[dpdk-dev] [PATCH v2 0/4] bnxt patches

[dpdk-dev] [PATCH v2 0/4] bnxt patches

[Somnath Kotur]

This set contains bug fixes for commits that made it into
19.08, so it's imperative that these go into 19.08 as well

Please apply

Kalesh AP (3):
  net/bnxt: fix to use chimp hwrm channel for few commands
  net/bnxt: fix endianness issue
  net/bnxt: reduce verbosity of a message

Santoshkumar Karanappa Rastapur (1):
  net/bnxt: fix extended port counter statistics

 drivers/net/bnxt/bnxt_hwrm.c  | 22 +++++++++++-----------
 drivers/net/bnxt/bnxt_stats.c | 24 ++++++++++++++----------
 2 files changed, 25 insertions(+), 21 deletions(-)

-- 
v2: Removed a commit from the set as it got in as part of 
Lance's commit (net/bnxt: use dedicated cpr for async events)
1.8.3.1

[dpdk-dev] [PATCH 1/4] net/bnxt: fix extended port counter statistics

[Somnath Kotur]

From: Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>

We were trying to fill in more rx extended stats than the size allocated
for stats causing segfault. Fixed this by adding an explicit check.
Rearranged the code to return statistic values in xstats_get as per the
names returned in xstats_get_names.

Fixes: f55e12f33416 ("net/bnxt: support extended port counters")

Signed-off-by: Rahul Gupta <rahul.gupta@broadcom.com>
Signed-off-by: Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
---
 drivers/net/bnxt/bnxt_stats.c | 24 ++++++++++++++----------
 1 file changed, 14 insertions(+), 10 deletions(-)

diff --git a/drivers/net/bnxt/bnxt_stats.c b/drivers/net/bnxt/bnxt_stats.c
index 4e74f8a..69ac2dd 100644
--- a/drivers/net/bnxt/bnxt_stats.c
+++ b/drivers/net/bnxt/bnxt_stats.c
@@ -427,8 +427,12 @@ int bnxt_dev_xstats_get_op(struct rte_eth_dev *eth_dev,
 	bnxt_hwrm_port_qstats(bp);
 	bnxt_hwrm_func_qstats_tx_drop(bp, 0xffff, &tx_drop_pkts);
 	bnxt_hwrm_ext_port_qstats(bp);
-	rx_port_stats_ext_cnt = bp->fw_rx_port_stats_ext_size / stat_size;
-	tx_port_stats_ext_cnt = bp->fw_tx_port_stats_ext_size / stat_size;
+	rx_port_stats_ext_cnt = RTE_MIN(RTE_DIM(bnxt_rx_ext_stats_strings),
+					(bp->fw_rx_port_stats_ext_size /
+					 stat_size));
+	tx_port_stats_ext_cnt = RTE_MIN(RTE_DIM(bnxt_tx_ext_stats_strings),
+					(bp->fw_tx_port_stats_ext_size /
+					 stat_size));
 
 	count = RTE_DIM(bnxt_rx_stats_strings) +
 		RTE_DIM(bnxt_tx_stats_strings) + 1/* For tx_drop_pkts */ +
@@ -463,22 +467,22 @@ int bnxt_dev_xstats_get_op(struct rte_eth_dev *eth_dev,
 	xstats[count].value = rte_le_to_cpu_64(tx_drop_pkts);
 	count++;
 
-	for (i = 0; i < tx_port_stats_ext_cnt; i++) {
-		uint64_t *tx_stats_ext = (uint64_t *)bp->hw_tx_port_stats_ext;
+	for (i = 0; i < rx_port_stats_ext_cnt; i++) {
+		uint64_t *rx_stats_ext = (uint64_t *)bp->hw_rx_port_stats_ext;
 
 		xstats[count].value = rte_le_to_cpu_64
-					(*(uint64_t *)((char *)tx_stats_ext +
-					 bnxt_tx_ext_stats_strings[i].offset));
+					(*(uint64_t *)((char *)rx_stats_ext +
+					 bnxt_rx_ext_stats_strings[i].offset));
 
 		count++;
 	}
 
-	for (i = 0; i < rx_port_stats_ext_cnt; i++) {
-		uint64_t *rx_stats_ext = (uint64_t *)bp->hw_rx_port_stats_ext;
+	for (i = 0; i < tx_port_stats_ext_cnt; i++) {
+		uint64_t *tx_stats_ext = (uint64_t *)bp->hw_tx_port_stats_ext;
 
 		xstats[count].value = rte_le_to_cpu_64
-					(*(uint64_t *)((char *)rx_stats_ext +
-					 bnxt_rx_ext_stats_strings[i].offset));
+					(*(uint64_t *)((char *)tx_stats_ext +
+					 bnxt_tx_ext_stats_strings[i].offset));
 
 		count++;
 	}
-- 
1.8.3.1

Re: [dpdk-dev] [PATCH 1/4] net/bnxt: fix extended port counter statistics

[David Marchand]

On Thu, Jul 25, 2019 at 7:05 AM Somnath Kotur
<somnath.kotur@broadcom.com> wrote:
>
> From: Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>
>
> We were trying to fill in more rx extended stats than the size allocated
> for stats causing segfault. Fixed this by adding an explicit check.
> Rearranged the code to return statistic values in xstats_get as per the
> names returned in xstats_get_names.
>
> Fixes: f55e12f33416 ("net/bnxt: support extended port counters")
>
> Signed-off-by: Rahul Gupta <rahul.gupta@broadcom.com>
> Signed-off-by: Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>
> Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
> ---
>  drivers/net/bnxt/bnxt_stats.c | 24 ++++++++++++++----------
>  1 file changed, 14 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/net/bnxt/bnxt_stats.c b/drivers/net/bnxt/bnxt_stats.c
> index 4e74f8a..69ac2dd 100644
> --- a/drivers/net/bnxt/bnxt_stats.c
> +++ b/drivers/net/bnxt/bnxt_stats.c

[snip]

> @@ -463,22 +467,22 @@ int bnxt_dev_xstats_get_op(struct rte_eth_dev *eth_dev,
>         xstats[count].value = rte_le_to_cpu_64(tx_drop_pkts);
>         count++;
>
> -       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
> -               uint64_t *tx_stats_ext = (uint64_t *)bp->hw_tx_port_stats_ext;
> +       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
> +               uint64_t *rx_stats_ext = (uint64_t *)bp->hw_rx_port_stats_ext;
>
>                 xstats[count].value = rte_le_to_cpu_64
> -                                       (*(uint64_t *)((char *)tx_stats_ext +
> -                                        bnxt_tx_ext_stats_strings[i].offset));
> +                                       (*(uint64_t *)((char *)rx_stats_ext +
> +                                        bnxt_rx_ext_stats_strings[i].offset));
>
>                 count++;
>         }
>
> -       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
> -               uint64_t *rx_stats_ext = (uint64_t *)bp->hw_rx_port_stats_ext;
> +       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
> +               uint64_t *tx_stats_ext = (uint64_t *)bp->hw_tx_port_stats_ext;
>
>                 xstats[count].value = rte_le_to_cpu_64
> -                                       (*(uint64_t *)((char *)rx_stats_ext +
> -                                        bnxt_rx_ext_stats_strings[i].offset));
> +                                       (*(uint64_t *)((char *)tx_stats_ext +
> +                                        bnxt_tx_ext_stats_strings[i].offset));
>
>                 count++;
>         }
> --
> 1.8.3.1
>

This whole hunk just adds some noise, right? or is there anything fixed in it?


-- 
David Marchand

Re: [dpdk-dev] [PATCH 1/4] net/bnxt: fix extended port counter statistics

[Somnath Kotur]

+Santosh

On Thu, Jul 25, 2019 at 12:52 PM David Marchand <david.marchand@redhat.com>
wrote:

> On Thu, Jul 25, 2019 at 7:05 AM Somnath Kotur
> <somnath.kotur@broadcom.com> wrote:
> >
> > From: Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>
> >
> > We were trying to fill in more rx extended stats than the size allocated
> > for stats causing segfault. Fixed this by adding an explicit check.
> > Rearranged the code to return statistic values in xstats_get as per the
> > names returned in xstats_get_names.
> >
> > Fixes: f55e12f33416 ("net/bnxt: support extended port counters")
> >
> > Signed-off-by: Rahul Gupta <rahul.gupta@broadcom.com>
> > Signed-off-by: Santoshkumar Karanappa Rastapur <
> santosh.rastapur@broadcom.com>
> > Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
> > ---
> >  drivers/net/bnxt/bnxt_stats.c | 24 ++++++++++++++----------
> >  1 file changed, 14 insertions(+), 10 deletions(-)
> >
> > diff --git a/drivers/net/bnxt/bnxt_stats.c
> b/drivers/net/bnxt/bnxt_stats.c
> > index 4e74f8a..69ac2dd 100644
> > --- a/drivers/net/bnxt/bnxt_stats.c
> > +++ b/drivers/net/bnxt/bnxt_stats.c
>
> [snip]
>
> > @@ -463,22 +467,22 @@ int bnxt_dev_xstats_get_op(struct rte_eth_dev
> *eth_dev,
> >         xstats[count].value = rte_le_to_cpu_64(tx_drop_pkts);
> >         count++;
> >
> > -       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
> > -               uint64_t *tx_stats_ext = (uint64_t
> *)bp->hw_tx_port_stats_ext;
> > +       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
> > +               uint64_t *rx_stats_ext = (uint64_t
> *)bp->hw_rx_port_stats_ext;
> >
> >                 xstats[count].value = rte_le_to_cpu_64
> > -                                       (*(uint64_t *)((char
> *)tx_stats_ext +
> > -
> bnxt_tx_ext_stats_strings[i].offset));
> > +                                       (*(uint64_t *)((char
> *)rx_stats_ext +
> > +
> bnxt_rx_ext_stats_strings[i].offset));
> >
> >                 count++;
> >         }
> >
> > -       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
> > -               uint64_t *rx_stats_ext = (uint64_t
> *)bp->hw_rx_port_stats_ext;
> > +       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
> > +               uint64_t *tx_stats_ext = (uint64_t
> *)bp->hw_tx_port_stats_ext;
> >
> >                 xstats[count].value = rte_le_to_cpu_64
> > -                                       (*(uint64_t *)((char
> *)rx_stats_ext +
> > -
> bnxt_rx_ext_stats_strings[i].offset));
> > +                                       (*(uint64_t *)((char
> *)tx_stats_ext +
> > +
> bnxt_tx_ext_stats_strings[i].offset));
> >
> >                 count++;
> >         }
> > --
> > 1.8.3.1
> >
>
> This whole hunk just adds some noise, right? or is there anything fixed in
> it?
>
>
> --
> David Marchand
>

Re: [dpdk-dev] [PATCH 1/4] net/bnxt: fix extended port counter statistics

[Santoshkumar Karanappa Rastapur]

On Thu, Jul 25, 2019 at 1:09 PM Somnath Kotur <somnath.kotur@broadcom.com>
wrote:

> +Santosh
>
> On Thu, Jul 25, 2019 at 12:52 PM David Marchand <david.marchand@redhat.com>
> wrote:
>
>> On Thu, Jul 25, 2019 at 7:05 AM Somnath Kotur
>> <somnath.kotur@broadcom.com> wrote:
>> >
>> > From: Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>
>> >
>> > We were trying to fill in more rx extended stats than the size allocated
>> > for stats causing segfault. Fixed this by adding an explicit check.
>> > Rearranged the code to return statistic values in xstats_get as per the
>> > names returned in xstats_get_names.
>> >
>> > Fixes: f55e12f33416 ("net/bnxt: support extended port counters")
>> >
>> > Signed-off-by: Rahul Gupta <rahul.gupta@broadcom.com>
>> > Signed-off-by: Santoshkumar Karanappa Rastapur <
>> santosh.rastapur@broadcom.com>
>> > Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
>> > ---
>> >  drivers/net/bnxt/bnxt_stats.c | 24 ++++++++++++++----------
>> >  1 file changed, 14 insertions(+), 10 deletions(-)
>> >
>> > diff --git a/drivers/net/bnxt/bnxt_stats.c
>> b/drivers/net/bnxt/bnxt_stats.c
>> > index 4e74f8a..69ac2dd 100644
>> > --- a/drivers/net/bnxt/bnxt_stats.c
>> > +++ b/drivers/net/bnxt/bnxt_stats.c
>>
>> [snip]
>>
>> > @@ -463,22 +467,22 @@ int bnxt_dev_xstats_get_op(struct rte_eth_dev
>> *eth_dev,
>> >         xstats[count].value = rte_le_to_cpu_64(tx_drop_pkts);
>> >         count++;
>> >
>> > -       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
>> > -               uint64_t *tx_stats_ext = (uint64_t
>> *)bp->hw_tx_port_stats_ext;
>> > +       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
>> > +               uint64_t *rx_stats_ext = (uint64_t
>> *)bp->hw_rx_port_stats_ext;
>> >
>> >                 xstats[count].value = rte_le_to_cpu_64
>> > -                                       (*(uint64_t *)((char
>> *)tx_stats_ext +
>> > -
>> bnxt_tx_ext_stats_strings[i].offset));
>> > +                                       (*(uint64_t *)((char
>> *)rx_stats_ext +
>> > +
>> bnxt_rx_ext_stats_strings[i].offset));
>> >
>> >                 count++;
>> >         }
>> >
>> > -       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
>> > -               uint64_t *rx_stats_ext = (uint64_t
>> *)bp->hw_rx_port_stats_ext;
>> > +       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
>> > +               uint64_t *tx_stats_ext = (uint64_t
>> *)bp->hw_tx_port_stats_ext;
>> >
>> >                 xstats[count].value = rte_le_to_cpu_64
>> > -                                       (*(uint64_t *)((char
>> *)rx_stats_ext +
>> > -
>> bnxt_rx_ext_stats_strings[i].offset));
>> > +                                       (*(uint64_t *)((char
>> *)tx_stats_ext +
>> > +
>> bnxt_tx_ext_stats_strings[i].offset));
>> >
>> >                 count++;
>> >         }
>> > --
>> > 1.8.3.1
>> >
>>
>> This whole hunk just adds some noise, right? or is there anything fixed
>> in it?
>>
>>
>> --
>> David Marchand
>>
>

In bnxt_dev_xstats_get_names_op, we were filling statistics names in
xstats_names in this order.

                bnxt_rx_stats_strings

                bnxt_tx_stats_strings

                bnxt_rx_ext_stats_strings

                bnxt_tx_ext_stats_strings



Where as in bnxt_dev_xstats_get_op, we were returning stats values in
xstats in this order.

                bnxt_rx_stats_strings

                bnxt_tx_stats_strings

                bnxt_tx_ext_stats_strings

                bnxt_rx_ext_stats_strings



We were ending up displaying extended Tx stats values against extended Rx
stats names and vice versa.

This above code fixes this order.

Regards
-Santosh

Re: [dpdk-dev] [PATCH 1/4] net/bnxt: fix extended port counter statistics

[David Marchand]

On Thu, Jul 25, 2019 at 9:54 AM Santoshkumar Karanappa Rastapur
<santosh.rastapur@broadcom.com> wrote:
>
>
>
> On Thu, Jul 25, 2019 at 1:09 PM Somnath Kotur <somnath.kotur@broadcom.com> wrote:
>>
>> +Santosh
>>
>> On Thu, Jul 25, 2019 at 12:52 PM David Marchand <david.marchand@redhat.com> wrote:
>>>
>>> On Thu, Jul 25, 2019 at 7:05 AM Somnath Kotur
>>> <somnath.kotur@broadcom.com> wrote:
>>> >
>>> > From: Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>
>>> >
>>> > We were trying to fill in more rx extended stats than the size allocated
>>> > for stats causing segfault. Fixed this by adding an explicit check.
>>> > Rearranged the code to return statistic values in xstats_get as per the
>>> > names returned in xstats_get_names.
>>> >
>>> > Fixes: f55e12f33416 ("net/bnxt: support extended port counters")
>>> >
>>> > Signed-off-by: Rahul Gupta <rahul.gupta@broadcom.com>
>>> > Signed-off-by: Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>
>>> > Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
>>> > ---
>>> >  drivers/net/bnxt/bnxt_stats.c | 24 ++++++++++++++----------
>>> >  1 file changed, 14 insertions(+), 10 deletions(-)
>>> >
>>> > diff --git a/drivers/net/bnxt/bnxt_stats.c b/drivers/net/bnxt/bnxt_stats.c
>>> > index 4e74f8a..69ac2dd 100644
>>> > --- a/drivers/net/bnxt/bnxt_stats.c
>>> > +++ b/drivers/net/bnxt/bnxt_stats.c
>>>
>>> [snip]
>>>
>>> > @@ -463,22 +467,22 @@ int bnxt_dev_xstats_get_op(struct rte_eth_dev *eth_dev,
>>> >         xstats[count].value = rte_le_to_cpu_64(tx_drop_pkts);
>>> >         count++;
>>> >
>>> > -       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
>>> > -               uint64_t *tx_stats_ext = (uint64_t *)bp->hw_tx_port_stats_ext;
>>> > +       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
>>> > +               uint64_t *rx_stats_ext = (uint64_t *)bp->hw_rx_port_stats_ext;
>>> >
>>> >                 xstats[count].value = rte_le_to_cpu_64
>>> > -                                       (*(uint64_t *)((char *)tx_stats_ext +
>>> > -                                        bnxt_tx_ext_stats_strings[i].offset));
>>> > +                                       (*(uint64_t *)((char *)rx_stats_ext +
>>> > +                                        bnxt_rx_ext_stats_strings[i].offset));
>>> >
>>> >                 count++;
>>> >         }
>>> >
>>> > -       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
>>> > -               uint64_t *rx_stats_ext = (uint64_t *)bp->hw_rx_port_stats_ext;
>>> > +       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
>>> > +               uint64_t *tx_stats_ext = (uint64_t *)bp->hw_tx_port_stats_ext;
>>> >
>>> >                 xstats[count].value = rte_le_to_cpu_64
>>> > -                                       (*(uint64_t *)((char *)rx_stats_ext +
>>> > -                                        bnxt_rx_ext_stats_strings[i].offset));
>>> > +                                       (*(uint64_t *)((char *)tx_stats_ext +
>>> > +                                        bnxt_tx_ext_stats_strings[i].offset));
>>> >
>>> >                 count++;
>>> >         }
>>> > --
>>> > 1.8.3.1
>>> >
>>>
>>> This whole hunk just adds some noise, right? or is there anything fixed in it?
>>>
>>>
>>> --
>>> David Marchand
>
>
>
> In bnxt_dev_xstats_get_names_op, we were filling statistics names in xstats_names in this order.
>
>                 bnxt_rx_stats_strings
>
>                 bnxt_tx_stats_strings
>
>                 bnxt_rx_ext_stats_strings
>
>                 bnxt_tx_ext_stats_strings
>
>
>
> Where as in bnxt_dev_xstats_get_op, we were returning stats values in xstats in this order.
>
>                 bnxt_rx_stats_strings
>
>                 bnxt_tx_stats_strings
>
>                 bnxt_tx_ext_stats_strings
>
>                 bnxt_rx_ext_stats_strings
>
>
>
> We were ending up displaying extended Tx stats values against extended Rx stats names and vice versa.
>
> This above code fixes this order.

Erf, I must have read your commitlog too quickly, or you could have
split it in two patches, anyway understood.


-- 
David Marchand

Re: [dpdk-dev] [PATCH 1/4] net/bnxt: fix extended port counter statistics

[Somnath Kotur]

+Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>

On Thu, Jul 25, 2019 at 10:35 AM Somnath Kotur <somnath.kotur@broadcom.com>
wrote:

> From: Santoshkumar Karanappa Rastapur <santosh.rastapur@broadcom.com>
>
> We were trying to fill in more rx extended stats than the size allocated
> for stats causing segfault. Fixed this by adding an explicit check.
> Rearranged the code to return statistic values in xstats_get as per the
> names returned in xstats_get_names.
>
> Fixes: f55e12f33416 ("net/bnxt: support extended port counters")
>
> Signed-off-by: Rahul Gupta <rahul.gupta@broadcom.com>
> Signed-off-by: Santoshkumar Karanappa Rastapur <
> santosh.rastapur@broadcom.com>
> Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
> ---
>  drivers/net/bnxt/bnxt_stats.c | 24 ++++++++++++++----------
>  1 file changed, 14 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/net/bnxt/bnxt_stats.c b/drivers/net/bnxt/bnxt_stats.c
> index 4e74f8a..69ac2dd 100644
> --- a/drivers/net/bnxt/bnxt_stats.c
> +++ b/drivers/net/bnxt/bnxt_stats.c
> @@ -427,8 +427,12 @@ int bnxt_dev_xstats_get_op(struct rte_eth_dev
> *eth_dev,
>         bnxt_hwrm_port_qstats(bp);
>         bnxt_hwrm_func_qstats_tx_drop(bp, 0xffff, &tx_drop_pkts);
>         bnxt_hwrm_ext_port_qstats(bp);
> -       rx_port_stats_ext_cnt = bp->fw_rx_port_stats_ext_size / stat_size;
> -       tx_port_stats_ext_cnt = bp->fw_tx_port_stats_ext_size / stat_size;
> +       rx_port_stats_ext_cnt = RTE_MIN(RTE_DIM(bnxt_rx_ext_stats_strings),
> +                                       (bp->fw_rx_port_stats_ext_size /
> +                                        stat_size));
> +       tx_port_stats_ext_cnt = RTE_MIN(RTE_DIM(bnxt_tx_ext_stats_strings),
> +                                       (bp->fw_tx_port_stats_ext_size /
> +                                        stat_size));
>
>         count = RTE_DIM(bnxt_rx_stats_strings) +
>                 RTE_DIM(bnxt_tx_stats_strings) + 1/* For tx_drop_pkts */ +
> @@ -463,22 +467,22 @@ int bnxt_dev_xstats_get_op(struct rte_eth_dev
> *eth_dev,
>         xstats[count].value = rte_le_to_cpu_64(tx_drop_pkts);
>         count++;
>
> -       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
> -               uint64_t *tx_stats_ext = (uint64_t
> *)bp->hw_tx_port_stats_ext;
> +       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
> +               uint64_t *rx_stats_ext = (uint64_t
> *)bp->hw_rx_port_stats_ext;
>
>                 xstats[count].value = rte_le_to_cpu_64
> -                                       (*(uint64_t *)((char
> *)tx_stats_ext +
> -
> bnxt_tx_ext_stats_strings[i].offset));
> +                                       (*(uint64_t *)((char
> *)rx_stats_ext +
> +
> bnxt_rx_ext_stats_strings[i].offset));
>
>                 count++;
>         }
>
> -       for (i = 0; i < rx_port_stats_ext_cnt; i++) {
> -               uint64_t *rx_stats_ext = (uint64_t
> *)bp->hw_rx_port_stats_ext;
> +       for (i = 0; i < tx_port_stats_ext_cnt; i++) {
> +               uint64_t *tx_stats_ext = (uint64_t
> *)bp->hw_tx_port_stats_ext;
>
>                 xstats[count].value = rte_le_to_cpu_64
> -                                       (*(uint64_t *)((char
> *)rx_stats_ext +
> -
> bnxt_rx_ext_stats_strings[i].offset));
> +                                       (*(uint64_t *)((char
> *)tx_stats_ext +
> +
> bnxt_tx_ext_stats_strings[i].offset));
>
>                 count++;
>         }
> --
> 1.8.3.1
>
>

[dpdk-dev] [PATCH 2/4] net/bnxt: fix to use chimp hwrm channel for few commands

[Somnath Kotur]

From: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>

Modified to send the tunnel redirect commands to Chimp as Kong does not
support these commands.

Fixes: 11e5e19695c7bfb3 ("net/bnxt: support redirecting tunnel packets to VF")

Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
---
 drivers/net/bnxt/bnxt_hwrm.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/net/bnxt/bnxt_hwrm.c b/drivers/net/bnxt/bnxt_hwrm.c
index 045ce4a..1587ca1 100644
--- a/drivers/net/bnxt/bnxt_hwrm.c
+++ b/drivers/net/bnxt/bnxt_hwrm.c
@@ -4524,10 +4524,10 @@ int bnxt_hwrm_ext_port_qstats(struct bnxt *bp)
 		bp->hwrm_cmd_resp_addr;
 	int rc = 0;
 
-	HWRM_PREP(req, CFA_REDIRECT_TUNNEL_TYPE_ALLOC, BNXT_USE_KONG(bp));
+	HWRM_PREP(req, CFA_REDIRECT_TUNNEL_TYPE_ALLOC, BNXT_USE_CHIMP_MB);
 	req.tunnel_type = type;
 	req.dest_fid = bp->fw_fid;
-	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_KONG(bp));
+	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_CHIMP_MB);
 	HWRM_CHECK_RESULT();
 
 	HWRM_UNLOCK();
@@ -4543,10 +4543,10 @@ int bnxt_hwrm_ext_port_qstats(struct bnxt *bp)
 		bp->hwrm_cmd_resp_addr;
 	int rc = 0;
 
-	HWRM_PREP(req, CFA_REDIRECT_TUNNEL_TYPE_FREE, BNXT_USE_KONG(bp));
+	HWRM_PREP(req, CFA_REDIRECT_TUNNEL_TYPE_FREE, BNXT_USE_CHIMP_MB);
 	req.tunnel_type = type;
 	req.dest_fid = bp->fw_fid;
-	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_KONG(bp));
+	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_CHIMP_MB);
 	HWRM_CHECK_RESULT();
 
 	HWRM_UNLOCK();
@@ -4561,9 +4561,9 @@ int bnxt_hwrm_tunnel_redirect_query(struct bnxt *bp, uint32_t *type)
 		bp->hwrm_cmd_resp_addr;
 	int rc = 0;
 
-	HWRM_PREP(req, CFA_REDIRECT_QUERY_TUNNEL_TYPE, BNXT_USE_KONG(bp));
+	HWRM_PREP(req, CFA_REDIRECT_QUERY_TUNNEL_TYPE, BNXT_USE_CHIMP_MB);
 	req.src_fid = bp->fw_fid;
-	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_KONG(bp));
+	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_CHIMP_MB);
 	HWRM_CHECK_RESULT();
 
 	if (type)
@@ -4582,10 +4582,10 @@ int bnxt_hwrm_tunnel_redirect_info(struct bnxt *bp, uint8_t tun_type,
 		bp->hwrm_cmd_resp_addr;
 	int rc = 0;
 
-	HWRM_PREP(req, CFA_REDIRECT_TUNNEL_TYPE_INFO, BNXT_USE_KONG(bp));
+	HWRM_PREP(req, CFA_REDIRECT_TUNNEL_TYPE_INFO, BNXT_USE_CHIMP_MB);
 	req.src_fid = bp->fw_fid;
 	req.tunnel_type = tun_type;
-	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_KONG(bp));
+	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_CHIMP_MB);
 	HWRM_CHECK_RESULT();
 
 	if (dst_fid)
-- 
1.8.3.1

[dpdk-dev] [PATCH 3/4] net/bnxt: fix endianness issue

[Somnath Kotur]

From: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>

Use rte_cpu_to_le_16/32 while parsing the hwrm command response.

Fixes: 11e5e19695c7bfb3 ("net/bnxt: support redirecting tunnel packets to VF")

Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
---
 drivers/net/bnxt/bnxt_hwrm.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/bnxt/bnxt_hwrm.c b/drivers/net/bnxt/bnxt_hwrm.c
index 1587ca1..39c71bc 100644
--- a/drivers/net/bnxt/bnxt_hwrm.c
+++ b/drivers/net/bnxt/bnxt_hwrm.c
@@ -4567,7 +4567,7 @@ int bnxt_hwrm_tunnel_redirect_query(struct bnxt *bp, uint32_t *type)
 	HWRM_CHECK_RESULT();
 
 	if (type)
-		*type = resp->tunnel_mask;
+		*type = rte_le_to_cpu_32(resp->tunnel_mask);
 
 	HWRM_UNLOCK();
 
@@ -4589,7 +4589,7 @@ int bnxt_hwrm_tunnel_redirect_info(struct bnxt *bp, uint8_t tun_type,
 	HWRM_CHECK_RESULT();
 
 	if (dst_fid)
-		*dst_fid = resp->dest_fid;
+		*dst_fid = rte_le_to_cpu_16(resp->dest_fid);
 
 	PMD_DRV_LOG(DEBUG, "dst_fid: %x\n", resp->dest_fid);
 
-- 
1.8.3.1

[dpdk-dev] [PATCH 4/4] net/bnxt: reduce verbosity of a message

[Somnath Kotur]

From: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>

Change verbosity of a message to DEBUG from ERROR.
This is just debug message.

Fixes: 3e92fd4e4ec0c7b ("net/bnxt: use dynamic log type")

Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
---
 drivers/net/bnxt/bnxt_hwrm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/bnxt/bnxt_hwrm.c b/drivers/net/bnxt/bnxt_hwrm.c
index 39c71bc..74b8d00 100644
--- a/drivers/net/bnxt/bnxt_hwrm.c
+++ b/drivers/net/bnxt/bnxt_hwrm.c
@@ -2261,7 +2261,7 @@ int bnxt_clear_hwrm_vnic_filters(struct bnxt *bp, struct bnxt_vnic_info *vnic)
 
 	STAILQ_FOREACH(flow, &vnic->flow_list, next) {
 		filter = flow->filter;
-		PMD_DRV_LOG(ERR, "filter type %d\n", filter->filter_type);
+		PMD_DRV_LOG(DEBUG, "filter type %d\n", filter->filter_type);
 		if (filter->filter_type == HWRM_CFA_EM_FILTER)
 			rc = bnxt_hwrm_clear_em_filter(bp, filter);
 		else if (filter->filter_type == HWRM_CFA_NTUPLE_FILTER)
-- 
1.8.3.1

Re: [dpdk-dev] [PATCH v2 0/4] bnxt patches

[Ferruh Yigit]

On 7/25/2019 5:59 AM, Somnath Kotur wrote:
> This set contains bug fixes for commits that made it into
> 19.08, so it's imperative that these go into 19.08 as well
> 
> Please apply
> 
> Kalesh AP (3):
>   net/bnxt: fix to use chimp hwrm channel for few commands
>   net/bnxt: fix endianness issue
>   net/bnxt: reduce verbosity of a message
> 
> Santoshkumar Karanappa Rastapur (1):
>   net/bnxt: fix extended port counter statistics

Series applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH 0/4] bnxt patches

[Ajit Khaparde]

On Wed, Jul 29, 2020 at 7:09 AM Somnath Kotur <somnath.kotur@broadcom.com>
wrote:

> fixes and cleanups in bnxt TRUFlow
>
> Kishore Padmanabha (2):
>   net/bnxt: configure loopback parif for full offload egress flows
>   net/bnxt: enable lookup default action record parif entry
>
> Somnath Kotur (1):
>   net/bnxt: cleanups and checks for ulp context allocation
>
> Venkat Duvvuru (1):
>   net/bnxt: fix vfrep port add when it's endpoint interface is down
>
>  drivers/net/bnxt/bnxt.h                         |  21 ++
>  drivers/net/bnxt/bnxt_cpr.c                     |  51 ++++
>  drivers/net/bnxt/bnxt_ethdev.c                  |  12 +-
>  drivers/net/bnxt/bnxt_hwrm.c                    |   4 +
>  drivers/net/bnxt/bnxt_hwrm.h                    |   2 +
>  drivers/net/bnxt/bnxt_reps.c                    |  88 +++++--
>  drivers/net/bnxt/tf_ulp/bnxt_ulp.c              |   2 +
>  drivers/net/bnxt/tf_ulp/ulp_def_rules.c         |   2 +-
>  drivers/net/bnxt/tf_ulp/ulp_rte_parser.c        |   6 +
>  drivers/net/bnxt/tf_ulp/ulp_template_db_class.c | 314
> ++++++++++++++----------
>  drivers/net/bnxt/tf_ulp/ulp_template_db_enum.h  |   3 +-
>  11 files changed, 356 insertions(+), 149 deletions(-)
>
> Patchset applied to dpdk-next-net-brcm.


> --
> 2.7.4
>
>

[dpdk-dev] [PATCH v2 0/4] bnxt patches

[Ajit Khaparde]

fixes and cleanups in bnxt TRUFlow

v1->v2: updated signed-off tags in commit logs.

Kishore Padmanabha (2):
  net/bnxt: fix loopback parif for egress flows
  net/bnxt: fix lookup for default parif action record

Somnath Kotur (1):
  net/bnxt: fix cleanup and check for ulp context alloc

Venkat Duvvuru (1):
  net/bnxt: fix vfrep add when VF interface is down

 drivers/net/bnxt/bnxt.h                       |  21 ++
 drivers/net/bnxt/bnxt_cpr.c                   |  51 +++
 drivers/net/bnxt/bnxt_ethdev.c                |  12 +-
 drivers/net/bnxt/bnxt_hwrm.c                  |   4 +
 drivers/net/bnxt/bnxt_hwrm.h                  |   2 +
 drivers/net/bnxt/bnxt_reps.c                  |  88 +++--
 drivers/net/bnxt/tf_ulp/bnxt_ulp.c            |   2 +
 drivers/net/bnxt/tf_ulp/ulp_def_rules.c       |   2 +-
 drivers/net/bnxt/tf_ulp/ulp_rte_parser.c      |   6 +
 .../net/bnxt/tf_ulp/ulp_template_db_class.c   | 314 +++++++++++-------
 .../net/bnxt/tf_ulp/ulp_template_db_enum.h    |   3 +-
 11 files changed, 356 insertions(+), 149 deletions(-)

-- 
2.21.1 (Apple Git-122.3)