[V2,04/14] net/hns3: add Rx and Tx bytes stats
Checks
Commit Message
From: "Min Hu (Connor)" <humin29@huawei.com>
In current HNS3 PMD, Rx/Tx bytes from packet stats are not
implemented.
This patch implemented Rx/Tx bytes using soft counters.
Rx/Tx bytes stats will be enabled if the macro
RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS is defined.
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Signed-off-by: Lijun Ou <oulijun@huawei.com>
---
drivers/net/hns3/hns3_rxtx.c | 24 ++++++++++++++++++++++++
drivers/net/hns3/hns3_rxtx_vec_neon.h | 15 +++++++++++++++
drivers/net/hns3/hns3_rxtx_vec_sve.c | 11 +++++++++++
drivers/net/hns3/hns3_stats.c | 22 ++++++++++++++++++----
4 files changed, 68 insertions(+), 4 deletions(-)
Comments
On 3/2/2021 1:58 PM, Lijun Ou wrote:
> From: "Min Hu (Connor)" <humin29@huawei.com>
>
> In current HNS3 PMD, Rx/Tx bytes from packet stats are not
> implemented.
>
> This patch implemented Rx/Tx bytes using soft counters.
> Rx/Tx bytes stats will be enabled if the macro
> RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS is defined.
>
> Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
> Signed-off-by: Lijun Ou <oulijun@huawei.com>
> ---
> drivers/net/hns3/hns3_rxtx.c | 24 ++++++++++++++++++++++++
> drivers/net/hns3/hns3_rxtx_vec_neon.h | 15 +++++++++++++++
> drivers/net/hns3/hns3_rxtx_vec_sve.c | 11 +++++++++++
> drivers/net/hns3/hns3_stats.c | 22 ++++++++++++++++++----
> 4 files changed, 68 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/hns3/hns3_rxtx.c b/drivers/net/hns3/hns3_rxtx.c
> index 5e79177..a8bd2cc 100644
> --- a/drivers/net/hns3/hns3_rxtx.c
> +++ b/drivers/net/hns3/hns3_rxtx.c
> @@ -2181,6 +2181,10 @@ hns3_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
> cksum_err);
> hns3_rxd_to_vlan_tci(rxq, rxm, l234_info, &rxd);
>
> +#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
> + /* Increment bytes counter */
> + rxq->basic_stats.bytes += rxm->pkt_len;
> +#endif
copy/paste from previous version:
Why statistics enabled only with macro?
It is not common to use macro to enable the stats, what do you think to remove
it, to be consistent with rest of the PMDs?
在 2021/3/3 21:28, Ferruh Yigit 写道:
> On 3/2/2021 1:58 PM, Lijun Ou wrote:
>> From: "Min Hu (Connor)" <humin29@huawei.com>
>>
>> In current HNS3 PMD, Rx/Tx bytes from packet stats are not
>> implemented.
>>
>> This patch implemented Rx/Tx bytes using soft counters.
>> Rx/Tx bytes stats will be enabled if the macro
>> RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS is defined.
>>
>> Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>> ---
>> drivers/net/hns3/hns3_rxtx.c | 24 ++++++++++++++++++++++++
>> drivers/net/hns3/hns3_rxtx_vec_neon.h | 15 +++++++++++++++
>> drivers/net/hns3/hns3_rxtx_vec_sve.c | 11 +++++++++++
>> drivers/net/hns3/hns3_stats.c | 22 ++++++++++++++++++----
>> 4 files changed, 68 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/net/hns3/hns3_rxtx.c b/drivers/net/hns3/hns3_rxtx.c
>> index 5e79177..a8bd2cc 100644
>> --- a/drivers/net/hns3/hns3_rxtx.c
>> +++ b/drivers/net/hns3/hns3_rxtx.c
>> @@ -2181,6 +2181,10 @@ hns3_recv_pkts(void *rx_queue, struct rte_mbuf
>> **rx_pkts, uint16_t nb_pkts)
>> cksum_err);
>> hns3_rxd_to_vlan_tci(rxq, rxm, l234_info, &rxd);
>> +#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
>> + /* Increment bytes counter */
>> + rxq->basic_stats.bytes += rxm->pkt_len;
>> +#endif
>
> copy/paste from previous version:
>
> Why statistics enabled only with macro?
> It is not common to use macro to enable the stats, what do you think to
> remove it, to be consistent with rest of the PMDs?
I'm sorry. I thought it was a success.
Firstly, the macro is used to control the statistics to ensure
performance and facilitate flexible usage. For example, the macro needs
to be disabled when high performance is required.
secondly the byte statistics of other vendors are implemented by reading
and writing registers. Therefore, macros are not used.By the way, the
MLX driver has a precedent (code snippets can be intercepted here).
> _______________________________________________
> Linuxarm mailing list -- linuxarm@openeuler.org
> To unsubscribe send an email to linuxarm-leave@openeuler.org
On 3/3/2021 2:08 PM, oulijun wrote:
>
>
> 在 2021/3/3 21:28, Ferruh Yigit 写道:
>> On 3/2/2021 1:58 PM, Lijun Ou wrote:
>>> From: "Min Hu (Connor)" <humin29@huawei.com>
>>>
>>> In current HNS3 PMD, Rx/Tx bytes from packet stats are not
>>> implemented.
>>>
>>> This patch implemented Rx/Tx bytes using soft counters.
>>> Rx/Tx bytes stats will be enabled if the macro
>>> RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS is defined.
>>>
>>> Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>>> ---
>>> drivers/net/hns3/hns3_rxtx.c | 24 ++++++++++++++++++++++++
>>> drivers/net/hns3/hns3_rxtx_vec_neon.h | 15 +++++++++++++++
>>> drivers/net/hns3/hns3_rxtx_vec_sve.c | 11 +++++++++++
>>> drivers/net/hns3/hns3_stats.c | 22 ++++++++++++++++++----
>>> 4 files changed, 68 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/drivers/net/hns3/hns3_rxtx.c b/drivers/net/hns3/hns3_rxtx.c
>>> index 5e79177..a8bd2cc 100644
>>> --- a/drivers/net/hns3/hns3_rxtx.c
>>> +++ b/drivers/net/hns3/hns3_rxtx.c
>>> @@ -2181,6 +2181,10 @@ hns3_recv_pkts(void *rx_queue, struct rte_mbuf
>>> **rx_pkts, uint16_t nb_pkts)
>>> cksum_err);
>>> hns3_rxd_to_vlan_tci(rxq, rxm, l234_info, &rxd);
>>> +#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
>>> + /* Increment bytes counter */
>>> + rxq->basic_stats.bytes += rxm->pkt_len;
>>> +#endif
>>
>> copy/paste from previous version:
>>
>> Why statistics enabled only with macro?
>> It is not common to use macro to enable the stats, what do you think to remove
>> it, to be consistent with rest of the PMDs?
> I'm sorry. I thought it was a success.
> Firstly, the macro is used to control the statistics to ensure performance and
> facilitate flexible usage. For example, the macro needs to be disabled when high
> performance is required.
> secondly the byte statistics of other vendors are implemented by reading and
> writing registers. Therefore, macros are not used.By the way, the MLX driver has
> a precedent (code snippets can be intercepted here).
It is not convenient for a user re-compile the DPDK to be able to get the ethdev
byte statistics, stats are not developer/debug information, end user may need
them. And this recompilation may not be an option for the distributed software.
How much performance affect it has if you enable it always?
And just to double check, isn't there a way to get this information from HW
without calculating it in the driver?
The compile time flags, specially after meson switch, hard to enable and a
little hidden, it is very easy to have broken code there, that is why it better
to prevent compile time flags as much as possible.
在 2021/3/3 22:24, Ferruh Yigit 写道:
> On 3/3/2021 2:08 PM, oulijun wrote:
>>
>>
>> 在 2021/3/3 21:28, Ferruh Yigit 写道:
>>> On 3/2/2021 1:58 PM, Lijun Ou wrote:
>>>> From: "Min Hu (Connor)" <humin29@huawei.com>
>>>>
>>>> In current HNS3 PMD, Rx/Tx bytes from packet stats are not
>>>> implemented.
>>>>
>>>> This patch implemented Rx/Tx bytes using soft counters.
>>>> Rx/Tx bytes stats will be enabled if the macro
>>>> RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS is defined.
>>>>
>>>> Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
>>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>>>> ---
>>>> drivers/net/hns3/hns3_rxtx.c | 24 ++++++++++++++++++++++++
>>>> drivers/net/hns3/hns3_rxtx_vec_neon.h | 15 +++++++++++++++
>>>> drivers/net/hns3/hns3_rxtx_vec_sve.c | 11 +++++++++++
>>>> drivers/net/hns3/hns3_stats.c | 22 ++++++++++++++++++----
>>>> 4 files changed, 68 insertions(+), 4 deletions(-)
>>>>
>>>> diff --git a/drivers/net/hns3/hns3_rxtx.c
>>>> b/drivers/net/hns3/hns3_rxtx.c
>>>> index 5e79177..a8bd2cc 100644
>>>> --- a/drivers/net/hns3/hns3_rxtx.c
>>>> +++ b/drivers/net/hns3/hns3_rxtx.c
>>>> @@ -2181,6 +2181,10 @@ hns3_recv_pkts(void *rx_queue, struct
>>>> rte_mbuf **rx_pkts, uint16_t nb_pkts)
>>>> cksum_err);
>>>> hns3_rxd_to_vlan_tci(rxq, rxm, l234_info, &rxd);
>>>> +#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
>>>> + /* Increment bytes counter */
>>>> + rxq->basic_stats.bytes += rxm->pkt_len;
>>>> +#endif
>>>
>>> copy/paste from previous version:
>>>
>>> Why statistics enabled only with macro?
>>> It is not common to use macro to enable the stats, what do you think
>>> to remove it, to be consistent with rest of the PMDs?
>> I'm sorry. I thought it was a success.
>> Firstly, the macro is used to control the statistics to ensure
>> performance and facilitate flexible usage. For example, the macro
>> needs to be disabled when high performance is required.
>> secondly the byte statistics of other vendors are implemented by
>> reading and writing registers. Therefore, macros are not used.By the
>> way, the MLX driver has a precedent (code snippets can be intercepted
>> here).
>
> It is not convenient for a user re-compile the DPDK to be able to get
> the ethdev byte statistics, stats are not developer/debug information,
> end user may need them. And this recompilation may not be an option for
> the distributed software.
>
> How much performance affect it has if you enable it always?
>
We theoretically analyzed that being on a critical path might have a
performance impact. Maybe the actual mountain is negligible.
> And just to double check, isn't there a way to get this information from
> HW without calculating it in the driver?
>
OK
> The compile time flags, specially after meson switch, hard to enable and
> a little hidden, it is very easy to have broken code there, that is why
> it better to prevent compile time flags as much as possible.
> .
After internal analysis, you can accept your comments and remove macros.
I will fix it.
>
@@ -2181,6 +2181,10 @@ hns3_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
cksum_err);
hns3_rxd_to_vlan_tci(rxq, rxm, l234_info, &rxd);
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ rxq->basic_stats.bytes += rxm->pkt_len;
+#endif
rx_pkts[nb_rx++] = rxm;
continue;
pkt_err:
@@ -2401,6 +2405,10 @@ hns3_recv_scattered_pkts(void *rx_queue,
cksum_err);
hns3_rxd_to_vlan_tci(rxq, first_seg, l234_info, &rxd);
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ rxq->basic_stats.bytes += first_seg->pkt_len;
+#endif
rx_pkts[nb_rx++] = first_seg;
first_seg = NULL;
continue;
@@ -3516,6 +3524,13 @@ hns3_tx_fill_hw_ring(struct hns3_tx_queue *txq,
for (i = 0; i < mainpart; i += PER_LOOP_NUM) {
hns3_tx_backup_4mbuf(tx_entry + i, pkts + i);
hns3_tx_setup_4bd(txdp + i, pkts + i);
+
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ uint32_t j;
+ for (j = 0; j < PER_LOOP_NUM; j++)
+ txq->basic_stats.bytes += pkts[i + j]->pkt_len;
+#endif
}
if (unlikely(leftover > 0)) {
for (i = 0; i < leftover; i++) {
@@ -3523,6 +3538,11 @@ hns3_tx_fill_hw_ring(struct hns3_tx_queue *txq,
pkts + mainpart + i);
hns3_tx_setup_1bd(txdp + mainpart + i,
pkts + mainpart + i);
+
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ txq->basic_stats.bytes += pkts[mainpart + i]->pkt_len;
+#endif
}
}
}
@@ -3661,6 +3681,10 @@ hns3_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
desc->tx.tp_fe_sc_vld_ra_ri |=
rte_cpu_to_le_16(BIT(HNS3_TXD_FE_B));
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ txq->basic_stats.bytes += tx_pkt->pkt_len;
+#endif
nb_hold += i;
txq->next_to_use = tx_next_use;
txq->tx_bd_ready -= i;
@@ -61,6 +61,11 @@ hns3_xmit_fixed_burst_vec(void *__restrict tx_queue,
for (i = 0; i < n; i++, tx_pkts++, tx_desc++) {
hns3_vec_tx(tx_desc, *tx_pkts);
tx_entry[i].mbuf = *tx_pkts;
+
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ txq->basic_stats.bytes += (*tx_pkts)->pkt_len;
+#endif
}
nb_commit -= n;
@@ -72,6 +77,11 @@ hns3_xmit_fixed_burst_vec(void *__restrict tx_queue,
for (i = 0; i < nb_commit; i++, tx_pkts++, tx_desc++) {
hns3_vec_tx(tx_desc, *tx_pkts);
tx_entry[i].mbuf = *tx_pkts;
+
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ txq->basic_stats.bytes += (*tx_pkts)->pkt_len;
+#endif
}
next_to_use += nb_commit;
@@ -116,6 +126,11 @@ hns3_desc_parse_field(struct hns3_rx_queue *rxq,
if (likely(bd_base_info & BIT(HNS3_RXD_L3L4P_B)))
hns3_rx_set_cksum_flag(pkt, pkt->packet_type,
cksum_err);
+
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ rxq->basic_stats.bytes += pkt->pkt_len;
+#endif
}
return retcode;
@@ -58,6 +58,11 @@ hns3_desc_parse_field_sve(struct hns3_rx_queue *rxq,
if (likely(key->bd_base_info[i] & BIT(HNS3_RXD_L3L4P_B)))
hns3_rx_set_cksum_flag(rx_pkts[i],
rx_pkts[i]->packet_type, cksum_err);
+
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ rxq->basic_stats.bytes += rx_pkts[i]->pkt_len;
+#endif
}
return retcode;
@@ -408,6 +413,12 @@ hns3_tx_fill_hw_ring_sve(struct hns3_tx_queue *txq,
svst1_scatter_u64offset_u64(pg, (uint64_t *)&txdp->tx.paylen,
offsets, svdup_n_u64(valid_bit));
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Increment bytes counter */
+ uint32_t idx;
+ for (idx = 0; idx < svcntd(); idx++)
+ txq->basic_stats.bytes += pkts[idx]->pkt_len;
+#endif
/* update index for next loop */
i += svcntd();
pkts += svcntd();
@@ -358,6 +358,7 @@ static const struct hns3_xstats_name_offset hns3_tx_queue_strings[] = {
HNS3_NUM_RESET_XSTATS)
static void hns3_tqp_stats_clear(struct hns3_hw *hw);
+static void hns3_tqp_basic_stats_clear(struct rte_eth_dev *dev);
/*
* Query all the MAC statistics data of Network ICL command ,opcode id: 0x0034.
@@ -543,16 +544,30 @@ hns3_stats_get(struct rte_eth_dev *eth_dev, struct rte_eth_stats *rte_stats)
return ret;
}
- /* Get the error stats of received packets */
+ /* Get the error stats and bytes of received packets */
for (i = 0; i < eth_dev->data->nb_rx_queues; i++) {
rxq = eth_dev->data->rx_queues[i];
if (rxq) {
cnt = rxq->err_stats.l2_errors +
rxq->err_stats.pkt_len_errors;
rte_stats->ierrors += cnt;
+
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ rte_stats->ibytes += rxq->basic_stats.bytes;
+#endif
}
}
+#ifdef RTE_LIBRTE_HNS3_PMD_SOFT_COUNTERS
+ /* Get the bytes of received packets */
+ struct hns3_tx_queue *txq;
+ for (i = 0; i < eth_dev->data->nb_tx_queues; i++) {
+ txq = eth_dev->data->tx_queues[i];
+ if (txq)
+ rte_stats->obytes += txq->basic_stats.bytes;
+ }
+#endif
+
rte_stats->oerrors = 0;
/*
* If HW statistics are reset by stats_reset, but a lot of residual
@@ -623,6 +638,7 @@ hns3_stats_reset(struct rte_eth_dev *eth_dev)
* their source.
*/
hns3_tqp_stats_clear(hw);
+ hns3_tqp_basic_stats_clear(eth_dev);
return 0;
}
@@ -807,7 +823,6 @@ hns3_rxq_basic_stats_get(struct rte_eth_dev *dev, struct rte_eth_xstat *xstats,
rxq_stats->packets =
stats->rcb_rx_ring_pktnum[i] > rxq_stats->errors ?
stats->rcb_rx_ring_pktnum[i] - rxq_stats->errors : 0;
- rxq_stats->bytes = 0;
for (j = 0; j < HNS3_NUM_RXQ_BASIC_STATS; j++) {
val = (char *)rxq_stats +
hns3_rxq_basic_stats_strings[j].offset;
@@ -836,7 +851,7 @@ hns3_txq_basic_stats_get(struct rte_eth_dev *dev, struct rte_eth_xstat *xstats,
txq_stats = &txq->basic_stats;
txq_stats->packets = stats->rcb_tx_ring_pktnum[i];
- txq_stats->bytes = 0;
+
for (j = 0; j < HNS3_NUM_TXQ_BASIC_STATS; j++) {
val = (char *)txq_stats +
hns3_txq_basic_stats_strings[j].offset;
@@ -1328,7 +1343,6 @@ hns3_dev_xstats_reset(struct rte_eth_dev *dev)
if (ret)
return ret;
- hns3_tqp_basic_stats_clear(dev);
hns3_tqp_dfx_stats_clear(dev);
/* Clear reset stats */