Ryzen 7000 Testing

[CaseySJ]

Where does one find these kernel patches?

…

Huh?? Just take another look at the post you quoted from.

 

[Galve2000]

Huh?? Just take another look at the post you quoted from.

So just copy, paste into the config.plist?

I think I have to finally face my fear of OC Configurator

 

[mariettosun]

@CaseySJ I am double checking if I have put all as you said
I have this:


have you the same device id as mine?

 

[ExtremeXT]

So after studying the patches, I've concluded that this is what they do:
Patch 1: Changes 3 (0x3) to 0x13 in this code snippet from the disassembled code.
alloc_dma_buffer(this,*(IOMapper **)(this + 0x130),(ulong)*local_50 << 4 | 8,0x200,3,'@',(IOBufferMemoryDescriptor **)pplVar18,(IODMACommand **)pplVar2);
Is changed to:
alloc_dma_buffer(this,*(IOMapper **)(this + 0x130),(ulong)*local_50 << 4 | 8,0x200,0x13,'@',(IOBufferMemoryDescriptor **)pplVar18,(IODMACommand **)pplVar2);

Patch 2: Changes 0x1 to 0x11
plVar12 = (long *)IOBufferMemoryDescriptor::withOptions(1,(ulong)(*puVar16 << 0xb),0x1000);
Is changed to:
plVar12 = (long *)IOBufferMemoryDescriptor::withOptions(0x11,(ulong)(*puVar16 << 0xb),0x1000);

Patch 3 and 4 are basically the same but change bytes in another function.

I wonder how exactly that increasing the byte by 0x10 would fix this, but unfortunately it's hard to figure out what the function alloc_dma_buffer actually does with only the disassembled code.

If the patches cannot be further trimmed, should them get PR-ed to be merged into the ForceAquantiaEthernet kernel quirk in OC? Here is the where the Aquantia patch is in the code.

OpenCorePkg/Library/OcAppleKernelLib/CommonPatches.c at master · acidanthera/OpenCorePkg

OpenCore bootloader. Contribute to acidanthera/OpenCorePkg development by creating an account on GitHub.

github.com

 

[mariettosun]

If the patches cannot be further trimmed, should them get PR-ed to be merged into the ForceAquantiaEthernet kernel quirk in OC?

For you does it work?

 

[ExtremeXT]

For you does it work?

I don't have any Aquantia card, so I can't test.

 

[CaseySJ]

So after studying the patches, I've concluded that this is what they do:
Patch 1: Changes 3 (0x3) to 0x13 in this code snippet from the disassembled code.
alloc_dma_buffer(this,*(IOMapper **)(this + 0x130),(ulong)*local_50 << 4 | 8,0x200,3,'@',(IOBufferMemoryDescriptor **)pplVar18,(IODMACommand **)pplVar2);
Is changed to:
alloc_dma_buffer(this,*(IOMapper **)(this + 0x130),(ulong)*local_50 << 4 | 8,0x200,0x13,'@',(IOBufferMemoryDescriptor **)pplVar18,(IODMACommand **)pplVar2);

Patch 2: Changes 0x1 to 0x11
plVar12 = (long *)IOBufferMemoryDescriptor::withOptions(1,(ulong)(*puVar16 << 0xb),0x1000);
Is changed to:
plVar12 = (long *)IOBufferMemoryDescriptor::withOptions(0x11,(ulong)(*puVar16 << 0xb),0x1000);

Patch 3 and 4 are basically the same but change bytes in another function.

I wonder how exactly that increasing the byte by 0x10 would fix this, but unfortunately it's hard to figure out what the function alloc_dma_buffer actually does with only the disassembled code.

If the patches cannot be further trimmed, should them get PR-ed to be merged into the ForceAquantiaEthernet kernel quirk in OC? Here is the where the Aquantia patch is in the code.

OpenCorePkg/Library/OcAppleKernelLib/CommonPatches.c at master · acidanthera/OpenCorePkg

OpenCore bootloader. Contribute to acidanthera/OpenCorePkg development by creating an account on GitHub.

github.com

I’ll create a Patch Theory post today. Meanwhile, your analysis of the patches is correct.

 

[CaseySJ]

@CaseySJ I am double checking if I have put all as you said
I have this:
View attachment 8825

have you the same device id as mine?

I tried it with my GC-AQC113C, which has (spoofed) device ID 0x94C0. Some suggestions:

• Please try cold start with cable disconnected
• After booting up, connect cable
• Also try Monterey 12.6.x

 

[mariettosun]

I don't have any Aquantia card, so I can't test.

I would try to have a more solid and successful database (Device id and platform) before to submit a PR
"IF" It is "AMD" related problem they check only if code is valid

 

[mariettosun]

• Please try cold start with cable disconnected
• After booting up, connect cable
• Also try Monterey 12.6.x

Tested, same behaviour
after few minutes it says connected but wrong ip and subnet mask and so on
Tested also @ different hardware speed different from 10Gbit

 

[CaseySJ]

Tested, same behaviour
after few minutes it says connected but wrong ip and subnet mask and so on
Tested also @ different hardware speed different from 10Gbit

View attachment 8826

This happened to me in Ventura once, but I made the port Inactive and then Active. I also did a cold boot.

If this problem continues, please test Monterey 12.3 or newer.

 

[mariettosun]

Tested also in Monterey 12.6.1
It is the same
Maybe not useful to test on trx40 and I hope other people could give a positive feedback

 

[CaseySJ]

Tested also in Monterey 12.6.1
It is the same
Maybe not useful to test on trx40 and I hope other people could give a positive feedback

Interesting…

• Does the TRX40 system crash when cable is connected?
• It may be useful to look at debug logs (boot argument apple-axgbe-debug=0x1 will generate system logs)
• We would then need to look for “interrupt_msi_handler”

log show --last boot | grep Aquantia > bootlog.txt

Then open bootlog.txt in any text editor and look for interrupt_msi_handler sections

We would then compare your logs to this:

Ryzen 7000 Testing

Log Analysis 4 of N Conditions: This is a comparison of how interrupts are set up Intel platform with driver from 12.6.x that works properly because AppleVTD is enabled The same number of MSI interrupts were found and registered, but here we see that each interrupt was remapped by AppleVTD The...

forum.amd-osx.com

 

[mariettosun]

Does the TRX40 system crash when cable is connected?

No, in both OS I use

I will test as you advice in few minutes
thanks

 

[mariettosun]

@CaseySJ
I have few lines in bootlog

 

[CaseySJ]

@CaseySJ
I have few lines in bootlog

Is this everything related to Aquantia from the system log? There are too few logs in this file. There should be thousands of log entries related to Aquantia.

Do you also have debug=0x100 boot argument? Not sure if that is required, but it is present in my configuration.

 

[mariettosun]

I have posted the entire file I have had
now trying with additional boot arg ( I did not use it)

 

[CaseySJ]

I have posted the entire file I have had
now trying with additional boot arg ( I did not use it)

Also please post a screenshot of those 4 patches. I know you don’t like Clover Configurator, but it shows these patches very clearly. Because I am sitting by the fireplace with just my iPad, I cannot open a plist editor. I actually avoid plist editors as much as possible (just my preference, don’t mean to upset anyone).

 

[mariettosun]

I know you don’t like Clover Configurator

not true
I like all software
CC destroyed (in the past) AMD kernel patches in Clover

 

[mariettosun]

a bit different but always few lines