sysfs.txt 6.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156
  1. GPIO Sysfs Interface for Userspace
  2. ==================================
  3. THIS ABI IS DEPRECATED, THE ABI DOCUMENTATION HAS BEEN MOVED TO
  4. Documentation/ABI/obsolete/sysfs-gpio AND NEW USERSPACE CONSUMERS
  5. ARE SUPPOSED TO USE THE CHARACTER DEVICE ABI. THIS OLD SYSFS ABI WILL
  6. NOT BE DEVELOPED (NO NEW FEATURES), IT WILL JUST BE MAINTAINED.
  7. Refer to the examples in tools/gpio/* for an introduction to the new
  8. character device ABI. Also see the userspace header in
  9. include/uapi/linux/gpio.h
  10. The deprecated sysfs ABI
  11. ------------------------
  12. Platforms which use the "gpiolib" implementors framework may choose to
  13. configure a sysfs user interface to GPIOs. This is different from the
  14. debugfs interface, since it provides control over GPIO direction and
  15. value instead of just showing a gpio state summary. Plus, it could be
  16. present on production systems without debugging support.
  17. Given appropriate hardware documentation for the system, userspace could
  18. know for example that GPIO #23 controls the write protect line used to
  19. protect boot loader segments in flash memory. System upgrade procedures
  20. may need to temporarily remove that protection, first importing a GPIO,
  21. then changing its output state, then updating the code before re-enabling
  22. the write protection. In normal use, GPIO #23 would never be touched,
  23. and the kernel would have no need to know about it.
  24. Again depending on appropriate hardware documentation, on some systems
  25. userspace GPIO can be used to determine system configuration data that
  26. standard kernels won't know about. And for some tasks, simple userspace
  27. GPIO drivers could be all that the system really needs.
  28. DO NOT ABUSE SYSFS TO CONTROL HARDWARE THAT HAS PROPER KERNEL DRIVERS.
  29. PLEASE READ THE DOCUMENT AT Documentation/driver-api/gpio/drivers-on-gpio.rst
  30. TO AVOID REINVENTING KERNEL WHEELS IN USERSPACE. I MEAN IT. REALLY.
  31. Paths in Sysfs
  32. --------------
  33. There are three kinds of entries in /sys/class/gpio:
  34. - Control interfaces used to get userspace control over GPIOs;
  35. - GPIOs themselves; and
  36. - GPIO controllers ("gpio_chip" instances).
  37. That's in addition to standard files including the "device" symlink.
  38. The control interfaces are write-only:
  39. /sys/class/gpio/
  40. "export" ... Userspace may ask the kernel to export control of
  41. a GPIO to userspace by writing its number to this file.
  42. Example: "echo 19 > export" will create a "gpio19" node
  43. for GPIO #19, if that's not requested by kernel code.
  44. "unexport" ... Reverses the effect of exporting to userspace.
  45. Example: "echo 19 > unexport" will remove a "gpio19"
  46. node exported using the "export" file.
  47. GPIO signals have paths like /sys/class/gpio/gpio42/ (for GPIO #42)
  48. and have the following read/write attributes:
  49. /sys/class/gpio/gpioN/
  50. "direction" ... reads as either "in" or "out". This value may
  51. normally be written. Writing as "out" defaults to
  52. initializing the value as low. To ensure glitch free
  53. operation, values "low" and "high" may be written to
  54. configure the GPIO as an output with that initial value.
  55. Note that this attribute *will not exist* if the kernel
  56. doesn't support changing the direction of a GPIO, or
  57. it was exported by kernel code that didn't explicitly
  58. allow userspace to reconfigure this GPIO's direction.
  59. "value" ... reads as either 0 (low) or 1 (high). If the GPIO
  60. is configured as an output, this value may be written;
  61. any nonzero value is treated as high.
  62. If the pin can be configured as interrupt-generating interrupt
  63. and if it has been configured to generate interrupts (see the
  64. description of "edge"), you can poll(2) on that file and
  65. poll(2) will return whenever the interrupt was triggered. If
  66. you use poll(2), set the events POLLPRI and POLLERR. If you
  67. use select(2), set the file descriptor in exceptfds. After
  68. poll(2) returns, either lseek(2) to the beginning of the sysfs
  69. file and read the new value or close the file and re-open it
  70. to read the value.
  71. "edge" ... reads as either "none", "rising", "falling", or
  72. "both". Write these strings to select the signal edge(s)
  73. that will make poll(2) on the "value" file return.
  74. This file exists only if the pin can be configured as an
  75. interrupt generating input pin.
  76. "active_low" ... reads as either 0 (false) or 1 (true). Write
  77. any nonzero value to invert the value attribute both
  78. for reading and writing. Existing and subsequent
  79. poll(2) support configuration via the edge attribute
  80. for "rising" and "falling" edges will follow this
  81. setting.
  82. GPIO controllers have paths like /sys/class/gpio/gpiochip42/ (for the
  83. controller implementing GPIOs starting at #42) and have the following
  84. read-only attributes:
  85. /sys/class/gpio/gpiochipN/
  86. "base" ... same as N, the first GPIO managed by this chip
  87. "label" ... provided for diagnostics (not always unique)
  88. "ngpio" ... how many GPIOs this manages (N to N + ngpio - 1)
  89. Board documentation should in most cases cover what GPIOs are used for
  90. what purposes. However, those numbers are not always stable; GPIOs on
  91. a daughtercard might be different depending on the base board being used,
  92. or other cards in the stack. In such cases, you may need to use the
  93. gpiochip nodes (possibly in conjunction with schematics) to determine
  94. the correct GPIO number to use for a given signal.
  95. Exporting from Kernel code
  96. --------------------------
  97. Kernel code can explicitly manage exports of GPIOs which have already been
  98. requested using gpio_request():
  99. /* export the GPIO to userspace */
  100. int gpiod_export(struct gpio_desc *desc, bool direction_may_change);
  101. /* reverse gpio_export() */
  102. void gpiod_unexport(struct gpio_desc *desc);
  103. /* create a sysfs link to an exported GPIO node */
  104. int gpiod_export_link(struct device *dev, const char *name,
  105. struct gpio_desc *desc);
  106. After a kernel driver requests a GPIO, it may only be made available in
  107. the sysfs interface by gpiod_export(). The driver can control whether the
  108. signal direction may change. This helps drivers prevent userspace code
  109. from accidentally clobbering important system state.
  110. This explicit exporting can help with debugging (by making some kinds
  111. of experiments easier), or can provide an always-there interface that's
  112. suitable for documenting as part of a board support package.
  113. After the GPIO has been exported, gpiod_export_link() allows creating
  114. symlinks from elsewhere in sysfs to the GPIO sysfs node. Drivers can
  115. use this to provide the interface under their own device in sysfs with
  116. a descriptive name.