feat: without-evm cli option in reth (#12134)

Co-authored-by: Matthias Seitz <matthias.seitz@outlook.de>

Cargo.lock

@@ -6616,6 +6616,7 @@ dependencies = [
  "ahash",
  "alloy-eips",
  "alloy-primitives",
+ "alloy-rlp",
  "arbitrary",
  "backon",
  "clap",

book/cli/reth/init-state.md

@@ -72,6 +72,22 @@ Database:
       --db.read-transaction-timeout <READ_TRANSACTION_TIMEOUT>
           Read transaction timeout in seconds, 0 means no timeout
 
+      --without-evm
+          Specifies whether to initialize the state without relying on EVM historical data.
+
+          When enabled, and before inserting the state, it creates a dummy chain up to the last EVM block specified. It then, appends the first block provided block.
+
+          - **Note**: **Do not** import receipts and blocks beforehand, or this will fail or be ignored.
+
+      --header <HEADER_FILE>
+          Header file containing the header in an RLP encoded format.
+
+      --total-difficulty <TOTAL_DIFFICULTY>
+          Total difficulty of the header.
+
+      --header-hash <HEADER_HASH>
+          Hash of the header.
+
   <STATE_DUMP_FILE>
           JSONL file with state dump.
 

crates/cli/commands/Cargo.toml

@@ -51,6 +51,7 @@ reth-trie-common = { workspace = true, optional = true }
 # ethereum
 alloy-eips.workspace = true
 alloy-primitives.workspace = true
+alloy-rlp.workspace = true
 
 itertools.workspace = true
 futures.workspace = true

crates/cli/commands/src/init_state.rs

@@ -1,79 +0,0 @@
-//! Command that initializes the node from a genesis file.
-
-use crate::common::{AccessRights, Environment, EnvironmentArgs};
-use alloy_primitives::B256;
-use clap::Parser;
-use reth_chainspec::{EthChainSpec, EthereumHardforks};
-use reth_cli::chainspec::ChainSpecParser;
-use reth_config::config::EtlConfig;
-use reth_db_common::init::init_from_state_dump;
-use reth_node_builder::NodeTypesWithEngine;
-use reth_provider::{providers::ProviderNodeTypes, ProviderFactory};
-
-use std::{fs::File, io::BufReader, path::PathBuf};
-use tracing::info;
-
-/// Initializes the database with the genesis block.
-#[derive(Debug, Parser)]
-pub struct InitStateCommand<C: ChainSpecParser> {
-    #[command(flatten)]
-    pub env: EnvironmentArgs<C>,
-
-    /// JSONL file with state dump.
-    ///
-    /// Must contain accounts in following format, additional account fields are ignored. Must
-    /// also contain { "root": \<state-root\> } as first line.
-    /// {
-    ///     "balance": "\<balance\>",
-    ///     "nonce": \<nonce\>,
-    ///     "code": "\<bytecode\>",
-    ///     "storage": {
-    ///         "\<key\>": "\<value\>",
-    ///         ..
-    ///     },
-    ///     "address": "\<address\>",
-    /// }
-    ///
-    /// Allows init at a non-genesis block. Caution! Blocks must be manually imported up until
-    /// and including the non-genesis block to init chain at. See 'import' command.
-    #[arg(value_name = "STATE_DUMP_FILE", verbatim_doc_comment)]
-    pub state: PathBuf,
-}
-
-impl<C: ChainSpecParser<ChainSpec: EthChainSpec + EthereumHardforks>> InitStateCommand<C> {
-    /// Execute the `init` command
-    pub async fn execute<N: NodeTypesWithEngine<ChainSpec = C::ChainSpec>>(
-        self,
-    ) -> eyre::Result<()> {
-        info!(target: "reth::cli", "Reth init-state starting");
-
-        let Environment { config, provider_factory, .. } = self.env.init::<N>(AccessRights::RW)?;
-
-        info!(target: "reth::cli", "Initiating state dump");
-
-        let hash = init_at_state(self.state, provider_factory, config.stages.etl)?;
-
-        info!(target: "reth::cli", hash = ?hash, "Genesis block written");
-        Ok(())
-    }
-}
-
-/// Initialize chain with state at specific block, from a file with state dump.
-pub fn init_at_state<N: ProviderNodeTypes>(
-    state_dump_path: PathBuf,
-    factory: ProviderFactory<N>,
-    etl_config: EtlConfig,
-) -> eyre::Result<B256> {
-    info!(target: "reth::cli",
-        path=?state_dump_path,
-        "Opening state dump");
-
-    let file = File::open(state_dump_path)?;
-    let reader = BufReader::new(file);
-
-    let provider_rw = factory.provider_rw()?;
-    let hash = init_from_state_dump(reader, &provider_rw.0, etl_config)?;
-    provider_rw.commit()?;
-
-    Ok(hash)
-}

crates/cli/commands/src/init_state/mod.rs

@@ -0,0 +1,132 @@
+//! Command that initializes the node from a genesis file.
+
+use crate::common::{AccessRights, Environment, EnvironmentArgs};
+use alloy_primitives::{B256, U256};
+use clap::Parser;
+use reth_chainspec::{EthChainSpec, EthereumHardforks};
+use reth_cli::chainspec::ChainSpecParser;
+use reth_db_common::init::init_from_state_dump;
+use reth_node_builder::NodeTypesWithEngine;
+use reth_primitives::SealedHeader;
+use reth_provider::{
+    BlockNumReader, DatabaseProviderFactory, StaticFileProviderFactory, StaticFileWriter,
+};
+
+use std::{fs::File, io::BufReader, path::PathBuf, str::FromStr};
+use tracing::info;
+
+pub mod without_evm;
+
+/// Initializes the database with the genesis block.
+#[derive(Debug, Parser)]
+pub struct InitStateCommand<C: ChainSpecParser> {
+    #[command(flatten)]
+    pub env: EnvironmentArgs<C>,
+
+    /// JSONL file with state dump.
+    ///
+    /// Must contain accounts in following format, additional account fields are ignored. Must
+    /// also contain { "root": \<state-root\> } as first line.
+    /// {
+    ///     "balance": "\<balance\>",
+    ///     "nonce": \<nonce\>,
+    ///     "code": "\<bytecode\>",
+    ///     "storage": {
+    ///         "\<key\>": "\<value\>",
+    ///         ..
+    ///     },
+    ///     "address": "\<address\>",
+    /// }
+    ///
+    /// Allows init at a non-genesis block. Caution! Blocks must be manually imported up until
+    /// and including the non-genesis block to init chain at. See 'import' command.
+    #[arg(value_name = "STATE_DUMP_FILE", verbatim_doc_comment)]
+    pub state: PathBuf,
+
+    /// Specifies whether to initialize the state without relying on EVM historical data.
+    ///
+    /// When enabled, and before inserting the state, it creates a dummy chain up to the last EVM
+    /// block specified. It then, appends the first block provided block.
+    ///
+    /// - **Note**: **Do not** import receipts and blocks beforehand, or this will fail or be
+    ///   ignored.
+    #[arg(long, default_value = "false")]
+    pub without_evm: bool,
+
+    /// Header file containing the header in an RLP encoded format.
+    #[arg(long, value_name = "HEADER_FILE", verbatim_doc_comment)]
+    pub header: Option<PathBuf>,
+
+    /// Total difficulty of the header.
+    #[arg(long, value_name = "TOTAL_DIFFICULTY", verbatim_doc_comment)]
+    pub total_difficulty: Option<String>,
+
+    /// Hash of the header.
+    #[arg(long, value_name = "HEADER_HASH", verbatim_doc_comment)]
+    pub header_hash: Option<String>,
+}
+
+impl<C: ChainSpecParser<ChainSpec: EthChainSpec + EthereumHardforks>> InitStateCommand<C> {
+    /// Execute the `init` command
+    pub async fn execute<N: NodeTypesWithEngine<ChainSpec = C::ChainSpec>>(
+        self,
+    ) -> eyre::Result<()> {
+        info!(target: "reth::cli", "Reth init-state starting");
+
+        let Environment { config, provider_factory, .. } = self.env.init::<N>(AccessRights::RW)?;
+
+        let static_file_provider = provider_factory.static_file_provider();
+        let provider_rw = provider_factory.database_provider_rw()?;
+
+        if self.without_evm {
+            // ensure header, total difficulty and header hash are provided
+            let header = self.header.ok_or_else(|| eyre::eyre!("Header file must be provided"))?;
+            let header = without_evm::read_header_from_file(header)?;
+
+            let header_hash =
+                self.header_hash.ok_or_else(|| eyre::eyre!("Header hash must be provided"))?;
+            let header_hash = B256::from_str(&header_hash)?;
+
+            let total_difficulty = self
+                .total_difficulty
+                .ok_or_else(|| eyre::eyre!("Total difficulty must be provided"))?;
+            let total_difficulty = U256::from_str(&total_difficulty)?;
+
+            let last_block_number = provider_rw.last_block_number()?;
+
+            if last_block_number == 0 {
+                without_evm::setup_without_evm(
+                    &provider_rw,
+                    &static_file_provider,
+                    // &header,
+                    // header_hash,
+                    SealedHeader::new(header, header_hash),
+                    total_difficulty,
+                )?;
+
+                // SAFETY: it's safe to commit static files, since in the event of a crash, they
+                // will be unwinded according to database checkpoints.
+                //
+                // Necessary to commit, so the header is accessible to provider_rw and
+                // init_state_dump
+                static_file_provider.commit()?;
+            } else if last_block_number > 0 && last_block_number < header.number {
+                return Err(eyre::eyre!(
+                    "Data directory should be empty when calling init-state with --without-evm-history."
+                ));
+            }
+        }
+
+        info!(target: "reth::cli", "Initiating state dump");
+
+        let file = File::open(self.state)?;
+        let reader = BufReader::new(file);
+
+        let hash = init_from_state_dump(reader, &provider_rw, config.stages.etl)?;
+
+        provider_rw.commit()?;
+
+        info!(target: "reth::cli", hash = ?hash, "Genesis block written");
+        Ok(())
+    }
+}

crates/cli/commands/src/init_state/without_evm.rs

@@ -1,5 +1,6 @@
 use alloy_primitives::{BlockNumber, B256, U256};
-use reth_optimism_primitives::bedrock::{BEDROCK_HEADER, BEDROCK_HEADER_HASH, BEDROCK_HEADER_TTD};
+use alloy_rlp::Decodable;
+
 use reth_primitives::{
     BlockBody, Header, SealedBlock, SealedBlockWithSenders, SealedHeader, StaticFileSegment,
 };
@@ -7,28 +8,42 @@ use reth_provider::{
     providers::StaticFileProvider, BlockWriter, StageCheckpointWriter, StaticFileWriter,
 };
 use reth_stages::{StageCheckpoint, StageId};
+
+use std::{fs::File, io::Read, path::PathBuf};
 use tracing::info;
 
-/// Creates a dummy chain (with no transactions) up to the last OVM block and appends the
-/// first valid Bedrock block.
-pub(crate) fn setup_op_mainnet_without_ovm<Provider>(
+/// Reads the header RLP from a file and returns the Header.
+pub(crate) fn read_header_from_file(path: PathBuf) -> Result<Header, eyre::Error> {
+    let mut file = File::open(path)?;
+    let mut buf = Vec::new();
+    file.read_to_end(&mut buf)?;
+
+    let header = Header::decode(&mut &buf[..])?;
+    Ok(header)
+}
+
+/// Creates a dummy chain (with no transactions) up to the last EVM block and appends the
+/// first valid block.
+pub fn setup_without_evm<Provider>(
     provider_rw: &Provider,
     static_file_provider: &StaticFileProvider,
+    header: SealedHeader,
+    total_difficulty: U256,
 ) -> Result<(), eyre::Error>
 where
     Provider: StageCheckpointWriter + BlockWriter,
 {
-    info!(target: "reth::cli", "Setting up dummy OVM chain before importing state.");
+    info!(target: "reth::cli", "Setting up dummy EVM chain before importing state.");
 
-    // Write OVM dummy data up to `BEDROCK_HEADER - 1` block
-    append_dummy_chain(static_file_provider, BEDROCK_HEADER.number - 1)?;
+    // Write EVM dummy data up to `header - 1` block
+    append_dummy_chain(static_file_provider, header.number - 1)?;
 
-    info!(target: "reth::cli", "Appending Bedrock block.");
+    info!(target: "reth::cli", "Appending first valid block.");
 
-    append_bedrock_block(provider_rw, static_file_provider)?;
+    append_first_block(provider_rw, static_file_provider, &header, total_difficulty)?;
 
     for stage in StageId::ALL {
-        provider_rw.save_stage_checkpoint(stage, StageCheckpoint::new(BEDROCK_HEADER.number))?;
+        provider_rw.save_stage_checkpoint(stage, StageCheckpoint::new(header.number))?;
     }
 
     info!(target: "reth::cli", "Set up finished.");
@@ -36,38 +51,30 @@ where
     Ok(())
 }
 
-/// Appends the first bedrock block.
+/// Appends the first block.
 ///
 /// By appending it, static file writer also verifies that all segments are at the same
 /// height.
-fn append_bedrock_block(
+fn append_first_block(
     provider_rw: impl BlockWriter,
     sf_provider: &StaticFileProvider,
+    header: &SealedHeader,
+    total_difficulty: U256,
 ) -> Result<(), eyre::Error> {
     provider_rw.insert_block(
-        SealedBlockWithSenders::new(
-            SealedBlock::new(
-                SealedHeader::new(BEDROCK_HEADER, BEDROCK_HEADER_HASH),
-                BlockBody::default(),
-            ),
-            vec![],
-        )
-        .expect("no senders or txes"),
+        SealedBlockWithSenders::new(SealedBlock::new(header.clone(), BlockBody::default()), vec![])
+            .expect("no senders or txes"),
     )?;
 
     sf_provider.latest_writer(StaticFileSegment::Headers)?.append_header(
-        &BEDROCK_HEADER,
-        BEDROCK_HEADER_TTD,
-        &BEDROCK_HEADER_HASH,
+        header,
+        total_difficulty,
+        &header.hash(),
     )?;
 
-    sf_provider
-        .latest_writer(StaticFileSegment::Receipts)?
-        .increment_block(BEDROCK_HEADER.number)?;
+    sf_provider.latest_writer(StaticFileSegment::Receipts)?.increment_block(header.number)?;
 
-    sf_provider
-        .latest_writer(StaticFileSegment::Transactions)?
-        .increment_block(BEDROCK_HEADER.number)?;
+    sf_provider.latest_writer(StaticFileSegment::Transactions)?.increment_block(header.number)?;
 
     Ok(())
 }

crates/optimism/cli/src/commands/init_state.rs

@@ -6,7 +6,8 @@ use reth_cli_commands::common::{AccessRights, Environment};
 use reth_db_common::init::init_from_state_dump;
 use reth_node_builder::NodeTypesWithEngine;
 use reth_optimism_chainspec::OpChainSpec;
-use reth_optimism_primitives::bedrock::BEDROCK_HEADER;
+use reth_optimism_primitives::bedrock::{BEDROCK_HEADER, BEDROCK_HEADER_HASH, BEDROCK_HEADER_TTD};
+use reth_primitives::SealedHeader;
 use reth_provider::{
     BlockNumReader, ChainSpecProvider, DatabaseProviderFactory, StaticFileProviderFactory,
     StaticFileWriter,
@@ -14,8 +15,6 @@ use reth_provider::{
 use std::{fs::File, io::BufReader};
 use tracing::info;
 
-mod bedrock;
-
 /// Initializes the database with the genesis block.
 #[derive(Debug, Parser)]
 pub struct InitStateCommandOp<C: ChainSpecParser> {
@@ -53,7 +52,12 @@ impl<C: ChainSpecParser<ChainSpec = OpChainSpec>> InitStateCommandOp<C> {
             let last_block_number = provider_rw.last_block_number()?;
 
             if last_block_number == 0 {
-                bedrock::setup_op_mainnet_without_ovm(&provider_rw, &static_file_provider)?;
+                reth_cli_commands::init_state::without_evm::setup_without_evm(
+                    &provider_rw,
+                    &static_file_provider,
+                    SealedHeader::new(BEDROCK_HEADER, BEDROCK_HEADER_HASH),
+                    BEDROCK_HEADER_TTD,
+                )?;
 
                 // SAFETY: it's safe to commit static files, since in the event of a crash, they
                 // will be unwinded according to database checkpoints.