Add simpasm header and footer and change to assembly code in dev/

willieyz · willieyz · commit 2552b274821a · 2025-10-27T14:23:52.000+08:00
- This commit add simpasm header and footer to all assembly code in dev/

- Change balign from 16 to 4 in poly_caddq_asm.S to match our
  standard setting.
- The llvm-objdump discards one of the consecutive labels, which causes
  the autogen script to fail. We removed the end label since it is not
  used anywhere, and also simplified this two in a separate commit.

Signed-off-by: willieyz &lt;willie.zhao@chelpis.com&gt;
diff --git a/dev/aarch64_clean/src/intt.S b/dev/aarch64_clean/src/intt.S
@@ -25,6 +25,7 @@
 
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64)
+/* simpasm: header-end */
 
 .macro mulmodq dst, src, const, idx0, idx1
         sqrdmulh t2.4s,      \src\().4s, \const\().s[\idx1\()]
@@ -367,4 +368,5 @@ layer1234_start:
         pop_stack
         ret
 
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 */
diff --git a/dev/aarch64_clean/src/mld_polyvecl_pointwise_acc_montgomery_l4.S b/dev/aarch64_clean/src/mld_polyvecl_pointwise_acc_montgomery_l4.S
@@ -4,6 +4,7 @@
 
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64)
+/* simpasm: header-end */
 
 .macro montgomery_reduce_long res, inl, inh
     uzp1   \res\().4s, \inl\().4s, \inh\().4s
@@ -109,4 +110,6 @@ l4_loop_start:
     cbnz count, l4_loop_start
 
     ret
+    
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 */
diff --git a/dev/aarch64_clean/src/mld_polyvecl_pointwise_acc_montgomery_l5.S b/dev/aarch64_clean/src/mld_polyvecl_pointwise_acc_montgomery_l5.S
@@ -4,6 +4,7 @@
 
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64)
+/* simpasm: header-end */
 
 .macro montgomery_reduce_long res, inl, inh
     uzp1   \res\().4s, \inl\().4s, \inh\().4s
@@ -117,4 +118,5 @@ l5_loop_start:
     cbnz count, l5_loop_start
 
     ret
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 */
diff --git a/dev/aarch64_clean/src/mld_polyvecl_pointwise_acc_montgomery_l7.S b/dev/aarch64_clean/src/mld_polyvecl_pointwise_acc_montgomery_l7.S
@@ -4,6 +4,7 @@
 
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64)
+/* simpasm: header-end */
 
 .macro montgomery_reduce_long res, inl, inh
     uzp1   \res\().4s, \inl\().4s, \inh\().4s
@@ -133,4 +134,5 @@ l7_loop_start:
     cbnz count, l7_loop_start
 
     ret
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 */
diff --git a/dev/aarch64_clean/src/ntt.S b/dev/aarch64_clean/src/ntt.S
@@ -25,6 +25,7 @@
 
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64)
+/* simpasm: header-end */
 
 .macro mulmodq dst, src, const, idx0, idx1
         sqrdmulh t2.4s,      \src\().4s, \const\().s[\idx1\()]
@@ -302,4 +303,5 @@ layer45678_start:
        pop_stack
        ret
 
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 */
diff --git a/dev/aarch64_clean/src/pointwise_montgomery.S b/dev/aarch64_clean/src/pointwise_montgomery.S
@@ -4,6 +4,7 @@
 
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64)
+/* simpasm: header-end */
 
 .macro montgomery_reduce_long res, inl, inh
     uzp1   \res\().4s, \inl\().4s, \inh\().4s
@@ -73,4 +74,5 @@ loop_start:
     cbnz count, loop_start
 
     ret
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 */
diff --git a/dev/aarch64_clean/src/poly_caddq_asm.S b/dev/aarch64_clean/src/poly_caddq_asm.S
@@ -5,14 +5,15 @@
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 .macro caddq inout
         ushr tmp.4s, \inout\().4s, #31
         mla \inout\().4s, tmp.4s, q_reg.4s
 .endm
 
 .global MLD_ASM_NAMESPACE(poly_caddq_asm)
-.balign 16
+.balign 4
 MLD_ASM_FN_SYMBOL(poly_caddq_asm)
         // Function signature: void mld_poly_caddq_asm(int32_t *a)
         // x0: pointer to polynomial coefficients
@@ -55,5 +56,5 @@ poly_caddq_loop:
         .unreq count
         .unreq q_reg
         .unreq tmp
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/poly_chknorm_asm.S b/dev/aarch64_clean/src/poly_chknorm_asm.S
@@ -5,6 +5,7 @@
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 .macro chknorm a
         abs \a\().4s, \a\().4s
@@ -58,5 +59,5 @@ poly_chknorm_loop:
         .unreq count
         .unreq bound
         .unreq flags
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/poly_decompose_32_asm.S b/dev/aarch64_clean/src/poly_decompose_32_asm.S
@@ -5,6 +5,7 @@
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 // a aliased with a0
 .macro decompose32 a1, a, temp
@@ -100,5 +101,5 @@ poly_decompose_32_loop:
         .unreq q_bound
         .unreq gamma2_2x
         .unreq barrett_const
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/poly_decompose_88_asm.S b/dev/aarch64_clean/src/poly_decompose_88_asm.S
@@ -5,6 +5,7 @@
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 // a aliased with a0
 .macro decompose88 a1, a, temp
@@ -98,5 +99,5 @@ poly_decompose_88_loop:
         .unreq q_bound
         .unreq gamma2_2x
         .unreq barrett_const
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/poly_use_hint_32_asm.S b/dev/aarch64_clean/src/poly_use_hint_32_asm.S
@@ -5,6 +5,7 @@
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 // a aliased with a0
 .macro decompose32 a1, a, temp
@@ -117,5 +118,5 @@ poly_use_hint_32_loop:
         .unreq gamma2_2x
         .unreq barrett_const
         .unreq mask_15
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/poly_use_hint_88_asm.S b/dev/aarch64_clean/src/poly_use_hint_88_asm.S
@@ -5,6 +5,7 @@
 #include "../../../common.h"
 
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 // a aliased with a0
 .macro decompose88 a1, a, temp
@@ -119,5 +120,5 @@ poly_use_hint_88_loop:
         .unreq gamma2_2x
         .unreq barrett_const
         .unreq const_43
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/polyz_unpack_17_asm.S b/dev/aarch64_clean/src/polyz_unpack_17_asm.S
@@ -6,6 +6,7 @@
 
  #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 .macro trim_map_17 a
         // Keep only 18 out of 24 bits in each 32-bit lane
@@ -99,5 +100,5 @@ polyz_unpack_17_loop:
         .unreq shifts
         .unreq mask
         .unreq gamma1
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/polyz_unpack_19_asm.S b/dev/aarch64_clean/src/polyz_unpack_19_asm.S
@@ -6,6 +6,7 @@
 
  #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) && !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 .macro trim_map_19 a
         // Keep only 20 out of 24 bits in each 32-bit lane
@@ -96,5 +97,5 @@ polyz_unpack_19_loop:
         .unreq shifts
         .unreq mask
         .unreq gamma1
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/rej_uniform_asm.S b/dev/aarch64_clean/src/rej_uniform_asm.S
@@ -7,6 +7,7 @@
  #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) && \
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 // We save the output on the stack first, and copy to the actual
 // output buffer only in the end. This is because the main loop can overwrite
@@ -405,5 +406,5 @@ rej_uniform_return:
 
 /* TODO: autogenerate this in autogen */
 #undef STACK_SIZE
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/rej_uniform_eta2_asm.S b/dev/aarch64_clean/src/rej_uniform_eta2_asm.S
@@ -7,6 +7,7 @@
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) && \
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 // We save the output on the stack first, and copy to the actual
 // output buffer only in the end. This is because the main loop can overwrite
@@ -223,7 +224,7 @@ rej_uniform_eta2_loop8:
 
         cmp buflen, #8
         b.hs rej_uniform_eta2_loop8
-rej_uniform_eta2_loop8_end:
+
 rej_uniform_eta2_memory_copy:
     // min = min(count,len)
     cmp  count, len
@@ -328,5 +329,5 @@ rej_uniform_eta2_final_copy:
     .unreq barrett_tmp
 
 #undef STACK_SIZE
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/aarch64_clean/src/rej_uniform_eta4_asm.S b/dev/aarch64_clean/src/rej_uniform_eta4_asm.S
@@ -7,6 +7,7 @@
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_AARCH64) && \
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 // We save the output on the stack first, and copy to the actual
 // output buffer only in the end. This is because the main loop can overwrite
@@ -220,7 +221,7 @@ rej_uniform_eta4_loop8:
 
         cmp buflen, #8
         b.hs rej_uniform_eta4_loop8
-rej_uniform_eta4_loop8_end:
+
 rej_uniform_eta4_memory_copy:
     // min = min(count,len)
     cmp  count, len
@@ -304,5 +305,5 @@ rej_uniform_eta4_final_copy:
     .unreq bits
 
 #undef STACK_SIZE
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_AARCH64 && !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/fips202/aarch64/src/keccak_f1600_x1_scalar_asm.S b/dev/fips202/aarch64/src/keccak_f1600_x1_scalar_asm.S
@@ -12,6 +12,7 @@
 #include "../../../../common.h"
 #if defined(MLD_FIPS202_AARCH64_NEED_X1_SCALAR) && \
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
@@ -316,6 +317,6 @@ keccak_f1600_x1_scalar_loop:
         ldp	x29, x30, [sp, #0x70]
         add	sp, sp, #0x80
         ret
-
+/* simpasm: footer-start */
 #endif /* MLD_FIPS202_AARCH64_NEED_X1_SCALAR && \
           !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/fips202/aarch64/src/keccak_f1600_x1_v84a_asm.S b/dev/fips202/aarch64/src/keccak_f1600_x1_v84a_asm.S
@@ -29,6 +29,7 @@
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
 
 #if defined(__ARM_FEATURE_SHA3)
+/* simpasm: header-end */
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
@@ -150,7 +151,7 @@ keccak_f1600_x1_v84a_loop:
         ldp	d14, d15, [sp, #0x30]
         add	sp, sp, #0x40
         ret
-
+/* simpasm: footer-start */
 #endif /* __ARM_FEATURE_SHA3 */
 
 #endif /* MLD_FIPS202_AARCH64_NEED_X1_V84A && !MLD_CONFIG_MULTILEVEL_NO_SHARED \
diff --git a/dev/fips202/aarch64/src/keccak_f1600_x2_v84a_asm.S b/dev/fips202/aarch64/src/keccak_f1600_x2_v84a_asm.S
@@ -29,6 +29,7 @@
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
 
 #if defined(__ARM_FEATURE_SHA3)
+/* simpasm: header-end */
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
@@ -204,7 +205,7 @@ keccak_f1600_x2_v84a_loop:
         ldp	d14, d15, [sp, #0x30]
         add	sp, sp, #0x40
         ret
-
+/* simpasm: footer-start */
 #endif /* __ARM_FEATURE_SHA3 */
 
 #endif /* MLD_FIPS202_AARCH64_NEED_X2_V84A && !MLD_CONFIG_MULTILEVEL_NO_SHARED \
diff --git a/dev/fips202/aarch64/src/keccak_f1600_x4_v8a_scalar_hybrid_asm.S b/dev/fips202/aarch64/src/keccak_f1600_x4_v8a_scalar_hybrid_asm.S
@@ -12,6 +12,7 @@
 #include "../../../../common.h"
 #if defined(MLD_FIPS202_AARCH64_NEED_X4_V8A_SCALAR_HYBRID) && \
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
@@ -1000,6 +1001,6 @@ keccak_f1600_x4_scalar_v8a_hybrid_done:
         ldp	x29, x30, [sp, #0x80]
         add	sp, sp, #0xe0
         ret
-
+/* simpasm: footer-start */
 #endif /* MLD_FIPS202_AARCH64_NEED_X4_V8A_SCALAR_HYBRID && \
           !MLD_CONFIG_MULTILEVEL_NO_SHARED */
diff --git a/dev/fips202/aarch64/src/keccak_f1600_x4_v8a_v84a_scalar_hybrid_asm.S b/dev/fips202/aarch64/src/keccak_f1600_x4_v8a_v84a_scalar_hybrid_asm.S
@@ -14,6 +14,7 @@
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
 
 #if defined(__ARM_FEATURE_SHA3)
+/* simpasm: header-end */
 
 /*
  * WARNING: This file is auto-derived from the mlkem-native source file
@@ -908,7 +909,7 @@ keccak_f1600_x4_scalar_v8a_v84a_hybrid_done:
         ldp	x29, x30, [sp, #0x80]
         add	sp, sp, #0xe0
         ret
-
+/* simpasm: footer-start */
 #endif /* __ARM_FEATURE_SHA3 */
 
 #endif /* MLD_FIPS202_AARCH64_NEED_X4_V8A_V84A_SCALAR_HYBRID && \
diff --git a/dev/x86_64/src/intt.S b/dev/x86_64/src/intt.S
@@ -20,6 +20,7 @@
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 #include "consts.h"
 
@@ -279,6 +280,6 @@ levels6t7	2
 levels6t7	3
 
 ret
-
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_X86_64_DEFAULT && !MLD_CONFIG_MULTILEVEL_NO_SHARED \
         */
diff --git a/dev/x86_64/src/ntt.S b/dev/x86_64/src/ntt.S
@@ -21,6 +21,7 @@
 #include "../../../common.h"
 #if defined(MLD_ARITH_BACKEND_X86_64_DEFAULT) && \
     !defined(MLD_CONFIG_MULTILEVEL_NO_SHARED)
+/* simpasm: header-end */
 
 #include "consts.h"
 
@@ -237,5 +238,6 @@ levels2t7	2
 levels2t7	3
 
 ret
+/* simpasm: footer-start */
 #endif /* MLD_ARITH_BACKEND_X86_64_DEFAULT && !MLD_CONFIG_MULTILEVEL_NO_SHARED \
         */
diff --git a/dev/x86_64/src/nttunpack.S b/dev/x86_64/src/nttunpack.S
diff --git a/dev/x86_64/src/pointwise.S b/dev/x86_64/src/pointwise.S
diff --git a/dev/x86_64/src/pointwise_acc_l4.S b/dev/x86_64/src/pointwise_acc_l4.S
diff --git a/dev/x86_64/src/pointwise_acc_l5.S b/dev/x86_64/src/pointwise_acc_l5.S
diff --git a/dev/x86_64/src/pointwise_acc_l7.S b/dev/x86_64/src/pointwise_acc_l7.S
diff --git a/scripts/lint b/scripts/lint
