Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   YingChi Long <me@inclyc.cn>
To:     bp@alien8.de
Cc:     chang.seok.bae@intel.com, dave.hansen@linux.intel.com,
        david.laight@aculab.com, hpa@zytor.com,
        linux-kernel@vger.kernel.org, me@inclyc.cn, mingo@redhat.com,
        ndesaulniers@google.com, pbonzini@redhat.com, tglx@linutronix.de,
        x86@kernel.org
Subject: Re: [PATCH RESEND v3] x86/fpu: use _Alignof to avoid UB in TYPE_ALIGN
Date:   Thu,  3 Nov 2022 02:07:19 +0800
Message-Id: <20221102180719.2115424-1-me@inclyc.cn>
In-Reply-To: <Y2KhCyofvfG5W5hE@zn.tnic>
References: <Y2KhCyofvfG5W5hE@zn.tnic>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

> What does that paragraph have to do with fixing the kernel?

The clang patch D133574 has been made to satisfy the requirements of WG14
N2350. Compiling the kernel with this patched clang allows me to test where
type definitions are used in the kernel in the first argument of offsetof.

> Is this patch going to be part of clang? If so, which version?

Yes. Probably clang-16 because this patch is not landed to LLVM codebase
currently. The kernel needs this patch to be successfully compiled, in order
not to break the ability of LLVM mainline to compile the kernel, I am happy to
not landing D133574 for now.

> Does gcc need it too?

Since WG14 N2350 is generally applied to the C standard, I feel that GCC should
reject/fire a warning pointing out type definitions within offsetof.

> If so, should a gcc bug be opened too?

I'm not very familiar with the GCC community, but I thought it should be good
to file a bug.

Link: https://reviews.llvm.org/D133574