Chinese Characters in IE11, Edge and Windows 10

I recently worked on an issue where all Windows 10 users were seeing two strange display issues on certain websites via Internet Explorer 11 and Edge. There were two noticeable symptoms:

  • Chinese characters would show in particular locations on many websites. These were often buttons, but sometimes other symbols.
  • Buttons would be completely blank. The buttons themselves worked, which you could either use if they had a graphical representation of the button still, or you knew where to click.

This was even presenting itself in Office 365 – I couldn’t see the Notifications, Settings or Help buttons, and they would instead show as blank boxes.

This was found while piloting Windows 10 from Windows 7. The visible options in Internet Explorer seemed identical. and other browsers weren’t affected – Chrome could display these sites perfectly fine.

I worked out what the problem and fix was (jump to the end if you want that now), but here’s the story on how we got to this broken state:

As part of prepping for Windows 10, I followed Microsoft’s Security Baseline documentation which contains a handy Excel spreadsheet, with recommendations on what Group Policy settings you should use for best security practises. I followed this (I’ve linked to a newer version) and made choices based on understanding each option, and what worked for us. There were very few settings I didn’t follow exactly.

One of these settings was ‘Untrusted Font Blocking‘. The document recommended enabling this, to stop untrusted fonts being used as they’re a security risk – the loading of a font can allow elevated privileges, and has been used before. Made sense to me, so I enabled it.

This is what Group Policy says about Untrusted Font Blocking:

This security feature provides a global setting to prevent programs from loading untrusted fonts. Untrusted fonts are any font installed outside of the %windir%\Fonts directory. This feature can be configured to be in 3 modes: On, Off, and Audit. By default, it is Off and no fonts are blocked. If you aren’t quite ready to deploy this feature into your organization, you can run it in Audit mode to see if blocking untrusted fonts causes any usability or compatibility issues.

Eventually with a lot of testing and googling, I tried disabling this option – and it worked. Once you know the fix to a problem, it’s really easy to work backwards to find out more about it.

It turns out that in simple terms, websites can present their own fonts to use. It may be easier to present an arrow that’s from a font, rather than making a graphic of a font. Usually the site will load the font on the fly, but blocking that means the site fails back to a ‘best match’ on the font, which seems to be a font for Chinese characters, or a font that has a blank character for the matched result. Makes sense.

Microsoft changed their mind on this recommendation, only a month ago from time of writing. That recommendation change is worth reading, as it explained why they did it, and why they’re now changing their mind. The good news is that you’re not losing security by abandoning this setting, as the way fonts are parsed has changed from kernel to sandboxed user mode.

TL;DR version:

Turn off Untrusted Font Blocking through either of these methods:

Group Policy – Disable or change to Not Configured: Computer Configuration > Policies > Administrative Templates > System > Mitigation Options > Untrusted Font Blocking

Registry Setting – HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Kernel\ – QWORD MitigationOptions

  • To turn this feature on. Type 1000000000000.
  • To turn this feature off. Type 2000000000000.
  • To audit with this feature. Type 3000000000000.Important
    Your existing MitigationOptions values should be saved during your update. For example, if the current value is 1000, your updated value should be 1000000001000

Security Group Management Script

Over at eNow Consulting’s blog, I submitted an article and script on Exchange Group Management. It’s been working great for me, and hopefully will help others. I had a similar requirement around Security Groups, and this is the result.

The script itself is almost identical, but I wanted to share it anyway. I think it’s a great demonstration that you can really customise a script for whatever purpose you have. If you want to know how the script works generally, read my post at eNow, but there’s only one line different.

Instead of creating a “New Distribution Group”, it’s creating a New AD Group. The whole command is a bit different in syntax, but it’s still doing the same thing – creating a group. If you only wanted to manage existing groups, and removed the line altogether, you could manage both email and security groups from the single script (assuming a since csv file contains everything you want).

Here’s the script:

# Script to populate members of Security Groups
Start-Transcript -path C:\Scripts\Admin\Logs\securitygroups.txt
$data = import-csv C:\Scripts\Admin\securitygroups.csv
foreach ($group in $data){
New-ADGroup -name $group.GroupName -GroupCategory Security -GroupScope Universal -Path “OU=Security Groups,DC=mydomain,DC=com,DC=au” -Description “Automatically Managed by  @AdamFowler_IT’s Script”
$users = Get-ADUser -SearchBase “ou=Users,dc=mydomain,dc=com,dc=au” -Filter $group.filter
Get-ADGroup -Identity $group.groupname | Set-ADObject -clear member
Add-ADGroupMember -Identity $group.groupname -Members $users

Ideally, you should intelligently create security groups based on criteria around how the business functions. For example, the Finance department can have their own security group, if their department is Finance. Makes sense right?

The catch though, is to NOT link any actual security to this group. You don’t want 30 different things (e.g. files, folders, sharepoint sites, anything you’d use a security group for) pointing to one group. What if the Finance folder needs to be accessed by the CEO of your company? You shouldn’t just add them to the group by adjusting the filter, because they’ll get access to the 29 OTHER things pointed at this group.

The way around this is to have a security group for every single separate thing you apply security to. Have a Finance drive? Then create an AD security group with a descriptive name, and then add the original Finance security group as a member. This way, if someone joins or leaves the Finance team, security will automatically apply. On top of that, if you need to give the CEO access to the Finance drive by this secondary group, knowing you’re only giving them access to that one thing.

One to one relationships on a security group and what it applies to, will make managing it in the future much easier. You could extend this even further, and have a security group for each job function – this would mean there is a CEO security group that contains the CEO, and you can then add that security group to anything they need. The biggest benefit of this is when your CEO quits and another one comes along, you can just add the new CEO to the CEO group and they’ll get the same access. Not sure what access the CEO gets? Check what the CEO security group is a member of, and all your smartly named security groups will be listed.

My last tip around security groups is to note down who’s in charge of the group in either the notes or description field. If a query comes up a year later, you may not remember who originally asked for the security. Having a person or a job title listed means you can quickly get approval for making membership changes to the group.

Thinking about how you’re going to manage things in the future and planning around it might be a bit more painful at the time, but it really pays off in the end.