1 /*
2 * Copyright (c) 1997, 2018, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation.
8 *
9 * This code is distributed in the hope that it will be useful, but WITHOUT
10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12 * version 2 for more details (a copy is included in the LICENSE file that
13 * accompanied this code).
14 *
15 * You should have received a copy of the GNU General Public License version
16 * 2 along with this work; if not, write to the Free Software Foundation,
17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 *
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 * or visit www.oracle.com if you need additional information or have any
21 * questions.
22 *
23 */
24
25 #include "precompiled.hpp"
26 #include "jvm.h"
27 #include "classfile/compactHashtable.inline.hpp"
28 #include "classfile/javaClasses.hpp"
29 #include "logging/logMessage.hpp"
30 #include "memory/metadataFactory.hpp"
31 #include "memory/metaspaceShared.hpp"
32 #include "runtime/vmThread.hpp"
33 #include "utilities/numberSeq.hpp"
34 #include <sys/stat.h>
35
36 /////////////////////////////////////////////////////
37 //
38 // The compact hash table writer implementations
39 //
40 CompactHashtableWriter::CompactHashtableWriter(int num_buckets,
41 CompactHashtableStats* stats) {
42 assert(DumpSharedSpaces, "dump-time only");
43 assert(num_buckets > 0, "no buckets");
44 _num_buckets = num_buckets;
45 _num_entries = 0;
46 _buckets = NEW_C_HEAP_ARRAY(GrowableArray<Entry>*, _num_buckets, mtSymbol);
47 for (int i=0; i<_num_buckets; i++) {
48 _buckets[i] = new (ResourceObj::C_HEAP, mtSymbol) GrowableArray<Entry>(0, true, mtSymbol);
49 }
50
51 _stats = stats;
52 _compact_buckets = NULL;
53 _compact_entries = NULL;
54 _num_empty_buckets = 0;
55 _num_value_only_buckets = 0;
56 _num_other_buckets = 0;
57 }
58
59 CompactHashtableWriter::~CompactHashtableWriter() {
60 for (int index = 0; index < _num_buckets; index++) {
61 GrowableArray<Entry>* bucket = _buckets[index];
62 delete bucket;
63 }
64
65 FREE_C_HEAP_ARRAY(GrowableArray<Entry>*, _buckets);
66 }
67
68 // Add a symbol entry to the temporary hash table
69 void CompactHashtableWriter::add(unsigned int hash, u4 value) {
70 int index = hash % _num_buckets;
71 _buckets[index]->append_if_missing(Entry(hash, value));
72 _num_entries++;
73 }
74
75 void CompactHashtableWriter::allocate_table() {
76 int entries_space = 0;
77 for (int index = 0; index < _num_buckets; index++) {
78 GrowableArray<Entry>* bucket = _buckets[index];
79 int bucket_size = bucket->length();
80 if (bucket_size == 1) {
81 entries_space++;
82 } else {
83 entries_space += 2 * bucket_size;
84 }
85 }
86
87 if (entries_space & ~BUCKET_OFFSET_MASK) {
88 vm_exit_during_initialization("CompactHashtableWriter::allocate_table: Overflow! "
89 "Too many entries.");
90 }
91
92 _compact_buckets = MetaspaceShared::new_ro_array<u4>(_num_buckets + 1);
93 _compact_entries = MetaspaceShared::new_ro_array<u4>(entries_space);
94
95 _stats->bucket_count = _num_buckets;
96 _stats->bucket_bytes = _compact_buckets->size() * BytesPerWord;
97 _stats->hashentry_count = _num_entries;
98 _stats->hashentry_bytes = _compact_entries->size() * BytesPerWord;
99 }
100
101 // Write the compact table's buckets
102 void CompactHashtableWriter::dump_table(NumberSeq* summary) {
103 u4 offset = 0;
104 for (int index = 0; index < _num_buckets; index++) {
105 GrowableArray<Entry>* bucket = _buckets[index];
106 int bucket_size = bucket->length();
107 if (bucket_size == 1) {
108 // bucket with one entry is compacted and only has the symbol offset
109 _compact_buckets->at_put(index, BUCKET_INFO(offset, VALUE_ONLY_BUCKET_TYPE));
110
111 Entry ent = bucket->at(0);
112 _compact_entries->at_put(offset++, ent.value());
113 _num_value_only_buckets++;
114 } else {
115 // regular bucket, each entry is a symbol (hash, offset) pair
116 _compact_buckets->at_put(index, BUCKET_INFO(offset, REGULAR_BUCKET_TYPE));
117
118 for (int i=0; i<bucket_size; i++) {
119 Entry ent = bucket->at(i);
120 _compact_entries->at_put(offset++, u4(ent.hash())); // write entry hash
121 _compact_entries->at_put(offset++, ent.value());
122 }
123 if (bucket_size == 0) {
124 _num_empty_buckets++;
125 } else {
126 _num_other_buckets++;
127 }
128 }
129 summary->add(bucket_size);
130 }
131
132 // Mark the end of the buckets
133 _compact_buckets->at_put(_num_buckets, BUCKET_INFO(offset, TABLEEND_BUCKET_TYPE));
134 assert(offset == (u4)_compact_entries->length(), "sanity");
135 }
136
137
138 // Write the compact table
139 void CompactHashtableWriter::dump(SimpleCompactHashtable *cht, const char* table_name) {
140 NumberSeq summary;
141 allocate_table();
142 dump_table(&summary);
143
144 int table_bytes = _stats->bucket_bytes + _stats->hashentry_bytes;
145 address base_address = address(MetaspaceShared::shared_rs()->base());
146 cht->init(base_address, _num_entries, _num_buckets,
147 _compact_buckets->data(), _compact_entries->data());
148
149 LogMessage(cds, hashtables) msg;
150 if (msg.is_info()) {
151 double avg_cost = 0.0;
152 if (_num_entries > 0) {
153 avg_cost = double(table_bytes)/double(_num_entries);
154 }
155 msg.info("Shared %s table stats -------- base: " PTR_FORMAT,
156 table_name, (intptr_t)base_address);
157 msg.info("Number of entries : %9d", _num_entries);
158 msg.info("Total bytes used : %9d", table_bytes);
159 msg.info("Average bytes per entry : %9.3f", avg_cost);
160 msg.info("Average bucket size : %9.3f", summary.avg());
161 msg.info("Variance of bucket size : %9.3f", summary.variance());
162 msg.info("Std. dev. of bucket size: %9.3f", summary.sd());
163 msg.info("Empty buckets : %9d", _num_empty_buckets);
164 msg.info("Value_Only buckets : %9d", _num_value_only_buckets);
165 msg.info("Other buckets : %9d", _num_other_buckets);
166 }
167 }
168
169 /////////////////////////////////////////////////////////////
170 //
171 // Customization for dumping Symbol and String tables
172
173 void CompactSymbolTableWriter::add(unsigned int hash, Symbol *symbol) {
174 uintx deltax = MetaspaceShared::object_delta(symbol);
175 // When the symbols are stored into the archive, we already check that
176 // they won't be more than MAX_SHARED_DELTA from the base address, or
177 // else the dumping would have been aborted.
178 assert(deltax <= MAX_SHARED_DELTA, "must not be");
179 u4 delta = u4(deltax);
180
181 CompactHashtableWriter::add(hash, delta);
182 }
183
184 void CompactStringTableWriter::add(unsigned int hash, oop string) {
185 CompactHashtableWriter::add(hash, oopDesc::encode_heap_oop(string));
186 }
187
188 void CompactSymbolTableWriter::dump(CompactHashtable<Symbol*, char> *cht) {
189 CompactHashtableWriter::dump(cht, "symbol");
190 }
191
192 void CompactStringTableWriter::dump(CompactHashtable<oop, char> *cht) {
193 CompactHashtableWriter::dump(cht, "string");
194 }
195
196 /////////////////////////////////////////////////////////////
197 //
198 // The CompactHashtable implementation
199 //
200
201 void SimpleCompactHashtable::serialize(SerializeClosure* soc) {
202 soc->do_ptr((void**)&_base_address);
203 soc->do_u4(&_entry_count);
204 soc->do_u4(&_bucket_count);
205 soc->do_ptr((void**)&_buckets);
206 soc->do_ptr((void**)&_entries);
207 }
208
209 bool SimpleCompactHashtable::exists(u4 value) {
210 assert(!DumpSharedSpaces, "run-time only");
211
212 if (_entry_count == 0) {
213 return false;
214 }
215
216 unsigned int hash = (unsigned int)value;
217 int index = hash % _bucket_count;
218 u4 bucket_info = _buckets[index];
219 u4 bucket_offset = BUCKET_OFFSET(bucket_info);
220 int bucket_type = BUCKET_TYPE(bucket_info);
221 u4* entry = _entries + bucket_offset;
222
223 if (bucket_type == VALUE_ONLY_BUCKET_TYPE) {
224 return (entry[0] == value);
225 } else {
226 u4*entry_max = _entries + BUCKET_OFFSET(_buckets[index + 1]);
227 while (entry <entry_max) {
228 if (entry[1] == value) {
229 return true;
230 }
231 entry += 2;
232 }
233 return false;
234 }
235 }
236
237 template <class I>
238 inline void SimpleCompactHashtable::iterate(const I& iterator) {
239 for (u4 i = 0; i < _bucket_count; i++) {
240 u4 bucket_info = _buckets[i];
241 u4 bucket_offset = BUCKET_OFFSET(bucket_info);
242 int bucket_type = BUCKET_TYPE(bucket_info);
243 u4* entry = _entries + bucket_offset;
244
245 if (bucket_type == VALUE_ONLY_BUCKET_TYPE) {
246 iterator.do_value(_base_address, entry[0]);
247 } else {
248 u4*entry_max = _entries + BUCKET_OFFSET(_buckets[i + 1]);
249 while (entry < entry_max) {
250 iterator.do_value(_base_address, entry[1]);
251 entry += 2;
252 }
253 }
254 }
255 }
256
257 template <class T, class N> void CompactHashtable<T, N>::serialize(SerializeClosure* soc) {
258 SimpleCompactHashtable::serialize(soc);
259 soc->do_u4(&_type);
260 }
261
262 class CompactHashtable_SymbolIterator {
263 SymbolClosure* const _closure;
264 public:
265 CompactHashtable_SymbolIterator(SymbolClosure *cl) : _closure(cl) {}
266 inline void do_value(address base_address, u4 offset) const {
267 Symbol* sym = (Symbol*)((void*)(base_address + offset));
268 _closure->do_symbol(&sym);
269 }
270 };
271
272 template <class T, class N> void CompactHashtable<T, N>::symbols_do(SymbolClosure *cl) {
273 CompactHashtable_SymbolIterator iterator(cl);
274 iterate(iterator);
275 }
276
277 class CompactHashtable_OopIterator {
278 OopClosure* const _closure;
279 public:
280 CompactHashtable_OopIterator(OopClosure *cl) : _closure(cl) {}
281 inline void do_value(address base_address, u4 offset) const {
282 narrowOop o = (narrowOop)offset;
283 _closure->do_oop(&o);
284 }
285 };
286
287 template <class T, class N> void CompactHashtable<T, N>::oops_do(OopClosure* cl) {
288 assert(_type == _string_table || _bucket_count == 0, "sanity");
289 CompactHashtable_OopIterator iterator(cl);
290 iterate(iterator);
291 }
292
293 // Explicitly instantiate these types
294 template class CompactHashtable<Symbol*, char>;
295 template class CompactHashtable<oop, char>;
296
297 #ifndef O_BINARY // if defined (Win32) use binary files.
298 #define O_BINARY 0 // otherwise do nothing.
299 #endif
300
301 ////////////////////////////////////////////////////////
302 //
303 // HashtableTextDump
304 //
305 HashtableTextDump::HashtableTextDump(const char* filename) : _fd(-1) {
306 struct stat st;
307 if (os::stat(filename, &st) != 0) {
308 quit("Unable to get hashtable dump file size", filename);
309 }
310 _size = st.st_size;
311 _fd = open(filename, O_RDONLY | O_BINARY, 0);
312 if (_fd < 0) {
313 quit("Unable to open hashtable dump file", filename);
314 }
315 _base = os::map_memory(_fd, filename, 0, NULL, _size, true, false);
316 if (_base == NULL) {
317 quit("Unable to map hashtable dump file", filename);
318 }
319 _p = _base;
320 _end = _base + st.st_size;
321 _filename = filename;
322 _prefix_type = Unknown;
323 _line_no = 1;
324 }
325
326 HashtableTextDump::~HashtableTextDump() {
327 os::unmap_memory((char*)_base, _size);
328 if (_fd >= 0) {
329 close(_fd);
330 }
331 }
332
333 void HashtableTextDump::quit(const char* err, const char* msg) {
334 vm_exit_during_initialization(err, msg);
335 }
336
337 void HashtableTextDump::corrupted(const char *p, const char* msg) {
338 char info[100];
339 jio_snprintf(info, sizeof(info),
340 "%s. Corrupted at line %d (file pos %d)",
341 msg, _line_no, (int)(p - _base));
342 quit(info, _filename);
343 }
344
345 bool HashtableTextDump::skip_newline() {
346 if (_p[0] == '\r' && _p[1] == '\n') {
347 _p += 2;
348 } else if (_p[0] == '\n') {
349 _p += 1;
350 } else {
351 corrupted(_p, "Unexpected character");
352 }
353 _line_no++;
354 return true;
355 }
356
357 int HashtableTextDump::skip(char must_be_char) {
358 corrupted_if(remain() < 1, "Truncated");
359 corrupted_if(*_p++ != must_be_char, "Unexpected character");
360 return 0;
361 }
362
363 void HashtableTextDump::skip_past(char c) {
364 for (;;) {
365 corrupted_if(remain() < 1, "Truncated");
366 if (*_p++ == c) {
367 return;
368 }
369 }
370 }
371
372 void HashtableTextDump::check_version(const char* ver) {
373 int len = (int)strlen(ver);
374 corrupted_if(remain() < len, "Truncated");
375 if (strncmp(_p, ver, len) != 0) {
376 quit("wrong version of hashtable dump file", _filename);
377 }
378 _p += len;
379 skip_newline();
380 }
381
382 void HashtableTextDump::scan_prefix_type() {
383 _p++;
384 if (strncmp(_p, "SECTION: String", 15) == 0) {
385 _p += 15;
386 _prefix_type = StringPrefix;
387 } else if (strncmp(_p, "SECTION: Symbol", 15) == 0) {
388 _p += 15;
389 _prefix_type = SymbolPrefix;
390 } else {
391 _prefix_type = Unknown;
392 }
393 skip_newline();
394 }
395
396 int HashtableTextDump::scan_prefix(int* utf8_length) {
397 if (*_p == '@') {
398 scan_prefix_type();
399 }
400
401 switch (_prefix_type) {
402 case SymbolPrefix:
403 *utf8_length = scan_symbol_prefix(); break;
404 case StringPrefix:
405 *utf8_length = scan_string_prefix(); break;
406 default:
407 tty->print_cr("Shared input data type: Unknown.");
408 corrupted(_p, "Unknown data type");
409 }
410
411 return _prefix_type;
412 }
413
414 int HashtableTextDump::scan_string_prefix() {
415 // Expect /[0-9]+: /
416 int utf8_length = 0;
417 get_num(':', &utf8_length);
418 if (*_p != ' ') {
419 corrupted(_p, "Wrong prefix format for string");
420 }
421 _p++;
422 return utf8_length;
423 }
424
425 int HashtableTextDump::scan_symbol_prefix() {
426 // Expect /[0-9]+ (-|)[0-9]+: /
427 int utf8_length = 0;
428 get_num(' ', &utf8_length);
429 if (*_p == '-') {
430 _p++;
431 }
432 int ref_num;
433 get_num(':', &ref_num);
434 if (*_p != ' ') {
435 corrupted(_p, "Wrong prefix format for symbol");
436 }
437 _p++;
438 return utf8_length;
439 }
440
441 jchar HashtableTextDump::unescape(const char* from, const char* end, int count) {
442 jchar value = 0;
443
444 corrupted_if(from + count > end, "Truncated");
445
446 for (int i=0; i<count; i++) {
447 char c = *from++;
448 switch (c) {
449 case '0': case '1': case '2': case '3': case '4':
450 case '5': case '6': case '7': case '8': case '9':
451 value = (value << 4) + c - '0';
452 break;
453 case 'a': case 'b': case 'c':
454 case 'd': case 'e': case 'f':
455 value = (value << 4) + 10 + c - 'a';
456 break;
457 case 'A': case 'B': case 'C':
458 case 'D': case 'E': case 'F':
459 value = (value << 4) + 10 + c - 'A';
460 break;
461 default:
462 ShouldNotReachHere();
463 }
464 }
465 return value;
466 }
467
468 void HashtableTextDump::get_utf8(char* utf8_buffer, int utf8_length) {
469 // cache in local vars
470 const char* from = _p;
471 const char* end = _end;
472 char* to = utf8_buffer;
473 int n = utf8_length;
474
475 for (; n > 0 && from < end; n--) {
476 if (*from != '\\') {
477 *to++ = *from++;
478 } else {
479 corrupted_if(from + 2 > end, "Truncated");
480 char c = from[1];
481 from += 2;
482 switch (c) {
483 case 'x':
484 {
485 jchar value = unescape(from, end, 2);
486 from += 2;
487 assert(value <= 0xff, "sanity");
488 *to++ = (char)(value & 0xff);
489 }
490 break;
491 case 't': *to++ = '\t'; break;
492 case 'n': *to++ = '\n'; break;
493 case 'r': *to++ = '\r'; break;
494 case '\\': *to++ = '\\'; break;
495 default:
496 corrupted(_p, "Unsupported character");
497 }
498 }
499 }
500 corrupted_if(n > 0, "Truncated"); // expected more chars but file has ended
501 _p = from;
502 skip_newline();
503 }
504
505 // NOTE: the content is NOT the same as
506 // UTF8::as_quoted_ascii(const char* utf8_str, int utf8_length, char* buf, int buflen).
507 // We want to escape \r\n\t so that output [1] is more readable; [2] can be more easily
508 // parsed by scripts; [3] quickly processed by HashtableTextDump::get_utf8()
509 void HashtableTextDump::put_utf8(outputStream* st, const char* utf8_string, int utf8_length) {
510 const char *c = utf8_string;
511 const char *end = c + utf8_length;
512 for (; c < end; c++) {
513 switch (*c) {
514 case '\t': st->print("\\t"); break;
515 case '\r': st->print("\\r"); break;
516 case '\n': st->print("\\n"); break;
517 case '\\': st->print("\\\\"); break;
518 default:
519 if (isprint(*c)) {
520 st->print("%c", *c);
521 } else {
522 st->print("\\x%02x", ((unsigned int)*c) & 0xff);
523 }
524 }
525 }
526 }