Re: [PATCH v6 3/7] Reimplement RLIMIT_NPROC on top of ucounts

[Jens Axboe <axboe@kernel.dk>]

On 2/22/21 3:11 AM, Alexey Gladkov wrote:
> On Sun, Feb 21, 2021 at 04:38:10PM -0700, Jens Axboe wrote:
>> On 2/15/21 5:41 AM, Alexey Gladkov wrote:
>>> diff --git a/fs/io-wq.c b/fs/io-wq.c
>>> index a564f36e260c..5b6940c90c61 100644
>>> --- a/fs/io-wq.c
>>> +++ b/fs/io-wq.c
>>> @@ -1090,10 +1091,7 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
>>>  		wqe->node = alloc_node;
>>>  		wqe->acct[IO_WQ_ACCT_BOUND].max_workers = bounded;
>>>  		atomic_set(&wqe->acct[IO_WQ_ACCT_BOUND].nr_running, 0);
>>> -		if (wq->user) {
>>> -			wqe->acct[IO_WQ_ACCT_UNBOUND].max_workers =
>>> -					task_rlimit(current, RLIMIT_NPROC);
>>> -		}
>>> +		wqe->acct[IO_WQ_ACCT_UNBOUND].max_workers = task_rlimit(current, RLIMIT_NPROC);
>>
>> This doesn't look like an equivalent transformation. But that may be
>> moot if we merge the io_uring-worker.v3 series, as then you would not
>> have to touch io-wq at all.
> 
> In the current code the wq->user is always set to current_user():
> 
> io_uring_create [1]
> `- io_sq_offload_create
>    `- io_init_wq_offload [2]
>       `-io_wq_create [3]

current vs other wasn't my concern, but we're always setting ->user so
the test was pointless. So looks fine to me.

-- 
Jens Axboe