<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <div class="moz-cite-prefix">On 13 October 2025 at 05:58 pm,
      Manivannan Sadhasivam wrote:<span style="white-space: pre-wrap">
</span></div>
    <blockquote type="cite"
cite="mid:mg2ahzgcwgm6h5mtgs4tsel3yrphrfqgfcjydfxgzgxq5h7kot@jtealdt6vvcz">
      <pre wrap="" class="moz-quote-pre">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
</pre>
    </blockquote>
    Hello Mani,<br>
    <br>
    Thanks for your help.<br>
    <br>
    The kernel doesn't compile with PCI_VENDOR_ID_FSL but it compiles
    with PCI_VENDOR_ID_FREESCALE.<br>
    <br>
    I tried it with the following patch:<br>
    <br>
    diff -rupN a/drivers/pci/quirks.c b/drivers/pci/quirks.c<br>
    --- a/drivers/pci/quirks.c    2025-10-12 22:42:36.000000000 +0200<br>
    +++ b/drivers/pci/quirks.c    2025-10-13 17:59:51.473097708 +0200<br>
    @@ -2525,6 +2525,16 @@ static void quirk_disable_aspm_l0s_l1(st<br>
      */<br>
     DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ASMEDIA, 0x1080,
    quirk_disable_aspm_l0s_l1);<br>
     <br>
    +<br>
    +static void quirk_disable_aspm_all(struct pci_dev *dev)<br>
    +{<br>
    +       pci_info(dev, "Disabling ASPM\n");<br>
    +       pci_disable_link_state(dev, PCIE_LINK_STATE_ALL);<br>
    +}<br>
    +<br>
    +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, PCI_ANY_ID,
    quirk_disable_aspm_all);<br>
    +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_FREESCALE, PCI_ANY_ID,
    quirk_disable_aspm_all);<br>
    +<br>
     /*<br>
      * Some Pericom PCIe-to-PCI bridges in reverse mode need the PCIe
    Retrain<br>
      * Link bit cleared after starting the link retrain process to
    allow this<br>
    <br>
    ---<br>
    <br>
    Unfortunately it doesn't solve the issue with pcieport 0001:00:00.0.<br>
    <br>
    Here is the dmesg of the RC1 of kernel 6.18 with this patch applied:
    <a class="moz-txt-link-freetext" href="https://github.com/user-attachments/files/22896410/dmesg_fsl_ppc.txt">https://github.com/user-attachments/files/22896410/dmesg_fsl_ppc.txt</a><br>
    <br>
    Cyrus Plus block diagram:
    <a class="moz-txt-link-freetext" href="https://github.com/chzigotzky/kernels/issues/17#issuecomment-3400086860">https://github.com/chzigotzky/kernels/issues/17#issuecomment-3400086860</a><br>
    <br>
    Thanks,<br>
    Christian<br>
    <br>
  </body>
</html>