[libc] Error fixes for mbrtowc and wcrtomb

libc/src/__support/wchar/CMakeLists.txt

@@ -27,6 +27,7 @@ add_object_library(
   SRCS 
     wcrtomb.cpp
   DEPENDS
+    libc.hdr.errno_macros
     libc.hdr.types.char32_t
     libc.hdr.types.size_t
     libc.hdr.types.wchar_t
@@ -43,6 +44,7 @@ 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/wchar/mbrtowc.cpp

@@ -7,6 +7,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "src/__support/wchar/mbrtowc.h"
+#include "hdr/errno_macros.h"
 #include "hdr/types/mbstate_t.h"
 #include "hdr/types/size_t.h"
 #include "hdr/types/wchar_t.h"
@@ -22,6 +23,8 @@ namespace internal {
 ErrorOr<size_t> mbrtowc(wchar_t *__restrict pwc, const char *__restrict s,
                         size_t n, mbstate *__restrict ps) {
   CharacterConverter char_conv(ps);
+  if (!char_conv.isValidState())
+    return Error(EINVAL);
   if (s == nullptr)
     return 0;
   size_t i = 0;
@@ -30,7 +33,7 @@ ErrorOr<size_t> mbrtowc(wchar_t *__restrict pwc, const char *__restrict s,
     int err = char_conv.push(static_cast<char8_t>(s[i]));
     // Encoding error
     if (err == -1)
-      return Error(-1);
+      return Error(EILSEQ);
   }
   auto wc = char_conv.pop_utf32();
   if (wc.has_value()) {

libc/src/__support/wchar/wcrtomb.cpp

@@ -11,6 +11,7 @@
 #include "src/__support/wchar/character_converter.h"
 #include "src/__support/wchar/mbstate.h"
 
+#include "hdr/errno_macros.h"
 #include "hdr/types/char32_t.h"
 #include "hdr/types/size_t.h"
 #include "hdr/types/wchar_t.h"
@@ -26,12 +27,15 @@ ErrorOr<size_t> wcrtomb(char *__restrict s, wchar_t wc,
 
   CharacterConverter cr(ps);
 
+  if (!cr.isValidState())
+    return Error(EINVAL);
+
   if (s == nullptr)
-    return Error(-1);
+    return Error(EILSEQ);
 
   int status = cr.push(static_cast<char32_t>(wc));
   if (status != 0)
-    return Error(status);
+    return Error(EILSEQ);
 
   size_t count = 0;
   while (!cr.isEmpty()) {

libc/src/wchar/mbrtowc.cpp

@@ -29,7 +29,7 @@ LLVM_LIBC_FUNCTION(size_t, mbrtowc,
                                    : reinterpret_cast<internal::mbstate *>(ps));
   if (!ret.has_value()) {
     // Encoding failure
-    libc_errno = EILSEQ;
+    libc_errno = ret.error();
     return -1;
   }
   return ret.value();

libc/src/wchar/wcrtomb.cpp

@@ -35,7 +35,7 @@ LLVM_LIBC_FUNCTION(size_t, wcrtomb,
                     : reinterpret_cast<internal::mbstate *>(ps));
 
   if (!result.has_value()) {
-    libc_errno = EILSEQ;
+    libc_errno = result.error();
     return -1;
   }
 

libc/test/src/wchar/CMakeLists.txt

@@ -31,10 +31,12 @@ add_libc_test(
     mbrtowc_test.cpp
   DEPENDS
     libc.src.__support.libc_errno
+    libc.src.__support.wchar.mbstate
     libc.src.string.memset
     libc.src.wchar.mbrtowc
     libc.hdr.types.mbstate_t
     libc.hdr.types.wchar_t
+    libc.test.UnitTest.ErrnoCheckingTest
 )
 
 add_libc_test(
@@ -72,6 +74,8 @@ add_libc_test(
     libc.hdr.types.wchar_t
     libc.hdr.types.mbstate_t
     libc.src.__support.libc_errno
+    libc.src.__support.wchar.mbstate
+    libc.test.UnitTest.ErrnoCheckingTest
 )
 
 add_libc_test(

libc/test/src/wchar/mbrtowc_test.cpp

@@ -9,25 +9,31 @@
 #include "hdr/types/mbstate_t.h"
 #include "hdr/types/wchar_t.h"
 #include "src/__support/libc_errno.h"
+#include "src/__support/wchar/mbstate.h"
 #include "src/string/memset.h"
 #include "src/wchar/mbrtowc.h"
+#include "test/UnitTest/ErrnoCheckingTest.h"
 #include "test/UnitTest/Test.h"
 
-TEST(LlvmLibcMBRToWC, OneByte) {
+using LlvmLibcMBRToWCTest = LIBC_NAMESPACE::testing::ErrnoCheckingTest;
+
+TEST_F(LlvmLibcMBRToWCTest, OneByte) {
   const char *ch = "A";
   wchar_t dest[2];
   // Testing if it works with nullptr mbstate_t
   mbstate_t *mb = nullptr;
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
   ASSERT_EQ(static_cast<char>(*dest), 'A');
   ASSERT_EQ(static_cast<int>(n), 1);
+  ASSERT_ERRNO_SUCCESS();
 
   // Should fail since we have not read enough
   n = LIBC_NAMESPACE::mbrtowc(dest, ch, 0, mb);
   ASSERT_EQ(static_cast<int>(n), -2);
+  ASSERT_ERRNO_SUCCESS();
 }
 
-TEST(LlvmLibcMBRToWC, TwoByte) {
+TEST_F(LlvmLibcMBRToWCTest, TwoByte) {
   const char ch[2] = {static_cast<char>(0xC2),
                       static_cast<char>(0x8E)}; // Ž car symbol
   wchar_t dest[2];
@@ -36,6 +42,7 @@ TEST(LlvmLibcMBRToWC, TwoByte) {
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 2, mb);
   ASSERT_EQ(static_cast<int>(*dest), 142);
   ASSERT_EQ(static_cast<int>(n), 2);
+  ASSERT_ERRNO_SUCCESS();
 
   // Should fail since we have not read enough
   n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
@@ -44,9 +51,10 @@ TEST(LlvmLibcMBRToWC, TwoByte) {
   n = LIBC_NAMESPACE::mbrtowc(dest, ch + 1, 1, mb);
   ASSERT_EQ(static_cast<int>(n), 1);
   ASSERT_EQ(static_cast<int>(*dest), 142);
+  ASSERT_ERRNO_SUCCESS();
 }
 
-TEST(LlvmLibcMBRToWC, ThreeByte) {
+TEST_F(LlvmLibcMBRToWCTest, ThreeByte) {
   const char ch[3] = {static_cast<char>(0xE2), static_cast<char>(0x88),
                       static_cast<char>(0x91)}; // ∑ sigma symbol
   wchar_t dest[2];
@@ -55,17 +63,20 @@ TEST(LlvmLibcMBRToWC, ThreeByte) {
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 3, mb);
   ASSERT_EQ(static_cast<int>(*dest), 8721);
   ASSERT_EQ(static_cast<int>(n), 3);
+  ASSERT_ERRNO_SUCCESS();
 
   // Should fail since we have not read enough
   n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
   ASSERT_EQ(static_cast<int>(n), -2);
+  ASSERT_ERRNO_SUCCESS();
   // Should pass after reading two more bytes
   n = LIBC_NAMESPACE::mbrtowc(dest, ch + 1, 2, mb);
   ASSERT_EQ(static_cast<int>(n), 2);
   ASSERT_EQ(static_cast<int>(*dest), 8721);
+  ASSERT_ERRNO_SUCCESS();
 }
 
-TEST(LlvmLibcMBRToWC, FourByte) {
+TEST_F(LlvmLibcMBRToWCTest, FourByte) {
   const char ch[4] = {static_cast<char>(0xF0), static_cast<char>(0x9F),
                       static_cast<char>(0xA4),
                       static_cast<char>(0xA1)}; // 🤡 clown emoji
@@ -75,27 +86,29 @@ TEST(LlvmLibcMBRToWC, FourByte) {
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 4, mb);
   ASSERT_EQ(static_cast<int>(*dest), 129313);
   ASSERT_EQ(static_cast<int>(n), 4);
-
+  ASSERT_ERRNO_SUCCESS();
   // Should fail since we have not read enough
   n = LIBC_NAMESPACE::mbrtowc(dest, ch, 2, mb);
   ASSERT_EQ(static_cast<int>(n), -2);
+  ASSERT_ERRNO_SUCCESS();
   // Should pass after reading two more bytes
   n = LIBC_NAMESPACE::mbrtowc(dest, ch + 2, 2, mb);
   ASSERT_EQ(static_cast<int>(n), 2);
   ASSERT_EQ(static_cast<int>(*dest), 129313);
+  ASSERT_ERRNO_SUCCESS();
 }
 
-TEST(LlvmLibcMBRToWC, InvalidByte) {
+TEST_F(LlvmLibcMBRToWCTest, InvalidByte) {
   const char ch[1] = {static_cast<char>(0x80)};
   wchar_t dest[2];
   mbstate_t *mb;
   LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
   ASSERT_EQ(static_cast<int>(n), -1);
-  ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
+  ASSERT_ERRNO_EQ(EILSEQ);
 }
 
-TEST(LlvmLibcMBRToWC, InvalidMultiByte) {
+TEST_F(LlvmLibcMBRToWCTest, InvalidMultiByte) {
   const char ch[4] = {static_cast<char>(0x80), static_cast<char>(0x00),
                       static_cast<char>(0x80),
                       static_cast<char>(0x00)}; // invalid sequence of bytes
@@ -105,18 +118,19 @@ TEST(LlvmLibcMBRToWC, InvalidMultiByte) {
   // Trying to push all 4 should error
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 4, mb);
   ASSERT_EQ(static_cast<int>(n), -1);
-  ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
+  ASSERT_ERRNO_EQ(EILSEQ);
   // Trying to push just the first one should error
   n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
   ASSERT_EQ(static_cast<int>(n), -1);
-  ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
+  ASSERT_ERRNO_EQ(EILSEQ);
   // Trying to push the second and third should correspond to null wc
   n = LIBC_NAMESPACE::mbrtowc(dest, ch + 1, 2, mb);
   ASSERT_EQ(static_cast<int>(n), 0);
   ASSERT_TRUE(*dest == L'\0');
+  ASSERT_ERRNO_SUCCESS();
 }
 
-TEST(LlvmLibcMBRToWC, InvalidLastByte) {
+TEST_F(LlvmLibcMBRToWCTest, InvalidLastByte) {
   // Last byte is invalid since it does not have correct starting sequence.
   // 0xC0 --> 11000000 starting sequence should be 10xxxxxx
   const char ch[4] = {static_cast<char>(0xF1), static_cast<char>(0x80),
@@ -127,10 +141,10 @@ TEST(LlvmLibcMBRToWC, InvalidLastByte) {
   // Trying to push all 4 should error
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 4, mb);
   ASSERT_EQ(static_cast<int>(n), -1);
-  ASSERT_EQ(static_cast<int>(libc_errno), EILSEQ);
+  ASSERT_ERRNO_EQ(EILSEQ);
 }
 
-TEST(LlvmLibcMBRToWC, ValidTwoByteWithExtraRead) {
+TEST_F(LlvmLibcMBRToWCTest, ValidTwoByteWithExtraRead) {
   const char ch[3] = {static_cast<char>(0xC2), static_cast<char>(0x8E),
                       static_cast<char>(0x80)};
   wchar_t dest[2];
@@ -140,9 +154,10 @@ TEST(LlvmLibcMBRToWC, ValidTwoByteWithExtraRead) {
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 3, mb);
   ASSERT_EQ(static_cast<int>(n), 2);
   ASSERT_EQ(static_cast<int>(*dest), 142);
+  ASSERT_ERRNO_SUCCESS();
 }
 
-TEST(LlvmLibcMBRToWC, TwoValidTwoBytes) {
+TEST_F(LlvmLibcMBRToWCTest, TwoValidTwoBytes) {
   const char ch[4] = {static_cast<char>(0xC2), static_cast<char>(0x8E),
                       static_cast<char>(0xC7), static_cast<char>(0x8C)};
   wchar_t dest[2];
@@ -152,21 +167,39 @@ TEST(LlvmLibcMBRToWC, TwoValidTwoBytes) {
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 2, mb);
   ASSERT_EQ(static_cast<int>(n), 2);
   ASSERT_EQ(static_cast<int>(*dest), 142);
+  ASSERT_ERRNO_SUCCESS();
   n = LIBC_NAMESPACE::mbrtowc(dest + 1, ch + 2, 2, mb);
   ASSERT_EQ(static_cast<int>(n), 2);
   ASSERT_EQ(static_cast<int>(*(dest + 1)), 460);
+  ASSERT_ERRNO_SUCCESS();
 }
 
-TEST(LlvmLibcMBRToWC, NullString) {
+TEST_F(LlvmLibcMBRToWCTest, NullString) {
   wchar_t dest[2] = {L'O', L'K'};
   mbstate_t *mb;
   LIBC_NAMESPACE::memset(&mb, 0, sizeof(mbstate_t));
   // reading on nullptr should return 0
   size_t n = LIBC_NAMESPACE::mbrtowc(dest, nullptr, 2, mb);
   ASSERT_EQ(static_cast<int>(n), 0);
   ASSERT_TRUE(dest[0] == L'O');
+  ASSERT_ERRNO_SUCCESS();
   // reading a null terminator should return 0
   const char *ch = "\0";
   n = LIBC_NAMESPACE::mbrtowc(dest, ch, 1, mb);
   ASSERT_EQ(static_cast<int>(n), 0);
+  ASSERT_ERRNO_SUCCESS();
+}
+
+TEST_F(LlvmLibcMBRToWCTest, InvalidMBState) {
+  const char ch[4] = {static_cast<char>(0xC2), static_cast<char>(0x8E),
+                      static_cast<char>(0xC7), static_cast<char>(0x8C)};
+  wchar_t dest[2] = {L'O', L'K'};
+  mbstate_t *mb;
+  LIBC_NAMESPACE::internal::mbstate inv;
+  inv.total_bytes = 6;
+  mb = reinterpret_cast<mbstate_t *>(&inv);
+  // invalid mbstate should error
+  size_t n = LIBC_NAMESPACE::mbrtowc(dest, ch, 2, mb);
+  ASSERT_EQ(static_cast<int>(n), -1);
+  ASSERT_ERRNO_EQ(EINVAL);
 }