Valery Reznic wrote:
The second point is that I don't need advice about how
it can be done. I can think of three or four ways not
mentioned here at all. They have disadvantages in relation
to what I implemented, mostly in how long it takes to figure
out that the daemon has, indeed, finished, and in how much
protocol specific changes they would require. I am asking
for opinion about that one specific way I chose, which works
beautifully, but is extremely non-standard.

    
OK, you said itself - it's work. The only problem a can think of is that it will be impossible to use gdb/strace on your already ptrace'd agent.

Valery.


  
Hi Shahar,

About disadvantages of ptrace  (IMHO):
1. controlling process should take care of signal processing (signals are delivered to the caller ??)
    or just resent them to the tracee.  I am not sure signal delivery is reliable
2. You introduce new daemon wasting  the  tiny RAM you have.


And if you please can you comment why atexit combined with execve()
  or SIGSEGV/SIGTERM/, handler calling the execve() cannot be more simple solution  for you case ?

Regards,
Lev.
      

_______________________________________________
Linux-il mailing list
[email protected]
http://mailman.cs.huji.ac.il/mailman/listinfo/linux-il

  

_______________________________________________
Linux-il mailing list
[email protected]
http://mailman.cs.huji.ac.il/mailman/listinfo/linux-il

Reply via email to