| 1 |
.. SPDX-License-Identifier: GPL-2.0 |
| 2 |
.. include:: <isonum.txt> |
| 3 |
|
| 4 |
========================================= |
| 5 |
Why using ACPI drivers is not a good idea |
| 6 |
========================================= |
| 7 |
|
| 8 |
:Copyright: |copy| 2026, Intel Corporation |
| 9 |
|
| 10 |
:Author: Rafael J. Wysocki <[email protected]> |
| 11 |
|
| 12 |
Even though binding drivers directly to struct acpi_device objects, also |
| 13 |
referred to as "ACPI device nodes", allows basic functionality to be provided |
| 14 |
at least in some cases, there are problems with it, related to general |
| 15 |
consistency, sysfs layout, power management operation ordering, and code |
| 16 |
cleanliness. |
| 17 |
|
| 18 |
First of all, ACPI device nodes represent firmware entities rather than |
| 19 |
hardware and in many cases they provide auxiliary information on devices |
| 20 |
enumerated independently (like PCI devices or CPUs). It is therefore generally |
| 21 |
questionable to assign resources to them because the entities represented by |
| 22 |
them do not decode addresses in the memory or I/O address spaces and do not |
| 23 |
generate interrupts or similar (all of that is done by hardware). |
| 24 |
|
| 25 |
Second, as a general rule, a struct acpi_device can only be a parent of another |
| 26 |
struct acpi_device. If that is not the case, the location of the child device |
| 27 |
in the device hierarchy is at least confusing and it may not be straightforward |
| 28 |
to identify the piece of hardware providing functionality represented by it. |
| 29 |
However, binding a driver directly to an ACPI device node may cause that to |
| 30 |
happen if the given driver registers input devices or wakeup sources under it, |
| 31 |
for example. |
| 32 |
|
| 33 |
Next, using system suspend and resume callbacks directly on ACPI device nodes |
| 34 |
is also questionable because it may cause ordering problems to appear. Namely, |
| 35 |
ACPI device nodes are registered before enumerating hardware corresponding to |
| 36 |
them and they land on the PM list in front of the majority of other device |
| 37 |
objects. Consequently, the execution ordering of their PM callbacks may be |
| 38 |
different from what is generally expected. Also, in general, dependencies |
| 39 |
returned by _DEP objects do not affect ACPI device nodes themselves, but the |
| 40 |
"physical" devices associated with them, which potentially is one more source |
| 41 |
of inconsistency related to treating ACPI device nodes as "real" device |
| 42 |
representation. |
| 43 |
|
| 44 |
All of the above means that binding drivers to ACPI device nodes should |
| 45 |
generally be avoided and so struct acpi_driver objects should not be used. |
| 46 |
|
| 47 |
Moreover, a device ID is necessary to bind a driver directly to an ACPI device |
| 48 |
node, but device IDs are not generally associated with all of them. Some of |
| 49 |
them contain alternative information allowing the corresponding pieces of |
| 50 |
hardware to be identified, for example represeted by an _ADR object return |
| 51 |
value, and device IDs are not used in those cases. In consequence, confusingly |
| 52 |
enough, binding an ACPI driver to an ACPI device node may even be impossible. |
| 53 |
|
| 54 |
When that happens, the piece of hardware corresponding to the given ACPI device |
| 55 |
node is represented by another device object, like a struct pci_dev, and the |
| 56 |
ACPI device node is the "ACPI companion" of that device, accessible through its |
| 57 |
fwnode pointer used by the ACPI_COMPANION() macro. The ACPI companion holds |
| 58 |
additional information on the device configuration and possibly some "recipes" |
| 59 |
on device manipulation in the form of AML (ACPI Machine Language) bytecode |
| 60 |
provided by the platform firmware. Thus the role of the ACPI device node is |
| 61 |
similar to the role of a struct device_node on a system where Device Tree is |
| 62 |
used for platform description. |
| 63 |
|
| 64 |
For consistency, this approach has been extended to the cases in which ACPI |
| 65 |
device IDs are used. Namely, in those cases, an additional device object is |
| 66 |
created to represent the piece of hardware corresponding to a given ACPI device |
| 67 |
node. By default, it is a platform device, but it may also be a PNP device, a |
| 68 |
CPU device, or another type of device, depending on what the given piece of |
| 69 |
hardware actually is. There are even cases in which multiple devices are |
| 70 |
"backed" or "accompanied" by one ACPI device node (e.g. ACPI device nodes |
| 71 |
corresponding to GPUs that may provide firmware interfaces for backlight |
| 72 |
brightness control in addition to GPU configuration information). |
| 73 |
|
| 74 |
This means that it really should never be necessary to bind a driver directly to |
| 75 |
an ACPI device node because there is a "proper" device object representing the |
| 76 |
corresponding piece of hardware that can be bound to by a "proper" driver using |
| 77 |
the given ACPI device node as the device's ACPI companion. Thus, in principle, |
| 78 |
there is no reason to use ACPI drivers and if they all were replaced with other |
| 79 |
driver types (for example, platform drivers), some code could be dropped and |
| 80 |
some complexity would go away. |