Subject: Re: [PATCH 01/12] extarray: define helpers for arrays defined in
 linker scripts
To: Peter Zijlstra <peterz@infradead.org>,
        Vegard Nossum <vegard.nossum@oracle.com>
References: <20161016151616.31451-1-vegard.nossum@oracle.com>
 <20161016151616.31451-2-vegard.nossum@oracle.com>
 <20161017083315.GA29322@worktop.vlan200.pylonone.local>
Cc: linux-kernel@vger.kernel.org,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        "Luis R . Rodriguez" <mcgrof@kernel.org>, stable@vger.kernel.org,
        Ming Lei <ming.lei@canonical.com>,
        Steven Rostedt <srostedt@redhat.com>
From: Jiri Slaby <jslaby@suse.cz>
Message-ID: <186f8242-3f8d-31cd-a8e8-9743bbc1c1fd@suse.cz>
Date: Mon, 17 Oct 2016 11:01:13 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101
 Thunderbird/45.3.0
MIME-Version: 1.0
In-Reply-To: <20161017083315.GA29322@worktop.vlan200.pylonone.local>
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1060
Lines: 29

On 10/17/2016, 10:33 AM, Peter Zijlstra wrote:
> On Sun, Oct 16, 2016 at 05:16:05PM +0200, Vegard Nossum wrote:
>> The test in this loop:
>>
>> 	for (b_fw = __start_builtin_fw; b_fw != __end_builtin_fw; b_fw++) {
>>
>> was getting completely compiled out by my gcc, 7.0.0 20160520. The result
>> was that the loop was going beyond the end of the builtin_fw array and
>> giving me a page fault when trying to dereference b_fw->name.
>>
>> This is because __start_builtin_fw and __end_builtin_fw are both declared
>> as (separate) arrays, and so gcc conludes that b_fw can never point to
>> __end_builtin_fw.
>>
> 
> Urgh, isn't that the kind of 'optimizations' we should shoot in the head
> for the kernel? Just like the -fno-strict-aliassing crap?

Unfortunately, there is no such switch for this optimization.

On the top of that, it's incorrect C according to the standard. So it is
as correct as expecting 0 printed here: 'int zero; printf("%d\n",
zero);'. It never worked, not even with gcc 4 and maybe older. We were
just lucky.

thanks,
-- 
js
suse labs