Arch Linux

jwintersinger

Member

Registered: 2024-06-18

Posts: 4

Framework 13 AMD problem: USB & bluetooth stopped working in Arch

Hello! I'm having a weird issue with my Framework 13 AMD running Arch Linux. After leaving the system unused and unplugged for a week, USB and Bluetooth stopped working, but only in Arch Linux -- Windows 11 is fine.

• Specifically, I'm trying to use a wireless Logitech keyboard and mouse, each with separate USB dongles; they are not connected via Bluetooth.

• The two dongles are plugged into my Dell monitor, which is connected via USB-C to my laptop.

• This configuration worked fine until I left the system unused and turned off for a week. It still works without issue in Windows 11. The same hardware works fine with my MacBook Pro M3.

• I am running the latest BIOS (3.05). The issue first appeared while running kernel 6.9.1, and has persisted since upgrading to kernel 6.9.5. The same problem occurs in kernel 6.6.34 LTS.

• I tried resetting the BIOS settings.

• I also tried using the "disconnect battery" option in the BIOS, then leaving the system off for 20 minutes.

• Based on other related reports, I also tried rebuilding my initrd without Bluetooth modules. This did not help.

• I tried using a cheap USB wireless mouse instead, which also didn't work. I also tried without success to use the Logitech mouse and keyboard with a separate USB-C hub without connecting my monitor. Notably my laptop's webcam also no longer works, suggesting something is wrong with the entire USB subsystem.

`dmesg` shows voluminous output related to Bluetooth and USB repeatedly resetting. While the system is running, the following messages are logged approximately every three seconds:

[  885.359092] Bluetooth: hci0: HW/SW Version: 0x008a008a, Build Time: 20240409163648
[  885.538317] usb 1-1: USB disconnect, device number 8
[  885.656405] usb 1-4: USB disconnect, device number 9
[  885.657116] usb 1-5: USB disconnect, device number 10
[  885.678149] Bluetooth: hci0: sending frame failed (-19)
[  885.678166] Bluetooth: hci0: Failed to send wmt patch dwnld (-19)
[  885.678213] Bluetooth: hci0: Failed to set up firmware (-19)
[  885.678218] Bluetooth: hci0: HCI Enhanced Setup Synchronous Connection command is advertised, but not supported.
[  885.715560] usb 3-1: USB disconnect, device number 88
[  886.894948] hub 1-0:1.0: USB hub found
[  886.894979] hub 1-0:1.0: 5 ports detected
[  886.899970] usb usb1: authorized to connect
[  886.900266] hub 2-0:1.0: USB hub found
[  886.900278] hub 2-0:1.0: 2 ports detected
[  886.902100] usb usb2: authorized to connect
[  886.961595] hub 3-0:1.0: USB hub found
[  886.961628] hub 3-0:1.0: 1 port detected
[  886.963287] usb usb3: authorized to connect
[  886.963476] hub 4-0:1.0: USB hub found
[  886.963496] hub 4-0:1.0: 1 port detected
[  886.964434] usb usb4: authorized to connect
[  886.978318] hub 5-0:1.0: USB hub found
[  886.978398] hub 5-0:1.0: 1 port detected
[  886.980133] usb usb5: authorized to connect
[  886.980451] hub 6-0:1.0: USB hub found
[  886.980493] hub 6-0:1.0: 1 port detected
[  886.983226] usb usb6: authorized to connect
[  886.995198] hub 7-0:1.0: USB hub found
[  886.995247] hub 7-0:1.0: 1 port detected
[  886.997039] usb usb7: authorized to connect
[  886.997369] hub 8-0:1.0: USB hub found
[  886.997410] hub 8-0:1.0: 1 port detected
[  886.999739] usb usb8: authorized to connect
[  887.044805] usb 1-1: new full-speed USB device number 11 using xhci_hcd
[  887.111466] usb 3-1: new high-speed USB device number 89 using xhci_hcd
[  887.222579] usb 1-1: New USB device found, idVendor=32ac, idProduct=0002, bcdDevice= 0.00
[  887.222590] usb 1-1: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[  887.222594] usb 1-1: Product: HDMI Expansion Card
[  887.222597] usb 1-1: Manufacturer: Framework
[  887.222601] usb 1-1: SerialNumber: 11AD1D00B21A3F111B0A0B00
[  887.282581] usb 3-1: New USB device found, idVendor=0bda, idProduct=5634, bcdDevice= 0.21
[  887.282591] usb 3-1: New USB device strings: Mfr=3, Product=1, SerialNumber=2
[  887.282596] usb 3-1: Product: Laptop Camera
[  887.282599] usb 3-1: Manufacturer: Generic
[  887.282602] usb 3-1: SerialNumber: 200901010001
[  887.285749] hid-generic 0003:32AC:0002.01C7: hiddev96,hidraw1: USB HID v1.11 Device [Framework HDMI Expansion Card] on usb-0000:c1:00.3-1/input1
[  887.294352] usb 3-1: Found UVC 1.00 device Laptop Camera (0bda:5634)
[  887.411450] usb 1-4: new full-speed USB device number 12 using xhci_hcd
[  887.583529] usb 1-4: New USB device found, idVendor=27c6, idProduct=609c, bcdDevice= 1.00
[  887.583535] usb 1-4: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[  887.583537] usb 1-4: Product: Goodix USB2.0 MISC
[  887.583538] usb 1-4: Manufacturer: Goodix Technology Co., Ltd.
[  887.583540] usb 1-4: SerialNumber: UIDC5FFF269_XXXX_MOC_B0
[  887.714775] usb 1-5: new high-speed USB device number 13 using xhci_hcd
[  887.858566] usb 1-5: New USB device found, idVendor=0e8d, idProduct=e616, bcdDevice= 1.00
[  887.858576] usb 1-5: New USB device strings: Mfr=5, Product=6, SerialNumber=7
[  887.858581] usb 1-5: Product: Wireless_Device
[  887.858584] usb 1-5: Manufacturer: MediaTek Inc.
[  887.858587] usb 1-5: SerialNumber: 000000000

Mashing keys on my Logitech wireless keyboard while its dongle is active leads to these errors:

Jun 18 10:07:18 mango kernel: logitech-djreceiver 0003:046D:C52B.003E: logi_dj_probe: logi_dj_recv_switch_to_dj_mode returned error:-19
Jun 18 10:07:24 mango kernel: logitech-djreceiver 0003:046D:C52B.0047: can't resubmit intr, 0000:c3:00.3-1.3/input2, status -19

Does anyone have any idea how to fix this?

gromit

Administrator

From: Germany

Registered: 2024-02-10

Posts: 1,525

Website

Re: Framework 13 AMD problem: USB & bluetooth stopped working in Arch

Could you test if the issue is also present on the latest mainline kernel?

sudo pacman -U https://pkgbuild.com/\~gromit/linux-bisection-kernels/linux-mainline-6.10rc4-1-x86_64.pkg.tar.zst

and confirm again that it was still working with 6.8:

sudo pacman -U https://archive.archlinux.org/packages/l/linux/linux-6.8.arch1-1-x86_64.pkg.tar.zst

Also could you please post a full dmesg?

cryptearth

Member

Registered: 2024-02-03

Posts: 2,114

Re: Framework 13 AMD problem: USB & bluetooth stopped working in Arch

when dual-booting with windows make sure to disable fast boot in windows - otherwise it can cause weired issues by hardware not properly reset

jwintersinger

Member

Registered: 2024-06-18

Posts: 4

Re: Framework 13 AMD problem: USB & bluetooth stopped working in Arch

Hi, Gromit and cryptearth. Thank you for your help! No luck so far, alas. This is an incredibly bizarre problem that somehow seems to be Arch-specific, and to have some relationship with the MediaTek MT7922 Bluetooth controller and/or USB subsystem in the Framework 13 AMD. I've never come across anything like this before.

Configurations that work:

• Windows 11 works fine -- there are no issues with USB or Bluetooth. (I effectively do not use the Windows installation, though -- the only time I've used it in six months is booting briefly into it a few weeks ago to do a BIOS upgrade.) My wireless Logitech keyboard and mouse both work fine.

• The Ubuntu 24.04 live installation launched from a USB flash drive with the Ubuntu installer also works fine. Here is the dmesg output from Ubuntu 24.04 with kernel 6.8. There are no errors produced with respect to Bluetooth or USB. My wireless Logitech keyboard and mouse both work fine.

The following configurations don't work. They produce voluminous errors with respect to USB and bluetooth, and the wireless keyboard and mouse don't work (except for temporary instances described below). Other USB devices (integrated webcam, USB-A flash drive) don't work at all or suffer transient failures.

• Arch with kernel 6.9.5.arch1-1 and linux-firmware-20240610.9c10a208-1 (the latest stable versions)

• Arch with kernel 6.10.0-rc4-1-mainline and linux-firmware-20240610.9c10a208-1 (with kernel version as per your suggestion). Here is the dmesg output from this configuration. Since the error messages that continuously repeat end up pushing messages from the initial boot out of the ring buffer, here is the output of "journalctl -kb0", which grew to ~15k lines in only 15 minutes of system uptime. Note that both of these files reflect output when the system is working on battery, without any devices plugged into it. For the Framework laptop modular ports, I'm using three USB-C modules and one HDMI module, but none have anything plugged in.

• Arch with kernel 6.10.0-rc4-1-mainline and linux-firmware 20240220.97b693d2-1 (with the firmware version representing my attempt to downgrade to a version that included a previous revision of the Mediatek firmware)

• Arch with kernel 6.8.9.arch1-2 and linux-firmware-20240409.1addd7dc-2 (which I downgraded to as an approximation of the configuration that was working for me before the kernel 6.9 series)

• Arch with kernel linux-lts 6.6.34-1 and linux-firmware-20240409.1addd7dc-2

• Arch environment launched from Arch installer on USB flash drive, either for versions 2024.06.01 (most recent, with kernel 6.9) or 2023.11.01 (which I believe is the installer I used when I first set up my laptop, with kernel 6.5)

In my typical use, I boot to a TTY with the "<hostname> login:" prompt, then login and manually start Wayland via "sway". For these nonworking configurations, after rebooting, before launching Wayland, while sitting at the TTY, I will often have messages related to Bluetooth failure printing to the TTY:

Jun 19 07:36:49 mango kernel: Bluetooth: hci0: sending frame failed (-19)
Jun 19 07:36:49 mango kernel: Bluetooth: hci0: Failed to send wmt patch dwnld (-19)
Jun 19 07:36:49 mango kernel: Bluetooth: hci0: Failed to set up firmware (-19)

In other cases, I'll have errors related to the USB subsystem:

Jun 19 08:07:13 mango kernel: usb 1-5: new high-speed USB device number 86 using xhci_hcd
Jun 19 08:07:13 mango kernel: usb 1-5: device not accepting address 85, error -62
Jun 19 08:07:02 mango kernel: usb 1-5: new high-speed USB device number 85 using xhci_hcd
Jun 19 08:07:02 mango kernel: usb 1-5: device descriptor read/64, error -110
Jun 19 08:06:46 mango kernel: usb 1-5: device descriptor read/64, error -110

Sometimes I'll get both. Oftentimes they'll be printing continuously, with a new message appearing every few seconds. Based on other sessions with other errors (and a still nonfunctioning configuration), I believe the "usb 1-5" device is the MediaTek Bluetooth/wifi controller:

Jun 19 07:37:04 mango kernel: usb 1-5: new high-speed USB device number 107 using xhci_hcd
Jun 19 07:36:52 mango kernel: usb 1-5: USB disconnect, device number 105
Jun 19 07:36:51 mango kernel: usb 1-5: SerialNumber: 000000000
Jun 19 07:36:51 mango kernel: usb 1-5: Manufacturer: MediaTek Inc.
Jun 19 07:36:51 mango kernel: usb 1-5: Product: Wireless_Device

On two occasions, in a nonworking Arch config, despite the errors printed to the TTY, the Logitech wireless keyboard *did* function via its USB dongle. However, as soon as I launched "sway", it stopped working. (I'm not sure if the timing is coincidental, but since this happened twice, I assume it wasn't.)

Other points of interest:

• I also tried running "systemctl disable power-profiles-daemon" and rebooting -- I was worried this might be the issue, since I recently removed TLP and switched to power-profiles-daemon. However, this didn't help.

• I find this issue incredibly puzzling. I've been using this laptop for six months without issue. Everything was working fine until June 8th, when I unplugged the laptop from power and left it on my desk for a week without touching it. When I returned on June 18th and turned on the system, this problem appeared; nothing changed in the software configuration in this interval, and the configuration with kernel 6.9.1 was working fine when I left on June 8th.

• If the Ubuntu install didn't work, I would attribute this issue to a hardware failure. Since Ubuntu works, and Arch previously worked, I thought the issue must be some corrupted state saved somewhere on the system, but the fact that the Arch installation environment is now also showing this failure (in two separate versions separated by six months) suggests this isn't the case.

• I would otherwise attribute this to a weird corrupted hardware state. But I can boot directly from a working Windows or Ubuntu configuration into a nonworking Arch configuration. Likewise, I can trigger the "battery disconnect" function in the BIOS, power off the system for a couple hours (so no state should linger), and then do a cold boot directly into a nonworking Arch installation.

• I tried downgrading to BIOS 3.03, which was the BIOS version I used since setting up the system. This didn't work. Upgrading to 3.05 again also didn't help.

• I also tried disabling fast boot in Windows 11, as per cryptearth's suggestion, rebooting into WIndows again, and then rebooting into Arch. This didn't help.

• I tried removing the four expansion card modules from the laptop (three USB-C, one HDMI). This didn't help.

• Interestingly, on some Ubuntu boots, if I examine "journalctl -kb0", I'll see some of the same error messages as listed in my journalctl output for Arch with respect to "usb 1-4: device descriptor read/64, error -71". However, other Ubuntu boots don't have this error, and usb 1-4 appears fine (as in my Ubuntu dmesg output above). In both cases, the wireless keyboard and mouse work fine.

I have no idea what to try next, but will very much appreciate any ideas. Thank you!

Last edited by jwintersinger (2024-06-19 15:28:42)

jwintersinger

Member

Registered: 2024-06-18

Posts: 4

Re: Framework 13 AMD problem: USB & bluetooth stopped working in Arch

I have new results to report that narrow the scope of the issue. These contradict some of the things I described previously. In the “broken” config below, all USB devices continuously reset every 2.5 seconds or so, and so no USB device is usable.

• Working: Arch installer 2024.06.01 with kernel 6.9.3 and linux-firmware 20240510 booted from USB flash drive. Here is the dmesg output and pacman -Q output.

• Broken: existing Arch installation. I installed exactly the same kernel, linux-firmware, and amd-ucode versions as used in the Arch installer. Here is the dmesg output and pacman -Q output.

Points of interest:

• In the "working" dmesg, there's a message about "Unpatched return thunk in use. This should not happen!" This error does *not* occur in the "broken" dmesg.

• In the "working" Arch installer, the system has booted successfully by 48 seconds in dmesg. Nothing else happens until I manually connect to my wifi network (around 150 seconds) and plug in my wireless keyboard and mouse dongles (around 275 seconds). Both the wireless keyboard and mouse work. No errors are produced in dmesg by these operations.

• In the "broken" install, the USB subsystem initializes around 0.74 s with the various "USB hub found" messages. This is followed by several "USB disconnect" messages starting around 4.4 seconds, and by 5.78 seconds, the system starts a continuous loop of having USB devices disconnect and then be reinitialized. This behaviour will continue indefinitely as long as the system remains on, completing a USB disconnect/reinitialize cycle every three seconds or so. This behaviour is with no interaction from me as the user, and with no external devices plugged into the laptop. As expected given the dmesg errors, USB devices (wireless mouse & keyboard, flash drives) do not work.

• In the "broken" install, I disabled any unnecessary systemd units (e.g., bluetooth, NetworkManager, power-profile-daemon).

What could differ between the Arch installer environment and my existing Arch install to explain this behaviour?

Last edited by jwintersinger (2024-06-21 15:44:00)

jwintersinger

Member

Registered: 2024-06-18

Posts: 4

Re: Framework 13 AMD problem: USB & bluetooth stopped working in Arch

Problem solved: I reinstalled Arch. I have no idea what unholy nonsense transpired in my other installation to cause USB to fail, especially given that the installation was only six months old and I updated it regularly, and I left /boot and /home untouched when reinstalling.