File: | src/repopage.c |
Warning: | line 402, column 7 Value stored to 'litofs' is never read |
1 | /* |
2 | * Copyright (c) 2007-2012, Novell Inc. |
3 | * |
4 | * This program is licensed under the BSD license, read LICENSE.BSD |
5 | * for further information |
6 | */ |
7 | |
8 | /* |
9 | * repopage.c |
10 | * |
11 | * Paging and compression functions for the vertical repository data. |
12 | * Vertical data is grouped by key, normal data is grouped by solvable. |
13 | * This makes searching for a string in vertical data fast as there's |
14 | * no need to skip over data if keys we're not interested in. |
15 | * |
16 | * The vertical data is split into pages, each page is compressed with a fast |
17 | * compression algorithm. These pages are read in on demand, not recently used |
18 | * pages automatically get dropped. |
19 | */ |
20 | |
21 | #define _XOPEN_SOURCE500 500 |
22 | |
23 | #include <sys/types.h> |
24 | #include <stdint.h> |
25 | #include <stdio.h> |
26 | #include <stdlib.h> |
27 | #include <string.h> |
28 | #include <unistd.h> |
29 | #include <assert.h> |
30 | #include <fcntl.h> |
31 | #include <time.h> |
32 | |
33 | #include "repo.h" |
34 | #include "repopage.h" |
35 | |
36 | |
37 | |
38 | #define BLOCK_SIZE(65536*1) (65536*1) |
39 | #if BLOCK_SIZE(65536*1) <= 65536 |
40 | typedef uint16_t Ref; |
41 | #else |
42 | typedef uint32_t Ref; |
43 | #endif |
44 | |
45 | /* |
46 | The format is tailored for fast decompression (i.e. only byte based), |
47 | and skewed to ASCII content (highest bit often not set): |
48 | |
49 | a 0LLLLLLL |
50 | - self-describing ASCII character hex L |
51 | b 100lllll <l+1 bytes> |
52 | - literal run of length l+1 |
53 | c 101oolll <8o> |
54 | - back ref of length l+2, at offset -(o+1) (o < 1 << 10) |
55 | d 110lllll <8o> |
56 | - back ref of length l+2+8, at offset -(o+1) (o < 1 << 8) |
57 | e 1110llll <8o> <8o> |
58 | - back ref of length l+3, at offset -(o+1) (o < 1 << 16) |
59 | f1 1111llll <8l> <8o> <8o> |
60 | - back ref, length l+19 (l < 1<<12), offset -(o+1) (o < 1<<16) |
61 | f2 11110lll <8l> <8o> <8o> |
62 | - back ref, length l+19 (l < 1<<11), offset -(o+1) (o < 1<<16) |
63 | g 11111lll <8l> <8o> <8o> <8o> |
64 | - back ref, length l+5 (l < 1<<11), offset -(o+1) (o < 1<<24) |
65 | |
66 | Generally for a literal of length L we need L+1 bytes, hence it is |
67 | better to encode also very short backrefs (2 chars) as backrefs if |
68 | their offset is small, as that only needs two bytes. Except if we |
69 | already have a literal run, in that case it's better to append there, |
70 | instead of breaking it for a backref. So given a potential backref |
71 | at offset O, length L the strategy is as follows: |
72 | |
73 | L < 2 : encode as 1-literal |
74 | L == 2, O > 1024 : encode as 1-literal |
75 | L == 2, have already literals: encode as 1-literal |
76 | O = O - 1 |
77 | L >= 2, L <= 9, O < 1024 : encode as c |
78 | L >= 10, L <= 41, O < 256 : encode as d |
79 | else we have either O >= 1024, or L >= 42: |
80 | L < 3 : encode as 1-literal |
81 | L >= 3, L <= 18, O < 65536 : encode as e |
82 | L >= 19, L <= 4095+18, O < 65536 : encode as f |
83 | else we have either L >= 4096+18 or O >= 65536. |
84 | O >= 65536: encode as 1-literal, too bad |
85 | (with the current block size this can't happen) |
86 | L >= 4096+18, so reduce to 4095+18 : encode as f |
87 | */ |
88 | |
89 | |
90 | static unsigned int |
91 | compress_buf(const unsigned char *in, unsigned int in_len, |
92 | unsigned char *out, unsigned int out_len) |
93 | { |
94 | unsigned int oo = 0; /* out-offset */ |
95 | unsigned int io = 0; /* in-offset */ |
96 | #define HS(65536) (65536) |
97 | Ref htab[HS(65536)]; |
98 | Ref hnext[BLOCK_SIZE(65536*1)]; |
99 | unsigned int litofs = 0; |
100 | memset(htab, -1, sizeof (htab)); |
101 | memset(hnext, -1, sizeof (hnext)); |
102 | while (io + 2 < in_len) |
103 | { |
104 | /* Search for a match of the string starting at IN, we have at |
105 | least three characters. */ |
106 | unsigned int hval = in[io] | in[io + 1] << 8 | in[io + 2] << 16; |
107 | unsigned int try, mlen, mofs, tries; |
108 | hval = (hval ^ (hval << 5) ^ (hval >> 5)) - hval * 5; |
109 | hval = hval & (HS(65536) - 1); |
110 | try = htab[hval]; |
111 | hnext[io] = htab[hval]; |
112 | htab[hval] = io; |
113 | mlen = 0; |
114 | mofs = 0; |
115 | |
116 | for (tries = 0; try != -1 && tries < 12; tries++) |
117 | { |
118 | if (try < io |
119 | && in[try] == in[io] && in[try + 1] == in[io + 1]) |
120 | { |
121 | mlen = 2; |
122 | mofs = (io - try) - 1; |
123 | break; |
124 | } |
125 | try = hnext[try]; |
126 | } |
127 | for (; try != -1 && tries < 12; tries++) |
128 | { |
129 | /* assert(mlen >= 2); */ |
130 | /* assert(io + mlen < in_len); */ |
131 | /* Try a match starting from [io] with the strings at [try]. |
132 | That's only sensible if TRY actually is before IO (can happen |
133 | with uninit hash table). If we have a previous match already |
134 | we're only going to take the new one if it's longer, hence |
135 | check the potentially last character. */ |
136 | if (try < io && in[try + mlen] == in[io + mlen]) |
137 | { |
138 | unsigned int this_len, this_ofs; |
139 | if (memcmp(in + try, in + io, mlen)) |
140 | goto no_match; |
141 | this_len = mlen + 1; |
142 | /* Now try extending the match by more characters. */ |
143 | for (; |
144 | io + this_len < in_len |
145 | && in[try + this_len] == in[io + this_len]; this_len++) |
146 | ; |
147 | #if 0 |
148 | unsigned int testi; |
149 | for (testi = 0; testi < this_len; testi++) |
150 | assert(in[try + testi] == in[io + testi])({ if (in[try + testi] == in[io + testi]) ; else __assert_fail ("in[try + testi] == in[io + testi]", "/home/brain/Projects/upstream/libsolv/src/repopage.c" , 150, __PRETTY_FUNCTION__); }); |
151 | #endif |
152 | this_ofs = (io - try) - 1; |
153 | /*if (this_ofs > 65535) |
154 | goto no_match; */ |
155 | #if 0 |
156 | assert(this_len >= 2)({ if (this_len >= 2) ; else __assert_fail ("this_len >= 2" , "/home/brain/Projects/upstream/libsolv/src/repopage.c", 156 , __PRETTY_FUNCTION__); }); |
157 | assert(this_len >= mlen)({ if (this_len >= mlen) ; else __assert_fail ("this_len >= mlen" , "/home/brain/Projects/upstream/libsolv/src/repopage.c", 157 , __PRETTY_FUNCTION__); }); |
158 | assert(this_len > mlen || (this_len == mlen && this_ofs > mofs))({ if (this_len > mlen || (this_len == mlen && this_ofs > mofs)) ; else __assert_fail ("this_len > mlen || (this_len == mlen && this_ofs > mofs)" , "/home/brain/Projects/upstream/libsolv/src/repopage.c", 158 , __PRETTY_FUNCTION__); }); |
159 | #endif |
160 | mlen = this_len, mofs = this_ofs; |
161 | /* If our match extends up to the end of input, no next |
162 | match can become better. This is not just an |
163 | optimization, it establishes a loop invariant |
164 | (io + mlen < in_len). */ |
165 | if (io + mlen >= in_len) |
166 | goto match_done; |
167 | } |
168 | no_match: |
169 | try = hnext[try]; |
170 | /*if (io - try - 1 >= 65536) |
171 | break;*/ |
172 | } |
173 | |
174 | match_done: |
175 | if (mlen) |
176 | { |
177 | /*fprintf(stderr, "%d %d\n", mlen, mofs);*/ |
178 | if (mlen == 2 && (litofs || mofs >= 1024)) |
179 | mlen = 0; |
180 | /*else if (mofs >= 65536) |
181 | mlen = 0;*/ |
182 | else if (mofs >= 65536) |
183 | { |
184 | if (mlen >= 2048 + 5) |
185 | mlen = 2047 + 5; |
186 | else if (mlen < 5) |
187 | mlen = 0; |
188 | } |
189 | else if (mlen < 3) |
190 | mlen = 0; |
191 | /*else if (mlen >= 4096 + 19) |
192 | mlen = 4095 + 19;*/ |
193 | else if (mlen >= 2048 + 19) |
194 | mlen = 2047 + 19; |
195 | /* Skip this match if the next character would deliver a better one, |
196 | but only do this if we have the chance to really extend the |
197 | length (i.e. our current length isn't yet the (conservative) |
198 | maximum). */ |
199 | if (mlen && mlen < (2048 + 5) && io + 3 < in_len) |
200 | { |
201 | unsigned int hval = |
202 | in[io + 1] | in[io + 2] << 8 | in[io + 3] << 16; |
203 | unsigned int try; |
204 | hval = (hval ^ (hval << 5) ^ (hval >> 5)) - hval * 5; |
205 | hval = hval & (HS(65536) - 1); |
206 | try = htab[hval]; |
207 | if (try < io + 1 |
208 | && in[try] == in[io + 1] && in[try + 1] == in[io + 2]) |
209 | { |
210 | unsigned int this_len; |
211 | this_len = 2; |
212 | for (; |
213 | io + 1 + this_len < in_len |
214 | && in[try + this_len] == in[io + 1 + this_len]; |
215 | this_len++) |
216 | ; |
217 | if (this_len >= mlen) |
218 | mlen = 0; |
219 | } |
220 | } |
221 | } |
222 | if (!mlen) |
223 | { |
224 | if (!litofs) |
225 | litofs = io + 1; |
226 | io++; |
227 | } |
228 | else |
229 | { |
230 | if (litofs) |
231 | { |
232 | unsigned litlen; |
233 | litofs--; |
234 | litlen = io - litofs; |
235 | /* fprintf(stderr, "lit: %d\n", litlen); */ |
236 | while (litlen) |
237 | { |
238 | unsigned int easy_sz; |
239 | /* Emit everything we can as self-describers. As soon as |
240 | we hit a byte we can't emit as such we're going to emit |
241 | a length descriptor anyway, so we can as well include |
242 | bytes < 0x80 which might follow afterwards in that run. */ |
243 | for (easy_sz = 0; |
244 | easy_sz < litlen && in[litofs + easy_sz] < 0x80; |
245 | easy_sz++) |
246 | ; |
247 | if (easy_sz) |
248 | { |
249 | if (oo + easy_sz >= out_len) |
250 | return 0; |
251 | memcpy(out + oo, in + litofs, easy_sz); |
252 | litofs += easy_sz; |
253 | oo += easy_sz; |
254 | litlen -= easy_sz; |
255 | if (!litlen) |
256 | break; |
257 | } |
258 | if (litlen <= 32) |
259 | { |
260 | if (oo + 1 + litlen >= out_len) |
261 | return 0; |
262 | out[oo++] = 0x80 | (litlen - 1); |
263 | while (litlen--) |
264 | out[oo++] = in[litofs++]; |
265 | break; |
266 | } |
267 | else |
268 | { |
269 | /* Literal length > 32, so chunk it. */ |
270 | if (oo + 1 + 32 >= out_len) |
271 | return 0; |
272 | out[oo++] = 0x80 | 31; |
273 | memcpy(out + oo, in + litofs, 32); |
274 | oo += 32; |
275 | litofs += 32; |
276 | litlen -= 32; |
277 | } |
278 | } |
279 | litofs = 0; |
280 | } |
281 | |
282 | /* fprintf(stderr, "ref: %d @ %d\n", mlen, mofs); */ |
283 | |
284 | if (mlen >= 2 && mlen <= 9 && mofs < 1024) |
285 | { |
286 | if (oo + 2 >= out_len) |
287 | return 0; |
288 | out[oo++] = 0xa0 | ((mofs & 0x300) >> 5) | (mlen - 2); |
289 | out[oo++] = mofs & 0xff; |
290 | } |
291 | else if (mlen >= 10 && mlen <= 41 && mofs < 256) |
292 | { |
293 | if (oo + 2 >= out_len) |
294 | return 0; |
295 | out[oo++] = 0xc0 | (mlen - 10); |
296 | out[oo++] = mofs; |
297 | } |
298 | else if (mofs >= 65536) |
299 | { |
300 | assert(mlen >= 5 && mlen < 2048 + 5)({ if (mlen >= 5 && mlen < 2048 + 5) ; else __assert_fail ("mlen >= 5 && mlen < 2048 + 5", "/home/brain/Projects/upstream/libsolv/src/repopage.c" , 300, __PRETTY_FUNCTION__); }); |
301 | if (oo + 5 >= out_len) |
302 | return 0; |
303 | out[oo++] = 0xf8 | ((mlen - 5) >> 8); |
304 | out[oo++] = (mlen - 5) & 0xff; |
305 | out[oo++] = mofs & 0xff; |
306 | out[oo++] = (mofs >> 8) & 0xff; |
307 | out[oo++] = mofs >> 16; |
308 | } |
309 | else if (mlen >= 3 && mlen <= 18) |
310 | { |
311 | assert(mofs < 65536)({ if (mofs < 65536) ; else __assert_fail ("mofs < 65536" , "/home/brain/Projects/upstream/libsolv/src/repopage.c", 311 , __PRETTY_FUNCTION__); }); |
312 | if (oo + 3 >= out_len) |
313 | return 0; |
314 | out[oo++] = 0xe0 | (mlen - 3); |
315 | out[oo++] = mofs & 0xff; |
316 | out[oo++] = mofs >> 8; |
317 | } |
318 | else |
319 | { |
320 | assert(mlen >= 19 && mlen <= 4095 + 19 && mofs < 65536)({ if (mlen >= 19 && mlen <= 4095 + 19 && mofs < 65536) ; else __assert_fail ("mlen >= 19 && mlen <= 4095 + 19 && mofs < 65536" , "/home/brain/Projects/upstream/libsolv/src/repopage.c", 320 , __PRETTY_FUNCTION__); }); |
321 | if (oo + 4 >= out_len) |
322 | return 0; |
323 | out[oo++] = 0xf0 | ((mlen - 19) >> 8); |
324 | out[oo++] = (mlen - 19) & 0xff; |
325 | out[oo++] = mofs & 0xff; |
326 | out[oo++] = mofs >> 8; |
327 | } |
328 | /* Insert the hashes for the compressed run [io..io+mlen-1]. |
329 | For [io] we have it already done at the start of the loop. |
330 | So it's from [io+1..io+mlen-1], and we need three chars per |
331 | hash, so the accessed characters will be [io+1..io+mlen-1+2], |
332 | ergo io+mlen+1 < in_len. */ |
333 | mlen--; |
334 | io++; |
335 | while (mlen--) |
336 | { |
337 | if (io + 2 < in_len) |
338 | { |
339 | unsigned int hval = |
340 | in[io] | in[io + 1] << 8 | in[io + 2] << 16; |
341 | hval = (hval ^ (hval << 5) ^ (hval >> 5)) - hval * 5; |
342 | hval = hval & (HS(65536) - 1); |
343 | hnext[io] = htab[hval]; |
344 | htab[hval] = io; |
345 | } |
346 | io++; |
347 | }; |
348 | } |
349 | } |
350 | /* We might have some characters left. */ |
351 | if (io < in_len && !litofs) |
352 | litofs = io + 1; |
353 | io = in_len; |
354 | if (litofs) |
355 | { |
356 | unsigned litlen; |
357 | litofs--; |
358 | litlen = io - litofs; |
359 | /* fprintf(stderr, "lit: %d\n", litlen); */ |
360 | while (litlen) |
361 | { |
362 | unsigned int easy_sz; |
363 | /* Emit everything we can as self-describers. As soon as we hit a |
364 | byte we can't emit as such we're going to emit a length |
365 | descriptor anyway, so we can as well include bytes < 0x80 which |
366 | might follow afterwards in that run. */ |
367 | for (easy_sz = 0; easy_sz < litlen && in[litofs + easy_sz] < 0x80; |
368 | easy_sz++) |
369 | ; |
370 | if (easy_sz) |
371 | { |
372 | if (oo + easy_sz >= out_len) |
373 | return 0; |
374 | memcpy(out + oo, in + litofs, easy_sz); |
375 | litofs += easy_sz; |
376 | oo += easy_sz; |
377 | litlen -= easy_sz; |
378 | if (!litlen) |
379 | break; |
380 | } |
381 | if (litlen <= 32) |
382 | { |
383 | if (oo + 1 + litlen >= out_len) |
384 | return 0; |
385 | out[oo++] = 0x80 | (litlen - 1); |
386 | while (litlen--) |
387 | out[oo++] = in[litofs++]; |
388 | break; |
389 | } |
390 | else |
391 | { |
392 | /* Literal length > 32, so chunk it. */ |
393 | if (oo + 1 + 32 >= out_len) |
394 | return 0; |
395 | out[oo++] = 0x80 | 31; |
396 | memcpy(out + oo, in + litofs, 32); |
397 | oo += 32; |
398 | litofs += 32; |
399 | litlen -= 32; |
400 | } |
401 | } |
402 | litofs = 0; |
Value stored to 'litofs' is never read | |
403 | } |
404 | return oo; |
405 | } |
406 | |
407 | static unsigned int |
408 | unchecked_decompress_buf(const unsigned char *in, unsigned int in_len, |
409 | unsigned char *out, |
410 | unsigned int out_len __attribute__((unused))) |
411 | { |
412 | unsigned char *orig_out = out; |
413 | const unsigned char *in_end = in + in_len; |
414 | while (in < in_end) |
415 | { |
416 | unsigned int first = *in++; |
417 | int o; |
418 | switch (first >> 4) |
419 | { |
420 | default: |
421 | /* This default case can't happen, but GCCs VRP is not strong |
422 | enough to see this, so make this explicitely not fall to |
423 | the end of the switch, so that we don't have to initialize |
424 | o above. */ |
425 | continue; |
426 | case 0: case 1: |
427 | case 2: case 3: |
428 | case 4: case 5: |
429 | case 6: case 7: |
430 | /* a 0LLLLLLL */ |
431 | /* fprintf (stderr, "lit: 1\n"); */ |
432 | *out++ = first; |
433 | continue; |
434 | case 8: case 9: |
435 | /* b 100lllll <l+1 bytes> */ |
436 | { |
437 | unsigned int l = first & 31; |
438 | /* fprintf (stderr, "lit: %d\n", l); */ |
439 | do |
440 | *out++ = *in++; |
441 | while (l--); |
442 | continue; |
443 | } |
444 | case 10: case 11: |
445 | /* c 101oolll <8o> */ |
446 | { |
447 | o = first & (3 << 3); |
448 | o = (o << 5) | *in++; |
449 | first = (first & 7) + 2; |
450 | break; |
451 | } |
452 | case 12: case 13: |
453 | /* d 110lllll <8o> */ |
454 | { |
455 | o = *in++; |
456 | first = (first & 31) + 10; |
457 | break; |
458 | } |
459 | case 14: |
460 | /* e 1110llll <8o> <8o> */ |
461 | { |
462 | o = in[0] | (in[1] << 8); |
463 | in += 2; |
464 | first = first & 31; |
465 | first += 3; |
466 | break; |
467 | } |
468 | case 15: |
469 | /* f1 1111llll <8o> <8o> <8l> */ |
470 | /* f2 11110lll <8o> <8o> <8l> */ |
471 | /* g 11111lll <8o> <8o> <8o> <8l> */ |
472 | { |
473 | first = first & 15; |
474 | if (first >= 8) |
475 | { |
476 | first = (((first - 8) << 8) | in[0]) + 5; |
477 | o = in[1] | (in[2] << 8) | (in[3] << 16); |
478 | in += 4; |
479 | } |
480 | else |
481 | { |
482 | first = ((first << 8) | in[0]) + 19; |
483 | o = in[1] | (in[2] << 8); |
484 | in += 3; |
485 | } |
486 | break; |
487 | } |
488 | } |
489 | /* fprintf(stderr, "ref: %d @ %d\n", first, o); */ |
490 | o++; |
491 | o = -o; |
492 | #if 0 |
493 | /* We know that first will not be zero, and this loop structure is |
494 | better optimizable. */ |
495 | do |
496 | { |
497 | *out = *(out - o); |
498 | out++; |
499 | } |
500 | while (--first); |
501 | #else |
502 | switch (first) |
503 | { |
504 | case 18: *out = *(out + o); out++; |
505 | case 17: *out = *(out + o); out++; |
506 | case 16: *out = *(out + o); out++; |
507 | case 15: *out = *(out + o); out++; |
508 | case 14: *out = *(out + o); out++; |
509 | case 13: *out = *(out + o); out++; |
510 | case 12: *out = *(out + o); out++; |
511 | case 11: *out = *(out + o); out++; |
512 | case 10: *out = *(out + o); out++; |
513 | case 9: *out = *(out + o); out++; |
514 | case 8: *out = *(out + o); out++; |
515 | case 7: *out = *(out + o); out++; |
516 | case 6: *out = *(out + o); out++; |
517 | case 5: *out = *(out + o); out++; |
518 | case 4: *out = *(out + o); out++; |
519 | case 3: *out = *(out + o); out++; |
520 | case 2: *out = *(out + o); out++; |
521 | case 1: *out = *(out + o); out++; |
522 | case 0: break; |
523 | default: |
524 | /* Duff duff :-) */ |
525 | switch (first & 15) |
526 | { |
527 | do |
528 | { |
529 | case 0: *out = *(out + o); out++; |
530 | case 15: *out = *(out + o); out++; |
531 | case 14: *out = *(out + o); out++; |
532 | case 13: *out = *(out + o); out++; |
533 | case 12: *out = *(out + o); out++; |
534 | case 11: *out = *(out + o); out++; |
535 | case 10: *out = *(out + o); out++; |
536 | case 9: *out = *(out + o); out++; |
537 | case 8: *out = *(out + o); out++; |
538 | case 7: *out = *(out + o); out++; |
539 | case 6: *out = *(out + o); out++; |
540 | case 5: *out = *(out + o); out++; |
541 | case 4: *out = *(out + o); out++; |
542 | case 3: *out = *(out + o); out++; |
543 | case 2: *out = *(out + o); out++; |
544 | case 1: *out = *(out + o); out++; |
545 | } |
546 | while ((int)(first -= 16) > 0); |
547 | } |
548 | break; |
549 | } |
550 | #endif |
551 | } |
552 | return out - orig_out; |
553 | } |
554 | |
555 | /**********************************************************************/ |
556 | |
557 | void repopagestore_init(Repopagestore *store) |
558 | { |
559 | memset(store, 0, sizeof(*store)); |
560 | store->pagefd = -1; |
561 | } |
562 | |
563 | void repopagestore_free(Repopagestore *store) |
564 | { |
565 | store->blob_store = solv_free(store->blob_store); |
566 | store->file_pages = solv_free(store->file_pages); |
567 | store->mapped_at = solv_free(store->mapped_at); |
568 | store->mapped = solv_free(store->mapped); |
569 | if (store->pagefd != -1) |
570 | close(store->pagefd); |
571 | store->pagefd = -1; |
572 | } |
573 | |
574 | |
575 | /**********************************************************************/ |
576 | |
577 | unsigned char * |
578 | repopagestore_load_page_range(Repopagestore *store, unsigned int pstart, unsigned int pend) |
579 | { |
580 | /* Make sure all pages from PSTART to PEND (inclusive) are loaded, |
581 | and are consecutive. Return a pointer to the mapping of PSTART. */ |
582 | unsigned char buf[REPOPAGE_BLOBSIZE(1 << 15)]; |
583 | unsigned int i, best, pnum; |
584 | |
585 | if (pstart == pend) |
586 | { |
587 | /* Quick check in case the requested page is already mapped */ |
588 | if (store->mapped_at[pstart] != -1) |
589 | return store->blob_store + store->mapped_at[pstart]; |
590 | } |
591 | else |
592 | { |
593 | /* Quick check in case all pages are already mapped and consecutive. */ |
594 | for (pnum = pstart; pnum <= pend; pnum++) |
595 | if (store->mapped_at[pnum] == -1 |
596 | || (pnum > pstart |
597 | && store->mapped_at[pnum] |
598 | != store->mapped_at[pnum-1] + REPOPAGE_BLOBSIZE(1 << 15))) |
599 | break; |
600 | if (pnum > pend) |
601 | return store->blob_store + store->mapped_at[pstart]; |
602 | } |
603 | |
604 | if (store->pagefd == -1 || !store->file_pages) |
605 | return 0; /* no backing file */ |
606 | |
607 | #ifdef DEBUG_PAGING |
608 | fprintf(stderrstderr, "PAGE: want %d pages starting at %d\n", pend - pstart + 1, pstart); |
609 | #endif |
610 | |
611 | /* Ensure that we can map the numbers of pages we need at all. */ |
612 | if (pend - pstart + 1 > store->nmapped) |
613 | { |
614 | unsigned int oldcan = store->nmapped; |
615 | store->nmapped = pend - pstart + 1; |
616 | if (store->nmapped < 4) |
617 | store->nmapped = 4; |
618 | store->mapped = solv_realloc2(store->mapped, store->nmapped, sizeof(store->mapped[0])); |
619 | for (i = oldcan; i < store->nmapped; i++) |
620 | store->mapped[i] = -1; |
621 | store->blob_store = solv_realloc2(store->blob_store, store->nmapped, REPOPAGE_BLOBSIZE(1 << 15)); |
622 | #ifdef DEBUG_PAGING |
623 | fprintf(stderrstderr, "PAGE: can map %d pages\n", store->nmapped); |
624 | #endif |
625 | } |
626 | |
627 | if (store->mapped_at[pstart] != -1) |
628 | { |
629 | /* assume forward search */ |
630 | best = store->mapped_at[pstart] / REPOPAGE_BLOBSIZE(1 << 15); |
631 | if (best + (pend - pstart) >= store->nmapped) |
632 | best = 0; |
633 | } |
634 | else if (store->mapped_at[pend] != -1) |
635 | { |
636 | /* assume backward search */ |
637 | best = store->mapped_at[pend] / REPOPAGE_BLOBSIZE(1 << 15); |
638 | if (best < pend - pstart) |
639 | best = store->nmapped - 1; |
640 | best -= pend - pstart; |
641 | } |
642 | else |
643 | { |
644 | /* choose some "random" location to avoid thrashing */ |
645 | best = (pstart + store->rr_counter++) % (store->nmapped - pend + pstart); |
646 | } |
647 | |
648 | /* So we want to map our pages from [best] to [best+pend-pstart]. |
649 | Use a very simple strategy, which doesn't make the best use of |
650 | our resources, but works. Throw away all pages in that range |
651 | (even ours) then copy around ours or read them in. */ |
652 | for (i = best, pnum = pstart; pnum <= pend; i++, pnum++) |
653 | { |
654 | unsigned int pnum_mapped_at; |
655 | unsigned int oldpnum = store->mapped[i]; |
656 | if (oldpnum != -1) |
657 | { |
658 | if (oldpnum == pnum) |
659 | continue; /* already have the correct page */ |
660 | /* Evict this page. */ |
661 | #ifdef DEBUG_PAGING |
662 | fprintf(stderrstderr, "PAGE: evict page %d from %d\n", oldpnum, i); |
663 | #endif |
664 | store->mapped[i] = -1; |
665 | store->mapped_at[oldpnum] = -1; |
666 | } |
667 | /* check if we can copy the correct content (before it gets evicted) */ |
668 | pnum_mapped_at = store->mapped_at[pnum]; |
669 | if (pnum_mapped_at != -1 && pnum_mapped_at != i * REPOPAGE_BLOBSIZE(1 << 15)) |
670 | { |
671 | void *dest = store->blob_store + i * REPOPAGE_BLOBSIZE(1 << 15); |
672 | #ifdef DEBUG_PAGING |
673 | fprintf(stderrstderr, "PAGECOPY: %d from %d to %d\n", pnum, pnum_mapped_at / REPOPAGE_BLOBSIZE(1 << 15), i); |
674 | #endif |
675 | memcpy(dest, store->blob_store + pnum_mapped_at, REPOPAGE_BLOBSIZE(1 << 15)); |
676 | store->mapped[pnum_mapped_at / REPOPAGE_BLOBSIZE(1 << 15)] = -1; |
677 | store->mapped[i] = pnum; |
678 | store->mapped_at[pnum] = i * REPOPAGE_BLOBSIZE(1 << 15); |
679 | } |
680 | } |
681 | |
682 | /* Everything is free now. Read in or copy the pages we want. */ |
683 | for (i = best, pnum = pstart; pnum <= pend; i++, pnum++) |
684 | { |
685 | void *dest = store->blob_store + i * REPOPAGE_BLOBSIZE(1 << 15); |
686 | if (store->mapped_at[pnum] != -1) |
687 | { |
688 | unsigned int pnum_mapped_at = store->mapped_at[pnum]; |
689 | if (pnum_mapped_at != i * REPOPAGE_BLOBSIZE(1 << 15)) |
690 | { |
691 | #ifdef DEBUG_PAGING |
692 | fprintf(stderrstderr, "PAGECOPY: %d from %d to %d\n", pnum, pnum_mapped_at / REPOPAGE_BLOBSIZE(1 << 15), i); |
693 | #endif |
694 | /* Still mapped somewhere else, so just copy it from there. */ |
695 | memcpy(dest, store->blob_store + pnum_mapped_at, REPOPAGE_BLOBSIZE(1 << 15)); |
696 | store->mapped[pnum_mapped_at / REPOPAGE_BLOBSIZE(1 << 15)] = -1; |
697 | } |
698 | } |
699 | else |
700 | { |
701 | Attrblobpage *p = store->file_pages + pnum; |
702 | unsigned int in_len = p->page_size; |
703 | unsigned int compressed = in_len & 1; |
704 | in_len >>= 1; |
705 | #ifdef DEBUG_PAGING |
706 | fprintf(stderrstderr, "PAGEIN: %d to %d", pnum, i); |
707 | #endif |
708 | if (pread(store->pagefd, compressed ? buf : dest, in_len, store->file_offset + p->page_offset) != in_len) |
709 | { |
710 | perror("mapping pread"); |
711 | return 0; |
712 | } |
713 | if (compressed) |
714 | { |
715 | unsigned int out_len; |
716 | out_len = unchecked_decompress_buf(buf, in_len, dest, REPOPAGE_BLOBSIZE(1 << 15)); |
717 | if (out_len != REPOPAGE_BLOBSIZE(1 << 15) && pnum < store->num_pages - 1) |
718 | { |
719 | #ifdef DEBUG_PAGING |
720 | fprintf(stderrstderr, "can't decompress\n"); |
721 | #endif |
722 | return 0; |
723 | } |
724 | #ifdef DEBUG_PAGING |
725 | fprintf(stderrstderr, " (expand %d to %d)", in_len, out_len); |
726 | #endif |
727 | } |
728 | #ifdef DEBUG_PAGING |
729 | fprintf(stderrstderr, "\n"); |
730 | #endif |
731 | } |
732 | store->mapped_at[pnum] = i * REPOPAGE_BLOBSIZE(1 << 15); |
733 | store->mapped[i] = pnum; |
734 | } |
735 | return store->blob_store + best * REPOPAGE_BLOBSIZE(1 << 15); |
736 | } |
737 | |
738 | unsigned int |
739 | repopagestore_compress_page(unsigned char *page, unsigned int len, unsigned char *cpage, unsigned int max) |
740 | { |
741 | return compress_buf(page, len, cpage, max); |
742 | } |
743 | |
744 | #define SOLV_ERROR_EOF3 3 |
745 | #define SOLV_ERROR_CORRUPT6 6 |
746 | |
747 | static inline unsigned int |
748 | read_u32(FILE *fp) |
749 | { |
750 | int c, i; |
751 | unsigned int x = 0; |
752 | |
753 | for (i = 0; i < 4; i++) |
754 | { |
755 | c = getc(fp)_IO_getc (fp); |
756 | if (c == EOF(-1)) |
757 | return 0; |
758 | x = (x << 8) | c; |
759 | } |
760 | return x; |
761 | } |
762 | |
763 | /* Try to either setup on-demand paging (using FP as backing |
764 | file), or in case that doesn't work (FP not seekable) slurps in |
765 | all pages and deactivates paging. */ |
766 | int |
767 | repopagestore_read_or_setup_pages(Repopagestore *store, FILE *fp, unsigned int pagesz, unsigned int blobsz) |
768 | { |
769 | unsigned int npages; |
770 | unsigned int i; |
771 | unsigned int can_seek; |
772 | unsigned int cur_page_ofs; |
773 | unsigned char buf[REPOPAGE_BLOBSIZE(1 << 15)]; |
774 | |
775 | if (pagesz != REPOPAGE_BLOBSIZE(1 << 15)) |
776 | { |
777 | /* We could handle this by slurping in everything. */ |
778 | return SOLV_ERROR_CORRUPT6; |
779 | } |
780 | can_seek = 1; |
781 | if ((store->file_offset = ftell(fp)) < 0) |
782 | can_seek = 0; |
783 | clearerr(fp); |
784 | if (can_seek) |
785 | store->pagefd = dup(fileno(fp)); |
786 | if (store->pagefd == -1) |
787 | can_seek = 0; |
788 | else |
789 | fcntl(store->pagefd, F_SETFD2, FD_CLOEXEC1); |
790 | |
791 | #ifdef DEBUG_PAGING |
792 | fprintf(stderrstderr, "can %sseek\n", can_seek ? "" : "NOT "); |
793 | #endif |
794 | npages = (blobsz + REPOPAGE_BLOBSIZE(1 << 15) - 1) / REPOPAGE_BLOBSIZE(1 << 15); |
795 | |
796 | store->num_pages = npages; |
797 | store->mapped_at = solv_malloc2(npages, sizeof(*store->mapped_at)); |
798 | |
799 | /* If we can't seek on our input we have to slurp in everything. |
800 | * Otherwise set up file_pages containing offest/length of the |
801 | * pages */ |
802 | if (can_seek) |
803 | store->file_pages = solv_malloc2(npages, sizeof(*store->file_pages)); |
804 | else |
805 | store->blob_store = solv_malloc2(npages, REPOPAGE_BLOBSIZE(1 << 15)); |
806 | cur_page_ofs = 0; |
807 | for (i = 0; i < npages; i++) |
808 | { |
809 | unsigned int in_len = read_u32(fp); |
810 | unsigned int compressed = in_len & 1; |
811 | in_len >>= 1; |
812 | #ifdef DEBUG_PAGING |
813 | fprintf(stderrstderr, "page %d: len %d (%scompressed)\n", |
814 | i, in_len, compressed ? "" : "not "); |
815 | #endif |
816 | if (can_seek) |
817 | { |
818 | Attrblobpage *p = store->file_pages + i; |
819 | cur_page_ofs += 4; |
820 | store->mapped_at[i] = -1; /* not mapped yet */ |
821 | p->page_offset = cur_page_ofs; |
822 | p->page_size = in_len * 2 + compressed; |
823 | if (fseek(fp, in_len, SEEK_CUR1) < 0) |
824 | { |
825 | /* We can't fall back to non-seeking behaviour as we already |
826 | read over some data pages without storing them away. */ |
827 | close(store->pagefd); |
828 | store->pagefd = -1; |
829 | return SOLV_ERROR_EOF3; |
830 | } |
831 | cur_page_ofs += in_len; |
832 | } |
833 | else |
834 | { |
835 | unsigned int out_len; |
836 | void *dest = store->blob_store + i * REPOPAGE_BLOBSIZE(1 << 15); |
837 | store->mapped_at[i] = i * REPOPAGE_BLOBSIZE(1 << 15); |
838 | /* We can't seek, so suck everything in. */ |
839 | if (fread(compressed ? buf : dest, in_len, 1, fp) != 1) |
840 | { |
841 | perror("fread"); |
842 | return SOLV_ERROR_EOF3; |
843 | } |
844 | if (compressed) |
845 | { |
846 | out_len = unchecked_decompress_buf(buf, in_len, dest, REPOPAGE_BLOBSIZE(1 << 15)); |
847 | if (out_len != REPOPAGE_BLOBSIZE(1 << 15) && i < npages - 1) |
848 | { |
849 | return SOLV_ERROR_CORRUPT6; |
850 | } |
851 | } |
852 | } |
853 | } |
854 | return 0; |
855 | } |
856 | |
857 | void |
858 | repopagestore_disable_paging(Repopagestore *store) |
859 | { |
860 | if (store->num_pages) |
861 | repopagestore_load_page_range(store, 0, store->num_pages - 1); |
862 | } |
863 | |
864 | #ifdef STANDALONE |
865 | |
866 | static void |
867 | transfer_file(FILE * from, FILE * to, int compress) |
868 | { |
869 | unsigned char inb[BLOCK_SIZE(65536*1)]; |
870 | unsigned char outb[BLOCK_SIZE(65536*1)]; |
871 | while (!feof (from) && !ferror (from)) |
872 | { |
873 | unsigned int in_len, out_len; |
874 | if (compress) |
875 | { |
876 | in_len = fread(inb, 1, BLOCK_SIZE(65536*1), from); |
877 | if (in_len) |
878 | { |
879 | unsigned char *b = outb; |
880 | out_len = compress_buf(inb, in_len, outb, sizeof (outb)); |
881 | if (!out_len) |
882 | b = inb, out_len = in_len; |
883 | if (fwrite(&out_len, sizeof (out_len), 1, to) != 1) |
884 | { |
885 | perror("write size"); |
886 | exit (1); |
887 | } |
888 | if (fwrite(b, out_len, 1, to) != 1) |
889 | { |
890 | perror("write data"); |
891 | exit (1); |
892 | } |
893 | } |
894 | } |
895 | else |
896 | { |
897 | if (fread(&in_len, sizeof(in_len), 1, from) != 1) |
898 | { |
899 | if (feof(from)) |
900 | return; |
901 | perror("can't read size"); |
902 | exit(1); |
903 | } |
904 | if (fread(inb, in_len, 1, from) != 1) |
905 | { |
906 | perror("can't read data"); |
907 | exit(1); |
908 | } |
909 | out_len = |
910 | unchecked_decompress_buf(inb, in_len, outb, sizeof(outb)); |
911 | if (fwrite(outb, out_len, 1, to) != 1) |
912 | { |
913 | perror("can't write output"); |
914 | exit(1); |
915 | } |
916 | } |
917 | } |
918 | } |
919 | |
920 | /* Just for benchmarking purposes. */ |
921 | static void |
922 | dumb_memcpy(void *dest, const void *src, unsigned int len) |
923 | { |
924 | char *d = dest; |
925 | const char *s = src; |
926 | while (len--) |
927 | *d++ = *s++; |
928 | } |
929 | |
930 | static void |
931 | benchmark(FILE * from) |
932 | { |
933 | unsigned char inb[BLOCK_SIZE(65536*1)]; |
934 | unsigned char outb[BLOCK_SIZE(65536*1)]; |
935 | unsigned int in_len = fread(inb, 1, BLOCK_SIZE(65536*1), from); |
936 | unsigned int out_len; |
937 | if (!in_len) |
938 | { |
939 | perror("can't read from input"); |
940 | exit(1); |
941 | } |
942 | |
943 | unsigned int calib_loop; |
944 | unsigned int per_loop; |
945 | unsigned int i, j; |
946 | clock_t start, end; |
947 | float seconds; |
948 | |
949 | #if 0 |
950 | calib_loop = 1; |
951 | per_loop = 0; |
952 | start = clock(); |
953 | while ((clock() - start) < CLOCKS_PER_SEC((__clock_t) 1000000) / 4) |
954 | { |
955 | calib_loop *= 2; |
956 | for (i = 0; i < calib_loop; i++) |
957 | dumb_memcpy(outb, inb, in_len); |
958 | per_loop += calib_loop; |
959 | } |
960 | |
961 | fprintf(stderrstderr, "memcpy:\nCalibrated to %d iterations per loop\n", |
962 | per_loop); |
963 | |
964 | start = clock(); |
965 | for (i = 0; i < 10; i++) |
966 | for (j = 0; j < per_loop; j++) |
967 | dumb_memcpy(outb, inb, in_len); |
968 | end = clock(); |
969 | seconds = (end - start) / (float) CLOCKS_PER_SEC((__clock_t) 1000000); |
970 | fprintf(stderrstderr, "%.2f seconds == %.2f MB/s\n", seconds, |
971 | ((long long) in_len * per_loop * 10) / (1024 * 1024 * seconds)); |
972 | #endif |
973 | |
974 | calib_loop = 1; |
975 | per_loop = 0; |
976 | start = clock(); |
977 | while ((clock() - start) < CLOCKS_PER_SEC((__clock_t) 1000000) / 4) |
978 | { |
979 | calib_loop *= 2; |
980 | for (i = 0; i < calib_loop; i++) |
981 | compress_buf(inb, in_len, outb, sizeof(outb)); |
982 | per_loop += calib_loop; |
983 | } |
984 | |
985 | fprintf(stderrstderr, "compression:\nCalibrated to %d iterations per loop\n", |
986 | per_loop); |
987 | |
988 | start = clock(); |
989 | for (i = 0; i < 10; i++) |
990 | for (j = 0; j < per_loop; j++) |
991 | compress_buf(inb, in_len, outb, sizeof(outb)); |
992 | end = clock(); |
993 | seconds = (end - start) / (float) CLOCKS_PER_SEC((__clock_t) 1000000); |
994 | fprintf(stderrstderr, "%.2f seconds == %.2f MB/s\n", seconds, |
995 | ((long long) in_len * per_loop * 10) / (1024 * 1024 * seconds)); |
996 | |
997 | out_len = compress_buf(inb, in_len, outb, sizeof(outb)); |
998 | |
999 | calib_loop = 1; |
1000 | per_loop = 0; |
1001 | start = clock(); |
1002 | while ((clock() - start) < CLOCKS_PER_SEC((__clock_t) 1000000) / 4) |
1003 | { |
1004 | calib_loop *= 2; |
1005 | for (i = 0; i < calib_loop; i++) |
1006 | unchecked_decompress_buf(outb, out_len, inb, sizeof(inb)); |
1007 | per_loop += calib_loop; |
1008 | } |
1009 | |
1010 | fprintf(stderrstderr, "decompression:\nCalibrated to %d iterations per loop\n", |
1011 | per_loop); |
1012 | |
1013 | start = clock(); |
1014 | for (i = 0; i < 10; i++) |
1015 | for (j = 0; j < per_loop; j++) |
1016 | unchecked_decompress_buf(outb, out_len, inb, sizeof(inb)); |
1017 | end = clock(); |
1018 | seconds = (end - start) / (float) CLOCKS_PER_SEC((__clock_t) 1000000); |
1019 | fprintf(stderrstderr, "%.2f seconds == %.2f MB/s\n", seconds, |
1020 | ((long long) in_len * per_loop * 10) / (1024 * 1024 * seconds)); |
1021 | } |
1022 | |
1023 | int |
1024 | main(int argc, char *argv[]) |
1025 | { |
1026 | int compress = 1; |
1027 | if (argc > 1 && !strcmp(argv[1], "-d")) |
1028 | compress = 0; |
1029 | if (argc > 1 && !strcmp(argv[1], "-b")) |
1030 | benchmark(stdinstdin); |
1031 | else |
1032 | transfer_file(stdinstdin, stdoutstdout, compress); |
1033 | return 0; |
1034 | } |
1035 | |
1036 | #endif |
1037 |