OSDev.org

I am glad to announce after some months of work - and more debugging sessions than good - that I've been able to stuff together a nice release of my Toy Project with the funny name - BlueIllusion. Here is the web: www.distantvoices.org. Click on blueillusion os - link, and there you are.

This Time, we have some quirks too, some olden quirks aren't yet resolved, but some features I've added: IO streams, the virtual consoles are working finally, and on an exception which canna be solved neatly, a process gets killed instead of a general system halt.

And Semaphores we have. Give the "semtest" tool a try. It causes some aftermath which I have to resolve, but it neatly shows the functionality of the semaphor ipc mechainsm.

Further you may try out the "argvtest" with some command line arguments and look what happens.

The GUI is disabled in this release for I've got to design some facility to redirect output streams to the GUI service. (Pipes it will be...)

One Note: Once you 've started a program and fs service fetches the data from disk (once cached, loading is quick, no sorrow there), don't start another program for this screws things up in a really queer way. I am working on multithreading and once that's done, such issues aren't problem anymore for I've got some neat ideas about scheduling of IO requests and worker threads...

@the_kung_f00_master: Check out the source code for three files: they each have an addendum like "with_english_transl" in their names. I've placed comments with translations from german to english in som functions which are important for vm86. - beside the vm86 process image containing the 16 bit code. Hope the comments are helpful.

Bugfix release online: The nasty fs service bug is resolved. Now it works without any problems.

)

Looking forward to your comments and critics, I wish you happy hoppy easter.

@pype: is it pasques en francais for easter?

stay safe!

I thought it was Paques. (but I'm not quite sure )

It is indeed "P?ques"
btw, i've been giving the latest disk image a test and it looks to stop its execution at


the 'spinning character' at topright corner goes on, characters get displayed and F11 lists running process (i guess so): idle,mmdrv,video, system, timer, pipedrv, fdcdv, mmsrv, no_name, fssrv and two other 'noname'.

Is there something more i can do/should expect ?

thanks!

as for the bug report: thanks, I appreciate it. Have to figure out this one for it looks like it starts up properly, but then, the first process started by init seems to land in the land of silicium nirvana.

I expect it to start up and then drop you at a console with: "You are welcome for an experiment " and a ">"-prompt.

You test it on that old toshiba laptop, do you? Maybe it's a kind of memory issue? I *stupidly* have hardcoded 64 mb into the mmdrv. This *can* be an issue if using that thing on a computer with less than 64 mb. *gnarfl* need to build in some memory calculation stuff.

stay safe

Yes, you should do that, it doesn't require that much code

Well, it teaches: the more diversity of testing hardware you have at hands, the earlier you discover your bugs

currently, I am sitting here, my own ole toshiba laptop in front of me running BlueIllusion (and it runs for that machine has 128 mb ram) and I am thinking about how to reproduce this error. The smallest entity of RAM I have at home is a 128 mb SDRAM. *grin*

@Dennis: right you are. This being comfortable in existing solutions doesn't give a good bang of the buck.

stay safe.

Good point. That was indeed the laptop (the only machine i can use for OS testing atm.) which barely has 48MB ...


could be helpful ;D

Hi

I have just tested Blueillusion OS on my:

900MHz Athlon t-bird, 128MB RAM, Voodoo3 2000
and
Toshiba Laptop, 486DX-4 75MHz, 8MB RAM, C&T 65545
and
bochs

on the t-bird, it gets to:

And an 'm' is placed in the top right hand corner of the screen, then it just stops, Keyboard LEDs don't change, nothing.

What's fertig in english BTW?

On the laptop:


First it gets an invalid opcode and then a GPF.

There are some very odd code and stack selectors in these!

oh! oh! also, on bochs, I get to:


And then nothing. Like what pype says, hitting F11 shows processes

I have megs=64

Hope this helps
srg

done.
Example: Es is fertig, it means: It is done/completed

@dennis: thats correct

@srg: Hmmmm ... seems like it is caught in a page fault. It is very likely, that my paging subroutines still are buggy for I have started some development in the realm of the pager.

It is a bit troublesome with the laptops. a Program which loads perfectly well on three computers (on a toshiba laptop with ati graphics and no vga bios and with 128 mb ram, on a asus board with athlon xp1700+ and 512 mb of ram, and on some pc at work) but craps out on others ... *shakes head*

In the bochs I don't seem to be able to read/write sectors from/to floppy. I don't have put efforts in finding the bug there.

The weird selectors stem from a process sent to nirvana. *gg* either by an overwritten page or by a pointer shoving the cpu to nowhere land.

Thanks for testing, it is very appreciated for I lack the diversity of hardware needed to clear out all the quirks and sorta ...

stay safe

@srg&pype:

Could you please be so kind and fetch the image I've uploaded a few minutes ago and give it a test on your machines? (srg: most interesting is the amd machine - t-bird ) It would help me very much.

What I have changed/discovered: I've omitted a crucial initialitation in the init process - and have sent the execve call into holy nirvana. I wonder why it didn't crap out on the three machines I am testing here nor in the bochs (with disabled floppy access *gnarfl*).

What you can expect: a proper boot sequence:
fssrv:init ...... fertig.

blue char starts spinning (now the idle task is running) and floppy starts scratching the disk again: init forks off the shells.

then, if everything is happy and ok, it should drop you to a shell:

BlueIllusion OS v0.034 - in 1, out 2
You are welcome for an experiment

>_

It is a big pity, that always the smallest things make so much difficulties and that one doesn't has each and every debug possibility at hands.

Thank you in advance, lads!

PS @srg: with laptops I still have some trouble. And with only one at hands, it is a bit difficult to remove all the quirks and mistakes causing it to fail.

Right

I've tried it on the t-bird and the only difference is that rather than an 'm' in the top right hand corner of the screen, there is an 'l'. Appart from that, no change.

BTW Does the fact that on that machine, both it's hard disks run from a Promise Ultra100 TX2 IDE card and NOT the onboard IDE. I know that it's loading from floppy, but do you try to read or write the the hard disk's controllers?

Also, what file system is the disk formatted in. Windows XP can't read it and on my main machine, bochs can't read it either, so I have to use the image rather than my written disk.

(Like Tim, I use Windows and cygwin rather than Linux)

srg

I've test it on 3 other machine, all failed I'm afraid

My Main laptop:
Athlon XP 1400+, 256MB RAM
Gets to fssrv: init........ fertig and then stops, not even the character!

My Main machine:
Athlon XP barton 2500+ @ 3200+ 512MB RAM
My Socket 7 Machine:
K6-III 450MHz, 128MB RAM

On these, it gets past fssrv: init........ fertig, and I have a charachter in the top right hand corner whizzing very quickly. There are some weired floppy drive noises (although I'm taking them to be normal) and then...............reset.

Your not having much luck with my machines

Hope This helps
srg

This shows me that there are quite some issues to clean out, which I couldna find on my own machines at home.

I'll take a look, whether I can get floppy of my fourth computer running and accepting a disk. It is an amd thunderbird and maybe it reveals some of the quirks.

weird floppy drive noises: at least it reaches the init process and starts loading a program (shell.bin)

By the way, I am using ext2 file system.

As for the hard disk controllers: no, I don't do anything with them yet. This is some task I am delayint until I've got the basic OS clean and debugged and stuffed with goodies like semaphores (already present), signals (active development atm).

Gonna activate some debug stuff, which might be annoying as hell (output of page faults f. ex., and context dumps). This way, I might have a chance to find the bug, because there is one, that's sure.