Sometimes the boot hangs - Intel HD 4600

[Alby128]

In the past few months, I ran into a kind of problem that I have never experienced in all these years experimenting with Hackintosh.

I set up Sierra 10.12.6 on a machine with this configuration:

 

Motherboard: ASUS H87M-E

CPU: i5-4570

GPU: Intel HD Graphics 4600 (GT2)

Audio: ALC887

LAN: RTL8111G

 

I used Clover 4114 as bootloader. To make the internal graphics work, I inject changes in DSDT tables from GFX0 to IGPU at boot time. Inject Intel is also enabled with ig-platform-id = 0x0a260006. Two kexts are used for audio and LAN. My config.plist file is attached.

 

Most of the times, everything works perfectly. Full QE/CI support, full resolution, no evident issues. Sometimes, the system does not boot at all. The boot hangs at "waiting for IGPU" and it seems that does not recognize the graphics card (no "PPGTT is enabled", for example, when the system does not boot). Even crazier, after one or two reboots without changing anything in the config.plist, the system starts again and everything works again, out of the blue. 

 

I suspected this behavior was due to some weird BIOS setting, but I already fixed the amount of GPU shared memory and this did not solve the issue. Do you have any idea of what might be the cause? The fact that the issue is not permanent or even deliberately reproducible is making me go crazy.

 

Thanks for your help,

Alberto

config.plist.zip

[Alby128]

Thanks, but I do not think this solves the issue.

With the updated layout-id, it does boot, but the error still happened other two times. On the bright side, now I have an exact copy of the point where the boot hangs: 

busy timeout[0], (60s), kextd wait(0): 'IGPU'

Any other ideas? I attached to this post the config.plist I updated according to your suggestion.

 

Alberto

config.plist.zip

[Alby128]

Attached to this post there is a complete boot log from one of those times in which the boot hangs. I hope it can help some of you guys to understand what is happening. 

 

Thanks,

Alberto

 

 

[Alby128]

Sorry for the late reply, but strangely I am not getting email notifications from this thread. 

 

Anyway, attached there is my whole clover folder. I thought I had generated my SSDT and DSDT, but I could not find them anymore in my ACPI/patched folder, so maybe something happened and I decided to remove them. Also, I am using iMac 14,1 as SMBIOS, so maybe that could also be an issue. 

 

Thanks for your help,

Alberto

clover.zip

[Alby128]

I tried to amend and simplify my config.plist following your suggestions. I changed my SMBIOS to iMac 14,2 , removed the fake GPU ID, removed the AICPUPM patch, and added a custom generated SSDT.aml in Clover's patched folder. The unknown CPU model warnings are gone, but the bug is still there. Attached to this post there is a copy of the edited config.plist and a pic of the boot log from a failed boot after these changes.

 

Shockingly, from a glimpse during a working boot, it seems that the boot log is much different when the boot works (many strings with long numbers). However, it runs too fast to take a pic and I am not aware of a way to log Clover boot logs. Can you help me with that?

 

Thanks again,

Alberto 

config.plist.zip

[Alby128]

I finally made a high-resolution video of both situations, and I found some differences in the logs between the two cases. I hope these data will help you understand what is going on. I am only uploading the parts that contain differences. 

 

At the beginning of the boot, when it works:

 

Same point, in the case in which it eventually hangs:

 

 

Then at the end of the boot, in the successful case:

 

And where the boot hangs in the other case:

 

 

The absence of the FakeSMC (and SMC) messages is particularly striking in the hangs case, but also it appears that there is a missing PCI device (11 vs 10) when the boot does not work. Any ideas?

 

Thanks,

Alberto

[Alby128]

Up... I also tried with IntelGraphicsFixup, with no luck.