commit caee99519a2f385975f7341a44cf9936a55e1335
author Kirill Makurin <maiddaisuki@outlook.com> Fri Aug 22 02:35:51 2025 +0900
committer Martin Storsjö <martin@martin.st> Mon Aug 25 12:41:01 2025 +0300
tree 72c62a81167b899bb3fd63da4e1fa86f026f43ad
parent 152d8c6b600718264385a3919d290660c771a109

crt: fix return value of mbrlen and mbrtowc functions with DBCS code pages

CRT's mbrlen and mbrtowc functions return incorrect value if
previous call to either of them has returned (size_t)-2 when
converting using DBCS code page.

Previously, mingw-w64's implementation of those functions was
implementing the same buggy behavior for compatibility with CRT.

This commit fixes return value of mbrlen and mbrtowc functions.
Tests for mbrlen and mbrtowc functions are updated accordingly.

Signed-off-by: Kirill Makurin <maiddaisuki@outlook.com>
Signed-off-by: Martin Storsjö <martin@martin.st>

mingw-w64-crt/misc/mbrtowc.c

diff --git a/mingw-w64-crt/misc/mbrtowc.c b/mingw-w64-crt/misc/mbrtowc.c
index be5830a..75c9f98 100644
--- a/mingw-w64-crt/misc/mbrtowc.c
+++ b/mingw-w64-crt/misc/mbrtowc.c
@@ -68,8 +68,12 @@
   /* Length of potential multibyte character */
   int length = 1;
 
+  /* Number of bytes consumed from `mbs` */
+  int bytes_consumed = 0;
+
   if (conversion_state.bytes[0]) {
     conversion_state.bytes[1] = mbs[0];
+    bytes_consumed = 1;
     length = 2;
   } else if (mb_cur_max == 2 && isleadbyte (mbs[0])) {
     conversion_state.bytes[0] = mbs[0];
@@ -81,9 +85,11 @@
     }
 
     conversion_state.bytes[1] = mbs[1];
+    bytes_consumed = 2;
     length = 2;
   } else {
     conversion_state.bytes[0] = mbs[0];
+    bytes_consumed = 1;
   }
 
   /* Store terminating '\0' */
@@ -116,7 +122,7 @@
     *state = 0;
   }
 
-  return length;
+  return bytes_consumed;
 
 eilseq:
   errno = EILSEQ;

mingw-w64-crt/testcases/t_mbrlen.c

diff --git a/mingw-w64-crt/testcases/t_mbrlen.c b/mingw-w64-crt/testcases/t_mbrlen.c
index 4d297e3..2e723a1 100644
--- a/mingw-w64-crt/testcases/t_mbrlen.c
+++ b/mingw-w64-crt/testcases/t_mbrlen.c
@@ -118,12 +118,9 @@
 
   /**
    * Complete multibyte character
-   *
-   * NOTE: return value does not conform to ISO C and POSIX.
-   * This behavior is implemented for consistency with CRT.
    */
 
-  assert (mbrlen ((char *) Multibyte + 1, 1, &state) == 2);
+  assert (mbrlen ((char *) Multibyte + 1, 1, &state) == 1);
   assert (mbsinit (&state));
   assert (errno == 0);
 

mingw-w64-crt/testcases/t_mbrtowc.c

diff --git a/mingw-w64-crt/testcases/t_mbrtowc.c b/mingw-w64-crt/testcases/t_mbrtowc.c
index 1b94207..cb11b14 100644
--- a/mingw-w64-crt/testcases/t_mbrtowc.c
+++ b/mingw-w64-crt/testcases/t_mbrtowc.c
@@ -136,13 +136,10 @@
 
   /**
    * Complete multibyte character
-   *
-   * NOTE: return value does not conform to ISO C and POSIX.
-   * This behavior is implemented for consistency with CRT.
    */
   wc = WEOF;
 
-  assert (mbrtowc (&wc, (char *) Multibyte + 1, 1, &state) == 2);
+  assert (mbrtowc (&wc, (char *) Multibyte + 1, 1, &state) == 1);
   assert (wc != WEOF);
   assert (mbsinit (&state));
   assert (errno == 0);