1 .\" Copyright (c) 2002, 2003 Hiten M. Pandya.
2 .\" All rights reserved.
4 .\" Redistribution and use in source and binary forms, with or without
5 .\" modification, are permitted provided that the following conditions
7 .\" 1. Redistributions of source code must retain the above copyright
8 .\" notice, this list of conditions, and the following disclaimer,
9 .\" without modification, immediately at the beginning of the file.
10 .\" 2. The name of the author may not be used to endorse or promote products
11 .\" derived from this software without specific prior written permission.
13 .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
14 .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
15 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
16 .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR, CONTRIBUTORS OR THE
17 .\" VOICES IN HITEN PANDYA'S HEAD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
18 .\" SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
19 .\" TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
20 .\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
21 .\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
22 .\" NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
23 .\" SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
25 .\" Copyright (c) 1996, 1997, 1998, 2001 The NetBSD Foundation, Inc.
26 .\" All rights reserved.
28 .\" This code is derived from software contributed to The NetBSD Foundation
29 .\" by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
30 .\" NASA Ames Research Center.
32 .\" Redistribution and use in source and binary forms, with or without
33 .\" modification, are permitted provided that the following conditions
35 .\" 1. Redistributions of source code must retain the above copyright
36 .\" notice, this list of conditions and the following disclaimer.
37 .\" 2. Redistributions in binary form must reproduce the above copyright
38 .\" notice, this list of conditions and the following disclaimer in the
39 .\" documentation and/or other materials provided with the distribution.
41 .\" THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
42 .\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
43 .\" TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
44 .\" PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
45 .\" BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
46 .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
47 .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
48 .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
49 .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
50 .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
51 .\" POSSIBILITY OF SUCH DAMAGE.
54 .\" $NetBSD: bus_dma.9,v 1.25 2002/10/14 13:43:16 wiz Exp $
61 .Nm bus_dma_tag_create ,
62 .Nm bus_dma_tag_destroy ,
63 .Nm bus_dmamap_create ,
64 .Nm bus_dmamap_destroy ,
66 .Nm bus_dmamap_load_bio ,
67 .Nm bus_dmamap_load_ccb ,
68 .Nm bus_dmamap_load_mbuf ,
69 .Nm bus_dmamap_load_mbuf_sg ,
70 .Nm bus_dmamap_load_uio ,
71 .Nm bus_dmamap_unload ,
73 .Nm bus_dmamem_alloc ,
75 .Nd Bus and Machine Independent DMA Mapping Interface
79 .Fn bus_dma_tag_create "bus_dma_tag_t parent" "bus_size_t alignment" \
80 "bus_addr_t boundary" "bus_addr_t lowaddr" "bus_addr_t highaddr" \
81 "bus_dma_filter_t *filtfunc" "void *filtfuncarg" "bus_size_t maxsize" \
82 "int nsegments" "bus_size_t maxsegsz" "int flags" "bus_dma_lock_t *lockfunc" \
83 "void *lockfuncarg" "bus_dma_tag_t *dmat"
85 .Fn bus_dma_tag_destroy "bus_dma_tag_t dmat"
87 .Fn bus_dmamap_create "bus_dma_tag_t dmat" "int flags" "bus_dmamap_t *mapp"
89 .Fn bus_dmamap_destroy "bus_dma_tag_t dmat" "bus_dmamap_t map"
91 .Fn bus_dmamap_load "bus_dma_tag_t dmat" "bus_dmamap_t map" "void *buf" \
92 "bus_size_t buflen" "bus_dmamap_callback_t *callback" "void *callback_arg" \
95 .Fn bus_dmamap_load_bio "bus_dma_tag_t dmat" "bus_dmamap_t map" \
96 "struct bio *bio" "bus_dmamap_callback_t *callback" "void *callback_arg" \
99 .Fn bus_dmamap_load_ccb "bus_dma_tag_t dmat" "bus_dmamap_t map" \
100 "union ccb *ccb" "bus_dmamap_callback_t *callback" "void *callback_arg" \
103 .Fn bus_dmamap_load_mbuf "bus_dma_tag_t dmat" "bus_dmamap_t map" \
104 "struct mbuf *mbuf" "bus_dmamap_callback2_t *callback" "void *callback_arg" \
107 .Fn bus_dmamap_load_mbuf_sg "bus_dma_tag_t dmat" "bus_dmamap_t map" \
108 "struct mbuf *mbuf" "bus_dma_segment_t *segs" "int *nsegs" "int flags"
110 .Fn bus_dmamap_load_uio "bus_dma_tag_t dmat" "bus_dmamap_t map" \
111 "struct uio *uio" "bus_dmamap_callback2_t *callback" "void *callback_arg" \
114 .Fn bus_dmamap_unload "bus_dma_tag_t dmat" "bus_dmamap_t map"
116 .Fn bus_dmamap_sync "bus_dma_tag_t dmat" "bus_dmamap_t map" \
119 .Fn bus_dmamem_alloc "bus_dma_tag_t dmat" "void **vaddr" \
120 "int flags" "bus_dmamap_t *mapp"
122 .Fn bus_dmamem_free "bus_dma_tag_t dmat" "void *vaddr" \
125 Direct Memory Access (DMA) is a method of transferring data
126 without involving the CPU, thus providing higher performance.
127 A DMA transaction can be achieved between device to memory,
128 device to device, or memory to memory.
132 API is a bus, device, and machine-independent (MI) interface to
134 It provides the client with flexibility and simplicity by
135 abstracting machine dependent issues like setting up
136 DMA mappings, handling cache issues, bus specific features
138 .Sh STRUCTURES AND TYPES
139 .Bl -tag -width indent
141 A machine-dependent (MD) opaque type that describes the
142 characteristics of DMA transactions.
143 DMA tags are organized into a hierarchy, with each child
144 tag inheriting the restrictions of its parent.
145 This allows all devices along the path of DMA transactions
146 to contribute to the constraints of those transactions.
147 .It Vt bus_dma_filter_t
148 Client specified address filter having the format:
149 .Bl -tag -width indent
151 .Fn "client_filter" "void *filtarg" "bus_addr_t testaddr"
154 Address filters can be specified during tag creation to allow
155 for devices whose DMA address restrictions cannot be specified
159 argument is specified by the client during tag creation to be passed to all
160 invocations of the callback.
163 argument contains a potential starting address of a DMA mapping.
164 The filter function operates on the set of addresses from
167 .Ql trunc_page(testaddr) + PAGE_SIZE - 1 ,
169 The filter function should return zero if any mapping in this range
170 can be accommodated by the device and non-zero otherwise.
171 .It Vt bus_dma_segment_t
172 A machine-dependent type that describes individual
174 It contains the following fields:
182 field contains the device visible address of the DMA segment, and
184 contains the length of the DMA segment.
185 Although the DMA segments returned by a mapping call will adhere to
186 all restrictions necessary for a successful DMA operation, some conversion
187 (e.g.\& a conversion from host byte order to the device's byte order) is
188 almost always required when presenting segment information to the device.
190 A machine-dependent opaque type describing an individual mapping.
191 One map is used for each memory allocation that will be loaded.
192 Maps can be reused once they have been unloaded.
193 Multiple maps can be associated with one DMA tag.
194 While the value of the map may evaluate to
196 on some platforms under certain conditions,
197 it should never be assumed that it will be
200 .It Vt bus_dmamap_callback_t
201 Client specified callback for receiving mapping information resulting from
205 .Fn bus_dmamap_load ,
206 .Fn bus_dmamap_load_bio
208 .Fn bus_dmamap_load_ccb .
209 Callbacks are of the format:
210 .Bl -tag -width indent
212 .Fn "client_callback" "void *callback_arg" "bus_dma_segment_t *segs" \
213 "int nseg" "int error"
218 is the callback argument passed to dmamap load functions.
223 arguments describe an array of
224 .Vt bus_dma_segment_t
225 structures that represent the mapping.
226 This array is only valid within the scope of the callback function.
227 The success or failure of the mapping is indicated by the
230 More information on the use of callbacks can be found in the
231 description of the individual dmamap load functions.
232 .It Vt bus_dmamap_callback2_t
233 Client specified callback for receiving mapping information resulting from
237 .Fn bus_dmamap_load_uio
239 .Fn bus_dmamap_load_mbuf .
241 Callback2s are of the format:
242 .Bl -tag -width indent
244 .Fn "client_callback2" "void *callback_arg" "bus_dma_segment_t *segs" \
245 "int nseg" "bus_size_t mapsize" "int error"
248 Callback2's behavior is the same as
249 .Vt bus_dmamap_callback_t
250 with the addition that the length of the data mapped is provided via
252 .It Vt bus_dmasync_op_t
253 Memory synchronization operation specifier.
254 Bus DMA requires explicit synchronization of memory with its device
255 visible mapping in order to guarantee memory coherency.
258 allows the type of DMA operation that will be or has been performed
259 to be communicated to the system so that the correct coherency measures
261 The operations are represented as bitfield flags that can be combined together,
262 though it only makes sense to combine PRE flags or POST flags, not both.
265 description below for more details on how to use these operations.
267 All operations specified below are performed from the host memory point of view,
268 where a read implies data coming from the device to the host memory, and a write
269 implies data going from the host memory to the device.
270 Alternatively, the operations can be thought of in terms of driver operations,
271 where reading a network packet or storage sector corresponds to a read operation
274 .Bl -tag -width ".Dv BUS_DMASYNC_POSTWRITE"
275 .It Dv BUS_DMASYNC_PREREAD
276 Perform any synchronization required prior to an update of host memory by the
278 .It Dv BUS_DMASYNC_PREWRITE
279 Perform any synchronization required after an update of host memory by the CPU
280 and prior to device access to host memory.
281 .It Dv BUS_DMASYNC_POSTREAD
282 Perform any synchronization required after an update of host memory by the
283 device and prior to CPU access to host memory.
284 .It Dv BUS_DMASYNC_POSTWRITE
285 Perform any synchronization required after device access to host memory.
287 .It Vt bus_dma_lock_t
288 Client specified lock/mutex manipulation method.
289 This will be called from
290 within busdma whenever a client lock needs to be manipulated.
291 In its current form, the function will be called immediately before
292 the callback for a DMA load operation that has been deferred with
294 and immediately after with
296 If the load operation does not need to be deferred, then it
297 will not be called since the function loading the map should
298 be holding the appropriate locks.
299 This method is of the format:
300 .Bl -tag -width indent
302 .Fn "lockfunc" "void *lockfunc_arg" "bus_dma_lock_op_t op"
307 argument is specified by the client during tag creation to be passed to all
308 invocations of the callback.
311 argument specifies the lock operation to perform.
315 implementations are provided for convenience.
316 .Fn busdma_lock_mutex
317 performs standard mutex operations on the sleep mutex provided via
320 will generate a system panic if it is called.
321 It is substituted into the tag when
326 .Fn bus_dma_tag_create
327 and is useful for tags that should not be used with deferred load operations.
328 .It Vt bus_dma_lock_op_t
329 Operations to be performed by the client-specified
331 .Bl -tag -width ".Dv BUS_DMA_UNLOCK"
333 Acquires and/or locks the client locking primitive.
334 .It Dv BUS_DMA_UNLOCK
335 Releases and/or unlocks the client locking primitive.
339 .Bl -tag -width indent
340 .It Fn bus_dma_tag_create "parent" "alignment" "boundary" "lowaddr" \
341 "highaddr" "*filtfunc" "*filtfuncarg" "maxsize" "nsegments" "maxsegsz" \
342 "flags" "lockfunc" "lockfuncarg" "*dmat"
343 Allocates a device specific DMA tag, and initializes it according to
344 the arguments provided:
345 .Bl -tag -width ".Fa filtfuncarg"
347 Indicates restrictions between the parent bridge, CPU memory, and the
349 Each device must use a master parent tag by calling
350 .Fn bus_get_dma_tag .
352 Alignment constraint, in bytes, of any mappings created using this tag.
353 The alignment must be a power of 2.
354 Hardware that can DMA starting at any address would specify
357 Hardware requiring DMA transfers to start on a multiple of 4K
361 Boundary constraint, in bytes, of the target DMA memory region.
362 The boundary indicates the set of addresses, all multiples of the
363 boundary argument, that cannot be crossed by a single
364 .Vt bus_dma_segment_t .
365 The boundary must be a power of 2 and must be no smaller than the
366 maximum segment size.
368 indicates that there are no boundary restrictions.
369 .It Fa lowaddr , highaddr
370 Bounds of the window of bus address space that
372 be directly accessed by the device.
373 The window contains all addresses greater than
375 and less than or equal to
377 For example, a device incapable of DMA above 4GB, would specify a
380 .Dv BUS_SPACE_MAXADDR
384 .Dv BUS_SPACE_MAXADDR_32BIT .
385 Similarly a device that can only perform DMA to addresses below
389 .Dv BUS_SPACE_MAXADDR
393 .Dv BUS_SPACE_MAXADDR_24BIT .
394 Some implementations requires that some region of device visible
395 address space, overlapping available host memory, be outside the
399 is used to bounce requests that would otherwise conflict with
400 the exclusion window.
402 Optional filter function (may be
404 to be called for any attempt to
405 map memory into the window described by
409 A filter function is only required when the single window described
414 cannot adequately describe the constraints of the device.
415 The filter function will be called for every machine page
416 that overlaps the exclusion window.
418 Argument passed to all calls to the filter function for this tag.
422 Maximum size, in bytes, of the sum of all segment lengths in a given
423 DMA mapping associated with this tag.
425 Number of discontinuities (scatter/gather segments) allowed
426 in a DMA mapped region.
427 If there is no restriction,
428 .Dv BUS_SPACE_UNRESTRICTED
431 Maximum size, in bytes, of a segment in any DMA mapped region associated
436 .Bl -tag -width ".Dv BUS_DMA_ALLOCNOW"
437 .It Dv BUS_DMA_ALLOCNOW
438 Pre-allocate enough resources to handle at least one map load operation on
440 If sufficient resources are not available,
443 This should not be used for tags that only describe buffers that will be
445 .Fn bus_dmamem_alloc .
446 Also, due to resource sharing with other tags, this flag does not guarantee
447 that resources will be allocated or reserved exclusively for this tag.
448 It should be treated only as a minor optimization.
449 .It Dv BUS_DMA_COHERENT
450 Indicate that the DMA engine and CPU are cache-coherent.
451 Cached memory may be used to back allocations created by
452 .Fn bus_dmamem_alloc .
454 .Fn bus_dma_tag_create ,
457 flag is currently implemented on arm64.
460 Optional lock manipulation function (may be
462 to be called when busdma
463 needs to manipulate a lock on behalf of the client.
470 Optional argument to be passed to the function specified by
473 Pointer to a bus_dma_tag_t where the resulting DMA tag will
479 if sufficient memory is not available for tag creation
480 or allocating mapping resources.
481 .It Fn bus_dma_tag_destroy "dmat"
482 Deallocate the DMA tag
485 .Fn bus_dma_tag_create .
489 if any DMA maps remain associated with
494 .It Fn bus_dmamap_create "dmat" "flags" "*mapp"
495 Allocates and initializes a DMA map.
496 Arguments are as follows:
497 .Bl -tag -width ".Fa nsegments"
502 .Bl -tag -width ".Dv BUS_DMA_COHERENT"
503 .It Dv BUS_DMA_COHERENT
504 Attempt to map the memory loaded with this map such that cache sync
505 operations are as cheap as possible.
506 This flag is typically set on maps when the memory loaded with these will
507 be accessed by both a CPU and a DMA engine, frequently such as control data
508 and as opposed to streamable data such as receive and transmit buffers.
509 Use of this flag does not remove the requirement of using
510 .Fn bus_dmamap_sync ,
511 but it may reduce the cost of performing these operations.
513 .Fn bus_dmamap_create ,
516 flag is currently implemented on sparc64.
521 where the resulting DMA map will be stored.
526 if sufficient memory is not available for creating the
527 map or allocating mapping resources.
528 .It Fn bus_dmamap_destroy "dmat" "map"
529 Frees all resources associated with a given DMA map.
530 Arguments are as follows:
531 .Bl -tag -width ".Fa dmat"
533 DMA tag used to allocate
536 The DMA map to destroy.
541 if a mapping is still active for
543 .It Fn bus_dmamap_load "dmat" "map" "buf" "buflen" "*callback" \
544 "callback_arg" "flags"
545 Creates a mapping in device visible address space of
549 associated with the DMA map
551 This call will always return immediately and will not block for any reason.
552 Arguments are as follows:
553 .Bl -tag -width ".Fa buflen"
555 DMA tag used to allocate
558 A DMA map without a currently active mapping.
560 A kernel virtual address pointer to a contiguous (in KVA) buffer, to be
561 mapped into device visible address space.
563 The size of the buffer.
564 .It Fa callback Fa callback_arg
565 The callback function, and its argument.
566 This function is called once sufficient mapping resources are available for
568 If resources are temporarily unavailable, this function will be deferred until
569 later, but the load operation will still return immediately to the caller.
570 Thus, callers should not assume that the callback will be called before the
571 load returns, and code should be structured appropriately to handle this.
572 See below for specific flags and error codes that control this behavior.
575 .Bl -tag -width ".Dv BUS_DMA_NOWAIT"
576 .It Dv BUS_DMA_NOWAIT
577 The load should not be deferred in case of insufficient mapping resources,
578 and instead should return immediately with an appropriate error.
579 .It Dv BUS_DMA_NOCACHE
580 The generated transactions to and from the virtual page are non-cacheable.
582 .Fn bus_dmamap_load ,
585 flag is currently implemented on sparc64.
589 Return values to the caller are as follows:
590 .Bl -tag -width ".Er EINPROGRESS"
592 The callback has been called and completed.
593 The status of the mapping has been delivered to the callback.
595 The mapping has been deferred for lack of resources.
596 The callback will be called as soon as resources are available.
597 Callbacks are serviced in FIFO order.
599 Note that subsequent load operations for the same tag that do not require
600 extra resources will still succeed.
601 This may result in out-of-order processing of requests.
602 If the caller requires the order of requests to be preserved,
603 then the caller is required to stall subsequent requests until a pending
604 request's callback is invoked.
606 The load request has failed due to insufficient resources, and the caller
607 specifically used the
611 The load request was invalid.
612 The callback has been called and has been provided the same error.
613 This error value may indicate that
623 argument used to create the dma tag
627 When the callback is called, it is presented with an error value
628 indicating the disposition of the mapping.
629 Error may be one of the following:
630 .Bl -tag -width ".Er EINPROGRESS"
632 The mapping was successful and the
634 callback argument contains an array of
635 .Vt bus_dma_segment_t
636 elements describing the mapping.
637 This array is only valid during the scope of the callback function.
639 A mapping could not be achieved within the segment constraints provided
640 in the tag even though the requested allocation size was less than maxsize.
642 .It Fn bus_dmamap_load_bio "dmat" "map" "bio" "callback" "callback_arg" "flags"
643 This is a variation of
645 which maps buffers pointed to by
649 may point to either a mapped or unmapped buffer.
650 .It Fn bus_dmamap_load_ccb "dmat" "map" "ccb" "callback" "callback_arg" "flags"
651 This is a variation of
653 which maps data pointed to by
658 may be any of the following types:
659 .Bl -tag -width ".Er CAM_DATA_SG_PADDR"
661 The data is a single KVA buffer.
663 The data is a single bus address range.
665 The data is a scatter/gather list of KVA buffers.
666 .It CAM_DATA_SG_PADDR
667 The data is a scatter/gather list of bus address ranges.
669 The data is contained in a
674 .Fn bus_dmamap_load_ccb
675 supports the following CCB XPT function codes:
677 .Bl -item -offset indent -compact
685 .It Fn bus_dmamap_load_mbuf "dmat" "map" "mbuf" "callback2" "callback_arg" \
687 This is a variation of
689 which maps mbuf chains
693 argument is also passed to the callback routine, which
694 contains the mbuf chain's packet header length.
697 flag is implied, thus no callback deferral will happen.
699 Mbuf chains are assumed to be in kernel virtual address space.
701 Beside the error values listed for
702 .Fn bus_dmamap_load ,
704 will be returned if the size of the mbuf chain exceeds the maximum limit of the
706 .It Fn bus_dmamap_load_mbuf_sg "dmat" "map" "mbuf" "segs" "nsegs" "flags"
708 .Fn bus_dmamap_load_mbuf
709 except that it returns immediately without calling a callback function.
710 It is provided for efficiency.
711 The scatter/gather segment array
713 is provided by the caller and filled in directly by the function.
716 argument is returned with the number of segments filled in.
717 Returns the same errors as
718 .Fn bus_dmamap_load_mbuf .
719 .It Fn bus_dmamap_load_uio "dmat" "map" "uio" "callback2" "callback_arg" "flags"
720 This is a variation of
722 which maps buffers pointed to by
727 argument is also passed to the callback routine, which contains the size of
733 flag is implied, thus no callback deferral will happen.
734 Returns the same errors as
735 .Fn bus_dmamap_load .
741 then it is assumed that the buffer,
744 .Fa "uio->uio_td->td_proc" Ns 's
746 User space memory must be in-core and wired prior to attempting a map
748 Pages may be locked using
750 .It Fn bus_dmamap_unload "dmat" "map"
752 Arguments are as follows:
753 .Bl -tag -width ".Fa dmam"
755 DMA tag used to allocate
758 The DMA map that is to be unloaded.
761 .Fn bus_dmamap_unload
762 will not perform any implicit synchronization of DMA buffers.
763 This must be done explicitly by a call to
765 prior to unloading the map.
766 .It Fn bus_dmamap_sync "dmat" "map" "op"
767 Performs synchronization of a device visible mapping with the CPU visible
768 memory referenced by that mapping.
769 Arguments are as follows:
770 .Bl -tag -width ".Fa dmat"
772 DMA tag used to allocate
775 The DMA mapping to be synchronized.
777 Type of synchronization operation to perform.
778 See the definition of
780 for a description of the acceptable values for
787 is the method used to ensure that CPU's and device's direct
788 memory access (DMA) to shared
790 For example, the CPU might be used to set up the contents of a buffer
791 that is to be made available to a device.
792 To ensure that the data are visible via the device's mapping of that
793 memory, the buffer must be loaded and a DMA sync operation of
794 .Dv BUS_DMASYNC_PREWRITE
795 must be performed after the CPU has updated the buffer and before the device
797 If the CPU modifies this buffer again later, another
798 .Dv BUS_DMASYNC_PREWRITE
799 sync operation must be performed before an additional device
801 Conversely, suppose a device updates memory that is to be read by a CPU.
802 In this case, the buffer must be loaded, and a DMA sync operation of
803 .Dv BUS_DMASYNC_PREREAD
804 must be performed before the device access is initiated.
805 The CPU will only be able to see the results of this memory update
806 once the DMA operation has completed and a
807 .Dv BUS_DMASYNC_POSTREAD
808 sync operation has been performed.
810 If read and write operations are not preceded and followed by the
811 appropriate synchronization operations, behavior is undefined.
812 .It Fn bus_dmamem_alloc "dmat" "**vaddr" "flags" "*mapp"
813 Allocates memory that is mapped into KVA at the address returned
816 and that is permanently loaded into the newly created
820 Arguments are as follows:
821 .Bl -tag -width ".Fa alignment"
823 DMA tag describing the constraints of the DMA mapping.
825 Pointer to a pointer that will hold the returned KVA mapping of
826 the allocated region.
828 Flags are defined as follows:
829 .Bl -tag -width ".Dv BUS_DMA_NOWAIT"
830 .It Dv BUS_DMA_WAITOK
831 The routine can safely wait (sleep) for resources.
832 .It Dv BUS_DMA_NOWAIT
833 The routine is not allowed to wait for resources.
834 If resources are not available,
837 .It Dv BUS_DMA_COHERENT
838 Attempt to map this memory in a coherent fashion.
840 .Fn bus_dmamap_create
841 above for a description of this flag.
843 .Fn bus_dmamem_alloc ,
846 flag is currently implemented on arm, arm64 and sparc64.
848 Causes the allocated memory to be set to all zeros.
849 .It Dv BUS_DMA_NOCACHE
850 The allocated memory will not be cached in the processor caches.
851 All memory accesses appear on the bus and are executed
854 .Fn bus_dmamem_alloc ,
857 flag is currently implemented on amd64 and i386 where it results in the
858 Strong Uncacheable PAT to be set for the allocated virtual address range.
863 where the resulting DMA map will be stored.
866 The size of memory to be allocated is
868 as specified in the call to
869 .Fn bus_dma_tag_create
873 The current implementation of
875 will allocate all requests as a single segment.
877 An initial load operation is required to obtain the bus address of the allocated
878 memory, and an unload operation is required before freeing the memory, as
880 .Fn bus_dmamem_free .
881 Maps are automatically handled by this function and should not be explicitly
882 allocated or destroyed.
884 Although an explicit load is not required for each access to the memory
885 referenced by the returned map, the synchronization requirements
888 section still apply and should be used to achieve portability on architectures
889 without coherent buses.
893 if sufficient memory is not available for completing
895 .It Fn bus_dmamem_free "dmat" "*vaddr" "map"
896 Frees memory previously allocated by
897 .Fn bus_dmamem_alloc .
900 Arguments are as follows:
901 .Bl -tag -width ".Fa vaddr"
905 Kernel virtual address of the memory.
907 DMA map to be invalidated.
911 Behavior is undefined if invalid arguments are passed to
912 any of the above functions.
913 If sufficient resources cannot be allocated for a given
918 routines that are not of type
920 will return 0 on success or an error
921 code on failure as discussed above.
925 routines will succeed if provided with valid arguments.
927 Two locking protocols are used by
929 The first is a private global lock that is used to synchronize access to the
930 bounce buffer pool on the architectures that make use of them.
931 This lock is strictly a leaf lock that is only used internally to
933 and is not exposed to clients of the API.
935 The second protocol involves protecting various resources stored in the tag.
938 operations are done through requests from the driver that created the tag,
939 the most efficient way to protect the tag resources is through the lock that
943 acts on its own without being called by the driver, the lock primitive
944 specified in the tag is acquired and released automatically.
945 An example of this is when the
947 callback function is called from a deferred context instead of the driver
949 This means that certain
951 functions must always be called with the same lock held that is specified in the
953 These functions include:
955 .Bl -item -offset indent -compact
959 .Fn bus_dmamap_load_bio
961 .Fn bus_dmamap_load_ccb
963 .Fn bus_dmamap_load_mbuf
965 .Fn bus_dmamap_load_mbuf_sg
967 .Fn bus_dmamap_load_uio
969 .Fn bus_dmamap_unload
974 There is one exception to this rule.
975 It is common practice to call some of these functions during driver start-up
976 without any locks held.
977 So long as there is a guarantee of no possible concurrent use of the tag by
978 different threads during this operation, it is safe to not hold a lock for
983 operations should not be called with the driver lock held, either because
984 they are already protected by an internal lock, or because they might sleep
985 due to memory or resource allocation.
986 The following functions must not be
987 called with any non-sleepable locks held:
989 .Bl -item -offset indent -compact
991 .Fn bus_dma_tag_create
993 .Fn bus_dmamap_create
998 All other functions do not have a locking protocol and can thus be
999 called with or without any system or driver locks held.
1008 .%A "Jason R. Thorpe"
1009 .%T "A Machine-Independent DMA Framework for NetBSD"
1010 .%J "Proceedings of the Summer 1998 USENIX Technical Conference"
1011 .%Q "USENIX Association"
1017 interface first appeared in
1022 API was adopted from
1024 for use in the CAM SCSI subsystem.
1025 The alterations to the original API were aimed to remove the need for
1027 .Vt bus_dma_segment_t
1028 array stored in each
1030 while allowing callers to queue up on scarce resources.
1034 interface was designed and implemented by
1036 of the Numerical Aerospace Simulation Facility, NASA Ames Research Center.
1037 Additional input on the
1039 design was provided by
1041 .An Chris Demetriou ,
1042 .An Charles Hannum ,
1045 .An Jonathan Stone ,
1053 benefits from the contributions of
1054 .An Justin T. Gibbs ,
1057 .An Matthew N. Dodd ,
1059 .An Maxime Henrion ,
1060 .An Jake Burkholder ,
1061 .An Takahashi Yoshihiro ,
1065 This manual page was written by
1068 .An Justin T. Gibbs .