Received-SPF: pass (google.com: domain of linux-kernel+bounces-155338-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Tue, 23 Apr 2024 07:05:53 -0700
In-Reply-To: <20240423045548.1324969-1-song@kernel.org>
Precedence: bulk
Mime-Version: 1.0
References: <20240423045548.1324969-1-song@kernel.org>
Message-ID: <ZifAQY9yS4U3oEkT@google.com>
Subject: Re: [PATCH] arch/Kconfig: Move SPECULATION_MITIGATIONS to arch/Kconfig
From: Sean Christopherson <seanjc@google.com>
To: Song Liu <song@kernel.org>
Cc: linux-kernel@vger.kernel.org, bpf@vger.kernel.org, netdev@vger.kernel.org, 
	kernel-team@meta.com, stable@vger.kernel.org, Ingo Molnar <mingo@kernel.org>, 
	Daniel Sneddon <daniel.sneddon@linux.intel.com>, Jakub Kicinski <kuba@kernel.org>
Content-Type: text/plain; charset="us-ascii"

On Mon, Apr 22, 2024, Song Liu wrote:
> SPECULATION_MITIGATIONS is currently defined only for x86. As a result,
> IS_ENABLED(CONFIG_SPECULATION_MITIGATIONS) is always false for other
> archs. f337a6a21e2f effectively set "mitigations=off" by default on
> non-x86 archs, which is not desired behavior. Jakub observed this
> change when running bpf selftests on s390 and arm64.
> 
> Fix this by moving SPECULATION_MITIGATIONS to arch/Kconfig so that it is
> available in all archs and thus can be used safely in kernel/cpu.c

Yeah, it's a known issue that we've been slow to fix because we've haven't come
to an agreement on exactly what the Kconfig should look like[1], though there's
general consensus to add CPU_MITIGATIONS in common code[2][3].

I'll poke Josh's thread and make sure a fix gets into rc6.

[1] https://lore.kernel.org/all/20240417001507.2264512-2-seanjc@google.com
[2] https://lore.kernel.org/all/20240420000556.2645001-2-seanjc@google.com
[3] https://lore.kernel.org/all/9d3c997264829d0e2b28718222724ae8f9e7d8b4.1713559768.git.jpoimboe@kernel.org