Re: [PATCH 9/9] io_uring: optimise io_get_cqe()

[Pavel Begunkov <asml.silence@gmail.com>]

On 4/12/22 17:06, Florian Schmaus wrote:
> On 12/04/2022 16.09, Pavel Begunkov wrote:
>> io_get_cqe() is expensive because of a bunch of loads, masking, etc.
>> However, most of the time we should have enough of entries in the CQ,
>> so we can cache two pointers representing a range of contiguous CQE
>> memory we can use. When the range is exhausted we'll go through a slower
>> path to set up a new range. When there are no CQEs avaliable, pointers
>> will naturally point to the same address.
>>
>> Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
>> ---
>>   fs/io_uring.c | 46 +++++++++++++++++++++++++++++++++++-----------
>>   1 file changed, 35 insertions(+), 11 deletions(-)
>>
>> diff --git a/fs/io_uring.c b/fs/io_uring.c
>> index b349a3c52354..f2269ffe09eb 100644
>> --- a/fs/io_uring.c
>> +++ b/fs/io_uring.c
>> @@ -416,6 +416,13 @@ struct io_ring_ctx {
>>       unsigned long        check_cq_overflow;
>>       struct {
>> +        /*
>> +         * We cache a range of free CQEs we can use, once exhausted it
>> +         * should go through a slower range setup, see __io_get_cqe()
>> +         */
>> +        struct io_uring_cqe    *cqe_cached;
>> +        struct io_uring_cqe    *cqe_santinel;
> 
> I think this should s/santinel/sentinel.

Indeed, thanks

-- 
Pavel Begunkov