On Fri, Apr 08, 2011 at 01:19:10AM +0200, Arnd Bergmann wrote:

On Friday 08 April 2011, Ryan Mallon wrote:

quoted

quoted

quoted

quoted

quoted

/sys/devices/soc/${NAME}/
/sys/devices/platform/soc${NUMBER}/

I prefer the second format here since the path is always the same which
makes it easier to write parsing tools. The name should be an entry in
the directory rather than the name of the directory itself.

In the first case, we would ensure that every directory under /sys/devices/soc
is an SOC device, so you would not need to parse the name at all, which is
even simpler. I'm open to other arguments either way, but I think this
one is not particularly important.

I think your proposal is best.

quoted

quoted

That would mean two separate instances of Linux, just like two systems.  
SMP across multiple SOCs doesn't make much sense, and I don't think they 
are likely to be designed for that ability either.

So we probably don't need the ability to have multiple SoC directories
under sysfs then?

There are systems that have multiple ones, they are just not as common.

quoted

For the other part of the question, do you think it makes sense for the
CPU's to be child devices of the SoC (and for the CPU devices to be
potentially symlinked from the SoC directory)?

No, the CPUs are children of /sys/devices/system, and we should not change
that. Symlinks sound like a good idea though.

Symlinks are a requirement as multiple cpus can be attached to a single
SoC.

What about multiple cpus that are attached to multiple SoCs?  Why even
try to describe this relationship, what would userspace get out of this
information?

thanks,

greg k-h