Tips & Tidbits

Edited by Scott Boyd, Editor

Tip Of The Month

MPW C Navigation Time Saver

MPW has a nice feature for C programmers: Using the tool CMarker, you can make a list of marks for all the functions in a source file.

Unfortunately, when you open the Mark menu, you can wind up with something unusable. For example, the list could be so long that it takes too much time to find what you’re looking for. Moreover, you are often looking for a function because you are in a piece of code in which you see it being called. It’s right there in front of you. It would be nice to be able to use that piece of context to get you to the function itself.

The following MPW Shell statement sets up a function key which jumps to a function when you highlight its name and press control-enter (or any other key you wish).

/* 1 */
setkey control-enter 'find `catenate "{Active}".§`"{Active}"'

Be sure to get all of the quotes right. MPW cares about the different quotes. Reading from left-to-right, the quotes are straight-, back-, double-, double-, back-, double-, double-, and single-quotes. The § character is option-6.

- Manuel Carlos Alcoba

INLINE ASSEMBLY IN PASCAL

Knowing how to write inline assembly within your pascal code can sometimes save you time and trouble. For example, I use an Inline routine that returns the current stack address for debugging purposes. The routine looks like this:

/* 2 */
INTERFACE
  FUNCTION GetStackTop: LongInt; INLINE
   $2E8F;
My code calls it just as if it were a regular function:
VAR
 stackAddress: LongInt;
...
stackAddress := GetStackTop;

When used properly, Inline assembly provides considerable speed improvements and allows you to enhance your Pascal code in ways not possible with straight Pascal.

Define your inline routines in the INTERFACE part of a file. In my example, the hexadecimal number $2E8F represents the instruction MOVE.L A7,(A7). The compiler places this instruction directly into the code. This reduces the number of instructions needed to write and compile this routine in assembly language and also avoids calling overhead for a function. No calling overhead simply means no JSR instructions are generated in the code. When I was learning how to write inline assembly, the biggest obstacle was lack of documentation on how to obtain the hexadecimal format (e.g. $2E8F) of assembly language instructions. You can laboriously try to compose the hexadecimal instruction from some 68000 assembly books, but this usually takes several minutes, is error prone, and requires that you know exactly what you are doing. There is a much easier way to obtain the hexadecimal format of any assembly language instruction, provided you have MPW assembler. This is what you need to do:

Suppose you want to find the hexadecimal format of the MOVE.L A7,(A7) instruction.

1) In MPW, create a new file called Example.a and enter the text:

/* 3 */
RoutineName PROC
 MOVE.L A7,(A7)
 END

2) Create a ‘.o’ file with the MPW ASM tool. In this case I call the file Example.a and the ASM tool creates an Example.a.o file:

/* 4 */
 Asm Example.a

3) Now disassemble the Example.a.o file with the dumpObj MPW tool. The dumpObj takes whatever is inside a compiled file and disassembles it back into assembly language instructions. In this case it disassembles the file Example.a.o into a file called Example.o.a:

/* 5 */
dumpobj 'Example.a.o' > 'Example.o.a'

4) The disassembled file contains the line

/* 6 */
00000000: 2E8F           '..'              MOVE.L     A7,(A7)

The second number on that line, “i.e. 2E8F”, is the hexadecimal format of the assembly instruction.

From now on, anytime you need a hexadecimal format of an instruction, just add the instruction to Example.a file and execute the ASM and dumpObj MPW tools. After you do this a couple of times you will realize this is a simple repetitious process. This usually takes less than a minute and is much easier and quicker than finding the instruction from a book.

- Marek Hajek

Another hex-for-opcode trick

Finding the hex for an instruction can be useful in a number of situations. One situation I’ve found myself in is sitting in Macsbug on a machine and needing to change a single instruction. For example, suppose Igot a branch condition backwards, and want to change the sense of the conditional branch from BLE to BGT. I’d like to change the instruction with a SM (set memory) Macsbug command, but first I have to figure out the value to set. I could guess (don’t laugh, I’ve done it before, and even got it right after just a few guesses on a couple of occasions), but a better way is to let someone do all the guessing (or table lookups) for me. The fastest tool I’ve found is TMONPro on some handy nearby machine. Here’s how:

1) Open up an assembly window. Set its address to PlayMem. That’s a chunk of memory that TMON sets aside when it loads. It’s there for you to play with.

2) Select the first line. It probably reads something like ORI.B #$00,D0. Regardless, select the whole line.

3) Now type in the instruction you want. It replaces the selection. Now press return or enter.

4) TMONPro assembles your instruction right then and there.

5) Open a memory window and point it at PlayMem. An easy way to do that is to command-click on PlayMem’s address over on the left-hand side of the assembly window. The memory window now shows the hex values for the instruction you just typed in.

Of course, you might also just use it on the machine you’re debugging. That’s even faster because now you can simply type over the line of assembly, press enter, and the code changes in place.

- Ned Flanders

Playing a sound asynchronously and continuously

When allocating your channel, pass a callbackproc. When playing the sound using bufferCmd, immediately play a callbackCmd as well. Your callback will be called when there is about 50ms of sound left to play, and can queue a new bufferCmd in the same channel (even though it's at interrupt time).

- Jon W{tte

[When asked about the { in his name, Jon replied :

“For reference, it's option-u a, or a-umlaut - ä. However, I'm trying to push the brace as the next character to be included in the Swedish naming character set :-)” ]