trx40 [Discussion] - TRX40 Bare Metal - Vanilla Patches

[iGPU]

4 minutes ago, meina222 said:

One more thing to report after some testing the shiny new BS install. Every once in a while I fail to boot and instead of getting deep into the log, my PC just reboots. Next time it usually succeeds. So there is still some quirk around maybe slides that causes this instability.

 

As I mentioned, I'm not using any Slide value. Look at post above, maybe you have a similar USB issue.

 

On a related issue, I think all desktop Hackintoshers should disable Energy Saver as shown below. This can reduce oddball crashes and reboots in many systems (Intel or AMD).

 

[meina222]

Correct - no slide here too. I pretty much took your official OC debug EFI and tweaked args, MMIO and now added some SSDT's to it for my hardware. The reboot happens at the very beginning before I even get a chance to see any text log - and not always, maybe once every 3-4 boots.

Edited August 29, 2020 by meina222

[iGPU]

This is a follow up on how to remove HS02 from XHCI. You cannot remove HS02 until you define something. The attached SSDT (8-SSDT-TRX40-BYD8-XHC-XHCI-complete.aml) defines all ports in both XHC and XHCI located at S0D2/DYB8.

 

Again, the SSDT defines all of the ports. If you need to remove any other port, such as for BT/Wifi, be my guest. I would suggest duplicating this SSDT and not actually using it. That is, delete whatever you wish from the duplicate and re-name the modified duplicate to something like 8-SSDT-TRX40-BYD8-XHC-XHCI.aml 

 

Needless to say, only load the modified SSDT (not the *-complete version).

 

The "8-" in the file name is because I have others SSDTs to load and like to define an order of loading. This is a holdover from Clover, which loads files based on file name; OpenCore forces you to specify the order of ACPI and Kext file loading based on file entry order in the ACPI/Add and Kernel/Add sections. So the "8-" is superfluous.

 

Shown below is an excerpt from the attached SSDT. Only the XHCI section is shown (the XHC section extends above with only the last PR10 port revealed).

 

Highlighted is the HS02 port that needs removing (based on my post above; your requirement may differ!). Simply delete the entire HS02 definition (or whatever other port on your system that needs removal). Re-save the file as described above and load that SSDT in OpenCore. On re-boot HS02 will no longer be displayed in IORE. Please note that this is reversible: if you add HS02 back to the SSDT and re-boot, HS02 will re-appear as these mods are non-destructive to hardware.

 

 

8-SSDT-TRX40-BYD8-XHC-XHCI-complete.aml.zip

 

Edited August 29, 2020 by iGPU

[meina222]

@iGPU - I left my IOReg running for the last 20 min. No such flickering. The only items that change (which is normal as I browse some stuff) are:

 

RootDomainUserClient

AMDRadeonX6000_AMDAccelCommandQueue

AMDRadeonX6000_AMDAccelSharedUserClient

AMDRadeonX6000_AMDGFX10GLContext

IOAudioEngineUserClient

IOAudioControlUserClient

IOSurfaceRootUserClient

IOTimeSyncDomainUserClient

 

I see a lot of those being stricken out, but none from USB ports.

Edited August 29, 2020 by meina222

[meina222]

@iGPU, My attempt to define my ports (and remove that PRT5 that is the AX200 BT as you showed me earlier) is attached.

 

Unlike Catalina, this one doesn't seem to work though. It's the last of my SSDT's. Attaching EFI, IOReg. Could you please help with checking if anything is off?

EFI.zip BSiMacProIOReg.zip

 

Edit: Digging thru my OC log I see this:

09:224 00:001 OC: Failed to find ACPI SSDT-TRX40-DESIGNARE-XHC

 

That's weird - file was there (forgot to add .aml but I don't think that matters if name matches).

 

Update: BT fixed. I erased and reapplied the SSDT and changed its name to have .aml (latter wasn't the issue I think but better for consistency). Must have had some subtle typo.

 

I still do get occasional failures to boot as if my MMIO is not correct but it is non-deterministic.

Edited August 29, 2020 by meina222

[Ploddles]

Happy to report that I managed to install BS onto an NVMe thanks to @iGPU's EFI. No issues with it at all at the moment.

 

The only remaining problem I have is the lack of any SATA drives once Catalina or BS load up. I can see them at the Opencore Picker and within Windows and Mint, so it is restricted to macOS. As I have pretty much no idea when it comes to SSDTs, I'm guessing the problem, and solution, will be in there somewhere.

 

Attached is my IOREG if anyone can point me in the right direction.

 

Thanks.

Ploddles IOREG.ioreg.zip

[meina222]

Some further observations on Big Sur:

 

• NVRAM works! Kudos, @iGPU

• Shutdown clicks and turns off and immediately re-boots back up. Still unresolved even after USB remap (which I think needs more work as my port types are not 100% correct)

• Whenever I reboot or shutdown (which turns into a reboot after the power off click), I cannot boot back in. Most of the time if not always. Stop is early on in the same manner as if my MMIO's are incorrect. But then when I power cycle the PC, I can boot back in without a problem. This is very weird and is the most serious issue so far.

Edited August 29, 2020 by meina222

[tsongz]

@meina222 is your WiFi / BT working in Big Sur with BCM4360?

 

Also, I noticed an update in Dortania's OC Guide regarding TRX40 - not sure if this is correlated but I imagine it has a lot to do with your work @iGPU.

 

Additionally - there's a new update with intelwifi kext that does officially support AX200 - I have the kext loaded and will test it on my next reboot on my PVE Big Sur (mainly because I almost bricked my BS by trying to fix the BCM4360 kext)

[meina222]

Yes, it does after disabling the AX200 with an SSDT as per @iGPU's post half a page above this.

 

The 2 big issues I have:

 

1. Not able to boot in after a shutdown or a restart (peculiar as I have not seen anyone else report this)

2. Shutdown itself still doesn't work but is close as we need to identify what causes the system to spring back to life after the click off.

 

@tsongz - the Dortania guide is discussed earlier in this topic. The MMIO address section of it is not accurate as it does not apply to many motherboards/configurations. I would rather follow @iGPU's posts and try to start from his EFI's.

Edited August 29, 2020 by meina222

[meina222]

After every reboot  get this 

 

 

[iGPU]

3 hours ago, meina222 said:

@iGPU, My attempt to define my ports (and remove that PRT5 that is the AX200 BT as you showed me earlier) is attached.

 

That's weird - file was there (forgot to add .aml but I don't think that matters if name matches).

 

Update: BT fixed. I erased and reapplied the SSDT and changed its name to have .aml (latter wasn't the issue I think but better for consistency). Must have had some subtle typo.

 

I still do get occasional failures to boot as if my MMIO is not correct but it is non-deterministic.

 

OK, I changed a few things, re-arranging SSDTs. I also commented out 2 DevProp entries. If DevProp are wrong, they'll mess up a boot; better none than wrong. (I'd need a PCI.txt output from Hackintool to verify. In fact, when I work up stuff I use that file and IORE to work on config.plist entries.) The newer SSDTs are more streamlined than what I'd uploaded earlier (now removed) in this thread (the 'renaming' SSDT is no longer required).

 

 

 

and I disabled the Aquantia patch since it's not applicable to your mobo:

 

 

I also removed un-used kext entries and swapped out the I211, replacing it with the I210 kext (IntelMausi).

 

 

EFI deleted.

 

 

 

Edited August 30, 2020 by iGPU

[Pavo]

So.... you has working NVRAM for TRX40? How did you get it working? This stupid Booter > Quirks > DisableVariableWrite variable is pissing me off.

[iGPU]

2 hours ago, Ploddles said:

Happy to report that I managed to install BS onto an NVMe thanks to @iGPU's EFI. No issues with it at all at the moment.

 

The only remaining problem I have is the lack of any SATA drives once Catalina or BS load up. I can see them at the Opencore Picker and within Windows and Mint, so it is restricted to macOS. As I have pretty much no idea when it comes to SSDTs, I'm guessing the problem, and solution, will be in there somewhere.

 

Attached is my IOREG if anyone can point me in the right direction.

 

Thanks.

Ploddles IOREG.ioreg.zip 1.27 MB · 0 downloads

 

I'm not certain why you're not seeing a SATA drive. I do nothing special for SATA drives. I only use the basic EFI I've previously uploaded with some customized SSDTs. However, none of the SSDT or anything I've done to the EFI or config.plist file specifically address SATA drives.

 

I can see the 1 of the 3 SATA drives I presently have running (there is a 4th 10TB drive I use for backup but it's a bit noisy and I only connect when backing up things). I have a Catalina back on a SATA drive as shown below. The "Untitled" drive is a Win10 NVMe drive (and Win10 is somehow corrupted and and won't boot... annoying but I dislike Windows so expect nothing less). The bottom drive is Cat-Bkup, the SATA drive (ignore the fact that it's yellow suggesting an external drive; it is not external).

 

The other 2 SATA drives contain Linux. They're described in bottom image.

 

 

 

All of my SATA drives appear under BYD8 (below XHC and XHCI). Since the bottom two are Linux. they don't appear on the macOS desktop (unless you add some extra software, but not worth the effort as it is risky to edit from a Mac).

 

 

 

PAVO:

25 minutes ago, Pavo said:

So.... you has working NVRAM for TRX40? How did you get it working? This stupid Booter > Quirks > DisableVariableWrite variable is pissing me off.

 

Yes, I got it working several days ago. It required setting up MmWhitelist. Since you and I have same set up, I'll send you my EFI for a trial run.

 

Edited August 30, 2020 by iGPU

[Pavo]

7 minutes ago, iGPU said:

 

PAVO:

Yes, I got it working several days ago. It required setting up MmWhitelist. Since you and I have same set up, I'll send you my EFI for a trial run.

When you do the MMIOWhitelist do you disable the DisableVariableWrite quirk? If I disable that stupid quirk, macOS boots but I don't get any video. Only with it enabled do I get video.

[iGPU]

3 minutes ago, Pavo said:

When you do the MMIOWhitelist do you disable the DisableVariableWrite quirk? If I disable that stupid quirk, macOS boots but I don't get any video. Only with it enabled do I get video.

 

Here I whipped this up for you just now. Only need to fill-in SSN stuff. I included my SSDTs for this board, only changing the fact that you use 1 Radeon VII and I have 2. This could change some DevProp stuff I left in place as I don't know your exact NVMe set up. So please re-check those.

 

I removed TB references since I think you're not using.

 

(EFI deleted after download.)

 

Edited August 30, 2020 by iGPU

[meina222]

@iGPU - I downloaded the EFI. you can delete it.

 

 

[Pavo]

12 minutes ago, iGPU said:

 

Here I whipped this up for you just now. Only need to fill-in SSN stuff. I included my SSDTs for this board, only changing the fact that you use 1 Radeon VII and I have 2. This could change some DevProp stuff I left in place as I don't know your exact NVMe set up. So please re-check those.

 

I removed TB references since I think you're not using.

 

(EFI deleted after download.)

 

Ok cool thanks, all I needed was the PMCR SSDT, all good now.

[iGPU]

5 minutes ago, Pavo said:

Ok cool thanks, all I needed was the PMCR SSDT, all good now.

 

Let me know how it works. I've certainly enjoyed using your software, so a pleasure if I can return a favor.

 

If you were to read through the old posts, I initially used a calculated slide value (slide=128 as boot arg), but saw no advantage and stopped using. The MmioWhitelist, WriteFlash=Yes, and other Quirk settings are all that seem to be needed for native NVRAM on our TRX40.

 

Edited August 30, 2020 by iGPU

[Pavo]

7 minutes ago, iGPU said:

 

Let me know how it works. I've certainly enjoyed using your software, so a pleasure if I can return a favor.

 

If you were to read through the old posts, I initially used a calculated slide value (slide=128 as boot arg), but saw no advantage and stopped using. The MmioWhitelist, WriteFlash=Yes, and other Quirk settings are all that seem to be needed for native NVRAM on our TRX40.

 

All good now, now.... has anyone tried install BigSur from installer? I can boot already installed BigSur, but remember trying earlier and couldn't get past second install phase.

[iGPU]

8 minutes ago, Pavo said:

All good now, now.... has anyone tried install BigSur from installer? I can boot already installed BigSur, but remember trying earlier and couldn't get past second install phase.

 

The EFI works for installation. I've installed from both USB and applications folder. Original here. But you have same EFI with extra SSDTs, so no advantage re-downloading.

 

If you want to remove Snapshot partition see here. For removal of seal during install, see here. (Remove seal during install, then remove Snapshot.)

[Driftwood]

Thanks Fabiosun for the help today. Here's a Macpro7,1 check working  apps. Apoologies for poor quality - my screeners software Camtasia doesnt work in BM  so I had to use some Telestream software and by the time Id edited it, then exported to see the dreaded Watermark on the trial! Had to double screen record the playback then, thru OBS and then output to You Tube!

 

:-) Anyways it shows you what works.

 

 

[Driftwood]

4 hours ago, meina222 said:

After every reboot  get

 

Check your MMIO skips  in Opencore [date].txt

 

Skips = 1 is good remember, but follow @fabiosun layout schema inserting YOUR MMIO addresses into each of the 19 Childs in MMIOwhitelist.

 

The order of your. YES and NO for each child should follow similar to the one fabiosun talks about and I posted earlier in terms of order of Yes/Nos. Note that theyre all mainly YES with a I think 5 No's towards the latter Childs.

 

Slide values dont appear to be worthwhile so Ive ignored them.

Edited August 30, 2020 by Driftwood

[iGPU]

34 minutes ago, Driftwood said:

 

Check your MMIO skips  in Opencore [date].txt

 

Skips = 1 is good remember, but follow @fabiosun layout schema inserting YOUR MMIO addresses into each of the 19 Childs in MMIOwhitelist.

 

The order of your. YES and NO for each child should follow similar to the one fabiosun talks about and I posted earlier in terms of order of Yes/Nos. Note that theyre all mainly YES with a I think 5 MOs towards the latter Childs.

 

Slide values dont appear to be worthwhile so Ive ignored them.

 

This is simply not true. My pattern is totally different.

 

Each mobo, BIOS ver and who knows what else will be different. I wouldn't give recommendations unless you truly know by having tested it, which, of course, is impossible due to our different mobos. 

 

Edited August 30, 2020 by iGPU

[Driftwood]

43 minutes ago, iGPU said:

This is simply not true. My pattern is totally different.

 

I worked out my own pattern a few days back working backwards as you mentioned, testing each address with a reboot, did it twice - one for csm disabled and one for Above 4G - and yes. their order was differing.

 

However, the fact that I got no Shutdown, Restart and Sleep proved that the backwards test isn't fool proof. Yes each motherboard has different addresses. But after fabiosun suggested his schema of order of `yes/No' and it working for both mine and his is either a brilliant coincidence or there's something in it.

 

Indeed you could try umpteen combinations of their order, but it does seem the skips are important in certain child numbers.

 

I don't think anyone knows the perfect answer yet.

 

No disrespect to your tireless work @iGPU

Edited August 30, 2020 by Driftwood

[fabiosun]

@Driftwood thank you for the credits for me and this site 

happy it is working well also for you

if others want to try check well number of page in mmio log description

do a conversion from your data and follow the schema I have proposed

i have tested with other 2 different users and all of them have the same behaviors shown in @driftwood video

 

My initial goal for mmio was to skip all mmio blacklisted address due devirtualizemmio on flag