blob: 123881f62219d444303b047154b25d39363fb9f8 [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
Joe Perches92e112f2013-12-13 11:36:22 -0700126all-zero entry. Definitions with static const are generally preferred.
127Use of the deprecated macro DEFINE_PCI_DEVICE_TABLE should be avoided.
128
129Each entry consists of:
Grant Grundler74da15e2006-12-25 01:06:35 -0700130
131 vendor,device Vendor and device ID to match (or PCI_ANY_ID)
132
Linus Torvalds1da177e2005-04-16 15:20:36 -0700133 subvendor, Subsystem vendor and device ID to match (or PCI_ANY_ID)
Grant Grundler74da15e2006-12-25 01:06:35 -0700134 subdevice,
135
136 class Device class, subclass, and "interface" to match.
137 See Appendix D of the PCI Local Bus Spec or
138 include/linux/pci_ids.h for a full list of classes.
139 Most drivers do not need to specify class/class_mask
140 as vendor/device is normally sufficient.
141
142 class_mask limit which sub-fields of the class field are compared.
143 See drivers/scsi/sym53c8xx_2/ for example of usage.
144
Linus Torvalds1da177e2005-04-16 15:20:36 -0700145 driver_data Data private to the driver.
Grant Grundler74da15e2006-12-25 01:06:35 -0700146 Most drivers don't need to use driver_data field.
147 Best practice is to use driver_data as an index
148 into a static list of equivalent device types,
149 instead of using it as a pointer.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700150
Linus Torvalds1da177e2005-04-16 15:20:36 -0700151
Grant Grundler74da15e2006-12-25 01:06:35 -0700152Most drivers only need PCI_DEVICE() or PCI_DEVICE_CLASS() to set up
153a pci_device_id table.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700154
Grant Grundler74da15e2006-12-25 01:06:35 -0700155New PCI IDs may be added to a device driver pci_ids table at runtime
156as shown below:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700157
158echo "vendor device subvendor subdevice class class_mask driver_data" > \
Grant Grundler74da15e2006-12-25 01:06:35 -0700159/sys/bus/pci/drivers/{driver}/new_id
160
161All fields are passed in as hexadecimal values (no leading 0x).
Jean Delvare6ba18632007-04-07 17:21:28 +0200162The vendor and device fields are mandatory, the others are optional. Users
163need pass only as many optional fields as necessary:
164 o subvendor and subdevice fields default to PCI_ANY_ID (FFFFFFFF)
Grant Grundler74da15e2006-12-25 01:06:35 -0700165 o class and classmask fields default to 0
166 o driver_data defaults to 0UL.
167
Jean Delvareb41d6cf2008-08-17 21:06:59 +0200168Note that driver_data must match the value used by any of the pci_device_id
169entries defined in the driver. This makes the driver_data field mandatory
170if all the pci_device_id entries have a non-zero driver_data value.
171
Grant Grundler74da15e2006-12-25 01:06:35 -0700172Once added, the driver probe routine will be invoked for any unclaimed
173PCI devices listed in its (newly updated) pci_ids list.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700174
175When the driver exits, it just calls pci_unregister_driver() and the PCI layer
176automatically calls the remove hook for all devices handled by the driver.
177
Grant Grundler74da15e2006-12-25 01:06:35 -0700178
1791.1 "Attributes" for driver functions/data
180
Linus Torvalds1da177e2005-04-16 15:20:36 -0700181Please mark the initialization and cleanup functions where appropriate
182(the corresponding macros are defined in <linux/init.h>):
183
184 __init Initialization code. Thrown away after the driver
185 initializes.
186 __exit Exit code. Ignored for non-modular drivers.
Grant Grundler74da15e2006-12-25 01:06:35 -0700187
Grant Grundler74da15e2006-12-25 01:06:35 -0700188Tips on when/where to use the above attributes:
189 o The module_init()/module_exit() functions (and all
190 initialization functions called _only_ from these)
191 should be marked __init/__exit.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700192
Grant Grundler74da15e2006-12-25 01:06:35 -0700193 o Do not mark the struct pci_driver.
194
Grant Grundler74da15e2006-12-25 01:06:35 -0700195 o Do NOT mark a function if you are not sure which mark to use.
196 Better to not mark the function than mark the function wrong.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700197
198
Grant Grundler74da15e2006-12-25 01:06:35 -0700199
2002. How to find PCI devices manually
201~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
202
203PCI drivers should have a really good reason for not using the
204pci_register_driver() interface to search for PCI devices.
205The main reason PCI devices are controlled by multiple drivers
206is because one PCI device implements several different HW services.
207E.g. combined serial/parallel port/floppy controller.
208
209A manual search may be performed using the following constructs:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700210
211Searching by vendor and device ID:
212
213 struct pci_dev *dev = NULL;
214 while (dev = pci_get_device(VENDOR_ID, DEVICE_ID, dev))
215 configure_device(dev);
216
217Searching by class ID (iterate in a similar way):
218
219 pci_get_class(CLASS_ID, dev)
220
221Searching by both vendor/device and subsystem vendor/device ID:
222
Grant Grundler74da15e2006-12-25 01:06:35 -0700223 pci_get_subsys(VENDOR_ID,DEVICE_ID, SUBSYS_VENDOR_ID, SUBSYS_DEVICE_ID, dev).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700224
Grant Grundler74da15e2006-12-25 01:06:35 -0700225You can use the constant PCI_ANY_ID as a wildcard replacement for
Linus Torvalds1da177e2005-04-16 15:20:36 -0700226VENDOR_ID or DEVICE_ID. This allows searching for any device from a
227specific vendor, for example.
228
Grant Grundler74da15e2006-12-25 01:06:35 -0700229These functions are hotplug-safe. They increment the reference count on
Linus Torvalds1da177e2005-04-16 15:20:36 -0700230the pci_dev that they return. You must eventually (possibly at module unload)
231decrement the reference count on these devices by calling pci_dev_put().
232
233
Linus Torvalds1da177e2005-04-16 15:20:36 -0700234
Grant Grundler74da15e2006-12-25 01:06:35 -07002353. Device Initialization Steps
236~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Linus Torvalds1da177e2005-04-16 15:20:36 -0700237
Grant Grundler74da15e2006-12-25 01:06:35 -0700238As noted in the introduction, most PCI drivers need the following steps
239for device initialization:
240
241 Enable the device
242 Request MMIO/IOP resources
243 Set the DMA mask size (for both coherent and streaming DMA)
244 Allocate and initialize shared control data (pci_allocate_coherent())
245 Access device configuration space (if needed)
246 Register IRQ handler (request_irq())
247 Initialize non-PCI (i.e. LAN/SCSI/etc parts of the chip)
248 Enable DMA/processing engines.
249
250The driver can access PCI config space registers at any time.
251(Well, almost. When running BIST, config space can go away...but
252that will just result in a PCI Bus Master Abort and config reads
253will return garbage).
254
255
2563.1 Enable the PCI device
257~~~~~~~~~~~~~~~~~~~~~~~~~
258Before touching any device registers, the driver needs to enable
259the PCI device by calling pci_enable_device(). This will:
260 o wake up the device if it was in suspended state,
261 o allocate I/O and memory regions of the device (if BIOS did not),
262 o allocate an IRQ (if BIOS did not).
263
264NOTE: pci_enable_device() can fail! Check the return value.
Grant Grundler74da15e2006-12-25 01:06:35 -0700265
266[ OS BUG: we don't check resource allocations before enabling those
267 resources. The sequence would make more sense if we called
268 pci_request_resources() before calling pci_enable_device().
269 Currently, the device drivers can't detect the bug when when two
270 devices have been allocated the same range. This is not a common
271 problem and unlikely to get fixed soon.
272
273 This has been discussed before but not changed as of 2.6.19:
274 http://lkml.org/lkml/2006/3/2/194
275]
276
277pci_set_master() will enable DMA by setting the bus master bit
278in the PCI_COMMAND register. It also fixes the latency timer value if
Ben Hutchings6a479072008-12-23 03:08:29 +0000279it's set to something bogus by the BIOS. pci_clear_master() will
280disable DMA by clearing the bus master bit.
Grant Grundler74da15e2006-12-25 01:06:35 -0700281
282If the PCI device can use the PCI Memory-Write-Invalidate transaction,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700283call pci_set_mwi(). This enables the PCI_COMMAND bit for Mem-Wr-Inval
284and also ensures that the cache line size register is set correctly.
Grant Grundler74da15e2006-12-25 01:06:35 -0700285Check the return value of pci_set_mwi() as not all architectures
Randy Dunlap694625c2007-07-09 11:55:54 -0700286or chip-sets may support Memory-Write-Invalidate. Alternatively,
287if Mem-Wr-Inval would be nice to have but is not required, call
288pci_try_set_mwi() to have the system do its best effort at enabling
289Mem-Wr-Inval.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700290
Grant Grundler74da15e2006-12-25 01:06:35 -0700291
2923.2 Request MMIO/IOP resources
293~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
294Memory (MMIO), and I/O port addresses should NOT be read directly
295from the PCI device config space. Use the values in the pci_dev structure
296as the PCI "bus address" might have been remapped to a "host physical"
297address by the arch/chip-set specific kernel support.
298
Paul Bolle395cf962011-08-15 02:02:26 +0200299See Documentation/io-mapping.txt for how to access device registers
Grant Grundler74da15e2006-12-25 01:06:35 -0700300or device memory.
301
302The device driver needs to call pci_request_region() to verify
303no other device is already using the same address resource.
304Conversely, drivers should call pci_release_region() AFTER
Linus Torvalds1da177e2005-04-16 15:20:36 -0700305calling pci_disable_device().
Grant Grundler74da15e2006-12-25 01:06:35 -0700306The idea is to prevent two devices colliding on the same address range.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700307
Grant Grundler74da15e2006-12-25 01:06:35 -0700308[ See OS BUG comment above. Currently (2.6.19), The driver can only
309 determine MMIO and IO Port resource availability _after_ calling
310 pci_enable_device(). ]
311
312Generic flavors of pci_request_region() are request_mem_region()
313(for MMIO ranges) and request_region() (for IO Port ranges).
314Use these for address resources that are not described by "normal" PCI
315BARs.
316
317Also see pci_request_selected_regions() below.
318
319
3203.3 Set the DMA mask size
321~~~~~~~~~~~~~~~~~~~~~~~~~
322[ If anything below doesn't make sense, please refer to
323 Documentation/DMA-API.txt. This section is just a reminder that
324 drivers need to indicate DMA capabilities of the device and is not
325 an authoritative source for DMA interfaces. ]
326
327While all drivers should explicitly indicate the DMA capability
328(e.g. 32 or 64 bit) of the PCI bus master, devices with more than
32932-bit bus master capability for streaming data need the driver
330to "register" this capability by calling pci_set_dma_mask() with
331appropriate parameters. In general this allows more efficient DMA
332on systems where System RAM exists above 4G _physical_ address.
333
334Drivers for all PCI-X and PCIe compliant devices must call
335pci_set_dma_mask() as they are 64-bit DMA devices.
336
337Similarly, drivers must also "register" this capability if the device
338can directly address "consistent memory" in System RAM above 4G physical
339address by calling pci_set_consistent_dma_mask().
340Again, this includes drivers for all PCI-X and PCIe compliant devices.
341Many 64-bit "PCI" devices (before PCI-X) and some PCI-X devices are
34264-bit DMA capable for payload ("streaming") data but not control
343("consistent") data.
344
345
3463.4 Setup shared control data
347~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
348Once the DMA masks are set, the driver can allocate "consistent" (a.k.a. shared)
349memory. See Documentation/DMA-API.txt for a full description of
350the DMA APIs. This section is just a reminder that it needs to be done
351before enabling DMA on the device.
352
353
3543.5 Initialize device registers
355~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
356Some drivers will need specific "capability" fields programmed
357or other "vendor specific" register initialized or reset.
358E.g. clearing pending interrupts.
359
360
3613.6 Register IRQ handler
362~~~~~~~~~~~~~~~~~~~~~~~~
Michael Opdenacker59c51592007-05-09 08:57:56 +0200363While calling request_irq() is the last step described here,
Grant Grundler74da15e2006-12-25 01:06:35 -0700364this is often just another intermediate step to initialize a device.
365This step can often be deferred until the device is opened for use.
366
367All interrupt handlers for IRQ lines should be registered with IRQF_SHARED
368and use the devid to map IRQs to devices (remember that all PCI IRQ lines
369can be shared).
370
371request_irq() will associate an interrupt handler and device handle
372with an interrupt number. Historically interrupt numbers represent
373IRQ lines which run from the PCI device to the Interrupt controller.
374With MSI and MSI-X (more below) the interrupt number is a CPU "vector".
375
376request_irq() also enables the interrupt. Make sure the device is
377quiesced and does not have any interrupts pending before registering
378the interrupt handler.
379
380MSI and MSI-X are PCI capabilities. Both are "Message Signaled Interrupts"
381which deliver interrupts to the CPU via a DMA write to a Local APIC.
382The fundamental difference between MSI and MSI-X is how multiple
383"vectors" get allocated. MSI requires contiguous blocks of vectors
384while MSI-X can allocate several individual ones.
385
386MSI capability can be enabled by calling pci_enable_msi() or
387pci_enable_msix() before calling request_irq(). This causes
388the PCI support to program CPU vector data into the PCI device
389capability registers.
390
391If your PCI device supports both, try to enable MSI-X first.
392Only one can be enabled at a time. Many architectures, chip-sets,
393or BIOSes do NOT support MSI or MSI-X and the call to pci_enable_msi/msix
394will fail. This is important to note since many drivers have
395two (or more) interrupt handlers: one for MSI/MSI-X and another for IRQs.
396They choose which handler to register with request_irq() based on the
397return value from pci_enable_msi/msix().
398
399There are (at least) two really good reasons for using MSI:
4001) MSI is an exclusive interrupt vector by definition.
401 This means the interrupt handler doesn't have to verify
402 its device caused the interrupt.
403
4042) MSI avoids DMA/IRQ race conditions. DMA to host memory is guaranteed
405 to be visible to the host CPU(s) when the MSI is delivered. This
406 is important for both data coherency and avoiding stale control data.
407 This guarantee allows the driver to omit MMIO reads to flush
408 the DMA stream.
409
410See drivers/infiniband/hw/mthca/ or drivers/net/tg3.c for examples
411of MSI/MSI-X usage.
412
413
414
4154. PCI device shutdown
416~~~~~~~~~~~~~~~~~~~~~~~
417
418When a PCI device driver is being unloaded, most of the following
419steps need to be performed:
420
421 Disable the device from generating IRQs
422 Release the IRQ (free_irq())
423 Stop all DMA activity
424 Release DMA buffers (both streaming and consistent)
425 Unregister from other subsystems (e.g. scsi or netdev)
426 Disable device from responding to MMIO/IO Port addresses
427 Release MMIO/IO Port resource(s)
428
429
4304.1 Stop IRQs on the device
431~~~~~~~~~~~~~~~~~~~~~~~~~~~
432How to do this is chip/device specific. If it's not done, it opens
433the possibility of a "screaming interrupt" if (and only if)
434the IRQ is shared with another device.
435
436When the shared IRQ handler is "unhooked", the remaining devices
437using the same IRQ line will still need the IRQ enabled. Thus if the
438"unhooked" device asserts IRQ line, the system will respond assuming
439it was one of the remaining devices asserted the IRQ line. Since none
440of the other devices will handle the IRQ, the system will "hang" until
441it decides the IRQ isn't going to get handled and masks the IRQ (100,000
442iterations later). Once the shared IRQ is masked, the remaining devices
443will stop functioning properly. Not a nice situation.
444
445This is another reason to use MSI or MSI-X if it's available.
446MSI and MSI-X are defined to be exclusive interrupts and thus
447are not susceptible to the "screaming interrupt" problem.
448
449
4504.2 Release the IRQ
451~~~~~~~~~~~~~~~~~~~
452Once the device is quiesced (no more IRQs), one can call free_irq().
453This function will return control once any pending IRQs are handled,
454"unhook" the drivers IRQ handler from that IRQ, and finally release
455the IRQ if no one else is using it.
456
457
4584.3 Stop all DMA activity
459~~~~~~~~~~~~~~~~~~~~~~~~~
460It's extremely important to stop all DMA operations BEFORE attempting
461to deallocate DMA control data. Failure to do so can result in memory
462corruption, hangs, and on some chip-sets a hard crash.
463
464Stopping DMA after stopping the IRQs can avoid races where the
465IRQ handler might restart DMA engines.
466
467While this step sounds obvious and trivial, several "mature" drivers
468didn't get this step right in the past.
469
470
4714.4 Release DMA buffers
472~~~~~~~~~~~~~~~~~~~~~~~
473Once DMA is stopped, clean up streaming DMA first.
474I.e. unmap data buffers and return buffers to "upstream"
475owners if there is one.
476
477Then clean up "consistent" buffers which contain the control data.
478
479See Documentation/DMA-API.txt for details on unmapping interfaces.
480
481
4824.5 Unregister from other subsystems
483~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
484Most low level PCI device drivers support some other subsystem
485like USB, ALSA, SCSI, NetDev, Infiniband, etc. Make sure your
486driver isn't losing resources from that other subsystem.
487If this happens, typically the symptom is an Oops (panic) when
488the subsystem attempts to call into a driver that has been unloaded.
489
490
4914.6 Disable Device from responding to MMIO/IO Port addresses
492~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
493io_unmap() MMIO or IO Port resources and then call pci_disable_device().
494This is the symmetric opposite of pci_enable_device().
495Do not access device registers after calling pci_disable_device().
496
497
4984.7 Release MMIO/IO Port Resource(s)
499~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
500Call pci_release_region() to mark the MMIO or IO Port range as available.
501Failure to do so usually results in the inability to reload the driver.
502
503
504
5055. How to access PCI config space
Linus Torvalds1da177e2005-04-16 15:20:36 -0700506~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -0700507
508You can use pci_(read|write)_config_(byte|word|dword) to access the config
Linus Torvalds1da177e2005-04-16 15:20:36 -0700509space of a device represented by struct pci_dev *. All these functions return 0
510when successful or an error code (PCIBIOS_...) which can be translated to a text
511string by pcibios_strerror. Most drivers expect that accesses to valid PCI
512devices don't fail.
513
Grant Grundler74da15e2006-12-25 01:06:35 -0700514If you don't have a struct pci_dev available, you can call
Linus Torvalds1da177e2005-04-16 15:20:36 -0700515pci_bus_(read|write)_config_(byte|word|dword) to access a given device
516and function on that bus.
517
Grant Grundler74da15e2006-12-25 01:06:35 -0700518If you access fields in the standard portion of the config header, please
Linus Torvalds1da177e2005-04-16 15:20:36 -0700519use symbolic names of locations and bits declared in <linux/pci.h>.
520
Grant Grundler74da15e2006-12-25 01:06:35 -0700521If you need to access Extended PCI Capability registers, just call
Linus Torvalds1da177e2005-04-16 15:20:36 -0700522pci_find_capability() for the particular capability and it will find the
523corresponding register block for you.
524
525
Linus Torvalds1da177e2005-04-16 15:20:36 -0700526
5276. Other interesting functions
528~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -0700529
Yijing Wanga37bee72013-09-02 14:34:40 +0800530pci_get_domain_bus_and_slot() Find pci_dev corresponding to given domain,
531 bus and slot and number. If the device is
532 found, its reference count is increased.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700533pci_set_power_state() Set PCI Power Management state (0=D0 ... 3=D3)
534pci_find_capability() Find specified capability in device's capability
535 list.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700536pci_resource_start() Returns bus start address for a given PCI region
537pci_resource_end() Returns bus end address for a given PCI region
538pci_resource_len() Returns the byte length of a PCI region
539pci_set_drvdata() Set private driver data pointer for a pci_dev
540pci_get_drvdata() Return private driver data pointer for a pci_dev
541pci_set_mwi() Enable Memory-Write-Invalidate transactions.
542pci_clear_mwi() Disable Memory-Write-Invalidate transactions.
543
544
Grant Grundler74da15e2006-12-25 01:06:35 -0700545
Linus Torvalds1da177e2005-04-16 15:20:36 -07005467. Miscellaneous hints
547~~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -0700548
549When displaying PCI device names to the user (for example when a driver wants
550to tell the user what card has it found), please use pci_name(pci_dev).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700551
552Always refer to the PCI devices by a pointer to the pci_dev structure.
553All PCI layer functions use this identification and it's the only
554reasonable one. Don't use bus/slot/function numbers except for very
555special purposes -- on systems with multiple primary buses their semantics
556can be pretty complex.
557
Linus Torvalds1da177e2005-04-16 15:20:36 -0700558Don't try to turn on Fast Back to Back writes in your driver. All devices
559on the bus need to be capable of doing it, so this is something which needs
560to be handled by platform and generic code, not individual drivers.
561
562
Grant Grundler74da15e2006-12-25 01:06:35 -0700563
Ingo Oeser9b860b82006-04-18 11:20:55 +02005648. Vendor and device identifications
565~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Ingo Oeser9b860b82006-04-18 11:20:55 +0200566
Michael S. Tsirkin37a9c502015-03-30 10:32:34 +0200567Do not add new device or vendor IDs to include/linux/pci_ids.h unless they
568are shared across multiple drivers. You can add private definitions in
569your driver if they're helpful, or just use plain hex constants.
Ingo Oeser9b860b82006-04-18 11:20:55 +0200570
Michael S. Tsirkin37a9c502015-03-30 10:32:34 +0200571The device IDs are arbitrary hex numbers (vendor controlled) and normally used
572only in a single location, the pci_device_id table.
Grant Grundler74da15e2006-12-25 01:06:35 -0700573
Michael S. Tsirkin37a9c502015-03-30 10:32:34 +0200574Please DO submit new vendor/device IDs to http://pciids.sourceforge.net/.
Grant Grundler74da15e2006-12-25 01:06:35 -0700575
576
Ingo Oeser9b860b82006-04-18 11:20:55 +0200577
5789. Obsolete functions
Linus Torvalds1da177e2005-04-16 15:20:36 -0700579~~~~~~~~~~~~~~~~~~~~~
Grant Grundler74da15e2006-12-25 01:06:35 -0700580
Linus Torvalds1da177e2005-04-16 15:20:36 -0700581There are several functions which you might come across when trying to
582port an old driver to the new PCI interface. They are no longer present
583in the kernel as they aren't compatible with hotplug or PCI domains or
584having sane locking.
585
Grant Grundler74da15e2006-12-25 01:06:35 -0700586pci_find_device() Superseded by pci_get_device()
587pci_find_subsys() Superseded by pci_get_subsys()
Yijing Wanga37bee72013-09-02 14:34:40 +0800588pci_find_slot() Superseded by pci_get_domain_bus_and_slot()
589pci_get_slot() Superseded by pci_get_domain_bus_and_slot()
Grant Grundler74da15e2006-12-25 01:06:35 -0700590
591
592The alternative is the traditional PCI device driver that walks PCI
593device lists. This is still possible but discouraged.
594
595
596
Grant Grundlerd48b5d32007-12-24 00:08:51 -070059710. MMIO Space and "Write Posting"
Grant Grundler74da15e2006-12-25 01:06:35 -0700598~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
599
600Converting a driver from using I/O Port space to using MMIO space
601often requires some additional changes. Specifically, "write posting"
602needs to be handled. Many drivers (e.g. tg3, acenic, sym53c8xx_2)
603already do this. I/O Port space guarantees write transactions reach the PCI
604device before the CPU can continue. Writes to MMIO space allow the CPU
605to continue before the transaction reaches the PCI device. HW weenies
606call this "Write Posting" because the write completion is "posted" to
607the CPU before the transaction has reached its destination.
608
609Thus, timing sensitive code should add readl() where the CPU is
610expected to wait before doing other work. The classic "bit banging"
611sequence works fine for I/O Port space:
612
613 for (i = 8; --i; val >>= 1) {
614 outb(val & 1, ioport_reg); /* write bit */
615 udelay(10);
616 }
617
618The same sequence for MMIO space should be:
619
620 for (i = 8; --i; val >>= 1) {
621 writeb(val & 1, mmio_reg); /* write bit */
622 readb(safe_mmio_reg); /* flush posted write */
623 udelay(10);
624 }
625
626It is important that "safe_mmio_reg" not have any side effects that
627interferes with the correct operation of the device.
628
629Another case to watch out for is when resetting a PCI device. Use PCI
630Configuration space reads to flush the writel(). This will gracefully
631handle the PCI master abort on all platforms if the PCI device is
632expected to not respond to a readl(). Most x86 platforms will allow
633MMIO reads to master abort (a.k.a. "Soft Fail") and return garbage
634(e.g. ~0). But many RISC platforms will crash (a.k.a."Hard Fail").
635