Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

CPU Panics - out of my wits...

Killerbob

Killerbob

macrumors 68000
Original poster
Jan 25, 2008
1,946
728
I am about ready to throw my towel in the ring, and I need some help...

Since the first Mojave 10.14.4 Beta, I have been getting CPU Panics on my Mac Pro 2013 (3.5GHz, 6 Core, 64GB RAM, 2xD700). I get them at all times, 1 minute after login, 6 hours after login, in the middle of doing nothing, and in the middle of using several applications. There seems to be no rhyme or reason. Some days I can go all day without a crash…

When I check the Panic Report, I see several instances of Panics, in multiple CPUs, but the BSD process are different from instance to instance. One crash it is expressvpnd, kernel_task, and networksetup. Another crash it is other processes.

I have tried all I know; 1) I booted into safe mode, and back to normal. 2) I reset SMC and PRAM. 3) I checked HW in the HW test, including the internal SSD. 4) I disconnected all peripherals, and only reconnected them one at a time. 5) I ran ONYX to clear out all caches, etc. 6) I turned off all auto starting programs, and only started using them again after a crash when they were not being used.

Basically, I don’t what else to do…

Is a reinstall really the next thing to try?

If so, how should I “migrate”? A) New Mojave install on top of existing setup. B) New Mojave install, and import latest Time Capsule backup? C) New install of everything?

Any suggestions are really appreciated!
 
This is a tough one. Typical source of CPU panic are hardware problems (I know it passed HW test, but something can be flaky) or kexts (kernel extensions). I would consider to do following (backup first!):
1. if possible, pull half of memory and run with 32GB for some time, if still failing, switch to the other half. This would check if it is memory failing. May be the problem is transient and did not show in HW test. Or I would run dedicated memory test (memtest, I think was what I run). Takes forever, especially on 64GB RAM, but in my case identified which stick was bad.
2. Install fresh system. Run and test without moving your stuff back. If stable, migrate from backup using migration assistant. If becomes unstable, it is software, likely one of non-apple kexts. Check what kexts you have which are non apple and if you need them - or even should have them.
3. It could be other hardware transient failure. Touch to identify. Good luck.
 
If you want to isolate to hardware issue, the only option you have to be sure is to get off the beta and do a complete erase and clean install, not migrating anything initially. Do this and proceed to cleanly reinstall all you apps. You can of course drag and drop content like pictures and documents, etc without worrying. If your issue goes away then great. If it does not, I would be leaning towards a bad GPU. the D500s and D700s were part of a recall. Some but not all had issues. If you are running any third party RAM this could also be the issue. So in order I'd say remove any third party RAM, then try reinstall, then if still messed up, look into the GPU issue.
 
I rebooted in Safe Mode last night, and ran Rember all night. The results are encouraging, and according to Rember "All tests passed", which at least means the RAM is OK, and probably the CPU as well.

Also, in that there was no crash for 10 hours (in safe mode), I assume it isn't the OS, but rather a KEXT or a software package that is causing the crashes.

What is a good way to troubleshoot that?
 
Best way would be to go into Console and look for kernel panics. Search for keyword "backtrace" which should point you to particular kexts involved with that kernel panic.
 
Thank you for the info.

So I checked the 3 kernel panic entires, and when looking for the backtrace entries I get;

1:
Backtrace (CPU 10), Frame : Return Address
Backtrace (CPU 6), Frame : Return Address
Kernel Extensions in backtrace:
com.apple.filesystems.hfs.kext(407.200.4)[67B4485D-062C-3D55-A9A5-DE09D4C89972]@0xffffff7f85f0d000->0xffffff7f85f74fff
dependency: com.apple.filesystems.hfs.encodings.kext(1)[1729DC55-062F-3973-ABA7-90F5122608EB]@0xffffff7f835be000
Backtrace (CPU 2), Frame : Return Address

2:
Backtrace (CPU 10), Frame : Return Address
Backtrace (CPU 11), Frame : Return Address
Kernel Extensions in backtrace:
com.apple.driver.AppleIntelCPUPowerManagement(220.0)[4470F19B-64B4-3298-AA5B-AEB12C1661D0]@0xffffff7f9691d000->0xffffff7f96946fff
Backtrace (CPU 5), Frame : Return Address
Kernel Extensions in backtrace:
com.apple.driver.AppleIntelCPUPowerManagement(220.0)[4470F19B-64B4-3298-AA5B-AEB12C1661D0]@0xffffff7f9691d000->0xffffff7f96946fff

3:
Backtrace (CPU 4), Frame : Return Address
Backtrace (CPU 3), Frame : Return Address
Kernel Extensions in backtrace:
com.apple.driver.AppleIntelCPUPowerManagement(220.0)[4470F19B-64B4-3298-AA5B-AEB12C1661D0]@0xffffff7f91f1d000->0xffffff7f91f46fff
Backtrace (CPU 11), Frame : Return Address
Kernel Extensions in backtrace:
com.apple.driver.AppleIntelCPUPowerManagement(220.0)[4470F19B-64B4-3298-AA5B-AEB12C1661D0]@0xffffff7f91f1d000->0xffffff7f91f46fff

What is this telling me?
 
Ehhh...I was expecting to see some third party kexts here based on what you said before. I have something you can try, not sure how effective it will be because I’ve never tried it but it’s something. Go back into safe mode and run kextstat in terminal to see if any of the above kexts are loaded in safe mode. I’m not sure which ones are necessary, but if you notice that one of the ones you listed above is not loaded in safe mode, then use kextload to load it in and see if your system becomes unstable.

Also pro tip, if you use kextstat and don’t want to read the whole list, use something like “kextstat | grep filesystems” to narrow results. In my example “filesystems” was just a word from one of your back traces.
 
I will try that tomorrow. Today, interestingly enough, my Mac Pro ran smoothly without crashes.

In that these are Apple kexts, perhaps the fix is to wait for the official 10.14.4, and see if that is more stable than the betas. If not, then try a reinstall on top of the existing installation - I think that refreshes the Apple files, right?
 
Killerbob said:
I will try that tomorrow. Today, interestingly enough, my Mac Pro ran smoothly without crashes.

In that these are Apple kexts, perhaps the fix is to wait for the official 10.14.4, and see if that is more stable than the betas. If not, then try a reinstall on top of the existing installation - I think that refreshes the Apple files, right?
Click to expand...
Just a question. Do you have a bootable external drive with any OSX older than Mojave?
l presently have my MacPro 6.1 running High Sierra but keep one or more older OSX in USB external bootable drives. If you don't have any I suppose it is easy to install any older OSX (just the OSX system) in for instance a flash pen drive or external drive so that the MacPro can be booted from it. (I apologize if Mojave makes problems to do so, with my High Sierra I can for instance boot an external Yosemite).
My idea is that if your hardware is faulty and causes the panics, it should show problems with whatever OSX you run while if another OSX works fine, it should not be a problem of hardware.
The MacPro will of course run much slower when externally booted but it should not panic because of that.
I apologize if it seems silly (and as already said I have no Mojave experience) but that is what I would do if I had constant crashes with my MacPro as you unfortunately have.
By the way my 64 GB RAM are third part, not the original 12 GB I got from Apple and till now never noticed any inconvenience since any RAM fullfilling Apple's specs should work OK. Apple is not a RAM manufacturer.
Good luck to solve your problems!
 
I have turned on the Mac Pro in Safe Mode, running Rember all night, and it didn't crash, and I have tested several times using Apple Hardware Test, so I do not think it is HW related. I do think it is SW related, and it seems that the kexts causing the Panics, are Apple ones.
 
Yeah that is odd, I think you might be at the point of starting a fresh install and being careful about what is connected to the mac. It is otherwise going to be very hard to diagnose what happened. Try not to have external drives connected during testing.
 
Then today, my Mac Pro has been running for 28 hours straight, and not a hiccup...
 
Seeing as it seems that when my Mac Pro has been running for 30 min. or more, it does not get into a Panic, is it possible that it is indeed a hardware issue, but more like a loose connection, and when the Mac is warm, the loose connection is not present?
 
I could be wrong, but I believe that most kernel panics result from either:
- problem with hardware
or
- probably when certain software encounters an error while trying to interact with certain hardware.
 
Hence, could it be that there is a "loose connection" to for instance a stick of RAM, and as long the Mac Pro is cold, it is present. Then when it gets warm, the connection is solid?
 
I updated from 10.14.3 to 10.14.4 on my 2013 Mac Pro (2.7GHz, 12 Core, 64GB RAM, 2xD700). During the upgrade I got a kernel panic. It continued to restart, attempt to continue install, panic.. etc.

So I attempted a full install of 10.14.4. Same issue persisted. After some time on the phone with Apple Support they told me to use High Sierra until they can resolve the issue. They believe it's software related and not hardware after running some test.

I'm now on High Sierra and have zero issues. I don't keep full TimeMachine system Backups so I wasn't able to do a full recovery. I think going forward I will. I'd would of liked to rollback to 10.14.3 but cannot.

I'm not seeing a ton of coverage about this issue. I'm hoping a fix comes soon. The support people said they'd alert the engineers of the issue I reported.
 
jsonarray said:
After some time on the phone with Apple Support they told me to use High Sierra until they can resolve the issue. They believe it's software related and not hardware after running some test.
Click to expand...

Thx! Good to know I am not alone, and that it may be OS related. That will be fixed by Apple - even if it has low priority. I just can’t think of having to reinstall everything right now...
 
jsonarray wrote:
"I'm now on High Sierra and have zero issues. I don't keep full TimeMachine system Backups so I wasn't able to do a full recovery. I think going forward I will. I'd would of liked to rollback to 10.14.3 but cannot."

If you want to be free of such headaches in the future, DO NOT rely on Time Machine backups.

Instead, use either CarbonCopyCloner or SuperDuper to create a BOOTABLE cloned backup of your drive BEFORE you do a critical upgrade.

IF you had a CCC/SD cloned backup, then "getting back to where you once belonged" would be little more than child's play.
It's THAT easy -- when you have the right setup.
 
  • Like
Reactions: Riwam
Unfortunately a lot of us are dependent on TM. I do have CCC, but I do not use it for a full backup of my macOS drive - perhaps I should.
 
Fishrrman said:
jsonarray wrote:
"I'm now on High Sierra and have zero issues. I don't keep full TimeMachine system Backups so I wasn't able to do a full recovery. I think going forward I will. I'd would of liked to rollback to 10.14.3 but cannot."

If you want to be free of such headaches in the future, DO NOT rely on Time Machine backups.

Instead, use either CarbonCopyCloner or SuperDuper to create a BOOTABLE cloned backup of your drive BEFORE you do a critical upgrade.

IF you had a CCC/SD cloned backup, then "getting back to where you once belonged" would be little more than child's play.
It's THAT easy -- when you have the right setup.
Click to expand...

I wasn't able to even boot back into the OS after this update. Just kept causing kernel panic. With TimeMachine I can do a restore from that from the recovery console.

How does your suggestion work in this scenario differently than what TimeMachine would do? Can it be used even if the OS cannot be booted into?
 
Fishrrman said:
jsonarray wrote:
"I'm now on High Sierra and have zero issues. I don't keep full TimeMachine system Backups so I wasn't able to do a full recovery. I think going forward I will. I'd would of liked to rollback to 10.14.3 but cannot."

If you want to be free of such headaches in the future, DO NOT rely on Time Machine backups.

Instead, use either CarbonCopyCloner or SuperDuper to create a BOOTABLE cloned backup of your drive BEFORE you do a critical upgrade.

IF you had a CCC/SD cloned backup, then "getting back to where you once belonged" would be little more than child's play.
It's THAT easy -- when you have the right setup.
Click to expand...
Does CCC ever go on sale?
 
jsonarray said:
I updated from 10.14.3 to 10.14.4 on my 2013 Mac Pro (2.7GHz, 12 Core, 64GB RAM, 2xD700). During the upgrade I got a kernel panic. It continued to restart, attempt to continue install, panic.. etc.

So I attempted a full install of 10.14.4. Same issue persisted. After some time on the phone with Apple Support they told me to use High Sierra until they can resolve the issue. They believe it's software related and not hardware after running some test.

I'm now on High Sierra and have zero issues. I don't keep full TimeMachine system Backups so I wasn't able to do a full recovery. I think going forward I will. I'd would of liked to rollback to 10.14.3 but cannot.

I'm not seeing a ton of coverage about this issue. I'm hoping a fix comes soon. The support people said they'd alert the engineers of the issue I reported.
Click to expand...

Exact same thing happened to my MacPro 2013 (3.7GHz Quad-core E6) when updating from 10.14.3 to 10.14.4. Constant Kernel Panics (CPU): "NMIPI for unresponsive processor". Rolled back to 10.14.3 via TimeMachine. Runs OK on 10.14.3. Also ran OK when Safe Booting on 10.14.4; but very slowly.

Hopefully Apple will acknowledge (and fix) this issue soon.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back