About RTX-51 Full to AT89C51ED2 run any os function

rtx51

I'd tried to port RTX-51 Full version to AT89C51ED2. The compiler and simulator can run without any errors or warnings. Even I can see the real-time signals varies high-low-high low in simulated logical analyzer. I can also download it into target successfully, but it cannot not run any os function include os_start_system().

Here is my questions:

1. What's wrong with my project?
2. Does someone knows what the difference between evaluation & registered version?

Please~
-----------------------------------------------------------------

//*****************************
Here is the uVision3 report
//*****************************
Build target 'Target 1'
User command #1: a51 rtxconf.a51 debug
A51 MACRO ASSEMBLER V8.00c - SN: Eval Version
COPYRIGHT KEIL ELEKTRONIK GmbH 1987 - 2006
ASSEMBLY COMPLETE. 0 WARNING(S), 0 ERROR(S)
assembling STARTUP.A51...
compiling main.c...
assembling L51_BANK.A51...
assembling RTXCONF.A51...
linking...
Program Size: data=87.0 xdata=1139 code=7399
creating hex file from "test"...
"test" - 0 Error(s), 0 Warning(s).

//*****************************
Here is my code in main.c
//*****************************
#include "rtx51.h"
#include "at89c51xd2.h"

sbit pin_0 = P2^7;
sbit pin_1 = P2^6;

long counter0 = 0; /* counter for task 0 */
long counter1 = 0; /* counter for task 1 */
long counter2 = 0; /* counter for task 2 */
long counter3 = 0; /* counter for task 2 */


void main (void)
{
pin_0 = 0; // it can run successfully!
pin_1 = 1; // it can run successfully!
os_start_system (4); // no response here!!!
}

task_create () _task_ 4 {
os_create_task (0);
os_create_task (1); /* start task 1 */

os_delete_task (4);
}

job0 () _task_ 0 {

while (1) { /* endless loop */
pin_0 = ~pin_0;
os_wait2 (K_TMO, 5); /* wait for timeout: 5 ticks */
}
}

job1 () _task_ 1 {
while (1) { /* endless loop */
pin_1 = ~pin_1;
os_wait2 (K_TMO, 4); /* wait for timeout: 4 ticks */
}
}