Re: Kernel Panic - interrupt / if_re related?

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

:I'm using a custom configured kernel (although I haven't done anything but
:comment out drivers I don't use) with the if_re network driver
:(which Joerg recently updated). Since I started using the if_re driver
:again I've noticed a storm of livelock messages in the console, and twice
:tonight I've had a kernel panic while in X (thankfully I have DBB
:UNATTENDED enabled). I've been updating my kernel and doing a buildworld,
:installworld, and kernel every day now for a while. My kernel is current
:as of the tcp_output.c commit Matthew Dillon made August 2nd.
:
:I'm using gcc2, and both panics have happened while doing a make world on
:one of the virtual consoles while in X.
:
:I've provided links to a backtrace, a full backtrace, pciconf -l -v
:output, vmstat -i output, and my config below:
:
:http://drevil.warpcore.org/bt.log
:http://drevil.warpcore.org/full_bt.log
:http://drevil.warpcore.org/pciconf.log
:http://drevil.warpcore.org/vmstat_i.log
:http://drevil.warpcore.org/DTRIM
:(my config for "debug trim")
:
:As you can see from the vmstat -i output, interrupt 11 for the re0
:interface generates an incredible amount of total interrupts (22047896 in
:39 minutes of uptime).
:
:I'll keep the vmcore for a few more days, but I'd rather not have to
:upload it anywhere on this cable connection (1gb, a few hundred mb
:compressed I imagine)...
:
:Thanks
    
    Oh that's weird.  The backtrace is impossible, but I can see how it
    could have been generated.  GDB has the stack frame confused but so
    many interrupts occured on that stack it is finding ghost linkages
    from previous interrupts.

    I think your gdb might be out of sync with your kernel and you need to
    sync the world (buildworld/installworld) to get anything meaningful out
    of it.

    It does look to be RE related to me.

					-Matt
					Matthew Dillon 
					<dillon@xxxxxxxxxxxxx>