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
//! A Framed flavor of BBQueue, useful for variable length packets
//!
//! This module allows for a `Framed` mode of operation,
//! where a size header is included in each grant, allowing for
//! "chunks" of data to be passed through a BBQueue, rather than
//! just a stream of bytes. This is convenient when receiving
//! packets of variable sizes.
//!
//! ## Example
//!
//! ```rust
//! # // bbqueue test shim!
//! # fn bbqtest() {
//! use bbqueue::BBBuffer;
//!
//! let bb: BBBuffer<1000> = BBBuffer::new();
//! let (mut prod, mut cons) = bb.try_split_framed().unwrap();
//!
//! // One frame in, one frame out
//! let mut wgrant = prod.grant(128).unwrap();
//! assert_eq!(wgrant.len(), 128);
//! for (idx, i) in wgrant.iter_mut().enumerate() {
//! *i = idx as u8;
//! }
//! wgrant.commit(128);
//!
//! let rgrant = cons.read().unwrap();
//! assert_eq!(rgrant.len(), 128);
//! for (idx, i) in rgrant.iter().enumerate() {
//! assert_eq!(*i, idx as u8);
//! }
//! rgrant.release();
//! # // bbqueue test shim!
//! # }
//! #
//! # fn main() {
//! # #[cfg(not(feature = "thumbv6"))]
//! # bbqtest();
//! # }
//! ```
//!
//! ## Frame header
//!
//! An internal header is required for each frame stored
//! inside of the `BBQueue`. This header is never exposed to end
//! users of the bbqueue library.
//!
//! A variable sized integer is used for the header size, and the
//! size of this header is based on the max size requested for the grant.
//! This header size must be factored in when calculating an appropriate
//! total size of your buffer.
//!
//! Even if a smaller portion of the grant is committed, the original
//! requested grant size will be used for the header size calculation.
//!
//! For example, if you request a 128 byte grant, the header size will
//! be two bytes. If you then only commit 32 bytes, two bytes will still
//! be used for the header of that grant.
//!
//! | Grant Size (bytes) | Header size (bytes) |
//! | :--- | :--- |
//! | 1..(2^7) | 1 |
//! | (2^7)..(2^14) | 2 |
//! | (2^14)..(2^21) | 3 |
//! | (2^21)..(2^28) | 4 |
//! | (2^28)..(2^35) | 5 |
//! | (2^35)..(2^42) | 6 |
//! | (2^42)..(2^49) | 7 |
//! | (2^49)..(2^56) | 8 |
//! | (2^56)..(2^64) | 9 |
//!
use crate::{Consumer, GrantR, GrantW, Producer};
use crate::{
vusize::{decode_usize, decoded_len, encode_usize_to_slice, encoded_len},
Result,
};
use core::{
cmp::min,
ops::{Deref, DerefMut},
};
/// A producer of Framed data
pub struct FrameProducer<'a, const N: usize> {
pub(crate) producer: Producer<'a, N>,
}
impl<'a, const N: usize> FrameProducer<'a, N> {
/// Receive a grant for a frame with a maximum size of `max_sz` in bytes.
///
/// This size does not include the size of the frame header. The exact size
/// of the frame can be set on `commit`.
pub fn grant(&mut self, max_sz: usize) -> Result<FrameGrantW<'a, N>> {
let hdr_len = encoded_len(max_sz);
Ok(FrameGrantW {
grant_w: self.producer.grant_exact(max_sz + hdr_len)?,
hdr_len: hdr_len as u8,
})
}
}
/// A consumer of Framed data
pub struct FrameConsumer<'a, const N: usize> {
pub(crate) consumer: Consumer<'a, N>,
}
impl<'a, const N: usize> FrameConsumer<'a, N> {
/// Obtain the next available frame, if any
pub fn read(&mut self) -> Option<FrameGrantR<'a, N>> {
// Get all available bytes. We never wrap a frame around,
// so if a header is available, the whole frame will be.
let mut grant_r = self.consumer.read().ok()?;
// Additionally, we never commit less than a full frame with
// a header, so if we have ANY data, we'll have a full header
// and frame. `Consumer::read` will return an Error when
// there are 0 bytes available.
// The header consists of a single usize, encoded in native
// endianess order
let frame_len = decode_usize(&grant_r);
let hdr_len = decoded_len(grant_r[0]);
let total_len = frame_len + hdr_len;
let hdr_len = hdr_len as u8;
debug_assert!(grant_r.len() >= total_len);
// Reduce the grant down to the size of the frame with a header
grant_r.shrink(total_len);
Some(FrameGrantR { grant_r, hdr_len })
}
}
/// A write grant for a single frame
///
/// NOTE: If the grant is dropped without explicitly commiting
/// the contents without first calling `to_commit()`, then no
/// frame will be comitted for writing.
#[derive(Debug, PartialEq)]
pub struct FrameGrantW<'a, const N: usize> {
grant_w: GrantW<'a, N>,
hdr_len: u8,
}
/// A read grant for a single frame
///
/// NOTE: If the grant is dropped without explicitly releasing
/// the contents, then no frame will be released.
#[derive(Debug, PartialEq)]
pub struct FrameGrantR<'a, const N: usize> {
grant_r: GrantR<'a, N>,
hdr_len: u8,
}
impl<'a, const N: usize> Deref for FrameGrantW<'a, N> {
type Target = [u8];
fn deref(&self) -> &Self::Target {
&self.grant_w.buf[self.hdr_len.into()..]
}
}
impl<'a, const N: usize> DerefMut for FrameGrantW<'a, N> {
fn deref_mut(&mut self) -> &mut [u8] {
&mut self.grant_w.buf[self.hdr_len.into()..]
}
}
impl<'a, const N: usize> Deref for FrameGrantR<'a, N> {
type Target = [u8];
fn deref(&self) -> &Self::Target {
&self.grant_r.buf[self.hdr_len.into()..]
}
}
impl<'a, const N: usize> DerefMut for FrameGrantR<'a, N> {
fn deref_mut(&mut self) -> &mut [u8] {
&mut self.grant_r.buf[self.hdr_len.into()..]
}
}
impl<'a, const N: usize> FrameGrantW<'a, N> {
/// Commit a frame to make it available to the Consumer half.
///
/// `used` is the size of the payload, in bytes, not
/// including the frame header
pub fn commit(mut self, used: usize) {
let total_len = self.set_header(used);
// Commit the header + frame
self.grant_w.commit(total_len);
}
/// Set the header and return the total size
fn set_header(&mut self, used: usize) -> usize {
// Saturate the commit size to the available frame size
let grant_len = self.grant_w.len();
let hdr_len: usize = self.hdr_len.into();
let frame_len = min(used, grant_len - hdr_len);
let total_len = frame_len + hdr_len;
// Write the actual frame length to the header
encode_usize_to_slice(frame_len, hdr_len, &mut self.grant_w[..hdr_len]);
total_len
}
/// Configures the amount of bytes to be commited on drop.
pub fn to_commit(&mut self, amt: usize) {
if amt == 0 {
self.grant_w.to_commit(0);
} else {
let size = self.set_header(amt);
self.grant_w.to_commit(size);
}
}
}
impl<'a, const N: usize> FrameGrantR<'a, N> {
/// Release a frame to make the space available for future writing
///
/// Note: The full frame is always released
pub fn release(mut self) {
// For a read grant, we have already shrunk the grant
// size down to the correct size
let len = self.grant_r.len();
self.grant_r.release_inner(len);
}
/// Set whether the read fram should be automatically released
pub fn auto_release(&mut self, is_auto: bool) {
self.grant_r
.to_release(if is_auto { self.grant_r.len() } else { 0 });
}
}