1What:		/sys/bus/pci/drivers/.../bind
2Date:		December 2003
3Contact:	linux-pci@vger.kernel.org
4Description:
5		Writing a device location to this file will cause
6		the driver to attempt to bind to the device found at
7		this location.	This is useful for overriding default
8		bindings.  The format for the location is: DDDD:BB:DD.F.
9		That is Domain:Bus:Device.Function and is the same as
10		found in /sys/bus/pci/devices/.  For example::
11
12		  # echo 0000:00:19.0 > /sys/bus/pci/drivers/foo/bind
13
14		(Note: kernels before 2.6.28 may require echo -n).
15
16What:		/sys/bus/pci/drivers/.../unbind
17Date:		December 2003
18Contact:	linux-pci@vger.kernel.org
19Description:
20		Writing a device location to this file will cause the
21		driver to attempt to unbind from the device found at
22		this location.	This may be useful when overriding default
23		bindings.  The format for the location is: DDDD:BB:DD.F.
24		That is Domain:Bus:Device.Function and is the same as
25		found in /sys/bus/pci/devices/. For example::
26
27		  # echo 0000:00:19.0 > /sys/bus/pci/drivers/foo/unbind
28
29		(Note: kernels before 2.6.28 may require echo -n).
30
31What:		/sys/bus/pci/drivers/.../new_id
32Date:		December 2003
33Contact:	linux-pci@vger.kernel.org
34Description:
35		Writing a device ID to this file will attempt to
36		dynamically add a new device ID to a PCI device driver.
37		This may allow the driver to support more hardware than
38		was included in the driver's static device ID support
39		table at compile time.  The format for the device ID is:
40		VVVV DDDD SVVV SDDD CCCC MMMM PPPP.  That is Vendor ID,
41		Device ID, Subsystem Vendor ID, Subsystem Device ID,
42		Class, Class Mask, and Private Driver Data.  The Vendor ID
43		and Device ID fields are required, the rest are optional.
44		Upon successfully adding an ID, the driver will probe
45		for the device and attempt to bind to it.  For example::
46
47		  # echo "8086 10f5" > /sys/bus/pci/drivers/foo/new_id
48
49What:		/sys/bus/pci/drivers/.../remove_id
50Date:		February 2009
51Contact:	Chris Wright <chrisw@sous-sol.org>
52Description:
53		Writing a device ID to this file will remove an ID
54		that was dynamically added via the new_id sysfs entry.
55		The format for the device ID is:
56		VVVV DDDD SVVV SDDD CCCC MMMM.	That is Vendor ID, Device
57		ID, Subsystem Vendor ID, Subsystem Device ID, Class,
58		and Class Mask.  The Vendor ID and Device ID fields are
59		required, the rest are optional.  After successfully
60		removing an ID, the driver will no longer support the
61		device.  This is useful to ensure auto probing won't
62		match the driver to the device.  For example::
63
64		  # echo "8086 10f5" > /sys/bus/pci/drivers/foo/remove_id
65
66What:		/sys/bus/pci/rescan
67Date:		January 2009
68Contact:	Linux PCI developers <linux-pci@vger.kernel.org>
69Description:
70		Writing a non-zero value to this attribute will
71		force a rescan of all PCI buses in the system, and
72		re-discover previously removed devices.
73
74What:		/sys/bus/pci/devices/.../msi_bus
75Date:		September 2014
76Contact:	Linux PCI developers <linux-pci@vger.kernel.org>
77Description:
78		Writing a zero value to this attribute disallows MSI and
79		MSI-X for any future drivers of the device.  If the device
80		is a bridge, MSI and MSI-X will be disallowed for future
81		drivers of all child devices under the bridge.  Drivers
82		must be reloaded for the new setting to take effect.
83
84What:		/sys/bus/pci/devices/.../msi_irqs/
85Date:		September, 2011
86Contact:	Neil Horman <nhorman@tuxdriver.com>
87Description:
88		The /sys/devices/.../msi_irqs directory contains a variable set
89		of files, with each file being named after a corresponding msi
90		irq vector allocated to that device.
91
92What:		/sys/bus/pci/devices/.../msi_irqs/<N>
93Date:		September 2011
94Contact:	Neil Horman <nhorman@tuxdriver.com>
95Description:
96		This attribute indicates the mode that the irq vector named by
97		the file is in (msi vs. msix)
98
99What:		/sys/bus/pci/devices/.../remove
100Date:		January 2009
101Contact:	Linux PCI developers <linux-pci@vger.kernel.org>
102Description:
103		Writing a non-zero value to this attribute will
104		hot-remove the PCI device and any of its children.
105
106What:		/sys/bus/pci/devices/.../pci_bus/.../rescan
107Date:		May 2011
108Contact:	Linux PCI developers <linux-pci@vger.kernel.org>
109Description:
110		Writing a non-zero value to this attribute will
111		force a rescan of the bus and all child buses,
112		and re-discover devices removed earlier from this
113		part of the device tree.
114
115What:		/sys/bus/pci/devices/.../rescan
116Date:		January 2009
117Contact:	Linux PCI developers <linux-pci@vger.kernel.org>
118Description:
119		Writing a non-zero value to this attribute will
120		force a rescan of the device's parent bus and all
121		child buses, and re-discover devices removed earlier
122		from this part of the device tree.
123
124What:		/sys/bus/pci/devices/.../reset
125Date:		July 2009
126Contact:	Michael S. Tsirkin <mst@redhat.com>
127Description:
128		Some devices allow an individual function to be reset
129		without affecting other functions in the same device.
130		For devices that have this support, a file named reset
131		will be present in sysfs.  Writing 1 to this file
132		will perform reset.
133
134What:		/sys/bus/pci/devices/.../vpd
135Date:		February 2008
136Contact:	Ben Hutchings <bwh@kernel.org>
137Description:
138		A file named vpd in a device directory will be a
139		binary file containing the Vital Product Data for the
140		device.  It should follow the VPD format defined in
141		PCI Specification 2.1 or 2.2, but users should consider
142		that some devices may have malformatted data.  If the
143		underlying VPD has a writable section then the
144		corresponding section of this file will be writable.
145
146What:		/sys/bus/pci/devices/.../virtfnN
147Date:		March 2009
148Contact:	Yu Zhao <yu.zhao@intel.com>
149Description:
150		This symbolic link appears when hardware supports the SR-IOV
151		capability and the Physical Function driver has enabled it.
152		The symbolic link points to the PCI device sysfs entry of the
153		Virtual Function whose index is N (0...MaxVFs-1).
154
155What:		/sys/bus/pci/devices/.../dep_link
156Date:		March 2009
157Contact:	Yu Zhao <yu.zhao@intel.com>
158Description:
159		This symbolic link appears when hardware supports the SR-IOV
160		capability and the Physical Function driver has enabled it,
161		and this device has vendor specific dependencies with others.
162		The symbolic link points to the PCI device sysfs entry of
163		Physical Function this device depends on.
164
165What:		/sys/bus/pci/devices/.../physfn
166Date:		March 2009
167Contact:	Yu Zhao <yu.zhao@intel.com>
168Description:
169		This symbolic link appears when a device is a Virtual Function.
170		The symbolic link points to the PCI device sysfs entry of the
171		Physical Function this device associates with.
172
173What:		/sys/bus/pci/slots/.../module
174Date:		June 2009
175Contact:	linux-pci@vger.kernel.org
176Description:
177		This symbolic link points to the PCI hotplug controller driver
178		module that manages the hotplug slot.
179
180What:		/sys/bus/pci/devices/.../label
181Date:		July 2010
182Contact:	Narendra K <narendra_k@dell.com>, linux-bugs@dell.com
183Description:
184		Reading this attribute will provide the firmware
185		given name (SMBIOS type 41 string or ACPI _DSM string) of
186		the PCI device.	The attribute will be created only
187		if the firmware	has given a name to the PCI device.
188		ACPI _DSM string name will be given priority if the
189		system firmware provides SMBIOS type 41 string also.
190Users:
191		Userspace applications interested in knowing the
192		firmware assigned name of the PCI device.
193
194What:		/sys/bus/pci/devices/.../index
195Date:		July 2010
196Contact:	Narendra K <narendra_k@dell.com>, linux-bugs@dell.com
197Description:
198		Reading this attribute will provide the firmware given instance
199		number of the PCI device.  Depending on the platform this can
200		be for example the SMBIOS type 41 device type instance or the
201		user-defined ID (UID) on s390. The attribute will be created
202		only if the firmware has given an instance number to the PCI
203		device and that number is guaranteed to uniquely identify the
204		device in the system.
205Users:
206		Userspace applications interested in knowing the
207		firmware assigned device type instance of the PCI
208		device that can help in understanding the firmware
209		intended order of the PCI device.
210
211What:		/sys/bus/pci/devices/.../acpi_index
212Date:		July 2010
213Contact:	Narendra K <narendra_k@dell.com>, linux-bugs@dell.com
214Description:
215		Reading this attribute will provide the firmware
216		given instance (ACPI _DSM instance number) of the PCI device.
217		The attribute will be created only if the firmware has given
218		an instance number to the PCI device. ACPI _DSM instance number
219		will be given priority if the system firmware provides SMBIOS
220		type 41 device type instance also.
221Users:
222		Userspace applications interested in knowing the
223		firmware assigned instance number of the PCI
224		device that can help in understanding the firmware
225		intended order of the PCI device.
226
227What:		/sys/bus/pci/devices/.../d3cold_allowed
228Date:		July 2012
229Contact:	Huang Ying <ying.huang@intel.com>
230Description:
231		d3cold_allowed is bit to control whether the corresponding PCI
232		device can be put into D3Cold state.  If it is cleared, the
233		device will never be put into D3Cold state.  If it is set, the
234		device may be put into D3Cold state if other requirements are
235		satisfied too.  Reading this attribute will show the current
236		value of d3cold_allowed bit.  Writing this attribute will set
237		the value of d3cold_allowed bit.
238
239What:		/sys/bus/pci/devices/.../sriov_totalvfs
240Date:		November 2012
241Contact:	Donald Dutile <ddutile@redhat.com>
242Description:
243		This file appears when a physical PCIe device supports SR-IOV.
244		Userspace applications can read this file to determine the
245		maximum number of Virtual Functions (VFs) a PCIe physical
246		function (PF) can support. Typically, this is the value reported
247		in the PF's SR-IOV extended capability structure's TotalVFs
248		element.  Drivers have the ability at probe time to reduce the
249		value read from this file via the pci_sriov_set_totalvfs()
250		function.
251
252What:		/sys/bus/pci/devices/.../sriov_numvfs
253Date:		November 2012
254Contact:	Donald Dutile <ddutile@redhat.com>
255Description:
256		This file appears when a physical PCIe device supports SR-IOV.
257		Userspace applications can read and write to this file to
258		determine and control the enablement or disablement of Virtual
259		Functions (VFs) on the physical function (PF). A read of this
260		file will return the number of VFs that are enabled on this PF.
261		A number written to this file will enable the specified
262		number of VFs. A userspace application would typically read the
263		file and check that the value is zero, and then write the number
264		of VFs that should be enabled on the PF; the value written
265		should be less than or equal to the value in the sriov_totalvfs
266		file. A userspace application wanting to disable the VFs would
267		write a zero to this file. The core ensures that valid values
268		are written to this file, and returns errors when values are not
269		valid.  For example, writing a 2 to this file when sriov_numvfs
270		is not 0 and not 2 already will return an error. Writing a 10
271		when the value of sriov_totalvfs is 8 will return an error.
272
273What:		/sys/bus/pci/devices/.../driver_override
274Date:		April 2014
275Contact:	Alex Williamson <alex.williamson@redhat.com>
276Description:
277		This file allows the driver for a device to be specified which
278		will override standard static and dynamic ID matching.  When
279		specified, only a driver with a name matching the value written
280		to driver_override will have an opportunity to bind to the
281		device.  The override is specified by writing a string to the
282		driver_override file (echo pci-stub > driver_override) and
283		may be cleared with an empty string (echo > driver_override).
284		This returns the device to standard matching rules binding.
285		Writing to driver_override does not automatically unbind the
286		device from its current driver or make any attempt to
287		automatically load the specified driver.  If no driver with a
288		matching name is currently loaded in the kernel, the device
289		will not bind to any driver.  This also allows devices to
290		opt-out of driver binding using a driver_override name such as
291		"none".  Only a single driver may be specified in the override,
292		there is no support for parsing delimiters.
293
294What:		/sys/bus/pci/devices/.../numa_node
295Date:		Oct 2014
296Contact:	Prarit Bhargava <prarit@redhat.com>
297Description:
298		This file contains the NUMA node to which the PCI device is
299		attached, or -1 if the node is unknown.  The initial value
300		comes from an ACPI _PXM method or a similar firmware
301		source.  If that is missing or incorrect, this file can be
302		written to override the node.  In that case, please report
303		a firmware bug to the system vendor.  Writing to this file
304		taints the kernel with TAINT_FIRMWARE_WORKAROUND, which
305		reduces the supportability of your system.
306
307What:		/sys/bus/pci/devices/.../revision
308Date:		November 2016
309Contact:	Emil Velikov <emil.l.velikov@gmail.com>
310Description:
311		This file contains the revision field of the PCI device.
312		The value comes from device config space. The file is read only.
313
314What:		/sys/bus/pci/devices/.../sriov_drivers_autoprobe
315Date:		April 2017
316Contact:	Bodong Wang<bodong@mellanox.com>
317Description:
318		This file is associated with the PF of a device that
319		supports SR-IOV.  It determines whether newly-enabled VFs
320		are immediately bound to a driver.  It initially contains
321		1, which means the kernel automatically binds VFs to a
322		compatible driver immediately after they are enabled.  If
323		an application writes 0 to the file before enabling VFs,
324		the kernel will not bind VFs to a driver.
325
326		A typical use case is to write 0 to this file, then enable
327		VFs, then assign the newly-created VFs to virtual machines.
328		Note that changing this file does not affect already-
329		enabled VFs.  In this scenario, the user must first disable
330		the VFs, write 0 to sriov_drivers_autoprobe, then re-enable
331		the VFs.
332
333		This is similar to /sys/bus/pci/drivers_autoprobe, but
334		affects only the VFs associated with a specific PF.
335
336What:		/sys/bus/pci/devices/.../p2pmem/size
337Date:		November 2017
338Contact:	Logan Gunthorpe <logang@deltatee.com>
339Description:
340		If the device has any Peer-to-Peer memory registered, this
341	        file contains the total amount of memory that the device
342		provides (in decimal).
343
344What:		/sys/bus/pci/devices/.../p2pmem/available
345Date:		November 2017
346Contact:	Logan Gunthorpe <logang@deltatee.com>
347Description:
348		If the device has any Peer-to-Peer memory registered, this
349	        file contains the amount of memory that has not been
350		allocated (in decimal).
351
352What:		/sys/bus/pci/devices/.../p2pmem/published
353Date:		November 2017
354Contact:	Logan Gunthorpe <logang@deltatee.com>
355Description:
356		If the device has any Peer-to-Peer memory registered, this
357	        file contains a '1' if the memory has been published for
358		use outside the driver that owns the device.
359
360What:		/sys/bus/pci/devices/.../link/clkpm
361		/sys/bus/pci/devices/.../link/l0s_aspm
362		/sys/bus/pci/devices/.../link/l1_aspm
363		/sys/bus/pci/devices/.../link/l1_1_aspm
364		/sys/bus/pci/devices/.../link/l1_2_aspm
365		/sys/bus/pci/devices/.../link/l1_1_pcipm
366		/sys/bus/pci/devices/.../link/l1_2_pcipm
367Date:		October 2019
368Contact:	Heiner Kallweit <hkallweit1@gmail.com>
369Description:	If ASPM is supported for an endpoint, these files can be
370		used to disable or enable the individual power management
371		states. Write y/1/on to enable, n/0/off to disable.
372
373What:		/sys/bus/pci/devices/.../power_state
374Date:		November 2020
375Contact:	Linux PCI developers <linux-pci@vger.kernel.org>
376Description:
377		This file contains the current PCI power state of the device.
378		The value comes from the PCI kernel device state and can be one
379		of: "unknown", "error", "D0", D1", "D2", "D3hot", "D3cold".
380		The file is read only.
381
382What:		/sys/bus/pci/devices/.../sriov_vf_total_msix
383Date:		January 2021
384Contact:	Leon Romanovsky <leonro@nvidia.com>
385Description:
386		This file is associated with a SR-IOV physical function (PF).
387		It contains the total number of MSI-X vectors available for
388		assignment to all virtual functions (VFs) associated with PF.
389		The value will be zero if the device doesn't support this
390		functionality. For supported devices, the value will be
391		constant and won't be changed after MSI-X vectors assignment.
392
393What:		/sys/bus/pci/devices/.../sriov_vf_msix_count
394Date:		January 2021
395Contact:	Leon Romanovsky <leonro@nvidia.com>
396Description:
397		This file is associated with a SR-IOV virtual function (VF).
398		It allows configuration of the number of MSI-X vectors for
399		the VF. This allows devices that have a global pool of MSI-X
400		vectors to optimally divide them between VFs based on VF usage.
401
402		The values accepted are:
403		 * > 0 - this number will be reported as the Table Size in the
404			 VF's MSI-X capability
405		 * < 0 - not valid
406		 * = 0 - will reset to the device default value
407
408		The file is writable if the PF is bound to a driver that
409		implements ->sriov_set_msix_vec_count().
410