Re: [PATCH v2] x86/suspend: fix false positive KASAN warning on suspend/resume

From: Dmitry Vyukov <dvyukov@google.com>
To: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>,
	"Rafael J. Wysocki" <rjw@rjwysocki.net>,
	Len Brown <len.brown@intel.com>, Pavel Machek <pavel@ucw.cz>,
	linux-pm@vger.kernel.org, LKML <linux-kernel@vger.kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@kernel.org>,
	Andy Lutomirski <luto@amacapital.net>,
	Scott Bauer <scott.bauer@intel.com>,
	"x86@kernel.org" <x86@kernel.org>,
	Alexander Potapenko <glider@google.com>,
	kasan-dev <kasan-dev@googlegroups.com>
Subject: Re: [PATCH v2] x86/suspend: fix false positive KASAN warning on suspend/resume
Date: Fri, 2 Dec 2016 15:02:46 +0100	[thread overview]
Message-ID: <CACT4Y+a1jSkEW2KVh+=6wg12qGEB0nf=zrcNoD7raFaPY8e5Vg@mail.gmail.com> (raw)
In-Reply-To: <20161202140147.gvj452hmlbxnstrg@treble>

On Fri, Dec 2, 2016 at 3:01 PM, Josh Poimboeuf <jpoimboe@redhat.com> wrote:
> On Fri, Dec 02, 2016 at 04:41:09PM +0300, Andrey Ryabinin wrote:
>>
>>
>> On 12/01/2016 11:31 PM, Josh Poimboeuf wrote:
>>
>> >  arch/x86/kernel/acpi/wakeup_64.S | 16 ++++++++++++++++
>> >  1 file changed, 16 insertions(+)
>> >
>> > diff --git a/arch/x86/kernel/acpi/wakeup_64.S b/arch/x86/kernel/acpi/wakeup_64.S
>> > index 169963f..1df9b75 100644
>> > --- a/arch/x86/kernel/acpi/wakeup_64.S
>> > +++ b/arch/x86/kernel/acpi/wakeup_64.S
>> > @@ -109,6 +109,22 @@ ENTRY(do_suspend_lowlevel)
>> >     movq    pt_regs_r14(%rax), %r14
>> >     movq    pt_regs_r15(%rax), %r15
>> >
>> > +#ifdef CONFIG_KASAN
>> > +   /*
>> > +    * The suspend path may have poisoned some areas deeper in the stack,
>> > +    * which we now need to unpoison.
>> > +    *
>> > +    * We can't call kasan_unpoison_task_stack_below() because it uses %gs
>> > +    * for 'current', which hasn't been set up yet.  Instead, calculate the
>> > +    * stack range manually and call kasan_unpoison_shadow().
>> > +    */
>> > +   movq    %rsp, %rdi
>> > +   andq    $CURRENT_MASK, %rdi
>> > +   movq    %rsp, %rsi
>> > +   xorq    %rdi, %rsi
>> > +   call    kasan_unpoison_shadow
>> > +#endif
>> > +
>>
>> Looks good, but in fact we can use kasan_unpoison_task_stack_below(). We just need to change it a little:
>>
>> diff --git a/mm/kasan/kasan.c b/mm/kasan/kasan.c
>> index 70c0097..e779236 100644
>> --- a/mm/kasan/kasan.c
>> +++ b/mm/kasan/kasan.c
>> @@ -80,7 +80,9 @@ void kasan_unpoison_task_stack(struct task_struct *task)
>>  /* Unpoison the stack for the current task beyond a watermark sp value. */
>>  asmlinkage void kasan_unpoison_task_stack_below(const void *watermark)
>>  {
>> -       __kasan_unpoison_stack(current, watermark);
>> +       void *base = (void *)((unsigned long)watermark & ~(THREAD_SIZE - 1));
>> +
>> +       kasan_unpoison_shadow(base, watermark - base);
>>  }
>>
>>
>> With this we don't have to calculate stack range in assembly.
>
> That is better indeed, will do a v3.

agree