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