libroot: Fix ICU locale implementation of wcsxfrm.

* The "size" parameter is the size of "out" not "in", and the
   return size_t parameter is supposed to always have the total amount
   of wchar_ts needed, not how many are actually used.
 * In the case where "outSize == 0", we set "requiredSize" and then
   return.

Fixes crashes seen in glib2 Unicode collation routines, which
are used in GTK file dialogs.

Thanks to PulkoMandy for glancing at this.

headers/private/libroot/locale/ICUCollateData.h

@@ -33,7 +33,7 @@ public:
 			status_t			Wcscoll(const wchar_t* a, const wchar_t* b,
 									int& out);
 			status_t			Wcsxfrm(wchar_t* out, const wchar_t* in,
-									size_t size, size_t& outSize);
+									size_t outSize, size_t& requiredSize);
 
 private:
 			status_t			_ToUnicodeString(const char* in,

headers/private/libroot/locale/ICULocaleBackend.h

@@ -64,7 +64,7 @@ public:
 	virtual	status_t			Wcscoll(const wchar_t* a, const wchar_t* b,
 									int& out);
 	virtual status_t			Wcsxfrm(wchar_t* out, const wchar_t* in,
-									size_t size, size_t& outSize);
+									size_t outSize, size_t& requiredSize);
 
 	virtual status_t			TZSet(const char* timeZoneID, const char* tz);
 	virtual	status_t			Localtime(const time_t* inTime,

headers/private/libroot/locale/LocaleBackend.h

@@ -147,7 +147,7 @@ public:
 	virtual	status_t			Wcscoll(const wchar_t* a, const wchar_t* b,
 									int& out) = 0;
 	virtual status_t			Wcsxfrm(wchar_t* out, const wchar_t* in,
-									size_t size, size_t& outSize) = 0;
+									size_t outSize, size_t& requiredSize) = 0;
 
 	virtual status_t			TZSet(const char* timeZoneID,
 									const char* tz) = 0;

src/system/libroot/add-ons/icu/ICUCollateData.cpp

@@ -174,18 +174,18 @@ ICUCollateData::Wcscoll(const wchar_t* a, const wchar_t* b, int& result)
 
 
 status_t
-ICUCollateData::Wcsxfrm(wchar_t* out, const wchar_t* in, size_t size,
-	size_t& outSize)
+ICUCollateData::Wcsxfrm(wchar_t* out, const wchar_t* in, size_t outSize,
+	size_t& requiredSize)
 {
 	if (in == NULL) {
-		outSize = 0;
+		requiredSize = 0;
 		return B_OK;
 	}
 
 	if (fCollator == NULL || strcmp(fPosixLocaleName, "POSIX") == 0) {
 		// handle POSIX here as the collator ICU uses for that (english) is
 		// incompatible in too many ways
-		outSize = wcslcpy(out, in, size);
+		requiredSize = wcslcpy(out, in, outSize);
 		for (const wchar_t* inIter = in; *inIter != 0; ++inIter) {
 			if (*inIter > 127)
 				return B_BAD_VALUE;
@@ -194,10 +194,13 @@ ICUCollateData::Wcsxfrm(wchar_t* out, const wchar_t* in, size_t size,
 	}
 
 	UnicodeString unicodeIn = UnicodeString::fromUTF32((UChar32*)in, -1);
-	size_t requiredSize = fCollator->getSortKey(unicodeIn, NULL, 0);
+	requiredSize = fCollator->getSortKey(unicodeIn, NULL, 0);
+
+	if (outSize == 0)
+		return B_OK;
 
 	uint8_t* buffer = (uint8_t*)out;
-	outSize = fCollator->getSortKey(unicodeIn, buffer, requiredSize);
+	fCollator->getSortKey(unicodeIn, buffer, outSize);
 
 	// convert 1-byte characters to 4-byte wide characters:
 	for (size_t i = 0; i < outSize; ++i)

src/system/libroot/add-ons/icu/ICULocaleBackend.cpp

@@ -308,12 +308,12 @@ ICULocaleBackend::Wcscoll(const wchar_t* a, const wchar_t* b, int& result)
 
 
 status_t
-ICULocaleBackend::Wcsxfrm(wchar_t* out, const wchar_t* in, size_t size,
-	size_t& outSize)
+ICULocaleBackend::Wcsxfrm(wchar_t* out, const wchar_t* in, size_t outSize,
+	size_t& requiredSize)
 {
 	ErrnoMaintainer errnoMaintainer;
 
-	return fCollateData.Wcsxfrm(out, in, size, outSize);
+	return fCollateData.Wcsxfrm(out, in, outSize, requiredSize);
 }