Getting the below error that I think is to do with headers from searching the web but I don’t understand how to fix.
jimmy@jimmy-NUC12:~$ ls /dev/apex_0
ls: cannot access '/dev/apex_0': No such file or directory
All other steps worked up to that point. coral-linux
NUC12 PRO
OS - Ubuntu 22.04.2 LTS
Docker version 24.0.4, build 3713ee1
I’m very new to Linux.
This is a new setup and only being used for Frigate at this point.
Thank you
“ls” the /dev directory and see what shows
ls /dev
jimmy@jimmy-NUC12:~/Documents/frigate$ ls /dev
autofs drm_dp_aux0 hugepages i2c-6 loop13 loop-control port sg0 tty12 tty25 tty38 tty50 tty63 ttyS17 ttyS3 vcs vcsu
block drm_dp_aux1 hwrng i2c-7 loop14 mapper ppp shm tty13 tty26 tty39 tty51 tty7 ttyS18 ttyS30 vcs1 vcsu1
bsg drm_dp_aux2 i2c-0 i2c-8 loop15 mcelog pps0 snapshot tty14 tty27 tty4 tty52 tty8 ttyS19 ttyS31 vcs2 vcsu2
btrfs-control drm_dp_aux3 i2c-1 i2c-9 loop16 mei0 psaux snd tty15 tty28 tty40 tty53 tty9 ttyS2 ttyS4 vcs3 vcsu3
bus ecryptfs i2c-10 initctl loop17 mem ptmx stderr tty16 tty29 tty41 tty54 ttyprintk ttyS20 ttyS5 vcs4 vcsu4
char fb0 i2c-11 input loop18 mqueue ptp0 stdin tty17 tty3 tty42 tty55 ttyS0 ttyS21 ttyS6 vcs5 vcsu5
console fd i2c-12 kmsg loop2 net pts stdout tty18 tty30 tty43 tty56 ttyS1 ttyS22 ttyS7 vcs6 vcsu6
core full i2c-13 kvm loop3 ng0n1 random tpm0 tty19 tty31 tty44 tty57 ttyS10 ttyS23 ttyS8 vcsa vfio
cpu fuse i2c-14 log loop4 null rfkill tpmrm0 tty2 tty32 tty45 tty58 ttyS11 ttyS24 ttyS9 vcsa1 vga_arbiter
cpu_dma_latency gpiochip0 i2c-15 loop0 loop5 nvme0 rtc tty tty20 tty33 tty46 tty59 ttyS12 ttyS25 udmabuf vcsa2 vhci
cuse hidraw0 i2c-2 loop1 loop6 nvme0n1 rtc0 tty0 tty21 tty34 tty47 tty6 ttyS13 ttyS26 uhid vcsa3 vhost-net
disk hidraw1 i2c-3 loop10 loop7 nvme0n1p1 sda tty1 tty22 tty35 tty48 tty60 ttyS14 ttyS27 uinput vcsa4 vhost-vsock
dma_heap hidraw2 i2c-4 loop11 loop8 nvme0n1p2 sda1 tty10 tty23 tty36 tty49 tty61 ttyS15 ttyS28 urandom vcsa5 zero
dri hpet i2c-5 loop12 loop9 nvram sda2 tty11 tty24 tty37 tty5 tty62 ttyS16 ttyS29 userio vcsa6 zfs
Do you have dual TPU version?
I would reboot.
Check again
If still no good go through coral instructions steps again
Thanks for the reply.
Did the coral instruction again, restarted multiple times. Removed the card, boot up/down, reinstalled it, did the coral install and still same error.
It is a M.2 A+E key single TPU.
One of the requirements is " *All systems require support for MSI-X as defined in the PCI 3.0 specification" how can this be checked?
Not sure but i am certain the nuc will be Ok with coral
Are you certain it is properly installed physically? Have you tried power off/remove/reseat/power on?
I forget command but there should be linux command to check for mounted devices. I will look for it.
Just FYI, “man” is handy command. It provides a manual for a linux command
“man ls” provide details of ls command for example
“man cat” provides details of cat command
Try the lshw command. This may show you if coral is mounted and seen by system i think. I will test this on my system later today. I imagine you will see some device with google or some other obvious indication of it being the coral
EDIT
“sudo lshw” shows below for coral. I believe it will show device even if driver is not present.
*-pci:0
description: PCI bridge
product: ASM1182e 2-Port PCIe x1 Gen2 Packet Switch
vendor: ASMedia Technology Inc.
physical id: 3
bus info: pci@0000:83:03.0
version: 00
width: 32 bits
clock: 33MHz
capabilities: pci normal_decode bus_master cap_list
configuration: driver=pcieport
resources: irq:43 ioport:fbc00000(size=2097152)
*-generic
description: Non-VGA unclassified device
product: Coral Edge TPU
vendor: Global Unichip Corp.
physical id: 0
bus info: pci@0000:84:00.0
version: 00
width: 64 bits
clock: 33MHz
capabilities: bus_master cap_list
configuration: driver=apex latency=0
resources: irq:42 memory:fbd00000-fbd03fff memory:fbc00000-fbcfffff
*-pci:1
description: PCI bridge
product: ASM1182e 2-Port PCIe x1 Gen2 Packet Switch
vendor: ASMedia Technology Inc.
physical id: 7
bus info: pci@0000:83:07.0
version: 00
width: 32 bits
clock: 33MHz
capabilities: pci normal_decode bus_master cap_list
configuration: driver=pcieport
resources: irq:44 ioport:fba00000(size=2097152)
*-generic
description: Non-VGA unclassified device
product: Coral Edge TPU
vendor: Global Unichip Corp.
physical id: 0
bus info: pci@0000:85:00.0
version: 00
width: 64 bits
clock: 33MHz
capabilities: bus_master cap_list
configuration: driver=apex latency=0
resources: irq:42 memory:fbb00000-fbb03fff memory:fba00000-fbafffff
Missed your reply. Thanks for the man command
I have started over, install Debian and same error.
here are the results for the command. It shows the Coral on pci:2, looks to be missing the driver=apex? Does the Unclaimed mean anything?
Thanks
*-pci:2
description: PCI bridge
product: Alder Lake PCI Express x1 Root Port #10
vendor: Intel Corporation
physical id: 1d.1
bus info: pci@0000:00:1d.1
version: 01
width: 32 bits
clock: 33MHz
capabilities: pci pciexpress msi pm normal_decode bus_master cap_list
configuration: driver=pcieport
resources: irq:124 ioport:6000000000(size=2097152)
*-generic UNCLAIMED
description: Non-VGA unclassified device
product: Coral Edge TPU
vendor: Global Unichip Corp.
physical id: 0
bus info: pci@0000:03:00.0
version: 00
width: 64 bits
clock: 33MHz
capabilities: pciexpress msix msi pm cap_list
configuration: latency=0
resources: iomemory:600-5ff iomemory:600-5ff memory:6000100000-6000103fff
The coral instructions explain how to install apex driver. I’m completely guessing here but i believe the driver is bit being install or used. That also may be reason for unclaimed by hardware and linux is still something i dont clearly understand.
Not sure where to point you other then telling you to go through install instruction for coral and its drivers. I believe the answer is there.
Just realized but did you update linux?
sudo apt update
sudo apt upgrade
sudo reboot
I read somewhere that the unclaimed means device isnt recognized by kernel. I presume you update already but gotta verify.
I was just logging on to say that. It now displays.
Thank you so much for your help.
What was the cause/fix?
Cause great question. Everything is new, might be a NUC thing.
Solution was disable secure boot in bios. Have no clue what is does but hope it doesn’t bite me in the rear later.
Secure boot I think validates drivers and software at boot.
Maybe it requires signed software and coral drivers not signed or cannot be validated. According to the link it was a NUC thing
I still can’t get the access to /dev/apex_0.
Safe Boot is also deactivated and all steps to this point went without failure.
My goal is to use my Coral m-2 single TPU in Scrypted (as LXC in Proxmox).
UPDATE: I am sorry, I finally managed by using this instruction and also with the comment “go into the /home folder” before that “git clone” command.
