Problem with mod_fastcgi and multithread app.

Hi,

I'm developing a multithreaded app under fastcgi. After issuing a few 
requests it dumps core with signal 11. Examining the core, the problem is here:

#0 0x8048a3e in do_something2(parameters here)
133      FCGX_FPrintF(request->err, "The address is %x", &var);
(gdb) l
{
130      char a[40974];
131      char *p1, *p2;
132      u_long var;
133      FCGX_FPrintF(request->err, "The address is %x", &var);
(gdb) print var
Cannot access memory at address 0xbfac8e60.

I've trimmed the app down to it's most basic form (it looks a little 
bit more complex than the example threaded.c) with only 130 lines of C code, 
doing nothing useful and compiled like this:

gcc -Wall -D_THREAD_SAFE -o app.fcgi app.o -L/usr/local/lib -lfcgi -pthread

It consistently dies in the same way after two or three requests. It seems like 
something is corrupting the process itself. If I remove line 130, the symptom 
dissaprears, although the program start to behave randomly. The size was select 
to mimic the stack shape the first time I saw the problem (is the size of the 
DB's variables that used to be there - the current version does not use or link 
against the DB libraries any more).

The platform is FreeBSD 4.8-4.10, Apache 1.3.29-1.3.33, mod_fastcgi-2.4.2 and 
fcgi-devkit-2.4.0, with FreeBSD's stock gcc.

I can send a sample program if needed.

Does anyone have a suggestion or pointer?

Kind regards and thanks in advance.


Fernando.

___________________________________
fastcgi-developers mailing list
http://fastcgi.com/fastcgi-developers/