-
Notifications
You must be signed in to change notification settings - Fork 296
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ibc: add withdrawal timestamp rounding chain rule #4065
Conversation
af17be2
to
24c161d
Compare
@@ -77,6 +77,15 @@ impl Ics20Withdrawal { | |||
anyhow::bail!("timeout time must be non-zero"); | |||
} | |||
|
|||
// in order to prevent clients from inadvertantly identifying themselves by their clock | |||
// skew, enforce that timeout time is rounded to the nearest even second | |||
if self.timeout_time % 2_000_000_000 != 0 { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What's the motivation for rounding to 2-second intervals rather than 1-second ones?
I might have miscommunicated in Discord, I was thinking of "even second" as meaning "on the second boundary" -- but looking at this I'm wondering what the right resolution should be. 1s is an "easy" answer, but what about 10 or 60, or greater?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't have a good intuition for 'what amount of rounding covers the skew of >99% of machines', but I think single seconds might not be enough. From a user's standpoint quantizing withdrawal timeouts to minutes seems fine, so maybe 1 minute would be better
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's set two rules:
- in consensus, require that timeout timestamps are truncated to one-minute boundaries
- in our clients, set the default timeout timestamp to be truncated to 1h boundaries
This leaves the possibility for someone who has a pressing need for finer grained timestamps to do so while setting a good default
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
For posterity, what's the reasoning behind baking a privacy setting like this into consensus? Is it to prevent mistakes with alternative clients?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, the "obvious" thing to do is set the timestamp as Utc.now() + timeout, and the existence of this rule means that that code will immediately break.
Updated the chain rule to require rounding to 1 minute, updated pcli to round to 1 hour |
@ValarDragon comments:
|
Changed the pcli default to round to nearest 10 minute, per suggestion |
this is a consensus-breaking change
Fixes #4064