nmosh will not start, 'free-c-callback-trampoline'

[David Banks]

Hi all,

I have a weird problem, one of my linux machines will not run the
current nmosh from git, although it will configure and compile without
any problem.

amoe@vlinder $ uname -a
Linux vlinder 2.6.32-5-686 #1 SMP Tue Mar 8 21:36:00 UTC 2011 i686 GNU/Linux

<git checkout, ./gen-git-build.sh, configure and make proceed without error>

amoe@vlinder $ ./nmosh
[sudo] password for root:

Exception:
error in raise: unhandled exception has occurred

#<type:condition(#<#<type:record-type-descriptor&assertion
#<type:record-type-descriptor&violation
#<type:record-type-descriptor&serious
#<type:record-type-descriptor&condition #f &conidion-uid #f #f () >
&serious-uid #f #f () > &violation-uid #f #f () > &assertion-uid #f #f
() >> #<#<type:record-type-descriptor&who
#<type:record-type-descriptor&condition #f &conidion-uid #f #f () >
&who-uid #f #f ((#f . who)) >free-c-callback-trampoline >
#<#<type:record-type-descriptor&message
#<type:record-type-descriptor&condition #f &conidion-uid #f #f () >
&message-uid #f #f ((#f . message)) >wrong number of arguments
(required 1, got 0) > #<#<type:record-type-descriptor&irritants
#<type:record-type-descriptor&condition #f &conidion-uid #f #f () >
&irritants-uid #f #f ((#f . irritants)) >(()) >) >

Stack trace:
1. throw: <subr>
2. %ffi-free-c-callback-trampoline: <subr>

The weird thing is that this error is unique to this machine. It is
running Debian squeeze, I've just installed a clean VM with the same
OS and arch, and nmosh builds and runs fine on that VM.
Does anyone know what the problem might be? There is nothing obvious
in the configure output. psyntax-mosh is not affected and runs fine.

I can provide config.status etc on request...

Cheers,
--
David Banks  <amo...@gmail.com>

[okuoku]

2011/5/23 David Banks <amo...@gmail.com>:

> Hi all,
>
> I have a weird problem, one of my linux machines will not run the
> current nmosh from git, although it will configure and compile without
> any problem.

- snip -

> The weird thing is that this error is unique to this machine.  It is
> running Debian squeeze, I've just installed a clean VM with the same
> OS and arch, and nmosh builds and runs fine on that VM.
> Does anyone know what the problem might be?  There is nothing obvious
> in the configure output.  psyntax-mosh is not affected and runs fine.

Sorry for inconvenience..

In short:
Current(git head) mosh corrupts its heap in very rare circumstances.
So i recommend you to use mosh(psyntax-mosh) instead of nmosh for a while.

Detailed:
Currently, I think this behavior triggered by a BoehmGC option
GC_DONT_ADD_BYTE_AT_END
which we applied to mosh recently. But i'm still investigating on this.

At least, it seems GC-related problem. So, this problem may or may not
occur on same arch, same OS, same machine...
because the problem depends on the memory layout of mosh/nmosh
execution environment.

[David Banks]

On 23 May 2011 12:24, okuoku <oku...@gmail.com> wrote:
> Current(git head) mosh corrupts its heap in very rare circumstances.
> So i recommend you to use mosh(psyntax-mosh) instead of nmosh for a while.

Ah, OK. Sound nasty to debug! :/
If you want me to run any diagnostics on my machine, let me know.
This is reproducible every time.