pull down to refresh

Monero Weekly #4
1 mXMR -5 mXMR \ 0 comments \ @untraceableflame 30 Oct monero

Monero Research Lab meeting - October 28, 2026Monero Research Lab meeting - October 28, 2026

The Monero Research Lab met on Wednesday, October 28, chaired by Rucknium. In updates, jberman is on FCMP++ integration followups, Rucknium on more Nyx eclipse-attack defense analysis, gingeropolous on Share-or-Perish and antiproxy research in monerosim, and vtnerd on SSL work plus lws wallet-backup review, with another SSL patch still needed. The main crypto item was #11586: get_tx_key/check_tx_key doxes the sender in 2-output transactions, where jeffro256's spec (Carrot PR #16) and implementation (monero PR #9559) are now ready for review. InProofs keep working for incoming payments and OutProofs for external outgoing payments; the trade-off lands on self-sends - internal enotes lose OutProofs and their InProofs take a different form that leaks whether the enote is internal. Nothing prevents making external enotes to yourself in transactions with more than 2 outputs to keep both proof types, a workaround Rucknium called a good outcome.

On FCMP++, jberman said the integration PRs are ready for review with followups implemented, audit quotes are being fielded with final discussion expected next week, and he personally hopes all code is ready in Q1. An unproven idea to use Bulletproofs++ for roughly 50% smaller membership proofs (a PoC exists) was parked as possibly interesting for a followup fork after Aaron's earlier reservations, with jeffro256 arguing it is better to move forward and swap in more efficient modules later than to wait on perfect solutions. The FCMP beta stressnet v3.1 shipped with the inflation patch for the 24-hour fork window, GUI fixes, faster scanning and some DoS prevention; stress testing stays "relaxed" for now, with Rucknium holding off on spamming until jberman reviews tx relay v2 logs from the v3 beta high-stress window, and the unusually high alt-chain count from the last stress run (#508) on the checklist for next time.

The P2P item ran longest: simulating the Shi et al. Nyx eclipse attack, Rucknium's patch (increasing white_list size and changing connection selection, PR #11527) defeats the attack, with the reassuring caveat that an adversary needs IPs spread across separate /24 subnets - at 10 nodes per /24 even unpatched nodes with the existing subnet deduplication do OK. ofrnxmr presented a broader companion patch set (whitelist bucketed into 256×16 and graylist into 256×32, /16 dedup instead of /24, test-before-evict, one outbound peer per netgroup) that performs roughly the same in the reviewed scenarios while aiming to handle more diverse adversary strategies; vtnerd supports landing a quick fix but flagged that the approach leans on IPv4 scarcity, and boog900 noted the paper demonstrated a real attack on the seed nodes, making a stopgap worthwhile while bigger p2p work proceeds. On Share or Perish (selfish mining mitigation), sims show SoP defeats the classic selfish miner, but bot-discovered attack strategies reduce its efficacy and a network with multiple honest mining nodes may push the math toward the attacker; Rucknium questioned the MDP methodology (an exact MDP should be an upper bound on the optimal adversary, and no solver library was imported) and gingeropolous will update the notes. A side debate over hashrate-weighted node selection was repeatedly deferred until its proposer writes it up.

News roundupNews roundup

Sources: Monero Observer · MRL meeting logs

write
compose
related posts