serai/coordinator/tributary/src/lib.rs

423 lines
13 KiB
Rust
Raw Normal View History

use core::{marker::PhantomData, fmt::Debug};
use std::{sync::Arc, io};
use async_trait::async_trait;
use zeroize::Zeroizing;
use ciphersuite::{Ciphersuite, Ristretto};
use scale::Decode;
use futures::{StreamExt, SinkExt, channel::mpsc::UnboundedReceiver};
use ::tendermint::{
ext::{BlockNumber, Commit, Block as BlockTrait, Network},
SignedMessageFor, SyncedBlock, SyncedBlockSender, SyncedBlockResultReceiver, MessageSender,
TendermintMachine, TendermintHandle,
};
pub use ::tendermint::Evidence;
use serai_db::Db;
use tokio::sync::RwLock;
mod merkle;
pub(crate) use merkle::*;
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
pub mod transaction;
pub use transaction::{TransactionError, Signed, TransactionKind, Transaction as TransactionTrait};
use crate::tendermint::tx::TendermintTx;
2023-04-12 00:24:27 +00:00
mod provided;
pub(crate) use provided::*;
pub use provided::ProvidedError;
2023-04-12 00:24:27 +00:00
mod block;
pub use block::*;
2023-04-12 15:13:48 +00:00
mod blockchain;
pub(crate) use blockchain::*;
2023-04-12 15:13:48 +00:00
2023-04-12 16:15:38 +00:00
mod mempool;
pub(crate) use mempool::*;
2023-04-12 16:15:38 +00:00
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
pub mod tendermint;
pub(crate) use crate::tendermint::*;
#[cfg(any(test, feature = "tests"))]
pub mod tests;
/// Size limit for an individual transaction.
Support multiple key shares per validator (#416) * Update the coordinator to give key shares based on weight, not based on existence Participants are now identified by their starting index. While this compiles, the following is unimplemented: 1) A conversion for DKG `i` values. It assumes the threshold `i` values used will be identical for the MuSig signature used to confirm the DKG. 2) Expansion from compressed values to full values before forwarding to the processor. * Add a fn to the DkgConfirmer to convert `i` values as needed Also removes TODOs regarding Serai ensuring validator key uniqueness + validity. The current infra achieves both. * Have the Tributary DB track participation by shares, not by count * Prevent a node from obtaining 34% of the maximum amount of key shares This is actually mainly intended to set a bound on message sizes in the coordinator. Message sizes are amplified by the amount of key shares held, so setting an upper bound on said amount lets it determine constants. While that upper bound could be 150, that'd be unreasonable and increase the potential for DoS attacks. * Correct the mechanism to detect if sufficient accumulation has occured It used to check if the latest accumulation hit the required threshold. Now, accumulations may jump past the required threshold. The required mechanism is to check the threshold wasn't prior met and is now met. * Finish updating the coordinator to handle a multiple key share per validator environment * Adjust stategy re: preventing noce reuse in DKG Confirmer * Add TODOs regarding dropped transactions, add possible TODO fix * Update tests/coordinator This doesn't add new multi-key-share tests, it solely updates the existing single key-share tests to compile and run, with the necessary fixes to the coordinator. * Update processor key_gen to handle generating multiple key shares at once * Update SubstrateSigner * Update signer, clippy * Update processor tests * Update processor docker tests
2023-11-04 23:26:13 +00:00
pub const TRANSACTION_SIZE_LIMIT: usize = 3_000_000;
/// Amount of transactions a single account may have in the mempool.
pub const ACCOUNT_MEMPOOL_LIMIT: u32 = 50;
/// Block size limit.
Support multiple key shares per validator (#416) * Update the coordinator to give key shares based on weight, not based on existence Participants are now identified by their starting index. While this compiles, the following is unimplemented: 1) A conversion for DKG `i` values. It assumes the threshold `i` values used will be identical for the MuSig signature used to confirm the DKG. 2) Expansion from compressed values to full values before forwarding to the processor. * Add a fn to the DkgConfirmer to convert `i` values as needed Also removes TODOs regarding Serai ensuring validator key uniqueness + validity. The current infra achieves both. * Have the Tributary DB track participation by shares, not by count * Prevent a node from obtaining 34% of the maximum amount of key shares This is actually mainly intended to set a bound on message sizes in the coordinator. Message sizes are amplified by the amount of key shares held, so setting an upper bound on said amount lets it determine constants. While that upper bound could be 150, that'd be unreasonable and increase the potential for DoS attacks. * Correct the mechanism to detect if sufficient accumulation has occured It used to check if the latest accumulation hit the required threshold. Now, accumulations may jump past the required threshold. The required mechanism is to check the threshold wasn't prior met and is now met. * Finish updating the coordinator to handle a multiple key share per validator environment * Adjust stategy re: preventing noce reuse in DKG Confirmer * Add TODOs regarding dropped transactions, add possible TODO fix * Update tests/coordinator This doesn't add new multi-key-share tests, it solely updates the existing single key-share tests to compile and run, with the necessary fixes to the coordinator. * Update processor key_gen to handle generating multiple key shares at once * Update SubstrateSigner * Update signer, clippy * Update processor tests * Update processor docker tests
2023-11-04 23:26:13 +00:00
// This targets a growth limit of roughly 45 GB a day, under load, in order to prevent a malicious
// participant from flooding disks and causing out of space errors in order processes.
Support multiple key shares per validator (#416) * Update the coordinator to give key shares based on weight, not based on existence Participants are now identified by their starting index. While this compiles, the following is unimplemented: 1) A conversion for DKG `i` values. It assumes the threshold `i` values used will be identical for the MuSig signature used to confirm the DKG. 2) Expansion from compressed values to full values before forwarding to the processor. * Add a fn to the DkgConfirmer to convert `i` values as needed Also removes TODOs regarding Serai ensuring validator key uniqueness + validity. The current infra achieves both. * Have the Tributary DB track participation by shares, not by count * Prevent a node from obtaining 34% of the maximum amount of key shares This is actually mainly intended to set a bound on message sizes in the coordinator. Message sizes are amplified by the amount of key shares held, so setting an upper bound on said amount lets it determine constants. While that upper bound could be 150, that'd be unreasonable and increase the potential for DoS attacks. * Correct the mechanism to detect if sufficient accumulation has occured It used to check if the latest accumulation hit the required threshold. Now, accumulations may jump past the required threshold. The required mechanism is to check the threshold wasn't prior met and is now met. * Finish updating the coordinator to handle a multiple key share per validator environment * Adjust stategy re: preventing noce reuse in DKG Confirmer * Add TODOs regarding dropped transactions, add possible TODO fix * Update tests/coordinator This doesn't add new multi-key-share tests, it solely updates the existing single key-share tests to compile and run, with the necessary fixes to the coordinator. * Update processor key_gen to handle generating multiple key shares at once * Update SubstrateSigner * Update signer, clippy * Update processor tests * Update processor docker tests
2023-11-04 23:26:13 +00:00
pub const BLOCK_SIZE_LIMIT: usize = 3_001_000;
pub(crate) const TENDERMINT_MESSAGE: u8 = 0;
pub(crate) const BLOCK_MESSAGE: u8 = 1;
pub(crate) const TRANSACTION_MESSAGE: u8 = 2;
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
#[allow(clippy::large_enum_variant)]
#[derive(Clone, PartialEq, Eq, Debug)]
pub enum Transaction<T: TransactionTrait> {
Tendermint(TendermintTx),
Application(T),
}
impl<T: TransactionTrait> ReadWrite for Transaction<T> {
fn read<R: io::Read>(reader: &mut R) -> io::Result<Self> {
let mut kind = [0];
reader.read_exact(&mut kind)?;
match kind[0] {
0 => {
let tx = TendermintTx::read(reader)?;
Ok(Transaction::Tendermint(tx))
}
1 => {
let tx = T::read(reader)?;
Ok(Transaction::Application(tx))
}
_ => Err(io::Error::new(io::ErrorKind::Other, "invalid transaction type")),
}
}
fn write<W: io::Write>(&self, writer: &mut W) -> io::Result<()> {
match self {
Transaction::Tendermint(tx) => {
writer.write_all(&[0])?;
tx.write(writer)
}
Transaction::Application(tx) => {
writer.write_all(&[1])?;
tx.write(writer)
}
}
}
}
impl<T: TransactionTrait> Transaction<T> {
pub fn hash(&self) -> [u8; 32] {
match self {
Transaction::Tendermint(tx) => tx.hash(),
Transaction::Application(tx) => tx.hash(),
}
}
pub fn kind(&self) -> TransactionKind<'_> {
match self {
Transaction::Tendermint(tx) => tx.kind(),
Transaction::Application(tx) => tx.kind(),
}
}
}
/// An item which can be read and written.
pub trait ReadWrite: Sized {
fn read<R: io::Read>(reader: &mut R) -> io::Result<Self>;
fn write<W: io::Write>(&self, writer: &mut W) -> io::Result<()>;
fn serialize(&self) -> Vec<u8> {
// BlockHeader is 64 bytes and likely the smallest item in this system
let mut buf = Vec::with_capacity(64);
self.write(&mut buf).unwrap();
buf
}
}
#[async_trait]
pub trait P2p: 'static + Send + Sync + Clone + Debug {
/// Broadcast a message to all other members of the Tributary with the specified genesis.
///
/// The Tributary will re-broadcast consensus messages on a fixed interval to ensure they aren't
/// prematurely dropped from the P2P layer. THe P2P layer SHOULD perform content-based
/// deduplication to ensure a sane amount of load.
async fn broadcast(&self, genesis: [u8; 32], msg: Vec<u8>);
}
#[async_trait]
impl<P: P2p> P2p for Arc<P> {
async fn broadcast(&self, genesis: [u8; 32], msg: Vec<u8>) {
(*self).broadcast(genesis, msg).await
}
}
#[derive(Clone)]
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
pub struct Tributary<D: Db, T: TransactionTrait, P: P2p> {
db: D,
genesis: [u8; 32],
network: TendermintNetwork<D, T, P>,
synced_block: Arc<RwLock<SyncedBlockSender<TendermintNetwork<D, T, P>>>>,
synced_block_result: Arc<RwLock<SyncedBlockResultReceiver>>,
messages: Arc<RwLock<MessageSender<TendermintNetwork<D, T, P>>>>,
p2p_meta_task_handle: Arc<tokio::task::AbortHandle>,
}
impl<D: Db, T: TransactionTrait, P: P2p> Drop for Tributary<D, T, P> {
fn drop(&mut self) {
self.p2p_meta_task_handle.abort();
}
}
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
impl<D: Db, T: TransactionTrait, P: P2p> Tributary<D, T, P> {
pub async fn new(
db: D,
genesis: [u8; 32],
start_time: u64,
key: Zeroizing<<Ristretto as Ciphersuite>::F>,
validators: Vec<(<Ristretto as Ciphersuite>::G, u64)>,
p2p: P,
) -> Option<Self> {
log::info!("new Tributary with genesis {}", hex::encode(genesis));
let validators_vec = validators.iter().map(|validator| validator.0).collect::<Vec<_>>();
let signer = Arc::new(Signer::new(genesis, key));
let validators = Arc::new(Validators::new(genesis, validators)?);
let mut blockchain = Blockchain::new(db.clone(), genesis, &validators_vec);
let block_number = BlockNumber(blockchain.block_number().into());
let start_time = if let Some(commit) = blockchain.commit(&blockchain.tip()) {
Commit::<Validators>::decode(&mut commit.as_ref()).unwrap().end_time
} else {
start_time
};
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
let proposal = TendermintBlock(
blockchain.build_block::<TendermintNetwork<D, T, P>>(validators.clone()).serialize(),
);
let blockchain = Arc::new(RwLock::new(blockchain));
let to_rebroadcast = Arc::new(RwLock::new(vec![]));
// Actively rebroadcast consensus messages to ensure they aren't prematurely dropped from the
// P2P layer
let p2p_meta_task_handle = Arc::new(
tokio::spawn({
let to_rebroadcast = to_rebroadcast.clone();
let p2p = p2p.clone();
async move {
loop {
let to_rebroadcast = to_rebroadcast.read().await.clone();
for msg in to_rebroadcast {
p2p.broadcast(genesis, msg).await;
}
tokio::time::sleep(core::time::Duration::from_secs(1)).await;
}
}
})
.abort_handle(),
);
let network =
TendermintNetwork { genesis, signer, validators, blockchain, to_rebroadcast, p2p };
let TendermintHandle { synced_block, synced_block_result, messages, machine } =
TendermintMachine::new(network.clone(), block_number, start_time, proposal).await;
tokio::spawn(machine.run());
Some(Self {
db,
genesis,
network,
synced_block: Arc::new(RwLock::new(synced_block)),
synced_block_result: Arc::new(RwLock::new(synced_block_result)),
messages: Arc::new(RwLock::new(messages)),
p2p_meta_task_handle,
})
}
pub fn block_time() -> u32 {
TendermintNetwork::<D, T, P>::block_time()
}
2023-04-20 09:05:17 +00:00
pub fn genesis(&self) -> [u8; 32] {
self.genesis
2023-04-20 09:05:17 +00:00
}
pub async fn block_number(&self) -> u32 {
self.network.blockchain.read().await.block_number()
}
pub async fn tip(&self) -> [u8; 32] {
self.network.blockchain.read().await.tip()
2023-04-15 04:41:48 +00:00
}
pub fn reader(&self) -> TributaryReader<D, T> {
TributaryReader(self.db.clone(), self.genesis, PhantomData)
}
2023-04-15 04:41:48 +00:00
pub async fn provide_transaction(&self, tx: T) -> Result<(), ProvidedError> {
self.network.blockchain.write().await.provide_transaction(tx)
}
pub async fn next_nonce(&self, signer: <Ristretto as Ciphersuite>::G) -> Option<u32> {
self.network.blockchain.read().await.next_nonce(signer)
}
// Returns Ok(true) if new, Ok(false) if an already present unsigned, or the error.
// Safe to be &self since the only meaningful usage of self is self.network.blockchain which
// successfully acquires its own write lock
pub async fn add_transaction(&self, tx: T) -> Result<bool, TransactionError> {
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
let tx = Transaction::Application(tx);
let mut to_broadcast = vec![TRANSACTION_MESSAGE];
tx.write(&mut to_broadcast).unwrap();
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
let res = self.network.blockchain.write().await.add_transaction::<TendermintNetwork<D, T, P>>(
true,
tx,
self.network.signature_scheme(),
);
if res == Ok(true) {
self.network.p2p.broadcast(self.genesis, to_broadcast).await;
}
res
}
async fn sync_block_internal(
&self,
block: Block<T>,
commit: Vec<u8>,
result: &mut UnboundedReceiver<bool>,
) -> bool {
let (tip, block_number) = {
let blockchain = self.network.blockchain.read().await;
(blockchain.tip(), blockchain.block_number())
};
if block.header.parent != tip {
log::debug!("told to sync a block whose parent wasn't our tip");
return false;
}
let block = TendermintBlock(block.serialize());
2023-05-10 03:45:05 +00:00
let mut commit_ref = commit.as_ref();
let Ok(commit) = Commit::<Arc<Validators>>::decode(&mut commit_ref) else {
log::error!("sent an invalidly serialized commit");
return false;
};
2023-05-10 03:45:05 +00:00
// Storage DoS vector. We *could* truncate to solely the relevant portion, trying to save this,
// yet then we'd have to test the truncation was performed correctly.
if !commit_ref.is_empty() {
log::error!("sent an commit with additional data after it");
return false;
}
if !self.network.verify_commit(block.id(), &commit) {
log::error!("sent an invalid commit");
return false;
}
let number = BlockNumber((block_number + 1).into());
self.synced_block.write().await.send(SyncedBlock { number, block, commit }).await.unwrap();
result.next().await.unwrap()
}
// Sync a block.
// TODO: Since we have a static validator set, we should only need the tail commit?
pub async fn sync_block(&self, block: Block<T>, commit: Vec<u8>) -> bool {
let mut result = self.synced_block_result.write().await;
self.sync_block_internal(block, commit, &mut result).await
}
// Return true if the message should be rebroadcasted.
pub async fn handle_message(&self, msg: &[u8]) -> bool {
// Acquire the lock now to prevent sync_block from being run at the same time
let mut sync_block = self.synced_block_result.write().await;
match msg.first() {
Some(&TRANSACTION_MESSAGE) => {
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
let Ok(tx) = Transaction::read::<&[u8]>(&mut &msg[1 ..]) else {
log::error!("received invalid transaction message");
return false;
};
// TODO: Sync mempools with fellow peers
// Can we just rebroadcast transactions not included for at least two blocks?
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
let res =
self.network.blockchain.write().await.add_transaction::<TendermintNetwork<D, T, P>>(
false,
tx,
self.network.signature_scheme(),
);
log::debug!("received transaction message. valid new transaction: {res:?}");
res == Ok(true)
}
Some(&TENDERMINT_MESSAGE) => {
2023-08-01 04:47:36 +00:00
let Ok(msg) =
SignedMessageFor::<TendermintNetwork<D, T, P>>::decode::<&[u8]>(&mut &msg[1 ..])
else {
log::error!("received invalid tendermint message");
return false;
};
self.messages.write().await.send(msg).await.unwrap();
false
}
Some(&BLOCK_MESSAGE) => {
let mut msg_ref = &msg[1 ..];
let Ok(block) = Block::<T>::read(&mut msg_ref) else {
log::error!("received invalid block message");
return false;
};
let commit = msg[(msg.len() - msg_ref.len()) ..].to_vec();
if self.sync_block_internal(block, commit, &mut sync_block).await {
log::debug!("synced block over p2p net instead of building the commit ourselves");
}
false
}
_ => false,
}
}
/// Get a Future which will resolve once the next block has been added.
pub async fn next_block_notification(
&self,
) -> impl Send + Sync + core::future::Future<Output = Result<(), impl Send + Sync>> {
let (tx, rx) = tokio::sync::oneshot::channel();
self.network.blockchain.write().await.next_block_notifications.push_back(tx);
rx
}
}
#[derive(Clone)]
Slash malevolent validators (#294) * add slash tx * ignore unsigned tx replays * verify that provided evidence is valid * fix clippy + fmt * move application tx handling to another module * partially handle the tendermint txs * fix pr comments * support unsigned app txs * add slash target to the votes * enforce provided, unsigned, signed tx ordering within a block * bug fixes * add unit test for tendermint txs * bug fixes * update tests for tendermint txs * add tx ordering test * tidy up tx ordering test * cargo +nightly fmt * Misc fixes from rebasing * Finish resolving clippy * Remove sha3 from tendermint-machine * Resolve a DoS in SlashEvidence's read Also moves Evidence from Vec<Message> to (Message, Option<Message>). That should meet all requirements while being a bit safer. * Make lazy_static a dev-depend for tributary * Various small tweaks One use of sort was inefficient, sorting unsigned || signed when unsigned was already properly sorted. Given how the unsigned TXs were given a nonce of 0, an unstable sort may swap places with an unsigned TX and a signed TX with a nonce of 0 (leading to a faulty block). The extra protection added here sorts signed, then concats. * Fix Tributary tests I broke, start review on tendermint/tx.rs * Finish reviewing everything outside tests and empty_signature * Remove empty_signature empty_signature led to corrupted local state histories. Unfortunately, the API is only sane with a signature. We now use the actual signature, which risks creating a signature over a malicious message if we have ever have an invariant producing malicious messages. Prior, we only signed the message after the local machine confirmed it was okay per the local view of consensus. This is tolerated/preferred over a corrupt state history since production of such messages is already an invariant. TODOs are added to make handling of this theoretical invariant further robust. * Remove async_sequential for tokio::test There was no competition for resources forcing them to be run sequentially. * Modify block order test to be statistically significant without multiple runs * Clean tests --------- Co-authored-by: Luke Parker <lukeparker5132@gmail.com>
2023-08-21 04:28:23 +00:00
pub struct TributaryReader<D: Db, T: TransactionTrait>(D, [u8; 32], PhantomData<T>);
impl<D: Db, T: TransactionTrait> TributaryReader<D, T> {
pub fn genesis(&self) -> [u8; 32] {
self.1
}
// Since these values are static once set, they can be safely read from the database without lock
// acquisition
pub fn block(&self, hash: &[u8; 32]) -> Option<Block<T>> {
Blockchain::<D, T>::block_from_db(&self.0, self.1, hash)
}
pub fn commit(&self, hash: &[u8; 32]) -> Option<Vec<u8>> {
Blockchain::<D, T>::commit_from_db(&self.0, self.1, hash)
}
pub fn parsed_commit(&self, hash: &[u8; 32]) -> Option<Commit<Validators>> {
self.commit(hash).map(|commit| Commit::<Validators>::decode(&mut commit.as_ref()).unwrap())
}
pub fn block_after(&self, hash: &[u8; 32]) -> Option<[u8; 32]> {
Blockchain::<D, T>::block_after(&self.0, self.1, hash)
}
pub fn time_of_block(&self, hash: &[u8; 32]) -> Option<u64> {
self
.commit(hash)
.map(|commit| Commit::<Validators>::decode(&mut commit.as_ref()).unwrap().end_time)
}
pub fn locally_provided_txs_in_block(&self, hash: &[u8; 32], order: &str) -> bool {
Blockchain::<D, T>::locally_provided_txs_in_block(&self.0, &self.1, hash, order)
}
// This isn't static, yet can be read with only minor discrepancy risks
pub fn tip(&self) -> [u8; 32] {
Blockchain::<D, T>::tip_from_db(&self.0, self.1)
}
}