perf(vm): general performance/space improvements (#482)

Jon-Becker · web-flow · commit 90e6622f0724 · 2024-08-25T15:12:17.000-07:00
diff --git a/crates/cfg/src/core/graph.rs b/crates/cfg/src/core/graph.rs
@@ -2,7 +2,7 @@ use alloy::primitives::U256;
 use eyre::{OptionExt, Result};
 use heimdall_common::utils::strings::encode_hex_reduced;
 use heimdall_vm::{
-    core::opcodes::{OpCodeInfo, JUMPDEST},
+    core::opcodes::{opcode_name, JUMPDEST},
     ext::exec::VMTrace,
 };
 use petgraph::{matrix_graph::NodeIndex, Graph};
@@ -21,7 +21,7 @@ pub fn build_cfg(
 
     // add the current operations to the cfg
     for operation in &vm_trace.operations {
-        let opcode_name = OpCodeInfo::from(operation.last_instruction.opcode).name();
+        let opcode_name = opcode_name(operation.last_instruction.opcode);
 
         let assembly = format!(
             "{} {} {}",
diff --git a/crates/decompile/src/utils/heuristics/arguments.rs b/crates/decompile/src/utils/heuristics/arguments.rs
@@ -4,7 +4,7 @@ use alloy::primitives::U256;
 use eyre::eyre;
 use heimdall_common::utils::strings::find_balanced_encapsulator;
 use heimdall_vm::core::{
-    opcodes::{OpCodeInfo, CALLDATALOAD, ISZERO},
+    opcodes::{opcode_name, CALLDATALOAD, ISZERO},
     types::{byte_size_to_type, convert_bitmask},
     vm::State,
 };
@@ -72,7 +72,7 @@ pub fn argument_heuristic(
                     debug!(
                         "instruction {} ({}) indicates argument {} is masked to {} bytes",
                         state.last_instruction.instruction,
-                        OpCodeInfo::from(state.last_instruction.opcode).name(),
+                        opcode_name(state.last_instruction.opcode),
                         arg_index,
                         mask_size_bytes
                     );
@@ -216,7 +216,7 @@ pub fn argument_heuristic(
                 debug!(
                     "instruction {} ({}) indicates argument {} may be a numeric type",
                     state.last_instruction.instruction,
-                    OpCodeInfo::from(state.last_instruction.opcode).name(),
+                    opcode_name(state.last_instruction.opcode),
                     arg_index
                 );
 
@@ -239,7 +239,7 @@ pub fn argument_heuristic(
                 debug!(
                     "instruction {} ({}) indicates argument {} may be a bytes type",
                     state.last_instruction.instruction,
-                    OpCodeInfo::from(state.last_instruction.opcode).name(),
+                    opcode_name(state.last_instruction.opcode),
                     arg_index
                 );
 
@@ -262,7 +262,7 @@ pub fn argument_heuristic(
                     debug!(
                         "instruction {} ({}) indicates argument {} may be a boolean",
                         state.last_instruction.instruction,
-                        OpCodeInfo::from(state.last_instruction.opcode).name(),
+                        opcode_name(state.last_instruction.opcode),
                         arg_index
                     );
 
diff --git a/crates/decompile/src/utils/heuristics/modifiers.rs b/crates/decompile/src/utils/heuristics/modifiers.rs
@@ -9,42 +9,29 @@ use tracing::debug;
 
 use crate::{core::analyze::AnalyzerState, interfaces::AnalyzedFunction, Error};
 
-use lazy_static::lazy_static;
-
-lazy_static! {
-    /// A list of opcodes that are considered non-pure (state accessing)
-    pub static ref NON_PURE_OPCODES: Vec<u8> = vec![
-        0x31, 0x32, 0x33, 0x3a, 0x3b, 0x3c, 0x40, 0x41, 0x42,
-        0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x54, 0x55, 0xf0,
-        0xf1, 0xf2, 0xf4, 0xf5, 0xfa, 0xff
-    ];
-    /// A list of opcodes that are considered non-view (state modifying)
-    pub static ref NON_VIEW_OPCODES: Vec<u8> = vec![
-        0x55, 0xf0, 0xf1, 0xf2, 0xf4, 0xf5, 0xfa, 0xff
-    ];
-}
-
 pub fn modifier_heuristic(
     function: &mut AnalyzedFunction,
     state: &State,
     _: &mut AnalyzerState,
 ) -> Result<(), Error> {
-    let opcode_name = OpCodeInfo::from(state.last_instruction.opcode).name();
+    let opcode_info = OpCodeInfo::from(state.last_instruction.opcode);
 
     // if any instruction is non-pure, the function is non-pure
-    if function.pure && NON_PURE_OPCODES.contains(&state.last_instruction.opcode) {
+    if function.pure && !opcode_info.is_pure() {
         debug!(
             "instruction {} ({}) indicates a non-pure function",
-            state.last_instruction.instruction, opcode_name
+            state.last_instruction.instruction,
+            opcode_info.name()
         );
         function.pure = false;
     }
 
     // if any instruction is non-view, the function is non-view
-    if function.view && NON_VIEW_OPCODES.contains(&state.last_instruction.opcode) {
+    if function.view && !opcode_info.is_view() {
         debug!(
             "instruction {} ({}) indicates a non-view function",
-            state.last_instruction.instruction, opcode_name
+            state.last_instruction.instruction,
+            opcode_info.name()
         );
         function.view = false;
     }
diff --git a/crates/decompile/src/utils/heuristics/solidity.rs b/crates/decompile/src/utils/heuristics/solidity.rs
@@ -1,7 +1,7 @@
 use alloy::primitives::U256;
 use alloy_dyn_abi::{DynSolType, DynSolValue};
 use heimdall_common::utils::strings::encode_hex_reduced;
-use heimdall_vm::core::{opcodes::OpCodeInfo, vm::State};
+use heimdall_vm::core::{opcodes::opcode_name, vm::State};
 
 use crate::{
     core::analyze::AnalyzerState,
@@ -188,7 +188,7 @@ pub fn solidity_heuristic(
                     function.logic.push(format!(
                         "(bool success, bytes memory ret0) = address({}).{}{}(abi.encode({}));",
                         address,
-                        OpCodeInfo::from(instruction.opcode).name().to_lowercase(),
+                        opcode_name(instruction.opcode).to_lowercase(),
                         modifier,
                         calldata
                             .iter()
diff --git a/crates/decompile/src/utils/heuristics/yul.rs b/crates/decompile/src/utils/heuristics/yul.rs
@@ -1,5 +1,5 @@
 use heimdall_common::utils::strings::encode_hex_reduced;
-use heimdall_vm::core::{opcodes::OpCodeInfo, vm::State};
+use heimdall_vm::core::{opcodes::opcode_name, vm::State};
 
 use crate::{
     core::analyze::AnalyzerState,
@@ -24,7 +24,7 @@ pub fn yul_heuristic(
             function.memory.insert(key, StorageFrame { operation });
             function.logic.push(format!(
                 "{}({}, {})",
-                OpCodeInfo::from(instruction.opcode).name().to_lowercase(),
+                opcode_name(instruction.opcode).to_lowercase(),
                 encode_hex_reduced(key),
                 instruction.input_operations[1].yulify()
             ));
@@ -80,7 +80,7 @@ pub fn yul_heuristic(
         0xff | 0xA0 | 0xA1 | 0xA2 | 0xA3 | 0xA4 => {
             function.logic.push(format!(
                 "{}({})",
-                OpCodeInfo::from(instruction.opcode).name().to_lowercase(),
+                opcode_name(instruction.opcode).to_lowercase(),
                 instruction
                     .input_operations
                     .iter()
diff --git a/crates/disassemble/src/core/mod.rs b/crates/disassemble/src/core/mod.rs
@@ -3,7 +3,7 @@ use std::time::Instant;
 use crate::{error::Error, interfaces::DisassemblerArgs};
 use eyre::eyre;
 use heimdall_common::utils::strings::encode_hex;
-use heimdall_vm::core::opcodes::OpCodeInfo;
+use heimdall_vm::core::opcodes::opcode_name;
 use tracing::{debug, info};
 
 pub async fn disassemble(args: DisassemblerArgs) -> Result<String, Error> {
@@ -45,7 +45,7 @@ pub async fn disassemble(args: DisassemblerArgs) -> Result<String, Error> {
                 } else {
                     format!("{:06x}", program_counter)
                 },
-                OpCodeInfo::from(opcode).name(),
+                opcode_name(opcode),
                 pushed_bytes
             )
             .as_str(),
diff --git a/crates/vm/benches/bench_ten_thousand_hashes.rs b/crates/vm/benches/bench_ten_thousand_hashes.rs
@@ -4,10 +4,10 @@ use heimdall_common::utils::strings::decode_hex;
 use heimdall_vm::core::vm::VM;
 use tokio::runtime::Runtime;
 
-fn test_fib(c: &mut Criterion) {
+fn test_ten_thousand_hashes(c: &mut Criterion) {
     let mut group = c.benchmark_group("heimdall_vm");
 
-    group.sample_size(500);
+    group.sample_size(100);
     group.bench_function(BenchmarkId::from_parameter("ten_thousand_hashes"), |b| {
         b.to_async::<Runtime>(Runtime::new().unwrap()).iter(|| async {
             // build the evm
@@ -33,5 +33,5 @@ fn test_fib(c: &mut Criterion) {
     group.finish();
 }
 
-criterion_group!(benches, test_fib);
+criterion_group!(benches, test_ten_thousand_hashes);
 criterion_main!(benches);
diff --git a/crates/vm/src/core/opcodes/mod.rs b/crates/vm/src/core/opcodes/mod.rs
@@ -75,7 +75,7 @@ impl OpCodeInfo {
 impl From<u8> for OpCodeInfo {
     #[inline]
     fn from(opcode: u8) -> Self {
-        OPCODE_INFO_JUMPTABLE[opcode as usize].unwrap_or(OpCodeInfo {
+        OPCODE_INFO_TABLE[opcode as usize].unwrap_or(OpCodeInfo {
             name: "unknown",
             inputs: 0,
             outputs: 0,
@@ -269,7 +269,7 @@ macro_rules! opcodes {
         )*
 
         /// Maps each opcode to its info.
-        pub const OPCODE_INFO_JUMPTABLE: [Option<OpCodeInfo>; 256] = {
+        pub const OPCODE_INFO_TABLE: [Option<OpCodeInfo>; 256] = {
             let mut map = [None; 256];
             let mut prev: u8 = 0;
             $(
@@ -287,9 +287,24 @@ macro_rules! opcodes {
             let _ = prev;
             map
         };
+
+        /// Maps each opcode to its name. (So we dont need to load [`OpCodeInfo`] to get the name)
+        pub const OPCODE_NAME_TABLE: [&'static str; 256] = {
+            let mut map = ["unknown"; 256];
+            $(
+                map[$val] = stringify!($name);
+            )*
+            map
+        };
     }
 }
 
+/// Get the name of an opcode.
+#[inline]
+pub fn opcode_name(opcode: u8) -> &'static str {
+    OPCODE_NAME_TABLE[opcode as usize]
+}
+
 opcodes! {
     0x00 => STOP => terminating;
 
diff --git a/crates/vm/src/core/opcodes/wrapped.rs b/crates/vm/src/core/opcodes/wrapped.rs
@@ -1,6 +1,6 @@
 use alloy::primitives::U256;
 
-use crate::core::opcodes::OpCodeInfo;
+use crate::core::opcodes::opcode_name;
 
 /// A WrappedInput can contain either a raw U256 value or a WrappedOpcode
 #[derive(Clone, Debug, PartialEq, Eq, Hash)]
@@ -30,7 +30,7 @@ impl std::fmt::Display for WrappedOpcode {
         write!(
             f,
             "{}({})",
-            OpCodeInfo::from(self.opcode).name(),
+            opcode_name(self.opcode),
             self.inputs.iter().map(|x| x.to_string()).collect::<Vec<_>>().join(", ")
         )
     }
diff --git a/crates/vm/src/core/vm.rs b/crates/vm/src/core/vm.rs
diff --git a/crates/vm/src/ext/lexers/yul.rs b/crates/vm/src/ext/lexers/yul.rs

Original file line number	Diff line number	Diff line change
`@@ -1,6 +1,6 @@`
`1`	`1`	`use alloy::primitives::U256;`
`2`	`2`
`3`		`-use crate::core::opcodes::OpCodeInfo;`
	`3`	`+use crate::core::opcodes::opcode_name;`
`4`	`4`
`5`	`5`	`/// A WrappedInput can contain either a raw U256 value or a WrappedOpcode`
`6`	`6`	`#[derive(Clone, Debug, PartialEq, Eq, Hash)]`
`@@ -30,7 +30,7 @@ impl std::fmt::Display for WrappedOpcode {`
`30`	`30`	`write!(`
`31`	`31`	`f,`
`32`	`32`	`"{}({})",`
`33`		`- OpCodeInfo::from(self.opcode).name(),`
	`33`	`+ opcode_name(self.opcode),`
`34`	`34`	`self.inputs.iter().map(\|x\| x.to_string()).collect::<Vec<_>>().join(", ")`
`35`	`35`	`)`
`36`	`36`	`}`