BCE(4)

📦 version 15.1, in freebsd 14.5, freebsd 15.1

💖 bce(4) — the QLogic NetXtreme II bestie 💖

okay so this is the man page for bce, bestie — it's the driver for the QLogic NetXtreme II family of PCI/PCIe Gigabit Ethernet adapters, and yes that includes the BCM5706, BCM5708, BCM5709 and BCM5716. she's a networking icon and she's about to explain herself 💅

✨ NAME ✨

bce — QLogic NetXtreme II (BCM5706/5708/5709/5716) PCI/PCIe Gigabit Ethernet adapter driver

🦋 SYNOPSIS 🦋

so if you wanna compile this driver into the kernel (main character energy, no modules), you drop these lines into your kernel configuration file:

device miibus
device bce

OR, alternative route, if you'd rather load her as a module at boot time, you put this line in loader.conf(5):

if_bce_load="YES"

both are valid, pick your vibe, you can't go wrong 🫶

🌸 DESCRIPTION 🌸

the bce driver supports QLogic's NetXtreme II product family, which includes the BCM5706, BCM5708, BCM5709 and BCM5716 Ethernet controllers. literally the whole squad 💜

okay so context: the NetXtreme II product family is composed of various Converged NIC (or CNIC) Ethernet controllers. these things support a TCP Offload Engine (TOE), Remote DMA (RDMA), and iSCSI acceleration, in addition to standard L2 Ethernet traffic — all on the same controller. she does it all, no cap 👑

here's the feature list that bce actually supports under FreeBSD:

IP/TCP/UDP checksum offload
Jumbo frames (up to 9022 bytes)
VLAN tag stripping
Interrupt coalescing
10/100/1000Mbps operation in full-duplex mode
10/100Mbps operation in half-duplex mode

and here are the media types bce supports — the whole lineup, no skips:

autoselect — enable autoselection of the media type and options. and if you're feeling controlling, the user can manually override the autoselected mode by adding media options to rc.conf(5). she'll let you, she's chill like that ✨
10baseT/UTP — set 10Mbps operation. the ifconfig(8) mediaopt option can also be used to select either full-duplex or half-duplex modes. tiny but mighty 🎀
100baseTX — set 100Mbps (Fast Ethernet) operation. the ifconfig(8) mediaopt option can also be used to select either full-duplex or half-duplex modes. fast-ish, still cute 🌈
1000baseSX — sets 1000Mbps operation. only full-duplex mode is supported at this speed. ok so no half-duplex here, that's just the rules 💅
1000baseT — set 1000baseT operation over twisted pair. only full-duplex mode is supported. same deal, full-duplex only, she said what she said 🔥
2500BaseSX — set 2500Mbps operation. only full-duplex mode is supported. (yes the capital B is on purpose, don't come for her 💀)

and these are the media options bce supports:

full-duplex — force full duplex operation. she's listening and talking, periodt 🫶
half-duplex — force half duplex operation. one at a time bestie, we're taking turns ✋

for more information on configuring this device, see ifconfig(8). that's your bestie for all the config stuff, iykyk 🌟

🔥 HARDWARE 🔥

the bce driver provides support for various NICs based on the QLogic NetXtreme II family of Gigabit Ethernet controllers, including the following — and i'm listing every single one because they all ate:

QLogic NetXtreme II BCM5706 1000Base-SX
QLogic NetXtreme II BCM5706 1000Base-T
QLogic NetXtreme II BCM5708 1000Base-SX
QLogic NetXtreme II BCM5708 1000Base-T
QLogic NetXtreme II BCM5709 1000Base-SX
QLogic NetXtreme II BCM5709 1000Base-T
QLogic NetXtreme II BCM5716 1000Base-T
Dell PowerEdge 1950 integrated BCM5708 NIC
Dell PowerEdge 2950 integrated BCM5708 NIC
Dell PowerEdge R710 integrated BCM5709 NIC
HP NC370F Multifunction Gigabit Server Adapter
HP NC370T Multifunction Gigabit Server Adapter
HP NC370i Multifunction Gigabit Server Adapter
HP NC371i Multifunction Gigabit Server Adapter
HP NC373F PCIe Multifunc Giga Server Adapter
HP NC373T PCIe Multifunction Gig Server Adapter
HP NC373i Multifunction Gigabit Server Adapter
HP NC373m Multifunction Gigabit Server Adapter
HP NC374m PCIe Multifunction Adapter
HP NC380T PCIe DP Multifunc Gig Server Adapter
HP NC382T PCIe DP Multifunction Gigabit Server Adapter
HP NC382i DP Multifunction Gigabit Server Adapter
HP NC382m DP 1GbE Multifunction BL-c Adapter

like… that's a lot of hardware and every one of them is valid 💖

🎀 SYSCTL VARIABLES 🎀

okay these next ones are tunable — the following variables are available as both sysctl(8) variables and loader(8) tunables. so you can poke them at runtime or set them before boot, whichever fits your situation. obsessed 🌸

hw.bce.verbose — enable/disable verbose logging and output to the console. useful for debugging (default 0). basically the "tell me everything" switch 🔍
hw.bce.msi_enable — enable/disable MSI support (default 1). interrupts but fancy ✨
hw.bce.tso_enable — enable/disable TSO support (default 1). offload queen 👑
hw.bce.strict_rx_mtu — enable/disable strict RX frame size checking (default 0). she's checking 👀
hw.bce.hdr_split — enable/disable frame header/payload splitting (default 1). split it, slay it 💅
hw.bce.rx_pages — set the number of memory pages assigned to receive packets by the driver. ⚠️ due to alignment issues, this value can only be of the set 1, 2, 4 or 8 (default 2). don't try to be creative here bestie, the alignment said no 🙈
hw.bce.tx_pages — set the number of memory pages assigned to transmit packets by the driver. ⚠️ due to alignment issues, this value can only be of the set 1, 2, 4 or 8 (default 2). same rules, same energy 🦋
hw.bce.rx_ticks — time in microsecond ticks to wait before generating a status block updates due to RX processing activity. values from 0-100 are valid. a value of 0 disables this status block update. ⚠️ cannot be set to 0 if hw.bce.rx_quick_cons_trip is also 0 (default 18). so if you zero one, don't zero the other, ok? 😭
hw.bce.rx_ticks_int — time in microsecond ticks to wait during RX interrupt processing before generating a status block update. values from 0-100 are valid. valid values are in the range from 0-100. a value of 0 disables this status block update (default 18). yes she says the range twice, she's thorough 💜
hw.bce.rx_quick_cons_trip — number of RX Quick BD Chain entries that must be completed before a status block is generated. values from 0-256 are valid. a value of 0 disables this status block update. ⚠️ cannot be set to 0 if hw.bce.rx_ticks is also 0 (default 6). it's giving mutual dependency 🫶
hw.bce.rx_quick_cons_trip_int — number of RX quick BD entries that must be completed before a status block is generated duing interrupt processing. values from 0-256 are valid. a value of 0 disables this status block update (default 6). (yes "duing" — that's the original's spelling, not mine, don't @ me 💀)
hw.bce.tx_ticks — time in microsecond ticks to wait before a status block update is generated due to TX activity. values from 0-100 are valid. a value of 0 disables this status block update. ⚠️ cannot be set to 0 if hw.bce.tx_quick_cons_trip is also 0 (default 80). same buddy-system rule as RX ✨
hw.bce.tx_ticks_int — time in microsecond ticks to wait in interrupt processing before a status block update is generated due to TX activity. values from 0-100 are valid. a value of 0 disables this status block update (default 80). transmission ticking, but make it interrupt-y 🔥
hw.bce.tx_cons_trip — how many TX Quick BD Chain entries that must be completed before a status block is generated. values from 0-100 are valid. a value of 0 disables this status block update. ⚠️ cannot be set to 0 if hw.bce.tx_ticks is also 0 (default 20). she keeps her promises, and she keeps her dependencies 💖
hw.bce.tx_cons_trip_int — how many TX Quick BD Chain entries that must be completed before a status block is generated during an interrupt. values from 0-100 are valid. a value of 0 disables this status block update (default 20). last one of this batch and she ate ⭐

💀 DIAGNOSTICS 💀

okay bestie, buckle up — this is the part where the driver tells you what went wrong. every one of these starts with bce%d: where the number is the interface unit, so you know exactly which card is being dramatic. some of these are "lol just a timeout" and some are straight-up fatal initialization errors where the whole thing just gave up. read carefully, no skipping 💅

bce%d: PCI memory allocation failed! — the driver hit a fatal initialization error. it's giving "couldn't even get out of bed."
bce%d: PCI map interrupt failed! — another fatal initialization error, this time about interrupts. also not slay.
bce%d: Unsupported controller revision (%c%d) — the driver literally does not support the revision of the controller you've got. wrong era, wrong vibe, incompatible.
bce%d: Controller initialization failed! — fatal initialization error, full stop. it tried, it really did.
bce%d: NVRAM test failed! — couldn't access the controller's NVRAM properly. NVRAM said "no ❤️".
bce%d: DMA resource allocation failed! — couldn't allocate DMA memory to set up the controller's host memory data structures. the plumbing didn't get installed.
bce%d: Interface allocation failed! — the driver could not create a network interface for the controller. no interface for you 🙈
bce%d: PHY probe failed! — couldn't access the PHY that the controller uses.
bce%d: Failed to setup IRQ! — couldn't initialize the IRQ handler.
bce%d: Error: PHY read timeout! — couldn't read a PHY register before the timeout period expired. clock ran out.
bce%d: PHY write timeout! — couldn't write to the PHY register because a timeout occurred.
bce%d: Timeout error reading NVRAM at offset 0x%08X! — per the docs, the driver could not write to NVRAM because a timeout occurred. yes it says "reading" in the message and "write" in the description, we're just reporting what it says, don't come for me 🫶
bce%d: Unknown Flash NVRAM found! — the driver doesn't recognize the NVRAM device being used, which means it can't access it correctly. mystery hardware, main character energy.
bce%d: Invalid NVRAM magic value! — the driver can't read NVRAM, or the NVRAM is corrupt. either way: bad.
bce%d: Invalid Manufacturing Information NVRAM CRC! — same deal: can't read NVRAM, or it's corrupt.
bce%d: Invalid Feature Configuration Information NVRAM CRC! — again: can't read NVRAM, or it's corrupt.
bce%d: DMA mapping error! — couldn't map memory into the DMA addressable space the controller needs.
bce%d: Could not allocate parent DMA tag! — couldn't allocate a PCI compatible DMA tag.
bce%d: Could not allocate status block DMA tag! — couldn't allocate a DMA tag for the controller's status block.
bce%d: Could not allocate status block DMA memory! — couldn't get DMA addressable memory for the controller's status block.
bce%d: Could not map status block DMA memory! — couldn't map the status block memory into the controller's DMA address space.
bce%d: Could not allocate statistics block DMA tag! — couldn't allocate a DMA tag for the controller's statistics block.
bce%d: Could not allocate statistics block DMA memory! — couldn't get DMA addressable memory for the statistics block.
bce%d: Could not map statistics block DMA memory! — couldn't map the statistics block memory into the controller's DMA address space.
bce%d: Could not allocate TX descriptor chain DMA tag! — couldn't allocate a DMA tag for the controller's TX chain. transmit side is struggling.
bce%d: Could not allocate TX descriptor chain DMA memory! — couldn't get DMA addressable memory for the TX chain.
bce%d: Could not map TX descriptor chain DMA memory! — couldn't map the TX descriptor chain memory into the controller's DMA address space.
bce%d: Could not allocate TX mbuf DMA tag! — couldn't allocate a DMA tag for the controller's TX mbuf memory.
bce%d: Unable to create TX mbuf DMA map! — couldn't map the TX mbuf memory into the controller's DMA address space.
bce%d: Could not allocate RX descriptor chain DMA tag! — couldn't allocate a DMA tag for the controller's RX chain. now the receive side wants attention too.
bce%d: Could not allocate RX descriptor chain — couldn't allocate DMA addressable memory for the controller's RX chain. (and yeah, that message just ends there. no exclamation mark, no nothing. iconic, honestly.)
bce%d: Could not map RX descriptor chain DMA memory! — couldn't map the RX descriptor chain memory into the controller's DMA address space.
bce%d: Could not allocate RX mbuf DMA tag! — couldn't allocate a DMA tag for the controller's RX mbuf memory.
bce%d: Unable to create RX mbuf DMA map! — couldn't map the RX mbuf memory into the controller's DMA address space.
bce%d: Firmware synchronization timeout! — couldn't sync with the firmware running on the controller. the firmware may be stopped or hung 😭
bce%d: Invalid Ethernet address! — couldn't read a valid Ethernet MAC address from NVRAM.
bce%d: Reset failed! — fatal initialization error. the reset did not reset.
bce%d: Byte swap is incorrect! — fatal initialization error, and this one's special: you're supposed to contact the author with details of the CPU architecture and system chipset in use. yes really, go tell them 🫡
bce%d: Firmware did not complete initialization! — fatal initialization error.
bce%d: Bootcode not running! — fatal initialization error.
bce%d: Error mapping mbuf into RX chain! — couldn't map an RX mbuf into DMA addressable memory.
bce%d: Error filling RX chain: rx_bd[0x%04X]! — couldn't allocate enough mbufs to fill the RX chain during initialization. the official advice: try increasing the number of mbufs available in the system, increase system memory, or — if you're using jumbo frames — make sure enough 9KB mbufs are available. 👑
bce%d: Failed to allocate new mbuf, incoming frame dropped! — couldn't grab a new mbuf for the RX chain, so it reused the mbuf from the received frame and dropped that incoming frame in the process. try increasing the number of mbufs available in the system or increase system memory.
bce%d: Controller reset failed! — a fatal initialization error has occurred.
bce%d: Controller initialization failed! — a fatal initialization error has occurred. (yes, it shows up twice in the list with different wording. we don't question the classics.)
bce%d: Block initialization failed! — a fatal initialization error has occurred.
bce%d: Error mapping mbuf into TX chain! — couldn't map a TX mbuf into DMA addressable memory.
bce%d: Error registering poll function! — the driver got an error while trying to register the poll function.
bce%d: Changing VLAN_MTU not supported. — changing the VLAN MTU is not currently supported by the driver. no cap, it's just not a thing yet.
bce%d: Cannot change VLAN_HWTAGGING while management firmware (ASF/IPMI/UMP) is running! — management firmware for ASF/IPMI/UMP requires that VLAN tag stripping be enabled in the controller. so you can't turn it off while that's running, bestie.
bce%d: Changing VLAN_HWTAGGING not supported! — disabling VLAN tag stripping is not currently supported by the driver.
bce%d: Watchdog timeout occurred, resetting! — the device stopped responding to the network, there's a problem with the cable connection, or a driver logic problem happened. could be any of the three, go check your cable first 🧚
bce%d: Fatal attention detected: 0x%08X! — a controller hardware failure has occurred. if the problem keeps happening, replace the controller. respectfully, it's cooked.

🫶 SUPPORT 🫶

got questions? hit up your QLogic approved reseller, or QLogic Technical Support at http://support.qlogic.com, or slide into their inbox at support@qlogic.com. they're literally there for this 💖

the friend group! go read them, iykyk ✨

🦋 HISTORY 🦋

the bce device driver first appeared in FreeBSD 6.1. a vintage icon, honestly. a little retro, still serving 🎀

👑 AUTHORS 👑

the bce driver was written by David Christensen <davidch@broadcom.com>. thank you for your service, king 💜

and that's the whole page, bestie! this one came from the Debian crew, dated June 4, 2012 — so yeah, it's a lil vintage, but the facts still eat. go forth and ifconfig responsibly 🌸💅

📦 where this page comes from

freebsd 14.5: base 14.5
freebsd 15.1: base 15.1
💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash-1017 on 2026-11-04: a language model's rewrite of the original, under its terms