Nominating http://codereview.chromium.org/13163 
for the branch. Sorry for yet another last minute nomination.

This is a 'two character' fix and is more than absolutely safe, but gives some good
PR to a small number of i18n-minded people :-)

Fixed on the release branch with r6467.

This will be in 154.32 or later.

I'll leave this open in case it still has not landed on trunk. Since the release 
branch is up-to-date, I'm moving it off the mstone:1.0 list.

Sorry for my late response.
I have committed this fix to trunk as r6500, and verified it fixes this problem.

The WebCore::FillNonBMPGlyphs() function (in 
"src/webkit/port/platform/graphics/GlyphPageTreeNodeWin.cpp") creates a 
mapping from non-bmp characters to its glyphs and it tries to retrieve the 
glyph of its i-th character with the 
WebCore::UniscribeHelper::FirstGlyphForCharacter() calls. Unfortunately, 
because of a mistake in its first argument, the function returns invalid 
glyphs for non-BMP characters. 

<b>What steps will reproduce the problem?</b>
1. Retrieve a copy of the code2001 font from http://www.code2000.net/ and 
install it;
2. Display the attached HTML file with Chrome. 

What is the expected output?
No duplicate entries in the table.

What do you see instead?
Chrome uses the same glyph for both '&amp;#x10000;' and '&amp;#x10001;', for both  
'&amp;#x10002;' and '&amp;#x10003;', etc.

<b>Please use labels and text to provide additional information.</b>