Re: [PATCH v2] arm64: add ptrace regsets for ptrauth key management

From: Kristina Martsenko <kristina.martsenko@arm.com>
To: Dave Martin <Dave.Martin@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Amit Kachhap <amit.kachhap@arm.com>,
	Will Deacon <will.deacon@arm.com>,
	linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v2] arm64: add ptrace regsets for ptrauth key management
Date: Tue, 22 Jan 2019 19:08:25 +0000	[thread overview]
Message-ID: <b8afbda4-c782-ceee-8294-a96a1fec02df@arm.com> (raw)
In-Reply-To: <20190116151254.GA3578@e103592.cambridge.arm.com>

On 16/01/2019 15:13, Dave Martin wrote:
> On Tue, Jan 15, 2019 at 07:32:30PM +0000, Kristina Martsenko wrote:
>> On 11/01/2019 13:58, Dave Martin wrote:
>>> On Thu, Jan 10, 2019 at 07:41:15PM +0000, Kristina Martsenko wrote:
>>>> On 10/01/2019 19:35, Kristina Martsenko wrote:
>>>> @@ -80,12 +65,12 @@ static inline unsigned long ptrauth_strip_insn_pac(unsigned long ptr)
>>>>  #define ptrauth_thread_init_user(tsk)					\
>>>>  do {									\
>>>>  	struct task_struct *__ptiu_tsk = (tsk);				\
>>>
>>> Not added by this patch, but __ptiu_tsk doesn't seem to do anything
>>> except make the subsquent lines more verbose than otherwise (and pollute
>>> the identifier namespace -- though unlikely to be a problem).
>>>
>>> It may not be worth dropping it now that it's there though.
>>
>> Using __ptiu_tsk prevents the argument (tsk) from being evaluated twice,
>> which could have side effects.
> 
> Ah, right.
> 
> Actually, could this be a function instead?  That would avoid multiple-
> evaulation a clean way.

If it were a function, then this file (pointer_auth.h) would need to
#include <linux/sched.h> (for the struct task_struct definition), which
would create a circular header dependency, since sched.h #includes
asm/processor.h, which #includes pointer_auth.h. Alternatively the
function could be moved to pointer_auth.c, but that would prevent it
from being inlined, so I'd prefer to keep it as a macro for now.

>>>> -	ptrauth_keys_init(&__ptiu_tsk->thread.keys_user);		\
>>>> -	ptrauth_keys_switch(&__ptiu_tsk->thread.keys_user);		\
>>>> +	ptrauth_keys_init(&__ptiu_tsk->thread.uw.keys_user);		\
>>>> +	ptrauth_keys_switch(&__ptiu_tsk->thread.uw.keys_user);		\
>>>>  } while (0)
>>>>  
>>>>  #define ptrauth_thread_switch(tsk)	\
>>>> -	ptrauth_keys_switch(&(tsk)->thread.keys_user)
>>>> +	ptrauth_keys_switch(&(tsk)->thread.uw.keys_user)
> 
> Similarly, can this be a function?

Same reasons as above.

> (Technically tsk may be evaulated twice in this macro.  Given the way
> these macros are used, I'm not sure that matters though.)

tsk is only used once here, so I think it's only evaluated once.

>> Also note that with this patch we have struct "user_pac_address_keys" in
>> struct ptrauth_keys, which may be confusing once we start using pointer
>> authentication in the kernel and use struct ptrauth_keys for kernel keys
>> as well, not just user keys.
> 
> Not a big deal either way.
> 
> The main thing to freeze now are the user ABI and the UAPI header.  We
> can refactor the kernel's internal implementation later if we want.

Ok.

Kristina

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel