Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Date:   Wed, 16 Aug 2023 15:22:19 +0100
From:   Will Deacon <will@kernel.org>
To:     Mark Brown <broonie@kernel.org>
Cc:     Catalin Marinas <catalin.marinas@arm.com>,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        David Spickett <David.Spickett@arm.com>
Subject: Re: [PATCH] arm64/ptrace: Ensure that the task sees ZT writes on
 first use
Message-ID: <20230816142219.GA13526@willie-the-truck>
References: <20230814-arm64-zt-ptrace-first-use-v1-1-fc8e8022140f@kernel.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230814-arm64-zt-ptrace-first-use-v1-1-fc8e8022140f@kernel.org>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On Mon, Aug 14, 2023 at 10:27:51PM +0100, Mark Brown wrote:
> When the value of ZT is set via ptrace we don't disable traps for SME.
> This means that when a the task has never used SME before then the value
> set via ptrace will never be seen by the target task since it will
> trigger a SME access trap which will flush the register state.
> 
> Disable SME traps when setting ZT, this means we also need to allocate
> storage for SVE if it is not already allocated, for the benefit of
> streaming SVE.
> 
> Fixes: f90b529bcbe5 ("arm64/sme: Implement ZT0 ptrace support")
> Signed-off-by: Mark Brown <broonie@kernel.org>
> ---
>  arch/arm64/kernel/ptrace.c | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/arch/arm64/kernel/ptrace.c b/arch/arm64/kernel/ptrace.c
> index 5b9b4305248b..254eb37e1f07 100644
> --- a/arch/arm64/kernel/ptrace.c
> +++ b/arch/arm64/kernel/ptrace.c
> @@ -1170,6 +1170,11 @@ static int zt_set(struct task_struct *target,
>  	if (!system_supports_sme2())
>  		return -EINVAL;
>  
> +	/* Ensure SVE storage in case this is first use of SME */
> +	sve_alloc(target, false);
> +	if (!target->thread.sve_state)
> +		return -ENOMEM;
> +
>  	if (!thread_za_enabled(&target->thread)) {
>  		sme_alloc(target);
>  		if (!target->thread.sme_state)
> @@ -1182,6 +1187,8 @@ static int zt_set(struct task_struct *target,
>  	if (ret == 0)
>  		target->thread.svcr |= SVCR_ZA_MASK;
>  
> +	set_tsk_thread_flag(target, TIF_SME);

Hmm, this is now weirdly inconsistent with za_set(), which doesn't touch
the thread flag unless the regset copy succeeds. Is that intentional?

Will