pq_crypto 0.3.2 → 0.5.0

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/pointwise_montgomery.S

@@ -0,0 +1,79 @@
+/* Copyright (c) The mldsa-native project authors
+ * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
+ */
+
+#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/pointwise_montgomery.S using scripts/simpasm. Do not modify it directly.
+ */
+
+#if defined(__ELF__)
+.section .note.GNU-stack,"",@progbits
+#endif
+
+.text
+.balign 4
+.global MLD_ASM_NAMESPACE(poly_pointwise_montgomery_asm)
+MLD_ASM_FN_SYMBOL(poly_pointwise_montgomery_asm)
+
+        .cfi_startproc
+        mov	w3, #0xe001             // =57345
+        movk	w3, #0x7f, lsl #16
+        dup	v0.4s, w3
+        mov	w3, #0x2001             // =8193
+        movk	w3, #0x380, lsl #16
+        dup	v1.4s, w3
+        mov	x3, #0x40               // =64
+
+Lpoly_pointwise_montgomery_loop_start:
+        ldr	q17, [x1, #0x10]
+        ldr	q18, [x1, #0x20]
+        ldr	q19, [x1, #0x30]
+        ldr	q16, [x1], #0x40
+        ldr	q21, [x2, #0x10]
+        ldr	q22, [x2, #0x20]
+        ldr	q23, [x2, #0x30]
+        ldr	q20, [x2], #0x40
+        smull	v24.2d, v16.2s, v20.2s
+        smull2	v25.2d, v16.4s, v20.4s
+        smull	v26.2d, v17.2s, v21.2s
+        smull2	v27.2d, v17.4s, v21.4s
+        smull	v28.2d, v18.2s, v22.2s
+        smull2	v29.2d, v18.4s, v22.4s
+        smull	v30.2d, v19.2s, v23.2s
+        smull2	v31.2d, v19.4s, v23.4s
+        uzp1	v16.4s, v24.4s, v25.4s
+        mul	v16.4s, v16.4s, v1.4s
+        smlsl	v24.2d, v16.2s, v0.2s
+        smlsl2	v25.2d, v16.4s, v0.4s
+        uzp2	v16.4s, v24.4s, v25.4s
+        uzp1	v17.4s, v26.4s, v27.4s
+        mul	v17.4s, v17.4s, v1.4s
+        smlsl	v26.2d, v17.2s, v0.2s
+        smlsl2	v27.2d, v17.4s, v0.4s
+        uzp2	v17.4s, v26.4s, v27.4s
+        uzp1	v18.4s, v28.4s, v29.4s
+        mul	v18.4s, v18.4s, v1.4s
+        smlsl	v28.2d, v18.2s, v0.2s
+        smlsl2	v29.2d, v18.4s, v0.4s
+        uzp2	v18.4s, v28.4s, v29.4s
+        uzp1	v19.4s, v30.4s, v31.4s
+        mul	v19.4s, v19.4s, v1.4s
+        smlsl	v30.2d, v19.2s, v0.2s
+        smlsl2	v31.2d, v19.4s, v0.4s
+        uzp2	v19.4s, v30.4s, v31.4s
+        str	q17, [x0, #0x10]
+        str	q18, [x0, #0x20]
+        str	q19, [x0, #0x30]
+        str	q16, [x0], #0x40
+        subs	x3, x3, #0x4
+        cbnz	x3, Lpoly_pointwise_montgomery_loop_start
+        ret
+        .cfi_endproc
+
+MLD_ASM_FN_SIZE(poly_pointwise_montgomery_asm)
+
+#endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/poly_caddq_asm.S

@@ -0,0 +1,53 @@
+/*
+ * Copyright (c) The mldsa-native project authors
+ * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
+ */
+#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_asm.S using scripts/simpasm. Do not modify it directly.
+ */
+
+#if defined(__ELF__)
+.section .note.GNU-stack,"",@progbits
+#endif
+
+.text
+.balign 4
+.global MLD_ASM_NAMESPACE(poly_caddq_asm)
+MLD_ASM_FN_SYMBOL(poly_caddq_asm)
+
+        .cfi_startproc
+        mov	w9, #0xe001             // =57345
+        movk	w9, #0x7f, lsl #16
+        dup	v4.4s, w9
+        mov	x1, #0x10               // =16
+
+Lpoly_caddq_loop:
+        ldr	q0, [x0]
+        ldr	q1, [x0, #0x10]
+        ldr	q2, [x0, #0x20]
+        ldr	q3, [x0, #0x30]
+        ushr	v5.4s, v0.4s, #0x1f
+        mla	v0.4s, v5.4s, v4.4s
+        ushr	v5.4s, v1.4s, #0x1f
+        mla	v1.4s, v5.4s, v4.4s
+        ushr	v5.4s, v2.4s, #0x1f
+        mla	v2.4s, v5.4s, v4.4s
+        ushr	v5.4s, v3.4s, #0x1f
+        mla	v3.4s, v5.4s, v4.4s
+        str	q1, [x0, #0x10]
+        str	q2, [x0, #0x20]
+        str	q3, [x0, #0x30]
+        str	q0, [x0], #0x40
+        subs	x1, x1, #0x1
+        b.ne	Lpoly_caddq_loop
+        ret
+        .cfi_endproc
+
+MLD_ASM_FN_SIZE(poly_caddq_asm)
+
+#endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/poly_chknorm_asm.S

@@ -0,0 +1,55 @@
+/*
+ * Copyright (c) The mldsa-native project authors
+ * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
+ */
+#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_asm.S using scripts/simpasm. Do not modify it directly.
+ */
+
+#if defined(__ELF__)
+.section .note.GNU-stack,"",@progbits
+#endif
+
+.text
+.balign 4
+.global MLD_ASM_NAMESPACE(poly_chknorm_asm)
+MLD_ASM_FN_SYMBOL(poly_chknorm_asm)
+
+        .cfi_startproc
+        dup	v20.4s, w1
+        eor	v21.16b, v21.16b, v21.16b
+        mov	x2, #0x10               // =16
+
+Lpoly_chknorm_loop:
+        ldr	q1, [x0, #0x10]
+        ldr	q2, [x0, #0x20]
+        ldr	q3, [x0, #0x30]
+        ldr	q0, [x0], #0x40
+        abs	v1.4s, v1.4s
+        cmge	v1.4s, v1.4s, v20.4s
+        orr	v21.16b, v21.16b, v1.16b
+        abs	v2.4s, v2.4s
+        cmge	v2.4s, v2.4s, v20.4s
+        orr	v21.16b, v21.16b, v2.16b
+        abs	v3.4s, v3.4s
+        cmge	v3.4s, v3.4s, v20.4s
+        orr	v21.16b, v21.16b, v3.16b
+        abs	v0.4s, v0.4s
+        cmge	v0.4s, v0.4s, v20.4s
+        orr	v21.16b, v21.16b, v0.16b
+        subs	x2, x2, #0x1
+        b.ne	Lpoly_chknorm_loop
+        umaxv	s21, v21.4s
+        fmov	w0, s21
+        and	w0, w0, #0x1
+        ret
+        .cfi_endproc
+
+MLD_ASM_FN_SIZE(poly_chknorm_asm)
+
+#endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/poly_decompose_32_asm.S

@@ -0,0 +1,85 @@
+/*
+ * Copyright (c) The mldsa-native project authors
+ * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
+ */
+#include "../../../common.h"
+
+#if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
+           (defined(MLD_CONFIG_MULTILEVEL_WITH_SHARED) || (MLD_CONFIG_PARAMETER_SET == 65 || MLD_CONFIG_PARAMETER_SET == 87))
+
+/*
+ * WARNING: This file is auto-derived from the mldsa-native source file
+ *   dev/aarch64_opt/src/poly_decompose_32_asm.S using scripts/simpasm. Do not modify it directly.
+ */
+
+#if defined(__ELF__)
+.section .note.GNU-stack,"",@progbits
+#endif
+
+.text
+.balign 4
+.global MLD_ASM_NAMESPACE(poly_decompose_32_asm)
+MLD_ASM_FN_SYMBOL(poly_decompose_32_asm)
+
+        .cfi_startproc
+        mov	w4, #0xe001             // =57345
+        movk	w4, #0x7f, lsl #16
+        dup	v20.4s, w4
+        mov	w5, #0xe100             // =57600
+        movk	w5, #0x7b, lsl #16
+        dup	v21.4s, w5
+        mov	w7, #0xfe00             // =65024
+        movk	w7, #0x7, lsl #16
+        dup	v22.4s, w7
+        mov	w11, #0x401             // =1025
+        movk	w11, #0x4010, lsl #16
+        dup	v23.4s, w11
+        mov	x3, #0x10               // =16
+
+Lpoly_decompose_32_loop:
+        ldr	q0, [x1]
+        ldr	q1, [x1, #0x10]
+        ldr	q2, [x1, #0x20]
+        ldr	q3, [x1, #0x30]
+        sqdmulh	v5.4s, v1.4s, v23.4s
+        srshr	v5.4s, v5.4s, #0x12
+        cmgt	v24.4s, v1.4s, v21.4s
+        mls	v1.4s, v5.4s, v22.4s
+        bic	v5.16b, v5.16b, v24.16b
+        add	v1.4s, v1.4s, v24.4s
+        sqdmulh	v6.4s, v2.4s, v23.4s
+        srshr	v6.4s, v6.4s, #0x12
+        cmgt	v24.4s, v2.4s, v21.4s
+        mls	v2.4s, v6.4s, v22.4s
+        bic	v6.16b, v6.16b, v24.16b
+        add	v2.4s, v2.4s, v24.4s
+        sqdmulh	v7.4s, v3.4s, v23.4s
+        srshr	v7.4s, v7.4s, #0x12
+        cmgt	v24.4s, v3.4s, v21.4s
+        mls	v3.4s, v7.4s, v22.4s
+        bic	v7.16b, v7.16b, v24.16b
+        add	v3.4s, v3.4s, v24.4s
+        sqdmulh	v4.4s, v0.4s, v23.4s
+        srshr	v4.4s, v4.4s, #0x12
+        cmgt	v24.4s, v0.4s, v21.4s
+        mls	v0.4s, v4.4s, v22.4s
+        bic	v4.16b, v4.16b, v24.16b
+        add	v0.4s, v0.4s, v24.4s
+        str	q5, [x0, #0x10]
+        str	q6, [x0, #0x20]
+        str	q7, [x0, #0x30]
+        str	q4, [x0], #0x40
+        str	q1, [x1, #0x10]
+        str	q2, [x1, #0x20]
+        str	q3, [x1, #0x30]
+        str	q0, [x1], #0x40
+        subs	x3, x3, #0x1
+        b.ne	Lpoly_decompose_32_loop
+        ret
+        .cfi_endproc
+
+MLD_ASM_FN_SIZE(poly_decompose_32_asm)
+
+#endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED && \
+          (MLD_CONFIG_MULTILEVEL_WITH_SHARED || MLD_CONFIG_PARAMETER_SET == 65 \
+          || MLD_CONFIG_PARAMETER_SET == 87) */

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/poly_decompose_88_asm.S

@@ -0,0 +1,85 @@
+/*
+ * Copyright (c) The mldsa-native project authors
+ * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
+ */
+#include "../../../common.h"
+
+#if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
+           (defined(MLD_CONFIG_MULTILEVEL_WITH_SHARED) || MLD_CONFIG_PARAMETER_SET == 44)
+
+/*
+ * WARNING: This file is auto-derived from the mldsa-native source file
+ *   dev/aarch64_opt/src/poly_decompose_88_asm.S using scripts/simpasm. Do not modify it directly.
+ */
+
+#if defined(__ELF__)
+.section .note.GNU-stack,"",@progbits
+#endif
+
+.text
+.balign 4
+.global MLD_ASM_NAMESPACE(poly_decompose_88_asm)
+MLD_ASM_FN_SYMBOL(poly_decompose_88_asm)
+
+        .cfi_startproc
+        mov	w4, #0xe001             // =57345
+        movk	w4, #0x7f, lsl #16
+        dup	v20.4s, w4
+        mov	w5, #0x6c00             // =27648
+        movk	w5, #0x7e, lsl #16
+        dup	v21.4s, w5
+        mov	w7, #0xe800             // =59392
+        movk	w7, #0x2, lsl #16
+        dup	v22.4s, w7
+        mov	w11, #0x581             // =1409
+        movk	w11, #0x5816, lsl #16
+        dup	v23.4s, w11
+        mov	x3, #0x10               // =16
+
+Lpoly_decompose_88_loop:
+        ldr	q0, [x1]
+        ldr	q1, [x1, #0x10]
+        ldr	q2, [x1, #0x20]
+        ldr	q3, [x1, #0x30]
+        sqdmulh	v5.4s, v1.4s, v23.4s
+        srshr	v5.4s, v5.4s, #0x11
+        cmgt	v24.4s, v1.4s, v21.4s
+        mls	v1.4s, v5.4s, v22.4s
+        bic	v5.16b, v5.16b, v24.16b
+        add	v1.4s, v1.4s, v24.4s
+        sqdmulh	v6.4s, v2.4s, v23.4s
+        srshr	v6.4s, v6.4s, #0x11
+        cmgt	v24.4s, v2.4s, v21.4s
+        mls	v2.4s, v6.4s, v22.4s
+        bic	v6.16b, v6.16b, v24.16b
+        add	v2.4s, v2.4s, v24.4s
+        sqdmulh	v7.4s, v3.4s, v23.4s
+        srshr	v7.4s, v7.4s, #0x11
+        cmgt	v24.4s, v3.4s, v21.4s
+        mls	v3.4s, v7.4s, v22.4s
+        bic	v7.16b, v7.16b, v24.16b
+        add	v3.4s, v3.4s, v24.4s
+        sqdmulh	v4.4s, v0.4s, v23.4s
+        srshr	v4.4s, v4.4s, #0x11
+        cmgt	v24.4s, v0.4s, v21.4s
+        mls	v0.4s, v4.4s, v22.4s
+        bic	v4.16b, v4.16b, v24.16b
+        add	v0.4s, v0.4s, v24.4s
+        str	q5, [x0, #0x10]
+        str	q6, [x0, #0x20]
+        str	q7, [x0, #0x30]
+        str	q4, [x0], #0x40
+        str	q1, [x1, #0x10]
+        str	q2, [x1, #0x20]
+        str	q3, [x1, #0x30]
+        str	q0, [x1], #0x40
+        subs	x3, x3, #0x1
+        b.ne	Lpoly_decompose_88_loop
+        ret
+        .cfi_endproc
+
+MLD_ASM_FN_SIZE(poly_decompose_88_asm)
+
+#endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED && \
+          (MLD_CONFIG_MULTILEVEL_WITH_SHARED || MLD_CONFIG_PARAMETER_SET == 44) \
+        */

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/poly_use_hint_32_asm.S

@@ -0,0 +1,102 @@
+/*
+ * Copyright (c) The mldsa-native project authors
+ * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
+ */
+#include "../../../common.h"
+
+#if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
+           (defined(MLD_CONFIG_MULTILEVEL_WITH_SHARED) || (MLD_CONFIG_PARAMETER_SET == 65 || MLD_CONFIG_PARAMETER_SET == 87))
+
+/*
+ * WARNING: This file is auto-derived from the mldsa-native source file
+ *   dev/aarch64_opt/src/poly_use_hint_32_asm.S using scripts/simpasm. Do not modify it directly.
+ */
+
+#if defined(__ELF__)
+.section .note.GNU-stack,"",@progbits
+#endif
+
+.text
+.balign 4
+.global MLD_ASM_NAMESPACE(poly_use_hint_32_asm)
+MLD_ASM_FN_SYMBOL(poly_use_hint_32_asm)
+
+        .cfi_startproc
+        mov	w4, #0xe001             // =57345
+        movk	w4, #0x7f, lsl #16
+        dup	v20.4s, w4
+        mov	w5, #0xe100             // =57600
+        movk	w5, #0x7b, lsl #16
+        dup	v21.4s, w5
+        mov	w7, #0xfe00             // =65024
+        movk	w7, #0x7, lsl #16
+        dup	v22.4s, w7
+        mov	w11, #0x401             // =1025
+        movk	w11, #0x4010, lsl #16
+        dup	v23.4s, w11
+        movi	v24.4s, #0xf
+        mov	x3, #0x10               // =16
+
+Lpoly_use_hint_32_loop:
+        ldr	q1, [x1, #0x10]
+        ldr	q2, [x1, #0x20]
+        ldr	q3, [x1, #0x30]
+        ldr	q0, [x1], #0x40
+        ldr	q5, [x2, #0x10]
+        ldr	q6, [x2, #0x20]
+        ldr	q7, [x2, #0x30]
+        ldr	q4, [x2], #0x40
+        sqdmulh	v17.4s, v1.4s, v23.4s
+        srshr	v17.4s, v17.4s, #0x12
+        cmgt	v25.4s, v1.4s, v21.4s
+        mls	v1.4s, v17.4s, v22.4s
+        bic	v17.16b, v17.16b, v25.16b
+        add	v1.4s, v1.4s, v25.4s
+        cmle	v1.4s, v1.4s, #0
+        orr	v1.4s, #0x1
+        mla	v17.4s, v1.4s, v5.4s
+        and	v17.16b, v17.16b, v24.16b
+        sqdmulh	v18.4s, v2.4s, v23.4s
+        srshr	v18.4s, v18.4s, #0x12
+        cmgt	v25.4s, v2.4s, v21.4s
+        mls	v2.4s, v18.4s, v22.4s
+        bic	v18.16b, v18.16b, v25.16b
+        add	v2.4s, v2.4s, v25.4s
+        cmle	v2.4s, v2.4s, #0
+        orr	v2.4s, #0x1
+        mla	v18.4s, v2.4s, v6.4s
+        and	v18.16b, v18.16b, v24.16b
+        sqdmulh	v19.4s, v3.4s, v23.4s
+        srshr	v19.4s, v19.4s, #0x12
+        cmgt	v25.4s, v3.4s, v21.4s
+        mls	v3.4s, v19.4s, v22.4s
+        bic	v19.16b, v19.16b, v25.16b
+        add	v3.4s, v3.4s, v25.4s
+        cmle	v3.4s, v3.4s, #0
+        orr	v3.4s, #0x1
+        mla	v19.4s, v3.4s, v7.4s
+        and	v19.16b, v19.16b, v24.16b
+        sqdmulh	v16.4s, v0.4s, v23.4s
+        srshr	v16.4s, v16.4s, #0x12
+        cmgt	v25.4s, v0.4s, v21.4s
+        mls	v0.4s, v16.4s, v22.4s
+        bic	v16.16b, v16.16b, v25.16b
+        add	v0.4s, v0.4s, v25.4s
+        cmle	v0.4s, v0.4s, #0
+        orr	v0.4s, #0x1
+        mla	v16.4s, v0.4s, v4.4s
+        and	v16.16b, v16.16b, v24.16b
+        str	q17, [x0, #0x10]
+        str	q18, [x0, #0x20]
+        str	q19, [x0, #0x30]
+        str	q16, [x0], #0x40
+        subs	x3, x3, #0x1
+        b.ne	Lpoly_use_hint_32_loop
+        ret
+        .cfi_endproc
+
+MLD_ASM_FN_SIZE(poly_use_hint_32_asm)
+
+#endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED && \
+          (MLD_CONFIG_MULTILEVEL_WITH_SHARED || MLD_CONFIG_PARAMETER_SET == 65 \
+          || MLD_CONFIG_PARAMETER_SET == 87) */

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/poly_use_hint_88_asm.S

@@ -0,0 +1,110 @@
+/*
+ * Copyright (c) The mldsa-native project authors
+ * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
+ */
+#include "../../../common.h"
+
+#if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
+           (defined(MLD_CONFIG_MULTILEVEL_WITH_SHARED) || MLD_CONFIG_PARAMETER_SET == 44)
+
+/*
+ * WARNING: This file is auto-derived from the mldsa-native source file
+ *   dev/aarch64_opt/src/poly_use_hint_88_asm.S using scripts/simpasm. Do not modify it directly.
+ */
+
+#if defined(__ELF__)
+.section .note.GNU-stack,"",@progbits
+#endif
+
+.text
+.balign 4
+.global MLD_ASM_NAMESPACE(poly_use_hint_88_asm)
+MLD_ASM_FN_SYMBOL(poly_use_hint_88_asm)
+
+        .cfi_startproc
+        mov	w4, #0xe001             // =57345
+        movk	w4, #0x7f, lsl #16
+        dup	v20.4s, w4
+        mov	w5, #0x6c00             // =27648
+        movk	w5, #0x7e, lsl #16
+        dup	v21.4s, w5
+        mov	w7, #0xe800             // =59392
+        movk	w7, #0x2, lsl #16
+        dup	v22.4s, w7
+        mov	w11, #0x581             // =1409
+        movk	w11, #0x5816, lsl #16
+        dup	v23.4s, w11
+        movi	v24.4s, #0x2b
+        mov	x3, #0x10               // =16
+
+Lpoly_use_hint_88_loop:
+        ldr	q1, [x1, #0x10]
+        ldr	q2, [x1, #0x20]
+        ldr	q3, [x1, #0x30]
+        ldr	q0, [x1], #0x40
+        ldr	q5, [x2, #0x10]
+        ldr	q6, [x2, #0x20]
+        ldr	q7, [x2, #0x30]
+        ldr	q4, [x2], #0x40
+        sqdmulh	v17.4s, v1.4s, v23.4s
+        srshr	v17.4s, v17.4s, #0x11
+        cmgt	v25.4s, v1.4s, v21.4s
+        mls	v1.4s, v17.4s, v22.4s
+        bic	v17.16b, v17.16b, v25.16b
+        add	v1.4s, v1.4s, v25.4s
+        cmle	v1.4s, v1.4s, #0
+        orr	v1.4s, #0x1
+        mla	v17.4s, v1.4s, v5.4s
+        cmgt	v25.4s, v17.4s, v24.4s
+        bic	v17.16b, v17.16b, v25.16b
+        umin	v17.4s, v17.4s, v24.4s
+        sqdmulh	v18.4s, v2.4s, v23.4s
+        srshr	v18.4s, v18.4s, #0x11
+        cmgt	v25.4s, v2.4s, v21.4s
+        mls	v2.4s, v18.4s, v22.4s
+        bic	v18.16b, v18.16b, v25.16b
+        add	v2.4s, v2.4s, v25.4s
+        cmle	v2.4s, v2.4s, #0
+        orr	v2.4s, #0x1
+        mla	v18.4s, v2.4s, v6.4s
+        cmgt	v25.4s, v18.4s, v24.4s
+        bic	v18.16b, v18.16b, v25.16b
+        umin	v18.4s, v18.4s, v24.4s
+        sqdmulh	v19.4s, v3.4s, v23.4s
+        srshr	v19.4s, v19.4s, #0x11
+        cmgt	v25.4s, v3.4s, v21.4s
+        mls	v3.4s, v19.4s, v22.4s
+        bic	v19.16b, v19.16b, v25.16b
+        add	v3.4s, v3.4s, v25.4s
+        cmle	v3.4s, v3.4s, #0
+        orr	v3.4s, #0x1
+        mla	v19.4s, v3.4s, v7.4s
+        cmgt	v25.4s, v19.4s, v24.4s
+        bic	v19.16b, v19.16b, v25.16b
+        umin	v19.4s, v19.4s, v24.4s
+        sqdmulh	v16.4s, v0.4s, v23.4s
+        srshr	v16.4s, v16.4s, #0x11
+        cmgt	v25.4s, v0.4s, v21.4s
+        mls	v0.4s, v16.4s, v22.4s
+        bic	v16.16b, v16.16b, v25.16b
+        add	v0.4s, v0.4s, v25.4s
+        cmle	v0.4s, v0.4s, #0
+        orr	v0.4s, #0x1
+        mla	v16.4s, v0.4s, v4.4s
+        cmgt	v25.4s, v16.4s, v24.4s
+        bic	v16.16b, v16.16b, v25.16b
+        umin	v16.4s, v16.4s, v24.4s
+        str	q17, [x0, #0x10]
+        str	q18, [x0, #0x20]
+        str	q19, [x0, #0x30]
+        str	q16, [x0], #0x40
+        subs	x3, x3, #0x1
+        b.ne	Lpoly_use_hint_88_loop
+        ret
+        .cfi_endproc
+
+MLD_ASM_FN_SIZE(poly_use_hint_88_asm)
+
+#endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED && \
+          (MLD_CONFIG_MULTILEVEL_WITH_SHARED || MLD_CONFIG_PARAMETER_SET == 44) \
+        */

data/ext/pqcrypto/vendor/mldsa-native/mldsa/src/native/aarch64/src/polyz_unpack_17_asm.S

@@ -0,0 +1,72 @@
+/*
+ * Copyright (c) The mldsa-native project authors
+ * Copyright (c) The mlkem-native project authors
+ * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
+ */
+
+ #include "../../../common.h"
+#if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED) && \
+           (defined(MLD_CONFIG_MULTILEVEL_WITH_SHARED) || MLD_CONFIG_PARAMETER_SET == 44)
+
+/*
+ * WARNING: This file is auto-derived from the mldsa-native source file
+ *   dev/aarch64_opt/src/polyz_unpack_17_asm.S using scripts/simpasm. Do not modify it directly.
+ */
+
+#if defined(__ELF__)
+.section .note.GNU-stack,"",@progbits
+#endif
+
+.text
+.balign 4
+.global MLD_ASM_NAMESPACE(polyz_unpack_17_asm)
+MLD_ASM_FN_SYMBOL(polyz_unpack_17_asm)
+
+        .cfi_startproc
+        ldr	q24, [x2]
+        ldr	q25, [x2, #0x10]
+        ldr	q26, [x2, #0x20]
+        ldr	q27, [x2, #0x30]
+        mov	x3, #0xfe00000000       // =1090921693184
+        mov	v28.d[0], x3
+        mov	x3, #0xfc               // =252
+        movk	x3, #0xfa, lsl #32
+        mov	v28.d[1], x3
+        movi	v29.4s, #0x3, msl #16
+        movi	v30.4s, #0x2, lsl #16
+        mov	x9, #0x10               // =16
+
+Lpolyz_unpack_17_loop:
+        ld1	{ v0.16b, v1.16b }, [x1]
+        add	x1, x1, #0x14
+        ld1	{ v2.16b }, [x1], #16
+        tbl	v4.16b, { v0.16b }, v24.16b
+        tbl	v5.16b, { v0.16b, v1.16b }, v25.16b
+        tbl	v6.16b, { v1.16b }, v26.16b
+        tbl	v7.16b, { v1.16b, v2.16b }, v27.16b
+        ushl	v4.4s, v4.4s, v28.4s
+        and	v4.16b, v4.16b, v29.16b
+        sub	v4.4s, v30.4s, v4.4s
+        ushl	v5.4s, v5.4s, v28.4s
+        and	v5.16b, v5.16b, v29.16b
+        sub	v5.4s, v30.4s, v5.4s
+        ushl	v6.4s, v6.4s, v28.4s
+        and	v6.16b, v6.16b, v29.16b
+        sub	v6.4s, v30.4s, v6.4s
+        ushl	v7.4s, v7.4s, v28.4s
+        and	v7.16b, v7.16b, v29.16b
+        sub	v7.4s, v30.4s, v7.4s
+        str	q5, [x0, #0x10]
+        str	q6, [x0, #0x20]
+        str	q7, [x0, #0x30]
+        str	q4, [x0], #0x40
+        subs	x9, x9, #0x1
+        b.ne	Lpolyz_unpack_17_loop
+        ret
+        .cfi_endproc
+
+MLD_ASM_FN_SIZE(polyz_unpack_17_asm)
+
+#endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED && \
+          (MLD_CONFIG_MULTILEVEL_WITH_SHARED || MLD_CONFIG_PARAMETER_SET == 44) \
+        */