[libc] Reland wchar string conversion mb to wc (#151048)

Added crash on nullptr to mbstowcs

---------

Co-authored-by: Sriya Pratipati <sriyap@google.com>

libc/config/linux/x86_64/entrypoints.txt

@@ -1273,6 +1273,9 @@ if(LLVM_LIBC_FULL_BUILD)
     libc.src.wchar.mbsinit
     libc.src.wchar.mbrtowc
     libc.src.wchar.mbtowc
+    libc.src.wchar.mbstowcs
+    libc.src.wchar.mbsrtowcs
+    libc.src.wchar.mbsnrtowcs
     libc.src.wchar.wcrtomb
     libc.src.wchar.wctomb
     libc.src.wchar.wcstombs

libc/include/wchar.yaml

@@ -53,6 +53,33 @@ functions:
       - type: wchar_t *__restrict
       - type: const char *__restrict
       - type: size_t
+  - name: mbsnrtowcs
+    standards:
+      - stdc
+    return_type: size_t
+    arguments:
+      - type: wchar_t *__restrict
+      - type: const char **__restrict
+      - type: size_t
+      - type: size_t
+      - type: mbstate_t *__restrict
+  - name: mbsrtowcs
+    standards:
+      - stdc
+    return_type: size_t
+    arguments:
+      - type: wchar_t *__restrict
+      - type: const char **__restrict
+      - type: size_t
+      - type: mbstate_t *__restrict
+  - name: mbstowcs
+    standards:
+      - stdc
+    return_type: size_t
+    arguments:
+      - type: wchar_t *__restrict
+      - type: const char *__restrict
+      - type: size_t
   - name: mbsinit
     standards:
       - stdc

libc/src/__support/wchar/CMakeLists.txt

@@ -60,14 +60,31 @@ add_object_library(
   SRCS
     mbrtowc.cpp
   DEPENDS
-  libc.hdr.errno_macros
-  libc.hdr.types.wchar_t
-  libc.hdr.types.size_t
-  libc.src.__support.common
-  libc.src.__support.error_or
-  libc.src.__support.macros.config
-  .character_converter
-  .mbstate
+    libc.hdr.errno_macros
+    libc.hdr.types.wchar_t
+    libc.hdr.types.size_t
+    libc.src.__support.common
+    libc.src.__support.error_or
+    libc.src.__support.macros.config
+    .character_converter
+    .mbstate
+)
+
+add_header_library(
+  mbsnrtowcs
+  HDRS
+    mbsnrtowcs.h
+  DEPENDS
+    libc.hdr.errno_macros
+    libc.hdr.types.wchar_t
+    libc.hdr.types.size_t
+    libc.src.__support.common
+    libc.src.__support.error_or
+    libc.src.__support.macros.config
+    libc.src.__support.macros.null_check
+    .character_converter
+    .mbstate
+    .string_converter
 )
 
 add_header_library(

libc/src/__support/wchar/mbsnrtowcs.h

@@ -0,0 +1,66 @@
+//===-- Implementation for mbsnrtowcs function ------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS_H
+#define LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS_H
+
+#include "hdr/errno_macros.h"
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/common.h"
+#include "src/__support/error_or.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/null_check.h"
+#include "src/__support/wchar/character_converter.h"
+#include "src/__support/wchar/mbstate.h"
+#include "src/__support/wchar/string_converter.h"
+
+namespace LIBC_NAMESPACE_DECL {
+namespace internal {
+
+LIBC_INLINE static ErrorOr<size_t> mbsnrtowcs(wchar_t *__restrict dst,
+                                              const char **__restrict src,
+                                              size_t nmc, size_t len,
+                                              mbstate *__restrict ps) {
+  LIBC_CRASH_ON_NULLPTR(src);
+  // Checking if mbstate is valid
+  CharacterConverter char_conv(ps);
+  if (!char_conv.isValidState())
+    return Error(EINVAL);
+
+  StringConverter<char8_t> str_conv(reinterpret_cast<const char8_t *>(*src), ps,
+                                    len, nmc);
+  size_t dst_idx = 0;
+  ErrorOr<char32_t> converted = str_conv.popUTF32();
+  while (converted.has_value()) {
+    if (dst != nullptr)
+      dst[dst_idx] = converted.value();
+    // null terminator should not be counted in return value
+    if (converted.value() == L'\0') {
+      if (dst != nullptr)
+        *src = nullptr;
+      return dst_idx;
+    }
+    dst_idx++;
+    converted = str_conv.popUTF32();
+  }
+
+  if (converted.error() == -1) { // if we hit conversion limit
+    if (dst != nullptr)
+      *src += str_conv.getSourceIndex();
+    return dst_idx;
+  }
+
+  return Error(converted.error());
+}
+
+} // namespace internal
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC___SUPPORT_WCHAR_MBSNRTOWCS_H

libc/src/wchar/CMakeLists.txt

@@ -184,6 +184,55 @@ add_entrypoint_object(
     libc.src.__support.wchar.mbstate
 )
 
+add_entrypoint_object(
+  mbstowcs
+  SRCS
+    mbstowcs.cpp
+  HDRS
+    mbstowcs.h
+  DEPENDS
+    libc.hdr.types.size_t
+    libc.hdr.types.wchar_t
+    libc.src.__support.common
+    libc.src.__support.macros.config
+    libc.src.__support.macros.null_check
+    libc.src.__support.libc_errno
+    libc.src.__support.wchar.mbstate
+    libc.src.__support.wchar.mbsnrtowcs
+)
+
+add_entrypoint_object(
+  mbsrtowcs
+  SRCS
+    mbsrtowcs.cpp
+  HDRS
+    mbsrtowcs.h
+  DEPENDS
+    libc.hdr.types.size_t
+    libc.hdr.types.wchar_t
+    libc.src.__support.common
+    libc.src.__support.macros.config
+    libc.src.__support.libc_errno
+    libc.src.__support.wchar.mbstate
+    libc.src.__support.wchar.mbsnrtowcs
+)
+
+add_entrypoint_object(
+  mbsnrtowcs
+  SRCS
+    mbsnrtowcs.cpp
+  HDRS
+    mbsnrtowcs.h
+  DEPENDS
+    libc.hdr.types.size_t
+    libc.hdr.types.wchar_t
+    libc.src.__support.common
+    libc.src.__support.macros.config
+    libc.src.__support.libc_errno
+    libc.src.__support.wchar.mbstate
+    libc.src.__support.wchar.mbsnrtowcs
+)
+
 add_entrypoint_object(
   wcstombs
   SRCS

libc/src/wchar/mbsnrtowcs.cpp

@@ -0,0 +1,39 @@
+//===-- Implementation of mbsnrtowcs --------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/wchar/mbsnrtowcs.h"
+
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/common.h"
+#include "src/__support/libc_errno.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/wchar/mbsnrtowcs.h"
+#include "src/__support/wchar/mbstate.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(size_t, mbsnrtowcs,
+                   (wchar_t *__restrict dst, const char **__restrict src,
+                    size_t nmc, size_t len, mbstate_t *__restrict ps)) {
+  static internal::mbstate internal_mbstate;
+  // If destination is null, ignore len
+  len = dst == nullptr ? SIZE_MAX : len;
+  auto ret = internal::mbsnrtowcs(
+      dst, src, nmc, len,
+      ps == nullptr ? &internal_mbstate
+                    : reinterpret_cast<internal::mbstate *>(ps));
+  if (!ret.has_value()) {
+    // Encoding failure
+    libc_errno = ret.error();
+    return -1;
+  }
+  return ret.value();
+}
+
+} // namespace LIBC_NAMESPACE_DECL

libc/src/wchar/mbsnrtowcs.h

@@ -0,0 +1,24 @@
+//===-- Implementation header for mbsnrtowcs ------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_WCHAR_MBSNRTOWCS_H
+#define LLVM_LIBC_SRC_WCHAR_MBSNRTOWCS_H
+
+#include "hdr/types/mbstate_t.h"
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+size_t mbsnrtowcs(wchar_t *__restrict dst, const char **__restrict src,
+                  size_t nmc, size_t len, mbstate_t *__restrict ps);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_WCHAR_MBSNRTOWCS_H

libc/src/wchar/mbsrtowcs.cpp

@@ -0,0 +1,39 @@
+//===-- Implementation of mbsrtowcs ---------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/wchar/mbsrtowcs.h"
+
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/common.h"
+#include "src/__support/libc_errno.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/wchar/mbsnrtowcs.h"
+#include "src/__support/wchar/mbstate.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(size_t, mbsrtowcs,
+                   (wchar_t *__restrict dst, const char **__restrict src,
+                    size_t len, mbstate_t *__restrict ps)) {
+  static internal::mbstate internal_mbstate;
+  // If destination is null, ignore len
+  len = dst == nullptr ? SIZE_MAX : len;
+  auto ret = internal::mbsnrtowcs(
+      dst, src, SIZE_MAX, len,
+      ps == nullptr ? &internal_mbstate
+                    : reinterpret_cast<internal::mbstate *>(ps));
+  if (!ret.has_value()) {
+    // Encoding failure
+    libc_errno = ret.error();
+    return -1;
+  }
+  return ret.value();
+}
+
+} // namespace LIBC_NAMESPACE_DECL

libc/src/wchar/mbsrtowcs.h

@@ -0,0 +1,24 @@
+//===-- Implementation header for mbsrtowcs -------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_WCHAR_MBSRTOWCS_H
+#define LLVM_LIBC_SRC_WCHAR_MBSRTOWCS_H
+
+#include "hdr/types/mbstate_t.h"
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+size_t mbsrtowcs(wchar_t *__restrict dst, const char **__restrict src,
+                 size_t len, mbstate_t *__restrict ps);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_WCHAR_MBSRTOWCS_H

libc/src/wchar/mbstowcs.cpp

@@ -0,0 +1,40 @@
+//===-- Implementation of mbstowcs ----------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/wchar/mbstowcs.h"
+
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/common.h"
+#include "src/__support/libc_errno.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/null_check.h"
+#include "src/__support/wchar/mbsnrtowcs.h"
+#include "src/__support/wchar/mbstate.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(size_t, mbstowcs,
+                   (wchar_t *__restrict pwcs, const char *__restrict s,
+                    size_t n)) {
+  LIBC_CRASH_ON_NULLPTR(s);
+  // If destination is null, ignore n
+  n = pwcs == nullptr ? SIZE_MAX : n;
+  static internal::mbstate internal_mbstate;
+  const char *temp = s;
+  auto ret = internal::mbsnrtowcs(pwcs, &temp, SIZE_MAX, n, &internal_mbstate);
+
+  if (!ret.has_value()) {
+    // Encoding failure
+    libc_errno = ret.error();
+    return -1;
+  }
+  return ret.value();
+}
+
+} // namespace LIBC_NAMESPACE_DECL

libc/src/wchar/mbstowcs.h

@@ -0,0 +1,22 @@
+//===-- Implementation header for mbstowcs --------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_WCHAR_MBSTOWCS_H
+#define LLVM_LIBC_SRC_WCHAR_MBSTOWCS_H
+
+#include "hdr/types/size_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+size_t mbstowcs(wchar_t *__restrict pwcs, const char *__restrict s, size_t n);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_WCHAR_MBSTOWCS_H

libc/test/src/wchar/CMakeLists.txt

@@ -75,6 +75,19 @@ add_libc_test(
     libc.test.UnitTest.ErrnoCheckingTest
 )
 
+add_libc_test(
+  mbstowcs_test
+  SUITE
+    libc_wchar_unittests
+  SRCS
+    mbstowcs_test.cpp
+  DEPENDS
+    libc.src.__support.libc_errno
+    libc.src.wchar.mbstowcs
+    libc.hdr.types.wchar_t
+    libc.test.UnitTest.ErrnoCheckingTest
+)
+
 add_libc_test(
   mblen_test
   SUITE
@@ -87,6 +100,22 @@ add_libc_test(
     libc.test.UnitTest.ErrnoCheckingTest
 )
 
+add_libc_test(
+  mbsrtowcs_test
+  SUITE
+    libc_wchar_unittests
+  SRCS
+    mbsrtowcs_test.cpp
+  DEPENDS
+    libc.src.__support.libc_errno
+    libc.src.__support.wchar.mbstate
+    libc.src.string.memset
+    libc.src.wchar.mbsrtowcs
+    libc.hdr.types.mbstate_t
+    libc.hdr.types.wchar_t
+    libc.test.UnitTest.ErrnoCheckingTest
+)
+
 add_libc_test(
   mbrlen_test
   SUITE
@@ -97,7 +126,23 @@ add_libc_test(
     libc.src.__support.libc_errno
     libc.src.__support.wchar.mbstate
     libc.src.string.memset
-    libc.src.wchar.mbrlen
+    libc.src.wchar.mbsrlen
+    libc.hdr.types.mbstate_t
+    libc.hdr.types.wchar_t
+    libc.test.UnitTest.ErrnoCheckingTest
+)
+
+add_libc_test(
+  mbsnrtowcs_test
+  SUITE
+    libc_wchar_unittests
+  SRCS
+    mbsnrtowcs_test.cpp
+  DEPENDS
+    libc.src.__support.libc_errno
+    libc.src.__support.wchar.mbstate
+    libc.src.string.memset
+    libc.src.wchar.mbsnrtowcs
     libc.hdr.types.mbstate_t
     libc.hdr.types.wchar_t
     libc.test.UnitTest.ErrnoCheckingTest

libc/test/src/wchar/mbsnrtowcs_test.cpp

@@ -0,0 +1,212 @@
+//===-- Unittests for mbsetowcs -------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "hdr/types/mbstate_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/libc_errno.h"
+#include "src/__support/macros/null_check.h"
+#include "src/__support/wchar/mbstate.h"
+#include "src/string/memset.h"
+#include "src/wchar/mbsnrtowcs.h"
+#include "test/UnitTest/ErrnoCheckingTest.h"
+#include "test/UnitTest/Test.h"
+
+using LlvmLibcMBSNRToWCSTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
+
+TEST_F(LlvmLibcMBSNRToWCSTest, OneByteOneChar) {
+  const char *ch = "A";
+  const char *original = ch;
+  wchar_t dest[2];
+  mbstate_t mb;
+  LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+  size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &ch, 1, 1, &mb);
+  ASSERT_EQ(static_cast<char>(*dest), 'A');
+  ASSERT_EQ(static_cast<int>(n), 1);
+  // Should point to null terminator now
+  ASSERT_EQ(ch, original + 1);
+  ASSERT_ERRNO_SUCCESS();
+
+  n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &ch, 1, 1, &mb);
+  ASSERT_EQ(static_cast<char>(dest[1]), '\0');
+  // Should not include null terminator
+  ASSERT_EQ(static_cast<int>(n), 0);
+  // Should now be a nullptr
+  ASSERT_EQ(ch, nullptr);
+  ASSERT_ERRNO_SUCCESS();
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, FourByteOneChar) {
+  const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
+  const char *original = src;
+  wchar_t dest[2];
+  mbstate_t mb;
+  LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+  // Not enough bytes for the full character
+  size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 3, 2, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(n), 0);
+  ASSERT_EQ(src, original + 3);
+  // Needs 2 more bytes (last byte of cat + null terminator)
+  n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 2, 2, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  // Does not include null terminator
+  ASSERT_EQ(static_cast<int>(n), 1);
+  ASSERT_EQ(src, nullptr);
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_TRUE(dest[1] == L'\0');
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, MixedNumberOfBytes) {
+  // 'A', sigma symbol 'Σ', recycling symbol '♻', laughing cat emoji '😹'
+  const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[5];
+  mbstate_t mb;
+  LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+
+  // Read 'A'
+  size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 1, 1, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<char>(dest[0]), 'A');
+  ASSERT_EQ(static_cast<int>(n), 1);
+  ASSERT_EQ(src, original + 1);
+
+  // Read sigma 'Σ'
+  n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &src, 2, 1, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(dest[1]), 931);
+  ASSERT_EQ(static_cast<int>(n), 1);
+  ASSERT_EQ(src, original + 3);
+
+  // Read recycling '♻'
+  n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 2, 5, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(n), 0);
+  ASSERT_EQ(src, original + 5);
+  n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 1, 1, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(n), 1);
+  ASSERT_EQ(src, original + 6);
+  ASSERT_EQ(static_cast<int>(dest[2]), 9851);
+
+  // Read laughing cat emoji '😹'
+  n = LIBC_NAMESPACE::mbsnrtowcs(dest + 3, &src, 4, 5, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(n), 1);
+  ASSERT_EQ(src, original + 10);
+  ASSERT_EQ(static_cast<int>(dest[3]), 128569);
+
+  n = LIBC_NAMESPACE::mbsnrtowcs(dest + 4, &src, 4, 4, nullptr);
+  ASSERT_TRUE(dest[4] == L'\0');
+  ASSERT_ERRNO_SUCCESS();
+  // Should not count null terminator in number
+  ASSERT_EQ(static_cast<int>(n), 0);
+  // Should now be a nullptr
+  ASSERT_EQ(src, nullptr);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, ReadLessThanStringLength) {
+  // Four laughing cat emojis "😹😹😹😹"
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[5] = {L'a', L'b', L'c', L'd', L'e'};
+  size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 100, 3, nullptr);
+  ASSERT_ERRNO_SUCCESS();
+  // Should have read 3 emojis
+  ASSERT_EQ(static_cast<int>(n), 3);
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[2]), 128569);
+  ASSERT_TRUE(dest[3] == L'd');
+  ASSERT_TRUE(dest[4] == L'e');
+  // Read three laughing cat emojis, 12 bytes
+  ASSERT_EQ(src, original + 12);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, InvalidFirstByte) {
+  // 0x80 is invalid first byte of mb character
+  const char *src =
+      "\x80\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[3];
+  size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 88, 88, nullptr);
+  // Should return error and set errno
+  ASSERT_EQ(static_cast<int>(n), -1);
+  ASSERT_ERRNO_EQ(EILSEQ);
+  // Should not update pointer
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, InvalidMiddleByte) {
+  // The 7th byte is invalid for a 4 byte character
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[3];
+  mbstate_t mb;
+  LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+  // Successfully read one character and first byte of the second character
+  size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 5, 88, &mb);
+  ASSERT_EQ(static_cast<int>(n), 1);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(src, original + 5);
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+
+  n = LIBC_NAMESPACE::mbsnrtowcs(dest + 1, &src, 5, 88, &mb);
+  // Should return error, set errno, and not update the pointer
+  ASSERT_EQ(static_cast<int>(n), -1);
+  ASSERT_ERRNO_EQ(EILSEQ);
+  ASSERT_EQ(src, original + 5);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, NullDestination) {
+  // Four laughing cat emojis "😹😹😹😹"
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  size_t n = LIBC_NAMESPACE::mbsnrtowcs(nullptr, &src, 88, 88, nullptr);
+  ASSERT_ERRNO_SUCCESS();
+  // Null destination should ignore len and read till end of string
+  ASSERT_EQ(static_cast<int>(n), 4);
+  // It should also not change the src pointer
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSNRToWCSTest, ErrnoChecks) {
+  // Two laughing cat emojis and invalid 3rd mb char (3rd byte of it)
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[5];
+  mbstate_t mb;
+  LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+  // First two bytes are valid --> should not set errno
+  size_t n = LIBC_NAMESPACE::mbsnrtowcs(dest, &src, 80, 2, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(n), 2);
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+  ASSERT_EQ(src, original + 8);
+
+  // Trying to read the 3rd byte should set errno
+  n = LIBC_NAMESPACE::mbsnrtowcs(dest + 2, &src, 4, 2, &mb);
+  ASSERT_ERRNO_EQ(EILSEQ);
+  ASSERT_EQ(static_cast<int>(n), -1);
+  // Should not move the pointer
+  ASSERT_EQ(src, original + 8);
+}
+
+#if defined(LIBC_ADD_NULL_CHECKS)
+TEST(LlvmLibcMBSNRToWCSTest, NullptrCrash) {
+  // Passing in a nullptr should crash the program.
+  EXPECT_DEATH(
+      [] { LIBC_NAMESPACE::mbsnrtowcs(nullptr, nullptr, 1, 1, nullptr); },
+      WITH_SIGNAL(-1));
+}
+#endif // LIBC_ADD_NULL_CHECKS

libc/test/src/wchar/mbsrtowcs_test.cpp

@@ -0,0 +1,185 @@
+//===-- Unittests for mbsetowcs -------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "hdr/types/mbstate_t.h"
+#include "hdr/types/wchar_t.h"
+#include "src/__support/libc_errno.h"
+#include "src/__support/macros/null_check.h"
+#include "src/__support/wchar/mbstate.h"
+#include "src/string/memset.h"
+#include "src/wchar/mbsrtowcs.h"
+#include "test/UnitTest/ErrnoCheckingTest.h"
+#include "test/UnitTest/Test.h"
+
+using LlvmLibcMBSRToWCSTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
+
+TEST_F(LlvmLibcMBSRToWCSTest, OneByteOneChar) {
+  const char *ch = "A";
+  const char *original = ch;
+  wchar_t dest[2];
+  mbstate_t mb;
+  LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+  size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &ch, 1, &mb);
+  ASSERT_EQ(static_cast<char>(*dest), 'A');
+  ASSERT_EQ(static_cast<int>(n), 1);
+  // Should point to null terminator now
+  ASSERT_EQ(ch, original + 1);
+  ASSERT_ERRNO_SUCCESS();
+
+  n = LIBC_NAMESPACE::mbsrtowcs(dest + 1, &ch, 1, &mb);
+  ASSERT_EQ(static_cast<char>(dest[1]), '\0');
+  // Should not include null terminator
+  ASSERT_EQ(static_cast<int>(n), 0);
+  // Should now be a nullptr
+  ASSERT_EQ(ch, nullptr);
+  ASSERT_ERRNO_SUCCESS();
+}
+
+TEST_F(LlvmLibcMBSRToWCSTest, FourByteOneChar) {
+  const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
+  wchar_t dest[2];
+  mbstate_t mb;
+  LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+  size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 2, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_TRUE(dest[1] == L'\0');
+  // Should not count null terminator in number
+  ASSERT_EQ(static_cast<int>(n), 1);
+  // Should now be a nullptr
+  ASSERT_EQ(src, nullptr);
+}
+
+TEST_F(LlvmLibcMBSRToWCSTest, MultiByteTwoCharacters) {
+  // Two laughing cat emojis "😹😹"
+  const char *src = "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  wchar_t dest[3];
+  mbstate_t mb;
+  LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
+  size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 3, &mb);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+  ASSERT_TRUE(dest[2] == L'\0');
+  // Should not count null terminator in number
+  ASSERT_EQ(static_cast<int>(n), 2);
+  // Should now be a nullptr
+  ASSERT_EQ(src, nullptr);
+}
+
+TEST_F(LlvmLibcMBSRToWCSTest, MixedNumberOfBytes) {
+  // 'A', sigma symbol 'Σ', recycling symbol '♻', laughing cat emoji '😹'
+  const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[5];
+  size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 4, nullptr);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<char>(dest[0]), 'A');
+  ASSERT_EQ(static_cast<int>(dest[1]), 931);
+  ASSERT_EQ(static_cast<int>(dest[2]), 9851);
+  ASSERT_EQ(static_cast<int>(dest[3]), 128569);
+  // Should point to null terminator (byte at 10th index)
+  ASSERT_EQ(src, original + 10);
+  ASSERT_EQ(static_cast<int>(n), 4);
+  n = LIBC_NAMESPACE::mbsrtowcs(dest + 4, &src, 4, nullptr);
+  ASSERT_TRUE(dest[4] == L'\0');
+  ASSERT_ERRNO_SUCCESS();
+  // Should not count null terminator in number
+  ASSERT_EQ(static_cast<int>(n), 0);
+  // Should now be a nullptr
+  ASSERT_EQ(src, nullptr);
+}
+
+TEST_F(LlvmLibcMBSRToWCSTest, ReadLessThanStringLength) {
+  // Four laughing cat emojis "😹😹😹😹"
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[5] = {L'a', L'b', L'c', L'd', L'e'};
+  size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 3, nullptr);
+  ASSERT_ERRNO_SUCCESS();
+  // Should have read 3 emojis
+  ASSERT_EQ(static_cast<int>(n), 3);
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[2]), 128569);
+  ASSERT_TRUE(dest[3] == L'd');
+  ASSERT_TRUE(dest[4] == L'e');
+  // Read three laughing cat emojis, 12 bytes
+  ASSERT_EQ(src, original + 12);
+}
+
+TEST_F(LlvmLibcMBSRToWCSTest, InvalidFirstByte) {
+  // 0x80 is invalid first byte of mb character
+  const char *src =
+      "\x80\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[3];
+  size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 3, nullptr);
+  // Should return error and set errno
+  ASSERT_EQ(static_cast<int>(n), -1);
+  ASSERT_ERRNO_EQ(EILSEQ);
+  // Should not update pointer
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSRToWCSTest, InvalidMiddleByte) {
+  // The 7th byte is invalid for a 4 byte character
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[3];
+  size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 5, nullptr);
+  // Should return error, set errno, and not update the pointer
+  ASSERT_EQ(static_cast<int>(n), -1);
+  ASSERT_ERRNO_EQ(EILSEQ);
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSRToWCSTest, NullDestination) {
+  // Four laughing cat emojis "😹😹😹😹"
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  size_t n = LIBC_NAMESPACE::mbsrtowcs(nullptr, &src, 2, nullptr);
+  ASSERT_ERRNO_SUCCESS();
+  // Null destination should ignore len and read till end of string
+  ASSERT_EQ(static_cast<int>(n), 4);
+  // It should also not change the src pointer
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSRToWCSTest, ErrnoChecks) {
+  // Two laughing cat emojis and invalid 3rd mb char (3rd byte of it)
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[5];
+  // First two bytes are valid --> should not set errno
+  size_t n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 2, nullptr);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(n), 2);
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+  ASSERT_EQ(src, original + 8);
+
+  // Trying to read the 3rd byte should set errno
+  n = LIBC_NAMESPACE::mbsrtowcs(dest, &src, 2, nullptr);
+  ASSERT_ERRNO_EQ(EILSEQ);
+  ASSERT_EQ(static_cast<int>(n), -1);
+  // Should not move the pointer
+  ASSERT_EQ(src, original + 8);
+}
+
+#if defined(LIBC_ADD_NULL_CHECKS)
+TEST(LlvmLibcMBSRToWCSTest, NullptrCrash) {
+  // Passing in a nullptr should crash the program.
+  EXPECT_DEATH([] { LIBC_NAMESPACE::mbsrtowcs(nullptr, nullptr, 1, nullptr); },
+               WITH_SIGNAL(-1));
+}
+#endif // LIBC_ADD_NULL_CHECKS

libc/test/src/wchar/mbstowcs_test.cpp

@@ -0,0 +1,171 @@
+//===-- Unittests for mbstowcs --------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "hdr/types/wchar_t.h"
+#include "src/__support/libc_errno.h"
+#include "src/__support/macros/null_check.h"
+#include "src/wchar/mbstowcs.h"
+#include "test/UnitTest/ErrnoCheckingTest.h"
+#include "test/UnitTest/Test.h"
+
+using LlvmLibcMBSToWCSTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
+
+TEST_F(LlvmLibcMBSToWCSTest, OneByteOneChar) {
+  const char *ch = "A";
+  const char *original = ch;
+  wchar_t dest[2];
+  size_t n = LIBC_NAMESPACE::mbstowcs(dest, ch, 1);
+  ASSERT_EQ(static_cast<char>(*dest), 'A');
+  ASSERT_EQ(static_cast<int>(n), 1);
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(ch, original);
+  ASSERT_ERRNO_SUCCESS();
+
+  n = LIBC_NAMESPACE::mbstowcs(dest + 1, ch + 1, 1);
+  ASSERT_EQ(static_cast<char>(dest[1]), '\0');
+  // Should not include null terminator
+  ASSERT_EQ(static_cast<int>(n), 0);
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(ch, original);
+  ASSERT_ERRNO_SUCCESS();
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, FourByteOneChar) {
+  const char *src = "\xf0\x9f\x98\xb9"; // laughing cat emoji 😹
+  const char *original = src;
+  wchar_t dest[2];
+  size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_TRUE(dest[1] == L'\0');
+  // Should not count null terminator in number
+  ASSERT_EQ(static_cast<int>(n), 1);
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, MultiByteTwoCharacters) {
+  // Two laughing cat emojis "😹😹"
+  const char *src = "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[3];
+  size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+  ASSERT_TRUE(dest[2] == L'\0');
+  // Should not count null terminator in number
+  ASSERT_EQ(static_cast<int>(n), 2);
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, MixedNumberOfBytes) {
+  // 'A', sigma symbol 'Σ', recycling symbol '♻', laughing cat emoji '😹'
+  const char *src = "A\xce\xa3\xe2\x99\xbb\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[5];
+  size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<char>(dest[0]), 'A');
+  ASSERT_EQ(static_cast<int>(dest[1]), 931);
+  ASSERT_EQ(static_cast<int>(dest[2]), 9851);
+  ASSERT_EQ(static_cast<int>(dest[3]), 128569);
+  ASSERT_TRUE(dest[4] == L'\0');
+  // Should not count null terminator in number
+  ASSERT_EQ(static_cast<int>(n), 4);
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, ReadLessThanStringLength) {
+  // Four laughing cat emojis "😹😹😹😹"
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[5] = {L'a', L'b', L'c', L'd', L'e'};
+  size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
+  ASSERT_ERRNO_SUCCESS();
+  // Should have read 3 emojis
+  ASSERT_EQ(static_cast<int>(n), 3);
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[2]), 128569);
+  ASSERT_TRUE(dest[3] == L'd');
+  ASSERT_TRUE(dest[4] == L'e');
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, InvalidFirstByte) {
+  // 0x80 is invalid first byte of mb character
+  const char *src =
+      "\x80\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  wchar_t dest[3];
+  size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 3);
+  // Should return error and set errno
+  ASSERT_EQ(static_cast<int>(n), -1);
+  ASSERT_ERRNO_EQ(EILSEQ);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, InvalidMiddleByte) {
+  // The 7th byte is invalid for a 4 byte character
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[3];
+  size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 5);
+  // Should return error and set errno
+  ASSERT_EQ(static_cast<int>(n), -1);
+  ASSERT_ERRNO_EQ(EILSEQ);
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, NullDestination) {
+  // Four laughing cat emojis "😹😹😹😹"
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  size_t n = LIBC_NAMESPACE::mbstowcs(nullptr, src, 2);
+  ASSERT_ERRNO_SUCCESS();
+  // Null destination should ignore len and read till end of string
+  ASSERT_EQ(static_cast<int>(n), 4);
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(src, original);
+}
+
+TEST_F(LlvmLibcMBSToWCSTest, ErrnoChecks) {
+  // Two laughing cat emojis and invalid 3rd mb char (3rd byte of it)
+  const char *src =
+      "\xf0\x9f\x98\xb9\xf0\x9f\x98\xb9\xf0\x9f\xf0\xb9\xf0\x9f\x98\xb9";
+  const char *original = src;
+  wchar_t dest[5];
+  // First two bytes are valid --> should not set errno
+  size_t n = LIBC_NAMESPACE::mbstowcs(dest, src, 2);
+  ASSERT_ERRNO_SUCCESS();
+  ASSERT_EQ(static_cast<int>(n), 2);
+  ASSERT_EQ(static_cast<int>(dest[0]), 128569);
+  ASSERT_EQ(static_cast<int>(dest[1]), 128569);
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(src, original);
+  // Trying to read the 3rd byte should set errno
+  n = LIBC_NAMESPACE::mbstowcs(dest, src + 2, 2);
+  ASSERT_ERRNO_EQ(EILSEQ);
+  ASSERT_EQ(static_cast<int>(n), -1);
+  // Making sure the pointer is not getting updated
+  ASSERT_EQ(src, original);
+}
+
+#if defined(LIBC_ADD_NULL_CHECKS)
+TEST(LlvmLibcMBSToWCSTest, NullptrCrash) {
+  // Passing in a nullptr should crash the program.
+  EXPECT_DEATH([] { LIBC_NAMESPACE::mbstowcs(nullptr, nullptr, 1); },
+               WITH_SIGNAL(-1));
+}
+#endif // LIBC_ADD_NULL_CHECKS