Blame view
Documentation/DMA-API-HOWTO.txt
32.8 KB
266921bdb
|
1 2 3 |
========================= Dynamic DMA mapping Guide ========================= |
1da177e4c
|
4 |
|
266921bdb
|
5 6 7 |
:Author: David S. Miller <davem@redhat.com> :Author: Richard Henderson <rth@cygnus.com> :Author: Jakub Jelinek <jakub@redhat.com> |
1da177e4c
|
8 |
|
216bf58f4
|
9 10 |
This is a guide to device driver writers on how to use the DMA API with example pseudo-code. For a concise description of the API, see |
1da177e4c
|
11 |
DMA-API.txt. |
266921bdb
|
12 13 |
CPU and DMA addresses ===================== |
77f2ea2f8
|
14 15 16 17 18 19 |
There are several kinds of addresses involved in the DMA API, and it's important to understand the differences. The kernel normally uses virtual addresses. Any address returned by kmalloc(), vmalloc(), and similar interfaces is a virtual address and can |
266921bdb
|
20 |
be stored in a ``void *``. |
77f2ea2f8
|
21 22 23 24 25 26 27 |
The virtual memory system (TLB, page tables, etc.) translates virtual addresses to CPU physical addresses, which are stored as "phys_addr_t" or "resource_size_t". The kernel manages device resources like registers as physical addresses. These are the addresses in /proc/iomem. The physical address is not directly useful to a driver; it must use ioremap() to map the space and produce a virtual address. |
3a9ad0b4f
|
28 29 30 31 32 |
I/O devices use a third kind of address: a "bus address". If a device has registers at an MMIO address, or if it performs DMA to read or write system memory, the addresses used by the device are bus addresses. In some systems, bus addresses are identical to CPU physical addresses, but in general they are not. IOMMUs and host bridges can produce arbitrary |
77f2ea2f8
|
33 |
mappings between physical and bus addresses. |
3a9ad0b4f
|
34 35 36 37 |
From a device's point of view, DMA uses the bus address space, but it may be restricted to a subset of that space. For example, even if a system supports 64-bit addresses for main memory and PCI BARs, it may use an IOMMU so devices only need to use 32-bit DMA addresses. |
266921bdb
|
38 |
Here's a picture and some examples:: |
77f2ea2f8
|
39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 |
CPU CPU Bus Virtual Physical Address Address Address Space Space Space +-------+ +------+ +------+ | | |MMIO | Offset | | | | Virtual |Space | applied | | C +-------+ --------> B +------+ ----------> +------+ A | | mapping | | by host | | +-----+ | | | | bridge | | +--------+ | | | | +------+ | | | | | CPU | | | | RAM | | | | Device | | | | | | | | | | | +-----+ +-------+ +------+ +------+ +--------+ | | Virtual |Buffer| Mapping | | X +-------+ --------> Y +------+ <---------- +------+ Z | | mapping | RAM | by IOMMU | | | | | | | | +-------+ +------+ During the enumeration process, the kernel learns about I/O devices and their MMIO space and the host bridges that connect them to the system. For example, if a PCI device has a BAR, the kernel reads the bus address (A) from the BAR and converts it to a CPU physical address (B). The address B is stored in a struct resource and usually exposed via /proc/iomem. When a driver claims a device, it typically uses ioremap() to map physical address B at a virtual address (C). It can then use, e.g., ioread32(C), to access the device registers at bus address A. If the device supports DMA, the driver sets up a buffer using kmalloc() or a similar interface, which returns a virtual address (X). The virtual memory system maps X to a physical address (Y) in system RAM. The driver can use virtual address X to access the buffer, but the device itself cannot because DMA doesn't go through the CPU virtual memory system. In some simple systems, the device can do DMA directly to physical address |
3a9ad0b4f
|
78 |
Y. But in many others, there is IOMMU hardware that translates DMA |
77f2ea2f8
|
79 80 81 |
addresses to physical addresses, e.g., it translates Z to Y. This is part of the reason for the DMA API: the driver can give a virtual address X to an interface like dma_map_single(), which sets up any required IOMMU |
3a9ad0b4f
|
82 |
mapping and returns the DMA address Z. The driver then tells the device to |
77f2ea2f8
|
83 84 |
do DMA to Z, and the IOMMU maps it to the buffer at address Y in system RAM. |
1da177e4c
|
85 86 87 88 89 90 91 |
So that Linux can use the dynamic DMA mapping, it needs some help from the drivers, namely it has to take into account that DMA addresses should be mapped only for the time they are actually used and unmapped after the DMA transfer. The following API will work of course even on platforms where no such |
216bf58f4
|
92 93 94 |
hardware exists. Note that the DMA API works with any bus independent of the underlying |
77f2ea2f8
|
95 96 97 |
microprocessor architecture. You should use the DMA API rather than the bus-specific DMA API, i.e., use the dma_map_*() interfaces rather than the pci_map_*() interfaces. |
1da177e4c
|
98 |
|
266921bdb
|
99 |
First of all, you should make sure:: |
1da177e4c
|
100 |
|
266921bdb
|
101 |
#include <linux/dma-mapping.h> |
1da177e4c
|
102 |
|
77f2ea2f8
|
103 |
is in your driver, which provides the definition of dma_addr_t. This type |
3a9ad0b4f
|
104 |
can hold any valid DMA address for the platform and should be used |
77f2ea2f8
|
105 |
everywhere you hold a DMA address returned from the DMA mapping functions. |
1da177e4c
|
106 |
|
266921bdb
|
107 108 |
What memory is DMA'able? ======================== |
1da177e4c
|
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 |
The first piece of information you must know is what kernel memory can be used with the DMA mapping facilities. There has been an unwritten set of rules regarding this, and this text is an attempt to finally write them down. If you acquired your memory via the page allocator (i.e. __get_free_page*()) or the generic memory allocators (i.e. kmalloc() or kmem_cache_alloc()) then you may DMA to/from that memory using the addresses returned from those routines. This means specifically that you may _not_ use the memory/addresses returned from vmalloc() for DMA. It is possible to DMA to the _underlying_ memory mapped into a vmalloc() area, but this requires walking page tables to get the physical addresses, and then translating each of those pages back to a kernel address using something like __va(). [ EDIT: Update this when we integrate Gerd Knorr's generic code which does this. ] |
21440d313
|
127 128 129 130 131 132 133 134 135 |
This rule also means that you may use neither kernel image addresses (items in data/text/bss segments), nor module image addresses, nor stack addresses for DMA. These could all be mapped somewhere entirely different than the rest of physical memory. Even if those classes of memory could physically work with DMA, you'd need to ensure the I/O buffers were cacheline-aligned. Without that, you'd see cacheline sharing problems (data corruption) on CPUs with DMA-incoherent caches. (The CPU could write to one word, DMA would write to a different one in the same cache line, and one of them could be overwritten.) |
1da177e4c
|
136 137 138 139 140 141 142 |
Also, this means that you cannot take the return of a kmap() call and DMA to/from that. This is similar to vmalloc(). What about block I/O and networking buffers? The block I/O and networking subsystems make sure that the buffers they use are valid for you to DMA from/to. |
9eb9e96e9
|
143 |
DMA addressing capabilities |
9fda5130d
|
144 |
=========================== |
1da177e4c
|
145 |
|
9eb9e96e9
|
146 147 148 |
By default, the kernel assumes that your device can address 32-bits of DMA addressing. For a 64-bit capable device, this needs to be increased, and for a device with limitations, it needs to be decreased. |
1da177e4c
|
149 |
|
9eb9e96e9
|
150 151 152 153 |
Special note about PCI: PCI-X specification requires PCI-X devices to support 64-bit addressing (DAC) for all transactions. And at least one platform (SGI SN2) requires 64-bit consistent allocations to operate correctly when the IO bus is in PCI-X mode. |
216bf58f4
|
154 |
|
9eb9e96e9
|
155 156 |
For correct operation, you must set the DMA mask to inform the kernel about your devices DMA addressing capabilities. |
216bf58f4
|
157 |
|
9eb9e96e9
|
158 |
This is performed via a call to dma_set_mask_and_coherent():: |
1da177e4c
|
159 |
|
4aa806b77
|
160 |
int dma_set_mask_and_coherent(struct device *dev, u64 mask); |
1da177e4c
|
161 |
|
9eb9e96e9
|
162 163 164 |
which will set the mask for both streaming and coherent APIs together. If you have some special requirements, then the following two separate calls can be used instead: |
1da177e4c
|
165 |
|
9eb9e96e9
|
166 |
The setup for streaming mappings is performed via a call to |
266921bdb
|
167 |
dma_set_mask():: |
4aa806b77
|
168 169 |
int dma_set_mask(struct device *dev, u64 mask); |
9eb9e96e9
|
170 |
The setup for consistent allocations is performed via a call |
266921bdb
|
171 |
to dma_set_coherent_mask():: |
4aa806b77
|
172 173 |
int dma_set_coherent_mask(struct device *dev, u64 mask); |
1da177e4c
|
174 |
|
9eb9e96e9
|
175 176 177 178 179 |
Here, dev is a pointer to the device struct of your device, and mask is a bit mask describing which bits of an address your device supports. Often the device struct of your device is embedded in the bus-specific device struct of your device. For example, &pdev->dev is a pointer to the device struct of a PCI device (pdev is a pointer to the PCI device struct of your device). |
1da177e4c
|
180 |
|
9eb9e96e9
|
181 182 183 184 185 186 187 |
These calls usually return zero to indicated your device can perform DMA properly on the machine given the address mask you provided, but they might return an error if the mask is too small to be supportable on the given system. If it returns non-zero, your device cannot perform DMA properly on this platform, and attempting to do so will result in undefined behavior. You must not use DMA on this device unless the dma_set_mask family of functions has returned success. |
1da177e4c
|
188 |
|
9eb9e96e9
|
189 |
This means that in the failure case, you have two options: |
1da177e4c
|
190 |
|
9eb9e96e9
|
191 192 |
1) Use some non-DMA mode for data transfer, if possible. 2) Ignore this device and do not initialize it. |
1da177e4c
|
193 |
|
9eb9e96e9
|
194 195 196 197 |
It is recommended that your driver print a kernel KERN_WARNING message when setting the DMA mask fails. In this manner, if a user of your driver reports that performance is bad or that the device is not even detected, you can ask them for the kernel messages to find out exactly why. |
1da177e4c
|
198 |
|
9eb9e96e9
|
199 |
The standard 64-bit addressing device would do something like this:: |
1da177e4c
|
200 |
|
9eb9e96e9
|
201 |
if (dma_set_mask_and_coherent(dev, DMA_BIT_MASK(64))) { |
77f2ea2f8
|
202 203 |
dev_warn(dev, "mydev: No suitable DMA available "); |
1da177e4c
|
204 205 |
goto ignore_this_device; } |
9eb9e96e9
|
206 207 |
If the device only supports 32-bit addressing for descriptors in the coherent allocations, but supports full 64-bits for streaming mappings |
a36d05386
|
208 |
it would look like this:: |
1da177e4c
|
209 |
|
9eb9e96e9
|
210 |
if (dma_set_mask(dev, DMA_BIT_MASK(64))) { |
77f2ea2f8
|
211 212 |
dev_warn(dev, "mydev: No suitable DMA available "); |
1da177e4c
|
213 214 |
goto ignore_this_device; } |
34c815fbd
|
215 216 217 218 |
The coherent mask will always be able to set the same or a smaller mask as the streaming mask. However for the rare case that a device driver only uses consistent allocations, one would have to check the return value from dma_set_coherent_mask(). |
1da177e4c
|
219 |
|
1da177e4c
|
220 |
Finally, if your device can only drive the low 24-bits of |
266921bdb
|
221 |
address you might do something like:: |
1da177e4c
|
222 |
|
216bf58f4
|
223 |
if (dma_set_mask(dev, DMA_BIT_MASK(24))) { |
77f2ea2f8
|
224 225 |
dev_warn(dev, "mydev: 24-bit DMA addressing not available "); |
1da177e4c
|
226 227 |
goto ignore_this_device; } |
4aa806b77
|
228 229 230 |
When dma_set_mask() or dma_set_mask_and_coherent() is successful, and returns zero, the kernel saves away this mask you have provided. The kernel will use this information later when you make DMA mappings. |
1da177e4c
|
231 232 233 234 235 236 237 |
There is a case which we are aware of at this time, which is worth mentioning in this documentation. If your device supports multiple functions (for example a sound card provides playback and record functions) and the various different functions have _different_ DMA addressing limitations, you may wish to probe each mask and only provide the functionality which the machine can handle. It |
216bf58f4
|
238 |
is important that the last call to dma_set_mask() be for the |
1da177e4c
|
239 |
most specific mask. |
266921bdb
|
240 |
Here is pseudo-code showing how this might be done:: |
1da177e4c
|
241 |
|
2c5510d4e
|
242 |
#define PLAYBACK_ADDRESS_BITS DMA_BIT_MASK(32) |
038f7d002
|
243 |
#define RECORD_ADDRESS_BITS DMA_BIT_MASK(24) |
1da177e4c
|
244 245 |
struct my_sound_card *card; |
216bf58f4
|
246 |
struct device *dev; |
1da177e4c
|
247 248 |
... |
216bf58f4
|
249 |
if (!dma_set_mask(dev, PLAYBACK_ADDRESS_BITS)) { |
1da177e4c
|
250 251 252 |
card->playback_enabled = 1; } else { card->playback_enabled = 0; |
77f2ea2f8
|
253 254 |
dev_warn(dev, "%s: Playback disabled due to DMA limitations ", |
1da177e4c
|
255 256 |
card->name); } |
216bf58f4
|
257 |
if (!dma_set_mask(dev, RECORD_ADDRESS_BITS)) { |
1da177e4c
|
258 259 260 |
card->record_enabled = 1; } else { card->record_enabled = 0; |
77f2ea2f8
|
261 262 |
dev_warn(dev, "%s: Record disabled due to DMA limitations ", |
1da177e4c
|
263 264 265 266 267 268 |
card->name); } A sound card was used as an example here because this genre of PCI devices seems to be littered with ISA chips given a PCI front end, and thus retaining the 16MB DMA addressing limitations of ISA. |
266921bdb
|
269 270 |
Types of DMA mappings ===================== |
1da177e4c
|
271 272 273 274 275 276 277 278 279 280 281 282 |
There are two types of DMA mappings: - Consistent DMA mappings which are usually mapped at driver initialization, unmapped at the end and for which the hardware should guarantee that the device and the CPU can access the data in parallel and will see updates made by each other without any explicit software flushing. Think of "consistent" as "synchronous" or "coherent". The current default is to return consistent memory in the low 32 |
3a9ad0b4f
|
283 |
bits of the DMA space. However, for future compatibility you should |
216bf58f4
|
284 |
set the consistent mask even if this default is fine for your |
1da177e4c
|
285 286 287 288 289 290 291 292 293 294 295 296 |
driver. Good examples of what to use consistent mappings for are: - Network card DMA ring descriptors. - SCSI adapter mailbox command data structures. - Device firmware microcode executed out of main memory. The invariant these examples all require is that any CPU store to memory is immediately visible to the device, and vice versa. Consistent mappings guarantee this. |
266921bdb
|
297 298 299 300 |
.. important:: Consistent DMA memory does not preclude the usage of proper memory barriers. The CPU may reorder stores to |
1da177e4c
|
301 302 303 |
consistent memory just as it may normal memory. Example: if it is important for the device to see the first word of a descriptor updated before the second, you must do |
266921bdb
|
304 |
something like:: |
1da177e4c
|
305 306 307 308 309 310 |
desc->word0 = address; wmb(); desc->word1 = DESC_VALID; in order to get correct behavior on all platforms. |
21440d313
|
311 312 313 314 |
Also, on some platforms your driver may need to flush CPU write buffers in much the same way as it needs to flush write buffers found in PCI bridges (such as by reading a register's value after writing it). |
216bf58f4
|
315 316 317 |
- Streaming DMA mappings which are usually mapped for one DMA transfer, unmapped right after it (unless you use dma_sync_* below) and for which hardware can optimize for sequential accesses. |
1da177e4c
|
318 |
|
11e285d8f
|
319 |
Think of "streaming" as "asynchronous" or "outside the coherency |
1da177e4c
|
320 321 322 323 324 325 326 327 328 329 330 |
domain". Good examples of what to use streaming mappings for are: - Networking buffers transmitted/received by a device. - Filesystem buffers written/read by a SCSI device. The interfaces for using this type of mapping were designed in such a way that an implementation can make whatever performance optimizations the hardware allows. To this end, when using such mappings you must be explicit about what you want to happen. |
216bf58f4
|
331 332 |
Neither type of DMA mapping has alignment restrictions that come from the underlying bus, although some devices may have such restrictions. |
21440d313
|
333 334 |
Also, systems with caches that aren't DMA-coherent will work better when the underlying buffers don't share cache lines with other data. |
1da177e4c
|
335 |
|
266921bdb
|
336 337 |
Using Consistent DMA mappings ============================= |
1da177e4c
|
338 339 |
To allocate and map large (PAGE_SIZE or so) consistent DMA regions, |
266921bdb
|
340 |
you should do:: |
1da177e4c
|
341 342 |
dma_addr_t dma_handle; |
216bf58f4
|
343 |
cpu_addr = dma_alloc_coherent(dev, size, &dma_handle, gfp); |
1da177e4c
|
344 |
|
266921bdb
|
345 |
where device is a ``struct device *``. This may be called in interrupt |
216bf58f4
|
346 |
context with the GFP_ATOMIC flag. |
1da177e4c
|
347 348 349 350 |
Size is the length of the region you want to allocate, in bytes. This routine will allocate RAM for that region, so it acts similarly to |
77f2ea2f8
|
351 |
__get_free_pages() (but takes size instead of a page order). If your |
1da177e4c
|
352 |
driver needs regions sized smaller than a page, you may prefer using |
216bf58f4
|
353 |
the dma_pool interface, described below. |
d7e02a931
|
354 355 356 357 358 359 |
The consistent DMA mapping interfaces, will by default return a DMA address which is 32-bit addressable. Even if the device indicates (via the DMA mask) that it may address the upper 32-bits, consistent allocation will only return > 32-bit addresses for DMA if the consistent DMA mask has been explicitly changed via dma_set_coherent_mask(). This is true of the dma_pool interface as well. |
216bf58f4
|
360 |
|
77f2ea2f8
|
361 |
dma_alloc_coherent() returns two values: the virtual address which you |
1da177e4c
|
362 363 |
can use to access it from the CPU and dma_handle which you pass to the card. |
3a9ad0b4f
|
364 |
The CPU virtual address and the DMA address are both |
1da177e4c
|
365 366 367 368 369 |
guaranteed to be aligned to the smallest PAGE_SIZE order which is greater than or equal to the requested size. This invariant exists (for example) to guarantee that if you allocate a chunk which is smaller than or equal to 64 kilobytes, the extent of the buffer you receive will not cross a 64K boundary. |
266921bdb
|
370 |
To unmap and free such a DMA region, you call:: |
1da177e4c
|
371 |
|
216bf58f4
|
372 |
dma_free_coherent(dev, size, cpu_addr, dma_handle); |
1da177e4c
|
373 |
|
216bf58f4
|
374 |
where dev, size are the same as in the above call and cpu_addr and |
77f2ea2f8
|
375 |
dma_handle are the values dma_alloc_coherent() returned to you. |
1da177e4c
|
376 377 378 |
This function may not be called in interrupt context. If your driver needs lots of smaller memory regions, you can write |
77f2ea2f8
|
379 |
custom code to subdivide pages returned by dma_alloc_coherent(), |
216bf58f4
|
380 |
or you can use the dma_pool API to do that. A dma_pool is like |
77f2ea2f8
|
381 |
a kmem_cache, but it uses dma_alloc_coherent(), not __get_free_pages(). |
1da177e4c
|
382 383 |
Also, it understands common hardware constraints for alignment, like queue heads needing to be aligned on N byte boundaries. |
266921bdb
|
384 |
Create a dma_pool like this:: |
1da177e4c
|
385 |
|
216bf58f4
|
386 |
struct dma_pool *pool; |
1da177e4c
|
387 |
|
2af9da863
|
388 |
pool = dma_pool_create(name, dev, size, align, boundary); |
1da177e4c
|
389 |
|
216bf58f4
|
390 |
The "name" is for diagnostics (like a kmem_cache name); dev and size |
1da177e4c
|
391 392 393 |
are as above. The device's hardware alignment requirement for this type of data is "align" (which is expressed in bytes, and must be a power of two). If your device has no boundary crossing restrictions, |
2af9da863
|
394 |
pass 0 for boundary; passing 4096 says memory allocated from this pool |
1da177e4c
|
395 |
must not cross 4KByte boundaries (but at that time it may be better to |
77f2ea2f8
|
396 |
use dma_alloc_coherent() directly instead). |
1da177e4c
|
397 |
|
266921bdb
|
398 |
Allocate memory from a DMA pool like this:: |
1da177e4c
|
399 |
|
216bf58f4
|
400 |
cpu_addr = dma_pool_alloc(pool, flags, &dma_handle); |
1da177e4c
|
401 |
|
2af9da863
|
402 403 |
flags are GFP_KERNEL if blocking is permitted (not in_interrupt nor holding SMP locks), GFP_ATOMIC otherwise. Like dma_alloc_coherent(), |
1da177e4c
|
404 |
this returns two values, cpu_addr and dma_handle. |
266921bdb
|
405 |
Free memory that was allocated from a dma_pool like this:: |
1da177e4c
|
406 |
|
216bf58f4
|
407 |
dma_pool_free(pool, cpu_addr, dma_handle); |
1da177e4c
|
408 |
|
77f2ea2f8
|
409 410 |
where pool is what you passed to dma_pool_alloc(), and cpu_addr and dma_handle are the values dma_pool_alloc() returned. This function |
1da177e4c
|
411 |
may be called in interrupt context. |
266921bdb
|
412 |
Destroy a dma_pool by calling:: |
1da177e4c
|
413 |
|
216bf58f4
|
414 |
dma_pool_destroy(pool); |
1da177e4c
|
415 |
|
77f2ea2f8
|
416 |
Make sure you've called dma_pool_free() for all memory allocated |
1da177e4c
|
417 418 |
from a pool before you destroy the pool. This function may not be called in interrupt context. |
266921bdb
|
419 420 |
DMA Direction ============= |
1da177e4c
|
421 422 423 |
The interfaces described in subsequent portions of this document take a DMA direction argument, which is an integer and takes on |
266921bdb
|
424 |
one of the following values:: |
1da177e4c
|
425 |
|
216bf58f4
|
426 427 428 429 |
DMA_BIDIRECTIONAL DMA_TO_DEVICE DMA_FROM_DEVICE DMA_NONE |
1da177e4c
|
430 |
|
77f2ea2f8
|
431 |
You should provide the exact DMA direction if you know it. |
1da177e4c
|
432 |
|
216bf58f4
|
433 434 |
DMA_TO_DEVICE means "from main memory to the device" DMA_FROM_DEVICE means "from the device to main memory" |
1da177e4c
|
435 436 437 438 439 440 441 |
It is the direction in which the data moves during the DMA transfer. You are _strongly_ encouraged to specify this as precisely as you possibly can. If you absolutely cannot know the direction of the DMA transfer, |
216bf58f4
|
442 |
specify DMA_BIDIRECTIONAL. It means that the DMA can go in |
1da177e4c
|
443 444 445 |
either direction. The platform guarantees that you may legally specify this, and that it will work, but this may be at the cost of performance for example. |
216bf58f4
|
446 |
The value DMA_NONE is to be used for debugging. One can |
1da177e4c
|
447 448 449 450 451 452 453 454 455 |
hold this in a data structure before you come to know the precise direction, and this will help catch cases where your direction tracking logic has failed to set things up properly. Another advantage of specifying this value precisely (outside of potential platform-specific optimizations of such) is for debugging. Some platforms actually have a write permission boolean which DMA mappings can be marked with, much like page protections in the user program address space. Such platforms can and do report errors in the |
216bf58f4
|
456 |
kernel logs when the DMA controller hardware detects violation of the |
1da177e4c
|
457 458 459 460 |
permission setting. Only streaming mappings specify a direction, consistent mappings implicitly have a direction attribute setting of |
216bf58f4
|
461 |
DMA_BIDIRECTIONAL. |
1da177e4c
|
462 |
|
be7db055d
|
463 464 465 |
The SCSI subsystem tells you the direction to use in the 'sc_data_direction' member of the SCSI command your driver is working on. |
1da177e4c
|
466 467 |
For Networking drivers, it's a rather simple affair. For transmit |
216bf58f4
|
468 |
packets, map/unmap them with the DMA_TO_DEVICE direction |
1da177e4c
|
469 |
specifier. For receive packets, just the opposite, map/unmap them |
216bf58f4
|
470 |
with the DMA_FROM_DEVICE direction specifier. |
1da177e4c
|
471 |
|
266921bdb
|
472 473 |
Using Streaming DMA mappings ============================ |
1da177e4c
|
474 475 476 477 478 |
The streaming DMA mapping routines can be called from interrupt context. There are two versions of each map/unmap, one which will map/unmap a single memory region, and one which will map/unmap a scatterlist. |
266921bdb
|
479 |
To map a single region, you do:: |
1da177e4c
|
480 |
|
216bf58f4
|
481 |
struct device *dev = &my_dev->dev; |
1da177e4c
|
482 483 484 |
dma_addr_t dma_handle; void *addr = buffer->ptr; size_t size = buffer->len; |
216bf58f4
|
485 |
dma_handle = dma_map_single(dev, addr, size, direction); |
b2dd83b37
|
486 |
if (dma_mapping_error(dev, dma_handle)) { |
8d7f62e6a
|
487 488 489 490 491 492 493 |
/* * reduce current DMA mapping usage, * delay and try again later or * reset driver. */ goto map_error_handling; } |
1da177e4c
|
494 |
|
266921bdb
|
495 |
and to unmap it:: |
1da177e4c
|
496 |
|
216bf58f4
|
497 |
dma_unmap_single(dev, dma_handle, size, direction); |
1da177e4c
|
498 |
|
8d7f62e6a
|
499 |
You should call dma_mapping_error() as dma_map_single() could fail and return |
f51f288e2
|
500 501 502 503 504 |
error. Doing so will ensure that the mapping code will work correctly on all DMA implementations without any dependency on the specifics of the underlying implementation. Using the returned address without checking for errors could result in failures ranging from panics to silent data corruption. The same applies to dma_map_page() as well. |
8d7f62e6a
|
505 |
|
77f2ea2f8
|
506 |
You should call dma_unmap_single() when the DMA activity is finished, e.g., |
1da177e4c
|
507 |
from the interrupt which told you that the DMA transfer is done. |
f311a724a
|
508 |
Using CPU pointers like this for single mappings has a disadvantage: |
1da177e4c
|
509 |
you cannot reference HIGHMEM memory in this way. Thus, there is a |
77f2ea2f8
|
510 |
map/unmap interface pair akin to dma_{map,unmap}_single(). These |
f311a724a
|
511 |
interfaces deal with page/offset pairs instead of CPU pointers. |
266921bdb
|
512 |
Specifically:: |
1da177e4c
|
513 |
|
216bf58f4
|
514 |
struct device *dev = &my_dev->dev; |
1da177e4c
|
515 516 517 518 |
dma_addr_t dma_handle; struct page *page = buffer->page; unsigned long offset = buffer->offset; size_t size = buffer->len; |
216bf58f4
|
519 |
dma_handle = dma_map_page(dev, page, offset, size, direction); |
b2dd83b37
|
520 |
if (dma_mapping_error(dev, dma_handle)) { |
8d7f62e6a
|
521 522 523 524 525 526 527 |
/* * reduce current DMA mapping usage, * delay and try again later or * reset driver. */ goto map_error_handling; } |
1da177e4c
|
528 529 |
... |
216bf58f4
|
530 |
dma_unmap_page(dev, dma_handle, size, direction); |
1da177e4c
|
531 532 |
Here, "offset" means byte offset within the given page. |
8d7f62e6a
|
533 534 |
You should call dma_mapping_error() as dma_map_page() could fail and return error as outlined under the dma_map_single() discussion. |
77f2ea2f8
|
535 |
You should call dma_unmap_page() when the DMA activity is finished, e.g., |
8d7f62e6a
|
536 |
from the interrupt which told you that the DMA transfer is done. |
266921bdb
|
537 |
With scatterlists, you map a region gathered from several regions by:: |
1da177e4c
|
538 |
|
216bf58f4
|
539 |
int i, count = dma_map_sg(dev, sglist, nents, direction); |
1da177e4c
|
540 |
struct scatterlist *sg; |
4c2f6d4c2
|
541 |
for_each_sg(sglist, sg, count, i) { |
1da177e4c
|
542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 |
hw_address[i] = sg_dma_address(sg); hw_len[i] = sg_dma_len(sg); } where nents is the number of entries in the sglist. The implementation is free to merge several consecutive sglist entries into one (e.g. if DMA mapping is done with PAGE_SIZE granularity, any consecutive sglist entries can be merged into one provided the first one ends and the second one starts on a page boundary - in fact this is a huge advantage for cards which either cannot do scatter-gather or have very limited number of scatter-gather entries) and returns the actual number of sg entries it mapped them to. On failure 0 is returned. Then you should loop count times (note: this can be less than nents times) and use sg_dma_address() and sg_dma_len() macros where you previously accessed sg->address and sg->length as shown above. |
266921bdb
|
559 |
To unmap a scatterlist, just call:: |
1da177e4c
|
560 |
|
216bf58f4
|
561 |
dma_unmap_sg(dev, sglist, nents, direction); |
1da177e4c
|
562 563 |
Again, make sure DMA activity has already finished. |
266921bdb
|
564 565 566 567 568 569 |
.. note:: The 'nents' argument to the dma_unmap_sg call must be the _same_ one you passed into the dma_map_sg call, it should _NOT_ be the 'count' value _returned_ from the dma_map_sg call. |
1da177e4c
|
570 |
|
77f2ea2f8
|
571 |
Every dma_map_{single,sg}() call should have its dma_unmap_{single,sg}() |
3a9ad0b4f
|
572 573 |
counterpart, because the DMA address space is a shared resource and you could render the machine unusable by consuming all DMA addresses. |
1da177e4c
|
574 575 576 |
If you need to use the same streaming DMA region multiple times and touch the data in between the DMA transfers, the buffer needs to be synced |
f311a724a
|
577 |
properly in order for the CPU and device to see the most up-to-date and |
1da177e4c
|
578 |
correct copy of the DMA buffer. |
77f2ea2f8
|
579 |
So, firstly, just map it with dma_map_{single,sg}(), and after each DMA |
266921bdb
|
580 |
transfer call either:: |
1da177e4c
|
581 |
|
216bf58f4
|
582 |
dma_sync_single_for_cpu(dev, dma_handle, size, direction); |
1da177e4c
|
583 |
|
266921bdb
|
584 |
or:: |
1da177e4c
|
585 |
|
216bf58f4
|
586 |
dma_sync_sg_for_cpu(dev, sglist, nents, direction); |
1da177e4c
|
587 588 589 590 |
as appropriate. Then, if you wish to let the device get at the DMA area again, |
f311a724a
|
591 |
finish accessing the data with the CPU, and then before actually |
266921bdb
|
592 |
giving the buffer to the hardware call either:: |
1da177e4c
|
593 |
|
216bf58f4
|
594 |
dma_sync_single_for_device(dev, dma_handle, size, direction); |
1da177e4c
|
595 |
|
266921bdb
|
596 |
or:: |
1da177e4c
|
597 |
|
216bf58f4
|
598 |
dma_sync_sg_for_device(dev, sglist, nents, direction); |
1da177e4c
|
599 600 |
as appropriate. |
266921bdb
|
601 602 603 |
.. note:: The 'nents' argument to dma_sync_sg_for_cpu() and |
7bc590b2f
|
604 605 606 |
dma_sync_sg_for_device() must be the same passed to dma_map_sg(). It is _NOT_ the count returned by dma_map_sg(). |
1da177e4c
|
607 |
After the last DMA transfer call one of the DMA unmap routines |
77f2ea2f8
|
608 609 610 |
dma_unmap_{single,sg}(). If you don't touch the data from the first dma_map_*() call till dma_unmap_*(), then you don't have to call the dma_sync_*() routines at all. |
1da177e4c
|
611 612 |
Here is pseudo code which shows a situation in which you would need |
266921bdb
|
613 |
to use the dma_sync_*() interfaces:: |
1da177e4c
|
614 615 616 617 |
my_card_setup_receive_buffer(struct my_card *cp, char *buffer, int len) { dma_addr_t mapping; |
216bf58f4
|
618 |
mapping = dma_map_single(cp->dev, buffer, len, DMA_FROM_DEVICE); |
be6c30956
|
619 |
if (dma_mapping_error(cp->dev, mapping)) { |
8d7f62e6a
|
620 621 622 623 624 625 626 |
/* * reduce current DMA mapping usage, * delay and try again later or * reset driver. */ goto map_error_handling; } |
1da177e4c
|
627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 |
cp->rx_buf = buffer; cp->rx_len = len; cp->rx_dma = mapping; give_rx_buf_to_card(cp); } ... my_card_interrupt_handler(int irq, void *devid, struct pt_regs *regs) { struct my_card *cp = devid; ... if (read_card_status(cp) == RX_BUF_TRANSFERRED) { struct my_card_header *hp; /* Examine the header to see if we wish * to accept the data. But synchronize * the DMA transfer with the CPU first * so that we see updated contents. */ |
216bf58f4
|
650 651 652 |
dma_sync_single_for_cpu(&cp->dev, cp->rx_dma, cp->rx_len, DMA_FROM_DEVICE); |
1da177e4c
|
653 654 655 656 |
/* Now it is safe to examine the buffer. */ hp = (struct my_card_header *) cp->rx_buf; if (header_is_ok(hp)) { |
216bf58f4
|
657 658 |
dma_unmap_single(&cp->dev, cp->rx_dma, cp->rx_len, DMA_FROM_DEVICE); |
1da177e4c
|
659 660 661 |
pass_to_upper_layers(cp->rx_buf); make_and_setup_new_rx_buf(cp); } else { |
3f0fb4e85
|
662 663 664 665 666 667 |
/* CPU should not write to * DMA_FROM_DEVICE-mapped area, * so dma_sync_single_for_device() is * not needed here. It would be required * for DMA_BIDIRECTIONAL mapping if * the memory was modified. |
1da177e4c
|
668 |
*/ |
1da177e4c
|
669 670 671 672 |
give_rx_buf_to_card(cp); } } } |
77f2ea2f8
|
673 674 675 |
Drivers converted fully to this interface should not use virt_to_bus() any longer, nor should they use bus_to_virt(). Some drivers have to be changed a little bit, because there is no longer an equivalent to bus_to_virt() in the |
1da177e4c
|
676 |
dynamic DMA mapping scheme - you have to always store the DMA addresses |
77f2ea2f8
|
677 678 |
returned by the dma_alloc_coherent(), dma_pool_alloc(), and dma_map_single() calls (dma_map_sg() stores them in the scatterlist itself if the platform |
1da177e4c
|
679 680 |
supports dynamic DMA mapping in hardware) in your driver structures and/or in the card registers. |
216bf58f4
|
681 682 |
All drivers should be using these interfaces with no exceptions. It is planned to completely remove virt_to_bus() and bus_to_virt() as |
1da177e4c
|
683 684 |
they are entirely deprecated. Some ports already do not provide these as it is impossible to correctly support them. |
266921bdb
|
685 686 |
Handling Errors =============== |
4ae9ca825
|
687 688 689 |
DMA address space is limited on some architectures and an allocation failure can be determined by: |
77f2ea2f8
|
690 |
- checking if dma_alloc_coherent() returns NULL or dma_map_sg returns 0 |
4ae9ca825
|
691 |
|
77f2ea2f8
|
692 |
- checking the dma_addr_t returned from dma_map_single() and dma_map_page() |
266921bdb
|
693 |
by using dma_mapping_error():: |
4ae9ca825
|
694 695 696 697 698 699 700 701 702 703 |
dma_addr_t dma_handle; dma_handle = dma_map_single(dev, addr, size, direction); if (dma_mapping_error(dev, dma_handle)) { /* * reduce current DMA mapping usage, * delay and try again later or * reset driver. */ |
8d7f62e6a
|
704 705 706 707 708 709 |
goto map_error_handling; } - unmap pages that are already mapped, when mapping error occurs in the middle of a multiple page mapping attempt. These example are applicable to dma_map_page() as well. |
266921bdb
|
710 |
Example 1:: |
8d7f62e6a
|
711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 |
dma_addr_t dma_handle1; dma_addr_t dma_handle2; dma_handle1 = dma_map_single(dev, addr, size, direction); if (dma_mapping_error(dev, dma_handle1)) { /* * reduce current DMA mapping usage, * delay and try again later or * reset driver. */ goto map_error_handling1; } dma_handle2 = dma_map_single(dev, addr, size, direction); if (dma_mapping_error(dev, dma_handle2)) { /* * reduce current DMA mapping usage, * delay and try again later or * reset driver. */ goto map_error_handling2; } ... map_error_handling2: dma_unmap_single(dma_handle1); map_error_handling1: |
266921bdb
|
738 739 740 741 742 743 |
Example 2:: /* * if buffers are allocated in a loop, unmap all mapped buffers when * mapping error is detected in the middle */ |
8d7f62e6a
|
744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 |
dma_addr_t dma_addr; dma_addr_t array[DMA_BUFFERS]; int save_index = 0; for (i = 0; i < DMA_BUFFERS; i++) { ... dma_addr = dma_map_single(dev, addr, size, direction); if (dma_mapping_error(dev, dma_addr)) { /* * reduce current DMA mapping usage, * delay and try again later or * reset driver. */ goto map_error_handling; } array[i].dma_addr = dma_addr; save_index++; } ... map_error_handling: for (i = 0; i < save_index; i++) { ... dma_unmap_single(array[i].dma_addr); |
4ae9ca825
|
775 |
} |
77f2ea2f8
|
776 |
Networking drivers must call dev_kfree_skb() to free the socket buffer |
4ae9ca825
|
777 778 779 780 781 782 783 |
and return NETDEV_TX_OK if the DMA mapping fails on the transmit hook (ndo_start_xmit). This means that the socket buffer is just dropped in the failure case. SCSI drivers must return SCSI_MLQUEUE_HOST_BUSY if the DMA mapping fails in the queuecommand hook. This means that the SCSI subsystem passes the command to the driver again later. |
266921bdb
|
784 785 |
Optimizing Unmap State Space Consumption ======================================== |
1da177e4c
|
786 |
|
216bf58f4
|
787 |
On many platforms, dma_unmap_{single,page}() is simply a nop. |
1da177e4c
|
788 789 790 791 792 793 794 |
Therefore, keeping track of the mapping address and length is a waste of space. Instead of filling your drivers up with ifdefs and the like to "work around" this (which would defeat the whole purpose of a portable API) the following facilities are provided. Actually, instead of describing the macros one by one, we'll transform some example code. |
216bf58f4
|
795 |
1) Use DEFINE_DMA_UNMAP_{ADDR,LEN} in state saving structures. |
266921bdb
|
796 |
Example, before:: |
1da177e4c
|
797 798 799 800 801 802 |
struct ring_state { struct sk_buff *skb; dma_addr_t mapping; __u32 len; }; |
266921bdb
|
803 |
after:: |
1da177e4c
|
804 805 806 |
struct ring_state { struct sk_buff *skb; |
216bf58f4
|
807 808 |
DEFINE_DMA_UNMAP_ADDR(mapping); DEFINE_DMA_UNMAP_LEN(len); |
1da177e4c
|
809 |
}; |
77f2ea2f8
|
810 |
2) Use dma_unmap_{addr,len}_set() to set these values. |
266921bdb
|
811 |
Example, before:: |
1da177e4c
|
812 813 814 |
ringp->mapping = FOO; ringp->len = BAR; |
266921bdb
|
815 |
after:: |
1da177e4c
|
816 |
|
216bf58f4
|
817 818 |
dma_unmap_addr_set(ringp, mapping, FOO); dma_unmap_len_set(ringp, len, BAR); |
1da177e4c
|
819 |
|
77f2ea2f8
|
820 |
3) Use dma_unmap_{addr,len}() to access these values. |
266921bdb
|
821 |
Example, before:: |
1da177e4c
|
822 |
|
216bf58f4
|
823 824 |
dma_unmap_single(dev, ringp->mapping, ringp->len, DMA_FROM_DEVICE); |
1da177e4c
|
825 |
|
266921bdb
|
826 |
after:: |
1da177e4c
|
827 |
|
216bf58f4
|
828 829 830 831 |
dma_unmap_single(dev, dma_unmap_addr(ringp, mapping), dma_unmap_len(ringp, len), DMA_FROM_DEVICE); |
1da177e4c
|
832 833 834 835 |
It really should be self-explanatory. We treat the ADDR and LEN separately, because it is possible for an implementation to only need the address in order to perform the unmap operation. |
266921bdb
|
836 837 |
Platform Issues =============== |
1da177e4c
|
838 839 840 841 842 843 |
If you are just writing drivers for Linux and do not maintain an architecture port for the kernel, you can safely skip down to "Closing". 1) Struct scatterlist requirements. |
e92ae527e
|
844 845 |
You need to enable CONFIG_NEED_SG_DMA_LENGTH if the architecture supports IOMMUs (including software IOMMU). |
1da177e4c
|
846 |
|
ce00f7feb
|
847 |
2) ARCH_DMA_MINALIGN |
2fd74e25d
|
848 849 850 851 852 |
Architectures must ensure that kmalloc'ed buffer is DMA-safe. Drivers and subsystems depend on it. If an architecture isn't fully DMA-coherent (i.e. hardware doesn't ensure that data in the CPU cache is identical to data in main memory), |
ce00f7feb
|
853 |
ARCH_DMA_MINALIGN must be set so that the memory allocator |
2fd74e25d
|
854 855 |
makes sure that kmalloc'ed buffer doesn't share a cache line with the others. See arch/arm/include/asm/cache.h as an example. |
ce00f7feb
|
856 |
Note that ARCH_DMA_MINALIGN is about DMA memory alignment |
2fd74e25d
|
857 858 859 |
constraints. You don't need to worry about the architecture data alignment constraints (e.g. the alignment constraints about 64-bit objects). |
1da177e4c
|
860 |
|
266921bdb
|
861 862 |
Closing ======= |
1da177e4c
|
863 |
|
a33f32244
|
864 |
This document, and the API itself, would not be in its current |
1da177e4c
|
865 866 |
form without the feedback and suggestions from numerous individuals. We would like to specifically mention, in no particular order, the |
266921bdb
|
867 |
following people:: |
1da177e4c
|
868 869 870 871 872 873 874 875 |
Russell King <rmk@arm.linux.org.uk> Leo Dagum <dagum@barrel.engr.sgi.com> Ralf Baechle <ralf@oss.sgi.com> Grant Grundler <grundler@cup.hp.com> Jay Estabrook <Jay.Estabrook@compaq.com> Thomas Sailer <sailer@ife.ee.ethz.ch> Andrea Arcangeli <andrea@suse.de> |
26bbb29a2
|
876 |
Jens Axboe <jens.axboe@oracle.com> |
1da177e4c
|
877 |
David Mosberger-Tang <davidm@hpl.hp.com> |