code-423n4 / 2023-10-zksync-findings

4 stars 0 forks source link

Gas Optimizations #732

Open c4-submissions opened 1 year ago

c4-submissions commented 1 year ago

See the markdown file with the details of this report here.

141345 commented 1 year ago

4 r 4 nc

UBS-01G: Significant Optimization of Overall Transaction Parsing r

LCH-01G: Inefficient Evaluation of Modulo nc

LCH-02G: Inefficient Evaluation of Multiplication nc

ALS-01G: Inefficient Mapping Lookups r

DDN-01G: Inefficient Memory Expansion r

DDN-02G: Inefficient Mapping Lookups d

LMP-01G: Sub-Optimal Fork of Solady r

MEL-01G: Inefficient Evaluation of Modulo d

MEL-02G: Inefficient Evaluation of Division d

ESR-01G: Inefficient Evaluation of Ternary Operator nc

MXO-01G: Inefficient Mapping Lookups d

LER-01G: Inefficient Mapping Lookups d

LWB-01G: Inefficient Mapping Lookups d

GSE-01G: Inefficient Iterator Increment Statement nc

BZS-01G: Inefficient Iterator Increment Statement d

c4-pre-sort commented 1 year ago

141345 marked the issue as sufficient quality report

c4-judge commented 11 months ago

GalloDaSballo marked the issue as grade-b

alex-ppg commented 11 months ago

Hey @141345, I would be keen to know why the following optimizations are marked as "d":

For example, DDN-02G refers to the following lookups:

Please keep in mind that a mapping entry lookup performs a keccak256 instruction under the hood (as well as potential abi.encode etc. methods) to construct the "key" (i.e. storage offset) to access an entry. I am more than happy to produce supplemental test cases showcasing the gas savings of each exhibit.

Additionally, BZS-01G is marked as d while GSE-01G is marked as nc which does not make sense as they relate to the exact same type of optimization.

GalloDaSballo commented 11 months ago

I have checked adding further refactorings to this report score and don't believe it would qualify for A at this time

I also agree with the Presorter Heuristic that a specific type of optimization (lookup) should be awarded only once

I recommend listing the instances in sequence instead of sending them file by file