Re: [PATCH v3] io_uring: limit inflight IO

[Jens Axboe <axboe@kernel.dk>]

On 11/9/19 12:16 PM, Pavel Begunkov wrote:
>> @@ -2992,6 +2992,30 @@ static bool io_get_sqring(struct io_ring_ctx *ctx, struct sqe_submit *s)
>>   	return false;
>>   }
>>   
>> +static bool io_sq_over_limit(struct io_ring_ctx *ctx, unsigned to_submit)
>> +{
>> +	unsigned inflight;
>> +
>> +	if ((ctx->flags & IORING_SETUP_CQ_NODROP) &&
>> +	    !list_empty(&ctx->cq_overflow_list))
>> +		return true;
>> +
>> +	/*
>> +	 * This doesn't need to be super precise, so only check every once
>> +	 * in a while.
>> +	 */
>> +	if ((ctx->cached_sq_head & ctx->sq_mask) !=
>> +	    ((ctx->cached_sq_head + to_submit) & ctx->sq_mask))
>> +		return false;
> 
> ctx->sq_mask = sq_entries - 1, e.g. 0x0000...ffff.
> Thus the code above is modular arithmetic (% sq_entries) and
> equivalent to:
> 
> if (to_submit != sq_entries)
> 	return false;
>   
> I suggest, the intention was:
> 
> if ((sq_head & ~mask) == ((sq_head + to_submit) & ~mask))
> 	return false;

Hah indeed, that was pretty silly. Fixed.

-- 
Jens Axboe