Data RAM access changes

[douggilliland]

@mjgpy3 I changed the Data RAM accesses to be byte/word/long addresses.

It should fill the exact same way from the assembler but the labels are off by power of 4 in their location/value.

For instance: firstVal: .long 0x12345678 secVal: .long 0xDEADBABA

secVal should be at address 0x4 (worked before but not after my hardware change) but it is 0x1.

00000020 42800000 lix r8,firstVal.lower 00000021 42800001 lix r8,secVal.lower

Can you shift the data label values left by 2?

The result of the above code would be:

00000020 42800000 lix r8,firstVal.lower 00000021 42800004 lix r8,secVal.lower

The reason I did this was that it was too painful to pack/unpack bytes into longs. For instance, readLine got really painful when I wanted to implement backspaces. It had to backtrack differently for longs that were already stored. If it's bytes then it "just works".

[mjgpy3]

@douggilliland does this look right for hello world?

:0400000048656C6C77
:040004006F2C2057E6
:040008006F726C6443
:04000C0000000000F0

[douggilliland]

@mjgpy3 I can't remember where hello world is ;)

[douggilliland]

@mjgpy3 I check in C073 which shows the problem.

[douggilliland]

It still doesn't work.

firstLong: .long 0x00000001 secondLong: .long 0x00000002

start: 00000000 42800000 lix r8,firstLong.lower 00000001 42800001 lix r8,secondLong.lower

s/b:

firstLong: .long 0x00000001 secondLong: .long 0x00000002

start: 00000000 42800000 lix r8,firstLong.lower 00000001 42800004 lix r8,secondLong.lower

[douggilliland]

Something about one of the changes broke other stuff.

I only get part of the prompt now.

[douggilliland]

@mjgpy3 Weird since I see it in the data file so it has to be something I broke.

[douggilliland]

@mjgpy3

[douggilliland]

@mjgpy3 The data should be contiguous.

I think when you did one of the left shifts it should not have been done. The one for the address of the label was correct to do (the last one you did). But a previous one wasn't.

[douggilliland]

@mjgpy3

I think line 428 was the commit that broke it.

[douggilliland]

I backed out the change to line 428 and it works. Committed.