------- Comment #6 from pault at gcc dot gnu dot org 2009-02-03 19:59 ------- I have just realised that this is a case of complete overlap that we miss completely in dependency analysis:
If one of the lhs or rhs is a full array, the stride is unity and one of lbound == start or ubound == end, then the arrays overlap. I feel a fix coming on.... Paul -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38863
