Adding Jian-Hong. There's already another patch under discussion to the same problem: https://patchwork.kernel.org/project/linux-pci/patch/20241115072200.37509-3-jhp@xxxxxxxxxxxxx/ -- i. On Wed, 27 Nov 2024, Ajay Agarwal wrote: > It is possible that the downstream port's L1ss registers were not > saved after the initial configuration performed in the function > aspm_calc_l12_info() during the child bus probe. If the upstream > port config space is saved-restored due to some reason, the > downstream port L1ss registers will be overwritten with stale > configuration due to the logic present in > pci_restore_aspm_l1ss_state(). So, attempt to save the downstream > port L1ss registers when we are at the upstream component. > > Signed-off-by: Ajay Agarwal <ajayagarwal@xxxxxxxxxx> > --- > drivers/pci/pcie/aspm.c | 13 +++++++++++++ > 1 file changed, 13 insertions(+) > > diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c > index cee2365e54b8..769a305fad63 100644 > --- a/drivers/pci/pcie/aspm.c > +++ b/drivers/pci/pcie/aspm.c > @@ -99,6 +99,19 @@ void pci_save_aspm_l1ss_state(struct pci_dev *pdev) > cap = &save_state->cap.data[0]; > pci_read_config_dword(pdev, l1ss + PCI_L1SS_CTL2, cap++); > pci_read_config_dword(pdev, l1ss + PCI_L1SS_CTL1, cap++); > + > + /* > + * It is possible that the downstream port's L1ss registers were not > + * saved after the initial configuration performed in the function > + * aspm_calc_l12_info() during the child bus probe. If the upstream port > + * config space is saved-restored due to some reason, the downstream > + * port L1ss registers will be overwritten with stale configuration due > + * to the logic present in pci_restore_aspm_l1ss_state(). So, attempt to > + * save the downstream port L1ss registers when we are at the upstream > + * component. > + */ > + if (!pcie_downstream_port(pdev)) > + pci_save_aspm_l1ss_state(pdev->bus->self); > } > > void pci_restore_aspm_l1ss_state(struct pci_dev *pdev) >
