BlueIllusion V0.031

[distantvoices]

Looong it's been now since the last test release of my project.

Disk Image

But now it is here:

It should boot up with some nice startup comments and end with "You are welcome for an experiment!" and the prompt. The floppy should do some work for a certain time and then stop spinning. (actually Blue Illusion "mounts" the floppy - fetches the superblock and the group descriptor table and thats it - this one is async at the moment, the user shell is not blocked! so attention attention.)

My floppy code refuses to work in the bochs and I am sure it is due to some error of mine but alas i lack the hankering to search for it. The Keyboard has a kind of bug. the first command one enters isn't recognized, but after that it works fine. Maybe there slips a char into the queue where no one shall slip in.

System requirements: at least 64 mb ram installed. a vesa 2.0 video card. a floppy drive. A PS2 Mouse. A Keyboard.

Why that much ram? Ha, it's hard coded in the start up parameters to assume that this muchmemory is installed. This will change in future releases. I'm just lazy to do it now.

What's new: Open, Read, Close a file, chosen by file name: Best you say open and '.', then read - it should spit out the contents of the root directory in a not correct way, and after that close. its all hidden behind the command 'floppy'

Well ... For your tests I thank you very much as well as for your comments. would be fine to receive some critics, comments, input.

stay safe

ps: source code will follow as soon as I've done the clean up.It is a mess as usual.

[Pype.Clicker]

great. i just got a laptop to run tests ... i'll be trying your image asap...

[BI lazy]

ad Toshiba laptops: I don't have a clue why, but the vm86 code craps out as soon as I try to query the VGA Bios for the available vbe modes on my Toshiba Satellite 1700-300.

If sbdy knows what is up with those machines, pls let me know.

[Pype.Clicker]

Looong it's been now since the last test release of my project.

Disk Image

The Keyboard has a kind of bug. the first command one enters isn't recognized, but after that it works fine.

i can see a

Code: Select all

00010894988e[KBD  ] controller_enQ(): OUTB set!
00010895223e[KBD  ] controller_enQ(): OUTB set!
00010895223e[KBD  ] controller_enQ(): OUTB set!
00010895223e[KBD  ] controller_enQ(): OUTB set!

on bochs output, maybe it'll help you a bit ...

ooh, and, btw, where can i submit an AZERTY map instead of the QWERTZ one you have :-p

[Pype.Clicker]

a few things:
- the F12 (process table) doesn't look like working
- what's the spinning blue char in top-right corner of the screen ?
- what are the files that i can try to read from the floppy ?
- the [backspace] key erases characters past the shell prompt (i.e. you can erase the whole screen if you're patient enough ...)
- no vbe2.0 available on bochs (doh)
- TEST:newcon is interpreted as "test.newcon" which is not recognized
- what does "Na damit wird des nix" means ??
- i can't find digits on the keyboard (1->7, 2->unhandled e0/e9, 3&4=>?, 5=>9, 6=>unhandled a7 (by bochs ... do), 8->1, ...)
i guess there must be something wrong with my bochs keyboard configuration ...

[distantvoices]

First of all thanks for testing, Pype!

the Process table key ... I've finally implemented it in a veeery hackish manner - well - for a first startoff it is sufficient to just have the system thread put the contents of the process queue on the screen, but it will be definitely change: the output will be put to a buffer from where it is copied to the use rprocess which requests it. - in form of a very long null terminated string so to say.

vbe2.0 in the bochs ... well the latest vga_lgpl_bios should do the work but alas, my implementation doesn't yet work with it.

the Keymap ... hmmm, maybe you send it to me and I'll have a look at how to build in a switch in order to change the keymap at needs? *ggg* shan't be any problem.

files to try: the root directory I recommend: just enter . as filename. For directories I've already a kind of reader, which fetches the entries one by one. another one is fsserver.bin, but I don't think the read command will like it very much as for now. There is nothing yet which checks whether a directory is read or a file.

the spinning blue char in top right corner of the screen indicates that the idle thread is veeery busy doing nothing. It will vanish as soon as the system has got enough stability and tools to be called a system instead of a mess. *lol*

About the backspace thing ... oh sweet jesus yes, I should indeed do something about this, because it leads to funny results indeed.

and "na damit wird des nix" is "No, this won't work"

No,nothing is wrong with your keyboard konfiguration in the bochs, but many things are wrong with my keymap: some holes are in it for keys where I've been yet a wee bit too lazy to fill sthg in.

'TEST:newcon' -> just enter newcon. in Text mode it won't do any thing, but in graphics mode it invokes a new gui-shell process.

Thank you again!

[Pype.Clicker]

on my newly-old toshiba laptop, GRUB loaded stuff fine, then i have:

Code: Select all

BlueIllusion OS - INIT
INIT: add driver tasks...
pager init ...
freier physik. speicher: frames - 4000
hoechste_addresse: 4000000

then the system looks to freeze, keeping a blinking cursor after "4000000"

no keypress looks helping ...

[distantvoices]

It does this to me too if I dare to boot it on my own toshiba laptop. I 'll have to look after this one, because it's too early for the vm86 task. After the mm-driver the video driver should initialize. There is definitely something unfriendly underway.

But I wonder for it boots on different 'normal' pcs without problems. hm ... no one has said that osdev is a thing full of easyness

Thanks for testing, Pype!

[Pype.Clicker]

imho, you should consider announcing what you're going to initialize before it could possibly fail, like

Code: Select all

 
   kprint("INIT: %s...", driver[i].pname);
   if (driver[i].init()) kprint("done\n");
   else {
      if (driver[i].flags&KDRIVER_OPTIONAL) {
         kprint("doh!");
         continue;
      } else {
         kprint("abort");
         panic();
      }
   }
}

[distantvoices]

well, a new release is out again, and I reckon, this one is one of my interesting ones, thou: the more BlueIllusion advances, the more bugs are hidden in the back ...

Well, I've set up a dedicated realm for it:

www.distantvoices.org
-->edv --> os


You'll find the latest disk image as well as the latest source code tree there in section Downloads.

Try out typing 'fork' and 'fork y', to check out some difference in behaviour. you may also type 'help'. at first the floppy drive should spring to life and fetch some blocks and after that, it should display a nice help text -> works only on console 1, for I don't yet pass parameters to the exec'd program. That's the next step: **argv and **env. If you type in anything, it should look in the actual directory for the executable you wanna execute. One thing: in the shell, first type "floppy",choose open and enter "." at the prompt. It will open the root directory of the floppy. After this, you can also say "ls" and "cd [dir]". cd doesn't yet work with this: "dir/dira...". I don't have implemented something small but crucial yet, which faciliates entering /&% and so forth.

Unfortunately, BlueIllusion still refuses to work on Toshiba Laptops, and the fdc driver has issues with the Bochs. But bug tracking in this direction is under way. I hope I find those damned things.

Thanks for feedback.

[Slasher]

You have your Z and Y key in the wrong places. pressing z key shows a y while pressing y shows a z

[BI lazy]

*gg*

That's because I am using a qwertz-layout.

Later, I'll implement loadable keymaps.

Thank you for testing, code slasher!

[Pype.Clicker]

dude! your disk image is damn long to write to a floppy ... what do you have in there ??

tested the stuff on my laptop again. MMSRV detected VBE2.0, then i got twice "kalibrieren... kalibrieren fertig... FLOPPY:3 alles ok ..."
and then a

Code: Select all

bestehender prozess - nr:64 adr:46207175 pgdir: f25067
exception ... : 13 in proz: 8
k->gs:23, k->fs:23 k->es:23 k->ds:23
k->eax:0 k->ecx:b000fde8 k->edx:2 k->ebx:0
k->esp:b000fee3, k->ebp:0, k->esi:b000ff8b, k->edi=c0023320
k->eip=129f, k->cs=1b, k->eflags: 13286
k->u_esp=b000ff33, k->u_ss: 23

cs:eip pairs seems weird ... is this in pmode ?

[distantvoices]

Hehe, that's because I 've just "cat"'ed the whole /dev/fd0 disk to a file. Thats ugly, I know. 've heard rumors 'thou', that there exists this bochs tool which is capagle of making disk image from nearby anything, so ... i shall see ...

Your output stems from Infringement of C-Commandment Nr. 1: Thou shalt not follow the NULL pointer. I just didn't set some list start elements to NULL upon init of the list. It is fixed, but I canna upload for my provider does some web server updating stuff for the next 24 hrs.

The error comes in protected mode and stems from a pagefault of a process, which has already allocated his adress space and traps into page fault exception because it tries to do something the system just doesn't like: trying to access an adress in nowhereland. MM service at the moment just looks whether the adress is present in the allocation tree or not, and then traps on a 'hlt' - thats the exception nr 13 you are quoting - instruction - because I 've not yet focussed on finishing the page fault handling stuff for already allocated adress spaces. Lazy me. It should add pages to adress space if the pagefault is in a valid adress area and it should kill the faulting process in case of an illegal access of memory.

What kind of graphics chipset do you have in your ol' toshiba laptop. I have an ati rage128 mobility graphics chipset, and this one does not support 'int 0x10' calls. Upon attempting such a call it lands directly in silicium nirvana of Unknown Instruction Exception.

Thanks for testing!

[Pype.Clicker]

"Chips And Technologies" graphic card (never heard of ...)
i can't find any chipset infos through that ugly mswin95 (no need to say i haven't reinstalled a system myself on it :p)

it's the toshiba 300CDS model, pentium 1 with 48 MB of RAM ...