1 /*
2 * Copyright (c) 2015, 2023, Oracle and/or its affiliates. All rights reserved.
3 *
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions
6 * are met:
7 *
8 * - Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 *
11 * - Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 *
15 * - Neither the name of Oracle nor the names of its
16 * contributors may be used to endorse or promote products derived
17 * from this software without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS
20 * IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
21 * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
23 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
24 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
25 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
26 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
27 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
28 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
29 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30 */
31
32 #include <assert.h>
33 #include <string.h>
34 #include <stdlib.h>
35
36 #include "endian.hpp"
37 #include "imageDecompressor.hpp"
38 #include "imageFile.hpp"
39 #include "inttypes.hpp"
40 #include "jni.h"
41 #include "osSupport.hpp"
42
43 // Map the full jimage, only with 64 bit addressing.
44 bool ImageFileReader::memory_map_image = sizeof(void *) == 8;
45
46 #ifdef WIN32
47 const char FileSeparator = '\\';
48 #else
49 const char FileSeparator = '/';
50 #endif
51
52 // Image files are an alternate file format for storing classes and resources. The
53 // goal is to supply file access which is faster and smaller than the jar format.
54 //
55 // (More detailed nodes in the header.)
56 //
57
58 // Compute the Perfect Hashing hash code for the supplied UTF-8 string.
59 s4 ImageStrings::hash_code(const char* string, s4 seed) {
60 assert(seed > 0 && "invariant");
61 // Access bytes as unsigned.
62 u1* bytes = (u1*)string;
63 u4 useed = (u4)seed;
64 // Compute hash code.
65 for (u1 byte = *bytes++; byte; byte = *bytes++) {
66 useed = (useed * HASH_MULTIPLIER) ^ byte;
67 }
68 // Ensure the result is not signed.
69 return (s4)(useed & 0x7FFFFFFF);
70 }
71
72 // Match up a string in a perfect hash table.
73 // Returns the index where the name should be.
74 // Result still needs validation for precise match (false positive.)
75 s4 ImageStrings::find(Endian* endian, const char* name, s4* redirect, u4 length) {
76 // If the table is empty, then short cut.
77 if (!redirect || !length) {
78 return NOT_FOUND;
79 }
80 // Compute the basic perfect hash for name.
81 s4 hash_code = ImageStrings::hash_code(name);
82 // Modulo table size.
83 s4 index = hash_code % length;
84 // Get redirect entry.
85 // value == 0 then not found
86 // value < 0 then -1 - value is true index
87 // value > 0 then value is seed for recomputing hash.
88 s4 value = endian->get(redirect[index]);
89 // if recompute is required.
90 if (value > 0 ) {
91 // Entry collision value, need to recompute hash.
92 hash_code = ImageStrings::hash_code(name, value);
93 // Modulo table size.
94 return hash_code % length;
95 } else if (value < 0) {
96 // Compute direct index.
97 return -1 - value;
98 }
99 // No entry found.
100 return NOT_FOUND;
101 }
102
103 // Test to see if UTF-8 string begins with the start UTF-8 string. If so,
104 // return non-NULL address of remaining portion of string. Otherwise, return
105 // NULL. Used to test sections of a path without copying from image string
106 // table.
107 const char* ImageStrings::starts_with(const char* string, const char* start) {
108 char ch1, ch2;
109 // Match up the strings the best we can.
110 while ((ch1 = *string) && (ch2 = *start)) {
111 if (ch1 != ch2) {
112 // Mismatch, return NULL.
113 return NULL;
114 }
115 // Next characters.
116 string++, start++;
117 }
118 // Return remainder of string.
119 return string;
120 }
121
122 // Inflates the attribute stream into individual values stored in the long
123 // array _attributes. This allows an attribute value to be quickly accessed by
124 // direct indexing. Unspecified values default to zero (from constructor.)
125 void ImageLocation::set_data(u1* data) {
126 // Deflate the attribute stream into an array of attributes.
127 u1 byte;
128 // Repeat until end header is found.
129 while ((data != NULL) && (byte = *data)) {
130 // Extract kind from header byte.
131 u1 kind = attribute_kind(byte);
132 assert(kind < ATTRIBUTE_COUNT && "invalid image location attribute");
133 if (kind == ATTRIBUTE_END) {
134 break;
135 }
136 // Extract length of data (in bytes).
137 u1 n = attribute_length(byte);
138 // Read value (most significant first.)
139 _attributes[kind] = attribute_value(data + 1, n);
140 // Position to next attribute by skipping attribute header and data bytes.
141 data += n + 1;
142 }
143 }
144
145 // Zero all attribute values.
146 void ImageLocation::clear_data() {
147 // Set defaults to zero.
148 memset(_attributes, 0, sizeof(_attributes));
149 }
150
151 // ImageModuleData constructor maps out sub-tables for faster access.
152 ImageModuleData::ImageModuleData(const ImageFileReader* image_file) :
153 _image_file(image_file),
154 _endian(image_file->endian()) {
155 }
156
157 // Release module data resource.
158 ImageModuleData::~ImageModuleData() {
159 }
160
161
162 // Return the module in which a package resides. Returns NULL if not found.
163 const char* ImageModuleData::package_to_module(const char* package_name) {
164 // replace all '/' by '.'
165 char* replaced = new char[(int) strlen(package_name) + 1];
166 assert(replaced != NULL && "allocation failed");
167 int i;
168 for (i = 0; package_name[i] != '\0'; i++) {
169 replaced[i] = package_name[i] == '/' ? '.' : package_name[i];
170 }
171 replaced[i] = '\0';
172
173 // build path /packages/<package_name>
174 const char* radical = "/packages/";
175 char* path = new char[(int) strlen(radical) + (int) strlen(package_name) + 1];
176 assert(path != NULL && "allocation failed");
177 strcpy(path, radical);
178 strcat(path, replaced);
179 delete[] replaced;
180
181 // retrieve package location
182 ImageLocation location;
183 bool found = _image_file->find_location(path, location);
184 delete[] path;
185 if (!found) {
186 return NULL;
187 }
188
189 // retrieve offsets to module name
190 int size = (int)location.get_attribute(ImageLocation::ATTRIBUTE_UNCOMPRESSED);
191 u1* content = new u1[size];
192 assert(content != NULL && "allocation failed");
193 _image_file->get_resource(location, content);
194 u1* ptr = content;
195 // sequence of sizeof(8) isEmpty|offset. Use the first module that is not empty.
196 u4 offset = 0;
197 for (i = 0; i < size; i+=8) {
198 u4 isEmpty = _endian->get(*((u4*)ptr));
199 ptr += 4;
200 if (!isEmpty) {
201 offset = _endian->get(*((u4*)ptr));
202 break;
203 }
204 ptr += 4;
205 }
206 delete[] content;
207 return _image_file->get_strings().get(offset);
208 }
209
210 // Manage a table of open image files. This table allows multiple access points
211 // to share an open image.
212 ImageFileReaderTable::ImageFileReaderTable() : _count(0), _max(_growth) {
213 _table = static_cast<ImageFileReader**>(calloc(_max, sizeof(ImageFileReader*)));
214 assert(_table != NULL && "allocation failed");
215 }
216
217 // Add a new image entry to the table.
218 void ImageFileReaderTable::add(ImageFileReader* image) {
219 if (_count == _max) {
220 _max += _growth;
221 _table = static_cast<ImageFileReader**>(realloc(_table, _max * sizeof(ImageFileReader*)));
222 }
223 _table[_count++] = image;
224 }
225
226 // Remove an image entry from the table.
227 void ImageFileReaderTable::remove(ImageFileReader* image) {
228 for (u4 i = 0; i < _count; i++) {
229 if (_table[i] == image) {
230 // Swap the last element into the found slot
231 _table[i] = _table[--_count];
232 break;
233 }
234 }
235
236 if (_count != 0 && _count == _max - _growth) {
237 _max -= _growth;
238 _table = static_cast<ImageFileReader**>(realloc(_table, _max * sizeof(ImageFileReader*)));
239 }
240 }
241
242 // Determine if image entry is in table.
243 bool ImageFileReaderTable::contains(ImageFileReader* image) {
244 for (u4 i = 0; i < _count; i++) {
245 if (_table[i] == image) {
246 return true;
247 }
248 }
249 return false;
250 }
251
252 // Table to manage multiple opens of an image file.
253 ImageFileReaderTable ImageFileReader::_reader_table;
254
255 SimpleCriticalSection _reader_table_lock;
256
257 // Locate an image if file already open.
258 ImageFileReader* ImageFileReader::find_image(const char* name) {
259 // Lock out _reader_table.
260 SimpleCriticalSectionLock cs(&_reader_table_lock);
261 // Search for an exist image file.
262 for (u4 i = 0; i < _reader_table.count(); i++) {
263 // Retrieve table entry.
264 ImageFileReader* reader = _reader_table.get(i);
265 // If name matches, then reuse (bump up use count.)
266 assert(reader->name() != NULL && "reader->name must not be null");
267 if (strcmp(reader->name(), name) == 0) {
268 reader->inc_use();
269 return reader;
270 }
271 }
272
273 return NULL;
274 }
275
276 // Open an image file, reuse structure if file already open.
277 ImageFileReader* ImageFileReader::open(const char* name, bool big_endian) {
278 ImageFileReader* reader = find_image(name);
279 if (reader != NULL) {
280 return reader;
281 }
282
283 // Need a new image reader.
284 reader = new ImageFileReader(name, big_endian);
285 if (reader == NULL || !reader->open()) {
286 // Failed to open.
287 delete reader;
288 return NULL;
289 }
290
291 // Lock to update
292 SimpleCriticalSectionLock cs(&_reader_table_lock);
293 // Search for an existing image file.
294 for (u4 i = 0; i < _reader_table.count(); i++) {
295 // Retrieve table entry.
296 ImageFileReader* existing_reader = _reader_table.get(i);
297 // If name matches, then reuse (bump up use count.)
298 assert(reader->name() != NULL && "reader->name still must not be null");
299 if (strcmp(existing_reader->name(), name) == 0) {
300 existing_reader->inc_use();
301 reader->close();
302 delete reader;
303 return existing_reader;
304 }
305 }
306 // Bump use count and add to table.
307 reader->inc_use();
308 _reader_table.add(reader);
309 return reader;
310 }
311
312 // Close an image file if the file is not in use elsewhere.
313 void ImageFileReader::close(ImageFileReader *reader) {
314 // Lock out _reader_table.
315 SimpleCriticalSectionLock cs(&_reader_table_lock);
316 // If last use then remove from table and then close.
317 if (reader->dec_use()) {
318 _reader_table.remove(reader);
319 delete reader;
320 }
321 }
322
323 // Return an id for the specified ImageFileReader.
324 u8 ImageFileReader::reader_to_ID(ImageFileReader *reader) {
325 // ID is just the cloaked reader address.
326 return (u8)reader;
327 }
328
329 // Validate the image id.
330 bool ImageFileReader::id_check(u8 id) {
331 // Make sure the ID is a managed (_reader_table) reader.
332 SimpleCriticalSectionLock cs(&_reader_table_lock);
333 return _reader_table.contains((ImageFileReader*)id);
334 }
335
336 // Return an id for the specified ImageFileReader.
337 ImageFileReader* ImageFileReader::id_to_reader(u8 id) {
338 assert(id_check(id) && "invalid image id");
339 return (ImageFileReader*)id;
340 }
341
342 // Constructor initializes to a closed state.
343 ImageFileReader::ImageFileReader(const char* name, bool big_endian) :
344 _module_data(NULL) {
345 // Copy the image file name.
346 int len = (int) strlen(name) + 1;
347 _name = new char[len];
348 assert(_name != NULL && "allocation failed");
349 strncpy(_name, name, len);
350 // Initialize for a closed file.
351 _fd = -1;
352 _endian = Endian::get_handler(big_endian);
353 _index_data = NULL;
354 }
355
356 // Close image and free up data structures.
357 ImageFileReader::~ImageFileReader() {
358 // Ensure file is closed.
359 close();
360 // Free up name.
361 if (_name) {
362 delete[] _name;
363 _name = NULL;
364 }
365
366 if (_module_data != NULL) {
367 delete _module_data;
368 }
369 }
370
371 // Open image file for read access.
372 bool ImageFileReader::open() {
373 // If file exists open for reading.
374 _fd = osSupport::openReadOnly(_name);
375 if (_fd == -1) {
376 return false;
377 }
378 // Retrieve the file size.
379 _file_size = osSupport::size(_name);
380 // Read image file header and verify it has a valid header.
381 size_t header_size = sizeof(ImageHeader);
382 if (_file_size < header_size ||
383 !read_at((u1*)&_header, header_size, 0) ||
384 _header.magic(_endian) != IMAGE_MAGIC ||
385 _header.major_version(_endian) != MAJOR_VERSION ||
386 _header.minor_version(_endian) != MINOR_VERSION) {
387 close();
388 return false;
389 }
390 // Size of image index.
391 _index_size = index_size();
392 // Make sure file is large enough to contain the index.
393 if (_file_size < _index_size) {
394 return false;
395 }
396 // Memory map image (minimally the index.)
397 _index_data = (u1*)osSupport::map_memory(_fd, _name, 0, (size_t)map_size());
398 assert(_index_data && "image file not memory mapped");
399 // Retrieve length of index perfect hash table.
400 u4 length = table_length();
401 // Compute offset of the perfect hash table redirect table.
402 u4 redirect_table_offset = (u4)header_size;
403 // Compute offset of index attribute offsets.
404 u4 offsets_table_offset = redirect_table_offset + length * (u4)sizeof(s4);
405 // Compute offset of index location attribute data.
406 u4 location_bytes_offset = offsets_table_offset + length * (u4)sizeof(u4);
407 // Compute offset of index string table.
408 u4 string_bytes_offset = location_bytes_offset + locations_size();
409 // Compute address of the perfect hash table redirect table.
410 _redirect_table = (s4*)(_index_data + redirect_table_offset);
411 // Compute address of index attribute offsets.
412 _offsets_table = (u4*)(_index_data + offsets_table_offset);
413 // Compute address of index location attribute data.
414 _location_bytes = _index_data + location_bytes_offset;
415 // Compute address of index string table.
416 _string_bytes = _index_data + string_bytes_offset;
417
418 // Initialize the module data
419 _module_data = new ImageModuleData(this);
420 // Successful open (if memory allocation succeeded).
421 return _module_data != NULL;
422 }
423
424 // Close image file.
425 void ImageFileReader::close() {
426 // Deallocate the index.
427 if (_index_data) {
428 osSupport::unmap_memory((char*)_index_data, (size_t)map_size());
429 _index_data = NULL;
430 }
431 // Close file.
432 if (_fd != -1) {
433 osSupport::close(_fd);
434 _fd = -1;
435 }
436
437 if (_module_data != NULL) {
438 delete _module_data;
439 _module_data = NULL;
440 }
441 }
442
443 // Read directly from the file.
444 bool ImageFileReader::read_at(u1* data, u8 size, u8 offset) const {
445 return (u8)osSupport::read(_fd, (char*)data, size, offset) == size;
446 }
447
448 // Find the location attributes associated with the path. Returns true if
449 // the location is found, false otherwise.
450 bool ImageFileReader::find_location(const char* path, ImageLocation& location) const {
451 // Locate the entry in the index perfect hash table.
452 s4 index = ImageStrings::find(_endian, path, _redirect_table, table_length());
453 // If is found.
454 if (index != ImageStrings::NOT_FOUND) {
455 // Get address of first byte of location attribute stream.
456 u1* data = get_location_data(index);
457 // Expand location attributes.
458 location.set_data(data);
459 // Make sure result is not a false positive.
460 return verify_location(location, path);
461 }
462 return false;
463 }
464
465 // Find the location index and size associated with the path.
466 // Returns the location index and size if the location is found, 0 otherwise.
467 u4 ImageFileReader::find_location_index(const char* path, u8 *size) const {
468 // Locate the entry in the index perfect hash table.
469 s4 index = ImageStrings::find(_endian, path, _redirect_table, table_length());
470 // If found.
471 if (index != ImageStrings::NOT_FOUND) {
472 // Get address of first byte of location attribute stream.
473 u4 offset = get_location_offset(index);
474 u1* data = get_location_offset_data(offset);
475 // Expand location attributes.
476 ImageLocation location(data);
477 // Make sure result is not a false positive.
478 if (verify_location(location, path)) {
479 *size = (jlong)location.get_attribute(ImageLocation::ATTRIBUTE_UNCOMPRESSED);
480 return offset;
481 }
482 }
483 return 0; // not found
484 }
485
486 // Verify that a found location matches the supplied path (without copying.)
487 bool ImageFileReader::verify_location(ImageLocation& location, const char* path) const {
488 // Manage the image string table.
489 ImageStrings strings(_string_bytes, _header.strings_size(_endian));
490 // Position to first character of the path string.
491 const char* next = path;
492 // Get module name string.
493 const char* module = location.get_attribute(ImageLocation::ATTRIBUTE_MODULE, strings);
494 // If module string is not empty.
495 if (*module != '\0') {
496 // Compare '/module/' .
497 if (*next++ != '/') return false;
498 if (!(next = ImageStrings::starts_with(next, module))) return false;
499 if (*next++ != '/') return false;
500 }
501 // Get parent (package) string
502 const char* parent = location.get_attribute(ImageLocation::ATTRIBUTE_PARENT, strings);
503 // If parent string is not empty string.
504 if (*parent != '\0') {
505 // Compare 'parent/' .
506 if (!(next = ImageStrings::starts_with(next, parent))) return false;
507 if (*next++ != '/') return false;
508 }
509 // Get base name string.
510 const char* base = location.get_attribute(ImageLocation::ATTRIBUTE_BASE, strings);
511 // Compare with basne name.
512 if (!(next = ImageStrings::starts_with(next, base))) return false;
513 // Get extension string.
514 const char* extension = location.get_attribute(ImageLocation::ATTRIBUTE_EXTENSION, strings);
515 // If extension is not empty.
516 if (*extension != '\0') {
517 // Compare '.extension' .
518 if (*next++ != '.') return false;
519 if (!(next = ImageStrings::starts_with(next, extension))) return false;
520 }
521 // True only if complete match and no more characters.
522 return *next == '\0';
523 }
524
525 // Return the resource for the supplied location offset.
526 void ImageFileReader::get_resource(u4 offset, u1* uncompressed_data) const {
527 // Get address of first byte of location attribute stream.
528 u1* data = get_location_offset_data(offset);
529 // Expand location attributes.
530 ImageLocation location(data);
531 // Read the data
532 get_resource(location, uncompressed_data);
533 }
534
535 // Return the resource for the supplied location.
536 void ImageFileReader::get_resource(ImageLocation& location, u1* uncompressed_data) const {
537 // Retrieve the byte offset and size of the resource.
538 u8 offset = location.get_attribute(ImageLocation::ATTRIBUTE_OFFSET);
539 u8 uncompressed_size = location.get_attribute(ImageLocation::ATTRIBUTE_UNCOMPRESSED);
540 u8 compressed_size = location.get_attribute(ImageLocation::ATTRIBUTE_COMPRESSED);
541 // If the resource is compressed.
542 if (compressed_size != 0) {
543 u1* compressed_data;
544 // If not memory mapped read in bytes.
545 if (!memory_map_image) {
546 // Allocate buffer for compression.
547 compressed_data = new u1[(size_t)compressed_size];
548 assert(compressed_data != NULL && "allocation failed");
549 // Read bytes from offset beyond the image index.
550 bool is_read = read_at(compressed_data, compressed_size, _index_size + offset);
551 assert(is_read && "error reading from image or short read");
552 } else {
553 compressed_data = get_data_address() + offset;
554 }
555 // Get image string table.
556 const ImageStrings strings = get_strings();
557 // Decompress resource.
558 ImageDecompressor::decompress_resource(compressed_data, uncompressed_data, uncompressed_size,
559 &strings, _endian);
560 // If not memory mapped then release temporary buffer.
561 if (!memory_map_image) {
562 delete[] compressed_data;
563 }
564 } else {
565 // Read bytes from offset beyond the image index.
566 bool is_read = read_at(uncompressed_data, uncompressed_size, _index_size + offset);
567 assert(is_read && "error reading from image or short read");
568 }
569 }
570
571 // Return the ImageModuleData for this image
572 ImageModuleData * ImageFileReader::get_image_module_data() {
573 return _module_data;
574 }