Description:
------------
mbstring functions seem to use old Unicode data, so the functions are not able to cope with new characters introduced to Unicode lately (e.g. mb_strtoupper ignores lower-case characters because it does not know them).

AFAICT (and I am no PHP expert, mind you), what is needed to be done is to update ext/mbstring/unicode_data.h (possibly just regenerating the file, as it says “It was generated by a modified version of ucgendat, part of the ucdata-2.5 package”).

Test script:
---------------
function test($str)
{
	$upper = mb_strtoupper($str, 'UTF-8');
	$len = strlen($upper);
	for ($i = 0; $i < $len; ++$i) echo dechex(ord($upper[$i])) . ' ';
	echo "\n";
}

// OK
test("\xF0\x90\x90\xB8");// U+10438 DESERET SMALL LETTER H (added in 3.1.0, March 2001)
// not OK
test("\xE2\xB0\xB0");	// U+2C30 GLAGOLITIC SMALL LETTER AZU (added in 4.1.0, March 2005)
test("\xD4\xA5");		// U+0525 CYRILLIC SMALL LETTER PE WITH DESCENDER (added in 5.2.0, October 2009)


Expected result:
----------------
f0 90 90 90
e2 b0 80
d4 a4

Actual result:
--------------
f0 90 90 90
e2 b0 b0
d4 a5