From: Huazhong Tan <tanhuazhong@huawei.com>
To: <davem@davemloft.net>
Cc: <netdev@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
<salil.mehta@huawei.com>, <yisen.zhuang@huawei.com>,
<linuxarm@huawei.com>, <jakub.kicinski@netronome.com>,
Yunsheng Lin <linyunsheng@huawei.com>,
Huazhong Tan <tanhuazhong@huawei.com>
Subject: [PATCH net-next 2/5] net: hns3: remove mailbox and reset work in hclge_main
Date: Sat, 14 Dec 2019 10:06:38 +0800 [thread overview]
Message-ID: <1576289201-57017-3-git-send-email-tanhuazhong@huawei.com> (raw)
In-Reply-To: <1576289201-57017-1-git-send-email-tanhuazhong@huawei.com>
From: Yunsheng Lin <linyunsheng@huawei.com>
There are three work (mbx_service_task, service_task,
rst_service_task) in the HNS3 driver, mbx_service_task is for
handling mailbox work, service_task is for periodic management
issue and rst_service_task is for reset related issue, which can
be handled in a single work.
This patch removes the mbx_service_task and rst_service_task
work, and moves the related handling to the service_task work
in order to remove concurrency between the three work and to
improve efficiency.
BTW, since stats_timer in struct hclge_hw_stats is not needed
anymore, so removes the definition of struct hclge_hw_stats,
and moves mac_stats into struct hclge_dev.
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
.../ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c | 10 ++
.../ethernet/hisilicon/hns3/hns3pf/hclge_main.c | 163 +++++++++++++--------
.../ethernet/hisilicon/hns3/hns3pf/hclge_main.h | 17 +--
.../net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c | 1 -
4 files changed, 118 insertions(+), 73 deletions(-)
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
index 112df34..92a99e4 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_debugfs.c
@@ -976,6 +976,14 @@ void hclge_dbg_dump_rst_info(struct hclge_dev *hdev)
dev_info(&hdev->pdev->dev, "hdev state: 0x%lx\n", hdev->state);
}
+static void hclge_dbg_dump_serv_info(struct hclge_dev *hdev)
+{
+ dev_info(&hdev->pdev->dev, "last_serv_processed: %lu\n",
+ hdev->last_serv_processed);
+ dev_info(&hdev->pdev->dev, "last_serv_cnt: %lu\n",
+ hdev->serv_processed_cnt);
+}
+
static void hclge_dbg_get_m7_stats_info(struct hclge_dev *hdev)
{
struct hclge_desc *desc_src, *desc_tmp;
@@ -1227,6 +1235,8 @@ int hclge_dbg_run_cmd(struct hnae3_handle *handle, const char *cmd_buf)
hclge_dbg_dump_reg_cmd(hdev, &cmd_buf[sizeof(DUMP_REG)]);
} else if (strncmp(cmd_buf, "dump reset info", 15) == 0) {
hclge_dbg_dump_rst_info(hdev);
+ } else if (strncmp(cmd_buf, "dump serv info", 14) == 0) {
+ hclge_dbg_dump_serv_info(hdev);
} else if (strncmp(cmd_buf, "dump m7 info", 12) == 0) {
hclge_dbg_get_m7_stats_info(hdev);
} else if (strncmp(cmd_buf, "dump ncl_config", 15) == 0) {
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index d862e9b..a0177e3 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -416,7 +416,7 @@ static int hclge_mac_update_stats_defective(struct hclge_dev *hdev)
{
#define HCLGE_MAC_CMD_NUM 21
- u64 *data = (u64 *)(&hdev->hw_stats.mac_stats);
+ u64 *data = (u64 *)(&hdev->mac_stats);
struct hclge_desc desc[HCLGE_MAC_CMD_NUM];
__le64 *desc_data;
int i, k, n;
@@ -453,7 +453,7 @@ static int hclge_mac_update_stats_defective(struct hclge_dev *hdev)
static int hclge_mac_update_stats_complete(struct hclge_dev *hdev, u32 desc_num)
{
- u64 *data = (u64 *)(&hdev->hw_stats.mac_stats);
+ u64 *data = (u64 *)(&hdev->mac_stats);
struct hclge_desc *desc;
__le64 *desc_data;
u16 i, k, n;
@@ -802,7 +802,7 @@ static void hclge_get_stats(struct hnae3_handle *handle, u64 *data)
struct hclge_dev *hdev = vport->back;
u64 *p;
- p = hclge_comm_get_stats(&hdev->hw_stats.mac_stats, g_mac_stats_string,
+ p = hclge_comm_get_stats(&hdev->mac_stats, g_mac_stats_string,
ARRAY_SIZE(g_mac_stats_string), data);
p = hclge_tqps_get_stats(handle, p);
}
@@ -815,8 +815,8 @@ static void hclge_get_mac_stat(struct hnae3_handle *handle,
hclge_update_stats(handle, NULL);
- mac_stats->tx_pause_cnt = hdev->hw_stats.mac_stats.mac_tx_mac_pause_num;
- mac_stats->rx_pause_cnt = hdev->hw_stats.mac_stats.mac_rx_mac_pause_num;
+ mac_stats->tx_pause_cnt = hdev->mac_stats.mac_tx_mac_pause_num;
+ mac_stats->rx_pause_cnt = hdev->mac_stats.mac_rx_mac_pause_num;
}
static int hclge_parse_func_status(struct hclge_dev *hdev,
@@ -2665,31 +2665,26 @@ static int hclge_mac_init(struct hclge_dev *hdev)
static void hclge_mbx_task_schedule(struct hclge_dev *hdev)
{
- if (!test_bit(HCLGE_STATE_CMD_DISABLE, &hdev->state) &&
+ if (!test_bit(HCLGE_STATE_REMOVING, &hdev->state) &&
!test_and_set_bit(HCLGE_STATE_MBX_SERVICE_SCHED, &hdev->state))
- queue_work_on(cpumask_first(&hdev->affinity_mask), system_wq,
- &hdev->mbx_service_task);
+ mod_delayed_work_on(cpumask_first(&hdev->affinity_mask),
+ system_wq, &hdev->service_task, 0);
}
static void hclge_reset_task_schedule(struct hclge_dev *hdev)
{
if (!test_bit(HCLGE_STATE_REMOVING, &hdev->state) &&
!test_and_set_bit(HCLGE_STATE_RST_SERVICE_SCHED, &hdev->state))
- queue_work_on(cpumask_first(&hdev->affinity_mask), system_wq,
- &hdev->rst_service_task);
+ mod_delayed_work_on(cpumask_first(&hdev->affinity_mask),
+ system_wq, &hdev->service_task, 0);
}
void hclge_task_schedule(struct hclge_dev *hdev, unsigned long delay_time)
{
- if (!test_bit(HCLGE_STATE_DOWN, &hdev->state) &&
- !test_bit(HCLGE_STATE_REMOVING, &hdev->state) &&
- !test_and_set_bit(HCLGE_STATE_SERVICE_SCHED, &hdev->state)) {
- hdev->hw_stats.stats_timer++;
- hdev->fd_arfs_expire_timer++;
+ if (!test_bit(HCLGE_STATE_REMOVING, &hdev->state))
mod_delayed_work_on(cpumask_first(&hdev->affinity_mask),
system_wq, &hdev->service_task,
delay_time);
- }
}
static int hclge_get_mac_link_status(struct hclge_dev *hdev)
@@ -2748,6 +2743,10 @@ static void hclge_update_link_status(struct hclge_dev *hdev)
if (!client)
return;
+
+ if (test_and_set_bit(HCLGE_STATE_LINK_UPDATING, &hdev->state))
+ return;
+
state = hclge_get_mac_phy_link(hdev);
if (state != hdev->hw.mac.link) {
for (i = 0; i < hdev->num_vmdq_vport + 1; i++) {
@@ -2761,6 +2760,8 @@ static void hclge_update_link_status(struct hclge_dev *hdev)
}
hdev->hw.mac.link = state;
}
+
+ clear_bit(HCLGE_STATE_LINK_UPDATING, &hdev->state);
}
static void hclge_update_port_capability(struct hclge_mac *mac)
@@ -3352,6 +3353,18 @@ static int hclge_set_all_vf_rst(struct hclge_dev *hdev, bool reset)
return 0;
}
+static void hclge_mailbox_service_task(struct hclge_dev *hdev)
+{
+ if (!test_and_clear_bit(HCLGE_STATE_MBX_SERVICE_SCHED, &hdev->state) ||
+ test_bit(HCLGE_STATE_CMD_DISABLE, &hdev->state) ||
+ test_and_set_bit(HCLGE_STATE_MBX_HANDLING, &hdev->state))
+ return;
+
+ hclge_mbx_handler(hdev);
+
+ clear_bit(HCLGE_STATE_MBX_HANDLING, &hdev->state);
+}
+
static int hclge_func_reset_sync_vf(struct hclge_dev *hdev)
{
struct hclge_pf_rst_sync_cmd *req;
@@ -3363,6 +3376,9 @@ static int hclge_func_reset_sync_vf(struct hclge_dev *hdev)
hclge_cmd_setup_basic_desc(&desc, HCLGE_OPC_QUERY_VF_RST_RDY, true);
do {
+ /* vf need to down netdev by mbx during PF or FLR reset */
+ hclge_mailbox_service_task(hdev);
+
ret = hclge_cmd_send(&hdev->hw, &desc, 1);
/* for compatible with old firmware, wait
* 100 ms for VF to stop IO
@@ -3939,36 +3955,19 @@ static void hclge_reset_subtask(struct hclge_dev *hdev)
hdev->reset_type = HNAE3_NONE_RESET;
}
-static void hclge_reset_service_task(struct work_struct *work)
+static void hclge_reset_service_task(struct hclge_dev *hdev)
{
- struct hclge_dev *hdev =
- container_of(work, struct hclge_dev, rst_service_task);
+ if (!test_and_clear_bit(HCLGE_STATE_RST_SERVICE_SCHED, &hdev->state))
+ return;
if (test_and_set_bit(HCLGE_STATE_RST_HANDLING, &hdev->state))
return;
- clear_bit(HCLGE_STATE_RST_SERVICE_SCHED, &hdev->state);
-
hclge_reset_subtask(hdev);
clear_bit(HCLGE_STATE_RST_HANDLING, &hdev->state);
}
-static void hclge_mailbox_service_task(struct work_struct *work)
-{
- struct hclge_dev *hdev =
- container_of(work, struct hclge_dev, mbx_service_task);
-
- if (test_and_set_bit(HCLGE_STATE_MBX_HANDLING, &hdev->state))
- return;
-
- clear_bit(HCLGE_STATE_MBX_SERVICE_SCHED, &hdev->state);
-
- hclge_mbx_handler(hdev);
-
- clear_bit(HCLGE_STATE_MBX_HANDLING, &hdev->state);
-}
-
static void hclge_update_vport_alive(struct hclge_dev *hdev)
{
int i;
@@ -3986,29 +3985,62 @@ static void hclge_update_vport_alive(struct hclge_dev *hdev)
}
}
-static void hclge_service_task(struct work_struct *work)
+static void hclge_periodic_service_task(struct hclge_dev *hdev)
{
- struct hclge_dev *hdev =
- container_of(work, struct hclge_dev, service_task.work);
+ unsigned long delta = round_jiffies_relative(HZ);
- clear_bit(HCLGE_STATE_SERVICE_SCHED, &hdev->state);
+ /* Always handle the link updating to make sure link state is
+ * updated when it is triggered by mbx.
+ */
+ hclge_update_link_status(hdev);
- if (hdev->hw_stats.stats_timer >= HCLGE_STATS_TIMER_INTERVAL) {
- hclge_update_stats_for_all(hdev);
- hdev->hw_stats.stats_timer = 0;
+ if (time_is_after_jiffies(hdev->last_serv_processed + HZ)) {
+ delta = jiffies - hdev->last_serv_processed;
+
+ if (delta < round_jiffies_relative(HZ)) {
+ delta = round_jiffies_relative(HZ) - delta;
+ goto out;
+ }
}
- hclge_update_port_info(hdev);
- hclge_update_link_status(hdev);
+ hdev->serv_processed_cnt++;
hclge_update_vport_alive(hdev);
+
+ if (test_bit(HCLGE_STATE_DOWN, &hdev->state)) {
+ hdev->last_serv_processed = jiffies;
+ goto out;
+ }
+
+ if (!(hdev->serv_processed_cnt % HCLGE_STATS_TIMER_INTERVAL))
+ hclge_update_stats_for_all(hdev);
+
+ hclge_update_port_info(hdev);
hclge_sync_vlan_filter(hdev);
- if (hdev->fd_arfs_expire_timer >= HCLGE_FD_ARFS_EXPIRE_TIMER_INTERVAL) {
+ if (!(hdev->serv_processed_cnt % HCLGE_ARFS_EXPIRE_INTERVAL))
hclge_rfs_filter_expire(hdev);
- hdev->fd_arfs_expire_timer = 0;
- }
- hclge_task_schedule(hdev, round_jiffies_relative(HZ));
+ hdev->last_serv_processed = jiffies;
+
+out:
+ hclge_task_schedule(hdev, delta);
+}
+
+static void hclge_service_task(struct work_struct *work)
+{
+ struct hclge_dev *hdev =
+ container_of(work, struct hclge_dev, service_task.work);
+
+ hclge_reset_service_task(hdev);
+ hclge_mailbox_service_task(hdev);
+ hclge_periodic_service_task(hdev);
+
+ /* Handle reset and mbx again in case periodical task delays the
+ * handling by calling hclge_task_schedule() in
+ * hclge_periodic_service_task().
+ */
+ hclge_reset_service_task(hdev);
+ hclge_mailbox_service_task(hdev);
}
struct hclge_vport *hclge_get_vport(struct hnae3_handle *handle)
@@ -6734,6 +6766,19 @@ static void hclge_reset_tqp_stats(struct hnae3_handle *handle)
}
}
+static void hclge_flush_link_update(struct hclge_dev *hdev)
+{
+#define HCLGE_FLUSH_LINK_TIMEOUT 100000
+
+ unsigned long last = hdev->serv_processed_cnt;
+ int i = 0;
+
+ while (test_bit(HCLGE_STATE_LINK_UPDATING, &hdev->state) &&
+ i++ < HCLGE_FLUSH_LINK_TIMEOUT &&
+ last == hdev->serv_processed_cnt)
+ usleep_range(1, 1);
+}
+
static void hclge_set_timer_task(struct hnae3_handle *handle, bool enable)
{
struct hclge_vport *vport = hclge_get_vport(handle);
@@ -6742,12 +6787,12 @@ static void hclge_set_timer_task(struct hnae3_handle *handle, bool enable)
if (enable) {
hclge_task_schedule(hdev, round_jiffies_relative(HZ));
} else {
- /* Set the DOWN flag here to disable the service to be
- * scheduled again
- */
+ /* Set the DOWN flag here to disable link updating */
set_bit(HCLGE_STATE_DOWN, &hdev->state);
- cancel_delayed_work_sync(&hdev->service_task);
- clear_bit(HCLGE_STATE_SERVICE_SCHED, &hdev->state);
+
+ /* flush memory to make sure DOWN is seen by service task */
+ smp_mb__before_atomic();
+ hclge_flush_link_update(hdev);
}
}
@@ -9269,10 +9314,6 @@ static void hclge_state_uninit(struct hclge_dev *hdev)
del_timer_sync(&hdev->reset_timer);
if (hdev->service_task.work.func)
cancel_delayed_work_sync(&hdev->service_task);
- if (hdev->rst_service_task.func)
- cancel_work_sync(&hdev->rst_service_task);
- if (hdev->mbx_service_task.func)
- cancel_work_sync(&hdev->mbx_service_task);
}
static void hclge_flr_prepare(struct hnae3_ae_dev *ae_dev)
@@ -9477,8 +9518,6 @@ static int hclge_init_ae_dev(struct hnae3_ae_dev *ae_dev)
timer_setup(&hdev->reset_timer, hclge_reset_timer, 0);
INIT_DELAYED_WORK(&hdev->service_task, hclge_service_task);
- INIT_WORK(&hdev->rst_service_task, hclge_reset_service_task);
- INIT_WORK(&hdev->mbx_service_task, hclge_mailbox_service_task);
/* Setup affinity after service timer setup because add_timer_on
* is called in affinity notify.
@@ -9512,6 +9551,8 @@ static int hclge_init_ae_dev(struct hnae3_ae_dev *ae_dev)
dev_info(&hdev->pdev->dev, "%s driver initialization finished.\n",
HCLGE_DRIVER_NAME);
+ hclge_task_schedule(hdev, round_jiffies_relative(HZ));
+
return 0;
err_mdiobus_unreg:
@@ -9534,7 +9575,7 @@ static int hclge_init_ae_dev(struct hnae3_ae_dev *ae_dev)
static void hclge_stats_clear(struct hclge_dev *hdev)
{
- memset(&hdev->hw_stats, 0, sizeof(hdev->hw_stats));
+ memset(&hdev->mac_stats, 0, sizeof(hdev->mac_stats));
}
static int hclge_set_mac_spoofchk(struct hclge_dev *hdev, int vf, bool enable)
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
index ebb4c6e..ad40cf6 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
@@ -208,13 +208,13 @@ enum HCLGE_DEV_STATE {
HCLGE_STATE_NIC_REGISTERED,
HCLGE_STATE_ROCE_REGISTERED,
HCLGE_STATE_SERVICE_INITED,
- HCLGE_STATE_SERVICE_SCHED,
HCLGE_STATE_RST_SERVICE_SCHED,
HCLGE_STATE_RST_HANDLING,
HCLGE_STATE_MBX_SERVICE_SCHED,
HCLGE_STATE_MBX_HANDLING,
HCLGE_STATE_STATISTICS_UPDATING,
HCLGE_STATE_CMD_DISABLE,
+ HCLGE_STATE_LINK_UPDATING,
HCLGE_STATE_MAX
};
@@ -454,11 +454,7 @@ struct hclge_mac_stats {
u64 mac_rx_ctrl_pkt_num;
};
-#define HCLGE_STATS_TIMER_INTERVAL (60 * 5)
-struct hclge_hw_stats {
- struct hclge_mac_stats mac_stats;
- u32 stats_timer;
-};
+#define HCLGE_STATS_TIMER_INTERVAL 300UL
struct hclge_vlan_type_cfg {
u16 rx_ot_fst_vlan_type;
@@ -549,7 +545,7 @@ struct key_info {
/* assigned by firmware, the real filter number for each pf may be less */
#define MAX_FD_FILTER_NUM 4096
-#define HCLGE_FD_ARFS_EXPIRE_TIMER_INTERVAL 5
+#define HCLGE_ARFS_EXPIRE_INTERVAL 5UL
enum HCLGE_FD_ACTIVE_RULE_TYPE {
HCLGE_FD_RULE_NONE,
@@ -712,7 +708,7 @@ struct hclge_dev {
struct hnae3_ae_dev *ae_dev;
struct hclge_hw hw;
struct hclge_misc_vector misc_vector;
- struct hclge_hw_stats hw_stats;
+ struct hclge_mac_stats mac_stats;
unsigned long state;
unsigned long flr_state;
unsigned long last_reset_time;
@@ -774,8 +770,6 @@ struct hclge_dev {
unsigned long service_timer_previous;
struct timer_list reset_timer;
struct delayed_work service_task;
- struct work_struct rst_service_task;
- struct work_struct mbx_service_task;
bool cur_promisc;
int num_alloc_vfs; /* Actual number of VFs allocated */
@@ -811,7 +805,8 @@ struct hclge_dev {
struct hlist_head fd_rule_list;
spinlock_t fd_rule_lock; /* protect fd_rule_list and fd_bmap */
u16 hclge_fd_rule_num;
- u16 fd_arfs_expire_timer;
+ unsigned long serv_processed_cnt;
+ unsigned long last_serv_processed;
unsigned long fd_bmap[BITS_TO_LONGS(MAX_FD_FILTER_NUM)];
enum HCLGE_FD_ACTIVE_RULE_TYPE fd_active_type;
u8 fd_en;
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c
index 0b433eb..a79d151 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c
@@ -635,7 +635,6 @@ static void hclge_handle_link_change_event(struct hclge_dev *hdev,
#define LINK_STATUS_OFFSET 1
#define LINK_FAIL_CODE_OFFSET 2
- clear_bit(HCLGE_STATE_SERVICE_SCHED, &hdev->state);
hclge_task_schedule(hdev, 0);
if (!req->msg[LINK_STATUS_OFFSET])
--
2.7.4
next prev parent reply other threads:[~2019-12-14 2:07 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-12-14 2:06 [PATCH net-next 0/5] net: hns3: some optimizaions related to work task Huazhong Tan
2019-12-14 2:06 ` [PATCH net-next 1/5] net: hns3: schedule hclgevf_service by using delayed workqueue Huazhong Tan
2019-12-14 2:06 ` Huazhong Tan [this message]
2019-12-14 2:06 ` [PATCH net-next 3/5] net: hns3: remove unnecessary work in hclgevf_main Huazhong Tan
2019-12-14 2:06 ` [PATCH net-next 4/5] net: hns3: allocate WQ with WQ_MEM_RECLAIM flag Huazhong Tan
2019-12-14 2:06 ` [PATCH net-next 5/5] net: hns3: do not schedule the periodic task when reset fail Huazhong Tan
2019-12-17 0:12 ` [PATCH net-next 0/5] net: hns3: some optimizaions related to work task David Miller
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=1576289201-57017-3-git-send-email-tanhuazhong@huawei.com \
--to=tanhuazhong@huawei.com \
--cc=davem@davemloft.net \
--cc=jakub.kicinski@netronome.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxarm@huawei.com \
--cc=linyunsheng@huawei.com \
--cc=netdev@vger.kernel.org \
--cc=salil.mehta@huawei.com \
--cc=yisen.zhuang@huawei.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;
as well as URLs for NNTP newsgroup(s).