aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--gcc/testsuite/ChangeLog5
-rw-r--r--gcc/testsuite/gfortran.dg/matmul_6.f9066
-rw-r--r--libgfortran/ChangeLog8
-rw-r--r--libgfortran/generated/matmul_l16.c4
-rw-r--r--libgfortran/generated/matmul_l4.c4
-rw-r--r--libgfortran/generated/matmul_l8.c4
-rw-r--r--libgfortran/m4/matmull.m44
7 files changed, 87 insertions, 8 deletions
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index 9f5aa26..4bfd523 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,8 @@
+2007-12-25 Thomas Koenig <tkoenig@gcc.gnu.org>
+
+ PR libfortran/34566
+ * gfortran.dg/matmul_6.f90: New test.
+
2007-12-23 Tobias Burnus <burnus@net-b.de>
PR fortran/34421
diff --git a/gcc/testsuite/gfortran.dg/matmul_6.f90 b/gcc/testsuite/gfortran.dg/matmul_6.f90
new file mode 100644
index 0000000..737c5c4
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/matmul_6.f90
@@ -0,0 +1,66 @@
+! { dg-do run }
+! PR 34566 - logical matmul used to give the wrong result.
+! We check this by running through every permutation in
+! multiplying two 3*3 matrices, and all permutations of multiplying
+! a 3-vector and a 3*3 matrices and checking against equivalence
+! with integer matrix multiply.
+program main
+ implicit none
+ integer, parameter :: ki=4
+ integer, parameter :: dimen=3
+ integer :: i, j, k
+ real, dimension(dimen,dimen) :: r1, r2
+ integer, dimension(dimen,dimen) :: m1, m2
+ logical(kind=ki), dimension(dimen,dimen) :: l1, l2
+ logical(kind=ki), dimension(dimen*dimen) :: laux
+ logical(kind=ki), dimension(dimen) :: lv
+ integer, dimension(dimen) :: iv
+
+ do i=0,2**(dimen*dimen)-1
+ forall (k=1:dimen*dimen)
+ laux(k) = btest(i, k-1)
+ end forall
+ l1 = reshape(laux,shape(l1))
+ m1 = ltoi(l1)
+
+ ! Check matrix*matrix multiply
+ do j=0,2**(dimen*dimen)-1
+ forall (k=1:dimen*dimen)
+ laux(k) = btest(i, k-1)
+ end forall
+ l2 = reshape(laux,shape(l2))
+ m2 = ltoi(l2)
+ if (any(matmul(l1,l2) .neqv. (matmul(m1,m2) /= 0))) then
+ call abort
+ end if
+ end do
+
+ ! Check vector*matrix and matrix*vector multiply.
+ do j=0,2**dimen-1
+ forall (k=1:dimen)
+ lv(k) = btest(j, k-1)
+ end forall
+ iv = ltoi(lv)
+ if (any(matmul(lv,l1) .neqv. (matmul(iv,m1) /=0))) then
+ call abort
+ end if
+ if (any(matmul(l1,lv) .neqv. (matmul(m1,iv) /= 0))) then
+ call abort
+ end if
+ end do
+ end do
+
+contains
+ elemental function ltoi(v)
+ implicit none
+ integer :: ltoi
+ real :: rtoi
+ logical(kind=4), intent(in) :: v
+ if (v) then
+ ltoi = 1
+ else
+ ltoi = 0
+ end if
+ end function ltoi
+
+end program main
diff --git a/libgfortran/ChangeLog b/libgfortran/ChangeLog
index 9d84e1b..8f32ca0 100644
--- a/libgfortran/ChangeLog
+++ b/libgfortran/ChangeLog
@@ -1,3 +1,11 @@
+2007-12-25 Thomas Koenig <tkoenig@gcc.gnu.org>
+
+ PR libfortran/34566
+ * m4/matmull.m4: Multiply xstride and ystride by correct kind.
+ * generated/matmul_l4.c: Regenerated.
+ * generated/matmul_l8.c: Regenerated.
+ * generated/matmul_l16.c: Regenerated.
+
2007-12-19 Tobias Burnus <burnus@net-b.de>
PR fortran/34530
diff --git a/libgfortran/generated/matmul_l16.c b/libgfortran/generated/matmul_l16.c
index c3cabdb..b2b86ec 100644
--- a/libgfortran/generated/matmul_l16.c
+++ b/libgfortran/generated/matmul_l16.c
@@ -152,7 +152,7 @@ matmul_l16 (gfc_array_l16 * const restrict retarray,
{
astride = a->dim[1].stride * a_kind;
count = a->dim[1].ubound + 1 - a->dim[1].lbound;
- xstride = a->dim[0].stride;
+ xstride = a->dim[0].stride * a_kind;
xcount = a->dim[0].ubound + 1 - a->dim[0].lbound;
}
if (GFC_DESCRIPTOR_RANK (b) == 1)
@@ -167,7 +167,7 @@ matmul_l16 (gfc_array_l16 * const restrict retarray,
{
bstride = b->dim[0].stride * b_kind;
assert(count == b->dim[0].ubound + 1 - b->dim[0].lbound);
- ystride = b->dim[1].stride;
+ ystride = b->dim[1].stride * b_kind;
ycount = b->dim[1].ubound + 1 - b->dim[1].lbound;
}
diff --git a/libgfortran/generated/matmul_l4.c b/libgfortran/generated/matmul_l4.c
index 531603b..9a6cb1d 100644
--- a/libgfortran/generated/matmul_l4.c
+++ b/libgfortran/generated/matmul_l4.c
@@ -152,7 +152,7 @@ matmul_l4 (gfc_array_l4 * const restrict retarray,
{
astride = a->dim[1].stride * a_kind;
count = a->dim[1].ubound + 1 - a->dim[1].lbound;
- xstride = a->dim[0].stride;
+ xstride = a->dim[0].stride * a_kind;
xcount = a->dim[0].ubound + 1 - a->dim[0].lbound;
}
if (GFC_DESCRIPTOR_RANK (b) == 1)
@@ -167,7 +167,7 @@ matmul_l4 (gfc_array_l4 * const restrict retarray,
{
bstride = b->dim[0].stride * b_kind;
assert(count == b->dim[0].ubound + 1 - b->dim[0].lbound);
- ystride = b->dim[1].stride;
+ ystride = b->dim[1].stride * b_kind;
ycount = b->dim[1].ubound + 1 - b->dim[1].lbound;
}
diff --git a/libgfortran/generated/matmul_l8.c b/libgfortran/generated/matmul_l8.c
index 0b9b0be..7d4e35e 100644
--- a/libgfortran/generated/matmul_l8.c
+++ b/libgfortran/generated/matmul_l8.c
@@ -152,7 +152,7 @@ matmul_l8 (gfc_array_l8 * const restrict retarray,
{
astride = a->dim[1].stride * a_kind;
count = a->dim[1].ubound + 1 - a->dim[1].lbound;
- xstride = a->dim[0].stride;
+ xstride = a->dim[0].stride * a_kind;
xcount = a->dim[0].ubound + 1 - a->dim[0].lbound;
}
if (GFC_DESCRIPTOR_RANK (b) == 1)
@@ -167,7 +167,7 @@ matmul_l8 (gfc_array_l8 * const restrict retarray,
{
bstride = b->dim[0].stride * b_kind;
assert(count == b->dim[0].ubound + 1 - b->dim[0].lbound);
- ystride = b->dim[1].stride;
+ ystride = b->dim[1].stride * b_kind;
ycount = b->dim[1].ubound + 1 - b->dim[1].lbound;
}
diff --git a/libgfortran/m4/matmull.m4 b/libgfortran/m4/matmull.m4
index b488632..54afa8a 100644
--- a/libgfortran/m4/matmull.m4
+++ b/libgfortran/m4/matmull.m4
@@ -154,7 +154,7 @@ sinclude(`matmul_asm_'rtype_code`.m4')dnl
{
astride = a->dim[1].stride * a_kind;
count = a->dim[1].ubound + 1 - a->dim[1].lbound;
- xstride = a->dim[0].stride;
+ xstride = a->dim[0].stride * a_kind;
xcount = a->dim[0].ubound + 1 - a->dim[0].lbound;
}
if (GFC_DESCRIPTOR_RANK (b) == 1)
@@ -169,7 +169,7 @@ sinclude(`matmul_asm_'rtype_code`.m4')dnl
{
bstride = b->dim[0].stride * b_kind;
assert(count == b->dim[0].ubound + 1 - b->dim[0].lbound);
- ystride = b->dim[1].stride;
+ ystride = b->dim[1].stride * b_kind;
ycount = b->dim[1].ubound + 1 - b->dim[1].lbound;
}
generated by cgit v1.1 at 2024-11-07 19:22:14 +0000