mbedtls: fix hw accelerated big-num mul if operand and result overlap

this issue is mainly exposed when using larger (4096) client key in TLS mutual auth,
since it uses multiplications > 2048 when mbedtls_mpi_mul_mpi is used in recursion,
which works only if both operands point to different location than result since
mpi_mult_mpi_overlong() called mbedtls_mpi_grow() to reallocate buffers used in previous
pointer arithmetics and thus corrupting it. Fixed by growing the mpi buffer before
calling mpi_mult_mpi_overlong()

components/mbedtls/port/esp32/esp_bignum.c

@@ -564,6 +564,9 @@ int mbedtls_mpi_mul_mpi( mbedtls_mpi *Z, const mbedtls_mpi *X, const mbedtls_mpi
         return ret;
     }
 
+    /* Grow Z to result size early, avoid interim allocations */
+    MBEDTLS_MPI_CHK( mbedtls_mpi_grow(Z, z_words) );
+
     /* If either factor is over 2048 bits, we can't use the standard hardware multiplier
        (it assumes result is double longest factor, and result is max 4096 bits.)
 
@@ -608,8 +611,6 @@ int mbedtls_mpi_mul_mpi( mbedtls_mpi *Z, const mbedtls_mpi *X, const mbedtls_mpi
 
     start_op(RSA_MULT_START_REG);
 
-    MBEDTLS_MPI_CHK( mbedtls_mpi_grow(Z, z_words) );
-
     wait_op_complete(RSA_MULT_START_REG);
 
     /* Read back the result */
@@ -716,9 +717,6 @@ static int mpi_mult_mpi_overlong(mbedtls_mpi *Z, const mbedtls_mpi *X, const mbe
     };
     mbedtls_mpi_init(&Ztemp);
 
-    /* Grow Z to result size early, avoid interim allocations */
-    mbedtls_mpi_grow(Z, z_words);
-
     /* Get result Ztemp = Yp * X (need temporary variable Ztemp) */
     MBEDTLS_MPI_CHK( mbedtls_mpi_mul_mpi(&Ztemp, X, &Yp) );
 

components/mbedtls/port/esp32s2beta/esp_bignum.c

@@ -449,6 +449,7 @@ int mbedtls_mpi_mul_mpi( mbedtls_mpi *Z, const mbedtls_mpi *X, const mbedtls_mpi
             return mpi_mult_mpi_failover_mod_mult(Z, X, Y, z_words);
         } else {
             /* Still too long for the hardware unit... */
+            mbedtls_mpi_grow(Z, z_words);
             if(y_words > x_words) {
                 return mpi_mult_mpi_overlong(Z, X, Y, y_words, z_words);
             } else {
@@ -573,9 +574,6 @@ static int mpi_mult_mpi_overlong(mbedtls_mpi *Z, const mbedtls_mpi *X, const mbe
     };
     mbedtls_mpi_init(&Ztemp);
 
-    /* Grow Z to result size early, avoid interim allocations */
-    mbedtls_mpi_grow(Z, z_words);
-
     /* Get result Ztemp = Yp * X (need temporary variable Ztemp) */
     MBEDTLS_MPI_CHK( mbedtls_mpi_mul_mpi(&Ztemp, X, &Yp) );