On Mon, Oct 14, 2019 at 06:00:09PM -0500, Bjorn Helgaas wrote: > From: Bjorn Helgaas <bhelgaas@xxxxxxxxxx> > > Dexuan, the important thing here is the first patch, which is your [1], > which I modified by doing pci_restore_state() as well as setting to D0: > > pci_set_power_state(pci_dev, PCI_D0); > pci_restore_state(pci_dev); > > I'm proposing some more patches on top. None are relevant to the problem > you're solving; they're just minor doc and other updates in the same area. > > Rafael, if you have a chance to look at these, I'd appreciate it. I tried > to make the doc match the code, but I'm no PM expert. > > [1] https://lore.kernel.org/r/KU1P153MB016637CAEAD346F0AA8E3801BFAD0@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx > > > Dexuan Cui (1): > PCI/PM: Always return devices to D0 when thawing > > Bjorn Helgaas (6): > PCI/PM: Correct pci_pm_thaw_noirq() documentation > PCI/PM: Clear PCIe PME Status even for legacy power management > PCI/PM: Run resume fixups before disabling wakeup events > PCI/PM: Make power management op coding style consistent > PCI/PM: Wrap long lines in documentation > PCI/MSI: Move power state check out of pci_msi_supported() > > Documentation/power/pci.rst | 38 +++++++------- > drivers/pci/msi.c | 6 +-- > drivers/pci/pci-driver.c | 99 ++++++++++++++++++------------------- > 3 files changed, 71 insertions(+), 72 deletions(-) Thanks Dexuan and Rafael for taking a look at these! I applied the first six to pci/pm and the last to pci/msi, all for v5.5.
