[PATCH net-next v2 9/9] octeon_ep: add heartbeat monitor

From: Veerasenareddy Burru
Date: Tue Nov 29 2022 - 08:10:14 EST


Monitor periodic heartbeat messages from device firmware.
Presence of heartbeat indicates the device is active and running.
If the heartbeat is missed for configured interval indicates
firmware has crashed and device is unusable; in this case, PF driver
stops and uninitialize the device.

Signed-off-by: Veerasenareddy Burru <vburru@xxxxxxxxxxx>
Signed-off-by: Abhijit Ayarekar <aayarekar@xxxxxxxxxxx>
---
v1 -> v2:
* reworked the patch for changes made to preceding patches.
* removed device status oct->status, as it is not required with the
modified implementation.

.../marvell/octeon_ep/octep_cn9k_pf.c | 9 +++++
.../ethernet/marvell/octeon_ep/octep_config.h | 6 +++
.../ethernet/marvell/octeon_ep/octep_main.c | 37 +++++++++++++++++++
.../ethernet/marvell/octeon_ep/octep_main.h | 7 ++++
.../marvell/octeon_ep/octep_regs_cn9k_pf.h | 2 +
5 files changed, 61 insertions(+)

diff --git a/drivers/net/ethernet/marvell/octeon_ep/octep_cn9k_pf.c b/drivers/net/ethernet/marvell/octeon_ep/octep_cn9k_pf.c
index 4840133477dc..9c6b2a95bc18 100644
--- a/drivers/net/ethernet/marvell/octeon_ep/octep_cn9k_pf.c
+++ b/drivers/net/ethernet/marvell/octeon_ep/octep_cn9k_pf.c
@@ -16,6 +16,9 @@
#define CTRL_MBOX_MAX_PF 128
#define CTRL_MBOX_SZ ((size_t)(0x400000 / CTRL_MBOX_MAX_PF))

+#define FW_HB_INTERVAL_IN_SECS 1
+#define FW_HB_MISS_COUNT 10
+
/* Names of Hardware non-queue generic interrupts */
static char *cn93_non_ioq_msix_names[] = {
"epf_ire_rint",
@@ -249,6 +252,10 @@ static void octep_init_config_cn93_pf(struct octep_device *oct)
conf->ctrl_mbox_cfg.barmem_addr = (void __iomem *)oct->mmio[2].hw_addr +
(0x400000ull * 8) +
(link * CTRL_MBOX_SZ);
+
+ conf->hb_interval = FW_HB_INTERVAL_IN_SECS;
+ conf->max_hb_miss_cnt = FW_HB_MISS_COUNT;
+
}

/* Setup registers for a hardware Tx Queue */
@@ -407,6 +414,8 @@ static int octep_poll_non_ioq_interrupts_cn93_pf(struct octep_device *oct)
octep_write_csr64(oct, CN93_SDP_EPF_OEI_RINT, reg0);
if (reg0 & CN93_SDP_EPF_OEI_RINT_DATA_BIT_MBOX)
queue_work(octep_wq, &oct->ctrl_mbox_task);
+ else if (reg0 & CN93_SDP_EPF_OEI_RINT_DATA_BIT_HBEAT)
+ atomic_set(&oct->hb_miss_cnt, 0);

handled = 1;
}
diff --git a/drivers/net/ethernet/marvell/octeon_ep/octep_config.h b/drivers/net/ethernet/marvell/octeon_ep/octep_config.h
index f208f3f9a447..df7cd39d9fce 100644
--- a/drivers/net/ethernet/marvell/octeon_ep/octep_config.h
+++ b/drivers/net/ethernet/marvell/octeon_ep/octep_config.h
@@ -200,5 +200,11 @@ struct octep_config {

/* ctrl mbox config */
struct octep_ctrl_mbox_config ctrl_mbox_cfg;
+
+ /* Configured maximum heartbeat miss count */
+ u32 max_hb_miss_cnt;
+
+ /* Configured firmware heartbeat interval in secs */
+ u32 hb_interval;
};
#endif /* _OCTEP_CONFIG_H_ */
diff --git a/drivers/net/ethernet/marvell/octeon_ep/octep_main.c b/drivers/net/ethernet/marvell/octeon_ep/octep_main.c
index cd0d77ceb868..751a3a9c576f 100644
--- a/drivers/net/ethernet/marvell/octeon_ep/octep_main.c
+++ b/drivers/net/ethernet/marvell/octeon_ep/octep_main.c
@@ -903,6 +903,38 @@ static void octep_intr_poll_task(struct work_struct *work)
msecs_to_jiffies(OCTEP_INTR_POLL_TIME_MSECS));
}

+/**
+ * octep_hb_timeout_task - work queue task to check firmware heartbeat.
+ *
+ * @work: pointer to hb work_struct
+ *
+ * Check for heartbeat miss count. Uninitialize oct device if miss count
+ * exceeds configured max heartbeat miss count.
+ *
+ **/
+static void octep_hb_timeout_task(struct work_struct *work)
+{
+ struct octep_device *oct = container_of(work, struct octep_device,
+ hb_task.work);
+
+ int miss_cnt;
+
+ atomic_inc(&oct->hb_miss_cnt);
+ miss_cnt = atomic_read(&oct->hb_miss_cnt);
+ if (miss_cnt < oct->conf->max_hb_miss_cnt) {
+ queue_delayed_work(octep_wq, &oct->hb_task,
+ msecs_to_jiffies(oct->conf->hb_interval * 1000));
+ return;
+ }
+
+ dev_err(&oct->pdev->dev, "Missed %u heartbeats. Uninitializing\n",
+ miss_cnt);
+ rtnl_lock();
+ if (netif_running(oct->netdev))
+ octep_stop(oct->netdev);
+ rtnl_unlock();
+}
+
/**
* octep_ctrl_mbox_task - work queue task to handle ctrl mbox messages.
*
@@ -979,6 +1011,10 @@ int octep_device_setup(struct octep_device *oct)
if (ret)
return ret;

+ atomic_set(&oct->hb_miss_cnt, 0);
+ INIT_DELAYED_WORK(&oct->hb_task, octep_hb_timeout_task);
+ queue_delayed_work(octep_wq, &oct->hb_task,
+ msecs_to_jiffies(oct->conf->hb_interval * 1000));
return 0;

unsupported_dev:
@@ -1004,6 +1040,7 @@ static void octep_device_cleanup(struct octep_device *oct)

octep_delete_pfvf_mbox(oct);
octep_ctrl_net_uninit(oct);
+ cancel_delayed_work_sync(&oct->hb_task);

oct->hw_ops.soft_reset(oct);
for (i = 0; i < OCTEP_MMIO_REGIONS; i++) {
diff --git a/drivers/net/ethernet/marvell/octeon_ep/octep_main.h b/drivers/net/ethernet/marvell/octeon_ep/octep_main.h
index ad6d324bd525..beedf8dc841d 100644
--- a/drivers/net/ethernet/marvell/octeon_ep/octep_main.h
+++ b/drivers/net/ethernet/marvell/octeon_ep/octep_main.h
@@ -286,6 +286,13 @@ struct octep_device {
bool poll_non_ioq_intr;
/* Work entry to poll non-ioq interrupts */
struct delayed_work intr_poll_task;
+
+ /* Firmware heartbeat timer */
+ struct timer_list hb_timer;
+ /* Firmware heartbeat miss count tracked by timer */
+ atomic_t hb_miss_cnt;
+ /* Task to reset device on heartbeat miss */
+ struct delayed_work hb_task;
};

static inline u16 OCTEP_MAJOR_REV(struct octep_device *oct)
diff --git a/drivers/net/ethernet/marvell/octeon_ep/octep_regs_cn9k_pf.h b/drivers/net/ethernet/marvell/octeon_ep/octep_regs_cn9k_pf.h
index f29c4344fc41..48051e23ef18 100644
--- a/drivers/net/ethernet/marvell/octeon_ep/octep_regs_cn9k_pf.h
+++ b/drivers/net/ethernet/marvell/octeon_ep/octep_regs_cn9k_pf.h
@@ -376,5 +376,7 @@

/* bit 0 for control mbox interrupt */
#define CN93_SDP_EPF_OEI_RINT_DATA_BIT_MBOX BIT_ULL(0)
+/* bit 1 for firmware heartbeat interrupt */
+#define CN93_SDP_EPF_OEI_RINT_DATA_BIT_HBEAT BIT_ULL(1)

#endif /* _OCTEP_REGS_CN9K_PF_H_ */
--
2.36.0