fix(ssa refactor): more comprehensive instruction simplification (#1735)

* chore(ssa refactor): more comprehensive instruction simplification

* chore(ssa refactor): cp working test

* chore(ssa refactor): allow HO functions; tidy

* chore(ssa refactor): also now works

* chore(ssa refactor): PR feedback

* Update crates/noirc_evaluator/src/ssa_refactor/ir/instruction.rs

---------

Co-authored-by: jfecher <[email protected]>

crates/nargo_cli/tests/test_data_ssa_refactor/6/Nargo.toml

@@ -0,0 +1,5 @@
+[package]
+authors = [""]
+compiler_version = "0.1"
+
+[dependencies]

crates/nargo_cli/tests/test_data_ssa_refactor/6/Prover.toml

@@ -0,0 +1,39 @@
+
+# hello as bytes
+# used : https://emn178.github.io/online-tools/sha256.html
+x = [104, 101, 108, 108, 111]
+
+result = [
+        0x2c,
+        0xf2,
+        0x4d,
+        0xba,
+        0x5f,
+        0xb0,
+        0xa3,
+        0x0e,
+        0x26,
+        0xe8,
+        0x3b,
+        0x2a,
+        0xc5,
+        0xb9,
+        0xe2,
+        0x9e,
+        0x1b,
+        0x16,
+        0x1e,
+        0x5c,
+        0x1f,
+        0xa7,
+        0x42,
+        0x5e,
+        0x73,
+        0x04,
+        0x33,
+        0x62,
+        0x93,
+        0x8b,
+        0x98,
+        0x24,
+]

crates/nargo_cli/tests/test_data_ssa_refactor/6/src/main.nr

@@ -0,0 +1,20 @@
+// Sha256 circuit where the input is 5 bytes
+// not five field elements since sha256 operates over
+// bytes.
+//
+// If you do not cast, it will take all the bytes from the field element!
+
+// Mimc input is an array of field elements
+// The function is called mimc_bn254 to emphasize its parameters are chosen for bn254 curve, it should be used only with a proving system using the same curve (e.g Plonk from Aztec)
+use dep::std;
+
+fn main(x: [u8; 5], result: pub [u8; 32]) {
+    let mut digest = std::hash::sha256(x);
+    digest[0] = 5 as u8;
+    digest = std::hash::sha256(x);
+    assert(digest == result);
+
+    let y = [12,45,78,41];
+    let h = std::hash::mimc_bn254(y);
+   assert(h == 18226366069841799622585958305961373004333097209608110160936134895615261821931);
+}

crates/nargo_cli/tests/test_data_ssa_refactor/merkle_insert/Nargo.toml

@@ -0,0 +1,7 @@
+
+        [package]
+        authors = [""]
+        compiler_version = "0.1"
+
+        [dependencies]
+

crates/nargo_cli/tests/test_data_ssa_refactor/merkle_insert/Prover.toml

@@ -0,0 +1,11 @@
+old_root = "0x285785b10eca49cf456b935f1c9787ff571f306c1bc62549c31a9199a633f9f8"
+old_leaf = "0x1cdcf02431ba623767fe389337d011df1048dcc24b98ed81cec97627bab454a0"
+old_hash_path = [
+    "0x1cdcf02431ba623767fe389337d011df1048dcc24b98ed81cec97627bab454a0",
+    "0x0b5e9666e7323ce925c28201a97ddf4144ac9d148448ed6f49f9008719c1b85b",
+    "0x22ec636f8ad30ef78c42b7fe2be4a4cacf5a445cfb5948224539f59a11d70775",
+]
+new_root = "0x2d05c2650e6c2ef02c6dc7fae7f517b8ac191386666c0b5a68130a8c11092f5f"
+leaf = "0x085ca53be9c9d95b57e6e5fc91c5d531ad9e63e85dd71af7e35562991774b435"
+index = "0"
+mimc_input = [12,45,78,41]

crates/nargo_cli/tests/test_data_ssa_refactor/merkle_insert/src/main.nr

@@ -0,0 +1,21 @@
+use dep::std;
+
+fn main(
+    old_root: Field,
+    old_leaf: Field,
+    old_hash_path: [Field; 3],
+    new_root: pub Field,
+    leaf: Field,
+    index: Field,
+    mimc_input: [Field; 4],
+) {
+    assert(old_root == std::merkle::compute_merkle_root(old_leaf, index, old_hash_path));
+
+    let calculated_root = std::merkle::compute_merkle_root(leaf, index, old_hash_path);
+    assert(new_root == calculated_root);
+
+    let h = std::hash::mimc_bn254(mimc_input);
+    // Regression test for PR #891
+    std::println(h);
+    assert(h == 18226366069841799622585958305961373004333097209608110160936134895615261821931);
+}

crates/noirc_evaluator/src/ssa_refactor/ir/instruction.rs

@@ -1,3 +1,5 @@
+use std::rc::Rc;
+
 use acvm::{acir::BlackBoxFunc, FieldElement};
 use iter_extended::vecmap;
 use num_bigint::BigUint;
@@ -239,33 +241,7 @@ impl Instruction {
         use SimplifyResult::*;
         match self {
             Instruction::Binary(binary) => binary.simplify(dfg),
-            Instruction::Cast(value, typ) => {
-                if let Some(constant) = dfg.get_numeric_constant(*value) {
-                    let src_typ = dfg.type_of_value(*value);
-                    match (typ, src_typ) {
-                        (
-                            Type::Numeric(NumericType::Unsigned { bit_size }),
-                            Type::Numeric(NumericType::Unsigned { .. }),
-                        )
-                        | (
-                            Type::Numeric(NumericType::Unsigned { bit_size }),
-                            Type::Numeric(NumericType::NativeField),
-                        ) => {
-                            let integer_modulus = BigUint::from(2u128).pow(*bit_size);
-                            let constant: BigUint = BigUint::from_bytes_be(&constant.to_be_bytes());
-                            let truncated = constant % integer_modulus;
-                            let truncated =
-                                FieldElement::from_be_bytes_reduce(&truncated.to_bytes_be());
-                            SimplifiedTo(dfg.make_constant(truncated, typ.clone()))
-                        }
-                        _ => None,
-                    }
-                } else if let Some(value) = (*typ == dfg.type_of_value(*value)).then_some(*value) {
-                    SimplifiedTo(value)
-                } else {
-                    None
-                }
-            }
+            Instruction::Cast(value, typ) => simplify_cast(*value, typ, dfg),
             Instruction::Not(value) => {
                 match &dfg[dfg.resolve(*value)] {
                     // Limit optimizing ! on constants to only booleans. If we tried it on fields,
@@ -329,7 +305,7 @@ impl Instruction {
                     None
                 }
             }
-            Instruction::Call { .. } => None,
+            Instruction::Call { func, arguments } => simplify_call(*func, arguments, dfg),
             Instruction::Allocate { .. } => None,
             Instruction::Load { .. } => None,
             Instruction::Store { .. } => None,
@@ -338,6 +314,110 @@ impl Instruction {
     }
 }
 
+/// Try to simplify this cast instruction. If the instruction can be simplified to a known value,
+/// that value is returned. Otherwise None is returned.
+fn simplify_cast(value: ValueId, dst_typ: &Type, dfg: &mut DataFlowGraph) -> SimplifyResult {
+    use SimplifyResult::*;
+    if let Some(constant) = dfg.get_numeric_constant(value) {
+        let src_typ = dfg.type_of_value(value);
+        match (src_typ, dst_typ) {
+            (Type::Numeric(NumericType::NativeField), Type::Numeric(NumericType::NativeField)) => {
+                // Field -> Field: use src value
+                SimplifiedTo(value)
+            }
+            (
+                Type::Numeric(NumericType::Unsigned { .. }),
+                Type::Numeric(NumericType::NativeField),
+            ) => {
+                // Unsigned -> Field: redefine same constant as Field
+                SimplifiedTo(dfg.make_constant(constant, dst_typ.clone()))
+            }
+            (
+                Type::Numeric(NumericType::NativeField | NumericType::Unsigned { .. }),
+                Type::Numeric(NumericType::Unsigned { bit_size }),
+            ) => {
+                // Field/Unsigned -> unsigned: truncate
+                let integer_modulus = BigUint::from(2u128).pow(*bit_size);
+                let constant: BigUint = BigUint::from_bytes_be(&constant.to_be_bytes());
+                let truncated = constant % integer_modulus;
+                let truncated = FieldElement::from_be_bytes_reduce(&truncated.to_bytes_be());
+                SimplifiedTo(dfg.make_constant(truncated, dst_typ.clone()))
+            }
+            _ => None,
+        }
+    } else if *dst_typ == dfg.type_of_value(value) {
+        SimplifiedTo(value)
+    } else {
+        None
+    }
+}
+
+/// Try to simplify this call instruction. If the instruction can be simplified to a known value,
+/// that value is returned. Otherwise None is returned.
+fn simplify_call(func: ValueId, arguments: &[ValueId], dfg: &mut DataFlowGraph) -> SimplifyResult {
+    use SimplifyResult::*;
+    let intrinsic = match &dfg[func] {
+        Value::Intrinsic(intrinsic) => *intrinsic,
+        _ => return None,
+    };
+    let constant_args: Option<Vec<_>> =
+        arguments.iter().map(|value_id| dfg.get_numeric_constant(*value_id)).collect();
+    let constant_args = match constant_args {
+        Some(constant_args) => constant_args,
+        Option::None => return None,
+    };
+    match intrinsic {
+        Intrinsic::ToBits(endian) => {
+            let field = constant_args[0];
+            let limb_count = constant_args[1].to_u128() as u32;
+            SimplifiedTo(constant_to_radix(endian, field, 2, limb_count, dfg))
+        }
+        Intrinsic::ToRadix(endian) => {
+            let field = constant_args[0];
+            let radix = constant_args[1].to_u128() as u32;
+            let limb_count = constant_args[1].to_u128() as u32;
+            SimplifiedTo(constant_to_radix(endian, field, radix, limb_count, dfg))
+        }
+        Intrinsic::BlackBox(_) | Intrinsic::Println | Intrinsic::Sort => None,
+    }
+}
+
+/// Returns a Value::Array of constants corresponding to the limbs of the radix decomposition.
+fn constant_to_radix(
+    endian: Endian,
+    field: FieldElement,
+    radix: u32,
+    limb_count: u32,
+    dfg: &mut DataFlowGraph,
+) -> ValueId {
+    let bit_size = u32::BITS - (radix - 1).leading_zeros();
+    let radix_big = BigUint::from(radix);
+    assert_eq!(BigUint::from(2u128).pow(bit_size), radix_big, "ICE: Radix must be a power of 2");
+    let big_integer = BigUint::from_bytes_be(&field.to_be_bytes());
+
+    // Decompose the integer into its radix digits in little endian form.
+    let decomposed_integer = big_integer.to_radix_le(radix);
+    let mut limbs = vecmap(0..limb_count, |i| match decomposed_integer.get(i as usize) {
+        Some(digit) => FieldElement::from_be_bytes_reduce(&[*digit]),
+        None => FieldElement::zero(),
+    });
+    if endian == Endian::Big {
+        limbs.reverse();
+    }
+
+    // For legacy reasons (see #617) the to_radix interface supports 256 bits even though
+    // FieldElement::max_num_bits() is only 254 bits. Any limbs beyond the specified count
+    // become zero padding.
+    let max_decomposable_bits: u32 = 256;
+    let limb_count_with_padding = max_decomposable_bits / bit_size;
+    while limbs.len() < limb_count_with_padding as usize {
+        limbs.push(FieldElement::zero());
+    }
+    let result_constants =
+        limbs.into_iter().map(|limb| dfg.make_constant(limb, Type::unsigned(bit_size))).collect();
+    dfg.make_array(result_constants, Rc::new(vec![Type::unsigned(bit_size)]))
+}
+
 /// The possible return values for Instruction::return_types
 pub(crate) enum InstructionResultType {
     /// The result type of this instruction matches that of this operand