Re: [PATCH 2/3] eventpoll: support non-blocking do_epoll_ctl() calls

[Jens Axboe <axboe@kernel.dk>]

On 1/22/20 9:20 AM, Jann Horn wrote:
> On Wed, Jan 22, 2020 at 5:02 PM Jens Axboe <axboe@kernel.dk> wrote:
>> Also make it available outside of epoll, along with the helper that
>> decides if we need to copy the passed in epoll_event.
> [...]
>> diff --git a/fs/eventpoll.c b/fs/eventpoll.c
>> index cd848e8d08e2..162af749ea50 100644
>> --- a/fs/eventpoll.c
>> +++ b/fs/eventpoll.c
> [...]
>> -static int do_epoll_ctl(int epfd, int op, int fd, struct epoll_event *epds)
>> +static inline int epoll_mutex_lock(struct mutex *mutex, int depth,
>> +                                  bool nonblock)
>> +{
>> +       if (!nonblock) {
>> +               mutex_lock_nested(mutex, depth);
>> +               return 0;
>> +       }
>> +       if (!mutex_trylock(mutex))
>> +               return 0;
>> +       return -EAGAIN;
> 
> The documentation for mutex_trylock() says:
> 
>  * Try to acquire the mutex atomically. Returns 1 if the mutex
>  * has been acquired successfully, and 0 on contention.
> 
> So in the success case, this evaluates to:
> 
>     if (!1)
>       return 0;
>     return -EAGAIN;
> 
> which is
> 
>     if (0)
>       return 0;
>     return -EAGAIN;
> 
> which is
> 
>     return -EAGAIN;
> 
> I think you'll have to get rid of the negation.

Doh indeed. I'll rework and run the test case, just rebased this and I
think I inadvertently used an older version. Ditto for the below.

-- 
Jens Axboe