On Fri, Apr 04, 2014 at 02:53:32PM -0600, Jeff Law wrote:
> Presumably when we wrote out the PCH -fsanitize wasn't being used
> and thus those builtins are not initialized. Right?
Right. I put a local hack into c_common_write_pch and that said:
wrote x86_64-unknown-linux-gnu/bits/stdc++.h.gch/O2g.gch, sanitize flag: 0
so builtins weren't initialized (see #define DEF_SANITIZER_BUILTIN for
why). O2g.gch is the file that is then read by c_common_read_pch.
> Assuming that's correct, your patch is fine with a comment to that effect.
Thanks. I'll wait a day or two for possible comments.
2014-04-05 Marek Polacek <pola...@redhat.com>
PR sanitizer/60745
* c-ubsan.c: Include asan.h.
(ubsan_instrument_return): Call initialize_sanitizer_builtins.
diff --git gcc/c-family/c-ubsan.c gcc/c-family/c-ubsan.c
index dc4d981..e89ebc1 100644
--- gcc/c-family/c-ubsan.c
+++ gcc/c-family/c-ubsan.c
@@ -29,6 +29,7 @@ along with GCC; see the file COPYING3. If not see
#include "ubsan.h"
#include "c-family/c-common.h"
#include "c-family/c-ubsan.h"
+#include "asan.h"
/* Instrument division by zero and INT_MIN / -1. If not instrumenting,
return NULL_TREE. */
@@ -185,6 +186,10 @@ ubsan_instrument_vla (location_t loc, tree size)
tree
ubsan_instrument_return (location_t loc)
{
+ /* It is possible that PCH zapped table with definitions of sanitizer
+ builtins. Reinitialize them if needed. */
+ initialize_sanitizer_builtins ();
+
tree data = ubsan_create_data ("__ubsan_missing_return_data", &loc,
NULL, NULL_TREE);
tree t = builtin_decl_explicit (BUILT_IN_UBSAN_HANDLE_MISSING_RETURN);
Marek
