Damnation

this got ommitted from the main post by mistake I think patch.exe -Np1 -u -l -i sp1_to_sp3(ACP2).patch

@Mov AX, 0xDEAD I want to try something - Have acpi.sys treat ACPI0010 device as a CPU Processor instead of just a generic device since it contains all the CPUs within it. Do you know how we could do this with acpi.sys source code?

@DrunkenTanker Which sites are accessible for you?

@Dietmar There is an Micro-ATX version of this board that is a bit cheaper, but even if more affordable there is a chance that it does not have this ACPI0010 device in it's CPU SSDT ACPI table that the full sized version does.

@Dietmar Will you be able to get your own ASRock B550 steel Legend for testing?

This might be the same issue being described here - https://patchwork.kernel.org/project/qemu-devel/patch/1495446630-365062-2-git-send-email-eyakovlev@virtuozzo.com/

@Mov AX, 0xDEAD Could you write a command line test program to check if HLT or MWAIT CPU command is working correctly in XP?

@DrunkenTanker The work around right now is to use XP compatible USB 1gb LAN adapters.

@Dietmar IIRC It's netio.sys that is giving us the most trouble getting NDIS6 working right?

A fix for this would be very much appreciated.

@Dietmar Here is mine with modded intelppm.sys

@Mov AX, 0xDEAD Could you try making this ACPI0010 device in a QEMU VM and see if it affects CPU power state? (assuming QEMU can do something like that) here it is in my CPU SSDT ---

@Mov AX, 0xDEAD Are you sure? I was thinking perhaps the ACPI driver can't control the CPU power state since the processors are now all inside this container device now instead of on their own outside the container like they are with my other machines, including my threadripper which does not have this issue. I really think this new ACPI0010 device is the cause as it's the only thing that's different from my other AMD systems. --- I tried an intelppm.sys that @Dietmar sent me earlier but it did not make any difference, still 140W idle. I noticed that windows 7 and later has an amdppm.sys driver - maybe this needs to be ported to XP? --- But if you're sure intelppm.sys will resolve the issue can you send me a patched version? --- P.S I also tried manually installing processor onto the "ACPI0007" devices in device manager, but I got code 10: cannot start, so I reverted

@Dietmar I think my way could potentially work since you have some ability to use internal unexported ntoskrnl code which you can't do with the normal ntoskrnl extender. But it could be that this approach will not work as intended. @Mov AX, 0xDEAD still knows way more about this than I do. To build just the ntoskrnl extender you need only to use Windows 7 DDK in a Windows 10 VM After installing 7 DDK Open "Free Build Environment" from the start menu and navigate to C:\NTOSKRNL_Emu-master_21jul2022_ndis6 then type bcz to build extender. --- To make ntosxlibs: To make ntosxlibs you will need leaked XP/2K3 source code, an XP VM to built it in, and a mass find/replace tool like powergrep to append ntoskrnX onto the end of all exported ntoskrnl functions so you don't run into conflicts later on. But this is all very complicated to describe so 1st thing to do would be to get razzle setup and running and building everything in the base folder of the Vanilla Win2k3 source tree. Once that is working we can continue from there.

@Mov AX, 0xDEAD I see there is code for handling the "ACPI Processor Container" in the Windows 10 acpi.sys driver under AcpiProcessorContainerIrpDispatch can you make an implementation of this for XP acpi.sys please? I'd like my CPU to not consume so much power.

@Dietmar Here is my latest ntoskrnl extender source code https://ufile.io/ew7wz951 It's building those ntosxlibs that's the most challenging part for my version, you can use the prebuilt ones for now, but I can help you to build your own if you like.

@Dietmar I made use of both the ntoskrnl extender source code as well as the leaked XP/2K3 source code with my attempts to make this work. I'll upload my source code on here when I get the chance. To compile ntoskrnl extender I used a Windows 10 VM and the Windows 7 DDK For my customized internal ntos libraries I used and XP VM and the razzle build environment. I'll upload my source code on here when I get the chance.

@Dietmar The board I'm using that has this ACPI container device is an ASRock B550 Steel Legend. If you can get one of these boards for yourself can you test it?

@Dietmar Do any of your boards have this "ACPI Processor Container Device" on them?

@GD 2W10 I think @George King backported the Windows 8.0 x64 version, it'd be part of his XP2ESD project. you can ask him.

@Mov AX, 0xDEAD So my processors are in a new ACPI 6.0 container - linux talks about it here - https://patchwork.kernel.org/project/linux-acpi/patch/1449065446-26115-2-git-send-email-sudeep.holla@arm.com/ Can you add in support for this to the ACPI driver? edit: It's called "ACPI Processor Container Device" on Windows 11 https://ibb.co/gWw1Nm0 and as "Generic Bus" on XP

@Mov AX, 0xDEAD I compared my 5950X system with my threadripper, On threadripper the CPUs are installed but not on my 5950X. I'm quite sure that this is the issue, but I'm not sure how to get the CPUs installed in Device manager. Any idea how to install them?

@GD 2W10 I am quite that a Surface Pro 1 will be UEFI only. There is experimental UEFI support for XP x64, but it's nowhere near stable.

@WinWord2000 If you can get me a copy of the driver I can check if the .sys file has it's dependencies met by the ntoskrnl extender. If dependencies are met That'll give you basic sound output at least. The bundled software still won't work though.

@WinWord2000 I think there should be some USB Sound Cards that will work with XP.