Merge branch 'refactor-ecc-mod' into master-updates

diff --cc ChangeLog
index 218fa39452df3897233c6ce9574dfb2eaae0228b,9dfd044cfbeed51a50e14263b6869a4ee79a72b7..fc8f9fc76648b555743e6552ff41b91e5a2a8681
--- 1/ChangeLog
--- 2/ChangeLog
+++ b/ChangeLog
@@@ -1,20 -1,90 +1,108 @@@
+ 2020-11-08  Niels Möller  <nisse@lysator.liu.se>
+ 
++      Merge refactoring of ecc modulo and reduce functions.
+       * eddsa-sign.c (_eddsa_sign_itch): Update, since now point
+       multiplication needs less scratch than point compression.
+       * eddsa-pubkey.c (_eddsa_public_key_itch): Likewise.
+ 
+       * ecc-internal.h: Update *_ITCH macros for point multiplication
+       and signatures. They need slightly less scratch after optimization
+       of the point addition functions.
+ 
+       * ecc-mul-m.c (ecc_mul_m): Reduce scratch need.
+       (ecc_mul_m): Optimize swapping, with only a single mpn_cnd_swap
+       per iteration.
+ 
+       * ecc-add-jja.c (ecc_add_jja): Reduce scratch need.
+       * ecc-add-jjj.c (ecc_add_jjj): Reduce scratch need.
+       * ecc-internal.h (ECC_ADD_JJA_ITCH, ECC_ADD_JJJ_ITCH): Now 5*size.
+       (ECC_MUL_M_ITCH): New 8*size.
+ 
+ 2020-11-06  Niels Möller  <nisse@lysator.liu.se>
+ 
+       After these changes, both curve25519 and curve448 need 4*size for
+       invert and 6*size for sqrt.
+       * ecc-curve448.c (ecc_mod_pow_446m224m1): Reduce scratch need.
+       (ecc_curve448_inv): Likewise.
+       (ecc_curve448_sqrt): Likewise.
+       * ecc-curve25519.c (ecc_curve25519_sqrt): Reduce scratch need.
+ 
+       * ecc-add-jja.c (ecc_add_jja): Delete an unneeded copy.
+ 
+ 2020-11-05  Niels Möller  <nisse@lysator.liu.se>
+ 
+       * ecc-dup-jj.c (ecc_dup_jj): Reduce scratch need.
+       * ecc-internal.h (ECC_DUP_JJ_ITCH): Now 4*size.
+ 
+ 2020-11-03  Niels Möller  <nisse@lysator.liu.se>
+ 
+       * ecc-dup-eh.c (ecc_dup_eh): Reduce scratch need.
+       * ecc-dup-th.c (ecc_dup_th): Analogous changes.
+       * ecc-internal.h (ECC_DUP_EH_ITCH, ECC_DUP_TH_ITCH): Now 3*size.
+ 
+       * ecc-internal.h (ecc_add_func): Document in-place operation.
+       * ecc-mul-a-eh.c (ecc_mul_a_eh): Fix call to ecc->add_hhh accordingly.
+       * testsuite/ecc-add-test.c (test_main): Likewise.
+ 
+       * ecc-add-eh.c (ecc_add_eh): Reduce scratch need.
+       * ecc-add-th.c (ecc_add_th): Analogous changes.
+       * ecc-add-ehh.c (ecc_add_ehh): Reduce scratch need.
+       * ecc-add-thh.c (ecc_add_thh): Analogous changes.
+       * ecc-internal.h (ECC_ADD_EH_ITCH, ECC_ADD_EHH_ITCH)
+       (ECC_ADD_TH_ITCH, ECC_ADD_THH_ITCH): Now 4*size.
+ 
+ 2020-11-02  Niels Möller  <nisse@lysator.liu.se>
+ 
+       * ecc-curve25519.c (ecc_mod_pow_252m3): Reduce scratch need.
+       (ecc_curve25519_inv): Likewise.
+       (ecc_curve25519_sqrt): Likewise.
+ 
+ 2020-11-01  Niels Möller  <nisse@lysator.liu.se>
+ 
+       * ecc-mod-arith.c (ecc_mod_mul, ecc_mod_sqr): Separate argument
+       for scratch area, reducing required size of result area. Update
+       all callers to naïvely keep using result in scratch area.
+       (ecc_mod_pow_2k, ecc_mod_pow_2k_mul): Simplified, also reducing
+       required size of result area.
+ 
+       * testsuite/testutils.c (test_ecc_point): Show curve bits on failure.
+ 
+ 2020-10-31  Niels Möller  <nisse@lysator.liu.se>
+ 
+       * ecc-internal.h (typedef ecc_mod_func): Updated all assembly
+       implementations.
+ 
+       * testsuite/ecc-mod-test.c (test_one): Extend tests, to also test
+       with different destination area.
+       * testsuite/ecc-redc-test.c (test_main): Likewise.
+ 
+ 2020-10-30  Niels Möller  <nisse@lysator.liu.se>
+ 
+       * ecc-internal.h (typedef ecc_mod_func): Add separate result
+       argument. Updated all C implementations and callers.
+ 
+ 2020-10-29  Niels Möller  <nisse@lysator.liu.se>
+ 
+       * ecc-mod.c (ecc_mod): More unified handling of final carry
+       folding. Also eliminates a goto statement.
+ 
 +2020-11-07  Niels Möller  <nisse@lysator.liu.se>
 +
 +      Merged initial powerpc64 implementation of chacha.
 +      * configure.ac: New command line option --enable-power-altivec.
 +      Update asm_path logic, and add altivec to FAT_TEST_LIST.
 +      * fat-ppc.c (get_ppc_features): Add logic to check for altivec and
 +      vsx support, and select aither C or altivec implementation of
 +      chacha_core.
 +      * powerpc64/p7/chacha-core-internal.asm: New file.
 +
 +2020-09-25  Niels Möller  <nisse@lysator.liu.se>
 +
 +      * powerpc64/p7/chacha-core-internal.asm: New file.
 +      * Makefile.in (distdir): Add powerpc64/p7.
 +
 +2020-10-29  Niels Möller  <nisse@lysator.liu.se>
 +
        * blowfish.c (blowfish_set_key): Add casts to uint32_t. Avoids
        undefined behavior, since shifting an 8-bit value left by 24 bits
        overflows the range of signed int. Reported by Guido Vranken.