On 10/19/21 7:44 AM, Christoph Hellwig wrote:
> On Tue, Oct 19, 2021 at 07:43:04AM -0600, Jens Axboe wrote:
>> On 10/19/21 7:39 AM, Christoph Hellwig wrote:
>>> If RQF_ELVPRIV is set RQF_ELV is by definition set as well.
>>>
>>> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
>>> ---
>>> block/blk-mq.c | 2 +-
>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/block/blk-mq.c b/block/blk-mq.c
>>> index 428e0e0fd5504..34392c439d2a8 100644
>>> --- a/block/blk-mq.c
>>> +++ b/block/blk-mq.c
>>> @@ -580,7 +580,7 @@ void blk_mq_free_request(struct request *rq)
>>> struct request_queue *q = rq->q;
>>> struct blk_mq_hw_ctx *hctx = rq->mq_hctx;
>>>
>>> - if (rq->rq_flags & (RQF_ELVPRIV | RQF_ELV)) {
>>> + if (rq->rq_flags & RQF_ELV) {
>>
>> Actually just fixed a bug there. RQF_ELV means "we have an IO
>> scheduler", and RQF_ELVPRIV means that plus "we have rq private data".
>> The above shouldn't check RQF_ELV at all, just PRIV.
>
> Well, in that case RQF_ELVPRIV can be replaced with
> RQF_ELV && !op_is_flush as in the next patch. But I can resend once I
> see the fix in a tree somewhere.
That'd be great. It's in for-5.16/block
--
Jens Axboe
