Re: Linux 2.6.27.27
[Posted July 22, 2009 by corbet]
| From: |
| Ian Lance Taylor <iant-AT-google.com> |
| To: |
| Krzysztof Oledzki <olel-AT-ans.pl> |
| Subject: |
| Re: Linux 2.6.27.27 |
| Date: |
| Wed, 22 Jul 2009 08:36:21 -0700 |
| Message-ID: |
| <m3k520buh6.fsf@google.com> |
| Cc: |
| Linus Torvalds <torvalds-AT-linux-foundation.org>,
Troy Moure <twmoure-AT-szypr.net>, Greg KH <gregkh-AT-suse.de>,
Linux Kernel Mailing List <linux-kernel-AT-vger.kernel.org>,
Andrew Morton <akpm-AT-linux-foundation.org>, stable-AT-kernel.org,
lwn-AT-lwn.net |
| Archive-link: |
| Article, Thread
|
Krzysztof Oledzki <olel@ans.pl> writes:
> BTW: here is a simple testcase for this bug:
>
> --- fno-strict-overflow-fixed-bug.c ---
> #include <stdio.h>
>
> int main() {
>
> unsigned char i;
>
> for (i = 0; i < 128; i++)
> printf("loop %u\n", i);
>
> return 0;
> }
> --- cut here ---
I can confirm that this is broken in gcc 4.2 using -fno-strict-overflow,
and works correctly in gcc 4.3 and later. Sigh.
Ian
(
Log in to post comments)