On Fri, Jan 14, 2022 at 09:41:35AM +0100, Jan Hubicka via Gcc-patches wrote:
> > > > > > --- a/gcc/ipa-split.c
> > > > > > +++ b/gcc/ipa-split.c
> > > > > > @@ -873,7 +873,7 @@ visit_bb (basic_block bb, basic_block return_bb,
> > > > > > gimple *stmt = gsi_stmt (bsi);
> > > > > > tree op;
> > > > > > ssa_op_iter iter;
> > > > > > - tree decl;
> > > > > > + tree decl = NULL_TREE;
> > > > > >
> > > > > > if (is_gimple_debug (stmt))
> > > > > > continue;
> > > > > > @@ -927,6 +927,16 @@ visit_bb (basic_block bb, basic_block
> > > > > > return_bb,
> Decl is initialized in
> if (gimple_code (stmt) == GIMPLE_CALL
> && (decl = gimple_call_fndecl (stmt)) != NULL_TREE
> && fndecl_built_in_p (decl, BUILT_IN_NORMAL))
>
> I think this is confusing. I would change it to
> if (gimple_code (stmt) == GIMPLE_CALL
> && (decl = gimple_call_fndecl (stmt)) != NULL_TREE
> {
> if (fndecl_built_in_p (decl, BUILT_IN_NORMAL))
> ... existing code ...
> if (decl && (lookup_attribute ("warning", DECL_ATTRIBUTES (decl))
> || lookup_attribute ("error", DECL_ATTRIBUTES (decl))))
> ... your code ...
> }
> OK with that change.
Perhaps even
/* Check builtins that prevent splitting. */
if (gimple_code (stmt) == GIMPLE_CALL)
if (tree decl = gimple_call_fndecl (stmt))
{
if (fndecl_built_in_p (decl, BUILT_IN_NORMAL))
...
if (lookup_attribute || lookup_attribute)
...
}
?
Jakub
