Re: [developer] fmd core dump

[Joerg Schilling <Joerg.Schilling@fokus.fraunhofer.de>]

Gabriele Bulfon <gbulfon@sonicle.com> wrote:

> Hi, I have a system (not really a recent illumos kernel, probably around 2012) that recently caused a couple of core.fmd.xxx dumps, here's what mdb is saying:
>
> bash-4.2# mdb core.fmd.1104
> Loading modules: [ fmd libumem.so.1 libc.so.1 libnvpair.so.1 libtopo.so.1 libuut il.so.1 libavl.so.1 libsysevent.so.1 eft.so ld.so.1 ]
> > $C
> fd98e988 libc.so.1`_lwp_kill+0x15(4, 6, 120a0, fef58000, fef58000, 4)
> fd98e9a8 libc.so.1`raise+0x2b(6, 0, fd98e9c0, feed83e9, 0, 0)
> fd98e9f8 libc.so.1`abort+0x10e(3a646d66, 4f424120, 203a5452, 75736e69, 63696666
> , 746e6569)
> fd98ee18 fmd_panic(8080ec0, fd98ee44, 1, 0)
> fd98ee38 fmd_panic+0x12(8080ec0, c, 3e8, ffb3dd87)
> fd98ee78 fmd_alloc+0x81(c, 1, 1dca2110, 0, 893c688, 84fd718)
> fd98eeb8 fmd_eventq_insert_at_head+0x43(890bb48, 91ec5b8, 0, 92f1ab2d)
> fd98eed8 fmd_module_gc+0x66(893c680, 0, 0, fd98eef8)
> fd98ef18 fmd_modhash_apply+0x3e(84fd718, 8073d50, 0, 0, 6c275b0e, 30cef3)
> fd98ef48 fmd_gc+0x28(80998c0, d, ff19063b, 30ceff, 84f8a48)
> fd98efa8 fmd_timerq_exec+0x127(84f8a40, 0, feda22a0, fef58000)
> fd98efc8 fmd_thread_start+0x5b(826cfb8, 0, 0, 0)
> fd98efe8 libc.so.1`_thrp_setup+0x88(feda2240)
> fd98eff8 libc.so.1`_lwp_start(feda2240, 0, 0, 0, 0, 0)

I had a similar problem half a year ago and it turned out that the reason was a 
bug in "dmake" that caused target specific macro assignements to be done twice 
(when in parallel mode) and as a result, the linker map file was used two times 
on the linker command line.

That caused shared libraries loaded by fmd to become currupted...

Jörg

-- 
 EMail:joerg@schily.net                    (home) Jörg Schilling D-13353 Berlin
    joerg.schilling@fokus.fraunhofer.de (work) Blog: http://schily.blogspot.com/
 URL: http://cdrecord.org/private/ http://sf.net/projects/schilytools/files/'