To Do #2

[GreatSoshiant]

Future Steps:

• Reading papers that conducted transaction-based research on the whole Ethereum blockchain to check how they did it. What are the requirements of the full node I/O we need (these papers) :

  • https://arxiv.org/abs/1812.05934
  • https://arxiv.org/abs/2101.06204
  • https://www.usenix.org/conference/usenixsecurity21/presentation/perez
  • https://www.usenix.org/conference/usenixsecurity20/presentation/zhang-mengya
  • https://arxiv.org/abs/2005.08278
  • https://www4.comp.polyu.edu.hk/~csxluo/SODA.pdf

• Adding beacon proxies

• Try to find a way to detect the admin of upgradeable proxies and then check if the admin is an EOA, Multi-Sig, or governance token (number of governors of the system: EOA is one. Multi-sig is easy to understand. For governance tokens, we need 2 things: 1. Distribution and allocation of the token to different addresses 2. Threshold of the governance token).

• We can find the number of admin changes on the proxies we found.

• We can find the number of upgrades that happened to a proxy contract. We can also check for the amount of upgrade each time happened to the proxy contract by using a code difference tool and comparing the old implementation contract with the newer version

• Finding a way to detect Call-based upgradeable contracts

  • Other patterns are not recognizable ( I have my reasons)

• Reading about Diamonds and Create2 upgradeability methods

  • Add the description
  • Update the evaluation table
  • Check if there is a way to find contracts that use this method

• Running a full node maybe!

• Add differences between delegate call and call based approaches on the Devs' view (how much they should change their code)

• We can do research on the plausibility as well because it's important for upgrade (for measurement study part)

[GreatSoshiant]

Sereum: Protecting Existing Smart Contracts Against Re-Entrancy Attacks

• They ran a full archival node and as they said they had a reasonable I/O
• They have used a modified version of Geth Client (they add some modules to it) and also they've changed it to speed up process of getting transaction traces
• Their model is a hybrid analysis similar to ours (Transaction trace + Bytecode Analysis)
• Our model need a decompiler in addition
• They re-executed 78 million transactions involved in 4.5 million Blocks
• They randomly sampled from their outcomes and check them manually like what we did

[GreatSoshiant]

The Eye of Horus: Spotting and Analyzing Attacks on Ethereum Smart Contracts

• They ran a full archival Geth client
• The used debug_traceTransaction method
• They changed the Tracer module of the Geth to just have their desired data on the output of transaction traces (for example they removed stack counter, memory for every instruction, etc.)
• They used the Ethereum ETL framework to retrieve a list of transactions for every smart contract deployed up to block 10 M and collected a total of 697,373,206 transactions and 3,362,876 contracts.
• Ethereum ETL needs access to a full archival node
• They filtered contracts without transactions and removed transactions that have a gas limit of 21,000 (do not execute code). Also, they skipped all the transactions that were part of the 2016 denial-of-service attacks, as these incur high execution times
• Their finaal dataset is 1,234,197 smart contracts consisting of 371,419,070 transactions
• Their Setup: 64 GB of memory and an Intel(R) Core(TM) i7-8700 CPU with 12 cores clocked at 3.2 GHz, running 64-bit Ubuntu 18.04.5 LTS. (used Geth version 1.9.9)

[GreatSoshiant]

Smart Contract Vulnerabilities: Vulnerable Does Not Imply Exploited

• They patched Geth client to speed up getting transaction traces

• I think they changed the tracer module of Geth to have their just their desired output

• maybe a helpful link on stackoverflow: https://ethereum.stackexchange.com/questions/40191/how-to-install-a-custom-tracer-in-geth/40465

• tracer module: https://github.com/ethereum/go-ethereum/tree/master/eth/tracers

• check how Horus changed the trace module

• they used data set of other papers which is not related to our topic

[GreatSoshiant]

TXSPECTOR: Uncovering Attacks in Ethereum from Transactions

• like other researches, they modified the Geth client to just have desired traces for transactions.

• unlike others, they modified EVM instead of trace module to log their desired field of transactions and op-codes

• " The traces were collected on an L8s v2 instance on the Microsoft Azure Cloud, with 8 VCPUs, 64GB RAM and 2TB SSD, running Ubuntu 18.04. Trace Extractor ran a full Ethereum node to collect bytecode-level traces, from the 0-t h block to the 7, 200, 000-t h block. Note that Ethereum has around 10180000 number of blocks (as in June 1st 2020) and it keeps growing exponentially. We cannot collect all of them for our experiment because it takes a huge amount of storage and also processing time. We therefore stop collecting at 7,200,000-th block, which has resulted in the size of 1,577 GB, containing 397,269,533 transactions in total. We store them in a Trace DB (implemented atop MongoDB) by Trace Extractor.

• They said that based on their logic it takes 1 second for each transaction to be processed!!

[GreatSoshiant]

Time-Travel Investigation: Towards Building A Scalable Attack Detection Framework on Ethereum

• they said that it will take 47 minutes for replaying 100 transactions!!!

• they used JS tracer module of geth which gives us the ability to filter traces as we want

  • https://geth.ethereum.org/docs/dapp/tracing-filtered

• They changed Geth tracer and the EVM to collect the state of the blockchain which helps them to re execute each transaction rather than all transactions in the same block

• their model is efficient when you want to target a specific transaction and not all transactions in a block

[GreatSoshiant]

SODA: A Generic Online Detection Framework for Smart Contracts

• they modified the Geth client to retrieve their desired data

• They have collected information of the internal transactions by adding code into the interpretation handler of the EVM instructions (CALL, CALLCODE,Create, Delegatecall, staticcall, self destruct) that can create internal transactions

• I'm not sure if we can collect the proxies just by Geth's EVM modification because we need two consecutive instructions.

[GreatSoshiant]

Relate the upgrade to the attacks or anything like this that happened before the upgrade