1 | /* Test and measure strcmp and wcscmp functions. |
2 | Copyright (C) 1999-2022 Free Software Foundation, Inc. |
3 | This file is part of the GNU C Library. |
4 | |
5 | The GNU C Library is free software; you can redistribute it and/or |
6 | modify it under the terms of the GNU Lesser General Public |
7 | License as published by the Free Software Foundation; either |
8 | version 2.1 of the License, or (at your option) any later version. |
9 | |
10 | The GNU C Library is distributed in the hope that it will be useful, |
11 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
13 | Lesser General Public License for more details. |
14 | |
15 | You should have received a copy of the GNU Lesser General Public |
16 | License along with the GNU C Library; if not, see |
17 | <https://www.gnu.org/licenses/>. */ |
18 | |
19 | #define TEST_MAIN |
20 | #ifdef WIDE |
21 | # define TEST_NAME "wcscmp" |
22 | #else |
23 | # define TEST_NAME "strcmp" |
24 | #endif |
25 | #include "test-string.h" |
26 | #include <support/test-driver.h> |
27 | |
28 | #ifdef WIDE |
29 | # include <wchar.h> |
30 | |
31 | # define L(str) L##str |
32 | # define STRCMP wcscmp |
33 | # define STRCPY wcscpy |
34 | # define STRLEN wcslen |
35 | # define MEMCPY wmemcpy |
36 | # define SIMPLE_STRCMP simple_wcscmp |
37 | # define STUPID_STRCMP stupid_wcscmp |
38 | # define CHAR wchar_t |
39 | # define UCHAR wchar_t |
40 | # define CHARBYTES 4 |
41 | # define CHARBYTESLOG 2 |
42 | # define CHARALIGN __alignof__ (CHAR) |
43 | # define MIDCHAR 0x7fffffff |
44 | # define LARGECHAR 0xfffffffe |
45 | # define CHAR__MAX WCHAR_MAX |
46 | # define CHAR__MIN WCHAR_MIN |
47 | |
48 | /* Wcscmp uses signed semantics for comparison, not unsigned */ |
49 | /* Avoid using substraction since possible overflow */ |
50 | |
51 | int |
52 | simple_wcscmp (const wchar_t *s1, const wchar_t *s2) |
53 | { |
54 | wchar_t c1, c2; |
55 | do |
56 | { |
57 | c1 = *s1++; |
58 | c2 = *s2++; |
59 | if (c2 == L'\0') |
60 | return c1 - c2; |
61 | } |
62 | while (c1 == c2); |
63 | |
64 | return c1 < c2 ? -1 : 1; |
65 | } |
66 | |
67 | int |
68 | stupid_wcscmp (const wchar_t *s1, const wchar_t *s2) |
69 | { |
70 | size_t ns1 = wcslen (s1) + 1; |
71 | size_t ns2 = wcslen (s2) + 1; |
72 | size_t n = ns1 < ns2 ? ns1 : ns2; |
73 | int ret = 0; |
74 | |
75 | wchar_t c1, c2; |
76 | |
77 | while (n--) { |
78 | c1 = *s1++; |
79 | c2 = *s2++; |
80 | if ((ret = c1 < c2 ? -1 : c1 == c2 ? 0 : 1) != 0) |
81 | break; |
82 | } |
83 | return ret; |
84 | } |
85 | |
86 | #else |
87 | # include <limits.h> |
88 | |
89 | # define L(str) str |
90 | # define STRCMP strcmp |
91 | # define STRCPY strcpy |
92 | # define STRLEN strlen |
93 | # define MEMCPY memcpy |
94 | # define SIMPLE_STRCMP simple_strcmp |
95 | # define STUPID_STRCMP stupid_strcmp |
96 | # define CHAR char |
97 | # define UCHAR unsigned char |
98 | # define CHARBYTES 1 |
99 | # define CHARBYTESLOG 0 |
100 | # define CHARALIGN 1 |
101 | # define MIDCHAR 0x7f |
102 | # define LARGECHAR 0xfe |
103 | # define CHAR__MAX CHAR_MAX |
104 | # define CHAR__MIN CHAR_MIN |
105 | |
106 | /* Strcmp uses unsigned semantics for comparison. */ |
107 | int |
108 | simple_strcmp (const char *s1, const char *s2) |
109 | { |
110 | int ret; |
111 | |
112 | while ((ret = *(unsigned char *) s1 - *(unsigned char*) s2++) == 0 && *s1++); |
113 | return ret; |
114 | } |
115 | |
116 | int |
117 | stupid_strcmp (const char *s1, const char *s2) |
118 | { |
119 | size_t ns1 = strlen (s1) + 1; |
120 | size_t ns2 = strlen (s2) + 1; |
121 | size_t n = ns1 < ns2 ? ns1 : ns2; |
122 | int ret = 0; |
123 | |
124 | while (n--) |
125 | if ((ret = *(unsigned char *) s1++ - *(unsigned char *) s2++) != 0) |
126 | break; |
127 | return ret; |
128 | } |
129 | #endif |
130 | |
131 | typedef int (*proto_t) (const CHAR *, const CHAR *); |
132 | |
133 | IMPL (STUPID_STRCMP, 1) |
134 | IMPL (SIMPLE_STRCMP, 1) |
135 | IMPL (STRCMP, 1) |
136 | |
137 | static int |
138 | check_result (impl_t *impl, |
139 | const CHAR *s1, const CHAR *s2, |
140 | int exp_result) |
141 | { |
142 | int result = CALL (impl, s1, s2); |
143 | if ((exp_result == 0 && result != 0) |
144 | || (exp_result < 0 && result >= 0) |
145 | || (exp_result > 0 && result <= 0)) |
146 | { |
147 | error (status: 0, errnum: 0, format: "Wrong result in function %s %d %d" , impl->name, |
148 | result, exp_result); |
149 | ret = 1; |
150 | return -1; |
151 | } |
152 | |
153 | return 0; |
154 | } |
155 | |
156 | static void |
157 | do_one_test (impl_t *impl, |
158 | const CHAR *s1, const CHAR *s2, |
159 | int exp_result) |
160 | { |
161 | if (check_result (impl, s1, s2, exp_result) < 0) |
162 | return; |
163 | } |
164 | |
165 | static void |
166 | do_test (size_t align1, size_t align2, size_t len, int max_char, |
167 | int exp_result) |
168 | { |
169 | size_t i; |
170 | |
171 | CHAR *s1, *s2; |
172 | |
173 | if (len == 0) |
174 | return; |
175 | |
176 | align1 &= 63; |
177 | if (align1 + (len + 1) * CHARBYTES >= page_size) |
178 | return; |
179 | |
180 | align2 &= 63; |
181 | if (align2 + (len + 1) * CHARBYTES >= page_size) |
182 | return; |
183 | |
184 | /* Put them close to the end of page. */ |
185 | i = align1 + CHARBYTES * (len + 2); |
186 | s1 = (CHAR *) (buf1 + ((page_size - i) / 16 * 16) + align1); |
187 | i = align2 + CHARBYTES * (len + 2); |
188 | s2 = (CHAR *) (buf2 + ((page_size - i) / 16 * 16) + align2); |
189 | |
190 | for (i = 0; i < len; i++) |
191 | s1[i] = s2[i] = 1 + (23 << ((CHARBYTES - 1) * 8)) * i % max_char; |
192 | |
193 | s1[len] = s2[len] = 0; |
194 | s1[len + 1] = 23; |
195 | s2[len + 1] = 24 + exp_result; |
196 | s2[len - 1] -= exp_result; |
197 | |
198 | FOR_EACH_IMPL (impl, 0) |
199 | do_one_test (impl, s1, s2, exp_result); |
200 | } |
201 | |
202 | static void |
203 | do_random_tests (void) |
204 | { |
205 | UCHAR *p1 = (UCHAR *) (buf1 + page_size - 512 * CHARBYTES); |
206 | UCHAR *p2 = (UCHAR *) (buf2 + page_size - 512 * CHARBYTES); |
207 | |
208 | for (size_t n = 0; n < ITERATIONS; n++) |
209 | { |
210 | /* for wcscmp case align1 and align2 mean here alignment |
211 | in wchar_t symbols, it equal 4*k alignment in bytes, we |
212 | don't check other alignments like for example |
213 | p1 = (wchar_t *)(buf1 + 1) |
214 | because it's wrong using of wchar_t type. */ |
215 | size_t align1 = random () & 31; |
216 | size_t align2; |
217 | if (random () & 1) |
218 | align2 = random () & 31; |
219 | else |
220 | align2 = align1 + (random () & 24); |
221 | size_t pos = random () & 511; |
222 | size_t j = align1 > align2 ? align1 : align2; |
223 | if (pos + j >= 511) |
224 | pos = 510 - j - (random () & 7); |
225 | size_t len1 = random () & 511; |
226 | if (pos >= len1 && (random () & 1)) |
227 | len1 = pos + (random () & 7); |
228 | if (len1 + j >= 512) |
229 | len1 = 511 - j - (random () & 7); |
230 | size_t len2; |
231 | if (pos >= len1) |
232 | len2 = len1; |
233 | else |
234 | len2 = len1 + (len1 != 511 - j ? random () % (511 - j - len1) : 0); |
235 | j = (pos > len2 ? pos : len2) + align1 + 64; |
236 | if (j > 512) |
237 | j = 512; |
238 | for (size_t i = 0; i < j; ++i) |
239 | { |
240 | p1[i] = random () & 255; |
241 | if (i < len1 + align1 && !p1[i]) |
242 | { |
243 | p1[i] = random () & 255; |
244 | if (!p1[i]) |
245 | p1[i] = 1 + (random () & 127); |
246 | } |
247 | } |
248 | for (size_t i = 0; i < j; ++i) |
249 | { |
250 | p2[i] = random () & 255; |
251 | if (i < len2 + align2 && !p2[i]) |
252 | { |
253 | p2[i] = random () & 255; |
254 | if (!p2[i]) |
255 | p2[i] = 1 + (random () & 127); |
256 | } |
257 | } |
258 | |
259 | int result = 0; |
260 | MEMCPY (p2 + align2, p1 + align1, pos); |
261 | if (pos < len1) |
262 | { |
263 | if (p2[align2 + pos] == p1[align1 + pos]) |
264 | { |
265 | p2[align2 + pos] = random () & 255; |
266 | if (p2[align2 + pos] == p1[align1 + pos]) |
267 | p2[align2 + pos] = p1[align1 + pos] + 3 + (random () & 127); |
268 | } |
269 | |
270 | if (p1[align1 + pos] < p2[align2 + pos]) |
271 | result = -1; |
272 | else |
273 | result = 1; |
274 | } |
275 | p1[len1 + align1] = 0; |
276 | p2[len2 + align2] = 0; |
277 | |
278 | FOR_EACH_IMPL (impl, 1) |
279 | { |
280 | int r = CALL (impl, (CHAR *) (p1 + align1), (CHAR *) (p2 + align2)); |
281 | /* Test whether on 64-bit architectures where ABI requires |
282 | callee to promote has the promotion been done. */ |
283 | asm ("" : "=g" (r) : "0" (r)); |
284 | if ((r == 0 && result) |
285 | || (r < 0 && result >= 0) |
286 | || (r > 0 && result <= 0)) |
287 | { |
288 | error (status: 0, errnum: 0, format: "Iteration %zd - wrong result in function %s (align in bytes: %zd, align in bytes: %zd, len1: %zd, len2: %zd, pos: %zd) %d != %d, p1 %p p2 %p" , |
289 | n, impl->name, (size_t) (p1 + align1) & 63, (size_t) (p1 + align2) & 63, len1, len2, pos, r, result, p1, p2); |
290 | ret = 1; |
291 | } |
292 | } |
293 | } |
294 | } |
295 | |
296 | static void |
297 | check (void) |
298 | { |
299 | CHAR *s1 = (CHAR *) (buf1 + 0xb2c); |
300 | CHAR *s2 = (CHAR *) (buf1 + 0xfd8); |
301 | |
302 | STRCPY(s1, L("abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrs" )); |
303 | STRCPY(s2, L("abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijkLMNOPQRSTUV" )); |
304 | |
305 | /* Check correct working for negatives values */ |
306 | |
307 | s1[0] = 1; |
308 | s2[0] = 1; |
309 | s1[1] = 1; |
310 | s2[1] = 1; |
311 | s1[2] = -1; |
312 | s2[2] = 3; |
313 | s1[3] = 0; |
314 | s2[3] = -1; |
315 | |
316 | /* Check possible overflow bug, actual more for wcscmp */ |
317 | |
318 | s1[7] = CHAR__MIN; |
319 | s2[7] = CHAR__MAX; |
320 | |
321 | size_t l1 = STRLEN (s1); |
322 | size_t l2 = STRLEN (s2); |
323 | |
324 | for (size_t i1 = 0; i1 < l1; i1++) |
325 | for (size_t i2 = 0; i2 < l2; i2++) |
326 | { |
327 | int exp_result = SIMPLE_STRCMP (s1: s1 + i1, s2: s2 + i2); |
328 | FOR_EACH_IMPL (impl, 0) |
329 | check_result (impl, s1: s1 + i1, s2: s2 + i2, exp_result); |
330 | } |
331 | |
332 | /* Test cases where there are multiple zero bytes after the first. */ |
333 | |
334 | for (size_t i = 0; i < 16 + 1; i++) |
335 | { |
336 | s1[i] = 0x00; |
337 | s2[i] = 0x00; |
338 | } |
339 | |
340 | for (size_t i = 0; i < 16; i++) |
341 | { |
342 | int exp_result; |
343 | |
344 | for (int val = 0x01; val < 0x100; val++) |
345 | { |
346 | for (size_t j = 0; j < i; j++) |
347 | { |
348 | s1[j] = val; |
349 | s2[j] = val; |
350 | } |
351 | |
352 | s2[i] = val; |
353 | |
354 | exp_result = SIMPLE_STRCMP (s1, s2); |
355 | FOR_EACH_IMPL (impl, 0) |
356 | check_result (impl, s1, s2, exp_result); |
357 | } |
358 | } |
359 | } |
360 | |
361 | static void |
362 | check2 (void) |
363 | { |
364 | /* To trigger bug 25933, we need a size that is equal to the vector |
365 | length times 4. In the case of AVX2 for Intel, we need 32 * 4. We |
366 | make this test generic and run it for all architectures as additional |
367 | boundary testing for such related algorithms. */ |
368 | size_t size = 32 * 4; |
369 | CHAR *s1 = (CHAR *) (buf1 + (BUF1PAGES - 1) * page_size); |
370 | CHAR *s2 = (CHAR *) (buf2 + (BUF1PAGES - 1) * page_size); |
371 | int exp_result; |
372 | |
373 | memset (s1, 'a', page_size); |
374 | memset (s2, 'a', page_size); |
375 | s1[(page_size / CHARBYTES) - 1] = (CHAR) 0; |
376 | s2[(page_size / CHARBYTES) - 1] = (CHAR) 0; |
377 | |
378 | /* Iterate over a size that is just below where we expect the bug to |
379 | trigger up to the size we expect will trigger the bug e.g. [99-128]. |
380 | Likewise iterate the start of two strings between 30 and 31 bytes |
381 | away from the boundary to simulate alignment changes. */ |
382 | for (size_t s = 99; s <= size; s++) |
383 | for (size_t s1a = 30; s1a < 32; s1a++) |
384 | for (size_t s2a = 30; s2a < 32; s2a++) |
385 | { |
386 | CHAR *s1p = s1 + (page_size / CHARBYTES - s) - s1a; |
387 | CHAR *s2p = s2 + (page_size / CHARBYTES - s) - s2a; |
388 | exp_result = SIMPLE_STRCMP (s1: s1p, s2: s2p); |
389 | FOR_EACH_IMPL (impl, 0) |
390 | check_result (impl, s1: s1p, s2: s2p, exp_result); |
391 | } |
392 | } |
393 | |
394 | static void |
395 | check3 (void) |
396 | { |
397 | size_t size = 0xd000 + 0x4000; |
398 | CHAR *s1, *s2; |
399 | CHAR *buffer1 = mmap (NULL, len: size, PROT_READ | PROT_WRITE, |
400 | MAP_PRIVATE | MAP_ANON, fd: -1, offset: 0); |
401 | CHAR *buffer2 = mmap (NULL, len: size, PROT_READ | PROT_WRITE, |
402 | MAP_PRIVATE | MAP_ANON, fd: -1, offset: 0); |
403 | if (buffer1 == MAP_FAILED || buffer1 == MAP_FAILED) |
404 | error (status: EXIT_UNSUPPORTED, errno, format: "mmap failed" ); |
405 | |
406 | s1 = (CHAR *) (buffer1 + 0x8f8 / sizeof (CHAR)); |
407 | s2 = (CHAR *) (buffer2 + 0xcff3 / sizeof (CHAR)); |
408 | |
409 | STRCPY(s1, L("/export/redhat/rpms/BUILD/java-1.8.0-openjdk-1.8.0.312.b07-2.fc35.x86_64/openjdk/langtools/src/share/classes/com/sun/tools/doclets/internal/toolkit/util/PathDocFileFactory.java" )); |
410 | STRCPY(s2, L("/export/redhat/rpms/BUILD/java-1.8.0-openjdk-1.8.0.312.b07-2.fc35.x86_64/openjdk/langtools/src/share/classes/com/sun/tools/doclets/internal/toolkit/taglets/ThrowsTaglet.java" )); |
411 | |
412 | int exp_result = SIMPLE_STRCMP (s1, s2); |
413 | FOR_EACH_IMPL (impl, 0) |
414 | check_result (impl, s1, s2, exp_result); |
415 | |
416 | munmap (addr: (void *) buffer1, len: size); |
417 | munmap (addr: (void *) buffer2, len: size); |
418 | } |
419 | |
420 | int |
421 | test_main (void) |
422 | { |
423 | size_t i; |
424 | |
425 | test_init (); |
426 | check(); |
427 | check2 (); |
428 | check3 (); |
429 | |
430 | printf (format: "%23s" , "" ); |
431 | FOR_EACH_IMPL (impl, 0) |
432 | printf (format: "\t%s" , impl->name); |
433 | putchar (c: '\n'); |
434 | |
435 | for (i = 1; i < 32; ++i) |
436 | { |
437 | do_test (CHARBYTES * i, CHARBYTES * i, len: i, MIDCHAR, exp_result: 0); |
438 | do_test (CHARBYTES * i, CHARBYTES * i, len: i, MIDCHAR, exp_result: 1); |
439 | do_test (CHARBYTES * i, CHARBYTES * i, len: i, MIDCHAR, exp_result: -1); |
440 | } |
441 | |
442 | for (i = 1; i < 10 + CHARBYTESLOG; ++i) |
443 | { |
444 | do_test (align1: 0, align2: 0, len: 2 << i, MIDCHAR, exp_result: 0); |
445 | do_test (align1: 0, align2: 0, len: 2 << i, LARGECHAR, exp_result: 0); |
446 | do_test (align1: 0, align2: 0, len: 2 << i, MIDCHAR, exp_result: 1); |
447 | do_test (align1: 0, align2: 0, len: 2 << i, LARGECHAR, exp_result: 1); |
448 | do_test (align1: 0, align2: 0, len: 2 << i, MIDCHAR, exp_result: -1); |
449 | do_test (align1: 0, align2: 0, len: 2 << i, LARGECHAR, exp_result: -1); |
450 | do_test (align1: 0, CHARBYTES * i, len: 2 << i, MIDCHAR, exp_result: 1); |
451 | do_test (CHARBYTES * i, CHARBYTES * (i + 1), len: 2 << i, LARGECHAR, exp_result: 1); |
452 | } |
453 | |
454 | for (i = 1; i < 8; ++i) |
455 | { |
456 | do_test (CHARBYTES * i, align2: 2 * CHARBYTES * i, len: 8 << i, MIDCHAR, exp_result: 0); |
457 | do_test (align1: 2 * CHARBYTES * i, CHARBYTES * i, len: 8 << i, LARGECHAR, exp_result: 0); |
458 | do_test (CHARBYTES * i, align2: 2 * CHARBYTES * i, len: 8 << i, MIDCHAR, exp_result: 1); |
459 | do_test (align1: 2 * CHARBYTES * i, CHARBYTES * i, len: 8 << i, LARGECHAR, exp_result: 1); |
460 | do_test (CHARBYTES * i, align2: 2 * CHARBYTES * i, len: 8 << i, MIDCHAR, exp_result: -1); |
461 | do_test (align1: 2 * CHARBYTES * i, CHARBYTES * i, len: 8 << i, LARGECHAR, exp_result: -1); |
462 | } |
463 | |
464 | do_random_tests (); |
465 | return ret; |
466 | } |
467 | |
468 | #include <support/test-driver.c> |
469 | |