[sanitizer] Add MemCpyAccessible (#112794)

A layer over `TryMemCpy` to copy only available pages.

Author: vitalybuka

compiler-rt/lib/sanitizer_common/sanitizer_common.h

@@ -275,6 +275,8 @@ bool IsAccessibleMemoryRange(uptr beg, uptr size);
 // the source range cannot be read, in which case the contents of `dest` are
 // undefined.
 bool TryMemCpy(void *dest, const void *src, uptr n);
+// Copies accessible memory, and zero fill inaccessible.
+void MemCpyAccessible(void *dest, const void *src, uptr n);
 
 // Error report formatting.
 const char *StripPathPrefix(const char *filepath,

compiler-rt/lib/sanitizer_common/sanitizer_common_libcdep.cpp

@@ -219,6 +219,32 @@ static void StopStackDepotBackgroundThread() {
 static void StopStackDepotBackgroundThread() {}
 #endif
 
+void MemCpyAccessible(void *dest, const void *src, uptr n) {
+  if (TryMemCpy(dest, src, n))
+    return;
+
+  const uptr page_size = GetPageSize();
+  uptr b = reinterpret_cast<uptr>(src);
+  uptr b_up = RoundUpTo(b, page_size);
+
+  uptr e = reinterpret_cast<uptr>(src) + n;
+  uptr e_down = RoundDownTo(e, page_size);
+
+  auto copy_or_zero = [dest, src](uptr beg, uptr end) {
+    const uptr udest = reinterpret_cast<uptr>(dest);
+    const uptr usrc = reinterpret_cast<uptr>(src);
+    void *d = reinterpret_cast<void *>(udest + (beg - usrc));
+    const uptr size = end - beg;
+    if (!TryMemCpy(d, reinterpret_cast<void *>(beg), size))
+      internal_memset(d, 0, size);
+  };
+
+  copy_or_zero(b, b_up);
+  for (uptr p = b_up; p < e_down; p += page_size)
+    copy_or_zero(p, p + page_size);
+  copy_or_zero(e_down, e);
+}
+
 }  // namespace __sanitizer
 
 SANITIZER_INTERFACE_WEAK_DEF(void, __sanitizer_sandbox_on_notify,

compiler-rt/lib/sanitizer_common/tests/sanitizer_posix_test.cpp

@@ -127,6 +127,43 @@ TEST(SanitizerCommon, TryMemCpyNull) {
   EXPECT_FALSE(TryMemCpy(dst.data(), nullptr, dst.size()));
 }
 
+TEST(SanitizerCommon, MemCpyAccessible) {
+  const int page_num = 1000;
+  const int page_size = GetPageSize();
+  InternalMmapVector<char> src(page_num * page_size);
+  std::iota(src.begin(), src.end(), 123);
+  std::vector<char> dst;
+  std::vector<char> exp = {src.begin(), src.end()};
+
+  // Protect some pages.
+  for (int i = 7; i < page_num; i *= 2) {
+    mprotect(src.data() + i * page_size, page_size, PROT_NONE);
+    std::fill(exp.data() + i * page_size, exp.data() + (i + 1) * page_size, 0);
+  }
+
+  dst.assign(src.size(), 0);
+  EXPECT_FALSE(TryMemCpy(dst.data(), src.data(), dst.size()));
+
+  // Full page aligned range with mprotect pages.
+  dst.assign(src.size(), 0);
+  MemCpyAccessible(dst.data(), src.data(), dst.size());
+  EXPECT_TRUE(std::equal(dst.begin(), dst.end(), exp.begin()));
+
+  // Misaligned range with mprotect pages.
+  size_t offb = 3;
+  size_t offe = 7;
+  dst.assign(src.size() - offb - offe, 0);
+  MemCpyAccessible(dst.data(), src.data() + offb, dst.size());
+  EXPECT_TRUE(std::equal(dst.begin(), dst.end(), exp.begin() + offb));
+
+  // Misaligned range with ends in mprotect pages.
+  offb = 3 + 7 * page_size;
+  offe = 7 + 14 * page_size;
+  dst.assign(src.size() - offb - offe, 0);
+  MemCpyAccessible(dst.data(), src.data() + offb, dst.size());
+  EXPECT_TRUE(std::equal(dst.begin(), dst.end(), exp.begin() + offb));
+}
+
 }  // namespace __sanitizer
 
 #endif  // SANITIZER_POSIX