d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
1af4a58503
freebsd-skq/share/man/man9/rman.9
John Baldwin cc981af204 Add new bus methods for mapping resources. 
Add a pair of bus methods that can be used to "map" resources for direct
CPU access using bus_space(9).  bus_map_resource() creates a mapping and
bus_unmap_resource() releases a previously created mapping.  Mappings are
described by 'struct resource_map' object.  Pointers to these objects can
be passed as the first argument to the bus_space wrapper API used for bus
resources.

Drivers that wish to map all of a resource using default settings
(for example, using uncacheable memory attributes) do not need to change.
However, drivers that wish to use non-default settings can now do so
without jumping through hoops.

First, an RF_UNMAPPED flag is added to request that a resource is not
implicitly mapped with the default settings when it is activated.  This
permits other activation steps (such as enabling I/O or memory decoding
in a device's PCI command register) to be taken without creating a
mapping.  Right now the AGP drivers don't set RF_ACTIVE to avoid using
up a large amount of KVA to map the AGP aperture on 32-bit platforms.
Once RF_UNMAPPED is supported on all platforms that support AGP this
can be changed to using RF_UNMAPPED with RF_ACTIVE instead.

Second, bus_map_resource accepts an optional structure that defines
additional settings for a given mapping.

For example, a driver can now request to map only a subset of a resource
instead of the entire range.  The AGP driver could also use this to only
map the first page of the aperture (IIRC, it calls pmap_mapdev() directly
to map the first page currently).  I will also eventually change the
PCI-PCI bridge driver to request mappings of the subset of the I/O window
resource on its parent side to create mappings for child devices rather
than passing child resources directly up to nexus to be mapped.  This
also permits bridges that do address translation to request suitable
mappings from a resource on the "upper" side of the bus when mapping
resources on the "lower" side of the bus.

Another attribute that can be specified is an alternate memory attribute
for memory-mapped resources.  This can be used to request a
Write-Combining mapping of a PCI BAR in an MI fashion.  (Currently the
drivers that do this call pmap_change_attr() directly for x86 only.)

Note that this commit only adds the MI framework.  Each platform needs
to add support for handling RF_UNMAPPED and thew new
bus_map/unmap_resource methods.  Generally speaking, any drivers that
are calling rman_set_bustag() and rman_set_bushandle() need to be
updated.

Discussed on:	arch
Reviewed by:	cem
Differential Revision:	https://reviews.freebsd.org/D5237
2016-05-20 17:57:47 +00:00

477 lines
13 KiB
Groff
Raw Blame History

.\"
.\" Copyright (c) 2003 Bruce M Simpson <bms@spc.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.
.\"
.\" $FreeBSD$
.\"
.Dd May 20, 2016
.Dt RMAN 9
.Os
.Sh NAME
.Nm rman ,
.Nm rman_activate_resource ,
.Nm rman_adjust_resource ,
.Nm rman_deactivate_resource ,
.Nm rman_fini ,
.Nm rman_init ,
.Nm rman_init_from_resource ,
.Nm rman_is_region_manager ,
.Nm rman_manage_region ,
.Nm rman_first_free_region ,
.Nm rman_last_free_region ,
.Nm rman_release_resource ,
.Nm rman_reserve_resource ,
.Nm rman_reserve_resource_bound ,
.Nm rman_make_alignment_flags ,
.Nm rman_get_start ,
.Nm rman_get_end ,
.Nm rman_get_device ,
.Nm rman_get_size ,
.Nm rman_get_flags ,
.Nm rman_set_mapping ,
.Nm rman_get_mapping ,
.Nm rman_set_virtual ,
.Nm rman_get_virtual ,
.Nm rman_set_bustag ,
.Nm rman_get_bustag ,
.Nm rman_set_bushandle ,
.Nm rman_get_bushandle ,
.Nm rman_set_rid ,
.Nm rman_get_rid
.Nd resource management functions
.Sh SYNOPSIS
.In sys/types.h
.In sys/rman.h
.Ft int
.Fn rman_activate_resource "struct resource *r"
.Ft int
.Fn rman_adjust_resource "struct resource *r" "rman_res_t start" "rman_res_t end"
.Ft int
.Fn rman_deactivate_resource "struct resource *r"
.Ft int
.Fn rman_fini "struct rman *rm"
.Ft int
.Fn rman_init "struct rman *rm"
.Ft int
.Fn rman_init_from_resource "struct rman *rm" "struct resource *r"
.Ft int
.Fn rman_is_region_manager "struct resource *r" "struct rman *rm"
.Ft int
.Fn rman_manage_region "struct rman *rm" "rman_res_t start" "rman_res_t end"
.Ft int
.Fn rman_first_free_region "struct rman *rm" "rman_res_t *start" "rman_res_t *end"
.Ft int
.Fn rman_last_free_region "struct rman *rm" "rman_res_t *start" "rman_res_t *end"
.Ft int
.Fn rman_release_resource "struct resource *r"
.Ft "struct resource *"
.Fo rman_reserve_resource
.Fa "struct rman *rm" "rman_res_t start" "rman_res_t end" "rman_res_t count"
.Fa "u_int flags" "struct device *dev"
.Fc
.Ft "struct resource *"
.Fo rman_reserve_resource_bound
.Fa "struct rman *rm" "rman_res_t start" "rman_res_t end" "rman_res_t count"
.Fa "rman_res_t bound" "u_int flags" "struct device *dev"
.Fc
.Ft uint32_t
.Fn rman_make_alignment_flags "uint32_t size"
.Ft rman_res_t
.Fn rman_get_start "struct resource *r"
.Ft rman_res_t
.Fn rman_get_end "struct resource *r"
.Ft "struct device *"
.Fn rman_get_device "struct resource *r"
.Ft rman_res_t
.Fn rman_get_size "struct resource *r"
.Ft u_int
.Fn rman_get_flags "struct resource *r"
.Ft void
.Fn rman_set_mapping "struct resource *r" "struct resource_map *map"
.Ft void
.Fn rman_get_mapping "struct resource *r" "struct resource_map *map"
.Ft void
.Fn rman_set_virtual "struct resource *r" "void *v"
.Ft "void *"
.Fn rman_get_virtual "struct resource *r"
.Ft void
.Fn rman_set_bustag "struct resource *r" "bus_space_tag_t t"
.Ft bus_space_tag_t
.Fn rman_get_bustag "struct resource *r"
.Ft void
.Fn rman_set_bushandle "struct resource *r" "bus_space_handle_t h"
.Ft bus_space_handle_t
.Fn rman_get_bushandle "struct resource *r"
.Ft void
.Fn rman_set_rid "struct resource *r" "int rid"
.Ft int
.Fn rman_get_rid "struct resource *r"
.Sh DESCRIPTION
The
.Nm
set of functions provides a flexible resource management abstraction.
It is used extensively by the bus management code.
It implements the abstractions of region and resource.
A region descriptor is used to manage a region; this could be memory or
some other form of bus space.
.Pp
Each region has a set of bounds.
Within these bounds, allocated segments may reside.
Each segment, termed a resource, has several properties which are
represented by a 16-bit flag register, as follows.
.Bd -literal
#define RF_ALLOCATED 0x0001 /* resource has been reserved */
#define RF_ACTIVE 0x0002 /* resource allocation has been activated */
#define RF_SHAREABLE 0x0004 /* resource permits contemporaneous sharing */
#define RF_FIRSTSHARE 0x0020 /* first in sharing list */
#define RF_PREFETCHABLE 0x0040 /* resource is prefetchable */
#define RF_UNMAPPED 0x0100 /* don't map resource when activating */
.Ed
.Pp
Bits 15:10 of the flag register are used to represent the desired alignment
of the resource within the region.
.Pp
The
.Fn rman_init
function initializes the region descriptor, pointed to by the
.Fa rm
argument, for use with the resource management functions.
It is required that the fields
.Va rm_type
and
.Va rm_descr
of
.Vt "struct rman"
be set before calling
.Fn rman_init .
The field
.Va rm_type
shall be set to
.Dv RMAN_ARRAY .
The field
.Va rm_descr
shall be set to a string that describes the resource to be managed.
The
.Va rm_start
and
.Va rm_end
fields may be set to limit the range of acceptable resource addresses.
If these fields are not set,
.Fn rman_init
will initialize them to allow the entire range of resource addresses.
It also initializes any mutexes associated with the structure.
If
.Fn rman_init
fails to initialize the mutex, it will return
.Er ENOMEM ; otherwise it will return 0 and
.Fa rm
will be initialized.
.Pp
The
.Fn rman_fini
function frees any structures associated with the structure
pointed to by the
.Fa rm
argument.
If any of the resources within the managed region have the
.Dv RF_ALLOCATED
flag set, it will return
.Er EBUSY ;
otherwise, any mutexes associated with the structure will be released
and destroyed, and the function will return 0.
.Pp
The
.Fn rman_manage_region
function establishes the concept of a region which is under
.Nm
control.
The
.Fa rman
argument points to the region descriptor.
The
.Fa start
and
.Fa end
arguments specify the bounds of the region.
If successful,
.Fn rman_manage_region
will return 0.
If the region overlaps with an existing region, it will return
.Er EBUSY .
If any part of the region falls outside of the valid address range for
.Fa rm ,
it will return
.Er EINVAL .
.Er ENOMEM
will be returned when
.Fn rman_manage_region
failed to allocate memory for the region.
.Pp
The
.Fn rman_init_from_resource
function is a wrapper routine to create a resource manager backed by an
existing resource.
It initializes
.Fa rm
using
.Fn rman_init
and then adds a region to
.Fa rm
corresponding to the address range allocated to
.Fa r
via
.Fn rman_manage_region .
.Pp
The
.Fn rman_first_free_region
and
.Fn rman_last_free_region
functions can be used to query a resource manager for its first
.Pq or last
unallocated region.
If
.Fa rm
contains no free region,
these functions will return
.Er ENOENT .
Otherwise,
.Fa *start
and
.Fa *end
are set to the bounds of the free region and zero is returned.
.Pp
The
.Fn rman_reserve_resource_bound
function is where the bulk of the
.Nm
logic is located.
It attempts to reserve a contiguous range in the specified region
.Fa rm
for the use of the device
.Fa dev .
The caller can specify the
.Fa start
and
.Fa end
of an acceptable range,
as well as a boundary restriction and required aligment,
and the code will attempt to find a free segment which fits.
The
.Fa start
argument is the lowest acceptable starting value of the resource.
The
.Fa end
argument is the highest acceptable ending value of the resource.
Therefore,
.Fa start No + Fa count No \- 1
must be \[<=]
.Fa end
for any allocation to happen.
The aligment requirement
.Pq if any
is specified in
.Fa flags .
The
.Fa bound
argument may be set to specify a boundary restriction such that an
allocated region may cross an address that is a multiple of the
boundary.
The
.Fa bound
argument must be a power of two.
It may be set to zero to specify no boundary restriction.
A shared segment will be allocated if the
.Dv RF_SHAREABLE
flag is set, otherwise an exclusive segment will be allocated.
If this shared segment already exists, the caller has its device
added to the list of consumers.
.Pp
The
.Fn rman_reserve_resource
function is used to reserve resources within a previously established region.
It is a simplified interface to
.Fn rman_reserve_resource_bound
which passes 0 for the
.Fa bound
argument.
.Pp
The
.Fn rman_make_alignment_flags
function returns the flag mask corresponding to the desired alignment
.Fa size .
This should be used when calling
.Fn rman_reserve_resource_bound .
.Pp
The
.Fn rman_is_region_manager
function returns true if the allocated resource
.Fa r
was allocated from
.Fa rm .
Otherwise,
it returns false.
.Pp
The
.Fn rman_adjust_resource
function is used to adjust the reserved address range of an allocated resource
to reserve
.Fa start
through
.Fa end .
It can be used to grow or shrink one or both ends of the resource range.
The current implementation does not support entirely relocating the resource
and will fail with
.Er EINVAL
if the new resource range does not overlap the old resource range.
If either end of the resource range grows and the new resource range would
conflict with another allocated resource,
the function will fail with
.Er EBUSY .
The
.Fn rman_adjust_resource
function does not support adjusting the resource range for shared resources
and will fail such attempts with
.Er EINVAL .
Upon success,
the resource
.Fa r
will have a start address of
.Fa start
and an end address of
.Fa end
and the function will return zero.
Note that none of the constraints of the original allocation request such
as alignment or boundary restrictions are checked by
.Fn rman_adjust_resource .
It is the caller's responsibility to enforce any such requirements.
.Pp
The
.Fn rman_release_resource
function releases the reserved resource
.Fa r .
It may attempt to merge adjacent free resources.
.Pp
The
.Fn rman_activate_resource
function marks a resource as active, by setting the
.Dv RF_ACTIVE
flag.
If this is a time shared resource, and the caller has not yet acquired
the resource, the function returns
.Er EBUSY .
.Pp
The
.Fn rman_deactivate_resource
function marks a resource
.Fa r
as inactive, by clearing the
.Dv RF_ACTIVE
flag.
If other consumers are waiting for this range, it will wakeup their threads.
.Pp
The
.Fn rman_get_start ,
.Fn rman_get_end ,
.Fn rman_get_size ,
and
.Fn rman_get_flags
functions return the bounds, size and flags of the previously reserved
resource
.Fa r .
.Pp
The
.Fn rman_set_bustag
function associates a
.Vt bus_space_tag_t
.Fa t
with the resource
.Fa r .
The
.Fn rman_get_bustag
function is used to retrieve this tag once set.
.Pp
The
.Fn rman_set_bushandle
function associates a
.Vt bus_space_handle_t
.Fa h
with the resource
.Fa r .
The
.Fn rman_get_bushandle
function is used to retrieve this handle once set.
.Pp
The
.Fn rman_set_virtual
function is used to associate a kernel virtual address with a resource
.Fa r .
The
.Fn rman_get_virtual
function can be used to retrieve the KVA once set.
.Pp
The
.Fn rman_set_mapping
function is used to associate a resource mapping with a resource
.Fa r .
The mapping must cover the entire resource.
Setting a mapping sets the associated
.Xr bus_space 9
handle and tag for
.Fa r
as well as the kernel virtual address if the mapping contains one.
These individual values can be retrieved via
.Fn rman_get_bushandle ,
.Fn rman_get_bustag ,
and
.Fn rman_get_virtual .
.Pp
The
.Fn rman_get_mapping
function can be used to retrieve the associated resource mapping once set.
.Pp
The
.Fn rman_set_rid
function associates a resource identifier with a resource
.Fa r .
The
.Fn rman_get_rid
function retrieves this RID.
.Pp
The
.Fn rman_get_device
function returns a pointer to the device which reserved the resource
.Fa r .
.Sh SEE ALSO
.Xr bus_activate_resource 9 ,
.Xr bus_adjust_resource 9 ,
.Xr bus_alloc_resource 9 ,
.Xr bus_map_resource 9 ,
.Xr bus_release_resource 9 ,
.Xr bus_space 9 ,
.Xr bus_set_resource 9 ,
.Xr mutex 9
.Sh AUTHORS
This manual page was written by
.An Bruce M Simpson Aq Mt bms@spc.org .
Reference in New Issue View Git Blame Copy Permalink