2020-07-21 11:07:41

by Huazhong Tan

[permalink] [raw]
Subject: [PATCH net 0/4] net: hns3: fixes for -net

There are some bugfixes for the HNS3 ethernet driver.

Jian Shen (1):
net: hns3: fix return value error when query MAC link status fail

Yunsheng Lin (3):
net: hns3: fix for not unmapping TX buffer correctly
net: hns3: fix for not calculating TX BD send size correctly
net: hns3: fix error handling for desc filling

drivers/net/ethernet/hisilicon/hns3/hnae3.h | 1 +
drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 24 +++++------
drivers/net/ethernet/hisilicon/hns3/hns3_enet.h | 2 -
.../ethernet/hisilicon/hns3/hns3pf/hclge_main.c | 49 ++++++++++------------
.../ethernet/hisilicon/hns3/hns3pf/hclge_main.h | 3 ++
5 files changed, 38 insertions(+), 41 deletions(-)

--
2.7.4


2020-07-21 11:09:39

by Huazhong Tan

[permalink] [raw]
Subject: [PATCH net 3/4] net: hns3: fix error handling for desc filling

From: Yunsheng Lin <[email protected]>

The content of the TX desc is automatically cleared by the HW
when the HW has sent out the packet to the wire. When desc filling
fails in hns3_nic_net_xmit(), it will call hns3_clear_desc() to do
the error handling, which miss zeroing of the TX desc and the
checking if a unmapping is needed.

So add the zeroing and checking in hns3_clear_desc() to avoid the
above problem. Also add DESC_TYPE_UNKNOWN to indicate the info in
desc_cb is not valid, because hns3_nic_reclaim_desc() may treat
the desc_cb->type of zero as packet and add to the sent pkt
statistics accordingly.

Fixes: 76ad4f0ee747 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08 SoC")
Signed-off-by: Yunsheng Lin <[email protected]>
Signed-off-by: Huazhong Tan <[email protected]>
---
drivers/net/ethernet/hisilicon/hns3/hnae3.h | 1 +
drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 8 ++++++++
2 files changed, 9 insertions(+)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hnae3.h b/drivers/net/ethernet/hisilicon/hns3/hnae3.h
index d041cac..088550d 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hnae3.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hnae3.h
@@ -77,6 +77,7 @@
((ring)->p = ((ring)->p - 1 + (ring)->desc_num) % (ring)->desc_num)

enum hns_desc_type {
+ DESC_TYPE_UNKNOWN,
DESC_TYPE_SKB,
DESC_TYPE_FRAGLIST_SKB,
DESC_TYPE_PAGE,
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index a814d99..b1bea030 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -1338,6 +1338,10 @@ static void hns3_clear_desc(struct hns3_enet_ring *ring, int next_to_use_orig)
unsigned int i;

for (i = 0; i < ring->desc_num; i++) {
+ struct hns3_desc *desc = &ring->desc[ring->next_to_use];
+
+ memset(desc, 0, sizeof(*desc));
+
/* check if this is where we started */
if (ring->next_to_use == next_to_use_orig)
break;
@@ -1345,6 +1349,9 @@ static void hns3_clear_desc(struct hns3_enet_ring *ring, int next_to_use_orig)
/* rollback one */
ring_ptr_move_bw(ring, next_to_use);

+ if (!ring->desc_cb[ring->next_to_use].dma)
+ continue;
+
/* unmap the descriptor dma address */
if (ring->desc_cb[ring->next_to_use].type == DESC_TYPE_SKB ||
ring->desc_cb[ring->next_to_use].type ==
@@ -1361,6 +1368,7 @@ static void hns3_clear_desc(struct hns3_enet_ring *ring, int next_to_use_orig)

ring->desc_cb[ring->next_to_use].length = 0;
ring->desc_cb[ring->next_to_use].dma = 0;
+ ring->desc_cb[ring->next_to_use].type = DESC_TYPE_UNKNOWN;
}
}

--
2.7.4

2020-07-21 11:09:56

by Huazhong Tan

[permalink] [raw]
Subject: [PATCH net 2/4] net: hns3: fix for not calculating TX BD send size correctly

From: Yunsheng Lin <[email protected]>

With GRO and fraglist support, the SKB can be aggregated to
a total size of 65535, and when that SKB is forwarded through
a bridge, the size of the SKB may be pushed to exceed the size
of 65535 when br_dev_queue_push_xmit() is called.

The max send size of BD supported by the HW is 65535, when a SKB
with a headlen of over 65535 is sent to the driver, the driver
needs to use multi BD to send the linear data, and the send size
of the last BD is calculated incorrectly by the driver who is
using '&' operation, which causes a TX error.

Use '%' operation to fix this problem.

Fixes: 3fe13ed95dd3 ("net: hns3: avoid mult + div op in critical data path")
Signed-off-by: Yunsheng Lin <[email protected]>
Signed-off-by: Huazhong Tan <[email protected]>
---
drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 2 +-
drivers/net/ethernet/hisilicon/hns3/hns3_enet.h | 2 --
2 files changed, 1 insertion(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 18f7623..a814d99 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -1135,7 +1135,7 @@ static int hns3_fill_desc(struct hns3_enet_ring *ring, void *priv,
}

frag_buf_num = hns3_tx_bd_count(size);
- sizeoflast = size & HNS3_TX_LAST_SIZE_M;
+ sizeoflast = size % HNS3_MAX_BD_SIZE;
sizeoflast = sizeoflast ? sizeoflast : HNS3_MAX_BD_SIZE;

/* When frag size is bigger than hardware limit, split this frag */
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
index 9f64077..9922c5f 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
@@ -165,8 +165,6 @@ enum hns3_nic_state {
#define HNS3_TXD_MSS_S 0
#define HNS3_TXD_MSS_M (0x3fff << HNS3_TXD_MSS_S)

-#define HNS3_TX_LAST_SIZE_M 0xffff
-
#define HNS3_VECTOR_TX_IRQ BIT_ULL(0)
#define HNS3_VECTOR_RX_IRQ BIT_ULL(1)

--
2.7.4

2020-07-21 11:10:16

by Huazhong Tan

[permalink] [raw]
Subject: [PATCH net 1/4] net: hns3: fix for not unmapping TX buffer correctly

From: Yunsheng Lin <[email protected]>

When a big TX buffer is sent using multi BD, the driver maps the
whole TX buffer, and unmaps it using info in desc_cb corresponding
to each BD, but only the info in the desc_cb of first BD is correct,
other info in desc_cb is wrong, which causes TX unmapping problem
when SMMU is on.

Only set the mapping and freeing info in the desc_cb of first BD to
fix this problem, because the TX buffer only need to be unmapped and
freed once.

Fixes: 1e8a7977d09f("net: hns3: add handling for big TX fragment")
Signed-off-by: Yunsheng Lin <[email protected]>
Signed-off-by: Huzhong Tan <[email protected]>
---
drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 14 +++-----------
1 file changed, 3 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index fe7d57ae..18f7623 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -1118,12 +1118,12 @@ static int hns3_fill_desc(struct hns3_enet_ring *ring, void *priv,
return -ENOMEM;
}

+ desc_cb->priv = priv;
desc_cb->length = size;
+ desc_cb->dma = dma;
+ desc_cb->type = type;

if (likely(size <= HNS3_MAX_BD_SIZE)) {
- desc_cb->priv = priv;
- desc_cb->dma = dma;
- desc_cb->type = type;
desc->addr = cpu_to_le64(dma);
desc->tx.send_size = cpu_to_le16(size);
desc->tx.bdtp_fe_sc_vld_ra_ri =
@@ -1140,13 +1140,6 @@ static int hns3_fill_desc(struct hns3_enet_ring *ring, void *priv,

/* When frag size is bigger than hardware limit, split this frag */
for (k = 0; k < frag_buf_num; k++) {
- /* The txbd's baseinfo of DESC_TYPE_PAGE & DESC_TYPE_SKB */
- desc_cb->priv = priv;
- desc_cb->dma = dma + HNS3_MAX_BD_SIZE * k;
- desc_cb->type = ((type == DESC_TYPE_FRAGLIST_SKB ||
- type == DESC_TYPE_SKB) && !k) ?
- type : DESC_TYPE_PAGE;
-
/* now, fill the descriptor */
desc->addr = cpu_to_le64(dma + HNS3_MAX_BD_SIZE * k);
desc->tx.send_size = cpu_to_le16((k == frag_buf_num - 1) ?
@@ -1158,7 +1151,6 @@ static int hns3_fill_desc(struct hns3_enet_ring *ring, void *priv,
/* move ring pointer to next */
ring_ptr_move_fw(ring, next_to_use);

- desc_cb = &ring->desc_cb[ring->next_to_use];
desc = &ring->desc[ring->next_to_use];
}

--
2.7.4

2020-07-21 11:10:28

by Huazhong Tan

[permalink] [raw]
Subject: [PATCH net 4/4] net: hns3: fix return value error when query MAC link status fail

From: Jian Shen <[email protected]>

Currently, PF queries the MAC link status per second by calling
function hclge_get_mac_link_status(). It return the error code
when failed to send cmdq command to firmware. It's incorrect,
because this return value is used as the MAC link status, which
0 means link down, and none-zero means link up. So fixes it.

Fixes: 46a3df9f9718 ("net: hns3: Add HNS3 Acceleration Engine & Compatibility Layer Support")
Signed-off-by: Jian Shen <[email protected]>
Signed-off-by: Huazhong tan <[email protected]>
---
.../ethernet/hisilicon/hns3/hns3pf/hclge_main.c | 49 ++++++++++------------
.../ethernet/hisilicon/hns3/hns3pf/hclge_main.h | 3 ++
2 files changed, 25 insertions(+), 27 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index d6bfdc6..bb4a632 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -2673,11 +2673,10 @@ void hclge_task_schedule(struct hclge_dev *hdev, unsigned long delay_time)
delay_time);
}

-static int hclge_get_mac_link_status(struct hclge_dev *hdev)
+static int hclge_get_mac_link_status(struct hclge_dev *hdev, int *link_status)
{
struct hclge_link_status_cmd *req;
struct hclge_desc desc;
- int link_status;
int ret;

hclge_cmd_setup_basic_desc(&desc, HCLGE_OPC_QUERY_LINK_STATUS, true);
@@ -2689,33 +2688,25 @@ static int hclge_get_mac_link_status(struct hclge_dev *hdev)
}

req = (struct hclge_link_status_cmd *)desc.data;
- link_status = req->status & HCLGE_LINK_STATUS_UP_M;
+ *link_status = (req->status & HCLGE_LINK_STATUS_UP_M) > 0 ?
+ HCLGE_LINK_STATUS_UP : HCLGE_LINK_STATUS_DOWN;

- return !!link_status;
+ return 0;
}

-static int hclge_get_mac_phy_link(struct hclge_dev *hdev)
+static int hclge_get_mac_phy_link(struct hclge_dev *hdev, int *link_status)
{
- unsigned int mac_state;
- int link_stat;
+ struct phy_device *phydev = hdev->hw.mac.phydev;
+
+ *link_status = HCLGE_LINK_STATUS_DOWN;

if (test_bit(HCLGE_STATE_DOWN, &hdev->state))
return 0;

- mac_state = hclge_get_mac_link_status(hdev);
-
- if (hdev->hw.mac.phydev) {
- if (hdev->hw.mac.phydev->state == PHY_RUNNING)
- link_stat = mac_state &
- hdev->hw.mac.phydev->link;
- else
- link_stat = 0;
-
- } else {
- link_stat = mac_state;
- }
+ if (phydev && (phydev->state != PHY_RUNNING || !phydev->link))
+ return 0;

- return !!link_stat;
+ return hclge_get_mac_link_status(hdev, link_status);
}

static void hclge_update_link_status(struct hclge_dev *hdev)
@@ -2725,6 +2716,7 @@ static void hclge_update_link_status(struct hclge_dev *hdev)
struct hnae3_handle *rhandle;
struct hnae3_handle *handle;
int state;
+ int ret;
int i;

if (!client)
@@ -2733,7 +2725,12 @@ static void hclge_update_link_status(struct hclge_dev *hdev)
if (test_and_set_bit(HCLGE_STATE_LINK_UPDATING, &hdev->state))
return;

- state = hclge_get_mac_phy_link(hdev);
+ ret = hclge_get_mac_phy_link(hdev, &state);
+ if (ret) {
+ clear_bit(HCLGE_STATE_LINK_UPDATING, &hdev->state);
+ return;
+ }
+
if (state != hdev->hw.mac.link) {
for (i = 0; i < hdev->num_vmdq_vport + 1; i++) {
handle = &hdev->vport[i].nic;
@@ -6524,14 +6521,15 @@ static int hclge_mac_link_status_wait(struct hclge_dev *hdev, int link_ret)
{
#define HCLGE_MAC_LINK_STATUS_NUM 100

+ int link_status;
int i = 0;
int ret;

do {
- ret = hclge_get_mac_link_status(hdev);
- if (ret < 0)
+ ret = hclge_get_mac_link_status(hdev, &link_status);
+ if (ret)
return ret;
- else if (ret == link_ret)
+ if (link_status == link_ret)
return 0;

msleep(HCLGE_LINK_STATUS_MS);
@@ -6542,9 +6540,6 @@ static int hclge_mac_link_status_wait(struct hclge_dev *hdev, int link_ret)
static int hclge_mac_phy_link_status_wait(struct hclge_dev *hdev, bool en,
bool is_phy)
{
-#define HCLGE_LINK_STATUS_DOWN 0
-#define HCLGE_LINK_STATUS_UP 1
-
int link_ret;

link_ret = en ? HCLGE_LINK_STATUS_UP : HCLGE_LINK_STATUS_DOWN;
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
index 46e6e0f..9bbdd45 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
@@ -317,6 +317,9 @@ enum hclge_link_fail_code {
HCLGE_LF_XSFP_ABSENT,
};

+#define HCLGE_LINK_STATUS_DOWN 0
+#define HCLGE_LINK_STATUS_UP 1
+
#define HCLGE_PG_NUM 4
#define HCLGE_SCH_MODE_SP 0
#define HCLGE_SCH_MODE_DWRR 1
--
2.7.4

2020-07-21 22:50:12

by David Miller

[permalink] [raw]
Subject: Re: [PATCH net 0/4] net: hns3: fixes for -net

From: Huazhong Tan <[email protected]>
Date: Tue, 21 Jul 2020 19:03:50 +0800

> There are some bugfixes for the HNS3 ethernet driver.

Series applied, thank you.