Quote from: Lucius on December 28, 2018, 02:47:19 PM

Interesting video, I have to admit I looked at the part which show Flashing the Ledger Nano S with custom firmware just because I use that HW. In this part of video we can see that is possible to flash Nano S with custom firmware, and in case they presented we see that instead HW you can turn on this device in miniature game console and play game snake.

Yet this is no threat that can affect current users since requires physical access to the device, but it show that Ledger still have no solution to prevent that device is flash with custom firmware. So if hackers find way to trick users with false firmware update, it is possible that this could be one of the vectors of the attack.

The worst possible scenario : Hackers hack official Ledger site, add fake firmware and try to get as many users as possible. Maybe it's not a true comparison, but who could have imagined a few days ago that hackers will use original Electrum wallet to steal hundreds, and probably thousands of BTC? [...]

The problem is less with being able to flash the Nano S with custom firmware, but rather with flashing the Nano S with custom firmware without the device noticing and warning the user. In this regard I have to tip my hat to SatoshiLabs that at least their firmware check was solid enough as to force these researchers to resort to a rather sophisticated attack on the hardware level (for what little good it brought, in the end). So at least in SatoshiLabs' case the scenario of hacking the update server and deploying malicious firmware appears to be non-viable.

Still, rather worrying, especially given the fact that for the Ledger Nano S an attack on the software level was sufficient. I think the Ledger vulnerabilities should be fairly straightforward to fix, about the Trezor One I'm not so sure, given the complexity of the issue. Worse still I wouldn't be surprised if one could mount a similar hardware-level attack on Ledger devices.

Regardless of would-be attackers requiring physical access to the device I still wouldn't shrug it off as a practical non-threat. Obviously once an attacker is able to attain physical access to your hardware wallet you'll likely have more acute problems than firmware integrity (ie. getting a "memory dump" from you, as a person, is likely more trivial than getting one from your hardware wallet). However at least to me personally results like these mostly serve as a stark reminder of how hard it is to get security right (ie. if it's possible to break the most popular, trusted and peer-reviewed hardware wallets, I don't even want to think about the rest of the market).

Quote from: gentlemand on December 28, 2018, 12:34:58 PM

[...]

The main thing I took away from it is using a 25th password saves you from quite a few sad outcomes.

I guess that depends on the attack vector. If the firmware itself is compromised, the 25th password is likely to get compromised as well. It definitely protects against memory dumps as described in the Trezor One attack though -- or at least it should buy enough time to move your coins before the attacker can access them.

---

Come to think of it, I'm now really worried about Ledger's update server getting compromised. I don't think compromising Ledger's update servers would be easy, especially unnoticed, but as long as their wallet's bootloader can be tricked an attack scenario as described by Lucius would allow for remotely compromising Ledger hardware wallets without direct physical access O.o