-
Notifications
You must be signed in to change notification settings - Fork 82
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
Feat(engine): Cross contract calls #560
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Everything looks good! Already reviewed what is in the code, I'll do a future review about what could be potentially missing from it.
I can't approve / reject this PR as I created it 🤦 . Let's discuss the comments I made, and then I'll approve in a comment.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
All in all looks good. I don't have any issues per se with the logic itself. I want to go once again when we reach a final review and try to come up with scenarios on how we could break this. I have not done a security analysis of this yet.
This is blocked until #565 is in, correct? |
Approve ✅ (I can't approve on GitHub interface, as I initiated the PR) |
We need to get better at not contributing to others' work directly, but via PR me thinks :P. It is awkward you can't exactly review your own PR in this case. |
@@ -30,12 +30,10 @@ mod costs { | |||
// TODO(#483): Determine the correct amount of gas | |||
pub(super) const EXIT_TO_ETHEREUM_GAS: EthGas = EthGas::new(0); | |||
|
|||
// TODO(#332): Determine the correct amount of gas | |||
pub(super) const FT_TRANSFER_GAS: NearGas = NearGas::new(100_000_000_000_000); | |||
pub(super) const FT_TRANSFER_GAS: NearGas = NearGas::new(10_000_000_000_000); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't want to block this PR, but I believe the rationale for this value should be attributed. Same for the one blow.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
/// For example, if an expensive EVM call ends with a NEAR cross contract call, then there may not be | ||
/// much gas left to perform it. In this case, the promise could be `Delayed` (stored in the router) | ||
/// and executed in a separate transaction with a fresh 300 Tgas available for it. | ||
Delayed(PromiseArgs), |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What if these are never consumed? Could we have some process to free these up if they are old? Or should we?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ultimately it is the user that has to cover the storage staking for stored promises they never used. I don't know why they would not use the promises they have scheduled. We could add a cancellation interface, but this adds more complexity to the design, and I do not think it is worth having that complexity for this edge case that may never come up.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Also this bring the question: who can destroy it and when? probably the answer is the same user that scheduled it, but is still a bit pointless! Agree on leaving it there until it is consumed!
pub const INITIALIZE_GAS: NearGas = NearGas::new(15_000_000_000_000); | ||
pub const UNWRAP_AND_REFUND_GAS: NearGas = NearGas::new(25_000_000_000_000); | ||
pub const WITHDRAW_GAS: NearGas = NearGas::new(30_000_000_000_000); | ||
pub const WITHDRAW_TO_NEAR_SELECTOR: [u8; 4] = [0x6b, 0x35, 0x18, 0x48]; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This should explain what the 4 bytes are meant to represent.
pub const VERSION_UPDATE_GAS: NearGas = NearGas::new(5_000_000_000_000); | ||
pub const INITIALIZE_GAS: NearGas = NearGas::new(15_000_000_000_000); | ||
pub const UNWRAP_AND_REFUND_GAS: NearGas = NearGas::new(25_000_000_000_000); | ||
pub const WITHDRAW_GAS: NearGas = NearGas::new(30_000_000_000_000); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
These gas costs should be explained.
const WNEAR_WITHDRAW_GAS: Gas = Gas(5_000_000_000_000); | ||
const WNEAR_REGISTER_GAS: Gas = Gas(5_000_000_000_000); | ||
const REFUND_GAS: Gas = Gas(5_000_000_000_000); | ||
const WNEAR_REGISTER_AMOUNT: u128 = 1_250_000_000_000_000_000_000; | ||
/// Must match arora_engine_precompiles::xcc::state::STORAGE_AMOUNT | ||
const REFUND_AMOUNT: u128 = 2_000_000_000_000_000_000_000_000; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
These gas costs should be explained.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
My only real complaint is that there is a lot of lacking documentation that will not only help ourselves, but also others and especially the auditors.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sorry, forgot to say that we require all the engine/src/lib.rs
new functions should also be feature gated.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's fine, @birchmd had informed me that this should not be an issue in the end, and I do agree.
…ile gas estimate tests
I'll address requests for more documentation in a follow-up PR. |
Feat(tracing): Implement callTracer (#538) Add docker as a prerequisites (#542) This was found by @mooori while running tests for first time. Update dependencies (#541) Update etc/self-contained-5bEgfRQ Decrease gas usage for some tests Merge: 2.6.1 into `develop` from `master` (#546) * Feat(standalone): Add debug tracing of remaining gas values (#391) * Balance type refactoring (#386) * Remove standalone binary (#403) * Eth Address cleanup (#387) * Fix(dep) Pass std feature on to rjson library (#402) * Feat(benchmarks): update gas bounds after wasm cost reduction (#406) * Feat(tests): Mock touching trie nodes (#408) * Fix: Only allow owner to call deploy_upgrade (#410) * Feat(tests): Uniswap multi-hop swap benchmark (#415) * Fix(engine): Return revert error message during contract deploy (#424) * Move engine transactions module to engine-transactions crate (#418) * Avoid using wasmer directly (#420) * Fix compilation after #418 merge * Use default git attributes for Cargo.lock (#421) * Switch to rebranded blake2 (#422) * Remove unused unused (#423) * Chore: make aurora compatible with latest nightlies (#425) * Chore: upgrade to the recent nightly (#426) * chore: upgrade to the latest nearcore version (#427) * chore: move to 2021 edition (#428) * Benchmark contract performing only pure arithmetic and memory operations (#429) * Fix(engine): Require chain_id (#432) * chore: make sure aurora is using the latest cost config (#435) * Fix(test): make uniswap benchmarks reproducible (#437) * Fix(test): lower gas limit on OOG test; this is needed because the solidity compiler got more efficient, not because of a regression in EVM gas metering correctness * Fix(engine): Cache generation values to avoid repeated state reads (#438) * Fix(engine): Optimize is_account_empty (#439) * Fix(test): update uniswap wasm fraction; this is needed because optimizations from #438 and #439 together lowered the fraction of gas spent on IO * Fix(engine): Upgrade to latest SputnikVM (#445) * Fix(engine): Simple cache to stop consecutive duplicate reads (#446) * Version update and change log Co-authored-by: Aleksey Kladov <aleksey@near.org> Co-authored-by: Evgeny Ukhanov <evgeny.ukhanov@aurora.dev> Co-authored-by: Michael Birch <michael.birch@aurora.dev> Release 2.5.0. (#460) * Engine: Upgrade to latest SputnikVM changes (#455) * Engine: further caching optimizations (#456) * Use EthTransactionKind reference to serialize into bytes (#457) * Test: Reproduce GdASJ3KESs8VegpFECTveCwLQp8fxw8yvsauNEmGb6pZ gas failure (#454) * Fix typo: promise_create (#452) * Fix(Engine): Transactions to the zero address are not the same as transactions with empty to field (#458) Co-authored-by: Marcelo Fornet <marcelo.fornet@aurora.dev> Co-authored-by: Michael Birch <michael.birch@aurora.dev> Release 2.5.1. (#470) * Feat(Engine): Precompiles for predecessor_account_id and current_account_id (#462) * Implement a NEAR native version of the Solidity pure arithmetic benchmark to compare EVM execution cost with direct wasm cost (#463) * Fix features related to panic_info_message (#466) * Fix(standalone): set predecessor_account_id appropriately when executing promise callbacks (#467) * Fix(engine): fix bug in checking if an address exists (#469) * Version update * Update unreleased link Co-authored-by: Joshua J. Bouw <joshua.j.bouw@aurora.dev> Release 2.5.2. (#475) * Revert "Feat(Engine): Precompiles for predecessor_account_id and current_account_id (#462)" Release 2.5.3. (#527) * Fix(precompile): ExitToNear ExitToEthereum vulnerability patch Fix vulnerability Include exploit contract * Release 2.5.3 notes * Update solidity version Co-authored-by: Michael Birch <michael.birch@aurora.dev> Release 2.6.0. (#494) * Feat(Engine): Precompiles for predecessor_account_id and current_account_id (#462) * Fix(precompiles): Allow native precompiles to work in the standalone engine (#473) * Standalone engine improvements (#478) * Fix(Engine): Predecessor id precompile works in view calls (#477) * Feat(standalone): Persist transaction data (#481) * Feat(Engine): Add custom precompile for NEAR prepaid_gas (#479) * Test: Reproduce 8ru7VEAEbyfZdbC1W2PYQv2cY3W92rbTToDEN4yTp8aZ gas failure (#485) * Feat(standalone): Function to get latest/earliest block (#482) * Engine optimization: cache all reads from NEAR state (#488) * Feat(standalone): Handle call to new method of Engine contract (#490) * Feat(engine-types): optional serde integration (#468) * Changed near-blake2 dependency (#484) * Feat(standalone): thread-safety and serde (#496) * Refactor repro tests to have less boilerplate (#497) * Add serde for TraceTransaction (#495) * Fix(standalone): include possible execution errors in the TransactionIncludedOutcome result field (#500) * Reproduce transaction FRcorNvFojoxBrdiVMTy9gRD3H8EYXXKau4feevMZmFV (#498) * Reproduce transaction 5bEgfRQ5TSJfN9XCqYkMr9cgBLToM7JmS1bNzKpDXJhT (#499) * Fix(connector): Return an error when storage cannot be read instead of panicking (#501) * Standalone: forward underlying rocksdb crate features (#502) * Feat(tests): Multisender benchmark (#503) * Feat(test): Reproduce D98vwmi44hAYs8KtX5aLne1zEkj3MUss42e5SkG2a4SC (#504) * Standalone engine storage saves the AccountId of the associated engine deployed on chain (#510) * Fix(precompiles): Fix secp256k1 run not returning empty slice on an incorrect V byte (#513) * Self-contained wasm contract for testing 5bEgfRQ (#514) * Update gas estimation code to use new data from NEAR protocol v53 (#517) * Fix: Legacy transactions must be allowed (#520) * Chore(precompiles): Include blake2 compression only (#528) * Build(deps): Upgrade `libsecp256k1` version 0.3.5 => 0.7.0 (#515) * Bump regex from 1.5.4 to 1.5.6 (#526) * Bump regex from 1.5.4 to 1.5.6 in /etc/ft-receiver (#525) * Bump zeroize_derive from 1.1.0 to 1.3.2 (#523) * Bump crossbeam-deque from 0.8.0 to 0.8.1 (#521) * Bump crossbeam-utils from 0.8.4 to 0.8.8 (#524) * Bump cross-fetch from 2.2.3 to 2.2.6 in /etc/eth-contracts (#508) * Bump simple-get from 2.8.1 to 2.8.2 in /etc/eth-contracts (#507) * Bump follow-redirects from 1.14.0 to 1.15.1 in /etc/eth-contracts (#529) * Bump shelljs from 0.8.4 to 0.8.5 in /etc/eth-contracts * Fix(ecrecover): Set malleability flag to `0` for ecrecover. (#474) Co-authored-by: Evgeny Ukhanov <evgeny.ukhanov@aurora.dev> Co-authored-by: Marcelo Fornet <marcelo.fornet@aurora.dev> Co-authored-by: Michael Birch <michael.birch@aurora.dev> Co-authored-by: Roman Hodulák <roman.hodulak@aurora.dev> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Release 2.6.1. (#537) * Fix: Don't allow fee stealing. * Fix: Don't allow bridge receipt forging. * Fix(engine): Correctly account for changes in total supply of ETH on Aurora (#536) Co-authored-by: Michael Birch <michael.birch@aurora.dev> Co-authored-by: Michael Birch <michael.birch@aurora.dev> Chore: Update js dependencies (#543) # Update JS dependencies A number of the JS dependencies were out of date and require to be updated. While a number of these were used almost exclusively for testing, the EVM ERC 20 contract is obviously used in production. Over the last year, we had carefully monitored to ensure that we have not been allowing security-related vulnerabilities into our production contracts. However updating these contracts is a good thing to do, and should be actively updated going forward as we do see slight performance gains in doing so. As a small bonus, this also introduces a force of updating the docker image used to compile some of the test / benchmark contracts. ## Related Issues There are no known related issues. ## Risk Analysis The EvmErc20 and EvmErc20V2 contracts have been recompiled due to the nature of these changes as the compiler has been updated. However, this is expected and intended. ## Testing None are required. Constant error messages (#535) * Added errors module and changed json errors * Set engine module errors as consts * Set connector module errors as consts * deposit_event const errors * extend const errors msgs chore: Cargo make integration (#378) Make files can be a bit messy and hard to read through the file. Luckily, a rust alternative exists which is quite clean and accomplishes the same goals in a Rust-friendly manner. What this accomplishes is a replace of `make` with `cargo make`. Why this is important is because we are able to get a cleaner make file that can have even better control with the use of profiles. That would make it quite difficult to run the run deployments on the wrong binaries as it'll correctly read the appropriate ENV file. On top of this, I had reworked the Github action workflows and I am getting no failures now. Which is good, less time spent with kickstarting it to work again. This also includes a monthly "sweep" to keep our footprint smaller. In the future, this can be improved by dropping `default_to_workspace = false` and adding a Makefile.toml to every single crate in our workspace. Do note, for whatever reason there was an efficiency being made on the `profile_erc20_get_balance()` test with this compilation which doesn't make itself immediately obvious. I have tried to figure out why this is but can not come up with a reason. I request reviewers to please try to notice why this is. This is low priority, and I have only been working on this while debugging precompiles. Chore: PR template (#548) <!-- Thanks for submitting a pull request! Here are some helpful tips: * Run all the tests locally and ensure that they are passing. * Run `cargo fmt --all` to ensure that the code is formatted. * Small commits and contributions which attempt one single goal is preferable. * If the idea changes or adds anything functional which will affect users an AIP discussion is required first on the Aurora forum: https://forum.aurora.dev/discussions/AIPs%20(Aurora%20Improvement%20Proposals). * Avoid breaking the public API (namely in engine/src/lib.rs) unless required. * If your PR is a WIP, ensure that you enable "draft" mode to ensure that it is not reviewed too early. * Your first PRs won't use the CI automatically unless started by a maintainer. If this is not your first PR, please do NOT abuse the CI resources. Checklist: - [ ] I have performed a self-review of my code - [ ] I have documented my code, particularly in the hard to understand areas - [ ] I have made corresponding changes to the documentation - [ ] I have added tests to prove my fix or new feature is effective and works - [ ] Any dependent changes have been merged - [ ] I have pre-squashed my commits into a single commit and rebased. --> ## Description This adds a PR template to ensure that PRs are of the highest quality by providing guidelines and a helpful checklist. ## How should this be reviewed Ensure that everything that could be said has been said in this template. Do comment below if we should add any more particulars. ## Additional information This is based on a number of other templates that other projects have adopted (mainly Rust projects) but also heavily modified and have included particulars to the project itself. Chore: Delete binaries (#553) Chore(engine): Update to latest version of SputnikVM (#565) Chore(CI): Update checkout to v3 (#566) Chore: Disallow as_conversions (#561) Fix feature gated import in hash precompile Change missed featured `into` to `try_into` birchmd fixes Cleaner usize conversion error Fix missing vec in hash Fix tests Fix V check Fix testnet env EVM account Change hex feature from alloc to std Improve u64 to u8 conversion in JSON Cargo fmt Chore: Update Rust crypto libraries (#568) Revert ripemd160, as the lib is deprecated Fix(engine-transactions): Prevent panic on empty input (#573) Feat(engine): Get promise results precompile (#575) Fix(engine): Return correct value for get_bridge_prover function (#581) Refactor(Tests): Move test contracts to test etc folder (#577) Co-authored-by: Michael Birch <michael.birch@aurora.dev> Chore: Cleanup dead dependencies (#571) Chore(Precompiles): Ripemd160 lib deprecated (#570) Feat(precompiles): Use near host functions for alt_bn128 when compiling wasm artifact (#540) Co-authored-by: Joshua J. Bouw <joshua@aurora.dev> Feat(engine): Cross contract calls (#560) Chore(eth-contracts): Update openzeppelin to 4.7.3 (#584) Fix: Make binaries on push (#585) 2.7.0 release prep
Feat(tracing): Implement callTracer (#538) Add docker as a prerequisites (#542) Update dependencies (#541) Chore: Update js dependencies (#543) Constant error messages (#535) chore: Cargo make integration (#378) Chore: PR template (#548) Chore: Delete binaries (#553) Chore(engine): Update to latest version of SputnikVM (#565) Chore(CI): Update checkout to v3 (#566) Chore: Disallow as_conversions (#561) Chore: Update Rust crypto libraries (#568) Fix(engine-transactions): Prevent panic on empty input (#573) Feat(engine): Get promise results precompile (#575) Fix(engine): Return correct value for get_bridge_prover function (#581) Refactor(Tests): Move test contracts to test etc folder (#577) Co-authored-by: Michael Birch <michael.birch@aurora.dev> Chore: Cleanup dead dependencies (#571) Chore(Precompiles): Ripemd160 lib deprecated (#570) Feat(precompiles): Use near host functions for alt_bn128 when compiling wasm artifact (#540) Co-authored-by: Joshua J. Bouw <joshua@aurora.dev> Feat(engine): Cross contract calls (#560) Chore(eth-contracts): Update openzeppelin to 4.7.3 (#584) Fix: Make binaries on push (#585) 2.7.0 release prep
Enable access to trigger cross contract call from addresses internal to Aurora, to NEAR Account Ids.
AIP link: aurora-is-near/AIPs#2
Forum discussion: https://forum.aurora.dev/t/cross-contract-calls/304
Notes on the design
Cross contract calls can be triggered by any aurora address calling a particular precompile. When the precompile is called, a new proxy NEAR contract will be deployed (only the first time) and the contract call will be routed through this account.