Compare commits

..
Author SHA1 Message Date
fanyang a7aae67c43 fix: eliminate unsafe code in packet construction and stats counters
Replace all UnsafeCell-based counters with safe alternatives:

- New ShardedCounter: per-thread TLS accumulation (thread_local crate)
  with periodic publish to static AtomicU64. Zero atomic RMW on hot path.
- ACL RuleStatsTracker: remove unsafe Arc<RuleStats> raw pointer
  mutation, use two ShardedCounter fields.
- ZCPacket: replace set_len on uninitialized BytesMut with
  write_bytes + copy_nonoverlapping before set_len.
- StatsManager UnsafeCounter/MetricData: remove UnsafeCell and
  unsafe impl Send/Sync, wrap ShardedCounter. last_updated uses
  AtomicU64 epoch millis.
- Throughput: replace UnsafeCell with AtomicU64.
- secure_datagram: fix grace-window test timestamp.
2026-07-11 20:48:19 +08:00
fanyangandGitHub f24735a86f perf: reduce packet buffer slicing churn (#2381)
* bench: add packet bytes extraction Criterion benchmark

Adds a Criterion benchmark under easytier/benches/ covering
ZCPacket::payload_bytes and tunnel_payload_bytes at 1280/4096-byte payload
sizes, using iter_batched so ZCPacket construction stays in the setup phase
and is excluded from the timed region.

- Register the [[bench]] entry in easytier/Cargo.toml.
- Document the bench and PACKET_BYTES_* env vars in benches/README.md.

* perf: reduce packet buffer slicing churn

Replace BytesMut::split_off with Buf::advance in ZCPacket bytes
extraction paths (payload_bytes, tunnel_payload_bytes, convert_type,
drop_foreign_header) and in TunZCPacketToBytes, and simplify the
copy_from_slice in new_from_payload.

When the buffer is in its unique (VEC) representation, split_off promotes
it to the shared (ARC) representation, allocating a Shared control block
and bumping the refcount on every call, and pins the buffer in shared
mode. advance only mutates the in-place ptr/len/cap fields, avoiding that
allocation/refcount churn on the TX hot path. The byte data itself is not
copied by either path.
2026-07-01 23:19:34 +08:00
7756a15cbe feat: add interface address fallback for IPv6 prefix detection (DHCPv6 IA_NA / SLAAC) (#2334)
feat: add IPv6 prefix fallback and NDP proxy for SLAAC/IA_NA

Add interface address fallback for IPv6 prefix detection when
route-based detection fails (no delegated prefix on LAN). Covers
DHCPv6 IA_NA / SLAAC where prefix is assigned to WAN without PD.
Uses getifaddrs() and Ipv6Inet to handle host bits.

Add NDP proxy sync for on-link prefixes. ISP router uses NDP to
resolve MACs; EasyTier /128 addresses on tun0 won't answer on
physical WAN. Periodic sync (30s) reads /128 routes via netlink
and manages proxy neigh entries. Auto-enables proxy_ndp sysctl.
Returns WAN ifindex from detection; static NDP_WAN_IFINDEX for
sync task. Spawns from provider reconcile when auto-detection
enabled.

Add tests: fallback default route interface, on-link NDP proxy,
interface prefix selection.

Fixes #2333

Co-authored-by: ririyeye <200610237@qq.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: KKRainbow <443152178@qq.com>
2026-07-01 12:34:46 +08:00
fanyangandGitHub 741460e1e4 bench: add Criterion TX throughput benchmark (#2397) 2026-06-30 22:10:34 +08:00
fanyangandGitHub 13412895c5 refactor(core): remove hotpath profiling (#2394)
hotpath adds noticeable overhead on hot paths, so remove the optional
profiling integration while keeping direct quanta::Instant timing.
2026-06-30 09:50:27 +08:00
26 changed files with 3091 additions and 557 deletions
-5
View File
@@ -51,11 +51,6 @@ jobs:
pnpm: true
token: ${{ secrets.GITHUB_TOKEN }}
- name: Install LTTng UST headers
run: |
sudo apt-get update
sudo apt-get install -y liblttng-ust-dev
- uses: actions-rust-lang/setup-rust-toolchain@v1
with:
components: rustfmt,clippy
+32
View File
@@ -0,0 +1,32 @@
# rust-analyzer config
# Skip the `easytier-gui` (Tauri) crate because its build scripts require
# system libraries webkit2gtk-4.1 / javascriptcoregtk-4.1 that are not
# installed on this host. Excluding it keeps rust-analyzer healthy for the
# rest of the workspace.
# Override the command used to run build scripts / collect build data.
[cargo.buildScripts]
overrideCommand = [
"cargo",
"check",
"--quiet",
"--workspace",
"--exclude",
"easytier-gui",
"--message-format=json",
"--all-targets",
"--keep-going",
]
# Override the command used for diagnostics-on-save.
[check]
overrideCommand = [
"cargo",
"check",
"--workspace",
"--exclude",
"easytier-gui",
"--message-format=json",
"--all-targets",
"--keep-going",
]
Generated
+138 -63
View File
@@ -129,6 +129,12 @@ dependencies = [
"libc",
]
[[package]]
name = "anes"
version = "0.1.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4b46cbb362ab8752921c97e041f5e366ee6297bd428a31275b9fcf1e380f7299"
[[package]]
name = "anstream"
version = "0.6.15"
@@ -772,29 +778,6 @@ dependencies = [
"serde",
]
[[package]]
name = "bindgen"
version = "0.65.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cfdf7b466f9a4903edc73f95d6d2bcd5baf8ae620638762244d3f60143643cc5"
dependencies = [
"bitflags 1.3.2",
"cexpr",
"clang-sys",
"lazy_static",
"lazycell",
"log",
"peeking_take_while",
"prettyplease",
"proc-macro2",
"quote",
"regex",
"rustc-hash 1.1.0",
"shlex",
"syn 2.0.117",
"which 4.4.2",
]
[[package]]
name = "bindgen"
version = "0.72.1"
@@ -808,7 +791,7 @@ dependencies = [
"proc-macro2",
"quote",
"regex",
"rustc-hash 2.1.0",
"rustc-hash",
"shlex",
"syn 2.0.117",
]
@@ -1162,6 +1145,12 @@ dependencies = [
"toml 0.9.12+spec-1.1.0",
]
[[package]]
name = "cast"
version = "0.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "37b2a672a2cb129a2e41c10b1224bb368f9f37a2b16b612598138befd7b37eb5"
[[package]]
name = "cc"
version = "1.2.10"
@@ -1271,6 +1260,33 @@ dependencies = [
"windows-targets 0.52.6",
]
[[package]]
name = "ciborium"
version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "42e69ffd6f0917f5c029256a24d0161db17cea3997d185db0d35926308770f0e"
dependencies = [
"ciborium-io",
"ciborium-ll",
"serde",
]
[[package]]
name = "ciborium-io"
version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "05afea1e0a06c9be33d539b876f1ce3692f4afea2cb41f740e7743225ed1c757"
[[package]]
name = "ciborium-ll"
version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "57663b653d948a338bfb3eeba9bb2fd5fcfaecb9e199e87e1eda4d9e8b240fd9"
dependencies = [
"ciborium-io",
"half",
]
[[package]]
name = "cidr"
version = "0.3.1"
@@ -1616,6 +1632,42 @@ dependencies = [
"cfg-if",
]
[[package]]
name = "criterion"
version = "0.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f2b12d017a929603d80db1831cd3a24082f8137ce19c69e6447f54f5fc8d692f"
dependencies = [
"anes",
"cast",
"ciborium",
"clap",
"criterion-plot",
"is-terminal",
"itertools 0.10.5",
"num-traits",
"once_cell",
"oorandom",
"plotters",
"rayon",
"regex",
"serde",
"serde_derive",
"serde_json",
"tinytemplate",
"walkdir",
]
[[package]]
name = "criterion-plot"
version = "0.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6b50826342786a51a89e2da3a28f1c32b06e387201bc2d19791f622c673706b1"
dependencies = [
"cast",
"itertools 0.10.5",
]
[[package]]
name = "critical-section"
version = "1.2.0"
@@ -2289,6 +2341,7 @@ dependencies = [
"clap_complete",
"clap_complete_nushell",
"console-subscriber",
"criterion",
"crossbeam",
"ctor 0.8.0",
"dashmap",
@@ -2321,8 +2374,6 @@ dependencies = [
"indoc",
"itertools 0.14.0",
"kcp-sys",
"lttng-ust",
"lttng-ust-generate",
"machine-uid",
"maplit",
"mimalloc",
@@ -2384,6 +2435,7 @@ dependencies = [
"tempfile",
"terminal_size",
"thiserror 1.0.63",
"thread_local",
"thunk-rs",
"tikv-jemalloc-ctl",
"tikv-jemalloc-sys",
@@ -4521,6 +4573,17 @@ dependencies = [
"once_cell",
]
[[package]]
name = "is-terminal"
version = "0.4.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3640c1c38b8e4e43584d8df18be5fc6b0aa314ce6ebf51b53313d4306cca8e46"
dependencies = [
"hermit-abi",
"libc",
"windows-sys 0.61.2",
]
[[package]]
name = "is-wsl"
version = "0.4.0"
@@ -4689,7 +4752,7 @@ source = "git+https://github.com/EasyTier/kcp-sys?rev=d7427c22d764deb1860a7d37ac
dependencies = [
"anyhow",
"auto_impl",
"bindgen 0.72.1",
"bindgen",
"bitflags 2.8.0",
"bytes",
"cc",
@@ -4736,12 +4799,6 @@ dependencies = [
"spin",
]
[[package]]
name = "lazycell"
version = "1.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "830d08ce1d1d941e6b30645f1a0eb5643013d835ce3779a5fc208261dbe10f55"
[[package]]
name = "leb128fmt"
version = "0.1.0"
@@ -4805,7 +4862,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4979f22fdb869068da03c9f7528f8297c6fd2606bc3a4affe42e6a823fdb8da4"
dependencies = [
"cfg-if",
"windows-targets 0.48.5",
"windows-targets 0.52.6",
]
[[package]]
@@ -4935,20 +4992,6 @@ version = "0.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "112b39cec0b298b6c1999fee3e31427f74f676e4cb9879ed1a121b43661a4154"
[[package]]
name = "lttng-ust"
version = "0.1.0"
source = "git+https://github.com/EasyTier/lttng-ust-rs.git#fb3a7a86e9c3d8edad34cc70fd48f908b745f8d9"
[[package]]
name = "lttng-ust-generate"
version = "0.1.1"
source = "git+https://github.com/EasyTier/lttng-ust-rs.git#fb3a7a86e9c3d8edad34cc70fd48f908b745f8d9"
dependencies = [
"bindgen 0.65.1",
"cc",
]
[[package]]
name = "mac"
version = "0.1.1"
@@ -5901,6 +5944,12 @@ dependencies = [
"portable-atomic",
]
[[package]]
name = "oorandom"
version = "11.1.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d6790f58c7ff633d8771f42965289203411a5e5c68388703c06e14f24770b41e"
[[package]]
name = "opaque-debug"
version = "0.3.1"
@@ -6268,12 +6317,6 @@ dependencies = [
"hmac",
]
[[package]]
name = "peeking_take_while"
version = "0.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "19b17cddbe7ec3f8bc800887bab5e717348c95ea2ca0b1bf0837fb964dc67099"
[[package]]
name = "pem"
version = "3.0.4"
@@ -6537,6 +6580,34 @@ dependencies = [
"time",
]
[[package]]
name = "plotters"
version = "0.3.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5aeb6f403d7a4911efb1e33402027fc44f29b5bf6def3effcc22d7bb75f2b747"
dependencies = [
"num-traits",
"plotters-backend",
"plotters-svg",
"wasm-bindgen",
"web-sys",
]
[[package]]
name = "plotters-backend"
version = "0.3.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "df42e13c12958a16b3f7f4386b9ab1f3e7933914ecea48da7139435263a4172a"
[[package]]
name = "plotters-svg"
version = "0.3.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "51bae2ac328883f7acdfea3d66a7c35751187f870bc81f94563733a154d7a670"
dependencies = [
"plotters-backend",
]
[[package]]
name = "pnet"
version = "0.35.0"
@@ -7118,7 +7189,7 @@ dependencies = [
"pin-project-lite",
"quinn-proto",
"quinn-udp",
"rustc-hash 2.1.0",
"rustc-hash",
"rustls",
"socket2 0.5.10",
"thiserror 2.0.11",
@@ -7139,7 +7210,7 @@ dependencies = [
"lru-slab",
"rand 0.9.1",
"ring",
"rustc-hash 2.1.0",
"rustc-hash",
"rustls",
"rustls-pki-types",
"rustls-platform-verifier",
@@ -7826,12 +7897,6 @@ dependencies = [
"wasm-bindgen",
]
[[package]]
name = "rustc-hash"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "08d43f7aa6b08d49f382cde6a7982047c3426db949b1424bc4b7ec9ae12c6ce2"
[[package]]
name = "rustc-hash"
version = "2.1.0"
@@ -9921,6 +9986,16 @@ dependencies = [
"zerovec",
]
[[package]]
name = "tinytemplate"
version = "1.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "be4d6b5f19ff7664e8c98d03e2139cb510db9b0a60b55f8e8709b689d939b6bc"
dependencies = [
"serde",
"serde_json",
]
[[package]]
name = "tinyvec"
version = "1.8.0"
@@ -11325,7 +11400,7 @@ version = "0.1.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cf221c93e13a30d793f7645a0e7762c55d169dbb0a49671918a2319d289b10bb"
dependencies = [
"windows-sys 0.48.0",
"windows-sys 0.59.0",
]
[[package]]
+24 -3
View File
@@ -28,6 +28,14 @@ path = "src/easytier-cli.rs"
name = "easytier"
path = "src/lib.rs"
[[bench]]
name = "tx_throughput"
harness = false
[[bench]]
name = "packet_bytes_extraction"
harness = false
[dependencies]
git-version = "0.3.9"
@@ -37,7 +45,6 @@ tracing-subscriber = { version = "0.3", features = [
"local-time",
"time",
] }
lttng-ust = { git = "https://github.com/EasyTier/lttng-ust-rs.git", optional = true }
derivative = "2.2.0"
derive_more = { version = "2.1.1", features = ["full"] }
console-subscriber = { version = "0.4.1", optional = true }
@@ -217,6 +224,7 @@ smoltcp = { git = "https://github.com/smoltcp-rs/smoltcp.git", rev = "0a926767a6
"async",
] }
parking_lot = { version = "0.12.0" }
thread_local = "1.1"
wildmatch = "2.3.4"
@@ -330,7 +338,6 @@ quote = "1"
thunk-rs = { git = "https://github.com/easytier/thunk.git", default-features = false, features = [
"win7",
] }
lttng-ust-generate = { git = "https://github.com/EasyTier/lttng-ust-rs.git", optional = true }
[target.'cfg(windows)'.build-dependencies]
reqwest = { version = "0.12.12", features = ["blocking"] }
@@ -338,6 +345,7 @@ zip = "4.0.0"
[dev-dependencies]
criterion = { version = "0.5", features = ["html_reports"] }
serial_test = "3.0.0"
rstest = "0.25.0"
futures-util = "0.3.31"
@@ -345,6 +353,14 @@ maplit = "1.0.2"
tempfile = "3.22.0"
ctor = "0.8.0"
[[bench]]
name = "acl_hotpath"
harness = false
[[bench]]
name = "zc_packet"
harness = false
[target.'cfg(target_os = "linux")'.dev-dependencies]
defguard_wireguard_rs = "0.4.2"
tokio-socks = "0.5.2"
@@ -404,9 +420,14 @@ jemalloc-prof = [
"jemalloc-sys/stats",
]
tracing = ["tokio/tracing", "dep:console-subscriber"]
lttng = ["dep:lttng-ust", "dep:lttng-ust-generate"]
magic-dns = ["dep:hickory-client", "dep:hickory-server"]
faketcp = ["dep:flume"]
zstd = ["dep:zstd"]
# Deprecated: hotpath profiling has been removed. These feature aliases are
# retained as no-ops so existing build scripts using `--features hotpath*`
# continue to work without pulling in any dependencies.
hotpath = []
hotpath-cpu = ["hotpath"]
hotpath-alloc = ["hotpath"]
# For Network Extension on macOS
macos-ne = []
+162
View File
@@ -0,0 +1,162 @@
# Benchmarks
Criterion benchmarks for EasyTier hot paths.
| Bench | What it measures |
| --------------------------- | -------------------------------------------------------------------------------- |
| `tx_throughput` | End-to-end TX injection path through `peer_manager::send_msg_by_ip` |
| `packet_bytes_extraction` | `ZCPacket::payload_bytes` / `tunnel_payload_bytes` extraction (advance hot path) |
## Packet Bytes Extraction
Criterion benchmark for `ZCPacket` bytes extraction — the methods touched by the
`advance`-based slicing refactor. Measures `payload_bytes` and
`tunnel_payload_bytes` at two payload sizes (1280, 4096). Setup
(`ZCPacket::new_with_payload`) runs in the benchmark harness's preparation
phase and is excluded from the timed region, so the numbers reflect only the
extraction call.
### Quick start
```bash
cargo bench --bench packet_bytes_extraction
```
Smoke run:
```bash
PACKET_BYTES_MEASUREMENT_SECS=2 \
PACKET_BYTES_WARMUP_SECS=1 \
PACKET_BYTES_SAMPLE_SIZE=10 \
cargo bench --bench packet_bytes_extraction -- --quiet
```
### Environment variables
| Variable | Default | Notes |
| ------------------------------- | ------- | ---------------------------- |
| `PACKET_BYTES_MEASUREMENT_SECS` | `10` | Criterion `measurement_time` |
| `PACKET_BYTES_WARMUP_SECS` | `3` | Criterion `warm_up_time` |
| `PACKET_BYTES_SAMPLE_SIZE` | `10` | Criterion `sample_size` (min 10) |
---
## TX Throughput Benchmark
Criterion benchmark for EasyTier's TX injection path (`peer_manager::send_msg_by_ip`).
## What it measures
The benchmark sets up two EasyTier instances (`hot-a` / `hot-b`) and drives
packets from `hot-a` to `hot-b` via `peer_manager.send_msg_by_ip`. This is the
same entry point `easytier-core` uses for daily forwarded traffic, so the
numbers reflect the real TX hot path: NIC pipeline → route lookup →
compress/encrypt → peer connection → tunnel send.
Two variants are reported per tunnel kind:
| Bench | What it measures |
| --------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `tx_throughput/<tunnel>` | Serial baseline. One send in flight at a time. Reports per-packet CPU cost (TX injection latency). |
| `tx_throughput/<tunnel>-saturate` | Spawns `TX_THROUGHPUT_INFLIGHT` tokio tasks that independently pump `send_msg_by_ip`. Reports the aggregate throughput ceiling the peer manager + tunnel can sustain across worker threads. |
> **Out of scope (by design):** TUN read/write (`no_tun = true`), compression
> (default `None`), reverse/RX-side measurement, multi-peer fanout. Add
> separate benchmarks if you need those.
## Quick start
### ring tunnel (no root, fastest)
```bash
cargo bench --bench tx_throughput
```
Smoke run (faster iteration):
```bash
TX_THROUGHPUT_MEASUREMENT_SECS=2 \
TX_THROUGHPUT_WARMUP_SECS=1 \
TX_THROUGHPUT_SAMPLE_SIZE=10 \
cargo bench --bench tx_throughput -- --quiet
```
### tcp / udp tunnels (requires Docker + root)
The benchmark creates a Docker network and registers each container's netns
under `/var/run/netns`, which requires root. Run the whole command under
`sudo`:
```bash
sudo TX_THROUGHPUT_TUNNEL=tcp \
TX_THROUGHPUT_MEASUREMENT_SECS=5 \
TX_THROUGHPUT_WARMUP_SECS=2 \
TX_THROUGHPUT_INFLIGHT=64 \
cargo bench --bench tx_throughput -- --quiet
sudo TX_THROUGHPUT_TUNNEL=udp cargo bench --bench tx_throughput -- --quiet
```
> If `sudo` cannot find `cargo`, use `sudo -E` or the absolute path
> (`$(which cargo)`).
## Environment variables
| Variable | Default | Notes |
| -------------------------------- | --------------------- | -------------------------------------- |
| `TX_THROUGHPUT_TUNNEL` | `ring` | `ring` / `tcp` / `udp` |
| `TX_THROUGHPUT_PKT_SIZE` | `1400` | IP total length in bytes |
| `TX_THROUGHPUT_WORKER_THREADS` | `4` | tokio worker threads |
| `TX_THROUGHPUT_INFLIGHT` | `64` | saturate-mode concurrency (task count) |
| `TX_THROUGHPUT_TUNNEL_PORT` | `35521` | tcp/udp listen port |
| `TX_THROUGHPUT_MEASUREMENT_SECS` | `10` | Criterion `measurement_time` |
| `TX_THROUGHPUT_WARMUP_SECS` | `3` | Criterion `warm_up_time` |
| `TX_THROUGHPUT_SAMPLE_SIZE` | `10` | Criterion `sample_size` (min 10) |
| `TX_THROUGHPUT_DOCKER_IMAGE` | `busybox:latest` | tcp/udp only |
| `TX_THROUGHPUT_DOCKER_NET` | `easytier-bench-<id>` | auto-generated unique name |
| `TX_THROUGHPUT_DOCKER_SUBNET` | `172.31.250.0/24` | |
| `TX_THROUGHPUT_DOCKER_IP_A` | `172.31.250.2` | |
| `TX_THROUGHPUT_DOCKER_IP_B` | `172.31.250.3` | |
## Parameter sweeps
```bash
# Packet size
for sz in 64 256 1400 9000; do
TX_THROUGHPUT_PKT_SIZE=$sz cargo bench --bench tx_throughput -- --quick
done
# Inflight depth (self-check: depth=1 should match serial baseline)
for d in 1 4 16 64 256; do
TX_THROUGHPUT_INFLIGHT=$d cargo bench --bench tx_throughput -- --quick
done
# Worker threads
for w in 1 2 4 8; do
TX_THROUGHPUT_WORKER_THREADS=$w cargo bench --bench tx_throughput -- --quick
done
```
## Interpreting results
- **`<tunnel>`** reports per-packet latency. Lower is better. Throughput
column here is "what one in-flight sender sustains".
- **`<tunnel>-saturate`** reports aggregate throughput across
`TX_THROUGHPUT_INFLIGHT` concurrent senders. If this matches the serial
baseline, the TX path is bottlenecked on an internal serialization point
(lock, single-threaded queue, etc.) rather than CPU or link bandwidth.
### Known finding (ring, single peer)
On the ring tunnel with a single destination peer, saturate does **not** beat
serial (observed ~277 MiB/s saturate vs ~288 MiB/s serial on a 4-worker
runtime). This points to a serialization point inside the peer-connection TX
path. Tunnels with real I/O await points (tcp/udp via Docker) are expected to
show a saturate > serial gap; verify with the sudo commands above.
## Output artifacts
Criterion writes HTML reports + SVG plots under
`easytier/target/criterion/`. Open `tx_throughput/<tunnel>/report/index.html`
or `.../<tunnel>-saturate/report/index.html` in a browser to inspect
distributions and regressions across runs.
+137
View File
@@ -0,0 +1,137 @@
use std::sync::Arc;
use criterion::{BenchmarkId, Criterion, criterion_group, criterion_main};
use easytier::common::acl_processor::{AclProcessor, PacketInfo};
use easytier::proto::acl::*;
use std::net::{IpAddr, Ipv4Addr};
fn make_acl_config() -> Acl {
let mut acl_config = Acl::default();
let mut acl_v1 = AclV1::default();
let mut chain = Chain {
name: "bench_inbound".to_string(),
chain_type: ChainType::Inbound as i32,
enabled: true,
..Default::default()
};
chain.rules.push(Rule {
name: "allow_all".to_string(),
priority: 100,
enabled: true,
action: Action::Allow as i32,
protocol: Protocol::Any as i32,
..Default::default()
});
acl_v1.chains.push(chain);
acl_config.acl_v1 = Some(acl_v1);
acl_config
}
fn make_packet_info() -> PacketInfo {
PacketInfo {
src_ip: IpAddr::V4(Ipv4Addr::new(192, 168, 1, 100)),
dst_ip: IpAddr::V4(Ipv4Addr::new(10, 0, 0, 1)),
src_port: Some(12345),
dst_port: Some(80),
protocol: Protocol::Tcp,
packet_size: 1024,
src_groups: Arc::new(vec![]),
dst_groups: Arc::new(vec![]),
}
}
fn bench_cache_hit_single(c: &mut Criterion) {
let rt = tokio::runtime::Runtime::new().unwrap();
let processor = rt.block_on(async { AclProcessor::new(make_acl_config()) });
let packet_info = make_packet_info();
// Prime the cache
let _ = processor.process_packet(&packet_info, ChainType::Inbound);
c.bench_function("acl_cache_hit_1t", |b| {
b.iter(|| {
std::hint::black_box(processor.process_packet(&packet_info, ChainType::Inbound));
});
});
}
fn bench_cache_hit_multi(c: &mut Criterion) {
let rt = tokio::runtime::Runtime::new().unwrap();
let mut group = c.benchmark_group("acl_cache_hit_multi");
for threads in [2, 4, 8] {
let processor = Arc::new(rt.block_on(async { AclProcessor::new(make_acl_config()) }));
let packet_info = Arc::new(make_packet_info());
// Prime the cache
let _ = processor.process_packet(&packet_info, ChainType::Inbound);
group.bench_with_input(
BenchmarkId::from_parameter(threads),
&threads,
|b, &threads| {
b.iter_custom(|iters| {
use std::sync::Barrier;
use std::thread;
let barrier = Arc::new(Barrier::new(threads + 1));
let per_thread = (iters / threads as u64) as usize;
let mut handles = Vec::with_capacity(threads);
for _ in 0..threads {
let processor = Arc::clone(&processor);
let packet_info = Arc::clone(&packet_info);
let barrier = Arc::clone(&barrier);
handles.push(thread::spawn(move || {
barrier.wait();
for _ in 0..per_thread {
std::hint::black_box(
processor.process_packet(&packet_info, ChainType::Inbound),
);
}
}));
}
let start = std::time::Instant::now();
barrier.wait();
for handle in handles {
handle.join().unwrap();
}
start.elapsed()
});
},
);
}
group.finish();
}
fn bench_unique_rule_match(c: &mut Criterion) {
let rt = tokio::runtime::Runtime::new().unwrap();
let processor = rt.block_on(async { AclProcessor::new(make_acl_config()) });
c.bench_function("acl_unique_rule_match_1t", |b| {
let mut i = 0usize;
b.iter(|| {
let mut packet_info = make_packet_info();
packet_info.src_port = Some((1024 + (i % 60_000)) as u16);
packet_info.src_ip = IpAddr::V4(Ipv4Addr::new(
10,
((i >> 16) & 0xff) as u8,
((i >> 8) & 0xff) as u8,
(i & 0xff) as u8,
));
std::hint::black_box(processor.process_packet(&packet_info, ChainType::Inbound));
i = i.wrapping_add(1);
});
});
}
criterion_group!(
benches,
bench_cache_hit_single,
bench_cache_hit_multi,
bench_unique_rule_match
);
criterion_main!(benches);
@@ -0,0 +1,65 @@
use std::hint::black_box;
use std::time::Duration;
use criterion::{BatchSize, Criterion, Throughput, criterion_group, criterion_main};
use easytier::tunnel::packet_def::ZCPacket;
const PAYLOAD_SIZES: &[usize] = &[1280, 4096];
fn env_parse<T: std::str::FromStr>(key: &str, default: T) -> T {
std::env::var(key)
.ok()
.and_then(|v| v.parse().ok())
.unwrap_or(default)
}
fn bench_payload_bytes(c: &mut Criterion) {
let mut group = c.benchmark_group("payload_bytes");
for &size in PAYLOAD_SIZES {
let data = vec![0u8; size];
group.throughput(Throughput::Bytes(size as u64));
group.bench_with_input(format!("{size}"), &data, |b, data| {
b.iter_batched(
|| ZCPacket::new_with_payload(black_box(data)),
|p| black_box(p).payload_bytes(),
BatchSize::SmallInput,
)
});
}
group.finish();
}
fn bench_tunnel_payload_bytes(c: &mut Criterion) {
let mut group = c.benchmark_group("tunnel_payload_bytes");
for &size in PAYLOAD_SIZES {
let data = vec![0u8; size];
group.throughput(Throughput::Bytes(size as u64));
group.bench_with_input(format!("{size}"), &data, |b, data| {
b.iter_batched(
|| ZCPacket::new_with_payload(black_box(data)),
|p| black_box(p).tunnel_payload_bytes(),
BatchSize::SmallInput,
)
});
}
group.finish();
}
fn criterion_config() -> Criterion {
let measurement_secs = env_parse("PACKET_BYTES_MEASUREMENT_SECS", 10u64);
let warmup_secs = env_parse("PACKET_BYTES_WARMUP_SECS", 3u64);
let sample_size = env_parse("PACKET_BYTES_SAMPLE_SIZE", 10usize).max(10);
Criterion::default()
.measurement_time(Duration::from_secs(measurement_secs))
.warm_up_time(Duration::from_secs(warmup_secs))
.sample_size(sample_size)
}
criterion_group! {
name = benches;
config = criterion_config();
targets = bench_payload_bytes, bench_tunnel_payload_bytes
}
criterion_main!(benches);
+472
View File
@@ -0,0 +1,472 @@
use std::{
net::IpAddr,
path::PathBuf,
process::{Command, Stdio},
str::FromStr,
sync::Arc,
sync::atomic::{AtomicU64, Ordering},
time::{Duration, Instant, SystemTime, UNIX_EPOCH},
};
use bytes::BytesMut;
use criterion::{Criterion, Throughput, criterion_group, criterion_main};
use easytier::{
common::config::{ConfigLoader, TomlConfigLoader},
instance::instance::Instance,
tunnel::{
packet_def::ZCPacket, ring::RingTunnelConnector, tcp::TcpTunnelConnector,
udp::UdpTunnelConnector,
},
};
const VIRTUAL_IP_A: &str = "10.144.144.1";
const VIRTUAL_IP_B: &str = "10.144.144.2";
const DEFAULT_DOCKER_SUBNET: &str = "172.31.250.0/24";
const DEFAULT_DOCKER_IP_A: &str = "172.31.250.2";
const DEFAULT_DOCKER_IP_B: &str = "172.31.250.3";
const DEFAULT_TUNNEL_PORT: u16 = 35521;
#[derive(Clone, Copy, Debug)]
enum TunnelKind {
Ring,
Tcp,
Udp,
}
impl TunnelKind {
fn as_str(self) -> &'static str {
match self {
TunnelKind::Ring => "ring",
TunnelKind::Tcp => "tcp",
TunnelKind::Udp => "udp",
}
}
}
impl FromStr for TunnelKind {
type Err = String;
fn from_str(value: &str) -> Result<Self, Self::Err> {
match value {
"ring" => Ok(TunnelKind::Ring),
"tcp" => Ok(TunnelKind::Tcp),
"udp" => Ok(TunnelKind::Udp),
other => Err(format!(
"unsupported TX_THROUGHPUT_TUNNEL={other:?}; expected ring, tcp, or udp"
)),
}
}
}
struct BenchTopology {
_docker: Option<DockerNetns>,
inst_a: Instance,
_inst_b: Instance,
dst: IpAddr,
packet: ZCPacket,
}
struct DockerNetns {
network: String,
container_a: String,
container_b: String,
netns_a: String,
netns_b: String,
ip_a: String,
netns_a_path: PathBuf,
netns_b_path: PathBuf,
}
impl DockerNetns {
fn create() -> Self {
let id = unique_id();
let image = env_string("TX_THROUGHPUT_DOCKER_IMAGE", "busybox:latest");
let network = env_string("TX_THROUGHPUT_DOCKER_NET", &format!("easytier-bench-{id}"));
let subnet = env_string("TX_THROUGHPUT_DOCKER_SUBNET", DEFAULT_DOCKER_SUBNET);
let ip_a = env_string("TX_THROUGHPUT_DOCKER_IP_A", DEFAULT_DOCKER_IP_A);
let ip_b = env_string("TX_THROUGHPUT_DOCKER_IP_B", DEFAULT_DOCKER_IP_B);
let container_a = format!("easytier-bench-a-{id}");
let container_b = format!("easytier-bench-b-{id}");
let netns_a = format!("easytier-bench-a-{id}");
let netns_b = format!("easytier-bench-b-{id}");
docker(&[
"network", "create", "--driver", "bridge", "--subnet", &subnet, &network,
]);
let mut docker_netns = Self {
network,
container_a,
container_b,
netns_a,
netns_b,
ip_a: ip_a.clone(),
netns_a_path: PathBuf::new(),
netns_b_path: PathBuf::new(),
};
docker_netns.start_container(&docker_netns.container_a, &ip_a, &image);
docker_netns.start_container(&docker_netns.container_b, &ip_b, &image);
let pid_a = docker(&["inspect", "-f", "{{.State.Pid}}", &docker_netns.container_a]);
let pid_b = docker(&["inspect", "-f", "{{.State.Pid}}", &docker_netns.container_b]);
docker_netns.netns_a_path = register_netns(&docker_netns.netns_a, &pid_a);
docker_netns.netns_b_path = register_netns(&docker_netns.netns_b, &pid_b);
docker_netns
}
fn start_container(&self, name: &str, ip: &str, image: &str) {
docker(&[
"run",
"-d",
"--name",
name,
"--network",
&self.network,
"--ip",
ip,
image,
"sleep",
"3600",
]);
}
}
impl Drop for DockerNetns {
fn drop(&mut self) {
let _ = std::fs::remove_file(&self.netns_a_path);
let _ = std::fs::remove_file(&self.netns_b_path);
docker_ignore(&["rm", "-f", &self.container_a, &self.container_b]);
docker_ignore(&["network", "rm", &self.network]);
}
}
fn bench_tx_throughput(c: &mut Criterion) {
let tunnel = env_string("TX_THROUGHPUT_TUNNEL", "ring")
.parse::<TunnelKind>()
.unwrap_or_else(|err| panic!("{err}"));
let packet_size = env_parse("TX_THROUGHPUT_PKT_SIZE", 1400usize);
const MIN_PKT_SIZE: usize = 28; // IPv4 (20) + UDP (8) header
assert!(
packet_size >= MIN_PKT_SIZE,
"TX_THROUGHPUT_PKT_SIZE={packet_size} is smaller than the minimum {MIN_PKT_SIZE} (IPv4+UDP headers)"
);
let worker_threads = env_parse("TX_THROUGHPUT_WORKER_THREADS", 4usize);
let inflight_depth = env_parse("TX_THROUGHPUT_INFLIGHT", 64usize).max(1);
let runtime = tokio::runtime::Builder::new_multi_thread()
.worker_threads(worker_threads)
.enable_all()
.build()
.expect("create tokio runtime");
let topology = runtime.block_on(setup_topology(tunnel, packet_size));
let peer_manager = topology.inst_a.get_peer_manager();
let packet = topology.packet.clone();
let dst = topology.dst;
eprintln!(
"tx_throughput: tunnel={} inflight={} workers={} pkt_size={}",
tunnel.as_str(),
inflight_depth.max(1),
worker_threads,
packet_size
);
let mut group = c.benchmark_group("tx_throughput");
group.throughput(Throughput::Bytes(packet_size as u64));
// Serial baseline: one packet in flight at a time.
// Measures per-packet CPU cost (TX injection latency).
group.bench_function(tunnel.as_str(), |b| {
b.iter_custom(|iterations| {
let pm = peer_manager.clone();
let pkt = packet.clone();
runtime.block_on(async move {
let start = Instant::now();
for _ in 0..iterations {
pm.send_msg_by_ip(pkt.clone(), dst, false)
.await
.expect("send packet by EasyTier IP");
}
start.elapsed()
})
});
});
// Saturate: spawn TX_THROUGHPUT_INFLIGHT worker tasks, each independently
// pumping send_msg_by_ip. Work is distributed across tokio worker threads,
// exposing the peer manager + tunnel's true aggregate throughput ceiling.
// With TX_THROUGHPUT_INFLIGHT=1 it degrades to the serial baseline.
group.bench_function(format!("{}-saturate", tunnel.as_str()), |b| {
b.iter_custom(|iterations| {
let pm = peer_manager.clone();
let pkt = packet.clone();
let concurrency = inflight_depth.min(iterations as usize).max(1);
runtime.block_on(async move {
let counter = Arc::new(AtomicU64::new(iterations));
let start = Instant::now();
let mut handles = Vec::with_capacity(concurrency);
for _ in 0..concurrency {
let pm = pm.clone();
let pkt = pkt.clone();
let counter = counter.clone();
handles.push(tokio::spawn(async move {
loop {
if counter
.fetch_update(Ordering::AcqRel, Ordering::Acquire, |cur| {
if cur > 0 { Some(cur - 1) } else { None }
})
.is_err()
{
return;
}
pm.send_msg_by_ip(pkt.clone(), dst, false)
.await
.expect("send packet by EasyTier IP");
}
}));
}
for h in handles {
h.await.expect("saturate worker task panicked");
}
start.elapsed()
})
});
});
group.finish();
runtime.block_on(async move {
drop(topology);
});
}
async fn setup_topology(tunnel: TunnelKind, packet_size: usize) -> BenchTopology {
let tunnel_port = env_parse("TX_THROUGHPUT_TUNNEL_PORT", DEFAULT_TUNNEL_PORT);
let docker = match tunnel {
TunnelKind::Ring => None,
TunnelKind::Tcp | TunnelKind::Udp => Some(DockerNetns::create()),
};
let (netns_a, netns_b) = match &docker {
Some(docker) => (Some(docker.netns_a.clone()), Some(docker.netns_b.clone())),
None => (None, None),
};
let listeners_a = match tunnel {
TunnelKind::Ring => Vec::new(),
TunnelKind::Tcp | TunnelKind::Udp => vec![
format!("{}://0.0.0.0:{}", tunnel.as_str(), tunnel_port)
.parse()
.unwrap(),
],
};
let mut inst_a = Instance::new(no_tun_config("hot-a", VIRTUAL_IP_A, netns_a, listeners_a));
let mut inst_b = Instance::new(no_tun_config("hot-b", VIRTUAL_IP_B, netns_b, Vec::new()));
inst_a.run().await.expect("inst_a run");
inst_b.run().await.expect("inst_b run");
match tunnel {
TunnelKind::Ring => inst_b
.get_conn_manager()
.add_connector(RingTunnelConnector::new(
format!("ring://{}", inst_a.id()).parse().unwrap(),
)),
TunnelKind::Tcp => inst_b
.get_conn_manager()
.add_connector(TcpTunnelConnector::new(
format!(
"tcp://{}:{}",
docker.as_ref().expect("tcp benchmark needs Docker").ip_a,
tunnel_port
)
.parse()
.unwrap(),
)),
TunnelKind::Udp => inst_b
.get_conn_manager()
.add_connector(UdpTunnelConnector::new(
format!(
"udp://{}:{}",
docker.as_ref().expect("udp benchmark needs Docker").ip_a,
tunnel_port
)
.parse()
.unwrap(),
)),
}
wait_for_routes(&inst_a, &inst_b).await;
BenchTopology {
_docker: docker,
inst_a,
_inst_b: inst_b,
dst: VIRTUAL_IP_B.parse().unwrap(),
packet: make_data_packet(VIRTUAL_IP_A, VIRTUAL_IP_B, packet_size),
}
}
async fn wait_for_routes(inst_a: &Instance, inst_b: &Instance) {
tokio::time::timeout(Duration::from_secs(15), async {
loop {
let routes_a = inst_a.get_peer_manager().list_routes().await;
let routes_b = inst_b.get_peer_manager().list_routes().await;
if !routes_a.is_empty() && !routes_b.is_empty() {
return;
}
tokio::time::sleep(Duration::from_millis(500)).await;
}
})
.await
.expect("EasyTier routes did not converge within 15s");
}
fn make_data_packet(src: &str, dst: &str, total_size: usize) -> ZCPacket {
use std::net::Ipv4Addr;
let hdr_len = 28;
let payload_len = total_size.saturating_sub(hdr_len);
let ip_total_len = (hdr_len + payload_len) as u16;
let mut buf = BytesMut::with_capacity(total_size);
buf.extend_from_slice(&[
0x45,
0x00,
(ip_total_len >> 8) as u8,
(ip_total_len & 0xff) as u8,
0x00,
0x00,
0x40,
0x00,
0x40,
0x11,
0x00,
0x00,
]);
let src: Ipv4Addr = src.parse().unwrap();
buf.extend_from_slice(&src.octets());
let dst: Ipv4Addr = dst.parse().unwrap();
buf.extend_from_slice(&dst.octets());
let udp_len = (8 + payload_len) as u16;
buf.extend_from_slice(&[
0x30,
0x39,
0xd4,
0x31,
(udp_len >> 8) as u8,
(udp_len & 0xff) as u8,
0x00,
0x00,
]);
buf.resize(total_size, 0xaa);
ZCPacket::new_with_payload(&buf)
}
fn no_tun_config(
name: &str,
ipv4: &str,
netns: Option<String>,
listeners: Vec<url::Url>,
) -> TomlConfigLoader {
let config = TomlConfigLoader::default();
config.set_inst_name(name.to_owned());
config.set_netns(netns);
config.set_ipv4(Some(ipv4.parse().unwrap()));
config.set_listeners(listeners);
let mut flags = config.get_flags();
flags.no_tun = true;
config.set_flags(flags);
config
}
fn register_netns(name: &str, pid: &str) -> PathBuf {
#[cfg(target_os = "linux")]
{
let dir = PathBuf::from("/var/run/netns");
std::fs::create_dir_all(&dir).expect("create /var/run/netns");
let path = dir.join(name);
let _ = std::fs::remove_file(&path);
std::os::unix::fs::symlink(format!("/proc/{pid}/ns/net"), &path)
.expect("link Docker netns into /var/run/netns");
path
}
#[cfg(not(target_os = "linux"))]
{
let _ = (name, pid);
panic!("Docker netns benchmark requires Linux");
}
}
fn docker(args: &[&str]) -> String {
let output = Command::new("docker")
.args(args)
.output()
.unwrap_or_else(|err| panic!("failed to run docker {args:?}: {err}"));
if !output.status.success() {
panic!(
"docker {:?} failed with status {:?}: {}",
args,
output.status.code(),
String::from_utf8_lossy(&output.stderr)
);
}
String::from_utf8_lossy(&output.stdout).trim().to_owned()
}
fn docker_ignore(args: &[&str]) {
let _ = Command::new("docker")
.args(args)
.stdout(Stdio::null())
.stderr(Stdio::null())
.status();
}
fn env_string(name: &str, default: &str) -> String {
std::env::var(name).unwrap_or_else(|_| default.to_owned())
}
fn env_parse<T>(name: &str, default: T) -> T
where
T: FromStr,
T::Err: std::fmt::Display,
{
match std::env::var(name) {
Ok(value) => value
.parse()
.unwrap_or_else(|err| panic!("invalid {name}={value:?}: {err}")),
Err(_) => default,
}
}
fn criterion_config() -> Criterion {
let measurement_secs = env_parse("TX_THROUGHPUT_MEASUREMENT_SECS", 10u64);
let warmup_secs = env_parse("TX_THROUGHPUT_WARMUP_SECS", 3u64);
let sample_size = env_parse("TX_THROUGHPUT_SAMPLE_SIZE", 10usize).max(10);
Criterion::default()
.measurement_time(Duration::from_secs(measurement_secs))
.warm_up_time(Duration::from_secs(warmup_secs))
.sample_size(sample_size)
}
fn unique_id() -> String {
let nanos = SystemTime::now()
.duration_since(UNIX_EPOCH)
.expect("system clock before UNIX epoch")
.as_nanos();
format!("{}-{nanos}", std::process::id())
}
criterion_group! {
name = benches;
config = criterion_config();
targets = bench_tx_throughput
}
criterion_main!(benches);
+51
View File
@@ -0,0 +1,51 @@
use criterion::{BenchmarkId, Criterion, criterion_group, criterion_main};
use easytier::tunnel::packet_def::{ZCPacket, ZCPacketType};
fn bench_new_with_payload(c: &mut Criterion) {
let mut group = c.benchmark_group("zc_new_with_payload");
for size in [64usize, 1500] {
let payload = vec![0xabu8; size];
group.bench_with_input(BenchmarkId::from_parameter(size), &payload, |b, payload| {
b.iter(|| {
std::hint::black_box(ZCPacket::new_with_payload(std::hint::black_box(payload)));
});
});
}
group.finish();
}
fn bench_new_for_foreign_network(c: &mut Criterion) {
let payload = vec![0xabu8; 64];
let foreign_packet = ZCPacket::new_with_payload(&payload);
let network_name = "bench-network".to_string();
c.bench_function("zc_new_for_foreign_network_64b", |b| {
b.iter(|| {
std::hint::black_box(ZCPacket::new_for_foreign_network(
std::hint::black_box(&network_name),
42,
std::hint::black_box(&foreign_packet),
));
});
});
}
fn bench_convert_type(c: &mut Criterion) {
let payload = vec![0xabu8; 64];
let packet = ZCPacket::new_with_payload(&payload);
c.bench_function("zc_convert_type_tcp_64b", |b| {
b.iter(|| {
let p = std::hint::black_box(packet.clone());
std::hint::black_box(p.convert_type(ZCPacketType::TCP));
});
});
}
criterion_group!(
benches,
bench_new_with_payload,
bench_new_for_foreign_network,
bench_convert_type
);
criterion_main!(benches);
-32
View File
@@ -133,35 +133,6 @@ fn check_locale() {
}
}
#[cfg(feature = "lttng")]
fn generate_lttng_tracepoints(out: &PathBuf) -> Result<(), Box<dyn std::error::Error>> {
use lttng_ust_generate::{CIntegerType, CTFType, Generator, LogLevel, Provider};
let mut provider = Provider::new("easytier");
provider
.create_class("marker")
.add_field("name", CTFType::SequenceText)
.add_field("value", CTFType::Integer(CIntegerType::U64))
.instantiate_with_level("mark", LogLevel::Info);
provider
.create_class("duration")
.add_field("name", CTFType::SequenceText)
.add_field("duration_micros", CTFType::Integer(CIntegerType::U64))
.instantiate_with_level("duration_micros", LogLevel::Info);
Generator::default()
.generated_lib_name("easytier_lttng_tracepoints")
.register_provider(provider)
.output_file_name(out.join("lttng_tracepoints.rs"))
.generate()
.map_err(|error| {
std::io::Error::other(format!("failed to generate LTTng tracepoints: {error:?}"))
})?;
Ok(())
}
fn main() -> Result<(), Box<dyn std::error::Error>> {
cfg_aliases! {
mobile: {
@@ -203,9 +174,6 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
}
let out = PathBuf::from(env::var("OUT_DIR")?);
#[cfg(feature = "lttng")]
generate_lttng_tracepoints(&out)?;
let descriptor = out.join("descriptors.bin");
let mut config = prost_build::Config::new();
+100 -68
View File
@@ -3,7 +3,8 @@ use std::{
net::{IpAddr, SocketAddr},
str::FromStr as _,
sync::Arc,
time::{Duration, SystemTime, UNIX_EPOCH},
sync::atomic::{AtomicU64, Ordering::Relaxed},
time::Duration,
};
use quanta::Instant;
@@ -76,7 +77,41 @@ pub struct FastLookupRule {
pub stateful: bool,
pub rate_limit: u32,
pub burst_limit: u32,
pub rule_stats: Arc<RuleStats>,
pub rule_stats: Arc<RuleStatsTracker>,
}
#[derive(Debug)]
pub struct RuleStatsTracker {
rule: Option<Rule>,
packets: AtomicU64,
bytes: AtomicU64,
}
impl RuleStatsTracker {
fn new(rule: Option<Rule>) -> Self {
Self {
rule,
packets: AtomicU64::new(0),
bytes: AtomicU64::new(0),
}
}
#[inline]
#[cfg_attr(feature = "hotpath", hotpath::measure(impl_type = "RuleStatsTracker"))]
fn increment(&self, packet_size: usize) {
self.packets.fetch_add(1, Relaxed);
self.bytes.fetch_add(packet_size as u64, Relaxed);
}
fn snapshot(&self) -> RuleStats {
RuleStats {
rule: self.rule.clone(),
stat: Some(StatItem {
packet_count: self.packets.load(Relaxed),
byte_count: self.bytes.load(Relaxed),
}),
}
}
}
// Cache key combining packet info and chain type
@@ -108,17 +143,17 @@ impl AclCacheKey {
}
// Cache entry with timestamp for LRU cleanup
#[derive(Debug, Clone)]
#[derive(Debug)]
pub(crate) struct AclCacheEntry {
pub action: Action,
pub matched_rule: RuleId,
pub last_access: Instant,
pub last_access: AtomicU64,
// New fields to track rule characteristics for proper cache behavior
pub conn_track_key: Option<String>,
pub rate_limit_keys: Vec<RateLimitKey>,
pub chain_type: ChainType,
pub acl_result: Option<AclResult>,
pub rule_stats_vec: Vec<Arc<RuleStats>>,
pub rule_stats_vec: Vec<Arc<RuleStatsTracker>>,
}
// Packet info extracted for ACL processing
@@ -211,7 +246,7 @@ pub struct AclProcessor {
default_outbound_action: Action,
default_forward_action: Action,
default_rule_stats: Arc<RuleStats>,
default_rule_stats: Arc<RuleStatsTracker>,
// Connection tracking table - shared across different processor instances if needed
conn_track: Arc<DashMap<String, ConnTrackEntry>>,
@@ -224,6 +259,13 @@ pub struct AclProcessor {
cache_max_size: usize,
cache_cleanup_interval: Duration,
// Coarse monotonic timestamp updated by the cleanup task, used to avoid
// calling Instant::now() on every cache hit.
coarse_millis: Arc<AtomicU64>,
// Hot-path counters that bypass the DashMap stats table
cache_hits: AtomicU64,
// Statistics
stats: Arc<DashMap<AclStatKey, u64>>,
@@ -259,18 +301,14 @@ impl AclProcessor {
default_outbound_action,
default_forward_action,
default_rule_stats: Arc::new(RuleStats {
rule: None,
stat: Some(StatItem {
packet_count: 0,
byte_count: 0,
}),
}),
default_rule_stats: Arc::new(RuleStatsTracker::new(None)),
conn_track: conn_track.unwrap_or_else(|| Arc::new(DashMap::new())),
rate_limiters: rate_limiters.unwrap_or_else(|| Arc::new(DashMap::new())),
rule_cache: Arc::new(DashMap::new()), // Always start with fresh cache
cache_max_size: 1024, // Limit cache to 1k entries
cache_cleanup_interval: Duration::from_secs(20), // Cleanup every 5 minutes
coarse_millis: Arc::new(AtomicU64::new(0)),
cache_hits: AtomicU64::new(0),
stats: stats.unwrap_or_else(|| Arc::new(DashMap::new())),
tasks,
};
@@ -374,11 +412,14 @@ impl AclProcessor {
let rule_cache = self.rule_cache.clone();
let cache_max_size = self.cache_max_size;
let cleanup_interval = self.cache_cleanup_interval;
let coarse_millis = self.coarse_millis.clone();
self.tasks.spawn(async move {
let mut interval = tokio::time::interval(cleanup_interval);
loop {
interval.tick().await;
let now = crate::common::stats_manager::now_monotonic_millis();
coarse_millis.store(now, Relaxed);
Self::cleanup_cache(&rule_cache, cache_max_size);
rule_cache.shrink_to_fit();
@@ -401,10 +442,9 @@ impl AclProcessor {
/// Clean up cache using LRU strategy
fn cleanup_cache(cache: &DashMap<AclCacheKey, AclCacheEntry>, max_size: usize) {
// remove cache not be used in last 15 second
let expired_timepoint = Instant::now()
.checked_sub(Duration::from_secs(15))
.unwrap_or(Instant::now());
cache.retain(|_, entry| entry.last_access > expired_timepoint);
let now = crate::common::stats_manager::now_monotonic_millis();
let cutoff = now.saturating_sub(15_000);
cache.retain(|_, entry| entry.last_access.load(Relaxed) > cutoff);
let current_size = cache.len();
if current_size <= max_size {
@@ -412,9 +452,9 @@ impl AclProcessor {
}
// Remove oldest entries (LRU cleanup)
let mut entries: Vec<(AclCacheKey, Instant)> = cache
let mut entries: Vec<(AclCacheKey, u64)> = cache
.iter()
.map(|entry| (entry.key().clone(), entry.value().last_access))
.map(|entry| (entry.key().clone(), entry.value().last_access.load(Relaxed)))
.collect();
// Sort by last_access (oldest first)
@@ -433,6 +473,7 @@ impl AclProcessor {
);
}
#[cfg_attr(feature = "hotpath", hotpath::measure(impl_type = "AclProcessor"))]
pub(crate) fn process_packet_with_cache_entry(
&self,
packet_info: &PacketInfo,
@@ -459,42 +500,41 @@ impl AclProcessor {
cache_entry.acl_result.clone().unwrap()
}
#[cfg_attr(feature = "hotpath", hotpath::measure(impl_type = "AclProcessor"))]
fn inc_cache_entry_stats(&self, cache_entry: &AclCacheEntry, packet_info: &PacketInfo) {
for rule_stats in cache_entry.rule_stats_vec.iter() {
// Use unsafe code to mutate the contents behind the Arc
let stat_ptr = rule_stats.stat.as_ref().unwrap() as *const StatItem as *mut StatItem;
unsafe {
(*stat_ptr).packet_count += 1;
(*stat_ptr).byte_count += packet_info.packet_size as u64;
}
rule_stats.increment(packet_info.packet_size);
}
}
pub fn get_rules_stats(&self) -> Vec<RuleStats> {
let mut stats: Vec<RuleStats> = Vec::new();
for rule in self.inbound_rules.iter() {
stats.push((*rule.rule_stats).clone());
stats.push(rule.rule_stats.snapshot());
}
for rule in self.outbound_rules.iter() {
stats.push((*rule.rule_stats).clone());
stats.push(rule.rule_stats.snapshot());
}
for rule in self.forward_rules.iter() {
stats.push((*rule.rule_stats).clone());
stats.push(rule.rule_stats.snapshot());
}
stats
}
/// Process a packet through ACL rules - Now lock-free!
#[cfg_attr(feature = "hotpath", hotpath::measure(impl_type = "AclProcessor"))]
pub fn process_packet(&self, packet_info: &PacketInfo, chain_type: ChainType) -> AclResult {
// Check cache first for performance
let cache_key = AclCacheKey::from_packet_info(packet_info, chain_type);
// If cache hit and can skip checks, return cached result
if let Some(mut cached) = self.rule_cache.get_mut(&cache_key) {
// Update last access time for LRU
cached.last_access = Instant::now();
self.increment_stat(AclStatKey::CacheHits);
// If cache hit and can skip checks, return cached result.
// Use get() (read lock) instead of get_mut() (write lock) and update
// last_access via AtomicU64, avoiding expensive Instant::now().
if let Some(cached) = self.rule_cache.get(&cache_key) {
cached
.last_access
.store(self.coarse_millis.load(Relaxed), Relaxed);
self.cache_hits.fetch_add(1, Relaxed);
return self.process_packet_with_cache_entry(packet_info, &cached);
}
@@ -516,7 +556,7 @@ impl AclProcessor {
let mut cache_entry = AclCacheEntry {
action: Action::Allow,
matched_rule: RuleId::Default,
last_access: Instant::now(),
last_access: AtomicU64::new(self.coarse_millis.load(Relaxed)),
conn_track_key: None,
rate_limit_keys: vec![],
chain_type,
@@ -581,8 +621,9 @@ impl AclProcessor {
// Cache the result with rule info
self.increment_stat(AclStatKey::RuleMatches);
self.inc_cache_entry_stats(&cache_entry, packet_info);
self.cache_result(&cache_key, cache_entry.clone());
return cache_entry.acl_result.clone().unwrap();
let result = cache_entry.acl_result.clone().unwrap();
self.cache_result(&cache_key, cache_entry);
return result;
}
let default_action = match chain_type {
@@ -618,8 +659,9 @@ impl AclProcessor {
// Cache the default result (no rule info)
self.inc_cache_entry_stats(&cache_entry, packet_info);
self.cache_result(&cache_key, cache_entry.clone());
cache_entry.acl_result.clone().unwrap()
let result = cache_entry.acl_result.clone().unwrap();
self.cache_result(&cache_key, cache_entry);
result
}
/// Get shared state for preserving across hot reloads
@@ -743,13 +785,11 @@ impl AclProcessor {
/// Check connection state for stateful rules
fn check_connection_state(&self, conn_track_key: &str, packet_info: &PacketInfo) {
let now = current_unix_secs();
self.conn_track
.entry(conn_track_key.to_string())
.and_modify(|x| {
x.last_seen = SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap()
.as_secs();
x.last_seen = now;
x.packet_count += 1;
x.byte_count += packet_info.packet_size as u64;
x.state = ConnState::Established as i32;
@@ -763,14 +803,8 @@ impl AclProcessor {
),
protocol: packet_info.protocol as i32,
state: ConnState::New as i32,
created_at: SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap()
.as_secs(),
last_seen: SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap()
.as_secs(),
created_at: now,
last_seen: now,
packet_count: 1,
byte_count: packet_info.packet_size as u64,
});
@@ -862,13 +896,7 @@ impl AclProcessor {
stateful: rule.stateful,
rate_limit: rule.rate_limit,
burst_limit: rule.burst_limit,
rule_stats: Arc::new(RuleStats {
rule: Some(rule.clone()),
stat: Some(StatItem {
packet_count: 0,
byte_count: 0,
}),
}),
rule_stats: Arc::new(RuleStatsTracker::new(Some(rule.clone()))),
}
}
@@ -894,6 +922,10 @@ impl AclProcessor {
.collect::<HashMap<_, _>>();
// Add cache statistics using enum keys
stats.insert(
AclStatKey::CacheHits.as_str(),
self.cache_hits.load(Relaxed),
);
stats.insert(AclStatKey::CacheSize.as_str(), self.rule_cache.len() as u64);
stats.insert(
AclStatKey::CacheMaxSize.as_str(),
@@ -908,14 +940,11 @@ impl AclProcessor {
conn_track: Arc<DashMap<String, ConnTrackEntry>>,
timeout_secs: u64,
) {
let current_time = SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap()
.as_secs();
let current_time = current_unix_secs();
let keys_to_remove: Vec<String> = conn_track
.iter()
.filter_map(|entry| {
if current_time - entry.last_seen > timeout_secs {
if current_time.saturating_sub(entry.last_seen) > timeout_secs {
Some(entry.key().clone())
} else {
None
@@ -930,11 +959,7 @@ impl AclProcessor {
/// Get cache hit rate
pub fn get_cache_hit_rate(&self) -> f64 {
let cache_hits = self
.stats
.get(&AclStatKey::CacheHits)
.map(|v| *v.value())
.unwrap_or(0);
let cache_hits = self.cache_hits.load(Relaxed);
let total_requests = cache_hits
+ self
.stats
@@ -950,6 +975,13 @@ impl AclProcessor {
}
}
fn current_unix_secs() -> u64 {
std::time::SystemTime::now()
.duration_since(std::time::UNIX_EPOCH)
.unwrap()
.as_secs()
}
// 新增辅助函数
fn parse_port_start(port_strs: &[String]) -> Option<u16> {
port_strs
+50
View File
@@ -219,6 +219,7 @@ pub struct GlobalCtx {
running_listeners: Mutex<Vec<url::Url>>,
advertised_ipv6_public_addr_prefix: Mutex<Option<cidr::Ipv6Cidr>>,
tun_device_name: Mutex<Option<String>>,
flags: ArcSwap<Flags>,
@@ -336,6 +337,7 @@ impl GlobalCtx {
running_listeners: Mutex::new(Vec::new()),
advertised_ipv6_public_addr_prefix: Mutex::new(None),
tun_device_name: Mutex::new(None),
flags: ArcSwap::new(Arc::new(flags)),
@@ -370,6 +372,24 @@ impl GlobalCtx {
}
}
fn set_tun_device_name(&self, name: Option<String>) {
*self.tun_device_name.lock().unwrap() = name;
}
pub(crate) fn set_tun_device_ready(&self, name: String) {
self.set_tun_device_name(Some(name.clone()));
self.issue_event(GlobalCtxEvent::TunDeviceReady(name));
}
pub(crate) fn set_tun_device_error(&self, error: String) {
self.set_tun_device_name(None);
self.issue_event(GlobalCtxEvent::TunDeviceError(error));
}
pub fn get_tun_device_name(&self) -> Option<String> {
self.tun_device_name.lock().unwrap().clone()
}
pub fn check_network_in_whitelist(&self, network_name: &str) -> Result<(), anyhow::Error> {
if self
.get_flags()
@@ -825,6 +845,36 @@ pub mod tests {
);
}
#[tokio::test]
async fn test_tun_device_name_tracks_explicit_runtime_state() {
let config = TomlConfigLoader::default();
let global_ctx = GlobalCtx::new(config);
assert_eq!(global_ctx.get_tun_device_name(), None);
global_ctx.issue_event(GlobalCtxEvent::TunDeviceReady("ignored".to_string()));
assert_eq!(global_ctx.get_tun_device_name(), None);
let mut subscriber = global_ctx.subscribe();
global_ctx.set_tun_device_ready("easytier0".to_string());
assert_eq!(
global_ctx.get_tun_device_name(),
Some("easytier0".to_string())
);
assert_eq!(
subscriber.recv().await.unwrap(),
GlobalCtxEvent::TunDeviceReady("easytier0".to_string())
);
global_ctx.set_tun_device_error("closed".to_string());
assert_eq!(global_ctx.get_tun_device_name(), None);
assert_eq!(
subscriber.recv().await.unwrap(),
GlobalCtxEvent::TunDeviceError("closed".to_string())
);
}
#[tokio::test]
async fn trusted_key_source_lookup_is_precise() {
let config = TomlConfigLoader::default();
+17
View File
@@ -177,3 +177,20 @@ pub(crate) fn list_ipv6_route_messages()
pub(crate) fn get_interface_index(name: &str) -> Result<u32, Error> {
netlink::NetlinkIfConfiger::get_interface_index(name)
}
#[cfg(target_os = "linux")]
pub(crate) fn add_ipv6_ndp_proxy(name: &str, address: Ipv6Addr) -> Result<(), Error> {
netlink::NetlinkIfConfiger::add_ipv6_ndp_proxy(name, address)
}
#[cfg(target_os = "linux")]
pub(crate) fn remove_ipv6_ndp_proxy(name: &str, address: Ipv6Addr) -> Result<(), Error> {
netlink::NetlinkIfConfiger::remove_ipv6_ndp_proxy(name, address)
}
#[cfg(target_os = "linux")]
pub(crate) fn list_ipv6_ndp_proxy(
name: &str,
) -> Result<std::collections::BTreeSet<Ipv6Addr>, Error> {
netlink::NetlinkIfConfiger::list_ipv6_ndp_proxy(name)
}
+104
View File
@@ -1,4 +1,5 @@
use std::{
collections::BTreeSet,
ffi::CString,
fmt::Debug,
net::{IpAddr, Ipv4Addr, Ipv6Addr},
@@ -16,6 +17,10 @@ use netlink_packet_core::{
use netlink_packet_route::{
AddressFamily, RouteNetlinkMessage,
address::{AddressAttribute, AddressMessage},
neighbour::{
NeighbourAddress, NeighbourAttribute, NeighbourFlags, NeighbourHeader, NeighbourMessage,
NeighbourState,
},
route::{
RouteAddress, RouteAttribute, RouteHeader, RouteMessage, RouteProtocol, RouteScope,
RouteType,
@@ -375,6 +380,105 @@ impl NetlinkIfConfiger {
pub(crate) fn list_ipv6_route_messages() -> Result<Vec<RouteMessage>, Error> {
Self::list_route_messages(AddressFamily::Inet6)
}
fn ipv6_ndp_proxy_message(name: &str, address: Ipv6Addr) -> Result<NeighbourMessage, Error> {
let mut message = NeighbourMessage::default();
message.header = NeighbourHeader {
family: AddressFamily::Inet6,
ifindex: Self::get_interface_index(name)?,
state: NeighbourState::Permanent,
flags: NeighbourFlags::Proxy,
kind: RouteType::Unicast,
};
message
.attributes
.push(NeighbourAttribute::Destination(NeighbourAddress::Inet6(
address,
)));
Ok(message)
}
pub(crate) fn add_ipv6_ndp_proxy(name: &str, address: Ipv6Addr) -> Result<(), Error> {
send_netlink_req_and_wait_one_resp(
RouteNetlinkMessage::NewNeighbour(Self::ipv6_ndp_proxy_message(name, address)?),
false,
)
}
pub(crate) fn remove_ipv6_ndp_proxy(name: &str, address: Ipv6Addr) -> Result<(), Error> {
send_netlink_req_and_wait_one_resp(
RouteNetlinkMessage::DelNeighbour(Self::ipv6_ndp_proxy_message(name, address)?),
true,
)
}
fn list_neighbour_messages(
address_family: AddressFamily,
) -> Result<Vec<NeighbourMessage>, Error> {
let mut message = NeighbourMessage::default();
message.header.family = address_family;
message.header.flags = NeighbourFlags::Proxy;
let s = send_netlink_req(
RouteNetlinkMessage::GetNeighbour(message),
NLM_F_REQUEST | NLM_F_DUMP,
)?;
let mut ret_vec = vec![];
let mut resp = Vec::<u8>::new();
loop {
if resp.is_empty() {
let (new_resp, _) = s.recv_from_full()?;
resp = new_resp;
}
let ret = NetlinkMessage::<RouteNetlinkMessage>::deserialize(&resp)
.with_context(|| "Failed to deserialize netlink neighbour message")?;
resp = resp.split_off(ret.buffer_len());
tracing::debug!("net link response <<< {:?}", ret);
match ret.payload {
NetlinkPayload::Error(e) => {
if e.code == NonZero::new(0) {
continue;
} else {
return Err(e.to_io().into());
}
}
NetlinkPayload::InnerMessage(RouteNetlinkMessage::NewNeighbour(m)) => {
ret_vec.push(m);
}
NetlinkPayload::Done(_) => {
break;
}
p => {
tracing::error!("Unexpected netlink response: {:?}", p);
return Err(anyhow::anyhow!("Unexpected netlink response").into());
}
}
}
Ok(ret_vec)
}
pub(crate) fn list_ipv6_ndp_proxy(name: &str) -> Result<BTreeSet<Ipv6Addr>, Error> {
let ifindex = Self::get_interface_index(name)?;
Ok(Self::list_neighbour_messages(AddressFamily::Inet6)?
.into_iter()
.filter(|message| {
message.header.ifindex == ifindex
&& message.header.flags.contains(NeighbourFlags::Proxy)
})
.filter_map(|message| {
message.attributes.into_iter().find_map(|attr| match attr {
NeighbourAttribute::Destination(NeighbourAddress::Inet6(addr)) => Some(addr),
_ => None,
})
})
.collect())
}
}
#[async_trait]
-23
View File
@@ -1,23 +0,0 @@
#[cfg(feature = "lttng")]
lttng_ust::import_tracepoints!(
concat!(env!("OUT_DIR"), "/lttng_tracepoints.rs"),
tracepoints
);
#[cfg(feature = "lttng")]
pub fn mark(name: &str, value: u64) {
tracepoints::easytier::mark(name, value);
}
#[cfg(not(feature = "lttng"))]
#[inline(always)]
pub fn mark(_name: &str, _value: u64) {}
#[cfg(feature = "lttng")]
pub fn duration_micros(name: &str, duration_micros: u64) {
tracepoints::easytier::duration_micros(name, duration_micros);
}
#[cfg(not(feature = "lttng"))]
#[inline(always)]
pub fn duration_micros(_name: &str, _duration_micros: u64) {}
+1 -1
View File
@@ -18,11 +18,11 @@ pub mod global_ctx;
pub mod idn;
pub mod ifcfg;
pub mod log;
pub mod lttng;
pub mod machine_id;
pub mod netns;
pub mod network;
pub mod os_info;
pub mod sharded_counter;
pub mod stats_manager;
pub mod stun;
pub mod stun_codec_ext;
+110
View File
@@ -0,0 +1,110 @@
use std::sync::atomic::{AtomicU64, Ordering::Relaxed};
const PUBLISH_INTERVAL: u64 = 256;
/// Counter optimized for long-lived worker threads.
///
/// Pending values below `PUBLISH_INTERVAL` live in `ThreadLocal` shards. The
/// shards are retained until this counter is dropped, so this is intended for
/// tokio workers or similarly long-lived threads rather than high-churn threads.
pub struct ShardedCounter {
published: AtomicU64,
locals: thread_local::ThreadLocal<AtomicU64>,
}
impl ShardedCounter {
pub fn new() -> Self {
Self {
published: AtomicU64::new(0),
locals: thread_local::ThreadLocal::new(),
}
}
#[inline]
#[cfg_attr(feature = "hotpath", hotpath::measure(impl_type = "ShardedCounter"))]
pub fn add(&self, delta: u64) {
let local = self.locals.get_or(|| AtomicU64::new(0));
let v = local.load(Relaxed).saturating_add(delta);
local.store(v, Relaxed);
if v >= PUBLISH_INTERVAL {
let pending = local.swap(0, Relaxed);
if pending > 0 {
self.published.fetch_add(pending, Relaxed);
}
}
}
#[inline]
pub fn inc(&self) {
self.add(1);
}
pub fn get(&self) -> u64 {
self.locals
.iter()
.fold(self.published.load(Relaxed), |total, local| {
total.saturating_add(local.load(Relaxed))
})
}
pub fn reset(&self) {
self.published.store(0, Relaxed);
for local in self.locals.iter() {
local.store(0, Relaxed);
}
}
}
impl Default for ShardedCounter {
fn default() -> Self {
Self::new()
}
}
impl std::fmt::Debug for ShardedCounter {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
f.debug_struct("ShardedCounter")
.field("value", &self.get())
.finish()
}
}
#[cfg(test)]
mod tests {
use super::*;
use std::{sync::Arc, thread};
#[test]
fn sharded_counter_get_includes_other_thread_locals() {
let counter = Arc::new(ShardedCounter::new());
let thread_counter = Arc::clone(&counter);
thread::spawn(move || {
for _ in 0..10 {
thread_counter.inc();
}
})
.join()
.unwrap();
assert_eq!(counter.get(), 10);
}
#[test]
fn sharded_counter_reset_clears_other_thread_locals() {
let counter = Arc::new(ShardedCounter::new());
let thread_counter = Arc::clone(&counter);
thread::spawn(move || {
for _ in 0..10 {
thread_counter.inc();
}
})
.join()
.unwrap();
counter.reset();
assert_eq!(counter.get(), 0);
}
}
+55 -150
View File
@@ -1,13 +1,16 @@
use crate::common::sharded_counter::ShardedCounter;
use dashmap::DashMap;
use quanta::Instant;
use serde::{Deserialize, Serialize};
use std::cell::UnsafeCell;
use std::fmt;
use std::sync::Arc;
use std::sync::atomic::{AtomicU64, Ordering};
use std::sync::{Arc, LazyLock};
use std::time::Duration;
use tokio::time::interval;
use tokio_util::task::AbortOnDropHandle;
static START_INSTANT: LazyLock<Instant> = LazyLock::new(Instant::now);
/// Predefined metric names for type safety
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Serialize, Deserialize)]
pub enum MetricName {
@@ -375,10 +378,10 @@ impl Default for LabelSet {
}
}
/// UnsafeCounter provides a high-performance counter using UnsafeCell
/// High-performance counter backed by sharded thread-local accumulation.
#[derive(Debug)]
pub struct UnsafeCounter {
value: UnsafeCell<u64>,
inner: ShardedCounter,
}
impl Default for UnsafeCounter {
@@ -390,121 +393,56 @@ impl Default for UnsafeCounter {
impl UnsafeCounter {
pub fn new() -> Self {
Self {
value: UnsafeCell::new(0),
inner: ShardedCounter::new(),
}
}
pub fn new_with_value(initial: u64) -> Self {
Self {
value: UnsafeCell::new(initial),
}
pub fn add(&self, delta: u64) {
self.inner.add(delta);
}
/// Increment the counter by the given amount
/// # Safety
/// This method is unsafe because it uses UnsafeCell. The caller must ensure
/// that no other thread is accessing this counter simultaneously.
pub unsafe fn add(&self, delta: u64) {
let ptr = self.value.get();
unsafe {
*ptr = (*ptr).saturating_add(delta);
}
pub fn inc(&self) {
self.inner.inc();
}
/// Increment the counter by 1
/// # Safety
/// This method is unsafe because it uses UnsafeCell. The caller must ensure
/// that no other thread is accessing this counter simultaneously.
pub unsafe fn inc(&self) {
unsafe {
self.add(1);
}
pub fn get(&self) -> u64 {
self.inner.get()
}
/// Get the current value of the counter
/// # Safety
/// This method is unsafe because it uses UnsafeCell. The caller must ensure
/// that no other thread is modifying this counter simultaneously.
pub unsafe fn get(&self) -> u64 {
let ptr = self.value.get();
unsafe { *ptr }
}
/// Reset the counter to zero
/// # Safety
/// This method is unsafe because it uses UnsafeCell. The caller must ensure
/// that no other thread is accessing this counter simultaneously.
pub unsafe fn reset(&self) {
let ptr = self.value.get();
unsafe {
*ptr = 0;
}
}
/// Set the counter to a specific value
/// # Safety
/// This method is unsafe because it uses UnsafeCell. The caller must ensure
/// that no other thread is accessing this counter simultaneously.
pub unsafe fn set(&self, value: u64) {
let ptr = self.value.get();
unsafe {
*ptr = value;
}
pub fn reset(&self) {
self.inner.reset();
}
}
// UnsafeCounter is Send + Sync because the safety is guaranteed by the caller
unsafe impl Send for UnsafeCounter {}
unsafe impl Sync for UnsafeCounter {}
/// MetricData contains both the counter and last update timestamp
/// Uses UnsafeCell for lock-free access
#[derive(Debug)]
struct MetricData {
counter: UnsafeCounter,
last_updated: UnsafeCell<Instant>,
last_updated: AtomicU64,
}
pub(crate) fn now_monotonic_millis() -> u64 {
Instant::now().duration_since(*START_INSTANT).as_millis() as u64
}
impl MetricData {
fn new() -> Self {
Self {
counter: UnsafeCounter::new(),
last_updated: UnsafeCell::new(Instant::now()),
last_updated: AtomicU64::new(now_monotonic_millis()),
}
}
fn new_with_value(initial: u64) -> Self {
Self {
counter: UnsafeCounter::new_with_value(initial),
last_updated: UnsafeCell::new(Instant::now()),
}
fn touch(&self) {
self.last_updated
.store(now_monotonic_millis(), Ordering::Relaxed);
}
/// Update the last_updated timestamp
/// # Safety
/// This method is unsafe because it uses UnsafeCell. The caller must ensure
/// that no other thread is accessing this timestamp simultaneously.
unsafe fn touch(&self) {
let ptr = self.last_updated.get();
unsafe {
*ptr = Instant::now();
}
}
/// Get the last updated timestamp
/// # Safety
/// This method is unsafe because it uses UnsafeCell. The caller must ensure
/// that no other thread is modifying this timestamp simultaneously.
unsafe fn get_last_updated(&self) -> Instant {
let ptr = self.last_updated.get();
unsafe { *ptr }
fn get_last_updated(&self) -> u64 {
self.last_updated.load(Ordering::Relaxed)
}
}
// MetricData is Send + Sync because the safety is guaranteed by the caller
unsafe impl Send for MetricData {}
unsafe impl Sync for MetricData {}
/// MetricKey uniquely identifies a metric with its name and labels
#[derive(Debug, Clone, PartialEq, Eq, Hash)]
struct MetricKey {
@@ -545,41 +483,23 @@ impl CounterHandle {
}
}
/// Increment the counter by the given amount
pub fn add(&self, delta: u64) {
unsafe {
self.metric_data.counter.add(delta);
self.metric_data.touch();
}
self.metric_data.counter.add(delta);
self.metric_data.touch();
}
/// Increment the counter by 1
pub fn inc(&self) {
unsafe {
self.metric_data.counter.inc();
self.metric_data.touch();
}
self.metric_data.counter.inc();
self.metric_data.touch();
}
/// Get the current value of the counter
pub fn get(&self) -> u64 {
unsafe { self.metric_data.counter.get() }
self.metric_data.counter.get()
}
/// Reset the counter to zero
pub fn reset(&self) {
unsafe {
self.metric_data.counter.reset();
self.metric_data.touch();
}
}
/// Set the counter to a specific value
pub fn set(&self, value: u64) {
unsafe {
self.metric_data.counter.set(value);
self.metric_data.touch();
}
self.metric_data.counter.reset();
self.metric_data.touch();
}
}
@@ -615,9 +535,7 @@ impl StatsManager {
loop {
interval.tick().await;
let Some(cutoff_time) = Instant::now().checked_sub(Duration::from_secs(180)) else {
continue;
};
let cutoff_millis = now_monotonic_millis().saturating_sub(180_000);
let Some(counters) = counters_clone.upgrade() else {
break;
@@ -625,7 +543,7 @@ impl StatsManager {
counters.retain(|_, metric_data: &mut Arc<MetricData>| {
Arc::strong_count(metric_data) > 1
|| unsafe { metric_data.get_last_updated() > cutoff_time }
|| metric_data.get_last_updated() >= cutoff_millis
});
counters.shrink_to_fit();
}
@@ -663,7 +581,7 @@ impl StatsManager {
let key = entry.key();
let metric_data = entry.value();
let value = unsafe { metric_data.counter.get() };
let value = metric_data.counter.get();
metrics.push(MetricSnapshot {
name: key.name,
@@ -696,7 +614,7 @@ impl StatsManager {
let key = MetricKey::new(name, labels.clone());
if let Some(metric_data) = self.counters.get(&key) {
let value = unsafe { metric_data.counter.get() };
let value = metric_data.counter.get();
Some(MetricSnapshot {
name,
labels: labels.clone(),
@@ -797,17 +715,9 @@ mod tests {
async fn test_unsafe_counter() {
let counter = UnsafeCounter::new();
unsafe {
assert_eq!(counter.get(), 0);
counter.inc();
assert_eq!(counter.get(), 1);
counter.add(5);
assert_eq!(counter.get(), 6);
counter.set(10);
assert_eq!(counter.get(), 10);
counter.reset();
assert_eq!(counter.get(), 0);
}
assert_eq!(counter.get(), 0);
counter.add(256);
assert_eq!(counter.get(), 256);
}
#[tokio::test]
@@ -853,11 +763,11 @@ mod tests {
let stats = StatsManager::new();
let counter1 = stats.get_simple_counter(MetricName::TrafficBytesTx);
counter1.set(100);
counter1.add(100);
let labels = LabelSet::new().with_label("status", "success");
let counter2 = stats.get_counter(MetricName::PeerRpcClientTx, labels);
counter2.set(50);
counter2.add(50);
let traffic_labels = LabelSet::new()
.with_label_type(LabelType::NetworkName("default".to_string()))
@@ -865,7 +775,7 @@ mod tests {
"87ede5a2-9c3d-492d-9bbe-989b9d07e742".to_string(),
));
let counter3 = stats.get_counter(MetricName::TrafficBytesTxByInstance, traffic_labels);
counter3.set(25);
counter3.add(25);
let prometheus_output = stats.export_prometheus();
@@ -885,7 +795,7 @@ mod tests {
let labels = LabelSet::new().with_label("peer", "test");
let counter = stats.get_counter(MetricName::PeerRpcClientTx, labels.clone());
counter.set(42);
counter.add(42);
let metric = stats
.get_metric(MetricName::PeerRpcClientTx, &labels)
@@ -902,11 +812,11 @@ mod tests {
stats
.get_simple_counter(MetricName::PeerRpcClientTx)
.set(10);
stats.get_simple_counter(MetricName::PeerRpcErrors).set(2);
.add(10);
stats.get_simple_counter(MetricName::PeerRpcErrors).add(2);
stats
.get_simple_counter(MetricName::TrafficBytesTx)
.set(100);
.add(100);
let rpc_metrics = stats.get_metrics_by_prefix("peer_rpc");
assert_eq!(rpc_metrics.len(), 2);
@@ -921,19 +831,15 @@ mod tests {
// 创建一些计数器
let counter1 = stats.get_simple_counter(MetricName::PeerRpcClientTx);
counter1.set(10);
counter1.add(10);
let labels = LabelSet::new().with_label("test", "value");
let counter2 = stats.get_counter(MetricName::TrafficBytesTx, labels);
counter2.set(20);
counter2.add(20);
// 验证计数器存在
assert_eq!(stats.metric_count(), 2);
// 注意:实际的清理测试需要等待3分钟,这在单元测试中不现实
// 这里我们只验证清理机制的基本结构是否正确
// 清理逻辑在后台线程中运行,会自动删除超过3分钟未更新的条目
// 验证计数器仍然可以正常工作
counter1.inc();
assert_eq!(counter1.get(), 11);
@@ -946,14 +852,14 @@ mod tests {
async fn test_cleanup_keeps_metrics_with_live_handles() {
let stats = StatsManager::new();
let counter = stats.get_simple_counter(MetricName::TrafficBytesForwarded);
counter.set(1);
counter.add(1);
let cutoff_time = Instant::now().checked_add(Duration::from_secs(1)).unwrap();
// Use a future cutoff so last_updated check always fails
let future_cutoff = now_monotonic_millis() + 1000;
stats
.counters
.retain(|_, metric_data: &mut Arc<MetricData>| {
Arc::strong_count(metric_data) > 1
|| unsafe { metric_data.get_last_updated() > cutoff_time }
Arc::strong_count(metric_data) > 1 || metric_data.get_last_updated() > future_cutoff
});
assert_eq!(stats.metric_count(), 1);
@@ -963,8 +869,7 @@ mod tests {
stats
.counters
.retain(|_, metric_data: &mut Arc<MetricData>| {
Arc::strong_count(metric_data) > 1
|| unsafe { metric_data.get_last_updated() > cutoff_time }
Arc::strong_count(metric_data) > 1 || metric_data.get_last_updated() > future_cutoff
});
assert_eq!(stats.metric_count(), 0);
}
+72 -9
View File
@@ -65,9 +65,9 @@ use crate::vpn_portal::{self, VpnPortal};
use super::dns_server::{MAGIC_DNS_FAKE_IP, runner::DnsRunner};
use super::listeners::ListenerManager;
use super::public_ipv6_provider::{
reconcile_public_ipv6_provider_runtime, run_public_ipv6_provider_reconcile_task,
should_run_public_ipv6_provider_reconcile, validate_public_ipv6_config,
validate_public_ipv6_config_values,
PublicIpv6ProviderReconcileTask, reconcile_public_ipv6_provider_runtime,
run_public_ipv6_provider_reconcile_task, should_run_public_ipv6_provider_reconcile,
validate_public_ipv6_config, validate_public_ipv6_config_values,
};
#[cfg(feature = "socks5")]
@@ -194,6 +194,44 @@ impl NicCtxContainer {
#[cfg(feature = "tun")]
type ArcNicCtx = Arc<Mutex<Option<NicCtxContainer>>>;
type ArcPublicIpv6ProviderTaskSlot = Arc<PublicIpv6ProviderTaskSlot>;
struct PublicIpv6ProviderTaskSlot {
task: Mutex<Option<PublicIpv6ProviderReconcileTask>>,
closing: AtomicBool,
}
impl PublicIpv6ProviderTaskSlot {
fn new() -> Self {
Self {
task: Mutex::new(None),
closing: AtomicBool::new(false),
}
}
async fn ensure_started(&self, global_ctx: &ArcGlobalCtx) {
let mut task = self.task.lock().await;
if self.closing.load(Ordering::Acquire) || task.is_some() {
return;
}
*task = run_public_ipv6_provider_reconcile_task(global_ctx);
}
async fn shutdown(&self) {
self.closing.store(true, Ordering::Release);
let task = self.task.lock().await.take();
if let Some(task) = task {
task.shutdown().await;
}
}
}
async fn ensure_public_ipv6_provider_reconcile_task(
global_ctx: &ArcGlobalCtx,
task_slot: &ArcPublicIpv6ProviderTaskSlot,
) {
task_slot.ensure_started(global_ctx).await;
}
pub struct InstanceRpcServerHook {
rpc_portal_whitelist: Vec<IpCidr>,
@@ -254,6 +292,7 @@ pub struct InstanceConfigPatcher {
socks5_server: Weak<Socks5Server>,
peer_manager: Weak<PeerManager>,
conn_manager: Weak<ManualConnectorManager>,
public_ipv6_provider_task: ArcPublicIpv6ProviderTaskSlot,
}
impl InstanceConfigPatcher {
@@ -324,7 +363,6 @@ impl InstanceConfigPatcher {
self.patch_mapped_listeners(patch.mapped_listeners).await?;
self.patch_connector(patch.connectors).await?;
let provider_reconcile_was_running = should_run_public_ipv6_provider_reconcile(&global_ctx);
let mut provider_config_changed = false;
if let Some(hostname) = patch.hostname {
global_ctx.set_hostname(hostname.clone());
@@ -362,10 +400,12 @@ impl InstanceConfigPatcher {
if provider_config_changed {
reconcile_public_ipv6_provider_runtime(&global_ctx).await;
let provider_reconcile_should_run =
should_run_public_ipv6_provider_reconcile(&global_ctx);
if !provider_reconcile_was_running && provider_reconcile_should_run {
run_public_ipv6_provider_reconcile_task(&global_ctx);
if should_run_public_ipv6_provider_reconcile(&global_ctx) {
ensure_public_ipv6_provider_reconcile_task(
&global_ctx,
&self.public_ipv6_provider_task,
)
.await;
}
}
@@ -647,6 +687,7 @@ pub struct Instance {
socks5_server: Arc<Socks5Server>,
proxy_cidrs_monitor: Option<AbortOnDropHandle<()>>,
public_ipv6_provider_task: ArcPublicIpv6ProviderTaskSlot,
global_ctx: ArcGlobalCtx,
}
@@ -734,6 +775,7 @@ impl Instance {
socks5_server,
proxy_cidrs_monitor: None,
public_ipv6_provider_task: Arc::new(PublicIpv6ProviderTaskSlot::new()),
global_ctx,
}
@@ -1034,7 +1076,11 @@ impl Instance {
.await?;
self.listener_manager.lock().await.run().await?;
self.peer_manager.run().await?;
run_public_ipv6_provider_reconcile_task(&self.global_ctx);
ensure_public_ipv6_provider_reconcile_task(
&self.global_ctx,
&self.public_ipv6_provider_task,
)
.await;
#[cfg(feature = "tun")]
{
@@ -1347,6 +1393,7 @@ impl Instance {
socks5_server: Arc::downgrade(&self.socks5_server),
peer_manager: Arc::downgrade(&self.peer_manager),
conn_manager: Arc::downgrade(&self.conn_manager),
public_ipv6_provider_task: self.public_ipv6_provider_task.clone(),
}
}
@@ -1602,6 +1649,7 @@ impl Instance {
}
pub async fn clear_resources(&mut self) {
self.public_ipv6_provider_task.shutdown().await;
self.peer_manager.clear_resources().await;
#[cfg(feature = "tun")]
let _ = self.nic_ctx.lock().await.take();
@@ -1787,6 +1835,21 @@ mod tests {
);
}
#[tokio::test]
async fn public_ipv6_provider_task_slot_does_not_restart_after_shutdown() {
let global_ctx = get_mock_global_ctx();
let slot = std::sync::Arc::new(super::PublicIpv6ProviderTaskSlot::new());
global_ctx.config.set_ipv6_public_addr_provider(true);
global_ctx
.config
.set_ipv6_public_addr_prefix(Some("2001:db8::/48".parse().unwrap()));
slot.shutdown().await;
super::ensure_public_ipv6_provider_reconcile_task(&global_ctx, &slot).await;
assert!(slot.task.lock().await.is_none());
}
#[tokio::test]
async fn validate_public_ipv6_patch_allows_enabling_auto_with_manual_ipv6() {
let global_ctx = get_mock_global_ctx();
File diff suppressed because it is too large Load Diff
+8 -9
View File
@@ -24,7 +24,7 @@ use crate::{
};
use byteorder::WriteBytesExt as _;
use bytes::{BufMut, BytesMut};
use bytes::{Buf, BufMut, BytesMut};
use cidr::{Ipv4Inet, Ipv6Inet};
use futures::{SinkExt, Stream, StreamExt, lock::BiLock, ready};
use pin_project_lite::pin_project;
@@ -180,12 +180,13 @@ impl ZCPacketToBytes for TunZCPacketToBytes {
assert!(payload_offset >= 4);
let ret = if self.has_packet_info {
let mut inner = inner.split_off(payload_offset - 4);
inner.advance(payload_offset - 4);
let proto = infer_proto(&inner[4..]);
self.fill_packet_info(&mut inner[0..4], proto)?;
inner
} else {
inner.split_off(payload_offset)
inner.advance(payload_offset);
inner
};
tracing::debug!(?ret, ?payload_offset, "convert zc packet to tun packet");
@@ -1361,12 +1362,11 @@ impl NicCtx {
}
self.global_ctx
.issue_event(GlobalCtxEvent::TunDeviceReady(nic.ifname().to_string()));
.set_tun_device_ready(nic.ifname().to_string());
ret
}
Err(err) => {
self.global_ctx
.issue_event(GlobalCtxEvent::TunDeviceError(err.to_string()));
self.global_ctx.set_tun_device_error(err.to_string());
return Err(err);
}
}
@@ -1405,12 +1405,11 @@ impl NicCtx {
match nic.create_dev_for_mobile(tun_fd).await {
Ok(ret) => {
self.global_ctx
.issue_event(GlobalCtxEvent::TunDeviceReady(nic.ifname().to_string()));
.set_tun_device_ready(nic.ifname().to_string());
ret
}
Err(err) => {
self.global_ctx
.issue_event(GlobalCtxEvent::TunDeviceError(err.to_string()));
self.global_ctx.set_tun_device_error(err.to_string());
return Err(err);
}
}
+6 -1
View File
@@ -173,7 +173,12 @@ impl EasyTierLauncher {
#[cfg(mobile)]
Self::run_routine_for_mobile(&instance, &data, &mut tasks).await;
instance.run().await?;
if let Err(err) = instance.run().await {
tasks.abort_all();
drop(tasks);
instance.clear_resources().await;
return Err(err.into());
}
#[cfg(feature = "ffi-dataplane")]
data.data_plane
+2 -6
View File
@@ -990,13 +990,9 @@ mod tests {
s.sync_root_key(root_key, 2, 2, true);
assert!(s.check_replay_for_test(2, 0, SecureDatagramDirection::AToB, now + 2));
let expires_at = s.sync_rx_grace_expires_at_ms.load(Ordering::Relaxed);
assert!(!s.check_replay_for_test(
0,
1,
SecureDatagramDirection::AToB,
now + SecureDatagramSession::SYNC_RX_GRACE_AFTER_MS + 3
));
assert!(!s.check_replay_for_test(0, 1, SecureDatagramDirection::AToB, expires_at + 1));
}
#[test]
+163 -48
View File
@@ -477,6 +477,12 @@ struct PublicIpv6Lab {
extra_bridges: [&'static str; 2],
}
#[derive(Clone, Copy)]
enum PublicIpv6LabTopology {
DelegatedPrefix,
OnLinkPrefix,
}
impl PublicIpv6Lab {
const PROVIDER_NS: &'static str = "net_a";
const CLIENT_NS: &'static str = "net_b";
@@ -490,11 +496,13 @@ impl PublicIpv6Lab {
const PROVIDER_DEFAULT_FROM: &'static str = "2001:db8:100::/64";
const PROVIDER_WAN_ADDR: &'static str = "2001:db8:ffff:1::2/64";
const UPSTREAM_WAN_ADDR: &'static str = "2001:db8:ffff:1::1/64";
const ON_LINK_PROVIDER_WAN_ADDR: &'static str = "2001:db8:100::2/64";
const ON_LINK_UPSTREAM_WAN_ADDR: &'static str = "2001:db8:100::1/64";
const UPSTREAM_SERVER_ADDR: &'static str = "2001:db8:ffff:2::1/64";
const SERVER_ADDR: &'static str = "2001:db8:ffff:2::100/64";
const SERVER_IP: &'static str = "2001:db8:ffff:2::100";
fn setup() -> Self {
fn setup_with_topology(topology: PublicIpv6LabTopology) -> Self {
prepare_linux_namespaces();
del_netns(Self::UPSTREAM_NS);
@@ -544,13 +552,23 @@ impl PublicIpv6Lab {
Self::SERVER_BRIDGE,
);
let (provider_wan_addr, upstream_wan_addr) = match topology {
PublicIpv6LabTopology::DelegatedPrefix => {
(Self::PROVIDER_WAN_ADDR, Self::UPSTREAM_WAN_ADDR)
}
PublicIpv6LabTopology::OnLinkPrefix => (
Self::ON_LINK_PROVIDER_WAN_ADDR,
Self::ON_LINK_UPSTREAM_WAN_ADDR,
),
};
run_ip_in_ns(
Self::PROVIDER_NS,
&["addr", "add", Self::PROVIDER_WAN_ADDR, "dev", "pubwan0"],
&["addr", "add", provider_wan_addr, "dev", "pubwan0"],
);
run_ip_in_ns(
Self::UPSTREAM_NS,
&["addr", "add", Self::UPSTREAM_WAN_ADDR, "dev", "upwan0"],
&["addr", "add", upstream_wan_addr, "dev", "upwan0"],
);
run_ip_in_ns(
Self::UPSTREAM_NS,
@@ -561,37 +579,56 @@ impl PublicIpv6Lab {
&["addr", "add", Self::SERVER_ADDR, "dev", "srv0"],
);
run_ip_in_ns(
Self::PROVIDER_NS,
&["link", "add", "pubprefix0", "type", "dummy"],
);
run_ip_in_ns(Self::PROVIDER_NS, &["link", "set", "pubprefix0", "up"]);
run_ip_in_ns(
Self::PROVIDER_NS,
&[
"-6",
"route",
"add",
Self::PROVIDER_PREFIX,
"dev",
"pubprefix0",
],
);
run_ip_in_ns(
Self::PROVIDER_NS,
&[
"-6",
"route",
"add",
"default",
"from",
Self::PROVIDER_DEFAULT_FROM,
"via",
"2001:db8:ffff:1::1",
"dev",
"pubwan0",
],
);
match topology {
PublicIpv6LabTopology::DelegatedPrefix => {
run_ip_in_ns(
Self::PROVIDER_NS,
&["link", "add", "pubprefix0", "type", "dummy"],
);
run_ip_in_ns(Self::PROVIDER_NS, &["link", "set", "pubprefix0", "up"]);
run_ip_in_ns(
Self::PROVIDER_NS,
&[
"-6",
"route",
"add",
Self::PROVIDER_PREFIX,
"dev",
"pubprefix0",
],
);
run_ip_in_ns(
Self::PROVIDER_NS,
&[
"-6",
"route",
"add",
"default",
"from",
Self::PROVIDER_DEFAULT_FROM,
"via",
"2001:db8:ffff:1::1",
"dev",
"pubwan0",
],
);
}
PublicIpv6LabTopology::OnLinkPrefix => {
run_ip_in_ns(
Self::PROVIDER_NS,
&[
"-6",
"route",
"add",
"default",
"via",
"2001:db8:100::1",
"dev",
"pubwan0",
],
);
}
}
run_ip_in_ns(
Self::SERVER_NS,
@@ -606,19 +643,21 @@ impl PublicIpv6Lab {
"srv0",
],
);
run_ip_in_ns(
Self::UPSTREAM_NS,
&[
"-6",
"route",
"add",
Self::PROVIDER_PREFIX,
"via",
"2001:db8:ffff:1::2",
"dev",
"upwan0",
],
);
if matches!(topology, PublicIpv6LabTopology::DelegatedPrefix) {
run_ip_in_ns(
Self::UPSTREAM_NS,
&[
"-6",
"route",
"add",
Self::PROVIDER_PREFIX,
"via",
"2001:db8:ffff:1::2",
"dev",
"upwan0",
],
);
}
run_sysctl_in_ns(Self::PROVIDER_NS, "net.ipv6.conf.all.forwarding=1");
run_sysctl_in_ns(Self::UPSTREAM_NS, "net.ipv6.conf.all.forwarding=1");
@@ -672,7 +711,15 @@ fn get_public_ipv6_config(
async fn init_public_ipv6_two_node(
client_inst_id: uuid::Uuid,
) -> (PublicIpv6Lab, Instance, Instance) {
let lab = PublicIpv6Lab::setup();
init_public_ipv6_two_node_with_topology(client_inst_id, PublicIpv6LabTopology::DelegatedPrefix)
.await
}
async fn init_public_ipv6_two_node_with_topology(
client_inst_id: uuid::Uuid,
topology: PublicIpv6LabTopology,
) -> (PublicIpv6Lab, Instance, Instance) {
let lab = PublicIpv6Lab::setup_with_topology(topology);
let provider_cfg = get_public_ipv6_config(
"provider_public_ipv6",
@@ -756,6 +803,13 @@ fn addr_exists_in_ns(ns: &str, dev: &str, needle: &str) -> bool {
run_ip_in_ns_output(ns, &["-6", "addr", "show", "dev", dev]).contains(needle)
}
fn ndp_proxy_exists_in_ns(ns: &str, dev: &str, addr: std::net::Ipv6Addr) -> bool {
let addr = addr.to_string();
run_ip_in_ns_output(ns, &["-6", "neigh", "show", "proxy", "dev", dev])
.lines()
.any(|line| line.split_whitespace().next() == Some(addr.as_str()))
}
#[tokio::test]
#[serial_test::serial]
pub async fn public_ipv6_auto_addr_end_to_end() {
@@ -878,6 +932,67 @@ pub async fn public_ipv6_auto_addr_end_to_end() {
drop_insts(vec![provider, client]).await;
}
#[tokio::test]
#[serial_test::serial]
pub async fn public_ipv6_auto_addr_on_link_ndp_proxy_end_to_end() {
let client_id = uuid::Uuid::parse_str("44444444-4444-4444-4444-444444444444").unwrap();
let (_lab, provider, client) =
init_public_ipv6_two_node_with_topology(client_id, PublicIpv6LabTopology::OnLinkPrefix)
.await;
wait_for_condition(
|| async {
provider
.get_global_ctx()
.get_advertised_ipv6_public_addr_prefix()
== Some(PublicIpv6Lab::PROVIDER_PREFIX.parse().unwrap())
},
Duration::from_secs(10),
)
.await;
let leased = wait_for_public_ipv6_addr(&client).await;
wait_for_public_ipv6_route(&provider, leased).await;
wait_for_condition(
|| async {
addr_exists_in_ns(
PublicIpv6Lab::CLIENT_NS,
PublicIpv6Lab::CLIENT_TUN,
&leased.to_string(),
) && route_exists_in_ns(
PublicIpv6Lab::PROVIDER_NS,
&format!("{} dev {}", leased.address(), PublicIpv6Lab::PROVIDER_TUN),
)
},
Duration::from_secs(10),
)
.await;
wait_for_condition(
|| async {
ndp_proxy_exists_in_ns(PublicIpv6Lab::PROVIDER_NS, "pubwan0", leased.address())
},
Duration::from_secs(20),
)
.await;
wait_for_condition(
|| async {
ping6_test(
PublicIpv6Lab::SERVER_NS,
leased.address().to_string().as_str(),
None,
)
.await
},
Duration::from_secs(20),
)
.await;
drop_insts(vec![provider, client]).await;
}
#[tokio::test]
#[serial_test::serial]
pub async fn public_ipv6_auto_addr_reconnect_reuses_same_address() {
+85 -26
View File
@@ -1,9 +1,10 @@
use bytes::Buf;
use bytes::Bytes;
use bytes::BytesMut;
use zerocopy::byteorder::*;
use zerocopy::AsBytes;
use zerocopy::FromBytes;
use zerocopy::FromZeroes;
use zerocopy::byteorder::*;
type DefaultEndian = LittleEndian;
@@ -485,8 +486,16 @@ impl ZCPacket {
let payload_off = ret.packet_type.get_packet_offsets().payload_offset;
let total_len = payload_off + payload.len();
ret.inner.reserve(total_len);
unsafe { ret.inner.set_len(total_len) };
ret.mut_payload()[..payload.len()].copy_from_slice(payload);
// SAFETY: `reserve` guarantees capacity >= total_len.
// We zero the header region and copy payload before advancing length,
// so every byte in [0..total_len) is initialized before any read.
unsafe {
let ptr = ret.inner.as_mut_ptr();
std::ptr::write_bytes(ptr, 0, payload_off);
std::ptr::copy_nonoverlapping(payload.as_ptr(), ptr.add(payload_off), payload.len());
ret.inner.set_len(total_len);
}
ret
}
@@ -494,12 +503,12 @@ impl ZCPacket {
let mut ret = Self::new_nic_packet();
ret.inner.reserve(cap);
let total_len = ret.packet_type.get_packet_offsets().payload_offset - packet_info_len;
unsafe { ret.inner.set_len(total_len) };
ret.inner.resize(total_len, 0);
ret
}
pub fn new_for_foreign_network(
network_name: &String,
network_name: &str,
dst_peer_id: u32,
foreign_zc_packet: &ZCPacket,
) -> Self {
@@ -508,26 +517,71 @@ impl ZCPacket {
foreign_network_hdr.get_header_len() + foreign_zc_packet.tunnel_payload().len();
let mut ret = Self::new_nic_packet();
let payload_off = ret.packet_type.get_packet_offsets().payload_offset;
ret.inner.reserve(payload_off + total_payload_len);
unsafe { ret.inner.set_len(payload_off + total_payload_len) };
let offsets = ret.packet_type.get_packet_offsets();
let payload_off = offsets.payload_offset;
let pm_hdr_off = offsets.peer_manager_header_offset;
let total_len = payload_off + total_payload_len;
ret.inner.reserve(total_len);
let fixed_hdr_len = std::mem::size_of::<ForeignNetworkPacketHeader>();
ret.mut_payload()[..fixed_hdr_len].copy_from_slice(foreign_network_hdr.as_bytes());
let name_offset = foreign_network_hdr.network_name_offset.get() as usize;
let name_len = foreign_network_hdr.network_name_len.get() as usize;
ret.mut_payload()[name_offset..name_offset + name_len]
.copy_from_slice(network_name.as_bytes());
let foreign_payload = foreign_zc_packet.tunnel_payload();
ret.mut_payload()[foreign_network_hdr.get_header_len()..]
.copy_from_slice(foreign_zc_packet.tunnel_payload());
// Construct the PeerManagerHeader on the stack so we can write it
// directly into the buffer, avoiding a separate mut_peer_manager_header()
// call after set_len.
let pm_hdr = PeerManagerHeader {
from_peer_id: 0.into(),
to_peer_id: 0.into(),
packet_type: PacketType::ForeignNetworkPacket as u8,
flags: 0,
forward_counter: 0,
reserved: 0,
len: U32::new(total_payload_len as u32),
};
let hdr = ret.mut_peer_manager_header().unwrap();
hdr.from_peer_id = 0.into();
hdr.to_peer_id = 0.into();
hdr.packet_type = PacketType::ForeignNetworkPacket as u8;
hdr.len.set(total_payload_len as u32);
// SAFETY: `reserve` guarantees capacity >= total_len.
// We zero only the tunnel-header reserved space [0..pm_hdr_off], write
// the PeerManagerHeader directly at pm_hdr_off, then copy the foreign
// network header, network name, and payload. Every byte in [0..total_len)
// is initialized before set_len.
unsafe {
let ptr = ret.inner.as_mut_ptr();
// Zero the tunnel header reserved space only (not the PM header region)
std::ptr::write_bytes(ptr, 0, pm_hdr_off);
// Write PeerManagerHeader directly
std::ptr::copy_nonoverlapping(
pm_hdr.as_bytes().as_ptr(),
ptr.add(pm_hdr_off),
std::mem::size_of::<PeerManagerHeader>(),
);
// Copy foreign network fixed header
std::ptr::copy_nonoverlapping(
foreign_network_hdr.as_bytes().as_ptr(),
ptr.add(payload_off),
fixed_hdr_len,
);
// Copy network name
std::ptr::copy_nonoverlapping(
network_name.as_ptr(),
ptr.add(payload_off + name_offset),
name_len,
);
// Copy foreign payload
std::ptr::copy_nonoverlapping(
foreign_payload.as_ptr(),
ptr.add(payload_off + foreign_network_hdr.get_header_len()),
foreign_payload.len(),
);
ret.inner.set_len(total_len);
}
ret
}
@@ -587,7 +641,8 @@ impl ZCPacket {
}
pub fn payload_bytes(mut self) -> BytesMut {
self.inner.split_off(self.payload_offset())
self.inner.advance(self.payload_offset());
self.inner
}
pub fn peer_manager_header(&self) -> Option<&PeerManagerHeader> {
@@ -652,11 +707,12 @@ impl ZCPacket {
}
pub fn tunnel_payload_bytes(mut self) -> BytesMut {
self.inner.split_off(
self.inner.advance(
self.packet_type
.get_packet_offsets()
.peer_manager_header_offset,
)
);
self.inner
}
pub fn convert_type(mut self, target_packet_type: ZCPacketType) -> Self {
@@ -697,12 +753,13 @@ impl ZCPacket {
.get_packet_offsets()
.peer_manager_header_offset;
let mut buf = BytesMut::with_capacity(new_pm_offset + tunnel_payload.len());
unsafe { buf.set_len(new_pm_offset) };
buf.resize(new_pm_offset, 0);
buf.extend_from_slice(tunnel_payload);
return Self::new_from_buf(buf, target_packet_type);
}
Self::new_from_buf(self.inner.split_off(new_offset), target_packet_type)
self.inner.advance(new_offset);
Self::new_from_buf(self.inner, target_packet_type)
}
pub fn into_bytes(self) -> Bytes {
@@ -748,8 +805,10 @@ impl ZCPacket {
let foreign_hdr_len = hdr.get_header_len();
Self::new_from_buf(
self.inner
.split_off(foreign_hdr_len + self.payload_offset()),
{
self.inner.advance(foreign_hdr_len + self.payload_offset());
self.inner
},
ZCPacketType::DummyTunnel,
)
}
+21 -32
View File
@@ -1,7 +1,4 @@
use std::{
cell::UnsafeCell,
sync::atomic::{AtomicU32, Ordering::Relaxed},
};
use std::sync::atomic::{AtomicU32, AtomicU64, Ordering::Relaxed};
pub struct WindowLatency {
latency_us_window: Vec<AtomicU32>,
@@ -63,34 +60,30 @@ impl WindowLatency {
#[derive(Debug)]
pub struct Throughput {
tx_bytes: UnsafeCell<u64>,
rx_bytes: UnsafeCell<u64>,
tx_packets: UnsafeCell<u64>,
rx_packets: UnsafeCell<u64>,
tx_bytes: AtomicU64,
rx_bytes: AtomicU64,
tx_packets: AtomicU64,
rx_packets: AtomicU64,
}
impl Clone for Throughput {
fn clone(&self) -> Self {
Self {
tx_bytes: UnsafeCell::new(unsafe { *self.tx_bytes.get() }),
rx_bytes: UnsafeCell::new(unsafe { *self.rx_bytes.get() }),
tx_packets: UnsafeCell::new(unsafe { *self.tx_packets.get() }),
rx_packets: UnsafeCell::new(unsafe { *self.rx_packets.get() }),
tx_bytes: AtomicU64::new(self.tx_bytes.load(Relaxed)),
rx_bytes: AtomicU64::new(self.rx_bytes.load(Relaxed)),
tx_packets: AtomicU64::new(self.tx_packets.load(Relaxed)),
rx_packets: AtomicU64::new(self.rx_packets.load(Relaxed)),
}
}
}
// add sync::Send and sync::Sync traits to Throughput
unsafe impl Send for Throughput {}
unsafe impl Sync for Throughput {}
impl Default for Throughput {
fn default() -> Self {
Self {
tx_bytes: UnsafeCell::new(0),
rx_bytes: UnsafeCell::new(0),
tx_packets: UnsafeCell::new(0),
rx_packets: UnsafeCell::new(0),
tx_bytes: AtomicU64::new(0),
rx_bytes: AtomicU64::new(0),
tx_packets: AtomicU64::new(0),
rx_packets: AtomicU64::new(0),
}
}
}
@@ -101,32 +94,28 @@ impl Throughput {
}
pub fn tx_bytes(&self) -> u64 {
unsafe { *self.tx_bytes.get() }
self.tx_bytes.load(Relaxed)
}
pub fn rx_bytes(&self) -> u64 {
unsafe { *self.rx_bytes.get() }
self.rx_bytes.load(Relaxed)
}
pub fn tx_packets(&self) -> u64 {
unsafe { *self.tx_packets.get() }
self.tx_packets.load(Relaxed)
}
pub fn rx_packets(&self) -> u64 {
unsafe { *self.rx_packets.get() }
self.rx_packets.load(Relaxed)
}
pub fn record_tx_bytes(&self, bytes: u64) {
unsafe {
*self.tx_bytes.get() += bytes;
*self.tx_packets.get() += 1;
}
self.tx_bytes.fetch_add(bytes, Relaxed);
self.tx_packets.fetch_add(1, Relaxed);
}
pub fn record_rx_bytes(&self, bytes: u64) {
unsafe {
*self.rx_bytes.get() += bytes;
*self.rx_packets.get() += 1;
}
self.rx_bytes.fetch_add(bytes, Relaxed);
self.rx_packets.fetch_add(1, Relaxed);
}
}