feat: add drop local user operation endpoint #610

dancoombs · 2024-02-17T20:56:33Z

Closes #567

Proposed Changes

Adds rundler_dropLocalUserOperation rpc endpoint
Requirements:
- The user operation must contain a sender/nonce pair this is present in the local mempool.
- The user operation must pass entrypoint.simulateValidation. I.e. it must have a valid signature and verificationGasLimit
- The user operation must have zero values for: preVerificationGas, callGasLimit, calldata, and maxFeePerGas
- The user operation must be in the pool for at least 10 blocks before getting dropped.

codecov · 2024-02-17T20:59:28Z

Codecov Report

Attention: 180 lines in your changes are missing coverage. Please review.

Comparison is base (20b725f) 57.86% compared to head (1e72e7d) 57.59%.

Additional details and impacted files

Files	Coverage Δ
crates/pool/src/mempool/error.rs	`76.19% <ø> (ø)`
crates/pool/src/mempool/mod.rs	`87.50% <ø> (ø)`
crates/pool/src/mempool/pool.rs	`95.65% <100.00%> (+0.16%)`	⬆️
crates/pool/src/server/mod.rs	`25.00% <ø> (ø)`
crates/provider/src/traits/entry_point.rs	`50.00% <ø> (ø)`
crates/rpc/src/eth/mod.rs	`0.00% <ø> (ø)`
crates/sim/src/simulation/simulation.rs	`78.11% <ø> (ø)`
crates/types/src/validation_results.rs	`57.53% <ø> (ø)`
bin/rundler/src/cli/node/mod.rs	`0.00% <0.00%> (ø)`
crates/pool/src/mempool/uo_pool.rs	`92.65% <98.86%> (+0.45%)`	⬆️
... and 13 more

Flag	Coverage Δ
unit-tests	`57.59% <39.18%> (-0.28%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
rundler binary	`0.00% <0.00%> (ø)`
builder	`51.30% <ø> (ø)`
dev	`0.00% <ø> (ø)`
pool	`64.62% <57.14%> (-0.22%)`	⬇️
provider	`0.87% <0.00%> (-0.04%)`	⬇️
rpc	`34.70% <0.00%> (-1.52%)`	⬇️
sim	`84.64% <0.00%> (+0.48%)`	⬆️
tasks	`∅ <ø> (∅)`
types	`86.82% <ø> (-4.45%)`	⬇️
utils	`14.96% <ø> (ø)`

denniswon

So uo hash is not needed for this api? What's the difference between uo hash bs uo Id (sender + nonce) for serving uniqueness?

Also after canceling the uo with nonce K, should nonce still be a new nonce value? (K+1) or K can still be the nonce of the next uo?

denniswon · 2024-02-18T23:27:48Z

crates/rpc/src/rundler.rs

+
+    /// Drops a user operation from the local mempool.
+    ///
+    /// Requirements:


Noted. So it costs gas for verification to cancel a UO. Why the 3rd requirement for this end point instead of just ignoring? Just curious

It doesn't cost any gas on chain, but the UO must contain a high enough gas limit such that we can simulate it. This value is set by the user because its part of the UO that is signed.

denniswon · 2024-02-18T23:35:02Z

crates/rpc/src/rundler.rs

+        if uo.pre_verification_gas != U256::zero()
+            || uo.call_gas_limit != U256::zero()
+            || uo.call_data.len() != 0
+            || uo.max_fee_per_gas != U256::zero()


Do these requirements exist for simulate uo operation?

I mean simulation verification

The reason for these requirements is to ensure that the signed UO is not executable onchain. That is, it should have 0 gas fees associated with it. The most important one here is max_fee_per_gas == 0, as that would cause the entrypoint to reject the UO. We could reduce this requirements set to just that.

Simulation verification when a normal UO is submitted doesn't have this requirement because the UO is meant to be included onchain.

dancoombs · 2024-02-19T16:00:00Z

So uo hash is not needed for this api? What's the difference between uo hash bs uo Id (sender + nonce) for serving uniqueness?

There is no API for a sender to lookup pending UOs. So if they find that they are stuck, i.e. getting replacement underpriced consistently, they can issue a cancel just knowing their address and their nonce. No need to also lookup a hash.

Also after canceling the uo with nonce K, should nonce still be a new nonce value? (K+1) or K can still be the nonce of the next uo?

K is still the nonce of the next UO, since nothing happened onchain.

dancoombs · 2024-02-19T16:00:48Z

This is missing the 10 block delay that we discussed in the ticket to prevent spam, need to add that before merge.

dancoombs · 2024-02-19T16:01:42Z

Also worth noting that this isn't a perfect drop mechanism. Its possible that the UO has already been bundled (or sent on the P2P network) before the drop.

0xfourzerofour

Looks good, just a nit

crates/pool/src/mempool/mod.rs

0xfourzerofour · 2024-02-21T16:46:18Z

crates/provider/src/ethers/entry_point.rs

@@ -46,21 +46,38 @@ where
        self.deref().address()
    }

-    async fn simulate_validation(
+    async fn get_simulate_validation_call(
        &self,
        user_op: UserOperation,
        max_validation_gas: u64,
    ) -> anyhow::Result<TypedTransaction> {
        let pvg = user_op.pre_verification_gas;


not a comment but having spaces between distinct code blocks is very handy for vim movements rip

denniswon reviewed Feb 18, 2024

View reviewed changes

dancoombs force-pushed the danc/balances-rpc branch 5 times, most recently from a727ed8 to c635e0a Compare February 21, 2024 13:02

dancoombs force-pushed the danc/drop-rpc branch from 75f76c2 to 34dd83d Compare February 21, 2024 15:27

dancoombs force-pushed the danc/balances-rpc branch 2 times, most recently from 5f327b9 to 85504b4 Compare February 21, 2024 15:29

Base automatically changed from danc/balances-rpc to main February 21, 2024 15:32

dancoombs force-pushed the danc/drop-rpc branch from 34dd83d to f8c1c93 Compare February 21, 2024 15:33

dancoombs marked this pull request as ready for review February 21, 2024 15:33

dancoombs requested review from dphilipson, 0xfourzerofour and alex-miao as code owners February 21, 2024 15:33

dancoombs force-pushed the danc/drop-rpc branch from f8c1c93 to 6616322 Compare February 21, 2024 15:59

0xfourzerofour approved these changes Feb 21, 2024

View reviewed changes

feat: add drop UO rpc endpoint

1e72e7d

dancoombs force-pushed the danc/drop-rpc branch from 6616322 to 1e72e7d Compare February 21, 2024 20:04

0xfourzerofour approved these changes Feb 21, 2024

View reviewed changes

dancoombs merged commit 44fae8a into main Feb 21, 2024
7 checks passed

dancoombs deleted the danc/drop-rpc branch February 21, 2024 20:08

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: add drop local user operation endpoint #610

feat: add drop local user operation endpoint #610

dancoombs commented Feb 17, 2024 •

edited

Loading

codecov bot commented Feb 17, 2024 •

edited

Loading

denniswon left a comment

denniswon Feb 18, 2024

dancoombs Feb 19, 2024

denniswon Feb 18, 2024

denniswon Feb 18, 2024

dancoombs Feb 19, 2024

dancoombs commented Feb 19, 2024

dancoombs commented Feb 19, 2024

dancoombs commented Feb 19, 2024

0xfourzerofour left a comment

0xfourzerofour Feb 21, 2024

feat: add drop local user operation endpoint #610

feat: add drop local user operation endpoint #610

Conversation

dancoombs commented Feb 17, 2024 • edited Loading

Proposed Changes

codecov bot commented Feb 17, 2024 • edited Loading

Codecov Report

denniswon left a comment

Choose a reason for hiding this comment

denniswon Feb 18, 2024

Choose a reason for hiding this comment

dancoombs Feb 19, 2024

Choose a reason for hiding this comment

denniswon Feb 18, 2024

Choose a reason for hiding this comment

denniswon Feb 18, 2024

Choose a reason for hiding this comment

dancoombs Feb 19, 2024

Choose a reason for hiding this comment

dancoombs commented Feb 19, 2024

dancoombs commented Feb 19, 2024

dancoombs commented Feb 19, 2024

0xfourzerofour left a comment

Choose a reason for hiding this comment

0xfourzerofour Feb 21, 2024

Choose a reason for hiding this comment

dancoombs commented Feb 17, 2024 •

edited

Loading

codecov bot commented Feb 17, 2024 •

edited

Loading