Adding BCM5720 NIC to rp5700 – five beeps, but RAM is good

[nabijaczleweli]

I have an rp5700, and a BCM5720 2x1G NIC (product brief). After inserting it in the PCIe x1 slot, on next boot the BIOS beeps in five-time bursts; according to the knowledge base, this means a "Pre-video memory error"; however my RAM is good and I have tried many configurations between six gigabytes (2x2G, 2x1G) and one gigabyte to the same effect: system boots correctly without the card, and five-beeps as soon as I boot with it inserted.

My CPU is an E6400 and I'm running BIOS 1.24 with microcode 1.06, both latest at time of writing; however, I've gone through all BIOS versions listed on the product page during my testing, to no avail.

The card firmware is version 20.8.4, with bootcode version 5720-v1.39.

I have verified that the slot works by plugging in a WLAN card, which worked, no problems.

I have verified that the card works by plugging it in a different computer – functions without problems.

When hot-plugged, upon rescan the WLAN card is picked up normally, but the NIC does not enumerate.

I've tried disabling Option ROM download, the onboard NIC in general, and all devices not critical to the boot process, to no avail.

I tried disabling Option ROM upload from the card as well, with no behavioural changes.

Any suggestions? Have I missed something obvious?

[nabijaczleweli]

By way of an update: I'm booting with "acpi=off noapic" to get the hotplug driver working, and "pci_hotplug.debug=Y pci_hotplug.debug_acpi=Y pciehp.pciehp_poll_mode=Y pciehp.pciehp_poll_time=1" for verbosity and debugging.

This is what I get in dmesg when hotplugging a JMicron SATA+PATA controller card [2229], unplugging it [2519], then plugging in the BCM5720 [2699], and unplugging it (afterward, no output).

[ 2229.386509] pcieport 0000:00:1c.0: pciehp_ctrl: Slot(2): Card present
[ 2229.387935] pcieport 0000:00:1c.0: pciehp_ctrl: Slot(2): Link Up
[ 2229.522513] pci 0000:20:00.0: [197b:2363] type 00 class 0x010601
[ 2229.524013] pci 0000:20:00.0: reg 0x24: [mem 0x00000000-0x00001fff]
[ 2229.525416] pci 0000:20:00.0: reg 0x30: [mem 0x00000000-0x0000ffff pref]
[ 2229.526867] pci 0000:20:00.0: PME# supported from D3hot
[ 2229.527001] pci 0000:20:00.1: [197b:2363] type 00 class 0x010185
[ 2229.527027] pci 0000:20:00.1: reg 0x10: [io  0x0000-0x0007]
[ 2229.527037] pci 0000:20:00.1: reg 0x14: [io  0x0000-0x0003]
[ 2229.527047] pci 0000:20:00.1: reg 0x18: [io  0x0000-0x0007]
[ 2229.527056] pci 0000:20:00.1: reg 0x1c: [io  0x0000-0x0003]
[ 2229.527066] pci 0000:20:00.1: reg 0x20: [io  0x0000-0x000f]
[ 2229.527197] pci 0000:20:00.0: disabling ASPM on pre-1.1 PCIe device.  You can enable it with 'pcie_aspm=force'
[ 2229.527218] pci 0000:20:00.0: BAR 6: assigned [mem 0xf0000000-0xf000ffff pref]
[ 2229.527220] pci 0000:20:00.0: BAR 5: assigned [mem 0xf0010000-0xf0011fff]
[ 2229.527226] pci 0000:20:00.1: BAR 4: assigned [io  0x3000-0x300f]
[ 2229.527231] pci 0000:20:00.1: BAR 0: assigned [io  0x3010-0x3017]
[ 2229.527236] pci 0000:20:00.1: BAR 2: assigned [io  0x3018-0x301f]
[ 2229.527241] pci 0000:20:00.1: BAR 1: assigned [io  0x3020-0x3023]
[ 2229.527246] pci 0000:20:00.1: BAR 3: assigned [io  0x3024-0x3027]
[ 2229.527252] pcieport 0000:00:1c.0: PCI bridge to [bus 20]
[ 2229.527255] pcieport 0000:00:1c.0:   bridge window [io  0x3000-0x3fff]
[ 2229.527259] pcieport 0000:00:1c.0:   bridge window [mem 0xf0000000-0xf01fffff]
[ 2229.527262] pcieport 0000:00:1c.0:   bridge window [mem 0x1a0000000-0x1a01fffff 64bit pref]
[ 2229.527271] pci 0000:20:00.0: async suspend disabled to avoid multi-function power-on ordering issue
[ 2229.527306] pci 0000:20:00.1: async suspend disabled to avoid multi-function power-on ordering issue
[ 2229.588199] pata_jmicron 0000:20:00.1: enabling device (0000 -> 0001)
[ 2229.588212] pata_jmicron 0000:20:00.1: found PCI INT B -> IRQ 11
[ 2229.590384] scsi host2: pata_jmicron
[ 2229.594957] scsi host3: pata_jmicron
[ 2229.595040] ata3: PATA max UDMA/100 cmd 0x3010 ctl 0x3020 bmdma 0x3000 irq 11
[ 2229.595042] ata4: PATA max UDMA/100 cmd 0x3018 ctl 0x3024 bmdma 0x3008 irq 11
[ 2229.604267] ahci 0000:20:00.0: version 3.0
[ 2229.604278] ahci 0000:20:00.0: enabling device (0000 -> 0002)
[ 2229.604290] ahci 0000:20:00.0: found PCI INT A -> IRQ 11
[ 2229.604294] ahci 0000:20:00.0: sharing IRQ 11 with 0000:00:02.0
[ 2229.614479] ahci 0000:20:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA mode
[ 2229.614482] ahci 0000:20:00.0: flags: 64bit ncq pm led clo pmp pio slum part 
[ 2229.615286] scsi host4: ahci
[ 2229.620362] scsi host5: ahci
[ 2229.620478] ata5: SATA max UDMA/133 abar m8192@0xf0010000 port 0xf0010100 irq 11
[ 2229.620481] ata6: SATA max UDMA/133 abar m8192@0xf0010000 port 0xf0010180 irq 11
[ 2229.930936] ata6: SATA link down (SStatus 0 SControl 300)
[ 2229.932884] ata5: SATA link down (SStatus 0 SControl 300)

[ 2519.303179] pcieport 0000:00:1c.0: pciehp_ctrl: Slot(2): Link Down
[ 2519.305085] pcieport 0000:00:1c.0: pciehp_ctrl: Slot(2): Card not present
[ 2519.312295] ahci 0000:20:00.0: AHCI controller unavailable!
[ 2519.314153] ata5: failed to stop engine (-19)
[ 2519.316117] ahci 0000:20:00.0: AHCI controller unavailable!
[ 2519.317981] ata6: failed to stop engine (-19)

[ 2699.556956] pcieport 0000:00:1c.0: pciehp_ctrl: Slot(2): Card present
[ 2701.596901] pcieport 0000:00:1c.0: Data Link Layer Link Active not set in 1000 msec
[ 2701.598606] pcieport 0000:00:1c.0: pciehp_ctrl: Failed to check link status

(The pciehp log names are a bit non-standard, as I've added this patch to my kernel – uname -a gives "Linux szarotka 5.5.0-1-amd64 #1 SMP Debian 5.5.13-2.1~szarotka2 (2020-04-15) x86_64 GNU/Linux". I thought other two patches in that directory would help, but they didn't change anything. Oh well.).

On the Buster kernel, however (no additional patches; Linux szarotka 4.19.0-8-amd64 #1 SMP Debian 4.19.98-1 (2020-01-26) x86_64 GNU/Linux), I get this:

[   50.179494] pciehp 0000:00:1c.0:pcie004: Slot(2): Card present
[   52.180619] pcieport 0000:00:1c.0: Data Link Layer Link Active not set in 1000 msec
[   53.686927] pciehp 0000:00:1c.0:pcie004: link training error: status 0x1001
[   53.689123] pciehp 0000:00:1c.0:pcie004: Failed to check link status

I've started to rebase the debian patchset for the current 5.6.4 kernel, but building and testing that will take time.

[DGroves · Accepted solution]

is this a server based card? if so then it may use the "SMB" bus and this is conflicting with you motherboards "SMB" bus as the commands on the card are not compatible with your motherboard

usually taping the cards SMB bus pins will bypass the error, but i still recommend a different card as you may have other issues

The pins you need to mask are:

B5 (SMCLK, SMBus clock) an B6 (SMDAT, SMBus data)

they are the 5/6 th pins on the top/componet side of the card going from the left to right

only the top pins 5/6 are covered not the underside pins

[nabijaczleweli]

Thank you, that worked! The BIOS stopped beeping immediately!

I'm not sure about the "serverness" of the card (I got it second-hand, but it does have a low-profile bracket, so you're most likely right), but that must've been it!

It'd still hard-halt during OpROM download (or, with it disabled, when it should've happened) – the solution to that was to enable all features marked "Enabled/Disable" in Broadcom's b57udiag.exe tool, and enable booting and OpROM download in the BIOS (though I believe this is the default config, or close to it).

With your fix the card works perfectly so far, and the config screen allows the configuration of the onboard NIC as well! Thanks again!