[GUIDE] DSDT override eGPU error 12 fix  

Page shortcut:  https://egpu.io/DSDT   |  dsdt

Guide: DSDT override eGPU error 12 fix (Windows only) [started [email protected] ]

 

here ◄ Mac owners please work through this thread to solve error 12 before doing a DSDT override
here ◄ pre-compiled DSDT overrides  - load or use as reference when making your own

 

Introduction

 

A Windows system's DSDT table root bridge definition (ACPI PNP0A08 or PNP0A03) is usually confined to a reserved 32-bit space (under 4GB) budgetted to be large enough to host the notebook's PCIe devices. A watermark TOLUD value is then set and locked in the system firmware. Windows OS honors the root bridge definition and will allocate PCIe devices within it. macOS ignores the root bridge constraints as too does Linux when booted with the 'pci=noCRS' parameter. Neither of those OS require a DSDT override and can allocate freely in the huge 64-bit PCIe address space.

 

When retrofitting a eGPU, an error 12 (This device cannot find enough free resources that it can use) can occur against an eGPU in Windows' device manager making it inoperable. This can indicate there is insufficient 32-bit addressing space available to host the eGPU. An eGPU requires a relatively large PCIe config space to allocate into. Decreasing TOLUD by reducing RAM to 2GB offers a somewhat impractical workaround. Rather, the definitive solution is below.

 

---

This three step solution removes Window's 32-bit PCIe allocation constraint in order to resolve the eGPU error 12:

 

Step 1. Create a dsdt-modified.aml DSDT file with a 36-bit root bridge

 

Step 2. Load your dsdt-modified.aml as registry override or in-memory substitution

Step 3. Confirm success with a 'Large Memory' area in Device Manager

 

 

macOS users: refer instead to Mikeal's post that covers these steps titled Windows 10 - Clover DSDT memory override [UEFI Windows on Macbooks only].

 

---

 

Step 1. Create a dsdt-modified.aml DSDT file with a 36-bit root bridge

 

i. Download and install required tools:

• Windows Binary Tools (WBT - Dec 2016) extracted to c:\dsdt directory.  [ newer WBT has parsing errors]
• Windows Driver Kit (WDK), which contains the Windows ASL Compiler (asl.exe)
• Notepad++ text editor with Search->Goto (line) menu for fast line editting if asl or iasl compilation fails below

 

ii. Copy WDK's ASL compiler into the c:\dsdt directory. Do this by opening Command Prompt (run as administrator) and then copy-and-paste the commands below.

mkdir c:\dsdt
c: & cd \dsdt
set 64bit_OS_asl="C:\Program Files (x86)\Windows Kits\10\Tools\x64\ACPIVerify\asl.exe"
set 32bit_OS_asl="C:\Program Files (x86)\Windows Kits\10\Tools\x86\ACPIVerify\asl.exe"
copy /y %32bit_OS_asl% c:\dsdt > nul & copy /y %64bit_OS_asl% c:\dsdt > nul
if not exist c:\dsdt\asl.exe echo ERROR: Failed to copy asl.exe to c:\dsdt

 

iii. Dump your ACPI tables to disk files (dsdt.asl and dsdt.dat) with these commands at Command Prompt (run as administrator). The created dsdt.asl is copied here as dsdt-modified.asl which is used later on to make our required modifications.

c: & cd \dsdt
acpidump -b -z
asl /u dsdt.dat
copy dsdt.asl dsdt-modified.asl

 

iv. You now can choose either the Intel method or Microsoft method (with blue dsdt-modified.dsl or magenta dsdt-modified.asl work file respectively) to generate a dsdt-modified.aml file, even trying both to maximize success. Consider:

- For systems other than Lenovo, use the Intel method as their DSDT usually has an Intel creation signature.

- Lenovo Thinkpad X220, T420, W530, T540P and likely other 2nd-4th gen i-core Lenovo Thinkpad systems are known to require the Microsoft method . The Intel method causes a "ACPI BIOS ERROR" on Windows bootup there.

 

OPTION 1: Use the Intel method

 

i. Save this >> refs.txt file << to your c:\dsdt folder. What is it used for? From tonymacx86: The iasl disassembler will attempt to guess the number of arguments [for unresolved symbols not defined in any file] but often guesses poorly. You can correct it by providing the external declarations in a refs.txt text file. It contains some common (and not so common) missing symbols . . .

 

ii. From the Command Prompt (admin), decompile dsdt.dat as dsdt.dsl . The refs.txt file is used here. dsdt.dsl is then copied as dsdt-modified.dsl on which we'll make the required changes on.

c: & cd \dsdt
iasl -da -dl -fe refs.txt dsdt.dat
copy dsdt.dsl dsdt-modified.dsl

 

iii. With Notepad++, open the resultant c:\dsdt\dsdt-modified.dsl file and search for ResourceProducer. Beneath it will be a series of "DWordMemory" resource entries. Under the last DWordMemory entry in that area, typically above the _CRS method, add a 'QWordMemory' (64-bit) entry as shown in red below. The range chosen is in the 36-bit range (< 64GB) to maintain compatibility with PAE-capable 32-bit Windows. A location above 48GB was chosen to alleviate issues with 32GB equipped systems. Here we use between 48.5GB to 56.25GB. Once systems start shipping with 64GB of RAM, this will need to be revised to use a 64-bit address.

 

DWordMemory (ResourceProducer, PosDecode, MinFixed, MaxFixed, Cacheable, ReadWrite,
   0x00000000,         // Granularity
   0x000A0000,         // Range Minimum
   0x000BFFFF,         // Range Maximum
   0x00000000,         // Translation Offset
   0x00020000,         // Length
   ,, , AddressRangeMemory, TypeStatic)
QWordMemory (ResourceProducer, PosDecode, MinFixed, MaxFixed, Cacheable, ReadWrite,
   0x0000000000000000, // Granularity
   0x0000000C20000000, // Range Minimum,  set it to 48.5GB
   0x0000000E0FFFFFFF, // Range Maximum,  set it to 56.25GB
   0x0000000000000000, // Translation Offset
   0x00000001F0000000, // Length calculated by Range Max - Range Min.
   ,, , AddressRangeMemory, TypeStatic)
})
   Method (_CRS, 0, Serialized)  // _CRS: Current Resource Settings

 

iv. Create a dsdt-modified.aml file . The  '-ve' disables warning messages.

c: & cd \dsdt
iasl -ve dsdt-modified.dsl

 

It is unlikely this will succeed first time as the compiler is very strict. Errors reported will need to be looked at with Notepad++ on the line they occur on. Search for a unique error keyword from your dsdt-modified.dsl error line within dsdt-modified.asl (it may look a bit different) and the just swap the lines above/below into your dsdt-modified.dsl file and compile. This was sufficient to get a Dell XPS 9350, Dell E6540, 2016 15" Macbook Pro DSDT override all done perfectly.

 

If stuck then see eGPU.io's public pre-compiled DSDT repository for various systems. If one exists for your system then can download it along with Winmerge  to merge the file changes into your dsdt-modified.dsl work file. There are also other fixes like described at [Guide] Patching LAPTOP DSDT/SSDTs (tonymacx86). Consider also asking for guidance at the DSDT-centric tonymacx86.com DSDT forums . Then try compiling your DSDT again.

 

v . Proceed to Step 2 to load your dsdt-modified.aml file.

 

OPTION 2: Use the Microsoft method

 

i. With Notepad++, find the working area in your c:\dsdt\dsdt-modified.asl file as shown in screenshots below. In that area, make the following two edits. These are annonated in those screenshots.

 

EDIT 1: use Windows' calc in programmers mode to add hexidecimal 2E (46 decimal) to your "Buffer(value)" line. Examples:

• "Buffer(0x1C0)" becomes "Buffer(0x1EE)"
• "Buffer(0x1D4)" becomes "Buffer(0x202)"

EDIT 2: replace "0x79, 0x00" in the hex table end with the following 46 hex values, also ending with "0x79, 0x00". This adds the same 36-bit QWordMemory resource entry shown in the Intel method but as a hex table.

 

[ copy-and-paste this table ]

0x8a, 0x2b, 0x00, 0x00, 0x0c, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x20, 0x0c, 0x00, 0x00, 0x00, 0xff, 0xff,
0xff, 0x0f, 0x0e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0xf0, 0x01, 0x00, 0x00, 0x00, 0x79, 0x00   

 

For Lenovo Systems [ CLICK on images to zoom ]

 

For other systems [ CLICK on images to zoom ]

 

ii. At the Command Prompt (admin), compile your dsdt-modified.asl file to generate a dsdt-modified.aml file:

c: & cd \dsdt
asl /Fo=dsdt-modified.aml dsdt-modified.asl

 

This will almost certainly fail to generate a dsdt-modified.aml on your first compile attempt due to to error lines requiring further editting of your dsdt-modified.asl file, like the ones below (further example here), followed by another compile attempt. A successful compilation with many warnings is OK.

- ATMC() - error: ATMC is not a method -> swap ATMC with \_SB_.PCI0.LPC_.EC__.ATMC()
- Zero - error: unexpected ASL term type -> add 'Zero' as bracketted argument to line above it, eg: GLIS (Zero)
- Arg0 - error: unexpected ASL term type -> add 'Argo0' as bracketted argument to line above it, eg: GDCK (Arg0)

 

iii. Open c:\dsdt with Windows explorer. Compare the size of your created dsdt-modified.aml file against the memory dumped dsdt.dat. They should be within +/- 10% of each other in size. If not, repeat the above process to make sure no mistakes were made.

A HP Elitebook 8440P saw the generated dsdt-modified.aml being only 15% the size of the memory dumped dsdt.dat. There this process simply did not work. For such cases use the Intel method instead.

 

iv . Proceed to Step 2 to load your dsdt-modified.aml file.

 

 

Step 2. Load your dsdt-modified.aml as a registry override or in-memory substitution

 

OPTION 1: Load your dsdt-modified.aml as a registry override with Windows test signing mode enabled

 

Here we load your dsdt-modified.aml as Windows registry DSDT override. Do note that an invalid dsdt-modified.aml loaded in this way can cause a BSOD on bootup. Furthermore, Windows test signing mode can be problematic for app compatibility. Both these issues can be avoided by using OPTION 2: Avoid test signing mode - load your dsdt-modified.aml as an in memory DSDT substitution.

 

i. At Command Prompt (admin) type the following.  WDK containing asl.exe must be installed per Step 1 (i) and (ii)  for this to work.

c: & cd \dsdt
asl /loadtable dsdt-modified.aml  

 

ii. Enable TESTSIGNING mode for the registry override to apply. At the Command Prompt (admin) type:

bcdedit -set TESTSIGNING ON

 

If get an error like below when do this, then disable SECURE BOOT in your BIOS.

An error has occurred setting the element data
The value is protected by Secure Boot policy and cannot be modified or deleted

 

iii. Reboot your system and check for 'large memory' in step 3.

 

OPTION 2: Avoid test signing mode - load your dsdt-modified.aml as an in memory DSDT substitution

 

Here we avoid the problematic registry override & test signing mode altogether by loading the DSDT as an in-memory substitution before Windows loads with these steps:

 

i. Disable the previous registry DSDT override and test signing mode.

bcdedit -set TESTSIGNING OFF

 

ii. check whether you are using a MBR or UEFI Windows installation by viewing Device Manager->Disk Drives->[double-click boot drive]->Volumes->Populate->Partition Style. If it says "MBR", then it's a MBR install. If it says "GPT" then it's a UEFI install. This will determine which next step to apply.

 

 

iii. [ MBR mode ] Load the dsdt-modified.aml as an in-memory substitution via nando's DIY eGPU Setup 1.35 in the next post.

 

iv. [ UEFI mode] Load the dsdt-modified.aml as an in-memory substitution via Clover bootloader as follows:

 

---

Update Feb-2019 >> Mac users are advised to use @goalque's automate-eGPU EFI instead of Clover to load your resultant DSDT override to avoid issue noted below in the BIG WARNING.

 

BIG WARNING by nando4 >> @Goalque has correctly identified that Clover loads a DSDT table in firmware volume and as such can brick a Macbook as this user found. If you proceed with using Clover to do a DSDT override the you do so at your own risk!! For risk-adverse users it is suggested to simply do a DSDT registry override and persevere with Windows' test signing mode until other solutions are found and presented.

---

 

- Mount your EFI volume as s: drive and backup your existing \EFI\BOOT\BOOTX64.efi file. At Command Prompt (admin) type:

mountvol s: /s
copy s:\EFI\BOOT\BOOTX64.efi s:\EFI\BOOT\BOOTX64.win

 

- Download the Clover ISO file and install it by extracting with 7-zip (64-bit only) the clover.tar.lzma->clover.tar->clover.pkg->\EFI folder to s:\EFI. Be sure to use the the 7-zip interface to extract to s: drive as Windows explorer refuses to allow viewing of the s: EFI volume.

 

- Copy your dsdt-modified.aml file as dsdt.aml to s:\EFI\CLOVER\ACPI\Windows, the directory Clover uses to preload it. If your source dsdt-modified.aml file is not in c:\dsdt then use this workaround to copy-and-paste it. Locate your dsdt-modified.aml file with Explorer, right-click, rename to dsdt.aml, right-click, copy. Then hit CTRL+ALT+DEL, task manager, File->New task. Explore the s:\EFI\CLOVER\ACPI\WINDOWS directory and paste it there.

 

Otherwise at the Command Prompt (admin) type:

copy c:\dsdt\dsdt-modified.aml s:\EFI\CLOVER\ACPI\WINDOWS‌\dsdt.aml

 

- Reboot via Clover -> Windows EFI menu and proceed to step 3 to confirm it worked.

 

 

Step 3. Confirm success with a 'large memory' area in Device Manager

 

Check you now have a new Large Memory entry in Device Manager as shown below to confirm your dsdt-modified.aml was a success:

 

Success stories

 

nando4 - HP ZBook 17 G2 + AKiTiO Thunder2 + GTX970. SUCCESS! Allocated the eGPU into 36-bit space.
Mikeal - 2016 15" Macbook Pro + Razer Core + GTX1070. SUCCESS! Allocated the eGPU into 36-bit space.
itsage - 2016 13" Macbook Pro + AKiTiO Thunder3 + GTX980Ti. SUCCESS! Allocated the eGPU into 36-bit space.
batist - 2013 15" Lenovo T540 + EXP GDC + GTX680. SUCCESS! Allocated the eGPU into 36-bit space.

 

FAQ

 

1. I still have an error 12 with the 'large memory' area present. How can I fix it?

i. [MBR/Legacy boot - not UEFI] Force allocate the eGPU into 36-bit PCI space before booting Windows with nando's DIY eGPU Setup 1.35 software:

• At Windows boot menu, select Setup 1.35
• Boot Setup 1.35 -> menu-based
• Select PCI compaction->Endpoint=56.25GB (36-bit)
• Select PCI compaction->Run compact. When prompted for the scope select eGPU.
• Select Chainloader->Test Run
• At Windows boot menu, select Windows.

 

Your result will be then be like shown in step 3 above, but you'll have the eGPU rather then the (example) HD8790M dGPU allocated into 36-bit PCI space.

 

ii. Use the bcdedit forcedisable option to override the advanced PCI Express features and use legacy PCI Express behavior as explained by Microsoft. madjester confirmed this fixed error 12 on 2015 13" MBP after doing a DSDT override here (thank you!)

bcdedit.exe -set PCIEXPRESS FORCEDISABLE

 

iii. Re plug'n'play the devices. Go into Device Manager -> Video cards, delete the eGPU. Also go into Device Manager -> System devices and delete the PCI Express Root Port x entries.

Window should now be clever enough to re-allocate the eGPU into 36-bit space upon realizing there is insufficient 32-bit space to host it.

Can go one step further and delete your eGPU NVidia/ATiAMD driver, restart the system and reload it. That may help as well.

 

iv. From here:

• remove all of the devices in Device Manager - Resources using the large memory space (All of the ones with a blue icon - not the USB icon one!)
• Hot Plug the eGPU (Within a few minutes the eGPU will be usable without error 12 and the other devices will also come back and start working again)
• Everything should work

  

 

2. How do I disable the registry DSDT override?

The easiest method is to disable TESTSIGNING. If your system fails to boot with the DSDT override in place then boot in Safe Mode and disable TEST SIGNING.

bcdedit -set TESTSIGNING OFF 

 

Remove the DSDT override from the registry:

c: & cd\dsdt
asl /loadtable -d dsdt-modified.aml

 

.. or remove this DSDT key in the registry. Windows will create a new one derived from your system bootup ACPI DSDT table upon restart:

HKEY_LOCAL_MACHINE\SYSTEM\ControlSetXXX\services\ACPI\Parameters\DSDT

* where XXX are 001, 002, 003 & so on.

and

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\ACPI\Parameters\DSDT