On Wed, Aug 23, 2023 at 10:44:47AM +0300, Mika Westerberg wrote: > I guess even without the 60s delay you see in the logs that the PCIe > link is down and Linux starts tearing the device stack towards the dock > on resume? Okay this is what happens (from your log): srp 22 18:19:50 fenix kernel: pcieport 0000:05:01.0: Data Link Layer Link Active not set in 1000 msec srp 22 18:19:50 fenix kernel: pcieport 0000:07:00.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: pcieport 0000:08:00.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: pcieport 0000:08:04.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: pcieport 0000:08:01.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: pcieport 0000:08:03.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: pcieport 0000:08:02.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: not ready 1023ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: not ready 1023ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: not ready 2047ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: not ready 2047ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: not ready 4095ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: not ready 4095ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: not ready 8191ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: not ready 8191ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: not ready 16383ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: not ready 16383ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: not ready 32767ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: not ready 32767ms after resume; waiting srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: not ready 65535ms after resume; giving up srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: not ready 65535ms after resume; giving up srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: ACPI: EC: event unblocked srp 22 18:19:50 fenix kernel: i915 0000:00:02.0: [drm] [ENCODER:94:DDI A/PHY A] is disabled/in DSI mode with an ungated DDI clock, gate it srp 22 18:19:50 fenix kernel: i915 0000:00:02.0: [drm] [ENCODER:102:DDI B/PHY B] is disabled/in DSI mode with an ungated DDI clock, gate it srp 22 18:19:50 fenix kernel: i915 0000:00:02.0: [drm] [ENCODER:118:DDI C/PHY C] is disabled/in DSI mode with an ungated DDI clock, gate it srp 22 18:19:50 fenix kernel: pcieport 0000:05:01.0: pciehp: Slot(1): Card not present srp 22 18:19:50 fenix kernel: pcieport 0000:08:04.0: pciehp: pcie_do_write_cmd: no response from device srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: Unable to change power state from D3cold to D0, device inaccessible srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: Controller not ready at resume -19 srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: PCI post-resume error -19! srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: HC died; cleaning up srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: Controller not ready at resume -19 srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: PCI post-resume error -19! srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: HC died; cleaning up srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: PM: dpm_run_callback(): pci_pm_resume+0x0/0xf0 returns -19 srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: PM: dpm_run_callback(): pci_pm_resume+0x0/0xf0 returns -19 srp 22 18:19:50 fenix kernel: xhci_hcd 0000:09:00.0: PM: failed to resume async: error -19 srp 22 18:19:50 fenix kernel: xhci_hcd 0000:0b:00.0: PM: failed to resume async: error -19 srp 22 18:19:50 fenix kernel: xhci_hcd 0000:3c:00.0: xHC error in resume, USBSTS 0x401, Reinit So directly after resume the PCIe link (tunnel) from the Thunderbolt host router PCIe downstream port does not get re-established and this brings down the whole device hierarchy behind that port. The delay just made it take longer but the actual problem is not the delay but why the tunnel is not re-established properly. Next question is that what's the Thunderbolt firmware version? You can check it throughs sysfs: /sys/bus/thunderbolt/devices/0-0/nvm_version. I see the BIOS you have seems to be quite recent (06/12/2023) so that probably should be good enough. @Chris Chiu, have you guys run Linux on that particular Lenovo system?
