[dpdk-dev] BCM57414 VF Mode: max_rx_em_flows value is inconsistent

[Christophe Grosse <christophe.grosse@6wind.com>]

Hello,

I am using DPDK 18.11.5 from dpdk-stable repository with a Broadcom
network adapter in VF mode :
Broadcom Limited BCM57414 NetXtreme-E 10Gb/25Gb RDMA Ethernet
Controller (rev 01)
# ./bnxtnvm   listdev
Broadcom P225p NetXtreme-E Dual-port 10Gb/25Gb Ethernet PCIe Adapter #1
Device Interface Name     : ntfp2
MACAddress                : 00-0A-F7-B6-E3-D1
PCI Device Name           : 0000:02:00.1

# ./bnxtnvm  -dev=ntfp2 devid
PCI VendorID         : 14e4
PCI DeviceID         : 16d7
PCI Subsys VendorID  : 14e4
PCI Subsys DeviceID  : 1402
PCI Device Name      : 0000:02:00:1

# ./bnxtnvm  -dev=ntfp2 pkgver
Active Package version    : 20.06.01.06
Package version on NVM    : 20.06.01.06

I added a log inside function __bnxt_hwrm_func_qcaps() and I notice
that max_rx_em_flows value is inconsistent. It changes right after
port init.

max_mac_addrs is also inconsistent because it depends on
max_rx_em_flows since commit :
http://scm.6wind.com/pub/dpdk.org/dpdk-stable/commit/?h=18.11&id=15f42f5e426d,

This is the log I have,  when I launch ethtool example application.
For port 1, max_rx_em_flows=51116 and then  max_rx_em_flows=61292

# ethtool -l 0-1 -n 4
EAL: Detected 12 lcore(s)
EAL: Detected 1 NUMA nodes
EAL: Multi-process socket /var/run/dpdk/rte/mp_socket
EAL: Some devices want iova as va but pa will be used because.. EAL:
vfio-noiommu mode configured
EAL: No free hugepages reported in hugepages-1048576kB
EAL: Probing VFIO support...
EAL: VFIO support initialized
EAL: WARNING: cpu flags constant_tsc=yes nonstop_tsc=no -> using
unreliable clock cycles !
EAL: PCI device 0000:00:03.0 on NUMA socket -1
EAL:   Invalid NUMA socket, default to 0
EAL:   probe driver: 1af4:1000 net_virtio
EAL: PCI device 0000:00:04.0 on NUMA socket -1
EAL:   Invalid NUMA socket, default to 0
EAL:   probe driver: 14e4:16dc net_bnxt
EAL:   using IOMMU type 8 (No-IOMMU)
bnxt_dev_init(): Broadcom NetXtreme driver bnxt

bnxt_hwrm_ver_get(): 1.7.6:20.6.107
bnxt_hwrm_ver_get(): Driver HWRM version: 1.9.2
__bnxt_hwrm_func_qcaps(): !!!!!!!!!!!! bp->max_l2_ctx= 4 + 51116 = 51120
bnxt_dev_init(): bnxt found at mem feb04000, node addr 0x1100800000M
EAL: PCI device 0000:00:05.0 on NUMA socket -1
EAL:   Invalid NUMA socket, default to 0
EAL:   probe driver: 14e4:16dc net_bnxt
bnxt_hwrm_ver_get(): 1.7.6:20.6.107
bnxt_hwrm_ver_get(): Driver HWRM version: 1.9.2
__bnxt_hwrm_func_qcaps(): !!!!!!!!!!!! bp->max_l2_ctx= 4 + 51116 = 51120
bnxt_dev_init(): bnxt found at mem feb0c000, node addr 0x1100908000M
Number of NICs: 2
!!!!!!!!!!!! dev_info.max_mac_addrs= 51120
!!!!!!!!!!!! dev_info.max_mac_addrs= 51120

Init port 0..
__bnxt_hwrm_func_qcaps(): !!!!!!!!!!!! bp->max_l2_ctx= 4 + 1 = 5
!!!!!!!!!!!! dev_info.max_mac_addrs= 5
bnxt_print_link_info(): Port 0 Link Down
bnxt_print_link_info(): Port 0 Link Up - speed 10000 Mbps - full-duplex

Init port 1..
__bnxt_hwrm_func_qcaps(): !!!!!!!!!!!! bp->max_l2_ctx= 4 + 61292 = 61296
!!!!!!!!!!!! dev_info.max_mac_addrs= 61296
bnxt_print_link_info(): Port 1 Link Down
bnxt_print_link_info(): Port 1 Link Up - speed 10000 Mbps - full-duplex
EthApp>

On master, I see that the new calculation for the number of L2 context
has been restricted to Whitney chip family.
Anyone knows if the chip/FW I am using is compatible with this new calculation ?

By the way, I tried to upgrade to FW 214.0.253.5 and Init is falling very early:
bnxt_hwrm_func_resc_qcaps(): error 1:0:00000000:00f9
bnxt_dev_init(): hwrm query capability failure rc: ffffffea
EAL: Requested device 0000:00:04.0 cannot be used


Best Regards
Chris