diff --git a/CHANGELOG.md b/CHANGELOG.md
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -3,6 +3,12 @@
- SPDX-License-Identifier: MPL-2.0
-->
+## v0.0.3
+
+* Minor tweaks for GHC 9.8 compatibility.
+* Types `Point` and `Affine` now have `PointKind` parameter be role `nominal`, as coercing those between point types will break invariants.
+* Bundled blst version bumped to v0.3.11.
+
## v0.0.2
* Prevent inlining of foreign calls. This fixes a potential efficiency issue,
diff --git a/README.md b/README.md
--- a/README.md
+++ b/README.md
@@ -8,7 +8,7 @@
HsBLST is low-level Haskell bindings and a high-level interface to [BLST](https://github.com/supranational/blst) -- a multilingual BLS12-381 signature library.
-This package includes sources from v0.3.10 version of BLST, licensed under
+This package includes sources from v0.3.11 version of BLST, licensed under
[Apache License 2.0](https://github.com/supranational/blst/blob/master/LICENSE)
## Build Instructions [↑](#hsblst)
diff --git a/c-source/README.md b/c-source/README.md
--- a/c-source/README.md
+++ b/c-source/README.md
@@ -1,4 +1,4 @@
-[](https://travis-ci.com/github/supranational/blst) [](https://github.com/supranational/blst/actions) [](https://lgtm.com/projects/g/supranational/blst/alerts/)
+[](https://travis-ci.com/github/supranational/blst) [](https://github.com/supranational/blst/actions)
@@ -56,6 +56,8 @@
- [Python](bindings/python)
- [Java](bindings/java)
- [Node.js](bindings/node.js)
+- [Emscripten](bindings/emscripten)
+- [C#](bindings/c%23)
## API
@@ -134,7 +136,7 @@
* The third-most significant bit is set if (and only if) this point is in compressed form _and_ it is not the point at infinity _and_ its y-coordinate is the lexicographically largest of the two associated with the encoded x-coordinate.
## Build
-The build process is very simple and only requires a C complier. It's integrated into the Go and Rust ecosystems, so that respective users would go about as they would with any other external module. Otherwise, a binary library would have to be compiled.
+The build process is very simple and only requires a C compiler. It's integrated into the Go and Rust ecosystems, so that respective users would go about as they would with any other external module. Otherwise, a binary library would have to be compiled.
### C static library
A static library called libblst.a can be built in the current working directory of the user's choice:
diff --git a/c-source/bindings/blst.h b/c-source/bindings/blst.h
--- a/c-source/bindings/blst.h
+++ b/c-source/bindings/blst.h
@@ -83,7 +83,7 @@
#ifndef SWIG
/*
- * BLS12-381-specifc Fr operations.
+ * BLS12-381-specific Fr operations.
*/
void blst_fr_add(blst_fr *ret, const blst_fr *a, const blst_fr *b);
void blst_fr_sub(blst_fr *ret, const blst_fr *a, const blst_fr *b);
@@ -102,7 +102,7 @@
void blst_scalar_from_fr(blst_scalar *ret, const blst_fr *a);
/*
- * BLS12-381-specifc Fp operations.
+ * BLS12-381-specific Fp operations.
*/
void blst_fp_add(blst_fp *ret, const blst_fp *a, const blst_fp *b);
void blst_fp_sub(blst_fp *ret, const blst_fp *a, const blst_fp *b);
@@ -126,7 +126,7 @@
void blst_lendian_from_fp(byte ret[48], const blst_fp *a);
/*
- * BLS12-381-specifc Fp2 operations.
+ * BLS12-381-specific Fp2 operations.
*/
void blst_fp2_add(blst_fp2 *ret, const blst_fp2 *a, const blst_fp2 *b);
void blst_fp2_sub(blst_fp2 *ret, const blst_fp2 *a, const blst_fp2 *b);
@@ -141,7 +141,7 @@
bool blst_fp2_sqrt(blst_fp2 *ret, const blst_fp2 *a);
/*
- * BLS12-381-specifc Fp12 operations.
+ * BLS12-381-specific Fp12 operations.
*/
void blst_fp12_sqr(blst_fp12 *ret, const blst_fp12 *a);
void blst_fp12_cyclotomic_sqr(blst_fp12 *ret, const blst_fp12 *a);
@@ -155,11 +155,11 @@
bool blst_fp12_is_equal(const blst_fp12 *a, const blst_fp12 *b);
bool blst_fp12_is_one(const blst_fp12 *a);
bool blst_fp12_in_group(const blst_fp12 *a);
-const blst_fp12 *blst_fp12_one();
+const blst_fp12 *blst_fp12_one(void);
#endif // SWIG
/*
- * BLS12-381-specifc point operations.
+ * BLS12-381-specific point operations.
*/
typedef struct { blst_fp x, y, z; } blst_p1;
typedef struct { blst_fp x, y; } blst_p1_affine;
@@ -180,13 +180,13 @@
bool blst_p1_in_g1(const blst_p1 *p);
bool blst_p1_is_equal(const blst_p1 *a, const blst_p1 *b);
bool blst_p1_is_inf(const blst_p1 *a);
-const blst_p1 *blst_p1_generator();
+const blst_p1 *blst_p1_generator(void);
bool blst_p1_affine_on_curve(const blst_p1_affine *p);
bool blst_p1_affine_in_g1(const blst_p1_affine *p);
bool blst_p1_affine_is_equal(const blst_p1_affine *a, const blst_p1_affine *b);
bool blst_p1_affine_is_inf(const blst_p1_affine *a);
-const blst_p1_affine *blst_p1_affine_generator();
+const blst_p1_affine *blst_p1_affine_generator(void);
typedef struct { blst_fp2 x, y, z; } blst_p2;
typedef struct { blst_fp2 x, y; } blst_p2_affine;
@@ -207,13 +207,13 @@
bool blst_p2_in_g2(const blst_p2 *p);
bool blst_p2_is_equal(const blst_p2 *a, const blst_p2 *b);
bool blst_p2_is_inf(const blst_p2 *a);
-const blst_p2 *blst_p2_generator();
+const blst_p2 *blst_p2_generator(void);
bool blst_p2_affine_on_curve(const blst_p2_affine *p);
bool blst_p2_affine_in_g2(const blst_p2_affine *p);
bool blst_p2_affine_is_equal(const blst_p2_affine *a, const blst_p2_affine *b);
bool blst_p2_affine_is_inf(const blst_p2_affine *a);
-const blst_p2_affine *blst_p2_affine_generator();
+const blst_p2_affine *blst_p2_affine_generator(void);
/*
* Multi-scalar multiplications and other multi-point operations.
@@ -337,6 +337,9 @@
#ifndef SWIG
void blst_miller_loop(blst_fp12 *ret, const blst_p2_affine *Q,
const blst_p1_affine *P);
+void blst_miller_loop_n(blst_fp12 *ret, const blst_p2_affine *const Qs[],
+ const blst_p1_affine *const Ps[],
+ size_t n);
void blst_final_exp(blst_fp12 *ret, const blst_fp12 *f);
void blst_precompute_lines(blst_fp6 Qlines[68], const blst_p2_affine *Q);
void blst_miller_loop_lines(blst_fp12 *ret, const blst_fp6 Qlines[68],
@@ -352,7 +355,7 @@
typedef struct blst_opaque blst_pairing;
#endif
-size_t blst_pairing_sizeof();
+size_t blst_pairing_sizeof(void);
void blst_pairing_init(blst_pairing *new_ctx, bool hash_or_encode,
const byte *DST DEFNULL, size_t DST_len DEFNULL);
const byte *blst_pairing_get_dst(const blst_pairing *ctx);
diff --git a/c-source/bindings/blst_aux.h b/c-source/bindings/blst_aux.h
--- a/c-source/bindings/blst_aux.h
+++ b/c-source/bindings/blst_aux.h
@@ -10,8 +10,14 @@
* depending on their proven/unproven worthiness.
*/
+void blst_fr_ct_bfly(blst_fr *x0, blst_fr *x1, const blst_fr *twiddle);
+void blst_fr_gs_bfly(blst_fr *x0, blst_fr *x1, const blst_fr *twiddle);
void blst_fr_to(blst_fr *ret, const blst_fr *a);
void blst_fr_from(blst_fr *ret, const blst_fr *a);
+#ifdef BLST_FR_PENTAROOT
+void blst_fr_pentaroot(blst_fr *ret, const blst_fr *a);
+void blst_fr_pentapow(blst_fr *ret, const blst_fr *a);
+#endif
void blst_fp_to(blst_fp *ret, const blst_fp *a);
void blst_fp_from(blst_fp *ret, const blst_fp *a);
@@ -50,7 +56,11 @@
void blst_sign_pk2_in_g2(byte out[96], blst_p1_affine *out_sig,
const blst_p1 *hash, const blst_scalar *SK);
+#ifdef __BLST_RUST_BINDGEN__
typedef struct {} blst_uniq;
+#else
+typedef struct blst_opaque blst_uniq;
+#endif
size_t blst_uniq_sizeof(size_t n_nodes);
void blst_uniq_init(blst_uniq *tree);
@@ -94,9 +104,14 @@
void blst_fr_from_hexascii(blst_fr *ret, const byte *hex);
void blst_fp_from_hexascii(blst_fp *ret, const byte *hex);
-size_t blst_p1_sizeof();
-size_t blst_p1_affine_sizeof();
-size_t blst_p2_sizeof();
-size_t blst_p2_affine_sizeof();
-size_t blst_fp12_sizeof();
+size_t blst_p1_sizeof(void);
+size_t blst_p1_affine_sizeof(void);
+size_t blst_p2_sizeof(void);
+size_t blst_p2_affine_sizeof(void);
+size_t blst_fp12_sizeof(void);
+
+/*
+ * Single-shot SHA-256 hash function.
+ */
+void blst_sha256(byte out[32], const byte *msg, size_t msg_len);
#endif
diff --git a/c-source/build/assembly.S b/c-source/build/assembly.S
--- a/c-source/build/assembly.S
+++ b/c-source/build/assembly.S
@@ -2,23 +2,22 @@
# if defined(__ELF__)
# if defined(__BLST_PORTABLE__)
# include "elf/sha256-portable-x86_64.s"
-# else
-# include "elf/sha256-x86_64.s"
+# define blst_sha256_block_data_order blst_sha256_block_ssse3
# endif
-# if defined(__ADX__) && !defined(__BLST_PORTABLE__)
+# include "elf/sha256-x86_64.s"
+# if defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "elf/ctx_inverse_mod_384-x86_64.s"
-# else
+# endif
+# if !defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "elf/ctq_inverse_mod_384-x86_64.s"
# endif
# include "elf/add_mod_384-x86_64.s"
# include "elf/add_mod_384x384-x86_64.s"
-# define __add_mod_384 __add_mont_384
-# define __sub_mod_384 __sub_mont_384
-# define __sub_mod_384x384 __sub_mont_384x384
-# if defined(__ADX__) && !defined(__BLST_PORTABLE__)
+# if defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "elf/mulx_mont_384-x86_64.s"
# include "elf/mulx_mont_256-x86_64.s"
-# else
+# endif
+# if !defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "elf/mulq_mont_384-x86_64.s"
# include "elf/mulq_mont_256-x86_64.s"
# endif
@@ -27,25 +26,20 @@
# include "elf/div3w-x86_64.s"
# include "elf/ct_is_square_mod_384-x86_64.s"
# elif defined(_WIN64) || defined(__CYGWIN__)
-# if defined(__BLST_PORTABLE__)
-# include "coff/sha256-portable-x86_64.s"
-# else
-# include "coff/sha256-x86_64.s"
-# endif
-# if defined(__ADX__) && !defined(__BLST_PORTABLE__)
+# include "coff/sha256-x86_64.s"
+# if defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "coff/ctx_inverse_mod_384-x86_64.s"
-# else
+# endif
+# if !defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "coff/ctq_inverse_mod_384-x86_64.s"
# endif
# include "coff/add_mod_384-x86_64.s"
# include "coff/add_mod_384x384-x86_64.s"
-# define __add_mod_384 __add_mont_384
-# define __sub_mod_384 __sub_mont_384
-# define __sub_mod_384x384 __sub_mont_384x384
-# if defined(__ADX__) && !defined(__BLST_PORTABLE__)
+# if defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "coff/mulx_mont_384-x86_64.s"
# include "coff/mulx_mont_256-x86_64.s"
-# else
+# endif
+# if !defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "coff/mulq_mont_384-x86_64.s"
# include "coff/mulq_mont_256-x86_64.s"
# endif
@@ -55,20 +49,19 @@
# include "coff/ct_is_square_mod_384-x86_64.s"
# elif defined(__APPLE__)
# include "mach-o/sha256-x86_64.s"
-# if defined(__ADX__) && !defined(__BLST_PORTABLE__)
+# if defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "mach-o/ctx_inverse_mod_384-x86_64.s"
-# else
+# endif
+# if !defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "mach-o/ctq_inverse_mod_384-x86_64.s"
# endif
# include "mach-o/add_mod_384-x86_64.s"
# include "mach-o/add_mod_384x384-x86_64.s"
-# define __add_mod_384 __add_mont_384
-# define __sub_mod_384 __sub_mont_384
-# define __sub_mod_384x384 __sub_mont_384x384
-# if defined(__ADX__) && !defined(__BLST_PORTABLE__)
+# if defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "mach-o/mulx_mont_384-x86_64.s"
# include "mach-o/mulx_mont_256-x86_64.s"
-# else
+# endif
+# if !defined(__ADX__) || defined(__BLST_PORTABLE__)
# include "mach-o/mulq_mont_384-x86_64.s"
# include "mach-o/mulq_mont_256-x86_64.s"
# endif
diff --git a/c-source/build/coff/add_mod_256-x86_64.s b/c-source/build/coff/add_mod_256-x86_64.s
--- a/c-source/build/coff/add_mod_256-x86_64.s
+++ b/c-source/build/coff/add_mod_256-x86_64.s
@@ -10,14 +10,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_add_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
subq $8,%rsp
@@ -81,13 +81,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_by_3_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
pushq %rbx
pushq %r12
@@ -161,14 +161,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_lshift_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -218,14 +218,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_rshift_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
subq $8,%rsp
@@ -307,14 +307,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_cneg_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -385,14 +385,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sub_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
subq $8,%rsp
@@ -454,10 +454,10 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_check_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
+ movq %rcx,%rdi
+ movq %rdx,%rsi
movq 0(%rdi),%rax
movq 8(%rdi),%r9
movq 16(%rdi),%r10
@@ -497,14 +497,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_add_n_check_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
subq $8,%rsp
@@ -573,14 +573,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sub_n_check_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
subq $8,%rsp
@@ -744,8 +744,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_add_mod_256_body:
.byte 1,0,9,0
.byte 0x00,0x34,0x01,0x00
@@ -753,7 +754,8 @@
.byte 0x00,0x74,0x04,0x00
.byte 0x00,0x64,0x05,0x00
.byte 0x00,0x22
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_add_mod_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -764,8 +766,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_by_3_mod_256_body:
.byte 1,0,11,0
.byte 0x00,0xc4,0x00,0x00
@@ -785,8 +788,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_lshift_mod_256_body:
.byte 1,0,11,0
.byte 0x00,0xc4,0x00,0x00
@@ -806,8 +810,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_rshift_mod_256_body:
.byte 1,0,9,0
.byte 0x00,0x34,0x01,0x00
@@ -815,7 +820,8 @@
.byte 0x00,0x74,0x04,0x00
.byte 0x00,0x64,0x05,0x00
.byte 0x00,0x22
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_rshift_mod_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -826,8 +832,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_cneg_mod_256_body:
.byte 1,0,11,0
.byte 0x00,0xc4,0x00,0x00
@@ -847,8 +854,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sub_mod_256_body:
.byte 1,0,9,0
.byte 0x00,0x34,0x01,0x00
@@ -856,7 +864,8 @@
.byte 0x00,0x74,0x04,0x00
.byte 0x00,0x64,0x05,0x00
.byte 0x00,0x22
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sub_mod_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -873,8 +882,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_add_n_check_mod_256_body:
.byte 1,0,9,0
.byte 0x00,0x34,0x01,0x00
@@ -882,7 +892,8 @@
.byte 0x00,0x74,0x04,0x00
.byte 0x00,0x64,0x05,0x00
.byte 0x00,0x22
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_add_n_check_mod_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -893,8 +904,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sub_n_check_mod_256_body:
.byte 1,0,9,0
.byte 0x00,0x34,0x01,0x00
@@ -902,7 +914,8 @@
.byte 0x00,0x74,0x04,0x00
.byte 0x00,0x64,0x05,0x00
.byte 0x00,0x22
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sub_n_check_mod_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/add_mod_384-armv8.S b/c-source/build/coff/add_mod_384-armv8.S
--- a/c-source/build/coff/add_mod_384-armv8.S
+++ b/c-source/build/coff/add_mod_384-armv8.S
@@ -786,6 +786,23 @@
ret
+.globl vec_select_32
+
+.def vec_select_32;
+.type 32;
+.endef
+.p2align 5
+vec_select_32:
+ dup v6.2d, x3
+ ld1 {v0.2d, v1.2d, v2.2d}, [x1],#48
+ cmeq v6.2d, v6.2d, #0
+ ld1 {v3.2d, v4.2d, v5.2d}, [x2],#48
+ bit v0.16b, v3.16b, v6.16b
+ bit v1.16b, v4.16b, v6.16b
+ bit v2.16b, v5.16b, v6.16b
+ st1 {v0.2d, v1.2d, v2.2d}, [x0]
+ ret
+
.globl vec_select_48
.def vec_select_48;
@@ -977,5 +994,63 @@
cmp x0, x1
csel x0, x1, x0, hi
prfm pldl1keep, [x0]
+ ret
+
+.globl vec_is_zero_16x
+
+.def vec_is_zero_16x;
+.type 32;
+.endef
+.p2align 5
+vec_is_zero_16x:
+ ld1 {v0.2d}, [x0], #16
+ lsr x1, x1, #4
+ sub x1, x1, #1
+ cbz x1, .Loop_is_zero_done
+
+.Loop_is_zero:
+ ld1 {v1.2d}, [x0], #16
+ orr v0.16b, v0.16b, v1.16b
+ sub x1, x1, #1
+ cbnz x1, .Loop_is_zero
+
+.Loop_is_zero_done:
+ dup v1.2d, v0.d[1]
+ orr v0.16b, v0.16b, v1.16b
+ mov x1, v0.d[0]
+ mov x0, #1
+ cmp x1, #0
+ csel x0, x0, xzr, eq
+ ret
+
+.globl vec_is_equal_16x
+
+.def vec_is_equal_16x;
+.type 32;
+.endef
+.p2align 5
+vec_is_equal_16x:
+ ld1 {v0.2d}, [x0], #16
+ ld1 {v1.2d}, [x1], #16
+ lsr x2, x2, #4
+ eor v0.16b, v0.16b, v1.16b
+
+.Loop_is_equal:
+ sub x2, x2, #1
+ cbz x2, .Loop_is_equal_done
+ ld1 {v1.2d}, [x0], #16
+ ld1 {v2.2d}, [x1], #16
+ eor v1.16b, v1.16b, v2.16b
+ orr v0.16b, v0.16b, v1.16b
+ b .Loop_is_equal
+ nop
+
+.Loop_is_equal_done:
+ dup v1.2d, v0.d[1]
+ orr v0.16b, v0.16b, v1.16b
+ mov x1, v0.d[0]
+ mov x0, #1
+ cmp x1, #0
+ csel x0, x0, xzr, eq
ret
diff --git a/c-source/build/coff/add_mod_384-x86_64.s b/c-source/build/coff/add_mod_384-x86_64.s
--- a/c-source/build/coff/add_mod_384-x86_64.s
+++ b/c-source/build/coff/add_mod_384-x86_64.s
@@ -10,14 +10,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_add_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -118,14 +118,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_add_mod_384x:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -186,14 +186,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_rshift_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -315,13 +315,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_div_by_2_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
pushq %rbx
pushq %r12
@@ -387,14 +387,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_lshift_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -528,13 +528,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_by_3_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
pushq %rbx
pushq %r12
@@ -595,13 +595,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_by_8_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
pushq %rbx
pushq %r12
@@ -669,13 +669,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_by_3_mod_384x:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
pushq %rbx
pushq %r12
@@ -752,13 +752,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_by_8_mod_384x:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
pushq %rbx
pushq %r12
@@ -845,14 +845,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_cneg_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -952,14 +952,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sub_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -1058,14 +1058,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sub_mod_384x:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -1124,13 +1124,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_by_1_plus_i_mod_384x:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
pushq %rbx
pushq %r12
@@ -1274,10 +1274,10 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sgn0_pty_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
+ movq %rcx,%rdi
+ movq %rdx,%rsi
.LSEH_body_sgn0_pty_mod_384:
movq 0(%rdi),%r8
@@ -1328,12 +1328,12 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sgn0_pty_mod_384x:
- movq %rcx,%rdi
- movq %rdx,%rsi
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
pushq %rbx
subq $8,%rsp
@@ -1439,6 +1439,35 @@
.byte 0xf3,0xc3
.LSEH_end_sgn0_pty_mod_384x:
+.globl vec_select_32
+
+.def vec_select_32; .scl 2; .type 32; .endef
+.p2align 5
+vec_select_32:
+ .byte 0xf3,0x0f,0x1e,0xfa
+
+ movd %r9d,%xmm5
+ pxor %xmm4,%xmm4
+ pshufd $0,%xmm5,%xmm5
+ movdqu (%rdx),%xmm0
+ leaq 16(%rdx),%rdx
+ pcmpeqd %xmm4,%xmm5
+ movdqu (%r8),%xmm1
+ leaq 16(%r8),%r8
+ pcmpeqd %xmm5,%xmm4
+ leaq 16(%rcx),%rcx
+ pand %xmm4,%xmm0
+ movdqu 0+16-16(%rdx),%xmm2
+ pand %xmm5,%xmm1
+ movdqu 0+16-16(%r8),%xmm3
+ por %xmm1,%xmm0
+ movdqu %xmm0,0-16(%rcx)
+ pand %xmm4,%xmm2
+ pand %xmm5,%xmm3
+ por %xmm3,%xmm2
+ movdqu %xmm2,16-16(%rcx)
+ .byte 0xf3,0xc3
+
.globl vec_select_48
.def vec_select_48; .scl 2; .type 32; .endef
@@ -1854,6 +1883,69 @@
prefetchnta (%rcx)
.byte 0xf3,0xc3
+.globl vec_is_zero_16x
+
+.def vec_is_zero_16x; .scl 2; .type 32; .endef
+.p2align 5
+vec_is_zero_16x:
+ .byte 0xf3,0x0f,0x1e,0xfa
+
+ shrl $4,%edx
+ movdqu (%rcx),%xmm0
+ leaq 16(%rcx),%rcx
+
+.Loop_is_zero:
+ decl %edx
+ jz .Loop_is_zero_done
+ movdqu (%rcx),%xmm1
+ leaq 16(%rcx),%rcx
+ por %xmm1,%xmm0
+ jmp .Loop_is_zero
+
+.Loop_is_zero_done:
+ pshufd $0x4e,%xmm0,%xmm1
+ por %xmm1,%xmm0
+.byte 102,72,15,126,192
+ incl %edx
+ testq %rax,%rax
+ cmovnzl %edx,%eax
+ xorl $1,%eax
+ .byte 0xf3,0xc3
+
+.globl vec_is_equal_16x
+
+.def vec_is_equal_16x; .scl 2; .type 32; .endef
+.p2align 5
+vec_is_equal_16x:
+ .byte 0xf3,0x0f,0x1e,0xfa
+
+ shrl $4,%r8d
+ movdqu (%rcx),%xmm0
+ movdqu (%rdx),%xmm1
+ subq %rcx,%rdx
+ leaq 16(%rcx),%rcx
+ pxor %xmm1,%xmm0
+
+.Loop_is_equal:
+ decl %r8d
+ jz .Loop_is_equal_done
+ movdqu (%rcx),%xmm1
+ movdqu (%rcx,%rdx,1),%xmm2
+ leaq 16(%rcx),%rcx
+ pxor %xmm2,%xmm1
+ por %xmm1,%xmm0
+ jmp .Loop_is_equal
+
+.Loop_is_equal_done:
+ pshufd $0x4e,%xmm0,%xmm1
+ por %xmm1,%xmm0
+.byte 102,72,15,126,192
+ incl %r8d
+ testq %rax,%rax
+ cmovnzl %r8d,%eax
+ xorl $1,%eax
+ .byte 0xf3,0xc3
+
.section .pdata
.p2align 2
.rva .LSEH_begin_add_mod_384
@@ -2042,8 +2134,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_add_mod_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2055,7 +2148,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_add_mod_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2066,8 +2160,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_add_mod_384x_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x03,0x00
@@ -2079,7 +2174,8 @@
.byte 0x00,0x74,0x0a,0x00
.byte 0x00,0x64,0x0b,0x00
.byte 0x00,0x82
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_add_mod_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2090,8 +2186,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_rshift_mod_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2103,7 +2200,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_rshift_mod_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2114,8 +2212,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_div_by_2_mod_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2127,7 +2226,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_div_by_2_mod_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2138,8 +2238,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_lshift_mod_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2151,7 +2252,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_lshift_mod_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2162,8 +2264,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_by_3_mod_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2175,7 +2278,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mul_by_3_mod_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2186,8 +2290,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_by_8_mod_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2199,7 +2304,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mul_by_8_mod_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2210,8 +2316,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_by_3_mod_384x_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2223,7 +2330,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mul_by_3_mod_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2234,8 +2342,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_by_8_mod_384x_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2247,7 +2356,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mul_by_8_mod_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2258,8 +2368,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_cneg_mod_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2271,7 +2382,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_cneg_mod_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2282,8 +2394,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sub_mod_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -2295,7 +2408,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sub_mod_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2306,8 +2420,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sub_mod_384x_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x03,0x00
@@ -2319,7 +2434,8 @@
.byte 0x00,0x74,0x0a,0x00
.byte 0x00,0x64,0x0b,0x00
.byte 0x00,0x82
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sub_mod_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2330,8 +2446,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_by_1_plus_i_mod_384x_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x07,0x00
@@ -2343,7 +2460,8 @@
.byte 0x00,0x74,0x0e,0x00
.byte 0x00,0x64,0x0f,0x00
.byte 0x00,0xc2
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mul_by_1_plus_i_mod_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2354,8 +2472,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sgn0_pty_mod_384_body:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -2371,8 +2490,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sgn0_pty_mod_384x_body:
.byte 1,0,9,0
.byte 0x00,0x34,0x01,0x00
@@ -2380,7 +2500,8 @@
.byte 0x00,0x74,0x04,0x00
.byte 0x00,0x64,0x05,0x00
.byte 0x00,0x22
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sgn0_pty_mod_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/add_mod_384x384-x86_64.s b/c-source/build/coff/add_mod_384x384-x86_64.s
--- a/c-source/build/coff/add_mod_384x384-x86_64.s
+++ b/c-source/build/coff/add_mod_384x384-x86_64.s
@@ -145,14 +145,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_add_mod_384x384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -202,14 +202,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sub_mod_384x384:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -280,8 +280,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_add_mod_384x384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -293,7 +294,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_add_mod_384x384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -304,8 +306,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sub_mod_384x384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -317,7 +320,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sub_mod_384x384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/ct_inverse_mod_256-armv8.S b/c-source/build/coff/ct_inverse_mod_256-armv8.S
--- a/c-source/build/coff/ct_inverse_mod_256-armv8.S
+++ b/c-source/build/coff/ct_inverse_mod_256-armv8.S
@@ -1,6 +1,7 @@
.text
.globl ct_inverse_mod_256
+
.def ct_inverse_mod_256;
.type 32;
.endef
@@ -62,14 +63,14 @@
madd x4, x16, x8, xzr // |u|*|f0|
madd x4, x17, x9, x4 // |v|*|g0|
str x4, [x0,#8*4]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*5]
stp x5, x5, [x0,#8*7]
madd x4, x12, x8, xzr // |u|*|f1|
madd x4, x13, x9, x4 // |v|*|g1|
str x4, [x0,#8*9]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*10]
stp x5, x5, [x0,#8*12]
eor x1, x1, #256 // flip-flop src |a|b|u|v|
diff --git a/c-source/build/coff/ct_inverse_mod_256-x86_64.s b/c-source/build/coff/ct_inverse_mod_256-x86_64.s
--- a/c-source/build/coff/ct_inverse_mod_256-x86_64.s
+++ b/c-source/build/coff/ct_inverse_mod_256-x86_64.s
@@ -1,6 +1,7 @@
.text
.globl ct_inverse_mod_256
+
.def ct_inverse_mod_256; .scl 2; .type 32; .endef
.p2align 5
ct_inverse_mod_256:
@@ -9,14 +10,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_ct_inverse_mod_256:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- movq %r9,%rcx
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ movq %r9,%rcx
pushq %rbx
pushq %r12
@@ -1188,8 +1189,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_ct_inverse_mod_256_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x86,0x00
@@ -1201,6 +1203,8 @@
.byte 0x00,0x74,0x8d,0x00
.byte 0x00,0x64,0x8e,0x00
.byte 0x00,0x01,0x8c,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_ct_inverse_mod_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/ct_inverse_mod_384-armv8.S b/c-source/build/coff/ct_inverse_mod_384-armv8.S
--- a/c-source/build/coff/ct_inverse_mod_384-armv8.S
+++ b/c-source/build/coff/ct_inverse_mod_384-armv8.S
@@ -1,6 +1,7 @@
.text
.globl ct_inverse_mod_383
+
.def ct_inverse_mod_383;
.type 32;
.endef
@@ -73,7 +74,7 @@
adds x3, x3, x5
adc x4, x4, x6
stp x3, x4, [x0,#8*6]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*8]
stp x5, x5, [x0,#8*10]
@@ -84,7 +85,7 @@
adds x3, x3, x5
adc x4, x4, x6
stp x3, x4, [x0,#8*12]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*14]
stp x5, x5, [x0,#8*16]
eor x1, x1, #256 // flip-flop src |a|b|u|v|
diff --git a/c-source/build/coff/ct_is_square_mod_384-armv8.S b/c-source/build/coff/ct_is_square_mod_384-armv8.S
--- a/c-source/build/coff/ct_is_square_mod_384-armv8.S
+++ b/c-source/build/coff/ct_is_square_mod_384-armv8.S
@@ -1,6 +1,7 @@
.text
.globl ct_is_square_mod_384
+
.def ct_is_square_mod_384;
.type 32;
.endef
diff --git a/c-source/build/coff/ct_is_square_mod_384-x86_64.s b/c-source/build/coff/ct_is_square_mod_384-x86_64.s
--- a/c-source/build/coff/ct_is_square_mod_384-x86_64.s
+++ b/c-source/build/coff/ct_is_square_mod_384-x86_64.s
@@ -1,6 +1,7 @@
.text
.globl ct_is_square_mod_384
+
.def ct_is_square_mod_384; .scl 2; .type 32; .endef
.p2align 5
ct_is_square_mod_384:
@@ -9,12 +10,12 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_ct_is_square_mod_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
pushq %rbp
+ movq %rcx,%rdi
+ movq %rdx,%rsi
pushq %rbx
pushq %r12
@@ -484,8 +485,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_ct_is_square_mod_384_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x43,0x00
@@ -497,6 +499,8 @@
.byte 0x00,0x74,0x4a,0x00
.byte 0x00,0x64,0x4b,0x00
.byte 0x00,0x01,0x49,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_ct_is_square_mod_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/ctq_inverse_mod_384-x86_64.s b/c-source/build/coff/ctq_inverse_mod_384-x86_64.s
--- a/c-source/build/coff/ctq_inverse_mod_384-x86_64.s
+++ b/c-source/build/coff/ctq_inverse_mod_384-x86_64.s
@@ -1,6 +1,8 @@
+.comm __blst_platform_cap,4
.text
.globl ct_inverse_mod_383
+
.def ct_inverse_mod_383; .scl 2; .type 32; .endef
.p2align 5
ct_inverse_mod_383:
@@ -9,12 +11,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_ct_inverse_mod_383:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz ct_inverse_mod_383$1
+#endif
pushq %rbp
pushq %rbx
@@ -1200,8 +1206,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_ct_inverse_mod_383_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x8b,0x00
@@ -1213,6 +1220,8 @@
.byte 0x00,0x74,0x92,0x00
.byte 0x00,0x64,0x93,0x00
.byte 0x00,0x01,0x91,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_ct_inverse_mod_383_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/ctx_inverse_mod_384-x86_64.s b/c-source/build/coff/ctx_inverse_mod_384-x86_64.s
--- a/c-source/build/coff/ctx_inverse_mod_384-x86_64.s
+++ b/c-source/build/coff/ctx_inverse_mod_384-x86_64.s
@@ -1,6 +1,7 @@
.text
.globl ctx_inverse_mod_383
+
.def ctx_inverse_mod_383; .scl 2; .type 32; .endef
.p2align 5
ctx_inverse_mod_383:
@@ -9,12 +10,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_ctx_inverse_mod_383:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+ct_inverse_mod_383$1:
pushq %rbp
pushq %rbx
@@ -812,7 +814,7 @@
movq 48(%rsi),%r10
- call __inner_loop_62
+ call __tail_loop_53
@@ -1514,9 +1516,9 @@
.byte 0xf3,0xc3
-.def __inner_loop_62; .scl 3; .type 32; .endef
+.def __tail_loop_53; .scl 3; .type 32; .endef
.p2align 5
-__inner_loop_62:
+__tail_loop_53:
.byte 0xf3,0x0f,0x1e,0xfa
movq $1,%rdx
@@ -1524,7 +1526,7 @@
xorq %r12,%r12
movq $1,%r13
-.Loop_62:
+.Loop_53:
xorq %rax,%rax
testq $1,%r8
movq %r10,%rbx
@@ -1551,7 +1553,7 @@
subq %rax,%rdx
subq %rbx,%rcx
subl $1,%edi
- jnz .Loop_62
+ jnz .Loop_53
.byte 0xf3,0xc3
@@ -1575,8 +1577,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_ctx_inverse_mod_383_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x8b,0x00
@@ -1588,6 +1591,8 @@
.byte 0x00,0x74,0x92,0x00
.byte 0x00,0x64,0x93,0x00
.byte 0x00,0x01,0x91,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_ctx_inverse_mod_383_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/div3w-armv8.S b/c-source/build/coff/div3w-armv8.S
--- a/c-source/build/coff/div3w-armv8.S
+++ b/c-source/build/coff/div3w-armv8.S
@@ -27,7 +27,7 @@
asr x3,x0,#63 // top bit -> mask
add x0,x0,x0 // Q <<= 1
subs x6,x4,x1 // R - D
- add x0,x0,#1 // Q + specilative bit
+ add x0,x0,#1 // Q + speculative bit
sbcs x7,x5,x2
sbc x0,x0,xzr // subtract speculative bit
diff --git a/c-source/build/coff/div3w-x86_64.s b/c-source/build/coff/div3w-x86_64.s
--- a/c-source/build/coff/div3w-x86_64.s
+++ b/c-source/build/coff/div3w-x86_64.s
@@ -8,11 +8,14 @@
.byte 0xf3,0x0f,0x1e,0xfa
movq %rdi,8(%rsp)
movq %rsi,16(%rsp)
- movq %rsp,%rax
+ movq %rsp,%r11
.LSEH_begin_div_3_limbs:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
+.LSEH_body_div_3_limbs:
movq (%rdi),%r8
movq 8(%rdi),%r9
@@ -45,9 +48,12 @@
orq %rcx,%rax
- movq 8(%rsp),%rdi
- movq 16(%rsp),%rsi
+.LSEH_epilogue_div_3_limbs:
+ mov 8(%rsp),%rdi
+ mov 16(%rsp),%rsi
+
.byte 0xf3,0xc3
+
.LSEH_end_div_3_limbs:
.globl quot_rem_128
@@ -57,11 +63,14 @@
.byte 0xf3,0x0f,0x1e,0xfa
movq %rdi,8(%rsp)
movq %rsi,16(%rsp)
- movq %rsp,%rax
+ movq %rsp,%r11
.LSEH_begin_quot_rem_128:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
+.LSEH_body_quot_rem_128:
movq %rdx,%rax
movq %rdx,%rcx
@@ -97,9 +106,12 @@
movq %rcx,%rax
- movq 8(%rsp),%rdi
- movq 16(%rsp),%rsi
+.LSEH_epilogue_quot_rem_128:
+ mov 8(%rsp),%rdi
+ mov 16(%rsp),%rsi
+
.byte 0xf3,0xc3
+
.LSEH_end_quot_rem_128:
@@ -114,11 +126,14 @@
.byte 0xf3,0x0f,0x1e,0xfa
movq %rdi,8(%rsp)
movq %rsi,16(%rsp)
- movq %rsp,%rax
+ movq %rsp,%r11
.LSEH_begin_quot_rem_64:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
+.LSEH_body_quot_rem_64:
movq %rdx,%rax
imulq 0(%rsi),%rdx
@@ -130,11 +145,104 @@
movq %r10,0(%rdi)
movq %rax,8(%rdi)
- movq 8(%rsp),%rdi
- movq 16(%rsp),%rsi
+.LSEH_epilogue_quot_rem_64:
+ mov 8(%rsp),%rdi
+ mov 16(%rsp),%rsi
+
.byte 0xf3,0xc3
+
.LSEH_end_quot_rem_64:
.section .pdata
.p2align 2
+.rva .LSEH_begin_div_3_limbs
+.rva .LSEH_body_div_3_limbs
+.rva .LSEH_info_div_3_limbs_prologue
+
+.rva .LSEH_body_div_3_limbs
+.rva .LSEH_epilogue_div_3_limbs
+.rva .LSEH_info_div_3_limbs_body
+
+.rva .LSEH_epilogue_div_3_limbs
+.rva .LSEH_end_div_3_limbs
+.rva .LSEH_info_div_3_limbs_epilogue
+
+.rva .LSEH_begin_quot_rem_128
+.rva .LSEH_body_quot_rem_128
+.rva .LSEH_info_quot_rem_128_prologue
+
+.rva .LSEH_body_quot_rem_128
+.rva .LSEH_epilogue_quot_rem_128
+.rva .LSEH_info_quot_rem_128_body
+
+.rva .LSEH_epilogue_quot_rem_128
+.rva .LSEH_end_quot_rem_128
+.rva .LSEH_info_quot_rem_128_epilogue
+
+.rva .LSEH_begin_quot_rem_64
+.rva .LSEH_body_quot_rem_64
+.rva .LSEH_info_quot_rem_64_prologue
+
+.rva .LSEH_body_quot_rem_64
+.rva .LSEH_epilogue_quot_rem_64
+.rva .LSEH_info_quot_rem_64_body
+
+.rva .LSEH_epilogue_quot_rem_64
+.rva .LSEH_end_quot_rem_64
+.rva .LSEH_info_quot_rem_64_epilogue
+
.section .xdata
.p2align 3
+.LSEH_info_div_3_limbs_prologue:
+.byte 1,0,5,0x0b
+.byte 0,0x74,1,0
+.byte 0,0x64,2,0
+.byte 0,0xb3
+.byte 0,0
+.long 0,0
+.LSEH_info_div_3_limbs_body:
+.byte 1,0,4,0
+.byte 0x00,0x74,0x01,0x00
+.byte 0x00,0x64,0x02,0x00
+.byte 0x00,0x00,0x00,0x00
+.LSEH_info_div_3_limbs_epilogue:
+.byte 1,0,4,0
+.byte 0x00,0x74,0x01,0x00
+.byte 0x00,0x64,0x02,0x00
+.byte 0x00,0x00,0x00,0x00
+
+.LSEH_info_quot_rem_128_prologue:
+.byte 1,0,5,0x0b
+.byte 0,0x74,1,0
+.byte 0,0x64,2,0
+.byte 0,0xb3
+.byte 0,0
+.long 0,0
+.LSEH_info_quot_rem_128_body:
+.byte 1,0,4,0
+.byte 0x00,0x74,0x01,0x00
+.byte 0x00,0x64,0x02,0x00
+.byte 0x00,0x00,0x00,0x00
+.LSEH_info_quot_rem_128_epilogue:
+.byte 1,0,4,0
+.byte 0x00,0x74,0x01,0x00
+.byte 0x00,0x64,0x02,0x00
+.byte 0x00,0x00,0x00,0x00
+
+.LSEH_info_quot_rem_64_prologue:
+.byte 1,0,5,0x0b
+.byte 0,0x74,1,0
+.byte 0,0x64,2,0
+.byte 0,0xb3
+.byte 0,0
+.long 0,0
+.LSEH_info_quot_rem_64_body:
+.byte 1,0,4,0
+.byte 0x00,0x74,0x01,0x00
+.byte 0x00,0x64,0x02,0x00
+.byte 0x00,0x00,0x00,0x00
+.LSEH_info_quot_rem_64_epilogue:
+.byte 1,0,4,0
+.byte 0x00,0x74,0x01,0x00
+.byte 0x00,0x64,0x02,0x00
+.byte 0x00,0x00,0x00,0x00
+
diff --git a/c-source/build/coff/mulq_mont_256-x86_64.s b/c-source/build/coff/mulq_mont_256-x86_64.s
--- a/c-source/build/coff/mulq_mont_256-x86_64.s
+++ b/c-source/build/coff/mulq_mont_256-x86_64.s
@@ -1,3 +1,4 @@
+.comm __blst_platform_cap,4
.text
.globl mul_mont_sparse_256
@@ -10,13 +11,17 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_mont_sparse_256:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_mont_sparse_256$1
+#endif
pushq %rbp
pushq %rbx
@@ -80,12 +85,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqr_mont_sparse_256:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_mont_sparse_256$1
+#endif
pushq %rbp
pushq %rbx
@@ -430,12 +439,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_from_mont_256:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz from_mont_256$1
+#endif
pushq %rbp
pushq %rbx
@@ -510,12 +523,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_redc_mont_256:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz redc_mont_256$1
+#endif
pushq %rbp
pushq %rbx
@@ -778,8 +795,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_mont_sparse_256_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -791,7 +809,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mul_mont_sparse_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -802,8 +821,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqr_mont_sparse_256_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -815,7 +835,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqr_mont_sparse_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -826,8 +847,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_from_mont_256_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -839,7 +861,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_from_mont_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -850,8 +873,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_redc_mont_256_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -863,7 +887,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_redc_mont_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/mulq_mont_384-x86_64.s b/c-source/build/coff/mulq_mont_384-x86_64.s
--- a/c-source/build/coff/mulq_mont_384-x86_64.s
+++ b/c-source/build/coff/mulq_mont_384-x86_64.s
@@ -1,3 +1,4 @@
+.comm __blst_platform_cap,4
.text
@@ -6,9 +7,9 @@
-.def __sub_mod_384x384; .scl 3; .type 32; .endef
+.def __subq_mod_384x384; .scl 3; .type 32; .endef
.p2align 5
-__sub_mod_384x384:
+__subq_mod_384x384:
.byte 0xf3,0x0f,0x1e,0xfa
movq 0(%rsi),%r8
@@ -73,9 +74,9 @@
.byte 0xf3,0xc3
-.def __add_mod_384; .scl 3; .type 32; .endef
+.def __addq_mod_384; .scl 3; .type 32; .endef
.p2align 5
-__add_mod_384:
+__addq_mod_384:
.byte 0xf3,0x0f,0x1e,0xfa
movq 0(%rsi),%r8
@@ -123,9 +124,9 @@
.byte 0xf3,0xc3
-.def __sub_mod_384; .scl 3; .type 32; .endef
+.def __subq_mod_384; .scl 3; .type 32; .endef
.p2align 5
-__sub_mod_384:
+__subq_mod_384:
.byte 0xf3,0x0f,0x1e,0xfa
movq 0(%rsi),%r8
@@ -135,7 +136,7 @@
movq 32(%rsi),%r12
movq 40(%rsi),%r13
-__sub_mod_384_a_is_loaded:
+__subq_mod_384_a_is_loaded:
subq 0(%rdx),%r8
movq 0(%rcx),%r14
sbbq 8(%rdx),%r9
@@ -182,13 +183,17 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_mont_384x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_mont_384x$1
+#endif
pushq %rbp
pushq %rbx
@@ -229,12 +234,12 @@
movq 8(%rsp),%rcx
leaq -48(%rsi),%rdx
leaq 40+192+48(%rsp),%rdi
- call __add_mod_384
+ call __addq_mod_384
movq 16(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq -48(%rdi),%rdi
- call __add_mod_384
+ call __addq_mod_384
leaq (%rdi),%rbx
leaq 48(%rdi),%rsi
@@ -244,17 +249,17 @@
leaq (%rdi),%rsi
leaq 40(%rsp),%rdx
movq 8(%rsp),%rcx
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq (%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq 40(%rsp),%rsi
leaq 40+96(%rsp),%rdx
leaq 40(%rsp),%rdi
- call __sub_mod_384x384
+ call __subq_mod_384x384
movq %rcx,%rbx
@@ -263,14 +268,14 @@
movq 0(%rsp),%rcx
movq 32(%rsp),%rdi
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
leaq 40+192(%rsp),%rsi
movq 0(%rsp),%rcx
leaq 48(%rdi),%rdi
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
leaq 328(%rsp),%r8
movq 0(%r8),%r15
@@ -304,12 +309,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqr_mont_384x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_mont_384x$1
+#endif
pushq %rbp
pushq %rbx
@@ -335,13 +344,13 @@
leaq 48(%rsi),%rdx
leaq 32(%rsp),%rdi
- call __add_mod_384
+ call __addq_mod_384
movq 16(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq 32+48(%rsp),%rdi
- call __sub_mod_384
+ call __subq_mod_384
movq 16(%rsp),%rsi
@@ -433,12 +442,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_382x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_382x$1
+#endif
pushq %rbp
pushq %rbx
@@ -528,18 +541,18 @@
leaq 32(%rsp),%rdx
movq 24(%rsp),%rcx
movq %rsi,%rdi
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq 0(%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq -96(%rdi),%rsi
leaq 32(%rsp),%rdx
leaq -96(%rdi),%rdi
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq 136(%rsp),%r8
movq 0(%r8),%r15
@@ -573,11 +586,15 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqr_382x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_382x$1
+#endif
pushq %rbp
pushq %rbx
@@ -628,7 +645,7 @@
leaq 48(%rsi),%rdx
leaq 48(%rdi),%rdi
- call __sub_mod_384_a_is_loaded
+ call __subq_mod_384_a_is_loaded
leaq (%rdi),%rsi
@@ -710,11 +727,15 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_384$1
+#endif
pushq %rbp
pushq %rbx
@@ -1039,10 +1060,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqr_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_384$1
+#endif
pushq %rbp
pushq %rbx
@@ -1286,12 +1311,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqr_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_mont_384$1
+#endif
pushq %rbp
pushq %rbx
@@ -1321,7 +1350,7 @@
movq 104(%rsp),%rbx
movq 112(%rsp),%rdi
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
leaq 120(%rsp),%r8
movq 120(%rsp),%r15
@@ -1358,12 +1387,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_redc_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz redc_mont_384$1
+#endif
pushq %rbp
pushq %rbx
@@ -1383,7 +1416,7 @@
movq %rdx,%rbx
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
movq 8(%rsp),%r15
@@ -1420,12 +1453,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_from_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz from_mont_384$1
+#endif
pushq %rbp
pushq %rbx
@@ -1795,9 +1832,9 @@
.byte 0xf3,0xc3
-.def __redc_tail_mont_384; .scl 3; .type 32; .endef
+.def __redq_tail_mont_384; .scl 3; .type 32; .endef
.p2align 5
-__redc_tail_mont_384:
+__redq_tail_mont_384:
.byte 0xf3,0x0f,0x1e,0xfa
addq 48(%rsi),%r14
@@ -1852,11 +1889,15 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sgn0_pty_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sgn0_pty_mont_384$1
+#endif
pushq %rbp
pushq %rbx
@@ -1934,11 +1975,15 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sgn0_pty_mont_384x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sgn0_pty_mont_384x$1
+#endif
pushq %rbp
pushq %rbx
@@ -2065,13 +2110,17 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mul_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_mont_384$1
+#endif
pushq %rbp
pushq %rbx
@@ -2733,14 +2782,18 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqr_n_mul_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
movq 48(%rsp),%r9
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_n_mul_mont_384$1
+#endif
pushq %rbp
pushq %rbx
@@ -2774,7 +2827,7 @@
movq 0(%rsp),%rcx
movq 16(%rsp),%rbx
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
movd %xmm1,%edx
leaq 0(%rdi),%rsi
@@ -2828,14 +2881,18 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqr_n_mul_mont_383:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
movq 48(%rsp),%r9
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_n_mul_mont_383$1
+#endif
pushq %rbp
pushq %rbx
@@ -3494,12 +3551,16 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqr_mont_382x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_mont_382x$1
+#endif
pushq %rbp
pushq %rbx
@@ -3858,8 +3919,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_mont_384x_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x29,0x00
@@ -3871,6 +3933,8 @@
.byte 0x00,0x74,0x30,0x00
.byte 0x00,0x64,0x31,0x00
.byte 0x00,0x01,0x2f,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mul_mont_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3881,8 +3945,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqr_mont_384x_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x11,0x00
@@ -3894,6 +3959,8 @@
.byte 0x00,0x74,0x18,0x00
.byte 0x00,0x64,0x19,0x00
.byte 0x00,0x01,0x17,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqr_mont_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3904,8 +3971,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_382x_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x11,0x00
@@ -3917,6 +3985,8 @@
.byte 0x00,0x74,0x18,0x00
.byte 0x00,0x64,0x19,0x00
.byte 0x00,0x01,0x17,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mul_382x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3927,8 +3997,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqr_382x_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -3940,7 +4011,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqr_382x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3951,8 +4023,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_384_body:
.byte 1,0,11,0
.byte 0x00,0xc4,0x00,0x00
@@ -3972,8 +4045,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqr_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -3985,7 +4059,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqr_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3996,8 +4071,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqr_mont_384_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x0f,0x00
@@ -4009,6 +4085,8 @@
.byte 0x00,0x74,0x16,0x00
.byte 0x00,0x64,0x17,0x00
.byte 0x00,0x01,0x15,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqr_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -4019,8 +4097,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_redc_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -4032,7 +4111,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_redc_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -4043,8 +4123,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_from_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -4056,7 +4137,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_from_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -4067,8 +4149,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sgn0_pty_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -4080,7 +4163,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sgn0_pty_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -4091,8 +4175,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sgn0_pty_mont_384x_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -4104,7 +4189,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sgn0_pty_mont_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -4115,8 +4201,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mul_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x03,0x00
@@ -4128,7 +4215,8 @@
.byte 0x00,0x74,0x0a,0x00
.byte 0x00,0x64,0x0b,0x00
.byte 0x00,0x82
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mul_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -4139,8 +4227,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqr_n_mul_mont_384_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x11,0x00
@@ -4152,6 +4241,8 @@
.byte 0x00,0x74,0x18,0x00
.byte 0x00,0x64,0x19,0x00
.byte 0x00,0x01,0x17,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqr_n_mul_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -4162,8 +4253,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqr_n_mul_mont_383_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x11,0x00
@@ -4175,6 +4267,8 @@
.byte 0x00,0x74,0x18,0x00
.byte 0x00,0x64,0x19,0x00
.byte 0x00,0x01,0x17,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqr_n_mul_mont_383_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -4185,8 +4279,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqr_mont_382x_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x11,0x00
@@ -4198,6 +4293,8 @@
.byte 0x00,0x74,0x18,0x00
.byte 0x00,0x64,0x19,0x00
.byte 0x00,0x01,0x17,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqr_mont_382x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/mulx_mont_256-x86_64.s b/c-source/build/coff/mulx_mont_256-x86_64.s
--- a/c-source/build/coff/mulx_mont_256-x86_64.s
+++ b/c-source/build/coff/mulx_mont_256-x86_64.s
@@ -10,13 +10,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mulx_mont_sparse_256:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
-
-
+mul_mont_sparse_256$1:
pushq %rbp
pushq %rbx
@@ -78,12 +79,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqrx_mont_sparse_256:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+sqr_mont_sparse_256$1:
pushq %rbp
pushq %rbx
@@ -342,12 +344,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_fromx_mont_256:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+from_mont_256$1:
pushq %rbp
pushq %rbx
@@ -422,12 +425,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_redcx_mont_256:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+redc_mont_256$1:
pushq %rbp
pushq %rbx
@@ -690,8 +694,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mulx_mont_sparse_256_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -703,7 +708,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mulx_mont_sparse_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -714,8 +720,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqrx_mont_sparse_256_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -727,7 +734,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqrx_mont_sparse_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -738,8 +746,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_fromx_mont_256_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -751,7 +760,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_fromx_mont_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -762,8 +772,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_redcx_mont_256_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -775,7 +786,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_redcx_mont_256_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/mulx_mont_384-x86_64.s b/c-source/build/coff/mulx_mont_384-x86_64.s
--- a/c-source/build/coff/mulx_mont_384-x86_64.s
+++ b/c-source/build/coff/mulx_mont_384-x86_64.s
@@ -6,9 +6,9 @@
-.def __sub_mod_384x384; .scl 3; .type 32; .endef
+.def __subx_mod_384x384; .scl 3; .type 32; .endef
.p2align 5
-__sub_mod_384x384:
+__subx_mod_384x384:
.byte 0xf3,0x0f,0x1e,0xfa
movq 0(%rsi),%r8
@@ -73,9 +73,9 @@
.byte 0xf3,0xc3
-.def __add_mod_384; .scl 3; .type 32; .endef
+.def __addx_mod_384; .scl 3; .type 32; .endef
.p2align 5
-__add_mod_384:
+__addx_mod_384:
.byte 0xf3,0x0f,0x1e,0xfa
movq 0(%rsi),%r8
@@ -123,9 +123,9 @@
.byte 0xf3,0xc3
-.def __sub_mod_384; .scl 3; .type 32; .endef
+.def __subx_mod_384; .scl 3; .type 32; .endef
.p2align 5
-__sub_mod_384:
+__subx_mod_384:
.byte 0xf3,0x0f,0x1e,0xfa
movq 0(%rsi),%r8
@@ -135,7 +135,7 @@
movq 32(%rsi),%r12
movq 40(%rsi),%r13
-__sub_mod_384_a_is_loaded:
+__subx_mod_384_a_is_loaded:
subq 0(%rdx),%r8
movq 0(%rcx),%r14
sbbq 8(%rdx),%r9
@@ -182,13 +182,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mulx_mont_384x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
-
-
+mul_mont_384x$1:
pushq %rbp
pushq %rbx
@@ -230,12 +231,12 @@
leaq (%rbx),%rsi
leaq -48(%rbx),%rdx
leaq 40+192+48(%rsp),%rdi
- call __add_mod_384
+ call __addx_mod_384
movq 24(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq -48(%rdi),%rdi
- call __add_mod_384
+ call __addx_mod_384
leaq (%rdi),%rbx
leaq 48(%rdi),%rsi
@@ -245,17 +246,17 @@
leaq (%rdi),%rsi
leaq 40(%rsp),%rdx
movq 8(%rsp),%rcx
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq (%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq 40(%rsp),%rsi
leaq 40+96(%rsp),%rdx
leaq 40(%rsp),%rdi
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq (%rcx),%rbx
@@ -264,14 +265,14 @@
movq 0(%rsp),%rcx
movq 32(%rsp),%rdi
call __mulx_by_1_mont_384
- call __redc_tail_mont_384
+ call __redx_tail_mont_384
leaq 40+192(%rsp),%rsi
movq 0(%rsp),%rcx
leaq 48(%rdi),%rdi
call __mulx_by_1_mont_384
- call __redc_tail_mont_384
+ call __redx_tail_mont_384
leaq 328(%rsp),%r8
movq 0(%r8),%r15
@@ -305,12 +306,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqrx_mont_384x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+sqr_mont_384x$1:
pushq %rbp
pushq %rbx
@@ -337,13 +339,13 @@
leaq 48(%rsi),%rdx
leaq 32(%rsp),%rdi
- call __add_mod_384
+ call __addx_mod_384
movq 24(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq 32+48(%rsp),%rdi
- call __sub_mod_384
+ call __subx_mod_384
movq 24(%rsp),%rsi
@@ -445,12 +447,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mulx_382x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+mul_382x$1:
pushq %rbp
pushq %rbx
@@ -540,18 +543,18 @@
leaq 32(%rsp),%rdx
movq 24(%rsp),%rcx
movq %rsi,%rdi
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq 0(%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq -96(%rdi),%rsi
leaq 32(%rsp),%rdx
leaq -96(%rdi),%rdi
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq 136(%rsp),%r8
movq 0(%r8),%r15
@@ -585,11 +588,12 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqrx_382x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
-
-
+sqr_382x$1:
pushq %rbp
pushq %rbx
@@ -640,7 +644,7 @@
leaq 48(%rsi),%rdx
leaq 48(%rdi),%rdi
- call __sub_mod_384_a_is_loaded
+ call __subx_mod_384_a_is_loaded
leaq (%rdi),%rsi
@@ -722,11 +726,12 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mulx_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
-
-
+mul_384$1:
pushq %rbp
pushq %rbx
@@ -950,10 +955,11 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqrx_384:
- movq %rcx,%rdi
- movq %rdx,%rsi
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+sqr_384$1:
pushq %rbp
pushq %rbx
@@ -1145,12 +1151,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_redcx_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+redc_mont_384$1:
pushq %rbp
pushq %rbx
@@ -1170,7 +1177,7 @@
movq %rdx,%rbx
call __mulx_by_1_mont_384
- call __redc_tail_mont_384
+ call __redx_tail_mont_384
movq 8(%rsp),%r15
@@ -1207,12 +1214,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_fromx_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+from_mont_384$1:
pushq %rbp
pushq %rbx
@@ -1473,9 +1481,9 @@
.byte 0xf3,0xc3
-.def __redc_tail_mont_384; .scl 3; .type 32; .endef
+.def __redx_tail_mont_384; .scl 3; .type 32; .endef
.p2align 5
-__redc_tail_mont_384:
+__redx_tail_mont_384:
.byte 0xf3,0x0f,0x1e,0xfa
addq 48(%rsi),%r14
@@ -1530,11 +1538,12 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sgn0x_pty_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
-
-
+sgn0_pty_mont_384$1:
pushq %rbp
pushq %rbx
@@ -1612,11 +1621,12 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sgn0x_pty_mont_384x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
-
-
+sgn0_pty_mont_384x$1:
pushq %rbp
pushq %rbx
@@ -1743,13 +1753,14 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_mulx_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
-
-
+mul_mont_384$1:
pushq %rbp
pushq %rbx
@@ -2215,12 +2226,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqrx_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+sqr_mont_384$1:
pushq %rbp
pushq %rbx
@@ -2287,14 +2299,15 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqrx_n_mul_mont_384:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
movq 48(%rsp),%r9
-
-
+sqr_n_mul_mont_384$1:
pushq %rbp
pushq %rbx
@@ -2379,14 +2392,15 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqrx_n_mul_mont_383:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
movq 40(%rsp),%r8
movq 48(%rsp),%r9
-
-
+sqr_n_mul_mont_383$1:
pushq %rbp
pushq %rbx
@@ -2831,12 +2845,13 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_sqrx_mont_382x:
+
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
movq %r9,%rcx
-
-
+sqr_mont_382x$1:
pushq %rbp
pushq %rbx
@@ -3205,8 +3220,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mulx_mont_384x_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x29,0x00
@@ -3218,6 +3234,8 @@
.byte 0x00,0x74,0x30,0x00
.byte 0x00,0x64,0x31,0x00
.byte 0x00,0x01,0x2f,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mulx_mont_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3228,8 +3246,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqrx_mont_384x_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x11,0x00
@@ -3241,6 +3260,8 @@
.byte 0x00,0x74,0x18,0x00
.byte 0x00,0x64,0x19,0x00
.byte 0x00,0x01,0x17,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqrx_mont_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3251,8 +3272,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mulx_382x_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x11,0x00
@@ -3264,6 +3286,8 @@
.byte 0x00,0x74,0x18,0x00
.byte 0x00,0x64,0x19,0x00
.byte 0x00,0x01,0x17,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mulx_382x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3274,8 +3298,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqrx_382x_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -3287,7 +3312,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqrx_382x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3298,8 +3324,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mulx_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x00,0x00
@@ -3311,7 +3338,8 @@
.byte 0x00,0x74,0x07,0x00
.byte 0x00,0x64,0x08,0x00
.byte 0x00,0x52
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mulx_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3322,8 +3350,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqrx_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -3335,7 +3364,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqrx_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3346,8 +3376,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_redcx_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -3359,7 +3390,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_redcx_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3370,8 +3402,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_fromx_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -3383,7 +3416,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_fromx_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3394,8 +3428,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sgn0x_pty_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -3407,7 +3442,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sgn0x_pty_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3418,8 +3454,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sgn0x_pty_mont_384x_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x01,0x00
@@ -3431,7 +3468,8 @@
.byte 0x00,0x74,0x08,0x00
.byte 0x00,0x64,0x09,0x00
.byte 0x00,0x62
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sgn0x_pty_mont_384x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3442,8 +3480,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_mulx_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x03,0x00
@@ -3455,7 +3494,8 @@
.byte 0x00,0x74,0x0a,0x00
.byte 0x00,0x64,0x0b,0x00
.byte 0x00,0x82
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_mulx_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3466,8 +3506,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqrx_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x03,0x00
@@ -3479,7 +3520,8 @@
.byte 0x00,0x74,0x0a,0x00
.byte 0x00,0x64,0x0b,0x00
.byte 0x00,0x82
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqrx_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3490,8 +3532,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqrx_n_mul_mont_384_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x05,0x00
@@ -3503,7 +3546,8 @@
.byte 0x00,0x74,0x0c,0x00
.byte 0x00,0x64,0x0d,0x00
.byte 0x00,0xa2
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqrx_n_mul_mont_384_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3514,8 +3558,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqrx_n_mul_mont_383_body:
.byte 1,0,17,0
.byte 0x00,0xf4,0x05,0x00
@@ -3527,7 +3572,8 @@
.byte 0x00,0x74,0x0c,0x00
.byte 0x00,0x64,0x0d,0x00
.byte 0x00,0xa2
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqrx_n_mul_mont_383_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
@@ -3538,8 +3584,9 @@
.byte 1,0,5,0x0b
.byte 0,0x74,1,0
.byte 0,0x64,2,0
-.byte 0,0x03
+.byte 0,0xb3
.byte 0,0
+.long 0,0
.LSEH_info_sqrx_mont_382x_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x11,0x00
@@ -3551,6 +3598,8 @@
.byte 0x00,0x74,0x18,0x00
.byte 0x00,0x64,0x19,0x00
.byte 0x00,0x01,0x17,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_sqrx_mont_382x_epilogue:
.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
diff --git a/c-source/build/coff/sha256-armv8.S b/c-source/build/coff/sha256-armv8.S
--- a/c-source/build/coff/sha256-armv8.S
+++ b/c-source/build/coff/sha256-armv8.S
@@ -10,11 +10,12 @@
//
// sha256_block procedure for ARMv8.
//
-// This module is stripped of scalar code paths, with raionale that all
+// This module is stripped of scalar code paths, with rationale that all
// known processors are NEON-capable.
//
// See original module at CRYPTOGAMS for further details.
+.comm __blst_platform_cap,4
.text
.p2align 6
@@ -188,6 +189,11 @@
.endef
.p2align 4
blst_sha256_block_data_order:
+ adrp x16,__blst_platform_cap
+ ldr w16,[x16,#:lo12:__blst_platform_cap]
+ tst w16,#1
+ b.ne .Lv8_entry
+
stp x29, x30, [sp, #-16]!
mov x29, sp
sub sp,sp,#16*4
diff --git a/c-source/build/coff/sha256-portable-x86_64.s b/c-source/build/coff/sha256-portable-x86_64.s
--- a/c-source/build/coff/sha256-portable-x86_64.s
+++ b/c-source/build/coff/sha256-portable-x86_64.s
@@ -1,3 +1,4 @@
+.comm __blst_platform_cap,4
.text
.globl blst_sha256_block_data_order
@@ -9,15 +10,21 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_blst_sha256_block_data_order:
+
+
+ pushq %rbp
+
+ movq %rsp,%rbp
+
movq %rcx,%rdi
movq %rdx,%rsi
movq %r8,%rdx
-
-
+#ifdef __BLST_PORTABLE__
+ testl $2,__blst_platform_cap(%rip)
+ jnz .Lblst_sha256_block_data_order$2
+#endif
pushq %rbx
- pushq %rbp
-
pushq %r12
pushq %r13
@@ -29,13 +36,14 @@
shlq $4,%rdx
subq $64+24,%rsp
+
+.LSEH_body_blst_sha256_block_data_order:
+
leaq (%rsi,%rdx,4),%rdx
movq %rdi,64+0(%rsp)
movq %rsi,64+8(%rsp)
movq %rdx,64+16(%rsp)
-.LSEH_body_blst_sha256_block_data_order:
-
movl 0(%rdi),%eax
movl 4(%rdi),%ebx
movl 8(%rdi),%ecx
@@ -1637,17 +1645,11 @@
leaq 64+24+48(%rsp),%r11
movq 64+24(%rsp),%r15
-
movq -40(%r11),%r14
-
movq -32(%r11),%r13
-
movq -24(%r11),%r12
-
- movq -16(%r11),%rbp
-
- movq -8(%r11),%rbx
-
+ movq -16(%r11),%rbx
+ movq -8(%r11),%rbp
.LSEH_epilogue_blst_sha256_block_data_order:
mov 8(%r11),%rdi
mov 16(%r11),%rsi
@@ -1657,6 +1659,7 @@
.LSEH_end_blst_sha256_block_data_order:
+#ifndef __BLST_PORTABLE__
.p2align 6
K256:
@@ -1742,6 +1745,7 @@
movq %r11,24(%rcx)
.byte 0xf3,0xc3
+#endif
.section .pdata
.p2align 2
.rva .LSEH_begin_blst_sha256_block_data_order
@@ -1759,26 +1763,30 @@
.section .xdata
.p2align 3
.LSEH_info_blst_sha256_block_data_order_prologue:
-.byte 1,0,5,0x0b
-.byte 0,0x74,1,0
-.byte 0,0x64,2,0
-.byte 0,0x03
-.byte 0,0
+.byte 1,4,6,0x05
+.byte 4,0x74,2,0
+.byte 4,0x64,3,0
+.byte 4,0x53
+.byte 1,0x50
+.long 0,0
.LSEH_info_blst_sha256_block_data_order_body:
.byte 1,0,18,0
.byte 0x00,0xf4,0x0b,0x00
.byte 0x00,0xe4,0x0c,0x00
.byte 0x00,0xd4,0x0d,0x00
.byte 0x00,0xc4,0x0e,0x00
-.byte 0x00,0x54,0x0f,0x00
-.byte 0x00,0x34,0x10,0x00
+.byte 0x00,0x34,0x0f,0x00
+.byte 0x00,0x54,0x10,0x00
.byte 0x00,0x74,0x12,0x00
.byte 0x00,0x64,0x13,0x00
.byte 0x00,0x01,0x11,0x00
+.byte 0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_blst_sha256_block_data_order_epilogue:
.byte 1,0,5,11
.byte 0x00,0x74,0x01,0x00
.byte 0x00,0x64,0x02,0x00
-.byte 0x00,0x03
-.byte 0x00,0x00
+.byte 0x00,0xb3
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
diff --git a/c-source/build/coff/sha256-x86_64.s b/c-source/build/coff/sha256-x86_64.s
--- a/c-source/build/coff/sha256-x86_64.s
+++ b/c-source/build/coff/sha256-x86_64.s
@@ -1,3 +1,4 @@
+.comm __blst_platform_cap,4
.text
.p2align 6
@@ -34,22 +35,23 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_blst_sha256_block_data_order_shaext:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
- subq $0x58,%rsp
-
- movaps %xmm6,-88(%r11)
-
- movaps %xmm7,-72(%r11)
+ pushq %rbp
- movaps %xmm8,-56(%r11)
+ movq %rsp,%rbp
- movaps %xmm9,-40(%r11)
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+.Lblst_sha256_block_data_order$2:
+ subq $0x50,%rsp
- movaps %xmm10,-24(%r11)
+ movaps %xmm6,-80(%rbp)
+ movaps %xmm7,-64(%rbp)
+ movaps %xmm8,-48(%rbp)
+ movaps %xmm9,-32(%rbp)
+ movaps %xmm10,-16(%rbp)
.LSEH_body_blst_sha256_block_data_order_shaext:
@@ -254,16 +256,18 @@
movdqu %xmm1,(%rdi)
movdqu %xmm2,16(%rdi)
- movaps -88(%r11),%xmm6
- movaps -72(%r11),%xmm7
- movaps -56(%r11),%xmm8
- movaps -40(%r11),%xmm9
- movaps -24(%r11),%xmm10
- movq %r11,%rsp
+ movaps -80(%rbp),%xmm6
+ movaps -64(%rbp),%xmm7
+ movaps -48(%rbp),%xmm8
+ movaps -32(%rbp),%xmm9
+ movaps -16(%rbp),%xmm10
+ movq %rbp,%rsp
+ popq %rbp
+
.LSEH_epilogue_blst_sha256_block_data_order_shaext:
- mov 8(%r11),%rdi
- mov 16(%r11),%rsi
+ mov 8(%rsp),%rdi
+ mov 16(%rsp),%rsi
.byte 0xf3,0xc3
@@ -278,13 +282,17 @@
movq %rsi,16(%rsp)
movq %rsp,%r11
.LSEH_begin_blst_sha256_block_data_order:
- movq %rcx,%rdi
- movq %rdx,%rsi
- movq %r8,%rdx
pushq %rbp
+ movq %rsp,%rbp
+
+ movq %rcx,%rdi
+ movq %rdx,%rsi
+ movq %r8,%rdx
+ testl $2,__blst_platform_cap(%rip)
+ jnz .Lblst_sha256_block_data_order$2
pushq %rbx
pushq %r12
@@ -296,21 +304,16 @@
pushq %r15
shlq $4,%rdx
- subq $104,%rsp
+ subq $88,%rsp
leaq (%rsi,%rdx,4),%rdx
- movq %rdi,0(%rsp)
-
- movq %rdx,16(%rsp)
- movaps %xmm6,32(%rsp)
-
- movaps %xmm7,48(%rsp)
-
- movaps %xmm8,64(%rsp)
-
- movaps %xmm9,80(%rsp)
+ movq %rdi,-64(%rbp)
- movq %rsp,%rbp
+ movq %rdx,-48(%rbp)
+ movaps %xmm6,-128(%rbp)
+ movaps %xmm7,-112(%rbp)
+ movaps %xmm8,-96(%rbp)
+ movaps %xmm9,-80(%rbp)
.LSEH_body_blst_sha256_block_data_order:
@@ -331,7 +334,7 @@
.p2align 4
.Lloop_ssse3:
movdqa K256+256(%rip),%xmm7
- movq %rsi,8(%rbp)
+ movq %rsi,-56(%rbp)
movdqu 0(%rsi),%xmm0
movdqu 16(%rsi),%xmm1
movdqu 32(%rsi),%xmm2
@@ -1356,9 +1359,9 @@
addl %r15d,%eax
movl %r8d,%r13d
addl %eax,%r14d
- movq 0(%rbp),%rdi
+ movq -64(%rbp),%rdi
movl %r14d,%eax
- movq 8(%rbp),%rsi
+ movq -56(%rbp),%rsi
addl 0(%rdi),%eax
addl 4(%rdi),%ebx
@@ -1370,7 +1373,7 @@
addl 28(%rdi),%r11d
leaq 64(%rsi),%rsi
- cmpq 16(%rbp),%rsi
+ cmpq -48(%rbp),%rsi
movl %eax,0(%rdi)
movl %ebx,4(%rdi)
@@ -1383,33 +1386,27 @@
jb .Lloop_ssse3
xorps %xmm0,%xmm0
- leaq 104+48(%rbp),%r11
-
movaps %xmm0,0(%rsp)
movaps %xmm0,16(%rsp)
movaps %xmm0,32(%rsp)
movaps %xmm0,48(%rsp)
- movaps 32(%rbp),%xmm6
- movaps 48(%rbp),%xmm7
- movaps 64(%rbp),%xmm8
- movaps 80(%rbp),%xmm9
- movq 104(%rbp),%r15
-
- movq -40(%r11),%r14
-
- movq -32(%r11),%r13
-
- movq -24(%r11),%r12
-
- movq -16(%r11),%rbx
+ movaps -128(%rbp),%xmm6
+ movaps -112(%rbp),%xmm7
+ movaps -96(%rbp),%xmm8
+ movaps -80(%rbp),%xmm9
+ movq -40(%rbp),%r15
+ movq -32(%rbp),%r14
+ movq -24(%rbp),%r13
+ movq -16(%rbp),%r12
+ movq -8(%rbp),%rbx
+ movq %rbp,%rsp
- movq -8(%r11),%rbp
+ popq %rbp
.LSEH_epilogue_blst_sha256_block_data_order:
- mov 8(%r11),%rdi
- mov 16(%r11),%rsi
+ mov 8(%rsp),%rdi
+ mov 16(%rsp),%rsi
- leaq (%r11),%rsp
.byte 0xf3,0xc3
.LSEH_end_blst_sha256_block_data_order:
@@ -1506,13 +1503,14 @@
.section .xdata
.p2align 3
.LSEH_info_blst_sha256_block_data_order_shaext_prologue:
-.byte 1,0,5,0x0b
-.byte 0,0x74,1,0
-.byte 0,0x64,2,0
-.byte 0,0x03
-.byte 0,0
+.byte 1,4,6,0x05
+.byte 4,0x74,2,0
+.byte 4,0x64,3,0
+.byte 4,0x53
+.byte 1,0x50
+.long 0,0
.LSEH_info_blst_sha256_block_data_order_shaext_body:
-.byte 1,0,15,0
+.byte 1,0,17,85
.byte 0x00,0x68,0x00,0x00
.byte 0x00,0x78,0x01,0x00
.byte 0x00,0x88,0x02,0x00
@@ -1520,41 +1518,45 @@
.byte 0x00,0xa8,0x04,0x00
.byte 0x00,0x74,0x0c,0x00
.byte 0x00,0x64,0x0d,0x00
-.byte 0x00,0xa2
+.byte 0x00,0x53
+.byte 0x00,0x92
+.byte 0x00,0x50
.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_blst_sha256_block_data_order_shaext_epilogue:
-.byte 1,0,5,11
+.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
.byte 0x00,0x64,0x02,0x00
-.byte 0x00,0x03
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_blst_sha256_block_data_order_prologue:
-.byte 1,0,5,0x0b
-.byte 0,0x74,1,0
-.byte 0,0x64,2,0
-.byte 0,0x03
-.byte 0,0
+.byte 1,4,6,0x05
+.byte 4,0x74,2,0
+.byte 4,0x64,3,0
+.byte 4,0x53
+.byte 1,0x50
+.long 0,0
.LSEH_info_blst_sha256_block_data_order_body:
-.byte 1,0,26,5
-.byte 0x00,0x68,0x02,0x00
-.byte 0x00,0x78,0x03,0x00
-.byte 0x00,0x88,0x04,0x00
-.byte 0x00,0x98,0x05,0x00
-.byte 0x00,0xf4,0x0d,0x00
-.byte 0x00,0xe4,0x0e,0x00
-.byte 0x00,0xd4,0x0f,0x00
-.byte 0x00,0xc4,0x10,0x00
-.byte 0x00,0x34,0x11,0x00
-.byte 0x00,0x74,0x14,0x00
-.byte 0x00,0x64,0x15,0x00
-.byte 0x00,0x03
-.byte 0x00,0x01,0x12,0x00
+.byte 1,0,25,133
+.byte 0x00,0x68,0x00,0x00
+.byte 0x00,0x78,0x01,0x00
+.byte 0x00,0x88,0x02,0x00
+.byte 0x00,0x98,0x03,0x00
+.byte 0x00,0xf4,0x0b,0x00
+.byte 0x00,0xe4,0x0c,0x00
+.byte 0x00,0xd4,0x0d,0x00
+.byte 0x00,0xc4,0x0e,0x00
+.byte 0x00,0x34,0x0f,0x00
+.byte 0x00,0x74,0x12,0x00
+.byte 0x00,0x64,0x13,0x00
+.byte 0x00,0x53
+.byte 0x00,0xf2
.byte 0x00,0x50
+.byte 0x00,0x00,0x00,0x00,0x00,0x00
+.byte 0x00,0x00,0x00,0x00
.LSEH_info_blst_sha256_block_data_order_epilogue:
-.byte 1,0,5,11
+.byte 1,0,4,0
.byte 0x00,0x74,0x01,0x00
.byte 0x00,0x64,0x02,0x00
-.byte 0x00,0x03
-.byte 0x00,0x00
+.byte 0x00,0x00,0x00,0x00
diff --git a/c-source/build/elf/add_mod_384-armv8.S b/c-source/build/elf/add_mod_384-armv8.S
--- a/c-source/build/elf/add_mod_384-armv8.S
+++ b/c-source/build/elf/add_mod_384-armv8.S
@@ -748,6 +748,21 @@
ret
.size sgn0_pty_mod_384x,.-sgn0_pty_mod_384x
+.globl vec_select_32
+.hidden vec_select_32
+.type vec_select_32,%function
+.align 5
+vec_select_32:
+ dup v6.2d, x3
+ ld1 {v0.2d, v1.2d, v2.2d}, [x1],#48
+ cmeq v6.2d, v6.2d, #0
+ ld1 {v3.2d, v4.2d, v5.2d}, [x2],#48
+ bit v0.16b, v3.16b, v6.16b
+ bit v1.16b, v4.16b, v6.16b
+ bit v2.16b, v5.16b, v6.16b
+ st1 {v0.2d, v1.2d, v2.2d}, [x0]
+ ret
+.size vec_select_32,.-vec_select_32
.globl vec_select_48
.hidden vec_select_48
.type vec_select_48,%function
@@ -929,3 +944,57 @@
prfm pldl1keep, [x0]
ret
.size vec_prefetch,.-vec_prefetch
+.globl vec_is_zero_16x
+.hidden vec_is_zero_16x
+.type vec_is_zero_16x,%function
+.align 5
+vec_is_zero_16x:
+ ld1 {v0.2d}, [x0], #16
+ lsr x1, x1, #4
+ sub x1, x1, #1
+ cbz x1, .Loop_is_zero_done
+
+.Loop_is_zero:
+ ld1 {v1.2d}, [x0], #16
+ orr v0.16b, v0.16b, v1.16b
+ sub x1, x1, #1
+ cbnz x1, .Loop_is_zero
+
+.Loop_is_zero_done:
+ dup v1.2d, v0.d[1]
+ orr v0.16b, v0.16b, v1.16b
+ mov x1, v0.d[0]
+ mov x0, #1
+ cmp x1, #0
+ csel x0, x0, xzr, eq
+ ret
+.size vec_is_zero_16x,.-vec_is_zero_16x
+.globl vec_is_equal_16x
+.hidden vec_is_equal_16x
+.type vec_is_equal_16x,%function
+.align 5
+vec_is_equal_16x:
+ ld1 {v0.2d}, [x0], #16
+ ld1 {v1.2d}, [x1], #16
+ lsr x2, x2, #4
+ eor v0.16b, v0.16b, v1.16b
+
+.Loop_is_equal:
+ sub x2, x2, #1
+ cbz x2, .Loop_is_equal_done
+ ld1 {v1.2d}, [x0], #16
+ ld1 {v2.2d}, [x1], #16
+ eor v1.16b, v1.16b, v2.16b
+ orr v0.16b, v0.16b, v1.16b
+ b .Loop_is_equal
+ nop
+
+.Loop_is_equal_done:
+ dup v1.2d, v0.d[1]
+ orr v0.16b, v0.16b, v1.16b
+ mov x1, v0.d[0]
+ mov x0, #1
+ cmp x1, #0
+ csel x0, x0, xzr, eq
+ ret
+.size vec_is_equal_16x,.-vec_is_equal_16x
diff --git a/c-source/build/elf/add_mod_384-x86_64.s b/c-source/build/elf/add_mod_384-x86_64.s
--- a/c-source/build/elf/add_mod_384-x86_64.s
+++ b/c-source/build/elf/add_mod_384-x86_64.s
@@ -1372,6 +1372,37 @@
.byte 0xf3,0xc3
.cfi_endproc
.size sgn0_pty_mod_384x,.-sgn0_pty_mod_384x
+.globl vec_select_32
+.hidden vec_select_32
+.type vec_select_32,@function
+.align 32
+vec_select_32:
+.cfi_startproc
+ .byte 0xf3,0x0f,0x1e,0xfa
+
+ movd %ecx,%xmm5
+ pxor %xmm4,%xmm4
+ pshufd $0,%xmm5,%xmm5
+ movdqu (%rsi),%xmm0
+ leaq 16(%rsi),%rsi
+ pcmpeqd %xmm4,%xmm5
+ movdqu (%rdx),%xmm1
+ leaq 16(%rdx),%rdx
+ pcmpeqd %xmm5,%xmm4
+ leaq 16(%rdi),%rdi
+ pand %xmm4,%xmm0
+ movdqu 0+16-16(%rsi),%xmm2
+ pand %xmm5,%xmm1
+ movdqu 0+16-16(%rdx),%xmm3
+ por %xmm1,%xmm0
+ movdqu %xmm0,0-16(%rdi)
+ pand %xmm4,%xmm2
+ pand %xmm5,%xmm3
+ por %xmm3,%xmm2
+ movdqu %xmm2,16-16(%rdi)
+ .byte 0xf3,0xc3
+.cfi_endproc
+.size vec_select_32,.-vec_select_32
.globl vec_select_48
.hidden vec_select_48
.type vec_select_48,@function
@@ -1799,6 +1830,73 @@
.byte 0xf3,0xc3
.cfi_endproc
.size vec_prefetch,.-vec_prefetch
+.globl vec_is_zero_16x
+.hidden vec_is_zero_16x
+.type vec_is_zero_16x,@function
+.align 32
+vec_is_zero_16x:
+.cfi_startproc
+ .byte 0xf3,0x0f,0x1e,0xfa
+
+ shrl $4,%esi
+ movdqu (%rdi),%xmm0
+ leaq 16(%rdi),%rdi
+
+.Loop_is_zero:
+ decl %esi
+ jz .Loop_is_zero_done
+ movdqu (%rdi),%xmm1
+ leaq 16(%rdi),%rdi
+ por %xmm1,%xmm0
+ jmp .Loop_is_zero
+
+.Loop_is_zero_done:
+ pshufd $0x4e,%xmm0,%xmm1
+ por %xmm1,%xmm0
+.byte 102,72,15,126,192
+ incl %esi
+ testq %rax,%rax
+ cmovnzl %esi,%eax
+ xorl $1,%eax
+ .byte 0xf3,0xc3
+.cfi_endproc
+.size vec_is_zero_16x,.-vec_is_zero_16x
+.globl vec_is_equal_16x
+.hidden vec_is_equal_16x
+.type vec_is_equal_16x,@function
+.align 32
+vec_is_equal_16x:
+.cfi_startproc
+ .byte 0xf3,0x0f,0x1e,0xfa
+
+ shrl $4,%edx
+ movdqu (%rdi),%xmm0
+ movdqu (%rsi),%xmm1
+ subq %rdi,%rsi
+ leaq 16(%rdi),%rdi
+ pxor %xmm1,%xmm0
+
+.Loop_is_equal:
+ decl %edx
+ jz .Loop_is_equal_done
+ movdqu (%rdi),%xmm1
+ movdqu (%rdi,%rsi,1),%xmm2
+ leaq 16(%rdi),%rdi
+ pxor %xmm2,%xmm1
+ por %xmm1,%xmm0
+ jmp .Loop_is_equal
+
+.Loop_is_equal_done:
+ pshufd $0x4e,%xmm0,%xmm1
+ por %xmm1,%xmm0
+.byte 102,72,15,126,192
+ incl %edx
+ testq %rax,%rax
+ cmovnzl %edx,%eax
+ xorl $1,%eax
+ .byte 0xf3,0xc3
+.cfi_endproc
+.size vec_is_equal_16x,.-vec_is_equal_16x
.section .note.GNU-stack,"",@progbits
.section .note.gnu.property,"a",@note
diff --git a/c-source/build/elf/ct_inverse_mod_256-armv8.S b/c-source/build/elf/ct_inverse_mod_256-armv8.S
--- a/c-source/build/elf/ct_inverse_mod_256-armv8.S
+++ b/c-source/build/elf/ct_inverse_mod_256-armv8.S
@@ -1,6 +1,7 @@
.text
.globl ct_inverse_mod_256
+.hidden ct_inverse_mod_256
.type ct_inverse_mod_256, %function
.align 5
ct_inverse_mod_256:
@@ -60,14 +61,14 @@
madd x4, x16, x8, xzr // |u|*|f0|
madd x4, x17, x9, x4 // |v|*|g0|
str x4, [x0,#8*4]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*5]
stp x5, x5, [x0,#8*7]
madd x4, x12, x8, xzr // |u|*|f1|
madd x4, x13, x9, x4 // |v|*|g1|
str x4, [x0,#8*9]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*10]
stp x5, x5, [x0,#8*12]
eor x1, x1, #256 // flip-flop src |a|b|u|v|
diff --git a/c-source/build/elf/ct_inverse_mod_256-x86_64.s b/c-source/build/elf/ct_inverse_mod_256-x86_64.s
--- a/c-source/build/elf/ct_inverse_mod_256-x86_64.s
+++ b/c-source/build/elf/ct_inverse_mod_256-x86_64.s
@@ -1,6 +1,7 @@
.text
.globl ct_inverse_mod_256
+.hidden ct_inverse_mod_256
.type ct_inverse_mod_256,@function
.align 32
ct_inverse_mod_256:
diff --git a/c-source/build/elf/ct_inverse_mod_384-armv8.S b/c-source/build/elf/ct_inverse_mod_384-armv8.S
--- a/c-source/build/elf/ct_inverse_mod_384-armv8.S
+++ b/c-source/build/elf/ct_inverse_mod_384-armv8.S
@@ -1,6 +1,7 @@
.text
.globl ct_inverse_mod_383
+.hidden ct_inverse_mod_383
.type ct_inverse_mod_383, %function
.align 5
ct_inverse_mod_383:
@@ -71,7 +72,7 @@
adds x3, x3, x5
adc x4, x4, x6
stp x3, x4, [x0,#8*6]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*8]
stp x5, x5, [x0,#8*10]
@@ -82,7 +83,7 @@
adds x3, x3, x5
adc x4, x4, x6
stp x3, x4, [x0,#8*12]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*14]
stp x5, x5, [x0,#8*16]
eor x1, x1, #256 // flip-flop src |a|b|u|v|
diff --git a/c-source/build/elf/ct_is_square_mod_384-armv8.S b/c-source/build/elf/ct_is_square_mod_384-armv8.S
--- a/c-source/build/elf/ct_is_square_mod_384-armv8.S
+++ b/c-source/build/elf/ct_is_square_mod_384-armv8.S
@@ -1,6 +1,7 @@
.text
.globl ct_is_square_mod_384
+.hidden ct_is_square_mod_384
.type ct_is_square_mod_384, %function
.align 5
ct_is_square_mod_384:
diff --git a/c-source/build/elf/ct_is_square_mod_384-x86_64.s b/c-source/build/elf/ct_is_square_mod_384-x86_64.s
--- a/c-source/build/elf/ct_is_square_mod_384-x86_64.s
+++ b/c-source/build/elf/ct_is_square_mod_384-x86_64.s
@@ -1,6 +1,7 @@
.text
.globl ct_is_square_mod_384
+.hidden ct_is_square_mod_384
.type ct_is_square_mod_384,@function
.align 32
ct_is_square_mod_384:
diff --git a/c-source/build/elf/ctq_inverse_mod_384-x86_64.s b/c-source/build/elf/ctq_inverse_mod_384-x86_64.s
--- a/c-source/build/elf/ctq_inverse_mod_384-x86_64.s
+++ b/c-source/build/elf/ctq_inverse_mod_384-x86_64.s
@@ -1,6 +1,8 @@
+.comm __blst_platform_cap,4
.text
.globl ct_inverse_mod_383
+.hidden ct_inverse_mod_383
.type ct_inverse_mod_383,@function
.align 32
ct_inverse_mod_383:
@@ -8,6 +10,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz ct_inverse_mod_383$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/elf/ctx_inverse_mod_384-x86_64.s b/c-source/build/elf/ctx_inverse_mod_384-x86_64.s
--- a/c-source/build/elf/ctx_inverse_mod_384-x86_64.s
+++ b/c-source/build/elf/ctx_inverse_mod_384-x86_64.s
@@ -1,6 +1,7 @@
.text
.globl ctx_inverse_mod_383
+.hidden ctx_inverse_mod_383
.type ctx_inverse_mod_383,@function
.align 32
ctx_inverse_mod_383:
@@ -8,6 +9,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+ct_inverse_mod_383$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -810,7 +812,7 @@
movq 48(%rsi),%r10
- call __inner_loop_62
+ call __tail_loop_53
@@ -1521,9 +1523,9 @@
.cfi_endproc
.size __inner_loop_31,.-__inner_loop_31
-.type __inner_loop_62,@function
+.type __tail_loop_53,@function
.align 32
-__inner_loop_62:
+__tail_loop_53:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -1532,7 +1534,7 @@
xorq %r12,%r12
movq $1,%r13
-.Loop_62:
+.Loop_53:
xorq %rax,%rax
testq $1,%r8
movq %r10,%rbx
@@ -1559,11 +1561,11 @@
subq %rax,%rdx
subq %rbx,%rcx
subl $1,%edi
- jnz .Loop_62
+ jnz .Loop_53
.byte 0xf3,0xc3
.cfi_endproc
-.size __inner_loop_62,.-__inner_loop_62
+.size __tail_loop_53,.-__tail_loop_53
.section .note.GNU-stack,"",@progbits
.section .note.gnu.property,"a",@note
diff --git a/c-source/build/elf/div3w-armv8.S b/c-source/build/elf/div3w-armv8.S
--- a/c-source/build/elf/div3w-armv8.S
+++ b/c-source/build/elf/div3w-armv8.S
@@ -25,7 +25,7 @@
asr x3,x0,#63 // top bit -> mask
add x0,x0,x0 // Q <<= 1
subs x6,x4,x1 // R - D
- add x0,x0,#1 // Q + specilative bit
+ add x0,x0,#1 // Q + speculative bit
sbcs x7,x5,x2
sbc x0,x0,xzr // subtract speculative bit
diff --git a/c-source/build/elf/div3w-x86_64.s b/c-source/build/elf/div3w-x86_64.s
--- a/c-source/build/elf/div3w-x86_64.s
+++ b/c-source/build/elf/div3w-x86_64.s
@@ -8,6 +8,8 @@
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
+
+
movq (%rdi),%r8
movq 8(%rdi),%r9
xorq %rax,%rax
@@ -39,8 +41,9 @@
orq %rcx,%rax
+
.byte 0xf3,0xc3
-.cfi_endproc
+.cfi_endproc
.size div_3_limbs,.-div_3_limbs
.globl quot_rem_128
.hidden quot_rem_128
@@ -50,6 +53,8 @@
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
+
+
movq %rdx,%rax
movq %rdx,%rcx
@@ -84,8 +89,9 @@
movq %rcx,%rax
+
.byte 0xf3,0xc3
-.cfi_endproc
+.cfi_endproc
.size quot_rem_128,.-quot_rem_128
@@ -100,6 +106,8 @@
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
+
+
movq %rdx,%rax
imulq 0(%rsi),%rdx
@@ -110,8 +118,9 @@
movq %r10,0(%rdi)
movq %rax,8(%rdi)
+
.byte 0xf3,0xc3
-.cfi_endproc
+.cfi_endproc
.size quot_rem_64,.-quot_rem_64
.section .note.GNU-stack,"",@progbits
diff --git a/c-source/build/elf/mulq_mont_256-x86_64.s b/c-source/build/elf/mulq_mont_256-x86_64.s
--- a/c-source/build/elf/mulq_mont_256-x86_64.s
+++ b/c-source/build/elf/mulq_mont_256-x86_64.s
@@ -1,3 +1,4 @@
+.comm __blst_platform_cap,4
.text
.globl mul_mont_sparse_256
@@ -9,6 +10,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_mont_sparse_256$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -73,6 +78,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_mont_sparse_256$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -419,6 +428,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz from_mont_256$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -494,6 +507,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz redc_mont_256$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/elf/mulq_mont_384-x86_64.s b/c-source/build/elf/mulq_mont_384-x86_64.s
--- a/c-source/build/elf/mulq_mont_384-x86_64.s
+++ b/c-source/build/elf/mulq_mont_384-x86_64.s
@@ -1,3 +1,4 @@
+.comm __blst_platform_cap,4
.text
@@ -6,9 +7,9 @@
-.type __sub_mod_384x384,@function
+.type __subq_mod_384x384,@function
.align 32
-__sub_mod_384x384:
+__subq_mod_384x384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -73,11 +74,11 @@
.byte 0xf3,0xc3
.cfi_endproc
-.size __sub_mod_384x384,.-__sub_mod_384x384
+.size __subq_mod_384x384,.-__subq_mod_384x384
-.type __add_mod_384,@function
+.type __addq_mod_384,@function
.align 32
-__add_mod_384:
+__addq_mod_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -125,11 +126,11 @@
.byte 0xf3,0xc3
.cfi_endproc
-.size __add_mod_384,.-__add_mod_384
+.size __addq_mod_384,.-__addq_mod_384
-.type __sub_mod_384,@function
+.type __subq_mod_384,@function
.align 32
-__sub_mod_384:
+__subq_mod_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -140,7 +141,7 @@
movq 32(%rsi),%r12
movq 40(%rsi),%r13
-__sub_mod_384_a_is_loaded:
+__subq_mod_384_a_is_loaded:
subq 0(%rdx),%r8
movq 0(%rcx),%r14
sbbq 8(%rdx),%r9
@@ -177,7 +178,7 @@
.byte 0xf3,0xc3
.cfi_endproc
-.size __sub_mod_384,.-__sub_mod_384
+.size __subq_mod_384,.-__subq_mod_384
.globl mul_mont_384x
.hidden mul_mont_384x
.type mul_mont_384x,@function
@@ -187,6 +188,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_mont_384x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -232,12 +237,12 @@
movq 8(%rsp),%rcx
leaq -48(%rsi),%rdx
leaq 40+192+48(%rsp),%rdi
- call __add_mod_384
+ call __addq_mod_384
movq 16(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq -48(%rdi),%rdi
- call __add_mod_384
+ call __addq_mod_384
leaq (%rdi),%rbx
leaq 48(%rdi),%rsi
@@ -247,17 +252,17 @@
leaq (%rdi),%rsi
leaq 40(%rsp),%rdx
movq 8(%rsp),%rcx
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq (%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq 40(%rsp),%rsi
leaq 40+96(%rsp),%rdx
leaq 40(%rsp),%rdi
- call __sub_mod_384x384
+ call __subq_mod_384x384
movq %rcx,%rbx
@@ -266,14 +271,14 @@
movq 0(%rsp),%rcx
movq 32(%rsp),%rdi
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
leaq 40+192(%rsp),%rsi
movq 0(%rsp),%rcx
leaq 48(%rdi),%rdi
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
leaq 328(%rsp),%r8
movq 0(%r8),%r15
@@ -303,6 +308,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_mont_384x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -333,13 +342,13 @@
leaq 48(%rsi),%rdx
leaq 32(%rsp),%rdi
- call __add_mod_384
+ call __addq_mod_384
movq 16(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq 32+48(%rsp),%rdi
- call __sub_mod_384
+ call __subq_mod_384
movq 16(%rsp),%rsi
@@ -427,6 +436,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_382x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -521,18 +534,18 @@
leaq 32(%rsp),%rdx
movq 24(%rsp),%rcx
movq %rsi,%rdi
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq 0(%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq -96(%rdi),%rsi
leaq 32(%rsp),%rdx
leaq -96(%rdi),%rdi
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq 136(%rsp),%r8
movq 0(%r8),%r15
@@ -562,6 +575,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_382x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -617,7 +634,7 @@
leaq 48(%rsi),%rdx
leaq 48(%rdi),%rdi
- call __sub_mod_384_a_is_loaded
+ call __subq_mod_384_a_is_loaded
leaq (%rdi),%rsi
@@ -695,6 +712,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1019,6 +1040,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1265,6 +1290,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1299,7 +1328,7 @@
movq 104(%rsp),%rbx
movq 112(%rsp),%rdi
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
leaq 120(%rsp),%r8
movq 120(%rsp),%r15
@@ -1332,6 +1361,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz redc_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1356,7 +1389,7 @@
movq %rdx,%rbx
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
movq 8(%rsp),%r15
.cfi_restore %r15
@@ -1389,6 +1422,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz from_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1762,9 +1799,9 @@
.cfi_endproc
.size __mulq_by_1_mont_384,.-__mulq_by_1_mont_384
-.type __redc_tail_mont_384,@function
+.type __redq_tail_mont_384,@function
.align 32
-__redc_tail_mont_384:
+__redq_tail_mont_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -1809,7 +1846,7 @@
.byte 0xf3,0xc3
.cfi_endproc
-.size __redc_tail_mont_384,.-__redc_tail_mont_384
+.size __redq_tail_mont_384,.-__redq_tail_mont_384
.globl sgn0_pty_mont_384
.hidden sgn0_pty_mont_384
@@ -1820,6 +1857,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sgn0_pty_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1898,6 +1939,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sgn0_pty_mont_384x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2025,6 +2070,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz mul_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2689,6 +2738,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_n_mul_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2727,7 +2780,7 @@
movq 0(%rsp),%rcx
movq 16(%rsp),%rbx
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
movd %xmm1,%edx
leaq 0(%rdi),%rsi
@@ -2777,6 +2830,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_n_mul_mont_383$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -3438,6 +3495,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,__blst_platform_cap(%rip)
+ jnz sqr_mont_382x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/elf/mulx_mont_256-x86_64.s b/c-source/build/elf/mulx_mont_256-x86_64.s
--- a/c-source/build/elf/mulx_mont_256-x86_64.s
+++ b/c-source/build/elf/mulx_mont_256-x86_64.s
@@ -9,6 +9,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_mont_sparse_256$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -71,6 +72,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_mont_sparse_256$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -332,6 +334,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+from_mont_256$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -407,6 +410,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+redc_mont_256$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/elf/mulx_mont_384-x86_64.s b/c-source/build/elf/mulx_mont_384-x86_64.s
--- a/c-source/build/elf/mulx_mont_384-x86_64.s
+++ b/c-source/build/elf/mulx_mont_384-x86_64.s
@@ -6,9 +6,9 @@
-.type __sub_mod_384x384,@function
+.type __subx_mod_384x384,@function
.align 32
-__sub_mod_384x384:
+__subx_mod_384x384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -73,11 +73,11 @@
.byte 0xf3,0xc3
.cfi_endproc
-.size __sub_mod_384x384,.-__sub_mod_384x384
+.size __subx_mod_384x384,.-__subx_mod_384x384
-.type __add_mod_384,@function
+.type __addx_mod_384,@function
.align 32
-__add_mod_384:
+__addx_mod_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -125,11 +125,11 @@
.byte 0xf3,0xc3
.cfi_endproc
-.size __add_mod_384,.-__add_mod_384
+.size __addx_mod_384,.-__addx_mod_384
-.type __sub_mod_384,@function
+.type __subx_mod_384,@function
.align 32
-__sub_mod_384:
+__subx_mod_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -140,7 +140,7 @@
movq 32(%rsi),%r12
movq 40(%rsi),%r13
-__sub_mod_384_a_is_loaded:
+__subx_mod_384_a_is_loaded:
subq 0(%rdx),%r8
movq 0(%rcx),%r14
sbbq 8(%rdx),%r9
@@ -177,7 +177,7 @@
.byte 0xf3,0xc3
.cfi_endproc
-.size __sub_mod_384,.-__sub_mod_384
+.size __subx_mod_384,.-__subx_mod_384
.globl mulx_mont_384x
.hidden mulx_mont_384x
.type mulx_mont_384x,@function
@@ -187,6 +187,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_mont_384x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -233,12 +234,12 @@
leaq (%rbx),%rsi
leaq -48(%rbx),%rdx
leaq 40+192+48(%rsp),%rdi
- call __add_mod_384
+ call __addx_mod_384
movq 24(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq -48(%rdi),%rdi
- call __add_mod_384
+ call __addx_mod_384
leaq (%rdi),%rbx
leaq 48(%rdi),%rsi
@@ -248,17 +249,17 @@
leaq (%rdi),%rsi
leaq 40(%rsp),%rdx
movq 8(%rsp),%rcx
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq (%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq 40(%rsp),%rsi
leaq 40+96(%rsp),%rdx
leaq 40(%rsp),%rdi
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq (%rcx),%rbx
@@ -267,14 +268,14 @@
movq 0(%rsp),%rcx
movq 32(%rsp),%rdi
call __mulx_by_1_mont_384
- call __redc_tail_mont_384
+ call __redx_tail_mont_384
leaq 40+192(%rsp),%rsi
movq 0(%rsp),%rcx
leaq 48(%rdi),%rdi
call __mulx_by_1_mont_384
- call __redc_tail_mont_384
+ call __redx_tail_mont_384
leaq 328(%rsp),%r8
movq 0(%r8),%r15
@@ -304,6 +305,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_mont_384x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -335,13 +337,13 @@
leaq 48(%rsi),%rdx
leaq 32(%rsp),%rdi
- call __add_mod_384
+ call __addx_mod_384
movq 24(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq 32+48(%rsp),%rdi
- call __sub_mod_384
+ call __subx_mod_384
movq 24(%rsp),%rsi
@@ -439,6 +441,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_382x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -533,18 +536,18 @@
leaq 32(%rsp),%rdx
movq 24(%rsp),%rcx
movq %rsi,%rdi
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq 0(%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq -96(%rdi),%rsi
leaq 32(%rsp),%rdx
leaq -96(%rdi),%rdi
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq 136(%rsp),%r8
movq 0(%r8),%r15
@@ -574,6 +577,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_382x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -629,7 +633,7 @@
leaq 48(%rsi),%rdx
leaq 48(%rdi),%rdi
- call __sub_mod_384_a_is_loaded
+ call __subx_mod_384_a_is_loaded
leaq (%rdi),%rsi
@@ -707,6 +711,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -933,6 +938,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1127,6 +1133,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+redc_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1151,7 +1158,7 @@
movq %rdx,%rbx
call __mulx_by_1_mont_384
- call __redc_tail_mont_384
+ call __redx_tail_mont_384
movq 8(%rsp),%r15
.cfi_restore %r15
@@ -1184,6 +1191,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+from_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1448,9 +1456,9 @@
.cfi_endproc
.size __mulx_by_1_mont_384,.-__mulx_by_1_mont_384
-.type __redc_tail_mont_384,@function
+.type __redx_tail_mont_384,@function
.align 32
-__redc_tail_mont_384:
+__redx_tail_mont_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -1495,7 +1503,7 @@
.byte 0xf3,0xc3
.cfi_endproc
-.size __redc_tail_mont_384,.-__redc_tail_mont_384
+.size __redx_tail_mont_384,.-__redx_tail_mont_384
.globl sgn0x_pty_mont_384
.hidden sgn0x_pty_mont_384
@@ -1506,6 +1514,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sgn0_pty_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1584,6 +1593,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sgn0_pty_mont_384x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1711,6 +1721,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2178,6 +2189,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2245,6 +2257,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_n_mul_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2330,6 +2343,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_n_mul_mont_383$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2776,6 +2790,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_mont_382x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/elf/sha256-armv8.S b/c-source/build/elf/sha256-armv8.S
--- a/c-source/build/elf/sha256-armv8.S
+++ b/c-source/build/elf/sha256-armv8.S
@@ -10,11 +10,12 @@
//
// sha256_block procedure for ARMv8.
//
-// This module is stripped of scalar code paths, with raionale that all
+// This module is stripped of scalar code paths, with rationale that all
// known processors are NEON-capable.
//
// See original module at CRYPTOGAMS for further details.
+.comm __blst_platform_cap,4
.text
.align 6
@@ -184,6 +185,11 @@
.type blst_sha256_block_data_order,%function
.align 4
blst_sha256_block_data_order:
+ adrp x16,__blst_platform_cap
+ ldr w16,[x16,#:lo12:__blst_platform_cap]
+ tst w16,#1
+ b.ne .Lv8_entry
+
stp x29, x30, [sp, #-16]!
mov x29, sp
sub sp,sp,#16*4
diff --git a/c-source/build/elf/sha256-portable-x86_64.s b/c-source/build/elf/sha256-portable-x86_64.s
--- a/c-source/build/elf/sha256-portable-x86_64.s
+++ b/c-source/build/elf/sha256-portable-x86_64.s
@@ -1,3 +1,4 @@
+.comm __blst_platform_cap,4
.text
.globl blst_sha256_block_data_order
@@ -8,33 +9,35 @@
.byte 0xf3,0x0f,0x1e,0xfa
- pushq %rbx
-.cfi_adjust_cfa_offset 8
-.cfi_offset %rbx,-16
pushq %rbp
.cfi_adjust_cfa_offset 8
-.cfi_offset %rbp,-24
+.cfi_offset %rbp,-16
+ movq %rsp,%rbp
+.cfi_def_cfa_register %rbp
+#ifdef __BLST_PORTABLE__
+ testl $2,__blst_platform_cap(%rip)
+ jnz .Lblst_sha256_block_data_order$2
+#endif
+ pushq %rbx
+.cfi_offset %rbx,-24
pushq %r12
-.cfi_adjust_cfa_offset 8
.cfi_offset %r12,-32
pushq %r13
-.cfi_adjust_cfa_offset 8
.cfi_offset %r13,-40
pushq %r14
-.cfi_adjust_cfa_offset 8
.cfi_offset %r14,-48
pushq %r15
-.cfi_adjust_cfa_offset 8
.cfi_offset %r15,-56
shlq $4,%rdx
subq $64+24,%rsp
-.cfi_adjust_cfa_offset 16*4+3*8
+
+.cfi_def_cfa %rsp,144
+
leaq (%rsi,%rdx,4),%rdx
movq %rdi,64+0(%rsp)
movq %rsi,64+8(%rsp)
movq %rdx,64+16(%rsp)
-
movl 0(%rdi),%eax
movl 4(%rdi),%ebx
movl 8(%rdi),%ecx
@@ -1636,23 +1639,23 @@
leaq 64+24+48(%rsp),%r11
.cfi_def_cfa %r11,8
movq 64+24(%rsp),%r15
-.cfi_restore %r15
movq -40(%r11),%r14
-.cfi_restore %r14
movq -32(%r11),%r13
-.cfi_restore %r13
movq -24(%r11),%r12
+ movq -16(%r11),%rbx
+ movq -8(%r11),%rbp
.cfi_restore %r12
- movq -16(%r11),%rbp
+.cfi_restore %r13
+.cfi_restore %r14
+.cfi_restore %r15
.cfi_restore %rbp
- movq -8(%r11),%rbx
.cfi_restore %rbx
-
leaq (%r11),%rsp
.byte 0xf3,0xc3
.cfi_endproc
.size blst_sha256_block_data_order,.-blst_sha256_block_data_order
+#ifndef __BLST_PORTABLE__
.align 64
.type K256,@object
K256:
@@ -1744,6 +1747,7 @@
.byte 0xf3,0xc3
.cfi_endproc
.size blst_sha256_hcopy,.-blst_sha256_hcopy
+#endif
.section .note.GNU-stack,"",@progbits
.section .note.gnu.property,"a",@note
diff --git a/c-source/build/elf/sha256-x86_64.s b/c-source/build/elf/sha256-x86_64.s
--- a/c-source/build/elf/sha256-x86_64.s
+++ b/c-source/build/elf/sha256-x86_64.s
@@ -1,3 +1,4 @@
+.comm __blst_platform_cap,4
.text
.align 64
@@ -33,6 +34,13 @@
.byte 0xf3,0x0f,0x1e,0xfa
+ pushq %rbp
+.cfi_adjust_cfa_offset 8
+.cfi_offset %rbp,-16
+ movq %rsp,%rbp
+.cfi_def_cfa_register %rbp
+.Lblst_sha256_block_data_order$2:
+
leaq K256+128(%rip),%rcx
movdqu (%rdi),%xmm1
movdqu 16(%rdi),%xmm2
@@ -234,6 +242,11 @@
movdqu %xmm1,(%rdi)
movdqu %xmm2,16(%rdi)
+.cfi_def_cfa_register %rsp
+ popq %rbp
+.cfi_adjust_cfa_offset -8
+.cfi_restore %rbp
+
.byte 0xf3,0xc3
.cfi_endproc
.size blst_sha256_block_data_order_shaext,.-blst_sha256_block_data_order_shaext
@@ -249,30 +262,27 @@
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
+ movq %rsp,%rbp
+.cfi_def_cfa_register %rbp
+ testl $2,__blst_platform_cap(%rip)
+ jnz .Lblst_sha256_block_data_order$2
pushq %rbx
-.cfi_adjust_cfa_offset 8
.cfi_offset %rbx,-24
pushq %r12
-.cfi_adjust_cfa_offset 8
.cfi_offset %r12,-32
pushq %r13
-.cfi_adjust_cfa_offset 8
.cfi_offset %r13,-40
pushq %r14
-.cfi_adjust_cfa_offset 8
.cfi_offset %r14,-48
pushq %r15
-.cfi_adjust_cfa_offset 8
.cfi_offset %r15,-56
shlq $4,%rdx
- subq $40,%rsp
-.cfi_adjust_cfa_offset 40
+ subq $24,%rsp
+
leaq (%rsi,%rdx,4),%rdx
- movq %rdi,0(%rsp)
+ movq %rdi,-64(%rbp)
- movq %rdx,16(%rsp)
- movq %rsp,%rbp
-.cfi_def_cfa_register %rbp
+ movq %rdx,-48(%rbp)
leaq -64(%rsp),%rsp
@@ -291,7 +301,7 @@
.align 16
.Lloop_ssse3:
movdqa K256+256(%rip),%xmm7
- movq %rsi,8(%rbp)
+ movq %rsi,-56(%rbp)
movdqu 0(%rsi),%xmm0
movdqu 16(%rsi),%xmm1
movdqu 32(%rsi),%xmm2
@@ -1316,9 +1326,9 @@
addl %r15d,%eax
movl %r8d,%r13d
addl %eax,%r14d
- movq 0(%rbp),%rdi
+ movq -64(%rbp),%rdi
movl %r14d,%eax
- movq 8(%rbp),%rsi
+ movq -56(%rbp),%rsi
addl 0(%rdi),%eax
addl 4(%rdi),%ebx
@@ -1330,7 +1340,7 @@
addl 28(%rdi),%r11d
leaq 64(%rsi),%rsi
- cmpq 16(%rbp),%rsi
+ cmpq -48(%rbp),%rsi
movl %eax,0(%rdi)
movl %ebx,4(%rdi)
@@ -1343,26 +1353,25 @@
jb .Lloop_ssse3
xorps %xmm0,%xmm0
- leaq 40+48(%rbp),%r11
-.cfi_def_cfa %r11,8
movaps %xmm0,0(%rsp)
movaps %xmm0,16(%rsp)
movaps %xmm0,32(%rsp)
movaps %xmm0,48(%rsp)
- movq 40(%rbp),%r15
-.cfi_restore %r15
- movq -40(%r11),%r14
-.cfi_restore %r14
- movq -32(%r11),%r13
-.cfi_restore %r13
- movq -24(%r11),%r12
+ movq -40(%rbp),%r15
+ movq -32(%rbp),%r14
+ movq -24(%rbp),%r13
+ movq -16(%rbp),%r12
+ movq -8(%rbp),%rbx
+ movq %rbp,%rsp
+.cfi_def_cfa_register %rsp
+ popq %rbp
+.cfi_adjust_cfa_offset -8
+.cfi_restore %rbp
.cfi_restore %r12
- movq -16(%r11),%rbx
+.cfi_restore %r13
+.cfi_restore %r14
+.cfi_restore %r15
.cfi_restore %rbx
- movq -8(%r11),%rbp
-.cfi_restore %rbp
-
- leaq (%r11),%rsp
.byte 0xf3,0xc3
.cfi_endproc
.size blst_sha256_block_data_order,.-blst_sha256_block_data_order
diff --git a/c-source/build/mach-o/add_mod_384-armv8.S b/c-source/build/mach-o/add_mod_384-armv8.S
--- a/c-source/build/mach-o/add_mod_384-armv8.S
+++ b/c-source/build/mach-o/add_mod_384-armv8.S
@@ -748,6 +748,21 @@
ret
+.globl _vec_select_32
+.private_extern _vec_select_32
+
+.align 5
+_vec_select_32:
+ dup v6.2d, x3
+ ld1 {v0.2d, v1.2d, v2.2d}, [x1],#48
+ cmeq v6.2d, v6.2d, #0
+ ld1 {v3.2d, v4.2d, v5.2d}, [x2],#48
+ bit v0.16b, v3.16b, v6.16b
+ bit v1.16b, v4.16b, v6.16b
+ bit v2.16b, v5.16b, v6.16b
+ st1 {v0.2d, v1.2d, v2.2d}, [x0]
+ ret
+
.globl _vec_select_48
.private_extern _vec_select_48
@@ -927,5 +942,59 @@
cmp x0, x1
csel x0, x1, x0, hi
prfm pldl1keep, [x0]
+ ret
+
+.globl _vec_is_zero_16x
+.private_extern _vec_is_zero_16x
+
+.align 5
+_vec_is_zero_16x:
+ ld1 {v0.2d}, [x0], #16
+ lsr x1, x1, #4
+ sub x1, x1, #1
+ cbz x1, Loop_is_zero_done
+
+Loop_is_zero:
+ ld1 {v1.2d}, [x0], #16
+ orr v0.16b, v0.16b, v1.16b
+ sub x1, x1, #1
+ cbnz x1, Loop_is_zero
+
+Loop_is_zero_done:
+ dup v1.2d, v0.d[1]
+ orr v0.16b, v0.16b, v1.16b
+ mov x1, v0.d[0]
+ mov x0, #1
+ cmp x1, #0
+ csel x0, x0, xzr, eq
+ ret
+
+.globl _vec_is_equal_16x
+.private_extern _vec_is_equal_16x
+
+.align 5
+_vec_is_equal_16x:
+ ld1 {v0.2d}, [x0], #16
+ ld1 {v1.2d}, [x1], #16
+ lsr x2, x2, #4
+ eor v0.16b, v0.16b, v1.16b
+
+Loop_is_equal:
+ sub x2, x2, #1
+ cbz x2, Loop_is_equal_done
+ ld1 {v1.2d}, [x0], #16
+ ld1 {v2.2d}, [x1], #16
+ eor v1.16b, v1.16b, v2.16b
+ orr v0.16b, v0.16b, v1.16b
+ b Loop_is_equal
+ nop
+
+Loop_is_equal_done:
+ dup v1.2d, v0.d[1]
+ orr v0.16b, v0.16b, v1.16b
+ mov x1, v0.d[0]
+ mov x0, #1
+ cmp x1, #0
+ csel x0, x0, xzr, eq
ret
diff --git a/c-source/build/mach-o/add_mod_384-x86_64.s b/c-source/build/mach-o/add_mod_384-x86_64.s
--- a/c-source/build/mach-o/add_mod_384-x86_64.s
+++ b/c-source/build/mach-o/add_mod_384-x86_64.s
@@ -1372,6 +1372,37 @@
.byte 0xf3,0xc3
.cfi_endproc
+.globl _vec_select_32
+.private_extern _vec_select_32
+
+.p2align 5
+_vec_select_32:
+.cfi_startproc
+ .byte 0xf3,0x0f,0x1e,0xfa
+
+ movd %ecx,%xmm5
+ pxor %xmm4,%xmm4
+ pshufd $0,%xmm5,%xmm5
+ movdqu (%rsi),%xmm0
+ leaq 16(%rsi),%rsi
+ pcmpeqd %xmm4,%xmm5
+ movdqu (%rdx),%xmm1
+ leaq 16(%rdx),%rdx
+ pcmpeqd %xmm5,%xmm4
+ leaq 16(%rdi),%rdi
+ pand %xmm4,%xmm0
+ movdqu 0+16-16(%rsi),%xmm2
+ pand %xmm5,%xmm1
+ movdqu 0+16-16(%rdx),%xmm3
+ por %xmm1,%xmm0
+ movdqu %xmm0,0-16(%rdi)
+ pand %xmm4,%xmm2
+ pand %xmm5,%xmm3
+ por %xmm3,%xmm2
+ movdqu %xmm2,16-16(%rdi)
+ .byte 0xf3,0xc3
+.cfi_endproc
+
.globl _vec_select_48
.private_extern _vec_select_48
@@ -1796,6 +1827,73 @@
cmpq %rsi,%rdi
cmovaq %rsi,%rdi
prefetchnta (%rdi)
+ .byte 0xf3,0xc3
+.cfi_endproc
+
+.globl _vec_is_zero_16x
+.private_extern _vec_is_zero_16x
+
+.p2align 5
+_vec_is_zero_16x:
+.cfi_startproc
+ .byte 0xf3,0x0f,0x1e,0xfa
+
+ shrl $4,%esi
+ movdqu (%rdi),%xmm0
+ leaq 16(%rdi),%rdi
+
+L$oop_is_zero:
+ decl %esi
+ jz L$oop_is_zero_done
+ movdqu (%rdi),%xmm1
+ leaq 16(%rdi),%rdi
+ por %xmm1,%xmm0
+ jmp L$oop_is_zero
+
+L$oop_is_zero_done:
+ pshufd $0x4e,%xmm0,%xmm1
+ por %xmm1,%xmm0
+.byte 102,72,15,126,192
+ incl %esi
+ testq %rax,%rax
+ cmovnzl %esi,%eax
+ xorl $1,%eax
+ .byte 0xf3,0xc3
+.cfi_endproc
+
+.globl _vec_is_equal_16x
+.private_extern _vec_is_equal_16x
+
+.p2align 5
+_vec_is_equal_16x:
+.cfi_startproc
+ .byte 0xf3,0x0f,0x1e,0xfa
+
+ shrl $4,%edx
+ movdqu (%rdi),%xmm0
+ movdqu (%rsi),%xmm1
+ subq %rdi,%rsi
+ leaq 16(%rdi),%rdi
+ pxor %xmm1,%xmm0
+
+L$oop_is_equal:
+ decl %edx
+ jz L$oop_is_equal_done
+ movdqu (%rdi),%xmm1
+ movdqu (%rdi,%rsi,1),%xmm2
+ leaq 16(%rdi),%rdi
+ pxor %xmm2,%xmm1
+ por %xmm1,%xmm0
+ jmp L$oop_is_equal
+
+L$oop_is_equal_done:
+ pshufd $0x4e,%xmm0,%xmm1
+ por %xmm1,%xmm0
+.byte 102,72,15,126,192
+ incl %edx
+ testq %rax,%rax
+ cmovnzl %edx,%eax
+ xorl $1,%eax
.byte 0xf3,0xc3
.cfi_endproc
diff --git a/c-source/build/mach-o/ct_inverse_mod_256-armv8.S b/c-source/build/mach-o/ct_inverse_mod_256-armv8.S
--- a/c-source/build/mach-o/ct_inverse_mod_256-armv8.S
+++ b/c-source/build/mach-o/ct_inverse_mod_256-armv8.S
@@ -1,6 +1,7 @@
.text
.globl _ct_inverse_mod_256
+.private_extern _ct_inverse_mod_256
.align 5
_ct_inverse_mod_256:
@@ -60,14 +61,14 @@
madd x4, x16, x8, xzr // |u|*|f0|
madd x4, x17, x9, x4 // |v|*|g0|
str x4, [x0,#8*4]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*5]
stp x5, x5, [x0,#8*7]
madd x4, x12, x8, xzr // |u|*|f1|
madd x4, x13, x9, x4 // |v|*|g1|
str x4, [x0,#8*9]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*10]
stp x5, x5, [x0,#8*12]
eor x1, x1, #256 // flip-flop src |a|b|u|v|
diff --git a/c-source/build/mach-o/ct_inverse_mod_256-x86_64.s b/c-source/build/mach-o/ct_inverse_mod_256-x86_64.s
--- a/c-source/build/mach-o/ct_inverse_mod_256-x86_64.s
+++ b/c-source/build/mach-o/ct_inverse_mod_256-x86_64.s
@@ -1,6 +1,7 @@
.text
.globl _ct_inverse_mod_256
+.private_extern _ct_inverse_mod_256
.p2align 5
_ct_inverse_mod_256:
diff --git a/c-source/build/mach-o/ct_inverse_mod_384-armv8.S b/c-source/build/mach-o/ct_inverse_mod_384-armv8.S
--- a/c-source/build/mach-o/ct_inverse_mod_384-armv8.S
+++ b/c-source/build/mach-o/ct_inverse_mod_384-armv8.S
@@ -1,6 +1,7 @@
.text
.globl _ct_inverse_mod_383
+.private_extern _ct_inverse_mod_383
.align 5
_ct_inverse_mod_383:
@@ -71,7 +72,7 @@
adds x3, x3, x5
adc x4, x4, x6
stp x3, x4, [x0,#8*6]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*8]
stp x5, x5, [x0,#8*10]
@@ -82,7 +83,7 @@
adds x3, x3, x5
adc x4, x4, x6
stp x3, x4, [x0,#8*12]
- asr x5, x4, #63 // sign extenstion
+ asr x5, x4, #63 // sign extension
stp x5, x5, [x0,#8*14]
stp x5, x5, [x0,#8*16]
eor x1, x1, #256 // flip-flop src |a|b|u|v|
diff --git a/c-source/build/mach-o/ct_is_square_mod_384-armv8.S b/c-source/build/mach-o/ct_is_square_mod_384-armv8.S
--- a/c-source/build/mach-o/ct_is_square_mod_384-armv8.S
+++ b/c-source/build/mach-o/ct_is_square_mod_384-armv8.S
@@ -1,6 +1,7 @@
.text
.globl _ct_is_square_mod_384
+.private_extern _ct_is_square_mod_384
.align 5
_ct_is_square_mod_384:
diff --git a/c-source/build/mach-o/ct_is_square_mod_384-x86_64.s b/c-source/build/mach-o/ct_is_square_mod_384-x86_64.s
--- a/c-source/build/mach-o/ct_is_square_mod_384-x86_64.s
+++ b/c-source/build/mach-o/ct_is_square_mod_384-x86_64.s
@@ -1,6 +1,7 @@
.text
.globl _ct_is_square_mod_384
+.private_extern _ct_is_square_mod_384
.p2align 5
_ct_is_square_mod_384:
diff --git a/c-source/build/mach-o/ctq_inverse_mod_384-x86_64.s b/c-source/build/mach-o/ctq_inverse_mod_384-x86_64.s
--- a/c-source/build/mach-o/ctq_inverse_mod_384-x86_64.s
+++ b/c-source/build/mach-o/ctq_inverse_mod_384-x86_64.s
@@ -1,6 +1,8 @@
+.comm ___blst_platform_cap,4
.text
.globl _ct_inverse_mod_383
+.private_extern _ct_inverse_mod_383
.p2align 5
_ct_inverse_mod_383:
@@ -8,6 +10,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz ct_inverse_mod_383$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/mach-o/ctx_inverse_mod_384-x86_64.s b/c-source/build/mach-o/ctx_inverse_mod_384-x86_64.s
--- a/c-source/build/mach-o/ctx_inverse_mod_384-x86_64.s
+++ b/c-source/build/mach-o/ctx_inverse_mod_384-x86_64.s
@@ -1,6 +1,7 @@
.text
.globl _ctx_inverse_mod_383
+.private_extern _ctx_inverse_mod_383
.p2align 5
_ctx_inverse_mod_383:
@@ -8,6 +9,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+ct_inverse_mod_383$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -810,7 +812,7 @@
movq 48(%rsi),%r10
- call __inner_loop_62
+ call __tail_loop_53
@@ -1523,7 +1525,7 @@
.p2align 5
-__inner_loop_62:
+__tail_loop_53:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -1532,7 +1534,7 @@
xorq %r12,%r12
movq $1,%r13
-L$oop_62:
+L$oop_53:
xorq %rax,%rax
testq $1,%r8
movq %r10,%rbx
@@ -1559,7 +1561,7 @@
subq %rax,%rdx
subq %rbx,%rcx
subl $1,%edi
- jnz L$oop_62
+ jnz L$oop_53
.byte 0xf3,0xc3
.cfi_endproc
diff --git a/c-source/build/mach-o/div3w-armv8.S b/c-source/build/mach-o/div3w-armv8.S
--- a/c-source/build/mach-o/div3w-armv8.S
+++ b/c-source/build/mach-o/div3w-armv8.S
@@ -25,7 +25,7 @@
asr x3,x0,#63 // top bit -> mask
add x0,x0,x0 // Q <<= 1
subs x6,x4,x1 // R - D
- add x0,x0,#1 // Q + specilative bit
+ add x0,x0,#1 // Q + speculative bit
sbcs x7,x5,x2
sbc x0,x0,xzr // subtract speculative bit
diff --git a/c-source/build/mach-o/div3w-x86_64.s b/c-source/build/mach-o/div3w-x86_64.s
--- a/c-source/build/mach-o/div3w-x86_64.s
+++ b/c-source/build/mach-o/div3w-x86_64.s
@@ -8,6 +8,8 @@
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
+
+
movq (%rdi),%r8
movq 8(%rdi),%r9
xorq %rax,%rax
@@ -39,8 +41,9 @@
orq %rcx,%rax
+
.byte 0xf3,0xc3
-.cfi_endproc
+.cfi_endproc
.globl _quot_rem_128
.private_extern _quot_rem_128
@@ -50,6 +53,8 @@
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
+
+
movq %rdx,%rax
movq %rdx,%rcx
@@ -84,8 +89,9 @@
movq %rcx,%rax
+
.byte 0xf3,0xc3
-.cfi_endproc
+.cfi_endproc
@@ -100,6 +106,8 @@
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
+
+
movq %rdx,%rax
imulq 0(%rsi),%rdx
@@ -110,6 +118,7 @@
movq %r10,0(%rdi)
movq %rax,8(%rdi)
+
.byte 0xf3,0xc3
-.cfi_endproc
+.cfi_endproc
diff --git a/c-source/build/mach-o/mulq_mont_256-x86_64.s b/c-source/build/mach-o/mulq_mont_256-x86_64.s
--- a/c-source/build/mach-o/mulq_mont_256-x86_64.s
+++ b/c-source/build/mach-o/mulq_mont_256-x86_64.s
@@ -1,3 +1,4 @@
+.comm ___blst_platform_cap,4
.text
.globl _mul_mont_sparse_256
@@ -9,6 +10,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz mul_mont_sparse_256$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -73,6 +78,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sqr_mont_sparse_256$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -419,6 +428,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz from_mont_256$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -494,6 +507,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz redc_mont_256$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/mach-o/mulq_mont_384-x86_64.s b/c-source/build/mach-o/mulq_mont_384-x86_64.s
--- a/c-source/build/mach-o/mulq_mont_384-x86_64.s
+++ b/c-source/build/mach-o/mulq_mont_384-x86_64.s
@@ -1,3 +1,4 @@
+.comm ___blst_platform_cap,4
.text
@@ -8,7 +9,7 @@
.p2align 5
-__sub_mod_384x384:
+__subq_mod_384x384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -77,7 +78,7 @@
.p2align 5
-__add_mod_384:
+__addq_mod_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -129,7 +130,7 @@
.p2align 5
-__sub_mod_384:
+__subq_mod_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -140,7 +141,7 @@
movq 32(%rsi),%r12
movq 40(%rsi),%r13
-__sub_mod_384_a_is_loaded:
+__subq_mod_384_a_is_loaded:
subq 0(%rdx),%r8
movq 0(%rcx),%r14
sbbq 8(%rdx),%r9
@@ -187,6 +188,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz mul_mont_384x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -232,12 +237,12 @@
movq 8(%rsp),%rcx
leaq -48(%rsi),%rdx
leaq 40+192+48(%rsp),%rdi
- call __add_mod_384
+ call __addq_mod_384
movq 16(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq -48(%rdi),%rdi
- call __add_mod_384
+ call __addq_mod_384
leaq (%rdi),%rbx
leaq 48(%rdi),%rsi
@@ -247,17 +252,17 @@
leaq (%rdi),%rsi
leaq 40(%rsp),%rdx
movq 8(%rsp),%rcx
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq (%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq 40(%rsp),%rsi
leaq 40+96(%rsp),%rdx
leaq 40(%rsp),%rdi
- call __sub_mod_384x384
+ call __subq_mod_384x384
movq %rcx,%rbx
@@ -266,14 +271,14 @@
movq 0(%rsp),%rcx
movq 32(%rsp),%rdi
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
leaq 40+192(%rsp),%rsi
movq 0(%rsp),%rcx
leaq 48(%rdi),%rdi
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
leaq 328(%rsp),%r8
movq 0(%r8),%r15
@@ -303,6 +308,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sqr_mont_384x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -333,13 +342,13 @@
leaq 48(%rsi),%rdx
leaq 32(%rsp),%rdi
- call __add_mod_384
+ call __addq_mod_384
movq 16(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq 32+48(%rsp),%rdi
- call __sub_mod_384
+ call __subq_mod_384
movq 16(%rsp),%rsi
@@ -427,6 +436,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz mul_382x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -521,18 +534,18 @@
leaq 32(%rsp),%rdx
movq 24(%rsp),%rcx
movq %rsi,%rdi
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq 0(%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq -96(%rdi),%rsi
leaq 32(%rsp),%rdx
leaq -96(%rdi),%rdi
- call __sub_mod_384x384
+ call __subq_mod_384x384
leaq 136(%rsp),%r8
movq 0(%r8),%r15
@@ -562,6 +575,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sqr_382x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -617,7 +634,7 @@
leaq 48(%rsi),%rdx
leaq 48(%rdi),%rdi
- call __sub_mod_384_a_is_loaded
+ call __subq_mod_384_a_is_loaded
leaq (%rdi),%rsi
@@ -695,6 +712,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz mul_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1019,6 +1040,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sqr_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1265,6 +1290,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sqr_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1299,7 +1328,7 @@
movq 104(%rsp),%rbx
movq 112(%rsp),%rdi
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
leaq 120(%rsp),%r8
movq 120(%rsp),%r15
@@ -1332,6 +1361,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz redc_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1356,7 +1389,7 @@
movq %rdx,%rbx
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
movq 8(%rsp),%r15
.cfi_restore %r15
@@ -1389,6 +1422,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz from_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1764,7 +1801,7 @@
.p2align 5
-__redc_tail_mont_384:
+__redq_tail_mont_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -1820,6 +1857,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sgn0_pty_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1898,6 +1939,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sgn0_pty_mont_384x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2025,6 +2070,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz mul_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2689,6 +2738,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sqr_n_mul_mont_384$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2727,7 +2780,7 @@
movq 0(%rsp),%rcx
movq 16(%rsp),%rbx
call __mulq_by_1_mont_384
- call __redc_tail_mont_384
+ call __redq_tail_mont_384
movd %xmm1,%edx
leaq 0(%rdi),%rsi
@@ -2777,6 +2830,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sqr_n_mul_mont_383$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -3438,6 +3495,10 @@
.byte 0xf3,0x0f,0x1e,0xfa
+#ifdef __BLST_PORTABLE__
+ testl $1,___blst_platform_cap(%rip)
+ jnz sqr_mont_382x$1
+#endif
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/mach-o/mulx_mont_256-x86_64.s b/c-source/build/mach-o/mulx_mont_256-x86_64.s
--- a/c-source/build/mach-o/mulx_mont_256-x86_64.s
+++ b/c-source/build/mach-o/mulx_mont_256-x86_64.s
@@ -9,6 +9,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_mont_sparse_256$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -71,6 +72,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_mont_sparse_256$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -332,6 +334,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+from_mont_256$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -407,6 +410,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+redc_mont_256$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/mach-o/mulx_mont_384-x86_64.s b/c-source/build/mach-o/mulx_mont_384-x86_64.s
--- a/c-source/build/mach-o/mulx_mont_384-x86_64.s
+++ b/c-source/build/mach-o/mulx_mont_384-x86_64.s
@@ -8,7 +8,7 @@
.p2align 5
-__sub_mod_384x384:
+__subx_mod_384x384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -77,7 +77,7 @@
.p2align 5
-__add_mod_384:
+__addx_mod_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -129,7 +129,7 @@
.p2align 5
-__sub_mod_384:
+__subx_mod_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -140,7 +140,7 @@
movq 32(%rsi),%r12
movq 40(%rsi),%r13
-__sub_mod_384_a_is_loaded:
+__subx_mod_384_a_is_loaded:
subq 0(%rdx),%r8
movq 0(%rcx),%r14
sbbq 8(%rdx),%r9
@@ -187,6 +187,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_mont_384x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -233,12 +234,12 @@
leaq (%rbx),%rsi
leaq -48(%rbx),%rdx
leaq 40+192+48(%rsp),%rdi
- call __add_mod_384
+ call __addx_mod_384
movq 24(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq -48(%rdi),%rdi
- call __add_mod_384
+ call __addx_mod_384
leaq (%rdi),%rbx
leaq 48(%rdi),%rsi
@@ -248,17 +249,17 @@
leaq (%rdi),%rsi
leaq 40(%rsp),%rdx
movq 8(%rsp),%rcx
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq (%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq 40(%rsp),%rsi
leaq 40+96(%rsp),%rdx
leaq 40(%rsp),%rdi
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq (%rcx),%rbx
@@ -267,14 +268,14 @@
movq 0(%rsp),%rcx
movq 32(%rsp),%rdi
call __mulx_by_1_mont_384
- call __redc_tail_mont_384
+ call __redx_tail_mont_384
leaq 40+192(%rsp),%rsi
movq 0(%rsp),%rcx
leaq 48(%rdi),%rdi
call __mulx_by_1_mont_384
- call __redc_tail_mont_384
+ call __redx_tail_mont_384
leaq 328(%rsp),%r8
movq 0(%r8),%r15
@@ -304,6 +305,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_mont_384x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -335,13 +337,13 @@
leaq 48(%rsi),%rdx
leaq 32(%rsp),%rdi
- call __add_mod_384
+ call __addx_mod_384
movq 24(%rsp),%rsi
leaq 48(%rsi),%rdx
leaq 32+48(%rsp),%rdi
- call __sub_mod_384
+ call __subx_mod_384
movq 24(%rsp),%rsi
@@ -439,6 +441,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_382x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -533,18 +536,18 @@
leaq 32(%rsp),%rdx
movq 24(%rsp),%rcx
movq %rsi,%rdi
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq 0(%rdi),%rsi
leaq -96(%rdi),%rdx
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq -96(%rdi),%rsi
leaq 32(%rsp),%rdx
leaq -96(%rdi),%rdi
- call __sub_mod_384x384
+ call __subx_mod_384x384
leaq 136(%rsp),%r8
movq 0(%r8),%r15
@@ -574,6 +577,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_382x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -629,7 +633,7 @@
leaq 48(%rsi),%rdx
leaq 48(%rdi),%rdi
- call __sub_mod_384_a_is_loaded
+ call __subx_mod_384_a_is_loaded
leaq (%rdi),%rsi
@@ -707,6 +711,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -933,6 +938,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1127,6 +1133,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+redc_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1151,7 +1158,7 @@
movq %rdx,%rbx
call __mulx_by_1_mont_384
- call __redc_tail_mont_384
+ call __redx_tail_mont_384
movq 8(%rsp),%r15
.cfi_restore %r15
@@ -1184,6 +1191,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+from_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1450,7 +1458,7 @@
.p2align 5
-__redc_tail_mont_384:
+__redx_tail_mont_384:
.cfi_startproc
.byte 0xf3,0x0f,0x1e,0xfa
@@ -1506,6 +1514,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sgn0_pty_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1584,6 +1593,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sgn0_pty_mont_384x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -1711,6 +1721,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+mul_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2178,6 +2189,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2245,6 +2257,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_n_mul_mont_384$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2330,6 +2343,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_n_mul_mont_383$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
@@ -2776,6 +2790,7 @@
.byte 0xf3,0x0f,0x1e,0xfa
+sqr_mont_382x$1:
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
diff --git a/c-source/build/mach-o/sha256-armv8.S b/c-source/build/mach-o/sha256-armv8.S
--- a/c-source/build/mach-o/sha256-armv8.S
+++ b/c-source/build/mach-o/sha256-armv8.S
@@ -10,11 +10,12 @@
//
// sha256_block procedure for ARMv8.
//
-// This module is stripped of scalar code paths, with raionale that all
+// This module is stripped of scalar code paths, with rationale that all
// known processors are NEON-capable.
//
// See original module at CRYPTOGAMS for further details.
+.comm ___blst_platform_cap,4
.text
.align 6
@@ -184,6 +185,11 @@
.align 4
_blst_sha256_block_data_order:
+ adrp x16,___blst_platform_cap@PAGE
+ ldr w16,[x16,___blst_platform_cap@PAGEOFF]
+ tst w16,#1
+ b.ne Lv8_entry
+
stp x29, x30, [sp, #-16]!
mov x29, sp
sub sp,sp,#16*4
diff --git a/c-source/build/mach-o/sha256-portable-x86_64.s b/c-source/build/mach-o/sha256-portable-x86_64.s
--- a/c-source/build/mach-o/sha256-portable-x86_64.s
+++ b/c-source/build/mach-o/sha256-portable-x86_64.s
@@ -1,3 +1,4 @@
+.comm ___blst_platform_cap,4
.text
.globl _blst_sha256_block_data_order
@@ -8,33 +9,35 @@
.byte 0xf3,0x0f,0x1e,0xfa
- pushq %rbx
-.cfi_adjust_cfa_offset 8
-.cfi_offset %rbx,-16
pushq %rbp
.cfi_adjust_cfa_offset 8
-.cfi_offset %rbp,-24
+.cfi_offset %rbp,-16
+ movq %rsp,%rbp
+.cfi_def_cfa_register %rbp
+#ifdef __BLST_PORTABLE__
+ testl $2,___blst_platform_cap(%rip)
+ jnz L$blst_sha256_block_data_order$2
+#endif
+ pushq %rbx
+.cfi_offset %rbx,-24
pushq %r12
-.cfi_adjust_cfa_offset 8
.cfi_offset %r12,-32
pushq %r13
-.cfi_adjust_cfa_offset 8
.cfi_offset %r13,-40
pushq %r14
-.cfi_adjust_cfa_offset 8
.cfi_offset %r14,-48
pushq %r15
-.cfi_adjust_cfa_offset 8
.cfi_offset %r15,-56
shlq $4,%rdx
subq $64+24,%rsp
-.cfi_adjust_cfa_offset 16*4+3*8
+
+.cfi_def_cfa %rsp,144
+
leaq (%rsi,%rdx,4),%rdx
movq %rdi,64+0(%rsp)
movq %rsi,64+8(%rsp)
movq %rdx,64+16(%rsp)
-
movl 0(%rdi),%eax
movl 4(%rdi),%ebx
movl 8(%rdi),%ecx
@@ -1636,23 +1639,23 @@
leaq 64+24+48(%rsp),%r11
.cfi_def_cfa %r11,8
movq 64+24(%rsp),%r15
-.cfi_restore %r15
movq -40(%r11),%r14
-.cfi_restore %r14
movq -32(%r11),%r13
-.cfi_restore %r13
movq -24(%r11),%r12
+ movq -16(%r11),%rbx
+ movq -8(%r11),%rbp
.cfi_restore %r12
- movq -16(%r11),%rbp
+.cfi_restore %r13
+.cfi_restore %r14
+.cfi_restore %r15
.cfi_restore %rbp
- movq -8(%r11),%rbx
.cfi_restore %rbx
-
leaq (%r11),%rsp
.byte 0xf3,0xc3
.cfi_endproc
+#ifndef __BLST_PORTABLE__
.p2align 6
K256:
@@ -1744,3 +1747,4 @@
.byte 0xf3,0xc3
.cfi_endproc
+#endif
diff --git a/c-source/build/mach-o/sha256-x86_64.s b/c-source/build/mach-o/sha256-x86_64.s
--- a/c-source/build/mach-o/sha256-x86_64.s
+++ b/c-source/build/mach-o/sha256-x86_64.s
@@ -1,3 +1,4 @@
+.comm ___blst_platform_cap,4
.text
.p2align 6
@@ -33,6 +34,13 @@
.byte 0xf3,0x0f,0x1e,0xfa
+ pushq %rbp
+.cfi_adjust_cfa_offset 8
+.cfi_offset %rbp,-16
+ movq %rsp,%rbp
+.cfi_def_cfa_register %rbp
+L$blst_sha256_block_data_order$2:
+
leaq K256+128(%rip),%rcx
movdqu (%rdi),%xmm1
movdqu 16(%rdi),%xmm2
@@ -234,6 +242,11 @@
movdqu %xmm1,(%rdi)
movdqu %xmm2,16(%rdi)
+.cfi_def_cfa_register %rsp
+ popq %rbp
+.cfi_adjust_cfa_offset -8
+.cfi_restore %rbp
+
.byte 0xf3,0xc3
.cfi_endproc
@@ -249,30 +262,27 @@
pushq %rbp
.cfi_adjust_cfa_offset 8
.cfi_offset %rbp,-16
+ movq %rsp,%rbp
+.cfi_def_cfa_register %rbp
+ testl $2,___blst_platform_cap(%rip)
+ jnz L$blst_sha256_block_data_order$2
pushq %rbx
-.cfi_adjust_cfa_offset 8
.cfi_offset %rbx,-24
pushq %r12
-.cfi_adjust_cfa_offset 8
.cfi_offset %r12,-32
pushq %r13
-.cfi_adjust_cfa_offset 8
.cfi_offset %r13,-40
pushq %r14
-.cfi_adjust_cfa_offset 8
.cfi_offset %r14,-48
pushq %r15
-.cfi_adjust_cfa_offset 8
.cfi_offset %r15,-56
shlq $4,%rdx
- subq $40,%rsp
-.cfi_adjust_cfa_offset 40
+ subq $24,%rsp
+
leaq (%rsi,%rdx,4),%rdx
- movq %rdi,0(%rsp)
+ movq %rdi,-64(%rbp)
- movq %rdx,16(%rsp)
- movq %rsp,%rbp
-.cfi_def_cfa_register %rbp
+ movq %rdx,-48(%rbp)
leaq -64(%rsp),%rsp
@@ -291,7 +301,7 @@
.p2align 4
L$loop_ssse3:
movdqa K256+256(%rip),%xmm7
- movq %rsi,8(%rbp)
+ movq %rsi,-56(%rbp)
movdqu 0(%rsi),%xmm0
movdqu 16(%rsi),%xmm1
movdqu 32(%rsi),%xmm2
@@ -1316,9 +1326,9 @@
addl %r15d,%eax
movl %r8d,%r13d
addl %eax,%r14d
- movq 0(%rbp),%rdi
+ movq -64(%rbp),%rdi
movl %r14d,%eax
- movq 8(%rbp),%rsi
+ movq -56(%rbp),%rsi
addl 0(%rdi),%eax
addl 4(%rdi),%ebx
@@ -1330,7 +1340,7 @@
addl 28(%rdi),%r11d
leaq 64(%rsi),%rsi
- cmpq 16(%rbp),%rsi
+ cmpq -48(%rbp),%rsi
movl %eax,0(%rdi)
movl %ebx,4(%rdi)
@@ -1343,26 +1353,25 @@
jb L$loop_ssse3
xorps %xmm0,%xmm0
- leaq 40+48(%rbp),%r11
-.cfi_def_cfa %r11,8
movaps %xmm0,0(%rsp)
movaps %xmm0,16(%rsp)
movaps %xmm0,32(%rsp)
movaps %xmm0,48(%rsp)
- movq 40(%rbp),%r15
-.cfi_restore %r15
- movq -40(%r11),%r14
-.cfi_restore %r14
- movq -32(%r11),%r13
-.cfi_restore %r13
- movq -24(%r11),%r12
+ movq -40(%rbp),%r15
+ movq -32(%rbp),%r14
+ movq -24(%rbp),%r13
+ movq -16(%rbp),%r12
+ movq -8(%rbp),%rbx
+ movq %rbp,%rsp
+.cfi_def_cfa_register %rsp
+ popq %rbp
+.cfi_adjust_cfa_offset -8
+.cfi_restore %rbp
.cfi_restore %r12
- movq -16(%r11),%rbx
+.cfi_restore %r13
+.cfi_restore %r14
+.cfi_restore %r15
.cfi_restore %rbx
- movq -8(%r11),%rbp
-.cfi_restore %rbp
-
- leaq (%r11),%rsp
.byte 0xf3,0xc3
.cfi_endproc
diff --git a/c-source/src/aggregate.c b/c-source/src/aggregate.c
--- a/c-source/src/aggregate.c
+++ b/c-source/src/aggregate.c
@@ -90,7 +90,7 @@
/*
* Optional |nbits|-wide |scalar| is used to facilitate multiple aggregated
- * signature vetification as discussed at
+ * signature verification as discussed at
* https://ethresear.ch/t/fast-verification-of-multiple-bls-signatures/5407.
* Usage pattern is not finalized yet, because (sig != NULL) is better and
* will be handled separately...
diff --git a/c-source/src/bulk_addition.c b/c-source/src/bulk_addition.c
--- a/c-source/src/bulk_addition.c
+++ b/c-source/src/bulk_addition.c
@@ -145,8 +145,7 @@
void prefix##s_add(ptype *sum, const ptype##_affine *const points[], \
size_t npoints) \
{ \
- /* Performance with 288K scratch is within 1-2-3% from optimal */ \
- const size_t stride = sizeof(ptype)==sizeof(POINTonE1) ? 2048 : 1024; \
+ const size_t stride = SCRATCH_LIMIT / sizeof(ptype); \
ptype *scratch = alloca((npoints > stride ? stride : npoints) * \
sizeof(ptype)); \
const ptype##_affine *point = NULL; \
@@ -162,6 +161,15 @@
npoints -= j; \
} \
}
+
+#ifndef SCRATCH_LIMIT
+# ifdef __wasm__
+# define SCRATCH_LIMIT (45 * 1024)
+# else
+ /* Performance with 144K scratch is within 1-2-3% from optimal */
+# define SCRATCH_LIMIT (144 * 1024)
+# endif
+#endif
ADDITION_BTREE(blst_p1, POINTonE1, 384, fp, BLS12_381_Rx.p2)
diff --git a/c-source/src/bytes.h b/c-source/src/bytes.h
--- a/c-source/src/bytes.h
+++ b/c-source/src/bytes.h
@@ -26,7 +26,7 @@
* 'if (n % sizeof(limb_t) == 0)' is omitted because it's cheaper
* to perform redundant stores than to pay penalty for
* mispredicted branch. Besides, some compilers unroll the
- * loop and remove redundant stores to 'restict'-ed storage...
+ * loop and remove redundant stores to 'restrict'-ed storage...
*/
ret[n / sizeof(limb_t)] = limb;
}
@@ -55,7 +55,7 @@
* 'if (n % sizeof(limb_t) == 0)' is omitted because it's cheaper
* to perform redundant stores than to pay penalty for
* mispredicted branch. Besides, some compilers unroll the
- * loop and remove redundant stores to 'restict'-ed storage...
+ * loop and remove redundant stores to 'restrict'-ed storage...
*/
ret[n / sizeof(limb_t)] = limb;
}
diff --git a/c-source/src/cpuid.c b/c-source/src/cpuid.c
new file mode 100644
--- /dev/null
+++ b/c-source/src/cpuid.c
@@ -0,0 +1,85 @@
+/*
+ * Copyright Supranational LLC
+ * Licensed under the Apache License, Version 2.0, see LICENSE for details.
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+#if (defined(__GNUC__) || defined(__clang__) || defined(__SUNPRO_C)) && !defined(_WIN32)
+__attribute__((visibility("hidden")))
+#endif
+int __blst_platform_cap = 0;
+
+#if defined(__x86_64__) || defined(__x86_64) || defined(_M_X64)
+
+# if defined(__GNUC__) || defined(__clang__) || defined(__SUNPRO_C)
+static void __cpuidex(int info[4], int func, int sub)
+{
+ int eax, ebx, ecx, edx;
+
+ __asm__("cpuid" : "=a"(eax), "=b"(ebx), "=c"(ecx), "=d"(edx)
+ : "a"(func), "c"(sub));
+
+ info[0] = eax;
+ info[1] = ebx;
+ info[2] = ecx;
+ info[3] = edx;
+}
+# else
+# include
+# endif
+
+# if defined(__GNUC__) || defined(__clang__)
+__attribute__((constructor))
+# endif
+static int __blst_cpuid(void)
+{
+ int info[4], cap = 0;
+
+ __cpuidex(info, 0, 0);
+ if (info[0] > 6) {
+ __cpuidex(info, 7, 0);
+ cap |= (info[1]>>19) & 1; /* ADX */
+ cap |= (info[1]>>28) & 2; /* SHA */
+ }
+
+ __blst_platform_cap = cap;
+
+ return 0;
+}
+
+# if defined(_MSC_VER) && !defined(__clang__)
+# pragma section(".CRT$XCU",read)
+__declspec(allocate(".CRT$XCU")) static int (*p)(void) = __blst_cpuid;
+# elif defined(__SUNPRO_C)
+# pragma init(__blst_cpuid)
+# endif
+
+#elif defined(__aarch64__) || defined(__aarch64)
+
+# if defined(__linux__) && (defined(__GNUC__) || defined(__clang__))
+extern unsigned long getauxval(unsigned long type) __attribute__ ((weak));
+
+__attribute__((constructor))
+static int __blst_cpuid(void)
+{
+ int cap = 0;
+
+ if (getauxval) {
+ unsigned long hwcap_ce = getauxval(16);
+ cap = (hwcap_ce>>6) & 1; /* SHA256 */
+ }
+
+ __blst_platform_cap = cap;
+
+ return 0;
+}
+# elif defined(__APPLE__) && (defined(__GNUC__) || defined(__clang__))
+__attribute__((constructor))
+static int __blst_cpuid()
+{
+ __blst_platform_cap = 1; /* SHA256 */
+ return 0;
+}
+# endif
+
+#endif
diff --git a/c-source/src/e1.c b/c-source/src/e1.c
--- a/c-source/src/e1.c
+++ b/c-source/src/e1.c
@@ -155,7 +155,7 @@
{
if (vec_is_zero(in->X, 2*sizeof(in->X))) {
bytes_zero(out, 96);
- out[0] = 0x40; /* infinitiy bit */
+ out[0] = 0x40; /* infinity bit */
} else {
(void)POINTonE1_affine_Serialize_BE(out, in);
}
@@ -178,7 +178,7 @@
{
if (vec_is_zero(in->Z, sizeof(in->Z))) {
bytes_zero(out, 96);
- out[0] = 0x40; /* infinitiy bit */
+ out[0] = 0x40; /* infinity bit */
} else {
(void)POINTonE1_Serialize_BE(out, in);
}
@@ -202,7 +202,7 @@
{
if (vec_is_zero(in->X, 2*sizeof(in->X))) {
bytes_zero(out, 48);
- out[0] = 0xc0; /* compressed and infinitiy bits */
+ out[0] = 0xc0; /* compressed and infinity bits */
} else {
limb_t sign = POINTonE1_affine_Compress_BE(out, in);
out[0] |= (unsigned char)(0x80 | ((sign & 2) << 4));
@@ -226,7 +226,7 @@
{
if (vec_is_zero(in->Z, sizeof(in->Z))) {
bytes_zero(out, 48);
- out[0] = 0xc0; /* compressed and infinitiy bits */
+ out[0] = 0xc0; /* compressed and infinity bits */
} else {
limb_t sign = POINTonE1_Compress_BE(out, in);
out[0] |= (unsigned char)(0x80 | ((sign & 2) << 4));
diff --git a/c-source/src/e2.c b/c-source/src/e2.c
--- a/c-source/src/e2.c
+++ b/c-source/src/e2.c
@@ -196,7 +196,7 @@
{
if (vec_is_zero(in->X, 2*sizeof(in->X))) {
bytes_zero(out, 192);
- out[0] = 0x40; /* infinitiy bit */
+ out[0] = 0x40; /* infinity bit */
} else {
(void)POINTonE2_affine_Serialize_BE(out, in);
}
@@ -219,7 +219,7 @@
{
if (vec_is_zero(in->Z, sizeof(in->Z))) {
bytes_zero(out, 192);
- out[0] = 0x40; /* infinitiy bit */
+ out[0] = 0x40; /* infinity bit */
} else {
(void)POINTonE2_Serialize_BE(out, in);
}
@@ -245,7 +245,7 @@
{
if (vec_is_zero(in->X, 2*sizeof(in->X))) {
bytes_zero(out, 96);
- out[0] = 0xc0; /* compressed and infinitiy bits */
+ out[0] = 0xc0; /* compressed and infinity bits */
} else {
limb_t sign = POINTonE2_affine_Compress_BE(out, in);
out[0] |= (unsigned char)(0x80 | ((sign & 2) << 4));
@@ -269,7 +269,7 @@
{
if (vec_is_zero(in->Z, sizeof(in->Z))) {
bytes_zero(out, 96);
- out[0] = 0xc0; /* compressed and infinitiy bits */
+ out[0] = 0xc0; /* compressed and infinity bits */
} else {
limb_t sign = POINTonE2_Compress_BE(out, in);
out[0] |= (unsigned char)(0x80 | ((sign & 2) << 4));
diff --git a/c-source/src/ec_mult.h b/c-source/src/ec_mult.h
--- a/c-source/src/ec_mult.h
+++ b/c-source/src/ec_mult.h
@@ -46,9 +46,10 @@
static limb_t booth_encode(limb_t wval, size_t sz)
{
limb_t mask = 0 - (wval >> sz); /* "sign" bit -> mask */
+ launder(mask);
wval = (wval + 1) >> 1;
- wval = (wval & ~mask) | ((0-wval) & mask);
+ wval = (wval ^ mask) - mask;
/* &0x1f, but <=0x10, is index in table, rest is extended "sign" bit */
return wval;
@@ -61,7 +62,7 @@
* pass order's bit-length, which is customarily publicly known, instead
* of the factual scalars' bit-lengths. This is facilitated by point
* addition subroutines implemented to handle points at infinity, which
- * are encoded as Z==0. [Doubling agorithms handle such points at
+ * are encoded as Z==0. [Doubling algorithms handle such points at
* infinity "naturally," since resulting Z is product of original Z.]
*/
#define POINT_MULT_SCALAR_WX_IMPL(ptype, SZ) \
diff --git a/c-source/src/exports.c b/c-source/src/exports.c
--- a/c-source/src/exports.c
+++ b/c-source/src/exports.c
@@ -19,7 +19,7 @@
#include "bytes.h"
/*
- * BLS12-381-specifc Fr shortcuts to assembly.
+ * BLS12-381-specific Fr shortcuts to assembly.
*/
void blst_fr_add(vec256 ret, const vec256 a, const vec256 b)
{ add_mod_256(ret, a, b, BLS12_381_r); }
@@ -39,6 +39,24 @@
void blst_fr_mul(vec256 ret, const vec256 a, const vec256 b)
{ mul_mont_sparse_256(ret, a, b, BLS12_381_r, r0); }
+void blst_fr_ct_bfly(vec256 x0, vec256 x1, const vec256 twiddle)
+{
+ vec256 x2;
+
+ mul_mont_sparse_256(x2, x1, twiddle, BLS12_381_r, r0);
+ sub_mod_256(x1, x0, x2, BLS12_381_r);
+ add_mod_256(x0, x0, x2, BLS12_381_r);
+}
+
+void blst_fr_gs_bfly(vec256 x0, vec256 x1, const vec256 twiddle)
+{
+ vec256 x2;
+
+ sub_mod_256(x2, x0, x1, BLS12_381_r);
+ add_mod_256(x0, x0, x1, BLS12_381_r);
+ mul_mont_sparse_256(x1, x2, twiddle, BLS12_381_r, r0);
+}
+
void blst_fr_sqr(vec256 ret, const vec256 a)
{ sqr_mont_sparse_256(ret, a, BLS12_381_r, r0); }
@@ -102,27 +120,26 @@
int blst_sk_mul_n_check(pow256 ret, const pow256 a, const pow256 b)
{
- vec256 a_fr, b_fr;
+ vec256 t[2];
const union {
long one;
char little;
} is_endian = { 1 };
+ bool_t is_zero;
if (((size_t)a|(size_t)b)%sizeof(limb_t) != 0 || !is_endian.little) {
- limbs_from_le_bytes(a_fr, a, sizeof(a_fr));
- limbs_from_le_bytes(b_fr, b, sizeof(a_fr));
- a = (const byte *)a_fr;
- b = (const byte *)b_fr;
+ limbs_from_le_bytes(t[0], a, sizeof(pow256));
+ limbs_from_le_bytes(t[1], b, sizeof(pow256));
+ a = (const byte *)t[0];
+ b = (const byte *)t[1];
}
- mul_mont_sparse_256(a_fr, (const limb_t *)a, BLS12_381_rRR,
- BLS12_381_r, r0);
- mul_mont_sparse_256(b_fr, (const limb_t *)b, BLS12_381_rRR,
- BLS12_381_r, r0);
- mul_mont_sparse_256(a_fr, a_fr, b_fr, BLS12_381_r, r0);
- from_mont_256(a_fr, a_fr, BLS12_381_r, r0);
- le_bytes_from_limbs(ret, a_fr, sizeof(a_fr));
+ mul_mont_sparse_256(t[0], BLS12_381_rRR, (const limb_t *)a, BLS12_381_r, r0);
+ mul_mont_sparse_256(t[0], t[0], (const limb_t *)b, BLS12_381_r, r0);
+ le_bytes_from_limbs(ret, t[0], sizeof(pow256));
+ is_zero = vec_is_zero(t[0], sizeof(vec256));
+ vec_zero(t, sizeof(t));
- return (int)(vec_is_zero(a_fr, sizeof(a_fr)) ^ 1);
+ return (int)(is_zero^1);
}
void blst_sk_inverse(pow256 ret, const pow256 a)
@@ -150,7 +167,7 @@
}
/*
- * BLS12-381-specifc Fp shortcuts to assembly.
+ * BLS12-381-specific Fp shortcuts to assembly.
*/
void blst_fp_add(vec384 ret, const vec384 a, const vec384 b)
{ add_fp(ret, a, b); }
@@ -284,7 +301,7 @@
}
/*
- * BLS12-381-specifc Fp2 shortcuts to assembly.
+ * BLS12-381-specific Fp2 shortcuts to assembly.
*/
void blst_fp2_add(vec384x ret, const vec384x a, const vec384x b)
{ add_fp2(ret, a, b); }
@@ -311,7 +328,7 @@
{ cneg_fp2(ret, a, is_zero(flag) ^ 1); }
/*
- * Scalar serialization/deseriazation
+ * Scalar serialization/deserialization.
*/
void blst_scalar_from_uint32(pow256 ret, const unsigned int a[8])
{
@@ -480,68 +497,75 @@
int blst_scalar_from_le_bytes(pow256 out, const unsigned char *bytes, size_t n)
{
- struct { vec256 out, digit, radix; } t;
+ size_t rem = (n - 1) % 32 + 1;
+ struct { vec256 out, digit; } t;
limb_t ret;
vec_zero(t.out, sizeof(t.out));
- vec_copy(t.radix, BLS12_381_rRR, sizeof(t.radix));
- while (n > 32) {
- limbs_from_le_bytes(t.digit, bytes, 32);
- from_mont_256(t.digit, t.digit, BLS12_381_r, r0);
- mul_mont_sparse_256(t.digit, t.digit, t.radix, BLS12_381_r, r0);
+ n -= rem;
+ limbs_from_le_bytes(t.out, bytes += n, rem);
+ mul_mont_sparse_256(t.out, BLS12_381_rRR, t.out, BLS12_381_r, r0);
+
+ while (n) {
+ limbs_from_le_bytes(t.digit, bytes -= 32, 32);
add_mod_256(t.out, t.out, t.digit, BLS12_381_r);
- mul_mont_sparse_256(t.radix, t.radix, BLS12_381_rRR, BLS12_381_r, r0);
- bytes += 32;
+ mul_mont_sparse_256(t.out, BLS12_381_rRR, t.out, BLS12_381_r, r0);
n -= 32;
}
- vec_zero(t.digit, sizeof(t.digit));
- limbs_from_le_bytes(t.digit, bytes, n);
- from_mont_256(t.digit, t.digit, BLS12_381_r, r0);
- mul_mont_sparse_256(t.digit, t.digit, t.radix, BLS12_381_r, r0);
- add_mod_256(t.out, t.out, t.digit, BLS12_381_r);
+ from_mont_256(t.out, t.out, BLS12_381_r, r0);
ret = vec_is_zero(t.out, sizeof(t.out));
le_bytes_from_limbs(out, t.out, 32);
- vec_zero(t.out, 2*sizeof(t.out));
+ vec_zero(&t, sizeof(t));
return (int)(ret^1);
}
int blst_scalar_from_be_bytes(pow256 out, const unsigned char *bytes, size_t n)
{
- struct { vec256 out, digit, radix; } t;
+ size_t rem = (n - 1) % 32 + 1;
+ struct { vec256 out, digit; } t;
limb_t ret;
vec_zero(t.out, sizeof(t.out));
- vec_copy(t.radix, BLS12_381_rRR, sizeof(t.radix));
- bytes += n;
- while (n > 32) {
- limbs_from_be_bytes(t.digit, bytes -= 32, 32);
- from_mont_256(t.digit, t.digit, BLS12_381_r, r0);
- mul_mont_sparse_256(t.digit, t.digit, t.radix, BLS12_381_r, r0);
+ limbs_from_be_bytes(t.out, bytes, rem);
+ mul_mont_sparse_256(t.out, BLS12_381_rRR, t.out, BLS12_381_r, r0);
+
+ while (n -= rem) {
+ limbs_from_be_bytes(t.digit, bytes += rem, 32);
add_mod_256(t.out, t.out, t.digit, BLS12_381_r);
- mul_mont_sparse_256(t.radix, t.radix, BLS12_381_rRR, BLS12_381_r, r0);
- n -= 32;
+ mul_mont_sparse_256(t.out, BLS12_381_rRR, t.out, BLS12_381_r, r0);
+ rem = 32;
}
- vec_zero(t.digit, sizeof(t.digit));
- limbs_from_be_bytes(t.digit, bytes -= n, n);
- from_mont_256(t.digit, t.digit, BLS12_381_r, r0);
- mul_mont_sparse_256(t.digit, t.digit, t.radix, BLS12_381_r, r0);
- add_mod_256(t.out, t.out, t.digit, BLS12_381_r);
+ from_mont_256(t.out, t.out, BLS12_381_r, r0);
ret = vec_is_zero(t.out, sizeof(t.out));
le_bytes_from_limbs(out, t.out, 32);
- vec_zero(t.out, 2*sizeof(t.out));
+ vec_zero(&t, sizeof(t));
return (int)(ret^1);
}
/*
- * Test facilitator
+ * Single-short SHA-256 hash function.
+ */
+#include "sha256.h"
+
+void blst_sha256(unsigned char md[32], const void *msg, size_t len)
+{
+ SHA256_CTX ctx;
+
+ sha256_init(&ctx);
+ sha256_update(&ctx, msg, len);
+ sha256_final(md, &ctx);
+}
+
+/*
+ * Test facilitator.
*/
void blst_scalar_from_hexascii(pow256 ret, const char *hex)
{ bytes_from_hexascii(ret, sizeof(pow256), hex); }
diff --git a/c-source/src/fields.h b/c-source/src/fields.h
--- a/c-source/src/fields.h
+++ b/c-source/src/fields.h
@@ -10,7 +10,7 @@
#include "consts.h"
/*
- * BLS12-381-specifc Fp shortcuts to assembly.
+ * BLS12-381-specific Fp shortcuts to assembly.
*/
static inline void add_fp(vec384 ret, const vec384 a, const vec384 b)
{ add_mod_384(ret, a, b, BLS12_381_P); }
@@ -49,7 +49,7 @@
{ redc_mont_384(ret, a, BLS12_381_P, p0); }
/*
- * BLS12-381-specifc Fp2 shortcuts to assembly.
+ * BLS12-381-specific Fp2 shortcuts to assembly.
*/
static inline void add_fp2(vec384x ret, const vec384x a, const vec384x b)
{ add_mod_384x(ret, a, b, BLS12_381_P); }
diff --git a/c-source/src/fp12_tower.c b/c-source/src/fp12_tower.c
--- a/c-source/src/fp12_tower.c
+++ b/c-source/src/fp12_tower.c
@@ -545,7 +545,7 @@
mul_by_u_plus_1_fp2(c1, c1);
mul_fp2(t0, a[0], a[1]);
sub_fp2(c1, c1, t0);
-
+
/* c2 = a1^2 - a0*a2 */
sqr_fp2(c2, a[1]);
mul_fp2(t0, a[0], a[2]);
@@ -733,7 +733,7 @@
/*
- * BLS12-381-specifc Fp12 shortcuts.
+ * BLS12-381-specific Fp12 shortcuts.
*/
void blst_fp12_sqr(vec384fp12 ret, const vec384fp12 a)
{ sqr_fp12(ret, a); }
diff --git a/c-source/src/multi_scalar.c b/c-source/src/multi_scalar.c
--- a/c-source/src/multi_scalar.c
+++ b/c-source/src/multi_scalar.c
@@ -399,7 +399,20 @@
size_t npoints, \
const byte *const scalars[], size_t nbits, \
ptype##xyzz scratch[]) \
-{ ptype##s_mult_pippenger(ret, points, npoints, scalars, nbits, scratch, 0); }
+{ \
+ if (npoints == 1) { \
+ prefix##_from_affine(ret, points[0]); \
+ prefix##_mult(ret, ret, scalars[0], nbits); \
+ return; \
+ } \
+ if ((npoints * sizeof(ptype##_affine) * 8 * 3) <= SCRATCH_LIMIT) { \
+ ptype##_affine *table = alloca(npoints * sizeof(ptype##_affine) * 8); \
+ ptype##s_precompute_wbits(table, 4, points, npoints); \
+ ptype##s_mult_wbits(ret, table, 4, npoints, scalars, nbits, NULL); \
+ return; \
+ } \
+ ptype##s_mult_pippenger(ret, points, npoints, scalars, nbits, scratch, 0); \
+}
DECLARE_PRIVATE_POINTXYZZ(POINTonE1, 384)
POINTXYZZ_TO_JACOBIAN_IMPL(POINTonE1, 384, fp)
diff --git a/c-source/src/no_asm.h b/c-source/src/no_asm.h
--- a/c-source/src/no_asm.h
+++ b/c-source/src/no_asm.h
@@ -16,9 +16,20 @@
# pragma GCC diagnostic ignored "-Wstatic-in-inline"
#endif
+#if !defined(__clang__) && !defined(__builtin_assume)
+# if defined(__GNUC__) && __GNUC__>=5
+# define __builtin_assume(condition) if (!(condition)) __builtin_unreachable()
+# elif defined(_MSC_VER)
+# define __builtin_assume(condition) __assume(condition)
+# else
+# define __builtin_assume(condition) (void)(condition)
+# endif
+#endif
+
static void mul_mont_n(limb_t ret[], const limb_t a[], const limb_t b[],
const limb_t p[], limb_t n0, size_t n)
{
+ __builtin_assume(n != 0 && n%2 == 0);
llimb_t limbx;
limb_t mask, borrow, mx, hi, tmp[n+1], carry;
size_t i, j;
@@ -64,6 +75,7 @@
}
mask = carry - borrow;
+ launder(mask);
for(i=0; i> LIMB_T_BITS) & 1;
}
mask = 0 - borrow;
+ launder(mask);
/* "mul_mont_n(ret[1], a[0], a[1], p, n0, NLIMBS(384));" */
mul_mont_nonred_n(ret[1], a[0], a[1], p, n0, NLIMBS(384));
@@ -642,6 +679,7 @@
borrow = (limb_t)(limbx >> LIMB_T_BITS) & 1;
}
mask = 0 - borrow;
+ launder(mask);
for (carry=0, i=0; i> (LIMB_T_BITS-1))
+#if defined(__GNUC__) || defined(__clang__)
+# define MSB(x) ({ limb_t ret = (x) >> (LIMB_T_BITS-1); launder(ret); ret; })
+#else
+# define MSB(x) ((x) >> (LIMB_T_BITS-1))
+#endif
static size_t num_bits(limb_t l)
{
@@ -704,6 +746,7 @@
static void ab_approximation_n(limb_t a_[2], const limb_t a[],
limb_t b_[2], const limb_t b[], size_t n)
{
+ __builtin_assume(n != 0 && n%2 == 0);
limb_t a_hi, a_lo, b_hi, b_lo, mask;
size_t i;
@@ -729,6 +772,7 @@
static void inner_loop_n(factors *fg, const limb_t a_[2], const limb_t b_[2],
size_t n)
{
+ __builtin_assume(n != 0);
llimb_t limbx;
limb_t f0 = 1, g0 = 0, f1 = 0, g1 = 1;
limb_t a_lo, a_hi, b_lo, b_hi, t_lo, t_hi, odd, borrow, xorm;
@@ -784,6 +828,7 @@
static limb_t cneg_n(limb_t ret[], const limb_t a[], limb_t neg, size_t n)
{
+ __builtin_assume(n != 0);
llimb_t limbx = 0;
limb_t carry;
size_t i;
@@ -799,6 +844,7 @@
static limb_t add_n(limb_t ret[], const limb_t a[], limb_t b[], size_t n)
{
+ __builtin_assume(n != 0);
llimb_t limbx;
limb_t carry;
size_t i;
@@ -814,6 +860,7 @@
static limb_t umul_n(limb_t ret[], const limb_t a[], limb_t b, size_t n)
{
+ __builtin_assume(n != 0);
llimb_t limbx;
limb_t hi;
size_t i;
@@ -831,6 +878,7 @@
const limb_t b[], limb_t *g_,
size_t n)
{
+ __builtin_assume(n != 0);
limb_t a_[n+1], b_[n+1], f, g, neg, carry, hi;
size_t i;
@@ -872,6 +920,7 @@
static limb_t smul_2n(limb_t ret[], const limb_t u[], limb_t f,
const limb_t v[], limb_t g, size_t n)
{
+ __builtin_assume(n != 0);
limb_t u_[n], v_[n], neg, hi;
/* |u|*|f_| */
@@ -895,6 +944,7 @@
static void ct_inverse_mod_n(limb_t ret[], const limb_t inp[],
const limb_t mod[], const limb_t modx[], size_t n)
{
+ __builtin_assume(n != 0 && n%2 == 0);
llimb_t limbx;
limb_t a[n], b[n], u[2*n], v[2*n], t[2*n];
limb_t a_[2], b_[2], sign, carry, top;
@@ -935,13 +985,13 @@
add_n(ret+n, ret+n, a, n);
}
-#define CT_INVERSE_MOD_IMPL(bits) \
-inline void ct_inverse_mod_##bits(vec##bits ret, const vec##bits inp, \
+#define CT_INVERSE_MOD_IMPL(bits, bits2) \
+inline void ct_inverse_mod_##bits(vec##bits2 ret, const vec##bits inp, \
const vec##bits mod, const vec##bits modx) \
{ ct_inverse_mod_n(ret, inp, mod, modx, NLIMBS(bits)); }
-CT_INVERSE_MOD_IMPL(256)
-CT_INVERSE_MOD_IMPL(384)
+CT_INVERSE_MOD_IMPL(256, 512)
+CT_INVERSE_MOD_IMPL(384, 768)
/*
* Copy of inner_loop_n above, but with |L| updates.
@@ -949,6 +999,7 @@
static limb_t legendre_loop_n(limb_t L, factors *fg, const limb_t a_[2],
const limb_t b_[2], size_t n)
{
+ __builtin_assume(n != 0);
llimb_t limbx;
limb_t f0 = 1, g0 = 0, f1 = 0, g1 = 1;
limb_t a_lo, a_hi, b_lo, b_hi, t_lo, t_hi, odd, borrow, xorm;
@@ -1010,6 +1061,7 @@
static bool_t ct_is_sqr_mod_n(const limb_t inp[], const limb_t mod[], size_t n)
{
+ __builtin_assume(n != 0 && n%2 == 0);
limb_t a[n], b[n], t[n];
limb_t a_[2], b_[2], neg, L = 0;
factors fg;
@@ -1103,6 +1155,7 @@
static limb_t quot_rem_n(limb_t *div_rem, const limb_t *divisor,
limb_t quotient, size_t n)
{
+ __builtin_assume(n != 0 && n%2 == 0);
llimb_t limbx;
limb_t tmp[n+1], carry, mask, borrow;
size_t i;
@@ -1123,6 +1176,7 @@
}
mask = 0 - borrow;
+ launder(mask);
/* if quotient was off by one, add divisor to the remainder */
for (carry=0, i=0; iX, Pptr->X);
+ neg_fp(Px2[i].X, Px2[i].X);
+ add_fp(Px2[i].Y, Pptr->Y, Pptr->Y);
+
+ vec_copy(Q[i].X, Qptr->X, 2*sizeof(Q[i].X));
+ vec_copy(T[i].X, Qptr->X, 2*sizeof(T[i].X));
+ vec_copy(T[i].Z, BLS12_381_Rx.p2, sizeof(T[i].Z));
+
+ if (++i == MILLER_LOOP_N_MAX || j == n-1) {
+ vec384fp12 tmp;
+ vec384fp6 *ret = j < MILLER_LOOP_N_MAX ? out : tmp;
+
+ /* first step is ret = 1^2*line, which is just ret = line */
+ start_dbl_n(ret, T, Px2, i); /* 0x2 */
+ add_n_dbl_n(ret, T, Q, Px2, i, 2); /* ..0xc */
+ add_n_dbl_n(ret, T, Q, Px2, i, 3); /* ..0x68 */
+ add_n_dbl_n(ret, T, Q, Px2, i, 9); /* ..0xd200 */
+ add_n_dbl_n(ret, T, Q, Px2, i, 32); /* ..0xd20100000000 */
+ add_n_dbl_n(ret, T, Q, Px2, i, 16); /* ..0xd201000000010000 */
+ conjugate_fp12(ret); /* account for z being negative */
+
+ if (j >= MILLER_LOOP_N_MAX)
+ mul_fp12(out, out, ret);
+
+ i = 0;
+ }
+ }
+}
+
void blst_final_exp(vec384fp12 ret, const vec384fp12 f)
{ final_exp(ret, f); }
diff --git a/c-source/src/pentaroot-addchain.h b/c-source/src/pentaroot-addchain.h
new file mode 100644
--- /dev/null
+++ b/c-source/src/pentaroot-addchain.h
@@ -0,0 +1,333 @@
+/*
+ * Copyright Supranational LLC
+ * Licensed under the Apache License, Version 2.0, see LICENSE for details.
+ * SPDX-License-Identifier: Apache-2.0
+ */
+/*
+ * The "magic" number is 1/5 modulo BLS12_381_r-1. Exponentiation to which
+ * yields 5th root of the base.
+ *
+ * Generated with 'addchain 20974350070050476191779096203274386335076221000211055129041463479975432473805'
+ * https://github.com/kwantam/addchain
+ * # Bos-Coster (win=4) : 307 (15)
+ * # Bos-Coster (win=10) : 307 (18)
+ * # Yacobi : 319 (16)
+ * # Bos-Coster (win=2) : 319 ( 5)
+ * # Bos-Coster (win=5) : 306 (19) <<<
+ * # Bos-Coster (win=7) : 311 (22)
+ * # Bos-Coster (win=9) : 313 (20)
+ * # Bos-Coster (win=3) : 314 ( 9)
+ * # Bos-Coster (win=6) : 309 (21)
+ * # Bos-Coster (win=8) : 309 (23)
+ * # Bergeron-Berstel-Brlek-Duboc : 334 ( 5)
+ */
+
+#define PENTAROOT_MOD_BLS12_381_r(out, inp, ptype) do { \
+ptype t[19]; \
+vec_copy(t[1], inp, sizeof(ptype)); /* 0: 1 */\
+sqr(t[7], t[1]); /* 1: 2 */\
+sqr(t[0], t[7]); /* 2: 4 */\
+sqr(t[2], t[0]); /* 3: 8 */\
+mul(t[10], t[2], t[1]); /* 4: 9 */\
+mul(t[3], t[10], t[7]); /* 5: b */\
+mul(t[1], t[10], t[0]); /* 6: d */\
+mul(t[5], t[3], t[0]); /* 7: f */\
+mul(t[9], t[10], t[2]); /* 8: 11 */\
+mul(t[4], t[3], t[2]); /* 9: 13 */\
+mul(t[15], t[5], t[2]); /* 10: 17 */\
+mul(t[8], t[15], t[2]); /* 11: 1f */\
+mul(t[13], t[8], t[7]); /* 12: 21 */\
+mul(t[14], t[8], t[0]); /* 13: 23 */\
+mul(t[12], t[13], t[0]); /* 14: 25 */\
+mul(t[6], t[8], t[2]); /* 15: 27 */\
+mul(t[11], t[14], t[2]); /* 16: 2b */\
+sqr(t[0], t[15]); /* 17: 2e */\
+mul(t[18], t[6], t[2]); /* 18: 2f */\
+mul(t[2], t[11], t[2]); /* 19: 33 */\
+mul(t[16], t[2], t[7]); /* 20: 35 */\
+mul(t[7], t[0], t[3]); /* 21: 39 */\
+mul(t[17], t[0], t[5]); /* 22: 3d */\
+/* sqr(t[0], t[0]); */ /* 23: 5c */\
+/* sqr(t[0], t[0]); */ /* 24: b8 */\
+/* sqr(t[0], t[0]); */ /* 25: 170 */\
+/* sqr(t[0], t[0]); */ /* 26: 2e0 */\
+/* sqr(t[0], t[0]); */ /* 27: 5c0 */\
+/* sqr(t[0], t[0]); */ /* 28: b80 */\
+/* sqr(t[0], t[0]); */ /* 29: 1700 */\
+sqr_n_mul(t[0], t[0], 7, t[18]); /* 30: 172f */\
+/* sqr(t[0], t[0]); */ /* 31: 2e5e */\
+/* sqr(t[0], t[0]); */ /* 32: 5cbc */\
+/* sqr(t[0], t[0]); */ /* 33: b978 */\
+/* sqr(t[0], t[0]); */ /* 34: 172f0 */\
+/* sqr(t[0], t[0]); */ /* 35: 2e5e0 */\
+/* sqr(t[0], t[0]); */ /* 36: 5cbc0 */\
+sqr_n_mul(t[0], t[0], 6, t[13]); /* 37: 5cbe1 */\
+/* sqr(t[0], t[0]); */ /* 38: b97c2 */\
+/* sqr(t[0], t[0]); */ /* 39: 172f84 */\
+/* sqr(t[0], t[0]); */ /* 40: 2e5f08 */\
+/* sqr(t[0], t[0]); */ /* 41: 5cbe10 */\
+/* sqr(t[0], t[0]); */ /* 42: b97c20 */\
+/* sqr(t[0], t[0]); */ /* 43: 172f840 */\
+sqr_n_mul(t[0], t[0], 6, t[17]); /* 44: 172f87d */\
+/* sqr(t[0], t[0]); */ /* 45: 2e5f0fa */\
+/* sqr(t[0], t[0]); */ /* 46: 5cbe1f4 */\
+/* sqr(t[0], t[0]); */ /* 47: b97c3e8 */\
+/* sqr(t[0], t[0]); */ /* 48: 172f87d0 */\
+/* sqr(t[0], t[0]); */ /* 49: 2e5f0fa0 */\
+/* sqr(t[0], t[0]); */ /* 50: 5cbe1f40 */\
+sqr_n_mul(t[0], t[0], 6, t[16]); /* 51: 5cbe1f75 */\
+/* sqr(t[0], t[0]); */ /* 52: b97c3eea */\
+/* sqr(t[0], t[0]); */ /* 53: 172f87dd4 */\
+/* sqr(t[0], t[0]); */ /* 54: 2e5f0fba8 */\
+/* sqr(t[0], t[0]); */ /* 55: 5cbe1f750 */\
+/* sqr(t[0], t[0]); */ /* 56: b97c3eea0 */\
+sqr_n_mul(t[0], t[0], 5, t[15]); /* 57: b97c3eeb7 */\
+/* sqr(t[0], t[0]); */ /* 58: 172f87dd6e */\
+/* sqr(t[0], t[0]); */ /* 59: 2e5f0fbadc */\
+/* sqr(t[0], t[0]); */ /* 60: 5cbe1f75b8 */\
+/* sqr(t[0], t[0]); */ /* 61: b97c3eeb70 */\
+/* sqr(t[0], t[0]); */ /* 62: 172f87dd6e0 */\
+/* sqr(t[0], t[0]); */ /* 63: 2e5f0fbadc0 */\
+sqr_n_mul(t[0], t[0], 6, t[15]); /* 64: 2e5f0fbadd7 */\
+/* sqr(t[0], t[0]); */ /* 65: 5cbe1f75bae */\
+/* sqr(t[0], t[0]); */ /* 66: b97c3eeb75c */\
+/* sqr(t[0], t[0]); */ /* 67: 172f87dd6eb8 */\
+/* sqr(t[0], t[0]); */ /* 68: 2e5f0fbadd70 */\
+/* sqr(t[0], t[0]); */ /* 69: 5cbe1f75bae0 */\
+/* sqr(t[0], t[0]); */ /* 70: b97c3eeb75c0 */\
+/* sqr(t[0], t[0]); */ /* 71: 172f87dd6eb80 */\
+/* sqr(t[0], t[0]); */ /* 72: 2e5f0fbadd700 */\
+sqr_n_mul(t[0], t[0], 8, t[14]); /* 73: 2e5f0fbadd723 */\
+/* sqr(t[0], t[0]); */ /* 74: 5cbe1f75bae46 */\
+/* sqr(t[0], t[0]); */ /* 75: b97c3eeb75c8c */\
+/* sqr(t[0], t[0]); */ /* 76: 172f87dd6eb918 */\
+/* sqr(t[0], t[0]); */ /* 77: 2e5f0fbadd7230 */\
+/* sqr(t[0], t[0]); */ /* 78: 5cbe1f75bae460 */\
+/* sqr(t[0], t[0]); */ /* 79: b97c3eeb75c8c0 */\
+/* sqr(t[0], t[0]); */ /* 80: 172f87dd6eb9180 */\
+/* sqr(t[0], t[0]); */ /* 81: 2e5f0fbadd72300 */\
+sqr_n_mul(t[0], t[0], 8, t[13]); /* 82: 2e5f0fbadd72321 */\
+/* sqr(t[0], t[0]); */ /* 83: 5cbe1f75bae4642 */\
+/* sqr(t[0], t[0]); */ /* 84: b97c3eeb75c8c84 */\
+/* sqr(t[0], t[0]); */ /* 85: 172f87dd6eb91908 */\
+/* sqr(t[0], t[0]); */ /* 86: 2e5f0fbadd723210 */\
+/* sqr(t[0], t[0]); */ /* 87: 5cbe1f75bae46420 */\
+/* sqr(t[0], t[0]); */ /* 88: b97c3eeb75c8c840 */\
+sqr_n_mul(t[0], t[0], 6, t[2]); /* 89: b97c3eeb75c8c873 */\
+/* sqr(t[0], t[0]); */ /* 90: 172f87dd6eb9190e6 */\
+/* sqr(t[0], t[0]); */ /* 91: 2e5f0fbadd72321cc */\
+/* sqr(t[0], t[0]); */ /* 92: 5cbe1f75bae464398 */\
+/* sqr(t[0], t[0]); */ /* 93: b97c3eeb75c8c8730 */\
+/* sqr(t[0], t[0]); */ /* 94: 172f87dd6eb9190e60 */\
+/* sqr(t[0], t[0]); */ /* 95: 2e5f0fbadd72321cc0 */\
+sqr_n_mul(t[0], t[0], 6, t[13]); /* 96: 2e5f0fbadd72321ce1 */\
+/* sqr(t[0], t[0]); */ /* 97: 5cbe1f75bae46439c2 */\
+/* sqr(t[0], t[0]); */ /* 98: b97c3eeb75c8c87384 */\
+/* sqr(t[0], t[0]); */ /* 99: 172f87dd6eb9190e708 */\
+/* sqr(t[0], t[0]); */ /* 100: 2e5f0fbadd72321ce10 */\
+/* sqr(t[0], t[0]); */ /* 101: 5cbe1f75bae46439c20 */\
+/* sqr(t[0], t[0]); */ /* 102: b97c3eeb75c8c873840 */\
+/* sqr(t[0], t[0]); */ /* 103: 172f87dd6eb9190e7080 */\
+sqr_n_mul(t[0], t[0], 7, t[12]); /* 104: 172f87dd6eb9190e70a5 */\
+/* sqr(t[0], t[0]); */ /* 105: 2e5f0fbadd72321ce14a */\
+/* sqr(t[0], t[0]); */ /* 106: 5cbe1f75bae46439c294 */\
+/* sqr(t[0], t[0]); */ /* 107: b97c3eeb75c8c8738528 */\
+/* sqr(t[0], t[0]); */ /* 108: 172f87dd6eb9190e70a50 */\
+/* sqr(t[0], t[0]); */ /* 109: 2e5f0fbadd72321ce14a0 */\
+/* sqr(t[0], t[0]); */ /* 110: 5cbe1f75bae46439c2940 */\
+/* sqr(t[0], t[0]); */ /* 111: b97c3eeb75c8c87385280 */\
+/* sqr(t[0], t[0]); */ /* 112: 172f87dd6eb9190e70a500 */\
+sqr_n_mul(t[0], t[0], 8, t[11]); /* 113: 172f87dd6eb9190e70a52b */\
+/* sqr(t[0], t[0]); */ /* 114: 2e5f0fbadd72321ce14a56 */\
+/* sqr(t[0], t[0]); */ /* 115: 5cbe1f75bae46439c294ac */\
+/* sqr(t[0], t[0]); */ /* 116: b97c3eeb75c8c873852958 */\
+/* sqr(t[0], t[0]); */ /* 117: 172f87dd6eb9190e70a52b0 */\
+/* sqr(t[0], t[0]); */ /* 118: 2e5f0fbadd72321ce14a560 */\
+/* sqr(t[0], t[0]); */ /* 119: 5cbe1f75bae46439c294ac0 */\
+sqr_n_mul(t[0], t[0], 6, t[1]); /* 120: 5cbe1f75bae46439c294acd */\
+/* sqr(t[0], t[0]); */ /* 121: b97c3eeb75c8c873852959a */\
+/* sqr(t[0], t[0]); */ /* 122: 172f87dd6eb9190e70a52b34 */\
+/* sqr(t[0], t[0]); */ /* 123: 2e5f0fbadd72321ce14a5668 */\
+/* sqr(t[0], t[0]); */ /* 124: 5cbe1f75bae46439c294acd0 */\
+/* sqr(t[0], t[0]); */ /* 125: b97c3eeb75c8c873852959a0 */\
+/* sqr(t[0], t[0]); */ /* 126: 172f87dd6eb9190e70a52b340 */\
+/* sqr(t[0], t[0]); */ /* 127: 2e5f0fbadd72321ce14a56680 */\
+/* sqr(t[0], t[0]); */ /* 128: 5cbe1f75bae46439c294acd00 */\
+sqr_n_mul(t[0], t[0], 8, t[2]); /* 129: 5cbe1f75bae46439c294acd33 */\
+/* sqr(t[0], t[0]); */ /* 130: b97c3eeb75c8c873852959a66 */\
+/* sqr(t[0], t[0]); */ /* 131: 172f87dd6eb9190e70a52b34cc */\
+/* sqr(t[0], t[0]); */ /* 132: 2e5f0fbadd72321ce14a566998 */\
+/* sqr(t[0], t[0]); */ /* 133: 5cbe1f75bae46439c294acd330 */\
+/* sqr(t[0], t[0]); */ /* 134: b97c3eeb75c8c873852959a660 */\
+/* sqr(t[0], t[0]); */ /* 135: 172f87dd6eb9190e70a52b34cc0 */\
+sqr_n_mul(t[0], t[0], 6, t[11]); /* 136: 172f87dd6eb9190e70a52b34ceb */\
+/* sqr(t[0], t[0]); */ /* 137: 2e5f0fbadd72321ce14a56699d6 */\
+/* sqr(t[0], t[0]); */ /* 138: 5cbe1f75bae46439c294acd33ac */\
+/* sqr(t[0], t[0]); */ /* 139: b97c3eeb75c8c873852959a6758 */\
+/* sqr(t[0], t[0]); */ /* 140: 172f87dd6eb9190e70a52b34ceb0 */\
+sqr_n_mul(t[0], t[0], 4, t[10]); /* 141: 172f87dd6eb9190e70a52b34ceb9 */\
+/* sqr(t[0], t[0]); */ /* 142: 2e5f0fbadd72321ce14a56699d72 */\
+/* sqr(t[0], t[0]); */ /* 143: 5cbe1f75bae46439c294acd33ae4 */\
+/* sqr(t[0], t[0]); */ /* 144: b97c3eeb75c8c873852959a675c8 */\
+/* sqr(t[0], t[0]); */ /* 145: 172f87dd6eb9190e70a52b34ceb90 */\
+/* sqr(t[0], t[0]); */ /* 146: 2e5f0fbadd72321ce14a56699d720 */\
+sqr_n_mul(t[0], t[0], 5, t[8]); /* 147: 2e5f0fbadd72321ce14a56699d73f */\
+/* sqr(t[0], t[0]); */ /* 148: 5cbe1f75bae46439c294acd33ae7e */\
+/* sqr(t[0], t[0]); */ /* 149: b97c3eeb75c8c873852959a675cfc */\
+/* sqr(t[0], t[0]); */ /* 150: 172f87dd6eb9190e70a52b34ceb9f8 */\
+/* sqr(t[0], t[0]); */ /* 151: 2e5f0fbadd72321ce14a56699d73f0 */\
+/* sqr(t[0], t[0]); */ /* 152: 5cbe1f75bae46439c294acd33ae7e0 */\
+/* sqr(t[0], t[0]); */ /* 153: b97c3eeb75c8c873852959a675cfc0 */\
+/* sqr(t[0], t[0]); */ /* 154: 172f87dd6eb9190e70a52b34ceb9f80 */\
+/* sqr(t[0], t[0]); */ /* 155: 2e5f0fbadd72321ce14a56699d73f00 */\
+/* sqr(t[0], t[0]); */ /* 156: 5cbe1f75bae46439c294acd33ae7e00 */\
+/* sqr(t[0], t[0]); */ /* 157: b97c3eeb75c8c873852959a675cfc00 */\
+/* sqr(t[0], t[0]); */ /* 158: 172f87dd6eb9190e70a52b34ceb9f800 */\
+/* sqr(t[0], t[0]); */ /* 159: 2e5f0fbadd72321ce14a56699d73f000 */\
+/* sqr(t[0], t[0]); */ /* 160: 5cbe1f75bae46439c294acd33ae7e000 */\
+/* sqr(t[0], t[0]); */ /* 161: b97c3eeb75c8c873852959a675cfc000 */\
+/* sqr(t[0], t[0]); */ /* 162: 172f87dd6eb9190e70a52b34ceb9f8000 */\
+sqr_n_mul(t[0], t[0], 15, t[9]); /* 163: 172f87dd6eb9190e70a52b34ceb9f8011 */\
+/* sqr(t[0], t[0]); */ /* 164: 2e5f0fbadd72321ce14a56699d73f0022 */\
+/* sqr(t[0], t[0]); */ /* 165: 5cbe1f75bae46439c294acd33ae7e0044 */\
+/* sqr(t[0], t[0]); */ /* 166: b97c3eeb75c8c873852959a675cfc0088 */\
+/* sqr(t[0], t[0]); */ /* 167: 172f87dd6eb9190e70a52b34ceb9f80110 */\
+/* sqr(t[0], t[0]); */ /* 168: 2e5f0fbadd72321ce14a56699d73f00220 */\
+/* sqr(t[0], t[0]); */ /* 169: 5cbe1f75bae46439c294acd33ae7e00440 */\
+/* sqr(t[0], t[0]); */ /* 170: b97c3eeb75c8c873852959a675cfc00880 */\
+/* sqr(t[0], t[0]); */ /* 171: 172f87dd6eb9190e70a52b34ceb9f801100 */\
+sqr_n_mul(t[0], t[0], 8, t[3]); /* 172: 172f87dd6eb9190e70a52b34ceb9f80110b */\
+/* sqr(t[0], t[0]); */ /* 173: 2e5f0fbadd72321ce14a56699d73f002216 */\
+/* sqr(t[0], t[0]); */ /* 174: 5cbe1f75bae46439c294acd33ae7e00442c */\
+/* sqr(t[0], t[0]); */ /* 175: b97c3eeb75c8c873852959a675cfc008858 */\
+/* sqr(t[0], t[0]); */ /* 176: 172f87dd6eb9190e70a52b34ceb9f80110b0 */\
+/* sqr(t[0], t[0]); */ /* 177: 2e5f0fbadd72321ce14a56699d73f0022160 */\
+sqr_n_mul(t[0], t[0], 5, t[8]); /* 178: 2e5f0fbadd72321ce14a56699d73f002217f */\
+/* sqr(t[0], t[0]); */ /* 179: 5cbe1f75bae46439c294acd33ae7e00442fe */\
+/* sqr(t[0], t[0]); */ /* 180: b97c3eeb75c8c873852959a675cfc00885fc */\
+/* sqr(t[0], t[0]); */ /* 181: 172f87dd6eb9190e70a52b34ceb9f80110bf8 */\
+/* sqr(t[0], t[0]); */ /* 182: 2e5f0fbadd72321ce14a56699d73f002217f0 */\
+/* sqr(t[0], t[0]); */ /* 183: 5cbe1f75bae46439c294acd33ae7e00442fe0 */\
+/* sqr(t[0], t[0]); */ /* 184: b97c3eeb75c8c873852959a675cfc00885fc0 */\
+/* sqr(t[0], t[0]); */ /* 185: 172f87dd6eb9190e70a52b34ceb9f80110bf80 */\
+/* sqr(t[0], t[0]); */ /* 186: 2e5f0fbadd72321ce14a56699d73f002217f00 */\
+/* sqr(t[0], t[0]); */ /* 187: 5cbe1f75bae46439c294acd33ae7e00442fe00 */\
+/* sqr(t[0], t[0]); */ /* 188: b97c3eeb75c8c873852959a675cfc00885fc00 */\
+sqr_n_mul(t[0], t[0], 10, t[7]); /* 189: b97c3eeb75c8c873852959a675cfc00885fc39 */\
+/* sqr(t[0], t[0]); */ /* 190: 172f87dd6eb9190e70a52b34ceb9f80110bf872 */\
+/* sqr(t[0], t[0]); */ /* 191: 2e5f0fbadd72321ce14a56699d73f002217f0e4 */\
+/* sqr(t[0], t[0]); */ /* 192: 5cbe1f75bae46439c294acd33ae7e00442fe1c8 */\
+/* sqr(t[0], t[0]); */ /* 193: b97c3eeb75c8c873852959a675cfc00885fc390 */\
+/* sqr(t[0], t[0]); */ /* 194: 172f87dd6eb9190e70a52b34ceb9f80110bf8720 */\
+/* sqr(t[0], t[0]); */ /* 195: 2e5f0fbadd72321ce14a56699d73f002217f0e40 */\
+sqr_n_mul(t[0], t[0], 6, t[6]); /* 196: 2e5f0fbadd72321ce14a56699d73f002217f0e67 */\
+/* sqr(t[0], t[0]); */ /* 197: 5cbe1f75bae46439c294acd33ae7e00442fe1cce */\
+/* sqr(t[0], t[0]); */ /* 198: b97c3eeb75c8c873852959a675cfc00885fc399c */\
+/* sqr(t[0], t[0]); */ /* 199: 172f87dd6eb9190e70a52b34ceb9f80110bf87338 */\
+/* sqr(t[0], t[0]); */ /* 200: 2e5f0fbadd72321ce14a56699d73f002217f0e670 */\
+/* sqr(t[0], t[0]); */ /* 201: 5cbe1f75bae46439c294acd33ae7e00442fe1cce0 */\
+sqr_n_mul(t[0], t[0], 5, t[4]); /* 202: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3 */\
+/* sqr(t[0], t[0]); */ /* 203: b97c3eeb75c8c873852959a675cfc00885fc399e6 */\
+/* sqr(t[0], t[0]); */ /* 204: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cc */\
+/* sqr(t[0], t[0]); */ /* 205: 2e5f0fbadd72321ce14a56699d73f002217f0e6798 */\
+/* sqr(t[0], t[0]); */ /* 206: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf30 */\
+/* sqr(t[0], t[0]); */ /* 207: b97c3eeb75c8c873852959a675cfc00885fc399e60 */\
+/* sqr(t[0], t[0]); */ /* 208: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cc0 */\
+/* sqr(t[0], t[0]); */ /* 209: 2e5f0fbadd72321ce14a56699d73f002217f0e67980 */\
+/* sqr(t[0], t[0]); */ /* 210: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf300 */\
+sqr_n_mul(t[0], t[0], 8, t[2]); /* 211: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf333 */\
+/* sqr(t[0], t[0]); */ /* 212: b97c3eeb75c8c873852959a675cfc00885fc399e666 */\
+/* sqr(t[0], t[0]); */ /* 213: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc */\
+/* sqr(t[0], t[0]); */ /* 214: 2e5f0fbadd72321ce14a56699d73f002217f0e679998 */\
+/* sqr(t[0], t[0]); */ /* 215: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3330 */\
+/* sqr(t[0], t[0]); */ /* 216: b97c3eeb75c8c873852959a675cfc00885fc399e6660 */\
+/* sqr(t[0], t[0]); */ /* 217: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc0 */\
+/* sqr(t[0], t[0]); */ /* 218: 2e5f0fbadd72321ce14a56699d73f002217f0e6799980 */\
+sqr_n_mul(t[0], t[0], 7, t[5]); /* 219: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f */\
+/* sqr(t[0], t[0]); */ /* 220: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e */\
+/* sqr(t[0], t[0]); */ /* 221: b97c3eeb75c8c873852959a675cfc00885fc399e6663c */\
+/* sqr(t[0], t[0]); */ /* 222: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78 */\
+/* sqr(t[0], t[0]); */ /* 223: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f0 */\
+/* sqr(t[0], t[0]); */ /* 224: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e0 */\
+/* sqr(t[0], t[0]); */ /* 225: b97c3eeb75c8c873852959a675cfc00885fc399e6663c0 */\
+/* sqr(t[0], t[0]); */ /* 226: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc780 */\
+/* sqr(t[0], t[0]); */ /* 227: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f00 */\
+/* sqr(t[0], t[0]); */ /* 228: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e00 */\
+sqr_n_mul(t[0], t[0], 9, t[2]); /* 229: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e33 */\
+/* sqr(t[0], t[0]); */ /* 230: b97c3eeb75c8c873852959a675cfc00885fc399e6663c66 */\
+/* sqr(t[0], t[0]); */ /* 231: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc */\
+/* sqr(t[0], t[0]); */ /* 232: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f198 */\
+/* sqr(t[0], t[0]); */ /* 233: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e330 */\
+/* sqr(t[0], t[0]); */ /* 234: b97c3eeb75c8c873852959a675cfc00885fc399e6663c660 */\
+/* sqr(t[0], t[0]); */ /* 235: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc0 */\
+/* sqr(t[0], t[0]); */ /* 236: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f1980 */\
+sqr_n_mul(t[0], t[0], 7, t[4]); /* 237: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f1993 */\
+/* sqr(t[0], t[0]); */ /* 238: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e3326 */\
+/* sqr(t[0], t[0]); */ /* 239: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664c */\
+/* sqr(t[0], t[0]); */ /* 240: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc98 */\
+/* sqr(t[0], t[0]); */ /* 241: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19930 */\
+/* sqr(t[0], t[0]); */ /* 242: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e33260 */\
+/* sqr(t[0], t[0]); */ /* 243: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664c0 */\
+/* sqr(t[0], t[0]); */ /* 244: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc980 */\
+/* sqr(t[0], t[0]); */ /* 245: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f199300 */\
+sqr_n_mul(t[0], t[0], 8, t[2]); /* 246: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f199333 */\
+/* sqr(t[0], t[0]); */ /* 247: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e332666 */\
+/* sqr(t[0], t[0]); */ /* 248: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664ccc */\
+/* sqr(t[0], t[0]); */ /* 249: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc9998 */\
+/* sqr(t[0], t[0]); */ /* 250: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f1993330 */\
+/* sqr(t[0], t[0]); */ /* 251: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e3326660 */\
+/* sqr(t[0], t[0]); */ /* 252: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664ccc0 */\
+/* sqr(t[0], t[0]); */ /* 253: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc99980 */\
+/* sqr(t[0], t[0]); */ /* 254: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933300 */\
+sqr_n_mul(t[0], t[0], 8, t[2]); /* 255: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333 */\
+/* sqr(t[0], t[0]); */ /* 256: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e33266666 */\
+/* sqr(t[0], t[0]); */ /* 257: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664ccccc */\
+/* sqr(t[0], t[0]); */ /* 258: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc999998 */\
+/* sqr(t[0], t[0]); */ /* 259: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f199333330 */\
+/* sqr(t[0], t[0]); */ /* 260: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e332666660 */\
+/* sqr(t[0], t[0]); */ /* 261: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664ccccc0 */\
+/* sqr(t[0], t[0]); */ /* 262: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc9999980 */\
+/* sqr(t[0], t[0]); */ /* 263: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f1993333300 */\
+sqr_n_mul(t[0], t[0], 8, t[2]); /* 264: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f1993333333 */\
+/* sqr(t[0], t[0]); */ /* 265: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e3326666666 */\
+/* sqr(t[0], t[0]); */ /* 266: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664ccccccc */\
+/* sqr(t[0], t[0]); */ /* 267: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc99999998 */\
+/* sqr(t[0], t[0]); */ /* 268: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333330 */\
+/* sqr(t[0], t[0]); */ /* 269: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e33266666660 */\
+/* sqr(t[0], t[0]); */ /* 270: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664ccccccc0 */\
+sqr_n_mul(t[0], t[0], 6, t[3]); /* 271: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb */\
+/* sqr(t[0], t[0]); */ /* 272: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc999999996 */\
+/* sqr(t[0], t[0]); */ /* 273: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333332c */\
+/* sqr(t[0], t[0]); */ /* 274: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e332666666658 */\
+/* sqr(t[0], t[0]); */ /* 275: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb0 */\
+/* sqr(t[0], t[0]); */ /* 276: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc9999999960 */\
+/* sqr(t[0], t[0]); */ /* 277: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333332c0 */\
+/* sqr(t[0], t[0]); */ /* 278: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e3326666666580 */\
+/* sqr(t[0], t[0]); */ /* 279: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb00 */\
+sqr_n_mul(t[0], t[0], 8, t[2]); /* 280: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb33 */\
+/* sqr(t[0], t[0]); */ /* 281: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc99999999666 */\
+/* sqr(t[0], t[0]); */ /* 282: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333332ccc */\
+/* sqr(t[0], t[0]); */ /* 283: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e33266666665998 */\
+/* sqr(t[0], t[0]); */ /* 284: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb330 */\
+/* sqr(t[0], t[0]); */ /* 285: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc999999996660 */\
+/* sqr(t[0], t[0]); */ /* 286: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333332ccc0 */\
+/* sqr(t[0], t[0]); */ /* 287: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e332666666659980 */\
+/* sqr(t[0], t[0]); */ /* 288: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb3300 */\
+sqr_n_mul(t[0], t[0], 8, t[2]); /* 289: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb3333 */\
+/* sqr(t[0], t[0]); */ /* 290: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc9999999966666 */\
+/* sqr(t[0], t[0]); */ /* 291: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333332ccccc */\
+/* sqr(t[0], t[0]); */ /* 292: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e3326666666599998 */\
+/* sqr(t[0], t[0]); */ /* 293: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb33330 */\
+/* sqr(t[0], t[0]); */ /* 294: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc99999999666660 */\
+/* sqr(t[0], t[0]); */ /* 295: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333332ccccc0 */\
+/* sqr(t[0], t[0]); */ /* 296: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e33266666665999980 */\
+/* sqr(t[0], t[0]); */ /* 297: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb333300 */\
+sqr_n_mul(t[0], t[0], 8, t[2]); /* 298: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb333333 */\
+/* sqr(t[0], t[0]); */ /* 299: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc999999996666666 */\
+/* sqr(t[0], t[0]); */ /* 300: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333332ccccccc */\
+/* sqr(t[0], t[0]); */ /* 301: 5cbe1f75bae46439c294acd33ae7e00442fe1ccf3331e332666666659999998 */\
+/* sqr(t[0], t[0]); */ /* 302: b97c3eeb75c8c873852959a675cfc00885fc399e6663c664cccccccb3333330 */\
+/* sqr(t[0], t[0]); */ /* 303: 172f87dd6eb9190e70a52b34ceb9f80110bf8733cccc78cc9999999966666660 */\
+/* sqr(t[0], t[0]); */ /* 304: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333332ccccccc0 */\
+sqr_n_mul(out, t[0], 6, t[1]); /* 305: 2e5f0fbadd72321ce14a56699d73f002217f0e679998f19933333332cccccccd */\
+} while(0)
diff --git a/c-source/src/pentaroot.c b/c-source/src/pentaroot.c
new file mode 100644
--- /dev/null
+++ b/c-source/src/pentaroot.c
@@ -0,0 +1,76 @@
+/*
+ * Copyright Supranational LLC
+ * Licensed under the Apache License, Version 2.0, see LICENSE for details.
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+#include "fields.h"
+
+static inline void mul_fr(vec256 ret, const vec256 a, const vec256 b)
+{ mul_mont_sparse_256(ret, a, b, BLS12_381_r, r0); }
+
+static inline void sqr_fr(vec256 ret, const vec256 a)
+{ sqr_mont_sparse_256(ret, a, BLS12_381_r, r0); }
+
+#ifdef __OPTIMIZE_SIZE__
+void blst_fr_pentaroot(vec256 out, const vec256 inp)
+{
+ static const byte pow[] = {
+ TO_BYTES(0x33333332cccccccd), TO_BYTES(0x217f0e679998f199),
+ TO_BYTES(0xe14a56699d73f002), TO_BYTES(0x2e5f0fbadd72321c)
+ };
+ size_t pow_bits = 254;
+ vec256 ret;
+
+ vec_copy(ret, inp, sizeof(ret)); /* ret = inp^1 */
+ --pow_bits; /* most significant bit is set, skip over */
+ while (pow_bits--) {
+ sqr_fr(ret, ret);
+ if (is_bit_set(pow, pow_bits))
+ mul_fr(ret, ret, inp);
+ }
+ vec_copy(out, ret, sizeof(ret)); /* out = ret */
+}
+#else
+# if 0
+/*
+ * "255"-bit variant omits full reductions at the ends of squarings,
+ * not implemented yet[?].
+ */
+static inline void sqr_n_mul_fr(vec256 out, const vec256 a, size_t count,
+ const vec256 b)
+{ sqr_n_mul_mont_255(out, a, count, BLS12_381_r, r0, b); }
+# else
+static void sqr_n_mul_fr(vec256 out, const vec256 a, size_t count,
+ const vec256 b)
+{
+ do {
+ sqr_fr(out, a);
+ a = out;
+ } while (--count);
+ mul_fr(out, out, b);
+}
+# endif
+
+# define sqr(ret,a) sqr_fr(ret,a)
+# define mul(ret,a,b) mul_fr(ret,a,b)
+# define sqr_n_mul(ret,a,n,b) sqr_n_mul_fr(ret,a,n,b)
+
+# include "pentaroot-addchain.h"
+void blst_fr_pentaroot(vec256 out, const vec256 inp)
+{ PENTAROOT_MOD_BLS12_381_r(out, inp, vec256); }
+# undef PENTAROOT_MOD_BLS12_381_r
+
+# undef sqr_n_mul
+# undef sqr
+# undef mul
+#endif
+
+void blst_fr_pentapow(vec256 out, const vec256 inp)
+{
+ vec256 tmp;
+
+ sqr_fr(tmp, inp);
+ sqr_fr(tmp, tmp);
+ mul_fr(out, tmp, inp);
+}
diff --git a/c-source/src/server.c b/c-source/src/server.c
--- a/c-source/src/server.c
+++ b/c-source/src/server.c
@@ -21,4 +21,10 @@
#include "consts.c"
#include "vect.c"
#include "exports.c"
-#include "rb_tree.c"
+#ifndef __BLST_CGO__
+# include "rb_tree.c"
+#endif
+#ifdef BLST_FR_PENTAROOT
+# include "pentaroot.c"
+#endif
+#include "cpuid.c"
diff --git a/c-source/src/vect.h b/c-source/src/vect.h
--- a/c-source/src/vect.h
+++ b/c-source/src/vect.h
@@ -61,7 +61,7 @@
typedef byte pow256[256/8];
/*
- * Internal Boolean type, Bolean by value, hence safe to cast to or
+ * Internal Boolean type, Boolean by value, hence safe to cast to or
* reinterpret as 'bool'.
*/
typedef limb_t bool_t;
@@ -147,7 +147,6 @@
# define mul_mont_384x mulx_mont_384x
# define sqr_mont_384x sqrx_mont_384x
# define sqr_mont_382x sqrx_mont_382x
-# define sqr_n_mul_mont_384x sqrx_n_mul_mont_384x
# define mul_382x mulx_382x
# define sqr_382x sqrx_382x
#endif
@@ -156,8 +155,6 @@
const vec384 p, limb_t n0);
void sqr_mont_384x(vec384x ret, const vec384x a, const vec384 p, limb_t n0);
void sqr_mont_382x(vec384x ret, const vec384x a, const vec384 p, limb_t n0);
-void sqr_n_mul_mont_384x(vec384x ret, const vec384x a, size_t count,
- const vec384 p, limb_t n0, const vec384x b);
void mul_382x(vec768 ret[2], const vec384x a, const vec384x b, const vec384 p);
void sqr_382x(vec768 ret[2], const vec384x a, const vec384 p);
@@ -213,11 +210,25 @@
# endif
#endif
+#if defined(__GNUC__) || defined(__clang__)
+# define launder(var) __asm__ __volatile__("" : "+r"(var))
+#else
+# define launder(var)
+#endif
+
static inline bool_t is_bit_set(const byte *v, size_t i)
-{ return (v[i/8] >> (i%8)) & 1; }
+{
+ bool_t ret = (v[i/8] >> (i%8)) & 1;
+ launder(ret);
+ return ret;
+}
static inline bool_t byte_is_zero(unsigned char c)
-{ return ((limb_t)(c) - 1) >> (LIMB_T_BITS - 1); }
+{
+ limb_t ret = ((limb_t)(c) - 1) >> (LIMB_T_BITS - 1);
+ launder(ret);
+ return ret;
+}
static inline bool_t bytes_are_zero(const unsigned char *a, size_t num)
{
@@ -235,9 +246,12 @@
{
limb_t ai, *ap = (limb_t *)a;
limb_t bi, *bp = (limb_t *)b;
- limb_t xorm, mask = (limb_t)0 - cbit;
+ limb_t xorm, mask;
size_t i;
+ launder(cbit);
+ mask = (limb_t)0 - cbit;
+
num /= sizeof(limb_t);
for (i = 0; i < num; i++) {
@@ -248,6 +262,7 @@
}
/* ret = bit ? a : b */
+void vec_select_32(void *ret, const void *a, const void *b, bool_t sel_a);
void vec_select_48(void *ret, const void *a, const void *b, bool_t sel_a);
void vec_select_96(void *ret, const void *a, const void *b, bool_t sel_a);
void vec_select_144(void *ret, const void *a, const void *b, bool_t sel_a);
@@ -256,8 +271,10 @@
static inline void vec_select(void *ret, const void *a, const void *b,
size_t num, bool_t sel_a)
{
+ launder(sel_a);
#ifndef __BLST_NO_ASM__
- if (num == 48) vec_select_48(ret, a, b, sel_a);
+ if (num == 32) vec_select_32(ret, a, b, sel_a);
+ else if (num == 48) vec_select_48(ret, a, b, sel_a);
else if (num == 96) vec_select_96(ret, a, b, sel_a);
else if (num == 144) vec_select_144(ret, a, b, sel_a);
else if (num == 192) vec_select_192(ret, a, b, sel_a);
@@ -266,7 +283,8 @@
if (0) ;
#endif
else {
- limb_t bi, *rp = (limb_t *)ret;
+ limb_t bi;
+ volatile limb_t *rp = (limb_t *)ret;
const limb_t *ap = (const limb_t *)a;
const limb_t *bp = (const limb_t *)b;
limb_t xorm, mask = (limb_t)0 - sel_a;
@@ -282,7 +300,11 @@
}
static inline bool_t is_zero(limb_t l)
-{ return (~l & (l - 1)) >> (LIMB_T_BITS - 1); }
+{
+ limb_t ret = (~l & (l - 1)) >> (LIMB_T_BITS - 1);
+ launder(ret);
+ return ret;
+}
static inline bool_t vec_is_zero(const void *a, size_t num)
{
@@ -290,6 +312,12 @@
limb_t acc;
size_t i;
+#ifndef __BLST_NO_ASM__
+ bool_t vec_is_zero_16x(const void *a, size_t num);
+ if ((num & 15) == 0)
+ return vec_is_zero_16x(a, num);
+#endif
+
num /= sizeof(limb_t);
for (acc = 0, i = 0; i < num; i++)
@@ -305,6 +333,12 @@
limb_t acc;
size_t i;
+#ifndef __BLST_NO_ASM__
+ bool_t vec_is_equal_16x(const void *a, const void *b, size_t num);
+ if ((num & 15) == 0)
+ return vec_is_equal_16x(a, b, num);
+#endif
+
num /= sizeof(limb_t);
for (acc = 0, i = 0; i < num; i++)
@@ -342,8 +376,8 @@
for (i = 0; i < num; i++)
rp[i] = 0;
-#if defined(__GNUC__) && !defined(__NVCC__)
- asm volatile("" : : "r"(ret) : "memory");
+#if defined(__GNUC__) || defined(__clang__)
+ __asm__ __volatile__("" : : "r"(ret) : "memory");
#endif
}
@@ -364,7 +398,7 @@
# pragma warning(disable: 4127 4189)
#endif
-#if !defined(__wasm__)
+#if !defined(__wasm__) && __STDC_HOSTED__-0 != 0
# include
#endif
diff --git a/hsblst.cabal b/hsblst.cabal
--- a/hsblst.cabal
+++ b/hsblst.cabal
@@ -1,11 +1,11 @@
-cabal-version: 1.12
+cabal-version: 3.0
-- This file has been generated from package.yaml by hpack version 0.34.4.
--
-- see: https://github.com/sol/hpack
name: hsblst
-version: 0.0.2
+version: 0.0.3
synopsis: Haskell bindings to BLST
description: HsBLST is low-level Haskell bindings and a high-level interface to [BLST](https://github.com/supranational/blst) -- a multilingual BLS12-381 signature library.
category: Cryptography
@@ -18,12 +18,10 @@
license-file: LICENSE
build-type: Simple
tested-with:
- GHC == 9.0.2
- , GHC == 9.2.8
- , GHC == 9.4.5
+ GHC == 9.4.8
+ , GHC == 9.6.3
+ , GHC == 9.8.1
extra-source-files:
- README.md
- CHANGELOG.md
c-source/build/coff/add_mod_256-x86_64.s
c-source/build/coff/add_mod_384-x86_64.s
c-source/build/coff/add_mod_384x384-x86_64.s
@@ -99,6 +97,7 @@
c-source/src/client_min_pk.c
c-source/src/client_min_sig.c
c-source/src/consts.c
+ c-source/src/cpuid.c
c-source/src/e1.c
c-source/src/e2.c
c-source/src/exp.c
@@ -110,6 +109,7 @@
c-source/src/map_to_g2.c
c-source/src/multi_scalar.c
c-source/src/pairing.c
+ c-source/src/pentaroot.c
c-source/src/rb_tree.c
c-source/src/recip.c
c-source/src/server.c
@@ -122,6 +122,7 @@
c-source/src/errors.h
c-source/src/fields.h
c-source/src/no_asm.h
+ c-source/src/pentaroot-addchain.h
c-source/src/point.h
c-source/src/recip-addchain.h
c-source/src/sha256.h
@@ -129,6 +130,9 @@
c-source/src/vect.h
c-source/bindings/blst.h
c-source/bindings/blst_aux.h
+extra-doc-files:
+ README.md
+ CHANGELOG.md
c-source/LICENSE
c-source/README.md
@@ -146,6 +150,8 @@
Crypto.BLST.Internal.Types
other-modules:
Paths_hsblst
+ autogen-modules:
+ Paths_hsblst
hs-source-dirs:
src
default-extensions:
@@ -201,16 +207,15 @@
c-source/bindings
c-sources:
c-source/src/server.c
- c-source/build/assembly.S
- build-tools:
- c2hs
build-depends:
- base >=4.15.1.0 && <4.18
- , deepseq >=1.4.5.0 && <1.5
+ base >=4.17 && <4.20
+ , deepseq >=1.4.5.0 && <1.5.1
, memory >=0.16.0 && <0.19
if impl(ghc >= 9.2)
ghc-options: -Wno-missing-kind-signatures
default-language: Haskell2010
+ asm-sources: c-source/build/assembly.S
+ build-tool-depends: c2hs:c2hs
test-suite hsblst-test
type: exitcode-stdio-1.0
@@ -222,6 +227,8 @@
Test.BLST.Tezos
Test.BLST.Util
Paths_hsblst
+ autogen-modules:
+ Paths_hsblst
hs-source-dirs:
test
default-extensions:
@@ -276,14 +283,14 @@
build-tool-depends:
tasty-discover:tasty-discover
build-depends:
- base >=4.15.1.0 && <4.18
- , bytestring >=0.10.12.1 && <0.12
- , hex-text >=0.1.0.6 && <0.2
+ base >=4.17 && <4.20
+ , base16-bytestring ==1.0.*
+ , bytestring >=0.10.12.1 && <0.13
, hsblst
, memory >=0.16.0 && <0.19
- , tasty >=1.4.2.1 && <1.5
+ , tasty >=1.4.2.1 && <1.6
, tasty-hunit >=0.10.0.3 && <0.11
- , text >=1.2.5.0 && <1.3 || >=2.0.1 && <2.1
+ , text >=1.2.5.0 && <1.3 || >=2.0.1 && <2.1 || >=2.1 && <2.2
if impl(ghc >= 9.2)
ghc-options: -Wno-missing-kind-signatures
if impl(ghc < 9.4)
diff --git a/src/Crypto/BLST/Internal/Bindings.chs b/src/Crypto/BLST/Internal/Bindings.chs
--- a/src/Crypto/BLST/Internal/Bindings.chs
+++ b/src/Crypto/BLST/Internal/Bindings.chs
@@ -4,6 +4,7 @@
#include
{-# OPTIONS_HADDOCK not-home #-}
+{-# OPTIONS_GHC -Wno-orphans #-}
-- | Lower-level bindings. Functions starting with @blst_@ are raw c2hs
-- bindings. Others are slightly higher level wrappers around those bindings.
diff --git a/src/Crypto/BLST/Internal/Bindings/Types.hs b/src/Crypto/BLST/Internal/Bindings/Types.hs
--- a/src/Crypto/BLST/Internal/Bindings/Types.hs
+++ b/src/Crypto/BLST/Internal/Bindings/Types.hs
@@ -2,6 +2,7 @@
-- SPDX-License-Identifier: MPL-2.0
{-# OPTIONS_HADDOCK not-home #-}
+{-# LANGUAGE RoleAnnotations #-}
-- | Types for lower-level bindings
module Crypto.BLST.Internal.Bindings.Types
@@ -29,6 +30,7 @@
type SizeOf :: Type -> Nat
type family SizeOf t
+type role Point nominal
-- | Point representation.
newtype Point (a :: PointKind) = Point ( SizedByteArray (SizeOf (Point a)) Bytes )
deriving stock (Show, Eq)
@@ -36,6 +38,7 @@
instance NFData (Point a) where
rnf = rnf @Bytes . unSizedByteArray . coerce
+type role Affine nominal
-- | Affine point representation.
newtype Affine (a :: PointKind) = Affine { unAffine :: SizedByteArray (SizeOf (Affine a)) Bytes }
deriving stock (Show, Eq)
diff --git a/src/Crypto/BLST/Internal/Types.hs b/src/Crypto/BLST/Internal/Types.hs
--- a/src/Crypto/BLST/Internal/Types.hs
+++ b/src/Crypto/BLST/Internal/Types.hs
@@ -2,6 +2,7 @@
-- SPDX-License-Identifier: MPL-2.0
{-# OPTIONS_HADDOCK not-home #-}
+{-# LANGUAGE RoleAnnotations #-}
-- | Types used for the high-level interface.
module Crypto.BLST.Internal.Types
@@ -35,6 +36,7 @@
type instance ByteSize 'Serialize SecretKey = SkSerializeSize
+type role PublicKey nominal
-- | Public key representation.
type PublicKey :: Curve -> Type
newtype PublicKey c = PublicKey (Affine (CurveToPkPoint c))
@@ -44,6 +46,7 @@
type instance ByteSize 'Serialize (PublicKey c) = SerializedSize (CurveToPkPoint c)
type instance ByteSize 'Compress (PublicKey c) = CompressedSize (CurveToPkPoint c)
+type role Signature nominal phantom
-- | Signature representation.
type Signature :: Curve -> EncodeMethod -> Type
newtype Signature c m = Signature (Affine (CurveToMsgPoint c))
diff --git a/test/Test/BLST/Util.hs b/test/Test/BLST/Util.hs
--- a/test/Test/BLST/Util.hs
+++ b/test/Test/BLST/Util.hs
@@ -16,12 +16,13 @@
import Data.ByteArray qualified as BA
import Data.ByteArray.Sized (SizedByteArray)
import Data.ByteArray.Sized qualified as AS
-import Data.Maybe (fromJust, fromMaybe)
+import Data.ByteString.Base16 qualified as B16
+import Data.Maybe (fromMaybe)
import Data.Proxy (Proxy(..))
import Data.Text (Text, unpack)
+import Data.Text.Encoding (decodeUtf8, encodeUtf8)
import GHC.TypeNats (KnownNat, natVal)
import System.IO.Unsafe (unsafePerformIO)
-import Text.Hex (decodeHex, encodeHex)
import Crypto.BLST.Internal.Bindings qualified as B
import Crypto.BLST.Internal.Classy qualified as C
@@ -29,14 +30,14 @@
fromHex :: forall n. (KnownNat n) => Text -> SizedByteArray n Bytes
fromHex arg = AS.convert
. fromMaybe (error $ "Expected " <> show (natVal @n Proxy) <> " bytes got " <> unpack arg)
- . AS.sizedByteArray . fromJust . decodeHex
+ . AS.sizedByteArray . B16.decodeLenient . encodeUtf8
$ arg
fromHex' :: BA.ByteArray a => Text -> a
-fromHex' = BA.convert . fromJust . decodeHex
+fromHex' = BA.convert . B16.decodeLenient . encodeUtf8
toHex :: BA.ByteArray a => SizedByteArray n a -> Text
-toHex = encodeHex . BA.convert . AS.unSizedByteArray
+toHex = decodeUtf8 . B16.encode . BA.convert . AS.unSizedByteArray
deserializePoint :: C.IsPoint p => Text -> B.Point p
deserializePoint = unsafePerformIO . C.fromAffine . deserializeAffine