MSP430 code works on TI/GNU-C but not on IAR

Other Parts Discussed in Thread: MSP430F6736

I added a short code at the end. The code is based on TI's example code for MSP430F6x series to output data on an UART port. I use MSP430F6736.

That code works fine on Code Composer Studio v6 with TI's or GNUC compilers. What I mean fine is that the MCU output the correct data when it is run. I can verify that the data is output by using an oscilloscope and UART is connected to a Linux device that receives data correctly.

I use IAR Workbench for MSP430 IDE v6.10.7, common components v7.0.5.3137. When I run the below code on IAR, it executes the code but does not output any signal at the UART port, so that I don't see any data on the Linux end device does either. I use the same settings on both IAR or CCS usage. Could you please tell me why the code does not work on the IAR or if I need some special definition or library for code to work on IAR?

I am using one of TI's EVM based on MSP430F6736. I am connected it on JTAG connector via one of TI's debugger MSP-FET430UIF.

Here is the build output

Building configuration: j - Debug 
Updating build tree... 

4  file(s) deleted. 
Updating build tree... 
main.c  
Warning[Pe111]: statement is unreachable C:\Users\user\Desktop\main.c 36
Linking 

Total number of errors: 0 
Total number of warnings: 1


Here is the code

#include <msp430.h>

void mUART_init(void){
      WDTCTL = WDTPW + WDTHOLD;                 // Stop WDT

     // Setup P1.2 UCA0RXD, P1.3 UCA0TXD
          P1SEL |= BIT2 | BIT3;                   // Set P1.2, P1.3 to non-IO
          P1DIR |= BIT2 | BIT3;                   // Enable UCA0RXD, UCA0TXD
        UCA0CTLW0 |= UCSWRST;                   // **Put state machine in reset**
      UCA0CTLW0 |= UCSSEL_2;                  // SMCLK
      UCA0BRW_L = 6;                          // 1MHz 9600 (see User's Guide)
          UCA0BRW_H = 0;                          // 1MHz 96000
     UCA0MCTLW = UCBRF_13 | UCOS16;          // Modln UCBRSx=0, UCBRFx=0x13,
                                            // over sampling
      UCA0CTLW0 &= ~UCSWRST;                  // **Initialize USCI state machine**
}

void mUART_send(uint32_t send){
    uint8_t MYtx[4];
      MYtx[0] = send;
      MYtx[1] = (send >>8);
      MYtx[2] = (send >>16);
      MYtx[3] = (send >>24);

     int i;
    for(i=0; i<4; i++){
         while (!(UCA0IFG&UCTXIFG));             // USCI_A0 TX buffer ready?
         UCA0TXBUF = MYtx[i];                  // TX -> RXed character
     }
}

int main(void){
    mUART_init();
        while(1)
          mUART_send(0xFEAB);
    return 0;
}

Project > options: MSP430F6736 is selected, code model is large, data model is small, no optimization is selected. Normal DLIB is selected