Re: Mojo 0.2.1 released
Posted: Sat May 23, 2009 11:43 am
Athlon x2 at 2.8 Ghz for me (that confirms your suspicions once more)
The Place to Start for Operating System Developers
https://f.osdev.org/
Indeed it does. Thanks for helping me.Tomaka17 wrote:Athlon x2 at 2.8 Ghz for me (that confirms your suspicions once more)
Yeah it annoys me too. But there is not much to do about it (assuming that it is the floppy version you tested). The main reason is that I load lots of graphics for the GUI and loading from floppy is slow. Since I have not (yet) implemented a floppy disk driver everything is loaded by GRUB during boot which makes everything seem much slower than if the loads where "spread out" more. The CD version should load in under 10 seconds and the floppy version without GUI in under 30 seconds. And it even used to be be worseAUsername wrote:It takes a little while to boot the kernel though and that kinda annoys me.
I can reproduce it but I will not spend a lot of time fixing it I think. Like the man complaining to the doctor that it hurts when he pokes his finger into his eye. The doctors advice: "Don't do that!"AUsername wrote:Pressing any key while loading causes Boch to throw an error: "kbd: OUTB set and command 0xa9 encountered"
It is not the GUI that is slow - it is Bochs Seriously, I get around 100 FPS in 640x480x32 and around 250 FPS in 640x480x8 using Qemu. But I am actually very impressed with Bochs' performance considering that I update the entire screen on each little change (using dirty lines to do partial updates is on my todo list).AUsername wrote:The puritygui is very slow and eventually it crashes Boch, when testing CPU usage it was almost always 100%
Even if Bochs operates in realtime rather then its default setting? ie using the pit: realtime=1 configuration option?It is not the GUI that is slow - it is Bochs
It has been quite a while since I have played with Bochs' configuration so I don't know - but I suspect it is still quite slow compared to eg. Qemu. But I still think that Bochs' is pretty amazing considering that it is pure emulation.neon wrote:Even if Bochs operates in realtime rather then its default setting? ie using the pit: realtime=1 configuration option?It is not the GUI that is slow - it is Bochs
Thanks, I hope you find some time to testneon wrote:(I have not tested your system yet; looks quite nice though )
I'm very happy to hear that it worked - and also that you didn't find any more bugs.AlexExtreme wrote:Works great with the updated build you posted, can't find any other bugs right now.
Thanks - but I can't take credit alone. The look is greatly inspired by OS X and actually also the earlier systems and Windows XP tooAlexExtreme wrote:I like your GUI style, btw
"inspired"? really? looks like you stole the hard disk icon, and a few other thingsclange wrote: inspired by OS X
I am actually a bit offended that you accuse me of stealing! That is a very serious accusation. So serious that I think you should have done your homework a bit better. I credit everything I use that is not my own - see http://softwarewizard.dk/mojo/index.php?page=credits. I clearly have a link to http://www.iconarchive.com/category/app ... esign.html which is an icon set containing this icon http://www.iconarchive.com/show/cats-ic ... -icon.html. The license clearly says "Free for non-commercial use". </rant>xDDunce wrote:"inspired"? really? looks like you stole the hard disk icon, and a few other thingsclange wrote: inspired by OS X
Thanks. That is one of my goal.xDDunce wrote: but aside from that, it looks REALLY good, and easy to use.
That sounds like a massive throw up by my kernel . I don't have Parallels myself (I use Fusion my Mac) so I can't test it. "can't deliver kernel message" means that an IRQ fired, was turned into an IPC message and then the receiving driver's message queues was full. Something weird is going on - I clearly have a massive bug in my kernel. It should never just shut down. I have a double fault handler setup with its own TSS and address space which should catch these errors... But clearly it doesn't happen. I will have to think a bit about what possible scenario could cause what you describe.xDDunce wrote:i tested the ISO image in parallels, and it crashed spectacularly. from the looks of it, it loads verbosely with a blue background text, and then something goes wrong... the screen fills with garbage, although the top line has a small amount of readable content (just says "timer: /" and then alot of garbage). after 2 seconds it then repeatedly prints "can't deliver kernel message" and then shuts down (could be a triple fault, parallels doesn't like them).
Thanks. I just tried this on my Mac http://www.kju-app.org/ and it works very well. It installed flawless in a few minutes and boots my system perfectly - almost. The RGB values are inverted so instead of a light blue background I now get a light pink backgroundxDDunce wrote:will be getting a native version of bochs working soon, and can't wait to see this OS working...
Looking good!
Cheers,
James.
Code: Select all
starting harddisk device driver [7]
internal kernel error in src/kernel/interrupt/interrupt.c at line 233:
failed expression: 'g_enter_kernel_time < g_exit_kernel_time'
Apology accepted. I just had to mention it. I care about rights - mine and others. I make a living making software - and I really, really, really like it when people pay me a lot of money for itxDDunce wrote:I apologize if I offended.
Thanks a lot.xDDunce wrote:To be honest, the icons look great. Just like the ENTIRE GUI.
Porting gcc and nasm (and the rest of the tools) is on my ToDo list - but just so much to do before I am ready for it. With my current plan I don't think it will happen the next 6-9 months. Being self hosting of course a goal.xDDunce wrote:Finally got round to testing it in Q and i'm loving it. Good work. Small lack in functionality, but I'm not complaining. If there was a gcc/nasm port for it, I would consider using it as a development environment, due to it's fast loading speed, simplicity AND good looks.
Hmmm, it should be impossible (like all other errors ). I keep track of the time stamp when entering and exiting the kernel to keep track of time. As usually it is probably my kernel throwing up. I have had this error before, but previously it suddenly disappeared (this always make me nervous). Could you try to reproduce the error with the kernel log level set to 4 and send me a screen shot of the result.NickJohnson wrote:I can't get the CD image working under VirtualBox, but even your text mode setup looks great under Bochs. I really like the Life simulator and Tetris game, although the latter crashed a couple of times. I can't wait to get the GUI working, because the screenshots look pretty impressive on your site.
By the way, the error I get in VirtualBox (OSE 2.2.2) is this:
Code: Select all
starting harddisk device driver [7] internal kernel error in src/kernel/interrupt/interrupt.c at line 233: failed expression: 'g_enter_kernel_time < g_exit_kernel_time'