[PATCH] Fix COMPONENT_REF expansion (PR rtl-optimization/77919)

[Jakub Jelinek]

Hi!

The following testcase ICEs on x86_64-linux with -O1, the problem is
that we expand assignment from COMPONENT_REF of MEM_REF into a V4SImode
SSA_NAME.  The MEM_REF has non-addressable DCmode var inside of it, and
type of a struct containing a single V4SImode element.

The bug seems to be that if the op0 (i.e. get_inner_reference expanded)
is a CONCAT and we want a reference that covers all bits of the CONCAT,
we just short path it and return it immediately, rather than trying
to convert it to the requested mode.

I've bootstrapped/regtested on x86_64-linux and i686-linux the following
patch, which takes the short path only if we want a complex mode.
The only place it makes a difference in both bootstraps/regtests was this
new testcase.

Though, perhaps COMPLEX_MODE_P (mode1) is also wrong, if mode1 isn't
GET_MODE (op0), then we still will return something with unexpected mode
(e.g. DCmode vs. CDImode); I wonder if for such mismatches we shouldn't
just force_reg (convert_modes ()) each CONCAT operand separately and
create a new CONCAT.  Do we have a guarantee that COMPLEX_MODE_P (GET_MODE 
(op0))
if op0 is CONCAT?

2016-10-27  Jakub Jelinek  <ja...@redhat.com>

        PR rtl-optimization/77919
        * expr.c (expand_expr_real_1) <normal_inner_ref>: Force CONCAT into
        MEM if mode1 is not a complex mode.

        * g++.dg/torture/pr77919.C: New test.

--- gcc/expr.c.jj       2016-10-27 20:50:22.699586175 +0200
+++ gcc/expr.c  2016-10-27 21:15:30.146309091 +0200
@@ -10421,7 +10421,8 @@ expand_expr_real_1 (tree exp, rtx target
        if (GET_CODE (op0) == CONCAT && !must_force_mem)
          {
            if (bitpos == 0
-               && bitsize == GET_MODE_BITSIZE (GET_MODE (op0)))
+               && bitsize == GET_MODE_BITSIZE (GET_MODE (op0))
+               && COMPLEX_MODE_P (mode1))
              {
                if (reversep)
                  op0 = flip_storage_order (GET_MODE (op0), op0);
--- gcc/testsuite/g++.dg/torture/pr77919.C.jj   2016-10-27 21:15:19.883440139 
+0200
+++ gcc/testsuite/g++.dg/torture/pr77919.C      2016-10-27 21:16:01.694906242 
+0200
@@ -0,0 +1,11 @@
+// PR rtl-optimization/77919
+// { dg-do compile }
+// { dg-additional-options "-Wno-psabi" }
+
+struct A { A (double) {} _Complex double i; };
+typedef int __attribute__ ((vector_size (16))) B;
+typedef struct { B b; } C;
+struct D { D (const B &x) : b (x) {} B b; };
+static inline B foo (const double *x) { C *a; a = (C *) x; return a->b; }
+static inline D baz (const A &x) { return foo ((double *) &x); }
+D b = baz (0);

        Jakub