Re: My progress on armhf
lsorense@csclub.uwaterloo.ca (Lennart Sorensen) writes:
> On Mon, Aug 22, 2011 at 06:53:49PM +0000, Hector Oron wrote:
>> 2011/8/22 Lennart Sorensen <lsorense@csclub.uwaterloo.ca>:
>> > Any suggestions for what to fix next?
>>
>> Pick the one you dislike the most =)
>> http://wiki.debian.org/ArmHardFloatTodo
>
> Some things look like the are already fixed. Who maintains the page?
> Who decides what the buildd should be building? Is there a state for
> it somewhere like buildd.debian.org has? Oh never mind, those are linked
> to from debian-ports.org.
>
>> Bootstrapping compilers are extra points
>
> Oh yes, that is always fun (having done that for a uclinux system). :)
>
>> BTW, I realized kernel image shipped with imx53qs does not have device
>> mapper for LVM support. I have been trying to compile FSL kernel
>> without luck at the moment. It would also be interesting getting
>> proper mainline kernel support for this device.
>
> I looked upstream, and 3.0 looks broken for i.MX53, but 3.1 should be OK,
> so once it is released I was thinking of trying to get a kernel going.
hm... I've yet to test something newer than 3.0rcX on my loco but at
least I'll probably add some stuff to my efika patches to fix bugs. Be
warned that my patches are based on imx tree so a little bit more
experimental. At least, if you find some, details welcome :)
>
> Freescale's ubuntu kernel has 767 patches on top of 2.6.35.3 which to
> me is just insane. A lot of that appears to have gone upstream to
> Linus's tree by now, so I figure it is worth seeing what is missing from
> 3.1 once released to see what else needs pushing upstream.
My experience of fsl code, is that the tree contains a lot of stuff with
randoms bugs and not so good quality. Things are improving and they're
trying to push more things upstream.
>
> Would be nice to have a debian built kernel eventually after all, which
> means it has to go upstream first.
I guess we're not so far from having needed stuff to be able to build a
kernel for buildd (I guess only ahci is missing) but for the rest, it's
lacking things like:
- regulator (look for da9053 at the lkml, there has been several tries
to get something merged but failed until now).
- ipu
- sound [ It's wip but delay by lack of regulator ].
- check about mx51 and mx50/53 built into a single kernel.
There are probably other missing bits but I don't remember them atm.
Arnaud
Reply to: