1/*
2 * Copyright (c) 1997, 2019, 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.hpp"
28#include "classfile/javaClasses.hpp"
29#include "logging/logMessage.hpp"
30#include "memory/dynamicArchive.hpp"
31#include "memory/heapShared.inline.hpp"
32#include "memory/metadataFactory.hpp"
33#include "memory/metaspaceShared.hpp"
34#include "runtime/vmThread.hpp"
35#include "utilities/numberSeq.hpp"
36#include <sys/stat.h>
37
38#if INCLUDE_CDS
39/////////////////////////////////////////////////////
40//
41// The compact hash table writer implementations
42//
43CompactHashtableWriter::CompactHashtableWriter(int num_entries,
44 CompactHashtableStats* stats) {
45 assert(DumpSharedSpaces || DynamicDumpSharedSpaces, "dump-time only");
46 assert(num_entries >= 0, "sanity");
47 _num_buckets = calculate_num_buckets(num_entries);
48 assert(_num_buckets > 0, "no buckets");
49
50 _num_entries_written = 0;
51 _buckets = NEW_C_HEAP_ARRAY(GrowableArray<Entry>*, _num_buckets, mtSymbol);
52 for (int i=0; i<_num_buckets; i++) {
53 _buckets[i] = new (ResourceObj::C_HEAP, mtSymbol) GrowableArray<Entry>(0, true, mtSymbol);
54 }
55
56 _stats = stats;
57 _compact_buckets = NULL;
58 _compact_entries = NULL;
59 _num_empty_buckets = 0;
60 _num_value_only_buckets = 0;
61 _num_other_buckets = 0;
62}
63
64CompactHashtableWriter::~CompactHashtableWriter() {
65 for (int index = 0; index < _num_buckets; index++) {
66 GrowableArray<Entry>* bucket = _buckets[index];
67 delete bucket;
68 }
69
70 FREE_C_HEAP_ARRAY(GrowableArray<Entry>*, _buckets);
71}
72
73size_t CompactHashtableWriter::estimate_size(int num_entries) {
74 int num_buckets = calculate_num_buckets(num_entries);
75 size_t bucket_bytes = MetaspaceShared::ro_array_bytesize<u4>(num_buckets + 1);
76
77 // In worst case, we have no VALUE_ONLY_BUCKET_TYPE, so each entry takes 2 slots
78 int entries_space = 2 * num_entries;
79 size_t entry_bytes = MetaspaceShared::ro_array_bytesize<u4>(entries_space);
80
81 return bucket_bytes
82 + entry_bytes
83 + SimpleCompactHashtable::calculate_header_size();
84}
85
86// Add a symbol entry to the temporary hash table
87void CompactHashtableWriter::add(unsigned int hash, u4 value) {
88 int index = hash % _num_buckets;
89 _buckets[index]->append_if_missing(Entry(hash, value));
90 _num_entries_written++;
91}
92
93void CompactHashtableWriter::allocate_table() {
94 int entries_space = 0;
95 for (int index = 0; index < _num_buckets; index++) {
96 GrowableArray<Entry>* bucket = _buckets[index];
97 int bucket_size = bucket->length();
98 if (bucket_size == 1) {
99 entries_space++;
100 } else if (bucket_size > 1) {
101 entries_space += 2 * bucket_size;
102 }
103 }
104
105 if (entries_space & ~BUCKET_OFFSET_MASK) {
106 vm_exit_during_initialization("CompactHashtableWriter::allocate_table: Overflow! "
107 "Too many entries.");
108 }
109
110 _compact_buckets = MetaspaceShared::new_ro_array<u4>(_num_buckets + 1);
111 _compact_entries = MetaspaceShared::new_ro_array<u4>(entries_space);
112
113 _stats->bucket_count = _num_buckets;
114 _stats->bucket_bytes = _compact_buckets->size() * BytesPerWord;
115 _stats->hashentry_count = _num_entries_written;
116 _stats->hashentry_bytes = _compact_entries->size() * BytesPerWord;
117}
118
119// Write the compact table's buckets
120void CompactHashtableWriter::dump_table(NumberSeq* summary) {
121 u4 offset = 0;
122 for (int index = 0; index < _num_buckets; index++) {
123 GrowableArray<Entry>* bucket = _buckets[index];
124 int bucket_size = bucket->length();
125 if (bucket_size == 1) {
126 // bucket with one entry is compacted and only has the symbol offset
127 _compact_buckets->at_put(index, BUCKET_INFO(offset, VALUE_ONLY_BUCKET_TYPE));
128
129 Entry ent = bucket->at(0);
130 _compact_entries->at_put(offset++, ent.value());
131 _num_value_only_buckets++;
132 } else {
133 // regular bucket, each entry is a symbol (hash, offset) pair
134 _compact_buckets->at_put(index, BUCKET_INFO(offset, REGULAR_BUCKET_TYPE));
135
136 for (int i=0; i<bucket_size; i++) {
137 Entry ent = bucket->at(i);
138 _compact_entries->at_put(offset++, u4(ent.hash())); // write entry hash
139 _compact_entries->at_put(offset++, ent.value());
140 }
141 if (bucket_size == 0) {
142 _num_empty_buckets++;
143 } else {
144 _num_other_buckets++;
145 }
146 }
147 summary->add(bucket_size);
148 }
149
150 // Mark the end of the buckets
151 _compact_buckets->at_put(_num_buckets, BUCKET_INFO(offset, TABLEEND_BUCKET_TYPE));
152 assert(offset == (u4)_compact_entries->length(), "sanity");
153}
154
155
156// Write the compact table
157void CompactHashtableWriter::dump(SimpleCompactHashtable *cht, const char* table_name) {
158 NumberSeq summary;
159 allocate_table();
160 dump_table(&summary);
161
162 int table_bytes = _stats->bucket_bytes + _stats->hashentry_bytes;
163 address base_address = address(SharedBaseAddress);
164 cht->init(base_address, _num_entries_written, _num_buckets,
165 _compact_buckets->data(), _compact_entries->data());
166
167 LogMessage(cds, hashtables) msg;
168 if (msg.is_info()) {
169 double avg_cost = 0.0;
170 if (_num_entries_written > 0) {
171 avg_cost = double(table_bytes)/double(_num_entries_written);
172 }
173 msg.info("Shared %s table stats -------- base: " PTR_FORMAT,
174 table_name, (intptr_t)base_address);
175 msg.info("Number of entries : %9d", _num_entries_written);
176 msg.info("Total bytes used : %9d", table_bytes);
177 msg.info("Average bytes per entry : %9.3f", avg_cost);
178 msg.info("Average bucket size : %9.3f", summary.avg());
179 msg.info("Variance of bucket size : %9.3f", summary.variance());
180 msg.info("Std. dev. of bucket size: %9.3f", summary.sd());
181 msg.info("Maximum bucket size : %9d", (int)summary.maximum());
182 msg.info("Empty buckets : %9d", _num_empty_buckets);
183 msg.info("Value_Only buckets : %9d", _num_value_only_buckets);
184 msg.info("Other buckets : %9d", _num_other_buckets);
185 }
186}
187
188/////////////////////////////////////////////////////////////
189//
190// The CompactHashtable implementation
191//
192
193void SimpleCompactHashtable::init(address base_address, u4 entry_count, u4 bucket_count, u4* buckets, u4* entries) {
194 _bucket_count = bucket_count;
195 _entry_count = entry_count;
196 _base_address = base_address;
197 if (DynamicDumpSharedSpaces) {
198 _buckets = DynamicArchive::buffer_to_target(buckets);
199 _entries = DynamicArchive::buffer_to_target(entries);
200 } else {
201 _buckets = buckets;
202 _entries = entries;
203 }
204}
205
206size_t SimpleCompactHashtable::calculate_header_size() {
207 // We have 5 fields. Each takes up sizeof(intptr_t). See WriteClosure::do_u4
208 size_t bytes = sizeof(intptr_t) * 5;
209 return bytes;
210}
211
212void SimpleCompactHashtable::serialize_header(SerializeClosure* soc) {
213 // NOTE: if you change this function, you MUST change the number 5 in
214 // calculate_header_size() accordingly.
215 soc->do_ptr((void**)&_base_address);
216 soc->do_u4(&_entry_count);
217 soc->do_u4(&_bucket_count);
218 soc->do_ptr((void**)&_buckets);
219 soc->do_ptr((void**)&_entries);
220}
221#endif // INCLUDE_CDS
222
223#ifndef O_BINARY // if defined (Win32) use binary files.
224#define O_BINARY 0 // otherwise do nothing.
225#endif
226
227////////////////////////////////////////////////////////
228//
229// HashtableTextDump
230//
231HashtableTextDump::HashtableTextDump(const char* filename) : _fd(-1) {
232 struct stat st;
233 if (os::stat(filename, &st) != 0) {
234 quit("Unable to get hashtable dump file size", filename);
235 }
236 _size = st.st_size;
237 _fd = os::open(filename, O_RDONLY | O_BINARY, 0);
238 if (_fd < 0) {
239 quit("Unable to open hashtable dump file", filename);
240 }
241 _base = os::map_memory(_fd, filename, 0, NULL, _size, true, false);
242 if (_base == NULL) {
243 quit("Unable to map hashtable dump file", filename);
244 }
245 _p = _base;
246 _end = _base + st.st_size;
247 _filename = filename;
248 _prefix_type = Unknown;
249 _line_no = 1;
250}
251
252HashtableTextDump::~HashtableTextDump() {
253 os::unmap_memory((char*)_base, _size);
254 if (_fd >= 0) {
255 close(_fd);
256 }
257}
258
259void HashtableTextDump::quit(const char* err, const char* msg) {
260 vm_exit_during_initialization(err, msg);
261}
262
263void HashtableTextDump::corrupted(const char *p, const char* msg) {
264 char info[100];
265 jio_snprintf(info, sizeof(info),
266 "%s. Corrupted at line %d (file pos %d)",
267 msg, _line_no, (int)(p - _base));
268 quit(info, _filename);
269}
270
271bool HashtableTextDump::skip_newline() {
272 if (_p[0] == '\r' && _p[1] == '\n') {
273 _p += 2;
274 } else if (_p[0] == '\n') {
275 _p += 1;
276 } else {
277 corrupted(_p, "Unexpected character");
278 }
279 _line_no++;
280 return true;
281}
282
283int HashtableTextDump::skip(char must_be_char) {
284 corrupted_if(remain() < 1, "Truncated");
285 corrupted_if(*_p++ != must_be_char, "Unexpected character");
286 return 0;
287}
288
289void HashtableTextDump::skip_past(char c) {
290 for (;;) {
291 corrupted_if(remain() < 1, "Truncated");
292 if (*_p++ == c) {
293 return;
294 }
295 }
296}
297
298void HashtableTextDump::check_version(const char* ver) {
299 int len = (int)strlen(ver);
300 corrupted_if(remain() < len, "Truncated");
301 if (strncmp(_p, ver, len) != 0) {
302 quit("wrong version of hashtable dump file", _filename);
303 }
304 _p += len;
305 skip_newline();
306}
307
308void HashtableTextDump::scan_prefix_type() {
309 _p++;
310 if (strncmp(_p, "SECTION: String", 15) == 0) {
311 _p += 15;
312 _prefix_type = StringPrefix;
313 } else if (strncmp(_p, "SECTION: Symbol", 15) == 0) {
314 _p += 15;
315 _prefix_type = SymbolPrefix;
316 } else {
317 _prefix_type = Unknown;
318 }
319 skip_newline();
320}
321
322int HashtableTextDump::scan_prefix(int* utf8_length) {
323 if (*_p == '@') {
324 scan_prefix_type();
325 }
326
327 switch (_prefix_type) {
328 case SymbolPrefix:
329 *utf8_length = scan_symbol_prefix(); break;
330 case StringPrefix:
331 *utf8_length = scan_string_prefix(); break;
332 default:
333 tty->print_cr("Shared input data type: Unknown.");
334 corrupted(_p, "Unknown data type");
335 }
336
337 return _prefix_type;
338}
339
340int HashtableTextDump::scan_string_prefix() {
341 // Expect /[0-9]+: /
342 int utf8_length = 0;
343 get_num(':', &utf8_length);
344 if (*_p != ' ') {
345 corrupted(_p, "Wrong prefix format for string");
346 }
347 _p++;
348 return utf8_length;
349}
350
351int HashtableTextDump::scan_symbol_prefix() {
352 // Expect /[0-9]+ (-|)[0-9]+: /
353 int utf8_length = 0;
354 get_num(' ', &utf8_length);
355 if (*_p == '-') {
356 _p++;
357 }
358 int ref_num;
359 get_num(':', &ref_num);
360 if (*_p != ' ') {
361 corrupted(_p, "Wrong prefix format for symbol");
362 }
363 _p++;
364 return utf8_length;
365}
366
367jchar HashtableTextDump::unescape(const char* from, const char* end, int count) {
368 jchar value = 0;
369
370 corrupted_if(from + count > end, "Truncated");
371
372 for (int i=0; i<count; i++) {
373 char c = *from++;
374 switch (c) {
375 case '0': case '1': case '2': case '3': case '4':
376 case '5': case '6': case '7': case '8': case '9':
377 value = (value << 4) + c - '0';
378 break;
379 case 'a': case 'b': case 'c':
380 case 'd': case 'e': case 'f':
381 value = (value << 4) + 10 + c - 'a';
382 break;
383 case 'A': case 'B': case 'C':
384 case 'D': case 'E': case 'F':
385 value = (value << 4) + 10 + c - 'A';
386 break;
387 default:
388 ShouldNotReachHere();
389 }
390 }
391 return value;
392}
393
394void HashtableTextDump::get_utf8(char* utf8_buffer, int utf8_length) {
395 // cache in local vars
396 const char* from = _p;
397 const char* end = _end;
398 char* to = utf8_buffer;
399 int n = utf8_length;
400
401 for (; n > 0 && from < end; n--) {
402 if (*from != '\\') {
403 *to++ = *from++;
404 } else {
405 corrupted_if(from + 2 > end, "Truncated");
406 char c = from[1];
407 from += 2;
408 switch (c) {
409 case 'x':
410 {
411 jchar value = unescape(from, end, 2);
412 from += 2;
413 assert(value <= 0xff, "sanity");
414 *to++ = (char)(value & 0xff);
415 }
416 break;
417 case 't': *to++ = '\t'; break;
418 case 'n': *to++ = '\n'; break;
419 case 'r': *to++ = '\r'; break;
420 case '\\': *to++ = '\\'; break;
421 default:
422 corrupted(_p, "Unsupported character");
423 }
424 }
425 }
426 corrupted_if(n > 0, "Truncated"); // expected more chars but file has ended
427 _p = from;
428 skip_newline();
429}
430
431// NOTE: the content is NOT the same as
432// UTF8::as_quoted_ascii(const char* utf8_str, int utf8_length, char* buf, int buflen).
433// We want to escape \r\n\t so that output [1] is more readable; [2] can be more easily
434// parsed by scripts; [3] quickly processed by HashtableTextDump::get_utf8()
435void HashtableTextDump::put_utf8(outputStream* st, const char* utf8_string, int utf8_length) {
436 const char *c = utf8_string;
437 const char *end = c + utf8_length;
438 for (; c < end; c++) {
439 switch (*c) {
440 case '\t': st->print("\\t"); break;
441 case '\r': st->print("\\r"); break;
442 case '\n': st->print("\\n"); break;
443 case '\\': st->print("\\\\"); break;
444 default:
445 if (isprint(*c)) {
446 st->print("%c", *c);
447 } else {
448 st->print("\\x%02x", ((unsigned int)*c) & 0xff);
449 }
450 }
451 }
452}
453