With MultiversX’s chain outage, everyone is praising the fact that “no funds were lost.” What I care more about is the timeline.
The postmortem that MvX developer Andrei Marinica shared on X was very straightforward: on the morning of 11/17, an address deployed 23 contracts in a row on shard 0, targeting an unpatched bug in their own fork of Wasmer. The team believed the attacker wanted to run arbitrary code on validators’ machines, potentially reaching their private keys in the worst case, so they asked all validators to shut down—and the chain went down.
Put the dates side by side:
April 30: The fix had already been submitted upstream to Wasmer, but hadn’t made it into a formal release.
May 12: MvX received a report about this stack pointer bug. They couldn’t reproduce it at the time, and the first fix was still wrong, so it was never released.
November 17: Someone actually attacked. It was discovered at 10 a.m. (EEST), the fix was written by 11 a.m., and a new version was ready by 4 p.m.
The same patch took six months to deal with under normal circumstances, but was done in six hours under attack. This wasn’t a question of technical difficulty; it was a question of priority.
The attack surface this time was also different from what people are used to. It wasn’t someone exploiting faulty logic in a DeFi contract. Instead, the contract planted machine code disguised as a numeric constant in memory during compilation, directly attacking the node process. According to the postmortem, the payload was meant to install a hook in memory and impersonate the owner of some contract. The team doesn’t even know which contract was being targeted. Audit reports cover contracts; they can’t tell you how far behind upstream your underlying fork of the compiler is.
I’m not entirely convinced by claims like “decentralization held up.” What held things together was a single word from the core team getting validators to shut down in unison. An X user posted that Bithumb had also paused EGLD withdrawals. It was the right call to shut things down this time, but it shows who holds the switch when an emergency hits.
Also, for now, the claims that “no funds were lost and there’s no need to roll back” are only what I’ve seen the team say about itself. Hexens’ research says the affected component was the Singlepass compiler in Wasmer versions 2.1.0 through 6.x, which is widely used in blockchains. If you hold tokens on chains that use their own Wasm virtual machines, it’s worth checking who maintains their VM and how many versions behind upstream they are.
#EGLD #MultiversX