fold-const.c (fold_binary): Do not fold multiplication by 1 or -1 for complex floating-point types if...

	* fold-const.c (fold_binary): Do not fold multiplication by 1 or
	-1 for complex floating-point types if honoring signed zeros.

testsuite:
	* gcc.dg/torture/complex-sign-mul-minus-one.c,
	gcc.dg/torture/complex-sign-mul-one.c: New tests.

From-SVN: r147295