linux.git
2 years agoMerge branch 'pci/ctrl/xilinx'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:42 +0000 (10:36 -0600)]
Merge branch 'pci/ctrl/xilinx'

- Fix whitespace issues (Michal Simek)

* pci/ctrl/xilinx:
  PCI: xilinx-nwl: Fix coding style violations

2 years agoMerge branch 'pci/ctrl/mvebu'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:41 +0000 (10:36 -0600)]
Merge branch 'pci/ctrl/mvebu'

- Switch to the gpiod API so we can make of_get_named_gpio_flags() private
  (Dmitry Torokhov)

* pci/ctrl/mvebu:
  PCI: mvebu: Switch to using gpiod API

2 years agoMerge branch 'pci/ctrl/aardvark'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:40 +0000 (10:36 -0600)]
Merge branch 'pci/ctrl/aardvark'

- Switch to using devm_gpiod_get_optional() so we can stop exporting
  devm_gpiod_get_from_of_node() (Dmitry Torokhov)

* pci/ctrl/aardvark:
  PCI: aardvark: Switch to using devm_gpiod_get_optional()

2 years agoMerge branch 'remotes/lorenzo/pci/misc'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:40 +0000 (10:36 -0600)]
Merge branch 'remotes/lorenzo/pci/misc'

- Register notifier if core_init_notifier is enabled in pci-epf-test
  (Kunihiko Hayashi)

- Fixup Kconfig indentation (Shunsuke Mie)

* remotes/lorenzo/pci/misc:
  PCI: endpoint: Fix Kconfig indent style
  PCI: pci-epf-test: Register notifier if only core_init_notifier is enabled

2 years agoMerge branch 'remotes/lorenzo/pci/vmd'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:39 +0000 (10:36 -0600)]
Merge branch 'remotes/lorenzo/pci/vmd'

- Restore MSI remapping configuration during resume because the
  configuration is cleared out by firmware when suspending (Nirmal Patel)

- Reset the hierarchy below VMD when probing the VMD; we attempted this
  before, but with the wrong device, so it didn't work (Francisco Munoz)

* remotes/lorenzo/pci/vmd:
  PCI: vmd: Fix secondary bus reset for Intel bridges
  PCI: vmd: Disable MSI remapping after suspend

2 years agoMerge branch 'remotes/lorenzo/pci/tegra'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:39 +0000 (10:36 -0600)]
Merge branch 'remotes/lorenzo/pci/tegra'

- Switch from devm_gpiod_get_from_of_node() to devm_fwnode_gpiod_get()
  (Dmitry Torokhov)

* remotes/lorenzo/pci/tegra:
  PCI: tegra: Switch to using devm_fwnode_gpiod_get

2 years agoMerge branch 'remotes/lorenzo/pci/qcom'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:38 +0000 (10:36 -0600)]
Merge branch 'remotes/lorenzo/pci/qcom'

- Add DT and driver support for SC8280XP/SA8540P basic interconnects where
  interconnect bandwidth must be requested before enabling interconnect
  clocks (Johan Hovold)

- Add 'dma-coherent' property (Johan Hovold)

* remotes/lorenzo/pci/qcom:
  dt-bindings: PCI: qcom: Allow 'dma-coherent' property
  PCI: qcom: Add basic interconnect support
  dt-bindings: PCI: qcom: Add SC8280XP/SA8540P interconnects

2 years agoMerge branch 'remotes/lorenzo/pci/mt7621'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:38 +0000 (10:36 -0600)]
Merge branch 'remotes/lorenzo/pci/mt7621'

- Add sentinel to mt7621_pcie_quirks_match[] to prevent oops when parsing
  the table (John Thomson)

* remotes/lorenzo/pci/mt7621:
  PCI: mt7621: Add sentinel to quirks table

2 years agoMerge branch 'remotes/lorenzo/pci/endpoint'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:37 +0000 (10:36 -0600)]
Merge branch 'remotes/lorenzo/pci/endpoint'

- Add a .release() callback for the Endpoint Controller library so an
  Endpoint driver is removable (Yoshihiro Shimoda)

- Fix pci-epf-vntb kernel-doc and whitespace (Frank Li)

- Fix pci-epf-vntb error path usage of pci_epc_mem_free_addr() (Frank Li)

- Remove pci-epf-vntb unused epf_db_phy (Frank Li)

- Fix pci-epf-vntb sparse warnings (Frank Li)

* remotes/lorenzo/pci/endpoint:
  PCI: endpoint: pci-epf-vntb: Fix sparse ntb->reg build warning
  PCI: endpoint: pci-epf-vntb: Fix sparse build warning for epf_db
  PCI: endpoint: pci-epf-vntb: Replace hardcoded 4 with sizeof(u32)
  PCI: endpoint: pci-epf-vntb: Remove unused epf_db_phy struct member
  PCI: endpoint: pci-epf-vntb: Fix call pci_epc_mem_free_addr() in error path
  PCI: endpoint: pci-epf-vntb: Fix struct epf_ntb_ctrl indentation
  PCI: endpoint: pci-epf-vntb: Clean up kernel_doc warning
  PCI: endpoint: Fix WARN() when an endpoint driver is removed

2 years agoMerge branch 'remotes/lorenzo/pci/dwc'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:37 +0000 (10:36 -0600)]
Merge branch 'remotes/lorenzo/pci/dwc'

- Fix n_fts[] array overrun (Vidya Sagar)

- Don't advertise PTM Responder role for Endpoints (Vidya Sagar)

- Fix qcom "reset assert" error message (Manivannan Sadhasivam)

- Downgrade "link didn't come up" message to dev_info (Vidya Sagar)

- Initialize PHY before deasserting core reset so the link comes up on
  boards where the PHY provides the reference clock (this was a regression
  in v6.0) (Sascha Hauer)

- Switch histb to the gpiod API (Dmitry Torokhov)

- Fix imx6sx and imx8mq clock names in DT binding (Serge Semin)

- Fix visconti MSI interrupt in DT binding (Serge Semin)

- Consolidate reset-gpio, cdm, windows info in common DT shared by both
  Root Port and Endpoint bindings (Serge Semin)

- Remove bus node from DT examples (Serge Semin)

- Add common phys, phy-names to DT (Serge Semin)

- Add default max-link-speed of Gen5 to DT (Serge Semin)

- Apply generic schema for generic device  (Serge Semin)

- Add default max-functions of 32 to DT (Serge Semin)

- Add common interrupts, interrupt-names to DT (Serge Semin)

- Add common regs, reg-names to DT (Serge Semin)

- Add common clocks, resets to DT (Serge Semin)

- Add dma-coherent to DT (Serge Semin)

- Apply common schema to Rockchip DT (Serge Semin)

- Add Baikal-T1 DT bindings (Serge Semin)

- Add dma-ranges support in DesignWare core (Serge Semin)

- Add dw_pcie_cap_is() for testing controller capabilities (Serge Semin)

- Add generic resources getter to DesignWare core (Serge Semin)

- Combine iATU detection procedures (Serge Semin)

- Add generic clock and reset names to DesignWare core (Serge Semin)

- Add Baikal-T1 PCIe controller driver (Serge Semin)

* remotes/lorenzo/pci/dwc:
  PCI: dwc: Add Baikal-T1 PCIe controller support
  PCI: dwc: Introduce generic platform clocks and resets
  PCI: dwc: Combine iATU detection procedures
  PCI: dwc: Introduce generic resources getter
  PCI: dwc: Introduce generic controller capabilities interface
  PCI: dwc: Introduce dma-ranges property support for RC-host
  dt-bindings: PCI: dwc: Add Baikal-T1 PCIe Root Port bindings
  dt-bindings: PCI: dwc: Apply common schema to Rockchip DW PCIe nodes
  dt-bindings: PCI: dwc: Add dma-coherent property
  dt-bindings: PCI: dwc: Add clocks/resets common properties
  dt-bindings: PCI: dwc: Add reg/reg-names common properties
  dt-bindings: PCI: dwc: Add interrupts/interrupt-names common properties
  dt-bindings: PCI: dwc: Add max-functions EP property
  dt-bindings: PCI: dwc: Apply generic schema for generic device only
  dt-bindings: PCI: dwc: Add max-link-speed common property
  dt-bindings: PCI: dwc: Add phys/phy-names common properties
  dt-bindings: PCI: dwc: Remove bus node from the examples
  dt-bindings: PCI: dwc: Detach common RP/EP DT bindings
  dt-bindings: visconti-pcie: Fix interrupts array max constraints
  dt-bindings: imx6q-pcie: Fix clock names for imx6sx and imx8mq
  PCI: histb: Switch to using gpiod API
  PCI: imx6: Initialize PHY before deasserting core reset
  PCI: dwc: Use dev_info for PCIe link down event logging
  PCI: qcom: Fix error message for reset_control_assert()
  PCI: designware-ep: Disable PTM capabilities for EP mode
  PCI: Add PCI_PTM_CAP_RES macro
  PCI: dwc: Fix n_fts[] array overrun

2 years agoMerge branch 'remotes/lorenzo/pci/brcmstb'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:36 +0000 (10:36 -0600)]
Merge branch 'remotes/lorenzo/pci/brcmstb'

- Enable Multi-MSI (Jim Quinlan)

- Wait for 100ms after PERST# deassert for power and clocks to stabilize
  (Jim Quinlan)

- Use readl_poll_timeout_atomic() instead of hand-rolled timeout loop (Jim
  Quinlan)

- Drop needless "inline" annotations (Jim Quinlan)

- Set RCB_MPS mode bit so data for reads up to MPS are returned in a single
  completion (Jim Quinlan)

* remotes/lorenzo/pci/brcmstb:
  PCI: brcmstb: Set RCB_{MPS,64B}_MODE bits
  PCI: brcmstb: Drop needless 'inline' annotations
  PCI: brcmstb: Replace status loops with read_poll_timeout_atomic()
  PCI: brcmstb: Wait for 100ms following PERST# deassert
  PCI: brcmstb: Enable Multi-MSI

2 years agoMerge branch 'remotes/lorenzo/pci/dt'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:36 +0000 (10:36 -0600)]
Merge branch 'remotes/lorenzo/pci/dt'

- Add ti,j721e-pci-host interrupt controller definition (Matt Ranostay)

- Add ti,j721e-pci-host interrupt properties (Matt Ranostay)

- Add ti,j721s2 host mode device-id (Matt Ranostay)

- Add mediatek-gen3 iommu, power properties (Jianjun Wang)

- Add mediatek-gen3 SoC-based clock names (Frank Wunderlich)

- Add mediatek-gen3 mt7986 support (Frank Wunderlich)

* remotes/lorenzo/pci/dt:
  dt-bindings: PCI: mediatek-gen3: add support for mt7986
  dt-bindings: PCI: mediatek-gen3: add SoC based clock config
  dt-bindings: PCI: Add host mode device-id for j721s2 platform
  dt-bindings: PCI: mediatek-gen3: Support mt8195
  dt-bindings: PCI: ti,j721e-pci-*: Add missing interrupt properties
  dt-bindings: PCI: ti,j721e-pci-host: add interrupt controller definition

2 years agoMerge branch 'pci/sysfs'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:35 +0000 (10:36 -0600)]
Merge branch 'pci/sysfs'

- Fix a double free in the error path of creating sysfs "resource%d"
  attributes (Sascha Hauer)

* pci/sysfs:
  PCI/sysfs: Fix double free in error path

2 years agoMerge branch 'pci/resource'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:34 +0000 (10:36 -0600)]
Merge branch 'pci/resource'

- Remove EfiMemoryMappedIO regions from the E820 map to allow PCI core to
  allocate BARs from them.  The only purpose of EfiMemoryMappedIO is to
  tell the OS to map things needed by EFI runtime services, so it's often
  used for PCI host bridge apertures.  If we can't allocate from those
  apertures, we can't hot-add devices (Bjorn Helgaas)

* pci/resource:
  x86/PCI: Use pr_info() when possible
  x86/PCI: Fix log message typo
  x86/PCI: Tidy E820 removal messages
  PCI: Skip allocate_resource() if too little space available
  efi/x86: Remove EfiMemoryMappedIO from E820 map

2 years agoMerge branch 'pci/portdrv'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:34 +0000 (10:36 -0600)]
Merge branch 'pci/portdrv'

- Squash portdrv_core.c and portdrv_pci.c into portdrv.c to make it easier
  to find things (Bjorn Helgaas)

- Allow AER service only for Root Ports & RCECs so portdrv can successfully
  bind to other devices that have AER but lack MSI (which they don't need
  for AER), which allows power management for those devices (Bjorn Helgaas)

* pci/portdrv:
  PCI/portdrv: Allow AER service only for Root Ports & RCECs
  PCI/portdrv: Unexport pcie_port_service_register(), pcie_port_service_unregister()
  PCI/portdrv: Move private things to portdrv.c
  PCI/portdrv: Squash into portdrv.c

2 years agoMerge branch 'pci/pm-agp'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:33 +0000 (10:36 -0600)]
Merge branch 'pci/pm-agp'

- Convert AGP efficeon, intel, amd-k7, ati, nvidia to generic power
  management (Bjorn Helgaas)

* pci/pm-agp:
  agp/via: Update to DEFINE_SIMPLE_DEV_PM_OPS()
  agp/sis: Update to DEFINE_SIMPLE_DEV_PM_OPS()
  agp/amd64: Update to DEFINE_SIMPLE_DEV_PM_OPS()
  agp/nvidia: Convert to generic power management
  agp/ati: Convert to generic power management
  agp/amd-k7: Convert to generic power management
  agp/intel: Convert to generic power management
  agp/efficeon: Convert to generic power management

2 years agoMerge branch 'pci/pm'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:33 +0000 (10:36 -0600)]
Merge branch 'pci/pm'

- Remove unused 'state' parameter to pci_legacy_suspend_late() (Bjorn
  Helgaas)

* pci/pm:
  PCI/PM: Remove unused 'state' parameter to pci_legacy_suspend_late()

2 years agoMerge branch 'pci/misc'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:32 +0000 (10:36 -0600)]
Merge branch 'pci/misc'

- Use METHOD_NAME__UID instead of plain string to make it easier to find
  all uses (Yipeng Zou)

* pci/misc:
  PCI/ACPI: Use METHOD_NAME__UID instead of plain string

2 years agoMerge branch 'pci/hotplug'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:32 +0000 (10:36 -0600)]
Merge branch 'pci/hotplug'

- Enable pciehp by default if USB4 is enabled because USB4/Thunderbolt
  tunneling depends on native PCIe hotplug (Albert Zhou)

- Make sure pciehp binds only to Downstream Ports, not Upstream Ports
  (Rafael J. Wysocki)

- Remove unused get_mode1_ECC_cap callback in shpchp (Ian Cowan)

- Enable pciehp Command Completed Interrupt only if supported to reduce
  confusion when looking at lspci output (Pali Rohár)

* pci/hotplug:
  PCI: pciehp: Enable Command Completed Interrupt only if supported
  PCI: shpchp: Remove unused get_mode1_ECC_cap callback
  PCI: acpiphp: Avoid setting is_hotplug_bridge for PCIe Upstream Ports
  PCI/portdrv: Set PCIE_PORT_SERVICE_HP for Root and Downstream Ports only
  PCI: pciehp: Enable by default if USB4 enabled

2 years agoMerge branch 'pci/enumeration'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:32 +0000 (10:36 -0600)]
Merge branch 'pci/enumeration'

- Only read/write PCIe Link 2 registers for devices with Links and PCIe
  Capability version >= 2 (Maciej W. Rozycki)

- Revert a patch that cleared PCI_STATUS during enumeration because it
  broke Linux guests on Apple's virtualization framework (Bjorn Helgaas)

- Assign PCI domain IDs using IDAs so IDs can be easily reused after
  loading/unloading host bridge drivers (Pali Rohár)

- Fix pci_device_is_present(), which previously always returned "false" for
  VFs because their vendor ID is always 0xfff (Michael S. Tsirkin)

- Check for alloc failure in pci_request_irq() (Zeng Heng)

* pci/enumeration:
  PCI: Check for alloc failure in pci_request_irq()
  PCI: Fix pci_device_is_present() for VFs by checking PF
  PCI: Assign PCI domain IDs by ida_alloc()
  Revert "PCI: Clear PCI_STATUS when setting up device"
  PCI: Access Link 2 registers only for devices with Links

2 years agoMerge branch 'pci/doe'
Bjorn Helgaas [Sat, 10 Dec 2022 16:36:31 +0000 (10:36 -0600)]
Merge branch 'pci/doe'

- Fix calculation of DOE length to account for the "0 means 2^18 DWORDs"
  special case (Li Ming)

* pci/doe:
  PCI/DOE: Fix maximum data object length miscalculation

2 years agox86/PCI: Use pr_info() when possible
Bjorn Helgaas [Fri, 9 Dec 2022 20:41:27 +0000 (14:41 -0600)]
x86/PCI: Use pr_info() when possible

Use pr_info() and similar when possible.  No functional change intended.

Link: https://lore.kernel.org/r/20221209205131.GA1726524@bhelgaas
Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2 years agox86/PCI: Fix log message typo
Bjorn Helgaas [Thu, 8 Dec 2022 19:03:41 +0000 (13:03 -0600)]
x86/PCI: Fix log message typo

Add missing word in the log message:

  - ... so future kernels can this automatically
  + ... so future kernels can do this automatically

Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com>
Link: https://lore.kernel.org/r/20221208190341.1560157-5-helgaas@kernel.org
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
2 years agox86/PCI: Tidy E820 removal messages
Bjorn Helgaas [Thu, 8 Dec 2022 19:03:40 +0000 (13:03 -0600)]
x86/PCI: Tidy E820 removal messages

These messages:

  clipped [mem size 0x00000000 64bit] to [mem size 0xfffffffffffa0000 64bit] for e820 entry [mem 0x0009f000-0x000fffff]

aren't as useful as they could be because (a) the resource is often
IORESOURCE_UNSET, so we print the size instead of the start/end and (b) we
print the available resource even if it is empty after removing the E820
entry.

Print the available space by hand to avoid the IORESOURCE_UNSET problem and
only if it's non-empty.  No functional change intended.

Link: https://lore.kernel.org/r/20221208190341.1560157-4-helgaas@kernel.org
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
2 years agoPCI: Skip allocate_resource() if too little space available
Bjorn Helgaas [Thu, 8 Dec 2022 19:03:39 +0000 (13:03 -0600)]
PCI: Skip allocate_resource() if too little space available

pci_bus_alloc_from_region() allocates MMIO space by iterating through all
the resources available on the bus.  The available resource might be
reduced if the caller requires 32-bit space or we're avoiding BIOS or E820
areas.

Don't bother calling allocate_resource() if we need more space than is
available in this resource.  This prevents some pointless and annoying
messages about avoided areas.

Link: https://lore.kernel.org/r/20221208190341.1560157-3-helgaas@kernel.org
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
2 years agoefi/x86: Remove EfiMemoryMappedIO from E820 map
Bjorn Helgaas [Thu, 8 Dec 2022 19:03:38 +0000 (13:03 -0600)]
efi/x86: Remove EfiMemoryMappedIO from E820 map

Firmware can use EfiMemoryMappedIO to request that MMIO regions be mapped
by the OS so they can be accessed by EFI runtime services, but should have
no other significance to the OS (UEFI r2.10, sec 7.2).  However, most
bootloaders and EFI stubs convert EfiMemoryMappedIO regions to
E820_TYPE_RESERVED entries, which prevent Linux from allocating space from
them (see remove_e820_regions()).

Some platforms use EfiMemoryMappedIO entries for PCI MMCONFIG space and PCI
host bridge windows, which means Linux can't allocate BAR space for
hot-added devices.

Remove large EfiMemoryMappedIO regions from the E820 map to avoid this
problem.

Leave small (< 256KB) EfiMemoryMappedIO regions alone because on some
platforms, these describe non-window space that's included in host bridge
_CRS.  If we assign that space to PCI devices, they don't work.  On the
Lenovo X1 Carbon, this leads to suspend/resume failures.

The previous solution to the problem of allocating BARs in these regions
was to add pci_crs_quirks[] entries to disable E820 checking for these
machines (see d341838d776a ("x86/PCI: Disable E820 reserved region clipping
via quirks")):

  Acer   DMI_PRODUCT_NAME    Spin SP513-54N
  Clevo  DMI_BOARD_NAME      X170KM-G
  Lenovo DMI_PRODUCT_VERSION *IIL*

Florent reported the BAR allocation issue on the Clevo NL4XLU.  We could
add another quirk for the NL4XLU, but I hope this generic change can solve
it for many machines without having to add quirks.

This change has been tested on Clevo X170KM-G (Konrad) and Lenovo Ideapad
Slim 3 (Matt) and solves the problem even when overriding the existing
quirks by booting with "pci=use_e820".

Link: https://bugzilla.kernel.org/show_bug.cgi?id=216565
Link: https://bugzilla.kernel.org/show_bug.cgi?id=206459#c78
Link: https://bugzilla.redhat.com/show_bug.cgi?id=1868899
Link: https://bugzilla.redhat.com/show_bug.cgi?id=2029207
Link: https://lore.kernel.org/r/20221208190341.1560157-2-helgaas@kernel.org
Reported-by: Florent DELAHAYE <kernelorg@undead.fr>
Tested-by: Konrad J Hambrick <kjhambrick@gmail.com>
Tested-by: Matt Hansen <2lprbe78@duck.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
2 years agoPCI/portdrv: Allow AER service only for Root Ports & RCECs
Bjorn Helgaas [Fri, 9 Dec 2022 17:01:00 +0000 (11:01 -0600)]
PCI/portdrv: Allow AER service only for Root Ports & RCECs

Previously portdrv allowed the AER service for any device with an AER
capability (assuming Linux had control of AER) even though the AER service
driver only attaches to Root Port and RCECs.

Because get_port_device_capability() included AER for non-RP, non-RCEC
devices, we tried to initialize the AER IRQ even though these devices
don't generate AER interrupts.

Intel DG1 and DG2 discrete graphics cards contain a switch leading to a
GPU.  The switch supports AER but not MSI, so initializing an AER IRQ
failed, and portdrv failed to claim the switch port at all.  The GPU itself
could be suspended, but the switch could not be put in a low-power state
because it had no driver.

Don't allow the AER service on non-Root Port, non-Root Complex Event
Collector devices.  This means we won't enable Bus Mastering if the device
doesn't require MSI, the AER service will not appear in sysfs, and the AER
service driver will not bind to the device.

Link: https://lore.kernel.org/r/20221207084105.84947-1-mika.westerberg@linux.intel.com
Link: https://lore.kernel.org/r/20221210002922.1749403-1-helgaas@kernel.org
Based-on-patch-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com>
2 years agoPCI: xilinx-nwl: Fix coding style violations
Michal Simek [Thu, 8 Dec 2022 12:38:50 +0000 (13:38 +0100)]
PCI: xilinx-nwl: Fix coding style violations

Fix code alignments and remove additional newline.

Link: https://lore.kernel.org/r/17c75e7003bb8c43a0f45ae3d7c45cac230ef852.1670503129.git.michal.simek@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2 years agoPCI: mvebu: Switch to using gpiod API
Dmitry Torokhov [Wed, 7 Dec 2022 21:07:10 +0000 (13:07 -0800)]
PCI: mvebu: Switch to using gpiod API

Switch the driver away from legacy gpio/of_gpio API to gpiod API, and
remove use of of_get_named_gpio_flags() which I want to make private to
gpiolib.

Link: https://lore.kernel.org/r/Y5EAft42YiT66mVj@google.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2 years agoPCI: pciehp: Enable Command Completed Interrupt only if supported
Pali Rohár [Tue, 27 Sep 2022 14:19:17 +0000 (16:19 +0200)]
PCI: pciehp: Enable Command Completed Interrupt only if supported

The No Command Completed Support bit in the Slot Capabilities register
indicates whether Command Completed Interrupt Enable is unsupported.

We already check whether No Command Completed Support bit is set in
pcie_wait_cmd(), and do not wait in this case.

Don't enable this Command Completed Interrupt at all if NCCS is set, so
that when users dump configuration space from userspace, the dump does not
confuse them by saying that Command Completed Interrupt is not supported,
but it is enabled.

Link: https://lore.kernel.org/r/20220927141926.8895-2-kabel@kernel.org
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Lukas Wunner <lukas@wunner.de>
2 years agoPCI: aardvark: Switch to using devm_gpiod_get_optional()
Dmitry Torokhov [Mon, 14 Nov 2022 18:42:25 +0000 (10:42 -0800)]
PCI: aardvark: Switch to using devm_gpiod_get_optional()

Switch the driver to the generic version of gpiod API (and away from
OF-specific variant), so that we can stop exporting
devm_gpiod_get_from_of_node().

Link: https://lore.kernel.org/r/Y3KMEZFv6dpxA+Gv@google.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Pali Rohár <pali@kernel.org>
2 years agodt-bindings: PCI: mediatek-gen3: add support for mt7986
Frank Wunderlich [Sun, 27 Nov 2022 11:41:38 +0000 (12:41 +0100)]
dt-bindings: PCI: mediatek-gen3: add support for mt7986

Add compatible string and clock-definition for mt7986. It needs 4 clocks
for PCIe, define them in binding.

Link: https://lore.kernel.org/r/20221127114142.156573-5-linux@fw-web.de
Signed-off-by: Frank Wunderlich <frank-w@public-files.de>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Matthias Brugger <matthias.bgg@gmail.com>
Acked-by: Jianjun Wang <jianjun.wang@mediatek.com>
2 years agodt-bindings: PCI: mediatek-gen3: add SoC based clock config
Frank Wunderlich [Sun, 27 Nov 2022 11:41:37 +0000 (12:41 +0100)]
dt-bindings: PCI: mediatek-gen3: add SoC based clock config

The PCIe driver covers different SOC which needing different clock
configs. Define them based on compatible.

Link: https://lore.kernel.org/r/20221127114142.156573-4-linux@fw-web.de
Signed-off-by: Frank Wunderlich <frank-w@public-files.de>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Matthias Brugger <matthias.bgg@gmail.com>
Acked-by: Jianjun Wang <jianjun.wang@mediatek.com>
2 years agodt-bindings: PCI: qcom: Allow 'dma-coherent' property
Johan Hovold [Mon, 5 Dec 2022 09:45:30 +0000 (10:45 +0100)]
dt-bindings: PCI: qcom: Allow 'dma-coherent' property

Devices on some PCIe buses may be cache coherent and must be marked as
such in the devicetree to avoid data corruption.

This is specifically needed on recent Qualcomm platforms like SC8280XP.

Link: https://lore.kernel.org/r/20221205094530.12883-1-johan+linaro@kernel.org
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>
2 years agoPCI: mt7621: Add sentinel to quirks table
John Thomson [Mon, 5 Dec 2022 20:46:45 +0000 (06:46 +1000)]
PCI: mt7621: Add sentinel to quirks table

Current driver is missing a sentinel in the struct soc_device_attribute
array, which causes an oops when assessed by the
soc_device_match(mt7621_pcie_quirks_match) call.

This was only exposed once the CONFIG_SOC_MT7621 mt7621 soc_dev_attr
was fixed to register the SOC as a device, in:

commit 7c18b64bba3b ("mips: ralink: mt7621: do not use kzalloc too early")

Fix it by adding the required sentinel.

Link: https://lore.kernel.org/lkml/26ebbed1-0fe9-4af9-8466-65f841d0b382@app.fastmail.com
Link: https://lore.kernel.org/r/20221205204645.301301-1-git@johnthomson.fastmail.com.au
Fixes: b483b4e4d3f6 ("staging: mt7621-pci: add quirks for 'E2' revision using 'soc_device_attribute'")
Signed-off-by: John Thomson <git@johnthomson.fastmail.com.au>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
2 years agoPCI: vmd: Fix secondary bus reset for Intel bridges
Francisco Munoz [Tue, 6 Dec 2022 00:16:37 +0000 (17:16 -0700)]
PCI: vmd: Fix secondary bus reset for Intel bridges

The reset was never applied in the current implementation because Intel
Bridges owned by VMD are parentless. Internally, pci_reset_bus() applies
a reset to the parent of the PCI device supplied as argument, but in this
case it failed because there wasn't a parent.

In more detail, this change allows the VMD driver to enumerate NVMe devices
in pass-through configurations when guest reboots are performed. There was
an attempted to fix this, but later we discovered that the code inside
pci_reset_bus() wasn’t triggering secondary bus resets. Therefore, we
updated the parameters passed to it, and now NVMe SSDs attached to VMD
bridges are properly enumerated in VT-d pass-through scenarios.

Link: https://lore.kernel.org/r/20221206001637.4744-1-francisco.munoz.ruiz@linux.intel.com
Fixes: 6aab5622296b ("PCI: vmd: Clean up domain before enumeration")
Signed-off-by: Francisco Munoz <francisco.munoz.ruiz@linux.intel.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Nirmal Patel <nirmal.patel@linux.intel.com>
Reviewed-by: Jonathan Derrick <jonathan.derrick@linux.dev>
2 years agoPCI: endpoint: pci-epf-vntb: Fix sparse ntb->reg build warning
Frank Li [Wed, 2 Nov 2022 14:10:14 +0000 (10:10 -0400)]
PCI: endpoint: pci-epf-vntb: Fix sparse ntb->reg build warning

  pci-epf-vntb.c:1128:33: sparse:     expected void [noderef] __iomem *base
  pci-epf-vntb.c:1128:33: sparse:     got struct epf_ntb_ctrl *reg

Add __iomem type cast in vntb_epf_peer_spad_read() and
vntb_epf_peer_spad_write().

Link: https://lore.kernel.org/r/20221102141014.1025893-8-Frank.Li@nxp.com
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Frank Li <frank.li@nxp.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Manivannan Sadhasivam <mani@kernel.org>
2 years agoPCI: endpoint: pci-epf-vntb: Fix sparse build warning for epf_db
Frank Li [Wed, 2 Nov 2022 14:10:13 +0000 (10:10 -0400)]
PCI: endpoint: pci-epf-vntb: Fix sparse build warning for epf_db

Use epf_db[i] dereference instead of readl() because epf_db is
in memory allocated by dma_alloc_coherent(), not I/O.

Remove useless/duplicated readl() in the process.

Link: https://lore.kernel.org/r/20221102141014.1025893-7-Frank.Li@nxp.com
Signed-off-by: Frank Li <frank.li@nxp.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
2 years agoPCI: endpoint: pci-epf-vntb: Replace hardcoded 4 with sizeof(u32)
Frank Li [Wed, 2 Nov 2022 14:10:12 +0000 (10:10 -0400)]
PCI: endpoint: pci-epf-vntb: Replace hardcoded 4 with sizeof(u32)

NTB spad entry item size is sizeof(u32), replace hardcoded 4 with it.

Link: https://lore.kernel.org/r/20221102141014.1025893-6-Frank.Li@nxp.com
Signed-off-by: Frank Li <frank.li@nxp.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Manivannan Sadhasivam <mani@kernel.org>
2 years agoPCI: endpoint: pci-epf-vntb: Remove unused epf_db_phy struct member
Frank Li [Wed, 2 Nov 2022 14:10:11 +0000 (10:10 -0400)]
PCI: endpoint: pci-epf-vntb: Remove unused epf_db_phy struct member

epf_db_phy member in struct epf_ntb is not used, remove it.

Link: https://lore.kernel.org/r/20221102141014.1025893-5-Frank.Li@nxp.com
Signed-off-by: Frank Li <frank.li@nxp.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Manivannan Sadhasivam <mani@kernel.org>
2 years agoPCI: endpoint: pci-epf-vntb: Fix call pci_epc_mem_free_addr() in error path
Frank Li [Wed, 2 Nov 2022 14:10:10 +0000 (10:10 -0400)]
PCI: endpoint: pci-epf-vntb: Fix call pci_epc_mem_free_addr() in error path

Replace pci_epc_mem_free_addr() with pci_epf_free_space() in the
error handle path to match pci_epf_alloc_space().

Link: https://lore.kernel.org/r/20221102141014.1025893-4-Frank.Li@nxp.com
Fixes: e35f56bb0330 ("PCI: endpoint: Support NTB transfer between RC and EP")
Signed-off-by: Frank Li <frank.li@nxp.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
2 years agoPCI: endpoint: pci-epf-vntb: Fix struct epf_ntb_ctrl indentation
Frank Li [Wed, 2 Nov 2022 14:10:09 +0000 (10:10 -0400)]
PCI: endpoint: pci-epf-vntb: Fix struct epf_ntb_ctrl indentation

Align the indentation of struct epf_ntb_ctrl with other structs in
the driver.

Link: https://lore.kernel.org/r/20221102141014.1025893-3-Frank.Li@nxp.com
Signed-off-by: Frank Li <frank.li@nxp.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
2 years agoPCI: endpoint: pci-epf-vntb: Clean up kernel_doc warning
Frank Li [Wed, 2 Nov 2022 14:10:08 +0000 (10:10 -0400)]
PCI: endpoint: pci-epf-vntb: Clean up kernel_doc warning

Cleanup warning found by scripts/kernel-doc.

Consolidate terms:

- host, host1 to HOST
- vhost, vHost, Vhost, VHOST2 to VHOST

Link: https://lore.kernel.org/r/20221102141014.1025893-2-Frank.Li@nxp.com
Signed-off-by: Frank Li <frank.li@nxp.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Manivannan Sadhasivam <mani@kernel.org>
2 years agoPCI: dwc: Add Baikal-T1 PCIe controller support
Serge Semin [Sun, 13 Nov 2022 19:13:01 +0000 (22:13 +0300)]
PCI: dwc: Add Baikal-T1 PCIe controller support

Baikal-T1 SoC is equipped with DWC PCIe v4.60a host controller. It can be
trained to work up to Gen.3 speed over up to x4 lanes. The host controller
is attached to the DW PCIe 3.0 PCS via the PIPE-4 interface, which in its
turn is connected to the DWC 10G PHY. The whole system is supposed to be
fed up with four clock sources: DBI peripheral clock, AXI application
clocks and external PHY/core reference clock generating the 100MHz signal.
In addition to that the platform provide a way to reset each part of the
controller: sticky/non-sticky bits, host controller core, PIPE interface,
PCS/PHY and Hot/Power reset signal. The driver also provides a way to
handle the GPIO-based PERST# signal.

Note due to the Baikal-T1 MMIO peculiarity we have to implement the DBI
interface accessors which make sure the IO operations are dword-aligned.

Link: https://lore.kernel.org/r/20221113191301.5526-21-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
2 years agoPCI: dwc: Introduce generic platform clocks and resets
Serge Semin [Sun, 13 Nov 2022 19:13:00 +0000 (22:13 +0300)]
PCI: dwc: Introduce generic platform clocks and resets

Currently almost each platform driver uses its own resets and clocks
naming in order to get the corresponding descriptors. It makes the code
harder to maintain and comprehend especially seeing the DWC PCIe core main
resets and clocks signals set hasn't changed much for about at least one
major IP-core release. So in order to organize things around these signals
we suggest to create a generic interface for them in accordance with the
naming introduced in the DWC PCIe IP-core reference manual:

Application clocks:
- "dbi"  - data bus interface clock (on some DWC PCIe platforms it's
           referred as "pclk", "pcie", "sys", "ahb", "cfg", "iface",
           "gio", "reg", "pcie_apb_sys");
- "mstr" - AXI-bus master interface clock (some DWC PCIe glue drivers
           refer to this clock as "port", "bus", "pcie_bus",
           "bus_master/master_bus/axi_m", "pcie_aclk");
- "slv"  - AXI-bus slave interface clock (also called as "port", "bus",
           "pcie_bus", "bus_slave/slave_bus/axi_s", "pcie_aclk",
           "pcie_inbound_axi").

Core clocks:
- "pipe" - core-PCS PIPE interface clock coming from external PHY (it's
           normally named by the platform drivers as just "pipe");
- "core" - primary clock of the controller (none of the platform drivers
           declare such a clock but in accordance with the ref. manual
           the devices may have it separately specified);
- "aux"  - auxiliary PMC domain clock (it is named by some platforms as
           "pcie_aux" and just "aux");
- "ref"  - Generic reference clock (it is a generic clock source, which
           can be used as a signal source for multiple interfaces, some
           platforms call it as "ref", "general", "pcie_phy",
           "pcie_phy_ref").

Application resets:
- "dbi"  - Data-bus interface reset (it's CSR interface clock and is
           normally called as "apb" though technically it's not APB but
           DWC PCIe-specific interface);
- "mstr" - AXI-bus master reset (some platforms call it as "port", "apps",
           "bus", "axi_m");
- "slv"  - ABI-bus slave reset (some platforms call it as "port", "apps",
           "bus", "axi_s").

Core resets:
- "non-sticky" - non-sticky CSR flags reset;
- "sticky"     - sticky CSR flags reset;
- "pipe"       - PIPE-interface (Core-PCS) logic reset (some platforms
                 call it just "pipe");
- "core"       - controller primary reset (resets everything except PMC
                 module, some platforms refer to this signal as "soft",
                 "pci");
- "phy"        - PCS/PHY block reset (strictly speaking it is normally
                 connected to the input of an external block, but the
                 reference manual says it must be available for the PMC
                 working correctly, some existing platforms call it
                 "pciephy", "phy", "link");
- "hot"        - PMC hot reset signal (also called as "sleep");
- "pwr"        - cold reset signal (can be referred as "pwr", "turnoff").

Bus reset:
- "perst" - PCIe standard signal used to reset the PCIe peripheral
            devices.

As you can see each platform uses it's own naming for basically the same
set of the signals. In the framework of this commit we suggest to add a
set of the clocks and reset signals resources, corresponding names and
identifiers for each denoted entity. At current stage the platforms will
be able to use the provided infrastructure to automatically request all
these resources and manipulate with them in the Host/EP init callbacks.
Alas it isn't that easy to create a common cold/hot reset procedure due to
too many platform-specifics in the procedure, like the external flags
exposure and the delays requirement.

Link: https://lore.kernel.org/r/20221113191301.5526-20-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
2 years agoPCI: dwc: Combine iATU detection procedures
Serge Semin [Sun, 13 Nov 2022 19:12:59 +0000 (22:12 +0300)]
PCI: dwc: Combine iATU detection procedures

Since the iATU CSR region is now retrieved in the DW PCIe resources getter
there is no much benefits in the iATU detection procedures splitting up.
Therefore let's join the iATU unroll/viewport detection procedure with the
rest of the iATU parameters detection code. The resultant method will be
as coherent as before, while the redundant functions will be eliminated
thus producing more readable code.

Link: https://lore.kernel.org/r/20221113191301.5526-19-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
2 years agoPCI: dwc: Introduce generic resources getter
Serge Semin [Sun, 13 Nov 2022 19:12:58 +0000 (22:12 +0300)]
PCI: dwc: Introduce generic resources getter

Currently the DW PCIe Root Port and Endpoint CSR spaces are retrieved in
the separate parts of the DW PCIe core driver. It doesn't really make
sense since the both controller types have identical set of the core CSR
regions: DBI, DBI CS2 and iATU/eDMA. Thus we can simplify the DW PCIe Host
and EP initialization methods by moving the platform-specific registers
space getting and mapping into a common method. It gets to be even more
justified seeing the CSRs base address pointers are preserved in the
common DW PCIe descriptor. Note all the OF-based common DW PCIe settings
initialization will be moved to the new method too in order to have a
single function for all the generic platform properties handling in single
place.

A nice side-effect of this change is that the pcie-designware-host.c and
pcie-designware-ep.c drivers are cleaned up from all the direct dw_pcie
storage modification, which makes the DW PCIe core, Root Port and Endpoint
modules more coherent.

Link: https://lore.kernel.org/r/20221113191301.5526-18-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agoPCI: dwc: Introduce generic controller capabilities interface
Serge Semin [Sun, 13 Nov 2022 19:12:57 +0000 (22:12 +0300)]
PCI: dwc: Introduce generic controller capabilities interface

Since in addition to the already available iATU unrolled mapping we are
about to add a few more DW PCIe platform-specific capabilities (CDM-check
and generic clocks/resets resources) let's add a generic interface to set
and get the flags indicating their availability. The new interface shall
improve maintainability of the platform-specific code.

Link: https://lore.kernel.org/r/20221113191301.5526-17-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
2 years agoPCI: dwc: Introduce dma-ranges property support for RC-host
Serge Semin [Sun, 13 Nov 2022 19:12:56 +0000 (22:12 +0300)]
PCI: dwc: Introduce dma-ranges property support for RC-host

In accordance with the generic PCIe Root Port DT-bindings the "dma-ranges"
property has the same format as the "ranges" property. The only difference
is in their semantics. The "dma-ranges" property describes the PCIe-to-CPU
memory mapping in opposite to the CPU-to-PCIe mapping of the "ranges"
property. Even though the DW PCIe controllers are normally equipped with
the internal Address Translation Unit which inbound and outbound tables
can be used to implement both properties semantics, it was surprising for
me to discover that the host-related part of the DW PCIe driver currently
supports the "ranges" property only while the "dma-ranges" windows are
just ignored. Having the "dma-ranges" supported in the driver would be
very handy for the platforms, that don't tolerate the 1:1 CPU-PCIe memory
mapping and require a customized PCIe memory layout. So let's fix that by
introducing the "dma-ranges" property support.

First of all we suggest to rename the dw_pcie_prog_inbound_atu() method to
dw_pcie_prog_ep_inbound_atu() and create a new version of the
dw_pcie_prog_inbound_atu() function. Thus we'll have two methods for the
RC and EP controllers respectively in the same way as it has been
developed for the outbound ATU setup methods.

Secondly aside with the memory window index and type the new
dw_pcie_prog_inbound_atu() function will accept CPU address, PCIe address
and size as its arguments. These parameters define the PCIe and CPU memory
ranges which will be used to setup the respective inbound ATU mapping. The
passed parameters need to be verified against the ATU ranges constraints
in the same way as it is done for the outbound ranges.

Finally the DMA-ranges detected for the PCIe controller need to be
converted to the inbound ATU entries during the host controller
initialization procedure. It will be done in the framework of the
dw_pcie_iatu_setup() method. Note before setting the inbound ranges up we
need to disable all the inbound ATU entries in order to prevent unexpected
PCIe TLPs translations defined by some third party software like
bootloaders.

Link: https://lore.kernel.org/r/20221113191301.5526-16-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
2 years agodt-bindings: PCI: dwc: Add Baikal-T1 PCIe Root Port bindings
Serge Semin [Sun, 13 Nov 2022 19:12:55 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Add Baikal-T1 PCIe Root Port bindings

Baikal-T1 SoC is equipped with DWC PCIe v4.60a Root Port controller, which
link can be trained to work on up to Gen.3 speed over up to x4 lanes. The
controller is supposed to be fed up with four clock sources: DBI
peripheral clock, AXI application Tx/Rx clocks and external PHY/core
reference clock generating the 100MHz signal. In addition to that the
platform provide a way to reset each part of the controller:
sticky/non-sticky bits, host controller core, PIPE interface, PCS/PHY and
Hot/Power reset signal. The Root Port controller is equipped with multiple
IRQ lines like MSI, system AER, PME, HP, Bandwidth change, Link
equalization request and eDMA ones. The registers space is accessed over
the DBI interface. There can be no more than four inbound or outbound iATU
windows configured.

Link: https://lore.kernel.org/r/20221113191301.5526-15-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Apply common schema to Rockchip DW PCIe nodes
Serge Semin [Sun, 13 Nov 2022 19:12:54 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Apply common schema to Rockchip DW PCIe nodes

As the DT-bindings description states the Rockchip PCIe controller is
based on the DW PCIe RP IP-core thus its DT-nodes are supposed to be
compatible with the common DW PCIe controller schema. Let's make sure they
are evaluated against it by referring to the snps,dw-pcie.yaml schema in
the allOf sub-schemas composition.

Link: https://lore.kernel.org/r/20221113191301.5526-14-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Add dma-coherent property
Serge Semin [Sun, 13 Nov 2022 19:12:53 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Add dma-coherent property

DW PCIe EP/RP AXI- and TRGT1-master interfaces are responsible for the
application memory access. They are used by the RP/EP PCIe buses (MWr/MWr
TLPs emitted by the peripheral PCIe devices) and the eDMA block. Since all
of them mainly involve the system memory and basically mean DMA we can
expect the corresponding platforms can be designed in a way to make sure
the transactions are cache-coherent. As such the DW PCIe DT-nodes can have
the 'dma-coherent' property specified. Let's permit it in the DT-bindings
then.

Link: https://lore.kernel.org/r/20221113191301.5526-13-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Add clocks/resets common properties
Serge Semin [Sun, 13 Nov 2022 19:12:52 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Add clocks/resets common properties

DW PCIe RP/EP reference manuals explicit define all the clocks and reset
requirements in [1] and [2]. Seeing the DW PCIe vendor-specific
DT-bindings have already started assigning random names to the same set of
the clocks and resets lines, let's define a generic names sets and add
them to the DW PCIe common DT-schema.

Note since there are DW PCI-based vendor-specific DT-bindings with the
custom names assigned to the same clocks and resets resources we have no
much choice but to add them to the generic DT-schemas in order to have the
schemas being applicable for such devices. These names are marked as
vendor-specific and should be avoided being used in new bindings in favor
of the generic names.

[1] Synopsys DesignWare Cores PCI Express Controller Databook - DWC PCIe
Root Port, Version 5.40a, March 2019, p.55 - 78.
[2] Synopsys DesignWare Cores PCI Express Controller Databook - DWC PCIe
Endpoint, Version 5.40a, March 2019, p.58 - 81.

Link: https://lore.kernel.org/r/20221113191301.5526-12-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Add reg/reg-names common properties
Serge Semin [Sun, 13 Nov 2022 19:12:51 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Add reg/reg-names common properties

Even though there is a more-or-less limited set of the CSR spaces can be
defined for each DW PCIe controller the generic DT-schema currently
doesn't specify much limitations on the reg-space names used for one or
another range. In order to prevent the vendor-specific controller schemas
further deviation from the generic interface let's fix that by introducing
the reg-names definition in the common DW PCIe DT-schemas and preserving
the generic "reg" and "reg-names" properties in there. New DW PCIe device
DT-bindings are encouraged to use the generic set of the CSR spaces
defined in the generic DW PCIe RP/EP DT-bindings, while the already
available vendor-specific DT-bindings can still apple the common
DT-schemas.

Note the number of reg/reg-names items need to be changed in the DW PCIe
EP DT-schema since aside with the "dbi" CSRs space these arrays can have
"dbi2", "addr_space", "atu", etc ranges.

Also note since there are DW PCIe-based vendor-specific DT-bindings with
the custom names assigned to the same CSR resources we have no much choice
but to add them to the generic DT-schemas in order to have the schemas
being applicable for such devices. These names are marked as
vendor-specific and should be avoided being used in new bindings in favor
of the generic names.

Link: https://lore.kernel.org/r/20221113191301.5526-11-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Add interrupts/interrupt-names common properties
Serge Semin [Sun, 13 Nov 2022 19:12:50 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Add interrupts/interrupt-names common properties

Currently the 'interrupts' and 'interrupt-names' properties are defined
being too generic to really describe any actual IRQ interface. Moreover
the DW PCIe End-point devices are left with no IRQ signals. All of that
can be fixed by adding the IRQ-related properties to the common DW PCIe
DT-schemas in accordance with the hardware reference manual. The DW PCIe
common DT-schema will contain the generic properties definitions with just
a number of entries per property, while the DW PCIe RP/EP-specific schemas
will have the particular number of items and the generic resource names
listed.

Note since there are DW PCI-based vendor-specific DT-bindings with the
custom names assigned to the same IRQ resources we have no much choice but
to add them to the generic DT-schemas in order to have the schemas being
applicable for such devices. These names are marked as vendor-specific and
should be avoided being used in new bindings in favor of the generic
names.

Link: https://lore.kernel.org/r/20221113191301.5526-10-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Add max-functions EP property
Serge Semin [Sun, 13 Nov 2022 19:12:49 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Add max-functions EP property

In accordance with [1] the CX_NFUNC IP-core synthesize parameter is
responsible for the number of physical functions to support in the EP
mode. Its upper limit is 32. Let's use it to constrain the number of
PCIe functions the DW PCIe EP DT-nodes can advertise.

[1] Synopsys DesignWare Cores PCI Express Controller Databook - DWC PCIe
Endpoint, Version 5.40a, March 2019, p. 887.

Link: https://lore.kernel.org/r/20221113191301.5526-9-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Apply generic schema for generic device only
Serge Semin [Sun, 13 Nov 2022 19:12:48 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Apply generic schema for generic device only

Having the generic compatible strings constraints with the 'any'+'generic
string' semantic implicitly encourages either to add new DW PCIe-based
DT-bindings with the generic compatible string attached or just forget
about adding new DT-bindings since the corresponding DT-node will be
evaluated anyway. Moreover having that semantic implemented in the
generic DT-schema causes the DT-validation tool to apply the schema twice:
first by implicit compatible-string-based selection and second by means of
the 'allOf: [ $ref ]' statement. Let's fix all of that by dropping the
compatible property constraints and selecting the generic DT-schema only
for the purely generic DW PCIe DT-nodes. The later is required since there
is a driver for such devices. (Though there are no such DT-nodes currently
defined in the kernel DT sources.)

Link: https://lore.kernel.org/r/20221113191301.5526-8-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Add max-link-speed common property
Serge Semin [Sun, 13 Nov 2022 19:12:47 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Add max-link-speed common property

In accordance with [1] DW PCIe controllers support up to Gen5 link speed.
Let's add the max-link-speed property upper bound to 5 then. The DT
bindings of the particular devices are expected to setup more strict
constraint on that parameter.

[1] Synopsys DesignWare Cores PCI Express Controller Databook, Version
5.40a, March 2019, p. 27

Link: https://lore.kernel.org/r/20221113191301.5526-7-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Add phys/phy-names common properties
Serge Semin [Sun, 13 Nov 2022 19:12:46 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Add phys/phy-names common properties

It's normal to have the DW PCIe RP/EP DT-nodes equipped with the explicit
PHY phandle references. There can be up to 16 PHYs attach in accordance
with the maximum number of supported PCIe lanes. Let's extend the common
DW PCIe controller schema with the 'phys' and 'phy-names' properties
definition. There two types PHY names are defined: preferred generic names
'^pcie[0-9]+$' and non-preferred vendor-specific names
'^pcie([0-9]+|-?phy[0-9]*)?$' so to match the names currently supported by
the DW PCIe platform drivers ("pcie": meson; "pciephy": qcom, imx6;
"pcie-phy": uniphier, rockchip, spear13xx; "pcie": intel-gw; "pcie-phy%d":
keystone, dra7xx; "pcie": histb, etc).

Link: https://lore.kernel.org/r/20221113191301.5526-6-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Remove bus node from the examples
Serge Semin [Sun, 13 Nov 2022 19:12:45 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Remove bus node from the examples

It's absolutely redundant seeing by default each node is embedded into its
own example-X node with address and size cells set to 1.

Link: https://lore.kernel.org/r/20221113191301.5526-5-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: dwc: Detach common RP/EP DT bindings
Serge Semin [Sun, 13 Nov 2022 19:12:44 +0000 (22:12 +0300)]
dt-bindings: PCI: dwc: Detach common RP/EP DT bindings

Currently both DW PCIe Root Port and End-point DT bindings are defined as
separate schemas. Carefully looking at them, at the hardware reference
manuals and seeing there is a generic part of the driver used by the both
RP and EP drivers we can greatly simplify the DW PCIe controller bindings
by moving some of the properties into the common DT schema. It concerns
the PERST GPIO control, number of lanes, number of iATU windows and CDM
check properties. They will be defined in the snps,dw-pcie-common.yaml
schema which will be referenced in the DW PCIe Root Port and End-point DT
bindings in order to evaluate the common for both of these controllers
properties. The rest of properties like reg{,-names}, clock{s,-names},
reset{s,-names}, etc will be consolidate there in one of the next commits.

Link: https://lore.kernel.org/r/20221113191301.5526-4-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: visconti-pcie: Fix interrupts array max constraints
Serge Semin [Sun, 13 Nov 2022 19:12:43 +0000 (22:12 +0300)]
dt-bindings: visconti-pcie: Fix interrupts array max constraints

In accordance with the way the device DT-node is actually defined in
arch/arm64/boot/dts/toshiba/tmpv7708.dtsi and the way the device is probed
by the DW PCIe driver there are two IRQs it actually has. It's MSI IRQ the
DT-bindings lack. Let's extend the interrupts property constraints then
and fix the schema example so one would be acceptable by the actual device
DT-bindings.

Link: https://lore.kernel.org/r/20221113191301.5526-3-Sergey.Semin@baikalelectronics.ru
Fixes: 17c1b16340f0 ("dt-bindings: pci: Add DT binding for Toshiba Visconti PCIe controller")
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>
Acked-by: Nobuhiro Iwamatsu <nobuhiro1.iwamatsu@toshiba.co.jp>
2 years agodt-bindings: imx6q-pcie: Fix clock names for imx6sx and imx8mq
Serge Semin [Sun, 13 Nov 2022 19:12:42 +0000 (22:12 +0300)]
dt-bindings: imx6q-pcie: Fix clock names for imx6sx and imx8mq

Originally as it was defined the legacy bindings the pcie_inbound_axi and
pcie_aux clock names were supposed to be used in the fsl,imx6sx-pcie and
fsl,imx8mq-pcie devices respectively. But the bindings conversion has been
incorrectly so now the fourth clock name is defined as "pcie_inbound_axi
for imx6sx-pcie, pcie_aux for imx8mq-pcie", which is completely wrong.
Let's fix that by conditionally apply the clock-names constraints based on
the compatible string content.

Link: https://lore.kernel.org/r/20221113191301.5526-2-Sergey.Semin@baikalelectronics.ru
Fixes: 751ca492f131 ("dt-bindings: PCI: imx6: convert the imx pcie controller to dtschema")
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Acked-by: Alexander Stein <alexander.stein@ew.tq-group.com>
2 years agoPCI: shpchp: Remove unused get_mode1_ECC_cap callback
Ian Cowan [Sat, 12 Nov 2022 14:28:57 +0000 (09:28 -0500)]
PCI: shpchp: Remove unused get_mode1_ECC_cap callback

The ->get_mode1_ECC_cap callback in the shpchp_hpc_ops struct is never
called, so remove it.

[bhelgaas: squash]
Link: https://lore.kernel.org/r/20221112142859.319733-2-ian@linux.cowan.aero
Link: https://lore.kernel.org/r/20221112142859.319733-3-ian@linux.cowan.aero
Link: https://lore.kernel.org/r/20221112142859.319733-4-ian@linux.cowan.aero
Signed-off-by: Ian Cowan <ian@linux.cowan.aero>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2 years agoPCI: acpiphp: Avoid setting is_hotplug_bridge for PCIe Upstream Ports
Rafael J. Wysocki [Mon, 21 Nov 2022 18:16:57 +0000 (19:16 +0100)]
PCI: acpiphp: Avoid setting is_hotplug_bridge for PCIe Upstream Ports

It is reported that on some systems pciehp binds to an Upstream Port and
attempts to operate it which causes devices below the Port to disappear
from the bus.

This happens because acpiphp sets dev->is_hotplug_bridge for that Port
(after receiving a Device Check notification on it from the platform
firmware via ACPI) during the enumeration of PCI devices.

get_port_device_capability() sees that dev->is_hotplug_bridge is set and
adds PCIE_PORT_SERVICE_HP to Port services, which allows pciehp to bind to
the Port in question.

Even though this particular problem can be addressed by making the
portdrv_core checks more robust, it also causes power management to work
differently on the affected systems which generally is not desirable (PCIe
Ports with dev->is_hotplug_bridge set have to pass additional tests to be
allowed to go into the D3hot/cold power states which affects runtime PM of
devices below these Ports).

For this reason, amend check_hotplug_bridge() with a PCIe type check to
prevent it from setting dev->is_hotplug_bridge for Upstream Ports.

Suggested-by: Lukas Wunner <lukas@wunner.de>
Link: https://lore.kernel.org/r/2262230.ElGaqSPkdT@kreacher
Reported-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Tested-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Lukas Wunner <lukas@wunner.de>
2 years agoPCI/portdrv: Set PCIE_PORT_SERVICE_HP for Root and Downstream Ports only
Rafael J. Wysocki [Mon, 21 Nov 2022 18:15:18 +0000 (19:15 +0100)]
PCI/portdrv: Set PCIE_PORT_SERVICE_HP for Root and Downstream Ports only

It is reported that on some systems pciehp binds to an Upstream Port and
attempts to operate it which causes devices below the Port to disappear
from the bus.

This happens because acpiphp sets dev->is_hotplug_bridge for that Port
(after receiving a Device Check notification on it from the platform
firmware via ACPI) during the enumeration of PCI devices.

get_port_device_capability() sees that dev->is_hotplug_bridge is set and
adds PCIE_PORT_SERVICE_HP to Port services (which allows pciehp to bind to
the Port in question) without consulting the PCIe type, which should be
either Root Port or Downstream Port for the hotplug capability to be
present.

Per PCIe r6.0, sec 7.5.3.2, the Slot Implemented bit is only valid for
Downstream Ports (including Root Ports), and PCIe hotplug depends on the
Slot Capabilities / Control / Status registers.

Make get_port_device_capability() more robust by adding a PCIe type check
to it before adding PCIE_PORT_SERVICE_HP to Port services which helps to
avoid the problem.

[bhelgaas: add spec citation]
Suggested-by: Lukas Wunner <lukas@wunner.de>
Link: https://lore.kernel.org/r/4786090.31r3eYUQgx@kreacher
Reported-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Lukas Wunner <lukas@wunner.de>
2 years agoPCI: Check for alloc failure in pci_request_irq()
Zeng Heng [Mon, 21 Nov 2022 02:00:29 +0000 (10:00 +0800)]
PCI: Check for alloc failure in pci_request_irq()

When kvasprintf() fails to allocate memory, it returns a NULL pointer.
Return error from pci_request_irq() so we don't dereference it.

[bhelgaas: commit log]
Fixes: 704e8953d3e9 ("PCI/irq: Add pci_request_irq() and pci_free_irq() helpers")
Link: https://lore.kernel.org/r/20221121020029.3759444-1-zengheng4@huawei.com
Signed-off-by: Zeng Heng <zengheng4@huawei.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2 years agoPCI/DOE: Fix maximum data object length miscalculation
Li Ming [Wed, 16 Nov 2022 01:56:37 +0000 (09:56 +0800)]
PCI/DOE: Fix maximum data object length miscalculation

Per PCIe r6.0, sec 6.30.1, a data object Length of 0x0 indicates 2^18
DWORDs (256K DW or 1MB) being transferred.  Adjust the value of data object
length for this case on both sending side and receiving side.

Don't bother checking whether Length is greater than SZ_1M because all
values of the 18-bit Length field are valid, and it is impossible to
represent anything larger than SZ_1M:

  0x00000    256K DW (1M bytes)
  0x00001       1 DW (4 bytes)
  ...
  0x3ffff  256K-1 DW (1M - 4 bytes)

[bhelgaas: commit log]
Link: https://lore.kernel.org/r/20221116015637.3299664-1-ming4.li@intel.com
Fixes: 9d24322e887b ("PCI/DOE: Add DOE mailbox support functions")
Signed-off-by: Li Ming <ming4.li@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Lukas Wunner <lukas@wunner.de>
Cc: stable@vger.kernel.org # v6.0+
2 years agoPCI: pciehp: Enable by default if USB4 enabled
Albert Zhou [Tue, 15 Nov 2022 11:38:56 +0000 (22:38 +1100)]
PCI: pciehp: Enable by default if USB4 enabled

Thunderbolt/USB4 PCIe tunneling depends on native PCIe hotplug.  Enable
pciehp by default if USB4 is enabled.

[bhelgaas: squash, update subject, commit logs, tidy whitespace]
Link: https://lore.kernel.org/r/20221115113857.35800-2-albert.zhou.50@gmail.com
Link: https://lore.kernel.org/r/20221115113857.35800-3-albert.zhou.50@gmail.com
Signed-off-by: Albert Zhou <albert.zhou.50@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Mika Westerberg <mika.westerberg@linux.intel.com>
2 years agoPCI: histb: Switch to using gpiod API
Dmitry Torokhov [Tue, 6 Sep 2022 20:43:00 +0000 (13:43 -0700)]
PCI: histb: Switch to using gpiod API

This patch switches the driver away from legacy gpio/of_gpio API to
gpiod API, and removes use of of_get_named_gpio_flags() which I want to
make private to gpiolib.

Link: https://lore.kernel.org/r/20220906204301.3736813-1-dmitry.torokhov@gmail.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
2 years agoPCI: Fix pci_device_is_present() for VFs by checking PF
Michael S. Tsirkin [Wed, 26 Oct 2022 06:11:21 +0000 (02:11 -0400)]
PCI: Fix pci_device_is_present() for VFs by checking PF

pci_device_is_present() previously didn't work for VFs because it reads the
Vendor and Device ID, which are 0xffff for VFs, which looks like they
aren't present.  Check the PF instead.

Wei Gong reported that if virtio I/O is in progress when the driver is
unbound or "0" is written to /sys/.../sriov_numvfs, the virtio I/O
operation hangs, which may result in output like this:

  task:bash state:D stack:    0 pid: 1773 ppid:  1241 flags:0x00004002
  Call Trace:
   schedule+0x4f/0xc0
   blk_mq_freeze_queue_wait+0x69/0xa0
   blk_mq_freeze_queue+0x1b/0x20
   blk_cleanup_queue+0x3d/0xd0
   virtblk_remove+0x3c/0xb0 [virtio_blk]
   virtio_dev_remove+0x4b/0x80
   ...
   device_unregister+0x1b/0x60
   unregister_virtio_device+0x18/0x30
   virtio_pci_remove+0x41/0x80
   pci_device_remove+0x3e/0xb0

This happened because pci_device_is_present(VF) returned "false" in
virtio_pci_remove(), so it called virtio_break_device().  The broken vq
meant that vring_interrupt() skipped the vq.callback() that would have
completed the virtio I/O operation via virtblk_done().

[bhelgaas: commit log, simplify to always use pci_physfn(), add stable tag]
Link: https://lore.kernel.org/r/20221026060912.173250-1-mst@redhat.com
Reported-by: Wei Gong <gongwei833x@gmail.com>
Tested-by: Wei Gong <gongwei833x@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: stable@vger.kernel.org
2 years agoPCI: imx6: Initialize PHY before deasserting core reset
Sascha Hauer [Tue, 1 Nov 2022 09:57:14 +0000 (10:57 +0100)]
PCI: imx6: Initialize PHY before deasserting core reset

When the PHY is the reference clock provider then it must be initialized
and powered on before the reset on the client is deasserted, otherwise
the link will never come up. The order was changed in cf236e0c0d59.
Restore the correct order to make the driver work again on boards where
the PHY provides the reference clock. This also changes the order for
boards where the Soc is the PHY reference clock divider, but this
shouldn't do any harm.

Link: https://lore.kernel.org/r/20221101095714.440001-1-s.hauer@pengutronix.de
Fixes: cf236e0c0d59 ("PCI: imx6: Do not hide PHY driver callbacks and refine the error handling")
Tested-by: Richard Zhu <hongxing.zhu@nxp.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
2 years agoPCI: endpoint: Fix Kconfig indent style
Shunsuke Mie [Mon, 15 Aug 2022 02:50:06 +0000 (11:50 +0900)]
PCI: endpoint: Fix Kconfig indent style

Change to follow the Kconfig style guide. This patch fixes to use tab
rather than space to indent, while help text is indented an additional
two spaces.

Link: https://lore.kernel.org/r/20220815025006.48167-1-mie@igel.co.jp
Fixes: e35f56bb0330 ("PCI: endpoint: Support NTB transfer between RC and EP")
Signed-off-by: Shunsuke Mie <mie@igel.co.jp>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
2 years agoPCI: qcom: Add basic interconnect support
Johan Hovold [Wed, 2 Nov 2022 09:07:05 +0000 (10:07 +0100)]
PCI: qcom: Add basic interconnect support

On Qualcomm platforms like SC8280XP and SA8540P, interconnect bandwidth
must be requested before enabling interconnect clocks.

Add basic support for managing an optional "pcie-mem" interconnect path
by setting a low constraint before enabling clocks and updating it after
the link is up.

Note that it is not possible for a controller driver to set anything but
a maximum peak bandwidth as expected average bandwidth will vary with
use case and actual use (and power policy?). This very much remains an
unresolved problem with the interconnect framework.

Also note that no constraint is set for the SC8280XP/SA8540P "cpu-pcie"
path for now as it is not clear what an appropriate constraint would be
(and the system does not crash when left unspecified).

Link: https://lore.kernel.org/r/20221102090705.23634-3-johan+linaro@kernel.org
Fixes: 70574511f3fc ("PCI: qcom: Add support for SC8280XP")
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>
Acked-by: Georgi Djakov <djakov@kernel.org>
2 years agodt-bindings: PCI: qcom: Add SC8280XP/SA8540P interconnects
Johan Hovold [Wed, 2 Nov 2022 09:07:04 +0000 (10:07 +0100)]
dt-bindings: PCI: qcom: Add SC8280XP/SA8540P interconnects

Add the missing SC8280XP/SA8540P "pcie-mem" and "cpu-pcie" interconnect
paths to the bindings.

Link: https://lore.kernel.org/r/20221102090705.23634-2-johan+linaro@kernel.org
Fixes: 76d777ae045e ("dt-bindings: PCI: qcom: Add SC8280XP to binding")
Fixes: 76c4207f4085 ("dt-bindings: PCI: qcom: Add SA8540P to binding")
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Acked-by: Manivannan Sadhasivam <mani@kernel.org>
2 years agoPCI: vmd: Disable MSI remapping after suspend
Nirmal Patel [Wed, 9 Nov 2022 14:26:52 +0000 (07:26 -0700)]
PCI: vmd: Disable MSI remapping after suspend

MSI remapping is disabled by VMD driver for Intel's Icelake and
newer systems in order to improve performance by setting
VMCONFIG_MSI_REMAP. By design VMCONFIG_MSI_REMAP register is cleared
by firmware during boot. The same register gets cleared when system
is put in S3 power state. VMD driver needs to set this register again
in order to avoid interrupt issues with devices behind VMD if MSI
remapping was disabled before.

Link: https://lore.kernel.org/r/20221109142652.450998-1-nirmal.patel@linux.intel.com
Fixes: ee81ee84f873 ("PCI: vmd: Disable MSI-X remapping when possible")
Signed-off-by: Nirmal Patel <nirmal.patel@linux.intel.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Francisco Munoz <francisco.munoz.ruiz@linux.intel.com>
2 years agoPCI: brcmstb: Set RCB_{MPS,64B}_MODE bits
Jim Quinlan [Tue, 11 Oct 2022 18:42:10 +0000 (14:42 -0400)]
PCI: brcmstb: Set RCB_{MPS,64B}_MODE bits

Set RCB_MPS mode bit so that data for PCIe read requests up to the size of
the Maximum Payload Size (MPS) are returned in one completion, and data for
PCIe read requests greater than the MPS are split at the specified Read
Completion Boundary setting.

Set RCB_64B so that the Read Compeletion Boundary is 64B.

Link: https://lore.kernel.org/r/20221011184211.18128-6-jim2101024@gmail.com
Signed-off-by: Jim Quinlan <jim2101024@gmail.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
2 years agoPCI: brcmstb: Drop needless 'inline' annotations
Jim Quinlan [Tue, 11 Oct 2022 18:42:09 +0000 (14:42 -0400)]
PCI: brcmstb: Drop needless 'inline' annotations

A number of inline functions are called rarely and/or are not
time-critical.  Take out the "inline" and let the compiler do its work.

Link: https://lore.kernel.org/r/20221011184211.18128-5-jim2101024@gmail.com
Signed-off-by: Jim Quinlan <jim2101024@gmail.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
2 years agoPCI: brcmstb: Replace status loops with read_poll_timeout_atomic()
Jim Quinlan [Tue, 11 Oct 2022 18:42:08 +0000 (14:42 -0400)]
PCI: brcmstb: Replace status loops with read_poll_timeout_atomic()

It would be nice to replace the PCIe link-up loop as well but
there are too many uses of this that do not poll (and the
read_poll_timeout uses "timeout==0" to loop forever).

Link: https://lore.kernel.org/r/20221011184211.18128-4-jim2101024@gmail.com
Signed-off-by: Jim Quinlan <jim2101024@gmail.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
2 years agoPCI: brcmstb: Wait for 100ms following PERST# deassert
Jim Quinlan [Tue, 11 Oct 2022 18:42:07 +0000 (14:42 -0400)]
PCI: brcmstb: Wait for 100ms following PERST# deassert

Be prudent and give some time for power and clocks to become stable.  As
described in the PCIe CEM specification sections 2.2 and 2.2.1; as well as
PCIe r5.0, 6.6.1.

Link: https://lore.kernel.org/r/20221011184211.18128-3-jim2101024@gmail.com
Signed-off-by: Jim Quinlan <jim2101024@gmail.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
2 years agoPCI: brcmstb: Enable Multi-MSI
Jim Quinlan [Tue, 11 Oct 2022 18:42:06 +0000 (14:42 -0400)]
PCI: brcmstb: Enable Multi-MSI

We always wanted to enable Multi-MSI but didn't have a test device until
recently.  In addition, there are some devices out there that will ask for
multiple MSI but refuse to work if they are only granted one.

Link: https://lore.kernel.org/r/20221011184211.18128-2-jim2101024@gmail.com
Signed-off-by: Jim Quinlan <jim2101024@gmail.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
2 years agodt-bindings: PCI: Add host mode device-id for j721s2 platform
Matt Ranostay [Fri, 28 Oct 2022 09:17:16 +0000 (02:17 -0700)]
dt-bindings: PCI: Add host mode device-id for j721s2 platform

Add unique device-id of 0xb013 for j721s2 platform to oneOf field.

Link: https://lore.kernel.org/r/20221028091716.21414-1-mranostay@ti.com
Signed-off-by: Matt Ranostay <mranostay@ti.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>
2 years agoPCI: dwc: Use dev_info for PCIe link down event logging
Vidya Sagar [Tue, 13 Sep 2022 10:12:37 +0000 (15:42 +0530)]
PCI: dwc: Use dev_info for PCIe link down event logging

Some of the platforms (like Tegra194 and Tegra234) have open slots and
not having an endpoint connected to the slot is not an error.
So, changing the macro from dev_err to dev_info to log the event.

Link: https://lore.kernel.org/r/20220913101237.4337-1-vidyas@nvidia.com
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Jon Hunter <jonathanh@nvidia.com>
Acked-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
2 years agodt-bindings: PCI: mediatek-gen3: Support mt8195
Jianjun Wang [Thu, 3 Nov 2022 02:56:54 +0000 (10:56 +0800)]
dt-bindings: PCI: mediatek-gen3: Support mt8195

In order to support mt8195 pcie node, update the yaml to support new
properties of iommu and power-domain, and update the reset-names
property to allow only one 'mac' name.

Link: https://lore.kernel.org/r/20221103025656.8714-2-tinghan.shen@mediatek.com
Signed-off-by: Jianjun Wang <jianjun.wang@mediatek.com>
Signed-off-by: TingHan Shen <tinghan.shen@mediatek.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Rob Herring <robh@kernel.org>
2 years agoPCI: qcom: Fix error message for reset_control_assert()
Manivannan Sadhasivam [Wed, 9 Nov 2022 09:40:39 +0000 (15:10 +0530)]
PCI: qcom: Fix error message for reset_control_assert()

Fix the error message to mention "assert" instead of "deassert".

Link: https://lore.kernel.org/r/20221109094039.25753-1-manivannan.sadhasivam@linaro.org
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
2 years agoPCI/sysfs: Fix double free in error path
Sascha Hauer [Tue, 8 Nov 2022 23:05:59 +0000 (17:05 -0600)]
PCI/sysfs: Fix double free in error path

When pci_create_attr() fails, pci_remove_resource_files() is called which
will iterate over the res_attr[_wc] arrays and frees every non NULL entry.
To avoid a double free here set the array entry only after it's clear we
successfully initialized it.

Fixes: b562ec8f74e4 ("PCI: Don't leak memory if sysfs_create_bin_file() fails")
Link: https://lore.kernel.org/r/20221007070735.GX986@pengutronix.de/
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: stable@vger.kernel.org
2 years agoPCI: Assign PCI domain IDs by ida_alloc()
Pali Rohár [Thu, 14 Jul 2022 18:41:30 +0000 (20:41 +0200)]
PCI: Assign PCI domain IDs by ida_alloc()

Replace assignment of PCI domain IDs from atomic_inc_return() to
ida_alloc().

Use two IDAs, one for static domain allocations (those which are defined in
device tree) and second for dynamic allocations (all other).

During removal of root bus / host bridge, also release the domain ID.  The
released ID can be reused again, for example when dynamically loading and
unloading native PCI host bridge drivers.

This change also allows to mix static device tree assignment and dynamic by
kernel as all static allocations are reserved in dynamic pool.

[bhelgaas: set "err" if "bus->domain_nr < 0"]
Link: https://lore.kernel.org/r/20220714184130.5436-1-pali@kernel.org
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2 years agoRevert "PCI: Clear PCI_STATUS when setting up device"
Bjorn Helgaas [Mon, 7 Nov 2022 21:31:08 +0000 (15:31 -0600)]
Revert "PCI: Clear PCI_STATUS when setting up device"

This reverts commit 6cd514e58f12b211d638dbf6f791fa18d854f09c.

Christophe Fergeau reported that 6cd514e58f12 ("PCI: Clear PCI_STATUS when
setting up device") causes boot failures when trying to start linux guests
with Apple's virtualization framework (for example using
https://developer.apple.com/documentation/virtualization/running_linux_in_a_virtual_machine?language=objc)

6cd514e58f12 only solved a cosmetic problem, so revert it to fix the boot
failures.

Link: https://bugzilla.redhat.com/show_bug.cgi?id=2137803
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2 years agoPCI/ACPI: Use METHOD_NAME__UID instead of plain string
Yipeng Zou [Fri, 4 Nov 2022 03:24:30 +0000 (11:24 +0800)]
PCI/ACPI: Use METHOD_NAME__UID instead of plain string

Replace the string "_UID" with the METHOD_NAME__UID macro so instances are
easier to find.

Link: https://lore.kernel.org/r/20221104032430.186424-1-zouyipeng@huawei.com
Signed-off-by: Yipeng Zou <zouyipeng@huawei.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2 years agoPCI: Access Link 2 registers only for devices with Links
Maciej W. Rozycki [Sat, 17 Sep 2022 12:03:09 +0000 (13:03 +0100)]
PCI: Access Link 2 registers only for devices with Links

PCIe r2.0, sec 7.8 added Link Capabilities/Status/Control 2 registers to
the PCIe Capability with Capability Version 2.

Previously we assumed these registers were implemented for all PCIe
Capabilities of version 2 or greater, but in fact they are only
implemented for devices with Links.

Update pcie_capability_reg_implemented() to check whether the device has
a Link.

[bhelgaas: commit log, squash export]
Link: https://lore.kernel.org/r/alpine.DEB.2.21.2209100057070.2275@angie.orcam.me.uk
Link: https://lore.kernel.org/r/alpine.DEB.2.21.2209100057300.2275@angie.orcam.me.uk
Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2 years agoPCI: pci-epf-test: Register notifier if only core_init_notifier is enabled
Kunihiko Hayashi [Thu, 25 Aug 2022 09:01:01 +0000 (18:01 +0900)]
PCI: pci-epf-test: Register notifier if only core_init_notifier is enabled

The pci_epf_test_notifier function should be installed also if only
core_init_notifier is enabled. Fix the current logic.

Link: https://lore.kernel.org/r/20220825090101.20474-1-hayashi.kunihiko@socionext.com
Fixes: 5e50ee27d4a5 ("PCI: pci-epf-test: Add support to defer core initialization")
Signed-off-by: Kunihiko Hayashi <hayashi.kunihiko@socionext.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Om Prakash Singh <omp@nvidia.com>
Acked-by: Kishon Vijay Abraham I <kishon@ti.com>
2 years agoPCI: tegra: Switch to using devm_fwnode_gpiod_get
Dmitry Torokhov [Mon, 5 Sep 2022 06:30:53 +0000 (23:30 -0700)]
PCI: tegra: Switch to using devm_fwnode_gpiod_get

[devm_]gpiod_get_from_of_node in drivers usage should be limited
so that gpiolib can be cleaned up; let's switch to the generic device
property API.

It may even help with handling secondary fwnodes when gpiolib is taught
to handle gpios described by swnodes.

Link: https://lore.kernel.org/r/20220903-gpiod_get_from_of_node-remove-v1-1-b29adfb27a6c@gmail.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
[lpieralisi@kernel.org: commit log]
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
2 years agoPCI: endpoint: Fix WARN() when an endpoint driver is removed
Yoshihiro Shimoda [Thu, 23 Jun 2022 00:38:17 +0000 (09:38 +0900)]
PCI: endpoint: Fix WARN() when an endpoint driver is removed

Since there is no release callback defined for the PCI EPC device,
the below warning is thrown by driver core when a PCI endpoint driver is
removed:

  Device 'e65d0000.pcie-ep' does not have a release() function, it is broken and must be fixed. See Documentation/core-api/kobject.rst.
  WARNING: CPU: 0 PID: 139 at drivers/base/core.c:2232 device_release+0x78/0x8c

Hence, add the release callback and also move the kfree(epc) from
pci_epc_destroy() so that the epc memory is freed when all references are
dropped.

Link: https://lore.kernel.org/r/20220623003817.298173-1-yoshihiro.shimoda.uh@renesas.com
Tested-by: Vidya Sagar <vidyas@nvidia.com>
Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
2 years agoPCI: designware-ep: Disable PTM capabilities for EP mode
Vidya Sagar [Mon, 19 Sep 2022 14:33:40 +0000 (20:03 +0530)]
PCI: designware-ep: Disable PTM capabilities for EP mode

Dual mode DesignWare PCIe IP has PTM capability enabled (if supported) even
in the EP mode. The PCIe compliance for the EP mode expects PTM
capabilities (ROOT_CAPABLE, RES_CAPABLE, CLK_GRAN) be disabled.
Hence disable PTM for the EP mode.

Link: https://lore.kernel.org/r/20220919143340.4527-3-vidyas@nvidia.com
Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Jingoo Han <jingoohan1@gmail.com>
2 years agoPCI: Add PCI_PTM_CAP_RES macro
Vidya Sagar [Mon, 19 Sep 2022 14:33:39 +0000 (20:03 +0530)]
PCI: Add PCI_PTM_CAP_RES macro

Add macro defining Responder capable bit in Precision Time Measurement
capability register.

Link: https://lore.kernel.org/r/20220919143340.4527-2-vidyas@nvidia.com
Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Jingoo Han <jingoohan1@gmail.com>
2 years agoPCI: dwc: Fix n_fts[] array overrun
Vidya Sagar [Mon, 26 Sep 2022 11:19:23 +0000 (16:49 +0530)]
PCI: dwc: Fix n_fts[] array overrun

commit aeaa0bfe89654 ("PCI: dwc: Move N_FTS setup to common setup")
incorrectly uses pci->link_gen in deriving the index to the
n_fts[] array also introducing the issue of accessing beyond the
boundaries of array for greater than Gen-2 speeds. This change fixes
that issue.

Link: https://lore.kernel.org/r/20220926111923.22487-1-vidyas@nvidia.com
Fixes: aeaa0bfe8965 ("PCI: dwc: Move N_FTS setup to common setup")
Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Acked-by: Jingoo Han <jingoohan1@gmail.com>
2 years agodt-bindings: PCI: ti,j721e-pci-*: Add missing interrupt properties
Matt Ranostay [Tue, 25 Oct 2022 08:19:09 +0000 (01:19 -0700)]
dt-bindings: PCI: ti,j721e-pci-*: Add missing interrupt properties

Both interrupts, and interrupt names weren't defined in both EP and host
yaml. Also define the only possible interrupt-name as link_state, and
maxItems of interrupts to one.

This patch resolves the following warning:

arch/arm64/boot/dts/ti/k3-j721s2-common-proc-board.dtb: pcie-ep@2910000: Unevaluated properties are not allowed ('interrupt-names', 'interrupts' were unexpected)
        From schema Documentation/devicetree/bindings/pci/ti,j721e-pci-ep.yaml

Link: https://lore.kernel.org/r/20221025081909.404107-3-mranostay@ti.com
Signed-off-by: Matt Ranostay <mranostay@ti.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>
2 years agodt-bindings: PCI: ti,j721e-pci-host: add interrupt controller definition
Matt Ranostay [Tue, 25 Oct 2022 08:19:08 +0000 (01:19 -0700)]
dt-bindings: PCI: ti,j721e-pci-host: add interrupt controller definition

Add missing 'interrupt-controller' property and related subnodes to resolve
the following warning:

arch/arm64/boot/dts/ti/k3-j721s2-common-proc-board.dtb: pcie@2910000: Unevaluated properties are not allowed ('interrupt-controller' was unexpected)
        From schema: Documentation/devicetree/bindings/pci/ti,j721e-pci-host.yaml

Link: https://lore.kernel.org/r/20221025081909.404107-2-mranostay@ti.com
Signed-off-by: Matt Ranostay <mranostay@ti.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>
2 years agoPCI/PM: Remove unused 'state' parameter to pci_legacy_suspend_late()
Bjorn Helgaas [Tue, 25 Oct 2022 19:35:02 +0000 (14:35 -0500)]
PCI/PM: Remove unused 'state' parameter to pci_legacy_suspend_late()

1a1daf097e21 ("PCI/PM: Remove unused pci_driver.suspend_late() hook")
removed the legacy .suspend_late() hook, which was the only user of the
"state" parameter to pci_legacy_suspend_late(), but it neglected to remove
the parameter.

Remove the unused "state" parameter to pci_legacy_suspend_late().

Link: https://lore.kernel.org/r/20221025193502.669091-1-helgaas@kernel.org
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2 years agoagp/via: Update to DEFINE_SIMPLE_DEV_PM_OPS()
Bjorn Helgaas [Tue, 25 Oct 2022 20:38:52 +0000 (15:38 -0500)]
agp/via: Update to DEFINE_SIMPLE_DEV_PM_OPS()

As of 1a3c7bb08826 ("PM: core: Add new *_PM_OPS macros, deprecate old
ones"), SIMPLE_DEV_PM_OPS() is deprecated in favor of
DEFINE_SIMPLE_DEV_PM_OPS(), which has the advantage that the PM callbacks
don't need to be wrapped with #ifdef CONFIG_PM or tagged with
__maybe_unused.

Convert to DEFINE_SIMPLE_DEV_PM_OPS().  No functional change intended.

Link: https://lore.kernel.org/r/20221025203852.681822-9-helgaas@kernel.org
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Dave Airlie <airlied@redhat.com>