Compiling ACPI v2.0 driver for Windows XP SP3 and Windows 2003 SP2 (x32/x64)

[Dietmar]

@Mov AX, 0xDEAD

Something strange happened:

I start this XP SP3 again on the MSI z690-A pro DDR4 and 12600 cpu with connected Windbg, but now only from breakpoint for acpi.sys driver entry (means without ed Kd_ACPI_Mask 0xFFFFFFFF).

And then hit in KD line just g.

Now, all 12 processors show the correct name 12600 in Device Manager.

So, your way for to fix the names works on this compi.

Also the message in registry for intelppm about INITSTARTFAILED is gone.

The cpus are now listed there on both sides the same as

0,1,10,11,2,3,4,5,6,7,8,9

Dietmar

Here is the wished reg entry, from now with working cpu names

https://ufile.io/hxdbepgg

 

Edited December 9, 2022 by Dietmar

[Dietmar]

@Mov AX, 0xDEAD

For to understand, what is going on,

I deleted all entries in registry services for intelppm.

Also I delete the entry in ENUM for Intel processors.

Voila, on next reboot with connected Windbg again only the cpu 1,2,3,4,5,6,7,10,11 have the correct entry in Device Manager 12600

and the 8,9 have the entry "Intel Processor".

After reboot with Windbg connected, again all processors are shown correct as 12600.

Here is the ENUM entry with the error "Intel Processor"

Dietmar

https://ufile.io/yg35vs8s

[Dietmar]

@Mov AX, 0xDEAD

And here is the ENUM entry for working 12600 entries for all cpu after 3 reboot with connected Windbg

Dietmar

https://ufile.io/fhd4s0pv

[Dietmar]

@Mov AX, 0xDEAD

When you compare genuine2 (error) with genuine3 (all ok)

you see, that the entry

"FriendlyName"="12th Gen Intel(R) Core(TM) i5-12600"

is missed for cpu 8 and 9 only in genuine2

Dietmar

[Mov AX, 0xDEAD]

49 minutes ago, Dietmar said:

@Mov AX, 0xDEAD

When you compare genuine2 (error) with genuine3 (all ok)

you see, that the entry

"FriendlyName"="12th Gen Intel(R) Core(TM) i5-12600"

is missed for cpu 8 and 9 only in genuine2

genuine2 (error) has many missed "ActiveService"="intelppm" on random cores

May be problem is out acpi.sys.....

Can you show how you patched SSDT to work with previous acpi.sys (where Processor(PR00, 0, 0x1810, 6) was inserted) ?
 

[Dietmar]

@Mov AX, 0xDEAD

First I try the patched SSDT from @Damnation .

But with new Bios, this gives always Bsod.

Infuscomus then put this patch into DSDT. I delete there all the now unnecessary entries for the 0007 device.

This works.

BUT: Via Benchmark tests and power consumption with this patched Bios, only the power cores where recogniced,

even in win10.

Here is the modded DSDT but your acpi.sys is better.

It looks, as now it only needs one reboot and all is ok with cpu's in Device Manager

Dietmar

https://ufile.io/nh6xuban

 

 

Edited December 9, 2022 by Dietmar

[Dietmar]

@Mov AX, 0xDEAD

Can you send me the acpi.sys FREE version from this new acpi.sys.

I will test

Dietmar

[Dietmar]

@Mov AX, 0xDEAD

I remember my tests with the acpi.sys from outerspace ) 6666.

With Ryzen cpu's, that only all cores are enumerated correct in Device Manager happens there

also sometimes only on second reboot. And all those Bios use the old processor definition.

So, this is not a special problem of the new processor definition via 0007 device.

This "not recognicing of all cpu's correct on first boot" happens at a different place

Dietmar

Edited December 9, 2022 by Dietmar

[UsefulAGKHelper]

3 hours ago, Dietmar said:

@Mov AX, 0xDEAD

When you compare genuine2 (error) with genuine3 (all ok)

you see, that the entry

"FriendlyName"="12th Gen Intel(R) Core(TM) i5-12600"

is missed for cpu 8 and 9 only in genuine2

Dietmar

"12th Gen Intel(R) Core(TM) i5-12600"

You managed to run Windows XP on newer processors than Coffee Lake?

I am amazed!

Edited December 9, 2022 by UsefulAGKHelper

[Mov AX, 0xDEAD]

12 hours ago, Dietmar said:

@Mov AX, 0xDEAD

First I try the patched SSDT from @Damnation .

But with new Bios, this gives always Bsod.

Infuscomus then put this patch into DSDT. I delete there all the now unnecessary entries for the 0007 device.

This works.

@Dietmar

1) Orig definition is PR00, PR01, ... , PR63

Modded is C001, C002, ... , C03F and not in same position as orig

Later DSDT has definition of _SB->PEPD->DEVY with link to _SB.PR00, _SB.PR01, ...

On modded bios, these links linked to nothing

2) Do you always get right names of all cores with modded bios and prev acpi.sys ?

3) Check PM for free version

[Dietmar]

@Mov AX, 0xDEAD

Yes, with this modded Bios always at once all cpu's (12900k, 12600, on 3 boards) are listed correct in Device Manager.

BUT: Only the power cores work, the E-cores are listed but just do nothing,

in XP SP3, win7 SP1, win8.1 all 32 bit version, and win10 (bit64).

I noticed the "notwork" of the E-cores from 12900k only via its power consumption (few Watt) against 25 Watt with all cores working at 0% load

and about 30% less cpu power in different benchmarks

Dietmar

PS: Thanks for free acpi.sys . I just set up a brandnew original XP SP3 with it and will soon report.

Edited December 10, 2022 by Dietmar

[Dietmar]

@Mov AX, 0xDEAD

Yesssa ,

with this free acpi.sys all cpu's are correct listed at once!

I make a test before last reboot of XP GUI install on other compi, loading registry HIVE system.

In ENUM for the Genuine Intel randomally some cores have friendly name and some not.

But all have service intelppm

Dietmar

Edited December 10, 2022 by Dietmar

[Damnation]

@Mov AX, 0xDEAD

Can I try this free acpi.sys on my system?

[XP4ever]

On 12/9/2022 at 1:11 PM, Dietmar said:

I have also an normal Lenovo x230 notebook.

On this notebook I make the same experience as you.

After I installed under XP SP3 the USB3 driver, it was impossible to work with it any longer.

And this behavior cant be reverted by deinstalling the USB2,3 driver.

Now I think, that the Intel chipset driver for USB2 and any USB3 driver are incompatible.

You have to make a new install of XP on this notebook without any chipset driver.

After this, use the AMD USB3 driver 145 from @daniel_k for this notebook,

Dietmar

Hi Dietmar,

Thanks for info, it really helps.

Now when I looked closer, it turns out that one of chipset support driver INF files installed has the following header:

; ****************************************************************************
; ****************************************************************************
; **    Filename:  PantUSB3.INF                                              
; **    Abstract:  Null driver for Intel(R) USB 3.0 devices
; **    Last Update: December 06, 2011 (Version 9.3.0 Build 1018)
; ****************************************************************************
; ****************************************************************************

huh, quite strange... looks like it intentionally has some USB3 stubs...

Anyways, I still believe there must be a way to remove that deriver without reinstalling entire system, as I did before for storage and other stuff when I relocated this SSD from eventually died X220 to this X230.

Another thing I'm concerning about - when I, say, broke somehow this USB driver, it requires specifically usbd.sys and usbhub.sys, may be it does not matter but I afraid it will not see AMD USB drivers anyway.

 

All, does anyone know the correct way to completely revert Intel INF Chipset Support installation?

[Damnation]

@Mov AX, 0xDEAD @Dietmar

Your most recent free build acpi.sys on my AMD system gives BSOD A5 0x11 0x8

I still have not yet setup KDNET on this system so no debug logs.

edit: @George King's 4th Dec build still works just fine.

Edited December 11, 2022 by Damnation