trx40 [Discussion] - TRX40 Bare Metal - Vanilla Patches

[Ploddles]

On 8/6/2021 at 6:37 AM, fabiosun said:

@Ploddlesin your config latest kernel patch is not needed for trx40 chipset:

 

 

And usually also in system that needs it (X370,x470,x570 and lower) , it produces worst GPU benchmark..

maybe you can try to skip it

 

Well spotted, I thought I had taken that one out but obviously I missed it. I have now deleted it and everything is still working. Thanks.

[fabiosun]

4 hours ago, Ploddles said:

 

Well spotted, I thought I had taken that one out but obviously I missed it. I have now deleted it and everything is still working. Thanks.

Updated your EFI in your OP section 🙂

thanks

 

[Arrakis]

Successful installion of Monterey from BigSur 11.5.1.

Update is scrolling without problems.

I did the installation from the EFi posted above here with two changes in the kexts (AirportItlwm-Mont.kext, BlueToolFixup.kext) and delete of the IntelBluetoothInjector.kext

WiFI and Bluetooth operational.

 

But I have the same problem as when I went from 11.3.1 to 11.4

No more ethernet connection.

The I210 ethernet modules are more recognized while the kexts are included in the OS so in theory recognized natively.

 

Edited : My config.plist Arrakis Monterey Open Core 0.7.2.zip 

 

Edited August 7, 2021 by Arrakis

[carlo_67]

Il 4/8/2021 at 07:06, iGPU ha scritto:

 

@fabiosun ,

 

I've begun having sudden crashes since using the new AMD Power Gadget (only tested in Monterey ß4).

 

The computer locks up after about 3 or 4 minutes or will even spontaneously re-boot from the login screen. It happens whether or not SMCAMDProcessor is enabled, so it would seem to be a problem with either AMDRyzenCPUPowerManagement or the Power Gadget app. I did not see this with the older version.

 

I now have the new kexts disabled and the computer has been stably running for hours, just like before.

Have you noticed any problems?

 

I personally have not found any problems, even changing in Misc / Security / ExposeSensitiveData from 0 to 8, it also detects my motherboard,

 

@fabiosun la mia carretta 🤬

[iGPU]

3 hours ago, carlo_67 said:

I personally have not found any problems, even changing in Misc / Security / ExposeSensitiveData from 0 to 8, it also detects my motherboard,

 

@fabiosun la mia carretta 🤬

 

This is interesting. I again tried and had crashes. I thought something might be amiss with Mß4, so I re-installed it through Recovery. Once more, AMD Power tools would crash the computer. I tried disabling various kexts and had no success. I think it is the app that is causing the crashes. I did not see any crashes with the older version from last fall with all things in the config file and BIOS being the same.

 

I've had ExposeSensitiveData set to 14 for some time (previously having had it set at 3 or 8). I'll give the 8 setting a go.

 

[Arrakis]

13 hours ago, carlo_67 said:

I personally have not found any problems, even changing in Misc / Security / ExposeSensitiveData from 0 to 8, it also detects my motherboard,

 

@fabiosun la mia carretta 🤬

I have the same problem as @iGPUwith the new version SMCAMDPProcessor.kext making my system unstable

and moreover I no longer have the temperature near the CPU in iStat Menus (I use this application because I have GPU temperatures and other constants ...)

I reverted to the previous version. The ExposeSensitiveData value is indeed 8 to recognize my motherboard.

Edited August 8, 2021 by Arrakis

[fabiosun]

@Arrakis @iGPU

luckily I don't have this problem with the new @trulyspinach app.

Instead, these days I discovered a serious problem that I did not know I had In the past, installing windows from scratch was quite complex on this machine

 

It was mandatory to disconnect all satas to be successful

In UEFI, then even worse

I wiped my Windows 10 disk to try ...

Moral of the story ...

I no longer have Windows and cannot install it in any way. :)

[Ploddles]

5 hours ago, fabiosun said:

@Arrakis @iGPU

luckily I don't have this problem with the new @trulyspinach app.

Instead, these days I discovered a serious problem that I did not know I had In the past, installing windows from scratch was quite complex on this machine

 

It was mandatory to disconnect all satas to be successful

In UEFI, then even worse

I wiped my Windows 10 disk to try ...

Moral of the story ...

I no longer have Windows and cannot install it in any way. 🙂

 

The trick to install Windows is to do it to an NVMe in slot 1 (closest to the CPU) , preferably after removing any other NVMe / Sata drives. If it is installed anywhere else them when it does a Windows Update it can/will wipe your OC EFI and replace it with a Windows/Microsoft EFI. It will also change your BIOS options to boot from the Windows disc once the install has finished, so afterwards you need to change the boot options and change it back to your OC EFI once you have put them back in.

[fabiosun]

2 minutes ago, Ploddles said:

 

The trick to install Windows is to do it to an NVMe in slot 1 (closest to the CPU) , preferably after removing any other NVMe / Sata drives. If it is installed anywhere else them when it does a Windows Update it can/will wipe your OC EFI and replace it with a Windows/Microsoft EFI. It will also change your BIOS options to boot from the Windows disc once the install has finished, so afterwards you need to change the boot options and change it back to your OC EFI once you have put them back in.

thank you for your advice

I did also with only my "old" 960pro Nvme disk on first nvme slot (near my cpu)

The best result is in UEFI configuration.

Windows flag

A blu empty screen

PC off (after 30s)

 

The best result in CSM (on)

It let me to chose disk to install (960 pro nvme)

format it the it can't install on it with an alphanumeric string error print out

 

I will try again but it is very weird

 

[Rocket88]

22 hours ago, Arrakis said:

Successful installion of Monterey from BigSur 11.5.1.

Update is scrolling without problems.

I did the installation from the EFi posted above here with two changes in the kexts (AirportItlwm-Mont.kext, BlueToolFixup.kext) and delete of the IntelBluetoothInjector.kext

WiFI and Bluetooth operational.

 

 

No more ethernet connection.

The I210 ethernet modules are more recognized while the kexts are included in the OS so in theory recognized natively.

 

I had to make a change to my Kernel patches to get ethernet to work under Monterey. I changed MaxKernel to 21.99.99.

[Ploddles]

@fabiosun I would "Load Optimised Defaults" first and then turn CSM OFF but don't change anything else. Then try and install Windows. If you still can't install Windows, only then turn CSM ON and try again.  Afterwards you should be able to change your BIOS settings back to what they need to be for macOS.

[iGPU]

2 hours ago, Rocket88 said:

I had to make a change to my Kernel patches to get ethernet to work under Monterey. I changed MaxKernel to 21.99.99.

 

Your reminder to update the MaxKernel is a good one, as this step is often over-looked when updating to a new macOS.

 

However, under Monterey, the problem isn't with Aquantia ethernet ports, but with I210 and I211 ports. These ports seem to be no longer properly responding to the kexts which were working under Big Sur.

 

I've literally spent a dozen hours or more working on the I211 problem (discussed here and here), and so far have had no success.

 

[Arrakis]

I may have to buy an Aquantia card ...😒

I'm not very enthusiastic about this idea ...😒

[iGPU]

13 hours ago, Arrakis said:

I have the same problem as @iGPUwith the new version SMCAMDPProcessor.kext making my system unstable

and moreover I no longer have the temperature near the CPU in iStat Menus (I use this application because I have GPU temperatures and other constants ...)

I reverted to the previous version. The ExposeSensitiveData value is indeed 8 to recognize my motherboard.

 

After several more crashes (everything locks up, freezing mouse–––the computer if left alone will re-boot; if impatient, a forced shutdown is required), I think I've isolated the problem...

 

Each new kext seems okay. I can have a stable computer if one or both are running. I can even run the AMD Power Gadget app and the computer is stable. The problem is if the app has the feature to "Launch in menu bar at login" enabled (StartAtLogin in the plist file discussed below) through it's menu, shown below.

 

 

Once this is enabled and the computer re-booted, the menu is no longer accessible to disable and now the computer locks up in less than 60 seconds after each re-boot due to the auto-login feature of the AMD Power Gadget app.

 

I had to boot into Big Sur (or use another EFI where the AMD Spinach kexts were not enabled) to remove the AMD Spinach plist file. This file is located in the User/Library/Preferences folder as shown below. You can either delete it or better is to edit and set "startAtLogin" to "NO" (and while you're at it, also set "startAtLoginAsked' to "YES" to avoid a pop-up window shown in Spoiler below).

 

Spoiler

 

 

 

In looking more deeply in the CrashReport log, one can find the following (an excerpt from that log file). I'll provide this info to the developer and maybe they can provide a fix.

 

 

The bottom line, in its present state, the AMD Power Gadget system is usable (at least for me), if the Auto-Login feature of the AMD Gadget app is not activated.

 

ADDENDUM:

 

I found out through playing with the wtf.spinach.AMD-Power-Gadget.plist file that the menu is inactivated if "StartAtLoginAsked" is enabled and "statusbarenabled" is set to "YES". I'm leaving "statusbarenabled" set to "NO" in order to have access to the AMD Power Gadget app's menu bar.

 

After writing the above, the system began locking up even if only the AMDRyzenCPUPowerManagement.kext was enabled (and no AMD Power Gadget app was running). This is after a morning of the computer running fine with both kexts enabled but no app running. It was only after enabling auto log-in did all the troubles start once more. Odd. I've presently disabled both kext files and have contacted the developer.

 

 

Spoiler

Menu is inaccessible if "statusbarenabled" set to "YES"!

 

 

 

 

 

[fabiosun]

19 hours ago, Ploddles said:

@fabiosun I would "Load Optimised Defaults" first and then turn CSM OFF but don't change anything else. Then try and install Windows. If you still can't install Windows, only then turn CSM ON and try again.  Afterwards you should be able to change your BIOS settings back to what they need to be for macOS.

No way @Ploddles

I reverted to an old bios and it is a bit better but always fails...

then...I have used my wife PC to first installation step 🙂

I think it is a combination of new gpu and bios to this very weird behaviour 

I have less problem to install any OSX in this pc rig...

[fabiosun]

@iGPU

 

I have enabled it to test..I rebooted

but app does not start in automatic , and to have it running I have to launch via usual icon

..so another bug here

I can't see in location you said any AMDSPinach plist

so maybe other problem here..

I am testing in Monterey b4

 

EDIT:

SIP Enabled here

 

 

EDIT2:

with sip disabled it starts automatically 

Let see if it hangs 

 

EDIT3:

Does not hang here and  I can see option to disable automatic app running

 

 

Edited August 9, 2021 by fabiosun
EDIT3:

[iGPU]

On 8/9/2021 at 7:56 AM, fabiosun said:

@iGPU

 

I have enabled it to test..I rebooted

but app does not start in automatic , and to have it running I have to launch via usual icon

..so another bug here

I can't see in location you said any AMDSPinach plist

so maybe other problem here..

I am testing in Monterey b4

 

EDIT:

SIP Enabled here

 

 

EDIT2:

with sip disabled it starts automatically 

Let see if it hangs 

 

EDIT3:

Does not hang here and  I can see option to disable automatic app running

 

 

 

In discussion with the developer, the freezing (hang) problem is solved by simply enabling the DummyPowerManagement quirk (inside Kernel/Emulate as shown in Spoiler).

 

Spoiler

 

 

I'm now of the opinion that, from now on, we should keep this quirk enabled for all TRX40 mobos. (I don't see any downsides to this view.) 

 

There are still issues with the AMD Power Gadget (AMDPG) app having the submenus disappear if "statusbarenabled" Is enabled (plist file location shown below). And I see this menu problem whether SIP is enabled or disabled. This bug needs to be fixed.

 

But at least now there seems to be no further computer lock-ups when using the AMDPG app!

 

 

[fabiosun]

Mmmh you know my way to think about not mandatory stuff..

I am trying automatic load of amd power gadget and I have no problem

i have to check  if I have enabled dummy power management 

usually I have not enabled it so I think no

For x570 is mandatory and also for @Ploddles and @Arrakisi think

but if I have understood well

arrakis has some hangs with new app..

I will try tomorrow menu disappearing bug

but I think I have not that problem🤞

@iGPU

by the way

this week I will receive icegiant elite cooler..

I hope it will better or similar of my actual AIO

 

 

[iGPU]

8 hours ago, fabiosun said:

Mmmh you know my way to think about not mandatory stuff..

I am trying automatic load of amd power gadget and I have no problem

i have to check  if I have enabled dummy power management 

usually I have not enabled it so I think no

For x570 is mandatory and also for @Ploddles and @Arrakisi think

but if I have understood well

arrakis has some hangs with new app..

I will try tomorrow menu disappearing bug

but I think I have not that problem🤞

@iGPU

by the way

this week I will receive icegiant elite cooler..

I hope it will better or similar of my actual AIO

 

 

 

I do not view recommended settings as mandatory, but beneficial. That is, I view recommended settings as a way of allowing most users to have the most trouble-free functionality as possible: the basic common denominator (so to speak) for all. 

 

It's like learning to drive a car: there are many steps and rules that are not always obvious at the start, but end up benefiting most people, allowing for an overall enjoyable experience. Those who later wish to race cars or do stunt tricks will abandon those basic rules, stretching the limits by acquiring more more esoteric knowledge.

 

Similarly, if we can get a generalized config file that is useful to as many users as possible, giving them an easy, enjoyable experience, then we've accomplished something good for most TRX40 users.

 

So far, the only unique settings, aside from AIC and specialized SSDT files, are the MmioWhitelist (which is unique for each mobo/manufacturer, and even those have much in common with each other) and Memory configurations (which also can be simplified into 2 or 3 basic settings). Later, those who wish to streamline their config files, can trim the config file, and erase portions, to their heart's content.

 

Using the DummyPowerManagement as an example, while I don't need this Quirk to boot, it now seems to allow me to use the AMDPG app whenever I wish, without having the computer freeze up. Meanwhile, GB mobo owners do need this Quirk. So why not include it as part of a basic, recommended config file for TRX40 users?  

 

 

[fabiosun]

3 hours ago, iGPU said:

 

I do not view recommended settings as mandatory, but beneficial. That is, I view recommended settings as a way of allowing the most users, to have the most trouble-free functionality as possible: the basic common denominator (so to speak) for all. 

 

It's like learning to drive a car: there are many steps and rules that are not always obvious at the start, but end up benefiting most people, allowing for an overall enjoyable experience. Those who later wish to race cars or do stunt tricks will abandon those basic rules and gain more more esoteric knowledge.

 

Similarly, if we can get a generalized config file that is useful to as many users as possible, giving them an easy, enjoyable experience, then we've accomplished something good for most TRX40 users.

 

So far, the only unique settings, aside from AIC and specialized SSDT files, are the MmioWhitelist (which is unique for each mobo/manufacturer, and even those have much in common with each other) and Memory configurations (which also can be simplified into 2 or 3 basic settings). Later, those who wish to streamline their config files, can trim the config file, and erase portions, to their heart's content.

 

Using the DummyPowerManagement as an example, while I don't need this Quirk to boot, it now seems to allow me to use the AMDPG app whenever I wish, without having the computer freeze up. Meanwhile, GB mobo owners do need this Quirk. So why not include it as part of a basic, recommended config file for TRX40 users?  

 

 

 

I agree with a lot of what you write, on the last part I just ask myself more questions:

 

1) what has changed in the app and in the kext to be now necessary to activate that quirk (which is quite important regarding power management)?

2) an old question of mine was the correlation between that quirk, the bios versions (MSR) and the patches we have to use..but I couldn't find an answer

 

However, on this platform the story is still to be written and in part we have all contributed to making some parts clearer.

 

I also have another question / curiosity

Why does the activation of the dummypower management quirk seem unnecessary here?

 

I try to explain better, under what conditions the use or not of a kernel patches that seems "unnecessary" to start the system or a quirk like the one mentioned could create a problem, but maybe by activating a simple autologin you can make malfunctioning a system that was basically stable?

[iGPU]

52 minutes ago, fabiosun said:

 

I agree with a lot of what you write, on the last part I just ask myself more questions:

 

1) what has changed in the app and in the kext to be now necessary to activate that quirk (which is quite important regarding power management)?

2) an old question of mine was the correlation between that quirk, the bios versions (MSR) and the patches we have to use..but I couldn't find an answer

 

However, on this platform the story is still to be written and in part we have all contributed to making some parts clearer.

 

I also have another question / curiosity

Why does the activation of the dummypower management quirk seem unnecessary here?

 

I try to explain better, under what conditions the use or not of a kernel patches that seems "unnecessary" to start the system or a quirk like the one mentioned could create a problem, but maybe by activating a simple autologin you can make malfunctioning a system that was basically stable?

 

As to why the DummyPowerManagement quirk might be important:

 

The developer asked me to look at the Kernel panic log when the computer froze. I did so and found the following (Spoiler; small extract from end of panic log).

 

Spoiler

MC4yLjAuMSkgICAgICAgICAgICAgICAgPERBMEVBREQyLTc4QUMtMzkxRS05QTYyLUYyQkNCRTM3QjQyRT4gIC9TeXN0ZW0vTGlicmFyeS9FeHRlbnNpb25zL3B0aHJlYWQua2V4dC9Db250ZW50cy9NYWNPUy9wdGhyZWFkCg==","notes":"Stackshot source : disk","macOSPanicString":"panic(cpu 0 caller 0xffffff801821efa7): CPU 63 has no HPET assigned to it\nPanicked task 0xffffff8829fef6a0: 296 threads: pid 0: kernel_task\nBacktrace (CPU 0), panicked thread: 0xffffff882a5816e0, Frame : Return Address\n0xffffffd1d514bc70 : 0xffffff8016ca34fd \n0xffffffd1d514bcc0 : 0xffffff8016dfcc35 \n0xffffffd1d514bd00 : 0xffffff8016dec603 \n0xffffffd1d514bd50 : 0xffffff8016c42a60 \n0xffffffd1d514bd70 : 0xffffff8016ca38cd \n0xffffffd1d514be90 : 0xffffff8016ca3086 \n0xffffffd1d514bef0 : 0xffffff8017519aa9 \n0xffffffd1d514bf60 : 0xffffff801821efa7 \n0xffffffd1d514bfa0 : 0xffffff8016c4218e \n      Kernel Extensions in backtrace:\n         com.apple.driver.AppleIntelCPUPowerManagement(222.0)[C976299C-7A0C-3F2C-8A3A-524D89DEF7C6]@0xffffff8018219000->0xffffff8018235fff\n\nProcess name corresponding to current thread (0xffffff882a5816e0): kernel_task\nBoot args: agdpmod=pikera \n\nMac OS version:\n21A5294g\n\nKernel version:\nDarwin Kernel Version 21.0.0: Thu Jul 22 19:52:31 PDT 2021; root:xnu-8019.0.72.141.3~2\/RELEASE_X86_64\nKernel UUID: 43AA07BC-5319-3B90-A977-9050C6C2F8D7\nKernelCache slide: 0x0000000016a00000\nKernelCache base:  0xffffff8016c00000\nKernel slide:      0x0000000016a10000\nKernel text base:  0xffffff8016c10000\n__HIB  text base: 0xffffff8016b00000\nSystem model name: MacPro7,1 (Mac-27AD2F918AE68F61)\nSystem shutdown begun: NO\nPanic diags file unavailable, panic occurred prior to initialization\nHibernation exit count: 0\n\nSystem uptime in nanoseconds: 90982254187\nLast Sleep:           absolute           base_tsc          base_nano\n  Uptime  : 0x000000152ef70980\n  Sleep   : 0x0000000000000000 0x0000000000000000 0x0000000000000000\n  Wake    : 0x0000000000000000 0x0000002478dab456 0x0000000000000000\n\n\n"}

 

 

To highlight the important part:

 

 "Kernel Extensions in backtrace:\n com.apple.driver.AppleIntelCPUPowerManagement (222.0)[C976299C-7A0C-3F2C-8A3A-524D89DEF7C6]@0xffffff8018219000->0xffffff8018235fff\n\nProcess name corresponding to current thread (0xffffff882a5816e0)".

 

He suggested that the problem with the freeze was probably due to macOS's Intel PM (power management), which would require the DummyPowerManagement quirk to be enabled to fix the kernel panic. I tried this and found that by enabling this quirk, the crash (kernel panic) stopped.

 

 

[fabiosun]

 

[Ploddles]

So who is going to try Monterey Beta 5 (21A5304g) first? 🙂

[Ploddles]

Anybody else lost connection to their iPad after the Big Sur update to 11.5.2?

 

It shows up in the sidebar but is just displaying a blank screen now so I can  no longer sync etc.

 

 

 

Edited August 11, 2021 by Ploddles

[fabiosun]

3 hours ago, Ploddles said:

So who is going to try Monterey Beta 5 (21A5304g) first? 🙂

 

Edited August 12, 2021 by fabiosun
updated :)