OSDev.org

This is new release featuring:

• The finer-grained locking in the kernel.
• The new, improved C++ system classes.
• The new, improved GUI server.
• Fixed ATAPI code.
• Doxygen documentation.

URL: http://www.djm.co.za/spoon/download.php
Screenshot: http://www.djm.co.za/spoon/vesa26072005.png

Requires: Please use a real PC (Pentium2+ or recent AMD) or use a newer version of VMWare. Anything else is risky.

Things to try out:

• Boot the ISO.
• Click on the "run" button in the clock window.
• Click OK or run to start the terminal.
• The terminal has a history. Press up to access the last command. Run it and watch the trailer. (Unfortunately, the trailer only half-works. It has some bad bytes. Don't expect it to finish.)
• F12 reboots.

Details:

1. The kernel locking now uses a series of reader-writer locks to only lock the processes or threads that are needed. This basically means more messages and pulses can be flying around through the kernel at any one time, improving responsiveness.

2. The new C++ system classes are just cool and only help to improve the stability of the system as well as making development easier for the programmer. I'm making a hybrid of the old BeAPI and some of the new Java API.

3. The new GUI has window moving, proper clipping, proper syncing, etc.

4. The fixed ATAPI code is actually a fix to the timing method used for timeouts. Some machines have an unusual CMOS situation where the time returned is off by a few minutes. Either that or my BCD to decimal conversion was off. (more likely).

5. The doxygen documentation is just plain cool. I need to document more.



Durand.


PS. I've already been notified of a bug which causes the GUI server to crash when a mouse button is clicked at the very very edge of a window.

Hmmm...didnt seem to work for me in VMWare. Both the AMD and the i686 version wouldnt move the mouse and the clock simply did nothing.


CPU: AMD Athlon 64 3000+
RAM: 64 MB

VMWare 5.0.0 running on Windows XP x64 Edition


Ill try it out on a real machine later.

Cool. A 64 bit machine? I've never tested on a 64 bit machine before. Actually, neither have I tested on VMWare 5+. Do you think VMWare running in a 64 bit environment means it can't virtualize 32 bits exactly as should?

You might need more than 64 MB of ram. I doubt it but I'm not sure.

Thanks for the feedback. Not many people have had any luck with this release. I don't know why.

If anyone has got it working, please let me know.

Hi, I have AMD duron 750 with 384 MiB memory.
Is 'athlon' version good for my processor?
I downloaded it, and when i run on real pc, i've got a double fault ;D
Is it corrent??

Durons are about equivalent to an i686.

Well, I'm glad it double-faulted. Do you have the previous CS:EIP of the double fault location and perhaps the original fault?

I was reading that the Duron is the cost-effective cousin of the Athlon and beats the P3 in performance. So, it sounds like it should work ... but apparently not.

Also! The kernel should have refused to run if the CPU was missing feature flags in the CPUID return information.

got that screenshot of the illegal-opcode-fault which mpeg2decode has generated on my vmware 4.5.1?

If not, I'll resend it today evening.

Yes, thanks I haven't had a chance to investigate yet. mpeg2decode crashing is very strange; it's not my code

durand wrote: Well, I'm glad it double-faulted. Do you have the previous CS:EIP of the double fault location and perhaps the original fault?

I was reading that the Duron is the cost-effective cousin of the Athlon and beats the P3 in performance. So, it sounds like it should work ... but apparently not.

Also! The kernel should have refused to run if the CPU was missing feature flags in the CPUID return information.

Note that the 750mhz duron is a stepping 3 duron, which has less features than the (now) more common stepping 6/7 duron. It could well be you assume the duron/athlon has those features whereas it doesn't in stepping 3. Check with AMD for their manuals.

You might want to try checking your input server, kbd, ps2mouse, ATA, ATPAI, and VFS modules for conflicts. I found that if i remove all of those from the GRUB list that it worked and the clock actually worked and told time.

Everything else just gave a Virtural PC Error.

I'm sorry. Virtual PC, Bochs, old versions of VMWare and QEMU are not supported. Each of those have several features missing or have certain bugs which cause problems.

Old versions of VMWare didn't handle CPU state correctly, QEMU doesn't support VBE2 out of the box and Bochs didn't support the new stuff last time I checked.

VMWare 4.5+ seems to handle everything properly, as well as a real PC. at least... that's the plan.

Bochs supports up to i686 with mmx, sse, sse2, PAE, PGE, PSE, x86-64 and 3dnow! (I'm not sure if it supports SSE3 though, it should have SYSENTER as well at least)

Most of this stuff is "non-standard" though, the pre-built packages from their site are only i586 with a minimal feature set, you need to build it yourself to get these features.

durand wrote: I'm sorry. Virtual PC, Bochs, old versions of VMWare and QEMU are not supported. Each of those have several features missing or have certain bugs which cause problems.

Old versions of VMWare didn't handle CPU state correctly, QEMU doesn't support VBE2 out of the box and Bochs didn't support the new stuff last time I checked.

VMWare 4.5+ seems to handle everything properly, as well as a real PC. at least... that's the plan.

The same thing happened in VMWare 5 as it did in VPC.

The same thing happened when i tried it on a real pc too.

Hi, I downloaded a iso version of spoon. I was testing it on VMWare 5.0 under linux. When grub was loaded all servers and clock:D - screen was filled on the brown color. Clock didn't work and i couldn't move the mosue pointer:(
My PC:
Duron 750MHZ
384 MB RAM

Tora OS wrote:
The same thing happened in VMWare 5 as it did in VPC.

The same thing happened when i tried it on a real pc too.

Thanks for the help. Of those processes you mentioned (input server, kbd, ps2mouse, ATA, ATPAI, and VFS), they're all very INT/IRQ intensive and all the bug reports I'm receiving mention the same thing... system hangs, no response. Sounds like a driver problem.

So, I've redone the PS2 mouse driver, specifically because it contained the most questionable code and also because I was told that it didn't work in more than 1 machine. The new code is better, cleaner, smaller and works correctly. I have a feeling that this was a problem.


Thanks for all the testing everyone is doing. If anyone has some more time, please give the new ISO a try and let me know if it works now.

http://www.djm.co.za/spoon/download.php

Um.. I'm busy uploading at the moment but they'll be done soon. The date of the ISO should be 31 July or 1 August, depending on when it finishes.