China DOS Union

The Tang monk and his disciples relied on "begging for alms from the common people" all the way to reach the Western Heaven and obtain the true scriptures. A political party needs party members to pay party fees to operate for a long time.

"Before the cavalry arrives, the food and grass come first". Ask the "person upstairs", how to raise funds?

Money - indeed a problem! How can modern people be satisfied with just a bowl and a钵? I really don't know how Linus did it. Besides amateur creation, there is another way, which is college students. But nowadays, Chinese college students may not be too willing to do such things @_@.

Support DOS, and more to discover DOS!

DOS will not die out, but will only develop slowly. It is the simplest and most useful operating system. The more primitive and basic something is, the more vitality it has. Please let everyone slowly experience these words of mine!!!

To be honest, after reading everyone's discussions, I really lost my way.......

Starting to talk about the development of the DOS kernel, how does it develop?
Expand to protected mode? This has a lot to do with the CPU architecture. Just like some people said, many BIOS interrupts can't be used. Even if there is V86,

Then I thought to myself, the development of an operating system is inseparable from two things, the CPU and the production concept of the operating system. The CPU is the hard truth. Although I have heard from others that if the CPU doesn't provide the CALL instruction, JMP can still be used, but it's better if it is provided. But the current problem is that to develop the DOS kernel, do we need to keep up with the CPU architecture and new mechanisms? Although the influence of software is very deep, such as DPMI in protected mode, or the incompatibility problem of IA64 architecture that made AMD64 shine brightly,......

However, we can't expect a dedicated CPU, a CPU for DOS (maybe possible). INTEL's development in the IA64 architecture can be said to be "starting over". INTEL itself also said that although the X86 architecture is classic, it is still not perfect. Although they show people how excellent the IA64 architecture is, Microsoft is hesitant. When XP was very popular, Microsoft first considered software portability. From the market perspective, there are countless X86 architecture software facing 32-bit Windows. Moreover, with its dominant position in the desktop field, once it shifts to IA64, the software also needs to be remade, and the cost is huge!!! Microsoft does not like IA64, and downward compatibility just gave AMD a beautiful battle.

Do you think the CPU is important or the OS is important???
If possible, people fundamentally value the OS. For most people, a substitute factor for the CPU is "speed". Just choose a fast CPU. But if you want to use a linux system, "I haven't heard of it. People who buy computers here all use win98" This is what I often heard when I just bought a computer in 2000.
It can be said that the first software people come into contact with is the operating system. From a programming perspective, it well masks the hardware. We are not playing with the CPU, we are making software, but just such a software, we cannot ignore the existence of the CPU. Our concepts are largely restricted by the CPU. So software is not all-powerful, and without it, it's absolutely impossible.........

I personally think Gates has a good vision. He can really see the essence of things. Take DOS for example. This system was not originally Microsoft's. Microsoft bought it. After improving it by itself, it was renamed MS-DOS.
Even up to the current VISTA, there is still the familiar DOS interface we know. Although some things are different,
Microsoft's developers think that in the progress of the times, some things need to be retained, and some things can't be retained.
So MS-DOS is retained partially. This has nothing to do with those DOS fanatics, and it's not for those low-level workers.

The biggest feature of DOS is simplicity and comprehensiveness. It can do graphics and images, watch audio and video, and surf the Internet,
But the shortcoming is that compared with modern operating systems, there are too many restrictions.
Regarding kernel development, if we want to retain all the characteristics of DOS, then most DOS software can be smoothly ported, but how to cooperate with the CPU? Rewrite the I/O interface and interrupts by ourselves? It's possible, but it's too difficult and complicated

Also, since it's development, how to adapt to multiple CPUs? Nowadays, there are dual-core, 3-core, and even 4-core. How can we make DOS balance the work of CPUs? Relying solely on TSR is not the way.

These are all issues to consider for the development of DOS. The stagnation of FREEDOS-32, I don't think it's that simple, because there still need to be device manufacturers willing to write drivers and programs for your operating system. Under the trend of some economic interests.......
It's too difficult.

I am at least somewhat relieved to see everyone discussing enthusiastically.

The thought of this issue mainly arises from seeing that DOS is not keeping up with development. Many new hardware and new functions cannot be developed under 16-bit DOS, and currently, 32-bit DOS cannot truly coordinate the work of various programs.

If we want to seek the most software and hardware support, actually, we can take the path of "compatible with Win32 API". Maybe we can go to support the development of HX Extender and turn it into a powerful 32-bit extension platform under DOS, so that various applications can be developed in Win32 mode on the HX platform, and even Win32-compatible drivers can be compatible.

This essentially is equivalent to the work of Win9x. It is also equivalent to making HX a sub-operating system on DOS. All 16-bit DOS calls are still handed over to the DOS kernel to complete, and Win32 calls are completed by the HX kernel. We just strip off the GUI of Win9x, and can inject more advanced concepts into the development of HX. Win9x is no longer supported by Microsoft, while we can use HX to support the latest hardware and technologies.

If we want to continue the development of DOS, this may be the most feasible path.

This year, our amateur group has achieved that in pure DOS, multiple tasks can be run.... However, currently it only supports English. As long as experts write (the relevant code), high-speed operation can be achieved.

Currently it's a cooperation issue...

Today, after carefully reviewing the specifications of FreeDOS-32, I found it very inspiring.

Although the FreeDOS-32 project has been stagnant since 2005 up to now, I still pay occasional attention to it.

Those specifications mention that DOS is a non-kernel system. Then, when I specifically looked at the design concept of the FreeDOS-32 kernel, I discovered that it allows user programs to run in ring 0 to ring 3, in user space or kernel space, without imposing any protection on the programs, and there is no memory paging. Although the kernel is designed for single-task and single-user, such a design makes it possible to write a driver-like program to implement multitasking, thus both continuing the original characteristics of DOS and supporting new technologies. Then I suddenly thought that making DOS into a non-kernel system is actually a good idea. Such a design might even be able to be compatible with Linux/Windows drivers through wrapper-like programs.

After reading some documents, I generally understood that FreeDOS-32 is going to achieve compatibility with 16-bit DOS programs and DPMI programs by rewriting all DOS interrupts and BIOS interrupts, and its native programs call FreeDOS-32's own system calls, and the executable file format is still ELF.

If I had the ability and energy, I would like to port a set of GNU to FreeDOS-32, of course natively not DPMI. Then I would port the HX Extender to it to be compatible with win32 console programs.

I don't know if everyone has tried Windows hibernation and then restarting.
This kind of startup is a bit like the previous DOS startup, direct and fast.

Does the new DOS need to support graphics? Does the new DOS need to support the mouse?
Does the new DOS also need to support multi-processes? How do multiple processes interact with the user?

If the above technologies cannot be realized, even if the new DOS is developed, there won't be many users.
If the above technologies are realized, it seems to be like w......!

The meaning of building 39 is clear, that is, to have 32-bit linear addresses without paging, and all programs run in RING0. Such a system needs to do the following:
1. Abandon the BIOS (in fact, NT, linux have all abandoned the BIOS), and use drivers to manage hardware. Considering that hardware manufacturers are unlikely to specially write drivers for us, we can only compile drivers by referring to linux drivers.
2. Single-tasking. If it is multi-tasking, the timing interrupt cannot be guaranteed to respond in time.
3. Since there is no protection, one program can damage the entire system, so it is basically considered useless for general users.
4. Pure 32-bit (or 64-bit), and abandon the BIOS, so there is basically no compatibility with the original DOS programs.

Regarding these 4 points, what do you think of the prospects? Why did win and linux choose the multi-layer protection mode at the beginning? This is mainly to avoid "one bad program affecting the stability of one system".

As for the so-called command line compatibility in many previous buildings, that is really nothing. Maybe it can be solved with a Java virtual machine.

[ Last edited by netwinxp on 2009-5-9 at 10:15 ]

The 32-bit DOS kernel is only in the default state where 32-bit linear addresses have no paging, all programs run at ring 0, and it is single-tasking. But since there are no restrictions on accessing hardware, it is certain that driver - like programs can be developed to implement multi-tasking and protect program behavior. At that time, things like no paging and ring 0 can all be changed. Once a multi-tasking driver program is made, it will definitely not allow programs to have direct access to hardware, and there will be no situation where "a bad program affects the stability of a system". In fact, this is equivalent to implementing a kernel on this 32-bit DOS. I think that if FreeDOS - 32 is developed, theoretically, the Linux kernel and the NT kernel can all run on it.

As for 16-bit compatibility, it can be solved through virtual machines, just like Windows.

[ Last edited by brglng on 2009 - 5 - 19 at 11:42 ]

32-bit linear addresses without paging, all programs run at ring0, single-task, but since there is no restriction on accessing hardware...

Since all are at Ring0, space and hardware can be accessed, then

it must be impossible to develop programs like driver to achieve multi-tasking and protect the program behavior

this must be impossible. All programs are at Ring0, then it has the ability to access and modify any place in memory, the situation of intentionally or unintentionally (pointer error) modifying key memory data or code cannot be avoided. Windows is protected by the CPU to achieve key data and code, leaving only a few trap doors for privilege switching, it has been repeatedly broken privilege restrictions.
ps: If you don't let the program run at Ring0, then the so-called advantage of easy access to hardware is gone, and there is no advantage compared with win, linux.

[ Last edited by netwinxp on 2009-5-21 at 13:58 ]

This amateur group has been operating. Since it is multi-tasking, it must protect the program behavior. Welcome DOS friends to participate.
See the website:
http://blog.tianya.cn/blogger/view_blog.asp?BlogName=xie2005
There are pictures in it. Next time, send dynamic pictures. Or http://xie2005.tianyablog.com

If you understand, multi-tasking can be realized on ancient machines. 2009-10-15

[ Last edited by aq2007 on 2009-10-18 at 17:13 ]

Mainly developed under DOS.

[ Last edited by aq2007 on 2009-10-18 at 17:11 ]