public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "vogel at pi2 dot physik dot uni-erlangen dot de" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug middle-end/32856] Invalid optimization in the face of aliasing
Date: Mon, 23 Jul 2007 10:09:00 -0000	[thread overview]
Message-ID: <20070723100851.32215.qmail@sourceware.org> (raw)
In-Reply-To: <bug-32856-3511@http.gcc.gnu.org/bugzilla/>



------- Comment #6 from vogel at pi2 dot physik dot uni-erlangen dot de  2007-07-23 10:08 -------
This program demonstrates the problem, it creates different output depending on
if compiled with or without optimisation.

Without optimisation, n->next is not cached:
n->next = 0xbfb01af0
n->next = 0xbfb01af8

With optimisation, n->next is cached, this is wrong:
n->next = 0xbfdb3da0
n->next = 0xbfdb3da0

Note that the pointer c will point exactly one pointer-width above the
structure a, so n->next->next->prev=n -- which corresponds to c->prev=n -- will
overwrite n->next with n.

#include <stdio.h>

struct node {
        struct node *next, *prev;
};

void foo(struct node* n) {
  printf("n->next = %p\n",n->next);
  n->next->next->prev=n;
  printf("n->next = %p\n",n->next);
};

int main(int argc,char **argv){
        struct node a = { },b = { };
        struct node *c = NULL;

        c = ((void*)&(a.next)) - sizeof(void*);
        b.next = c;
        a.next = &b;

        foo(&a);
}


-- 

vogel at pi2 dot physik dot uni-erlangen dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |vogel at pi2 dot physik dot
                   |                            |uni-erlangen dot de


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32856


  parent reply	other threads:[~2007-07-23 10:09 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-07-22 17:06 [Bug c/32856] New: " felix-gcc at fefe dot de
2007-07-22 17:09 ` [Bug c/32856] " felix-gcc at fefe dot de
2007-07-22 17:12 ` felix-gcc at fefe dot de
2007-07-22 19:27 ` [Bug middle-end/32856] " pinskia at gcc dot gnu dot org
2007-07-22 23:01 ` falk at debian dot org
2007-07-22 23:08 ` felix-gcc at fefe dot de
2007-07-22 23:19 ` falk at debian dot org
2007-07-23 10:09 ` vogel at pi2 dot physik dot uni-erlangen dot de [this message]
2007-07-23 14:17 ` falk at debian dot org
2007-07-24 12:09 ` rguenth at gcc dot gnu dot org
2007-07-25  8:24 ` falk at debian dot org

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=20070723100851.32215.qmail@sourceware.org \
    --to=gcc-bugzilla@gcc.gnu.org \
    --cc=gcc-bugs@gcc.gnu.org \
    /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).