576 lines · plain
1===================2ACPI on Arm systems3===================4 5ACPI can be used for Armv8 and Armv9 systems designed to follow6the BSA (Arm Base System Architecture) [0] and BBR (Arm7Base Boot Requirements) [1] specifications. Both BSA and BBR are publicly8accessible documents.9Arm Servers, in addition to being BSA compliant, comply with a set10of rules defined in SBSA (Server Base System Architecture) [2].11 12The Arm kernel implements the reduced hardware model of ACPI version135.1 or later. Links to the specification and all external documents14it refers to are managed by the UEFI Forum. The specification is15available at http://www.uefi.org/specifications and documents referenced16by the specification can be found via http://www.uefi.org/acpi.17 18If an Arm system does not meet the requirements of the BSA and BBR,19or cannot be described using the mechanisms defined in the required ACPI20specifications, then ACPI may not be a good fit for the hardware.21 22While the documents mentioned above set out the requirements for building23industry-standard Arm systems, they also apply to more than one operating24system. The purpose of this document is to describe the interaction between25ACPI and Linux only, on an Arm system -- that is, what Linux expects of26ACPI and what ACPI can expect of Linux.27 28 29Why ACPI on Arm?30----------------31Before examining the details of the interface between ACPI and Linux, it is32useful to understand why ACPI is being used. Several technologies already33exist in Linux for describing non-enumerable hardware, after all. In this34section we summarize a blog post [3] from Grant Likely that outlines the35reasoning behind ACPI on Arm systems. Actually, we snitch a good portion36of the summary text almost directly, to be honest.37 38The short form of the rationale for ACPI on Arm is:39 40- ACPI’s byte code (AML) allows the platform to encode hardware behavior,41 while DT explicitly does not support this. For hardware vendors, being42 able to encode behavior is a key tool used in supporting operating43 system releases on new hardware.44 45- ACPI’s OSPM defines a power management model that constrains what the46 platform is allowed to do into a specific model, while still providing47 flexibility in hardware design.48 49- In the enterprise server environment, ACPI has established bindings (such50 as for RAS) which are currently used in production systems. DT does not.51 Such bindings could be defined in DT at some point, but doing so means Arm52 and x86 would end up using completely different code paths in both firmware53 and the kernel.54 55- Choosing a single interface to describe the abstraction between a platform56 and an OS is important. Hardware vendors would not be required to implement57 both DT and ACPI if they want to support multiple operating systems. And,58 agreeing on a single interface instead of being fragmented into per OS59 interfaces makes for better interoperability overall.60 61- The new ACPI governance process works well and Linux is now at the same62 table as hardware vendors and other OS vendors. In fact, there is no63 longer any reason to feel that ACPI only belongs to Windows or that64 Linux is in any way secondary to Microsoft in this arena. The move of65 ACPI governance into the UEFI forum has significantly opened up the66 specification development process, and currently, a large portion of the67 changes being made to ACPI are being driven by Linux.68 69Key to the use of ACPI is the support model. For servers in general, the70responsibility for hardware behaviour cannot solely be the domain of the71kernel, but rather must be split between the platform and the kernel, in72order to allow for orderly change over time. ACPI frees the OS from needing73to understand all the minute details of the hardware so that the OS doesn’t74need to be ported to each and every device individually. It allows the75hardware vendors to take responsibility for power management behaviour without76depending on an OS release cycle which is not under their control.77 78ACPI is also important because hardware and OS vendors have already worked79out the mechanisms for supporting a general purpose computing ecosystem. The80infrastructure is in place, the bindings are in place, and the processes are81in place. DT does exactly what Linux needs it to when working with vertically82integrated devices, but there are no good processes for supporting what the83server vendors need. Linux could potentially get there with DT, but doing so84really just duplicates something that already works. ACPI already does what85the hardware vendors need, Microsoft won’t collaborate on DT, and hardware86vendors would still end up providing two completely separate firmware87interfaces -- one for Linux and one for Windows.88 89 90Kernel Compatibility91--------------------92One of the primary motivations for ACPI is standardization, and using that93to provide backward compatibility for Linux kernels. In the server market,94software and hardware are often used for long periods. ACPI allows the95kernel and firmware to agree on a consistent abstraction that can be96maintained over time, even as hardware or software change. As long as the97abstraction is supported, systems can be updated without necessarily having98to replace the kernel.99 100When a Linux driver or subsystem is first implemented using ACPI, it by101definition ends up requiring a specific version of the ACPI specification102-- its baseline. ACPI firmware must continue to work, even though it may103not be optimal, with the earliest kernel version that first provides support104for that baseline version of ACPI. There may be a need for additional drivers,105but adding new functionality (e.g., CPU power management) should not break106older kernel versions. Further, ACPI firmware must also work with the most107recent version of the kernel.108 109 110Relationship with Device Tree111-----------------------------112ACPI support in drivers and subsystems for Arm should never be mutually113exclusive with DT support at compile time.114 115At boot time the kernel will only use one description method depending on116parameters passed from the boot loader (including kernel bootargs).117 118Regardless of whether DT or ACPI is used, the kernel must always be capable119of booting with either scheme (in kernels with both schemes enabled at compile120time).121 122 123Booting using ACPI tables124-------------------------125The only defined method for passing ACPI tables to the kernel on Arm126is via the UEFI system configuration table. Just so it is explicit, this127means that ACPI is only supported on platforms that boot via UEFI.128 129When an Arm system boots, it can either have DT information, ACPI tables,130or in some very unusual cases, both. If no command line parameters are used,131the kernel will try to use DT for device enumeration; if there is no DT132present, the kernel will try to use ACPI tables, but only if they are present.133If neither is available, the kernel will not boot. If acpi=force is used134on the command line, the kernel will attempt to use ACPI tables first, but135fall back to DT if there are no ACPI tables present. The basic idea is that136the kernel will not fail to boot unless it absolutely has no other choice.137 138Processing of ACPI tables may be disabled by passing acpi=off on the kernel139command line; this is the default behavior.140 141In order for the kernel to load and use ACPI tables, the UEFI implementation142MUST set the ACPI_20_TABLE_GUID to point to the RSDP table (the table with143the ACPI signature "RSD PTR "). If this pointer is incorrect and acpi=force144is used, the kernel will disable ACPI and try to use DT to boot instead; the145kernel has, in effect, determined that ACPI tables are not present at that146point.147 148If the pointer to the RSDP table is correct, the table will be mapped into149the kernel by the ACPI core, using the address provided by UEFI.150 151The ACPI core will then locate and map in all other ACPI tables provided by152using the addresses in the RSDP table to find the XSDT (eXtended System153Description Table). The XSDT in turn provides the addresses to all other154ACPI tables provided by the system firmware; the ACPI core will then traverse155this table and map in the tables listed.156 157The ACPI core will ignore any provided RSDT (Root System Description Table).158RSDTs have been deprecated and are ignored on arm64 since they only allow159for 32-bit addresses.160 161Further, the ACPI core will only use the 64-bit address fields in the FADT162(Fixed ACPI Description Table). Any 32-bit address fields in the FADT will163be ignored on arm64.164 165Hardware reduced mode (see Section 4.1 of the ACPI 6.1 specification) will166be enforced by the ACPI core on arm64. Doing so allows the ACPI core to167run less complex code since it no longer has to provide support for legacy168hardware from other architectures. Any fields that are not to be used for169hardware reduced mode must be set to zero.170 171For the ACPI core to operate properly, and in turn provide the information172the kernel needs to configure devices, it expects to find the following173tables (all section numbers refer to the ACPI 6.5 specification):174 175 - RSDP (Root System Description Pointer), section 5.2.5176 177 - XSDT (eXtended System Description Table), section 5.2.8178 179 - FADT (Fixed ACPI Description Table), section 5.2.9180 181 - DSDT (Differentiated System Description Table), section182 5.2.11.1183 184 - MADT (Multiple APIC Description Table), section 5.2.12185 186 - GTDT (Generic Timer Description Table), section 5.2.24187 188 - PPTT (Processor Properties Topology Table), section 5.2.30189 190 - DBG2 (DeBuG port table 2), section 5.2.6, specifically Table 5-6.191 192 - APMT (Arm Performance Monitoring unit Table), section 5.2.6, specifically Table 5-6.193 194 - AGDI (Arm Generic diagnostic Dump and Reset Device Interface Table), section 5.2.6, specifically Table 5-6.195 196 - If PCI is supported, the MCFG (Memory mapped ConFiGuration197 Table), section 5.2.6, specifically Table 5-6.198 199 - If booting without a console=<device> kernel parameter is200 supported, the SPCR (Serial Port Console Redirection table),201 section 5.2.6, specifically Table 5-6.202 203 - If necessary to describe the I/O topology, SMMUs and GIC ITSs,204 the IORT (Input Output Remapping Table, section 5.2.6, specifically205 Table 5-6).206 207 - If NUMA is supported, the following tables are required:208 209 - SRAT (System Resource Affinity Table), section 5.2.16210 211 - SLIT (System Locality distance Information Table), section 5.2.17212 213 - If NUMA is supported, and the system contains heterogeneous memory,214 the HMAT (Heterogeneous Memory Attribute Table), section 5.2.28.215 216 - If the ACPI Platform Error Interfaces are required, the following217 tables are conditionally required:218 219 - BERT (Boot Error Record Table, section 18.3.1)220 221 - EINJ (Error INJection table, section 18.6.1)222 223 - ERST (Error Record Serialization Table, section 18.5)224 225 - HEST (Hardware Error Source Table, section 18.3.2)226 227 - SDEI (Software Delegated Exception Interface table, section 5.2.6,228 specifically Table 5-6)229 230 - AEST (Arm Error Source Table, section 5.2.6,231 specifically Table 5-6)232 233 - RAS2 (ACPI RAS2 feature table, section 5.2.21)234 235 - If the system contains controllers using PCC channel, the236 PCCT (Platform Communications Channel Table), section 14.1237 238 - If the system contains a controller to capture board-level system state,239 and communicates with the host via PCC, the PDTT (Platform Debug Trigger240 Table), section 5.2.29.241 242 - If NVDIMM is supported, the NFIT (NVDIMM Firmware Interface Table), section 5.2.26243 244 - If video framebuffer is present, the BGRT (Boot Graphics Resource Table), section 5.2.23245 246 - If IPMI is implemented, the SPMI (Server Platform Management Interface),247 section 5.2.6, specifically Table 5-6.248 249 - If the system contains a CXL Host Bridge, the CEDT (CXL Early Discovery250 Table), section 5.2.6, specifically Table 5-6.251 252 - If the system supports MPAM, the MPAM (Memory Partitioning And Monitoring table), section 5.2.6,253 specifically Table 5-6.254 255 - If the system lacks persistent storage, the IBFT (ISCSI Boot Firmware256 Table), section 5.2.6, specifically Table 5-6.257 258 259If the above tables are not all present, the kernel may or may not be260able to boot properly since it may not be able to configure all of the261devices available. This list of tables is not meant to be all inclusive;262in some environments other tables may be needed (e.g., any of the APEI263tables from section 18) to support specific functionality.264 265 266ACPI Detection267--------------268Drivers should determine their probe() type by checking for a null269value for ACPI_HANDLE, or checking .of_node, or other information in270the device structure. This is detailed further in the "Driver271Recommendations" section.272 273In non-driver code, if the presence of ACPI needs to be detected at274run time, then check the value of acpi_disabled. If CONFIG_ACPI is not275set, acpi_disabled will always be 1.276 277 278Device Enumeration279------------------280Device descriptions in ACPI should use standard recognized ACPI interfaces.281These may contain less information than is typically provided via a Device282Tree description for the same device. This is also one of the reasons that283ACPI can be useful -- the driver takes into account that it may have less284detailed information about the device and uses sensible defaults instead.285If done properly in the driver, the hardware can change and improve over286time without the driver having to change at all.287 288Clocks provide an excellent example. In DT, clocks need to be specified289and the drivers need to take them into account. In ACPI, the assumption290is that UEFI will leave the device in a reasonable default state, including291any clock settings. If for some reason the driver needs to change a clock292value, this can be done in an ACPI method; all the driver needs to do is293invoke the method and not concern itself with what the method needs to do294to change the clock. Changing the hardware can then take place over time295by changing what the ACPI method does, and not the driver.296 297In DT, the parameters needed by the driver to set up clocks as in the example298above are known as "bindings"; in ACPI, these are known as "Device Properties"299and provided to a driver via the _DSD object.300 301ACPI tables are described with a formal language called ASL, the ACPI302Source Language (section 19 of the specification). This means that there303are always multiple ways to describe the same thing -- including device304properties. For example, device properties could use an ASL construct305that looks like this: Name(KEY0, "value0"). An ACPI device driver would306then retrieve the value of the property by evaluating the KEY0 object.307However, using Name() this way has multiple problems: (1) ACPI limits308names ("KEY0") to four characters unlike DT; (2) there is no industry309wide registry that maintains a list of names, minimizing re-use; (3)310there is also no registry for the definition of property values ("value0"),311again making re-use difficult; and (4) how does one maintain backward312compatibility as new hardware comes out? The _DSD method was created313to solve precisely these sorts of problems; Linux drivers should ALWAYS314use the _DSD method for device properties and nothing else.315 316The _DSM object (ACPI Section 9.14.1) could also be used for conveying317device properties to a driver. Linux drivers should only expect it to318be used if _DSD cannot represent the data required, and there is no way319to create a new UUID for the _DSD object. Note that there is even less320regulation of the use of _DSM than there is of _DSD. Drivers that depend321on the contents of _DSM objects will be more difficult to maintain over322time because of this; as of this writing, the use of _DSM is the cause323of quite a few firmware problems and is not recommended.324 325Drivers should look for device properties in the _DSD object ONLY; the _DSD326object is described in the ACPI specification section 6.2.5, but this only327describes how to define the structure of an object returned via _DSD, and328how specific data structures are defined by specific UUIDs. Linux should329only use the _DSD Device Properties UUID [4]:330 331 - UUID: daffd814-6eba-4d8c-8a91-bc9bbf4aa301332 333Common device properties can be registered by creating a pull request to [4] so334that they may be used across all operating systems supporting ACPI.335Device properties that have not been registered with the UEFI Forum can be used336but not as "uefi-" common properties.337 338Before creating new device properties, check to be sure that they have not339been defined before and either registered in the Linux kernel documentation340as DT bindings, or the UEFI Forum as device properties. While we do not want341to simply move all DT bindings into ACPI device properties, we can learn from342what has been previously defined.343 344If it is necessary to define a new device property, or if it makes sense to345synthesize the definition of a binding so it can be used in any firmware,346both DT bindings and ACPI device properties for device drivers have review347processes. Use them both. When the driver itself is submitted for review348to the Linux mailing lists, the device property definitions needed must be349submitted at the same time. A driver that supports ACPI and uses device350properties will not be considered complete without their definitions. Once351the device property has been accepted by the Linux community, it must be352registered with the UEFI Forum [4], which will review it again for consistency353within the registry. This may require iteration. The UEFI Forum, though,354will always be the canonical site for device property definitions.355 356It may make sense to provide notice to the UEFI Forum that there is the357intent to register a previously unused device property name as a means of358reserving the name for later use. Other operating system vendors will359also be submitting registration requests and this may help smooth the360process.361 362Once registration and review have been completed, the kernel provides an363interface for looking up device properties in a manner independent of364whether DT or ACPI is being used. This API should be used [5]; it can365eliminate some duplication of code paths in driver probing functions and366discourage divergence between DT bindings and ACPI device properties.367 368 369Programmable Power Control Resources370------------------------------------371Programmable power control resources include such resources as voltage/current372providers (regulators) and clock sources.373 374With ACPI, the kernel clock and regulator framework is not expected to be used375at all.376 377The kernel assumes that power control of these resources is represented with378Power Resource Objects (ACPI section 7.1). The ACPI core will then handle379correctly enabling and disabling resources as they are needed. In order to380get that to work, ACPI assumes each device has defined D-states and that these381can be controlled through the optional ACPI methods _PS0, _PS1, _PS2, and _PS3;382in ACPI, _PS0 is the method to invoke to turn a device full on, and _PS3 is for383turning a device full off.384 385There are two options for using those Power Resources. They can:386 387 - be managed in a _PSx method which gets called on entry to power388 state Dx.389 390 - be declared separately as power resources with their own _ON and _OFF391 methods. They are then tied back to D-states for a particular device392 via _PRx which specifies which power resources a device needs to be on393 while in Dx. Kernel then tracks number of devices using a power resource394 and calls _ON/_OFF as needed.395 396The kernel ACPI code will also assume that the _PSx methods follow the normal397ACPI rules for such methods:398 399 - If either _PS0 or _PS3 is implemented, then the other method must also400 be implemented.401 402 - If a device requires usage or setup of a power resource when on, the ASL403 should organize that it is allocated/enabled using the _PS0 method.404 405 - Resources allocated or enabled in the _PS0 method should be disabled406 or de-allocated in the _PS3 method.407 408 - Firmware will leave the resources in a reasonable state before handing409 over control to the kernel.410 411Such code in _PSx methods will of course be very platform specific. But,412this allows the driver to abstract out the interface for operating the device413and avoid having to read special non-standard values from ACPI tables. Further,414abstracting the use of these resources allows the hardware to change over time415without requiring updates to the driver.416 417 418Clocks419------420ACPI makes the assumption that clocks are initialized by the firmware --421UEFI, in this case -- to some working value before control is handed over422to the kernel. This has implications for devices such as UARTs, or SoC-driven423LCD displays, for example.424 425When the kernel boots, the clocks are assumed to be set to reasonable426working values. If for some reason the frequency needs to change -- e.g.,427throttling for power management -- the device driver should expect that428process to be abstracted out into some ACPI method that can be invoked429(please see the ACPI specification for further recommendations on standard430methods to be expected). The only exceptions to this are CPU clocks where431CPPC provides a much richer interface than ACPI methods. If the clocks432are not set, there is no direct way for Linux to control them.433 434If an SoC vendor wants to provide fine-grained control of the system clocks,435they could do so by providing ACPI methods that could be invoked by Linux436drivers. However, this is NOT recommended and Linux drivers should NOT use437such methods, even if they are provided. Such methods are not currently438standardized in the ACPI specification, and using them could tie a kernel439to a very specific SoC, or tie an SoC to a very specific version of the440kernel, both of which we are trying to avoid.441 442 443Driver Recommendations444----------------------445DO NOT remove any DT handling when adding ACPI support for a driver. The446same device may be used on many different systems.447 448DO try to structure the driver so that it is data-driven. That is, set up449a struct containing internal per-device state based on defaults and whatever450else must be discovered by the driver probe function. Then, have the rest451of the driver operate off of the contents of that struct. Doing so should452allow most divergence between ACPI and DT functionality to be kept local to453the probe function instead of being scattered throughout the driver. For454example::455 456 static int device_probe_dt(struct platform_device *pdev)457 {458 /* DT specific functionality */459 ...460 }461 462 static int device_probe_acpi(struct platform_device *pdev)463 {464 /* ACPI specific functionality */465 ...466 }467 468 static int device_probe(struct platform_device *pdev)469 {470 ...471 struct device_node node = pdev->dev.of_node;472 ...473 474 if (node)475 ret = device_probe_dt(pdev);476 else if (ACPI_HANDLE(&pdev->dev))477 ret = device_probe_acpi(pdev);478 else479 /* other initialization */480 ...481 /* Continue with any generic probe operations */482 ...483 }484 485DO keep the MODULE_DEVICE_TABLE entries together in the driver to make it486clear the different names the driver is probed for, both from DT and from487ACPI::488 489 static struct of_device_id virtio_mmio_match[] = {490 { .compatible = "virtio,mmio", },491 { }492 };493 MODULE_DEVICE_TABLE(of, virtio_mmio_match);494 495 static const struct acpi_device_id virtio_mmio_acpi_match[] = {496 { "LNRO0005", },497 { }498 };499 MODULE_DEVICE_TABLE(acpi, virtio_mmio_acpi_match);500 501 502ASWG503----504The ACPI specification changes regularly. During the year 2014, for instance,505version 5.1 was released and version 6.0 substantially completed, with most of506the changes being driven by Arm-specific requirements. Proposed changes are507presented and discussed in the ASWG (ACPI Specification Working Group) which508is a part of the UEFI Forum. The current version of the ACPI specification509is 6.5 release in August 2022.510 511Participation in this group is open to all UEFI members. Please see512http://www.uefi.org/workinggroup for details on group membership.513 514It is the intent of the Arm ACPI kernel code to follow the ACPI specification515as closely as possible, and to only implement functionality that complies with516the released standards from UEFI ASWG. As a practical matter, there will be517vendors that provide bad ACPI tables or violate the standards in some way.518If this is because of errors, quirks and fix-ups may be necessary, but will519be avoided if possible. If there are features missing from ACPI that preclude520it from being used on a platform, ECRs (Engineering Change Requests) should be521submitted to ASWG and go through the normal approval process; for those that522are not UEFI members, many other members of the Linux community are and would523likely be willing to assist in submitting ECRs.524 525 526Linux Code527----------528Individual items specific to Linux on Arm, contained in the Linux529source code, are in the list that follows:530 531ACPI_OS_NAME532 This macro defines the string to be returned when533 an ACPI method invokes the _OS method. On Arm534 systems, this macro will be "Linux" by default.535 The command line parameter acpi_os=<string>536 can be used to set it to some other value. The537 default value for other architectures is "Microsoft538 Windows NT", for example.539 540ACPI Objects541------------542Detailed expectations for ACPI tables and object are listed in the file543Documentation/arch/arm64/acpi_object_usage.rst.544 545 546References547----------548[0] https://developer.arm.com/documentation/den0094/latest549 document Arm-DEN-0094: "Arm Base System Architecture", version 1.0C, dated 6 Oct 2022550 551[1] https://developer.arm.com/documentation/den0044/latest552 Document Arm-DEN-0044: "Arm Base Boot Requirements", version 2.0G, dated 15 Apr 2022553 554[2] https://developer.arm.com/documentation/den0029/latest555 Document Arm-DEN-0029: "Arm Server Base System Architecture", version 7.1, dated 06 Oct 2022556 557[3] http://www.secretlab.ca/archives/151,558 10 Jan 2015, Copyright (c) 2015,559 Linaro Ltd., written by Grant Likely.560 561[4] _DSD (Device Specific Data) Implementation Guide562 https://github.com/UEFI/DSD-Guide/blob/main/dsd-guide.pdf563 564[5] Kernel code for the unified device565 property interface can be found in566 include/linux/property.h and drivers/base/property.c.567 568 569Authors570-------571- Al Stone <al.stone@linaro.org>572- Graeme Gregory <graeme.gregory@linaro.org>573- Hanjun Guo <hanjun.guo@linaro.org>574 575- Grant Likely <grant.likely@linaro.org>, for the "Why ACPI on ARM?" section576