[WIP] Windows Vista Extended Kernel

[asdf2345]

On 7/28/2020 at 5:21 AM, WindowsVista said:

I opened up the firefox.exe file in  CFF Explorer and set the entry point in ntdll.dll from RtlQueryPerformanceCounter to NtQueryInformationProcess

There's your fix

[WindowsVista]

I modified shell32.dll since I decided to change some icons however, I think it removed certain functions since Firefox kept crashing afterwards. I re applied the original extended kernel shell32.dll and it worked perfectly fine just llike before. I'm definitely sure that I modified the extended kernel shell32. I don't know the cause of this. I used resource hacker to modify the dll file to replace icons and manually change the file copy animation color.

[asdf2345]

Was playing around with Chromium versions to see what the newest version that allows me to install addons is

52 and 53 work, but I can't install addons, 54-62 doesn't work at all, 58 throws a DEP error, and anything 63 through 72 work, but addons don't install.

 

Another thing I discovered is that @greenhillmaniac's 15.101 driver works without any dll replacing, so this might help with making newer AMD drivers work. Will try 15.7 soon

[win32]

32 minutes ago, asdf2345 said:

Another thing I discovered is that @greenhillmaniac's 15.101 driver works without any dll replacing, so this might help with making newer AMD drivers work. Will try 15.7 soon

That's because of those user32 functions I added. But I haven't touched powerprof.dll, which apparently caused problems as well.

15.7.1 misses ntoskrnl function sprintf_s. And the 16.x series adds RtlDowncaseUnicodeChar.

3 hours ago, WindowsVista said:

I modified shell32.dll since I decided to change some icons however, I think it removed certain functions since Firefox kept crashing afterwards. I re applied the original extended kernel shell32.dll and it worked perfectly fine just llike before. I'm definitely sure that I modified the extended kernel shell32. I don't know the cause of this. I used resource hacker to modify the dll file to replace icons and manually change the file copy animation color.

That may be because I put my new code and export table below the rsrc/reloc sections, which doesn't correspond to MS convention, yet seems to the easiest way to do it on x64. I'm gonna have to check out whatever reshack does to these files myself.

[asdf2345]

On 8/18/2020 at 11:46 PM, win32 said:

That's because of those user32 functions I added. But I haven't touched powerprof.dll, which apparently caused problems as well.

15.7.1 misses ntoskrnl function sprintf_s. And the 16.x series adds RtlDowncaseUnicodeChar.

That may be because I put my new code and export table below the rsrc/reloc sections, which doesn't correspond to MS convention, yet seems to the easiest way to do it on x64. I'm gonna have to check out whatever reshack does to these files myself.

Powerprof appears to only cause issues with CCC

Looks to be this missing function that the 15.6 CCC uses, but it does open fine after the error, so idk

 

On 8/19/2020 at 12:23 AM, asdf2345 said:

Powerprof appears to only cause issues with CCC

Looks to be this missing function that the 15.6 CCC uses, but it does open fine after the error, so idk

CCC from 15.6, and @greenhillmaniac's 15.101 driver

[WindowsVista]

Well I've found the string

I'm a little stuck at this point since I haven't had a lot of experience

This is to do with the windows vista boot animation color BTW.

 

[win32]

That's the right string, but you will have to look at the three green colours, 01 65 53, 00 92 89 and 04 DC 8E. In RGB colour values, those are 083 101 001, 137 146 000 and 142 220 004 respectively.

So look for the RGB values for the shades of pink you want to use, reverse them and convert to hex. Then replace those three values in the string.

You should probably copy your ntoskrnl and give it a different name like ntkrnlmp.exe.

And then

bcdedit /copy {current} /d "Windows Vista Custom Boot Screen"

And then remember the long number/ GUID value that you get from this message:

"The entry was successfully copied to {01234567-89ab-cdef-00ff-fff000ffffff}" (for example)

so you can do this:

bcdedit /set {01234567-89ab-cdef-00ff-fff000ffffff} kernel ntkrnlmp.exe

And then there is the code signing enforcement that Vista and up introduces. I made a cracked ntoskrnl for that purpose, but you may also want to do this:

bcdedit /set {01234567-89ab-cdef-00ff-fff000ffffff} nointegritychecks 1

 

Edited August 19, 2020 by win32

[asdf2345]

Rust is going to require SHA-2 in Oct 2020, so you'll need to see about making an extended kernel that has SHA-2 capability

[tamarindojuice]

--

Edited December 7, 2020 by tamarindojuice

[win32]

On 8/20/2020 at 8:12 PM, asdf2345 said:

Rust is going to require SHA-2 in Oct 2020, so you'll need to see about making an extended kernel that has SHA-2 capability

Wouldn't the OS-level SHA-2 support be sufficient? And which Rust are you talking about? The programming language upon which newer Firefoxes are built, or the game?

1 minute ago, tamarindojuice said:

Do you think that Haswell support fixing that nasty boot and service intialization fail bugs is something coming on the way?

I would LOVE to run Vista on my rig but it's unfortunatelly an I7 4790k

Not asking for ETA or anything else but just out of curiosity, theres anything else that holds you from getting this working right now?

And... Do you think that kind of fix will come to your EX Kernel anytime soon?

I'm trying. I'd really love to have Vista running stable on my Kaby Lake laptop as well. 2012R2 looks like Windows 1, and having Vista's UI would be a major improvement (I'm also going to work on getting graphics drivers like my Intel HD 620 ones working too).

I had previously narrowed it down to the function LoadLibraryExW, and I recently installed Windows 7 build 6519 which has a very similar version of the function, and also doesn't have those bugs.

But simply transplanting the function failed, as it resulted in a BSOD where "the system couldn't find the file specified (@98w9asoise590dj.dll [or similar nonsense file name])". But after trying a few more things, I think I've narrowed down the problems to one particular part of the function. But making a good patch will be very challenging.

What also doesn't help is that this laptop has broken hinges, so I can't bring it to where I am half of the time.

[asdf2345]

43 minutes ago, win32 said:

Wouldn't the OS-level SHA-2 support be sufficient? And which Rust are you talking about? The programming language upon which newer Firefoxes are built, or the game?

I'm talking about the game. Well, it's more of the anti-cheat used. Easy Anti-Cheat is moving to SHA-2 for some reason.

On my main Windows 7 PC, I get this error message instead, telling me to download the SHA-2 update from 2019, which I'm assuming is the same KB number as the one that makes Vista 6.0.6003.

Edited August 23, 2020 by asdf2345

[win32]

15 minutes ago, asdf2345 said:

On my main Windows 7 PC, I get this error message instead, telling me to download the SHA-2 update from 2019, which I'm assuming is the same KB number as the one that makes Vista 6.0.6003.

So it appears that all of the files that are modified to accommodate SHA-2 support are not part of the extended kernel, except for winload.exe (which is just a minor patch to allow for the patched ntoskrnl to work). So the extended kernel should not interfere with SHA-2. I believe that some extended kernel users are installing new SHA-2 signed updates.

Edited August 23, 2020 by win32

[tamarindojuice]

--

 

Edited December 7, 2020 by tamarindojuice

[ilwalaihr]

11 hours ago, win32 said:

I'm trying. I'd really love to have Vista running stable on my Kaby Lake laptop as well. 2012R2 looks like Windows 1, and having Vista's UI would be a major improvement (I'm also going to work on getting graphics drivers like my Intel HD 620 ones working too).

I had previously narrowed it down to the function LoadLibraryExW, and I recently installed Windows 7 build 6519 which has a very similar version of the function, and also doesn't have those bugs.

But simply transplanting the function failed, as it resulted in a BSOD where "the system couldn't find the file specified (@98w9asoise590dj.dll [or similar nonsense file name])". But after trying a few more things, I think I've narrowed down the problems to one particular part of the function. But making a good patch will be very challenging.

What also doesn't help is that this laptop has broken hinges, so I can't bring it to where I am half of the time.

I got my Ryzen PC to work, thanks to some help on this forum, but I'm now trying to get the Extended Kernel to work, there are some things that I don't understand but that I have to do though.

I'm on Build 0.6.6002 by the way.

There are no 000BCF** Offsets here.

Edited August 23, 2020 by ilwalaihr

[asdf2345]

I think 99% of us are using the dlls given to us by @win32, so just ask.

Also, about Ryzen, what BIOS is your board running?