-
Notifications
You must be signed in to change notification settings - Fork 146
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
Adds decaf group #115
base: master
Are you sure you want to change the base?
Adds decaf group #115
Conversation
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.
Mostly looks good to me, most of my comments relate to style.
sign/ed448/ed448.go
Outdated
_ = goldilocks.Curve{}.ScalarBaseMult(s).ToBytes(pair.public[:]) | ||
P := &goldilocks.Point{} | ||
goldilocks.Curve{}.ScalarBaseMult(P, s) | ||
enc, _ := P.MarshalBinary() |
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.
Error being ignored intentionally?
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.
Encoding always succeeds, but Go's standard MarshalBinary interface doesn't care. Maybe it's worthwhile having a separate PackInto
function that writes into a buffer instead of returning it. Saves an allocation and then you don't have an error to confuse people about.
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.
Since this error will not occur, added a panic if it happens. Not the best way to handle this error.
I will consider adding PackInto.
ecc/goldilocks/point.go
Outdated
fp.Sub(u, u, &one) // u = y^2-1 | ||
fp.Sub(v, v, &one) // v = dy^2-1 | ||
isQR := fp.InvSqrt(&P.x, u, v) // x = sqrt(u/v) | ||
if !isQR { |
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.
This is not constant time.
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.
updated take a look
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.
Looks good.
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.
It might be hard though to write constant time algorithms using this function as it returns an error instead of an int. In go-ristretto I have two variants: Equals
and EqualsI
. The former returns the idiomatic bool
and the latter an int
, which allows one to write constant time code.
ecc/goldilocks/twist.go
Outdated
var isZero int | ||
if k.IsZero() { | ||
if kk.IsZero() { |
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.
Not constant time.
I'll check this tomorrow @armfazh |
Nice! Have you run it against the vectors of: https://sourceforge.net/p/ed448goldilocks/code/ci/master/tree/test/ ? |
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.
We should talk more about where the formulas come from and how to make clear the code matches them.
ecc/goldilocks/curve.go
Outdated
// Double returns 2P. | ||
func (Curve) Double(P *Point) *Point { R := *P; R.Double(); return &R } | ||
// Double calculates R = 2P. | ||
func (Curve) Double(R, P *Point) { *R = *P; R.Double() } |
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.
This is a change to the external API of the package.
93a4a24
to
c97b01e
Compare
f120857
to
ec347f5
Compare
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'll continue my review next week. I can't concentrate on the maths with this heat wave.
ecc/decaf/decaf.go
Outdated
func IsValid(a *Elt) bool { return ted448.IsOnCurve(&a.p) } | ||
|
||
// Identity returns the identity element of the group. | ||
func Identity() *Elt { return &Elt{ted448.Identity()} } |
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 think the Go compiler gets rid of the allocation here.
@bwesterb Addressed requested changes. |
We should add here the draft-ristretto-decaf test vectors, which is pretty easy as they are the multiplicatives of the generator. We can also add here the one-way-map of the draft. I also did a small implementation here of it: https://github.com/claucece/sage-ristretto255-decaf448 in sage. |
ecc/goldilocks/decaf.go
Outdated
} | ||
|
||
s := &fp.Elt{} | ||
copy(s[:], b[:fp.Size]) |
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.
Mmm.. is there a reason why this constrain should be enforced b[:fp.Size]
?
@armfazh what's the plan for this PR? |
Just pushed changes that update this PR. |
It's been a while back, but I believe I haven't gone through all the maths in detail. If you want a detailed review, then I'll need to dedicate a day or two to it. Is there a deadline? |
Squashed commit of the following: commit ec347f5 Author: armfazh <[email protected]> Date: Fri Jul 24 15:28:34 2020 -0700 Adding decaf group. commit 796b37e Author: armfazh <[email protected]> Date: Wed Jul 22 19:44:07 2020 -0700 Updates internal packages of Ed448. commit f5957e7 Author: armfazh <[email protected]> Date: Wed Jul 22 19:11:15 2020 -0700 Updating Decaf documentation. commit 6f573e6 Author: armfazh <[email protected]> Date: Wed Jul 22 13:56:12 2020 -0700 Updating documentation of ted448 internal package. commit 7e80bfc Author: armfazh <[email protected]> Date: Wed Jun 10 01:56:31 2020 -0700 Adapting ed448 for using the internal ted448 package. commit 44ce6c5 Author: armfazh <[email protected]> Date: Tue Jun 9 23:36:17 2020 -0700 ted448 point public fields. commit 13dd30d Author: armfazh <[email protected]> Date: Tue Jun 9 20:32:32 2020 -0700 Moving twist implementation to an internal package. commit 4cdcb71 Author: armfazh <[email protected]> Date: Mon Jun 1 01:30:22 2020 -0700 Solving scalar constant-time operations. commit 26b9ea4 Author: armfazh <[email protected]> Date: Thu May 21 09:21:36 2020 -0700 Review comments on formulas. commit ff821b5 Author: armfazh <[email protected]> Date: Tue May 19 16:54:36 2020 -0700 Adding tests for detecting decaf/point invalid encodings. commit 3881ea8 Author: armfazh <[email protected]> Date: Fri May 15 15:13:53 2020 -0700 One test for decaf, unmarshaling straight-line code, and check for errors. commit 9b048c0 Author: armfazh <[email protected]> Date: Thu May 14 18:12:57 2020 -0700 Updating interface for decaf and curve. commit a99153c Author: armfazh <[email protected]> Date: Thu May 14 16:10:04 2020 -0700 Adding goldilocks documentation. commit a62d6bd Author: armfazh <[email protected]> Date: Thu May 14 14:15:02 2020 -0700 Adding decaf v1.1 and kat tests. commit c72bdfa Author: armfazh <[email protected]> Date: Wed May 13 13:30:11 2020 -0700 Removing non-used fp functions. commit d4fc865 Author: armfazh <[email protected]> Date: Tue May 12 11:47:57 2020 -0700 Adding some helper functions. commit 6173c83 Author: armfazh <[email protected]> Date: Tue May 12 05:15:55 2020 -0700 Decaf decoding working. More tests needed. commit daa36c1 Author: armfazh <[email protected]> Date: Mon May 11 16:16:16 2020 -0700 Decaf encoding is working, except by the choice of generator. commit 8933f6c Author: armfazh <[email protected]> Date: Thu May 7 01:19:47 2020 -0700 Decaf encoding requires cannon sqrt. commit 9479b45 Author: armfazh <[email protected]> Date: Wed Apr 29 14:51:21 2020 -0700 Adding support for decaf quotient group.
Includes a package for the decaf group.
Decaf is specified https://datatracker.ietf.org/doc/html/draft-irtf-cfrg-ristretto255-decaf448-03, this implementation is fully compatible with draft. Test vectors for group operations and hashing to group are passing.
Decaf (circl/group) is a safe group abstraction on top of the Goldilocks (edwards448) curve (circl/ecc/goldilocks).
Ed448 also uses Goldilocks curve but only requires specific operations.
Internally, Goldilocks curve is implemented with ted448, a twist of the curve, since it provides faster point arithmetic.