Legacy PCI INTx interrupts are represented in the PCI_INTERRUPT_PIN
register using the range 1-4, which matches our enum pci_interrupt_pin.
This is however not ideal for an IRQ domain, where with 4 interrupts we
would ideally have a domain of size 4 & hwirq numbers in the range 0-3.

Different PCI host controller drivers have handled this in different ways.
Of those under drivers/pci/ which register an INTx IRQ domain, we have:

  - pcie-altera uses the range 1-4 in device trees and an IRQ domain of
    size 5 to cover that range, with entry 0 wasted.

  - pcie-xilinx & pcie-xilinx-nwl use the range 1-4 in device trees but
    register an IRQ domain of size 4, which doesn't cover the hwirq=4/INTD
    case leading to that interrupt being broken.

  - pci-ftpci100 & pci-aardvark use the range 0-3 in both device trees & as
    hwirq numbering in the driver & IRQ domain.

In order to introduce some level of consistency in at least the hwirq
numbering used by the drivers & IRQ domains, this patch introduces a new
pci_irqd_intx_xlate() helper function which drivers using the 1-4 range in
device trees can assign as the xlate callback for their INTx IRQ domain.
This translates the 1-4 range into a 0-3 range, allowing us to use an IRQ
domain of size 4 & avoid a wasted entry. Further patches will make use of
this in drivers to allow them to use an IRQ domain of size 4 for legacy
INTx interrupts without breaking INTD.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

We currently have a definition of enum pci_interrupt_pin in a header
specific to PCI endpoints - linux/pci-epf.h. In order to allow for use of
this enum from PCI host code in a future commit, move its definition to
linux/pci.h & include that from linux/pci-epf.h.

Additionally we add a PCI_NUM_INTX macro which indicates the number of PCI
INTx interrupts, and will be used alongside enum pci_interrupt_pin in
further patches.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
[bhelgaas: move enum pci_interrupt_pin outside #ifdef CONFIG_PCI]
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

Per PCIe r3.1, sec 2.2.6.2 and 7.8.4, a Requester may not use 8-bit Tags
unless its Extended Tag Field Enable is set, but all Receivers/Completers
must handle 8-bit Tags correctly regardless of their Extended Tag Field
Enable.

Some devices do not handle 8-bit Tags as Completers, so add a quirk for
them.  If we find such a device, we disable Extended Tags for the entire
hierarchy to make peer-to-peer DMA possible.

The Broadcom HT2100 seems to have issues with handling 8-bit tags.  Mark it
as broken.

The pci_walk_bus() in the quirk handles devices we've enumerated in the
past, and pci_configure_device() handles devices we enumerate in the
future.

Fixes: 60db3a4d ("PCI: Enable PCIe Extended Tags if supported")
Link: https://bugzilla.redhat.com/show_bug.cgi?id=1467674Reported-and-tested-by: NWim ten Have <wim.ten.have@oracle.com>
Signed-off-by: NSinan Kaya <okaya@codeaurora.org>
[bhelgaas: changelog, tweak messages, rename bit and quirk]
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

The pci_error_handlers->reset_notify() method had a flag to indicate
whether to prepare for or clean up after a reset.  The prepare and done
cases have no shared functionality whatsoever, so split them into separate
methods.

[bhelgaas: changelog, update locking comments]
Link: http://lkml.kernel.org/r/20170601111039.8913-3-hch@lst.deSigned-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

Here we delete the static pdev_fixup_irq() function which is currently what
pci_fixup_irqs() uses to actually assign the IRQs and replace it with the
pci_assign_irq() function which changes the interface and uses the new
function pointers stored in the host bridge structure.

Eventually this will allow pci_fixup_irqs() to be removed entirely and the
new deferred assignment code path will call pci_assign_irq() directly.
However to ensure current users continue to work, a new implementation of
pci_fixup_irqs() is introduced which simply wraps the functionality of
pci_assign_irq().
Signed-off-by: NMatthew Minter <matt@masarand.com>
[lorenzo.pieralisi@arm.com: reworked comments/log]
Signed-off-by: NLorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

In order to defer IRQ assignment arches must be able to register functions
to map and swizzle interrupts.  These registered functions are stored in
the pci_host_bridge struct.
Signed-off-by: NMatthew Minter <matt@masarand.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

The pci_scan_root_bus_bridge() function allows passing a parameterized
struct pci_host_bridge and scanning the resulting PCI bus; since the struct
msi_controller is part of the struct pci_host_bridge and the struct
pci_host_bridge can now be passed to pci_scan_root_bus_bridge() explicitly,
there is no need for a scan interface with a MSI controller parameter.

With all PCI host controller drivers and platform code relying on
pci_scan_root_bus_msi() converted over to pci_scan_root_bus_bridge() the
pci_scan_root_bus_msi() becomes obsolete and can be removed.
Signed-off-by: NLorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

With the introduction of pci_scan_root_bus_bridge() there is no need to
export pci_register_host_bridge() to other kernel subsystems other than the
PCI compilation unit that needs it.

Make pci_register_host_bridge() static to its compilation unit and convert
the existing drivers usage over to pci_scan_root_bus_bridge().
Signed-off-by: NLorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>

The current pci_scan_root_bus() interface is made up of two main code
paths:

  - pci_create_root_bus()
  - pci_scan_child_bus()

pci_create_root_bus() is a wrapper function that allows to create a struct
pci_host_bridge structure, initialize it with the passed parameters and
register it with the kernel.

As the struct pci_host_bridge require additional struct members,
pci_create_root_bus() parameters list has grown in time, making it unwieldy
to add further parameters to it in case the struct pci_host_bridge gains
more members fields to augment its functionality.

Since PCI core code provides functions to allocate struct pci_host_bridge,
instead of forcing the pci_create_root_bus() interface to add new
parameters to cater for new struct pci_host_bridge functionality, it is
more suitable to add an interface in PCI core code to scan a PCI bus
straight from a struct pci_host_bridge created and customized by each
specific PCI host controller driver.

Add a pci_scan_root_bus_bridge() function to allow PCI host controller
drivers to create and initialize struct pci_host_bridge and scan the
resulting bus.
Signed-off-by: NLorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>

Struct pci_host_bridge can be allocated by PCI host bridge drivers which
usually allocate and map memory through devm managed interfaces.

Add a devm version for the pci_alloc_host_bridge() interface to simplify
PCI host controller driver porting and simplify the driver failure paths.
Signed-off-by: NLorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>

Commit a52d1443 ("PCI: Export host bridge registration interface")
exported the pci_alloc_host_bridge() interface so that PCI host controllers
drivers can make use of it.

Introduce pci_alloc_host_bridge() kernel counterpart to free the host
bridge data structures, pci_free_host_bridge(), export it and update kernel
functions releasing host bridge objects allocated memory to make use of it.
Signed-off-by: NLorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>

After previous changes it is not necessary to distinguish between
device wakeup for run time and device wakeup from system sleep states
any more, so rework the PCI device wakeup settings code accordingly.
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: NBjorn Helgaas <bhelgaas@google.com>

The pme_interrupt flag in struct pci_dev is set when PMEs generated
by the device are going to be signaled via root port PME interrupts.

Ironically enough, that information is only used by the code setting
up device wakeup through ACPI which returns as soon as it sees the
pme_interrupt flag set while setting up "remote runtime wakeup".
That is questionable, however, because in theory there may be PCIe
devices using out-of-band PME signaling under root ports handled
by the native PME code or devices requiring wakeup power setup to be
carried out by AML.  For such devices, ACPI wakeup should be invoked
regardless of whether or not native PME signaling is used in general.

For this reason, drop the pme_interrupt flag and rework the code
using it which then allows the ACPI-based device wakeup handling
in PCI to be consolidated to use one code path for both "runtime
remote wakeup" and system wakeup (from sleep states).
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: NBjorn Helgaas <bhelgaas@google.com>

The test for INTx masking via PCI_COMMAND_INTX_DISABLE performed in
pci_intx_mask_supported() should be done before the device can be used.
This is to avoid writing PCI_COMMAND while the driver owns the device, in
case that has any effect on MSI/MSI-X interrupts.

Move the content of pci_intx_mask_supported() to pci_intx_mask_broken() and
call it from pci_setup_device().

The test result can be queried at any time later using the same
pci_intx_mask_supported() interface as before (though with changed
implementation), so callers (uio, vfio) should be unaffected.
Signed-off-by: NPiotr Gregor <piotrgregor@rsyncme.org>
[bhelgaas: changelog, remove quirk check, remove locking, move
dev->broken_intx_masking assignment to caller]
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Reviewed-by: NAlex Williamson <alex.williamson@redhat.com>
Acked-by: NMichael S. Tsirkin <mst@redhat.com>

After a Function-Level Reset, PCI states need to be restored.  Save PASID
features and PRI reqs cached.

[bhelgaas: search for capability only if PRI/PASID were enabled]
Signed-off-by: NCQ Tang <cq.tang@intel.com>
Signed-off-by: NAshok Raj <ashok.raj@intel.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: Jean-Phillipe Brucker <jean-philippe.brucker@arm.com>
Cc: David Woodhouse <dwmw2@infradead.org>

Device drivers need to check if an IOMMU enabled ATS, PRI and PASID in
order to know when they can use the SVM API.  Cache PRI and PASID bits in
the pci_dev structure, similarly to what is currently done for ATS.
Signed-off-by: NJean-Philippe Brucker <jean-philippe.brucker@arm.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

We need to return an error for any call that asks for MSI / MSI-X
vectors only, so that non-trivial fallback logic can work properly.

Also valid dev->irq and use the "correct" errno value based on feedback
from Linus.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reported-by: NSteven Rostedt <rostedt@goodmis.org>
Fixes: aff17164 ("PCI: Provide sensible IRQ vector alloc/free routines")
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

pci_call_probe() can called recursively when a physcial function is probed
and the probing creates virtual functions, which are populated via
pci_bus_add_device() which in turn can end up calling pci_call_probe()
again.

The code has an interesting way to prevent recursing into the workqueue
code.  That's accomplished by a check whether the current task runs already
on the numa node which is associated with the device.

While that works to prevent the recursion into the workqueue code, it's
racy versus normal execution as there is no guarantee that the node does
not vanish after the check.

There is another issue with this code. It dereferences cpumask_of_node()
unconditionally without checking whether the node is available.

Make the detection reliable by:

 - Mark a probed device as 'is_probed' in pci_call_probe()
 
 - Check in pci_call_probe for a virtual function. If it's a virtual
   function and the associated physical function device is marked
   'is_probed' then this is a recursive call, so the call can be invoked in
   the calling context.

 - Add a check whether the node is online before dereferencing it.
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Acked-by: NIngo Molnar <mingo@kernel.org>
Acked-by: NBjorn Helgaas <bhelgaas@google.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-pci@vger.kernel.org
Cc: Sebastian Siewior <bigeasy@linutronix.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/20170524081548.771457199@linutronix.de

Some drivers - like i915 - may not support the system suspend direct
complete optimization due to differences in their runtime and system
suspend sequence.  Add a flag that when set resumes the device before
calling the driver's system suspend handlers which effectively disables
the optimization.

Needed by a future patch fixing suspend/resume on i915.

Suggested by Rafael.
Signed-off-by: NImre Deak <imre.deak@intel.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Acked-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: stable@vger.kernel.org

The introduction of the pci_remap_cfgspace() interface allows PCI host
controller drivers to map PCI config space through a dedicated kernel
interface. Current PCI host controller drivers use the devm_ioremap_*()
devres interfaces to map PCI configuration space regions so in order to
update them to the new pci_remap_cfgspace() mapping interface a new set of
devres interfaces should be implemented so that PCI host controller drivers
can make use of them.

Introduce two new functions in the PCI kernel layer and Devres
documentation:

- devm_pci_remap_cfgspace()
- devm_pci_remap_cfg_resource()

so that PCI host controller drivers can make use of them to map PCI
configuration space regions.
Signed-off-by: NLorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Cc: Jonathan Corbet <corbet@lwn.net>

Currently we opencode the FLR sequence in lots of place; export a core
helper instead.  We split out the probing for FLR support as all the
non-core callers already know their hardware.

Note that in the new pci_has_flr() function the quirk check has been moved
before the capability check as there is no point in reading the capability
in this case.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

This will need to call into an arch-provided pci_iobar_pfn() function.
Signed-off-by: NDavid Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

Starting to leave behind the legacy of the pci_mmap_page_range() interface
which takes "user-visible" BAR addresses.  This takes just the resource and
offset.

For now, both APIs coexist and depending on the platform, one is
implemented as a wrapper around the other.
Signed-off-by: NDavid Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

In all cases we know which BAR it is.  Passing it in means that arch code
(or generic code; watch this space) won't have to go looking for it again.
Signed-off-by: NDavid Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

These are small wrappers around request_threaded_irq() and free_irq(),
which dynamically allocate space for the device name so that drivers don't
need to keep static buffers for these around.  Additionally it works with
device-relative vector numbers to make the usage easier, and force the
IRQF_SHARED flag on given that it has no runtime overhead and should be
supported by all PCI devices.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Reviewed-by: NThomas Gleixner <tglx@linutronix.de>

This is relatively esoteric, and knowing that we don't have it makes life
easier in some cases rather than just an eventual -EINVAL from
pci_mmap_page_range().
Signed-off-by: NDavid Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

We can declare it <linux/pci.h> even on platforms where it isn't going to
be defined.  There's no need to have it littered through the various
<asm/pci.h> files.
Signed-off-by: NDavid Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

Most of the almost-identical versions of pci_mmap_page_range() silently
ignore the 'write_combine' argument and give uncached mappings.

Yet we allow the PCIIOC_WRITE_COMBINE ioctl in /proc/bus/pci, expose the
'resourceX_wc' file in sysfs, and allow an attempted mapping to apparently
succeed.

To fix this, introduce a macro arch_can_pci_mmap_wc() which indicates
whether the platform can do a write-combining mapping.  On x86 this ends up
being pat_enabled(), while the few other platforms that support it can just
set it to a literal '1'.
Signed-off-by: NDavid Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

Add a new quirk flag PCI_DEV_FLAGS_BRIDGE_XLATE_ROOT to limit the DMA alias
search to go no further than the bridge where the IOMMU unit is attached.

The flag will be used to indicate a bridge device which forwards the
address translation requests to the IOMMU, i.e., where the interrupt and
DMA requests leave the PCIe hierarchy and go into the system blocks.

Usually this happens at the PCI RC, so this flag is not needed.  But on
systems where there are bridges that introduce aliases above the IOMMU,
this flag prevents pci_for_each_dma_alias() from generating aliases that
the IOMMU will never see.

The function pci_for_each_dma_alias() is updated to stop when it see a
bridge with this flag set.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=195447Signed-off-by: NJayachandran C <jnair@caviumnetworks.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Acked-by: NDavid Daney <david.daney@cavium.com>

Unused now that all callers switched to pci_alloc_irq_vectors.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NBjorn Helgaas <bhelgaas@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Per Intel Specification Update 335553-002 (see link below), some 82579
network adapters advertise a Function Level Reset (FLR) capability, but
they can hang when an FLR is triggered.

To reproduce the problem, attach the device to a VM, then detach and try to
attach again.

Add a quirk to prevent the use of FLR on these devices.

[bhelgaas: changelog, comments]
Link: http://www.intel.com/content/dam/www/public/us/en/documents/specification-updates/82579lm-82579v-gigabit-network-connection-spec-update.pdfSigned-off-by: NSasha Neftin <sasha.neftin@intel.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

Detect on probe whether a PCI device is part of a Thunderbolt controller.
Intel uses a Vendor-Specific Extended Capability (VSEC) with ID 0x1234
on such devices.  Detect presence of this VSEC and cache it in a newly
added is_thunderbolt bit in struct pci_dev.

Also, add a helper to check whether a given PCI device is situated on a
Thunderbolt daisy chain (i.e., below a PCI device with is_thunderbolt
set).

The necessity arises from the following:

* If an external Thunderbolt GPU is connected to a dual GPU laptop,
  that GPU is currently registered with vga_switcheroo even though it
  can neither drive the laptop's panel nor be powered off by the
  platform.  To vga_switcheroo it will appear as if two discrete
  GPUs are present.  As a result, when the external GPU is runtime
  suspended, vga_switcheroo will cut power to the internal discrete GPU
  which may not be runtime suspended at all at this moment.  The
  solution is to not register external GPUs with vga_switcheroo, which
  necessitates a way to recognize if they're on a Thunderbolt daisy
  chain.

* Dual GPU MacBook Pros introduced 2011+ can no longer switch external
  DisplayPort ports between GPUs.  (They're no longer just used for DP
  but have become combined DP/Thunderbolt ports.)  The driver to switch
  the ports, drivers/platform/x86/apple-gmux.c, needs to detect presence
  of a Thunderbolt controller and, if found, keep external ports
  permanently switched to the discrete GPU.

v2: Make kerneldoc for pci_is_thunderbolt_attached() more precise,
    drop portion of commit message pertaining to separate series.
    (Bjorn Helgaas)

Cc: Andreas Noever <andreas.noever@gmail.com>
Cc: Michael Jamet <michael.jamet@intel.com>
Cc: Tomas Winkler <tomas.winkler@intel.com>
Cc: Amir Levy <amir.jer.levy@intel.com>
Acked-by: NBjorn Helgaas <bhelgaas@google.com>
Signed-off-by: NLukas Wunner <lukas@wunner.de>
Link: http://patchwork.freedesktop.org/patch/msgid/0ab165a4a35c0b60f29d4c306c653ead14fcd8f9.1489145162.git.lukas@wunner.de

Add a new state to pci_dev to be set when it is unexpectedly disconnected.
The PCI driver tear down functions can observe this new device state so
they may skip operations that will fail.

The pciehp and pcie-dpc drivers are aware when the link is down, so these
set the flag when their handlers detect the device is disconnected.
Tested-by: NKrishna Dhulipala <krishnad@fb.com>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NWei Zhang <wzhang@fb.com>

Replace the inline PCI device config read and write accessors with exported
functions.  This is preparing for these functions to make use of private
data.
Tested-by: NKrishna Dhulipala <krishnad@fb.com>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NWei Zhang <wzhang@fb.com>

pci_msi_shutdown() and pci_msix_shutdown() are used only in
drivers/pci/msi.c, so make them static.
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

Next patch will use the API to get the node from vector for nvme device
Signed-off-by: NShaohua Li <shli@fb.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NBjorn Helgaas <bhelgaas@google.com>
Signed-off-by: NJens Axboe <axboe@fb.com>

No hardware seems to actually call .link_reset(), and no driver implements
it as more than a nop stub.

Drop mentions of the callback from everywhere.  It's dropped from the
documentation as well, but the doc really needs to be updated to reflect
reality better (e.g., on PCIe, slot reset is the link reset).  This will be
done in a later patch.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

Now that pci_bus_type has num_vf callback set, dev_num_vf can be
implemented in a bus type independent way and the check for whether a
PCI device is being handled in rtnetlink can be dropped.
Signed-off-by: NPhil Sutter <phil@nwl.cc>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

All multi-MSI allocations are now done through pci_irq_alloc_vectors(), so
remove the old pci_enable_msi_range() and pci_enable_msi_exact()
interfaces.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

Allow PCI host bridge drivers to use the new host bridge interfaces to
register their host bridge.
Signed-off-by: NThierry Reding <treding@nvidia.com>
Signed-off-by: NBjorn Helgaas <helgaas@kernel.org>