ewah: support platforms that require aligned reads
[git] / ewah / ewah_io.c
1 /**
2  * Copyright 2013, GitHub, Inc
3  * Copyright 2009-2013, Daniel Lemire, Cliff Moon,
4  *      David McIntosh, Robert Becho, Google Inc. and Veronika Zenz
5  *
6  * This program is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU General Public License
8  * as published by the Free Software Foundation; either version 2
9  * of the License, or (at your option) any later version.
10  *
11  * This program is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
14  * GNU General Public License for more details.
15  *
16  * You should have received a copy of the GNU General Public License
17  * along with this program; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
19  */
20 #include "git-compat-util.h"
21 #include "ewok.h"
22
23 int ewah_serialize_native(struct ewah_bitmap *self, int fd)
24 {
25         uint32_t write32;
26         size_t to_write = self->buffer_size * 8;
27
28         /* 32 bit -- bit size for the map */
29         write32 = (uint32_t)self->bit_size;
30         if (write(fd, &write32, 4) != 4)
31                 return -1;
32
33         /** 32 bit -- number of compressed 64-bit words */
34         write32 = (uint32_t)self->buffer_size;
35         if (write(fd, &write32, 4) != 4)
36                 return -1;
37
38         if (write(fd, self->buffer, to_write) != to_write)
39                 return -1;
40
41         /** 32 bit -- position for the RLW */
42         write32 = self->rlw - self->buffer;
43         if (write(fd, &write32, 4) != 4)
44                 return -1;
45
46         return (3 * 4) + to_write;
47 }
48
49 int ewah_serialize_to(struct ewah_bitmap *self,
50                       int (*write_fun)(void *, const void *, size_t),
51                       void *data)
52 {
53         size_t i;
54         eword_t dump[2048];
55         const size_t words_per_dump = sizeof(dump) / sizeof(eword_t);
56         uint32_t bitsize, word_count, rlw_pos;
57
58         const eword_t *buffer;
59         size_t words_left;
60
61         /* 32 bit -- bit size for the map */
62         bitsize =  htonl((uint32_t)self->bit_size);
63         if (write_fun(data, &bitsize, 4) != 4)
64                 return -1;
65
66         /** 32 bit -- number of compressed 64-bit words */
67         word_count =  htonl((uint32_t)self->buffer_size);
68         if (write_fun(data, &word_count, 4) != 4)
69                 return -1;
70
71         /** 64 bit x N -- compressed words */
72         buffer = self->buffer;
73         words_left = self->buffer_size;
74
75         while (words_left >= words_per_dump) {
76                 for (i = 0; i < words_per_dump; ++i, ++buffer)
77                         dump[i] = htonll(*buffer);
78
79                 if (write_fun(data, dump, sizeof(dump)) != sizeof(dump))
80                         return -1;
81
82                 words_left -= words_per_dump;
83         }
84
85         if (words_left) {
86                 for (i = 0; i < words_left; ++i, ++buffer)
87                         dump[i] = htonll(*buffer);
88
89                 if (write_fun(data, dump, words_left * 8) != words_left * 8)
90                         return -1;
91         }
92
93         /** 32 bit -- position for the RLW */
94         rlw_pos = (uint8_t*)self->rlw - (uint8_t *)self->buffer;
95         rlw_pos = htonl(rlw_pos / sizeof(eword_t));
96
97         if (write_fun(data, &rlw_pos, 4) != 4)
98                 return -1;
99
100         return (3 * 4) + (self->buffer_size * 8);
101 }
102
103 static int write_helper(void *fd, const void *buf, size_t len)
104 {
105         return write((intptr_t)fd, buf, len);
106 }
107
108 int ewah_serialize(struct ewah_bitmap *self, int fd)
109 {
110         return ewah_serialize_to(self, write_helper, (void *)(intptr_t)fd);
111 }
112
113 int ewah_read_mmap(struct ewah_bitmap *self, void *map, size_t len)
114 {
115         uint8_t *ptr = map;
116
117         self->bit_size = get_be32(ptr);
118         ptr += sizeof(uint32_t);
119
120         self->buffer_size = self->alloc_size = get_be32(ptr);
121         ptr += sizeof(uint32_t);
122
123         self->buffer = ewah_realloc(self->buffer,
124                 self->alloc_size * sizeof(eword_t));
125
126         if (!self->buffer)
127                 return -1;
128
129         /*
130          * Copy the raw data for the bitmap as a whole chunk;
131          * if we're in a little-endian platform, we'll perform
132          * the endianness conversion in a separate pass to ensure
133          * we're loading 8-byte aligned words.
134          */
135         memcpy(self->buffer, ptr, self->buffer_size * sizeof(uint64_t));
136         ptr += self->buffer_size * sizeof(uint64_t);
137
138 #if __BYTE_ORDER != __BIG_ENDIAN
139         {
140                 size_t i;
141                 for (i = 0; i < self->buffer_size; ++i)
142                         self->buffer[i] = ntohll(self->buffer[i]);
143         }
144 #endif
145
146         self->rlw = self->buffer + get_be32(ptr);
147
148         return (3 * 4) + (self->buffer_size * 8);
149 }
150
151 int ewah_deserialize(struct ewah_bitmap *self, int fd)
152 {
153         size_t i;
154         eword_t dump[2048];
155         const size_t words_per_dump = sizeof(dump) / sizeof(eword_t);
156         uint32_t bitsize, word_count, rlw_pos;
157
158         eword_t *buffer = NULL;
159         size_t words_left;
160
161         ewah_clear(self);
162
163         /* 32 bit -- bit size for the map */
164         if (read(fd, &bitsize, 4) != 4)
165                 return -1;
166
167         self->bit_size = (size_t)ntohl(bitsize);
168
169         /** 32 bit -- number of compressed 64-bit words */
170         if (read(fd, &word_count, 4) != 4)
171                 return -1;
172
173         self->buffer_size = self->alloc_size = (size_t)ntohl(word_count);
174         self->buffer = ewah_realloc(self->buffer,
175                 self->alloc_size * sizeof(eword_t));
176
177         if (!self->buffer)
178                 return -1;
179
180         /** 64 bit x N -- compressed words */
181         buffer = self->buffer;
182         words_left = self->buffer_size;
183
184         while (words_left >= words_per_dump) {
185                 if (read(fd, dump, sizeof(dump)) != sizeof(dump))
186                         return -1;
187
188                 for (i = 0; i < words_per_dump; ++i, ++buffer)
189                         *buffer = ntohll(dump[i]);
190
191                 words_left -= words_per_dump;
192         }
193
194         if (words_left) {
195                 if (read(fd, dump, words_left * 8) != words_left * 8)
196                         return -1;
197
198                 for (i = 0; i < words_left; ++i, ++buffer)
199                         *buffer = ntohll(dump[i]);
200         }
201
202         /** 32 bit -- position for the RLW */
203         if (read(fd, &rlw_pos, 4) != 4)
204                 return -1;
205
206         self->rlw = self->buffer + ntohl(rlw_pos);
207         return 0;
208 }