Ryzen 7000 Testing

[CaseySJ]

Please, any EFI, tips for a 7950X with Asus X670E -F Gaming Wifi ?

Hello @thunder72fr,

Please have a look at this post:

Ryzen 7000 Testing

@etorix and @CaseySJ Thank you both for your insight, especially since this will be my first machine with a TB4 maple ridge controller. Regarding maple ridge, an old friend of mine actually had a pretty nice setup with a Gigabyte Z590i Vision D board with TB4 that served him well. I think he...

forum.amd-osx.com

 

[fabiosun]

But if RestrictEvents is disabled, then incremental update works?

Yes

 

[CaseySJ]

Interim Update #22:

• I described the microphone and line-in problem on InsanelyMac, in the VoodooHDA section
• User Slice who has been developing and maintaining VoodooHDA responded a few hours ago
• We'll try some straightforward options first and go from there
• The thread on InsanelyMac is located here:

VoodooHDA 3.0.2

www.insanelymac.com

 

[Moozuki]

I use Resrtrictevents on my AMD system, my updates average 4 GB in size.
Can't remember when the last time I had to download a full installer.

 

[CaseySJ]

I use Resrtrictevents on my AMD system, my updates average 4 GB in size.
Can't remember when the last time I had to download a full installer.

Are you using MacPro7,1 as your system product name and running Ventura?

 

[Moozuki]

iMacPro, just updated to Ventura today. Is this only happening on MacPro7,1 ?

 

[CaseySJ]

iMacPro, just updated to Ventura today. Is this only happening on MacPro7,1 ?

Correct — I’m told this happens only with MacPro7,1.

 

[Moozuki]

Ok, please feel free to ignore my post then! I must have missed that info.....

 

[ToloEnt]

Please, any EFI, tips for a 7950X with Asus X670E -F Gaming Wifi ?

I'm trying, with this one, maybe it will serve you as a starting point at least.

 

[fabiosun]

I'm trying, with this one, maybe it will serve you as a starting point at least.

Debug/target is not set well and this EFI misses MMIO whitelist (mandatory for X670/e chipset to boot fine (as working EFI users demonstrate in the past)

 

[CaseySJ]

** Ventura Incremental macOS Updates with RestrictEvents Enabled **

As we know, incremental macOS updates fail to occur when using MacPro7,1 with Ventura. To solve this problem, try adding revpatch boot argument as shown using standard version of RestrictEvents:

revpatch=auto,sbvmm,asset

Just confirmed this on Intel Z690, but not yet on AMD Ryzen​

 

[CaseySJ]

It shall lie there unopened until one day...

 

[fabiosun]

revpatch=auto,sbvmm,asset

@CaseySJ @allusers
Happy new year

it does not work here

 

[CaseySJ]

@CaseySJ @allusers
Happy new year

it does not work here

There are few options remaining, but if incremental update works on AMD when RestrictEvents is disabled, we can consider the following:

• When we don't specify revpatch, it gets set to auto, which means memtab,pci,cpuname are used for MacPro7,1
• So let's see what happens if we reduce these options one by one:
  • Test 1:
    • revpatch=memtab,cpuname
  • Test 2:
    • revpatch=cpuname
  • Test 3:
    • revpatch=memtab

 

[fabiosun]

revpatch=memtab,cpuname

it doesn't work*

revpatch=cpuname

it works!**

revpatch=memtab

it doesn't work*


* but memory full slot detection is good (old graphics representation)


** but memory full slot detection is wrong (new MacPro7.1 graphic representation)



4 should be 8

 

[CaseySJ]

@mariettosun

Okay, now we’re getting somewhere. Can you please try:

• Test 4:
  • revpatch=pci,cpuname

If this doesn’t break incremental updates, then we can focus on memory slot configuration…

 

[fabiosun]

revpatch=pci,cpuname

it breaks incremental upgrade*

* memory slot is good but with old memory graphics slots representation (same memtab graphics option and results)

 

[CaseySJ]

it breaks incremental upgrade*

* memory slot is good but with old memory graphics slots representation (same memtab graphics option and results)

I'm surprised or perhaps puzzled why memtab causes incremental macOS update failure with MacPro7,1. The code only implements this for:

• MacBookAir
• MacBookPro10

We can see that needsMemPatch is only set to true for these two products:

            if (enableMemoryUiPatching | enablePciUiPatching) {
                // Rename existing values to invalid ones to avoid matching.
                if (strcmp(di.modelIdentifier, "MacPro7,1") == 0) {
                    // on 13.0 MacPro7,1 string literal is inlined, but "MacPro7," will do the matching.
                    modelFindPatch = "MacPro7,";
                    modelReplPatch = "HacPro7,";
                    // partial matching, thus exclude '\0'.
                    modelFindSize  = sizeof("MacPro7,") - 1;
                    DBGLOG("rev", "detected MP71");
                } else if (strncmp(di.modelIdentifier, "MacBookAir", strlen("MacBookAir")) == 0) {
                    needsMemPatch = true;
                    modelFindPatch = "MacBookAir";
                    modelReplPatch = "HacBookAir";
                    modelFindSize  = sizeof("MacBookAir");
                    DBGLOG("rev", "detected MBA");
                } else if (strncmp(di.modelIdentifier, "MacBookPro10", strlen("MacBookPro10")) == 0) {
                    needsMemPatch = true;
                    modelFindPatch = "MacBookPro10";
                    modelReplPatch = "HacBookPro10";
                    modelFindSize  = sizeof("MacBookPro10");
                    DBGLOG("rev", "detected MBP10");
                }

                if (modelFindPatch != nullptr) {
                    binPathSystemInformation = getKernelVersion() >= KernelVersion::Catalina ? binPathSystemInformationCatalina : binPathSystemInformationLegacy;
                }
            }

But if pci is set, then we get this:

        // Disable notification prompts for mismatched memory configuration on MacPro7,1
        if (strcmp(info->modelIdentifier, "MacPro7,1") == 0) {
            if (strstr(value, "pci", strlen("pci")) || strstr(value, "auto", strlen("auto"))) {
                if (getKernelVersion() >= KernelVersion::Catalina) {
                    DBGLOG("rev", "disabling PCIe & memory notifications");
                    procBlacklist[i++] = (char *)"/System/Library/CoreServices/ExpansionSlotNotification";
                    procBlacklist[i++] = (char *)"/System/Library/CoreServices/MemorySlotNotification";
                }
            }
        }

We can see that both ExpansionSlotNotification and MemorySlotNotification are disabled with either pci or auto for MacPro7,1. But memtab should not have an effect for MacPro7,1 unless of course I'm not reading the code properly!

 

[fabiosun]

I can't say why, but here only in this condition system updates well:

 

[CaseySJ]

I can't say why, but here only in this condition system updates well:

Because memory slot and PCI slot notifications are enabled, do you see any of those notifications? Or do you think the system works well with just revpatch=cpuname?