onibla Posted November 17, 2021 Share Posted November 17, 2021 Hi guys, I need your help. After several hours I'm able to start Big Sur on this machine: CPU: Intel Xeon-D (Broadwell-E) 1581 Mainboard: GXD2-ITX (unknown) https://www.amazon.it/HUNSN-Computer-Windows-GeForce-GTX1650/dp/B08X3FD4KL) Ram: 32GB Kingston Viper Video: Sapphire R7 250 (Cape Verde)(injected with SSDT-GPU-SPOOF.aml) Audio: ALC 4030 chip (Activated with voodooHDA) WiFi: Intel dual band 3168NGW But the system freeze after 5 minutes...I'm not able to solve this problem!!! IM becoming crazy CPUFriend.kext has been generate with ./CPUFriendFriend.command and CPUFriendDataProvider.kext has been generated with ./ResourceConverter.sh --kext /Users/osx/Desktop/ssdt.aml trying to manage 16 core. But I have no functional speedstep. The processor is ever at 1.8GHz also configuring TDP. Attached my EFI, my Darwin dumper and my bootlog. EFI.zip bootlog.zip DarwinDumper.webarchive.zip Link to comment Share on other sites More sharing options...
etorix Posted November 17, 2021 Share Posted November 17, 2021 (edited) Amazing find and amazing pick! Some crazy manufacturer has forcefully turned a Xeon D-1581 for edge servers into a desktop… and is selling the whole thing, including a GTX1650, for less than 700 E while any Supermicro X10SDV-16C motherboard retails for about 2300E!!! Neither the rationale nor the economics make any sense to me. It's already remarkable that you boot at all, especially with a Nvidia GTX spoofed as an AMD Radeon ( ). Realistically, this machine should be used with High Sierra and the last web drivers—I suppose the GPU cannot be swapped. The multiple CPU entries suggest that SSDT-UNC is needed to remove three of four sockets present in the DSDT. Edited November 18, 2021 by etorix update Link to comment Share on other sites More sharing options...
onibla Posted November 17, 2021 Author Share Posted November 17, 2021 Thanks etorix for reply. Yes seems to be a very nice desktop...I have changed Nvidia with real R7 250 to have metal support. I have already added UNC patch boot the system and patch testing both with CPUid1Data: C3060300 00000000 00000000 00000000 that D4060300 00000000 00000000 00000000 to boot the system, but the freeze problem persist Link to comment Share on other sites More sharing options...
etorix Posted November 18, 2021 Share Posted November 18, 2021 So the GPU can be changed, despite the fixed back connectors. This is getting more and more intriguing. Pity this machine is crippled with 2* SO-DIMM slots and a Realtek 1 GbE NIC (the Xeon-D SoC only needs a X557 PHY to provide server-grade 10 GbE)… Cooling may also be an issue. Even a 35 W Xeon D-1528 benefits from active cooling; a D-1581 should require more airflow than can be provided by the slits of the case. Is there any kernel panic associated with the freeze? Any hint as to the cause? Looking at the DSDT, I see no EC0 (it just needs a fake EC). I've found the system bus; there is a MCHC already. X99-CP00-XCPM.aml does nothing: It uses a wrong path for the processor, lacks its associated DTGP method; and, as far as I understand, if it worked, it would just duplicate what's already done by SSDT-PLUG. But none of that could explain the freeze. SSDT-EC-USBX.aml SSDT-SBUS.aml 1 Link to comment Share on other sites More sharing options...
Baio77 Posted November 18, 2021 Share Posted November 18, 2021 @onibla could you pass ioreg of your hack ??? Thanks 1 Link to comment Share on other sites More sharing options...
onibla Posted November 18, 2021 Author Share Posted November 18, 2021 (edited) @etorix thank you very much for the reply. Performing a fresh install on another ssd I have solved the freeze problem. Now remains the CPU manage issue. I have disabled X99-CP00-XCPM.aml and use your .aml files (SSDT-EC-USBX.aml and SSDT-SBUS.aml) I have already generated CPUFriend.kext and CPUFriendDataProvider.kext but nothing is changed. The same with ssdt-CPU.aml generated by ssdtPRGen. I not have speedstep and utilization is ever at 99%. How I can resolve these issue? There are other ways? Is it useful to make Bios screenshots of CPU management? Thanks again ssdt-CPU.aml Edited November 19, 2021 by onibla Link to comment Share on other sites More sharing options...
onibla Posted November 19, 2021 Author Share Posted November 19, 2021 @Baio77 thank you for your reply. attached the ioreg ioreg.zip Link to comment Share on other sites More sharing options...
Baio77 Posted November 19, 2021 Share Posted November 19, 2021 (edited) As I imagined, SSDT-Spoof does not work as you see addresses are different so SSDT is not read I show you some examples in photos If you want to fix the problems plaguing your hack you have to give the EFI a clean up. Example you have DSDT with various patches built in, but at the same time you use rename _DSM to XDSM in configplist. This is bad because you undo all DSM patches built into DSDT Edited November 19, 2021 by Baio77 1 Link to comment Share on other sites More sharing options...
etorix Posted November 19, 2021 Share Posted November 19, 2021 @onibla Can you provide the complete original ACPI tables? Use OpenCore DEBUG with Misc>Debug>SysReport: True and Target: 67; NO patched DSDT under ACPI. Then zip and post the SysReport folder recovered from the EFI partition. It is not necessary to successfully boot. This would help understand your configuration and how to reproduce it "the OpenCore way", with SSDTs and little or no patches. Link to comment Share on other sites More sharing options...
Baio77 Posted November 19, 2021 Share Posted November 19, 2021 (edited) Debug EFI Debug EFI 0.7.5.zip Test only DSDT in ACPI folder . I have added every patch in DSDTDSDT.aml keep only these rename in configplist This in quirks ACPI Edited November 19, 2021 by Baio77 Link to comment Share on other sites More sharing options...
onibla Posted November 19, 2021 Author Share Posted November 19, 2021 @etorix please find attached the ACPI origin extracted using clover 5122 by F4 at boot ACPI origin HSUNS.zip PS: If necessary and not sufficient I will extract them also using OC procedure suggested by you as soon I get home!!!!! @Baio77 as soon as I get home I will try to boot it following your suggestions and I'll update you... Thanks again 1 Link to comment Share on other sites More sharing options...
etorix Posted November 19, 2021 Share Posted November 19, 2021 Thanks! Either method gets the same table. An OpenCore SysReport would provide a list and devices and their PCI path (as PCIInfo.txt), but this information is already available from the IOReg. I'll have a look at this (and try to resist pulling the trigger on a barebone BM29… even desperately crippled by the motherboard, it's incredible to get a system for less than half the likely list price of the D-1581 alone). Link to comment Share on other sites More sharing options...
Baio77 Posted November 19, 2021 Share Posted November 19, 2021 Test this EFI in USB , if start ioreg https://drive.google.com/drive/folders/1HtjPQNcUgNdNKS8uqCsqW9DT32zc66i2?usp=sharing Link to comment Share on other sites More sharing options...
onibla Posted November 19, 2021 Author Share Posted November 19, 2021 (edited) @Baio77 and @etorix thank you very much for the support. @Baio77 using the new EFI the system not start, but stop here. Edited November 19, 2021 by onibla Link to comment Share on other sites More sharing options...
onibla Posted November 19, 2021 Author Share Posted November 19, 2021 (edited) Attached a clean copy of EFI folder that I'm using now. Almost all seems to be ok except the CPU management as showed in the picture. Frequency locked and utilization at 99% 🤔 in bootlog i noted kernel: (IOPlatformPluginFamily) X86PlatformPlugin::getCPUCStates - Unexpected data for C1 kernel: (IOPlatformPluginFamily) X86PlatformPlugin::getCPUCStates - Unexpected data for C1 kernel: (IOPlatformPluginFamily) X86PlatformPlugin::publishACPIStates - Failed to get max non-turbo PState. Set max non-turbo PState to default value 1 kernel: (IOPlatformPluginFamily) X86PlatformPlugin::publishACPIStates - Failed to get max non-turbo PState. Set max non-turbo PState to default value 1 kernel: (IOPlatformPluginFamily) X86PlatformPlugin::getCPUCStates - Unexpected data for C1 kernel: (IOPlatformPluginFamily) X86PlatformPlugin::getCPUCStates - Unexpected data for C1 kernel: (IOPlatformPluginFamily) X86PlatformPlugin::getCPUCStates - Unexpected data for C1 kernel: (IOPlatformPluginFamily) X86PlatformPlugin::getCPUCStates - Unexpected data for C1 kernel: (IOPlatformPluginFamily) X86PlatformPlugin::getCPUCStates - Unexpected data for C1 kernel: (IOPlatformPluginFamily) X86PlatformPlugin::getCPUCStates - Unexpected data for C1 EFI_clean.zip Edited November 19, 2021 by onibla Link to comment Share on other sites More sharing options...
Baio77 Posted November 19, 2021 Share Posted November 19, 2021 (edited) If start ioreg observe the order of the kext so it must be more or less EFI_clean.zip Edited November 19, 2021 by Baio77 Link to comment Share on other sites More sharing options...
Guest 5T33Z0 Posted November 19, 2021 Share Posted November 19, 2021 (edited) Try generating a SSDT-PM using ssdtPRGen. But you have have to add this line to the broadwell.cfg inside the ssdtPRGen folder in the Library: D-1585,65,500,1800,2400,16,32 Ypu should also drop CpuPm and Cpu0Ist before doing this. Edited November 19, 2021 by 5T33Z0 Link to comment Share on other sites More sharing options...
onibla Posted November 20, 2021 Author Share Posted November 20, 2021 (edited) @5T33Z0 thank you for reply.... I have already done this procedure with no success.. clearly I have added the ssdt into ACPI. Edited November 21, 2021 by onibla Link to comment Share on other sites More sharing options...
onibla Posted November 20, 2021 Author Share Posted November 20, 2021 @etorix @Baio77 @5T33Z0 probably the problem about the managing of CPU speedstep is here: Model of my CPU is 86 (hex=56), while the kernel recognize 3D (decimal=61). But I have not idea how to solve this!!!!!!!!! 😟 Link to comment Share on other sites More sharing options...
etorix Posted November 21, 2021 Share Posted November 21, 2021 (edited) @onibla That's one possible clue… Have you checked in another OS that the CPU can actually reach turbo frequencies? An issue (at least for me) is that you have a very Clover-ish setting with a modified DSDT and kernel patches. There are three xcpm patchs, a C6/C7 patch which could potentially mess up with power management, and an IOPCIFamily patch whose purpose is not apparent. I'd prefer to have a more typical OpenCore setting with only quirks and SSDTs, but it will be some time before I receive my BM29 (couldn't resist…). Here is an EFI I would start with, based on OpenCore 0.7.5 and the default guide, which suggests iMacPro1,1; no enabled patch, no custom frequency vector (yet). Needs serial numbers and rename Sample.plist to config.plist. Note: Keep your serial numbers private! Remove them before posting EFI folder or config.plist. EFI.zip Edited November 21, 2021 by etorix added EFI Link to comment Share on other sites More sharing options...
onibla Posted November 22, 2021 Author Share Posted November 22, 2021 @etorix thank you! I will start with the attached EFI to delineate the best configuration to boot my pc. I'll keep in touch Link to comment Share on other sites More sharing options...
onibla Posted November 24, 2021 Author Share Posted November 24, 2021 (edited) ...and finally, here we are! I would like to sincerely thank @etorix and @Baio77 for the support to realize my new hackintosh. good night and... See you soon onibla EFI.zip Edited November 24, 2021 by onibla 1 Link to comment Share on other sites More sharing options...
etorix Posted November 24, 2021 Share Posted November 24, 2021 Excellent news! From the EFI I see you have power management without CPUFriend and its data provider but that you've added a few ACPI renames and kernel patches. Could you comment on what they do? Link to comment Share on other sites More sharing options...
Guest 5T33Z0 Posted November 24, 2021 Share Posted November 24, 2021 (edited) @etorix Looks like he force-enabled X86 Platform plugin with a Kernel patch "Xcpm_bootstrap" so SSDT-Plug.aml works. But this also requires AppleXcpmCfgLock and AppleXpmExtraMsrs Kernel Quirks to make the whol construct work. Edited November 24, 2021 by 5T33Z0 Link to comment Share on other sites More sharing options...
onibla Posted November 24, 2021 Author Share Posted November 24, 2021 Hi @etorix and @5T33Z0 and thank you again for the support. DSM -> XDSM DSDT replacement patch seems to be necessary to inject SSDT on my configuration. While I have activated FPU_->MATH, TMR_->TIMR based on origin DSDT. They seems to be necessary to recognize some variables (I don't know which ones) as real Mac. Disabling even just one of these patch I lost the CPUManagement. Now I'm finalizing the last few things to improve it further, and then I'll test macOS Monterey!!!! I'ii update soon 1 Link to comment Share on other sites More sharing options...
Recommended Posts