Fix Armv8.1-M x4 xor_bytes alignment fallback by bremoran · Pull Request #1766 · pq-code-package/mlkem-native

bremoran · 2026-06-29T08:38:03Z

Corrects the xor-bytes unaligned access

hanno-becker · 2026-06-30T15:41:42Z

+        ((uintptr_t)data3 + byte_prefix)) &
+       3u) != 0u)
+  {
+    return MLK_NATIVE_FUNC_FALLBACK;


I don't think we can use fallback here, because the C version it not bit-interleaved?

ping @bremoran

Armv8.1-M: handle unaligned x4 xor_bytes inputs Avoid using vldrw.u32 for x4 Keccak xor_bytes inputs that are not 32-bit aligned. The aligned path still uses the existing gather-word loads, while unaligned input falls through to a predicated byte-load loop that also handles the final partial lane. Update the prologue byte loads to advance to the next lane start, then perform a single alignment check before choosing the gather-word path. Extend the x4 Keccak unit test to exercise all byte alignments for the input streams. Fixes #1759. Signed-off-by: Brendan Moran <brendan.moran@arm.com>

oqs-bot · 2026-07-09T13:48:13Z

CBMC Results (ML-KEM-512)

Full Results (191 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	1482s	1605s	-7.7%
`mlk_indcpa_enc`	✅	244s	259s	-6%
`mlk_indcpa_keypair_derand`	✅	200s	215s	-7%
`mlk_poly_rej_uniform`	✅	124s	143s	-13%
`mlk_rej_uniform_c`	✅	102s	123s	-17%
`rej_uniform_native_x86_64`	✅	55s	63s	-13%
`mlk_polyvec_basemul_acc_montgomery_cached_c`	✅	47s	53s	-11%
`poly_ntt_native`	✅	33s	40s	-18%
`rej_uniform_native_aarch64`	✅	33s	38s	-13%
`mlk_poly_reduce_native`	✅	31s	34s	-9%
`mlk_ntt_layer`	✅	27s	34s	-21%
`mlk_keccak_squeezeblocks_x4`	✅	24s	24s	+0%
`mlk_fqmul`	✅	17s	16s	+6%
`keccakf1600x4_permute_native_x4`	✅	16s	17s	-6%
`mlk_indcpa_dec`	✅	14s	16s	-12%
`mlk_poly_decompress_d10_native`	✅	14s	14s	+0%
`mlk_poly_decompress_d4_native`	✅	14s	15s	-7%
`mlk_poly_sub`	✅	10s	7s	+43%
`mlk_polyvec_add`	✅	10s	12s	-17%
`mlk_keccak_squeezeblocks`	✅	9s	8s	+12%
`mlk_poly_frommsg`	✅	9s	10s	-10%
`polyvec_basemul_acc_montgomery_cached_native`	✅	8s	8s	+0%
`mlk_keccak_squeeze_once`	✅	7s	7s	+0%
`mlk_poly_frombytes_native`	✅	7s	9s	-22%
`mlk_poly_rej_uniform_x4`	✅	7s	7s	+0%
`kem_check_pk`	✅	6s	4s	+50%
`mlk_keccak_absorb_once_x4`	✅	6s	6s	+0%
`mlk_ntt_butterfly_block`	✅	6s	8s	-25%
`kem_dec`	✅	5s	3s	+67%
`kem_keypair_derand`	✅	5s	2s	+150%
`mlk_ct_sel_int16`	✅	5s	3s	+67%
`mlk_keccakf1600_extract_bytes (big endian)`	✅	5s	1s	+400%
`mlk_poly_add`	✅	5s	5s	+0%
`mlk_poly_getnoise_eta1_4x`	✅	5s	3s	+67%
`mlk_poly_ntt`	✅	5s	8s	-38%
`mlk_poly_tobytes_c`	✅	5s	2s	+150%
`poly_compress_d4_native_x86_64`	✅	5s	2s	+150%
`poly_decompress_d10_native_x86_64`	✅	5s	7s	-29%
`mlk_enc_getnoise_eta1_eta2`	✅	4s	4s	+0%
`mlk_invntt_layer`	✅	4s	6s	-33%
`mlk_keccak_absorb_once`	✅	4s	5s	-20%
`mlk_keccakf1600_permute`	✅	4s	2s	+100%
`mlk_keccakf1600_permute_c`	✅	4s	6s	-33%
`mlk_keccakf1600x4_extract_bytes`	✅	4s	3s	+33%
`mlk_keccakf1600x4_xor_bytes_c`	✅	4s	2s	+100%
`mlk_matvec_mul`	✅	4s	2s	+100%
`mlk_poly_compress_dv`	✅	4s	4s	+0%
`mlk_poly_decompress_d4_c`	✅	4s	4s	+0%
`mlk_poly_invntt_tomont_c`	✅	4s	2s	+100%
`mlk_polyvec_basemul_acc_montgomery_cached`	✅	4s	2s	+100%
`mlk_polyvec_permute_bitrev_to_custom_native`	✅	4s	4s	+0%
`mlk_scalar_compress_d5`	✅	4s	2s	+100%
`mlk_scalar_decompress_d4`	✅	4s	2s	+100%
`mlk_shake256x4`	✅	4s	2s	+100%
`mlk_value_barrier_i32`	✅	4s	2s	+100%
`mlk_value_barrier_u8`	✅	4s	2s	+100%
`poly_decompress_d4_native_x86_64`	✅	4s	4s	+0%
`poly_frombytes_native_x86_64`	✅	4s	3s	+33%
`intt_native_aarch64`	✅	3s	2s	+50%
`keccak_f1600_x4_native_aarch64_v84a`	✅	3s	1s	+200%
`keccakf1600_permute_native`	✅	3s	3s	+0%
`keccakf1600x4_extract_bytes_native`	✅	3s	2s	+50%
`kem_check_sk`	✅	3s	2s	+50%
`mlk_check_pct`	✅	3s	2s	+50%
`mlk_ct_cmov_zero`	✅	3s	2s	+50%
`mlk_gen_matrix`	✅	3s	3s	+0%
`mlk_gen_matrix_serial`	✅	3s	2s	+50%
`mlk_keccakf1600x4_permute`	✅	3s	1s	+200%
`mlk_keccakf1600x4_xor_bytes`	✅	3s	3s	+0%
`mlk_keypair_getnoise_eta1`	✅	3s	2s	+50%
`mlk_poly_cbd_eta1`	✅	3s	2s	+50%
`mlk_poly_compress_d10_native`	✅	3s	2s	+50%
`mlk_poly_compress_d4`	✅	3s	3s	+0%
`mlk_poly_compress_d4_c`	✅	3s	4s	-25%
`mlk_poly_compress_d4_native`	✅	3s	2s	+50%
`mlk_poly_compress_du`	✅	3s	1s	+200%
`mlk_poly_decompress_d11`	✅	3s	1s	+200%
`mlk_poly_decompress_d11_c`	✅	3s	3s	+0%
`mlk_poly_decompress_d11_native`	✅	3s	2s	+50%
`mlk_poly_decompress_d5_native`	✅	3s	2s	+50%
`mlk_poly_frombytes_c`	✅	3s	5s	-40%
`mlk_poly_getnoise_eta1_4x_native`	✅	3s	4s	-25%
`mlk_poly_mulcache_compute_c`	✅	3s	2s	+50%
`mlk_poly_mulcache_compute_native`	✅	3s	3s	+0%
`mlk_poly_tomont`	✅	3s	1s	+200%
`mlk_polyvec_ntt`	✅	3s	2s	+50%
`mlk_polyvec_reduce`	✅	3s	4s	-25%
`mlk_scalar_decompress_d10`	✅	3s	2s	+50%
`mlk_value_barrier_u32`	✅	3s	2s	+50%
`poly_invntt_tomont_native`	✅	3s	2s	+50%
`poly_mulcache_compute_native_x86_64`	✅	3s	2s	+50%
`poly_reduce_native_aarch64`	✅	3s	4s	-25%
`poly_tobytes_native_aarch64`	✅	3s	3s	+0%
`poly_tomont_native_x86_64`	✅	3s	3s	+0%
`polyvec_basemul_acc_montgomery_cached_k2_native_aarch64`	✅	3s	4s	-25%
`polyvec_basemul_acc_montgomery_cached_k3_native_aarch64`	✅	3s	3s	+0%
`polyvec_basemul_acc_montgomery_cached_k3_native_x86_64`	✅	3s	1s	+200%
`keccak_f1600_x1_native_aarch64_v84a`	✅	2s	3s	-33%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	2s	2s	+0%
`keccakf1600x4_xor_bytes_native`	✅	2s	3s	-33%
`kem_enc`	✅	2s	4s	-50%
`kem_enc_derand`	✅	2s	4s	-50%
`kem_keypair`	✅	2s	4s	-50%
`mlk_barrett_reduce`	✅	2s	1s	+100%
`mlk_ct_cmask_neg_i16`	✅	2s	3s	-33%
`mlk_ct_cmask_nonzero_u16`	✅	2s	2s	+0%
`mlk_ct_get_optblocker_i32`	✅	2s	4s	-50%
`mlk_ct_get_optblocker_u8`	✅	2s	2s	+0%
`mlk_ct_memcmp`	✅	2s	2s	+0%
`mlk_ct_sel_uint8`	✅	2s	1s	+100%
`mlk_keccakf1600x4_extract_bytes_c`	✅	2s	3s	-33%
`mlk_poly_cbd_eta2`	✅	2s	3s	-33%
`mlk_poly_compress_d10_c`	✅	2s	3s	-33%
`mlk_poly_compress_d11`	✅	2s	2s	+0%
`mlk_poly_compress_d11_c`	✅	2s	1s	+100%
`mlk_poly_compress_d11_native`	✅	2s	3s	-33%
`mlk_poly_compress_d5_c`	✅	2s	3s	-33%
`mlk_poly_compress_d5_native`	✅	2s	2s	+0%
`mlk_poly_decompress_d10_c`	✅	2s	3s	-33%
`mlk_poly_decompress_du`	✅	2s	2s	+0%
`mlk_poly_decompress_dv`	✅	2s	2s	+0%
`mlk_poly_getnoise_eta1122_4x`	✅	2s	2s	+0%
`mlk_poly_invntt_tomont`	✅	2s	2s	+0%
`mlk_poly_ntt_c`	✅	2s	2s	+0%
`mlk_poly_reduce_c`	✅	2s	2s	+0%
`mlk_poly_tobytes_native`	✅	2s	3s	-33%
`mlk_poly_tomont_native`	✅	2s	4s	-50%
`mlk_poly_tomsg`	✅	2s	2s	+0%
`mlk_polyvec_compress_du`	✅	2s	3s	-33%
`mlk_polyvec_decompress_du`	✅	2s	2s	+0%
`mlk_polyvec_frombytes`	✅	2s	6s	-67%
`mlk_polyvec_mulcache_compute`	✅	2s	3s	-33%
`mlk_polyvec_permute_bitrev_to_custom`	✅	2s	1s	+100%
`mlk_polyvec_tomont`	✅	2s	2s	+0%
`mlk_rej_uniform`	✅	2s	2s	+0%
`mlk_scalar_compress_d1`	✅	2s	4s	-50%
`mlk_scalar_compress_d10`	✅	2s	2s	+0%
`mlk_scalar_compress_d11`	✅	2s	2s	+0%
`mlk_scalar_compress_d4`	✅	2s	2s	+0%
`mlk_scalar_signed_to_unsigned_q`	✅	2s	1s	+100%
`mlk_sha3_256`	✅	2s	3s	-33%
`mlk_shake128_squeezeblocks`	✅	2s	2s	+0%
`mlk_shake128x4_absorb_once`	✅	2s	3s	-33%
`mlk_shake128x4_squeezeblocks`	✅	2s	1s	+100%
`mlk_shake256`	✅	2s	1s	+100%
`ntt_native_x86_64`	✅	2s	5s	-60%
`poly_compress_d11_native_x86_64`	✅	2s	2s	+0%
`poly_compress_d5_native_x86_64`	✅	2s	2s	+0%
`poly_decompress_d11_native_x86_64`	✅	2s	2s	+0%
`poly_decompress_d5_native_x86_64`	✅	2s	1s	+100%
`poly_getnoise_eta1122_4x_native`	✅	2s	3s	-33%
`poly_mulcache_compute_native_aarch64`	✅	2s	3s	-33%
`poly_reduce_native_x86_64`	✅	2s	2s	+0%
`poly_tobytes_native_x86_64`	✅	2s	3s	-33%
`poly_tomont_native_aarch64`	✅	2s	2s	+0%
`polyvec_basemul_acc_montgomery_cached_k2_native_x86_64`	✅	2s	3s	-33%
`polyvec_basemul_acc_montgomery_cached_k4_native_x86_64`	✅	2s	1s	+100%
`intt_native_x86_64`	✅	1s	2s	-50%
`keccak_f1600_x1_native_aarch64`	✅	1s	2s	-50%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	1s	2s	-50%
`keccak_f1600_x4_native_avx2`	✅	1s	1s	+0%
`mlk_ct_cmask_nonzero_u8`	✅	1s	2s	-50%
`mlk_ct_get_optblocker_u32`	✅	1s	1s	+0%
`mlk_keccakf1600_extract_bytes`	✅	1s	2s	-50%
`mlk_keccakf1600_xor_bytes`	✅	1s	4s	-75%
`mlk_keccakf1600_xor_bytes (big endian)`	✅	1s	3s	-67%
`mlk_montgomery_reduce`	✅	1s	1s	+0%
`mlk_poly_compress_d10`	✅	1s	1s	+0%
`mlk_poly_compress_d5`	✅	1s	2s	-50%
`mlk_poly_decompress_d10`	✅	1s	2s	-50%
`mlk_poly_decompress_d4`	✅	1s	2s	-50%
`mlk_poly_decompress_d5`	✅	1s	2s	-50%
`mlk_poly_decompress_d5_c`	✅	1s	2s	-50%
`mlk_poly_frombytes`	✅	1s	3s	-67%
`mlk_poly_getnoise_eta2`	✅	1s	1s	+0%
`mlk_poly_mulcache_compute`	✅	1s	3s	-67%
`mlk_poly_reduce`	✅	1s	1s	+0%
`mlk_poly_tobytes`	✅	1s	1s	+0%
`mlk_poly_tomont_c`	✅	1s	1s	+0%
`mlk_polymat_permute_bitrev_to_custom`	✅	1s	4s	-75%
`mlk_polyvec_invntt_tomont`	✅	1s	2s	-50%
`mlk_polyvec_tobytes`	✅	1s	2s	-50%
`mlk_scalar_decompress_d11`	✅	1s	2s	-50%
`mlk_scalar_decompress_d5`	✅	1s	3s	-67%
`mlk_sha3_512`	✅	1s	5s	-80%
`mlk_shake128_absorb_once`	✅	1s	1s	+0%
`ntt_native_aarch64`	✅	1s	3s	-67%
`nttunpack_native_x86_64`	✅	1s	1s	+0%
`poly_compress_d10_native_x86_64`	✅	1s	4s	-75%
`polyvec_basemul_acc_montgomery_cached_k4_native_aarch64`	✅	1s	1s	+0%
`rej_uniform_native`	✅	1s	2s	-50%
`sys_check_capability`	✅	1s	3s	-67%

oqs-bot · 2026-07-09T13:50:19Z

CBMC Results (ML-KEM-1024)

Full Results (191 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	1682s	1648s	+2.1%
`mlk_indcpa_enc`	✅	373s	350s	+7%
`mlk_indcpa_keypair_derand`	✅	164s	166s	-1%
`mlk_poly_rej_uniform`	✅	137s	129s	+6%
`mlk_rej_uniform_c`	✅	112s	123s	-9%
`mlk_polyvec_basemul_acc_montgomery_cached_c`	✅	68s	73s	-7%
`rej_uniform_native_x86_64`	✅	59s	56s	+5%
`poly_ntt_native`	✅	36s	36s	+0%
`mlk_poly_reduce_native`	✅	35s	33s	+6%
`rej_uniform_native_aarch64`	✅	35s	34s	+3%
`mlk_ntt_layer`	✅	29s	28s	+4%
`mlk_keccak_squeezeblocks_x4`	✅	25s	25s	+0%
`polyvec_basemul_acc_montgomery_cached_native`	✅	24s	26s	-8%
`keccakf1600x4_permute_native_x4`	✅	18s	18s	+0%
`mlk_fqmul`	✅	18s	17s	+6%
`mlk_polyvec_add`	✅	17s	17s	+0%
`mlk_poly_decompress_d11_native`	✅	15s	14s	+7%
`mlk_poly_decompress_d5_native`	✅	13s	14s	-7%
`mlk_indcpa_dec`	✅	10s	9s	+11%
`mlk_ntt_butterfly_block`	✅	9s	8s	+12%
`mlk_poly_frommsg`	✅	9s	10s	-10%
`mlk_polyvec_ntt`	✅	9s	10s	-10%
`mlk_keccak_squeeze_once`	✅	8s	8s	+0%
`mlk_keccak_squeezeblocks`	✅	8s	8s	+0%
`mlk_poly_frombytes_native`	✅	8s	8s	+0%
`mlk_poly_sub`	✅	8s	8s	+0%
`mlk_invntt_layer`	✅	7s	3s	+133%
`mlk_poly_rej_uniform_x4`	✅	7s	6s	+17%
`poly_decompress_d11_native_x86_64`	✅	7s	5s	+40%
`kem_dec`	✅	6s	3s	+100%
`mlk_gen_matrix_serial`	✅	6s	3s	+100%
`mlk_keccak_absorb_once_x4`	✅	6s	7s	-14%
`mlk_poly_add`	✅	6s	5s	+20%
`intt_native_x86_64`	✅	5s	2s	+150%
`mlk_ct_sel_int16`	✅	5s	3s	+67%
`mlk_gen_matrix`	✅	5s	4s	+25%
`mlk_keccakf1600_permute_c`	✅	5s	6s	-17%
`mlk_poly_ntt`	✅	5s	5s	+0%
`mlk_rej_uniform`	✅	5s	1s	+400%
`poly_compress_d11_native_x86_64`	✅	5s	3s	+67%
`mlk_ct_cmask_nonzero_u8`	✅	4s	2s	+100%
`mlk_enc_getnoise_eta1_eta2`	✅	4s	5s	-20%
`mlk_keccak_absorb_once`	✅	4s	4s	+0%
`mlk_keccakf1600x4_xor_bytes`	✅	4s	4s	+0%
`mlk_matvec_mul`	✅	4s	4s	+0%
`mlk_poly_compress_d11`	✅	4s	2s	+100%
`mlk_poly_decompress_d5`	✅	4s	1s	+300%
`mlk_poly_frombytes_c`	✅	4s	5s	-20%
`mlk_poly_invntt_tomont`	✅	4s	2s	+100%
`mlk_poly_tobytes`	✅	4s	2s	+100%
`mlk_polyvec_permute_bitrev_to_custom_native`	✅	4s	1s	+300%
`mlk_scalar_decompress_d11`	✅	4s	2s	+100%
`mlk_scalar_decompress_d5`	✅	4s	2s	+100%
`mlk_shake128x4_squeezeblocks`	✅	4s	3s	+33%
`poly_compress_d10_native_x86_64`	✅	4s	1s	+300%
`poly_decompress_d5_native_x86_64`	✅	4s	4s	+0%
`polyvec_basemul_acc_montgomery_cached_k4_native_aarch64`	✅	4s	5s	-20%
`keccak_f1600_x1_native_aarch64`	✅	3s	1s	+200%
`keccak_f1600_x4_native_aarch64_v84a`	✅	3s	2s	+50%
`kem_check_sk`	✅	3s	3s	+0%
`kem_enc`	✅	3s	1s	+200%
`kem_enc_derand`	✅	3s	3s	+0%
`mlk_ct_cmov_zero`	✅	3s	2s	+50%
`mlk_keccakf1600_extract_bytes`	✅	3s	1s	+200%
`mlk_poly_cbd_eta2`	✅	3s	4s	-25%
`mlk_poly_compress_d10_native`	✅	3s	2s	+50%
`mlk_poly_compress_d11_c`	✅	3s	3s	+0%
`mlk_poly_compress_d11_native`	✅	3s	3s	+0%
`mlk_poly_compress_d4_native`	✅	3s	1s	+200%
`mlk_poly_compress_d5_native`	✅	3s	3s	+0%
`mlk_poly_decompress_d11_c`	✅	3s	1s	+200%
`mlk_poly_decompress_d4_c`	✅	3s	4s	-25%
`mlk_poly_getnoise_eta1122_4x`	✅	3s	3s	+0%
`mlk_poly_getnoise_eta1_4x_native`	✅	3s	4s	-25%
`mlk_poly_mulcache_compute_c`	✅	3s	4s	-25%
`mlk_poly_ntt_c`	✅	3s	2s	+50%
`mlk_poly_tomont_c`	✅	3s	2s	+50%
`mlk_poly_tomont_native`	✅	3s	2s	+50%
`mlk_polymat_permute_bitrev_to_custom`	✅	3s	3s	+0%
`mlk_polyvec_frombytes`	✅	3s	2s	+50%
`mlk_polyvec_tobytes`	✅	3s	2s	+50%
`mlk_polyvec_tomont`	✅	3s	2s	+50%
`mlk_scalar_compress_d1`	✅	3s	2s	+50%
`mlk_scalar_decompress_d10`	✅	3s	1s	+200%
`mlk_shake128x4_absorb_once`	✅	3s	2s	+50%
`mlk_shake256x4`	✅	3s	3s	+0%
`mlk_value_barrier_i32`	✅	3s	2s	+50%
`mlk_value_barrier_u32`	✅	3s	2s	+50%
`ntt_native_x86_64`	✅	3s	2s	+50%
`nttunpack_native_x86_64`	✅	3s	2s	+50%
`poly_decompress_d4_native_x86_64`	✅	3s	1s	+200%
`poly_frombytes_native_x86_64`	✅	3s	3s	+0%
`poly_getnoise_eta1122_4x_native`	✅	3s	1s	+200%
`poly_invntt_tomont_native`	✅	3s	1s	+200%
`poly_mulcache_compute_native_x86_64`	✅	3s	4s	-25%
`poly_reduce_native_aarch64`	✅	3s	1s	+200%
`poly_tomont_native_aarch64`	✅	3s	3s	+0%
`polyvec_basemul_acc_montgomery_cached_k2_native_aarch64`	✅	3s	3s	+0%
`polyvec_basemul_acc_montgomery_cached_k2_native_x86_64`	✅	3s	3s	+0%
`polyvec_basemul_acc_montgomery_cached_k3_native_aarch64`	✅	3s	4s	-25%
`sys_check_capability`	✅	3s	1s	+200%
`intt_native_aarch64`	✅	2s	2s	+0%
`keccak_f1600_x1_native_aarch64_v84a`	✅	2s	2s	+0%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	2s	1s	+100%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	2s	1s	+100%
`keccak_f1600_x4_native_avx2`	✅	2s	1s	+100%
`keccakf1600_permute_native`	✅	2s	3s	-33%
`kem_check_pk`	✅	2s	3s	-33%
`kem_keypair`	✅	2s	2s	+0%
`mlk_barrett_reduce`	✅	2s	2s	+0%
`mlk_check_pct`	✅	2s	5s	-60%
`mlk_ct_cmask_nonzero_u16`	✅	2s	2s	+0%
`mlk_ct_get_optblocker_u32`	✅	2s	4s	-50%
`mlk_keccakf1600_extract_bytes (big endian)`	✅	2s	2s	+0%
`mlk_keccakf1600_permute`	✅	2s	2s	+0%
`mlk_keccakf1600_xor_bytes`	✅	2s	4s	-50%
`mlk_keccakf1600_xor_bytes (big endian)`	✅	2s	2s	+0%
`mlk_keccakf1600x4_permute`	✅	2s	1s	+100%
`mlk_keccakf1600x4_xor_bytes_c`	✅	2s	2s	+0%
`mlk_keypair_getnoise_eta1`	✅	2s	1s	+100%
`mlk_montgomery_reduce`	✅	2s	3s	-33%
`mlk_poly_compress_d10`	✅	2s	3s	-33%
`mlk_poly_compress_d4`	✅	2s	3s	-33%
`mlk_poly_compress_d5`	✅	2s	2s	+0%
`mlk_poly_compress_d5_c`	✅	2s	2s	+0%
`mlk_poly_compress_dv`	✅	2s	2s	+0%
`mlk_poly_decompress_d10_c`	✅	2s	3s	-33%
`mlk_poly_decompress_d11`	✅	2s	3s	-33%
`mlk_poly_decompress_d4`	✅	2s	2s	+0%
`mlk_poly_decompress_d4_native`	✅	2s	2s	+0%
`mlk_poly_decompress_d5_c`	✅	2s	1s	+100%
`mlk_poly_decompress_dv`	✅	2s	2s	+0%
`mlk_poly_frombytes`	✅	2s	3s	-33%
`mlk_poly_getnoise_eta1_4x`	✅	2s	3s	-33%
`mlk_poly_getnoise_eta2`	✅	2s	3s	-33%
`mlk_poly_invntt_tomont_c`	✅	2s	3s	-33%
`mlk_poly_mulcache_compute`	✅	2s	3s	-33%
`mlk_poly_mulcache_compute_native`	✅	2s	2s	+0%
`mlk_poly_reduce`	✅	2s	2s	+0%
`mlk_poly_tobytes_c`	✅	2s	3s	-33%
`mlk_poly_tomont`	✅	2s	1s	+100%
`mlk_poly_tomsg`	✅	2s	4s	-50%
`mlk_polyvec_basemul_acc_montgomery_cached`	✅	2s	2s	+0%
`mlk_polyvec_compress_du`	✅	2s	1s	+100%
`mlk_polyvec_decompress_du`	✅	2s	2s	+0%
`mlk_polyvec_invntt_tomont`	✅	2s	2s	+0%
`mlk_polyvec_mulcache_compute`	✅	2s	2s	+0%
`mlk_polyvec_permute_bitrev_to_custom`	✅	2s	5s	-60%
`mlk_scalar_compress_d11`	✅	2s	2s	+0%
`mlk_scalar_compress_d4`	✅	2s	2s	+0%
`mlk_scalar_signed_to_unsigned_q`	✅	2s	1s	+100%
`mlk_sha3_256`	✅	2s	2s	+0%
`mlk_sha3_512`	✅	2s	1s	+100%
`mlk_shake128_squeezeblocks`	✅	2s	1s	+100%
`ntt_native_aarch64`	✅	2s	1s	+100%
`poly_compress_d4_native_x86_64`	✅	2s	3s	-33%
`poly_compress_d5_native_x86_64`	✅	2s	1s	+100%
`poly_mulcache_compute_native_aarch64`	✅	2s	4s	-50%
`poly_reduce_native_x86_64`	✅	2s	5s	-60%
`poly_tobytes_native_aarch64`	✅	2s	3s	-33%
`poly_tobytes_native_x86_64`	✅	2s	1s	+100%
`poly_tomont_native_x86_64`	✅	2s	3s	-33%
`polyvec_basemul_acc_montgomery_cached_k3_native_x86_64`	✅	2s	1s	+100%
`polyvec_basemul_acc_montgomery_cached_k4_native_x86_64`	✅	2s	3s	-33%
`rej_uniform_native`	✅	2s	4s	-50%
`keccakf1600x4_extract_bytes_native`	✅	1s	3s	-67%
`keccakf1600x4_xor_bytes_native`	✅	1s	2s	-50%
`kem_keypair_derand`	✅	1s	2s	-50%
`mlk_ct_cmask_neg_i16`	✅	1s	4s	-75%
`mlk_ct_get_optblocker_i32`	✅	1s	2s	-50%
`mlk_ct_get_optblocker_u8`	✅	1s	1s	+0%
`mlk_ct_memcmp`	✅	1s	3s	-67%
`mlk_ct_sel_uint8`	✅	1s	2s	-50%
`mlk_keccakf1600x4_extract_bytes`	✅	1s	2s	-50%
`mlk_keccakf1600x4_extract_bytes_c`	✅	1s	2s	-50%
`mlk_poly_cbd_eta1`	✅	1s	1s	+0%
`mlk_poly_compress_d10_c`	✅	1s	3s	-67%
`mlk_poly_compress_d4_c`	✅	1s	4s	-75%
`mlk_poly_compress_du`	✅	1s	3s	-67%
`mlk_poly_decompress_d10`	✅	1s	2s	-50%
`mlk_poly_decompress_d10_native`	✅	1s	2s	-50%
`mlk_poly_decompress_du`	✅	1s	2s	-50%
`mlk_poly_reduce_c`	✅	1s	1s	+0%
`mlk_poly_tobytes_native`	✅	1s	2s	-50%
`mlk_polyvec_reduce`	✅	1s	2s	-50%
`mlk_scalar_compress_d10`	✅	1s	1s	+0%
`mlk_scalar_compress_d5`	✅	1s	2s	-50%
`mlk_scalar_decompress_d4`	✅	1s	2s	-50%
`mlk_shake128_absorb_once`	✅	1s	3s	-67%
`mlk_shake256`	✅	1s	4s	-75%
`mlk_value_barrier_u8`	✅	1s	3s	-67%
`poly_decompress_d10_native_x86_64`	✅	1s	5s	-80%

oqs-bot · 2026-07-09T13:53:46Z

CBMC Results (ML-KEM-768)

Full Results (191 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	1768s	1768s	+0.0%
`mlk_indcpa_keypair_derand`	✅	333s	342s	-3%
`mlk_indcpa_enc`	✅	259s	252s	+3%
`mlk_poly_rej_uniform`	✅	157s	148s	+6%
`mlk_rej_uniform_c`	✅	128s	128s	+0%
`rej_uniform_native_x86_64`	✅	61s	61s	+0%
`mlk_polyvec_basemul_acc_montgomery_cached_c`	✅	51s	56s	-9%
`mlk_ntt_layer`	✅	43s	41s	+5%
`poly_ntt_native`	✅	39s	45s	-13%
`rej_uniform_native_aarch64`	✅	38s	38s	+0%
`mlk_poly_reduce_native`	✅	37s	36s	+3%
`mlk_keccak_squeezeblocks_x4`	✅	26s	24s	+8%
`mlk_fqmul`	✅	19s	17s	+12%
`mlk_poly_decompress_d10_native`	✅	16s	13s	+23%
`polyvec_basemul_acc_montgomery_cached_native`	✅	16s	16s	+0%
`keccakf1600x4_permute_native_x4`	✅	15s	17s	-12%
`mlk_poly_decompress_d4_native`	✅	15s	16s	-6%
`mlk_polyvec_add`	✅	12s	12s	+0%
`mlk_poly_frombytes_native`	✅	11s	8s	+38%
`mlk_indcpa_dec`	✅	9s	11s	-18%
`mlk_keccak_squeezeblocks`	✅	9s	9s	+0%
`mlk_poly_frommsg`	✅	9s	10s	-10%
`mlk_keccak_absorb_once_x4`	✅	8s	6s	+33%
`mlk_ntt_butterfly_block`	✅	8s	8s	+0%
`mlk_keccak_squeeze_once`	✅	7s	7s	+0%
`mlk_poly_ntt`	✅	7s	9s	-22%
`mlk_keccakf1600_permute_c`	✅	6s	5s	+20%
`mlk_poly_add`	✅	6s	5s	+20%
`mlk_poly_frombytes_c`	✅	6s	5s	+20%
`mlk_poly_rej_uniform_x4`	✅	6s	6s	+0%
`mlk_poly_sub`	✅	6s	8s	-25%
`mlk_keccakf1600x4_extract_bytes_c`	✅	5s	2s	+150%
`mlk_poly_compress_d11`	✅	5s	2s	+150%
`ntt_native_aarch64`	✅	5s	2s	+150%
`poly_decompress_d10_native_x86_64`	✅	5s	6s	-17%
`poly_decompress_d4_native_x86_64`	✅	5s	5s	+0%
`intt_native_x86_64`	✅	4s	4s	+0%
`keccakf1600x4_extract_bytes_native`	✅	4s	1s	+300%
`kem_dec`	✅	4s	2s	+100%
`kem_keypair`	✅	4s	1s	+300%
`mlk_ct_sel_int16`	✅	4s	2s	+100%
`mlk_invntt_layer`	✅	4s	5s	-20%
`mlk_keccak_absorb_once`	✅	4s	3s	+33%
`mlk_keccakf1600x4_extract_bytes`	✅	4s	4s	+0%
`mlk_keccakf1600x4_xor_bytes`	✅	4s	3s	+33%
`mlk_montgomery_reduce`	✅	4s	1s	+300%
`mlk_poly_decompress_d11_c`	✅	4s	2s	+100%
`mlk_poly_decompress_d5_native`	✅	4s	2s	+100%
`mlk_poly_decompress_du`	✅	4s	2s	+100%
`mlk_poly_mulcache_compute`	✅	4s	3s	+33%
`mlk_poly_tomont_native`	✅	4s	2s	+100%
`mlk_polymat_permute_bitrev_to_custom`	✅	4s	3s	+33%
`mlk_polyvec_permute_bitrev_to_custom`	✅	4s	5s	-20%
`poly_compress_d11_native_x86_64`	✅	4s	3s	+33%
`poly_frombytes_native_x86_64`	✅	4s	4s	+0%
`keccak_f1600_x1_native_aarch64_v84a`	✅	3s	3s	+0%
`keccakf1600x4_xor_bytes_native`	✅	3s	4s	-25%
`kem_enc`	✅	3s	1s	+200%
`kem_enc_derand`	✅	3s	3s	+0%
`kem_keypair_derand`	✅	3s	3s	+0%
`mlk_ct_cmask_nonzero_u16`	✅	3s	2s	+50%
`mlk_enc_getnoise_eta1_eta2`	✅	3s	2s	+50%
`mlk_gen_matrix`	✅	3s	5s	-40%
`mlk_gen_matrix_serial`	✅	3s	2s	+50%
`mlk_keccakf1600_extract_bytes`	✅	3s	4s	-25%
`mlk_keccakf1600_extract_bytes (big endian)`	✅	3s	3s	+0%
`mlk_keccakf1600_permute`	✅	3s	2s	+50%
`mlk_keccakf1600_xor_bytes`	✅	3s	3s	+0%
`mlk_keccakf1600_xor_bytes (big endian)`	✅	3s	3s	+0%
`mlk_matvec_mul`	✅	3s	3s	+0%
`mlk_poly_compress_d5_native`	✅	3s	3s	+0%
`mlk_poly_compress_dv`	✅	3s	2s	+50%
`mlk_poly_decompress_d10`	✅	3s	2s	+50%
`mlk_poly_decompress_d11`	✅	3s	1s	+200%
`mlk_poly_decompress_d4_c`	✅	3s	5s	-40%
`mlk_poly_decompress_d5`	✅	3s	4s	-25%
`mlk_poly_decompress_d5_c`	✅	3s	2s	+50%
`mlk_poly_frombytes`	✅	3s	2s	+50%
`mlk_poly_getnoise_eta1_4x`	✅	3s	3s	+0%
`mlk_poly_getnoise_eta1_4x_native`	✅	3s	5s	-40%
`mlk_poly_invntt_tomont_c`	✅	3s	2s	+50%
`mlk_poly_mulcache_compute_c`	✅	3s	4s	-25%
`mlk_poly_mulcache_compute_native`	✅	3s	1s	+200%
`mlk_poly_tobytes_c`	✅	3s	2s	+50%
`mlk_poly_tomont`	✅	3s	4s	-25%
`mlk_polyvec_frombytes`	✅	3s	1s	+200%
`mlk_polyvec_permute_bitrev_to_custom_native`	✅	3s	3s	+0%
`mlk_polyvec_reduce`	✅	3s	2s	+50%
`mlk_rej_uniform`	✅	3s	2s	+50%
`mlk_scalar_decompress_d4`	✅	3s	2s	+50%
`mlk_value_barrier_i32`	✅	3s	3s	+0%
`ntt_native_x86_64`	✅	3s	2s	+50%
`poly_compress_d4_native_x86_64`	✅	3s	2s	+50%
`poly_decompress_d11_native_x86_64`	✅	3s	3s	+0%
`poly_getnoise_eta1122_4x_native`	✅	3s	2s	+50%
`poly_invntt_tomont_native`	✅	3s	4s	-25%
`poly_mulcache_compute_native_aarch64`	✅	3s	2s	+50%
`poly_mulcache_compute_native_x86_64`	✅	3s	2s	+50%
`poly_reduce_native_aarch64`	✅	3s	1s	+200%
`poly_reduce_native_x86_64`	✅	3s	3s	+0%
`polyvec_basemul_acc_montgomery_cached_k2_native_x86_64`	✅	3s	3s	+0%
`polyvec_basemul_acc_montgomery_cached_k3_native_x86_64`	✅	3s	5s	-40%
`polyvec_basemul_acc_montgomery_cached_k4_native_x86_64`	✅	3s	3s	+0%
`rej_uniform_native`	✅	3s	3s	+0%
`intt_native_aarch64`	✅	2s	3s	-33%
`keccak_f1600_x1_native_aarch64`	✅	2s	1s	+100%
`keccak_f1600_x4_native_aarch64_v84a`	✅	2s	2s	+0%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	2s	2s	+0%
`keccak_f1600_x4_native_avx2`	✅	2s	2s	+0%
`keccakf1600_permute_native`	✅	2s	3s	-33%
`kem_check_pk`	✅	2s	3s	-33%
`kem_check_sk`	✅	2s	5s	-60%
`mlk_barrett_reduce`	✅	2s	1s	+100%
`mlk_check_pct`	✅	2s	1s	+100%
`mlk_ct_cmask_neg_i16`	✅	2s	1s	+100%
`mlk_ct_cmask_nonzero_u8`	✅	2s	2s	+0%
`mlk_ct_cmov_zero`	✅	2s	3s	-33%
`mlk_ct_get_optblocker_u32`	✅	2s	1s	+100%
`mlk_ct_get_optblocker_u8`	✅	2s	1s	+100%
`mlk_ct_memcmp`	✅	2s	1s	+100%
`mlk_keccakf1600x4_permute`	✅	2s	2s	+0%
`mlk_keypair_getnoise_eta1`	✅	2s	3s	-33%
`mlk_poly_cbd_eta1`	✅	2s	1s	+100%
`mlk_poly_cbd_eta2`	✅	2s	3s	-33%
`mlk_poly_compress_d10`	✅	2s	3s	-33%
`mlk_poly_compress_d10_c`	✅	2s	2s	+0%
`mlk_poly_compress_d10_native`	✅	2s	1s	+100%
`mlk_poly_compress_d11_c`	✅	2s	2s	+0%
`mlk_poly_compress_d11_native`	✅	2s	1s	+100%
`mlk_poly_compress_d4_c`	✅	2s	4s	-50%
`mlk_poly_compress_d4_native`	✅	2s	3s	-33%
`mlk_poly_compress_d5`	✅	2s	3s	-33%
`mlk_poly_compress_du`	✅	2s	1s	+100%
`mlk_poly_decompress_d11_native`	✅	2s	1s	+100%
`mlk_poly_decompress_d4`	✅	2s	3s	-33%
`mlk_poly_decompress_dv`	✅	2s	2s	+0%
`mlk_poly_getnoise_eta1122_4x`	✅	2s	1s	+100%
`mlk_poly_getnoise_eta2`	✅	2s	2s	+0%
`mlk_poly_ntt_c`	✅	2s	3s	-33%
`mlk_poly_reduce`	✅	2s	3s	-33%
`mlk_poly_reduce_c`	✅	2s	2s	+0%
`mlk_poly_tobytes`	✅	2s	3s	-33%
`mlk_poly_tobytes_native`	✅	2s	1s	+100%
`mlk_poly_tomont_c`	✅	2s	2s	+0%
`mlk_poly_tomsg`	✅	2s	6s	-67%
`mlk_polyvec_basemul_acc_montgomery_cached`	✅	2s	3s	-33%
`mlk_polyvec_compress_du`	✅	2s	3s	-33%
`mlk_polyvec_decompress_du`	✅	2s	3s	-33%
`mlk_polyvec_invntt_tomont`	✅	2s	3s	-33%
`mlk_polyvec_mulcache_compute`	✅	2s	3s	-33%
`mlk_polyvec_ntt`	✅	2s	2s	+0%
`mlk_polyvec_tomont`	✅	2s	2s	+0%
`mlk_scalar_compress_d1`	✅	2s	1s	+100%
`mlk_scalar_compress_d10`	✅	2s	3s	-33%
`mlk_scalar_compress_d11`	✅	2s	2s	+0%
`mlk_scalar_compress_d5`	✅	2s	3s	-33%
`mlk_scalar_decompress_d10`	✅	2s	4s	-50%
`mlk_scalar_decompress_d11`	✅	2s	2s	+0%
`mlk_scalar_decompress_d5`	✅	2s	4s	-50%
`mlk_scalar_signed_to_unsigned_q`	✅	2s	2s	+0%
`mlk_sha3_512`	✅	2s	2s	+0%
`mlk_shake128_absorb_once`	✅	2s	1s	+100%
`mlk_shake128_squeezeblocks`	✅	2s	2s	+0%
`mlk_shake128x4_absorb_once`	✅	2s	2s	+0%
`mlk_shake128x4_squeezeblocks`	✅	2s	2s	+0%
`mlk_shake256`	✅	2s	3s	-33%
`mlk_shake256x4`	✅	2s	5s	-60%
`nttunpack_native_x86_64`	✅	2s	3s	-33%
`poly_compress_d5_native_x86_64`	✅	2s	1s	+100%
`poly_decompress_d5_native_x86_64`	✅	2s	2s	+0%
`poly_tobytes_native_x86_64`	✅	2s	3s	-33%
`polyvec_basemul_acc_montgomery_cached_k3_native_aarch64`	✅	2s	1s	+100%
`polyvec_basemul_acc_montgomery_cached_k4_native_aarch64`	✅	2s	1s	+100%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	1s	1s	+0%
`mlk_ct_get_optblocker_i32`	✅	1s	3s	-67%
`mlk_ct_sel_uint8`	✅	1s	3s	-67%
`mlk_keccakf1600x4_xor_bytes_c`	✅	1s	2s	-50%
`mlk_poly_compress_d4`	✅	1s	5s	-80%
`mlk_poly_compress_d5_c`	✅	1s	2s	-50%
`mlk_poly_decompress_d10_c`	✅	1s	3s	-67%
`mlk_poly_invntt_tomont`	✅	1s	6s	-83%
`mlk_polyvec_tobytes`	✅	1s	4s	-75%
`mlk_scalar_compress_d4`	✅	1s	3s	-67%
`mlk_sha3_256`	✅	1s	1s	+0%
`mlk_value_barrier_u32`	✅	1s	1s	+0%
`mlk_value_barrier_u8`	✅	1s	3s	-67%
`poly_compress_d10_native_x86_64`	✅	1s	2s	-50%
`poly_tobytes_native_aarch64`	✅	1s	3s	-67%
`poly_tomont_native_aarch64`	✅	1s	1s	+0%
`poly_tomont_native_x86_64`	✅	1s	3s	-67%
`polyvec_basemul_acc_montgomery_cached_k2_native_aarch64`	✅	1s	2s	-50%
`sys_check_capability`	✅	1s	1s	+0%

bremoran · 2026-07-10T16:49:45Z

@hanno-becker how does the fix look to you?

bremoran requested a review from a team as a code owner June 29, 2026 08:38

mkannwischer added the needs-mldsa-native-port label Jun 29, 2026

hanno-becker reviewed Jun 30, 2026

View reviewed changes

bremoran mentioned this pull request Jul 8, 2026

C-Only fix for unaligned inputs to keccak x4 xor #1794

Closed

bremoran force-pushed the issue-1759-armv81m-xor-bytes branch from 889857e to ec577f6 Compare July 9, 2026 13:09

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Fix Armv8.1-M x4 xor_bytes alignment fallback - #1766

Fix Armv8.1-M x4 xor_bytes alignment fallback#1766
bremoran wants to merge 1 commit into
mainfrom
issue-1759-armv81m-xor-bytes

bremoran commented Jun 29, 2026

Uh oh!

hanno-becker Jun 30, 2026

Uh oh!

hanno-becker Jul 6, 2026

Uh oh!

oqs-bot commented Jul 9, 2026

Uh oh!

oqs-bot commented Jul 9, 2026

Uh oh!

oqs-bot commented Jul 9, 2026

Uh oh!

bremoran commented Jul 10, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Conversation

bremoran commented Jun 29, 2026

Uh oh!

hanno-becker Jun 30, 2026

Choose a reason for hiding this comment

Uh oh!

hanno-becker Jul 6, 2026

Choose a reason for hiding this comment

Uh oh!

oqs-bot commented Jul 9, 2026

CBMC Results (ML-KEM-512)

Uh oh!

oqs-bot commented Jul 9, 2026

CBMC Results (ML-KEM-1024)

Uh oh!

oqs-bot commented Jul 9, 2026

CBMC Results (ML-KEM-768)

Uh oh!

bremoran commented Jul 10, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants