p4c-dpdk can compile PNA programs into spec files. When I try to load it into the bf_switchd target, and run bfrt.<program_name>.enable, I get "Invalid struct field size." error. I think this is related to some of the PNA metadata fields are 3 bits like bit<3> pna_pre_input_metadata_pass

However, when I hack the .spec file to make them bit<8>, I got "Pipeline build error."

Could you supply some sample programs that can load and run on this DPDK target? Maybe the l3_simple_demo example that is mentioned in the tools folder.

Hello, we're trying to run a simple main.p4 using p4-dpdk-target: main.p4:

#include <core.p4>
#include <psa.p4>

struct empty_metadata_t {}
struct headers_t {}

struct local_metadata_t {}

parser packet_parser(
  packet_in packet,
  out headers_t headers,
  inout local_metadata_t local_metadata,
  in psa_ingress_parser_input_metadata_t standard_metadata,
  in empty_metadata_t resub_meta,
  in empty_metadata_t recirc_meta) {
    state start {
        transition accept;
    }
}

control packet_deparser(
    packet_out packet,
    out empty_metadata_t clone_i2e_meta,
    out empty_metadata_t resubmit_meta,
    out empty_metadata_t normal_meta,
    inout headers_t headers,
    in local_metadata_t local_metadata,
    in psa_ingress_output_metadata_t istd) {
    apply {
    }
}

control ingress(
  inout headers_t headers,
  inout local_metadata_t local_metadata1,
  in psa_ingress_input_metadata_t piim,
  inout psa_ingress_output_metadata_t piom) {
    apply {
      if (piim.ingress_port == (PortId_t)0) {
        piom.egress_port = (PortId_t)1;
      } else {
        piom.egress_port = (PortId_t)0;
      }
    }
}

control egress(inout headers_t headers, inout local_metadata_t local_metadata, in psa_egress_input_metadata_t istd, inout psa_egress_output_metadata_t ostd) {
    apply {
    }
}

parser egress_parser(packet_in buffer, out headers_t headers, inout local_metadata_t local_metadata, in psa_egress_parser_input_metadata_t istd, in empty_metadata_t normal_meta, in empty_metadata_t clone_i2e_meta, in empty_metadata_t clone_e2e_meta) {
    state start {
        transition accept;
    }
}

control egress_deparser(packet_out packet, out empty_metadata_t clone_e2e_meta, out empty_metadata_t recirculate_meta, inout headers_t headers, in local_metadata_t local_metadata, in psa_egress_output_metadata_t istd, in psa_egress_deparser_input_metadata_t edstd) {
    apply {
    }
}

IngressPipeline(packet_parser(), ingress(), packet_deparser()) ip;

EgressPipeline(egress_parser(), egress(), egress_deparser()) ep;

PSA_Switch(ip, PacketReplicationEngine(), ep, BufferingQueueingEngine()) main;

We downloaded and installed p4c.

we installed p4-dpdk-target using @jafingerhut 's script and @Yi-Tseng 's script, (apparently one of the scripts did not work properly, but we do not know why.

we compile with:

p4c-dpdk main.p4 \
    -o main.spec \
    --arch psa \
    --bf-rt-schema main.bfrt.json \
    --context context.json \
    --p4runtime-files main.pb.txt \
    --p4runtime-format text

we try to run bf_switched with the following:

# Basic environment variables we need
export SDE=$HOME/sde
export SDE_INSTALL=$SDE/install
export LD_LIBRARY_PATH=$SDE_INSTALL/lib:$SDE_INSTALL/lib64:$SDE_INSTALL/lib/x86_64-linux-gnu/:/usr/local/lib64:/usr/local/lib
# We need to put SDE executable path before the system one since the bfrt_python
# needs to use Python libraries from it and use the Python executable from SDE
export PATH=$SDE_INSTALL/bin:$PATH
hash -r

# For security reason, the PATH and LD_LIBRARY_PATH won't pass to root user even if we use "sudo -E"
# We must pass them in sudo to make sure it is correct.
sudo -E PATH=$PATH LD_LIBRARY_PATH=$LD_LIBRARY_PATH $SDE_INSTALL/bin/bf_switchd --install-dir $SDE_INSTALL --conf switch_config.json

but here we encounter an error :

bf_switchd: system services initialized
bf_switchd: loading conf_file switch_config.json...
bf_switchd: processing device configuration...
Configuration for dev_id 0
  Family        : dpdk
bf_switchd: processing P4 configuration...
P4 profile for dev_id 0
P4 EAL args dummy -n 2 -c 0xC
Debug CLI disable
num mempool objs 1
  mempool_name: MEMPOOL0
  buffer_size:  2304
  pool_size:    1024
  cache_size:   256
  numa_node:    0
num fixed functions 0
num P4 programs 1
  p4_name: l1switch
  bfrt_config: /home/p4project/p4-dpdk-target-notes/l1switch/tdi.json
  port_config: /home/p4project/p4-dpdk-target-notes/l1switch/ports.json
  p4_pipeline_name: pipe
  core_id: 1
  numa_node: 0
    context: /home/p4project/p4-dpdk-target-notes/l1switch/context.json
    config: /home/p4project/p4-dpdk-target-notes/l1switch/main.spec
  Pipes in scope [0 1 2 3 ]
  Timer values []
  Mirror Config
    n_slots: 4
    n_sessions: 64
    fast_clone: 0
  diag:
  accton diag:
  non_default_port_ppgs: 0
  SAI default initialize: 1
Operational mode set to DPDK MODEL
Initialized the device types using PAL handler registration
STUB:port_mgr_platform_init DPDK
bf_switchd: drivers initialized
EAL: Detected CPU lcores: 16
EAL: Detected NUMA nodes: 2
EAL: Detected shared linkage of DPDK
EAL: Multi-process socket /var/run/dpdk/rte/mp_socket
EAL: Selected IOVA mode 'PA'
EAL: VFIO support initialized
EAL: DPDK is running on a NUMA system, but is compiled without NUMA support.
EAL: This will have adverse consequences for performance and usability.
EAL: Please use --legacy-mem option, or recompile with NUMA support.
EAL: Device 0000:01:00.0 is not NUMA-aware
EAL: Device 0000:01:00.1 is not NUMA-aware
EAL: Device 0000:03:00.0 is not NUMA-aware
EAL: Device 0000:07:00.0 is not NUMA-aware
TELEMETRY: No legacy callbacks, legacy socket not created
RING: Cannot reserve memory
Error: Thread initialization failed (-1)
2022-12-27 16:19:13.789453 BF_PORT ERROR - parse_port_config_json:539: Could not open configuration file: /home/p4project/p4-dpdk-target-notes/l1switch/ports.json.

2022-12-27 16:19:13.789498 BF_PORT ERROR - port_mgr_config_import : Error in parsing the port config file
2022-12-27 16:19:13.789549 BF_DVM ERROR - Device add handling failed for dev 0, sts No system resources (2), Client port_mgr
2022-12-27 16:19:13.789594 BF_DVM ERROR - Device add failed for dev 0, sts No system resources (2)
ERROR: bf_device_add failed(2) for dev_id 0
ERROR: bf_switchd_device_add failed(2) for dev_id 0

bf_switchd: initialized 0 devices
bf_switchd_switchsai_lib_init
 starting SAI
bf_switchd: spawning cli server thread
bf_switchd: spawning driver shell
bf_switchd: server started - listening on port 9999

        ********************************************
        *      WARNING: Authorised Access Only     *
        ********************************************


bfshell>

We tried looking up documentation for ports.json or switch_confiog.json, but alas we could not find any.

any help regarding how should the ports.json and switch_config.json should contain would be helpful. please know that we are still learning and don't completely understand things that may be obvious to the more experienced.

thank you

This commits removes AVX512 support and enables generic CPU options for the DPDK build. The IPDK CI builds on Azure VMs which support AVX512. However, since Virtualbox doesn't support this (see [1] and [2]), we should skip this for p4-dpdk-target when building DPDK.

Fixes: https://github.com/ipdk-io/ipdk/issues/118

[1] https://forums.virtualbox.org/viewtopic.php?t=86047 [2] https://forums.virtualbox.org/viewtopic.php?f=7&t=87131

Signed-off-by: Kyle Mestery [email protected]

Hello, we're trying to build and install but we're experiencing a lot of errors,

we'll be more than happy to get a list of repositories we need to clone and build and prerequisites we need to install. If there is a complete setup.sh somewhere it would be great. thank you for your time :)

Hello, Has anyone gone through the situation I'm going through to link a physical NIC to p4-dpdk-target . I already have no idea what to do to successfully execute the binding. The problem only occurs when using BFRT, I tested it with the DPDK examples and the card links normally. The error that appears when trying to attach the physical card to bf_switchd is how it cannot read the ports.json file, but this only happens with physical cards.

Log syslog. Oct 19 23:48:13 master-node dummy[942929]: EAL: Probe PCI driver: net_nfp_pf (19ee:4000) device: 0000:41:00.0 (socket 1) Oct 19 23:48:13 master-node kernel: [34440.907257] igb_uio 0000:41:00.0: uio device registered with irq 87 MInha dúvida ao efetuar os teste descritos é:Oct 19 23:58:07 master-node dummy[957429]: 2022-10-19 23:58:07.648645 ERROR (BF_PORT:(file=null):(func=null):0) - port_config_json_parse_ports:507: Failed to add port LINK0 Oct 19 23:59:07 master-node dummy[957429]: 2022-10-19 23:58:07.648667 ERROR (BF_PORT:(file=null):(func=null):0) - parse_port_config_json:586: Failed to parse ports from Port Config JSON Oct 19 23:59:07 master-node dummy[957429]: 2022-10-19 23:58:07.648705 ERROR (BF_PORT:(file=null):(func=null):0) - port_mgr_config_import : Error in parsing the port config file Oct 19 23:59:07 master-node dummy[957429]: 2022-10-19 23:58:07.648744 ERROR (BF_DVM:(file=null):(func=null):0) - Device add handling failed for dev 0, sts No system resources (2), Client port_mgr Oct 19 23:59:07 master-node dummy[957429]: 2022-10-19 23:58:07.648764 ERROR (BF_DVM:(file=null):(func=null):0) - Device add failed for dev 0, sts No system resources (2)

Log bfshell LINK CREATE: Dev:0000:41:00.0 Args: dev_hotplug_enabled: 0 2022-10-19 09:54:00.196125 BF_LLD ERROR - Creation of Link Port LINK0 failed

2022-10-19 09:54:00.196285 BF_PORT ERROR - port_config_json_parse_ports:507: Failed to add port LINK0 2022-10-19 09:54:00.196322 BF_PORT ERROR - parse_port_config_json:586: Failed to parse ports from Port Config JSON 2022-10-19 09:54:00.196377 BF_PORT ERROR - port_mgr_config_import : Error in parsing the port config file 2022-10-19 09:54:00.196436 BF_DVM ERROR - Device add handling failed for dev 0, sts No system resources (2), Client port_mgr 2022-10-19 09:54:00.196466 BF_DVM ERROR - Device add failed for dev 0, sts No system resources (2) ERROR: bf_device_add failed(2) for dev_id 0 ERROR: bf_switchd_device_add failed(2) for dev_id 0

bf_switchd: initialized 0 devices bf_switchd_switchsai_lib_init starting SAI bf_switchd: spawning cli server thread bf_switchd: spawning driver shell bf_switchd: server started - listening on port 9999

    ********************************************
    *      WARNING: Authorised Access Only     *
    ********************************************

Every help is welcome, Thanks.

Hello again,

I am trying to run the example that is given. I have successfully compiled it according to the instructions and created a conf-file following the examples given. However, after running the switch I get the following errors:

2022-07-27 20:48:11.005078 BF_BFRT ERROR - TdiInfo:150 Error creating Table:$PORT_FP_IDX_INFO
2022-07-27 20:48:11.005909 BF_BFRT ERROR - TdiInfo:150 Error creating Table:$PORT_HDL_INFO
2022-07-27 20:48:11.006688 BF_BFRT ERROR - TdiInfo:150 Error creating Table:$PORT_STR_INFO
2022-07-27 20:48:11.007621 BF_BFRT ERROR - parseContextJson:184 context Table object not found for "$PORT"
2022-07-27 20:48:11.008474 BF_BFRT ERROR - parseContextJson:184 context Table object not found for "$PORT_STAT"
2022-07-27 20:48:11.009262 BF_BFRT ERROR - parseContextJson:184 context Table object not found for "ip.ingress.ipv4_host"
2022-07-27 20:48:11.010026 BF_BFRT ERROR - parseContextJson:184 context Table object not found for "ip.ingress.ipv4_lpm"
2022-07-27 20:48:11.010804 BF_BFRT ERROR - parseContextJson:184 context Table object not found for "ip.ingress.l2_fwd"
2022-07-27 20:48:11.011562 BF_BFRT ERROR - parseContextJson:184 context Table object not found for "ip.ingress.mymac"

Also when trying to enable the p4 program I get the following

------> bfrt.simple_l2l3_lpm.enable()
2022-07-27 20:51:16.399251 BF_PIPE ERROR - Number of ports added to DPDK Pipeline pipe in input direction should be a power of 2
2022-07-27 20:51:16.400320 BF_PIPE ERROR - Failed to Build Pipeline pipe
2022-07-27 20:51:16.401184 BF_BFRT ERROR - bf_rt_enable_pipeline:120 failed

I am not sure what exactly is the problem and if I am missing some file, or maybe change something?

Hello,

I have been trying to run some examples in p4-dpdk and I am not able to understand how you bind the interfaces to the switch using a port-config file.

I see that in the bf_switchd config file there exists a port-config file, but I couldn't find any example of such a port-config file somewhere in the repository.

Added support to configure the timer values from the conf file. Added a new timer array in conf file to hold the timer values. CT_timer_unit.txt CT_timer_unit_test.txt

The example json config for the bf_switchd reference application has unnecessary commas on lines 6 and 29 that cause the program to fail to process the file. In the readme the example json given is

    {
        "chip_list": [
            {
                "id": "asic-0",
                "chip_family": "dpdk",
                "instance": 0,
            }
        ],
        "instance": 0,
        "p4_devices": [
            {
                "device-id": 0,
                "p4_programs": [
                    {
                        "program-name": "<P4 program name>",
                        "cpu_numa_node": "0",
                        "bfrt-config": "share/dpdk/<P4 program name>/bf-rt.json",
                        "port-config": "share/dpdk/<P4 program name>/ports_topology.json",
                        "eal-args": "dummy -n 4 -c 3",
                        "p4_pipelines": [
                            {
                                "p4_pipeline_name": "pipe",
                                "context": "share/dpdk/<P4 program name>/pipe/context.json",
                                "config": "share/dpdk/<P4 program name>/pipe/<P4 program name>.spec",
                                "path": "share/dpdk/<P4 program name>"
                            }
                        ]
                    }
                ],
            }
        ]
    }

When it should be

    {
        "chip_list": [
            {
                "id": "asic-0",
                "chip_family": "dpdk",
                "instance": 0
            }
        ],
        "instance": 0,
        "p4_devices": [
            {
                "device-id": 0,
                "p4_programs": [
                    {
                        "program-name": "<P4 program name>",
                        "cpu_numa_node": "0",
                        "bfrt-config": "share/dpdk/<P4 program name>/bf-rt.json",
                        "port-config": "share/dpdk/<P4 program name>/ports_topology.json",
                        "eal-args": "dummy -n 4 -c 3",
                        "p4_pipelines": [
                            {
                                "p4_pipeline_name": "pipe",
                                "context": "share/dpdk/<P4 program name>/pipe/context.json",
                                "config": "share/dpdk/<P4 program name>/pipe/<P4 program name>.spec",
                                "path": "share/dpdk/<P4 program name>"
                            }
                        ]
                    }
                ]
            }
        ]
    }

Note the removed commas after "instance": 0 and the second to last ].

Here is the structure of submodules of this repository, and the submodules of those repositories, as of 2021-Dec-21:

+ p4-dpdk-target
  + target-syslibs @ f646177
    + gperftools @ f7c6fb6
    + zlog @ 876099f
  + target-utils @ 61969dd
    + cJSON @ d348621
    + cpython @ 3d8993a
    + target-syslibs @ 580b5e5    Repeat of target-syslibs above.  Why?
    + tommyds @ 5ada275
    + xxHash @ 94e5f23
  + tdi @ 21da36e
    + target-syslibs @ 4f765f5    Repeat of target-syslibs above.  Why?
    + target-utils @ 243a99e      Repeat of target-utils above.    Why?

Is it intentional that so many submodules are repeated? If so, why?

The -j48 in the ninja command line requires at least about 8 GB of RAM on the system in order to avoid random process crashes due to the Linux kernel OOM process killer: https://github.com/p4lang/p4-dpdk-target/blob/0baaa0cd296008bb307cb750cd2419036dd804bc/src/lld/dpdk/Makefile.am#L5

https://www.kernel.org/doc/gorman/html/understand/understand016.html

It also does not help builds go any faster on a system with relatively few CPU cores, e.g. 4 to 8.

Consider either hard-coding a smaller number (perhaps -j8 ?) or perhaps make it configurable via an environment variable.

Hello, I want to run a simple program, however I get a ERROR: bf_device_add failed(2) for dev_id 0 ERROR: bf_switchd_device_add failed(2) for dev_id 0 erros. I suspect it is because ports.json.

I have searched the web for an example but alas, couldn't find any.

I would love a detailed explanation \ guide \ example that works, for how to write my ports.json..

thank you for your time !

Error:

023-01-08 13:29:26.912067 BF_PORT ERROR - port_config_json_parse_ports:507: Failed to add port ens33
2023-01-08 13:29:26.912226 BF_PORT ERROR - parse_port_config_json:586: Failed to parse ports from Port Config JSON
2023-01-08 13:29:26.912395 BF_PORT ERROR - port_mgr_config_import : Error in parsing the port config file
2023-01-08 13:29:26.912580 BF_DVM ERROR - Device add handling failed for dev 0, sts No system resources (2), Client port_mgr 
2023-01-08 13:29:26.912706 BF_DVM ERROR - Device add failed for dev 0, sts No system resources (2)
ERROR: bf_device_add failed(2) for dev_id 0
ERROR: bf_switchd_device_add failed(2) for dev_id 0

Hi, after changing the .json to:

Ports: `{ "ports": [ { "dev_port": 3, "port_name": "ens3", "mempool_name": "MEMPOOL0", "pipe_name": "pipe", "pipe_in":"pipe", "pipe_out":"pipe", "port_dir": "default", "port_in_id": 0, "port_out_id": 0, "port_type": "tap", "tap_port_attributes": { "mtu": 1500 } }, { "dev_port": 5, "port_name": "ens5", "mempool_name": "MEMPOOL0", "pipe_name": "pipe", "port_dir": "default", "port_in_id": 1, "pipe_in":"pipe", "pipe_out":"pipe", "port_out_id": 1, "port_type": "tap", "tap_port_attributes": { "mtu": 1500 } }

]

}`

switch config:

{ "chip_list": [ { "id": "asic-0", "chip_family": "dpdk", "instance": 0 } ], "instance": 0, "p4_devices": [ { "device-id": 0, "eal-args": "-n 1 -c 0x1", "mempools": [ { "name": "MEMPOOL0", "buffer_size": 2304, "pool_size": 1024, "cache_size": 256, "numa_node": 0 } ], "p4_programs": [ { "program-name": "l1switch", "bfrt-config": "/home/p4project/l1switch/tdi.json", "port-config": "/home/p4project/l1switch/ports.json", "p4_pipelines": [ { "p4_pipeline_name": "pipe", "core_id": 1, "numa_node": 0, "context": "/home/p4project/l1switch/context.json", "config": "/home/p4project/l1switch/main.spec", "pipe_scope": [ 0, 1, 2, 3 ], "path": "/home/p4project/l1switch/" } ] } ] } ] }

build.sh :

`#!/bin/bash p4c-dpdk main.p4
-o main.spec
--arch psa
--bf-rt-schema main.bfrt.json
--context context.json
--p4runtime-files main.pb.txt
--p4runtime-format text
--tdi tdi.json

Basic environment variables we need

export SDE=$HOME/sde export SDE_INSTALL=$SDE/install export LD_LIBRARY_PATH=$SDE_INSTALL/lib:$SDE_INSTALL/lib64:$SDE_INSTALL/lib/x86_64-linux-gnu/:/usr/local/lib64:/usr/local/lib

We need to put SDE executable path before the system one since the bfrt_python

needs to use Python libraries from it and use the Python executable from SDE

export PATH=$SDE_INSTALL/bin:$PATH hash -r

For security reason, the PATH and LD_LIBRARY_PATH won't pass to root user even if we use "sudo -E"

We must pass them in sudo to make sure it is correct.

sudo -E PATH=$PATH LD_LIBRARY_PATH=$LD_LIBRARY_PATH $SDE_INSTALL/bin/bf_switchd --install-dir $SDE_INSTALL --conf switch_config.json `

but sill see issues:

`p4project@p4project-S5520HC:~/l1switch$ ./build.sh Install dir: /home/p4project/sde/install (0x564c18724c20) bf_switchd: system services initialized bf_switchd: loading conf_file switch_config.json... bf_switchd: processing device configuration... Configuration for dev_id 0 Family : dpdk bf_switchd: processing P4 configuration... P4 profile for dev_id 0 P4 EAL args -n 1 -c 0x1 Debug CLI disable num mempool objs 1 mempool_name: MEMPOOL0 buffer_size: 2304 pool_size: 1024 cache_size: 256 numa_node: 0 num fixed functions 0 num P4 programs 1 p4_name: l1switch bfrt_config: /home/p4project/l1switch/tdi.json port_config: /home/p4project/l1switch/ports.json p4_pipeline_name: pipe core_id: 1 numa_node: 0 context: /home/p4project/l1switch/context.json config: /home/p4project/l1switch/main.spec Pipes in scope [0 1 2 3 ] Timer values [] Mirror Config n_slots: 4 n_sessions: 64 fast_clone: 0 diag: accton diag: non_default_port_ppgs: 0 SAI default initialize: 1 Operational mode set to DPDK MODEL Initialized the device types using PAL handler registration STUB:port_mgr_platform_init DPDK bf_switchd: drivers initialized EAL: Detected CPU lcores: 16 EAL: Detected NUMA nodes: 2 EAL: Detected shared linkage of DPDK EAL: Multi-process socket /var/run/dpdk/rte/mp_socket EAL: Selected IOVA mode 'PA' EAL: VFIO support initialized EAL: DPDK is running on a NUMA system, but is compiled without NUMA support. EAL: This will have adverse consequences for performance and usability. EAL: Please use --legacy-mem option, or recompile with NUMA support. EAL: Device 0000:01:00.0 is not NUMA-aware EAL: Device 0000:01:00.1 is not NUMA-aware EAL: Device 0000:03:00.0 is not NUMA-aware EAL: Using IOMMU type 8 (No-IOMMU) EAL: Ignore mapping IO port bar(2) EAL: Probe PCI driver: net_ixgbe (8086:10fb) device: 0000:03:00.0 (socket -1) EAL: Device 0000:07:00.0 is not NUMA-aware EAL: Ignore mapping IO port bar(2) EAL: Probe PCI driver: net_ixgbe (8086:10fb) device: 0000:07:00.0 (socket -1) TELEMETRY: No legacy callbacks, legacy socket not created

bf_switchd: dev_id 0 initialized

bf_switchd: initialized 1 devices bf_switchd_switchsai_lib_init starting SAI bf_switchd: spawning cli server thread bf_switchd: spawning driver shell bf_switchd: server started - listening on port 9999

    ********************************************
    *      WARNING: Authorised Access Only     *
    ********************************************

bfshell> bfrt_python cwd : /home/p4project/l1switch

We've found 1 p4 programs for device 0: l1switch Creating tree for dev 0 and program l1switch

Devices found : [0] /home/p4project/sde/install/lib/python3.10/site-packages/IPython/core/history.py:226: UserWarning: IPython History requires SQLite, your history will not be saved warn("IPython History requires SQLite, your history will not be saved") Python 3.10.8 (aaaf517424:aaaf517424, Dec 27 2022, 12:59:27) [GCC 11.3.0] Type 'copyright', 'credits' or 'license' for more information IPython 7.31.1 -- An enhanced Interactive Python. Type '?' for help.

In [1]: bfrt ------> bfrt() Available symbols: dump - Command enable - Command info - Command l1switch - Node port - Node

bfrt> l1switch.enable ----> l1switch.enable() 2022-12-28 18:27:04.276807 BF_PIPE ERROR - error :thread pipeline enable 2022-12-28 18:27:04.276956 BF_PIPE ERROR - Failed to Build Pipeline pipe 2022-12-28 18:27:04.277010 BF_BFRT ERROR - bf_rt_enable_pipeline:120 failed `

What should we write in switch_config.json and ports.json? We would appreciate a link to a full tutorial \ guide on how to write these .json files.

Thank you

Hello, Does anyone know if there is any way to pass operating parameters (mtu, bsize ...) to the network card to be linked in the BF_Switchd through the .json file for port configuration? If yes, how? Or if there is any other way to do this with bf_Switchd running. Note: I use an Intel 82599ES 10-Gigabit SFI/SFP+ network card linked in 'link' mode. My port configuration is the one below, I believe I can assign these parameters in "link_port_attributes", but how/which ones are accepted?
{ "ports": [ { "dev_port": 0, "port_name": "LINK0", "mempool_name": "MEMPOOL0", "port_dir": "default", "port_in_id": 0, "pipe_in": "pipe", "port_out_id": 0, "pipe_out": "pipe", "port_type": "link", "net_port" : 0, "link_port_attributes": { "pcie_bdf": "0000:b3:00.0" } } ]

Grateful.

Problem Statement

While using P4Runtime on p4-DPDK-target ovs I want to gather all counter values in single query. This functionality is available in behavioral model and described in P4Runtime spec/proto, which should be target agnostic.

1. I create CounterEntry ce = counter_entry["someCounter"]
2. I edit manually some counter entries, intentionally. Example: ce.index=10 ce.packet_count=101 ce.modify
3. If I want to see only single counter entry, data is visible ce?
4. But, If i want to see all entries in single query, data is not visible. Commands used: ce.index=None list(ce.read())

Steps to reproduce & Current output

P4 program with some counter (let's say 1024 in size, PACKETS type) ce = counter_entry["someCounter"] ce.index=10 ce.packet_count=101 ce.modify ce.index=11 ce.packet_count=2000 ce.modify ce.index=None list(ce.read())

output:

...
 counter_id: 317030940 ("someCounter")
index {
  index: 10
}
data {
  
}
,
 counter_id: 317030940 ("someCounter")
index {
  index: 11
}
data {
  
}
...

Expected output

...
 counter_id: 317030940 ("someCounter")
index {
  index: 10
}
data {
   packet_count: 101
}
,
 counter_id: 317030940 ("someCounter")
index {
  index: 11
}
data {
   packet_count: 2000
}
...

When loading a PNA program (like the PNA simple_lpm) with bf_pal_device_add(dev_id, &device_profile) we are seeing this error message:

2022-11-09 21:55:56.747712 BF_PIPE ERROR - Error -17 at line 23: Struct registration error.
.
2022-11-09 21:55:56.747830 BF_PIPE ERROR - Failed to Build Pipeline pipe
2022-11-09 21:55:56.747854 BF_DVM ERROR - Device add handling failed for dev 0, sts Unexpected error (15), Client pipe-mgr
2022-11-09 21:55:56.747871 BF_DVM ERROR - Device add failed for dev 0, sts Unexpected error (15)

The same build does work with PSA version of the pipeline.

Could be related to #16