Re: [PATCH] thunderbolt: Fix PCIe device enumeration with delayed rescan

From: Jayi Li

Date: Sun Feb 08 2026 - 20:31:23 EST



在 2026/2/4 20:37, Mika Westerberg 写道:
On Wed, Feb 04, 2026 at 10:37:58AM +0800, Jayi Li wrote:
在 2026/2/3 18:07, Mika Westerberg 写道:
On Tue, Feb 03, 2026 at 06:00:06PM +0800, Jayi Li wrote:
I'm not sure if this is relevant to this issue, but sharing just in case.
Thanks for sharing!

It could be. What device this is?
The device is Targus DOCK221.
Is the host Intel or AMD (well can you share bit more details about the
topology)? Then if you block runtime PM of the PCIe Downstream Port that
leads to the TB3 device like:

# echo on > /sys/bus/pci/devices/DEVICE/power/control

Does it work?
The host is ASMedia.
Ah okay. I don't have any experience with ASMedia host. What system/laptop
comes with that?

 ASMedia Host (0-0, domain0)
         |
        └─ Port 3 ──→ Thunderbolt 3 Dock (0-3)

I tried disabling runtime PM by writing 'on' to power/control for the
downstream port, but it did not work.
I think that's integrated into the SoC, right? So it's a PCIe root port
used for tunneling. What does 'sudo lspci -vv' say?
Thanks for your help. The issue I encountered occurred during a testing phase. I will try to verify it later on an Intel/AMD host system and provide updates if available. Sorry that I can't offer more details at the moment.