Merge 8.7

3 files changed, 10 insertions, 2 deletions

diff --git a/generic/tclTomMathDecls.h b/generic/tclTomMathDecls.h
index 9a13a1a..d6d8baf 100644
--- a/generic/tclTomMathDecls.h
+++ b/generic/tclTomMathDecls.h
@@ -589,6 +589,9 @@ extern const TclTomMathStubs *tclTomMathStubsPtr;
 #undef mp_isodd
 #define mp_iseven(a) (!mp_isodd(a))
 #define mp_isodd(a)  (((a)->used != 0 && (((a)->dp[0] & 1) != 0)) ? MP_YES : MP_NO)
+#undef mp_sqr
+#define mp_sqr(a,b) mp_mul(a,a,b)
+
 #undef TCL_STORAGE_CLASS
 #define TCL_STORAGE_CLASS DLLIMPORT
 
diff --git a/libtommath/bn_mp_mul.c b/libtommath/bn_mp_mul.c
index 91707cd..b00334d 100644
--- a/libtommath/bn_mp_mul.c
+++ b/libtommath/bn_mp_mul.c
@@ -12,12 +12,14 @@ mp_err mp_mul(const mp_int *a, const mp_int *b, mp_int *c)
        digs = a->used + b->used + 1;
    mp_sign neg = (a->sign == b->sign) ? MP_ZPOS : MP_NEG;
 
-   if (MP_HAS(S_MP_BALANCE_MUL) &&
+   if (a == b) {
+       return mp_sqr(a,c);
+   } else if (MP_HAS(S_MP_BALANCE_MUL) &&
        /* Check sizes. The smaller one needs to be larger than the Karatsuba cut-off.
         * The bigger one needs to be at least about one MP_KARATSUBA_MUL_CUTOFF bigger
         * to make some sense, but it depends on architecture, OS, position of the
         * stars... so YMMV.
-        * Using it to cut the input into slices small enough for s_mp_mul_digs_fast
+        * Using it to cut the input into slices small enough for fast_s_mp_mul_digs
         * was actually slower on the author's machine, but YMMV.
         */
        (min_len >= MP_KARATSUBA_MUL_CUTOFF) &&
diff --git a/libtommath/tommath_private.h b/libtommath/tommath_private.h
index 7cef443..60c8838 100644
--- a/libtommath/tommath_private.h
+++ b/libtommath/tommath_private.h
@@ -249,6 +249,9 @@ MP_DEPRECATED(s_mp_toom_mul) mp_err mp_toom_mul(const mp_int *a, const mp_int *b
 MP_DEPRECATED(s_mp_toom_sqr) mp_err mp_toom_sqr(const mp_int *a, mp_int *b);
 MP_DEPRECATED(s_mp_reverse) void bn_reverse(unsigned char *s, int len);
 
+#undef mp_sqr
+#define mp_sqr TclBN_mp_sqr
+
 #define MP_GET_ENDIANNESS(x) \
    do{\
       int16_t n = 0x1;                                          \