blob: bccf602a87f5c2054b146c826725ac183863a8ad [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001
Grant Grundler74da15e2006-12-25 01:06:35 -07002 How To Write Linux PCI Drivers
3
4 by Martin Mares <mj@ucw.cz> on 07-Feb-2000
5 updated by Grant Grundler <grundler@parisc-linux.org> on 23-Dec-2006
Linus Torvalds1da177e2005-04-16 15:20:36 -07006
7~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -07008The world of PCI is vast and full of (mostly unpleasant) surprises.
9Since each CPU architecture implements different chip-sets and PCI devices
10have different requirements (erm, "features"), the result is the PCI support
11in the Linux kernel is not as trivial as one would wish. This short paper
12tries to introduce all potential driver authors to Linux APIs for
13PCI device drivers.
14
15A more complete resource is the third edition of "Linux Device Drivers"
16by Jonathan Corbet, Alessandro Rubini, and Greg Kroah-Hartman.
17LDD3 is available for free (under Creative Commons License) from:
18
19 http://lwn.net/Kernel/LDD3/
20
21However, keep in mind that all documents are subject to "bit rot".
22Refer to the source code if things are not working as described here.
23
24Please send questions/comments/patches about Linux PCI API to the
25"Linux PCI" <linux-pci@atrey.karlin.mff.cuni.cz> mailing list.
26
Linus Torvalds1da177e2005-04-16 15:20:36 -070027
28
290. Structure of PCI drivers
30~~~~~~~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -070031PCI drivers "discover" PCI devices in a system via pci_register_driver().
32Actually, it's the other way around. When the PCI generic code discovers
33a new device, the driver with a matching "description" will be notified.
34Details on this below.
35
36pci_register_driver() leaves most of the probing for devices to
37the PCI layer and supports online insertion/removal of devices [thus
38supporting hot-pluggable PCI, CardBus, and Express-Card in a single driver].
39pci_register_driver() call requires passing in a table of function
40pointers and thus dictates the high level structure of a driver.
41
42Once the driver knows about a PCI device and takes ownership, the
43driver generally needs to perform the following initialization:
Linus Torvalds1da177e2005-04-16 15:20:36 -070044
45 Enable the device
Grant Grundler74da15e2006-12-25 01:06:35 -070046 Request MMIO/IOP resources
47 Set the DMA mask size (for both coherent and streaming DMA)
48 Allocate and initialize shared control data (pci_allocate_coherent())
49 Access device configuration space (if needed)
50 Register IRQ handler (request_irq())
51 Initialize non-PCI (i.e. LAN/SCSI/etc parts of the chip)
52 Enable DMA/processing engines
53
54When done using the device, and perhaps the module needs to be unloaded,
55the driver needs to take the follow steps:
56 Disable the device from generating IRQs
57 Release the IRQ (free_irq())
58 Stop all DMA activity
59 Release DMA buffers (both streaming and coherent)
60 Unregister from other subsystems (e.g. scsi or netdev)
61 Release MMIO/IOP resources
Linus Torvalds1da177e2005-04-16 15:20:36 -070062 Disable the device
63
Grant Grundler74da15e2006-12-25 01:06:35 -070064Most of these topics are covered in the following sections.
65For the rest look at LDD3 or <linux/pci.h> .
Linus Torvalds1da177e2005-04-16 15:20:36 -070066
67If the PCI subsystem is not configured (CONFIG_PCI is not set), most of
Grant Grundler74da15e2006-12-25 01:06:35 -070068the PCI functions described below are defined as inline functions either
69completely empty or just returning an appropriate error codes to avoid
70lots of ifdefs in the drivers.
Linus Torvalds1da177e2005-04-16 15:20:36 -070071
72
Linus Torvalds1da177e2005-04-16 15:20:36 -070073
Grant Grundler74da15e2006-12-25 01:06:35 -0700741. pci_register_driver() call
75~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
76
77PCI device drivers call pci_register_driver() during their
78initialization with a pointer to a structure describing the driver
79(struct pci_driver):
80
81 field name Description
82 ---------- ------------------------------------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -070083 id_table Pointer to table of device ID's the driver is
84 interested in. Most drivers should export this
85 table using MODULE_DEVICE_TABLE(pci,...).
Grant Grundler74da15e2006-12-25 01:06:35 -070086
87 probe This probing function gets called (during execution
88 of pci_register_driver() for already existing
89 devices or later if a new device gets inserted) for
90 all PCI devices which match the ID table and are not
91 "owned" by the other drivers yet. This function gets
92 passed a "struct pci_dev *" for each device whose
93 entry in the ID table matches the device. The probe
94 function returns zero when the driver chooses to
95 take "ownership" of the device or an error code
96 (negative number) otherwise.
97 The probe function always gets called from process
98 context, so it can sleep.
99
100 remove The remove() function gets called whenever a device
101 being handled by this driver is removed (either during
102 deregistration of the driver or when it's manually
103 pulled out of a hot-pluggable slot).
104 The remove function always gets called from process
105 context, so it can sleep.
106
Linus Torvalds1da177e2005-04-16 15:20:36 -0700107 suspend Put device into low power state.
Grant Grundler74da15e2006-12-25 01:06:35 -0700108 suspend_late Put device into low power state.
109
110 resume_early Wake device from low power state.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700111 resume Wake device from low power state.
Grant Grundler74da15e2006-12-25 01:06:35 -0700112
113 (Please see Documentation/power/pci.txt for descriptions
114 of PCI Power Management and the related functions.)
115
Grant Grundler74da15e2006-12-25 01:06:35 -0700116 shutdown Hook into reboot_notifier_list (kernel/sys.c).
117 Intended to stop any idling DMA operations.
118 Useful for enabling wake-on-lan (NIC) or changing
119 the power state of a device before reboot.
120 e.g. drivers/net/e100.c.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700121
Randy Dunlap4b5ff462008-03-10 17:16:32 -0700122 err_handler See Documentation/PCI/pci-error-recovery.txt
Linus Torvalds1da177e2005-04-16 15:20:36 -0700123
Grant Grundler74da15e2006-12-25 01:06:35 -0700124
125The ID table is an array of struct pci_device_id entries ending with an
Andrew Morton9f9351b2008-03-10 11:43:34 -0700126all-zero entry; use of the macro DEFINE_PCI_DEVICE_TABLE is the preferred
Jonas Bonn90a1ba02008-02-22 11:02:21 +0100127method of declaring the table. Each entry consists of:
Grant Grundler74da15e2006-12-25 01:06:35 -0700128
129 vendor,device Vendor and device ID to match (or PCI_ANY_ID)
130
Linus Torvalds1da177e2005-04-16 15:20:36 -0700131 subvendor, Subsystem vendor and device ID to match (or PCI_ANY_ID)
Grant Grundler74da15e2006-12-25 01:06:35 -0700132 subdevice,
133
134 class Device class, subclass, and "interface" to match.
135 See Appendix D of the PCI Local Bus Spec or
136 include/linux/pci_ids.h for a full list of classes.
137 Most drivers do not need to specify class/class_mask
138 as vendor/device is normally sufficient.
139
140 class_mask limit which sub-fields of the class field are compared.
141 See drivers/scsi/sym53c8xx_2/ for example of usage.
142
Linus Torvalds1da177e2005-04-16 15:20:36 -0700143 driver_data Data private to the driver.
Grant Grundler74da15e2006-12-25 01:06:35 -0700144 Most drivers don't need to use driver_data field.
145 Best practice is to use driver_data as an index
146 into a static list of equivalent device types,
147 instead of using it as a pointer.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700148
Linus Torvalds1da177e2005-04-16 15:20:36 -0700149
Grant Grundler74da15e2006-12-25 01:06:35 -0700150Most drivers only need PCI_DEVICE() or PCI_DEVICE_CLASS() to set up
151a pci_device_id table.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700152
Grant Grundler74da15e2006-12-25 01:06:35 -0700153New PCI IDs may be added to a device driver pci_ids table at runtime
154as shown below:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700155
156echo "vendor device subvendor subdevice class class_mask driver_data" > \
Grant Grundler74da15e2006-12-25 01:06:35 -0700157/sys/bus/pci/drivers/{driver}/new_id
158
159All fields are passed in as hexadecimal values (no leading 0x).
Jean Delvare6ba18632007-04-07 17:21:28 +0200160The vendor and device fields are mandatory, the others are optional. Users
161need pass only as many optional fields as necessary:
162 o subvendor and subdevice fields default to PCI_ANY_ID (FFFFFFFF)
Grant Grundler74da15e2006-12-25 01:06:35 -0700163 o class and classmask fields default to 0
164 o driver_data defaults to 0UL.
165
Jean Delvareb41d6cf2008-08-17 21:06:59 +0200166Note that driver_data must match the value used by any of the pci_device_id
167entries defined in the driver. This makes the driver_data field mandatory
168if all the pci_device_id entries have a non-zero driver_data value.
169
Grant Grundler74da15e2006-12-25 01:06:35 -0700170Once added, the driver probe routine will be invoked for any unclaimed
171PCI devices listed in its (newly updated) pci_ids list.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700172
173When the driver exits, it just calls pci_unregister_driver() and the PCI layer
174automatically calls the remove hook for all devices handled by the driver.
175
Grant Grundler74da15e2006-12-25 01:06:35 -0700176
1771.1 "Attributes" for driver functions/data
178
Linus Torvalds1da177e2005-04-16 15:20:36 -0700179Please mark the initialization and cleanup functions where appropriate
180(the corresponding macros are defined in <linux/init.h>):
181
182 __init Initialization code. Thrown away after the driver
183 initializes.
184 __exit Exit code. Ignored for non-modular drivers.
Grant Grundler74da15e2006-12-25 01:06:35 -0700185
Grant Grundler74da15e2006-12-25 01:06:35 -0700186Tips on when/where to use the above attributes:
187 o The module_init()/module_exit() functions (and all
188 initialization functions called _only_ from these)
189 should be marked __init/__exit.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700190
Grant Grundler74da15e2006-12-25 01:06:35 -0700191 o Do not mark the struct pci_driver.
192
Grant Grundler74da15e2006-12-25 01:06:35 -0700193 o Do NOT mark a function if you are not sure which mark to use.
194 Better to not mark the function than mark the function wrong.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700195
196
Grant Grundler74da15e2006-12-25 01:06:35 -0700197
1982. How to find PCI devices manually
199~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
200
201PCI drivers should have a really good reason for not using the
202pci_register_driver() interface to search for PCI devices.
203The main reason PCI devices are controlled by multiple drivers
204is because one PCI device implements several different HW services.
205E.g. combined serial/parallel port/floppy controller.
206
207A manual search may be performed using the following constructs:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700208
209Searching by vendor and device ID:
210
211 struct pci_dev *dev = NULL;
212 while (dev = pci_get_device(VENDOR_ID, DEVICE_ID, dev))
213 configure_device(dev);
214
215Searching by class ID (iterate in a similar way):
216
217 pci_get_class(CLASS_ID, dev)
218
219Searching by both vendor/device and subsystem vendor/device ID:
220
Grant Grundler74da15e2006-12-25 01:06:35 -0700221 pci_get_subsys(VENDOR_ID,DEVICE_ID, SUBSYS_VENDOR_ID, SUBSYS_DEVICE_ID, dev).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222
Grant Grundler74da15e2006-12-25 01:06:35 -0700223You can use the constant PCI_ANY_ID as a wildcard replacement for
Linus Torvalds1da177e2005-04-16 15:20:36 -0700224VENDOR_ID or DEVICE_ID. This allows searching for any device from a
225specific vendor, for example.
226
Grant Grundler74da15e2006-12-25 01:06:35 -0700227These functions are hotplug-safe. They increment the reference count on
Linus Torvalds1da177e2005-04-16 15:20:36 -0700228the pci_dev that they return. You must eventually (possibly at module unload)
229decrement the reference count on these devices by calling pci_dev_put().
230
231
Linus Torvalds1da177e2005-04-16 15:20:36 -0700232
Grant Grundler74da15e2006-12-25 01:06:35 -07002333. Device Initialization Steps
234~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Linus Torvalds1da177e2005-04-16 15:20:36 -0700235
Grant Grundler74da15e2006-12-25 01:06:35 -0700236As noted in the introduction, most PCI drivers need the following steps
237for device initialization:
238
239 Enable the device
240 Request MMIO/IOP resources
241 Set the DMA mask size (for both coherent and streaming DMA)
242 Allocate and initialize shared control data (pci_allocate_coherent())
243 Access device configuration space (if needed)
244 Register IRQ handler (request_irq())
245 Initialize non-PCI (i.e. LAN/SCSI/etc parts of the chip)
246 Enable DMA/processing engines.
247
248The driver can access PCI config space registers at any time.
249(Well, almost. When running BIST, config space can go away...but
250that will just result in a PCI Bus Master Abort and config reads
251will return garbage).
252
253
2543.1 Enable the PCI device
255~~~~~~~~~~~~~~~~~~~~~~~~~
256Before touching any device registers, the driver needs to enable
257the PCI device by calling pci_enable_device(). This will:
258 o wake up the device if it was in suspended state,
259 o allocate I/O and memory regions of the device (if BIOS did not),
260 o allocate an IRQ (if BIOS did not).
261
262NOTE: pci_enable_device() can fail! Check the return value.
Grant Grundler74da15e2006-12-25 01:06:35 -0700263
264[ OS BUG: we don't check resource allocations before enabling those
265 resources. The sequence would make more sense if we called
266 pci_request_resources() before calling pci_enable_device().
267 Currently, the device drivers can't detect the bug when when two
268 devices have been allocated the same range. This is not a common
269 problem and unlikely to get fixed soon.
270
271 This has been discussed before but not changed as of 2.6.19:
272 http://lkml.org/lkml/2006/3/2/194
273]
274
275pci_set_master() will enable DMA by setting the bus master bit
276in the PCI_COMMAND register. It also fixes the latency timer value if
Ben Hutchings6a479072008-12-23 03:08:29 +0000277it's set to something bogus by the BIOS. pci_clear_master() will
278disable DMA by clearing the bus master bit.
Grant Grundler74da15e2006-12-25 01:06:35 -0700279
280If the PCI device can use the PCI Memory-Write-Invalidate transaction,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700281call pci_set_mwi(). This enables the PCI_COMMAND bit for Mem-Wr-Inval
282and also ensures that the cache line size register is set correctly.
Grant Grundler74da15e2006-12-25 01:06:35 -0700283Check the return value of pci_set_mwi() as not all architectures
Randy Dunlap694625c2007-07-09 11:55:54 -0700284or chip-sets may support Memory-Write-Invalidate. Alternatively,
285if Mem-Wr-Inval would be nice to have but is not required, call
286pci_try_set_mwi() to have the system do its best effort at enabling
287Mem-Wr-Inval.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700288
Grant Grundler74da15e2006-12-25 01:06:35 -0700289
2903.2 Request MMIO/IOP resources
291~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
292Memory (MMIO), and I/O port addresses should NOT be read directly
293from the PCI device config space. Use the values in the pci_dev structure
294as the PCI "bus address" might have been remapped to a "host physical"
295address by the arch/chip-set specific kernel support.
296
Paul Bolle395cf962011-08-15 02:02:26 +0200297See Documentation/io-mapping.txt for how to access device registers
Grant Grundler74da15e2006-12-25 01:06:35 -0700298or device memory.
299
300The device driver needs to call pci_request_region() to verify
301no other device is already using the same address resource.
302Conversely, drivers should call pci_release_region() AFTER
Linus Torvalds1da177e2005-04-16 15:20:36 -0700303calling pci_disable_device().
Grant Grundler74da15e2006-12-25 01:06:35 -0700304The idea is to prevent two devices colliding on the same address range.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700305
Grant Grundler74da15e2006-12-25 01:06:35 -0700306[ See OS BUG comment above. Currently (2.6.19), The driver can only
307 determine MMIO and IO Port resource availability _after_ calling
308 pci_enable_device(). ]
309
310Generic flavors of pci_request_region() are request_mem_region()
311(for MMIO ranges) and request_region() (for IO Port ranges).
312Use these for address resources that are not described by "normal" PCI
313BARs.
314
315Also see pci_request_selected_regions() below.
316
317
3183.3 Set the DMA mask size
319~~~~~~~~~~~~~~~~~~~~~~~~~
320[ If anything below doesn't make sense, please refer to
321 Documentation/DMA-API.txt. This section is just a reminder that
322 drivers need to indicate DMA capabilities of the device and is not
323 an authoritative source for DMA interfaces. ]
324
325While all drivers should explicitly indicate the DMA capability
326(e.g. 32 or 64 bit) of the PCI bus master, devices with more than
32732-bit bus master capability for streaming data need the driver
328to "register" this capability by calling pci_set_dma_mask() with
329appropriate parameters. In general this allows more efficient DMA
330on systems where System RAM exists above 4G _physical_ address.
331
332Drivers for all PCI-X and PCIe compliant devices must call
333pci_set_dma_mask() as they are 64-bit DMA devices.
334
335Similarly, drivers must also "register" this capability if the device
336can directly address "consistent memory" in System RAM above 4G physical
337address by calling pci_set_consistent_dma_mask().
338Again, this includes drivers for all PCI-X and PCIe compliant devices.
339Many 64-bit "PCI" devices (before PCI-X) and some PCI-X devices are
34064-bit DMA capable for payload ("streaming") data but not control
341("consistent") data.
342
343
3443.4 Setup shared control data
345~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
346Once the DMA masks are set, the driver can allocate "consistent" (a.k.a. shared)
347memory. See Documentation/DMA-API.txt for a full description of
348the DMA APIs. This section is just a reminder that it needs to be done
349before enabling DMA on the device.
350
351
3523.5 Initialize device registers
353~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
354Some drivers will need specific "capability" fields programmed
355or other "vendor specific" register initialized or reset.
356E.g. clearing pending interrupts.
357
358
3593.6 Register IRQ handler
360~~~~~~~~~~~~~~~~~~~~~~~~
Michael Opdenacker59c51592007-05-09 08:57:56 +0200361While calling request_irq() is the last step described here,
Grant Grundler74da15e2006-12-25 01:06:35 -0700362this is often just another intermediate step to initialize a device.
363This step can often be deferred until the device is opened for use.
364
365All interrupt handlers for IRQ lines should be registered with IRQF_SHARED
366and use the devid to map IRQs to devices (remember that all PCI IRQ lines
367can be shared).
368
369request_irq() will associate an interrupt handler and device handle
370with an interrupt number. Historically interrupt numbers represent
371IRQ lines which run from the PCI device to the Interrupt controller.
372With MSI and MSI-X (more below) the interrupt number is a CPU "vector".
373
374request_irq() also enables the interrupt. Make sure the device is
375quiesced and does not have any interrupts pending before registering
376the interrupt handler.
377
378MSI and MSI-X are PCI capabilities. Both are "Message Signaled Interrupts"
379which deliver interrupts to the CPU via a DMA write to a Local APIC.
380The fundamental difference between MSI and MSI-X is how multiple
381"vectors" get allocated. MSI requires contiguous blocks of vectors
382while MSI-X can allocate several individual ones.
383
384MSI capability can be enabled by calling pci_enable_msi() or
385pci_enable_msix() before calling request_irq(). This causes
386the PCI support to program CPU vector data into the PCI device
387capability registers.
388
389If your PCI device supports both, try to enable MSI-X first.
390Only one can be enabled at a time. Many architectures, chip-sets,
391or BIOSes do NOT support MSI or MSI-X and the call to pci_enable_msi/msix
392will fail. This is important to note since many drivers have
393two (or more) interrupt handlers: one for MSI/MSI-X and another for IRQs.
394They choose which handler to register with request_irq() based on the
395return value from pci_enable_msi/msix().
396
397There are (at least) two really good reasons for using MSI:
3981) MSI is an exclusive interrupt vector by definition.
399 This means the interrupt handler doesn't have to verify
400 its device caused the interrupt.
401
4022) MSI avoids DMA/IRQ race conditions. DMA to host memory is guaranteed
403 to be visible to the host CPU(s) when the MSI is delivered. This
404 is important for both data coherency and avoiding stale control data.
405 This guarantee allows the driver to omit MMIO reads to flush
406 the DMA stream.
407
408See drivers/infiniband/hw/mthca/ or drivers/net/tg3.c for examples
409of MSI/MSI-X usage.
410
411
412
4134. PCI device shutdown
414~~~~~~~~~~~~~~~~~~~~~~~
415
416When a PCI device driver is being unloaded, most of the following
417steps need to be performed:
418
419 Disable the device from generating IRQs
420 Release the IRQ (free_irq())
421 Stop all DMA activity
422 Release DMA buffers (both streaming and consistent)
423 Unregister from other subsystems (e.g. scsi or netdev)
424 Disable device from responding to MMIO/IO Port addresses
425 Release MMIO/IO Port resource(s)
426
427
4284.1 Stop IRQs on the device
429~~~~~~~~~~~~~~~~~~~~~~~~~~~
430How to do this is chip/device specific. If it's not done, it opens
431the possibility of a "screaming interrupt" if (and only if)
432the IRQ is shared with another device.
433
434When the shared IRQ handler is "unhooked", the remaining devices
435using the same IRQ line will still need the IRQ enabled. Thus if the
436"unhooked" device asserts IRQ line, the system will respond assuming
437it was one of the remaining devices asserted the IRQ line. Since none
438of the other devices will handle the IRQ, the system will "hang" until
439it decides the IRQ isn't going to get handled and masks the IRQ (100,000
440iterations later). Once the shared IRQ is masked, the remaining devices
441will stop functioning properly. Not a nice situation.
442
443This is another reason to use MSI or MSI-X if it's available.
444MSI and MSI-X are defined to be exclusive interrupts and thus
445are not susceptible to the "screaming interrupt" problem.
446
447
4484.2 Release the IRQ
449~~~~~~~~~~~~~~~~~~~
450Once the device is quiesced (no more IRQs), one can call free_irq().
451This function will return control once any pending IRQs are handled,
452"unhook" the drivers IRQ handler from that IRQ, and finally release
453the IRQ if no one else is using it.
454
455
4564.3 Stop all DMA activity
457~~~~~~~~~~~~~~~~~~~~~~~~~
458It's extremely important to stop all DMA operations BEFORE attempting
459to deallocate DMA control data. Failure to do so can result in memory
460corruption, hangs, and on some chip-sets a hard crash.
461
462Stopping DMA after stopping the IRQs can avoid races where the
463IRQ handler might restart DMA engines.
464
465While this step sounds obvious and trivial, several "mature" drivers
466didn't get this step right in the past.
467
468
4694.4 Release DMA buffers
470~~~~~~~~~~~~~~~~~~~~~~~
471Once DMA is stopped, clean up streaming DMA first.
472I.e. unmap data buffers and return buffers to "upstream"
473owners if there is one.
474
475Then clean up "consistent" buffers which contain the control data.
476
477See Documentation/DMA-API.txt for details on unmapping interfaces.
478
479
4804.5 Unregister from other subsystems
481~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
482Most low level PCI device drivers support some other subsystem
483like USB, ALSA, SCSI, NetDev, Infiniband, etc. Make sure your
484driver isn't losing resources from that other subsystem.
485If this happens, typically the symptom is an Oops (panic) when
486the subsystem attempts to call into a driver that has been unloaded.
487
488
4894.6 Disable Device from responding to MMIO/IO Port addresses
490~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
491io_unmap() MMIO or IO Port resources and then call pci_disable_device().
492This is the symmetric opposite of pci_enable_device().
493Do not access device registers after calling pci_disable_device().
494
495
4964.7 Release MMIO/IO Port Resource(s)
497~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
498Call pci_release_region() to mark the MMIO or IO Port range as available.
499Failure to do so usually results in the inability to reload the driver.
500
501
502
5035. How to access PCI config space
Linus Torvalds1da177e2005-04-16 15:20:36 -0700504~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -0700505
506You can use pci_(read|write)_config_(byte|word|dword) to access the config
Linus Torvalds1da177e2005-04-16 15:20:36 -0700507space of a device represented by struct pci_dev *. All these functions return 0
508when successful or an error code (PCIBIOS_...) which can be translated to a text
509string by pcibios_strerror. Most drivers expect that accesses to valid PCI
510devices don't fail.
511
Grant Grundler74da15e2006-12-25 01:06:35 -0700512If you don't have a struct pci_dev available, you can call
Linus Torvalds1da177e2005-04-16 15:20:36 -0700513pci_bus_(read|write)_config_(byte|word|dword) to access a given device
514and function on that bus.
515
Grant Grundler74da15e2006-12-25 01:06:35 -0700516If you access fields in the standard portion of the config header, please
Linus Torvalds1da177e2005-04-16 15:20:36 -0700517use symbolic names of locations and bits declared in <linux/pci.h>.
518
Grant Grundler74da15e2006-12-25 01:06:35 -0700519If you need to access Extended PCI Capability registers, just call
Linus Torvalds1da177e2005-04-16 15:20:36 -0700520pci_find_capability() for the particular capability and it will find the
521corresponding register block for you.
522
523
Linus Torvalds1da177e2005-04-16 15:20:36 -0700524
5256. Other interesting functions
526~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -0700527
Linus Torvalds1da177e2005-04-16 15:20:36 -0700528pci_find_slot() Find pci_dev corresponding to given bus and
529 slot numbers.
530pci_set_power_state() Set PCI Power Management state (0=D0 ... 3=D3)
531pci_find_capability() Find specified capability in device's capability
532 list.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700533pci_resource_start() Returns bus start address for a given PCI region
534pci_resource_end() Returns bus end address for a given PCI region
535pci_resource_len() Returns the byte length of a PCI region
536pci_set_drvdata() Set private driver data pointer for a pci_dev
537pci_get_drvdata() Return private driver data pointer for a pci_dev
538pci_set_mwi() Enable Memory-Write-Invalidate transactions.
539pci_clear_mwi() Disable Memory-Write-Invalidate transactions.
540
541
Grant Grundler74da15e2006-12-25 01:06:35 -0700542
Linus Torvalds1da177e2005-04-16 15:20:36 -07005437. Miscellaneous hints
544~~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -0700545
546When displaying PCI device names to the user (for example when a driver wants
547to tell the user what card has it found), please use pci_name(pci_dev).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700548
549Always refer to the PCI devices by a pointer to the pci_dev structure.
550All PCI layer functions use this identification and it's the only
551reasonable one. Don't use bus/slot/function numbers except for very
552special purposes -- on systems with multiple primary buses their semantics
553can be pretty complex.
554
Linus Torvalds1da177e2005-04-16 15:20:36 -0700555Don't try to turn on Fast Back to Back writes in your driver. All devices
556on the bus need to be capable of doing it, so this is something which needs
557to be handled by platform and generic code, not individual drivers.
558
559
Grant Grundler74da15e2006-12-25 01:06:35 -0700560
Ingo Oeser9b860b82006-04-18 11:20:55 +02005618. Vendor and device identifications
562~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Ingo Oeser9b860b82006-04-18 11:20:55 +0200563
Kulikov Vasiliyfa964e12010-07-10 18:35:43 +0400564One is not required to add new device ids to include/linux/pci_ids.h.
Grant Grundler74da15e2006-12-25 01:06:35 -0700565Please add PCI_VENDOR_ID_xxx for vendors and a hex constant for device ids.
Ingo Oeser9b860b82006-04-18 11:20:55 +0200566
Grant Grundler74da15e2006-12-25 01:06:35 -0700567PCI_VENDOR_ID_xxx constants are re-used. The device ids are arbitrary
568hex numbers (vendor controlled) and normally used only in a single
569location, the pci_device_id table.
570
571Please DO submit new vendor/device ids to pciids.sourceforge.net project.
572
573
Ingo Oeser9b860b82006-04-18 11:20:55 +0200574
5759. Obsolete functions
Linus Torvalds1da177e2005-04-16 15:20:36 -0700576~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -0700577
Linus Torvalds1da177e2005-04-16 15:20:36 -0700578There are several functions which you might come across when trying to
579port an old driver to the new PCI interface. They are no longer present
580in the kernel as they aren't compatible with hotplug or PCI domains or
581having sane locking.
582
Grant Grundler74da15e2006-12-25 01:06:35 -0700583pci_find_device() Superseded by pci_get_device()
584pci_find_subsys() Superseded by pci_get_subsys()
585pci_find_slot() Superseded by pci_get_slot()
586
587
588The alternative is the traditional PCI device driver that walks PCI
589device lists. This is still possible but discouraged.
590
591
592
Grant Grundlerd48b5d32007-12-24 00:08:51 -070059310. MMIO Space and "Write Posting"
Grant Grundler74da15e2006-12-25 01:06:35 -0700594~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
595
596Converting a driver from using I/O Port space to using MMIO space
597often requires some additional changes. Specifically, "write posting"
598needs to be handled. Many drivers (e.g. tg3, acenic, sym53c8xx_2)
599already do this. I/O Port space guarantees write transactions reach the PCI
600device before the CPU can continue. Writes to MMIO space allow the CPU
601to continue before the transaction reaches the PCI device. HW weenies
602call this "Write Posting" because the write completion is "posted" to
603the CPU before the transaction has reached its destination.
604
605Thus, timing sensitive code should add readl() where the CPU is
606expected to wait before doing other work. The classic "bit banging"
607sequence works fine for I/O Port space:
608
609 for (i = 8; --i; val >>= 1) {
610 outb(val & 1, ioport_reg); /* write bit */
611 udelay(10);
612 }
613
614The same sequence for MMIO space should be:
615
616 for (i = 8; --i; val >>= 1) {
617 writeb(val & 1, mmio_reg); /* write bit */
618 readb(safe_mmio_reg); /* flush posted write */
619 udelay(10);
620 }
621
622It is important that "safe_mmio_reg" not have any side effects that
623interferes with the correct operation of the device.
624
625Another case to watch out for is when resetting a PCI device. Use PCI
626Configuration space reads to flush the writel(). This will gracefully
627handle the PCI master abort on all platforms if the PCI device is
628expected to not respond to a readl(). Most x86 platforms will allow
629MMIO reads to master abort (a.k.a. "Soft Fail") and return garbage
630(e.g. ~0). But many RISC platforms will crash (a.k.a."Hard Fail").
631