Merge tulin@bk-internal.mysql.com:/home/bk/mysql-4.1

into poseidon.ndb.mysql.com:/home/tomas/mysql-4.1

mysql-test/r/ctype_sjis.result

@@ -41,3 +41,9 @@ C0C1C2C3C4C5C6C7C8C9CACBCCCDCECF
 select hex(CONVERT(@utf84 USING sjis));
 hex(CONVERT(@utf84 USING sjis))
 D0D1D2D3D4D5D6D7D8D9DADBDCDDDEDF
+create table t1 (a char(10) character set sjis);
+insert into t1 values (0x878A);
+select hex(a) from t1;
+hex(a)
+878A
+drop table t1;

mysql-test/t/ctype_sjis.test

@@ -32,3 +32,12 @@ select hex(CONVERT(@utf81 USING sjis));
 select hex(CONVERT(@utf82 USING sjis));
 select hex(CONVERT(@utf83 USING sjis));
 select hex(CONVERT(@utf84 USING sjis));
+
+#
+# Allow to insert extra CP932 characters
+# into a SJIS column
+#
+create table t1 (a char(10) character set sjis);
+insert into t1 values (0x878A);
+select hex(a) from t1;
+drop table t1;

strings/ctype-sjis.c

@@ -4563,6 +4563,40 @@ uint my_numcells_sjis(CHARSET_INFO *cs __attribute__((unused)),
   return clen;
 }
 
+/*
+  Returns a well formed length of a SJIS string.
+  CP932 additional characters are also accepted.
+*/
+static
+uint my_well_formed_len_sjis(CHARSET_INFO *cs __attribute__((unused)),
+                             const char *b, const char *e, uint pos)
+{
+  const char *b0= b;
+  while (pos && b < e)
+  {
+    /*
+      Cast to int8 for extra safety.
+      "char" can be unsigned by default
+      on some platforms.
+    */
+    if (((int8)b[0]) >= 0)
+    {
+      /* Single byte character */
+      b+= 1;
+    }
+    else  if (issjishead((uchar)*b) && (e-b)>1 && issjistail((uchar)b[1]))
+    {
+      /* Double byte character */
+      b+= 2;
+    }
+    else
+    {
+      /* Wrong byte sequence */
+      break;
+    }
+  }
+  return b - b0;
+}
 
 
 static MY_COLLATION_HANDLER my_collation_ci_handler =
@@ -4586,7 +4620,7 @@ static MY_CHARSET_HANDLER my_charset_handler=
   mbcharlen_sjis,
   my_numchars_mb,
   my_charpos_mb,
-  my_well_formed_len_mb,
+  my_well_formed_len_sjis,
   my_lengthsp_8bit,
   my_numcells_sjis,
   my_mb_wc_sjis,	/* mb_wc */

strings/ctype-utf8.c

@@ -2127,7 +2127,12 @@ int my_strcasecmp_utf8(CHARSET_INFO *cs, const char *s, const char *t)
   {
     my_wc_t s_wc,t_wc;
     
-    if (s[0] >= 0)
+    /*
+      Cast to int8 for extra safety.
+      char can be unsigned by default
+      on some platforms.
+    */
+    if (((int8)s[0]) >= 0)
     {
       /* 
         s[0] is between 0 and 127.