Ebpf

eBPF

Purpose

Guide agents through writing, loading, and debugging eBPF programs using libbpf, bpftrace, and bpftool. Covers map types, program types, verifier errors, XDP networking, and CO-RE portability.

Triggers

• "How do I write an eBPF program to trace system calls?"
• "My eBPF program fails with a verifier error"
• "How do I use bpftrace to trace kernel events?"
• "How do I share data between kernel eBPF and userspace?"
• "How do I write an XDP program for packet filtering?"
• "How do I make my eBPF program portable across kernel versions (CO-RE)?"

Workflow

1. Choose the right tool

Goal?
├── One-liner kernel tracing / scripting → bpftrace
├── Production eBPF program with userspace → libbpf (C) or aya (Rust)
├── Inspect loaded programs and maps → bpftool
└── High-performance packet processing → XDP + libbpf

2. bpftrace — quick kernel tracing

# Trace all execve calls with comm and args
bpftrace -e 'tracepoint:syscalls:sys_enter_execve { printf("%s %s\n", comm, str(args->filename)); }'

# Count syscalls by process
bpftrace -e 'tracepoint:raw_syscalls:sys_enter { @[comm] = count(); }'

# Latency histogram for read() syscall
bpftrace -e '
  tracepoint:syscalls:sys_enter_read { @start[tid] = nsecs; }
  tracepoint:syscalls:sys_exit_read  { @us = hist((nsecs - @start[tid]) / 1000); delete(@start[tid]); }'

# List available tracepoints
bpftrace -l 'tracepoint:syscalls:*'
bpftrace -l 'kprobe:tcp_*'

3. libbpf skeleton — minimal C program

// counter.bpf.c — kernel-side
#include <vmlinux.h>
#include <bpf/bpf_helpers.h>

struct {
    __uint(type, BPF_MAP_TYPE_HASH);
    __type(key, u32);
    __type(value, u64);
    __uint(max_entries, 1024);
} call_count SEC(".maps");

SEC("tracepoint/syscalls/sys_enter_read")
int trace_read(struct trace_event_raw_sys_enter *ctx)
{
    u32 pid = bpf_get_current_pid_tgid() >> 32;
    u64 *cnt = bpf_map_lookup_elem(&call_count, &pid);
    if (cnt)
        (*cnt)++;
    else {
        u64
        bpf_map_update_elem(&call_count, &pid, &one, BPF_ANY);
    }
    return 0;
}

char LICENSE[] SEC("license") = "GPL";

// counter.c — userspace loader
#include "counter.skel.h"

int main(void) {
    struct counter_bpf *skel = counter_bpf__open();
    if (!skel || counter_bpf__load(skel))
        return 1;
    counter_bpf__attach(skel);
    // read map, print results
    counter_bpf__destroy(skel);
}

# Build with libbpf 1.x
clang -g -O2 -target bpf -D__TARGET_ARCH_x86 -I/usr/include/bpf \
      -c counter.bpf.c -o counter.bpf.o
bpftool gen skeleton counter.bpf.o > counter.skel.h
gcc -o counter counter.c -lbpf -lelf -lz

libbpf 1.x API changes:

// Open and load (replaces older bpf_object__open/load split patterns)
struct counter_bpf *skel = counter_bpf__open();
counter_bpf__load(skel);
counter_bpf__attach(skel);

// Or explicit file open
struct bpf_object *obj = bpf_object__open_file("counter.bpf.o", NULL);
bpf_object__load(obj);

// Skeleton generation (always via bpftool)
// bpftool gen skeleton counter.bpf.o name counter > counter.skel.h

4. eBPF map types

Use BPF_MAP_TYPE_RINGBUF over PERF_EVENT_ARRAY for new code — lower overhead, variable-size records.

5. Verifier error triage

# Get detailed verifier log
bpftool prog load prog.bpf.o /sys/fs/bpf/prog type kprobe \
    2>&1 | head -100

# Check loaded programs
bpftool prog list
bpftool prog dump xlated id 42

6. XDP programs

// xdp_drop_icmp.bpf.c
#include <vmlinux.h>
#include <bpf/bpf_helpers.h>
#include <bpf/bpf_endian.h>

SEC("xdp")
int xdp_filter(struct xdp_md *ctx)
{
    void *data_end = (void *)(long)ctx->data_end;
    void *data     = (void *)(long)ctx->data;
    struct ethhdr *eth = data;

    if ((void *)(eth + 1) > data_end)
        return XDP_PASS;

    if (bpf_ntohs(eth->h_proto) != ETH_P_IP)
        return XDP_PASS;

    struct iphdr *ip = (void *)(eth + 1);
    if ((void *)(ip + 1) > data_end)
        return XDP_PASS;

    if (ip->protocol == IPPROTO_ICMP)
        return XDP_DROP;

    return XDP_PASS;
}
char LICENSE[] SEC("license") = "GPL";

# Attach XDP program to interface
ip link set dev eth0 xdp obj xdp_drop_icmp.bpf.o sec xdp
# Remove
ip link set dev eth0 xdp off
# Use native (driver) mode for best performance
ip link set dev eth0 xdp obj prog.bpf.o sec xdp mode native

XDP return codes: XDP_PASS, XDP_DROP, XDP_TX (hairpin), XDP_REDIRECT.

7. CO-RE — compile once, run everywhere

CO-RE (Compile Once - Run Everywhere) uses BTF type info to relocate field accesses at load time.

// Use BTF-based field access (CO-RE aware)
#include <vmlinux.h>        // generated from running kernel's BTF
#include <bpf/bpf_core_read.h>

SEC("kprobe/tcp_connect")
int trace_connect(struct pt_regs *ctx)
{
    struct sock *sk = (struct sock *)PT_REGS_PARM1(ctx);
    u16 dport = BPF_CORE_READ(sk, __sk_common.skc_dport);
    // BPF_CORE_READ relocates the field offset at load time
    bpf_printk("connect to port %d\n", bpf_ntohs(dport));
    return 0;
}

# Generate vmlinux.h from running kernel
bpftool btf dump file /sys/kernel/btf/vmlinux format c > vmlinux.h

# Verify BTF is enabled
ls /sys/kernel/btf/vmlinux

8. BPF ring buffer vs perf buffer

// Ring buffer (preferred for new programs — better perf, no per-CPU loss)
struct {
    __uint(type, BPF_MAP_TYPE_RINGBUF);
    __uint(max_entries, 256 * 1024);
} rb SEC(".maps");

SEC("kprobe/sys_open")
int handle_open(struct pt_regs *ctx)
{
    struct event *e = bpf_ringbuf_reserve(&rb, sizeof(*e), 0);
    if (!e)
        return 0;
    e->pid = bpf_get_current_pid_tgid() >> 32;
    bpf_ringbuf_submit(e, 0);
    // bpf_ringbuf_discard(e, 0) on error paths
    return 0;
}

9. BPF iterators

struct {
    __uint(type, BPF_MAP_TYPE_PROG_ARRAY);
    __uint(max_entries, 1);
} iter_prog SEC(".maps");

SEC("iter/task")
int dump_tasks(struct bpf_iter__task *ctx)
{
    struct task_struct *task = ctx->task;
    if (task)
        bpf_seq_printf(ctx->meta->seq, "%d %s\n", task->tgid, task->comm);
    return 0;
}

# Read iterator output from userspace
bpftool prog tracelog   # or attach iter to seq_file reader
cat /sys/kernel/debug/tracing/trace_pipe

Iterators walk kernel data structures (tasks, maps, TCP sockets) without kprobe overhead per element.

10. BPF atomics

// GCC/Clang atomic builtins in BPF programs (kernel 5.12+)
static __always_inline void inc_counter(__u32 *counter)
{
    __sync_fetch_and_add(counter, 1);
}

// Use for per-CPU or map-backed counters under concurrent probes

Prefer per-CPU array maps for high-frequency counters; use atomics when aggregating into a single map value.

For the full map types reference, see references/ebpf-map-types.md.

Related skills

• Use skills/observability/ebpf-rust for Aya framework Rust eBPF programs
• Use skills/profilers/linux-perf for perf-based tracing without eBPF
• Use skills/runtimes/binary-hardening for seccomp-bpf syscall filtering
• Use skills/low-level-programming/linux-kernel-modules for kernel module development
• Use skills/async-io/af-xdp for XDP_REDIRECT to AF_XDP sockets