Interrupt storm-like behaviour on Netgate RCC-VE 2440



  • Hey, I got a bit of a weird problem I have been unable to solve:

    After a recent firmware update (ADI_RCCVE-01.00.00.12-nodebug) the second mPCIe slot started working (for more information see here). It all seems to work great, except the second slot is generating an interrupt storm-like behavior. The interrupts even completely saturate one of the CPU cores. It doesn't look like an IRQ conflict, but I'm not entirely sure about the IRQ routing on this device.

    Anyone have any ideas?


  • Netgate Administrator

    The behaviour stays with the slot if you switch the devices? Or only use one device?

    What devices are you using there? What OS are you running?

    Steve



  • @stephenw10:

    The behaviour stays with the slot if you switch the devices? Or only use one device?

    Yes, it does, both of it. Every wireless card plugged into the second mPCIe port causes an interrupt storm.

    @stephenw10:

    What devices are you using there? What OS are you running?

    I tried different Atheros wireless cards (for example WLE200NX and WLE600VX). I tried PfSense, FreeBSD and OpenBSD. Linux also, but the kernel disables the interrupt on boot, but this leads to other problems.

    Thank you for your help!


  • Netgate Administrator

    Hmm, the Interrupt storm issue was supposed to have been fixed in v10.

    https://github.com/ADIEngineering/adi_coreboot_public/blob/master/releases/ADI_RCCVE-01.00.00.10/ADI Engineering RCCVE Bootrom Software Release Notes.pdf

    You're sure you're running v12? It's hard to imagine you'd get that wrong but….

    Steve



  • Ah, that's "good" news - I'm not the first one  ;)

    No, I'm sure about that. I manually updated to version 12 and recently installed the package which allows PfSense-controlled updates to the BIOS. It definitely shows I'm up to date.

    EDIT: I'm not sure I understand the BIOS' release notes. Before version 10/12 I could not use the second mPCIe slot at all - and others neither.

    EDIT2: Possibly the interrupt has been disabled in the ACPI table so that the Linux kernel ignores it? On boot, Linux disabled the IRQ in question - but obviously that's really bad for networking hardware. Packet drops all over the place…


  • Netgate Administrator

    Mmm, I don't have the details unfortunately I wasn't involved in that.

    Probably better to drop an email to Netgate support and we can take it from there.

    Steve



  • @stephenw10:

    Mmm, I don't have the details unfortunately I wasn't involved in that.

    Probably better to drop an email to Netgate support and we can take it from there.

    Thank you for your help so far, I just did.


Log in to reply