Jump to content

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


Mov AX, 0xDEAD

Recommended Posts

@Andalu

I change on the Ramsey XP the whole HIVE system against my working one.

On Safe Mode, the Ramsey XP boots for few seconds to full desktop, without any error message, then reboots.

On normal boot, I get the Bsod 7F again.

This is really crazy

Dietmar

EDIT: That the Ramsey XP boots in Safe Mode is an indicator for an driver conflict of the nvme.sys driver.

Edited by Dietmar
Link to comment
Share on other sites


@Dietmar

to enter on the desktop with the XP from Ramsey you can try to delete winxpup.inf from the prepared ISO.
In this way you will encounter many errors for missing files (the first one during TXT mode for winxpup.inf) but by clicking "esc and OK" on the various messages the installation can be completed anyway.

XP so installed will be almost unusable, and unfortunately, you will not be able to install the video driver either, which I believe is the factor that causes a change in XP's configuration that leads to the conflict with the nvme driver we are testing.

With other NVMe drivers and in XP x64 with installed the driver we are testing the NVMe drive and the video card share the same IRQ 16 but I never encountered the BSOD 7F.

Edited by Andalu
Link to comment
Share on other sites

@Andalu

It looks, as if XP uses only the model for IRQ 0...17 (not enlarged).

And with this, about 10 devices have IRQ 16. This does not like this new nvme.sys driver.

I catch the moment of fail: 9 devices have already IRQ16. With the graphikcard NVIDIA install, this next device wants IRQ16 also.

This is too much, gives Bsod 7F on next reboot. But the compi starts in Safe Mode, without graphikcard on IRQ16.

 

@Mov AX, 0xDEAD

Do you have an idea, how to enable the enlarged IRQ (0...255?) under XP SP3

Dietmar

 

Link to comment
Share on other sites

@Andalu

I succeed to boot with this new nvme.sys driver on the Samsung 970 Pro,

and graphikcard gt730 Nvidia and all the devices under IRQ16.

For this I setup my original XP SP3 *.iso, modd it via Ramsey, but edit all by hand. I use my new acpi.sys with OS fakes.

The setup took place via Firadisk and the Kai driver for HD on my Asrock extreme z690 board with 12900k cpu.

Then, I install really everything on this HD.

And at the end also the new nvme.sys driver by hand.

This XP SP3 on HD I "show" to the 970 Pro, means I connect the 970 Pro.

In registry I set Start = 4 for the stornvme.sys.

Then I copy via AOMEI 7.0 this whole HD to the 970 Pro.

Voila, boots very fast via the new nvme.sys,

no Bsod

Dietmar

Edited by Dietmar
Link to comment
Share on other sites

3 hours ago, Dietmar said:

@Andalu

It looks, as if XP uses only the model for IRQ 0...17 (not enlarged).

And with this, about 10 devices have IRQ 16. This does not like this new nvme.sys driver.

On same IRQ can be many devices, this is concept, when irq signal triggered, windows call handler for each driver in cycle. Driver then check hardware registers of device and if it say "interrupt was not initiated by my device", windows call next driver, etc.

i think problem in nvme/storport  itself, maybe it expects MSI type of interrupt or uses too much stack memory or some else

3 hours ago, Dietmar said:

@Mov AX, 0xDEAD

Do you have an idea, how to enable the enlarged IRQ (0...255?) under XP SP3

cannot help with this, i don't know how pci.sys works

Link to comment
Share on other sites

@Mov AX, 0xDEAD

When I compare nvme2.txt with nvme3.txt I see, that the crash belongs to the graphik driver

together with the nvme.sys, very near to the end of the boot process

Dietmar

 

nvme2.txt

MiSessionWideReserveImageAddress: NO Code Sharing on \SystemRoot\System32\drivers\dxg.sys, Address 0xbf000000

*** Fatal System Error: 0x0000007f
                       (0x00000008,0xBA3D0D70,0x00000000,0x00000000)

 

nvme3.txt

MiSessionWideReserveImageAddress: NO Code Sharing on \SystemRoot\System32\nv4_disp.dll, Address 0xbd012000
8A21A080 8 (0x886fb008): ACPIDeviceIrpDeviceFilterRequest - Transition to D2
8A21A080 8 (0x8a0a7730): ACPIDevicePowerProcessPhase0DeviceSubPhase1 = 0x00000000
8A21A080 8 ACPIDeviceCompleteGenericPhase = 0x00000000
8A21A080 8 (0x8a0a7730): ACPIDevicePowerProcessPhase0DeviceSubPhase2

 

Edited by Dietmar
Link to comment
Share on other sites

@Andalu

This means, that this new nvme.sys driver is not usable on normal XP SP3 install.

IF you succeed to install it, it is stable. No crash then, all ok. All nvme devices are recogniced.

The Silicon nvme driver you succeed to install, but then behaves unstable, brr..

The Toshiba nvme driver recognices only Toshiba nvme devices, there it is stable.

 

So, until now only the Microsoft stornvme.sys works correct always under XP SP3 with any nvme device,

the OFA 1.3 nvme driver (works with most nvme devices), is stable

and sometimes the Samsung nvme driver (does not recognice a lot of non Samsung devices), is stable.

 

Until now no trim under XP SP3 on nvme and no working CrystalDiskInfo.

Today again I make a very strange discovery with the Samsung 970 Pro 1 TB nvme device:

After about 17% zero out with WinHex, it hangs.

When you trim it via win10, you can wipe out 50%, again crash.

And the most most bad: from 3970 MB/sec speed goes down to 20 MB/sec, much less than any harddisk.

When you connect the Samsung 970 Pro with an USB encloser, only about 50% files that you copy, are in real there.

 

When I use the Optane 900P or 905P, such crazy bad things never happen and it is from feeling

always faster than any other nvme

Dietmar

 

Edited by Dietmar
Link to comment
Share on other sites

@Dietmar

I still thinking this is stack problem:

Quote

.text:00413976                 push    ebp
.text:00413977                 mov     ebp, esp
.text:00413979                 sub     esp, 0AA8h
.text:0041397F                 lea     eax, [ebp+var_AA8]
.text:00413985                 push    esi

1) 413976 - some procedure entry inside nvme,sys, esp at this point is ba294f9c+AA8= BA295A44 or BA295A40 or BA295A48

It allocate local buffer on kernel stack with size AA8 = 2728 Bytes

I think stack pool ended at BA295000, 4k aligned

ba294f9c is below stack pool so get BSOD when driver PUSH ESI to this memory

 

2) Win7/Win8 kernels have kernel KeExpandKernelStackAndCallout(), it is emulated by NTOSKRNL Emu_Extender

If nvme uses it and some wrong with emulation code (i never tested this procedure) -> driver tried to allocate more stack pool, but really it was not happen, so stack is same size.

 

Edited by Mov AX, 0xDEAD
Link to comment
Share on other sites

@Mov AX, 0xDEAD

Do you have an idea, what I can test more with this new nvme.sys driver?

May be, with the working XP SP3 install with this nvme.sys as scsi boot driver?

Or what can be changed, tested in the NTOSKRNL Emu_Extender ?

It is a nice driver, because it recognices ALL nvme devices and if it is work, it is stable.

I also make an try with the last version of the acpi.sys 6666 from outerspace ).

This gives exact the same Bsod 7F, (0x08,..)

Dietmar

EDIT: Can I enlarge the stack by hand in this nvme driver by hand in

.text:00413979                 sub     esp, 0AA8h

Edited by Dietmar
Link to comment
Share on other sites

This is all so interesting to me. I'd love to install XP on my MSI Z690-A Pro WiFi D4 for 3DMark01 runs with my 13900K on Ln2! I have no idea where to even start. Would someone be able to provide assistance, or maybe even an image if they've had it working on the same or similar board? Thank you

 

Link to comment
Share on other sites

4 minutes ago, Dietmar said:

@RandyOC

I have the MSI z690-A Pro DDR4 just without wifi.

I succeed after loong testing to install XP SP3 to this board with 12900k cpu.

Please tell me, which drivers you want and I make an xp.iso for you

Dietmar

that would be unbelievable mate! thank you!! if you can do USB, that would be amazing, but I'm sure I could work without it. I need an NVIDIA drivers for 900 series, and that's about it. 

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...