Re: [tree-ssa] Removing redundant loads

[law at redhat dot com]

In message <wvln0h40yjy.fsf@prospero.boston.redhat.com>, Jason Merrill writes:
 >> For example, if you pass it this tree for both operands, you'll get a
 >> false back from operand_b because this expression has tree_side_effects
 >> set.
 >
 >Why does that expression have TREE_SIDE_EFFECTS set?
Because we unconditionally set TREE_SIDE_EFFECTS on STMT_EXPRs.  That
gets propagated up to the COMPONENT_REFs.

The testcase for the code in question looks something like:

struct tree_decl
{
  unsigned in_system_header_flag:1;
};
union tree_node
{
  struct tree_decl decl;
};
typedef union tree_node *tree;
static int
redeclaration_error_message (olddecl)
     tree olddecl;
{
  if (({olddecl;})->decl.in_system_header_flag)
    ;
}