Re: [PPC] Boot problems after the pci-v6.18-changes

On 13 October 2025 at 05:58 pm, Manivannan Sadhasivam wrote:
 > Either the Root Port could be triggering these AER messages due to 
ASPM issue or
 > due to the endpoint connected downstream.
 >
 > If possible, please share the whole dmesg log instead of the snippet 
so that we
 > can be sure from where the AER messages are coming from.
 >
 > You can also add the below quirk and check:
 >
 > DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_FSL, 0x0451, 
quirk_disable_aspm_all);
 >
 > But it would be better to get the whole dmesg.
 >
 > - Mani

Hello Mani,

Thanks for your help.

The kernel doesn't compile with PCI_VENDOR_ID_FSL but it compiles with 
PCI_VENDOR_ID_FREESCALE.

I tried it with the following patch:

diff -rupN a/drivers/pci/quirks.c b/drivers/pci/quirks.c

--- a/drivers/pci/quirks.c    2025-10-12 22:42:36.000000000 +0200
+++ b/drivers/pci/quirks.c    2025-10-13 17:59:51.473097708 +0200

@@ -2525,6 +2525,16 @@ static void quirk_disable_aspm_l0s_l1(st
   */
  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ASMEDIA, 0x1080, 

quirk_disable_aspm_l0s_l1);

+
+static void quirk_disable_aspm_all(struct pci_dev *dev)
+{
+       pci_info(dev, "Disabling ASPM\n");
+       pci_disable_link_state(dev, PCIE_LINK_STATE_ALL);
+}
+
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, PCI_ANY_ID, 
quirk_disable_aspm_all);
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_FREESCALE, PCI_ANY_ID, 
quirk_disable_aspm_all);
+
  /*
   * Some Pericom PCIe-to-PCI bridges in reverse mode need the PCIe Retrain
   * Link bit cleared after starting the link retrain process to allow this

---

Unfortunately it doesn't solve the issue with pcieport 0001:00:00.0.

Here is the dmesg of the RC1 of kernel 6.18 with this patch applied: 
https://github.com/user-attachments/files/22896410/dmesg_fsl_ppc.txt

Cyrus Plus block diagram: 
https://github.com/chzigotzky/kernels/issues/17#issuecomment-3400086860

Thanks,
Christian