[vbox-dev] Updating VirtualBox consumption of IMAGE_LOAD_CONFIG_DIRECTORY*
jkunkee at microsoft.com
Fri Nov 9 05:25:33 UTC 2018
That is the puzzle piece I was missing. The app compat team just confirmed that the fix works; we'll look forward to the 5.2.22 release for it to be available for customers.
Sent from Outlook<http://aka.ms/weboutlook>
From: vbox-dev-bounces at virtualbox.org <vbox-dev-bounces at virtualbox.org> on behalf of Klaus Espenlaub <klaus.espenlaub at oracle.com>
Sent: Thursday, November 8, 2018 6:53:48 AM
To: vbox-dev at virtualbox.org
Subject: Re: [vbox-dev] Updating VirtualBox consumption of IMAGE_LOAD_CONFIG_DIRECTORY*
the 5.2 test builds are always "not quite next 5.2.x release" builds, so if they have the working fix then it will automatically end up in the next release. In this case it will be 5.2.22. Of course the same fix will be also in 6.0, but that's further in the future since we just started beta testing.
On 08.11.2018 01:37, Jon Kunkee wrote:
I appreciate the pointer to testing builds; we’re doing some internal validation using them and will post back when we know the fix is good.
As we’re preparing for the upcoming Windows release, we’d like to make sure customers have more than just test builds to use; is there any idea when this will be in an actual release? Would it make sense to expect this to land in a 5.2 update, or is this only going to land in 6.0?
From: Jon Kunkee
Sent: Thursday, September 20, 2018 4:30 PM
To: vbox-dev at virtualbox.org<mailto:vbox-dev at virtualbox.org>
Cc: Divya Bhat <divyabh at microsoft.com><mailto:divyabh at microsoft.com>
Subject: Updating VirtualBox consumption of IMAGE_LOAD_CONFIG_DIRECTORY*
I’m a software engineer on in the Windows group at Microsoft. My team recently added a pointer-sized field to IMAGE_LOAD_CONFIG_DIRECTORY32 and IMAGE_LOAD_CONFIG_DIRECTORY64 that is non-NULL on 32-bit x86 Windows 10 systems in the current Insider Preview.
It seems that VirtualBox does some extensive validation of this structure, so it now exits with an error when it sees this field in use.
I filed a ticket about it:
If I grokked the bug filing page right, it’s proper to ask about it here too.
Adjusting the VirtualBox validation to handle a new field in this struct has been done before, so hopefully it’s easy to do again. (I linked two of the previous tickets in the History section.)
In the longer term, is there a way for the validation to not break so easily?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the vbox-dev