[PATCH] arm64: percpu: Make this_cpu accessors pre-empt safe

[Steve Capper]

On 19 March 2015 at 15:44, Mark Rutland <mark.rutland at arm.com> wrote:
> Hi Steve,
>

Hey Mark,

> Thanks for putting this together!
>
> On Thu, Mar 19, 2015 at 02:52:31PM +0000, Steve Capper wrote:
>> this_cpu operations were implemented for arm64 in:
>>  5284e1b arm64: xchg: Implement cmpxchg_double
>>  f97fc81 arm64: percpu: Implement this_cpu operations
>>
>> Unfortunately, it is possible for pre-emption to take place between
>> address generation and data access. This can lead to cases where data
>> is being manipulated by this_cpu for a different CPU than it was
>> called on. Which effectively breaks the spec.
>>
>> This patch disables pre-emption for the this_cpu operations
>> guaranteeing that address generation and data manipulation.
>
> Shouldn't that last sentence end with "occur on the same CPU", or
> something like that?

Gah! Yes it should, thank you.

>
> [...]
>
>> +/*
>> + * Modules aren't allowed to use preempt_enable_no_resched, and it is
>> + * undef'ed. If we are unable to use preempt_enable_no_resched, then
>> + * fallback to the standard preempt_enable.
>> + */
>> +#ifdef preempt_enable_no_resched
>> +#define __pcp_preempt_enable()       preempt_enable_no_resched()
>> +#else
>> +#define __pcp_preempt_enable()       preempt_enable()
>> +#endif /* preempt_enable_no_resched */
>
> I think it would be worth mentioning in the comment why we want to use
> preempt_enable_no_resched where possible (e.g. read-modify-cmpxchg
> sequences where we want to have as few retries as possible).

Good point, I'll add something.


>
> Other than those points, the patch looks good to me, feel free to add:
>
> Reviewed-by: Mark Rutland <mark.rutland at arm.com>

Thanks!

>
> It's a shame there don't seem to be any this_cpu_* self-tests; I've
> booted a kernel with this applied, but I didn't have anything that
> exploded without this, so I'd feel uneasy giving a Tested-by.

There is the percpu_test module, I ran this and it appeared to pass.
Also, I ran the traditional hackbench tests.

Thanks Mark, I'll send out a V2 shortly with the amended commit log
and expanded comment.

Cheers,
-- 
Steve

>
> Mark.