Well, I have reprogrammed the MATT card from my PC as that was the only way. I PM Alex to inspect the ROM whenever possible.
@tsialex is there a way to disable sip on the MATT card externally?
Last edited:
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
GOP Boot Screens part II(Clover)
- Thread starter DearthnVader
- Start date
- Sort by reaction score
@tsialex is there a way to disable sip on the MATT card externally?
The SIP settings are stored in nvram.
Glad you were able to reprogram your MATT card, and thanks for being our guinea pig for this test.
Not a problem for the testing I can safely do tests with the MATT card, but the inspection done by Alex on the ROM is discouraging. He quickly looked in the ROM and found a complete mess in the NVRAM section. He can elaborate on this when possible.
UEFITool may not handle Apple firmwares correctly, if it is corrupting the nvram.
About the AGP Driver, it seems to be some sort of shim, the injects itself between a display connected to a supported graphic card and the OS or Boot Picker.
I'll be running a few tests to see if I can learn anything new about how/why it works on the 3,1 for GOP.
I am not sure whether it is the UEFI tool or Clover... I wonder if newer driver from nMP would be a better alternative rather than the older driver from 3.1.
Not sure the 6,1 has the AGP Driver, likely the CoreEG2.efi.
![Nick [D]vB](/data/avatars/m/1132/1132239.jpg?1552018909){kind=avatar}
Hi all, during my iMac firmware tests I found CoreEG2 can co-exist with the existing AGP driver, so you just inject it before the free space rather than replace the AGP DXE's body. I'd try the MP6,1 version first but if you wanted to try loading / injecting the AMD EFI from the iMac 18,3 (7853E756-CA34-4296-90BB-DEDC91BD64B6) you may want to use the CoreEG2 from that machine instead, you would probably need to strip the existing GOP from the vbios first though?? If you post a dump of your RX580 bios I can try integrating both, is it a dual bios card?
Last edited:
Thanks Nick, on the 3,1 CoreEG2 doesn't seem to do anything.
Code:
E2 00000010 ? - - - - <UNKNOWN> \EFI\CLOVER\driver
E3 00000010 ? - - - - <UNKNOWN> \EFI\CLOVER\driver
E4 00000010 ? - - - - <UNKNOWN> \EFI\CLOVER\driverI'm not sure injecting it into the FW would change that, but maybe it will do something on the 5,1. The 5,1 seems to load AMD GOP roms, but not nVidia GOP roms.
We never get boot screens on the 5,1, but do one the 3,1, with custom GOP enabled boot loaders.
Shame it doesn't work with the 3,1 - some of the older iMacs I tested didn't know what to do with it either. Not sure it could help load standard PC GOPs anyway, probably best get onto the boot-loader developers about the 5,1 issue?
It would still be interesting to test CoreEG2 with the iMac 18,3 GOP on a 5,1 with the RX580. You could inject both into the bootrom but probably best replace the existing GOP in the vbios.
If you post a full vbios dump I will update it, if we are very lucky you might get a bootscreen on one of the outputs, but even if you don't it will be interesting to see what actually gets loaded / linked.
It would still be interesting to test CoreEG2 with the iMac 18,3 GOP on a 5,1 with the RX580. You could inject both into the bootrom but probably best replace the existing GOP in the vbios.
If you post a full vbios dump I will update it, if we are very lucky you might get a bootscreen on one of the outputs, but even if you don't it will be interesting to see what actually gets loaded / linked.
Last edited:
Download This is the rom for the rx580 I have, but it already works for my 3,1.
I don't own a 5,1, but maybe it will be useful if you mod it.
Yes, you can insert any settings into the streams of the NVRAM volume. I'm making a list of useful thing to insert, like SIP disable/NVIDIA web drivers.
@startergo sent me the dump the inserted the AGP driver and the dump from the MATT card after the tests. Clover is writing binary blocks into the first stream of the NVRAM volume and totally erasing the second stream.
Perhaps Clover don't know how to interact with Macs that don't have PRAM, with NVRAM volume into the SPI flash.
[doublepost=1542661806][/doublepost]
It's not UEFITool, I'm using it daily for more than 6 months, done more than a hundred BootROMs for different Macs and never got a problem with it.
The BootROM that you inserted the AGP driver have a valid NVRAM volume, the one from the MATT card is corrupted.
- MP3,1 has a 32Kbit PRAM SPI flash and a 16MBit FWB flash for the BootROM.
- MP5,1 has a 32Mbit SPI Flash for the whole BootROM, NVRAM volume inside it.
- MP6,1 has a 64Mbit SPI Flash for the whole BootROM, NVRAM volume inside it.
Last edited:
What I have noticed during the tests, there is a safe configuration for Clover:
In the SMBIOS section you need to select predefined MacPro 5,1 :
Download DarwinDumper: https://bitbucket.org/blackosx/darwindumper/downloads/
Select (SIP and Antivirus must be disabled)
Run the dump
In the DMI tables 1, 2 and 3 you will find (note I have selected private dumps, so the serials are hidden. For your purposes you should disable the private dumps):
In the NVRAM section you will find:
Substitute the original values for the SMBIOS and RT- Variables page.
Disable all kext injection and tweaks in the other pages.
You may enable debug options in the BOOT page
Save the config file in the Clover folder.
If you leave the SMBIOS page empty (like I did in my last test) you may get NVRAM corruption.
Anyway it is recommended using Refind, Refit or Next Loader as they are designed for Mac unlike Clover.
In the SMBIOS section you need to select predefined MacPro 5,1 :
Download DarwinDumper: https://bitbucket.org/blackosx/darwindumper/downloads/
Select (SIP and Antivirus must be disabled)
Run the dump
In the DMI tables 1, 2 and 3 you will find (note I have selected private dumps, so the serials are hidden. For your purposes you should disable the private dumps):
Code:
System Information
Manufacturer: Apple Inc.
Product Name: MacPro5,1
Version: 0.0
Serial Number: H*********H
UUID: 96********************************0A
Wake-up Type: Power Switch
SKU Number: System SKU#
Family: MacPro
ase Board Information
Manufacturer: Apple Inc.
Product Name: Mac-F221BEC8
Version: Not Specified
Serial Number: J***********C
Asset Tag: 0
Features:
Board is a hosting board
Board is removable
Board is replaceable
Location In Chassis: Part Component
Chassis Handle: 0x0096
Type: Processor+Memory Module
Contained Object Handles: 0
Chassis Information
Manufacturer: Apple Inc.
Type: Tower
Lock: Not Present
Version: Mac-F221BEC8
Serial Number: H*********H
Asset Tag: Not Specified
Boot-up State: Safe
Power Supply State: Safe
Thermal State: Other
Security Status: Other
OEM Information: 0x00000000
Height: Unspecified
Number Of Power Cords: Unspecified
Contained Elements: 0In the NVRAM section you will find:
Code:
4D1EDE05-38C7-4A6A-9CC6-4BCCA8B38C14:FirmwareFeatures
------------------------------------------------------------------------------
000000: 17 54 0c c0 |.T..|
------------------------------------------------------------------------------
4D1EDE05-38C7-4A6A-9CC6-4BCCA8B38C14:FirmwareFeaturesMask
------------------------------------------------------------------------------
000000: 3f ff 1f ff |?...|
------------------------------------------------------------------------------
4D1EDE05-38C7-4A6A-9CC6-4BCCA8B38C14:MLB
------------------------------------------------------------------------------
0000****************************************************************H9A|
------------------------------------------------------------------------------
4D1EDE05-38C7-4A6A-9CC6-4BCCA8B38C14:ROM
------------------------------------------------------------------------------
000***********************************************************..|Disable all kext injection and tweaks in the other pages.
You may enable debug options in the BOOT page
Save the config file in the Clover folder.
If you leave the SMBIOS page empty (like I did in my last test) you may get NVRAM corruption.
Anyway it is recommended using Refind, Refit or Next Loader as they are designed for Mac unlike Clover.
ROMs attached, with just the iMac 18,3 GOP and with that plus the CoreEG2 from MP6,1 or IM18,3. Not expecting any of them to work on the 3,1 tbh but I'd still try it, without clover to start with.
Not sure if CoreEG2 will load on either MP as I think this trick only works from a PCI Rev 3 vbios, could try converting the headers if needs be but would be easier to just inject it into the bootrom.
Startergo it would be good to see if the 5,1 can load the GOP without EG2 first, also test without Clover. All the usual heath warnings apply, no one else try these unless you have a bootrom backup and means to restore it!
Last edited:
This ROM Ellesmere_IM183GOP.rom loaded fine, but I did not see boot screen. Tried next loader (refind) still no boot screen.
The second one Ellesmere_IM183GOP_IM183EG2.rom bricked my ROM on the first boot. No chime at all. Now I don't know how to reflash the video card
[doublepost=1542684150][/doublepost]Removed the AMD card and booted only with GT-120. So the system ROM is not bricked, but is not passing the chime. The question is how to flash the video card back?
Sorry to hear that, I'd assumed it was a dual-bios card. Have you tried booting with both cards installed? Can you drop it into a PC to re-flash it? (non UEFI boot) Failing that you can short the EEPROM at boot to reflash it, can you post a high res photo of the back of the card?
i Imagine the card will be able to be flashed back in a PC
it seems like most DXE drivers from the iMac18,3 etc will cause an older EFI64 Mac to just hard lock up, so if you embed it into the BootROM for example you will effectively brick the machine, so in this case the MP5,1 is loading the EG2 DXE driver and hard locking up.
im actually surprised the Radeon Pro 580 EFI did not do that on its own.
(@tsialex and I found the iMac18,3 out the hard way after he bricked a MP5,1 injecting an iMac18,3s NVMe driver into BootROM and then I did some testing from an EFI shell and discovered the moment the driver is loaded the Machine hard locks up)
it seems like most DXE drivers from the iMac18,3 etc will cause an older EFI64 Mac to just hard lock up, so if you embed it into the BootROM for example you will effectively brick the machine, so in this case the MP5,1 is loading the EG2 DXE driver and hard locking up.
im actually surprised the Radeon Pro 580 EFI did not do that on its own.
(@tsialex and I found the iMac18,3 out the hard way after he bricked a MP5,1 injecting an iMac18,3s NVMe driver into BootROM and then I did some testing from an EFI shell and discovered the moment the driver is loaded the Machine hard locks up)
No, it is a single BIOS. I may try it on my PC's. One of them has Intel Video through the CPU. The other one has a bios flashing feature in the Asus BIOS as long as I can boot to BIOS with 2 cards.
[doublepost=1542714384][/doublepost]
The first ROM I tested worked actually quite well and though I did not get a boot screen, maybe it is worth testing it more with different GOP drivers for refind
I'd just flash it in windows, I think that ASUS feature is only for the system bios. If the card still locks up in the PC you may need to set Legacy / CSM boot mode, if you then can't get into windows you can flash from a bootable USB stick, let me know if you need help with that.
The vbios with CoreEG2 from the MP6,1 is probably a much safer bet, I have used it to load the M380 GOP from the iMac 17,1. You could try it if you're feeling brave but no probs if you don't want to risk it. As you have the MATT card it might be better to inject it into the bootrom anyway.
I'd be interested to know if the first vbios without EG2 actually loaded the GOP, did you see anything different in system profiler? The iMacs have a debug LED that lights up when the video EFI is loaded, don't think the MPs have one for that??
The vbios with CoreEG2 from the MP6,1 is probably a much safer bet, I have used it to load the M380 GOP from the iMac 17,1. You could try it if you're feeling brave but no probs if you don't want to risk it. As you have the MATT card it might be better to inject it into the bootrom anyway.
I'd be interested to know if the first vbios without EG2 actually loaded the GOP, did you see anything different in system profiler? The iMacs have a debug LED that lights up when the video EFI is loaded, don't think the MPs have one for that??
Code:
loadpcirom test.romThe iMac's EG2 driver also hard locked my 3,1. I'll have to see what the others do with my RX580 installed.
Yeh, the MP6,1 version seems like a better fit, but the GOP may actually load without it. Boot screens are a very long shot as I've not patched the framebuffer yet, I'd still try different outputs as one of the DisplayPorts should match.