- wanningD·Last edited by wanningD on Feb 9, 2023Linked with GitHubContributed by
There is no commentSelect some text and then click Comment, or simply add a comment to this page from below to start a discussion.Ethereum Protocol Fellowship 14th week update
Here is my mentor's (@https://github.com/MariusVanDerWijden) implementation based on the version I pushed before.
https://github.com/ethereum/go-ethereum/compare/master…MariusVanDerWijden:txpool-attacks?expand=1
The previous attack defenses are being tested on the real network and the refactor and benchmarking results are not significantly worse than the current version:
name old time/op new time/op delta
ListAdd-24 83.5ms ± 2% 128.5ms ± 0% ~ (p=0.333 n=2+2)
PendingDemotion100-24 85.8ns ± 0% 84.9ns ± 1% ~ (p=0.667 n=2+2)
PendingDemotion1000-24 82.2ns ± 2% 84.1ns ± 2% ~ (p=0.667 n=2+2)
PendingDemotion10000-24 82.3ns ± 1% 84.7ns ± 0% ~ (p=0.333 n=2+2)
FuturePromotion100-24 1.99ns ± 1% 2.07ns ± 0% ~ (p=0.333 n=2+2)
FuturePromotion1000-24 1.96ns ± 1% 1.96ns ± 1% ~ (p=0.667 n=2+2)
FuturePromotion10000-24 1.99ns ± 1% 1.95ns ± 1% ~ (p=0.667 n=2+2)
BatchInsert100-24 11.7ms ± 0% 12.0ms ± 3% ~ (p=1.000 n=2+2)
BatchInsert1000-24 85.0ms ± 1% 83.2ms ± 3% ~ (p=0.667 n=2+2)
BatchInsert10000-24 605ms ± 0% 615ms ± 1% ~ (p=0.333 n=2+2)
BatchLocalInsert100-24 9.85ms ± 2% 9.84ms ± 2% ~ (p=1.000 n=2+2)
BatchLocalInsert1000-24 86.1ms ± 3% 88.0ms ± 0% ~ (p=1.000 n=2+2)
BatchLocalInsert10000-24 772ms ± 3% 770ms ± 0% ~ (p=1.000 n=2+2)
InsertRemoteWithAllLocals-24 27.0ms ± 2% 26.2ms ± 1% ~ (p=0.333 n=2+2)
MultiAccountBatchInsert-24 190µs ± 4% 182µs ± 5% ~ (p=0.667 n=2+2)
name old alloc/op new alloc/op delta
ListAdd-24 31.2MB ± 0% 55.2MB ± 0% ~ (p=0.333 n=2+2)
PendingDemotion100-24 0.00B 0.00B ~ (all equal)
PendingDemotion1000-24 0.00B 0.00B ~ (all equal)
PendingDemotion10000-24 0.00B 0.00B ~ (all equal)
FuturePromotion100-24 0.00B 0.00B ~ (all equal)
FuturePromotion1000-24 0.00B 0.00B ~ (all equal)
FuturePromotion10000-24 0.00B 0.00B ~ (all equal)
BatchInsert100-24 299kB ± 0% 341kB ± 1% ~ (p=0.333 n=2+2)
BatchInsert1000-24 2.98MB ± 0% 3.37MB ± 0% ~ (p=0.333 n=2+2)
BatchInsert10000-24 24.9MB ± 0% 26.8MB ± 0% ~ (p=0.333 n=2+2)
BatchLocalInsert100-24 2.30MB ± 3% 2.53MB ± 2% ~ (p=0.333 n=2+2)
BatchLocalInsert1000-24 36.4MB ± 0% 38.9MB ± 5% ~ (p=0.333 n=2+2)
BatchLocalInsert10000-24 437MB ± 0% 444MB ± 0% ~ (p=0.333 n=2+2)
InsertRemoteWithAllLocals-24 5.93MB ± 3% 5.53MB ± 3% ~ (p=0.333 n=2+2)
MultiAccountBatchInsert-24 14.4kB ±11% 17.2kB ± 3% ~ (p=0.333 n=2+2)
name old allocs/op new allocs/op delta
ListAdd-24 1.70M ± 0% 2.90M ± 0% ~ (p=0.333 n=2+2)
PendingDemotion100-24 0.00 0.00 ~ (all equal)
PendingDemotion1000-24 0.00 0.00 ~ (all equal)
PendingDemotion10000-24 0.00 0.00 ~ (all equal)
FuturePromotion100-24 0.00 0.00 ~ (all equal)
FuturePromotion1000-24 0.00 0.00 ~ (all equal)
FuturePromotion10000-24 0.00 0.00 ~ (all equal)
BatchInsert100-24 5.92k ± 0% 7.68k ± 1% ~ (p=0.333 n=2+2)
BatchInsert1000-24 57.7k ± 0% 74.1k ± 0% ~ (p=0.333 n=2+2)
BatchInsert10000-24 508k ± 0% 583k ± 0% ~ (p=0.333 n=2+2)
BatchLocalInsert100-24 6.25k ± 0% 8.88k ± 0% ~ (p=0.333 n=2+2)
BatchLocalInsert1000-24 61.9k ± 0% 88.9k ± 0% ~ (p=0.333 n=2+2)
BatchLocalInsert10000-24 618k ± 0% 877k ± 0% ~ (p=0.333 n=2+2)
InsertRemoteWithAllLocals-24 46.0k ± 0% 59.8k ± 0% ~ (p=0.333 n=2+2)
MultiAccountBatchInsert-24 104 ± 4% 118 ± 1% ~ (p=0.333 n=2+2)
We both think the ED3 part needs more thinking and will work on it in the future.
Read more
Ethereum Protocol Fellowship 15th week update
After the discussion with mentor Marius, we agree that the modified version is working well on the ED1,2,4, but may cause the txpool to be flooded with underpriced transactions as they can not be replaced if a future transaction exists. So the new version contains the defense of two attacks: future transaction evicting a pending transaction a transaction overspending the funds of a sender Here is the merge request on go-ethereum mempool we just pushed. https://github.com/ethereum/go-ethereum/pull/26648
Feb 9, 2023Ethereum Protocol Fellowship 10th week update
Self Introduction I’m a PhD student in Syracuse University in Blockchain security area, and now working on the Ethereum Mempool Attack Detection&Defence. Very excited joining in the fellowship and make some contribution to the Ethereum Portocol. Progress As I finished the first part of the defense and test it successfully on the mainnet, I started to dive into the more general way of Dos Attack defense which can make the codes more ellegent and easy to read and can save time and space cost. For now, here is the strategy to mitigate ADAMS attacks. The idea is to enforce that a valid transaction always has a higher priority in admission than an invalid transaction, despite their prices. Figure belw depicts the priority among transactions of different types, that is, parent valid transactions, child valid transactions and invalid transactions (including future and latent-overdraft transactions). Only among transactions of the same type, price-based priority is enforced. We call it validity-first admission priority. We enforce validity-first priority by two proposed admission policies:
Jan 20, 2023Ethereum Protocol Fellowship 12th week update
Self Introduction I’m a PhD student in Syracuse University in Blockchain security area, and now working on the Ethereum Mempool Attack Detection&Defence. Very excited joining in the fellowship and make some contribution to the Ethereum Portocol. Progress Attacks For precvious attacks, we introduced four patterns ED1-4 to defence, here are other patterns. Pattern LD1: It locks the mempool by directly crafting future transactions. On an initially empty mempool, it sends the same transactions with Pattern ED1.
Jan 19, 2023Ethereum Protocol Fellowship 8th week update
Self Introduction I’m a PhD student in Syracuse University in Blockchain security area, and now working on the Ethereum Mempool Attack Detection&Defence. Very excited joining in the fellowship and make some contribution to the Ethereum Portocol. Process As I just recovered from Covid-19, I didn’t do much work last week. But this week I'm getting better, so I keep on doing the test of the previous deter attack defense. How do we test the success of our defense on the net? Start private node We start a private node and start an interactive session using Geth console connected to the node we established.
Dec 22, 2022
HackMD