Kupyna: implemented hashing function #621

AnarchistHoneybun · 2024-10-11T16:20:29Z

Scrapped previous PR since it ran into a lot of problems. I'm aware that #601 exists, creating this one to check against the repo's test suite as I iterate on my implementation, and potentially make it the main pr for kupyna if the other one becomes inactive.

Currently the implementation does everything the paper describes (testing leaves a bit to be desired) so step one is to clean up any linter complaints, and then proceed to making it no-std compatible.

AnarchistHoneybun · 2024-10-11T16:33:55Z

@newpavlov could you give me some idea on why the tests are failing right now? I don't think I've touched the Cargo.toml let alone the version of digest; checked and it's the same version as that present in some other toml's. Any idea why this could be happening?

kupyna/Cargo.toml

kupyna/LICENSE

kupyna/README.md

kupyna/src/main.rs

kupyna/src/sub_units/t_xor_plus/tests/test_individual_layers.rs

kupyna/.gitignore

kupyna/Cargo.toml

AnarchistHoneybun · 2024-10-11T18:11:33Z

Might be a bit unrelated but when I try cargo run on my local it gives me this error:

error: failed to select a version for `digest`.
    ... required by package `kupyna v0.1.0 (/home/vrin/kupyna_hashes_contrib/kupyna)`
versions that meet the requirements `=0.11.0-pre.9` are: 0.11.0-pre.9

all possible versions conflict with previously selected packages.

  previously selected package `digest v0.11.0-pre.8`
    ... which satisfies dependency `digest = "=0.11.0-pre.8"` of package `ascon-hash v0.3.0-pre (/home/vrin/kupyna_hashes_contrib/ascon-hash)`

failed to select a version for `digest` which could resolve this conflict

now this worked fine before updating the toml. Post toml update the CI is fine, but the main function is failing. Is this expected behaviour?

newpavlov · 2024-10-11T20:01:03Z

Try to rebase your branch to master.

AnarchistHoneybun · 2024-10-12T04:21:33Z

I don't have a lot (read any) experience with rebasing, sadly. Could you explain a bit more?

newpavlov · 2024-10-12T04:42:14Z

You should be able to rebase to this repos master using something like this (it accounts for the fact that you've forked from a fork):

git remote add upstream [email protected]:RustCrypto/hashes.git
git pull upstream
git checkout main_repo_contrib
git rebase upstream/master
git push --force

Note that I do not guarantee correctness of these commands since I wrote them from memory.

kupyna/src/sub_units/t_xor_plus/tests/test_individual_layers.rs

…sh code size

… dependencies

… the row and col constants

jkoudys · 2025-03-15T14:56:48Z

@newpavlov I'm pushing these in on https://github.com/AnarchistHoneybun/kupyna_hashes_contrib/pull/7

Do you have any examples of non-byte-aligned bit streams for hashing? You can take a 1-bit, 510-bit, etc. message hash in kupyna, and the C reference app does it. We have some tests that I'm trying to convert from the C version to do one of the bitwise hashes but I can't figure out how the shorter length gets passed down into the hasher.

Is it totally fine to just make this version support only byte-bound messages? I'm sure it's useful even without being bitwise. I assume that could go in on a future PR.

newpavlov · 2025-03-15T18:13:13Z

IIRC most hash functions theoretically support processing messages with non-byte lengths. But it's a problematic capability to properly support on modern general purpose computers and it has an extremely low practical value, so cryptographic libraries generally just ignore it.

Is it totally fine to just make this version support only byte-bound messages?

Yes, see the traits issue I linked above.

AnarchistHoneybun · 2025-03-15T18:27:47Z

Hey @newpavlov could you also do a review and lmk if there's anything left out? I wanted to finalize this and then discuss the possibility of working on something larger like this (I remember yescrypt coming up in discussion) for GSOC if possible. Thank you!

jkoudys · 2025-03-16T03:31:10Z

@AnarchistHoneybun I have the style updates, and the new unit tests, waiting on https://github.com/AnarchistHoneybun/kupyna_hashes_contrib/pull/7 . That PR's going into the same branch AnarchistHoneybun:main_repo_contrib for this PR to RustCrypto, so you'll just need to accept that PR.

kupyna/tests/kup256.rs

newpavlov · 2025-03-17T12:44:42Z

kupyna/src/lib.rs

+pub type KupynaShort<OutSize> = CoreWrapper<KupynaShortCore<OutSize>>;
+
+/// Kupyna-48 hasher state
+pub type Kupyna48 = CoreWrapper<KupynaShortCore<U6>>;


I don't think Kupyna48 is intended to be used in practice, so we probably do not need an alias for it. It's probably provided only for testing purposes, so you can just define it in tests/kup48.rs.

kupyna/tests/kup48.rs

kupyna/src/lib.rs

newpavlov · 2025-03-17T12:57:24Z

kupyna/src/lib.rs

+
+    fn finalize_variable_core(&mut self, buffer: &mut Buffer<Self>, out: &mut Output<Self>) {
+        let total_message_len_bits =
+            (((self.blocks_len * 64) + (buffer.size() - buffer.remaining()) as u64) * 8) as u128;


To prevent theoretical overflows it's better to write it like this:

let msg_len_bytes = (self.blocks_len as u128) * Self::BlockSize::U128 + (buffer.get_pos() as u128); let msg_len_bits = 8 * msg_len_bytes;

could you explain the issue with the current one a bit

self.blocks_len * 64 * 8 may overflow u64, while (self.blocks_len as u128) * 64 * 8 is guaranteed to not overflow. It's mostly a theoretical issue (since you would need to process more than 2^64 bits to trigger it), but in addition to that, I think the suggested variant is also a bit easier to read.

newpavlov · 2025-03-17T13:18:45Z

kupyna/src/compress512.rs

+
+fn rotate_rows(mut state: Matrix) -> Matrix {
+    const ROWS: usize = 8;
+    let cols = 8;


Use the COLS constant here.

newpavlov · 2025-03-17T13:20:03Z

kupyna/src/compress512.rs

+}
+
+fn rotate_rows(mut state: Matrix) -> Matrix {
+    const ROWS: usize = 8;


Why not define this constant at the top of the module?

newpavlov · 2025-03-17T13:21:19Z

kupyna/src/compress512.rs

+
+fn matrix_to_block(matrix: Matrix) -> [u8; 64] {
+    const ROWS: usize = 8;
+    const COLS: usize = 8;


Why do you redefine COLS here?

newpavlov · 2025-03-17T13:24:46Z

kupyna/src/compress1024.rs

+    let mut result = [0u8; 128];
+    for i in 0..128 {
+        result[i] = a[i] ^ b[i];
+    }


You can write it as core::array::from_fn(|i| a[i] ^ b[i]). You also could define a generic function:

pub fn xor<const N: usize>(a: [u8; N], b: [u8; N]) -> [u8; N] { core::array::from_fn(|i| a[i] ^ b[i]) }

newpavlov · 2025-03-17T13:29:53Z

kupyna/src/compress512.rs

+    let mut prev_vector_u8 = [0u8; 64];
+    for (i, &value) in prev_vector.iter().enumerate() {
+        let bytes = value.to_be_bytes();
+        prev_vector_u8[i * 8..(i + 1) * 8].copy_from_slice(&bytes);


Usually it's better to transform message block into [u64; 8], than transforming the state into bytes.

Try to look into re-implementing t_xor_l and t_plus_l in terms of [u64; 8]? Do not forget to measure performance while doing it.

newpavlov · 2025-03-23T00:33:52Z

@AnarchistHoneybun
If you don't plan to work on the comments above in the near future, I guess we could merge this PR in its current form and leave the improvements for later PRs.

AnarchistHoneybun · 2025-03-23T02:20:33Z

@AnarchistHoneybun
If you don't plan to work on the comments above in the near future, I guess we could merge this PR in its current form and leave the improvements for later PRs.

I'm sorry I've just been stuck with some college assignments, planned to finish these improvements latest by this weekend. Merging it now seems fine as we could atleast have the function in the repo and then work on the mentioned details, but I'm fine with waiting too.

jkoudys · 2025-03-23T03:06:10Z

How about merge at this level, then I can branch directly and finish the remaining?

newpavlov reviewed Oct 11, 2024

View reviewed changes

kupyna/Cargo.toml Outdated Show resolved Hide resolved

newpavlov reviewed Oct 12, 2024

View reviewed changes

kupyna/src/sub_units/t_xor_plus/tests/test_individual_layers.rs Outdated Show resolved Hide resolved

jkoudys and others added 22 commits October 14, 2024 00:02

Add kupyna as a new member

0db03dd

add a basic first-past cargo based on ascon's

2517c46

Add readme

d5532d8

update usage in readme

7e5766e

First pass, pushing up of @AnarchistHoneybun's matrix XORing code

2d01f22

fix: update correct crate name

bf41d89

Add kupyna to the list of algos

190b351

update algo link

9d2eada

update author creds for kupyna readme

2d0218e

updating toml

60ad1de

resolving clippy warnings/errors

d212a3f

rustfmt pass

36d271d

new gitignore

350e568

library updated to use a struct

1310735

renamed base level function kept for b/w compatibility

7b508cd

removed state size param from constructor to make dependent on the ha…

d9dc773

…sh code size

rustfmt pass

1bf9cad

added state mtrx col+row fields & num round field on hash struct

11c9b53

refactored to pass borrow of hasher struct to downstream functions

6346cff

passing borrow of struct further downstream to eliminate all constant…

96b4322

… dependencies

module refactored to pass borrow of struct around, eliminates need of…

6db4b7e

… the row and col constants

tests are broken now

97c31bf

AnarchistHoneybun added 6 commits March 13, 2025 20:14

merge upstream

4d03e51

rm error messages

7470d47

gitignore update

67899fe

merge resolve

02dad37

fmt pass

94114ab

fmt pass

Loading
Loading status checks…

7020609

AnarchistHoneybun requested a review from newpavlov March 13, 2025 15:42

newpavlov reviewed Mar 17, 2025

View reviewed changes

AnarchistHoneybun added 2 commits March 18, 2025 22:16

tests update

98bb111

update loops to use zip+chunk format

Loading
Loading status checks…

c903550

newpavlov added 4 commits March 24, 2025 03:27

remove redundant empty line

cc12acc

Define Kupyna48 in kup48.rs

123ab45

Remove Kupyna48 from lib.rs

4b948d0

Fix import warning

c6cc14c

newpavlov mentioned this pull request Mar 24, 2025

Missing hash functions #1

Open

18 tasks

newpavlov added 6 commits March 24, 2025 03:39

Tweak message length calculation

1653375

fmt

4faa551

fix

fdd00f0

fix

162a565

Update MSRV badge

9aa0bf1

Update MSRV badge

1c8ee92

newpavlov merged commit 97292d6 into RustCrypto:master Mar 24, 2025
228 checks passed

newpavlov mentioned this pull request Mar 24, 2025

kupyna: improve perfromance #675

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Kupyna: implemented hashing function #621

Kupyna: implemented hashing function #621

AnarchistHoneybun commented Oct 11, 2024

AnarchistHoneybun commented Oct 11, 2024

AnarchistHoneybun commented Oct 11, 2024

newpavlov commented Oct 11, 2024

AnarchistHoneybun commented Oct 12, 2024

newpavlov commented Oct 12, 2024

jkoudys commented Mar 15, 2025

newpavlov commented Mar 15, 2025

AnarchistHoneybun commented Mar 15, 2025

jkoudys commented Mar 16, 2025

newpavlov Mar 17, 2025 •

edited

Loading

newpavlov Mar 17, 2025

AnarchistHoneybun Mar 18, 2025

newpavlov Mar 18, 2025 •

edited

Loading

newpavlov Mar 17, 2025

newpavlov Mar 17, 2025

newpavlov Mar 17, 2025

newpavlov Mar 17, 2025

newpavlov Mar 17, 2025

newpavlov commented Mar 23, 2025

AnarchistHoneybun commented Mar 23, 2025

jkoudys commented Mar 23, 2025

Kupyna: implemented hashing function #621

Kupyna: implemented hashing function #621

Conversation

AnarchistHoneybun commented Oct 11, 2024

AnarchistHoneybun commented Oct 11, 2024

AnarchistHoneybun commented Oct 11, 2024

newpavlov commented Oct 11, 2024

AnarchistHoneybun commented Oct 12, 2024

newpavlov commented Oct 12, 2024

jkoudys commented Mar 15, 2025

newpavlov commented Mar 15, 2025

AnarchistHoneybun commented Mar 15, 2025

jkoudys commented Mar 16, 2025

newpavlov Mar 17, 2025 • edited Loading

Choose a reason for hiding this comment

newpavlov Mar 17, 2025

Choose a reason for hiding this comment

AnarchistHoneybun Mar 18, 2025

Choose a reason for hiding this comment

newpavlov Mar 18, 2025 • edited Loading

Choose a reason for hiding this comment

newpavlov Mar 17, 2025

Choose a reason for hiding this comment

newpavlov Mar 17, 2025

Choose a reason for hiding this comment

newpavlov Mar 17, 2025

Choose a reason for hiding this comment

newpavlov Mar 17, 2025

Choose a reason for hiding this comment

newpavlov Mar 17, 2025

Choose a reason for hiding this comment

newpavlov commented Mar 23, 2025

AnarchistHoneybun commented Mar 23, 2025

jkoudys commented Mar 23, 2025

newpavlov Mar 17, 2025 •

edited

Loading

newpavlov Mar 18, 2025 •

edited

Loading