Re: [PATCH] pci: quirks: Disable native PCIe hotplug on MSI Claw A8 root bridge

On Thu, Aug 6, 2026 at 12:19 PM <[email protected]> wrote:
>
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [High] The quirk disables native PCIe hotplug on the shared host bridge rather than specifically for the target root port, which breaks hotplug functionality for dynamically enumerated Thunderbolt/USB4 devices system-wide.
> --
>
> --- Patch [1]: [PATCH] pci: quirks: Disable native PCIe hotplug on MSI Claw A8 root bridge ---
> commit 841400e3b6d906dff0f94738835128127178d2e6
> Author: Derek J. Clark <[email protected]>
>
> pci: quirks: Disable native PCIe hotplug on MSI Claw A8 root bridge
>
> This commit adds a quirk for the MSI Claw A8 to prevent s2idle resume failures
> caused by the Realtek card reader. It disables native PCIe hotplug on the host
> bridge when the specific root port is found.
>
> > diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> > --- a/drivers/pci/quirks.c
> > +++ b/drivers/pci/quirks.c
>
> [ ... ]
>
> > +static void quirk_claw_a8_no_native_hotplug(struct pci_dev *pdev)
> > +{
> > +     struct pci_host_bridge *bridge;
> > +
> > +     if (!dmi_match(DMI_BOARD_NAME, "MS-1T8K"))
> > +             return;
> > +
> > +     if (pdev->bus->number != 0 ||
> > +         PCI_SLOT(pdev->devfn) != 2 ||
> > +         PCI_FUNC(pdev->devfn) != 2)
> > +             return;
> > +
> > +     bridge = pci_find_host_bridge(pdev->bus);
> > +     if (!bridge)
> > +             return;
> > +     dev_info(&pdev->dev, "disabling native PCIe hotplug\n");
> > +     bridge->native_pcie_hotplug = false;
>
> [Severity: High]
> Does setting bridge->native_pcie_hotplug = false here break hotplug
> functionality for dynamically enumerated Thunderbolt/USB4 devices on this
> system?
>
> Because the pci_host_bridge is a shared structure, modifying it globally
> prevents the pciehp driver from claiming any hotplug-capable PCIe port under
> that domain. While this might be the only fixed slot reporting HotPlug+ during
> boot, downstream ports on dynamically hot-plugged PCIe switches (like
> Thunderbolt docks, eGPUs, or NVMe enclosures) rely on the host bridge's native
> hotplug capability to function.
>
> Would it be safer to clear pdev->is_pciehp and pdev->is_hotplug_bridge on the
> specific root port instead of modifying the shared host bridge?
>

Hi robot, and any humans reading this,

This identified something I missed when scoping the patch. The two USB
ports on top of the unit are USB4/TB capable and are affected by the
root bridge quirk. In a previous version of the patch I did try
pdev->is_hotplug_bridge = false to no effect, but was not aware of
pdev->is_pciehp. I will try with both and see if that resolves it at
the lower scope.

Thanks,
Derek

> > +}
> > +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x150b,
> > +                      quirk_claw_a8_no_native_hotplug);
>
> --
> Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1