High Sierra wake problem (x79)

[frontend-dev]

While quite some folks seem to have wake problems with High Sierra, the symptoms (and therefore possible solutions) seem to be different, dpending on the hardware (and maybe something else?). I already tried those pmset changes and in fact I did the changes you proposed anyways ... but it does not solve the problem. Also your hardware setup is quite different. The "lead symptoms" I - and others - experience seem to be related to X79 chipsets, for example those RIVE and R4BE boards, or at least related to chipsets used with Sandy Bridge and / or Ivy Bridge CPUs. "Lead symptoms" are in my case unability to wake (only fans ramping up, but nothing else), reset button not working, therefore forced to make a "hard shutdown", and of course all those blinking LEDs on the mobo immediatly after waking. Also no traces whatsoever in the logs, indicating that the OS is not even aware of any wake. Looks like the problem is quite "low level", maybe something with ACPI. Also wake does work just fine with 10.11, 10.12 and Windows on the same machine.

 

[b0bbyd33]

I don't know if this helps but I have a bios code display on my motherboard (Patsburg C602 - very similar to X79), same symptoms as everyone else, Insta reset

the codes shown on wake
32 - 38 - 88 - E1 - 00 - 19 - A8 - AF - A8 - B8 - 88 - 6A - 5A - motherboard reset - sequence repeats

on Sierra after 5A
32 - 38 - 96 - 4F - 60 - 61 - 62 - 63 - 88 - 69 - 70 - 71 - 79 - 9A - 94 - 96 - AA - B2 and so on

the aptio codes are documented here : https://www.ami.com/ami_downloads/Aptio_V_Status_Codes.pdf

5A = Internal CPU error
32 = CPU post-memory initialization is started

is everyone on APTIO bios? that might be related

 

[frontend-dev]

My codes go a bit like this (hard to say what is a separate code and what is an artefact due to my recording as LEDs "lag" a bit):

32 - E1 - 00 - 10 - 19 - 27 - 26 - b9 - b5 - 5A

But I am not sure if that all makes sense, for example code "00" is simply unused. But at least I see a lot of the same codes as you do (am on APTIO as well).

And you are using similar hardware indeed. So it seems to mainly affect 2nd & 3rd generation Intel Core (Sandy Bridge/Ivy Bridge) CPUs with their respective chipsets (X79, C602 etc) and of course 10.13

I also attached a video to this post so you can see how it all actually looks like. Also note there are various other LEDs blinking, including the HDD Activity LED on the case, as one can see directly at the beginning when I wake the machine.

What I want to know in general is what it means if the status LED endlessly cycles like this after wake? I never have seen anything like this before.

Another thing: you say your machine resets on its own? For me even the reset button does not work anymore then, the only way out is a "hard shutdown", by pressing the power button for about 3-5 seconds.

 

[b0bbyd33]

Another thing: you say your machine resets on its own? For me even the reset button does not work anymore then, the only way out is a "hard shutdown", by pressing the power button for about 3-5 seconds.

it resets in that it immediately goes back through that bios sequence, only a hard power off gets it out of that state. I don't have a reset button available on my case.

 

[frontend-dev]

Ah, understand. So yeah this is apparently exactly the same behavior - and probably the same reason for this behavior - as in my case. Now, we need a solution

 

[russell664]

I don't know if this helps but I have a bios code display on my motherboard (Patsburg C602 - very similar to X79), same symptoms as everyone else, Insta reset

the codes shown on wake
32 - 38 - 88 - E1 - 00 - 19 - A8 - AF - A8 - B8 - 88 - 6A - 5A - motherboard reset - sequence repeats

on Sierra after 5A
32 - 38 - 96 - 4F - 60 - 61 - 62 - 63 - 88 - 69 - 70 - 71 - 79 - 9A - 94 - 96 - AA - B2 and so on

the aptio codes are documented here : https://www.ami.com/ami_downloads/Aptio_V_Status_Codes.pdf

5A = Internal CPU error
32 = CPU post-memory initialization is started

is everyone on APTIO bios? that might be related

APTIO BIOS, I'm assuming that is the stock one supplied by Asus?
I am aware that there are custom BIOS's for the Asus RIVE at least, though I'm pretty sure that some of the official 2013 Mac Pro users are having sleep issues too. As you say, this is very low level because of the CPU initialisation.

Out of question, what C-States are enabled in everyones BIOS?
All of my possible C-States are enabled, the CPU down clocks to 1.20ghz when idle and ramps up to 4.675ghz based on a custom SSDT-1.aml. Though, I have tested with a standard 4.5ghz SSDT-1.aml and there was no difference.

I wonder if the sleep commands are different for High Sierra and breaks at ACPI level.

 

[frontend-dev]

APTIO BIOS, I'm assuming that is the stock one supplied by Asus?

Yes I tested with 801 and the older 701. I even did a CMOS reset atone time.

As you say, this is very low level because of the CPU initialisation.

Are you sure it is because of CPU initialisation? At least this would narrow down possible causes a LOT.

Out of question, what C-States are enabled in everyones BIOS?
All of my possible C-States are enabled, the CPU down clocks to 1.20ghz when idle and ramps up to 4.675ghz based on a custom SSDT-1.aml. Though, I have tested with a standard 4.5ghz SSDT-1.aml and there was no difference.

My C-States are all enabled as well. But I recall testing limited states (at least the default which has limited states AFAIK) with the same results. I did not test every possible combination though, maybe it is a good idea to look at this a bit more and test other settings as well.

I wonder if the sleep commands are different for High Sierra and breaks at ACPI level.

Yeah, good question. At least it is apparently 10.13 which broke wake. Older versions worked just fine for me.

 

[otherwhaler]

A couple months ago, I installed a debug kernel and tried to investigate the ACPI messages around sleep and wake via FireWire debugging, using some of the parameters here, among others. I also tried comparing it to a real Mac, but in both cases I didn't see anything obvious, and I’m not really sure what to look for.

I had been hoping to catch some of the debug messages that are in Clover (via EDKII?) around saving and restoring the S3 resume vector. But they don’t seem to be logged anywhere that I can actually access. (Although that does make some sense, since control hasn't even been handed back to the kernel yet at that point.)

 

[frontend-dev]

But they don’t seem to be logged anywhere that I can actually access. (Although that does make some sense, since control hasn't even been handed back to the kernel yet at that point.)

Yeah, seems that the OS / kernel is not even aware of any wake, seems like the machine is stuck at some earlier point. So maybe the problem is not how the machine wakes, but the way it is put into sleep? I mean something must be different, since wake works on the same machine with earlier macOS versions (and windows).

 

[b0bbyd33]

Just because it was asked, my motherboard is a Lenovo D30 motherboard. Not sure who the OEM for the motherboard is, but it is using an APTIO bios with the MSR patch applied. Its not just an ASUS problem.