On Mon, Jul 21, 2025 at 11:02:01AM GMT, Johan Hovold wrote: > On Wed, Jul 16, 2025 at 12:16:42PM +0530, Manivannan Sadhasivam wrote: > > On Wed, Jul 16, 2025 at 10:24:23AM GMT, Krishna Chaitanya Chundru wrote: > > > > How about using enable_device in host bridge, without pci_enable_device > > > call the endpoints can't start the transfers. May be we can use that. > > > > Q: Who is going to call pci_enable_device()? > > A: The PCI client driver > > > > This is same as relying on BUS_NOTIFY_BIND_DRIVER notifier. > > It seems to me that enable_device() may be a better fit if we're only > going to enable ASPM for devices with a driver (or when enabled through > sysfs). > > PCI core will already have placed the device in D0, and this avoids > dealing with notifiers. > I'm planning to drop this series in favor of this patch (with one yet-to-be-submitted patch for pcie-qcom on top): https://lore.kernel.org/linux-pci/20250720190140.2639200-1-david.e.box@xxxxxxxxxxxxxxx/ This patch is more elegant compared to this series and also avoids the issue we are discussing. - Mani -- மணிவண்ணன் சதாசிவம்
