Merge pull request #80 from Chia-Network/20240406-assign-cse-fuzz

fix for ability to lift a CSE out of an assign form when it depends on bindings in the assign and fuzz infra for testing

resources/tests/cse-bad-letstar.clsp

@@ -0,0 +1,12 @@
+(mod (blob)
+  (include *standard-cl-23*)
+
+  (defun test-f (blob)
+    (let*
+      ((first (list 1 2 3))
+       (copies (list (r first) (r first))))
+       (c (r first) copies)
+      )
+    )
+  (test-f blob)
+)

resources/tests/cse-bad.clsp

@@ -0,0 +1,12 @@
+(mod (blob)
+  (include *standard-cl-23*)
+
+  (defun test-f (blob)
+    (assign
+      first (list 1 2 3)
+      copies (list (r first) (r first))
+      (c (r first) copies)
+      )
+    )
+  (test-f blob)
+)

resources/tests/cse-overlap.clsp

@@ -0,0 +1,15 @@
+(mod (Q)
+  (include *standard-cl-23*)
+
+  (defun F (X)
+    (assign
+      A1 (r (r X))
+      A2 (c A1 (r X))
+      A3 (c A2 (r (r X)))
+      A4 (assign B1 (c A1 (r (r X))) (list B1 (r (r X))))
+      (list A1 A2 A3 A4)
+      )
+    )
+
+  (F (list Q 99 100 101 102))
+  )

resources/tests/cse-tricky-basic.clsp

@@ -0,0 +1,25 @@
+(mod (move)
+  (include *standard-cl-23*)
+
+  (defun fhi (ha ca) (sha256 ha ca))
+  (defun nfh (ha ca) (fhi ha ca))
+
+  (defun testf1 (ca)
+    (assign-lambda
+      fc (+ ca 1)
+      sc (* ca 11)
+      (assign-lambda
+        tango 121
+        (if (= ca 13)
+          (nfh fc sc)
+          (if (= ca 15)
+            (fhi fc sc)
+            tango
+            )
+          )
+        )
+      )
+    )
+
+  (testf1 move)
+  )

src/compiler/compiler.rs

@@ -1,6 +1,6 @@
 use num_bigint::ToBigInt;
 use std::borrow::Borrow;
-use std::collections::HashMap;
+use std::collections::{HashMap, HashSet};
 use std::fs;
 use std::path::PathBuf;
 use std::rc::Rc;
@@ -22,6 +22,8 @@ use crate::compiler::srcloc::Srcloc;
 use crate::compiler::{BasicCompileContext, CompileContextWrapper};
 use crate::util::Number;
 
+pub const FUZZ_TEST_PRE_CSE_MERGE_FIX_FLAG: usize = 1;
+
 lazy_static! {
     pub static ref STANDARD_MACROS: String = {
         indoc! {"(
@@ -84,6 +86,7 @@ pub struct DefaultCompilerOpts {
     pub start_env: Option<Rc<SExp>>,
     pub disassembly_ver: Option<usize>,
     pub prim_map: Rc<HashMap<Vec<u8>, Rc<SExp>>>,
+    pub diag_flags: Rc<HashSet<usize>>,
     pub dialect: AcceptedDialect,
 }
 
@@ -208,6 +211,9 @@ impl CompilerOpts for DefaultCompilerOpts {
     fn get_search_paths(&self) -> Vec<String> {
         self.include_dirs.clone()
     }
+    fn diag_flags(&self) -> Rc<HashSet<usize>> {
+        self.diag_flags.clone()
+    }
 
     fn set_dialect(&self, dialect: AcceptedDialect) -> Rc<dyn CompilerOpts> {
         let mut copy = self.clone();
@@ -264,6 +270,11 @@ impl CompilerOpts for DefaultCompilerOpts {
         copy.prim_map = prims;
         Rc::new(copy)
     }
+    fn set_diag_flags(&self, flags: Rc<HashSet<usize>>) -> Rc<dyn CompilerOpts> {
+        let mut copy = self.clone();
+        copy.diag_flags = flags;
+        Rc::new(copy)
+    }
 
     fn read_new_file(
         &self,
@@ -330,6 +341,7 @@ impl DefaultCompilerOpts {
             dialect: AcceptedDialect::default(),
             prim_map: create_prim_map(),
             disassembly_ver: None,
+            diag_flags: Rc::new(HashSet::default()),
         }
     }
 }

src/compiler/comptypes.rs

@@ -420,6 +420,9 @@ pub trait CompilerOpts {
     fn prim_map(&self) -> Rc<HashMap<Vec<u8>, Rc<SExp>>>;
     /// Specifies the search paths we're carrying.
     fn get_search_paths(&self) -> Vec<String>;
+    /// Specifies flags that were passed down to various consumers.  This is
+    /// open ended for various purposes, such as diagnostics.
+    fn diag_flags(&self) -> Rc<HashSet<usize>>;
 
     /// Set the dialect.
     fn set_dialect(&self, dialect: AcceptedDialect) -> Rc<dyn CompilerOpts>;
@@ -444,6 +447,8 @@ pub trait CompilerOpts {
     fn set_start_env(&self, start_env: Option<Rc<SExp>>) -> Rc<dyn CompilerOpts>;
     /// Set the primitive map in use so we can add custom primitives.
     fn set_prim_map(&self, new_map: Rc<HashMap<Vec<u8>, Rc<SExp>>>) -> Rc<dyn CompilerOpts>;
+    /// Set the flags this CompilerOpts holds.  Consumers can examine these.
+    fn set_diag_flags(&self, new_flags: Rc<HashSet<usize>>) -> Rc<dyn CompilerOpts>;
 
     /// Using the search paths list we have, try to read a file by name,
     /// Returning the expanded path to the file and its content.
@@ -516,6 +521,9 @@ pub trait HasCompilerOptsDelegation {
     fn override_get_search_paths(&self) -> Vec<String> {
         self.compiler_opts().get_search_paths()
     }
+    fn override_diag_flags(&self) -> Rc<HashSet<usize>> {
+        self.compiler_opts().diag_flags()
+    }
 
     fn override_set_dialect(&self, dialect: AcceptedDialect) -> Rc<dyn CompilerOpts> {
         self.update_compiler_opts(|o| o.set_dialect(dialect))
@@ -547,6 +555,9 @@ pub trait HasCompilerOptsDelegation {
     fn override_set_start_env(&self, start_env: Option<Rc<SExp>>) -> Rc<dyn CompilerOpts> {
         self.update_compiler_opts(|o| o.set_start_env(start_env))
     }
+    fn override_set_diag_flags(&self, flags: Rc<HashSet<usize>>) -> Rc<dyn CompilerOpts> {
+        self.update_compiler_opts(|o| o.set_diag_flags(flags))
+    }
     fn override_set_prim_map(
         &self,
         new_map: Rc<HashMap<Vec<u8>, Rc<SExp>>>,
@@ -610,6 +621,9 @@ impl<T: HasCompilerOptsDelegation> CompilerOpts for T {
     fn get_search_paths(&self) -> Vec<String> {
         self.override_get_search_paths()
     }
+    fn diag_flags(&self) -> Rc<HashSet<usize>> {
+        self.override_diag_flags()
+    }
 
     fn set_dialect(&self, dialect: AcceptedDialect) -> Rc<dyn CompilerOpts> {
         self.override_set_dialect(dialect)
@@ -644,6 +658,9 @@ impl<T: HasCompilerOptsDelegation> CompilerOpts for T {
     fn set_prim_map(&self, new_map: Rc<HashMap<Vec<u8>, Rc<SExp>>>) -> Rc<dyn CompilerOpts> {
         self.override_set_prim_map(new_map)
     }
+    fn set_diag_flags(&self, new_flags: Rc<HashSet<usize>>) -> Rc<dyn CompilerOpts> {
+        self.override_set_diag_flags(new_flags)
+    }
     fn read_new_file(
         &self,
         inc_from: String,

src/compiler/frontend.rs

@@ -33,7 +33,7 @@ fn collect_used_names_binding(body: &Binding) -> Vec<Vec<u8>> {
     collect_used_names_bodyform(body.body.borrow())
 }
 
-fn collect_used_names_bodyform(body: &BodyForm) -> Vec<Vec<u8>> {
+pub fn collect_used_names_bodyform(body: &BodyForm) -> Vec<Vec<u8>> {
     match body {
         BodyForm::Let(_, letdata) => {
             let mut result = Vec::new();

src/compiler/optimize/above22.rs

@@ -6,6 +6,8 @@ use clvmr::allocator::Allocator;
 
 use crate::classic::clvm_tools::stages::stage_0::TRunProgram;
 
+#[cfg(any(test, feature = "fuzz"))]
+use crate::compiler::compiler::FUZZ_TEST_PRE_CSE_MERGE_FIX_FLAG;
 use crate::compiler::comptypes::{
     BodyForm, CompileErr, CompileForm, CompilerOpts, DefunData, HelperForm, PrimaryCodegen,
 };
@@ -25,6 +27,18 @@ use crate::compiler::StartOfCodegenOptimization;
 #[derive(Default, Clone)]
 pub struct Strategy23 {}
 
+#[cfg(any(test, feature = "fuzz"))]
+fn enable_cse_merge_fix_so_can_be_disabled_for_tests(opts: Rc<dyn CompilerOpts>) -> bool {
+    !opts
+        .diag_flags()
+        .contains(&FUZZ_TEST_PRE_CSE_MERGE_FIX_FLAG)
+}
+
+#[cfg(not(any(test, feature = "fuzz")))]
+fn enable_cse_merge_fix_so_can_be_disabled_for_tests(_opts: Rc<dyn CompilerOpts>) -> bool {
+    true
+}
+
 impl Strategy23 {
     pub fn new() -> Self {
         Strategy23 {}
@@ -36,13 +50,20 @@ impl Optimization for Strategy23 {
         &mut self,
         _allocator: &mut Allocator,
         _runner: Rc<dyn TRunProgram>,
-        _opts: Rc<dyn CompilerOpts>,
+        opts: Rc<dyn CompilerOpts>,
         mut p0: CompileForm,
     ) -> Result<CompileForm, CompileErr> {
         let mut rebuilt_helpers = Vec::new();
+        let enable_merge_disable_for_tests =
+            enable_cse_merge_fix_so_can_be_disabled_for_tests(opts.clone());
         for h in p0.helpers.iter() {
             if let HelperForm::Defun(inline, d) = h {
-                let function_body = cse_optimize_bodyform(&h.loc(), h.name(), d.body.borrow())?;
+                let function_body = cse_optimize_bodyform(
+                    &h.loc(),
+                    h.name(),
+                    enable_merge_disable_for_tests,
+                    d.body.borrow(),
+                )?;
                 // Ok we've got a body that is now a let stack.
                 let new_helper = HelperForm::Defun(
                     *inline,