riscure / jlsca Goto Github PK

After updating Julia to v.1.0.0 and using the newest Jlsca this happens:

_.julia\packages\Jlsca\PpvWi\examples> julia main-condavg.jl examples/aes128_mc_ciph_c8bb4cbd817f9bc6ff762dc4298a29e4.trs ERROR: LoadError: LoadError: UndefVarError: @everywhere not defined_
Any ideas?

Sidenote: Tests run inside julia repl ("test Jlsca") work/pass, but the examples no longer work.

Hi,

I was working on a CPA problem that I was bashing my head in using jlsca and which was trivially solved by lascar and I wanted to understand the difference / my mistake. Here is the data to reproduce:

cpa_data.zip

Here is my code for jlsca:

using Jlsca.Trs
using Jlsca.Sca
using Jlsca.Align
using Plots
using Statistics

fname = "/home/kali/machines/self/hw/cpa/nMY.ts"
keyByteNum = 1
condavg = false;

trs = InspectorTrace(fname)

attack = AesSboxAttack()
attack.xor = true
anal = IncrementalCPA() # Also tested  CPA()

anal.leakages = [HW()] # Also tested with ID(), Bit(0) and all bits

params = DpaAttack(attack,anal)
params.maxCols = 1042
params.maxColsPost = 1042
params.dataOffset = 1
params.targetOffsets = collect(1:16)
params.updateInterval = 10
params.knownKey = missing
params.keepraw = true

getTrs() = trs

@time rankdata = sca(DistributedTrace(getTrs), params, 1, nrTraces);

And here is my code for lascar, which runs perfectly:

import numpy as np 
from lascar import Session, CpaEngine, ConsoleOutputMethod, MatPlotLibOutputMethod, RankProgressionOutputMethod, ScoreProgressionOutputMethod, DictOutputMethod
from lascar.container import TraceBatchContainer
from lascar.tools.aes import sbox

traces = np.load("traces.npy")
values = np.load("pts.npy")


engines = [
        CpaEngine(f"cpa{i}",  lambda v, k, z=i: sbox[v[z] ^ k], range(256))
        for i in range(16)
]


t = TraceBatchContainer(traces, values)
s = Session(t)
s.add_engines(engines)
s.output_method = ConsoleOutputMethod(*engines)

s.run()

If I understand correctly, lascar is using an identity leakage model, but I'm not 100% sure if it doesn't take the hamming weight somewhere still. For the rest, the lascar code - to me - looks identical to what I try to do in jlsca, but I'm probably missing an important step.

ParallelIncrementalCPATestWithInterval() fails with EOFError.

Dear, Riscure.
Are planning add support PQC guessing and DL in Jlsca?

For example add support DL via Flux.jl or TensorFlow.jl Julia packages.

I used the Jlsca library to perform power analysis attacks for my master's thesis experiment and I would like to know how the rank data is generated. I spent quite a bit of time looking through the code, but I'm not very familiar with the Julia programming language.

Here is the attack that I am running:

trs = InspectorTrace(insfname)

params = DpaAttack(AesSboxAttack(), IncrementalCPA())
params.dataOffset = 1
params.attack.direction = FORWARD
params.analysis.leakages = [HW()]
numberOfTraces = length(trs);

rankData = sca(trs, params, 1, numberOfTraces)
key = getKey(params,rankData)

Dear Riscure
Are you can implement guesser & selected function for Jlsca?
For example simplifided version of ECC module in Inspector SCA software.