upstream/opnsense-src
1
0
Fork 0
mirror of https://github.com/opnsense/src.git synced 2026-06-03 22:02:58 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

ena: Improve reset reason statistics

Browse source 
The driver uses different reset reasons.
Some of them are counted and presented in the driver statistics.
There are cases where statistics are counted on a ring level,
but these are zeroed after a reset procedure takes place.

This commit makes the following changes:
1. Add statistics for the unrepresented reset reasons.
2. Add reset reasons which are counted on a ring level,
to be also global for better tracking.

Approved by: cperciva (mentor)
Sponsored by: Amazon, Inc.

(cherry picked from commit 89ce3f6314f6feba0e6626be51832d44df611218)
This commit is contained in:
Osama Abboud 2024-08-07 06:24:19 +00:00 committed by Osama Abboud
parent b1718de62f
commit 89940eed91
3 changed files with 58 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
sys/dev/ena/ena.c
View file

@ -3014,7 +3014,6 @@ check_for_missing_keep_alive(struct ena_adapter *adapter)
time = getsbinuptime() - timestamp;
if (unlikely(time > adapter->keep_alive_timeout)) {
ena_log(adapter->pdev, ERR, "Keep alive watchdog timeout.\n");
counter_u64_add(adapter->dev_stats.wd_expired, 1);
ena_trigger_reset(adapter, ENA_REGS_RESET_KEEP_ALIVE_TO);
}
}
@ -3026,7 +3025,6 @@ check_for_admin_com_state(struct ena_adapter *adapter)
if (unlikely(ena_com_get_admin_running_state(adapter->ena_dev) == false)) {
ena_log(adapter->pdev, ERR,
"ENA admin queue is not in running state!\n");
counter_u64_add(adapter->dev_stats.admin_q_pause, 1);
ena_trigger_reset(adapter, ENA_REGS_RESET_ADMIN_TO);
}
}

42
sys/dev/ena/ena.h
View file

@ -381,6 +381,14 @@ struct ena_stats_dev {
counter_u64_t interface_up;
counter_u64_t interface_down;
counter_u64_t admin_q_pause;
counter_u64_t total_resets;
counter_u64_t os_trigger;
counter_u64_t missing_tx_cmpl;
counter_u64_t bad_rx_req_id;
counter_u64_t bad_tx_req_id;
counter_u64_t bad_rx_desc_num;
counter_u64_t invalid_state;
counter_u64_t missing_intr;
};
struct ena_hw_stats {
@ -519,6 +527,29 @@ struct ena_adapter {
extern struct sx ena_global_lock;
#define ENA_RESET_STATS_ENTRY(reset_reason, stat) \
[reset_reason] = { \
.stat_offset = offsetof(struct ena_stats_dev, stat) / sizeof(u64), \
.has_counter = true \
}
struct ena_reset_stats_offset {
int stat_offset;
bool has_counter;
};
static const struct ena_reset_stats_offset resets_to_stats_offset_map[ENA_REGS_RESET_LAST] = {
ENA_RESET_STATS_ENTRY(ENA_REGS_RESET_KEEP_ALIVE_TO, wd_expired),
ENA_RESET_STATS_ENTRY(ENA_REGS_RESET_ADMIN_TO, admin_q_pause),
ENA_RESET_STATS_ENTRY(ENA_REGS_RESET_OS_TRIGGER, os_trigger),
ENA_RESET_STATS_ENTRY(ENA_REGS_RESET_MISS_TX_CMPL, missing_tx_cmpl),
ENA_RESET_STATS_ENTRY(ENA_REGS_RESET_INV_RX_REQ_ID, bad_rx_req_id),
ENA_RESET_STATS_ENTRY(ENA_REGS_RESET_INV_TX_REQ_ID, bad_tx_req_id),
ENA_RESET_STATS_ENTRY(ENA_REGS_RESET_TOO_MANY_RX_DESCS, bad_rx_desc_num),
ENA_RESET_STATS_ENTRY(ENA_REGS_RESET_DRIVER_INVALID_STATE, invalid_state),
ENA_RESET_STATS_ENTRY(ENA_REGS_RESET_MISS_INTERRUPT, missing_intr),
};
int ena_up(struct ena_adapter *adapter);
void ena_down(struct ena_adapter *adapter);
int ena_restore_device(struct ena_adapter *adapter);
@ -547,6 +578,17 @@ ena_trigger_reset(struct ena_adapter *adapter,
enum ena_regs_reset_reason_types reset_reason)
{
if (likely(!ENA_FLAG_ISSET(ENA_FLAG_TRIGGER_RESET, adapter))) {
const struct ena_reset_stats_offset *ena_reset_stats_offset =
&resets_to_stats_offset_map[reset_reason];
if (ena_reset_stats_offset->has_counter) {
uint64_t *stat_ptr = (uint64_t *)&adapter->dev_stats +
ena_reset_stats_offset->stat_offset;
counter_u64_add((counter_u64_t)(*stat_ptr), 1);
}
counter_u64_add(adapter->dev_stats.total_resets, 1);
adapter->reset_reason = reset_reason;
ENA_FLAG_SET_ATOMIC(ENA_FLAG_TRIGGER_RESET, adapter);
}

16
sys/dev/ena/ena_sysctl.c
View file

@ -280,6 +280,22 @@ ena_sysctl_add_stats(struct ena_adapter *adapter)
"Network interface down count");
SYSCTL_ADD_COUNTER_U64(ctx, child, OID_AUTO, "admin_q_pause",
CTLFLAG_RD, &dev_stats->admin_q_pause, "Admin queue pauses");
SYSCTL_ADD_COUNTER_U64(ctx, child, OID_AUTO, "os_trigger", CTLFLAG_RD,
&dev_stats->os_trigger, "OS trigger count");
SYSCTL_ADD_COUNTER_U64(ctx, child, OID_AUTO, "missing_tx_cmpl", CTLFLAG_RD,
&dev_stats->missing_tx_cmpl, "Missing TX completions resets count");
SYSCTL_ADD_COUNTER_U64(ctx, child, OID_AUTO, "bad_rx_req_id", CTLFLAG_RD,
&dev_stats->bad_rx_req_id, "Bad RX req id count");
SYSCTL_ADD_COUNTER_U64(ctx, child, OID_AUTO, "bad_tx_req_id", CTLFLAG_RD,
&dev_stats->bad_tx_req_id, "Bad TX req id count");
SYSCTL_ADD_COUNTER_U64(ctx, child, OID_AUTO, "bad_rx_desc_num", CTLFLAG_RD,
&dev_stats->bad_rx_desc_num, "Bad RX descriptors number count");
SYSCTL_ADD_COUNTER_U64(ctx, child, OID_AUTO, "invalid_state", CTLFLAG_RD,
&dev_stats->invalid_state, "Driver invalid state count");
SYSCTL_ADD_COUNTER_U64(ctx, child, OID_AUTO, "missing_intr", CTLFLAG_RD,
&dev_stats->missing_intr, "Missing interrupt count");
SYSCTL_ADD_COUNTER_U64(ctx, child, OID_AUTO, "total_resets", CTLFLAG_RD,
&dev_stats->total_resets, "Total resets count");
for (i = 0; i < adapter->num_io_queues; ++i, ++tx_ring, ++rx_ring) {
snprintf(namebuf, QUEUE_NAME_LEN, "queue%d", i);