OSDev.org

Hello,

I was wondering if anyone thinks that there would be a need for a real mode multitasking OS kernel. I have written one that works pretty good. It sets up a stack for each task and launching a new task is as simple as doing:
I don't think there is a need on modern pc's, but how about embedded stuff?

From what I heard, the standard for x86 embedded is the 486. Long story short, it means that protected mode has a better ROI.

Combuster wrote:From what I heard, the standard for x86 embedded is the 486. Long story short, it means that protected mode has a better ROI.

If you're looking for a better ROI, you might as well just look into something ARM based.

And ARM is almost easier than x86, until you reach things like virtual memory and interrupts

Hi,

Is there a market for an OS designed for ARM?

I've always assumed that systems that use ARM fall into one of these categories:

• Custom hardware with a custom "bare metal" application; where anything more than a boot loader and some libraries just gets in the way (no OS desired).
• Smartphones, PDAs and mobile phones; where an OS is desired, but it's impossible to change the OS and an OS developer would need to convince the manufacturers to throw away all the software they've already been using and adopt an untested alternative OS (more chance of being attacked by a flock of flying pigs).
• A few PDAs, notebooks, laptops and thin clients, where the OS can be replaced by the user; that are so rare that an alternative OS's target market would be smaller than the target market for the old/obsolete Playstation 1.

In addition, all existing 80x86 CPU manufacturers are currently targeting the small device market - VIA has for ages, Intel's Atom, AMD's Geode, SiS Vortex86, etc. I don't think it'll be long before most of the larger small systems (which are an alternative ARM OS's only market) shift to 80x86, if they haven't already.

Like it or not, to me it looks like the ROI for ARM is almost zero.

Of course I could be entirely wrong - I don't even know if any of the equipment I own uses ARM (I'm guessing my ethernet switches and KVMs might, but then they'd fall into the "custom hardware with a custom application" category).


Cheers,

Brendan

Brendan wrote:Like it or not, to me it looks like the ROI for ARM is almost zero.

But then again, ROI for any homebrew OS is 0 :).


JAL

Brendan wrote:Hi,

Is there a market for an OS designed for ARM?

http://www.openpandora.org?

That Pandora thing looks cool, but I think one of those mini-laptops would be better. And the laptops are around the same price.

[EDIT]
I take my previous statement back. That thing is a lot smaller than I realized. I watched it running Ubuntu in the video and it's got some nice speed there. Makes me want to learn a little more about ARM.

Well, now that I've thought about it a little more, multitasking in real mode is very easy to do. So if there was a need for it I'm sure a company could find someone to code it. Heck it only took me about 2 days.

Hi,

Solar wrote:http://www.openpandora.org?

Hehe - while you're pre-ordering your Pandora and waiting for it to exist, you might like to pre-order my OS for the Pandora and wait for that to exist too....


Cheers,

Brendan

Some of us have been part of the Pandora project from the start and have been work on a OS for it .
Its the next step up from the GP2x and for those who are not part of the dev team or can not wait, you can get started on beagleboard http://beagleboard.org/

ARM is the way to go, everything from the i-pod to the i-pone users it.

Hi,

Dex wrote:Some of us have been part of the Pandora project from the start and have been work on a OS for it .
Its the next step up from the GP2x and for those who are not part of the dev team or can not wait, you can get started on beagleboard http://beagleboard.org/

Sorry - I didn't mean to poke fun at these projects. I just think that if half of the people that buy these systems install your OS then you'd have 6 end-users...

50% of a small market (with no competition) can be much better than 1% of a large market like 80x86 (with lots of competition). However, even for obscure systems like these you're still competing against Microsoft (Windows Embedded) and Linux, so you're not really gaining much.

Dex wrote:ARM is the way to go, everything from the i-pod to the i-pone users it.

Everything from the proprietory locked-down iPod all the way to the proprietory locked-down iPhone?

I really liked this part of the iPodLinux wikipedia page:

Wikipedia wrote:The bootloader for the 4th generation iPod was extracted by Nils Schneider, a German computer science student. Previous software methods to extract the necessary bootloader to enable Linux to be ported no longer worked. Bernard Leach had previously discovered how to operate the piezo buzzer inside the iPod. Schneider was able to use this program with some modifications to make a series of clicks for each byte of bootloader. The extraction process took 22 hours to complete and required Schneider to construct a soundproof box to prevent outside interference with the process.

IMHO the only good thing about the iPod is that it's small enough to use as a suppository (although Apple don't recommend using this method when returning them to the place of purchase)...


Cheers,

Brendan

Just a note, "iPodLinux" is a port of uClinux to the older generations of iPods, the new iPhone/iTouch devices use a scaled down version of OS X.. and there is a project related to porting Linux to them is called "iPhoneLinux", and AFAIK, they only have a bootloader written at the moment.. but they've reverse engineered quite a bit.

I wouldn't completely belittle hobbyists interested in writing OS's for embedded devices, if anything, it looks like it takes a great deal of work.

http://www.iphonelinux.org/index.php/Main_Page

Presently, I think the OpenMoko phones are rather neat.. they appear to be semi-open.

Mods: Perhaps this topic should be... split?