Subject: Re: [PATCH] init/Kconfig: extend -Wno-array-bounds to gcc 13
To:     Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Cc:     Kees Cook <kees@kernel.org>,
        Nick Desaulniers <ndesaulniers@google.com>,
        Kees Cook <keescook@chromium.org>, mhiramat@kernel.org,
        akpm@linux-foundation.org, masahiroy@kernel.org,
        paulmck@kernel.org, hannes@cmpxchg.org, ojeda@kernel.org,
        thunder.leizhen@huawei.com, christophe.leroy@csgroup.eu,
        vbabka@suse.cz, linux-kernel@vger.kernel.org
References: <20230306220947.1982272-1-trix@redhat.com>
 <CAKwvOdkd7pUu_oxt4=CW3fxXz2y7evtU9J5Gd7oQuyi1MkDMCw@mail.gmail.com>
 <C01C39FE-E776-48DA-84F2-402B49A705A8@kernel.org>
 <899b639b-3c93-b96d-863d-9e25d1824dc3@redhat.com>
 <CANiq72kKmVzQYU7aOy4k0GsrgV=jfi_fLk2TvtWVGV_sXc8e2w@mail.gmail.com>
From:   Tom Rix <trix@redhat.com>
Message-ID: <e7eef530-16b8-0a5d-273d-98aa3f673e61@redhat.com>
Date:   Tue, 7 Mar 2023 05:28:53 -0800
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.10.1
MIME-Version: 1.0
In-Reply-To: <CANiq72kKmVzQYU7aOy4k0GsrgV=jfi_fLk2TvtWVGV_sXc8e2w@mail.gmail.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Content-Language: en-US
Precedence: bulk


On 3/7/23 3:42 AM, Miguel Ojeda wrote:
> On Tue, Mar 7, 2023 at 2:07 AM Tom Rix <trix@redhat.com> wrote:
>> The build with fail generally with gcc 13.
>>
>> The warnings could be cleaned without having an error, but I looked at
>> multiple errors, none of them were real.
>>
>> imo this is a broken compiler option.
> I am not sure I understand -- my reading of Kees' message is that he
> would prefer to get the warning (rather than the kernel) fixed before
> GCC 13 releases.

yes, that would be ideal.

But anyone using gcc 13 in the meanwhile will have a broken build

and any other aspect of testing the kernel with gcc 13 would have

to be deferred to after the fix, if it happens.

>
> Are you asking to have the option disabled until GCC 13 releases and
> reevaluate then? How many warnings are you getting? Are those actual
> errors or `-Werror`?

With W=1, there are 40 error:'s, sorry I do not have the default logs at 
hand.

I looked about 10 of them over the weekend, they we all bogus.

So gcc needs to be fixed first, just disabling with -Wnoerror will

give folks bad problems that do not need fixing.


I am asking that we at least temporarily disable this option so

the build break is fixed.

Tom

>
> Cheers,
> Miguel
>