------- Additional Comments From tobi at gcc dot gnu dot org 2005-01-21 22:58
-------
Unfortunately, this patch is not sufficient to fix this issue, as the logical
constants are not dealt with correctly by the middle-end.
The following testcase shouldn't abort:
program test
implicit none
logical(1), parameter :: t1 = .TRUE., f1 = .FALSE.
logical(2), parameter :: t2 = .TRUE., f2 = .FALSE.
logical(4), parameter :: t4 = .TRUE., f4 = .FALSE.
logical(8), parameter :: t8 = .TRUE., f8 = .FALSE.
character*2 :: t(4), f(4)
write(t(1),*) t1
write(f(1),*) f1
write(t(2),*) t2
write(f(2),*) f2
write(t(3),*) t4
write(f(3),*) f4
write(t(4),*) t8
write(f(4),*) f8
if (any(t .ne. " T")) call abort
if (any(f .ne. " F")) call abort
end
It does, with or without my patch. Without my patch it only fails for f(4),
i.e. for LOGICAL*8, with my patch it fails for f(2) through to f(4), i.e. for
all logical types which are not byte-sized.
While the .original-dump looks correct with my patch, and not correct without
it, the logical constants get merged in an unhealthy way:
.align 4
.LC1:
.long 1
.align 4
.LC2:
.long 0
(without my patch, with my patch the alignments disappear, which of course
explains why this fails for anything wider than 32 bits, or byte respectively)
To the potential middle-end hacker looking into this: Fortran's LOGICAL is a
boolean type which is defined as:
new_type = make_unsigned_type (bit_size);
TREE_SET_CODE (new_type, BOOLEAN_TYPE);
TYPE_MAX_VALUE (new_type) = build_int_cst (new_type, 1);
TYPE_PRECISION (new_type) = 1;
(from trans-types.c, around line 420)
bit_size is one of 8, 16, 32, or 64.
--
What |Removed |Added
----------------------------------------------------------------------------
AssignedTo|tobi at gcc dot gnu dot org |unassigned at gcc dot gnu
| |dot org
Status|ASSIGNED |NEW
Keywords|patch |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19543
