Hi,
I see a regression test failure in my mingw-vista port
when I invoke the command
make check MULTIBYTE=euc_jp NO_LOCALE=yes
.
It causes a crash at tsearch.
The crash seems to occur when the server encoding isn't
UTF-8 with no locale.
The attached is a patch to avoid the crash.
regards,
Hiroshi Inoue
Index: backend/utils/mb/mbutils.c
===================================================================
RCS file: /projects/cvsroot/pgsql/src/backend/utils/mb/mbutils.c,v
retrieving revision 1.78
diff -c -r1.78 mbutils.c
*** backend/utils/mb/mbutils.c 22 Jan 2009 10:09:48 -0000 1.78
--- backend/utils/mb/mbutils.c 17 Feb 2009 21:59:26 -0000
***************
*** 575,580 ****
--- 575,584 ----
wchar2char(char *to, const wchar_t *from, size_t tolen)
{
size_t result;
+ #ifdef WIN32
+ int encoding = GetDatabaseEncoding();
+ bool useWcstombs = !(encoding == PG_UTF8 || lc_ctype_is_c());
+ #endif
if (tolen == 0)
return 0;
***************
*** 584,602 ****
* On Windows, the "Unicode" locales assume UTF16 not UTF8 encoding,
* and for some reason mbstowcs and wcstombs won't do this for us,
* so we use MultiByteToWideChar().
*/
! if (GetDatabaseEncoding() == PG_UTF8)
{
! result = WideCharToMultiByte(CP_UTF8, 0, from, -1, to, tolen,
NULL, NULL);
/* A zero return is failure */
! if (result <= 0)
result = -1;
else
{
- Assert(result <= tolen);
/* Microsoft counts the zero terminator in the result */
! result--;
}
}
else
--- 588,624 ----
* On Windows, the "Unicode" locales assume UTF16 not UTF8 encoding,
* and for some reason mbstowcs and wcstombs won't do this for us,
* so we use MultiByteToWideChar().
+ * Also note wcstombs/mbstowcs is unavailable when LC_CTYPE is C.
*/
! if (!useWcstombs)
{
! int utf8len = tolen;
! char *utf8str = to;
!
! if (encoding != PG_UTF8)
! {
! utf8len = pg_encoding_max_length(PG_UTF8) * tolen;
! utf8str = palloc(utf8len + 1);
! }
! utf8len = WideCharToMultiByte(CP_UTF8, 0, from, -1, utf8str,
utf8len,
NULL, NULL);
/* A zero return is failure */
! if (utf8len <= 0)
result = -1;
else
{
/* Microsoft counts the zero terminator in the result */
! result = utf8len - 1;
! if (encoding != PG_UTF8)
! {
! char *mbstr =
pg_do_encoding_conversion((unsigned char *) utf8str, result, PG_UTF8, encoding);
! result = strlcpy(to, mbstr, tolen);
! if (utf8str != to)
! pfree(utf8str);
! if (mbstr != utf8str)
! pfree(mbstr);
! }
! Assert(result <= tolen);
}
}
else
***************
*** 618,637 ****
char2wchar(wchar_t *to, size_t tolen, const char *from, size_t fromlen)
{
size_t result;
if (tolen == 0)
return 0;
#ifdef WIN32
! /* See WIN32 "Unicode" comment above */
! if (GetDatabaseEncoding() == PG_UTF8)
{
/* Win32 API does not work for zero-length input */
! if (fromlen == 0)
result = 0;
else
{
! result = MultiByteToWideChar(CP_UTF8, 0, from, fromlen,
to, tolen - 1);
/* A zero return is failure */
if (result == 0)
result = -1;
--- 640,672 ----
char2wchar(wchar_t *to, size_t tolen, const char *from, size_t fromlen)
{
size_t result;
+ #ifdef WIN32
+ int encoding = GetDatabaseEncoding();
+ bool useMbstowcs = !(encoding == PG_UTF8 || lc_ctype_is_c());
+ #endif
if (tolen == 0)
return 0;
#ifdef WIN32
! if (!useMbstowcs)
{
+ int utf8len = fromlen;
+ unsigned char *utf8str = (unsigned char *) from;
+
+ if (encoding != PG_UTF8)
+ {
+ utf8str = pg_do_encoding_conversion(from, fromlen,
encoding, PG_UTF8);
+ if (utf8str != from)
+ utf8len = strlen(utf8str);
+ }
+ /* See WIN32 "Unicode" comment above */
/* Win32 API does not work for zero-length input */
! if (utf8len == 0)
result = 0;
else
{
! result = MultiByteToWideChar(CP_UTF8, 0, utf8str,
utf8len, to, tolen - 1);
/* A zero return is failure */
if (result == 0)
result = -1;
***************
*** 643,648 ****
--- 678,685 ----
/* Append trailing null wchar (MultiByteToWideChar()
does not) */
to[result] = 0;
}
+ if (utf8str != from)
+ pfree(utf8str);
}
else
#endif /* WIN32 */
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
