Re: [BUG] thunderbolt: USB4 v2 TBGAA tunnel creation crash in TMU enhanced uni-directional mode

[Mika Westerberg <mika.westerberg@linux.intel.com>]

Hi,

On Wed, May 27, 2026 at 09:19:56PM +0200, Sebastian Loscher wrote:
> Hey Mika,
> 
> Please find attached some more tbtraces (and dmesg however I think they tell
> you less of a story).
> 
> I have the previously used cable and repeated the usual boot, plugin. Then I
> dumped the files and rebooted and repeated with a 2nd cable (it's from a three
> months old Lenovo and it charges with 140W, says TB in the package; the
> original cable came with the dock and was only unpacked the day before I
> reached out to you). I attached two different dumps for the 2nd, no reboot in
> between.
> You wondered about the initial couple of disconnects in the logs and blamed the
> cable - rightfully so - but here are some more details, maybe not fully obvious
> from the logs:
> Onbiously there is a GPU in the dock that runs at 100% fan speed, expected of
> course no PC tells it to do otherwise. When I plug the USB cable in there is
> 50% chance that the fans go down in speed for a second during the initial
> handshake (did so with tbtrace_cable1 and cable2_2 | but fails at speed
> negotiation) or it just keeps spinning high (tbtrace_cable2) which seems like,
> at least when I watched dmesg, that it doesn't go beyond the initial handshake
> and disconnects right away. That was, in the previous test, for me the reason
> to plugin it in and out multiple times until I could ‘hear’ something more was
> happening.
> 
> A bit adventurous description but maybe the extra info is worth it.

I noticed in your dmesg you have pcie_port_pm=off, can you remove that from
the command line, and add "thunderbolt.dyndbg=+p" instead and try again?
Then provide full dmesg?