MIPS assembly programming Q&A

This document was produced to answer some of the (repeated) 
questions about programming in MIPS assembly language and using 
the SPIM simulators. 

Q: How do I input and output numbers?

A: The spim/xspim simulator provides a simple I/O facility through 
the syscall instruction. Section 1.5 in the documentation explains how 
to use syscall to do I/O.

Q: My program works correctly.  However at the end of a run it gives 
me an "Attempt to execute non-instruction at 0x00400090" .
Because my loop ends the program, I assume that there must be 
some code that tells spim to end the run.   How do I exit the 
program?

A:  There are two ways to exit a program:

1. use the syscall command to exit the program (see section 1.5)

2. Save the content of $ra register and use "    jr    $ra"  to exit.   
When you enter your program $ra holds the return address for the 
operating system code that initiated your program. In effect  running 
your program looks like a procedure call to the OS. Returning from 
that procedure is equivalent to exiting your program.

Q: I get what looks like an endless  loop with a whole bunch of 
statements like these:

Exception 7  [Bad address in data/stack read]  occurred and ignored
Exception 4  [Unaligned address in inst/data fetch]  occurred and 
ignored
Exception 5  [Unaligned address in store]  occurred and ignored

What does it mean?

A:  
> Exception 7  [Bad address in data/stack read]  occurred and ignored
 
It means that your program is has a load instruction  with a bad 
address. That address is not inside the data or stack segment of the 
address space. Look at your loads and see which instruction does not 
use a good address. I suggest using breakpoints or single step 
execution, looking at all load/store instructions checking what 
addresses your programs is using.

> Exception 4  [Unaligned address in inst/data fetch]  occurred and 
ignored
>Exception 5  [Unaligned address in store]  occurred and ignored

This means that your load/store instructions are loading a word but 
the byte address is not on a word boundary. Remember!  When 
loading a word, the lower order two bits must be 0 (The address 
must be divisible by 4!) Look at your code, you must be incrementing 
the address by 1 instead of 4. Also use  the ".align 2 " directive to the 
assembler.

Q: What is wrong with : "  lw $t2, 0x100000100($t1)"

A: You are using a 32 bit number in the immediate field. This does 
not work since the immediate field is only 16-bit long!!! I am not 
sure what exactly happens here, but it can't be right!

Load the address into a register first and then increment the register 
to address consecutive words that is, use:

      la    $t1  0x100000100

to load the start address into register $t1. Increment register $t1 
by 4(!) to address the next word.

Q: I have finished and submitted my program, called Fibonachi, and 
it works.  However, After I have loaded and run the program once in 
xspim then xspim doesn't seem to run the program correctly a 
second time.  I do not know why this would occur and I checked over 
my program and found no errors, but I was wondering why this 
would occur.  The program runs fine if xspim is quit and restarted 
and then the program is loaded once again.

A:  The load "file" command in the spim simulator documentation 
says:  " If the file has already been read into SPIM, the system 
should be cleared befor loading (see reinitialize, below) or global 
symbols will be multiply defined."

I suspect that you loaded the file twice and have multiply defined 
symbols.