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:   Thu, 22 Sep 2022 14:38:51 +0100
From:   Will Deacon <will@kernel.org>
To:     Liao Chang <liaochang1@huawei.com>
Cc:     catalin.marinas@arm.com, guoren@kernel.org,
        paul.walmsley@sifive.com, palmer@dabbelt.com,
        aou@eecs.berkeley.edu, rostedt@goodmis.org, mhiramat@kernel.org,
        maz@kernel.org, alexandru.elisei@arm.com,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        linux-csky@vger.kernel.org, linux-riscv@lists.infradead.org
Subject: Re: [PATCH V2 3/3] arm64/kprobe: Optimize the performance of
 patching single-step slot
Message-ID: <20220922133850.GC12095@willie-the-truck>
References: <20220913033454.104519-1-liaochang1@huawei.com>
 <20220913033454.104519-4-liaochang1@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220913033454.104519-4-liaochang1@huawei.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On Tue, Sep 13, 2022 at 11:34:54AM +0800, Liao Chang wrote:
> Single-step slot would not be used until kprobe is enabled, that means
> no race condition occurs on it under SMP, hence it is safe to pacth ss
> slot without stopping machine.
> 
> Signed-off-by: Liao Chang <liaochang1@huawei.com>
> ---
>  arch/arm64/kernel/probes/kprobes.c | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm64/kernel/probes/kprobes.c b/arch/arm64/kernel/probes/kprobes.c
> index d1d182320245..5902e33fd3b6 100644
> --- a/arch/arm64/kernel/probes/kprobes.c
> +++ b/arch/arm64/kernel/probes/kprobes.c
> @@ -44,11 +44,10 @@ post_kprobe_handler(struct kprobe *, struct kprobe_ctlblk *, struct pt_regs *);
>  static void __kprobes arch_prepare_ss_slot(struct kprobe *p)
>  {
>  	kprobe_opcode_t *addr = p->ainsn.api.insn;
> -	void *addrs[] = {addr, addr + 1};
> -	u32 insns[] = {p->opcode, BRK64_OPCODE_KPROBES_SS};
>  
>  	/* prepare insn slot */
> -	aarch64_insn_patch_text(addrs, insns, 2);
> +	aarch64_insn_write(addr, p->opcode);
> +	aarch64_insn_write(addr + 1, BRK64_OPCODE_KPROBES_SS);
>  
>  	flush_icache_range((uintptr_t)addr, (uintptr_t)(addr + MAX_INSN_SIZE));

Hmm, so it looks like prior to your change we were doing the cache
maintebnance twice: once in aarch64_insn_patch_text() from stop machine
context and then again in the flush_icache_range() call above.

I suppose the cleanest thing would be to drop the flush_icache_range()
call and then use aarch64_insn_patch_text_nosync() instead of
aarch64_insn_write() in your change.

Will