xref: /linux/Documentation/bpf/linux-notes.rst (revision f4738f56d1dc62aaba69b33702a5ab098f1b8c63)
1.. contents::
2.. sectnum::
3
4==========================
5Linux implementation notes
6==========================
7
8This document provides more details specific to the Linux kernel implementation of the eBPF instruction set.
9
10Byte swap instructions
11======================
12
13``BPF_FROM_LE`` and ``BPF_FROM_BE`` exist as aliases for ``BPF_TO_LE`` and ``BPF_TO_BE`` respectively.
14
15Jump instructions
16=================
17
18``BPF_CALL | BPF_X | BPF_JMP`` (0x8d), where the helper function
19integer would be read from a specified register, is not currently supported
20by the verifier.  Any programs with this instruction will fail to load
21until such support is added.
22
23Maps
24====
25
26Linux only supports the 'map_val(map)' operation on array maps with a single element.
27
28Linux uses an fd_array to store maps associated with a BPF program. Thus,
29map_by_idx(imm) uses the fd at that index in the array.
30
31Variables
32=========
33
34The following 64-bit immediate instruction specifies that a variable address,
35which corresponds to some integer stored in the 'imm' field, should be loaded:
36
37=========================  ======  ===  =========================================  ===========  ==============
38opcode construction        opcode  src  pseudocode                                 imm type     dst type
39=========================  ======  ===  =========================================  ===========  ==============
40BPF_IMM | BPF_DW | BPF_LD  0x18    0x3  dst = var_addr(imm)                        variable id  data pointer
41=========================  ======  ===  =========================================  ===========  ==============
42
43On Linux, this integer is a BTF ID.
44
45Legacy BPF Packet access instructions
46=====================================
47
48As mentioned in the `ISA standard documentation
49<instruction-set.html#legacy-bpf-packet-access-instructions>`_,
50Linux has special eBPF instructions for access to packet data that have been
51carried over from classic BPF to retain the performance of legacy socket
52filters running in the eBPF interpreter.
53
54The instructions come in two forms: ``BPF_ABS | <size> | BPF_LD`` and
55``BPF_IND | <size> | BPF_LD``.
56
57These instructions are used to access packet data and can only be used when
58the program context is a pointer to a networking packet.  ``BPF_ABS``
59accesses packet data at an absolute offset specified by the immediate data
60and ``BPF_IND`` access packet data at an offset that includes the value of
61a register in addition to the immediate data.
62
63These instructions have seven implicit operands:
64
65* Register R6 is an implicit input that must contain a pointer to a
66  struct sk_buff.
67* Register R0 is an implicit output which contains the data fetched from
68  the packet.
69* Registers R1-R5 are scratch registers that are clobbered by the
70  instruction.
71
72These instructions have an implicit program exit condition as well. If an
73eBPF program attempts access data beyond the packet boundary, the
74program execution will be aborted.
75
76``BPF_ABS | BPF_W | BPF_LD`` (0x20) means::
77
78  R0 = ntohl(*(u32 *) ((struct sk_buff *) R6->data + imm))
79
80where ``ntohl()`` converts a 32-bit value from network byte order to host byte order.
81
82``BPF_IND | BPF_W | BPF_LD`` (0x40) means::
83
84  R0 = ntohl(*(u32 *) ((struct sk_buff *) R6->data + src + imm))
85