Linux RDMA and InfiniBand development
 help / color / mirror / Atom feed
From: Zhu Yanjun <yanjun.zhu@linux.dev>
To: Honggang LI <honggangli@163.com>
Cc: zyjzyj2000@gmail.com, jgg@ziepe.ca, leon@kernel.org,
	linux-rdma@vger.kernel.org
Subject: Re: [PATCH 1/1] RDMA/rxe: Fix the qp flush warnings in req
Date: Sat, 26 Oct 2024 07:52:31 +0200	[thread overview]
Message-ID: <267804d9-df81-489e-8d8b-6dcee7859ae0@linux.dev> (raw)
In-Reply-To: <ZxxMqWp0oCJxPq64@fc39>


在 2024/10/26 3:58, Honggang LI 写道:
> On Fri, Oct 25, 2024 at 05:20:36PM +0200, Zhu Yanjun wrote:
>> ---
>>   drivers/infiniband/sw/rxe/rxe_req.c | 6 ++++--
>>   1 file changed, 4 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/infiniband/sw/rxe/rxe_req.c b/drivers/infiniband/sw/rxe/rxe_req.c
>> index 479c07e6e4ed..87a02f0deb00 100644
>> --- a/drivers/infiniband/sw/rxe/rxe_req.c
>> +++ b/drivers/infiniband/sw/rxe/rxe_req.c
>> @@ -663,10 +663,12 @@ int rxe_requester(struct rxe_qp *qp)
>>   	if (unlikely(qp_state(qp) == IB_QPS_ERR)) {
>>   		wqe = __req_next_wqe(qp);
>>   		spin_unlock_irqrestore(&qp->state_lock, flags);
>> -		if (wqe)
>> +		if (wqe) {
>> +			wqe->status = IB_WC_WR_FLUSH_ERR;
>                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>
> Why not update wqe->status in function `flush_send_wqe()` ?

flush_send_wqe is to handle the cqe in cq queue. Please see the source 
code as below.

static int flush_send_wqe(struct rxe_qp *qp, struct rxe_send_wqe *wqe)
{
     struct rxe_cqe cqe = {};
     struct ib_wc *wc = &cqe.ibwc;
     struct ib_uverbs_wc *uwc = &cqe.uibwc;
     int err;

     if (qp->is_user) {
         uwc->wr_id = wqe->wr.wr_id;
         uwc->status = IB_WC_WR_FLUSH_ERR;
         uwc->qp_num = qp->ibqp.qp_num;
     } else {
         wc->wr_id = wqe->wr.wr_id;
         wc->status = IB_WC_WR_FLUSH_ERR;
         wc->qp = &qp->ibqp;
     }

     err = rxe_cq_post(qp->scq, &cqe, 0);
     if (err)
         rxe_dbg_cq(qp->scq, "post cq failed, err = %d\n", err);

     return err;
}

This error occurs in send queue. Please see the source code as below.

static struct rxe_send_wqe *__req_next_wqe(struct rxe_qp *qp)
{
     struct rxe_queue *q = qp->sq.queue;
     unsigned int index = qp->req.wqe_index;
     unsigned int prod;

     prod = queue_get_producer(q, QUEUE_TYPE_FROM_CLIENT);
     if (index == prod)
         return NULL;
     else
         return queue_addr_from_index(q, index);
}

This is why we should set the error status in send queue error handler.

Thanks,

Zhu Yanjun

>
> thanks
>
-- 
Best Regards,
Yanjun.Zhu


  reply	other threads:[~2024-10-26  5:52 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-10-25 15:20 [PATCH 1/1] RDMA/rxe: Fix the qp flush warnings in req Zhu Yanjun
2024-10-26  1:58 ` Honggang LI
2024-10-26  5:52   ` Zhu Yanjun [this message]
2024-10-30 12:22 ` Leon Romanovsky

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=267804d9-df81-489e-8d8b-6dcee7859ae0@linux.dev \
    --to=yanjun.zhu@linux.dev \
    --cc=honggangli@163.com \
    --cc=jgg@ziepe.ca \
    --cc=leon@kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=zyjzyj2000@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox