Based on kernel version 3.16. Page generated on 2014-08-06 21:36 EST.
1 Booting AArch64 Linux 2 ===================== 3 4 Author: Will Deacon <firstname.lastname@example.org> 5 Date : 07 September 2012 6 7 This document is based on the ARM booting document by Russell King and 8 is relevant to all public releases of the AArch64 Linux kernel. 9 10 The 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 12 counterpart. EL2 is the hypervisor level and exists only in non-secure 13 mode. EL3 is the highest priority level and exists only in secure mode. 14 15 For the purposes of this document, we will use the term `boot loader' 16 simply to define all software that executes on the CPU(s) before control 17 is passed to the Linux kernel. This may include secure monitor and 18 hypervisor code, or it may just be a handful of instructions for 19 preparing a minimal boot environment. 20 21 Essentially, the boot loader should provide (as a minimum) the 22 following: 23 24 1. Setup and initialise the RAM 25 2. Setup the device tree 26 3. Decompress the kernel image 27 4. Call the kernel image 28 29 30 1. Setup and initialise RAM 31 --------------------------- 32 33 Requirement: MANDATORY 34 35 The boot loader is expected to find and initialise all RAM that the 36 kernel will use for volatile data storage in the system. It performs 37 this in a machine dependent manner. (It may use internal algorithms 38 to automatically locate and size all RAM, or it may use knowledge of 39 the RAM in the machine, or any other method the boot loader designer 40 sees fit.) 41 42 43 2. Setup the device tree 44 ------------------------- 45 46 Requirement: MANDATORY 47 48 The device tree blob (dtb) must be placed on an 8-byte boundary within 49 the first 512 megabytes from the start of the kernel image and must not 50 cross a 2-megabyte boundary. This is to allow the kernel to map the 51 blob using a single section mapping in the initial page tables. 52 53 54 3. Decompress the kernel image 55 ------------------------------ 56 57 Requirement: OPTIONAL 58 59 The AArch64 kernel does not currently provide a decompressor and 60 therefore requires decompression (gzip etc.) to be performed by the boot 61 loader if a compressed Image target (e.g. Image.gz) is used. For 62 bootloaders that do not implement this requirement, the uncompressed 63 Image target is available instead. 64 65 66 4. Call the kernel image 67 ------------------------ 68 69 Requirement: MANDATORY 70 71 The decompressed kernel image contains a 64-byte header as follows: 72 73 u32 code0; /* Executable code */ 74 u32 code1; /* Executable code */ 75 u64 text_offset; /* Image load offset */ 76 u64 res0 = 0; /* reserved */ 77 u64 res1 = 0; /* reserved */ 78 u64 res2 = 0; /* reserved */ 79 u64 res3 = 0; /* reserved */ 80 u64 res4 = 0; /* reserved */ 81 u32 magic = 0x644d5241; /* Magic number, little endian, "ARM\x64" */ 82 u32 res5 = 0; /* reserved */ 83 84 85 Header notes: 86 87 - code0/code1 are responsible for branching to stext. 88 - when booting through EFI, code0/code1 are initially skipped. 89 res5 is an offset to the PE header and the PE header has the EFI 90 entry point (efi_stub_entry). When the stub has done its work, it 91 jumps to code0 to resume the normal boot process. 92 93 The image must be placed at the specified offset (currently 0x80000) 94 from the start of the system RAM and called there. The start of the 95 system RAM must be aligned to 2MB. 96 97 Before jumping into the kernel, the following conditions must be met: 98 99 - Quiesce all DMA capable devices so that memory does not get 100 corrupted by bogus network packets or disk data. This will save 101 you many hours of debug. 102 103 - Primary CPU general-purpose register settings 104 x0 = physical address of device tree blob (dtb) in system RAM. 105 x1 = 0 (reserved for future use) 106 x2 = 0 (reserved for future use) 107 x3 = 0 (reserved for future use) 108 109 - CPU mode 110 All forms of interrupts must be masked in PSTATE.DAIF (Debug, SError, 111 IRQ and FIQ). 112 The CPU must be in either EL2 (RECOMMENDED in order to have access to 113 the virtualisation extensions) or non-secure EL1. 114 115 - Caches, MMUs 116 The MMU must be off. 117 Instruction cache may be on or off. 118 The address range corresponding to the loaded kernel image must be 119 cleaned to the PoC. In the presence of a system cache or other 120 coherent masters with caches enabled, this will typically require 121 cache maintenance by VA rather than set/way operations. 122 System caches which respect the architected cache maintenance by VA 123 operations must be configured and may be enabled. 124 System caches which do not respect architected cache maintenance by VA 125 operations (not recommended) must be configured and disabled. 126 127 - Architected timers 128 CNTFRQ must be programmed with the timer frequency and CNTVOFF must 129 be programmed with a consistent value on all CPUs. If entering the 130 kernel at EL1, CNTHCTL_EL2 must have EL1PCTEN (bit 0) set where 131 available. 132 133 - Coherency 134 All CPUs to be booted by the kernel must be part of the same coherency 135 domain on entry to the kernel. This may require IMPLEMENTATION DEFINED 136 initialisation to enable the receiving of maintenance operations on 137 each CPU. 138 139 - System registers 140 All writable architected system registers at the exception level where 141 the kernel image will be entered must be initialised by software at a 142 higher exception level to prevent execution in an UNKNOWN state. 143 144 The requirements described above for CPU mode, caches, MMUs, architected 145 timers, coherency and system registers apply to all CPUs. All CPUs must 146 enter the kernel in the same exception level. 147 148 The boot loader is expected to enter the kernel on each CPU in the 149 following manner: 150 151 - The primary CPU must jump directly to the first instruction of the 152 kernel image. The device tree blob passed by this CPU must contain 153 an 'enable-method' property for each cpu node. The supported 154 enable-methods are described below. 155 156 It is expected that the bootloader will generate these device tree 157 properties and insert them into the blob prior to kernel entry. 158 159 - CPUs with a "spin-table" enable-method must have a 'cpu-release-addr' 160 property in their cpu node. This property identifies a 161 naturally-aligned 64-bit zero-initalised memory location. 162 163 These CPUs should spin outside of the kernel in a reserved area of 164 memory (communicated to the kernel by a /memreserve/ region in the 165 device tree) polling their cpu-release-addr location, which must be 166 contained in the reserved region. A wfe instruction may be inserted 167 to reduce the overhead of the busy-loop and a sev will be issued by 168 the primary CPU. When a read of the location pointed to by the 169 cpu-release-addr returns a non-zero value, the CPU must jump to this 170 value. The value will be written as a single 64-bit little-endian 171 value, so CPUs must convert the read value to their native endianness 172 before jumping to it. 173 174 - CPUs with a "psci" enable method should remain outside of 175 the kernel (i.e. outside of the regions of memory described to the 176 kernel in the memory node, or in a reserved area of memory described 177 to the kernel by a /memreserve/ region in the device tree). The 178 kernel will issue CPU_ON calls as described in ARM document number ARM 179 DEN 0022A ("Power State Coordination Interface System Software on ARM 180 processors") to bring CPUs into the kernel. 181 182 The device tree should contain a 'psci' node, as described in 183 Documentation/devicetree/bindings/arm/psci.txt. 184 185 - Secondary CPU general-purpose register settings 186 x0 = 0 (reserved for future use) 187 x1 = 0 (reserved for future use) 188 x2 = 0 (reserved for future use) 189 x3 = 0 (reserved for future use)