AVR Freaks

I wrote some simple SPI slave code a while back that has worked fine for some time. I use an atmega32u4 and have it wired to a harness. This Harness connects to a master SPI device. I had to remake my harness as I lost the old one. Now the pair will not communicate.  So logically, I broke out my analyzer to have a look. I have attached a screen shot showing this.

The code I run matches what we see here..

	command=Spi_read(1, 0xff);
	Spi_read(1, 0x41);
	Spi_read(1, 0x5a);
	Spi_read(1, 0xF1);
	Spi_read(0, 0xF2);

Here is the SPI function for reference.

unsigned char Spi_read(char ack, unsigned char data)
{
    SPDR = data;
    while(!(SPSR & (1<<SPIF)))
    if (ack)
    {
        ATT_ON();
        _delay_us(1);
        ATT_OFF();
    }
    return SPDR;
}

When I debug out my command I get 0xff. If I put my command= on the next line

i.e.     command=Spi_read(1, 0x41);

I get 0x41 and same for the other 3 lines of code. Command should not be the data I'm sending. It should be the reply but it for some reason is showing the data sent to the mater.  I verified the SPI function is waiting for the communication to end. So the SPI function seems to be running successfully. I can see its sending the right data so the miso/mosi are not swapped and I have checked that harness many times over. From a practical sense it almost has to be the code, but it never changed. I reverted from my SNV....  What on earth would make it echo back what I send where the analyzer clearly shows that's not what comes back? IS there anything that would make the SPDR register not load the data from the temporary register?