refactor: simplify ChainIndex, hoist out checkpoints (#3220)

Co-authored-by: Aatif Syed <38045910+aatifsyed@users.noreply.github.com>
ChainSafe · Jul 19, 2023 · 68a0fd8 · 68a0fd8
1 parent df5c5a3
commit 68a0fd8
Show file tree

Hide file tree

Showing 21 changed files with 397 additions and 489 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -29,6 +29,9 @@
 
 - [#3189](https://github.com/ChainSafe/forest/issues/3189): Changed the database
   organisation to use multiple columns. The database will need to be recreated.
+- [#3220](https://github.com/ChainSafe/forest/pull/3220): Removed the
+  `forest-cli chain validate-tipset-checkpoints` and
+  `forest-cli chain tipset-hash` commands.
 
 ### Added
 

diff --git a/build/known_blocks.yaml b/build/known_blocks.yaml
@@ -0,0 +1,52 @@
+# This file maps epochs to block headers for calibnet and mainnet. Forest use
+# this mapping to quickly identify the origin network of a tipset.
+#
+# Block headers can be inspected on filfox:
+# - https://filfox.info/en/block/bafy2bzacebnfm6dvxo7sm5thcxnv3kttoamb53uxycnvtdxgk5mh7d73qlly2
+# - https://calibration.filfox.info/en/block/bafy2bzacedhkkz76zdekpexha55b42eop42e24ajmajm26wws4nbvtq7louvu
+#
+# This file was generated with `forest-cli archive checkpoints`
+calibnet:
+  691200: bafy2bzacedhkkz76zdekpexha55b42eop42e24ajmajm26wws4nbvtq7louvu
+  604800: bafy2bzacec7prcdtac2blaeblzsorwpndtbjsz3k7tiyvwmellajinaaavpyo
+  518400: bafy2bzacecforpkd3h6hb46r5u3sqipcpzhws7itojwr65fj75nujftheeqya
+  432000: bafy2bzaceb2bryss7ixglbvpjyr2wcsdh3v4stgr7jarhbxdkzrzgfc7jo54s
+  345600: bafy2bzaceczzut4ogoavwc3hhrbucgzwm4ty5ifrbozhuce4mm3iqcapbg2l2
+  259200: bafy2bzaceajlesebltywlns4ctneyqtva7zut4fwrbihafm6c3rfydy5i4bw2
+  172800: bafy2bzacebluce5la2lbyjlgmup6mx4xytnnilsf4oy6v4omon2ivyduqwlsq
+  86400: bafy2bzaceatx7tlwdhez6vyias5qlhaxa54vjftigbuqzfsmdqduc6jdiclzc
+mainnet:
+  2937600: bafy2bzacec3jsm2p6v4vtyim6cehl33bbf3qa4whdptjmbkm4bcmsrkow2ddk
+  2851200: bafy2bzacecs64y3uj6r7ikajjiahkgzrdqb246dprus2jipgsnck5amq2ekpa
+  2764800: bafy2bzaceco5ad55adwn5qyzmjzthgpxulciqlj5ydj5jdqomcmya5w4laqw2
+  2678400: bafy2bzacecoypmmzncy4sxno6kbqwps4ylr7tqf5qcdrxvewiycmg7z5a4akg
+  2592000: bafy2bzacec6gkk66djzdlvwpm3ij4dmgl27jxt5qhusgslhcprrrue46gugei
+  2505600: bafy2bzaceahcpwgeu7echjebrcdc2kow6ma4qws5kljyydrujrnat7mo4goe4
+  2419200: bafy2bzacechvixq5nbkhhlcdxgtcax6u5ti4owzywt2jaxxmky3wedlodhtfm
+  2332800: bafy2bzaceapj6btckm2uurejddv4unroox4sq3ntorseizn72ph6wbnhpxnck
+  2246400: bafy2bzacebr2jcw3esw6kvd7oi6h76c4zj2t3ncmab2psgpchsk6vswmshmoi
+  2160000: bafy2bzacedrntfl4tsa4xb2jkqpegbkt6t2cbnj6osfsha2hwhuury4s5wn2c
+  2073600: bafy2bzacedct5jwgjedkrvskseihsab7koqxolhnbveihtygob24sbmf26zrk
+  1987200: bafy2bzacec7yhpo4o7jhdfcbnl4mevulp2ifxg6ojhub6xfrg3jdfajuxx3oc
+  1900800: bafy2bzacedvvcatpaowihbobz53s6oiwcq5sujdrrswztf65nhgoxqr77rlio
+  1814400: bafy2bzaced4dyvpbogdif4xfnqiqxu2pjb3xqjkzs7muycfgq3snv7ksorvnc
+  1728000: bafy2bzaceca3bvwdu67rlxh6fr4fbtggii3pjt4tcrtfmsqzt5epc4y6emqxa
+  1641600: bafy2bzacecsq3z3xpfe3dszo4ftt53g44dex2ubya3f3wtz7vkbsnpksri6cy
+  1555200: bafy2bzacebnfm6dvxo7sm5thcxnv3kttoamb53uxycnvtdxgk5mh7d73qlly2
+  1468800: bafy2bzacedo7ufcauavu27ptpbjdxquvd6zzlnv7wduysrgms3wrhfdxmifwu
+  1382400: bafy2bzacea7z2hklqikueykuf5ayb6ulmnijj3nuhtg3dya4ttojmr5m2rug6
+  1296000: bafy2bzacecky7lbzjkqeqcoglrgfgto76ztwgb5gxaqmrsxfvjqsjr7r427uk
+  1209600: bafy2bzacediunbjqtmbxdp7xynkifhcvgzv5uhjwhb6aocgbskd55d7zw6uv6
+  1123200: bafy2bzaceax6x7iildnssg6s3rzinru7odinha6jfwmk5rxpgqxh6rjkh2ysm
+  1036800: bafy2bzacebg5enb2gsbydtpegmjvwifhq7bsgibllqrczhn56aartok2hsz5c
+  950400: bafy2bzacecwpmsp4reyhbim2xpulhxnsp3l6duo267vsbwsiyyiszpgw5isfi
+  864000: bafy2bzacea7yrzlkpee6fnw2ckoxv3rxxidq7hlhk732znxqwtoqsqzgw5sf4
+  777600: bafy2bzacealxm7nc25xfo7jzzaal2ruakr3n2xwcq6wt7c3ntj4ksczlumf56
+  691200: bafy2bzaceavrlfce6qxtlrahevzfijpzrpgt2uwkfae3333z4gafd4lj7wodk
+  604800: bafy2bzacedexv5uhodwwhaxs7hqo7fgwx33oo22pswxc73gyrawdmnbzl3h7y
+  518400: bafy2bzacedhzi7hqtwelbwcejvq45xvcqlcrymersv2dlfqh3niqs4ohdpm5i
+  432000: bafy2bzacebutfwbn55lid7g4nsxr4dwje7tcie4npyidsfjvvc5ohnem4yaue
+  345600: bafy2bzaceavqxds47ihtcrfiihpxzlef6jvhbdnbpfzxsk6h4lczokxkl7bcw
+  259200: bafy2bzacecebsgo4guk6t6tukqux265cbzcav6pg6bihlay2z3y3mp3ositws
+  172800: bafy2bzaceajphhjy3saqf573ewxhfyu7dwwb7yodcwass64kevbktlk4wcawu
+  86400: bafy2bzacec7ywitfdjr5hkglybvgavg3hzo7oluwnmfo55aa6h2qw2lyz4nac
diff --git a/scripts/gen_coverage_report.sh b/scripts/gen_coverage_report.sh
@@ -39,7 +39,6 @@ SNAPSHOT_PATH=$(find "$TMP_DIR" -name \*.car | head -n 1)
 cov forest --chain calibnet --encrypt-keystore false --import-snapshot "$SNAPSHOT_PATH" --height=-200 --detach --track-peak-rss --save-token "$TOKEN_PATH"
 cov forest-cli sync wait
 cov forest-cli sync status
-cov forest-cli chain validate-tipset-checkpoints
 cov forest-cli --chain calibnet db gc
 cov forest-cli --chain calibnet db stats
 cov forest-cli snapshot export

diff --git a/scripts/tests/calibnet_other_check.sh b/scripts/tests/calibnet_other_check.sh
@@ -9,9 +9,6 @@ source "$(dirname "$0")/harness.sh"
 
 forest_init
 
-echo "Validating checkpoint tipset hashes"
-$FOREST_CLI_PATH chain validate-tipset-checkpoints
-
 echo "Running database garbage collection"
 forest_check_db_stats
 $FOREST_CLI_PATH db gc
@@ -20,20 +17,6 @@ forest_check_db_stats
 echo "Testing js console"
 $FOREST_CLI_PATH attach --exec 'showPeers()'
 
-# Get the checkpoint hash at epoch 424000. This output isn't affected by the
-# number of recent state roots we store (2k at the time of writing) and this
-# output should never change.
-echo "Checkpoint hash test"
-EXPECTED_HASH="Chain:           calibnet
-Epoch:           424000
-Checkpoint hash: 8cab45fd441c1fb68d2fd7e45d5e9ef9a5d3b45f68b414ab3e244233dd8e37fc4dacffc8966b2dc8804d4abf92c8a57efda743e26db6805a77a4feac19478293"
-ACTUAL_HASH=$($FOREST_CLI_PATH --chain calibnet chain tipset-hash 424000)
-if [[ $EXPECTED_HASH != "$ACTUAL_HASH" ]]; then
-  printf "Invalid tipset hash:\n%s" "$ACTUAL_HASH"
-  printf "Expected:\n%s" "$EXPECTED_HASH"
-  exit 1
-fi
-
 echo "Test dev commands (which could brick the node/cause subsequent snapshots to fail)"
 
 echo "Test subcommand: chain set-head"

diff --git a/scripts/tests/forest_cli_check.sh b/scripts/tests/forest_cli_check.sh
@@ -55,6 +55,10 @@ pushd "$(mktemp --directory)"
     if "$FOREST_CLI_PATH" snapshot validate --check-network mainnet "$validate_me"; then
         exit 1
     fi
+
+    : : check that it contains at least one expected checkpoint
+    # If calibnet is reset or the checkpoint interval is changed, this check has to be updated
+    "$FOREST_CLI_PATH" archive checkpoints "$validate_me" | grep bafy2bzaceatx7tlwdhez6vyias5qlhaxa54vjftigbuqzfsmdqduc6jdiclzc
 rm -- *
 popd
 
diff --git a/src/blocks/header/mod.rs b/src/blocks/header/mod.rs
@@ -36,9 +36,9 @@ pub mod json;
 /// use cid::multihash::MultihashDigest;
 ///
 /// BlockHeader::builder()
-///     .message_receipts(Cid::new_v1(DAG_CBOR, Identity.digest(&[]))) // required
-///     .state_root(Cid::new_v1(DAG_CBOR, Identity.digest(&[]))) // required
-///     .miner_address(Address::new_id(0)) // required
+///     .message_receipts(Cid::new_v1(DAG_CBOR, Identity.digest(&[]))) // optional
+///     .state_root(Cid::new_v1(DAG_CBOR, Identity.digest(&[]))) // optional
+///     .miner_address(Address::new_id(0)) // optional
 ///     .messages(Cid::new_v1(DAG_CBOR, Identity.digest(&[]))) // optional
 ///     .beacon_entries(Vec::new()) // optional
 ///     .winning_post_proof(Vec::new()) // optional
@@ -54,7 +54,7 @@ pub mod json;
 ///     .build()
 ///     .unwrap();
 /// ```
-#[derive(Clone, Debug, Builder)]
+#[derive(Clone, Debug, Default, Builder)]
 #[builder(name = "BlockHeaderBuilder")]
 pub struct BlockHeader {
     // CHAIN LINKING
@@ -84,6 +84,7 @@ pub struct BlockHeader {
 
     // MINER INFO
     /// `miner_address` is the address of the miner actor that mined this block
+    #[builder(default)]
     miner_address: Address,
 
     // STATE

diff --git a/src/blocks/tipset.rs b/src/blocks/tipset.rs
@@ -1,19 +1,20 @@
 // Copyright 2019-2023 ChainSafe Systems
 // SPDX-License-Identifier: Apache-2.0, MIT
 
-use std::fmt;
+use std::{fmt, sync::OnceLock};
 
-use crate::shim::address::Address;
-use crate::shim::clock::ChainEpoch;
+use crate::networks::{calibnet, mainnet};
+use crate::shim::{address::Address, clock::ChainEpoch};
 use crate::utils::cid::CidCborExt;
-use ahash::{HashSet, HashSetExt};
+use ahash::{HashMap, HashSet, HashSetExt};
 use anyhow::Context as _;
 use cid::Cid;
 use fvm_ipld_blockstore::Blockstore;
 use fvm_ipld_encoding::CborStore;
 use num::BigInt;
 use once_cell::sync::OnceCell;
 use serde::{Deserialize, Serialize};
+use std::str::FromStr;
 use tracing::info;
 
 use super::{Block, BlockHeader, Error, Ticket};
@@ -265,6 +266,49 @@ impl Tipset {
             })
         })
     }
+
+    /// Fetch the genesis block header for a given tipset.
+    pub fn genesis(&self, store: impl Blockstore) -> anyhow::Result<BlockHeader> {
+        // Scanning through millions of epochs to find the genesis is quite
+        // slow. Let's use a list of known blocks to short-circuit the search.
+        // The blocks are hash-chained together and known blocks are guaranteed
+        // to have a known genesis.
+        #[derive(Serialize, Deserialize)]
+        struct KnownHeaders {
+            calibnet: HashMap<ChainEpoch, String>,
+            mainnet: HashMap<ChainEpoch, String>,
+        }
+
+        static KNOWN_HEADERS: OnceLock<KnownHeaders> = OnceLock::new();
+        let headers = KNOWN_HEADERS.get_or_init(|| {
+            serde_yaml::from_str(include_str!("../../build/known_blocks.yaml")).unwrap()
+        });
+
+        let calibnet_cid = Cid::from_str(calibnet::GENESIS_CID).unwrap();
+        let mainnet_cid = Cid::from_str(mainnet::GENESIS_CID).unwrap();
+
+        for tipset in self.clone().chain(&store) {
+            // Search for known calibnet and mainnet blocks
+            for (genesis_cid, known_blocks) in [
+                (calibnet_cid, &headers.calibnet),
+                (mainnet_cid, &headers.mainnet),
+            ] {
+                if let Some(known_block_cid) = known_blocks.get(&tipset.epoch()) {
+                    if known_block_cid == &tipset.min_ticket_block().cid().to_string() {
+                        return store
+                            .get_cbor(&genesis_cid)?
+                            .ok_or_else(|| anyhow::anyhow!("Genesis block missing from database"));
+                    }
+                }
+            }
+
+            // If no known blocks are found, we'll eventually hit the genesis tipset.
+            if tipset.epoch() == 0 {
+                return Ok(tipset.min_ticket_block().clone());
+            }
+        }
+        anyhow::bail!("Genesis block not found")
+    }
 }
 
 /// `FullTipset` is an expanded version of a tipset that contains all the blocks

diff --git a/src/chain/store/chain_store.rs b/src/chain/store/chain_store.rs
@@ -9,7 +9,7 @@ use crate::interpreter::BlockMessages;
 use crate::ipld::{walk_snapshot, WALK_SNAPSHOT_PROGRESS_EXPORT};
 use crate::libp2p_bitswap::{BitswapStoreRead, BitswapStoreReadWrite};
 use crate::message::{ChainMessage, Message as MessageTrait, SignedMessage};
-use crate::networks::{ChainConfig, NetworkChain};
+use crate::networks::ChainConfig;
 use crate::shim::clock::ChainEpoch;
 use crate::shim::{
     address::Address, econ::TokenAmount, executor::Receipt, message::Message,
@@ -43,7 +43,7 @@ use tokio::sync::{
 use tracing::{debug, info, warn};
 
 use super::{
-    index::{checkpoint_tipsets, ChainIndex},
+    index::{ChainIndex, ResolveNullTipset},
     tipset_tracker::TipsetTracker,
     Error,
 };
@@ -74,7 +74,7 @@ pub struct ChainStore<DB> {
     pub db: Arc<DB>,
 
     /// Used as a cache for tipset `lookbacks`.
-    chain_index: ChainIndex<DB>,
+    pub chain_index: ChainIndex<DB>,
 
     /// Tracks blocks for the purpose of forming tipsets.
     tipset_tracker: TipsetTracker<DB>,
@@ -255,11 +255,6 @@ where
         self.chain_index.load_tipset(tsk)
     }
 
-    /// Returns Tipset key hash from key-value store from provided CIDs
-    pub fn tipset_hash_from_keys(&self, tsk: &TipsetKeys) -> String {
-        checkpoint_tipsets::tipset_hash(tsk)
-    }
-
     /// Determines if provided tipset is heavier than existing known heaviest
     /// tipset
     fn update_heaviest<S>(&self, ts: Arc<Tipset>) -> Result<(), Error>
@@ -300,92 +295,6 @@ where
         let _did_work = file.remove(cid);
     }
 
-    /// Returns the tipset behind `tsk` at a given `height`.
-    /// If the given height is a null round:
-    /// - If `prev` is `true`, the tipset before the null round is returned.
-    /// - If `prev` is `false`, the tipset following the null round is returned.
-    #[tracing::instrument(skip(self, ts))]
-    pub fn tipset_by_height(
-        &self,
-        height: ChainEpoch,
-        ts: Arc<Tipset>,
-        prev: bool,
-    ) -> Result<Arc<Tipset>, Error> {
-        if height > ts.epoch() {
-            return Err(Error::Other(
-                "searching for tipset that has a height less than starting point".to_owned(),
-            ));
-        }
-        if height == ts.epoch() {
-            return Ok(ts);
-        }
-
-        let mut lbts = self.chain_index.get_tipset_by_height(ts.clone(), height)?;
-
-        if lbts.epoch() < height {
-            warn!(
-                "chain index returned the wrong tipset at height {}, using slow retrieval",
-                height
-            );
-            lbts = self
-                .chain_index
-                .get_tipset_by_height_without_cache(ts, height)?;
-        }
-
-        if lbts.epoch() == height || !prev {
-            Ok(lbts)
-        } else {
-            self.chain_index.load_tipset(lbts.parents())
-        }
-    }
-
-    pub fn validate_tipset_checkpoints(
-        &self,
-        from: Arc<Tipset>,
-        network: &NetworkChain,
-    ) -> Result<(), Error> {
-        info!(
-            "Validating {network} tipset checkpoint hashes from: {}",
-            from.epoch()
-        );
-
-        let Some(mut hashes) = checkpoint_tipsets::get_tipset_hashes(network) else {
-            info!("No checkpoint tipsets found for network: {network}, skipping validation.");
-            return Ok(());
-        };
-
-        let mut ts = from;
-        let tipset_hash = checkpoint_tipsets::tipset_hash(ts.key());
-        hashes.remove(&tipset_hash);
-
-        loop {
-            let pts = self.chain_index.load_tipset(ts.parents())?;
-            let tipset_hash = checkpoint_tipsets::tipset_hash(ts.key());
-            hashes.remove(&tipset_hash);
-
-            ts = pts;
-
-            if ts.epoch() == 0 {
-                break;
-            }
-        }
-
-        if !hashes.is_empty() {
-            return Err(Error::Other(format!(
-                "Found tipset hash(es) on {network} that are no longer valid: {hashes:?}"
-            )));
-        }
-
-        if !checkpoint_tipsets::validate_genesis_cid(&ts, network) {
-            return Err(Error::Other(format!(
-                "Genesis cid {:?} on {network} network does not match with one stored in checkpoint registry",
-                ts.key().cid()
-            )));
-        }
-
-        Ok(())
-    }
-
     /// Finds the latest beacon entry given a tipset up to 20 tipsets behind
     pub fn latest_beacon_entry(&self, ts: &Tipset) -> Result<BeaconEntry, Error> {
         let check_for_beacon_entry = |ts: &Tipset| {
@@ -594,7 +503,8 @@ where
         round: ChainEpoch,
     ) -> Result<TipsetKeys, Error> {
         let ts = self
-            .tipset_by_height(round, tipset, true)
+            .chain_index
+            .tipset_by_height(round, tipset, ResolveNullTipset::TakeOlder)
             .map_err(|e| Error::Other(format!("Could not get tipset by height {e:?}")))?;
         Ok(ts.key().clone())
     }
@@ -647,7 +557,12 @@ where
         }
 
         let next_ts = self
-            .tipset_by_height(lbr + 1, heaviest_tipset.clone(), false)
+            .chain_index
+            .tipset_by_height(
+                lbr + 1,
+                heaviest_tipset.clone(),
+                ResolveNullTipset::TakeNewer,
+            )
             .map_err(|e| Error::Other(format!("Could not get tipset by height {e:?}")))?;
         if lbr > next_ts.epoch() {
             return Err(Error::Other(format!(