projects / linux-block.git / blame
| Commit | Line | Data |
|---|---|---|
| 6166da0a DT |
1 | .. contents:: |
| 2 | .. sectnum:: | |
| 3 | ||
| 4 | ========================== | |
| 5 | Linux implementation notes | |
| 6 | ========================== | |
| 7 | ||
| 8 | This document provides more details specific to the Linux kernel implementation of the eBPF instruction set. | |
| 9 | ||
| 9a0bf213 DT |
10 | Byte 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 | ||
| c1f9e14e DT |
15 | Jump instructions |
| 16 | ================= | |
| 17 | ||
| 18 | ``BPF_CALL | BPF_X | BPF_JMP`` (0x8d), where the helper function | |
| 19 | integer would be read from a specified register, is not currently supported | |
| 20 | by the verifier. Any programs with this instruction will fail to load | |
| 21 | until such support is added. | |
| 22 | ||
| 6166da0a DT |
23 | Legacy BPF Packet access instructions |
| 24 | ===================================== | |
| 25 | ||
| 26 | As mentioned in the `ISA standard documentation <instruction-set.rst#legacy-bpf-packet-access-instructions>`_, | |
| 27 | Linux has special eBPF instructions for access to packet data that have been | |
| 28 | carried over from classic BPF to retain the performance of legacy socket | |
| 29 | filters running in the eBPF interpreter. | |
| 30 | ||
| 31 | The instructions come in two forms: ``BPF_ABS | <size> | BPF_LD`` and | |
| 32 | ``BPF_IND | <size> | BPF_LD``. | |
| 33 | ||
| 34 | These instructions are used to access packet data and can only be used when | |
| 35 | the program context is a pointer to a networking packet. ``BPF_ABS`` | |
| 36 | accesses packet data at an absolute offset specified by the immediate data | |
| 37 | and ``BPF_IND`` access packet data at an offset that includes the value of | |
| 38 | a register in addition to the immediate data. | |
| 39 | ||
| 40 | These instructions have seven implicit operands: | |
| 41 | ||
| 42 | * Register R6 is an implicit input that must contain a pointer to a | |
| 43 | struct sk_buff. | |
| 44 | * Register R0 is an implicit output which contains the data fetched from | |
| 45 | the packet. | |
| 46 | * Registers R1-R5 are scratch registers that are clobbered by the | |
| 47 | instruction. | |
| 48 | ||
| 49 | These instructions have an implicit program exit condition as well. If an | |
| 50 | eBPF program attempts access data beyond the packet boundary, the | |
| 51 | program execution will be aborted. | |
| 52 | ||
| 53 | ``BPF_ABS | BPF_W | BPF_LD`` (0x20) means:: | |
| 54 | ||
| 55 | R0 = ntohl(*(u32 *) ((struct sk_buff *) R6->data + imm)) | |
| 56 | ||
| 57 | where ``ntohl()`` converts a 32-bit value from network byte order to host byte order. | |
| 58 | ||
| 59 | ``BPF_IND | BPF_W | BPF_LD`` (0x40) means:: | |
| 60 | ||
| 61 | R0 = ntohl(*(u32 *) ((struct sk_buff *) R6->data + src + imm)) |