Re: [PATCH v14 1/4] asm-generic,arm64: create task variant of access_ok

From: Gregory Price <gregory.price@memverge.com>
To: Oleg Nesterov <oleg@redhat.com>
Cc: Arnd Bergmann <arnd@arndb.de>,
	Gregory Price <gourry.memverge@gmail.com>,
	linux-kernel@vger.kernel.org, linux-doc@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org,
	Linux-Arch <linux-arch@vger.kernel.org>,
	avagin@gmail.com, Peter Zijlstra <peterz@infradead.org>,
	Andy Lutomirski <luto@kernel.org>,
	krisman@collabora.com, Thomas Gleixner <tglx@linutronix.de>,
	Jonathan Corbet <corbet@lwn.net>, shuah <shuah@kernel.org>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Will Deacon <will@kernel.org>,
	Mark Rutland <mark.rutland@arm.com>,
	tongtiangen@huawei.com, Robin Murphy <robin.murphy@arm.com>
Subject: Re: [PATCH v14 1/4] asm-generic,arm64: create task variant of access_ok
Date: Wed, 29 Mar 2023 01:37:40 -0400	[thread overview]
Message-ID: <ZCPOpClZ3hOQCs7a@memverge.com> (raw)
In-Reply-To: <20230329171322.GB4477@redhat.com>

On Wed, Mar 29, 2023 at 07:13:22PM +0200, Oleg Nesterov wrote:
> On 03/28, Gregory Price wrote:
> >
> > Not sure how I should proceed here,
> 
> Can't we just kill this access_ok() in set_syscall_user_dispatch() ?
> I don't think it buys too much.
> 
> Oleg.
> 
> diff --git a/kernel/entry/syscall_user_dispatch.c b/kernel/entry/syscall_user_dispatch.c
> index 0b6379adff6b..d2e516ece52b 100644
> --- a/kernel/entry/syscall_user_dispatch.c
> +++ b/kernel/entry/syscall_user_dispatch.c
> @@ -43,11 +43,7 @@ bool syscall_user_dispatch(struct pt_regs *regs)
>  		return false;
>  
>  	if (likely(sd->selector)) {
> -		/*
> -		 * access_ok() is performed once, at prctl time, when
> -		 * the selector is loaded by userspace.
> -		 */
> -		if (unlikely(__get_user(state, sd->selector))) {
> +		if (unlikely(get_user(state, sd->selector))) {
>  			force_exit_sig(SIGSEGV);
>  			return true;
>  		}
> @@ -86,9 +82,6 @@ int set_syscall_user_dispatch(unsigned long mode, unsigned long offset,
>  		if (offset && offset + len <= offset)
>  			return -EINVAL;
>  
> -		if (selector && !access_ok(selector, sizeof(*selector)))
> -			return -EFAULT;
> -
>  		break;
>  	default:
>  		return -EINVAL;
> 

The result of this would be either a task calling via prctl or a tracer
calling via ptrace would be capable of setting selector to a bad pointer
and producing a SIGSEGV on the next system call.

It's a pretty small footgun, but maybe that's reasonable?

From a user perspective, debugging this behavior would be nightmarish.
Your call to prctl/ptrace would succeed and the process would continue
to execute until the next syscall - at which point you incur a SIGSEGV,
and depending on the syscall that could mean anything?

Everything feels bad here lol.

~Gregory