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