[compiler-rt] 69ebac7 - [win/asan] Don't intercept memset etc. in ntdll (#120397)

[via llvm-commits]

Author: Hans Wennborg
Date: 2024-12-20T11:03:17+01:00
New Revision: 69ebac7ad6ae1db9bb19cf3a19ea978af6034ca3

URL: https://github.com/llvm/llvm-project/commit/69ebac7ad6ae1db9bb19cf3a19ea978af6034ca3
DIFF: https://github.com/llvm/llvm-project/commit/69ebac7ad6ae1db9bb19cf3a19ea978af6034ca3.diff

LOG: [win/asan] Don't intercept memset etc. in ntdll (#120397)

When ntdll was added to the list of of "interesting DLLs" list (in
d58230b9dcb3b312a2da8f874daa0cc8dc27da9b), the intention was not to
intercept the "mini CRT" functions it exports. OverrideFunction would
only intercept the *first* function it found when searching the list of
DLLs, and ntdll was put last in that list.

However, after 42cdfbcf3e92466754c175cb0e1e237e9f66749e,
OverrideFunction intercepts *all* matching functions in those DLLs. As
a side-effect, the runtime would now intercept functions like memset
etc. also in ntdll.

This causes a problem when ntdll-internal functions like
RtlDispatchException call the intercepted memset, which tries to
inspect uncommitted shadow memory, raising an exception, and getting
stuck in that loop until the stack overflows.

Since we never intended to intercept ntdll's memset etc., the simplest
fix seems to be to actively ignore ntdll when intercepting those
functions.

Fixes #114793

Added: 
    

Modified: 
    compiler-rt/lib/interception/interception_win.cpp

Removed: 
    


################################################################################
diff  --git a/compiler-rt/lib/interception/interception_win.cpp b/compiler-rt/lib/interception/interception_win.cpp
index a5897274521e92..bd85c50a083a64 100644
--- a/compiler-rt/lib/interception/interception_win.cpp
+++ b/compiler-rt/lib/interception/interception_win.cpp
@@ -213,6 +213,18 @@ static char* _strchr(char* str, char c) {
   return nullptr;
 }
 
+static int _strcmp(const char *s1, const char *s2) {
+  while (true) {
+    unsigned c1 = *s1;
+    unsigned c2 = *s2;
+    if (c1 != c2) return (c1 < c2) ? -1 : 1;
+    if (c1 == 0) break;
+    s1++;
+    s2++;
+  }
+  return 0;
+}
+
 static void _memset(void *p, int value, size_t sz) {
   for (size_t i = 0; i < sz; ++i)
     ((char*)p)[i] = (char)value;
@@ -1177,8 +1189,7 @@ static void **InterestingDLLsAvailable() {
     "libc++.dll",     // libc++
     "libunwind.dll",  // libunwind
 #  endif
-    // NTDLL should go last as it exports some functions that we should
-    // override in the CRT [presumably only used internally].
+    // NTDLL must go last as it gets special treatment in OverrideFunction.
     "ntdll.dll",
     NULL
   };
@@ -1235,7 +1246,7 @@ uptr InternalGetProcAddress(void *module, const char *func_name) {
 
   for (DWORD i = 0; i < exports->NumberOfNames; i++) {
     RVAPtr<char> name(module, names[i]);
-    if (!strcmp(func_name, name)) {
+    if (!_strcmp(func_name, name)) {
       DWORD index = ordinals[i];
       RVAPtr<char> func(module, functions[index]);
 
@@ -1281,9 +1292,22 @@ uptr InternalGetProcAddress(void *module, const char *func_name) {
 
 bool OverrideFunction(
     const char *func_name, uptr new_func, uptr *orig_old_func) {
+  static const char *kNtDllIgnore[] = {
+    "memcmp", "memcpy", "memmove", "memset"
+  };
+
   bool hooked = false;
   void **DLLs = InterestingDLLsAvailable();
   for (size_t i = 0; DLLs[i]; ++i) {
+    if (DLLs[i + 1] == nullptr) {
+      // This is the last DLL, i.e. NTDLL. It exports some functions that
+      // we only want to override in the CRT.
+      for (const char *ignored : kNtDllIgnore) {
+        if (_strcmp(func_name, ignored) == 0)
+          return hooked;
+      }
+    }
+
     uptr func_addr = InternalGetProcAddress(DLLs[i], func_name);
     if (func_addr &&
         OverrideFunction(func_addr, new_func, orig_old_func)) {
@@ -1337,7 +1361,7 @@ bool OverrideImportedFunction(const char *module_to_patch,
       RVAPtr<IMAGE_IMPORT_BY_NAME> import_by_name(
           module, name_table->u1.ForwarderString);
       const char *funcname = &import_by_name->Name[0];
-      if (strcmp(funcname, function_name) == 0)
+      if (_strcmp(funcname, function_name) == 0)
         break;
     }
   }