1 .. SPDX-License-Identifier: GPL-2.0
3 ==========================
4 ACPI _OSI and _REV methods
5 ==========================
7 An ACPI BIOS can use the "Operating System Interfaces" method (_OSI)
8 to find out what the operating system supports. Eg. If BIOS
9 AML code includes _OSI("XYZ"), the kernel's AML interpreter
10 can evaluate that method, look to see if it supports 'XYZ'
11 and answer YES or NO to the BIOS.
13 The ACPI _REV method returns the "Revision of the ACPI specification
16 This document explains how and why the BIOS and Linux should use these methods.
17 It also explains how and why they are widely misused.
22 Linux runs on two groups of machines -- those that are tested by the OEM
23 to be compatible with Linux, and those that were never tested with Linux,
24 but where Linux was installed to replace the original OS (Windows or OSX).
26 The larger group is the systems tested to run only Windows. Not only that,
27 but many were tested to run with just one specific version of Windows.
28 So even though the BIOS may use _OSI to query what version of Windows is running,
29 only a single path through the BIOS has actually been tested.
30 Experience shows that taking untested paths through the BIOS
31 exposes Linux to an entire category of BIOS bugs.
32 For this reason, Linux _OSI defaults must continue to claim compatibility
33 with all versions of Windows.
35 But Linux isn't actually compatible with Windows, and the Linux community
36 has also been hurt with regressions when Linux adds the latest version of
37 Windows to its list of _OSI strings. So it is possible that additional strings
38 will be more thoroughly vetted before shipping upstream in the future.
39 But it is likely that they will all eventually be added.
41 What should an OEM do if they want to support Linux and Windows
42 using the same BIOS image? Often they need to do something different
43 for Linux to deal with how Linux is different from Windows.
44 Here the BIOS should ask exactly what it wants to know:
46 _OSI("Linux-OEM-my_interface_name")
47 where 'OEM' is needed if this is an OEM-specific hook,
48 and 'my_interface_name' describes the hook, which could be a
49 quirk, a bug, or a bug-fix.
51 In addition, the OEM should send a patch to upstream Linux
52 via the linux-acpi@vger.kernel.org mailing list. When that patch
53 is checked into Linux, the OS will answer "YES" when the BIOS
54 on the OEM's system uses _OSI to ask if the interface is supported
55 by the OS. Linux distributors can back-port that patch for Linux
56 pre-installs, and it will be included by all distributions that
57 re-base to upstream. If the distribution can not update the kernel binary,
58 they can also add an acpi_osi=Linux-OEM-my_interface_name
59 cmdline parameter to the boot loader, as needed.
61 If the string refers to a feature where the upstream kernel
62 eventually grows support, a patch should be sent to remove
63 the string when that support is added to the kernel.
65 That was easy. Read on, to find out how to do it wrong.
67 Before _OSI, there was _OS
68 ==========================
70 ACPI 1.0 specified "_OS" as an
71 "object that evaluates to a string that identifies the operating system."
73 The ACPI BIOS flow would include an evaluation of _OS, and the AML
74 interpreter in the kernel would return to it a string identifying the OS:
76 Windows 98, SE: "Microsoft Windows"
77 Windows ME: "Microsoft WindowsME:Millenium Edition"
78 Windows NT: "Microsoft Windows NT"
80 The idea was on a platform tasked with running multiple OS's,
81 the BIOS could use _OS to enable devices that an OS
82 might support, or enable quirks or bug workarounds
83 necessary to make the platform compatible with that pre-existing OS.
85 But _OS had fundamental problems. First, the BIOS needed to know the name
86 of every possible version of the OS that would run on it, and needed to know
87 all the quirks of those OS's. Certainly it would make more sense
88 for the BIOS to ask *specific* things of the OS, such
89 "do you support a specific interface", and thus in ACPI 3.0,
90 _OSI was born to replace _OS.
92 _OS was abandoned, though even today, many BIOS look for
93 _OS "Microsoft Windows NT", though it seems somewhat far-fetched
94 that anybody would install those old operating systems
95 over what came with the machine.
97 Linux answers "Microsoft Windows NT" to please that BIOS idiom.
98 That is the *only* viable strategy, as that is what modern Windows does,
99 and so doing otherwise could steer the BIOS down an untested path.
101 _OSI is born, and immediately misused
102 =====================================
104 With _OSI, the *BIOS* provides the string describing an interface,
105 and asks the OS: "YES/NO, are you compatible with this interface?"
107 eg. _OSI("3.0 Thermal Model") would return TRUE if the OS knows how
108 to deal with the thermal extensions made to the ACPI 3.0 specification.
109 An old OS that doesn't know about those extensions would answer FALSE,
110 and a new OS may be able to return TRUE.
112 For an OS-specific interface, the ACPI spec said that the BIOS and the OS
113 were to agree on a string of the form such as "Windows-interface_name".
115 But two bad things happened. First, the Windows ecosystem used _OSI
116 not as designed, but as a direct replacement for _OS -- identifying
117 the OS version, rather than an OS supported interface. Indeed, right
118 from the start, the ACPI 3.0 spec itself codified this misuse
119 in example code using _OSI("Windows 2001").
121 This misuse was adopted and continues today.
123 Linux had no choice but to also return TRUE to _OSI("Windows 2001")
124 and its successors. To do otherwise would virtually guarantee breaking
125 a BIOS that has been tested only with that _OSI returning TRUE.
127 This strategy is problematic, as Linux is never completely compatible with
128 the latest version of Windows, and sometimes it takes more than a year
129 to iron out incompatibilities.
131 Not to be out-done, the Linux community made things worse by returning TRUE
132 to _OSI("Linux"). Doing so is even worse than the Windows misuse
133 of _OSI, as "Linux" does not even contain any version information.
134 _OSI("Linux") led to some BIOS' malfunctioning due to BIOS writer's
135 using it in untested BIOS flows. But some OEM's used _OSI("Linux")
136 in tested flows to support real Linux features. In 2009, Linux
137 removed _OSI("Linux"), and added a cmdline parameter to restore it
138 for legacy systems still needed it. Further a BIOS_BUG warning prints
139 for all BIOS's that invoke it.
141 No BIOS should use _OSI("Linux").
143 The result is a strategy for Linux to maximize compatibility with
144 ACPI BIOS that are tested on Windows machines. There is a real risk
145 of over-stating that compatibility; but the alternative has often been
146 catastrophic failure resulting from the BIOS taking paths that
147 were never validated under *any* OS.
152 Since _OSI("Linux") went away, some BIOS writers used _REV
153 to support Linux and Windows differences in the same BIOS.
155 _REV was defined in ACPI 1.0 to return the version of ACPI
156 supported by the OS and the OS AML interpreter.
158 Modern Windows returns _REV = 2. Linux used ACPI_CA_SUPPORT_LEVEL,
159 which would increment, based on the version of the spec supported.
161 Unfortunately, _REV was also misused. eg. some BIOS would check
162 for _REV = 3, and do something for Linux, but when Linux returned
163 _REV = 4, that support broke.
165 In response to this problem, Linux returns _REV = 2 always,
166 from mid-2015 onward. The ACPI specification will also be updated
167 to reflect that _REV is deprecated, and always returns 2.
169 Apple Mac and _OSI("Darwin")
170 ============================
172 On Apple's Mac platforms, the ACPI BIOS invokes _OSI("Darwin")
173 to determine if the machine is running Apple OSX.
175 Like Linux's _OSI("*Windows*") strategy, Linux defaults to
176 answering YES to _OSI("Darwin") to enable full access
177 to the hardware and validated BIOS paths seen by OSX.
178 Just like on Windows-tested platforms, this strategy has risks.
180 Starting in Linux-3.18, the kernel answered YES to _OSI("Darwin")
181 for the purpose of enabling Mac Thunderbolt support. Further,
182 if the kernel noticed _OSI("Darwin") being invoked, it additionally
183 disabled all _OSI("*Windows*") to keep poorly written Mac BIOS
184 from going down untested combinations of paths.
186 The Linux-3.18 change in default caused power regressions on Mac
187 laptops, and the 3.18 implementation did not allow changing
188 the default via cmdline "acpi_osi=!Darwin". Linux-4.7 fixed
189 the ability to use acpi_osi=!Darwin as a workaround, and
190 we hope to see Mac Thunderbolt power management support in Linux-4.11.