Bug#1101195: linux-image-6.1.0-32-amd64: Dell Venue 11 Pro 7139 fails to detect dock's ethernet port when connected at boot

[Salvatore Bonaccorso <carnil@debian.org>]

Control: tags -1 + moreinfo

Hi Patrick

On Mon, Mar 24, 2025 at 12:34:34PM -0400, Patrick Hibbs wrote:
> Package: src:linux
> Version: 6.1.129-1
> Severity: normal
> X-Debbugs-Cc: hibbsncc1701@gmail.com
> 
> Dear Maintainer,
> 
> The Dell Venue 11 Pro 7139 has a dock (Dell K10a (proprietary)) that is used to
> provide additional I/O to the device. Most of that functionality works out of
> the box under linux, with the exception of the dock's ethernet port. (A ASIX
> AX88179 USB 3.0 Gigabit Ethernet chipset embedded in the dock itself.)
> 
> That port only works if the Venue is booted without being connected to the dock
> during POST / initial kernel start. Otherwise linux will fail to detect the
> dock's ethernet port at all, (no dmesg output at all), despite everything else
> on the dock working. No amount of unplugging and replugging in the dock will
> fix it. (Even unplugging the dock from it's own separate power source does
> nothing to resolve the issue.) In this case the only solution is a full
> shutdown and reboot. (With the dock disconnected from the Venue during the
> reboot.)
> 
> This puts a lot of repetitive strain on the dock's spring loaded connector as
> it means that everytime linux reboots, the Venue must be removed from the dock
> and then reattached for the ethernet port to continue working.
> 
> No amount of tweaking settings in the Venue's BIOS / firmware seems to fix it.
> (Fastboot disabled, UEFI Secure boot disabled, Disable Legacy ROMs, etc.) And
> there is no option in the BIOS / firmware to disable network booting entirely.
> The BIOS / Firmware just seems to be putting the USB NIC into an unknown state
> that linux can't recover from, if it detects the card during POST. A state that
> is unique to the dock NIC itself. Plugging in a different USB ethernet device
> (even one with a similar chipset) works as expected even if that different non-
> dock NIC is connected during POST.
> 
> I've attached some dmesg output from both a boot with the dock connected and
> without (and what booting without looks like after linux is booted and the dock
> attached.)

As I understand your description this is *not* a regression from a
previously working 6.1.y version correct? 

You could test please newer kernel versions either from backports or
directly from trixie/unstable and can you report back if those fix the
problem?

Regards,
Salvatore