Incident report: Denial-of-Service (Feather Wallet)

I just noticed your username. Thank you for the project, Feather is amazing. I have a question though. I know Ruckinum ran an analysis and thinks this is not a black marble flood, but I can't help but think it's a way go somehow break the anonymity of monero, whether just sent amounts, or received amounts, which would still give a wealth of information.

I don't believe this is a random (D)DoS/spam. This is a deanon attack. I know it in my gut. I don't know enough about the internals of monero but I think you might.

Specifically...

The bug was triggered when the number of RingCT outputs on the blockchain exceeded 100 million

For instance, this transaction was constructed using a manipulated output distribution. Can you determine what the true spend is? Notice that all ring members are older than 1y 200d except for one 6-day-old output. Unless the user checks the ring on a block explorer and knows what to look out for, they would not notice that their transactions are being fingerprinted.

My understanding is that the 16 (or 15+real?) rings are all real, prior transactions. Are the transactions reused? If not, then they exhaust the supply of rings and now have great statistical advantage going forward. If they are reused, then they can tell the real spend by discarding any spend that's been used more than once. Is that correct?

I can't help but believe this is part of something larger, along with all the previous attacks in the last 2 years and now Samourai, Liquid pulling out of US, attacks on tor, RISAA and mandatory KYC on US cloud providers and domain registrars...

On that note, we've known LE has tools for years now (Chainalysis and 1-2 others) that can in some circumstances give a lead on who a target is, likely via statistical analysis. The tools are only available to law enforcement so the methods aren't known. My thoughts are, in no particular order:

They run or have compromised a lot of 'activist' nodes and xpubs are sent to the nodes in light wallets, unsure if this is how it works, or if that was unique to Samourai's whirlpool design. If this was the case, light wallets use currently online available servers, so chances are a user connects their wallet to tens of servers. Users who run their own nodes would be unaffected but I think the majority of monero users use light nodes.
They have tools that monitor public ledger chains, and watch the amounts in/out. You use an exchange service to trade $500 of BTC to XMR, the amounts (fees included) are correlated over time, leading to known persons selling via KYC services. Probably least likely option but unsure how XMR works in depth.
They run and/or work with (gag order) no-KYC major services that would have that information, as well as other more 'centralized' helpful no-KYC exchange services that know exactly what amount and address the funds are going to and where they came from.

8 comments

Again, thanks to the featherwallet's developer for theses fixes. All theses investigations about spam attacks really shows how membership proof implementations matters
- In version 2.6.7 you can send the same coin without waiting for the 10 block confirmation.
  I can't reproduce this. The 10 block lock-time is a consensus rule, so the transaction would get rejected by the network. Is that what happened?
  
  Oops I was wrong, it looks like I have a penny left in the sending wallet so it was just a lucky coincidence on amounts sent during testing.
  
  I just noticed your username. Thank you for the project, Feather is amazing. I have a question though. I know Ruckinum ran an analysis and thinks this is not a black marble flood, but I can't help but think it's a way go somehow break the anonymity of monero, whether just sent amounts, or received amounts, which would still give a wealth of information.
  I don't believe this is a random (D)DoS/spam. This is a deanon attack. I know it in my gut. I don't know enough about the internals of monero but I think you might.
  Specifically...
  The bug was triggered when the number of RingCT outputs on the blockchain exceeded 100 million
  For instance, this transaction was constructed using a manipulated output distribution. Can you determine what the true spend is? Notice that all ring members are older than 1y 200d except for one 6-day-old output. Unless the user checks the ring on a block explorer and knows what to look out for, they would not notice that their transactions are being fingerprinted.
  My understanding is that the 16 (or 15+real?) rings are all real, prior transactions. Are the transactions reused? If not, then they exhaust the supply of rings and now have great statistical advantage going forward. If they are reused, then they can tell the real spend by discarding any spend that's been used more than once. Is that correct?
  I can't help but believe this is part of something larger, along with all the previous attacks in the last 2 years and now Samourai, Liquid pulling out of US, attacks on tor, RISAA and mandatory KYC on US cloud providers and domain registrars...
  On that note, we've known LE has tools for years now (Chainalysis and 1-2 others) that can in some circumstances give a lead on who a target is, likely via statistical analysis. The tools are only available to law enforcement so the methods aren't known. My thoughts are, in no particular order:
  They run or have compromised a lot of 'activist' nodes and xpubs are sent to the nodes in light wallets, unsure if this is how it works, or if that was unique to Samourai's whirlpool design. If this was the case, light wallets use currently online available servers, so chances are a user connects their wallet to tens of servers. Users who run their own nodes would be unaffected but I think the majority of monero users use light nodes.
  They have tools that monitor public ledger chains, and watch the amounts in/out. You use an exchange service to trade $500 of BTC to XMR, the amounts (fees included) are correlated over time, leading to known persons selling via KYC services. Probably least likely option but unsure how XMR works in depth.
  They run and/or work with (gag order) no-KYC major services that would have that information, as well as other more 'centralized' helpful no-KYC exchange services that know exactly what amount and address the funds are going to and where they came from.
thank you