blob: ef3a04fcad655f2efd9989f9b810314233c90bee [file] [log] [blame]
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -03001============================================
2Dynamic DMA mapping using the generic device
3============================================
Linus Torvalds1da177e2005-04-16 15:20:36 -07004
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -03005:Author: James E.J. Bottomley <James.Bottomley@HansenPartnership.com>
Linus Torvalds1da177e2005-04-16 15:20:36 -07006
7This document describes the DMA API. For a more gentle introduction
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -06008of the API (and actual examples), see Documentation/DMA-API-HOWTO.txt.
Linus Torvalds1da177e2005-04-16 15:20:36 -07009
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060010This API is split into two pieces. Part I describes the basic API.
11Part II describes extensions for supporting non-consistent memory
12machines. Unless you know that your driver absolutely has to support
13non-consistent platforms (this is usually only legacy platforms) you
14should only use the API described in part I.
Linus Torvalds1da177e2005-04-16 15:20:36 -070015
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -030016Part I - dma_API
17----------------
Linus Torvalds1da177e2005-04-16 15:20:36 -070018
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -030019To get the dma_API, you must #include <linux/dma-mapping.h>. This
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060020provides dma_addr_t and the interfaces described below.
Linus Torvalds1da177e2005-04-16 15:20:36 -070021
Yinghai Lu3a9ad0b2015-05-27 17:23:51 -070022A dma_addr_t can hold any valid DMA address for the platform. It can be
23given to a device to use as a DMA source or target. A CPU cannot reference
24a dma_addr_t directly because there may be translation between its physical
25address space and the DMA address space.
Linus Torvalds1da177e2005-04-16 15:20:36 -070026
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060027Part Ia - Using large DMA-coherent buffers
Linus Torvalds1da177e2005-04-16 15:20:36 -070028------------------------------------------
29
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -030030::
31
32 void *
33 dma_alloc_coherent(struct device *dev, size_t size,
34 dma_addr_t *dma_handle, gfp_t flag)
Linus Torvalds1da177e2005-04-16 15:20:36 -070035
36Consistent memory is memory for which a write by either the device or
37the processor can immediately be read by the processor or device
David Brownell21440d32006-04-01 10:21:52 -080038without having to worry about caching effects. (You may however need
39to make sure to flush the processor's write buffers before telling
40devices to read that memory.)
Linus Torvalds1da177e2005-04-16 15:20:36 -070041
42This routine allocates a region of <size> bytes of consistent memory.
Linus Torvalds1da177e2005-04-16 15:20:36 -070043
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060044It returns a pointer to the allocated region (in the processor's virtual
Linus Torvalds1da177e2005-04-16 15:20:36 -070045address space) or NULL if the allocation failed.
46
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060047It also returns a <dma_handle> which may be cast to an unsigned integer the
Yinghai Lu3a9ad0b2015-05-27 17:23:51 -070048same width as the bus and given to the device as the DMA address base of
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060049the region.
50
Linus Torvalds1da177e2005-04-16 15:20:36 -070051Note: consistent memory can be expensive on some platforms, and the
52minimum allocation length may be as big as a page, so you should
53consolidate your requests for consistent memory as much as possible.
54The simplest way to do that is to use the dma_pool calls (see below).
55
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060056The flag parameter (dma_alloc_coherent() only) allows the caller to
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -030057specify the ``GFP_`` flags (see kmalloc()) for the allocation (the
Randy Dunlapa12e2c62007-07-31 00:38:17 -070058implementation may choose to ignore flags that affect the location of
FUJITA Tomonorif5a69f42010-03-10 15:23:43 -080059the returned memory, like GFP_DMA).
Linus Torvalds1da177e2005-04-16 15:20:36 -070060
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -030061::
62
63 void *
64 dma_zalloc_coherent(struct device *dev, size_t size,
65 dma_addr_t *dma_handle, gfp_t flag)
Andrew Morton842fa692011-11-02 13:39:33 -070066
67Wraps dma_alloc_coherent() and also zeroes the returned memory if the
68allocation attempt succeeded.
69
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -030070::
71
72 void
73 dma_free_coherent(struct device *dev, size_t size, void *cpu_addr,
74 dma_addr_t dma_handle)
Linus Torvalds1da177e2005-04-16 15:20:36 -070075
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060076Free a region of consistent memory you previously allocated. dev,
77size and dma_handle must all be the same as those passed into
78dma_alloc_coherent(). cpu_addr must be the virtual address returned by
79the dma_alloc_coherent().
Linus Torvalds1da177e2005-04-16 15:20:36 -070080
David Brownellaa248862007-08-10 13:10:27 -070081Note that unlike their sibling allocation calls, these routines
82may only be called with IRQs enabled.
83
Linus Torvalds1da177e2005-04-16 15:20:36 -070084
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060085Part Ib - Using small DMA-coherent buffers
Linus Torvalds1da177e2005-04-16 15:20:36 -070086------------------------------------------
87
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -030088To get this part of the dma_API, you must #include <linux/dmapool.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070089
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060090Many drivers need lots of small DMA-coherent memory regions for DMA
Linus Torvalds1da177e2005-04-16 15:20:36 -070091descriptors or I/O buffers. Rather than allocating in units of a page
92or more using dma_alloc_coherent(), you can use DMA pools. These work
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -060093much like a struct kmem_cache, except that they use the DMA-coherent allocator,
Linus Torvalds1da177e2005-04-16 15:20:36 -070094not __get_free_pages(). Also, they understand common hardware constraints
Randy Dunlapa12e2c62007-07-31 00:38:17 -070095for alignment, like queue heads needing to be aligned on N-byte boundaries.
Linus Torvalds1da177e2005-04-16 15:20:36 -070096
97
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -030098::
99
Linus Torvalds1da177e2005-04-16 15:20:36 -0700100 struct dma_pool *
101 dma_pool_create(const char *name, struct device *dev,
102 size_t size, size_t align, size_t alloc);
103
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600104dma_pool_create() initializes a pool of DMA-coherent buffers
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105for use with a given device. It must be called in a context which
106can sleep.
107
Christoph Lametere18b8902006-12-06 20:33:20 -0800108The "name" is for diagnostics (like a struct kmem_cache name); dev and size
Linus Torvalds1da177e2005-04-16 15:20:36 -0700109are like what you'd pass to dma_alloc_coherent(). The device's hardware
110alignment requirement for this type of data is "align" (which is expressed
111in bytes, and must be a power of two). If your device has no boundary
112crossing restrictions, pass 0 for alloc; passing 4096 says memory allocated
113from this pool must not cross 4KByte boundaries.
114
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300115::
Linus Torvalds1da177e2005-04-16 15:20:36 -0700116
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300117 void *
118 dma_pool_zalloc(struct dma_pool *pool, gfp_t mem_flags,
119 dma_addr_t *handle)
Sean O. Stalleyad823622015-09-08 15:02:27 -0700120
121Wraps dma_pool_alloc() and also zeroes the returned memory if the
122allocation attempt succeeded.
123
124
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300125::
126
127 void *
128 dma_pool_alloc(struct dma_pool *pool, gfp_t gfp_flags,
129 dma_addr_t *dma_handle);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700130
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600131This allocates memory from the pool; the returned memory will meet the
132size and alignment requirements specified at creation time. Pass
133GFP_ATOMIC to prevent blocking, or if it's permitted (not
134in_interrupt, not holding SMP locks), pass GFP_KERNEL to allow
135blocking. Like dma_alloc_coherent(), this returns two values: an
Bjorn Helgaasf311a722014-05-20 16:56:27 -0600136address usable by the CPU, and the DMA address usable by the pool's
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600137device.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700138
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300139::
Linus Torvalds1da177e2005-04-16 15:20:36 -0700140
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300141 void
142 dma_pool_free(struct dma_pool *pool, void *vaddr,
143 dma_addr_t addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700144
Linus Torvalds1da177e2005-04-16 15:20:36 -0700145This puts memory back into the pool. The pool is what was passed to
Bjorn Helgaasf311a722014-05-20 16:56:27 -0600146dma_pool_alloc(); the CPU (vaddr) and DMA addresses are what
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147were returned when that routine allocated the memory being freed.
148
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300149::
Linus Torvalds1da177e2005-04-16 15:20:36 -0700150
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300151 void
152 dma_pool_destroy(struct dma_pool *pool);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700153
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600154dma_pool_destroy() frees the resources of the pool. It must be
Linus Torvalds1da177e2005-04-16 15:20:36 -0700155called in a context which can sleep. Make sure you've freed all allocated
156memory back to the pool before you destroy it.
157
158
159Part Ic - DMA addressing limitations
160------------------------------------
161
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300162::
163
164 int
165 dma_set_mask_and_coherent(struct device *dev, u64 mask)
Russell King4aa806b2013-06-26 13:49:44 +0100166
167Checks to see if the mask is possible and updates the device
168streaming and coherent DMA mask parameters if it is.
169
170Returns: 0 if successful and a negative error if not.
171
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300172::
173
174 int
175 dma_set_mask(struct device *dev, u64 mask)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700176
177Checks to see if the mask is possible and updates the device
178parameters if it is.
179
180Returns: 0 if successful and a negative error if not.
181
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300182::
183
184 int
185 dma_set_coherent_mask(struct device *dev, u64 mask)
FUJITA Tomonori6a1961f2010-03-10 15:23:39 -0800186
187Checks to see if the mask is possible and updates the device
188parameters if it is.
189
190Returns: 0 if successful and a negative error if not.
191
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300192::
193
194 u64
195 dma_get_required_mask(struct device *dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700196
John Keller175add12008-11-24 16:47:17 -0600197This API returns the mask that the platform requires to
198operate efficiently. Usually this means the returned mask
Linus Torvalds1da177e2005-04-16 15:20:36 -0700199is the minimum required to cover all of memory. Examining the
200required mask gives drivers with variable descriptor sizes the
201opportunity to use smaller descriptors as necessary.
202
203Requesting the required mask does not alter the current mask. If you
John Keller175add12008-11-24 16:47:17 -0600204wish to take advantage of it, you should issue a dma_set_mask()
205call to set the mask to the value returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700206
207
208Part Id - Streaming DMA mappings
209--------------------------------
210
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300211::
212
213 dma_addr_t
214 dma_map_single(struct device *dev, void *cpu_addr, size_t size,
215 enum dma_data_direction direction)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700216
217Maps a piece of processor virtual memory so it can be accessed by the
Yinghai Lu3a9ad0b2015-05-27 17:23:51 -0700218device and returns the DMA address of the memory.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600220The direction for both APIs may be converted freely by casting.
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300221However the dma_API uses a strongly typed enumerator for its
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222direction:
223
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300224======================= =============================================
FUJITA Tomonorif5a69f42010-03-10 15:23:43 -0800225DMA_NONE no direction (used for debugging)
226DMA_TO_DEVICE data is going from the memory to the device
227DMA_FROM_DEVICE data is coming from the device to the memory
228DMA_BIDIRECTIONAL direction isn't known
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300229======================= =============================================
Linus Torvalds1da177e2005-04-16 15:20:36 -0700230
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300231.. note::
Linus Torvalds1da177e2005-04-16 15:20:36 -0700232
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300233 Not all memory regions in a machine can be mapped by this API.
234 Further, contiguous kernel virtual space may not be contiguous as
235 physical memory. Since this API does not provide any scatter/gather
236 capability, it will fail if the user tries to map a non-physically
237 contiguous piece of memory. For this reason, memory to be mapped by
238 this API should be obtained from sources which guarantee it to be
239 physically contiguous (like kmalloc).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700240
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300241 Further, the DMA address of the memory must be within the
242 dma_mask of the device (the dma_mask is a bit mask of the
243 addressable region for the device, i.e., if the DMA address of
244 the memory ANDed with the dma_mask is still equal to the DMA
245 address, then the device can perform DMA to the memory). To
246 ensure that the memory allocated by kmalloc is within the dma_mask,
247 the driver may specify various platform-dependent flags to restrict
248 the DMA address range of the allocation (e.g., on x86, GFP_DMA
249 guarantees to be within the first 16MB of available DMA addresses,
250 as required by ISA devices).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700251
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300252 Note also that the above constraints on physical contiguity and
253 dma_mask may not apply if the platform has an IOMMU (a device which
254 maps an I/O DMA address to a physical memory address). However, to be
255 portable, device driver writers may *not* assume that such an IOMMU
256 exists.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700257
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300258.. warning::
Linus Torvalds1da177e2005-04-16 15:20:36 -0700259
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300260 Memory coherency operates at a granularity called the cache
261 line width. In order for memory mapped by this API to operate
262 correctly, the mapped region must begin exactly on a cache line
263 boundary and end exactly on one (to prevent two separately mapped
264 regions from sharing a single cache line). Since the cache line size
265 may not be known at compile time, the API will not enforce this
266 requirement. Therefore, it is recommended that driver writers who
267 don't take special care to determine the cache line size at run time
268 only map virtual regions that begin and end on page boundaries (which
269 are guaranteed also to be cache line boundaries).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700270
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300271 DMA_TO_DEVICE synchronisation must be done after the last modification
272 of the memory region by the software and before it is handed off to
273 the device. Once this primitive is used, memory covered by this
274 primitive should be treated as read-only by the device. If the device
275 may write to it at any point, it should be DMA_BIDIRECTIONAL (see
276 below).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700277
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300278 DMA_FROM_DEVICE synchronisation must be done before the driver
279 accesses data that may be changed by the device. This memory should
280 be treated as read-only by the driver. If the driver needs to write
281 to it at any point, it should be DMA_BIDIRECTIONAL (see below).
282
283 DMA_BIDIRECTIONAL requires special handling: it means that the driver
284 isn't sure if the memory was modified before being handed off to the
285 device and also isn't sure if the device will also modify it. Thus,
286 you must always sync bidirectional memory twice: once before the
287 memory is handed off to the device (to make sure all memory changes
288 are flushed from the processor) and once before the data may be
289 accessed after being used by the device (to make sure any processor
290 cache lines are updated with data that the device may have changed).
291
292::
293
294 void
295 dma_unmap_single(struct device *dev, dma_addr_t dma_addr, size_t size,
296 enum dma_data_direction direction)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700297
298Unmaps the region previously mapped. All the parameters passed in
299must be identical to those passed in (and returned) by the mapping
300API.
301
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300302::
303
304 dma_addr_t
305 dma_map_page(struct device *dev, struct page *page,
306 unsigned long offset, size_t size,
307 enum dma_data_direction direction)
308
309 void
310 dma_unmap_page(struct device *dev, dma_addr_t dma_address, size_t size,
311 enum dma_data_direction direction)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700312
313API for mapping and unmapping for pages. All the notes and warnings
314for the other mapping APIs apply here. Also, although the <offset>
315and <size> parameters are provided to do partial page mapping, it is
316recommended that you never use these unless you really know what the
317cache width is.
318
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300319::
Niklas Söderlund6f3d8792016-08-10 13:22:16 +0200320
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300321 dma_addr_t
322 dma_map_resource(struct device *dev, phys_addr_t phys_addr, size_t size,
323 enum dma_data_direction dir, unsigned long attrs)
324
325 void
326 dma_unmap_resource(struct device *dev, dma_addr_t addr, size_t size,
327 enum dma_data_direction dir, unsigned long attrs)
Niklas Söderlund6f3d8792016-08-10 13:22:16 +0200328
329API for mapping and unmapping for MMIO resources. All the notes and
330warnings for the other mapping APIs apply here. The API should only be
331used to map device MMIO resources, mapping of RAM is not permitted.
332
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300333::
334
335 int
336 dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700337
Niklas Söderlund6f3d8792016-08-10 13:22:16 +0200338In some circumstances dma_map_single(), dma_map_page() and dma_map_resource()
339will fail to create a mapping. A driver can check for these errors by testing
340the returned DMA address with dma_mapping_error(). A non-zero return value
341means the mapping could not be created and the driver should take appropriate
342action (e.g. reduce current DMA mapping usage or delay and try again later).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700343
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300344::
345
David Brownell21440d32006-04-01 10:21:52 -0800346 int
347 dma_map_sg(struct device *dev, struct scatterlist *sg,
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300348 int nents, enum dma_data_direction direction)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700349
Yinghai Lu3a9ad0b2015-05-27 17:23:51 -0700350Returns: the number of DMA address segments mapped (this may be shorter
FUJITA Tomonori1d678f32008-12-01 13:14:01 -0800351than <nents> passed in if some elements of the scatter/gather list are
352physically or virtually adjacent and an IOMMU maps them with a single
353entry).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354
355Please note that the sg cannot be mapped again if it has been mapped once.
356The mapping process is allowed to destroy information in the sg.
357
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600358As with the other mapping interfaces, dma_map_sg() can fail. When it
Linus Torvalds1da177e2005-04-16 15:20:36 -0700359does, 0 is returned and a driver must take appropriate action. It is
360critical that the driver do something, in the case of a block driver
361aborting the request or even oopsing is better than doing nothing and
362corrupting the filesystem.
363
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300364With scatterlists, you use the resulting mapping like this::
David Brownell21440d32006-04-01 10:21:52 -0800365
366 int i, count = dma_map_sg(dev, sglist, nents, direction);
367 struct scatterlist *sg;
368
FUJITA Tomonori79eb0142008-09-18 09:35:28 -0700369 for_each_sg(sglist, sg, count, i) {
David Brownell21440d32006-04-01 10:21:52 -0800370 hw_address[i] = sg_dma_address(sg);
371 hw_len[i] = sg_dma_len(sg);
372 }
373
374where nents is the number of entries in the sglist.
375
376The implementation is free to merge several consecutive sglist entries
377into one (e.g. with an IOMMU, or if several pages just happen to be
378physically contiguous) and returns the actual number of sg entries it
379mapped them to. On failure 0, is returned.
380
381Then you should loop count times (note: this can be less than nents times)
382and use sg_dma_address() and sg_dma_len() macros where you previously
383accessed sg->address and sg->length as shown above.
384
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300385::
386
David Brownell21440d32006-04-01 10:21:52 -0800387 void
388 dma_unmap_sg(struct device *dev, struct scatterlist *sg,
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300389 int nents, enum dma_data_direction direction)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700390
Randy Dunlapa12e2c62007-07-31 00:38:17 -0700391Unmap the previously mapped scatter/gather list. All the parameters
Linus Torvalds1da177e2005-04-16 15:20:36 -0700392must be the same as those and passed in to the scatter/gather mapping
393API.
394
395Note: <nents> must be the number you passed in, *not* the number of
Yinghai Lu3a9ad0b2015-05-27 17:23:51 -0700396DMA address entries returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700397
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300398::
399
400 void
401 dma_sync_single_for_cpu(struct device *dev, dma_addr_t dma_handle,
402 size_t size,
403 enum dma_data_direction direction)
404
405 void
406 dma_sync_single_for_device(struct device *dev, dma_addr_t dma_handle,
407 size_t size,
408 enum dma_data_direction direction)
409
410 void
411 dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg,
412 int nents,
413 enum dma_data_direction direction)
414
415 void
416 dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg,
417 int nents,
418 enum dma_data_direction direction)
FUJITA Tomonori9705ef72010-03-10 15:23:17 -0800419
Bjorn Helgaasf311a722014-05-20 16:56:27 -0600420Synchronise a single contiguous or scatter/gather mapping for the CPU
FUJITA Tomonori9705ef72010-03-10 15:23:17 -0800421and device. With the sync_sg API, all the parameters must be the same
422as those passed into the single mapping API. With the sync_single API,
423you can use dma_handle and size parameters that aren't identical to
424those passed into the single mapping API to do a partial sync.
425
FUJITA Tomonori9705ef72010-03-10 15:23:17 -0800426
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300427.. note::
428
429 You must do this:
430
431 - Before reading values that have been written by DMA from the device
432 (use the DMA_FROM_DEVICE direction)
433 - After writing values that will be written to the device using DMA
434 (use the DMA_TO_DEVICE) direction
435 - before *and* after handing memory to the device if the memory is
436 DMA_BIDIRECTIONAL
FUJITA Tomonori9705ef72010-03-10 15:23:17 -0800437
438See also dma_map_single().
439
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300440::
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700441
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300442 dma_addr_t
443 dma_map_single_attrs(struct device *dev, void *cpu_addr, size_t size,
444 enum dma_data_direction dir,
445 unsigned long attrs)
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700446
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300447 void
448 dma_unmap_single_attrs(struct device *dev, dma_addr_t dma_addr,
449 size_t size, enum dma_data_direction dir,
450 unsigned long attrs)
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700451
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300452 int
453 dma_map_sg_attrs(struct device *dev, struct scatterlist *sgl,
454 int nents, enum dma_data_direction dir,
455 unsigned long attrs)
456
457 void
458 dma_unmap_sg_attrs(struct device *dev, struct scatterlist *sgl,
459 int nents, enum dma_data_direction dir,
460 unsigned long attrs)
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700461
462The four functions above are just like the counterpart functions
463without the _attrs suffixes, except that they pass an optional
Krzysztof Kozlowski00085f12016-08-03 13:46:00 -0700464dma_attrs.
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700465
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600466The interpretation of DMA attributes is architecture-specific, and
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700467each attribute should be documented in Documentation/DMA-attributes.txt.
468
Krzysztof Kozlowski00085f12016-08-03 13:46:00 -0700469If dma_attrs are 0, the semantics of each of these functions
470is identical to those of the corresponding function
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700471without the _attrs suffix. As a result dma_map_single_attrs()
472can generally replace dma_map_single(), etc.
473
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300474As an example of the use of the ``*_attrs`` functions, here's how
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700475you could pass an attribute DMA_ATTR_FOO when mapping memory
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300476for DMA::
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700477
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300478 #include <linux/dma-mapping.h>
479 /* DMA_ATTR_FOO should be defined in linux/dma-mapping.h and
480 * documented in Documentation/DMA-attributes.txt */
481 ...
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700482
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300483 unsigned long attr;
484 attr |= DMA_ATTR_FOO;
485 ....
486 n = dma_map_sg_attrs(dev, sg, nents, DMA_TO_DEVICE, attr);
487 ....
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700488
489Architectures that care about DMA_ATTR_FOO would check for its
490presence in their implementations of the mapping and unmapping
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300491routines, e.g.:::
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700492
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300493 void whizco_dma_map_sg_attrs(struct device *dev, dma_addr_t dma_addr,
494 size_t size, enum dma_data_direction dir,
495 unsigned long attrs)
496 {
497 ....
498 if (attrs & DMA_ATTR_FOO)
499 /* twizzle the frobnozzle */
500 ....
501 }
Arthur Kepnera75b0a22008-04-29 01:00:31 -0700502
Linus Torvalds1da177e2005-04-16 15:20:36 -0700503
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300504Part II - Advanced dma usage
505----------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -0700506
FUJITA Tomonorif5a69f42010-03-10 15:23:43 -0800507Warning: These pieces of the DMA API should not be used in the
508majority of cases, since they cater for unlikely corner cases that
509don't belong in usual drivers.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700510
511If you don't understand how cache line coherency works between a
512processor and an I/O device, you should not be using this part of the
513API at all.
514
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300515::
516
517 void *
Christoph Hellwig2fd523c2017-08-25 17:06:13 +0200518 dma_alloc_attrs(struct device *dev, size_t size, dma_addr_t *dma_handle,
519 gfp_t flag, unsigned long attrs)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700520
Christoph Hellwig2fd523c2017-08-25 17:06:13 +0200521Identical to dma_alloc_coherent() except that when the
522DMA_ATTR_NON_CONSISTENT flags is passed in the attrs argument, the
523platform will choose to return either consistent or non-consistent memory
524as it sees fit. By using this API, you are guaranteeing to the platform
525that you have all the correct and necessary sync points for this memory
526in the driver should it choose to return non-consistent memory.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700527
528Note: where the platform can return consistent memory, it will
529guarantee that the sync points become nops.
530
531Warning: Handling non-consistent memory is a real pain. You should
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600532only use this API if you positively know your driver will be
Linus Torvalds1da177e2005-04-16 15:20:36 -0700533required to work on one of the rare (usually non-PCI) architectures
534that simply cannot make consistent memory.
535
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300536::
537
538 void
Christoph Hellwig2fd523c2017-08-25 17:06:13 +0200539 dma_free_attrs(struct device *dev, size_t size, void *cpu_addr,
540 dma_addr_t dma_handle, unsigned long attrs)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700541
Christoph Hellwig2fd523c2017-08-25 17:06:13 +0200542Free memory allocated by the dma_alloc_attrs(). All parameters common
543parameters must identical to those otherwise passed to dma_fre_coherent,
544and the attrs argument must be identical to the attrs passed to
545dma_alloc_attrs().
Linus Torvalds1da177e2005-04-16 15:20:36 -0700546
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300547::
548
549 int
550 dma_get_cache_alignment(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700551
Randy Dunlapa12e2c62007-07-31 00:38:17 -0700552Returns the processor cache alignment. This is the absolute minimum
Linus Torvalds1da177e2005-04-16 15:20:36 -0700553alignment *and* width that you must observe when either mapping
554memory or doing partial flushes.
555
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300556.. note::
Linus Torvalds1da177e2005-04-16 15:20:36 -0700557
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300558 This API may return a number *larger* than the actual cache
559 line, but it will guarantee that one or more cache lines fit exactly
560 into the width returned by this call. It will also always be a power
561 of two for easy alignment.
562
563::
564
565 void
566 dma_cache_sync(struct device *dev, void *vaddr, size_t size,
567 enum dma_data_direction direction)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700568
Christoph Hellwig2fd523c2017-08-25 17:06:13 +0200569Do a partial sync of memory that was allocated by dma_alloc_attrs() with
570the DMA_ATTR_NON_CONSISTENT flag starting at virtual address vaddr and
Linus Torvalds1da177e2005-04-16 15:20:36 -0700571continuing on for size. Again, you *must* observe the cache line
572boundaries when doing this.
573
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300574::
575
576 int
577 dma_declare_coherent_memory(struct device *dev, phys_addr_t phys_addr,
578 dma_addr_t device_addr, size_t size, int
579 flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700580
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600581Declare region of memory to be handed out by dma_alloc_coherent() when
Linus Torvalds1da177e2005-04-16 15:20:36 -0700582it's asked for coherent memory for this device.
583
Bjorn Helgaasf311a722014-05-20 16:56:27 -0600584phys_addr is the CPU physical address to which the memory is currently
585assigned (this will be ioremapped so the CPU can access the region).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700586
Yinghai Lu3a9ad0b2015-05-27 17:23:51 -0700587device_addr is the DMA address the device needs to be programmed
Bjorn Helgaas88a984b2014-05-20 16:54:22 -0600588with to actually address this memory (this will be handed out as the
Randy Dunlapa12e2c62007-07-31 00:38:17 -0700589dma_addr_t in dma_alloc_coherent()).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700590
591size is the size of the area (must be multiples of PAGE_SIZE).
592
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600593flags can be ORed together and are:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700594
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300595- DMA_MEMORY_MAP - request that the memory returned from
596 dma_alloc_coherent() be directly writable.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700597
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300598- DMA_MEMORY_IO - request that the memory returned from
599 dma_alloc_coherent() be addressable using read()/write()/memcpy_toio() etc.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700600
Randy Dunlapa12e2c62007-07-31 00:38:17 -0700601One or both of these flags must be present.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700602
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300603- DMA_MEMORY_INCLUDES_CHILDREN - make the declared memory be allocated by
604 dma_alloc_coherent of any child devices of this one (for memory residing
605 on a bridge).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700606
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300607- DMA_MEMORY_EXCLUSIVE - only allocate memory from the declared regions.
608 Do not allow dma_alloc_coherent() to fall back to system memory when
609 it's out of memory in the declared region.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610
611The return value will be either DMA_MEMORY_MAP or DMA_MEMORY_IO and
612must correspond to a passed in flag (i.e. no returning DMA_MEMORY_IO
613if only DMA_MEMORY_MAP were passed in) for success or zero for
614failure.
615
616Note, for DMA_MEMORY_IO returns, all subsequent memory returned by
617dma_alloc_coherent() may no longer be accessed directly, but instead
618must be accessed using the correct bus functions. If your driver
619isn't prepared to handle this contingency, it should not specify
620DMA_MEMORY_IO in the input flags.
621
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300622As a simplification for the platforms, only **one** such region of
Linus Torvalds1da177e2005-04-16 15:20:36 -0700623memory may be declared per device.
624
625For reasons of efficiency, most platforms choose to track the declared
626region only at the granularity of a page. For smaller allocations,
627you should use the dma_pool() API.
628
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300629::
630
631 void
632 dma_release_declared_memory(struct device *dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700633
634Remove the memory region previously declared from the system. This
635API performs *no* in-use checking for this region and will return
636unconditionally having removed all the required structures. It is the
Randy Dunlapa12e2c62007-07-31 00:38:17 -0700637driver's job to ensure that no parts of this memory region are
Linus Torvalds1da177e2005-04-16 15:20:36 -0700638currently in use.
639
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300640::
641
642 void *
643 dma_mark_declared_memory_occupied(struct device *dev,
644 dma_addr_t device_addr, size_t size)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700645
646This is used to occupy specific regions of the declared space
647(dma_alloc_coherent() will hand out the first free region it finds).
648
Randy Dunlapa12e2c62007-07-31 00:38:17 -0700649device_addr is the *device* address of the region requested.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700650
Randy Dunlapa12e2c62007-07-31 00:38:17 -0700651size is the size (and should be a page-sized multiple).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700652
653The return value will be either a pointer to the processor virtual
654address of the memory, or an error (via PTR_ERR()) if any part of the
655region is occupied.
Joerg Roedel187f9c32009-01-09 16:28:07 +0100656
657Part III - Debug drivers use of the DMA-API
658-------------------------------------------
659
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600660The DMA-API as described above has some constraints. DMA addresses must be
Joerg Roedel187f9c32009-01-09 16:28:07 +0100661released with the corresponding function with the same size for example. With
662the advent of hardware IOMMUs it becomes more and more important that drivers
663do not violate those constraints. In the worst case such a violation can
664result in data corruption up to destroyed filesystems.
665
666To debug drivers and find bugs in the usage of the DMA-API checking code can
667be compiled into the kernel which will tell the developer about those
668violations. If your architecture supports it you can select the "Enable
669debugging of DMA-API usage" option in your kernel configuration. Enabling this
670option has a performance impact. Do not enable it in production kernels.
671
672If you boot the resulting kernel will contain code which does some bookkeeping
673about what DMA memory was allocated for which device. If this code detects an
674error it prints a warning message with some details into your kernel log. An
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300675example warning message may look like this::
Joerg Roedel187f9c32009-01-09 16:28:07 +0100676
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300677 WARNING: at /data2/repos/linux-2.6-iommu/lib/dma-debug.c:448
678 check_unmap+0x203/0x490()
679 Hardware name:
680 forcedeth 0000:00:08.0: DMA-API: device driver frees DMA memory with wrong
681 function [device address=0x00000000640444be] [size=66 bytes] [mapped as
682 single] [unmapped as page]
683 Modules linked in: nfsd exportfs bridge stp llc r8169
684 Pid: 0, comm: swapper Tainted: G W 2.6.28-dmatest-09289-g8bb99c0 #1
685 Call Trace:
686 <IRQ> [<ffffffff80240b22>] warn_slowpath+0xf2/0x130
687 [<ffffffff80647b70>] _spin_unlock+0x10/0x30
688 [<ffffffff80537e75>] usb_hcd_link_urb_to_ep+0x75/0xc0
689 [<ffffffff80647c22>] _spin_unlock_irqrestore+0x12/0x40
690 [<ffffffff8055347f>] ohci_urb_enqueue+0x19f/0x7c0
691 [<ffffffff80252f96>] queue_work+0x56/0x60
692 [<ffffffff80237e10>] enqueue_task_fair+0x20/0x50
693 [<ffffffff80539279>] usb_hcd_submit_urb+0x379/0xbc0
694 [<ffffffff803b78c3>] cpumask_next_and+0x23/0x40
695 [<ffffffff80235177>] find_busiest_group+0x207/0x8a0
696 [<ffffffff8064784f>] _spin_lock_irqsave+0x1f/0x50
697 [<ffffffff803c7ea3>] check_unmap+0x203/0x490
698 [<ffffffff803c8259>] debug_dma_unmap_page+0x49/0x50
699 [<ffffffff80485f26>] nv_tx_done_optimized+0xc6/0x2c0
700 [<ffffffff80486c13>] nv_nic_irq_optimized+0x73/0x2b0
701 [<ffffffff8026df84>] handle_IRQ_event+0x34/0x70
702 [<ffffffff8026ffe9>] handle_edge_irq+0xc9/0x150
703 [<ffffffff8020e3ab>] do_IRQ+0xcb/0x1c0
704 [<ffffffff8020c093>] ret_from_intr+0x0/0xa
705 <EOI> <4>---[ end trace f6435a98e2a38c0e ]---
Joerg Roedel187f9c32009-01-09 16:28:07 +0100706
707The driver developer can find the driver and the device including a stacktrace
708of the DMA-API call which caused this warning.
709
710Per default only the first error will result in a warning message. All other
711errors will only silently counted. This limitation exist to prevent the code
712from flooding your kernel log. To support debugging a device driver this can
713be disabled via debugfs. See the debugfs interface documentation below for
714details.
715
716The debugfs directory for the DMA-API debugging code is called dma-api/. In
717this directory the following files can currently be found:
718
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300719=============================== ===============================================
720dma-api/all_errors This file contains a numeric value. If this
Joerg Roedel187f9c32009-01-09 16:28:07 +0100721 value is not equal to zero the debugging code
722 will print a warning for every error it finds
Matt LaPlante19f59462009-04-27 15:06:31 +0200723 into the kernel log. Be careful with this
724 option, as it can easily flood your logs.
Joerg Roedel187f9c32009-01-09 16:28:07 +0100725
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300726dma-api/disabled This read-only file contains the character 'Y'
Joerg Roedel187f9c32009-01-09 16:28:07 +0100727 if the debugging code is disabled. This can
728 happen when it runs out of memory or if it was
729 disabled at boot time
730
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300731dma-api/error_count This file is read-only and shows the total
Joerg Roedel187f9c32009-01-09 16:28:07 +0100732 numbers of errors found.
733
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300734dma-api/num_errors The number in this file shows how many
Joerg Roedel187f9c32009-01-09 16:28:07 +0100735 warnings will be printed to the kernel log
736 before it stops. This number is initialized to
737 one at system boot and be set by writing into
738 this file
739
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300740dma-api/min_free_entries This read-only file can be read to get the
Joerg Roedel187f9c32009-01-09 16:28:07 +0100741 minimum number of free dma_debug_entries the
742 allocator has ever seen. If this value goes
743 down to zero the code will disable itself
744 because it is not longer reliable.
745
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300746dma-api/num_free_entries The current number of free dma_debug_entries
Joerg Roedel187f9c32009-01-09 16:28:07 +0100747 in the allocator.
748
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300749dma-api/driver-filter You can write a name of a driver into this file
Joerg Roedel016ea682009-05-22 21:57:23 +0200750 to limit the debug output to requests from that
751 particular driver. Write an empty string to
752 that file to disable the filter and see
753 all errors again.
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300754=============================== ===============================================
Joerg Roedel016ea682009-05-22 21:57:23 +0200755
Joerg Roedel187f9c32009-01-09 16:28:07 +0100756If you have this code compiled into your kernel it will be enabled by default.
757If you want to boot without the bookkeeping anyway you can provide
758'dma_debug=off' as a boot parameter. This will disable DMA-API debugging.
759Notice that you can not enable it again at runtime. You have to reboot to do
760so.
761
Joerg Roedel016ea682009-05-22 21:57:23 +0200762If you want to see debug messages only for a special device driver you can
763specify the dma_debug_driver=<drivername> parameter. This will enable the
764driver filter at boot time. The debug code will only print errors for that
765driver afterwards. This filter can be disabled or changed later using debugfs.
766
Joerg Roedel187f9c32009-01-09 16:28:07 +0100767When the code disables itself at runtime this is most likely because it ran
768out of dma_debug_entries. These entries are preallocated at boot. The number
769of preallocated entries is defined per architecture. If it is too low for you
770boot with 'dma_debug_entries=<your_desired_number>' to overwrite the
771architectural default.
Shuah Khan6c9c6d62012-10-08 11:08:06 -0600772
Mauro Carvalho Chehab03158a72017-05-14 07:27:52 -0300773::
774
775 void
776 debug_dma_mapping_error(struct device *dev, dma_addr_t dma_addr);
Shuah Khan6c9c6d62012-10-08 11:08:06 -0600777
778dma-debug interface debug_dma_mapping_error() to debug drivers that fail
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600779to check DMA mapping errors on addresses returned by dma_map_single() and
Shuah Khan6c9c6d62012-10-08 11:08:06 -0600780dma_map_page() interfaces. This interface clears a flag set by
781debug_dma_map_page() to indicate that dma_mapping_error() has been called by
782the driver. When driver does unmap, debug_dma_unmap() checks the flag and if
783this flag is still set, prints warning message that includes call trace that
784leads up to the unmap. This interface can be called from dma_mapping_error()
Bjorn Helgaas77f2ea22014-04-30 11:20:53 -0600785routines to enable DMA mapping error check debugging.