->match_length > 1 + next->strstart))
        return;

    window = s->window;

    match = window - current->match_length + 1 + next->match_start;
    orig  = window - current->match_length + 1 + next->strstart;

    /* quick exit check.. if this fails then don't bother with anything else */
    if (LIKELY(*match != *orig))
        return;

    c = *current;
    n = *next;

    /* step one: try to move the "next" match to the left as much as possible */
    limit = next->strstart > MAX_DIST(s) ? next->strstart - (Pos)MAX_DIST(s) : 0;

    match = window + n.match_start - 1;
    orig = window + n.strstart - 1;

    while (*match == *orig) {
        if (UNLIKELY(c.match_length < 1))
            break;
        if (UNLIKELY(n.strstart <= limit))
            break;
        if (UNLIKELY(n.match_length >= 256))
            break;
        if (UNLIKELY(n.match_start <= 1))
            break;

        n.strstart--;
        n.match_start--;
        n.match_length++;
        c.match_length--;
        match--;
        orig--;
        changed++;
    }

    if (!changed)
        return;

    if (c.match_length <= 1 && n.match_length != 2) {
        n.orgstart++;
        *current = c;
        *next = n;
    } else {
        return;
    }
}

Z_INTERNAL block_state deflate_medium(deflate_state *s, int flush) {
    /* Align the first struct to start on a new cacheline, this allows us to fit both structs in one cacheline */
    ALIGNED_(16) struct match current_match = {0};
                 struct match next_match = {0};

    /* For levels below 5, don't check the next position for a better match */
    int early_exit = s->level < 5;

    for (;;) {
        uint32_t hash_head = 0;    /* head of the hash chain */
        int bflush = 0;       /* set if current block must be flushed */
        int64_t dist;

        /* Make sure that we always have enough lookahead, except
         * at the end of the input file. We need STD_MAX_MATCH bytes
         * for the next match, plus WANT_MIN_MATCH bytes to insert the
         * string following the next current_match.
         */
        if (s->lookahead < MIN_LOOKAHEAD) {
            PREFIX(fill_window)(s);
            if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
                return need_more;
            }
            if (UNLIKELY(s->lookahead == 0))
                break; /* flush the current block */
            next_match.match_length = 0;
        }

        /* Insert the string window[strstart .. strstart+2] in the
         * dictionary, and set hash_head to the head of the hash chain:
         */

        /* If we already have a future match from a previous round, just use that */
        if (!early_exit && next_match.match_length > 0) {
            current_match = next_match;
            next_match.match_length = 0;
        } else {
            hash_head = 0;
            if (s->lookahead >= WANT_MIN_MATCH) {
                hash_head = quick_insert_string(s, s->strstart);
            }

            current_match.strstart = (uint16_t)s->strstart;
            current_match.orgstart = current_match.strstart;

            /* Find the longest match, discarding those <= prev_length.
             * At this point we have always match_length < WANT_MIN_MATCH
             */

            dist = (int64_t)s->strstart - hash_head;
            if (dist <= MAX_DIST(s) && dist > 0 && hash_head != 0) {
                /* To simplify the code, we prevent matches with the string
                 * of window index 0 (in particular we have to avoid a match
                 * of the string with itself at the start of the input file).
                 */
                current_match.match_length = (uint16_t)FUNCTABLE_CALL(longest_match)(s, hash_head);
                current_match.match_start = (uint16_t)s->match_start;
                if (UNLIKELY(current_match.match_length < WANT_MIN_MATCH))
                    current_match.match_length = 1;
                if (UNLIKELY(current_match.match_start >= current_match.strstart)) {
                    /* this can happen due to some restarts */
                    current_match.match_length = 1;
                }
            } else {
                /* Set up the match to be a 1 byte literal */
                current_match.match_start = 0;
                current_match.match_length = 1;
            }
        }

        insert_match(s, current_match);

        /* now, look ahead one */
        if (LIKELY(!early_exit && s->lookahead > MIN_LOOKAHEAD && (uint32_t)(current_match.strstart + current_match.match_length) < (s->window_size - MIN_LOOKAHEAD))) {
            s->strstart = current_match.strstart + current_match.match_length;
            hash_head = quick_insert_string(s, s->strstart);

            next_match.strstart = (uint16_t)s->strstart;
            next_match.orgstart = next_match.strstart;

            /* Find the longest match, discarding those <= prev_length.
             * At this point we have always match_length < WANT_MIN_MATCH
             */

            dist = (int64_t)s->strstart - hash_head;
            if (dist <= MAX_DIST(s) && dist > 0 && hash_head != 0) {
                /* To simplify the code, we prevent matches with the string
                 * of window index 0 (in particular we have to avoid a match
                 * of the string with itself at the start of the input file).
                 */
                next_match.match_length = (uint16_t)FUNCTABLE_CALL(longest_match)(s, hash_head);
                next_match.match_start = (uint16_t)s->match_start;
                if (UNLIKELY(next_match.match_start >= next_match.strstart)) {
                    /* this can happen due to some restarts */
                    next_match.match_length = 1;
                }
                if (next_match.match_length < WANT_MIN_MATCH)
                    next_match.match_length = 1;
                else
                    fizzle_matches(s, &current_match, &next_match);
            } else {
                /* Set up the match to be a 1 byte literal */
                next_match.match_start = 0;
                next_match.match_length = 1;
            }

            s->strstart = current_match.strstart;
        } else {
            next_match.match_length = 0;
        }

        /* now emit the current match */
        bflush = emit_match(s, current_match);

        /* move the "cursor" forward */
        s->strstart += current_match.match_length;

        if (UNLIKELY(bflush))
            FLUSH_BLOCK(s, 0);
    }
    s->insert = s->strstart < (STD_MIN_MATCH - 1) ? s->strstart : (STD_MIN_MATCH - 1);
    if (flush == Z_FINISH) {
        FLUSH_BLOCK(s, 1);
        return finish_done;
    }
    if (UNLIKELY(s->sym_next))
        FLUSH_BLOCK(s, 0);

    return block_done;
}
#endif

Coverage Report

Created: 2026-02-26 06:53

Line	Count	Source
1		/* deflate_medium.c -- The deflate_medium deflate strategy
2		*
3		* Copyright (C) 2013 Intel Corporation. All rights reserved.
4		* Authors:
5		* Arjan van de Ven <arjan@linux.intel.com>
6		*
7		* For conditions of distribution and use, see copyright notice in zlib.h
8		*/
9		#ifndef NO_MEDIUM_STRATEGY
10		#include "zbuild.h"
11		#include "deflate.h"
12		#include "deflate_p.h"
13		#include "functable.h"
14		#include "insert_string_p.h"
15
16		struct match {
17		uint16_t match_start;
18		uint16_t match_length;
19		uint16_t strstart;
20		uint16_t orgstart;
21		};
22
23	43.0M	static int emit_match(deflate_state *s, struct match match) {
24	43.0M	int bflush = 0;
25
26		/* matches that are not long enough we need to emit as literals */
27	43.0M	if (match.match_length < WANT_MIN_MATCH) {
28	80.8M	while (match.match_length) {
29	40.4M	bflush += zng_tr_tally_lit(s, s->window[match.strstart]);
30	40.4M	s->lookahead--;
31	40.4M	match.strstart++;
32	40.4M	match.match_length--;
33	40.4M	}
34	40.4M	return bflush;
35	40.4M	}
36
37	2.59M	check_match(s, match.strstart, match.match_start, match.match_length);
38
39	2.59M	bflush += zng_tr_tally_dist(s, match.strstart - match.match_start, match.match_length - STD_MIN_MATCH);
40
41	2.59M	s->lookahead -= match.match_length;
42	2.59M	return bflush;
43	43.0M	}
44
45	43.0M	static void insert_match(deflate_state *s, struct match match) {
46	43.0M	if (UNLIKELY(s->lookahead <= (unsigned int)(match.match_length + WANT_MIN_MATCH)))
47	4.63k	return;
48
49		/* matches that are not long enough we need to emit as literals */
50	43.0M	if (LIKELY(match.match_length < WANT_MIN_MATCH)) {
51	40.4M	match.strstart++;
52	40.4M	match.match_length--;
53	40.4M	if (UNLIKELY(match.match_length > 0)) {
54	0	if (match.strstart >= match.orgstart) {
55	0	if (match.strstart + match.match_length - 1 >= match.orgstart) {
56	0	insert_string(s, match.strstart, match.match_length);
57	0	} else {
58	0	insert_string(s, match.strstart, match.orgstart - match.strstart + 1);
59	0	}
60	0	match.strstart += match.match_length;
61	0	match.match_length = 0;
62	0	}
63	0	}
64	40.4M	return;
65	40.4M	}
66
67		/* Insert new strings in the hash table only if the match length
68		* is not too large. This saves time but degrades compression.
69		*/
70	2.61M	if (match.match_length <= 16 * s->max_insert_length && s->lookahead >= WANT_MIN_MATCH) {
71	2.55M	match.match_length--; /* string at strstart already in table */
72	2.55M	match.strstart++;
73
74	2.55M	if (LIKELY(match.strstart >= match.orgstart)) {
75	2.52M	if (LIKELY(match.strstart + match.match_length - 1 >= match.orgstart)) {
76	2.52M	insert_string(s, match.strstart, match.match_length);
77	2.52M	} else {
78	0	insert_string(s, match.strstart, match.orgstart - match.strstart + 1);
79	0	}
80	2.52M	} else if (match.orgstart < match.strstart + match.match_length) {
81	25.3k	insert_string(s, match.orgstart, match.strstart + match.match_length - match.orgstart);
82	25.3k	}
83	2.55M	match.strstart += match.match_length;
84	2.55M	match.match_length = 0;
85	2.55M	} else {
86	61.2k	match.strstart += match.match_length;
87	61.2k	match.match_length = 0;
88
89	61.2k	if (match.strstart >= (STD_MIN_MATCH - 2))
90	61.2k	quick_insert_string(s, match.strstart + 2 - STD_MIN_MATCH);
91
92		/* If lookahead < WANT_MIN_MATCH, ins_h is garbage, but it does not
93		* matter since it will be recomputed at next deflate call.
94		*/
95	61.2k	}
96	2.61M	}
97
98	801k	static void fizzle_matches(deflate_state s, struct match current, struct match *next) {
99	801k	unsigned char *window;
100	801k	unsigned char match, orig;
101	801k	struct match c, n;
102	801k	int changed = 0;
103	801k	Pos limit;
104		/* step zero: sanity checks */
105
106	801k	if (current->match_length <= 1)
107	467k	return;
108
109	334k	if (UNLIKELY(current->match_length > 1 + next->match_start))
110	278	return;
111
112	333k	if (UNLIKELY(current->match_length > 1 + next->strstart))
113	0	return;
114
115	333k	window = s->window;
116
117	333k	match = window - current->match_length + 1 + next->match_start;
118	333k	orig = window - current->match_length + 1 + next->strstart;
119
120		/* quick exit check.. if this fails then don't bother with anything else */
121	333k	if (LIKELY(match != orig))
122	185k	return;
123
124	148k	c = *current;
125	148k	n = *next;
126
127		/* step one: try to move the "next" match to the left as much as possible */
128	148k	limit = next->strstart > MAX_DIST(s) ? next->strstart - (Pos)MAX_DIST(s) : 0;
129
130	148k	match = window + n.match_start - 1;
131	148k	orig = window + n.strstart - 1;
132
133	1.84M	while (match == orig) {
134	1.73M	if (UNLIKELY(c.match_length < 1))
135	3.85k	break;
136	1.73M	if (UNLIKELY(n.strstart <= limit))
137	0	break;
138	1.73M	if (UNLIKELY(n.match_length >= 256))
139	37.7k	break;
140	1.69M	if (UNLIKELY(n.match_start <= 1))
141	34	break;
142
143	1.69M	n.strstart--;
144	1.69M	n.match_start--;
145	1.69M	n.match_length++;
146	1.69M	c.match_length--;
147	1.69M	match--;
148	1.69M	orig--;
149	1.69M	changed++;
150	1.69M	}
151
152	148k	if (!changed)
153	86.3k	return;
154
155	62.1k	if (c.match_length <= 1 && n.match_length != 2) {
156	25.3k	n.orgstart++;
157	25.3k	*current = c;
158	25.3k	*next = n;
159	36.7k	} else {
160	36.7k	return;
161	36.7k	}
162	62.1k	}
163
164	1.96k	Z_INTERNAL block_state deflate_medium(deflate_state *s, int flush) {
165		/* Align the first struct to start on a new cacheline, this allows us to fit both structs in one cacheline */
166	1.96k	ALIGNED_(16) struct match current_match = {0};
167	1.96k	struct match next_match = {0};
168
169		/* For levels below 5, don't check the next position for a better match */
170	1.96k	int early_exit = s->level < 5;
171
172	43.0M	for (;;) {
173	43.0M	uint32_t hash_head = 0; /* head of the hash chain */
174	43.0M	int bflush = 0; /* set if current block must be flushed */
175	43.0M	int64_t dist;
176
177		/* Make sure that we always have enough lookahead, except
178		* at the end of the input file. We need STD_MAX_MATCH bytes
179		* for the next match, plus WANT_MIN_MATCH bytes to insert the
180		* string following the next current_match.
181		*/
182	43.0M	if (s->lookahead < MIN_LOOKAHEAD) {
183	87.3k	PREFIX(fill_window)(s);
184	87.3k	if (s->lookahead < MIN_LOOKAHEAD && flush == Z_NO_FLUSH) {
185	437	return need_more;
186	437	}
187	86.9k	if (UNLIKELY(s->lookahead == 0))
188	1.31k	break; /* flush the current block */
189	85.6k	next_match.match_length = 0;
190	85.6k	}
191
192		/* Insert the string window[strstart .. strstart+2] in the
193		* dictionary, and set hash_head to the head of the hash chain:
194		*/
195
196		/* If we already have a future match from a previous round, just use that */
197	43.0M	if (!early_exit && next_match.match_length > 0) {
198	18.4M	current_match = next_match;
199	18.4M	next_match.match_length = 0;
200	24.5M	} else {
201	24.5M	hash_head = 0;
202	24.5M	if (s->lookahead >= WANT_MIN_MATCH) {
203	24.5M	hash_head = quick_insert_string(s, s->strstart);
204	24.5M	}
205
206	24.5M	current_match.strstart = (uint16_t)s->strstart;
207	24.5M	current_match.orgstart = current_match.strstart;
208
209		/* Find the longest match, discarding those <= prev_length.
210		* At this point we have always match_length < WANT_MIN_MATCH
211		*/
212
213	24.5M	dist = (int64_t)s->strstart - hash_head;
214	24.5M	if (dist <= MAX_DIST(s) && dist > 0 && hash_head != 0) {
215		/* To simplify the code, we prevent matches with the string
216		* of window index 0 (in particular we have to avoid a match
217		* of the string with itself at the start of the input file).
218		*/
219	9.14M	current_match.match_length = (uint16_t)FUNCTABLE_CALL(longest_match)(s, hash_head);
220	9.14M	current_match.match_start = (uint16_t)s->match_start;
221	9.14M	if (UNLIKELY(current_match.match_length < WANT_MIN_MATCH))
222	7.33M	current_match.match_length = 1;
223	9.14M	if (UNLIKELY(current_match.match_start >= current_match.strstart)) {
224		/* this can happen due to some restarts */
225	228	current_match.match_length = 1;
226	228	}
227	15.4M	} else {
228		/* Set up the match to be a 1 byte literal */
229	15.4M	current_match.match_start = 0;
230	15.4M	current_match.match_length = 1;
231	15.4M	}
232	24.5M	}
233
234	43.0M	insert_match(s, current_match);
235
236		/* now, look ahead one */
237	43.0M	if (LIKELY(!early_exit && s->lookahead > MIN_LOOKAHEAD && (uint32_t)(current_match.strstart + current_match.match_length) < (s->window_size - MIN_LOOKAHEAD))) {
238	18.4M	s->strstart = current_match.strstart + current_match.match_length;
239	18.4M	hash_head = quick_insert_string(s, s->strstart);
240
241	18.4M	next_match.strstart = (uint16_t)s->strstart;
242	18.4M	next_match.orgstart = next_match.strstart;
243
244		/* Find the longest match, discarding those <= prev_length.
245		* At this point we have always match_length < WANT_MIN_MATCH
246		*/
247
248	18.4M	dist = (int64_t)s->strstart - hash_head;
249	18.4M	if (dist <= MAX_DIST(s) && dist > 0 && hash_head != 0) {
250		/* To simplify the code, we prevent matches with the string
251		* of window index 0 (in particular we have to avoid a match
252		* of the string with itself at the start of the input file).
253		*/
254	5.82M	next_match.match_length = (uint16_t)FUNCTABLE_CALL(longest_match)(s, hash_head);
255	5.82M	next_match.match_start = (uint16_t)s->match_start;
256	5.82M	if (UNLIKELY(next_match.match_start >= next_match.strstart)) {
257		/* this can happen due to some restarts */
258	195	next_match.match_length = 1;
259	195	}
260	5.82M	if (next_match.match_length < WANT_MIN_MATCH)
261	5.02M	next_match.match_length = 1;
262	801k	else
263	801k	fizzle_matches(s, &current_match, &next_match);
264	12.6M	} else {
265		/* Set up the match to be a 1 byte literal */
266	12.6M	next_match.match_start = 0;
267	12.6M	next_match.match_length = 1;
268	12.6M	}
269
270	18.4M	s->strstart = current_match.strstart;
271	24.5M	} else {
272	24.5M	next_match.match_length = 0;
273	24.5M	}
274
275		/* now emit the current match */
276	43.0M	bflush = emit_match(s, current_match);
277
278		/* move the "cursor" forward */
279	43.0M	s->strstart += current_match.match_length;
280
281	43.0M	if (UNLIKELY(bflush))
282	43.0M	FLUSH_BLOCK(s, 0);
283	43.0M	}
284	1.31k	s->insert = s->strstart < (STD_MIN_MATCH - 1) ? s->strstart : (STD_MIN_MATCH - 1);
285	1.31k	if (flush == Z_FINISH) {
286	1.31k	FLUSH_BLOCK(s, 1);
287	1.25k	return finish_done;
288	1.31k	}
289	0	if (UNLIKELY(s->sym_next))
290	0	FLUSH_BLOCK(s, 0);
291
292	0	return block_done;
293	0	}
294		#endif