Moar string interceptors: strstr, strcasestr, strcspn, strpbrk

[m.guseva]

Summary

This is a patch for Issue 346: moar string interceptors: strstr, strcasestr, strcspn, strpbrk

New interceptors check all bytes in second argument but in first argument check range is based on return value. Some of existing interceptors also uses this approach (like asan strchr) while other just check the full length of s1(like strstr in tsan). I prefered the first approach as more accurate. What do you find more convinient for these interceptors?

To solve overlap of asan-specific replace_str flag and new common replace_strstr/replace_strcspn flags I've chosen more priority for replace_str flag: when it is false all str interceptors(including new ones) are disabled.

[m.guseva]

Ping.

[glider]

Sorry, looks like this has slipped through the cracks. I'll take a look now.

Can you please upload the whole context using `svn diff
--diff-cmd=diff -x -U999999` next time?

[glider]

There are several places where the shadow checks are too aggressive compared to what the glibc implementations of these functions do.
Because the standards are unlikely to define which optimizations are applicable to each function, I think we'd better be more conservative here.

lib/sanitizer_common/sanitizer_common_interceptors.inc

260

If haystack is shorter than the needle, there's no point in reading the whole contents of needle.
The generic glibc implementation of strstr() indeed makes this optimization (not sure about the SSE-based ones).

302

Only in the case when s1 is not an empty string, otherwise there's no point in reading s2.

312

I never realized all these functions crash if we pass NULL as one of the arguments.
Fortunately we don't need to handle that.

320

This is a bit pessimistic. For the following call:

strpbrk("foobar", "fb")

only the first byte of |s2| will be read.

lib/sanitizer_common/sanitizer_flags.cc

161 ↗

(On Diff #15621)

It's not clear why replace_strcspn controls the behavior of strpbrk()
I suggest to either rename the flag to "replace_strcspn_strpbrk" or introduce the "replace_strpbrk" flag.
My dream is to ultimately have a dynamically generated "replace_xyz" flag for almost every interceptor, so the second option is better IMO.

[ygribov]

Because the standards are unlikely to define which optimizations are applicable to each function,
I think we'd better be more conservative here.

That's questionable. Some users (e.g. me) would very much prefer to get a warning even about potential errors. Can we have a flag for this like we already have for strict_memcmp and alloc_dealloc_mismatch?

[joerg]

The interceptors should enforce the contract. For all string functions without explicit size argument, this means that the input is a valid nul-terminated string. Different implementations take different optimisation short cuts, but depending on such implementation details is part of what the interceptors should expose.

[ygribov]

The interceptors should enforce the contract.

I believe sanitizer folks are in unfortunate situation where they have to deal with tons of legacy crap which relies heavily on glibc quirks (there's a lot of code like this, some in binary form with no ability to recompile). So I can feel their pain. On the other hand we shouldn't force users which deal with saner codebases (me or Joerg) to use overly conservative tool.

[kcc]

As long as the aggressive behavior can be disabled by a flag (similar to what we have to do with strict_memcmp, or just by disabling the whole interceptor logic) I am ok with this.

[ygribov]

As long as the aggressive behavior can be disabled by a flag ... I am ok with this.

I guess we can do this for already existing interceptors as well? What setting should be default?

[kcc]

In D6056#13, @ygribov wrote:

As long as the aggressive behavior can be disabled by a flag ... I am ok with this.

I guess we can do this for already existing interceptors as well?

Well, if you want to ...

What setting should be default?

I am fine with the most aggressive default

[m.guseva]

Hi all
I've uploaded another patch for new string interceptors. As it was discussed in comments above I've introduced new flag "strict_str" in order to control aggressive checks for strings arguments. So I've also updated all existing interceptors with "const char*" arguments and added strict checks wherever they were missed.

[kcc]

I like the whole thing, the more interceptors the better.
Statistically, such large changes in interceptors are risky, so please don't commit before the end of holidays in RU, even if you get LGTM before that.
Please also add a documentation blob to the commit message so that we can later copy-paste it into
https://code.google.com/p/address-sanitizer/wiki/Flags

glider, please apply this patch and try it on chrome.

lib/asan/asan_rtl.cc

130 ↗

(On Diff #17561)

Is this necessary?
Such things are sometimes unavoidable, but they complicate the code, the understanding and the documentation.
I'd rather prefer users to add more flags...

test/asan/TestCases/atoi_strict.c

4

also, add a 3-rd run with no flag so that we test the default behaviour.
same for other tests.

[m.guseva]

lib/asan/asan_rtl.cc

130 ↗

(On Diff #17561)

If we remove it there will be still some uncertainty in documentation. From user's point of view it's not clear which interceptors replace_str controls and which not. I supposed it should control all of them and overwrite other replace_str* flags in non default case. Otherwise I think it must be specified that replace_str controls only some subset of string interceptiors while for rest there are custom flags. Shouldn't we to fix the replace_str description for that case?

[m.guseva]

test/asan/TestCases/atoi_strict.c

4

Added in Diff 17587.

[glider]

lib/msan/msan_interceptors.cc

96

Am I right that you're only using this macro with n=0?

lib/sanitizer_common/sanitizer_common_interceptors.inc

328

Please put a space before the colon.

2431

I suggest putting the UNUSED function under the #ifdef below and remove the UNUSED attribute.

2457

You don't need IsValidStrtolBase anymore:

bool is_valid_base = (base == 0) || (2 <= base && base <= 36);
if (is_valid_base) {
  ...
}
COMMON_INTERCEPTOR_READ_STRING(ctx, nptr, is_valid_base ? ...)

4796

Chrome's net_unittests crash for me with the following stacktrace:

Program received signal SIGSEGV, Segmentation fault.
__sanitizer::internal_strlen (s=s@entry=0x0)
    at /usr/local/google/ssd/llvm/projects/compiler-rt/lib/sanitizer_common/sanitizer_libc.cc:155
155	  while (s[i]) i++;
(gdb) bt
#0  __sanitizer::internal_strlen (s=s@entry=0x0)
    at /usr/local/google/ssd/llvm/projects/compiler-rt/lib/sanitizer_common/sanitizer_libc.cc:155
#1  0x000000000059f080 in __interceptor_dlopen (filename=filename@entry=0x0, 
    flag=flag@entry=1)
    at /usr/local/google/ssd/llvm/projects/compiler-rt/lib/asan/../sanitizer_common/sanitizer_common_interceptors.inc:4759
#2  0x00007ffff5167ec2 in pr_FindSymbolInProg (
    name=0x7ffff517f29f "nspr_use_zone_allocator") at prmem.c:98
#3  0x00007ffff5167fad in _PR_InitZones () at prmem.c:154
...

We'll need to add yet another dlopen() test once we figure out what's wrong.

lib/sanitizer_common/sanitizer_flags.cc

167 ↗

(On Diff #17587)

I think we need to be consistent in naming the flags that disable interceptors.
There's already "intercept_tls_get_addr", so we either need to change it (no need to do that in this CL) or rename your flags into "intercept_something".
(Note that "intrin" in "replace_intrin" isn't a function name, so you don't need to be consistent with that).

173 ↗

(On Diff #17587)

I think "strict_string_checks" is more informative, but that's bikeshedding.

lib/tsan/rtl/tsan_interceptors.cc

245

nit: spare whitespace before ')'

test/asan/TestCases/atoi_strict.c

15

Can you please add a space after the 'CHECK:' label here and below?

test/asan/TestCases/strstr-1.c

14

Note that you can use #ifdef directives and different check prefixes to combine related test cases into one file.

test/asan/Unit/lit.site.cfg.in

33

Why do you need this line?

test/sanitizer_common/TestCases/strcasestr.c

4

Why? Is that because strcasestr() is unavailable on Windows?

[ygribov]

lib/msan/msan_interceptors.cc

96

I think Maria wanted added n to anticipate future work on conservative length checking.

lib/sanitizer_common/sanitizer_common_interceptors.inc

4796

Looks like COMMON_INTERCEPTOR_READ_STRING needs a check for NULL.

test/asan/Unit/lit.site.cfg.in

33

AFAIR unit tests are not strict_str-safe.

[glider]

test/asan/Unit/lit.site.cfg.in

33

Is that something we may want to fix?
If so, please file a bug and add a TODO

[m.guseva]

lib/msan/msan_interceptors.cc

96

ygribov
I think Maria wanted added n to anticipate future work on conservative length checking.

Exactly.

lib/sanitizer_common/sanitizer_common_interceptors.inc

2431

These functions were moved from asan_interceptors.cc source where they are used in strtol-related interceptors. Some of them(strtol\atoi\atol) are not under any if directive. Is there any better place where to put this helpers to be shared by asan and common interceptors?

4796

Here is my bug. It needs check for NULL before COMMON_INTERCEPTOR_READ_STRING. Cause dlopen accepts null filename pointer as a normal argument and then there is nothing to check here. In other interceptors (e.g strsmthing) NULL pointer arg is an error case. So the program should fail with segfault. And it will in internal_strlen. I'll recheck other modified interceptors for such cases.

lib/sanitizer_common/sanitizer_flags.cc

167 ↗

(On Diff #17587)

The "replace_" flags were suggested in original issue. I agree it should be aligned with existing one.

test/asan/Unit/lit.site.cfg.in

33

ygribov
AFAIR unit tests are not strict_str-safe.

Right. I saw fails caused by checks relied on standard optimizations in string functions. And I didn't want to remove these tests at all cause they seems meaningful.

[glider]

lib/sanitizer_common/sanitizer_common_interceptors.inc

2431

Oh, sorry, I've missed that.
It's fine to keep these functions as is (I still think IsValidStrtolBase is unnecessary and can be removed)

lib/sanitizer_common/sanitizer_flags.cc

167 ↗

(On Diff #17587)

Let's see what Kostya and Alexey think.

[kcc]

Sorry for the delay in the review. Frankly, I've lost the track of who is waiting for whose reply.
Large changes are hard to review, I wonder If you could split it into a few smaller ones?

[ygribov]

I thought Maria was going to upload a new revision.

[m.guseva]

Hi all
I've prepared new patch with fixes suggested in glider's review. I've also done merge with recent trunk. Please take a look.

Regarding splitting the patch:
indeed there two main changes in this patch:

1. new strict_string_checks run-time flag introduced and applied for existing common, asan, msan and tsan interceptors. New asan tests added for this flag.
2. new interceptors added as it was suggested in original request(strstr, strcasestr, strcspn, strpbrk) with strict_string_checks flag supported. New common and asan tests added for them.

If it wil be easier I can split patch in these two correspondingly. If so should I attach both to this revision? Or create new one?

[m.guseva]

test/sanitizer_common/TestCases/strcasestr.c

4

Yes. I've referenced printf-*.c tests as printf is not intercepted on Windows. Althouth there is a difference because strcasestr is unavailable at all on Windows so it won't even compile. What do you think is more correct way to exclude it?

[kcc]

Yes, please split it.
If there is a good logical split point -- always try to do that.
Code reviews often have O(N^2) complexity, you don't want N to be too large.

[kcc]

(Just create two fresh patches and abandon this one)

[m.guseva]

Submitted first separate patch: New "strict_string_checks" run-time flag http://reviews.llvm.org/D7123
The second one(new interceptors) depends on this so I'll submit it when the first review passed.

[m.guseva]

Ok, now I am abandoning this revision - let's move all discussions to new splitted patches. So I'm looking for your feedback in http://reviews.llvm.org/D7123 regarding strict_string_checks flag.

lib/asan/asan_interceptors.cc

#define ACCESS_MEMORY_RANGE(ctx, offset, size, isWrite) do {            \
     }                                                                   \
   } while (0)
 
 #define ASAN_READ_RANGE(ctx, offset, size) \
   ACCESS_MEMORY_RANGE(ctx, offset, size, false)
 #define ASAN_WRITE_RANGE(ctx, offset, size) \
   ACCESS_MEMORY_RANGE(ctx, offset, size, true)
 
+#define ASAN_READ_STRING(ctx, s, n)                             \
+  ASAN_READ_RANGE((ctx), (s),                                   \
+    common_flags()->strict_string_checks ? internal_strlen(s) + 1 : (n) )
+
 // Behavior of functions like "memcpy" or "strcpy" is undefined
 // if memory intervals overlap. We report error in this case.
 // Macro is used to avoid creation of new frames.
 static inline bool RangesOverlap(const char *offset1, uptr length1,
                                  const char *offset2, uptr length2) {
   return !((offset1 + length1 <= offset2) || (offset2 + length2 <= offset1));
 }
 #define CHECK_RANGES_OVERLAP(name, _offset1, length1, _offset2, length2) do { \
INTERCEPTOR(char*, strchr, const char *str, int c) {
   // used.
   if (asan_init_is_running) {
     return REAL(strchr)(str, c);
   }
   ENSURE_ASAN_INITED();
   char *result = REAL(strchr)(str, c);
   if (flags()->replace_str) {
     uptr bytes_read = (result ? result - str : REAL(strlen)(str)) + 1;
-    ASAN_READ_RANGE(ctx, str, bytes_read);
+    ASAN_READ_STRING(ctx, str, bytes_read);
   }
   return result;
 }
 
 #if ASAN_INTERCEPT_INDEX
 # if ASAN_USE_ALIAS_ATTRIBUTE_FOR_INDEX
 INTERCEPTOR(char*, index, const char *string, int c)
   ALIAS(WRAPPER_NAME(strchr));

 INTERCEPTOR(char*, strcat, char *to, const char *from) {  // NOLINT
   void *ctx;
   ASAN_INTERCEPTOR_ENTER(ctx, strcat);  // NOLINT
   ENSURE_ASAN_INITED();
   if (flags()->replace_str) {
     uptr from_length = REAL(strlen)(from);
     ASAN_READ_RANGE(ctx, from, from_length + 1);
     uptr to_length = REAL(strlen)(to);
-    ASAN_READ_RANGE(ctx, to, to_length);
+    ASAN_READ_STRING(ctx, to, to_length);
     ASAN_WRITE_RANGE(ctx, to + to_length, from_length + 1);
     // If the copying actually happens, the |from| string should not overlap
     // with the resulting string starting at |to|, which has a length of
     // to_length + from_length + 1.
     if (from_length > 0) {
       CHECK_RANGES_OVERLAP("strcat", to, from_length + to_length + 1,
                            from, from_length + 1);
     }
   }
   return REAL(strcat)(to, from);  // NOLINT
 }
 
 INTERCEPTOR(char*, strncat, char *to, const char *from, uptr size) {
   void *ctx;
   ASAN_INTERCEPTOR_ENTER(ctx, strncat);
   ENSURE_ASAN_INITED();
   if (flags()->replace_str) {
     uptr from_length = MaybeRealStrnlen(from, size);
     uptr copy_length = Min(size, from_length + 1);
     ASAN_READ_RANGE(ctx, from, copy_length);
     uptr to_length = REAL(strlen)(to);
-    ASAN_READ_RANGE(ctx, to, to_length);
+    ASAN_READ_STRING(ctx, to, to_length);
     ASAN_WRITE_RANGE(ctx, to + to_length, from_length + 1);
     if (from_length > 0) {
       CHECK_RANGES_OVERLAP("strncat", to, to_length + copy_length + 1,
                            from, copy_length);
     }
   }
   return REAL(strncat)(to, from, size);
 }
INTERCEPTOR(uptr, strnlen, const char *s, uptr maxlen) {
   uptr length = REAL(strnlen)(s, maxlen);
   if (flags()->replace_str) {
     ASAN_READ_RANGE(ctx, s, Min(length + 1, maxlen));
   }
   return length;
 }
 #endif  // ASAN_INTERCEPT_STRNLEN
 
-static inline bool IsValidStrtolBase(int base) {
-  return (base == 0) || (2 <= base && base <= 36);
-}
-
-static inline void FixRealStrtolEndptr(const char *nptr, char **endptr) {
-  CHECK(endptr);
-  if (nptr == *endptr) {
-    // No digits were found at strtol call, we need to find out the last
-    // symbol accessed by strtoll on our own.
-    // We get this symbol by skipping leading blanks and optional +/- sign.
-    while (IsSpace(*nptr)) nptr++;
-    if (*nptr == '+' || *nptr == '-') nptr++;
-    *endptr = const_cast<char *>(nptr);
-  }
-  CHECK(*endptr >= nptr);
-}
-
 INTERCEPTOR(long, strtol, const char *nptr,  // NOLINT
             char **endptr, int base) {
   void *ctx;
   ASAN_INTERCEPTOR_ENTER(ctx, strtol);
   ENSURE_ASAN_INITED();
   if (!flags()->replace_str) {
     return REAL(strtol)(nptr, endptr, base);
   }
   char *real_endptr;
   long result = REAL(strtol)(nptr, &real_endptr, base);  // NOLINT
-  if (endptr != 0) {
-    *endptr = real_endptr;
-  }
-  if (IsValidStrtolBase(base)) {
-    FixRealStrtolEndptr(nptr, &real_endptr);
-    ASAN_READ_RANGE(ctx, nptr, (real_endptr - nptr) + 1);
-  }
+  StrtolFixAndCheck(ctx, nptr, endptr, real_endptr, base);
   return result;
 }
 
 INTERCEPTOR(int, atoi, const char *nptr) {
   void *ctx;
   ASAN_INTERCEPTOR_ENTER(ctx, atoi);
 #if SANITIZER_MAC
   if (UNLIKELY(!asan_inited)) return REAL(atoi)(nptr);
 #endif
   ENSURE_ASAN_INITED();
   if (!flags()->replace_str) {
     return REAL(atoi)(nptr);
   }
   char *real_endptr;
   // "man atoi" tells that behavior of atoi(nptr) is the same as
   // strtol(nptr, 0, 10), i.e. it sets errno to ERANGE if the
   // parsed integer can't be stored in *long* type (even if it's
   // different from int). So, we just imitate this behavior.
   int result = REAL(strtol)(nptr, &real_endptr, 10);
   FixRealStrtolEndptr(nptr, &real_endptr);
-  ASAN_READ_RANGE(ctx, nptr, (real_endptr - nptr) + 1);
+  ASAN_READ_STRING(ctx, nptr, (real_endptr - nptr) + 1);
   return result;
 }
 
 INTERCEPTOR(long, atol, const char *nptr) {  // NOLINT
   void *ctx;
   ASAN_INTERCEPTOR_ENTER(ctx, atol);
 #if SANITIZER_MAC
   if (UNLIKELY(!asan_inited)) return REAL(atol)(nptr);
 #endif
   ENSURE_ASAN_INITED();
   if (!flags()->replace_str) {
     return REAL(atol)(nptr);
   }
   char *real_endptr;
   long result = REAL(strtol)(nptr, &real_endptr, 10);  // NOLINT
   FixRealStrtolEndptr(nptr, &real_endptr);
-  ASAN_READ_RANGE(ctx, nptr, (real_endptr - nptr) + 1);
+  ASAN_READ_STRING(ctx, nptr, (real_endptr - nptr) + 1);
   return result;
 }
 
 #if ASAN_INTERCEPT_ATOLL_AND_STRTOLL
 INTERCEPTOR(long long, strtoll, const char *nptr,  // NOLINT
             char **endptr, int base) {
   void *ctx;
   ASAN_INTERCEPTOR_ENTER(ctx, strtoll);
   ENSURE_ASAN_INITED();
   if (!flags()->replace_str) {
     return REAL(strtoll)(nptr, endptr, base);
   }
   char *real_endptr;
   long long result = REAL(strtoll)(nptr, &real_endptr, base);  // NOLINT
-  if (endptr != 0) {
-    *endptr = real_endptr;
-  }
-  // If base has unsupported value, strtoll can exit with EINVAL
-  // without reading any characters. So do additional checks only
-  // if base is valid.
-  if (IsValidStrtolBase(base)) {
-    FixRealStrtolEndptr(nptr, &real_endptr);
-    ASAN_READ_RANGE(ctx, nptr, (real_endptr - nptr) + 1);
-  }
+  StrtolFixAndCheck(ctx, nptr, endptr, real_endptr, base);
   return result;
 }
 
 INTERCEPTOR(long long, atoll, const char *nptr) {  // NOLINT
   void *ctx;
   ASAN_INTERCEPTOR_ENTER(ctx, atoll);
   ENSURE_ASAN_INITED();
   if (!flags()->replace_str) {
     return REAL(atoll)(nptr);
   }
   char *real_endptr;
   long long result = REAL(strtoll)(nptr, &real_endptr, 10);  // NOLINT
   FixRealStrtolEndptr(nptr, &real_endptr);
-  ASAN_READ_RANGE(ctx, nptr, (real_endptr - nptr) + 1);
+  ASAN_READ_STRING(ctx, nptr, (real_endptr - nptr) + 1);
   return result;
 }
 #endif  // ASAN_INTERCEPT_ATOLL_AND_STRTOLL
 
 static void AtCxaAtexit(void *unused) {
   (void)unused;
   StopInitOrderChecking();
 }

lib/msan/msan_interceptors.cc

 
 // Check that [x, x+n) range is unpoisoned unless we are in a nested
 // interceptor.
 #define CHECK_UNPOISONED(x, n)                             \
   do {                                                     \
     if (!IsInInterceptorScope()) CHECK_UNPOISONED_0(x, n); \
   } while (0);
 
+#define CHECK_UNPOISONED_STRING(x, n)                           \
+  CHECK_UNPOISONED((x),                                         \
+    common_flags()->strict_string_checks ? internal_strlen(x) + 1 : (n) )
+
 INTERCEPTOR(SIZE_T, fread, void *ptr, SIZE_T size, SIZE_T nmemb, void *file) {
   ENSURE_MSAN_INITED();
   SIZE_T res = REAL(fread)(ptr, size, nmemb, file);
   if (res > 0)
     __msan_unpoison(ptr, res *size);
   return res;
 }
 
 #if !SANITIZER_FREEBSD
 INTERCEPTOR(SIZE_T, fread_unlocked, void *ptr, SIZE_T size, SIZE_T nmemb,
             void *file) {
   ENSURE_MSAN_INITED();
   SIZE_T res = REAL(fread_unlocked)(ptr, size, nmemb, file);
   if (res > 0)
     __msan_unpoison(ptr, res *size);
   return res;
 }
 #define MSAN_MAYBE_INTERCEPT_FREAD_UNLOCKED INTERCEPT_FUNCTION(fread_unlocked)
 #else
 #define MSAN_MAYBE_INTERCEPT_FREAD_UNLOCKED
 #endif
 
 INTERCEPTOR(SSIZE_T, readlink, const char *path, char *buf, SIZE_T bufsiz) {
   ENSURE_MSAN_INITED();
+  CHECK_UNPOISONED_STRING(path, 0)
   SSIZE_T res = REAL(readlink)(path, buf, bufsiz);
   if (res > 0)
     __msan_unpoison(buf, res);
   return res;
 }
 
 INTERCEPTOR(void *, memcpy, void *dest, const void *src, SIZE_T n) {
   return __msan_memcpy(dest, src, n);

 INTERCEPTOR(SIZE_T, strnlen, const char *s, SIZE_T n) {
   ENSURE_MSAN_INITED();
   SIZE_T res = REAL(strnlen)(s, n);
   SIZE_T scan_size = (res == n) ? res : res + 1;
   CHECK_UNPOISONED(s, scan_size);
   return res;
 }
 
-// FIXME: Add stricter shadow checks in str* interceptors (ex.: strcpy should
-// check the shadow of the terminating \0 byte).
-
 INTERCEPTOR(char *, strcpy, char *dest, const char *src) {  // NOLINT
   ENSURE_MSAN_INITED();
   GET_STORE_STACK_TRACE;
   SIZE_T n = REAL(strlen)(src);
+  CHECK_UNPOISONED_STRING(src + n, 0);
   char *res = REAL(strcpy)(dest, src);  // NOLINT
   CopyPoison(dest, src, n + 1, &stack);
   return res;
 }
 
 INTERCEPTOR(char *, strncpy, char *dest, const char *src, SIZE_T n) {  // NOLINT
   ENSURE_MSAN_INITED();
   GET_STORE_STACK_TRACE;
   SIZE_T copy_size = REAL(strnlen)(src, n);
   if (copy_size < n)
     copy_size++;  // trailing \0
   char *res = REAL(strncpy)(dest, src, n);  // NOLINT
   CopyPoison(dest, src, copy_size, &stack);
   __msan_unpoison(dest + copy_size, n - copy_size);
   return res;
 }
 
 INTERCEPTOR(char *, stpcpy, char *dest, const char *src) {  // NOLINT
   ENSURE_MSAN_INITED();
   GET_STORE_STACK_TRACE;
   SIZE_T n = REAL(strlen)(src);
+  CHECK_UNPOISONED_STRING(src + n, 0);
   char *res = REAL(stpcpy)(dest, src);  // NOLINT
   CopyPoison(dest, src, n + 1, &stack);
   return res;
 }
 
 INTERCEPTOR(char *, strdup, char *src) {
   ENSURE_MSAN_INITED();
   GET_STORE_STACK_TRACE;
   // On FreeBSD strdup() leverages strlen().
   InterceptorScope interceptor_scope;
   SIZE_T n = REAL(strlen)(src);
+  CHECK_UNPOISONED_STRING(src + n, 0);
   char *res = REAL(strdup)(src);
   CopyPoison(res, src, n + 1, &stack);
   return res;
 }
 
 #if !SANITIZER_FREEBSD
 INTERCEPTOR(char *, __strdup, char *src) {
   ENSURE_MSAN_INITED();
   GET_STORE_STACK_TRACE;
   SIZE_T n = REAL(strlen)(src);
+  CHECK_UNPOISONED_STRING(src + n, 0);
   char *res = REAL(__strdup)(src);
   CopyPoison(res, src, n + 1, &stack);
   return res;
 }
 #define MSAN_MAYBE_INTERCEPT___STRDUP INTERCEPT_FUNCTION(__strdup)
 #else
 #define MSAN_MAYBE_INTERCEPT___STRDUP
 #endif
INTERCEPTOR(char *, gcvt, double number, SIZE_T ndigit, char *buf) {
   return res;
 }
 
 INTERCEPTOR(char *, strcat, char *dest, const char *src) {  // NOLINT
   ENSURE_MSAN_INITED();
   GET_STORE_STACK_TRACE;
   SIZE_T src_size = REAL(strlen)(src);
   SIZE_T dest_size = REAL(strlen)(dest);
+  CHECK_UNPOISONED_STRING(src + src_size, 0);
+  CHECK_UNPOISONED_STRING(dest + dest_size, 0);
   char *res = REAL(strcat)(dest, src);  // NOLINT
   CopyPoison(dest + dest_size, src, src_size + 1, &stack);
   return res;
 }
 
 INTERCEPTOR(char *, strncat, char *dest, const char *src, SIZE_T n) {  // NOLINT
   ENSURE_MSAN_INITED();
   GET_STORE_STACK_TRACE;
   SIZE_T dest_size = REAL(strlen)(dest);
   SIZE_T copy_size = REAL(strnlen)(src, n);
+  CHECK_UNPOISONED_STRING(dest + dest_size, 0);
   char *res = REAL(strncat)(dest, src, n);  // NOLINT
   CopyPoison(dest + dest_size, src, copy_size, &stack);
   __msan_unpoison(dest + dest_size + copy_size, 1); // \0
   return res;
 }
 
 // Hack: always pass nptr and endptr as part of __VA_ARGS_ to avoid having to
 // deal with empty __VA_ARGS__ in the case of INTERCEPTOR_STRTO.
for (; *envp; ++envp) {
     __msan_unpoison(*envp, REAL(strlen)(*envp) + 1);
   }
   // Trailing NULL pointer.
   __msan_unpoison(envp, sizeof(*envp));
 }
 
 INTERCEPTOR(int, setenv, const char *name, const char *value, int overwrite) {
   ENSURE_MSAN_INITED();
+  CHECK_UNPOISONED_STRING(name, 0)
   int res = REAL(setenv)(name, value, overwrite);
   if (!res) UnpoisonEnviron();
   return res;
 }
 
 INTERCEPTOR(int, putenv, char *string) {
   ENSURE_MSAN_INITED();
   int res = REAL(putenv)(string);

lib/sanitizer_common/sanitizer_common_interceptors.inc

 #define COMMON_INTERCEPTOR_ENTER_NOIGNORE(ctx, ...) \
   COMMON_INTERCEPTOR_ENTER(ctx, __VA_ARGS__)
 #endif
 
 #ifndef COMMON_INTERCEPTOR_NOTHING_IS_INITIALIZED
 #define COMMON_INTERCEPTOR_NOTHING_IS_INITIALIZED (0)
 #endif
 
+#define COMMON_INTERCEPTOR_READ_STRING(ctx, s, n)                   \
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, s,                           \
+      common_flags()->strict_string_checks ? internal_strlen(s) + 1 : n )
+
 struct FileMetadata {
   // For open_memstream().
   char **addr;
   SIZE_T *size;
 };
 
 struct CommonInterceptorMetadata {
   enum {
UNUSED static void DeleteInterceptorMetadata(void *addr) {
   CHECK(h.exists());
 }
 #endif  // SI_NOT_WINDOWS
 
 #if SANITIZER_INTERCEPT_TEXTDOMAIN
 INTERCEPTOR(char*, textdomain, const char *domainname) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, textdomain, domainname);
-  char* domain = REAL(textdomain)(domainname);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, domainname, 0);
+  char *domain = REAL(textdomain)(domainname);
   if (domain) {
     COMMON_INTERCEPTOR_INITIALIZE_RANGE(domain, REAL(strlen)(domain) + 1);
   }
   return domain;
 }
 #define INIT_TEXTDOMAIN COMMON_INTERCEPT_FUNCTION(textdomain)
 #else
 #define INIT_TEXTDOMAIN
INTERCEPTOR(int, strcmp, const char *s1, const char *s2) {
   COMMON_INTERCEPTOR_ENTER(ctx, strcmp, s1, s2);
   unsigned char c1, c2;
   uptr i;
   for (i = 0;; i++) {
     c1 = (unsigned char)s1[i];
     c2 = (unsigned char)s2[i];
     if (c1 != c2 || c1 == '\0') break;
   }
-  COMMON_INTERCEPTOR_READ_RANGE(ctx, s1, i + 1);
-  COMMON_INTERCEPTOR_READ_RANGE(ctx, s2, i + 1);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, s1, i + 1);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, s2, i + 1);
   return CharCmpX(c1, c2);
 }
 
 INTERCEPTOR(int, strncmp, const char *s1, const char *s2, uptr size) {
   if (COMMON_INTERCEPTOR_NOTHING_IS_INITIALIZED)
     return internal_strncmp(s1, s2, size);
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, strncmp, s1, s2, size);
INTERCEPTOR(int, strcasecmp, const char *s1, const char *s2) {
   COMMON_INTERCEPTOR_ENTER(ctx, strcasecmp, s1, s2);
   unsigned char c1 = 0, c2 = 0;
   uptr i;
   for (i = 0;; i++) {
     c1 = (unsigned char)s1[i];
     c2 = (unsigned char)s2[i];
     if (CharCaseCmp(c1, c2) != 0 || c1 == '\0') break;
   }
-  COMMON_INTERCEPTOR_READ_RANGE(ctx, s1, i + 1);
-  COMMON_INTERCEPTOR_READ_RANGE(ctx, s2, i + 1);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, s1, i + 1);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, s2, i + 1);
   return CharCaseCmp(c1, c2);
 }
 
 INTERCEPTOR(int, strncasecmp, const char *s1, const char *s2, SIZE_T n) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, strncasecmp, s1, s2, n);
   unsigned char c1 = 0, c2 = 0;
   uptr i;

 
 #define INIT_STRCASECMP COMMON_INTERCEPT_FUNCTION(strcasecmp)
 #define INIT_STRNCASECMP COMMON_INTERCEPT_FUNCTION(strncasecmp)
 #else
 #define INIT_STRCASECMP
 #define INIT_STRNCASECMP
 #endif
 
+
+#if SANITIZER_INTERCEPT_STRSTR
+
+static inline void StrstrCheck(void *ctx, char *r, const char *s1,
+                               const char *s2 ) {
+    uptr len1 = REAL(strlen)(s1) + 1;
+    uptr len2 = REAL(strlen)(s2) + 1;
+    COMMON_INTERCEPTOR_READ_STRING(ctx, s1, r ? r - s1 + len2 : len1);
+    COMMON_INTERCEPTOR_READ_STRING(ctx, s2, r ? len2 :
+                                   (len2 <= len1 ? len2 - 1 : len1));
+}
+
+INTERCEPTOR(char*, strstr, const char *s1, const char *s2) {
+  if COMMON_INTERCEPTOR_NOTHING_IS_INITIALIZED
+    return internal_strstr(s1, s2);
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, strstr, s1, s2);
+  char *r = REAL(strstr)(s1, s2);
+  if (common_flags()->intercept_strstr) {
+    StrstrCheck(ctx, r, s1, s2);
+  }
+  return r;
+}
+
+#define INIT_STRSTR COMMON_INTERCEPT_FUNCTION(strstr);
+
+INTERCEPTOR(char*, strcasestr, const char *s1, const char *s2) {
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, strcasestr, s1, s2);
+  char *r = REAL(strcasestr)(s1, s2);
+  if (common_flags()->intercept_strstr) {
+    StrstrCheck(ctx, r, s1, s2);
+  }
+  return r;
+}
+
+#define INIT_STRCASESTR COMMON_INTERCEPT_FUNCTION(strcasestr);
+
+#else
+#define INIT_STRSTR
+#define INIT_STRCASESTR
+#endif
+
+
+#if SANITIZER_INTERCEPT_STRCSPN
+
+INTERCEPTOR(SIZE_T, strcspn, const char *s1, const char *s2) {
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, strcspn, s1, s2);
+  SIZE_T r = REAL(strcspn)(s1, s2);
+  if (common_flags()->intercept_strcspn) {
+    COMMON_INTERCEPTOR_READ_STRING(ctx, s2,
+        s1 && *s1 ? internal_strchr(s2, s1[r]) - s2 + 1 : 0);
+    COMMON_INTERCEPTOR_READ_STRING(ctx, s1, r + 1);
+  }
+  return r;
+}
+
+#define INIT_STRCSPN COMMON_INTERCEPT_FUNCTION(strcspn);
+
+INTERCEPTOR(char *, strpbrk, const char *s1, const char *s2) {
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, strpbrk, s1, s2);
+  char *r = REAL(strpbrk)(s1, s2);
+  if (common_flags()->intercept_strpbrk) {
+    COMMON_INTERCEPTOR_READ_STRING(ctx, s2,
+        r ? internal_strchr(s2, *r) - s2 + 1 :
+        *s1 ? REAL(strlen)(s2) + 1 : 0);
+    COMMON_INTERCEPTOR_READ_STRING(ctx, s1,
+        r ? r - s1 + 1 : REAL(strlen)(s1) + 1);
+  }
+  return r;
+}
+
+#define INIT_STRPBRK COMMON_INTERCEPT_FUNCTION(strpbrk);
+
+#else
+#define INIT_STRCSPN
+#define INIT_STRPBRK
+#endif
+
 #if SANITIZER_INTERCEPT_MEMCHR
 INTERCEPTOR(void*, memchr, const void *s, int c, SIZE_T n) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, memchr, s, c, n);
   void *res = REAL(memchr)(s, c, n);
   uptr len = res ? (char *)res - (const char *)s + 1 : n;
   COMMON_INTERCEPTOR_READ_RANGE(ctx, s, len);
   return res;
INTERCEPTOR(char *, strptime, char *s, char *format, __sanitizer_tm *tm) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, strptime, s, format, tm);
   if (format)
     COMMON_INTERCEPTOR_READ_RANGE(ctx, format, REAL(strlen)(format) + 1);
   // FIXME: under ASan the call below may write to freed memory and corrupt
   // its metadata. See
   // https://code.google.com/p/address-sanitizer/issues/detail?id=321.
   char *res = REAL(strptime)(s, format, tm);
-  if (res) {
-    COMMON_INTERCEPTOR_READ_RANGE(ctx, s, res - s);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, s, res ? res - s : 0);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, format, 0);
+  if (res && tm) {
     // Do not call unpoison_tm here, because strptime does not, in fact,
     // initialize the entire struct tm. For example, tm_zone pointer is left
     // uninitialized.
-    if (tm) COMMON_INTERCEPTOR_WRITE_RANGE(ctx, tm, sizeof(*tm));
+    COMMON_INTERCEPTOR_WRITE_RANGE(ctx, tm, sizeof(*tm));
   }
   return res;
 }
 #define INIT_STRPTIME COMMON_INTERCEPT_FUNCTION(strptime);
 #else
 #define INIT_STRPTIME
 #endif
 
static int wrapped_gl_stat(const char *s, void *st) {
   return pglob_copy->gl_stat(s, st);
 }
 
 INTERCEPTOR(int, glob, const char *pattern, int flags,
             int (*errfunc)(const char *epath, int eerrno),
             __sanitizer_glob_t *pglob) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, glob, pattern, flags, errfunc, pglob);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, pattern, 0);
   __sanitizer_glob_t glob_copy = {
       0,                  0,                   0,
       0,                  wrapped_gl_closedir, wrapped_gl_readdir,
       wrapped_gl_opendir, wrapped_gl_lstat,    wrapped_gl_stat};
   if (flags & glob_altdirfunc) {
     Swap(pglob->gl_closedir, glob_copy.gl_closedir);
     Swap(pglob->gl_readdir, glob_copy.gl_readdir);
     Swap(pglob->gl_opendir, glob_copy.gl_opendir);
INTERCEPTOR(int, glob, const char *pattern, int flags,
   return res;
 }
 
 INTERCEPTOR(int, glob64, const char *pattern, int flags,
             int (*errfunc)(const char *epath, int eerrno),
             __sanitizer_glob_t *pglob) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, glob64, pattern, flags, errfunc, pglob);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, pattern, 0);
   __sanitizer_glob_t glob_copy = {
       0,                  0,                   0,
       0,                  wrapped_gl_closedir, wrapped_gl_readdir,
       wrapped_gl_opendir, wrapped_gl_lstat,    wrapped_gl_stat};
   if (flags & glob_altdirfunc) {
     Swap(pglob->gl_closedir, glob_copy.gl_closedir);
     Swap(pglob->gl_readdir, glob_copy.gl_readdir);
     Swap(pglob->gl_opendir, glob_copy.gl_opendir);
INTERCEPTOR(char *, inet_ntop, int af, const void *src, char *dst, u32 size) {
   // https://code.google.com/p/address-sanitizer/issues/detail?id=321.
   char *res = REAL(inet_ntop)(af, src, dst, size);
   if (res) COMMON_INTERCEPTOR_WRITE_RANGE(ctx, res, REAL(strlen)(res) + 1);
   return res;
 }
 INTERCEPTOR(int, inet_pton, int af, const char *src, void *dst) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, inet_pton, af, src, dst);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, src, 0);
   // FIXME: figure out read size based on the address family.
   // FIXME: under ASan the call below may write to freed memory and corrupt
   // its metadata. See
   // https://code.google.com/p/address-sanitizer/issues/detail?id=321.
   int res = REAL(inet_pton)(af, src, dst);
   if (res == 1) {
     uptr sz = __sanitizer_in_addr_sz(af);
     if (sz) COMMON_INTERCEPTOR_WRITE_RANGE(ctx, dst, sz);

 }
 
 #define INIT_GET_CURRENT_DIR_NAME \
   COMMON_INTERCEPT_FUNCTION(get_current_dir_name);
 #else
 #define INIT_GET_CURRENT_DIR_NAME
 #endif
 
+UNUSED static inline bool IsValidStrtolBase(int base) {
+  return (base == 0) || (2 <= base && base <= 36);
+}
+
+UNUSED static inline void FixRealStrtolEndptr(const char *nptr, char **endptr) {
+  CHECK(endptr);
+  if (nptr == *endptr) {
+    // No digits were found at strtol call, we need to find out the last
+    // symbol accessed by strtoll on our own.
+    // We get this symbol by skipping leading blanks and optional +/- sign.
+    while (IsSpace(*nptr)) nptr++;
+    if (*nptr == '+' || *nptr == '-') nptr++;
+    *endptr = const_cast<char *>(nptr);
+  }
+  CHECK(*endptr >= nptr);
+}
+
+UNUSED static inline void StrtolFixAndCheck(void *ctx, const char *nptr,
+                             char **endptr, char *real_endptr, int base) {
+  if (endptr != 0) {
+    *endptr = real_endptr;
+    COMMON_INTERCEPTOR_WRITE_RANGE(ctx, endptr, sizeof(*endptr));
+  }
+  // If base has unsupported value, strtol can exit with EINVAL
+  // without reading any characters. So do additional checks only
+  // if base is valid.
+  bool is_valid_base = (base == 0) || (2 <= base && base <= 36);
+  if (is_valid_base) {
+    FixRealStrtolEndptr(nptr, &real_endptr);
+  }
+  COMMON_INTERCEPTOR_READ_STRING(ctx, nptr, is_valid_base ?
+                                 (real_endptr - nptr) + 1 : 0);
+}
+
+
 #if SANITIZER_INTERCEPT_STRTOIMAX
 INTERCEPTOR(INTMAX_T, strtoimax, const char *nptr, char **endptr, int base) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, strtoimax, nptr, endptr, base);
   // FIXME: under ASan the call below may write to freed memory and corrupt
   // its metadata. See
   // https://code.google.com/p/address-sanitizer/issues/detail?id=321.
-  INTMAX_T res = REAL(strtoimax)(nptr, endptr, base);
-  if (endptr) COMMON_INTERCEPTOR_WRITE_RANGE(ctx, endptr, sizeof(*endptr));
+  char *real_endptr;
+  INTMAX_T res = REAL(strtoimax)(nptr, &real_endptr, base);
+  StrtolFixAndCheck(ctx, nptr, endptr, real_endptr, base);
   return res;
 }
 
 INTERCEPTOR(INTMAX_T, strtoumax, const char *nptr, char **endptr, int base) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, strtoumax, nptr, endptr, base);
   // FIXME: under ASan the call below may write to freed memory and corrupt
   // its metadata. See
   // https://code.google.com/p/address-sanitizer/issues/detail?id=321.
-  INTMAX_T res = REAL(strtoumax)(nptr, endptr, base);
-  if (endptr) COMMON_INTERCEPTOR_WRITE_RANGE(ctx, endptr, sizeof(*endptr));
+  char *real_endptr;
+  INTMAX_T res = REAL(strtoumax)(nptr, &real_endptr, base);
+  StrtolFixAndCheck(ctx, nptr, endptr, real_endptr, base);
   return res;
 }
 
 #define INIT_STRTOIMAX                  \
   COMMON_INTERCEPT_FUNCTION(strtoimax); \
   COMMON_INTERCEPT_FUNCTION(strtoumax);
 #else
 #define INIT_STRTOIMAX

 #else
 #define INIT_TEMPNAM
 #endif
 
 #if SANITIZER_INTERCEPT_PTHREAD_SETNAME_NP
 INTERCEPTOR(int, pthread_setname_np, uptr thread, const char *name) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER(ctx, pthread_setname_np, thread, name);
+  COMMON_INTERCEPTOR_READ_STRING(ctx, name, 0);
   COMMON_INTERCEPTOR_SET_PTHREAD_NAME(ctx, thread, name);
   return REAL(pthread_setname_np)(thread, name);
 }
 #define INIT_PTHREAD_SETNAME_NP COMMON_INTERCEPT_FUNCTION(pthread_setname_np);
 #else
 #define INIT_PTHREAD_SETNAME_NP
 #endif
 

 #else
 #define INIT_FCLOSE
 #endif
 
 #if SANITIZER_INTERCEPT_DLOPEN_DLCLOSE
 INTERCEPTOR(void*, dlopen, const char *filename, int flag) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER_NOIGNORE(ctx, dlopen, filename, flag);
+  if (filename) COMMON_INTERCEPTOR_READ_STRING(ctx, filename, 0);
   void *res = REAL(dlopen)(filename, flag);
   COMMON_INTERCEPTOR_LIBRARY_LOADED(filename, res);
   return res;
 }
 
 INTERCEPTOR(int, dlclose, void *handle) {
   void *ctx;
   COMMON_INTERCEPTOR_ENTER_NOIGNORE(ctx, dlclose, handle);
static void InitializeCommonInterceptors() {
   static u64 metadata_mem[sizeof(MetadataHashMap) / sizeof(u64) + 1];
   interceptor_metadata_map = new((void *)&metadata_mem) MetadataHashMap();
 
   INIT_TEXTDOMAIN;
   INIT_STRCMP;
   INIT_STRNCMP;
   INIT_STRCASECMP;
   INIT_STRNCASECMP;
+  INIT_STRSTR;
+  INIT_STRCASESTR;
+  INIT_STRCSPN;
+  INIT_STRPBRK;
   INIT_MEMCHR;
   INIT_MEMRCHR;
   INIT_READ;
   INIT_PREAD;
   INIT_PREAD64;
   INIT_READV;
   INIT_PREADV;
   INIT_PREADV64;

lib/sanitizer_common/sanitizer_flags.inc

 COMMON_FLAG(bool, symbolize_inline_frames, true,
             "Print inlined frames in stacktraces. Defaults to true.")
 COMMON_FLAG(const char *, stack_trace_format, "DEFAULT",
             "Format string used to render stack frames. "
             "See sanitizer_stacktrace_printer.h for the format description. "
             "Use DEFAULT to get default format.")
 COMMON_FLAG(bool, no_huge_pages_for_shadow, true,
             "If true, the shadow is not allowed to use huge pages. ")
+COMMON_FLAG(bool, intercept_strstr, true,
+            "If set, uses custom wrappers for strstr and strcasestr functions "
+            "to find more errors.")
+COMMON_FLAG(bool, intercept_strcspn, true,
+            "If set, uses custom wrappers for strcspn function "
+            "to find more errors.")
+COMMON_FLAG(bool, intercept_strpbrk, true,
+            "If set, uses custom wrappers for strpbrk function "
+            "to find more errors.")
+COMMON_FLAG(bool, strict_string_checks, true,
+            "If set check that string arguments are properly null-terminated")

lib/sanitizer_common/sanitizer_platform_interceptors.h

 
 #if SANITIZER_IOS
 # define SI_IOS 1
 #else
 # define SI_IOS 0
 #endif
 
 #define SANITIZER_INTERCEPT_STRCMP 1
+#define SANITIZER_INTERCEPT_STRSTR 1
+#define SANITIZER_INTERCEPT_STRCASESTR SI_NOT_WINDOWS
+#define SANITIZER_INTERCEPT_STRCSPN 1
+#define SANITIZER_INTERCEPT_STRPBRK 1
 #define SANITIZER_INTERCEPT_TEXTDOMAIN SI_LINUX_NOT_ANDROID
 #define SANITIZER_INTERCEPT_STRCASECMP SI_NOT_WINDOWS
 #define SANITIZER_INTERCEPT_MEMCHR 1
 #define SANITIZER_INTERCEPT_MEMRCHR SI_LINUX
 
 #define SANITIZER_INTERCEPT_READ   SI_NOT_WINDOWS
 #define SANITIZER_INTERCEPT_PREAD  SI_NOT_WINDOWS
 #define SANITIZER_INTERCEPT_WRITE  SI_NOT_WINDOWS

lib/tsan/rtl/tsan_interceptors.cc

 #define TSAN_INTERCEPTOR(ret, func, ...) INTERCEPTOR(ret, func, __VA_ARGS__)
 #define TSAN_INTERCEPT(func) INTERCEPT_FUNCTION(func)
 #if SANITIZER_FREEBSD
 # define TSAN_INTERCEPT_VER(func, ver) INTERCEPT_FUNCTION(func)
 #else
 # define TSAN_INTERCEPT_VER(func, ver) INTERCEPT_FUNCTION_VER(func, ver)
 #endif
 
+#define TSAN_READ_STRING(thr, pc, s, n)                             \
+  MemoryAccessRange(thr, pc, (uptr)s,                               \
+    common_flags()->strict_string_checks ? internal_strlen(s) + 1 : n, false)
+
 #define BLOCK_REAL(name) (BlockingCall(thr), REAL(name))
 
 struct BlockingCall {
   explicit BlockingCall(ThreadState *thr)
       : thr(thr)
       , ctx(SigCtx(thr)) {
     for (;;) {
       atomic_store(&ctx->in_blocking_func, 1, memory_order_relaxed);
TSAN_INTERCEPTOR(void*, memmove, void *dst, void *src, uptr n) {
   MemoryAccessRange(thr, pc, (uptr)src, n, false);
   return REAL(memmove)(dst, src, n);
 }
 
 TSAN_INTERCEPTOR(char*, strchr, char *s, int c) {
   SCOPED_TSAN_INTERCEPTOR(strchr, s, c);
   char *res = REAL(strchr)(s, c);
   uptr len = res ? (char*)res - (char*)s + 1 : internal_strlen(s) + 1;
-  MemoryAccessRange(thr, pc, (uptr)s, len, false);
+  TSAN_READ_STRING(thr, pc, s, len);
   return res;
 }
 
 TSAN_INTERCEPTOR(char*, strchrnul, char *s, int c) {
   SCOPED_TSAN_INTERCEPTOR(strchrnul, s, c);
   char *res = REAL(strchrnul)(s, c);
   uptr len = (char*)res - (char*)s + 1;
-  MemoryAccessRange(thr, pc, (uptr)s, len, false);
+  TSAN_READ_STRING(thr, pc, s, len);
   return res;
 }
 
 TSAN_INTERCEPTOR(char*, strrchr, char *s, int c) {
   SCOPED_TSAN_INTERCEPTOR(strrchr, s, c);
   MemoryAccessRange(thr, pc, (uptr)s, internal_strlen(s) + 1, false);
   return REAL(strrchr)(s, c);
 }

 TSAN_INTERCEPTOR(char*, strncpy, char *dst, char *src, uptr n) {
   SCOPED_TSAN_INTERCEPTOR(strncpy, dst, src, n);
   uptr srclen = internal_strnlen(src, n);
   MemoryAccessRange(thr, pc, (uptr)dst, n, true);
   MemoryAccessRange(thr, pc, (uptr)src, min(srclen + 1, n), false);
   return REAL(strncpy)(dst, src, n);
 }
 
-TSAN_INTERCEPTOR(const char*, strstr, const char *s1, const char *s2) {
-  SCOPED_TSAN_INTERCEPTOR(strstr, s1, s2);
-  const char *res = REAL(strstr)(s1, s2);
-  uptr len1 = internal_strlen(s1);
-  uptr len2 = internal_strlen(s2);
-  MemoryAccessRange(thr, pc, (uptr)s1, len1 + 1, false);
-  MemoryAccessRange(thr, pc, (uptr)s2, len2 + 1, false);
-  return res;
-}
-
 TSAN_INTERCEPTOR(char*, strdup, const char *str) {
   SCOPED_TSAN_INTERCEPTOR(strdup, str);
   // strdup will call malloc, so no instrumentation is required here.
   return REAL(strdup)(str);
 }
 
 static bool fix_mmap_addr(void **addr, long_t sz, int flags) {
   if (*addr) {
if (res == 0) {
     Acquire(thr, pc, (uptr)s);
   }
   return res;
 }
 
 #if !SANITIZER_FREEBSD
 TSAN_INTERCEPTOR(int, __xstat, int version, const char *path, void *buf) {
   SCOPED_TSAN_INTERCEPTOR(__xstat, version, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(__xstat)(version, path, buf);
 }
 #define TSAN_MAYBE_INTERCEPT___XSTAT TSAN_INTERCEPT(__xstat)
 #else
 #define TSAN_MAYBE_INTERCEPT___XSTAT
 #endif
 
 TSAN_INTERCEPTOR(int, stat, const char *path, void *buf) {
 #if SANITIZER_FREEBSD
   SCOPED_TSAN_INTERCEPTOR(stat, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(stat)(path, buf);
 #else
   SCOPED_TSAN_INTERCEPTOR(__xstat, 0, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(__xstat)(0, path, buf);
 #endif
 }
 
 #if !SANITIZER_FREEBSD
 TSAN_INTERCEPTOR(int, __xstat64, int version, const char *path, void *buf) {
   SCOPED_TSAN_INTERCEPTOR(__xstat64, version, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(__xstat64)(version, path, buf);
 }
 #define TSAN_MAYBE_INTERCEPT___XSTAT64 TSAN_INTERCEPT(__xstat64)
 #else
 #define TSAN_MAYBE_INTERCEPT___XSTAT64
 #endif
 
 #if !SANITIZER_FREEBSD
 TSAN_INTERCEPTOR(int, stat64, const char *path, void *buf) {
   SCOPED_TSAN_INTERCEPTOR(__xstat64, 0, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(__xstat64)(0, path, buf);
 }
 #define TSAN_MAYBE_INTERCEPT_STAT64 TSAN_INTERCEPT(stat64)
 #else
 #define TSAN_MAYBE_INTERCEPT_STAT64
 #endif
 
 #if !SANITIZER_FREEBSD
 TSAN_INTERCEPTOR(int, __lxstat, int version, const char *path, void *buf) {
   SCOPED_TSAN_INTERCEPTOR(__lxstat, version, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(__lxstat)(version, path, buf);
 }
 #define TSAN_MAYBE_INTERCEPT___LXSTAT TSAN_INTERCEPT(__lxstat)
 #else
 #define TSAN_MAYBE_INTERCEPT___LXSTAT
 #endif
 
 TSAN_INTERCEPTOR(int, lstat, const char *path, void *buf) {
 #if SANITIZER_FREEBSD
   SCOPED_TSAN_INTERCEPTOR(lstat, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(lstat)(path, buf);
 #else
   SCOPED_TSAN_INTERCEPTOR(__lxstat, 0, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(__lxstat)(0, path, buf);
 #endif
 }
 
 #if !SANITIZER_FREEBSD
 TSAN_INTERCEPTOR(int, __lxstat64, int version, const char *path, void *buf) {
   SCOPED_TSAN_INTERCEPTOR(__lxstat64, version, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(__lxstat64)(version, path, buf);
 }
 #define TSAN_MAYBE_INTERCEPT___LXSTAT64 TSAN_INTERCEPT(__lxstat64)
 #else
 #define TSAN_MAYBE_INTERCEPT___LXSTAT64
 #endif
 
 #if !SANITIZER_FREEBSD
 TSAN_INTERCEPTOR(int, lstat64, const char *path, void *buf) {
   SCOPED_TSAN_INTERCEPTOR(__lxstat64, 0, path, buf);
+  TSAN_READ_STRING(thr, pc, path, 0);
   return REAL(__lxstat64)(0, path, buf);
 }
 #define TSAN_MAYBE_INTERCEPT_LSTAT64 TSAN_INTERCEPT(lstat64)
 #else
 #define TSAN_MAYBE_INTERCEPT_LSTAT64
 #endif
 
 #if !SANITIZER_FREEBSD

 }
 #define TSAN_MAYBE_INTERCEPT_FSTAT64 TSAN_INTERCEPT(fstat64)
 #else
 #define TSAN_MAYBE_INTERCEPT_FSTAT64
 #endif
 
 TSAN_INTERCEPTOR(int, open, const char *name, int flags, int mode) {
   SCOPED_TSAN_INTERCEPTOR(open, name, flags, mode);
+  TSAN_READ_STRING(thr, pc, name, 0);
   int fd = REAL(open)(name, flags, mode);
   if (fd >= 0)
     FdFileCreate(thr, pc, fd);
   return fd;
 }
 
 #if !SANITIZER_FREEBSD
 TSAN_INTERCEPTOR(int, open64, const char *name, int flags, int mode) {
   SCOPED_TSAN_INTERCEPTOR(open64, name, flags, mode);
+  TSAN_READ_STRING(thr, pc, name, 0);
   int fd = REAL(open64)(name, flags, mode);
   if (fd >= 0)
     FdFileCreate(thr, pc, fd);
   return fd;
 }
 #define TSAN_MAYBE_INTERCEPT_OPEN64 TSAN_INTERCEPT(open64)
 #else
 #define TSAN_MAYBE_INTERCEPT_OPEN64
 #endif
 
 TSAN_INTERCEPTOR(int, creat, const char *name, int mode) {
   SCOPED_TSAN_INTERCEPTOR(creat, name, mode);
+  TSAN_READ_STRING(thr, pc, name, 0);
   int fd = REAL(creat)(name, mode);
   if (fd >= 0)
     FdFileCreate(thr, pc, fd);
   return fd;
 }
 
 #if !SANITIZER_FREEBSD
 TSAN_INTERCEPTOR(int, creat64, const char *name, int mode) {
   SCOPED_TSAN_INTERCEPTOR(creat64, name, mode);
+  TSAN_READ_STRING(thr, pc, name, 0);
   int fd = REAL(creat64)(name, mode);
   if (fd >= 0)
     FdFileCreate(thr, pc, fd);
   return fd;
 }
 #define TSAN_MAYBE_INTERCEPT_CREAT64 TSAN_INTERCEPT(creat64)
 #else
 #define TSAN_MAYBE_INTERCEPT_CREAT64
#endif
   TSAN_INTERCEPT(memcpy);
   TSAN_INTERCEPT(memmove);
   TSAN_INTERCEPT(memcmp);
   TSAN_INTERCEPT(strchr);
   TSAN_INTERCEPT(strchrnul);
   TSAN_INTERCEPT(strrchr);
   TSAN_INTERCEPT(strcpy);  // NOLINT
   TSAN_INTERCEPT(strncpy);
-  TSAN_INTERCEPT(strstr);
   TSAN_INTERCEPT(strdup);
 
   TSAN_INTERCEPT(pthread_create);
   TSAN_INTERCEPT(pthread_join);
   TSAN_INTERCEPT(pthread_detach);
 
   TSAN_INTERCEPT_VER(pthread_cond_init, "GLIBC_2.3.2");
   TSAN_INTERCEPT_VER(pthread_cond_signal, "GLIBC_2.3.2");

test/asan/TestCases/atoi_strict.c

+// Test strict_string_checks option in atoi function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+#include <assert.h>
+#include <stdlib.h>
+
+int main(int argc, char **argv) {
+  char p[] = {'1', 'b'};
+  char z = 0;
+  int r = atoi(p);
+  // CHECK: READ of size
+  // CHECK: '{{[p|z]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == 1);
+  return 0;
+}

test/asan/TestCases/atol_strict.c

+// Test strict_string_checks option in atol function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+#include <assert.h>
+#include <stdlib.h>
+
+int main(int argc, char **argv) {
+  char p[] = {'1', 'b'};
+  char z = 0;
+  long r = atol(p);
+  // CHECK: READ of size
+  // CHECK: '{{[p|z]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == 1);
+  return 0;
+}

test/asan/TestCases/atoll_strict.c

+// Test strict_string_checks option in atoll function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+#include <assert.h>
+#include <stdlib.h>
+
+int main(int argc, char **argv) {
+  char p[] = {'1', 'b'};
+  char z = 0;
+  long long r = atoll(p);
+  // CHECK: READ of size
+  // CHECK: '{{[p|z]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == 1);
+  return 0;
+}

test/asan/TestCases/strcasestr-1.c

+// Test haystack overflow in strcasestr function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+
+// Test intercept_strstr asan option
+// RUN: ASAN_OPTIONS=intercept_strstr=false %run %t 2>&1
+
+// There's no interceptor for strcasestr on Windows
+// XFAIL: win32
+
+#define _GNU_SOURCE
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s2[] = "c";
+  char s1[] = {'a', 'b'};
+  char s3 = 0;
+  r = strcasestr(s1, s2);
+  // CHECK:'s{{[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == 0);
+  return 0;
+}

test/asan/TestCases/strcasestr-2.c

+// Test needle overflow in strcasestr function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+
+// Test intercept_strstr asan option
+// RUN: ASAN_OPTIONS=intercept_strstr=false %run %t 2>&1
+
+// There's no interceptor for strcasestr on Windows
+// XFAIL: win32
+
+#define _GNU_SOURCE
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s1[] = "ab";
+  char s2[] = {'c'};
+  char s3 = 0;
+  r = strcasestr(s1, s2);
+  assert(r == 0);
+  // CHECK:'s{{[2|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  return 0;
+}

test/asan/TestCases/strcasestr_strict.c

+// Test strict_string_checks option in strcasestr function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+// There's no interceptor for strcasestr on Windows
+// XFAIL: win32
+
+#define _GNU_SOURCE
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s2[] = {"Ab"};
+  char s1[] = {'a', 'b', 'c'};
+  char s3 = 0;
+  r = strcasestr(s1, s2);
+  // CHECK: READ of size
+  // CHECK: '{{s[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == s1);
+  return 0;
+}

test/asan/TestCases/strcat_strict.c

+// Test strict_string_checks option in strcat function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s --check-prefix=CHECK-STRICT --check-prefix=CHECK
+// RUN: ASAN_OPTIONS=strict_string_checks=false not  %run %t 2>&1 | FileCheck %s --check-prefix=CHECK
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s --check-prefix=CHECK-STRICT --check-prefix=CHECK
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s2[] = {"a"};
+  char s1[] = {'b'};
+  char s3[] = {'\0', 'z', 'z'};
+  r = strcat(s1, s2);
+  // CHECK-STRICT: READ of size
+  // CHECK: '{{s[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  return 0;
+}

test/asan/TestCases/strchr_strict.c

+// Test strict_string_checks option in strchr function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char c = 'a';
+  char s[] = {'a', 'b', 'c'};
+  char z = 0;
+  r = strchr(s, c);
+  // CHECK: '{{[s|z]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == s);
+  return 0;
+}

test/asan/TestCases/strcspn-1.c

+// Test string s1 overflow in strcspn function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+
+// Test intercept_strcspn asan option
+// RUN: ASAN_OPTIONS=intercept_strcspn=false %run %t 2>&1
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  size_t r;
+  char s2[] = "ab";
+  char s1[] = {'c', 'd'};
+  char s3 = 0;
+  r = strcspn(s1, s2);
+  // CHECK:'s{{[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r >= sizeof(s1));
+  return 0;
+}

test/asan/TestCases/strcspn-2.c

+// Test stopset overflow in strcspn function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+
+// Test intercept_strcspn asan option
+// RUN: ASAN_OPTIONS=intercept_strcspn=false %run %t 2>&1
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  size_t r;
+  char s1[] = "ab";
+  char s2[] = {'c', 'd'};
+  char s3 = 0;
+  r = strcspn(s1, s2);
+  // CHECK:'s{{[2|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == sizeof(s1) - 1);
+  return 0;
+}

test/asan/TestCases/strcspn_strict.c

+// Test strict_string_checks option in strcspn function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  size_t r;
+  char s2[] = "ab";
+  char s1[] = {'c', 'a'};
+  char s3 = 0;
+  r = strcspn(s1, s2);
+  // CHECK: READ of size
+  // CHECK: '{{s[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == 1);
+  return 0;
+}

test/asan/TestCases/strncat_strict.c

+// Test strict_string_checks option in strncat function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s --check-prefix=CHECK-STRICT --check-prefix=CHECK
+// RUN: ASAN_OPTIONS=strict_string_checks=false not  %run %t 2>&1 | FileCheck %s --check-prefix=CHECK
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s --check-prefix=CHECK-STRICT --check-prefix=CHECK
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s2[] = {"a"};
+  char s1[] = {'b'};
+  char s3[] = {'\0', 'z', 'z'};
+  r = strncat(s1, s2, 1);
+  // CHECK-STRICT: READ of size
+  // CHECK: '{{s[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  return 0;
+}

test/asan/TestCases/strpbrk-1.c

+// Test string s1 overflow in strpbrk function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+
+// Test intercept_strpbrk asan option
+// RUN: ASAN_OPTIONS=intercept_strpbrk=false %run %t 2>&1
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r;
+  char s2[] = "ab";
+  char s1[] = {'c', 'd'};
+  char s3[] = "a";
+  r = strpbrk(s1, s2);
+  // CHECK:'s{{[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r <= s3);
+  return 0;
+}

test/asan/TestCases/strpbrk-2.c

+// Test stopset overflow in strpbrk function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+
+// Test intercept_strpbrk asan option
+// RUN: ASAN_OPTIONS=intercept_strpbrk=false %run %t 2>&1
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r;
+  char s1[] = "a";
+  char s2[] = {'b', 'c'};
+  char s3 = 0;
+  r = strpbrk(s1, s2);
+  // CHECK:'s{{[2|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == (r ? s1 : 0));
+  return 0;
+}

test/asan/TestCases/strpbrk_strict.c

+// Test strict_string_checks option in strpbrk function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r;
+  char s2[] = "ab";
+  char s1[] = {'c', 'a'};
+  char s3 = 0;
+  r = strpbrk(s1, s2);
+  // CHECK: READ of size
+  // CHECK: '{{s[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == s1 + 1);
+  return 0;
+}

test/asan/TestCases/strstr-1.c

+// Test haystack overflow in strstr function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+
+// Test intercept_strstr asan option
+// RUN: ASAN_OPTIONS=intercept_strstr=false %run %t 2>&1
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s2[] = "c";
+  char s1[] = {'a', 'b'};
+  char s3 = 0;
+  r = strstr(s1, s2);
+  // CHECK:'s{{[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == 0);
+  return 0;
+}

test/asan/TestCases/strstr-2.c

+// Test needle overflow in strstr function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+
+// Test intercept_strstr asan option
+// RUN: ASAN_OPTIONS=intercept_strstr=false %run %t 2>&1
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s1[] = "ab";
+  char s2[] = {'c'};
+  char s3 = 0;
+  r = strstr(s1, s2);
+  // CHECK:'s{{[2|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == 0);
+  return 0;
+}

test/asan/TestCases/strstr_strict.c

+// Test strict_string_checks option in strstr function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s2[] = {"ab"};
+  char s1[] = {'a', 'b', 'c'};
+  char s3 = 0;
+  r = strstr(s1, s2);
+  // CHECK: READ of size
+  // CHECK: '{{s[1|3]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == s1);
+  return 0;
+}

test/asan/TestCases/strtol_strict.c

+// Test strict_string_checks option in strtol function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+#include <assert.h>
+#include <stdlib.h>
+
+int main(int argc, char **argv) {
+  char *endptr;
+  char p[] = {'1', 'b'};
+  char z = 0;
+  long r = strtol(p, &endptr, 2);
+  // CHECK: READ of size
+  // CHECK: '{{[p|z]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == 1);
+  return 0;
+}

test/asan/TestCases/strtoll_strict.c

+// Test strict_string_checks option in strtoll function
+// RUN: %clang_asan %s -o %t && not %run %t 2>&1 | FileCheck %s
+// RUN: ASAN_OPTIONS=strict_string_checks=false %run %t 2>&1
+// RUN: ASAN_OPTIONS=strict_string_checks=true not %run %t 2>&1 | FileCheck %s
+
+#include <assert.h>
+#include <stdlib.h>
+
+int main(int argc, char **argv) {
+  char *endptr;
+  char p[] = {'1', 'b'};
+  char z = 0;
+  long long r = strtol(p, &endptr, 2);
+  // CHECK: READ of size
+  // CHECK: '{{[p|z]}}' <== Memory access at offset {{[0-9]+ .*}}flows this variable
+  assert(r == 1);
+  return 0;
+}

test/asan/Unit/lit.site.cfg.in

else:
   test_dir = "default"
 config.test_exec_root = os.path.join("@COMPILER_RT_BINARY_DIR@",
                                      "lib", "asan", "tests", test_dir)
 
 config.test_source_root = config.test_exec_root
 
 # Set LD_LIBRARY_PATH to pick dynamic runtime up properly.
 push_ld_library_path(config, config.compiler_rt_libdir)
+
+# Disable strict str checks in unit tests
+config.environment['ASAN_OPTIONS'] = 'strict_string_checks=false'

test/sanitizer_common/TestCases/strcasestr.c

+// RUN: %clang %s -o %t && %run %t 2>&1
+
+// There's no interceptor for strcasestr on Windows
+// XFAIL: win32
+
+#define _GNU_SOURCE
+#include <assert.h>
+#include <string.h>
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s1[] = "aB";
+  char s2[] = "b";
+  r = strcasestr(s1, s2);
+  assert(r == s1 + 1);
+  return 0;
+}

test/sanitizer_common/TestCases/strcspn.c

+// RUN: %clang %s -o %t && %run %t 2>&1
+
+#include <assert.h>
+#include <string.h>
+
+int main(int argc, char **argv) {
+  size_t r;
+  char s1[] = "ad";
+  char s2[] = "cd";
+  r = strcspn(s1, s2);
+  assert(r == 1);
+  return 0;
+}

test/sanitizer_common/TestCases/strpbrk.c

+// RUN: %clang %s -o %t && %run %t 2>&1
+
+#include <assert.h>
+#include <string.h>
+
+
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s1[] = "ad";
+  char s2[] = "cd";
+  r = strpbrk(s1, s2);
+  assert(r == s1 + 1);
+  return 0;
+}

test/sanitizer_common/TestCases/strstr.c

+// RUN: %clang %s -o %t && %run %t 2>&1
+
+#include <assert.h>
+#include <string.h>
+int main(int argc, char **argv) {
+  char *r = 0;
+  char s1[] = "ab";
+  char s2[] = "b";
+  r = strstr(s1, s2);
+  assert(r == s1 + 1);
+  return 0;
+}