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:   Wed, 29 Mar 2023 17:15:16 +0200
From:   Oleg Nesterov <oleg@redhat.com>
To:     Gregory Price <gourry.memverge@gmail.com>
Cc:     linux-kernel@vger.kernel.org, linux-doc@vger.kernel.org,
        linux-arm-kernel@lists.infradead.org, linux-arch@vger.kernel.org,
        avagin@gmail.com, peterz@infradead.org, luto@kernel.org,
        krisman@collabora.com, tglx@linutronix.de, corbet@lwn.net,
        shuah@kernel.org, catalin.marinas@arm.com, arnd@arndb.de,
        will@kernel.org, mark.rutland@arm.com, tongtiangen@huawei.com,
        robin.murphy@arm.com, Gregory Price <gregory.price@memverge.com>
Subject: Re: [PATCH v14 1/4] asm-generic,arm64: create task variant of
 access_ok
Message-ID: <20230329151515.GA913@redhat.com>
References: <20230328164811.2451-1-gregory.price@memverge.com>
 <20230328164811.2451-2-gregory.price@memverge.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230328164811.2451-2-gregory.price@memverge.com>
User-Agent: Mutt/1.5.24 (2015-08-30)
Precedence: bulk

Hmm. I am not comfortable with this change...

I won't really argue because I don't have a better solution and because
I think we don't really care as long as task_set_syscall_user_dispatch()
is the only user of task_access_ok(), but still...

OK, so this version changes set_syscall_user_dispatch() to use
task_access_ok() instead of access_ok() because task != current.

On 03/28, Gregory Price wrote:
>
> If the architecture does not implement task_access_ok, the operation
> reduces to access_ok and the task argument is discarded.

No, with this patch it reduces to __access_ok(). And this already doesn't
look very good to me, but this is minor.

> --- a/include/asm-generic/access_ok.h
> +++ b/include/asm-generic/access_ok.h
> @@ -45,4 +45,14 @@ static inline int __access_ok(const void __user *ptr, unsigned long size)
>  #define access_ok(addr, size) likely(__access_ok(addr, size))
>  #endif
>
> +/*
> + * Some architectures may have special features (such as ARM MTE)
> + * that require handling if access_ok is called on a pointer from one
> + * task in the context of another.  On most architectures this operation
> + * is equivalent to simply __access_ok.
> + */
> +#ifndef task_access_ok
> +#define task_access_ok(task, addr, size) likely(__access_ok(addr, size))
> +#endif

Lets ignore arm64.

This look as if access_ok() or __access_ok() doesn't depend on task, but
this is not true in general. Say, TASK_SIZE_MAX can check is_32bit_task()
test_thread_flag(TIF_32BIT...) and this uses "current".

Again, we probably do not care, but I don't like the fact task_access_ok()
looks as if task_access_ok(task) returns the same result as "task" calling
access_ok().

Oleg.