[Arm-netbook] Linus wingeing about ARM
luke.leighton
luke.leighton at gmail.com
Fri Oct 12 07:58:02 BST 2012
On Fri, Oct 12, 2012 at 7:50 AM, Henrik Nordström
<henrik at henriknordstrom.net> wrote:
> ons 2012-10-10 klockan 18:14 +0100 skrev luke.leighton:
>
>> naah. i don't believe it. it still doesn't solve the issue of
>> having to compile code to suit a particular pipeline architecture, and
>> it doesn't even begin to solve the diversity of hard-coded devices
>> (and neither does devicetree - devicetree just moves the problem into
>> devicetree files).
>
> It's not a magical solution, but a huge step forward. The change is on
> very many levels including a massive tree reorganisation to make it more
> maintainable and strongly pushing developers into not duplicating code
> for different SoCs sharing the same or similar IP blocks.
wishing that to happen does not make it happen, especially when SoC
vendors do not communicate with upstream and repeatedly copy rather
than consolidate.
also, SoC vendors have nothing to do with hardware device manufacturers.
so it still does not resolve the issue of having to customise a
linux-kernel-plus-a-devicetree-file to a device.
previous situation: ODM uniquely customises a linux kernel to a device.
future situation: ODM uniquely customises a linux kernel plus a
devicetree file to a device.
so the linux kernel has pushed the problem away and out of their
responsibility? so what. actually creating a real hardware device is
still just as much a nightmare.
l.
More information about the arm-netbook
mailing list