This was an oversight on my part when I converted extract_component
to use gimple_build_assign instead of force_gimple_operand_gsi.
I had provided a special case for VCE of a SSA_NAME but I missed
that the same issue would be caused with invariants too and invariants
would show up with VCE; I assumed they would be folded.
This changes the check to include invariants too.
Bootstrapped and tested on x86_64-linux-gnu.
PR tree-optimization/124086
gcc/ChangeLog:
* tree-complex.cc (extract_component): Extend the check
for ssa names for VCE to include invariants.
gcc/testsuite/ChangeLog:
* c-c++-common/torture/pr124086-1.c: New test.
* g++.dg/torture/pr124086-1.C: New test.
Signed-off-by: Andrew Pinski <[email protected]>
---
gcc/testsuite/c-c++-common/torture/pr124086-1.c | 15 +++++++++++++++
gcc/testsuite/g++.dg/torture/pr124086-1.C | 14 ++++++++++++++
gcc/tree-complex.cc | 12 ++++++++----
3 files changed, 37 insertions(+), 4 deletions(-)
create mode 100644 gcc/testsuite/c-c++-common/torture/pr124086-1.c
create mode 100644 gcc/testsuite/g++.dg/torture/pr124086-1.C
diff --git a/gcc/testsuite/c-c++-common/torture/pr124086-1.c
b/gcc/testsuite/c-c++-common/torture/pr124086-1.c
new file mode 100644
index 00000000000..6bca0431e4c
--- /dev/null
+++ b/gcc/testsuite/c-c++-common/torture/pr124086-1.c
@@ -0,0 +1,15 @@
+/* { dg-do compile } */
+/* { dg-options "" } */
+/* PR tree-optimization/124086 */
+
+
+typedef __attribute__((__vector_size__(2*sizeof(long double)))) int V;
+int j;
+
+void
+foo()
+{
+ V v = (V){0, -0xd};
+ _Complex long double t = *(_Complex long double *)&v;
+ j = __real__ t;
+}
diff --git a/gcc/testsuite/g++.dg/torture/pr124086-1.C
b/gcc/testsuite/g++.dg/torture/pr124086-1.C
new file mode 100644
index 00000000000..f01bd35ce3c
--- /dev/null
+++ b/gcc/testsuite/g++.dg/torture/pr124086-1.C
@@ -0,0 +1,14 @@
+// { dg-do compile { target lp64 } }
+
+// PR tree-optimization/124086
+
+int j;
+int *v;
+void
+foo()
+{
+ int t;
+ v = &t;
+ auto t1 = __builtin_bit_cast(_Complex int, v);
+ j = __real__ t1;
+}
diff --git a/gcc/tree-complex.cc b/gcc/tree-complex.cc
index 90401925f26..0937cb442bf 100644
--- a/gcc/tree-complex.cc
+++ b/gcc/tree-complex.cc
@@ -664,14 +664,18 @@ extract_component (gimple_stmt_iterator *gsi, tree t,
bool imagpart_p,
}
case VIEW_CONVERT_EXPR:
- /* Getting the real/imag parts of a VCE of a ssa-name requires
- to place the complex into a ssa name before getting the
- 2 parts.
+ /* Getting the real/imag parts of a VCE of a ssa-name
+ (or gimple invariant) requires to place the complex
+ into a ssa name before getting the 2 parts.
As `IMAGPART_EXPR<VIEW_CONVERT_EXPR<a_BN>>` is an invalid
gimple. This will only show up when gimplifying it.
Note this creates an extra copy. The call to
force_gimple_operand_gsi would create one too. */
- if (TREE_CODE (TREE_OPERAND (t, 0)) == SSA_NAME)
+ tree expr;
+ expr = TREE_OPERAND (t, 0);
+ if (TREE_CODE (expr) == SSA_NAME
+ || (is_gimple_min_invariant (expr)
+ && TREE_CODE (expr) != STRING_CST))
{
gcc_assert (gimple_p);
tree new_cplx = make_ssa_name (TREE_TYPE (t));
--
2.43.0
