New features/APIs:

• __ksym extern variables support for getting kernel symbol addresses;
• BPF XDP link support;
• bpf_program__set_autoload() to disable loading and verifying specific BPF programs;
• support for attaching map elements BPF iterator programs;
• new getters/setters for more control over BPF map definitions;
• all destructor-like APIs (e.g., perf_buffer__free() and bpf_object__close()) now accept pointers returned on error (in addition to NULL and valid pointers) and ignore them, no need to guard destructors with extra checks now;
• bpf_link__detach() for force-detaching link, while it's still alive;
• btf__parse_raw() and btf__parse() APIs for more convenient and flexible BTF parsing.

New BPF program types supported:

• cgroup/sock_release (BPF_CGROUP_INET_SOCK_RELEASE);
• cgroup/sock_create is an alias for cgroup/sock now;
• xdp_cpumap;
• sk_lookup.

BPF-side APIs:

• most BPF helpers now return long for better code generation;
• bpf_endian.h now can be included along the vmlinux.h;
• bpf_get_task_stack();
• bpf_skc_to_tcp6_sock(), bpf_skc_to_udp6_sock();
• bpf_skc_to_{tcp, tcp_timewait, tcp_request}_sock();
• PT_REGS macros fixes for MIPS architecture.

Bug fixes:

• CO-RE relocations in .text section (in sub-programs) are now performed properly;
• vmlinux BTF is not loaded unnecessarily twice;
• perf_buffer__new() can be used on old kernels down to at least 4.9 version;
• libbpf's internal hashmap fixes for 32-bit architectures;
• few BTF sanitization bugs and memory leaks fixed;
• btf_dump handling of GCC built-in types for Arm NEON fixed;
• BTF-defined map-in-map initialization fixed for 32-bit architectures;
• various BTF fixes for 32-bit architectures.

New features:

• BTF-defined map-in-map support;
• bpf_link support expanded to support new kernel BPF link objects;
• BPF_ENABLE_STATS API;
• new BPF ringbuf map support, added ring_buffer API for consuming;
• bpf_link-backed netns attachment (flow_dissector).

New BPF program types supported:

• BPF iterator programs;
• getpeername, getsockname;
• xdp_devmap.

New BPF helpers:

• bpf_ktime_get_boot_ns();
• bpf_seq_printf(), bpf_seq_write();
• offsetof()/container_of() macro added to bpf_helpers.h;
• bpf_sk_cgroup_id(), bpf_sk_ancestor_cgroup_id();
• bpf_csum_level().

Bug fixes:

• bpf_object__load() better error code propagation;
• few memory leaks and corruptions fixed;
• register naming in PT_REGS s390 macros;
• .bss pre-setting through skeleton is now supported as well.

New BPF program types supported:

• lsm;
• fmod_ret.

New APIs:

• bpf_link pinning;
• bpf_link-based cgroup attachments;
• bpf_program__set_attach_target() to customize freplace BPF programs;
• EXPECTED_FD field support in bpf_set_link_xdp_fd_opts()
• bpf_map__set_initial_value() to specify intial value of read-only maps.

New BPF helpers:

• bpf_read_branch_records();
• bpf_get_ns_current_pid_tgid();
• get_netns_cookie();
• bpf_xdp_output();
• bpf_sk_assign();
• PT_REGS macro fixes;
• CO-RE variants of PT_REGS macros.

Bug fixes:

• CO-RE relocation edge cases;
• expected_attach_type handling fixes at load time;
• fixes in hanling kernels without BTF support;
• internal map sanitization improvements.

Major new features:

• BPF skeleton support;
• Kconfig extern variables support;
• STRUCT_OPS support;
• support for BPF program extensions;
• cgroup MULTI-mode support (bpf_prog_attach_xattr() API).
• bpf_send_signal_thread() BPF helper;

Usability improvements:

• BPF CO-RE improvements (flexible array, LDX/ST/STX instructions, improved conditional relocations);
• generic bpf_program__attach() API;
• SK_REUSEPORT program type support;
• bpf_link_disconnect();
• bpf_find_kernel_btf() API exposed;
• large instruction limit probing added;
• improved error message for RLIMIT_MEMLOCK.

Fixes:

• fix perf_buffer handling of offline/missing CPUs;
• various other fixes and usability improvements.

New features/APIs:

• new extensible bpf_object__open_{file,mem} APIs and DECLARE_LIBBPF_OPTS() macro to go with them;
• bpf_helpers.h, bpf_endian.h, and bpf_tracing.h are now distributed with libbpf;
• BPF CO-RE: added field size, field existence, and bitfield relocation support;
• BPF CO-RE: BPF_CORE_READ(), bpf_core_field_exists(), bpf_core_field_size() and other BPF CO-RE related helpers available through bpf_core_read.h header;
• bpf_object__open() API now auto-detects program type from its section name;
• BPF_PROG_TRACING programs support (incuding BTF-typed raw tracepoints, fentry/fexit programs);
• mmap() support for BPF global variables;
• declarative map pinning support added;
• probe_read_{user,kernel}[_str]() BPF helpers added;
• bpf_get_link_xdp_info() function to get more XDP information added;
• a bunch of other AF_XDP changes;

Usability improvements:

• no need for int version SEC('version') = 1; anymore;
• raw_tp/tp and uprobe/uretprobe section prefixes added;
• new bpf_program__get_{type,expected_attach_type} getters;
• preserve error code on program load failure;

Fixes (most notable ones):

• btf_dump padding handling;
• bpf_object__name() returning name, not path;
• ELF section handling off-by-one bug fix;
• mem leak/double free fix in BPF program relocation code;

Changes:

• bpf_btf_get_next_id() API to iterate over system's BTF objects;
• libbpf_set_print() now returns previously set print callback;
• libbpf versioning, build, and packaging improvements;
• convenience helpers for working with BTF types;
• experimental BPF CO-RE relocation support;
• various AF_XDP fixes and enhancements;
• lots of bug fixes and small improvements all around the library from many contributors.

N.B. This is a first release of libbpf that uses symbol versioning. Please let us know if you encounter any issue because of that.

Changes:

• BTF-defined maps;
• tracing attachment APIs and bpf_link abstraction;
• perf buffer API;
• BTF-to-C conversion API;
• btf__parse_elf API for loading .BTF from ELF files;
• libbpf_num_possible_cpus() added;
• passing through prog_flags through bpf_object__open;
• new attach types detection added;
• various bug fixes, code clean up, etc;

First tagged release