1 //===-- dfsan.cpp ---------------------------------------------------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This file is a part of DataFlowSanitizer.
10 //
11 // This file defines the custom functions listed in done_abilist.txt.
12 //===----------------------------------------------------------------------===//
13 
14 #include "sanitizer_common/sanitizer_common.h"
15 #include "sanitizer_common/sanitizer_internal_defs.h"
16 #include "sanitizer_common/sanitizer_linux.h"
17 
18 #include "dfsan/dfsan.h"
19 
20 #include <arpa/inet.h>
21 #include <assert.h>
22 #include <ctype.h>
23 #include <dlfcn.h>
24 #include <link.h>
25 #include <poll.h>
26 #include <pthread.h>
27 #include <pwd.h>
28 #include <sched.h>
29 #include <signal.h>
30 #include <stdarg.h>
31 #include <stdint.h>
32 #include <stdio.h>
33 #include <stdlib.h>
34 #include <string.h>
35 #include <sys/resource.h>
36 #include <sys/select.h>
37 #include <sys/stat.h>
38 #include <sys/time.h>
39 #include <sys/types.h>
40 #include <time.h>
41 #include <unistd.h>
42 
43 using namespace __dfsan;
44 
45 #define CALL_WEAK_INTERCEPTOR_HOOK(f, ...)                                     \
46   do {                                                                         \
47     if (f)                                                                     \
48       f(__VA_ARGS__);                                                          \
49   } while (false)
50 #define DECLARE_WEAK_INTERCEPTOR_HOOK(f, ...) \
51 SANITIZER_INTERFACE_ATTRIBUTE SANITIZER_WEAK_ATTRIBUTE void f(__VA_ARGS__);
52 
53 extern "C" {
54 SANITIZER_INTERFACE_ATTRIBUTE int
55 __dfsw_stat(const char *path, struct stat *buf, dfsan_label path_label,
56             dfsan_label buf_label, dfsan_label *ret_label) {
57   int ret = stat(path, buf);
58   if (ret == 0)
59     dfsan_set_label(0, buf, sizeof(struct stat));
60   *ret_label = 0;
61   return ret;
62 }
63 
64 SANITIZER_INTERFACE_ATTRIBUTE int __dfsw_fstat(int fd, struct stat *buf,
65                                                dfsan_label fd_label,
66                                                dfsan_label buf_label,
67                                                dfsan_label *ret_label) {
68   int ret = fstat(fd, buf);
69   if (ret == 0)
70     dfsan_set_label(0, buf, sizeof(struct stat));
71   *ret_label = 0;
72   return ret;
73 }
74 
75 SANITIZER_INTERFACE_ATTRIBUTE char *__dfsw_strchr(const char *s, int c,
76                                                   dfsan_label s_label,
77                                                   dfsan_label c_label,
78                                                   dfsan_label *ret_label) {
79   for (size_t i = 0;; ++i) {
80     if (s[i] == c || s[i] == 0) {
81       if (flags().strict_data_dependencies) {
82         *ret_label = s_label;
83       } else {
84         *ret_label = dfsan_union(dfsan_read_label(s, i + 1),
85                                  dfsan_union(s_label, c_label));
86       }
87 
88       // If s[i] is the \0 at the end of the string, and \0 is not the
89       // character we are searching for, then return null.
90       if (s[i] == 0 && c != 0) {
91         return nullptr;
92       }
93       return const_cast<char *>(s + i);
94     }
95   }
96 }
97 
98 DECLARE_WEAK_INTERCEPTOR_HOOK(dfsan_weak_hook_memcmp, uptr caller_pc,
99                               const void *s1, const void *s2, size_t n,
100                               dfsan_label s1_label, dfsan_label s2_label,
101                               dfsan_label n_label)
102 
103 SANITIZER_INTERFACE_ATTRIBUTE int __dfsw_memcmp(const void *s1, const void *s2,
104                                                 size_t n, dfsan_label s1_label,
105                                                 dfsan_label s2_label,
106                                                 dfsan_label n_label,
107                                                 dfsan_label *ret_label) {
108   CALL_WEAK_INTERCEPTOR_HOOK(dfsan_weak_hook_memcmp, GET_CALLER_PC(), s1, s2, n,
109                              s1_label, s2_label, n_label);
110   const char *cs1 = (const char *) s1, *cs2 = (const char *) s2;
111   for (size_t i = 0; i != n; ++i) {
112     if (cs1[i] != cs2[i]) {
113       if (flags().strict_data_dependencies) {
114         *ret_label = 0;
115       } else {
116         *ret_label = dfsan_union(dfsan_read_label(cs1, i + 1),
117                                  dfsan_read_label(cs2, i + 1));
118       }
119       return cs1[i] - cs2[i];
120     }
121   }
122 
123   if (flags().strict_data_dependencies) {
124     *ret_label = 0;
125   } else {
126     *ret_label = dfsan_union(dfsan_read_label(cs1, n),
127                              dfsan_read_label(cs2, n));
128   }
129   return 0;
130 }
131 
132 DECLARE_WEAK_INTERCEPTOR_HOOK(dfsan_weak_hook_strcmp, uptr caller_pc,
133                               const char *s1, const char *s2,
134                               dfsan_label s1_label, dfsan_label s2_label)
135 
136 SANITIZER_INTERFACE_ATTRIBUTE int __dfsw_strcmp(const char *s1, const char *s2,
137                                                 dfsan_label s1_label,
138                                                 dfsan_label s2_label,
139                                                 dfsan_label *ret_label) {
140   CALL_WEAK_INTERCEPTOR_HOOK(dfsan_weak_hook_strcmp, GET_CALLER_PC(), s1, s2,
141                              s1_label, s2_label);
142   for (size_t i = 0;; ++i) {
143     if (s1[i] != s2[i] || s1[i] == 0 || s2[i] == 0) {
144       if (flags().strict_data_dependencies) {
145         *ret_label = 0;
146       } else {
147         *ret_label = dfsan_union(dfsan_read_label(s1, i + 1),
148                                  dfsan_read_label(s2, i + 1));
149       }
150       return s1[i] - s2[i];
151     }
152   }
153   return 0;
154 }
155 
156 SANITIZER_INTERFACE_ATTRIBUTE int
157 __dfsw_strcasecmp(const char *s1, const char *s2, dfsan_label s1_label,
158                   dfsan_label s2_label, dfsan_label *ret_label) {
159   for (size_t i = 0;; ++i) {
160     if (tolower(s1[i]) != tolower(s2[i]) || s1[i] == 0 || s2[i] == 0) {
161       if (flags().strict_data_dependencies) {
162         *ret_label = 0;
163       } else {
164         *ret_label = dfsan_union(dfsan_read_label(s1, i + 1),
165                                  dfsan_read_label(s2, i + 1));
166       }
167       return s1[i] - s2[i];
168     }
169   }
170   return 0;
171 }
172 
173 DECLARE_WEAK_INTERCEPTOR_HOOK(dfsan_weak_hook_strncmp, uptr caller_pc,
174                               const char *s1, const char *s2, size_t n,
175                               dfsan_label s1_label, dfsan_label s2_label,
176                               dfsan_label n_label)
177 
178 SANITIZER_INTERFACE_ATTRIBUTE int __dfsw_strncmp(const char *s1, const char *s2,
179                                                  size_t n, dfsan_label s1_label,
180                                                  dfsan_label s2_label,
181                                                  dfsan_label n_label,
182                                                  dfsan_label *ret_label) {
183   if (n == 0) {
184     *ret_label = 0;
185     return 0;
186   }
187 
188   CALL_WEAK_INTERCEPTOR_HOOK(dfsan_weak_hook_strncmp, GET_CALLER_PC(), s1, s2,
189                              n, s1_label, s2_label, n_label);
190 
191   for (size_t i = 0;; ++i) {
192     if (s1[i] != s2[i] || s1[i] == 0 || s2[i] == 0 || i == n - 1) {
193       if (flags().strict_data_dependencies) {
194         *ret_label = 0;
195       } else {
196         *ret_label = dfsan_union(dfsan_read_label(s1, i + 1),
197                                  dfsan_read_label(s2, i + 1));
198       }
199       return s1[i] - s2[i];
200     }
201   }
202   return 0;
203 }
204 
205 SANITIZER_INTERFACE_ATTRIBUTE int
206 __dfsw_strncasecmp(const char *s1, const char *s2, size_t n,
207                    dfsan_label s1_label, dfsan_label s2_label,
208                    dfsan_label n_label, dfsan_label *ret_label) {
209   if (n == 0) {
210     *ret_label = 0;
211     return 0;
212   }
213 
214   for (size_t i = 0;; ++i) {
215     if (tolower(s1[i]) != tolower(s2[i]) || s1[i] == 0 || s2[i] == 0 ||
216         i == n - 1) {
217       if (flags().strict_data_dependencies) {
218         *ret_label = 0;
219       } else {
220         *ret_label = dfsan_union(dfsan_read_label(s1, i + 1),
221                                  dfsan_read_label(s2, i + 1));
222       }
223       return s1[i] - s2[i];
224     }
225   }
226   return 0;
227 }
228 
229 SANITIZER_INTERFACE_ATTRIBUTE void *__dfsw_calloc(size_t nmemb, size_t size,
230                                                   dfsan_label nmemb_label,
231                                                   dfsan_label size_label,
232                                                   dfsan_label *ret_label) {
233   void *p = calloc(nmemb, size);
234   dfsan_set_label(0, p, nmemb * size);
235   *ret_label = 0;
236   return p;
237 }
238 
239 SANITIZER_INTERFACE_ATTRIBUTE size_t
240 __dfsw_strlen(const char *s, dfsan_label s_label, dfsan_label *ret_label) {
241   size_t ret = strlen(s);
242   if (flags().strict_data_dependencies) {
243     *ret_label = 0;
244   } else {
245     *ret_label = dfsan_read_label(s, ret + 1);
246   }
247   return ret;
248 }
249 
250 
251 static void *dfsan_memcpy(void *dest, const void *src, size_t n) {
252   dfsan_label *sdest = shadow_for(dest);
253   const dfsan_label *ssrc = shadow_for(src);
254   internal_memcpy((void *)sdest, (const void *)ssrc, n * sizeof(dfsan_label));
255   return internal_memcpy(dest, src, n);
256 }
257 
258 static void dfsan_memset(void *s, int c, dfsan_label c_label, size_t n) {
259   internal_memset(s, c, n);
260   dfsan_set_label(c_label, s, n);
261 }
262 
263 SANITIZER_INTERFACE_ATTRIBUTE
264 void *__dfsw_memcpy(void *dest, const void *src, size_t n,
265                     dfsan_label dest_label, dfsan_label src_label,
266                     dfsan_label n_label, dfsan_label *ret_label) {
267   *ret_label = dest_label;
268   return dfsan_memcpy(dest, src, n);
269 }
270 
271 SANITIZER_INTERFACE_ATTRIBUTE
272 void *__dfsw_memset(void *s, int c, size_t n,
273                     dfsan_label s_label, dfsan_label c_label,
274                     dfsan_label n_label, dfsan_label *ret_label) {
275   dfsan_memset(s, c, c_label, n);
276   *ret_label = s_label;
277   return s;
278 }
279 
280 SANITIZER_INTERFACE_ATTRIBUTE char *
281 __dfsw_strdup(const char *s, dfsan_label s_label, dfsan_label *ret_label) {
282   size_t len = strlen(s);
283   void *p = malloc(len+1);
284   dfsan_memcpy(p, s, len+1);
285   *ret_label = 0;
286   return static_cast<char *>(p);
287 }
288 
289 SANITIZER_INTERFACE_ATTRIBUTE char *
290 __dfsw_strncpy(char *s1, const char *s2, size_t n, dfsan_label s1_label,
291                dfsan_label s2_label, dfsan_label n_label,
292                dfsan_label *ret_label) {
293   size_t len = strlen(s2);
294   if (len < n) {
295     dfsan_memcpy(s1, s2, len+1);
296     dfsan_memset(s1+len+1, 0, 0, n-len-1);
297   } else {
298     dfsan_memcpy(s1, s2, n);
299   }
300 
301   *ret_label = s1_label;
302   return s1;
303 }
304 
305 SANITIZER_INTERFACE_ATTRIBUTE ssize_t
306 __dfsw_pread(int fd, void *buf, size_t count, off_t offset,
307              dfsan_label fd_label, dfsan_label buf_label,
308              dfsan_label count_label, dfsan_label offset_label,
309              dfsan_label *ret_label) {
310   ssize_t ret = pread(fd, buf, count, offset);
311   if (ret > 0)
312     dfsan_set_label(0, buf, ret);
313   *ret_label = 0;
314   return ret;
315 }
316 
317 SANITIZER_INTERFACE_ATTRIBUTE ssize_t
318 __dfsw_read(int fd, void *buf, size_t count,
319              dfsan_label fd_label, dfsan_label buf_label,
320              dfsan_label count_label,
321              dfsan_label *ret_label) {
322   ssize_t ret = read(fd, buf, count);
323   if (ret > 0)
324     dfsan_set_label(0, buf, ret);
325   *ret_label = 0;
326   return ret;
327 }
328 
329 SANITIZER_INTERFACE_ATTRIBUTE int __dfsw_clock_gettime(clockid_t clk_id,
330                                                        struct timespec *tp,
331                                                        dfsan_label clk_id_label,
332                                                        dfsan_label tp_label,
333                                                        dfsan_label *ret_label) {
334   int ret = clock_gettime(clk_id, tp);
335   if (ret == 0)
336     dfsan_set_label(0, tp, sizeof(struct timespec));
337   *ret_label = 0;
338   return ret;
339 }
340 
341 static void unpoison(const void *ptr, uptr size) {
342   dfsan_set_label(0, const_cast<void *>(ptr), size);
343 }
344 
345 // dlopen() ultimately calls mmap() down inside the loader, which generally
346 // doesn't participate in dynamic symbol resolution.  Therefore we won't
347 // intercept its calls to mmap, and we have to hook it here.
348 SANITIZER_INTERFACE_ATTRIBUTE void *
349 __dfsw_dlopen(const char *filename, int flag, dfsan_label filename_label,
350               dfsan_label flag_label, dfsan_label *ret_label) {
351   void *handle = dlopen(filename, flag);
352   link_map *map = GET_LINK_MAP_BY_DLOPEN_HANDLE(handle);
353   if (map)
354     ForEachMappedRegion(map, unpoison);
355   *ret_label = 0;
356   return handle;
357 }
358 
359 struct pthread_create_info {
360   void *(*start_routine_trampoline)(void *, void *, dfsan_label, dfsan_label *);
361   void *start_routine;
362   void *arg;
363 };
364 
365 static void *pthread_create_cb(void *p) {
366   pthread_create_info pci(*(pthread_create_info *)p);
367   free(p);
368   dfsan_label ret_label;
369   return pci.start_routine_trampoline(pci.start_routine, pci.arg, 0,
370                                       &ret_label);
371 }
372 
373 SANITIZER_INTERFACE_ATTRIBUTE int __dfsw_pthread_create(
374     pthread_t *thread, const pthread_attr_t *attr,
375     void *(*start_routine_trampoline)(void *, void *, dfsan_label,
376                                       dfsan_label *),
377     void *start_routine, void *arg, dfsan_label thread_label,
378     dfsan_label attr_label, dfsan_label start_routine_label,
379     dfsan_label arg_label, dfsan_label *ret_label) {
380   pthread_create_info *pci =
381       (pthread_create_info *)malloc(sizeof(pthread_create_info));
382   pci->start_routine_trampoline = start_routine_trampoline;
383   pci->start_routine = start_routine;
384   pci->arg = arg;
385   int rv = pthread_create(thread, attr, pthread_create_cb, (void *)pci);
386   if (rv != 0)
387     free(pci);
388   *ret_label = 0;
389   return rv;
390 }
391 
392 struct dl_iterate_phdr_info {
393   int (*callback_trampoline)(void *callback, struct dl_phdr_info *info,
394                              size_t size, void *data, dfsan_label info_label,
395                              dfsan_label size_label, dfsan_label data_label,
396                              dfsan_label *ret_label);
397   void *callback;
398   void *data;
399 };
400 
401 int dl_iterate_phdr_cb(struct dl_phdr_info *info, size_t size, void *data) {
402   dl_iterate_phdr_info *dipi = (dl_iterate_phdr_info *)data;
403   dfsan_set_label(0, *info);
404   dfsan_set_label(0, const_cast<char *>(info->dlpi_name),
405                   strlen(info->dlpi_name) + 1);
406   dfsan_set_label(
407       0, const_cast<char *>(reinterpret_cast<const char *>(info->dlpi_phdr)),
408       sizeof(*info->dlpi_phdr) * info->dlpi_phnum);
409   dfsan_label ret_label;
410   return dipi->callback_trampoline(dipi->callback, info, size, dipi->data, 0, 0,
411                                    0, &ret_label);
412 }
413 
414 SANITIZER_INTERFACE_ATTRIBUTE int __dfsw_dl_iterate_phdr(
415     int (*callback_trampoline)(void *callback, struct dl_phdr_info *info,
416                                size_t size, void *data, dfsan_label info_label,
417                                dfsan_label size_label, dfsan_label data_label,
418                                dfsan_label *ret_label),
419     void *callback, void *data, dfsan_label callback_label,
420     dfsan_label data_label, dfsan_label *ret_label) {
421   dl_iterate_phdr_info dipi = { callback_trampoline, callback, data };
422   *ret_label = 0;
423   return dl_iterate_phdr(dl_iterate_phdr_cb, &dipi);
424 }
425 
426 SANITIZER_INTERFACE_ATTRIBUTE
427 char *__dfsw_ctime_r(const time_t *timep, char *buf, dfsan_label timep_label,
428                      dfsan_label buf_label, dfsan_label *ret_label) {
429   char *ret = ctime_r(timep, buf);
430   if (ret) {
431     dfsan_set_label(dfsan_read_label(timep, sizeof(time_t)), buf,
432                     strlen(buf) + 1);
433     *ret_label = buf_label;
434   } else {
435     *ret_label = 0;
436   }
437   return ret;
438 }
439 
440 SANITIZER_INTERFACE_ATTRIBUTE
441 char *__dfsw_fgets(char *s, int size, FILE *stream, dfsan_label s_label,
442                    dfsan_label size_label, dfsan_label stream_label,
443                    dfsan_label *ret_label) {
444   char *ret = fgets(s, size, stream);
445   if (ret) {
446     dfsan_set_label(0, ret, strlen(ret) + 1);
447     *ret_label = s_label;
448   } else {
449     *ret_label = 0;
450   }
451   return ret;
452 }
453 
454 SANITIZER_INTERFACE_ATTRIBUTE
455 char *__dfsw_getcwd(char *buf, size_t size, dfsan_label buf_label,
456                     dfsan_label size_label, dfsan_label *ret_label) {
457   char *ret = getcwd(buf, size);
458   if (ret) {
459     dfsan_set_label(0, ret, strlen(ret) + 1);
460     *ret_label = buf_label;
461   } else {
462     *ret_label = 0;
463   }
464   return ret;
465 }
466 
467 SANITIZER_INTERFACE_ATTRIBUTE
468 char *__dfsw_get_current_dir_name(dfsan_label *ret_label) {
469   char *ret = get_current_dir_name();
470   if (ret) {
471     dfsan_set_label(0, ret, strlen(ret) + 1);
472   }
473   *ret_label = 0;
474   return ret;
475 }
476 
477 SANITIZER_INTERFACE_ATTRIBUTE
478 int __dfsw_gethostname(char *name, size_t len, dfsan_label name_label,
479                        dfsan_label len_label, dfsan_label *ret_label) {
480   int ret = gethostname(name, len);
481   if (ret == 0) {
482     dfsan_set_label(0, name, strlen(name) + 1);
483   }
484   *ret_label = 0;
485   return ret;
486 }
487 
488 SANITIZER_INTERFACE_ATTRIBUTE
489 int __dfsw_getrlimit(int resource, struct rlimit *rlim,
490                      dfsan_label resource_label, dfsan_label rlim_label,
491                      dfsan_label *ret_label) {
492   int ret = getrlimit(resource, rlim);
493   if (ret == 0) {
494     dfsan_set_label(0, rlim, sizeof(struct rlimit));
495   }
496   *ret_label = 0;
497   return ret;
498 }
499 
500 SANITIZER_INTERFACE_ATTRIBUTE
501 int __dfsw_getrusage(int who, struct rusage *usage, dfsan_label who_label,
502                      dfsan_label usage_label, dfsan_label *ret_label) {
503   int ret = getrusage(who, usage);
504   if (ret == 0) {
505     dfsan_set_label(0, usage, sizeof(struct rusage));
506   }
507   *ret_label = 0;
508   return ret;
509 }
510 
511 SANITIZER_INTERFACE_ATTRIBUTE
512 char *__dfsw_strcpy(char *dest, const char *src, dfsan_label dst_label,
513                     dfsan_label src_label, dfsan_label *ret_label) {
514   char *ret = strcpy(dest, src);  // NOLINT
515   if (ret) {
516     internal_memcpy(shadow_for(dest), shadow_for(src),
517                     sizeof(dfsan_label) * (strlen(src) + 1));
518   }
519   *ret_label = dst_label;
520   return ret;
521 }
522 
523 SANITIZER_INTERFACE_ATTRIBUTE
524 long int __dfsw_strtol(const char *nptr, char **endptr, int base,
525                        dfsan_label nptr_label, dfsan_label endptr_label,
526                        dfsan_label base_label, dfsan_label *ret_label) {
527   char *tmp_endptr;
528   long int ret = strtol(nptr, &tmp_endptr, base);
529   if (endptr) {
530     *endptr = tmp_endptr;
531   }
532   if (tmp_endptr > nptr) {
533     // If *tmp_endptr is '\0' include its label as well.
534     *ret_label = dfsan_union(
535         base_label,
536         dfsan_read_label(nptr, tmp_endptr - nptr + (*tmp_endptr ? 0 : 1)));
537   } else {
538     *ret_label = 0;
539   }
540   return ret;
541 }
542 
543 SANITIZER_INTERFACE_ATTRIBUTE
544 double __dfsw_strtod(const char *nptr, char **endptr,
545                        dfsan_label nptr_label, dfsan_label endptr_label,
546                        dfsan_label *ret_label) {
547   char *tmp_endptr;
548   double ret = strtod(nptr, &tmp_endptr);
549   if (endptr) {
550     *endptr = tmp_endptr;
551   }
552   if (tmp_endptr > nptr) {
553     // If *tmp_endptr is '\0' include its label as well.
554     *ret_label = dfsan_read_label(
555         nptr,
556         tmp_endptr - nptr + (*tmp_endptr ? 0 : 1));
557   } else {
558     *ret_label = 0;
559   }
560   return ret;
561 }
562 
563 SANITIZER_INTERFACE_ATTRIBUTE
564 long long int __dfsw_strtoll(const char *nptr, char **endptr, int base,
565                        dfsan_label nptr_label, dfsan_label endptr_label,
566                        dfsan_label base_label, dfsan_label *ret_label) {
567   char *tmp_endptr;
568   long long int ret = strtoll(nptr, &tmp_endptr, base);
569   if (endptr) {
570     *endptr = tmp_endptr;
571   }
572   if (tmp_endptr > nptr) {
573     // If *tmp_endptr is '\0' include its label as well.
574     *ret_label = dfsan_union(
575         base_label,
576         dfsan_read_label(nptr, tmp_endptr - nptr + (*tmp_endptr ? 0 : 1)));
577   } else {
578     *ret_label = 0;
579   }
580   return ret;
581 }
582 
583 SANITIZER_INTERFACE_ATTRIBUTE
584 unsigned long int __dfsw_strtoul(const char *nptr, char **endptr, int base,
585                        dfsan_label nptr_label, dfsan_label endptr_label,
586                        dfsan_label base_label, dfsan_label *ret_label) {
587   char *tmp_endptr;
588   unsigned long int ret = strtoul(nptr, &tmp_endptr, base);
589   if (endptr) {
590     *endptr = tmp_endptr;
591   }
592   if (tmp_endptr > nptr) {
593     // If *tmp_endptr is '\0' include its label as well.
594     *ret_label = dfsan_union(
595         base_label,
596         dfsan_read_label(nptr, tmp_endptr - nptr + (*tmp_endptr ? 0 : 1)));
597   } else {
598     *ret_label = 0;
599   }
600   return ret;
601 }
602 
603 SANITIZER_INTERFACE_ATTRIBUTE
604 long long unsigned int __dfsw_strtoull(const char *nptr, char **endptr,
605                                        dfsan_label nptr_label,
606                                        int base, dfsan_label endptr_label,
607                                        dfsan_label base_label,
608                                        dfsan_label *ret_label) {
609   char *tmp_endptr;
610   long long unsigned int ret = strtoull(nptr, &tmp_endptr, base);
611   if (endptr) {
612     *endptr = tmp_endptr;
613   }
614   if (tmp_endptr > nptr) {
615     // If *tmp_endptr is '\0' include its label as well.
616     *ret_label = dfsan_union(
617         base_label,
618         dfsan_read_label(nptr, tmp_endptr - nptr + (*tmp_endptr ? 0 : 1)));
619   } else {
620     *ret_label = 0;
621   }
622   return ret;
623 }
624 
625 SANITIZER_INTERFACE_ATTRIBUTE
626 time_t __dfsw_time(time_t *t, dfsan_label t_label, dfsan_label *ret_label) {
627   time_t ret = time(t);
628   if (ret != (time_t) -1 && t) {
629     dfsan_set_label(0, t, sizeof(time_t));
630   }
631   *ret_label = 0;
632   return ret;
633 }
634 
635 SANITIZER_INTERFACE_ATTRIBUTE
636 int __dfsw_inet_pton(int af, const char *src, void *dst, dfsan_label af_label,
637                      dfsan_label src_label, dfsan_label dst_label,
638                      dfsan_label *ret_label) {
639   int ret = inet_pton(af, src, dst);
640   if (ret == 1) {
641     dfsan_set_label(dfsan_read_label(src, strlen(src) + 1), dst,
642                     af == AF_INET ? sizeof(struct in_addr) : sizeof(in6_addr));
643   }
644   *ret_label = 0;
645   return ret;
646 }
647 
648 SANITIZER_INTERFACE_ATTRIBUTE
649 struct tm *__dfsw_localtime_r(const time_t *timep, struct tm *result,
650                               dfsan_label timep_label, dfsan_label result_label,
651                               dfsan_label *ret_label) {
652   struct tm *ret = localtime_r(timep, result);
653   if (ret) {
654     dfsan_set_label(dfsan_read_label(timep, sizeof(time_t)), result,
655                     sizeof(struct tm));
656     *ret_label = result_label;
657   } else {
658     *ret_label = 0;
659   }
660   return ret;
661 }
662 
663 SANITIZER_INTERFACE_ATTRIBUTE
664 int __dfsw_getpwuid_r(id_t uid, struct passwd *pwd,
665                       char *buf, size_t buflen, struct passwd **result,
666                       dfsan_label uid_label, dfsan_label pwd_label,
667                       dfsan_label buf_label, dfsan_label buflen_label,
668                       dfsan_label result_label, dfsan_label *ret_label) {
669   // Store the data in pwd, the strings referenced from pwd in buf, and the
670   // address of pwd in *result.  On failure, NULL is stored in *result.
671   int ret = getpwuid_r(uid, pwd, buf, buflen, result);
672   if (ret == 0) {
673     dfsan_set_label(0, pwd, sizeof(struct passwd));
674     dfsan_set_label(0, buf, strlen(buf) + 1);
675   }
676   *ret_label = 0;
677   dfsan_set_label(0, result, sizeof(struct passwd*));
678   return ret;
679 }
680 
681 SANITIZER_INTERFACE_ATTRIBUTE
682 int __dfsw_poll(struct pollfd *fds, nfds_t nfds, int timeout,
683                 dfsan_label dfs_label, dfsan_label nfds_label,
684                 dfsan_label timeout_label, dfsan_label *ret_label) {
685   int ret = poll(fds, nfds, timeout);
686   if (ret >= 0) {
687     for (; nfds > 0; --nfds) {
688       dfsan_set_label(0, &fds[nfds - 1].revents, sizeof(fds[nfds - 1].revents));
689     }
690   }
691   *ret_label = 0;
692   return ret;
693 }
694 
695 SANITIZER_INTERFACE_ATTRIBUTE
696 int __dfsw_select(int nfds, fd_set *readfds, fd_set *writefds,
697                   fd_set *exceptfds, struct timeval *timeout,
698                   dfsan_label nfds_label, dfsan_label readfds_label,
699                   dfsan_label writefds_label, dfsan_label exceptfds_label,
700                   dfsan_label timeout_label, dfsan_label *ret_label) {
701   int ret = select(nfds, readfds, writefds, exceptfds, timeout);
702   // Clear everything (also on error) since their content is either set or
703   // undefined.
704   if (readfds) {
705     dfsan_set_label(0, readfds, sizeof(fd_set));
706   }
707   if (writefds) {
708     dfsan_set_label(0, writefds, sizeof(fd_set));
709   }
710   if (exceptfds) {
711     dfsan_set_label(0, exceptfds, sizeof(fd_set));
712   }
713   dfsan_set_label(0, timeout, sizeof(struct timeval));
714   *ret_label = 0;
715   return ret;
716 }
717 
718 SANITIZER_INTERFACE_ATTRIBUTE
719 int __dfsw_sched_getaffinity(pid_t pid, size_t cpusetsize, cpu_set_t *mask,
720                              dfsan_label pid_label,
721                              dfsan_label cpusetsize_label,
722                              dfsan_label mask_label, dfsan_label *ret_label) {
723   int ret = sched_getaffinity(pid, cpusetsize, mask);
724   if (ret == 0) {
725     dfsan_set_label(0, mask, cpusetsize);
726   }
727   *ret_label = 0;
728   return ret;
729 }
730 
731 SANITIZER_INTERFACE_ATTRIBUTE
732 int __dfsw_sigemptyset(sigset_t *set, dfsan_label set_label,
733                        dfsan_label *ret_label) {
734   int ret = sigemptyset(set);
735   dfsan_set_label(0, set, sizeof(sigset_t));
736   return ret;
737 }
738 
739 SANITIZER_INTERFACE_ATTRIBUTE
740 int __dfsw_sigaction(int signum, const struct sigaction *act,
741                      struct sigaction *oldact, dfsan_label signum_label,
742                      dfsan_label act_label, dfsan_label oldact_label,
743                      dfsan_label *ret_label) {
744   int ret = sigaction(signum, act, oldact);
745   if (oldact) {
746     dfsan_set_label(0, oldact, sizeof(struct sigaction));
747   }
748   *ret_label = 0;
749   return ret;
750 }
751 
752 SANITIZER_INTERFACE_ATTRIBUTE
753 int __dfsw_gettimeofday(struct timeval *tv, struct timezone *tz,
754                         dfsan_label tv_label, dfsan_label tz_label,
755                         dfsan_label *ret_label) {
756   int ret = gettimeofday(tv, tz);
757   if (tv) {
758     dfsan_set_label(0, tv, sizeof(struct timeval));
759   }
760   if (tz) {
761     dfsan_set_label(0, tz, sizeof(struct timezone));
762   }
763   *ret_label = 0;
764   return ret;
765 }
766 
767 SANITIZER_INTERFACE_ATTRIBUTE void *__dfsw_memchr(void *s, int c, size_t n,
768                                                   dfsan_label s_label,
769                                                   dfsan_label c_label,
770                                                   dfsan_label n_label,
771                                                   dfsan_label *ret_label) {
772   void *ret = memchr(s, c, n);
773   if (flags().strict_data_dependencies) {
774     *ret_label = ret ? s_label : 0;
775   } else {
776     size_t len =
777         ret ? reinterpret_cast<char *>(ret) - reinterpret_cast<char *>(s) + 1
778             : n;
779     *ret_label =
780         dfsan_union(dfsan_read_label(s, len), dfsan_union(s_label, c_label));
781   }
782   return ret;
783 }
784 
785 SANITIZER_INTERFACE_ATTRIBUTE char *__dfsw_strrchr(char *s, int c,
786                                                    dfsan_label s_label,
787                                                    dfsan_label c_label,
788                                                    dfsan_label *ret_label) {
789   char *ret = strrchr(s, c);
790   if (flags().strict_data_dependencies) {
791     *ret_label = ret ? s_label : 0;
792   } else {
793     *ret_label =
794         dfsan_union(dfsan_read_label(s, strlen(s) + 1),
795                     dfsan_union(s_label, c_label));
796   }
797 
798   return ret;
799 }
800 
801 SANITIZER_INTERFACE_ATTRIBUTE char *__dfsw_strstr(char *haystack, char *needle,
802                                                   dfsan_label haystack_label,
803                                                   dfsan_label needle_label,
804                                                   dfsan_label *ret_label) {
805   char *ret = strstr(haystack, needle);
806   if (flags().strict_data_dependencies) {
807     *ret_label = ret ? haystack_label : 0;
808   } else {
809     size_t len = ret ? ret + strlen(needle) - haystack : strlen(haystack) + 1;
810     *ret_label =
811         dfsan_union(dfsan_read_label(haystack, len),
812                     dfsan_union(dfsan_read_label(needle, strlen(needle) + 1),
813                                 dfsan_union(haystack_label, needle_label)));
814   }
815 
816   return ret;
817 }
818 
819 SANITIZER_INTERFACE_ATTRIBUTE int __dfsw_nanosleep(const struct timespec *req,
820                                                    struct timespec *rem,
821                                                    dfsan_label req_label,
822                                                    dfsan_label rem_label,
823                                                    dfsan_label *ret_label) {
824   int ret = nanosleep(req, rem);
825   *ret_label = 0;
826   if (ret == -1) {
827     // Interrupted by a signal, rem is filled with the remaining time.
828     dfsan_set_label(0, rem, sizeof(struct timespec));
829   }
830   return ret;
831 }
832 
833 SANITIZER_INTERFACE_ATTRIBUTE int
834 __dfsw_socketpair(int domain, int type, int protocol, int sv[2],
835                   dfsan_label domain_label, dfsan_label type_label,
836                   dfsan_label protocol_label, dfsan_label sv_label,
837                   dfsan_label *ret_label) {
838   int ret = socketpair(domain, type, protocol, sv);
839   *ret_label = 0;
840   if (ret == 0) {
841     dfsan_set_label(0, sv, sizeof(*sv) * 2);
842   }
843   return ret;
844 }
845 
846 // Type of the trampoline function passed to the custom version of
847 // dfsan_set_write_callback.
848 typedef void (*write_trampoline_t)(
849     void *callback,
850     int fd, const void *buf, ssize_t count,
851     dfsan_label fd_label, dfsan_label buf_label, dfsan_label count_label);
852 
853 // Calls to dfsan_set_write_callback() set the values in this struct.
854 // Calls to the custom version of write() read (and invoke) them.
855 static struct {
856   write_trampoline_t write_callback_trampoline = nullptr;
857   void *write_callback = nullptr;
858 } write_callback_info;
859 
860 SANITIZER_INTERFACE_ATTRIBUTE void
861 __dfsw_dfsan_set_write_callback(
862     write_trampoline_t write_callback_trampoline,
863     void *write_callback,
864     dfsan_label write_callback_label,
865     dfsan_label *ret_label) {
866   write_callback_info.write_callback_trampoline = write_callback_trampoline;
867   write_callback_info.write_callback = write_callback;
868 }
869 
870 SANITIZER_INTERFACE_ATTRIBUTE int
871 __dfsw_write(int fd, const void *buf, size_t count,
872              dfsan_label fd_label, dfsan_label buf_label,
873              dfsan_label count_label, dfsan_label *ret_label) {
874   if (write_callback_info.write_callback) {
875     write_callback_info.write_callback_trampoline(
876         write_callback_info.write_callback,
877         fd, buf, count,
878         fd_label, buf_label, count_label);
879   }
880 
881   *ret_label = 0;
882   return write(fd, buf, count);
883 }
884 } // namespace __dfsan
885 
886 // Type used to extract a dfsan_label with va_arg()
887 typedef int dfsan_label_va;
888 
889 // Formats a chunk either a constant string or a single format directive (e.g.,
890 // '%.3f').
891 struct Formatter {
892   Formatter(char *str_, const char *fmt_, size_t size_)
893       : str(str_), str_off(0), size(size_), fmt_start(fmt_), fmt_cur(fmt_),
894         width(-1) {}
895 
896   int format() {
897     char *tmp_fmt = build_format_string();
898     int retval =
899         snprintf(str + str_off, str_off < size ? size - str_off : 0, tmp_fmt,
900                  0 /* used only to avoid warnings */);
901     free(tmp_fmt);
902     return retval;
903   }
904 
905   template <typename T> int format(T arg) {
906     char *tmp_fmt = build_format_string();
907     int retval;
908     if (width >= 0) {
909       retval = snprintf(str + str_off, str_off < size ? size - str_off : 0,
910                         tmp_fmt, width, arg);
911     } else {
912       retval = snprintf(str + str_off, str_off < size ? size - str_off : 0,
913                         tmp_fmt, arg);
914     }
915     free(tmp_fmt);
916     return retval;
917   }
918 
919   char *build_format_string() {
920     size_t fmt_size = fmt_cur - fmt_start + 1;
921     char *new_fmt = (char *)malloc(fmt_size + 1);
922     assert(new_fmt);
923     internal_memcpy(new_fmt, fmt_start, fmt_size);
924     new_fmt[fmt_size] = '\0';
925     return new_fmt;
926   }
927 
928   char *str_cur() { return str + str_off; }
929 
930   size_t num_written_bytes(int retval) {
931     if (retval < 0) {
932       return 0;
933     }
934 
935     size_t num_avail = str_off < size ? size - str_off : 0;
936     if (num_avail == 0) {
937       return 0;
938     }
939 
940     size_t num_written = retval;
941     // A return value of {v,}snprintf of size or more means that the output was
942     // truncated.
943     if (num_written >= num_avail) {
944       num_written -= num_avail;
945     }
946 
947     return num_written;
948   }
949 
950   char *str;
951   size_t str_off;
952   size_t size;
953   const char *fmt_start;
954   const char *fmt_cur;
955   int width;
956 };
957 
958 // Formats the input and propagates the input labels to the output. The output
959 // is stored in 'str'. 'size' bounds the number of output bytes. 'format' and
960 // 'ap' are the format string and the list of arguments for formatting. Returns
961 // the return value vsnprintf would return.
962 //
963 // The function tokenizes the format string in chunks representing either a
964 // constant string or a single format directive (e.g., '%.3f') and formats each
965 // chunk independently into the output string. This approach allows to figure
966 // out which bytes of the output string depends on which argument and thus to
967 // propagate labels more precisely.
968 //
969 // WARNING: This implementation does not support conversion specifiers with
970 // positional arguments.
971 static int format_buffer(char *str, size_t size, const char *fmt,
972                          dfsan_label *va_labels, dfsan_label *ret_label,
973                          va_list ap) {
974   Formatter formatter(str, fmt, size);
975 
976   while (*formatter.fmt_cur) {
977     formatter.fmt_start = formatter.fmt_cur;
978     formatter.width = -1;
979     int retval = 0;
980 
981     if (*formatter.fmt_cur != '%') {
982       // Ordinary character. Consume all the characters until a '%' or the end
983       // of the string.
984       for (; *(formatter.fmt_cur + 1) && *(formatter.fmt_cur + 1) != '%';
985            ++formatter.fmt_cur) {}
986       retval = formatter.format();
987       dfsan_set_label(0, formatter.str_cur(),
988                       formatter.num_written_bytes(retval));
989     } else {
990       // Conversion directive. Consume all the characters until a conversion
991       // specifier or the end of the string.
992       bool end_fmt = false;
993       for (; *formatter.fmt_cur && !end_fmt; ) {
994         switch (*++formatter.fmt_cur) {
995         case 'd':
996         case 'i':
997         case 'o':
998         case 'u':
999         case 'x':
1000         case 'X':
1001           switch (*(formatter.fmt_cur - 1)) {
1002           case 'h':
1003             // Also covers the 'hh' case (since the size of the arg is still
1004             // an int).
1005             retval = formatter.format(va_arg(ap, int));
1006             break;
1007           case 'l':
1008             if (formatter.fmt_cur - formatter.fmt_start >= 2 &&
1009                 *(formatter.fmt_cur - 2) == 'l') {
1010               retval = formatter.format(va_arg(ap, long long int));
1011             } else {
1012               retval = formatter.format(va_arg(ap, long int));
1013             }
1014             break;
1015           case 'q':
1016             retval = formatter.format(va_arg(ap, long long int));
1017             break;
1018           case 'j':
1019             retval = formatter.format(va_arg(ap, intmax_t));
1020             break;
1021           case 'z':
1022           case 't':
1023             retval = formatter.format(va_arg(ap, size_t));
1024             break;
1025           default:
1026             retval = formatter.format(va_arg(ap, int));
1027           }
1028           dfsan_set_label(*va_labels++, formatter.str_cur(),
1029                           formatter.num_written_bytes(retval));
1030           end_fmt = true;
1031           break;
1032 
1033         case 'a':
1034         case 'A':
1035         case 'e':
1036         case 'E':
1037         case 'f':
1038         case 'F':
1039         case 'g':
1040         case 'G':
1041           if (*(formatter.fmt_cur - 1) == 'L') {
1042             retval = formatter.format(va_arg(ap, long double));
1043           } else {
1044             retval = formatter.format(va_arg(ap, double));
1045           }
1046           dfsan_set_label(*va_labels++, formatter.str_cur(),
1047                           formatter.num_written_bytes(retval));
1048           end_fmt = true;
1049           break;
1050 
1051         case 'c':
1052           retval = formatter.format(va_arg(ap, int));
1053           dfsan_set_label(*va_labels++, formatter.str_cur(),
1054                           formatter.num_written_bytes(retval));
1055           end_fmt = true;
1056           break;
1057 
1058         case 's': {
1059           char *arg = va_arg(ap, char *);
1060           retval = formatter.format(arg);
1061           va_labels++;
1062           internal_memcpy(shadow_for(formatter.str_cur()), shadow_for(arg),
1063                           sizeof(dfsan_label) *
1064                               formatter.num_written_bytes(retval));
1065           end_fmt = true;
1066           break;
1067         }
1068 
1069         case 'p':
1070           retval = formatter.format(va_arg(ap, void *));
1071           dfsan_set_label(*va_labels++, formatter.str_cur(),
1072                           formatter.num_written_bytes(retval));
1073           end_fmt = true;
1074           break;
1075 
1076         case 'n': {
1077           int *ptr = va_arg(ap, int *);
1078           *ptr = (int)formatter.str_off;
1079           va_labels++;
1080           dfsan_set_label(0, ptr, sizeof(ptr));
1081           end_fmt = true;
1082           break;
1083         }
1084 
1085         case '%':
1086           retval = formatter.format();
1087           dfsan_set_label(0, formatter.str_cur(),
1088                           formatter.num_written_bytes(retval));
1089           end_fmt = true;
1090           break;
1091 
1092         case '*':
1093           formatter.width = va_arg(ap, int);
1094           va_labels++;
1095           break;
1096 
1097         default:
1098           break;
1099         }
1100       }
1101     }
1102 
1103     if (retval < 0) {
1104       return retval;
1105     }
1106 
1107     formatter.fmt_cur++;
1108     formatter.str_off += retval;
1109   }
1110 
1111   *ret_label = 0;
1112 
1113   // Number of bytes written in total.
1114   return formatter.str_off;
1115 }
1116 
1117 extern "C" {
1118 SANITIZER_INTERFACE_ATTRIBUTE
1119 int __dfsw_sprintf(char *str, const char *format, dfsan_label str_label,
1120                    dfsan_label format_label, dfsan_label *va_labels,
1121                    dfsan_label *ret_label, ...) {
1122   va_list ap;
1123   va_start(ap, ret_label);
1124   int ret = format_buffer(str, ~0ul, format, va_labels, ret_label, ap);
1125   va_end(ap);
1126   return ret;
1127 }
1128 
1129 SANITIZER_INTERFACE_ATTRIBUTE
1130 int __dfsw_snprintf(char *str, size_t size, const char *format,
1131                     dfsan_label str_label, dfsan_label size_label,
1132                     dfsan_label format_label, dfsan_label *va_labels,
1133                     dfsan_label *ret_label, ...) {
1134   va_list ap;
1135   va_start(ap, ret_label);
1136   int ret = format_buffer(str, size, format, va_labels, ret_label, ap);
1137   va_end(ap);
1138   return ret;
1139 }
1140 
1141 // Default empty implementations (weak). Users should redefine them.
1142 SANITIZER_INTERFACE_WEAK_DEF(void, __sanitizer_cov_trace_pc_guard, u32 *) {}
1143 SANITIZER_INTERFACE_WEAK_DEF(void, __sanitizer_cov_trace_pc_guard_init, u32 *,
1144                              u32 *) {}
1145 SANITIZER_INTERFACE_WEAK_DEF(void, __sanitizer_cov_pcs_init, void) {}
1146 SANITIZER_INTERFACE_WEAK_DEF(void, __sanitizer_cov_trace_pc_indir, void) {}
1147 
1148 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_cmp, void) {}
1149 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_cmp1, void) {}
1150 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_cmp2, void) {}
1151 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_cmp4, void) {}
1152 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_cmp8, void) {}
1153 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_const_cmp1,
1154                              void) {}
1155 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_const_cmp2,
1156                              void) {}
1157 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_const_cmp4,
1158                              void) {}
1159 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_const_cmp8,
1160                              void) {}
1161 SANITIZER_INTERFACE_WEAK_DEF(void, __dfsw___sanitizer_cov_trace_switch, void) {}
1162 }  // extern "C"
1163