http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56318
Tobias Burnus <burnus at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |missed-optimization
--- Comment #2 from Tobias Burnus <burnus at gcc dot gnu.org> 2013-02-14
11:29:55 UTC ---
Simplified test case:
!------------------------------------
integer, parameter :: A(3,2) = reshape([1,2,3,4,5,6],[3,2])
integer, parameter :: B(2,3) = reshape([1,1,1,1,1,1],[2,3])
integer, parameter :: m1 = 1
print '(3i3)', m1*matmul(A,B)
end
!------------------------------------
Expected:
5 7 9
5 7 9
5 7 9
Actual result:
9 12 8
9 12 8
4 6 8
a) BUG 1: The matrix is not always simplified
The "m1*" is crucial. When gfc_simplify_matmul is called initially, it returns
NULL as "is_constant_array_expr (matrix_a)" is false.
The "m1*" causes a re-evaluation of the RHS expression, namely in
gfc_simplify_expr:
for (ap = p->value.function.actual; ap; ap = ap->next)
if (gfc_simplify_expr (ap->expr, type) == FAILURE)
That converts an EXPR_VARIABLE with flavor FL_PARAMETER into an EXPR_ARRAY,
which can then be processed by calling gfc_simplify_matmul
b) BUG 2: matmul gives the wrong result
