-
Notifications
You must be signed in to change notification settings - Fork 177
/
nestloopindexexecutor.cpp
472 lines (417 loc) · 19.3 KB
/
nestloopindexexecutor.cpp
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
/* This file is part of VoltDB.
* Copyright (C) 2008-2010 VoltDB L.L.C.
*
* This file contains original code and/or modifications of original code.
* Any modifications made by VoltDB L.L.C. are licensed under the following
* terms and conditions:
*
* VoltDB is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* VoltDB is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with VoltDB. If not, see <http://www.gnu.org/licenses/>.
*/
/* Copyright (C) 2008 by H-Store Project
* Brown University
* Massachusetts Institute of Technology
* Yale University
*
* Permission is hereby granted, free of charge, to any person obtaining
* a copy of this software and associated documentation files (the
* "Software"), to deal in the Software without restriction, including
* without limitation the rights to use, copy, modify, merge, publish,
* distribute, sublicense, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
*
* The above copyright notice and this permission notice shall be
* included in all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT
* IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*/
#include <vector>
#include <string>
#include "nestloopindexexecutor.h"
#include "common/debuglog.h"
#include "common/tabletuple.h"
#include "common/FatalException.hpp"
#include "execution/VoltDBEngine.h"
#include "expressions/abstractexpression.h"
#include "plannodes/nestloopindexnode.h"
#include "plannodes/indexscannode.h"
#include "storage/table.h"
#include "storage/persistenttable.h"
#include "storage/temptable.h"
#include "indexes/tableindex.h"
#include "storage/tableiterator.h"
#include "storage/tablefactory.h"
#ifdef ANTICACHE
#include "anticache/AntiCacheEvictionManager.h"
#endif
using namespace voltdb;
bool NestLoopIndexExecutor::p_init(AbstractPlanNode* abstract_node,
const catalog::Database* catalog_db, int* tempTableMemoryInBytes)
{
VOLT_TRACE("init NLIJ Executor");
assert(tempTableMemoryInBytes);
node = dynamic_cast<NestLoopIndexPlanNode*>(abstract_node);
assert(node);
inline_node = dynamic_cast<IndexScanPlanNode*>(node->getInlinePlanNode(PLAN_NODE_TYPE_INDEXSCAN));
assert(inline_node);
join_type = node->getJoinType();
m_lookupType = inline_node->getLookupType();
//
// We need exactly one input table and a target table
//
assert(node->getInputTables().size() == 1);
Table* input_table = node->getInputTables()[0];
assert(input_table);
Table* target_table = inline_node->getTargetTable();
assert(target_table);
VOLT_DEBUG("Initializing Output PlanColumns for %s", node->debug().c_str());
// Our output table will have all the columns from outer and inner table
std::vector<boost::shared_ptr<const TableColumn> > columns;
// For passing to plan node counterpart
const TupleSchema *first = input_table->schema();
const TupleSchema *second = target_table->schema();
TupleSchema *schema = TupleSchema::createTupleSchema(first, second);
int combinedColumnCount =
input_table->columnCount() + target_table->columnCount();
std::string *columnNames = new std::string[combinedColumnCount];
std::vector<int> outputColumnGuids;
int cur_index = 0;
// 2012-02-08
// The NestLoopIndexPlanNode will have proper output column information, so
// we can rely on that instead of trying to mash together the PlanColumn guids
// from the child node and the target table. Without this, then doing an aggregate
// right after this won't work properly.
// copy from outer table (input table)
for (int col_ctr = 0, col_cnt = input_table->columnCount();
col_ctr < col_cnt;
col_ctr++, cur_index++) {
columnNames[cur_index] = input_table->columnName(col_ctr);
outputColumnGuids.push_back(node->getOutputColumnGuids()[cur_index]);
VOLT_TRACE("[%d] %s [guid=%d]\n",
cur_index, columnNames[cur_index].c_str(), outputColumnGuids[cur_index]);
// outputColumnGuids.
// push_back(node->getChildren()[0]->getOutputColumnGuids()[col_ctr]);
} // FOR
// copy from inner table (target table)
for (int col_ctr = 0, col_cnt = target_table->columnCount();
col_ctr < col_cnt;
col_ctr++, cur_index++) {
columnNames[cur_index] = target_table->columnName(col_ctr);
outputColumnGuids.push_back(node->getOutputColumnGuids()[cur_index]);
// HACK: Since our output columns for the inline IndexScan now include the outer table,
// we need to use the cur_index as the proper offset to get the right guid
// outputColumnGuids.push_back(inline_node->getOutputColumnGuids()[cur_index]);
VOLT_TRACE("[%d] %s [guid=%d]\n",
cur_index, columnNames[cur_index].c_str(), outputColumnGuids[cur_index]);
} // FOR
// create the output table
node->setOutputTable(TableFactory::getTempTable(node->getInputTables()[0]->databaseId(),
"temp", schema, columnNames, tempTableMemoryInBytes));
// Set the mapping of column names to column indexes in output tables
node->setOutputColumnGuids(outputColumnGuids);
// clean up
delete[] columnNames;
//
// Make sure that we actually search keys
//
int num_of_searchkeys = (int)inline_node->getSearchKeyExpressions().size();
if (num_of_searchkeys == 0) {
VOLT_ERROR("There are no search key expressions for the internal"
" PlanNode '%s' of PlanNode '%s'",
inline_node->debug().c_str(), node->debug().c_str());
return false;
}
for (int ctr = 0; ctr < num_of_searchkeys; ctr++) {
if (inline_node->getSearchKeyExpressions()[ctr] == NULL) {
VOLT_ERROR("The search key expression at position '%d' is NULL for"
" internal PlanNode '%s' of PlanNode '%s'",
ctr, inline_node->debug().c_str(), node->debug().c_str());
return false;
}
}
// output must be a temp table
output_table = dynamic_cast<TempTable*>(node->getOutputTable());
assert(output_table);
inner_table = dynamic_cast<PersistentTable*>(inline_node->getTargetTable());
assert(inner_table);
inner_catalogTable = catalog_db->tables().get(inner_table->name());
assert(node->getInputTables().size() == 1);
outer_table = node->getInputTables()[0];
assert(outer_table);
//
// Grab the Index from our inner table
// We'll throw an error if the index is missing
//
index = inner_table->index(inline_node->getTargetIndexName());
if (index == NULL) {
VOLT_ERROR("Failed to retreive index '%s' from inner table '%s' for"
" internal PlanNode '%s'",
inline_node->getTargetIndexName().c_str(),
inner_table->name().c_str(), inline_node->debug().c_str());
return false;
}
index_values = TableTuple(index->getKeySchema());
index_values_backing_store = new char[index->getKeySchema()->tupleLength()];
index_values.move( index_values_backing_store - TUPLE_HEADER_SIZE);
index_values.setAllNulls();
return true;
}
bool NestLoopIndexExecutor::p_execute(const NValueArray ¶ms, ReadWriteTracker *tracker)
{
VOLT_TRACE ("executing NestLoopIndex...");
assert (node == dynamic_cast<NestLoopIndexPlanNode*>(abstract_node));
assert(node);
assert (inline_node == dynamic_cast<IndexScanPlanNode*>(node->getInlinePlanNode(PLAN_NODE_TYPE_INDEXSCAN)));
assert(inline_node);
assert (output_table == dynamic_cast<TempTable*>(node->getOutputTable()));
assert(output_table);
//inner_table is the table that has the index to be used in this executor
assert (inner_table == dynamic_cast<PersistentTable*>(inline_node->getTargetTable()));
assert(inner_table);
//outer_table is the input table that have tuples to be iterated
assert(node->getInputTables().size() == 1);
assert (outer_table == node->getInputTables()[0]);
assert (outer_table);
VOLT_TRACE ("outer table:\n %s", outer_table->debug().c_str());
VOLT_TRACE ("inner table:\n %s", inner_table->debug().c_str());
//
// Substitute parameter to SEARCH KEY Note that the expressions
// will include TupleValueExpression even after this substitution
//
int num_of_searchkeys = (int)inline_node->getSearchKeyExpressions().size();
for (int ctr = 0; ctr < num_of_searchkeys; ctr++) {
VOLT_TRACE("Search Key[%d] before substitution:\n%s",
ctr, inline_node->getSearchKeyExpressions()[ctr]->debug(true).c_str());
inline_node->getSearchKeyExpressions()[ctr]->substitute(params);
VOLT_TRACE("Search Key[%d] after substitution:\n%s",
ctr, inline_node->getSearchKeyExpressions()[ctr]->debug(true).c_str());
}
// end expression
AbstractExpression* end_expression = inline_node->getEndExpression();
if (end_expression) {
end_expression->substitute(params);
VOLT_TRACE("End Expression:\n%s", end_expression->debug(true).c_str());
}
// post expression
AbstractExpression* post_expression = inline_node->getPredicate();
if (post_expression != NULL) {
post_expression->substitute(params);
VOLT_TRACE("Post Expression:\n%s", post_expression->debug(true).c_str());
}
// Anti-Cache Variables
#ifdef ANTICACHE
AntiCacheEvictionManager* eviction_manager = executor_context->getAntiCacheEvictionManager();
bool hasEvictedTable = (eviction_manager != NULL && inner_table->getEvictedTable() != NULL);
bool blockingMergeSuccessful = false;
#endif
//
// OUTER TABLE ITERATION
//
TableTuple outer_tuple(outer_table->schema());
TableTuple inner_tuple(inner_table->schema());
TableIterator outer_iterator(outer_table);
int num_of_outer_cols = outer_table->columnCount();
int num_of_inner_cols = inner_table->columnCount();
assert (outer_tuple.sizeInValues() == outer_table->columnCount());
assert (inner_tuple.sizeInValues() == inner_table->columnCount());
TableTuple &join_tuple = output_table->tempTuple();
while (outer_iterator.next(outer_tuple)) {
VOLT_TRACE("outer_tuple:%s",
outer_tuple.debug(outer_table->name()).c_str());
outer_table->updateTupleAccessCount();
//
// Now use the outer table tuple to construct the search key
// against the inner table
//
assert (index_values.getSchema()->columnCount() == num_of_searchkeys || m_lookupType == INDEX_LOOKUP_TYPE_GT);
for (int ctr = num_of_searchkeys - 1; ctr >= 0 ; --ctr) {
index_values.
setNValue(ctr,
inline_node->getSearchKeyExpressions()[ctr]->eval(&outer_tuple, NULL));
}
VOLT_TRACE("Searching %s, isEvicted: %d", index_values.debug("").c_str(), outer_tuple.isEvicted());
//
// In order to apply the Expression trees in our join, we need
// to put the outer and inner tuples together into a single
// tuple. The column references in the Expressions have
// already been offset to accomodate this
//
for (int col_ctr = 0; col_ctr < num_of_outer_cols; ++col_ctr) {
join_tuple.setNValue(col_ctr, outer_tuple.getNValue(col_ctr));
}
//
// Our index scan on the inner table is going to have three parts:
// (1) Lookup tuples using the search key
//
// (2) For each tuple that comes back, check whether the
// end_expression is false. If it is, then we stop
// scanning. Otherwise...
//
// (3) Check whether the tuple satisfies the post expression.
// If it does, then add it to the output table
//
// Use our search key to prime the index iterator
// The loop through each tuple given to us by the iterator
//
if (m_lookupType == INDEX_LOOKUP_TYPE_EQ) {
index->moveToKey(&index_values);
} else if (m_lookupType == INDEX_LOOKUP_TYPE_GT) {
index->moveToGreaterThanKey(&index_values);
} else if (m_lookupType == INDEX_LOOKUP_TYPE_GTE) {
index->moveToKeyOrGreater(&index_values);
} else {
return false;
}
bool match = false;
while ((m_lookupType == INDEX_LOOKUP_TYPE_EQ &&
!(inner_tuple = index->nextValueAtKey()).isNullTuple()) ||
(m_lookupType != INDEX_LOOKUP_TYPE_EQ &&
!(inner_tuple = index->nextValue()).isNullTuple()))
{
VOLT_TRACE("Searching inner!");
match = true;
inner_table->updateTupleAccessCount();
// Anti-Cache Evicted Tuple Tracking
#ifdef ANTICACHE
#ifdef ANTICACHE_COUNTER
inner_tuple.setTempMergedFalse();
#endif
blockingMergeSuccessful = false;
// We are pointing to an entry for an evicted tuple
VOLT_TRACE("If condition before check: %d %d!", hasEvictedTable, inner_tuple.isEvicted());
if (hasEvictedTable && inner_tuple.isEvicted()) {
VOLT_TRACE("Tuple in NestLoopIndexScan is evicted %s", inner_catalogTable->name().c_str());
// Tell the EvictionManager's internal tracker that we touched this mofo
eviction_manager->recordEvictedAccess(inner_catalogTable, &inner_tuple);
// Pavlo: 2014-07-09
// If the tuple is evicted, then we can't continue with the rest of stuff below us.
// There is nothing else we can do with it (i.e., check expressions).
// I don't know why this wasn't here in the first place?
// MJG: 2014-02-20
// If we can merge now, let's merge
// TODO: possibly an alternate codepath that simply looks through all the tuples
// for evicted tuples and then see if we have any non-blockable accesses
if (eviction_manager->hasBlockableEvictedAccesses()) {
//VOLT_ERROR("From sync!");
#ifdef ANTICACHE_COUNTER
if (eviction_manager->m_update_access)
#endif
blockingMergeSuccessful = eviction_manager->blockingMerge();
} else {
//VOLT_ERROR("From abrt!");
blockingMergeSuccessful = false;
//eviction_manager->blockingMerge();
continue;
}
}
if (blockingMergeSuccessful) {
//printf("Scan from merge.\n");
VOLT_TRACE("grabbing tuple again");
if (m_lookupType == INDEX_LOOKUP_TYPE_EQ) {
index->moveToKey(&index_values);
inner_tuple = index->nextValueAtKey();
} else {
if (m_lookupType == INDEX_LOOKUP_TYPE_GT)
index->moveToGreaterThanKey(&index_values);
else
index->moveToKeyOrGreater(&index_values);
inner_tuple = index->nextValue();
}
#ifdef ANTICACHE_COUNTER
inner_tuple.setTempMergedFalse();
#endif
if (inner_tuple.isNullTuple()) {
VOLT_INFO("We've got a null tuple for some reason");
}
VOLT_TRACE("Merged Tuple: %s", inner_tuple.debug(inner_table->name()).c_str());
}
#endif
VOLT_TRACE("inner_tuple:%s",
inner_tuple.debug(inner_table->name()).c_str());
//
// Append the inner values to the end of our join tuple
//
for (int col_ctr = 0; col_ctr < num_of_inner_cols; ++col_ctr) {
join_tuple.setNValue(col_ctr + num_of_outer_cols,
inner_tuple.getNValue(col_ctr));
}
VOLT_TRACE("join_tuple tuple: %s",
join_tuple.debug(output_table->name()).c_str());
//
// First check whether the end_expression is now false
//
if (end_expression != NULL &&
end_expression->eval(&join_tuple, NULL).isFalse()) {
VOLT_TRACE("End Expression evaluated to false, stopping scan");
break;
}
//
// Then apply our post-predicate to do further filtering
//
if (post_expression == NULL ||
post_expression->eval(&join_tuple, NULL).isTrue()) {
//
// Try to put the tuple into our output table
//
VOLT_TRACE("MATCH: %s",
join_tuple.debug(output_table->name()).c_str());
output_table->insertTupleNonVirtual(join_tuple);
#ifdef ANTICACHE
if (hasEvictedTable) {
// update the tuple in the LRU eviction chain
eviction_manager->updateTuple(inner_table, &inner_tuple, false);
}
#endif
}
#if defined(ANTICACHE) && defined(ANTICACHE_COUNTER)
if (inner_tuple.isTempMerged()) {
//printf("isTempMerged?\n");
delete[] inner_tuple.address();
}
#endif
} // WHILE
//
// Left Outer Join
//
if (!match && join_type == JOIN_TYPE_LEFT) {
//
// Append NULLs to the end of our join tuple
//
for (int col_ctr = 0; col_ctr < num_of_inner_cols; ++col_ctr) {
const int index = col_ctr + num_of_outer_cols;
NValue value = join_tuple.getNValue(index);
value.setNull();
join_tuple.setNValue(col_ctr + num_of_outer_cols, value);
}
output_table->insertTupleNonVirtual(join_tuple);
}
} // WHILE
#ifdef ANTICACHE
// throw exception indicating evicted blocks are needed
if (hasEvictedTable && eviction_manager->hasEvictedAccesses()) {
eviction_manager->throwEvictedAccessException();
}
#endif
VOLT_TRACE ("result table:\n %s", output_table->debug().c_str());
return (true);
}
NestLoopIndexExecutor::~NestLoopIndexExecutor() {
delete [] index_values_backing_store;
}