Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Critical Update on DAO Vulnerability

blog.ethereum.org

Post reply on HN
Print view View original

91–100 of 629 posts

Re: Critical Update on DAO Vulnerability

#93

Post by forgotpwtomain »

brohee wrote: post #81
lazaroclapp wrote: post #71

Earlier quoted context omitted.

I think they could also be shorting ETH, in dollars/euros/rubles/yuans, in which case destroying the currency would be excellent business for them, even if they can't extract what they stole.

Any (presumably buoyant) entity actually offering that possibility?

Poloniex seems to be the exchange with the largest volume (I somehow doubt anyone is buying long-positions larger than that) - so it looks like they could at least cash out several thousand bitcoins - although it's been a while since I've been on their site and if they offer API access some of those positions might be automatically diverted as the price moves.

[0] http://coinmarketcap.com/currencies/ethereum/#markets

[1] https://poloniex.com/exchange#btc_eth

Re: Critical Update on DAO Vulnerability

#94

Post by benmmurphy »

I'm not expert on etherium code (just started looking at it now) but it looks like the DAO didn't look for similar issues with the latest security fix.

https://github.com/slockit/DAO/commit/f01f3bd8df5e1e222dde62...

             reward = rewardAccount.balance
but if you grep payOut then you see a similar broken pattern where it modifies the state after the call instead of before it.

        if(_toMembers) {
            if (!DAOrewardAccount.payOut(dao.rewardAccount(), reward))
                throw;
            }
        else {
            if (!DAOrewardAccount.payOut(dao, reward))
                throw;
        }
        DAOpaidOut[msg.sender] += reward;
but apparently this is not how the DAO is being drained because there are no rewards at the moment.

this is a good summary of the problem:

https://blog.ethereum.org/2016/06/10/smart-contract-security...

and should scare you about the security of smart contracts base on etherium.

EDIT: mm.. maybe it is safe because the addresses dao/dao.rewardAccount() can't be controlled by attackers

Re: Critical Update on DAO Vulnerability

#95

Post by a2800276 »

nneonneo wrote: post #30

The provided link is just a page showing a bunch of transactions. For someone like me, who is not so intimate with the Ethereum terminology in use (but who is still interested in the DAO, as an observer), could someone provide a layman's explanation of what's going on? Somewhat more specifically, I'm wondering the following: - At a high level, what does this attack actually consist of? - How does ethereum "go missing…

> - At a high level, what does this attack actually consist of?

At a high level, The Dao is like Pokemon. Casual observation suggests that it may be a more or less consistent world internally and that a lot of people are very excited about it. A few of those people even claim to fully understand how it works. But there's a lot of fat guys with acne blindly spending on Pokemon cards thinking it's a wise investment.

Unfortunately, the fans are so excited that they jump up and down clapping all the time and forget to breath while they're talking. This makes it very difficult to understand what they're trying to say. If you manage to understand, you will hear something like:

"Most Pokémon have only one type. However, EX Team Magma vs Team Aqua introduced Dual-type Pokémon, which have two different types. For a while, all existing dual-type cards had either Darkness or Metal as their secondary type, with the exception of certain Pokémon cards with the Dual Armor Poké-Body, such as Medicham from the EX Crystal Guardians expansion, which can have multiple types when certain energy are attached."

Now what possibly happened during the attack is that Pokemon cards are spontaneously catching fire! This is particularly troubling, because Pokemon just came out with a super-duper fire defence card and told everyone to buy it for lots and lots of money. And everybody did.

Now while all of this is clearly very troubling for heavily invested Pokemon fans. But in the grand scheme of things, well a butterfly flapped it's wings in China: there's a theoretical possibility that this affects anything, but in practice no one but fans cares.

> - How does ethereum "go missing" in a distributed blockchain, where you can see all the transaction endpoints?

1. keep the exp you earned 2. lose some money 3. wind back at the pokemon center at the pokemon league 4. have to start over on the e4

> Who loses and who gains from an attack of this scale?

Pokemon fans, obviously.

> How severe could this attack be - does it pose an existential threat to The DAO (or Ethereum, more broadly)?

Synchronoise when used by Pokémon like Umbreon. It deals damage to every Pokémon that shares the user's type, so in Umbreon's case it hits every Dark type Pokémon. Only it doesn't, because it's a Psychic move and always fails.

Re: Critical Update on DAO Vulnerability

#96

Post by malux85 »

jpereira wrote: post #84
narrator wrote: post #74

I always though that Etherium had a huge attack surface. Each script has to be security audited, etc. That's the thing about Bitcoin. It's as simple as possible while still being secure and useful and has been beat up and audited by the best security pros in the world. Distributed Systems are not easy. Secure distributed systems with Byzantine fault tolerance are even harder. Etherium is just trying to do too much.

Doesn't this show an issue with the Distributed Systems on Ethereum, with every script that has to be audited individually, and not with the platform itself? I'm with you on the fact that proper auditing is an absolute must, as this DAO fiasco shows, but I don't think this event exposes any flaws in the Ethereum platform itself.

Agreed - this is a problem with the DAO, not with the ethereum network itself.

Re: Critical Update on DAO Vulnerability

#97

Post by modeless »

narrator wrote: post #74

I always though that Etherium had a huge attack surface. Each script has to be security audited, etc. That's the thing about Bitcoin. It's as simple as possible while still being secure and useful and has been beat up and audited by the best security pros in the world. Distributed Systems are not easy. Secure distributed systems with Byzantine fault tolerance are even harder. Etherium is just trying to do too much.

I've held a similar opinion, but lately with the run up in the Ethereum price I was worried that I was missing something other people saw. Apparently this isn't a vulnerability in Ethereum itself, but it may still have serious consequences for the network.

There are plenty of proposals to add complexity to the Bitcoin system. I hope that people promoting those proposals pay attention to this example of the problems complexity will inevitably cause.

Re: Critical Update on DAO Vulnerability

#98

Post by tinkerrr »

jpereira wrote: post #86
tinkerrr wrote: post #78

Earlier quoted context omitted.

Also remember that this isn't the first time this idea of 'lets rollback a blockchain due to a hack/attack' has been floated or even tried. The first major blockchain rollback almost completely destroyed the cryptocurrency [1]. Since then, some major hacks have happened and the community/developers rejected the idea of a rollback [2][3]. That was a hard lesson, and hopefully the current developers will learn from the…

One could argue that at this point in its history it stands to benefit more from that efficiency than it would from more robust decentralization. When adoption is low it stands to reason centralization would be high, and so it enables them to move faster to increase adoption, and hence lower centralization as other players and stakeholders enter the game. However it does also stand the risk of allowing the current in…

There is no doubt that centralization is more efficient than decentralization. The problem comes when developers or other stakeholders don't understand or underplay this feature.

Also, centralization/decentralization is a a scale, not discrete boolean values.

It will definitely be interesting to see how this plays out. Good luck to everyone involved.

Re: Critical Update on DAO Vulnerability

#100

Post by rlpb »

joosters wrote: post #26

Just remember, when the developers inevitably appear with suggestions about how to stop the hack, roll back the blockchain, or come up with other schemes to block the hackers, they are showing everyone that all the talk of blockchains being decentralised, or being beyond the control of governments or other powers... is a complete lie. If this hack can be stopped, then it demonstrates that the currency can be manipula…

If there is consensus that this is a problem and shouldn't be allowed, then there presumably is also consensus for even a hard fork if necessary. As an outside observer, I don't see why this is a demonstration that the decentralised system can be manipulated, is not fault tolerant, or censored.

Since if this is a problem and will be fixed, it will presumably be by consensus. We already know that with consensus the entire system can be changed (or forked, if you like). This is no secret and there was never any claim otherwise. This is exactly what they mean by decentralised.

What should worry you is if a change happens without consensus, but there is no such indication here.

Post reply on HN