Implemented DES #2

gsingh93 · 2017-01-25T22:05:12Z

This is an implementation of DES. The included benchmark gives 2 MB/s performance, I'm not sure if that's good or bad. Most of the trickiness with the implementation is the fact that the most significant bit is considered bit zero instead of the least significant bit. It's possible I did some unnecessary shifting/arithmetic to account for this.

Note that the crate name des is taken on crates.io, so we might have to change that name.

I'm planning on adding more test cases. I have a file of (input, key, output) tuples that I need to parse and turn into the correct format for tests used in this repo.

gsingh93 · 2017-01-26T00:10:37Z

Hmm, now that I think about it, it would probably be cleaner to just use big endian integers and work from the LSB instead of using little endian and working from the MSB. I'll experiment with that.

newpavlov · 2017-01-26T09:01:19Z

Thank you!
I will review your PR on this weekend. Regarding performance, this paper gives DES 21 MB/s, so there is room for improvement. You can use perf and flamegraphs for finding hot spots, additionally try to play with #[inline(always)] and loop unrolling, they can significantly improve performance.

Also you can implement 3DES, not sure if it's better to use the separate crate for it or include in this one.

newpavlov · 2017-01-26T09:13:19Z

Also check this implementation (it uses rayon for parallelism, so it dependent on std), I will try to contact owner of this crate later to ask if he would like to cooperate with us.

newpavlov · 2017-01-29T07:44:26Z

I've updated your code with optimizations I borrowed from antoyo's des-rs, it makes code less readable, but significantly faster. It could be nice to add tests which will check optimized transformations against straightforward ones written by you. I did similar test in magma. Also I've moved subkeys generation into new(), so if state will be reused subkeys will not be generated again. I'll add benchmarks of cipher initialization later.

I haven't modified apply_sboxes, probably it can be optimized too.

Currently I am getting 17 MB/s without initialization. Meanwhile OpenSSL implementation gives approximately 50 MB/s, so there is still room for improvement.

EDIT: Forgot about functions inlining, using it gives 19 MB/s respectively.

newpavlov · 2017-01-29T07:54:11Z

des/Cargo.toml

+
+[dependencies]
+generic-array = "0.5"
+block-cipher-trait = "0.2"


You use old version of crates:

generic-array = "0.6" block-cipher-trait = "0.3"

(Forgot to update blowfish)

newpavlov · 2017-01-29T07:55:30Z

des/src/lib.rs

+}
+
+impl BlockCipherFixKey for Des {
+    type KeySize = U8;


Shouldn't it be U7?

newpavlov · 2017-01-29T07:56:34Z

des/src/lib.rs

+    fn encrypt_block(&self, input: &Block<U8>, output: &mut Block<U8>) {
+        let block = read_u64_be(input);
+        let res = self.encrypt(block);
+        write_u64_be(output, res);


I've added read_u64_be to byte-tools. Don't forget to run cargo update to get new version.

antoyo · 2017-02-18T03:42:19Z

I've been invited to participate to this discussion, so here I am :p .
I won't have much time to work on this, unfortunately, but I can give you access to the des crate on crates.io.
I don't mind adding the Apache license.

How can I give you a write access to the des crate?
Is it throught the crates.io website?
I cannot access it anymore: the login doesn't work for me in Firefox and the other browsers that I use.
I already had this issue in the past, but I don't remember what was the fix.
Is it related to third-party cookies (I tried accepting them)?

newpavlov · 2017-02-18T16:57:30Z

@antoyo
Thank you for your participation!

How can I give you a write access to the des crate?

You can do it using cargo with this command: cargo owner --add newpavlov. More information about it available here.

I cannot access it anymore: the login doesn't work for me in Firefox and the other browsers that I use.

You don't need to be authenticated on crates.io if your cargo token is intact. Your problems with cargo.io could be because of privacy settings, as it uses github for authentication. As a backup plan you can contact Alex Crichton.

antoyo · 2017-02-18T17:39:16Z

I added you as an owner.
Thanks for your interest in my work.
I can answer any question about the implementation.

newpavlov · 2017-02-21T09:44:39Z

@antoyo
Thank you! I will publish new version after merge.

Meanwhile can you compare your S-box implementation with gsingh93's implementation? As I understand in the later transformed S-boxes are used to improve performance. Simple tests show that this implementation is a bit faster than yours, but maybe you know about additional optimizations which can be applied here?

newpavlov · 2017-11-17T09:25:54Z

Sorry it took so long!

I will merge this PR, although before publishing I'll need to make some changes.

newpavlov · 2017-11-17T10:58:30Z

BTW it looks like you had a bug in decryption part.

newpavlov · 2017-11-17T12:59:58Z

@gsingh93
If you'll have time please take a look at #12.

gsingh93 · 2017-11-17T19:23:51Z

Ack. Fairly busy for a while, so if you or anyone can fix that would be great. Otherwise, I'll take a look when I have time.

gsingh93 added 12 commits January 23, 2017 00:24

des: add stub struct and impls

eda805d

des: added constants and implemented sbox function

4d54315

des: implemented apply_pboxa

71b53f1

des: implemented round function and encrypt_block

7f02620

des: implemented feistl function

2e482cd

des: implemented BlockCipherFixKey::new()

480205f

des: fixed FINAL_PBOX

8fbe114

des: implemented key schedule and decryption function

1e75cd4

des: fixed remaining bugs

8dae2c2

des: added test

7083dd9

des: added benchmark

03c945e

des: added license files

86dd5d3

newpavlov added 3 commits January 29, 2017 11:21

Optimization changes in lib.rs

cb9fb8f

Removed constants which no more necessary

3c415cb

Updated copyrights

f40647b

newpavlov reviewed Jan 29, 2017

View reviewed changes

newpavlov mentioned this pull request Apr 24, 2017

Missing block ciphers #1

Open

19 tasks

newpavlov merged commit bafc264 into RustCrypto:master Nov 17, 2017

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Implemented DES #2

Implemented DES #2

gsingh93 commented Jan 25, 2017

gsingh93 commented Jan 26, 2017

newpavlov commented Jan 26, 2017

newpavlov commented Jan 26, 2017 •

edited

Loading

newpavlov commented Jan 29, 2017 •

edited

Loading

newpavlov Jan 29, 2017 •

edited

Loading

newpavlov Jan 29, 2017

newpavlov Jan 29, 2017

antoyo commented Feb 18, 2017 •

edited

Loading

newpavlov commented Feb 18, 2017

antoyo commented Feb 18, 2017

newpavlov commented Feb 21, 2017 •

edited

Loading

newpavlov commented Nov 17, 2017

newpavlov commented Nov 17, 2017

newpavlov commented Nov 17, 2017

gsingh93 commented Nov 17, 2017

Implemented DES #2

Implemented DES #2

Conversation

gsingh93 commented Jan 25, 2017

gsingh93 commented Jan 26, 2017

newpavlov commented Jan 26, 2017

newpavlov commented Jan 26, 2017 • edited Loading

newpavlov commented Jan 29, 2017 • edited Loading

newpavlov Jan 29, 2017 • edited Loading

Choose a reason for hiding this comment

newpavlov Jan 29, 2017

Choose a reason for hiding this comment

newpavlov Jan 29, 2017

Choose a reason for hiding this comment

antoyo commented Feb 18, 2017 • edited Loading

newpavlov commented Feb 18, 2017

antoyo commented Feb 18, 2017

newpavlov commented Feb 21, 2017 • edited Loading

newpavlov commented Nov 17, 2017

newpavlov commented Nov 17, 2017

newpavlov commented Nov 17, 2017

gsingh93 commented Nov 17, 2017

newpavlov commented Jan 26, 2017 •

edited

Loading

newpavlov commented Jan 29, 2017 •

edited

Loading

newpavlov Jan 29, 2017 •

edited

Loading

antoyo commented Feb 18, 2017 •

edited

Loading

newpavlov commented Feb 21, 2017 •

edited

Loading