[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Mon, 5 Nov 2018 11:42:29 +0100
From: Ard Biesheuvel <ard.biesheuvel@...aro.org>
To: Nicolas Pitre <nicolas.pitre@...aro.org>
Cc: linux-arm-kernel <linux-arm-kernel@...ts.infradead.org>,
Russell King - ARM Linux <linux@...linux.org.uk>, Kees Cook <keescook@...omium.org>,
Arnd Bergmann <arnd@...db.de>, Emese Revfy <re.emese@...il.com>,
Kernel Hardening <kernel-hardening@...ts.openwall.com>, Laura Abbott <labbott@...hat.com>,
Marc Zyngier <marc.zyngier@....com>
Subject: Re: [RFC/RFT PATCH] ARM: smp: add support for per-task stack canaries
On 4 November 2018 at 21:36, Nicolas Pitre <nicolas.pitre@...aro.org> wrote:
> On Sun, 4 Nov 2018, Ard Biesheuvel wrote:
>
>> On ARM, we currently only change the value of the stack canary when
>> switching tasks if the kernel was built for UP. On SMP kernels, this
>> is impossible since the stack canary value is obtained via a global
>> symbol reference, which means
>> a) all running tasks on all CPUs must use the same value
>> b) we can only modify the value when no kernel stack frames are live,
>> which is effectively never.
>>
>> So instead, use a GCC plugin to add a RTL pass that replaces each
>> reference to the address of the __stack_chk_guard symbol with an
>> expression that produces the address of the 'stack_canary' field
>> that is added to struct thread_info. This way, each task will use
>> its own randomized value.
>>
>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@...aro.org>
>> ---
>> Known issues:
>> - The plugin requires the values of THREAD_SIZE and the offset of
>> stack_canary in struct thread_info at a really early stage in the
>> build, so I had to redefine them in the plugin source.
>
> Can't you rely on the asm-offsets.c mechanism before the plugin is used?
>
Not currently, no. The plugins are built with the host compiler before
the target compiler is invoked, and asm-offsets is built with the
target compiler.
Kees: any ideas on how to resolve this?
>> Questions:
>> - Are there any execution contexts except for the decompressor and the
>> EFI stub where we should disable this? KVM perhaps?
>
> Most certainly head-inflate-data.c... who knows if someone will use an
> XIP kernel on an SMP system someday.
>
That is a bit tricky, unfortunately: building head-inflate-data.c with
the stack protector disabled is easy, but we'll also need to build
zlib_inflate() in the same way.
> [...]
>> +#ifdef CONFIG_STACKPROTECTOR_PER_TASK
>> + BUILD_BUG_ON(offsetof(struct thread_info, stack_canary) != 100);
>> + BUILD_BUG_ON(THREAD_SIZE != 0x2000);
>
> You should at least define those values in a header file that gets
> included by both this code and the plugin source. But asm-offsets.c
> would be best.
>
Sure. I will put the right hand sides in a shared header as #defines
and use those in the plugin source.
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
