[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Possible ports

From: Vladimir 'φ-coder/phcoder' Serbinenko
Subject: Possible ports
Date: Tue, 30 Mar 2010 00:39:31 +0200
User-agent: Mozilla-Thunderbird (X11/20091109)

Hello, I would like to outline the usefulness of possible ports. It's
not exhaustive it's more like list of remarks.
1) the smallest embedded systems and microcontrollers. These ones simply
don't have any space to store multiple kernels or any interactivity
whatsoever. On such platforms the only grub usefulness remaining will
probably be supplying an upcoming multiboot2 protocol and perhaps
disaster recovery but on such systems one usually wants to shave off
every possible byte and they often use specialised ISA so basically
everything needs to be rewritten or heavily ported so even this
usefullness is reduced.

2) Bigger embedded systems. Those may have enough space for multiple
kernels and grub as well as some means to communicate. This would allow
a disaster recovery by choosing another kernel or changing kernel
parameters. Additionally these systems are likely to pick a more
standard ISA so it will allow more easy porting of payloads to this
platforms by using multiboot2
Intresting platforms: too many to list.

3) Handhelds and game consoles. These ones are pretty much like normal
desktop computer except having less power and a different communication
system. So additionally to normal port interface features would have to
be adapted but then all usual grub advantages kick in.
Intresting ISAs: ARM, SuperH, PowerPC (+Cell).
Interesting platforms: too many to list.

4) Standrard desktops and laptops. It's the most interesting target,
doesn't even need to be discussed.
The only systems I'm aware of in this category and not yet implemented
are Fuloong mini and ARM based systems

5a) Deprecated hardware. Not first target but is acceptable if doesn't
put too much problems for other ports.
The pre-PCI systems are 12+ years old and while some of them will live
for at least another 15 years they run out of parts which are even just
electrically compatible. Taking this road may leave us with a situation
when only about 10 working machines exist in the world, 2 users
interested, 0 developpers and huge effort spent.
Example ISAs: m68k, vax, sparc32

5b) PCI systems are another story. With GRUB having always more native
drivers these systems can use commodity PCI parts even if they aren't
supported by normal firmware as long as fimware is able to locate grub.
If e.g. a keyboard controller is burned
Interesting ISAs: Alpha.
Interesting platforms: Alphas, Old world macintoshes, SGI mips.

6) Paravirtualized environments are often used for security reasons. In
this context it's preferable to have a bootloader inside virtual machine
rather than outside. Some paravirtualized environments are carbon copy
of some real machines. Those need no special support but other platforms do
Interesting platforms: xen
Interesting ISAs: i386, x86_64, ppc, ia64

7) Servers have quite a special requirement and usually aren't rebooted
often yet alone changing of bootloaders. But it's a worthy target as it
often facilitates reusing kernels from desktop class. The problem is to
get such machine for testing. They often cost a lot and usually used
with thousand of simultaneous users logged i so you can't really ask
someone to have 10 reboots in one day to test new grub
Interesting architectures: ia64, system z.
I'm not aware of any other architecture in this category not currently
supported since we have support for i386 BIOS and EFI, PPC OFW and sparc

Vladimir 'φ-coder/phcoder' Serbinenko

Attachment: signature.asc
Description: OpenPGP digital signature

reply via email to

[Prev in Thread] Current Thread [Next in Thread]