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;
Date:   Mon, 7 Aug 2023 17:06:17 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Florian Weimer <fweimer@redhat.com>
Cc:     Marco Elver <elver@google.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Kees Cook <keescook@chromium.org>,
        Guenter Roeck <linux@roeck-us.net>,
        Mark Rutland <mark.rutland@arm.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        Marc Zyngier <maz@kernel.org>,
        Oliver Upton <oliver.upton@linux.dev>,
        James Morse <james.morse@arm.com>,
        Suzuki K Poulose <suzuki.poulose@arm.com>,
        Zenghui Yu <yuzenghui@huawei.com>,
        Catalin Marinas <catalin.marinas@arm.com>,
        Will Deacon <will@kernel.org>,
        Nathan Chancellor <nathan@kernel.org>,
        Nick Desaulniers <ndesaulniers@google.com>,
        Tom Rix <trix@redhat.com>, Miguel Ojeda <ojeda@kernel.org>,
        linux-arm-kernel@lists.infradead.org, kvmarm@lists.linux.dev,
        linux-kernel@vger.kernel.org, llvm@lists.linux.dev,
        Dmitry Vyukov <dvyukov@google.com>,
        Alexander Potapenko <glider@google.com>,
        kasan-dev@googlegroups.com, linux-toolchains@vger.kernel.org
Subject: Re: [PATCH v2 1/3] compiler_types: Introduce the Clang
 __preserve_most function attribute
Message-ID: <20230807150617.GB569857@hirez.programming.kicks-ass.net>
References: <20230804090621.400-1-elver@google.com>
 <87il9rgjvw.fsf@oldenburg.str.redhat.com>
 <CANpmjNN4h2+i3LUG__GHha849PZ3jK=mBoFQWpSz4jffXB4wrw@mail.gmail.com>
 <87pm3zf2qi.fsf@oldenburg.str.redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <87pm3zf2qi.fsf@oldenburg.str.redhat.com>
Precedence: bulk

On Mon, Aug 07, 2023 at 02:36:53PM +0200, Florian Weimer wrote:

> I think the GCC vs Clang thing is expected to work today, isn't it?
> Using the Clang-based BPF tools with a GCC-compiled kernel requires a
> matching ABI.

Nope, all bets are off. There is no module ABI, in the widest possible
sense.

There's all sorts of subtle breakage, I don't remember the exact
details, but IIRC building the kernel with a compiler that has
asm-goto-output and modules with a compiler that doesn't have it gets
you fireworks.

We absolutely do even bother tracking any of this.

There's also things like GCC plugins, they can randomly (literally in
the case of struct randomization) change things around that isn't
compatible with what the other compiler does -- or perhaps even a later
version of the same compiler.