mainline merge (was Re: Linux 2.6.23-rc8 for OpenMoko)

[Harald Welte]

On Sun, Sep 30, 2007 at 12:21:14PM +0200, Christoph Hellwig wrote:
> On Sun, Sep 30, 2007 at 12:14:31PM +0200, Pavel Mateja wrote:
> > > > Ben, perhaps some of these should be submitted to lkml.
> > >
> > > Yes, certainly the changed include path fixes at the very least look
> > > very appropriate
> > > to merge upstream.
> > >
> > > I'm not sure how the openmoko kernel team is working but is there some
> > > reason why you can't send these patches upstream yourself?
> > 
> > Hello.
> > Patches for ARM tree should be checked in 
> > linux-arm-kernel at lists.arm.linux.org.uk first.
> > See http://www.arm.linux.org.uk/developer
> > I'm affraid merge window for next release is over now.
> 
> So what?  You need to start the process at some point to get it merged.

well, yes and no.  It is my experience that patches are usually not well
received if submitted at 'inconvenient' times.

> Just to second Roland's question I was rather surprised to see such
> a big patch queue for a high profile project like this.  Are the
> openmoko developers to time contrained to send it upstead or are there
> some reasons you thing it's not ready.  If it's the first problem people
> like me and and maybe Roland could help out with it a little.

yes, it is mostly a time constraint problem.  If somebody wanted to help
out with this, I'm more than happy!

Prime candidates for merging are, IMHO:

* the s3cmci driver (s3c_mci.patch + the ones on top)
* the qt2410 machine support (qt2410-base, qt2410-s3c_mci-pdata)
* pcf50606.patch  (absolutely required for GTA01)
* gta01-core.patch
* gta01-jbt6k74 (required to display anything on the LCM)
* gta01-inputdevice (just for the single button)
* gta01-no_nand_partitions.patch (merged with gta01-core.patch)

with those patches above, the mainline kernel would actually boot on
GTA01 hardware and support most of its features.

please don't merge yet
* gta01-backlight (will build on generic PWM support)
* gta01-vibrator (will build on generic PWM support)
* gta01-power_control (source files need to be moved)
* ts0710.patch (very far from complete)
* hxd8-core.patch (bitrot, needs internal update)
* gta02-core.patch (the hardware is still in flux)


> All the arm-related patches look pretty clean so I wouldn't expect
> major obstackles.  

Well, the biggest issue is the dependency of both GTA01 and GTA02 on the
pcf50606 and pcf50633 patches, respectively.  And I still don't think
there is any proper infrastructure for things like power management
units in the kernel.  So the pcf506* patches will likely cause some
coughing, especially by you :)

There are some other issues, like the audio amplifier using a reserved
I2C address (to avoid some philips patent on it), which needs an ugly
hack in the i2c core layer.

Apart from that, I'm happy to see that you think most of my code is
clean :)

> What makes me wonder a little is the yaffs2 patch, though.  Not only
> is yaffs2 an utter piece of crap, but it's also not used at laest on
> my neo (fortunately!).

we support more devices than GTA01 from that kernel :)

-- 
- Harald Welte <laforge at openmoko.org>          	        http://openmoko.org/
============================================================================
Software for the world's first truly open Free Software mobile phone