On Wed, Sep 27, 2023 at 04:31:43PM +0200, Lukas Wunner wrote: > On Wed, Sep 27, 2023 at 03:47:32PM +0300, Mika Westerberg wrote: > > This is not a Linux defect. The firmware is expected to create that > > tunnel so regardless of the "delay" the devices are already back. This > > is not happening. > > I recall that newer chips can be switched over to software connection > manager at runtime. > > Can we determine that the ICM firmware failed to do what it should, > kick it out and let the software connection manager take over? No that's not possible. In Macs it was "partially" possible but even there you lose all the PM and the like. I don't even want to speculate what happens if you run the same on PCs. There is an option to "force" this but I do not recommend this (pass thunderbolt.start_icm=1 in the command line). This is of course completely different with USB4 where software CM is the only option.
