On Fri, Nov 03, 2023 at 02:07:49PM -0500, Mario Limonciello wrote: > Downstream drivers are getting the wrong values from > pcie_bandwidth_available() which is causing problems for performance > of eGPUs. > > This series overhauls Thunderbolt related device detection and uses > the changes to change the behavior of pcie_bandwidth_available(). > > NOTE: This series is currently based on top of v6.6 + this change that > will be merged for 6.7: > Link: https://patchwork.freedesktop.org/patch/564738/ Thanks, Mario, I'll look at this soon after v6.7-rc1 (probably Nov 12), so the amdgpu patch should be in mainline by then. > v1->v2: > * Rename is_thunderbolt > * Look for _DSD instead of link > * Drop pci_is_thunderbolt_attached() from all drivers > * Adjust links > * Adjust commit messages > * Add quirk for Tiger Lake > > Mario Limonciello (9): > drm/nouveau: Switch from pci_is_thunderbolt_attached() to > dev_is_removable() > drm/radeon: Switch from pci_is_thunderbolt_attached() to > dev_is_removable() > PCI: Drop pci_is_thunderbolt_attached() > PCI: Move the `PCI_CLASS_SERIAL_USB_USB4` definition to common header > PCI: pciehp: Move check for is_thunderbolt into a quirk > PCI: Rename is_thunderbolt to is_tunneled > PCI: ACPI: Detect PCIe root ports that are used for tunneling > PCI: Exclude PCIe ports used for tunneling in > pcie_bandwidth_available() > PCI: Add a quirk to mark 0x8086 : 0x9a23 as supporting PCIe tunneling > > drivers/gpu/drm/nouveau/nouveau_vga.c | 6 +- > drivers/gpu/drm/radeon/radeon_device.c | 4 +- > drivers/gpu/drm/radeon/radeon_kms.c | 2 +- > drivers/pci/hotplug/pciehp_hpc.c | 6 +- > drivers/pci/pci-acpi.c | 16 ++++++ > drivers/pci/pci.c | 76 +++++++++++++++++--------- > drivers/pci/probe.c | 2 +- > drivers/pci/quirks.c | 31 +++++++++++ > drivers/platform/x86/apple-gmux.c | 2 +- > drivers/thunderbolt/nhi.h | 2 - > include/linux/pci.h | 25 +-------- > include/linux/pci_ids.h | 1 + > 12 files changed, 109 insertions(+), 64 deletions(-) > > -- > 2.34.1 >