[Bug 221073] xHCI host controller dies on resume from s2idle on AMD Strix Halo [1022:1587]

[bugzilla-daemon@kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=221073

--- Comment #31 from Mathias Nyman (mathias.nyman@linux.intel.com) ---
you lspci showed MSI-X are used for this controller.
c1:00.4 USB controller [0c03]: Advanced Micro Devices, Inc. [AMD] Strix Halo
USB 3.1 xHCI [1022:1587] (prog-if 30 [XHCI])
        ...
        Capabilities: [a0] MSI: Enable- Count=1/1 Maskable- 64bit+
        Capabilities: [c0] MSI-X: Enable+ Count=1 Masked-

Are the other xHCI hosts also using MSI-X, or just this one?

Are there any changes in any of the PCI MSI and MSI-X capabilities fields 
after resume, like address field?
check with lspci -vvv before and after resume.

Does moving the xHC irq to CPU0 before suspend help?
(check actual xhci irq number from /proc/interrupts, assume 25 in example)
echo 1 > /proc/irq/25/smp_affinity

Another thing to try would be to force MSI interrupt instead of MSI-X.
Not sure if there is an easy way to do this, couldn't find a kernel parameter
for it.

One way to do it is to modify xhci driver:

diff --git a/drivers/usb/host/xhci-pci.c b/drivers/usb/host/xhci-pci.c
index 585b2f3117b0..3acb6ad86f4e 100644
--- a/drivers/usb/host/xhci-pci.c
+++ b/drivers/usb/host/xhci-pci.c
@@ -167,7 +167,7 @@ static int xhci_try_enable_msi(struct usb_hcd *hcd)

        /* TODO: Check with MSI Soc for sysdev */
        xhci->nvecs = pci_alloc_irq_vectors(pdev, 1, xhci->nvecs,
-                                           PCI_IRQ_MSIX | PCI_IRQ_MSI);
+                                           PCI_IRQ_MSI);
        if (xhci->nvecs < 0) {
                xhci_dbg_trace(xhci, trace_xhci_dbg_init,
                               "failed to allocate IRQ vectors");

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.