Merge pull request #297 from pawanjay176/optimized_ghost

[WIP] Add OptimizedLMDGhost fork choice rule and tests

eth2/fork_choice/src/lib.rs

@@ -22,6 +22,7 @@ extern crate types;
 
 pub mod bitwise_lmd_ghost;
 pub mod longest_chain;
+pub mod optimized_lmd_ghost;
 pub mod slow_lmd_ghost;
 
 use db::stores::BeaconBlockAtSlotError;
@@ -30,6 +31,7 @@ use types::{BeaconBlock, ChainSpec, Hash256};
 
 pub use bitwise_lmd_ghost::BitwiseLMDGhost;
 pub use longest_chain::LongestChain;
+pub use optimized_lmd_ghost::OptimizedLMDGhost;
 pub use slow_lmd_ghost::SlowLMDGhost;
 
 /// Defines the interface for Fork Choices. Each Fork choice will define their own data structures
@@ -101,4 +103,6 @@ pub enum ForkChoiceAlgorithm {
     SlowLMDGhost,
     /// An optimised version of bitwise LMD-GHOST by Vitalik.
     BitwiseLMDGhost,
+    /// An optimised implementation of LMD ghost.
+    OptimizedLMDGhost,
 }

eth2/fork_choice/src/optimized_lmd_ghost.rs

@@ -0,0 +1,453 @@
+//! The optimised bitwise LMD-GHOST fork choice rule.
+extern crate bit_vec;
+
+use crate::{ForkChoice, ForkChoiceError};
+use db::{
+    stores::{BeaconBlockStore, BeaconStateStore},
+    ClientDB,
+};
+use log::{debug, trace};
+use std::cmp::Ordering;
+use std::collections::HashMap;
+use std::sync::Arc;
+use types::{
+    readers::BeaconBlockReader, validator_registry::get_active_validator_indices, BeaconBlock,
+    ChainSpec, Hash256, Slot, SlotHeight,
+};
+
+//TODO: Pruning - Children
+//TODO: Handle Syncing
+
+// NOTE: This uses u32 to represent difference between block heights. Thus this is only
+// applicable for block height differences in the range of a u32.
+// This can potentially be parallelized in some parts.
+
+/// Compute the base-2 logarithm of an integer, floored (rounded down)
+#[inline]
+fn log2_int(x: u64) -> u32 {
+    if x == 0 {
+        return 0;
+    }
+    63 - x.leading_zeros()
+}
+
+fn power_of_2_below(x: u64) -> u64 {
+    2u64.pow(log2_int(x))
+}
+
+/// Stores the necessary data structures to run the optimised lmd ghost algorithm.
+pub struct OptimizedLMDGhost<T: ClientDB + Sized> {
+    /// A cache of known ancestors at given heights for a specific block.
+    //TODO: Consider FnvHashMap
+    cache: HashMap<CacheKey<u64>, Hash256>,
+    /// Log lookup table for blocks to their ancestors.
+    //TODO: Verify we only want/need a size 16 log lookup
+    ancestors: Vec<HashMap<Hash256, Hash256>>,
+    /// Stores the children for any given parent.
+    children: HashMap<Hash256, Vec<Hash256>>,
+    /// The latest attestation targets as a map of validator index to block hash.
+    //TODO: Could this be a fixed size vec
+    latest_attestation_targets: HashMap<u64, Hash256>,
+    /// Block storage access.
+    block_store: Arc<BeaconBlockStore<T>>,
+    /// State storage access.
+    state_store: Arc<BeaconStateStore<T>>,
+    max_known_height: SlotHeight,
+}
+
+impl<T> OptimizedLMDGhost<T>
+where
+    T: ClientDB + Sized,
+{
+    pub fn new(
+        block_store: Arc<BeaconBlockStore<T>>,
+        state_store: Arc<BeaconStateStore<T>>,
+    ) -> Self {
+        OptimizedLMDGhost {
+            cache: HashMap::new(),
+            ancestors: vec![HashMap::new(); 16],
+            latest_attestation_targets: HashMap::new(),
+            children: HashMap::new(),
+            max_known_height: SlotHeight::new(0),
+            block_store,
+            state_store,
+        }
+    }
+
+    /// Finds the latest votes weighted by validator balance. Returns a hashmap of block_hash to
+    /// weighted votes.
+    pub fn get_latest_votes(
+        &self,
+        state_root: &Hash256,
+        block_slot: Slot,
+        spec: &ChainSpec,
+    ) -> Result<HashMap<Hash256, u64>, ForkChoiceError> {
+        // get latest votes
+        // Note: Votes are weighted by min(balance, MAX_DEPOSIT_AMOUNT) //
+        // FORK_CHOICE_BALANCE_INCREMENT
+        // build a hashmap of block_hash to weighted votes
+        let mut latest_votes: HashMap<Hash256, u64> = HashMap::new();
+        // gets the current weighted votes
+        let current_state = self
+            .state_store
+            .get_deserialized(&state_root)?
+            .ok_or_else(|| ForkChoiceError::MissingBeaconState(*state_root))?;
+
+        let active_validator_indices = get_active_validator_indices(
+            &current_state.validator_registry[..],
+            block_slot.epoch(spec.slots_per_epoch),
+        );
+
+        for index in active_validator_indices {
+            let balance = std::cmp::min(
+                current_state.validator_balances[index],
+                spec.max_deposit_amount,
+            ) / spec.fork_choice_balance_increment;
+            if balance > 0 {
+                if let Some(target) = self.latest_attestation_targets.get(&(index as u64)) {
+                    *latest_votes.entry(*target).or_insert_with(|| 0) += balance;
+                }
+            }
+        }
+        trace!("Latest votes: {:?}", latest_votes);
+        Ok(latest_votes)
+    }
+
+    /// Gets the ancestor at a given height `at_height` of a block specified by `block_hash`.
+    fn get_ancestor(
+        &mut self,
+        block_hash: Hash256,
+        target_height: SlotHeight,
+        spec: &ChainSpec,
+    ) -> Option<Hash256> {
+        // return None if we can't get the block from the db.
+        let block_height = {
+            let block_slot = self
+                .block_store
+                .get_deserialized(&block_hash)
+                .ok()?
+                .expect("Should have returned already if None")
+                .slot;
+
+            block_slot.height(spec.genesis_slot)
+        };
+
+        // verify we haven't exceeded the block height
+        if target_height >= block_height {
+            if target_height > block_height {
+                return None;
+            } else {
+                return Some(block_hash);
+            }
+        }
+        // check if the result is stored in our cache
+        let cache_key = CacheKey::new(&block_hash, target_height.as_u64());
+        if let Some(ancestor) = self.cache.get(&cache_key) {
+            return Some(*ancestor);
+        }
+
+        // not in the cache recursively search for ancestors using a log-lookup
+        if let Some(ancestor) = {
+            let ancestor_lookup = self.ancestors
+                [log2_int((block_height - target_height - 1u64).as_u64()) as usize]
+                .get(&block_hash)
+                //TODO: Panic if we can't lookup and fork choice fails
+                .expect("All blocks should be added to the ancestor log lookup table");
+            self.get_ancestor(*ancestor_lookup, target_height, &spec)
+        } {
+            // add the result to the cache
+            self.cache.insert(cache_key, ancestor);
+            return Some(ancestor);
+        }
+
+        None
+    }
+
+    // looks for an obvious block winner given the latest votes for a specific height
+    fn get_clear_winner(
+        &mut self,
+        latest_votes: &HashMap<Hash256, u64>,
+        block_height: SlotHeight,
+        spec: &ChainSpec,
+    ) -> Option<Hash256> {
+        // map of vote counts for every hash at this height
+        let mut current_votes: HashMap<Hash256, u64> = HashMap::new();
+        let mut total_vote_count = 0;
+
+        trace!("Clear winner at block height: {}", block_height);
+        // loop through the latest votes and count all votes
+        // these have already been weighted by balance
+        for (hash, votes) in latest_votes.iter() {
+            if let Some(ancestor) = self.get_ancestor(*hash, block_height, spec) {
+                let current_vote_value = current_votes.get(&ancestor).unwrap_or_else(|| &0);
+                current_votes.insert(ancestor, current_vote_value + *votes);
+                total_vote_count += votes;
+            }
+        }
+        // Check if there is a clear block winner at this height. If so return it.
+        for (hash, votes) in current_votes.iter() {
+            if *votes > total_vote_count / 2 {
+                // we have a clear winner, return it
+                return Some(*hash);
+            }
+        }
+        // didn't find a clear winner
+        None
+    }
+
+    // Finds the best child (one with highest votes)
+    fn choose_best_child(&self, votes: &HashMap<Hash256, u64>) -> Option<Hash256> {
+        if votes.is_empty() {
+            return None;
+        }
+
+        // Iterate through hashmap to get child with maximum votes
+        let best_child = votes.iter().max_by(|(child1, v1), (child2, v2)| {
+            let mut result = v1.cmp(v2);
+            // If votes are equal, choose smaller hash to break ties deterministically
+            if result == Ordering::Equal {
+                // Reverse so that max_by chooses smaller hash
+                result = child1.cmp(child2).reverse();
+            }
+            result
+        });
+
+        Some(*best_child.unwrap().0)
+    }
+}
+
+impl<T: ClientDB + Sized> ForkChoice for OptimizedLMDGhost<T> {
+    fn add_block(
+        &mut self,
+        block: &BeaconBlock,
+        block_hash: &Hash256,
+        spec: &ChainSpec,
+    ) -> Result<(), ForkChoiceError> {
+        // get the height of the parent
+        let parent_height = self
+            .block_store
+            .get_deserialized(&block.parent_root)?
+            .ok_or_else(|| ForkChoiceError::MissingBeaconBlock(block.parent_root))?
+            .slot()
+            .height(spec.genesis_slot);
+
+        let parent_hash = &block.parent_root;
+
+        // add the new block to the children of parent
+        (*self
+            .children
+            .entry(block.parent_root)
+            .or_insert_with(|| vec![]))
+        .push(block_hash.clone());
+
+        // build the ancestor data structure
+        for index in 0..16 {
+            if parent_height % (1 << index) == 0 {
+                self.ancestors[index].insert(*block_hash, *parent_hash);
+            } else {
+                // TODO: This is unsafe. Will panic if parent_hash doesn't exist. Using it for debugging
+                let parent_ancestor = self.ancestors[index][parent_hash];
+                self.ancestors[index].insert(*block_hash, parent_ancestor);
+            }
+        }
+        // update the max height
+        self.max_known_height = std::cmp::max(self.max_known_height, parent_height + 1);
+        Ok(())
+    }
+
+    fn add_attestation(
+        &mut self,
+        validator_index: u64,
+        target_block_root: &Hash256,
+        spec: &ChainSpec,
+    ) -> Result<(), ForkChoiceError> {
+        // simply add the attestation to the latest_attestation_target if the block_height is
+        // larger
+        trace!(
+            "Adding attestation of validator: {:?} for block: {}",
+            validator_index,
+            target_block_root
+        );
+        let attestation_target = self
+            .latest_attestation_targets
+            .entry(validator_index)
+            .or_insert_with(|| *target_block_root);
+        // if we already have a value
+        if attestation_target != target_block_root {
+            trace!("Old attestation found: {:?}", attestation_target);
+            // get the height of the target block
+            let block_height = self
+                .block_store
+                .get_deserialized(&target_block_root)?
+                .ok_or_else(|| ForkChoiceError::MissingBeaconBlock(*target_block_root))?
+                .slot()
+                .height(spec.genesis_slot);
+
+            // get the height of the past target block
+            let past_block_height = self
+                .block_store
+                .get_deserialized(&attestation_target)?
+                .ok_or_else(|| ForkChoiceError::MissingBeaconBlock(*attestation_target))?
+                .slot()
+                .height(spec.genesis_slot);
+            // update the attestation only if the new target is higher
+            if past_block_height < block_height {
+                trace!("Updating old attestation");
+                *attestation_target = *target_block_root;
+            }
+        }
+        Ok(())
+    }
+
+    /// Perform lmd_ghost on the current chain to find the head.
+    fn find_head(
+        &mut self,
+        justified_block_start: &Hash256,
+        spec: &ChainSpec,
+    ) -> Result<Hash256, ForkChoiceError> {
+        debug!(
+            "Starting optimised fork choice at block: {}",
+            justified_block_start
+        );
+        let block = self
+            .block_store
+            .get_deserialized(&justified_block_start)?
+            .ok_or_else(|| ForkChoiceError::MissingBeaconBlock(*justified_block_start))?;
+
+        let block_slot = block.slot();
+        let state_root = block.state_root();
+        let mut block_height = block_slot.height(spec.genesis_slot);
+
+        let mut current_head = *justified_block_start;
+
+        let mut latest_votes = self.get_latest_votes(&state_root, block_slot, spec)?;
+
+        // remove any votes that don't relate to our current head.
+        latest_votes
+            .retain(|hash, _| self.get_ancestor(*hash, block_height, spec) == Some(current_head));
+
+        // begin searching for the head
+        loop {
+            debug!(
+                "Iteration for block: {} with vote length: {}",
+                current_head,
+                latest_votes.len()
+            );
+            // if there are no children, we are done, return the current_head
+            let children = match self.children.get(&current_head) {
+                Some(children) => children.clone(),
+                None => {
+                    debug!("Head found: {}", current_head);
+                    return Ok(current_head);
+                }
+            };
+
+            // logarithmic lookup blocks to see if there are obvious winners, if so,
+            // progress to the next iteration.
+            let mut step =
+                power_of_2_below(self.max_known_height.saturating_sub(block_height).as_u64()) / 2;
+            while step > 0 {
+                trace!("Current Step: {}", step);
+                if let Some(clear_winner) = self.get_clear_winner(
+                    &latest_votes,
+                    block_height - (block_height % step) + step,
+                    spec,
+                ) {
+                    current_head = clear_winner;
+                    break;
+                }
+                step /= 2;
+            }
+            if step > 0 {
+                trace!("Found clear winner: {}", current_head);
+            }
+            // if our skip lookup failed and we only have one child, progress to that child
+            else if children.len() == 1 {
+                current_head = children[0];
+                trace!(
+                    "Lookup failed, only one child, proceeding to child: {}",
+                    current_head
+                );
+            }
+            // we need to find the best child path to progress down.
+            else {
+                trace!("Searching for best child");
+                let mut child_votes = HashMap::new();
+                for (voted_hash, vote) in latest_votes.iter() {
+                    // if the latest votes correspond to a child
+                    if let Some(child) = self.get_ancestor(*voted_hash, block_height + 1, spec) {
+                        // add up the votes for each child
+                        *child_votes.entry(child).or_insert_with(|| 0) += vote;
+                    }
+                }
+                // given the votes on the children, find the best child
+                current_head = self
+                    .choose_best_child(&child_votes)
+                    .ok_or(ForkChoiceError::CannotFindBestChild)?;
+                trace!("Best child found: {}", current_head);
+            }
+
+            // didn't find head yet, proceed to next iteration
+            // update block height
+            block_height = self
+                .block_store
+                .get_deserialized(&current_head)?
+                .ok_or_else(|| ForkChoiceError::MissingBeaconBlock(current_head))?
+                .slot()
+                .height(spec.genesis_slot);
+            // prune the latest votes for votes that are not part of current chosen chain
+            // more specifically, only keep votes that have head as an ancestor
+            for hash in latest_votes.keys() {
+                trace!(
+                    "Ancestor for vote: {} at height: {} is: {:?}",
+                    hash,
+                    block_height,
+                    self.get_ancestor(*hash, block_height, spec)
+                );
+            }
+            latest_votes.retain(|hash, _| {
+                self.get_ancestor(*hash, block_height, spec) == Some(current_head)
+            });
+        }
+    }
+}
+
+/// Type for storing blocks in a memory cache. Key is comprised of block-hash plus the height.
+#[derive(PartialEq, Eq, Hash)]
+pub struct CacheKey<T> {
+    block_hash: Hash256,
+    block_height: T,
+}
+
+impl<T> CacheKey<T> {
+    pub fn new(block_hash: &Hash256, block_height: T) -> Self {
+        CacheKey {
+            block_hash: *block_hash,
+            block_height,
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    pub fn test_power_of_2_below() {
+        assert_eq!(power_of_2_below(4), 4);
+        assert_eq!(power_of_2_below(5), 4);
+        assert_eq!(power_of_2_below(7), 4);
+        assert_eq!(power_of_2_below(24), 16);
+        assert_eq!(power_of_2_below(32), 32);
+        assert_eq!(power_of_2_below(33), 32);
+        assert_eq!(power_of_2_below(63), 32);
+    }
+
+    #[test]
+    pub fn test_power_of_2_below_large() {
+        let pow: u64 = 1 << 24;
+        for x in (pow - 20)..(pow + 20) {
+            assert!(power_of_2_below(x) <= x, "{}", x);
+        }
+    }
+}

eth2/fork_choice/tests/tests.rs

@@ -3,7 +3,7 @@
 extern crate beacon_chain;
 extern crate bls;
 extern crate db;
-//extern crate env_logger; // for debugging
+// extern crate env_logger; // for debugging
 extern crate fork_choice;
 extern crate hex;
 extern crate log;
@@ -15,8 +15,10 @@ pub use beacon_chain::BeaconChain;
 use bls::Signature;
 use db::stores::{BeaconBlockStore, BeaconStateStore};
 use db::MemoryDB;
-//use env_logger::{Builder, Env};
+// use env_logger::{Builder, Env};
-use fork_choice::{BitwiseLMDGhost, ForkChoice, ForkChoiceAlgorithm, LongestChain, SlowLMDGhost};
+use fork_choice::{
+    BitwiseLMDGhost, ForkChoice, ForkChoiceAlgorithm, LongestChain, OptimizedLMDGhost, SlowLMDGhost,
+};
 use ssz::ssz_encode;
 use std::collections::HashMap;
 use std::sync::Arc;
@@ -27,6 +29,18 @@ use yaml_rust::yaml;
 
 // Note: We Assume the block Id's are hex-encoded.
 
+#[test]
+fn test_optimized_lmd_ghost() {
+    // set up logging
+    // Builder::from_env(Env::default().default_filter_or("trace")).init();
+
+    test_yaml_vectors(
+        ForkChoiceAlgorithm::OptimizedLMDGhost,
+        "tests/lmd_ghost_test_vectors.yaml",
+        100,
+    );
+}
+
 #[test]
 fn test_bitwise_lmd_ghost() {
     // set up logging
@@ -212,6 +226,10 @@ fn setup_inital_state(
 
     // the fork choice instantiation
     let fork_choice: Box<ForkChoice> = match fork_choice_algo {
+        ForkChoiceAlgorithm::OptimizedLMDGhost => Box::new(OptimizedLMDGhost::new(
+            block_store.clone(),
+            state_store.clone(),
+        )),
         ForkChoiceAlgorithm::BitwiseLMDGhost => Box::new(BitwiseLMDGhost::new(
             block_store.clone(),
             state_store.clone(),