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

Re: Donation of ARM CubieTruck to anyone interested in developing ARM-Debian-desktop



On Sun, Dec 8, 2013 at 7:39 PM, Luc Verhaegen <libv@skynet.be> wrote:
> On Sun, Dec 08, 2013 at 04:46:17PM +0000, Tim Fletcher wrote:
>> On 08/12/13 16:36, Alexey Eromenko wrote:
>>> If USB is not working then whole Cubietruck is not working.
>>> Because I plug the keyboard into USB.
>>>
>>> For my Cubietruck USB is working on Linux 3.4
>>
>> USB works fine on the 3.4 kernel, the point I was making is that I am
>> now running debian jessie on a mainline kernel. This was raised as a
>> problem during earlier discussions about the Cubietruck / A20.
>>
>> There are a number of things that aren't perfect but it's seemed to me
>> to be a major step forward for Debian on the A20 to have a mainline
>> fully OSS kernel.
>
> If you are running mainline, then you do not get to complain about bits
> or pieces not working. I would've thought that that was clear.

 ... let's think this through logically.

 a) debian-arm like all distros is a disparate mess as far as kernel
support is concerned (for reasons which we all know, and i've
summarised a number of times.  summary again: there's no BIOS, pretty
much zero common ground between designs, and unlike power-pc where the
hardware is both sparse and very x86-like, device-tree doesn't help).

 b) very few fabless semiconductor companies have the time, resources
or patience to go mainline: it's too hostile, too alien, and
completely contrary to business practices.  their focus is android,
not gnu/linux distros such as debian: one stable kernel per SoC is
more than adequate for them to make several million in sales.

 c) we on the other hand _do_ have the patience and a different focus,
but it *is* still taking time, to clear up the mess and the trail of
debris left behind.  that patience (and time) is however limited, so
has to be dished out sparingly.

 d) the extreme low-cost of the allwinner SoCs alongside their
attempts (despite a complete arse-of-a-manager's direct violations of
the Directors explicit instructions) to be GPL-compliant means that
it's worthwhile mainlining the allwinner kernel source.

 e) there is a lot of catching up to do.

 f) debian has not _quite_ caught up with the mainline catching up
that the sunxi community is patiently and slowly working its way
through.

so, luc, i think it fair to say that alexey and tim were, far from
complaining, merely pointing out something very useful, which is that
work is successfully being done, and that, by implication, at some
point in the future, thanks to a considerable amount of work being
done, the allwinner SoCs _will_ end up in a debian-released kernel.

l.


Reply to: