[Clfs-support] CLFS stops booting after mounting root
Andreas Hofer
Andreas.Hofer at dsa-ac.de
Tue Dec 11 12:10:07 PST 2007
Joe Ciccone schrieb:
> Andreas Hofer wrote:
>>
>> My target is a Vortex86SX (486 instruction set without FPU) and my
>> host is a Core2Duo notebook. The funny thing is that I already managed
>> to build the embedded version of CLFS (VersionSVN-0.0.1-20070715-x86)
>> and run it successfully on the target.
>>
>>
> I'm assuming you followed the chroot method in the book, which is why
> you're having problems. The book is not stuctured for x86_64 -> i486 or
> i686 -> i486. The CLFS_TARGET enviornment variable only affects /tools.
> It does not determine the final system's target triplet. That is
> detected by automake's config.guess. 2 easy ways around this, load a
> uname hack into the kernel that makes uname -m return i486, then build
> the final system. Or wrap uname and transform the output.
>
> The sysroot book is going to be a way around this barrier once it's
> complete. There's still a lot of holes to patch before I would use it on
> your everyday system. Not to say the builds aren't very stable, they
> just still have a lot of influence from the host. Clfs-embedded uses the
> same method as Clfs-sysroot, which would explain why that worked.
>
>
Joe,
after some googling around I found the uname_hack at
http://cross-lfs.org/~jciccone/uname_hack.tar.bz2
I tried to compile the module for kernel version 2.6.22, without luck.
From the error messages it seems like the whole uname stuff in the
kernel has undergone major modifications since the code for the module
was created. I tried to figure out how to modify the code but that is
beyond my scope at the moment. Maybe you know how to change the code for
the new kernel API? Or alternatively, would it be possible to entirely
substitute all instances of config.guess by a script returning the
desired string? Though, of course the module way is more elegant.
IMHO an up to date uname-hack should be part os CLFS since often the
target system will differ.
So, any help to update the code is appreciated.
Cheers,
Andreas
More information about the Clfs-support
mailing list