Jump to content

[Discussion] - TRX40 Bare Metal - Vanilla Patches


Recommended Posts

@iGPU, @Driftwood - the MMIO was fine. In my case I needed to provide a slide=80, which I calculated using memmap. I think before it was a matter of luck as I succeeded sometimes and failed sometimes and the appearance of success on cold boot might have to do with the memory being reset, or just pure bias based on small samples where in fact it was always random.

 

OC log itself tells you actually the slide range and mine turns out to be the lower end of it.

 

04:892 00:001 OCABC: Only 176/256 slide values are usable!

104:893 00:001 OCABC: Valid slides - 80-255

 

I independently verified that 80 is indeed the least address under 256 in my memmap that is available (the derived from the Dortania formula address matching 79.5 which needs to be rounded to 80). Now my warm reboots work fine.
 

 

I now have only the shutdown problem (and maybe sleep haven't tried it), which everyone else faces.

Link to post
Share on other sites
  • Supervisor


I believe that more or less our motherboards have the same problems and behaviors

analyzing the logs of some of them I saw that the number of pages contained in the mmio groups are identical ..

obviously I don't know if the contents of the same

But the scheme I tested on my MSI also works on the asrock creator and it certainly worked also on MSI creator

in the various tests you are doing it is very simple to do this test with a basic EFI

now i am trying to mix this with Efi proposed by @iGPU because i assume that with his EFI is it possible to do a fresh installation of big sur os

this is the latest goal for me to reach

for now i can do a fresh install with old Osx but not with the newer one

Link to post
Share on other sites

Meaning that if MMIO is bad it won't boot (as opposed to boot sometimes). If MMIO boots sometimes but then randomly fails, one needs to look at slide. In my case leaving slide to default was causing these reboot failures, which look very similar to if MMIO is not devirtualised in the log. For me right value is slide=80. Now I get consistent boots.

 

Geekbench 5 loves bare metal and Big Sur 😎 - I think this is the highest I've seen and almost matches Linux! New OC fixes single core scores:

 

https://browser.geekbench.com/v5/cpu/3482941

 

Cinebench is unchanged at 21K same as VM

Geekbench Metal is also 75K same as VM.

 

I am amazed still at how good VM is given the stability and predictability I get from it.

Edited by meina222
Link to post
Share on other sites
6 minutes ago, fabiosun said:


I believe that more or less our motherboards have the same problems and behaviors

analyzing the logs of some of them I saw that the number of pages contained in the mmio groups are identical ..

obviously I don't know if the contents of the same

But the scheme I tested on my MSI also works on the asrock creator and it certainly worked also on MSI creator

in the various tests you are doing it is very simple to do this test with a basic EFI

now i am trying to mix this with Efi proposed by @iGPU because i assume that with his EFI is it possible to do a fresh installation of big sur os

this is the latest goal for me to reach

for now i can do a fresh install with old Osx but not with the newer one

You won't believe it, but I had the most twisted "successful" installation ever.

 

I 1st started with @iGPU's EFI, but would get freezes at 3-4 min in. I would try to erase all my partitions for clean install. But erasing a partition doesn't mean the files in it are deleted! So when the installer froze it did it AFTER it re-created the partitions but BEFORE it overwrite the install target disk EFI, so my old Catalina EFI survived! And because of an obscure option I used in the Catalina config.plist (enable bootstrap) on next reboot OC autoselected that without me realizing (as if someone changes the default boot option in BIOS), so I actually succeeded installing it using the 0.6.0 OC Catalina EFI and not the 0.6.1 by iGPU. Don't ask me how! I keep that EFI still though - still not sure if I re-try the install, if it will work or not with 0.6.1.

Link to post
Share on other sites
  • Supervisor

The sense of my writing above is different

you said shutdown is not working or it has a not clear behaviour

have you tested in a different way as i proposed?

i am asking before i miss only gigabyte board and i would like to have a confirmation if it is working also there

in my opinion from many users reports, gigabyte is the more problematic brand in trx40 osx configuration process

i am not sure of this because i love some gigabyte board design, but i also know their support is not so good sometimes

Link to post
Share on other sites

@fabiosun - I probably wasn't clear enough. I had 2 problems - 1) Start up (especially after reboot) would not always work - OC would fail in same way as if MMIO was not devirtualised. But sometimes it worked - it was non-deterministic (for me at least). As MMIO succeeded using cold reboot when doing the test described in Dortania, I figured it must be the slide as the slide is related to KASLR which is randomized. Now slide=80 fixes that issue for me. No more non-determinism. 2) Shutdown (or rather reboot after it). This was partly related to MMIO, and then partly related to what I describe next:

 

@iGPU, @Driftwood. I fixed shutdown. It works now, no problem. I turned off "Wake up on LAN" in my BIOS power settings. Now this puppy goes to full power down without an issue. Maybe all boards have this setting. I remembered I wanted to try that but didn't get a chance as my Catalina bare metal got corrupted so forgot. Until now. Hope it works for you too.

 

p.s. forgot to try sleep. Signing off, will try it tomorrow.

Edited by meina222
Link to post
Share on other sites
  • Supervisor

I see some EFI uploaded and then deleted

 

I can understand because some people use uploaded EFi in a not good way also rei-loading modified in a bad way and it looses original author intentions..but this behavior is not good for people is not present in that time EFI is uploaded (different time region)

 

So if EFI is only for personal use of a single user is better to not upload in public otherwise the reading of this thread is more complicated (it is a my personal opinion as simple user)

 

I hope you understand and I appreciate also your opinion in this subject

 

About MMIO

I would like to see an Opencore debug  log from a Gigabyte user and from a MSI Creator user

thank you

Edited by fabiosun
paginating
Link to post
Share on other sites

@iGPU I used your last version (EFI-v061-8e2755f-BareMetal-USB-installer-Public-Debug.zip), but unfortunately still the same issue with the installer. 😞

 

In the appendix you can find a video of the problem, debug txt and my config.plist. What could be the problem here?

 

Many thanks for your help in advance.

 

My hardware:

 

 

CPU: AMD 3970x

Mainboard: Gigabyte TRX40 Aorus Xtreme

GPU: AORUS Radeon RX 5700 XT 8G

Memory: 256GB  3200MHz - (F4-3200C16Q2-256GTZR) G.Skill Trident Z

SSD: Corsair MP600 2TB M.2 NVMe (PCIe 4.0)

config.plist.zip

opencore_debug_txt.zip

SysReport.zip

Edited by rressl
Link to post
Share on other sites
3 minutes ago, fabiosun said:

@rressl waiting from iGPU answer

 

you have:

  • Intel® X550-AT2 Dual 10GbE BASE-T LAN with cFosSpeed

are you sure kext you use is good for you?

Then

How do you have prepared Catalina drive?

 

 

I created it like this:

 

Format usb drive and crate the installer with "sudo /Applications/Install\ macOS\ Catalina.app/Contents/Resources/createinstallmedia --volume /Volumes/MyVolume" and then mount it with MountEFI.command and copy the EFI folder.

Link to post
Share on other sites
8 hours 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.

 

On my MB, the Gigabyte Aorus Xtreme, it was just a case of dong the MMIO Whitelisting and using the NVRAM SSDT. Since doing the Whitelisting I have had no problems with crashes, freezing  etc. My only problem at the moment is the lack of any SATA drives showing up and being accessible. They show up in Windows and at the OC picker (one of them has an old CCC bootable backup on it and presents as an option to load at the picker stage) but disappear as soon as macOS starts up. If I select the CCC clone it starts to load but crashes out - I'm guessing because it disappears as the SATA drives disappear.

Link to post
Share on other sites
  • Supervisor
13 minutes ago, rressl said:

 

Ohh okay. Simply formatted as apfs volume with Disk Utility. Just like I would do it with a "real" mac. 🙂 

 

try to comment your ethernet kext in your config and also, if you want try, rebuild your usb EFI   macOS installer with this simple app:

 

 

about your onboard ethernet:

https://www.elitemacx86.com/threads/fix-intel-x550-at2-10-gbit-card-ubuntu-eeprom-mod.327/

not tested and done by me but @DSM2 passes here sometimes 🙂

 

thanks @tomnic for this link

by the way thank you for your log

it confirms my idea 😉

 

Edited by fabiosun
deleted efi word
  • Like 2
Link to post
Share on other sites
8 minutes ago, fabiosun said:

try to comment your ethernet kext in your config and also, if you want try, rebuild your usb EFI   macOS installer with this simple app:

 

 

Thanks for the tips, but unfortunately the same problem. 😞

 

Can I have the issues because of the SSD's? (Corsair MP600 2TB M.2 NVMe (PCIe 4.0))

Link to post
Share on other sites
  • Supervisor

never seen that problem..but you can easily try with also a mechanical disk if you have

 

have you tested also installer app I have suggested?

it is based on create install media method...

Link to post
Share on other sites
1 minute ago, fabiosun said:

never seen that problem..but you can easily try with also a mechanical disk if you have

 

have you tested also installer app I have suggested?

it is based on create install media method...

 

Yes, I also tested with the installer app.

I mean suitable not because it is an SSD, but because it is PCIe 4.0.

Link to post
Share on other sites
  • Supervisor

others user use that drive

fresh installation could be a problem, but usually it is EFI related problem /bios settings

 

Many users solve cloning a working OSX (from a real Mac) and then booting with a proper EFI

 

 

Link to post
Share on other sites
5 minutes ago, tomnic said:

@rressl try to change in the bios the parameters you have for your sata interface, disable / enable things like sata aggressive link and hot plugging

I have tried and unfortunately exactly the same problem. I also tried to reset the UEFI. 😞

Link to post
Share on other sites
  • Supervisor
13 minutes ago, Ploddles said:

@rressl, the network kext you need for your MB is attached. I have the same MB as you.

@fabiosun FYI.

 

 

SmallTreeIntel8259x.kext.zip 96.69 kB · 1 download

 

for others Gigabyte users..

@Ploddles are you using others kext like some created from rehab man to inject or for you is useful only to have this you uploaded with modified lan id?

 

Edited by fabiosun
typo
Link to post
Share on other sites

@fabiosun,

 

Attached is my EFI for your info.

 

@rressl, try this. Just put your serial numbers etc in and you will need to add the pikera boot argument as you have a different dGPU. If you want, later today I can save my BIOS settings profile to a file that you can upload to yours to try, just let me know if you want it.

EFI.zip

Link to post
Share on other sites
  • fabiosun changed the title to [Discussion] - TRX40 Bare Metal - Vanilla Patches

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.