Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

10.15 beta 4 AppleIntelMCEReporter #424

Closed
Pavo-IM opened this issue Jul 17, 2019 · 41 comments

Comments

@Pavo-IM
Copy link

@Pavo-IM Pavo-IM commented Jul 17, 2019

Is there away we can get a new patch added to Lilu to patch AppleIntelMCEReporter.kext. Apple has introduced a new controller inside the driver called AppleIntelMCEInteruptController which has included the iMacPro1,1, MacPro6,1 and assume the new MacPro7,1 board-ids. This is causing kernel panic for Ivy Bridge-EP CPUs.
10.15 beta 3 AppleIntelMCEReporter.kext Info.plist
https://imgur.com/S9De4FP
10.15 beta 4 AppleIntelMCEReporter.kext Ino.plist
https://imgur.com/ACDbmQf

@PMheart

This comment has been minimized.

Copy link
Contributor

@PMheart PMheart commented Jul 17, 2019

Hello,

I have several ideas to test:

  • Patch the plist itself (by replacing MP61 to something else)
  • If there were no harm to totally block the kext, it could also be done via OC Kext Block.
@PMheart PMheart closed this Jul 17, 2019
@PMheart PMheart reopened this Jul 17, 2019
@PMheart

This comment has been minimized.

Copy link
Contributor

@PMheart PMheart commented Jul 17, 2019

Sorry for the close. Reopening.

@vit9696 vit9696 added the project:oc label Jul 17, 2019
@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Jul 17, 2019

It makes sense to me to block kext with OC, Lilu has nothing to do with it. Closing as not an issue really.

@vit9696 vit9696 closed this Jul 17, 2019
@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Jul 17, 2019

Use OC Kernel > Block did not work. This is from OC logs:
00:000 00:000 OC: Prelink blocker com.apple.driver.AppleIntelMCEReporter init failure - Not Found
I am assuming Kernel > Block section can only block kexts that is in the Prelink and not during booting phase. The KP is happening on update phase and fresh install phase.

@vit9696 vit9696 reopened this Jul 17, 2019
@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Jul 17, 2019

Well, ok, do provide a panic log now then with keepsyms=1 in boot arguments and the kext file attached.

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Jul 17, 2019

@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Jul 17, 2019

Please reread what I said, make good sense of it, and provide the necessary data. Thank you.

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Jul 17, 2019

Oh my bad here is the kext, didn't read that part lol.
AppleIntelMCEReporter.kext.zip

@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Jul 17, 2019

It's keepsyns=1 instead of keepsyms=1 too, and thus no symbols are visible ^^

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Jul 17, 2019

IMG_0017

@PMheart

This comment has been minimized.

Copy link
Contributor

@PMheart PMheart commented Jul 17, 2019

You are still leaving keepsyNs=1...

@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Jul 17, 2019

E21CCF38-5231-442C-9532-7739BB1A8BF4

Listen, I am patient, but if this continues, I will just close the issue until you learn how to provide bugreports. Thanks for understanding and please do save your own and my time.

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Jul 17, 2019

Ok I am little confused because when I provided the OC log output using Kernel > Block. You said and I quote "Well, ok, do provide a panic log now then with keepsyms=1 in boot arguments and the kext file attached." I provided a pic of the panic with keepsyns=1 already in it but forgot to attach the kext. I attached the kext next and then I posted a pic of the panic still with keepsyns=1 and now you are saying that I am not providing the information that you have asked for. So which is it? Do you want the a pic of the kernel panic with or without keepsyns=1 as a bootarg?

@PMheart

This comment has been minimized.

Copy link
Contributor

@PMheart PMheart commented Jul 17, 2019

We want the bugreport with keepsyMs=1 (M is just for emphasis, which should be LOWercase in fact).

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Jul 17, 2019

That is not what @vit9696 said, he said specifically keepsyns=1 instead of keepsyms=1.

@InsanelyHack

This comment has been minimized.

Copy link

@InsanelyHack InsanelyHack commented Jul 17, 2019

Well, ok, do provide a panic log now then with keepsyms=1 in boot arguments and the kext file attached.

This is what you put not keepsyns=1 it’s in the OC Pdf as keepsyms=1, it’s confusing as you have listed both N and M, so which is it?

@PMheart

This comment has been minimized.

Copy link
Contributor

@PMheart PMheart commented Jul 17, 2019

You must be misunderstanding what @vit9696 said then.

Well, anyway, do not use keepsyNs=1 until you write the code for Apple. ;)

@InsanelyHack

This comment has been minimized.

Copy link

@InsanelyHack InsanelyHack commented Jul 17, 2019

Right so both screenshots had keepsyns, okay now i see it. yes now i understood his reply.

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Jul 17, 2019

Here is one with keepsyms=1
IMG_0018

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Jul 17, 2019

It looks like its checking something on the memory banks for the CPUs.

@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Jul 17, 2019

Well, to just disable this kext you could return 0 from __ZN21AppleIntelMCEReporter5probeEP9IOServicePi. Such a simple solution should work fine, so feel free to write a Lilu plugin for it.

The crash itself is caused by out of bounds array access:
Снимок экрана 2019-07-18 в 1 40 46

You have 2 physical CPUs, and this driver does not support it, and does not allocate enough memory, see __ZN13AppleIntelMCE10Controller4InitEv.

@vit9696 vit9696 closed this Jul 17, 2019
PMheart added a commit to acidanthera/CPUFriend that referenced this issue Jul 18, 2019
@PMheart

This comment has been minimized.

Copy link
Contributor

@PMheart PMheart commented Jul 18, 2019

Added solution to this issue at the master of CPUFriend, please compile and give it a try. Thanks.

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Aug 3, 2019

10.15 beta 5 has removed AppleIntelMCEInteruptController from AppleIntelMCEReporter.kext Info.plist. It has been revered back to what 10.15 beta 3 was.

@mrmiller

This comment has been minimized.

Copy link

@mrmiller mrmiller commented Aug 26, 2019

Just to add to the public record, AppleIntelIMCEInterruptController is back in beta 6. The patch in CPUFriend does successfully work around it.

@Leoyzen

This comment has been minimized.

Copy link

@Leoyzen Leoyzen commented Sep 24, 2019

I still face the similar problem, and add Lilu/CPUFriend in kext/others of clover directory not work for me. Can you give some help or direction? the error info is here, much appreciate.
We use Penryn in kvm vm hackintosh, seem the cpu family not support in catalina.
image

@Leoyzen

This comment has been minimized.

Copy link

@Leoyzen Leoyzen commented Sep 24, 2019

Another similar error may related:
image

@DuIslingr

This comment has been minimized.

Copy link

@DuIslingr DuIslingr commented Sep 25, 2019

image0
Trying to boot Catalina beta 9 with opencore on Qemu KVM and I keep getting this panic. CPUFriend seems to be broken. Whats interesting is Clover doesnt have this problem though. Can boot just fine via clover and dont even need cpufriend.

@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Sep 26, 2019

@Leoyzen @DuIslingr all that is pretty strange. The screen above shows __ZN21AppleIntelMCEReporter5probeEP9IOServicePi was not found, and this is how CPUFriend kills the kext. @PMheart please check and update the signature.

@vit9696 vit9696 reopened this Sep 26, 2019
@vit9696 vit9696 mentioned this issue Sep 26, 2019
10 of 17 tasks complete
@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Sep 26, 2019

The latest kext has all its symbols stripped, so it is not like we can easily disable it as before. For now consider using a mac model different from mp61, mp71, imp11.

@naveenkrdy

This comment has been minimized.

Copy link

@naveenkrdy naveenkrdy commented Sep 26, 2019

Here is a small codeless kext to disable it.
AppleMCEReporterDisabler.kext.zip

And also any idea why the kext blocking mechanism of OC is not working for this kext ?
and i have also observed that blocking AppleIntelCPUpowermanagement.kext is also not working and due to that i had to use nullcpu kext .

@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Sep 26, 2019

I think the issue is in kextd reloading the kext after blocking. It will be nice if you could create a separate bugreport for failing kext blocking with examples of working and not working kext blocks. I think we could try increasing kext version in addition to kmod patching or maybe nulling the constructors…

@naveenkrdy

This comment has been minimized.

Copy link

@naveenkrdy naveenkrdy commented Sep 27, 2019

@vit9696 sure! will create a seperate bugreport for it.

@zhang-wenchao

This comment has been minimized.

Copy link

@zhang-wenchao zhang-wenchao commented Oct 24, 2019

@naveenkrdy Me too. I must needs AppleMCEReporterDisabler.kext and NullCPUPowerManagement.kext. Thank you for your work.

@vit9696 CPUFriend not work

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Oct 24, 2019

Well looks like the Disablerkext isn't working for 10.15.1 b3. Guess I'll have to use a different SMBIOS until CPUFriend or something else has figured out how to disable this from getting loaded.

@zhang-wenchao

This comment has been minimized.

Copy link

@zhang-wenchao zhang-wenchao commented Oct 26, 2019

My CPU is Intel Xeon Gold 5117,must fackcpuid 0x0306E0 to boot。

@mrmiller

This comment has been minimized.

Copy link

@mrmiller mrmiller commented Oct 26, 2019

Are you running OpenCore? I submitted patches a month or so ago that allow Xeon Golds (and others) to boot without a fake CPU ID. That is unrelated to this issue and workaround though.

@zhang-wenchao

This comment has been minimized.

Copy link

@zhang-wenchao zhang-wenchao commented Oct 26, 2019

This is causing kernel panic for Ivy Bridge-EP CPUs, If my don't use fakecpuid, 10.15 should work.

@Pavo-IM

This comment has been minimized.

Copy link
Author

@Pavo-IM Pavo-IM commented Oct 26, 2019

Here is a pic of the new panic of AppleIntelMCEReporter
Screenshot 2019-10-26 12 41 03

@Pavo-IM Pavo-IM mentioned this issue Oct 29, 2019
9 of 9 tasks complete
@startergo

This comment has been minimized.

Copy link

@startergo startergo commented Nov 18, 2019

I can report the same issue in Catalina 10.15.1 with a real MacPro5,1 and iMacPro1.1 board ID. It does not happen in Mojave. Please see here:
https://www.insanelymac.com/forum/topic/338516-opencore-discussion/?do=findComment&comment=2697758

@Gengik84

This comment has been minimized.

Copy link

@Gengik84 Gengik84 commented Dec 2, 2019

HI
I know it's not a good solution but for now it's the only way for use some smbios.
I avoid KP by simply putting the original kext from S/L/E into EFI's folder.
I hope for a better solution and Apple will also fix some bugs about it.

@vit9696

This comment has been minimized.

Copy link
Contributor

@vit9696 vit9696 commented Dec 4, 2019

Yes, we hope too. I removed the patch from CPUFriend as it does not work anyway, thus closing the issue.

@vit9696 vit9696 closed this Dec 4, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.