Skip to contentSkip to conversation

net/bnxt: allow configuring vector mode

PATCH In review Stephen Hemminger 5 participants 10 messages Mar 5, 2020 – Apr 24, 2020 Network drivers

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net/bnxt: allow configuring vector mode

The bnxt driver has vector mode but it has limitations. For example, rte_flow mark won't work in vector mode. For this reason the user should be able to disable vector mode as part of the config.

Make the configuration use the same as other drivers with vector mode: ixge, i40e, ...

This will also make future support of vector mode on other architectures possible.

Fixes bc4a000f2f53 ("net/bnxt: implement SSE vector mode")
---
 config/common_base             |  1 +
 drivers/net/bnxt/Makefile      |  2 +-
 drivers/net/bnxt/bnxt_ethdev.c | 10 +++-------
 drivers/net/bnxt/bnxt_ring.c   |  4 ++--
 drivers/net/bnxt/bnxt_rxr.c    |  3 +++
 5 files changed, 10 insertions(+), 10 deletions(-)
config/common_base +1 -0
index 7ca2f28b19c8..3c719c7cbfc3 100644
@@ -219,6 +219,7 @@ CONFIG_RTE_LIBRTE_BNX2X_DEBUG_PERIODIC=n
 # Compile burst-oriented Broadcom BNXT PMD driver
 #
 CONFIG_RTE_LIBRTE_BNXT_PMD=y
+CONFIG_RTE_LIBRTE_BNXT_INC_VECTOR=y
 
 #
 # Compile burst-oriented Chelsio Terminator (CXGBE) PMD
drivers/net/bnxt/Makefile +1 -1
index b77532b8179b..85d6df6b5aa0 100644
@@ -40,7 +40,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_BNXT_PMD) += bnxt_irq.c
 SRCS-$(CONFIG_RTE_LIBRTE_BNXT_PMD) += bnxt_util.c
 SRCS-$(CONFIG_RTE_LIBRTE_BNXT_PMD) += rte_pmd_bnxt.c
 ifeq ($(CONFIG_RTE_ARCH_X86), y)
-SRCS-$(CONFIG_RTE_LIBRTE_BNXT_PMD) += bnxt_rxtx_vec_sse.c
+SRCS-$(CONFIG_RTE_LIBRTE_BNXT_INC_VECTOR) += bnxt_rxtx_vec_sse.c
 endif
 
 #
drivers/net/bnxt/bnxt_ethdev.c +3 -7
index 18aa313fd8c1..701de33784b2 100644
@@ -757,8 +757,7 @@ bnxt_receive_function(struct rte_eth_dev *eth_dev)
 {
 	struct bnxt *bp = eth_dev->data->dev_private;
 
-#ifdef RTE_ARCH_X86
-#ifndef RTE_LIBRTE_IEEE1588
+#ifdef RTE_BNXT_INC_VECTOR
 	/*
 	 * Vector mode receive can be enabled only if scatter rx is not
 	 * in use and rx offloads are limited to VLAN stripping and
@@ -787,7 +786,6 @@ bnxt_receive_function(struct rte_eth_dev *eth_dev)
 		    eth_dev->data->port_id,
 		    eth_dev->data->scattered_rx,
 		    eth_dev->data->dev_conf.rxmode.offloads);
-#endif
 #endif
 	bp->flags &= ~BNXT_FLAG_RX_VECTOR_PKT_MODE;
 	return bnxt_recv_pkts;
@@ -796,8 +794,7 @@ bnxt_receive_function(struct rte_eth_dev *eth_dev)
 static eth_tx_burst_t
 bnxt_transmit_function(__rte_unused struct rte_eth_dev *eth_dev)
 {
-#ifdef RTE_ARCH_X86
-#ifndef RTE_LIBRTE_IEEE1588
+#ifdef RTE_BNXT_INC_VECTOR
 	/*
 	 * Vector mode transmit can be enabled only if not using scatter rx
 	 * or tx offloads.
@@ -815,7 +812,6 @@ bnxt_transmit_function(__rte_unused struct rte_eth_dev *eth_dev)
 		    eth_dev->data->port_id,
 		    eth_dev->data->scattered_rx,
 		    eth_dev->data->dev_conf.txmode.offloads);
-#endif
 #endif
 	return bnxt_xmit_pkts;
 }
@@ -2234,7 +2230,7 @@ int bnxt_mtu_set_op(struct rte_eth_dev *eth_dev, uint16_t new_mtu)
 	new_pkt_size = new_mtu + RTE_ETHER_HDR_LEN + RTE_ETHER_CRC_LEN +
 		       VLAN_TAG_SIZE * BNXT_NUM_VLANS;
 
-#ifdef RTE_ARCH_X86
+#ifdef RTE_BNXT_INC_VECTOR
 	/*
 	 * If vector-mode tx/rx is active, disallow any MTU change that would
 	 * require scattered receive support.
drivers/net/bnxt/bnxt_ring.c +2 -2
index d6e4e8a28138..9a4a0f21418d 100644
@@ -608,7 +608,7 @@ int bnxt_alloc_hwrm_rx_ring(struct bnxt *bp, int queue_index)
 		bnxt_db_write(&rxr->ag_db, rxr->ag_prod);
 	}
 	rxq->index = queue_index;
-#ifdef RTE_ARCH_X86
+#ifdef RTE_BNXT_INC_VECTOR
 	bnxt_rxq_vec_setup(rxq);
 #endif
 
@@ -713,7 +713,7 @@ int bnxt_alloc_hwrm_rings(struct bnxt *bp)
 		bnxt_db_write(&rxr->rx_db, rxr->rx_prod);
 		bnxt_db_write(&rxr->ag_db, rxr->ag_prod);
 		rxq->index = i;
-#ifdef RTE_ARCH_X86
+#ifdef RTE_BNXT_INC_VECTOR
 		bnxt_rxq_vec_setup(rxq);
 #endif
 	}
drivers/net/bnxt/bnxt_rxr.c +3 -0
index bef9720f590c..f9087bea89b9 100644
@@ -18,6 +18,9 @@
 #include "hsi_struct_def_dpdk.h"
 #ifdef RTE_LIBRTE_IEEE1588
 #include "bnxt_hwrm.h"
+#ifdef RTE_BNXT_INC_VECTOR
+#error "bnxt: IEEE1588 is incompatiable with vector mode"
+#endif
 #endif
 
 /*
Signature · 1 line
2.20.1
Acked-by on v1 1/1
#1 lore raw

Hi Stephen,

Quoted 1 line · Stephen Hemminger

<snip>

Quoted 3 lines

Make the configuration use the same as other drivers with vector mode: ixge, i40e, ...

s/ixge/ixgbe/?

Quoted 1 line

<snip>

Quoted 4 lines

This will also make future support of vector mode on other architectures possible.

Fixes bc4a000f2f53 ("net/bnxt: implement SSE vector mode")

<snip>

drivers/net/bnxt/bnxt_rxr.c line 23

Quoted 2 lines
drivers/net/bnxt/bnxt_rxr.c +2 -0
+#error "bnxt: IEEE1588 is incompatiable with vector mode"
+#endif

s/incompatiable/incompatible/

This was this approach taken in the initial patch set to add bnxt vector mode support,
however based on feedback the current approach was used instead. That discussion can be found here:

      http://patches.dpdk.org/patch/53683/

If mark support could be treated as a receive offload, it would be possible to choose
the non-vector receive handler based on whether mark is enabled. Adding a kvargs option to disable vector mode might be another possibility. Otherwise, a build-time
configuration option does seem to be useful.

LGTM... with the above:

Regards,
Lance

#2 lore raw
Quoted 36 lines · On Thu, 5 Mar 2020 15:10:48 -0500 Lance Richardson…

Hi Stephen,

Quoted 1 line · Stephen Hemminger

<snip>

Quoted 3 lines

Make the configuration use the same as other drivers with vector mode: ixge, i40e, ...

s/ixge/ixgbe/?

Quoted 1 line

<snip>

Quoted 4 lines

This will also make future support of vector mode on other architectures possible.

Fixes bc4a000f2f53 ("net/bnxt: implement SSE vector mode")

<snip>

drivers/net/bnxt/bnxt_rxr.c line 23

Quoted 2 lines
drivers/net/bnxt/bnxt_rxr.c +2 -0
+#error "bnxt: IEEE1588 is incompatiable with vector mode"
+#endif  

s/incompatiable/incompatible/

This was this approach taken in the initial patch set to add bnxt vector mode support,
however based on feedback the current approach was used instead. That discussion can be found here:

      http://patches.dpdk.org/patch/53683/

If mark support could be treated as a receive offload, it would be possible to choose
the non-vector receive handler based on whether mark is enabled. Adding a kvargs option to disable vector mode might be another possibility. Otherwise, a build-time
configuration option does seem to be useful.

Quoted 3 lines

LGTM... with the above:

What ever solution must be consistent across all drivers.

#3 lore raw
Quoted 39 lines · Stephen Hemminger
Quoted 36 lines · On Thu, 5 Mar 2020 15:10:48 -0500 Lance Richardson…

Hi Stephen,

Quoted 1 line · Stephen Hemminger

<snip>

Quoted 3 lines

Make the configuration use the same as other drivers with vector mode: ixge, i40e, ...

s/ixge/ixgbe/?

Quoted 1 line

<snip>

Quoted 4 lines

This will also make future support of vector mode on other architectures possible.

Fixes bc4a000f2f53 ("net/bnxt: implement SSE vector mode")

<snip>

drivers/net/bnxt/bnxt_rxr.c line 23

Quoted 2 lines
drivers/net/bnxt/bnxt_rxr.c +2 -0
+#error "bnxt: IEEE1588 is incompatiable with vector mode"
+#endif  

s/incompatiable/incompatible/

This was this approach taken in the initial patch set to add bnxt vector mode support,
however based on feedback the current approach was used instead. That discussion can be found here:

      http://patches.dpdk.org/patch/53683/

If mark support could be treated as a receive offload, it would be possible to choose
the non-vector receive handler based on whether mark is enabled. Adding a kvargs option to disable vector mode might be another possibility. Otherwise, a build-time
configuration option does seem to be useful.

Quoted 6 lines
Quoted 3 lines

LGTM... with the above:

What ever solution must be consistent across all drivers.

I saw Ajit merged this patch to brcm tree, but I am not sure about it. We have already removed this compile time option from some PMDs, and driver tries to detect to use or not to use vectorization transparently.

This config is also a problem for the meson, which always sets the flag in a hardcoded way.

But also I am not sure about to need to enable/disable vectorization explicitly, this patch seems because of this need. As far as I remember in the past this type of runtime configuration rejected to not make driver configuration more complex.

#4 lore raw
Quoted 59 lines · Ferruh Yigit
Quoted 21 lines · Stephen Hemminger
Quoted 18 lines · On Thu, 5 Mar 2020 15:10:48 -0500 Lance Richardson…

Hi Stephen,

Quoted 1 line · Stephen Hemminger

<snip>

Quoted 3 lines

Make the configuration use the same as other drivers with vector mode: ixge, i40e, ...

s/ixge/ixgbe/?

Quoted 1 line

<snip>

Quoted 4 lines

This will also make future support of vector mode on other architectures possible.

Fixes bc4a000f2f53 ("net/bnxt: implement SSE vector mode")

<snip>

Quoted 8 lines
Quoted 8 lines

drivers/net/bnxt/bnxt_rxr.c line 23

Quoted 2 lines
drivers/net/bnxt/bnxt_rxr.c +2 -0
+#error "bnxt: IEEE1588 is incompatiable with vector mode"
+#endif

s/incompatiable/incompatible/

This was this approach taken in the initial patch set to add bnxt vector mode support,
however based on feedback the current approach was used instead. That

discussion

Quoted 6 lines
Quoted 6 lines

can be found here:

      http://patches.dpdk.org/patch/53683/

If mark support could be treated as a receive offload, it would be possible to choose
the non-vector receive handler based on whether mark is enabled. Adding

a kvargs

Quoted 10 lines
Quoted 7 lines

option to disable vector mode might be another possibility. Otherwise, a build-time
configuration option does seem to be useful.

LGTM... with the above:

What ever solution must be consistent across all drivers.

I saw Ajit merged this patch to brcm tree, but I am not sure about it. We have
already removed this compile time option from some PMDs, and driver tries to
detect to use or not to use vectorization transparently.

This config is also a problem for the meson, which always sets the flag in a
hardcoded way.

Quoted 7 lines

But also I am not sure about to need to enable/disable vectorization explicitly,
this patch seems because of this need. As far as I remember in the past this
type of runtime configuration rejected to not make driver configuration more
complex.

Since we need a way to disable or enable vector mode.
May be a dev parameter could be used?
That way it would not interfere with the meson builds and also allow a user/application to set vector mode setting as desired.

#5 lore raw

31/03/2020 16:31, Ajit Khaparde:

Quoted 69 lines · Ferruh Yigit
Quoted 60 lines · Stephen Hemminger
Quoted 21 lines
Quoted 18 lines · On Thu, 5 Mar 2020 15:10:48 -0500 Lance Richardson…

Hi Stephen,

Quoted 1 line · Stephen Hemminger

<snip>

Quoted 3 lines

Make the configuration use the same as other drivers with vector mode: ixge, i40e, ...

s/ixge/ixgbe/?

Quoted 1 line

<snip>

Quoted 4 lines

This will also make future support of vector mode on other architectures possible.

Fixes bc4a000f2f53 ("net/bnxt: implement SSE vector mode")

<snip>

Quoted 8 lines
Quoted 8 lines

drivers/net/bnxt/bnxt_rxr.c line 23

Quoted 2 lines
drivers/net/bnxt/bnxt_rxr.c +2 -0
+#error "bnxt: IEEE1588 is incompatiable with vector mode"
+#endif

s/incompatiable/incompatible/

This was this approach taken in the initial patch set to add bnxt vector mode support,
however based on feedback the current approach was used instead. That

discussion

Quoted 6 lines
Quoted 6 lines

can be found here:

      http://patches.dpdk.org/patch/53683/

If mark support could be treated as a receive offload, it would be possible to choose
the non-vector receive handler based on whether mark is enabled. Adding

a kvargs

Quoted 10 lines
Quoted 7 lines

option to disable vector mode might be another possibility. Otherwise, a build-time
configuration option does seem to be useful.

LGTM... with the above:

What ever solution must be consistent across all drivers.

I saw Ajit merged this patch to brcm tree, but I am not sure about it. We have
already removed this compile time option from some PMDs, and driver tries to
detect to use or not to use vectorization transparently.

This config is also a problem for the meson, which always sets the flag in a
hardcoded way.

Quoted 7 lines

But also I am not sure about to need to enable/disable vectorization explicitly,
this patch seems because of this need. As far as I remember in the past this
type of runtime configuration rejected to not make driver configuration more
complex.

Quoted 1 line

Since we need a way to disable or enable vector mode.

Why do you need to disable vector optimization?
Is it for debugging?

Quoted 3 lines

May be a dev parameter could be used?
That way it would not interfere with the meson builds and also allow a user/application to set vector mode setting as desired.

Yes diasbling optimization could be done in devargs.

#6 lore raw
Quoted 70 lines · On Tue, 31 Mar 2020 17:43:55 +0200 Thomas Monjalon <[email protected]> wrote:

31/03/2020 16:31, Ajit Khaparde:

Quoted 69 lines · Ferruh Yigit
Quoted 67 lines · Stephen Hemminger
Quoted 21 lines
Quoted 18 lines · On Thu, 5 Mar 2020 15:10:48 -0500 Lance Richardson…

Hi Stephen,

Quoted 1 line · Stephen Hemminger

<snip>

Quoted 3 lines

Make the configuration use the same as other drivers with vector mode: ixge, i40e, ...

s/ixge/ixgbe/?

Quoted 1 line

<snip>

Quoted 4 lines

This will also make future support of vector mode on other architectures possible.

Fixes bc4a000f2f53 ("net/bnxt: implement SSE vector mode")

<snip>

Quoted 8 lines
Quoted 8 lines

drivers/net/bnxt/bnxt_rxr.c line 23

Quoted 2 lines
drivers/net/bnxt/bnxt_rxr.c +2 -0
+#error "bnxt: IEEE1588 is incompatiable with vector mode"
+#endif  

s/incompatiable/incompatible/

This was this approach taken in the initial patch set to add bnxt vector mode support,
however based on feedback the current approach was used instead. That

discussion

Quoted 6 lines
Quoted 6 lines

can be found here:

      http://patches.dpdk.org/patch/53683/

If mark support could be treated as a receive offload, it would be possible to choose
the non-vector receive handler based on whether mark is enabled. Adding

a kvargs

Quoted 10 lines
Quoted 7 lines

option to disable vector mode might be another possibility. Otherwise, a build-time
configuration option does seem to be useful.

LGTM... with the above:

What ever solution must be consistent across all drivers.

I saw Ajit merged this patch to brcm tree, but I am not sure about it. We have
already removed this compile time option from some PMDs, and driver tries to
detect to use or not to use vectorization transparently.

This config is also a problem for the meson, which always sets the flag in a
hardcoded way.

But also I am not sure about to need to enable/disable vectorization explicitly,
this patch seems because of this need. As far as I remember in the past this
type of runtime configuration rejected to not make driver configuration more
complex.

Quoted 4 lines
Quoted 1 line

Since we need a way to disable or enable vector mode.

Why do you need to disable vector optimization?
Is it for debugging?

The rte_flow mark operation does not work with the vector optimization.

The choice to use vector mode is done by the driver earlier in the initialization process, and then when application programs rte_flow it has a problem; the flow create would fail.

#7 lore raw

31/03/2020 17:58, Stephen Hemminger:

Quoted 24 lines
Quoted 22 lines · On Tue, 31 Mar 2020 17:43:55 +0200 Thomas Monjalon <[email protected]> wrote:

31/03/2020 16:31, Ajit Khaparde:

Quoted 20 lines · Ferruh Yigit
Quoted 10 lines

I saw Ajit merged this patch to brcm tree, but I am not sure about it. We have
already removed this compile time option from some PMDs, and driver tries to
detect to use or not to use vectorization transparently.

This config is also a problem for the meson, which always sets the flag in a
hardcoded way.

Quoted 7 lines

But also I am not sure about to need to enable/disable vectorization explicitly,
this patch seems because of this need. As far as I remember in the past this
type of runtime configuration rejected to not make driver configuration more
complex.

Since we need a way to disable or enable vector mode.

Quoted 8 lines
Quoted 2 lines

Why do you need to disable vector optimization?
Is it for debugging?

The rte_flow mark operation does not work with the vector optimization.

The choice to use vector mode is done by the driver earlier in the initialization process, and then when application programs rte_flow it has a problem; the flow create would fail.

Could we imagine switching the Rx implementation dynamically between 2 bursts?

#8 lore raw
In reply to Stephen Hemminger
Quoted 8 lines · Stephen Hemminger

The bnxt driver has vector mode but it has limitations. For example, rte_flow mark won't work in vector mode. For this reason the user should be able to disable vector mode as part of the config.

Make the configuration use the same as other drivers with vector mode: ixge, i40e, ...

Quoted 8 lines

This will also make future support of vector mode on other architectures possible.

Fixes bc4a000f2f53 ("net/bnxt: implement SSE vector mode")

Patch applied to dpdk-next-net-brcm. Thanks

Quoted 127 lines
---
 config/common_base             |  1 +
 drivers/net/bnxt/Makefile      |  2 +-
 drivers/net/bnxt/bnxt_ethdev.c | 10 +++-------
 drivers/net/bnxt/bnxt_ring.c   |  4 ++--
 drivers/net/bnxt/bnxt_rxr.c    |  3 +++
 5 files changed, 10 insertions(+), 10 deletions(-)
config/common_base +1 -0
index 7ca2f28b19c8..3c719c7cbfc3 100644
@@ -219,6 +219,7 @@ CONFIG_RTE_LIBRTE_BNX2X_DEBUG_PERIODIC=n
 # Compile burst-oriented Broadcom BNXT PMD driver
 #
 CONFIG_RTE_LIBRTE_BNXT_PMD=y
+CONFIG_RTE_LIBRTE_BNXT_INC_VECTOR=y
 
 #
 # Compile burst-oriented Chelsio Terminator (CXGBE) PMD
drivers/net/bnxt/Makefile +1 -1
index b77532b8179b..85d6df6b5aa0 100644
@@ -40,7 +40,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_BNXT_PMD) += bnxt_irq.c
 SRCS-$(CONFIG_RTE_LIBRTE_BNXT_PMD) += bnxt_util.c
 SRCS-$(CONFIG_RTE_LIBRTE_BNXT_PMD) += rte_pmd_bnxt.c
 ifeq ($(CONFIG_RTE_ARCH_X86), y)
-SRCS-$(CONFIG_RTE_LIBRTE_BNXT_PMD) += bnxt_rxtx_vec_sse.c
+SRCS-$(CONFIG_RTE_LIBRTE_BNXT_INC_VECTOR) += bnxt_rxtx_vec_sse.c
 endif
 
 #
drivers/net/bnxt/bnxt_ethdev.c +2 -6
index 18aa313fd8c1..701de33784b2 100644
@@ -757,8 +757,7 @@ bnxt_receive_function(struct rte_eth_dev *eth_dev)
 {
        struct bnxt *bp = eth_dev->data->dev_private;
 
-#ifdef RTE_ARCH_X86
-#ifndef RTE_LIBRTE_IEEE1588
+#ifdef RTE_BNXT_INC_VECTOR
        /*
         * Vector mode receive can be enabled only if scatter rx is not
         * in use and rx offloads are limited to VLAN stripping and
@@ -787,7 +786,6 @@ bnxt_receive_function(struct rte_eth_dev *eth_dev)
                    eth_dev->data->port_id,
                    eth_dev->data->scattered_rx,
                    eth_dev->data->dev_conf.rxmode.offloads);
-#endif
 #endif
        bp->flags &= ~BNXT_FLAG_RX_VECTOR_PKT_MODE;
        return bnxt_recv_pkts;
@@ -796,8 +794,7 @@ bnxt_receive_function(struct rte_eth_dev *eth_dev)
 static eth_tx_burst_t
 bnxt_transmit_function(__rte_unused struct rte_eth_dev *eth_dev)
 {
-#ifdef RTE_ARCH_X86
-#ifndef RTE_LIBRTE_IEEE1588
+#ifdef RTE_BNXT_INC_VECTOR
        /*
         * Vector mode transmit can be enabled only if not using scatter rx
         * or tx offloads.
@@ -815,7 +812,6 @@ bnxt_transmit_function(__rte_unused struct rte_eth_dev
 *eth_dev)
                    eth_dev->data->port_id,
                    eth_dev->data->scattered_rx,
                    eth_dev->data->dev_conf.txmode.offloads);
-#endif
 #endif
        return bnxt_xmit_pkts;

}

drivers/net/bnxt/bnxt_ethdev.c +1 -1
@@ -2234,7 +2230,7 @@ int bnxt_mtu_set_op(struct rte_eth_dev *eth_dev,
 uint16_t new_mtu)
        new_pkt_size = new_mtu + RTE_ETHER_HDR_LEN + RTE_ETHER_CRC_LEN +
                       VLAN_TAG_SIZE * BNXT_NUM_VLANS;
 
-#ifdef RTE_ARCH_X86
+#ifdef RTE_BNXT_INC_VECTOR
        /*
         * If vector-mode tx/rx is active, disallow any MTU change that

would
* require scattered receive support.

drivers/net/bnxt/bnxt_ring.c +2 -2
index d6e4e8a28138..9a4a0f21418d 100644
@@ -608,7 +608,7 @@ int bnxt_alloc_hwrm_rx_ring(struct bnxt *bp, int
 queue_index)
                bnxt_db_write(&rxr->ag_db, rxr->ag_prod);
        }
        rxq->index = queue_index;
-#ifdef RTE_ARCH_X86
+#ifdef RTE_BNXT_INC_VECTOR
        bnxt_rxq_vec_setup(rxq);
 #endif
@@ -713,7 +713,7 @@ int bnxt_alloc_hwrm_rings(struct bnxt *bp)
                bnxt_db_write(&rxr->rx_db, rxr->rx_prod);
                bnxt_db_write(&rxr->ag_db, rxr->ag_prod);
                rxq->index = i;
-#ifdef RTE_ARCH_X86
+#ifdef RTE_BNXT_INC_VECTOR
                bnxt_rxq_vec_setup(rxq);
 #endif
        }
drivers/net/bnxt/bnxt_rxr.c +3 -1
index bef9720f590c..f9087bea89b9 100644
@@ -18,6 +18,9 @@
 #include "hsi_struct_def_dpdk.h"
 #ifdef RTE_LIBRTE_IEEE1588
 #include "bnxt_hwrm.h"
+#ifdef RTE_BNXT_INC_VECTOR
+#error "bnxt: IEEE1588 is incompatiable with vector mode"
+#endif
 #endif
 
 /*
--

2.20.1

#9 lore raw

PATCH 1/1 net/bnxt: disable vector mode if flow mark is used

In reply to Stephen Hemminger

Since vector mode can't be used with flow mark actions. The choice of vector or non-vector mode is done dynamically earlier in the initialization process (config) and the user application has no direct control over the selection.

Therefore the best resolution to the conflict is for the driver to disable vector mode if it needs to.

Fixes 94eb699bc82e ("net/bnxt: support flow mark action")
---
 drivers/net/bnxt/bnxt_flow.c | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)
drivers/net/bnxt/bnxt_flow.c +5 -8
index 44734272fbea..7715a7b45173 100644
@@ -18,6 +18,7 @@
 #include "bnxt_hwrm.h"
 #include "bnxt_ring.h"
 #include "bnxt_rxq.h"
+#include "bnxt_rxr.h"
 #include "bnxt_vnic.h"
 #include "hsi_struct_def_dpdk.h"
 
@@ -1405,14 +1406,10 @@ bnxt_validate_and_parse_flow(struct rte_eth_dev *dev,
 	case RTE_FLOW_ACTION_TYPE_MARK:
 		if (bp->flags & BNXT_FLAG_RX_VECTOR_PKT_MODE) {
 			PMD_DRV_LOG(DEBUG,
-				    "Disable vector processing for mark\n");
-			rte_flow_error_set(error,
-					   ENOTSUP,
-					   RTE_FLOW_ERROR_TYPE_ACTION,
-					   act,
-					   "Disable vector processing for mark");
-			rc = -rte_errno;
-			goto ret;
+				    "Disabling vector processing for mark\n");
+
+			bp->flags &= ~BNXT_FLAG_RX_VECTOR_PKT_MODE;
+			dev->rx_pkt_burst = &bnxt_recv_pkts;
 		}
 
 		if (bp->mark_table == NULL) {
Signature · 1 line
2.20.1