Re: Multiboot overhaul

[Mikhail Vorozhtsov <mikhail.vorozhtsov@gmail.com>]

Marco Gerards wrote:
> Mikhail Vorozhtsov <mikhail.vorozhtsov@gmail.com> writes:
> 
> Hi,
> 
>> I'm working on a microkernel for x86_64 (with looking forward to ppc64
>> and sparc64 ports). As you know, microkernels are quite useless without
>> user-space servers that provide "classic" OS services. But microkernels
>> are so useless that they even cannot load servers from disk :). That's
>> why bootloader need to load servers images directly into memory in
>> addition to image of microkernel itself. For i386-pc we have multiboot
>> and it "just works". But it's really too i386-pc-specific. We also
>> have draft of next multiboot version, but I think it still lacks
>> flexibility (e.g. single 'flags' field for all image format, platform,
>> and machine related features requests, no versioning). So I've
>> implemented new loader (patch attached) focused on exensibility and
>> versioning.
> 
> Feel free to give feedback on the next multiboot.  I think you
> understand we can't just accept a loader for something that is not
> discussed and most people are happy with. 
Of course, posting the patch was just the shortest way to show people 
the idea. Otherwise it would've taken many lines of text to describe 
everything.
> What's on the wiki is the result of discussion, but it is still open for
 > discussion. Did you look at the information about this on the wiki?
Yes, but I think my solution is too different from multiboot to be 
merged with it. It's just two ways of doing things, both of them will 
work. To say the truth, if I had found multiboot2 implemented I would've 
just started to using it. But I didn't, so I said to myself "Let's 
design a loading scheme, it should be fun" :)