Bitcoin Optech Newsletter #103

This week’s newsletter summarizes a newly published fee ransom attack against LN users, links to continued discussion about the attack against LN atomicity, and shares a reminder about collision attacks on RIPEMD160-based addresses in multiparty protocols. Also included are our regular sections with popular questions and answers from the Bitcoin StackExchange, a list of releases and release candidates published this week, and notable changes to popular Bitcoin infrastructure projects.

Action items

None this week.

News

  • LN fee ransom attack: René Pickhardt publicly disclosed a vulnerability to the Lightning-Dev mailing list that he had previously privately disclosed to LN implementation maintainers almost a year ago. In the current LN protocol, each time the channel state is updated, the party who initiated opening the channel must commit to paying any onchain transaction fees for the unilateral close transaction. The party paying the fees also gets to choose the feerate used—but the counterparty’s security also depends on the feerate being appropriate. That means the counterparty may close the channel at any time if they think the chosen feerate has become too low for current market conditions. To avoid such unnecessary closes, BOLT2 gives an example of the fee-paying party selecting fees five times higher than the minimum reasonable estimate—high enough that the counterparty should be satisfied even if they’re using a different fee estimation algorithm.

    BOLT2 also allows channels to route up to 483 payments simultaneously, each of which requires a 43 vbyte P2WSH output, for a total of about 20,000 vbytes of data that needs to be added to the chain relatively quickly—meaning it may need to pay a high feerate. If that feerate is five times higher than strictly necessary, this can easily result in paying more than $100 USD in transaction fees at current bitcoin prices. Additionally, if the commitment transaction is confirmed, the HTLCs then need to be settled (again using time-sensitive transactions that may need to pay a high feerate). If the victim was the party routing those payments outbound, they’ll need to pay an additional transaction fee to recover each payment, which could make the attack two or three times more costly for them.

    Fees are paid to miners, so there’s no direct motivation to perform this attack, but if the attacker has the means to quickly contact the victim, it’s possible they can offer to settle the HTLCs offchain in return for a ransom, allowing the attacker to profit.

    Pickhardt concludes his email with several ideas for addressing the problem, none of which he finds completely satisfactory. The mitigation originally implemented in Eclair and later implemented in C-Lightning (see Newsletter #59) is for LN nodes to limit the number of pending payments, keeping transactions small and total fees low. Another mitigation in development is anchor outputs, which allow feerates to be selected when the channel is closed—eliminating the need to overestimate fees in order to prevent premature channel closures. Several other ideas are mentioned, but Pickhardt asks readers to contemplate the problem and suggest any other possible solutions.

  • Continued discussion about LN atomicity attack: Bastien Teinturier posted to the Lightning-Dev mailing list with a link to a detailed description of the LN commitment protocol, its weaknesses, and proposals to address those weaknesses. The document teaches readers everything they should need to know to understand the attack against LN atomicity described in Newsletter #95 as well as several proposed mitigations. The document’s clear writing tied together several fragmented threads from previous discussions. This led to a renewed evaluation of several previously proposed solutions, including concerns about the effectiveness of the “alternative anchor proposal” and a suggested procedure to use the pay-for-signature scriptless script to trustlessly pay a third party for the final signature needed to complete an adaptor signature.

  • Reminder about collision attack risks on two-party ECDSA: cryptographer Jonas Nick replied to the Bitcoin-Dev mailing list thread about a proposed CoinSwap implementation (see Newsletter #100) reminding developers that P2PKH, P2WPKH, and P2SH addresses which use the 160-bit RIPEMD160 hash are vulnerable to collision attacks that reduce its security to 80 bits when multiple parties collaborate to create an address using a naive protocol (see our description of this weakness in legacy P2SH addresses). Although this was previously only a concern for users of P2SH multisig, it applies in contexts, such as CoinSwap, where it’s proposed that two users could share a P2PKH or P2WPKH address. It’s possible to avoid this problem, but it requires that the two-party ECDSA protocol be designed to include an extra commitment procedure, which Nick notes some two-party ECDSA protocols and implementations already do.

Selected Q&A from Bitcoin StackExchange

Bitcoin StackExchange is one of the first places Optech contributors look for answers to their questions—or when we have a few spare moments to help curious or confused users. In this monthly feature, we highlight some of the top-voted questions and answers posted since our last update.

Releases and release candidates

New releases and release candidates for popular Bitcoin infrastructure projects. Please consider upgrading to new releases or helping to test release candidates.

Notable code and documentation changes

Notable changes this week in Bitcoin Core, C-Lightning, Eclair, LND, Rust-Lightning, libsecp256k1, Bitcoin Improvement Proposals (BIPs), and Lightning BOLTs.

  • Bitcoin Core #19260 disconnects peers that send the BIP37 filterclear message if the local node isn’t accepting bloom filters (as advertised using the BIP111 NODE_BLOOM service flag). It was previously proposed that nodes starting in Initial Blockchain Download (IBD) could register as non-relaying peers to avoid receiving recent transactions while they were still downloading large numbers of blocks. When they finished syncing, they could then transition to receiving relayed transactions by sending a filterclear message. However, it was more recently proposed that this could be done instead using the BIP133 feefilter message. That eliminates any need for a non-bloom node to support the filterclear message and so this PR removes that feature.

  • Bitcoin Core #19133 adds a bitcoin-cli -generate parameter (note leading dash) to replace the functionality of the generate RPC, which was removed from Bitcoin Core in version 0.19.0.1. The new implementation avoids unnecessary dependencies between the wallet and other components. Providing a client-side alias for the former RPC is useful for manual testing and makes it easier to update broken documentation.

  • Bitcoin Core #18027 adds two options to the GUI’s File menu for working with Partially Signed Bitcoin Transactions (PSBTs): Load PSBT from file and Load PSBT from clipboard. When one of those options is clicked and a PSBT is loaded, a dialog is provided that allows the user to sign an incomplete PSBT if this wallet has its key, broadcast a complete PSBT, or copy or save the PSBT for processing with another tool such as HWI. Combined with other recent PSBT-related GUI changes (see Newsletters #74 and #82) and HWI’s own GUI, this makes it possible to use a PSBT-based process with Bitcoin Core for the first time without using any RPCs.

  • Bitcoin Core #16377 updates the walletcreatefundedpsbt and fundrawtransaction RPCs. These RPCs normally use the wallet to automatically choose which UTXOs to spend in an unsigned transaction, but they also allow the user to specify one or more UTXOs they want to spend in that transaction. Previously, if the UTXOs selected by the user weren’t enough to pay for all of the transaction’s outputs, the wallet would automatically select more UTXOs to spend. But, if a user is manually selecting UTXOs, they may have some reason they don’t want to spend additional UTXOs, so the RPCs will now fail by default if the user manually selects any UTXOs. This default may be overridden using the new add_inputs parameter to both RPCs.

  • Eclair #1461 adds several API endpoints that forward to Bitcoin Core RPCs for relaying that program’s wallet balances and other information. The goal is to make it easier to integrate Eclair with the Ride The Lightning node management dashboard.

  • Bitcoin Core #19071 adds documentation describing how developers can contribute to the new and experimental Bitcoin Core GUI repository. Pull requests related to the GUI should be made into this new repository, which will be bidirectionally synced with the main repository using the monotree development model used by the Linux Kernel Project. There are no direct user-visible changes from this split—users will still receive the GUI in the official packaged versions of Bitcoin Core or when building using --with-gui from the source code in the main repository.

    This split is an experiment designed to determine whether using different repositories for different subsystems will help people interested in a particular subsystem focus on that part of the project. For example, someone using the GitHub Watch Repository feature will receive fewer issue and PR status updates each day when watching the Bitcoin Core GUI repository rather than the main repository, making it easier for them to monitor the project. Conversely, developers watching the main project, who may not all be interested in the GUI, will no longer need to receive notifications about it. In the best case, it’s hoped that this improved focus can speed up development, which may lead to discussion about creating monotree repositories for other subsystems. In the worst case, it’s feared the split could slow down development—but, if that happens, the experiment can be easily terminated and development can return to using a single repository (monorepo).