mz_ore/id_gen.rs
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557
// Copyright Materialize, Inc. and contributors. All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License in the LICENSE file at the
// root of this repository, or online at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
//! ID generation utilities.
use hibitset::BitSet;
use std::borrow::Borrow;
use std::fmt;
use std::hash::Hash;
use std::marker::PhantomData;
use std::ops::{AddAssign, Sub};
use std::sync::atomic::{AtomicU64, Ordering};
use std::sync::{Arc, Mutex};
use uuid::Uuid;
use rand::rngs::StdRng;
use rand::{Rng, SeedableRng};
use crate::cast::CastFrom;
/// Manages the allocation of unique IDs.
#[derive(Debug, Clone)]
pub struct Gen<Id> {
id: u64,
phantom: PhantomData<Id>,
}
impl<Id> Default for Gen<Id> {
fn default() -> Self {
Self {
id: 0,
phantom: PhantomData,
}
}
}
impl<Id: From<u64>> Gen<Id> {
/// Allocates a new identifier of type `Id` and advances the generator.
pub fn allocate_id(&mut self) -> Id {
let id = self.id;
self.id += 1;
id.into()
}
}
/// A generator of u64-bit IDs.
pub type IdGen = Gen<u64>;
/// Manages the allocation of unique IDs.
///
/// Atomic version of `Gen`, for sharing between threads.
#[derive(Debug)]
pub struct AtomicGen<Id> {
id: AtomicU64,
phantom: PhantomData<Id>,
}
impl<Id> Default for AtomicGen<Id> {
fn default() -> Self {
Self {
id: AtomicU64::new(0),
phantom: PhantomData,
}
}
}
impl<Id: From<u64> + Default> AtomicGen<Id> {
/// Allocates a new identifier of type `Id` and advances the generator.
pub fn allocate_id(&self) -> Id {
// The only purpose of the atomic here is to ensure every caller receives a distinct ID,
// there are no requirements on the order in which IDs are produced and there is no other
// state protected by this atomic. `Relaxed` ordering is therefore sufficient.
let id = self.id.fetch_add(1, Ordering::Relaxed);
id.into()
}
}
/// A generator of u64-bit IDs.
///
/// Atomic version of `IdGen`, for sharing between threads.
pub type AtomicIdGen = AtomicGen<u64>;
/// IdAllocator common traits.
pub trait IdGenerator:
From<u8> + AddAssign + Sub + PartialOrd + Copy + Eq + Hash + Ord + serde::Serialize + fmt::Display
{
}
impl<T> IdGenerator for T where
T: From<u8>
+ AddAssign
+ Sub
+ PartialOrd
+ Copy
+ Eq
+ Hash
+ Ord
+ serde::Serialize
+ fmt::Display
{
}
/// Manages allocation of numeric IDs.
#[derive(Debug, Clone)]
pub struct IdAllocator<A: IdAllocatorInner>(pub Arc<Mutex<A>>);
/// Common trait for id allocators.
pub trait IdAllocatorInner: std::fmt::Debug + Send {
/// Name of the allocator.
const NAME: &'static str;
/// Construct an allocator with the given range. Returned ids will be OR'd with `mask`. `mask`
/// must not have any bits that could be set by a number <= `max`.
fn new(min: u32, max: u32, mask: u32) -> Self;
/// Allocate a new id.
fn alloc(&mut self) -> Option<u32>;
/// Deallocate a used id, making it available for reuse.
fn remove(&mut self, id: u32);
}
/// IdAllocator using a HiBitSet.
#[derive(Debug, Clone)]
pub struct IdAllocatorInnerBitSet {
next: StdRng,
min: u32,
max: u32,
mask: u32,
used: BitSet,
}
impl IdAllocatorInner for IdAllocatorInnerBitSet {
const NAME: &'static str = "hibitset";
fn new(min: u32, max: u32, mask: u32) -> Self {
let total = usize::cast_from(max - min);
assert!(total < BitSet::BITS_PER_USIZE.pow(4));
IdAllocatorInnerBitSet {
next: StdRng::from_entropy(),
min,
max,
mask,
used: BitSet::new(),
}
}
fn alloc(&mut self) -> Option<u32> {
let range = self.min..=self.max;
let init = self.next.gen_range(range);
let mut next = init;
loop {
// Because hibitset has a hard maximum of 64**4 (~16 million), subtract the min in case
// max is above that. This is safe because we already asserted above that `max - min <
// 64**4`.
let stored = next - self.min;
if !self.used.add(stored) {
assert!(
next & self.mask == 0,
"chosen ID must not intersect with mask:\n{:#034b}\n{:#034b}",
next,
self.mask
);
return Some(next | self.mask);
}
// Existing value, increment and try again. Wrap once we hit max back to min.
next = if next == self.max { self.min } else { next + 1 };
// We fully wrapped around. BitSet doesn't have a rank or count method, so we can't
// compute this early.
if next == init {
return None;
}
}
}
fn remove(&mut self, id: u32) {
let id = (!self.mask) & id;
let stored = id - self.min;
self.used.remove(stored);
}
}
impl<A: IdAllocatorInner> IdAllocator<A> {
/// Creates a new `IdAllocator` that will assign IDs between `min` and
/// `max`, both inclusive.
pub fn new(min: u32, max: u32, mask: u32) -> IdAllocator<A> {
assert!(min <= max);
if mask != 0 && max > 0 {
// mask_check is all 1s in any bit set by all numbers >= max. Assert that the mask
// doesn't share any bits with those.
let mask_check = (1 << (max.ilog2() + 1)) - 1;
assert_eq!(mask & mask_check, 0, "max and mask share bits");
}
let inner = A::new(min, max, mask);
IdAllocator(Arc::new(Mutex::new(inner)))
}
/// Allocates a new ID randomly distributed between min and max.
///
/// Returns `None` if the allocator is exhausted.
///
/// The ID associated with the [`IdHandle`] will be freed when all of the
/// outstanding [`IdHandle`]s have been dropped.
pub fn alloc(&self) -> Option<IdHandle<u32, A>> {
let inner = Arc::new(internal::IdHandleInner::new(self)?);
Some(IdHandle::Dynamic(inner))
}
// Attempt to allocate a new ID. We want the ID to be randomly distributed in the range. To do
// this, choose a random candidate. Check if it's already in the used set, and increment in a
// loop until an unused slot is found. Ideally we could ask used set what its next used or
// unused id is after some given X, but that's not part of the API. This means that when the
// range is large and the used set is near full, we will spend a lot of cycles looking for an
// open slot. However, we limit the number of connections to the thousands, and connection IDs
// have 20 bits (~1 million) of space, so it is not currently possible to enter that state.
fn alloc_internal(&self) -> Option<u32> {
let mut inner = self.0.lock().expect("lock poisoned");
inner.alloc()
}
fn free_internal(&self, id: u32) {
let mut inner = self.0.lock().expect("lock poisoned");
inner.remove(id);
}
}
/// A clone-able owned reference to an ID.
///
/// Once all of the [`IdHandle`]s referencing an ID have been dropped, we will then free the ID
/// for later re-use.
#[derive(Debug)]
pub enum IdHandle<T, A: IdAllocatorInner> {
/// An ID "allocated" at compile time.
///
/// Note: It is *entirely* up to the caller to make sure the provided ID is
/// not used by a dynamic ID allocator.
Static(T),
/// An ID allocated at runtime, gets freed once all handles have been dropped.
Dynamic(Arc<internal::IdHandleInner<T, A>>),
}
impl<T: Clone, A: IdAllocatorInner> Clone for IdHandle<T, A> {
fn clone(&self) -> Self {
match self {
IdHandle::Static(t) => IdHandle::Static(t.clone()),
IdHandle::Dynamic(handle) => IdHandle::Dynamic(Arc::clone(handle)),
}
}
}
impl<T: IdGenerator, A: IdAllocatorInner> IdHandle<T, A> {
/// Returns the raw ID inside of this handle.
///
/// Use with caution! It is easy for a raw ID to outlive the handle from
/// which it came. You are responsible for ensuring that your use of the raw
/// ID does not lead to ID reuse bugs.
pub fn unhandled(&self) -> T {
*self.borrow()
}
}
impl<T: IdGenerator, A: IdAllocatorInner> PartialEq for IdHandle<T, A> {
fn eq(&self, other: &Self) -> bool {
self.unhandled() == other.unhandled()
}
}
impl<T: IdGenerator, A: IdAllocatorInner> Eq for IdHandle<T, A> {}
impl<T: IdGenerator, A: IdAllocatorInner> PartialOrd for IdHandle<T, A> {
fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
Some(self.cmp(other))
}
}
impl<T: IdGenerator, A: IdAllocatorInner> Ord for IdHandle<T, A> {
fn cmp(&self, other: &Self) -> std::cmp::Ordering {
self.unhandled().cmp(&other.unhandled())
}
}
impl<T, A: IdAllocatorInner> Borrow<T> for IdHandle<T, A> {
fn borrow(&self) -> &T {
match self {
IdHandle::Static(id) => id,
IdHandle::Dynamic(inner) => &inner.id,
}
}
}
impl<T: IdGenerator, A: IdAllocatorInner> fmt::Display for IdHandle<T, A> {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
self.unhandled().fmt(f)
}
}
impl<T: IdGenerator, A: IdAllocatorInner> serde::Serialize for IdHandle<T, A> {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::Serializer,
{
self.unhandled().serialize(serializer)
}
}
mod internal {
use std::sync::Arc;
use crate::cast::CastFrom;
use crate::id_gen::{IdAllocator, IdAllocatorInner};
#[derive(Debug)]
pub struct IdHandleInner<T, A: IdAllocatorInner> {
/// A handle to the [`IdAllocator`] used to allocated the provided id.
pub(super) allocator: IdAllocator<A>,
/// The actual ID that was allocated.
pub(super) id: T,
stored: u32,
}
impl<T, A: IdAllocatorInner> IdHandleInner<T, A>
where
T: CastFrom<u32>,
{
pub fn new(allocator: &IdAllocator<A>) -> Option<Self> {
let stored = allocator.alloc_internal()?;
Some(IdHandleInner {
allocator: IdAllocator(Arc::clone(&allocator.0)),
id: T::cast_from(stored),
stored,
})
}
}
impl<T, A: IdAllocatorInner> Drop for IdHandleInner<T, A> {
fn drop(&mut self) {
// Release our ID for later re-use.
self.allocator.free_internal(self.stored);
}
}
}
/// Number of bits the org id is offset into a connection id.
pub const ORG_ID_OFFSET: usize = 19;
/// Max (inclusive) connection id that can be produced.
pub const MAX_ORG_ID: u32 = (1 << ORG_ID_OFFSET) - 1;
/// Extracts the lower 12 bits from an org id. These are later used as the [31, 20] bits of a
/// connection id to help route cancellation requests.
pub fn org_id_conn_bits(uuid: &Uuid) -> u32 {
let lower = uuid.as_u128();
let lower = (lower & 0xFFF) << ORG_ID_OFFSET;
let lower: u32 = lower.try_into().expect("must fit");
lower
}
/// Returns the portion of the org's UUID present in connection id.
pub fn conn_id_org_uuid(conn_id: u32) -> String {
const UPPER: [char; 16] = [
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F',
];
// Extract UUID from conn_id: upper 12 bits excluding the first.
let orgid = usize::try_from((conn_id >> ORG_ID_OFFSET) & 0xFFF).expect("must cast");
// Convert the bits into a 3 char string and inject into the resolver template.
let mut dst = String::with_capacity(3);
dst.push(UPPER[(orgid >> 8) & 0xf]);
dst.push(UPPER[(orgid >> 4) & 0xf]);
dst.push(UPPER[orgid & 0xf]);
dst
}
/// Generate a random temporary ID.
///
/// Concretely we generate a UUIDv4 and return the last 12 characters for maximum uniqueness.
///
/// Note: the reason we use the last 12 characters is because the bits 6, 7, and 12 - 15
/// are all hard coded <https://www.rfc-editor.org/rfc/rfc4122#section-4.4>.
/// ```
/// use mz_ore::id_gen::temp_id;
///
/// let temp = temp_id();
/// assert_eq!(temp.len(), 12);
/// assert!(temp.is_ascii());
/// ```
pub fn temp_id() -> String {
let temp_uuid = uuid::Uuid::new_v4().as_hyphenated().to_string();
temp_uuid.chars().rev().take_while(|c| *c != '-').collect()
}
#[cfg(test)]
mod tests {
use std::collections::BTreeMap;
use crate::assert_none;
use super::*;
#[crate::test]
fn test_conn_org() {
let uuid = Uuid::parse_str("9e37ec59-56f4-450a-acbd-18ff14f10ca8").unwrap();
let lower = org_id_conn_bits(&uuid);
let org_lower_uuid = conn_id_org_uuid(lower);
assert_eq!(org_lower_uuid, "CA8");
}
#[crate::test]
fn test_id_gen() {
test_ad_allocator::<IdAllocatorInnerBitSet>();
}
// Test masks and maxs that intersect panic.
#[crate::test]
#[should_panic]
fn test_mask_intersect<A: IdAllocatorInner>() {
let env_lower = org_id_conn_bits(&uuid::Uuid::from_u128(u128::MAX));
let ida = IdAllocator::<IdAllocatorInnerBitSet>::new(
1 << ORG_ID_OFFSET,
1 << ORG_ID_OFFSET,
env_lower,
);
let id = ida.alloc().unwrap();
assert_eq!(id.unhandled(), (0xfff << ORG_ID_OFFSET) | MAX_ORG_ID);
}
fn test_ad_allocator<A: IdAllocatorInner>() {
test_id_alloc::<A>();
test_static_id_sorting::<A>();
test_id_reuse::<A>();
test_display::<A>();
test_map_lookup::<A>();
test_serialization::<A>();
test_mask::<A>();
test_mask_envd::<A>();
}
fn test_mask<A: IdAllocatorInner>() {
let ida = IdAllocator::<A>::new(1, 1, 0xfff << 20);
let id = ida.alloc().unwrap();
assert_eq!(id.unhandled(), (0xfff << 20) | 1);
}
// Test that the random conn id and and uuid each with all bits set don't intersect.
fn test_mask_envd<A: IdAllocatorInner>() {
let env_lower = org_id_conn_bits(&uuid::Uuid::from_u128(u128::MAX));
let ida = IdAllocator::<A>::new(MAX_ORG_ID, MAX_ORG_ID, env_lower);
let id = ida.alloc().unwrap();
assert_eq!(id.unhandled(), (0xfff << ORG_ID_OFFSET) | MAX_ORG_ID);
}
fn test_id_alloc<A: IdAllocatorInner>() {
let ida = IdAllocator::<A>::new(3, 5, 0);
let id3 = ida.alloc().unwrap();
let id4 = ida.alloc().unwrap();
let id5 = ida.alloc().unwrap();
assert_ne!(id3, id4);
assert_ne!(id3, id5);
assert_ne!(id4, id5);
drop(id4);
let _id4 = ida.alloc().unwrap();
drop(id5);
drop(id3);
let _id5 = ida.alloc().unwrap();
let _id3 = ida.alloc().unwrap();
match ida.alloc() {
Some(id) => panic!(
"id allocator returned {}, not expected id exhaustion error",
id
),
None => (),
}
}
fn test_static_id_sorting<A: IdAllocatorInner>() {
let ida = IdAllocator::<A>::new(0, 0, 0);
let id0 = ida.alloc().unwrap();
let id1 = IdHandle::Static(1);
assert!(id0 < id1);
let ida = IdAllocator::<A>::new(1, 1, 0);
let id0 = IdHandle::Static(0);
let id1 = ida.alloc().unwrap();
assert!(id0 < id1);
}
fn test_id_reuse<A: IdAllocatorInner>() {
let allocator = IdAllocator::<A>::new(10, 11, 0);
let id_a = allocator.alloc().unwrap();
let a = id_a.unhandled();
let id_a_clone = id_a.clone();
// a should not get freed.
drop(id_a);
// There are only two slots, so trying to allocate 2 more should fail the second time.
let _id_b = allocator.alloc().unwrap();
assert_none!(allocator.alloc());
// a should get freed since all outstanding references have been dropped.
drop(id_a_clone);
// We should re-use a.
let id_c = allocator.alloc().unwrap();
assert_eq!(id_c.unhandled(), a);
}
fn test_display<A: IdAllocatorInner>() {
let allocator = IdAllocator::<A>::new(65_000, 65_000, 0);
let id_a = allocator.alloc().unwrap();
assert_eq!(id_a.unhandled(), 65_000);
// An IdHandle should use the inner type's Display impl.
let id_display = format!("{id_a}");
let val_display = format!("{}", id_a.unhandled());
assert_eq!(id_display, val_display);
}
fn test_map_lookup<A: IdAllocatorInner>() {
let allocator = IdAllocator::<A>::new(99, 101, 0);
let id_a = allocator.alloc().unwrap();
let a = id_a.unhandled();
let mut btree = BTreeMap::new();
btree.insert(id_a, "hello world");
// We should be able to lookup an IdHandle, based on just the value.
let entry = btree.remove(&a).unwrap();
assert_eq!(entry, "hello world");
assert!(btree.is_empty());
}
fn test_serialization<A: IdAllocatorInner>() {
let allocator = IdAllocator::<A>::new(42, 42, 0);
let id_a = allocator.alloc().unwrap();
assert_eq!(id_a.unhandled(), 42);
// An IdHandle should serialize the same as the inner value.
let id_json = serde_json::to_string(&id_a).unwrap();
let val_json = serde_json::to_string(&id_a.unhandled()).unwrap();
assert_eq!(id_json, val_json);
}
}