RTOS/MSP430F5438A: SPI code problem ... UCB1RXBUF always zero

Tool/software: TI-RTOS

I am having a problem writing SPI code for my MSP430 to talk to a FLASH chip.

I've tried writing both a polling routine, and an Interrupt ISR, and result is the same.  No matter what I try, UCB1RXBUF is always zero.

Here is my init routine:

void hal_spi_init(void)
{
    ////////////////////////////////////////////////////////////////////////////
    // Here are all the relevant IO Pins:
    //          FLASH_SS                P3.6        UCB1STE/CS#     (active low)
    //          FLASH_MOSI              P3.7        UCB1SIMO/SI
    //          FLASH_MISO              P5.4        UCB1SOMI/SO
    //          FLASH_SCLK              P5.5        UCB1CLK/SCK

    P3DIR |= FLASH_SS;                      // P3.6 FLASH_SS is an output pin
    P3OUT |= FLASH_SS;                      // active low, (disable)

    // set special mode for MOSI (P3.7), MISO (P5.4), SCLK (P5.5)
    GPIO_setAsPeripheralModuleFunctionOutputPin(GPIO_PORT_P3, FLASH_MOSI);
    GPIO_setAsPeripheralModuleFunctionInputPin(GPIO_PORT_P5, FLASH_MISO);
    GPIO_setAsPeripheralModuleFunctionOutputPin(GPIO_PORT_P5, FLASH_SCLK);

    // Initialize SPI in Master mode
    USCI_B_SPI_initMasterParam param = {0};
    param.selectClockSource = USCI_B_SPI_CLOCKSOURCE_SMCLK;
    param.clockSourceFrequency = UCS_getSMCLK();
    param.desiredSpiClock = 1000000;
    param.msbFirst = USCI_B_SPI_MSB_FIRST;
    param.clockPhase = USCI_B_SPI_PHASE_DATA_CAPTURED_ONFIRST_CHANGED_ON_NEXT;
    param.clockPolarity = USCI_B_SPI_CLOCKPOLARITY_INACTIVITY_LOW;      //or USCI_B_SPI_CLOCKPOLARITY_INACTIVITY_HIGH;
    bool rc = USCI_B_SPI_initMaster(USCI_B1_BASE, &param);
    assert(rc == STATUS_SUCCESS);

    // Enable SPI module
    USCI_B_SPI_enable(USCI_B1_BASE);

    // Disable interrupts
    USCI_B_SPI_clearInterrupt(USCI_B1_BASE,
                              USCI_B_SPI_RECEIVE_INTERRUPT|USCI_B_SPI_TRANSMIT_INTERRUPT);
    USCI_B_SPI_disableInterrupt(USCI_B1_BASE,
                              USCI_B_SPI_RECEIVE_INTERRUPT|USCI_B_SPI_TRANSMIT_INTERRUPT);
}

Next, here is the spi_transfer routine:

void hal_spi_transfer(
        uint8_t *txbuf, uint8_t *rxbuf, size_t txrxlen, uint8_t flags)
{
    if (flags & SPI_CS_ENABLE)
        flash_ss_active();

    // initiate the first byte transmit
    for ( ; txrxlen; --txrxlen) {

        uint8_t tbyte = txbuf ? *txbuf++ : 0;

        // USCI_B_SPI_transmitData(USCI_B1_BASE, tbyte);
        UCB1TXBUF = tbyte;
        UCB1IFG &= ~UCRXIFG;     // Clear RXIFG as transmit always sets the RX flag

        // wait until the RX byte is completely received
        while (!(UCB1IFG & UCRXIFG))
            ;

        uint8_t rbyte = UCB1RXBUF;        // USCI_B_SPI_receiveData(USCI_B1_BASE);

        if (rxbuf != NULL)
            *rxbuf++ = rbyte;
    }

    if (flags & SPI_CS_DISABLE)
        flash_ss_inactive();
}

When this routine is called, only zeros are read.

On the logic analyzer, I can see that data is properly on the SPI bus.   In the following example, we send the FLASH_READID command (hex 9F), and then read the 0x56 bytes of the chip configuration info:

    hal_spi_send_byte(FLASH_READID, SPI_CS_ENABLE);
    hal_spi_transfer(NULL, flash_CFI, sizeof flash_CFI, SPI_CS_DISABLE);

On the Logic Analyzer, we see all the SPI data:

And we can see on the blue line (probe 6) that the Flash chip received the command, and is returning the CFI data.   I can zoom in and examine the data, and it is absolutely correct.

I've also put an oscilloscope probe on the CPU pin (pin 51), and I get the same result.  The voltage levels look ok  (3v or so).

However, the 'flash_CFI[]' array is always zero.

Is it possibly my hardware is damaged and the SPI receiver doesn't work?

-allan schwartz