USB 2.0 devices not recognized on USB 3.0 ports after remap and 11.3.1 update

[UtterDisbelief]

My bad.
Hopefully this one works.

Yes, that's fine.

Well, as I mentioned before USBInjectAll.kext is active and ruling the roost:

​

and:

So, try as I suggested in post #7 and things will change - hopefully for the better.

 

[Navigator16]

No change.
Foremost, thanks for your guidance and patience working this thru with me.

Kernel entry in config.plist:
I deleted the USBInjectAll.kext entry. Before I just had it disabled.
USBPorts.kext is enabled with no Executable Path.
XhciPortLimit is set to false.

The Kexts folder contains USBPorts.kext and xxUSBInjectAll.kext. I used a prefix of "xx" just in case I needed the USBInjectAll.kext at a later time.

 

[Navigator16]

Something else I just found.... where is RM,USBInjectAll Boolean True coming from?
There are no references to USBInjectAll.

 

[UtterDisbelief]

Something else I just found.... where is RM,USBInjectAll Boolean True coming from?
There are no references to USBInjectAll.


View attachment 518512

Sorry - and with great respect - your previous post does not ring true ...



Perhaps you have a duplicate in the Library/Extensions folder?

If USBPorts.kext was active and USBInjectAll.kext was disabled you would not get 26x ports, which is what you latest IOReg update still shows.

If you had neither kext you would get 15x ports.

If you had USBPorts.kext enabled but no XhciPortLimit quirk and no USBInjectAll.kext, you would have configured ports - assuming the System-Definition is as you have encoded in the kext.

 

[Navigator16]

It keeps getting better and better....
My machine has 3 SSD's mounted.
One of the SSD's is a "Test" System currently running macOS 11.3.1 and you know it has USB problems...
Another SSD is a "Production" system currently running macOS 11.2.3 with no USB issues (so far).
The 3rd QOne is used strictly for Time Machine backups.
The

 

[Navigator16]

The TEST system has verbose turned on with a 5 second delay.
The PROD system has verbose turned off with a 3 second delay.
I've noticed when booting into the PROD system it goes into verbose mode -- it shouldn't.
I changed the PROD EFI system boot delay to 10 seconds and booted into the TEST system.
The TEST system waited 10 seconds then booted into its TEST OS just fine -- it should only have waited 5 seconds, not 10.

Both the TEST and the PROD SSD's are identical Samsung V-Nand SSD 970 EVO NVMe PCIe M.2 2280 - 500 GB SSD's. The only thing different about them is their serial #'s and their mount points on the mothberboard.

Is it possible that the OC boot loader is confused which EFI to boot to?

 

[P1LGRIM]

Is it possible that the OC boot loader is confused which EFI to boot to?

It is possible that you need to clear NVRAM between booting different disks.

 

[Navigator16]

The boot issue is resolved. In BIOS the Boot Option Priorities were disabled and the only Boot option available was something called Opencore. I never questioned this before as I could launch into either TEST or PROD without any problems but apparently the PROD EFI was being loaded every time. Occasionally BIOS has an issue which requires an immediate reboot but I'm able to get around this problem and now can selectively choose which SSD I wish to boot to: TEST or PROD.

PROD (11.2.3) is stable with no issues.
TEST (11.3.1) now fails on the following command:
Virtual IOReturn IONVMeController :: CreateSubmissionQueue (uint16_t, uint8_t) :: 2844: SQ index=1 entrysize=64

In the TEST config.plist I added USBPorts.kext and cleared the Executable path, disabled USBInjectAll.kext and set XhciPortLimit to false. Does anyone know why I am now getting an NVMe error?