xref: /freebsd/share/man/man9/pci.9 (revision e5cbf0e881fa1851912be77c62aa7ab17f844c3d)

.\"
.\" Copyright (c) 2005 Bruce M Simpson <bms@FreeBSD.org>
.\" All rights reserved.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\"    notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\"    notice, this list of conditions and the following disclaimer in the
.\"    documentation and/or other materials provided with the distribution.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.Dd March 27, 2025
.Dt PCI 9
.Os
.Sh NAME
.Nm pci ,
.Nm pci_alloc_msi ,
.Nm pci_alloc_msix ,
.Nm pci_clear_pme ,
.Nm pci_disable_busmaster ,
.Nm pci_disable_io ,
.Nm pci_enable_busmaster ,
.Nm pci_enable_io ,
.Nm pci_enable_pme ,
.Nm pci_find_bsf ,
.Nm pci_find_cap ,
.Nm pci_find_dbsf ,
.Nm pci_find_device ,
.Nm pci_find_extcap ,
.Nm pci_find_htcap ,
.Nm pci_find_next_cap ,
.Nm pci_find_next_extcap ,
.Nm pci_find_next_htcap ,
.Nm pci_find_pcie_root_port ,
.Nm pci_get_id ,
.Nm pci_get_max_payload ,
.Nm pci_get_max_read_req ,
.Nm pci_get_powerstate ,
.Nm pci_get_vpd_ident ,
.Nm pci_get_vpd_readonly ,
.Nm pci_has_pm ,
.Nm pci_iov_attach ,
.Nm pci_iov_attach_name ,
.Nm pci_iov_detach ,
.Nm pci_msi_count ,
.Nm pci_msix_count ,
.Nm pci_msix_pba_bar ,
.Nm pci_msix_table_bar ,
.Nm pci_pending_msix ,
.Nm pci_read_config ,
.Nm pci_release_msi ,
.Nm pci_remap_msix ,
.Nm pci_restore_state ,
.Nm pci_save_state ,
.Nm pci_set_max_read_req ,
.Nm pci_set_powerstate ,
.Nm pci_write_config ,
.Nm pcie_adjust_config ,
.Nm pcie_flr ,
.Nm pcie_get_max_completion_timeout ,
.Nm pcie_read_config ,
.Nm pcie_wait_for_pending_transactions ,
.Nm pcie_write_config
.Nd PCI bus interface
.Sh SYNOPSIS
.In sys/bus.h
.In dev/pci/pcireg.h
.In dev/pci/pcivar.h
.Ft int
.Fn pci_alloc_msi "device_t dev" "int *count"
.Ft int
.Fn pci_alloc_msix "device_t dev" "int *count"
.Ft void
.Fn pci_clear_pme "device_t dev"
.Ft int
.Fn pci_disable_busmaster "device_t dev"
.Ft int
.Fn pci_disable_io "device_t dev" "int space"
.Ft int
.Fn pci_enable_busmaster "device_t dev"
.Ft int
.Fn pci_enable_io "device_t dev" "int space"
.Ft void
.Fn pci_enable_pme "device_t dev"
.Ft device_t
.Fn pci_find_bsf "uint8_t bus" "uint8_t slot" "uint8_t func"
.Ft int
.Fn pci_find_cap "device_t dev" "int capability" "int *capreg"
.Ft device_t
.Fn pci_find_dbsf "uint32_t domain" "uint8_t bus" "uint8_t slot" "uint8_t func"
.Ft device_t
.Fn pci_find_device "uint16_t vendor" "uint16_t device"
.Ft int
.Fn pci_find_extcap "device_t dev" "int capability" "int *capreg"
.Ft int
.Fn pci_find_htcap "device_t dev" "int capability" "int *capreg"
.Ft int
.Fn pci_find_next_cap "device_t dev" "int capability" "int start" "int *capreg"
.Ft int
.Fn pci_find_next_extcap "device_t dev" "int capability" "int start" "int *capreg"
.Ft int
.Fn pci_find_next_htcap "device_t dev" "int capability" "int start" "int *capreg"
.Ft device_t
.Fn pci_find_pcie_root_port "device_t dev"
.Ft int
.Fn pci_get_id "device_t dev" "enum pci_id_type type" "uintptr_t *id"
.Ft int
.Fn pci_get_max_payload "device_t dev"
.Ft int
.Fn pci_get_max_read_req "device_t dev"
.Ft int
.Fn pci_get_powerstate "device_t dev"
.Ft int
.Fn pci_get_vpd_ident "device_t dev" "const char **identptr"
.Ft int
.Fn pci_get_vpd_readonly "device_t dev" "const char *kw" "const char **vptr"
.Ft bool
.Fn pci_has_pm "device_t dev"
.Ft int
.Fn pci_msi_count "device_t dev"
.Ft int
.Fn pci_msix_count "device_t dev"
.Ft int
.Fn pci_msix_pba_bar "device_t dev"
.Ft int
.Fn pci_msix_table_bar "device_t dev"
.Ft int
.Fn pci_pending_msix "device_t dev" "u_int index"
.Ft uint32_t
.Fn pci_read_config "device_t dev" "int reg" "int width"
.Ft int
.Fn pci_release_msi "device_t dev"
.Ft int
.Fn pci_remap_msix "device_t dev" "int count" "const u_int *vectors"
.Ft void
.Fn pci_restore_state "device_t dev"
.Ft void
.Fn pci_save_state "device_t dev"
.Ft int
.Fn pci_set_max_read_req "device_t dev" "int size"
.Ft int
.Fn pci_set_powerstate "device_t dev" "int state"
.Ft void
.Fn pci_write_config "device_t dev" "int reg" "uint32_t val" "int width"
.Ft uint32_t
.Fo pcie_adjust_config
.Fa "device_t dev"
.Fa "int reg"
.Fa "uint32_t mask"
.Fa "uint32_t val"
.Fa "int width"
.Fc
.Ft bool
.Fn pcie_flr "device_t dev" "u_int max_delay" "bool force"
.Ft int
.Fn pcie_get_max_completion_timeout "device_t dev"
.Ft uint32_t
.Fn pcie_read_config "device_t dev" "int reg" "int width"
.Ft bool
.Fn pcie_wait_for_pending_transactions "device_t dev" "u_int max_delay"
.Ft void
.Fn pcie_write_config "device_t dev" "int reg" "uint32_t val" "int width"
.Ft void
.Fn pci_event_fn "void *arg" "device_t dev"
.Fn EVENTHANDLER_REGISTER "pci_add_device" "pci_event_fn"
.Fn EVENTHANDLER_DEREGISTER "pci_delete_resource" "pci_event_fn"
.In dev/pci/pci_iov.h
.Ft int
.Fn pci_iov_attach "device_t dev" "nvlist_t *pf_schema" "nvlist_t *vf_schema"
.Ft int
.Fo pci_iov_attach_name
.Fa "device_t dev"
.Fa "nvlist_t *pf_schema"
.Fa "nvlist_t *vf_schema"
.Fa "const char *fmt"
.Fa "..."
.Fc
.Ft int
.Fn pci_iov_detach "device_t dev"
.Sh DESCRIPTION
The
.Nm
set of functions are used for managing PCI devices.
The functions are split into several groups:
raw configuration access,
locating devices,
device information,
device configuration,
and
message signaled interrupts.
.Ss Raw Configuration Access
The
.Fn pci_read_config
function is used to read data from the PCI configuration
space of the device
.Fa dev ,
at offset
.Fa reg ,
with
.Fa width
specifying the size of the access.
.Pp
The
.Fn pci_write_config
function is used to write the value
.Fa val
to the PCI configuration
space of the device
.Fa dev ,
at offset
.Fa reg ,
with
.Fa width
specifying the size of the access.
.Pp
The
.Fn pcie_adjust_config
function is used to modify the value of a register in the PCI-express
capability register set of device
.Fa dev .
The offset
.Fa reg
specifies a relative offset in the register set with
.Fa width
specifying the size of the access.
The new value of the register is computed by modifying bits set in
.Fa mask
to the value in
.Fa val .
Any bits not specified in
.Fa mask
are preserved.
The previous value of the register is returned.
.Pp
The
.Fn pcie_read_config
function is used to read the value of a register in the PCI-express
capability register set of device
.Fa dev .
The offset
.Fa reg
specifies a relative offset in the register set with
.Fa width
specifying the size of the access.
.Pp
The
.Fn pcie_write_config
function is used to write the value
.Fa val
to a register in the PCI-express capability register set of device
.Fa dev .
The offset
.Fa reg
specifies a relative offset in the register set with
.Fa width
specifying the size of the access.
.Pp
.Em NOTE :
Device drivers should only use these functions for functionality that
is not available via another
.Fn pci
function.
.Ss Locating Devices
The
.Fn pci_find_bsf
function looks up the
.Vt device_t
of a PCI device, given its
.Fa bus ,
.Fa slot ,
and
.Fa func .
The
.Fa slot
number actually refers to the number of the device on the bus,
which does not necessarily indicate its geographic location
in terms of a physical slot.
Note that in case the system has multiple PCI domains,
the
.Fn pci_find_bsf
function only searches the first one.
Actually, it is equivalent to:
.Bd -literal -offset indent
pci_find_dbsf(0, bus, slot, func);
.Ed
.Pp
The
.Fn pci_find_dbsf
function looks up the
.Vt device_t
of a PCI device, given its
.Fa domain ,
.Fa bus ,
.Fa slot ,
and
.Fa func .
The
.Fa slot
number actually refers to the number of the device on the bus,
which does not necessarily indicate its geographic location
in terms of a physical slot.
.Pp
The
.Fn pci_find_device
function looks up the
.Vt device_t
of a PCI device, given its
.Fa vendor
and
.Fa device
IDs.
Note that there can be multiple matches for this search; this function
only returns the first matching device.
.Ss Device Information
The
.Fn pci_find_cap
function is used to locate the first instance of a PCI capability
register set for the device
.Fa dev .
The capability to locate is specified by ID via
.Fa capability .
Constant macros of the form
.Dv PCIY_xxx
for standard capability IDs are defined in
.In dev/pci/pcireg.h .
If the capability is found, then
.Fa *capreg
is set to the offset in configuration space of the capability register set,
and
.Fn pci_find_cap
returns zero.
If the capability is not found or the device does not support capabilities,
.Fn pci_find_cap
returns an error.
The
.Fn pci_find_next_cap
function is used to locate the next instance of a PCI capability
register set for the device
.Fa dev .
The
.Fa start
should be the
.Fa *capreg
returned by a prior
.Fn pci_find_cap
or
.Fn pci_find_next_cap .
When no more instances are located
.Fn pci_find_next_cap
returns an error.
.Pp
The
.Fn pci_has_pm
function returns true if
.Fa dev
supports power management.
.Pp
The
.Fn pci_find_extcap
function is used to locate the first instance of a PCI-express
extended capability register set for the device
.Fa dev .
The extended capability to locate is specified by ID via
.Fa capability .
Constant macros of the form
.Dv PCIZ_xxx
for standard extended capability IDs are defined in
.In dev/pci/pcireg.h .
If the extended capability is found, then
.Fa *capreg
is set to the offset in configuration space of the extended capability
register set, and
.Fn pci_find_extcap
returns zero.
If the extended capability is not found or the device is not a
PCI-express device,
.Fn pci_find_extcap
returns an error.
The
.Fn pci_find_next_extcap
function is used to locate the next instance of a PCI-express
extended capability register set for the device
.Fa dev .
The
.Fa start
should be the
.Fa *capreg
returned by a prior
.Fn pci_find_extcap
or
.Fn pci_find_next_extcap .
When no more instances are located
.Fn pci_find_next_extcap
returns an error.
.Pp
The
.Fn pci_find_htcap
function is used to locate the first instance of a HyperTransport capability
register set for the device
.Fa dev .
The capability to locate is specified by type via
.Fa capability .
Constant macros of the form
.Dv PCIM_HTCAP_xxx
for standard HyperTransport capability types are defined in
.In dev/pci/pcireg.h .
If the capability is found, then
.Fa *capreg
is set to the offset in configuration space of the capability register set,
and
.Fn pci_find_htcap
returns zero.
If the capability is not found or the device is not a HyperTransport device,
.Fn pci_find_htcap
returns an error.
The
.Fn pci_find_next_htcap
function is used to locate the next instance of a HyperTransport capability
register set for the device
.Fa dev .
The
.Fa start
should be the
.Fa *capreg
returned by a prior
.Fn pci_find_htcap
or
.Fn pci_find_next_htcap .
When no more instances are located
.Fn pci_find_next_htcap
returns an error.
.Pp
The
.Fn pci_find_pcie_root_port
function walks up the PCI device hierarchy to locate the PCI-express root
port upstream of
.Fa dev .
If a root port is not found,
.Fn pci_find_pcie_root_port
returns
.Dv NULL .
.Pp
The
.Fn pci_get_id
function is used to read an identifier from a device.
The
.Fa type
flag is used to specify which identifier to read.
The following flags are supported:
.Bl -hang -width ".Dv PCI_ID_RID"
.It Dv PCI_ID_RID
Read the routing identifier for the device.
.It Dv PCI_ID_MSI
Read the MSI routing ID.
This is needed by some interrupt controllers to route MSI and MSI-X interrupts.
.El
.Pp
The
.Fn pci_get_vpd_ident
function is used to fetch a device's Vital Product Data
.Pq VPD
identifier string.
If the device
.Fa dev
supports VPD and provides an identifier string,
then
.Fa *identptr
is set to point at a read-only, null-terminated copy of the identifier
string,
and
.Fn pci_get_vpd_ident
returns zero.
If the device does not support VPD or does not provide an identifier
string,
then
.Fn pci_get_vpd_ident
returns an error.
.Pp
The
.Fn pci_get_vpd_readonly
function is used to fetch the value of a single VPD read-only keyword
for the device
.Fa dev .
The keyword to fetch is identified by the two character string
.Fa kw .
If the device supports VPD and provides a read-only value for the
requested keyword,
then
.Fa *vptr
is set to point at a read-only, null-terminated copy of the value,
and
.Fn pci_get_vpd_readonly
returns zero.
If the device does not support VPD or does not provide the requested
keyword,
then
.Fn pci_get_vpd_readonly
returns an error.
.Pp
The
.Fn pcie_get_max_completion_timeout
function returns the maximum completion timeout configured for the device
.Fa dev
in microseconds.
If the
.Fa dev
device is not a PCI-express device,
.Fn pcie_get_max_completion_timeout
returns zero.
When completion timeouts are disabled for
.Fa dev ,
this function returns the maxmimum timeout that would be used if timeouts
were enabled.
.Pp
The
.Fn pcie_wait_for_pending_transactions
function waits for any pending transactions initiated by the
.Fa dev
device to complete.
The function checks for pending transactions by polling the transactions
pending flag in the PCI-express device status register.
It returns
.Dv true
once the transaction pending flag is clear.
If transactions are still pending after
.Fa max_delay
milliseconds,
.Fn pcie_wait_for_pending_transactions
returns
.Dv false .
If
.Fa max_delay
is set to zero,
.Fn pcie_wait_for_pending_transactions
performs a single check;
otherwise,
this function may sleep while polling the transactions pending flag.
.Nm pcie_wait_for_pending_transactions
returns
.Dv true
if
.Fa dev
is not a PCI-express device.
.Ss Device Configuration
The
.Fn pci_enable_busmaster
function enables PCI bus mastering for the device
.Fa dev ,
by setting the
.Dv PCIM_CMD_BUSMASTEREN
bit in the
.Dv PCIR_COMMAND
register.
The
.Fn pci_disable_busmaster
function clears this bit.
.Pp
The
.Fn pci_enable_io
function enables memory or I/O port address decoding for the device
.Fa dev ,
by setting the
.Dv PCIM_CMD_MEMEN
or
.Dv PCIM_CMD_PORTEN
bit in the
.Dv PCIR_COMMAND
register appropriately.
The
.Fn pci_disable_io
function clears the appropriate bit.
The
.Fa space
argument specifies which resource is affected; this can be either
.Dv SYS_RES_MEMORY
or
.Dv SYS_RES_IOPORT
as appropriate.
Device drivers should generally not use these routines directly.
The PCI bus will enable decoding automatically when a
.Dv SYS_RES_MEMORY
or
.Dv SYS_RES_IOPORT
resource is activated via
.Xr bus_alloc_resource 9
or
.Xr bus_activate_resource 9 .
.Pp
The
.Fn pci_get_max_payload
function returns the current maximum TLP payload size in bytes for a
PCI-express device.
If the
.Fa dev
device is not a PCI-express device,
.Fn pci_get_max_payload
returns zero.
.Pp
The
.Fn pci_get_max_read_req
function returns the current maximum read request size in bytes for a
PCI-express device.
If the
.Fa dev
device is not a PCI-express device,
.Fn pci_get_max_read_req
returns zero.
.Pp
The
.Fn pci_set_max_read_req
sets the PCI-express maximum read request size for
.Fa dev .
The requested
.Fa size
may be adjusted,
and
.Fn pci_set_max_read_req
returns the actual size set in bytes.
If the
.Fa dev
device is not a PCI-express device,
.Fn pci_set_max_read_req
returns zero.
.Pp
The
.Fn pci_get_powerstate
function returns the current power state of the device
.Fa dev .
If the device does not support power management capabilities, then the default
state of
.Dv PCI_POWERSTATE_D0
is returned.
The following power states are defined by PCI:
.Bl -hang -width ".Dv PCI_POWERSTATE_UNKNOWN"
.It Dv PCI_POWERSTATE_D0
State in which device is on and running.
It is receiving full power from the system and delivering
full functionality to the user.
.It Dv PCI_POWERSTATE_D1
Class-specific low-power state in which device context may or
may not be lost.
Buses in this state cannot do anything to the bus, to
force devices to lose context.
.It Dv PCI_POWERSTATE_D2
Class-specific low-power state in which device context may or
may not be lost.
Attains greater power savings than
.Dv PCI_POWERSTATE_D1 .
Buses in this state can cause devices to lose some context.
Devices
.Em must
be prepared for the bus to be in this state or higher.
.It Dv PCI_POWERSTATE_D3
State in which the device is off and not running.
Device context is lost, and power from the device can
be removed.
.It Dv PCI_POWERSTATE_UNKNOWN
State of the device is unknown.
.El
.Pp
The
.Fn pci_set_powerstate
function is used to transition the device
.Fa dev
to the PCI power state
.Fa state .
If the device does not support power management capabilities or
it does not support the specific power state
.Fa state ,
then the function will fail with
.Er EOPNOTSUPP .
.Pp
The
.Fn pci_clear_pme
function is used to clear any pending PME# signal and disable generation
of power management events.
.Pp
The
.Fn pci_enable_pme
function is used to enable generation of power management events before
suspending a device.
.Pp
The
.Fn pci_iov_attach
function is used to advertise that the given device
.Pq and associated device driver
supports PCI Single-Root I/O Virtualization
.Pq SR-IOV .
A driver that supports SR-IOV must implement the
.Xr PCI_IOV_INIT 9 ,
.Xr PCI_IOV_ADD_VF 9
and
.Xr PCI_IOV_UNINIT 9
methods.
This function should be called during the
.Xr DEVICE_ATTACH 9
method.
If this function returns an error, it is recommended that the device driver
still successfully attaches, but runs with SR-IOV disabled.
The
.Fa pf_schema
and
.Fa vf_schema
parameters are used to define what device-specific configuration parameters the
device driver accepts when SR-IOV is enabled for the Physical Function
.Pq PF
and for individual Virtual Functions
.Pq VFs
respectively.
See
.Xr pci_iov_schema 9
for details on how to construct the schema.
If either the
.Pa pf_schema
or
.Pa vf_schema
is invalid or specifies parameter names that conflict with parameter names that
are already in use,
.Fn pci_iov_attach
will return an error and SR-IOV will not be available on the PF device.
If a driver does not accept configuration parameters for either the PF device
or the VF devices, the driver must pass an empty schema for that device.
The SR-IOV infrastructure takes ownership of the
.Fa pf_schema
and
.Fa vf_schema
and is responsible for freeing them.
The driver must never free the schemas itself.
.Pp
The
.Fn pci_iov_attach_name
function is a variant of
.Fn pci_iov_attach
that allows the name of the associated character device in
.Pa /dev/iov
to be specified by
.Fa fmt .
The
.Fn pci_iov_attach
function uses the name of
.Fa dev
as the device name.
.Pp
The
.Fn pci_iov_detach
function is used to advise the SR-IOV infrastructure that the driver for the
given device is attempting to detach and that all SR-IOV resources for the
device must be released.
This function must be called during the
.Xr DEVICE_DETACH 9
method if
.Fn pci_iov_attach
was successfully called on the device and
.Fn pci_iov_detach
has not subsequently been called on the device and returned no error.
If this function returns an error, the
.Xr DEVICE_DETACH 9
method must fail and return an error, as detaching the PF driver while VF
devices are active would cause system instability.
This function is safe to call and will always succeed if
.Fn pci_iov_attach
previously failed with an error on the given device, or if
.Fn pci_iov_attach
was never called on the device.
.Pp
The
.Fn pci_save_state
and
.Fn pci_restore_state
functions can be used by a device driver to save and restore standard PCI
config registers.
The
.Fn pci_save_state
function must be invoked while the device has valid state before
.Fn pci_restore_state
can be used.
If the device is not in the fully-powered state
.Pq Dv PCI_POWERSTATE_D0
when
.Fn pci_restore_state
is invoked,
then the device will be transitioned to
.Dv PCI_POWERSTATE_D0
before any config registers are restored.
.Pp
The
.Fn pcie_flr
function requests a Function Level Reset
.Pq FLR
of
.Fa dev .
If
.Fa dev
is not a PCI-express device or does not support Function Level Resets via
the PCI-express device control register,
.Dv false
is returned.
Pending transactions are drained by disabling busmastering and calling
.Fn pcie_wait_for_pending_transactions
before resetting the device.
The
.Fa max_delay
argument specifies the maximum timeout to wait for pending transactions as
described for
.Fn pcie_wait_for_pending_transactions .
If
.Fn pcie_wait_for_pending_transactions
fails with a timeout and
.Fa force
is
.Dv false ,
busmastering is re-enabled and
.Dv false
is returned.
If
.Fn pcie_wait_for_pending_transactions
fails with a timeout and
.Fa force
is
.Dv true ,
the device is reset despite the timeout.
After the reset has been requested,
.Nm pcie_flr
sleeps for at least 100 milliseconds before returning
.Dv true .
Note that
.Nm pcie_flr
does not save and restore any state around the reset.
The caller should save and restore state as needed.
.Ss Message Signaled Interrupts
Message Signaled Interrupts
.Pq MSI
and
Enhanced Message Signaled Interrupts
.Pq MSI-X
are PCI capabilities that provide an alternate method for PCI
devices to signal interrupts.
The legacy INTx interrupt is available to PCI devices as a
.Dv SYS_RES_IRQ
resource with a resource ID of zero.
MSI and MSI-X interrupts are available to PCI devices as one or more
.Dv SYS_RES_IRQ
resources with resource IDs greater than zero.
A driver must ask the PCI bus to allocate MSI or MSI-X interrupts
using
.Fn pci_alloc_msi
or
.Fn pci_alloc_msix
before it can use MSI or MSI-X
.Dv SYS_RES_IRQ
resources.
A driver is not allowed to use the legacy INTx
.Dv SYS_RES_IRQ
resource if MSI or MSI-X interrupts have been allocated,
and attempts to allocate MSI or MSI-X interrupts will fail if the
driver is currently using the legacy INTx
.Dv SYS_RES_IRQ
resource.
A driver is only allowed to use either MSI or MSI-X,
but not both.
.Pp
The
.Fn pci_msi_count
function returns the maximum number of MSI messages supported by the
device
.Fa dev .
If the device does not support MSI,
then
.Fn pci_msi_count
returns zero.
.Pp
The
.Fn pci_alloc_msi
function attempts to allocate
.Fa *count
MSI messages for the device
.Fa dev .
The
.Fn pci_alloc_msi
function may allocate fewer messages than requested for various
reasons including requests for more messages than the device
.Fa dev
supports,
or if the system has a shortage of available MSI messages.
On success,
.Fa *count
is set to the number of messages allocated and
.Fn pci_alloc_msi
returns zero.
The
.Dv SYS_RES_IRQ
resources for the allocated messages will be available at consecutive
resource IDs beginning with one.
If
.Fn pci_alloc_msi
is not able to allocate any messages,
it returns an error.
Note that MSI only supports message counts that are powers of two;
requests to allocate a non-power of two count of messages will fail.
.Pp
The
.Fn pci_release_msi
function is used to release any allocated MSI or MSI-X messages back
to the system.
If any MSI or MSI-X
.Dv SYS_RES_IRQ
resources are allocated by the driver or have a configured interrupt
handler,
this function will fail with
.Er EBUSY .
The
.Fn pci_release_msi
function returns zero on success and an error on failure.
.Pp
The
.Fn pci_msix_count
function returns the maximum number of MSI-X messages supported by the
device
.Fa dev .
If the device does not support MSI-X,
then
.Fn pci_msix_count
returns zero.
.Pp
The
.Fn pci_msix_pba_bar
function returns the offset in configuration space of the Base Address Register
.Pq BAR
containing the MSI-X Pending Bit Array (PBA) for device
.Fa dev .
The returned value can be used as the resource ID with
.Xr bus_alloc_resource 9
and
.Xr bus_release_resource 9
to allocate the BAR.
If the device does not support MSI-X,
then
.Fn pci_msix_pba_bar
returns -1.
.Pp
The
.Fn pci_msix_table_bar
function returns the offset in configuration space of the BAR
containing the MSI-X vector table for device
.Fa dev .
The returned value can be used as the resource ID with
.Xr bus_alloc_resource 9
and
.Xr bus_release_resource 9
to allocate the BAR.
If the device does not support MSI-X,
then
.Fn pci_msix_table_bar
returns -1.
.Pp
The
.Fn pci_alloc_msix
function attempts to allocate
.Fa *count
MSI-X messages for the device
.Fa dev .
The
.Fn pci_alloc_msix
function may allocate fewer messages than requested for various
reasons including requests for more messages than the device
.Fa dev
supports,
or if the system has a shortage of available MSI-X messages.
On success,
.Fa *count
is set to the number of messages allocated and
.Fn pci_alloc_msix
returns zero.
For MSI-X messages,
the resource ID for each
.Dv SYS_RES_IRQ
resource identifies the index in the MSI-X table of the
corresponding message.
A resource ID of one maps to the first index of the MSI-X table;
a resource ID two identifies the second index in the table, etc.
The
.Fn pci_alloc_msix
function assigns the
.Fa *count
messages allocated to the first
.Fa *count
table indices.
If
.Fn pci_alloc_msix
is not able to allocate any messages,
it returns an error.
Unlike MSI,
MSI-X does not require message counts that are powers of two.
.Pp
The BARs containing the MSI-X vector table and PBA must be
allocated via
.Xr bus_alloc_resource 9
before calling
.Fn pci_alloc_msix
and must not be released until after calling
.Fn pci_release_msi .
Note that the vector table and PBA may be stored in the same BAR or in
different BARs.
.Pp
The
.Fn pci_pending_msix
function examines the
.Fa dev
device's PBA
to determine the pending status of the MSI-X message at table index
.Fa index .
If the indicated message is pending,
this function returns a non-zero value;
otherwise,
it returns zero.
Passing an invalid
.Fa index
to this function will result in undefined behavior.
.Pp
As mentioned in the description of
.Fn pci_alloc_msix ,
MSI-X messages are initially assigned to the first N table entries.
A driver may use a different distribution of available messages to
table entries via the
.Fn pci_remap_msix
function.
Note that this function must be called after a successful call to
.Fn pci_alloc_msix
but before any of the
.Dv SYS_RES_IRQ
resources are allocated.
The
.Fn pci_remap_msix
function returns zero on success,
or an error on failure.
.Pp
The
.Fa vectors
array should contain
.Fa count
message vectors.
The array maps directly to the MSI-X table in that the first entry in
the array specifies the message used for the first entry in the MSI-X
table,
the second entry in the array corresponds to the second entry in the
MSI-X table,
etc.
The vector value in each array index can either be zero to indicate
that no message should be assigned to the corresponding MSI-X table entry,
or it can be a number from one to N
.Po
where N is the count returned from the previous call to
.Fn pci_alloc_msix
.Pc
to indicate which of the allocated messages should be assigned to the
corresponding MSI-X table entry.
.Pp
If
.Fn pci_remap_msix
succeeds,
each MSI-X table entry with a non-zero vector will have an associated
.Dv SYS_RES_IRQ
resource whose resource ID corresponds to the table index as described
above for
.Fn pci_alloc_msix .
MSI-X table entries that with a vector of zero will not have an
associated
.Dv SYS_RES_IRQ
resource.
Additionally,
if any of the original messages allocated by
.Fn pci_alloc_msix
are not used in the new distribution of messages in the MSI-X table,
they will be released automatically.
Note that if a driver wishes to use fewer messages than were allocated by
.Fn pci_alloc_msix ,
the driver must use a single, contiguous range of messages beginning
with one in the new distribution.
The
.Fn pci_remap_msix
function will fail if this condition is not met.
.Ss Device Events
The
.Va pci_add_device
event handler is invoked every time a new PCI device is added to the system.
This includes the creation of Virtual Functions via SR-IOV.
.Pp
The
.Va pci_delete_device
event handler is invoked every time a PCI device is removed from the system.
.Pp
Both event handlers pass the
.Vt device_t
object of the relevant PCI device as
.Fa dev
to each callback function.
Both event handlers are invoked while
.Fa dev
is unattached but with valid instance variables.
.Sh SEE ALSO
.Xr pci 4 ,
.Xr pciconf 8 ,
.Xr bus_alloc_resource 9 ,
.Xr bus_dma 9 ,
.Xr bus_release_resource 9 ,
.Xr bus_setup_intr 9 ,
.Xr bus_teardown_intr 9 ,
.Xr devclass 9 ,
.Xr device 9 ,
.Xr driver 9 ,
.Xr eventhandler 9 ,
.Xr rman 9
.Rs
.%B FreeBSD Developers' Handbook
.%T NewBus
.%U https://docs.freebsd.org/en/books/developers-handbook/
.Re
.Rs
.%A Shanley
.%A Anderson
.%B PCI System Architecture
.%N 2nd Edition
.%I Addison-Wesley
.%O ISBN 0-201-30974-2
.Re
.Sh AUTHORS
.An -nosplit
This manual page was written by
.An Bruce M Simpson Aq Mt bms@FreeBSD.org
and
.An John Baldwin Aq Mt jhb@FreeBSD.org .
.Sh BUGS
The kernel PCI code has a number of references to
.Dq "slot numbers" .
These do not refer to the geographic location of PCI devices,
but to the device number assigned by the combination of the PCI IDSEL
mechanism and the platform firmware.
This should be taken note of when working with the kernel PCI code.
.Pp
The PCI bus driver should allocate the MSI-X vector table and PBA internally
as necessary rather than requiring the caller to do so.