CodeQL: Add mldsa/ declaration visibility check

[hanno-becker]

CodeQL is GitHub's semantic code analysis engine: it builds a typed AST database from a build trace, then answers queries written in QL (a Datalog-flavoured language) against it.

This commit adds a custom query that flags any externally-linked declaration under mldsa/ lacking the appropriate visibility annotation (MLD_INTERNAL_API, MLD_EXTERNAL_API, MLD_API_QUALIFIER for functions; MLD_INTERNAL_DATA_DECLARATION / MLD_INTERNAL_DATA_DEFINITION for variables), catching new cross-TU symbols that miss the macro required for single-CU builds. Assembly symbols (asm) and empty-CU placeholders (*empty_cu) are excluded.

A scripts/ql driver wraps the CLI: it manages the CodeQL database (auto-rebuilding when sources change), dispatches queries, and runs query-specific post-processors that add colored output and pass/fail logic. The mldsa-declarations post-processor is registered for the visibility check; other queries get plain table output.

A codeql.yml workflow invoked from all.yml runs the check in CI.

The new query flagged two missing annotations:
mld_polyeta_unpack in poly_kl.c and mld_prepare_domain_separation_prefix in sign.c lacked MLD_INTERNAL_API / MLD_EXTERNAL_API, respectively. Both are fixed in this commit, too.

Licensing: CodeQL uses a GitHub-specific license and is marked 'unfree' in nixpkgs. It is free for open-source analysis and GitHub code scanning, covering our use case. The nix flake does not globally allowlist the package; developers opt in via
NIXPKGS_ALLOW_UNFREE=1 nix develop --impure .#codeql
to avoid silent acceptance of the non-OSS license.

[mkannwischer]

what's the benefit of installing codeql via nix in CI? Wouldn't github/codeql-action/init be the easier and less intrusive path?

[mkannwischer]

Or do we expect this to be very version sensitive? That's the only reason I can think of.

[hanno-becker]

I felt it's in keeping with our existing setup, and yes, more robust towards version changes. But I am OK either way.

[mkannwischer]

Thanks @hanno-becker. Thanks for looking into this and finding the two places where we were missing the markers.

I have some questions. If we plan to scale this, I think it makes sense to try to simplify this as much as possible.

[mkannwischer]

I'm not sure I see the value of this 640-line Python script.
What's the benefit over just changing the rules to actually throw errors and the workflow to

  - uses: github/codeql-action/init@x                                                                                                                                                                     
    with:                                                                                                            
      languages: cpp
      queries: ./test/codeql/mldsa-declarations.ql

One could still keep the ql script for local testing, but that sounds more like 10 line bash script rather than a 640 line Python script.

[hanno-becker]

Can you elaborate? What are you suggesting?

[mkannwischer]

I'll give it a try.

[hanno-becker]

Thanks! If we can directly trigger an error from the .ql that may be easier indeed.

[mkannwischer]

The following diff would allow us to install it unconditionally:

diff --git a/flake.nix b/flake.nix
index 3231ee183..83c7721d1 100644
--- a/flake.nix
+++ b/flake.nix
@@ -50,6 +50,8 @@
         {
           _module.args.pkgs = import inputs.nixpkgs {
             inherit system;
+            config.allowUnfreePredicate = pkg:
+              builtins.elem (inputs.nixpkgs.lib.getName pkg) [ "codeql" ];
             overlays = [
               (_:_: {
                 clang_22 = pkgs-unstable.clang_22;

With an appropriate comment explaining the license, I believe that may be preferable?

[mkannwischer]

If we expect the developer to run this, this should be included in the default shell.

[hanno-becker]

I would like the non-free license to be explicit, so the lack of convenience in the current PR is by design.

[mkannwischer]

If we consider this non-free, then I prefer to not use it at all.
There seem to be free alternatives. Let me try that.

[hanno-becker]

If we consider this non-free, then I prefer to not use it at all.

No need to be undifferentiated here. The license terms cover our use case. But I don't want to expose a nix shell which hides from the user the fact that not all use cases are covered. For that reason I also don't want it in the default shell, which I want to continue to work as-is.

[mkannwischer]

What happens if we forget to bump this? I would prefer leaving out this comment as this is guaranteed to be outdated soon.

[hanno-becker]

Ok!

[oqs-bot]

CBMC Results (ML-DSA-44)

Full Results (185 proofs)

Proof	Status	Current	Previous	Change
**TOTAL**	✅	1869s	1689s	+10.7%
sign_verify_internal	✅	241s	220s	+10%
rej_uniform_native	✅	136s	116s	+17%
polyvecl_pointwise_acc_montgomery_c	✅	126s	100s	+26%
poly_pointwise_montgomery_c	✅	122s	102s	+20%
mld_invntt_layer	✅	102s	89s	+15%
mld_ct_memcmp	✅	91s	75s	+21%
mld_attempt_signature_generation	✅	48s	48s	+0%
mld_ntt_layer	✅	46s	41s	+12%
polyvec_matrix_expand_eager	✅	33s	31s	+6%
fqmul	✅	31s	27s	+15%
sign_keypair_internal	✅	26s	23s	+13%
keccakf1600x4_permute_native	✅	24s	24s	+0%
rej_uniform	✅	24s	21s	+14%
sign_signature_internal	✅	24s	22s	+9%
poly_chknorm_c	✅	21s	17s	+24%
polyt0_unpack	✅	20s	21s	-5%
sign_pk_from_sk	✅	19s	18s	+6%
mld_ntt_butterfly_block	✅	17s	15s	+13%
rej_uniform_c	✅	17s	14s	+21%
polyveck_decompose	✅	14s	13s	+8%
poly_uniform_eta_4x	✅	13s	14s	-7%
poly_invntt_tomont_c	✅	12s	8s	+50%
poly_uniform_4x	✅	12s	11s	+9%
keccak_absorb_once_x4	✅	11s	9s	+22%
poly_add	✅	11s	11s	+0%
polyz_unpack_c	✅	11s	12s	-8%
mld_check_pct	✅	10s	9s	+11%
polyvec_matrix_expand_eager_serial	✅	10s	10s	+0%
polyveck_chknorm	✅	10s	7s	+43%
mld_compute_pack_z	✅	8s	7s	+14%
polyeta_unpack	✅	8s	5s	+60%
polyveck_add	✅	8s	4s	+100%
keccak_absorb	✅	7s	7s	+0%
keccakf1600_permute	✅	7s	8s	-12%
pointwise_acc_native_x86_64	✅	7s	6s	+17%
poly_power2round	✅	7s	8s	-12%
polyvec_matrix_pointwise_montgomery_eager	✅	7s	7s	+0%
polyveck_reduce	✅	7s	5s	+40%
polyvecl_unpack_eta	✅	7s	3s	+133%
sign_signature_pre_hash_internal	✅	7s	5s	+40%
unpack_sk	✅	7s	5s	+40%
keccakf1600_permute_native	✅	6s	7s	-14%
pointwise_acc_native_aarch64	✅	6s	6s	+0%
poly_caddq_c	✅	6s	5s	+20%
poly_chknorm_native_aarch64	✅	6s	3s	+100%
poly_decompose	✅	6s	2s	+200%
poly_uniform_eta	✅	6s	3s	+100%
poly_uniform_gamma1_4x	✅	6s	6s	+0%
polyt1_unpack	✅	6s	3s	+100%
polyveck_power2round	✅	6s	6s	+0%
polyveck_sub	✅	6s	4s	+50%
polyvecl_ntt	✅	6s	4s	+50%
rej_eta_native	✅	6s	5s	+20%
sign_verify_pre_hash_shake256	✅	6s	6s	+0%
keccak_squeezeblocks_x4	✅	5s	3s	+67%
keccakf1600_xor_bytes	✅	5s	3s	+67%
mld_ct_cmask_nonzero_u8	✅	5s	3s	+67%
mld_sample_s1_s2	✅	5s	6s	-17%
mld_sample_s1_s2_serial	✅	5s	3s	+67%
poly_caddq_native_aarch64	✅	5s	4s	+25%
poly_challenge	✅	5s	7s	-29%
poly_shiftl	✅	5s	6s	-17%
poly_uniform	✅	5s	4s	+25%
poly_uniform_gamma1	✅	5s	3s	+67%
poly_use_hint	✅	5s	3s	+67%
rej_eta	✅	5s	3s	+67%
shake256_absorb	✅	5s	3s	+67%
sign_signature_extmu	✅	5s	4s	+25%
sign_verify	✅	5s	4s	+25%
sign_verify_pre_hash_internal	✅	5s	4s	+25%
unpack_hints	✅	5s	4s	+25%
unpack_sig	✅	5s	4s	+25%
decompose	✅	4s	4s	+0%
intt_native_x86_64	✅	4s	2s	+100%
keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid	✅	4s	3s	+33%
keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid	✅	4s	4s	+0%
keccakf1600_extract_bytes (big endian)	✅	4s	2s	+100%
mld_ct_cmask_neg_i32	✅	4s	1s	+300%
mld_h	✅	4s	4s	+0%
mld_prepare_domain_separation_prefix	✅	4s	2s	+100%
mld_value_barrier_u8	✅	4s	2s	+100%
ntt_native_aarch64	✅	4s	3s	+33%
ntt_native_x86_64	✅	4s	4s	+0%
pack_pk	✅	4s	3s	+33%
pack_sig_h_poly	✅	4s	1s	+300%
pack_sk_s1	✅	4s	2s	+100%
pointwise_native_x86_64	✅	4s	4s	+0%
poly_decompose_c	✅	4s	5s	-20%
poly_decompose_native	✅	4s	3s	+33%
poly_make_hint	✅	4s	3s	+33%
poly_reduce	✅	4s	3s	+33%
poly_sub	✅	4s	3s	+33%
poly_use_hint_c	✅	4s	5s	-20%
polyt1_pack	✅	4s	2s	+100%
polyveck_caddq	✅	4s	5s	-20%
polyveck_invntt_tomont	✅	4s	4s	+0%
polyveck_ntt	✅	4s	3s	+33%
polyveck_pointwise_poly_montgomery	✅	4s	5s	-20%
polyveck_shiftl	✅	4s	5s	-20%
polyveck_unpack_eta	✅	4s	2s	+100%
polyveck_use_hint	✅	4s	5s	-20%
polyvecl_pack_eta	✅	4s	5s	-20%
polyvecl_uniform_gamma1	✅	4s	4s	+0%
polyvecl_uniform_gamma1_serial	✅	4s	5s	-20%
polyw1_pack	✅	4s	5s	-20%
polyz_unpack_native	✅	4s	3s	+33%
power2round	✅	4s	4s	+0%
shake256x4_squeezeblocks	✅	4s	2s	+100%
sign	✅	4s	5s	-20%
sign_open	✅	4s	5s	-20%
sign_verify_extmu	✅	4s	7s	-43%
use_hint	✅	4s	2s	+100%
fqscale	✅	3s	3s	+0%
keccak_init	✅	3s	2s	+50%
keccakf1600x4_extract_bytes	✅	3s	2s	+50%
make_hint	✅	3s	2s	+50%
mld_ct_get_optblocker_u8	✅	3s	2s	+50%
mld_ct_sel_int32	✅	3s	3s	+0%
montgomery_reduce	✅	3s	1s	+200%
pack_sig_c	✅	3s	4s	-25%
pack_sk_rho_key_tr_s2_t0	✅	3s	3s	+0%
pointwise_native_aarch64	✅	3s	4s	-25%
poly_caddq	✅	3s	2s	+50%
poly_caddq_native	✅	3s	4s	-25%
poly_chknorm	✅	3s	2s	+50%
poly_invntt_tomont	✅	3s	3s	+0%
poly_invntt_tomont_native	✅	3s	3s	+0%
poly_ntt_c	✅	3s	3s	+0%
poly_permute_bitrev_to_custom_optional	✅	3s	2s	+50%
poly_permute_bitrev_to_custom_optional_native	✅	3s	5s	-40%
poly_pointwise_montgomery	✅	3s	5s	-40%
polyt0_pack	✅	3s	5s	-40%
polyveck_pack_eta	✅	3s	4s	-25%
polyveck_pack_t0	✅	3s	3s	+0%
polyveck_pack_w1	✅	3s	5s	-40%
polyvecl_chknorm	✅	3s	3s	+0%
polyvecl_pointwise_acc_montgomery_native	✅	3s	6s	-50%
polyz_pack	✅	3s	3s	+0%
polyz_unpack	✅	3s	3s	+0%
rej_eta_c	✅	3s	4s	-25%
shake128_squeeze	✅	3s	4s	-25%
shake128x4_squeezeblocks	✅	3s	5s	-40%
shake256	✅	3s	3s	+0%
shake256_release	✅	3s	2s	+50%
shake256_squeeze	✅	3s	3s	+0%
shake256x4_absorb_once	✅	3s	3s	+0%
sign_keypair	✅	3s	3s	+0%
sign_signature	✅	3s	3s	+0%
sign_signature_pre_hash_shake256	✅	3s	6s	-50%
sys_check_capability	✅	3s	2s	+50%
caddq	✅	2s	4s	-50%
keccak_f1600_x1_native_aarch64_v84a	✅	2s	2s	+0%
keccak_finalize	✅	2s	2s	+0%
keccak_squeeze	✅	2s	4s	-50%
keccakf1600x4_xor_bytes	✅	2s	2s	+0%
mld_ct_abs_i32	✅	2s	2s	+0%
mld_ct_cmask_nonzero_u32	✅	2s	2s	+0%
mld_ct_get_optblocker_i64	✅	2s	2s	+0%
mld_keccakf1600_extract_bytes	✅	2s	3s	-33%
mld_value_barrier_u32	✅	2s	1s	+100%
pack_sig_z	✅	2s	2s	+0%
poly_chknorm_native	✅	2s	3s	-33%
poly_ntt_native	✅	2s	3s	-33%
poly_pointwise_montgomery_native	✅	2s	1s	+100%
poly_use_hint_native	✅	2s	2s	+0%
polyveck_unpack_t0	✅	2s	3s	-33%
polyvecl_pointwise_acc_montgomery	✅	2s	4s	-50%
polyvecl_unpack_z	✅	2s	2s	+0%
reduce32	✅	2s	3s	-33%
shake128_init	✅	2s	2s	+0%
shake128x4_absorb_once	✅	2s	3s	-33%
shake256_init	✅	2s	2s	+0%
unpack_pk	✅	2s	4s	-50%
keccak_f1600_x1_native_aarch64	✅	1s	3s	-67%
keccak_f1600_x4_native_aarch64_v84a	✅	1s	2s	-50%
keccakf1600_xor_bytes (big endian)	✅	1s	3s	-67%
keccakf1600x4_permute	✅	1s	4s	-75%
mld_ct_get_optblocker_u32	✅	1s	4s	-75%
mld_value_barrier_i64	✅	1s	4s	-75%
poly_ntt	✅	1s	1s	+0%
polyeta_pack	✅	1s	5s	-80%
shake128_absorb	✅	1s	2s	-50%
shake128_finalize	✅	1s	3s	-67%
shake128_release	✅	1s	4s	-75%
shake256_finalize	✅	1s	1s	+0%

[oqs-bot]

CBMC Results (ML-DSA-65)

Full Results (185 proofs)

Proof	Status	Current	Previous	Change
**TOTAL**	✅	2320s	2008s	+15.5%
polyvecl_pointwise_acc_montgomery_c	✅	294s	220s	+34%
sign_verify_internal	✅	256s	219s	+17%
polyvec_matrix_expand_eager	✅	193s	168s	+15%
rej_uniform_native	✅	139s	117s	+19%
poly_pointwise_montgomery_c	✅	132s	99s	+33%
mld_invntt_layer	✅	106s	94s	+13%
mld_ct_memcmp	✅	92s	74s	+24%
mld_attempt_signature_generation	✅	55s	49s	+12%
mld_ntt_layer	✅	49s	41s	+20%
fqmul	✅	32s	28s	+14%
sign_keypair_internal	✅	29s	28s	+4%
sign_signature_internal	✅	29s	26s	+12%
keccakf1600x4_permute_native	✅	27s	24s	+12%
polyvec_matrix_expand_eager_serial	✅	27s	27s	+0%
polyt0_unpack	✅	23s	17s	+35%
rej_uniform	✅	23s	20s	+15%
sign_pk_from_sk	✅	22s	19s	+16%
poly_chknorm_c	✅	20s	17s	+18%
polyveck_pointwise_poly_montgomery	✅	19s	15s	+27%
mld_ntt_butterfly_block	✅	18s	19s	-5%
rej_uniform_c	✅	17s	13s	+31%
polyveck_decompose	✅	16s	15s	+7%
polyveck_power2round	✅	15s	14s	+7%
poly_uniform_eta_4x	✅	14s	12s	+17%
polyveck_add	✅	14s	12s	+17%
poly_add	✅	13s	12s	+8%
poly_uniform_4x	✅	12s	11s	+9%
keccak_absorb_once_x4	✅	11s	10s	+10%
mld_check_pct	✅	10s	10s	+0%
poly_invntt_tomont_c	✅	10s	8s	+25%
polyvec_matrix_pointwise_montgomery_eager	✅	10s	9s	+11%
polyz_unpack_c	✅	10s	11s	-9%
keccakf1600_permute_native	✅	9s	8s	+12%
polyvecl_ntt	✅	9s	8s	+12%
keccak_absorb	✅	8s	7s	+14%
polyveck_ntt	✅	8s	7s	+14%
polyveck_shiftl	✅	8s	7s	+14%
polyveck_sub	✅	8s	6s	+33%
keccakf1600_permute	✅	7s	8s	-12%
ntt_native_x86_64	✅	7s	3s	+133%
pointwise_acc_native_aarch64	✅	7s	5s	+40%
pointwise_acc_native_x86_64	✅	7s	4s	+75%
polyveck_caddq	✅	7s	7s	+0%
polyveck_chknorm	✅	7s	6s	+17%
polyveck_invntt_tomont	✅	7s	6s	+17%
polyveck_use_hint	✅	7s	8s	-12%
sign	✅	7s	7s	+0%
sign_signature	✅	7s	3s	+133%
sign_verify	✅	7s	5s	+40%
poly_ntt	✅	6s	2s	+200%
poly_shiftl	✅	6s	4s	+50%
polyveck_reduce	✅	6s	8s	-25%
polyvecl_chknorm	✅	6s	3s	+100%
polyvecl_pointwise_acc_montgomery	✅	6s	3s	+100%
sign_signature_pre_hash_shake256	✅	6s	4s	+50%
sign_verify_pre_hash_shake256	✅	6s	3s	+100%
unpack_hints	✅	6s	6s	+0%
unpack_sk	✅	6s	4s	+50%
mld_compute_pack_z	✅	5s	4s	+25%
mld_sample_s1_s2	✅	5s	8s	-38%
pack_sk_rho_key_tr_s2_t0	✅	5s	2s	+150%
pack_sk_s1	✅	5s	3s	+67%
poly_caddq_c	✅	5s	3s	+67%
poly_caddq_native	✅	5s	3s	+67%
poly_chknorm	✅	5s	2s	+150%
poly_chknorm_native	✅	5s	4s	+25%
poly_decompose_c	✅	5s	4s	+25%
poly_power2round	✅	5s	5s	+0%
poly_uniform	✅	5s	8s	-38%
poly_uniform_eta	✅	5s	3s	+67%
poly_uniform_gamma1_4x	✅	5s	1s	+400%
polyveck_pack_t0	✅	5s	2s	+150%
rej_eta_native	✅	5s	3s	+67%
shake256	✅	5s	3s	+67%
shake256_init	✅	5s	4s	+25%
shake256_release	✅	5s	2s	+150%
sign_open	✅	5s	4s	+25%
sign_signature_pre_hash_internal	✅	5s	8s	-38%
intt_native_x86_64	✅	4s	2s	+100%
keccak_f1600_x1_native_aarch64	✅	4s	5s	-20%
keccak_f1600_x1_native_aarch64_v84a	✅	4s	2s	+100%
keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid	✅	4s	1s	+300%
keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid	✅	4s	4s	+0%
keccak_squeezeblocks_x4	✅	4s	6s	-33%
mld_h	✅	4s	2s	+100%
mld_prepare_domain_separation_prefix	✅	4s	5s	-20%
mld_sample_s1_s2_serial	✅	4s	5s	-20%
pointwise_native_x86_64	✅	4s	4s	+0%
poly_chknorm_native_aarch64	✅	4s	2s	+100%
poly_decompose	✅	4s	3s	+33%
poly_decompose_native	✅	4s	5s	-20%
poly_permute_bitrev_to_custom_optional	✅	4s	3s	+33%
poly_permute_bitrev_to_custom_optional_native	✅	4s	2s	+100%
poly_pointwise_montgomery_native	✅	4s	2s	+100%
poly_reduce	✅	4s	3s	+33%
poly_use_hint	✅	4s	2s	+100%
poly_use_hint_c	✅	4s	4s	+0%
polyeta_unpack	✅	4s	4s	+0%
polyveck_pack_w1	✅	4s	2s	+100%
polyveck_unpack_t0	✅	4s	5s	-20%
polyvecl_pointwise_acc_montgomery_native	✅	4s	6s	-33%
polyvecl_unpack_eta	✅	4s	2s	+100%
polyz_unpack	✅	4s	4s	+0%
power2round	✅	4s	5s	-20%
shake128_finalize	✅	4s	3s	+33%
sign_signature_extmu	✅	4s	7s	-43%
sign_verify_extmu	✅	4s	5s	-20%
caddq	✅	3s	4s	-25%
fqscale	✅	3s	3s	+0%
keccak_f1600_x4_native_aarch64_v84a	✅	3s	3s	+0%
keccak_finalize	✅	3s	5s	-40%
keccakf1600_extract_bytes (big endian)	✅	3s	3s	+0%
keccakf1600_xor_bytes	✅	3s	1s	+200%
keccakf1600x4_extract_bytes	✅	3s	1s	+200%
mld_ct_cmask_neg_i32	✅	3s	2s	+50%
mld_ct_get_optblocker_u8	✅	3s	3s	+0%
mld_ct_sel_int32	✅	3s	3s	+0%
mld_value_barrier_u8	✅	3s	2s	+50%
montgomery_reduce	✅	3s	3s	+0%
ntt_native_aarch64	✅	3s	2s	+50%
pack_sig_c	✅	3s	3s	+0%
pack_sig_z	✅	3s	4s	-25%
pointwise_native_aarch64	✅	3s	4s	-25%
poly_caddq	✅	3s	6s	-50%
poly_caddq_native_aarch64	✅	3s	3s	+0%
poly_challenge	✅	3s	3s	+0%
poly_ntt_c	✅	3s	3s	+0%
poly_sub	✅	3s	2s	+50%
poly_uniform_gamma1	✅	3s	3s	+0%
polyeta_pack	✅	3s	3s	+0%
polyt0_pack	✅	3s	4s	-25%
polyveck_unpack_eta	✅	3s	2s	+50%
polyvecl_pack_eta	✅	3s	3s	+0%
polyvecl_uniform_gamma1_serial	✅	3s	4s	-25%
polyz_pack	✅	3s	2s	+50%
rej_eta	✅	3s	2s	+50%
shake128_squeeze	✅	3s	2s	+50%
shake128x4_absorb_once	✅	3s	1s	+200%
shake256_finalize	✅	3s	3s	+0%
sign_keypair	✅	3s	3s	+0%
sys_check_capability	✅	3s	1s	+200%
unpack_pk	✅	3s	4s	-25%
use_hint	✅	3s	4s	-25%
keccak_squeeze	✅	2s	2s	+0%
keccakf1600_xor_bytes (big endian)	✅	2s	3s	-33%
keccakf1600x4_permute	✅	2s	3s	-33%
make_hint	✅	2s	3s	-33%
mld_ct_abs_i32	✅	2s	4s	-50%
mld_ct_cmask_nonzero_u32	✅	2s	4s	-50%
mld_ct_get_optblocker_i64	✅	2s	3s	-33%
mld_ct_get_optblocker_u32	✅	2s	2s	+0%
mld_keccakf1600_extract_bytes	✅	2s	3s	-33%
mld_value_barrier_u32	✅	2s	4s	-50%
pack_pk	✅	2s	6s	-67%
pack_sig_h_poly	✅	2s	2s	+0%
poly_invntt_tomont	✅	2s	2s	+0%
poly_invntt_tomont_native	✅	2s	4s	-50%
poly_make_hint	✅	2s	3s	-33%
poly_ntt_native	✅	2s	4s	-50%
poly_pointwise_montgomery	✅	2s	2s	+0%
poly_use_hint_native	✅	2s	1s	+100%
polyt1_pack	✅	2s	2s	+0%
polyt1_unpack	✅	2s	1s	+100%
polyveck_pack_eta	✅	2s	4s	-50%
polyvecl_uniform_gamma1	✅	2s	5s	-60%
polyvecl_unpack_z	✅	2s	6s	-67%
polyw1_pack	✅	2s	2s	+0%
polyz_unpack_native	✅	2s	2s	+0%
reduce32	✅	2s	2s	+0%
rej_eta_c	✅	2s	2s	+0%
shake128_absorb	✅	2s	2s	+0%
shake128_init	✅	2s	4s	-50%
shake128_release	✅	2s	3s	-33%
shake128x4_squeezeblocks	✅	2s	2s	+0%
shake256_absorb	✅	2s	2s	+0%
shake256x4_absorb_once	✅	2s	5s	-60%
shake256x4_squeezeblocks	✅	2s	2s	+0%
sign_verify_pre_hash_internal	✅	2s	6s	-67%
unpack_sig	✅	2s	4s	-50%
decompose	✅	1s	1s	+0%
keccak_init	✅	1s	3s	-67%
keccakf1600x4_xor_bytes	✅	1s	2s	-50%
mld_ct_cmask_nonzero_u8	✅	1s	2s	-50%
mld_value_barrier_i64	✅	1s	2s	-50%
shake256_squeeze	✅	1s	2s	-50%

[oqs-bot]

CBMC Results (ML-DSA-87)

Full Results (185 proofs)

Proof	Status	Current	Previous	Change
**TOTAL**	✅	2465s	2253s	+9.4%
polyvecl_pointwise_acc_montgomery_c	✅	351s	267s	+31%
sign_verify_internal	✅	297s	282s	+5%
polyvec_matrix_expand_eager	✅	150s	146s	+3%
rej_uniform_native	✅	133s	126s	+6%
poly_pointwise_montgomery_c	✅	129s	106s	+22%
mld_invntt_layer	✅	104s	98s	+6%
mld_ct_memcmp	✅	90s	80s	+12%
mld_attempt_signature_generation	✅	87s	84s	+4%
polyvec_matrix_expand_eager_serial	✅	64s	63s	+2%
sign_keypair_internal	✅	50s	50s	+0%
mld_ntt_layer	✅	46s	42s	+10%
fqmul	✅	31s	29s	+7%
sign_pk_from_sk	✅	28s	27s	+4%
sign_signature_internal	✅	28s	24s	+17%
keccakf1600x4_permute_native	✅	23s	24s	-4%
polyt0_unpack	✅	22s	17s	+29%
rej_uniform	✅	22s	24s	-8%
poly_chknorm_c	✅	20s	18s	+11%
mld_ntt_butterfly_block	✅	18s	16s	+12%
polyveck_decompose	✅	18s	17s	+6%
rej_uniform_c	✅	16s	14s	+14%
polyveck_chknorm	✅	15s	13s	+15%
polyveck_sub	✅	15s	19s	-21%
polyveck_add	✅	13s	14s	-7%
poly_add	✅	12s	12s	+0%
poly_uniform_eta_4x	✅	12s	14s	-14%
polyvec_matrix_pointwise_montgomery_eager	✅	12s	13s	-8%
keccak_absorb_once_x4	✅	11s	9s	+22%
poly_uniform_4x	✅	11s	12s	-8%
pointwise_acc_native_x86_64	✅	10s	8s	+25%
poly_decompose_c	✅	10s	7s	+43%
polyveck_pointwise_poly_montgomery	✅	10s	10s	+0%
mld_compute_pack_z	✅	9s	10s	-10%
poly_invntt_tomont_c	✅	9s	9s	+0%
poly_power2round	✅	9s	7s	+29%
polyeta_unpack	✅	9s	9s	+0%
polyveck_invntt_tomont	✅	9s	11s	-18%
polyvecl_ntt	✅	9s	10s	-10%
keccakf1600_permute	✅	8s	8s	+0%
keccakf1600_permute_native	✅	8s	8s	+0%
polyveck_power2round	✅	8s	11s	-27%
rej_eta_native	✅	8s	4s	+100%
sign	✅	8s	8s	+0%
keccak_absorb	✅	7s	7s	+0%
mld_check_pct	✅	7s	9s	-22%
mld_sample_s1_s2	✅	7s	7s	+0%
pointwise_acc_native_aarch64	✅	7s	7s	+0%
poly_use_hint_c	✅	7s	3s	+133%
polyveck_ntt	✅	7s	6s	+17%
keccak_finalize	✅	6s	2s	+200%
mld_prepare_domain_separation_prefix	✅	6s	4s	+50%
poly_caddq_native_aarch64	✅	6s	3s	+100%
polyveck_caddq	✅	6s	5s	+20%
polyveck_reduce	✅	6s	7s	-14%
polyveck_shiftl	✅	6s	3s	+100%
polyveck_unpack_eta	✅	6s	2s	+200%
polyveck_use_hint	✅	6s	7s	-14%
polyvecl_chknorm	✅	6s	3s	+100%
polyvecl_pointwise_acc_montgomery_native	✅	6s	1s	+500%
polyvecl_uniform_gamma1	✅	6s	4s	+50%
polyvecl_uniform_gamma1_serial	✅	6s	7s	-14%
polyz_unpack_c	✅	6s	4s	+50%
sign_verify	✅	6s	6s	+0%
unpack_hints	✅	6s	6s	+0%
unpack_sk	✅	6s	7s	-14%
mld_h	✅	5s	3s	+67%
pack_pk	✅	5s	4s	+25%
poly_caddq_native	✅	5s	2s	+150%
poly_ntt_c	✅	5s	3s	+67%
poly_shiftl	✅	5s	4s	+25%
polyeta_pack	✅	5s	3s	+67%
polyveck_pack_t0	✅	5s	4s	+25%
polyvecl_unpack_z	✅	5s	3s	+67%
shake256_release	✅	5s	1s	+400%
shake256_squeeze	✅	5s	2s	+150%
sign_signature	✅	5s	4s	+25%
sign_signature_pre_hash_shake256	✅	5s	6s	-17%
sign_verify_extmu	✅	5s	5s	+0%
caddq	✅	4s	6s	-33%
decompose	✅	4s	2s	+100%
fqscale	✅	4s	3s	+33%
intt_native_x86_64	✅	4s	4s	+0%
keccak_f1600_x1_native_aarch64	✅	4s	2s	+100%
keccak_squeezeblocks_x4	✅	4s	5s	-20%
mld_ct_cmask_nonzero_u32	✅	4s	5s	-20%
mld_sample_s1_s2_serial	✅	4s	7s	-43%
mld_value_barrier_i64	✅	4s	1s	+300%
ntt_native_aarch64	✅	4s	4s	+0%
pack_sig_c	✅	4s	2s	+100%
poly_caddq_c	✅	4s	4s	+0%
poly_challenge	✅	4s	3s	+33%
poly_chknorm_native	✅	4s	5s	-20%
poly_pointwise_montgomery_native	✅	4s	4s	+0%
poly_uniform	✅	4s	4s	+0%
poly_uniform_eta	✅	4s	3s	+33%
poly_uniform_gamma1_4x	✅	4s	5s	-20%
poly_use_hint_native	✅	4s	4s	+0%
polyveck_unpack_t0	✅	4s	6s	-33%
polyvecl_pack_eta	✅	4s	3s	+33%
polyvecl_pointwise_acc_montgomery	✅	4s	3s	+33%
polyvecl_unpack_eta	✅	4s	5s	-20%
polyz_unpack	✅	4s	3s	+33%
polyz_unpack_native	✅	4s	3s	+33%
power2round	✅	4s	2s	+100%
rej_eta_c	✅	4s	5s	-20%
shake256_init	✅	4s	2s	+100%
sign_open	✅	4s	4s	+0%
sign_verify_pre_hash_shake256	✅	4s	4s	+0%
unpack_pk	✅	4s	4s	+0%
keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid	✅	3s	4s	-25%
keccak_init	✅	3s	1s	+200%
keccak_squeeze	✅	3s	3s	+0%
keccakf1600_extract_bytes (big endian)	✅	3s	2s	+50%
keccakf1600_xor_bytes	✅	3s	2s	+50%
keccakf1600_xor_bytes (big endian)	✅	3s	2s	+50%
mld_ct_abs_i32	✅	3s	2s	+50%
mld_ct_cmask_neg_i32	✅	3s	2s	+50%
mld_ct_get_optblocker_i64	✅	3s	2s	+50%
mld_ct_sel_int32	✅	3s	2s	+50%
mld_value_barrier_u32	✅	3s	2s	+50%
pack_sk_s1	✅	3s	3s	+0%
pointwise_native_aarch64	✅	3s	3s	+0%
poly_caddq	✅	3s	3s	+0%
poly_chknorm	✅	3s	2s	+50%
poly_chknorm_native_aarch64	✅	3s	3s	+0%
poly_decompose	✅	3s	3s	+0%
poly_invntt_tomont	✅	3s	3s	+0%
poly_make_hint	✅	3s	4s	-25%
poly_permute_bitrev_to_custom_optional	✅	3s	2s	+50%
poly_pointwise_montgomery	✅	3s	3s	+0%
poly_uniform_gamma1	✅	3s	2s	+50%
poly_use_hint	✅	3s	4s	-25%
polyt0_pack	✅	3s	2s	+50%
polyt1_pack	✅	3s	1s	+200%
polyt1_unpack	✅	3s	1s	+200%
polyveck_pack_eta	✅	3s	3s	+0%
polyveck_pack_w1	✅	3s	3s	+0%
polyw1_pack	✅	3s	1s	+200%
polyz_pack	✅	3s	3s	+0%
reduce32	✅	3s	2s	+50%
rej_eta	✅	3s	3s	+0%
shake128_finalize	✅	3s	2s	+50%
shake256_absorb	✅	3s	2s	+50%
shake256_finalize	✅	3s	2s	+50%
shake256x4_absorb_once	✅	3s	4s	-25%
sign_keypair	✅	3s	2s	+50%
sign_signature_extmu	✅	3s	6s	-50%
unpack_sig	✅	3s	3s	+0%
keccak_f1600_x1_native_aarch64_v84a	✅	2s	2s	+0%
keccak_f1600_x4_native_aarch64_v84a	✅	2s	3s	-33%
keccakf1600x4_extract_bytes	✅	2s	3s	-33%
keccakf1600x4_permute	✅	2s	1s	+100%
keccakf1600x4_xor_bytes	✅	2s	3s	-33%
make_hint	✅	2s	2s	+0%
mld_ct_cmask_nonzero_u8	✅	2s	2s	+0%
mld_ct_get_optblocker_u32	✅	2s	2s	+0%
mld_ct_get_optblocker_u8	✅	2s	2s	+0%
mld_value_barrier_u8	✅	2s	3s	-33%
montgomery_reduce	✅	2s	3s	-33%
ntt_native_x86_64	✅	2s	4s	-50%
pack_sig_h_poly	✅	2s	5s	-60%
pack_sig_z	✅	2s	3s	-33%
pack_sk_rho_key_tr_s2_t0	✅	2s	4s	-50%
pointwise_native_x86_64	✅	2s	3s	-33%
poly_decompose_native	✅	2s	3s	-33%
poly_invntt_tomont_native	✅	2s	3s	-33%
poly_ntt	✅	2s	2s	+0%
poly_ntt_native	✅	2s	4s	-50%
poly_reduce	✅	2s	4s	-50%
poly_sub	✅	2s	2s	+0%
shake128_init	✅	2s	4s	-50%
shake128_release	✅	2s	3s	-33%
shake128_squeeze	✅	2s	1s	+100%
shake128x4_absorb_once	✅	2s	3s	-33%
shake128x4_squeezeblocks	✅	2s	2s	+0%
shake256x4_squeezeblocks	✅	2s	2s	+0%
sign_signature_pre_hash_internal	✅	2s	3s	-33%
sign_verify_pre_hash_internal	✅	2s	3s	-33%
sys_check_capability	✅	2s	3s	-33%
use_hint	✅	2s	2s	+0%
keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid	✅	1s	2s	-50%
mld_keccakf1600_extract_bytes	✅	1s	3s	-67%
poly_permute_bitrev_to_custom_optional_native	✅	1s	4s	-75%
shake128_absorb	✅	1s	1s	+0%
shake256	✅	1s	4s	-75%

[hanno-becker]

Shelving for now, @mkannwischer is not convinced of the approach and will explore alternatives first. We can always come back to this.

[mkannwischer]

Reopening so we don't forget this exists.