Hey all,

I come here with a problem that I haven't been able to solve after various tries:

I work on a computer with KDE neon (latest verion - 5.26) in a DELL Precision Tower 7290, in which we asked the company to add a PCI Thundebolt card (0000:03:00.0 PCI bridge: Intel Corporation JHL6540 Thunderbolt 3 Bridge (C step) [Alpine Ridge 4C 2016] (rev 02) (prog-if 00 [Normal decode])). This week, we wanted to install the 10gb network adaptor by QNAP QNA-T310G1S, which should be connected on the Thunderbolt port.

I first tested the QNAP adaptor in my Windows computer, and it worked fine, so the hardware is ok, and then I tried in the KDE computer. However, I can't make it work. I connect it to the thunderbolt port and on dmesg I get this message and it doesn't show as a new device to be added/authorized (my bios settings is to ask for authorization):

I tried installing the drivers according to the company's instructions (i.e. using this driver), but not working. I'm wondering if instead could be a problem with the Thunderbolt card itself...

A few things I noticed: when I turn on the computer, the thunderbolt subsystem is always off and I have to turn it on manually with sudo boltctrl; I tried connecting a display cable to the TB card, but it doesn't show any image.

I saw some people suggesting BIOS or firmware updates, but nothing concrete. I also tried boltctl list, and bolt ctl domains, but it doesn't show anything! One thing that might be relevant is that, even without showing anything in the system, the QNAP adaptor turns on and if I connect my cellphone to the thunderbolt port, it starts charging...

Would anyone be able to help me?

.Hello, eduardacenteno & welcome to LQ. This is above my pay grade, but here goes:

Just reading your post, you have been logical and thorough in what you did. For me, I'd check the Thubderbolt card, as you have not seen evidence of that working. I'd examine the bootup logs to see

1. Is there a kernel module sought or used?
2. Is there firmware wanted or loaded?
3. Can you get anything working through it? You only need voltage to charge a mobile phone.

Your having to start this manually is bad. Your box sounds like a serious enterprise, so I'm presuming RHEL or similar. They would have an init script for that card.

Having to run 'sudo boltctrl' sounds a bit like having to run 'ifconfig eth0 up.' That brings the interface up, but doesn't set anything. What output do you get from that? Does anything acknowledge it?
Let's go through the basics:

1. Do you have a pcie-4x slot with a decent supply of lanes?
2. Does your m/b have a thunderbolt header & is it connected?
3. Have you anything working through this already?

Hey! thanks!

I have some update that is probably very relevant:

So, I followed the instruction in this post (changing the GRUB_CMDLINE_LINUX_DEFAULT); and It improved: the device is now recognized, but not completely. I still get this message:

[ 52.683880] Initializing XFRM netlink socket

[ 92.294428] pci_bus 0000:27: busn_res: can not insert [bus 27-26] under [bus 03-26] (conflicts with (null) [bus 03-26])

[ 92.294501] pcieport 0000:03:00.0: PCI bridge to [bus 27-26]

[ 92.294516] pcieport 0000:03:00.0: bridge window [mem 0x90100000-0xa02fffff]

[ 92.294527] pci_bus 0000:27: busn_res: [bus 27-26] end is updated to 27

[ 92.294533] pci_bus 0000:27: busn_res: can not insert [bus 27] under [bus 03-26] (conflicts with (null) [bus 03-26])

[ 92.294545] pcieport 0000:03:00.0: devices behind bridge are unusable because [bus 27] cannot be assigned for them

[ 92.294555] pci_bus 0000:27: Allocating resources

[ 92.294560] pcieport 0000:03:00.0: bridge window [io 0x1000-0x0fff] to [bus 27] add_size 1000

[ 92.294567] pcieport 0000:03:00.0: bridge window [mem 0x00100000-0x000fffff 64bit pref] to [bus 27] add_size 40000000 add_align 100000

[ 92.294580] pcieport 0000:03:00.0: BAR 15: no space for [mem size 0x40000000 64bit pref]

[ 92.294585] pcieport 0000:03:00.0: BAR 15: failed to assign [mem size 0x40000000 64bit pref]

[ 92.294589] pcieport 0000:03:00.0: BAR 13: no space for [io size 0x1000]

[ 92.294593] pcieport 0000:03:00.0: BAR 13: failed to assign [io size 0x1000]

[ 92.294598] pcieport 0000:03:00.0: BAR 15: no space for [mem size 0x40000000 64bit pref]

[ 92.294602] pcieport 0000:03:00.0: BAR 15: failed to assign [mem size 0x40000000 64bit pref]

[ 92.294605] pcieport 0000:03:00.0: BAR 13: no space for [io size 0x1000]

[ 92.294609] pcieport 0000:03:00.0: BAR 13: failed to assign [io size 0x1000]


Now the QNAP device is recognized well on boltctl but still cannot make it a network connection. It should appear when I use ip addr show, but isn't there!


● QNAP Systems, Inc. QNA-T310G1S
├─ type: peripheral
├─ name: QNA-T310G1S
├─ vendor: QNAP Systems, Inc.
├─ uuid: 00d30a59-0928-5600-ffff-ffffffffffff
├─ generation: Thunderbolt 3
├─ status: authorized
│ ├─ domain: c7010000-0070-6718-a3a0-098f4e11e102
│ ├─ rx speed: 40 Gb/s = 2 lanes * 20 Gb/s
│ ├─ tx speed: 40 Gb/s = 2 lanes * 20 Gb/s
│ └─ authflags: none
├─ authorized: Fri 18 Nov 2022 16:53:08 UTC
├─ connected: Fri 18 Nov 2022 16:52:12 UTC
└─ stored: Fri 18 Nov 2022 16:53:08 UTC
├─ policy: auto
└─ key: no

At a glance, it seems like 'ifconfig etho up'. The thing is talking, but cannot be allocated I/O or interrupts so it cannot function. The driver/module is not agreeing with the system on any resources. My first suggestion would top get support or get your money back.

Chase up your logs - the kernel is very verbose. Look for a module being loaded. Examine whatever package gave you 'boltctl.' If it a script, or a binary? If it's a script, post or pastebin it. Any other useful utilities there, like 'boltdiag'? Any config files? Any software?