2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.8 1998-03-18 09:23:55 adam
8 * Blocks are stored in chunks on free list - up to factor 2 in speed.
9 * Fixed bug that could occur in block category rearrangemen.
11 * Revision 1.7 1998/03/11 11:18:18 adam
12 * Changed the isc_merge to take into account the mfill (minimum-fill).
14 * Revision 1.6 1998/03/06 13:54:03 adam
15 * Fixed two nasty bugs in isc_merge.
17 * Revision 1.5 1997/02/12 20:42:43 adam
18 * Bug fix: during isc_merge operations, some pages weren't marked dirty
19 * even though they should be. At this point the merge operation marks
20 * a page dirty if the previous page changed at all. A better approach is
21 * to mark it dirty if the last key written changed in previous page.
23 * Revision 1.4 1996/11/08 11:15:31 adam
24 * Number of keys in chain are stored in first block and the function
25 * to retrieve this information, isc_pp_num is implemented.
27 * Revision 1.3 1996/11/04 14:08:59 adam
28 * Optimized free block usage.
30 * Revision 1.2 1996/11/01 13:36:46 adam
31 * New element, max_blocks_mem, that control how many blocks of max size
32 * to store in memory during isc_merge.
33 * Function isc_merge now ignores delete/update of identical keys and
34 * the proper blocks are then non-dirty and not written in flush_blocks.
36 * Revision 1.1 1996/11/01 08:59:15 adam
37 * First version of isc_merge that supports update/delete.
49 struct isc_merge_block {
50 int offset; /* offset in r_buf */
51 int block; /* block number of file (0 if none) */
52 int dirty; /* block is different from that on file */
55 static void opt_blocks (ISAMC is, struct isc_merge_block *mb, int ptr)
58 for (i = 0; i<ptr; i++)
61 if (no_dirty*3 < ptr*2)
64 for (i = 0; i<ptr; i++)
66 int tmp, j, j_min = -1;
67 for (j = i; j<ptr; j++)
69 if (j_min < 0 || mb[j_min].block > mb[j].block)
73 tmp = mb[j_min].block;
74 mb[j_min].block = mb[i].block;
80 static void flush_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
81 char *r_buf, int *firstpos, int cat, int last,
86 for (i = 0; i<ptr; i++)
88 /* consider this block number */
91 mb[i].block = isc_alloc_block (is, cat);
95 /* consider next block pointer */
96 if (last && i == ptr-1)
98 else if (!mb[i+1].block)
100 mb[i+1].block = isc_alloc_block (is, cat);
105 opt_blocks (is, mb, ptr);
106 for (i = 0; i<ptr; i++)
109 ISAMC_BLOCK_SIZE ssize = mb[i+1].offset - mb[i].offset;
113 /* skip rest if not dirty */
116 assert (mb[i].block);
118 *firstpos = mb[i].block;
119 if (is->method->debug > 2)
120 logf (LOG_LOG, "isc: skip ptr=%d size=%d %d %d",
121 i, ssize, cat, mb[i].block);
122 ++(is->files[cat].no_skip_writes);
129 *firstpos = mb[i].block;
130 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_1;
131 ssize += ISAMC_BLOCK_OFFSET_1;
133 memcpy (src+sizeof(int)+sizeof(ssize), numkeys,
135 if (is->method->debug > 2)
136 logf (LOG_LOG, "isc: flush ptr=%d numk=%d size=%d nextpos=%d",
137 i, *numkeys, (int) ssize, mb[i+1].block);
141 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_N;
142 ssize += ISAMC_BLOCK_OFFSET_N;
143 if (is->method->debug > 2)
144 logf (LOG_LOG, "isc: flush ptr=%d size=%d nextpos=%d",
145 i, (int) ssize, mb[i+1].block);
147 memcpy (src, &mb[i+1].block, sizeof(int));
148 memcpy (src+sizeof(int), &ssize, sizeof(ssize));
149 isc_write_block (is, cat, mb[i].block, src);
153 static int get_border (ISAMC is, struct isc_merge_block *mb, int ptr,
154 int cat, int firstpos)
156 /* Border set to initial fill or block size depending on
157 whether we are creating a new one or updating and old one.
160 int fill = mb[ptr].block ? is->method->filecat[cat].bsize :
161 is->method->filecat[cat].ifill;
162 int off = (ptr||firstpos) ? ISAMC_BLOCK_OFFSET_N : ISAMC_BLOCK_OFFSET_1;
166 return mb[ptr].offset + fill - off;
169 ISAMC_P isc_merge (ISAMC is, ISAMC_P ipos, ISAMC_I data)
172 char i_item[128], *i_item_ptr;
173 int i_more, i_mode, i;
176 char f_item[128], *f_item_ptr;
179 struct isc_merge_block mb[200];
183 char r_item_buf[128]; /* temporary result output */
184 char *r_buf; /* block with resulting data */
185 int r_offset = 0; /* current offset in r_buf */
186 int ptr = 0; /* pointer */
187 void *r_clientData; /* encode client data */
191 r_clientData = (*is->method->code_start)(ISAMC_ENCODE);
192 r_buf = is->merge_buf + 128;
194 pp = isc_pp_open (is, ipos);
195 /* read first item from file. make sure f_more indicates no boundary */
197 f_more = isc_read_item (pp, &f_item_ptr);
202 if (is->method->debug > 1)
203 logf (LOG_LOG, "isc: isc_merge begin %d %d", cat, pp->pos);
205 /* read first item from i */
207 i_more = (*data->read_item)(data->clientData, &i_item_ptr, &i_mode);
209 mb[ptr].block = pp->pos; /* is zero if no block on disk */
213 border = get_border (is, mb, ptr, cat, firstpos);
214 while (i_more || f_more)
216 char *r_item = r_item_buf;
221 /* block to block boundary in the original file. */
225 /* the resulting output is of the same category as the
228 if (r_offset <= mb[ptr].offset +is->method->filecat[cat].mfill)
230 /* the resulting output block is too small/empty. Delete
231 the original (if any)
233 if (is->method->debug > 3)
234 logf (LOG_LOG, "isc: release A");
236 isc_release_block (is, pp->cat, mb[ptr].block);
237 mb[ptr].block = pp->pos;
244 /* indicate new boundary based on the original file */
245 mb[++ptr].block = pp->pos;
246 mb[ptr].dirty = (mb[ptr-1].dirty > 1) ? 1 : 0;
247 mb[ptr].offset = r_offset;
248 if (is->method->debug > 3)
249 logf (LOG_LOG, "isc: bound ptr=%d,offset=%d",
251 if (cat==is->max_cat && ptr >= is->method->max_blocks_mem)
253 /* We are dealing with block(s) of max size. Block(s)
254 except 1 will be flushed.
256 if (is->method->debug > 2)
257 logf (LOG_LOG, "isc: flush A %d sections", ptr);
258 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
260 mb[0].block = mb[ptr-1].block;
261 mb[0].dirty = mb[ptr-1].dirty;
262 memcpy (r_buf, r_buf + mb[ptr-1].offset,
263 mb[ptr].offset - mb[ptr-1].offset);
266 mb[1].block = mb[ptr].block;
267 mb[1].dirty = mb[ptr].dirty;
268 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
270 r_offset = mb[ptr].offset;
274 border = get_border (is, mb, ptr, cat, firstpos);
281 cmp = (*is->method->compare_item)(i_item, f_item);
282 if (cmp == 0) /* insert i=f */
284 if (!i_mode) /* delete item? */
288 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
290 /* is next input item the same as current except
291 for the delete flag? */
292 cmp = (*is->method->compare_item)(i_item, f_item);
293 if (!cmp && i_mode) /* delete/insert nop? */
295 /* yes! insert as if it was an insert only */
296 memcpy (r_item, i_item, i_item_ptr - i_item);
298 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
303 /* no! delete the item */
310 memcpy (r_item, f_item, f_item_ptr - f_item);
314 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
319 f_more = isc_read_item (pp, &f_item_ptr);
321 else if (cmp > 0) /* insert f */
323 memcpy (r_item, f_item, f_item_ptr - f_item);
326 f_more = isc_read_item (pp, &f_item_ptr);
330 if (!i_mode) /* delete item which isn't there? */
332 logf (LOG_FATAL, "Inconsistent register at offset %d",
336 memcpy (r_item, i_item, i_item_ptr - i_item);
340 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
343 if (r_item) /* insert resulting item? */
345 char *r_out_ptr = r_buf + r_offset;
348 (*is->method->code_item)(ISAMC_ENCODE, r_clientData,
349 &r_out_ptr, &r_item);
350 new_offset = r_out_ptr - r_buf;
354 if (border < new_offset && border >= r_offset)
356 if (is->method->debug > 2)
357 logf (LOG_LOG, "isc: border %d %d", ptr, border);
358 /* Max size of current block category reached ...
359 make new virtual block entry */
362 mb[ptr].offset = r_offset;
363 if (cat == is->max_cat && ptr >= is->method->max_blocks_mem)
365 /* We are dealing with block(s) of max size. Block(s)
366 except one will be flushed. Note: the block(s) are
367 surely not the last one(s).
369 if (is->method->debug > 2)
370 logf (LOG_LOG, "isc: flush B %d sections", ptr-1);
371 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
373 mb[0].block = mb[ptr-1].block;
374 mb[0].dirty = mb[ptr-1].dirty;
375 memcpy (r_buf, r_buf + mb[ptr-1].offset,
376 mb[ptr].offset - mb[ptr-1].offset);
379 mb[1].block = mb[ptr].block;
380 mb[1].dirty = mb[0].dirty;
381 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
382 memcpy (r_buf + mb[1].offset, r_buf + r_offset,
383 new_offset - r_offset);
384 new_offset = (new_offset - r_offset) + mb[1].offset;
387 border = get_border (is, mb, ptr, cat, firstpos);
389 r_offset = new_offset;
391 if (cat < is->max_cat && ptr >= is->method->filecat[cat].mblocks)
393 /* Max number blocks in current category reached ->
394 must switch to next category (with larger block size)
398 (is->files[cat].no_remap)++;
399 /* delete all original block(s) read so far */
400 for (i = 0; i < ptr; i++)
402 isc_release_block (is, pp->cat, mb[i].block);
403 /* also delete all block to be read in the future */
406 /* remap block offsets */
407 assert (mb[j].offset == 0);
411 mb[ptr].offset = r_offset;
412 for (i = 1; i < ptr; i++)
414 int border = is->method->filecat[cat].ifill -
415 ISAMC_BLOCK_OFFSET_1 + mb[j].offset;
416 if (is->method->debug > 3)
417 logf (LOG_LOG, "isc: remap %d border=%d", i, border);
418 if (mb[i+1].offset > border && mb[i].offset <= border)
420 if (is->method->debug > 3)
421 logf (LOG_LOG, "isc: to %d %d", j, mb[i].offset);
424 mb[j].offset = mb[i].offset;
427 if (is->method->debug > 2)
428 logf (LOG_LOG, "isc: remap from %d to %d sections to cat %d",
431 border = get_border (is, mb, ptr, cat, firstpos);
432 if (is->method->debug > 3)
433 logf (LOG_LOG, "isc: border=%d r_offset=%d", border, r_offset);
436 if (mb[ptr].offset < r_offset)
437 { /* make the final boundary offset */
440 mb[ptr].offset = r_offset;
443 { /* empty output. Release last block if any */
444 if (cat == pp->cat && mb[ptr].block)
446 if (is->method->debug > 3)
447 logf (LOG_LOG, "isc: release C");
448 isc_release_block (is, pp->cat, mb[ptr].block);
455 if (is->method->debug > 2)
456 logf (LOG_LOG, "isc: flush C, %d sections", ptr);
460 /* we have to patch initial block with num keys if that
462 if (numKeys != isc_pp_num (pp))
464 if (is->method->debug > 2)
465 logf (LOG_LOG, "isc: patch num keys firstpos=%d num=%d",
467 bf_write (is->files[cat].bf, firstpos, ISAMC_BLOCK_OFFSET_N,
468 sizeof(numKeys), &numKeys);
472 { /* we haven't flushed initial block yet and there surely are some
473 blocks to flush. Make first block dirty if numKeys differ */
474 if (numKeys != isc_pp_num (pp))
477 /* flush rest of block(s) in r_buf */
478 flush_blocks (is, mb, ptr, r_buf, &firstpos, cat, 1, &numKeys);
480 (*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
483 if (is->method->debug > 1)
484 logf (LOG_LOG, "isc: isc_merge return %d %d", cat, firstpos);
486 return cat + firstpos * 8;