* [PATCH] Fix -fsanitize=bool -fnon-call-exceptions (PR sanitizer/63913)
@ 2014-11-18 21:57 Jakub Jelinek
2014-11-19 9:13 ` Richard Biener
0 siblings, 1 reply; 2+ messages in thread
From: Jakub Jelinek @ 2014-11-18 21:57 UTC (permalink / raw)
To: Richard Biener; +Cc: gcc-patches
Hi!
This patch fixes instrumentation of bool/enum loads if they could
throw. We want to keep the EH on the load, and push
further statements on the fallthru edge.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2014-11-18 Jakub Jelinek <jakub@redhat.com>
PR sanitizer/63913
* ubsan.c: Include tree-eh.h.
(instrument_bool_enum_load): Handle loads that can throw.
* g++.dg/ubsan/pr63913.C: New test.
--- gcc/ubsan.c.jj 2014-11-14 15:11:49.000000000 +0100
+++ gcc/ubsan.c 2014-11-18 09:29:20.409209556 +0100
@@ -67,6 +67,7 @@ along with GCC; see the file COPYING3.
#include "dfp.h"
#include "builtins.h"
#include "tree-object-size.h"
+#include "tree-eh.h"
/* Map from a tree to a VAR_DECL tree. */
@@ -1135,7 +1136,9 @@ instrument_bool_enum_load (gimple_stmt_i
|| TREE_CODE (gimple_assign_lhs (stmt)) != SSA_NAME)
return;
+ bool can_throw = stmt_could_throw_p (stmt);
location_t loc = gimple_location (stmt);
+ tree lhs = gimple_assign_lhs (stmt);
tree ptype = build_pointer_type (TREE_TYPE (rhs));
tree atype = reference_alias_ptr_type (rhs);
gimple g = gimple_build_assign (make_ssa_name (ptype, NULL),
@@ -1145,9 +1148,24 @@ instrument_bool_enum_load (gimple_stmt_i
tree mem = build2 (MEM_REF, utype, gimple_assign_lhs (g),
build_int_cst (atype, 0));
tree urhs = make_ssa_name (utype, NULL);
- g = gimple_build_assign (urhs, mem);
- gimple_set_location (g, loc);
- gsi_insert_before (gsi, g, GSI_SAME_STMT);
+ if (can_throw)
+ {
+ gimple_assign_set_lhs (stmt, urhs);
+ g = gimple_build_assign_with_ops (NOP_EXPR, lhs, urhs, NULL_TREE);
+ gimple_set_location (g, loc);
+ edge e = find_fallthru_edge (gimple_bb (stmt)->succs);
+ gsi_insert_on_edge_immediate (e, g);
+ gimple_assign_set_rhs_with_ops (gsi, MEM_REF, mem, NULL_TREE);
+ update_stmt (stmt);
+ *gsi = gsi_for_stmt (g);
+ g = stmt;
+ }
+ else
+ {
+ g = gimple_build_assign (urhs, mem);
+ gimple_set_location (g, loc);
+ gsi_insert_before (gsi, g, GSI_SAME_STMT);
+ }
minv = fold_convert (utype, minv);
maxv = fold_convert (utype, maxv);
if (!integer_zerop (minv))
@@ -1169,8 +1187,11 @@ instrument_bool_enum_load (gimple_stmt_i
gimple_set_location (g, loc);
gsi_insert_after (gsi, g, GSI_NEW_STMT);
- gimple_assign_set_rhs_with_ops (&gsi2, NOP_EXPR, urhs, NULL_TREE);
- update_stmt (stmt);
+ if (!can_throw)
+ {
+ gimple_assign_set_rhs_with_ops (&gsi2, NOP_EXPR, urhs, NULL_TREE);
+ update_stmt (stmt);
+ }
gsi2 = gsi_after_labels (then_bb);
if (flag_sanitize_undefined_trap_on_error)
--- gcc/testsuite/g++.dg/ubsan/pr63913.C.jj 2014-11-18 09:34:10.603981487 +0100
+++ gcc/testsuite/g++.dg/ubsan/pr63913.C 2014-11-18 09:33:47.000000000 +0100
@@ -0,0 +1,12 @@
+// PR sanitizer/63913
+// { dg-do compile }
+// { dg-options "-fsanitize=bool -fnon-call-exceptions" }
+
+struct B { B (); ~B (); };
+
+double
+foo (bool *x)
+{
+ B b;
+ return *x;
+}
Jakub
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] Fix -fsanitize=bool -fnon-call-exceptions (PR sanitizer/63913)
2014-11-18 21:57 [PATCH] Fix -fsanitize=bool -fnon-call-exceptions (PR sanitizer/63913) Jakub Jelinek
@ 2014-11-19 9:13 ` Richard Biener
0 siblings, 0 replies; 2+ messages in thread
From: Richard Biener @ 2014-11-19 9:13 UTC (permalink / raw)
To: Jakub Jelinek; +Cc: gcc-patches
On Tue, 18 Nov 2014, Jakub Jelinek wrote:
> Hi!
>
> This patch fixes instrumentation of bool/enum loads if they could
> throw. We want to keep the EH on the load, and push
> further statements on the fallthru edge.
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
>
> 2014-11-18 Jakub Jelinek <jakub@redhat.com>
>
> PR sanitizer/63913
> * ubsan.c: Include tree-eh.h.
> (instrument_bool_enum_load): Handle loads that can throw.
>
> * g++.dg/ubsan/pr63913.C: New test.
>
> --- gcc/ubsan.c.jj 2014-11-14 15:11:49.000000000 +0100
> +++ gcc/ubsan.c 2014-11-18 09:29:20.409209556 +0100
> @@ -67,6 +67,7 @@ along with GCC; see the file COPYING3.
> #include "dfp.h"
> #include "builtins.h"
> #include "tree-object-size.h"
> +#include "tree-eh.h"
>
> /* Map from a tree to a VAR_DECL tree. */
>
> @@ -1135,7 +1136,9 @@ instrument_bool_enum_load (gimple_stmt_i
> || TREE_CODE (gimple_assign_lhs (stmt)) != SSA_NAME)
> return;
>
> + bool can_throw = stmt_could_throw_p (stmt);
> location_t loc = gimple_location (stmt);
> + tree lhs = gimple_assign_lhs (stmt);
> tree ptype = build_pointer_type (TREE_TYPE (rhs));
> tree atype = reference_alias_ptr_type (rhs);
> gimple g = gimple_build_assign (make_ssa_name (ptype, NULL),
> @@ -1145,9 +1148,24 @@ instrument_bool_enum_load (gimple_stmt_i
> tree mem = build2 (MEM_REF, utype, gimple_assign_lhs (g),
> build_int_cst (atype, 0));
> tree urhs = make_ssa_name (utype, NULL);
> - g = gimple_build_assign (urhs, mem);
> - gimple_set_location (g, loc);
> - gsi_insert_before (gsi, g, GSI_SAME_STMT);
> + if (can_throw)
> + {
> + gimple_assign_set_lhs (stmt, urhs);
> + g = gimple_build_assign_with_ops (NOP_EXPR, lhs, urhs, NULL_TREE);
> + gimple_set_location (g, loc);
> + edge e = find_fallthru_edge (gimple_bb (stmt)->succs);
> + gsi_insert_on_edge_immediate (e, g);
> + gimple_assign_set_rhs_with_ops (gsi, MEM_REF, mem, NULL_TREE);
Please use gimple_assign_set_rhs_from_tree for single-rhs trees.
Ok with that change.
Thanks,
RIchard.
> + update_stmt (stmt);
> + *gsi = gsi_for_stmt (g);
> + g = stmt;
> + }
> + else
> + {
> + g = gimple_build_assign (urhs, mem);
> + gimple_set_location (g, loc);
> + gsi_insert_before (gsi, g, GSI_SAME_STMT);
> + }
> minv = fold_convert (utype, minv);
> maxv = fold_convert (utype, maxv);
> if (!integer_zerop (minv))
> @@ -1169,8 +1187,11 @@ instrument_bool_enum_load (gimple_stmt_i
> gimple_set_location (g, loc);
> gsi_insert_after (gsi, g, GSI_NEW_STMT);
>
> - gimple_assign_set_rhs_with_ops (&gsi2, NOP_EXPR, urhs, NULL_TREE);
> - update_stmt (stmt);
> + if (!can_throw)
> + {
> + gimple_assign_set_rhs_with_ops (&gsi2, NOP_EXPR, urhs, NULL_TREE);
> + update_stmt (stmt);
> + }
>
> gsi2 = gsi_after_labels (then_bb);
> if (flag_sanitize_undefined_trap_on_error)
> --- gcc/testsuite/g++.dg/ubsan/pr63913.C.jj 2014-11-18 09:34:10.603981487 +0100
> +++ gcc/testsuite/g++.dg/ubsan/pr63913.C 2014-11-18 09:33:47.000000000 +0100
> @@ -0,0 +1,12 @@
> +// PR sanitizer/63913
> +// { dg-do compile }
> +// { dg-options "-fsanitize=bool -fnon-call-exceptions" }
> +
> +struct B { B (); ~B (); };
> +
> +double
> +foo (bool *x)
> +{
> + B b;
> + return *x;
> +}
>
> Jakub
>
>
--
Richard Biener <rguenther@suse.de>
SUSE LINUX GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendoerffer, HRB 21284
(AG Nuernberg)
Maxfeldstrasse 5, 90409 Nuernberg, Germany
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2014-11-19 8:57 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-11-18 21:57 [PATCH] Fix -fsanitize=bool -fnon-call-exceptions (PR sanitizer/63913) Jakub Jelinek
2014-11-19 9:13 ` Richard Biener
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).