root42Mar 8

I have an issue with mounting a cheap smart watch in #linux: Device is seen by the kernel, but never registered as a storage device /dev/sd*:

usb 1-2: new full-speed USB device number 3 using xhci_hcd
usb 1-2: New USB device found, idVendor=0e8d, idProduct=0002, bcdDevice= 1.00
...
usb-storage 1-2:1.0: USB Mass Storage device detected
scsi host0: usb-storage 1-2:1.0

Show threadroot42Mar 8
Ok, the plot thickens. When booting from the Linux Mint USB stick, the USB ports work. But when booting the installed version, USB ports are broken. Why??
Show threadroot42Mar 9
So my current working hypothesis is: Live Linux Mint uses kernel 6.14, my installation uses kernel 6.17. This is probably what causes the regression?
So I might want to downgrade the kernel...
#linux #linuxmint
Show threadroot42Mar 9
Nope, that was not it. I downgraded to Linux 6.14.0-37, but USB still doesn't work. So it must be something with my installation! But what?
Show threadroot42Mar 9
I am onto something: Attaching an OLD Apple keyboard with a USB-C to A adapter WORKS. But all my USB3.x devices do NOT work!
So how do I tell the USB controller to also work with USB3 devices...? Which works on the live thumbdrive!
Show threadroot42
Yes, so it seems the controller is stuck in EHCI mode, instead of switching to xHCI for USB3 devices. Bummer... how to switch it back...?
Mar 9 at 9:59amWeb
Show threadroot42Mar 9
Next hypothesis: These are Thunderbolt ports. And if a device connects in Thunderbolt mode (USB3/4), it will not work. Only the old USB1 devices work.
boltctl doesn't show anything, except for the controller. So my guess is that something in the bolt configuration is wonky. There IS a bit of a hint in the dmesg about this:
thunderbolt 0000:06:00.0: device links to tunneled native ports are missing!
Show threadroot42Mar 10
So I think I have a workaround. If I do "sudo lsusb -v", the devices start working! Without sudo and without -v nothing happens. It seems sudo lsusb -v does some magic that kickstarts the device detection. What is it and how can I make it happen automatically...?
Show threadAri [APz] SovijärviMar 10
@root42 Reading all this reminds me of a really weird Thinkpad I had years back. It had Thunderbolt ports and those had really weird compatibility issues with random devices in Linux. But only with the TB ports, the standard USB-C ports were fine. I assume the device you have is an Apple laptop, which most likely has even more undocumented and standard breaking behind the scenes firmware magic built into it as per their style.
Show threadroot42Mar 10
@apzpins the Thinkpads used to have messed up controllers. Intel screwed them up and after a certain number of uses they would get bricked. Was the T480s if I remember correctly.
Show threadchebraMar 10
@root42 That's... interesting. Just for fun I opened the lsusb source code and there is actually one place which calls `dumpdev` only when `verblevel > 0` which is increased by `-v` https://github.com/gregkh/usbutils/blob/master/lsusb.c#L3828 Actually the same thing should be happening when you want info of only one device by using `-D` https://github.com/gregkh/usbutils/blob/master/lsusb.c#L3759 I'm not claiming I understand any of this, I just thought it might be easy to see what `-v` does differently.
usbutils/lsusb.c at master · gregkh/usbutils

USB utilities for Linux, including lsusb. Contribute to gregkh/usbutils development by creating an account on GitHub.

GitHub
Show threadSebastiaanMar 9

@root42 Is your USB controller properly listed in `lspci`? And if it is, what kernel options does it use (see with `lspci -s <address of controller> -k`. For example, here's mine:

04:00.3 USB controller: Advanced Micro Devices, Inc. [AMD] Renoir/Cezanne USB 3.1
Subsystem: Advanced Micro Devices, Inc. [AMD] Renoir/Cezanne USB 3.1
Kernel driver in use: xhci_hcd
Kernel modules: xhci_pci

Show threadroot42Mar 9

@sebastiaanfranken Seems so:

07:00.0 USB controller: Intel Corporation JHL6540 Thunderbolt 3 USB Controller (C step) [Alpine Ridge 4C
2016] (rev 02)
Subsystem: Intel Corporation JHL6540 Thunderbolt 3 USB Controller (C step) [Alpine Ridge 4C 2016]
Kernel driver in use: xhci_hcd

Show threadSebastiaanMar 9

@root42 Ok, so that's good. Also the fact the proper hcd (xhci_hcd) module is loaded is good. Anything related to it in your system journal? And if you want you can get more output / info with

> sudo lspci -s 07:00.0 -k -v

Show threadroot42Mar 9
@sebastiaanfranken First the lspci output
Show threadroot42Mar 9
@sebastiaanfranken Then syslog:
https://privatebin.net/?83d299418ae3e05a#2ZQSiucLStRySg7yKv35MQ799oqivhZVM8ftmKLAyxQ8
PrivateBin

Visit this link to see the note. Giving the URL to anyone allows them to access the note, too.

PrivateBin
Show threadrevenge of the boltzmann brainMar 9
@root42 Check the pins in the connectors. I had a problem like this where in a thumb drive some pins belonging to USB3 were bent / damaged and it would not work in a USB3 port. Although it worked fine in a USB2 port, those 4 pins were not damaged.
Show threadrevenge of the boltzmann brainMar 9
@root42 with USB-C it could also be the wrong type of cable.
Show threadroot42Mar 9
@robert no. It works in the Live thumbdrive. Only the proper install on the ssd is botched. The ports work fine.
Show threadroot42Mar 9
@robert the USB C thumbdrive connects without a cable. It works in the live system, not in the installed system.