On 20 Nov 10:59, Richard Biener wrote:
> On Tue, Nov 19, 2013 at 9:29 PM, Ilya Enkovich <enkovich....@gmail.com> wrote:
> > On 19 Nov 12:33, Jeff Law wrote:
> >> On 11/19/13 05:13, Ilya Enkovich wrote:
> >> >On 19 Nov 13:00, Richard Biener wrote:
> >> >>I'd say not in the gimplifier either but in varpool (symbol table) code
> >> >>where the symbols are ultimatively registered with?
> >> >
> >> >Something like that?
> >> >
> >> >--- a/gcc/varpool.c
> >> >+++ b/gcc/varpool.c
> >> >@@ -151,6 +151,10 @@ varpool_node_for_decl (tree decl)
> >> > node = varpool_create_empty_node ();
> >> > node->decl = decl;
> >> > symtab_register_node (node);
> >> >+
> >> >+ if (DECL_NIITIAL (decl))
> >> >+ chkp_register_var_initializer (decl);
> >> >+
> >> > return node;
> >> > }
> >> Yea, I think that's what Richi is suggesting.
> >> jeff
> >>
> >
> > Great! Here is a full version of the patch. Bootstrap, make check and MPX
> > tests are OK with the change.
>
> Please don't do this in varpool_node_for_decl but in
> varpool_finalize_decl.
>
> Richard.
>
Here it is.
Thanks,
Ilya
--
gcc/
2013-11-20 Ilya Enkovich <ilya.enkov...@intel.com>
* cgraphunit.c: Include tree-chkp.h.
(varpool_finalize_decl): Register statically
initialized decls in Pointer Bounds Checker.
diff --git a/gcc/cgraphunit.c b/gcc/cgraphunit.c
index 8ab274b..6b6ec55 100644
--- a/gcc/cgraphunit.c
+++ b/gcc/cgraphunit.c
@@ -201,6 +201,7 @@ along with GCC; see the file COPYING3. If not see
#include "regset.h" /* FIXME: For reg_obstack. */
#include "context.h"
#include "pass_manager.h"
+#include "tree-chkp.h"
/* Queue of cgraph nodes scheduled to be added into cgraph. This is a
secondary queue used during optimization to accommodate passes that
@@ -830,6 +831,9 @@ varpool_finalize_decl (tree decl)
finished. */
if (cgraph_state == CGRAPH_STATE_FINISHED)
varpool_assemble_decl (node);
+
+ if (DECL_INITIAL (decl))
+ chkp_register_var_initializer (decl);
}
/* EDGE is an polymorphic call. Mark all possible targets as reachable
