Re: [PATCH] USB: hcd-pci: Do not disable an already disabled device on remove

[Alan Stern]

On Sun, Oct 18, 2026 at 12:41:02PM -0400, Cole Leavitt wrote:
> suspend_common() calls pci_disable_device() and resume_common()
> re-enables the device. If the host controller is hot-removed during
> system resume before its ->resume() callback ran (for example pciehp
> handling a Thunderbolt/USB4 dock or host router that did not come back),
> usb_hcd_pci_remove() drops the enable count a second time:
> 
>   xhci_hcd 0000:b2:00.0: disabling already-disabled device
>   WARNING: drivers/pci/pci.c:2198 at pci_disable_device+0xac/0xc0, CPU#9: irq/128-pciehp/225
>   Call Trace:
>    xhci_pci_remove+0x89/0xd0
>    pci_device_remove+0x4a/0xb0
>    device_release_driver_internal+0x19e/0x200
>    pci_stop_bus_device+0x63/0x70
>    pci_stop_and_remove_bus_device+0x16/0x30
>    pciehp_unconfigure_device+0x96/0x170
>    pciehp_disable_slot+0x66/0x100
>    pciehp_handle_presence_or_link_change+0x62/0x190
>    pciehp_ist+0x14f/0x1d0
> 
> (usb_hcd_pci_remove() tail-calls pci_disable_device(), hence it is not
> in the trace.)
> 
> Only disable the device if it is still enabled.
> 
> Fixes: a0d4922da2e4 ("USB: fix up suspend and resume for PCI host controllers")
> Signed-off-by: Cole Leavitt <cole@xxxxxxxxx>
> ---
> 
> Notes:
>     Tested on a Lenovo ThinkPad P16 Gen 3 (Core Ultra 9 275HX, JHL9580
>     Barlow Ridge host, I226 Ethernet), in daily use since 2026-09-17.

Acked-by: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>

>  drivers/usb/core/hcd-pci.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/usb/core/hcd-pci.c b/drivers/usb/core/hcd-pci.c
> index cd223475917e..ae5175fefd91 100644
> --- a/drivers/usb/core/hcd-pci.c
> +++ b/drivers/usb/core/hcd-pci.c
> @@ -348,7 +348,13 @@ void usb_hcd_pci_remove(struct pci_dev *dev)
>  	usb_put_hcd(hcd);
>  	if ((hcd_driver_flags & HCD_MASK) < HCD_USB3)
>  		pci_free_irq_vectors(dev);
> -	pci_disable_device(dev);
> +	/*
> +	 * suspend_common() disables the device. If the controller is
> +	 * hot-removed during system resume before resume_common() had a
> +	 * chance to re-enable it, there is nothing left to disable.
> +	 */
> +	if (pci_is_enabled(dev))
> +		pci_disable_device(dev);
>  }
>  EXPORT_SYMBOL_GPL(usb_hcd_pci_remove);
>  
> 
> base-commit: 32c52ec87d6245ee2d12ed5dd509e40062939b50
> -- 
> 2.54.0
>