Build Identifier: http://hg.mozilla.org/mozilla-central/rev/d0edf8086809 Mozilla/5.0 (Windows NT 6.1; WOW64; rv:25.0) Gecko/20100101 Firefox/25.0 ID:20130803030205 Steps To reproduce: 1. Open http://ybb.softbank.jp/support/connect/adsl/pdf/4g_set01.pdf Actual results: All font display as rectangle Expected results: Font should display properly Regression window(m-c) Good: http://hg.mozilla.org/mozilla-central/rev/0c1663454e49 Mozilla/5.0 (Windows NT 6.1; WOW64; rv:24.0) Gecko/20130517 Firefox/24.0 ID:20130517123846 Bad: http://hg.mozilla.org/mozilla-central/rev/ecdfb8bb501e Mozilla/5.0 (Windows NT 6.1; WOW64; rv:24.0) Gecko/20130517 Firefox/24.0 ID:20130517143042 Pushlog: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=0c1663454e49&tochange=ecdfb8bb501e Regressed by: aaf5d70ba693 Ryan VanderMeulen — Bug 871530 - Update pdf.js to version 0.8.169. r=dtownsend

Can you tell us what version of DirectWrite you have (in about:support)?

DirectWrite Enabled true (6.1.7601.18126)

Needinfo on :bdahl with the next steps here.:bdahl do you need any additional QA help to help with reproduction here or is the issue understood ?

It would be helpful if someone with with DirectWrite 6.1 could run git bisect on pdf.js and find the true regression commit. All of my test machines have autoupgraded to DirectWrite 6.2 with IE10. Bisecting pdf.js is pretty easy as it only requires refreshing a page, if you someone needs help stop by #pdfjs.

Adding qawanted and needinfo'ing :tracy so QA team can help with comment #5

pinged bdahl in irc. This requires physical hardware. All I have access to is VM for Win7. Alice, this is a big ask, but might you be able help with the git bisect per comment #5?

Mihaela, might you or someone from SV be able to look into finding an accurate regression range? Rquires a Win 7 or Vista machine with DirectWrite 6.1 on it.

(In reply to Tracy Walker [:tracy] from comment #7) > pinged bdahl in irc. This requires physical hardware. All I have access to > is VM for Win7. Alice, this is a big ask, but might you be able help with > the git bisect per comment #5? Ok, I will try. but I am not familiar with git.

I made an attempt at this but my Windows 7 system already has DirectWrite 6.2 and does not reproduce this. I tried to restore to a few different restore points but each time it failed with some extremely vague Windows OS error. I've managed to find a Windows Vista ISO but this failed to transfer to a USB drive using Microsoft's own tool so I'm not left trying to burn this to a DVD. Unfortunately I have run out of time to work on this today. I can resume tomorrow if we've made no progress on this. FWIW, is it worth trying to fix this for an old version of DirectWrite vs advising users to update their systems?

I've been fighting with Vista for a day now and still haven't managed to get it installed. Alice, any help you can give here in bisecting the range would be greatly appreciated.

Today I managed to get a clean copy of Windows Vista installed and am currently installing some required software updates, Service Pack 2 in particular. Hopefully I can be in a position to test this sometime tomorrow.

I could reproduce the issue on Windows Vista x86 (DirectWrite 7) using latest Nightly, Aurora and Fx 24 beta 6, but not reproduced on Win 7 x64 with DirectWrite 6.2. (20130829030201) Mozilla/5.0 (Windows NT 6.0; rv:26.0) Gecko/20100101 Firefox/26.0 (20130826004002) Mozilla/5.0 (Windows NT 6.0; rv:25.0) Gecko/20100101 Firefox/25.0 (20130826142034) Mozilla/5.0 (Windows NT 6.0; rv:24.0) Gecko/20100101 Firefox/24.0 DirectWrite Enabled true (7.0.6002.23097) I cloned pdfjs, but encountered errors on viewing localhost:8000/web/viewer.html page. I'll continue investigating this tomorrow, if still necessary with this version of DirectWrite .

After installing Vista and all mandatory updates I now have DirectWrite 7.0.6002.18409 activated in Firefox. Firefox 23.0.1 does not reproduce this bug but Firefox 24.0 Beta does. It would seem, at least on Windows Vista that the version of DirectWrite does not matter.

Results from git bisect: ------------------------ e128f8bb472bb81b21c13240bea28cf9eb1a354e is the first bad commit commit e128f8bb472bb81b21c13240bea28cf9eb1a354e Author: Brendan Dahl <brendan.dahl@gmail.com> Date: Wed May 1 16:42:25 2013 -0700 Fix glyph selection for CID fonts that don't actually have CID font files. :040000 040000 d795b5e9789a4cc342c800271535b838233f6fda f8033ce241127f5fc860d2759ea56a06af8d38f7 M src :040000 040000 d765d79ab017164be98ecb7ed98f83e40a988c33 5767ec1bd02b58f5deed208159f40b5a739a1818 M test

needinfo'ing :bdahl here , to help understand if we can perform a backout(if low risk) for our second last beta ? Any idea how common this issue may be in the wild , and a possible general use case to understand the wider user impact this bug may have ? Would anyone using this D2D version on windows be affected here ?

(In reply to bhavana bajaj [:bajaj] from comment #17) > Would anyone using this D2D version on windows be affected here? Anecdotally, *any* version of DirectWrite on Windows Vista seems to be affected whereas only DirectWrite <6.2 on Windows 7+ seems to be affected. While advising Windows 7+ users to update their DirectWrite version (7.0 is available on Windows Update) this will not resolve the issue for Windows Vista users (see comment 14).

(In reply to bhavana bajaj [:bajaj] from comment #17) > needinfo'ing :bdahl here , to help understand if we can perform a backout(if > low risk) for our second last beta ? A backout would cause regressions on at least two PDFs that I know of for all platforms. I should have access to a vista machine tomorrow so I can try to get a fix for DirectWrite 6.1, but I wouldn't want to put this in beta at this point.

Fixed by Bug 947260

Marking this WONTFIX for Firefox 27 and earlier since it's unlikely bug 947260 will get uplifted.

Verified fixed based on comment 20.