Jump to content

CFGLock - unlock (MSR 0xE2)


98 posts in this topic

Recommended Posts

On 3/26/2021 at 9:08 AM, Slice said:

Then into Shell and then into ControlMsrE2.

The application is now compiled with Clover and will be included into next Clover release.

Great news @Slice but please Slice and @Jief_Machak kindly add in the config file sample a relative "Tool" entry/section for this EFI tool as a guide for others that have never used such entries before? Thank you.

wait and see. With ControlMsrE2 integrated, Clover could toggle the MSR E2 bit to 0 automatically, without any user input...
 

 

36 minutes ago, BuXb said:

wait and see. With ControlMsrE2 integrated, Clover could toggle the MSR E2 bit to 0 automatically, without any user input...
 

 

It is interesting idea but it should be optional because it works not in all cases and has options what to choose manually.

On 3/25/2021 at 6:39 PM, BuXb said:

No.. first boot into your COMPUTER's boot menu, THEN into Clover

 

 

Intuitively, that make sense; however, booting first into my rig's boot menu (F9 in my case) does not expose Msr 0xE2 to ControlMsrE2.  VerifyMsrE2 reports Msr 0xE2 as locked, but that may be a false reading produced by VerifyMsrE2 when it can't find Msr 0xE2 - not sure.  I fear that my rig's BIOS has the same issue as @MacKonsti 's NUC8.

5 hours ago, tonyx86 said:

 

Intuitively, that make sense; however, booting first into my rig's boot menu (F9 in my case) does not expose Msr 0xE2 to ControlMsrE2.  VerifyMsrE2 reports Msr 0xE2 as locked, but that may be a false reading produced by VerifyMsrE2 when it can't find Msr 0xE2 - not sure.  I fear that my rig's BIOS has the same issue as @MacKonsti 's NUC8.

I had similar issue with BIOS F20 on my comp #1. First core unlocked but other cores locked. The register 0xE2 exists separately for each core.

ControlMsrE2 knows this issue and shows all cores.

Hi everyone, so this procedure attempt is not getting any results on my NUC8 unfortunately, allow me to explain:

  • Enable UEFI shell in BIOS; reboot; select F10 to boot into NUC's UEFI shell; switch to FS0: and ran ControlMsrE2..efi interactive; the computer freezes.
  • Press F10 for boot selector; select main Clover drive; Clover bootloader is shown; select UEFI shell; switch to FS1: and ran ControlMsrE2.efi interactive; no search string is found.

I tried "cfg" and "msr" and "cpu" no results at all. I tried "lock" but the only result was for "BootNum Lock" so that's something i.e. ControlMsrE2 looks into NVRAM. So either the firmware is not exposing these MSR 0xE2 at all or the names are totally non-standard.

 

If you have any idea as to why the BIOS native UEFI shell freezes during ControlMsrE2 it would be great. But did I follow the procedure you mentioned @tonyx86 and @Slice ? I am not sure why this procedure would "catch" the registers like that, in my mind (but I have no concrete knowledge how the boot process works on NUC either).

 

Thanks

The latest nighty of OpenCore contains an updated version of ControlMsrE2. The process of how to use it is described on page 37, chpt. 7.8 under "AppleCpuPmCfgLock"

Edited by 5T33Z0
  • Like 1
16 minutes ago, MacKonsti said:

Hi everyone, so this procedure attempt is not getting any results on my NUC8 unfortunately, allow me to explain:

  • Enable UEFI shell in BIOS; reboot; select F10 to boot into NUC's UEFI shell; switch to FS0: and ran ControlMsrE2..efi interactive; the computer freezes.
  • Press F10 for boot selector; select main Clover drive; Clover bootloader is shown; select UEFI shell; switch to FS1: and ran ControlMsrE2.efi interactive; no search string is found.

I tried "cfg" and "msr" and "cpu" no results at all. I tried "lock" but the only result was for "BootNum Lock" so that's something i.e. ControlMsrE2 looks into NVRAM. So either the firmware is not exposing these MSR 0xE2 at all or the names are totally non-standard.

 

If you have any idea as to why the BIOS native UEFI shell freezes during ControlMsrE2 it would be great. But did I follow the procedure you mentioned @tonyx86 and @Slice ? I am not sure why this procedure would "catch" the registers like that, in my mind (but I have no concrete knowledge how the boot process works on NUC either).

 

Thanks

The tool ControlMsrE2.efi requires some protocols provided by UEFI BIOS or Clover or Shell so why you have different results. I proposed most complex way BIOS->Clover->Shell->ControlMsrE2.efi which is tested to be OK.

Anyway if you read this topic from the begin you may see the method is not always working. It is not working on my Dell Latitude E6430. I can't unlock 0xE2 there.

  • Thanks 1
1 hour ago, Slice said:

The tool ControlMsrE2.efi requires some protocols provided by UEFI BIOS or Clover or Shell so why you have different results.
I proposed most complex way BIOS->Clover->Shell->ControlMsrE2.efi which is tested to be OK.

Anyway if you read this topic from the begin you may see the method is not always working. It is not working on my Dell Latitude E6430. I can't unlock 0xE2 there.

 

Just to be 100% sure by "BIOS" in "BIOS->Clover->UEFI Shell->ControlMsrE2" you do mean the boot selector screen of the PC (BIOS) correct? Because I cannot launch Clover from BIOS like on some other PCs (e.g. my old MSI) where you can select (one-time only) a drive/partition to launch from within the BIOS Setup screen.

 

Yes it seems the MSR 0xE2 parameter/variable is not exposed at all outside the NUC8 firmware or uses totally non-standard naming. NUC8 uses Intel Visual BIOS firmware; but NUC10 uses a different BIOS maker, and there I can unlock Cfg as it is found correctly.

 

Edited by MacKonsti
48 minutes ago, MacKonsti said:

 

Just to be 100% sure by "BIOS" in "BIOS->Clover->UEFI Shell->ControlMsrE2" you do mean the boot selector screen of the PC (BIOS) correct? Because I cannot launch Clover from BIOS like on some other PCs (e.g. my old MSI) where you can select (one-time only) a drive/partition to launch from within the BIOS Setup screen.

 

Yes it seems the MSR 0xE2 parameter/variable is not exposed at all outside the NUC8 firmware or uses totally non-standard naming. NUC8 uses Intel Visual BIOS firmware; but NUC10 uses a different BIOS maker, and there I can unlock Cfg as it is found correctly.

 

Sorry, what hardware did you mean? Are you know mrs 0xE2 exists from SandyBridge and up?

  • Thanks 1

Yes sorry I meant the BIOS of my MSI Core i7-4790K apologies forgot to mention it. The MSI BIOS has Cfg Lock menu but also allows to boot to a selected drive directly, I meant this "direct boot" feature really. On NUC you have to press F10 to get boot menu first and then select Clover drive/partition but I am not sure if this process "exposes" the BIOS parameters in NVRAM in some different way.

 

Anyway, I stop posting and flooding the thread, as NUC8 is pretty much a dead-end, but NUC10 worked even with older CfgLock.efi too. Thanks all for your support. I will follow the discussion, of course.

 

Edited by MacKonsti
9 hours ago, Slice said:

Are you know msr 0xE2 exists from Sandy Bridge and up?

Slice, where do you have that information from? I googled and did not find anything confirming that.

3 hours ago, BuXb said:

Slice, where do you have that information from? I googled and did not find anything confirming that.

At least as early as Haswell.

 

This seems to imply that MSR 0xE2 is applicable to Sandy Bridge.

Edited by tonyx86
Added Dortania link for Sandy Bridge
12 hours ago, BuXb said:

Slice, where do you have that information from? I googled and did not find anything confirming that.

www.intel.com

PS. It is always a complex quest to find a documentation on the site. I have a collection of some documents and may upload some documents here.

@Slice as you predicted, no results when searching for terms like MSR E2, CFG lock etc. on intel.com

 

Anyway, as long as your sure this register was introduced with Sandy Bridge CPUs, that's fine.

 

Do you also happen to know which version of macOS / OS X Apple had first request access to the MSR 0x02 register?

19 minutes ago, BuXb said:

@Slice as you predicted, no results when searching for terms like MSR E2, CFG lock etc. on intel.com

 

Anyway, as long as your sure this register was introduced with Sandy Bridge CPUs, that's fine.

 

Do you also happen to know which version of macOS / OS X Apple had first request access to the MSR 0x02 register?

Снимок экрана 2021-03-30 в 22.14.57.png

 

 

The problem with locked MSR 0xE2 exists since first system supported SandyBridge (10.7 may be?)

 Снимок экрана 2021-03-30 в 22.16.13.png

  • Thanks 1

I looked carefully and see that 0xE2 exist also on 1st generation CPU Nehalem family.

  • Like 1

Core 2 Duo. Any OS: NO, NO

CfgLock doesn't exists and no sense to patch for it.

  • Like 1

@Slice Ok, Nehalem - good you double-checked.. I also pointed zhen-zen and dreamwhite to your post so they can include that info in the Cfg Lock/ MSR E2 part of their guide.

 

Andrey1970's table: according to your latest finding, shouldn't it be Nehalem instead Haswell??

 

43 minutes ago, BuXb said:

Andrey1970's table: according to your latest finding, shouldn't it be Nehalem instead Haswell??

 

No.

Nehalem in the table it Ivy and older

  • Like 1
×
×
  • Create New...