OSDev.org

the 'vertical retrace and mouse' discussion has beeing copied in a dedicated thread.

New Version of BlueIllusion online:

Source Code

Disk Image

Known caveats:

1. the VESA 2.0 graphics doesn't work in the bochs. so it falls back to 320*200 .
2. In the bochs, the graphics server/video driver combo doesn't perform well, tough on any real machine it rocks - if that machine talks VESA. *gg*
3. On toshiba laptops it is not likely to work: something with the vm86 mode doesn't work well with them. Maybe i am getting something wrong with toshiba laptop bios and it's locations in memory.
4. Moving the mouse prior to entering any key directly after startup causes the keyboard not to work anymore. I 'll track this down.
5. Be careful of switching back and forth between text mode and gui mode or trying to invoke the gui again, when operating in gui mode already (I REALLY should fix this).
What is not yet tested/implemented properly:
1. process exit. I don't have buildt a proper framework for this yet, but I gonna need it.
2. Process loading: at the moment, I rely on grub to load files for me and put them in a location in memory.
3. filesystem: None yet. I gonna implement ext2-fs. No own design at the moment .

Stay safe !

ps: typing help after start up reveals some secrets ];->

nothing happened
booted, i typed gui... and got a black screen and that was it....

this is with current vmware 4.0.5 build 6030 and 128mb memory setting. scsi hd, ide cdrom, usb and bridged ethernet.

i had some bizzaro keyboard stuff. like pressing - gave me 'beta' sign... etc.. but you seem to know about dodgy keyboard

also, first type gui, command unrecognised, gui second time works...

if i start it without focus (vmware), it dies at speichserver init...
but if i mouse click inside vmware to give the vmachine focus during bios boot up, it loads up properly...

Thanks for testing and feedback!

Hmmm, the keyboard... some features aren't implemented yet: numeric keypad f. ex. As I have buildt a german keyboard layout, maybe it doesn't work well with vmware?! With the bochs it produces weird chars, especially on the german umlaut keys: ?,?,?.

You got a blank screen after starting the gui service? oy vey, how can i track this down? It's very likely some bug I've overseen.

Yesterday, I've got a blank screen after gui service start up because I wanted to paint controls at the wrong place - didn't work. At least I've got the window dragging stuff alright.

other:
Now, I am doing some redesign: not the video driver will be responsible for text output in graphics mode anymore but the gui service. the input/output streams are simply redirected.

stay safe

it was the main keyboard (my laptop has no numeric side keypad.)... i will boot it on my spare amd 700 duron 512mb box and see what happens.

Blue Illusion and Mobius tested...

Test system:

• 120 MHz Pentium CPU
• 16 meg RAM
• S3 PCI video board with 1 meg RAM and VBE 1.2 BIOS
• Windows Sound System sound board (the original! not a WSS compatible)
• 3-button Logitech PnP serial mouse on COM1
• Internal hardware modem on COM2
• No working CD-ROM
• No working Ethernet

Blue Illusion

The source code tarball would be a lot smaller if you delete
the binaries and the 'dump' files...

Reboots immediately after loading and displaying this message:
   MMSRV: VBE2.0 linear framebuffer at 0

The BIOS in my video board is only VBE 1.2

Under Bochs with 8 meg RAM, I get:
   BlueIllusion OS - INIT
   INIT: add driver tasks...
   pager init...
   video-subsystem init...
   video: maus init...
   timer initialisieren...
   system task init...
   floppy init...

   Event type: PANIC
   Device: [CPU ]
   Message: prefetch: running in bogus memory

Mobius

I disabled the PS/2 mouse and enabled the serial mouse in SYSTEM.PRO before booting. I also commented out the entire "KernelDebug" section. The console shell seemed to work; with two command lines on the top and bottom halves of the screen. Running /Mobius/winmgr caused an exception and register dump, though.

Just for grins, I re-enabled the PS/2 mouse driver in SYSTEM.PRO. This time, the GUI worked. It displayed an alert that said "No mice found. Check /System/Boot/System.pro" Pressing a key then caused an exception.

I substituted an old 2-button serial mouse, and I tried enabling both the PS/2 and serial mouse drivers, but the GUI still crashed.

Since I have an S3 Trio video board (vendor 0x5333, device 0x8811), I tried enabling the 'pci' device in the 'Devices' section of SYSTEM.PRO. When I started the GUI, I got a black screen. My monitor said the sync frequencies were out of range (horizontal sync = 62.6 KHz, vertical sync = 139 Hz). Seems like these frequencies are twice what they should be?

The online help shows that some commands require a backslash, but because the US and UK keyboard layouts differ, I can't type a backslash character.

The reboot-edit cycle was pretty tedious, so I tried loading
Mobius with MBLOAD. I unzipped mobius.tar.gz, since MBLOAD
still doesn't support gzip. Booting failed thus:
-- Mounting mobius.tar.gz on / as tarfs
Assertion failed: ramdisk_tar.c, linke 415
data == tar->end

Mobius under Bochs with 8 meg RAM:
keyboard.drv: not an executable (0)
ps2mouse.drv: not an executable (Thread kernel/1:
Interrupt 14 at 18:c0020c24: e001fa06
Page fault: page not present writing in kernel mode

I'd like to see something like the OSID card that lists system requirements (minimum CPU, RAM, VBE version), what hardware is supported (serial mouse!), and what testing has already been done on the OS (runs under Bochs, runs on a 486 system, etc.)

@chris: Thank you for testing!

Ow ow ow, the dump files and the binaries ... I admit, I've just tarballed the actual development tree. I'll upload a cleaned up source code tarball.

ok, here comes the requested information:

requirements:
video board - vbe 2.0 bios
ram - 64 mb
pointing device - ps2 mouse or emulated ps2 mouse via USB
keyboard is a must.

testing done: in the bochs - with 64 mb ram emulated:
*should* boot up. Should say: No vbe 2.0 available - no framebuffer. The gui should fall back to 320*200 mode if activated.

actual testing hardware: Athlon xp 1700 +, mainboard with asus bios, 512 mb ram, usb mouse, ps2 keyboard, creative nvidia2 3d video board with 32 mb, floppy, harddisks, a bios flasher connected to the second ide port, scsi cdrom. On this machine, BlueIllusion works.

It also works on some spare PC at work, which has 386 mb installed. cpu: pentium III 500 mhz. Graphics board has vbe 2.0 bios.

BlueIllusion refuses to work on toshiba laptops.

Stay safe

Chris: Thanks for testing mine, too! The GUI clearly needs more work, but I knew that. Mobius should work with MBLOAD if you reconfigure it to load all the files individually, so I'll experiment with that.

The S3 Trio code is a bit of black art -- my own Trio card only half works (it doesn't seem to work in any machine manufactured after 1996). I might drop that and rely on VESA instead.

The 8MB stuff -- hmm, I'll have to do some testing on small-memory configurations.

Well, it's good to see that it does something somewhere...

Clicker

Everything seemed to work fine... until stopped at
"Map-viewer bound to pannel 3 done."

Not sure if this is appropriate but pressing any keys after this point did nothing.

BTW, I like the debugging idea.

The Mobius

I attached an image showing where The Mobius stopped. I really like the clean manner in which The Mobius boots, though.




** This may to be too useful but I don't really know what else to say about it.. I'm new at this

[attachment deleted by admin]

xxchrisxx: Thanks for trying Mobius. You might have more luck with a more recent release; the ones on the Sourceforge download list haven't been updated in ages (years). The most recent stuff is at http://www.themobius.co.uk/sdk/.

The latest two builds are most likely to boot because they were released after I fixed some stupid mistakes which stopped the kernel from running on most machines (stupid, like, using a pointer variable's address with INVLPG instead of the address it points to; and forgetting to clear newly-allocated page tables).

xxchrisxx wrote: Clicker

Everything seemed to work fine... until stopped at
"Map-viewer bound to pannel 3 done."

Not sure if this is appropriate but pressing any keys after this point did nothing.

Can you tell me which version you tried (where you get the image from) and if there's something in the logs that looks unusual (Shift+F3, if you're still allowed to switch screens ...).

As far as i can tell from the current codebase, it should be trying to initialize "system threads" just after ...

BTW, I like the debugging idea.

Thx. I'm kinda proud of my kernel debugging system

Pype.Clicker wrote: Can you tell me which version you tried (where you get the image from) and if there's something in the logs that looks unusual (Shift+F3, if you're still allowed to switch screens ...).

As far as i can tell from the current codebase, it should be trying to initialize "system threads" just after ...

I downloaded version 0.8.8 located here: http://prdownloads.sourceforge.net/clic ... z?download

Thanks for trying Mobius. You might have more luck with a more recent release; the ones on the Sourceforge download list haven't been updated in ages (years). The most recent stuff is at http://www.themobius.co.uk/sdk/.

I'll shall download the newest image and get back to you

Pype.Clicker wrote: Can you tell me which version you tried (where you get the image from) and if there's something in the logs that looks unusual (Shift+F3, if you're still allowed to switch screens ...).

Here is a screenshot from Bochs

http://www.geocities.com/idontthinkalot/untitled.JPG

Tim Robinson: I just downloaded the new floppy images and everything booted fine. Really impressive! I only wish I had a kernel in those stages. I played around in the console for a while (dos cmds eh? then ran winmgr... it started fine. Then I ran the Hello World program, which also went fine. Great stuff !

xxchrisxx wrote: Here is a screenshot from Bochs
http://www.geocities.com/idontthinkalot/untitled.JPG

According to that screenshot, it appears that the system initialization ran fine (the "System Initialization:complete" line)

try Shift+F4, CTRL+F5 to play with modules.