Hi all,
svset_impl::fold lowers svset intrinsics into a copy of the entire tuple
and then a subsequent update of the given tuple member to the new vector
value, e.g.:
D.14149 = svset2_f32 (a, 0, v);
gets lowered into:
D.14149 = a;
D.14149.__val[0] = v;
. Because both stmts use the lhs of the original call stmt, the code
unshares the expressions. As svset_impl::fold notes:
/* [...]
The fold routines expect the replacement statement to have the
same lhs as the original call, so return the copy statement
rather than the field update. */
ultimately gsi_replace asserts that:
gimple_get_lhs (orig_stmt) == gimple_get_lhs (stmt)
i.e. requiring pointer equality, but the copy stmt we build is:
gassign *copy = gimple_build_assign (unshare_expr (f.lhs), rhs_tuple);
and notably the unshare_expr means that the pointer equality test above
fails. For VAR_DECLs (as above) it happens to work, since those aren't
unshared. mostly_copy_tree_r (called from unshare_expr) has:
/* Stop at types, decls, constants like copy_tree_r. */
else if (TREE_CODE_CLASS (code) == tcc_type
|| TREE_CODE_CLASS (code) == tcc_declaration
|| TREE_CODE_CLASS (code) == tcc_constant)
*walk_subtrees = 0;
but for MEM_EXPRs (as with the tescase in the PR) we *do* actually
unshare them, and thus trip the assert in gsi_replace (or since
r17-1237-g4fb2541debcca1, ICE in completely_unused, but the root cause
is the same).
We want the stmt returned from svset_impl::fold (copy) to have a
pointer-identical lhs as the original stmt, so this patch just changes
which use of f.lhs is unshared. We keep copy's use as per the original
stmt, and unshare the use in the update stmt instead.
Bootstrapped/regtested on aarch64-linux-gnu, OK for trunk and backports
to all active branches after some time to bake?
Thanks,
Alex
gcc/ChangeLog:
PR target/125818
* config/aarch64/aarch64-sve-builtins-base.cc (svset_impl::fold): Fix
unsharing to ensure the returned stmt has the exact same lhs as the
original.
gcc/testsuite/ChangeLog:
PR target/125818
* gcc.target/aarch64/torture/pr125818.c: New test.
---
gcc/config/aarch64/aarch64-sve-builtins-base.cc | 4 ++--
gcc/testsuite/gcc.target/aarch64/torture/pr125818.c | 6 ++++++
2 files changed, 8 insertions(+), 2 deletions(-)
create mode 100644 gcc/testsuite/gcc.target/aarch64/torture/pr125818.c
diff --git a/gcc/config/aarch64/aarch64-sve-builtins-base.cc b/gcc/config/aarch64/aarch64-sve-builtins-base.cc
index 1fa7473283d..963f0adfda1 100644
--- a/gcc/config/aarch64/aarch64-sve-builtins-base.cc
+++ b/gcc/config/aarch64/aarch64-sve-builtins-base.cc
@@ -2992,12 +2992,12 @@ public:
The fold routines expect the replacement statement to have the
same lhs as the original call, so return the copy statement
rather than the field update. */
- gassign *copy = gimple_build_assign (unshare_expr (f.lhs), rhs_tuple);
+ gassign *copy = gimple_build_assign (f.lhs, rhs_tuple);
/* Get a reference to the individual vector. */
tree field = tuple_type_field (TREE_TYPE (f.lhs));
tree lhs_array = build3 (COMPONENT_REF, TREE_TYPE (field),
- f.lhs, field, NULL_TREE);
+ unshare_expr (f.lhs), field, NULL_TREE);
tree lhs_vector = build4 (ARRAY_REF, TREE_TYPE (rhs_vector),
lhs_array, index, NULL_TREE, NULL_TREE);
gassign *update = gimple_build_assign (lhs_vector, rhs_vector);
diff --git a/gcc/testsuite/gcc.target/aarch64/torture/pr125818.c b/gcc/testsuite/gcc.target/aarch64/torture/pr125818.c
new file mode 100644
index 00000000000..875092000df
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/torture/pr125818.c
@@ -0,0 +1,6 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.2-a+sve" } */
+#include <arm_sve.h>
+void f(svfloat32x2_t *tp, const svfloat32x2_t *tuple, const svfloat32_t *vp) {
+ *tp = svset2_f32(*tuple, 0, *vp);
+}
