As a public blockchain system, Ethereum enables the deployment and execution of smart contracts, which are essential to its digital economy and facilitate over 1 million daily transactions with a total volume exceeding $4 billion.

To safeguard Ethereum’s computational resources from DoS attacks and tax Ethereum transactions, a fee called \textit{gas} is charged for executing each smart contract. Solidity, the official programming language of Ethereum, simplifies smart contract development by concealing the complexities of Ethereum Virtual Machine (EVM). However, it also obscures how gas fees are charged for each piece of Solidity code. Consequently, Solidity programmers could unknowingly write code snippets that unnecessarily cause more gas fees. These issues, which we refer to as gas wastes, can lead to significant monetary losses for users.

To understand gas waste, we conduct an empirical study on Solidity code snippets and Ethereum on-chain traces. The studied 100 snippets are extracted from five popular Solidity applications, including 54 specific to Solidity. The on-chain traces are collected from ten million Ethereum transactions over ten days, revealing a total gas fee of $160 million. We categorize these gas wastes based on the data store areas (stack, memory, storage, calldata) and analyze the root causes. In total, our analysis yields 11 insights and four suggestions. Most of these insights (except Insight 4) have not been reported in previous literature. While some of them (Insight 4 and Suggestions 1, 3, and 4) overlap with findings from existing papers, our focus is on understanding how gas is wasted in practice and we provide real data to demonstrate how frequently they impact the real world and their monetary consequences, making the reporting of these findings still valuable in practical terms.

To identify gas wastes in Solidity programs, we develop \textit{PeCatch}, a suite of six static checkers. We evaluate PeCatch on the latest versions of the five studied Solidity applications and four additional open-source Solidity projects. PeCatch detects a total of 302 previously unknown gas wastes in the benchmark programs, significantly more than existing techniques, while reporting zero false positives. Additionally, we pinpoint 14 bugs causing gas wastes in the Solidity compiler. Fixing the detected wastes and compiler bugs could save $0.76 million every day.