Re: [PATCH v2] PCI: Make d3cold_allowed sysfs attribute read only

[Bjorn Helgaas]

On Thu, Oct 05, 2023 at 01:56:27PM -0500, Mario Limonciello wrote:
> On 10/5/2023 13:53, Bjorn Helgaas wrote:
> > On Wed, Oct 04, 2023 at 09:47:31AM -0500, Mario Limonciello wrote:
> > > Before d3cold was stable userspace was allowed to influence the kernel's
> > > decision of whether to enable d3cold for a device by a sysfs file
> > > `d3cold_allowed`.  This potentially allows userspace to break the suspend
> > > for the system.
> > 
> > Is "Before d3cold was stable" referring to a "d3cold" read-only
> > variable, or to Linux functionality of using D3cold, or ...?
> 
> I was referring to the previous thread's comments when I asked about the
> history on it.
> 
> > In what sense does the `d3cold_allowed` sysfs file break suspend?
> 
> SoCs might not be able to get into their deepest sleep state if userspace
> messes with it.
> 
> > > For debugging purposes `pci_port_pm=` can be used to control whether
> > > a PCI port will go into D3cold and runtime PM can be turned off by
> > > sysfs on PCI end points.
> > 
> > I guess this should be "pcie_port_pm=", which affects *all* PCIe
> > ports?
> 
> Yes.
> 
> > Which sysfs file turns off runtime PM for endpoints?
> 
> /sys/bus/pci/devices/*/power/control

To close the loop on this, I think these are questions that should be
answered in the commit log (actually, that's usually the case when I
have questions, because future readers of the git history may have the
same questions, and it's not practical to dig the answers out of the
lore archive).

Bjorn