pq_crypto 0.6.4 → 0.6.5

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{poly_compress_d5_avx2_asm.S → mlkem_poly_compress_d5_avx2_asm.S}

@@ -31,11 +31,41 @@
 #include "../../../common.h"
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED) && \
+    (!defined(MLK_CONFIG_NO_ENCAPS_API) || \
+     !defined(MLK_CONFIG_NO_DECAPS_API)) && \
     (defined(MLK_CONFIG_MULTILEVEL_WITH_SHARED) || MLKEM_K == 4)
+/*yaml
+  Name: poly_compress_d5_avx2_asm
+  Description: x86_64 AVX2 polynomial compression (d=5)
+  Signature: void mlk_poly_compress_d5_avx2_asm(uint8_t *r, const int16_t *a, const uint8_t *data)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 160
+      permissions: write-only
+      c_parameter: uint8_t *r
+      description: Output compressed polynomial
+    rsi:
+      type: buffer
+      size_bytes: 512
+      permissions: read-only
+      c_parameter: const int16_t *a
+      description: Input polynomial (256 x int16_t)
+    rdx:
+      type: buffer
+      size_bytes: 32
+      permissions: read-only
+      c_parameter: const uint8_t *data
+      description: Precomputed compression constants
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/poly_compress_d5_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_poly_compress_d5_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -213,7 +243,8 @@ MLK_ASM_FN_SYMBOL(poly_compress_d5_avx2_asm)
 MLK_ASM_FN_SIZE(poly_compress_d5_avx2_asm)
 
 #endif /* MLK_ARITH_BACKEND_X86_64_DEFAULT && !MLK_CONFIG_MULTILEVEL_NO_SHARED \
-          && (MLK_CONFIG_MULTILEVEL_WITH_SHARED || MLKEM_K == 4) */
+          && (!MLK_CONFIG_NO_ENCAPS_API || !MLK_CONFIG_NO_DECAPS_API) && \
+          (MLK_CONFIG_MULTILEVEL_WITH_SHARED || MLKEM_K == 4) */
 
 #if defined(__ELF__)
 .section .note.GNU-stack,"",%progbits

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{poly_decompress_d10_avx2_asm.S → mlkem_poly_decompress_d10_avx2_asm.S}

@@ -31,11 +31,40 @@
 #include "../../../common.h"
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED) && \
+    !defined(MLK_CONFIG_NO_DECAPS_API) && \
     (defined(MLK_CONFIG_MULTILEVEL_WITH_SHARED) || MLKEM_K == 2 || MLKEM_K == 3)
+/*yaml
+  Name: poly_decompress_d10_avx2_asm
+  Description: x86_64 AVX2 polynomial decompression (d=10)
+  Signature: void mlk_poly_decompress_d10_avx2_asm(int16_t *r, const uint8_t *a, const uint8_t *data)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: write-only
+      c_parameter: int16_t *r
+      description: Output polynomial (256 x int16_t)
+    rsi:
+      type: buffer
+      size_bytes: 320
+      permissions: read-only
+      c_parameter: const uint8_t *a
+      description: Input compressed polynomial
+    rdx:
+      type: buffer
+      size_bytes: 32
+      permissions: read-only
+      c_parameter: const uint8_t *data
+      description: Precomputed decompression constants
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/poly_decompress_d10_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_poly_decompress_d10_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -220,8 +249,8 @@ MLK_ASM_FN_SYMBOL(poly_decompress_d10_avx2_asm)
 MLK_ASM_FN_SIZE(poly_decompress_d10_avx2_asm)
 
 #endif /* MLK_ARITH_BACKEND_X86_64_DEFAULT && !MLK_CONFIG_MULTILEVEL_NO_SHARED \
-          && (MLK_CONFIG_MULTILEVEL_WITH_SHARED || MLKEM_K == 2 || MLKEM_K == \
-          3) */
+          && !MLK_CONFIG_NO_DECAPS_API && (MLK_CONFIG_MULTILEVEL_WITH_SHARED || \
+          MLKEM_K == 2 || MLKEM_K == 3) */
 
 #if defined(__ELF__)
 .section .note.GNU-stack,"",%progbits

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{poly_decompress_d11_avx2_asm.S → mlkem_poly_decompress_d11_avx2_asm.S}

@@ -33,11 +33,40 @@
 #include "../../../common.h"
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED) && \
+    !defined(MLK_CONFIG_NO_DECAPS_API) && \
     (defined(MLK_CONFIG_MULTILEVEL_WITH_SHARED) || MLKEM_K == 4)
+/*yaml
+  Name: poly_decompress_d11_avx2_asm
+  Description: x86_64 AVX2 polynomial decompression (d=11)
+  Signature: void mlk_poly_decompress_d11_avx2_asm(int16_t *r, const uint8_t *a, const uint8_t *data)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: write-only
+      c_parameter: int16_t *r
+      description: Output polynomial (256 x int16_t)
+    rsi:
+      type: buffer
+      size_bytes: 352
+      permissions: read-only
+      c_parameter: const uint8_t *a
+      description: Input compressed polynomial
+    rdx:
+      type: buffer
+      size_bytes: 128
+      permissions: read-only
+      c_parameter: const uint8_t *data
+      description: Precomputed decompression constants
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/poly_decompress_d11_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_poly_decompress_d11_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -270,7 +299,8 @@ MLK_ASM_FN_SYMBOL(poly_decompress_d11_avx2_asm)
 MLK_ASM_FN_SIZE(poly_decompress_d11_avx2_asm)
 
 #endif /* MLK_ARITH_BACKEND_X86_64_DEFAULT && !MLK_CONFIG_MULTILEVEL_NO_SHARED \
-          && (MLK_CONFIG_MULTILEVEL_WITH_SHARED || MLKEM_K == 4) */
+          && !MLK_CONFIG_NO_DECAPS_API && (MLK_CONFIG_MULTILEVEL_WITH_SHARED || \
+          MLKEM_K == 4) */
 
 #if defined(__ELF__)
 .section .note.GNU-stack,"",%progbits

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{poly_decompress_d4_avx2_asm.S → mlkem_poly_decompress_d4_avx2_asm.S}

@@ -31,11 +31,40 @@
 #include "../../../common.h"
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED) && \
+    !defined(MLK_CONFIG_NO_DECAPS_API) && \
     (defined(MLK_CONFIG_MULTILEVEL_WITH_SHARED) || MLKEM_K == 2 || MLKEM_K == 3)
+/*yaml
+  Name: poly_decompress_d4_avx2_asm
+  Description: x86_64 AVX2 polynomial decompression (d=4)
+  Signature: void mlk_poly_decompress_d4_avx2_asm(int16_t *r, const uint8_t *a, const uint8_t *data)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: write-only
+      c_parameter: int16_t *r
+      description: Output polynomial (256 x int16_t)
+    rsi:
+      type: buffer
+      size_bytes: 128
+      permissions: read-only
+      c_parameter: const uint8_t *a
+      description: Input compressed polynomial
+    rdx:
+      type: buffer
+      size_bytes: 32
+      permissions: read-only
+      c_parameter: const uint8_t *data
+      description: Precomputed decompression constants
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/poly_decompress_d4_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_poly_decompress_d4_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -172,8 +201,8 @@ MLK_ASM_FN_SYMBOL(poly_decompress_d4_avx2_asm)
 MLK_ASM_FN_SIZE(poly_decompress_d4_avx2_asm)
 
 #endif /* MLK_ARITH_BACKEND_X86_64_DEFAULT && !MLK_CONFIG_MULTILEVEL_NO_SHARED \
-          && (MLK_CONFIG_MULTILEVEL_WITH_SHARED || MLKEM_K == 2 || MLKEM_K == \
-          3) */
+          && !MLK_CONFIG_NO_DECAPS_API && (MLK_CONFIG_MULTILEVEL_WITH_SHARED || \
+          MLKEM_K == 2 || MLKEM_K == 3) */
 
 #if defined(__ELF__)
 .section .note.GNU-stack,"",%progbits

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{poly_decompress_d5_avx2_asm.S → mlkem_poly_decompress_d5_avx2_asm.S}

@@ -32,11 +32,40 @@
 #include "../../../common.h"
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED) && \
+    !defined(MLK_CONFIG_NO_DECAPS_API) && \
     (defined(MLK_CONFIG_MULTILEVEL_WITH_SHARED) || MLKEM_K == 4)
+/*yaml
+  Name: poly_decompress_d5_avx2_asm
+  Description: x86_64 AVX2 polynomial decompression (d=5)
+  Signature: void mlk_poly_decompress_d5_avx2_asm(int16_t *r, const uint8_t *a, const uint8_t *data)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: write-only
+      c_parameter: int16_t *r
+      description: Output polynomial (256 x int16_t)
+    rsi:
+      type: buffer
+      size_bytes: 160
+      permissions: read-only
+      c_parameter: const uint8_t *a
+      description: Input compressed polynomial
+    rdx:
+      type: buffer
+      size_bytes: 96
+      permissions: read-only
+      c_parameter: const uint8_t *data
+      description: Precomputed decompression constants
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/poly_decompress_d5_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_poly_decompress_d5_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -185,7 +214,8 @@ MLK_ASM_FN_SYMBOL(poly_decompress_d5_avx2_asm)
 MLK_ASM_FN_SIZE(poly_decompress_d5_avx2_asm)
 
 #endif /* MLK_ARITH_BACKEND_X86_64_DEFAULT && !MLK_CONFIG_MULTILEVEL_NO_SHARED \
-          && (MLK_CONFIG_MULTILEVEL_WITH_SHARED || MLKEM_K == 4) */
+          && !MLK_CONFIG_NO_DECAPS_API && (MLK_CONFIG_MULTILEVEL_WITH_SHARED || \
+          MLKEM_K == 4) */
 
 #if defined(__ELF__)
 .section .note.GNU-stack,"",%progbits

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{poly_mulcache_compute_avx2_asm.S → mlkem_poly_mulcache_compute_avx2_asm.S}

@@ -6,10 +6,38 @@
 #include "../../../common.h"
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED)
+/*yaml
+  Name: poly_mulcache_compute_avx2_asm
+  Description: x86_64 AVX2 mulcache computation
+  Signature: void mlk_poly_mulcache_compute_avx2_asm(int16_t *out, const int16_t *in, const int16_t *qdata)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 256
+      permissions: write-only
+      c_parameter: int16_t *out
+      description: Output mulcache (128 x int16_t)
+    rsi:
+      type: buffer
+      size_bytes: 512
+      permissions: read-only
+      c_parameter: const int16_t *in
+      description: Input polynomial (256 x int16_t)
+    rdx:
+      type: buffer
+      size_bytes: 1248
+      permissions: read-only
+      c_parameter: const int16_t *qdata
+      description: Precomputed constants (624 x int16_t)
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/poly_mulcache_compute_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_poly_mulcache_compute_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{polyvec_basemul_acc_montgomery_cached_k2_avx2_asm.S → mlkem_polyvec_basemul_acc_montgomery_cached_k2_avx2_asm.S}

@@ -7,10 +7,44 @@
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED) && \
     (defined(MLK_CONFIG_MULTILEVEL_WITH_SHARED) || MLKEM_K == 2)
+/*yaml
+  Name: polyvec_basemul_acc_montgomery_cached_k2_avx2_asm
+  Description: x86_64 AVX2 base multiplication with accumulation (k=2)
+  Signature: void mlk_polyvec_basemul_acc_montgomery_cached_k2_avx2_asm(int16_t *r, const int16_t *a, const int16_t *b, const int16_t *b_cache)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: read/write
+      c_parameter: int16_t *r
+      description: Output polynomial (256 x int16_t)
+    rsi:
+      type: buffer
+      size_bytes: 1024
+      permissions: read-only
+      c_parameter: const int16_t *a
+      description: Input polyvec a (2 x 256 x int16_t)
+    rdx:
+      type: buffer
+      size_bytes: 1024
+      permissions: read-only
+      c_parameter: const int16_t *b
+      description: Input polyvec b (2 x 256 x int16_t)
+    rcx:
+      type: buffer
+      size_bytes: 512
+      permissions: read-only
+      c_parameter: const int16_t *b_cache
+      description: Mulcache for b (2 x 128 x int16_t)
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/polyvec_basemul_acc_montgomery_cached_k2_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_polyvec_basemul_acc_montgomery_cached_k2_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{polyvec_basemul_acc_montgomery_cached_k3_avx2_asm.S → mlkem_polyvec_basemul_acc_montgomery_cached_k3_avx2_asm.S}

@@ -7,10 +7,44 @@
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED) && \
     (defined(MLK_CONFIG_MULTILEVEL_WITH_SHARED) || MLKEM_K == 3)
+/*yaml
+  Name: polyvec_basemul_acc_montgomery_cached_k3_avx2_asm
+  Description: x86_64 AVX2 base multiplication with accumulation (k=3)
+  Signature: void mlk_polyvec_basemul_acc_montgomery_cached_k3_avx2_asm(int16_t *r, const int16_t *a, const int16_t *b, const int16_t *b_cache)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: read/write
+      c_parameter: int16_t *r
+      description: Output polynomial (256 x int16_t)
+    rsi:
+      type: buffer
+      size_bytes: 1536
+      permissions: read-only
+      c_parameter: const int16_t *a
+      description: Input polyvec a (3 x 256 x int16_t)
+    rdx:
+      type: buffer
+      size_bytes: 1536
+      permissions: read-only
+      c_parameter: const int16_t *b
+      description: Input polyvec b (3 x 256 x int16_t)
+    rcx:
+      type: buffer
+      size_bytes: 768
+      permissions: read-only
+      c_parameter: const int16_t *b_cache
+      description: Mulcache for b (3 x 128 x int16_t)
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/polyvec_basemul_acc_montgomery_cached_k3_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_polyvec_basemul_acc_montgomery_cached_k3_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{polyvec_basemul_acc_montgomery_cached_k4_avx2_asm.S → mlkem_polyvec_basemul_acc_montgomery_cached_k4_avx2_asm.S}

@@ -7,10 +7,44 @@
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED) && \
     (defined(MLK_CONFIG_MULTILEVEL_WITH_SHARED) || MLKEM_K == 4)
+/*yaml
+  Name: polyvec_basemul_acc_montgomery_cached_k4_avx2_asm
+  Description: x86_64 AVX2 base multiplication with accumulation (k=4)
+  Signature: void mlk_polyvec_basemul_acc_montgomery_cached_k4_avx2_asm(int16_t *r, const int16_t *a, const int16_t *b, const int16_t *b_cache)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: read/write
+      c_parameter: int16_t *r
+      description: Output polynomial (256 x int16_t)
+    rsi:
+      type: buffer
+      size_bytes: 2048
+      permissions: read-only
+      c_parameter: const int16_t *a
+      description: Input polyvec a (4 x 256 x int16_t)
+    rdx:
+      type: buffer
+      size_bytes: 2048
+      permissions: read-only
+      c_parameter: const int16_t *b
+      description: Input polyvec b (4 x 256 x int16_t)
+    rcx:
+      type: buffer
+      size_bytes: 1024
+      permissions: read-only
+      c_parameter: const int16_t *b_cache
+      description: Mulcache for b (4 x 128 x int16_t)
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/polyvec_basemul_acc_montgomery_cached_k4_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_polyvec_basemul_acc_montgomery_cached_k4_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{reduce_avx2_asm.S → mlkem_reduce_avx2_asm.S}

@@ -27,10 +27,26 @@
 
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED)
+/*yaml
+  Name: reduce_avx2_asm
+  Description: x86_64 AVX2 modular reduction
+  Signature: void mlk_reduce_avx2_asm(int16_t *r)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: read/write
+      c_parameter: int16_t *r
+      description: Input/output polynomial (256 x int16_t)
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/reduce_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_reduce_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{rej_uniform_avx2_asm.S → mlkem_rej_uniform_avx2_asm.S}

@@ -22,10 +22,43 @@
 
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED)
+/*yaml
+  Name: rej_uniform_avx2_asm
+  Description: x86_64 AVX2 rejection sampling
+  Signature: uint64_t mlk_rej_uniform_avx2_asm(int16_t *r, const uint8_t *buf, unsigned buflen, const uint8_t *table)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: write-only
+      c_parameter: int16_t *r
+      description: Output buffer (256 x int16_t)
+    rsi:
+      type: buffer
+      size_bytes: rdx
+      permissions: read-only
+      c_parameter: const uint8_t *buf
+      description: Input buffer
+    rdx:
+      type: scalar
+      c_parameter: unsigned buflen
+      description: Length of input buffer (must be multiple of 12)
+      test_with: 504  # MLKEM_GEN_MATRIX_NBLOCKS * MLK_XOF_RATE
+    rcx:
+      type: buffer
+      size_bytes: 4096
+      permissions: read-only
+      c_parameter: const uint8_t *table
+      description: Lookup table
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/rej_uniform_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_rej_uniform_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -38,7 +71,7 @@ MLK_ASM_FN_SYMBOL(rej_uniform_avx2_asm)
         .cfi_adjust_cfa_offset 0x210
         xorl %eax, %eax
         testq %rdx, %rdx
-        je Lrej_uniform_asm_end
+        je Lmlk_rej_uniform_asm_end
         movabsq $0xd010d010d010d01, %rax # imm = 0xD010D010D010D01
         movq %rax, %xmm0
         pinsrq $0x1, %rax, %xmm0
@@ -53,7 +86,7 @@ MLK_ASM_FN_SYMBOL(rej_uniform_avx2_asm)
         movq $0x0, %r8
         movq $0x5555, %r9            # imm = 0x5555
 
-Lrej_uniform_asm_loop_start:
+Lmlk_rej_uniform_asm_loop_start:
         movq (%rsi,%r8), %xmm2
         pinsrd $0x2, 0x8(%rsi,%r8), %xmm2
         pshufb %xmm4, %xmm2
@@ -73,12 +106,12 @@ Lrej_uniform_asm_loop_start:
         popcntq %r11, %r11
         addq %r11, %rax
         cmpq $0x100, %rax            # imm = 0x100
-        jae Lrej_uniform_asm_final_copy
+        jae Lmlk_rej_uniform_asm_final_copy
         addq $0xc, %r8
         cmpq %r8, %rdx
-        ja Lrej_uniform_asm_loop_start
+        ja Lmlk_rej_uniform_asm_loop_start
 
-Lrej_uniform_asm_final_copy:
+Lmlk_rej_uniform_asm_final_copy:
         movq $0x100, %rcx            # imm = 0x100
         cmpq $0x100, %rax            # imm = 0x100
         cmovaq %rcx, %rax
@@ -87,7 +120,7 @@ Lrej_uniform_asm_final_copy:
         shlq %rcx
         rep  movsb (%rsi), %es:(%rdi)
 
-Lrej_uniform_asm_end:
+Lmlk_rej_uniform_asm_end:
         addq $0x210, %rsp            # imm = 0x210
         .cfi_adjust_cfa_offset -0x210
         retq

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/native/x86_64/src/{tomont_avx2_asm.S → mlkem_tomont_avx2_asm.S}

@@ -24,11 +24,28 @@
 
 #include "../../../common.h"
 #if defined(MLK_ARITH_BACKEND_X86_64_DEFAULT) && \
-    !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED)
+    !defined(MLK_CONFIG_MULTILEVEL_NO_SHARED) && \
+    !defined(MLK_CONFIG_NO_KEYPAIR_API)
+/*yaml
+  Name: tomont_avx2_asm
+  Description: x86_64 AVX2 Montgomery conversion
+  Signature: void mlk_tomont_avx2_asm(int16_t *r)
+  ABI:
+    Architecture: x86_64
+    CallingConvention: SysV
+    Features: [AVX2]
+    rdi:
+      type: buffer
+      size_bytes: 512
+      permissions: read/write
+      c_parameter: int16_t *r
+      description: Input/output polynomial (256 x int16_t)
+*/
+
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
- *   dev/x86_64/src/tomont_avx2_asm.S using scripts/simpasm. Do not modify it directly.
+ *   dev/x86_64/src/mlkem_tomont_avx2_asm.S using scripts/simpasm. Do not modify it directly.
  */
 
 .text
@@ -148,7 +165,7 @@ MLK_ASM_FN_SYMBOL(tomont_avx2_asm)
 MLK_ASM_FN_SIZE(tomont_avx2_asm)
 
 #endif /* MLK_ARITH_BACKEND_X86_64_DEFAULT && !MLK_CONFIG_MULTILEVEL_NO_SHARED \
-        */
+          && !MLK_CONFIG_NO_KEYPAIR_API */
 
 #if defined(__ELF__)
 .section .note.GNU-stack,"",%progbits

data/ext/pqcrypto/vendor/mlkem-native/mlkem/src/poly.c

@@ -105,6 +105,7 @@ __contract__(
   return res;
 }
 
+#if !defined(MLK_CONFIG_NO_KEYPAIR_API)
 /* Reference: `poly_tomont()` in the reference implementation @[REF]. */
 MLK_STATIC_TESTABLE void mlk_poly_tomont_c(mlk_poly *r)
 __contract__(
@@ -142,6 +143,7 @@ void mlk_poly_tomont(mlk_poly *r)
 
   mlk_poly_tomont_c(r);
 }
+#endif /* !MLK_CONFIG_NO_KEYPAIR_API */
 
 /**
  * Constant-time conversion of signed representatives modulo MLKEM_Q within
@@ -241,6 +243,7 @@ void mlk_poly_add(mlk_poly *r, const mlk_poly *b)
   }
 }
 
+#if !defined(MLK_CONFIG_NO_DECAPS_API)
 /* Reference: `poly_sub()` in the reference implementation @[REF].
  *            - We use destructive version (output=first input) to avoid
  *              reasoning about aliasing in the CBMC specification */
@@ -259,6 +262,7 @@ void mlk_poly_sub(mlk_poly *r, const mlk_poly *b)
     r->coeffs[i] = (int16_t)(r->coeffs[i] - b->coeffs[i]);
   }
 }
+#endif /* !MLK_CONFIG_NO_DECAPS_API */
 
 #include "zetas.inc"
 
@@ -470,6 +474,7 @@ void mlk_poly_ntt(mlk_poly *r)
 }
 
 
+#if !defined(MLK_CONFIG_NO_ENCAPS_API) || !defined(MLK_CONFIG_NO_DECAPS_API)
 /* Compute one layer of inverse NTT */
 
 /* Reference: Embedded into `invntt()` in the reference implementation @[REF] */
@@ -568,8 +573,8 @@ void mlk_poly_invntt_tomont(mlk_poly *r)
 
   mlk_poly_invntt_tomont_c(r);
 }
-
-#else /* !MLK_CONFIG_MULTILEVEL_NO_SHARED */
+#endif /* !MLK_CONFIG_NO_ENCAPS_API || !MLK_CONFIG_NO_DECAPS_API */
+#else  /* !MLK_CONFIG_MULTILEVEL_NO_SHARED */
 
 MLK_EMPTY_CU(mlk_poly)