pq_crypto 0.6.4 → 0.6.6

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{ntt_aarch64_asm.S → mldsa_ntt_aarch64_asm.S}

@@ -14,20 +14,53 @@
  *   Becker, Hwang, Kannwischer, Yang, Yang
  *   https://eprint.iacr.org/2021/986
  *
+ * - [NeonNTT_Autoformalised]
+ *   Neon NTT - (Auto)formalised
+ *   Hanno Becker
+ *   https://eprint.iacr.org/2026/1223
+ *
  * - [SLOTHY_Paper]
  *   Fast and Clean: Auditable high-performance assembly via constraint solving
  *   Abdulrahman, Becker, Kannwischer, Klein
  *   https://eprint.iacr.org/2022/1303
  */
 
-/* AArch64 ML-DSA forward NTT following @[NeonNTT] and @[SLOTHY_Paper] */
+/* AArch64 ML-DSA forward NTT following @[NeonNTT], @[SLOTHY_Paper], and @[NeonNTT_Autoformalised] */
+
+/*yaml
+  Name: ntt_aarch64_asm
+  Description: AArch64 ML-DSA forward NTT
+  Signature: void mld_ntt_aarch64_asm(int32_t r[256], const int32_t zetas_l123456[144], const int32_t zetas_l78[384])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: read/write
+      c_parameter: int32_t r[256]
+      description: Input/output polynomial (256 x int32_t)
+    x1:
+      type: buffer
+      size_bytes: 576
+      permissions: read-only
+      c_parameter: const int32_t zetas_l123456[144]
+      description: Twiddle factors for layers 1-6 (144 x int32_t)
+    x2:
+      type: buffer
+      size_bytes: 1536
+      permissions: read-only
+      c_parameter: const int32_t zetas_l78[384]
+      description: Twiddle factors for layers 7-8 (384 x int32_t)
+*/
 
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) &&  !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/ntt_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_ntt_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -138,7 +171,7 @@ MLD_ASM_FN_SYMBOL(ntt_aarch64_asm)
         sub v15.4s, v10.4s, v26.4s
         sub x4, x4, #0x2
 
-Lntt_layer123_start:
+Lmld_ntt_layer123_start:
         add v31.4s, v10.4s, v26.4s
         mul v17.4s, v19.4s, v1.s[2]
         add v26.4s, v15.4s, v23.4s
@@ -216,7 +249,7 @@ Lntt_layer123_start:
         sqrdmulh v12.4s, v19.4s, v1.s[3]
         sub v15.4s, v10.4s, v26.4s
         subs x4, x4, #0x1
-        cbnz x4, Lntt_layer123_start
+        cbnz x4, Lmld_ntt_layer123_start
         add v13.4s, v10.4s, v26.4s
         mls v18.4s, v5.4s, v7.s[0]
         str q22, [x0, #0x180]
@@ -378,7 +411,7 @@ Lntt_layer123_start:
         sub v31.4s, v6.4s, v9.4s
         sub x4, x4, #0x1
 
-Lntt_layer45678_start:
+Lmld_ntt_layer45678_start:
         add v2.4s, v13.4s, v12.4s
         sqrdmulh v5.4s, v30.4s, v20.4s
         sub v25.4s, v13.4s, v12.4s
@@ -544,7 +577,7 @@ Lntt_layer45678_start:
         trn2 v10.2d, v22.2d, v1.2d
         mul v28.4s, v30.4s, v3.4s
         subs x4, x4, #0x1
-        cbnz x4, Lntt_layer45678_start
+        cbnz x4, Lmld_ntt_layer45678_start
         add v9.4s, v6.4s, v9.4s
         sqrdmulh v6.4s, v30.4s, v20.4s
         ldur q24, [x2, #-0xa0]

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{pointwise_montgomery_aarch64_asm.S → mldsa_pointwise_montgomery_aarch64_asm.S}

@@ -2,6 +2,28 @@
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
 
+/*yaml
+  Name: poly_pointwise_montgomery_aarch64_asm
+  Description: AArch64 pointwise Montgomery multiplication of two polynomials
+  Signature: void mld_poly_pointwise_montgomery_aarch64_asm(int32_t a[256], const int32_t b[256])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: read/write
+      c_parameter: int32_t a[256]
+      description: Input/output polynomial (256 x int32_t)
+    x1:
+      type: buffer
+      size_bytes: 1024
+      permissions: read-only
+      c_parameter: const int32_t b[256]
+      description: Input polynomial (256 x int32_t)
+*/
+
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) &&                                      \
     (!defined(MLD_CONFIG_NO_SIGN_API) || !defined(MLD_CONFIG_NO_VERIFY_API) || \
@@ -10,7 +32,7 @@
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/pointwise_montgomery_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_pointwise_montgomery_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -27,7 +49,7 @@ MLD_ASM_FN_SYMBOL(poly_pointwise_montgomery_aarch64_asm)
         dup v1.4s, w3
         mov x3, #0x40               // =64
 
-Lpoly_pointwise_montgomery_loop_start:
+Lmld_poly_pointwise_montgomery_loop_start:
         ldr q16, [x0]
         ldr q17, [x0, #0x10]
         ldr q18, [x0, #0x20]
@@ -69,7 +91,7 @@ Lpoly_pointwise_montgomery_loop_start:
         str q19, [x0, #0x30]
         str q16, [x0], #0x40
         subs x3, x3, #0x4
-        cbnz x3, Lpoly_pointwise_montgomery_loop_start
+        cbnz x3, Lmld_poly_pointwise_montgomery_loop_start
         ret
         .cfi_endproc
 

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{poly_caddq_aarch64_asm.S → mldsa_poly_caddq_aarch64_asm.S}

@@ -2,13 +2,29 @@
  * Copyright (c) The mldsa-native project authors
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
+/*yaml
+  Name: poly_caddq_aarch64_asm
+  Description: AArch64 conditional addition of q to each coefficient
+  Signature: void mld_poly_caddq_aarch64_asm(int32_t a[256])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: read/write
+      c_parameter: int32_t a[256]
+      description: Input/output polynomial (256 x int32_t)
+*/
+
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/poly_caddq_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_poly_caddq_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -22,7 +38,7 @@ MLD_ASM_FN_SYMBOL(poly_caddq_aarch64_asm)
         dup v4.4s, w9
         mov x1, #0x10               // =16
 
-Lpoly_caddq_loop:
+Lmld_poly_caddq_loop:
         ldr q0, [x0]
         ldr q1, [x0, #0x10]
         ldr q2, [x0, #0x20]
@@ -40,7 +56,7 @@ Lpoly_caddq_loop:
         str q3, [x0, #0x30]
         str q0, [x0], #0x40
         subs x1, x1, #0x1
-        b.ne Lpoly_caddq_loop
+        b.ne Lmld_poly_caddq_loop
         ret
         .cfi_endproc
 

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{poly_chknorm_aarch64_asm.S → mldsa_poly_chknorm_aarch64_asm.S}

@@ -2,13 +2,34 @@
  * Copyright (c) The mldsa-native project authors
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
+/*yaml
+  Name: poly_chknorm_aarch64_asm
+  Description: AArch64 infinity-norm bound check on polynomial coefficients
+  Signature: int mld_poly_chknorm_aarch64_asm(const int32_t a[256], int32_t B)
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: read-only
+      c_parameter: const int32_t a[256]
+      description: Input polynomial (256 x int32_t)
+    x1:
+      type: scalar
+      c_parameter: int32_t B
+      description: Norm bound
+      test_with: 131072  # representative non-negative bound (1 << 17)
+*/
+
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/poly_chknorm_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_poly_chknorm_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -21,7 +42,7 @@ MLD_ASM_FN_SYMBOL(poly_chknorm_aarch64_asm)
         eor v21.16b, v21.16b, v21.16b
         mov x2, #0x10               // =16
 
-Lpoly_chknorm_loop:
+Lmld_poly_chknorm_loop:
         ldr q1, [x0, #0x10]
         ldr q2, [x0, #0x20]
         ldr q3, [x0, #0x30]
@@ -39,7 +60,7 @@ Lpoly_chknorm_loop:
         cmge v0.4s, v0.4s, v20.4s
         orr v21.16b, v21.16b, v0.16b
         subs x2, x2, #0x1
-        b.ne Lpoly_chknorm_loop
+        b.ne Lmld_poly_chknorm_loop
         umaxv s21, v21.4s
         fmov w0, s21
         and w0, w0, #0x1

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{poly_decompose_32_aarch64_asm.S → mldsa_poly_decompose_32_aarch64_asm.S}

@@ -2,6 +2,28 @@
  * Copyright (c) The mldsa-native project authors
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
+/*yaml
+  Name: poly_decompose_32_aarch64_asm
+  Description: AArch64 coefficient decomposition (alpha = (Q-1)/32)
+  Signature: void mld_poly_decompose_32_aarch64_asm(int32_t a1[256], int32_t a0[256])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: write-only
+      c_parameter: int32_t a1[256]
+      description: Output high-part polynomial (256 x int32_t)
+    x1:
+      type: buffer
+      size_bytes: 1024
+      permissions: read/write
+      c_parameter: int32_t a0[256]
+      description: Input polynomial / output low-part (256 x int32_t)
+*/
+
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_NO_SIGN_API) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
@@ -9,7 +31,7 @@
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/poly_decompose_32_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_poly_decompose_32_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -32,7 +54,7 @@ MLD_ASM_FN_SYMBOL(poly_decompose_32_aarch64_asm)
         dup v23.4s, w11
         mov x3, #0x10               // =16
 
-Lpoly_decompose_32_loop:
+Lmld_poly_decompose_32_loop:
         ldr q0, [x1]
         ldr q1, [x1, #0x10]
         ldr q2, [x1, #0x20]
@@ -70,7 +92,7 @@ Lpoly_decompose_32_loop:
         str q3, [x1, #0x30]
         str q0, [x1], #0x40
         subs x3, x3, #0x1
-        b.ne Lpoly_decompose_32_loop
+        b.ne Lmld_poly_decompose_32_loop
         ret
         .cfi_endproc
 

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{poly_decompose_88_aarch64_asm.S → mldsa_poly_decompose_88_aarch64_asm.S}

@@ -2,6 +2,28 @@
  * Copyright (c) The mldsa-native project authors
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
+/*yaml
+  Name: poly_decompose_88_aarch64_asm
+  Description: AArch64 coefficient decomposition (alpha = (Q-1)/88)
+  Signature: void mld_poly_decompose_88_aarch64_asm(int32_t a1[256], int32_t a0[256])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: write-only
+      c_parameter: int32_t a1[256]
+      description: Output high-part polynomial (256 x int32_t)
+    x1:
+      type: buffer
+      size_bytes: 1024
+      permissions: read/write
+      c_parameter: int32_t a0[256]
+      description: Input polynomial / output low-part (256 x int32_t)
+*/
+
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_NO_SIGN_API) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
@@ -9,7 +31,7 @@
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/poly_decompose_88_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_poly_decompose_88_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -32,7 +54,7 @@ MLD_ASM_FN_SYMBOL(poly_decompose_88_aarch64_asm)
         dup v23.4s, w11
         mov x3, #0x10               // =16
 
-Lpoly_decompose_88_loop:
+Lmld_poly_decompose_88_loop:
         ldr q0, [x1]
         ldr q1, [x1, #0x10]
         ldr q2, [x1, #0x20]
@@ -70,7 +92,7 @@ Lpoly_decompose_88_loop:
         str q3, [x1, #0x30]
         str q0, [x1], #0x40
         subs x3, x3, #0x1
-        b.ne Lpoly_decompose_88_loop
+        b.ne Lmld_poly_decompose_88_loop
         ret
         .cfi_endproc
 

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{poly_use_hint_32_aarch64_asm.S → mldsa_poly_use_hint_32_aarch64_asm.S}

@@ -2,6 +2,28 @@
  * Copyright (c) The mldsa-native project authors
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
+/*yaml
+  Name: poly_use_hint_32_aarch64_asm
+  Description: AArch64 hint application (alpha = (Q-1)/32)
+  Signature: void mld_poly_use_hint_32_aarch64_asm(int32_t a[256], const int32_t h[256])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: read/write
+      c_parameter: int32_t a[256]
+      description: Input/output polynomial (256 x int32_t)
+    x1:
+      type: buffer
+      size_bytes: 1024
+      permissions: read-only
+      c_parameter: const int32_t h[256]
+      description: Hint polynomial (256 x int32_t)
+*/
+
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_NO_VERIFY_API) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
@@ -9,7 +31,7 @@
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/poly_use_hint_32_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_poly_use_hint_32_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -33,7 +55,7 @@ MLD_ASM_FN_SYMBOL(poly_use_hint_32_aarch64_asm)
         movi v24.4s, #0xf
         mov x3, #0x10               // =16
 
-Lpoly_use_hint_32_loop:
+Lmld_poly_use_hint_32_loop:
         ldr q1, [x0, #0x10]
         ldr q2, [x0, #0x20]
         ldr q3, [x0, #0x30]
@@ -87,7 +109,7 @@ Lpoly_use_hint_32_loop:
         str q19, [x0, #0x30]
         str q16, [x0], #0x40
         subs x3, x3, #0x1
-        b.ne Lpoly_use_hint_32_loop
+        b.ne Lmld_poly_use_hint_32_loop
         ret
         .cfi_endproc
 

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{poly_use_hint_88_aarch64_asm.S → mldsa_poly_use_hint_88_aarch64_asm.S}

@@ -2,6 +2,28 @@
  * Copyright (c) The mldsa-native project authors
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
+/*yaml
+  Name: poly_use_hint_88_aarch64_asm
+  Description: AArch64 hint application (alpha = (Q-1)/88)
+  Signature: void mld_poly_use_hint_88_aarch64_asm(int32_t a[256], const int32_t h[256])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: read/write
+      c_parameter: int32_t a[256]
+      description: Input/output polynomial (256 x int32_t)
+    x1:
+      type: buffer
+      size_bytes: 1024
+      permissions: read-only
+      c_parameter: const int32_t h[256]
+      description: Hint polynomial (256 x int32_t)
+*/
+
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_NO_VERIFY_API) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
@@ -9,7 +31,7 @@
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/poly_use_hint_88_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_poly_use_hint_88_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -33,7 +55,7 @@ MLD_ASM_FN_SYMBOL(poly_use_hint_88_aarch64_asm)
         movi v24.4s, #0x2b
         mov x3, #0x10               // =16
 
-Lpoly_use_hint_88_loop:
+Lmld_poly_use_hint_88_loop:
         ldr q1, [x0, #0x10]
         ldr q2, [x0, #0x20]
         ldr q3, [x0, #0x30]
@@ -95,7 +117,7 @@ Lpoly_use_hint_88_loop:
         str q19, [x0, #0x30]
         str q16, [x0], #0x40
         subs x3, x3, #0x1
-        b.ne Lpoly_use_hint_88_loop
+        b.ne Lmld_poly_use_hint_88_loop
         ret
         .cfi_endproc
 

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{mld_polyvecl_pointwise_acc_montgomery_l4_aarch64_asm.S → mldsa_polyvecl_pointwise_acc_montgomery_l4_aarch64_asm.S}

@@ -2,13 +2,41 @@
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
 
+/*yaml
+  Name: polyvecl_pointwise_acc_montgomery_l4_aarch64_asm
+  Description: AArch64 pointwise multiply-accumulate of length-4 polynomial vectors
+  Signature: void mld_polyvecl_pointwise_acc_montgomery_l4_aarch64_asm(int32_t r[256], const int32_t a[4][256], const int32_t b[4][256])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: write-only
+      c_parameter: int32_t r[256]
+      description: Output polynomial (256 x int32_t)
+    x1:
+      type: buffer
+      size_bytes: 4096
+      permissions: read-only
+      c_parameter: const int32_t a[4][256]
+      description: Input polynomial vector a (4 x 256 x int32_t)
+    x2:
+      type: buffer
+      size_bytes: 4096
+      permissions: read-only
+      c_parameter: const int32_t b[4][256]
+      description: Input polynomial vector b (4 x 256 x int32_t)
+*/
+
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
            (defined(MLD_CONFIG_MULTILEVEL_WITH_SHARED) || MLDSA_L == 4)
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/mld_polyvecl_pointwise_acc_montgomery_l4_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_polyvecl_pointwise_acc_montgomery_l4_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -25,7 +53,7 @@ MLD_ASM_FN_SYMBOL(polyvecl_pointwise_acc_montgomery_l4_aarch64_asm)
         dup v1.4s, w3
         mov x3, #0x40               // =64
 
-Lpolyvecl_pointwise_acc_montgomery_l4_loop_start:
+Lmld_polyvecl_pointwise_acc_montgomery_l4_loop_start:
         ldr q17, [x1, #0x10]
         ldr q18, [x1, #0x20]
         ldr q19, [x1, #0x30]
@@ -115,7 +143,7 @@ Lpolyvecl_pointwise_acc_montgomery_l4_loop_start:
         str q19, [x0, #0x30]
         str q16, [x0], #0x40
         subs x3, x3, #0x4
-        cbnz x3, Lpolyvecl_pointwise_acc_montgomery_l4_loop_start
+        cbnz x3, Lmld_polyvecl_pointwise_acc_montgomery_l4_loop_start
         ret
         .cfi_endproc
 

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{mld_polyvecl_pointwise_acc_montgomery_l5_aarch64_asm.S → mldsa_polyvecl_pointwise_acc_montgomery_l5_aarch64_asm.S}

@@ -2,13 +2,41 @@
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
 
+/*yaml
+  Name: polyvecl_pointwise_acc_montgomery_l5_aarch64_asm
+  Description: AArch64 pointwise multiply-accumulate of length-5 polynomial vectors
+  Signature: void mld_polyvecl_pointwise_acc_montgomery_l5_aarch64_asm(int32_t r[256], const int32_t a[5][256], const int32_t b[5][256])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: write-only
+      c_parameter: int32_t r[256]
+      description: Output polynomial (256 x int32_t)
+    x1:
+      type: buffer
+      size_bytes: 5120
+      permissions: read-only
+      c_parameter: const int32_t a[5][256]
+      description: Input polynomial vector a (5 x 256 x int32_t)
+    x2:
+      type: buffer
+      size_bytes: 5120
+      permissions: read-only
+      c_parameter: const int32_t b[5][256]
+      description: Input polynomial vector b (5 x 256 x int32_t)
+*/
+
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
            (defined(MLD_CONFIG_MULTILEVEL_WITH_SHARED) || MLDSA_L == 5)
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/mld_polyvecl_pointwise_acc_montgomery_l5_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_polyvecl_pointwise_acc_montgomery_l5_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -25,7 +53,7 @@ MLD_ASM_FN_SYMBOL(polyvecl_pointwise_acc_montgomery_l5_aarch64_asm)
         dup v1.4s, w3
         mov x3, #0x40               // =64
 
-Lpolyvecl_pointwise_acc_montgomery_l5_loop_start:
+Lmld_polyvecl_pointwise_acc_montgomery_l5_loop_start:
         ldr q17, [x1, #0x10]
         ldr q18, [x1, #0x20]
         ldr q19, [x1, #0x30]
@@ -131,7 +159,7 @@ Lpolyvecl_pointwise_acc_montgomery_l5_loop_start:
         str q19, [x0, #0x30]
         str q16, [x0], #0x40
         subs x3, x3, #0x4
-        cbnz x3, Lpolyvecl_pointwise_acc_montgomery_l5_loop_start
+        cbnz x3, Lmld_polyvecl_pointwise_acc_montgomery_l5_loop_start
         ret
         .cfi_endproc
 

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/{mld_polyvecl_pointwise_acc_montgomery_l7_aarch64_asm.S → mldsa_polyvecl_pointwise_acc_montgomery_l7_aarch64_asm.S}

@@ -2,13 +2,41 @@
  * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
  */
 
+/*yaml
+  Name: polyvecl_pointwise_acc_montgomery_l7_aarch64_asm
+  Description: AArch64 pointwise multiply-accumulate of length-7 polynomial vectors
+  Signature: void mld_polyvecl_pointwise_acc_montgomery_l7_aarch64_asm(int32_t r[256], const int32_t a[7][256], const int32_t b[7][256])
+  ABI:
+    Architecture: aarch64
+    CallingConvention: AAPCS64
+    Features: [NEON]
+    x0:
+      type: buffer
+      size_bytes: 1024
+      permissions: write-only
+      c_parameter: int32_t r[256]
+      description: Output polynomial (256 x int32_t)
+    x1:
+      type: buffer
+      size_bytes: 7168
+      permissions: read-only
+      c_parameter: const int32_t a[7][256]
+      description: Input polynomial vector a (7 x 256 x int32_t)
+    x2:
+      type: buffer
+      size_bytes: 7168
+      permissions: read-only
+      c_parameter: const int32_t b[7][256]
+      description: Input polynomial vector b (7 x 256 x int32_t)
+*/
+
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
            (defined(MLD_CONFIG_MULTILEVEL_WITH_SHARED) || MLDSA_L == 7)
 
 /*
  * WARNING: This file is auto-derived from the mldsa-native source file
- *   dev/aarch64_opt/src/mld_polyvecl_pointwise_acc_montgomery_l7_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/aarch64_opt/src/mldsa_polyvecl_pointwise_acc_montgomery_l7_aarch64_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -25,7 +53,7 @@ MLD_ASM_FN_SYMBOL(polyvecl_pointwise_acc_montgomery_l7_aarch64_asm)
         dup v1.4s, w3
         mov x3, #0x40               // =64
 
-Lpolyvecl_pointwise_acc_montgomery_l7_loop_start:
+Lmld_polyvecl_pointwise_acc_montgomery_l7_loop_start:
         ldr q17, [x1, #0x10]
         ldr q18, [x1, #0x20]
         ldr q19, [x1, #0x30]
@@ -163,7 +191,7 @@ Lpolyvecl_pointwise_acc_montgomery_l7_loop_start:
         str q19, [x0, #0x30]
         str q16, [x0], #0x40
         subs x3, x3, #0x4
-        cbnz x3, Lpolyvecl_pointwise_acc_montgomery_l7_loop_start
+        cbnz x3, Lmld_polyvecl_pointwise_acc_montgomery_l7_loop_start
         ret
         .cfi_endproc