Re: [PATCH 1/1] aio: make sure the input "timeout" value is valid

[Leizhen (ThunderTown)]

On 2017/12/14 3:31, Matthew Wilcox wrote:
> On Wed, Dec 13, 2017 at 11:27:00AM -0500, Jeff Moyer wrote:
>> Matthew Wilcox <willy@xxxxxxxxxxxxx> writes:
>>
>>> On Wed, Dec 13, 2017 at 09:42:52PM +0800, Zhen Lei wrote:
>>>> Below information is reported by a lower kernel version, and I saw the
>>>> problem still exist in current version.
>>>
>>> I think you're right, but what an awful interface we have here!
>>> The user must not only fetch it, they must validate it separately?
>>> And if they forget, then userspace is provoking undefined behaviour?  Ugh.
>>> Why not this:
>>
>> Why not go a step further and have get_timespec64 check for validity?
>> I wonder what caller doesn't want that to happen...
I tried this before. But I found some places call get_timespec64 in the following function.
If we do the check in get_timespec64, the check will be duplicated.

For example:
static long do_pselect(int n, fd_set __user *inp, fd_set __user *outp,
....
	if (get_timespec64(&ts, tsp))
		return -EFAULT;

	to = &end_time;
	if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec))

int poll_select_set_timeout(struct timespec64 *to, time64_t sec, long nsec)
{
	struct timespec64 ts = {.tv_sec = sec, .tv_nsec = nsec};

	if (!timespec64_valid(&ts))
		return -EINVAL;

> 
> There are some which don't today.  I'm hoping Deepa takes this and goes
> off and fixes them all up.
As my search results, just the case I mentioned above, which may cause duplicate check.
So if we don't care the slightly performance drop, maybe we should do timespec64_valid
check in get_timespec64. I can try this in v2. Otherwise, use your method.

> 
> .
> 

-- 
Thanks!
BestRegards