------- Comment #10 from jvdelisle at gcc dot gnu dot org 2008-01-29 03:33 ------- Here is a modified test case that illustrates the problem better. It was just dumb luck that the test case was passing on some platforms. The following segfaults on x86-64 with current trunk.
module abuse_mod implicit none integer i character(8), parameter :: HEX1 = '40490FDB' integer(1), parameter :: MSKa1(len(HEX1)/2) = (/(i,i=1,len(HEX1),2)/) integer(1), parameter :: ARR1(len(HEX1)/2) = (/( MSKa1(i), i=1,len(HEX1),2)/) end module abuse_mod program test use abuse_mod print '(8i2)', (/(i,i=1,len(HEX1),2)/) print '(8i2)', MSKa1 print '(8i2)', ARR1 end program test Just the right memory locations in ar->as->upper[i]->value.integer happen to be zero. There is no ar->as->upper[i]->value.integer in this test case. ar->as->upper[i] has expr_type == EXPR_FUNCTION. Now to figure out the proper fix. ;) -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34828
