High Sierra Random Freezing

[thefsb]

Yesterday I could work all day without trouble.

This morning I turned Backblaze Backup on again and the system froze with the monitors on while I was out. I was rather expecting that. If I find time I'll try creating synthetic filesystem load and conflict.

 

[Thireus]

Do you have a KP dump to share @thefsb?

Have a look in /Library/Logs/DiagnosticReports and open file names starting with "Kernel_", that's where you'll find your Kernel dumps. Post them all here so we can check if the dumps are consistant.

So far I'm at 1d 11h uptime. My last Kernel dump didn't contain any kernel extension in backtrace.

As for now it appears that my BIOS was not patched in full, which could write to MSR 0xE2 and thus lead to freezing! So I've just used pike's patch to prevent this, let's see how it turns out this time.

Edit: Analysis correction.

 

[thefsb]

Yes I do! Thanks for pointing out that location. All the diag logs starting with Kernel are all attached.

I'm not sure these are coincident with the freezes. The system froze yesterday morning but there's no corresponding log.

Here's the start of an example

Sun Jan 28 11:44:25 2018

Event:               GPU Reset
Date/Time:           Sun Jan 28 11:44:25 2018
Application:         plugin-container
Path:
Tailspin:            /Library/Logs/DiagnosticReports/gpuRestart2018-01-28-114425.tailspin
GPUSubmission Trace ID: 0
OS Version:          Mac OS X Version 10.13.3 (Build 17D47)
Graphics Hardware:   Intel HD Graphics 4600
Signature:           8013

Report Data:

===========================================================================
                         Intel GPU Hang Summary
                              version 1.0.0
===========================================================================

-- Summary of Status for ALL Rings --
MAIN Ring       is hung in a batch buffer        <-- ROOT CAUSE!
MEDIA Ring      is idle
BLT Ring        is waiting on a semaphore
VEBOX Ring      is idle

 

[thefsb]

I checked and can't see anything in /Library/Logs/DiagnosticReports with a modification time that could correspond to yesterday's freeze.

 

[Thireus]

Well, from what I can tell and it's a bit obvious now, your random freezes are related to your HD Graphics 4600 Intel Graphics not functioning properly.

I can't help you any further now, but at least you know where to look at now!

 

[thefsb]

Like I said above, the timing of these GPU reset events does not coincide at all with the freezes. I don't think they are related.

Nearly all the GPU resets were on different days from freezes.

 

[Thireus]

I think it's because the reports get generated upon next boot. My Kernel dump are dated when I reboot the system after a crash. i.e. if the system was crashed at 1am and I notice it and restart it at 8am, my Kernel dump date will be 8am, not 1am.

Let me know if that helps.

 

[thefsb]

The system only froze once when I was not there to witness it. Then I power it off and on to resume work. These are the
BOOT_TIMEs of those incidents in the last couple of weeks:

Jan 22 15:43
Jan 22 18:10
Jan 25 14:45
Jan 26 12:15
Jan 27 11:56
Jan 28 14:58
Jan 31 13:03

These are the GPU resets:

Jan 17 18:18
Jan 17 18:18
Jan 18 17:05
Jan 18 17:20
Jan 18 17:20
Jan 18 17:22
Jan 18 17:22
Jan 21 11:57
Jan 28 11:44
Jan 28 11:44
Jan 29 13:47

And I can look up the datetimes of the GPU resets in system logs and see the usual chatter before an after. It seems these GPU resets are pretty benign.

 

[Thireus]

I agree, no correlation then.

You don't have any NVidia driver installed and loaded, do you?

Could you check which System Preferences >Energy Saver settings you're using please?

I've just tried to disable "hard disk to sleep when possible" to see if that helps on my side. Unfortunately my system freezes happen after ~1d of uptime, so really hard to diagnose.

 

[thefsb]

I didn't add any nvidia stuff.

Currently display sleep = 12 min, computer sleep = 17 min (although sleep stopped working recently, probably config changes in BIOS). Sleep disks is checked. Wake on ethernet is not. But I doubt these settings are related since I've had a bunch of different values.