Blob Blame History Raw
From: Yangyang Li <liyangyang20@huawei.com>
Date: Fri, 9 Aug 2019 17:41:01 +0800
Subject: RDMA/hns: Modify pi vlaue when cq overflows
Patch-mainline: v5.4-rc1
Git-commit: d7e5ca88d60c0b5e166e33bfe356e961824ce0f4
References: jsc#SLE-8449

When exiting "for loop", the actual value of pi will be
increased by 1, which is compatible with the next calculation.
But when pi is equal to "ci + hr_cq-> ib_cq.cqe", the "break"
was called and the pi is actual value, it will lead one cqe
still existing, so the "==" should be modify to ">".

Signed-off-by: Yangyang Li <liyangyang20@huawei.com>
Link: https://lore.kernel.org/r/1565343666-73193-5-git-send-email-oulijun@huawei.com
Signed-off-by: Doug Ledford <dledford@redhat.com>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -2407,7 +2407,7 @@ static void __hns_roce_v2_cq_clean(struc
 
 	for (prod_index = hr_cq->cons_index; get_sw_cqe_v2(hr_cq, prod_index);
 	     ++prod_index) {
-		if (prod_index == hr_cq->cons_index + hr_cq->ib_cq.cqe)
+		if (prod_index > hr_cq->cons_index + hr_cq->ib_cq.cqe)
 			break;
 	}