[Nix-dev] Different behavior of cryopid on two "identical" machines

[Marco Maggesi]

Hi,

sorry for the late replay but I do not have very busy these days.

I moved the two virtual machines on the same physical host and the  
different
behavior persists. Thus I can now exclude that this depends on the  
Host OS or
on the version of VirtualBox.

I tried to use strace & gdb to get some insight but I'm not competent
enough to understand what is going.  Perhaps someone else on this list  
is
able to do better than me:

I put three files in the directory

   http://web.math.unifi.it/users/maggesi/Hello/

- hello.c        a test program.
- hello          the compiled version.
- hello.freeze   the binary freezed with cryopid.

When I run ./hello.freeze the program crashes immediately after one  
line of output.

I would be glad to know if other people could test cryopid (especially  
on i686)
and say what they get.

To test crypoid you can follow these steps:

0. install cryopid: nix-env -i cryopid
1. open a console
2. gcc -o hello hello.c
3. ./hello
4. the program output its PID and a sequence of progressive numbers.
5. open another console
6. crypid hello.freeze <PID>
7. test the generated binary: ./hello.freeze

Thanks again,
Marco

On Jan 4, 2010, at 3:14 PM, Lluís Batlle wrote:

> Can you try to use cryopid with some debug information, use gdb, ... ?
>
> 2010/1/4 Marco Maggesi <maggesi at math.unifi.it>:
>>
>> On Jan 3, 2010, at 1:55 PM, Lluís Batlle wrote:
>>
>>> Hello and a good new year!
>>>
>>> 2009/12/31 Marco Maggesi <maggesi at math.unifi.it>:
>>>>
>>>> Hi,
>>>>
>>>> I have two (almost) identical machines but cryopid works in one and
>>>> fails in the other.
>>>
>>> What problem do you see, specifically, to assert that it does not  
>>> work?
>>>
>>> Regards,
>>> Lluís.
>>
>>
>> Hi Lluís,
>>
>> I can freeze a program with cryopid and everything seems to work,
>> but the generated binary doesn't work, i.e., crashes immediately
>> after invocation.
>>
>> The binaries frozen with the "working" machine run without problems
>> on both machines.
>>
>> Thanks,
>> Marco