lkml.org 
[lkml]   [2019]   [Jul]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH V2 net-next 04/11] net: hns3: fix mis-counting IRQ vector numbers issue
Date
From: Yonglong Liu <liuyonglong@huawei.com>

The num_msi_left means the vector numbers of NIC, but if the
PF supported RoCE, it contains the vector numbers of NIC and
RoCE(Not expected).

This may cause interrupts lost in some case, because of the
NIC module used the vector resources which belongs to RoCE.

This patch corrects the value of num_msi_left to be equals to
the vector numbers of NIC, and adjust the default tqp numbers
according to the value of num_msi_left. Also, adds a little
cleanup about checking whether the device is supporting RoCE
before allocating IRQ vector.

Fixes: 46a3df9f9718 ("net: hns3: Add HNS3 Acceleration Engine & Compatibility Layer Support")
Signed-off-by: Yonglong Liu <liuyonglong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c | 2 ++
drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c | 12 ++++++++++--
2 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index 3c64d70..9329aab 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -1476,6 +1476,8 @@ static int hclge_vport_setup(struct hclge_vport *vport, u16 num_tqps)
nic->ae_algo = &ae_algo;
nic->numa_node_mask = hdev->numa_node_mask;

+ num_tqps = min_t(u16, hdev->roce_base_msix_offset - 1, num_tqps);
+
ret = hclge_knic_setup(vport, num_tqps,
hdev->num_tx_desc, hdev->num_rx_desc);
if (ret)
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
index a13a0e1..db84782 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
@@ -287,6 +287,14 @@ static int hclgevf_get_queue_info(struct hclgevf_dev *hdev)
memcpy(&hdev->rss_size_max, &resp_msg[2], sizeof(u16));
memcpy(&hdev->rx_buf_len, &resp_msg[4], sizeof(u16));

+ /* if irq is not enough, let tqps have the same value of irqs,
+ * to make sure one irq just bind to one tqp, this can improve
+ * the performance
+ */
+ hdev->num_tqps = min_t(u16, hdev->roce_base_msix_offset - 1,
+ hdev->num_tqps);
+ hdev->rss_size_max = min_t(u16, hdev->rss_size_max, hdev->num_tqps);
+
return 0;
}

@@ -2208,7 +2216,7 @@ static int hclgevf_init_msi(struct hclgevf_dev *hdev)
int vectors;
int i;

- if (hnae3_get_bit(hdev->ae_dev->flag, HNAE3_DEV_SUPPORT_ROCE_B))
+ if (hnae3_dev_roce_supported(hdev))
vectors = pci_alloc_irq_vectors(pdev,
hdev->roce_base_msix_offset + 1,
hdev->num_msi,
@@ -2495,7 +2503,7 @@ static int hclgevf_query_vf_resource(struct hclgevf_dev *hdev)

req = (struct hclgevf_query_res_cmd *)desc.data;

- if (hnae3_get_bit(hdev->ae_dev->flag, HNAE3_DEV_SUPPORT_ROCE_B)) {
+ if (hnae3_dev_roce_supported(hdev)) {
hdev->roce_base_msix_offset =
hnae3_get_field(__le16_to_cpu(req->msixcap_localid_ba_rocee),
HCLGEVF_MSIX_OFT_ROCEE_M,
--
2.7.4
\
 
 \ /
  Last update: 2019-07-26 05:28    [W:0.087 / U:0.860 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site