1 ==============================
2 Memory Layout on AArch64 Linux
3 ==============================
5 Author: Catalin Marinas <catalin.marinas@arm.com>
7 This document describes the virtual memory layout used by the AArch64
8 Linux kernel. The architecture allows up to 4 levels of translation
9 tables with a 4KB page size and up to 3 levels with a 64KB page size.
11 AArch64 Linux uses either 3 levels or 4 levels of translation tables
12 with the 4KB page configuration, allowing 39-bit (512GB) or 48-bit
13 (256TB) virtual addresses, respectively, for both user and kernel. With
14 64KB pages, only 2 levels of translation tables, allowing 42-bit (4TB)
15 virtual address, are used but the memory layout is the same.
17 ARMv8.2 adds optional support for Large Virtual Address space. This is
18 only available when running with a 64KB page size and expands the
19 number of descriptors in the first level of translation.
21 TTBRx selection is given by bit 55 of the virtual address. The
22 swapper_pg_dir contains only kernel (global) mappings while the user pgd
23 contains only user (non-global) mappings. The swapper_pg_dir address is
24 written to TTBR1 and never written to TTBR0.
27 AArch64 Linux memory layout with 4KB pages + 4 levels (48-bit)::
30 -----------------------------------------------------------------------
31 0000000000000000 0000ffffffffffff 256TB user
32 ffff000000000000 ffff7fffffffffff 128TB kernel logical memory map
33 [ffff600000000000 ffff7fffffffffff] 32TB [kasan shadow region]
34 ffff800000000000 ffff80007fffffff 2GB modules
35 ffff800080000000 fffffbffefffffff 124TB vmalloc
36 fffffbfff0000000 fffffbfffdffffff 224MB fixed mappings (top down)
37 fffffbfffe000000 fffffbfffe7fffff 8MB [guard region]
38 fffffbfffe800000 fffffbffff7fffff 16MB PCI I/O space
39 fffffbffff800000 fffffbffffffffff 8MB [guard region]
40 fffffc0000000000 fffffdffffffffff 2TB vmemmap
41 fffffe0000000000 ffffffffffffffff 2TB [guard region]
44 AArch64 Linux memory layout with 64KB pages + 3 levels (52-bit with HW support)::
47 -----------------------------------------------------------------------
48 0000000000000000 000fffffffffffff 4PB user
49 fff0000000000000 ffff7fffffffffff ~4PB kernel logical memory map
50 [fffd800000000000 ffff7fffffffffff] 512TB [kasan shadow region]
51 ffff800000000000 ffff80007fffffff 2GB modules
52 ffff800080000000 fffffbffefffffff 124TB vmalloc
53 fffffbfff0000000 fffffbfffdffffff 224MB fixed mappings (top down)
54 fffffbfffe000000 fffffbfffe7fffff 8MB [guard region]
55 fffffbfffe800000 fffffbffff7fffff 16MB PCI I/O space
56 fffffbffff800000 fffffbffffffffff 8MB [guard region]
57 fffffc0000000000 ffffffdfffffffff ~4TB vmemmap
58 ffffffe000000000 ffffffffffffffff 128GB [guard region]
61 Translation table lookup with 4KB pages::
63 +--------+--------+--------+--------+--------+--------+--------+--------+
64 |63 56|55 48|47 40|39 32|31 24|23 16|15 8|7 0|
65 +--------+--------+--------+--------+--------+--------+--------+--------+
68 | | | | | [11:0] in-page offset
69 | | | | +-> [20:12] L3 index
70 | | | +-----------> [29:21] L2 index
71 | | +---------------------> [38:30] L1 index
72 | +-------------------------------> [47:39] L0 index
73 +----------------------------------------> [55] TTBR0/1
76 Translation table lookup with 64KB pages::
78 +--------+--------+--------+--------+--------+--------+--------+--------+
79 |63 56|55 48|47 40|39 32|31 24|23 16|15 8|7 0|
80 +--------+--------+--------+--------+--------+--------+--------+--------+
83 | | | | [15:0] in-page offset
84 | | | +----------> [28:16] L3 index
85 | | +--------------------------> [41:29] L2 index
86 | +-------------------------------> [47:42] L1 index (48-bit)
87 | [51:42] L1 index (52-bit)
88 +----------------------------------------> [55] TTBR0/1
91 When using KVM without the Virtualization Host Extensions, the
92 hypervisor maps kernel pages in EL2 at a fixed (and potentially
93 random) offset from the linear mapping. See the kern_hyp_va macro and
94 kvm_update_va_mask function for more details. MMIO devices such as
95 GICv2 gets mapped next to the HYP idmap page, as do vectors when
96 ARM64_SPECTRE_V3A is enabled for particular CPUs.
98 When using KVM with the Virtualization Host Extensions, no additional
99 mappings are created, since the host kernel runs directly in EL2.
101 52-bit VA support in the kernel
102 -------------------------------
103 If the ARMv8.2-LVA optional feature is present, and we are running
104 with a 64KB page size; then it is possible to use 52-bits of address
105 space for both userspace and kernel addresses. However, any kernel
106 binary that supports 52-bit must also be able to fall back to 48-bit
107 at early boot time if the hardware feature is not present.
109 This fallback mechanism necessitates the kernel .text to be in the
110 higher addresses such that they are invariant to 48/52-bit VAs. Due
111 to the kasan shadow being a fraction of the entire kernel VA space,
112 the end of the kasan shadow must also be in the higher half of the
113 kernel VA space for both 48/52-bit. (Switching from 48-bit to 52-bit,
114 the end of the kasan shadow is invariant and dependent on ~0UL,
115 whilst the start address will "grow" towards the lower addresses).
117 In order to optimise phys_to_virt and virt_to_phys, the PAGE_OFFSET
118 is kept constant at 0xFFF0000000000000 (corresponding to 52-bit),
119 this obviates the need for an extra variable read. The physvirt
120 offset and vmemmap offsets are computed at early boot to enable
123 As a single binary will need to support both 48-bit and 52-bit VA
124 spaces, the VMEMMAP must be sized large enough for 52-bit VAs and
125 also must be sized large enough to accommodate a fixed PAGE_OFFSET.
127 Most code in the kernel should not need to consider the VA_BITS, for
128 code that does need to know the VA size the variables are
131 VA_BITS constant the *maximum* VA space size
133 VA_BITS_MIN constant the *minimum* VA space size
135 vabits_actual variable the *actual* VA space size
138 Maximum and minimum sizes can be useful to ensure that buffers are
139 sized large enough or that addresses are positioned close enough for
144 To maintain compatibility with software that relies on the ARMv8.0
145 VA space maximum size of 48-bits, the kernel will, by default,
146 return virtual addresses to userspace from a 48-bit range.
148 Software can "opt-in" to receiving VAs from a 52-bit space by
149 specifying an mmap hint parameter that is larger than 48-bit.
155 maybe_high_address = mmap(~0UL, size, prot, flags,...);
157 It is also possible to build a debug kernel that returns addresses
158 from a 52-bit space by enabling the following kernel config options:
162 CONFIG_EXPERT=y && CONFIG_ARM64_FORCE_52BIT=y
164 Note that this option is only intended for debugging applications
165 and should not be used in production.