Wow, I haven't posted on this board in ages.

I'm having some trouble with some newly added sector loading code in my OS. It compiles fine, but when you try to load sector 2 on a floppy disk, Bochs panics with the message "00068675875p[CPU0 ] >>PANIC<< prefetch: RIP > CS.limit". Some links to the code:

Kernel code: http://protonix.asmhackers.net/protonix/v2/build1/devel/NEWPRO.ASM

Program on sector 2 (that i'm trying to load): http://protonix.asmhackers.net/protonix/v2/build1/devel/test.asm

Bochs log: http://protonix.asmhackers.net/protonix/v2/build1/devel/log.txt

Project website:
http://protonix.asmhackers.net

Project forum:
http://protonix.asmhackers.net/board

Thanks,
Anthony

Hi,

I didn't check your code yet, and I don't use bochs, but from the log:

read() returned 0
RIP > CS.limit

The read() returned 0, this indicated success? I think bochs might keep track of CS:IP in one variable, and calls it RIP (Real Instruction Pointer?). From my point of view it looks like somehow this RIP exceeds any address addressable with the CS segment.

You could try the following options to see if it helps:
- output some text after a successfull read, but before the jump, to see where the error is
- debug your code to see if the jmp far turns out the way you planned it
- reset the file system before each read retry (the way it's documented )

Good luck!

OK, the sector loads fine, so it's either the jump or the code that I'm jumping to.

Oh, I thought that it meant "rest in peace CS limit.."

I used to use another assembler, which couldn't handle



Instead, you had to



Thats why I brought up the "debug to see if the jmp was assembled correctly". Code looks fine to me, after a successfull read there cannot be any operation but the jmp.

I am assuming that RIP is a part of AMD64 (for my own edification)?

I use two pushes and RETF to jump to the offending code that I've loaded from my boot sector:



I may have these reveresed and backwards (depending on your ORG for your code to load).

Awwight what version of bochs do you use? I've searched for your error message a bit, found a changelog for bochs 2.1 that says:

deltre:

It looks like 2.2 Pre 3:

Thanks. I'll try those things when I get home.

None of the alternate "jumping" methods worked. Any ideas?

Well, I've finally gotten everything sorted out. Turns out that iwas just a problem with the int 0x13 options. So, the OS was loading the sectors and jumping OK, just to the wrong sector.

Anyway, it's all fixed now, and I released version 512DevOS 0.02 (formerly Protonix and BS/OS) with sector loading support at 12:07AM EDT this morning. Don't ask why I did a midnight release; I just figured out the problem (with help from Benjamin David Lunt) that late.

So grab a copy at http://512dev.us.tt
Be sure to join our forums, no new members/activity in a while...

I like to code assembler by moonlight. lol j/k