> > Now afunc writes to __var_5_mma only indirectly so I think it is correct
> > that
> > we optimize the conditional out.
> >
> > Easy fix would be to add -fno-ipa-modref, but perhaps someone with
> > better understanding of Fortran would help me to improve the testcase so
> > the calls to matmul_r4 remains reachable?
>
> I think the two matmul_r4 cases were missed optimizations before so just
> changing the expected number of calls to zero is the correct fix here. Indeed
> we can now statically determine the matrices are not large and so only
> keep the inline copy.
I have updated the matmul as follows.
gcc/testsuite/ChangeLog:
2021-11-11 Jan Hubicka <hubi...@ucw.cz>
* gfortran.dg/inline_matmul_17.f90: Fix template
diff --git a/gcc/testsuite/gfortran.dg/inline_matmul_17.f90
b/gcc/testsuite/gfortran.dg/inline_matmul_17.f90
index d2ca8e2948a..cff4b6ce5e2 100644
--- a/gcc/testsuite/gfortran.dg/inline_matmul_17.f90
+++ b/gcc/testsuite/gfortran.dg/inline_matmul_17.f90
@@ -45,4 +45,4 @@ program main
c = matmul(a, bfunc())
if (any(c-d /= 0)) STOP 6
end program main
-! { dg-final { scan-tree-dump-times "matmul_r4" 2 "optimized" } }
+! { dg-final { scan-tree-dump-not "matmul_r4" "optimized" } }
