This week’s newsletter includes action items and news related to last week’s security release of Bitcoin Core 0.16.3 and Bitcoin Core 0.17RC4, popular questions and answers from the Bitcoin StackExchange over the past month, and short descriptions of notable merges made to popular Bitcoin infrastructure projects.
● Upgrade to Bitcoin Core 0.16.3 to fix CVE-2018-17144: as widely reported early Friday (UTC), the denial-of-service vulnerability described in last week’s Optech newsletter is now known to allow miners to trick affected systems into accepting invalid bitcoins.
As of this writing, it’s believed that a majority of large Bitcoin services and miners have upgraded, likely ensuring that any blocks exploiting the bug will be quickly reorganized out of the most proof-of-work chain—reducing the risk for SPV clients and non-upgraded nodes.
If you don’t plan to upgrade or if you use an SPV client, you should consider waiting for more confirmations than you usually do (30 confirmations—about 5 hours worth—is a normal recommendation in these sort of situations, as that’s enough time for people to notice a problem and get warnings published). Otherwise, upgrading to one of the following versions remains highly recommended for any system, especially those systems handling money:
● Allocate time to test Bitcoin Core 0.17RC4: Bitcoin Core has uploaded binaries for 0.17 Release Candidate (RC) 4. Testing is greatly appreciated and can help ensure the quality of the final release.
● CVE-2018-17144: the initial and subsequent disclosures of information about this bug were the only significant news this week. For more information, we suggest reading the following sources:
We’re aware of several very insightful people currently reflecting upon the bug, its ultimate causes, and possible methods for reducing the risk of future serious bugs. An especially good venue for Bitcoin Core internal discussion will be during the October 8th though 10th CoreDev.tech meetings following the Tokyo Scaling Bitcoin conference. We plan to follow up with links to any significant conclusions that are published.
Optech thanks the original reporter, Awemany, for his responsible disclosure as well as the following developers who unhesitatingly made the time to quickly confirm the issue, address it, and quietly provide round-the-clock monitoring for attempts to exploit the then-undisclosed inflation risk: Pieter Wuille, Gregory Maxwell, Wladimir van der Laan, Cory Fields, Suhas Daftuar, Alex Morcos, and Matt Corallo.
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 of time to help answer other people’s questions. In this monthly feature, we highlight some of the top voted questions and answers made since our last update.
● How does CVE-2018-17144 work? Andrew Chow provides a detailed explanation of how Bitcoin Core can be crashed or tricked into accepting multiple spends of the same input in versions vulnerable to this bug.
● Why doesn’t Bitcoin use UDP instead of TCP? Gregory Maxwell describes a case where important Bitcoin software does already use UDP and then details the reasons why UDP support isn’t implemented in popular full node software. He concludes with a description of some potential benefits that might be available if UDP support was implemented.
● How likely are you to get blacklisted by an exchange if you use Wasabi wallet’s CoinJoin mixing? Wasabi Wallet author Adam Ficsor explains that nothing stops exchanges from refusing funds mixed through Wasabi, but that several features of Wasabi (such as a required anonymity set of 100) can help make blocking users bad for business. Alternatively, he links to a tool that may allow users to circumvent an address blacklist.
● What’s the minimum number for a Bitcoin private key? Answers from Mark Erhardt and Gregory Maxwell were provided within a minute of each other, but a humorous rephrasing of Maxwell’s answer by Nate Eldredge has more upvotes than either answer as of this writing.
Notable commits this week in Bitcoin Core, LND, and C-lightning. Reminder: new merges to Bitcoin Core are made to its master development branch and are unlikely to become part of the upcoming 0.17 release—you’ll probably have to wait until version 0.18 in about six months from now.
● Bitcoin Core #13152: when connected to the peer-to-peer network, nodes share the IP addresses of other nodes they’ve heard about and these addresses are stored in a database that Bitcoin Core queries when it wants to open a new connection. This PR adds a new RPC command,
getnodeaddresses, that returns one or more of these addresses. This can be useful in conjunction with tools like bitcoin-submittx.
● LND #1738: the logic for validating channel updates has been moved to the routing package so that it’s available both in routing (to handle failed payment sessions) and the gossiper (where it was handled before). This fixes issue #1707 (and implements a test case for it) that may have allowed a node to trick one of its peers into believing a different peer had a routing failure, thus possibly redirecting traffic to the malicious node.
● C-Lightning #1945 now provides a
gossipwithtool that allows you to receive gossip from a node independently of lightningd or even to send the remote node a message. This tool is used for additional testing of lightningd’s gossip component.
● C-Lightning #1954 now complies with updates to BOLT7 by splitting the previous
flagsfield for the
listchannelsRPC into two new fields:
channel_flags. Also code comments and references to BOLT2 and BOLT11 have been updated.
● C-Lightning #1905 has significantly expanded the in-code documentation of its secrets module. The documentation is remarkably good (and, at times, quite humorous). See hsmd.c. The code comments even document other code comments:
/*~ You'll find FIXMEs like this scattered through the code. * Sometimes they suggest simple improvements which someone like * yourself should go ahead an implement. Sometimes they're deceptive * quagmires which will cause you nothing but grief. You decide! */ /* FIXME: We should cache these. */ get_channel_seed(&c->id, c->dbid, &channel_seed); derive_funding_key(&channel_seed, &funding_pubkey, &funding_privkey);