As Neil mentioned - where do you operate the chip select signal? You note that it is tied to 0v - I think you'll find the the cs signal is critical as the internal logic uses it to reset and latch the data the was shifted. Also, why do you have assembler code to do the shifting? Why not do it in 'c' and not create the potential for more problems? Since we don't know what compiler you're using - how can we check that you're passing the parameters in the correct registers to the assembler code?

I think once you control the cs signal, at least one of your problems will disappear.