Live data from Hacker News

LastPass breach: The significance of these password iterations

palant.info

11–17 of 17 posts

Re: LastPass breach: The significance of these password iterations

#11
post #8

>GTX 1080 Ti graphics card (cost factor: less than $1000) can be used to test 346,000 guesses per second. >GeForce RTX 4090 graphics card could test more than 88,000 guesses per second! Guessing we're missing a zero there?

Disclaimer: I’m the author of this article.

I don’t think so. 346,000 guesses four years ago was for the old default: 5,000 iterations. 88,000 guesses is on more powerful hardware but with the new default: 100,100 iterations.

Re: LastPass breach: The significance of these password iterations

#12

The article is missing key data. The password iterations that are set low are client side. The server side is different. The writer of the article needs to retract. https://support.lastpass.com/help/about-password-iterations-...

Disclaimer: I’m the author of this article.

I’m not missing anything. It’s LastPass who finally need to retract this article. I proved back in 2018 that server-side iterations are misimplemented and have no security effect. That’s why they increased the client-side value in the first place. See https://palant.info/2018/07/09/is-your-lastpass-data-really-...

Re: LastPass breach: The significance of these password iterations

#13
post #6

Can attackers can easily tell the 1 and 500 iteration databases and focus their resources in breaching those ones?

Disclaimer: I’m the author of this article.

Yes. The number of iterations is presumably stored in the same customers database that they stole. Even if not: the number of iterations can be queried via a public API, anyone can do it if they know the email address.

Re: LastPass breach: The significance of these password iterations

#14
post #10

As I feared, I changed the iterations myself at some point, and they never "migrated" it to the new value. So it's above the old default, but well below the recommended number of iterations. I don't suppose it being a non-obvious value makes it any more secure? Is an attacker brute forcing the thing likely to try obvious default values first and then give up if they don't work? Or will they simply +1 the iteration co…

Disclaimer : I’m the author of this article. No, the iteration count is no secret. It’s even exposed via a public API, anyone can query it if they know the email address.

…why would you even do that?

Re: LastPass breach: The significance of these password iterations

#15
post #10

Earlier quoted context omitted.

Disclaimer : I’m the author of this article. No, the iteration count is no secret. It’s even exposed via a public API, anyone can query it if they know the email address.

…why would you even do that?

The number of iterations is needed for login. The user enters their email address and password, and the app needs to know (before they actually log in) how many iterations to apply. There are approaches like the OPAQUE protocol which avoid having the iterations count in the open, but LastPass didn’t implement that. To their defense, OPAQUE is relatively new.

Re: LastPass breach: The significance of these password iterations

#16
post #11
post #8

>GTX 1080 Ti graphics card (cost factor: less than $1000) can be used to test 346,000 guesses per second. >GeForce RTX 4090 graphics card could test more than 88,000 guesses per second! Guessing we're missing a zero there?

Disclaimer : I’m the author of this article. I don’t think so. 346,000 guesses four years ago was for the old default: 5,000 iterations. 88,000 guesses is on more powerful hardware but with the new default: 100,100 iterations.

Ah gotcha.

Re: LastPass breach: The significance of these password iterations

#17
post #10

As I feared, I changed the iterations myself at some point, and they never "migrated" it to the new value. So it's above the old default, but well below the recommended number of iterations. I don't suppose it being a non-obvious value makes it any more secure? Is an attacker brute forcing the thing likely to try obvious default values first and then give up if they don't work? Or will they simply +1 the iteration co…

Disclaimer : I’m the author of this article. No, the iteration count is no secret. It’s even exposed via a public API, anyone can query it if they know the email address.

Well, balls. Thanks for confirming.
Post reply on HN