Re: [PATCH v3] ARM: smp: add support for per-task stack canaries

From: Guenter Roeck
Date: Wed Mar 11 2020 - 17:39:38 EST


On 3/11/20 1:45 PM, Russell King - ARM Linux admin wrote:
> On Wed, Mar 11, 2020 at 11:47:20AM -0700, Kees Cook wrote:
>> On Wed, Mar 11, 2020 at 11:31:13AM -0700, Guenter Roeck wrote:
>>> On 3/11/20 10:21 AM, Kees Cook wrote:
>>>> On Mon, Mar 09, 2020 at 09:49:31AM -0700, Guenter Roeck wrote:
>>>>> On Thu, Dec 06, 2018 at 09:32:57AM +0100, 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
>>>>>> on any CPU, 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.
>>>>>>
>>>>>> Cc: Russell King <linux@xxxxxxxxxxxxxxx>
>>>>>> Cc: Kees Cook <keescook@xxxxxxxxxxxx>
>>>>>> Cc: Emese Revfy <re.emese@xxxxxxxxx>
>>>>>> Cc: Arnd Bergmann <arnd@xxxxxxxx>
>>>>>> Cc: Laura Abbott <labbott@xxxxxxxxxx>
>>>>>> Cc: kernel-hardening@xxxxxxxxxxxxxxxxxx
>>>>>> Acked-by: Nicolas Pitre <nico@xxxxxxxxxx>
>>>>>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx>
>>>>>
>>>>> Since this patch is in the tree, cc-option no longer works on
>>>>> the arm architecture if CONFIG_STACKPROTECTOR_PER_TASK is enabled.
>>>>>
>>>>> Any idea how to fix that ?
>>>>
>>>> I thought Arnd sent a patch to fix it and it got picked up?
>>>>
>>>
>>> Yes, but the fix is not upstream (it is only in -next), and I missed it.
>>
>> Ah, yes, I found it again now too; it went through rmk's tree.
>>
>> For thread posterity:
>>
>> ARM: 8961/2: Fix Kbuild issue caused by per-task stack protector GCC plugin
>> https://www.arm.linux.org.uk/developer/patches/viewpatch.php?id=8961/2
>
> It's in my fixes branch, waiting for me to do my (now usual) push
> of fixes to Linus.
>
> I'm not sure whether the above discussion is suggesting that there's
> a problem with this patch, or whether it's trying to encourage me
> to send it up to Linus. I _think_ there's the suggestion that it
> causes a regression, hmm?
> ARM: 8961/2: Fix Kbuild issue caused by per-task stack protector GCC plugin

Commit 189af4657186da08 ("ARM: smp: add support for per-task stack canaries")
caused a regression. "ARM: 8961/2: Fix Kbuild issue caused by per-task stack
protector GCC plugin" fixes it.

Guenter