Re: Linux emulation problem

[christos%astron.com@localhost (Christos Zoulas)]

In article 
<0554472E-C717-48B6-831F-6F3E4E6670CD%physik.uni-wuppertal.de@localhost>,
Torsten Harenberg  <harenberg%physik.uni-wuppertal.de@localhost> wrote:
>Dear list,
>
>sorry, this is not that much port-amd64 related, but as I had several
>experienced answers here already, I would like to ask those of you with
>linux emulation experience to have a short look into this:
>
>As already described before, I try to run a linux binary (EMC^2
>Networker client) on a NetBSD/amd64 machine running under Xen.
>
>As I have learned before, I have upgraded this machine now to CURRENT
>(both kernel and world):
>
>NetBSD mail.physik.uni-wuppertal.de 5.99.53 NetBSD 5.99.53 (XEN3_DOMU)
>#0: Mon Jun 20 09:41:08 CEST 2011 
>root%mail.physik.uni-wuppertal.de@localhost:/usr/obj/sys/arch/amd64/compile/XEN3_DOMU
>amd64
>
>So Linux emulation runs now in principle. 
>
>However, I still have trouble to run the (unfortunately closed source)
>daemon provided by EMC^2, which ends in a core dump.
>
>I run it with ktrace and I think the critical part is here
>
> 14571  14571 nsrexecd CALL  bind(3,0x7f7fffff7d60,0xc)
> 14571  14571 nsrexecd MISC  linux sockaddr: 12, 100000000000000000000000
> 14571  14571 nsrexecd MISC  new sockaddr: 12, 0c2200000000000000000000
> 14571  14571 nsrexecd RET   bind -1 errno -95 Operation not supported
>
>Although I'm not sure, why this should crash the application.
>
>I have tried both suse100 and suse113 emulation layers, the behavior is
>this same.
>
>Full kdumps are here:
>
>http://www.atlas.uni-wuppertal.de/~harenber/suse113_kdump
>http://www.atlas.uni-wuppertal.de/~harenber/suse100_kdump
>
>Maybe one of you has a clue...

Great, now that you are on current it is a lot easier to help. Please
compile a kernel with DEBUG_LINUX. This will dump the contents of the
sockaddr we are barfing on. I think 0xc (12) is a tad small so I am
wondering what we are trying to do here.

christos