From 152001ef817c3dc4576ed950308f101ae0811115 Mon Sep 17 00:00:00 2001
From: blackshirt <blackshirt@users.noreply.github.com>
Date: Thu, 11 Sep 2025 05:30:06 +0000
Subject: [PATCH] x.crypto.ascon: improves one-shot of ascon hasing functions

---
 vlib/x/crypto/ascon/README.md       |   2 +-
 vlib/x/crypto/ascon/bench/aead.v    |  75 +++++++++++
 vlib/x/crypto/ascon/bench/hashxof.v | 197 ++++++++++++++++++++++++++++
 vlib/x/crypto/ascon/bench/sum.v     |  36 +++++
 vlib/x/crypto/ascon/digest.v        |  36 +++++
 vlib/x/crypto/ascon/hash.v          |  14 +-
 vlib/x/crypto/ascon/util.v          |   1 -
 vlib/x/crypto/ascon/xof.v           |  23 ++--
 8 files changed, 361 insertions(+), 23 deletions(-)
 create mode 100644 vlib/x/crypto/ascon/bench/aead.v
 create mode 100644 vlib/x/crypto/ascon/bench/hashxof.v
 create mode 100644 vlib/x/crypto/ascon/bench/sum.v

diff --git a/vlib/x/crypto/ascon/README.md b/vlib/x/crypto/ascon/README.md
index 96faf63c9c..acd03e933d 100644
--- a/vlib/x/crypto/ascon/README.md
+++ b/vlib/x/crypto/ascon/README.md
@@ -7,7 +7,7 @@ thats provides Authenticated Encryption, Hash, and Extendable Output Functions.
 See the [NIST.SP.800-232 Standard](https://doi.org/10.6028/NIST.SP.800-232) for more detail.
 
 This module mostly implements all the features availables on the document.
-Its currently implements:
+Its currently implements: 
 - `Ascon-Hash256`, Ascon-based hashing implementation that produces 256-bits output.
 - `Ascon-XOF128`, Ascon-based eXtendable Output Function (XOF) where the output size of 
 the hash of the message can be selected by the user.
diff --git a/vlib/x/crypto/ascon/bench/aead.v b/vlib/x/crypto/ascon/bench/aead.v
new file mode 100644
index 0000000000..c7d1cd1043
--- /dev/null
+++ b/vlib/x/crypto/ascon/bench/aead.v
@@ -0,0 +1,75 @@
+// This benchmark is for Ascon-AEAD128 in `x.crypto.ascon` compared to
+// already stocked `x.crypto.cacha20poly1305 for AEAD functionalities.
+//
+// Here is output in my tests, first item was `x.crypto.ascon` and the later
+// for `x.crypto.chacha20poly1305` on encryption or decryption part.
+//
+// Encryption..
+// -----------
+// Iterations:      10000          Total Duration:   26.008ms      ns/op:       2600       B/op:     16    allocs/op:     17
+// Iterations:      10000          Total Duration:  158.865ms      ns/op:      15886       B/op:     16    allocs/op:     16
+//
+// Decryption..
+// -----------
+// Iterations:      10000          Total Duration:   29.091ms      ns/op:       2909       B/op:      6    allocs/op:      8
+// Iterations:      10000          Total Duration:  158.373ms      ns/op:      15837       B/op:      8    allocs/op:     12
+//
+import encoding.hex
+import x.benchmark
+import x.crypto.ascon
+import x.crypto.chacha20poly1305
+
+// randomly generated key and nonce, 16-bytes of ascon key and 32-bytes of chacha20poly1305 key.
+const key_ascon = hex.decode('7857bfb462c654d1d1b02971be021235')!
+const key_cpoly = hex.decode('9d9603f4fc460e273b80795ea50eab5873c04f589226c7d591b5336feb32fcba')!
+
+// 16-bytes ascon-nonce
+const ascon_nonce = hex.decode('8b521028fb54591472d8d8ee14430835')!
+
+// 12-bytes chacha20poly1305 nonce
+const cpoly_nonce = hex.decode('9a3c83e4236ea9a2c4e482da')!
+
+const ad = 'Ascon-AEAD128 additional data'.bytes()
+const msg = 'Ascon-AEAD128 benchmarking message'.bytes()
+
+// expected ciphertext for aead128 := 4b21a18cbca65b11aaf73dc74241c89bfcec96a4c8973ae696a938e0a591e846c4eb7b2906664f2318c0fd6ec1c56424aa9b
+const ciphertext_aead128 = hex.decode('4b21a18cbca65b11aaf73dc74241c89bfcec96a4c8973ae696a938e0a591e846c4eb7b2906664f2318c0fd6ec1c56424aa9b')!
+
+fn bench_ascon_aead128_encrypt() ! {
+	_ := ascon.encrypt(key_ascon, ascon_nonce, ad, msg)!
+}
+
+fn bench_ascon_aead128_decrypt() ! {
+	_ := ascon.decrypt(key_ascon, ascon_nonce, ad, ciphertext_aead128)!
+}
+
+// expected ciphertext for chacha20poly1305
+const ciphertext_chachapoly1305 = hex.decode('67dea3c65f0f326bcf587f024140a85d9535790d9b16129210a2289eda43bb9b62746450026fc1baf466bcb8a181843cd424')!
+
+fn bench_chacha20poly1305_encrypt() ! {
+	_ := chacha20poly1305.encrypt(msg, key_cpoly, cpoly_nonce, ad)!
+}
+
+fn bench_chacha20poly1305_decrypt() ! {
+	_ := chacha20poly1305.decrypt(ciphertext_chachapoly1305, key_cpoly, cpoly_nonce, ad)!
+}
+
+fn main() {
+	cf := benchmark.BenchmarkDefaults{
+		n: 10000
+	}
+	println('Encryption..')
+	println('-----------')
+	mut b0 := benchmark.setup(bench_ascon_aead128_encrypt, cf)!
+	b0.run()
+	mut b1 := benchmark.setup(bench_chacha20poly1305_encrypt, cf)!
+	b1.run()
+
+	println('')
+	println('Decryption..')
+	println('-----------')
+	mut b2 := benchmark.setup(bench_ascon_aead128_decrypt, cf)!
+	b2.run()
+	mut b3 := benchmark.setup(bench_chacha20poly1305_decrypt, cf)!
+	b3.run()
+}
diff --git a/vlib/x/crypto/ascon/bench/hashxof.v b/vlib/x/crypto/ascon/bench/hashxof.v
new file mode 100644
index 0000000000..6b7ec6ce54
--- /dev/null
+++ b/vlib/x/crypto/ascon/bench/hashxof.v
@@ -0,0 +1,197 @@
+// Ascon-Hash256 (and Ascon-XOF128) benchmark compared to builtin
+// crypto.sha256 (for sum256) and sha3.shake256 (for xof outputing 256-bits)
+//
+// This benchmark code was adapted from argon2 benchmark by @fleximus, the creator argon2 module.
+// Credit tributed to @fleximus
+// See https://gist.github.com/fleximus/db5b867a9a37da46340db61bdac6e696
+//
+// Output
+// ======
+// Sum and Xof 256-bits output performance comparison
+// ============================================================
+// Iterations per test: 10000
+// --------------------------------------------------------------------------------------------------
+//   Data Size |   Ascon256 |     Sha256 |    Ratio 256 || AsconXof128 |   Shake256 |  Ratio (Xof) |
+// --------------------------------------------------------------------------------------------------
+//         4 B |    24.00ms |    33.00ms |        0.73x ||     24.00ms |   208.00ms |        0.12x |
+//         6 B |    23.00ms |    53.00ms |        0.45x ||     25.00ms |   287.00ms |        0.08x |
+//         8 B |    35.00ms |    37.00ms |        0.95x ||     26.00ms |   202.00ms |        0.18x |
+//        16 B |    30.00ms |    37.00ms |        0.83x ||     30.00ms |   205.00ms |        0.15x |
+//        64 B |    55.00ms |    61.00ms |        0.89x ||     53.00ms |   241.00ms |        0.23x |
+//        75 B |    61.00ms |    57.00ms |        1.07x ||     58.00ms |   182.00ms |        0.34x |
+//       256 B |   154.00ms |   123.00ms |        1.25x ||    144.00ms |   398.00ms |        0.39x |
+//       512 B |   273.00ms |   216.00ms |        1.26x ||    265.00ms |   779.00ms |        0.35x |
+//      1025 B |   610.00ms |   401.00ms |        1.52x ||    509.00ms |      1.37s |        0.45x |
+// --------------------------------------------------------------------------------------------------
+//       Total |      1.27s |      1.02s |        1.24x ||      1.14s |      3.87s |         0.294x|
+// --------------------------------------------------------------------------------------------------
+//
+// Per-operation averages:
+//   Ascon256:         14108 ns per hash
+//   Sha256:           11360 ns per hash
+//   AsconXof128:      12648 ns per hash
+//   Shake256:         43036 ns per hash
+//
+module main
+
+import time
+import crypto.sha3
+import crypto.sha256
+import x.crypto.ascon
+
+const benchmark_iterations = 10000
+
+// We include more small size because, Ascon-Hash256 working with more smaller block size.
+const test_data_sizes = [
+	4, // below Ascon-Hash256 block size
+	6, // Still below Ascon-Hash256 block size
+	8, // align with Ascon-Hash256 block size
+	16, // Small data
+	64, // Medium data
+	75, // above 64-bytes block
+	256, // Large data
+	512,
+	1025,
+]
+
+fn generate_test_data(size int) []u8 {
+	mut data := []u8{len: size}
+	for i in 0 .. size {
+		data[i] = u8(i % 256)
+	}
+	return data
+}
+
+fn benchmark_ascon_sha256(data []u8, iterations int) time.Duration {
+	start := time.now()
+	for _ in 0 .. iterations {
+		_ := ascon.sum256(data)
+	}
+	return time.since(start)
+}
+
+fn benchmark_sha256_sum256(data []u8, iterations int) time.Duration {
+	start := time.now()
+	for _ in 0 .. iterations {
+		_ := sha256.sum256(data)
+	}
+	return time.since(start)
+}
+
+// for eXtendable output functions (XOF)
+fn benchmark_ascon_xof128_32(data []u8, iterations int) time.Duration {
+	start := time.now()
+	for _ in 0 .. iterations {
+		_ := ascon.xof128(data, 32) or { panic(err) }
+	}
+	return time.since(start)
+}
+
+fn benchmark_sha3_shake256(data []u8, iterations int) time.Duration {
+	start := time.now()
+	for _ in 0 .. iterations {
+		_ := sha3.shake256(data, 32)
+	}
+	return time.since(start)
+}
+
+fn format_duration(d time.Duration) string {
+	if d.microseconds() < 1000 {
+		return '${d.microseconds():6}μs'
+	} else if d.milliseconds() < 1000 {
+		return '${f64(d.milliseconds()):6.2f}ms'
+	} else {
+		return '${f64(d.seconds()):6.2f}s'
+	}
+}
+
+const data_title = 'Data Size'
+const ascon_sum256_title = 'Ascon256'
+const sha256_title = 'Sha256'
+const ascon_xof128_title = 'AsconXof128'
+const sha3_shake256_title = 'Shake256'
+const ratio_ascon256_w_sha256 = 'Ratio 256'
+const ratio_asconxof128_w_shake256 = 'Ratio (Xof)'
+
+fn main() {
+	println('')
+	println('Sum and Xof 256-bits output performance comparison')
+	println('============================================================')
+	println('Iterations per test: ${benchmark_iterations}')
+
+	println('${'-'.repeat(98)}')
+	println('${data_title:12} | ${ascon_sum256_title:10} | ${sha256_title:10} | ${ratio_ascon256_w_sha256:12} || ${ascon_xof128_title:10} | ${sha3_shake256_title:10} | ${ratio_asconxof128_w_shake256:12} |')
+	println('${'-'.repeat(98)}')
+
+	mut total_ascon256 := time.Duration(0)
+	mut total_sha256 := time.Duration(0)
+	mut total_shake256 := time.Duration(0)
+	mut total_asconxof128 := time.Duration(0)
+
+	for size in test_data_sizes {
+		test_data := generate_test_data(size)
+
+		// Warm up
+		_ := ascon.sum256(test_data)
+		_ := sha256.sum256(test_data)
+
+		_ := ascon.xof128(test_data, 32)!
+		_ := sha3.shake256(test_data, 32)
+
+		// Benchmark Ascon-HASH256
+		ascon256_time := benchmark_ascon_sha256(test_data, benchmark_iterations)
+
+		// Benchmark Sha256 implementation
+		sha256_time := benchmark_sha256_sum256(test_data, benchmark_iterations)
+
+		// Benchmark Sha3 shake256 implementation
+		shake256_time := benchmark_sha3_shake256(test_data, benchmark_iterations)
+
+		// Benchmark AsconXof128 256-bits output
+		asconxof128_time := benchmark_ascon_xof128_32(test_data, benchmark_iterations)
+
+		// Calculate ratio ascon256 / sha256
+		ratio_ascon256_sha256 := f64(ascon256_time.nanoseconds()) / f64(sha256_time.nanoseconds())
+
+		// Calculate ratio asconxof128 / shake256
+		ratio_asconxof128_shake256 := f64(asconxof128_time.nanoseconds()) / f64(shake256_time.nanoseconds())
+
+		ascon256_str := format_duration(ascon256_time)
+		sha256_str := format_duration(sha256_time)
+		asconxof128_str := format_duration(asconxof128_time)
+		shake256_str := format_duration(shake256_time)
+
+		ratio_ascon256_sha256_str := '${ratio_ascon256_sha256:6.2f}x'
+		ratio_asconxof128_shake256_str := '${ratio_asconxof128_shake256:6.2f}x'
+
+		println('${size:10} B | ${ascon256_str:10} | ${sha256_str:10} | ${ratio_ascon256_sha256_str:12} || ${asconxof128_str:11} | ${shake256_str:10} | ${ratio_asconxof128_shake256_str:12} |')
+
+		total_ascon256 += ascon256_time
+		total_sha256 += sha256_time
+
+		total_asconxof128 += asconxof128_time
+		total_shake256 += shake256_time
+	}
+
+	println('${'-'.repeat(98)}')
+
+	// Overall performance comparison
+	overall_ascon256_w_sha256_ratio := f64(total_ascon256.nanoseconds()) / f64(total_sha256.nanoseconds())
+	overall_asconxof128_w_shake256_ratio := f64(total_asconxof128.nanoseconds()) / f64(total_shake256.nanoseconds())
+	total_title := 'Total'
+	println('${total_title:12} | ${format_duration(total_ascon256):10} | ${format_duration(total_sha256):10} | ${overall_ascon256_w_sha256_ratio:11.2f}x || ${format_duration(total_asconxof128):11} | ${format_duration(total_shake256):10} | ${overall_asconxof128_w_shake256_ratio:12.2f}x|')
+	println('${'-'.repeat(98)}')
+
+	println('')
+	println('Per-operation averages:')
+	avg_ascon256 := total_ascon256.nanoseconds() / (benchmark_iterations * test_data_sizes.len)
+	avg_sha256 := total_sha256.nanoseconds() / (benchmark_iterations * test_data_sizes.len)
+	avg_shake256 := total_shake256.nanoseconds() / (benchmark_iterations * test_data_sizes.len)
+	avg_asconxof128 := total_asconxof128.nanoseconds() / (benchmark_iterations * test_data_sizes.len)
+
+	println('  Ascon256:\t ${avg_ascon256:8} ns per hash')
+	println('  Sha256:\t ${avg_sha256:8} ns per hash')
+	println('  AsconXof128:\t ${avg_asconxof128:8} ns per hash')
+	println('  Shake256:\t ${avg_shake256:8} ns per hash')
+	println('')
+}
diff --git a/vlib/x/crypto/ascon/bench/sum.v b/vlib/x/crypto/ascon/bench/sum.v
new file mode 100644
index 0000000000..3739d0efff
--- /dev/null
+++ b/vlib/x/crypto/ascon/bench/sum.v
@@ -0,0 +1,36 @@
+import time
+import x.crypto.ascon
+
+// Before:
+// Benchmarking ascon.sum256 ...
+// Average ascon.sum256 time: 8 µs
+// Benchmarking ascon.sum256 ...
+// Average ascon.sum256 time: 6 µs
+
+// For xof128 (32 bytes)
+// Benchmarking ascon.xof128 ...
+// Average ascon.xof128 time: 7 µs
+// Benchmarking ascon.xof128 ...
+// Average ascon.xof128 time: 6 µs
+
+// For cxof128 32 bytes
+// Benchmarking ascon.cxof128 ...
+// Average ascon.cxof128 time: 9 µs
+// Benchmarking ascon.sum256 ...
+// Average ascon.cxof128 time: 7 µs
+//
+fn main() {
+	iterations := 1000
+	msg := [u8(0xff)].repeat(100)
+
+	println('Benchmarking ascon.sum256 ...')
+	mut total_sum_time := i64(0)
+	for _ in 0 .. iterations {
+		sw := time.new_stopwatch()
+		_ := ascon.sum256(msg)
+		elapsed := sw.elapsed().microseconds()
+		total_sum_time += elapsed
+	}
+	avg_sum_time := total_sum_time / iterations
+	println('Average ascon.sum256 time: ${avg_sum_time} µs')
+}
diff --git a/vlib/x/crypto/ascon/digest.v b/vlib/x/crypto/ascon/digest.v
index 6617b6e255..89a7dac7e1 100644
--- a/vlib/x/crypto/ascon/digest.v
+++ b/vlib/x/crypto/ascon/digest.v
@@ -133,3 +133,39 @@ fn (mut d Digest) squeeze(mut dst []u8) int {
 
 	return pos
 }
+
+@[direct_array_access; inline]
+fn ascon_generic_hash(mut s State, msg_ []u8, size int) []u8 {
+	// Assumed state was correctly initialized
+	// Absorbing the message
+	mut msg := msg_.clone()
+	for msg.len >= block_size {
+		s.e0 ^= binary.little_endian_u64(msg[0..block_size])
+		unsafe {
+			msg = msg[block_size..]
+		}
+		ascon_pnr(mut s, ascon_prnd_12)
+	}
+	// Absorb the last partial message block
+	s.e0 ^= load_bytes(msg, msg.len)
+	s.e0 ^= pad(msg.len)
+
+	// Squeezing phase
+	//
+	// The squeezing phase begins after msg is absorbed with an
+	// permutation 𝐴𝑠𝑐𝑜𝑛-𝑝[12] to the state:
+	ascon_pnr(mut s, ascon_prnd_12)
+	mut out := []u8{len: size}
+	mut pos := 0
+	mut clen := out.len
+	for clen >= block_size {
+		binary.little_endian_put_u64(mut out[pos..pos + 8], s.e0)
+		ascon_pnr(mut s, ascon_prnd_12)
+		pos += block_size
+		clen -= block_size
+	}
+	// final output, the resulting 256-bit digest is the concatenation of hash blocks
+	store_bytes(mut out[pos..], s.e0, clen)
+
+	return out
+}
diff --git a/vlib/x/crypto/ascon/hash.v b/vlib/x/crypto/ascon/hash.v
index 2ec0dda97e..60c02af506 100644
--- a/vlib/x/crypto/ascon/hash.v
+++ b/vlib/x/crypto/ascon/hash.v
@@ -42,13 +42,13 @@ const hash256_initial_state = State{
 }
 
 // sum256 creates an Ascon-Hash256 checksum for bytes on msg and produces a 256-bit hash.
-pub fn sum256(msg []u8) []u8 {
-	mut h := new_hash256()
-	_ := h.write(msg) or { panic(err) }
-	h.Digest.finish()
-	mut dst := []u8{len: hash256_size}
-	_ := h.Digest.squeeze(mut dst)
-	return dst
+pub fn sum256(msg_ []u8) []u8 {
+	// This is single-shot function, so, no need to use Hash256 opaque that process
+	// message in streaming way. To reduce this overhead, use raw processing instead.
+	//
+	// Initialize state
+	mut s := hash256_initial_state
+	return ascon_generic_hash(mut s, msg_, hash256_size)
 }
 
 // Hash256 is an opaque provides an implementation of Ascon-Hash256 from NIST.SP.800-232 standard.
diff --git a/vlib/x/crypto/ascon/util.v b/vlib/x/crypto/ascon/util.v
index 0618303648..345683ea66 100644
--- a/vlib/x/crypto/ascon/util.v
+++ b/vlib/x/crypto/ascon/util.v
@@ -5,7 +5,6 @@
 // Utility helpers used across the module
 module ascon
 
-import math.bits
 import encoding.binary
 
 // clear_bytes clears the bytes of x in n byte
diff --git a/vlib/x/crypto/ascon/xof.v b/vlib/x/crypto/ascon/xof.v
index febbf98968..eea14e35f1 100644
--- a/vlib/x/crypto/ascon/xof.v
+++ b/vlib/x/crypto/ascon/xof.v
@@ -30,13 +30,11 @@ const xof128_initial_state = State{
 
 // xof128 creates an Ascon-XOF128 checksum of msg with specified desired size of output.
 pub fn xof128(msg []u8, size int) ![]u8 {
-	mut x := new_xof128(size)
-	_ := x.write(msg)!
-	x.Digest.finish()
-	mut out := []u8{len: size}
-	_ := x.Digest.squeeze(mut out)
-	x.reset()
-	return out
+	if size > max_hash_size {
+		return error('xof128: invalid size')
+	}
+	mut s := xof128_initial_state
+	return ascon_generic_hash(mut s, msg, size)
 }
 
 // xof128_64 creates a 64-bytes of Ascon-XOF128 checksum of msg.
@@ -170,13 +168,10 @@ const cxof128_initial_state = State{
 
 // cxof128 creates an Ascon-CXOF128 checksum of msg with supplied size and custom string cs.
 pub fn cxof128(msg []u8, size int, cs []u8) ![]u8 {
-	mut cx := new_cxof128(size, cs)!
-	_ := cx.write(msg)!
-	cx.Digest.finish()
-	mut out := []u8{len: size}
-	_ := cx.Digest.squeeze(mut out)
-	cx.reset()
-	return out
+	// Initialize CXof128 state with precomputed-value and absorb the customization string
+	mut s := cxof128_initial_state
+	cxof128_absorb_custom_string(mut s, cs)
+	return ascon_generic_hash(mut s, msg, size)
 }
 
 // cxof128_64 creates a 64-bytes of Ascon-CXOF128 checksum of msg with supplied custom string in cs.