On 3.2.2025 13:46, Greg Kroah-Hartman wrote:
On Mon, Feb 03, 2025 at 01:00:26PM +0100, nb@xxxxxxxxxxx wrote:
From: Nicolai Buchwitz <nb@xxxxxxxxxxx>
Before commit 25f51b76f90f1 ("xhci-pci: Make xhci-pci-renesas a proper
modular driver"), the xhci-pci driver handled the Renesas uPD72020x
USB3
PHY and only utilized features of xhci-pci-renesas when no external
firmware EEPROM was attached. This allowed devices with a valid
firmware
stored in EEPROM to function without requiring xhci-pci-renesas.
That commit changed the behavior, making xhci-pci-renesas responsible
for
handling these devices entirely, even when firmware was already
present
in EEPROM. As a result, unnecessary warnings about missing firmware
files
appeared, and more critically, USB functionality broke whens
CONFIG_USB_XHCI_PCI_RENESAS was not enabled—despite previously
workings
without it.
Fix this by ensuring that devices are only handed over to
xhci-pci-renesas
if the config option is enabled. Otherwise, restore the original
behavior
and handle them as standard xhci-pci devices.
Signed-off-by: Nicolai Buchwitz <nb@xxxxxxxxxxx>
Fixes: 25f51b76f90f ("xhci-pci: Make xhci-pci-renesas a proper modular
driver")
---
drivers/usb/host/xhci-pci.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/usb/host/xhci-pci.c b/drivers/usb/host/xhci-pci.c
index 2d1e205c14c60..4ce80d8ac603e 100644
--- a/drivers/usb/host/xhci-pci.c
+++ b/drivers/usb/host/xhci-pci.c
@@ -654,9 +654,11 @@ int xhci_pci_common_probe(struct pci_dev *dev,
const struct pci_device_id *id)
EXPORT_SYMBOL_NS_GPL(xhci_pci_common_probe, "xhci");
static const struct pci_device_id pci_ids_reject[] = {
+#if IS_ENABLED(CONFIG_USB_XHCI_PCI_RENESAS)
/* handled by xhci-pci-renesas */
{ PCI_DEVICE(PCI_VENDOR_ID_RENESAS, 0x0014) },
{ PCI_DEVICE(PCI_VENDOR_ID_RENESAS, 0x0015) },
+#endif
{ /* end: all zeroes */ }
};
Have you seen:
https://lore.kernel.org/r/20250128104529.58a79bfc@foxbook
?
Hi Greg.
Thanks, I must have overlooked Michal's patch when I initially stumbled
over the issue.
Which one is correct?
I guess both, as Michal is implementing the same slightly different.
My approach was to to keep the changes less invasive as possible and
thus make it possible to use pci_ids_reject[] for further exceptions in
the xhci-pci driver. In Michael's patch the list is specifically used
for blacklisting the Renesas devices and cannot easily be expanded for
other controllers. Either approach is fine with me, so lets move the
discussion to the patch which came first.
thanks,
greg k-h
best,
Nicolai
