1                        Booting AArch64 Linux
   2                        =====================
   4Author: Will Deacon <>
   5Date  : 07 September 2012
   7This document is based on the ARM booting document by Russell King and
   8is relevant to all public releases of the AArch64 Linux kernel.
  10The AArch64 exception model is made up of a number of exception levels
  11(EL0 - EL3), with EL0 and EL1 having a secure and a non-secure
  12counterpart.  EL2 is the hypervisor level and exists only in non-secure
  13mode. EL3 is the highest priority level and exists only in secure mode.
  15For the purposes of this document, we will use the term `boot loader'
  16simply to define all software that executes on the CPU(s) before control
  17is passed to the Linux kernel.  This may include secure monitor and
  18hypervisor code, or it may just be a handful of instructions for
  19preparing a minimal boot environment.
  21Essentially, the boot loader should provide (as a minimum) the
  241. Setup and initialise the RAM
  252. Setup the device tree
  263. Decompress the kernel image
  274. Call the kernel image
  301. Setup and initialise RAM
  33Requirement: MANDATORY
  35The boot loader is expected to find and initialise all RAM that the
  36kernel will use for volatile data storage in the system.  It performs
  37this in a machine dependent manner.  (It may use internal algorithms
  38to automatically locate and size all RAM, or it may use knowledge of
  39the RAM in the machine, or any other method the boot loader designer
  40sees fit.)
  432. Setup the device tree
  46Requirement: MANDATORY
  48The device tree blob (dtb) must be no bigger than 2 megabytes in size
  49and placed at a 2-megabyte boundary within the first 512 megabytes from
  50the start of the kernel image. This is to allow the kernel to map the
  51blob using a single section mapping in the initial page tables.
  543. Decompress the kernel image
  57Requirement: OPTIONAL
  59The AArch64 kernel does not currently provide a decompressor and
  60therefore requires decompression (gzip etc.) to be performed by the boot
  61loader if a compressed Image target (e.g. Image.gz) is used.  For
  62bootloaders that do not implement this requirement, the uncompressed
  63Image target is available instead.
  664. Call the kernel image
  69Requirement: MANDATORY
  71The decompressed kernel image contains a 32-byte header as follows:
  73  u32 magic     = 0x14000008;   /* branch to stext, little-endian */
  74  u32 res0      = 0;            /* reserved */
  75  u64 text_offset;              /* Image load offset */
  76  u64 res1      = 0;            /* reserved */
  77  u64 res2      = 0;            /* reserved */
  79The image must be placed at the specified offset (currently 0x80000)
  80from the start of the system RAM and called there. The start of the
  81system RAM must be aligned to 2MB.
  83Before jumping into the kernel, the following conditions must be met:
  85- Quiesce all DMA capable devices so that memory does not get
  86  corrupted by bogus network packets or disk data.  This will save
  87  you many hours of debug.
  89- Primary CPU general-purpose register settings
  90  x0 = physical address of device tree blob (dtb) in system RAM.
  91  x1 = 0 (reserved for future use)
  92  x2 = 0 (reserved for future use)
  93  x3 = 0 (reserved for future use)
  95- CPU mode
  96  All forms of interrupts must be masked in PSTATE.DAIF (Debug, SError,
  97  IRQ and FIQ).
  98  The CPU must be in either EL2 (RECOMMENDED in order to have access to
  99  the virtualisation extensions) or non-secure EL1.
 101- Caches, MMUs
 102  The MMU must be off.
 103  Instruction cache may be on or off.
 104  Data cache must be off and invalidated.
 105  External caches (if present) must be configured and disabled.
 107- Architected timers
 108  CNTFRQ must be programmed with the timer frequency.
 109  If entering the kernel at EL1, CNTHCTL_EL2 must have EL1PCTEN (bit 0)
 110  set where available.
 112- Coherency
 113  All CPUs to be booted by the kernel must be part of the same coherency
 114  domain on entry to the kernel.  This may require IMPLEMENTATION DEFINED
 115  initialisation to enable the receiving of maintenance operations on
 116  each CPU.
 118- System registers
 119  All writable architected system registers at the exception level where
 120  the kernel image will be entered must be initialised by software at a
 121  higher exception level to prevent execution in an UNKNOWN state.
 123The boot loader is expected to enter the kernel on each CPU in the
 124following manner:
 126- The primary CPU must jump directly to the first instruction of the
 127  kernel image.  The device tree blob passed by this CPU must contain
 128  for each CPU node:
 130    1. An 'enable-method' property. Currently, the only supported value
 131       for this field is the string "spin-table".
 133    2. A 'cpu-release-addr' property identifying a 64-bit,
 134       zero-initialised memory location.
 136  It is expected that the bootloader will generate these device tree
 137  properties and insert them into the blob prior to kernel entry.
 139- Any secondary CPUs must spin outside of the kernel in a reserved area
 140  of memory (communicated to the kernel by a /memreserve/ region in the
 141  device tree) polling their cpu-release-addr location, which must be
 142  contained in the reserved region.  A wfe instruction may be inserted
 143  to reduce the overhead of the busy-loop and a sev will be issued by
 144  the primary CPU.  When a read of the location pointed to by the
 145  cpu-release-addr returns a non-zero value, the CPU must jump directly
 146  to this value.
 148- Secondary CPU general-purpose register settings
 149  x0 = 0 (reserved for future use)
 150  x1 = 0 (reserved for future use)
 151  x2 = 0 (reserved for future use)
 152  x3 = 0 (reserved for future use)
 153 kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.