Re: Debian ARM architectures and subarchitectures
Luke Kenneth Casson Leighton <firstname.lastname@example.org> writes:
> On Mon, Apr 4, 2011 at 2:03 PM, Arnaud Patard <email@example.com> wrote:
>> Hector Oron <firstname.lastname@example.org> writes:
>>> I am not sure how current Debian subarchitectures map to. Is it just
>>> a giving name? Do they map to platform devices in linux kernel? Do
>>> they map to machine devices?
>> I've not checked but I guess it's following the name of the mach-*
>> directories in the kernel.
>>> If I were to add support for Freescales' i.MX51 cores, which would
>>> be the agreed subarchitecture?
>>> * mx51
>>> * mx5
>>> * mx5x
>>> * imx51
>>> * imx5
>>> * imx5x
>> I would say mx5 as in mach-mx5 in the kernel but someone needs to check
>> if it's really possible to have mx50/mx51/mx53 in same kernel. I
>> remember some people on l-a-k ml saying it was not possible.
> ignore what i wrote - arnaud's making a good point. bizarre as the
> discrepancy between the official name from freescale and the choice by
> lakml to drop the "i" is, following the exact same choice made by the
> linux kernel team(s) makes a lot more sense than inventing something
> from arbitrary choice.
Freescale is/(was ?) using mxc for common "part" of their SoC and
was using mach-mx51 for imx51 in their bsp tree. iirc, they've now
dropped mach-mx51 in favor of mach-m5 like in mainline. Don't know about
how was handled imx50/53 in their tree.