[Partysip-dev] Re: partysip compiling

[Tim Na]

Hi, Ralf.
I found the reason for the bus error (or seg fault).
in pspm_uap_init() function, core->uap->uap_plugins wasn't initilized to NULL.
So I added the line for the initialization.

int  pspm_uap_init     (pspm_uap_t **pspm)
{
  int i;
  (*pspm) = (pspm_uap_t*) smalloc(sizeof(pspm_uap_t));
  if (*pspm==NULL) return -1;

  i = module_init(& ((*pspm)->module), MOD_THREAD, "uap");
  if (i!=0) goto mui_error1;

  (*pspm)->uap_plugins = NULL;                    /* Added by Tim Na */

  /*  (*pspm)->uap_pending = NULL; */

  (*pspm)->uas_request = (fifo_t*)smalloc(sizeof(fifo_t));
  if ((*pspm)->uas_request==NULL) goto mui_error2;


After this, I wondered how this code worked in Linux machine without any 
problem. (Red Hat 7.1)
As I traced the fuction, amazingly, after the smallloc of (*pspm), uap_plugins 
was initialized to NULL.!!!
I wonder how that magic happened there without explicit assignment... :-(  
(perhaps I wasn't paying attention during the lecture in pointers..)
Any comment on this?

Anyway, it works now.. : )

----- Original Message ----- 
From: "Ralf Mahlo" <address@hidden>
To: "Tim Na" <address@hidden>
Sent: Friday, May 17, 2002 2:47 PM
Subject: partysip compiling


> Hi Tim Na,
> 
> thank you very much for sending me your changes...
> Today I'll start to look deeper into the source code. I think, there could be 
> a problem with
> "ADD_ELEMENT" , defined in /ppl/include/ppl.h , because the bus error 
> happens, when 'first_element
> != NULL' ... I'll check this...
> 
> Regards, Ralf
> 
> P.S.: Are you directly from Korea?

 What do you mean "directly"?  I am working in Korea right now if that is what 
you are asking. : )
> 
>