Re: [PATCH 2/2] PCI/ASPM: Avoid L0s and L1 on Freescale Root Ports

Manivannan Sadhasivam <[email protected]> Fri, 7 Nov 2025 11:39:50 +0530
Newsgroups gmane.linux.kernel.pci,gmane.linux.ports.ppc64.devel,gmane.linux.debian.ports.powerpc,gmane.linux.kernel
Message-ID <ogkty663ld7fe3qmbxyil47pudidenqeikol5prk7n3qexpteq@h77qi3sg5xo4>
On Thu, Nov 06, 2025 at 12:36:39PM -0600, Bjorn Helgaas wrote:
> From: Bjorn Helgaas <[email protected]>
> 
> Christian reported that f3ac2ff14834 ("PCI/ASPM: Enable all ClockPM and
> ASPM states for devicetree platforms") broke booting on the A-EON X5000.
> 
> Fixes: f3ac2ff14834 ("PCI/ASPM: Enable all ClockPM and ASPM states for devicetree platforms")
> Fixes: df5192d9bb0e ("PCI/ASPM: Enable only L0s and L1 for devicetree platforms"
> )
> Reported-by: Christian Zigotzky <[email protected]>
> Link: https://lore.kernel.org/r/[email protected]
> Signed-off-by: Bjorn Helgaas <[email protected]>
> ---
>  drivers/pci/quirks.c | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
> 
> diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> index 214ed060ca1b..44e780718953 100644
> --- a/drivers/pci/quirks.c
> +++ b/drivers/pci/quirks.c
> @@ -2525,6 +2525,18 @@ static void quirk_disable_aspm_l0s_l1(struct pci_dev *dev)
>   */
>  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ASMEDIA, 0x1080, quirk_disable_aspm_l0s_l1);
>  
> +/*
> + * Remove ASPM L0s and L1 support from cached copy of Link Capabilities so
> + * aspm.c won't try to enable them.
> + */
> +static void quirk_disable_aspm_l0s_l1_cap(struct pci_dev *dev)
> +{
> +	dev->lnkcap &= ~PCI_EXP_LNKCAP_ASPM_L0S;
> +	dev->lnkcap &= ~PCI_EXP_LNKCAP_ASPM_L1;
> +	pci_info(dev, "ASPM: L0s L1 removed from Link Capabilities to work around device defect\n");
> +}
> +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_FREESCALE, 0x0451, quirk_disable_aspm_l0s_l1_cap);

From the commit message of the earlier version [1] you shared:

    Removing advertised features prevents aspm.c from enabling them, even if
    users try to enable them via sysfs or by building the kernel with
    CONFIG_PCIEASPM_POWERSAVE or CONFIG_PCIEASPM_POWER_SUPERSAVE.

Going by this reasoning, shouldn't we be doing this for the other quirks
(quirk_disable_aspm_l0s_l1/quirk_disable_aspm_l0s) as well?

- Mani

[1] https://lore.kernel.org/linux-pci/20251105220925.GA1926619@bhelgaas

-- 
மணிவண்ணன் சதாசிவம்