优化数据对象decode + profile validate；benchmark对比routinator geomean 0.8

benchmark/routinator_object_bench/Cargo.toml

@@ -0,0 +1,8 @@
+[package]
+name = "routinator-object-bench"
+version = "0.1.0"
+edition = "2024"
+publish = false
+
+[dependencies]
+rpki = { version = "=0.19.1", features = ["repository"] }

benchmark/routinator_object_bench/src/main.rs

@@ -0,0 +1,552 @@
+use rpki::repository::cert::Cert;
+use rpki::repository::crl::Crl;
+use rpki::repository::manifest::Manifest;
+use rpki::repository::roa::Roa;
+use rpki::repository::aspa::Aspa;
+use rpki::repository::resources::{AsResources, IpResources};
+
+use std::hint::black_box;
+use std::path::{Path, PathBuf};
+use std::time::Instant;
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord)]
+enum ObjType {
+    Cer,
+    Crl,
+    Manifest,
+    Roa,
+    Aspa,
+}
+
+impl ObjType {
+    fn parse(s: &str) -> Result<Self, String> {
+        match s {
+            "cer" => Ok(Self::Cer),
+            "crl" => Ok(Self::Crl),
+            "manifest" => Ok(Self::Manifest),
+            "roa" => Ok(Self::Roa),
+            "aspa" => Ok(Self::Aspa),
+            _ => Err("type must be one of: cer, crl, manifest, roa, aspa".into()),
+        }
+    }
+
+    fn as_str(self) -> &'static str {
+        match self {
+            ObjType::Cer => "cer",
+            ObjType::Crl => "crl",
+            ObjType::Manifest => "manifest",
+            ObjType::Roa => "roa",
+            ObjType::Aspa => "aspa",
+        }
+    }
+
+    fn ext(self) -> &'static str {
+        match self {
+            ObjType::Cer => "cer",
+            ObjType::Crl => "crl",
+            ObjType::Manifest => "mft",
+            ObjType::Roa => "roa",
+            ObjType::Aspa => "asa",
+        }
+    }
+}
+
+#[derive(Clone, Debug)]
+struct Sample {
+    obj_type: ObjType,
+    name: String,
+    path: PathBuf,
+}
+
+#[derive(Clone, Debug)]
+struct Config {
+    dir: PathBuf,
+    type_filter: Option<ObjType>,
+    sample_filter: Option<String>,
+    fixed_iters: Option<u64>,
+    warmup_iters: u64,
+    rounds: u64,
+    min_round_ms: u64,
+    max_adaptive_iters: u64,
+    strict: bool,
+    cert_inspect: bool,
+    out_csv: Option<PathBuf>,
+    out_md: Option<PathBuf>,
+}
+
+fn usage_and_exit(err: Option<&str>) -> ! {
+    if let Some(err) = err {
+        eprintln!("error: {err}");
+        eprintln!();
+    }
+    eprintln!(
+        "Usage:\n\
+  cargo run --release --manifest-path rpki/benchmark/routinator_object_bench/Cargo.toml -- [OPTIONS]\n\
+\n\
+Options:\n\
+  --dir <PATH>            Fixtures root dir (default: ../../tests/benchmark/selected_der_v2)\n\
+  --type <cer|crl|manifest|roa|aspa>    Filter by type\n\
+  --sample <NAME>         Filter by sample name (e.g. p50)\n\
+  --iters <N>             Fixed iterations per round (optional; otherwise adaptive)\n\
+  --warmup-iters <N>      Warmup iterations (default: 50)\n\
+  --rounds <N>            Rounds (default: 5)\n\
+  --min-round-ms <MS>     Adaptive: minimum round time (default: 200)\n\
+  --max-iters <N>         Adaptive: maximum iters (default: 1_000_000)\n\
+  --strict <true|false>   Strict DER where applicable (default: true)\n\
+  --cert-inspect          Also run Cert::inspect_ca/inspect_ee where applicable (default: false)\n\
+  --out-csv <PATH>        Write CSV output\n\
+  --out-md <PATH>         Write Markdown output\n\
+"
+    );
+    std::process::exit(2);
+}
+
+fn parse_bool(s: &str, name: &str) -> bool {
+    match s {
+        "1" | "true" | "TRUE" | "yes" | "YES" => true,
+        "0" | "false" | "FALSE" | "no" | "NO" => false,
+        _ => usage_and_exit(Some(&format!("{name} must be true/false"))),
+    }
+}
+
+fn parse_u64(s: &str, name: &str) -> u64 {
+    s.parse::<u64>()
+        .unwrap_or_else(|_| usage_and_exit(Some(&format!("{name} must be an integer"))))
+}
+
+fn default_samples_dir() -> PathBuf {
+    PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("../../tests/benchmark/selected_der_v2")
+}
+
+fn parse_args() -> Config {
+    let mut dir: PathBuf = default_samples_dir();
+    let mut type_filter: Option<ObjType> = None;
+    let mut sample_filter: Option<String> = None;
+    let mut fixed_iters: Option<u64> = None;
+    let mut warmup_iters: u64 = 50;
+    let mut rounds: u64 = 5;
+    let mut min_round_ms: u64 = 200;
+    let mut max_adaptive_iters: u64 = 1_000_000;
+    let mut strict: bool = true;
+    let mut cert_inspect: bool = false;
+    let mut out_csv: Option<PathBuf> = None;
+    let mut out_md: Option<PathBuf> = None;
+
+    let mut args = std::env::args().skip(1);
+    while let Some(arg) = args.next() {
+        match arg.as_str() {
+            "--dir" => dir = PathBuf::from(args.next().unwrap_or_else(|| usage_and_exit(None))),
+            "--type" => {
+                type_filter = Some(ObjType::parse(
+                    &args.next().unwrap_or_else(|| usage_and_exit(None)),
+                )
+                .unwrap_or_else(|e| usage_and_exit(Some(&e))))
+            }
+            "--sample" => {
+                sample_filter = Some(args.next().unwrap_or_else(|| usage_and_exit(None)))
+            }
+            "--iters" => {
+                fixed_iters = Some(parse_u64(
+                    &args.next().unwrap_or_else(|| usage_and_exit(None)),
+                    "--iters",
+                ))
+            }
+            "--warmup-iters" => {
+                warmup_iters = parse_u64(
+                    &args.next().unwrap_or_else(|| usage_and_exit(None)),
+                    "--warmup-iters",
+                )
+            }
+            "--rounds" => {
+                rounds = parse_u64(&args.next().unwrap_or_else(|| usage_and_exit(None)), "--rounds")
+            }
+            "--min-round-ms" => {
+                min_round_ms = parse_u64(
+                    &args.next().unwrap_or_else(|| usage_and_exit(None)),
+                    "--min-round-ms",
+                )
+            }
+            "--max-iters" => {
+                max_adaptive_iters = parse_u64(
+                    &args.next().unwrap_or_else(|| usage_and_exit(None)),
+                    "--max-iters",
+                )
+            }
+            "--strict" => {
+                strict = parse_bool(
+                    &args.next().unwrap_or_else(|| usage_and_exit(None)),
+                    "--strict",
+                )
+            }
+            "--cert-inspect" => cert_inspect = true,
+            "--out-csv" => out_csv = Some(PathBuf::from(args.next().unwrap_or_else(|| usage_and_exit(None)))),
+            "--out-md" => out_md = Some(PathBuf::from(args.next().unwrap_or_else(|| usage_and_exit(None)))),
+            "-h" | "--help" => usage_and_exit(None),
+            _ => usage_and_exit(Some(&format!("unknown argument: {arg}"))),
+        }
+    }
+
+    if warmup_iters == 0 {
+        usage_and_exit(Some("--warmup-iters must be > 0"));
+    }
+    if rounds == 0 {
+        usage_and_exit(Some("--rounds must be > 0"));
+    }
+    if min_round_ms == 0 {
+        usage_and_exit(Some("--min-round-ms must be > 0"));
+    }
+    if max_adaptive_iters == 0 {
+        usage_and_exit(Some("--max-iters must be > 0"));
+    }
+    if let Some(n) = fixed_iters {
+        if n == 0 {
+            usage_and_exit(Some("--iters must be > 0"));
+        }
+    }
+
+    Config {
+        dir,
+        type_filter,
+        sample_filter,
+        fixed_iters,
+        warmup_iters,
+        rounds,
+        min_round_ms,
+        max_adaptive_iters,
+        strict,
+        cert_inspect,
+        out_csv,
+        out_md,
+    }
+}
+
+fn read_samples(root: &Path) -> Vec<Sample> {
+    let mut out = Vec::new();
+    for obj_type in [
+        ObjType::Cer,
+        ObjType::Crl,
+        ObjType::Manifest,
+        ObjType::Roa,
+        ObjType::Aspa,
+    ] {
+        let dir = root.join(obj_type.as_str());
+        let rd = match std::fs::read_dir(&dir) {
+            Ok(rd) => rd,
+            Err(_) => continue,
+        };
+        for ent in rd.flatten() {
+            let path = ent.path();
+            if path.extension().and_then(|s| s.to_str()) != Some(obj_type.ext()) {
+                continue;
+            }
+            let name = path
+                .file_stem()
+                .and_then(|s| s.to_str())
+                .unwrap_or("unknown")
+                .to_string();
+            out.push(Sample { obj_type, name, path });
+        }
+    }
+    out.sort_by(|a, b| a.obj_type.cmp(&b.obj_type).then_with(|| a.name.cmp(&b.name)));
+    out
+}
+
+fn choose_iters_adaptive<F: FnMut()>(mut op: F, min_round_ms: u64, max_iters: u64) -> u64 {
+    let min_secs = (min_round_ms as f64) / 1e3;
+    let mut iters: u64 = 1;
+    loop {
+        let start = Instant::now();
+        for _ in 0..iters {
+            op();
+        }
+        let elapsed = start.elapsed().as_secs_f64();
+        if elapsed >= min_secs {
+            return iters;
+        }
+        if iters >= max_iters {
+            return iters;
+        }
+        iters = (iters.saturating_mul(2)).min(max_iters);
+    }
+}
+
+fn count_ip(res: &IpResources) -> u64 {
+    if res.is_inherited() {
+        return 1;
+    }
+    let Ok(blocks) = res.to_blocks() else {
+        return 0;
+    };
+    blocks.iter().count() as u64
+}
+
+fn count_as(res: &AsResources) -> u64 {
+    if res.is_inherited() {
+        return 1;
+    }
+    let Ok(blocks) = res.to_blocks() else {
+        return 0;
+    };
+    blocks.iter().count() as u64
+}
+
+fn complexity(obj_type: ObjType, bytes: &[u8], strict: bool, cert_inspect: bool) -> u64 {
+    match obj_type {
+        ObjType::Cer => {
+            let cert = Cert::decode(bytes).expect("decode cert");
+            if cert_inspect {
+                if cert.is_ca() {
+                    cert.inspect_ca(strict).expect("inspect ca");
+                } else {
+                    cert.inspect_ee(strict).expect("inspect ee");
+                }
+            }
+            count_ip(cert.v4_resources())
+                .saturating_add(count_ip(cert.v6_resources()))
+                .saturating_add(count_as(cert.as_resources()))
+        }
+        ObjType::Crl => {
+            let crl = Crl::decode(bytes).expect("decode crl");
+            crl.revoked_certs().iter().count() as u64
+        }
+        ObjType::Manifest => {
+            let mft = Manifest::decode(bytes, strict).expect("decode manifest");
+            if cert_inspect {
+                mft.cert().inspect_ee(strict).expect("inspect ee");
+            }
+            mft.content().len() as u64
+        }
+        ObjType::Roa => {
+            let roa = Roa::decode(bytes, strict).expect("decode roa");
+            if cert_inspect {
+                roa.cert().inspect_ee(strict).expect("inspect ee");
+            }
+            roa.content().iter().count() as u64
+        }
+        ObjType::Aspa => {
+            let asa = Aspa::decode(bytes, strict).expect("decode aspa");
+            if cert_inspect {
+                asa.cert().inspect_ee(strict).expect("inspect ee");
+            }
+            asa.content().provider_as_set().len() as u64
+        }
+    }
+}
+
+fn decode_profile(obj_type: ObjType, bytes: &[u8], strict: bool, cert_inspect: bool) {
+    match obj_type {
+        ObjType::Cer => {
+            let cert = Cert::decode(black_box(bytes)).expect("decode cert");
+            if cert_inspect {
+                if cert.is_ca() {
+                    cert.inspect_ca(strict).expect("inspect ca");
+                } else {
+                    cert.inspect_ee(strict).expect("inspect ee");
+                }
+            }
+            black_box(cert);
+        }
+        ObjType::Crl => {
+            let crl = Crl::decode(black_box(bytes)).expect("decode crl");
+            black_box(crl);
+        }
+        ObjType::Manifest => {
+            let mft = Manifest::decode(black_box(bytes), strict).expect("decode manifest");
+            if cert_inspect {
+                mft.cert().inspect_ee(strict).expect("inspect ee");
+            }
+            black_box(mft);
+        }
+        ObjType::Roa => {
+            let roa = Roa::decode(black_box(bytes), strict).expect("decode roa");
+            if cert_inspect {
+                roa.cert().inspect_ee(strict).expect("inspect ee");
+            }
+            black_box(roa);
+        }
+        ObjType::Aspa => {
+            let asa = Aspa::decode(black_box(bytes), strict).expect("decode aspa");
+            if cert_inspect {
+                asa.cert().inspect_ee(strict).expect("inspect ee");
+            }
+            black_box(asa);
+        }
+    }
+}
+
+#[derive(Clone, Debug)]
+struct ResultRow {
+    obj_type: String,
+    sample: String,
+    size_bytes: usize,
+    complexity: u64,
+    avg_ns_per_op: f64,
+    ops_per_sec: f64,
+}
+
+fn render_markdown(title: &str, rows: &[ResultRow]) -> String {
+    let mut out = String::new();
+    out.push_str(&format!("# {title}\n\n"));
+    out.push_str("| type | sample | size_bytes | complexity | avg ns/op | ops/s |\n");
+    out.push_str("|---|---|---:|---:|---:|---:|\n");
+    for r in rows {
+        out.push_str(&format!(
+            "| {} | {} | {} | {} | {:.2} | {:.2} |\n",
+            r.obj_type, r.sample, r.size_bytes, r.complexity, r.avg_ns_per_op, r.ops_per_sec
+        ));
+    }
+    out
+}
+
+fn render_csv(rows: &[ResultRow]) -> String {
+    let mut out = String::new();
+    out.push_str("type,sample,size_bytes,complexity,avg_ns_per_op,ops_per_sec\n");
+    for r in rows {
+        let sample = r.sample.replace('"', "\"\"");
+        out.push_str(&format!(
+            "{},{},{},{},{:.6},{:.6}\n",
+            r.obj_type,
+            format!("\"{}\"", sample),
+            r.size_bytes,
+            r.complexity,
+            r.avg_ns_per_op,
+            r.ops_per_sec
+        ));
+    }
+    out
+}
+
+fn create_parent_dirs(path: &Path) {
+    if let Some(parent) = path.parent() {
+        std::fs::create_dir_all(parent).unwrap_or_else(|e| {
+            panic!("create_dir_all {}: {e}", parent.display());
+        });
+    }
+}
+
+fn write_text_file(path: &Path, content: &str) {
+    create_parent_dirs(path);
+    std::fs::write(path, content).unwrap_or_else(|e| panic!("write {}: {e}", path.display()));
+}
+
+fn main() {
+    let cfg = parse_args();
+    let mut samples = read_samples(&cfg.dir);
+    if samples.is_empty() {
+        usage_and_exit(Some(&format!(
+            "no samples found under: {}",
+            cfg.dir.display()
+        )));
+    }
+
+    if let Some(t) = cfg.type_filter {
+        samples.retain(|s| s.obj_type == t);
+        if samples.is_empty() {
+            usage_and_exit(Some(&format!("no sample matched --type {}", t.as_str())));
+        }
+    }
+    if let Some(filter) = cfg.sample_filter.as_deref() {
+        samples.retain(|s| s.name == filter);
+        if samples.is_empty() {
+            usage_and_exit(Some(&format!("no sample matched --sample {filter}")));
+        }
+    }
+
+    println!("# Routinator baseline (rpki crate) decode benchmark (selected_der_v2)");
+    println!();
+    println!("- dir: {}", cfg.dir.display());
+    println!("- strict: {}", cfg.strict);
+    println!("- cert_inspect: {}", cfg.cert_inspect);
+    if let Some(t) = cfg.type_filter {
+        println!("- type: {}", t.as_str());
+    }
+    if let Some(s) = cfg.sample_filter.as_deref() {
+        println!("- sample: {}", s);
+    }
+    if let Some(n) = cfg.fixed_iters {
+        println!("- iters: {} (fixed)", n);
+    } else {
+        println!(
+            "- warmup: {} iters, rounds: {}, min_round: {}ms (adaptive iters, max {})",
+            cfg.warmup_iters, cfg.rounds, cfg.min_round_ms, cfg.max_adaptive_iters
+        );
+    }
+    if let Some(p) = cfg.out_csv.as_ref() {
+        println!("- out_csv: {}", p.display());
+    }
+    if let Some(p) = cfg.out_md.as_ref() {
+        println!("- out_md: {}", p.display());
+    }
+    println!();
+
+    println!("| type | sample | size_bytes | complexity | avg ns/op | ops/s |");
+    println!("|---|---|---:|---:|---:|---:|");
+
+    let mut rows: Vec<ResultRow> = Vec::with_capacity(samples.len());
+    for sample in &samples {
+        let bytes = std::fs::read(&sample.path)
+            .unwrap_or_else(|e| panic!("read {}: {e}", sample.path.display()));
+        let size_bytes = bytes.len();
+        let complexity = complexity(sample.obj_type, bytes.as_slice(), cfg.strict, cfg.cert_inspect);
+
+        for _ in 0..cfg.warmup_iters {
+            decode_profile(sample.obj_type, bytes.as_slice(), cfg.strict, cfg.cert_inspect);
+        }
+
+        let mut per_round_ns_per_op = Vec::with_capacity(cfg.rounds as usize);
+        for _round in 0..cfg.rounds {
+            let iters = if let Some(n) = cfg.fixed_iters {
+                n
+            } else {
+                choose_iters_adaptive(
+                    || decode_profile(sample.obj_type, bytes.as_slice(), cfg.strict, cfg.cert_inspect),
+                    cfg.min_round_ms,
+                    cfg.max_adaptive_iters,
+                )
+            };
+            let start = Instant::now();
+            for _ in 0..iters {
+                decode_profile(sample.obj_type, bytes.as_slice(), cfg.strict, cfg.cert_inspect);
+            }
+            let elapsed = start.elapsed();
+            let total_ns = elapsed.as_secs_f64() * 1e9;
+            per_round_ns_per_op.push(total_ns / (iters as f64));
+        }
+
+        let avg_ns = per_round_ns_per_op.iter().sum::<f64>() / (per_round_ns_per_op.len() as f64);
+        let ops_per_sec = 1e9_f64 / avg_ns;
+
+        println!(
+            "| {} | {} | {} | {} | {:.2} | {:.2} |",
+            sample.obj_type.as_str(),
+            sample.name,
+            size_bytes,
+            complexity,
+            avg_ns,
+            ops_per_sec
+        );
+
+        rows.push(ResultRow {
+            obj_type: sample.obj_type.as_str().to_string(),
+            sample: sample.name.clone(),
+            size_bytes,
+            complexity,
+            avg_ns_per_op: avg_ns,
+            ops_per_sec,
+        });
+    }
+
+    if let Some(path) = cfg.out_md.as_ref() {
+        let md = render_markdown(
+            "Routinator baseline (rpki crate) decode+inspect (selected_der_v2)",
+            &rows,
+        );
+        write_text_file(path, &md);
+        eprintln!("Wrote {}", path.display());
+    }
+    if let Some(path) = cfg.out_csv.as_ref() {
+        let csv = render_csv(&rows);
+        write_text_file(path, &csv);
+        eprintln!("Wrote {}", path.display());
+    }
+}