WhateverGreen + 6900 XT = crash

[atanvarno]

I am trying to add WEG 1.5.5 into the EFI for my build with 6900 XT, still based on OC 0.7.5.
Without WEG, build works more of less fine but I noticed some occasional lagging with UI redraws and was testing will WEG sort it out.

With WEG, it sometimes starts up fine and sometimes it fails to transfer into full graphics login mode with this crash report:

panic(cpu 4 caller 0xffffff7f9d143aa1): "GDDR6 Long Training Failed !!!
"@/System/Volumes/Data/SWE/macOS/BuildRoots/9dc5c14c79/Library/Caches/com.apple.xbs/Sources/GPUDriversAMD/GPUDriversAMD-4.6.21/Common/IONDRV/ATI/AMDUniversalFramebuffer/AMDUniversalFramebuffer/Controller/Controllers/AmdRadeonController.cpp:1812
Backtrace (CPU 4), Frame : Return Address
0xffffffa0b47639f0 : 0xffffff800e48c26d mach_kernel : _handle_debugger_trap + 0x3fd
0xffffffa0b4763a40 : 0xffffff800e5d3993 mach_kernel : _kdp_i386_trap + 0x143
0xffffffa0b4763a80 : 0xffffff800e5c3f8a mach_kernel : _kernel_trap + 0x55a
0xffffffa0b4763ad0 : 0xffffff800e430a2f mach_kernel : _return_from_trap + 0xff
0xffffffa0b4763af0 : 0xffffff800e48ba8d mach_kernel : _DebuggerTrapWithState + 0xad
0xffffffa0b4763c10 : 0xffffff800e48bd83 mach_kernel : _panic_trap_to_debugger + 0x273
0xffffffa0b4763c80 : 0xffffff800ec9c7ba mach_kernel : _panic + 0x54
0xffffffa0b4763cf0 : 0xffffff7f9d143aa1 com.apple.kext.AMDRadeonX6000Framebuffer : __ZN34AMDRadeonX6000_AmdRadeonController10doGPUPanicEPKcz + 0x171
0xffffffa0b4763d80 : 0xffffff7f9d0fe328 com.apple.kext.AMDRadeonX6000Framebuffer : __ZN39AMDRadeonX6000_AmdRadeonControllerNavi219doGddr6LongTrainingEv + 0x7c
0xffffffa0b4763da0 : 0xffffff7f9d0fe184 com.apple.kext.AMDRadeonX6000Framebuffer : __ZN39AMDRadeonX6000_AmdRadeonControllerNavi25startEP9IOService + 0x90
0xffffffa0b4763dd0 : 0xffffff800ebc484b mach_kernel : __ZN9IOService14startCandidateEPS_ + 0xeb
0xffffffa0b4763e40 : 0xffffff800ebc43de mach_kernel : __ZN9IOService15probeCandidatesEP12OSOrderedSet + 0xcce
0xffffffa0b4763ef0 : 0xffffff800ebc357f mach_kernel : __ZN9IOService14doServiceMatchEj + 0x30f
0xffffffa0b4763f50 : 0xffffff800ebc62f6 mach_kernel : __ZN15_IOConfigThread4mainEPvi + 0x186
0xffffffa0b4763fa0 : 0xffffff800e43013e mach_kernel : _call_continuation + 0x2e
      Kernel Extensions in backtrace:
         com.apple.kext.AMDRadeonX6000Framebuffer(4.0.6)[C3F2639B-A637-320C-AE35-75B267A45AC6]@0xffffff7f9d0fa000->0xffffff7f9d374fff
            dependency: com.apple.AppleGraphicsDeviceControl(6.3.5)[2943A57C-11EA-349E-ABF5-0AC7DED914EA]@0xffffff7fa7b50000->0xffffff7fa7b53fff
            dependency: com.apple.iokit.IOACPIFamily(1.4)[5B13B648-2033-3C43-9806-CC47FED932BC]@0xffffff8010a44000->0xffffff8010a45fff
            dependency: com.apple.iokit.IOGraphicsFamily(585.2)[18F5DD47-7688-37AB-A74A-5AF8B45A4E0F]@0xffffff7fa933c000->0xffffff7fa936afff
            dependency: com.apple.iokit.IOPCIFamily(2.9)[C14212B3-F922-3073-B54C-9DE243057AF4]@0xffffff8010f6b000->0xffffff8010f93fff

Process name corresponding to current thread: kernel_task
Boot args: keepsyms=1 agdpmod=ignore brcmfx-driver=2 brcmfx-country=RS chunklist-security-epoch=0 -chunklist-no-rev2-dev chunklist-security-epoch=0 -chunklist-no-rev2-dev

Mac OS version:
20G314

Kernel version:
Darwin Kernel Version 20.6.0: Wed Nov 10 22:23:07 PST 2021; root:xnu-7195.141.14~1/RELEASE_X86_64
Kernel UUID: C67468D8-72C1-36BC-A0DA-CD5FA852AD32
KernelCache slide: 0x000000000e200000
KernelCache base:  0xffffff800e400000
Kernel slide:      0x000000000e210000
Kernel text base:  0xffffff800e410000
__HIB  text base: 0xffffff800e300000
System model name: MacPro7,1 (Mac-27AD2F918AE68F61)
System shutdown begun: NO
Panic diags file available: YES (0x0)
Hibernation exit count: 0

System uptime in nanoseconds: 27571889971
Last Sleep:           absolute           base_tsc          base_nano
  Uptime  : 0x000000066b69aa91
  Sleep   : 0x0000000000000000 0x0000000000000000 0x0000000000000000
  Wake    : 0x0000000000000000 0x000000177c3a2061 0x0000000000000000

Any ideas are welcome, what to look into.

 

[Edhawk]

Aren't you supposed to use the agdpmod=pikera boot argument with an RX 6900/6800 or 6600 dGPU?

Try swapping agdpmod=ignore for agdpmod=pikera and see if that changes the behaviour of the dGPU.

 

[Shaneee]

If either of the boot args don't fix it I'd suggest opening a ticket on the acidanthera bug tracker.

 

[atanvarno]

Aren't you supposed to use the agdpmod=pikera boot argument with an RX 6900/6800 or 6600 dGPU?
Try swapping agdpmod=ignore for agdpmod=pikera and see if that changes the behaviour of the dGPU.

pikera is only for Radeon 5000 series, I believe.

 

[atanvarno]

Additional issue I encountered recently (most likely after 11.6.2 update) is significant lag in UI redraws, to the point where macOS WindowServer crashes (watchdogd kills it after being non responsive for 2mins)

Chasing all this through logs I found a lot of entries for this launcher helper for AMD Power Gadget app:

com.apple.xpc.launchd[1] (wtf.spinach.APGLaunchHelper[675]): Could not find and/or execute program specified by service: 155: Refusing to execute/trust quarantined program/file: wtf.spinach.APGLaunchHelper
com.apple.xpc.launchd[1] (wtf.spinach.APGLaunchHelper[675]): Service exited with abnormal code: 78
com.apple.xpc.launchd[1] (wtf.spinach.APGLaunchHelper): Service only ran for 0 seconds. Pushing respawn out by 10 seconds.

I uninstalled it completely – including app, launcher helper and prefs file – and also turned off AMDSMC & Radeon kexts from OpenCore.
And here it is – 2h in, no lag, no lockups, no crashes All this with WEG active and agdpmod=ignore
Will leave it like this for few days and then re-enable the kexts one by one.

Not sure what happened — I've been using AMDSMCProcessor for over a year and latest PowerGadget app for months. Thus my assumption is that 11.6.2 update put the APGLaunchHelper under quarantine for some reason and the process repeatedly attempted to restart.

 

[atanvarno]

Bleh, worked until it went to sleep then started lagging again. This is playing with me

Anw, back to non-WEG, which works.

 

[Edhawk]

agdpmod=pikera is also used with the 6000 series cards. Have you tried using it in place of the agdpmod=ignore boot argument?

 

[atanvarno]

agdpmod=pikera is also used with the 6000 series cards. Have you tried using it in place of the agdpmod=ignore boot argument?

Thanks for the nudge to try it - I've been running it with pikera argument for half a day, across few sleep/wake cycles and it holds up.

 

[madazzle]

I've also had some weird issues with my RX 6900 XT on the newer versions of OpenCore. I can't even come through the MacOS Installation or when I had my MacOS Installation use it for more than 1-2 minutes. It always revolved around the Windowserver but on lower Versions of OpenCore particularly 0.7.5 I had zero literally zero problems.

 

[madazzle]

I've also had some weird issues with my RX 6900 XT on the newer versions of OpenCore. I can't even come through the MacOS Installation or when I had my MacOS Installation use it for more than 1-2 minutes. It always revolved around the Windowserver but on lower Versions of OpenCore particularly 0.7.5 I had zero literally zero problems.

I currently can't get it to run even with the tips and tricks the people here suggested if pikera, ignore or amdprocessor kexts disabled nothing works for me till now-

 

[Middleman]

I currently can't get it to run even with the tips and tricks the people here suggested if pikera, ignore or amdprocessor kexts disabled nothing works for me till now-

What system do you have and which OS are you trying to run?