Thread (109 messages) 109 messages, 10 authors, 2007-04-27

Re: [PATCH] Start split out of common open firmware code

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: 2007-04-12 06:36:38

On Wed, 2007-04-11 at 23:25 -0700, David Miller wrote:
From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: Thu, 12 Apr 2007 15:50:20 +1000
quoted
On Thu, 2007-04-12 at 15:34 +1000, Stephen Rothwell wrote:
quoted
This creates kernel/openfw.c (depending on CONFIG_OPENFW) and puts the
first trivially common bits from the prom.c files into it.
Is this the right place ?

I want to split flat DT from live DT in separate files, then we have
of_device.c and of_platform.c that will need merging too, and possibly
the filesystem... that's a lot to put in kernel/
We could make drivers/openfw/{device,platform}.c, maybe that would
work best?
Yeah, we could have something like:

 - base.c (the core stuff that stephen just merged)
 - parse.c (the parsers for irq tree, "reg"/"ranges", ...)
 - device.c
 - platform.c
(there is some work needed for the above to consolidate properly, I
suppose I'm the one who should do it since I'm to blame for most of that
code :-)
 - fs.c

In fact, we could even have trampoline.c for the powerpc prom_init
thingy though I'm not sure there is any interest in having that made
common with sparc, so it might as well stay where it is (in fact, we've
been considering moving it to the zImage wrapper and using the flattened
format to pass the DT along).

In addition, there are things that relate to the flat device-tree which
I don't know how much we want to make generic, both walking it, but also
things to generate one, mostly for use by iSeries and zImage wrappers...

Note that for the filesystem, we really want to stick to our binary
representation while you really want to stick to your text rep (we
decided to disagree iirc :-) I think we could keep a lot in common still
and just have a mount option to indicate how the file content is
exposed.

It's not really 2.6.22 material as far as I'm concerned as I'm pretty
busy with other things, unless of course Stephen feel like doing it
all :-) I'll see what I can do for the next cycle though.

Ben.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help