[compiler-v2] Add Bytecode verification at the end of compilation (ap…

…tos-labs#11923)

third_party/move/move-compiler-v2/Cargo.toml

@@ -18,6 +18,7 @@ move-command-line-common = { path = "../move-command-line-common" }
 #move-package = { path = "../tools/move-package" }
 move-compiler = { path = "../move-compiler" }
 move-core-types = { path = "../move-core/types" }
+move-disassembler = { path = "../tools/move-disassembler" }
 move-ir-types = { path = "../move-ir/types" }
 move-model = { path = "../move-model" }
 move-stackless-bytecode = { path = "../move-model/bytecode" }
@@ -42,7 +43,6 @@ serde = { version = "1.0.124", features = ["derive"] }
 anyhow = "1.0.52"
 datatest-stable = "0.1.1"
 move-command-line-common = { path = "../move-command-line-common" }
-move-disassembler = { path = "../tools/move-disassembler" }
 move-ir-types = { path = "../move-ir/types" }
 move-prover-test-utils = { path = "../move-prover/test-utils" }
 move-stdlib = { path = "../move-stdlib" }

third_party/move/move-compiler-v2/src/lib.rs

@@ -25,15 +25,19 @@ use anyhow::bail;
 use codespan_reporting::term::termcolor::{ColorChoice, StandardStream, WriteColor};
 pub use experiments::*;
 use log::{debug, info, log_enabled, trace, Level};
+use move_binary_format::binary_views::BinaryIndexedView;
+use move_command_line_common::files::FileHash;
 use move_compiler::{
  compiled_unit::{
- AnnotatedCompiledModule, AnnotatedCompiledScript, AnnotatedCompiledUnit, CompiledUnit,
- FunctionInfo,
+ verify_units, AnnotatedCompiledModule, AnnotatedCompiledScript, AnnotatedCompiledUnit,
+ CompiledUnit, FunctionInfo,
  },
  diagnostics::FilesSourceText,
  shared::{known_attributes::KnownAttribute, unique_map::UniqueMap},
 };
-use move_model::{model::GlobalEnv, PackageInfo};
+use move_disassembler::disassembler::Disassembler;
+use move_ir_types::location;
+use move_model::{add_move_lang_diagnostics, model::GlobalEnv, PackageInfo};
 use move_stackless_bytecode::function_target_pipeline::{
  FunctionTargetPipeline, FunctionTargetsHolder, FunctionVariant,
 };
@@ -60,7 +64,7 @@ pub fn run_move_compiler(
  let mut env = run_checker(options.clone())?;
  check_errors(&env, error_writer, "checking errors")?;
 
- trace!("After context check, GlobalEnv={}", env.dump_env());
+ trace!("After context check, GlobalEnv=\n{}", env.dump_env());
 
  // Flow-insensitive checks on AST
  flow_insensitive_checkers::check_for_unused_vars_and_params(&mut env);
@@ -69,15 +73,15 @@ pub fn run_move_compiler(
  check_errors(&env, error_writer, "checking errors")?;
 
  trace!(
- "After flow-insensitive checks, GlobalEnv={}",
+ "After flow-insensitive checks, GlobalEnv=\n{}",
  env.dump_env()
  );
 
  // Run inlining.
  inliner::run_inlining(&mut env);
  check_errors(&env, error_writer, "inlining")?;
 
- debug!("After inlining, GlobalEnv={}", env.dump_env());
+ debug!("After inlining, GlobalEnv=\n{}", env.dump_env());
 
  // Run code generator
  let mut targets = run_bytecode_gen(&env);
@@ -110,8 +114,17 @@ pub fn run_move_compiler(
 
  let modules_and_scripts = run_file_format_gen(&env, &targets);
  check_errors(&env, error_writer, "assembling errors")?;
- let annotated = annotate_units(modules_and_scripts);
- Ok((env, annotated))
+
+ debug!(
+ "File format bytecode:\n{}",
+ disassemble_compiled_units(&modules_and_scripts)?
+ );
+
+ let annotated_units = annotate_units(modules_and_scripts);
+ run_bytecode_verifier(&annotated_units, &mut env);
+ check_errors(&env, error_writer, "bytecode verification errors")?;
+
+ Ok((env, annotated_units))
 }
 
 /// Run the type checker and return the global env (with errors if encountered). The result
@@ -151,7 +164,7 @@ pub fn run_checker(options: Options) -> anyhow::Result<GlobalEnv> {
 // compilation, create an entry in the functions target holder which encapsulate info
 // like the generated bytecode.
 pub fn run_bytecode_gen(env: &GlobalEnv) -> FunctionTargetsHolder {
- info!("Bytecode Generation");
+ info!("Stackless bytecode Generation");
  let mut targets = FunctionTargetsHolder::default();
  let mut todo = BTreeSet::new();
  let mut done = BTreeSet::new();
@@ -232,6 +245,30 @@ fn add_default_optimization_pipeline(pipeline: &mut FunctionTargetPipeline) {
  pipeline.add_processor(Box::new(UnreachableCodeRemover {}));
 }
 
+/// Disassemble the given compiled units and return the disassembled code as a string.
+pub fn disassemble_compiled_units(units: &[CompiledUnit]) -> anyhow::Result<String> {
+ let disassembled_units: anyhow::Result<Vec<_>> = units
+ .iter()
+ .map(|unit| {
+ let view = match unit {
+ CompiledUnit::Module(module) => BinaryIndexedView::Module(&module.module),
+ CompiledUnit::Script(script) => BinaryIndexedView::Script(&script.script),
+ };
+ Disassembler::from_view(view, location::Loc::new(FileHash::empty(), 0, 0))
+ .and_then(|d| d.disassemble())
+ })
+ .collect();
+ Ok(disassembled_units?.concat())
+}
+
+/// Run the bytecode verifier on the given compiled units and add any diagnostics to the global env.
+pub fn run_bytecode_verifier(units: &[AnnotatedCompiledUnit], env: &mut GlobalEnv) {
+ let diags = verify_units(units);
+ if !diags.is_empty() {
+ add_move_lang_diagnostics(env, diags);
+ }
+}
+
 /// Report any diags in the env to the writer and fail if there are errors.
 pub fn check_errors<W: WriteColor>(
  env: &GlobalEnv,