Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   "H. Peter Anvin (Intel)" <hpa@zytor.com>
To:     Borislav Petkov <bp@alien8.de>, Ingo Molnar <mingo@redhat.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Andy Lutomirski <luto@kernel.org>
Cc:     x86 mailing list <x86@kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        "H. Peter Anvin (Intel)" <hpa@zytor.com>
Subject: [PATCH v3 0/2] x86/asm: avoid register pressure from the init case in static_cpu_has()
Date:   Fri, 10 Sep 2021 12:59:07 -0700
Message-Id: <20210910195910.2542662-1-hpa@zytor.com>
In-Reply-To: <20210908171716.3340120-1-hpa@zytor.com>
References: <20210908171716.3340120-1-hpa@zytor.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

gcc will sometimes manifest the address of boot_cpu_data in a register
as part of constant propagation. When multiple static_cpu_has() are
used this may foul the mainline code with a register load which will
only be used on the fallback path, which is unused after
initialization.

Explicitly force gcc to use immediate (rip-relative) addressing for
the fallback path, thus removing any possible register use from
static_cpu_has().

However, currently there is no convenient way to make gcc generate a
%rip-relative immediate reference without splitting code into i386 and
x86-64 versions, so add a new macro to <asm/asm.h> for this purpose.

Changes in v3:
--------------
* Clarify the subject line

Changes in v2:
--------------
* Add new macro to <asm/asm.h>
* *Actually* generate the %rip-relative addressing mode.

 arch/x86/include/asm/asm.h        |  5 +++++
 arch/x86/include/asm/cpufeature.h | 13 +++++++++----
 2 files changed, 14 insertions(+), 4 deletions(-)