Fixes runpm breakage mainly on Nvidia GPUs as they are not able to resume.
Works perfectly with this workaround applied.
RFC comment: We are quite sure that there is a higher amount of bridges affected by this, but I was only testing it on my own machine for now.
I've stresstested runpm by doing 5000 runpm cycles with that patch applied and never saw it fail.
I mainly wanted to get a discussion going on if that's a feasable workaround indeed or if we need something better.
I am also sure, that the nouveau driver itself isn't at fault as I am able to reproduce the same issue by poking into some PCI registers on the PCIe bridge to put the GPU into D3cold as it's done in ACPI code.
I've written a little python script to reproduce this issue without the need of loading nouveau: https://raw.githubusercontent.com/karolherbst/pci-stub-runpm/master/nv_runpm...
Signed-off-by: Karol Herbst kherbst@redhat.com Cc: Bjorn Helgaas bhelgaas@google.com Cc: Lyude Paul lyude@redhat.com Cc: linux-pci@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: nouveau@lists.freedesktop.org --- drivers/pci/pci.c | 39 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 39 insertions(+)
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c index 088fcdc8d2b4..9dbd29ced1ac 100644 --- a/drivers/pci/pci.c +++ b/drivers/pci/pci.c @@ -799,6 +799,42 @@ static inline bool platform_pci_bridge_d3(struct pci_dev *dev) return pci_platform_pm ? pci_platform_pm->bridge_d3(dev) : false; }
+/* + * some intel bridges cause serious issues with runpm if the client device + * is put into D1/D2/D3hot before putting the client into D3cold via + * platform means (generally ACPI). + * + * skipping this makes runpm work perfectly fine on such devices. + * + * As far as we know only skylake and kaby lake SoCs are affected. + */ +static unsigned short intel_broken_d3_bridges[] = { + /* kbl */ + 0x1901, +}; + +static inline bool intel_broken_pci_pm(struct pci_bus *bus) +{ + struct pci_dev *bridge; + int i; + + if (!bus || !bus->self) + return false; + + bridge = bus->self; + if (bridge->vendor != PCI_VENDOR_ID_INTEL) + return false; + + for (i = 0; i < ARRAY_SIZE(intel_broken_d3_bridges); i++) { + if (bridge->device == intel_broken_d3_bridges[i]) { + pci_err(bridge, "found broken intel bridge\n"); + return true; + } + } + + return false; +} + /** * pci_raw_set_power_state - Use PCI PM registers to set the power state of * given PCI device @@ -827,6 +863,9 @@ static int pci_raw_set_power_state(struct pci_dev *dev, pci_power_t state) if (state < PCI_D0 || state > PCI_D3hot) return -EINVAL;
+ if (state != PCI_D0 && intel_broken_pci_pm(dev->bus)) + return 0; + /* * Validate current state: * Can enter D0 from any state, but if we can only go deeper