BrokenThorn Entertainment

In addition why do we need to null ds and es? ds and es are used implicitly by several common operations which we would want to perform during boot. In the example given, we are getting our running system into a consistant state. Depending on your nasm settings (the ORG statement), the chances are ...

You will get errors because you are delaring an extern variable but not linking.

As a result, nasm cant work out where that jump is supposed to go, so cant make a valid executable.

~Andrew

D'oh - I have been doing C89 for so long that without syntax highlighting, i completely missed the fact that that was a declaration rather than just another line of code.

~Andrew

If you are looking for some constructive criticism: Your #defines at the top should have the 'U' suffix to imply unsigned. It changes the behaviour of the arithmatic in certain places. Change HmmHeader to be more like this: typedef struct HmmHeader { uint32_t addr; HmmHeader *next; uint32_t allocate...

Virtual environments tend not to have an MMIO hole at 15 MB because it is not backed by the legacy hardware which requires it. As for your specific example, I would say that the number is simply being reported wrongly; The only way virtual environments want to communicate RAM information is throug t...

I would agree that that calculation looks wrong. However, the information in the multiboot header is advisory only, and even then only gives you the available memory up to the first MMIO hole which is typically at 15MB for compatability. Realistically, the only way to accuratly determine the extent ...

The problem is that BIOS service routines can not be used in 32bit mode, which means that when the bootloader jumps into 32bit mode, it can no longer load anything from disk without a proper scsi driver.

It is certainly possible to use memory above the 1Mb limit in Real mode. This is known as Unreal Mode, and works because of a quirk with how the segment registers work in 32bit mode. In 32bit mode, a lot of information is cached in non-accessable bits in the CPU. This saves the CPU needing to repeat...

Hey, Sorry to call you out on this, but IA64 is the Intel Itanium architecture, not 64bit x86 archtecture. The naming is very confusing. 32bit: x86 (most common name) IA-32 (Intel name - default in their manuals) i386 (Linux 32bit stuff, including systems code) i686 (Linux 32bit kernel level stuff) ...

And a somewhat belated Happy New Year from me.

(Internetless at the time)

~Andrew

Copy and Move are very simple when thinking about what they are actually doing. With copy, you want to duplicate the data into another file. So, your copy program needs to open 1 file for reading, and create a new file for writing. Then just loop reading data from 1 file and writing it to the other,...

Ah - I misread you question. You can ignore about clipboards for your purposes, although they most certainly are not just a gui concept. a simple command line mv is trivial if you have a filesystem driver and some standard library functions for accessing the filesystem driver. If you lack those, the...

First of all, you need a clipboard. It doesn't nesseserally have to be in the kernel, but it does need to be system wide accessable. You need to have an API to read, write and clear the clipboard. Then, your command prompt keystroke interpreter should recognise certain keystoke combinations and use ...

The FAT12/16/32 spec is available as a .doc from the Microsoft Website, or look on http://www.ntfs.com/#fat%20file%20system

However, on a floppy disk, you really want to be using FAT12 as it has less filesystem overhead, meaning you can store more regular data on the data.

~Andrew

I had noticed that it had got substantially better recently.

Thanks,

~Andrew