implement calldatacopy

Signed-off-by: xermicus <cyrill@parity.io>

.gitignore

@@ -4,3 +4,5 @@
 .DS_Store
 /*.sol
 /*.yul
+/*.ll
+/*.s

Cargo.lock

@@ -55,6 +55,36 @@ dependencies = [
  "bytes",
 ]
 
+[[package]]
+name = "alloy-sol-macro"
+version = "0.6.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e86ec0a47740b20bc5613b8712d0d321d031c4efc58e9645af96085d5cccfc27"
+dependencies = [
+ "const-hex",
+ "dunce",
+ "heck 0.4.1",
+ "indexmap",
+ "proc-macro-error",
+ "proc-macro2",
+ "quote",
+ "syn 2.0.52",
+ "syn-solidity",
+ "tiny-keccak",
+]
+
+[[package]]
+name = "alloy-sol-types"
+version = "0.6.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ad09ec5853fa700d12d778ad224dcdec636af424d29fad84fb9a2f16a5b0ef09"
+dependencies = [
+ "alloy-primitives",
+ "alloy-sol-macro",
+ "const-hex",
+ "serde",
+]
+
 [[package]]
 name = "anyhow"
 version = "1.0.81"
@@ -470,6 +500,12 @@ dependencies = [
  "subtle",
 ]
 
+[[package]]
+name = "dunce"
+version = "1.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "56ce8c6da7551ec6c462cbaf3bfbc75131ebbfa1c944aeaa9dab51ca1c5f0c3b"
+
 [[package]]
 name = "ecdsa"
 version = "0.16.9"
@@ -711,6 +747,12 @@ dependencies = [
  "unicode-segmentation",
 ]
 
+[[package]]
+name = "heck"
+version = "0.4.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "95505c38b4572b2d910cecb0281560f54b440a19336cbbcb27bf6ce6adc6f5a8"
+
 [[package]]
 name = "hex"
 version = "0.4.3"
@@ -1378,6 +1420,7 @@ name = "revive-integration"
 version = "0.1.0"
 dependencies = [
  "alloy-primitives",
+ "alloy-sol-types",
  "era-compiler-llvm-context",
  "hex",
  "parity-scale-codec",
@@ -1737,7 +1780,7 @@ version = "0.4.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "dcb5ae327f9cc13b68763b5749770cb9e048a99bd9dfdfa58d0cf05d5f64afe0"
 dependencies = [
- "heck",
+ "heck 0.3.3",
  "proc-macro-error",
  "proc-macro2",
  "quote",
@@ -1772,6 +1815,18 @@ dependencies = [
  "unicode-ident",
 ]
 
+[[package]]
+name = "syn-solidity"
+version = "0.6.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cb3d0961cd53c23ea94eeec56ba940f636f6394788976e9f16ca5ee0aca7464a"
+dependencies = [
+ "paste",
+ "proc-macro2",
+ "quote",
+ "syn 2.0.52",
+]
+
 [[package]]
 name = "tap"
 version = "1.0.1"

Cargo.toml

@@ -31,6 +31,7 @@ polkavm-linker = { git = "https://github.com/koute/polkavm.git" }
 polkavm = { git = "https://github.com/koute/polkavm.git" } 
 parity-scale-codec = "3.6"
 alloy-primitives = "0.6"
+alloy-sol-types = "0.6"
 
 [workspace.dependencies.inkwell]
 git = "https://github.com/TheDan64/inkwell.git"

README.md

@@ -2,7 +2,7 @@
 
 YUL and EVM bytecode recompiler to LLVM, targetting RISC-V on PolkaVM.
 
-Code bases of [frontend](https://github.com/matter-labs/era-compiler-solidity) and [code generator](https://github.com/matter-labs/era-compiler-llvm-context) are forked adapted from ZKSync `zksolc`.
+Code bases of [frontend](https://github.com/matter-labs/era-compiler-solidity) and [code generator](https://github.com/matter-labs/era-compiler-llvm-context) are forked and adapted from ZKSync `zksolc`.
 
 # Status
 
@@ -12,8 +12,9 @@ Currently, primary goal of this codebase is to allow for benchmarks comparing pe
 
 The project is in a very early PoC phase. Don't yet expect the produced code to be working nor to be correct for anything more than a basic flipper contract at the current stage.
 
-- [ ] Efficient implementations of byte swaps, memset, memmove and the like
+- [ ] Efficient implementations of byte swaps, memset, memmove, mulmod and the like
 - [ ] Use `drink` for integration tests once we have 64bit support in PolkaVM
+- [ ] Use PolkaVM allocator for heap space
 - [ ] Exercice `schlau` and possibly `smart-bench` benchmark cases
 - [ ] Tests currently rely on the binary being in $PATH, which is very annoying and requires `cargo install` all the times
 - [ ] Define how to do deployments

crates/integration/Cargo.toml

@@ -11,4 +11,5 @@ polkavm = { workspace = true }
 parity-scale-codec = { workspace = true }
 revive-solidity = { path = "../solidity" }
 era-compiler-llvm-context = { path = "../llvm-context" }
-alloy-primitives = { workspace = true }
+alloy-primitives = { workspace = true }
+alloy-sol-types = { workspace = true }

crates/integration/src/lib.rs

@@ -27,7 +27,8 @@ pub fn compile_blob(contract_name: &str, source_code: &str) -> Vec<u8> {
 
 #[cfg(test)]
 mod tests {
-    use alloy_primitives::U256;
+    use alloy_primitives::{FixedBytes, Keccak256, I256, U256};
+    use alloy_sol_types::{sol, SolCall};
 
     use crate::mock_runtime::{self, State};
 
@@ -46,6 +47,37 @@ mod tests {
         assert_eq!(state.storage[&U256::ZERO], U256::ZERO);
     }
 
+    #[test]
+    fn hash_keccak_256() {
+        sol!(
+            #[derive(Debug, PartialEq, Eq)]
+            contract TestSha3 {
+                function test(string memory _pre) external payable returns (bytes32);
+            }
+        );
+        let source = r#"contract TestSha3 {
+            function test(string memory _pre) external payable returns (bytes32 hash) {
+                hash = keccak256(bytes(_pre));
+            }
+        }"#;
+        let code = crate::compile_blob("TestSha3", source);
+
+        let param = "hello";
+        let input = TestSha3::testCall::new((param.to_string(),)).abi_encode();
+
+        let state = State::new(input);
+        let (instance, export) = mock_runtime::prepare(&code);
+        let state = crate::mock_runtime::call(state, &instance, export);
+
+        assert_eq!(state.output.flags, 0);
+
+        let mut hasher = Keccak256::new();
+        hasher.update(param);
+        let expected = hasher.finalize();
+        let received = FixedBytes::<32>::from_slice(&state.output.data);
+        assert_eq!(expected, received);
+    }
+
     #[test]
     fn erc20() {
         let _ = crate::compile_blob("ERC20", include_str!("../contracts/ERC20.sol"));
@@ -54,8 +86,8 @@ mod tests {
     #[test]
     fn triangle_number() {
         let code = crate::compile_blob("Computation", include_str!("../contracts/Computation.sol"));
-        let param = alloy_primitives::U256::try_from(13).unwrap();
-        let expected = alloy_primitives::U256::try_from(91).unwrap();
+        let param = U256::try_from(13).unwrap();
+        let expected = U256::try_from(91).unwrap();
 
         // function triangle_number(int64)
         let mut input = 0x0f760610u32.to_be_bytes().to_vec();
@@ -67,16 +99,15 @@ mod tests {
 
         assert_eq!(state.output.flags, 0);
 
-        let received =
-            alloy_primitives::U256::from_be_bytes::<32>(state.output.data.try_into().unwrap());
+        let received = U256::from_be_bytes::<32>(state.output.data.try_into().unwrap());
         assert_eq!(received, expected);
     }
 
     #[test]
     fn odd_product() {
         let code = crate::compile_blob("Computation", include_str!("../contracts/Computation.sol"));
-        let param = alloy_primitives::I256::try_from(5i32).unwrap();
-        let expected = alloy_primitives::I256::try_from(945i64).unwrap();
+        let param = I256::try_from(5i32).unwrap();
+        let expected = I256::try_from(945i64).unwrap();
 
         // function odd_product(int32)
         let mut input = 0x00261b66u32.to_be_bytes().to_vec();
@@ -88,8 +119,7 @@ mod tests {
 
         assert_eq!(state.output.flags, 0);
 
-        let received =
-            alloy_primitives::I256::from_be_bytes::<32>(state.output.data.try_into().unwrap());
+        let received = I256::from_be_bytes::<32>(state.output.data.try_into().unwrap());
         assert_eq!(received, expected);
     }
 }

crates/integration/src/mock_runtime.rs

@@ -2,7 +2,7 @@
 //! TODO: Switch to drink! once RISCV is ready in polkadot-sdk
 use std::collections::HashMap;
 
-use alloy_primitives::U256;
+use alloy_primitives::{Keccak256, U256};
 use parity_scale_codec::Encode;
 use polkavm::{
     Caller, Config, Engine, ExportIndex, GasMeteringKind, InstancePre, Linker, Module,
@@ -168,6 +168,27 @@ fn link_host_functions(engine: &Engine) -> Linker<State> {
         )
         .unwrap();
 
+    linker
+        .func_wrap(
+            "hash_keccak_256",
+            |caller: Caller<State>,
+             input_ptr: u32,
+             input_len: u32,
+             out_ptr: u32|
+             -> Result<(), Trap> {
+                let (mut caller, _) = caller.split();
+
+                let pre = caller.read_memory_into_vec(input_ptr, input_len)?;
+
+                let mut hasher = Keccak256::new();
+                hasher.update(&pre);
+                caller.write_memory(out_ptr, &hasher.finalize()[..])?;
+
+                Ok(())
+            },
+        )
+        .unwrap();
+
     linker
 }
 

crates/llvm-context/src/eravm/context/mod.rs

@@ -671,15 +671,13 @@ where
                     "offset_ptrtoint",
                 )?;
                 let pointer_value = unsafe {
-                    self.builder
-                        .build_gep(
-                            self.byte_type(),
-                            heap_pointer.value.as_pointer_value(),
-                            &[offset],
-                            "heap_offset_via_gep",
-                        )
-                        .unwrap()
-                };
+                    self.builder.build_gep(
+                        self.byte_type(),
+                        heap_pointer.value.as_pointer_value(),
+                        &[offset],
+                        "heap_offset_via_gep",
+                    )
+                }?;
                 let value = self
                     .builder()
                     .build_load(pointer.r#type, pointer_value, name)?;
@@ -806,6 +804,7 @@ where
                         )
                         .unwrap()
                 };
+
                 let value = self.build_byte_swap(value.as_basic_value_enum());
 
                 let instruction = self.builder.build_store(pointer_value, value).unwrap();

crates/llvm-context/src/eravm/context/pointer.rs

@@ -78,6 +78,7 @@ impl<'ctx> Pointer<'ctx> {
             "Stack pointers cannot be addressed"
         );
 
+        let offset = context.safe_truncate_int_to_i32(offset).unwrap();
         let value = context
             .builder
             .build_int_to_ptr(

crates/llvm-context/src/eravm/evm/calldata.rs

@@ -59,29 +59,36 @@ pub fn copy<'ctx, D>(
 where
     D: Dependency + Clone,
 {
-    // TODO: Untested
-    let destination = Pointer::new_with_offset(
-        context,
-        AddressSpace::Heap,
-        context.byte_type(),
-        destination_offset,
-        "calldata_copy_destination_pointer",
-    );
+    let heap_pointer = context
+        .get_global(crate::eravm::GLOBAL_HEAP_MEMORY_POINTER)?
+        .value
+        .as_pointer_value();
+    let destination = unsafe {
+        context.builder().build_gep(
+            context.byte_type(),
+            heap_pointer,
+            &[destination_offset],
+            "calldata_pointer_with_offset",
+        )
+    }?;
+
     let calldata_pointer = context
         .get_global(crate::eravm::GLOBAL_CALLDATA_POINTER)?
         .value
         .as_pointer_value();
-    let source = context.build_gep(
-        Pointer::new(context.byte_type(), AddressSpace::Stack, calldata_pointer),
-        &[source_offset],
-        context.field_type().as_basic_type_enum(),
-        "calldata_pointer_with_offset",
-    );
+    let source = unsafe {
+        context.builder().build_gep(
+            context.byte_type(),
+            calldata_pointer,
+            &[source_offset],
+            "calldata_pointer_with_offset",
+        )
+    }?;
 
     context.build_memcpy(
         context.intrinsics().memory_copy_from_generic,
-        destination,
-        source,
+        Pointer::new(context.byte_type(), AddressSpace::Stack, destination),
+        Pointer::new(context.byte_type(), AddressSpace::Stack, source),
         size,
         "calldata_copy_memcpy_from_child",
     )

crates/llvm-context/src/eravm/evm/crypto.rs

@@ -9,12 +9,52 @@ use crate::eravm::Dependency;
 /// Translates the `sha3` instruction.
 ///
 pub fn sha3<'ctx, D>(
-    _context: &mut Context<'ctx, D>,
-    _offset: inkwell::values::IntValue<'ctx>,
-    _length: inkwell::values::IntValue<'ctx>,
+    context: &mut Context<'ctx, D>,
+    offset: inkwell::values::IntValue<'ctx>,
+    length: inkwell::values::IntValue<'ctx>,
 ) -> anyhow::Result<inkwell::values::BasicValueEnum<'ctx>>
 where
     D: Dependency + Clone,
 {
-    todo!()
+    let offset_casted = context.safe_truncate_int_to_i32(offset)?;
+    let heap_pointer = context.get_global(crate::eravm::GLOBAL_HEAP_MEMORY_POINTER)?;
+    let input_pointer = unsafe {
+        context.builder().build_gep(
+            context.byte_type(),
+            heap_pointer.value.as_pointer_value(),
+            &[offset_casted],
+            "heap_offset_via_gep",
+        )
+    }?;
+    let input_pointer_casted = context.builder().build_ptr_to_int(
+        input_pointer,
+        context.integer_type(32),
+        "input_pointer_casted",
+    )?;
+
+    let length_casted = context.safe_truncate_int_to_i32(length)?;
+
+    let output_pointer = context.build_alloca(context.field_type(), "output_pointer");
+    let output_pointer_casted = context.builder().build_ptr_to_int(
+        output_pointer.value,
+        context.integer_type(32),
+        "output_pointer_casted",
+    )?;
+
+    let function = context
+        .module()
+        .get_function("hash_keccak_256")
+        .expect("is declared");
+
+    context.builder().build_call(
+        function,
+        &[
+            input_pointer_casted.into(),
+            length_casted.into(),
+            output_pointer_casted.into(),
+        ],
+        "call_seal_hash_keccak_256",
+    )?;
+
+    Ok(context.build_byte_swap(context.build_load(output_pointer, "sha3_output")?))
 }

crates/pallet-contracts-pvm-llapi/src/polkavm_guest.c

@@ -12,6 +12,15 @@ void * memset(void *b, int c, size_t len) {
     return b;
 }
 
+void * memcpy(void *dst, const void *_src, size_t len) {
+    uint8_t *dest = dst;
+    const uint8_t *src = _src;
+
+    while (len--) *dest++ = *src++;
+
+    return dst;
+}
+
 
 // Exports