brintos

brintos / linux-shallow public Read only

0
0
Text · 9.4 KiB · abad335 Raw
193 lines · plain
1Rules on how to access information in sysfs2===========================================3 4The kernel-exported sysfs exports internal kernel implementation details5and depends on internal kernel structures and layout. It is agreed upon6by the kernel developers that the Linux kernel does not provide a stable7internal API. Therefore, there are aspects of the sysfs interface that8may not be stable across kernel releases.9 10To minimize the risk of breaking users of sysfs, which are in most cases11low-level userspace applications, with a new kernel release, the users12of sysfs must follow some rules to use an as-abstract-as-possible way to13access this filesystem. The current udev and HAL programs already14implement this and users are encouraged to plug, if possible, into the15abstractions these programs provide instead of accessing sysfs directly.16 17But if you really do want or need to access sysfs directly, please follow18the following rules and then your programs should work with future19versions of the sysfs interface.20 21- Do not use libsysfs22    It makes assumptions about sysfs which are not true. Its API does not23    offer any abstraction, it exposes all the kernel driver-core24    implementation details in its own API. Therefore it is not better than25    reading directories and opening the files yourself.26    Also, it is not actively maintained, in the sense of reflecting the27    current kernel development. The goal of providing a stable interface28    to sysfs has failed; it causes more problems than it solves. It29    violates many of the rules in this document.30 31- sysfs is always at ``/sys``32    Parsing ``/proc/mounts`` is a waste of time. Other mount points are a33    system configuration bug you should not try to solve. For test cases,34    possibly support a ``SYSFS_PATH`` environment variable to overwrite the35    application's behavior, but never try to search for sysfs. Never try36    to mount it, if you are not an early boot script.37 38- devices are only "devices"39    There is no such thing like class-, bus-, physical devices,40    interfaces, and such that you can rely on in userspace. Everything is41    just simply a "device". Class-, bus-, physical, ... types are just42    kernel implementation details which should not be expected by43    applications that look for devices in sysfs.44 45    The properties of a device are:46 47    - devpath (``/devices/pci0000:00/0000:00:1d.1/usb2/2-2/2-2:1.0``)48 49      - identical to the DEVPATH value in the event sent from the kernel50        at device creation and removal51      - the unique key to the device at that point in time52      - the kernel's path to the device directory without the leading53        ``/sys``, and always starting with a slash54      - all elements of a devpath must be real directories. Symlinks55        pointing to /sys/devices must always be resolved to their real56        target and the target path must be used to access the device.57        That way the devpath to the device matches the devpath of the58        kernel used at event time.59      - using or exposing symlink values as elements in a devpath string60        is a bug in the application61 62    - kernel name (``sda``, ``tty``, ``0000:00:1f.2``, ...)63 64      - a directory name, identical to the last element of the devpath65      - applications need to handle spaces and characters like ``!`` in66        the name67 68    - subsystem (``block``, ``tty``, ``pci``, ...)69 70      - simple string, never a path or a link71      - retrieved by reading the "subsystem"-link and using only the72        last element of the target path73 74    - driver (``tg3``, ``ata_piix``, ``uhci_hcd``)75 76      - a simple string, which may contain spaces, never a path or a77        link78      - it is retrieved by reading the "driver"-link and using only the79        last element of the target path80      - devices which do not have "driver"-link just do not have a81        driver; copying the driver value in a child device context is a82        bug in the application83 84    - attributes85 86      - the files in the device directory or files below subdirectories87        of the same device directory88      - accessing attributes reached by a symlink pointing to another device,89        like the "device"-link, is a bug in the application90 91    Everything else is just a kernel driver-core implementation detail92    that should not be assumed to be stable across kernel releases.93 94- Properties of parent devices never belong into a child device.95    Always look at the parent devices themselves for determining device96    context properties. If the device ``eth0`` or ``sda`` does not have a97    "driver"-link, then this device does not have a driver. Its value is empty.98    Never copy any property of the parent-device into a child-device. Parent99    device properties may change dynamically without any notice to the100    child device.101 102- Hierarchy in a single device tree103    There is only one valid place in sysfs where hierarchy can be examined104    and this is below: ``/sys/devices.``105    It is planned that all device directories will end up in the tree106    below this directory.107 108- Classification by subsystem109    There are currently three places for classification of devices:110    ``/sys/block,`` ``/sys/class`` and ``/sys/bus.`` It is planned that these will111    not contain any device directories themselves, but only flat lists of112    symlinks pointing to the unified ``/sys/devices`` tree.113    All three places have completely different rules on how to access114    device information. It is planned to merge all three115    classification directories into one place at ``/sys/subsystem``,116    following the layout of the bus directories. All buses and117    classes, including the converted block subsystem, will show up118    there.119    The devices belonging to a subsystem will create a symlink in the120    "devices" directory at ``/sys/subsystem/<name>/devices``,121 122    If ``/sys/subsystem`` exists, ``/sys/bus``, ``/sys/class`` and ``/sys/block``123    can be ignored. If it does not exist, you always have to scan all three124    places, as the kernel is free to move a subsystem from one place to125    the other, as long as the devices are still reachable by the same126    subsystem name.127 128    Assuming ``/sys/class/<subsystem>`` and ``/sys/bus/<subsystem>``, or129    ``/sys/block`` and ``/sys/class/block`` are not interchangeable is a bug in130    the application.131 132- Block133    The converted block subsystem at ``/sys/class/block`` or134    ``/sys/subsystem/block`` will contain the links for disks and partitions135    at the same level, never in a hierarchy. Assuming the block subsystem to136    contain only disks and not partition devices in the same flat list is137    a bug in the application.138 139- "device"-link and <subsystem>:<kernel name>-links140    Never depend on the "device"-link. The "device"-link is a workaround141    for the old layout, where class devices are not created in142    ``/sys/devices/`` like the bus devices. If the link-resolving of a143    device directory does not end in ``/sys/devices/``, you can use the144    "device"-link to find the parent devices in ``/sys/devices/``, That is the145    single valid use of the "device"-link; it must never appear in any146    path as an element. Assuming the existence of the "device"-link for147    a device in ``/sys/devices/`` is a bug in the application.148    Accessing ``/sys/class/net/eth0/device`` is a bug in the application.149 150    Never depend on the class-specific links back to the ``/sys/class``151    directory.  These links are also a workaround for the design mistake152    that class devices are not created in ``/sys/devices.`` If a device153    directory does not contain directories for child devices, these links154    may be used to find the child devices in ``/sys/class.`` That is the single155    valid use of these links; they must never appear in any path as an156    element. Assuming the existence of these links for devices which are157    real child device directories in the ``/sys/devices`` tree is a bug in158    the application.159 160    It is planned to remove all these links when all class device161    directories live in ``/sys/devices.``162 163- Position of devices along device chain can change.164    Never depend on a specific parent device position in the devpath,165    or the chain of parent devices. The kernel is free to insert devices into166    the chain. You must always request the parent device you are looking for167    by its subsystem value. You need to walk up the chain until you find168    the device that matches the expected subsystem. Depending on a specific169    position of a parent device or exposing relative paths using ``../`` to170    access the chain of parents is a bug in the application.171 172- When reading and writing sysfs device attribute files, avoid dependency173    on specific error codes wherever possible. This minimizes coupling to174    the error handling implementation within the kernel.175 176    In general, failures to read or write sysfs device attributes shall177    propagate errors wherever possible. Common errors include, but are not178    limited to:179 180	``-EIO``: The read or store operation is not supported, typically181	returned by the sysfs system itself if the read or store pointer182	is ``NULL``.183 184	``-ENXIO``: The read or store operation failed185 186    Error codes will not be changed without good reason, and should a change187    to error codes result in user-space breakage, it will be fixed, or the188    the offending change will be reverted.189 190    Userspace applications can, however, expect the format and contents of191    the attribute files to remain consistent in the absence of a version192    attribute change in the context of a given attribute.193