noInk
I launched under Windows 2008 R2, "ReadBoot.bat F: SECTOR", where - F: my flash drive with DUET (EDK2015).
The DBR.bin file (512 bytes) was created, and the MBR.bin file was of zero length.

@vasyosuol_24

Until the MBR is empty and the DBR malformed DUET won't boot from that USB.
The application could also became incompatible or I could have made\uploaded a wrong archive.
Anyway attach the DBR chance are that right now the GenBootSector is failing to read the MBR duo the many changes Microsoft and Intel silently introduced in these past year, in both software and hardware.

noInk
I tried to start the process (ReadBoot.bat) in Windows 10 (LTSC 2019, Microsoft Windows [Version 10.0.17763.107]) - everything is the same, MBR.bin is empty. The CRC (MD5) files of DBR.bin are the same in both cases, 2008R2 & LTSC 2019.
https://www.upload.ee/files/11900543/MBR_DBR.rar.html
I tried to recreate the boot flash in Windows 10 - everything is the same as in Windows 2008R2.
What else can you try?

@vasyosuol_24

You need to use some hexadecimal utility to check the RAW data on the USB DISK and READ the MBR out, should be another 512 byte on start of the physical disk.
The DBR seems fine so, the MBR is probably right too.

noInk
Suppose incorrect / damaged boot flash drive? How can this be? They, flash drives are different, I used several different USB-flash drives. The OSs under which USB-flash drives were created are also different. And the computers themselves are different, damage to the USB-flash drive is eliminated. I see the problem so far for some reason that the bootable USB-flash DUET-REFIND on my hardware is not created correctly, but how will the RAW data check on the USB drive help me in this case? And what utility (example) can we talk about?
It seems that one could try a different way of generating DUET-REFIND USB-flash, for example, a backup of a known-working USB-flash (DUET-REFIND) by third-party programs. Say symantec ghost or any similar.
Is it possible to use a completely different bootable medium, such as a CD / DVD?

@vasyosuol_24

Since the DBR seems correct I'm assuming it's an issue with the board and the way this DUET version work.

Look, this will somehow help you determine if the MBR is correct for the booting sequence of DUET, nothing else.
Download HxD and locate the EXTRA TAB select OPEN DISK.
On the new form select the "Physical Disk" should be under "Logical Disk".
With only the DUET USB as removable attached select the "REMOVABLE DISK 1" and click OK with the READ ONLY option checked.
Now select the 0 - 1FF block and copy that 512 BYTE portion on a new file.
That portion is the actual MBR of the USB, the other (DBR) is on the removable "Logical Disk" and it's partially tied with the error somehow you described.

M.B.R. .S.t.a.r.t.!
B.S.t.a.r.t.
 
 

If your board read "BStartart!" and die, that's the problem. 0 Space/Align = memory lock.

Hi!

I would one more time to thank all helpful people here, especially @noInk for step by step instruction and help, Lost_N_BIOS for advices, SkOrPn, and Fernando.
It wasn't easy, and it take me longer than I would like, but my new NVMe drive works without a problem on Gigabyte Technology Co. Ltd. Z68P-DS3 (Socket 1155) with sandy bridge CPU using duet/refind method provided here.
This motherboard is launching quite long (memory tests, after this - AHCI etc - and thats on fast boot) so additional 2-3 seconds for duet are really not a problem.
I would like to add, that computer I use had last clean install around windows 8. After this It just adding new versions and patches. I don't like t to do this, and do this very seldom - and I think majority of people here not doing it for living so -
My biggest problem was my ignorance on this topic and not STICKING PRECISELY TO THE INSTRUCTION.

I used step by step solution provided by noInk. I downloaded and used "Build 2020"

What problems I had and what helped:

- USBsetup.exe didn't saw my USB drives.
It took me most of my time, and I tried to do this using command line as in youtube tutorials made by SkOrPn. If it is like this, find another computer and make start diske here.
It was probably bug in windows (after reinstall my windows 10 recognise USB keys as removable, so it wasn't hardware problem)





- My DUET/REFIND disk was stucking with red Rerror! message
I just added additional file provided somewhere in this thread by noINK, and it finally went to menu.

DUET/REFIND didnt saw my existing WIN 10 but I'm sure there is remedy for this, but I didn't search for it. I'm sure its because it was working in Legacy mode
I made new windows instal USB using Rufus ON DIFFERENT USB. (no problem here)

I tried GPT AND MBR but windows instaler always refused to install on new disk




- I finally opened case and disabled all additional disks
- NVME should be GPT, but installer still say that he could not use this partition or create new one. But when It was only one disk, I was able to delete this partition and make new one from the same menu (I highlited where is this function - here it is disabled - maybe because of MBR disk or maybe because of other disk IDK)

Windows 10 Installing procedure wasn't exactly painless.

-When instaler restarts 1st time - youll need to remove USB with windows install.

- I had one full stop (black screen and nothing for long time) during install. I just restarted computer and installation was proceeding as nothing happens.

That's all folks. Sorry it is in so weak english and not very precise, but frankly I'm quite exhausted and I'm not really good with hardware and its nomenclature (and motherboard naming its a hell ;))
I hope my observations will help someone, like this thread and nice helpful people here helped me. I hope world will thank you in some way too :).

@vasyosuol_24

I've sort of dirty quick fixed the GenBootSector utility again. Data sector division has been set static to FAT32 cluster count value.
If you re-download and unpack within the DUET_EDK2015_REFIND directory this archive the following command from an Admin command prompt should now work correctly on any updated windows version with unavailable FAT16 (or less) support.

ReadBoot Drive_Letter: SECTOR
 
 

Locate the new DBR and MBR files and attach\post or PM the files.


(20-06-2020)
UTILITY: Adjusted the cluster count to FAT32 value.

Zitat von noInk im Beitrag #803

---

@vasyosuol_24

I've sort of dirty quick fixed the GenBootSector utility again. Data sector division has been set static to FAT32 cluster count value.

---

@vasyosuol_24

The behavior doesn't change because it was already an requirement for the USB to be FAT32.
Both DBR and MBR are correct, unfortunately like I wrote it's a board BIOS implementation issue with the way the memory hold together.
Unless the bypass to the issue is proposed and documented enough to be quickly implemented I'm not going to look at it.
GIGABYTE (PHOENIX|AWARD) with source code and tool availability decided to make an entirely separate build.
You need to find an alternative, or propose the workaround.

noInk
That is, as I understand it, with DUET-REFIND I have no way?
This is not such behavior on the only board - I have at least 4 of them

@vasyosuol_24

The only option I can quickly give is to look for the NVME option ROM to determine the outcome.

noInk
Understood thanks. Apparently not fate. Thank you for your help and time spent.

@noink


i got it finally. Thanks for your patience.

Now i am able to reproduce the error.

It is the "EuP Ready" must be "Disabled" and VT-d musst be disabled too.

The big problem was if i changed the VT-d and press F10 nothing changed the same long load time and windows got stucked. I had to change the value save and unplug the main power cable and boot again. Holy sh**!

Now i have 17 sec. I came form 60 sec.


without DUET it is:

The orginal SMBIOS is all ready 2.5.

with DUET 2015 it is 2.5

with the older 2015 version from march it is 3.0


Are the improvements in the versions not usefull ?



I read something on post #620

VT is not required to be disabled ( disabling or enabling certain bios setting, unless prepared before may require a full OS install).

but then someone wrote in post #749

Zitat von NPina im Beitrag #749

---

This is Windows compatibility or (*force us buy new hardware), this has nothing to do with DUET, DUET do his job whit success.
I think the latest version compatible with our old* Intel VT-d is the Windows 10 version 1803 or below
Our Intel VT-d feature is too old* my friend
We cannot install any version of windows 10 above 1803 with VT-d Enable on our machines, in anyways.

Can you give me your numbers, readings and writings to give me an idea?
I think it will be around 1200MB's ~ 900MB's, because of the PCI-E 2.0 limitation...

Thanks and regards

---

@mujuli

Larger table require actual data to to fulfill the query. There no macro filler implemented on DUET to fulfill the missing entry\data size.
The 2.5 version as been selected for compatibility purpose.
A board with larger table will be either reduced to 2.5 or uses it's own higher version.

About VT: both are true it depend by the CPU/CHIPSET/BIOS/OS and how the specification get implemented and exposed.


If your board expose the VT (x)-(D) feature try to set those IOV feature individually.
On mine DX58SO2 I'm using these setting:
Interrupt Remapping ENABLED
FLR Capability DISABLED
ALS DISABLED

Find also the EXECUTE DISABLE BIT and set it to ENABLE


Good you found the related option, some register require a full power cut to wear down but I don't understand the use-case and the working scenario as your expectation.
Your board might have inverted meaning on it's settings. BIOS setting will control the BOARD behavior. Check on the manual the meaning of the option.