From: Bernd Schmidt <bernds_cb1@t-online.de>
To: Martin Sebor <msebor@gmail.com>,
Bernd Schmidt <bschmidt@redhat.com>,
Gcc Patch List <gcc-patches@gcc.gnu.org>,
Joseph Myers <joseph@codesourcery.com>
Subject: Re: [PATCH] c/67882 - improve -Warray-bounds for invalid offsetof
Date: Tue, 20 Oct 2015 13:21:00 -0000 [thread overview]
Message-ID: <56263F80.1090203@t-online.de> (raw)
In-Reply-To: <56215158.5040404@gmail.com>
[-- Attachment #1: Type: text/plain, Size: 1342 bytes --]
On 10/16/2015 09:34 PM, Martin Sebor wrote:
> Thank you for the review. Attached is an updated patch that hopefully
> addresses all your comments. I ran the check_GNU_style.sh script on
> it to make sure I didn't miss something. I've also added replies to
> a few of your comments below.
Ok, thanks. However - the logic around the context struct in the patch
still seems fairly impenetrable to me, and I've been wondering whether a
simpler approach wouldn't work just as well. I came up with the patch
below, which just passes a tree_code context to recursive invocations
and increasing the upper bound by one only if not in an ARRAY_REF or
COMPONENT_REF context.
As far as I can tell, this produces the same warnings on the testcase
(modulo differences in type printout), except for the final few FA5_7
testcases, which I had doubts about anyway:
typedef struct FA5_7 {
int i;
char a5_7 [5][7];
} FA5_7;
__builtin_offsetof (FA5_7, a5_7 [0][7]), // { dg-warning
"index" }
__builtin_offsetof (FA5_7, a5_7 [1][7]), // { dg-warning
"index" }
__builtin_offsetof (FA5_7, a5_7 [5][0]), // { dg-warning
"index" }
__builtin_offsetof (FA5_7, a5_7 [5][7]), // { dg-warning
"index" }
Why wouldn't at least the first two be convered by the
one-past-the-end-is-ok rule?
Bernd
[-- Attachment #2: offsetof.diff --]
[-- Type: text/x-patch, Size: 2096 bytes --]
Index: gcc/c-family/c-common.c
===================================================================
--- gcc/c-family/c-common.c (revision 229049)
+++ gcc/c-family/c-common.c (working copy)
@@ -10589,11 +10589,11 @@ c_common_to_target_charset (HOST_WIDE_IN
traditional rendering of offsetof as a macro. Return the folded result. */
tree
-fold_offsetof_1 (tree expr)
+fold_offsetof_1 (tree expr, enum tree_code ctx)
{
tree base, off, t;
-
- switch (TREE_CODE (expr))
+ tree_code code = TREE_CODE (expr);
+ switch (code)
{
case ERROR_MARK:
return expr;
@@ -10617,7 +10617,7 @@ fold_offsetof_1 (tree expr)
return TREE_OPERAND (expr, 0);
case COMPONENT_REF:
- base = fold_offsetof_1 (TREE_OPERAND (expr, 0));
+ base = fold_offsetof_1 (TREE_OPERAND (expr, 0), code);
if (base == error_mark_node)
return base;
@@ -10634,7 +10634,7 @@ fold_offsetof_1 (tree expr)
break;
case ARRAY_REF:
- base = fold_offsetof_1 (TREE_OPERAND (expr, 0));
+ base = fold_offsetof_1 (TREE_OPERAND (expr, 0), code);
if (base == error_mark_node)
return base;
@@ -10649,8 +10649,9 @@ fold_offsetof_1 (tree expr)
&& !tree_int_cst_equal (upbound,
TYPE_MAX_VALUE (TREE_TYPE (upbound))))
{
- upbound = size_binop (PLUS_EXPR, upbound,
- build_int_cst (TREE_TYPE (upbound), 1));
+ if (ctx != ARRAY_REF && ctx != COMPONENT_REF)
+ upbound = size_binop (PLUS_EXPR, upbound,
+ build_int_cst (TREE_TYPE (upbound), 1));
if (tree_int_cst_lt (upbound, t))
{
tree v;
Index: gcc/c-family/c-common.h
===================================================================
--- gcc/c-family/c-common.h (revision 229049)
+++ gcc/c-family/c-common.h (working copy)
@@ -1029,7 +1029,7 @@ extern bool c_dump_tree (void *, tree);
extern void verify_sequence_points (tree);
-extern tree fold_offsetof_1 (tree);
+extern tree fold_offsetof_1 (tree, tree_code ctx = ERROR_MARK);
extern tree fold_offsetof (tree);
/* Places where an lvalue, or modifiable lvalue, may be required.
next prev parent reply other threads:[~2015-10-20 13:20 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-10-09 2:55 Martin Sebor
2015-10-15 21:59 ` [PING] " Martin Sebor
2015-10-16 12:28 ` Bernd Schmidt
2015-10-16 17:27 ` Joseph Myers
2015-10-16 19:34 ` Martin Sebor
2015-10-20 13:21 ` Bernd Schmidt [this message]
2015-10-20 15:33 ` Martin Sebor
2015-10-20 15:52 ` Bernd Schmidt
2015-10-20 16:57 ` Martin Sebor
2015-10-20 17:11 ` Joseph Myers
2015-10-20 19:10 ` Martin Sebor
2015-10-20 16:54 ` Joseph Myers
2015-10-20 20:36 ` Bernd Schmidt
2015-10-20 22:19 ` Joseph Myers
2015-10-23 11:17 ` Bernd Schmidt
2015-10-23 15:15 ` Martin Sebor
2015-10-23 16:53 ` Joseph Myers
2015-10-23 17:45 ` Bernd Schmidt
2015-10-23 20:54 ` Martin Sebor
2015-10-26 11:44 ` Bernd Schmidt
2015-10-26 11:51 ` Jakub Jelinek
2015-10-26 12:01 ` Bernd Schmidt
2015-10-26 12:04 ` Jakub Jelinek
2015-10-26 12:32 ` Richard Biener
2015-10-27 11:18 ` Bernd Schmidt
2015-11-03 19:15 ` Martin Sebor
2015-11-07 23:38 ` Segher Boessenkool
2015-11-09 22:46 ` Martin Sebor
2015-11-10 0:02 ` Joseph Myers
-- strict thread matches above, loose matches on Subject: below --
2015-10-09 2:49 Martin Sebor
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=56263F80.1090203@t-online.de \
--to=bernds_cb1@t-online.de \
--cc=bschmidt@redhat.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=joseph@codesourcery.com \
--cc=msebor@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).