OSDev.org

SOme news, some changes:

the GUI as ugly as it still looks, has got some widget stuff. The cooperation with the launcher application has become more important.

writing ext2-filesystem access (deleting a file is not fully implemented - I lack the time for that stuff currently - but better times will come *gg*)

one can use the mkdir command to create a directory.
The argvtest program creates a file which gets the first argument as name. ls should show the file in the directory.

ls has got some flags - ls h should reveal them.

Try out the dropdown menue Start - launch some applications and then give the wins menue a kick. It should show all open windows. *gg* the application "hangy" might be something for fun. ];->

well - enough of the bragging, here it is:

http://www.distantvoices.org/download/b ... mg.tar.bz2

mark: It works in vmware,qemu and on a real pc. The Bochs - it has worked on my last try - obviously, I have finally resolved some of the things bochs didn't like.

neat!

bochs seems to complain a bit, though ... first of

00000537766i[BIOS ] ata0-1: PCHS=20/16/63 translation=none LCHS=20/16/63
00000854428i[BIOS ] int13_harddisk: function 41, unmapped device for DL=82
00000858924i[BIOS ] int13_harddisk: function 08, unmapped device for DL=82
00000863472i[BIOS ] *** int 15h function AX=00C0, BX=0000 not yet supported!
00000867951i[BIOS ] *** int 15h function AX=5300, BX=0000 not yet supported!
00000872365i[BIOS ] *** int 15h function AX=5304, BX=0000 not yet supported!
00087232796e[KBD ] controller_enQ(): OUTB set!
00087233065e[KBD ] controller_enQ(): OUTB set!
00087233065e[KBD ] controller_enQ(): OUTB set!
00087233065e[KBD ] controller_enQ(): OUTB set!

just when displaying "hoechste_address: 1F9F000"

then issuing a couple of

00191178602e[DMA ] write to command register: value(04h) not 0x00
00191414985e[DMA ] write to command register: value(04h) not 0x00
00191489247e[DMA ] write to command register: value(04h) not 0x00
00191551938e[DMA ] write to command register: value(04h) not 0x00
00191626200e[DMA ] write to command register: value(04h) not 0x00

and finally, saying that "No VBE2.0 available" (which suprising comes from "MMSRV" ?)

i wasn't able to type "help" as i wish ... the system firstly was uber-slow (taking several seconds to respond to one keypress) and finally said "keyboard buffer full" after i tried to switch to another console ...

hm. QEMU seems to like it better and even found a framebuffer with VBE2 ...

But still, when i'm typing "gui", nothing occurs (well, the command terminates and the shell is ready again ... )

the bochs ... it is really helpless.

qemu - it surprises me that this has issues. Which version are you using? I've successfully tested it on version 0.6.

How much memory have you installed on the machine you're testing it on?

the mm service will be renamed to process service and be responsible for al lthe thread and process management stuff. I'm gonna remove the vbe from this service and stuff it into the gui service once and for all. *grmpf*

QEMU PC emulator version 0.6.0, Copyright (c) 2003-2004 Fabrice Bellard
usage: qemu [options] [disk_image]
-m megs set virtual RAM size to megs MB [default=128]

It's a 512MB machine or something ...

That's nearby the configuration I am using. Hm. I just wonder.

For bochs to be vbe2 capable, just point the vgabios parameter to the lgpl'd vbebios version from plex32 iirc.

I've put an interims version of blueillusion online.

Just check out the help command. It should give you all available commands. help [command] gives some detailed text about the specified command.

Then, you may try out the edit,touch,rm,rmdir,mkdir and killall commands.

YOu can, if you are very desperate, type in GUI to start the gui mode. It has heavy clipping issues and currently there is some bug which causes it to bail out in a double page fault upon accessing the text label control (with settext).

Thanks for your testing! I appreciate it!

I think I triggered the double page fault you were talking about. I'm using VMWare 4.0.0 build 4460.

nested pf!; pid: 3, pde: 0, address, 18BF000, free pages: 26063
gs: 10, fs: 10, es: 10, ds: 10
eax: 18BF000, ecx: 10, edx: C002D124, ebx: 0
esp: 18BF000, ebp: C013AD3C, esi: C002D1AC, edi: C002D198
eip: C0000FB4, cs: 8, eflags: 286
u_esp: C002D21C, ss: C000FAD3
double page fault caught in system process! Halting!

(*phew* next time, i'll just upload a screenshot.)

I'll give it a go on three real separate computers here and give the results tomorrow.

THanks for testing, durand!

Have uploaded a new floppy image yesterday. Something interresting to play around: the remove-file-from-filesystem stuff works initially. Have to implement double/triple indirect blocks removing stuff. Maybe this will be done in form of a recursive function.

Clipping issues should be gone for the main part of gui operations - only the window dragging mode reveals some weird stuff occasionally - but that vanishes.

the start menue doesn't pop up immediately, it needs a few clicks around to reveal the task bar and ... oh well, issues after issues: welcome to debugging sessions darker than night I'm working on it, so the floppy image is uploaded each day to represent the latest "nightly development build"

If you want to give it a shot, have my thanks I appreciate bug reports.

stay safe.