mirror of
https://github.com/bspeice/aeron-rs
synced 2024-12-21 05:28:10 -05:00
Start porting over existing tests from Aeron
This commit is contained in:
parent
b902f580c9
commit
b567336bb0
@ -62,6 +62,20 @@ pub trait AtomicBuffer: Deref<Target = [u8]> + DerefMut<Target = [u8]> {
|
||||
})
|
||||
}
|
||||
|
||||
/// Overlay a mutable value on the buffer.
|
||||
///
|
||||
/// NOTE: Has the potential to cause undefined behavior if alignment is incorrect
|
||||
fn overlay_mut<T>(&mut self, offset: IndexT) -> Result<&mut T>
|
||||
where
|
||||
T: Sized
|
||||
{
|
||||
self.bounds_check(offset, size_of::<T>() as IndexT)
|
||||
.map(|_| {
|
||||
let offset_ptr = unsafe { self.as_mut_ptr().offset(offset as isize) };
|
||||
unsafe { &mut *(offset_ptr as *mut T) }
|
||||
})
|
||||
}
|
||||
|
||||
/// Overlay a struct on a buffer, and perform a volatile read
|
||||
///
|
||||
/// ```rust
|
||||
@ -155,6 +169,11 @@ pub trait AtomicBuffer: Deref<Target = [u8]> + DerefMut<Target = [u8]> {
|
||||
self.overlay_volatile::<i64>(offset)
|
||||
}
|
||||
|
||||
/// Read an `i64` value from the buffer without performing any synchronization
|
||||
fn get_i64(&self, offset: IndexT) -> Result<i64> {
|
||||
self.overlay::<i64>(offset).map(|i| *i)
|
||||
}
|
||||
|
||||
/// Perform a volatile write of an `i64` value
|
||||
///
|
||||
/// ```rust
|
||||
@ -167,6 +186,18 @@ pub trait AtomicBuffer: Deref<Target = [u8]> + DerefMut<Target = [u8]> {
|
||||
self.write_volatile::<i64>(offset, value)
|
||||
}
|
||||
|
||||
/// Write an `i64` value into the buffer without performing any synchronization
|
||||
///
|
||||
/// ```rust
|
||||
/// # use aeron_rs::client::concurrent::AtomicBuffer;
|
||||
/// let mut buffer = vec![0u8; 8];
|
||||
/// buffer.put_i64(0, 12);
|
||||
/// assert_eq!(buffer.get_i64(0), Ok(12));
|
||||
/// ```
|
||||
fn put_i64(&mut self, offset: IndexT, value: i64) -> Result<()> {
|
||||
self.overlay_mut::<i64>(offset).map(|i| *i = value)
|
||||
}
|
||||
|
||||
/// Write the contents of one buffer to another. Does not perform any synchronization
|
||||
fn put_bytes<B>(
|
||||
&mut self,
|
||||
@ -209,6 +240,11 @@ pub trait AtomicBuffer: Deref<Target = [u8]> + DerefMut<Target = [u8]> {
|
||||
self.overlay_volatile::<i32>(offset)
|
||||
}
|
||||
|
||||
/// Read an `i32` value from the buffer without performing any synchronization
|
||||
fn get_i32(&self, offset: IndexT) -> Result<i32> {
|
||||
self.overlay::<i32>(offset).map(|i| *i)
|
||||
}
|
||||
|
||||
/// Perform a volatile write of an `i32` into the buffer
|
||||
///
|
||||
/// ```rust
|
||||
@ -220,6 +256,11 @@ pub trait AtomicBuffer: Deref<Target = [u8]> + DerefMut<Target = [u8]> {
|
||||
fn put_i32_ordered(&mut self, offset: IndexT, value: i32) -> Result<()> {
|
||||
self.write_volatile::<i32>(offset, value)
|
||||
}
|
||||
|
||||
/// Return the total number of bytes in this buffer
|
||||
fn capacity(&self) -> IndexT {
|
||||
self.len() as IndexT
|
||||
}
|
||||
}
|
||||
|
||||
impl AtomicBuffer for Vec<u8> {}
|
||||
|
@ -2,7 +2,7 @@
|
||||
use crate::client::concurrent::AtomicBuffer;
|
||||
use crate::util::bit::align;
|
||||
use crate::util::{bit, AeronError, IndexT, Result};
|
||||
use std::ops::Deref;
|
||||
use std::ops::{Deref, DerefMut};
|
||||
|
||||
/// Description of the Ring Buffer schema.
|
||||
pub mod buffer_descriptor {
|
||||
@ -97,10 +97,16 @@ pub mod record_descriptor {
|
||||
record_offset + HEADER_LENGTH
|
||||
}
|
||||
|
||||
pub(super) fn length_offset(record_offset: IndexT) -> IndexT {
|
||||
/// Return the position of the record length field given a record's starting position
|
||||
pub fn length_offset(record_offset: IndexT) -> IndexT {
|
||||
record_offset
|
||||
}
|
||||
|
||||
/// Return the position of the record message type field given a record's starting position
|
||||
pub fn type_offset(record_offset: IndexT) -> IndexT {
|
||||
record_offset + size_of::<i32>() as IndexT
|
||||
}
|
||||
|
||||
pub(super) fn record_length(header: i64) -> i32 {
|
||||
header as i32
|
||||
}
|
||||
@ -110,6 +116,8 @@ pub mod record_descriptor {
|
||||
}
|
||||
}
|
||||
|
||||
const INSUFFICIENT_CAPACITY: IndexT = -2;
|
||||
|
||||
/// Multi-producer, single-consumer ring buffer implementation.
|
||||
pub struct ManyToOneRingBuffer<A>
|
||||
where
|
||||
@ -150,6 +158,11 @@ where
|
||||
.unwrap()
|
||||
}
|
||||
|
||||
/// Return the total number of bytes in this buffer
|
||||
pub fn capacity(&self) -> IndexT {
|
||||
self.capacity
|
||||
}
|
||||
|
||||
/// Write a message into the ring buffer
|
||||
pub fn write<B>(
|
||||
&mut self,
|
||||
@ -157,7 +170,7 @@ where
|
||||
source: &B,
|
||||
source_index: IndexT,
|
||||
length: IndexT,
|
||||
) -> Result<()>
|
||||
) -> Result<bool>
|
||||
where
|
||||
B: AtomicBuffer,
|
||||
{
|
||||
@ -168,6 +181,10 @@ where
|
||||
let required = bit::align(record_len as usize, record_descriptor::ALIGNMENT as usize);
|
||||
let record_index = self.claim_capacity(required as IndexT)?;
|
||||
|
||||
if record_index == INSUFFICIENT_CAPACITY {
|
||||
return Ok(false);
|
||||
}
|
||||
|
||||
// UNWRAP: `claim_capacity` performed bounds checking
|
||||
self.buffer
|
||||
.put_i64_ordered(
|
||||
@ -189,7 +206,7 @@ where
|
||||
.put_i32_ordered(record_descriptor::length_offset(record_index), record_len)
|
||||
.unwrap();
|
||||
|
||||
Ok(())
|
||||
Ok(true)
|
||||
}
|
||||
|
||||
/// Read messages from the ring buffer and dispatch to `handler`, up to `message_count_limit`
|
||||
@ -266,7 +283,7 @@ where
|
||||
fn claim_capacity(&mut self, required: IndexT) -> Result<IndexT> {
|
||||
// QUESTION: Is this mask how we handle the "ring" in ring buffer?
|
||||
// Would explain why we assert buffer capacity is a power of two during initialization
|
||||
let mask = self.capacity - 1;
|
||||
let mask: IndexT = self.capacity - 1;
|
||||
|
||||
// UNWRAP: Known-valid offset calculated during initialization
|
||||
let mut head = self
|
||||
@ -279,28 +296,18 @@ where
|
||||
let mut padding: IndexT;
|
||||
// Note the braces, making this a do-while loop
|
||||
while {
|
||||
// UNWRAP: Known-valid offset calculated during initialization
|
||||
tail = self
|
||||
.buffer
|
||||
.get_i64_volatile(self.tail_position_index)
|
||||
.unwrap();
|
||||
tail = self.buffer.get_i64_volatile(self.tail_position_index)?;
|
||||
let available_capacity = self.capacity - (tail - head) as IndexT;
|
||||
|
||||
if required > available_capacity {
|
||||
// UNWRAP: Known-valid offset calculated during initialization
|
||||
head = self
|
||||
.buffer
|
||||
.get_i64_volatile(self.head_position_index)
|
||||
.unwrap();
|
||||
head = self.buffer.get_i64_volatile(self.head_position_index)?;
|
||||
|
||||
if required > (self.capacity - (tail - head) as IndexT) {
|
||||
return Err(AeronError::InsufficientCapacity);
|
||||
return Ok(INSUFFICIENT_CAPACITY);
|
||||
}
|
||||
|
||||
// UNWRAP: Known-valid offset calculated during initialization
|
||||
self.buffer
|
||||
.put_i64_ordered(self.head_cache_position_index, head)
|
||||
.unwrap();
|
||||
.put_i64_ordered(self.head_cache_position_index, head)?;
|
||||
}
|
||||
|
||||
padding = 0;
|
||||
@ -315,45 +322,32 @@ where
|
||||
let mut head_index = (head & i64::from(mask)) as IndexT;
|
||||
|
||||
if required > head_index {
|
||||
// UNWRAP: Known-valid offset calculated during initialization
|
||||
head = self
|
||||
.buffer
|
||||
.get_i64_volatile(self.head_position_index)
|
||||
.unwrap();
|
||||
head = self.buffer.get_i64_volatile(self.head_position_index)?;
|
||||
head_index = (head & i64::from(mask)) as IndexT;
|
||||
|
||||
if required > head_index {
|
||||
return Err(AeronError::InsufficientCapacity);
|
||||
return Ok(INSUFFICIENT_CAPACITY);
|
||||
}
|
||||
|
||||
// UNWRAP: Known-valid offset calculated during initialization
|
||||
self.buffer
|
||||
.put_i64_ordered(self.head_cache_position_index, head)
|
||||
.unwrap();
|
||||
.put_i64_ordered(self.head_cache_position_index, head)?;
|
||||
}
|
||||
|
||||
padding = to_buffer_end_length;
|
||||
}
|
||||
|
||||
// UNWRAP: Known-valid offset calculated during initialization
|
||||
!self
|
||||
.buffer
|
||||
.compare_and_set_i64(
|
||||
self.tail_position_index,
|
||||
tail,
|
||||
tail + i64::from(required) + i64::from(padding),
|
||||
)
|
||||
.unwrap()
|
||||
!self.buffer.compare_and_set_i64(
|
||||
self.tail_position_index,
|
||||
tail,
|
||||
tail + i64::from(required) + i64::from(padding),
|
||||
)?
|
||||
} {}
|
||||
|
||||
if padding != 0 {
|
||||
// UNWRAP: Known-valid offset calculated during initialization
|
||||
self.buffer
|
||||
.put_i64_ordered(
|
||||
tail_index,
|
||||
record_descriptor::make_header(padding, record_descriptor::PADDING_MSG_TYPE_ID),
|
||||
)
|
||||
.unwrap();
|
||||
self.buffer.put_i64_ordered(
|
||||
tail_index,
|
||||
record_descriptor::make_header(padding, record_descriptor::PADDING_MSG_TYPE_ID),
|
||||
)?;
|
||||
tail_index = 0;
|
||||
}
|
||||
|
||||
@ -367,6 +361,11 @@ where
|
||||
Ok(())
|
||||
}
|
||||
}
|
||||
|
||||
/// Return the largest possible message size for this buffer
|
||||
pub fn max_msg_length(&self) -> IndexT {
|
||||
self.max_msg_length
|
||||
}
|
||||
}
|
||||
|
||||
impl<A> Deref for ManyToOneRingBuffer<A>
|
||||
@ -380,6 +379,15 @@ where
|
||||
}
|
||||
}
|
||||
|
||||
impl<A> DerefMut for ManyToOneRingBuffer<A>
|
||||
where
|
||||
A: AtomicBuffer,
|
||||
{
|
||||
fn deref_mut(&mut self) -> &mut Self::Target {
|
||||
&mut self.buffer
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use crate::client::concurrent::ringbuffer::{record_descriptor, ManyToOneRingBuffer};
|
||||
@ -453,7 +461,7 @@ mod tests {
|
||||
let mut ring_buffer =
|
||||
ManyToOneRingBuffer::new(vec![0u8; BUFFER_SIZE]).expect("Invalid buffer size");
|
||||
|
||||
let mut source_buffer = &mut [12u8, 0, 0, 0, 0, 0, 0, 0][..];
|
||||
let source_buffer = &mut [12u8, 0, 0, 0, 0, 0, 0, 0][..];
|
||||
let source_len = source_buffer.len() as IndexT;
|
||||
let type_id = 1;
|
||||
ring_buffer
|
||||
@ -475,7 +483,7 @@ mod tests {
|
||||
let mut ring_buffer =
|
||||
ManyToOneRingBuffer::new(vec![0u8; BUFFER_SIZE]).expect("Invalid buffer size");
|
||||
|
||||
let mut source_buffer = &mut [12u8, 0, 0, 0, 0, 0, 0, 0][..];
|
||||
let source_buffer = &mut [12u8, 0, 0, 0, 0, 0, 0, 0][..];
|
||||
let source_len = source_buffer.len() as IndexT;
|
||||
let type_id = 1;
|
||||
ring_buffer
|
||||
|
134
aeron-rs/tests/many_to_one_ring_buffer.rs
Normal file
134
aeron-rs/tests/many_to_one_ring_buffer.rs
Normal file
@ -0,0 +1,134 @@
|
||||
/// Tests based on the C++ tests included with Aeron
|
||||
use aeron_rs::client::concurrent::ringbuffer::{
|
||||
buffer_descriptor, record_descriptor, ManyToOneRingBuffer,
|
||||
};
|
||||
use aeron_rs::client::concurrent::AtomicBuffer;
|
||||
use aeron_rs::util::bit::align;
|
||||
use aeron_rs::util::IndexT;
|
||||
use std::ops::Deref;
|
||||
|
||||
const CAPACITY: usize = 1024;
|
||||
const BUFFER_SZ: usize = CAPACITY + buffer_descriptor::TRAILER_LENGTH as usize;
|
||||
const ODD_BUFFER_SZ: usize = (CAPACITY - 1) + buffer_descriptor::TRAILER_LENGTH as usize;
|
||||
|
||||
const MSG_TYPE_ID: i32 = 101;
|
||||
const HEAD_COUNTER_INDEX: IndexT = 1024 as IndexT + buffer_descriptor::HEAD_POSITION_OFFSET;
|
||||
const TAIL_COUNTER_INDEX: IndexT = 1024 as IndexT + buffer_descriptor::TAIL_POSITION_OFFSET;
|
||||
|
||||
#[test]
|
||||
fn should_calculate_capacity_for_buffer() {
|
||||
let buffer = ManyToOneRingBuffer::new(vec![0u8; BUFFER_SZ]).unwrap();
|
||||
|
||||
assert_eq!(AtomicBuffer::capacity(buffer.deref()), BUFFER_SZ as IndexT);
|
||||
assert_eq!(buffer.capacity(), CAPACITY as IndexT);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn should_throw_for_capacity_not_power_of_two() {
|
||||
let buffer = ManyToOneRingBuffer::new(vec![0u8; ODD_BUFFER_SZ]);
|
||||
|
||||
assert!(buffer.is_err());
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn should_throw_when_max_message_size_exceeded() {
|
||||
let mut buffer = ManyToOneRingBuffer::new(vec![0u8; BUFFER_SZ]).unwrap();
|
||||
|
||||
let bytes = vec![0u8; buffer.max_msg_length() as usize + 1];
|
||||
let write_res = buffer.write(MSG_TYPE_ID, &bytes, 0, bytes.len() as IndexT);
|
||||
|
||||
assert!(write_res.is_err());
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn should_write_to_empty_buffer() {
|
||||
let tail: IndexT = 0;
|
||||
let tail_index: IndexT = 0;
|
||||
let length: IndexT = 8;
|
||||
let record_length: IndexT = length + record_descriptor::HEADER_LENGTH;
|
||||
let src_index: IndexT = 0;
|
||||
let aligned_record_length: IndexT = align(
|
||||
record_length as usize,
|
||||
record_descriptor::ALIGNMENT as usize,
|
||||
) as IndexT;
|
||||
|
||||
let mut buffer = ManyToOneRingBuffer::new(vec![0u8; BUFFER_SZ]).unwrap();
|
||||
let src_bytes = vec![0u8; BUFFER_SZ];
|
||||
|
||||
assert!(buffer
|
||||
.write(MSG_TYPE_ID, &src_bytes, src_index, length)
|
||||
.unwrap());
|
||||
|
||||
assert_eq!(
|
||||
buffer.get_i32(record_descriptor::length_offset(tail_index)),
|
||||
Ok(record_length)
|
||||
);
|
||||
assert_eq!(
|
||||
buffer.get_i32(record_descriptor::type_offset(tail_index)),
|
||||
Ok(MSG_TYPE_ID)
|
||||
);
|
||||
assert_eq!(
|
||||
buffer.get_i64(TAIL_COUNTER_INDEX),
|
||||
Ok((tail + aligned_record_length) as i64)
|
||||
);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn should_reject_write_when_insufficient_space() {
|
||||
let length: IndexT = 100;
|
||||
let head: IndexT = 0;
|
||||
let tail: IndexT = head + (CAPACITY - align((length - record_descriptor::ALIGNMENT) as usize, record_descriptor::ALIGNMENT as usize)) as IndexT;
|
||||
let src_index: IndexT = 0;
|
||||
|
||||
let mut buffer = ManyToOneRingBuffer::new(vec![0u8; BUFFER_SZ]).unwrap();
|
||||
buffer.put_i64(HEAD_COUNTER_INDEX, head as i64).unwrap();
|
||||
buffer.put_i64(TAIL_COUNTER_INDEX, tail as i64).unwrap();
|
||||
|
||||
let src_bytes = vec![0u8; BUFFER_SZ];
|
||||
let write_res = buffer.write(MSG_TYPE_ID, &src_bytes, src_index, length);
|
||||
|
||||
assert_eq!(write_res, Ok(false));
|
||||
assert_eq!(buffer.get_i64(TAIL_COUNTER_INDEX), Ok(tail as i64));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn should_reject_write_when_buffer_full() {
|
||||
let length: IndexT = 8;
|
||||
let head: IndexT = 0;
|
||||
let tail: IndexT = head + CAPACITY as IndexT;
|
||||
let src_index: IndexT = 0;
|
||||
|
||||
let mut buffer = ManyToOneRingBuffer::new(vec![0u8; BUFFER_SZ]).unwrap();
|
||||
buffer.put_i64(HEAD_COUNTER_INDEX, head as i64).unwrap();
|
||||
buffer.put_i64(TAIL_COUNTER_INDEX, tail as i64).unwrap();
|
||||
|
||||
let src_bytes = vec![0u8; BUFFER_SZ];
|
||||
let write_res = buffer.write(MSG_TYPE_ID, &src_bytes, src_index, length);
|
||||
assert_eq!(write_res, Ok(false));
|
||||
assert_eq!(buffer.get_i64(TAIL_COUNTER_INDEX), Ok(tail as i64));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn should_insert_padding_record_plus_message_on_buffer_wrap() {
|
||||
let length: IndexT = 100;
|
||||
let record_length: IndexT = length + record_descriptor::HEADER_LENGTH;
|
||||
let aligned_record_length = align(record_length as usize, record_descriptor::ALIGNMENT as usize) as IndexT;
|
||||
let tail: IndexT = CAPACITY as IndexT - record_descriptor::ALIGNMENT;
|
||||
let head: IndexT = tail - (record_descriptor::ALIGNMENT * 4);
|
||||
let src_index: IndexT = 0;
|
||||
|
||||
let mut buffer = ManyToOneRingBuffer::new(vec![0u8; BUFFER_SZ]).unwrap();
|
||||
buffer.put_i64(HEAD_COUNTER_INDEX, head as i64).unwrap();
|
||||
buffer.put_i64(TAIL_COUNTER_INDEX, tail as i64).unwrap();
|
||||
|
||||
let src_bytes = vec![0u8; BUFFER_SZ];
|
||||
let write_res = buffer.write(MSG_TYPE_ID, &src_bytes, src_index, length);
|
||||
assert_eq!(write_res, Ok(true));
|
||||
|
||||
assert_eq!(buffer.get_i32(record_descriptor::type_offset(tail)), Ok(record_descriptor::PADDING_MSG_TYPE_ID));
|
||||
assert_eq!(buffer.get_i32(record_descriptor::length_offset(tail)), Ok(record_descriptor::ALIGNMENT));
|
||||
|
||||
assert_eq!(buffer.get_i32(record_descriptor::length_offset(0)), Ok(record_length));
|
||||
assert_eq!(buffer.get_i32(record_descriptor::type_offset(0)), Ok(MSG_TYPE_ID));
|
||||
assert_eq!(buffer.get_i64(TAIL_COUNTER_INDEX), Ok((tail + aligned_record_length + record_descriptor::ALIGNMENT) as i64));
|
||||
}
|
Loading…
Reference in New Issue
Block a user