fix(levm)!: fix jumpdests for large initcodes #5254

iovoid · 2025-11-10T20:07:47Z

Caution

This changes the serialization of Code, and thus requires re-syncing.

Motivation

While executing the state tests in debug mode, a debug assert was triggered.

The calculation of valid jumpdests assumes the code will be less than 2^16 bytes in length, however this isn't true before Shanghai.

Description

This PR adds support for up to 2^32 bytes of initcode.

github-actions · 2025-11-10T20:10:36Z

Lines of code report

Total lines added: 2
Total lines removed: 1
Total lines changed: 3

Detailed view

+------------------------------------------------------------------------+-------+------+
| File                                                                   | Lines | Diff |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/storage/store_db/rocksdb.rs                              | 1603  | -1   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/opcode_handlers/stack_memory_storage_flow.rs | 250   | +2   |
+------------------------------------------------------------------------+-------+------+

pablodeymo · 2025-11-10T20:17:28Z

crates/storage/store_db/rocksdb.rs


        for (code_hash, code) in update_batch.code_updates {
-            let mut buf = Vec::with_capacity(6 + code.bytecode.len() + 2 * code.jump_targets.len());
+            let mut buf = Vec::with_capacity(6 + code.bytecode.len() + 4 * code.jump_targets.len());


I think it's better to use size_of to avoid hard-coding this lenght

Improved in 9d129ce

pablodeymo · 2025-11-10T20:20:59Z

crates/storage/store_db/rocksdb.rs

-                .chunks_exact(2)
-                .map(|c| u16::from_le_bytes([c[0], c[1]]))
+                .chunks_exact(4)
+                .map(|c| u32::from_le_bytes([c[0], c[1], c[2], c[3]]))


Talking with @Oppen, we think that it's better to encode the jump_target in RLP, which will be smaller than 32 bits for the common case.

Better is a strong word for any use of RLP. But in the tradeoff IO probably wins, and RLP is better for IO than always using all the bytes.

Done in 084c8e2

github-actions · 2025-11-10T20:24:46Z

Benchmark Results Comparison

No significant difference was registered for any benchmark run.

Detailed Results

Benchmark Results: BubbleSort

Command	Mean [s]	Min [s]	Max [s]	Relative
`main_revm_BubbleSort`	2.962 ± 0.020	2.939	2.988	1.00 ± 0.01
`main_levm_BubbleSort`	3.248 ± 0.082	3.202	3.467	1.10 ± 0.03
`pr_revm_BubbleSort`	2.954 ± 0.019	2.937	3.000	1.00
`pr_levm_BubbleSort`	3.233 ± 0.021	3.203	3.272	1.09 ± 0.01

Benchmark Results: ERC20Approval

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ERC20Approval`	960.3 ± 5.8	951.7	968.4	1.00
`main_levm_ERC20Approval`	1137.5 ± 10.8	1123.7	1158.8	1.18 ± 0.01
`pr_revm_ERC20Approval`	970.9 ± 9.7	960.9	987.4	1.01 ± 0.01
`pr_levm_ERC20Approval`	1140.9 ± 5.6	1130.9	1150.4	1.19 ± 0.01

Benchmark Results: ERC20Mint

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ERC20Mint`	132.6 ± 2.3	130.8	138.4	1.00
`main_levm_ERC20Mint`	168.7 ± 5.8	165.0	184.8	1.27 ± 0.05
`pr_revm_ERC20Mint`	133.7 ± 2.1	131.7	138.3	1.01 ± 0.02
`pr_levm_ERC20Mint`	168.6 ± 1.4	167.2	172.0	1.27 ± 0.02

Benchmark Results: ERC20Transfer

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ERC20Transfer`	230.5 ± 3.8	225.7	237.2	1.00 ± 0.02
`main_levm_ERC20Transfer`	285.5 ± 3.1	280.2	288.8	1.24 ± 0.02
`pr_revm_ERC20Transfer`	230.1 ± 2.6	227.4	234.8	1.00
`pr_levm_ERC20Transfer`	288.6 ± 5.5	285.0	304.0	1.25 ± 0.03

Benchmark Results: Factorial

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_Factorial`	225.3 ± 2.5	223.4	231.7	1.01 ± 0.01
`main_levm_Factorial`	264.0 ± 1.9	262.4	268.9	1.18 ± 0.01
`pr_revm_Factorial`	223.5 ± 0.9	222.3	225.2	1.00
`pr_levm_Factorial`	263.1 ± 1.7	260.9	265.8	1.18 ± 0.01

Benchmark Results: FactorialRecursive

Command	Mean [s]	Min [s]	Max [s]	Relative
`main_revm_FactorialRecursive`	1.607 ± 0.042	1.543	1.661	1.00
`main_levm_FactorialRecursive`	8.248 ± 0.029	8.193	8.287	5.13 ± 0.14
`pr_revm_FactorialRecursive`	1.650 ± 0.019	1.615	1.675	1.03 ± 0.03
`pr_levm_FactorialRecursive`	8.282 ± 0.045	8.206	8.349	5.15 ± 0.14

Benchmark Results: Fibonacci

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_Fibonacci`	205.7 ± 1.9	202.5	210.1	1.00 ± 0.02
`main_levm_Fibonacci`	249.2 ± 3.1	247.0	257.0	1.22 ± 0.03
`pr_revm_Fibonacci`	205.0 ± 3.4	202.8	214.4	1.00
`pr_levm_Fibonacci`	250.1 ± 3.3	246.5	256.5	1.22 ± 0.03

Benchmark Results: FibonacciRecursive

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_FibonacciRecursive`	829.1 ± 11.0	815.8	849.1	1.11 ± 0.02
`main_levm_FibonacciRecursive`	744.4 ± 6.8	737.1	756.3	1.00
`pr_revm_FibonacciRecursive`	859.7 ± 9.4	841.2	872.2	1.15 ± 0.02
`pr_levm_FibonacciRecursive`	745.3 ± 7.5	738.3	759.2	1.00 ± 0.01

Benchmark Results: ManyHashes

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ManyHashes`	8.2 ± 0.1	8.2	8.3	1.00
`main_levm_ManyHashes`	9.6 ± 0.1	9.4	9.8	1.16 ± 0.01
`pr_revm_ManyHashes`	8.3 ± 0.1	8.2	8.6	1.00 ± 0.01
`pr_levm_ManyHashes`	9.5 ± 0.1	9.5	9.6	1.15 ± 0.01

Benchmark Results: MstoreBench

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_MstoreBench`	264.1 ± 4.7	260.9	274.4	1.08 ± 0.02
`main_levm_MstoreBench`	243.5 ± 3.2	238.9	249.6	1.00
`pr_revm_MstoreBench`	264.2 ± 3.7	261.3	273.0	1.09 ± 0.02
`pr_levm_MstoreBench`	245.4 ± 1.8	243.5	248.8	1.01 ± 0.02

Benchmark Results: Push

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_Push`	290.9 ± 1.0	289.2	292.5	1.00
`main_levm_Push`	298.0 ± 1.4	295.8	300.1	1.02 ± 0.01
`pr_revm_Push`	290.9 ± 1.3	289.7	294.2	1.00 ± 0.01
`pr_levm_Push`	299.0 ± 2.6	296.6	304.5	1.03 ± 0.01

Benchmark Results: SstoreBench_no_opt

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_SstoreBench_no_opt`	166.7 ± 3.7	161.1	170.7	1.86 ± 0.07
`main_levm_SstoreBench_no_opt`	89.5 ± 2.7	87.5	95.8	1.00
`pr_revm_SstoreBench_no_opt`	167.3 ± 5.0	161.4	178.2	1.87 ± 0.08
`pr_levm_SstoreBench_no_opt`	90.1 ± 2.0	87.9	94.1	1.01 ± 0.04

JereSalo · 2025-11-10T21:22:25Z

crates/common/types/account.rs

    pub bytecode: Bytes,
-    // TODO: Consider using Arc<[u16]> (needs to enable serde rc feature)
-    pub jump_targets: Vec<u16>,
+    // TODO: Consider using Arc<[u32]> (needs to enable serde rc feature)


Can we add a comment here or anywhere you consider convenient about the reason why we use 32 bits? So nobody changes it to u16 back again
If you want you can link EIP-3860 but it's not necessary

No, it would be a good idea to link to the EIP. Also to the failing EEST test.

When in doubt, overcommenting is better than undercommenting. Don't let school tell you otherwise.

Added comment in da77868

fix jumpdests for large initcodes

226f91a

github-actions bot assigned iovoid Nov 10, 2025

pablodeymo reviewed Nov 10, 2025

View reviewed changes

JereSalo reviewed Nov 10, 2025

View reviewed changes

iovoid added 5 commits November 11, 2025 12:33

use sizeof instead of constants

9d129ce

save jump_targets as rlp

084c8e2

add comment explaining why u32

da77868

clippy

8c7b633

fix deserialization

4d3ef13

iovoid marked this pull request as ready for review November 11, 2025 18:58

iovoid requested a review from a team as a code owner November 11, 2025 18:58

Merge branch 'main' into jumpdest_large_initcode

4a7156a

MegaRedHand approved these changes Nov 11, 2025

View reviewed changes

jrchatruc approved these changes Nov 11, 2025

View reviewed changes

Merge branch 'main' into jumpdest_large_initcode

6f974b4

jrchatruc enabled auto-merge November 11, 2025 19:45

jrchatruc added this pull request to the merge queue Nov 11, 2025

Merged via the queue into main with commit 8089824 Nov 11, 2025
51 checks passed

jrchatruc deleted the jumpdest_large_initcode branch November 11, 2025 20:45

fix(levm)!: fix jumpdests for large initcodes #5254

fix(levm)!: fix jumpdests for large initcodes #5254

Uh oh!

Conversation

iovoid commented Nov 10, 2025

Uh oh!

github-actions bot commented Nov 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Lines of code report

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented Nov 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Results Comparison

Benchmark Results: BubbleSort

Benchmark Results: ERC20Approval

Benchmark Results: ERC20Mint

Benchmark Results: ERC20Transfer

Benchmark Results: Factorial

Benchmark Results: FactorialRecursive

Benchmark Results: Fibonacci

Benchmark Results: FibonacciRecursive

Benchmark Results: ManyHashes

Benchmark Results: MstoreBench

Benchmark Results: Push

Benchmark Results: SstoreBench_no_opt

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

github-actions bot commented Nov 10, 2025 •

edited

Loading

github-actions bot commented Nov 10, 2025 •

edited

Loading