All of lore.kernel.org
 help / color / mirror / Atom feed
From: Weihang Li <liweihang@huawei.com>
To: <dledford@redhat.com>, <jgg@nvidia.com>
Cc: <leon@kernel.org>, <linux-rdma@vger.kernel.org>,
	<linuxarm@openeuler.org>
Subject: [PATCH for-next 05/12] RDMA/hns: Remove the reserved WQE of SRQ
Date: Sat, 30 Jan 2021 16:58:03 +0800	[thread overview]
Message-ID: <1611997090-48820-6-git-send-email-liweihang@huawei.com> (raw)
In-Reply-To: <1611997090-48820-1-git-send-email-liweihang@huawei.com>

From: Wenpeng Liang <liangwenpeng@huawei.com>

Each SRQs contain an reserved WQE, it is inappropriate and should be
removed.

Fixes: c7bcb13442e1 ("RDMA/hns: Add SRQ support for hip08 kernel mode")
Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
---
 drivers/infiniband/hw/hns/hns_roce_device.h | 1 +
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c  | 6 +++---
 drivers/infiniband/hw/hns/hns_roce_srq.c    | 6 ++++--
 3 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 916e031..d6a846b 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -54,6 +54,7 @@
 /* Hardware specification only for v1 engine */
 #define HNS_ROCE_MIN_CQE_NUM			0x40
 #define HNS_ROCE_MIN_WQE_NUM			0x20
+#define HNS_ROCE_MIN_SRQ_WQE_NUM		1
 
 /* Hardware specification only for v1 engine */
 #define HNS_ROCE_MAX_INNER_MTPT_NUM		0x7
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index b13775a..49a8456 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -860,7 +860,7 @@ int hns_roce_srqwq_overflow(struct hns_roce_srq *srq, int nreq)
 	unsigned int cur;
 
 	cur = idx_que->head - idx_que->tail;
-	return cur + nreq >= srq->wqe_cnt - 1;
+	return cur + nreq >= srq->wqe_cnt;
 }
 
 static int find_empty_entry(struct hns_roce_idx_que *idx_que,
@@ -5338,7 +5338,7 @@ static int hns_roce_v2_modify_srq(struct ib_srq *ibsrq,
 		return -EINVAL;
 
 	if (srq_attr_mask & IB_SRQ_LIMIT) {
-		if (srq_attr->srq_limit >= srq->wqe_cnt)
+		if (srq_attr->srq_limit > srq->wqe_cnt)
 			return -EINVAL;
 
 		mailbox = hns_roce_alloc_cmd_mailbox(hr_dev);
@@ -5401,7 +5401,7 @@ static int hns_roce_v2_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr)
 				  SRQC_BYTE_8_SRQ_LIMIT_WL_S);
 
 	attr->srq_limit = limit_wl;
-	attr->max_wr = srq->wqe_cnt - 1;
+	attr->max_wr = srq->wqe_cnt;
 	attr->max_sge = srq->max_gs - srq->rsv_sge;
 
 out:
diff --git a/drivers/infiniband/hw/hns/hns_roce_srq.c b/drivers/infiniband/hw/hns/hns_roce_srq.c
index 47e66fe..5d20b30 100644
--- a/drivers/infiniband/hw/hns/hns_roce_srq.c
+++ b/drivers/infiniband/hw/hns/hns_roce_srq.c
@@ -320,7 +320,7 @@ int hns_roce_create_srq(struct ib_srq *ib_srq,
 
 	max_sge = proc_srq_sge(hr_dev, srq, !!udata);
 
-	if (init_attr->attr.max_wr >= hr_dev->caps.max_srq_wrs ||
+	if (init_attr->attr.max_wr > hr_dev->caps.max_srq_wrs ||
 	    init_attr->attr.max_sge > max_sge) {
 		ibdev_err(&hr_dev->ib_dev,
 			  "SRQ config error, depth = %u, sge = %d\n",
@@ -331,7 +331,9 @@ int hns_roce_create_srq(struct ib_srq *ib_srq,
 	mutex_init(&srq->mutex);
 	spin_lock_init(&srq->lock);
 
-	srq->wqe_cnt = roundup_pow_of_two(init_attr->attr.max_wr + 1);
+	init_attr->attr.max_wr = max_t(u32, init_attr->attr.max_wr,
+				       HNS_ROCE_MIN_SRQ_WQE_NUM);
+	srq->wqe_cnt = roundup_pow_of_two(init_attr->attr.max_wr);
 	srq->max_gs =
 		roundup_pow_of_two(init_attr->attr.max_sge + srq->rsv_sge);
 	init_attr->attr.max_wr = srq->wqe_cnt;
-- 
2.8.1


  parent reply	other threads:[~2021-01-30  9:30 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-01-30  8:57 [PATCH for-next 00/12] RDMA/hns: Several fixes and cleanups of RQ/SRQ Weihang Li
2021-01-30  8:57 ` [PATCH for-next 01/12] RDMA/hns: Allocate one more recv SGE for HIP08 Weihang Li
2021-01-30  8:58 ` [PATCH for-next 02/12] RDMA/hns: Bugfix for checking whether the srq is full when post wr Weihang Li
2021-01-30  8:58 ` [PATCH for-next 03/12] RDMA/hns: Force srq_limit to 0 when creating SRQ Weihang Li
2021-01-30  8:58 ` [PATCH for-next 04/12] RDMA/hns: Fixed wrong judgments in the goto branch Weihang Li
2021-01-30  8:58 ` Weihang Li [this message]
2021-01-30  8:58 ` [PATCH for-next 06/12] RDMA/hns: Refactor hns_roce_create_srq() Weihang Li
2021-01-30  8:58 ` [PATCH for-next 07/12] RDMA/hns: Refactor code about SRQ Context Weihang Li
2021-01-30  8:58 ` [PATCH for-next 08/12] RDMA/hns: Use new interfaces to write SRQC Weihang Li
2021-01-30  8:58 ` [PATCH for-next 09/12] RDMA/hns: Refactor post recv flow Weihang Li
2021-01-30  8:58 ` [PATCH for-next 10/12] RDMA/hns: Clear remaining unused sges when post_recv Weihang Li
2021-01-30  8:58 ` [PATCH for-next 11/12] RDMA/hns: Refactor hns_roce_v2_post_srq_recv() Weihang Li
2021-01-30  8:58 ` [PATCH for-next 12/12] RDMA/hns: Add verification of QP type when post_recv Weihang Li
2021-02-08 23:43 ` [PATCH for-next 00/12] RDMA/hns: Several fixes and cleanups of RQ/SRQ Jason Gunthorpe

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=1611997090-48820-6-git-send-email-liweihang@huawei.com \
    --to=liweihang@huawei.com \
    --cc=dledford@redhat.com \
    --cc=jgg@nvidia.com \
    --cc=leon@kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=linuxarm@openeuler.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.