https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70509
--- Comment #4 from Zdenek Sojka <zsojka at seznam dot cz> ---
(In reply to Marc Glisse from comment #3)
> Rereading this a few years later, I have no idea why I used that type.
> bitsize_int would make more sense...
I've already tried this:
--- gcc/tree-ssa-forwprop.c (revision 234698)
+++ gcc/tree-ssa-forwprop.c (working copy)
@@ -1773,7 +1773,7 @@
if (code == VEC_PERM_EXPR)
{
- tree p, m, index, tem;
+ tree p, m, tem;
unsigned nelts;
m = gimple_assign_rhs3 (def_stmt);
if (TREE_CODE (m) != VECTOR_CST)
@@ -1790,9 +1790,8 @@
p = gimple_assign_rhs2 (def_stmt);
idx -= nelts;
}
- index = build_int_cst (TREE_TYPE (TREE_TYPE (m)), idx * size);
tem = build3 (BIT_FIELD_REF, TREE_TYPE (op),
- unshare_expr (p), op1, index);
+ unshare_expr (p), op1, bitsize_int(idx * size));
gimple_assign_set_rhs1 (stmt, tem);
fold_stmt (gsi);
update_stmt (gsi_stmt (*gsi));
x86_64 bootstrap finished (the code is likely not triggered), regtest running.
