Re: Support for sunxi-based ARM systems in d-i
On Sun, 2014-05-18 at 19:41 +0200, Karsten Merker wrote:
> On Fri, May 16, 2014 at 09:59:58AM +0100, Ian Campbell wrote:
> [enabling flash-kernel to run inside d-i on machines listed
> in the flash-kernel machine db]
> > I think I've got it working, there were a few yakks to shave but I have
> > now done a full install from sid on a cubietruck. I'll clean up the
> > patches (rough versions attached), push and upload ASAP (hopefully
> > tonight).
> attached is a small patch against flash-kernel to add machine db
> entries for the Cubieboard 1/2 and the Mele A1000. These three
> are generally supported by the armmp kernel and have SATA
> connectors, i.e. with linux-image-3.14-1-armmp 3.14.4-1 (in
> testing since yesterday) it should be possible to do Debian
> installations on a SATA disk attached to them.
Ah, I didn't realise it had migrated, great!
flash-kernel still has 4 days to go, I think we should wait for 3.19 to
migrate to testing before uploading any more, so I think we should hold
off on your patch until then if that is OK with you?
The big stumbling block at the moment is finding a u-boot with AHCI
support. I've got a tree I can point people at until that gets sorted
> I unfortunately don't have access to any of these models, so this
> is purely based on information from the linux-sunxi wiki and from
> the dts files in the kernel source. Testers welcome :-).
I did consider doing a sweep over the dtb files which the kernel builds
and just adding them all, but then I was thinking it might be best to
fault them in as they are tested, any opinions?
> I'll probably get access to a Cubieboard2 sometime next week and
> will test an installation on it, but for the Cubieboard 1 and the
> Mele somebody else would have to give them a try.
I tested on cubieboard2 this morning and it worked ok
(my /etc/flash-kernel/db entry is identical to the one in your patch).
I'll write something up for the wiki shortly (perhaps this week,
although there is a beer festival in town...). I've got notes for the
fully integrated with flash-kernel case as well as the case where that
isn't present yet (i.e. how to bodge the first boot and manually
configure f-k), which should make it possible for people to get
something working and file a wishlist bug against f-k.
I've got an A20 based MELE of some sort on my desk at work, but it's
been sat there for months and I still haven't bothered to solder the
UART cable together, I wouldn't be on me getting round to it any time