[Bug 2148735] Kernel logs, boot history, lspci, lsusb, and report text

Logs and hardware details collected from affected host. ** Attachment added: "Kernel logs, boot history, lspci, lsusb, and report text" https://bugs.launchpad.net/bugs/2148735/+attachment/5962747/+files/linux-kernel-usb-hard-reset-report.tar.gz -- You received this bug notification because you are subscribed to linux in Ubuntu. Matching subscriptions: Bgg, Bmail, Nb https://bugs.launchpad.net/bugs/2148735 Title: AMD host hard reset during sustained USB/ADB access, data fabric sync flood event Status in linux package in Ubuntu: New Bug description: # Draft Bug Report: AMD Host Hard Reset During USB Access To Lenovo A7600 Tablet ## Summary Linux host hard-resets during sustained USB access to Android tablets over ADB. The reset happened with a Lenovo A7600-H and also with an Amazon Fire HD 10. It occurred with different USB cables, with and without a powered USB hub, and before updating to the current kernel, so it is not known to be specific to one tablet, one cable, one hub, or `6.17.0-20-generic`. ## System - Distribution: Linux Mint 22.3 (Zena), Ubuntu noble base - Current affected kernel: `6.17.0-20-generic` - Current version signature: `Ubuntu 6.17.0-20.20~24.04.1-generic 6.17.13` - Previous affected kernel before update: `6.14.0-37-generic` - Previous version signature from boot logs: `Ubuntu 6.14.0-37.37~24.04.1-generic 6.14.11` - Architecture: x86_64 - Machine hostname: `sven-X600M-STX-R2-0` - CPU/platform: AMD system; kernel reports AMD data fabric reset reason ## Affected Devices - Lenovo A7600-H: - USB device: `17ef:7731 Lenovo Lenovo A7600` - Android build: `A7600H_A442_000_029_150324_ROW` - Android version: 4.4.2 - Amazon Fire HD 10: - Also triggered the same type of host reset during USB tablet access - Used a different USB cable - Also reproduced with a powered USB hub ## What Happened During ADB access, especially long `adb pull` operations from shared storage, the host hard-resets. The session is lost and the next boot reports a crash, not a clean shutdown. The failure happened: - while the tablet was attached directly - while the tablet was attached through a powered USB hub - with more than one USB cable - before and after the kernel was updated from `6.14.0-37-generic` to `6.17.0-20-generic` - with this Lenovo A7600-H tablet - previously with an Amazon Fire HD 10 tablet as well ## Expected Result ADB read operations should either complete or fail gracefully. The host should not hard-reset. ## Actual Result The host resets. After reboot, logs show: ```text x86/amd: Previous system reset reason [0x08000800]: an uncorrected error caused a data fabric sync flood event ``` Other relevant log lines: ```text usb: port power management may be unreliable usb 1-2.3.1: New USB device found, idVendor=17ef, idProduct=7731 usb 1-2.3.1: Product: Lenovo A7600 smscufx: I2C access timed out, resetting I2C hardware smscufx: Timeout waiting for I2C BUSY to clear ``` ## USB Topology At Time Of Investigation The tablet was on a hub path shared with several USB devices: - VIA Labs USB hub - Microchip/SMSC UFX6000 USB display adapter using `smscufx` - ASIX USB Ethernet adapter - C-Media USB audio adapter - USB mass-storage device `lsusb` included: ```text Bus 001 Device 004: ID 17ef:7731 Lenovo Lenovo A7600 Bus 001 Device 006: ID 0424:9d01 Microchip Technology, Inc. (formerly SMSC) UFX6000 Bus 001 Device 007: ID 0d8c:000c C-Media Electronics, Inc. Audio Adapter Bus 001 Device 008: ID 0b95:772b ASIX Electronics Corp. AX88772B Bus 002 Device 004: ID 1234:5678 Brain Actuated Technologies VLI Product String ``` ## Reproduction Steps 1. Boot Linux Mint 22.3 with kernel `6.17.0-20-generic`. 2. Connect Lenovo A7600-H Android tablet with USB debugging enabled. 3. Confirm ADB authorization. 4. Start a long ADB read operation, for example: ```bash adb pull -a /storage/emulated/legacy ./sdcard-backup ``` 5. Wait for sustained transfer. 6. Host hard-resets. ## Notes - This is a host stability issue. The tablet remains bootable after the host reset. - Because it reproduced with two tablets and different cables, it is less likely to be a single-device or single-cable fault. - Because the host resets during read-only operations, it is unsafe to use this host for SP Flash Tool or firmware flashing until resolved. - A useful next diagnostic would be reproducing with a minimal USB setup: direct USB 2.0 port, no dock, no USB display, no USB Ethernet, no USB audio, no USB mass storage, and USB autosuspend disabled. - The bug should not be described as a confirmed regression unless a known-good kernel and first-bad kernel are identified. Current evidence says it affects at least `6.14.0-37-generic` and `6.17.0-20-generic`. To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2148735/+subscriptions