1                        Dynamic DMA mapping
   2                        ===================
   4                 David S. Miller <>
   5                 Richard Henderson <>
   6                  Jakub Jelinek <>
   8This document describes the DMA mapping system in terms of the pci_
   9API.  For a similar API that works for generic devices, see
  12Most of the 64bit platforms have special hardware that translates bus
  13addresses (DMA addresses) into physical addresses.  This is similar to
  14how page tables and/or a TLB translates virtual addresses to physical
  15addresses on a CPU.  This is needed so that e.g. PCI devices can
  16access with a Single Address Cycle (32bit DMA address) any page in the
  1764bit physical address space.  Previously in Linux those 64bit
  18platforms had to set artificial limits on the maximum RAM size in the
  19system, so that the virt_to_bus() static scheme works (the DMA address
  20translation tables were simply filled on bootup to map each bus
  21address to the physical page __pa(bus_to_virt())).
  23So that Linux can use the dynamic DMA mapping, it needs some help from the
  24drivers, namely it has to take into account that DMA addresses should be
  25mapped only for the time they are actually used and unmapped after the DMA
  28The following API will work of course even on platforms where no such
  29hardware exists, see e.g. arch/x86/include/asm/pci.h for how it is implemented on
  30top of the virt_to_bus interface.
  32First of all, you should make sure
  34#include <linux/pci.h>
  36is in your driver. This file will obtain for you the definition of the
  37dma_addr_t (which can hold any valid DMA address for the platform)
  38type which should be used everywhere you hold a DMA (bus) address
  39returned from the DMA mapping functions.
  41                         What memory is DMA'able?
  43The first piece of information you must know is what kernel memory can
  44be used with the DMA mapping facilities.  There has been an unwritten
  45set of rules regarding this, and this text is an attempt to finally
  46write them down.
  48If you acquired your memory via the page allocator
  49(i.e. __get_free_page*()) or the generic memory allocators
  50(i.e. kmalloc() or kmem_cache_alloc()) then you may DMA to/from
  51that memory using the addresses returned from those routines.
  53This means specifically that you may _not_ use the memory/addresses
  54returned from vmalloc() for DMA.  It is possible to DMA to the
  55_underlying_ memory mapped into a vmalloc() area, but this requires
  56walking page tables to get the physical addresses, and then
  57translating each of those pages back to a kernel address using
  58something like __va().  [ EDIT: Update this when we integrate
  59Gerd Knorr's generic code which does this. ]
  61This rule also means that you may use neither kernel image addresses
  62(items in data/text/bss segments), nor module image addresses, nor
  63stack addresses for DMA.  These could all be mapped somewhere entirely
  64different than the rest of physical memory.  Even if those classes of
  65memory could physically work with DMA, you'd need to ensure the I/O
  66buffers were cacheline-aligned.  Without that, you'd see cacheline
  67sharing problems (data corruption) on CPUs with DMA-incoherent caches.
  68(The CPU could write to one word, DMA would write to a different one
  69in the same cache line, and one of them could be overwritten.)
  71Also, this means that you cannot take the return of a kmap()
  72call and DMA to/from that.  This is similar to vmalloc().
  74What about block I/O and networking buffers?  The block I/O and
  75networking subsystems make sure that the buffers they use are valid
  76for you to DMA from/to.
  78                        DMA addressing limitations
  80Does your device have any DMA addressing limitations?  For example, is
  81your device only capable of driving the low order 24-bits of address
  82on the PCI bus for SAC DMA transfers?  If so, you need to inform the
  83PCI layer of this fact.
  85By default, the kernel assumes that your device can address the full
  8632-bits in a SAC cycle.  For a 64-bit DAC capable device, this needs
  87to be increased.  And for a device with limitations, as discussed in
  88the previous paragraph, it needs to be decreased.
  90pci_alloc_consistent() by default will return 32-bit DMA addresses.
  91PCI-X specification requires PCI-X devices to support 64-bit
  92addressing (DAC) for all transactions. And at least one platform (SGI
  93SN2) requires 64-bit consistent allocations to operate correctly when
  94the IO bus is in PCI-X mode. Therefore, like with pci_set_dma_mask(),
  95it's good practice to call pci_set_consistent_dma_mask() to set the
  96appropriate mask even if your device only supports 32-bit DMA
  97(default) and especially if it's a PCI-X device.
  99For correct operation, you must interrogate the PCI layer in your
 100device probe routine to see if the PCI controller on the machine can
 101properly support the DMA addressing limitation your device has.  It is
 102good style to do this even if your device holds the default setting,
 103because this shows that you did think about these issues wrt. your
 106The query is performed via a call to pci_set_dma_mask():
 108        int pci_set_dma_mask(struct pci_dev *pdev, u64 device_mask);
 110The query for consistent allocations is performed via a call to
 113        int pci_set_consistent_dma_mask(struct pci_dev *pdev, u64 device_mask);
 115Here, pdev is a pointer to the PCI device struct of your device, and
 116device_mask is a bit mask describing which bits of a PCI address your
 117device supports.  It returns zero if your card can perform DMA
 118properly on the machine given the address mask you provided.
 120If it returns non-zero, your device cannot perform DMA properly on
 121this platform, and attempting to do so will result in undefined
 122behavior.  You must either use a different mask, or not use DMA.
 124This means that in the failure case, you have three options:
 1261) Use another DMA mask, if possible (see below).
 1272) Use some non-DMA mode for data transfer, if possible.
 1283) Ignore this device and do not initialize it.
 130It is recommended that your driver print a kernel KERN_WARNING message
 131when you end up performing either #2 or #3.  In this manner, if a user
 132of your driver reports that performance is bad or that the device is not
 133even detected, you can ask them for the kernel messages to find out
 134exactly why.
 136The standard 32-bit addressing PCI device would do something like
 139        if (pci_set_dma_mask(pdev, DMA_BIT_MASK(32))) {
 140                printk(KERN_WARNING
 141                       "mydev: No suitable DMA available.\n");
 142                goto ignore_this_device;
 143        }
 145Another common scenario is a 64-bit capable device.  The approach
 146here is to try for 64-bit DAC addressing, but back down to a
 14732-bit mask should that fail.  The PCI platform code may fail the
 14864-bit mask not because the platform is not capable of 64-bit
 149addressing.  Rather, it may fail in this case simply because
 15032-bit SAC addressing is done more efficiently than DAC addressing.
 151Sparc64 is one platform which behaves in this way.
 153Here is how you would handle a 64-bit capable device which can drive
 154all 64-bits when accessing streaming DMA:
 156        int using_dac;
 158        if (!pci_set_dma_mask(pdev, DMA_BIT_MASK(64))) {
 159                using_dac = 1;
 160        } else if (!pci_set_dma_mask(pdev, DMA_BIT_MASK(32))) {
 161                using_dac = 0;
 162        } else {
 163                printk(KERN_WARNING
 164                       "mydev: No suitable DMA available.\n");
 165                goto ignore_this_device;
 166        }
 168If a card is capable of using 64-bit consistent allocations as well,
 169the case would look like this:
 171        int using_dac, consistent_using_dac;
 173        if (!pci_set_dma_mask(pdev, DMA_BIT_MASK(64))) {
 174                using_dac = 1;
 175                consistent_using_dac = 1;
 176                pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64));
 177        } else if (!pci_set_dma_mask(pdev, DMA_BIT_MASK(32))) {
 178                using_dac = 0;
 179                consistent_using_dac = 0;
 180                pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32));
 181        } else {
 182                printk(KERN_WARNING
 183                       "mydev: No suitable DMA available.\n");
 184                goto ignore_this_device;
 185        }
 187pci_set_consistent_dma_mask() will always be able to set the same or a
 188smaller mask as pci_set_dma_mask(). However for the rare case that a
 189device driver only uses consistent allocations, one would have to
 190check the return value from pci_set_consistent_dma_mask().
 192Finally, if your device can only drive the low 24-bits of
 193address during PCI bus mastering you might do something like:
 195        if (pci_set_dma_mask(pdev, DMA_BIT_MASK(24))) {
 196                printk(KERN_WARNING
 197                       "mydev: 24-bit DMA addressing not available.\n");
 198                goto ignore_this_device;
 199        }
 201When pci_set_dma_mask() is successful, and returns zero, the PCI layer
 202saves away this mask you have provided.  The PCI layer will use this
 203information later when you make DMA mappings.
 205There is a case which we are aware of at this time, which is worth
 206mentioning in this documentation.  If your device supports multiple
 207functions (for example a sound card provides playback and record
 208functions) and the various different functions have _different_
 209DMA addressing limitations, you may wish to probe each mask and
 210only provide the functionality which the machine can handle.  It
 211is important that the last call to pci_set_dma_mask() be for the
 212most specific mask.
 214Here is pseudo-code showing how this might be done:
 216        #define PLAYBACK_ADDRESS_BITS   DMA_BIT_MASK(32)
 217        #define RECORD_ADDRESS_BITS     0x00ffffff
 219        struct my_sound_card *card;
 220        struct pci_dev *pdev;
 222        ...
 223        if (!pci_set_dma_mask(pdev, PLAYBACK_ADDRESS_BITS)) {
 224                card->playback_enabled = 1;
 225        } else {
 226                card->playback_enabled = 0;
 227                printk(KERN_WARN "%s: Playback disabled due to DMA limitations.\n",
 228                       card->name);
 229        }
 230        if (!pci_set_dma_mask(pdev, RECORD_ADDRESS_BITS)) {
 231                card->record_enabled = 1;
 232        } else {
 233                card->record_enabled = 0;
 234                printk(KERN_WARN "%s: Record disabled due to DMA limitations.\n",
 235                       card->name);
 236        }
 238A sound card was used as an example here because this genre of PCI
 239devices seems to be littered with ISA chips given a PCI front end,
 240and thus retaining the 16MB DMA addressing limitations of ISA.
 242                        Types of DMA mappings
 244There are two types of DMA mappings:
 246- Consistent DMA mappings which are usually mapped at driver
 247  initialization, unmapped at the end and for which the hardware should
 248  guarantee that the device and the CPU can access the data
 249  in parallel and will see updates made by each other without any
 250  explicit software flushing.
 252  Think of "consistent" as "synchronous" or "coherent".
 254  The current default is to return consistent memory in the low 32
 255  bits of the PCI bus space.  However, for future compatibility you
 256  should set the consistent mask even if this default is fine for your
 257  driver.
 259  Good examples of what to use consistent mappings for are:
 261        - Network card DMA ring descriptors.
 262        - SCSI adapter mailbox command data structures.
 263        - Device firmware microcode executed out of
 264          main memory.
 266  The invariant these examples all require is that any CPU store
 267  to memory is immediately visible to the device, and vice
 268  versa.  Consistent mappings guarantee this.
 270  IMPORTANT: Consistent DMA memory does not preclude the usage of
 271             proper memory barriers.  The CPU may reorder stores to
 272             consistent memory just as it may normal memory.  Example:
 273             if it is important for the device to see the first word
 274             of a descriptor updated before the second, you must do
 275             something like:
 277                desc->word0 = address;
 278                wmb();
 279                desc->word1 = DESC_VALID;
 281             in order to get correct behavior on all platforms.
 283             Also, on some platforms your driver may need to flush CPU write
 284             buffers in much the same way as it needs to flush write buffers
 285             found in PCI bridges (such as by reading a register's value
 286             after writing it).
 288- Streaming DMA mappings which are usually mapped for one DMA transfer,
 289  unmapped right after it (unless you use pci_dma_sync_* below) and for which
 290  hardware can optimize for sequential accesses.
 292  This of "streaming" as "asynchronous" or "outside the coherency
 293  domain".
 295  Good examples of what to use streaming mappings for are:
 297        - Networking buffers transmitted/received by a device.
 298        - Filesystem buffers written/read by a SCSI device.
 300  The interfaces for using this type of mapping were designed in
 301  such a way that an implementation can make whatever performance
 302  optimizations the hardware allows.  To this end, when using
 303  such mappings you must be explicit about what you want to happen.
 305Neither type of DMA mapping has alignment restrictions that come
 306from PCI, although some devices may have such restrictions.
 307Also, systems with caches that aren't DMA-coherent will work better
 308when the underlying buffers don't share cache lines with other data.
 311                 Using Consistent DMA mappings.
 313To allocate and map large (PAGE_SIZE or so) consistent DMA regions,
 314you should do:
 316        dma_addr_t dma_handle;
 318        cpu_addr = pci_alloc_consistent(pdev, size, &dma_handle);
 320where pdev is a struct pci_dev *. This may be called in interrupt context.
 321You should use dma_alloc_coherent (see DMA-API.txt) for buses
 322where devices don't have struct pci_dev (like ISA, EISA).
 324This argument is needed because the DMA translations may be bus
 325specific (and often is private to the bus which the device is attached
 328Size is the length of the region you want to allocate, in bytes.
 330This routine will allocate RAM for that region, so it acts similarly to
 331__get_free_pages (but takes size instead of a page order).  If your
 332driver needs regions sized smaller than a page, you may prefer using
 333the pci_pool interface, described below.
 335The consistent DMA mapping interfaces, for non-NULL pdev, will by
 336default return a DMA address which is SAC (Single Address Cycle)
 337addressable.  Even if the device indicates (via PCI dma mask) that it
 338may address the upper 32-bits and thus perform DAC cycles, consistent
 339allocation will only return > 32-bit PCI addresses for DMA if the
 340consistent dma mask has been explicitly changed via
 341pci_set_consistent_dma_mask().  This is true of the pci_pool interface
 342as well.
 344pci_alloc_consistent returns two values: the virtual address which you
 345can use to access it from the CPU and dma_handle which you pass to the
 348The cpu return address and the DMA bus master address are both
 349guaranteed to be aligned to the smallest PAGE_SIZE order which
 350is greater than or equal to the requested size.  This invariant
 351exists (for example) to guarantee that if you allocate a chunk
 352which is smaller than or equal to 64 kilobytes, the extent of the
 353buffer you receive will not cross a 64K boundary.
 355To unmap and free such a DMA region, you call:
 357        pci_free_consistent(pdev, size, cpu_addr, dma_handle);
 359where pdev, size are the same as in the above call and cpu_addr and
 360dma_handle are the values pci_alloc_consistent returned to you.
 361This function may not be called in interrupt context.
 363If your driver needs lots of smaller memory regions, you can write
 364custom code to subdivide pages returned by pci_alloc_consistent,
 365or you can use the pci_pool API to do that.  A pci_pool is like
 366a kmem_cache, but it uses pci_alloc_consistent not __get_free_pages.
 367Also, it understands common hardware constraints for alignment,
 368like queue heads needing to be aligned on N byte boundaries.
 370Create a pci_pool like this:
 372        struct pci_pool *pool;
 374        pool = pci_pool_create(name, pdev, size, align, alloc);
 376The "name" is for diagnostics (like a kmem_cache name); pdev and size
 377are as above.  The device's hardware alignment requirement for this
 378type of data is "align" (which is expressed in bytes, and must be a
 379power of two).  If your device has no boundary crossing restrictions,
 380pass 0 for alloc; passing 4096 says memory allocated from this pool
 381must not cross 4KByte boundaries (but at that time it may be better to
 382go for pci_alloc_consistent directly instead).
 384Allocate memory from a pci pool like this:
 386        cpu_addr = pci_pool_alloc(pool, flags, &dma_handle);
 388flags are SLAB_KERNEL if blocking is permitted (not in_interrupt nor
 389holding SMP locks), SLAB_ATOMIC otherwise.  Like pci_alloc_consistent,
 390this returns two values, cpu_addr and dma_handle.
 392Free memory that was allocated from a pci_pool like this:
 394        pci_pool_free(pool, cpu_addr, dma_handle);
 396where pool is what you passed to pci_pool_alloc, and cpu_addr and
 397dma_handle are the values pci_pool_alloc returned. This function
 398may be called in interrupt context.
 400Destroy a pci_pool by calling:
 402        pci_pool_destroy(pool);
 404Make sure you've called pci_pool_free for all memory allocated
 405from a pool before you destroy the pool. This function may not
 406be called in interrupt context.
 408                        DMA Direction
 410The interfaces described in subsequent portions of this document
 411take a DMA direction argument, which is an integer and takes on
 412one of the following values:
 419One should provide the exact DMA direction if you know it.
 421PCI_DMA_TODEVICE means "from main memory to the PCI device"
 422PCI_DMA_FROMDEVICE means "from the PCI device to main memory"
 423It is the direction in which the data moves during the DMA
 426You are _strongly_ encouraged to specify this as precisely
 427as you possibly can.
 429If you absolutely cannot know the direction of the DMA transfer,
 430specify PCI_DMA_BIDIRECTIONAL.  It means that the DMA can go in
 431either direction.  The platform guarantees that you may legally
 432specify this, and that it will work, but this may be at the
 433cost of performance for example.
 435The value PCI_DMA_NONE is to be used for debugging.  One can
 436hold this in a data structure before you come to know the
 437precise direction, and this will help catch cases where your
 438direction tracking logic has failed to set things up properly.
 440Another advantage of specifying this value precisely (outside of
 441potential platform-specific optimizations of such) is for debugging.
 442Some platforms actually have a write permission boolean which DMA
 443mappings can be marked with, much like page protections in the user
 444program address space.  Such platforms can and do report errors in the
 445kernel logs when the PCI controller hardware detects violation of the
 446permission setting.
 448Only streaming mappings specify a direction, consistent mappings
 449implicitly have a direction attribute setting of
 452The SCSI subsystem tells you the direction to use in the
 453'sc_data_direction' member of the SCSI command your driver is
 454working on.
 456For Networking drivers, it's a rather simple affair.  For transmit
 457packets, map/unmap them with the PCI_DMA_TODEVICE direction
 458specifier.  For receive packets, just the opposite, map/unmap them
 459with the PCI_DMA_FROMDEVICE direction specifier.
 461                  Using Streaming DMA mappings
 463The streaming DMA mapping routines can be called from interrupt
 464context.  There are two versions of each map/unmap, one which will
 465map/unmap a single memory region, and one which will map/unmap a
 468To map a single region, you do:
 470        struct pci_dev *pdev = mydev->pdev;
 471        dma_addr_t dma_handle;
 472        void *addr = buffer->ptr;
 473        size_t size = buffer->len;
 475        dma_handle = pci_map_single(pdev, addr, size, direction);
 477and to unmap it:
 479        pci_unmap_single(pdev, dma_handle, size, direction);
 481You should call pci_unmap_single when the DMA activity is finished, e.g.
 482from the interrupt which told you that the DMA transfer is done.
 484Using cpu pointers like this for single mappings has a disadvantage,
 485you cannot reference HIGHMEM memory in this way.  Thus, there is a
 486map/unmap interface pair akin to pci_{map,unmap}_single.  These
 487interfaces deal with page/offset pairs instead of cpu pointers.
 490        struct pci_dev *pdev = mydev->pdev;
 491        dma_addr_t dma_handle;
 492        struct page *page = buffer->page;
 493        unsigned long offset = buffer->offset;
 494        size_t size = buffer->len;
 496        dma_handle = pci_map_page(pdev, page, offset, size, direction);
 498        ...
 500        pci_unmap_page(pdev, dma_handle, size, direction);
 502Here, "offset" means byte offset within the given page.
 504With scatterlists, you map a region gathered from several regions by:
 506        int i, count = pci_map_sg(pdev, sglist, nents, direction);
 507        struct scatterlist *sg;
 509        for_each_sg(sglist, sg, count, i) {
 510                hw_address[i] = sg_dma_address(sg);
 511                hw_len[i] = sg_dma_len(sg);
 512        }
 514where nents is the number of entries in the sglist.
 516The implementation is free to merge several consecutive sglist entries
 517into one (e.g. if DMA mapping is done with PAGE_SIZE granularity, any
 518consecutive sglist entries can be merged into one provided the first one
 519ends and the second one starts on a page boundary - in fact this is a huge
 520advantage for cards which either cannot do scatter-gather or have very
 521limited number of scatter-gather entries) and returns the actual number
 522of sg entries it mapped them to. On failure 0 is returned.
 524Then you should loop count times (note: this can be less than nents times)
 525and use sg_dma_address() and sg_dma_len() macros where you previously
 526accessed sg->address and sg->length as shown above.
 528To unmap a scatterlist, just call:
 530        pci_unmap_sg(pdev, sglist, nents, direction);
 532Again, make sure DMA activity has already finished.
 534PLEASE NOTE:  The 'nents' argument to the pci_unmap_sg call must be
 535              the _same_ one you passed into the pci_map_sg call,
 536              it should _NOT_ be the 'count' value _returned_ from the
 537              pci_map_sg call.
 539Every pci_map_{single,sg} call should have its pci_unmap_{single,sg}
 540counterpart, because the bus address space is a shared resource (although
 541in some ports the mapping is per each BUS so less devices contend for the
 542same bus address space) and you could render the machine unusable by eating
 543all bus addresses.
 545If you need to use the same streaming DMA region multiple times and touch
 546the data in between the DMA transfers, the buffer needs to be synced
 547properly in order for the cpu and device to see the most uptodate and
 548correct copy of the DMA buffer.
 550So, firstly, just map it with pci_map_{single,sg}, and after each DMA
 551transfer call either:
 553        pci_dma_sync_single_for_cpu(pdev, dma_handle, size, direction);
 557        pci_dma_sync_sg_for_cpu(pdev, sglist, nents, direction);
 559as appropriate.
 561Then, if you wish to let the device get at the DMA area again,
 562finish accessing the data with the cpu, and then before actually
 563giving the buffer to the hardware call either:
 565        pci_dma_sync_single_for_device(pdev, dma_handle, size, direction);
 569        pci_dma_sync_sg_for_device(dev, sglist, nents, direction);
 571as appropriate.
 573After the last DMA transfer call one of the DMA unmap routines
 574pci_unmap_{single,sg}. If you don't touch the data from the first pci_map_*
 575call till pci_unmap_*, then you don't have to call the pci_dma_sync_*
 576routines at all.
 578Here is pseudo code which shows a situation in which you would need
 579to use the pci_dma_sync_*() interfaces.
 581        my_card_setup_receive_buffer(struct my_card *cp, char *buffer, int len)
 582        {
 583                dma_addr_t mapping;
 585                mapping = pci_map_single(cp->pdev, buffer, len, PCI_DMA_FROMDEVICE);
 587                cp->rx_buf = buffer;
 588                cp->rx_len = len;
 589                cp->rx_dma = mapping;
 591                give_rx_buf_to_card(cp);
 592        }
 594        ...
 596        my_card_interrupt_handler(int irq, void *devid, struct pt_regs *regs)
 597        {
 598                struct my_card *cp = devid;
 600                ...
 601                if (read_card_status(cp) == RX_BUF_TRANSFERRED) {
 602                        struct my_card_header *hp;
 604                        /* Examine the header to see if we wish
 605                         * to accept the data.  But synchronize
 606                         * the DMA transfer with the CPU first
 607                         * so that we see updated contents.
 608                         */
 609                        pci_dma_sync_single_for_cpu(cp->pdev, cp->rx_dma,
 610                                                    cp->rx_len,
 611                                                    PCI_DMA_FROMDEVICE);
 613                        /* Now it is safe to examine the buffer. */
 614                        hp = (struct my_card_header *) cp->rx_buf;
 615                        if (header_is_ok(hp)) {
 616                                pci_unmap_single(cp->pdev, cp->rx_dma, cp->rx_len,
 617                                                 PCI_DMA_FROMDEVICE);
 618                                pass_to_upper_layers(cp->rx_buf);
 619                                make_and_setup_new_rx_buf(cp);
 620                        } else {
 621                                /* Just sync the buffer and give it back
 622                                 * to the card.
 623                                 */
 624                                pci_dma_sync_single_for_device(cp->pdev,
 625                                                               cp->rx_dma,
 626                                                               cp->rx_len,
 627                                                               PCI_DMA_FROMDEVICE);
 628                                give_rx_buf_to_card(cp);
 629                        }
 630                }
 631        }
 633Drivers converted fully to this interface should not use virt_to_bus any
 634longer, nor should they use bus_to_virt. Some drivers have to be changed a
 635little bit, because there is no longer an equivalent to bus_to_virt in the
 636dynamic DMA mapping scheme - you have to always store the DMA addresses
 637returned by the pci_alloc_consistent, pci_pool_alloc, and pci_map_single
 638calls (pci_map_sg stores them in the scatterlist itself if the platform
 639supports dynamic DMA mapping in hardware) in your driver structures and/or
 640in the card registers.
 642All PCI drivers should be using these interfaces with no exceptions.
 643It is planned to completely remove virt_to_bus() and bus_to_virt() as
 644they are entirely deprecated.  Some ports already do not provide these
 645as it is impossible to correctly support them.
 647                Optimizing Unmap State Space Consumption
 649On many platforms, pci_unmap_{single,page}() is simply a nop.
 650Therefore, keeping track of the mapping address and length is a waste
 651of space.  Instead of filling your drivers up with ifdefs and the like
 652to "work around" this (which would defeat the whole purpose of a
 653portable API) the following facilities are provided.
 655Actually, instead of describing the macros one by one, we'll
 656transform some example code.
 6581) Use DECLARE_PCI_UNMAP_{ADDR,LEN} in state saving structures.
 659   Example, before:
 661        struct ring_state {
 662                struct sk_buff *skb;
 663                dma_addr_t mapping;
 664                __u32 len;
 665        };
 667   after:
 669        struct ring_state {
 670                struct sk_buff *skb;
 671                DECLARE_PCI_UNMAP_ADDR(mapping)
 672                DECLARE_PCI_UNMAP_LEN(len)
 673        };
 675   NOTE: DO NOT put a semicolon at the end of the DECLARE_*()
 676         macro.
 6782) Use pci_unmap_{addr,len}_set to set these values.
 679   Example, before:
 681        ringp->mapping = FOO;
 682        ringp->len = BAR;
 684   after:
 686        pci_unmap_addr_set(ringp, mapping, FOO);
 687        pci_unmap_len_set(ringp, len, BAR);
 6893) Use pci_unmap_{addr,len} to access these values.
 690   Example, before:
 692        pci_unmap_single(pdev, ringp->mapping, ringp->len,
 693                         PCI_DMA_FROMDEVICE);
 695   after:
 697        pci_unmap_single(pdev,
 698                         pci_unmap_addr(ringp, mapping),
 699                         pci_unmap_len(ringp, len),
 700                         PCI_DMA_FROMDEVICE);
 702It really should be self-explanatory.  We treat the ADDR and LEN
 703separately, because it is possible for an implementation to only
 704need the address in order to perform the unmap operation.
 706                        Platform Issues
 708If you are just writing drivers for Linux and do not maintain
 709an architecture port for the kernel, you can safely skip down
 710to "Closing".
 7121) Struct scatterlist requirements.
 714   Struct scatterlist must contain, at a minimum, the following
 715   members:
 717        struct page *page;
 718        unsigned int offset;
 719        unsigned int length;
 721   The base address is specified by a "page+offset" pair.
 723   Previous versions of struct scatterlist contained a "void *address"
 724   field that was sometimes used instead of page+offset.  As of Linux
 725   2.5., page+offset is always used, and the "address" field has been
 726   deleted.
 7282) More to come...
 730                        Handling Errors
 732DMA address space is limited on some architectures and an allocation
 733failure can be determined by:
 735- checking if pci_alloc_consistent returns NULL or pci_map_sg returns 0
 737- checking the returned dma_addr_t of pci_map_single and pci_map_page
 738  by using pci_dma_mapping_error():
 740        dma_addr_t dma_handle;
 742        dma_handle = pci_map_single(pdev, addr, size, direction);
 743        if (pci_dma_mapping_error(pdev, dma_handle)) {
 744                /*
 745                 * reduce current DMA mapping usage,
 746                 * delay and try again later or
 747                 * reset driver.
 748                 */
 749        }
 751                           Closing
 753This document, and the API itself, would not be in it's current
 754form without the feedback and suggestions from numerous individuals.
 755We would like to specifically mention, in no particular order, the
 756following people:
 758        Russell King <>
 759        Leo Dagum <>
 760        Ralf Baechle <>
 761        Grant Grundler <>
 762        Jay Estabrook <>
 763        Thomas Sailer <>
 764        Andrea Arcangeli <>
 765        Jens Axboe <>
 766        David Mosberger-Tang <>