Skip to content

Add Memory utility library #5189

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 27 commits into from
Jul 10, 2025
Merged
Show file tree
Hide file tree
Changes from 19 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions .changeset/dull-students-eat.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
---
'openzeppelin-solidity': minor
---

`Memory`: Add library with utilities to manipulate memory
1 change: 1 addition & 0 deletions contracts/mocks/Stateless.sol
Original file line number Diff line number Diff line change
Expand Up @@ -46,6 +46,7 @@ import {SignatureChecker} from "../utils/cryptography/SignatureChecker.sol";
import {SignedMath} from "../utils/math/SignedMath.sol";
import {StorageSlot} from "../utils/StorageSlot.sol";
import {Strings} from "../utils/Strings.sol";
import {Memory} from "../utils/Memory.sol";
import {Time} from "../utils/types/Time.sol";

contract Dummy1234 {}
104 changes: 104 additions & 0 deletions contracts/utils/Memory.sol
Original file line number Diff line number Diff line change
@@ -0,0 +1,104 @@
// SPDX-License-Identifier: MIT

pragma solidity ^0.8.20;

/**
* @dev Utilities to manipulate memory.
*
* Memory is a contiguous and dynamic byte array in which Solidity stores non-primitive types.
* This library provides functions to manipulate pointers to this dynamic array.
*
* WARNING: When manipulating memory, make sure to follow the Solidity documentation
* guidelines for https://docs.soliditylang.org/en/v0.8.20/assembly.html#memory-safety[Memory Safety].
*/
library Memory {
type Pointer is bytes32;

/// @dev Returns a `Pointer` to the current free `Pointer`.
function getFreePointer() internal pure returns (Pointer ptr) {
assembly ("memory-safe") {
ptr := mload(0x40)
}
}

/// @dev Sets the free `Pointer` to a specific value.
///
/// WARNING: Everything after the pointer may be overwritten.
function setFreePointer(Pointer ptr) internal pure {
assembly ("memory-safe") {
mstore(0x40, ptr)
}
}

/// @dev Returns a `Pointer` to the content of a `bytes` buffer. Skips the length word.
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Everything from here onward (line 33 to 103) was recently added with no discussion about the usecase. I was personally happy with having utils to "cleanup" memory manually, which AFAIK was the original idea behind this PR.

Some of the things here ressemble `Bytes._unsafeReadBytesOffset, which was left private (and not internal) because we could not guarantee the memory safety. This library goes way furter, with assembly marked memory safe when its not something we know (depends on the user input).

Using the copy function, its quite easy to break stuff by writting to the wrong locations (for example to 0x60).

Copy link
Member Author

@ernestognw ernestognw Jun 10, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The background for these changes is that they're used by the RLP library and we previously kept private:
0332ffe#diff-0399261b2cd954212dda0147b86044380b879ab21e1d1b89c564cb48f3552216L304-L328

I agree with your take. However I think it's worth pursuing this pattern since it composes pretty well imo (as long as we add proper WARNINGS and docs). See how it looks in some places in RLP:

function readRawBytes(Item memory item) internal pure returns (bytes memory) {
    uint256 itemLength = item.length;
    bytes memory result = new bytes(itemLength);
    result.contentPointer().copy(item.ptr, itemLength);

    return result;
}

I would be fine removing these functions. Want to hear your thoughts on the pattern

Copy link

@0xClandestine 0xClandestine Jul 4, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Rather than exposing these in Memory when they're only used in the RLP library, maybe make them private within the RLP library?

Would also consider constraining destPtr to be greater than fmp offset.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@Amxx Worth noting the original proveth implementation does essentially the same, but manually, without mcopy.

  /*
  * @param src Pointer to source
  * @param dest Pointer to destination
  * @param len Amount of memory to copy from the source
  */
  function copy(uint src, uint dest, uint len) private pure {
      if (len == 0) return;

      // copy as many word sizes as possible
      for (; len >= WORD_SIZE; len -= WORD_SIZE) {
          assembly {
              mstore(dest, mload(src))
          }

          src += WORD_SIZE;
          dest += WORD_SIZE;
      }

      // left over bytes. Mask is used to remove unwanted bytes from the word
      uint mask = 256 ** (WORD_SIZE - len) - 1;
      assembly {
          let srcpart := and(mload(src), not(mask)) // zero out src
          let destpart := and(mload(dest), mask) // retrieve the bytes
          mstore(dest, or(destpart, srcpart))
      }
  }

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I removed all new controversial functions. Opened #5792 instead. Also updated the RLP PR (#5680) accordingly as suggested @0xClandestine

function contentPointer(bytes memory buffer) internal pure returns (Pointer) {
return addOffset(asPointer(buffer), 32);
}

/**
* @dev Copies `length` bytes from `srcPtr` to `destPtr`. Equivalent to https://www.evm.codes/?fork=cancun#5e[`mcopy`].
*
* WARNING: Reading or writing beyond the allocated memory bounds of either pointer
* will result in undefined behavior and potential memory corruption.
*/
function copy(Pointer destPtr, Pointer srcPtr, uint256 length) internal pure {
assembly ("memory-safe") {
mcopy(destPtr, srcPtr, length)
}
}

/**
* @dev Extracts a `bytes1` from a `Pointer`. `offset` starts from the most significant byte.
*
* NOTE: Will return `0x00` if `offset` is larger or equal to `32`.
*/
function loadByte(Pointer ptr, uint256 offset) internal pure returns (bytes1 v) {
bytes32 word = load(ptr);
assembly ("memory-safe") {
v := byte(offset, word)
}
}

/// @dev Extracts a `bytes32` from a `Pointer`.
function load(Pointer ptr) internal pure returns (bytes32 v) {
assembly ("memory-safe") {
v := mload(ptr)
}
}

/// @dev Adds an offset to a `Pointer`.
function addOffset(Pointer ptr, uint256 offset) internal pure returns (Pointer) {
return asPointer(bytes32(asUint256(ptr) + offset));
}

/// @dev `Pointer` to `bytes32`. Expects a pointer to a properly ABI-encoded `bytes` object.
function asBytes32(Pointer ptr) internal pure returns (bytes32) {
return Pointer.unwrap(ptr);
}

/// @dev `Pointer` to `uint256`. Expects a pointer to a properly ABI-encoded `bytes` object.
function asUint256(Pointer ptr) internal pure returns (uint256) {
return uint256(asBytes32(ptr));
}

/// @dev `bytes32` to `Pointer`. Expects a pointer to a properly ABI-encoded `bytes` object.
function asPointer(bytes32 value) internal pure returns (Pointer) {
return Pointer.wrap(value);
}

/// @dev Returns a `Pointer` to the `value`'s header (i.e. includes the length word).
function asPointer(bytes memory value) internal pure returns (Pointer) {
bytes32 ptr;
assembly ("memory-safe") {
ptr := value
}
return asPointer(ptr);
}

/// @dev `Pointer` to `bytes`. Expects a pointer to a properly ABI-encoded `bytes` object.
function asBytes(Pointer ptr) internal pure returns (bytes memory b) {
assembly ("memory-safe") {
b := ptr
}
}
}
3 changes: 3 additions & 0 deletions contracts/utils/README.adoc
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,7 @@ Miscellaneous contracts and libraries containing utility functions you can use t
* {Panic}: A library to revert with https://docs.soliditylang.org/en/v0.8.20/control-structures.html#panic-via-assert-and-error-via-require[Solidity panic codes].
* {Comparators}: A library that contains comparator functions to use with the {Heap} library.
* {CAIP2}, {CAIP10}: Libraries for formatting and parsing CAIP-2 and CAIP-10 identifiers.
* {Memory}: A utility library to manipulate memory.
* {Blockhash}: A library for accessing historical block hashes beyond the standard 256 block limit utilizing EIP-2935's historical blockhash functionality.
* {Time}: A library that provides helpers for manipulating time-related objects, including a `Delay` type.

Expand Down Expand Up @@ -134,6 +135,8 @@ Ethereum contracts have no native concept of an interface, so applications must

{{CAIP10}}

{{Memory}}

{{Blockhash}}

{{Time}}
51 changes: 50 additions & 1 deletion docs/modules/ROOT/pages/utilities.adoc
Original file line number Diff line number Diff line change
Expand Up @@ -263,7 +263,7 @@ Some use cases require more powerful data structures than arrays and mappings of
- xref:api:utils.adoc#EnumerableSet[`EnumerableSet`]: A https://en.wikipedia.org/wiki/Set_(abstract_data_type)[set] with enumeration capabilities.
- xref:api:utils.adoc#EnumerableMap[`EnumerableMap`]: A `mapping` variant with enumeration capabilities.
- xref:api:utils.adoc#MerkleTree[`MerkleTree`]: An on-chain https://wikipedia.org/wiki/Merkle_Tree[Merkle Tree] with helper functions.
- xref:api:utils.adoc#Heap.sol[`Heap`]: A
- xref:api:utils.adoc#Heap.sol[`Heap`]: A https://en.wikipedia.org/wiki/Binary_heap[binary heap] to store elements with priority defined by a compartor function.

The `Enumerable*` structures are similar to mappings in that they store and remove elements in constant time and don't allow for repeated entries, but they also support _enumeration_, which means you can easily query all stored entries both on and off-chain.

Expand Down Expand Up @@ -461,6 +461,55 @@ await instance.multicall([
]);
----

=== Memory

The xref:api:utils.adoc#Memory[`Memory`] library provides functions for advanced use cases that require granular memory management. A common use case is to avoid unnecessary memory expansion costs when performing repeated operations that allocate memory in a loop. Consider the following example:

[source,solidity]
----
function processMultipleItems(uint256[] memory items) internal {
for (uint256 i = 0; i < items.length; i++) {
bytes memory tempData = abi.encode(items[i], block.timestamp);
// Process tempData...
}
}
----

Note that each iteration allocates new memory for `tempData`, causing the memory to expand continuously. This can be optimized by resetting the memory pointer between iterations:

[source,solidity]
----
function processMultipleItems(uint256[] memory items) internal {
Memory.Pointer ptr = Memory.getFreePointer(); // Cache pointer
for (uint256 i = 0; i < items.length; i++) {
bytes memory tempData = abi.encode(items[i], block.timestamp);
// Process tempData...
Memory.setFreePointer(ptr); // Reset pointer for reuse
}
}
----

This way, memory allocated for `tempData` in each iteration is reused, significantly reducing memory expansion costs when processing many items.

==== Copying memory buffers

The `Memory` library provides a `copy` function that allows copying data between memory locations. This is useful when you need to extract a segment of data from a larger buffer or when you want to avoid unnecessary memory allocations. The following example demonstrates how to copy a segment of data from a source buffer:

[source,solidity]
----
function copyDataSegment(bytes memory source, uint256 offset, uint256 length)
internal pure returns (bytes memory result) {

result = new bytes(length);
Memory.Pointer srcPtr = Memory.addOffset(Memory.contentPointer(source), offset);
Memory.Pointer destPtr = Memory.contentPointer(result);

Memory.copy(destPtr, srcPtr, length);
}
----

IMPORTANT: Manual memory management increases gas costs and prevents compiler optimizations. Only use these functions after profiling confirms they're necessary. By default, Solidity handles memory safely - using this library without understanding memory layout and safety may be dangerous. See the https://docs.soliditylang.org/en/v0.8.20/internals/layout_in_memory.html[memory layout] and https://docs.soliditylang.org/en/v0.8.20/assembly.html#memory-safety[memory safety] documentation for details.

=== Historical Block Hashes

xref:api:utils.adoc#Blockhash[`Blockhash`] provides L2 protocol developers with extended access to historical block hashes beyond Ethereum's native 256-block limit. By leveraging https://eips.ethereum.org/EIPS/eip-2935[EIP-2935]'s history storage contract, the library enables access to block hashes up to 8,191 blocks in the past, making it invaluable for L2 fraud proofs and state verification systems.
Expand Down
67 changes: 67 additions & 0 deletions test/utils/Memory.t.sol
Original file line number Diff line number Diff line change
@@ -0,0 +1,67 @@
// SPDX-License-Identifier: MIT

pragma solidity ^0.8.20;

import {Test} from "forge-std/Test.sol";
import {Memory} from "@openzeppelin/contracts/utils/Memory.sol";

contract MemoryTest is Test {
using Memory for *;

// - first 0x80 bytes are reserved (scratch + FMP + zero)
uint256 constant START_PTR = 0x80;
// - moving the free memory pointer to far causes OOG errors
uint256 constant END_PTR = type(uint24).max;

function testGetSetFreePointer(uint256 seed) public pure {
bytes32 ptr = bytes32(bound(seed, START_PTR, END_PTR));
ptr.asPointer().setFreePointer();
assertEq(Memory.getFreePointer().asBytes32(), ptr);
}

function testSymbolicContentPointer(uint256 seed) public pure {
Memory.Pointer ptr = bytes32(bound(seed, START_PTR, END_PTR)).asPointer();
assertEq(ptr.asBytes().contentPointer().asBytes32(), ptr.addOffset(32).asBytes32());
}

function testCopy(bytes memory data, uint256 destSeed) public pure {
uint256 minDestPtr = Memory.getFreePointer().asUint256();
Memory.Pointer destPtr = bytes32(bound(destSeed, minDestPtr, minDestPtr + END_PTR)).asPointer();
destPtr.addOffset(data.length + 32).setFreePointer();
destPtr.copy(data.asPointer(), data.length + 32);
bytes memory copiedData = destPtr.asBytes();
assertEq(data.length, copiedData.length);
for (uint256 i = 0; i < data.length; i++) {
assertEq(data[i], copiedData[i]);
}
}

function testLoadByte(uint256 seed, uint256 index, bytes32 value) public pure {
index = bound(index, 0, 31);
Memory.Pointer ptr = bytes32(bound(seed, START_PTR, END_PTR)).asPointer();

assembly ("memory-safe") {
mstore(ptr, value)
}

bytes1 expected;
assembly ("memory-safe") {
expected := byte(index, value)
}
assertEq(ptr.loadByte(index), expected);
}

function testLoad(uint256 seed, bytes32 value) public pure {
Memory.Pointer ptr = bytes32(bound(seed, START_PTR, END_PTR)).asPointer();
assembly ("memory-safe") {
mstore(ptr, value)
}
assertEq(ptr.load(), value);
}

function testSymbolicAddOffset(uint256 seed, uint256 offset) public pure {
offset = bound(offset, 0, type(uint256).max - END_PTR);
Memory.Pointer ptr = bytes32(bound(seed, START_PTR, END_PTR)).asPointer();
assertEq(ptr.addOffset(offset).asUint256(), ptr.asUint256() + offset);
}
}
91 changes: 91 additions & 0 deletions test/utils/Memory.test.js
Original file line number Diff line number Diff line change
@@ -0,0 +1,91 @@
const { ethers } = require('hardhat');
const { expect } = require('chai');
const { loadFixture } = require('@nomicfoundation/hardhat-network-helpers');

async function fixture() {
const mock = await ethers.deployContract('$Memory');

return { mock };
}

describe('Memory', function () {
beforeEach(async function () {
Object.assign(this, await loadFixture(fixture));
});

describe('free pointer', function () {
it('sets free memory pointer', async function () {
const ptr = ethers.toBeHex(0xa0, 32);
await expect(this.mock.$setFreePointer(ptr)).to.not.be.reverted;
});

it('gets free memory pointer', async function () {
await expect(this.mock.$getFreePointer()).to.eventually.equal(
ethers.toBeHex(0x80, 32), // Default pointer
);
});
});

it('load extracts a word', async function () {
const ptr = await this.mock.$getFreePointer();
await expect(this.mock.$load(ptr)).to.eventually.equal(ethers.toBeHex(0, 32));
});

it('loadByte extracts a byte', async function () {
const ptr = await this.mock.$getFreePointer();
await expect(this.mock.$loadByte(ptr, 0)).to.eventually.equal(ethers.toBeHex(0, 1));
});

it('contentPointer', async function () {
const data = ethers.toUtf8Bytes('hello world');
const result = await this.mock.$contentPointer(data);
expect(result).to.equal(ethers.toBeHex(0xa0, 32)); // 0x80 is the default free pointer (length)
});

describe('addOffset', function () {
it('addOffset', async function () {
const basePtr = ethers.toBeHex(0x80, 32);
const offset = 32;
const expectedPtr = ethers.toBeHex(0xa0, 32);

await expect(this.mock.$addOffset(basePtr, offset)).to.eventually.equal(expectedPtr);
});

it('addOffsetwraps around', async function () {
const basePtr = ethers.toBeHex(0x80, 32);
const offset = 256;
const expectedPtr = ethers.toBeHex(0x180, 32);
await expect(this.mock.$addOffset(basePtr, offset)).to.eventually.equal(expectedPtr);
});
});

describe('pointer conversions', function () {
it('asBytes32 / asPointer', async function () {
const ptr = ethers.toBeHex('0x1234', 32);
await expect(this.mock.$asBytes32(ptr)).to.eventually.equal(ptr);
await expect(this.mock.$asPointer(ethers.Typed.bytes32(ptr))).to.eventually.equal(ptr);
});

it('asBytes / asPointer', async function () {
const ptr = await this.mock.$asPointer(ethers.Typed.bytes(ethers.toUtf8Bytes('hello world')));
expect(ptr).to.equal(ethers.toBeHex(0x80, 32)); // Default free pointer
await expect(this.mock.$asBytes(ptr)).to.eventually.equal(ethers.toBeHex(0x20, 32));
});

it('asUint256', async function () {
const value = 0x1234;
const ptr = ethers.toBeHex(value, 32);
await expect(this.mock.$asUint256(ptr)).to.eventually.equal(value);
});
});

describe('memory operations', function () {
it('copy', async function () {
await expect(this.mock.$copy(ethers.toBeHex(0x80, 32), ethers.toBeHex(0xc0, 32), 32)).to.not.be.reverted;
});

it('copy with zero length', async function () {
await expect(this.mock.$copy(ethers.toBeHex(0x80, 32), ethers.toBeHex(0xc0, 32), 0)).to.not.be.reverted;
});
});
});